introduction to volume 9, issue 1 this issue contains 6 papers, written in chinese with english abstracts. the papers cover four topics related to risk analysis, which are incomplete geospatial data, earthquake, meteorology and crop water requirement. the paper “geospatial information diffusion technology supporting by background data” by chongfu huang, introduces well known information diffusion technology to supplement incomplete geospatial data to be complete. the diffusion is done with background data, which plays a role as a bridge to diffuse the information carried by the observations, obtained from observed units, to gap units. the suggested method has obvious advantages over the geographic weighted regression and the artificial neural network for inferring the observations in gap units. there are two papers in earthquake topic. the paper “study on the changes of industrial structure and its causes in sichuan province before and after the `5.12 wenchuan earthquake' ” by dongdong zhu, et al., uses the input-output table, the deviation share analysis method, structural decomposition method and industrial correlation analysis method to analyze the impact of the wenchuan earthquake on various industries in sichuan province. the second paper “rapid assessment of disaster loss and spatial distribution of intensity residuals in linfen area” by xiaofei han, et al., concludes that the probability of earthquake occurrence of high-intensity impact in the linfen basin is higher than that of the mountains on both sides of the basin, and the difference is significant. in calculating the seismic intensity, the difference in the seismic geological background and the complexity and homogeneity of the site conditions have been considered. there are also two papers related to meteorology. the paper “study on meteorological service policy for agricultural insurance in hebei province under the background of climate change” by kaicheng xing and shujun guo, with the climate models under moderate and high emission scenarios, given the distribution characteristics of mean annual and seasonal air temperature in hebei province in the first 50 years of the 21st century, and suggest some countermeasures to optimize the agricultural insurance policy and improve the meteorological service effect of agricultural insurance. the second paper “rainstorm warning information in beijing: exploring the local perceptions and views” by fangping wang, et al., using the descriptive statistics and non-parametric test methods and the public's cognition, evaluates and expectation of rainstorm warning information in beijing. the results show that more than 80% of the public can recognize the importance of rainstorm warning. the paper “the study of estimation on maize irrigation water requirement in the northwest of liaoning province” by xiaojing liu, et al., using penman-montes equation, effective coefficient method and cropland soil moisture index, estimated irrigation water requirement of maize at different growth stages of the northwest of liaoning province. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response vol. 9(1), march (2019), p. 1 doi: https://doi.org/10.2991/jracr.b.190328.007; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style the crisis of undercoordination lotfi a. zadeh* computer science division, department of electrical engineering and computer sciences university of california, 729 soda hall #1776, berkeley, ca 94720-1776, usa e-mail: zadeh@eecs.berkeley.edu abstract technological progress—especially in the realms of information processing, communication, transportation, commerce and distribution of energy—is leading to an accelerated growth in interdependence on all levels of modern society. a well-known result in systems analysis is that highly interdependent large-scale systems carry a high risk of catastrophic failures. to ensure stability, an increase in interdependence must be accompanied by an increase in coordination/regulation. in democratic societies there is an imbalance between interdependence and coordination/regulation. a manifestation of this imbalance is a chronic crisis of undercoordination. in large measure, the debt crisis of 2008 may be viewed as an eruption of the crisis of undercoordination. keywords: interdependence, coordination/regulation, undercoordination. *professor emeritus and director, berkeley initiative in soft computing (bisc), department of eecs, university of california, berkeley, ca 947201776; tel: 510-642-4959; fax: 510-642-1712. the painful debt crisis is not as much in the news today as it was a while ago, but it may flare up again. cyprus is a case in point. what is widely unrecognized—and what should be a cau se for worry—is that the debt to gdp ratio in the united states stands at 105%, not far behind greece’s 157% and far above china’s 15%. the debt crisis poses a serious threat to the financial stability of the world’s banking system. among the many contributing causes of the debt crisis, a cause that stands out in importance is the high interconnectedness of banking systems, with a failure of one system triggering failures of others. in large measure, high interconnectedness is a s ide effect of technological progress—progress which facilitates communication, trade, travel, networking and transfer of material and intellectual resources. in a 1974 paper entitled "on the analysis of large scale systems," i drew attention to what i called the crisis of undercoordination. in essence, the crisis of undercoordination relates to insufficiency of coordination in highly interconnected large-scale systems. two doctrines emerge from an analysis of behavior of such systems. the first doctrine is: highly interconnected large-scale systems are intrinsically vulnerable to catastrophic failures. the second doctrine is: to maintain stability of a highly interconnected large-scale system, an increase in the degree of interconnectedness must be accompanied by a commensurate increase in the degree of coordination. today, these doctrines are more or less conventional wisdom. however, what is not recognized to the extent that it should is that the doctrines in question apply to all large-scale systems -physical, social and economic. journal of risk analysis and crisis response, vol. 4, no. 2 (june 2014), 59-60 published by atlantis press copyright: the authors 59 willieb typewritten text received 17 april 2014 willieb typewritten text accepted 18 april 2014 willieb typewritten text l.a. zadeh the european debt crisis is a quintessential instance of the first doctrine. there are many other instances. an example drawn from the past is the massive failure of the electric power distribution system in the northeast in 1965. in that event, in the absence of coordination, the tripping of local circuit breakers spread unchecked through the whole system. the first doctrine is descriptive. the second doctrine is prescriptive. in general, coordination involves regulation, control and higher levels of taxation. in democratic societies, governments are responsive to the wishes of the electorate. nobody likes regulation, control and higher taxes. as a consequence, there is a chronic imbalance between the degree of interconnectedness and the degree of coordination. the crisis of undercoordination is a manifestation of this imbalance. in democratic societies the crisis of undercoordination is a chronic crisis which erupts when conditions are ripe. in this perspective, the european debt crisis may be viewed as an eruption of the crisis of undercoordination. the balance between interconnectedness and coordination is achieved without resistance when there is a co mpelling need for safety and stability, as in air traffic control. in democratic societies, the path of least resistance for dealing with undercoordination involves borrowing money, with future generations repaying the debt. this is the path of least resistance because future generations have no vote. in the united states, in particular, the crisis of undercoordination reflects—more so than in most other countries—the deep-seated tradition of distaste for big government. viewed against this backdrop, the credo of the republican party is in direct conflict with the second doctrine. in europe, the second doctrine was not considered when the european union was formed, resulting in a jump in the degree of interconnectedness. had it been employed, the chances are that we would not be faced with the european debt crisis. the crisis of undercoordination suggests a sobering thought. most democratic societies—with the exception of societies which have highly educated and sophisticated electorates, such as scandinavian countries—have become addicted to living on borrowed money, harboring seeds of self-destruction brought about by catastrophic failures of economic and financial systems. is this what awaits the united states? a troubling thought is that the 16 trillion dollar debt of the united states may well be a ticking bomb. published by atlantis press copyright: the authors 60 prespective research on the contagion mechanism of associated credit risk in the supply chain xiaofeng xie1,*, , yang yang2, jing gu3, zongfang zhou1 1school of management and economics, west china hospital, sichuan university, sichuan, china 2school of economics mathematics, southwestern university of finance and economics, sichuan, china 3school of economics, sichuan university, sichuan, china 1. 引言 伴随着市场经济的快速发展以及市场竞争的日趋激烈，竞争格 局已由企业之间的竞争转向供应链之间的竞争。供应链是指围 绕核心企业，通过对商品流、资金流、物流、信息流的控制， 涉及原材料采购、产品生产、产品销售等所有环节的由供应 a r t i c l e i n f o article history received 08 november 2019 accepted 17 january 2020 keywords supply chain associated relationships associated credit risk risk source contagion mechanism a b s t r a c t in recent years, the credit crisis of hainan airlines holding co., ltd., (hna), suntech, bogang, dunan, chengxing and other enterprises, as well as the fact that many enterprises and financial institutions in the supply chain suffer heavy losses caused by the credit crisis, make the associated credit risk in the supply chain widely concerned by the industry and academia, and also warn the importance of commercial banks to strengthen the credit risk management of supply chain customers. in order to correctly understand and understand the associated credit risk in the supply chain, this paper reveals the contagion mechanism of the associated credit risk in the supply chain based on multiple perspectives such as the origin, contagion path and basic characteristics of the associated credit risk in the supply chain. the research shows that factors such as supply chain enterprises’ own factors, the association relationship in the supply chain, the external environment of the supply chain and the relationship with the core enterprises form the contagion source of the associated credit risk in the supply chain. trade credit, loan guarantee and interpersonal relationship constitute the contagion channel of associated credit risk in the supply chain. the contagion effect caused by the association relationship in the supply chain makes the associated credit risk in the supply chain present the characteristics of infectivity, systematization and spillover. this study is helpful for the regulatory authorities to strengthen the management of enterprise credit risk in the supply chain. *corresponding author. email: xiexiaofeng@wchscu.cn 1基金项目：本文研究得到自然科学基金项目：71271043 & 71701066 & 71671144 & 71871447的支持。 通讯作者简介：谢小凤（1981—），女（汉族），重庆潼南人， 四川大学华西医院，博士后，副教授，研究方向：供应链管理，信用风险管理，养 老风险。 journal of risk analysis and crisis response vol. 10(1); april (2020), pp. 19–22 doi: https://doi.org/10.2991/jracr.k.200507.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr 供应链上关联信用风险传染机制研究1 1经济与管理学院，四川大学 华西医院，四川 成都 610041 2西南财经大学 经济数学学院，四川 成都 6111302 3四川大学 经济学院，四川 成都 610065 关键词 供应链 关联关系 关联信用风险 产生根源 传染机制 摘要 近年来，海航、尚德、渤钢、盾安、承兴等企业的信用危机，以及由此引发的诸多供应链上企业以及金融机构损失 惨重的事实，都使得供应链上关联信用风险受到业界和学界的广泛关注，也警示着商业银行加强对供应链客户信用 风险管理的重要性。为正确认识和理解供应链上关联信用风险，本文分别基于供应链上关联信用风险产生根源、传 染路径以及基本特征等多个视角，揭示了供应链上关联信用风险的传染机制。研究表明，供应链企业自身因素、供 应链上关联关系、供应链外部环境以及与核心企业关系状况等因素形成了供应链上关联信用风险传染源；商业信 用、贷款担保以及人际关联等供应链上典型关联关系构成了供应链上关联信用风险传染渠道。供应链上关联关系导 致的传染效应使供应链上关联信用风险呈现出传染性、系统性和外溢性等特征。本文研究有助于监管当局加强对供 应链上企业信用风险的管理。 © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). http://orcid.org/0000-0002-4640-9800 mailto:xiexiaofeng%40wchscu.cn?subject= https://doi.org/10.2991/jracr.k.200507.001 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ 20 x. xie et al. / journal of epidemiology and global health 10(1) 19–22 商、制造商、分销商、消费者所形成的整体功能网链结构。在 供应链环境下，由于链上企业之间实际业务的往来，供应链上 显然存在交易关联关系。随着供应链上企业联系的日益紧密 和广泛，它们之间往往还存在如商业信用、关联担保、债权债 务、交叉持股等导致的资产相关性，从而，供应链上还普遍存 在着资产关联关系。当供应链上企业管理者之间具有人际交往 或者管理者存在多重身份时，此时，供应链上存在人际关联关 系。各种各样的关联关系，使供应链上形成了一个紧密的信用 风险传染网络，一旦供应链上某家企业出现信用危机，将导致 与之关联的供应链上其他企业也遭受损失，进而增大了供应链 整体的信用风险。当供应链上某些企业发生信用违约时，导致 与其关联的供应链上其他企业也发生信用违约或者违约概率增 大，称这类信用风险为供应链上关联信用风险[1–2] 。 目前，供应链上关联信用风险及其传染效应已逐渐引起国 内外学者的关注。供应链上关联信用风险的研究文献大致 可划分为两类：供应链上信用风险，以及供应链上信用风 险之间的传染。在供应链上信用风险研究方面，walton和 gupta（1999）[3] 认为信息化水平的高低是影响供应链金融 信用风险的核心因素。hu和ansell（2007）[4] 引入宏观经 济变量以及企业财务指标，基于资源优势理论，运用专家评 分法分析供应链上企业信用风险的影响因素。tsai（2008） [5] 认为抵押资产减值的风险导致了供应链上企业的信用风 险。dyckman（2011）[6] 认为融资企业质押物所产生的风险 与融资结算周期的长短具有密切关系。goh等（2007）[7] 提 出了一个多阶段全球供应链网络问题的随机模型，对供应链 网络中企业的信用风险进行了度量。kull和closs（2007）[8] 以资源依赖理论为基础，度量了供应链上企业的信用风险。 江孝感等（2007）[9] 从供应链横向及纵向关系视角，对供应 链上企业的信用风险进行了综合评价。 在供应链上信用风险传染方面，love和zaidi（2010）[10] 在 具有债务关联的供应链上企业网络中研究了关联信用风险传 染问题。albuquerque等（2015）[11] 实证分析了跨国企业之 间商业信用风险的传染现象。chang等（2015）[12] 研究了 企业信用风险增大对行业内竞争对手信用风险，以及供应商 与客户信用风险的影响。jacobson和schedvin（2015）[13] 分 析了商业信用对供应链上企业信用风险的影响，研究表明供 应商的违约概率与其提供的商业信用额度正相关。我国学者 杨扬（2015）[14] 构建跳跃-扩散条件下的结构化模型，探 讨了商业信用风险对商业银行信用风险的传染效应。陈艺云 （2012）[15] 在核心企业违约的系统传染性以及非核心企业 违约的直接传染性前提下，构建数理模型分析了供应链上关 联信用风险的传染。张圣忠等（2013）[16] 构架了包括风险 源、风险载体和风险接受者的分析框架，研究了供应链上信 用风险传染的影响因素及内在机理。根据上述对供应链上关 联信用风险相关文献的回顾，本文将提炼出供应链上关联信 用风险的产生原因、传染路径和基本特征，试图揭示供应链 上关联信用风险的传染机制，为金融机构监管供应链客户信 用风险管理提供决策依据和新思路。 2. 供应链上关联信用风险的根源 美国经济学家弗兰克奈特在专著《风险、不确定性和利润》中 指出，风险是能够被量化的不确定性，表示收益或者损失存在 极大的不确定性。信用风险是指在资金借贷活动中，债权人面 临的资金可能遭受损失的风险。将债务人到期不能或不愿履行 借款协议，发生信用违约行为，致使债权人遭受损失的风险称 为信用风险。关联信用风险是指当某些关联信用主体违约时， 会对与其存在关联关系的其他信用主体的经营造成影响，可能 导致它们也违约或者违约概率增大。供应链上关联信用风险是 关联信用风险在供应链情境下的特例，是由供应链上关联关系 导致在供应链上企业之间产生的一类特殊信用风险。 供应链上关联信用风险与关联信用风险产生的原因既具有“ 共性”，也具有“特殊性”。在关联信用风险的研究中，信用主 体自身因素、信用主体之间的关联关系和外部宏观环境是导 致关联信用风险的主要因素。两者产生原因的“共性”在于， 供应链上企业自身经营状况、供应链上企业之间的关联关系 和供应链外部宏观环境也是导致供应链上关联信用风险的重 要因素。然而，在供应链情景下，供应链上关联信用风险的 产生，还与供应链上企业和核心企业之间的关系密不可分。 假如供应链上企业与核心企业关系密切，一方面，即使供应 链上企业在经营过程中出现财务危机，也能及时获得核心企 业的援助，缓解经营危机，从而防止了供应链上关联信用风 险的发生；另一方面，也使供应链上企业更容易获得外部融 资服务，从而具备了较强的风险抵御能力，使其不易受到供 应链上其他企业信用风险的传染，有效阻断了供应链上关联 信用风险的传染。因此，其“特殊性”在于，供应链上关联信 用风险的发生还受到供应链上核心企业信用状况，以及链上 企业与核心企业的关系这两个因素的影响。本文在借鉴前人 对关联信用风险原因分析的基础上，将供应链上关联信用风 险产生的原因归纳为供应链企业自身因素、供应链上关联关 系、供应链外部环境、与核心企业的关系状况等四个方面。 2.1. 供应链企业自身因素 相对于一般企业而言，供应链上企业之间的利益相关性更 强，一旦供应链上某家企业经营出现异常，危机将迅速传导 给供应链上其他企业，导致整条供应链收益受损。从风险传 播者的视角，供应链上企业经营和财务状况越理想，自身信 用风险越小，对供应链上其他企业经营活动产生负面影响的 可能性就越小，导致供应链上关联信用风险的可能性就越 低。从风险接受者的视角，如果其经营状况较好，资金流动 性较理想，就具备了较强的抗风险能力，当供应链上其他企 业发生违约行为时，就不容易被其信用风险传染，相应地供 应链上关联信用风险产生的可能性就小。因此，供应链上企 业自身经营状况直接影响供应链上关联信用风险的产生。 2.2. 供应链上关联关系 供应链上企业之间的交易、资产或者人际等关联关系，使供 应链上企业之间形成利益共同体，导致信用风险极易在供应 链上复制和传播，进而影响供应链整体收益。一方面，供应 链上关联关系使资源可以在供应链上更有效地分配，使供应 链上的中小型企业也能获得经营所需的资源，从而提升供应 链整体效益。另一方面，供应链上关联关系将供应链上企业 进行信用捆绑，给供应链上其他企业带来额外的风险，可能 导致供应链上更多关联企业惨遭损失，最终危及供应链的整 体运行。因此，供应链上企业之间的关联关系，是供应链上 关联信用风险传染的桥梁和纽带，是产生供应链上关联信用 风险必不可少的要素。 2.3. 供应链外部环境 宏观经济和政治环境使供应链上企业具有相似的经济基本 面，或面临相同的外部冲击。当宏观经济环境或国家政策发 x. xie et al. / journal of epidemiology and global health 10(1) 19–22 21 生改变时，将会使供应链上所有企业都受到不同程度的影 响，可能使链上企业普遍出现信用状况恶化，进而导致了供 应链上关联信用风险。在货币政策宽松时期，企业可以较为 便利地获得外部融资，发生信用风险的可能性较小。一般而 言，当宏观经济环境发生极端变化时，企业发生信用违约几 乎难以幸免，严重时可能影响行业内所有企业，甚至造成严 重的经济危机。 此外，由于供应链上企业多为中小型企业，其经营规模小、 固定资产少、财务状况不透明，使其信用状况难以被信贷机 构准确甄别。如果其因恶意欠款、不诚信经营等失信行为所 接受的惩罚低于其带来的收益时，这种失信行为会不断发 展，进而导致供应链上关联信用风险的产生。因此，当前信 用制度的不健全和惩戒制度的不完善，也是造成供应链上关 联信用风险发生的外在因素。 2.4. 与核心企业的关系状况 当供应链上企业因经营资金不足，向商业银行申请融资服务 时，商业银行主要通过对链上核心企业的信用状况，以及链 上企业与核心企业的合作关系进行评估，以此作为是否对链 上企业发放贷款的主要依据。因此，供应链上企业的融资约 束与供应链上核心企业的信用状况以及供应链上的关系状况 密切相关。 与供应链上核心企业的关系状况将对供应链上企业产生两种 效应：直接效应和间接效应。直接效应是指，当供应链上企 业陷入财务危机后，如果其与核心企业关系紧密，经营状况 良好的核心企业可以提供商业信用、资金拆借或者信用担保 等援助行为，有效降低链上企业违约的概率，进而降低引发 供应链上关联信用风险的可能性。间接效应是指，如果产品 的市场前景较好，且供应链上企业与核心企业存在紧密的合 作关系，其更容易获得商业银行提供的融资服务，进而在源 头上抑制了供应链上关联信用风险的发生。反之，若供应链 上企业与核心企业关系松散，因不能及时得到核心企业和商 业银行的救助，财务状况不断恶化，导致信用风险持续增 大，进而增加了供应链上关联信用风险发生的可能性。因 此，供应链上企业与核心企业的关系状况，直接影响供应链 上关联信用风险产生的可能性。 3. 供应链上关联信用风险的传染路径 供应链上关联信用风险的传染是指链上一家企业违约，引起 链上其他企业也违约或者违约概率增大，反映了供应链上企 业之间违约的相互联系。因此，供应链上关联信用风险传染 的途径根本在于供应链上企业之间具体的关联关系。在供应 链情景下，核心企业和上下游企业之间的“赊销”行为形成的 关联关系，供应链上贷款担保形成的关联关系，以及供应链 上企业管理层之间人际关系形成的关联关系等等，都是供应 链上关联信用风险的主要传染途径。 3.1. 商业信用 作为一种最古老的融资方式，商业信用广泛存在于供应链上 的往来交易中。商业信用，可以使供应链上中小企业的信用 得到增级，通过将资金注入供应链上核心企业，进而分配至 整条供应链，使供应链上企业之间资金、债务往来更加紧 密。供应链上企业之间的商业信用，使供应链上企业形成以“ 商业信用”关系为纽带的信用联盟，协调了供应链上资金资源 的合理分配，保障了供应链上运行环境的稳定。然而，一旦 供应链上某家企业陷入财务危机，财务危机将沿着商业信用 传染渠道在供应链上不断蔓延，引发违约的连锁效应。 3.2. 贷款担保 供应链上的贷款担保是指，在供应链上企业向商业银行等金 融机构融通资金的过程中，由供应链上核心企业为其债务提 供担保，当供应链上企业客观上不能或者主观上不愿按约履 行债务时，由供应链上核心企业按照担保合同承担相应的偿 还责任，从而保障了商业银行等金融机构债权的实现。对于 存在贷款担保关系的供应链上企业而言，当贷款的供应链上 企业违约时，为其担保的供应链上其他企业需要承担连带赔 偿责任，也可能因此陷入财务困境，从而使信用风险在整 个供应链上不断循环和放大，导致了供应链上的关联信用风 险。 3.3. 人际关联 在当前市场经济活动中，人际关联已发展为一类广泛存在的 经济关系。如果供应链上企业管理层之间存在如朋友、亲戚 等人际关系，当供应链上企业出现经营危机时，与其存在人 际关联的供应链上其他企业往往会伸出援手，通过提供资 金、代偿等行为进行救助。例如，2016年，乐视系陷入财务 困境，贾跃亭向朋友们求助，上至融创中国董事长孙宏斌， 下至乐视网高管、投资者和供应商都伸出援手，帮助乐视暂 缓了财务危机；2017年，留学培训公司“小马过河”资金链断 裂，创始人许建军向朋友们借钱，暂时渡过了劫难。石军伟 等（2007）[17] 和钱锡红等（2009）[18] 研究发现，企业管 理层的人际关系网络能显著正向影响企业成长。一方面，救 助行为可以帮助供应链上企业缓解资金压力，进而降低了违 约风险；然而，另一方面，救助行为也可能引火烧身，使救 助企业自身也陷入财务困境，导致违约风险增大。人际关联 关系，可能使信用风险在供应链上被吸收和消化，也可以使 信用风险在供应链上复制和放大，导致了供应链上的关联信 用风险。 4. 供应链上关联信用风险的基本特征 4.1. 供应链上关联信用风险具有传染性 对于供应链上企业而言，它们之间复杂的关联关系，往往会 产生“一荣俱荣，一损俱损”的现象。供应链上企业之间普遍 存在着各种各样的关联关系，当供应链上某家企业违约时， 不仅会给与之关联的供应链上其他企业的经营活动造成影 响，而且与之关联的供应链上其他企业往往会对其进行“救 助”，可能使自身周转资金紧张而陷入财务危机。因此，当供 应链上一家企业发生信用违约时，在一定程度上意味着供应 链上其它企业也可能存在着较高的信用风险水平。 4.2. 供应链上关联信用风险具有系统性 由于供应链上存在多家上下游企业，其触角已涉及全球范围 内多个行业和地区。目前，供应链已演变为多家上下游企 业构成的复杂供应链系统。关联关系在供应链系统上形成 22 x. xie et al. / journal of epidemiology and global health 10(1) 19–22 风险传染网络，当供应链系统上某家或者某些企业发生信用 违约时，信用风险将很快在整个供应链系统上传播、循环和 放大，进而演变为供应链上的系统性风险。特别是，在经济 全球化和业务多元化背景下，供应链系统上的企业可能涉及 到多个行业，这就使得供应链上信用风险不容易被分散和缓 释。因而，供应链上关联信用风险呈现出系统性风险特征。 4.3. 供应链上关联信用风险具有外溢性 供应链上关联信用风险不仅会在供应链内部传播，还会对资 本市场中的金融机构产生影响。商业银行等金融机构与供应 链上企业之间由信贷或者担保等形成的关联关系，使它们组 成利益共同体，一旦供应链上关联信用风险呈现传染效应， 将迅速波及到资本市场中的金融机构。比如，当供应链上企 业无法按期足额偿还贷款，将损害相关金融机构的利益，金 融机构信贷风险相应增加。因此，对商业银行等金融机构而 言，供应链上关联信用风险存在显著的外溢效应。 5. 总结 通过对供应链上关联信用风险相关研究成果的梳理和总结， 本文剖析了供应链上关联信用风险概念的内涵，提炼了供应 链上关联信用风险的产生原因、传染路径和基本特征，进而 揭示了供应链上关联信用风险的传染机制。研究表明，供应 链上关联信用风险的传染源主要包括供应链企业自身因素、 供应链上关联关系、供应链外部环境、与核心企业的关系状 况等关键因素。供应链上关联信用风险的传染渠道除商业信 用、贷款担保等供应链上典型关联关系以外，还受到供应链 上下游企业管理者之间的人际关联关系的影响。供应链上关 联关系导致的传染效应使供应链上关联信用风险呈现出传染 性、系统性和外溢性等特征。本文研究有助于金融机构理解 和识别供应链上关联信用风险的传染，为监管当局针对供应 链客户信用风险管理提供了新的思路。 conflicts of interest we declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, “ 供应链上关联信用风险传染机制研究”. authors’ contribution 作者谢小凤和杨杨负责提炼和总结关键研究结论，并撰写论 文，作者顾婧和周宗放负责论文设计构思、资料收集、结果 分析等研究工作。 acknowledgments this study was supported by national natural science foundation of china (grants: 71271043 & 71701066 & 71671144 & 71871447). 参考文献 [1] li y, zhou z. associated credit risk contagion between enterprises based on the imperfect immunization scenario. proc comput sci 2015;55:336–40. [2] xie xf, yang y, zhou ym, zhou zf. the clustering analysis on data and lending strategy of supply chain with information asymmetry method. cluster comput 2019;22:10299–307. [3] walton sv, gupta jnd. electronic data interchange for process change in an integrated supply chain. int j oper prod manage 1999;19:372–88. [4] hu yc, ansell j. measuring retail company performance using credit scoring techniques. eur j oper res 2007;183:1595–606. [5] tsai cy. on supply chain cash flow risks. decision support syst 2008;44:1031–42. [6] dyckman b. supply chain finance: risk mitigation and revenue growth. j corp treas manage 2011;4:168–73. [7] goh m, lim jys, meng f. a stochastic model for risk management in global supply chain networks. eur j oper res 2007;182:164–73. [8] kull t, closs d. the risk of second-tier supplier failures in serial supply chains: implications for order policies and distributor autonomy. eur j oper res 2008;186:1158–74. [9] 江孝感, 陈丰琳, 王凤. 基于供应链网络的风险分析与评估方 法. 东南大学学报(自然科学版) 2007;2:355–60. [10] love i, zaidi r. trade credit, bank credit and financial crisis. int rev finance 2010;10:125–47. [11] albuquerque r, ramadorai t, watugala sw. trade credit and cross-country predictable firm returns. j finan econ 2015;115:596–613. [12] chang jh, hung mw, tsai ft. credit contagion and competitive effects of bond rating downgrades along the supply chain. finan res lett 2015;15:232–8. [13] jacobson t, von schedvin e. trade credit and the propagation of corporate failure: an empirical analysis. econometrica 2015;83:1315–71. [14] 杨扬. 基于赊销行为的供应链企业信用风险. 系统工程 2015;12:55–60. [15] 陈艺云. 违约传染与供应链金融的信用风险测度. 统计与决 策 2012;1:33–5. [16] 张圣忠, 庞春媛, 李倩. 供应链违约风险传染的形成机理及影 响因素研究. 商业经济研究 2013;29:51–2. [17] 石军伟, 胡立君, 付海艳. 企业社会资本的功效结构:基于中国 上市公司的实证研究. 中国工业经济 2007;227:84–93. [18] 钱锡红, 徐万里,李孔岳. 企业家三维关系网络与企业成长研 究. 中国工业经济 2009;250:87–97. https://doi.org/10.1016/j.procs.2015.07.058 https://doi.org/10.1016/j.procs.2015.07.058 https://doi.org/10.1016/j.procs.2015.07.058 https://doi.org/10.1007/s10586-017-1281-x https://doi.org/10.1007/s10586-017-1281-x https://doi.org/10.1007/s10586-017-1281-x https://doi.org/10.1108/01443579910254222 https://doi.org/10.1108/01443579910254222 https://doi.org/10.1108/01443579910254222 https://doi.org/10.1016/j.ejor.2006.09.101 https://doi.org/10.1016/j.ejor.2006.09.101 https://doi.org/10.1016/j.dss.2007.12.006 https://doi.org/10.1016/j.dss.2007.12.006 https://doi.org/10.1016/j.ejor.2006.08.028 https://doi.org/10.1016/j.ejor.2006.08.028 https://doi.org/10.1016/j.ejor.2007.02.028 https://doi.org/10.1016/j.ejor.2007.02.028 https://doi.org/10.1016/j.ejor.2007.02.028 https://doi.org/10.1111/j.1468-2443.2009.01100.x https://doi.org/10.1111/j.1468-2443.2009.01100.x https://doi.org/10.1016/j.jfineco.2014.10.007 https://doi.org/10.1016/j.jfineco.2014.10.007 https://doi.org/10.1016/j.jfineco.2014.10.007 https://doi.org/10.1016/j.frl.2015.10.006 https://doi.org/10.1016/j.frl.2015.10.006 https://doi.org/10.1016/j.frl.2015.10.006 https://doi.org/10.3982/ecta12148 https://doi.org/10.3982/ecta12148 https://doi.org/10.3982/ecta12148 microsoft word the combustible materials remote sensing ration calculation and fire risk dynamic.doc 1 the combustible materials remote sensing ration calculation and fire risk dynamic monitor in the seasons of withered grass in xilingoule grassland zhuo yi graduate school of caas, beijing 100081, beijing liu guixiang* grassland research institute, caas, hohhot 010010, inner mongolia yu fengming inner mongolia climate center hohhot 010051,, inner mongolia abstract the amount of combustible material is the primary factor which influences the occurrence and developing of fire. from analyzing the field survey data during the six months of the whole seasons of withered grass in xilinguole, we obtained the different types grassland dynamic changing rates, which include the meadow steppe, steppe and desert steppe. on this basic, we estimated the amount of combustible material in each month by modis. so we realized the big quality and fast dynamic monitoring of combustible stock in seasons of withered grass from prairie. ues many factors data : the meteorological drought index(spi), the combustible stock, vegetation continuous and snow conditions , return fire risk index , which apply the important parameter for the alarming of prairie fire. key words: grassland fire; combustible materials; amount of herbage; remote sensing; dynamic monitoring 锡林郭勒草原牧区枯草季可燃物遥感定量反演与火险动态监测研究 卓义 1 1 中国农业科学院研究生院, 北京 100081, 北京 zhuoyii@126.com 刘桂香 2* 2 中国农业科学院草原研究所, 呼和浩特 010010, 内蒙古 liugx804@163.com 于凤鸣 3 内蒙古气候中心，呼和浩特 010051，内蒙古 yfm19790129@126.com 摘要 可燃物存量是影响草原火灾发生与发展的首要因子。通过对锡林郭勒盟整个枯草季 6 个月份的野外调查数据进行分析， 获得了草甸草原 、典型草原和荒漠草原等不同草地类型可燃物的动态变化率，在此基础上利用 modis 遥感数据估算了各月 * 通讯作者,e-mail :liugx804@163. com 基金项目:公益性行业科研专项“草原火灾应急管理技术”（200903041），中央级公益性科研所基本科研业务费专项“基于gis与遥感技术的锡林郭 勒草原地区干旱灾害监测、评估研究”，中央级公益性科研所基本科研业务费专项“锡林郭勒草原干旱灾害风险评估研究”资助。 作者简介:卓义,女,内蒙古呼伦贝尔人,在读博士研究生,主要研究草地资源生态环境与遥感监测技术研究。 journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 65 author’s names 份的可燃物存量。实现了原草区枯草季节可燃物存量的大面积、快速的动态监测。应用气象干旱指数 spi 与地表可燃物 存量、植被连续状况、雪被状况等数据叠加计算的多因子火险指数，有效表达了地表可燃物致灾的危险程度， 为草原火灾预警提供重要参数。 关键词: 草原火灾；可燃物；牧草存量；遥感；动态监测 1. 引言 可燃物是草原火发生、发展的物质基础。构成可 燃物的两在主体为干枯牧草和动物粪便。动物粪便 在草原上的分布零散,一般形不成连续排列,而且其易 燃性比牧草弱的多，在牧草燃烧之前通常不会首先 燃烧。草原上的枯黄牧草是最重要的可燃物,是草原 火燃烧的主体成分和基础。在草原火灾风险研究 中，可燃物作为草原火灾暴露性的首要影响因子， 实现对枯黄牧草动态的、定量的监测对草原牧区火 灾的火险管理有着重要的意义。 枯黄牧草作为草原可燃物的主体，其存量随时空 变化处于动态变化中并决定着草原牧区火灾的风险 程度。利用遥感数据对草原生长季的产草量进行监 测与估产方面的研究已有了广泛的应用。对于枯草 季节的牧草存量由于枯草的光谱特征与土壤背景相 似程度较高，虽然少数学者对应用noaa/avhrr、 modis等数据建立了枯草遥感监测模型[1]，但效果 并不理想，这已经成为了草原火灾风险的定量化研 究一个障碍。本研究应用成熟的生长季节产草量遥 感反演模型计算牧草枯黄前的鲜牧草存量，然后通 过牧草的干鲜比例以及牧草在枯草季节存量的动态 变化率来估算枯草季节各月份可燃物枯草的动态存 量。这种方法绕过常用遥感数据对枯黄牧草监测能 力有限的瓶颈，实现了原草区枯草季节可燃物存量 的大面积、快速的动态监测。 1. 研究区概况 锡林郭勒草原位于内蒙古自治区中东部，地处北 纬41°35′~46°46′，东经111°09′~119°58 ′。土地总面积 202 580 km2, 其中天然草场 197 000km2，占草场总面积的97.2%。地形以高平原为 主体，平坦开阔。气候类型为中温带半干旱大陆性 气候，冬季寒冷，夏季炎热，降雨不均，雨热同 期。土壤类型多样，地带性土壤有灰色森林土、黑 钙土、栗钙土、棕钙土，局部地区分布有风沙土， 隐域性土壤主要有草甸土、沼泽土及盐碱土。锡林 郭勒草原是亚欧大陆草原区亚欧中部亚区一个古老 的植物地理区域,草地资源类型多样，有草甸草原、 典型草原、荒漠草原、荒漠、草甸等亚类（图1）。 锡林郭勒草原在类型上具有一定的典型性和完整 性，在空间分布上具有一定的连续性。该区域作为 研究区域研究可燃物的动态空间分布更具有典型 性。 图 1 锡林郭勒盟草原分布图 fig.1 the grassland distribution of xilinguole 2. 数据来源与研究方法 2.1. 地面数据采集设计 地面数据采集（图2），选取了锡林郭勒盟境内 四种具有代表性的草地类型(草甸草原、典型草原、 荒漠草原和沙地植被) 作为研究对象，样地共计51 个，其中包括草甸草原15 个、典型草原21 个、荒漠 草原15 个。 published by atlantis press copyright: the authors 66 author’s names 图 2 样地分布图 fig.2 the sample plot distribution 我国北方可燃物的动态通常每年9月中旬后草原 牧草干枯变黄，进入草原防火季节。此时的草原可 燃物存量处于最高峰，而后由于牲畜的采食践踏和 自然损耗，草原可燃物量越来越少，到翌年5月底牧 草返青，牲畜采食青草而不再啃食枯草时草原可燃 物量降到最低[2]。可燃物在枯草季节有着不段减少 的规律，为了获得可燃物的动态变化率，于在一整 个枯草季节中从2007年10 月份～2008年4 月份每月 中旬进行一次野外实测,测量方法为样方法。草本样 方大小为1m ×1m ,有灌木或高大禾草的草地采用10m ×10m 的样方,每个样地做3 次重复。将干枯牧草齐地 面刈割,带回室内于85 ℃烘箱内烘至恒重，然后称 重。分析地面数据的可燃物干重从而获得不同类型 草地的可燃物在枯草季节的动态变化率。 %100 10 ×= c ci i w w v （1） 其中， i v 为对应月份的可燃物保存率， ci w 为 对应月份的可燃物存量， 10c w 为10月份的枯草季节 初始的可燃物存量。据此，可燃物在枯草季节的动 态变化率向量v 为： [ ]4321121110 ,,,,,, vvvvvvvv = 其中， 10v =1。 2.2. 遥感数据源与遥感监测方法 在遥感数据的选取上，应用modis数据2007年9 月 上 旬 的 250 米 分 辨 的 红 近 红 外 波 段 数 据 计 算 ndvi，然后进行整旬的最大合成，并与草地类型图 叠加，将非草地地区赋值为0予以剔除。 以往对枯草季节枯黄牧草的存量遥感估算模型都 着眼于实时监测，试图应用对枯草敏感的波段数据 建立枯草存量的反演模型，但是由于的枯草的光谱 特征与土壤背景相似程度较高，以及modis传感器 的设计上对地面植被的监测主要着眼于绿色植被的 绿度。因此，可以说modis数据直接实时监测枯草 存量的能力相对于对生长季产草量的监测的能力是 非常有限。在本研究设计上针对枯草存量的定量监 测提出了一个新的思路，应用牧草枯黄前的遥感反 演的产草量数据计算这些牧草的干物质量，然后通 过枯草季节存量的动态变化率来估算枯草季节各月 份可燃物枯草的动态存量。这种方法避开了遥感传 感器对枯黄牧草不敏感的问题。 生长季产草量能够估算枯草季枯草存量是由于生 长季的产草量是枯草季枯草的物质基础来源，生长 季的产草量决定了枯草季枯草存量的基数。在我国 北方9月的牧草开始干枯变黄，10月中旬全部干枯。 如果不考虑秋季放牧的影响，9月初的地面干物质量 与10月中旬枯草季节初期的枯草存量近似相等,根据 这一特点计算出枯草季节初始的可燃物存量。首 先，选用成熟的遥感产草量监测模型计算9月初的地 面产草量，本文采用《中国草原产草量遥感监测》 （徐斌，杨秀春等）所建立的产草量模型[3]，该模 型的建立使用了大量的中国草原地区地面数据，模 型有很好的适普性，实测值与模拟值的相关系数可 达0.77。基计算公式为： ndvi f ew 9841.43585.19= （2） 其中， fw 为单位面积产草量，是牧草的鲜重 量，单位为g/m2。 然后，根据不同草原类型的干鲜比 [4]折算得到 枯草季节初始的可燃物存量： c w w fc = （3） 其中： cw 为单位面积可燃物重量(单位：g/m 2)， fw 为鲜草重量，c为不同草地类型的干鲜比系数。 表格 1 各类型草地干鲜比系数 table 1 types of grassland ratio of withered grass to fresh grass 草地类型 干鲜比系数 温性草甸草原类 1︰3.2 温性草原类 1︰3.0 温性荒漠草原类 1︰2.7 最后，应用地面数据分析所得的各类型草地可燃 物在枯草季节的动态变化率向量v ，估算各月份的 地面可燃物存量，最终实现枯草可燃物动态监测。 icci vww ×= （4） 其中， ciw 为对应i月的单位面积可燃物重量(单 位：g/m2)。 2.3. 火险等级方法 火险等级划分依据火险指数划分，指数是应用 气象数据与地面数据相结合的多因子加权法建立 。 所应用到的影响火险的因子有：可燃物存量、植被 连续程度、积雪覆盖和气象干旱指数（spi）。可燃 物作为草原火发生、发展的物质基础，是孕灾环境 的核心，其地表存量与连续程度决定了孕灾环境中 published by atlantis press copyright: the authors 67 paper title (4 words maximum) 致灾的危险程度。锡林郭勒盟枯草季中期为冬季， 冬季积雪覆盖也是影响火险程度的重要因子。气象 干旱指标可以表征可燃物的干燥程度。这种依据气 象数据与地表数据的多因子指数，立体的表达了地 表可燃物致灾的危险程度。 植被连续程度对火的蔓延有着至关重要的影 响，利用 1km 网格土地利用图选取植被区，以 5km×5km（100 个像元）为窗口计算植被类土地利 用类型所占百分比赋值于窗口的中心像元得到植被 连续程度图值域范围[0，1]。 雪被划分采用modis数据计算归一化差分积 雪指数（ndsi）。其原理为：雪有很强的可见光反 射 和 强 的 短 红 外 吸 收 特 征 ， 因 此 使 用 第 2 通 道 （841～876nm）、第4通道（545～565nm）和第6通 道（1628～1652nm）进行ndsi的计算和积雪判识。 ndsi=（ch4－ch6)／（ch4＋ch6） （5） 一般地当ndsi＞0.4且ch2反射率＞11%、ch4 反射率＞10%时判定为雪。雪被区赋值为1，非雪被 赋值为0。 内蒙古地区地表水补给形式较为单一，以大气 降水为主。降水的充盈与亏损直接影响地表植被的 干燥程度。因此，降水是影响火险程度的重要指 标。气象干旱指标选用标准化降水指数spi，它的基 本原理为，首先假设降水服从gamma分布，然后 用gamma分布概率来描述降水量的变化，最后再 经正态标准化求得spi值。研究表明降水分布比起正 态分布、泊松分布来更接近于gamma分布，spi可 以灵活应用于对不同时间段，不同时间尺度，不同 地区的降水量变化的比较，是具有适普性的干旱监 测指标。其具体算法如下[5，6]： 假设某一站点某一时段降水量为x，则其gamma 分布的概率密度函数为： βα α χ αβ /1 )( 1 )( xexg −− γ = （6） )(αγ 为gamma函数，α 为形状参数， β 为尺度 参数，二者可用最大似然法拟合求得。 ) 3 4 11( 4 1 a a ++=α （7） α β x = （8） n x xa ∑−= )ln()ln( （9） 其中n为降水数据的时间序列长度。在给定的时 段序列长度下降水量为x的累积概率为： dxexdxxgxg x axx β α αβ / 1 00 )( 1 )()( − − ∫∫ γ == （10） 设 β/xt = dtet a xg t at − − ∫γ = 1 0)( 1 )( （11） 这样就构造出一个标准的不完全 gamma 函 数，在 idl 中应用 igamma 函数可直接求得 g(x)。但求得的 g(x)不包括 x 为 0 的情况，现实中 x 可以为 0，这种情况下累积概率应表示为： )()1()( xgqqxh −+= （12） q 为给定时间序列里降水为 0 的次数与时间序 列长度的比值。求 h（x）的正太分布函数： dxexh z∫ ∞ −= 0 2/2 2 1 )( π （13） 近似计算得： 5.0)(0 ≤xh< ) 1 ( 3 3 2 21 2 210 tdtdtd tctcc tspiz +++ ++ −−== （14） 2))(( 1 ln xh t = （15） 0.1)(5.0 ≤xh< ) 1 ( 3 3 2 21 2 210 tdtdtd tctcc tspiz +++ ++ −+== （16） 2))(0.1( 1 ln xh t − = （17） 其中：c0 =2.515 517， c1=0.802 853，c2=0.010 328，d1=1.432 788，d2=0.189 269，d3 =0.001 308。 以研究区内73个气象站点1970年到2009年的降 水数据为基础，建立降水的累积概率分布函数，代 入2007年10至2008年4月的逐旬的前一个月降水数 据，得到逐旬的一个月尺度的spi数据。图3到图5为 各类型草原区逐旬spi值平均值构成的时间序列。从 中可以看出，2007到2008年的枯草季节中，草甸草 原区10月和2月偏干；典型草原区10月、12月、2月 和3月偏干；荒漠草原区12月和2月偏干。 published by atlantis press copyright: the authors 68 author’s names 图 3 草甸草原区 spi 时间序列 fig.3 time serial of spi in meadow steppe region steppe and desert steppe 图 4 典型草原区 spi 时间序列 fig.4 time serial of spi in steppe region 图 5 荒漠草原区 spi 时间序列 fig.5 time serial of spi in desert steppe region 表格 2 标准化降水指数干旱等级划分 table 2 the grade dividing standards of drought base on standard precipitation index 等 级 类 别 spi 值 grade class the value of spi 1 无旱 -0.5<spi 2 轻旱 -1.0<spi≤-0.5 3 中旱 -1.5<spi≤-1.0 4 重旱 -2.0<spi≤-1.5 5 特旱 spi≤-2.0 为了满足空间数据叠加的需要，根据气象数据 的地理坐标位置，将 73 个气象站点 spi 数据用 kriging 法进行空间插值 [7] 。插值过程中选用的 kriging参数为变程为样点间的平均距离的8倍，基值 为1，块金值为0。插值后的数据，与遥感数据统一 在相同的地理坐标系下，空间分辨率与遥感数据相 同为1km网格数据。 图 6 2007 年 10 月 spi 分布图 fig.6 spi index in october, 2007 图 7 2008 年 1 月 spi 分布图 fig.7 spi index in january, 2008 图 8 2008 年 4 月 spi 分布图 fig.8 spi index in april, 2008 图 6图 8 为 空 间 插 值 后 的 枯 草 季 初 期 （ 10 月）、中期（1月）、末期（4月）spi分布图。应用 spi网格图与对应月份可燃物存量数据(w)、雪被数 据(if)和植被连续程度数据(pi)相结合，建立枯草季 可燃物火险指数fi(fire index)公式(18)： fi =w/200×(spi+3)/6×pi×(if = ture)×0.5 + published by atlantis press copyright: the authors 69 paper title (4 words maximum) w/200× (spi-3)/(-6)×pi×(if = false) (18) 表格 3 火险等级划分 table 3 the grade dividing standards of drought base on standard precipitation index 等 级 类 别 fi 值 grade class the value of fi 1 低火险 0.01＜fi 2 中低火险 0.01≤fi<0.05 3 中高火险 0.05≤fi<0.1 4 高火险 0.15≤fi<0.25 5 极高火险 fi≥0.25 火险指数值域分布范围在 0 到 1 之间。表 3 为 火险等级划分表。 3. 结果与分析 3.1. 锡林锡林郭勒草原牧区枯草季可燃物的动态变 化 草甸草原属温性草甸草原类平原丘陵草甸草原亚 类，主要为贝加尔针茅草原群系和羊草草原群系。 根据表2显示枯草季初期（10月份）草甸草原可燃物 存量平均100.20g/m 2 。可燃物变化率下降趋势相对平 缓，11月份至翌年4月份的可燃物变化率枯草季初期 的 86.% 、 20.36% ， 牧 草保 存 率 平 均月 递 减 率 为 13.27%。 表格 4 草甸草原可燃物月动态 table 4 combustible materials dynamic of meadow steppe 调查时间 （年/月） 可燃物存量 均值（g/㎡） 变化率 （%） 2007/10 100.20 100.0 2007/11 86.53 86.4 2007/12 63.93 63.8 2008/01 55.13 55.0 2008/02 40.87 40.8 2008/03 26.87 26.8 2008/04 20.40 20.4 典型草原属温性典型草原类平原丘陵典型草原亚 类，主要群系为大针茅草原群系和克氏针茅草原群 系。根据表3显示可燃物变化率呈明显降低趋势，逐 月递减幅度相差较大。典型草原10月份可燃物存量 平均为73.91 g/m2。典型草原牧草现存量的递减趋势 分为两个阶段，第一阶段为10月份到12月份，其中 11 月 份 和 12 月 份 牧 草 保 存 率 分 别 为 68.39% 和 42.68%，可燃物变化率的降低速度较快，月平均递 减率为28.66%；第二阶段为后四个月，可燃物变化 率从33.95%减到14.76%，四个月变化率的降低幅度 为28.12%，平均月递减率仅为7.03%。 表格 5 典型草原可燃物月动态 table 5 combustible materials dynamic of steppe 调查时间 （年/月） 可燃物存量 均值（g/㎡） 变化率 （%） 2007/10 73.91 100.00 2007/11 50.55 68.39 2007/12 31.55 42.68 2008/01 25.10 33.95 2008/02 18.57 25.12 2008/03 13.50 18.27 2008/04 10.91 14.76 荒漠草原属温性荒漠草原类平原丘陵荒漠草原亚 类，主要群系为戈壁针茅草原群系、多根葱草原群 系和小叶锦鸡儿草原群系。如表4所示，荒漠草原可 燃物变化率与典型草原率变化趋势类似。荒漠草原 10月份的可燃物存量平均为51.85g/m2，荒漠草原可 燃物存量的递减趋势分为两个阶段，第一阶段为10 月份到12月份，两个月的减少幅度为64.69%，平均 月递减率为32.25%；第二阶段为后四个月，变化率 分别为22.55%、16.17%、6.84%和3.32%，四个月降 低幅度为31.99%，平均月递减率为8.00%。 表格 6 荒漠草原可燃物月动态 table 6 combustible materials dynamic of desert steppe 调查时间 （年/月） 可燃物存量 均值（g/㎡） 变化率 （%） 2007/10 51.85 100.00 2007/11 32.38 62.46 2007/12 18.31 35.31 2008/01 11.69 22.55 2008/02 8.38 16.17 2008/03 3.55 6.84 2008/04 1.72 3.32 3.2. 可燃物存量遥感监测结果 空间分布上，锡林郭勒盟草原牧区的可燃物存量 为从东北向西南递减分布，这与本地区的草甸草 原、典型草原和荒漠草原为东北向西南分布相一 致。可燃物存量最高值出现在东乌珠穆沁旗东北的 林缘草甸地区；最低值出现在苏尼特右旗北部以及 二连浩特地区。在行政区划的分布上西乌珠穆沁旗 可燃物均值最高，二连浩特市的均值最低。时间分 布上，草甸草原枯草季在期间的变化幅度最大，可 燃物在枯草季初期平均值为103 g/m2，到了枯草季 末期均值下降为21 g/m2。典型草原枯草季初期的可 燃物在平均值为65 g/m2，末期均值下降为10 g/m2。 荒漠草原可燃物在枯草季初期平均值为26 g/m2，末 期均值下降为1 g/m2。 10月份，由于是在枯草季节的初期，可燃物的存 量处于较高的水平，东、西乌珠穆沁旗的东部地区 可燃存量在150 g/m2以上。锡盟中部的锡林浩特市 位于典型草原区，其可燃存量为65 g/m2左右。西部 的苏尼特左、右旗发育有荒漠草原，其可燃存量多 低于30 g/m2。 published by atlantis press copyright: the authors 70 author’s names 图 9 锡林郭勒盟 2007 年 10 月可燃物存量分布图 fig.9 the amount of combustible materials of xilinguole in october, 2007 图 10 锡林郭勒盟 2007 年 10 月可燃物东北西南向剖面图 fig.10 the northeast to southwest profile of combustible materials distribution of xilinguole in october, 2007 1月份(图3，图4)，可燃物存量荒漠草原下降的最 快，苏尼特左、右旗平均存量为6 g/m2。锡林浩特 市等典型草原地区，其可燃存量为22 g/m2左右。 东、西乌珠穆沁旗草甸草原地区可燃存量均值为63 g/m2。 图 11 锡林郭勒盟 2008 年 1 月可燃物存量分布图 fig.11 the amount of combustible materials of xilinguole in january, 2008 图 12 锡林郭勒盟 2008 年 1 月可燃物东北西南向剖面图 fig.12 the northeast to southwest profile of combustible materials distribution of xilinguole in january, 2008 枯草季末期4月份(图5，图6)，可燃物存量处于最 低值，荒漠草原平均存量仅为1 g/m2。典型草原地 区，其可燃存量为10 g/m2左右。东、西乌珠穆沁旗 草甸草原地区可燃存量均值为21 g/m2。 图 13 锡林郭勒盟 2008 年 4 月可燃物存量分布图 fig.13 the amount of combustible materials of xilinguole in april, 2008 图 14 锡林郭勒盟 2008 年 4 月可燃物东北西南向剖面图 fig.14 the northeast to southwest profile of combustible materials distribution of xilinguole in april, 2008 3.3 火险等级评估结果 叠加气象指标 spi 数据、雪被数据、植被连续 数据以及可燃物存量数据计算利用火险指数。应用 决策树对火险指数数据进行分类生成火险等级图。 published by atlantis press copyright: the authors 71 paper title (4 words maximum) 图 15 2007 年 10 月火险等级图 fig.15 the fire risk in october, 2007 2007 年 10 月枯草季初期，锡林郭勒盟火险程度 多中高火险和高火险。可燃物存量较高，其中以东乌 珠穆沁旗东部的草甸草原存量最高。西乌珠穆沁旗 东南部降水偏少，天气干旱，这两个原因形成东部 极高火险区。西乌珠穆沁旗西部，锡林郭勒市、阿巴嘎 旗东部以及南部五个旗多高火险。阿巴嘎旗西部、苏尼特 左旗以及苏尼特右旗以中高火险为主，中低火险区在其中 零星分布。 图 16 2008 年 1 月火险等级图 fig.16 the fire risk in january, 2008 2008 年 1 月枯草季中期，大部分地区为中低火 险区。极高火险区零星分部于西乌珠穆沁旗东部。 东乌珠穆沁旗东部和西乌珠穆沁旗东部降主要为高 火险区。西乌珠穆沁旗西部、正蓝旗、正镶白旗南部、 多伦县以及太仆寺旗多为中高火险。其余的大部分地区为 中低火险区。低火险区零星分布于苏尼特左旗和苏尼特右 旗。 图 17 2008 年 4 月火险等级图 fig.17 the fire risk in april, 2008 2008 年 4 月枯草季末期，大部分地区为低火险 区。东乌珠穆沁旗东部和西乌珠穆沁旗东部有中高 火险区分布。东乌珠穆沁旗西部、西乌珠穆沁旗中西 部、正蓝旗南部、正镶白旗南部、多伦县以及太仆寺旗多 为中低火险。其余的大部分地区为低火险区。 表格 7 火险等级面积表(单位：km2) table 7 fire risk area statistic(unit: km2) 低 中低 中高 高 极高 10 月 0 19386 78179 75622 25928 11 月 0 56645 74623 64359 3488 12 月 0 27747 86102 69292 15974 1 月 1308 121603 41857 32310 2037 2 月 64 95516 53378 48170 1987 3 月 32662 148973 17069 392 19 4 月 112823 68290 17021 973 8 表格 8 火险等级面积百分比表 table 8 fire risk percent of area statistic 低 中低 中高 高 极高 10 月 0.0% 9.7% 39.3% 38.0% 13.0% 11 月 0.0% 28.4% 37.5% 32.3% 1.8% 12 月 0.0% 13.9% 43.2% 34.8% 8.0% 1 月 0.7% 61.1% 21.0% 16.2% 1.0% 2 月 0.0% 48.0% 26.8% 24.2% 1.0% 3 月 16.4% 74.8% 8.6% 0.2% 0.0% 4 月 56.7% 34.3% 8.5% 0.5% 0.0% 表 7、表 8 为火险等级面积统计表。2007 年 10 月，中高火险和高火险地区占总区域面积最多，分 别为 39.3%和 38.0%。10 月无低火险区分布。11 月 中低、中高和高火险区面积相近，均占总面积的 30%左右；极高火险区面积由上月的 25928 km2，缩 减为 3488 km2，缩减面积占锡林郭勒盟总面积的 11.2%。12 月中高火险区的面积最大占全盟总面积 published by atlantis press copyright: the authors 72 author’s names 的 43.2% ；高为险区的面积居次，面积比例为 34.8%；中低风险区占全盟的面积比为 13.9%；极高 风险区的面积较上月有所增加，从 3488 km2 增长为 15974 km2。2007 年 1 月到 12 月均元低火险区分 布。2008 年 1 月全盟 61.1%的区域为中低火险区； 其次为中高火险和高火险区，分别占总面积比为 21.0%、16.2%；另外低火险区和极高火险区均有少 量分布。2 月中低火险区面积下降为 48%；中高火 险和高火险区的面积均有上升，分别为 26.8%、 24.2%。3 月中低火险区面积最大占全盟总面积的 74.8%；低火险区面积增长为全盟面积的 16.4%。4 月全盟 56.7%的地区为低火险区；中低火险区占总 面积的 34.3%；另有少量的中高火险区和高火险区 分布。 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 低 中低 中高 高 极高 初期 中期 末期 图 18 火险等级变化图 fig.18 the fire risk transformation in january, 2008 图 18 为各火险等级面积比在枯草季不同时期 的变化对比图。由图可见，初期火险等级程中低、 中高到高、极高均匀正态分布状；中期，以中低火 险区为主，中高火险、高火险区次之，低火险区和 极高火险区少量分布。末期以低火险区分布最多， 中低、中高火险区面积依次递减，高火险区和极高 火险区分布极少。 4. 结论与讨论 枯黄牧草作为枯草季节可燃物的主体，实现动态 监测对草原火灾的风险管理有个重要的意义。本文 在地面实测数据的支持下，获得了锡林郭勒盟地区 不同类型草地的可燃物逐月的动态变化率。结果表 明枯草季草原火灾可燃物的存量程逐月递减趋势， 其中草甸草原月平均变化率为13.27%；典型草原与 荒漠草原动态变化程两个阶段，第一阶段为10月份 到12月份，第二阶段为后四个月，典型草原地区的 可燃物动态变化的第一阶段可燃物变化率的降低速 度较快，月平均递减率为28.66%；第二阶段平均月 递减率仅为7.03%，荒漠草原地区的可燃物动态变化 的第一阶段可燃物变化率月平均递减率为32.25%； 第二阶段平均月递减率仅为8%。 锡林郭勒盟草原牧区的可燃物的月动态遥感监测 结果表明在空间分布上，锡林郭勒盟草原牧区的可 燃物存量为从东北向西南递减分布，可燃物存量最 高值出现在东乌珠穆沁旗东北的林缘草甸地区；最 低值出现在苏尼特右旗北部以及二连浩特地区。时 间分布上，可燃物的存量程逐月递减趋势，草甸草 原枯草季在期间的变化幅度最大，典型草原次之。 荒漠草原可燃物变化幅度最小。 火险等级面积比在枯草季不同时期火险程度不 同。初期火险等级分布以中低、中高为主，分别占 全锡盟总面积比为39.3%、38.0%。；中期，以中低 火险区为主，中高火险、高火险区次之，面积比分 别为61.1%、21.0%和16.2%；末期以低火险区分布 最多，占全盟总面积56.7%，中低火险区面积次，所 占之面积比34.3%。 应用成熟的产草量遥感反演模型反演生长季后 期的产草量，结合植被的干湿比例数据可以获得枯 草季节初始的可燃物存量，再根据可燃物在枯草季 节存量的动态变化率，可以估算出枯草季节可燃物 的动态变化，这种方法绕过了常用遥感数据对枯黄 牧草监测能力有限的瓶颈，实现了原草区枯草季节 可燃物存量的大面积、快速的动态监测。这一方法 反演的精度主要取决于生长季产草量遥感模型的精 度，在实际应用中可以根据不同的研究区域选取更 适合当地的产草量模型，从而提高监测的精度。依 据气象数据与多种地表数据的建立的多因子火险指 数，立体的表达了地表可燃物致灾的危险程度。如 果进一步使用长时间序列记录的草原火灾数据，计 算火灾发生频度，然后与火险指数相进行统计分析 获具有明确统计学意义的火险等级划分方案，则能 使本文的监测方法实用性得到进一步提高。 参考文献 1. 裴浩，李云鹏，范一大，利用气象卫星 noaa ／ avhrr 资料监测温带草原枯草季节牧草现存量的初 步研究，中国草地，06：11-14，1995 2. 色音巴图，草原可燃物动态研究，中国草地，05： 48-52，2009 3. 徐斌，杨秀春， 覃志豪，刘海启，缪建明，中国草 原产草量遥感监测，生态学报，02：54-60，2007 4. 陈世荣，草原火灾遥感监测与预警方法研究，中国博 士论文全文数据库，2006 年 5. seiler r. a., hayes m., bressan l. using the standardized precipitation index for flood risk monitoring [j]. international journal of climatology, 2002, 22(11): 1365−1376 6. gb/t 20481-2006 气象干旱等级 7. 闫殿武.idl 可视化工具入门与提高[m].北京.机械工业 出版社.2003 published by atlantis press copyright: the authors 73 paper title (4 words maximum) 8. 宋连春，邓振偏，蓝安详.干旱 [m].北京:气象出版 社，2003. 9. 史培军；三论灾害研究的理论与实践田；自然灾害学 报，2002，11(3):1-9. 10. 张继权，李宁；主要气象灾害风险评价与管理的数量 化方法及其应用；北京:北京师范大学出版社，2007. 11. 张继权，冈田宪夫，多多纳裕一；综合自然灾害风险 管，城市与减灾，2005，2 12. 黄崇福；自然灾害风险分析理论与实践[m]北京:科学 出版社，2005. 1-7. 13. 黄崇福,刘新立,周国贤,等.以历史灾情资料为依据的农 业自然灾害风险评估方法[j].自然灾害学报, 1998, 7(2):1-8 14. 牛建明．气候变化对内蒙古草原分布和生产力影响的 预测研究．草地学报，2001，9(4)：277～282 published by atlantis press copyright: the authors 74 基于主体的渔船回港避风疏散仿真 received 4 july 2015 accepted 19 august 2015 agent-based simulation of fish boats evacuation hanping zhao 1,2 , huiyan ding 1,2 , han wang 1,2 1 key laboratory of environment change and natural disaster, ministry of education, beijing 100875, china. 2 academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china abstract emergency management agencies will organize people to evacuee when disasters are forecasted. however, contradictions between individual behavior and group objective appear frequently during large-scale evacuations, the problem is obviously in the process of fisher boats evacuation. so, this paper set up fisher boats evacuation simulation model considering individual desire of back harbor. in the agent-based model, typhoon and harbor are inactive agents, government and fisher boats are active agents. fisher boats make decisions whether back to harbor and which harbor to be choose according of typhoon forecast information and government instruction. the model uses the harbors of zhejiang province as a case. key words: typhoon; fisher boat evacuation; multi-agent simulation 基于主体的渔船回港避风疏散仿真 赵晗萍 1,2,丁慧彦 1,2,王晗 1,2 1.北京师范大学环境演变与自然灾害教育部重点实验室,中国北京，100875； 2.北京师范大学减灾与应急管理研究院,中国北京，100875 摘要：灾害应急管理中政府部门会根据灾害预估采取各种转移安置的措施，但在大规模转移安置过程中对群众的安排组织工 作非常繁重，经常会出现个体行为与集体组织出现矛盾的情况，这种现象在台风来临前组织渔船回港避风中也非常明显。因 此在考虑渔船回港的意愿基础上设计，设计渔船回港避风仿真模型。模型中台风与渔港作为非活动主体，政府和渔船作为活 动主体，渔船根据台风灾害信息、政府指令及渔港状态等要素做出是否回港和选择港口的决策。选取浙江省主要渔港作为案 例背景进行仿真实验。 关键词：台风；渔船疏散；多主体仿真 1. 引言 台风灾害是中国东南沿海地区遭受的最频繁、最 严重的自然灾害之一，平均每年有 7.2 场台风登陆中 国[1]。台风灾害破坏力极强，台风中心风力可达 12 级 甚至更高，台风大风往往伴随着暴雨、风暴潮等，给 沿海居民的生产生活造成十分严重的影响。2011 年在 西北太平洋和南海共有 20 个台风生成，其中 7 个在我 国沿海地区登陆，共造成 23 人死亡，7 人失踪，直接 经济损失 192.56 亿元。2000 年以来，台风灾害在中国 造成的年平均直接经济损失达到 318.4 亿元[2]。 在台风灾害可能导致的各种损失和破坏中，海上 作业的船只受到的影响最为直接。在中国东南沿海地 区，海洋渔业是主导产业之一，渔船数量众多，目前， 中国已有海洋渔船 31.61 万艘。渔船以小型、木质渔 船为主，钢制、大型渔船较少，旧渔船多，新造渔船 少，船龄普遍偏高，10 年以上船龄的渔船占 60.8% [3] ， 渔船本身安全性较低。因此，在台风灾害应急响应中， journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 152-160 published by atlantis press copyright: the authors 152 及时有效的渔船回港避风是一个不容忽视的终于环节。 在应急疏散中涉及人员众多，难以在短时间内让 大规模群众完全听从指挥，极易出现个体按照自我意 愿自主行动的现象。因此在应急中应充分考虑个体行 动的意愿和偏好， helbing 等人（2000） [4] 在《nature》 上发表了研究成果，通过人群在恐慌状态下的行为构 建了恐慌状态下的人群逃散模型。hao-che wu 等 （2012） [5] 在美国卡特里娜和丽塔飓风之后以调查问 卷的形式分析疏散时人们的路径选择行为。胡红等 （2007） [6] 根据将驾驶员对突发事件的心理行为反应 调查，对 gipps 跟驰模型进行了改进，建立了车速与 安全车距的关系模型。huizhaotu 等（2010） [7] 为研究 疏散人员的驾驶行为对整个疏散过程的作用，采用 paramics 微观仿真软件对疏散人员的驾驶行为进行不 同情景的模拟仿真，研究车辆加速度、最大行驶速度、 平均距离、最小间距等因素对仿真结果的影响。stern 等人（1989） [8] 利用行为模拟研究了不同类型家庭的 行动（外出、在家或睡觉）为整体预警信息传播的影 响。uchida（2012） [9] 研究了在滑坡与预警中人群疏散 概率变化的现象。 考虑个体行为因素后，系统建模则多数采用基于 多主体仿真（agent-based simulation：abs）手段模拟 人群疏散行为。abs 是一种微观仿真技术，仿真模型 中主体具有自主性和目标性，可以与环境和其他主体 进行交互（railsback 等人，2006） [10] 。abs 技术在应 急响应建模中已经有了广泛的应用，chen 和 zhan （2008） [11] 研究了城市交通疏散模型；banerjee 等 （2009） [12] 针对大规模拥堵问题建立仿真模型； magesh 等（2012） [13] 提出基于 abs 的预警信息扩散 下疏散仿真的框架，其中利用概率原理模拟了疏散中 警示邻居的行为。hui 等人（2008） [14] 利用 abs 模型 研究个体行为如何影响预警信息如何在社区间传播。 对于渔船疏散问题，在台风来临时候管理渔船回 港避风时候更加明显，渔民倾向于在海上尽量多停留 可以多打些鱼，但同时面临的被台风打击的风险也会 随之增加。渔民在海上相对分散，很难进行有效组织 管理，因此本文针对渔民疏散问题进行研究，在对渔 民回港避风意愿调研基础上，设计了基于主体的渔船 回港避风仿真模型。 2. 渔船回港基于主体自主决策仿真建模 2.1 渔船回港仿真结构 在台风灾害渔船回港避风的问题中，有政府、渔 船、台风和港口 4 个要素，它们组成了一个互动的灾 害应急响应系统。仿真的结构与给要素之间的关系， 如图 1 所示： 图 1. 基于主体自主决策仿真结构 台风 港口 渔船1 渔船2 渔船m … 编号 家乡港口 风险偏好 政府决策： 催促回港 位置 类型 回港状态 回港编号 决策： 是否回港 决策： 港口选择 容许回港时间 海面风力 中心风力 编号 位置 等级 回 港 状 态 交 互 渔船 延迟回港时间 港口选择受到催促次数 渔船1 渔船2 渔船m„ 输出 published by atlantis press copyright: the authors 153 政府可以获得台风的预报和预警信息，通过对渔 船和港口情况的监测，监督渔船回港避风，做出“催促 回港”的决策，在系统中只有一个。渔船同样可以得到 台风的预警信息，并能感知自身所处位置气象信息， 根据对于自身台风风险状况的认知，和政府的相关指 令，做出“是否回港”及“港口选择”的决策，在系统中 有多个。政府获得渔船的位置信息和回港状况，渔船 则对政府“督促回港”的指令的反应。 台风具有风级、路径、等属性，港口具有等级、 容量等属性，它们的状态都会随时发生变化，它们是 仿真的外部环境。渔船和政府会根据它们的变化而做 出决策，而渔船回港状态的改变，也会对港口产生影 响，引起其容量属性的变化。 2.2 渔船回港仿真主体要素设计 2.2.1 外部环境主体 外部环境是非活动的主体，人在短时间内无法干 预和改变的主体，但外部环境状态会影响活动主体的 决策。台风和港口具体属性设置如表 1 所示。大风预 报图进行模拟见图 2。 2.2.2 活动主体行为定义 活动主体具有自主决策能力，它可以感知外界信 息并做出决策。对于不同的主体，可以感知的信息和 决策内容如表 2 所示。 2.2.3 渔船回港决策行为模式 2011 年 6 月 14 日-17 日，在浙江省岱山县就渔船 图 2. 中央气象台台风大风预报图（来源：中国天气网[68]） 表 1. 台风和港口属性列表 主体名称 属性 台风 容许回港时间 t：根据台风预报信息，估计台风到达的时间，规定所有应急活动需在该时间 内完成 台风中心风力 wc：预报时刻台风中心的风力等级 海面任意点风力预报 wi：气象部门对于海面大风情况的预报。在这里，使用气象部门发布的 大风预报图进行模拟（图 2）。对于预报时段内的任意时刻，海面上任意点的风力值服从正 态分布。在 10 级风圈内时：均值为(10+wc)/2，标准差满足 6= wc-10；在 7 级风圈内时：均 值为 8.5，标准差为 0.5；在 7 级风圈外时：均值为 3.5，标准差为 1.17 港口 港口编号：可提供避风的港口编号 港口等级 l：对于不同的港口等级，具有不同的可容纳渔船类型集合{hsl} 港口位置：由坐标 x 和 y 来表示港口的位置 published by atlantis press copyright: the authors 154 回港避风中渔民的决策行为因素进行了调研。综合调 研中从乡镇管理渔业管理部门和渔民两方面了解的结 果，可以看到，目前渔船在回港避风中，政府管理部 门更多承担的是“监督”的角色，即保证在渔船在一定 的时间内进入港口。但在回港的具体行动上，是由渔 船自主决策的，主要有如下方面： 1. 回港时机：渔船根据的台风风级、走向等气象信 息，回港需要的时间，政府监督情况及经济收益 等因素，判断开始回港的具体时刻。 2. 港口选择：根据回港需要的时间，判断是“回家” 避风，还是“就近”避风。 （1）是否回港 渔船当前时刻是否回港的决策，主要基于渔船根 据风力对自身风险情况的判断，及自身风险偏好 pri 情况来决定，是渔船基于风险评估的应急响应决策。 根据调研情况，将渔民风险认知的因素简单设定为台 风中心风力 wc 和自身所处海面的风力 wi。渔船根据 台风中心风力 wc 判断此次台风的破坏程度，根据自 身所处海面的风力 wi 判断自己当前状况的危险程度。 渔民根据台风风力的情况，而做出的回港决策处 理成一种模糊逻辑关系，即：对于不同的风险偏好类 型的渔民，在不同的台风情况下，以一定的概率会决 定回港。在不同情况下，决定回港的概率如表 3 和 4 所示。表中的参数均根据调研情况假设得到。 表 3 给出了在不同台风中心风力 wc 情况下，判 断自己的理论回港时间 ti 提前容许回港时间 t 的小时 数 t’i，在不同的提前程度 t’下，决定回港的概率 pbi。 如表中带下划线的数字表示：在台风中心风力为 9 级 时，若风险厌恶型渔船的需要回港时间在 24 小时到 12 小时之间，有 0.1 的概率选择回港。 表 2. 政府和渔船的属性、感知信息和决策内容 主体名称 属性 内容 政府 感知 信息 台风：容许回港时间 t（台风预计登陆的时间）； 港口：港口的位置（坐标 x 和 y）； 渔船：渔船的位置（坐标 x 和 y），回港状态 ibi，回港编号 shi，渔船类型 k。 决策 内容 政府主要通过与渔船的交互，做出是否催促该渔船回港的决策。政府根据渔船回港需 要的时间，当渔船的回港时间 ti 超过剩余容许回港时间 tr 的一半时，决定对该渔船“催 促回港” 渔船 属性 渔船编号：海上一系列渔船的编号； 渔船位置：由坐标 x 和 y 来表示渔船当前的位置； 渔船类型 k：渔船具有大、中、小三种类型。每种类型渔船对应一种代表船型，有相 应的渔船长、宽和设计航速； 家乡港口 hh：渔船长期作业，家庭所在的港口； 风险偏好 pr：渔船对于风险的喜好程度，具有偏好、中性、厌恶三种。不同风险偏好 的渔船会有不同的决策方式，越偏好风险的渔船，在台风中越倾向于晚回港； 回港状态 ib：渔船是否已经启动回港； 回港编号 sh：渔船前往避风的港口编号。 感知 信息 台风：容许回港时间 t（预计登陆的时间），台风中心风力 wc，自身所处海面的风力 w； 港口：港口的位置（坐标 x 和 y），可容纳渔船类型{hsl}； 政府：政府“催促回港”的命令； 渔船：距自身一定半径 r 内渔船的是否回港决策。 决策 内容 渔船根据外部环境和政府信息，做出是否回港和港口选择的决策。具体决策准则在 1.2.3.小节阐述 published by atlantis press copyright: the authors 155 表 4 给出了渔船根据当前所处位置的台风风力 wi 决定回港的概率 pbi。如表中带下画线的数字表 示，当渔船所处位置的风力为 8 级时，风险厌恶型 渔船选择回港的概率为 0.1。 除对环境的响应之外，渔船会和政府进行交互。 渔船“是否回港”决策会受到政府 “催促”回港次数 的影响，其回港的概率 pbi 设定如表 5 所示。如表 中带下划线的数字表示，当政府催促回港 1 次时， 风险厌恶型渔船，选择回港的概率为 0.5。 此外渔船之间也会进行交互，渔船会受到与自 身距离一定范围内的渔船的决策的影响。在影响半 径 r 内，决定回港渔船的比例，会影响不同风险偏 好的渔船做出回港的决策，回港概率 pbi 设定如表 6 所示。如表中带下划线的数字表明，对于风险厌恶 型渔船，当影响半径内有 0~10%的渔船决定回港时， 它有 0.5 的概率可能选择回港。 表 5. 根据政府催促回港次数渔船回港概率表 1 次 2 次 3 次 4 次 5 次 风险厌恶 0.5 1 1 1 1 风险中性 0.2 0.5 0.8 1 1 风险偏好 0 0.2 0.5 0.8 1 表 6. 根据影响半径内回港渔船比例 渔船回港概率表 0~10% 10%~30% 30%~50% 50%~80% 80%~100% 风险厌恶 0.1 0.5 0.8 1 1 风险中性 0 0.1 0.5 0.8 1 风险偏好 0 0 0 0.1 0.3 为使逻辑关系更容易处理，认为渔船基于上述 四种因素的判断是独立的，无论哪种因素使得渔船 做出“回港”的决定，渔船最终都会回港。 （2）港口选择 对于所有的渔船，首选港口均为家乡港口 hh， 渔船根据估计得到的提前回到港口的时间，以一定 的概率选择就近避风，决策概率 pni 如表 7。如表中 带下划线的数字 0.8，表示根据预计回港时间，仅能 使渔船在容许回港时间前 2 小时回到港口时，风险 偏好型渔船选择就近避风的概率为 0.8。 表 3. 根据台风中心风力 渔船回港概率表 中心风力 <10 级 10 级 11 级 12 级 提前时间 24 12 8 4 24 12 8 4 24 12 8 4 24 12 8 4 风险厌恶 0 0 0 0 0.1 0.3 0.5 0.8 0.5 0.8 1 1 0.8 1 1 1 风险中性 0 0 0 0 0 0.1 0.2 0.3 0.3 0.5 0.8 1 0.5 0.8 1 1 风险偏好 0 0 0 0 0 0 0 0 0 0.1 0.2 0.3 0.1 0.3 0.5 0.8 中心风力 13 级 14 级 15 级 >15 级 提前时间 24 12 8 4 24 12 8 4 24 12 8 4 24 12 8 4 风险厌恶 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 风险中性 0.8 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 风险偏好 0.3 0.5 0.8 1 0.5 0.8 1 1 0.8 1 1 1 1 1 1 1 表 4. 根据当前位置风力渔船回港概率表 <8 级 8 级 9 级 10 级 11 级 >11 级 风险厌恶 0 0.1 0.5 1 1 1 风险中性 0 0 0.1 0.8 1 1 风险偏好 0 0 0 0.5 0.8 1 published by atlantis press copyright: the authors 156 表 7. 不同提前回港时间下渔船决定就近避风的概率 <0 小时 0-2 2-4 4-8 8-12 12-16 16-20 >20 小 时 风险厌恶 1 1 1 0.8 0.5 0.3 0.1 0 风险中性 1 1 0.8 0.5 0.3 0.1 0 0 风险偏好 1 0.8 0.5 0.3 0.1 0 0 0 2.3 渔船回港个体自主决策仿真模型 当气象部门发布台风预报信息时，确定了一个 渔船的容许回港时间 t，政府部门的渔船回港避风 应急行动启动。根据当前时刻的台风中心实际风力、 台风大风预报、渔船位置状态，运行仿真模型，以 考察渔船回港时机和选择港口。 仿真模型的时间间隔为 1 小时，当所有渔船的 回港状态 ibi 变成 1 后，完成 1 次仿真，记录渔船回 港编号 shi 和相应的渔船的回港时间 ti。回港时间 为海上运行时间 ti 和延迟回港时间 twi 的总和。 在每一时刻，对于每条回港状态为 0 的渔船， 进行一次是否回港决策。渔船依次按照政府催促回 港次数、影响半径内回港渔船比例、台风中心风力 和当前位置风力判断是否需要回港。一旦上述任意 条件得到满足，渔船则做出回港的决策，即渔船的 回港状态 ibi 变更为 1，并跳出是否回港判断。根据 渔船回到家乡港口的航行时间和当前时刻的延迟时 间，按照港口选择的概率表，判断渔船是选择回到 家乡港口还是就近港口，记录渔船回港编号 shi，和 回港时间 ti。 在时刻末期，政府部门对所有渔船依次判断， 是否需要在下一时刻对其进行“催促回港”。对于所 有未回港的渔船，计算其影响半径内回港渔船的比 例。 将整个仿真模型运行多次，评估每一次仿真得 到回港方案的风险，对所有仿真结果取平均值，即 可评估渔船在个体自主回港状态下的风险情况。 3. 浙江省台风渔船回港避风个体自主决策仿 真案例 3.1 数据处理与说明 根据《浙江省沿海标准渔港布局与建设规划》 [15] ，截止到 2006 年底，浙江省年共有海洋渔船 34506 艘，各类渔港 208 座。计划在十年内，将海洋机动 渔船数控制在 34500 艘左右；到 2012 年底，新增 有效避风港池面积 2000 万平方米，由 2006 年的 1650 万平方米增加到 3650 万平方米。 考虑到模型规模和电脑运行速度，为说明问题， 并不失代表性，本示例假设有 3500 条渔船，前往 20 个不同港口避风。 3.1.1 港口数据 根据《浙江省沿海标准渔港近期建设项目布局 图》[15]，浙江省在 2012 年重点建设一、二、三级 渔港 69 座，图中还显示了国家级中心渔港和等级以 下渔港，所有共计 81 座。在这 81 座港口中，进行 分层随机抽样，在中心渔港、一级渔港、二级渔港、 三级渔港和等级以下渔港中，各选取 2，4，5，3， 6 个。 港口具有地理坐标、等级、避风面积、可容纳 渔船等级等属性，基本参数如表 8 所示：  港口的地理坐标通过数字化布局图时，使用的 投影坐标得到。  港口的等级由布局图中的专题图符号可以看 出。  渔港的避风水域面积通过一定的合理假设产 生随机数模拟得到： 1）基本保证所有 208 座渔港总避风面积在 3650 万平方米左右。 2）上一等级渔港避风面积是下一级的两倍左 右。 3）同等级渔港避风面积可以在一定范围内变 化。  港口的可容纳渔船等级根据渔港等级假设得 出。 3.1.2 渔船数据 根据 2010 年 9 月在江苏南通和 2011 年 6 月在 浙江岱山的关于渔船管理的调研，目前渔船上的信 息系统已基本配备完善，在台风来临时，海上需避 风渔船的地理位置及其他船只属性，可以由管理部 门通过信息管理系统实时获取。 在本示例中，待回港渔船采取在 arcgis 中生 成随机点的方式产生。假设浙江附近海域有 3500 只渔船需要回港避风，其中有大型渔船 20 只，在距 海岸线 200-300km 的范围内作业，中型渔船 1480 只，距海岸线 130-200km，小型渔船 2000 只，距海 岸线 80-150km。根据《标准渔港设计规范》中的船 型尺度表，大型渔船代表船型选择“开创号”，中型 渔船代表船型选择 hg818 型渔船，小型渔船选择 10kw 木质渔船。相关参数如表 9。 published by atlantis press copyright: the authors 157 表 8. 渔港参数表 编号 名称 坐标 x（m） 坐标 y（m） 等级 面积（m2） 可容纳船型 大 中 小 0 沥港渔港 1597857 3315665 二级渔港 172033.3 0 1 1 1 舟山中心渔港 1628658 3308746 中心渔港 492321.6 1 1 1 2 湖头渡渔港 1596071 3264441 等级以下渔港 26711.21 0 0 1 3 鸿屿渔港 1583684 3254174 等级以下渔港 35314.03 0 0 1 4 桐照渔港 1579555 3251607 一级渔港 321154.8 1 1 1 5 国庆渔港 1570627 3245581 二级渔港 155020.2 0 1 1 6 圆山渔港 1582903 3245135 三级渔港 72745.2 0 1 1 7 白歧渔港 1583684 3212659 等级以下渔港 47123.76 0 0 1 8 胡陈港渔港 1594174 3214222 二级渔港 162601.8 0 1 1 9 鹤浦渔港 1620846 3215003 一级渔港 374460 1 1 1 10 海游渔港 1579331 3202504 等级以下渔港 45335.96 0 0 1 11 路桥金清渔港 1600758 3139785 一级渔港 311549.2 1 1 1 12 坎门中心渔港 1578885 3085772 中心渔港 665499.4 1 1 1 13 杏湾渔港 1554445 3090682 二级渔港 154217.4 0 1 1 14 元觉花岗渔港 1570404 3064456 等级以下渔港 22716.84 0 0 1 15 兰田渔港 1539044 3066800 三级渔港 69297.2 0 1 1 16 西湾渔港 1527996 3028298 三级渔港 74508.33 0 1 1 17 炎亭渔港 1531121 3006202 一级渔港 328513.5 1 1 1 18 石砰渔港 1530898 3003077 二级渔港 176833.5 0 1 1 19 中墩渔港 1521188 2994595 等级以下渔港 36340.99 0 0 1 表 9. 船型参数表（来源：sc/t 9010-2000 标准渔港设计 规范） 渔船类型 代表船型 全长 （m） 型宽 （m） 船速 （km/h） 大 开创号 92 15 15 中 hg818 型 渔船 31.88 6 10 小 10kw 木 质渔船 7.59 3 5 渔船“家乡港口”属性，参考港口可接纳渔船类 型和港口面积因素，按照渔船编号依次将大型渔船 平均分配到 0 级和 1 级港口，将中型渔船平均分配 到 0 级-3 级港口，将小型渔船平均分配到 2 级-4 级 港口。对于“风险偏好”属性，设定每条渔船属于风 险偏好型、风险中性型和风险厌恶型的概率为：0.1， 0.7，0.2。通过产生随机数，得到每条渔船的风险偏 好值。 规定回港时间为 48 小时，影响半径为：50 千 米，使用 2011 年 8 月 5 日的“台风梅花”的预警信息， 台风梅花中心风力 14 级，大风预报图为图 2 所示。 需做说明的是，图 1 为 24 小时的大风预报数据，理 论上应该在 8 月 5 日同时获得一张 24-48 小时大风 预报图。由于数据获取限制，认为 48 小时内，大风 的预报均为图 2。 使用 c#语言编写程序，运行 5.3 中的仿真模型 200 次，记录的渔船延迟回港时间、政府催促次数 和回港方案。 3.2 模型运行结果 3.2.1 延迟回港结果 在 200 次仿真过程中，所有的 3500 只渔船，共 有 652 只曾发生过延迟回港的现象，占总渔船数量 的 18.63%，出现的最大延迟时间为 44 号渔船的 14 小时，计算每只渔船平均的延迟回港时间，该渔船 也具有最大的平均延迟回港时间，为 3.785 小时。 在 200 次仿真中，渔船延迟回港结果的统计表 10 所示： 表 10. 200 次仿真延迟回港结果统计 延迟回港 船只数量 最大延迟 回港时间 所有渔船累积 延迟回港时间 有延迟渔船 200 次仿真平均延迟时间 最大值 527 14 677 3.785 最小值 482 4 577 0.005 平均值 506 8.11 617.125 0.9465 published by atlantis press copyright: the authors 158 考察渔船在仿真过程中回港的情况，计算随仿 真时间的推移，回港渔船数量的变化。计算在 200 次仿真中每一时刻已回港渔船的平均值，如图 3 所 示（大型渔船数量见副坐标轴）。 总体来看，本次回港避风中，“延迟回港”的现 象不是很严重，这主要是因为本次台风中心风力已 达 14 级，多数的渔船都认为这是比较严重的状况， 希望自己能够提前到港的时间较宽裕。 3.2.2 催促回港结果 在 200 次仿真过程中，所有的 3500 只渔船，共 有 156 只渔船受到过政府的“催促回港”，占总渔船 数量的 4.46%，出现的最大催促回港次数为 2670 号 渔船的 5 次。在 200 次仿真中，渔船受到催促回港 结果的统计如表 11 所示： 图 3. 仿真过程渔船回港情况 表 11. 200 次仿真催促回港结果统计 催促回港 船只数量 最大 催促回港次数 所有渔船累积 催促回港次数 有催促渔船 200 次仿真平均催 促次数 最大值 115 5 97 0.925 最小值 88 2 53 0.005 平均值 101.64 3.54 76.645 0.4913 3.2.3 回港方案结果 在本模型中，渔船可能选择“回家避风”或“就近 避风”。从 200 次仿真的结果来看，所有的 3500 只 渔船中，有 1022 只渔船在所有的 200 次仿真中，均 选择了“就近避风”，说明它们在应急响应启动时， 所处的位置距家乡港口较远，在规定时间内很难安 全到达；而有 1286 只渔船在 200 次仿真中，均选择 了“回家避风”，它们所处的位置恰好有充足的时间 安全回港。 4. 结论和讨论 本文从应急响应个体的自主决策行为出发，设 计了渔船、政府、台风和港口的仿真模型，基于调 研了解的基本事实，建立了渔船回港的决策规则。 在仿真中，通过对不同风险偏好渔船的“延迟回港” 决策和“港口选择”决策的逻辑规则设计，及政府 “催促回港”决策设计，模拟了渔船在回港避风中 的延迟回港情况和港口选择情况。应急个体在自然 灾害中的自主应急行为和对管理部门应急安排的自 主执行，将影响自然灾害应急响应的完成状况，因 此，研究自然灾害应急响应个体的自主决策行为在 应急管理中有十分重要的意义。 本文通过仿真的方式模拟渔船在回港避风中的 自主决策行为，在决策规则建立时，仅根据调研的 基本事实进行了基于概率的逻辑设计，这使得模型 对于渔船自主回港决策行为的模拟只能反映基本情 况。在下一步的研究中，应该对应急个体的风险认 知、风险识别和风险判断进行更为细致的研究，得 到更为准确而符合实际意义的逻辑关系与参数，建 立更为符合实际的应急个体在决策行为规则。 本文受国家自然科学基金（41471424）；中央高 校基本科研业务费专项资金资助 published by atlantis press copyright: the authors 159 参考文献 [1] 周俊华. 西北太平洋热带气旋风险分析. 自然灾害 学报, 2004, 13(3): 146-151. zhou j h. risk analysis of tropical cyclones in the northwestern pacific. journal of natural disasters, 2004, 13(3): 146-151. [2] 中国气象局. 2011 年我国灾害性天气特点[eb/ol]. http://www.cma.gov.cn/2011zwxx/2011zyjgl/201203/t 20120305_163510.html characteristics of china's disastrous weather in 2011[eb/ol]. china meteorological administration. http://www.cma.gov.cn/2011zwxx/2011zyjgl/201203/t 20120305_163510.html [3] 冯 华 . 我 国 渔 船 数 量 世 界 第 一 . 人 民 日 报 , 2012-3-26(2). http://paper.people.com.cn/rmrb/html/2012-03/26/nw. d110000renmrb_20120326_8-02.htm feng h. china fishing boats are the first in the world。 people's daily, 2012-3-26(2) http://paper.people.com.cn/rmrb/html/2012-03/26/nw. d110000renmrb_20120326_8-02.htm [4] d. helbing, i. farkas, t. vicsek simulating dynamical features of escape panic. nature, 407 (2000), pp. 487–490 [5] hao-che wu, michael k.lindell, carla s.prater. logistics of hurricane evacuation in hurricanes katrina and rita. transportation research part f, 2012, 15(4):445–461. [6] 胡红, 刘小明, 杨孝宽. 基于最小安全间距的应急 交通疏散车辆跟驰模型. 北京工业大学学报, 2007, 33(10): 1070-1074. hu h, liu x m, yang x k. a car-following model of emergency evacuation based on minimum safety distance. journal of beijing university of technology, 2007, 33(10): 1070-1074. [7] tu h z, tamminga guus, drolenga hans, et al. evacuation plan of the city of almere: simulating the impact of driving behavior on evacuation clearance time. proceedings of the 1st international conference on evacuation modeling and management, netherlands, 2010: 67-75. [8] stern, e., sinuany-stern, z., 1989. a behavioural-based simulation model for urban evacuation. papers of the regional science association 66, 87–103. [9] uchida, k., a model evaluating effect of disaster warning issuance conditions on “cry wolf syndrome” in the case of a landslide. european journal of operational research 218 (2012) 530–537 [10] railsback, s.f., lytinen, s.l., jackson, s.k., 2006. agent-based simulation platforms: review and development recommendations. simulation 82 (9), 609–623. [11] chen, x., zhan, b.f., 2008. agent-based modeling and simulation of urban evacuation: relative effectiveness of simultaneous and staged evacuation strategies. journal of the operational research society 59, 25–33. [12] banerjee, b., abukmail, a., kraemer, l., 2009. layered intelligence for agent-based crowd simulation. simulation 85, 621–633. doi:10.1177/0037549709340659. [13] magesh nagarajan, duncan shaw, pavel albores disseminating a warning message to evacuate: a simulation study of the behaviour of neighbours. european journal of operational research 220 (2012) 810–819 [14] hui, c., goldberg, m., magdon-ismail, m., wallace, w.a., 2008. micro-simulation of diffusion of warnings. in: fiedrich, f., van de walle, b. (eds.), proceedings of the 5th international iscram conference, washington, dc, usa. [15] 浙江省发展和改革委员会, 浙江省海洋与渔业局. 浙江省沿海 标准渔港布 局与建设规 划 [eb/ol]. http://www.zjoaf.gov.cn/zfxxgk/ghjh/zxgh/2008/09/12/ ww092008091100008.shtml zhejiang development and reform commission, zhejiang ocean and fishery bureau. zhejiang province coastal port layout planning and construction standards[eb/ol]. http://www.zjoaf.gov.cn/zfxxgk/ghjh/zxgh/2008/09/12/ ww092008091100008.shtml published by atlantis press copyright: the authors 160 http://www.cma.gov.cn/2011zwxx/2011zyjgl/201203/t20120305_163510.html http://www.cma.gov.cn/2011zwxx/2011zyjgl/201203/t20120305_163510.html 中小商业银行票据业务风险与防范 bill business risk and prevention of small and medium commercial bank yanling hu civil, commercial and economic law school, china university of political science and law beijing, 100088, china e-mail: lxfhyl@163.com abstract with the increasingly fierce competition in china's banking industry, the traditional deposit and loan business has been difficult to meet the needs of the development of banking business itself. bill business as a new intermediary business has been a new point of profit growth for small and medium commercial banks (smcb). however, with the development of bill business, the relevant problems have been exposed. this paper proposes corresponding countermeasures on these problems and focuses on the risks should be noticed in doing bill business and the preventive measures to smbcs. keywords: small and medium commercial bank; bill business; risk; prevention 1. introduction in the recent years, the bill business of smcb develops rapidly, which has been playing a positive role in nurturing relation between bank and enterprise, enhancing service function, optimizing asset structure, decentralizing business risk, and to a certain extent, broadens financing channels, supports the development of smcb (chun, 2014). however, oppositely, there is idea bias in some smcbs, they are obsessed with market share but weak in risk awareness. there is no strict internal control and management on bill business and there are many irregularities and risk issues. if these problems can not be solved timely, they will affect the healthy development of bill business market and the sound operation of smbc. 2. the main risks in bill business 2.1. credit risk in the absence or deficiency of margin, issuing or discounting acceptance bills leads to much bank advance and thereby results in credit exposure risk. mainly as follows: first, do not charge the margin as specified proportion or the amount is low. second, shortfall guarantee to the shortfall part of margin or warranty is invalid, for example: the amount of guarantee contract is less than the shortfall part of margin, the guarantee company is lack of guarantee capacity, the pledge is not in comply with relevant provisions, the procedures of pledge is not timely or normally handled, guaranteed by affiliates or mutual guarantee, the guarantee contract is without agreement result form the board, etc (xue and xiao, 2014). third, use their own loan as margin to issue acceptance bill, or transferring loan to fixed deposit as journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 233-237 published by atlantis press copyright: the authors 233 willieb typewritten text received 22 september 2014 willieb typewritten text accepted 19 november 2014 willieb typewritten text pledge to apply for acceptance bill. these practices of the bank actually transfer the acceptance risk to credit risk, but it finally do not ascertain the payment source and also inflates the size of deposit and loan. through expanding the size of deposit, loan and bill business in such non-normal ways, forms a vicious circle like this “enterprise uses loan fund as margin→bank issues acceptance bill→enterprise discounts the acceptance→enterprise uses the discounted money as margin →bank issues acceptance again →enterprise discount acceptance again....” in the iterative process, credit expands, loan multiplies illusorily, forms credit bubble. fig.1 the date of commicial bills and bills discounted in 2013 and 2012 0 10 20 30 40 2013 2012 commicial bills （trillion yuan） bills discounted (trillion yuan) 2.2. financing risk financing risk is caused by commercial bank issuing or discounting acceptance bill without real transaction background or unknown real transaction background. it’s characters are: no contract or using false or invalid contract; issuing or discounting beyond contract amount or date; no copy of value-added tax invoice, reusing the same invoice or forged invoice (duan, 2013). some banks, under the guise of enterprises, open accounts illegally. they use these accounts to issue acceptance bills and then make them discounted in other banks, bring money back to their own bank and make their saving in increase. they call it “financing deposits”. issuing or discounting bills without real transaction, the bank is in violation of relevant provisions of payment and settlement method and commercial bank bill acceptance and discount and rediscount interim measures. therefore, doing like this the banks are undertaking financing risk and at the same time, potential legal risk. 2.3. morality risk first, affiliated enterprises, affiliated transaction parties collude with each other, to cash bank funds. the main way is: drawer and payee are affiliated enterprises or affiliated transaction parties, the drawer use proportional margin to apply for acceptance, then the insufficient part is guaranteed by the payee, finally the money flows from the payee’s account to the drawer’s account, realizing the reverse flow of capital. second, some unscrupulous people use forged or altered notes, "cloning" bills or notes "switching" to defraud the bank funds consciously and make the bank face financial loss. third, the bank’s internal stuff do not strictly control the “three checks”: the check of legitimacy and compliance of items on bills, the check of endorsement verification, the check of authenticity of bills, or someone illegally and irregularly embezzles bank bills, thus leave bank suffers financial losses. 2.4. interest rate risk and liquidity risk the four state-owned commercial banks seldom provide credit to small and medium enterprises, their loan and the loan of small project is usually carried out this way: the small and medium banks discount acceptance bills, and then they rediscount from the four state-owned commercial banks. smcbs earn the spread in bill market. once market funds are tight, rediscount interest rates rise, small banks will face interest rate risk, and will lead to severe liquidity risks (jue, 2012). 2.5. handle risk first, discount in advance and check afterwards--reverse handle. in order to win business and enhance competitiveness, some banks ask enterprises to issue letter of commitment for discounting. first, however, the checking date is a few days or half a month late than the expenditure presentation date (bo, 2012). the way of discounting acceptance without assured checking carries certain risks and makes it difficult to ensure the safety of bank capital. second, the bill endorsement obtained by discount and pledge is illegal and incomplete. mainly like this: open endorsement when discounted, discontinuous multiple endorsement, etc. this not only goes against bills of exchange act, but also makes the bank published by atlantis press copyright: the authors 234 impossible to claim bill right when bill dispute or bill loss happened, what’s more this is easy to trigger morality risk. 2.6. internal management risk the foundation work of some smcb is weak, there is safety risk in bill kept and used, what’s more, ineffective internal controls, resulting in bills cases occur frequently, risk and loss continue to produce. at present most smcbs, in vary degrees, have the issue of archives incomplete and irregularities, such as, acceptance agreement, transaction contract are incomplete, the basic documents of acceptance applier is incomplete and convention records are not detailed (dan, 2009). some banks exist account and the actual inconsistent, account and account inconsistent, don’t implement “three separate charge” of seal, mortgage, and voucher. there is no supervision over seal take-over, make errors in accounting margin, use administrative seal instead of settlement seal in the consignment collection of acceptance bill. 3. reasons of bills business risk 3.1. backward invoice anti-counterfeit means in-conformity information between bank, enterprise and backward involves anti-counterfeit means. some enterprise is weak in legal awareness, they forge contract or invoice to defraud bank funds. what’s worse, the sharing mechanism is imperfect, such as banks haven’t networked with tax department, can not query the authenticity of the invoice; the credit registration information system also lacks information about acceptance bill discount. in addition, the invoice and contract offered by companies are not matched, a few enterprises can not offer value-added invoice, which brings difficulties for banks to check the real transactions background. because of lacking unified bill issue, query, and certificate system and institution, the electronization level of bills is low, leaving an opportunity to “fake bill” and” cloning bill”. 3.2. imperfect legal system the provisions of the current bills of exchange act， payment and settlement method and commercial bank bill acceptance and discount and rediscount interim measures are too general, have not made judicial interpretations about discount, transfer discount, rediscount business that involve bill business, and lack of legal basis in regulation. first, there is no monitoring index to acceptance business. in 2002, pboc canceled the requirement that the total amount of acceptance bills can be lager than the 5% of the ending balance last year, and there is no other guidance issued. for small financial institutions with small assets scale and low risk control ability, lacking of external constraints, is likely to cause the blind development. second, there are no unified regulations to government of acceptance margin, way of charging and amount of margin and no define and punishment standard to transferring loan to margin. third, there are no regulatory measures to enterprise’s practice of opening accounts in different names and reusing value-added invoice to discount in different banks. 3.3. management malposition and enlarge business blindly smcbs are restricted from operating zones, most of them have the problem of financial strain and small business scale. bill business has the scale expansion function, can bring banks extra business. therefore smcbs use bill business as the main way to enlarge business scale. thorough offering loan or discounting for enterprise, banks make the money as margin or fixed deposit and then issue bills to enterprises, handling like this in circle, can realize the both growth of loan and deposit. second, bill business can dilute non-performing loan (npl): through increasing bill business to enlarge loan scale, which makes the denominator bigger, the npl rate smaller. meanwhile, bill business can adjust profit, add operation revenue: through buy and sell bills to gain interests in advance, to adjust profit condition; when loan is used as margin, banks earn the loan interest, but pay deposit interest to enterprise, there for the interest margin becomes the revenue of the bank. 3.4. incomplete internal structure and weak risk consciousness smcbs have generally established internal control system, but the operation process and responsibility published by atlantis press copyright: the authors 235 system is not unified and strict enough to cover all the risk points, to effectively prevent risk. some banks don’t include all bill financing to the credit of a single customer, which is not good for controlling and preventing single customer’s credit risk; some banks haven’t implemented “three checks”, can not clearly grasp the credit condition of enterprise; in a few banks, the training and periodic rotation of stuff in important position have not been formed to a system, which is bad for prevent operation risk and morality risk; there are also banks whose auditor is short-staffed, bring about the weak supervision over bill business. 3.5. discriminated by weak position smcbs especially municipal commercial banks don’t have a nationwide network, big banks agent their bill inquiry. some agent banks discriminate small banks, deliberately delay the inquiry response. since the slow response, the longest about one month, forces the authorized bank to discount in advance for enterprises in urgent need, or to simply discount at sight for big customers, forming handle risk in bill business. 4. bill business risk prevention and management 4.1. strengthen internal control and improve the risk prevention mechanism smcbs should adhere to the "internal control priority" principle, focus on risk management, strengthen the building of internal control mechanism, set up normative ticketing, registration, inquiry procedures and systems with strong operability, intensify the review and check of the authenticity of bills, and ensure the authenticity of bill transactions and the security of finance. meanwhile, each risk point in bill business procedure should be carefully predicted and analyzed. each risk point should be set up with "firewall", and risk management procedure should be evaluated regularly. smcbs should also enhance staff training to raise the awareness of business risks, improve the staff's defensive skills and build up noble work ethics and strictly perform operation, check and supervision system (shan and qian, 2012). 4.2. correct business ideas and enhance the consciousness of risk prevention smcbs should stop being obsessed with scale and speed, correctly handle the relationship between bill business development and risk prevention and avoid pursuing bill gains at the cost of assuming tremendous advances risks. comprehensive, coordinated and sustainable development view and scientific evaluation mechanism should be set up. the development goal and scheme of bill business should be established according to its own characteristics and actual situation. it should be avoided to be obsessed with scale and speed. business development should be based on controllable risk. 4.3. strengthen compliance regulation and promote the healthy development of bill business regulatory authorities should be fully aware of the potential risks in bill business, strengthen the efforts to on-site inspection, mainly examine bill acceptance and the authenticity of discount business’s background, discounting cash flow, loans as deposit, inner control regime and risks in bill business etc (ba and bei, 2005). special attention should be paid to the regulation of the effectiveness of internal control, operant behavior and each risk point. regulatory authorities should further strengthen the penalties for violations in bills business, strengthen supervision inspection, timely correct and handle illegal behaviors to maximize risk prevention and control. 4.4. perfect loan registration information system and solve the problem of asymmetric information pbc and regulatory authority should establish bank acceptance’s issuing and discounting business database, achieve national network query function, improve working efficiency, connect with tax department, timely recognize the authenticity of vat ticket copies or originals provided by enterprises, enhance the examination of trade background and prevent and mitigate risks. 4.5. revise the system and regulate the bill market some items in the current commercial instrument law, payment and settlement method, commercial bank bill acceptance and discount and rediscount interim published by atlantis press copyright: the authors 236 measures are not suitable for the development status of bills. admittance and retreating system for enterprises in bill business, the management of bill financing, bill transaction, bill credit evaluation, the price agreement of bill transaction, capital settlement and business risk reserve system etc. should be standardized as soon as possible. from a regulatory point of view, specific regulations regarding commercial bank bill business risk control should be made. first, bill financing should be wholly included in the range of credit authorization and customer credit to facilitate external regulation and internal control. second, directive monitoring index about total acceptance and acceptance proportion of single or ten household at most, and control regime regarding issuing to offset companies should be formulated to eliminate bubbles in bill market. third, security deposit raking management should be implemented or the zero norm of security should be regulated. the bill risk exposure should be linked with capital scale to reduce the acceptance risks of banks. 5. conclusions as a new intermediary business, bill business has been a new point of profit growth for small and medium commercial banks (smcb). however, with the development of bill business, the risks have been exposed. in order to prevent the bill business risk, smcbs should take the following measure. first, smcbs should stop being obsessed with scale and speed, correctly handle the relationship between bill business development and risk prevention and avoid pursuing bill gains at the cost of assuming tremendous advances risks. second, clearly defined standards and punishment regulation regarding margin loads and affiliates obtaining bank credit should be formulated. third, require all the financial institutions that the original value-added invoice should be stamped, noting discount amount, discount institution name, to prevent rediscount by the same invoice from the very beginning. references s. s. ba and j. bei, bills market research on development, accounting monthly (3) (2005) 15-16. m. x. bo, risks and measures currently exist in bill business, accountants (18) (2012) 25-26. l. w. chun, analysis of commercial bank’s bill business risks and preventive measures, market weekly (2) (2014) 80-82. z. dan, key point of china's commercial banks in the field of bill business risk prevention, financial economics (18) (2009) 110-111. y. duan, joint-stock commercial bank bill business risk, financial era (24) (2013) 76. z. jue, the business risk of commercial bank bills, rethinking times of financial (35) (2012) 109. c. shan and x. z. qian, characteristics and the influence of bank bill business, china financial (5) (2012) 82-83. c. xue and l. y. xiao, internal control problems of commercial banks’ bill business, modern economic information (16) (2014) 297-299. published by atlantis press copyright: the authors 237 1. introduction 2. the main risks in bill business 2.1. credit risk 2.2. financing risk 2.3. morality risk 2.4. interest rate risk and liquidity risk 2.6. internal management risk 3. reasons of bills business risk 3.1. backward invoice anti-counterfeit means 3.2. imperfect legal system 3.3. management malposition and enlarge business blindly 3.4. incomplete internal structure and weak risk consciousness 3.5. discriminated by weak position 4. bill business risk prevention and management 4.1. strengthen internal control and improve the risk prevention mechanism 4.2. correct business ideas and enhance the consciousness of risk prevention 4.3. strengthen compliance regulation and promote the healthy development of bill business 4.4. perfect loan registration information system and solve the problem of asymmetric information 4.5. revise the system and regulate the bill market 5. conclusions references atlantis press journal style study on risk evaluation based on occupational exposure evaluation and carcinogenic risk simulation* qian zhang1, deyin huang1, mao liu2 1 tianjin bohai chemical industry group co., tianjin 300051, china 2 research center of urban public safety nankai university, tianjin 300071, china abstract in order to provide the quantitative risk evaluation for 1,3-butadiene-expose, quantitative method using physiologically based pharmacokinetic model and dose-response model is applied to the carcinogenic risk evaluation for 1,3-butadiene-exposed workers. first, the internal dose of inhaled-1,3-butadiene in human beings is simulated by exposure related dose estimating model recommended by usepa. second, the internal dose is induced into calculating formula of carcinogenic risk. in order to determine the uncertainty of the cancer risk, monte carlo simulation is used to analyze the risk probability distribution. a large chemical enterprise in tianjin is selected as the case study and carcinogenic risk evaluation for 1,3-butadiene-exposed workers is calculated. the results show that the cancer risk of 1,3-butadiene-exposed workers obviously exceeds 1×10-4 the maximum acceptable risk level. it is necessary to take relevant measures to reduce risk. keywords: risk evaluation, 1,3-butadiene, occupational exposure evaluation, carcinogenic risk simulation. * this study was supported by municipal key science and technology support project of tianjin (no. 13zczdsy02300). 1. introduction in recent years, as china's rapid industrial development, harmful factors of industrial production on human health increasingly have aroused widespread concern. as an important organic chemical material, 1,3-butadiene (bd) is widely used in synthetic rubber, synthetic resin, synthetic fibers, plasticizers and other petrochemicals and manufacturing industries. the international community is increasingly concerned about the carcinogenicity of 1,3 butadiene. occupational epidemiology studies have shown that the carcinogenic risk of works suffering from 1,3-butadiene occupational exposure increased significantly. it is particularly important to assess the carcinogenic risk suffering from 1,3-butadiene occupational exposure. in the normal production situation, most of the monitoring data of hazardous substances is below the occupational exposure limits, but some human definite carcinogen such as bd has carcinogenic effects on human even if in the condition of low concentration. with development of the health risk assessment methods, the concept of the internal dose has been introduced into carcinogenic risk assessment. the internal dose is the exposure material quantity which remains in human body after absorption and transformation. more and more researchers apply physiologically based pharmacokinetic (pbpk) model into chemical dose response evaluation studies [1] and use internal dose instead of exposure dose in the risk assessment, which is considered more convincing and accurate[2-3]. the human internal dose of chemical substances calculated by pbpk model relates to several complicated metabolism differential equations. in order to solve the complex pbpk model calculations, researchers often use program methods, and the calculation program is generally written in acsl[4-5]. shi jie et al. adopted the trichloroethylene pbpk model existing in exposure related dose estimating model (erdem) to simulate human internal dose, which eliminated the need for complex programming. however, there is no research explaining how to self-build a pbpk model. this study journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 228-232 published by atlantis press copyright: the authors 228 willieb typewritten text received 20 august 2014 willieb typewritten text accepted 27 october 2014 willieb typewritten text willieb typewritten text elaborates for this problem, and introduces the internal dose simulated by this method into the carcinogenic risk calculation. the research content and method are shown in fig. 1. firstly, in erdem software, 1,3-butadiene internal dose is simulated by pbpk model which is built using metabolic parameters in the literature. then, we substitute the internal dose in the multistage dose-response model to obtain the cancer risk. finally, uncertainties in cancer risk assessment are analyzed by monte carlo simulation. figure 1. research content and method 2. software simulation of dose using erdem i. software brief introduction exposure related dose estimating model (erdem) is a pbpk model simulation software in acsl language and developed by epa’s national exposure research laboratory. the metabolism of toxic substances in the body can be simulated by choosing compartments and entering metabolic and physiological parameters. the database includes data of epidemiological investigations and toxicology experiment, and also includes approved pbpk models of many toxicant and its metabolites. erdem provides the flexibility either to use existing models and to build new pbpk models to address specific science questions. ii. running process fig. 2 shows the running process which erdem builds pbpk model and simulates the internal dose. the running steps are as follows: ①new modeling--set the simulated chemical name and molecular weight in the appropriate interface. fill in the name of model to be built, the exposing object to be studied, units and so on. ② pbpk model organization and parameters setup--select compartment and metabolic process, and input the organization parameters, metabolic rate, metabolic constants and so on. ③exposure parameters setup--input exposure route, time, concentration, etc. ④ output option setup--erdem can analyze chemical substances and their metabolites in different compartment, and give different output forms, such as concentration, auc, volume, etc. so this step can set the output options to get the anticipant result forms. ⑤ simulation running and result output--according to output options, erdem can give the internal dose quantity and curve over time. figure 2. running processes of erdem 3. 1,3-butadiene pbpk modeling the 1,3butadiene exposure route considered in this paper is inhalation. when bd enters human body, it is metabolized to monoepoxide — — 1,2–epoxybutene, which will produce a variety of active metabolites. the metabolites can be conjugated with biological macromolecule, which results in changes of genetic material and inducing tumor [6-8]. the metabolic process of bd into 1,2epoxybutene is the first step of the following metabolic process. according to related literature parameters [9], this paper build 1,3-butadiene pbpk models of mouse and human, and compared with the experiment data. model includes six compartments, such as lung, blood, fat, liver, vessel-rich compartment and vessel-poor compartment. metabolism to the monoepoxide is ascribed to the entire liver and is assumed to follow simple michaelis-menten kinetics. no further metabolism of epoxybutene is considered. s. fustinoni et al. [10] pointed out when exposure concentration was low, the concentration of the bd which remained in human body after metabolism can be used as biomarker. this paper use pbpk model to calculate bd blood concentration d, which is regarded as the internal dose of internal dose-response relation p(d). parameters and pbpk model are respectively shown as table 1 and fig. 3. new modeling pbpk model organization and parameters setup exposure parameters setup output option setup simulation running and result output uncertainty analysis risk probability distribution external dose internal dose pbpk model erdem software simulation dose-response model crystal ball software monte carlo cancer risk published by atlantis press copyright: the authors 229 figure 3. pbpk model by the respiratory metabolism after intake of bd table 1. 1,3-butadine pbpk model parameters parameters mouse human parameters mouse human physiological parameters volume proportion weight (kg) 0.028 70 blood 0.05 0.077 cardiac output (l/h) 1.044 660 fat 0.04 0.144 alveolar ventilation (l/h) 2.64 1,200 liver 0.062 0.025 blood flow proportion vessel-rich compartment 0.05 0.037 fat 0.05 0.036 vessel-poor compartment 0.78 0.547 liver 0.16 0.16 partition coefficients vessel-rich compartment 0.52 0.446 air 1.5 vessel-poor compartment 0.19 0.361 fat 118.2 metabolic parameters liver 5.49 liver v (nmol/h/mg) 155.4 70.8 vessel-rich compartment 5.34 liver km (mm) 0.002 0.00514 vessel-poor compartment 5.26 iv. verification of simulation results in order to verify pbpk model simulation results, the mouse metabolic internal doses simulated by erdem software are compared with biological experimental data. himmelstein et al. measured blood concentrations of 1,3-butadiene in mice exposed by nose-only inhalation to 62.5, 625, and 1,250 ppm 1,3-butadiene for 6 h. steady-state blood concentrations of 1,3-butadiene were achieved by 2 h. bd internal dose accord with the experimental data as shown in the fig.4, hence, pbpk model can predict and analyze the metabolism situation of bd in the body. however, there is no available 1,3-butadiene metabolic experimental data in human body, so human internal dose results are no further verified. figure 4. experimental data and simulation results 4. carcinogenic risk calculation dose response function of internal dose (d) to carcinogenic probability can be solved by several dose response model, such as probit model, logit model, weibull model, multi hit model, one hit, multistage model and so on, where multistage model is used most widely. usepa indicate that multistage model should be recommended in the research of dose response relationship. according to the usepa 1,3-butadiene risk assessment guideline[9], this paper uses linear multistage model shown as follows: )]dq...dqd1q(exp[1)d(p kk 2 2 1 10 ++++−−= , parameter qi≥0，i=0,1,…,k. (1) in latest health assessment of 1,3-butadiene[9], integrated risk information system (iris) point out that lots of epidemiological studies implicate bd is related to occupational contact crowd inducing hodgkin’s disease, leukemia, lymphosarcoma, lung cancer and so on. this paper attends to the calculation of occupational carcinogenic risk probability 1,3-butadiene. a 2-year chronic inhalation toxicity and carcinogenicity study on effects of 1,3-butadiene on b6c3f1 mice was conducted by national toxicology program (ntp). based on tumor incidence of mouse and multistage model, carcinogenic lung lung blood fat compartment vessel-rich compartment vessel-poor compartment liver venous blood active metabolite inhalation expiration metabolize arterial blood published by atlantis press copyright: the authors 230 probability function of 1,3-butadiene exposed workers is shown as follows: h a 2 )0(p )0(p ]5124.0d299.23d44.281exp[1 )d(p × −×−×−− = (2) where p(d) represents the lifetime risk (probability) of cancer at internal dose d, p(0)a represents background risk value of animal cancer, which is 0.12 obtained from experimental result, p(0)h=200 × 10-6 represents background risk value of human cancer. 5. uncertainty analysis in risk assessment, uncertainty which induces evaluation result cannot guarantee the reliability is always caused by randomicity of the objective world, our lack of knowledge and errors of evaluation method[12]. uncertainty of carcinogenic risk assessment in this paper is from the calculation of the internal dose and risk, which can be known from the research line. pbpk modeling in the calculation of the internal dose exists errors caused by animal experiment, and parameters also can cause uncertainties. another uncertainty factor is interspecific differences and own errors of the dose response relationship in the process of carcinogenic risk calculation. these can affect the real risk reflection of calculation results in different degree, and induce uncertainties of the carcinogenic risk. monte carlo analysis (mca) is one of the commonly used uncertainty analysis methods[13], and applies probability method to analyze the uncertainty and risk assessment characterization. monte carlo simulation of cancer risk in this paper is operated by crystal ball, which analyzes the uncertainty from risk probability distribution. 6. case study this method of carcinogenic risk assessment for 1,3-butadiene is applied into a large chemical enterprise in tianjin. the concentration monitoring method of hazardous material in the air of workplaces complies with solvent desorption gas chromatography in determining 1,3-butadiene regulated in “determination of alkenes in the air of workplace” (gbz/t160.39-2007). method of air sample collection complies with “specifications of air sampling for hazardous substances monitoring in the workplace” (gbz159-2004). field monitoring is made on 1,3-budatiene concentration in the air of workplaces, and 112 samples are collected in three consecutive days. according to the measured data of bd, internal dose of 1,3-butadiene exposed workers are simulated by erdem software. bd blood concentration after in vivo metabolism is obtained by setting continuous exposure for 8 hours, and substituted in (2) of dose response relationship to get 1,3-butadiene carcinogenic risk value. table 2 shows the mathematical statistics results of bd actual measured concentration and the simulatioin results of internal dose and carcinogenic risk. finally, the uncertainty of carcinogenic risk is analyzed by monte carlo simulation. fig. 5 and 6 respectively show the risk probability distribution and probability distribution obtained by crystal ball software. the results show that the cancer risk of workers in this enterprise is 6.52×10-4~1.24×10-3, and the average value is 7.20×10-4. generally, the risk level of 1×10-4 is used as maximum acceptable risk[14]. the risk values are all above 1×10-4 according to the simulation results by crystal ball, as shown in the table 2. table 2. actual monitoring concentration, internal dose by simulation and cancer risk average value median value minimum value maximum value monitoring concentration /mg/m3 3.669 2.1 0.1 27.3 internal dose /mg/m3 2.3 1.3 0 27.2 carcinogenic risk 7.20×10-4 6.99×10-4 6.52×10-4 1.24×10-3 figure 5. probability distribution of 1,3-butadiene cancer risk published by atlantis press copyright: the authors 231 figure 6. percentile distribution of cancer risk 7. discussion and conclusion 1)to provide the theoretical basis of quantitative research for occupational health risk in china, carcinogenic risk for 1,3-butadiene-exposed workers is evaluated by physiologically based pharmacokinetic (pbpk) model and multistage dose-response relationship. firstly, internal dose of 1,3-butadiene is analyzed by erdem which is a pbpk model simulation software recommended by usepa. then, carcinogenic risk is calculated by substituting internal dose into the function of carcinogenic risk. finally, risk probability distribution is simulated by monte carlo method to analyze the uncertainty. 2) this research method of carcinogenic risk evaluation for 1,3-butadiene-exposed workers is applied in the case of a chemical enterprise in tianjin. the result shows that occupational carcinogenic risk for bd exposed is higher than acceptable risk level of chemical enterprise, and it should adopt corresponding measures to reduce risk. 3) although uncertainty is analyzed, pbpk model, parameters and experimental research related in dose-response relationship are cited from literature data which is mostly foreign. these cause variety of uncertainties, and in order to more conform to reality using localization data in the research is the main problems for further investigation. references [1] krewski d, withey j r, ku lung-fa, et al. applications of physiologic pharmacokinetic modeling in carcinogenic risk assessment[j]. environmental health perspectives, 1994, vol. 102(11): 37-50. [2] liu shuyuan, wang hongqi.health risk assessment of organic pollution in a groundwater source[j]. environmental science and technology, 2014, vol. 37(2): 174-177. [3] zhang qian, liu mao huang de-yin. study on carcin ogenic risk evaluation of benzene-exposed workers[j]. ch ina safety science journal, 2011, vol. 21(5): 143-147. [4] lu ya-song, raymond s h yang. introduction to physiologically based pharmacokinetic model construction: an example of 1,1,1-trichloroethane[j]. journal of labour medicine, 2006, vol. 23(4): 330-338. [5] shi jie. analysis of acute health risk brought by accidental toxic gas leakage[d]. tianjin: nankai university , 2009. [6] tan hong-shan, miu wen-bin, sun pin, et al. updated research on 1,3-butadiene adducts. occupational health and emergency rescue, 2009, vol. 27(6): 303~308. [7] liu nan, cheng juan, li bin, et al. research progress on 1,3-butadiene biomarkers. foreign medical sciences(section of hygiene), 2007, vol. 34(6): 352~357 [8] cheng xue-mei, jiao yan-ni, chen jin-dong, et al. stu dy on urine biomarkers in 1, 3-butadiene exposed workers [j]. chinese journal of industrial hygiene and occupationa l diseases, 2012, vol. 30(9): 661-666. [9] u.s. environmental protection agency. health assessment of 1,3-butadiene (epa/600/p-98/001f). washington, dc, 2002 [10] s. fustinoni, l. perbellini, l. soleo, et al. biological monitoring in occupational exposure to low levels of 1,3-butadiene. toxicology letters, vol. 149(1-3):353-360. [11] matthew w. himmelstein, max j. turner, bahman asgharian, et al.comparison of blood concentrations of 1,3-butadiene and butadiene epoxides in mice and rats exposed to 1,3-butadiene by inhalation. carcinogenesis, 1994, vol. 15(8): 1479~1486 [12] wang li , huang de-yin, liu mao. application of monte-carlo simulation method in cancer risk assessment f or benzene exposure [j]. journal of safety and environment, 2011, vol. 11(5): 231-235. [13] sassi g, magnetti vernai a, ruggeri b. quantitative estimation of uncertainty in human risk analysis[j]. journal of hazardous materials, 2007, vol. 145(1-2): 296-304. [14] li yang, zhou chang-yu, zhang bo-jun. study on acceptable level of risk in petrochemical industry[j]. journal of safety and environment, 2007, vol. 7(6): 116-119. published by atlantis press copyright: the authors 232 1. introduction 2. software simulation of dose using erdem 3. 1,3-butadiene pbpk modeling 4. carcinogenic risk calculation 5. uncertainty analysis 6. case study 7. discussion and conclusion 陕甘宁地区旱灾发生的气候影响因素分析 received 4 december 2015 accepted 15 march 2016 the analysis on the effecting factor of drought disease in qingyang, gansu xiaodong wang the college of mine and coal, inner mongolia university of science and technology, baotou 014010, china abstract in recent years, the global climate is becoming warmer and warmer. the discussion about the global warming effecting on is also more and more intensified. the paper analyzes change of air temperature and precipitation change occurred from last half century to the beginning of this century in the qingyang region, gansu province, china. considering the relation between el-nino, la-nina and drought disease, we want to find out the reason of drought disease under the global warming, and to contribute to the theory studying of relieving poverty. the analysis and concluding is that drought disease is worse and worse in the region, air temperature and precipitation is negative correlation obviously, the rising of air temperature is main reason of drought disease. key words: global warming, precipitation change, drought disease, el-nino, la-nina, qingyang 甘肃庆阳地区旱灾发生的气候影响因素分析 王晓冬 内蒙古科技大学矿业与煤炭学院，包头 014010, 中国 摘要：近年来，全球气候变暖已经成为不争的事实，针对气候变暖所带来影响的讨论成 为一个热点。本文通过分析上世纪中期至本世纪初的 50 年间甘肃庆阳地区的气温、降水变化 趋势及厄尔尼诺、拉尼娜事件与干旱气候的关系，试图探讨全球气候变化下本区干旱的情况及 其产生原因，为减轻本区由此产生贫困的问题提供理论依据。分析认为：甘肃庆阳地区干旱情 况有越来越严重的趋势，本区降水与气温存在着明显的负相关性。这说明气温的升高是干旱发 生的主要因素之一，厄尔尼诺及拉尼娜事件与干旱的发生也有一定的关系。 关键词：全球气候变暖; 降水变化; 干旱；厄尔尼诺；拉尼娜； 庆阳地区 journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 103-108 published by atlantis press copyright: the authors 103 x.d. wang / the analysis on the effecting factor of drought disease in qingyang, gansu 随着全球气候的日益变暖，极端天气现象 频发，在这之中，地区降水变化的不均衡表现 得尤为突出，不仅是国家与国家之间降水变化 不均衡，即使在同一国家内部，不同地区之间 降水变化的不均衡性表现得也相当明显。拿中 国来说，常常是北方大旱的同时南方则连续洪 涝，因此分析全球气候变化对不同地区的影响 对预防极端天气变化，减少自然灾害发生所带 来的生命和财产等损失有着重要意义。 甘肃庆阳地区是陕甘宁革命老区重要组 成部分，又是全国著名的国家级贫困地区，全 区具有典型的黄土高原地形、地貌，是典型的 旱作农业区，由于地表水的缺乏，农业生产对 大气降水依赖程度比较高。随着全球气候的变 化，本区干旱程度的变化日益明显，严重影响 并制约着本区的农业生产。为配合国家新的五 年计划中关于扶贫工作的安排部署，分析造成 干旱的气候影响因素对本区发展农业生产、摆 脱贫困、实现国家战略目标有着重要意义。本 区位于甘肃省东部，介于 106 45´–108 45e 与 35 10´–37 20´n 之间，辖区有西峰、庆阳、镇 原、宁县、正宁、合水、华池和环县七县一市， 包括 146 个乡镇，1488 个行政村，11064 个自 然村，总人口约 251.46 万，其中农业人口 223.75 万[1]。本区是陇东黄土高原的重要组成 部分，有典型的黄土高原沟壑、丘陵地貌。辖 区农业人口占总人口比重 90%以上，农业是当 地支柱产业，但年平均降雨量在 480-660mm， 全区 92%的耕地为旱作农业，因此对环境的依 赖程度较高。2000 年的农民人均纯收入为 1272 元 1999 年的农民人均纯收入为 1380 元，农民 收入仍然受自然灾害影响 [1] 。全区人均 gdp 为 2278 元仅占全国和全省平均水平 32.4%和 57.9%，社会经济的总体发展水平仍相当落后 [2] 。本区 8 个县市中就有一个国家级贫困县和 一个省级贫困县，因此如何尽早摆脱贫困，对 完成国家未来几年的经济发展目标有着重要 意义。而气候变化对农业的影响是首当其冲的 重要指标，特别是全球气候变化下日益干旱的 特点严重制约着本区农业发展，如何根据气候 变化规律和干旱情况指导农业发展，积极预防 和应对也是行之有效的办法之一。 1. 数据与分析方法 通过对甘肃庆阳地区西峰、庆阳、镇原、 宁县、正宁、合水、华池和环县 8 个县市由 1951 到 2000 年降水情况的分析，对比本区气温变 化趋势与厄尔尼诺及拉尼娜现象发生规律，寻 找干旱与气温和极端事件之间的关系，分析干 旱发生的影响因素，为本区经济发展提供理论 依据。 2. 干旱情况日益加剧 2.1 年降水量逐年减少 据 1970 至 2005 年降水数据绘制时间序列 图（图 1）可知，本区年降水量呈逐年减少的 趋势，且降幅较大，表明干旱程度加剧。且通 过相关分析可知，降水变化与年份存在正相关 且相关系数较高，表明降水量减少与时间变化 的相关性明显。 图 1 庆阳地区年降水变化趋势 fig.1 the yearly average precipitation change of qing-yang region 2.2 四季降水变化不均衡 利用庆阳地区春、夏、秋、冬四季降水统 计数据作趋势图(图 2-5)分析可知，春季，本区 降水减少但趋势不明显，降水随时间的相关系 数 r=0.0003，表明几十年来春季的干旱情况随 时间变化的影响不大。 图 2 庆阳地区春季降水变化趋势 fig.2 the spring season average precipitation change of qing-yang region published by atlantis press copyright: the authors 104 x.d. wang / the analysis on the effecting factor of drought disease in qingyang, gansu 图 3 庆阳地区夏季降水变化趋势 fig.3 the summer season average precipitation change of qing-yang region 图 4 庆阳地区秋季降水变化趋势 fig.4 the autumn season average precipitation change of qing-yang region 图 5 庆阳地区冬季降水变化趋势 fig.5 the winter season average precipitation change of qing-yang region 夏季，与春、秋、冬三季降水逐渐减少的 趋势不同，夏季降水呈明显上升趋势，且增幅 较大，相关系数 r=0.0157，相关性明显。表明， 夏季降水随时间变化有明显增加的趋势。 秋季降水呈大幅下降趋势，下降幅度为每 年 1.56 毫米，相关系数较大为 0.1095，表明降 水随时间变化下降幅度相当明显，且两者相关 性很大。 冬季降水也呈明显下降趋势，下降幅度为 每年 0.23 毫米，相对秋季降水下降情况来看， 虽然总体下降幅度没有秋季大，但相对而言， 其下降情况也较严重。 由以上分析可知，四季当中，春季降水趋 势保持平稳，降幅不大，夏季降水不仅没有减 少反而呈现升高的趋势，且升高幅度较大，秋 冬两季降水明显呈下降趋势且幅度较大，表明 四季降水变化趋势不均衡。 本区由于主导产业为农业，所以降水的季 节分配对本区经济影响较大，总体来讲，夏季 降水量的增加一定程度上对本区农作物生长 有一定的好处，但夏季降水的增长幅度为每年 0.78 毫米，虽然增长幅度较大，但相对秋冬两 季总共的减少幅度为 1.79 毫米而言，总体全年 降水仍呈现巨大的减少趋势，由此表明本区降 水逐年减少，干旱情况日益严重。 除全年降水量的多寡对农业生产有较大 影响外，年内的四季降水分配情况也对农作物 的生长影响较大。中国是典型的季风气候，在 气温和降水的组合分配过程中水热同期的组 合特点一定程度上有助于农作物生长，也就是 说，夏季气温最高同时夏季又是一年中降水量 最大的季节对作物生长是有好处的，同时夏季 的降水有逐年增加的趋势对本区农耕经济有 力。但同时秋冬季降水急剧减少，会使土壤含 水量大量减少，而春季降水又不足以弥补损 失，使土地墒情受到严重影响，因此要想弥补 由于秋冬降水减少所带来的影响，如何改善农 业耕种制度也是刻不容缓的工作。 2.3 降水距平值为负的年份较多 根据年降水低于平均值的情况将 1954 年 至 2000 年年降水与年均降水量之差绘制降水 距平图（图 6）可知，距平值为负值的年份较 多，且有些年份甚至低于-20%。 图 6 庆阳地区降水距平变化 fig.6 the comparing yearly average precipitation change of qing-yang region -30 -20 -10 0 10 20 30 40 1 9 5 4 1 9 5 8 1 9 6 2 1 9 6 6 1 9 7 0 1 9 7 4 1 9 7 8 1 9 8 2 1 9 8 6 1 9 9 0 1 9 9 4 1 9 9 8 降 水 距 平 值 （ % ） published by atlantis press copyright: the authors 105 x.d. wang / the analysis on the effecting factor of drought disease in qingyang, gansu 根据降水距平变化情况（图 6）可以看出， 庆阳地区降水低于平均值的干旱年份较多，特 别是上世纪 90 年代以来，低于平均值的年份 有增加趋势。表明本区降水量有逐年减少的趋 势，而干旱情况则日益严重。对从 1950 年至 2000 年的降水距平为负值的年份做统计分析， 绘制负距平年份图表（表 1），可以看出，50 年间降水负距平的年份共有 17 年次，尤其是 进入 70 年代以来干旱年份更多、更加频繁， 其中低于-20%的极端干旱年份共有 5 年，分别 是 1974、1975、1985、1994、2000 年。 表 1 庆阳地区降水距平为负值（干旱）的年份 tab.1 years of precipitation difference is below 0 (drought disease) 年代 甘肃庆阳 50 1955 1956 1958 60 1960 1961 1963 1966 70 1972 1974 1975 1977 1979 80 1982 1983 1985 1989 90 1990 1992 1994 1996 1998 2000 3. 庆阳地区干旱发生的因素分析 3.1 气候变暖是干旱发生的主要因素之一 根据甘肃庆阳地区 1970 到 2005 年间的气 温变化情况分析（如图 7），气温总体呈上升 趋势，其线性方程为：y=0.042x+8.391，表明 气温以每年平均 0.042 的速度递增，远高于全 球平均增温幅度 0.03℃/年。相关系数 0.697， 所以变暖的趋势很明显。 图 7 庆阳地区 1970-2005 年平均气温变化图 fig.7 the yearly average temperature change of qing-yang region 世界气候变化越来越受到关注,有一点是 公认的，即近百年来气候存在着明显增温的趋 势。政府间气候变化委员会（ipcc）指出近百 年来气候增温约 0.5℃区[3] ，其第三次评估报 告指出，20 世纪全球平均温度升高 0.6±0.2℃， 80 年代和 90 年代是近 100 年来最温暖的 20 年 [4] 。据专家推测，如果目前温室气体排放趋 势继续下去，地球表层将以 0.3℃/10a[5] 的速 度增温。 气候变暖对降水量分布及多寡变化的影 响以及由此产生的一系列对农业生产带来的 后果问题，长期以来一直是人们关注的焦点, 许多学者在做了大量分析研究的基础上纷纷 提出自己的观点，一般认为,气候变暖后降水会 变得更加干燥和不稳定 [6] 。气候变暖很大程度 上会引起干旱的发生从而影响农业生产。在我 国，西北、华北和东北大部分地区是增温最显 著的地区 [7] 。kellogg 认为中国主要农业区很可 能变干旱，对粮食生产有严重影响 [8] 。 美国学者 parry 和 swaminathan 也指出， 全球升温将使中国北方和华中的土壤水分减 少，农业减产 [9] 。还有人认为：全球变暖将导 致世界主要粮食产区的中纬度和大陆中部地 区将变干旱 [10] ，国外实验也表明，对于生长季 节来说，在气温升高 1℃的情况下，灌溉需求 增加 12%，易发生旱灾[11]。 因此伴随着全球气候变暖本区气温也呈 现了上升趋势，并且与此同时随着本区气温总 体上升干旱情况也越来越严重，干旱的发生又 直接影响农业生产，而本区农业占经济收入的 过高比重也最终成为本区贫困发生的重要因 素。 3.2四季气温均升高但增幅不同 根据本区四季气温数据资料做趋势分析 (如图 8-11)可知，春、夏、秋、冬四季气温均 呈现增加趋势，表明在年均温逐年上升的同 时，四季气温也呈现出逐年上升的趋势，但明 显增温的幅度不同，其中秋季增温幅度最小， 而冬季增温幅度最大。且相关系数也最大，表 明年均温的逐年增加最主要的表现在冬季气 温随时间变化增加趋势最大的基础上。 图 8 庆阳地区春季气温变化趋势 fig.8 the spring season average temperature change of qing-yang region published by atlantis press copyright: the authors 106 x.d. wang / the analysis on the effecting factor of drought disease in qingyang, gansu 图 9 庆阳地区夏季气温变化趋势 fig.9.the summer season average temperature change of qing-yang region 图 10 庆阳地区秋季气温变化趋势 fig.10 the autumn season average temperature change of qing-yang region 图 11 庆阳地区冬季气温变化趋势 fig.11 the winter season average temperature change of qing-yang region 对比降水的季节变化，气温的季节变化一 方面有助于农作物生长，特别是夏季气温升高 的同时夏季降水不减增，雨热同期对本区的旱 作农业有一定好处。但另一方面，秋冬季气温 特别是冬季气温增加最明显的同时秋冬季降 水却大幅度减少，其中冬季降水减少幅度最 大。这种水热组合状况必然会对农业产生一定 影响。俗话说，麦收隔年墒，秋冬季的升温会 加剧土壤水分蒸发，同时秋冬季降水又呈现下 降趋势，这就使得保存在土壤中的水分锐减， 而本区属于黄土高原地形地貌，地表水、地下 水相对较少，农业生产更多依靠大气降水，虽 然秋冬季的土壤干旱不会给农作物造成较大 影响，但来年开春，由于春季升温的同时降水 量并没有增加的趋势，使得秋冬季土壤损失的 水分不能及时得到有力的补充，所以极易造成 春旱，而同时春季正是作物出苗、分蘖等重要 时期，所以会对农业产生较大影响从而影响本 区经济的经济发展。 本区现有的农作物从成熟季节来看有夏 熟和秋熟两种，夏熟作物主要指三月播种六月 收割的春小麦，或是可越冬来年六月收割的冬 小麦，秋熟作物指春季播种秋季收割的粮食作 物如玉米、马铃薯和油料作物如胡麻等。从水 热组合情况来看，夏季的雨热同期更适合秋熟 作物的生长，而夏熟作物因为成长季主要集中 在春季和夏初，而此时夏季降水还没有充分来 临，上一年的秋冬季暖而干的气候已经使土壤 储存水分大量蒸发，且春季降水变化不大，又 使得没有充足的降水进行弥补，所以土壤处在 严重缺水期，因此夏熟作物不能很好生长。所 以如何改变耕种制度也是应对气候变化，提高 农业生产，实现脱贫致富目标的主要途径之 一。 3.3气温与干旱的关系 利用庆阳地区 1950 年至 2000 年降水数据 资料作两者的相关趋势分析（如图 12）可知， 二 者 呈 明 显 负 相 关 性 ， 其 中 线 性 方 程 为 y=-58.432+1046 且相关系数较高 r=0.1574， 表明本区气温升高的同时降水明显较少，即暖 干现象明显。说明气温的变化是造成本区干旱 的主要原因之一。 图 12 庆阳地区降水与气温变化关系 fig.12.the relation of precipitation and temperature change of qing-yang region 3.4气候极端事件与干旱的关系 气候极端事件是指某一地点或地区从统 计分布的观点来看不常有或极少发生的天气 事件 [12] 。厄尔尼诺和拉尼娜现象就是其中一 种。厄尔尼诺又称南方涛动，主要指在东南太 平洋与印度洋及印尼地区之间的反相气压振 动，是一种海水偏暖对大气产生影响的海气互 动现象。拉尼娜则与之相反，是由于海水温度 published by atlantis press copyright: the authors 107 x.d. wang / the analysis on the effecting factor of drought disease in qingyang, gansu 偏低所导致的气候异常。这两种事件的出现常 常引起世界范围的气候异常，表现为使大陆夏 秋两季降水分布南北不均，易出现南涝北旱现 象 [13] 所以 2002 年前后长江流域应重点防御 重大洪涝灾害，而华北、特别是西北地区应重 点防御重大旱灾的发生 [14] 。 据世界气象组织的报道：1951 到 2000 年 间，厄尔尼诺发生了 14 次，发生年代分别为 1953、1957~1958、1963、1965~1966、1969、 1972~1973、1976、1982~1983、1986~1987、 1991~1992 、 1993 、 1994~1995 、 1996 、 1997~1998 ，由此可以看进入 90 年代以来厄 尔尼诺发生频繁且具有持续时间长、间隔时间 短的特点 [12] 。厄尔尼诺年及拉尼娜年世界气候 变化较大，极易引起降水的地区分配异常，导 致局部地区涝灾、旱灾频繁发生，如 1998 的 厄尔尼诺年，我国长江流域、嫩江、松花江流 域出现特大涝灾。而此时甘肃庆阳地区却面临 旱灾的威胁 [15] 。在上世纪后 50 年以来的 14 次 厄尔尼诺年中，降水量低于平均值的年份有 12 次：1958、1963、1966、1972、1982、1983、 1986、1992、1993、1994、1996、1998。由以 上分析表明，厄尔尼诺及拉尼娜发生年份庆阳 地区旱灾发生严重，厄尔尼诺及拉尼娜与本区 旱灾有明显的相关性。 4. 结论: （1）本区气温在升高的同时降水总体减少， 暖干化趋势明显。 （2）全球气候变暖是影响本区干旱频繁发生 的主要原因。 （3）厄尔尼诺年、拉尼娜年是旱灾的多发期， 且厄尔尼诺、拉尼娜连续发生年份，旱情 严重。 参考文献 [1] 庆阳年鉴编纂委员会.庆阳年鉴，2001. [2] 周立华等.庆阳地区农村生态经济发展模式与政 策建议.干旱地区农业研究，2002，3. [3] j.t. houghton, l.g. meira filho, b.a. callander, et al. (eds.), climate change 1995: the science of climate change (cambridge university press, cambridge, 1996). [4] 高峰，孙成权，曲建升. 全球气候变化的新认识 ---ipcc,第三次气候评价报告第一工作组报告 概要.地球科学进展, 2001, 16(3):442-445 [5] 蔡运龙，s. marry.全球气候变化下中国农业的 脆弱性与适应对策. 地理学报， 1996（3）： 202-210. [6] 延军平等.跨世纪全球环境问题及行为对策.北 京：科学出版社，1999，50-51. [7] 丁一汇，戴晓苏.中国近百年的温度变化. 气象 学报, 1994, 20(2): 19-26. [8] w.w. kellogg, mankind´s impact on climate: the evolution of an awareness. climatic chang, 1987, 10(2), 113-136. [9] m.l. parry, m.s. swaminathan, effects of climate chang on food production (cambridge university press, cambridge, 1992). [10] 蔡运龙等.全球气候变化下中国农业的脆弱性 与适应对策. 地理学报，1996，3. [11] 谢南金. 中国西北干旱气候变化与预测研究. 第三卷，北京：气象出版社，2000，339-344 [12] 丁一汇，张锦，宋亚芳.天气和气候极端事件的 变化及其与全球变暖的关系. 气象，28（3）， 3. [13] 赵振国. 厄尔尼诺现象对北半球大气环流和中 国降水的影响.大气科学，1996，20（4），422-428. [14] 栗珂，刘耀武，杨文峰，徐小红，郑小华.el nino 事件的概率预测研究. 热带气象学报，2001， 05-25. [15] 王晓冬，李瑞英.20 世纪 50 年代以来陕甘宁地 区旱情分析.内蒙古师范大学学报（自然科学汉 文版），2008，07-15. 作者简介 王晓冬，（1969—），女，内蒙古包头人，内 蒙古科技大学副教授，硕士，研究方向为区域 资源与灾害。 与灾害。 published by atlantis press copyright: the authors 108 1. 数据与分析方法 2. 干旱情况日益加剧 3. 庆阳地区干旱发生的因素分析 4. 结论: << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <feff005400610074006f0020006e006100730074006100760065006e00ed00200070006f0075017e0069006a007400650020006b0020007600790074007600e101590065006e00ed00200064006f006b0075006d0065006e0074016f002000410064006f006200650020005000440046002c0020006b00740065007200e90020007300650020006e0065006a006c00e90070006500200068006f006400ed002000700072006f0020006b00760061006c00690074006e00ed0020007400690073006b00200061002000700072006500700072006500730073002e002000200056007900740076006f01590065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f007400650076015900ed007400200076002000700072006f006700720061006d0065006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076011b006a016100ed00630068002e> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <feff004b0069007600e1006c00f30020006d0069006e0151007300e9006701710020006e0079006f006d00640061006900200065006c0151006b00e90073007a00ed007401510020006e0079006f006d00740061007400e100730068006f007a0020006c006500670069006e006b00e1006200620020006d0065006700660065006c0065006c0151002000410064006f00620065002000500044004600200064006f006b0075006d0065006e00740075006d006f006b0061007400200065007a0065006b006b0065006c0020006100200062006500e1006c006c00ed007400e10073006f006b006b0061006c0020006b00e90073007a00ed0074006800650074002e0020002000410020006c00e90074007200650068006f007a006f00740074002000500044004600200064006f006b0075006d0065006e00740075006d006f006b00200061007a0020004100630072006f006200610074002000e9007300200061007a002000410064006f00620065002000520065006100640065007200200035002e0030002c0020007600610067007900200061007a002000610074007400f3006c0020006b00e9007301510062006200690020007600650072007a006900f3006b006b0061006c0020006e00790069007400680061007400f3006b0020006d00650067002e> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <feff005500740069006c0069007a006500200065007300730061007300200063006f006e00660069006700750072006100e700f50065007300200064006500200066006f0072006d00610020006100200063007200690061007200200064006f00630075006d0065006e0074006f0073002000410064006f0062006500200050004400460020006d00610069007300200061006400650071007500610064006f00730020007000610072006100200070007200e9002d0069006d0070007200650073007300f50065007300200064006500200061006c007400610020007100750061006c00690064006100640065002e0020004f007300200064006f00630075006d0065006e0074006f00730020005000440046002000630072006900610064006f007300200070006f00640065006d0020007300650072002000610062006500720074006f007300200063006f006d0020006f0020004100630072006f006200610074002000650020006f002000410064006f00620065002000520065006100640065007200200035002e0030002000650020007600650072007300f50065007300200070006f00730074006500720069006f007200650073002e> /rum <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> /rus <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> /sky <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> /slv <feff005400650020006e006100730074006100760069007400760065002000750070006f0072006100620069007400650020007a00610020007500730074007600610072006a0061006e006a006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b006900200073006f0020006e0061006a007000720069006d00650072006e0065006a016100690020007a00610020006b0061006b006f0076006f00730074006e006f0020007400690073006b0061006e006a00650020007300200070007200690070007200610076006f0020006e00610020007400690073006b002e00200020005500730074007600610072006a0065006e006500200064006f006b0075006d0065006e0074006500200050004400460020006a00650020006d006f0067006f010d00650020006f0064007000720065007400690020007a0020004100630072006f00620061007400200069006e002000410064006f00620065002000520065006100640065007200200035002e003000200069006e0020006e006f00760065006a01610069006d002e> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style received 25 june 2015 accepted 15 november 2015 the variable characteristics and response to climatic factors of the runoff in the downstream areas of the yellow river under the background of global change mei hong1,∗, dong wang2, wenhua zeng1, chenchen ma3, liang zhao1 1research center of ocean environment numerical simulation, institute of meteorology and oceanography, pla university of science and technology, china 2department of hydrosciences, school of earth sciences and engineering, nanjing university, china 3lianyungang meteorological bureau, lianyungang 222006, china abstract in this paper, the runoff of huayuankou and lijin hydrologic stations of the downstream areas of the yellow river from 1951 to 2012 were calculated and analyzed in order to explore the runoff variable characteristics and their response to climatic factors. the following conclusions can be drawn:(1)annual runoff of the river shows a clear downward trend as a whole, especially in the 1990s, but shifted to increase in the early 21st century.(2)runoff can be affected by climate change; climatic factors and runoff change in the same time or precedes the latter.(3)the response of runoff to climatic factors is not stable, in other words, changing with time; (4)the response of runoff to precipitation is more sensitive than that of temperature.(5)precipitation and temperature both contribute to the change of runoff, but the former is a positive contribution and the latter is opposite. keywords: downstream areas of the yellow river, runoff change, climate change, response, quantitative analysis 全球变化背景下黄河下游径流的变化特征及对气候因子的响应 洪梅 1，王栋 2，曾文华 1，马晨晨 3，赵亮 1 1. 解放军理工大学气象海洋学院军事海洋教研中心，江苏 南京 211101 2. 南京大学地球科学与工程学院水科学系，江苏 南京 210046 3. 连云港气象局，江苏 连云港 222006 摘要: 选取 1951-2012 年黄河流域下游两个水文站实测径流时间序列资料进行了统计分析和特征诊断。研究表明：1）近 60 年，黄河流域下游的径流量总体呈明显下降趋势， 21 世纪之后转呈增长的趋势； 2）径流年季变化峰型不同，径流量主要 集中在汛期，且 90 年代之后峰型较之前平缓；3）径流地域分布极不均匀，越靠近黄河下游越不均匀；4）流域内气候增暖 效应明显，降水总体呈减少趋势，气候对径流影响显著且呈现时变特征；5）降水对径流以正贡献为主，温度对径流以负贡 献为主，且降水量正贡献绝对值高于温度负贡献绝对值。研究工作旨在为应对气候变化和我国水资源利用提供决策参考。 关键词：黄河下游；径流变化；气候变化；响应；定量分析 ∗ corresponding author: flowerrainhm@126.com 资助项目：国家自然科学基金 （no.41276088; no.41375002; no.41306010) 资助 journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 257-263 published by atlantis press copyright: the authors 257 1. 引言 径流是一定时期内，流域中气候因素和下垫面 各种自然地理因素以及人类活动等综合作用的产物。 径流的形成过程是多种因素相互作用、相互联系的复 杂的自然现象，径流的变化不仅影响着人类社会系统 的安全，同时也影响着自然生态系统的健康(郝振纯 等，2006；徐东霞，2008)。21 世纪，全球经历着以 变暖为主要特征的变化过程。1880-2012 年全球平均 温度升高了 0.85°c，且这种增暖趋势仍将持续。在 全球气候变暖背景下，全球河川径流基本上呈现出下 降的趋势。ipcc 第四次评估报告（2007）预测，到 21 世纪中期，某些中纬度和热带干旱地区的河川年 平均径流将会减 10%-30%。 对黄河流域径流变化及其与气候、人类活动等因 素关系的研究（蓝永超等，2010）表明：气温升高和 降水减少导致黄河流域径流量总体呈减少趋势，径流 量变化既有波动性特征，又表现出一定的年际趋势， 且与气候因子变化关系密切。在全球变化背景下，黄 河流域是气候变暖的显著区域，空间上呈北高南低特 点，气温的地区分布特点是由南向北、由东向西逐渐 降低，尤以 20 世纪 80 年代以来，黄河流域气温增暖 最为显著。20 世纪 90 年代，黄河流域降水减少最为 明显，进入 21 世纪，黄河流域降水略有增加趋势 （刘吉峰和范旻昊，2010）。气候变化对黄河流域径 流变化规律及其气候响应形式也将发生变化，但相关 研究不多，且偏重于定性分析，对黄河中下游地区的 研究更为欠缺。 针对上述问题，本文拟基于黄河流域近 60a 的 实测径流资料和气象资料，运用统计学等方法分析径 流在不同时间尺度上的变化特征、探讨径流与气候影 响因子的内在关联、诊断气候影响因子对径流变化的 贡献和机理，为适应和应对气候变化，合理规划水资 源利用提供科学依据。 2. 径流的时间序列变化规律 河川径流是一种非平稳信号，并不存在真正意 义上的周期性，而时而以这种周期变化，时而以另一 周期变化，并且同一时期又包含各种时间尺度的周期 变化，在时域中存在多层次时间尺度结构和局部化特 征。采用 morlet 小波变换对两站的年平均径流进行了 周期性分析。morlet 小波变换的原理可以详见相关参 考文献（grinsted and moore，2004）。 黄河利津水文站是万里黄河上的最后一个水文 站，黄河水务的封笔之作。黄河利津水文站建于 1934 年 6 月，1937 年 11 月因抗站停测。解放后， 1950 年 1 月重新设站，现归黄委会山东水文水资源局 管理。黄河花园口水文站设立于 1938 年 7 月，上距 河源约 4700 公里，下距河口 770 公里，集水面积 73 万平方公里，占黄河流域总面积的 97%，是黄河最重 要的水沙控制站。这两个水文站的主要任务是：控制 黄河入海水、沙量，为黄河下游防洪、防凌、水资源 统一调度提供水情：研究和探索水文要素变化规律， 为黄河下游河道治理、水沙资源利用以及黄河三角洲 开发等搜集水文资料：对外承担各类水文测验项目， 地形、河道测量等。所以本文选用这两个站点来对其 径流和气候因子进行分析。 小波分析发现，近60年，花园口站存在2-4年、912年、27-30年的周期，强信号分别存在于：20世纪 50年代中期到70年代、60年代初期到70年代中期、80 年代。利津站存在2-4年、9-12年的周期，强信号分别 存在于：20世纪50年代末期到60年代末期、50年代初 期到80年代中期。两站均是2-4年为第一主周期。这 表明黄河下游径流存在较显著的年际和年代际变化特 征。 2.1. 年代际变化 通过计算，两站近 60 年（1951-2012 年）的径流 平均值分别为：374.5×108 m3、302.9×108 m3。表 1 给出了 1951-2012 年花园口站和利津站各年代径流的 平均值以及距平百分比，20 世纪 50 年代和 60 年代， 两站平均径流均大大高于各自多年平均值，属于丰水 年份，特别是 60 年代，两水文站年平均径流均达最 高值，分别为 505.9×108 m3、501.4×108 m3。70 年 代，两站平均径流均下降到多年平均值附近，80 年 代，花园口站平均径流量增加到高于多年平均值，而 利津站则有所减少，略低于多年平均值，到 90 年 代，利津站年平均径流达最低值，仅为 143.6×108 m3，与 60 年代相比减少了 71.4%。花园口站 21 世纪 的平均径流相比 60 年代减少了 50.3%，利津站 21 世 纪与 20 世纪 90 年代相比则略有上升。根据表 1 各年 代距平百分比并对应前文的丰枯划分标准以及黄河流 域气温统计分析(具体见下文)可知，黄河下游在气候 相对冷的 50 年代、60 年代处于丰水期，70 年代、80 年代处于平水期，而 90 年代以后则进入枯水期，与 气温增暖的时期相同，现在正处于枯水期。 mann-kendall 的秩次相关检验法可以检验序列的 突变特点，通过此方法进行的检验结果表明：上述水 文站的径流序列表现出突变特征，其中花园口站的径 流突变发生在 1985 年，利津站的径流突变发生在 1978 年，这表明上述水文站出现的径流减少可能是其 年代际尺度上的一种转折性或突变现象。 published by atlantis press copyright: the authors 258 表 1 黄河下游各年代径流量（r/108m3）及距平百分比(%) table 1 the decadal runoff and its anomaly in the downstream areas of the yellow river 测站 1951-1959 年 1960-1969 年 1970-1979 年 1980-1989 年 1990-1999 年 2000-2012 年 径流 量 距平 百分比 径流 量 距平 百分比 径流 量 距平 百分比 径流 量 距平 百分比 径流 量 距平 百分比 径流 量 距平 百分比 花园口 487.6 29.2 505.9 34.5 381.6 1.5 411.7 9.50 256.9 -31.7 251.3 -38.4 利 津 476.8 56.9 501.4 63.7 311.2 1.6 285.9 -6.7 143.6 -53.1 158.9 -54.0 2.2. 年际变化特征 近 60 年黄河下游的径流量总体呈现下降趋势。 这与上中游年降水量减少、用水量增加有关。但在 2002 年以后又呈现出有一定的转折上升趋势。花园口 站年径流最大值（861.4 亿立方米）出现在 1964 年， 最小值（142.6 亿立方米）出现在 1997 年，最大、最 小值之比为 6.04。利津站的最大、最小值出现年份与 花园口站一致，其极值比达 42.5。表明黄河流域下游 径流的年际振荡现象十分显著，且越靠近下游越剧 烈。 近 60 年的径流气候变化倾向率（表 2）表明： 夏季和秋季径流的递减率大，而冬季则是最小。且利 津站每一季节的减少量都比位于其上游的花园口站更 多。两站四季径流变化（图 1）分析表明：2002 年之 前，各季节的径流都呈现一致的下降趋势；从 2002 年左右开始，春季、夏季、冬季两站的径流均呈现上 升趋势，且夏季增加更为明显，秋季则呈减少趋势。 由此推知，21 世纪黄河下游径流的上升趋势主要是由 于夏季径流的增加导致，该现象与 20 世纪末期以来 全球气候变暖的趋势有所缓和是一致的（沈永平和王 国亚，2013）。 图 1 黄河花园口站、利津站 1951-2012 年平均径流量(亿立方米) fig.1 the annual runoff in huayuankou and lijin hydrologic stations in the yellow river basin during 1951 2012(108/m3) 表 2 黄河下游径流(108m3/10a)变化趋势 table 2 annual and seasonal changes of runoff in the downstream areas of the yellow river(108m3/10a) 测站 全年 春季 夏季 秋季 冬季 花园口 -51.3 -4.48 -18.9 -22.6 -1.9 利 津 -71.9 -13.6 -26 -30.2 -5.7 published by atlantis press copyright: the authors 259 利用累积距平方法对每个站年径流分析表明：两 站径流变化大致可分为三个阶段（图中灰色虚线划分 的三个区域）：相对多水期、相对平稳期和显著下降 期。两站第一阶段的时间段基本一致，均在 20 世纪 50 年代和 60 年代；第二阶段略有差别，花园口站是 在 70 年代和 80 年代，利津站则是在 70 年代到 80 年 代中期；第三阶段花园口站是在 90 年代之后，利津 站是在 80 年代中期以后。由上可知， 1951 年以来， 黄河流域下游径流大体均经历了丰-正常-枯 3 个变化 阶段，虽然两站丰枯变化的分界点有些差异，但自 20 世纪 80 年代末 90 年代初以来持续偏枯的事实是可 以确定的，上面分析得到的结论与采用丰枯等级划分 标准所得到的结果基本一致。 2.3. 年内变化年内分配规律 由于受降水、气温等气候要素季节性变化的影 响，导致径流的补给也出现明显的季节性变化，决定 了径流年内分配随季节的周期性变化和非均匀变化。 研究径流年内分配特征的方法主要包括：月、季、汛 期-非汛期径流量占年径流量的百分比等。本文采用 变差系数来分析黄河下游径流的年内分配特征。 图 2 表明：花园口站和利津站多年平均的径流年 内分配曲线（虚线）非常相似，均为“双峰型”，丰枯 季节变化尤为明显。两站均在 8 月有一个较大峰值， 但是春汛水量都不是很大，略不同的是，花园口站在 春汛期的峰值稍比利津站明显一点。冬季径流量明显 低于其他季节，仅占全年总量的 11.7%，全年径流量 基本上集中在 7-10 月份，占全年总量的 60.7%。 20 世纪 50 年代，两站径流年内分配呈现双峰结 构，但春汛期的峰型坡度比较缓和，而 8 月份的峰型 则很陡峭；60 年代两站均呈现多峰型，但春汛期的峰 型依旧很平缓，利津站在 8 月和 10 月均有一个峰 值；70 年代、80 年代花园口站峰型特征与 60 年代的 类似，只是 80 年代其中一个峰值提前了一个月，落 在 8 月；代利津站 70 年代也呈双峰型，80 年代则则 是多峰型；90 年代花园口站是 3 月和 8 月的双峰型， 利津站则是 8 月的单峰型；进入 21 世纪后，径流减 少明显，花园口站峰值分布在 3 月、6 月、10 月，利 津站峰值分布在 6 月、10 月，但 10 月的很平缓，相 比于前面大多数年代，两站夏季径流均有增加，且出 现峰值，这也进一步验证了上面近 10 年的径流增加 是由于夏季径流上升造成的结论。 另外，通过比较两站各年代径流峰型曲线与多年 平均值曲线可以看出：两站 90 年代 8 月份的峰值均 夹在 21 世纪后峰值之间，且气候变暖后的 90 年代及 以后峰型远不及以前陡高，明显低于多年平均值。 图 2 花园口站（左）、利津站（右）各年代径流年内分配曲线(108/m3) fig.2 monthly variations of the runoff for various decades at huayuankou(left) and lijin(right) hydrologic stations(108/m3) 径流变差系数是反映径流变化的参数之一。计算 得到近 60 年花园口站和利津站的月径流变差系数均 很大，说明各月的年际丰枯变化均比较剧烈；花园口 站的峰值在 6 月和 10 月，利津站的峰值则在 5 月和 10 月，这些月份的径流年际振荡尤为剧烈，花园口站 夏季和秋季的径流变化多、最不稳定，而利津站径流 则是春末和秋季（尤其是春季）最变化多端；花园口 站春季的径流变差系数最小，径流变化比较平缓，而 利津站径流变差系数则是冬季最小，且所有月份的值 都大于对应月份花园口站的值，说明利津站径流年际 丰枯变化更为剧烈。 3. 径流对气候因子的响应 气候因子对水循环过程的影响是复杂和多层次的, 气候系统通过降水、气温、日照、风、相对湿度等因 子直接或间接地影响着水循环过程，其中降水是最为 直接影响的因子，气温等则是间接影响。通过统计近 60 年的降水、温度资料发现：黄河流域多年平均降水 published by atlantis press copyright: the authors 260 全球变化背景下黄河下游径流的变化特征及对气候因子的响应 量为 507mm，其中夏季的平均降水量最大，达 293 mm，秋季次之，春季降水量比秋季略小，冬季则几 乎没有降水，平均降水量仅为 15mm；多年平均温度 呈明显上升趋势，其中冬季增温最为明显，即暖冬现 象。 3.1. 径流对降水的响应 降水是径流的直接来源。黄河流域年降水线性 趋势和 5 年滑动平均曲线表明，黄河流域年降水总体 呈弱的下降趋势,进入 21 世纪之后有增加的趋势。通 过计算各年代降水距平值，发现降水表现出阶段性特 点：50、60、70 年代的降水均是正距平，降水偏多， 其中 50 年代距平值为 27mm，降水量最多；80、90 年代均为负距平，降水偏少；进入 21 世纪之后，降 水有增多趋势，比多年平均值略低，这种降水年代际 分型特点与径流的变化规律基本吻合。 对两站各时段径流序列与同期降水进行相关分 析(表 3)发现：年平均降水量与两水文站年径流的相 关系数分别为 0.56 和 0.6，均超过了 0.01 的置信水 平。除冬季之外，降水量与径流量在各时段均存在正 相关关系（通过显著性检验），秋季降水与径流相关 性最好，夏季次之，且夏季相关系数与秋季的相差不 大，但是冬季降水和径流的相关性没有通过显著性检 验，这是因为，冬季的降水本来就很少，径流来源主 要靠地下水和上游的冰雪融水。总之，降水与径流的 关系在汛期是较好的，相反，在非汛期（特别是冬 季）的关系则较差。 表 3 各时段径流量与降水、温度的相关系数 table 3 annual and seasonal correlation coefficients between precipitation temperature and runoff 测站 全年 春季 夏季 秋季 冬季 降水 温度 降水 温度 降水 温度 降水 温度 降水 温度 花园口站 0.56 -0.64 0.34 0.48 0.5 0.17 -0.35 -0.41 -0.48 -0.31 利津站 0.6 -0.67 0.3 0.5 0.52 0.05 -0.49 -0.34 -0.47 -0.56 注：r0.05=0.25；r0.01=0.32. 3.2. 径流对温度的响应 在气候变化背景下，近 60 年来，黄河流域气温 总体呈上升趋势，增温速率约 0.29/10a(图 3a)，高于 全国平均气温的增温趋势。 从图 3(b)看出：20 世纪 80 年代以后，增温趋势 明显，90 年代以后这种增暖趋势大大超过了显著性水 平 0.05 临 界 线 ， 甚 至 超过 了 0.001 显 著 性 水 平 (u0.05=1.96,u0.001=2.56)。根据 uf 与 ub 交线的位置， 确定气温突变年份是 1988 年。 统计分析表明，黄河流域气温变化大致可分为两 个阶段：20 世纪 50 年代到 80 年代中期，黄河流域气 温比较平稳；80 年代中后期开始气温呈上升趋势，90 年代以后升温明显加快，达 0.36/10a。同样地，温度 距平百分率曲线的振荡比径流的弱得多，甚至比降水 的幅度还小。表 4 给出了黄河流域年平均气候倾向率 变化情况，各季温度与全年温度变化趋势一致，其中 冬季的升温幅度最大(0.45/10a)。冬季气温的大幅度升 高 势 必 对 黄 河 流 域 凌 情 造 成 影 响 （ 杨 特 群 等 ， 2009）。分析各时段温度与两站径流的相关系数(表 4) 发现：两站径流与温度在全年和各个季节均存在明显 的负相关关系（通过了显著性检验），说明黄河流域 温度升高对径流减少有着不可忽视的影响。 结合降水和温度两个因子和两站径流的变化可以 发现：气温突变增暖之后的 20 世纪 90 年代，黄河流 域的气候变得暖而干燥， 与此同时，径流也显著减 少，甚至出现断流，但 21 世纪之后，降水有所增 加，气候向着暖湿化方向发展，所以径流量也随着增 加。 4. 降水、气温因子对径流变化的贡献分析 为进一步定量分析两气候因子在径流变化中所占 的比重，本文运用 lu 等（2010）所使用的方法，利 用月径流资料和月气象资料，计算降水、气温两个主 要气候因子对花园口站和利津站径流变化的贡献，以 此来定量分析气候因子对径流变化的贡献。径流的自 然变化(s)主要受降水(p)和温度(t)的影响，故径流的 变化可以用下式来解释： s s s p t p t ∂ ∂ ∆ = ∆ + ∆ ∂ ∂ ，其 中 , s s p t ∂ ∂ ∂ ∂ 分别是降水和温度对径流变化的贡献率， 近似取为对应因子的多元回归系数。先利用气候因 子、径流的各月值计算出贡献率，并将此贡献率与各 因子标准差相乘，即可得到每个气候因子对径流年内 变化的贡献值，文中所指“正贡献”是指径流因气候因 子的变化而增加，“负贡献”是指径流因气候因子的变 化而减少。 published by atlantis press copyright: the authors 261 图 3 黄河流域年平均气温变化(a)和 mk 检验(b) fig.3 annual average variation of temperature in the reaches of the yellow river and mk test 表 4 黄河流域平均气温气候倾向率(/10a) table 4 tendency rates of average temperature in the reaches of yellow river(/10a) 因子 全年 春季 夏季 秋季 冬季 温度 0.29 0.32 0.14 0.45 0.24 分析计算结果发现（图 4）：降水和气温因子对 径流的变化有所不同，对这两个水文站而言，降水对 径流的贡献以正贡献为主，温度对径流的贡献以负贡 献为主，且都是在夏季和秋季最为明显，冬季和春季 图 4 黄河流域降水和温度对径流的贡献(108/m3) fig.4 monthly variations of the contributions of precipitation and temperature to runoff in the downstream areas of the yellow river (108/m3) (a) (b) published by atlantis press copyright: the authors 262 的贡献相对较弱。在降水对径流的正贡献期间，两 个站都表现出明显的双峰型特征：其中全年最高峰 值点都出现在 8 月，7-11 月期间花园口、利津两站 的正贡献分别达 14.6 亿立方米、17.7 亿立方米；次 峰值点均出现在 10 月，两站的正贡献分别为 11.1 亿 立方米、11.7 亿立方米；两站温度对径流的负贡献 都呈现单峰型特征，在 9 月份的负贡献绝对值均达 到最大，分别为-14.4 亿立方米和-14.8 亿立方米，刚 好对应两站降水正贡献的谷值，即降水对径流影响 最弱时温度对径流影响达最强。花园口、利津两站 全年降水对径流的正贡献平均值分别是 4.6 亿立方 米和 5.1 亿立方米，温度对径流的负贡献平均值分 别是-3.3 亿立方米和-4.4 亿立方米。 5. 结论与讨论 本文根据 1951-2012 年的气象和水文观测资 料，对黄河流域下游的径流变化规律、径流对降 水、气温因子的响应以及这种响应随时间的变化特 征和降水、气温因子对径流的贡献等问题进行了统 计分析和特征诊断，揭示了一些有意义的现象特 征，主要研究结果： (1)近 60 年来，黄河流域下游径流量总体呈现 下降趋势，表现出突变特征：20 世纪 60 年代以前处 于丰水期，70 年代和 80 年代处于平水期，90 年代 以后进入枯水期，但目前是有由枯转丰的趋势；径 流年际变化振荡剧烈，且呈现偏态分布，枯水年份 不仅发生概率高，且持续性好，最长连枯年份可达 21 年之久。 (2)黄河流域多雨期主要集中在 20 世纪 50~60 年 代以及 70 年代，80 年代和 90 年代均为少雨期，进 入 21 世纪之后降水量有增多的趋势；气温一直呈增 加趋势，并在 1988 年呈现突变特征，冬季升温率最 高，即暖冬现象明显；气候因子与径流量在有的时 段呈同期变化或者先于径流量变化。 (3) 无论是年径流还是各季节径流与同期降水、 气温因子的相关性都是不稳定的，且这种响应随时 间呈非平稳变化；径流对降水量变化的响应比对温 度变化的响应更敏感。 (4) 降水对径流以正贡献为主，温度对径流以负 贡献为主，且都是在夏季和秋季最明显，冬季和春 季的贡献相对较弱。 参考文献 grinsted j c and moore s. j, application of the cross wavelet transform and wavelet coherence to geophysical time series. nonlinear processes in geophysics 11(2004), 561566. 郝振纯，王加虎，李丽，等.气候变化对黄河源区水资源 的影响．冰川冻土 28(1) (2006), 1-7． 蓝永超，文军，赵国辉，等.黄河河源区径流对气候变化 的敏感性分析.冰川冻土 32(1) (2010),175—182. 刘吉峰，范旻昊.黄河流域气候变化特点及趋势.中国水利 学会 2010 学术年会论文集,2010. lu er,take e s,manoj. the relationships between climatic and hydrological changes in the upper mississippi river basin: a swat and multi-gcm study. journal of hydrometerorology, 11 (2010), 437-451. 徐东霞.近 50 年嫩江流域径流变化特征及其影响因素分析. 北京：中国科学院研究生院.2008． parry m，canziani o，palutikof j，et al. climate change 2007:impacts, adaptation and vulnerability – contribution of working groupⅱof the forth assessment report of the intergovernmental panel on climate change. (cambridge，uk：cambridge university, 2007) p.122. 沈永平，王国亚.ipcc 第一工作组第五次评估报告对全球 气候变化认知的最新科学要点. 冰川冻土, 35(5) (2013), 1068-1070. 杨特群，饶素秋，陈冬伶. 1951 年以来黄河流域气温和降 水的变化特点分析.人民黄河，31(10) (2009), 76-77. published by atlantis press copyright: the authors 263 http://xueshu.baidu.com/s?wd=author%3a%28%e9%99%88%e5%86%ac%e4%bc%b6%29%20%e9%bb%84%e6%b2%b3%e6%b0%b4%e5%88%a9%e5%a7%94%e5%91%98%e4%bc%9a%e6%b0%b4%e6%96%87%e5%b1%80&tn=se_baiduxueshu_c1gjeupa&ie=utf-8&subtitle=person 1. 引言 2. 径流的时间序列变化规律 2.1. 年代际变化 table 1 the decadal runoff and its anomaly in the downstream areas of the yellow river 2.2. 年际变化特征 2.3. 年内变化年内分配规律 3. 径流对气候因子的响应 3.1. 径流对降水的响应 table 3 annual and seasonal correlation coefficients between precipitation temperature and runoff 3.2. 径流对温度的响应 4. 降水、气温因子对径流变化的贡献分析 fig.3 annual average variation of temperature in the reaches of the yellow river and mk test 5. 结论与讨论 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <feff005500740069006c006900730065007a00200063006500730020006f007000740069006f006e00730020006100660069006e00200064006500200063007200e900650072002000640065007300200064006f00630075006d0065006e00740073002000410064006f00620065002000500044004600200070006f0075007200200075006e00650020007100750061006c0069007400e90020006400270069006d007000720065007300730069006f006e00200070007200e9007000720065007300730065002e0020004c0065007300200064006f00630075006d0065006e00740073002000500044004600200063007200e900e90073002000700065007500760065006e0074002000ea0074007200650020006f007500760065007200740073002000640061006e00730020004100630072006f006200610074002c002000610069006e00730069002000710075002700410064006f00620065002000520065006100640065007200200035002e0030002000650074002000760065007200730069006f006e007300200075006c007400e90072006900650075007200650073002e> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <feff005500740069006c0069007a007a006100720065002000710075006500730074006500200069006d0070006f007300740061007a0069006f006e00690020007000650072002000630072006500610072006500200064006f00630075006d0065006e00740069002000410064006f00620065002000500044004600200070006900f900200061006400610074007400690020006100200075006e00610020007000720065007300740061006d0070006100200064006900200061006c007400610020007100750061006c0069007400e0002e0020004900200064006f00630075006d0065006e007400690020005000440046002000630072006500610074006900200070006f00730073006f006e006f0020006500730073006500720065002000610070006500720074006900200063006f006e0020004100630072006f00620061007400200065002000410064006f00620065002000520065006100640065007200200035002e003000200065002000760065007200730069006f006e006900200073007500630063006500730073006900760065002e> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <feff005500740069006c0069007a006500200065007300730061007300200063006f006e00660069006700750072006100e700f50065007300200064006500200066006f0072006d00610020006100200063007200690061007200200064006f00630075006d0065006e0074006f0073002000410064006f0062006500200050004400460020006d00610069007300200061006400650071007500610064006f00730020007000610072006100200070007200e9002d0069006d0070007200650073007300f50065007300200064006500200061006c007400610020007100750061006c00690064006100640065002e0020004f007300200064006f00630075006d0065006e0074006f00730020005000440046002000630072006900610064006f007300200070006f00640065006d0020007300650072002000610062006500720074006f007300200063006f006d0020006f0020004100630072006f006200610074002000650020006f002000410064006f00620065002000520065006100640065007200200035002e0030002000650020007600650072007300f50065007300200070006f00730074006500720069006f007200650073002e> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word mr. ganga tuladhar's article_final in two columns disaster management system in nepal – policy issues and solutions gangalal tuladhar ph.d. fellow at the ehime university, center for disaster management informatics research 3 bunkyo-cho, matsuyama, 790-8577, ehime, japan email: gangalaltuladhar@yahoo.com abstract nepal is highly vulnerable to natural and human induced disasters. complex geology, variable climatic conditions, active tectonic processes, rapid urbanization, unplanned settlements, lack of public awareness, increasing population, weak economic condition and low literacy rate have made her vulnerable to disasters. therefore, advanced technology and adequate resources are needed to alleviate the disasters in nepal. training and other educational programmes are extremely needed to raise awareness. nepal needs proactive disaster management act and policy for disaster risk reduction activities. key words: disasters; awareness; policy; reduction; training. 1. background nepal is prone to various types of disasters like: floods, landslides, fire, epidemics, earthquakes, avalanche, windstorm, hailstorm, lightning, glacier lake outbursts (glofs), drought and so on due to the rugged and fragile geophysical structure, very high peaks, high angle of slopes, complex geology, variable climatic conditions, active tectonic processes, rapid urbanization, unplanned settlements, increasing population, weak economic condition and low literacy rate. apart from the above reasons, the lack of coordination among disaster management stakeholders, dearth of resources, shortage of technical manpower, low level of public awareness, difficult geo-physical condition, absence of modern technology are other issues that have been found as the major obstacles to manage the disasters in nepal (chhetri 2001). more importantly, the absence of a proactive disaster management act and policy has been found as the main impediments for the disaster governance in nepal.11 the himalayan region of nepal can be considered as one of the severest flood hazard zone of the world. heavy precipitation, high wetness and steepness of watersheds and river channels contribute to flood magnitudes.4 mainly, the middle hills are prone to landslides and the tarai to flood and fire. thus, flood, landslide and fire disasters are recurrent disasters in nepal. these disasters occur almost every year in one part of the country or the other causing heavy loss of lives and damage to physical properties.10 many government reports indicate that over the last 30 years floods, landslides, fires, avalanches and epidemics kill hundreds of people and destroy property worth millions of dollars. they also have a negative impact on the nation’s development program. in addition to the above factors, the losses from disasters are increasing in the absence of proactive disaster management policies, laws and preparedness and risk reduction programs. existing laws that deal with disasters do not address them in totality as the law is limited to immediate disaster response. as a result such focus as does occur is limited to disaster response and relief rather than complete approaches including planning, preparedness and recovery.11 the earthquake of 1934 a.d., 1980 a.d. 1988 a.d. the flood of july, 1993 a.d. and the flood of 2008 a.d. are the most devastating natural disasters in nepal which not only caused heavy losses of human lives and physical properties but also adversely affected the development process of the country as a whole.1 in this way, nepal stands at the top most disaster prone countries of the world. in a nut cell it is a hot spot of disasters. please see the table 1 below for the losses of journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 166-172 published by atlantis press copyright: the authors 166 administrateur texte tapé à la machine received 15 september 2012 administrateur texte tapé à la machine accepted 12 october 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine ganga lal tuladhar human lives and property losses due to various types of natural and human induced disasters in nepal from 1971 to 2011. table 1: occurrence and effect of disasters on human lives and property losses in nepal 1971-2011 year no of records no of deaths missing injury affected population destroyed houses damaged houses 1971 115 311 2 55 860 131 142 1972 115 173 37 88 902 771 86 1973 207 214 9 317 7846 1957 160 1974 231 507 43 725 19917 2615 859 1975 145 263 38 133 37612 2051 36 1976 234 304 143 93 12002 4957 448 1977 204 162 54 195 6822 1347 462 1978 304 471 401 90 15172 3132 75 1979 204 640 0 114 62025 2061 68 1980 216 423 67 506 8614 14348 13650 1981 178 258 201 434 45513 1246 1004 1982 164 683 19 24 5159 1039 37 1983 171 492 15 122 4197 1384 1207 1984 368 1091 24 611 12418 2568 485 1985 172 229 7 77 5160 1475 63 1986 113 289 0 34 5163 1160 21 1987 121 122 0 68 13548 1041 6115 1988 337 1327 15 8205 3766 23202 41182 1989 303 352 5 1419 20087 4813 1377 1990 207 512 35 4107 7995 1209 1366 1991 415 1097 26 179 53441 1392 202 1992 410 998 63 29 23383 6225 79 1993 889 1812 82 304 653336 21249 21673 1994 427 1175 46 1253 183848 3175 517 1995 416 1158 35 1484 696515 9685 15898 1996 369 1147 98 1579 623216 19638 13923 1997 561 1331 4 944 217599 4549 1046 1998 429 1154 16 304 508900 15978 477 1999 481 1409 40 422 65159 4046 697 2000 664 708 46 342 29770 3038 1810 2001 1211 1902 80 3465 73652 6308 2350 2002 1170 899 105 12081 398748 14059 5479 2003 931 971 86 3438 565002 1974 761 2004 1020 1098 60 224 331929 1617 3339 2005 477 333 101 164 110647 1449 539 2006 524 596 101 5859 9265 1927 8512 2007 877 636 891 4694 86876 9456 1466 2008 1526 944 129 1266 483465 16029 3292 2009 1584 1944 140 1303 377792 3760 9124 2010 1555 839 146 473 143048 4263 8453 published by atlantis press copyright: the authors 167 disaster management system in nepal – policy issues and solutions relief and treatment sub-committee regional disaster relief committee supply, shelter and rehabilitation sub-committee district disaster relief committee local disaster relief committee central disaster relief committee 2011 1517 859 143 991 376073 7322 7878 total 21562 31833 3553 58215 6306442 229646 176358 source: desinventar, 2012 ; url: www.desinventar.net, www.nset.org.np; for more detail: nset 2. structure of disaster management system in nepal natural disaster relief act (ndra), 1982 was enacted in 1982 a.d. before the enactment of the act, disaster management activities were carried out in an unorganized way. therefore, to perform the disaster management activities in an organized way, ndra has constituted an organizational structure or system as shown below in figure 1:6 figure 1 the above figure is adopted from the book of dr. meen poudyal chhetri.10 according to the ndr act, 1982 central natural disaster relief committee (cndrc) has been constituted under the chairmanship of the home minister in order to formulate and implement the policies and programs relating to the natural disaster relief work and to undertake other necessary measures related thereof. moreover, the central committee prepares specific norms of relief assistance to be given to the disaster victims of the affected area in cash and/or in kind through the district natural disaster relief committee (dndrc).6 the central committee may constitute relief and treatment sub-committee (rtsc) and supply, shelter and rehabilitation sub-committee (ssrsc) which provides necessary advice and suggestions to the central committee, helps to execute policies and directives of the central committee and operates effectively the rescue, relief and rehabilitation works during mega disasters.2 published by atlantis press copyright: the authors 168 ganga lal tuladhar 3. the role of the focal agency the ministry of home affairs is the focal agency for disaster management in nepal which is guided and directed by the central disaster relief committee headed by the home minister in the matters of disaster management. the ministry formulates and implements national disaster policies, plans, programs and policies. primarily, the ministry and its partners are focused to carryout rescue operations and provide relief materials to the disaster affected population during disasters. being the nodal agency, the ministry of home affairs has the responsibilities of release and distribution of funds and resources to the affected population and area through the regional and district administration offices. the ministry of home affairs has the responsibilities to coordinate the activities relating to disaster preparedness, mitigation, reconstruction and rehabilitation works with other disaster management related agencies. it has its network throughout the country in the form of the regional administration office, district administration office and the district police office which are the field offices to carry out rescue and relief works during disasters.2 in the capacity of the focal agency, the ministry carries out various types of public awareness raising programs on disaster management. a national emergency operation center (neoc) is into operation in the premises of the ministry since december 2010. neoc sends informative messages through mass media in order to make the people aware of the natural disasters. the function of disaster data collection and dissemination are also being done by the neoc. the ministry has central database system which publishes annual report, maps, booklets, pamphlets, posters and other materials for disaster information for professionals and concerned agencies and people. however, it is realized by the disaster stakeholders and general public that the disaster prevention, preparedness and rehabilitation works are often ignored or underserved. while the casualty and property losses are in increasing trend.8 4. preparedness strategy of the government after the devastating floods and landslide disaster of 1993 july in which 1336 people lost their lives and 85,451 families were affected the government of nepal has been serious in the management of natural disasters in the country. therefore, the government prepared a national action plan on disaster management in 1996 which has specified priority activities to be undertaken in the field of disaster management to be implemented by the concerned agencies.5 the formulation and implementation of national strategy for disaster risk management (nsdrm) in 2009 is another step to mainstream disaster risk reduction activities in nepal.7 with regard to disaster management, both (the plan and strategy) depict in matrix form the priority item groups and activities together with the responsible executing agency and the cooperating agency in national hazard assessment, awareness raising, training, information system, land use plan, disaster reduction policy, regional and sub-regional cooperation between and among countries and establishment of documentation center on disasters.11 therefore, the government aims to:  reduce the loss of life and property;  address the sufferings of the people;  effect land use planning;  prepare and develop hazard maps of the disaster prone areas;  make the people aware of natural disasters;  make coordination among different agencies involved in disaster management;  mobilize the non-governmental sector in rescue and relief works as well as awareness raising programs;  mobilize internal and external resources for rehabilitation and reconstruction;  provide training at the grassroots level for the management of disasters;  build capacity on disaster management. the above are the main strategies that the government has taken after the lessons learnt from the flood and landslide disaster of 1993. 5. problems9 the disaster risk reduction plans and policies lack adequate attention in the field of pre-disaster works. large parts of the population in the country are still not aware of natural disasters. though, the country is suffering from a number of disasters every year, the studies regarding to identify most published by atlantis press copyright: the authors 169 disaster management system in nepal – policy issues and solutions vulnerable places and types of disasters are still inadequate. prioritized hazards and their preparedness and mitigation efforts are completely lacking. similarly, the pre, during and post disaster activities for different hazards are yet to be established. in addition to the above drawbacks, lack of proper policy and legal environment is the biggest impediment for an effective disaster management system in nepal. other major problems can be pin pointed as following: (i) limited resources (ii) remote, rural and difficult geophysical condition of the country (iii) inadequacy of infrastructure facilities (iv) low literacy rate, lack of public awareness and the misconception of the people (some people especially illiterate rural people think that the natural disasters are the act of god. they seldom know that preventive measures can reduce the impact of natural disasters). (v) absence of modern technology including early warning systems (vi) ineffective coordination (vii) reactive approach (viii) there is a small number of ngos willing to work in disaster preparedness, mitigation and rehabilitation works (ix) rapid urbanization and unplanned settlements. 6. solutions to solve the problems9 in view of the above problems, in brief, the following solutions can be suggested to address them: (i) avail adequate funds and resources (ii) infrastructure development (iii) mass education, literacy and awareness campaign in order to educate and aware the people. (iv) advancement in technology including the development of early warning systems (v) effective coordination among disaster related agencies (vi) the act should be amended or revisited in order to make it proactive (vii) a separate agency should be constituted to address the whole cycle of disaster preparedness, mitigation, response and rehabilitation works (viii) ngos need to be motivated to work in the disaster affected remote areas (ix) planned settlements are needed. building code should be strictly implemented. we know that natural disasters cannot be stopped. however, the extent of disasters can be reduced, if preventive measures be taken in due time for which pragmatic government policies and public awareness raising programs are of utmost importance. for this the government and the community should work in a cooperative manner. moreover, the effects of natural disasters in various parts of the world have shown the necessity to intensify international cooperation for disaster mitigation. above all, in the time of natural disasters local people are the main responders. usually local police, local administration and the central government come into the forefront bit later. often it is bit late after international communities appear at the disaster site. therefore, it is necessary to strengthen local communities. nevertheless, international communities should be involved in disaster preparedness and reconstruction works.10 7. the bill keeping in view the need and importance of a proactive law for the disaster management system in nepal, the nepal centre for disaster management (ncdm)1 drafted a new disaster management bill and policy in the year 2007, and submitted them to the government for consideration. the bill and policy drafted by the ncdm was an outcome of rigorous work done by its members who organized a series of meetings, workshops and interactions with disaster management stakeholders from central to local level in nepal in the course of finalizing the draft.3 1 nepal centre for disaster management (ncdm) is a specialized agency in disaster management policy and education in nepal. it is represented by high level professionals from academia, bureaucracy, media etc. published by atlantis press copyright: the authors 170 ganga lal tuladhar the bill and policy were revised by the relevant government agencies since then. after a series campaigns and pressures from the civil society; particularly from ncdm and dpnet-nepal, the bill was forwarded to the constituent assembly in april 2012. in the meantime, the constituent assembly was dissolved on 27 may 2012. as a result, the bill could not become an act and now it has to wait for the future constituent assembly or parliament. we hope that the bill will be approved soon. actually, the bill when it becomes the act and starts functioning will open avenue for effective disaster management system in nepal. 8. way forward at present, lack of proactive policy and legal instruments are the biggest impediments. such condition hinders replication of the successful cases to other places although there are high potentials for the same. absence of organizational outfit at the highest level that could be tasked to provide intellectual and at ministerial leadership is seriously noted in nepal. in fact, the country could learn from the experiences of the other countries like china, japan, india, bangladesh and many other nations where the prime minister or other top level authority lead or guide the disaster risk reduction activities. thus, improving on the role of leadership and the creation of the national disaster risk management council or disaster management authority, backed up by appropriate legislative instrument, seems to be the priority agenda for nepal. 9. conclusions it is a big concern among the disaster affected population and also the disaster management professionals that the drr activities are limited in carrying out immediate rescue and relief works on an ad hoc manner. hence, there is the need as well demand for vigorous change in the existing legal system. the start point is the enactment of the proposed disaster management bill which is supposed to pay due attention to disaster preparedness, mitigation and rehabilitation works and reduce the heavy loss of lives and properties by adopting proactive actions. above mentioned solutions could help as the tool to solve the existing problems to a greater extent. disaster mitigation, early warning system, insurance system, emergency rescue and relief operation, rehabilitation and recovery plans should involve activities such as training, post-disaster evaluation, monitoring of relief works, review and cooperation and coordination of central, district and local level preparedness and research works. recovery planning should involve immediate recovery as well as long-term programs. political determination, effective and efficient policy implementation and people's participation is of great importance for functional disaster governance. finally, emergency response planning and capacity enhancement, strengthening policy and legal environment, efficient and effective reconstruction and rehabilitation works are particularly necessary. to attain this goal a separate ministry or council or authority for disaster management in nepal is extremely desirable for effective disaster prevention, preparedness, response and recovery. published by atlantis press copyright: the authors 171 disaster management system in nepal – policy issues and solutions references 1. "1993 flood damage assessment, general infrastructure, final report main text" (1993), gon-undp, smec/cemat. 2. "disaster management in nepal a profile (1994)", gon, ministry of home affairs, kathmandu, nepal. 3. "draft disaster management bill and policy" prepared and submitted to the government by nepal center for disaster management (ncdm) in 2007. 4. khanal, n.r. (1998), "water induced disaster in nepal", a paper presented in an international seminar on water induced disaster held from 4 to 6 november, 1998 in lalitpur, nepal. 5. "national action plan on disaster management in nepal (1996)", gon, ministry of home affairs, kathmandu, nepal. 6. "natural calamity (relief) act, 1982", the government of nepal. 7. “national strategy for disaster risk management, 2009”, the government of nepal. 8. poudyal chhetri, m.b.p. (2000), "flood hazards in nepal", a paper presented in the first technical meeting on mitigation, management and control of floods in south asia, 24-25 january 2000, new delhi, india 9. poudyal chhetri, m.b. (1999), "disaster management in nepal: problems and solutions " an article published in the book on natural disaster management, edited by jon ingleton, tudor rose, holdings limited, leicester, england. 10. poudyal chhetri, m.b. and bhattarai, d. (2001), "mitigation and management of floods in nepal," a book published by, gon, ministry of home affairs, kathmandu, nepal. 11. poudyal chhetri, m.b. (2011), " disaster risk reduction: policy implications for nepal, australia and beyond,” post doctorate dissertation completed at the queensland university of technology, brisbane, australia. published by atlantis press copyright: the authors 172 2 received 14 october 2016 accepted 5 november 2016 public risk perception and risk communication of typhoon disaster zhihai shang, liping li department of geography, lingnan normal university, zhanjiang 524048, china abstract public risk perception is the basis for effective natural disaster risk communication, and the latter can strongly influence the former. take typhoon disaster as an example, this paper constructs the theory model of 12 factors affecting risk perception from three dimensions by public individual characteristics, risk situational characteristics, risk communication efficiency, which influence the input, output and transmission of risk information. then the theory was applied in typhoon disaster in zhanjiang city, and it comes out that the theory is reliable. finally, the contents of typhoon disaster risk perception capacity construction were given in this paper as risk communication capacity building, special platform construction and cultural environment of risk communication, which is the only way for effective risk communication. keywords: risk perception; risk communication; typhoon disaster 公众台风灾害风险认知与风险沟通探讨 尚志海，李丽苹 岭南师范学院地理系，广东 湛江 524048 摘 要：公众灾害风险认知是自然灾害风险沟通的基础，风险沟通又强烈地影响着风险认知。以台风灾害 为例，首先构建了自然灾害风险认知影响因素的理论体系，从风险信息的输入、输出和传播三个方面，提 出公众个体属性、风险情景特征及风险沟通效率 3 个维度的 12 个影响因素。接着，将灾害风险认知影响因 素的理论应用于湛江市台风灾害实践中加以检验，研究结果验证了理论的可靠性。最后，提出有效风险沟 通是提升公众风险认知能力的唯一路径，通过政府风险沟通能力、风险沟通专门平台和风险沟通文化环境 建设来实现有效风险沟通。 关键词：风险认知；风险沟通；台风灾害 《2015~2030 年仙台减少灾害风险框架》提出， 为实现仙台框架的预期成果和目标，第一个优先领 域是理解灾害风险，其中就包括风险沟通与交流 [1] 。 自然灾害风险沟通是灾害风险利益相关者之间的交 流工具，风险沟通的基础是了解公众风险认知 [2,3] 。 以往的经验显示，无论风险评估技术和结果多精细， 它本身也不大可能引发公众主动参与减灾行动 [4] 。 一般来说，高水平的风险认知才会激发人们采取风 险管理措施 [5,6] 。自然灾害风险沟通对公众风险认知 有着直接的影响，风险沟通方式不当，很容易导致 公众产生认知上的偏差 [7] 。 从 20 世纪 60 年代开始，西方学者就进行了灾 害风险认知研究 [11] ，探讨了公众风险认知的影响因 素 [8-10] ，并进行了滑坡、洪水、地震、台风灾害风 journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 213-220 published by atlantis press copyright: the authors 213 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 险认知与风险沟通的理论与实践探索 [12-16] 。汶川地 震发生后，国内学者才开始关注灾害风险认知与风 险沟通研究，但是研究进展不大 [17] ，分别以 “灾害 认知”、“灾害感知”、“灾害沟通”为检索词， 在中国期刊网中对论文篇名进行模糊检索，结果显 示到 2016 年 8 月 31 日为止上述三类期刊论文数量 仅分别为 58 篇、26 篇和 16 篇。由此可见，国内灾 害风险认知与风险沟通研究亟待加强，尤其是台风 等极端灾害事件的风险认知研究。 台风是中国沿海地区普遍发生且破坏力极强的 自然灾害，本文提出了公众灾害风险认知及其影响 因素的理论模型，并将其应用于湛江市居民台风灾 害风险认知研究中，最后提出了公众台风灾害风险 认知能力提升，即有效风险沟通的建设内容。 1. 研究方法与数据来源 在理论体系构建上，本文采用历史文献法，综 合运用灾害学、心理学、社会学等多学科的理论， 诠释台风灾害风险认知及其影响因素之间的关系。 在灾害实证分析上，本文采用问卷调查法，对湛江 市进行实地调研获得真实数据，并对调查数据进行 统计分析，从而得到相应的研究结果。 1.1 调查问卷设计 本问卷的设计包括两个部分：个人基本信息； 台风灾害风险认知和风险沟通情况。第一部分具体 包括被调查者的性别、年龄、居住地点、文化程度、 从事职业；第二部分包括被调查区域台风灾害的易 发性与危害性，台风灾害对居民的心理伤害，灾害 风险信息获取渠道，居民对灾害预警信息的响应情 况等。正式调研之前，笔者进行了预调查，并根据 调查结果对正式调查问卷进行了完善，使其更加切 合本次调查的目的，可以更真实地了解湛江居民对 台风灾害的认知情况。 1.2 调查地点选取 本文的研究区域为湛江市，台风是湛江市典型 灾害，每年平均会有 1~2 个台风登陆，而且湛江市 是广东沿海地区台风灾害风险最大的城市 [18] 。例如， 2015 年 10 月 4 日第 22 号台风“彩虹”以强台风级 在湛江市坡头区沿海登陆，登陆时中心最大风力达 16 级，是 1949 年以来十月登陆广东最强的台风。 在“彩虹”肆虐下，湛江市遭受了极大损失，直接 经济损失约为 210 亿元。由于台风“彩虹”给中国 南部造成严重影响，在美国檀香山举行的第 48 届台 风委员会会议上，中国提出将“彩虹”除名。 湛江市现辖雷州市、吴川市、廉江市 3 个县级 市，徐闻县、遂溪县 2 个县和赤坎区、霞山区、坡 头区、麻章区与湛江市经济技术开发区，其中雷州 市和徐闻县是遭受台风灾害侵袭最频繁的区域。根 据湛江市台风灾害的空间特征及历史灾情，本次调 查选取了“彩虹”重灾区坡头区、遂溪县及台风易 发区雷州市和徐闻县作为样本点，同时以随机抽样 方法调查这四个区域 235 名居民对台风灾害的认知 情况，各个样本点的调查数量基本一致，所占比例 都在 25%左右。 1.3 基本情况统计 本次调查是在台风“彩虹”发生 3 个多月后， 于 2016 年 1 月 10 日至 26 日实施。共发放问卷 235 份，回收问卷 235 份，其中无效问卷 24 份，最终有 效问卷 211 份，有效回收率 89.8%。样本数据的基 本情况统计是指对有效调查问卷的全部被调查者的 基本信息数据进行一个统计性表述，了解样本的大 致状况。调查样本中男女比例分别为 47.9% 和 52.1%，相差不大；按照联合国人口年龄划分标准， 样本以 18~44 岁的青年人为主，占了总数的 60.7%； 绝大多数具有中等及其以上的教育背景，小学及以 下学历的被调查者只占 9%；从职业构成来看，在校 学生和农林牧渔业生产人员的比例最高，占了总数 的 61.1%。 published by atlantis press copyright: the authors 214 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 2. 公众灾害风险认知及其影响因素分析 2.1 理论模型 风险认知是人们的主观心理活动，它是用来描 述人们对风险的态度和直觉判断的概念，广义上也 包括人们对风险的一般评估和反应 [8] ，同时也反映 了灾害利益相关者之间风险沟通的效果。不同的人 对灾害的风险认知是有差异的。slovic 认为风险认 知的影响因素有两个基本维度：忧虑性维度和未知 风险维度 [7] 。谢晓非等认为至少有 15 种风险认知影 响因素：自愿性、可控性、熟悉性、公正性、利益、 易理解性、不确定性、恐惧、对机构的信任、可逆 性、个人利害关系、伦理道德、自然或人为风险、 受害者特征、潜在的伤害程度 [9] 。在前人研究的基 础上 [19] ，根据公众感知和评估灾害风险信息的过程， 本文认为灾害风险认知的影响因素可以概括为公众 个体属性、风险情景特征、风险沟通效率三个维度， 分别对应着“谁来感知、感知什么和如何感知”。 公众个体属性影响着风险信息的输入，风险情景特 征影响着风险信息的输出，风险沟通效率作用于风 险信息的传输过程，三者共同作用塑造着公众灾害 风险认知的特征（表 1）。 表 1. 公众灾害风险认知的影响因素 三个维度 具体因素 公众个体属性 自然属性：性别、年龄 社会属性：文化程度、从事职业 风险情景特征 灾害易发性、灾害距离感、 事件危害性、灾害恐怖感 风险沟通效率 信息清晰度、信息可信性、 信息可达性、公众响应度 公众个体属性包括公众个体的自然属性和社会 属性，自然属性包括性别、年龄，两者对灾害风险 认知的影响比较复杂，虽然学者们都认为它们之间 有联系，但是其复杂程度也不容忽视。在不同性别 人群的灾害风险认知差异研究上，目前有三种不同 观点：一是女性的灾害认知程度不如男性 [20] ；二是 女性的灾害认知能力强于男性 [21] ；三是男女在灾害 风险的认知上基本相同 [22] 。而从年龄特征来看，灾 害认知能力最强的年龄段也有差别，包括：20～30 岁、19~25岁以及40岁以上等[20-22]。公众个体的社会 属性包括文化程度、从事职业。社会属性比自然属 性与灾害风险认知的关系更清晰，一般来说，公众 文化程度越高，其对灾害风险情景的评估越理性， 对灾害风险信息的理解越清晰，获取灾害风险信息 的渠道越广泛，信息越能被准确、及时地传递给居 民本人，风险沟通的参与度和话语权越大。公众从 事的职业与其经济水平有直接联系，经济水平越高， 灾害风险认知意愿和能力越强。 风险情景特征包括灾害易发性、灾害距离感、 事件危害性、灾害恐怖感，这四个因素分别从灾害 的时、空、强三个特征影响公众灾害风险认知，灾 害的强度包括客观特征和主观感受两个方面。首先， 灾害易发性影响着公众对灾害的熟悉程度，风险认 知程度会随灾害熟悉程度不同而不同。第二，灾害 距离公众的远近，包括空间距离和时间距离也会影 响公众灾害风险认知；公众对灾害的风险认知程度 会随时空距离增加而降低。第三，人们往往更看重 灾害事件的危害性，而不是其发生频率 [23] 。死亡与 失踪人数、直接经济损失集中显示了灾害的危害性， 因此极大地影响着公众的风险认知。最后，灾害恐 怖感影响着公众在面临灾害时的情绪和心理，灾害 事件所导致的恐慌，其危害程度可能远远大于灾害 本身 [9] 。灾害越恐怖，越能引起公众的风险认知， 这正是地震比干旱更能引发公众关注的原因。 风险沟通效率包括信息清晰度、信息可信性、 信息可达性、公众响应度。灾害风险信息本身的性 质与传播环境都会影响风险认知，体现为风险信息 是否清晰、是否可信与可达。信息的清晰度和可信 性是公众接收及关注信息的基础，人们不会在模糊、 可疑的信息上浪费时间和精力。灾害风险信息的可 published by atlantis press copyright: the authors 215 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 达性要通过风险沟通渠道来实现。只有通过传统媒 体和新媒体帮助公众建立理性的风险认知，才能将 灾害的负面影响降低到可接受水平 [24] 。此外，灾害 风险沟通效率的构建依赖于沟通各方的互动过程， 除了需要政府、专家、媒体传播信息外，还需要公 众积极响应，才能实现有效沟通，风险沟通又强烈 地影响着风险认知。 2.2 实证分析 2.1.1 公众个体属性与风险认知的关系 （1）自然属性 不同性别居民的台风灾害风险认知水平不同， 具体来说，湛江市居民中男性对防灾减灾实践技能 的掌握程度好于女性；但是就台风灾害理论知识的 了解程度来说，女性强于男性。另外，不同年龄段 的居民，其台风灾害风险认知表现出不同特征。调 查显示：湛江市居民中对台风灾害知识掌握最好的 是 30~44 岁年龄段的人群，而对防灾技能掌握最好 的则是 45~59 岁年龄段的人群。 （2）社会属性 随居民文化程度的不断提高，湛江市居民的台 风灾害风险认知水平及防灾减灾能力也不断增强， 大专及以上学历人员对台风灾害的认知水平最强， 初中或高中学历人员对台风灾害的认知水平较强， 小学及以下学历人员对台风灾害的认知水平最弱。 在不同从业人员中，对台风灾害的认知水平及其防 灾减灾能力差异较大，以国家机关、企业单位职工 认知水平最高。 通过上述分析，可以发现：在影响湛江市居民 台风灾害风险认知的公众个体属性中，性别虽有影 响，但是男女在不同风险认知能力上各有千秋，性 别对灾害风险认知的影响要具体问题具体分析； 30~60 岁中青年的风险认知水平最强，随着年龄的 不断增长，台风防灾减灾经验也是不断丰富的；湛 江市居民的公众灾害风险认知水平与受教育水平成 正相关，也与从事职业有一定关系。 2.2.2 风险情景特征与风险认知的关系 （1）灾害易发性 台风是湛江市最频繁的自然灾害，居民早已对 其习以为常，并积累了一定的抗台经验。调查结果 显示：居民对台风登陆前的防范措施是基本一致的， 81.0%的居民会提前准备好食物、饮用水、手电筒、 蜡烛和常用药品；75.8%的居民会修检房屋和加固门 窗；58.3%的居民会检查家中电路、煤气等设施的安 全情况。正是因为灾害易发性增强了公众的风险认 知水平，所以他们才会有相应的防灾减灾行为。 （2）灾害距离感 本文的调查对象都位于湛江市，与台风的空间 和时间距离都很近，居民对台风灾害风险的感受都 很深。尤其是 80.6%的居民都认为 2015 年的“彩 虹”对他们的影响大，其中又有超过一半的人认为 台风对其生活影响非常大，58.3%的居民受到的最大 影响是停电。这一情况也与事实一致，彩虹发生期 间湛江市区几乎全城停电停水，通信中断。刚刚发 生在公众身边的台风灾害，无疑会给他们带来巨大 的冲击，会使他们对台风灾害风险有更深的认知， 大多数人会及时改善风险应对行为。 （3）事件危害性 台风是湛江市最严重的自然灾害，调查显示湛 江居民认为台风灾害造成的影响主要以财产破坏和 出行不便为主，经济损失是对居民最大的危害。湛 江市每年平均会有 1~2 个台风登陆，由于台风的不 确定性和多变性，难以把握其规律，每次台风都造 成重大损失。但即使台风灾害经常发生，如果其给 居民造成的不利影响很小甚至可以忽略不计，那么 台风作为一种自然事件也不足以引发人们的关注。 这也许可以用来解释为什么有 55.5%的湛江市居民 对台风灾害信息的关注度较低，因为遭受直接损失 published by atlantis press copyright: the authors 216 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 的人毕竟是少数。 （4）灾害恐怖感 根据问卷调查和现场访谈显示，台风造成的居 民心理情绪的变化主要表现为居民的负面情绪及认 知。每次台风灾害后，居民心理都会产生不同程度 的变化，当谈及台风灾害时，他们都表现出伤心、 悲愤和害怕的情绪，其中认为台风会造成心理恐慌 的人群比例高达 60.7%。湛江市虽然也面临其他自 然灾害的影响，比如干旱、暴雨，但与台风相比， 其恐怖感偏低，因此不会产生像台风一样的强的负 面心理，也就没有引发公众足够的灾害风险认知。 湛江市台风灾害频发，居民与台风灾害的时空 距离都近在咫尺，且灾害危害严重，民众心理恐慌 程度高，因此上述影响因素综合在一起，极大地影 响着公众灾害风险认知。 2.2.3 风险沟通效率与风险认知的关系 （1）信息清晰度、可信性和可达性 与其他自然灾害相比，台风灾害预警信息的清 晰度和可信性都较高，一般是由政府应急办、三防 办、气象部门联合发布，例如台风“彩虹”的预警 信息非常详细，中央气象台 10 月 3 日 23 时 30 分发 布台风红色预警：今年第 22 号台风“彩虹”已于今 天（3 日）晚上 11 点钟由台风级加强为强台风级， 晚上 11 点钟其中心位于海南省文昌市以东大约 210 公里的南海北部海面上。预计，“彩虹”将以每小时 20-25 公里的速度向西偏北方向移动，强度继续加 强，最强可达 15-16 级，并将于 4 日上午在海南文 昌到广东湛江一带沿海登陆。此外，预警信息还包 括大风预报和降雨预报。 从上述信息可以发现，台风灾害风险的信息越 清晰，比如灾害发生时间、空间、强度的信息越明 确，公众越能及时做出响应；并且随着台风的登陆， 台风灾害信息会更清晰明确，这也会增强公众对灾 害风险的认知。如湛江市对台风彩虹的预警级别从 橙色上升到黄色，直至红色预警，通过预警级别的 提升，也从侧面反映了风险信息的清晰度和可信性， 因此公众会根据台风级别的提升，加深对台风灾害 风险的认知，从而采取适当的防灾减灾措施。 在台风灾害风险信息的可达性方面，目前政府 相关部门主要利用电视、广播、互联网、手机短信 等渠道向社会发布预警信号。在这些信息渠道中， 被调查者中 85.3%以上的居民是通过手机了解风险 信息，之后是电视广播、电脑、亲友告知、社区宣 传、报纸杂志（表 2）。从这些数据可以发现新媒体 在台风灾害风险沟通中的重要地位，以手机短信、 微博、微信为代表的新媒体成为了人们获取风险信 息的最主要渠道，其开放性、即时性、交互性特征， 增强了普通居民对灾害风险信息的获取与交换。 （2）台风风险沟通中的公众响应度 即使可靠的台风灾害风险信息表达清晰，并且 及时传递到了公众手中，但是如果公众不积极对灾 害风险预警做出响应，风险信息也只是单向的传递 而已。公众参与风险沟通的响应度主要表现在台风 来临前的准备和防护措施、台风来袭时的做法与表 现。调查结果显示：91.5%的居民表示接到台风预警 信号后，就会及时做好防灾减灾准备。公众对灾害 风险信息的响应度越高，说明公众的风险认知能力 越强，越能积极应对灾害风险。 因此，风险沟通效率的高低强烈地影响着公众 表 2. 公众对台风灾害风险沟通渠道的选择情况 排序 1 2 3 4 5 6 沟通渠道 手机 电视广播 电脑 亲友告知 社区宣传 报纸杂志 选择比例 85.3% 68.70% 66.80% 28.00% 25.10% 11.80% published by atlantis press copyright: the authors 217 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 灾害风险认知水平的大小，只有发布清晰、可信、 可达的台风灾害风险预警信息，公众才能真正地感 知灾害风险，并积极响应，从而最终降低灾害损失。 3. 有效风险沟通是影响风险认知的关键 在影响公众灾害风险认知的三大因素中，公众 个体属性和风险情景特征一旦形成，一般难以改变， 因此提高公众台风灾害风险认知能力的关键在于提 高风险沟通效率。根据影响风险沟通效率的四个因 素，本文提出从三个层面的建设来完善有效风险沟 通（表 3），从而提高风险信息的清晰度、可信性、 可达性以及公众的响应度，最终推动公众灾害风险 认知能力的提升。 表 3. 有效风险沟通的建设内容 三个层面 具体内容 政府风险沟通能力 组织管理能力、风险评估能力、 沟通协调能力 风险沟通专门平台 网络技术、手机终端、专门网站 风险沟通文化环境 大众传媒、风险教育、社会参与 3.1 政府台风灾害风险沟通能力建设 政府台风灾害风险沟通能力建设是风险沟通有 效实施的关键。政府灾害风险沟通能力包括组织管 理能力、风险评估能力、沟通协调能力。 政府的组织管理能力，具体体现在风险沟通机 构、风险沟通制度、风险沟通保障、人员队伍配备。 与风险沟通及风险管理相比，中国目前只重视应急 管理，而忽视灾害风险管理。在应急管理方面，根 据《中华人民共和国突发事件应对法》、《国家突 发公共事件总体应急预案》等的规定，应急管理组 织结构比较完善，国家层面有国务院应急管理办公 室，地方各级人民政府是本行政区域突发公共事件 应急管理工作的行政领导机构，负责本行政区域各 类突发公共事件的应对工作。有了组织机构和制度， 其他方面的建设就迎刃而解了。因此，建议政府有 关部门以应急管理为基础，开展风险管理及风险沟 通的组织管理工作。 政府的风险评估能力非常重要，风险评估结果 是风险信息的来源，目前导致公众对政府风险管理 能力缺乏信心的一个重要原因就是公众不信任政府 的风险评估能力及结果。公开、公正、科学地进行 灾害风险评估和建立预警系统，是台风灾害风险沟 通工作的必然要求。政府的风险评估不仅仅是指其 工作人员进行的内部评估，还应该广泛吸纳相关专 家和普通公众的评估，使风险评估在尽可能广的范 围内实施。 在对公众的调查中，居民认为台风灾害风险管 理的不足，除了基础设施建设外，第二大不足就是 政府协调能力较弱。政府风险沟通协调能力的提升， 需要国家所有行政和立法机构在国家和地方各级充 分参与，明确划分利益相关者的责任，包括专家、 媒体和公众的责任，以确保伙伴合作、职责和问责 相得益彰。政府必须能够与其他利益相关者进行积 极的互动与协调，才能把握民众的心理和行为动向。 总之，只有政府部门做好风险沟通的顶层设计 与统筹协调，公众才有可能在政府的帮助下获得清 晰、及时的风险信息，并增强对政府的信任度，从 而不断提升灾害风险认知水平和能力。 3.2 台风灾害风险沟通专门平台建设 有效风险沟通的开展需要沟通平台的辅助，如 此利益相关者才能实现平等、及时、有效沟通，从 而提升公众的风险认知水平。以互联网为主要标志 的新媒体使得灾害风险信息海量出现，并且散落在 政府网站、微博、微信等媒介中。另外，新媒体的 出现使得灾害风险信息不再被政府和专家垄断，每 个人都可以成为风险信息发布者。因此，台风灾害 风险沟通专门平台建设非常重要。 目前灾害风险信息非常零散且陈旧，即使在灾 害应急管理上也存在同样的问题。打开中国政府网 published by atlantis press copyright: the authors 218 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster —应急管理的网页（http://www.gov.cn/yjgl/），里面 的信息和动态还停留在 2014 年 2 月。全国第一个应 急管理专门网站是广东省人民政府应急管理办公室 网站，虽然网站的内容比较丰富，但是访问量较少， 没有起到风险沟通的应有作用。应急管理的经验提 醒政府管理者，台风灾害风险沟通专门平台建设需 要注意两个关键问题：一是平台上的风险信息必须 进行数据分类 [17] ；二是政府应该重视基于新媒体的 沟通平台建设。 现有灾害风险应急沟通平台存在着沟通主体分 类和灾害信息分类不清，缺乏互动尤其是即时互动 渠道。因此，本文建议以网络技术为核心，以手机 终端为载体，建立自然灾害风险沟通专门网站，考 虑到不同类别用户的需求，以适当形式编制和更新 灾害风险数据和风险地图。此外，要通过政务微博、 政务微信，尤其是微信公众号收集意见、倾听民意、 发布信息、服务大众，例如中国气象局官方微信， 可在此基础上建立灾害风险微信公众号。 因此，政府管理者应该根据新时代灾害风险信 息传播的特征，整合资源，建立台风灾害风险沟通 专门平台，并且公开、及时地与风险专家、媒体和 公众互动，促进信息的分享与交流。 3.3 台风灾害风险沟通文化环境构建 风险沟通的有效实施不仅依赖于组织、人力和 硬件的建设，还依赖于文化环境的构建。在风险沟 通时不能够脱离社会文化结构，如果不了解社会文 化的价值和信仰，就难以了解公众的风险认知，也 难以进行有效的风险沟通 [25] 。 风险沟通文化环境的构建很大程度上依赖于大 众媒体，尤其是新媒体具有信息沟通、舆论导向和 稳定社会的重要功能，因此媒体在报道台风灾害时 应该秉承中立的原则，不随意夸大缩小灾害风险。 以 2015 年台风“彩虹”和 2016 年台风“妮妲”的 报道为例，百度搜索中“台风妮妲广州”的信息量 是“台风彩虹湛江”的 5 倍多，由此可见媒体报道 的偏好。在媒体和社会的关注下，广州在妮妲登陆 前就启动全市 i 级应急响应，这也是历史以来广州 市首次发布防台风防汛全民动员令，但最终台风只 是擦过广州南沙区。而湛江市在彩虹影响下全城停 水、停电数天，交通和通讯受阻严重，但主流媒体 只是在灾后才有少量报道。如何建立良好的媒体文 化也是风险沟通文化建设的重点。 除了大众媒体之外，风险沟通文化的构建还有 赖于利益相关者的共同努力。在政府方面，可以努 力发挥教育的功能，尤其是学校教育，使公众拥有 准确、恰当、充足的风险知识。在风险教育中，政 府部门必须培育风险文化，并鼓励利益相关者参与 风险知识的教育与培训。另外，在政府的引导下， 应鼓励全社会的积极参与，且进一步增强专家、媒 体和公众的风险责任感，只有人们对其做出的决定 承担责任，才有可能实现灾害风险管理的目标，从 而推动包容、积极和非歧视的风险沟通文化建设。 风险沟通的完善过程也就是风险文化建构的过 程，有效的风险沟通应该使风险信息与文化情景交 融，公众只有认可和接纳风险沟通文化，其风险认 知水平才能真正提升。 4. 结论 在台风灾害风险管理的过程中，政府管理者只 有了解公众的风险认知特征，发布清晰、可信、及 时的风险信息，并使风险预警被公众接受，才能使 风险沟通完成既定目标，有效降低灾害损失。本文 在理论和实证研究的基础上，得出了以下结论： （1）分析了 3 个维度的灾害风险认知影响因 素，即公众个体属性、风险情景特征和风险沟通效 率，每个维度又分别包括 3 个影响因素，探讨了各 个因素对公众灾害风险认知的影响情况。（2）以湛 江市台风灾害为例，具体分析了 12 个影响因素与公 众灾害风险认知的关系。研究发现，湛江市居民的 published by atlantis press copyright: the authors 219 z.h. shang and l.p. li / public risk perception and risk communication of typhoon disaster 台风灾害风险认知特征及其影响因素与理论模型基 本一致。（3）风险沟通是影响公众风险认知的关键， 有效风险沟通的建设包括风险沟通能力建设、风险 沟通专门平台建设和风险沟通文化环境构建，从而 改善风险信息的清晰度、可信性和可达性，努力打 造正面的公众参与风险沟通的文化环境，实现有效 风险沟通。 致谢：本文受到国家自然科学基金项目（41606053 ）、广东省自然科学基金项目（2016a030310364） 和湛江市哲学社会科学 2016 年度规划项目（ zj16yb09）资助，在此表示感谢。 参考文献： [1] the united nations office for disaster risk reduction. sendai framework for disaster risk reduction 2015 2030.[2015-03-18].http://www.unisdr.org/we/coordinate/sen dai-framework. [2] 尚志海, 刘希林. 自然灾害风险管理关键问题探讨.灾害 学, 2014, 29 (2): 159-164. [3] johnson b b. risk communication: a mental models approach. risk analysis, 2002, 22 (4): 813-814. [4] unisdr. global assessment report on disaster risk reduction.[2015-12-04].http://www.preventionweb.net/engl ish/hyogo/gar/2015/en/home/index.html. [5] raška p. flood risk perception in central-eastern european members states of the eu: a review. natural hazards, 2015, 79 (3): 2163-2179. [6] steelman t a, mccaffrey s. best practices in risk and crisis communication: implications for natural hazards management. natural hazards, 2013, 65 (1): 683-705. [7] slovic p. perception of risk. science, 1987, 236 (4799): 280-285. [8] gregory r, mendelsohn r. perceived risk, dread, and benefits. risk analysis, 1993, 13 (3): 259-264. [9] 谢晓非, 郑蕊. 风险沟通与公众理性. 心理科学进展, 2003, 11 (4): 375-381. [10] gierlach e, belsher b e, beutler l e. cross-cultural differences in risk perceptions of disasters. risk analysis, 2010, 30 (10): 1539-1549. [11] starr c. social benefit versus technological risk. science, 1969, 165 (3899): 1232-1238. [12] li g m. tropical cyclone risk perceptions in darwin, australia: a comparison of different residential groups. natural hazards, 2009, 48 (3): 365-382. [13] terpstra t, lindell m k, gutteling j m. does communicating (flood) risk affect (flood) risk perceptions? results of a quasi-experimental study. risk analysis, 2009, 29 (8): 1141-1155. [14] finlay p. j., fell r. landslides: risk perception and acceptance. canadian geotechnical journal, 1997, 34 (2): 169-188. [15] maidl e, buchecker m. raising risk preparedness by flood risk communication. natural hazards and earth system sciences, 2015, 15 (7): 1577-1595. [16] ickert j, stewart i s. earthquake risk communication as dialogue-insights from a workshop in istanbul's urban renewal neighbourhoods. natural hazards and earth system sciences, 2016, 16 (5): 1157-1173. [17] 刘良明, 范一大, 李明, 等. 重大自然灾害风险沟通研 究. 北京: 测绘出版社, 2012. [18] 尚志海, 李晓雁. 广东省沿海地区台风灾害风险评价. 岭南师范学院学报, 2015, 36 (3): 136-142. [19] 时振刚, 张作义, 薛澜. 核能风险接受性研究. 核科学 与工程, 2002, 22 (3): 193-198. [20] 苏桂武, 马宗晋, 王若嘉, et al. 汶川地震灾区民众认知 与响应地震灾害的特点及其减灾宣教意义——以四川 省德阳市为例. 地震地质, 2008, 30 (4): 877-894. [21] 李景宜, 周旗, 严瑞. 国民灾害感知能力测评指标体系 研究. 自然灾害学报, 2002, 11 (4): 129-134. [22] 戴薇. 广州居民灾害风险感知研究. 兰州: 兰州大学, 2014. [23] 李华强, 范春梅, 贾建民, et al. 突发性灾害中的公众风 险感知与应急管理——以 5·12 汶川地震为例. 管理世 界, 2009, 25 (6): 52-60. [24] 梁哲, 许洁虹, 李纾, et al. 突发公共安全事件的风险沟 通难题——从心理学角度的观察. 自然灾害学报, 2008, 17 (2): 25-30. [25] 王威. 风险沟通的文化变量分析. 新闻界, 2014, (24): 58-61. published by atlantis press copyright: the authors 220 http://www.unisdr.org/we/coordinate/sendai-framework http://www.unisdr.org/we/coordinate/sendai-framework http://www.preventionweb.net/english/hyogo/gar/2015/en/home/index.html http://www.preventionweb.net/english/hyogo/gar/2015/en/home/index.html << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <feff06270633062a062e062f0645002006470630064700200627064406250639062f0627062f0627062a002006440625064606340627062100200648062b062706260642002000410064006f00620065002000500044004600200645062a064806270641064206290020064406440637062806270639062900200641064a00200627064406450637062706280639002006300627062a0020062f0631062c0627062a002006270644062c0648062f0629002006270644063906270644064a0629061b0020064a06450643064600200641062a062d00200648062b0627062606420020005000440046002006270644064506460634062306290020062806270633062a062e062f062706450020004100630072006f0062006100740020064800410064006f006200650020005200650061006400650072002006250635062f0627063100200035002e0030002006480627064406250635062f062706310627062a0020062706440623062d062f062b002e0635062f0627063100200035002e0030002006480627064406250635062f062706310627062a0020062706440623062d062f062b002e> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <feff0054006900650074006f0020006e006100730074006100760065006e0069006100200070006f0075017e0069007400650020006e00610020007600790074007600e100720061006e0069006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b0074006f007200e90020007300610020006e0061006a006c0065007001610069006500200068006f0064006900610020006e00610020006b00760061006c00690074006e00fa00200074006c0061010d00200061002000700072006500700072006500730073002e00200056007900740076006f00720065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f00740076006f00720069016500200076002000700072006f006700720061006d006f006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076016100ed00630068002e> /slv <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> /suo <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> /sve <feff0041006e007600e4006e00640020006400650020006800e4007200200069006e0073007400e4006c006c006e0069006e006700610072006e00610020006f006d002000640075002000760069006c006c00200073006b006100700061002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400200073006f006d002000e400720020006c00e4006d0070006c0069006700610020006600f60072002000700072006500700072006500730073002d007500740073006b00720069006600740020006d006500640020006800f600670020006b00760061006c0069007400650074002e002000200053006b006100700061006400650020005000440046002d0064006f006b0075006d0065006e00740020006b0061006e002000f600700070006e00610073002000690020004100630072006f0062006100740020006f00630068002000410064006f00620065002000520065006100640065007200200035002e00300020006f00630068002000730065006e006100720065002e> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice research article 基于 dea-tobit 模型的创业板上市公司管理层能力度量研究 research on the management capability measurement of gem listed companies based on dea tobit model feng xia1,2,*,†, zhang mu1,† 1school of big data application and economics, guizhou university of finance and economics, guiyang, guizhou 550025, china 2guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang, guizhou 550025, china a r t i c l e i n f o article history received 14 september 2020 accepted 09 november 2020 keywords dea-tobit model management management capability a b s t r a c t with the rise of the growth enterprise market (gem), more and more investors have begun to pay attention to it and begin to consider investing in related listed companies. therefore, the healthy and stable development of the gem is of great significance to investors. the management of the company is the actual operator of the company. using this as a starting point, analyzing the management capabilities of listed companies on the gem has very important theoretical value and practical significance for the sustainable development of the gem. taking the listed companies on the growth enterprise market of china from 2014 to 2018 as the research sample, this paper uses the two-stage data envelopment analysis-tobit model to measure the management ability, and analyzes the differences of the management ability among the listed companies with different characteristics. the research results show that there are differences in management capabilities between listed companies with different characteristics. specifically, the management ability of manufacturing companies is higher than that of non-manufacturing enterprises; the management ability of relatively large companies is higher than that of relatively small companies; the management ability of companies located in the eastern region is higher than that of companies in the middle companies in the western region. *corresponding author. email: 1843283748@qq.com †国家自然科学基金项目（地区科学基金项目）：“基于文本信息的科技型中小企 业信用风险识别机理研究”，项目批准号：71861003。 journal of risk analysis and crisis response vol. 10(4); december (2020), pp. 138–146 doi: https://doi.org/10.2991/jracr.k.201214.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr 关键词 dea-tobit 模型 管理层 管理层能力 中图分类号 f832.42 文献标识码 a 摘要 随着创业板的兴起，越来越多的投资者开始对其进行关注，并且开始考虑投资于相关上市公司，因此创业板市场的 健康稳定发展对投资者意义重大。公司管理层是公司的实际经营者，以此为出发点，分析创业板上市公司的管理层 能力，对创业板的持续发展具有十分重要的理论价值和现实意义。以 2014–2018 年中国创业板上市公司为研究样 本，运用数据包络分析法（dea-tobit）两阶段模型衡量管理层能力，对不同特征上市公司之间管理层能力的差异性 进行分析。研究结果表明:具有不同特征的上市公司之间的管理层能力存在差异。具体来说，制造业公司的管理层能 力高于非制造业企业；规模相对较大的公司的管理层能力高于规模相对较小的公司；位于东部地区的公司的管理层 能力高于位于中西部地区的公司。 © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 1. 引言 管理层能力是企业所拥有的最重要的无形资产之一，同时也 是管理层人力资本的外在体现，直接关系到企业的发展前 景。管理层能力是上市公司成功运营的重要因素，对上市公 司政策的制定、企业行为经济后果有着重要的影响。在财 务和经济学文献中，管理层能力是资本市场上一个重要的元 素，能力较强的管理层能够发现公司潜在的特质，他们能够 精准地将公司特有的价值向外部信息使用者传递，这就减少 了股票市场中上市公司与外部信息使用者之间的信息不对 称，引导社会资金优化配置，促进创新型行业发展，同时对 我国的产业结构优化起到推动作用。 管理层的界定是研究其具有管理能力的基础，因此，精准衡量 管理层能力显得至关重要。目前，对管理层的界定还没有统一 标准。通过对现有文献的梳理发现，在不同的制度背景下， 对管理层的概念和范围有所不同。国外的研究一般将管理层的 范围界定为以 ceo 为核心研究对象。国内的研究对管理层的 界定是从广义和狭义两个角度对其进行界定。从广义角度上 看，管理层一般包括公司年报上披露的公司的董事、监事和高 级管理人员(何威风等,2016; 姚立杰等,2018)[1-2]；从狭义 角度上界定，管理层包括年报中披露的总经理、副总经理、财 务总监、董事会秘书和其他高级管理人员，不含董事会和监事 会成员(薛安伟,2018; 李彬和秦淑倩,2016)[3-4]。根据本文 的研究目的和所选用的管理层能力的计量方法，结合我国的实 际情况和数据的可获得性，本文将管理层界定为公司的高级管 理人员。依据 2014 年 3 月起施行的 《中华人民共和国公司法》[5]第 217 条规定：高级管理人 员，具体是指公司的经理、副经理、上市公司董事会秘书， mailto:1843283748%40qq.com?subject= https://doi.org/10.2991/jracr.k.201214.002 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 139 财务负责人和其他公司章程规定的人员，也就是目前我国上 市公司年度报告中所披露的包括总经理、总裁、副总经理、 副总裁、财务总监、技术总监、总工程师、总经济师以及担 任管理职位的董事会成员和董事会秘书等会对公司经营决策 产生重要影响的高管成员。 良好的管理层能力能够为企业带来长远的发展空间与客观的 收益，同时也能满足投资者对上市公司的要求。本文通过衡 量创业板上市公司的管理层能力水平，为上市公司调整管理 层结构提供参考，对于促进创业板市场良性发展，具有极为 重要的理论意义。 2. 理论基础 管理层能力即为公司管理团队成员所具有的经营管理能力。 管理层能力就是管理层利用企业成本资源为企业创收的能力( 潘前进等，2015)[6]。在我国，每个管理层能力会受到管理 者背景特征、性格特征、思想观念和人际关系等众多因素的 影响，并且最终会体现在公司生产 效率、资本运营、战略规 划与执行等涉及公司经营与发展的各个方面。因此，识别一 个衡量管理层能力的可靠代理变量是非常复杂的，包含感知 能力、可信性、魅力、正直、诚实以及想象力，同样还有其 他不可量化的因素。 2.1. 用代理变量间接评价管理层能力 学者们主要从公司特征和管理层特征两个方面选取指标替代 衡量管理者能力。从公司自身特征的角度，thomas j（2010） [7]采用公司规模的大小衡量管理层能力，还有学者用公司业 绩相关指标，如行业调整后的历史股票回报率和盈余预测误 差(samuel b, 2016)[8]作为管理层能力的衡量指标。kathleen fuller (2002)[9]和 dan s (2016)[10]运用行业调整 的股票收益作为管理层能力的代理变量进行研究，他们发 现，越有能力的管理层拥有更多的雇佣机会。bok baik,paul a (2018)[11]发现在管理层被媒体识别为明星管理层之后， 公司获得显著较低的股票收益、较高的管理层薪酬以及较高 的盈余管理水平。但是这些指标由于包含管理层控制之外的 信息而饱受争议。chien-chiang lee (2018)[12]同样运用行 业调整的股票收益、ceo 任期和媒体引用度作为管理层能力 的衡量指标，发现越有能力的管理层拥有更高的薪酬－业绩 敏感度。然而，尽管这些指标在一定程度上反映了管理层能 力，但是仅从一个层面评价管理层能力会存在较大争议。 2.2. 管理者固定效应模型 利用固定效应模型估计管理层能力，就是以管理者产生变更 的公司为研究样本，将管理者变更前后的公司业绩等指标进 行对比，以衡量管理者能力的高低(bamber 等,2010;陈仕华 等,2013)[13-14]。国内学者马利军、周威(2012)[15]提出了 用股票市场反应和再任职市场反应代替管理层能力。不过以 上都是立足于突发事件，基于管理者固定效应这种方法可以 测量管理者个体对公司产生的影响，可以剔除公司本身特征 所产生的影响,然而，这种方法也存在一定的局限性。首先， 研究样本必须是在样本期间经历过管理者更替的公司，这使 公司在样本选择上存在较大偏差。其次，管理者的影响往往 具有一定的持续性，管理者发生变更并不代表企业不会受到 之前管理者的影响。 2.3. 数据包络分析(dea)法 数据包络分析法是利用线性规划的方法，根据所设定的投入 与产出指标，对具有可比性的业务单元进行相对效率有效性 评价和比较的一种数量分析方法。运用包络分析法衡量管理 层能力，就是将公司看成单独的决策单元(dmu)，管理层能力 就可以用管理者利用既定资源创造产出的相对效率值来表示 (leverty 和 grace 2012)[16]。陈德球和步丹璐(2015)[17] 运用数据包络分析法衡量管理层能力。这个方法可以综合评 价管理层的管理能力，但却无法将公司本身特征的影响与管 理层能力的影响区分开来。 2.4. dea-tobit 两阶段模型 为了改进数据包络分析法的缺陷，demerjian 等人(2012) [18]提出了 dea-tobit 两阶段的方法衡量管理层能力。该方 法先用包络分析法计算样本公司的相对效率值，然后再从企 业效 率中将公司特有的影响因素剔除，将管理者所贡献的部 分分离出来即为管理层能力。 demerjian 等人对该方法进行 了有效性检验，结果表明该方法可以更加精确地计量管理层能 力，在实证研究中优于其他计量指标，已成为应用最为广泛的 管理层能力的计量方法。自 demerjian 等人(2012)提出这一 衡量方法后，国内学者普遍采用这一指标对管理层能力进行 衡量，并对其经济效果进行探究（吴育辉等，2017）[19]。 张铁铸、沙曼(2014)[20]运用同样的 方法，研究发现管理层 能力越低，运用其职权在职消费的行为越普遍。谢建(2016) [21]在研究 管理层能力与企业避税问题时发现，管理层能 力可以降低企避税活动。张敦力、江新峰（2015） [22]利用 demerjian(2012)对管理层能力的衡量指标实证检验了管理层 能力对公司业绩预告的影响。何威风、刘巍(2015)[23]研究 发现，管理层能力与企业的审计收费负相关，并且这种负相 关关系主要存在于国有上市公司。许宁宁（2017）[24]、沈 烈和郭阳生（2017）[25]的研究表明，管理层能力的提高有 助于识别企业的内部控制缺陷。肖明、李海涛(2017)[26]基 于沪深 a 股上市公司数据，检验了管理层能力对企业并购的 影响。本文同样采用demerjian 等人 (2012) 构建的管理层 能力衡量指标，对创业板上市公司的管理层能力进行研究。 3. 基于 dea-tobit 的管理层能力度量模型 3.1. 一阶段数据包络分析法(dea) 数据包络分析法(dea)最初由charnes, cooper和rhodes (1978) [27]提出，这就是ccr模型。该模型是将farrell(1957)所提 出的“两投入一产出”的模式，推广至“多投入多产出”的 模式，并利用线性规划和对偶定理，求出待评估单位的生产 前沿，凡落在边界上的决策单位称为dea 有效率，其效率值 为1;而其他未落在边界上的smu则称为dea无效率，其效率值 介于0与1之间。dea方法的特点或者说优势在于以下几点:首 先，它是一种可以用于评价具有多投入、多产出的决策单位 运营效率的方法。由于dea不需要指定投入产出的生产函数形 态，因此它可以评价具有较复杂生产关系的决策单位的效率; 其次，它具有单位不变性，即dea衡量dmu 的效率不受投入产 出数据所选择单位的影响;第三，dea可以进行差异分析、敏 感度分析和效率分析，从而可以进一步了解决策单位资源使 用的情况，帮助管理者进行经营决策。当然，dea也存在一些 弱点，如对于投入产出为负的情况必须经过一定的人为调整 140 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 才能进行相对合适的效率测度，并且易受投入产出项人为选 择不当的影响。 3.2. 二阶段 tobit 回归 在第一阶段利用dea方法测算了上市公司运营效率之后，在 第二阶段将运营效率作为因变量，运营效率的影响因素作为 自变量，通过回归模型来实证考察自变量与因变量之间的关 系。由于前面我们测算出来的运营效率值都处于0和1之间， 最大值为1，具有被切割或截断的特点，因此选择tobit回归 模型。 tobit回归模型属于因变量受到限制的一种模型，其概 念最早是由tobin(1958)[28]提出， 然后由经济学家 goldberger(1964)[29]首度采用。如果要分析的数据具有这 样的特点:因变量的数值是切割或片段的情况时，那么普通最 小二乘法(ols)就不再适用于估计回归系数，这时 遵循最大 似然法概念的tobit模型就成为估计回归系数的一个较好选 择。由于dea方法所估计出的效率值都介于0与1之间，最大值 为1，如果采用最小二乘法来估计，可能由于无法完整地呈现 数据而导致估计偏差，因此本文决定采用tobit回归模型来分 析运营效率的影响因素。 3.3. 基于 dea-tobit 模型的管理层能力研究 识别一个衡量管理层能力的可靠代理变量是非常复杂的，因 为一家公司管理层的声誉性评价是多维度的，包含感知能 力、可信性、魅力、正直、诚实以及想象力，同样还有其他 不可量化的因素。本文亦采用demerjian等人(2012)提出的 dea-tobit两阶段模型并借鉴张铁铸和沙曼 (2014) 、肖明和 李海涛(2017)的研究，同时，考虑到中国会计处理的实际情 况和数据的可获得性，根据我国国情进行一定的修正来衡量 管理层能力。具体计算过程如下: 在第一阶段，本文按照公式（1）采用数据包络分析法分年 度分行业计算各个公司的运营效率（firm efficiency） ，采 用多阶段变动规模dea模型计算，并选择投入导向。该模型假 设，在年初给定一定资源的情形下，企业通过一年的努力， 在这一年期内取得的经营成果。由于不同行业、不同资源投 入的比例不同，因此，在计算各公司的投入-产出比率时按行 业计算，要控制行业变量的影响，然后按行业将所得到的行 业最高效率值作为该行业的基准进行标准化，得到一系列公 司效率值。 首先采用数据包络分析软件max dea7.0的ccr模型分年度分行 业测算各个公司的运营效率。该模型可以简单理解为，短期 内在企业资源耗费不变的情形下，获得最大生产经营成果。 该dea效率值(firm efficiency)的范围介于[0,1]之间，投 入-产出模型表述如下： firm efficiency sale cost sg ppe gw in = + + + + + v v a v v r d v v 1 2 3 4 5 6 & & tan (1) 其中，产出变量为营业收入(sale)，是 dea 分析中唯一一 个产出变量，投入变量分别为营业成本(cost), 销售及管理 费用(sg&a) 、固定资产净额 (ppe) 、研发支出(r&d), 商 誉(goodwill)和无形资产净额(intangible)。营业收入、营 业成本、销售及管理费用是本期末数据，无形资产净额、商 誉、固定资产净额、研发支出是本期初数据。 由于投入产出变量的选择会直接影响上市公司的运营效率，因 此对投入产出变量的选择非常慎重。公司进行生产经营所产生 的最直接的产出就是企业收入，因此将营业收入作为唯一产出 变量，同时进行投入变量的选择时综合短期和长期因素，既包 括营业成本、销管费用， 这两个与当期收入水平存在直接相 关关系的短期投入变量，也包括固定资产净值、无形资产净 值、商誉净值和研发支出四个与当期收入不存在明显直接关系 但是会从长远角度影响收入水平的长期投入变量。 在第二阶段，尽管生产效率值也可以作为管理层能力的替代 变量，且已有学者在研究中采用（岳修奎等,2019)[30]，但 这种方法计算出的管理层能力指标会掺杂其他因素的干扰， 导致变量的准确性和研究结果的可信度降低。需要剔除超出 管理层能力以外的公司因素（如规模优势等）对运营效率的 影响，按照公式 (2) 进行 tobit 回归, 得到残差为管理者 对公司效率的影响, 也就是管理者能力。tobit 回归模型表 示如下: fe size ms fcf age divers it it it it it it = + + + + + + + a a a a a a 0 1 2 3 4 5 year inddustry +åå e (2) 本文选取影响公司整体营运效率的特定因素包括以下五个方 面:总资产规模(size)、市场占有率(ms)、自由现金流 (fcf)、 上市年限(age)、公司多元化程度（divers）、σyear是年度 虚拟变量，反映年度固定效应， σ industry 是行业虚拟变 量，反映行业固定效应。 年度虚拟变量主要用来控制宏观经济的影响，行业虚拟变量 主要用来控制行业差异带来的影响。所谓虚拟变量，就是把 原来的一个多分类变量转化为多个二分变量，总的来说就 是， 如果多分类变量有 k 个类别，则可以转化为 k-1 个二 分变量。本文提到的年度虚拟变量的具体计算为：2014-2018 年可转化为 4 个二分类的变量，即“2014”虚拟变量取值为 1 时，其余年份均为 0，“2015”虚拟变量取值为 1 时，其 余年份均为 0，其余年份虚拟变量以此类推。行业虚拟变量 的计算方法同理。上市公司所属的行业不同，会直接影响公 司的运营效率，因此，在计算管理层能力时选择控制年度因 素、行业因素。 以上所述因素对上市公司的运营效率具有显著影响，通过模 型(2)，用上市公司的特征因素对公司运营效率进行回归，在 排除了公司特有因素规模、市场占有率、自由现金流、上市 年限、最终控制人性质，以及货币政策、财政政策和产业政 策等对企业运营效率的影响后， 管理层能力对企业运营效率 产生的效应将进入残差项，残差项包含的信息就是管理层能 力对企业运营效率的影响，即为管理层能力的度量。相关变 量的名称及计量方法，详见表 1。 3.4. 数据来源与样本筛选 本文所用指标的原始数据来自国泰安数据库(csmar)、锐思数 据库，本文选取创业板上市公司 2014-2018 年五年的相关数 据和信息为研究样本，在获取初始样本后，对样本进行了如 下筛选: (1) 剔除当年上市的上市公司，因为管理者能力测度的指标 要用到上年的数据。 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 141 (2) 由于金融类上市公司与其他类型上市公司的信息披露标准 存在较大差异，无法准确计算管理层能力，所以文章本处 选择与大多研究保持一致，即删除金融类上市公司。 (3) 剔除 st 类以及财务数据严重缺失的上市公司。 (4) 剔除样本量较少的教育及居民服务、修理和其他服务业 两个行业的上市公司，因为在管理层能力的计算过程中 需要分行业进行回归，需要保证足够的样本量。 最终，共有 300 个创业板上市公司纳入了研究的范畴，共得 到 1500 个面板数据。本文使用 max dea7.0 软件和 stata 16.0 统计分析软件进行数据处理。 采用中国证监会于 2012 年最新修订的《上市公司行业分类 指引》作为行业分类的标准， 剔除了可能影响实证结果的金 融、保险业与样本量较少行业的上市公司，最终研究样本按 行业划分为 13 类，如表 2 所示。 4. 实证检验与分析 4.1. 上市公司运营效率描述性统计 本章基于 dea 模型对创业板上市公司运营效率进行测算，选 择 ccr 模型计算运营效率， 得到的是上市公司的综合运营效 率，篇幅有限，详细数据见附表 1。关于上市公司运营效率估 计结果的描述性统计分析信息，见表 3。依据表中内容可知， 我国上市公司运营效率的平均值为 0.854，虽然这一数值并不 低，但其平均产出水平还是和最佳产出水平间存在一定偏离， 即上市公司的投入产出中还存在着较大改进空间。上市公司运 营效率的中位数为 0.858， 而且其标准差也仅为 0.124，水 平差异并不大。上市公司运营效率最小值为 0.25，最大值为 1，这说明虽然上市公司运营效率整体差异并不大，但仍有少 数公司的运营效率较低，同时， 也有少数公司的运营效率处 于极佳水平。 为了更直观地呈现上市公司运营效率的分布状况，本文绘制 了上市公司运营效率的频数 分布图，如图 1 所示。从图 1 可知，对于绝大多数上市公司而言，其运营效率值都分布于 0.85 的两侧，仅有极少数上市公司运营效率值分布于 0.25 附近，这进一步印证了此前的分析结论，即对于绝大数上市 公司而言，其平均运营效率约为 0.85，仅有极少数公司的运 营效率值偏低。 4.2. 上市公司管理层能力差异分析 4.2.1. tobit 回归结果分析 研究结果发现，资产规模、市场占有率、自由现金流、上市 年限、公司多元化程度对创业板上市公司运营效率的影响在 统计上显著。因此，tobit 回归方程成功剔除了超出管理层 能力以外的公司因素对公司运营效率的影响，残差能准确反 映管理层能力的大小。 模型回归结果表明，除上市公司年限外，其他均与公司运营 效率呈正相关。上市公司年限与营运效率负相关，随着上市 公司年限的增加，折旧成本越大，从而增大了运营成本，运 营效率也越小。企业规模与运营效率显著正相关，表示企业 规模越大，硬件和人力资源越完善，运营效率就越高;市场 占有率与企业运营效率显著正相关，市场占有份额越大，企 业经营、竞争能力就越强，获得比竞争对手更强的市场影 响力，运营效率也相对较高；自由现金流与公司运营效率显 著正相关，公司可利用的现金越多，越会加大生产投入， 投入-产出比率也会提高；公司多元化程度越高，说明管理层 能同时管理多项业务，增强公司竞争力表 4。 公司管理层能力的值域范围在-1 和 1 之间，篇幅有限，表 5 仅展示 2014 年的创业板上市公司的管理层能力，详细数据 见附表 1。 由于管理层能力是估计的残差，因此 2014-2018 年我国创业板上 市公司管理层能力 (tobit 回归残差)平均值偏小，仅为-0.02， 中位数为-0.02，最小值为-0.59，最大值为 0.25， 标准差为 0.12。这说明我国上市公司管理层能力存在较大差异，即不同 公司的管理者利用既定资源创造收入的能力明显不同。不同公 司的管理层能力必然会存在一定差异，本文从不同行业、不同 规模、不同地区等三个角度来分析管理层能力的差异表 6。 为了直观呈现创业板上市公司关于管理层能力的分布情况，本 文绘制了管理层能力频数分布图，如图 2 所示。从图中可以 看出，管理层能力频数分布直方图右偏，绝大部分上市公 司 表 1 | 管理层能力计算使用变量的定义 变量 变量名称 变量定义 size 企业规模 以公司总资产的自然对数衡量 ms 市场占有率 以公司营业收入占所处细分行业营业 收入总额的比例衡量 fcf 自由现金流 以公司自由现金流除以总资产衡量 age 上市年限 以上市年数的自然对数衡量 divers 公司多元化程度 企业各部分收入的平方除以总收入 的平方 year 年度虚拟变量 运用 stata，所属年份取值为 1，其余 年份均取值为 0 industry 行业虚拟变量 运用 stata，所属行业取值为 1，其余 行业均取值为 0 表 2 | 数据样本的行业分布情况 行业代码 行业名称 样本数量 所占比重 a 农、林、牧、渔业 25 0.02 b 采矿业 15 0.01 c 制造业 960 0.64 d 电力、热力、燃气及水生产 和供应业 10 0.01 e 建筑业 15 0.01 f 批发和零售业 10 0.01 g 交通运输、仓储和邮政业 10 0.01 i 信息传输、软件和信息技术 服务业 305 0.20 l 租赁和商务服务业 35 0.02 m 科学研究和技术服务业 35 0.02 n 水利、环境和公共设施管 理业 45 0.03 q 卫生和社会工作 20 0.01 r 文化、体育和娱乐业 15 0.01 表 3 | 上市公司运营效率的统计分析结果 变量 中位数 平均数 标准差 最小值 最大值 样本数 fe 0.858 0.854 0.124 0.25 1 1500 142 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 的管理层能力数值分布在 0 的附近，说明管理层能力整体而 言偏低，且管理层能力相对低的企业多于管理层能力相对高的 企业，且不同管理层能力水平上的公司数目存在较大差异。 4.2.2. 不同行业间管理层能力差异分析 根据中国证监会于 2012 年最新修订的行业分类标准，剔除 了可能影响实证结果的金融、保险业与样本量较少行业的上 市公司，最终研究样本共十三个行业门类。而在我国创业板 上市的公司中，超过半数的企业都归属于制造业。因此，检 验制造行业和非制造行业企业间管理层能力差异具有必要 性。基于这一实证分析要求，文章本处将全部全样本划分为 制造行业子样本和非制造行业子样本，并据此描述统计分析 其管理层能力水平，详细分析见表 7。 依表可知，电力、热力、燃气及水生产和供应业的管理层能力 排第一，建筑业的管理层能力紧跟其后，剩余行业的管理层能 力排序分别为：农、林、牧、渔业；租赁和商务服务业； 批发 和零售业；科学研究和技术服务业；制造业；采矿业；信息传 输、软件和信息技术服务业；卫生和社会工作；交通运输、仓 储和邮政业；文化、体育和娱乐业；水利、环境和公共设施管 理业。从整体来看，制造行业上市公司管理层均值为-0.017， 而非制造行业上市公司管理层能力均值为-0.026，可见制造行 业上市公司管理层能力均值高于非制造行业样本公司管理层能 力均值，且基于中位数的分析结论具有一致性。由此可见，同 非制造行业企业相比， 制造行业企业的管理层能力较高。这 可能是制造业上市公司大都是发展较为成熟的企业，拥有比较 全面的制度流程，相对于其他行业企业，管理层做出错误决策 而影响公司运营效率的可能性较小。 4.2.3. 不同企业资产规模间管理层能力差异分析 前期相关研究显示，上市公司管理行为与效果常常会与规模 差异层面产生分异，所以接下来拟分析管理层能力与资产规 模差异层面的差异性表现。本文以 2014-2018 年分行业公 司资产规模的中位数为分组标准，高于行业年度中位数的样 本分类为较大规模公司，而低于行业年度中位数的样本纳为 较小规模公司，并且据此分别统计分析其管理层能力，以检 验不同企业规模下管理层能力的差异性，详细数据见表 8。 依表可知，大规模公司管理层能力均值为-0.014，而小规模 公司管理层能力均值为-0.024， 可见大规模上市公司管理层 能力均值高于小规模上市公司管理层能力均值，且基于中位 数的 分析结论具有一致性。这在一定程度上说明公司规模越 大，管理层能力越高。其中的原因可 能是规模较大的公司往 往是行业中的引领者，更有实力和资源聘用能力较强的管理 层，而能 力较强的人才也往往更倾向于选择规模较大的平台 寻求职业发展。 图 1 | 创业板上市公司运营效率的频数分布图. 表 4 | tobit 模型回归结果 fe coef. std. err. t p > t size 0.0453362 0.005851 7.75 0 ms 0.3774359 0.069477 5.43 0 fcf 0.3174455 0.055065 5.76 0 age –0.090014 0.023662 –3.8 0 divers 0.0443751 0.015054 2.95 0.003 year_2 –0.0279936 0.013077 –2.14 0.032 year_3 –0.0018165 0.015592 2.12 0.0907 year_4 0.0237775 0.018277 1.3 0.193 year_5 –0.0102274 0.02075 2.87 0.022 industry_2 0.1434838 0.057472 2.5 0.013 industry_3 0.0681127 0.058893 1.16 0.248 industry_4 0.1385067 0.057374 2.41 0.016 industry_5 0.1797663 0.063391 2.84 0.005 industry_6 0.1314749 0.062756 2.1 0.036 industry_7 0.2290961 0.081319 2.82 0.005 industry_8 0.241252 0.072421 3.33 0.001 industry_9 0.1811743 0.059521 3.04 0.002 industry_10 0.0034389 0.066308 1.05 0.259 industry_11 0.093948 0.057514 1.63 0.103 industry_12 0.1570207 0.058935 2.66 0.008 industry_13 –0.0378795 0.060415 –0.63 0.531 _cons 0.4013336 0.197071 2.04 0.042 * p < 0.1, ** p < 0.05, *** p < 0.01. 注：回归结果数据由 tobit 回归模型整理可得. f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 143 表 5 | 2014 年创业板上市公司的管理层能力（ma） 名称 ma 名称 ma 名称 ma 名称 ma 名称 ma 特锐德 0.082 华谊嘉信 0.079 盈康生命 –0.06818 易华录 0.178229 苏交科 0.021555 神州泰岳 –0.042 三聚环保 0.043 宋城演艺 –0.26948 佳讯飞鸿 0.160901 国瓷材料 0.058965 乐普医疗 –0.109 当升科技 –0.053 中金环境 –0.22022 电科院 0.141099 安科瑞 0.033009 南风股份 –0.091 数字政通 0.012 香雪制药 –0.52393 千山药机 0.118242 飞利信 –0.08507 探路者 –0.203 gq 视讯 0.023 天舟文化 –0.44611 东方电热 0.057555 朗玛信息 –0.25825 莱美药业 –0.036 国民技术 0.136 量子生物 –0.00804 鸿利智汇 –0.13093 利德曼 0.014701 汉威科技 –0.199 思创医惠 –0.023 世纪瑞尔 –0.05327 金运激光 –0.00033 荣科科技 –0.10334 天海防务 –0.204 数码科技 0.028 昌红科技 –0.07474 银禧科技 –0.01247 吴通控股 –0.10671 安科生物 –0.189 易成新能 0.028 科融环境 –0.17388 科大智能 0.094654 蓝英装备 –0.06913 立思辰 –0.126 恒信东方 0.003 科泰电源 –0.15672 正海磁材 0.119768 博雅生物 –0.02895 鼎汉技术 0.049 奥克股份 –0.136 安居宝 –0.00903 光韵达 –0.01028 三六五网 –0.17562 华测检测 0.057 劲胜智能 –0.112 神雾环保 0.059954 富瑞特装 –0.10255 利亚德 –0.12556 新宁物流 0.035 海默科技 –0.108 恒泰艾普 0.144548 拓尔思 –0.00868 蓝盾股份 –0.12906 亿纬锂能 0.053 银之杰 –0.168 振东制药 0.099095 永利股份 0.053204 三诺生物 –0.16365 爱尔眼科 0.086 康芝药业 –0.116 新研股份 0.166576 洲明科技 0.048296 富春股份 –0.14475 北陆药业 0.067 荃银高科 0.165 秀强股份 0.102984 金城医药 –0.01951 汉鼎宇佑 –0.0414 网宿科技 –0.114 长信科技 0.136 华中数控 –0.01254 开尔新材 –0.03061 长方集团 0.043062 中元股份 –0.046 文化长城 0.086 雷曼光电 –0.16172 方直科技 0.07636 聚飞光电 0.099747 硅宝科技 –0.097 盛运环保 0.087 先锋新材 –0.07544 上海新阳 0.097198 云意电气 0.035633 银江股份 –0.145 金通灵 0.176 通源石油 –0.30265 美晨生态 0.085261 远方信息 0.038118 大禹节水 0.075 科新机电 0.044 天瑞仪器 –0.22308 冠昊生物 –0.04959 慈星股份 –0.0442 机器人 0.163 国联水产 0.100 东方国信 0.018412 飞力达 –0.07795 中际旭创 –0.06223 华星创业 0.132 华伍股份 0.064 迪威迅 0.054084 瑞丰光电 –0.03794 宜通世纪 –0.02826 红日药业 0.159 易联众 0.051 万达信息 0.053395 佳云科技 –0.17173 任子行 –0.05167 金亚科技 0.216 智云股份 –0.050 天晟新材 0.108691 迪安诊断 –0.20236 邦讯技术 –0.13973 阳普医疗 –0.093 高新兴 –0.017 汉得信息 –0.064 天玑科技 –0.01542 天山生物 –0.07507 宝通科技 –0.143 精准信息 0.119 东富龙 –0.07061 宝莱特 0.1179 戴维医疗 0.110522 金龙机电 –0.083 双林股份 0.151 中电环保 0.033834 融捷健康 0.151948 掌趣科技 0.034064 同花顺 –0.081 乐视退 0.127 智慧松德 0.01666 新开普 0.120442 晶盛机电 0.021794 中科电气 0.039 西部牧业 0.074 元力股份 –0.00796 依米康 0.188776 珈伟新能 0.110122 超图软件 –0.083 吉药控股 0.061 中海达 –0.12223 初灵信息 –0.26174 博晖创新 –0.08394 新宙邦 –0.091 新开源 0.031 腾邦国际 –0.18562 金信诺 –0.20574 麦捷科技 –0.08084 数知科技 –0.062 华仁药业 0.035 四方达 –0.13688 卫宁健康 –0.12979 海达股份 0.008366 九洲电气 –0.097 向日葵 –0.040 华峰超纤 –0.1402 仟源医药 –0.12781 硕贝德 0.05618 回天新材 –0.149 万讯自控 0.084 佐力药业 –0.08058 常山药业 –0.02809 华灿光电 0.124425 星辉娱乐 –0.005 顺网科技 0.114 捷成股份 0.072919 星星科技 –0.1591 旋极信息 0.04373 赛为智能 0.190 中航电测 0.014 东软载波 –0.02582 开山股份 –0.22011 德威新材 –0.05445 华力创通 0.149 长盈精密 0.082 力源信息 –0.00825 精锻科技 –0.05265 凯利泰 0.007714 台基股份 0.009 坚瑞沃能 0.035 通裕重工 –0.00621 新天科技 0.020727 宜安科技 –0.06742 天源迪科 –0.116 嘉寓股份 0.149 永清环保 –0.08247 新莱应材 –0.06726 海伦钢琴 –0.07786 合康新能 0.151 东方日升 0.086739 美亚柏科 0.027424 雅本化学 –0.14174 苏大维格 0.177658 福瑞股份 0.145 瑞普生物 0.074095 神农科技 –0.00513 巴安水务 –0.35707 天壕环境 0.043534 世纪鼎利 0.111 经纬辉开 0.056357 维尔利 –0.07809 隆华科技 –0.19895 兆日科技 –0.03229 三五互联 0.042 智飞生物 0.083516 科斯伍德 0.065024 佳创视讯 –0.29528 迪森股份 0.007353 中青宝 0.168 亚光科技 0.03792 佳士科技 0.080301 通光线缆 –0.03721 润和软件 0.157395 欧比特 –0.077 汇川技术 0.109068 福安药业 –0.39496 兴源环境 0.058533 科恒股份 –0.05342 鼎龙股份 –0.094 聆达股份 0.129614 长荣股份 –0.10734 尔康制药 0.007182 麦克奥迪 –0.09679 万邦达 –0.093 锦富技术 0.109404 铁汉生态 –0.0631 联建光电 3.66e–05 天银机电 –0.11367 中创环保 –0.148 泰胜风能 0.10479 纳川股份 –0.04506 华宇软件 0.00747 联创股份 –0.08632 万顺新材 –0.179 新国都 0.117809 翰宇药业 –0.18375 开能健康 –0.02436 太空智造 –0.1626 蓝色光标 0.103 英唐智控 0.125644 高盟新材 0.039489 和佳医疗 0.037235 红宇新材 –0.00486 旗天科技 0.106 青松股份 0.17919 海伦哲 0.026708 阳光电源 0.018902 泰格医药 –0.07329 中能电气 –0.118 大富科技 0.023828 聚光科技 0.017611 梅安森 0.026251 长亮科技 0.06905 天龙集团 –0.160 宝利国际 –0.05924 舒泰神 0.14913 三丰智能 0.017033 金卡智能 –0.02762 海兰信 –0.185 信维通信 –0.01432 天喻信息 0.207429 华昌达 –0.06618 华鹏飞 –0.18039 三川智慧 –0.066 先河环保 0.12335 理邦仪器 –0.11919 和晶科技 0.024886 永贵电器 –0.11199 安诺其 –0.149 晨光生物 0.125266 欣旺达 –0.00934 紫天科技 0.050218 北信源 –0.08451 南都电源 –0.100 晓程科技 0.096873 青岛中程 –0.07006 金明精机 –0.00579 东土科技 –0.13326 金利华电 –0.084 和顺电气 –0.01073 天泽信息 –0.15363 三盛教育 –0.07405 蒙草生态 –0.14135 碧水源 –0.055 沃森生物 0.133796 森远股份 –0.12165 温州宏丰 –0.05101 光一科技 –0.12045 144 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 图 2 | 创业板上市公司管理层能力的频数分布图. 表 6 | 管理层能力描述性统计 变量 均值 中位数 最大值 最小值 标准差 样本量 管理层能力（ma） –0.02 –0.02 0.25 –0.59 0.12 1500 注：由 tobit 回归的残差结果整理可得. 表 7 | 不同行业管理层能力 行业名称 平均数 标准差 中位数 最小值 最大值 样本数 农、林、牧、渔业 0.021676 0.09885 0.022031 –0.1603 0.2354 25 采矿业 –0.03181 0.124512 –0.02301 –0.3027 0.1445 15 制造业 –0.01665 0.114448 –0.01415 –0.524 0.255 960 电力、燃气及水生产和供应业 0.035476 0.110728 0.041661 –0.1851 0.2071 10 建筑业 0.029501 0.087533 0.015647 –0.0987 0.1485 15 批发和零售业 0.005252 0.072001 –0.00211 –0.0811 0.1256 10 交通运输、仓储和邮政业 –0.04251 0.10742 –0.00649 –0.2702 0.0861 10 信息传输、软件和信息技术服务业 –0.03233 0.127515 –0.02376 –0.5181 0.2438 305 租赁和商务服务业 0.005313 0.123306 3.66e–05 –0.323 0.2102 35 科学研究和技术服务业 –0.00334 0.111069 –0.00628 –0.2183 0.2028 35 水利、环境和公共设施管理业 –0.07676 0.126398 –0.05692 –0.5855 0.1829 45 卫生和社会工作 –0.04138 0.100515 –0.0479 –0.2766 0.1334 20 文化、体育和娱乐业 –0.04613 0.155246 –0.017 –0.4461 0.179 15 表 8 | 不同规模下的管理 样本量 均值 中位数 最小值 最大值 标准差 大规模公司 800 –0.014 –0.011 –0.553 0.247 0.117 小规模公司 700 –0.024 –0.023 –0.593 0.250 0.118 4.2.4. 不同地区间管理层能力差异分析 在社会经济发展的过程中，由于自然资源禀赋的不同和受到 历史、文化等方面的影响， 不同地区的经济发展水平必然会 存在差异。而我国幅员辽阔，地理上自然环境的差异尤为明 显，再加上不同地区战略布局和政策导向的不同，各个地区 经济不平衡发展，存在较大差距。基于此，本文依据传统的 地区划分方法，将上市公司所在注册地划分为东部、中部和 西部三个地区。 从样本分布来看，注册地位于东部地区的样本有 1210 个， 占比高达 80.7%，注册地位于中部地区的样本有 195 个，注 册地位于西部地区的样本有 95 个，可见我国的上市公司主 要 集中在东部地区。因此，本文将依据公司注册地将全部样 本划分为东部地区、中部地区、西部地区三组，旨在探究不 同地区间管理层能力是否存在显著差异，详细数据见表 9。 山东、广东和海南;中部地区包括 8 个省份，具体为山西、 吉林、黑龙江、安徽、江西、河 南、湖北和湖南;西部地区 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 145 表 9 | 不同地区管理层能力 样本量 均值 中位数 最小值 最大值 标准差 东部地区 1210 –0.013 –0.005 –0.46 0.255 0.118 中部地区 195 –0.019 –0.013 –0.466 0.255 0.111 西部地区 95 –0.034 –0.0353 –0.464 0.235 0.132 注：东部地区包含 11 个省市，具体为北京、天津、上海、辽宁、河北、江苏、浙江、福建. 包括 12 个省市自治区，具体为重庆、广西、四川、贵州、 云南、陕西、甘肃、青海、内蒙古、西藏、宁夏和新疆。 依表可知，东部地区样本公司的管理层能力均值为-0.013， 中部地区样本公司的管理层能力均值为-0.019，西部地区样 本公司的管理层能力均值为-0.034，可见东部地区上市公司 管理层能力高于中部地区以及西部地区的管理层的能力，且 基于中位数的分析结论具有一致性。由此可见，在经济社会 发展越为发达的地区，上市公司的管理层能力越高，可能是 由于高水平的管理人才往往聚集在北京、上海、广东等经济 发达的地区，这与东部地区人才密集的结构特征是一致的。 5. 结论 本文首先基于 dea 模型，利用上市公司的投入、产出数据计 算了上市公司的运营效率。在此基础上，将上市公司运营效 率设定为被解释变量，在系统控制影响运营效率的除管理层 能力外的诸多因素后，在残差项中捕捉管理层能力的信息。 最终形成的主要研究结论可概括为如下两个方面: 第一，上市公司管理层能力的结果显示，我国上市公司管理 层能力的平均值为-0.02， 而且，上市公司的管理层能力的 标准差为 0.12，这意味着管理层能力在不同样本公司间的波 动较大。管理层能力显著影响上市公司的运营效率,管理层所 具有的经营管理能力越强,企业运营不足的程度就会越小,企 业的运营效率也就越高。 第二，我国上市公司管理层能力表现出较强差异性。本文 基于行业特征、规模特征和区域特征，对我国创业板上市 公司管理层能力表现出的差异性进行分析。制造业样本公 司的管理层能力高于非制造业样本公司的管理层能力；大 规模公司的管理层能力高于小规模公司的管理层能力；相 对于经济社会发展落后的中西部地区而言，东部地区样本 公司的管理层能力相对更高，这些数据揭示的规律与我们 的主观判断一致。 conflicts of interest the authors declare they have no conflicts of interest. supplementary material supplementary data related to this article can be found at https:// doi.org/10.2991/jracr.k.201214.002. 参考文献 [1] weifeng h, wei l, kaili h. managerial ability and enterprise risk-taking. china soft sci. 2016;107–18. [2] lijie y, ying z. management ability, innovation level and innovation efficiency. account res 2018;70–7. [3] anwei x. research on the impact of cross-border m&a on corporate management efficiency——an empirical analysis based on propensity score matching method. int trade issues 2018;24–36. [4] bin l. shuqian q. management capability, investment bank reputation and m&a performance response. j beijing jiaotong univ (soc sci ed) 2016;15:61–70. [5] the company law of the people’s republic of china. communiqué of the standing committee of the national people’s congress of the people’s republic of china. 2014, pp. 53–75. [6] qianjin p. research on management capability, earnings quality and capital allocation efficiency. central university of finance and economics; 2015. [7] chemmanur tj, paeglis i, simonyan k. management quality and equity issue characteristics: a comparison of seos and ipos. financial manage 2010;39:1601–42. [8] bonsall iv sb, holzman er, miller bp. managerial ability and credit risk assessment. manage sci 2017;63:1271–656. [9] fuller k, netter j, stegemoller m. what do returns to acquiring firms tell us? evidence from firms that make many acquisitions. j finance 2002;57:1763–93. [10] dhaliwal ds, lamoreaux pt, litov lp, neyland jb. shared auditors in mergers and acquisitions. j account econ 2016;61:49–76. [11] baik b, brockman pa, farber db, (sunghan) lee s. managerial ability and the quality of firms’ information environment. j account audit finance 2018;33:506–27. [12] lee cc, wang cw, chiu wc, tien ts. managerial ability and corporate investment opportunity. int rev financial anal 2018;57:65–76. [13] bamber ls, (xuefeng) jiang j, wang iy. what’s my style? the influence of top managers on voluntary corporate financial disclosure. account rev 2010;85:1131–62. [14] shihua c, guangsheng j, changchong l. director connection, target company selection and m&a performance——based on the research perspective of information asymmetry between m&a parties. manage world 2013;117–132+187–188. [15] lijun m, wei z. ceo competence and corporate performance——based on the data of ceo resignation of chinese listed companies. oper res manage 2012;21:200–7. [16] leverty jt, grace mf. dupes or incompetents? an examination of management’s impact on firm distress. j risk insur 2012;79:751–83. [17] chen d, bu d. management ability, power characteristics and salary gap. j shanxi univ finance econ 2015;37:91–101. [18] demerjian p, lev b, mcvay s. quantifying managerial ability: a new measure and validity tests. manage sci 2012;58:1229–48. [19] yuhui w, shinong w, zhihua w. management ability, information disclosure quality and corporate credit rating. econ manage 2017;39:165–80. https://doi.org/10.2991/jracr.k.201214.002 https://doi.org/10.2991/jracr.k.201214.002 https://doi.org/10.1111/j.1755-053x.2010.01124.x https://doi.org/10.1111/j.1755-053x.2010.01124.x https://doi.org/10.1111/j.1755-053x.2010.01124.x https://doi.org/10.1287/mnsc.2015.2403 https://doi.org/10.1287/mnsc.2015.2403 https://doi.org/10.1093/rfs/hhg031 https://doi.org/10.1093/rfs/hhg031 https://doi.org/10.1093/rfs/hhg031 https://doi.org/10.1016/j.jacceco.2015.01.005 https://doi.org/10.1016/j.jacceco.2015.01.005 https://doi.org/10.1016/j.jacceco.2015.01.005 https://doi.org/10.1177/0148558x17742820 https://doi.org/10.1177/0148558x17742820 https://doi.org/10.1177/0148558x17742820 https://doi.org/10.1016/j.irfa.2018.02.007 https://doi.org/10.1016/j.irfa.2018.02.007 https://doi.org/10.2308/accr.2010.85.4.1131 https://doi.org/10.2308/accr.2010.85.4.1131 https://doi.org/10.2308/accr.2010.85.4.1131 https://doi.org/10.1111/j.1539-6975.2011.01443.x https://doi.org/10.1111/j.1539-6975.2011.01443.x https://doi.org/10.1111/j.1539-6975.2011.01443.x http://dx.doi.org/10.1287/mnsc.1110.1487%0d http://dx.doi.org/10.1287/mnsc.1110.1487%0d 146 f. xia and z. mu / journal of risk analysis and crisis response 10(4) 138–146 [20] tiezhu z, sharman. study on management ability, power and in-service consumption. nankai manage rev 2014;17:63–72. [21] jian x, guoping t, yurou x. the ability of management, the nature of property rights and corporate tax avoidance. j jiangxi univ finance econ 2016;43–59. [22] dunli z, xinfeng j. managerial ability and corporate investment herding behavior: the moderating effect based on salary equity. account res 2015;41–8. [23] he w, liu w. corporate manager ability and audit fees. account res 2015;82–9. [24] xu n. management capability and internal control: empirical evidence from listed companies in china. audit res 2017;80–8. [25] shen l, guo y. managerial competence and internal control quality: restrain or promote?. j zhongnan univ econ law 2017;58–67+159. [26] xiao m, li h. research on the impact of management capability on corporate m&a. manage world 2017;184–5. [27] charnes a, cooper ww, rhodes e. measuring the efficiency of decision making units. eur j oper res 1978;2:429–44. [28] tobin j. estimation of relationship for limited dependent variables. econometrica 1958;26:24–36. [29] goldberger as. econometric theory. new york: john wiley & sons, inc.; 1964. [30] yue x, liu c, xu m, ma c. why should we pay attention to management’s forecasting ability. china asset appraisal 2019;4–9. https://doi.org/10.1016/0377-2217(78)90138-8 https://doi.org/10.1016/0377-2217(78)90138-8 a study of the impact on soybean potential under climate change received 31 july 2016 accepted 25 march 2016 通讯作者：樊冬丽 ，女，1976 年出生，山西太原人，讲师，博士，主要从事科学研究。通讯地址：201418，上海市奉贤区海泉路 100 号， 上海应用技术学院，e-mail：fandongli2002@163.com。 a study of the impact on soybean potential under climate change qiuying ding1, zhan tian1,2, dongli fan1, laixiang sun2,3, guenther fischer4 1shanghai institute of technology, shanghai 21400, china 2shanghai climate center, shanghai meteorological bureau, shanghai 21400, 20030, china 3department of geographic sciences, university of maryland, college park 20742, usa 4international institute of applied system analysis, laxenburg 2361, austria abstract soybean is one of the important oil crops in china. however the supply and demand of soybean is at stake currently. the demand keeps increasing and the self-sufficient keeps decreasing. more seriously, climate change will bring obvious impact on the growth and development, planting pattern, planting area, potential production of soybean, etc. therefore, assessment of the impact of soybean production under climate change is quite essential for improving the self-sufficient and guaranteeing the safety of oil crops. this study will extend and improve the parameters of soybean in agricultural ecology zone (aez) based on the 22 soybean observation stations in the major planting area from 1981-2011 to achieve china-aez. and then simulate the impact of climate change on soybean. the results show that: the simulation of china-aez has been improved a lot. in 2050s, the total soybean potential will increased by 7123 thousand tons. the total suitable planting area will increased by 3589 thousand hectare. but the average potential will decreased by 55 kg/ha. from the spatial scale, the soybean potential will increase in northeast china and northwest china. soybean potential will decrease in the other area of china under climate change. keywords: soybean, climate change, aez, production potential 气候变化对我国大豆生产潜力的影响研究 丁秋莹 1, 田展 1,2, 樊冬丽 1, 孙来祥 2,3, guenther fischer4 1 上海应用技术学院，上海 21400，中国 2 上海气象局上海市气候中心，上海 20030,中国 3 马里兰大学帕克分校地理科学系，大学公园市 20742，马里兰州，美国 4 国际应用系统研究所，拉克森堡 2361，奥地利 摘 要：大豆是我国主要的油料作物之一。然而，目前我国大豆供给形势不容乐观。一方面需求增加，自给 率较低；另一方面，气候变化将会影响我国大豆生长发育、种植模式、种植布局、生产潜力等，因此客观 准确地评价气候变化对我国大豆生产的影响为提高我国大豆自给率、保障油料安全具有重大意义。本研究 基于 1981-2011 年我国大豆主产区的 22 个农业气象站点观测数据，扩充并改进 aez(agricultural ecology zone) 大豆的品种库参数，利用 aez 模型模拟了未来气候变化对我国大豆生产潜力的影响。结果显示：调整改进 后的 china-aez 模型对我国大豆主产区的区域模拟能力得到了较大的提高。总体来看，气候变化影响下，到 2050s 我国大豆总生产潜力将会比 1990s 增加 7123 千吨，约为 4.15%。大豆总适宜面积增加 3589 千公顷， 约为 6.15%。平均单产生产潜力减少 55kg/ha，约为 1.89%。区域上来看，气候变化将会引起东北以及西北 地区大豆生产潜力增加，但其他区域均小幅减产。 关键词：大豆；气候变化；aez；生产潜力 添加：国家自然基金（41371110，4167113 和 41601049）和中国气象局气候变化专项（ccsf2011330、 ccsf201110）资助。 journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 95-102 published by atlantis press copyright: the authors 95 q. ding et al. / a study of the impact on soybean potential under climate change 我国油脂资源短缺，食用油供给形势严峻，油 料长期存在着严重的供需矛盾。其中大豆是一种很 有价值的油料作物，是重要的畜饲料，生物燃料， 蛋白质等来源 [1] ，随着过去几十年间我国经济发展以 及人们生活水平的改善，大豆需求量显著增加，并 且预计这种增长趋势在未来还会持续 [2] 。然而多年来 我国油料作物生产潜力徘徊不前，大豆生产潜力也 连续走低 [3] 。2003年开始，我国进口大豆数量首次超 过国产大豆。未来气候变化将会严重影响我国大豆 生产，进一步加大大豆生产的不稳定性，供需矛盾 还将进一步加剧。因此明确气候变化对大豆生产的 影响，对保障国家食用油供给安全具有实际的重要 意义。 过去短时期内全球气候发生了迅速变暖，我国 年平均地表大气温度在过去50年内升高了1.1℃[4]， 这一增幅要高于全球地表平均温度变化。全国平均 降水总量变化不显著，年代际波动较大，降水趋于 集中，全国平均暴雨和极端强降水事件的频率和强 度都有所增长 [5,6] 。并且气候变暖会增加蒸散 [7] ，干 旱发生的频率和强度都会加强 [8] 。此外，随着全球变 暖，异常偏冷性事件减少减轻；而异常偏暖性极端 事件增多增强；气候暖干化趋势会加重病害的发生 [9] 。近年来极端气候事件发生的形式也更加多样性 [10] 。气候变化势必会给农业生产带来严重影响。干 旱、极端天气等增加了农业生产的风险，也将加大 大豆生产潜力的波动性。气候变暖影响下，东北地 区作物生育期延长，作物种植边界北扩，能够促进 大豆的种植边界北移，有利于扩大高纬度地区大豆 的种植面积 [11] 。 从目前相关研究来看，作物模拟模型是一种较 为系统综合的分析方法，能够综合评估农业系统与 土壤、气候等因子之间的相互作用。随着科学技术 的发展进步和消费需求的增加，作物模型无论在研 究深度还是应用广度上都取得了显著的成就。aez 模型是一种空间尺度的土地生产潜力评估模型，该 模型在模拟作物生产潜力时，采用的自动匹配算法 能够依据气候资源数据和lut（土地利用单元）属性 数据逐格点尝试所有可能性，因此模型能够自动筛 选出最优的作物种植品种、种植模式等，这使得aez 模型在区域作物潜在生产力评估中具有明显优势。 本研究基于我国多年大豆观测数据通过改进模型的 品种库参数以及算法，建立适合我国大豆评估的 aez-china模型，评估1990s（1981-2010）baseline条 件下与2050s(2041-2070)未来气候变化下大豆生产潜 力。 1. 研究资料与方法 1.1. 研究资料 土壤数据是从 hwsd（harmonized world soil database）土壤数据库中提取而来，该数据库提供全 球范围内 1km×1km 分辨率格网水平的土壤信息， 土壤数据分为顶层（0-30cm）和下层（30-100cm）， 土壤属性包括排水速率、土壤深度、容积密度、有 机碳有机质含量、土壤 ph 以及阳离子交换量等数据 信息，可以供模型直接提取使用。 基准条件下的逐日气候观测数据来自国家气候 中心，全国共设有 743 个气象站点，主要的气象要 素包括逐日最低气温、最高气温、降水、日照时数、 相对湿度以及风速，本研究选用 1981-2010 的气候数 据作为基准时段。未来气候变化数据是对未来温室 气体排放的不同情景进行的假设，已有研究表明 ipcc ar4 气候模式结果对东亚和中国的气候变化具 有较好的模拟能力 [12] ，设计了多种排放情景，对比 研究表明，近年来温室气体增长率与 sres a1b 情景 下的温室气体排放水平最为接近，能够一定程度上 代表未来气候发展的方向，因此本研究中采用 a1b 排放情景作为未来气候变化的趋势。 表 1 的大豆观测数据来自中国气象局信息中心 多年观测整理所得，涵盖了全国主要大豆种植省份。 根据站点记录数据的完整性以及地理位置分布，从 中国气象局提供的数据中挑选 22 个大豆站点，数据 年份包括 1981-2011 年，数据内容包括：站点基本信 息（站点名称、经纬度、海拔高度），栽培信息（种 植品种、熟性、耕作方式等），生长发育的详细信 息（播种、出苗、开花、成熟等），生产潜力及相 关信息（种植密度、粒重、总生产潜力、茎秆重等） 以及主要的管理措施（施肥、灌溉、收获等）。该 表中的生产潜力是根据站点历年的纪录数据，剔除 个别存在病虫草害的年份以及奇高的产量，奇高生 published by atlantis press copyright: the authors 96 q. ding et al. / a study of the impact on soybean potential under climate change 表 1. 大豆观测站点基本信息 省份 站点 经度 纬度 生产潜力（kg/ha） 生育期（天） 黑龙江 德都 127.35 46.08 4959 134 黑龙江 巴彦 126.15 48.47 5448 137 黑龙江 嫩江 125.23 49.17 4234 145 辽宁 阜新 121.72 42.08 3243 134 辽宁 盖州 122.35 40.42 4216 131 辽宁 海城 122.72 40.88 4124 135 吉林 辽源 125.08 42.92 4067 143 内蒙古 扎兰屯 122.73 48.00 4946 135 陕西 绥德 110.22 37.50 2499 166 陕西 延安 109.50 36.60 4088 154 新疆 莎车 77.10 38.43 3711 143 河北 黄骅 117.35 38.37 4692 95 河南 国营 114.40 33.75 4071 117 山东 莒县 118.83 35.58 3015 89 江苏 丰县 116.58 34.68 3729 103 江苏 盱眙 118.02 33.00 3445 106 安徽 蒙城 116.53 33.28 4132 100 安徽 寿县 116.78 32.55 2345 103 江西 龙南 114.82 24.92 3009 102 江西 南康 114.75 25.67 2872 96 江西 泰和 114.92 26.80 3045 95 湖南 怀化 107.97 27.55 2820 101 产潜力可能是存在记录误差，个别年份的记录生产 潜力在 8000kg/ha，根据查阅到的大豆最高生产潜力 在 6000 kg/ha 左右。然后从剩余观测年份中挑选中值 作为每个站点大豆观测生产潜力。从观测数据来看， 东 北 地 区 特 别 是 黑 龙 江 省 大 豆 产 量 较 高 ， 在 4000kg/ha 以上，而南方大部分地区包括江苏、安徽、 江西的大部分观测站点大豆产量较低，在 3000kg/ha 左右。生育期长度呈现北长南短的分布规律。在灌 溉方式上，观测数据中扎兰屯、蒙城、莎车站点有 灌溉纪录，其他多数站点为雨养种植模式。 1.2. 研究方法 aez模型同时考虑光、温、水、土、投入水平等 影响生物生产潜力形成的因素及指标。该模型能够 进行土地适宜性以及土地生产潜力评价，能够得到 农业生态区划图、土地适宜性评价图以及农作物生 产潜力等信息，为气候变化、土地承载力等研究提 供了依据。根据假定的投入水平、灌溉供给以及管 published by atlantis press copyright: the authors 97 q. ding et al. / a study of the impact on soybean potential under climate change 理条件，定义特定作物在气候、土壤、地形以及海 拔因素引起的限制因子，以此计算得到作物的最大 生产潜力和生产潜力。以土地利用单元（lut）作为 计算潜在生产力的基本单位，每一种土地利用单元 包含的信息有：收货系数、最大叶面积指数、最大 光合速率等，耕作方式和投入需求，作物残茬和作 物副产品系数。采用的自动匹配算法会依据气候资 源数据和lut属性数据逐格点尝试所有可能性，因此 模型能够自动筛选出最优的作物种植品种、种植模 式等，这使得aez模型在区域作物潜在生产力评估中 具有明显优势，主要步骤是： （1）通过气候和土地资源的清查，建立起研究 区域内逐格点中每个评价单元的气候（温度、降水、 风速、日照时数、相对湿度）、土壤、地形等资源 数据库，同时定量评估区域内农业气候资源，包括 积温、湿润指数等； （2）根据已有资源数据库及其他lut属性信息， 进行作物与环境匹配，计算所有作物每个lut的光温 生产潜力，以及在土壤、地形、水分、管理水平等 限制条件下的作物最高生产潜力； （3）计算由于其它限制因子导致的作物生产潜 力损失，如干旱、霜冻、作物病虫害等，并进一步 考虑土壤与地形因素对作物最终潜在生产潜力的影 响。在模拟我国小麦生产潜力中已得到应用 [13] 。但 该模型对农作物生长过程的机理性解释相对而言较 为粗略，作物生长的微观基础比较薄弱。 2. aez 模型的验证 根据大豆观测数据扩充 aez 模型的大豆品种 库，同时结合大豆的观测数据进行对比，调整大品 种系数。此外根据大豆的观测数据的空间分布，基 于生态系统类型调整 aez 模型的算法，形成适应于 我国大豆生产的 aez 模型版本。以此建立跨尺度模 拟的中国大豆评估模型 china-aez[14]，综合评估未 来气候变化影响下我国大豆生产潜力的变化。 （1）aez 模型验证 在 aez 模型中，生育期长度是品种的一个关键 属性，首先对比原 aez 模型得到的生育期分布图与 实际观测值进行比较，验证原始 aez 模型对我国大 豆的模拟能力。在 aez 模型中，生育期作为 lut 的 一个重要属性，根据大豆普遍生育期长度，以 15 天 间隔分成几个代表性生育期。为更明显的比较观测 值与 aez 模型模拟的结果，将观测站点的品种生育 期长度根据 aez 模型的品种库分为 4 类。从图 1 生 育期长度的对比中，我们可以看到两个主要的不同 点： 1）观测站点中存在 150 天生育期的品种，而在 原始 aez 模型的品种库中仅有其他三种类型（105 天，120 天，135 天）的生育期长度的品种；2）原 始 aez 模型模拟的结果显示在南方、华北大部分地 区选择生育期较长的品种，而在东北地区则选择了 生育期较短的品种，呈现为南长北短的趋势，而实 际观测结果与 aez 模型模拟的结果相反，呈现南短 图 1. 1990s 下原大豆生育期长度分布实际情况（左）以及 aez 模型模拟情况（右） published by atlantis press copyright: the authors 98 q. ding et al. / a study of the impact on soybean potential under climate change 北长的趋势，这主要是因为 aez 模型是采用的自动 算法，所考虑的因素主要是气候、土壤等客观因素 的影响，没有考虑区域内复种指数、作物的轮作等 种植者决定的主观因素，而在我国大部分地区，大 豆并不是一种主作物，农民会优先种植小麦、水稻 等收益较大的作物，从而导致大豆可种植的生育期 就会缩短。 （2）aez 模型的调试 针对原始 aez 模型的不足，通过不断试验模拟， 与观测值进行比较验证。本研究中我们做了两点主 要的改进：1)在 aez 模型的品种库中增加生育期为 150 天的品种，并根据观测数据补充和改进 aez 模 型的其他相关品种参数，包括收货系数、最大叶面 积指数、积温等，使之与我国大豆种植生长发育情 况更为吻合。从表 2 中改进的品种参数结果来看， 原有品种收货系数和最大叶面积指数整体上增加， 可增加模拟的生产潜力；适宜最低积温降低、适宜 最高积温提高，扩大大豆生长发育对温度的要求范 围，使之适宜于我国南方热量资源丰富的地区。 aez 模型能够根据气候、土壤资源生成八种不 同的耕作系统类型，通过比较观测值与耕作系统的 关系，发现大豆生育期长度与耕作系统分布之间存 在一定的相关性，即在单作区内大豆观测生育期长 度较长而在多熟区内大豆生育期较短。基于作物耕 作系统调整 aez 模型的自动算法，使其在一个耕作 系统内根据限制品种的选择，一种作物耕作系统内 选择特定的品种，代替原有的完全最优自动算法。 根据表 3 进行品种选择：其中单作区内大豆生 表 2. aez 模型中大豆品种及主要品种参数变化对比（原：原 aez 模型；改：改进后的 aez 模型） 品种 生育期长 度 收获 系数 最大叶面 积指数 适宜生育期 最低积温 次适宜生育 期最低积温 适宜生育期 最高积温 次适宜生育 期最高积温 温带和亚热带大 豆 1（原） 105 0.3 4.0 2200 1850 2600 3150 温带和亚热带大 豆 1（改） 105 0.38 4.0 2100 1850 2800 3150 温带和亚热带大 豆 2（原） 120 0.35 2.5 2400 2000 3000 3600 温带和亚热带大 豆 2（改） 120 0.37 4.2 2300 2000 3150 3500 温带和亚热带大 豆 3（原） 135 0.35 3.0 2600 2150 3400 4050 温带和亚热带大 豆 3（改） 135 0.36 4.5 2500 2150 3450 3800 温带和亚热带大 豆 4（增） 150 0.35 4.5 2700 2300 3750 4100 表 3. aez 模型不同耕作系统品种的选择 耕作系统类型 生育期长度类型的选择 no cropping 从四种类型中选择最优生育期 single cropping limited cropping double cropping 从 105 天或 120 天选择较优生育期 double cropping with rice double rice cropping 选择 105 天生育期长度 triple cropping triple rice cropping published by atlantis press copyright: the authors 99 q. ding et al. / a study of the impact on soybean potential under climate change 育期长度的选择不受限制，模型根据大豆最大生产 潜力选择品种类型，一般情况下，在气候资源丰富 的条件下，作物生产潜力与生育期长度之间有一定 的相关性，优先选择生育期长的品种；在两熟区， 大豆生育期长度的选择受轮作影响，限定在 105 天 和 120 天之间选择；在三熟区，大豆生育期的选择 更短，仅限定在 105 天。 3. 结果分析 3.1. 大豆生产潜力模拟评估 我国大豆种植方式以雨养为主，因此本研究主 要评估雨养条件下大豆生产潜力和适宜区划。基准 气候条件下，aez 模拟得到的我国大豆主产区集中 在东北平原、华北平原、长江中下游平原以及四川 盆地等地势较为平坦的地区。模型模拟的结果与我 国大豆主产区有很好的一致性，历史相关资料表明 我国大豆主产区包括北方春大豆（包括东北地区、 黄土高原、西北地区）、黄淮海夏大豆（包括晋冀 中部、黄淮海流域）、长江流域春夏大豆（包括长 江流域、云贵高原）、东南春夏秋大豆（浙江、福 建、江西、台湾、湖南、广东、广西大部）以及华 南四季大豆（广东、广西、云南和福建南部）。图 2 模拟 1990s 大豆生产潜力分布：其中大豆生产潜力最 大的区域位于东北省境内，很多区域在 4000kg/ha 以 上，这与该区属于一熟制，大豆生育期较长有关， 模拟的大豆生产潜力与观测的大豆生产潜力较为相 近，特别是黑龙江省的德都、巴彦站点附近区域大 豆生产潜力处于极高水平；华北平原的山东、河北、 河南部分区域大豆生产潜力接近 4000kg/ha，而实际 观测中河北的黄骅、河南的国营大豆生产潜力也高 于 4000kg/ha；虽然南方水热资源丰富，但改进后的 aez 模型对品种的选择受区域内熟制分布的限制， 考虑多熟制轮作的影响，因而不能充分利用气候资 源，大豆生产潜力不如东北地区高，在 3000kg/ha 左 右，与观测的站点（安徽的蒙城、寿县，江西的龙 南、南康、泰和以及湖北的怀化）的大豆生产潜力 相近。以上结果表明改进后的 aez 模型模拟得到的 大豆生产潜力空间分布能够较好的反映我国观测种 植大豆生产潜力情况，aez 对我国大豆生产具有较 好的模拟能力。 图 3 结果表明：到 2050s 未来气候变化下，在使 用最优播期和限制性的适宜品种的情况下，总体上 气候变化对大豆生产的影响呈现北增南减的趋势， 东北以及华北部分地区受全球变暖的影响，冻害、 大雪等低温灾害发生频率减少，有相当一部分地区 大豆增产 500kg/ha 以上，且黑龙江、内蒙古以及河 北省部分地区增产幅度在 1000kg/ha，仅有黑龙江西 部以及辽宁中部地区有小幅减产，均少于 250 kg/ha； 华北、华中南部以及南方大部分地区呈现减产趋势， 主要原因是华北地区处于熟制交界地带，气候变暖 下该区熟制增加，缩短了大豆适宜生育期长度的选 择，导致大豆减产；而华中南部以及南方地区可能 是由于该区热量资源本身较为丰富，加上气候变暖， 高温热浪的灾害增加，不利于作物生长，导致大豆 生育期缩短，从而导致大豆减产，但是大部分地区 减产幅度较小，均低于 500 kg/ha；山东、河北、河 南以及安徽、江苏北部等处于熟制过度地带，大豆 生育期长度变化较大，大豆减产趋势较为明显，高 于 500 kg/ha。 3.2. 大豆适应种植区域分析 表 4 结果显示，气候变化影响下，我国大豆适 宜种植面积增加 3589 千公顷，总产量增加 7123 千 吨，但是大豆单位面积产量减少 55kg/ha。 aez 模型将我国分为华北、东北、华东、华中、 东南、西南、西藏高原、西北八个区域统计分析， 其中华北地区包括北京市、天津市、河北省、山西 省、山东省、河南省和湖北省，东北地区包括辽宁 省、吉林省和黑龙江省，华东地区包括上海市、江 苏省、浙江省和安徽省，华中地区包括江西省和湖 南省，东南地区包括福建省、广东省、广西省和海 南省，西南地区包括重庆市、四川省、贵州省和云 南省，高原地区包括西藏自治区和青海省，西北地 区包括新疆自治区、内蒙古自治区、陕西省、甘肃 省和宁夏自治区。 在使用了最优播期和最优限制性品种选择计算 得到的大豆生产潜力中，表 5 统计结果显示，雨养 条件下，大豆主产区集中在华北、东北、西南和西 北地区，1990s 以上地区大豆产量占大豆总产的 published by atlantis press copyright: the authors 100 q. ding et al. / a study of the impact on soybean potential under climate change 表 4. 大豆生产总体统计结果 时间 适宜面积（千公顷） 总产（千吨） 单产（kg/ha） 1990s 58332 171616 2942 2050s 61921 178739 2887 表 5. 大豆生产分区域统计结果 区域 适宜面积（千公顷） 总产（千吨） 单产（kg/ha） 1990s 2050s 1990s 2050s 1990s 2050s 华北 12645 13029 35337 33984 3105 2898 东北 15461 15538 57274 63526 4116 4543 华东 4250 4271 11981 10335 3133 2689 华中 2253 2253 6146 5556 3031 2739 华南 4074 4138 9628 8615 2626 2313 西南 11253 12260 29646 28687 2927 2600 青藏高 原 40 125 93 278 2561 2472 西北 8356 10307 21511 27758 2860 2993 71.2%。1990s 下，华北和东北地区是大豆总生产潜 力最高的地区，占总生产潜力的 52.3%，气候变化影 响下大豆总产有小幅增加，为 7123 千吨，除西北和 东北两个地区，其他地区大豆总生产潜力均有减少。 适宜面积增加 3589 千公顷，增加明显的地区为华北、 西南和西北地区，该区位于高纬度地区和山区，主 要驱动因素可能是气候变暖影响下，该区热量资源 增加。从平均生产潜力来看，1990s 下大豆均产最高 的地区位于东北，平均生产潜力高于 4000kg/ha, 其 他地区在 3000kg/ha 左右，气候变化下大豆总体平均 生产潜力降低 55 kg/ha，除东北和西北地区大豆均产 有增加外，其他地区大豆均有小幅减产。 4. 结论及讨论 本研究基于大豆观测值对 aez 模型的大豆品种 库参数进行有效的扩充和改进，改进后的 aez 模型 能够更为准确地模拟我国大豆生产潜力及种植区 划。未来气候变化下，总体来看气候变化有利于东 北、西北以及华北北部地区大豆生产，但不利于华 北大部、南方大部分大豆生产。统计结果显示大豆 总产和总适宜 面积增加，平均生产潜力减少。区域上来看，气候 变化下带来的热量资源增加有利于东北及西北地区 大豆生产，华北以及南方大部分地区均有小幅减产。 虽然本研究中 aez 模型对大豆遗传参数进行了 调整改进，但对适宜播期的模拟仍然是采用最优播 期，并未考虑多熟制内其他主作物对大豆种植播期 的限制，今后还需进一步探讨多熟制内的作物轮作 模式，以期得到更为准确的结果。 参考文献 [1]masuda t, goldsmith p d. world soybean demand: an elasticity analysis and long-term projections. national soybean research laboratory university of illinois at urbana–champaign, urbana, illinois, 2009. [2]masuda t, goldsmith p d. world soybean production: area harvested, yield, and long-term projections. international food and agribusiness management review, 2009, 12(4): 143-162. 图 2. 1990s 大豆生产潜力模拟 图 3. 2050s 大豆生产潜力的变化（kg/ha）. published by atlantis press copyright: the authors 101 q. ding et al. / a study of the impact on soybean potential under climate change [3]李秀娟, 刘喜元, 李晓伟, 等.2009年呼伦贝尔市岭东地 区大豆生产的气象条件分析. 现代农业科技, 2010, (20):297, 299. [4]ding y h, ren g y, zhao z, et al. detection, causes and projection of climate change over china: an overview of recent progress. advances in atmospheric sciences, 2007, 24(6):954-971. [5]任国玉, 封国林, 严中伟. 中国极端气候变化观测研究 回顾与展望. 气候与环境研究, 2010, 15(4):337-353. [6]王会军, 孙建奇, 祝亚丽. 中国极端气候及东亚地区能 量和水分循环研究的若干近期进展. 自然杂志, 2012, 34(1):10-17. [7]左德鹏, 徐宗学, 李景玉, 等. 气候变化情景下渭河流 域潜在蒸散量时空变化特征 . 水科学进展 , 2011, 22(4):455-461. [8]gamble d w, campbell d, allen t l, et al. climate change, drought, and jamaican agriculture: local knowledge and the climate record. annals of the association of american geographers, 2010, 100(4):880-893. [9]王丽, 霍治国, 张蕾, 等. 气候变化对中国农作物病害 发生的影响[j]. 生态学杂志, 2012, 31(7):1673-1684. [10]chen h, zhang h, xue c. chinese extreme climate events and agricultural meteorological services. challenges and opportunities in agrometeorology. springer berlin heidelberg, 2011:435-459. [11]田展, 丁秋莹, 梁卓然,等. 气候变化对中国油料作物 的影响研究进展. 中国农学通报, 2014, (15):1-6. [12]施小英, 徐祥德, 徐影. 中国 600 个站气温和 ipcc 模 式产品气温的比较. 气象, 2005, 31(7):49-53. [13]田展,钟洪麟,施润和,等. estimating potential yield of wheat production in china based on cross-scale data-model fusion. frontiers of earth science, 2012, 6(4): 364-372. [14]fan d, ding q, tian z, et al. simulating the adaptive measures of soybean production to climate change in china: based on cross-scale model coupling. emerging economies, risk and development, and intelligent technology: proceedings of the 5th international conference on risk analysis and crisis response, june 1-3, 2015, tangier, morocco. crc press, 2015: 143. published by atlantis press copyright: the authors 102 http://link.springer.com/search?facet-author=%22chen+huailiang%22 http://link.springer.com/search?facet-author=%22zhang+hongwei%22 http://link.springer.com/search?facet-author=%22xue+changying%22 1. 研究资料与方法 2. aez模型的验证 3. 结果分析 4. 结论及讨论 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style received 26 september 2015 accepted 20 january 2016 high-rise building group regional fire risk assessment model based on ahp wei zhu, qiuju you beijing research center of urban systems engineering, beijing 100035, p.r. china beijing key laboratory of operation safety of gas, heating and underground pipelines, beijing 100035, p.r. china e-mail: zhuweianquan@126.com, yqjbyq@163.com abstract based on analytic hierarchy process (ahp), a regional fire risk assessment model is developed for high-rise building group considering the fire danger degree, basic characters of the high-rise building group and fire control capability. the impact factors, values of typical risk parameter and the weight of each index are analyzed. an assessment program is further developed using matlab, to perform the risk assessment procedures. the model is then applied to the regional fire risk evaluation of a high-rise building group. keywords: high rise building, fire, risk assessment, ahp 1. introduction with the development of city construction and the increase of population density, land is becoming a limited resource. in order to overcome the problem, more and more high-rise buildings are built. high-rise building fires have many special characteristics, such as the diversity of blazing, factors, various ways of fires spreading, the difficulty of evacuation and saving activities. due to the importance of fire safety for the high-rise building, many studies have been conducted to analyze the fire risk of the high-rise building. chen et al.(chen et al. 2012) has analyzed the fire accident using the accident tree analysis method and the principal theory of safety system engineering. liu et al.(liu et al. 2012) has analyzed the factors that can cause high-rise building fires, and a partial event tree has been built. sun et al.(sun & luo, 2014) has madea case study for a super high-rise building to expatiate the procedure and methodology of fire risk assessment for super high-rise buildings. cowlard et al.(cowlardet al.2013) has made some studies on the fire safety design for tall buildings. for a high-rise building group in a community, the fire safety is more important. because of the complex of itself, such as more high-rise buildings, limited spacing between buildings, once a fire happened, fire spreading may arise. as a typical disaster in urban public safety, high-rise building group fire should attract more attention and be treated seriously. however, very limited work has been reported for fire risk assessment of high-rise building group. here, based on the characteristic of high-rise building group, analytic hierarchy process (ahp) is used to develop a regional fire risk assessment model for high-rise building group. as a risk assessment method, ahp has been used by many researchers in evaluation of fire risk, see (wang et al.2012; ren,2012; zhang, 2013; gao, 2014; and omidvariet al. 2015) for more details. then the model is applied to the fire risk evaluation of a high-rise building group. 2. analytic hierarchy process the first step of ahp is to establish ahp structure index system. the ahp structure index system can be divided into several layers according to the influence factors, and the first layer only has one factor. after that the structure judgment matrix should be built. for the factors which are below a same upper factor are compared with each other to confirm the weights of each factors for the same upper factor based on table 1, and then the structure judgment matrix is obtained as eq. (1). 11 12 1 21 22 2 1 2 3 ⋅ ⋅ ⋅   ⋅ ⋅ ⋅ =  ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅ ⋅     n n n n n nn a a a a a a a a a a a (1) journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 31-37 published by atlantis press copyright: the authors 31 wei zhu and qiuju you / building group fire risk whereaij is the result, n is the number of factors. obviously, aij>0, aii=1, and aij=1/aji. in order to obtain the weight of each index, the eigenvector and the latent root of matrix should be counted, and the sum and product method is used as shown below. maxλ=aw w (2) wherew is the eigenvector,λmaxis the latent root of matrix, and they can be calculated as 1 1 1 1, 2, ,= = = = = ⋅ ⋅ ⋅ ∑ ∑ ∑ ij j i n n kj k j a w i n a (3) ( ) max 1 λ = = ∑ n i i i aw nw (4) then the consistency of the matrix should be tested with the index cr, = ci cr ri (5) max 1 λ − = − n ci n (6) and the value of ri is shown in table 2 for different matrix order. table 2. ri for different matrix order order 1 2 3 4 5 6 7 ri 0 0 0.58 0.90 1.12 1.24 1.32 if cr<0.10, the consistency of judgment matrix is good and the eigenvector is the weight of each index, or, the judgment matrix should be rebuilt. 3. a fire risk assessment model for high-rise building group region 3.1 comprehensive assessment factors based on the causes of high-rise building group, the comprehensive factors are shown in fig 1. the hierarchical model is divided into four layers. the first layer is the target layer, and the only factor was fire risk of high-rise building group. the second and third layer were criterion layers, the second layer include five factors. every factor contains its subset of indicators. the last layer is index layer, and the model includes twenty five indexes. 3.2weight of each factors according to the hierarchical model (fig. 1), the judgment matrix can be built. based on table 1, the results of the pair-wise comparison of the target layer are expressed as eq.7. and using eqs. 2-4, the weight of each factors in second layer ua, ub, uc, ud, ue is obtained w=(0.313 0.141 0.234 0.234 0.078). here, in order to make the operation convenient, a procedure is developed using matlab as shown in fig. 2. 1 2 2 2 3 1 2 1 1 2 1 2 2 1 2 2 1 1 3 1 2 2 1 1 3 1 3 1 2 1 3 1 3 1        =        a (7) table 1. importance degree. i,j compare aij signification iis as important as j 1 iis as important as j iis a little more important than j 3 the former factor is a little more important than the latter i is obviously more important than j 5 the former factor is obviously more important than the latter iis much more important than j 7 the former factor is much more important than the latter iis extremely more important than j 9 the former factor is extremely more important than the latter 2,4,6,8 the middle value between the above neighboring judgments published by atlantis press copyright: the authors 32 wei zhu and qiuju you / building group fire risk result of subset ua,ub,uc,ud, are shown respectively as eqs. 8-11. and the weight of the factors in layer three is obtained, and wa=(0.4 0.2 0.4); wb=(0.297 0.257 0.057 0.178 0.112 0.099); wc=(0.065 0.218 0.262 0.105 0.175 0.175); wd=(0.072 0.339 0.589). 1 2 1 1 2 1 1 2 1 2 1    =      aa (8) 1 1 5 2 3 3 1 1 3 2 3 3 1 5 1 3 1 1 3 1 2 1 2 1 2 1 2 3 1 1 3 1 3 1 3 2 1 1 1 1 3 1 3 2 1 3 1 1         =            ba (9) 1 1 2 1 3 1 2 1 3 1 3 2 1 1 2 2 2 3 1 1 3 2 2 2 1 2 1 3 1 1 2 1 2 3 1 2 1 2 2 1 1 3 1 2 1 2 2 1 1 ca         =           (10) 1 1 5 1 7 5 1 1 3 7 3 1 da    =      (11) result of subset ua1,ua2,ua3,ub6, are shown respectively as eqs. 12-15. and the weight of the factors in layer four is obtained, and wa1=(0.225 0.067 0.539 0.169); wa2=(0.833 0.167); wa3=(0.5 0.5); wb6=(0.333 0.667). 1 3 1 4 2 1 3 1 1 5 1 3 4 5 1 5 1 2 3 1 5 1      =       a1a (12) 1 5 1 5 1   =     a2a (13) 1 1 1 1   =     a3a (14) 1 1 2 2 1   =     b6a (15) according to the hierarchical model and the weight of every layer, the weight of every factor can be achieved. the characters of the buildings ua comprehensive fire extinguishing capability ub evacuation capability uc fire safety management ud others (arson, firecracker) ue internal condition of the building ua1 internal condition of the building ua2 high risk huildings ua3 height of the buildings ua11 finishing material ua12 population density ua13 electrical aging ua14 the spacing between adjacent buildings ua21 fire passage ua22 market ua31 ballroom, ktv, restaurant ua32 fire detecting and automatic alarm system ub1 fire extinguishing system ub2 fire elevator ub3 fire hydrant ub4 part time fire brigade ub5 professional fire brigade ub6 the arriving time of fire brigade ub61 the facility of the fire brigade ub62 fire compartment and smoke bay uc1 refuge storey uc2 smoke control system uc3 fire alarm address system uc4 evacuation distance uc5 evacuation exit and emergency exit uc6 fire-control publicity ud1 facility maintenance ud3 fire patrol ud2 fire risk of high-rise building group u fig. 1. fire risk assessment hierarchical model of the high-rise building group. fig. 2. judgment matrix and consistency. au bu cu du eu weight pass consistency of the matrix au bu cu du eu published by atlantis press copyright: the authors 33 wei zhu and qiuju you / building group fire risk the weight is shown in the following w=(0.028, 0.008, 0.067, 0.021, 0.052, 0.010, 0.062, 0.062, 0.041, 0.036, 0.008, 0.025, 0.026, 0.004, 0.009, 0.015, 0.051, 0.061, 0.024, 0.040, 0.040, 0.016, 0.079, 0.137, 0.078). 3.3comprehensive assessment model we could not quantify fire risk of the high-rise building group only based on the weight of the factors. in this study, the score of each factor of a certain high-rise building group should be determined with the aid of the expert scoring. and table 3 shows the scoring criteria. during the assessment, each factor has a score on the basis of table 3. and then it will be multiplied by its weight. the comprehensive score r of the high-rise building group will be obtained by the summation of all the factors. the relationship between the fire risk grades and the scale is shown detailed as follow 2 ≤r<4 low-grade risk,4 ≤r<6 moderate risk, 6 ≤r<8 high risk, 8 ≤r≤ 10super highrisk 4. application for case study then a high-rise building group has been selected for example. the area of the whole high-rise building group is 12.28 hectares, and the total construction area is 700000 square meters with ground floor total area of 190000 square meters. it contains 18 apartments, 2 office buildings, 4 small offices and a large number of annexes. supporting facilities include kindergarten and club. the height of the highest building is 100m. the scores of each factors is shown in table 4. here, the matlab procedure is also used to make the assessment based on the comprehensive assessment model built in this study, and part of the assessment process is shown in fig. 3. according to the score and weight of every factor the assessment of the fire risk of the high-rise building group can be done as shown in fig. 3. the quantitative result of the building group is 4.85681. the fire risk grade is moderate risk. 5. concluding remarks based on the analytic hierarchy process (ahp), a regional fire risk assessment model is developed for high-rise building group. fire risk assessment hierarchical model is built which contains twenty five impact factors, and the weights of each index are obtained using ahp method. marking criterion of each factor is provided. an assessment program is further developed using matlab, to perform the risk analysis and evaluation procedures. the model is then applied to the regional fire risk evaluation of a high-rise building group. published by atlantis press copyright: the authors 34 wei zhu and qiuju you / building group fire risk table 3. marking criterion of each factor. 2 3 4 factors and score ua ua1 ua11(m) >100 24 -100 <24 10 6 2 ua12 various and messy general less and organized 10 6 2 ua13(person/km 2) >3000 15003000 800 1500 300-800 <300 10 8 6 4 2 ua14 bad general good 10 6 2 ua2 ua21 nonstandard standard for main building standard 10 6 2 ua22 congested partly occupied expedite 10 6 2 ua3 ua31 many and concentrate d some and slightly concentrated general less andslightly dispersed rare and dispersed 10 8 6 4 2 ua32 many and concentrate d some and slightly concentrated general less andslightly dispersed rare and dispersed 10 8 6 4 2 ub ub1 broken normal running with general performance good 10 6 2 ub2 broken normal running with general performance good 10 6 2 ub3 ≤1or broken ≥2 ≥2 and normal running 10 6 2 ub4 insufficient or small cover area sufficient with small cover area sufficient with adequate coverage 10 2 2 ub5 unorganized general organized and high quality 10 6 2 ub6 ub61(min) ≥25 20-25 15-20 10-15 ≤10 10 8 6 4 2 ub62 bad general good 10 6 2 uc uc1 ≥2times of requirement ≥requirement and ≤2times requirement standard 10 6 2 uc2 nonstandard refugestorey but partially standard standard 10 6 2 published by atlantis press copyright: the authors 35 wei zhu and qiuju you / building group fire risk table 4. score of each factor. factors score factors score height of the buildings ua11 6 finishing material ua12 6 population density ua13 10 electrical aging ua14 6 the spacing between adjacent buildings ua21 6 fire passage ua22 6 market ua31 6 ballroom, ktv, restaurant ua32 6 fire detecting and automatic alarm system ub1 2 fire extinguishing system ub2 2 fire elevator ub3 2 fire hydrant ub4 2 part time fire brigade ub5 10 the arriving time of fire brigade ub61 6 the facility of the fire brigade ub62 6 fire compartment and smoke bay uc1 6 refuge storeyuc2 2 smoke control system uc3 2 fire alarm address system uc4 2 evacuation distance uc5 6 evacuation exit and emergency exit uc6 6 fire-control publicity ud1 6 fire patrol ud2 6 facility maintenance ud3 6 arson, firecracker ue 2 table 3. (continued) 2 3 4 factors and score uc uc3 poor performance or broken normal running with general quality good 10 6 2 uc4 poor performance or broken normal running with general quality good 10 6 2 uc5(m) >50 20-50 <20 10 6 2 uc6 congested partly occupied expedite 10 6 2 ud ud1 bad general good 10 6 2 ud2 unscheduled or overlooked scheduled with poor performance scheduled with good performance 10 6 2 ud3 unscheduled or overlooked scheduled with poor performance scheduled with good performance 10 6 2 ue arson, firecracker bad general good 10 6 2 published by atlantis press copyright: the authors 36 wei zhu and qiuju you / building group fire risk acknowledgment this research is supported by this research is supported by national science foundation for distinguished young scholars of china (no. 11202004), national science and technology program of china (no. 2015bak12b00), project of bjast (no. pxm2014-178215-000007). the authors deeply appreciate the support. references h.t. chen, l.l. lou, j.z.qiu, accident cause analysis and evacuation countermeasures on the high rise building fires,international symposium on safety science andengineering in china (procedia engineering), 43 (2012) 23-27. x.y. liu, h. zhang, q.m. zhu, factor analysis of high-rise building fires reasons and fire protection measures, international symposium on safety science and technology (procedia engineering), 45 (2012) 643-648. x.q. sun, m.c. luo, fire risk assessment for super high-rise buildings, procedia engineering, 71 (2014) 492-501. a.cowlard, a. bittern, c.abecassis-empis, j. torero, fire safety design for tall buildings,procedia engineering, 62 (2013) 169-181. q.k. wang, s. pan, on influence factors of wuhan housing industry based on the ahp, systems engineering procedia, 3 (2012) 158-165. s.y. ren, assessment on logistics warehouse fire risk based on analytic hierarchy process,international symposium on safety science and technology (procedia engineering), 45 (2012) 59-63. y. zhang, analysis on comprehensive risk assessment for urban fire: the case of haikou city,procedia engineering, 52 (2013) 618-623. j.p. gao, z.s. xu, d.l. liu, h.h. cao, application of the model based on fuzzy consistent matrix and ahp in the assessment of fire risk of subway tunnel,procedia engineering, 71 (2014) 591-596. m. omidvari, n. mansouri, j. nouri, a pattern of fire risk assessment and emergency management in educational center laboratories,safety science, 73 (2015) 34-42. published by atlantis press copyright: the authors 37 1. introduction 2. analytic hierarchy process 3. a fire risk assessment model for high-rise building group region 4. application for case study 5. concluding remarks acknowledgment references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /usedeviceindependentcolorforimages /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <feff005500740069006c0069007a007a006100720065002000710075006500730074006500200069006d0070006f007300740061007a0069006f006e00690020007000650072002000630072006500610072006500200064006f00630075006d0065006e00740069002000410064006f00620065002000500044004600200070006900f900200061006400610074007400690020006100200075006e00610020007000720065007300740061006d0070006100200064006900200061006c007400610020007100750061006c0069007400e0002e0020004900200064006f00630075006d0065006e007400690020005000440046002000630072006500610074006900200070006f00730073006f006e006f0020006500730073006500720065002000610070006500720074006900200063006f006e0020004100630072006f00620061007400200065002000410064006f00620065002000520065006100640065007200200035002e003000200065002000760065007200730069006f006e006900200073007500630063006500730073006900760065002e> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <feff0055007300740061007700690065006e0069006100200064006f002000740077006f0072007a0065006e0069006100200064006f006b0075006d0065006e007400f300770020005000440046002000700072007a0065007a006e00610063007a006f006e00790063006800200064006f002000770079006400720075006b00f30077002000770020007700790073006f006b00690065006a0020006a0061006b006f015b00630069002e002000200044006f006b0075006d0065006e0074007900200050004400460020006d006f017c006e00610020006f007400770069006500720061010700200077002000700072006f006700720061006d006900650020004100630072006f00620061007400200069002000410064006f00620065002000520065006100640065007200200035002e0030002000690020006e006f00770073007a0079006d002e> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice research article 基于dea的战略性新兴产业上市公司股权融资效率测度 equity financing efficiency measurement of listed companies in strategic emerging industries based on dea si-si li1,2, mu zhang1,* 1school of big data application and economics, guizhou university of finance and economics, guiyang 550025, china 2guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang 550025, china article info article history received 20 august 2020 accepted 12 october 2020 keywords strategic emerging industries listed companies equity financing equity financing efficiency dea methods 关键词 战略性新兴产业 上市公司 股权融资 股权融资效率 dea方法 中图分类号 f832.42 文献标识码 a abstract equity financing is an important part of corporate capital structure decision-making. the level of financing efficiency is of great significance to the survival and development of listed companies in strategic emerging industries. in order to measure the equity financing efficiency of listed companies in strategic emerging industries, this paper uses data envelopment analysis (dea) model to measure and study, taking 208 listed companies of strategic emerging industries as samples, through three input indicators. the paper calculates the dea evaluation value of each decision-making unit from 2014 to 2018, and analyzes the efficiency of equity financing in different years and the comparative analysis among seven industries. the conclusion of this paper is: according to dea–bcc model, the equity financing efficiency of listed companies in strategic emerging industries is generally low, mainly distributed in the lower efficiency range, pure technical efficiency and scale efficiency are mostly distributed in the higher efficiency range every year, most of the returns to scale are decreasing, and the equity financing efficiency exists unbalanced development among industries. according to dea malmquist model, the total factor productivity of seven strategic emerging industries tends to decline, the highest is new energy vehicles, which is 0.985, and the lowest is energy conservation and environmental protection, which is 0.951. the impact of technological change will be greater, followed by the change of scale efficiency. we should constantly improve the internal management level of the company, carry out technological innovation, select the correct financing channels, and coordinate the use of funds to create the maximum value. 摘 要 股权融资是公司资本结构决策的重要内容，融资效率的高低对于战略性新兴产业上市公司的生存和发展具有重要的 意义，为了测度战略性新兴产业上市公司股权融资效率，本文通过使用dea模型来进行测度研究，以208家战略性 新兴产业上市公司为样本，通过3个投入指标、3个产出指标，计算得出各决策单元的2014年-2018年的dea评价值， 并进行效率分析，以及分析股权融资效率在不同年份之间的变化和七大产业之间的比较分析。本文结论为：根据 dea-bcc模型得出，战略性新兴产业上市公司股权融资效率总体比较低，主要分布在较低效率区间，纯技术效率 和规模效率每年大多分布在较高效率区间，规模报酬多数表现为规模报酬递减，股权融资效率存在产业间的非均衡 发展，七大产业中生物医药股权融资效率最高，五年综合效率平均值为0.7558，新能源最低，为0.6463；根据deamalmquist模型结果得出，七大战略性新兴产业的全要素生产率变动总体是趋于下降的，全要素生产率变动最高的是 新能源汽车，为0.985，最低是节能环保，为0.951，全要素生产率变动受技术变动作用的影响会更大一点，其次是规 模效率变动。不断提高公司内部管理水平并进行技术创新，选择正确融资渠道，统筹资金使用为其创造最大价值。 © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/) . journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 101–112 doi: https://doi.org/10.2991/jracr.k.201027.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr *corresponding author. email: rim_007@163.com 1. 引言 改革开放以来，我国的经济飞速增长，无论是在综合国力还 是竞争力方面，都呈明显上升趋势，但是国际金融危机暴露 出我国的代工产业非常薄弱，国家不能完全局限于传统产业， 在金融危机后，国际技术经济也出现新动向。2009年国务院 总理温家宝在9月召开战略性新兴产业发展座谈会，在会上， 温家宝强调，发展战略性新兴产业，是中国立足当前渡难 关、着眼长远水平的重大战略选择。2010年9月审议并原则通 过《国务院关于加快培育和发展战略性新兴产业的决定》， 确定战略性新兴产业的重点发展方向：节能环保、新一代信息 技术、生物、高端装备制造、新能源、新材料和新能源汽车7 个产业。国家随后又推出《战略性新兴产业发展十二五规划》 。2016年11月印发了《“十三五”国家战略性新兴产业发展规 划》，2018年6月签署了《关于共同发起设立战略性新兴产业 发展基金的战略合作备忘录》，提出设立国家级战略性新兴产 业发展基金。表明了发展战略性新兴产业刻不容缓，战略性新 兴产业对经济社会发展和战略全局起到重大引领和推动作用。 习近平总书记在党的十九大报告中就明确指出创新是引领发 展的第一动力，是构建我国现代化经济体系的重要战略支 撑，战略性新兴产业代表新一轮科技革命和产业变革的发展 http://creativecommons.org/licenses/by-nc/4.0/ https://doi.org/10.2991/jracr.k.201027.001 https://www.atlantis-press.com/journals/jracr mailto: rim_007@163.com 102 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 方向，同时也是培育发展新动能获取未来竞争新优势的关键 领域，对于传统产业的改造升级也是重要方面，因此发展战 略性新兴产业对于构建现代化的经济体系，持续增强我国的 创新力和竞争力具有十分重要的意义。国家大力推动供给侧 结构性改革，社会上的金融资本开始从传统行业转向战略性 新兴产业，金融支持效率越高，战略性新兴产业的创新能力 和发展水平就会越高。 战略性新兴产业作为国民经济增长和产业结构转型的核心动 力之一，为了能在世界未来经济发展过程中占据重要地位， 大多数国家都在紧迫地制定其发展战略，我国目前处于战略 性新兴产业初始阶段，在竞争力方面偏弱，战略性新兴产业 在在技术、自主知识产权、资源配置方面等方面出现问题。 我国的上市公司表现出强股权融资偏好的特性，近年来，新 闻里经常爆出上市公司利用股权融资进行圈钱，从而使得上 市公司资源配置效率低，股权融资效率不高，战略性新兴产 业上市公司的股权融资效率会影响其生存发展，如何提高战 略性新兴产业中的金融支持效率，对于公司的良性发展来说 至关重要。企业在资本市场通过发行股票的方式以最低成本 和最小风险筹集到所需资金，并利用筹集的资金为企业创造 最大价值的能力，这就是股权融资效率，股权融资是公司资 本结构决策的重要内容，融资效率的高低对于战略性新兴产 业上市公司的生存和发展具有重要的意义。 2. 国内外文献综述 2.1. 效率研究 关 于 效 率 的 定 义 ， 国 外 学 者 从 不 同 方 面 提 出 自 己 的 意 见，p.r.gregory和r.c.stuart (1985) [1]认为效率是一个经济系 统在特定时点 (静态效率) 或在特定时期内 (动态效率) 对可获 得的资源 (包括知识) 的有效利用。e.f.fama (1970) [2]提出有 效市场假说，开启了国外关于金融市场效率的研究，此后国 外其他学者做了大量关于融资效率理论和实证的研究。认为 资本市场的效率在于能否完全反映各种与市场有关的信息， 并根据反映信息的程度定义了三种不同水平的市场效率： 弱式有效性、半强式有效性、强式有效性。e.m.tobin (1978) [3]指出用tobinq值来衡量上市公司配置效率。a.d.kunt和 r.levine (1996) [4]从功能的角度实证研究了股票市场的效 率，研究认为股票市场的作用不是表现在股票融资数量，而 是集中体现为提高资本配置效率。r.d.f.harris (1997) [5]对比 发达国家和欠发达国家股票市场，发现欠发达国家的股票市 场的效率很弱，发达国家的股票市场效率明显高于欠发达国 家。 2.2. 股权融资效率研究 刘力昌和冯根福 (2004) [6]以股权融资净总额、股权集中度、 股票非流通性、资产负债率为输入指标，净资产收益率、主 营业务收入增长率、托宾q值为输出指标，对47家首次发行 股票的上市公司股权融资效率进行测度。陶云燕 (2006) [7]以 平均每股收益、平均净资产收益率、平均主营业务收入增长 率、平均净利润增长率为考核指标，运用dea方法研究了我 国1999年至后面的3年中深沪a股上市公司的融资效率情况。 蔡玲君 (2011) [8]以股权融资净总额、股权集中度、股票非 流通度、资产负债率为输入指标，主营业务收入增长率、净 资产收益率、托宾q值为输出指标，运用dea方法对40家上 市高科技公司2006-2009年的股权融资效率进行测度。王重润 等 (2016) [9]以资产总额、资产负债率、融资成本率为输入指 标，净资产收益率、总资产周转率、主营业务收入增长率、 每股收益率为输出指标，通过 dea-malmquist 模型，选择 69 家在 2012 年之前挂牌新三板的中小公司为样本，分析了 2013-2015 年样本公司融资效率变化趋势。杨国佐等人 (2017) [10]以资产总额、主营业务成本、财务费用、资产负债率作 为输入指标，总资产周转率、净资产收益率、主营业务收入 增长率为输出指标，运用dea方法，对597家新三板挂牌公司 2012-2014年的股权融资效率进行测度。 综上所述，从已有文献来看，已有学者们运用dea模型，通 过财务性指标和非财务性指标对公司进行股权融资效率测 度，但大部分研究停留在分析单个行业的融资效率，并且近 年来研究战略性新兴产业上市公司股权融资效率的还较少， 用dea-malmquist 指数法来分析它的就更少了，所以本文选 取2014-2018年具有代表性的208家战略性新兴产业上市公司 的数据，使得研究结果更具有时效性，通过实收资本、股权 集中度、资产负债率3个投入指标、净资产收益率、主营业务 收益率、托宾q值3个产出指标，计算得出各决策单元的2014 年-2018年的dea评价值，并进行效率分析，但是dea-bcc 模型在研究股权融资效率评价方面缺乏动态性，所以同时采 用了dea-malmquist 指数法，从而能够更加客观地评价战略 性新兴产业上市公司的股权融资效率。最后根据实证结果有 针对性的提出如何提高我国战略性新兴产业上市公司股权融 资效率的建议。本文结构安排如下： 首先介绍论文的研究背景、意义以及目的，然后对研究股权 融资效率的国内外相关文献进行了梳理，并作简要的评述， 阐述本文的研究内容、结构，再对战略性新兴产业上市公 司股权融资效率进行测度与分析，此部分首先对dea的三个 模型进行了简要的介绍，然后构建股权融资效率评价指标体 系，再通过前文介绍的dea模型对该产业上市公司的股权融 资效率进行实证研究，最后基于dea-bcc模型的计算结果分 析、基于 malmquist 模型的计算结果分析和产业比较分析， 对上市公司的股权融资效率进行评价。论文最后先总结论文 的研究结论，其次基于前文的实证研究对提高战略性新兴产 业上市公司的股权融资效率提出建议。 3. dea模型 3.1. ccr模型 ccr模型是第一个dea模型，也是应用最广泛的dea模型 [11]。假设有n个生产决策单元dmuj (j = 1, 2, ..., n)，每 个dmu都有m项输入xj = (x1j, x2j, ..., xmj) t，s项输出，yj = (y1j, y2j, ..., ysj) t 则第j0个dmu的效率评估模型为： max . . , h u y v x s t h u y v x u v j t t j t j t j 0 0 0 1 0 0 = = ≤ ≥ ≥         (1) 其中，hj为第j个dmu的相对效率值，v = (v1, v2, ..., vm) t 和 u = (u1, u2, ..., us) t分别为投入和产出的权重向量。 式 (1) 是分式规划形式，若对其进行charnes-cooper变换，可 以得到线性规划模型： s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 103 max . . , , m w m w w m t t j t j t y s t x y x 0 0 0 1 0 0 − ≥ = ≥ ≥      (2) 利用线性规划的对偶理论，并引入松弛变量s+和s_，可以得到 如下的对偶规划模型： min . . q e l q l − +( )  + = − = − + − = + = ∑ e s e s s t x s x y s y t t j j j n j j j n � 0 1 0 1 ∑∑ ≥ ≥ ≥            + −lj s s0 0 0, , (3) 其中， (x0, y0) 为dmuj0的输入输出，e为非阿基米德无穷 小，ê和e分别为元素为1的m维向量和s维向量。 当式 (3) 的最优解为θ*、λ*、s−*和s+*时，关于dmuj0有如下结 论： (1) 若θ* = 1，则称dmuj0为弱dea有效； (2) 若θ* = 1， 且s* = s+* = 0，则称dmuj0为dea有效； (3) 若θ* < 1， 则 称 dm uj 0为 de a无 效 。 c 2r模 型 假 设 规 模 报 酬 不 变 ， 若dmuj0为dea有效，则技术和规模同时有效。 3.2. bcc模型 规模报酬不变 (crs) 的假设与实际差距较大。为解决这一问 题，1984年banker、charnes和cooper为生产可能集合建立 凸性性质、无效率性质、射线无限制性质和最小外插性质 等四项公理，并引进了shepherd距离函数的概念，将技术效 率 (technology efficiency，te) 分解为纯技术效率 (pure technology efficiency，pte) 和规模效率 (scale efficiency，se)， 即：te=pte×se。通过增加对权重λ的约束条件：iλ = 1，建 立如下的规模报酬可变模型 (亦称bcc模型或bc2模型) [12]： min . . , , , , q l q l l l s t x x x y i j t j j j t j j j t j ≤ ≥ = ≥ =        = = ∑ ∑ 0 1 0 1 1 0 1 2 …     (4) 其中，i = (1, 1, ..., 1) 1×t。 目标函数求得的即是纯技术效率 (pte) ，根据se=te/pte可以 求出规模效率 (se) 。纯技术效率 (pte) 是测度当规模报酬可变 时，决策单元与生产前沿面的距离；规模效率 (se) 是测度当规 模报酬可变时的生产前沿面与规模报酬不变时的生产前沿面的 距离。 3.3. dea-malmquist模型 dea-malmquist方法是将dea方法与malmquist 模型相结 合，用以测度全要素生产率。具体来讲首先依据dea方法 确定生产前沿面，然后得到距离函数，利用距离函数构造 tfp指数，以此来衡量全要素生产率的增长率。对于距离 函数而言，可以用投入与产出这两个角度加以定义，基于 投入的距离函数是在给定产出条件下投入向量向前沿面压 缩的程度；基于产出的距离函数是在给定投入时产出的最 大扩张程度；由于距离函数分产出导向 (output orientated) 与投入导向 (input orientated) ，因此可以按投入、产出分 别构造malmquist tfp指数。规模报酬可变情况下，产出 malmquist tfp指数的变化可能是由技术效率变动、生产 技术变动的结果及规模效率变动的结果，因此可以把产出 malmquist tfp指数分解，分别对三部分进行测算[13]。表 达式为： m x y x y d x y d x y d x y d t t t t t t t t t t t t t + + + + + + + ( ) = ( )( ) × ( )1 1 1 1 1 1 1, , , , , , tt t tx y+ ( )        1 1 2 , (5) effch d x y d x y t t t t t t = ( ) ( ) + +1 1, , (6) tech d x y d x y d x y d x y t t t t t t t t t t t t = ( ) ( ) × ( ) ( )     + + + + + + 1 1 1 1 1 1 , , , ,     1 2 (7) tfpch effch tech pech tech= × = ×( ) ×sech (8) 其中， (xt, yt) 和 (xt+1, yt+1) 分别表示t时期和t+1时期的投入产出 向量，若m指数>1，表明效率提高；若m指数<1，表明效率 降低[13]。 4. 指标体系与样本数据 4.1. 股权融资效率测度指标体系 本文选用 dea进行实证研究，而 dea模型的构建和评价 首先就是要选好输入指标和输出指标，在建立输入输出指 标体系时应注意三点：一要准确客观地反映此次的评价 目标；二是要注意投入产出指标间的关系，因为投入指标 和产出指标他们之间可能有很多线性关系；三是要确保投 入产出指标数据的可获得性以及多样性。本文选取实收资 本、股权集中度和资产负债率作为输入指标，选取净资产 收益率、主营业务收入增长率和托宾q值作为输出指标， 因为战略性新兴产业上市公司进行股权融资是通过花费一 定成本来筹得资金，公司再通过合理运用这笔资金获得利 润，指标公式见表1。 (1) 输入指标 1) 实收资本 实收资本是股东将其出资财产实际缴付到公司后公 司实际支配的资本，也就是企业创立或者增加资金 时实际收到的现金或者等价物，它作为公司的财产 可用于生产经营项目，实收资本构成比例是企业向 投资者分配利润或股份分配的依据。 2) 股权集中度 股权集中度用最大股东持股比例来表示，他是衡量公司 股权分散化的数量化指标，也能反映公司的稳定性和公 司结构。本文主要用第一大股东持股比例来衡量股权集 中度情况。 104 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 3) 资产负债率 资产负债率又称举债经营比例，是企业总负债与总资产 的比值，一方面可以反映企业借得资金后资金的运用能 力，一方面可以反映贷款人发放贷款的资金是否安全。 (2) 输出指标 4) 净资产收益率 净资产收益率又称股东权益收益率，是净利润与所有者 权益的比例，反映企业净资产可以创造多少的净利润的 能力，净资产收益率越高，可以表明公司的盈利能力越 高，本文使用该指标主要是该指标能体现资金的使用效 率。 5) 主营业务收入增长率 主营业务收入增长是公司从主要业务活动中取得的营业 收入，它反映公司的盈利状况和变化趋势也可以反映公 司发展状况，主营业务收入增长率越高，表明企业发展 前景越好，通常与股价成正相关关系。 6) 托宾 q值 托宾q值是市场价值与重置价值之间的比率，q=1是标准 值，q值越高，即超过1说明企业的投资回报率越高，企 业此时会增加投资，反之企业会选择持有股票。 4.2. 样本数据 本文所有数据是来自于国泰安csmar系列研究数据库、锐思 金融研究数据库以及同花顺数据库，从三个数据库中选取了 208家具有一定代表性、可比较性的公司作为研究对象。由于 dea模型方法，决策单元dmu的数量应至少大于投入产出指 标总数的2倍以上才具有实际意义。而本文的决策单元为208 个，投入产出指标总数为6个，满足模型的基本要求。因此 本文并从国泰安csmar系列研究数据库、锐思金融研究数据 库以及同花顺数据库中选取并整理了2014-2018年度各决策单 元 (即样本公司) 的投入产出指标数据。具体的原始数据详见 附录a，本文采用的软件是deap2.1，由于在使用软件时，所 要求的原始数据不能存在负值，也不能存在0值，但是输出 指标中净资产收益率和主营业务收入增长率等指标里面存在 负值，并且各指标间量纲可能存在差异，因此为避免数据量 纲差异导致测算误差并保证模型测算的准确性，所以在运用 deap2.1软件时，先将原始数据进行标准化处理，去除量纲差 异并将数据转化为[0, 1]区间的数值，具体操作方法如下： x x x x xij ij ij ij ij ∗ = + − − ×0 1 0 9. min( ) max( ) min( ) . (9) 1) 样本公司的描述性统计 为使评价结果准确反映我国战略性新兴产业股权融资效率的 整体状况，在本文样本公司的选取中，全面考虑了战略性新 兴产业下的七大子产业，其中节能环保产业共选取29家，占 14%；新一代信息技术产业共选取27家，占13%；生物医药产 业共选取34家，占16%；高端装备制造产业共选取29家，占 14%；新能源产业共选取31家，占15%；新材料产业共选取33 家，占16%；新能源汽车产业共选取25家，占12%。样本公司 产业分布情况如下图1所示。 2) 样本指标的描述性统计 在进行实证分析之前，先对样本指标进行简单的描述性统计 分析，样本指标包括三个输入指标，三个输出指标，分析结 果见表2。 从表2可以看出，战略性新兴产业上市公司股权融资效率的 输入指标中实收资本资金总额均值在2014年达到最高为134.65 千万元，最低在2015年，有91.643千万元，五年均值为118.494 千万元，表明企业创立或者增加资金时实际收到的现金或 者等价物规模比较大，但由于平均标准差达到了136.728， 可以看出不同的战略性新兴产业上市公司在不同时期内收 到的资金总额的差异很大。股权集中度在这五年的均值相差 不大，五年均值为34.28%，平均标准差13.74%，可以说明战 略性新兴产业上市公司的股权集中度较高，公司股权分散化 较低，大股东拥有较高的决策治理权，也能反映公司的稳定 性和公司结构较好。资产负债率最高在2014年，有51.746%， 五年均值为44.213%，平均标准差18.705%，表明大多数战略性 新兴企业上市公司较少的通过负债来获得资金，贷款人发放 贷款的资金较安全。战略性新兴产业上市公司股权融资效率 的输出指标中净资产收益率五年均值为8.855%，平均标准差 24.344%，主营业务收入增长率五年均值为0.441%，平均标准 差为2.116%，说明整体上看战略性新兴产业上市公司有一定 的发展潜能，盈利能力、成长能力都相对稳定，托宾q值的 图1 | 样本公司产业分布情况. 表1 | 股权融资效率评价指标体系 指标 公式 输入指标 实收资本 实收资本≈股权融资 股权集中度 第一大股东持股比例=第一大股东持有股份数÷总发行股份数×100% 资产负债率 资产负债率=总负债÷总资产×100% 输出指标 净资产收益率 净资产收益率=税后利润÷净资产×100% 主营业务收入增长率 主营业务收入增长率= (本期主营业务收入-上期主营业务收入) ÷上期主营业务收入×100% 托宾q值 托宾q值=市场价值÷资产重置成本 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 105 五年均值为288.14%，平均标准差为2082.84%，公司的市值明 显高于重置成本，表明公司可以通过发行公司的股票来获得 更多的收益，但标准差较高，说明样本上市公司在各个时间 阶段的差异较大。 5. 实证分析 5.1. 基于dea-bcc模型的计算结果分析 根据deap2.1软件下的dea值，参照类似参考文献，将dea 值假定为a, a分为4个部分，取值范围都是在0到1之间，其中 0<a<0.5是属于低效率的，0.5≤a<0.8是属于较低效率的，低 效率和较低效率都表明公司股权融资效率不高，都是无效率 的，0.8≤a<1表明公司的股权融资效率有较高的水平，a=1 时，这时候公司达到了最优股权融资效率[14]，通过运用 deap2.1软件对样本公司经过无量纲化的数据进行处理后，五 年数据统一计算，测算出208家公司2014年至2018年的综合效 率值、纯技术效率值、规模效率值和规模报酬，具体数据见 附录b。 1) 从综合效率进行分析 综合效率是对决策单元的资源配置能力、资源使用效率 等多方面能力的综合衡量与评价，一般认为，综合效率= 纯技术效率×规模效率，综合效率等于1时，表明该决策 单元的投入产出是有效的。从表3来看，在2014年有9家 公司达到了综合效率1，占比4.33%，分别是欢瑞世纪、京 威股份、美尚生态、长园集团、航发动力、中海油服、 中材节能、伟明环保、老百姓，表明这9家公司在2014年 达到了技术有效和规模有效，2015年减少了5家公司，只 有4家综合效率为1，只占比1.92%，分别为欢瑞世纪、长 园集团、广誉远、航发动力，2016年只有欢瑞世纪这一 家综合效率为1，2017年和2018年208家公司均未达到综 合效率为1。所有公司2014年综合效率平均值为逐年递减 状态，并且208家公司综合效率差异大。从图2蓝色线可 以看出综合效率等于1在五年间是逐年递减的，红色线也 是一条下降的曲线，在2014年达到最高，有105家公司处 于综合效率较高阶段，占比50.48%，2018年一样，位于最 低点，有39家公司，占比18.75%。灰色线是一条上升的曲 线，2014年达到最低点，有77家公司处于综合效率较低 阶段，占比37.02%，2018年达到最高点，有123家公司， 占比59.13%，黄色线逐年上升但占比较低。可以看出5年 间，占大部分比例的综合效率比较高的公司向综合效率 低的公司转移，总体来看公司的股权融资效率是逐步在 下降。 2) 从纯技术效率进行分析 纯技术效率是公司由于管理和技术等因素影响的生产效 率，当纯技术效率到达1的时，表明在目前的技术水平 上，公司投入资源，资源得到了有效利用，由表4可知， 在纯技术效率等于1方面，最多是在2014年，有11家上市 公司纯技术效率达到了1，占5.29%，分别为欢瑞世纪、 京威股份、维尔利、美尚生态、文一科技、长园集团、 航发动力、中海油服、中材节能、伟明环保、老百姓， 表示在目前的技术水平上，上述决策单元投入资源得到 了有效利用，2015年减少了5家，只有欢瑞世纪、美尚生 态、长园集团、广誉远、航发动力、老百姓达到了纯技 术效率为1，占比2.88%，2016年又减少了3家，只有欢瑞 世纪、长园集团、老百姓达到了纯技术效率为1，2017年 和2018年没有一家上市公司纯技术效率达到了1，并且5 年总体来看占比很小，纯技术效率的均值分布在较高效 率区，但是纯技术效率为较优状态的公司基本在减少， 较低效率水平的公司在逐年递增，这表明较高效率水平 的公司在向较低效率水平公司转移，说明公司的管理和 技术有一定的下降，所有公司2014年纯技术效率平均值 为0.8454，2018年为0.7607，可以看到所有公司每年纯技 术效率平均值是逐年递减的，每年208家公司超过当年纯 技术效率平均值的公司人数是逐年递增的，208家公司纯 技术效率分布差异大，所以公司应该加强公司内部的管 理和加大对先进的科学技术的投入，不断加强稳固并完 善公司内部制度，进一步增加公司达到纯技术效率为1的 公司数量，从而使得公司的股权融资效率得到提高。 3) 从规模效率进行分析 规模效率是指产业结构通过优化配置对产出单元所发生 作用的大小，是由于公司规模因素影响的生产效率， 从表5可以看出，2014年规模技术效率达到1的有11家， 占比5.29%，有欢瑞世纪、京威股份、美尚生态、中天科 技、长园集团、航发动力、中海油服、中材节能、伟明 环保、福斯特、老百姓，说明这11家公司能够充分将自 身资源投入到公司中去，实现公司的最优配置，有171家 公司到达了较高效率水平，占比82.21%，处于低效率和较 低效率的公司达到了12.5%，2015年，规模效率达到了1的 公司减少至4家，有欢瑞世纪、长园集团、广誉远、航发 动力，达到较高效率水平的公司也减少了6家，为79.33% ，而处于低效率和较低效率的公司增加到了18.75%，2016 年只有欢瑞世纪综合效率达到了1，2017年和2018年均 表2 | 输入、输出指标的描述性统计量 变量 (单位) 时间 样本数 最小值 最大值 平均值 标准差 实收资 本 ( 千万) 2014年 208 3.480 1102.557 134.650 115.593 2015年 208 4.483 1102.557 91.643 133.503 2016年 208 5.000 1102.557 112.218 139.559 2017年 208 6.135 1168.346 121.575 143.670 2018年 208 10.252 1168.346 132.401 151.314 股权集 中度 (%) 2014年 208 5.300 74.300 41.300 13.800 2015年 208 5.300 90.000 33.600 13.800 2016年 208 5.300 94.300 32.800 14.300 2017年 208 7.800 77.300 31.900 13.300 2018年 208 7.800 77.000 31.800 13.500 资产负 债率 (%) 2014年 208 3.545 95.260 51.746 18.935 2015年 208 5.170 88.201 40.229 18.079 2016年 208 5.859 89.653 40.904 18.998 2017年 208 6.144 90.242 43.561 18.559 2018年 208 7.056 91.295 44.624 18.954 净资 产收 益率 (%) 2014年 208 –123.895 831.668 17.750 58.799 2015年 208 –99.025 33.683 6.812 10.377 2016年 208 –70.221 38.816 7.697 8.725 2017年 208 –15.182 36.880 9.013 6.749 2018年 208 –485.967 29.570 3.003 37.068 主营业 务收 入增 长率 (%) 2014年 208 –1.000 3.936 0.100 0.501 2015年 208 –1.672 33.076 0.701 3.006 2016年 208 –0.469 53.969 0.560 3.747 2017年 208 –0.303 23.998 0.619 2.170 2018年 208 –0.592 15.583 0.223 1.154 托宾q值 (%) 2014年 208 93.500 71594.500 148.700 4950.700 2015年 208 102.100 72962.900 673.800 5041.800 2016年 208 95.500 1982.400 248.300 180.000 2017年 208 95.400 1389.400 215.200 148.400 2018年 208 80.200 875.800 154.700 93.300 106 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 没有达到，处于较高效率的公司基本是在逐年递减，处 于低效率和较低效率的公司却是在逐年递增。所有公司 2014年规模效率平均值为0.9061，2018年为0.829，可以 看到所有公司每年规模效率平均值是逐年递减的，并且 208家公司纯技术效率分布差异大。两个表说明公司产量 的增加比例与公司投入各生产要素增加的比例不相等， 没有达到规模有效状态，公司的融资效率处于较低的水 平。 4) 从规模报酬进行分析 从表6样本上市公司规模报酬分布情况可以看出，2014 年规模报酬不变的有11家公司，占比5.19%，这是五年最 高，说明这11家公司在2014年里公司产量的增加比例与 公司投入各生产要素增加的比例相等，实现了规模有效 状态，有欢瑞世纪、京威股份、美尚生态、中天科技、 长园集团、航发动力、中海油服、中材节能、中材节 能、福斯特、老百姓，2017年和2018年没有一家公司达到 了规模报酬不变，从5年总体看，达到规模报酬不变约每 年3.2家，占比1.538%，只有很少部分公司实现了规模有效 状态，大部分公司处于规模报酬递减阶段，平稳在97.98% ，这些公司应该采取一定手段来减少生产要素的投入， 如果盲目的扩张，可能会使得公司的生产投入不能与当 前的生产水平相匹配，从而导致公司产出减少，公司规 模效益下降，融资效率低下，从2014年到2018年的规模报 表3 | 样本上市公司综合效率分布情况 效率区间 2014年 2015年 2016年 2017年 2018年 公司数 占比 公司数 占比 公司数 占比 公司数 占比 公司数 占比 综合效率 1 9 4.33% 4 1.92% 1 0.48% 0 0.00% 0 0.00% 0.8<=a<1 105 50.48% 84 40.38% 64 30.77% 55 26.44% 39 18.75% 0.5<=a<0.8 77 37.02% 98 47.12% 108 51.92% 117 56.25% 123 59.13% 0<a<0.5 17 8.17% 22 10.58% 35 16.83% 36 17.31% 46 22.12% 所有公司年平均值 0.7753 0.7320 0.6873 0.6656 0.6397 与公司五年平均值 比较 >=0.7753 90 >=0.732 94 >=0.6873 117 >=0.6656 124 >=0.6397 138 <0.7753 118 <0.7320 114 <0.6873 91 <0.6656 84 <0.6397 70 表4 | 样本上市公司纯技术效率分布情况 效率区间 2014年 2015年 2016年 2017年 2018年 公司数 占比 公司数 占比 公司数 占比 公司数 占比 公司数 占比 纯技术效 率 1 11 5.29% 6 2.88% 3 1.44% 0 0.00% 0 0.00% 0.8<=a<1 136 65.38% 119 57.21% 101 48.56% 97 46.63% 90 43.27% 0.5<=a<0.8 61 29.33% 83 39.90% 103 49.52% 110 52.88% 116 55.77% 0<a<0.5 0 0.00% 0 0.00% 1 0.48% 1 0.48% 2 0.96% 所有公司年平均值 0.8454 0.8188 0.7946 0.7779 0.7607 与公司五年平均值 比较 >=0.8454 84 >=0.8188 97 >=0.7946 113 >=0.7779 123 >=0.7607 133 <0.8454 124 <0.8188 111 <0.7946 95 <0.7779 85 <0.7607 75 图2 | 战略性新兴产业样本上市公司综合效率情况图. s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 107 酬递增的公司来看，比例总体很小，这部分公司需要资 金来扩大生产规模以提高产出。 综合前面对战略性新兴产业样本上市公司进行综合效率、纯 技术效率、规模效率以及规模报酬的分析，可以看出5年来， 战略性新兴产业样本上市公司的股权融资效率处于比较低的 水平，并且股权融资效率是逐年递减的状态，所以公司应该 加强公司内部的管理、调整生产规模和加大对先进的科学技 术的投入，不断加强稳固并完善公司内部制度。 5.2. 基于 malmquist 模型的计算结果及分析 利用deap version 2.1软件，运用dea-malmquis模型，5年统 一计算，计算得出 2014－2018 年年度平均malmquist 指数变 化及分解以及208家战略性新兴产业上市公司平均malmquist 指数变化及分解，结果见表7、表8、图3和附录c。 1) 战略新兴产业股权融资效率总体趋势分析 malmquist指数能动态反映208家战略性新兴产业上市公司 股权融资效率变化趋势，并且全要素生产率变动=技术效 率变动×技术变动，技术效率变动=纯技术效率变动×规 模效率变动，效率的整体变化趋势从表6和图3来看，总 体呈下降趋势，在研究期间每年的全要素生产率变动都 小于1，说明208家战略性新兴产业上市公司股权融资效 率处于下降阶段，全要素生产率变动是先下降再上升在 下降，呈波动状态，说明对于战略性新兴产业的改进还 可以从技术、规模和资源等方面寻找解决途径。分解来 看，2014年-2018年全要素生产率变动、技术变动、技术 效率变动的指数平均值分别为 0.949、0.971和0.977，所 以在技术变动作用年均下降 2.9% 和技术效率变动作用年 均下降 2.3% 的综合作用下，全要素生产率变动年均下降 5.1%，即我国战略性新兴产业发展的总体股权融资效率 年均下降4.1%，从图3中可以看出，技术效率变动大体呈 上升趋势，技术变动大体呈下降趋势，相比之下，全要 素生产率变动受技术变动作用的影响会更大一点。分年 度来看，2014年-2015年全要素生产率变动为0.949，2015 年-2016年全要素生产率变动为0.927，说明在原有的基础 之上这种情况没有得到改善，2016年-2017年增加到0.963， 但在2017年-2018年又减少到0.957，始终小于1。说明这 几年的融资效率整体没有太大的改善，技术变动、技术 效率变动几年间都是小于1的，说明了公司总体效率低下 是由技术效率变动和技术变动共同引起的，所以应该在 金融技术创新方面加大力度，公司对于资源的合理配置 也是公司需要不断提升的。 由 于 技 术 效 率 变 动 =纯 技 术 效 率 变 动 × 规 模 效 率 变 动，2014年-2018年技术效率变动、纯技术效率变动、规 模效率变动平均值分别为 0.977、1.029和 0.95，纯技术效 率变动作用年均上升2.9% ，规模效率变动作用年均下降 5% ，技术效率变动年均下降 2.3%，在均值上纯技术效率 变动是大于1的，而规模效率变动是小于1的，图4可以看 出纯技术效率变动虽然是呈下降趋势的，规模效率变动 呈上升趋势，但纯技术效率变动每年都达到了1，表明技 术效率变动的下降是由规模效率变动的下降引起的，说 表5 | 样本上市公司规模效率分布情况 效率区间 2014年 2015年 2016年 2017年 2018年 公司数 占比 公司数 占比 公司数 占比 公司数 占比 公司数 占比 规模效率 1 11 5.29% 4 1.92% 1 0.48% 0 0.00% 0 0.00% 0.8<=a<1 171 82.21% 165 79.33% 150 72.12% 153 73.56% 143 68.75% 0.5<=a<0.8 22 10.58% 31 14.90% 49 23.56% 47 22.60% 55 26.44% 0<a<0.5 4 1.92% 8 3.85% 8 3.85% 8 3.85% 10 4.81% 所有公司年平均值 0.9061 0.8817 0.8506 0.8426 0.829 与公司五年平均值 比较 >=0.9061 97 >=0.8817 125 >=0.8506 146 >=0.8426 149 >=0.829 152 <0.9061 111 <0.8817 83 <0.8506 62 <0.8426 59 <0.829 56 表6 | 样本上市公司规模报酬分布情况 变化 2014年 2015年 2016年 2017年 2018年 规模报酬 递增 1 0.48% 2 0.96% 2 0.96% 0 0.00% 0 0.00% 不变 11 5.29% 4 1.92% 1 0.48% 0 0.00% 0 0.00% 递减 196 94.23% 202 97.12% 205 98.56% 208 100.00% 208 100.00% 表7 | 2014–2018 年年度平均malmquist 指数变化及分解 年份 技术效率变动 技术变动 纯技术效率变动 规模效率变动 全要素生产率变动 2014–2015 0.951 0.997 1.109 0.858 0.949 2015–2016 0.975 0.951 1.006 0.968 0.927 2016–2017 0.998 0.965 1 0.998 0.963 2017–2018 0.985 0.971 1.003 0.983 0.957 均值 0.977 0.971 1.029 0.95 0.949 108 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 明了全要素生产率变动的下降与公司规模很大关系，表 明技术进步的提高主要从公司规模着手。 2) 208家战略性新兴产业上市公司效率变化分析 将malmquist输出结果假定为b，从表8和图4可以看 出，208家战略性新兴产业上市公司中有20家年均全要 素生产率变动超过了1，占比9.62%，在0.9-1区间有160 家，占比76.92%，占208家战略性新兴产业上市公司的绝 大多数，技术效率变动中有81家超过了1，占比38.94%， 在0.9-1区间有112家，占比53.85%，技术变动在0.9-1区 间有199家，占比95.67%，而纯技术效率变动超过1的公 司占大部分，有190家上市公司，规模效率变动大多分布 在0.9-1区间，有147家上市公司，可以看出208家企业有 127家企业是技术效率变动下降的趋势，有4家企业是技 术变动下降的趋势，只有18家企业是纯技术效率变动下 降的趋势，有178家企业是规模效率变动下降的趋势，有 188家企业处于全要素生产率变动下降趋势。 通过附录c具体进行分析，从技术变动来看，只有东旭 蓝天技术变动达到了1，华控赛格、美锦能源、协鑫集 成三家公司超过了1，其余204家公司的技术变动都是小 于1的，平均减少2.9%，表明了七大战略性新兴产业的技 术变动总体是趋于下降的，技术创新还不够，整体股权 融资效率低下。从全要素生产增长率变动来看，有启迪 古汉、飞马国际、聚光科技3家公司全要素生产增长率变 动达到了1，华控赛格、东方市场、许继电气等17家公司 超过了1，其余188家公司的全要素增长率变动都是小于 1的， 全要素生产增长率变动平均值为0.949，从整体上 图4 | 208家战略性新兴产业上市公司效率变化分析. 图3 | 2014–2018 年年度平均malmquist 指数变化及分解图. 表8 | 208家战略性新兴产业上市公司效率变化分析 效率区间 分布 年均技术效率 年均技术 年均纯技术效率 年均规模效率 年均全要素生产率 变动 变动 变动 变动 变动 企业个数 比例 企业个数 比例 企业个数 比例 企业个数 比例 企业个数 比例 0.6<=h<0.8 2 0.96% 1 0.48% 0 0.00% 2 0.96% 2 0.96% 0.8<=a<0.9 13 6.25% 4 1.92% 0 0.00% 29 13.94% 26 12.50% 0.9<=a<1 112 53.85% 199 95.67% 18 8.65% 147 70.67% 160 76.92% a>=1.0 81 38.94% 4 1.92% 190 91.35% 30 14.42% 20 9.62% 总计 208 100% 208 100% 208 100% 208 100% 208 100% s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 109 看208家战略性新兴产业上市公司全要素生产增长率变动 平均减少5.1%，技术效率变动减少2.3%，说明了战略性新 兴产业大多数产业都应该在金融技术工具、产品创新和 公司规模方面进行提高和完善，建立良好的金融发展环 境，构建完备的金融发展体系，让金融资源得到充分利 用。 5.3. 产业比较分析 1) 根据dea-bcc模型五年统一计算得出的结果，整理出战 略性新兴产业七个子产业每年的综合效率有效情况、综 合效率平均值，结果见表9、表10、图5。 战略性新兴产业包括七大子产业，根据原始数据可以得 到这七大产业股权融资效率每年的综合效率平均值的变 化，从表9和表10可以看出，2014年节能环保只有3家公 司达到了综合效率1，占比1.44%，高端装备制造和新能 源汽车只有2家公司达到了综合效率1，占比0.96%，新 一代信息技术和生物医药只有1家公司达到了综合效率 1，占比0.48%，其余2大产业均没有达到。2015年生物 医药、高端装备制造、新能源汽车只有1家达到了综合 效率1，其余4家均没有达到，2015年、2016年、2017 年七大产业没有一家达到综合效率为1，说明七大产业 总体股权融资效率很低。战略性新兴产业上市公司的融 资效率五年均值为 0.6987，距离综合有效值1还有很大 的提升空间，生物医药最高，为0.7558，新能源最低， 为0.6463。从图5可以发现我国战略性新兴产业上市公 司的股权融资效率存在产业间的非均衡发展。其中，从 综合效率平均值来看，生物医药每年都保持最高，新一 代信息技术2014年-2016年保持在第二，新材料在2017 年-2018年保持第二，新能源每年都最低，其余产业排 名每年在变动，并且七大产业整体趋于下降趋势。 2) 七大产业平均malmquist 指数变化及分解 从表11总体来看，2014年-2018年的年均全要素生产率变 动最高的是新能源汽车，为0.961，新能源汽车的技术效 率变动和技术变动分别为0.985和0.980，纯技术效率变动 为1.030，规模效率变动为0.960，全要素生产率变动在技 术变动下降1.5%和技术效率变动下降2%的综合作用下，全 要素生产率变动下降3.9%，技术效率变动在纯技术效率变 动作用下上升3%，在规模效率变动下下降4%，所以技术 效率变动在两者共同作用下下降1.5%，相比之下，全要素 生产率变动受技术变动作用的影响会更大一点，其次是 规模效率变动作用。年均全要素生产率变动最低的是节 能环保，为0.951，节能环保全要素生产率变动下降4.9% ，是在技术变动作用下降2.5%和技术效率变动下降2.5%共 同作用下产生的结果，而技术效率变动下降完全是由于 规模效率变动的下降引起的，因为纯技术效率变动是上 升2.8%，而规模效率变动下降5%，七大产业总体效率的 下降主要是技术效率变动的下降和技术变动的下降，而 技术效率变动的下降是由规模效率变动下降引起的，因 为样本期间内，七大产业的技术效率变动和技术变动都 是小于1的，纯技术效率变动大于1，正是由于三者的共 同下降使得七大产业的股权融资效率平均值在样本期间 内都是处于下降趋势，表明整个样本观测期间七大战略 性新兴产业发展的总体效率是下降的，七大战略性新兴 产业在技术创新和资源配置等方面都需要进行完善和提 高。 表 9 | 七 大 产 业 综 合 效 率 有 效 情 况 年 份 效 率 节 能 环 保 新 一 代 信 息 技 术 生 物 医 药 高 端 装 备 制 造 新 能 源 新 材 料 新 能 源 汽 车 个 数 占 比 个 数 占 比 个 数 占 比 个 数 占 比 个 数 占 比 个 数 占 比 个 数 占 比 20 14 年 有 效 3 1. 44 % 1 0. 48 % 1 0. 48 % 2 0. 96 % 0 0. 00 % 0 0. 00 % 2 0. 96 % 非 有 效 26 12 .5 0% 26 12 .5 0% 33 15 .8 7% 27 12 .9 8% 31 14 .9 0% 33 15 .8 7% 23 11 .0 6% 20 15 年 有 效 0 0. 00 % 0 0. 00 % 1 0. 48 % 1 0. 48 % 0 0. 00 % 0 0. 00 % 1 0. 48 % 非 有 效 29 13 .9 4% 27 12 .9 8% 33 15 .8 7% 28 13 .4 6% 31 14 .9 0% 33 15 .8 7% 24 11 .5 4% 20 16 年 有 效 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 非 有 效 29 13 .9 4% 27 12 .9 8% 34 0. 00 % 29 13 .9 4% 31 14 .9 0% 33 15 .8 7% 25 12 .0 2% 20 17 年 有 效 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 非 有 效 29 13 .9 4% 27 12 .9 8% 34 0. 00 % 29 13 .9 4% 31 14 .9 0% 33 15 .8 7% 25 12 .0 2% 20 18 年 有 效 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 0 0. 00 % 非 有 效 29 13 .9 4% 27 12 .9 8% 34 0. 00 % 29 13 .9 4% 31 14 .9 0% 33 15 .8 7% 25 12 .0 2% 110 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 表10 | 综合效率平均值 产业名称 2014年 2015年 2016年 2017年 2018年 五年平均 节能环保 0.7905 0.7471 0.6683 0.6551 0.6343 0.6991 新一代信息技术 0.7964 0.7575 0.7257 0.6758 0.6454 0.7202 生物医药 0.8231 0.7875 0.7534 0.7202 0.6947 0.7558 高端装备制造 0.7482 0.6916 0.6653 0.6521 0.6292 0.6773 新能源 0.7147 0.6900 0.6277 0.6207 0.5783 0.6463 新材料 0.7846 0.7399 0.6989 0.6806 0.6612 0.7130 新能源汽车 0.7643 0.7002 0.6617 0.6444 0.6252 0.6791 战略性新兴产业整体 0.7746 0.7305 0.6859 0.6641 0.6383 0.6987 表11 | 平均 malmquist 指数变化及分解 时间 2014年-2015年 2015年-2016年 2016年-2017年 2017年-2018年 年度均值 节能环保 全要素生产率变动 0.984 0.895 0.973 0.954 0.951 技术效率变动 0.978 0.937 0.999 0.988 0.975 技术变动 1.001 0.957 0.975 0.966 0.975 纯技术效率变动 1.110 1.000 1.000 1.002 1.028 规模效率变动 0.878 0.937 0.998 0.985 0.950 新一代信息技术 全要素生产率变动 0.949 0.947 0.940 0.974 0.952 技术效率变动 0.961 1.008 0.988 1.003 0.990 技术变动 0.989 0.944 0.952 0.968 0.963 纯技术效率变动 1.102 1.008 0.998 1.001 1.027 规模效率变动 0.877 1.000 0.989 1.002 0.967 生物医药 全要素生产率变动 0.956 0.948 0.956 0.966 0.957 技术效率变动 0.965 0.996 0.980 0.984 0.981 技术变动 0.991 0.953 0.975 0.982 0.975 纯技术效率变动 1.103 1.006 0.999 1.004 1.028 规模效率变动 0.881 0.990 0.981 0.980 0.958 高端装备制造 全要素生产率变动 0.934 0.957 0.986 0.961 0.959 技术效率变动 0.930 1.019 1.021 0.985 0.989 技术变动 1.005 0.943 0.966 0.976 0.972 纯技术效率变动 1.124 1.009 1.002 1.002 1.034 规模效率变动 0.834 1.010 1.018 0.983 0.961 新能源 全要素生产率变动 0.973 0.922 0.989 0.927 0.953 技术效率变动 0.971 0.973 1.023 0.968 0.984 技术变动 1.004 0.951 0.969 0.959 0.971 纯技术效率变动 1.140 1.005 1.002 1.004 1.038 规模效率变动 0.857 0.968 1.021 0.964 0.953 新材料 全要素生产率变动 0.947 0.945 0.972 0.973 0.959 技术效率变动 0.956 0.993 1.007 0.999 0.989 技术变动 0.991 0.952 0.968 0.974 0.971 纯技术效率变动 1.099 1.012 1.000 1.004 1.029 规模效率变动 0.876 0.983 1.007 0.995 0.965 新能源汽车 全要素生产率变动 0.968 0.932 0.968 0.978 0.961 技术效率变动 0.957 0.975 1.012 0.996 0.985 技术变动 1.008 0.968 0.961 0.982 0.980 纯技术效率变动 1.113 1.004 1.002 1.001 1.030 规模效率变动 0.864 0.970 1.010 0.995 0.960 总体 全要素生产率变动 0.959 0.935 0.969 0.962 0.949 技术效率变动 0.960 0.986 1.004 0.989 0.977 技术变动 0.998 0.953 0.967 0.973 0.971 纯技术效率变动 1.113 1.006 1.001 1.003 1.031 规模效率变动 0.867 0.980 1.003 0.986 0.959 图5 | 七大产业综合效率平均值. s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 111 具体分析来看，7大战略性新兴产业股权融资效率下降的环比 区间都有4个，其中节能环保2个环比区间下降的原因是由于 技术效率变动下降引起的，也就是金融资源未能得到充分有 效配置，而这两个环比区间的技术效率变动下降的原因都是 由于规模效率变动下降引起的，企业其余2个环比区间主要是 由于技术水平下降引起的。新一代信息技术、生物医药、高 端装备制造、新能源、新材料和新能源汽车有1个环比区间下 降的原因主要是由于技术效率变动下降引起的，而这六大战 略性新兴产业技术效率变动下降的原因都是由于规模效率变 动下降引起的，纯技术效率变动上升的幅度小于规模效率变 动下降的幅度，其余3个环比区间都是主要是由于技术水平下 降引起的。总体来看，七大战略性新兴产业股权融资效率的 提高既要注重公司技术创新能力的提高，又要在资源配置方 面加大力度，并且在规模效率方面应该引起足够重视，采取 适当的方法增加产出减少投入，当企业的经营规模扩大时， 只有企业的产出增加比例大于投入比例，那么企业的经营规 模才能达到规模效益，而不会规模效率下降。 6. 结论与建议 本文通过dea模型来对战略性新兴产业上市公司股权融资效 率进行测度研究得出如下结论： 1) 根据5.1得出，208家战略性新兴产业上市公司的股权融资 效率普遍很低，综合效率在2014年大多分布在较高效率 区间，公司占比50.48%，其余分布在较低效率区间，并且 为逐年递增，从2015年47.12%到2018年59.13%，纯技术效 率和规模效率每年大多分布在较高效率区间，所以公司 内部技术水平还可以再提高，规模报酬多数表现为规模 报酬递减，说明大多数企业都应该增加自身规模，股权 融资效率是逐年递减的状态，企业应该通过生产规模、 加大对先进的科学技术的投入和不断加强稳固并完善公 司内部制度来促进企业股权融资效率的提高。 2) 根据5.2得出，效率的整体变化是呈下降趋势，5年的全要 素生产率变动都小于1，说明208家战略性新兴产业上市 公司股权融资效率处于下降阶段。全要素生产率变动为 0.949，全要素生产率变动年均下降5.1%，并且全要素生 产率变动受技术变动作用的影响会更大一点，其次是规 模效率变动，七大战略性新兴产业的技术变动总体是趋 于下降的，208家战略性新兴产业大多处于全要素生产率 变动下降趋势，技术创新还不够，整体股权融资效率低 下，公司需要及时更新设备，不断加强产能，大多数产 业都应该在金融技术工具、产品创新和公司规模方面进 行提高和完善，建立良好的金融发展环境，构建完备的 金融发展体系，让金融资源得到充分利用。 3) 根据5.3第一部分得出，七大产业里面每年达到综合效率 有效只有零星几家，融资效率五年均值为0.6987，达到综 合有效值1还有很大的提升空间，其中生物医药最高，新 能源最低，所以七大产业总体股权融资效率效率还是很 低，我国战略性新兴产业上市公司的股权融资效率存在 产业间的非均衡发展，七大产业整体趋于下降趋势，要 整体提高战略性新兴产业股权融资效率，重点突破新能 源产业的股权融资效率。 4) 根据5.3第二部分得出，年均全要素生产率变动最高的是 新能源汽车，为0.961，最低的是节能环保，为0.951，节 能环保全要素生产率变动下降4.9%，七大产业的全要素生 产率变动都是小于1的，技术效率变动和技术变动两者的 共同下降使得七大产业的股权融资效率平均值在样本期 间内都是处于下降趋势，总体来看，七大战略性新兴产 业股权融资效率的提高既要注重公司技术创新能力的提 高，又要在资源配置方面加大力度，公司要不断提高内 部的管理水平以及进行技术创新，企业应该根据产业本 身的特色，并且考虑多方面因素来正确选择融资渠道， 让产能发挥最大，资本得到最优配置。 本文通过对208家战略性新兴产业上市公司2014年到2018年股 权融资效率进行研究分析后，发现战略性新兴产业上市公司 的股权融资效率普遍较低，给出以下建议，首先是由于战略 性新兴产业发展还不够成熟，这些上市公司无论是在内部结 构还是管理水平和管理体系方面都还存在很多不足，所以企 业应该不断完善优化企业的内部机制，从而提高公司获得外 部资金的能力，也能提高自身利用资金使得资金得到最优配 置的能力。其次政府应该大力支持战略性新兴产业的金融技 术创新，企业要进行产品、业务、制作技术、服务等方面进 行创新。最后，七大产业的全要素生产率都是小于1的，技术 进步变化、技术效率变化指数也都是小于1的，所以七大产业 融资效率的提高需要公司不断提高生产管理水平，提高企业 的市场竞争力。 conflicts of interest the authors declare they have no conflicts of interest. acknowledgments to the regional project of national natural science foundation of china (71861003) and the innovative exploration and new academic seedlings project of guizhou university of finance and economics (guizhou-science cooperation platform talents [2018] 5774-016) for their support. supplementary materials supplementary data related to this article can be found at https:// doi.org/10.2991/jracr.k.201027.001. references [1] gregory pr, stuart rc. comparative economic systems. boston: houghton mifflin; 1985. [2] fama ef. efficient capital markets:a review of theory and empirical work. j finan 1970;25:383–417. [3] tobin em. light regulation of specific mrna species in lemna gibba l. g-3. proc natl acad sci 1978;75:4749–53. [4] kunt ad, levine r. stock markets, corporate finance, and economic growth: an overview. the world bank econ rev 1996;2:223–39. [5] harris rdf. stock markets and development: a re-assessment. eur econ rev 1997;41:139–46. [6] 刘力昌, 冯根福, 张道宏, 毛红霞. 基于dea的上市公司股权 融资效率评价[j].系统工程, 2004;22:55–59. liu lc, feng gf, zhang dh. evaluation of equity financing efficiency of listed companies based on dea. syst eng 2004;22:55–9. https://doi.org/10.2991/jracr.k.201027.001 https://doi.org/10.2991/jracr.k.201027.001 https://doi.org/10.2307/2325486 https://doi.org/10.2307/2325486 https://doi.org/10.1073/pnas.75.10.4749 https://doi.org/10.1073/pnas.75.10.4749 https://www.jstor.org/stable/3990061 https://www.jstor.org/stable/3990061 https://www.jstor.org/stable/3990061 https://doi.org/10.1016/s0014-2921(96)00021-9 https://doi.org/10.1016/s0014-2921(96)00021-9 112 s.-s. li and m. zhang / journal of risk analysis and crisis response 10(3) 101–112 [7] 陶云燕. 我国上市公司股权融资效率研究[d]. 湖南:南华 大学, 2006. tao y. research on equity financing efficiency of china’s listed companies [d]. hunan: nanhua university, 2006. [8] 蔡玲君. 我国上市高科技企业的股权融资效率研究：基于 dea模型[d].上海:华东师范大学, 2011. cai l. research on equity financing efficiency of listed high-tech enterprises in china: based on dea model [d]. shanghai: east china normal university, 2011. [9] 王重润,王赞.“新三板”挂牌企业融资效率分析[j].上海 金融, 2016;70–5. wang cr, wang z. analysis of financing efficiency of listed companies on the “new third board” [j]. shanghai finan 2016;70–5. [10] 杨国佐, 张峰, 陈紫怡.新三板挂牌公司融资效率实证分析 [j].财经理论与实践, 2017;38:48–53. yang gz, zhang f, chen z. empirical analysis of financing efficiency of companies listed on the new third board. finan theory pract 2017;38:48–53. [11] charnes a, cooper ww, rhodes e. measuring the efficiency of decision making units. eur j oper res 1978;2:429–44. [12] banker rd, charnes a, cooper ww. some models for estimating technical and scale inefficiencies in data envelopment analysis. manage sci 1984;30:1031–142. [13] 李志珍,韩瑛,伊文英,张强.陕西省全要素生产率变化趋势 及对经济增长的贡献—基于dea-malmquist指数方法[j].西 部金融, 2017;51–7. li zz, han y, yi wy, zhang q. change trend of total factor productivity in shaanxi province and its contribution to economic growth: based on dea malmquist index method [j]. western finan 2017;51–7. [14] 毕超,李逍雯.新三板企业股权融资效率及影响因素研究 [j].财会通讯, 2019;31–4. bi c, li xw. research on equity financing efficiency and influencing factors of new third board enterprises [j]. account commun 2019;31–4. https://doi.org/10.1016/0377-2217(78)90138-8 https://doi.org/10.1016/0377-2217(78)90138-8 https://doi.org/10.1287/mnsc.30.9.1078 https://doi.org/10.1287/mnsc.30.9.1078 https://doi.org/10.1287/mnsc.30.9.1078 microsoft word review and prospect of emergency logistics under uncertainty conditions.doc review and prospect of emergency logistics under uncertainty conditions hanping zhao1,2*, chence niu1,2, tingting zhang1,2, sida cai1,2 1key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing 100875, china 2academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education the peoples' republic of china, beijing normal university, beijing 100875, china received march 21, 2018 accepted april 20, 2018 abstract emergency logistics is the key process of disaster relief activities. abruptness and inaccurate prediction of disaster make emergency logistics operated under the condition of uncertainty; these uncertainties are essentially different from ones in traditional business decision. to seek the core and future of research in emergency logistics optimization, the mainly research of emergency logistics under uncertainty conditions in recent years were reviewed. the characteristics of uncertainty appearing in emergency logistics process were analyzed, and the expression of uncertain factors, demand forecasts and objective function definition in decision-making optimization model were summarized; moreover, this article analyzed the key points of research in aspects of emergency logistics operations, uncertainties origination and emergency risks, and provided useful reference and directions for future research. keywords: emergency logistics, uncertainty, demand forecasts, optimization 不确定条件下的应急物流研究综述与展望 赵晗萍 1,2*, 牛晨策 1,2, 张婷婷 1,2, 蔡思达 1,2 1. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875 2. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875 摘 要：突发事件发生具有突发性和无法准确预测性的特点，这导致应急物流在构建与运作过程中存在不 确定性。这种不确定性与传统商业决策问题中的不确定性有着本质差别。为了确定应急物流研究的核心和 方向，本文对近些年来国内外不确定条件下应急物流优化方面的研究进行了综述: 从灾害的本身特点出 发，阐述了应急物流问题的特征; 从不确定性因素的刻画、需求预测与决策目标定义三个方面总结应急物 流主要研究进展；最后，本文建议，应该将应急物流运作、不确定因素的来源和应急过程风险评估等三个 方向，作为应急物流问题的研究重点。 关键词：应急物流，不确定性，需求预测，优化 应急物流是为了满足紧急情况下受灾人员的应 急需求，从救灾点向突发事件发生地，对应急物 作者简介：赵晗萍（1977-），女，汉，副教授，博士， 研究方向：灾害风险评价、离散系统仿真、应急响应技术 资助项目: 国家自然科学基金(41471424)；中央高校基本 科研业务费专项资金资助 资、信息以及服务的有效流动进行计划、管理与控 制的过程 [1] 。它是应急响应的核心活动，它能否有 效的运作是保证灾后幸存者维持生命和生活的关键 [2] 。应急物流要根据灾情或灾害风险评估结果来确 定救灾物资的需求，然后根据需求信息安排物资的 供应，从各级物资储备库或潜在物资的供应商调运 到受灾地点，发放到物资需求者的手中。 101 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 国内外对应急物流问题越来越关注，一些学者 对应急物流方面研究做了详尽综述。如，kovacs 等 [3] 对灾害救助中的物流运营方面做了详细综述；陈 丽群 [4] 对应急物流中有关物资调运的研究进行分析 总结；李创 [5] 归纳整理了近些年国内外应急物流的 研究文献，并从基础理论、物资调运和运输优化、 系统构建三个方面入手，对国内外应急物流的研究 现状进行了综述分析。高文军等[6]从基本理论、系 统构建、风险、存在问题、存在问题对策及相关模 型六个方面对应急物流的研究进展进行了综述； galindo 和 batta[7]在 2013 年从灾害管理中的运用方 法、参与环节、灾害的类型、研究贡献和研究过程 中的假设这几个方面对灾害管理研究现状进行综述 分析。目前，应急物流决策优化问题的研究是从传 统运筹学模型改进与扩展而来，其主要分类如图 1 所示。 但应急物流问题研究必须要正视其自身的属性 特征，在突发事件的应急响应过程中，面对的往往 是个不可预测极端不确定的环境，应急救助过程中 常常发生各种极端干扰（公路网条件差、内部冲突 和有限的技术资源等）而导致应急供应链无法正常 运行甚至中断，发展中国家的灾害管理面临的这一 问题更加显著。bbc（2011年）的报告，在最近十 年遭受自然灾害影响的27亿人中，有99%的人生活 在发展中国家[8]。在2008年5·12汶川地震发生后，途 经重灾区的g317和g213道路中断，导致无法获取灾 情信息、救援力量无法进入灾区等应急响应活动受 阻。 不确定性是应急管理中必须要重点关注的因 素，解决应急物流问题也要从本质上分析其不确定 特征、刻画不确定因素、解决需求预测问题、合理 定义应急决策目标。因此本文将在分析应急物流不 确定因素特点基础上，总结应急物流研究进展并展 望未来主要研究方向。 1. 应急物流问题特征分析 1.1 不完备的需求信息 自然灾害的发生难以精确预测，不具有周期 性，加之人类行为活动的不可预测性，应急物流的 需求不确定性的复杂程度要高于传统物流问题。灾 害发生后短期灾情速报信息通常是不完备不准确 的。如，2014年云南“8·03”鲁甸地震由于偏远地区村 寨在地震中有大量山体垮塌，部分村寨几乎整村被 掩埋，救援力量到达后才得以进行详细统计。8月3 日当日统计的死亡人数为221人，但到了8月７日截 止到19:00时的死亡人数已经达到了615人，灾情信 息在短时间内发生意料之外的波动。此外，需求信 息的发布者并不是物资的需求人――通常灾害救助 的政府管理者或者慈善团体才是发布需求信息的主 体；需求信息的发布存在不确定性――需求的信息 在各级物资调配部门之间传递过程中，会由于各种 原因导致需求信息的进一步变异或失真。 1.2 多样化的不确定因素 应急物流系统不是一个稳定运行的日常系统， 不仅是需求难以预测，其供应的主体是由很多渠道 组成的，包括政府、非政府组织、企业、个人等。 在应急状态下对物资的供应并不是按照需求来供 给，其中会有诸多因素（如紧迫性、媒体舆论等） 来影响不同受灾区域的供给数量。在物资调运中会 有意外干扰毁坏交通路网和信息通讯导致供应中断 或延迟、甚至在有的极端情况下提前准备好的库存 会在灾害发生后被毁于一旦，这使得应急物资在生 产和运输过程中也会发生不确定变化。因此应急物 流系统中不确定因素种类是多样的。应急物流问题 也会因致灾因子、需求物资类型、应急过程不同而 应急物流问题 物资储备库应急 出救点选址问题 储备物资存 储问题 物资分配问题 物资运输调度问 题 大规模交通疏散问 题 运输模型库存 （报童）模型 p-中心模型 vrp模型 最大流模型 集合覆盖模型 图 1 应急物流问题模型分类 102 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 存在本质差异，比如不可预测地震和可以预测的台 风，前者侧重于灾后响应后者侧重于灾前备灾，前 者对于需求无法预报后者可以进行概率预报。因此 针对不同应急情景和问题要选取恰当不确定因素描 述方式。 1.3 非逐利的决策目标 应急物流不单追求在物资成本、运输成本、库 存成本等直接成本的最小化，而是更关注在应急过 程中如何保证公众安全，如由于物资短缺而产生的 人员伤亡损失，物资公平性分配的满意度评价等。 因此如何选取决策目标的评价指标是一个关键问 题。应急决策过程中会出现各种不确定条件甚至是 极端事件（低概率高损失）干扰，那么如何建立一 个合理目标度量方式，使得决策方案在发生各种不 确定变化情况下依然稳健也是需要关注的问题。 针对考虑不确定性因素与干扰条件下的应急物 流决策优化问题的研究，表 1 将主要文献进行分类 汇总。在第 2 节中，将具体介绍其研究方法。 2. 不确定应急物流问题处理方法 一个考虑不确定性应急优化问题基本表达形式 如下： jjxgts bxf j ,,2,1,0),(.. ),,(maxmin/ l=≤ζ ζ (1) 其中，ζ 为不确定性参数，x 为决策向量，b 为参数 向量； ),,( ζbxf 为目标函数，表达了衡量应急物流 运行效率的标准，如成本、时间、需求被满足程度 等； 0),( ≤ζxg j 为约束条件函数，通常应急过程中 受到人力、财力、时间、物资的限制，还会受到交 通网络、决策时间逻辑等条件约束。对于包含不确 定因素的应急物流决策问题，首先要找到适合于模 型中的不确定性因素的刻画方法，然后确定合适的 决策目标，并用有效的算法求解这个问题。 2.1 不确定因素刻画方法 目前应急物流优化问题中，描述不确定性主要 有以下几种的方法： （1）区间不确定集合 式(1)中不确定因素 ζ 可看作一个对称有界的随 机变量，在对称区间 ),( 00 ζζζζ +− 内取值，即区 间型不确定集合 { }),(| 00 ζζζζζζ +−∈=ξ 其中 0ζ 为区间的均值， ζ 则定义了区间的边界。 应用区间不确定集合估计不确定性比较粗糙， 但可以方便的转化成为线性规划模型，目前有一些 学者[9-11]利用这种方法估计应急需求，并采用鲁棒优 化技术解决该问题。 （2）情景分析法 情景分析法将利用构成所研究的系统问题的关 键事件（因素）及每个事件和事件之间的发展变化 的概率（影响程度），对事件未来的发展状态或结 果进行分析与描述[39]。按照灾害等级或者影响程度 将灾后划分为不同情景 ω, ω=1,2,…,s，并定义情景 ω 发生概率为 p(ω)，且 ∑ = = s p 1 1)( ω ω 。用 ωζ 记在情 景 ω 中不确定性因素向量。 情景分析法是目前应急问题中较为常用的一种 方法。应急问题是需求推动的，因此需求是最常用 的划分情景的依据[17] [15] [16]，而交通是应急物流运 行的承载体也是容易产生不确定的因素[20]。从本质 上而言，应急问题不确定性来源于灾害不确定性， 因此利用灾害风险分析的结果划分情景则更加有依 据[40] [21]。 灾害发展是一个连贯的影响过程并且影响因素 复杂，barbarosoglu[41]根据地震的震级划分地震情景 （es，earthquake scenario）；然后根据每种地震情 景 下 的 影 响 划 分 不 同 的 影 响 情 景 （ is ， impact scenario）。王旭坪等[39]利用具有不确定性的多属性 向量对应急路径每个路段未来可能的情景加以分 析。 （3）概率分布法 如果根据经验或数据估计不确定参数的近似分布， 可以用概率分布描述不确定参数，即令 ζ 服从分布 )(⋅φ 。 如，shen 等[30]假设运输时间和需求点需求量是 服从对数正态分布，从而建立了大规模灾害下的两 阶段车辆路径模型；beamon 和 kotleba[42]假设需求 服从均匀分布，建立了库存控制模型，以确定在长 期应急响应中最优订货量和订货点；he 和 hu 等[23] 则假设需求点的需求为泊松分布，为应急供应链系 统建立了一个多重营救模型。 （4）机会约束法 在应急条件下，要求绝对满足某些约束条件是 不现实的，机会约束规划（chance constrained programming）模型可以解决这样的问题[43, 44]，为允 103 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 许所作决策在一定程度上不满足约束条件，即约束 条件满足概率不小于某一置信水平。即令 ( ) αθζθ ≥=≤ },...2,1,0,pr{ pxg 其中 gθ(x,ζ), θ=1,2,…,p, 为随机约束函数， 是预先设 定的置信水平。 比如，beraldi 等[28, 29]在研究急救医疗服务问题 时利用机会约束条件描述服务车辆以一定概率满足 表 1 主要文献分类汇总 解决问题 不确定性因素 目标函数构成 不确定 性实现 方法 突发事 件 选址 分配 运输 库存 需求 供应 时间 其他 参数 成本 时间 满意 度 其他 求解方法 作者 地震 √ √ √ √ √ lingo 王晶等[9] 洪水 √ √ √ 未满足补偿 lingo 张玲等[10] 台风 √ √ 受困人群暴 露度 cplex yao 等 [11] 区间不 确定集 合 突发 √ √ √ √ 仿射调节鲁棒 优化 aharon 等 [12] 洪水 √ √ √ √ √ saa 算法 chang 等[13] 地震 √ √ √ √ 道路运输 能力 √ gms/osl barbarosoglu[ 14] 突发 √ √ 满足需求数 量 cplex jia 等 [15] 地震 √ √ √ √ 成本 √ √ 均衡规划 bozorgi-amiri[16] 飓风 √ √ √ √ 拉格朗日 l-shaped 方法 rawls 等 [17] 地震 √ √ √ √ √ √ gams/cplex mete 等[18] 突发 √ √ 安全性、 畅通度 √ 蒙特卡罗模拟 和遗传算法的 混合智能方法 王旭坪等[19] 突发 √ √ √ 可靠路径搜索 算法 缪成等 [20] 突发 √ √ gams/cplex salmeron 等[21] 突发 √ √ √ √ 成本、运 输条件 √ 粒子群优化算 法 bozorgiamiri 等[22]] 台风 √ √ √ √ 遗传算法 he 等[23] 突发 √ √ 未满足货物 库存模型 chakravarty [2 4] 地震 √ √ √ √ √ √ 运输成本 √ cplex、拉格 朗日松弛算法 döyen [25] 情景分 析法 地震 √ √ √ √ √ √ gams/cplex mete[26] 突发 √ √ 混合智能算法 孙莉[27] 突发 √ √ √ √ cplex beraldi 等[28] 突发 √ √ √ √ aimms &cplex beraldi 等 [29]机会约 束 突发 √ √ √ √ √ 未满足需求 数量 cplex、禁忌 启发式 shen 等 [30] 突发 √ 运输速度 √ dijkstra 算法和 蚁群遗传算法 yuan 等 [31] 概率分 布 突发 √ √ √ 库存模型 beamon[32] 地震 √ √ √ √ 马尔科夫决策 分析 王炜等 [33] 随机过 程 突发 √ √ 灾害发生 概率 √ 人口数量 马尔科夫过程 分析 economou 等 [34] 突发 √ √ √ √ 基于仿真优化 算法 刘春林 [35] 地震 √ √ √ √ 风险 √ √ 风险 混沌优化算法 zheng 等[36] 突发 √ √ √ 模拟退火算 法 杨勃等 [37] 模糊集 突发 √ √ √ 成本 √ 库存模型 钱佳[38] 104 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 需求量。shen 等[30]研究了大规模突发事件下的车辆 规划问题时用机会约束表示了需求和时间的不确定 性；孙莉[27]对不确定环境下应急物资配送问题，将 车辆在路径上的行走时间和运输风险设为模糊变 量，建立了不确定环境下的机会约束规划模型。 （5）随机过程 随机过程理论常被用于刻画具有时变特点的不 确定性。令 …… nxxxx ,,,, 321 随机变量的一个数 列， nx 的值是在时间 n 的状态。如果 1n +x 对于过去 状态的条件概率分布仅是 nx 的一个函数，则 ( ) )(,,,, 12101 nnnn xxxpxxxxxxp ==…= ++ 由于马尔科夫过程的特殊性质，在一般物流问 题中经常被采用，chakravarthy 和 daniel[45]将需求假 设为马尔科夫过程，补货期假设为负指数分布，并 证明供应过程也可以看作马尔科夫过程。feeney 和 sherbrooke[46]假定需求是任意复合泊松过程。在应 急过程中，王玮等[33]通过马尔科夫决策过程实现应 急 资 源 调 度 方 案 的 动 态 优 化 。 economou 和 fakinos[34]运用马尔可夫决策方法，刻画了灾害变化 下应急资源调配的动态优化过程。 概率分布和随机过程都需要足够的历史数据积 累估计统计分布，多适用于较常规的应急问题，如 消防和医疗急救等。但在极端灾害的应急物流方面 还没有足够的历史数据积累，很多情况下对于不确 定参数单纯的统计方法难以完全适用。 （6）模糊集 在巨灾的应急物流管理过程中，通常是在缺乏 数据经验和信息条件下，面对复杂问题要运转一个 复杂系统。有很多难以定量的指标，它们没有分明 的数量界限。模糊数是一个解决定性指标有效手段 （zadeh，1965）[47]，将模型中的不确定变量设为模 糊变量 ξ ，其对应的模糊集为 ξ ，然后用 )(ξμ g 来 表示模型中该因素对应的隶属函数。模糊隶属度的 形式有很多种，比如三角模糊函数、梯形模糊函 数、区间函数等。 zheng 和 ling[36]把公路运输时间、需求量、救 灾物资的期望到达时间和预计到达时间、购置额外 物资的上限、价格等不确定变量设为三角模糊变 量。刘春林[35]利用拟梯形模糊隶属度关系定义应急 供应时间满意度和数量满意度。杨勃[37]采用模糊数 来描述出救点到受灾点的时间。汪传旭和邓先明 [48]，田军[49]等探讨了受灾点需求为模糊变量时的应 急救援车辆路径与物资运输优化问题。behret 和 kahraman[50]将需求、存货、短缺成本看作一个三角 形隶属函数模糊集并且随时间模糊集的模糊性降 低，预测越来越准确，具有时间动态性。 （7）函数表达法 在应急响应过程中，由于灾后影响，有一些因 素会随着时间或其他连续因素的变化而变化。即用 )t（ξ 表示随时间变化的不确定因素。函数表达法最 常用于描述应急交通中交通流量变化状态，因为交 通流问题已经有了较完备的研究基础。但在其他方 面函数建立需要大量数据和精确的物理过程描述。 如，yuan 等[31]考虑到随着灾害发生后的时间和影响 范围的扩大，运输速度是不确定的。因此引入速度 随时间连续变化的函数来描述速度的不确定性. thm 0 hmhm hme) βα ××= vtv （ 其中， )hm tv （ 为灾害条件下 t 时刻弧段（h,m）上的 速度； 0 hmv 为正常条件下弧段(h,m)上的速度； hmα 和 hmβ 为在弧段(h,m)的参数，根据弧段到灾害中心 的距离推断出来。 fiedrich 等人[51]利用指数函数描述了受伤人数随 时间变化的规律，建立了动态规划模型，解决地震 灾害发生后的应急资源配置问题。 以上几种方法有时并不是单独使用的，可以多 种方法结合起来。如，孙莉[30]对不确定环境下应急 物资配送问题进行研究，根据实际情况将车辆在路 径上的行走时间和运输风险设为模糊变量，建立了 不确定环境下的机会约束规划模型；beraldi 等[29]针 对了在应急系统中的选址问题，用不同的情景来描 述需求的变化，建立了两阶段的机会约束规划模 型；shen 等[30]研究了大规模突发事件下的车辆规划 问题并根据实际情况将问题分为两个阶段－规划阶 段和运营阶段，假设运输时间和需求点的需求量是 服从对数正态分布的变量，建立了两阶段的机会约 束规划模型。 2.2 不确定需求预测 在传统物流中，供应商和零售商对于商品的历 史信息记录较为完备，市场表现较为稳定，时间序 列分析技术作为一项比较灵活的动态分析模型被广 泛应用，但这些模型比较适用于较长时间的需求预 测和随机规律刻画。在应急物流决策中，准确的需 求信息是保障和驱动所有决策正确实施的前提。应 急响应的大部分活动都是在紧迫、随时变化甚至是 通讯不畅通或中断的情况下进行的，需求的预测只 能根据灾前风险评估结果和应急状态下受灾点的实 时信息进行。 105 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 在灾害发生之前，可以根据灾害影响和区域社 会经济特征估算需求，如聂高众等 [52] 提出了基于地 区和季节系数城市灾害事故救援力量需求量的预测 模型。王楠等 [53] 选取受灾人口、直接经济损失、受 灾面积、灾害强度等4个因素，建立了他们与救援物 资需求的回归预测模型。这种预测模型适用于可预 报的灾害或者为了应急储备做出粗略预估，精确度 较低。 而在灾害发生之后，可能无法及时获取需求信 息，即使可以获取，需求信息来源混杂，各级政 府、灾民自身、企业和ngo等等各类组织都会发布 需求信息，针对这种情况，sheu[54]应用信息熵模型 计算每个发布信息的权重，进而用熵权法确定预测 动态救助需求。这个方法在目前信息传播渠道多样 化情况下是值得借鉴的。 对于具有时变特点的不确定需求，贝叶斯更新 是广泛应用于传统物流的方法[55-58]。在应急响应 中，通过灾情上报制度、无人机侦查、遥感等手 段，可以有效获取相关附加信息。因此可以利用贝 叶斯更新方法更加准确预测需求。如 lodree 和 taskin[59, 60]使用贝叶斯分析将不断获得的飓风风速 信息融合到应对飓风的物资储备决策中，建立了一 个修正的报童库存模型。choi 等[55]比较了两类贝叶 斯更新模型，预测季节性产品的均值和方差先验概 率。詹沙磊等 [61] 利用贝叶斯定理，对灾害情景的发 生概率进行更新和修正。 2.3 不确定优化决策目标 应急物流问题决策往往不是以成本或收益为目 标，而是考虑受灾公众的救助情况，如最小化伤亡 人数（fiedrich[51]），最小化物资到达或延迟时间 （yuan 等[31]、zheng 等[36]、mete 等[18]、孙莉[27]、 缪成等[20]、he 等[23]），最大化受灾群众的满意度 （刘春林[35]、bozorgi-amiri[16]、王旭坪等[19]）；最 小化未满足 需求物资（ shen 等 [30] 、 jia 等 [15] 、 chakravarty[24]）。应急物流往往是一个复杂系统， 要兼顾时间、费用、损失等各方面目标，因此很多 学者构建多目标函数[16, 27, 62-65]作为决策目标。此外 应急问题设计到民生，还要考虑公平性[64]和均衡性 [63]。 除了应急物流问题决策目标的内容，如何选择 决策目标的度量也非常关键。不确定优化通常采用 期望值准则，无法计算解析解时可以采取样本均值 逼 近 法 （ saa ： sample average approximation method）[66]。 但基于期望值准则的优化无法体现决策者的偏 好，并且应急救助追求目标不仅是传统决策优化意 义上找到一个最高效的决策方案，更重要的是要兼 顾这个决策方案的稳健性，让其在极端的干扰下依 然可以比较稳定持续的运行。鲁棒优化（mulvey 等 [67]）是一个可行选择，其目的是找到一个近似最优 解，使它对任意的不确定性参数观测值不敏感。目 前鲁棒优化主要采取两种处理方式，针对使用离散 情景描述不确定问题[68-73]，用遗憾值或惩罚函数控 制在不同情景下最优目标值差距。针对连续的区间 不确定集合来描述不确定因素的问题[9-12, 74]。通常将 不确定因素转化为不确定集合的鲁棒对应问题。 突发事件的发生尤其是极端事件很多是高损失 小概率事件，在计算过程中会被忽略。haimes[75]提 出分割多目标风险法（the partitioned multi-objective risk method，pmrm），该方法引入条件期望函 数，假定损害落入在一个特定的超越概率范围内。 这种方法可以在解决小概率事件风险度量问题，也 可以考虑将其纳入应急问题的优化目标中。 3. 主要挑战和研究展望 应急物流问题不仅相较于传统物流问题多了不 确定的因素，几个本质性的理论问题亟待解决： 应急物流运作过程刻画：实际的应急物流过程 是一个涉及政府部门、慈善组织、企业和个人等多 个节点和层级的供应体系。对应急物流的运作过程 而言，应急供应链不是商业利益驱动，而是纯粹的 需求拉动，并且运作的目的不是为了满足商业利益 最大化，而是尽量使需求满足程度增加。应急供应 链节点之间的物资流通不是完全按照契约关系，而 且不同供应方的信息不完全透明。信息的传递也不 是按照层级的关系逐级传递，因此对于应急物流乃 至应急供应链的运作模式还需要改变传统商业物流 问题的模式，从其供需关系的角度进行重新刻画。 确定不确定因素的来源：多数研究关注应急物 流不确定刻画，如果要从根本上解决不确定性问 题，需要从在应急物流运作分析基础上，探讨应急 物流中信息与物资的传递变化规律，进而发现信息 不确定的根源。此外，由社会舆论导向、信息传播 等因素产生的不确定性定量刻画与分析还相对薄 弱。目前无人机、遥感等技术手段可以用于灾情损 失核查，这些实时信息的更新对应急物流运作不确 定性条件的影响也需要充分考虑。 应急过程风险评估：应急物流决策活动以减轻 灾害产生的风险后果为目标，因此在自然灾害风险 评估体系中应纳入应急物流过程的风险要素。灾害 106 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 风险评估理论框架包括致灾因子危险性与承灾体脆 弱性评估两部分，多数是以损失评估为结果，并不 涉及与减轻灾害风险相关的决策活动，导致风险决 策与风险评估脱节。显而易见，应急响应各个环节 的决策活动对于灾害风险最终结果是有影响的，但 这种动态决策活动如何影响风险结果还没有理论指 导。如果将风险评估应用于指导决策，必须要将应 急过程风险内容加入到风险评估环节。 参考文献 [1] sheu j b. an emergency logistics distribution approach for quick response to urgent relief demand in disasters. transportation research part e: logistics and transportation review, 2007, 43(6): 687-709. [2] yi w, kumar a. ant colony optimization for disaster relief operations. transportation research part e: logistics and transportation review, 2007, 43(6): 660-672. [3] kovács g, spens k m. humanitarian logistics in disaster relief operations. international journal of physical distribution and logistics management, 2007, 37(2): 99114. [4] 陈丽群. 应急物流中物资调运研究综述. 物流科技, 2008, (12): 4-6. chen l q. renew on the research of material distribution in emergency logistics . logistics sci-tech, 2008, 31(12):4-6. [5] 李创. 国内外应急物流研究综述. 华东经济管理, 2013, (06): 160-165. li c. review on domestic and foreign research on emergency logistics. east china economic management, 2013, (06): 160-165. [6] 高文军, 陈菊红, 胡飞虎. 我国应急物流研究综述与展望. 物流科技, 2009, (08): 6-10. gao w j, chen j h, hu f h et al.the retrospect and prospect of chinese emergency logistics study. logistics scitech,2009,32(8):6-10． [7] galindo g, batta r. review of recent developments in or/ms research in disaster operations management. european journal of operational research, 2013, 230(2): 201-211. [8] news b. counting disaster: who's dying where? http://www.bbc.co.uk/news/science-environment15592761 . 2011-11-05. [9] 王晶, 张玲, 黄钧，等. 基于不确定需求的鲁棒应急物流 系统. 数学的实践与认识, 2009, 39(20): 53-60. wang j, zhang l, huang j. the robust emergency logistics system based on uncertain demand mathematics in prantice and theory, 2009, 39(20):53-60. [10] 张玲, 王晶, 黄钧. 不确定需求下应急资源配置的鲁棒优 化方法. 系统科学与数学, 2010, 30(10): 1283-1292. zhang l, wang j, huang j. robust optimal resource allocation model for uncertain demands.. journal of systems science and mathematical sciences, 2010, 30(10): 1283-1292. [11] yao t, mandala s r, chung b d. evacuation transportation planning under uncertainty: a robust optimization approach. networks & spatial economics, 2009, 9(2): 171-189. [12] ben-tal a, chung b d, mandala s r, et al. robust optimization for emergency logistics planning: risk mitigation in humanitarian relief supply chains. transportation research part b methodological, 2011, 45(8): 1177-1189. [13] chang m s, tseng y l, chen j w. a scenario planning approach for the flood emergency logistics preparation problem under uncertainty. transportation research part e logistics & transportation review, 2012, 43(6): 737-754. [14] barbarosoğlu g, arda y. a two-stage stochastic programming framework for transportation planning in disaster response. journal of the operational research society, 2004, volume 55(1): 43-53(11). [15] jia h, ordóñez f, dessouky m. a modeling framework for facility location of medical services for large-scale emergencies. iie transactions, 2007, 39(1): 41-55. [16] bozorgi-amiri a, jabalameli m s, al-e-hashem s m j m. a multi-objective robust stochastic programming model for disaster relief logistics under uncertainty. operations research-spektrum, 2013, 35(4): 905-933. [17] rawls c g, turnquist m a. pre-positioning of emergency supplies for disaster response// pre-positioning of emergency supplies for disaster response. ieee international symposium on technology and society. 1-9. [18] mete h o, zabinsky z b. stochastic optimization of medical supply location and distribution in disaster management. international journal of production economics, 2010, 126(1): 76-84. [19] 王旭坪, 李小龙, 郭武斌. 基于情景分析的应急路径选择 研究. 运筹与管理, 2012, (5): 67-72. wang x p, li x l, guo w b, et al. emergency path selection based on scenario analysis . operations research and management science, 2012,(5):67-72. [20] 缪成, 吴启迪, 许维胜. 突发灾害下可靠路径搜索模型与 算法. 计算机工程与应用, 2007, 43(28): 1-3. miao c, wu q d, xu w s .model and algorithm of reliable path finding under sudden-onset disaster. computer engineering and applications,2007, 43(28):1-3． [21] salmerón j, apte a. stochastic optimization for natural disaster asset prepositioning. production & operations management, 2010, 19(5): 561–574. [22] bozorgi-amiri a, jabalameli m s, alinaghian m, et al. a modified particle swarm optimization for disaster relief logistics under uncertain environment. international journal of advanced manufacturing technology, 2011, 60(1-4): 357-371. [23] xinhua he w h. modeling relief demands in an emergency supply chain system under large-scale disasters based on a queuing network. scientific world journal, 2014, 2014(1): 195053-195053. 107 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 [24] chakravarty a k. humanitarian relief chain: rapid response under uncertainty. international journal of production economics, 2014, 151(151): 146-157. [25] döyen a, aras n, barbarosoğlu g. a two-echelon stochastic facility location model for humanitarian relief logistics. optimization letters, 2011, 6(6): 1123-1145. [26] mete h o, zabinsky z b. preparing for disasters: medical supply location and distribution. [27] 孙莉. 不确定环境下应急物资配送优化研究. 兰州交通大 学, 2013. [28] beraldi p, bruni m e, conforti d. designing robust emergency medical service via stochastic programming. european journal of operational research, 2004, 158(1): 183-193. [29] beraldi p, bruni m e. a probabilistic model applied to emergency service vehicle location. european journal of operational research, 2009, 196(1): 323-331. [30] shen z, dessouky m, ordóñez f. stochastic vehicle routing problem for large-scale emergencies. ise working paper, 2007. [31] yuan y, wang d. path selection model and algorithm for emergency logistics management. computers & industrial engineering, 2009, 56(3): 1081-1094. [32] beamon b m, kotleba s a. inventory modelling for complex emergencies in humanitarian relief operations. international journal of logistics, 2006, 9(1): 1-18. [33] 王炜, 刘茂, 王丽. 基于马尔科夫决策过程的应急资源调 度方案的动态优化. 南开大学学报:自然科学版, 2010, (3): 18-23. w w, liu m, wang l. the dynamic optimal method of emergency resources deployment planning based on markov decision process. acta scientiarum naturalium universitatis nankaiensis. 2010, (3): 18-23. [34] economou a, fakinos d. a continuous-time markov chain under the influence of a regulating point process and applications in stochastic models with catastrophes. european journal of operational research, 2003, 149(3): 625-640. [35] 刘春林, 施建军, 李春雨. 模糊应急系统组合优化方案选 择问题的研究. 管理工程学报, 2002, 16(2): 25-28. liu c l,shi j j,li c y,et al.selection of the combinatorial optimal scheme for fuzzy emergency system. journal of industrial engineering and engineering management, 2002, 16(2):25-28. [36] zheng y j, ling h f. emergency transportation planning in disaster relief supply chain management: a cooperative fuzzy optimization approach. soft computing, 2013, 17(7): 1301-1314. [37] 杨勃, 李小林, 杜冰. 模糊环境下应急系统多目标调度问 题求解. 系统管理学报, 2013, 22(4): 518-525. yang b,li x l,du b,et al. solving multi-objective emergency scheduling problem in fuzzy environments. journal of systems & management, 2013, 22(4):518-525. [38] 钱佳. 不确定环境下模糊应急物资库存模型. 上海海事大 学学报, 2010, 31(1): 33-38. qian j. fuzzy inventory models of emergency supplies under uncertain environment. journal of shanghai maritime university, 2013, 22(4):518-525. [39] 王旭坪, 李小龙, 郭武斌. 基于情景分析的应急路径选择 研究. 运筹与管理, 2012, (5): 67-72. li x l, li x l, guo w b. emergency path selection based on scenario analysis. operations research and management science, 2012, (5): 67-72. [40] chang m s, tseng y l, chen j w. a scenario planning approach for the flood emergency logistics preparation problem under uncertainty. transportation research part e: logistics and transportation review, 2007, 43(6): 737-754. [41] barbarosolu g, arda y. a two-stage stochastic programming framework for transportation planning in disaster response. journal of the operational research society, 2004, 55(1): 43-53. [42] kotleba b m b, a. s. inventory modelling for complex emergencies in humanitarian relief operations. international journal of logistics, 2006, 9(1): 1-18. [43] 张英楠, 牟德一, 李辉. 基于机会约束规划的航班应急调 度问题研究. 中国安全科学学报, 2012, 22(12): 82. zhang y n, mou d y, li h ,et al.solving flight emergency scheduling problem based on chance-constrained programming. china safety science journal, 2012, 22(12):82-88. [44] charnes a, cooper w w. chance-constrained programming. management science, 1959, 6(1): 73-79. [45] chakravarthy s r, daniel j k. a markovian inventory system with random shelf time and back orders. computers & industrial engineering, 2004, 47(4): 315-337. [46] feeney g j, sherbrooke c c. the s-1, s inventory policy under compound poisson demand. manage sci, 1966, 12(5): 391-411. [47] zadeh l a. fuzzy sets . information & control, 1965, 8(65): 338–353. [48] 汪传旭, 邓先明. 模糊环境下多出救点应急救援车辆路径 与物资运输优化研究. 系统管理学报, 2011, 20(3): 269275. wanf c x, deng x m. multi-depot emergency vehicle routing and transportation optimization with fuzzy variables. journal of systems & management, 2011, 20(3): 269-275. [49] 田军, 马文正, 汪应洛,等. 应急物资配送动态调度的粒子 群算法. 系统工程理论与实践, 2011, 31(5): 898-906. tian j, ma w z, wang y l, et al. emergency supplies distributing and vehicle routes programming based on particle swarm optimization. systems engineering—theory & practice, 2011, 31(5): 898-906. [50] behret h, kahraman c. a multi ‐ period newsvendor problem with pre‐season extension under fuzzy demand. journal of business economics & management, 2010, 11(4): 17. [51] fiedrich f, gehbauer f, rickers u. optimized resource allocation for emergency response after earthquake disasters. safety science, 2000, 35(1–3): 41-57. 108 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 [52] 聂高众, 高建国, 苏桂武,等. 地震应急救助需求的模型化 处理—来自地震震例的经验分析. 资源科学, 2001, 23(1): 69-76. nie g z, gao j g, su g w, et al. models on rapid judgment for the emergent rescue needs during earthquake―by analysis on post-earthquake events.resources science,2001, 1, 23(1): 69-76. [53] 王楠, 刘勇, 曾敏刚. 自然灾害应急物流的物资分配策略 研究. 第五届中国物流学术年会，2006. [54] sheu j b. dynamic relief-demand management for emergency logistics operations under large-scale disasters. transportation research part e logistics & transportation review, 2010, 46(1): 1-17. [55] choi t m, li d, yan h. quick response policy with bayesian information updates. european journal of operational research, 2006, 170(3): 788-808. [56] choi t m, li d, yan h. optimal single ordering policy with multiple delivery modes and bayesian information updates. computers & operations research, 2004, 31(12): 1965-1984. [57] iyer a v, bergen m e. quick response in manufacturerretailer channels. management science, 1997, 43(4): 559570. [58] eppen g d, iyer a v. improved fashion buying with bayesian updates. operations research, 1997, 45(6): 805819. [59] taskin s, lodree e j. inventory decisions for emergency supplies based on hurricane count predictions. international journal of production economics, 2010, 126(1): 66-75. [60] lodree e j, taskin s. supply chain planning for hurricane response with wind speed information updates. computers & operations research, 2009, 36(1): 2-15. [61] 詹沙磊, 刘南. 基于灾情信息更新的应急物资配送多目标 随机规划模型. 系统工程理论与实践, 2013, 33(1): 159166. zhan s l, liu n. multi-objective stochastic programming model for relief allocation based on disaster scenario information updates. systems engineering —theory & practice, 2013, 33(1): 159-166. [62] gong q, batta r. allocation and reallocation of ambulances to casualty clusters in a disaster relief operation. iie transactions, 2007, 39(1): 27-39. [63] 朱建明. 损毁情景下应急设施选址的多目标决策方法. 系 统工程理论与实践, 2015, (3): 720-727. zhu j m. methods of multi-objective decision-making for emergency facility location problem under failure scenario. systems engineering —theory & practice, 2015, (3): 720727. [64] huang m, smilowitz k, balcik b. models for relief routing: equity, efficiency and efficacy. procedia social and behavioral sciences, 2011, 17(1): 416-437. [65] zhang z h, jiang h. a robust counterpart approach to the bi-objective emergency medical service design problem. applied mathematical modelling, 2014, 38(3): 1033-1040. [66] pagnoncelli b k, ahmed s, shapiro a. sample average approximation method. mat puc. [67] mulvey j m, zenios s a. robust optimization of largescale systems. operations research, 1995, 43(2): 264-281. [68] 刘波, 李波, 李砚. 不确定条件下应急资源布局的鲁棒双 层优化模型. 计算机工程与应用, 2013, 49(16): 13-17. liu b, li b, li y. robust bi-level optimization model for emergency resource location and allocation under uncertain condition .computer engineering and applications, 2013,(16):13-17. [69] 姜涛, 朱金福. 应急设施鲁棒优化选址模型及算法. 交通 运输工程学报, 2007, 7(5): 101-105. jiang t, zhu j f. robust optimization model and algorithm of emergency establishment .journal of traffic and transportation engineering, 2007, 7(5):101-105. [70] 王保华, 何世伟. 不确定环境下物流中心选址鲁棒优化模 型及其算法. 交通运输系统工程与信息, 2009, 9(2): 69-74. wang b h, he s w. robust optimization model and algorithm for logistics center location and allocation under uncertain environment. journal of transportation systems engineering and information technology, 2009, 9(2): 69-74. [71] 钟慧玲, 庄楠, 张冠湘，等. α-鲁棒的危险品道路运输应 急设施选址问题. 系统工程理论与实践, 2013, 33(5): 1262-1268. zhong h l, zhuang n, zhang g x, et al. α -robust emergency facilities location problems for hazardous materials transportation. systems engineering —theory & practice, 2013, 33(5): 1262-1268. [72] 朱雷, 黎建强, 汪明. 不确定条件下应急管理人力供应链 多功能资源配置鲁棒优化问题. 系统工程理论与实践, 2015, (3): 736-742. zhu l, li j q, wang m. multi-resource robust optimization of emergency human resource supply chain management under uncertainty. systems engineering —theory & practice, 2015, (3): 736-742. [73] 陈涛, 黄钧, 朱建明. 基于信息更新的两阶段鲁棒-随机优 化调配模型研究. 中国管理科学, 2015, 23(10): 67-77. chen t, huang j, zhu j m. two-stage robust-stochastic decision model for relief allocation based on disaster scenario information updata. chinese journal of management science, 2015, 23(10): 67-77. [74] yao t, bental a, chung b d, et al. robust optimization for dynamic traffic assignment under demand uncertainty// robust optimization for dynamic traffic assignment under demand uncertainty. transportation research board 89th annual meeting, 2010. [75] 海姆斯. 风险建模、评估和管理. 西安: 西安交通大学出 版社, 2007. 109 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 101-109 atlantis press journal style received 11 october 2014 accepted 28 january 2015 research on society risk evolution mechanism and counter measures in severe emergency infectious disease — in the case of h7n9 avian influenza xuanhua xu, chenguang cai *, chunhong wang school of business, central south university, changsha 410083, china abstract in this paper, the evolution and control of society risks caused by severe emergency infectious disease will be analyzed and studied. firstly, the evolution chain of society risks caused by severe emergency infectious disease is constructed to analyze the evolution rule of society risks and identify the essential factors in counter measures. then, the system dynamics model is established and employed to simulate the effects of society risk control with various countermeasures. finally, based on the simulation results, a conclusion is drawn: improving medical treatment capacity, strengthening quarantine level under epidemic situation and enhancing the effective ness of dealing with public opinions are effective in controlling society risks. especially when the three aspects mentioned above are promoted at the same time, the society risk control can achieve a more notable effect. keywords: severe emergency infectious disease; society risk; evolution; counter measure; disaster chain; system dynamics. *corresponding author: ccg169@126.com 1. introduction in recent years, severe emergency infectious diseases broke out frequently, such as sras in 2003, h5n1avian influenza in 2009, eye disease in 2012, h7n9 avian influenza in 2010, etc. on one hand, the occurrence of epidemic poses threats to human health and life. on the other hand, the increasingly cumulative society risk caused by epidemic gives rise to various social riots and exerts immense adverse effects on social stability and economic order. therefore, it is of great necessity to carry out research on evolution mechanism and countermeasures of society risk given rise by severe infectious disease. by analyzing the evolution process of society risk caused by severe emergency infectious disease, the basic rule of society risk evolution is grasped and the key elements during society risk evolution is identified. on this basis, the coping strategy is developed to improve the effectiveness of social risk control and reducing the probability of society risk outbreak. the dynamic evolution process of disaster has its own law and mechanism. therefore, only by grasping the evolution law and mechanism of disaster can the social risk initiated thereof be reduced to a minimum. over the past, enormous studies have been conducted on the evolution mechanism of flood, earthquake, extreme weather, drought and other natural disasters, focusing on forming a complete disaster chain and identifying potential risks through simulating the evolution process of disasters. in addition, a suitable method is necessarily proposed in studying the evolution mechanism of risk. li etc. [5] made a cross coupling analysis of the over two thousand disasters from 2001 to 2010 by statistical method, and summarized the occurrence rule of disasters in recent ten years in our country. peng [6] studied the influencing mechanism of gas seepage generated in journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 54-65 published by atlantis press copyright: the authors 54 sudden disaster by conducting the thermal coupling experiment. wind [7] investigated the post-flood psychological health of community residents in the towns and countryside of northern england, and studied the evolution rule of residents’ mental health. he [8] proposed a chaotic differential evolution algorithm, providing a better method to predict the evolution process of flood disaster. tinguaro, etc. [9] proposed a data-based bipolar intellectual decision support system prototype, where society risks after natural disaster were managed through the construction and application of induced rules. on the part of identifying and evaluating disaster risk, du[10] and birkmann [11] conducted some researches on identifying possible risks after natural disasters. guikema [12] put forward a risk assessment system on natural disaster and concrete method based on a large amount of statistical data, and applied it in the field of infrastructure system. from the perspective of meteorology, geography, disaster science and environmental science, zhang etc. [14] proposed a risk identification and assessment method of droughts based on gis (geographical information systems). zio [15] described the issues concerning the post-disaster risk identification resulted from such disaster as nuclear and gas leak, and proposed a risk assessment and management framework. jiang [16] utilized the fuzzy mathematics theory and method, and applied the fuzzy comprehensive evaluation method, fuzzy clustering and fuzzy similarity method into the risk identification, classification and assessment of floods, providing an effective method for risk management of flood disaster. in the field of risk coping strategy, hu and zhu [17, 18],with meteorological disaster as the research object, described the method of preventing disaster risk. batabyal[19], targeting at financial risks of disaster, presented the theory of financial risk management caused by natural disaster and its warning model. liu [20] put forward a decision analysis method of risk based on prospect theories, and applied it into the emergency response risk decision analysis. huang [21] applied three risk analysis models of natural disaster into practical decision-making problems. on the basis of the utility function theory, tamura [22] presented the risk analysis and decision model for natural disasters. at present, research on social risks mainly focuses on natural disaster itself. nevertheless, studies on social risks caused by infectious disease are far from sufficient. thus, this paper, with social risks caused by major infectious diseases as the research subject, is to construct the evolution chain of social risk to describe its process, and on this basis, to identify the key elements of risk response strategy and simulate the effectiveness of risk control with different strategies by means of the system dynamics model. the finding obtained in this paper is of certain reference value to develop and implement the prevention and control plan of severe infectious disease. 2. society risk evolution mechanism in severe emergency infectious disease social risk is defined as the possibility of giving rise to social conflicts and endangering social stability and order, namely, the possibility of social crisis outbreak. once this possibility becomes a reality, social risk is transformed into social conflict, producing disastrous effects on social stability and order. during sars and avian influenza outbreak, many events of social disruption effects due to epidemic situation broke out nationally, for example, many college students and migrant workers rushed to their hometown during sars, which formed a so-called “homebound rush”; during avian influenza, some people deceived by rumors sparked a run on medical supplies, daily necessities, and some unscrupulous merchants seized the opportunity to bid up prices, disrupting social economic order; some lawbreakers began to publicize feudal superstitions, inciting a crowd riot. the cases cited above are all caused by the accumulation of social risks. the social risk system (fig. 1) mainly covers three basic elements: hazard-formative factor, hazardinducing environment, and hazard-affected body. take social risk caused by avian influenza as an example, the outbreak of avian influenza is the incident that touches off the generation and evolution of social risk, which is thus defined as hazard-formative factor. hazardformative environment falls into two types: natural environment and social environment. in terms of natural environment, epidemic spreads very rapidly because of bird migration in spring and fall, and meanwhile the warm and humid climate in spring also provides favorable conditions for the spreading of disease. in the part of social environment, most of poultry farms are short of necessary monitoring and prevention measures against new virus, and in addition, the spreading of rumors and psychological fear of human beings published by atlantis press copyright: the authors 55 accelerates the formation and accumulation of social risks. hazard-affected body is mainly the human beings with psychological fear and vulnerability of social environment. when the number of residents with fear accumulates to a certain number, social risk will burst out, and then a variety of group social events will occur, causing negative influences on social development and stability. the evolution chain of society risk caused by severe emergent infectious disease is shown in fig. 2. when bird flu outbreaks at a certain moment, social risk began to generate; as time passes by, the number of people infected and the death toll are on constant increase, and all kinds of rumors began to spread, causing the psychological fear of residents to soar. in turn, the fear of residents also intensifies the spreading of rumors to a certain extent, making more people infected by rumors. thus, social risks began to accumulate rapidly. the enlargement of social risk is mainly manifested in two aspects: intensifying of psychological fear of residents and the increasing number of people infected by rumors. once social risk accumulates to a certain extent, social risk will burst out, giving rise to a series of mass events. meanwhile, these group social events will in turn speed up the accumulation of social risks. hazard-inducing environments hazard -affected bodies fears of residents; vulnerability of social environment birds migratory; weather conditions; regulatory loopholes of farm hazard -formative factors outbreak of avian influenza fig. 1. social risk system of avian influenza. increasing number of people infected by rumors epidemic breakout increasing number of infected people increasing death toll rumors beginning to spread increasing level of resident’s fear increasing number of mass incidents stage of risk generating stage of risk spreading stage of risk assembling stage of risk outbreak fig.2. evolution chain of social risk of avian in fluenza. published by atlantis press copyright: the authors 56 3. running head society risk control strategies of severe emergent infectious disease 3.1. key elements of the strategy from fig. 1 and fig. 2, it is known that there are primarily three influence factors in the formation and evolution of social risk: the death toll of infections, the spreading degree of epidemic and rumors. therefore, "strengthening the medical treatment ability", "enhancing the quarantine level during epidemic situation "and "promoting the effectiveness of handling public opinion are the key elements in formulating the risk control strategy. (1) strengthening the medical treatment ability to reduce the mortality of patients. the purpose of strengthening the medical treatment ability is to save the life of infections and minimize the mortality rate. once a person is found to infect with bird flu, he or she should be subject to early report, early quarantine and early treatment. moreover, the development of vaccine and its application in clinical trials should be accelerated as soon as possible. (2) enhancing the quarantine level to prevent the further spreading of epidemic. bird flu and sras has varied mode of infection. according to the existing studies, avian influenza can only be transmitted through "bird to bird" and "bird to human", excluding "person to person". however, in practice, once a bird-flu patient is diagnosed, quarantine treatment is conducted firstly, and meanwhile his or her contacts must be put in quarantine for a few days. when avian influenza outbreaks, the poultry farm and surrounding farms need to be disinfected and all the poultry therein should be killed, and vehicles carrying the livestock and poultry should be disinfected. through controlling the infection source and cutting off the mode of infection, the epidemic will be prevented from spreading to the surrounding area. (3) handling public opinions effectively to reduce their impacts on the masses. people’s fear of the epidemic is mainly because h7n9 avian influenza virus is a new one with few known information, which causes uneasiness over the mass psychology. especially when the number of infections and the death toll continues to rise, people’s fear will intensity and rumors begin to spread. confronted with the above situation, a series of measures should be taken to clarify the rumors timely and effectively, crack down on such illegal activities as spreading of rumors by network, publications and other means so as to reduce the effect of rumors on residents. 3.2. a system dynamics model over emergency infectious disease epidemic situation of social risk control effect in order to display the effect of social risk control emergency strategy on emergency infectious diseasein a more visual way, the system dynamics (sd) model is constructed to simulate the process of society risk control. the recovery rate, the source processing efficiency and the rumor clarification rate are optioned to represent the medical treatment capacity, the epidemic quarantine level and rumor processing effectiveness of emergency strategy, respectively. take h7n9 outbreak in shanghai as an example, based on the relationship between different variables, relevant parameters and formula are set as shown in fig.3. (1) basic situation of the epidemic area. shanghai city has an area of 8064.3 square kilometers and a total of 23.83 million populations. the area is distributed with dozens of poultry farms, with each producing about 200000 feather on average. by the end of july 16, 2013, a total of 30 h7n9 infected cases, including 11 death cases, has been reported in shanghai. the mortality was 36.7%, slightly higher than the national average. according to the relevant statistics, during the outbreak, mass incidents caused by epidemic have up to dozens of times, which produced a huge influence on the normal life of residents and social stability. (2) medical treatment of avian influenza. h7n9 is a highly lethal avian influenza. if the patient fails to receive timely and effective treatment, their life would be threatened. therefore, the recovery rate of avian flu is set as the index to evaluate the treatment, namely, the higher the treatment level, the better the medical effect on the avian influenza, and the lower the mortality rate. the infection probability of avian influenza through contact is 6.12% to 7.63%, sothe parameter of the infection probability of avian influenza through contact is set as random uniform (0.0612, 0.0763, 0.001). the infection rate refers to the number of newly infected patients per day, which is associated with the infection probability through contact and the number of people having contact with the virus. thus, the infection rate of avian influenza is set as: the infection probability of avian influenza through contact* the number of people contacting the virus. published by atlantis press copyright: the authors 57 (3) epidemic quarantine and control. the infection mechanism of h7n9 avian influenza is different from that of sars. it has not confirmed thath7n9 avian influenza can be transmitted by "person to person". the major infection source of h7n9 avian influenza is the bird and poultry with avian influenza virus. the epidemic quarantine and control, on one hand, is to disinfect the farm where avian influenza breaks out so as to reduce the probability of human exposure to the infection source. on the other hand, disinfection of the ground and vehicles used in the farm in epidemic area should be strengthened to cut off the transmission route of virus and reduce the diffusion rate of virus. the infection treatment efficiency index is introduced to indicate the quarantine and control level of an epidemic. there is a positive correlation between the infection treatment efficiency and the control level of epidemic situation. according to relevant statistical data, it can be obtained: the disinfection level of infected avian =50000* processing efficiency of infectious source (the quarantine and control level of an epidemic). the diffusion coefficient of epidemic occurred in a farms refers to the number of newly infected farms nearby at unit time when bird flu hits one farm, which is mainly infected by the characteristics of avian influenza virus and bird migration. avian influenza often take place at the beginning of the spring or autumn until the rising or falling temperature is unsuitable for virus survival as time passes by. in addition, migratory birds are obviously seasonal. thus, spring and autumn are two high-occurrence seasons of bird flu. the diffusion coefficient of epidemic taken place in a farm changes with time when the climate is warming or cooling, and the scale of bird migratory gradually decreases until the end. it can be concluded that the diffusion coefficient of an epidemic is in a negative correlation with time. according to relevant statistical data concerning the spreading of h7n9 bird flu in shanghai, the independent variable “time” is set to number of infectionsinfection rate region's population infection probability of avian influenza through contact processing efficiency of infectious source (the quarantine and control level of an epidemic) recovery rate of avian flu (medical treatment) mortality of infections number of rehabilitation death toll of infections resident number of infected by rumor increasing rate of the number of residents infected by rumors residents fear index frequency of mass incidentsoutbreak of massincidents <time> probability of rumor spreading number of residents having contact with avian influenza virus trust degree of clarification information diffusion coefficient of rumor total area of the region diffusion coefficient of the epidemic in the farm number of patients from other regions into the local region virus contact probability decrease rate of the number of residents infected by rumors threshold value of mass incidents breakout effectiveness of disease prevention and control measures rumor clarification rate (the level of public opinion processing) number of infected poultry decrease in the number of sick poultry average size of farms number of farms with epidemic outbreaks diffusion rate of farms outbreak average infection rate increase in the number of sick poultry disinfection level of infected avian quarantine level of epidemic number of group social events occurred that are psychologically acceptable to residents maximum number of infections that are psychologically acceptable to residents maximum number of deaths that are psychologically acceptable to residents fig. 3. system dynamics model of social risk control of disease epidemic situation. published by atlantis press copyright: the authors 58 fit the diffusion coefficient of the farm with epidemic: the diffusion coefficient of the epidemic in the farm=max ((1.8254* (1/ (1.4525*time+1)) 0.000256*time), 0). the increase rate of farms with epidemic outbreak is the number of newly infected farms at unit time, which is associated with the diffusion coefficient,the quarantine level and the number of farms infected. the increase rate of farms with epidemic outbreak = the diffusion coefficient of epidemic in the farm * (1-the quarantine level of epidemic) * the number of farms with epidemic outbreaks. according to the occurred cases, most of patients have contact with poultry before infection, so the probability ofinfection is related with the number of resident who isexposed to the infection. based on actual data,the equation is obtained: the probability of infection through contact = the number of infected poultry *0.85*10^-11, the number of residents having contact with avian influenza virus mainly includes local ones and those coming from other places. so the number of residents having contact with avian influenza virus = the total population in the area * probability of people having contact with virus + the number of patients from other regions into the local region. (4) social risk processing. relevant parameters of social risk comprise the number of rumors, the number of residents infected by rumors and the number of group events, etc. among them, the resident fear index refers to the fear of ordinary residents for epidemic situation, the number of infections and the number of group social events caused by epidemic. it can be expressed as the following equation: the residents fear index =0.2* frequency of mass incidents/ the number of group social events occurred that are psychologically acceptable to residents +0.4* the number of infected people /maximum number of infections that are psychologically acceptable to residents +0.4* deaths /maximum number of deaths that are psychologically acceptable to residents. the psychological fear index of residents is between [0, 1], the greater the index, the more fear the residents have psychologically. we use questionnaire to determine the coefficients mentioned in the equation above. i n this paper, the number of maximum acceptable group social events is set 100, the number of maximum acceptable infections is set 100 and the number of maximum acceptable deaths is set 50. the number of residents infected by rumors is connected with the spreading rate of rumors, the number of rumors and the effectiveness of rumor processing. the increase rate of residents infected by rumors =probability of rumor spreading* diffusion coefficient of rumor * (1+0.002* the resident number of infected by rumor/ (1+ the resident number of infected by rumor)). decreasing the number of residents infected by rumors is mainly achieved by eliminating the rumor influence by clarifications. the effectiveness of rumor refuting is mainly concerned with the rumor clarification rate and the trust degree of residents for clarification information. the rumor clarification rate refers to the ratio between the number of rumors clarified and the number of spreading rumors, the greater the rate is, the more people will be affected. the trust degree of clarification information mainly lies in the effectiveness of epidemic prevention and control measures, namely, the more effective the epidemic prevention and control measures is, the greater the trust degree of residents have. the decrease rate of the number of residents infected by rumors = rumor clarification rate (the level of public opinion processing) * the resident number of infected by rumor *the trust degree of clarification information. the frequency of mass incidents is the main criteria to evaluate the control level of social risks, which is mainly related to the number of residents infected by rumors. according to the survey, we find the threshold value of mass incident outbreak is 30000 persons average, meanly that it will break out once the number of residents infected by rumors exceeds the threshold. 3.3. social risk control effect under different conditions in initial state, supposing the number of farms with epidemic outbreak is 1, the simulation step size is 1 day and the end time is 100 days, the system dynamics model is then used to simulate the change of social risks under different medical treatment level, epidemic quarantine control level and effectiveness of public opinion handling. according to the actual prevention and control situation of avian influenza in shanghai, in the situation 0, the three exogenous variables, including the recovery rate, source processing efficiency and rumors clarification rate, are assigned as 0.66, 0.6 and 0.7 respectively, which are then used in the system dynamics model for simulation. as the simulation published by atlantis press copyright: the authors 59 results are basically consistent with actual data, this model is proven to be effective and the 3 exogenous variables are set as the initial parameters. 3.3.1. impact of medical treatment ability on risk society control on the premise that other conditions remains unchanged, the medical treatment level is improved, with the recovery rate grew by 10% and 20% respectively. that is to say, in situation 1, the recovery rate = 0.73; in situation 2, the recovery rate = 0.89; in situation 0, the recovery rate is the initial parameter. the simulation results are shown in the fig. 4 to fig. 6 below. as shown in fig. 4 to fig.6, with the improvement of medical treatment level, the recovery rate of infections was improved, the psychological fear index fig.4. impact of different medical treatment level on residents fear index fig.5. impact of different medical treatment level on the number of residents infected by rumor. fig.6. impact of different medical treatment level on the frequency of mass incidents. published by atlantis press copyright: the authors 60 of residents leveled off after rising to a certain degree. compared with the initial stage, the number of residents affected by rumors shows a downward trend, the frequency of mass incidents is fewer, indicating a more obvious effectiveness in social risk control. 3.3.2. impact of the quarantine and control level of an epidemic on society risk control on the premise that other conditions remains unchanged, the control level of an epidemic is improved, the processing efficiency of infection source is increased by 10% and 20% respectively. that is to fig.7. impact of different quarantine and control level on residents fear index fig.8. impact of different quarantine and control level on the number of residents infected by rumor. fig.9. impact of different quarantine and control level on the frequency of mass incidents. published by atlantis press copyright: the authors 61 say, the processing efficiency of infection source is set as 0.66 in situation 3,0.72 in situation 4and the initial parameters in situation 0. the simulation results are shown in fig. 7 to 9 below. as shown in fig. 7 to fig. 9, with the improvement of processing efficiency of infection source, the rise of residents fear index gradually leveled off and the number of residents infected by rumors is on the decrease, thus group social events will not occur basically. 3.3.3. impact of the rumor processing level on risk society control on the premise that other conditions remains unchanged, the efficiency of public opinion processing is improved, with the rumor clarification rate increased by 10% and 20% respectively. that is to say, the rumor processing efficiency is 0.77 in situation 5, 0.84 in situation 6 and is the initial parameter in situation 0. the simulation results are shown in fig. 10 to fig. 12. as the rumor clarification rate rises, the inflection point of resident fear index becomes low gradually, the increase rate of the number of residents infected by fig.10. impact of different rumor processing level on residents fear index. fig.11. impact of different rumor processing level on the number of residents infected by rumo. fig.12. impact of different rumor processing level on the frequency of mass incidents. published by atlantis press copyright: the authors 62 rumor is slowing, and the frequency of less mass incidents is decreasing. the above results indicate that social risks can be controlled, but the control effect is still not ideal. 3.3.4. impact of coordination strategy on social risk control collaborative strategy is to control the social risks caused by epidemic through improving the medical treatment capacity, quarantine level and rumor clarification level at the same time. suppose that the above three aspects are increased by 10% and 20%, respectively. that's to say, in situation 7, the recovery rate = 0.73, the processing efficiency of infection source = 0.66, the rumor clarification rate = 0.77; in situation 7, the recovery rate = 0.8, the processing efficiency of infection source =0.71, and the rumor clarification rate= 0.84; in situation 0, they are all the initial parameters. the simulation results are shown in fig. 13 to fig. 15. as shown in fig. 13 to 15, the higher level the ability of collaborative strategy, the lower the increase rate of resident fear and the number of residents infected fig.13. impact of different collaborative strategies on residents fear index. fig.14. impact of different collaborative strategies on the number of residents infected by rumor. fig.15. impact of different collaborative strategies on the frequency of mass incidents. published by atlantis press copyright: the authors 63 by rumors and the more effective the social risks control. compared with the coping strategy aiming at improving a single aspect, the collaborative strategy can realize effective control over all indexes related with social risks, thus obtaining more obvious control effects. therefore, in the actual operation, collaborative strategy should be given priority to cope with social risks caused by severe emergent epidemics. 4. conclusion this paper is mainly aimed at studying the key components, generation and evolution process of social risks caused by severe emergent epidemic, and simulating the effect of social risk control with different strategies by means of the system dynamics model. the simulation results indicated that it is effective to control social risks through improving the medical treatment ability, strengthening the quarantine control level and enhancing the effectiveness of public opinion handling, especially when the above-mentioned three aspects are strengthened at the same time. acknowledgements this paper was supported by the key program of nation al social science foundation of china (no. 12azd109) and the national natural science foundation of china (no. 71171202). references [1] w. zhu, c. k. chen, d.x. ji and y. f. sun, analysis on the risk and evolution process of rainstorm disaster in cities of north china, j. catast. 26(3) (2011) 88-91. [2] c.d. chen, y. f. sun and z. li, characteristic analysis of evolution and derivation chain of risk events caused by snow and ice disasters, j. catast.24(1)(2009)18-21. [3] z.l.xie and z. j. ma, analysis and simulation of evolution mechanism of urban seismic secondary disasters,j. nat. disast.21（3）(2012)155-163. [4] r. zhou,c. q. guo,q. j. fu and l. y. pan,study on the drought and flood disasters formation mechanism in karst regions of middle guangxi, pro. eng. 28(1) (2012) 277-281. [5] r. q. li, s. l. shi, q. f. nian and m. jiang, research on coalmine gas accident rules in china in recent decade,j. china safety sci. 21(9)(2012)143-151. [6] s. j. peng, j. xu and h. w. yang, experimental study on the influence mechanism of gas seepage on coal and gas outburst disaster, safety sci.50(4) (2012) 816-821. [7] t. r. wind and i. h. komproe, the mechanisms that associate community social capital with post-disaster mental health: a multilevel model, social. sci. amp; med. 75(9)(2012) 1715-1720. [8] y. he, j. zhou and p. kou, a fuzzy clustering iterative model using chaotic differential evolution algorithm for evaluating flood disaster, expert. syst. appl.38(8)(2012) 10060-10065. [9] r. j. tinguaro, b. vitoriano and j. montero, a general methodology for data-based rule building and its application to natural disaster management, comput. oper. res. 39(4)(2012) 863-873. [10] x. y. du and x. f. lin, conceptual model on regional natural disaster risk assessment, pro. eng. 45(2) (2012) 96-100. [11] j. birkmann, risk and vulnerability indicators at different scales: applicability, usefulness and policy implications, enviro. hazards,7(2007) 20-21. [12] s. d. guikema, natural disaster risk analysis for critical infrastructure systems: an approach based on statistical learning theory,reliab. eng. syst. safe. 94(4)(2009) 855–86. [13] n. altay and w. g. green, or/ms research in disaster operations management, eur. j. oper. res. 175 (2006)4 75–493. [14] j. q. zhang, risk assessment of drought disaster in the maize-growing region of songliao plain, china,agr. ecosyst. environ. 02(2) (2004) 133–153. [15] e. zioa and t. avenc, industrial disasters: extreme events, extremely rare. some reflections on the treatment of uncertainties in the assessment of the associated risks,process safe environ. 91(1)(2012) 3135. [16] w. g. jiang, l. deng, l. y. chen,j. j. wu andj. li, risk assessment and validation of flood disaster based on fuzzy mathematics, prog. nat. sci.19(10)(2009) 1419–1425. [17] a. j. hu, n. li and y. d. zhu, theory of meteorological disasters comprehensive risk prevention modes: the reflection of cold rain and snow freeze disaster in southern china 2008. prog. geo. 2(2010)159-165. [18] y. d.zhu, a. j. hu, y. p. xiong andy. he, economic development and the weather risk management(china financial economic press, beijing,2006). [19] a. a. batabyal, h. beladl, aspects of the theory of financial risk management for natural disasters, appl. math. lett.14(2001) 875-880. [20] y. liu, z. p. fan and y. zhang, risk decision analysis in emergency response: a method based on cumulative prospect theory, comput. oper. res, 42(1)(2012)75-82. published by atlantis press copyright: the authors 64 [21] c. f. huang and h. inoue, soft risk maps of natural disasters and their applications to decision-making, inform. sci. 177(2007) 1583–1592. [22] h. tamura, k. yamamoto, s. tomiyama, i. hatono, modeling and analysis of decision making problem for mitigating natural disaster risks. eur. j. oper. res.122(2000) 461-468 published by atlantis press copyright: the authors 65 1. introduction 2. society risk evolution mechanism in severe emergency infectious disease 3. running head society risk control strategies of severe emergent infectious disease 3.1. key elements of the strategy 3.2. a system dynamics model over emergency infectious disease epidemic situation of social risk control effect 3.3. social risk control effect under different conditions 3.3.1. impact of medical treatment ability on risk society control 4. conclusion acknowledgements references microsoft word volume 13, issue 2-3 journal of risk analysis and crisis response, 2023, 13(2), 136-147 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i2.361 136 article analysis of the impact of corporate income tax incentives on digital economy enterprises shu-xian wang 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: wangshuxian0426@163.com; tel.: +86-13598629108 received: february 27, 2023; accepted: may 31, 2023; published: june 30, 2023 abstract: this paper evaluates the effect of income tax preferences for digital economy enterprises based on the data of digital economy enterprises in beijing, tianjin and hebei regions from 20172021 (in line with the narrower caliber of the national bureau of statistics). the research results show that corporate income tax incentives have a significant impact on r&d investment of digital economy enterprises and provide a better incentive for digital economy enterprises to increase their r&d investment efforts. tax incentives are heterogeneous in nature of property rights for digital economy enterprises, among which there is a greater promotion for private enterprises. finally, this paper further expands the study on the factors influencing the innovative capability of enterprises and proposes policy recommendations on the relationship between tax preferences and innovation capability of enterprises based on the empirical results. keywords: digital economy; tax incentives; enterprise innovation 1. introduction the development level of china's digital economy has made a remarkable leap to a higher stage in the 14th five-year plan period. the value of china's digital economy industry is also steadily increasing, and the value added of the core industries of the digital economy is also increasing as a proportion of gdp, reaching 7.8% by 2020. at the same time, in the context of a unified market, the government has called for accelerated digitization, enabling the integration of online and offline, promoting joint development between regions and using the digital economy to break down regional boundaries. in september 2020, beijing introduced the "1+3" policy on digital economy development, which focuses on the creation of a digital trade pilot zone, the construction of a pilot project on the safe management of cross-border data flows and the establishment of the beijing international big data exchange, in order to accelerate the development of the digital economy. the city of tianjin pointed out that tianjin, as a bridge for the collaborative development of the beijing-tianjin-hebei region, will continue to accelerate the integration and docking of data resources in the beijing-tianjin-hebei region, play a bridging role, collaborate to build data parks and strengthen the cooperation of data enterprises among the three regions; hebei proposed that it will focus on supporting the development of the xiong'an new area in the region, put the focus of digital economy development on the digital transformation of the manufacturing industry, and collaborate with the development strategy of beijing-tianjin and national events. shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 137 in addition, in 2021, the beijing-tianjin-hebei digital economy alliance will be established in tianjin. the alliance is a spontaneous collaboration between the alliance is a spontaneous collaboration of more than 50 organizations from industries, universities, research institutes, financial institutions and associations related to the digital economy. it is a non-profit organization with a wide range of fields and a strong openness. the alliance focuses on the development of the digital economy, strengthens cooperation and communication between various entities, and contributes to the healthy development of the digital economy in the beijing-tianjin-hebei region. as technology continues to develop and evolve, most traditional industries are now striving to integrate technological innovations with the digital economy, which has become the trend of development in the world today. the replicability and reusability of data in the digital economy gives companies that enter the digital economy market first a clear development and innovation, the digital economy has become a global trend. therefore, by increasing tax incentives, the government encourages the transformation and upgrading of enterprises and helps them to accelerate the pace of technological innovation. this is also conducive to china gaining a head start in the development of the digital economy in the world competition and improving our digital competitiveness and international status. in the current situation, the government has introduced various preferential policies, including tax incentives, in order to encourage the development of innovation in enterprises and to enhance the innovation ability of enterprises. corporate income tax incentives are the most typical of all tax incentives. one is the preferential income tax rate of 15% for high-tech enterprises, and the other is to add deductions for r&d expenses when calculating taxable income, and indirect preferences such as tax exemption for technology transfer. the regulatory role of tax incentives for the development of enterprises in the digital economy has become necessary. this paper focuses on the first situation of corporate income tax incentives. through empirical analysis, this paper takes digital economy enterprises from 2017 to 2021 as the research object, and finally includes 142 digital economy enterprises in the study. after the empirical analysis of the tax incentives for these digital economy enterprises, we propose recommendations for the innovative development of digital economy enterprises based on the empirical results. 2. review of the literature the government has a certain incentive for the development and innovation of enterprises through fiscal and taxation instruments. such incentives are important not only to accelerate the digital transformation of enterprises, but also to contribute to the innovative transformation of society. for this reason, some scholars have studied the theoretical basis and effects of tax incentives. the first is the question of whether tax incentives can, to a certain extent, motivate enterprises to invest more in r&d. song qing et al. (2021) conducted a heterogeneity test on gem listed companies and found that tax incentives can significantly promote r&d investment in different regions. chen dong (2020) found that tax incentives have a significant effect on the r&d investment of enterprises. in summary, it can be concluded that the incentive effect of tax incentives is very significant when enterprises increase their investment in r&d to make them innovate faster. the companies studied above are all traditional companies. in the digital context, it is still unknown whether the government's tax incentives can stimulate the r&d investment of digital economy enterprises. the existing literature has mostly explored such issues from a theoretical level. shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 138 ma hongfan et al. (2021) argue that the development of enterprises should follow the development trend of the digital economy, and with the continuous improvement of the tax governance system and the enhancement of governance capacity, the tax governance approach should be transformed and upgraded in the direction of digitalization; li hui et al. (2021) argue that for the issue of digital economy and consumption tax collection and administration, the principle of consumption tax collection should be adhered to create a fair business environment for enterprises, which also makes tax collection and administration regarding the digital economy to be more regulated. some scholars have also conducted relevant studies on taxation and the development of the digital economy from the empirical level. based on provincial panel data, ye xu et al. (2021) found that the digital economy significantly influenced the employment structure of industries, sectors and skills, and generally contributed to the transformation and upgrading of the employment structure in the direction of high technology; yu changlin et al. (2021) argued that government subsidies, tax incentives and other industrial policies could continuously promote technological innovation in the digital economy. a comprehensive review of literature reveals that the existing studies on digital economy enterprises and tax incentives are more focused on theoretical analysis, while empirical studies are lacking. most of the studies have been conducted from a macro perspective, with relatively little research on micro enterprises. therefore, this paper analyzes the impact of tax incentives on digital economy enterprises in the beijing-tianjin-hebei region from an empirical perspective by constructing an econometric model and conducting heterogeneity tests to analyze the impact of tax incentives on digital economy enterprises in the beijing-tianjin-hebei region from different perspectives. 3. theoretical analysis and research hypothesis there are two types of tax preferences regarding china's corporate income tax, one is a direct incentive based on a 15% tax rate preference for high-tech enterprises, and the other is an indirect incentive represented by the tax base of r&d expenses plus deductions. by reducing the effective tax rate of enterprises through tax incentives, the government allows digital economy enterprises to have more funds to invest in innovation and r&d, thus promoting innovative transformation of enterprises. based on the results of previous studies by scholars on tax incentives and corporate innovation, we can know that tax incentives have a certain effect on corporate innovation. for example, chu deyin et al. (2017) conducted a research analysis on strategic emerging industries and concluded that tax incentives for corporate income tax would promote the r&d investment costs of the industry. therefore, this paper takes the digital economy industry as the research object and analyses the impact of corporate income tax incentives on the innovative development of digital economy enterprises. based on the above analysis, this paper proposes the following hypotheses. hypothesis 1: the innovative development of digital economy enterprises can be continuously improved by reducing the corporate income tax rate, considering other relevant variables. hypothesis 2: corporate income tax incentives have different effects on the innovation incentives of different digital economy enterprises. 4. research design 4.1. sample selection and variable description shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 139 4.1.1. sample selection this paper takes 142 digital economy enterprises in beijing-tianjin-hebei region as the research object, and the selected data range is the data range is from 2017 to 2021. in order to ensure the reliability of the research results, this paper refers to gao peiyong et al. (2013) and shen si et al. (2021), and excludes enterprises with missing or negative operating revenues, operating costs, total profits and r&d expenses, as well as enterprises exempted from corporate income tax due to the conversion of operating cultural institutions into enterprises. the final sample of 478 enterprises was obtained. to eliminate the effect of extreme outliers, all continuous variables are winsorized at 1% and 99%. 4.1.2. description of variables explanatory variable: corporate innovation investment (r&d). this indicator is replaced by r&d expenditure divided by operating revenue, following the example of shen (2021). core explanatory variable: tax-reduction. according to jia junxue et al. (2016), the since corporate income tax benefits are the most important part of tax benefits enjoyed by enterprises, this paper adopts the effective tax rate of corporate income tax as a proxy indicator of tax benefits. control variables: enterprise age, enterprise size, profitability and cost. the specific calculation methods and data sources are shown in table 1. table 1. selection of variables and description of indicators. variable type variable name variable symbols explanation of indicators data sources explained variables corporate innovation investment r&d r&d costs /operating income annual reports by company explanatory variables tax benefits tax-reduction in terms of the effective corporate income tax to express the tax rate control variables age of business age ln (current year company establishment year + 1) size of business size logarithm of operating income profitability profit profits / operating income operating cost ratio cost operating costs / operating income 4.2. econometric model this paper uses the least square (ols) method to test the impact of tax incentives for digital economy firms on firms' innovation investment. the specific model is constructed as follows: r&dit=β0+β1tax-reductionit+β2ageit+β3sizeit+β4profitit+β5cost+indk+yeart +μit (1) where the subscript i represents the firm and t represents the year; the firm's innovation input (r&d) is the explanatory variable and tax-reduction is the explanatory variable; the remaining shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 140 variables are control variables; and μ is a random disturbance term. in this paper, we choose a twoway fixed effects model controlling for industry and year. 4.3. descriptive statistics the table below shows the descriptive statistics for each variable. the standard deviation of r&d is 0.072, with the maximum and minimum values of 0.379 and 0.009 respectively, which indicates that there is some variation in the level of innovation investment among different enterprises. the standard deviation of tax-reduction is 0.020, with a maximum and minimum value of 0.15 and 0.1 respectively, indicating that there is some variation in the effective tax rate between firms. the average age of digital economy enterprises is only three years, which indicates that the development of digital economy enterprises is on the rise, and therefore the tax incentives can be used to stimulate the innovation ability of digital economy enterprises to a certain extent. there are also significant differences in size, profitability and cost across digital economy firms. table 2. descriptive statistics of the variables. variables sample size average value standard deviation median max min r&d 478 0.101 0.072 0.083 0.009 0.379 tax-reduction 478 0.140 0.020 0.15 0.1 0.15 age 478 3.025 0.229 3.091 2.485 3.434 size 478 21.109 1.271 20.989 18.324 25.264 profit 478 0.145 0.102 0.122 0.012 0.494 cost 478 0.594 0.187 0.607 0.069 0.937 note: continuous variables are winsorized at the 1% and 99% levels. 5. empirical analysis 5.1. full sample basis regression before conducting the regression, the model needs to be tested for multicollinearity. according to the results in table 3, the vif value of the core explanatory variable, i.e., tax incentives, is 1.09, while the values of the other control variables are all below 10, so it can be concluded that there is no multicollinearity in the regression model of this paper. table 3. multicollinearity test. variables vif 1/vif tax-reduction 1.09 0.916 age 1.11 0.902 size 1.23 0.810 profit 1.68 0.597 cost 1.80 0.555 then, a full-sample regression of the data was conducted to observe the impact of corporate income tax incentives on firms' innovation investment. table 4 shows the regression results. the first column shows the results without controlling for other influencing factors, while the second column shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 141 shows the results after controlling for related influencing factors. the coefficients of tax incentives are significantly negative in both columns. in column (1), the estimated coefficient of tax incentives is -0.798, which indicates that the reduction of corporate income tax rate of digital economy enterprises can promote the r&d investment of enterprises and thus improve the innovative development of enterprises, thus hypothesis 1 is confirmed. according to the results in column (2), the estimated coefficient of the tax benefit is -0.236, which indicates that the reduction of the corporate income tax rate of digital economy enterprises can promote the r&d investment of enterprises when other influencing factors are considered, which is basically consistent with the results without considering the influence of other control variables. table 4. corporate income tax incentives and corporate innovation investment. variables r&d （1） （2） tax-reduction -0.798*** （-4.78） -0.236** （-2.12） age - -0.051*** （-5.25） size - -0.003* （-1.66） profit - -0.234*** （-8.80） cost - -0.330*** （-21.67） industry effect control control _cons 0.203*** （8.41） 0.585*** （11.86） r2 0.0614 0.6032 n 478 478 note: *, **, *** denote significant at the 10%, 5% and 1% levels, respectively. the parentheses represent the tvalue, the same below. 5.2. sub-sample regressions 5.2.1. heterogeneity of property rights the property right attributes of enterprises may affect the effect of the effective tax rate of corporate income tax on the innovation investment of enterprises. therefore, this paper examines the heterogeneity of the research sample based on property rights. first, 142 digital economy enterprises are classified into three categories according to their property rights: private enterprises, state-owned enterprises, and other enterprises. the results in column (1) of table 5 show that there is a negative and significant relationship between corporate income tax incentives and innovation investment of digital economy enterprises, which indicates that reducing the effective corporate income tax rate can stimulate innovation investment of private enterprises, and hypothesis 2 is confirmed. after controlling for relevant shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 142 influencing factors, comparing (2), (4) and (6), for private firms, reducing the effective corporate income tax rate can significantly stimulate private firms' innovation investment through tax incentives, while the incentive effect for soes and other firms is not significant, which may be due to the low innovation efficiency of other firms and the small sample size of soes. table 5. tax incentives and firms' investment in innovation: private, state, and other. variables private state-owned other （1） （2） （3） （4） （5） （6） tax-reduction -1.108*** （-4.67） -0.415*** （-2.64） -2.059 （-0.36） -0.542 （-0.41） -0.409 （-1.63） 0.083 （0.50） age - -0.009 （-0.82） - -0.091** （-3.66） - -0.150*** （-7.58） size - -0.009*** （-3.05） - -0.045** （-4.33） - 0.007** （2.16） profit - -0.220** （-7.04） - -0.068 （-0.70） - -0.275*** （-5.25） cost - -0.310*** （-16.99） - 0.117 （1.04） - -0.379*** （-13.44） industry effect control control control control control control _cons 0.268*** （7.57） 0.600*** （9.10） 0.373 （0.43） 1.261** （5.88） 0.130*** （3.56） 0.664*** （7.81） r2 0.0777 0.6477 0.4847 0.9931 0.0985 0.6264 n 268 268 10 10 200 200 note: *, **, *** indicate significant at the 10%, 5%, and 1% levels, respectively. 5.2.2. regional heterogeneity since the economic development in beijing, tianjin and hebei are different, and the digital economy enterprises are unevenly distributed, and there is less data on digital economy enterprises in hebei, this paper divides the digital economy enterprises in beijing, tianjin and hebei into beijing, tianjin and hebei regions for regression analysis. according to the table below, comparing columns (1) and (3), the estimated coefficient of tax incentives for digital economy enterprises in beijing is significantly negative, while the estimated coefficient of tax incentives for digital economy enterprises in tianjin and hebei is not significant. it can be concluded that, without considering other influencing factors, the tax incentives of corporate income tax can significantly promote the innovation investment of digital economy enterprises in beijing, while there is no significant effect on the other two regions. then, comparing columns (2) and (4), the estimated coefficients of tax incentives for digital economy enterprises in beijing, tianjin and hebei are all significantly negative, controlling for the relevant influencing factors, indicating that under certain conditions, the tax incentives given by the government can effectively motivate shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 143 enterprises to increase the level of innovation investment and thus enhance the innovative development of digital economy enterprises. in addition, the absolute values of the estimated coefficients of tax incentives in tianjin and hebei are larger than those in beijing, which indicates that the incentive effect of corporate income tax incentives on enterprises' innovation investment is more significant for digital economy enterprises in tianjin and hebei. table 6. tax incentives and enterprise innovation investment: beijing, tianjin, and hebei regions. variables beijing tianjin and hebei （1） （2） （3） （4） tax-reduction -0.802*** （-4.83） -0.262** （-2.35） -87.604 （-1.46） -59.912* （-1.72） age - -0.049*** （-4.88） - -0.053** （-2.04） size - -0.003 （-1.47） - -0.012** （-2.10） profit - -0.246*** （-8.88） - -0.120 （-1.52） cost - -0.334 （-21.39） - -0.449*** （-6.41） industry effect control control control control _cons 0.204*** （8.34） 0.579*** （11.42） 13.225 （1.47） 9.843* （1.88） r2 0.0621 0.6142 0.0690 0.7266 n 439 439 39 39 note: *, **, *** indicate significant at the 10%, 5% and 1% levels, respectively. 5.3. robustness test 5.3.1. two-way causality test if the tax incentives of corporate income tax and enterprises' innovation investment are causal, i.e., while the tax incentives of corporate income tax affect the level of enterprises' innovation investment, enterprises may enjoy more tax incentives by increasing the level of innovation investment. then, the model is endogenous. in this regard, we lagged the core explanatory variable, i.e., corporate income tax incentives, by one period for robustness testing. the results are shown in the table below. this indicates that reducing the effective corporate income tax rate and increasing the tax incentives can promote the innovation investment of enterprises and enhance their innovation capability. this is consistent with the results of the full sample regression (table 4), indicating that the regression results are robust. shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 144 table 7. robustness test of tax incentives with one lag. variables r&d （1） （2） tax-reductiont-1 -0.788*** （0.71） -0.223* （2.20） age - -0.048*** （-4.17） size - -0.002 （-0.90） profit - -0.206*** （-6.11） cost - -0.334*** （-18.37） industry effect control control _cons 0.202*** （6.92） 0.555*** （9.58） r2 0.0641 0.6209 n 322 322 note: *, **, *** indicate significant at the 10%, 5% and 1% levels, respectively. 5.3.2. causality identification test of propensity score matching table 8. causal identification of corporate income tax incentives and firm innovation: estimation based on psm. variables r&d k=1 nuclear matching radius matching （1） （2） （3） att average processing effect （dummy variable：tax-reduction） 0.101* （0.015） 0.101** （0.010） 0.101** （0.010） control variables yes yes yes constant term 8.737*** （4.23） 8.737*** （4.23） 8.737*** （4.23） log-likelihood value -188.624 -188.624 -188.624 pseudo r 0.0310 0.0310 0.0310 observations 541 541 541 note: *, **, *** indicate significant at the 10%, 5% and 1% levels, respectively. in this paper, the propensity score matching method (psm) is used to investigate the causal relationship between corporate income tax incentives and corporate innovation investment by using shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 145 enterprises that enjoy corporate income tax incentives as the experimental group and enterprises that do not enjoy corporate income tax incentives as the control group. in this paper, the sample data is expanded to 541 items, and the digital economy enterprises that do not enjoy the eitc are set as the control group 0 and the digital economy enterprises that enjoy the eitc are set as the treatment group 1. in this paper, we use "one-to-one with put-back nearest neighbor matching (k=1)" for empirical analysis, followed by "kernel matching" and "radius matching" for robustness analysis of the results of the above method. the results of the above methods are then analyzed using "kernel matching" and "radius matching". as shown in the following table, columns (1), (2) and (3) are significantly positive, indicating that corporate income tax incentives promote innovation investment, while the opposite is not true. this is consistent with the findings obtained from the above regressions and proves that the results are robust. 6. conclusions and implications 6.1. research conclusions the development of digital economy plays an increasingly important role in the development process of the world today, and the future is still oriented towards the development of digital economy. as an important force to promote the modernization of china, digital economy occupies an important position in the development process of china. however, how we should promote the innovative development of digital economy enterprises and enhance our economic development through digital economy is still a problem to be solved today. based on the data of digital economy enterprises in beijing-tianjin-hebei region from 2017 to 2021 (in line with the narrow caliber of the national bureau of statistics), this paper examines the effect of tax incentives for 142 digital economy enterprises. finally, the following conclusions are drawn: firstly, with the control of other influencing factors, by reducing the tax rate, the enterprises are motivated to increase their innovation investment and enhance their innovation capability; secondly, through the heterogeneity test, it is concluded that increasing the tax incentives for enterprises has a more significant incentive effect for private enterprises, which makes the innovation investment of private enterprises significantly improved; thirdly, for the beijing-tianjin-hebei region. the thirdly, there is a significant difference between the digital economy enterprises in beijing, tianjin and hebei, and the tax incentives for digital economy enterprises in beijing can significantly promote the innovative development of digital economy enterprises in the region, while the impact of tianjin and hebei is relatively small. 6.2. research implications first, tax incentives should be increased. tax incentives can, to a certain extent, promote the r&d investment of digital enterprises and stimulate their ability to innovate and develop. the innovation ability of enterprises needs to be supported by the government, and tax incentives can better promote enterprises to invest in r&d, to improve their innovative ability and thus enhance their development competitiveness. second, different policies should be formulated for enterprises with different property rights and different levels of r&d investment. while increasing the tax incentives for private enterprises, it is also necessary to pay attention to the degree of tax incentives for state-owned enterprises and other shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 146 enterprises. for enterprises in beijing, the government should continuously strengthen tax incentives to stimulate the development of regional digital economy enterprises and improve their innovation capacity. finally, it is important to stimulate the flourishing of more large-format enterprises through tax incentives. since there is a large gap between the development of digital economy enterprises in tianjin, hebei and beijing, the beijing-tianjin-hebei region should pay more attention to the coordinated development of the region while promoting the development of its own digital economy enterprises, closely following the integrated development of beijing-tianjin-hebei digital economy, giving play to the demonstration and pulling role of the beijing region, thus driving the quantity and quality of digital economy enterprises in tianjin and hebei, and continuously injecting vitality into the market. funding: this research received no external funding. conflicts of interest: the author declares no conflict of interest. references [1] song q, liu y h. the degree of market competition, r&d investment and innovative output of small and medium-sized technology enterprises: a conditional process analysis based on the regulation of venture capital[j]. china soft science, 2021(10):182-192. [2] song qing, yang xue. tax incentives, business environment and innovative performance of enterprises[j]. china science and technology forum, 2021(5):99-107. doi: https://doi.org/10.13580/j.cnki.fstc.2021.05.011. [3] chen d, xing l. tax incentives and corporate r&d investment: a perspective of internal control[j]. modern economic discussion, 2020(12):80-90. doi: https://doi.org/10.13891/j.cnki.mer.2020.12.012. [4] jia junxue, ying shiwei. fiscal decentralization and corporate tax incentives: an analysis based on the perspective of local government competition[j]. china industry industrial economics, 2016(10):23-39. doi: https://doi.org/10.19581/j.cnki.ciejournal.2016.10.003. [5] shen s, liu w l. analysis of tax and fee reduction effect of digital economy enterprises--shenzhen as an example[j]. tax economic research, 2021,26(03):33-43. doi: https://doi.org/10.16340/j.cnki.ssjjyj.2021.03.005. [6] gao peiyong, mao jie. the size, structure and effectiveness of indirect tax incentives: empirical evidence from a national tax survey[j]. china industrial economy, 2013(12):143-155. doi: https://doi.org/10.19581/j.cnki.ciejournal.2013.12.014. [7] liang fushan. a study on the heterogeneous effect of tax incentives on r&d investment of enterprises[j]. tax research, 2021(3):134-143. doi: https://doi.org/10.19376/j.cnki.cn11-1011/f.2021.03.021. [8] chen d, xing l. tax incentives and corporate r&d investment: a perspective of internal control[j]. modern economic discussion, 2020(12):80-90. doi: https://doi.org/10.13891/j.cnki.mer.2020.12.012. [9] wang wanjun, shen kunrong, zhou shaodong, qin yong. on-the-job training, r&d investment and corporate innovation[j]. economic and management research, 2015(12):123-130. doi: https://doi.org/10.13502/j.cnki.issn1000-7636.2015.12.017. [10] fan y, li h n. tax administration, tax compliance and tax preference: an assessment of the policy effect of golden tax phase iii project[j]. finance and trade economy, 2020(5):51-66. doi: https://doi.org/10.19795/j.cnki.cn11-1166/f.20200512.004. [11] ye xu, du yunhan, he wenjun. structural effects of employment in digital economy development[j]. finance and trade research, 2021(4):1-13. doi: https://doi.org/10.19337/j.cnki.34-1093/f.2021.04.001. [12] ma hongfan, xu ling, liu guoping. digital economy, tax impact and tax governance change[j]. taxation research, 2021(4):84-91. doi: https://doi.org/10.19376/j.cnki.cn11-1011/f.2021.04.014. [13] li hui, zhang cheng. realistic dilemma of taxation in digital economy, international comparison and policy suggestions[j]. economic system reform reform, 2021(2):167-172. [14] yu changlin, yang guoge, du mingyue. industrial policy and technological innovation in china's digital economy[j]. statistical research, 2021(1):51-64. doi: https://doi.org/10.19343/j.cnki.11-1302/c.2021.01.005. shu-xian wang / journal of risk analysis and crisis response, 2023, 13(2), 136-147 doi: https://doi.org/10.54560/jracr.v13i2.361 147 [15] sun zheng, liang zhanshuo. research on china's digital economy tax policy in the context of "two transformations"[j]. reform and strategy, 2022,38(01):107-116. doi: https://doi.org/10.16331/j.cnki.issn1002736x.2022.01.011. [16] preacher k j, hayes a f. spss and sas procedures for estimating indirect effects in simple mediation models[j]. behavior research methods, 2004,36(4):717-731. [17] valeria bucci. the effect of fiscal incentives on business r&d[j]. international journal of economics finance and management sciences, 2019(3):88-91. [18] zhang h m, li t, chen y z. dynamic comprehensive evaluation of ecological environment of 12 provinces and cities in western china[j]. journal of risk analysis and crisis response, 2019,9(3):156-161. doi: https://doi.org/10.2991/jracr.k.191024.005. [19] zhang m, luo q. a systematic literature review on the influence mechanism of digital finance on high quality economic development [j]. journal of risk analysis and crisis response, 2022,12(1):45-54. doi: https://doi.org/10.54560/jracr.v12i1.321. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). research article multi-hazard awareness, risk perception and fear to earthquakes: the case of high-school students in mexico city jaime santos-reyes*, grupo de investigación: “saracs”, sepi-esime, zac., instituto politécnico nacional, cdmx 07738, mexico 1. introduction natural hazards in combination with a community that lacked an adequate preparation for such events, have caused a considerable damage to people, property, and economical losses [1,2]. for example, those communities in seismic prone regions, worldwide, are vulnerable to earthquakes given the fact that they lack earthquake resistant homes, earthquake early warning systems, among others [1,2] (i.e., earthquakes still cannot be predicted [3,4]). in addition to the individual natural hazards (e.g. earthquakes), there is an increasing evidence that hazards are found in combination with other threats, i.e., natural, environmental, psychosocial, technological, sociological, etc. [5–13]. the interconnectedness of these multi-hazards has been recognised in the sendai framework for disaster risk reduction in 2015 and the sustainable development goals [11]. these organizations, among others, have encouraged communities to develop an adaptation strategy to deal with such threats [11,14]. in order to be able to develop such strategies, it becomes necessary to be aware of such multi-hazards and the risk associated with them. within this context, risk perception plays a fundamental role in adopting some form of protective action to mitigate the impact of these hazards [11,15]. however, it has been found that there is not a proportional relationship between the level of risk perception and the rate of taking protective actions, i.e., higher level of risk perception is not always linked to protective action (“risk perception paradox”) [11,15]. risk perception has also been envisaged as having cognitive and emotional components [16]; for example, the six ‘basic’ emotions proposed by ekman [17]: i.e., fear, anger, happiness, sadness, disgust, surprise. several studies have been conducted and reported in the literature on emotional and behavioural reactions to earthquakes [18–24]. in the present case study, we are dealing with fear to earthquakes; according to the oxford dictionary, fear is defined as “the bad feeling that you have when you are in danger, when something bad might happen, or when a particular thing frightens you” [25]. in this context, fear to earthquakes may be defined as a feeling that something bad might happen to those experiencing one. in relation to earthquakes, previous research has shown that fear has been the most frequent emotion experienced by individuals during earthquake occurrence [18–20,23]. however, results have also shown that other emotions are experienced by earthquake survivors, e.g., in a study conducted on the umbria-marche earthquake in italy, 13.1% of the respondents experienced ‘surprise’, 12.1% ‘sadness’, etc. [23]. hence, it may be argued that during an earthquake occurrence, people experience several types of emotions. research has also been conducted on issues related to hazard awareness, preparedness, vulnerability and resilience, among others, and widely reported in the literature [e.g. 26–32]. for example, in a multi-hazard awareness study conducted in morocco; it has been found that floods and earthquakes represent the top two threats to local communities [26]. the paper presents the results of a cross-sectional study on the hazards awareness and risk perception for the case of high school students in mexico city. the paper gives an account of the key findings of the research project. a r t i c l e i n f o article history received 18 december 2019 accepted 16 september 2020 keywords earthquake multi-hazards awareness risk perception negative emotions mexico city a b s t r a c t a cross-sectional study was conducted in 2015 for the case of students at a high school in mexico city; the sample size was n = 302. the aim of the study was to assess the level of hazard awareness, risk perception, and the negative emotion related to fear during earthquakes. some of the main findings were the following: (a) regarding the multi-hazards (natural, environmental, psychosocial, technological, sociological) considered in the study, earthquakes and ‘crime & delinquency’ were perceived as the top two threats to the participants of the study, and in that order; (b) the perceived hazards that may cause physical harm were earthquakes and ‘crime & delinquency’ and in that order; (c) women fear the most during an earthquake [χ2 (2, n = 301) = 17.614, p < 0.001]. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *email: jrsantosr@hotmail.com journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 91–96 doi: https://doi.org/10.2991/jracr.k.200923.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-3758-9862 http://creativecommons.org/licenses/by-nc/4.0/ mailto:jrsantosr%40hotmail.com?subject= https://doi.org/10.2991/jracr.k.200923.001 https://www.atlantis-press.com/journals/jracr 92 j. santos-reyes / journal of risk analysis and crisis response 10(3) 91–96 2. materials and methods a cross-sectional survey has been conducted to several preparatory schools in mexico city in 2015; the schools were in different parts within the city (i.e., in the three geotechnical zones). however, only the collected data of one of these schools are presented here, and with a sample size, n = 302. the survey questionnaire included over fifty questions, however, only a few have been considered herein, and some of them are based on ronan et al. [32]. the results presented here include those associated with hazard awareness, risk perception, and some psychological issues in relation to negative emotions (i.e., fear) during earthquake occurrence. in order to analyse the descriptive information, frequency analysis has been conducted. further, the validity of the working hypothesis, on the other hand, was considered at a significance level of α = 0.05. furthermore, the relationship between the independent and dependent variables have been assessed by applying the contingency tables and pearson’s chi-square tests. table 1 shows the demographic characteristics of the sample considered in the present analysis. it can be seen that the range of the age of the respondents varied from 14 to 19 years old (m = 16.39; sd = 1.203). the percentage of men was higher than women (i.e., 55.0% vs. 45.0%). finally, in the capital city, the schooling system covers two shifts: i.e., the morning and afternoon shifts. in general, the morning shift covers from 07:00 to 14:00 h (55.3%); the afternoon shift, on the other hand, covers from 16:00 to 22:00 p.m. (44.7%). 3. results and discussion 3.1. multi-hazard awareness to assess the multi-hazard awareness of respondents of the present study, the following question was included in the questionnaire: “which are the two most likely hazards that could affect you at school and home?”. the types of hazards considered as possible response to the question were the following: natural hazards (“earthquakes”, “floods”, “hurricanes”, “landslides”, “volcanic eruptions”), environmental (“environmental pollution”), psychosocial (“bullying”), and sociological (“crime & delinquency”). it should be highlighted that all these hazards are present in the capital city [33] (see section 3.5 for further details). the results are shown in table 2. when considering natural hazards, the perceived two hazards that may affect the participants at home were earthquakes (85.5%) and floods (16%). however, when considering all types of hazards considered in the analysis, the results show that the two perceived hazards that could affect the participants were earthquake (85.5%) and ‘crime & delinquency’ (64.1%). in general, respondents perceived that earthquakes represents the major threat to them when at home. but, what about the two perceived hazards at school? again, the results show that earthquakes are the cause of concern by the respondents, with 81.6%; this was followed by ‘crime & delinquency’ and ‘bullying’, both with 47.3%, when considering all types of hazards (table 2). in summary, the results highlighted that the following types of hazards were those that participants most worry about both at home and at school: natural (earthquake), environmental (environmental pollution), psychosocial (bullying), and sociological (‘crime & delinquency’). 3.2. multi-hazard risk perception in an attempt to assess the perceived vulnerability and risk, the following question included in the questionnaire: “should one of these events occur in the future, how likely is it that it could hurt you or your family?” it should be emphasised that within the response options, a technological hazard was included this time (i.e., ‘fire & explosion’). the results are shown in table 3. the possible responses to the question were the following: “1 = unlikely”, “2 = a chance”, and “3 = likely”. the top four hazards that have been perceived as having a “better than a chance” to cause physical harm to the participants of the table 1 | demographic characteristics of the participants of the study, n = 302 variables n (%) age (years) 14 12 (4.0) 15 69 (22.8) 16 75 (24.8) 17 95 (31.5) 18 38 (12.6) 19 13 (4.3) gender women 136 (45.0) men 166 (55.0) schooling mode system morning 167 (55.3) afternoon 135 (44.7) table 2 | perceived multi-hazards at home and school hazards at home hazards at school hazard % hazard % floods 16 floods 8.2 hurricanes 2.3 hurricanes 2.0 earthquake 85.5 earthquake 81.6 landslide 6.6 landslide 1.6 volcanic eruption 3.1 volcanic eruption 1.6 environmental pollution 20.7 environmental pollution 10.6 crime & delinquency 64.1 crime & delinquency 47.3 bullying 1.6 bullying 47.3 table 3 | perceived event occurrence and physical risk perception hazard % “likely” mean sd floods 24.1 2.04 0.664 earthquake 87.6 2.85 0.427 landslide 10.7 1.66 0.663 volcanic eruption 14.0 1.75 0.687 environmental pollution 69.2 2.66 0.534 crime & delinquency 81.9 2.80 0.458 fire & explosion 65.6 2.60 0.602 j. santos-reyes / journal of risk analysis and crisis response 10(3) 91–96 93 study were: earthquakes (sd = 2.85), ‘crime & delinquency’ (sd = 2.80), environmental pollution (sd = 2.66), and ‘fire & explosion’ (sd = 2.60) (see section 3.5 for a discussion on these findings). on the other hand, the bottom three were floods, volcanic eruption, and landslide. 3.3. experiencing earthquakes the questionnaire included a question intended to elucidate whether the participants have experienced earthquakes. in particular, respondents were asked the following question: “have you experienced an earthquake?” the possible answers were either “yes” or “no”. the results showed that most of the participants have experienced an earthquake (98.7%; 298/302). this may be explained given the fact that in 2014, an unusual number of earthquakes occurred in that particular year (it should be highlighted that the study was conducted in 2015) [33]. for example, there were three relatively strong earthquakes with magnitudes ranging from m5.4, m5.9, and m7.9 [33]. these events caused panic and anxiety among the residents of the city; however, there were neither property damages, nor life threatening situation. 3.4. level of fear during earthquakes one of the factors related to psychological issues that has been addressed in the study is that related to the level of fear during earthquakes. in the context of the previous question (section 3.3), students were asked the following question: “what was your level of fear during that earthquake?” respondents were asked to rate their answers according to the following options: 1 = “not at all”, 2 = “little”, 3 = “a lot” (that is, ‘fear’ was measured with three levels; however, it should be highlighted that in similar studies, the intensity of fear was measured as a continuous variable, for example in prati et al. [24], respondents were asked to rate their perceived intensity of fear on a scale ranging from “0 to 100” [p. 104].). overall, the frequency data show that most of the participants experienced “little” fear during the earthquake (59.8%; 180/302) (figure 1). further, almost 40% of the participants of the study reported no fear at all (20.6%; 62/302) and 19.6% reported “a lot” of fear (59/302). in order to assess the relationship of the variables considered in the study (gender, age, and the schooling mode system, see table 1), cross-tabulations were constructed, and the results are shown in tables 4–7 (the percentages given in these tables are within the variables shown in columns; the total percentages in columns may not add up to 100% because of decimal rounding. further, differences in total n = 302 are due to missing values in items.). when conducting chi-square tests, the results showed that neither the variable age [χ2 (2, n = 301) = 0.969, p = 0.616, cramer’s v = 0.057], nor the schooling mode system [χ2 (2, n = 301) = 2.605, p = 0.272, cramer’s v = 0.093] were associated with the level of fear during earthquakes. figure 1 | responses to the question related to fear during an earthquake occurrence. table 4 | results of the relationship between gender and the level of fear to earthquakes variables male female total n (%) n (%) n (%) “not at all” 41 (30.1) 21 (12.7) 62 (20.6) “little” 78 (57.4) 102 (61.8) 180 (59.8) “a lot” 17 (12.5) 42 (25.5) 59 (19.6) table 5 | results of the relationship between the age category and the level of fear variables ≤16 ≥17 total n (%) n (%) n (%) “not at all” 33 (21.3) 29 (19.9) 62 (20.6) “little” 95 (61.3) 85 (58.2) 180 (59.8) “a lot” 27 (17.4) 32 (21.9) 59 (19.6) table 6 | results of the relationship between the schooling mode system and the level of fear variables morning shift afternoon shift total n (%) n (%) n (%) “not at all” 30 (18.1) 32 (23.7) 62 (20.6) “little” 106 (63.9) 74 (54.8) 180 (59.8) “a lot” 30 (18.1) 29 (21.5) 59 (19.6) table 7 | results of the relationship between the gender and the level of fear of the participants of the study when controlling for the age group category variables male female n (%) n (%) ≤16 “not at all” 20 (31.3) 13 (14.3) “little” 37 (57.8) 58 (63.7) “a lot” 7 (10.9) 20 (22.0) ≥17 “not at all” 21 (29.2) 8 (10.8) “little” 41 (56.9) 44 (59.5) “a lot” 10 (13.9) 22 (29.7) 94 j. santos-reyes / journal of risk analysis and crisis response 10(3) 91–96 however, it was found that the variable related to gender was associated with the level of fear of earthquakes of the participants of the study, χ2 (2, n = 301) = 17.614, p < 0.001, cramer’s v = 0.242. in particular, it was found that women feared “a lot” during the experienced earthquake when compared with men (i.e., 25.4% vs. 12.5%). to further investigate the influence of the variable related to the category of age, a multivariable analysis (i.e., by ‘controlling’ for the age categories) was conducted, and the results are shown in table 7. it has been found that only those participants (men and women) with the age category ≥17 years old were associated with “little” fear during the earthquake, χ2 (2, n = 146) = 10.408, p = 0.005, cramer’s v = 0.267. in particular, the results highlighted that men experienced “little” fear when compared to women (i.e., 29.2% vs. 10.8%). 3.5. discussion regarding the multi-hazard awareness and risk perception, the results showed that respondents’ perceptions may be regarded as consistent with what has been happening in the city; i.e., prior to 2015 when the study was conducted (section 2) (it may be argued that these results are relevant even today). that is, the residents of the capital city have been exposed to these multi-hazards [33–43]. for example, earthquakes and ‘crime & delinquency’ were the top two hazards that were perceived as threats at home and school. these were followed by environmental and technological hazards. each of these will be briefly discussed. effectively, mexico city (and those communities living along the pacific coast of the country) is prone to earthquakes [19,33]. for example, one of the strongest earthquakes that hit the capital city was in 1985 (m8.1); it is believed that thousands of people were killed [19,33]. that is, earthquakes occur all the time. this finding has demonstrated that students were aware of such events at the time of the study. for example, the two most recent strong earthquakes that occurred on 07 (m8.2) and 19 (m7.1) september 2017. it is thought the 19 september earthquake affected more than 28 million people, 54,000 schools, 5700 hospitals, etc. [44]. therefore, the findings are consistent with the perceived earthquake threats of the participants of the study. ‘crime & delinquency’, on the other hand, was the second hazard that was perceived as a threat to the participants of the study at home and school. there has not been (and there is not) a day without news on the mass media related to crime in the capital city; several studies on this have been reported in the literature [33,35,38–40]. it is believed that during 1990–2005, the pattern of violent crime rate decreased during this period [39]. however, since december 2006 when the federal government launched what is now called the “drug war” [i.e., a military offensive against the so called ‘drug traffic organizations’ (dtos)], there has been a surge in violent crime rates [39]. enamorado et al. [39] reported that in 2005, “the total rate of homicides was close to 11 per 100,000 individuals”; further, the authors reported that “by 2010, it was 18.5 per 100,000 individuals” (p. 129). furthermore, it is believed that in 2005, the number of people being killed by non-drug related crime was 7000 [39]. according to the 2015 child and youth survey [40], 75.4% of the children (6–9 years old), 59.7% (10–13), and 70.3% (14–17) felt unsafe on the streets. further, the study concluded that most mexicans felt unsafe on the streets and this is more accentuated in children and young boys. hence, regarding ‘crime & delinquency’, the results of the study are consistent with what has been happening and reported in the literature. the results also showed that one of the worries of the students is that related to the ‘environmental pollution’ in the capital city. several studies have given an account on this and reported in the literature [9,34,36,41,43]. further, on 20 november 1989, an ‘environmental contingency’ plan was implemented in the city. an environmental contingency has been defined as “an eventual and transitory situation declared by the authorities, when a high concentration of o3 and/or pm10 or pm2.5 pollutants occurs or is anticipated, based on objective analyses, forecasts or monitoring of environmental air pollution, derived from human activities or natural phenomena that affect the health of the population or the environment” [43]. for example, one of the measures taken has been that motor cars are not allowed to transit in the city during certain days [41,43] (other measures such as children staying indoors, etc., can be found in gobierno [43]). again, the results are consistent with environmental hazards in the city. finally, the fourth technological hazard that students were worried about is that related to ‘fire & explosion’. this is a typical case of what is called ‘low frequency-high impact’ events (it may be argued that earthquakes follow this pattern too). there have been several fire and explosion events in the metropolitan area of the capital city. for example, four of the major events occurred in the period between 2000 and 2008; other examples are given in santos-reyes and gouzeva [33] and lozanao et al. [42]. in 2000, a fire and explosion occurred involving ten trucks; in 2008, there was a spill of a hazardous material (i.e. kerosene) that forced the evacuation of 150 families in the metropolitan area of the city. again, the participants of the study showed their level of awareness of the existing technological hazards, such as ‘fire & explosion’. it is worth mentioning that research on the psychological issues related to emotions started since the 70s. for example, ekman [17] proposed the ‘basic’ emotions, namely: fear, happiness, surprise, anger, sadness, and disgust. some authors have used these ‘basic’ emotions to understand the emotional response of people during an earthquake occurrence [19,23,24]. however, in our study we did not used these categories of emotions, but the perceived level of fear during an earthquake occurrence was assessed with the following responses: “not at all”, “little”, and “a lot” (section 3.4). according to the frequency data in our case study, 19.6% of the respondents reported “a lot” of fear during an earthquake (59.8% responded “little” and 20.6% “not at all”). similar findings have been reported in a study that was conducted in new zealand. that is, about 14% of children reported often feeling ‘upset or scare’ when thinking or talking about earthquakes; 25% reported no upset, and 60% reported some level of fear [32]. when conducting chi-square tests, it has been found that gender was significantly associated with fear. that is, women fear “a lot” of j. santos-reyes / journal of risk analysis and crisis response 10(3) 91–96 95 earthquakes when compared to men. when considering the age of the respondents, on the other hand, the age category ≥17 years old was associated with “little” fear of earthquakes. the limitations of the study were the following: first, the design of the study was non probability, which means that the findings should not be generalised to the whole population of high school students of the capital city; second, the sample size of the present study was relatively small, i.e., n = 302; future research may involve the analysis of the collected data on several high schools in the city. third, the present study was conducted with no earthquake disaster as a precedent; that is, prior to 2015 (when the study was conducted) several minor earthquakes occurred but not damage was reported [33]. further research may include the analysis by considering the geotechnical seismic zones of the capital city where the school facilities are located. 4. conclusion the paper has presented the results of a cross-sectional study on the participants’ level of multi-hazard awareness, risk perception and emotional reaction related to fear during earthquakes. the study was conducted in 2015; the sample size was n = 302, and for the case of a high school located in mexico city. the key findings/ conclusions are summarised as follows: (a) the perceived two hazards that may affect students at home and school were earthquakes and ‘crime & delinquency’ and in that order. (b) the perceived two hazards that may cause physical harm (table 3) to the participants and family were earthquakes and ‘crime & delinquency’ and in that order. (c) most of the participants of the study have experienced an earthquake 97.8% (298/302). (d) regarding the negative emotion related to fear during earthquakes, 19.6% experienced “a lot” of fear, 59.8% “little” and 20.6% feared “not at all” of earthquakes. (e) the chi-squared results showed that women experienced “a lot” of fear during earthquakes when compared to men. (f ) when considering the multi-hazards (natural, environmental, psychosocial, technological, sociological) considered in the present study, earthquakes were perceived as the top threat to the students considered in the analysis. (g) more generally, these findings may assist key decision makers on disaster reduction in the capital city to device programmes aiming at better educating children to multi hazards. by doing this, children may be better prepared to mitigate the impact of earthquakes. conflicts of interest the author declares no conflicts of interest. acknowledgments this project was funded by sip-ipn under the following grants: conacyt no.248219 & sip-ipn: no. 20201790. references [1] united nations office for disaster risk reduction (unisdr). making development sustainable: the future of disaster risk management. global assessment report on disaster risk reduction. geneva, switzerland: unisdr; 2015. [2] international federation of the red cross and red crescent societies (ifrc). world disasters report. geneva: ifrc; 2014. [3] kagan yy. assessment of schemes for earthquake prediction: are earthquakes predictable? geophys j int 1997;13:505–25. [4] knopoff l. earthquake prediction: the scientific challenge. proc natl acad sci u s a 1996;93:3719–20. [5] cutter sl, mitchel jt, scott ms. revealing the vulnerability of people and places: a case study of georgetown county, south carolina. ann assoc am geograph 2000;90:713–37. [6] o’brian k, pelling m, patwardhan a, hallegatte s, maskrey a, oki t, et al. toward a sustainable and resilient future. in: field cb, barros v, stocker tf, quin d, dokken dj, ebi kl, et al., editors. managing the risks of extreme events and disasters to advance climate change adaptation. a special report of working groups i and ii of the intergovernmental panel on climate change (ipcc). cambridge: cambridge university press; 2012, pp. 437–86. [7] health and safety professionals alliance (haspa). psychological hazards: bullying, aggression, and violence. the core body of knowledge for generalist ohs professionals. tullamarine, vic: safety institute of australia; 2012. [8] meier rf, short jr jf. crime as hazard: perceptions of risk and seriousness. criminology 1985;23:389–400. [9] kar j, vaughan ma, liu z, omar ah, trepte cr, tackett j, et al. detection of pollution outflow from mexico city using calipso lidar measurements. remote sens environ 2015;169:205–11. [10] sanchez ey, represa s, mellado d, balbi kb, acquesta ad, colman lerner je, et al. risk analysis of technological hazards: simulation of scenarios and application of a local vulnerability index. j hazard mater 2018;352:101–10. [11] sullivan-wiley ka, short gianotti ag. risk perception in a multi-hazard environment. world dev 2017;97:138–52. [12] granados hd, jenkins s. chapter 13 extreme volcanic risks 1: mexico city. volcanic hazards risks disasters 2015:315–54. [13] zobin vm, martínez a. quantification of the 1998–1999 explosion sequence at popocatepetl volcano, mexico. j volcano geother res 2010;194:165–73. [14] shaw r, izumi t. civil society organization and disaster risk reduction: the asian dilemma. japan: springer science business; 2014. [15] lindell mk, perry rw. the protective action decision model: theoretical modifications and additional evidence. risk anal 2012;32:616–32. [16] loewenstein gf, weber eu, hsee ck, welch n. risk as feelings. psychol bull 2001;127:267–86. [17] ekman p. an argument for basic emotions. cogn emotion 1992;6:169–200. https://doi.org/10.1073/pnas.93.9.3719 https://doi.org/10.1073/pnas.93.9.3719 https://doi.org/10.1111/0004-5608.00219 https://doi.org/10.1111/0004-5608.00219 https://doi.org/10.1111/0004-5608.00219 https://doi.org/10.1111/j.1745-9125.1985.tb00346.x https://doi.org/10.1111/j.1745-9125.1985.tb00346.x https://doi.org/10.1016/j.rse.2015.08.009 https://doi.org/10.1016/j.rse.2015.08.009 https://doi.org/10.1016/j.rse.2015.08.009 https://doi.org/10.1016/j.jhazmat.2018.03.034 https://doi.org/10.1016/j.jhazmat.2018.03.034 https://doi.org/10.1016/j.jhazmat.2018.03.034 https://doi.org/10.1016/j.jhazmat.2018.03.034 https://doi.org/10.1016/j.worlddev.2017.04.002 https://doi.org/10.1016/j.worlddev.2017.04.002 https://doi.org/10.1016/b978-0-12-396453-3.00013-7 https://doi.org/10.1016/b978-0-12-396453-3.00013-7 https://doi.org/10.1016/j.jvolgeores.2010.05.008 https://doi.org/10.1016/j.jvolgeores.2010.05.008 https://doi.org/10.1016/j.jvolgeores.2010.05.008 https://doi.org/10.1111/j.1539-6924.2011.01647.x https://doi.org/10.1111/j.1539-6924.2011.01647.x https://doi.org/10.1111/j.1539-6924.2011.01647.x https://doi.org/10.1037/0033-2909.127.2.267 https://doi.org/10.1037/0033-2909.127.2.267 96 j. santos-reyes / journal of risk analysis and crisis response 10(3) 91–96 [18] alexander d. behaviour during earthquakes: a southern italian example. int j mass emerg disasters 1990;8:5–29. [19] santos-reyes j, gouzeva t. mexico city’s residents emotional and behavioural reactions to the 19 september 2017 earthquake. environ res 2020;186:109482. [20] bourque lb, russell la, goltz jd. human behaviour during and immediately after the earthquake. in: bolton pa, editor. the loma priets, california earthquake of october 17, 1989: public response. professional paper 1553-b. washington, dc: united states geological survey; 1993, pp. b3–b22. [21] goltz jd, bourque lb. earthquakes and human behaviour: a sociological perspective. int j disaster risk reduct 2017;21:251–65. [22] lindell mk, prater cs, wu hc, huang sk, johnston dm, becker js, et al. immediate behavioural responses to earthquakes in christchurch, new zealand, and hitachi, japan. disasters 2016;40:85–111. [23] prati g, catufi v, pietrantoni l. emotional and behavioural reactions to tremors of the umbria-marche earthquake. disasters 2012;36:439–51. [24] prati g, saccinto e, pietrantoni l, pérez-pastor c. the 2012 northern italy earthquakes: modelling human behaviour. nat hazards 2013;69:99–113. [25] oxford dictionary. definition of fear. 2020. available from: https://www.oxfordlearnersdictionaries.com/definition/american_ english/fear_1#:~:text=noun-,noun,child%20was%20shaking%20 with%20fear. [26] ivčević a, bertoldo r, mazurek h, siame l, guignard s, moussa ab, et al. local risk awareness and precautionary behaviour in a multi-hazard region of north morocco. int j disaster risk reduct 2020;50:101724. [27] granger k, jones t, leiba m, scott g. community risk in cairns: a provisional multi hazard risk assessment (agso cities project report no. 1). canberra, australia: australian geological survey organization; 1999. [28] beck e, andré-poyaud i, davoine pa, chardonnel s, lutoff c. risk perception and social vulnerability to earthquakes in grenoble (french alps). j risk res 2012;15:1245–60. [29] shaw r, shiwaku k, kobayashi h, kobayashi m. linking experience, education, perception and earthquake preparedness. disaster prev manage 2004;13:39–49. [30] marincioni f, appiotti f, ferretti m, antinori c, melonaro p, pusceddu a, et al. perception and communication of seismic risk: the 6 april 2009 l’aquila earthquake case study. earthquake spectra 2012;28:159–83. [31] rao ll, hana r, ren xp, baia xw, zhenga r, liu h, et al. disadvantage and prosocial behavior: the effects of the wenchuan earthquake. evol human behav 2011;32:63–9. [32] ronan kr, johnston dm, daly m, fairley r. school children’s risk perceptions and preparedness: a hazards education survey. austral j disaster trauma stud 2001;1:32. [33] santos-reyes j, gouzeva t, santos-reyes g. earthquake risk perception and mexico city’s public safety. proc eng 2014;84:662–71. [34] son y, osornio-vargas ar, o’neill ms, hystad p, texcalacsangrador jl, ohman-strickland p, et al. land use regression models to assess air pollution exposure in mexico city using finer spatial and temporal input parameters. sci total environ 2018;639:40–8. [35] murphy te, rossi ma. following the poppy trail: origins and consequences of mexican drug cartels. j dev econ 2020;143:102433. [36] silva-quiroz r, rivera al, ordoñez p, gay-garcia c, frank a. atmospheric blockages as trigger of environmental contingencies in mexico city. heliyon 2019;5:e02099. [37] urquieta e, varon j. mexico city’s petroleos mexicanos explosion: disaster management and air medical transport. air med j 2014;33:309–13. [38] baysan c, burke m, gonzález f, hsiang s, miguel e. non-economic factors in violence: evidence from organized crime, suicides and climate in mexico. j econ behav organ 2019;168:434–52. [39] enamorado t, lópez-calva l, rodríguez-castelán c, winkler h. income inequality and violent crime: evidence from mexico’s drug war. j dev econ 2016;120:128–43. [40] instituto nacional de estadística, geografía e informatica (inegi). encuesta nacional de victimización y percepción sobre seguridad publica (envipe), mexico: inegi; 2015. [41] davies lw. the effect of driving restrictions on air quality in mexico city. j polit econ 2008;116:38–81. [42] lozano a, muñoz a, antún jp, granados f, guarneros l. analysis of hazmat transportation accidents in congested urban areas, based on actual accidents in mexico. proc soc behav sci 2010;2:6053–64. [43] gobierno cdmx. announcement of the program to prevent and respond to atmospheric environmental contingencies in mexico city. gaceta cdmx; 2019 (in spanish). available from: http:// www.aire.cdmx.gob.mx/descargas/ultima-hora/calidad-aire/ pcaa/gaceta_oficial_cdmx.pdf. [44] geotechnical extreme events reconnaissance association (geer). geotechnical engineering reconnaissance of the 19 september 2017 mw 7.1 puebla-mexico city earthquake, version 2.0. report no. geer-055a. 2017. available from: http://geerassociation.org/ administrator/components/com_geer_reports/geerfiles/unam_ geer_sept2017_v2_final.pdf. https://doi.org/10.1016/j.envres.2020.109482 https://doi.org/10.1016/j.envres.2020.109482 https://doi.org/10.1016/j.envres.2020.109482 https://doi.org/10.1016/j.ijdrr.2016.12.007 https://doi.org/10.1016/j.ijdrr.2016.12.007 https://doi.org/10.1016/j.ijdrr.2016.12.007 https://doi.org/10.1111/disa.12133 https://doi.org/10.1111/disa.12133 https://doi.org/10.1111/disa.12133 https://doi.org/10.1111/disa.12133 https://doi.org/10.1111/j.1467-7717.2011.01264.x https://doi.org/10.1111/j.1467-7717.2011.01264.x https://doi.org/10.1111/j.1467-7717.2011.01264.x https://doi.org/10.1007/s11069-013-0688-9 https://doi.org/10.1007/s11069-013-0688-9 https://doi.org/10.1007/s11069-013-0688-9 https://www.oxfordlearnersdictionaries.com/definition/american_english/fear_1%23:~:text%3dnoun-%2cnoun%2cchild%2520was%2520shaking%2520with%2520fear https://www.oxfordlearnersdictionaries.com/definition/american_english/fear_1%23:~:text%3dnoun-%2cnoun%2cchild%2520was%2520shaking%2520with%2520fear https://www.oxfordlearnersdictionaries.com/definition/american_english/fear_1%23:~:text%3dnoun-%2cnoun%2cchild%2520was%2520shaking%2520with%2520fear https://doi.org/10.1016/j.ijdrr.2020.101724 https://doi.org/10.1016/j.ijdrr.2020.101724 https://doi.org/10.1016/j.ijdrr.2020.101724 https://doi.org/10.1016/j.ijdrr.2020.101724 https://doi.org/10.1080/13669877.2011.652649 https://doi.org/10.1080/13669877.2011.652649 https://doi.org/10.1080/13669877.2011.652649 https://doi.org/10.1108/09653560410521689 https://doi.org/10.1108/09653560410521689 https://doi.org/10.1108/09653560410521689 https://doi.org/10.1193/1.3672928 https://doi.org/10.1193/1.3672928 https://doi.org/10.1193/1.3672928 https://doi.org/10.1193/1.3672928 https://doi.org/10.1016/j.evolhumbehav.2010.07.002 https://doi.org/10.1016/j.evolhumbehav.2010.07.002 https://doi.org/10.1016/j.evolhumbehav.2010.07.002 https://doi.org/10.1016/j.proeng.2014.10.484 https://doi.org/10.1016/j.proeng.2014.10.484 https://doi.org/10.1016/j.scitotenv.2018.05.144 https://doi.org/10.1016/j.scitotenv.2018.05.144 https://doi.org/10.1016/j.scitotenv.2018.05.144 https://doi.org/10.1016/j.scitotenv.2018.05.144 https://doi.org/10.1016/j.jdeveco.2019.102433 https://doi.org/10.1016/j.jdeveco.2019.102433 https://doi.org/10.1016/j.heliyon.2019.e02099 https://doi.org/10.1016/j.heliyon.2019.e02099 https://doi.org/10.1016/j.heliyon.2019.e02099 https://doi.org/10.1016/j.amj.2014.07.004 https://doi.org/10.1016/j.amj.2014.07.004 https://doi.org/10.1016/j.amj.2014.07.004 https://doi.org/10.1016/j.jebo.2019.10.021 https://doi.org/10.1016/j.jebo.2019.10.021 https://doi.org/10.1016/j.jebo.2019.10.021 https://doi.org/10.1016/j.jdeveco.2015.12.004 https://doi.org/10.1016/j.jdeveco.2015.12.004 https://doi.org/10.1016/j.jdeveco.2015.12.004 https://doi.org/10.1016/j.sbspro.2010.04.018 https://doi.org/10.1016/j.sbspro.2010.04.018 https://doi.org/10.1016/j.sbspro.2010.04.018 http://www.aire.cdmx.gob.mx/descargas/ultima-hora/calidad-aire/pcaa/gaceta_oficial_cdmx.pdf http://www.aire.cdmx.gob.mx/descargas/ultima-hora/calidad-aire/pcaa/gaceta_oficial_cdmx.pdf http://www.aire.cdmx.gob.mx/descargas/ultima-hora/calidad-aire/pcaa/gaceta_oficial_cdmx.pdf http://geerassociation.org/administrator/components/com_geer_reports/geerfiles/unam_geer_sept2017_v2_final.pdf http://geerassociation.org/administrator/components/com_geer_reports/geerfiles/unam_geer_sept2017_v2_final.pdf http://geerassociation.org/administrator/components/com_geer_reports/geerfiles/unam_geer_sept2017_v2_final.pdf received 30 june 2015 accepted 9 august 2015 evaluating the three methods of goodness of fit test for frequency analysis xiankui zeng, dong wang * , jichun wu key laboratory of surficial geochemistry, ministry of education, school of earth sciences and engineering, nanjing university, nanjing, china. *email: wangdong@nju.edu.cn abstract in hydrological statistics, the traditional assessment of goodness of fit test is interested in the testing precision to the sample generated from supposed pdf. however, the ability to reject the hypotheses when the supposed pdf is different from real pdf should be also emphasized. in addition, the sensitivity of test method to series length is important in hydrological analysis. three methods of goodness of fit test that include chi-square (c-s), kolmogorov-smirnov (k-s), and anderson-darling (a-d) tests are applied in this study. the results of power test indicate that the most powerful tests for normal, uniform, p3, and weibull distribution are k-s, c-s, and a-d tests, respectively. the test method with the best comprehensive power is c-s test, followed by k-s and a-d test. keywords: test method, frequency analysis, chi-square, kolmogorov-smirnov, anderson-darling 1 introduction frequency analysis is an important method in hydrological statistics, which has been applied in uncertainty assessment of hydrological process extensively, such as flood control, drought resistance, and safety design of water conservancy project (raju and kumar, 2011; saf, 2009; xu et al., 2010; lee et al., 2012). this method serves many purposes in statistical data analysis by fitting a probability distribution function (pdf) to a series of observations. then, the fitted pdf provides summary information about the data, and allows predictions beyond the range of data used for parameter estimation. this prediction is based on the information of past events in order to define the probabilities of future events. therefore, the prediction is attached a risk of failure. the frequency analysis does not predict the future with certainty, but it provides good models for explaining and making efficient use of the events that had occurred in the past (khaliq et al., 2006; smakhtin, 2001; yang et al., 2010). there are three steps in frequency analysis (khaliq et al., 2006): (i) selecting a suitable probability distribution function (pdf); (ii) estimating the parameters of pdf based on samples; (iii) assessing the uncertainty of objective of interest in prescribed confidence level. hereinto, how to select a suitable pdf is the most crucial step for frequency analysis, because the following parameter estimation is influenced by the supposed pdf. therefore, the probability distribution characteristics of objective and its uncertainty assessment are determined by the selected pdf. thus, plenty of studies focus on determining the pdf that fits best the observed hydrological series. beard (1974) estimated the flood records at 300 stations in the usa by 8 different models, and the results showed that log-p3 and log-normal fitted the observations best. by applying the available methods to flood records which lengths varied between 44 and 97 year, the us water resources council (1976) recommended the use of log-p3 distribution, with a regional skew coefficient in the case of short records. the flood studies report of the nerc (1975) showed that the generalized extreme value (gev) distribution was recommended for the use in regional flood frequency analysis. vogel et al., (1993) analyzed the flood data at 61 stations in australia with records of 20 years or more. they found the best pdf at one region was even different for winter and summer-dominated rainfall regimes. recently, shabri and ariff (2009) found that the generalized logistic distribution was the most suitable pdf to fit the data of maximum daily rainfalls for stations in selangor and kuala lumpur. journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 178-187 published by atlantis press copyright: the authors 178 mailto:wangdong@nju.edu.cn song and singh (2010) analyzed the joint probability distribution of different hydrological periods in the wei river basin, china. the results showed that the drought duration and inter-arrival time followed the weibull distribution, and the drought severity followed the gamma distribution. similar to mentioned literatures, many others researchers made the goodness of fit tests for different types of hydrological series, and the best pdfs were not the same at different regions or periods (e.g., gunasekara and cunnane,1991; onoz and bayazit, 1995; onoz and bayazit, 2001; bi et al., 2010; lansigan, 2009; sandoval and raynal-villasenor, 2008; shao et al., 2004). therefore, there is not universal applicability to the selection of best pdf, and it should be selected based on the practical observations and effective comparison and testing. the choice of the suitable probability distribution function (pdf) is almost arbitrary as no physical basis is available to rationalize the use of any particular function. a common way is selecting one from several often-used probability distribution functions (pdfs) by goodness of fit test. as a result, the result of frequency analysis is affected by the power of the goodness of fit test. in the context of probability statistics, there is a wealth of the literatures on the goodness of fit test, such as pombo et al. (2015), durot and reboul (2010), goegebeur and guillou (2010), lee (2010), saldana-zepeda et al. (2010), stephens (1974a), stephens (1974b), woodruff et al. (1984). nevertheless, for hydrological frequency analysis, except for the precision and efficiency of the goodness of fit test, some issues should also be given special consideration. generally, there are two key points should be included. (i) the influence of series length on the goodness of fit test. it is hard to ensure the length of observations in hydrological process, the insufficiency of the sample size makes it difficult to choose the best pdf; (ii) the pdf for fitting observations is unknown in the ordinary course of events, it should be selected from several alternative pdfs. therefore, not only is the precision important for goodness of fit test, but also the accuracy that selecting the suitable pdf and rejecting others from several alternative pdf is crucial. the assessment to the goodness of fit test is common in hydrological literatures. according to the flood studies report of the nerc (1975), the conventional chi-square (c-s) and kolmogorov-smirnov (k-s) goodness of fit tests were insensitive to departures from the assumed distribution for the samples size of 30 to 60. ahmad et al. (1988) assessed the k-s, anderson-darling (a-d), and cramer-von mises (c-vm) goodness of fit tests by testing gamma distribution from several alternative distributions, which were gamma, weibull, normal, log-normal, uniform, and beta distribution. the sample size for each distribution was varied from 5 to 30, each test was repeated for 5000 times. the power of goodness of fit test was defined as the proportion of rejection to the hypothesis that the samples used for test were not from gamma distribution, but from another distribution. the testing result showed that c-vm test was most powerful, followed by a-d test, but the differences between these tests were very slight. nevertheless, the sample sizes of that study (5-30) was not widely representative in hydrological series, and the definition of power of test was not reasonable by only considering the proportion of rejections. onoz and bayazit (1995) studied the flood data of 19 stations with record lengths in the range 60 to 165, several test methods were applied to search for the best probability distribution of flood series. they found that c-s test was highly sensitive to the location of the data near the class limits, small errors in the parameters of the distribution may have significant effects on the test results, and the data inside the class intervals was insignificant. thus, c-s was not suitable for application in flood flows. in addition, a-d test was only limit to test gev distribution, as the critical values were not available for other distributions. therefore, this study was not a comprehensive assessment to the power of goodness of fit tests. shahabuddin et al. (2009) investigated the power of several goodness of fit test such as k-s, a-d, c-vm, etc., and the critical values of each test were obtained by monte carlo simulation. by comparing with the power, the modified k-s test was found to be the best. in that study, however, the power of test was defined as the proportion of hypotheses that were rejected. in addition, the parameters of pdf were assumed known in hypothesis, and the samples used for testing were generated from same distribution, only with small difference in location or scale parameter. for hydrological series, however, the parameters of supposed distribution are unknown, and should be estimated by observations, and the observation data may be from the supposed pdf, or from other distributions. therefore, the conclusions from shahabuddin et al. (2009) may be not applicable for frequency analysis in hydrological series. as the special requirement in hydrologic frequency analysis, the supposed probability distribution published by atlantis press copyright: the authors 179 function (pdf) is arbitrary chosen from several common used functions. moreover, the parameters of pdf should be estimated, and the length of series cannot be guaranteed. therefore, the assessment of goodness of fit test should consider not only the testing precision to the sample from known distribution function, but also emphasis on the ability to reject the hypotheses when the supposed pdf is different from the real one. in this paper, three often used goodness of fit test methods, c-s, k-s, and a-d tests were assessed. the samples used for test were generated from 4 common functions: normal, p3, weibull, and uniform. the paper is organized as follows: an introduction is first provided for the frequency analysis and goodness of fit test in hydrologic frequency analysis. it follows the description to the methods, including c-s, k-s, and a-d tests, the probability distribution functions and parameters estimation, and samples generation. next, the statistic results of test, power comparison of the three test methods, and a discussion on the application of the goodness of fit test in hydrologic frequency analysis are presented. finally, the main conclusions drawn from the analysis are presented. 2 methods the methods used in this paper include goodness of fit tests, c-s, k-s, a-d tests, parameter estimation for probability distribution function (pdf), and samples generation. 2.1 goodness of fit test assuming (x1, x2, …, xn) is the samples from population x, making a hypothesis: h0: f(x) = f0(x), where f0(x) is the alternative pdf (supposed pdf) with the parameters estimated by samples. chi-square test chi-square (c-s) test is a simple and convenient method for hypothesis test, it is related to the overall fit, the process can be written as follows (zhang, luo, 2000): (1) choosing k-1 numbers: -∞ < t1 < t2 < … < tk-1 < +∞, k ≈ 1.87(n-1)0.4, and the number axis is partitioned into k intervals, (-∞, t1], (t1, t2], …, (tk-2, tk-1], (tk-1, +∞]. (2) collecting the number of samples dropped into the i-th interval ni, i=1, 2, … , k, and then calculating the probability of the population which obeys alternative pdf fallen into the i-th interval: 1 1 0 1 2 1 2 0 2 0 1 1 2 1 0 1 0 2 1 0 1 ( ) ( ) ( ) ( ) ( ) ...... ( ) ( ) ( ) ( ) 1 ( ) k k k k k k k k p p x t f t p p t x t f t f t p p t x t f t f t p p t x f t                               (1) (3) constructing a statistics: 2 2 1 ( )k i i i i n np np      (2) which obeys chi-square distribution with the degree of freedom m, m=k-1, or m=k-1-r when there are r independent parameters of f0(x) need to be estimated by samples. and specifying a significance level α, if p (χ 2 ≥ χ2 1-α) ≥ α, then accept the hypothesis, otherwise reject it. kolmogorov-smirnov test kolmogorov-smirnov (k-s) test measures the greatest discrepancy between the observed and hypothesized distribution. the process can be summarized as follows (melo et al., 2009; wang, wang, 2010): (1) sorting the samples x (x1, x2, …, xn) by ascending order, and storing it to a new vector x′ (x′1, x ′ 2, …, x ′ n), calculating the empirical distribution function: 1 1 0, ( ) , 1, n k k n x x k f x x x x n x x               (3) (2) calculating the k-s statistics d(n): 1 ( ) 0 0 0 1 max ( ) ( ) 1 max ( ) , ( ) n n n x x x i i i n d f x f x i i f x f x n n               (4) specifying a significance level α, if p (d(n) ≥ d(n)(1-α)) ≥ α, then accept the hypothesis, otherwise reject it. anderson-darling test anderson-darling (a-d) test emphasizes discrepancies in both tails of the distribution, and that is often of prime importance in hydrologic frequency analysis. the process can be written as follows (coronel-brizio, hernandez-montoya, 2010; zhang et al., 2009): (1) sorting the samples x (x1, x2, …, xn) by ascending order, and storing it to a new vector x′ (x′1, x ′ 2, …, x ′ n), calculating the empirical distribution function (eq. 3). (2) calculating the a-d statistics a2 n : 2 1 1 [(2 1) log ( ) + (2 1 2 ) log(1 ( )) 1, 2,..., n n i i i a n i f x n n i f x i n           (5) published by atlantis press copyright: the authors 180 and specifying a significance level α, if a2 n < a2 n (1-α), then accept the hypothesis, otherwise reject it. however, the critical values of the distributions in this study are unavailable directly at the whole range of series length, and so, they are obtained by simulations, for which the following simulation steps are taken: (1) generating a sample of length n from the selected distribution, and estimating the parameters based on sample. (2) calculating the a-d statistics a2 n . (3) based on monte carlo simulation, step (1) – step (2) are repeated 10000 times, and the a-d statistics are sorted by ascending order, the critical values at the significance level α = 0.01, 0.05, 0.1 are the values at 9900th, 9500th, 9000th of the statistics series. 2.2 probability distribution function and parameters estimation four probability distribution functions are chosen for goodness of fit test in this study. two of this have two parameters (normal, uniform), and other two have three parameters (p3, weibull). these distributions have been selected because p3 and weibull distribution are currently applied in the hydrologic frequency analysis in various countries frequently. moreover, normal and uniform distributions have a significant different characteristics of probability distribution compared with other two distributions (fig. 1), in addition, they are often used in other field, especially in hydrogeology field (hassan et al., 2009; kuczera, parent, 1998). therefore, the power of the goodness of fit test can be assessed comprehensively. parameters of the p3 distribution are estimated by the principle of maximum entropy (chen et al., 2002; singh, 1987; singh, singh, 1985), and maximum likelihood method is used to estimate the parameters of weibull distribution (qiao and tsokos, 1994; qiao and tsokos, 1995). 2.3 samples generation the sample is fundamental to the power test of goodness of fit test in this paper. the assessment of the test methods is influenced by the characteristics of samples. therefore, the characteristics of generated samples should be keeping identical with their probability density function as much as possible. the samples of 4 probability density functions (normal, p3, weibull, and uniform distribution) are generated based on the procedure of random number generation of imsl (imsl, 1997). fig. 1. the probability density curves of normal, uniform, p3, and weibull distributions. published by atlantis press copyright: the authors 181 3 results and discussion the procedure of this power test is described as fig. 2. the four samples with length n were generated from normal, p3, weibull, and uniform distribution, and the supposed probability density function (pdf) in null hypothesis was set as one of four above functions respectively. after that, according to the hypothesis, estimating the parameters of pdf by sample, and testing all the hypotheses by c-s, k-s, and a-d tests in turn. the length of series n was varied from 20 to 300 with interval 10, each sample with length n was repeated for 2000 times based on monte carlo simulation, and the sample was generated randomly for each running. the significance levels of chi-square (c-s), kolmogorov-smirnov (k-s), and anderson-darling (a-d) test are all set as 0.05, the acceptance proportion for each hypothesis is denoted as follows: _ _ _ ( ) _ _ ( ) / 2000p m s h i m s h i (6) which m is the method used for goodness of fit test, s is the function used for generating samples, h is the supposed function in hypothesis, i = 1, 2, 3, …29, indicates the length of series varies from 20 to 300, m_s_h(i) is the accepting times for the specific hypothesis. as showed by fig. 3 – fig. 5, for the three test methods, in most instances, the accepting proportion is increase with the length of series when the sample fig. 2. the flow chart of this study. published by atlantis press copyright: the authors 182 is generated from supposed probability density function (pdf), and decrease when the sample is not generated from supposed function. in fact, it is attribute to the characteristics of sample which is closer to its pdf when the sample size is increased. however, when samples are generated from normal distribution, the hypotheses with weibull as the assumed function have very high accepting proportion for all three goodness of fit tests. the accepting proportions of the hypotheses in which the actual and supposed pdf both are p3 distribution are low for c-s and k-s tests. k-s and a-d tests are easy to accept the hypotheses when the supposed function is weibull distribution and the actual function is p3 distribution. furthermore, the k-s test is ineffective for testing the normal function from the samples generated by weibull distribution, and a-d test is very easy to accept the hypotheses when the supposed function is p3 and the samples are generated from weibull or uniform distribution. the ideal goodness of fit test should not only keep high testing precision for the samples generated from the supposed pdf, but also control the false accepting proportion when the samples are generated from other functions. therefore, a formula is proposed for describing the power of goodness of fit test for a specific pdf: _ _ ( ) _ _ _ ( ) (1 _ _ _ ( )) * 3 s h s h power m s i p m s h i p m s h i     (7) the symbols in eq. (7) are same as in eq. (6). fig. 6 exhibits the powers of the three goodness of fit test methods for 4 specific pdfs. the power is increase with the length of series for each test method. moreover, a-d test is faster to get a stable power value and the starting power is higher for each pdf (except uniform distribution) than c-s and k-s tests. for the 4 pdfs, the most powerful goodness of fit test for normal, uniform, p3, and weibull distribution are k-s, c-s, a-d, and a-d test respectively. for the test methods, c-s test is good at testing uniform, normal, and weibull distribution, but bad at p3 distribution. k-s test is powerful for testing normal and weibull distribution, and weak at p3 distribution. different from other two test methods, a-d test is good at p3, normal, and weibull distribution, and ineffective to uniform distribution. considering the powers for different probability density functions, the comprehensive power for each goodness of fit test is described as follows: fig. 3. the accepting proportion of hypotheses by chi-square test. published by atlantis press copyright: the authors 183 fig. 4. the accepting proportion of hypotheses by kolmogorov-smirnov test. fig. 5. the accepting proportion of hypotheses by anderson-darling test. published by atlantis press copyright: the authors 184 fig. 6. the powers of chi-square, kolmogorov-smirnov, and anderson-darling tests for specific probability density function. fig. 7. the comprehensive powers of chi-square, kolmogorov-smirnov, and anderson-darling tests. _ _ _ 4 s power m s power m   (8) the symbols in eq. (7) are same as in eq. (6). fig. 7 shows that the comprehensive powers for the tests are all increase with the series length. the goodness of fit test with the best performance is c-s test, the second is k-s test, and the last is a-d test, nevertheless, the difference among the powers of 3 test methods is not so significant. from the testing results above, it is concluded that the three test methods used in this study measure the different aspects of the goodness of fit. the c-s test is related to the overall fitting. therefore, it is appropriate to select c-s as the goodness of fit test when nothing is known about the samples. in addition, c-s test is powerful for distinguishing uniform distribution from other distributions and distinguishing other distributions when the sample is generated from uniform distribution. the power of c-s test is not so sensitive to the length of series (fig. 3). the k-s test measures the greatest discrepancy between the sample and supposed probability density function (pdf). it has the highest precision for testing the samples which generated from supposed function, and there is no other advantage to others except this. moreover, the k-s test is most insensitive to the length of series (fig. 4). the a-d test is emphasis on the differences at the both tails. thus, the power for specific pdf is most sensitive to the length of series, and it reach to a stable value quickly (fig. 5). in addition, the a-d test is extremely inefficient for testing pdfs when the samples are generated from uniform distribution. according to fig. 3 – fig. 5, it is obvious to found that the power for testing normal distribution from weibull samples, and testing weibull distribution from normal sample are fairly low for all the 3 test methods. referring to the probability density functions of normal and weibull distribution: 2 2 2 1 ( ) ( ; , ) exp[ ] 22 x f x        (9) 1 ( ; , , ) ( ) exp[ ( ) ], 0 x f x x x                    (10) published by atlantis press copyright: the authors 185 in fact, the above 2 distributions can be both transformed from exponential gamma distribution: 1 1 ( ; , , , ) ( ) exp( ( ) ) ( ) aa b f x a b c x c b x c a          (11) when the following conditions are satisfied respectively: 1 1 1 , , 0, 2 2 2 a b c     (12) 1 1, , ,a b c           (13) therefore, the normal and weibull distribution can be mutual transformed in some special condition, and it is hard to test them from each other with high precision. to sum up the above arguments, each goodness of fit test has its own advantage and disadvantage. as a result of the power test in this paper adopted the different evaluation criterion from other studies, which considering the ability to distinguish the sample from both real and false distributions. therefore, the a-d test is not always the best test method as reported in previous literatures. in order to acquire best testing effect in hydrological frequency analysis, it is beneficial to collect the prior information about the research variable as much as possible, and then select corresponding method. for example, from previous literatures, the observations always represent normal, weibull, and p3 distribution, and then a-d test should be selected for testing, because the a-d test has greatest comprehensive power when uniform distribution is excluded. in addition, the a-d test should be adopted when the length of series is not so large. when nothing is known about the probability distribution characteristics of samples, the c-s test is preferred. in addition, parameters estimation of probability distribution functions is always included in the procedure of goodness of fit test. thus, the methods used for estimating parameters may have influence on the results of hypothesis tests. moreover, how to effectively distinguish normal distribution (or weibull distribution) from weibull distribution (or normal distribution) is the focus of future work. 4 conclusions by choosing chi-square (c-s), kolmogorov smirnov (k-s), and anderson-darling (a-d) test as the methods of goodness of fit test, the supposed pdf in hypothesis and the function used for generating sample are selected as normal, uniform, p3, and weibull distributions. based on the comparison of goodness of fit test methods, following conclusions are demonstrated: (1) each goodness of fit test has its own advantage and disadvantage. the most powerful goodness of fit test method for normal, uniform, p3, and weibull distribution are k-s, c-s, a-d, and a-d test respectively. c-s test is good at testing uniform, normal, and weibull distribution, but bad at p3 distribution. k-s test is powerful for testing normal and weibull distribution, and weak at p3 distribution. a-d test is good at p3, normal, and weibull distribution, and ineffective to uniform distribution. (2) the goodness of fit test with the best comprehensive power is c-s test, the second is k-s test, and the last is a-d test. it is beneficial to search for the prior information about the research variables as much as possible, and then select corresponding method. acknowledgements this study was supported by the national basic research program (2013cb956503), national natural science fund of china (no. 41302181, 41172207, 51190091 and 51409161), and program for new century excellent talents in university (ncet-12-0262), science and technology program of jiangsu province (no. bk20140080). references ahmad, m.i., sinclair, c.d., spurr, b.d., 1988. assessment of flood frequency models using empirical distribution function statistics. water resour res, 24(8), 1323-1328. beard, l., 1974) flood flow frequency techniques. center of res. in water resour., university of texas, austin. bi, t.a.g., soro, g.e., dao, a., kouassi, f.w., srohourou, b., 2010. frequency analysis and new cartography of extremes daily rainfall events in cote d'ivoire. journal of applied sciences, 1684-1694. chen, y.f., hou, y., van gelder, p., zhigui, s., 2002. study of parameter estimation methods for pearson-iii distribution in flood frequency analysis. iahs-aish p, 271, 263-269. coronel-brizio, h.f., hernandez-montoya, a.r., 2010. the anderson-darling test of fit for the power-law distribution from left-censored samples. physica a-statistical mechanics and its applications, 389(17), 3508-3515. national environment research council, 1975. flood studies report, vols. 1-5. nerc, london. us water resources council, 1976. guidelines for determining flood flow frequency, bull. 17, hydrol. comm., washington, d.c. durot, c., reboul, l., 2010. goodness-of-fit test for monotone functions. scand j stat, 37(3), 422-441. goegebeur, y., guillou, a., 2010. goodness-of-fit testing for weibull-type behavior. j stat plan infer, 140(6), published by atlantis press copyright: the authors 186 1417-1436. gunasekara, t.a.g., cunnane, c., 1991. expected probabilities of exceedance for nonnormal flood distributions. j hydrol, 128(1-4), 101-113. hassan, a.e., bekhit, h.m., chapman, j.b., 2009. using markov chain monte carlo to quantify parameter uncertainty and its effect on predictions of a groundwater flow model. environ modell softw, 24(6), 749-763. imsl, 1997. fortran subroutines for mathematical applications, imsl inc., houston, texas, america. vol 2, 1165-1217. khaliq, m.n., ouarda, t.b.m.j., ondo, j.c., gachon, p., bobee, b., 2006. frequency analysis of a sequence of dependent and/or non-stationary hydro-meteorological observations, a review. j hydrol, 329(3-4): 534-552. kuczera, g., parent, e., 1998. monte carlo assessment of parameter uncertainty in conceptual catchment models, the metropolis algorithm. j hydrol, 211(1-4), 69-85. lansigan, f.p., 2009. frequency analysis of extreme hydrologic events and assessment of water stress in a changing climate in the philippines. from headwaters to the ocean: hydrological changes and watershed management, 497-501. lee, j.j., kwon, h.h., kim, t. w., 2012. spatio-temporal analysis of extreme precipitation regimes across south korea and its application to regionalization. journal of hydro-environment research 6(2), 101-110. lee, s., 2010. on the goodness of fit test for discretely observed sample from diffusion processes, divergence measure approach. j korean math soc, 47(6), 1137-1146. melo, i., tomasik, b., torrieri, g.., vogel, s., bleicher, m., korony, s., gintner, m., 2009. kolmogorov-smirnov test and its use for the identification of fireball fragmentation. phys rev c, 80(2), artn 024904. onoz, b., bayazit, m., 1995. best-fit distributions of largest available flood samples. j hydrol, 167(1-4), 195-208. onoz, b., bayazit, m., 2001. power distribution for low streamflows. j hydrol eng, 6(5), 429-435. pombo, s., de oliveira r. p., 2015. evaluation of extreme precipitation estimates from trmm in angola. j hydrol, 523: 663 679. qiao, h.z., tsokos, c.p., 1994. parameter-estimation of the weibull probability-distribution. mathematics and computers in simulation, 37(1), 47-55. qiao, h.z., tsokos, c.p., 1995. estimation of the 3-parameter weibull probability-distribution. mathematics and computers in simulation, 39(1-2), 173-185. raju, k.s., and kumar, d. n., 2011. classification of microwatersheds based on morphological characteristics. journal of hydro-environment research 5(2), 101-109. saf, b., 2009. regional flood frequency analysis using l moments for the buyuk and kucuk menderes river basins of turkey. j hydrol eng, 14(8), 783-794. saldana-zepeda, d.p., vaquera-huerta, h., arnold, b.c., 2010. a goodness of fit test for the pareto distribution in the presence of type ii censoring, based on the cumulative hazard function. computational statistics, data analysis, 54(4), 833-842. sandoval, c.e., raynal-villasenor, j., 2008. trivariate generalized extreme value distribution in flood frequency analysis. hydrolog sci j, 53(3), 550-567. shabri, a., ariff, n.m., 2009. frequency analysis of maximum daily rainfalls via l-moment approach. sains malaysiana, 38(2), 149-158. shahabuddin, f.a.a., ibrahim, k., jemain, a.a., 2009. on the comparison of several goodness of fit tests, with application to wind speed data. energy problems and environmental engineering, 394-398. shao, q.x., wong, h., xia, j., ip, w.c., 2004. models for extremes using the extended three-parameter burr xii system with application to flood frequency analysis. hydrolog sci j, 49(4), 685-702. singh, v.p., 1987. derivation of the pearson type (pt) iii distribution by using the principle of maximum-entropy (pome)-reply. j hydrol, 90(3-4), 355-357. singh, v.p., singh, k., 1985. derivation of the pearson type (pt) iii distribution by using the principle of maximum-entropy (pome). j hydrol, 80(3-4), 197-214. smakhtin, v.u., 2001. low flow hydrology: a review. j hydrol, 240(3-4), 147-186. song, s.b., singh, v.p., 2010. frequency analysis of droughts using the plackett copula and parameter estimation by genetic algorithm. stoch env res risk a, 24(5), 783-805. stephens, m.a., 1974a. components of goodness-of-fit statistics. ann i h poincare b, 10(1), 37-54. stephens, m.a., 1974b. edf statistics for goodness of fit and some comparisons. j am stat assoc, 69(347), 730-737. vogel, r.m., mcmahon, t.a., chiew, f.h.s., 1993. floodflow frequency model selection in australia. j hydrol, 146(1-4), 421-449. wang, f.g., wang, x.d., 2010. fast and robust modulation classification via kolmogorov-smirnov test. ieee t commun, 58(8), 2324-2332. woodruff, b.w., viviano, p.j., moore, a.h., dunne, e.j., 1984. modified goodness-of-fit tests for gamma-distributions with unknown location and scale-parameters. ieee t reliab, 33(3), 241-245. xu, y.p., booij, m.j., tong, y.b., 2010. uncertainty analysis in statistical modeling of extreme hydrological events. stoch env res risk a, 24(5), 567-578. yang, t., xu, c.y., shao, q.x., chen, x., 2010. regional flood frequency and spatial patterns analysis in the pearl river delta region using l-moments approach. stoch env res risk a, 24(2), 165-182. yoo, c., jung, k.s., kim, t.w., 2005. rainfall frequency analysis using a mixed gamma distribution, evaluation of the global warming effect on daily rainfall. hydrol process, 19(19), 3851-3861. zhang, d.p., luo, y.l., 2000. applied probability and statistic. higher education press, beijing. in chinese zhang, w., yu, s., zhang, g.., 2009. anderson-darling test for clutter distribution identification. chinese journal of scientific instrument, 30(3), 631-635. in chinese with english abstract. published by atlantis press copyright: the authors 187 http://apps.webofknowledge.com/full_record.do?product=ua&search_mode=generalsearch&qid=1&sid=u2xadiqmubtrb7zb1ei&page=1&doc=9 http://apps.webofknowledge.com/full_record.do?product=ua&search_mode=generalsearch&qid=1&sid=u2xadiqmubtrb7zb1ei&page=1&doc=9 research article a sales forecasting model for the consumer goods with holiday effects mu zhang1,*, xiao-nan huang1,2, chang-bing yang1,2 1school of big data application and economics, guizhou university of finance and economics, guiyang 550025, china 2guizhou institution for technology innovation and entrepreneurship investment, guizhou university of finance and economics, guiyang 550025, china 1. introduction sales forecasting refers to the estimation of sales quantity and sales amount of all products or specific products in a specific time in the future [1]. sales forecasting is based on the full consideration of various influence factors in the future, combined with the actual sales performance of enterprises, through certain analysis methods to put forward practical sales objectives. through sales forecasting, the initiative of sales staff can be mobilized, and the products can be sold as soon as possible, so as to complete the transformation from use value to value. meanwhile, the enterprises can set production by sales, arrange production according to sales forecasting data, and avoid overstock of products. it can be seen that the accurate and reliable sales forecasting is of great significance to the formulation of enterprises marketing plan, safety inventory, normal operation of cash flow and so on. at present, the commonly used qualitative forecasting methods in sales forecasting include: senior manager’s opinion method, salesperson’s opinion method, buyer’s expectation method and delphi method, etc. in addition, the commonly used quantitative forecasting methods in sales forecasting include: time series analysis method [1], regression and correlation analysis method [1], artificial intelligence technology represented by neural network [2] and support vector regression machine [3], etc. the qualitative forecasting method is simple and easy, but it has strong subjectivity. it is difficult to make an accurate explanation for the future change trend, to make a quantitative explanation for the interaction degree between various forecasting objectives, to estimate the error of the forecasting results and to evaluate the reliability of the forecasting results. the quantitative forecasting method is less affected by subjective factors, which overcomes the shortcomings of qualitative forecasting method. however, there are so-called holiday effects in the sales of many consumer goods in reality, and their sales data have the characteristics of double trend change of time series, that is, the overall trend variability and seasonal volatility [4,5], which leads to the decrease of prediction accuracy and the increase of prediction risk of quantitative forecasting method. the so-called holiday effects originally refer to the abnormal phenomenon of stock index return rate in the pre-holiday trading days and post-holiday trading days compared with other trading days due to the change of people’s mood and behavior caused by the holiday in the stock market [6]. with the deepening of research, some scholars found that the holiday effects widely exist in consumer goods, tourism, film and television and many other fields. in terms of theoretical research on the holiday effects in consumer goods market, hua [7] thought that for the phenomenon of holiday shopping, the economists call it “the period of consumption function mutation”, the sociologists call it “the period of consumption behavior impulse”, and the business counselors call it “holiday effects”. chen [8] summed up the main characteristics of holiday economy, the impact of holiday economy on economic life, the consumption potential of holiday economy and its development countermeasures. in terms of empirical research on the holiday effects in consumer goods market, li [9] made a sample analysis on the sales volume of social science books in jingdong mall in recent 3 years, which shows that there are holiday and weekend negative effects in book online sales. gui and han [10] took the monthly data of total retail sales of consumer goods from january 2000 article info article history received 09 june 2020 accepted 23 june 2020 keywords consumer goods sales forecasting holiday effects seasonal decomposition arima model seasonal factor abstract in reality, there are so-called holiday effects in the sales of many consumer goods, and their sales data have the characteristics of double trend change of time series. in view of this, by introducing the seasonal decomposition and arima model, this paper proposes a sales forecasting model for the consumer goods with holiday effects. first, a dummy variable model is constructed to test the holiday effects in consumer goods market. second, using the seasonal decomposition, the seasonal factor is separated from the original series, and the seasonally adjusted series is then obtained. through the arima model, a trend forecast to the seasonally adjusted series is further carried out. finally, according to the multiplicative model, refilling the trend forecast value with the seasonal factor, thus, the final sales forecast results of the consumer goods with holiday effects can be obtained. taking the cigarettes sales in g city, guizhou, china as an example, the feasibility and effectiveness of this new model is verified by the example analysis results. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: rim_007@163.com journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 69–76 doi: https://doi.org/10.2991/jracr.k.200709.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto: rim_007@163.com https://doi.org/10.2991/jracr.k.200709.001 https://www.atlantis-press.com/journals/jracr 70 m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 to december 2009 as the research object, and used bayes seasonal adjustment model to measure the holiday effects of household consumption. 곽영식 et al. [11] demonstrated empirically that replacement intervals of mobile phones sold in china online b2c are influenced by purchase points such as holidays. cheng et al. [12] analyzed the fluctuations of the three fresh cut flower yields of roses, gypsophila and gerbera before and after the festival, which had the largest trading volume in kunming international flower trading center, the festival trading rules of fresh cut flowers in the auction market were then determined basically. cai et al. [13] proposed an oracally efficient estimation for dense functional data with holiday effects, and applied it to analysis the sporting goods sales data. in view of the sales forecast problem of the consumer goods with holiday effects, in xie et al. [14], the time series of cigarettes sales was divided into trend component and cycle component. based on the circular back propagation network model, the two components were predicted respectively, and the overall predicted value of cigarettes sales volume was obtained through combination. in zou [15], according to the seasonal volatility of cigarettes sales, the paper selected the seasonal index prediction model as the basic prediction model of cigarettes sales, by introducing the lunar date, the prediction process and its system implementation method based on the lunar date were proposed. however, the former is limited by the defects of bp neural network, such as “black box”, difficulty in structure determination, low training efficiency and so on [16]; the latter assumes that the year-on-year growth rate of sales in the forecast month is equal to the year-on-year growth rate of total sales in the previous 12 months in trend prediction, its rationality is worth discussing. seasonal decomposition is to decompose the components of time series into four kinds: long-term trend factor (t), seasonal change factor (s), cycle change factor (c) and random fluctuation factor (i), etc. these components are then separated from time series to study their influence on the change of time series [17]. at present, the seasonal decomposition has been widely used in tax analysis and prediction, tourism passenger flow analysis and prediction, automobile sales analysis, agricultural product price prediction and other fields [17–20]. in addition, the arima model not only considers the dependence of economic phenomena on time series, but also considers the interference of random fluctuation in the process of economic prediction, so that the arima model has a relatively high prediction accuracy for the short-term trend of economic operation [21,22]. in view of this, by introducing the seasonal decomposition and arima model, this paper proposes a sales forecasting model for the consumer goods with holiday effects. this paper is structured as follows: section 2 constructs a dummy variable model to test the holiday effects in consumer goods market. section 3 proposes a new model to forecast the sales of consumer goods with the holiday effects. section 4 describes the example analysis results and section 5 concludes this paper. 2. a test model of holiday effects in consumer goods market according to the connotation of holiday effects in consumer goods market and the test principle of holiday effects in stock market, the following dummy variable model is constructed to test the holiday effects in consumer goods market. s i c d i it t t( ) ( ) ( ) pre= + +a e (1) where, s(i)t is the sales volume of the i-th consumer goods in the t-th period (month or day); d i t( ) pre is the dummy variable, pre is the last period before the holiday, when t is the last period before the holiday, d i t( ) pre = 1, otherwise d i t( ) pre = 0; c is the intercept term of the model; a is the estimation coefficient of d i t( ) pre, when a is significantly not zero, it indicates that there is significant holiday (pre-holiday) effects; e (i)t is the random disturbance term. in this paper, the weighted least square method is used for parameter estimation, where the weight w is the reciprocal of the estimated value of residual obtained by ordinary least square method. 3. a sales forecasting model for the consumer goods with holiday effects 3.1. seasonal decomposition generally speaking, the information of time series can come from the following four aspects [17]: (1) long-term trend refers to the continuous change of phenomena in a certain direction over a long period of time. the long-term trend is the result of the influence of some fixed and fundamental factors. (2) seasonal change refers to the regular change with the change of seasons in a year under the influence of nature and other factors. (3) cycle change refers to the regular period change with several years (months and quarters) as a certain period. the cycle change is different from the one-way continuous change of long-term trend and the fixed period rule of seasonal fluctuation, which is difficult to identify. (4) random fluctuation refers to the irregular change of phenomena influenced by accidental factors. seasonal decomposition can subjectively decompose time series into four kinds of factors: long-term trend factor (t), seasonal change factor (s), cycle change factor (c) and random fluctuation factor (i), i.e. time series can be considered as a function of these four factors, it can be expressed as: y f t s c it t t t t= ( , , , ) (2) where, yt represents the time series, tt represents the long-term trend factor (may be linear trend, may also be periodic fluctuation or long period fluctuation), st represents the seasonal factor (refers to the fluctuation with fixed amplitude and period, for example, the calendar effect is the common seasonal factor), ct represents the cycle change factor and it represents the random fluctuation factor (may be regarded as error). the function f includes the additive model yt = tt + st + ct + it and the multiplicative model yt = tt + st + ct + it where, the multiplicative model is often used [23]. in the sales of consumer goods with holiday effects, the quantity of each period is affected by many different factors. for example, the monthly sales volume will be affected by some factors, such as residents’ purchasing power, commodity price, commodity quality, customers’ preferences, seasonal change and so on. according m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 71 to the historical sales data of consumer goods with holiday effects, the sales of consumer goods with holiday effects have the characteristics of double trend change of time series, that is, the overall trend variability and seasonal volatility [5]. in view of this feature, using the seasonal decomposition, the seasonal factor is separated from the original series, and the seasonally adjusted series is then obtained. 3.2. trend forecast arima model is a time series analysis model put forward by box et al. in the 1970s. its basic idea is as follows [21,22]: some time series are a group of random variables depending on time t, although the single series value of time series has uncertainty, the change of the whole series has certain regularity, therefore, which can be approximately described by the corresponding mathematical model. time series model is a model based on its past value and random disturbance term. its concrete form is x x x xt t t p t p t t q t q= + + + + + + +-j j j e q e q e1 1 2 2 1 1... ... (3) the above formula shows that a random time series can be generated by an autoregressive moving average process, that is, it can be interpreted by its own past or lag values and random interference terms. if the time series is stable, that is, its behavior will not change with the passage of time, the future can be predicted according to the past behavior of the series. arima(p, d, q) model is called differential autoregressive moving average model, where ar is autoregressive, p is the number of autoregressive terms; ma is the moving average, q is the number of moving average terms; i is single integer, d is the number of difference times (order) to make time series become stable series. the specific modeling steps of arima(p, d, q) model include [21,22]: stability test and processing, model recognition, model order determination and model test, etc. the characteristics of double trend prediction are the importance of the ranking of observation values, and the correlation between the front-back observation values and the same period ratio, that is, the correlation between the prediction point and the observation point close to each other is strong, while the correlation between the prediction point and the observation point far away from each other is weak [14]. therefore, the arima(p, d, q) model is established to predict the trend of the seasonally adjusted series in subsection 3.1, and the trend prediction value in the prediction period is then obtained. 3.3. sales forecast of the consumer goods with holiday effects according to the multiplicative model of the seasonal decomposition, refilling the trend forecast value obtained in subsection 3.2 with the seasonal factor separated out in subsection 3.1, that is, the corresponding seasonal factor is multiplied by the trend prediction value of the seasonally adjusted series in the prediction period, thus, the final sales forecast results of the consumer goods with holiday effects can be obtained. 4. example analysis 4.1. sample data the original series of total monthly sales volume of cigarettes in g city, guizhou, china from 2007 to 2010 was shown in figure 1. the original data were from g city branch, guizhou tobacco company of china (see table 1). it can be seen from figure 1 that from 2007 to 2010, the original series shows the general characteristics of rising the bottom gradually and emerging two seasonal peaks in january and september every year. therefore, it can be preliminarily judged that the total monthly sales volume of cigarettes in g city have the characteristics of double trend change of time series. 4.2. test of the holiday effects in this paper, the spring festival and national day were selected as the holidays to investigate the holiday effects of cigarettes sales in g city. in order to facilitate comparison, the sample data were divided into two subgroups: the last month before the holiday (referred to as pre-holiday) and all other months (referred to as other). we used excel software to make descriptive statistical analysis of sample data, and used eviews7.2 software to do regression analysis of sample data. the descriptive statistical analysis results of the original series were shown in table 2. it can be seen from table 1 that the pre holiday average value is 21381.04 boxes, which is far higher than the other average value of 15631.69 boxes, reaching 1.3678 times. this preliminary shows that the cigarettes sales in g city have obvious pre-holiday effects. in addition, from the standard deviation point of view, the pre-holiday standard deviation is 3371.942 boxes, which is much higher than the other standard deviation of 1843.548 boxes, reflecting the greater pre-holiday volatility. therefore, the follow-up needs to carry on the regression analysis to get the general conclusion. using equation (1), a dummy variable model was constructed, and the pre-holiday effects of cigarettes sales in g city were then test. the test results were shown in table 3. figure 1 | original series. 72 m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 table 1 | seasonal decomposition time (month-year) original series (box) moving average series (box) ratio of original series to moving average series (%) seasonal factor (%) seasonally adjusted series (box) smoothed trend-cycle series (box) irregular (error) component jan-2007 20242.14 — — 146.3 13839.430 14024.348 0.987 feb-2007 13116.56 — — 91.3 14365.099 14122.436 1.017 mar-2007 13955.97 — — 98.5 14162.779 14318.610 0.989 apr-2007 14054.03 — — 96.7 14534.791 14498.119 1.003 may-2007 13811.13 — — 93.7 14739.944 14628.354 1.008 jun-2007 14045.40 — — 95.5 14708.150 14707.775 1.000 jul-2007 14883.09 14615.6077 101.8 100.7 14786.690 14750.661 1.002 aug-2007 14433.64 14516.1466 99.4 98.5 14657.462 14760.819 0.993 sep-2007 17377.90 14585.9138 119.1 116.4 14924.713 14852.390 1.005 oct-2007 12893.89 14601.8745 88.3 87.5 14744.025 14932.850 0.987 nov-2007 14535.67 14686.0441 99.0 95.0 15307.709 14826.872 1.032 dec-2007 12037.88 14896.2546 80.8 80.0 15041.271 14645.788 1.027 jan-2008 19048.61 14980.3642 127.2 146.3 13023.418 14375.738 0.906 feb-2008 13953.76 15134.6609 92.2 91.3 15281.998 14580.863 1.048 mar-2008 14147.50 15310.2532 92.4 98.5 14357.145 15027.746 0.955 apr-2008 15064.06 15469.7871 97.4 96.7 15579.378 15707.082 0.992 may-2008 16333.66 15598.9116 104.7 93.7 17432.113 16218.742 1.075 jun-2008 15054.72 15727.3979 95.7 95.5 15765.091 16420.960 0.960 jul-2008 16734.65 15838.1249 105.7 100.7 16626.259 16556.047 1.004 aug-2008 16540.75 16429.4141 100.7 98.5 16797.244 16562.548 1.014 sep-2008 19292.31 16409.1248 117.6 116.4 16568.869 16654.500 0.995 oct-2008 14443.38 16632.9389 86.8 87.5 16515.855 16671.881 0.991 nov-2008 16077.51 16768.1978 95.9 95.0 16931.437 16852.494 1.005 dec-2008 13366.60 16763.3228 79.7 80.0 16701.505 16805.299 0.994 jan-2009 26144.08 16901.4492 154.7 146.3 17874.548 16785.723 1.065 feb-2009 13710.29 16926.9555 81.0 91.3 15015.350 16546.676 0.907 mar-2009 16833.27 16918.0162 99.5 98.5 17082.714 16782.121 1.018 apr-2009 16687.17 17051.2879 97.9 96.7 17258.007 17021.664 1.014 may-2009 16275.16 17202.2155 94.6 93.7 17369.680 17293.307 1.004 jun-2009 16712.23 17187.1725 97.2 95.5 17500.818 17227.649 1.016 jul-2009 17040.72 17318.4606 98.4 100.7 16930.352 17164.595 0.986 aug-2009 16433.48 17374.2949 94.6 98.5 16688.309 17322.005 0.963 sep-2009 20891.57 17811.1228 117.3 116.4 17942.366 17561.015 1.022 oct-2009 16254.51 17956.3201 90.5 87.5 18586.865 17831.815 1.042 nov-2009 15896.99 18068.2965 88.0 95.0 16741.333 17890.310 0.936 dec-2009 14942.06 18050.0652 82.8 80.0 18670.030 18389.016 1.015 jan-2010 26814.09 18155.8828 147.7 146.3 18332.632 18826.967 0.974 feb-2010 18952.23 18306.9824 103.5 91.3 20756.256 19328.173 1.074 mar-2010 18575.64 18439.0649 100.7 98.5 18850.901 18981.047 0.993 apr-2010 18030.89 18467.9041 97.6 96.7 18647.690 18611.562 1.002 may-2010 16056.38 18346.4800 87.5 93.7 17136.192 18216.417 0.941 jun-2010 17982.04 18648.9966 96.4 95.5 18830.546 18352.578 1.026 jul-2010 18853.92 18587.4788 101.4 100.7 18731.804 18425.337 1.017 aug-2010 18018.47 — — 98.5 18297.878 18287.517 1.001 sep-2010 21237.64 — — 116.4 18239.582 18272.450 0.998 oct-2010 14797.42 — — 87.5 16920.699 18268.385 0.926 nov-2010 19527.19 — — 95.0 20564.345 18410.893 1.117 dec-2010 14203.85 — — 80.0 17747.637 18482.148 0.960 table 3 shows that the constant term c is 15645.03 and the regression coefficient a is 5585.926 in this model. the t-statistic of a is 38.36343, and the corresponding probability p-value is 0.0000, which shows that the regression coefficient a is significant at the 1% significance level. the goodness-of-fit (r2) of the model is 0.9696, and the adjusted goodness-of-fit (adjusted r2) is 0.9690, which shows that the fitting effect of the model is good, and the independent variable can explain 96.96% difference of the dependent variable. the observed value of f-statistic is 1471.753, which is higher than the critical value (7.220) of f-test at the 1% significance level. therefore, the hypothesis of zero is rejected, which shows that the linear relationship between dependent variable and independent variable is very significant, and a linear model can be established. the above results show that the cigarettes sales in g city have the statistically significant positive pre-holiday effects. 4.3. seasonal decomposition results according to subsection 3.1, using the seasonal decomposition function provided by spss18.0 software, we selected m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 73 figure 3 | seasonal factor. figure 2 | original series and seasonally adjusted series. table 2 | descriptive statistical analysis results subgroups average value/box maximal value/box minimal value/box standard deviation/box sample sizes multiple of average value pre-holiday 21381.04 26814 17378 3371.942 8 1.3678 other 15631.69 19527 12038 1843.548 40 1.0000 multiple of average value, pre-holiday average value/other average value. table 3 | test results of the pre-holiday effects variable coefficient std. error t-statistic prob. c 15645.03 141.3023 110.7203 0.0000 d i t( ) pre 5585.926 145.6055 38.36343 0.0000 “multiplicative” model (system default option) in the “model type” option group, and selected “all points equal” in “moving average weight”, thus, we can get the seasonal decomposition table, as shown in table 1. based on the original series, according to the specified moving average method and period, we obtained the moving average series. its purpose was to eliminate the seasonal factors in the original series and get the trend value of the original series. according to the multiplicative model, the trend factors were removed from the original series, and the ratio of original series to moving average series was then obtained. we averaged the ratio of the same month in each year to get 12 averages, and then divided the 12 averages by the total ratio average, therefore, the seasonal factor for each month can be obtained. after removing the seasonal component from the original series, we got the seasonally adjusted series, and then smooth it to get the smoothed trend-cycle series. finally, the irregular (error) component was obtained by eliminating the cyclic fluctuation factors from the seasonally adjusted series [23]. the original series and seasonally adjusted series were shown in figure 2. in figure 2, it can be seen that the original series shows the characteristics of fluctuating growth in annual cycle, and the seasonally adjusted series (i.e. the corrected monthly effects series) shows a steady growth trend in four years. the seasonal factor was shown in figure 3. in figure 3, the seasonal factor fluctuates regularly in a 12-month period. it can be found that the sales volume of cigarettes in january and september are larger than that in other months. figures 2 and 3 further verify the observations in subsection 4.1. 4.4. trend forecast results according to subsection 3.2, we used spss18.0 software to build the arima(p, d, q) model and forecast the trend of the seasonally adjusted series. (1) stability test and processing. the augmented dickey–fuller (adf) test results shown that the seasonally adjusted series was a first-order stationary series, i.e. d = 1. (2) model recognition. because the partial autocorrelation function (pacf) graph and acf graph of the stationary series were tailed, it can be determined that the stationary series was suitable for the arma model. (3) model order determination. the determining order method based on the optimal criterion function was used to determine the order of the model [24], i.e. we selected the group of orders minimizing the akaike’s information criterion as the ideal order. through the comparative test, the arima(1, 1, 1) model was selected. (4) model test. through observing the adf test, acf graph and pacf graph of the residual series of the arima(1, 1, 1) model, it was confirmed that the residual series was the white noise series. thus, the arima(1, 1, 1) model was the best prediction model for the stationary series. in the “criteria” option group, we selected the “nonseasonal arima”, “none transformation” and “include constant in model”; selected the “display forecasts” in the “statistics” option group; set january to march 2011 as the forecast period in the “selection” option group; and other items adopted the default options. the trend forecast results of the seasonally adjusted series were shown in table 4. 74 m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 table 4 | the trend forecast results of the seasonally adjusted series months 2011–01 2011–02 2011–03 forecast value/box 19426.1 19430.3 19547.5 upper limit value/box 21313.2 21319.1 21436.5 lower limit value/box 17539.0 17541.5 17658.6 table 5 | the forecast results of the total monthly sales volume of cigarettes in g city from january to march 2011 months actual value/box forecast value/box error/% upper limit value/box error/% lower limit value/box error/% 2011–01 30756.98 28420.36 −7.60 31181.21 1.38 25659.50 −16.57 2011–02 18614.07 17739.86 −4.70 19464.35 4.57 16015.38 −13.96 2011–03 19315.71 19254.33 −0.32 21114.98 9.32 17393.69 −9.95 error = (forecast value − actual value)/actual value. table 6 | comparative analysis results months actual value/box nonseasonal arima(2, 1, 1) product seasonal arima(1, 1, 1) (1, 0, 1) winters multiplicative model our model forecast value/box error/% forecast value/box error/% forecast value/box error/% forecast value/box error/% 2011–01 30756.98 20631.91 −32.92 26933.74 −12.43 27552.18 −10.42 28420.36 −7.60 2011–02 18614.07 18583.61 −0.16 18669.34 0.30 17834.00 −4.19 17739.86 −4.70 2011–03 19315.71 19305.21 −0.05 19465.53 0.78 18958.20 −1.85 19254.33 −0.32 absolute average error/% — 11.05 — 4.30 — 5.49 — 4.20 error = (forecast value − actual value)/actual value. 4.5. sales forecast results according to subsection 3.3, refilling the trend forecast value in table 4 with the seasonal factor in table 1, i.e. the seasonal factor of the corresponding month was multiplied by the trend forecast value of the seasonally adjusted series, thus, the forecast results of the total monthly sales volume of cigarettes in g city from january to march 2011 were obtained, as shown in table 5. table 5 shows the relative error between the forecast value obtained by our model and the actual value of the total monthly sales volume of cigarettes in g city. the absolute average error after calculation is 4.20%, which shows that our model can better predict the change trend of the total monthly sales volume of cigarettes in g city, and can effectively simulate the seasonal, periodic and random characteristics of cigarettes sales. in addition, our model also gives the upper and lower limits of the forecast value, which provide a space for the empirical adjustment of the forecast value. 4.6. comparative analysis in order to facilitate comparison, based on the original series, we also adopted respectively the nonseasonal arima model, product seasonal arima model [25] and seasonal exponential smoothing method winters multiplicative model to forecast the total monthly sales volume of cigarettes in g city from january to march 2011, the results were shown in table 6. where, the nonseasonal arima (2, 1, 1) model and the product seasonal arima(1, 1, 1)(1, 0 ,1) model were finally determined as the best prediction models for the stationary series after the modeling steps of stability test and processing, model recognition, model order determination and model test. it can be seen from table 6 that the absolute average error of our model is only 4.20%, which is lower than 5.49% of the winters multiplicative model and 11.05% of the nonseasonal arima (2, 1, 1) model, which is also slightly lower than 4.30% of the product seasonal arima(1, 1, 1)(1, 0, 1) model. especially in january 2011, when the seasonal fluctuation is prominent, the prediction error of our model is only −7.60%, which is significantly lower than the other three models. which means that the prediction accuracy of our model is better, which shows good applicability in the double trend prediction and can meet the needs of practical application. 5. discussion from the results and analysis of the previous section, we observed that our model was able to obtain the slightly higher prediction accuracy than the product seasonal arima(1, 1, 1)(1, 0, 1) model [our method was 95.8%, and the product seasonal arima(1, 1, 1) (1, 0, 1) model was 95.7%], and the higher prediction accuracy than the winters multiplicative model and the nonseasonal arima (2, 1, 1) model [the winters multiplicative model was 94.51%, and the nonseasonal arima(2, 1, 1) model was 88.95%]. compared with the product seasonal arima, winters multiplicative model and the nonseasonal arima model, our method can obtain the slightly higher or higher prediction accuracy. since our model divides the sales forecasting into three stages. first, using the seasonal decomposition, the seasonal factor is separated from the original series, and the seasonally adjusted series is then obtained. second, using the arima model, a trend forecast to the seasonally adjusted series is further carried out. finally, according to the multiplicative model, refilling the trend forecast value with the seasonal factor, thus, the final sales forecast results of the consumer goods with holiday effects can be obtained. m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 75 in addition, compared with literature [14], our model avoids the limitation of bp neural network in trend prediction. compared with literature [15], our model overcomes the irrationality of hypothesis premise in trend prediction. obviously, the advantage of our model is that it can give full play to the advantages of seasonal decomposition in dealing with seasonal change factor, and the advantages of arima model in trend prediction at the same time. therefore, our model has high prediction accuracy and low prediction risk in sales forecasting of the consumer goods with holiday effects. it should be noticed that the time span of the original series in this paper is from 2007 to 2010, and there are only 48 sample data, which may weaken the persuasiveness of the example analysis results. however, due to the trade secrets involved, the data problem cannot be solved temporarily. 6. conclusion there are so-called holiday effects in the sales of many consumer goods in reality, and their sales data have the characteristics of double trend change of time series, which leads to the decrease of prediction accuracy and the increase of prediction risk of quantitative prediction method. in view of this, by introducing the seasonal decomposition and arima model, this paper proposed a sales forecasting model for the consumer goods with holiday effects. taking the cigarettes sales in g city, guizhou, china as an example, the feasibility and effectiveness of this new model was verified by the example analysis results. this paper provides a new method and idea for improving the sales forecasting accuracy of the consumer goods with holiday effects, which has high practical value. the basic law of prediction is that the shorter the prediction period, the stronger the prediction ability of quantitative prediction method; the longer the prediction period, the weaker the prediction ability of quantitative prediction method. therefore, in practical application, it is necessary to establish a sales forecasting system based on our model to make real-time dynamic sales forecast for the consumer goods with holiday effects. it should be noticed that the seasonal decomposition artificially decomposes time series into four fixed components; however, its scientific nature needs to be further verified. in addition, since the information used in a single prediction method is limited, only one method is used to predict the same problem, the prediction accuracy is often not high, and the prediction risk is large [26]. therefore, in order to improve the forecast effect, it is worth further study to establish a sales combination forecasting model for the consumer goods with holiday effects [27]. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution mz contributed in conceived and designed the experiments. cy contributed in performed the experiments. xh contributed in analyzed the data. mz and xh contributed in paper drafting and final revision. acknowledgments to the regional project of national natural science foundation of china (71861003) and the innovative exploration and new academic seedlings project of guizhou university of finance and economics (guizhou-science cooperation platform talents [2018] 5774-016) for their support. references [1] mentzer jt, bienstock cc. sales forecasting management (translated by luo y b et al.). beijing: posts & telecom press; 2004. [2] zhong dt, zhang y. study on the improvement of bp neural network to the prediction method of tobacco sales volume. ind technol econ 2007;26:115–18. [3] wei hy, wang jh, he w. study on the support vector machines model for sales volume prediction and parameters selection. acta simul syst sin 2005;17:33–6. [4] luo yh, lu yg, li b. hybrid forecast model of cigarette sales. comput eng des 2009;30:4744–7. [5] luo yh, lv yg, li b. hybrid forecast model for cigarette sales based on arma. appl res comput 2009;26:2664–8. [6] chen q, xia yt. an empirical study of the holiday effects on the real estate stock index in sse. j chongqing technol bus univ (west forum) 2009;19:103–5. [7] hua yl. correct treatment of long-term benefits and healthy utilization of holiday effects. comm res 1996;25–6. [8] chen yx. “festival effect” reconstructs business form — looking at holiday economy from shanghai market in the new spring of the dragon year. capital shanghai 2000;48–51. [9] li jj. analysis on the holiday effects of book online sales — an empirical study based on jingdong. china publ j 2014;19–23. [10] gui wl, han zz. china’s consumption seasonal adjustment and holiday effect estimation. stat res 2015;32:60–8. [11] 곽영식, 홍재원, 남용식, et al. a study on the holiday effect of purchase interval of mobile phone in chinese online b2c market. j korea soc comput inform 2019;24:205–10. [12] cheng sg, chen w, chen hd. study on the festival effect of fresh cut flowers in yunnan — an empirical analysis based in the transaction price of the auction market. price theor pract 2019;71–4. [13] cai l, li l, huang s, ma l, yang l. oracally efficient estimation for dense functional data with holiday effects. test 2020;29: 282–306. [14] xie xf, xie df, zou p. study and application of double trend time series prediction models of cigarette sell based on cbp. control theor appl 2007;24:1015–20. [15] zou ym. research on lunar-based time sequence forecasting model and its applications on tobacco sales forecasting. hohhot: inner mongolia university; 2008. [16] wang c, li w. credit risk assessment in commercial banks: projection pursuit discriminant model. j ind eng eng manage 2000;14:43–6. [17] zhang l, xu j. the trends and seasonal change of industrial and commercial tax of china from 1950 to 1984. stat res 2007;24:34–8. [18] wu y, huang z. a research into the fluctuation of inbound tourists flow in jiangsu province. econ survey 2010;117–21. [19] yin xp, wang yx. an empirical analysis on the influencing factors of china’s automobile sales. stat decision 2011;98–100. https://doi.org/10.1007/s11749-019-00655-5 https://doi.org/10.1007/s11749-019-00655-5 https://doi.org/10.1007/s11749-019-00655-5 76 m. zhang et al. / journal of risk analysis and crisis response 10(2) 69–76 [20] li gq, xu sw, li zm, dong xx. study on short-term forecasting methods and modeling of agro-product market price: forecasting based on the time series models. j china agric univ 2011;16:172–8. [21] box gep, jenkins gm, reinsel gc. time series analysis: forecasting and control. (translated by gu l, fan jc). 3rd ed. beijing: china statistics press; 1997. [22] gao t. econometric analysis method and modeling eviews application and example. 2nd ed. beijing: tsinghua university press; 2006. [23] chen c, zou y. spss 15.0 chinese version of the common functions and application examples jingjiang. beijing: publishing house of electronics industry; 2009. [24] jiang y. about the application of arima model in the estimation of guangxi social fixed assets investment. appl stat manage 2006;25:588–92. [25] sun y, jia c, qin s. forecasting the hospitalization expenses of injury children with the arima model. appl stat manage 2007;26:1124–8. [26] clemen rt. combining forecasts: a review and annotated bibliography. int j forecast 1989;5:559–83. [27] yan t, qi s. exploring the source of the holiday-effect of stock market: based on the earnings rate of shanghai composite index and shenzhen component index. reform 2011;124–8. https://doi.org/10.1016/0169-2070(89)90012-5 https://doi.org/10.1016/0169-2070(89)90012-5 research article patterns of fatal explosion accidents in china lei pang1,2, , zhiwen zhang1, kai yang1,2, , pengfei lv1,2, siheng sun1,2,* 1school of safety engineering, beijing institute of petrochemical technology, beijing 102617, china 2institute of chemical hazard detection and control, beijing academy of safety engineering and technology, beijing 102617, china 1. introduction china is a developing country with rapid economic growth. however, large numbers of safety accidents happen every year due to various reasons like unbalanced economic development between provinces and insufficient economic or technical investment. not only have frequent safety accidents casted adverse effects on the rapid development of china’s national economy, but they have also resulted in enormous economic losses. explosions are accidents with high fatality and disability rates among all the safety accidents. for example, there were 493 fatal accidents in china in 2018, 51 or 10.35% of which were explosion accidents. of the 1860 deaths caused, 198 died in explosion accidents, which accounted for 10.65% of the total. according to incomplete statistics, nearly 2000 fatal explosion accidents (fea) that caused about 15,000 deaths were reported in china during 2001–2018. for example, the extraordinarily serious aluminum dust explosion that happened in zhongrong metalwork plant, kunshan, jiangsu province on august 2, 2014 had killed 75 people and wounded 185 others on the day. in august 12, 2015, the extraordinarily serious explosion in tianjin binhai new area had left 165 people dead and 74 others missing. with heavy casualties, explosion accidents have become the costliest disasters in china over the past few years. however, the occurrence of explosion accidents is a combined result of multiple factors. to effectively prevent and control accidents in such a severe situation, it is of great practical significance to study the patterns of fea. many researchers have considered the patterns of accidents by focusing on the circumstances in which accidents are more likely to occur. considering that china is a major coal producer and mine gas explosions take up about 25% of coal mine accidents [1], the characteristics of mine gas explosion accidents in chinese were analyzed using the statistical theory from various perspectives like the number of gas explosions, deaths, ignition source, time of accident and area [2–5]. in a study by zheng et al. [6], a statistical method was used to investigate the 106 fatal coal dust explosions in china during 1949–2007. the statistical characteristics of accidentrelated factors were analyzed, such as space, time, volatilization ratio of coal dust, ignition source and type of accident. wu [7] and li et al. [8] focused on hazardous chemical accidents that occurred in china during “the 11th five-year plan” and “the 12th five-year plan”. data statistics of fireworks explosion accidents that happened in china during 2006–2015 were carried out by zhao et al. [9] by means of clustering analysis of the overall situations and regions of these accidents using statistical product and service solutions (spss), based on which the law of the occurrence of fireworks safety accidents in china was summarized. in a study by li et al. [10], the safety accidents in china during 2010–2014 were counted. a general trend of china’s safety accidents was obtained by them by analyzing the overall situations of accidents, the occurrence trend of serious and extraordinarily serious accidents and the death toll in the safety accidents in various industries. on that basis, some rational countermeasures and suggestions were made. a r t i c l e i n f o article history received 30 april 2020 accepted 10 july 2020 keywords fatal explosion accident pattern statistical analysis cramer’s coefficient a b s t r a c t explosion accidents usually lead to serious fatality and disability. to investigate the patterns of fatal explosion accidents (fea), a database containing 1835 fea that happened in china during 2001–2018 was built. statistical analysis was made based on the year, month, time interval of day, day of week, province, type of accident, place of occurrence and severity of accident. the result shown that, (1) fea were more likely to occur in november; (2) 08: 00-10: 00 and 14: 00-16: 00 were two time intervals of frequent accident occurrence; (3) hazardous chemical fea mostly happened on tuesday and wednesday, whereas pressurized container fea often happened on saturday; (4) coal mine gas fea still accounted for a dominant proportion with the highest fatality rate. in addition, the correlation was measured by calculating cramer’s coefficients, which indicated that the severities and types of fea were significantly correlated with month. to be specific, (1) major and extraordinarily serious fea were more likely to occur in november; (2) most of gas fea were observed in april; (3) gunpowder and dust fea happened most commonly in may; (4) more fireworks fea happened in november. the patterns found in this study will provide explosion accident prevention and control with references. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: jiushiwossh@163.com journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 54–63 doi: https://doi.org/10.2991/jracr.k.200714.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0001-8529-4068 http://orcid.org/0000-0002-0927-8864 http://creativecommons.org/licenses/by-nc/4.0/ mailto: jiushiwossh@163.com https://doi.org/10.2991/jracr.k.200714.001 https://www.atlantis-press.com/journals/jracr%20 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 55 previous studies mostly focus on some certain types of explosion accidents that are generally divided into fatal and nonfatal ones. therefore, it is necessary to delve into fea on a general basis for the sake of accident prevention and control. there are many reasons for explosion accidents, such as safety management, backward technology, lack of preventive measures and so on. to further identify the patterns of fea, the statistics and analysis of explosion accidents in china during 2001–2018 were carried out by probing the statistical information of explosion accidents in this study, thereby providing accident prevention and control with countermeasures and facilitating the implementation of safety supervision. 2. date and methods 2.1. data collection to ensure the authenticity and validity of accident information, the data used in this study were collected from ministry of emergency management of the people’s republic of china (prc) [11], national bureau of statistics of prc [12] and safety management network [13]. data statistics covered a total number of 1858 fea that happened in all 34 provinces, municipalities and autonomous regions of china (including hong kong, macao and taiwan) from january 1, 2001 to december 31, 2018. due to the discontinuity of data statistics in some regions, 1835 explosion accidents that caused 13,930 deaths were considered in this study. in addition, for lack detailed explanation of cause was given from the data source, this paper only analyzes the objective attributes of the accident, such as the year, month, time interval of day, day of week, province, type of accident, etc. 2.2. data collation the fea were sorted and analyzed by the following key factors: (1) the accidents were sorted by geographical location (province, municipality and autonomous region), the year (2001–2018), month (january–december) and day of week (monday– sunday); (2) statistics of the time of fea were given by the specific time interval of day (8: 00-10: 00, 10: 00-12: 00, 12: 00-14: 00, 14: 00-16: 00, 16: 00-18: 00, 18: 00-22: 00 and 22: 00-08: 00); (3) fea were counted by type, namely mine gas explosion (ta1), fireworks explosion (ta2), hazardous chemical explosion (ta3), fuel gas explosion (ta4), pressurized container explosion (ta5), gunpowder explosion (ta6), dust explosion (ta7) and other explosions (ta8); (4) the severity of each fatal explosion accident was determined by death toll. according to regulations on the reporting, investigation and handling of production safety accidents (2007) issued by state council of prc, accident severity can be divided into four levels: minor accidents (sa1), serious accidents (sa2), major accidents (sa3) and extraordinarily serious accident (sa4), which refer to the accidents that cause one death to two deaths, three to nine deaths, 10–29 deaths and over 30 deaths each, respectively. 2.3. data analysis methods frequency analysis was used to analyze the basic characteristics of accident occurrence, followed by the investigation of the correlation between various accident factors using correlation coefficient analysis of spss. whether there was significant difference between two factors was verified by calculating the corresponding p-value. in addition, the correlation between different factors was identified based on cramer’s coefficient v and phi correlation coefficient j. the relevant calculation formula is shown in formulas (1)–(3). c 2 2 = ´ æ è ç ö ø ÷ ´ å ij ij i j i j x x x x x x x (1) j c = 2 x (2) v x a b = ´ c 2 1[ ( , ) ]min (3) a and b represent the number of rows and columns, respectively. 3. result analysis and discussion 3.1. frequency analysis in this section, statistical description of fea was made by the year, month, time of day and day of week, province, type of accident (ta), place of occurrence (po) and severity of accident (sa). 3.1.1. year the annual variation trend of the fea in china during 2001–2018 is shown in figure 1, which indicates that the death toll of explosion accidents declined with fluctuations year by year. during 2001–2018, there were 1835 fea that caused 13,930 deaths in total. the accident frequency, average death toll per accident and average daily death toll were 102, 774 and 2.12, respectively. by contrast, the number of people who died from building construction was 1.57/day on average in china [14]. 2005 was the year with highest death toll from explosion accidents, which left 1830 deaths in total. it can be seen from tables 1 and 2 that 98 (60.5% of the total) of the 162 fea that occurred in 2005 were mine gas explosion accidents; there were 10 extraordinarily serious explosion accidents that caused 843 deaths in the same year, which accounted for 46.1% of the total annual death toll. there were three fea that killed over 100 people each, all of which were mine gas explosion accidents. from a provincial perspective, these three accidents happened in liaoning province, heilongjiang province and hebei province and caused 214, 171 and 108 deaths, respectively. 3.1.2. month the distribution of explosive fatal accidents by month is shown in figure 2. it can be seen that november was the month with the 56 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 figure 1 | distribution of fatal explosive accidents by year. table 2 | summary of severity of explosion accidents in 2005 severity of accident sa1 sa2 sa3 sa4 total fea (no.) 19 105 28 10 162 fatalities (no.) 30 468 489 843 1830 table 1 | summary of types of fea in 2005 type of accident ta1 ta2 ta3 ta4 ta5 ta6 ta7 ta8 total no. 98 22 6 1 11 16 2 6 162 largest number of fea (169) and the highest death toll (1813), february was the month with the fewest fea, and that the month with the lowest death toll was september. there is a persistent febrile period after the weather turns warm from march to august, during which more fea happened and resulted in a higher death toll. one reason is that distractedness and fatigue are more likely to occur in warm weather. continuing high temperature can result in significant physical and mental changes to operators, which casts adverse effects on production safety and thus leads to accidents. however, both the frequency of fea and the death toll as caused were found to slightly decline in june. one possible reason is that china has implemented a “production safety month” activity in june every year since 2002. this activity is aimed to improve universal safety awareness, remove accident potential to the largest extent and thus effectively curb the occurrence of accidents by publicizing a series of guidelines of production safety and popularizing the knowledge on production safety-related laws and regulations. due to the high inflammability, explosibility and volatility of most explosive materials, the dry weather in november makes explosion accidents frequent [15]. in addition, the dry weather tends to fatigue workers, which is also a reason why fatal accidents happen frequently. compared with november, the occurrence frequency of fea and the death toll as caused were both lower in september. one possible reason is that china is planning large-scale celebrations in this month every year as national day approaches, during which the law enforcement on safety will be improved to ensure the risks during the national day is controllable. lower accident frequencies were also observed during major festivals in other countries [16]. it can also be seen that there were fewer explosion accidents in february (a month in winter), which was partly because low temperature is to the advantage of the storage and transportation of explosives and helps prevent them from reaching the temperature of explosion. moreover, spring festival, the most important traditional festival of china, is around february. all units are required to carry out complete year-end safety inspections and employees also start to enjoy this holiday or take working days off before and after spring festival. hence, fea occurred less often in february. 3.1.3. time interval of day according to statistics, the number of explosion accidents differed significantly time intervals of a day and followed a certain regularity (figure 3). in the daytime, explosion accidents happened most frequently during 08: 00-10: 00 and 14: 00-16: 00, during which there were 275 and 256 explosion accidents occurred, respectively. these two time intervals are exactly the time when people start working. in addition, this was also directly associated with people’s routines. most of chinese office workers stay up late, which results in operational errors due to poor performance and inattention. considering that 10: 00-12: 00 and 16: 00-18: 00 are commuting hours when people are distracted for lack of energy due to starvation or fatigue, accidents are more likely to occur. fea still occurred frequently when it came to the time from 22:00 to 08:00 in the next morning. according to the analysis of the accident types during this period (as shown in table 3), mine gas explosion accidents accounted for the largest proportion, accounting for 55.8% of the total. moreover, the mine gas explosions that occurred during this time interval took up 25.9% of the total number of mine gas explosion accidents, most of which happened in small-scale local coal mines [3]. inadequate illumination at night makes it difficult for workers to discover accident potentials, specially, gas leakage that can lead to a fatal explosion accident upon an ignition source. having been working for long, workers are too tired to concentrate themselves on mechanical operation. fea are more likely to happen in the context of operational error or negligence in management. it can also be seen that there were the fewest fea during 12: 01-14: 00 (noontime), which is because people are either taking noon breaks or having lunch with mechanical equipment in the outage state. l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 57 table 3 | summary of type of fea from 22:00-08:00 during 2001–2018 type of accident ta1 ta2 ta3 ta4 ta5 ta6 ta7 ta8 total no. 241 39 25 37 27 35 10 18 432 figure 2 | distribution of fea by month. figure 4 | distribution of fea by day of week. figure 3 | distribution of fea by time interval of day. 3.1.4. day of week as shown in figure 4, the statistics of china’s fea suggest that there were the fewest fea on sunday, which is probably because people do not work on weekends. from the perspective of explosion accident types, mine gas explosions were still the most common accidents; hazardous chemical explosions mostly happened during transportation on tuesday and wednesday. most of the pressurized container explosions occurred on saturday. one reason is that container reliability declines after continuous service for a whole week. another reason is that people take weekend breaks on saturday and sunday, which makes accident hazards difficult to be spotted in time. it is thus recommended that employees give emphasis on pressurized container inspection by the end of every friday. lopez et al. [17] and campolieti and hyatt [18] found that there were more accidents on monday, known as “monday effect”. however, the statistical analysis of china’s fea did not show evident monday effect. neither can people concentrate on work immediately after the weekend break, nor are they flexible enough to identify and handle potential safety hazards. this indeed was part of the reason why accidents happened more frequently on monday. 3.1.5. province the distribution of explosive fatal accidents by province is given in figure 5. geographical difference has led to varied industrial orientations between various provinces and municipalities. for example, coastal provinces and municipalities develop manufacturing industry and trade; shanxi province primarily depends on coal mines; some inland provinces and municipalities give priority to chemical industry. this has resulted in significant difference in the number of fea between various provinces. it can be seen from the figure that the fea in shanxi province, guizhou province and heilongjiang province caused the most deaths, which were 1739 (122 accidents), 1357 (224 accidents) and 1136 (80 accidents), respectively. further analysis of the fea that occurred in these three provinces by type of accident (figure 6) indicated that mine gas explosion was the primary type of fea. specifically, mine gas explosions accounted for 63.9%, 87.5% and 75.0% of the total explosion accidents in these provinces, respectively. this was closely associated with the situations of these provinces. with economic cycle considered, the annual gdps of various chinese provinces during 2008–2017 were counted based on global data (figure 7). these, together with the death toll and the number of accidents included in the database, were used to analyze the relationship between the death rate per 100 million yuan of gdp and death toll. according to analysis result, the death rate per 100 million yuan of gdp of guizhou province (0.0026) was the highest, followed by shanxi province (0.0024) and heilongjiang province (0.0023). one possible reason for the high death rate per 100 million 58 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 figure 5 | distribution of fea by province. figure 7 | distribution of gdp and the fatalities of fea. figure 6 | distribution of fea of shanxi/guizhou/heilongjiang. yuan of gdp of guizhou province is that there is unbalanced systematic development between various regions of this province. they have neglected the danger of safety accidents whilst pursuing gdp growth. in addition, unbalanced social development, poor organizational management and inadequate infrastructure can also contribute to a higher death rate per 100 million yuan of gdp. although the fea had resulted in high death tolls in both hunan province and henan province, their death rates per 100 million yuan of gdp was lowered by higher gdps. from the perspective of economics, development will inevitably invite accidents that lead to economic losses and causalities. thus, handling the relationship between production safety and social economic development to promote sustainable economic development is a long-term issue of safety economics [19]. l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 59 table 4 | distribution of fea by severity of accident severity of accident fea fatalities number percentage (%) cumulative percentage (%) number percentage (%) cumulative percentage (%) sa1 229 12.5 12.5 354 2.5 2.5 sa2 1306 71.2 83.7 6030 43.3 45.8 sa3 242 13.2 96.9 3982 28.6 74.4 sa4 58 3.1 100 3564 25.6 100 3.1.6. severity of accident as shown in table 4, the number of serious accidents occupied a dominant position in the four accident levels and accounted for 71.2% of the total. this indicated that the explosion accidents that caused the most deaths in china were serious accidents with a death toll of 3–9 people each and they contributed 43.3% of the total death toll. it can be seen from the type of accident divided by accident level that serious accidents accounted for a majority in all types of fea (figure 8). the number of extraordinarily serious accidents was 58, which was the smallest and took up only 3.1% of the total. specifically, 79.3% (46) of them were mine gas explosions; there were nine explosion accidents that caused over 100 deaths each. since 2010, only one extraordinarily serious accident had been reported, namely the hazardous chemical explosion that killed 165 people in tianjin in 2015. in addition, no more fea that caused over 30 deaths each were reported in both 2017 and 2018, which had indirectly evidenced that china is paying more and more attention to production safety and the overall safety situation has been improved over the past few years. 3.1.7. type of accident as shown in figure 9, mine gas explosion, fireworks explosion, fuel gas explosion and gunpowder explosion were the top four types of fea. there were 930 mine gas explosion accidents that resulted in 8761 deaths during 2001–2018, which accounted for 50.7% of the total number of fea and 62.9% of total death toll as caused. to be specific, 46 of them were extraordinarily serious accidents and killed 2902 people in total, which took up 33.1% of the total death toll caused by mine gas explosions. however, the number of mine gas explosion accidents showed a generally downward trend [20,21], as shown in table 5. it can thus be figure 9 | distribution of fea by type of accident. figure 8 | distribution of type of fea by severity of accident. 60 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 figure 10 | distribution of fea by place of occurrence. table 5 | distribution of gas fea by year year 2001 2002 2003 2004 2005 2006 2007 2008 2009 fea (no.) 157 156 135 107 98 66 43 27 24 year 2010 2011 2012 2013 2014 2015 2016 2017 2018 fea (no.) 26 22 11 18 14 7 9 9 1 table 6 | distribution of fea by severity of accident and month sa1 sa2 sa3 sa4 sum january 21 115 14 1 151 february 19 67 20 4 110 march 17 108 22 6 153 april 27 110 23 4 164 may 8 134 28 4 174 june 16 124 15 4 159 july 19 118 15 7 159 august 24 103 20 5 152 september 18 86 16 1 121 october 19 112 26 5 162 november 20 106 32 11 169 december 21 123 11 6 161 sum 229 1306 242 58 1835 inferred that strengthening mine gas monitoring, early waring and extraction control is still the focus. moreover, dust explosions have frequently occurred in recent years. for example, five out of the 51 fea reported in 2018 were dust explosions, which caused 16 deaths in total. 3.1.8. place of occurrence it can be seen from figure 10 that a majority of fea happened in coal mines, which accounted for 54% of the total. one reason is that china is a major coal producer with densely distributed coal mines and inadequate mine gas techniques. factories contributed to the second largest number of fea, which took up 17% of the total. one reason for the frequent explosions in factories is that equipment like boilers, reactors and pressurized containers can explode during production and processing in various plants. also, recent years has witnessed a number of explosions of wood dust, grain dust and metal dust. another reason is that some hazardous chemicals are more likely to cause explosions in chemical enterprises due to their unique properties. residential buildings were the third most common places of fea, which is because residents’ safety awareness and their ability to handle fuel gas leakage are still insufficient in the context of rapid fuel gas popularization in cities. for this reason, fea due to fuel gas leakage, tank corrosion and other factors are constantly being reported. in addition, fireworks factories were also places of frequent explosions. this is because spring festival is a traditional chinese festival, during which people will set of fireworks to celebrate it. however, some fireworks manufacturers pursue pecuniary interests at the cost of production safety. this, together with the backward production technology, negligence of safety training, imperfect safety management system and illegal production despite repeated prohibitions, has led to fea in these factories. fewer fea happened in other places, but they still resulted in casualties and property losses. 3.2. correlation coefficient analysis to further identify the patterns of fea, the correlation between the factors of 1835 fatal explosions except year was investigated. to begin with, the correlation analysis was made with severity of accident (sa) as the dependent variable and month day of week (dw), time interval of day (td) and type of accident (ta) as independent variables, followed by another correlation analysis with ta as the dependent variable and month dw and td as independent variables. the contingency table was used in this study. the coefficients of the correlation between various factors are shown in table 6 (confidence interval: 95%). according to the table of r × c (r > 2 and c > 2), the correlation was verified using cramer’s coefficient. if cramer’s coefficient between two factors was larger than 0.1 (p < 0.05), they are considered to be significantly correlated. take the correlation between sa and month as an example. the statistics of relevant accidents are shown in table 6. c 2 2 2 21 151 229 1835 151 229 = ´ æ èç ö ø÷ ´ = ´ æ èç ö ø÷ ´ å ij ij i j i j x x x x x x x 11835 115 151 1306 1835 151 1306 1835 11 161 242 1835 2 + ´ æ èç ö ø÷ ´ +¼ + ´ ææ èç ö ø÷ ´ + ´ æ èç ö ø÷ ´ = = 2 2 2 161 242 1835 6 161 58 1835 161 58 1835 55 4. j c xx v x a b = = = ´ = ´ 55 4 1835 0 1738 1 55 4 1835 4 12 1 2 . . [ ( , ) ] . [ ( , ) c min min ]] .= 0 1003 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 61 figure 11 | cumulative percentage of sa vs. month. figure 12 | cumulative percentage of ta vs. month. c 2 2 2 21 151 229 1835 151 229 = ´ æ èç ö ø÷ ´ = ´ æ èç ö ø÷ ´ å ij ij i j i j x x x x x x x 11835 115 151 1306 1835 151 1306 1835 11 161 242 1835 2 + ´ æ èç ö ø÷ ´ +¼ + ´ ææ èç ö ø÷ ´ + ´ æ èç ö ø÷ ´ = = 2 2 2 161 242 1835 6 161 58 1835 161 58 1835 55 4. j c xx v x a b = = = ´ = ´ 55 4 1835 0 1738 1 55 4 1835 4 12 1 2 . . [ ( , ) ] . [ ( , ) c min min ]] .= 0 1003 other data calculation results are shown in table 7. it can be seen that there was significant correlation between sa and month (v = 0.1003, p < 0.05) and between ta and month (v = 0.1052, p < 0.01). it can also be seen from table 6 that the sig. values p of the correlation between sa and dw, td and ta were larger than 0.05, so there were no significant difference between them, indicating that the severity of a fatal explosion accident was uncorrelated with dw, td and ta. likewise, the sig. values p of the correlation between ta and td was also larger than 0.05 and there was no significant difference between them. although the p-value of the correlation between ta and dw was smaller than 0.01, the cramer’s coefficient was 0.0740, which indicated that there was no significant correlation between them. thus, the type of a fatal explosion accident was uncorrelated with dw and td. to obtain more useful information from the factors of significant correlation, the internal relations between sa and month and between ta and month of the 1835 fea were investigated. the relationship between sa and month is shown in figure 11. as shown in figure 11, major and extraordinarily serious accidents were more likely to occur in november. statistics suggested that there were 32 major fea and 11 extraordinarily serious fea in november, which accounted for 13.2% and 19.0% of the totals, respectively. the relationship between ta and month of fea is presented in figure 12, which indicates that there were more fuel gas explosions table 7 | contingency table for correlation coefficients between factors dependent factor independent factor chi-square df. sig. p phi i cramer’s v sa month 55.4 47 0.002** 0.1738 0.1003 dw 24.13 27 0.570 0.1147 0.0662 td 29.53 27 0.304 0.1269 0.0732 ta 109.8 31 0.349 0.2446 0.1412 ta month 142.18 95 0.000** 0.2784 0.1052 dw 60.29 55 0.009** 0.1813 0.0740 td 42.63 55 0.292 0.1524 0.0622 *p < 0.05, **p < 0.01. 62 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 in april. 90% of fuel gas explosions were caused by improper use, valves and rubber hoses, etc. gunpowder and dust explosions occurred more frequently in may because gunpowder and explosive dust may generate heat upon moisture and are thus more likely to explode as air humidity rises when the rainy season starts from late april to may every year. in november, fireworks explosions were constantly reported since november is the peak season of fireworks production and sales by the end of the year [9,22]. workers in fireworks manufacturers are more prone to drop their guard as they are busy fulfilling the production targets, which can result in explosion accidents. in addition, many unlicensed small workshops recklessly disregard production safety, leading to an increase an increase in the number of fatal accidents caused by fireworks explosion. 4. conclusion and suggestions explosion accidents feature high fatality and disability rates due to their intrinsic danger. this study is aimed to investigate the patterns of fea that happened in practice in china, thereby providing the prevention and control of similar accidents with some references. according to the research above, the following conclusions can be reached: frequency statistics suggest that: (a) november is the month with the largest number of fea and the highest death toll as caused. thus, regulators need to strengthen daily supervision, especially safety inspections on fireworks manufacturers upon the spring festival. after the festival, however, it is advisable that safety managers take measures to restore the post awareness of workers from the festival atmosphere as soon as possible; (b) in terms of work hours, 08: 00-10: 00 and 14: 00-16: 00 are two time intervals when explosion accidents are more likely to happen. hence, it is critical the enterprise safety managers hold pre-shift mobilization meetings to strengthen employees’ safety consciousness; (c) although there is no typical “monday effect” in fea, it is still necessary for enterprises to carry out safety training once on every monday before work; (d) in addition, most of the hazardous chemical explosions occur on tuesday and wednesday and pressurized container explosions mostly on saturday; (e) fatal mine gas explosions still account for a dominant proportion in fea, which is especially true in guizhou province, shanxi province and heilongjiang province. for this reason, coal mine supervision should be strengthened; (f ) discussion from the place of occurrence, a majority of fea happened in coal mines, accounted for 54% of the total. the government still needs to strengthen the supervision of coal mines, promote the coal mine gas collection and control technology research, to reduce the occurrence of explosion accidents. the correlation analysis shown that: there is significant correlation both between ta and month (cramer’s v = 0.1003, p < 0.05) and between sa and month (cramer’s v = 0.1052, p < 0.01) of fea, whereas other factors are not significantly correlated. major and extraordinarily serious accidents are more likely to occur in november, fuel gas explosions in april, gunpowder and dust explosions in may and fireworks explosions in november. therefore, safety supervisory authorities need to carry out specific rectification to ensure production safety. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments the authors appreciate the financial support from the beijing natural science foundation-municipal education committee joint funding project (no. kz201910017020), the beijing science and technology nova program (no. z181100006218092), and the training funded project of the beijing youth top-notch talents of china (no. 2016000026833zk05). references [1] xiao s. causation analysis method and application research of gas explosion accidents in coal mines. beijing, china: china university of mining & technology; 2018. [2] zhu y, wang d, shao z, xu c, zhu x, qi x, et al. a statistical analysis of coalmine fires and explosions in china. proc safety environ protect 2019;121:357–66. [3] yang cl, li xc. reason and statistical rule of especially serious gas and coal dust explosion accidents in coal mine. coal technol 2015;34:309–11. [4] zhang j, david c, xu k, you g. focusing on the patterns and characteristics of extraordinarily severe gas explosion accidents in chinese coal mines. process safety environ protect 2018;117:390–8. [5] zhou x, jiang mx. statistical analysis and safety management on china’s coal mine gas accident from 2006 to 2015. ieti trans bus manage sci 2016;1:29–38. [6] zheng yp, feng cg, jing gx, qian xm, li xj, liu zy, et al. a statistical analysis of coal mine accidents caused by coal dust explosions in china. j loss prevent process ind 2009;22:528–32. [7] wu z, zhang s, zhang y, shi c, liu n, yang gl. statistical analysis of hazardous chemicals accidents occurring in china during 2006-2010. j safety sci technol 2011;7:5–9. [8] li j, feng y, yu y, yu l, wu j. statistics and suggestions on hazardous chemicals accidents in china during the twelfth five-year plan. j wuhan univ technol (inf manage eng ed) 2016;38:543–7. [9] zhao ch, jiang lr, fei q, li y. statistical analysis and prediction of fireworks accidents in china based on spss software. fire tech prod inf 2018;31:10–14. [10] li l, liu x, kang h, yan ld. the analysis of safety production situation in china based on the 2010-2014 accident statistics. j changchun inst technol (nat sci ed) 2015;16:77–9. [11] ministry of emergency management of the people’s republic of china. investigation report on extremely serious accident. available from: https://www.mem.gov.cn/gk/sgcc/tbzdsgdcbg/ [12] national bureau of statistics of china, national data. annual gdp of china’s provinces and cities from 2008 to 2017. available from: http://data.stats.gov.cn/easyquery.htm?cn=c01 [13] security management website. case analysis of explosion accident. available from: http://www.safehoo.com/case/case/blow/. [14] shao b, hu z, liu q, chen s, he w. fatal accident patterns of building construction activities in china. safety sci 2019;111:253–63. https://doi.org/10.1016/j.psep.2018.11.013 https://doi.org/10.1016/j.psep.2018.11.013 https://doi.org/10.1016/j.psep.2018.11.013 https://doi.org/10.1016/j.psep.2018.05.002 https://doi.org/10.1016/j.psep.2018.05.002 https://doi.org/10.1016/j.psep.2018.05.002 https://doi.org/10.1016/j.jlp.2009.02.010 https://doi.org/10.1016/j.jlp.2009.02.010 https://doi.org/10.1016/j.jlp.2009.02.010 http://www.safehoo.com/case/case/blow/ https://doi.org/10.1016/j.ssci.2018.07.019 https://doi.org/10.1016/j.ssci.2018.07.019 l. pang et al. / journal of risk analysis and crisis response 10(2) 54–63 63 [15] duo yq, liu yn, hu xs. statistical analysis on dust explosion accidents occurring in china during 2009-2013. j safety sci technol 2015;11:186–90. [16] amiri m, ardeshir a, zarandi mhf. risk-based analysis of construction accidents in iran during 2007-2011-meta analyze study. iran j pub health 2014;43:507–22. [17] lópez arquillos a, rubio romero jc, gibb a. analysis of construction accidents in spain, 2003-2008. j safety res 2012;43:381–8. [18] campolieti m, hyatt de. further evidence on the “monday effect” in workers’ compensation. ilr rev 2006;59:438–50. [19] luo y, tian sc. safety economics. beijing: journal of henan polytechnic university, china lab soc secur press; 2010. [20] liu yz, zhang lh. statistical analysis of gas death accidents in coal mines from january 2003 to june 2005. j henan polytech univ 2005;24:259–62. [21] li x, xu m. statistics and analysis of domestic coal and gas outburst accidents in recent years and prevention measures. mining mach 2010;38:13–6. [22] yang n. statistical analysis on firework accidents occurring in china during 2007-2011. j safety sci technol 2013;9: 72–7. https://pubmed.ncbi.nlm.nih.gov/26005662/ https://pubmed.ncbi.nlm.nih.gov/26005662/ https://pubmed.ncbi.nlm.nih.gov/26005662/ https://doi.org/10.1016/j.jsr.2012.07.005 https://doi.org/10.1016/j.jsr.2012.07.005 https://doi.org/10.1016/j.jsr.2012.07.005 atlantis press journal style received 31 august 2016 accepted 23 september 2016 cfd analysis of oxygen and carbon dioxide recovery during ventilation in municipal confined space yan liu, cong tan, dongliang liu beijing municipal institute of labour protection, beijing 100054, china e-mail: liuyandhn@163.com abstract confined space is a special working place. to ensure the safety, the forced ventilation must be carried out before entering the confined space. in this paper, ventilation process in municipal heating confine space is simulated by cfd numerical simulation method, and the distribution of temperature, oxygen and carbon dioxide in municipal heating confined space during ventilation process is studied. the simulation results show that the airflow and temperature are obviously stratified in vertical direction during the ventilation. the temperature is low and the airflow velocity is high at the bottom part of the tunnel. while the temperature is high and the airflow velocity is low at the top of the tunnel. by contrast, recovery of the oxygen and carbon dioxide content is relatively faster at the bottom than it at the top. when the ventilation volume flow rates are 6000 m3•h-1, 8000 m3•h-1,10000 m3•h-1, 12000 m3•h-1, and 14000 m3•h-1 respectively, the time when the temperature reach stable and gas fraction return to normal are 45 minutes, 32 minutes, 22 minutes, 15 minutes, 10 minutes respectively. keywords: cfd; numerical simulation; municipal confined space; ventilation; oxygen content 1. introduction according to the definition of the occupational safety and health administration of the united states, the confined space is defined as1: (1) is large enough and so configured that an employee can bodily enter and perform assigned work; (2) has limited or restricted means for entry or exit; and(3) is not designed for continuous employee occupancy. municipal heating pipe tunnel belongs to confined space. because of longterm closure and poor natural ventilation, worker will suffer hypoxia phenomenon when performing some operations in it2. many countries focus on the safety of the confined space operations. the united states3, britain4, france5, australia6 and other countries have developed a number of standards for confined space operations. niosh has made a thorough research on the safety of the confined space, and published a series of safety publications about the confined space operations. for example, criteria for a recommended standard, working in confined spaces7, request for assistance in preventing occupational fatalities in confined spaces 8and a guide to safety in confined spaces 9, which play a positive role in the prevention of the confined space accident. great attention is also paid to the safety of the confined space working in china. gao10 ,fan11, zhong 12, chen 13 and other researchers have carried out some research on the hazard identification, accident prevention, control measures, supervision and management of the confined space. the current research results show that there may be hydrogen sulfide, ammonia, methane and other toxic, flammable or explosive gas in the atmosphere of confined space, and oxygen deficiency in very common. oviedo14 studied concentration and distribution of hydrogen sulfide in sewer system. j m santos15, v. blanes-vidal16 studied the production rate of nh3, co2, journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 178-185 published by atlantis press copyright: the authors 178 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery h2s respectively. mi 17, peng18 studied the accumulation of combustible gas in sewage in the mechanism, and the gas evaluation model of explosion risk of city sewers and septic tanks as well as the warning model of gas explosion are built. wang, bao19 designed a remote monitoring and early warning system for municipal sewer harmful gas based on infrared sensor technology and gprs wireless data transmission so as to realize the function of real-time monitoring and monitoring of the concentration of combustible gas in municipal sewage. confined space operation procedure issued by the united states occupational safety and health department requires that before entering the confined space, the concentration of oxygen, toxic gases and flammable gas must be detected. also, no one is accessed to entering the confined space before eliminating the harmful gas by forced ventilation. standard titled interim provisions on the safety management and supervision of the limited space operation of industry and trade enterprises issued in 2013 by state administration of work safety puts forward that the confined space operations should strictly abide by the principle of "ventilation first, reinspection second, and operations last", which is reemphasized in standard titled five provisions on the safety operation of confined space issued in 2014 requiring that no operation should be carried out if there is no ventilation or monitor result are not up to standard. in conclusion, the forced ventilation is an essential step to ensure the safety of the workers before entering the confined space. cfd numerical simulation method has been widely used in the study of gas flow and mixed gas distribution. m.siddiqui20 studied risk control and assessment of indoor heavy gas leakage using cfd method. s.g. giannissi21 studied the hydrogen leakage and diffusion in the confined space under the condition of natural ventilation using cfd method. zhao22 optimized ventilation scheme in the septic tank confined space using cfd method. in this paper, the process of forced ventilation in heating confined space is simulated using cfd method. and the recovery law of temperature, oxygen, and carbon dioxide under different ventilation condition is obtained, which can provide theoretical basis and technical support for forced ventilation before operation. 2. basic equations of numerical simulation because the maximum air flow speed in the process of ventilation is far lower than the speed of sound (340m·s1), so the air in municipal heating confined space can be regarded as an incompressible fluid. and the ventilation process belongs to unsteady single-phase multicomponent diffusion issues without chemical reaction, so the flow is governed by the conservation of mass, momentum, and energy balance and species transmission equation in cfd. the multi-component three-dimensional unsteady turbulent flow is described by equations (1), (2), (3), (4) and (5) respectively. mass conservation equation (also known as continuity equation): ( )i i 0u t x ρ ρ ∂ ∂ + = ∂ ∂ ; (1) momentum balance equation: ( ) ( ) iji i j i i j i j p u u u g f t x x x t ρ ρ ρ ∂∂ ∂ ∂ + = − + + + ∂ ∂ ∂ ∂ ; (2) energy equation: ( ) ( ) ( ) ref i i c pc i ij heff i i eff t t e u e p t x t k j c dt u s x x ρ ρ t ∂ ∂ + + =  ∂ ∂  ∂ ∂ − + +  ∂ ∂  ∑ ∫ ; ref 2 c pc 2 t t v e c dtω= +∑ ∫ (3) species transmission equation: ( ) ( ) cc i c ii i cu dt x x x ω ρω ρ ω ρ  ∂∂ ∂ ∂ + =   ∂ ∂ ∂ ∂  ; (4) k-ε turbulence equations: k equation: ( ) ti k i i k i k u k g x x x µ ρ µ ρε σ   ∂ ∂ ∂ = + + −   ∂ ∂ ∂   ; ε equation: ( ) 2 t ε1 i k ε2 i i ε i ck u g c x x x k k µ ε ε ρ ε µ ρ σ   ∂ ∂ ∂ = + + −   ∂ ∂ ∂   (5) where 2 t μ k cµ ρ ε = , j jik t i j i u uu g x x x µ  ∂ ∂∂ = +  ∂ ∂ ∂  . here, ρ is the gas density, kg·m-3; v is the fluid velocity, m·s-1;uiis the velocity component in the ith direction, m·s-1; xi is coordinate in the ith direction, m; p is the fluid pressure, pa; τij is the stress tensor, n·m -2; gi is the component of gravitational acceleration in the ith direction, n·kg-1; fi is external volume force in the ith direction, n; e is a fluid energy, j;keff is the effective published by atlantis press copyright: the authors 179 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery thermal conductivity; jc is the diffusion flux of component c; sh is the volume heat source; ωc is the mass fraction of component c ; dc is the diffusion coefficient of component c; cpc is the specific heat capacity of component c, j·(kg·k)-1; k is the turbulent kinetic energy, m2·s–2; ε is the turbulent kinetic energy dissipation rate, m2·s-3; gk is the rate of change of turbulent kinetic energy, μvis the viscosity coefficient of laminar flow, μtis the coefficient of turbulent viscosity, pa·s, cε1, cε2, cμ, σε, σk, tref are constants, taken as 1.44, 1.92, 0.09, 1.3, 1.0, 298.15k. 3. numerical simulation model 3.1 geometric model a heating confined space in beijing is selected to do the simulation. the heating chamber is 6 meters long, 5 meters wide and 6 meters high. the distance from the roof of the chamber to the floor (i.e. shaft deep) is 4 meters. two entrance shafts locate at the diagonal of the chamber, and below the shaft, there is a metal platform and a metal ladder. there are arch pipe ditch which is 3 meters wide and 2.5 meters high on both sides of the chamber. in the pipe tunnel, there is a ventilation shaft with the diameter of 0.7 m which is 100 meters away from the chamber. two diameters 0.8m heating pipes (including thermal insulation layer) are built along the pipe tunnel. in the chamber, the heating pipe is divided into two branch pipes with the diameter of 0.6m which pass through the side chamber. and the valves are installed at the interface. because the platform, the ladder handrail, and the value structure of the pipeline are small, and they have little influence on the distribution of the flow field of the chamber, they are ignored to simplify the model and reduce the calculation. according to the actual situation, a threedimensional geometric model is built, as is shown in fig.1. the origin (0, 0, 0) of the model is set at the center of the heating chamber floor. the x axis is along the tunnel, the y axis is the horizontal direction vertical towards the tube groove and the vertical upward is set as the z axis. 9 observation points are set and the coordinates are v1(0,0,1),v2(0,0,3),v3(0,0,5.9), v4(50,0,0.5), v5(50,0,1),v6(50,0,1.9),v7(99,0,0.5), v8(99,0,1), and v9(99,0,1.9). fig.1. geometric model of heating confined space. 3.2 simulation parameters in the boundary conditions setting, the ventilation rates are set as 6000, 8000, 10000, 12000 and 14000m3·h-1 respectively. so, according to area of the inlet, velocity inlet is set with the air flow velocity 4.33, 5.77, 7.22, 8.66 and 10.1m·s-1 respectively. outflow is set in the exit. the surface of the heating pipeline is set as the wall of heat source. and the temperature is 60℃. original gas content, internal temperature and other parameters are set according to the actual results of the field trial, which are shown in table 1. table 1. simulation parameters settings. name parameter settings solving type unsteady energy model on turbulence model k-ε two-equation model component model component delivery model the operating temperature (k) 288 acceleration of gravity (m·s-2) z-axis direction: -9.81 operating pressure (pa) 101325 type inlet boundary speed inlet inlet velocity (m·s-1) 4.33, 5.77, 7.22, 8.66, 10.1 inlet species fraction o2:20.9%; n2:78%; co2: 0.03%; h2o:0.03%; other:1.07% hydraulic diameter (m) 0.7 turbulence intensity (%) 2.6 outlet boundary type wall temperature (k) outflow 333 pressure velocity coupling simplec algorithm the initial value o2:12%; n2 81%;co2: 2%; h2o:2.5%; other: 2.5% time step (s) 1 number of steps 3600 convergence precision 0.00 1 the geometric model is imported to cfd software, and the parameters are set according to the table 1. the parallel computation is carried out using 32-nuclear published by atlantis press copyright: the authors 180 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery work station. the result shows that the residual value (including continuous residual value, velocity residual value, energy residual value, k residual value, ε residual value, and residual values of all species fraction) is less than 0.001. therefore, the calculation results can reach the convergence requirements. 4. numerical simulation results 4.1 distribution of flow field in the process of ventilation, the distribution of flow field has a direct influence on the ventilation effect. at the beginning of the ventilation, the internal stationary gas starts to flow. the original gas distribution has great influence on the flow field. in order to avoid this effect, the flow field after 10 minutes ventilation (according to the simulation results, in the tunnel, the original gas distribution has no longer effect on the air flow field after 10 minutes) was selected to analyze. for example, when the ventilation volume flow rate is 6000 m3·h-1, velocity vector diagram of airflow and airflow velocity contour of x cross section are shown in fig. 2 and fig. 3 respectively. as is shown in fig.2 and fig.3, fresh air is driven vertically into the chamber when ventilation begins. due to the block of the platform, the airflow diffuses to the central part of the chamber. the airflow forms a large eddy current in the chamber and the airflow velocity distribution is extremely uneven. the airflow at the central chamber is fast, and velocity can reach above 1m·s-1.however, the airflow at the top and bottom of the chamber is slow, and velocity is only 0.1 m·s-1. airflow direction change rapidly when the air flows from the chamber to the tunnel, and vortex is formed at the top of the pipe tunnel .therefore, some airflow reverse at the top of tunnel within 10m away from chamber. airflow is relatively stable in the tunnel beyond 10m. air flows mainly along the bottom of the tunnel to both sides of the ventilation shaft. the airflow velocity is less than 0.1 m·s-1at the bottom of the tunnel. hot air rise and accumulate at the top of the tunnel according to temperature field, which make air flow faster at the bottom of the tunnel than it at the bottom. the distribution of airflow velocity field directly affects the recovery time of oxygen content. therefore, much attention should be paid to the oxygen content in the corner of both top and bottom of the chamber, as well as that in the top of the tunnel. fig.2. velocity vector of airflow in tunnel. fig.3. airflow velocity contours of different cross sections along the pipe tunnel. 4.2 distribution of temperature the temperature of hot water or hot air inside the heating pipeline is as high as 140℃.although measures of heat preservation and heat insulation are used, heat is still transferred into the pipe ditch space. surface temperature of the thermal insulation layer is about 60℃ according to measurement. when the ventilation volume flow rate is 6000 m3·h-1, temperature in different sections in the tunnel and chamber after 10mins ventilation is shown in fig.4 and temperature change of different observation points is shown in fig.5. in fig.4, temperature decreases fast in the chamber during the process of ventilation. and temperature at middle area of the chamber is lower compared with it at the top and the bottom. that is because there is a turbulent area at the top of the chamber, and air flows more slowly at the top area than it at other area, which makes temperature decrease more slowly at the top area. at the bottom of the chamber, the heating pipe release heat continuously, so the air temperature around the pipe is relatively high, but the area is small. temperature at the top of the tunnel is higher than it at the bottom of the tunnel. temperature at the area above the heating pipe is 11℃ higher than that below the heating pipe. high temperature area around the heating pipe is large. due to the effect of thermal lift, hot air rises up and accumulates at the top. also, the distribution of temperature field also influences the published by atlantis press copyright: the authors 181 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery distribution of airflow field. the air flows faster at the bottom compared with it at the top, which further aggravates the uneven distribution of the temperature field. fig.5 shows the temperature at each observation point decreases to a stable value after 25 minutes when the ventilation volume flow rate is 6000 m3·h-1. thereafter temperature in the space reaches a steady state and temperature at the observation point （50，0 ， 1.9 ） is the highest, which is about 49 ℃ . temperature at the obviation point （0，0，3）in the chamber is relatively low, which is 27℃. fig.4. temperature contours of different sections along the pipe tunnel. fig.5. temperature of different observation points as a function of time. 4.3 distribution of oxygen when ventilation volume flow rate is 6000 m3·h-1, oxygen distributions of different cross sections after 10mins ventilation in the chamber and the pipe tunnel are shown in fig.6. oxygen content distributions along the tunnel of different heights are shown in fig.7. oxygen content time histories of different observation points are shown in fig.8. fig.6. oxygen content contours of different cross sections along the tunnel. fig.7. oxygen content curves along x-direction line at different height from the tunnel floor. fig.8. oxygen content histories of different observation points as a function of time. fig.6 shows that oxygen content distribution law is the same with the airflow velocity distribution during the ventilation process. the oxygen content in the chamber recovers rapidly and distributes evenly. however, the oxygen distribution of the cross section in the tunnel is different. the oxygen content is high at the bottom of the chamber and is low at the top. in fig.7, oxygen content along the x-direction line 0.5m height from the floor presents linear decrease. the oxygen content decreases and then increases with the increase of the height from the floor. especially along the line z=1.9 m, 65 m meters from the chamber, the oxygen content is the lowest (15.2%). the distribution of published by atlantis press copyright: the authors 182 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery oxygen content along the line illustrates that the thermal lift of the original hot air in the pipe tunnel affects the distribution of flow field, and thereby affects the recovery of oxygen content. fig.8 shows that the recovery of the oxygen content is relatively fast at the beginning of the ventilation, and then gradually becomes slow till oxygen content returns to normal (20.9%). after 10mins ventilation, the oxygen content at the observation point（0,0,5.9）reaches 19.5%( value below this is hypoxia). and 40mins later, it reaches 20.9%. the recovery of the oxygen content at the observation point（50,0,1.9）in the middle part of the pipe tunnel is slow. it takes 20mins to reach 19.5% and 40mins to reach 20.9%. it can be concluded that even though the recovery rate of oxygen content is different between the pipe tunnel and the chamber, the time it takes to reach the normal value (20.9%) is almost the same. 4.4 distribution of carbon dioxide when ventilation volume flow rate is 6000 m3·h-1, carbon dioxide content contours of different cross sections after 10mins ventilation in the chamber and the pipe tunnel are shown in fig.9. carbon dioxide content distributions along the tunnel of different heights are shown in fig.10. carbon dioxide content time histories of different observation points are shown in fig.11. fig.9 shows that carbon dioxide content is high at the top of pipe tunnel and is low at the bottom during the ventilation process, which is opposite to oxygen distribution. fig.10 shows that carbon dioxide content at height 0.5m away from the floor increases linearly along the tunnel. however carbon dioxide content increases at first and then decreases significantly with the increase of height from floor. in fig.11, carbon dioxide content decreasing rate of each observation point in the chamber and the pipe tunnel is different, but the time it takes to reach the normal value of each observation point is the same. furthermore, carbon dioxide recovery time and oxygen recovery time is the same. fig.9. carbon dioxide content contours of different cross sections along the tunnel. fig.10. carbon dioxide content curves along x-direction line at different height from the tunnel floor. fig.11. carbon dioxide content histories of different observation points as a function of time. 4.5 influence of the ventilation volume flow rate on the ventilation effect in order to study change of temperature, oxygen, carbon dioxide along with time, the ventilation processes of different volume flow rate (6000m3·h-1, 8000m3·h1,10000m3·h-1,12000 m3·h-1 and 14000 m3·h-1) are simulated. the observation point v6 （ 50,0,1.9 ） is selected to analyze,because at this point, the decrease of the temperature as well as the recovery of carbon dioxide and oxygen is slow. the temperature, carbon dioxide and oxygen content time histories are shown in fig.12, fig.13 and fig.14 respectively. published by atlantis press copyright: the authors 183 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery fig.12. temperature histories of different ventilation flow rate as a function of time. fig.13. oxygen content histories of different ventilation flow rate as a function of time. fig.14. carbon dioxide content histories of different ventilation flow rate as a function of time. as can be seen from fig.12, with the increase of ventilation flow rate, temperature in the tunnel decreased more quickly, and the final temperature is lower. in this heating confined space, when the ventilation flow rate is set as 6000m3·h-1, 8000m3·h-1, 10000m3·h-1, 12000m3·h-1 and 14000 m3·h-1, it will take 25 mins, 18 mins, 15 mins, 12 mins and 10 mins respectively to reach steady state,and the final temperature is 48 ℃ , 46 ℃ , 43 ℃ , 40 ℃ and 37 ℃ respectively. fig.13 and fig.14 show that time of oxygen to reach normal content is the same with time of carbon dioxide to reach normal content. that is, in the process of ventilation, it will take the same of time of low content gas recovery and high content of gas decreased. in this heating confined space, recovery time of all simulated gas of the several ventilation volume flow rate are 45mins, 32mins, 22mins, 15mins and 10mins respectively. considering time to reach stale temperature, recovery time of oxygen content and carbon dioxide, the internal temperature of steady state and so on, when setting the ventilation volume flow rate as 6000m3·h-1, 8000m3·h-1, 10000m3·h-1, 12000 m3·h-1 and 14000 m3·h-1, the ventlilation time should not be less than 45mins, 32mins, 22mins, 15mins and 10mins respectively. 5. conclusion  because cfd numerical simulation method can reflect the process of ventilation intuitively and comprehensively, it can be used to simulate the process of ventilation in confined space. and the result can provide theoretical and technical supports for actual works.  distributions of the airflow fields during ventilation are affected by both the internal structures and the temperature fields in the municipal heating confined space.with the effect of the two factor, air flows fast at the bottom of pipe tunnel and slow at the top.  distributions of the airflow fields and the temperature fields can affect each other during the process of ventilation, which cause higher temperature at the top of the pipe tunnel and lower temperature at the bottom.  because time to reach normal oxygen content and time to reach normal carbon dioxide content is the same during ventilation, oxygen content can be used as the only index to evaluate ventilation effects. and it is better to use oxygen content close to the floor in the middle of trench as major parameter because recovery time is longer in the section.  ventilation times of municipal heating confined space should not be less than 45, 32, 22, 15, 10 min when the ventilation volume flow rate are 6000, 8000, 10000, 12000, 14000 m3·h-1. safety factor should be multiplied in practical application. published by atlantis press copyright: the authors 184 y. liu et al. / cfd analysis of oxygen and carbon dioxide recovery 6. analysis and discussion more and more attention is paid to the municipal confined space work safety. relevant laws and regulations are more and more perfect. and forced ventilation before entering confined space is an essential step. however, at present, the ventilation method and ventilation time are based on experience. the lack of theoretical basis not only affects the work efficiency, but also cannot guarantee the safety of the workers. in this paper, a numerical simulation method is used to simulate the process of forced ventilation in a heating confined space. the distribution of temperature, oxygen content and carbon dioxide content in the heating confined space is obtained. the recovery rules of internal temperature, oxygen and carbon dioxide under different ventilation conditions are further studied, and the optimal ventilation time of different air volume is determined. the research results can provide theoretical basis and technical support for the forced ventilation of heating confined space. however, there are still some deficiencies in the research of this paper. the section size of the pipe, the length of the pipe tunnel, the setting position and the number of ventilation shaft will affect ventilation. the author will make further research in the future work. acknowledgements this paper is supported by the beijing natural science foundation (no.8152014), funding for postdoctoral scientific research activities in beijing (no.2016zz-82). references 1. osha, permit‐required confined spaces. (washington, 1998). 2. d.burlet-vienney, y.chinniah, a.bahloul, et al.,occupational safety during interventions in confined spaces,safety sci.79(2015)19-28. 3. 29 c.f.r., 1910.146, permit-required confined spaces for general industry. (washington, d.c.,1993) 4. government of united kingdom, the confined spaces regulations 1997. (1997). 5. institut national de rechercheet de sécurité (inrs), espacesconfinés – guidepratique de ventilation n8, ed 703. confined spaces – practical guide for ventilation – n8. (inrs, paris, 2010). 6. standards australia, safe working in a confined space. (sydney, 2001). 7. niosh, criteria for a recommended standard: working in confined spaces. (dhhs publication, morgantown, 1979), pp.80-106. 8. niosh, request for assistance in preventing occupational fatalities in confined spaces. (dhhs publication, morgantown , 1986),pp. 86-110. 9. niosh, a guide to safety in confined spaces. (dhhs publication, morgantown,1987) ,pp.87-113. 10. j.m.gao, l.j.wei. studieson, hazard identification and risk control of confined space, industrial safety and environmental protection. 6(2005)63-54. 11. y.h. fan, s.k. wang, g.c. chen, et al., prevention of asphyxiation and poisoning accidents when working in confined space, j safety sci of china.16(5) (2006)89-93. 12. y. y. zhong, the study on preventive measures of hydrogen sulfide poisoning accidents in confined working space.(shanghai jiao tong university, shanghai, 2009). 13. b. chen, h. l. wang, j. f. zhang, et al., generating mechanism and control countermeasures of explosive gas in urban drainage limited space, journal of safety science and technology. 07(5) (2011)67-71. 14. e r oviedo, d johnson, h shipley, evaluation of hydrogen sulfide concentration and control in a sewer system, environmental technology.33(10-12) (2010)1207-1215. 15. j m santos, v kreim, j m guillot, et al., an experimental determination of the h2s overall mass transfer coefficient from quiescent surfaces at wastewater treatment plants, atmospheric environment. 60(11)(2012)18-24. 16. v blanes-vidal, m guàrdia, x r dai, et al., emissions of nh3, co2 and h2s during swine wastewater management: characterization of transient emissions after air-liquid interface disturbances, atmospheric environment. 54(2012) 408-418. 17. li mi, study on explosion risk assessment and early warning mechanism of city sewers and septic tank gas. (chongqing university, chongqing, 2010). 18. shujuan peng. study on risk assessment and early warning of gas explosion in the urban sewage pipeline in mountainous area. (chongqing university, chongqing, 2012). 19. liang bao. development and application of the monitoring and early warning system of city sewer and septic tank combustible gas. (chongqing university, chongqing, 2009). 20. m.siddiqui, s.jayanti, t.swaminathan, cfd analysis of dense gas dispersion in indoor environment for risk assessment and risk mitigation, j hazard materials. 209–210(1) (2012):177-185. 21. s.g.giannissi, v.shentsov, d.melideo, et al., cfd benchmark on hydrogen release and dispersion in confined, naturally ventilated space with one vent. int j hydrogen energy. 40(5) (2015)2415-2429. 22. j.zhao, manbeck h b, murphy d j, computational fluid dynamics simulations of gas evacuation and o2 recovery times for fan-ventilated confined-space manure pits, studies in foreign education.51(6) (2008)2135-2149. published by atlantis press copyright: the authors 185 1. introduction 2. basic equations of numerical simulation 3. numerical simulation model 3.1 geometric model 3.2 simulation parameters 4. numerical simulation results 4.1 distribution of flow field 4.2 distribution of temperature 4.3 distribution of oxygen 4.4 distribution of carbon dioxide 4.5 influence of the ventilation volume flow rate on the ventilation effect 5. conclusion 6. analysis and discussion << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice journal of risk analysis and crisis response vol. 10(1); april (2020), pp. 23–25 doi: https://doi.org/10.2991/jracr.k.200415.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article the diamond princess cruise: an accidentally experimental model of covid-19 liyuan liu* college of life science, beijing normal university, beijing 100875, china a r t i c l e i n f o article history received 10 march 2020 accepted 28 march 2020 keywords diamond princess cruise sars-cov2 covid-19 viral pneumonia infecting model residential building infection quarantine evacuation 关键词 钻石公主号邮轮 sars-cov2 covid-19 传染模型 居民楼传染 隔离 疏散 a b s t r a c t on february 2020, the diamond princess cruise, which left japan, was obliged to stay at sea for 14 days-quarantine after it was found that a hong kong passenger had been diagnosed sars-cov2 pneumonia after disembarking. as a result shocked the world, an outbreak of infection occurred on board; the number of infected people increased rapidly; and about 20% of the population infected. finally, the quarantine is completely failed and has to be ended officially. the crowd was evacuated back to their own country. the incident can be seen as an experimental model of virus infection in an independently enclosed building, showing powerful air-borne transmission of the virus to pass through public ventilation ducts and crevices in doors and windows. there have trend that no is immune; there have no racial differences and all world people are generally susceptible. the populations of asymptomatic recessive infection are large and are a dangerous infection source. the failure of closed quarantine indicates that evacuation is the best quarantine and protection. to sum up, mistake quarantine = forced infection; ineffective quarantine = condoning spreading; excessive quarantine = wasting resources. the only right way is to evacuate the infected immediately and disperse the uninfected quickly. *email: liu.liyuan@bnu.edu.cn 钻石公主号的重大启示: 意外的新冠病毒肺炎传染实验模型 刘里远* 北京师范大学生命科学学院，北京100875，中国 摘要 2020年2月1 日，从日本出发的钻石公主号邮轮，因为发现一名香港游客下船后被确诊为新型病毒肺炎 (covid-2019),邮 轮被迫停留在海上进行14天的隔离。结果，船上爆发传染，感染人数高速上升，约五分之一人群感染，震惊世界。最 后被迫终止隔离，疏散人群。该事件可以看作是一栋独立封闭大楼内的病毒传染实验模型，显示出病毒经公用通风管 道及门窗缝隙的强大空气传播能力，具有无一幸免的感染趋势，也无种族差异性，全世界人普遍易感。无症状的隐性 感染群体巨大，是危险的传播源。封闭隔离的失败，表明疏散才是最好的隔离和保护。总之，错误的隔离＝强制感 染；无效的隔离＝纵容扩散；过度的隔离＝浪费资源。唯一正确的办法是即时疏散感染者，分散未感染者。 © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 1. 引言 当2019新型冠状病毒肺炎 (covid-2019)，在武汉暴发并在亚 洲部分国家流行的时候 [1–3]，钻主公方号邮轮暴发了严重 传染。 钻主公主号邮轮，由于一名乘客在船上呆了六天，下船后发 烧咳嗽，并被测试为新型冠状病毒 (sars-cov2) 阳性，整 座船被封锁隔离在海上，船上所有3,700人都被隔离14天。如 此以来，邮轮孤零零在漂泊在海上，与世隔绝。结果，演成 了一个高度“理想”的新型冠状病毒肺炎的传染实验的居民 楼模型。模型的“设计”和运行都非常好，但实验的结果却 非常不好。在隔离过程中，感染病例的数量激增到621多例， 隔离完全失败了，最后，官方不得不宣布结束隔离，疏散撤 离 [4,5]。以惨痛代价换来的深刻教训，我们需要认真学习。 2. 邮轮概况 钻石公主号 (英文名diamond princess)位列全球十五大最豪 华邮轮之一，奢华无比，堪称一座移动的海上五星级酒店， 吃、喝、玩、乐，一应俱全。共有客舱一千三百三十七间， mailto:liu.liyuan%40bnu.edu.cn?subject= http://creativecommons.org/licenses/by-nc/4.0/ 24 l. liu / journal of risk analysis and crisis response 10(1) 23–25 其中有七百四十八间客舱带有私人露天阳台，游客可以不出 客房，直接凭栏观赏海上美景。 远远望去，钻石公主号邮轮恰似一座足有十多层楼房高的小 山，乳白色的船身在阳光的直射下，显得十分明艳。因此， 钻石公主号邮轮可以看作是新冠状病毒肺炎在一栋居民大楼 里传染传播的实验模型。 3. 事件 此次钻石公主号邮轮搭载有2666名来自世界各地的乘客(其中 近一半为日本公民)及1045名船员，总人数达到3700余人，涉 及到50多个国家和地区。看来，这是相当“理想”的世界性 人类病毒传染实验基地。其中半数为日本人，有约380名美籍 乘客（超过40人确诊感染），约330名香港居民（感染者超过 22 名），255名加拿大籍公民（感染人数不祥）。 因为有一名已下船的香港乘客在2月1日确诊新型冠状病毒 肺炎，邮轮于2月3日晚提前返回至日本横滨港，从2月5日起 横滨港附近开始为期14天的“海上隔离”。与此同时，日本 厚生劳动省开始对船上人员进行检疫工作。短短一周的时 间里，“钻石公主号”邮轮上的确诊人数陆续突破一百、两 百、三百的大关。 4. 过程 1) 1月20日，钻石公主号邮轮从日本横滨出发。 2) 1 月 25 日，邮轮上一名 80 岁的乘客在香港下船，随后 出现发烧症状，2 月 1 日病毒检测呈阳性，确认感染新 型冠状病毒。 3) 2月3日邮轮返回横滨港，远离码头。 4) 2 月 4 日，邮轮上3700 多名乘客和船员被要求留在船上 接受隔离检疫。船上的乘客被要求呆在自己的房间里， 戴上口罩，每天只能在甲板上行走几分钟，并与其他乘 客保持 2 米左右的距离，三餐会送到乘客的房间里。 5) 2月5日，日本厚生劳动省宣布，邮轮上所有人员需在海 上隔离14天。 6) 截至 2 月 10 日，日本官方仅对船上 3700 多名乘客中 的 439 人进行了冠状病毒检测，并称检测用品短缺。 7) 截至 2 月 12 日，船上一共检测了 492 人。同日，一名 负责检查船上乘客的日本卫生官员也被检测出病毒阳性。 8) 2月15日，新增67名新型冠状病毒肺炎确诊病例，其中38 人没有任何症状。截至15日傍晚，邮轮上共确诊285例新 型冠状病毒肺炎病例。这些患病的人中，还包括了一名 日本卫生官员和一名负责搬运病人的日本消防员。病毒 检测呈阳性的人可以离开邮轮接受治疗，但剩下的人至 少要在船上呆到 2 月 19 日。日本政府此前表示，从2月 19日起，检测结果呈阴性的乘客可以陆续下船。 9) 2月16日，客轮已在横滨港靠岸，接受大规模检疫及隔离 观察，又得出了289人次的新型冠状病毒检测结果，其中 70人确认感染，38人无相关症状。 10) 截至2月16日，已对船上的1219人进行了病毒检测。已经 确诊感染新型冠状病毒肺炎总数达355例(其中111人无症 状)，均被紧急送往医院。 11) 截至2月19日，完成全船人员检疫，感染人数达621人 （图1）。 12) 疏散以后，至3月15左右，最终确诊人数达712人，死亡7 例。 5. 结果 1) 这个在完全与世界隔离的海上“实验大楼”里，没有新 的外来传染源，病毒在船中递次传染。 2) 明确的病源只有一个人，在近一个月时间里，从已经 检测的1219人中（应是优先检测有症状者），已知感染 了355人，占检验人数的29%，数目庞大。其中111人无 症状，占已检阳性人员31%。带毒而没有症状，不是个 别，而是相当高的比例，是普遍现象！ 3) 所检1219人，减去有症状的244人（355-111），无症状者 为955人，其检出的阳性111人，占无症状者比例为11.6%。 那么，未检测的2500人中，包括无症状的感染者可能有 290人。加上那已检出无症状态111人，总计401人。即， 感染无症状者401人，明显超过感染有症状者355人，超 过一倍的人，是无症状的病毒携带者！ 4) 最终的感染人数达712人，应当是上面的无症状者相继发 病了，而有的可能自己痊愈了而没有发病。 5) 随着时间的推移，船上的确诊人数持续大规模增长，有 让全船全体感染的趋势。 6) 十万火急之下，美、加、意、澳等决定派专机接回自己 的未感染同胞，回国隔离观察。已病者，还是留在日本 治疗。 7) 最终的感染率，高达19%，占五分之一，是相当严重的 传染了。 8) 撤离后新增感染人数，仅91人，实际上只相当于前期的 部分无症状病毒携带者。这充分表明，撤离终止了继续 传染！ 6. 模型意义 钻石公主号邮轮实际上就是一栋居民大楼或酒店大楼的传染 模型，甚至一个小区的传染模型，还可以是一艘航母的近似 型，值得我们要加倍地重视、珍惜和利用。酒店、商场、办 公楼等皆有中央空调系统，与钻石公主号完全相似。居民楼 图1| 钻石公主号邮轮确认病例随时间快速增加。 l. liu / journal of risk analysis and crisis response 10(1) 23–25 25 内的厨房和厕所都有公用通风道，让各家实际相通。电梯在 高楼内的上下运行，就像抽风的活塞一样，搅动混合着各楼 层的空气。一旦楼中或小区里，发生了病例，就会以自家为 中心，通过空气途径向邻居传播开去。因此，要防火防盗防 邻居。虽然粪便也是传播途径，但在设施良好的楼房里，这 个途径的作用是相当次要的。为此，作者在《防治肺炎的十 大家庭隔离技巧》中，特别强调，要阻断家里与邻里上下左 右的共用通风管道。除了强调的洗手外，还要注意洗脸漱 口，甚至清洗鼻孔。常喝开水，可以清洗咽部。 这个“模型”的“实验”结果 ，重要意义是： 1) 最好的防护是隔离，最好的隔离是距离，而不是口罩和 墙壁。疏散与分散，才是最好的隔离与安全保障。 2) 要把已感染的病人，即时转移到医院去。传染病医院最 好设在人烟稀少的郊区。 3) 让其他人最好暂时离开新发病的局部疫区。惹不起，如 果躲得起的话，尽量暂时躲一躲。 4) 无症状的隐性感染者，是一个大群体，防不胜防。害人 之心不可有，但防人之心不可无。保持距离，越“远” 越好，乡村原野，是很好的避难所。 5) 社会急需简便快捷的无症状病毒携带者的普查方法。 6) 只有盾而没有矛，只防守而不进攻，是不行的。全人类 的重大目标和顶尖人才，需要从更小的芯片、更遥远的 太空中回头，对付这幽灵般的病毒。 鉴于：新加坡 、马来西亚和印度，都不断出现新病例，说明 夏天的气候并不能阻止这个新病毒的传染和发病，但可以指 望发病少些。看来，我们最后的指望只能是疫苗了。 当新冠肺炎世界大流行后 [6]，此模型仍然具有重要的现实 意义。 附： 最新恐惧邮轮事件: 在美国的至尊公主号邮轮，搭载了约3500 名乘客，2月11号起航，疫情爆发后已紧急返回旧金山，但一 直都不被允许靠岸。到3月6日，仅仅有46人接受了病毒检测， 这46人中就有21人确诊新冠肺炎，包括19名船员和2名游客。 最新航母感染事件: 美国海军“里根号”和“罗斯福号”航 空母舰上，均检出新冠肺炎阳性病例，随时推移，数量不断 在上升。其中，“罗斯福号”的感染病例已超过30例，并将 继续攀升。 7. 结论与讨论 1) 如不撤离，传染会持续严重扩大，指数性上升，或有无 一幸免之趋势。 2) 隔离失败，船上的隔离导致更多的感染，表明消毒和隔 离的巨大局限性。 3) 病毒在船体内或围绕船体流动，显示强大的空气传播能 力。船体内，其中央空调系统包括外部新鲜空气和内部空 气的混合循环，病毒在很长的通风道中流动着，这应当是 导致前期传染的重要原因。隔离后，通常会关闭内循环。 其次，病毒本身可以经窗户和门缝，以及开关门窗时的气 流，在船体内流动。可以说，整个船体都宠罩在病毒的充 斥及包围中。外面抽来的所谓新鲜空气，也不再新鲜。 4) 世界各国人种皆被感染，表明全人类的普遍易感性，没 有种族区别。特别是美加澳白人都大量感染了，所谓针 对黄种人的特异性甚至阴谋论，应当休止了。国难当 前，需要的是朋友，而不是敌人。 加拿大多伦多大学流行病学教授大卫 • 费子曼（davidfisman） 说：正是出于这个原因，医生和健康研究人员越来越质疑这种 隔离；这相当于是把一群人困在一个装有病毒的大容器里； 所以我认为这种‘隔离’正在促进病毒的传播。出于同样的原 因，哈佛大学流行病学家迈克尔 • 米娜 (michael mina) 在推 特上称这种隔离是不仁道的。 本来指望隔离14天后，大家都没事了，结果却是异乎寻常的 糟糕，豪华邮变成恐怖邮轮，甚至死亡邮轮。由于病毒强大 的人传人效应，有将全船人都感染的趋向，有无一幸免的可 能性，难以独善其身。 在一群素质比较高、条件相当好的人群中，在已经认真相互隔 离的情况下，在开阔的海面上，却发生了如此大规模的流行和 传染，显示出强大的空气传播能力，远远超出直接接触传染 途径和近距离飞沫传染范围。因为少数病毒本身可以单独飘 浮于空气，并不需要借助飞沫或形成气溶胶。其一般过程是： 飞沫—气溶胶—裸病毒。病人咳出的飞沫，在2-3米范围形成 扇形传播区；飞沫中的水分迅速蒸发，形成更小的气溶胶颗 粒，飘散得更远；气溶胶微粒进一步失去水分，形成裸病毒， 随风而逝，产生更远的传播范围。在此过程中，病毒数量逐渐 减少，但风险依然存在，这就是为什么在什么地方都需要戴口 罩的原因。此外，也可以直接以裸病毒的形式，从肺中出来。 参考文献 [1] li q, guan x, wu p, wang x, zhou l, tong y, et al. early transmission dynamics in wuhan, china, of novel coronavirus– infected pneumonia. n engl j med 2020;382:1199–207. [2] wu f, zhao s, yu b, chen ym, wang w, song zg, et al. a new coronavirus associated with human respiratory disease in china. nature 2020;579:265–9. [3] guan wj, ni zy, hu y, liang wh, ou cq, he jx, et al. clinical characteristics of 2019 novel coronavirus infection in china. n engl j med 2020;382:1708–20. [4] rocklöv j, sjödin h, wilder-smith a. covid-19 outbreak on the diamond princess cruise ship: estimating the epidemic potential and effectiveness of public health countermeasures. j travel med 2020;27;1–7. [5] the princess cruises’ official website: cruises p. princess cruises: diamond princess coronavirus & quarantine updates notices & advisories princess cruises website: @princesscruises; 2020. available from: https://www.princess.com/news/notices_and_ advisories/notices/diamondprincess-update.html [6] cohen j, kupferschmidt k. strategies shift as coronavirus pandemic looms. science 2020;367:962–3. https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1093/jtm/taaa030 https://doi.org/10.1093/jtm/taaa030 https://doi.org/10.1093/jtm/taaa030 https://doi.org/10.1093/jtm/taaa030 https://www.princess.com/news/notices_and_advisories/notices/diamondprincess-update.html https://www.princess.com/news/notices_and_advisories/notices/diamondprincess-update.html https://doi.org/10.1126/science.367.6481.962 https://doi.org/10.1126/science.367.6481.962 received 10 june 2015 accepted 26 october 2015 regional risk assessment of earthquake-triggered landslides yingying tian key laboratory of active tectonics and volcano, institute of geology, china earthquake administration, beijing 100029, china school of engineering and technology, china university of geosciences (beijing), beijing 100083, china e-mail: t190403098@163.com chong xu key laboratory of active tectonics and volcano, institute of geology, china earthquake administration, beijing 100029, china e-mail: xuchong@ies.ac.cn jian chen school of engineering and technology, chinauniversity of geosciences (beijing), beijing 100083, china e-mail:jianchen@cugb.edu.cn abstract great earthquakes occurring in mountainous areas can trigger large-scale landslides, leading to serious geological disasters. thus, in recent years, especially after the 2008 wenchuan earthquake, much attention has been focused on the research about regional risk assessment of seismic landslides in china and elsewhere in the world. such study is based on the engineering geological analogy, and its purpose is to estimate the risks of earthquake-induced landslides for the regions with the same or similar environment. in light of previous work, such assessment includes 2 tasks: establishment of seismic landslide database and evaluation of potential landslides using mathematical statistics models. the interpretation of regional seismic landslides is the basis for building the landslide database. the common methods for risk assessment include the evidence-weight model, certainty factor method (cf)and information value model, logistic regression model (lr), artificial neural networks (ann), support vector machine method (svm), newmark displacement model,analytic hierarchy process (ahp), and so forth. this paper presents a review on these methods, and an outlook on the advancement of this research field in the future. keywords:seismic, landslide, hazard assessment. journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 234-245 published by atlantis press copyright: the authors 234 区域地震滑坡危险性评价 田颖颖 中国地震局地质研究所，活动构造与火山重点实验室，北京 100029 中国地质大学（北京），工程技术学院，北京 100083 e-mail: t190403098@163.com 许冲 中国地震局地质研究所，活动构造与火山重点实验室，北京 100029 e-mail: xuchong@ies.ac.cn 陈剑 中国地质大学（北京），工程技术学院，北京 100083 e-mail:jianchen@cugb.edu.cn 摘要 山区大地震通常触发大规模滑坡并诱发严重的地质灾害。近年来，特别是 2008 年汶川地震之后，国内 学者高度重视区域地震滑坡危险性评价研究。它以工程地质类比法为理论基础，去估计具有相同或类似环 境条件的区域尺度地震滑坡的危险性。从前人工作来看，区域危险性评价主要包括 2 方面：建立地震滑坡 数据库和运用数学统计模型进行危险性评价。数据库建立以区域地震滑坡解译为基础，地震滑坡危险性评 价常用方法包括：证据权重模型、确定性系数法（cf）、信息量法、逻辑回归模型（lr）、人工神经网络（ann）、 支持向量机法（svm）、newmark 位移模型和层次分析法（ahp）等。本文对各种评价方法做了综述，并对 地震滑坡危险评价研究趋势提出了一些看法。 关键词：地震；滑坡；危险性评价 1. 引言 地震滑坡泛指由地震所诱发的次生地质灾害，包 括崩、滑、流等。在地震发生之后，地震滑坡波及面 积之广、发生频率之高、运动速度之快、灾害损失之 严重决定其致灾能力不亚于地震本身，因此引起了广 泛关注。早在公元前 1789 年中国和公元前 373 年或 372 年希腊就已经有了地震滑坡的记录，第一次正式 被用来进行科学研究的大地震为 1783 年发生在意大 利的 calabria 地震（keefer, 2002）。近年来国内外高 震级地震频发，地震及其诱发的次生地质灾害造成了 严重的人员伤亡和经济损失。1920 年 12 月 16 日宁 夏海原 ms8.5 级地震，诱发了至少 805 处大型黄土滑 坡，其所占的面积约 5 万 km2（吴玲, 2012; li 等, 2013b）。1994 年 1 月 17 日美国加州的 northridge mw6.7 级地震在 1 万 km2 的面积内触发了至少 11,000 处滑坡，且滑坡的面积多大于 1000 km2，造成了重大 的损失（harp and jibson, 1996）。1999 年 9 月 21 日台 湾 ms7.6 级集集地震，引发 13,175 处滑坡，造成了重 大的经济损失（lee, 2013）。2008 年 5 月 12 日四川 汶川 ms8.0 级地震诱发了 197,481 处滑坡灾害，直接 造成了 2 万人死亡和巨大的经济损失（殷跃平, 2009）。 2010 年 1 月 12 日海地太子港 mw7.0 级地震触发了至 少 30,828 处滑坡，造成 23 万多人遇难和近 80 亿美元 的经济损失（calais 等, 2010; 许冲, 2013a）。2010 年 4 月 14 日青海玉树 ms7.1 级地震在面积约 1455.3km2 的矩形区域内诱发了 2,036 处滑坡，造成的直接经济 损失约 60 万元（许冲等, 2011）。2013 年 4 月 20 日芦 山 ms7.0 级地震解译出滑坡 3,883 处，累计受灾人口 达 38.3 万人（许冲, 2013b）。2013 年 7 月 22 日甘肃 岷县漳县 ms 6.6 级地震中仅“永光村 1#滑坡” 滑坡 （体积约 4.2×104m3）和“永光村 2#滑坡”滑坡（体积 published by atlantis press copyright: the authors 235 约 3.3×104m3），就造成了永光村 4 社瞬间被夷为废 墟，造成了至少 12 人遇难（许冲等, 2013）。鉴于地 震滑坡的巨大致灾能力，地震滑坡灾害危险性评价显 得尤为重要。 2. 地震滑坡危险性评价概述 进行地震滑坡危险性评价主要有两种思路：一是 对大型单体地震滑坡进行物理力学性质分析，并用力 学模型去模拟滑坡发生的环境，达到危险性评价的目 的；二是基于地震、地质和地形地貌等影响地震滑坡 发生的因素，对区域性地震滑坡进行统计分析。本文 主要针对区域性地震滑坡的危险性评价方法进行论 述。 区域性地震滑坡危险性评价利用工程地质类比 法思想，根据地震后已经发生滑坡的空间分布信息反 演地震滑坡涉及范围内的滑坡易发条件，从而进行危 险性区划，为以后相似地质环境下的防灾减灾工作提 供参考。影响地震滑坡的因素有很多，包括地震因素、 地质因素和地形地貌因素，由于这些因素具有不确定 性和模糊性，加之有些因素很难获得或量化，因此地 震滑坡危险性评价十分复杂，目前还没有成熟的方法 （陈晓利等, 2008; 庄建琦等, 2010）。 早期人们对地震滑坡研究主要依靠实地调查。20 世纪 30 至 40 年代，航空照片的出现，成为了记录地 震滑坡发生的重要手段。航空照片使得我们可以很方 便地辨别滑坡的位置以及规模，大大减少了野外工作 量，为滑坡的编目提供了较为便捷的途径。20 世纪 80 年代高精度卫星影像的出现，进一步推动了地震滑 坡数据库的建立（keefer, 2002）。同时随着 gis 的出 现，地震滑坡危险性评价过程中复杂空间分析的需求 得到了很好地解决，进一步刺激了地震滑坡危险性评 价的发展。 通过野外实地调查和各种遥感数据解译可以获 得客观、详细、完整的滑坡编目；然后结合地震、地 质及地形资料，进而获得地震滑坡的空间分布规律。 最后借助各种数学模型进行滑坡危险性评价。进行地 震滑坡危险性评价时常常参考研究比较成熟的降雨 诱发滑坡危险性评价的研究成果（葛华等, 2013）。 3. 地震滑坡危险性评价的步骤 地震滑坡危险性评价的步骤可简要概括为： （1）进行滑坡编目。详细精确的滑坡编目是地 震滑坡危险性分析工作中必不可少的一部分（harp 等, 2011; 许冲和徐锡伟, 2014; xu, 2015）。借助连续且覆 盖率高的高分辨率遥感影像，用点要素或面要素去标 识滑坡，圈定出滑坡的轮廓和位置，并对特殊的滑坡 灾害进行现场调查和核实，创建地震滑坡数据库，这 是进行地震滑坡危险性评价的基础。尤其是 2008 年 汶川地震之后，出现了多个基于地震事件的同震滑坡 数据库成果，如 2010 年甘肃玉树地震（xu and xu, 2014）、2010 年海地地震（xu et al., 2014a）、2013 年 四川芦山地震（xu et al., 2015a; xu et al., 2015b）、2013 年甘肃岷县地震（xu et al., 2014b）、2014 年云南鲁甸 地震（许冲 et al., 2014）等。其中 2008 年四川汶川 地震同震滑坡数据库是记录最详细、滑坡数量最多的 （xu et al., 2014c）。 （2）开展地震滑坡危险性评价。选取影响地震 滑坡发生的地震、地质、地形地貌等因素，基于 gis 和各种评价模型进行地震滑坡危险性评价，并制作危 险性区划图。 影响因素 滑坡数据 地质图 地形图 地质因子 地形因子 地震数据 地震因子 遥感影像 滑坡编目 危险性评价及区划 评价模型 图 1.地震滑坡危险性评价流程 fig. 1.the steps of earthquake-induced landslide hazard assessment. gis published by atlantis press copyright: the authors 236 4. 地震滑坡危险性评价的方法 通过总结国内外研究者针对区域性地震滑坡危 险性评价所做的工作，危险性评价的方法大致可分为 2 大类：（1）数据驱动方法，通过研究已发生的真实 滑坡分布数据，统计各影响因子对地震滑坡的贡献权 重，从而得到地震滑坡危险性评价结果；（2）知识驱 动方法，主要通过专家知识或者机器学习事件发生的 规律，为各影响因子赋权重，从而判断滑坡在研究区 内的分布规律（谢洪斌等, 2011; 许冲, 2014）。 常用的评价方法主要有证据权重模型、确定性系 数法（cf）、信息量法、逻辑回归模型（lr）、人工 神经网络（ann）、支持向量机模型（svm）、newmark 位移模型以及层次分析（ahp）法等。 4.1.数据驱动方法 4.1.1.证据权重模型 证据权重模型属于数据驱动方法，是一种离散的 统计方法，其实质是根据已发生的滑坡事件来推测在 什么情况下较容易发生滑坡。它是以贝叶斯概率统计 为基础，假设各影响滑坡发生的因子证据层之间相互 独立，通过计算已有滑坡事件发生时的先验概率以及 各影响因子对滑坡发生与否的贡献值（权重），然后 基于卡方（ 2χ ）检验，对两两因子证据层进行条件 独立性检验，筛选出相关性较小的因子组合，并计算 滑坡危险度的大小（后验概率），进而得到滑坡危险 性分布图（lee 和 choi, 2004; 王志旺等, 2007; dahal et al., 2008; 赵艳南和牛瑞卿, 2010）。 { } { } | ln | i p f l w p f l + = . (1) { } { } | ln | i p f l w p f l − = . (2) ( )f i iw w w+ −= − . (3) 其中， { } { }| |p f l p f l 代表滑坡发生的充分率； { } { }| |p f l p f l 代表滑坡发生的必要率； iw+ 表示 当前影响因子级别发生滑坡的概率， iw − 代表影响因 子级别以外的部分发生滑坡的概率，二者的差值代表 该影响因子级别发生滑坡的权重，即 fw 。可以用 fw 衡量影响因子对滑坡的重要性，当 fw 为正，表示该 影响因子级别有利于发生滑坡，为负表示不易发生滑 坡，为 0 表示该影响因子级别对滑坡影响较小。 证据权重模型由于条件独立的约束，忽略了各评 价因子之间的相互作用，但是避免了因子权重赋值的 人为主观性。正负权重值的引入又兼顾了滑坡区域和 不滑区域权重，其评价结果也具有很高的参考价值。 谢洪斌等（2011）将模糊概率和条件模糊概率引 入证据权模型对 2008 年汶川地震震区岷江流域雁门 乡至映秀段进行了地震滑坡危险度区划，认为 80% 以上的滑坡位于高和极高危险区。xu 等（2012b）、 xu 等（2012c）基于证据权法分别对 2008 年汶川地 震震区清水河流域的 2321 处滑坡和 2010 年玉树地震 诱发的 2036 处滑坡进行了危险性区划，得到模型的 成功率分别为 71.82%和 80%。张艳玲等（2012）利 用证据权法完成了 2008 年汶川地震后 12 个极重灾县 市的滑坡危险性分区，模型正确率达 81%。 4.1.2.确定性系数法 确定性系数法（centainty factor, cf）是一种二 元统计方法，cf 由 shortliffe 和 buchanan 于 1975 年 提出，后经 heckerman（1986）改进的一个概率函数， 利用工程地质类比思想，在假定可以根据已发生的地 震滑坡和其诱发因素之间的统计关系进行滑坡危险 性评价的基础上，来分析各影响因素对滑坡事件发生 的敏感程度（兰恒星等, 2002, 2003）。该方法也完全 基于实际滑坡数据，避免了不滑样本选取的随意性， 对于大区域地震滑坡危险性评价来说具有较强的适 用性。 cf 函数具体表示为： ( ) ( ) [ ] , 1 , 1 1,1 . a s a s a s a s a s s a pp pp pp pp pp pp pp pp pp pp pp pp cf cf − ≥ − − < −   = ∈ −   (4) 其中， ppa 为地震滑坡事件在因子分类 a 中发生 的条件概率，即代表因子分类 a 中存在的地震滑坡面 积百分比； pps 为地震滑坡事件在整个研究区 a 中发 生的先验概率，可表示为整个研究区地震滑坡的面积 百分比，为一定值（许冲等, 2010a）。 cf >0 表示地 震滑坡变形失稳的确定性高，此单元为滑坡易发区； cf <0 则表示此单元不易发生滑坡； cf 接近 0 代表 事件发生的确定性难以确定，即不能确定此单元是否 有利于发生滑坡。 陈晓利等（2009）通过搜集前人的研究成果和龙 陵地震滑坡的相关资料，对 1976 年龙陵地震诱发的 滑坡运用 cf 法进行了影响因子的敏感性评价，确定 published by atlantis press copyright: the authors 237 了研究区内最有利于地震滑坡发生的数值区间。许冲 等（2010a, 2010b）运用 cf 法，基于不同的影响因子 以及因子组合对 2008 年汶川地震滑坡进行了影响因 子敏感性研究以及易发性评价，确定了滑坡因子的易 发区间以及危险性区划图。刘丽娜等（2014a）利用 cf 法对 2013 年芦山地震滑坡的影响因子敏感性进行 了分析，对不同的地形因子间以及地质与地震因子间 的相互影响进行了探索研究。 4.1.3.信息量法 信息量模型首先由晏同珍等（1989）基于信息论 引入了信息量法进行滑坡预测。具体到地震滑坡来 说，通过对已发生地震滑坡进行分析，把影响地震滑 坡各种因素的实测值转化为信息量值，用信息量值的 大小来衡量各影响因子与地震滑坡之间的作用程度 （阮沈勇和黄润秋, 2001）。该方法也是基于实际滑坡 数据得到因子的权重的一种二元统计方法，避免了因 子权重赋值的主观性。有效样本数越多，模型评价结 果越准确。但要注意评价因子选择的人为主观性。 信息量在实际计算时可用样本频率来估算条件 概率，则各影响因素 ix 对地震滑坡事件 a 提供的信息 量为： ( ) ( ) ( ) | , ln lni ii i i p x a n n i x a p x s s = = . (5) 那么假设参加评价的因子有 n 个，则单个评价单 元内总的信息量 ii ： ( ) 1 1 1 , a ln n n n i i i i i i i ii n n n s i i x s s n s= = = = = = .∑ ∑ ∑ (6) 其中， ( ),ii x a 为滑坡分布条件下出现 ix 的概率， 为研究区内出现 ix 的概率， s 为研究区总单元 数，n 为研究区滑坡所占的单元总数， is 为研究区内 存在因素 ix 的单元个数， in 为分布在因素 ix 内的特 定类别内的滑坡单元数。 i in s 为滑坡在某影响因子 特定类别内的分布密度；n s 为研究区的滑坡面积百 分比，为一定值（高克昌等, 2006）。 刘应辉（2009）运用信息量法对 2008 年汶川地 震震区都汶公路沿线崩塌滑坡灾害等进行了研究，结 果表明该模型的评价效果较好。陶舒等（2010）等采 用信息量法与逻辑回归模型以 2008 年汶川地震后的 汶川县北部的地震次生滑坡灾害为例进行了敏感性 评价，分别得到了 2 幅敏感性评估图，结果表明信息 量法在极高敏感区内的评价精度也很高。牛全福等 （2011）运用信息量模型得到了 2010 年玉树地震诱 发滑坡的危险性区划图。xu（2013）对 2013 年芦山 地震震区 3878 处地震滑坡运用基于专家知识和信息 量模型分别进行了地震滑坡危险性评价，最终认为基 于信息量模型的芦山地震震区危险性评价精度较高。 4.1.4.逻辑回归模型 逻辑回归模型（logistic regression, lr）是通过 探索一个二值因变量和多个自变量之间形成的回归 关系，从而预测一定条件下某一事件的发生概率。逻 辑回归的优势在于其最优输出由最小二乘法给定，且 进行统计分析时，自变量可以是连续的，也可以是离 散的（邢秋菊等, 2004; 刘艺梁等, 2010）。逻辑回归 模型作为一种多元统计方法，具有操作方便、性能 稳定、计算过程受人为干扰小，且利用回归系数较 好地体现了评价因子之间的相互关系。这种模型适 用于大区域地震滑坡危险性评价研究，需要大量的 滑坡样本和不滑样本。该模型也是区域地质灾害研 究最常用的统计模型。 设 p 为发生地震滑坡的概率，取值范围[0, 1]， ( )1 p− 为不发生地震滑坡的概率，将 p/(1-p)取自然对 数 ( )( )ln 1p p− ，记为 log pit ，以 p 为因变量，建立 线性回归方程： 1 1log = + m mit p x xα β β+ + . (7) 1 1 1 1 exp( + ) 1 exp( + ) m m m m x x p x x α β β α β β + + = . + + +   (8) 其中，α 为常数， 1β 为回归系数。 garcia-rodriguez 等（2008）利用逻辑回归模型， 对 2001 年萨尔瓦多地震所诱发的滑坡灾害进行了危 险性评价，认为地形粗糙度和土地类型是影响该地区 滑坡发生的最重要的因素，最后得到了该地区相对的 滑坡敏感性分布图。su 等（2010）基于 gis 和逻辑 回归模型对 2008 年汶川地震后青川县的地震滑坡进 行了危险性区划，65.3%的滑坡分布在高和极高危险 性区域内。许冲和徐锡伟（2012c）针对 2010 年玉树 地震的地震滑坡进行了解译，并运用逻辑回归模型建 立了滑坡危险性指数分布图，对比实际滑坡分布图与 滑坡危险性指数图得到模型的正确率达到了 83.21%。 xu 等（2013）运用逻辑回归模型对 2008 年汶川地震 所诱发的近 20 万处滑坡进行了危险性区划，实际滑 坡分布与敏感指数图对比的结果表明，模型的成功率 83.751%，预测率 86.930%。李晓璇和马海建（2013）、 文海家等（2014）分别基于逻辑回归模型对汶川地震 重灾区汶川县的地震滑坡进行了危险性评价，李晓璇 和马海建等利用 998 处滑坡样本、998 处非滑坡样本 点基于坡度、断裂带、地形起伏度、岩性和烈度 5 个 因子进行分析，认为 72.4%的灾害点位于高和极高危 险区；文海家等利用 149 个滑坡样本点、751 个不滑 published by atlantis press copyright: the authors 238 样本点基于高程、坡度、坡位、坡向、岩性、微地貌 等 12 个因子进行分析，认为 61.2%的滑坡位于高和极 高危险区，模型的正确率高达 94%。li 等（2013a） 对 2008 年汶川地震震区 37 个重灾县中的前 10 个县 的 43842 处地震滑坡运用逻辑回归方法进行了地震滑 坡危险性评价，结果表明在占研究区总面积 3.9%的高 和极高危险区内分布有 73.3%的滑坡，模型的成功率 也达到了 80%。 4.1.5.人工神经网络 人工神经网络（artifieial neural newtork, ann）， 是通过模拟人脑的思维，把不同层的大量神经元连接 成一个复杂的网络，用具有代表性的已知样本对神经 元之间的连接进行训练，在不断地训练和检验的过程 中实现输出与目标的一致，进而建立起不同输入层隐层-输出层之间的非线性映射关系并对未知样本进 行预测（ermini 等, 2005; chen 等, 2009）。其中，bp 神经网络（back-propagation neural network, 反向传 播神经网络）模型是目前应用最广泛的的一种神经网 络模型。对于地震滑坡危险性评价来说，就是将不同 的影响因子和具有代表性的滑坡样本作为输入层，通 过隐层处理后输出，若输出结果与预测样本不符，输 出误差将朝着误差最小的方向通过隐层向输入层传 递，从而修正输入层与隐层以及隐层与输出结果的权 重和阈值，通过反复训练，最后得到与实际滑坡分布 一致的危险性评价模型，用于对未发生滑坡区域进行 危险性区划。 gis 已知样本 高程 坡度 岩性 …… 断裂 烈度 水系 ann 是否满足条件 输入 输出 否 预 测 样 本 完成样本训练 是 图 2.神经网络的基本原理 fig. 2.basic principles of the neural network. lee 和 evangelista（2006）基于 gis 和 bp 神经 网络针对 1990 年发生在菲律宾碧瑶市的地震所触发 的滑坡样本进行训练，最后将所得神经网络模型和已 知滑坡进行验证对比，模型预测的成功率达到了 93.2%。陈晓利等（2006）利用径向基概率神经网络 自学习的特性，以 1976 年龙陵地震触发的 32 个大型 滑坡和滑坡群所在的区域为研究区，基于区内已知的 滑坡单元和未发生滑坡的单元为训练样本，通过样本 的自主学习训练对待识别样本进行判断，最终的识别 精 度 达 到 了 89.9% 。 garcía-rodríguez 和 malpica （2010）基于人工神经网络法针对 2001 年 ei salvador 地震所诱发的滑坡灾害以及影响边坡稳定性的各因 素进行了危险性评价，其中有 83%的滑坡位于极高和 高危险区。许冲和徐锡伟（2012a）利用 bp 神经网络 以滑坡的质心位置和滑坡源区坡体高程最大位置为 训练样本对 2010 年玉树地震滑坡进行危险性区划， 最终模型的预测率都达到了 81%。li 等（2012）对 2008 年汶川地震后青川县的 855 处地震滑坡以及震前 的 473 处降雨滑坡运用 bp 神经网络模型进行了危险 性评价，基于地震诱发滑坡的神经网络模型成功率为 98%。 人工神经网络虽然在工程领域得到了认可，但当 滑坡数量和评价因子数量较多时，计算量较大是十分 突出的问题；并且人工神经网络对于输入层的优化存 在着黑箱操作问题，很难解释和验证其反演结果。 4.1.6.支持向量机 支持向量机（support vector machines, svm）由 vapnik（2000）创建，是一种基于统计的模型，能够 published by atlantis press copyright: the authors 239 在样本较少的情况下，通过核函数将线性不可分的数 据转换成高维空间的线性可分数据。基于结构风险最 小原则和求解受限二次型使得支持向量机法可以在 理论上获得最优解，因此支持向量机在有效滑坡样本 和影响因子较少情况下，可以获得较合理的危险性评 价结果。其中，支持向量机模型的重要组成部分是核 函数。常见的核函数主要有（cristianini 和 scholkopf, 2002; 戴福初等, 2008; 谭龙等, 2014）： 线性核函数（linear）： ( ), ti j i jk x x x x= . (9) 多项式核函数（polynomial）： ( ) ( ), , 0dti j i jk x x x x rγ γ= + > . (10) 径向基核函数（radial basis function）： ( ) ( ) 2 , i jx xi jk x x e γ− − = . (11) s 形核函数（sigmoid）： ( )( , ) tan γ += .ti j i jx xk x rx (12) γ 、 r 、 d 是核函数的参数，计算时需要人工输 入。其中，基于径向基核函数的支持向量机模型得到 的评价结果效果最好。 xu 和 xu（2012）、许冲和徐锡伟（2012）基于 不同的核函数（线性核函数、多项式核函数、径向基 核函数和 s 形核函数）运用支持向量机法对 2010 年 玉树地震所诱发的 2036 处滑坡进行了空间预测和危 险性分级，并指出径向基核函数是最适于该地区地震 滑坡空间预测的模型，其模型正确率达 84.16%。xu 等（2012a）以 2008 年汶川震区涪江流域的地震滑坡 为研究对象，基于支持向量机法，选用了线性核函数、 多项式核函数和径向基核函数进行了滑坡敏感性区 划，模型的成功率和预测率都在 70%以上，且径向基 函数的向量机模型成功率和预测率最高。 4.1.7.newmark 位移模型 newmark 位移模型假设地震触发的滑坡体为刚 性块体，基于极限平衡理论，通过计算滑体在地震荷 载作用下沿着危险滑动面所发生的永久位移量来评 价斜坡的稳定性，而永久位移可以通过对地震荷载加 速度和滑体滑动的临界加速度的差值进行积分得到， 所以该方法的重点转移到了求取滑坡体本身的临界 加速度和安全系数上。临界加速度和安全系数可以通 过岩土体的力学性质（有效内聚力、有效内摩擦角、 岩土体重度等）以及坡度等计算得到，这就使得该方 法在地震滑坡危险性评价中得到了普及（newmark, 1965; jibson 等, 2000; 王涛等, 2013）。 正是由于 newmark 位移方法利用临界加速度 或 newmark 位移等来评价地震滑坡的危险性，所 以该方法不需要滑坡样本数据，仅根据震区的岩土 体强度和坡度来判断危险区域。但这种方法忽略了 区域地质体的局部差异性和复杂性以及坡体的规 模。因此当缺乏较详细区域岩土体的实测强度参 数、地震动参数等资料时，newmark 位移方法的精 度还有待提高。 refice 和 capolongo（2002）基于简化的 newmark 方法对意大利南部 irpinian 地区的地震滑坡进行了危 险性评价，与其他评价方法的结果基本一致。jibson 等（2000）对 1994 年 northridge 地震所诱发的滑坡 利用 newmark 永久位移法进行了分析，得到了滑坡 的危险性区划图。陈晓利等（2013）利用 newmark 模型对 2013 年芦山震区烈度大于 vii 的区域内的地 震滑坡进行了空间分布预测，虽然预测结果与实际滑 坡分布情况有一定的出入，但该方法仍表现出了较高 的可行性。 4.2.知识驱动方法 层次分析法（analytichierarchy process, ahp）属 于常用的知识驱动模型，是由美国匹兹堡大学教授 saaty 在 20 世纪 70 年代中期提出的面向多因素、多 层次、多目标的复杂系统和难于完全用定量的方法来 分析与决策的复杂系统工程问题的处理方法，其思想 是通过构造判断矩阵对比两两影响因子，确定诸因子 的相对重要性（saaty, 1988; 陈晓利, 2007）。判断矩 阵标度及含义见表 1。 published by atlantis press copyright: the authors 240 表 1.层次分析法判断矩阵的标度和含义 table.1.the scale and the meanings of the ahp matrix. 标度值 含义 1 表示两个因素相比，具有相同重要性 3 表示两个因素相比，前者比后者稍重要 5 表示两个因素相比，前者比后者明显重要 7 表示两个因素相比，前者比后者强烈重要 9 表示两个因素相比，前者比后者极端重要 2，4，6，8 表示上述相邻判断的中间值 倒数 与上述影响情况相反 ahp 法作为一种多指标分析方法，其基本操作流 程为由专家将影响地震滑坡发生的因素进行简单比 较和计算，得到每个因子合理的权重，然后进行危险 性评价。这是一种定性和定量相结合的方法。虽然带 有一定的主观性，但是通过比较影响因子得到的权重 避免了较不合理的情况。这种方法不需要选择滑坡样 本，仅依靠专家知识和影响因子基础数据即可实现， 操作方便快捷，精度较高。 许冲等（2009）利用 ahp 法对 2008 年汶川地震 震区的 48007 个滑坡进行了危险性评价，模型成功率 达到了 79.656%。杜军和杨青华（2009）、张建强等 （2009）基于 ahp 模型，确定了 2008 年汶川震区汶 川县的地震次生地质灾害各评价因子的权重，并进行 了风险性评估，结果与实际灾害分布一致。tang 等 （2009）基于专家知识，利用 ahp 法获取 2008 年汶 川地震震区青川县地震滑坡影响因子的权重并进行 了该地区地震滑坡危险性评价，然后又将该评价模型 应用于北川县城地震滑坡进行危险性评价，都得到了 较好的结果。刘丽娜等（2014b）使用 ahp 法采用多 名专家估计方案对 2013 年芦山地震震区一个矩形区 域内的滑坡进行了危险性评价，模型的成功率高达 85.41%。 5. 实例分析 以 2008 年汶川地震震区嘉陵江涪江一个支流流 域为研究区，通过实地调查和对覆盖整个研究区的 1m 分辨率的航片进行目视解译，得到地震诱发滑坡 944 处。选取了坡度、坡向、高程、曲率、距河流的 距离、地形湿度指数（twi）、距主干道路的距离以 及距地表破裂带的距离、pga 和岩性等 10 个影响地 震滑坡发生的因素，分别基于二元统计模型、逻辑回 归模型、人工神经网络、线性核函数向量机模型、多 项式核函数向量机模型、径向基核函数向量机模型进 行了危险性评价，最终得到了研究区内的地震滑坡危 险性分区图（图 3）（xu 等, 2012c）。 运用实际滑坡数量百分比累加-危险性面积百分 比累积曲线的曲线下面积方法对这 6 种模型评价结果 的成功率和预测率进行定量检验。各模型的曲线下面 积见图 4。从图中可以看出，模型的成功率和预测率 由高到低依次为：逻辑回归模型、基于径向基函数的 支持向量机模型、人工神经网络、基于多项式函数的 支持向量机模型、二元统计模型、线性函数向量机模 型。其中逻辑回归模型的成功率和预测率均最高 （auc 分别为 80.34%和 80.27%）。 分析结果表明，在该研究区内，基于逻辑回归模 型的地震滑坡危险性评价结果最好，其次为基于径向 基函数的支持向量机模型。 图 3 涪江流域地震滑坡危险性分区图（xu 等, 2012c） fig.3. landslide susceptibility mapsbased on different models in fu river basin. (a)二元统计模型；(b) 逻辑回归模 型；(c) 人工神经网络；(d) 线性核函数向量机模型；(e) 多 项式核函数向量机模型；(f) 径向基核函数向量机模型。 published by atlantis press copyright: the authors 241 图 4.滑坡危险性评价结果评价曲线（xu 等, 2012c） fig.4. area under curve (auc) representing the success rate and prediction rate of the models used. (a) 成功率曲线；(b) 预测率曲线。 6. 结论与展望 6.1.结论 （1）随着 gis 的发展，特别是其空间分析功能 的完善，通过结合 gis 和各种评价方法，地震滑坡危 险性评价变得更加便捷，评价精度和效率也得到了很 大程度的提高。利用 gis 的制图功能，也可做出各种 危险性区划图件，方便灾区支援和减灾工作。 （2）地震滑坡编目是地震滑坡危险性评价的数 据基础。通过各种评价模型分析各影响因子对地震滑 坡的贡献，从而进行地震滑坡危险性区划。常用的地 震滑坡危险性评价的模型有证据权重模型、确定性系 数法（cf）、信息量法、逻辑回归模型（lr）、人工 神经网络（ann）、支持向量机法（svm）、newmark 位移模型和层次分析法（ahp）等。这些方法各有利 弊：数据驱动方法中，证据权重模型、确定性系数法 和信息量法均基于实际滑坡数据，避免了不滑样本选 取的随意性和因子赋值的主观性，适用于存在大量有 效滑坡样本的区域，在评价时要尽量将影响因子考虑 全面。逻辑回归模型也适用于存在大量滑坡样本的区 域，但是对于不滑样本的选取没有统一的标准，这往 往会影响评价的结果。人工神经网络和支持向量机模 型由于处理过程的复杂性，数据量较大，适用于小区 域地震滑坡危险性评价。newmark 位移法将岩土体的 力学性质引入区域滑坡危险性评价中，仅根据震区的 岩土体强度和坡度来判断危险区域，忽略了区域地质 体的局部差异性和复杂性以及坡体的规模。因此只适 用于具备较详细区域岩土体的实测强度参数、地震动 参数等资料的小区域地震滑坡危险性评价。文中以 2008 年汶川地震震区嘉陵江涪江一个支流流域为例， 基于二元统计模型、逻辑回归模型、人工神经网络、 线性核函数向量机模型、多项式核函数向量机模型、 径向基核函数向量机模型，对研究区内的 944 个地震 滑坡进行了危险性评价。结果表明，基于逻辑回归模 型的地震滑坡危险性评价结果最适用于本研究区。 6.2.展望 （1）虽然地震滑坡危险性评价已经进入定量阶 段，但由于每种评价方法自身的适用性和局限性，加 之地震的预测十分困难，从而造成了危险性评价因子 和评价结果的各种不确定性和模糊性。另外，由于地 震滑坡形成的力学机理和环境条件十分复杂，所以目 前地震滑坡危险性评价的方法仍有待改进，必要时可 以综合两种或多种评价模型进行评价。 （2）较好的评价结果必须建立在对历史地震滑 坡研究较充分的基础上，但在进行地震滑坡解译的时 候，震后第一时间的遥感影像并不容易获得，影像的 质量、分辨率和云量等都影响编目结果的准确性。 （3）选取评价因子以及对评价因子赋权重时， 需考虑地震滑坡影响因子选取的合理性、全面性以及 最优因素组合，尽量降低外界因素对评价结果的影 响，使评价结果更加客观。 （4）传统的地震滑坡统计分析方法仅考虑了影 响滑坡发生的表层因素，而这些因素大多是静态的， 这就很可能忽略了对滑坡发生的起关键作用的因素， 比如断裂深部的形态以及坡体内含水量等对滑坡发 生的影响。因此，进行危险性评价时要基于滑坡所处 的地质背景，同时也要关注对滑坡自身形态以及滑坡 物源区、流通区以及堆积区的形态和规模等与各影响 因素之间的关系，并加强对区域地震滑坡危险性评价 的物理或数值模拟研究。 致谢 本文由国家自然科学基金（41472202）资助。 published by atlantis press copyright: the authors 242 参考文献 d. k. keefer, investigating landslides caused by earthquakes–a historical review,surveys in geophysics, 23(6) (2002) 473-510. w. l. li, r. q. huang, x. j. pei and x. c. zhang, historical co-seismic landslide inventory with google earth: a case study of 1920 haiyuan earthquake, china,global view of engineering geology and the environment. boca raton: crc press, (2013a) 179-184. 吴玲, 基于 gis 和 rs 什邡市山区地震滑坡主要影响因子 的研究. (四川农业大学, 四川, 2012). e. l. harp and r. w. jibson, landslides triggered by the 1994 northridge, california, earthquake,bulletin of the seismological society of america, 86(1b) (1996) s319-s332. c.t. lee, re-evaluation of factors controlling landslides triggered by the 1999 chi–chi earthquake, in earthquake-induced landslides (springer, 2013), pp. 213-224. 殷跃平, 汶川八级地震滑坡特征分析, 工程地质学报, 17(01) (2009) 29-38. e. calais, a. freed, g. mattioli, f. amelung, s. jónsson, p. jansma, s.h. hong, t. dixon, c. prépetit and r. momplaisir, transpressional rupture of an unmapped fault during the 2010 haiti earthquake,nature geoscience, 3(11) (2010) 794-799. 许冲, 2010 年海地太子港地震触发滑坡危险性区划和合理 性检验, 科技导报, 31(12) (2013a) 42-47. 许冲, 徐锡伟, 戴福初, 王彦颖, 2010 年 4 月 14 日玉树 地震滑坡空间分布与控制变量分析, 工程地质学报, 19(4) (2011) 505-510. 许冲, 2013 年芦山地震 ms7.0 级滑坡易发性快速评价方法. 科技导报, 31(28/29) (2013b) 15-23. 许冲, 徐锡伟, 郑文俊, 闵伟, 任治坤 and 李志强, 2013 年甘肃岷县漳县 6.6 级地震触发滑坡及其构造分析, 地震地质, 35(3) (2013) 616-626. 陈晓利, 祁生文,叶洪, 基于 gis 的地震滑坡危险性的模糊 综合评价研究, 北京大学学报: 自然科学版, 44(3) (2008) 434-438. 庄建琦, 崔鹏, 葛永刚, 朱颖彦, 刘应辉, 裴来政, “5·12” 汶川地震崩塌滑坡危险性评价——以都汶公路沿线为 例, 岩石力学与工程学报, 29(增 2) (2010) 3735-3742. 葛华, 陈启国, 王德伟, 地震滑坡危险性评价及编图——以 映秀震中区为例, 中国地质, 40(2) (2013) 644-652. e. l. harp, d. k. keefer, h. p. sato and h. yagi, landslide inventories: the essential part of seismic landslide hazard analyses. engineering geology, 122(1) (2011) 9-21. 许冲, 徐锡伟, 21 世纪初几次大地震事件触发滑坡基础数 据建设, 地震地质, 36(1) (2014) 90-104. c. xu, preparation of earthquake-triggered landslide inventory maps using remote sensing and gis technologies: principles and case studies,geoscience frontiers, 6(6) (2015) 825-836. c. xu and x. w. xu, statistical analysis of landslides caused by the mw 6.9 yushu, china, earthquake of april 14, 2010,natural hazards, 72(2) (2014) 871-893. c. xu, j. b. h. shyu and x. w. xu, landslides triggered by the 12 january 2010 port-au-prince, haiti, mw = 7.0 earthquake: visual interpretation, inventory compiling, and spatial distribution statistical analysis,natural hazards and earth system sciences, 14(7) (2014a) 1789-1818. c. xu, x. w. xu and j. b. h. shyu, database and spatial distribution of landslides triggered by the lushan, china mw 6.6 earthquake of 20 april 2013,geomorphology, 248((2015a) 77-92. c. xu, x. xu, j. b. h. shyu, m. gao, x. tan, y. ran and w. zheng, landslides triggered by the 20 april 2013 lushan, china, mw 6.6 earthquake from field investigations and preliminary analyses,landslides, 12(2) (2015b) 365-385. c. xu, x. w. xu, j. b. h. shyu, w. j. zheng and w. min, landslides triggered by the 22 july 2013 minxian-zhangxian, china, mw 5.9 earthquake: inventory compiling and spatial distribution analysis,journal of asian earth sciences, 92(2014b) 125-142. 许冲, 徐锡伟, 沈玲玲, 窦帅, 吴赛儿, 田颖颖, 李西, 2014 年鲁甸 ms 6.5 地震触发滑坡编录及其对一些地震参数 的指示, 地震地质, 36(4) (2014) 1186-1203. c. xu, x. xu, x. yao and f. dai, three (nearly) complete inventories of landslides triggered by the may 12, 2008 wenchuan mw 7.9 earthquake of china and their spatial distribution statistical analysis,landslides, 11(3) (2014c) 441-461. 谢洪斌, 杨雪, 罗真富, 模糊证据权法在地震滑坡危险度区 划中的应用, 中国安全科学学报, 21(8) (2011) 164-170. 许冲, 2008 年汶川地震前的中国大陆地震滑坡研究, 科技 导报, 32(16) (2014) 63-77. s. lee and j. choi, landslide susceptibility mapping using gis and the weight-of-evidence model,international journal of geographical information science, 18(8) (2004) 789-814. 王志旺, 李端有, 王湘桂, 证据权法在滑坡危险度区划研究 中的应用, 岩土工程学报, 29(8) (2007) 1268-1273. r. k. dahal, s. hasegawa, a. nonomura, m. yamanaka, s. dhakal and p. paudyal, predictive modelling of rainfall-induced landslide hazard in the lesser himalaya of nepal based on weights-of. geomorphology, 102(3) (2008) 496–510. 赵艳南, 牛瑞卿, 基于证据权法的滑坡危险性区划探索, 地 理与地理信息科学, 26(6) (2010) 19-23. c. xu, x. xu, y. h. lee, x. tan, g. yu and f. dai, the 2010 yushu earthquake triggered landslide hazard mapping using gis and weight of evidence modeling,environmental earth sciences, 66(6) (2012a) 1603-1616. c. xu, x. w. xu, f. c. dai, j. z. xiao, x. b. tan and r. m. yuan, landslide hazard mapping using gis and weight of evidence model in qingshui river watershed of 2008 wenchuan earthquake struck region,journal of earth science, 23(1) (2012b) 97-120. 张艳玲, 南征兵, 周平根, 利用证据权法实现滑坡易发性区 划, 水文地质工程地质, 39(2) (2012) 121-125. e. h. shortliffe and b. g. buchanan, a model of inexact reasoning in medicine. mathematical biosciences, 23(3) (1975) 351-379. d. heckerman, probabilistic interpretations for mycin's published by atlantis press copyright: the authors 243 certainty factors,arxiv preprint arxiv:1304.3419, (2013). 兰恒星, 伍法权, 王思敬, 基于 gis 的滑坡 cf 多元回归模 型及其应用, 山地学报, 20(06) (2002) 732-737. 兰恒星, 王苓涓, 周成虎, 云南小江流域滑坡关键影响因子 研究. 中国地质灾害与防治学报, 40(01) (2003) 103-109. 许冲, 戴福初, 姚鑫, 陈剑, 涂新斌, 曹琰波, 肖建章, 基于 gis 的汶川地震滑坡灾害影响因子确定性系数分析, 岩 石力学与工程学报, 29(增 1) (2010a) 2972-2981. 陈晓利, 冉洪流, 祁生文, 1976 年龙陵地震诱发滑坡的影响 因子敏感性分析, 北京大学学报(自然科学版), 45(1) (2009) 104-110. 许冲, 戴福初, 姚鑫, 赵洲, 肖建章, 基于 gis 与确定性系 数分析方法的汶川地震滑坡易发性评价, 工程地质学报, 18(1) (2010b) 15-26. 刘丽娜, 许冲, 陈剑, gis 支持下基于 cf 方法的 2013 年 芦山地震滑坡因子敏感性分析, 工程地质学报, 22(6) (2014a) 1176-1186. 晏同珍, 伍法权, 殷坤龙, 滑坡系统静动态规律及斜坡不稳 定性空时定量预测, 地球科学-中国地质大学学报, 14(2) (1989) 117-133. 阮沈勇, 黄润秋, 基于 gis 的信息量法模型在地质灾害危险 性区划中的应用, 成都理工学院学报, 28(01) (2001) 89-92. 高克昌, 崔鹏, 赵纯勇, 韦方强, 基于地理信息系统和信息 量模型的滑坡危险性评价——以重庆万州为例, 岩石力 学与工程学报, 25(5) (2006) 991-996. 刘应辉, 汶川地震区都汶公路沿线崩塌滑坡灾害特征与评 价.（兰州大学, 兰州, 2009). 陶舒, 胡德勇, 赵文吉, 范一大, 王志恒, 基于信息量与逻 辑回归模型的次生滑坡灾害敏感性评价——以汶川县 北部为例, 地理研究, 29(09) (2010) 1594-1605. 牛全福, 程维明, 兰恒星, 刘勇, 颉耀文, 基于信息量模型 的玉树地震次生地质灾害危险性评价, 山地学报, 29(02) (2011) 243-249. c. xu, assessment of earthquake-triggered landslide susceptibility based on expert knowledge and information value methods: a case study of the 20 april 2013 lushan, china mw6. 6 earthquake,disaster advances, 6(13) (2013) 119-130. 邢秋菊, 赵纯勇, 高克昌, 郭跃, 基于 gis 的滑坡危险性逻 辑回归评价研究, 地理与地理信息科学, 20(03) (2004) 49-51. 刘艺梁, 殷坤龙, 刘斌, 逻辑回归和人工神经网络模型在滑 坡灾害空间预测中的应用, 水文地质工程地质, 37(5) (2010) 92-96. m. j. garcia-rodriguez, j. a. malpica, b. benito and m. diaz, susceptibility assessment of earthquake-triggered landslides in el salvador using logistic regression,geomorphology, 95(3-4) (2008) 172-191. f. su, p. cui, j. zhang and l. xiang, susceptibility assessment of landslides caused by the wenchuan earthquake using a logistic regression model.,journal of mountain science, 7(3) (2010) 234-245. 许冲, 徐锡伟, 逻辑回归模型在玉树地震滑坡危险性评价 中的应用与检验, 工程地质学报, 20(3) (2012a) 326-333. c. xu, x. xu, f. dai, z. wu, h. he, f. shi, x. wu and s. xu, application of an incomplete landslide inventory, logistic regression model and its validation for landslide susceptibility mapping related to the may 12, 2008 wenchuan earthquake of china,natural hazards, 68(2) (2013) 883-900. 李晓璇, 马海建, 基于 logistic 模型的地震次生崩滑危险 性评价——以汶川县为例, 地震, 33(2) (2013) 63-70. 文海家, 胡东萍, 王桂林, 汶川县地震滑坡易发性 lr 与 nn 评价比较研究, 土木工程学报, 47(增 1) (2014) 17-23. w. l. li, r. q. huang, c. tang, q. xu and c. van westen, co-seismic landslide inventory and susceptibility mapping in the 2008 wenchuan earthquake disaster area, china,journal of mountain science, 10(3) (2013b) 339-354. l. ermini, f. catani and n. casagli, artificial neural networks applied to landslide susceptibility assessment,geomorphology, 66(1) (2005) 327-343. c. h. chen, c. c. ke and c. l. wang, a back-propagation network for the assessment of susceptibility to rock slope failure in the eastern portion of the southern cross-island highway in taiwan,environmental geology, 57(4) (2009) 723-733. s. lee and d. evangelista, earthquake-induced landslide-susceptibility mapping using an artificial neural network,natural hazards and earth system science, 6(5) (2006) 687-695. 陈晓利, 赵健, 叶洪, 应用径向基概率神经网络研究地震滑 坡, 地震地质, 28(3) (2006) 430-440. m. garcía-rodríguez and j. malpica, assessment of earthquake-triggered landslide susceptibility in el salvador based on an artificial neural network model,natural hazards and earth system science, 10(6) (2010) 1307-1315. 许冲, 徐锡伟, 基于 gis 与 ann 模型的地震滑坡易发性 区划, 地质科技情报, 31(3) (2012b) 116-121. y. li, g. chen, c. tang, g. zhou and l. zheng, rainfall and earthquake-induced landslide susceptibility assessment using gis and artificial neural network,natural hazards and earth system sciences, 12(8) (2012) 2719-2729. v. vapnik (eds), the nature of statistical learning theory (springer science & business media, 2000). n. cristianini and b. scholkopf, support vector machines and kernel methods: the new generation of learning machines,ai magazine, 23(3) (2002) 31-41. 戴福初, 姚鑫, 谭国焕, 滑坡灾害空间预测支持向量机模型 及其应用, 地学前缘, 14(6) (2008) 153-159. 谭龙, 陈冠, 王思源, 孟兴民, 逻辑回归与支持向量机模型 在滑坡敏感性评价中的应用, 工程地质学报, 22(01) (2014) 56-63. c. xu and x. w. xu, spatial prediction models for seismic landslides based on support vector machine and varied kernel functions: a case study of the 14 april 2010 yushu earthquake in china,chinese journal of geophysics, 55(6) (2012) 666-679. 许冲, 徐锡伟, 基于不同核函数的 2010 年玉树地震滑坡 空间预测 模型研 究 , 地球物 理学报 , 55(9) (2012c) 2994-3005. published by atlantis press copyright: the authors 244 c. xu, x. xu, f. dai and a. k. saraf, comparison of different models for susceptibility mapping of earthquake triggered landslides related with the 2008 wenchuan earthquake in china,computers & geosciences, 46(2012c) 317-329. n. m. newmark, effects of earthquakes on dams and embankments,geotechnique, 15(2) (1965) 139-160. r. w. jibson, e. l. harp and j. a. michael, a method for producing digital probabilistic seismic landslide hazard maps,engineering geology, 58(3) (2000) 271-289. 王涛, 吴树仁, 石菊松, 辛鹏, 基于简化 newmark 位移模 型的区域地震滑坡危险性快速评估——以汶川 m_s8. 0 级地震为例, 工程地质学报, 21(1) (2013) 16-24. a. refice and d. capolongo, probabilistic modeling of uncertainties in earthquake-induced landslide hazard assessment,computers & geosciences, 28(6) (2002) 735-749. 陈晓利, 袁仁茂,d 庾露, newmark 方法在芦山地震诱发滑 坡分布预测研究中的应用, 地震地质, 35(3) (2013) 661-670. t. l. saaty (eds), what is the analytic hierarchy process? (springer, 1988). 陈晓利, 人工智能在地震滑坡危险性评价中的应用. (中国 地震局地质研究所, 北京, 2007). 许冲, 戴福初, 姚鑫, 陈剑, 涂新斌, 孙瑜, 王志一, gis 支 持下基于层次分析法的汶川地震区滑坡易发性评价, 岩 石力学与工程学报, 28(增 2) (2009) 3978-3985. 杜军 and 杨青华, 基于 gis 与 ahp 耦合的汶川震后次生地 质灾害风险评估. 中国水土保持, 11) (2009) 14-16. 张建强, 范建容, 严冬, 郭芬芬, 苏凤环, 地震诱发崩塌滑 坡敏感性评价——以北川县为例, 四川大学学报(工程 科学版), 41(03) (2009) 140-145. c. tang, j. zhu and j. liang, emergency assessment of seismic landslide susceptibility: a case study of the 2008 wenchuan earthquake affected area,earthquake engineering and engineering vibration, 8(2) (2009) 207-217. 刘丽娜, 许冲, 徐锡伟, 陈剑, gis 支持下基于 ahp 方法的 2013 年芦山地震区滑坡危险性评价, 灾害学, 29(04) (2014b) 183-191. published by atlantis press copyright: the authors 245 regional risk assessment of earthquake-triggered landslides 山区大地震通常触发大规模滑坡并诱发严重的地质灾害。近年来，特别是2008年汶川地震之后，国内学者高度重视区域地震滑坡危险性评价研究。它以工程地质类比法为理论基础，去估计具有相同或类似环境条件的区域尺度地震滑坡的危险性。从前人工作来看，区域危险性评价主要包括2方面：建立地震滑坡数据库和运用数学统计模型进行危险性评价。数据库建立以区域地震滑坡解译为基础，地震滑坡危险性评价常用方法包括：证据权重模型、确定性系数法（cf）、信息量法、逻辑回归模型（lr）、人工神经网络（ann）、支持向量机法（svm）、n... 关键词：地震；滑坡；危险性评价 1. 引言 2. 地震滑坡危险性评价概述 3. 地震滑坡危险性评价的步骤 4. 地震滑坡危险性评价的方法 4.1.数据驱动方法 4.1.1.证据权重模型 4.1.2.确定性系数法 4.1.3.信息量法 4.1.4.逻辑回归模型 4.1.5.人工神经网络 4.1.6.支持向量机 4.1.7.newmark位移模型 4.2.知识驱动方法 5. 实例分析 6. 结论与展望 6.1.结论 6.2.展望 致谢 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <feff004200720075006700200069006e0064007300740069006c006c0069006e006700650072006e0065002000740069006c0020006100740020006f007000720065007400740065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400650072002c0020006400650072002000620065006400730074002000650067006e006500720020007300690067002000740069006c002000700072006500700072006500730073002d007500640073006b007200690076006e0069006e00670020006100660020006800f8006a0020006b00760061006c0069007400650074002e0020004400650020006f007000720065007400740065006400650020005000440046002d0064006f006b0075006d0065006e0074006500720020006b0061006e002000e50062006e00650073002000690020004100630072006f00620061007400200065006c006c006500720020004100630072006f006200610074002000520065006100640065007200200035002e00300020006f00670020006e0079006500720065002e> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 119–120 doi: https://doi.org/10.2991/jracr.k.200522.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr corrigendum corrigendum to “competency of the infantry troops of the nepalese army in disaster response” [journal of risk analysis and crisis response 9(2), (2019), 62–73] shrijan bahadur malla1, shuichi hasegawa2, ranjan kumar dahal3 faculty of engineering, graduate school of engineering, kagawa university, 2217-20 hayashi-cho, takamatsu, kagawa 761-0396, japan department of safety systems construction engineering, kagawa university, takamatsu city, japan central department of geology, tribhuvan university, kritipur, nepal123 in the original article, there were mistakes in introduction section paragraph 11, paragraph 13, paragraph 14 and paragraph 15. these paragraphs contain wrong abbreviation ‘skk’. the abbreviation should be ‘ssk’. the wrong abbreviation was also written in paragraph 1 of method section, table 1 and paragraph 2 of effect of rank and training in competition section, and paragraph 1 and 3 of discussion section. similarly paragraph 11 and paragraph 15 of introduction section, and paragraph 1 of method of analysis section also contain wrong word ‘nepali’. they should be replaced by word ‘nepalese’. the corrected paragraphs and table are given below. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). doi of original article: https://doi.org/10.2991/jracr.k.190703.002 1. introduction but in contrary to this urgency, disaster response preparedness of those regular infantry troops has remained in shadow. it is because the organization is moreover focused on strengthening its war-waging capability. the secondary mandate like disaster management has been facing chronic resource shortage (poudel, 2016). in such a scenario full-fledged preparedness of such troops in disaster response is hardly attainable. and this reality was clearly evident during the response operations of gorkha earthquake 2015. various national and international after-action reports of gorkha earthquake have emphasized that there is a need to ‘enhance search and rescue capability of nepalese army’ (moha, 2015; moha, 2016). they have also stated that the troops were less trained and less equipped for disaster response (manandhar et al., 2017; grunewald & burlet, 2016). furthermore, disaster response knowledge consists of two categories i.e. ‘soft skill knowledge (ssk)’ and ‘technical skill knowledge (tsk)’ (barelli et al., 2014). the ssk is non-technical, interpersonal and cognitive like principles, leadership, teamwork, and communication42 whereas the tsk is procedural and systematic to function technical work like search and rescue, debris management, dead body management, casualty evacuation etc (poul, 2018). and both the knowledge is prerequisite for the normal infantry troops since they have to carry out multiple functions during a disaster scenario. the gorkha earthquake is an example how these same troops carried out series of response operations like search and rescue, casualty evacuation, debris management, dead body management simultaneously (barsky et al., 2007). when it comes to competency building, the national strategy for disaster risk management 2009, the national framework for disaster management 2013 and the disaster risk reduction and mitigation act 2017 have also guided the nepalese army to carry out preparedness activities of its troops. similarly, the nepalese army disaster management doctrine has also defined that training, exercise, and rehearsal are the regular tasks for the competency building. in this backdrop, it is important that the knowledge and preparedness of normal infantry troops should be studied. but hardly this issue has garnered interest in the academic research field. julia hornyacsek (2018) admits that the ‘disaster response competency of such regular infantry forces has hardly been studied’. in this regard, this research aims to study the disaster response competency of regular infantry troops of the nepalese army setting following research questions (1) whether the ssk and tsk of the regular infantry troops of the nepalese army adequate enough to carry out effective response operations? (3) is there any gap in the ssk and tsk, and preparedness of the regular infantry troops of the nepalese army? https://doi.org/10.2991/jracr.k.200522.001 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ https://doi.org/10.2991/jracr.k.190703.002 120 s.b. malla et al. / journal of risk analysis and crisis response 10(3) 119–120 2.4. method of analysis fundamentally, the descriptive analysis method and inferential analysis method were adopted during research that helped to examine the relationship between the nepalese army’s disaster response competency building initiative and its effect on personnel competency. for the purpose, three key independent variables i.e. ssk, tsk, and disaster preparedness adaptation were considered as study factors. these variables were again analyzed on the basis of two dependent variables i.e. ‘rank’ (officer, jco and other ranks) and ‘training’ (trained and untrained troops). whether these variables make any significant difference in the study participants’ knowledge and preparedness or not was the aim of the analysis process. similarly during the second survey prevailing perception of senior army officers on the engagement of the nepalese army in disaster response was analyzed. 3.4. effect of rank and training in the competency table 1 correlation matrix of competency competency ssk tsk preparedness activities ssk pearson correlation 1 0.205* –0.051 sig. (2-tailed) 0.041 0.566 tsk pearson correlation 0.205* 1 –0.193* sig. (2-tailed) 0.041 0.041 preparedness activities pearson correlation –0.051 –0.193* 1 sig. (2-tailed) 0.566 0.041 *correlation is significant at the 0.05 level (2-tailed). to study whether the participants’ training in disaster response makes any significant difference in their competency linear regression test was conducted. as per the equation derived (f(3, 82) = 11.384, p < 0.000) the disaster response training of participants is statistically significant to their response competency as a whole (see tables 2). but when compared individually, coefficients test result illustrated that only the training has statistically significant relation with preparedness activities where the p-value is less than 0.05 (p < 0.000) and t value is 5.539 (see table 3). that means the trained troops were more prepared than untrained troops, whereas with ssk and tsk training factor do not have any significant relation since p values are higher than 0.05. 4. discussion this research has ultimately helped to understand the competency of the general infantry troops of the nepalese army in disaster response operations. indeed, in the army, junior officers and jcos are the key personalities. they are the one to lead disaster response operations in the field. but the result showed that the junior officers and jcos have limited ssk. the issues of civil-military relation, a priority of their engagement during a disaster, and the responsibilities of local government in disaster management are much important at their respective level. it is because at the local level they are the one to play interface between civilian authorities, civil societies and responding troops. they have to establish effective communication, coordination, and cooperation with those civilian bodies and civil communities at the time of preparedness and response. in the void of such knowledge, they can’t drive the venture in an effective way. in the aspect of training, more than 50% of the participants were found participated in some kind of disaster-related training and exercise in their career. it is true that following the essence of national policies and guidelines (moha, 2017; nepalese army, 2017) the organization has credibly invested to train its infantry troops in disaster response training like collapsed structure search and rescue training, medical first responders training, dead body management training, deep water rescue training, and sphere training (nepalese army, 2017). but the result of ssk and tsk doesn’t warrant qualitative conduct of such training. the understandings of trained participants are low in many fundamental issues. and there is not much difference amongst trained troops’ and untrained troops’ responses. surprising only 22% of the trained participants replied that search and rescue is the most suitable tasks for them for disaster response. this sort of gap clearly indicates the need for better orientation of the infantry troops in disaster response venture. further research is suggested to investigate in-depth issues of this aspect. atlantis press journal style analogies in entity risk mechanics milík tichý department of civil engineering, czech technical university prague 16629, czech republic e-mail: milik.tichy@volny.cz abstract companies exposed to hazards can be paralleled to building and engineering structures subjected to physical loads; see. important analogies can be identified in various specific concepts, e.g., in size effect, instability, fatigue, and in many others. keywords: entity, structure, company, risk, hazard, risk concern, structural mechanics, size effect, instability 1. introduction in (m. tichý, 2012) basic analogies between notions of structural mechanics and risk analysis were shown. for this purpose, structure subjected to loads was paralleled to company exposed to hazards. similarly as loads create stress and strain in structures, risk, rs, and risk concern, cn, are generated in companies whenever hazards are identified by individuals or groups involved directly or indirectly in company's activities. the science of risk management is considerably younger than structural mechanics. though the explicit definition of risk dates back to 1718, cf. (m. tichý, 2012), a systematic theoretical approach started only in 1968. see the seminal paper by (c. starr, 1969) that moved the risk theory from the space of the games theory and actuarial sciences space to the economic and societal decision making area. 2. analogies discussed for the reader's convenience, basic analogies between loads and hazards identified in (m. tichý, 2012) are summarized in table 1. in this paper, the idea of load/hazard analogies is being further developed. some specific notions of structural mechanics are matched with notions of company risk analysis to show the viability of the analogies concept. table 1. load/hazard analogies as identified in (m. tichý, 2012) level structural mechanics  risk analysis external load, l  hazard, hz internal load effect, f  hazard scenario, sc stress, σ  risk, rs strain, ε  concern, cn constitutive law stress-strain function [σ, ε]  risk-concern function [rs, cn] governing requirement structural reliability requirement  risk acceptability requirement to avoid misunderstandings, it has to be emphasized that, similarly as in (m. tichý, 2012), behavior of companies exposed to hazards is being here paralleled to the behavior of load-carrying structures subjected to physical loads, but definitely not to the behavior of structures exposed to hazards. note also that risk analysis is not compared with structural mechanics; journal of risk analysis and crisis response, vol. 4, no. 2 (june 2014), 96-107 published by atlantis press copyright: the authors 96 willieb typewritten text received 12 october 2013 willieb typewritten text accepted 23 april 2014 willieb typewritten text milík tichý they both are undeniably two independent branches of knowledge, which have nothing in common. at many points of this paper, detailed listings of examples could be given to illustrate the concepts and notions presented. unfortunately, the number of examples had to be limited. the reader will surely understand this restriction; space remains thus open for his/her creative imagination. 3. review of concepts and notions in general, the original concept of "risk triplet", introduced by (s. kaplan & b. j. garrick, 1981), is respected by the author; see also (s. kaplan, 1997). risk analysis notions applied in (m. tichý, 2012), i.e., hazard, hazard scenario, and risk, are used. upside aspects of risks are not pursued. the meaning and definition of "risk" are not discussed here. "risk" is a complicated, multidimensional concept. the definition of risk has been recently discussed in a series of papers; see, e.g., (h. merkelsen, 2011). only hazards are being studied here. note that hazard uncertainties are not identical with risk uncertainties. it happens in risk analyses that a hazard of high subjective severity gets paired with a very low risk, owing to very small probabilities of materialization of possible hazard scenarios. though the value of risk is taken as a s tarting point of decisions in general, hazard circumstances exist where decisions shall be governed by hazard severity first and foremost. inversely, it happens that some hazards get overrated at the cost of hazards that remain unobserved. recall that in (m. tichý, 2012) the term "entity" universally refers to all objects and processes of any kind – tangible, intangible, or mixed. most entities are components of supra-entities, and, similarly, many entities are composed of sub-entities. hierarchies of entities can be thus easily found in whatever framework. a component can be a cl early defined unit in the respective entity, or a unit spread in the systems. the links between units belong also to the set of entity components, as well as the respective joints. throughout the paper, the summary term person stands either for individuals or for groups, both of diverse qualities and positions. a person can be internal, existing in the framework of the company (e.g., owners, stockholders, ceo, staff, management departments, individual employees, an ad hoc expert team), or external, existing beyond the company and possessing some permanent or transient relation to it (as, e.g., customers, business partners, lawyers, clients, banks, government).—there of course are several other classifications of persons; e.g., natural and judicial persons are distinguished in legal procedures. time and space must never be ignored in any analyses. while time has always the same physical quality, several kinds of spaces can be of interest in risk analyses. in addition to the traditional space, defined geometrically, a management space, marketing space, operation space, cost space, and others shall be considered, taking into account their specific, time dependent properties. modifying attributes like slow/fast/quick, small /large, light/heavy, immediate/brief/long, sudden/ gradual, insignificant/significant and similar shall be, in general, understood as indicative, without any particular quantifiable meaning. e.g., a statement saying that "something is small" shall be examined and assessed in the particular context discussed. similarly, attributes empirical/exact, must be taken in view of circumstances. semi-exact and semi-empirical solutions are frequent in any science, except mathematics. 4. hazards recall here briefly some general qualities of hazards: basically, hazards originate from one of two distinct groups of sources: (1) anthropogenic sources – hazards come from human activity/non-activity and human decisions/nondecisions (e.g., careless market analyses, insurance fraud, insufficient supervision of employees) (2) natural sources – hazards come from phenomena independent on human activities (e.g., seismic waves, inclement weather) hazards from sources of both origins are frequently interwoven; a typical example: floods in areas where the river flow (nature) is regulated by longitudinal or transversal dams (man). hazards can be fully subjective, created and dwelling solely in minds of persons, or objective, generated in minds by existing or expected phenomena, tangible or not. the boundary between objectivity and subjectivity is in most situations fuzzy. it is important to stress that a hazard from whatever source, single or mixed, objective or subjective, identified in the risk analysis of a project, may never get materialized during the lifetime of the entity at all. published by atlantis press copyright: the authors 97 analogies in entity risk mechanics anyway, hazards of any kind breed feelings of uncertainty, risk concern, and possibly fear or panic. since any uncertainty costs money in various forms, hazards of whatever nature or origin affect funds and money flows. therefore, hazards can irreversibly damage the entity exposed, without having been materialized. problems of hazard qualification will not be discussed in this study. hazard severity or any other measure of hazard is a h ighly subjective concept, composed of many factors that are mainly associated with human cognitive properties. the perception of hazards and the risk concern are often willingly or unwillingly affected by media (d. gardner, 2009). 5. structure and company as before in (m. tichý, 2012), two common entities, structure and company, are taken as objects of this study. attention will be paid to some selected particular concepts of structural mechanics, aiming at their transfer into the language of companies. the reader is referred to table 2, where the principal characteristics tabel 2. some characteristics of the two entities considered characteristic structure company basic role in economy mainly passive: carrying, sheltering; consuming money mainly active: producing goods or supplying services; generating profit system properties adjustable closed system adjustable open system adaptability properties rigid with necessary flexibility and ductility flexible and ductile with necessary rigidity governing parameters strength and stiffness assets and liabilities; sales and earnings; liquidity and profitability environmental specification specified in physical terms specified in economic, social, and geographic terms components carrying members (beams, columns, foundations); well outlined in majority individuals (e.g., employees) and groups (e.g., management teams) links and joints (internal and external) tangible; defined predominantly intangible; administrative, managerial, societal; tangible links and joints exist description predominantly in physical terms predominantly in business, financial, industrial, and analogous terms calculation models exact or quasi-exact models exist speculative models prevail physical testing widely used impracticable input properties tested and controlled not tested or testable with difficulties only data availability data available in sufficient quantity and quality data are scarce, and must be, in majority, estimated uncertainties low degree of uncertainty and fuzziness high degree of uncertainty and fuzziness research possibilities physical and theoretical theoretical legal aspects standards and codes of practice contracts based on common law or civil law; regulations lifetime often longer than expected often shorter than expected published by atlantis press copyright: the authors 98 milík tichý of the two entities are listed, with no claim to completeness. note that structures are always exposed to hazards of various origin (e.g., tornadoes, terrorist attacks, bad maintenance, misuse), and, similarly, companies are always subjected to loads (e.g., debts, taxes, fees for legal processing). while hazards to structures are similar to hazards to companies, loads on companies are entirely different from loads on structures. note further: a s tructure is one of several components of a constructed facility, i.e. of a building, highway bridge, tunnel, etc.; a company is one of many components of the economic system. the main features of loads and hazards, important for the examination of the load/hazard analogies are shown in table. 3. it is generally known that loads can bring a structure to failure. hazards affect a company in similar way. the sole rumors of hazards can have negative contribution to the image of the company, so that its business health becomes doubtful, and, as a result, the company's image starts to deteriorate. banks watch for signs of problems and for possible leaks of internal rumors, and credits and bank sureties might be refused. hazards of this kind can get materialized; this may stimulate a partial or entire collapse of the company's business activities, leading to its decline, and finally to its dissolution. 6. particular concepts and notions 6.1. defects the many definitions of defect can be summarized in a simple statement: a defect is any noxious deviation from the expected properties of a system/object/process that were assumed in its creation and that are expected during its lifetime. defects are system properties that can be inherent, inherited, and implanted. a defect-free system does not exist; whether such a s tatement is true or not, can be neither confirmed, nor challenged. the common quality of defects is randomness. it may be argued that many defects, e.g., those caused by sabotage, violence, or war activities are not random. yet, even sabotage, street violence, and war are, in the long run, random sociopolitical phenomena. sooner or later, due to the presence and development of a s pecific defect, stress and strain gets created in the structural material, with undesired effects on the reliability of the structure. therefore, defects can be regarded as a sort of structural loads. yet, in structural design, defects are never treated in such a way. not only material and members may be defective. defects of the loading process and also defects of use and maintenance are enemies of construction projects. any defect, like any load, affects the reliability of the structure and can result in its partial or total collapse. defects can be also a cause of loss of a company can become defective from its very beginning, or, more exactly, the management processes can be defective even before the company gets ever set up. innate defects generate future hazards to the company, since their existence and possible materialization can be followed by direct or indirect financial loss, be it i mmediate or delayed. such possibilities are often recognized only too late. it should be mentioned that the company itself can be seemingly defect free; nevertheless, defects can be r ooted in the environment and, unfortunately, in human decisions. typically, unfair business practice, which is in fact a market table 3. characteristic properties of structural loads and company-affecting hazards compared property loads on structures hazards to companies general character tangible; objective; passive intangible; predominantly subjective; active deterministic models physical – well established and well defined managerial – scarce and fuzzy random uncertainties acceptably described by estimated probability models models not available; uncertainties mainly linked to the character of business published by atlantis press copyright: the authors 99 analogies in entity risk mechanics serviceability of the constructed facility. in structural design, regulations do not provide any explicit guidance on how to deal with the random nature of defects. the problem is covered reactively by additional reliability parameters, and proactively by quality control at various phases and levels of construction projects. defect, is a dangerous source of hazards. that can further result in internal strains, disputes and litigation, loss of mar ket position, and many other deleterious consequences. u nder specific conditions risk concern can reach fear or panic. hazard feelings provoked by observed defects can create increased risk concern in persons involved. 6.2. failure and collapse as soon as a system is borne, it starts its way to gradual decay and final death. in terms of history, the decay may be brief or long. on the route, the system can sometimes fail to meet the requirements set or expected by owners, users, the public, and other individuals or groups interested. periods exist when a system is close to stopping its service, followed possibly by periods of regeneration. nevertheless, after having passed all turns and oscillations of development, the system collapses. depending on the intrinsic properties of the system and its environment, a co llapse can have different forms: e.g., partial or total degradation, ruin. under increasing load a structure goes primarily through a stage when its actual or future normal use is threatened, limited, or entirely restricted. finally it starts to become out of use. the reliability margin of the structureload system gets consumed, and the structure collapses either partially or totally. two basic modes of structural failure exist: ductile and brittle. typically, the former failure mode is signaled there are many obvious causes of a company failure: bad management, collapse of markets, fraud, corporate crime, and many, many others. yet, a company can simply fail by accumulation of hazards. whenever hazards increase in number and severity, the persons' risk concern grows and may achieve the distress, dread, and panic levels. at the same time, the risk capacity of the company gets weakened, and staying in business may become by easy to understand warnings of pending or approaching loss of reliability. whenever brittle fracture is imminent, no signs of pending disturbances are given to untrained individuals (e.g., users of the constructed facility) or even to experienced structural engineers. a structure made of brittle material or functioning as a quasibrittle system would collapse at a specific level of load. yet, at the same load level, a similar structure made of ductile material properties and subjected to same loads may survive. difficult or impossible. this phenomenon is usually underrated and misjudged; hazards accumulate step-by-step and remain unobserved until the hazard portfolio had reached an unacceptable level. the consequential damage may become excessive. companies can suddenly collapse without any preliminary warning, or with warnings that are not taken seriously by persons concerned, particularly by the external ones. many examples can be given. 6.3. progressive collapse any working stationary or non-stationary system can collapse at an unexpected moment, in an unexpected way, for strange reasons, at surprising conditions, and with unpredictable collapse scenarios. a simple and isolated defect of any kind, tangible or intangible, locally significant or insignificant, can trigger uncontrolled behavior, distributed along timeline and over spaces considered. in structures, the failure network belonging to the formation of a single defect is characterized by a system of loads and corresponding load effects that differ from those met under defect free conditions. the intensity and arrangement of load effects change because of the spreading structural deterioration. changes in the carrying system whenever a company is found to be exposed to a hazard that had not been predicted, a s tudy of possible hazard scenarios shall be performed. it happens that a chain of hazards is thus identified. nevertheless, it is hardly possible to estimate all hazard scenarios conceivable. detailed awareness of a hazard may or may not published by atlantis press copyright: the authors 100 milík tichý generate further structural damage, and so a s ort of butterfly effect can start. the process can either get stopped by the properties of the load-structure system or further degenerate up to collapse. structures can collapse either gradually or suddenly. in the first case, systems emit significant warnings like cracks, deflections and other deformations; such effects can arouse the sensory system in humans or in man-made indicators construed for this purpose. they can be considered precursors of imminent failure. because of uncertainties of the input information, architects and engineers are not able to predict, during the design period, all possible failure scenarios. it thus happens that many structural systems fail in an unpredicted and very surprising way. reduce its severity. typically however, any accumulation and regrouping of hazards have a n egative effect upon the internal persons. as a rule, external persons can get worried, too. accumulation and chaining of hazards increase feelings of uncertainty, thus amplifying the existing risk concern. the randomness of hazards and of possible hazard scenarios affects all persons involved. to avoid a domino effect in hazard spread, advanced companies arrange for business continuity management, which, in case of adverse events, ensures an effective response to hazard scenarios predicted, and sets barriers to spread of consequences. the mere existence of disaster recovery plans and similar proactive or reactive documents may have a favorable effect on internal persons, as far as their hazard feelings are concerned. 6.4. instability problems instability is a concept applied in various branches of science and technology as well as in many other branches – psychology, chemistry, social sciences, in politics. though the term "instability" has qualitatively different meanings in each branch, the terms stable, metastable, unstable, and indifferent designate the quality of the respective state of equilibrium in whatever, tangible or intangible, connotation. in structural mechanics three basic types of instability are dealt with: instability of shape: a load-subjected structure changes the state of deformations and internal forces in such a way that sudden, sometimes disastrous changes in the system's shape and arrangement ensue. instability of position: because of overload and other reasons, a s tiff part of the carrying system can start moving with respect to its environment; displacements may affect the state of equilibrium and result in toppling, sliding, and submerging or uplift. dynamic instability: under quickly repeated loads structural members and the structure as a complex can start vibrating. due to dynamic instability phenomena, the amplitude of vibrations can increase without any increase of the load level. the rating agencies often declare that the business standing of a specific company or of an entire business branch is "stable" or "unstable". in the vocabulary of rating, attributes "indifferent" or "metastable" sometimes appear, too. under hazards, the company can go through changes unobserved by persons involved. subconscious hazard feelings of unknown origin can create risk concern without any ostensible reason. sudden collapse of the company can ensue. a company exposed to a set of subjective hazards can be forced to move the business into new spaces, with the aim of avoiding materialization of hazards and escaping risk in such a way. vanishing and reappearing of hazards, may disturb the company's internal equilibrium state. 6.5. non-linearity nothing in the material world happens along a straight line, on a p erfect plane, and in a p erfect space. perfect linearity does not exist. seen from distance, everything is curved, more or less. in the everyday life, nonlinearity is either not recognizable at all or without substantial importance for correct descriptions and assessments of phenomena. however, there definitely are many instances where the non-linearity must not be neglected. for many reasons, the behavior of any loaded structure is, in general, non-linear. although a a minor change in hazards affecting a s ingle component, be it d istant from the company's published by atlantis press copyright: the authors 101 analogies in entity risk mechanics perfect linear behavior is non-existent, it is typically one of starting assumptions in structural mechanics. three main sources of non-linearity of the performance of structures can be recognized:  material properties  geometric properties of the structure  load properties combined together, effects of these sources produce a complex pattern of non-linearity. efficient calculation models covering the non-linearity features are nowadays available in the structural mechanics' tool kit; numerical solutions are widely used. note here that soil and rock belong to the family of carrying materials. their properties, definitely non-linear, are often chaotic and thus very close to the substance in which companies exist. much experience can be drawn from this fact. management core or not, may cause severe changes in the comprehensive set of hazards the system is normally exposed to. this is a typical attribute of the non-linear behavior of hazards. no special or general description of the hazard non-linearity is available. most hazards and hazard scenarios are too subjective to allow for such a description. empirical approach is being accepted as entirely sufficient, and the nonlinear amplification of hazard-related problems is tacitly taken as a common fact. the non-linearity of company’s response to hazards is mainly caused by a familiar phenomenon: remedies used in the prevention and in the elimination of the hazards estimated can generate new hazards that can be more severe than the hazards having been removed. consequently, new risks can arise by the management of hazards itself, since the risk portfolio can change its composition and magnitude in dependence of the properties of the company system. 6.6. constitutive law, elasticity and plasticity it is generally not known that mathematical models of phenomena governing the behavior of entities must be based on specific functions, called constitutive laws. yet, at various levels of everyday communication, many notions that are typically bound to constitutive laws, are met and used. fig. 1. stress-strain diagram of a structural material: a – material softens with growing stress/strain, b – material hardens with growing stress/strain expressions like elastic, plastic, inelastic, elasticplastic, and their derivations typically appearing with scientific and engineering connotations are read and heard in social, business, and other framework. fig. 2. risk-concern diagram of a company: a – risk-averse company, b – risk-reverse company; ult – ultimate risk capacity e.g., price elasticity and elastic/inelastic demand often occur in business vocabulary; their meaning is obviously very distant from that being perceived in physical context. these notions are used by nonengineers with only a superficial understanding of what they actually mean and what is their actual importance. a ε σ ultε ult 0 σ b ε σ 0 ultσ ultε ult ult a b rs rs cn cn rsult cnult cnult rsult 0 0 ult ult published by atlantis press copyright: the authors 102 milík tichý often the term "elastic" is substituted for "linear", and "elastic-plastic" or "plastic" for "non-linear", or viceversa; such simplifications may get confusing, and should be either avoided or specified. in general, it has always to be taken into account that whatever constitutive laws are proposed and applied, they are estimates of the objective reality, simplified more or less, depending on the problem solved. consider a s tructural material the stress-strain diagram, [σ, ε], of which, is shown in simple form in fig. 1a. it is assumed that there the loading is monotonous up to the ultimate capacity level, ult[σult, εult]. the diagram shows a straight segment followed by a cu rved segment, and it flattens at increasing stress, σ , and at increasing strain, ε. in plain words: at the beginning, when the stress in material grows, the strain follows the changing stress almost proportionally. then, starting from a cer tain level of stress on, the strain increments accelerate, and the particular diagram becomes concave. it seems like the material is yielding to the load. another simple diagram is shown in fig. 1b. there, the curved segment is convex; in other words, the material is resisting the load growth. in mathematical terms, the two distinct segments of behavior in both drawings are linear and non-linear, respectively. in the beginning of the operating process of a company, hazard-related risk and risk-related concern, cn, remain proportional (see the riskconcern diagram in fig. 2). the company management deals with hazards and risks in regular way, and, as a result, the risk acceptability requirements are complied with. assume now that the hazard portfolio starts to grow in size and magnitude, i.e., the number of hazards and their subjective severities increase (note that both the number and severity of hazards are subjective quantities). then, • a soft company becomes ductile and yields to hazard accumulation • a hard company becomes stiff, defies hazards with less risk concern, but is a l ikely candidate for failure without sufficient warning. removal of hazards may reduce or entirely remove risks, but in majority of cases this does not remove hazard the problem gets more involved if also unloading and possibly also reloading processes are taken into account. fig. 3 shows a very simple case of unloading and reloading. the diagram shows an important particularity: after complete unloading, a certain residual strain, εres , remains in the material. the diagrams shown in fig. 1 through 3 are very simplified. they do n ot say anything about the character of stress (compression, tension), nor do they reflect situations when the load changes from positive to negative, etc. feelings in full. as a rule, persons involved do not believe the hazards were gone. residual risk concern, cnres, remains (for a soft company see fig. 4), and may even continue to increase. constitutive laws related to the behavior of non-engineering entities in hazard environments are a cardinal topic of entity risk mechanics. unfortunately, but not surprisingly, they have not been defined and studied as per today. fig. 3. stress-strain diagram under loading and unloading; max – maximum level of load applied, ult – ultimate capacity possible, εres – residual strain after unloading fig. 4. risk-concern diagram at occurrence and after removal of hazards (a soft company case); max – maximum corresponding to most unfavorable hazard pattern achieved, cnres – residual risk concern 0 σ ultε εε ultσ maxσ εmaxres ult max rs cncnmax rsmax 0 cnres max published by atlantis press copyright: the authors 103 analogies in entity risk mechanics 6.7. kinematics and dynamics wherever movements of solid bodies are possible, two closely related disciplines enter the theoretical models, namely, kinematics that deals with geometric aspects of motion without paying attention to its causes, and dynamics covering the energy aspects of the motion, including properties and behavior of the motion sources. kinematics and dynamics are encountered in many science fields, starting with astronomy, passing through fluid mechanics, and ending perhaps in the domains of microbiology and psychology. in structural mechanics, solid bodies are often subjected to kinematic and dynamic examination. a structure subjected to a fast moving load of whatever origin is likely to fail at lower level of load even if the load effects do not achieve their maximum assumed level expected under conditions of continuous and slowly increasing loading. sometimes, however, the failure level under dynamic loading can be higher than under static load. the principal backgrounds to these phenomena are changes of the stress-strain function under quickly changing loading rates. this feature governs the dynamic behavior of the loadstructure system in an intermingled way. e.g., phenomena like resonance and damping are typically conditioned by the above properties. a company affected by a sudden and unexpected incidence of one or more hazards tends to get damaged earlier than under a slow and continual development of the hazard pattern. while, in the latter case, the company would be able to develop sufficient risk capacity, the sudden occurrence of hazards may affect the available risk capacity in deteriorating way. obviously, the respective risk-concern diagram, [rs, cn], gets modified under hazard impact (fig. 5). though the risk concern does not substantially intensify, higher risks are accepted, and no signals of disastrous hazards and possible linked-up hazard scenarios are felt by persons involved. yet, if hazards change quickly, and their character and severity fluctuate irregularly, the company is not able to adjust and develop an adequate risk capacity. the phenomena of risk resonance and risk damping can be easily identified in the behavior of companies exposed to dynamic hazards. fig. 5. risk-concern diagram after hazard impact 6.8. fatigue repetitions, regular or irregular, of any phenomenon, tangible or intangible, cause changes in the systems affected. such changes remain often unnoticed, but they may suddenly surprise individuals or groups concerned. in certain cases even the originating phenomenon can become modified by its own repetitions itself. the summary term for this phenomenon is fatigue; there is obviously no need to explain its meaning. the process of repeated stress and strain, caused by repeated loadings, affects the strength of material. with certain materials the carrying capacity of the structure by such process may be not markedly affected. however, many structures subjected to repeated loading can often fail at lower level of load than it would be achieved at single monotonously growing load. of course, the maximum level of load governs the phenomenon. for some materials and in some companies exposed to repeated hazards, hazard feelings in persons get aroused, and the severity of hazards is felt menacing. the ensuing risk concerns grow. the company gets fatigued of eliminating the repetitive hazards, and can collapse, unexpectedly and prematurely. on the other hand, many companies adjust their build-up, so as to develop more risk capacity under repeated hazards during their residual expected lifetime. rs cncnult rsult 0 cnimp rsimp hazard impact situation normal situation hazard impact ult ult after imp published by atlantis press copyright: the authors 104 milík tichý structures subjected to repetitions of load, their resistance gets, for a short period, enhanced, but it degrades under repeated loadings later on. notions like low-cycle fatigue and high-cycle fatigue are well known in structural mechanics. each of the phenomena is characterized by its specific fracture pattern. the problems of fatigue of materials and structures is an important issue for structures of various kind. the strengthening phase comes about. finally, companies harassed by repeating hazard occurrences can entirely change their business plans and activities, and even switch to new markets. it depends of course on the frequency and severity of the threatening hazards, and on the adaptability of the individual company. 6.9. rheology rheology is a cross-disciplinary science branch that deals with the time-dependent fluid behavior of loaded and unloaded matter. for a rigorous scientist, fluidity is a common property of matter (panta rhei), but in the everyday practice only that matter is examined as fluid, the flow of which can be observed by human sensory tools. for specific rheological reasons, strain of material subjected to constant stress grows, and, as a r esult, deformations of the structure increase; this phenomenon is known as creep. reversely, under constant strain, stress in material decreases; this phenomenon is called relaxation. moreover, the very strength of material subjected to stress, increasing or not, diminishes with time. because of these material properties, the load effects in the structure subjected consider a company whose organization, location, business plan, and other operation parameters have been stabilized in time. because of rheological phenomena, timedependent long-time changes of components, i.e., units, links, and joints, happen. however, they always get observed with certain delay. persons involved start to perceive such changes and take them as sources of intensifying hazards; this obviously results in to a f ixed load are changing. as a rule, the ensuing changes are irreversible, and affect the reliability of the constructed facility in general. because of these effects, rheological phenomena belong to the structural loads category. structural mechanics is able to describe these phenomena and their consequences by exact or semi-exact models with more or less empirical background. humidity and temperature of the environment play a significant role in such models. growing risk concern. in this way, the risk capacity of the company can weaken. the internal hazard-generated tensions are time-dependent; they either develop or, quite the reverse, fade in time. because of the creeping development of hazards, persons get often hazard-indifferent, or even hazard-blind; this results in carelessness and in ignoring any direct or indirect warnings. behavior of individuals in war zones shall be mentioned here. similarly, people are getting apathetic about terrorist threats. 6.10. size effect the effect of size of an object or of a process, natural or man-made of whatever kinds on their general behavior in time and space is a s ignificant factor well known in many areas: in business and politics, as well as in engineering and management sciences, in agriculture, ecology, and in other spheres. note that the term size can have many meanings: geometric, economic, societal, and other. under increasing load, the behavior of a structure or a s tructural member is affected by their size. the nature of this phenomenon depends primarily upon the properties of the structural material and then also upon the layout of the structure. in large-size solid bodies made of homogeneous brittle material, fracture arrives in a given business environment, the behavior of large companies is in many aspects substantially different from that of small companies. company's readiness to deal with hazards and risks, flexibility, willingness to diversification, and many other properties affect its conduct in the environment in dependence on company's size. note that the notion published by atlantis press copyright: the authors 105 analogies in entity risk mechanics statistically at lower loads than in bodies of small size. however, if the material is ductile, possibilities of sudden failure become less prominent, and the respective structure can survive an increase in load, even when one or more carrying members had been already damaged. time and space conditions play an important role both in brittle and ductile materials. note that space in structural mechanics is, as a rule, geometric. for detailed information on size effect and related problems see, e.g., (z. p. bažant, 2002) and (x. z. hu & k. duan, 2010). of size may have different meanings. further, not only the business size of the company may govern its behavior: the size of the customers, of the field of activities, etc. may have a dominant effect. it cannot be asserted that large size companies are more flexible than small ones. it depends on the company's build-up, whether upside or downside features of the size effect prevail. while some large companies may be typically less sensitive to hazards than small ones, the contrary is often true. large size may have negative effect on the risk capacity. fig. 6. size effect, structure of brittle material fig. 7. size effect, company exposed to hazard; a – small company, failure of one weak element can be catastrophic, b – large company can survive figs. 6 and 7 may help understand the analogy of the size effects in structures and companies. under a certain level of load, a local crack can cause failure of a large body (fig. 6a). in an assembly of small bodies (fig. 6b), the member with the crack will fail under the same level of stress as in the body in fig. 6a, while the other members will reach higher stress levels. thus, the mean level of failure stress (strength) will be higher for sets of small bodies than for a l arge body. size has a unfavorable effect in this case. in a s mall company (fig. 7a), the failure of a weak component, w, may have disastrous consequences, while (fig. 7b) a w eak component may not affect the existence of a l arge company in its entirety. size has a favorable effect in this case. it all depends on what is the meaning of size in the particular case considered. 7. general remarks the ten analogies presented above show that concepts and notions frequent in the domain of structural mechanics have their parallels in the domain of risk analyses. the above list is far from being complete, and each item can be further elaborated in detail. it is obvious that the basics of the load/hazard analogies concept proposed in (m. tichý, 2012) can be further extended, and fundamentals of the entity risk mechanics can be built up. concrete notions of structural mechanics are transferred to the domain of abstract notions related to the behavior of companies or of other non-engineering entities exposed to tangible or intangible hazards coming from various sources. the transfer flows in one direction, i.e., from concrete to abstract (fig. 8). the transfer concept proposed is similar to that applied in the description of the development of languages; see (g. deutscher, 2005). fig. 8 – transfer of concrete notions to abstract notions the route from the first perception of a hazard to the evaluation of the respective risk can never become straightforward. it even happens that the speculative a b crack load hz b w hz a w load notions (concrete) tr an sf er fu nc tio n hazard notions (abstract) published by atlantis press copyright: the authors 106 milík tichý process gets not finalized at all, since necessary data and models are missing. the load/hazard analogies concept might stimulate solutions which would otherwise escape the attention of risk analyzers. at the present stage of development it is obviously not possible to formulate analytical models of entity risk mechanics, erm, that would transfer the engineering knowledge of load subjected structures to hazard exposed entities. the availability of such models seems rather remote even in close futurity. however, the mere understanding of the transfer of notions may affect and improve the thinking of experts, and stipulate development of new ideas. the way to a complete system of erm will logically be lengthy and tedious. nevertheless, the possibilities offered by the load/hazard analogies are, in the author's opinion, wide-ranging and promising. 8. conclusions and suggestions (i) the basic analogies between load-subjected structures and hazard-exposed companies, identified in (m. tichý, 2012), can be further developed and extended to specific problems of the two entities examined. (ii) the analogies discussed and many others can be helpful in case of special problems of risk management and risk analysis, like those of risk allocation, risk prevention, insurance decisions, risk forecasting, and others. (iii) a structural engineer, when properly instructed and guided, would discover unexpected hazards and hazard scenarios of the project even if their sources were beyond the construction domain. (iv) at the beginning of the project decision stage, it may be sensible to include into the body of experts a structural engineer experienced in structural mechanics as a cross-thinking help. (v) since engineering systems of any type (transport, electrical, and others) are subjected to loads, and all social, economic, educational, and other nonengineering entities are exposed to hazards, the concept of load/hazard analogies can be used in many other load-subjected/hazard-exposed entity tuplets. (vi) empirical and speculative research is necessary to develop concepts of erm in full. in particular, study of the risk-concern function, [rs, cn], shall be prioritized. acknowledgements the author would like to thank professor jiří šejnoha, czech technical university, for his insightful suggestions concerning some aspects of the analogies presented. however, any errors of whatever kind are my responsibility. references m. tichý, entity risk mechanics, j. of risk analysis and crisis response. 2(2) (2012), 107-113. c. starr, social benefit versus technological risk, science, 165 (1969), 1232-1238 (adapted from a symposium paper presented by starr in 1968). s. kaplan and b. j. garrick, on the quantitative definition of risk, risk analysis, 1(1) (1981), 11-27. s. kaplan, the words of risk analysis, risk analysis, 17(4) (1997), 407-417. h. merkelsen, the constitutive element of probabilistic agency in risk: a semantic analysis of risk, danger, chance, and hazard. j. of risk research, 14(7) (2011), 881-897. d. gardner, risk. the science and politics of fear (virgin books, london, 2009). j. schijve, fatigue of structures and materials, 2nd edn. (springer, berlin, 2009). z. p. bažant, scaling of structural strength (butterworth-heinemann, london, 2002). x. z. hu and k. duan, mechanism behind the size effect phenomenon, j. of eng. mechanics asce, 136(1) (2010), 60-68. g. deutscher, the unfolding of language (arrow books, london, 2005), 115-143. published by atlantis press copyright: the authors 107 1. introduction 2. analogies discussed 3. review of concepts and notions 4. hazards 5. structure and company 6. particular concepts and notions 6.1. defects 6.2. failure and collapse 6.3. progressive collapse 6.4. instability problems 6.5. non-linearity 6.6. constitutive law, elasticity and plasticity 6.7. kinematics and dynamics 6.8. fatigue 6.9. rheology 6.10. size effect 7. general remarks 8. conclusions and suggestions acknowledgements references research article game theoretic strategies for supplier capability assessment and manufacturing order allocation cheng-kuang wu1,*, , yu-min chuang2 1school of computer science, zhaoqing university, zhaoqing road, zhaoqing city, guangdong province 526021, p. r. china 2department of industrial and systems engineering, chung yuan christian university, taiwan 1. introduction with the globalization of supply chain management, organizations have to find ways to reduce costs through global sourcing, to increase their competitive advantage. supplier evaluation or selection plays an important role in the manufacturing process, particularly when the available budget is restricted in manufacturer expenditures. each enterprise might have its supplier selection criteria, and from these criteria devise their method of selection [1]. each supply chain includes a number of manufacturers, with finite budgets, which need to be allocated to all suppliers in all parts of the manufacturing process. some manufacturers are faced with budgetary limitations. each manufacturer has multiple suppliers each of which in turn possesses different specific capabilities (such as delivery performance and cost of manufacturing). supply chain management (scm) is thus needed to perform a feasible allocation of suppliers during manufacturing process flows. poirier and reiter [2] explained that each member of the supply chain channel which extends from the supplier to the customers must be integrated into the design for marketing, procurement, distribution, and other activities of each organization. dickson [3] used questionnaire responses to identify 23 factors of influence for supplier selection criteria. their study showed product quality, delivery, and past performance history to be critical factors in supplier selection. weber et al. [4] found in a study of vendor selection that suppliers of equipment, output capacity, and technological capability are also related to supplier evaluation for the manufacturing process. chan et al. [5] and maurizio et al. [6] indicated that the cost of the product, which measures supplier capability from the perspective of the supplier, is crucial to support the manufacturing flow. there is a trade-off between increased manufacturer performance and a decrease in manufacturing costs in the selection of multiple suppliers in scm. manufacturer performance shows an upward slope in figure 1 but manufacturer demand states that all other factors remaining equal, supplier service quality is better with higher manufacturer demand, because higher service quality improves manufacturer performance. in other words, the lower the manufacturer’s demand, and the worse the service quality, leading to reduce manufacturer performance. the downward slope in figure 1 indicates that the lower the manufacturer demand, the lower the cost that will be spent on that good. in other words, a smaller manufacturer demand produces higher supplier service quality which results in decreasing cost to the manufacturer; see the downward slope. jayaraman et al. [7] indicated that it is more likely to produce high-quality, low-cost products with the capabilities of suppliers and the requirements of manufacturers. this means that the ideal of high-level manufacturer performance and low-level budgets is difficult to achieve. consequently, we need to find the balance between lower cost (or budgets) and higher manufacturer performance; see figure 1. in previous research, the supplier selection process has been investigated taking the manufacturer’s interests as the starting point but ignoring the point of views from the suppliers. this study applies a r t i c l e i n f o article history received 14 september 2020 accepted 13 october 2020 keywords supplier power value nash equilibrium shapley value supplier allocation a b s t r a c t an effective method is required to determine the amount and priority for the deployment of suppliers for multiple manufacturing processes, particularly when the available budget for each manufacturing process is limited. in this study, we propose an integrated approach for supplier assessment that consists of two game theory models which are designed to recommend manufacturers on how best to choose suppliers given budgetary limitations. in the first model, the interactive behaviors between the key factors representing the manufacturer and the supplier are modeled and analyzed as a two-player and zero-sum game, after which the supplier power value (spv) is derived from the pure or mixed strategy nash equilibrium. in the second model, 12 spvs are used to compute a shapley value for each supplier, in terms of the thresholds of the majority levels in one manufacturing process. the shapley values are then applied to create an allocated set of limited manufacturing orders for suppliers. the experimental results present that the manufacturer can use our approach to quantitatively evaluate the suppliers and easily allocate suppliers within one manufacturing process. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: shapleyvalue@hotmail.com journal of risk analysis and crisis response vol. 10(4); november (2020), pp. 121–129 doi: https://doi.org/10.2991/jracr.k.201014.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-5466-9987 http://creativecommons.org/licenses/by-nc/4.0/ mailto:shapleyvalue%40hotmail.com?subject= https://doi.org/10.2991/jracr.k.201014.002 https://www.atlantis-press.com/journals/jracr 122 c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 a two-stage game model to solve the supplier allocation problem. first, we establish the interactions between the manufacturer and the supplier as a two-player strategy, in a non-cooperative finite and zero-sum game. the proposed payoff functions are calculated by the competency measures (i.e., manufacturer’s requirements and supplier’s capabilities) for four assessment of factors (i.e., quality of products, cost of manufacturing, the technology of the supplier, and delivery performance) affecting the manufacturing flow. the supplier power value (spv) is derived from the payoff matrix. in the second model, a manufacturing process of all suppliers is playing a cooperative game. the shapley value is adopted as a measure to calculate the marginal contribution of all suppliers in the manufacturing process and the mutually agreed upon the allocation of manufacturing budgets. multiple suppliers are united into the coalition groups for calculation of the majority thresholds in such a way as to distribute fair and feasible supplier deployment in a single manufacturing flow. in the experimental simulation, the proposed models are applied to compute the spv for each supplier and verify that the shapley value does indeed offer significant assistance for the purpose of decision-making and efficient allocation of suppliers in the supply chain. 2. literature review within the past few decades, researchers have developed a variety of optimization methods in order to optimize multi-supplier selection and order quantity allocation, to improve supply chain performance. hosseini and barker [8] considered three resilience-based supplier selection criteria: absorptive (e.g., surplus inventory), adaptive (e.g., rerouting), and restorative (e.g., technical resource restoration) capacities in their development of a bayesian network formulation for supplier evaluation and selection. their model depended upon expert judgment. in addition, bayesian networks are usually very complicated because of the large number of variables used to capture causality. dogan and aydin [9] proposed an integrated model which provides some advantages when analysing supplier selection problems. their goal was to identify uncertainties more clearly and utilizes the buyer’s specific domain knowledge. they provided a supplier figure 1 | the balance between lower budgets and higher manufacturer performance in scm. assessment and selection process for first-tier suppliers in the automotive industry, but their model also relies upon expert opinions and judgments. erdem and göçen [10] developed an analytic hierarchy approach, which captured both qualitative and quantitative criteria for supplier evaluation, including cost, quality, logistics, and technology. based on these factors of assessments, a goal programming model was proposed to distribute orders among suppliers. their model has been applied to the decision support system, which presents dynamic, flexible, and fast decision-making. mohammed et al. [11] and nazari-shirkouhi et al. [12] proposed a fuzzy analytic hierarchy process. the multi-objective programming model adopt the final pareto solution to choose suppliers in a supply chain under various uncertainties (such as cost, demand). scott et al. [13] proposed a three-stage decision support system. their first stage process is run for supplier evaluation and selection according to the stakeholders’ requirements, evaluation criteria, and supplier’s characteristics (i.e., capacity, price, quality criteria), after which the supplier performance score is derived for each supplier. in the second stage, the optimization algorithm is used for order allocation based on the supplier performance score. in the third stage, their decision support system is validated using a monte-carlo simulation. mendoza and ventura [14] proposed the nonlinear programming model that selects the best supplier set and determines the appropriate order quantity allocation, given the supplier’s ability and quality constraints so as to minimize annual ordering, inventory holdings, and procurement costs. demirtas and üstün [15] also proposed a two-stage model that utilizes tangible and intangible elements when selecting the best supplier and defines the best quantity among the selected suppliers to maximize the total value of the purchase and minimize defect rate. four different plastic molding companies working with refrigerator factories were analysed according to 14 criteria grouped into four clusters: benefits, opportunities, costs, and risks. huang et al. [16] proposed a solution concept derived from cooperative game theory (i.e., shapley value) to obtain the marginal contributions of suppliers so as to select the best suppliers. wang and li [17] expanded supplier evaluation in the case of variable returns to scale for the development of a new nash bargaining game, including a data envelopment analysis (dea) model for supplier evaluation. their model uses universal weights for evaluation based on the nash bargaining game dea model can fairly evaluate and rank all suppliers. all suppliers are motivated to accept pareto solutions for nash bargaining games. the above supplier evaluation and order allocation methods are designed for improving the supply chain performance with more complicated calculations. this study extends the supplier evaluation model of wu et al. [18] to more explicit in which the interactions between manufacturer and supplier are combined with manufacturing requirements. the interactions of six critical factors (i.e., quality of products, cost of manufacturing, the technology of the supplier, delivery performance, service demand, and priority) between the supplier and the manufacture are modelled and analysed as a non-cooperative game to create a spv with which to evaluate a given supplier. and then the shapley value is applied in a simple cooperative game to easily and quickly calculate a fair manufacturing order allocation based on the spv. c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 123 when the supply chains face unpredictable damage caused by sudden natural disasters (such as the outbreak of covid-19, hurricanes or earthquakes), or man-made threats (such as terrorist attacks or strikes), they need to meet customer requirements and recover from the interruption [19,20]. hosseini et al. [21] proposed a supply chain resilience method that integrated markov chain and dynamic bayesian network methods to find the potential high-risk paths in the supply chain and to prioritize the emergency response and recovery strategies. one contribution of the proposed approach as compared to these above researches is that we consider the interactions between the manufacturer and the supplier in a supplier assessment game, which fairly and reasonably distributes manufacturing budgets. 3. the proposed framework the two models proposed here consider a 12-supplier, one manufacturing process of the manufacturer; see figure 2. two games are constructed for the economical allocation of suppliers. one is a supplier assessment game which calculates 12 spvs during one manufacturing process. the other is a manufacturing order allocation game which efficiently distributes the limited manufacturing orders to all suppliers within one manufacturing process. a simplified workflow chart outlining the principles of the supplier deployment model for this manufacturing process appears in figure 2. in the first step, the interactions between the manufacturer and the supplier are modelled as a two-player, zero-sum, and non-cooperative game. after considering the information needed to evaluate each supplier, such as the quality of the products, cost of manufacturing, the technology of the supplier, delivery performance, and the manufacturer’s requirements (i.e., services demanded and priority), the first model calculates two player’s payoff for their strategy combination in the matrix. given the payoff matrix, a unique spv is calculated for each supplier from the expected payoff using the pure (or mixed strategy) nash equilibrium. in the second step, a cooperative game model (i.e., manufacturing order allocation game) is applied to generate a power index for the deployment of suppliers based on the shapley value. from the total spvs of all suppliers, the majority of spv values is computed to generate the majority threshold in one manufacturing process. then the 12 shapley values (s1, s2, …, s12) are calculated based on the threshold. finally, the proposed model generates the appropriate manufacturing order allocation vector (e1, e2, …, e12) for the allocation of the suppliers to a manufacturer in the supply chain. 3.1. twelve supplier assessment games from the 12 supplier assessment games in the first stage, we obtain spvs for the suppliers, i = 1, 2, …, 12. in each simultaneous game, two players implement a single static step. manufacturer and supplier behaviours are captured with a two-person and zero-sum game. it is assumed that the manufacturer and the supplier are rational players. they form a set of noncooperative players i = {i1, i2}, where i1 is the manufacturer evaluating supplier capability; and i2 is the supplier. dickson [3] conducted a questionnaire survey of about 300 commercial organizations (mainly manufacturing companies), asking the purchasing managers of these companies the question: what are the important factors in choosing a supplier. he found that quality, price, delivery, technical capabilities, and service are the most critical factors in the supplier selection. cheraghi et al. [22] also reviewed more than 110 research papers to find these five critical factors for successful supplier selection. the parameters for determining the measures for evaluating a good supplier are defined below. assume that the manufacturer is player 1 in a supplier assessment game. each manufacturer has a decision maker, who in turn has specific suppliers with which to prepare for product manufacturing figure 2 | workflow for supplier capability assessment and manufacturing order allocation. 124 c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 processes. here, u denotes the set of player 1’s strategies: u = {u1, u2} = {services demand, priority}. the greater the productivity of the manufacturing process, the larger the supplier capacities needed. w denotes the set of manufacturer service requirements for each manufacturing process in a supply chain. w = {w1, w2, w3, w4}. the variable wk denotes the number of product manufacturing services required u1 with the suppliers’ capacity dk, k = 1 – 4, where wk = 1 is the lowest level and wk = 4 is the highest, as shown in table 1. each supplier is assigned a priority p on a scale of 1–4 indicating the significance of the supplier’s capacity dk. p denotes the set of manufacturer priority requirements for each manufacturing process in a supply chain, p = {p1, p2, p3, p4}. the variable pk denotes the manufacturing priority required u2 with the suppliers’ capacity dk, k = 1–4, where pk = 1 is the lowest level and pk = 4 is the highest, as shown in table 1. player 1 has two strategies and player 2 has four strategies in this game, therefore there exists 36 (4! × 4!) potential combinations. this study presents six cases in the simulation; see table 1. the supplier is player 2 and d denotes the set of player 2’s strategies: d = {d1, d2, d3, d4} = {quality of products, cost of manufacturing, the technology of the supplier, and delivery performance}. c denotes the set of capacities available from one supplier: c = {c1, c2, c3, c4}. ck denotes the number of capacities available at the supplier’s capacity dk, k = 1–4, where ck = 1 is the lowest level and ck = 10 is the highest, as shown in table 2. the yield rate of the delivered product refers to the quality of the products. in this study, the supplier’s delivery product yield rate is used as the criterion for the scoring. the highest score is 10 and the lowest is 1. the higher the yield rate of the delivered product, the higher the score supplier obtains. assuming that the highest yield rate of the delivered product is 100%, the score is reduced by 2% of the yield rate of the delivered product, and the lowest delivery product yield is 82%. for example, the supplier delivers 100 products, of which three are defective, so the yield rate of the delivered product is 97%, and for comparison see table 2 where between 96% and 98%, the score is 8. the quality of products is defined by c1, as the first capacity of the ith supplier which is product yield rate, given by c1 = number of good units number of all units % (1) the cost of manufacturing is defined in this study using the ratio of the manufacturer’s accepted price to the supplier’s offer price as the criterion for the scoring. the manufacturer’s accepted price is the price at which the manufacturer agrees to buy the supplier product. the supplier’s offer price is the price that the supplier provides a selling product. we assume the supplier’s offer price is bigger than or equal to the manufacturer’s accepted price. the cost of manufacturing which defines c2 as the second capacity of the ith supplier is given by c2 10= ´ manufacturer’s accepted price supplier’s offer price (2) in table 2, the highest score of the c2 is 10 and the lowest is 1. the delivery performance is defined using the proportion of on-time delivery of products as the criterion for the scoring. the highest score is 10 and the lowest is 1. the higher the proportion of historical on-time deliveries by the supplier, the higher the score. when the on-time delivery rate is 100%, the score level is reduced by 2% proportionally; see table 2. the delivery performance which defines c3 as the third capacity of ith supplier as given by c3 = number of on time delivery number of delivery %. (3) the technology of the supplier defines c4 as the fourth capacity of ith supplier, given by c c pk4 3 3 = = − −    min usl lsla d a d , , (4) where cpk is the process capability ratio for an off-center process or process capability index. this study applies the proposed method of montgomery [23] to compute this process capability index. in table 2, usl represents the upper specification limit of the manufacturing process. lsl presents the lower specification limit of the manufacturing process. a is the mean of the manufacturing process and δ represents the deviation of the manufacturing process. this study uses the process capability c = {c1, c2, c3, c4} of the supplier as the criterion for the scoring. the reference c rating table is shown in table 2. the highest score is 10 and the lowest is 1. the higher the c value of the supplier, the higher the score. we assume that 12 suppliers from a to l are evaluated by one manufacturer. the number of variables c1, c2, c3, and c4 in table 2 are transferred to levels 1–10 so as to produce table 3. table 1 | manufacturer x’s requests for service demand and priority: six cases no manufacturer (player 1) supplier (player 2) quality of products cost of manu­ facturing delivery perfor mance the technology of the supplier case 1 service demand 1 3 2 4 priority 4 1 3 2 case 2 service demand 4 1 2 3 priority 1 2 3 4 case 3 service demand 3 2 4 1 priority 1 4 2 3 case 4 service demand 1 4 3 2 priority 4 2 1 3 case 5 service demand 3 4 2 1 priority 1 2 4 3 case 6 service demand 1 3 4 2 priority 4 1 2 3 table 2 | capability level of supplier level quality of products (c1) (%) cost of manufacturing (c2) delivery performance (c3) (%) the technology of the supplier (c4) [23] 10 100 (manufacturer’s accepted price)/ (supplier’s offer price) * 10 100 2 ≦ c4 9 98 98 1.83 ≦ c4 < 2 8 96 96 1.67 ≦ c4 < 1.83 7 94 94 1.5 ≦ c4 < 1.67 6 92 92 1.33 ≦ c4 < 1.5 5 90 90 1.17 ≦ c4 < 1.33 4 88 88 1 ≦ c4 < 1.17 3 86 86 0.83 ≦ c4 < 1 2 84 84 0.67 ≦ c4 < 0.83 1 82 82 c4 < 0.67 c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 125 this study assumes that the manufacturer receives a payoff from the supplier during their interactions, and the higher the supplier’s ability to supply resources to the manufacturer during the manufacturing process, the higher the manufacturer’s payoff. in other words, the greater the supplier’s loss, the greater the gain to the manufacturer because the more the resources the supplier has to expend to complete the order, the higher the manufacturer’s gains. in this game, both players will simultaneously make strategic decisions. thus, a 2 × 4 payoff matrix is created for the supplier assessment game based on the two players’ strategies and interactions as shown in table 4. the payoff to player 1 (the manufacturer) for choosing a strategy when player 2 (the supplier) makes his or her selection can be represented as a gain for the manufacturer or a loss to the supplier. in this model, a summation of the gains of the manufacturer is depicted, and the manufacturer tries to maximize the supplier’s losses (efforts) while the supplier tries to minimize their losses. the manufacturer obtains a positive payoff (+), which means that he or she gains a profit from the supplier’s capacity responses. the supplier obtains a negative payoff (−), which means that he or she pays based on the manufacturer’s manufacturing requirements. in table 4, the payoff for the two strategies for player 1 (the manufacturer) when player 2 (the supplier) chooses four strategies in response can be formulated as p1 1 1 2 2 3 3 4 4 1 1 2 2 3 3 4 4 1 1 1 = + + + + + + + = + + ( ) ( ) ( ) ( w c w c w c w c p c p c p c p c w p c ww p c w p c w p c2 2 2 3 3 3 4 4 4+ + + + +) ( ) ( ) . (5) the supplier assessment game is a zero-sum game; thus, the payoff function of player 2 (the supplier) is given by p2 1 1 1 2 2 2 3 3 3 4 4 4= + + + + + + +(( ) ( ) ( ) ( ) ).w p c w p c w p c w p c (6) this study assumes that player 1 (manufacturer) and player 2 (supplier) are selfish and will try to maximize their own utility when player 1 randomly commits to a service demand or priority. player 2 also randomizes the availability of his or her capabilities (i.e., quality of products, cost of manufacturing, delivery performance, and technology of the supplier). thus, a mixed strategy nash equilibrium pair (p*, q*) exists in the normal form of the game if the game has no pure strategy n.e., which is an optimal strategy [24–26]. the prevent-exploitation method [27] can be applied to calculate the mixed strategy n.e. for the 2 × 4 payoff matrix. player 1’s expected payoff is computed when player 1 (manufacturer) and player 2 (supplier) play mixed strategies p and q, respectively. the mixed strategy n.e. for the probability vector is p* = {p*(u1), p *(u2)} with actions {u1, u2} for the manufacturer and the probability vector is q* = {q*(d1), q *(d2), q *(d3), q *(d4)}, with actions {d1, d2, d3, d4} for the supplier. player 1’s (manufacturer’s) expected payoff for a pure or mixed strategy n.e. is defined as the spv of the ith supplier. here, vi is defined as the ith spv, given by v p q p u q d u d u di j k j k j k j k= = î = = ååp p1 1 2 1 4 1( ) ( ) ( ) ( ), , , , . * * * * * * * * n e (7) the higher the supplier’s ability to supply resources to the manufacturer during the manufacturing process, the higher the manufacturer’s payoff. therefore, vi is obtained from the expected payoff of both players’ optimal strategies which represents the spv of the ith supplier in the first game model. the next proposed model uses the value of vi to compute the shapley value for each supplier within the cooperative game. 3.2. manufacturing order allocation game in this second game, the interactions of all suppliers in the manufacturing process are likened to the playing of a cooperative game. we assume that the manufacturer is going to evaluate 12 suppliers (i.e., for manufacturing process x); see figure 2 and table 3. a useful method is needed to decide the number and priority for the allocation of suppliers to one manufacturing process, especially when available budgets for the manufacturing process is limited. this study applies a resource reallocation game of wu and hu [28] to deal with the allocation or selection of suppliers. shapley value is a solution concept of this game which is conceived of as a power index for resource allocation in a cooperative game [29]. in this study, shapley value is applied to create a fair manufacturing order allocation of suppliers for the given product manufacturing process. the second model applies the concept of the majority coalition in a table 3 | availability of the supplier’s capabilities supplier (player 2) quality of products (c1) cost of manufacturing (c2) delivery performance (c3) technique of supplier (c4) manufacturer (player 1) a 10 8 7 6 x b 8 9 6 8 c 6 6 7 9 d 7 7 6 7 e 8 6 7 7 f 6 8 10 8 g 7 6 8 9 h 5 8 9 7 i 8 6 9 10 j 6 7 10 6 k 9 7 7 7 l 7 8 7 8 table 4 | payoff matrix for supplier assessment game manufacturer (player 1) supplier (player 2) quality of products (d1) cost of manu­ facturing (d2) delivery performance (d3) the technology of the supplier (d4) service demand (u1) w1c1 w2c2 w3c3 w4c4 priority (u2) p1c1 p2c2 p3c3 p4c4 126 c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 voting game to compute the power index of each supplier so as to allocate fair supplier deployment in a single manufacturing process. a majority of voters can pass any bill in a voting game. the power of a voter depends on how critical that voter is to form a winning coalition. similarly, when the sum of the spv of some suppliers passes the threshold of the majority level, the formation of a winning coalition is enabled, and the power index of each supplier can be computed. wu et al. [30,31] proposed a resource deployment game for emergency responses. according to their model, this study also defines y: v → r+ as a one-to-one function by assigning a positive real number to each item of v (i.e., spv) and y(0) = 0, v = {v1, v2, …, v12}. the majority of level h is derived from a majority of all spvs for a manufacturing process which represents the corresponding threshold value mh. supplier allocation for the manufacturing process is based on the concept of the majority of spvs. given the output vector of all spvs for the manufacturing process, the majority level is h, if the sum of the spvs is greater than or equal to mh: h v m m v i n i h h i iif = =å å ³ = 1 1 12 2 , . (8) where the threshold m v h i i= = ∑ 1 12 2 . all supplier spvs can be grouped according to the majority level h. the threshold value mh is half of the summation of all spvs. all suppliers can be modelled as a 12-person cooperative game, where v = {v1, v2, ..., v12}, which includes a set of players (i.e., suppliers) and each subset c ⊂ v, where vi ≠ 0, ∀vi ∈ c is called a coalition. the coalition of c supplier groups in the mh threshold of the majority level, and each subset of c (coalition), represents the observed capability pattern for the majority level h. the aggregated value of the coalition is defined as the sum of the spvs for a manufacturing process, y(c), and is called a coalition function. now let y c vi( ) = å , vi ∈ v, c ⊂ x be the value of coalition c with a cardinality of c. the shapley value of the ith supplier is defined by s c n c n y c y c ii c v i c n = -ì î å ( )!( )! ! [ ( ) ( { })]; 1 (9) þ = î ¼¢ î ¢ ìås c n c n ii c v i c n ( )!( )! ! , , . 1 1 12 (10) this manufacturing order allocation game is represented by a characteristic (or coalition) function y that chooses a value 0 or 1. coalition cʹ is called the winning coalition if y(cʹ) = 1 and losing if y(cʹ) = 0. this is a simple and majority game with 12 voters (i.e., spvs) [32], meaning 12 suppliers, and can be represented by the vector [mh; v1, …, v12], where v1 denotes the number of votes cast by the first supplier, and mh denotes the number of votes needed for a winning coalition. the winning coalitions are then precisely those coalitions c´ with enough votes; that is, cʹ wins if and only if i c i hv m î ¢ å ³ , i ∈ cʹ. equation (9) can be simplified to eq. (10), because the term y(c) − y(c − {i}) will always have a value of 0 or 1, taking a value of 1 whenever cʹ is a winning coalition. if cʹ is not a winning coalition, the term y(c) − y(c − {i}) has a value of 0 [33]. the shapley value of the ith supplier output indicates the relative spv value for the threshold mh of the majority level. the shapley value represents the strength of the supplier’s capabilities, which the manufacturer should consider when assigning suppliers to meet manufacturing requirements for one process. the shapley value in the ith supplier is applied to compute the number of the supplier’s manufacturing orders for the mh threshold of the majority level. this study assumes that 12 suppliers are selected by one manufacturer. given mh thresholds of the majority level in the manufacturing process of manufacturer x, the amount of manufacturing budget allocated to be assigned to the ith supplier is defined by e s o ii i= ´ î ¼all 1 12, , (11) here, oall is the total amount of manufacturing budget available for the manufacturing process. the amount allocated to the ith supplier for process ei is derived from the shapley value for the ith supplier si multiplied by oall the total amount of manufacturing budget available. the proposed model not only obtains the appropriate shapley value vector of twelve suppliers (s1, s2, …, s12) but also calculates their manufacturing order allocation vector (e1, e2, …, e12). 4. experimental results a numerical example to illustrate the application of the proposed framework is presented in this section. it is found that 12 suppliers are critical for the manufacturing process of one manufacturer after evaluating them for the supply chain management. the information in tables 1–3 regarding manufacturer priority, evaluation types, and requests, and capacities available from each of the suppliers and the manufacturer is used in the simulation. simulated sets of supply and demand measures for 12 suppliers and one manufacturer are randomly generated, as shown in tables 1–3. first, 12 hypothetical parameter amounts are given to model the payoff matrix. we apply the gambit method [34] to calculate 12 supplier’s spvs from the numerical examples and create one nash equilibrium in each strategic game; see figure 3. then, the spv of each supplier is utilized to compute each supplier’s shapley value based on one manufacturing process. in this paper, the majority levels are designed for the x manufacturer. matlab is adopted to calculate the supplier’s shapley values for the manufacturing process of x manufacturer; see figure 4. 4.1. numerical analysis of supplier assessment a numerical simulation is conducted to determine whether the spvs calculated in the first stage of the supplier assessment game are optimal for both the supplier and the manufacturer. we assume that the proposed game is a zero-sum game which exists a mixed-strategy nash equilibrium. when the supplier and manufacturer choose their mixed nash equilibrium strategies, the figure 3 | twelve spvs for supplier assessment. c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 127 manufacturer’s expected payoff is vi and the supplier’s expected payoff is the negative of vi; see eq. (7). equilibrium is reached when the manufacturer, for the given q-mixed strategies, chooses its p-mixed strategies to maximize eq. (7) (i.e., the manufacturer’s expected payoff ). simultaneously the supplier, for the given p-mixed strategies, chooses its q-mixed strategies to maximize the negative of eq. (7) (i.e., the supplier’s expected payoff ) or to minimize the same equation. according to eq. (7), the expected payoff of the manufacturer’s optimal strategies is represented by the spv. this study presents six cases for validation; see table 1 and figure 5a–5f. each figure 4 | twelve shapley values for the manufacturing process of x manufacturer. figure 5 | the spv is represented as a saddle point when the supplier chooses the: (a) cost of manufacturing (d2) and delivery performance (d3) mixed strategy; (b) quality of products (d1) and cost of manufacturing (d2) mixed strategy; (c) quality of products (d1) and the technology of the supplier (d4) mixed strategy; (d) delivery performance (d3) and technology of the supplier (d4) mixed strategy; (e) quality of products (d1) and delivery performance (d3) mixed strategy; (f ) cost of manufacturing (d2) and technology of the supplier (d4) mixed strategy. a b c d e f 128 c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 table 5 | twelve examples for computing the spvs, shapley values, and allocation amount for suppliers manufacturing process supplier spvs threshold value mh shapley value si allocation amount for supplier ei manufacturer x a 17.043 mh = 83.85 0.114 1140.0 b 15.750 0.095 945.2 c 13.200 0.079 786.4 d 14.700 0.089 894.7 e 11.200 0.057 570.0 f 12.923 0.074 743.1 g 10.889 0.052 519.5 h 12.174 0.068 678.2 i 17.143 0.115 1154.4 j 12.000 0.063 634.9 k 15.750 0.101 1010.1 l 14.933 0.092 923.5 total 167.7 1 10,000 case plays a 2 × 4 payoff matrix game. the “preventexploitation method” [27] is applied to find the mixed strategy nash equilibrium in the 2 × 4 payoff matrix game. in the first case, two mixed strategies are chosen for the manufacturer: services demand (u1) and priority (u1). two mixed strategies are chosen for the supplier: cost of manufacturing (d2) and delivery performance (d3), and two strategies are deleted: quality of products (d1) and technology of supplier (d4). thus, this game can be simplified to 2 × 2 payoff matrix. consider eq. (7) as a function of all the p-mixed strategies and the q-mixed strategies and draw it in three-dimensional space. the saddle-shaped front and rear cross-sections are valley or u-shaped with the minimum value in the middle, while the lateral cross-section looks like a peak or inverted u-shape with the maximum value in the middle. if the supplier or manufacturer has just two pure strategies, p(u1) and q(d2) can be determined by an integer (1 or 0). if the probability of the manufacturer choosing the first strategy is p(u1), then the probability of choosing the second strategy is p(u2) = 1 − p(u1); if the probability of the supplier choosing the first strategy is q(d2), then the probability of choosing the second strategy is q(d3) = 1 − q(d2). the graph for this supplier assessment game is drawn in three dimensions; see figure 5a, where the xand y-axis are on the horizontal plane, and the z-axis points vertically upward. the p(u1) of the manufacturer is shown along the x-axis, the q(d2) of the supplier along the y-axis, and the value v of the expected payoff along the z-axis. a cross-section of the graph along the x direction will show the maximum value of v with respect to p(u1) as a peak. a cross-section along the y direction will show that v is minimized with respect to q(d2), and therefore appears as a valley. thus, the graph is saddle-shaped, as illustrated in figure 5a. the mixed strategies n.e. v1 (i.e., spv) is called a saddle point from which the manufacturer and supplier cannot deviate. similarly, as shown in figure 5b, in the second case, we also delete two of the manufacturer’s strategies: delivery performance (d3) and technology of the supplier (d4) leaving the two mixed strategies of quality of products (d1) and cost of manufacturing (d2). the numerical simulation shows that the saddle point of this game is the spv (v2), as shown in figure 5b. the third case also can be simplified to a 2 × 2 payoff matrix, with the two mixed strategies of quality of products (d1) and technology of the supplier (d4). we find that the saddle point of this game is the spv (v3); see figure 5c. the fourth case presents two mixed strategies of delivery performance (d3) and technology of the supplier (d4). the saddle point of this game is the spv (v4); see figure 5d. in the fifth and the sixth case, we also find that the saddle point of the game to be the spv (v5) and the spv (v6); see figure 5e and 5f. the value of vi (i.e., spv) is the mixed strategy n.e., that is, the simultaneous maximum with respect to the expected payoff to the manufacturer and the minimum with respect to the expected payoff for the supplier. this is called the minimax value of the zero-sum game. thus, this value (i.e., spv) of the supplier assessment game is optimal for the manufacturer and the supplier. the higher spv supplier obtains, the higher the expected payoff the manufacturer gains and the more orders manufacturer is willing to give to the supplier in the manufacturing process. the spv appears to offer useful information concerning the decision making of supplier assessment. 4.2. validation of manufacturing order allocation in this experimental example, the administrator of the scm is responsible for the deployment of a supplier order budget of 10,000 and provides central management and monitoring with consideration of the majority levels (for the manufacturing process of x manufacturer) for 12 suppliers. his/her goal is to find the most effective allocation of suppliers for the manufacturing process given one majority level. the 12 spvs are calculated exactly using eqs. (1)–(7). figure 3 shows the sequence of the 12 suppliers’ spvs as an output vector. we use eq. (8) to obtain the threshold of the majority level mh = 83.85 according to the spv vector output. using these threshold values, we apply eq. (10) to calculate the exact shapley values for the suppliers as shown in figure 4 and table 5. then, a manufacturing order allocation plan comprised of a minimum set of suppliers for the manufacturing process using eq. (11); see table 5. as can be seen in table 5, from supplier a to l, the higher the spv, the higher the shapley value (such as for suppliers a and i). the shapley value shows the relative importance of the suppliers. suppliers with a higher shapley value receive more order quantities than suppliers with a lower shapley value. in contrast, the lower the spv value, the lower the shapley value (as for suppliers e and g), so the fewer order quantities received. 5. conclusion and future work in the proposed framework, two game theory models are applied to develop the supply chain decision and analysis process. the first model creates a nash equilibrium game strategy to find the spv for each supplier, for optimal supplier assessment. the second model uses the spvs to compute a shapley value for each supplier given the majority threshold. the shapley values are then used to build a rating system for the deployment of suppliers within one manufacturing flow. in this way, we can provide a fair and equitable allocation of a manufacturing order. the simulation results demonstrate the c.-k. wu and y.-m. chuang / journal of risk analysis and crisis response 10(4) 121–129 129 suitability of the two game theory models for the determination of the deployment of suppliers. the experiments help us to demonstrate that the framework connecting the nash equilibrium and shapley values, will enable the manufacturer to prioritize their budget for a manufacturing process. the experimental results confirm that the framework can assist managers in assessing manufacturer demands and supplier’s capacities for manufacturing order allocation. in future work, we will use real data to verify the suitability of the proposed framework for the manufacturing flow management process. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution both authors designed the model and the computational framework and analyzed the data. all authors discussed the experimental results and contributed to the final manuscript. acknowledgments the authors appreciate the time and effort of the editors and reviewers in providing constructive comments which have helped to improve the manuscript. references [1] choi ty, hartley jl. an exploration of supplier selection practices across the supply chain. j oper manage 1996;14:333–43. [2] poirier cc, reiter se. supply chain optimization: building the strongest total business network. 1st ed. san francisco, us: berrett-koehler publishers; 1996. [3] dickson gw. an analysis of vendor selection systems and decisions. j purchasing 1966;2:5–17. [4] weber ca, current jr, benton wc. vendor selection criteria and methods. eur j oper res 1991;50:2–18. [5] chan tsf, kumar n. global supplier development considering risk factors using fuzzy extended ahp-based approach. omega 2007;35:417–31. [6] maurizio b, alberto p. from traditional purchasing to supplier management: a fuzzy logic-based approach to supplier selection. int j logist 2002;5:235–55. [7] jayaraman v, srivastava r, benton wc. supplier selection and order quantity allocation: a comprehensive model. j supply chain manage 1999;35:50–8. [8] hosseini s, barker k. a bayesian network model for resilience based supplier selection. int j prod econ 2016;180:68–87. [9] dogan i, aydin n. combining bayesian networks and total cost of ownership method for supplier selection analysis. comput ind eng 2011;61:1072–85. [10] erdem as, göçen e. development of a decision support system for supplier evaluation and order allocation. expert syst appl 2012;39:4927–37. [11] mohammed a, setchi r, filip m, harris i, li x. an integrated methodology for a sustainable two-stage supplier selection and order allocation problem. j clean prod 2018;192:99–114. [12] nazari-shirkouhi s, shakouri h, javadi b, keramati a. supplier selection and order allocation problem using a two-phase fuzzy multi objective linear programming. appl math modell 2013;37:9308–23. [13] scott j, ho w, dey pk, talluri s. a decision support system for supplier selection and order allocation in stochastic, multi-stakeholder and multi-criteria environments. int j prod econ 2015;166:226–37. [14] mendoza a, ventura ja. analytical models for supplier selection and order quantity allocation. appl math modell 2012;36:3826–35. [15] demirtas ea, üstün ö. an integrated multiobjective decision making process for supplier selection and order allocation. omega 2008;36:76–90. [16] huang y, huang gq, liu x. cooperative game-theoretic approach for supplier selection, pricing and inventory decisions in a multi-level supply chain. in: proceedings of the international multiconference of engineers and computer scientists. hong kong: imecs; 2012. [17] wang m, li y. supplier evaluation based on nash bargaining game model. expert syst appl 2014;41:4181–5. [18] wu ck, wu d, chen ym, chuang ym. a game theory approach for evaluating and assigning suppliers in supply chain management. in: 30th international conference on game theory at stony brook. new york: stony brook center for game theory; 2019. [19] hosseini s, al khaled a. a hybrid ensemble and ahp approach for resilient supplier selection. j intell manuf 2019;30:207–28. [20] hosseini s, ivanov d, dolgui a. review of quantitative methods for supply chain resilience analysis. transport res e 2019;125:285–307. [21] hosseini s, ivanov d, dolgui a. ripple effect modelling of supplier disruption: integrated markov chain and dynamic bayesian network approach. int j prod res 2020;58:3284–303. [22] cheraghi sh, dadashzadeh m, subramanian m. critical success factors for supplier selection: an update. j appl bus res 2004;20: 91–108. [23] montgomery dc. introduction to statistical quality control. 5th ed. new york: john wiley & sons; 2005. [24] gibbons r. a primer in game theory. hemel hempstead, uk: harvester wheatsheaf; 1992. [25] hansen t. on the approximation of nash equilibrium points in an n-person noncooperative game. siam j appl math 1974;26:622–37. [26] lemke ce, howson jr jt. equilibrium points of bimatrix games. j soc ind appl math 1964;12:413–23. [27] dixit ak, skeath s. games of strategy. new york: w. w. norton & company; 2007, p. 574. [28] wu ck, hu x. a game theory approach for multi-agent system resources allocation against outside threats. j risk anal crisis resp 2019;9:134–44. [29] roth ae, verrecchia re. the shapley value as applied to cost allocation: a reinterpretation. j account res 1979;17:295–303. [30] chen ym, wu d, wu ck. a game theory approach for evaluating terrorist threats and deploying response agents in urban environments. j homeland secur emerg manage 2009;6. [31] wu ck. a game theory approach for assessing risk value and deploying search-and-rescue resources after devastating tsunamis. environ res 2018;162:18–26. [32] wu ck, chen ym, wu d, chi cl. a game theory approach for assessment of risk and deployment of police patrols in response to criminal activity in san francisco. risk anal 2020;40:534–49. [33] owen g. game theory. 3rd ed. new york, ny: academic press; 1996. [34] mckelvey rd, mclennan am, turocy tl. gambit: software tools for game theory, version 16.0.1. 2016. available from: http://www.gambit-project.org. https://doi.org/10.1016/s0272-6963%2896%2900091-5 https://doi.org/10.1016/s0272-6963%2896%2900091-5 https://doi.org/10.1111/j.1745-493x.1966.tb00818.x https://doi.org/10.1111/j.1745-493x.1966.tb00818.x https://doi.org/10.1016/0377-2217(91)90033-r https://doi.org/10.1016/0377-2217(91)90033-r https://doi.org/10.1016/j.omega.2005.08.004 https://doi.org/10.1016/j.omega.2005.08.004 https://doi.org/10.1016/j.omega.2005.08.004 https://doi.org/10.1080/1367556021000026691 https://doi.org/10.1080/1367556021000026691 https://doi.org/10.1080/1367556021000026691 https://doi.org/10.1111/j.1745-493x.1999.tb00237.x https://doi.org/10.1111/j.1745-493x.1999.tb00237.x https://doi.org/10.1111/j.1745-493x.1999.tb00237.x https://doi.org/10.1016/j.ijpe.2016.07.007 https://doi.org/10.1016/j.ijpe.2016.07.007 https://doi.org/10.1016/j.cie.2011.06.021 https://doi.org/10.1016/j.cie.2011.06.021 https://doi.org/10.1016/j.cie.2011.06.021 https://doi.org/10.1016/j.eswa.2011.10.024 https://doi.org/10.1016/j.eswa.2011.10.024 https://doi.org/10.1016/j.eswa.2011.10.024 https://doi.org/10.1016/j.jclepro.2018.04.131 https://doi.org/10.1016/j.jclepro.2018.04.131 https://doi.org/10.1016/j.jclepro.2018.04.131 https://doi.org/10.1016/j.apm.2013.04.045 https://doi.org/10.1016/j.apm.2013.04.045 https://doi.org/10.1016/j.apm.2013.04.045 https://doi.org/10.1016/j.ijpe.2014.11.008 https://doi.org/10.1016/j.ijpe.2014.11.008 https://doi.org/10.1016/j.ijpe.2014.11.008 https://doi.org/10.1016/j.apm.2011.11.025 https://doi.org/10.1016/j.apm.2011.11.025 https://doi.org/10.1016/j.omega.2005.11.003 https://doi.org/10.1016/j.omega.2005.11.003 https://doi.org/10.1016/j.omega.2005.11.003 https://doi.org/10.1016/j.eswa.2013.12.044 https://doi.org/10.1016/j.eswa.2013.12.044 https://doi.org/10.1007/s10845-016-1241-y https://doi.org/10.1007/s10845-016-1241-y https://doi.org/10.1016/j.tre.2019.03.001 https://doi.org/10.1016/j.tre.2019.03.001 https://doi.org/10.1016/j.tre.2019.03.001 https://doi.org/10.1080/00207543.2019.1661538 https://doi.org/10.1080/00207543.2019.1661538 https://doi.org/10.1080/00207543.2019.1661538 https://doi.org/10.19030/jabr.v20i2.2209 https://doi.org/10.19030/jabr.v20i2.2209 https://doi.org/10.19030/jabr.v20i2.2209 https://doi.org/10.1137/0126058 https://doi.org/10.1137/0126058 https://doi.org/10.1137/0126058 https://doi.org/10.1137/0112033 https://doi.org/10.1137/0112033 https://doi.org/10.2991/jracr.k.191024.003 https://doi.org/10.2991/jracr.k.191024.003 https://doi.org/10.2991/jracr.k.191024.003 https://doi.org/10.2307/2490320 https://doi.org/10.2307/2490320 https://doi.org/10.2202/1547-7355.1488 https://doi.org/10.2202/1547-7355.1488 https://doi.org/10.2202/1547-7355.1488 https://doi.org/10.1016/j.envres.2017.12.008 https://doi.org/10.1016/j.envres.2017.12.008 https://doi.org/10.1016/j.envres.2017.12.008 https://doi.org/10.1111/risa.13411 https://doi.org/10.1111/risa.13411 https://doi.org/10.1111/risa.13411 journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 27–36 doi: https://doi.org/10.2991/jracr.k.200415.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article study on mutation mechanism of victim’s psychological behavior state after major natural disasters xuanhua xu*, yushi liu, zitao zhang school of business, central south university, changsha 410083, china 1. introduction with the increasingly prominent problem of global climate change, the frequency of all kinds of natural disasters is increasing, and the scope of impact is becoming wider and wider. according to the statistics of the civil affairs bureau, all kinds of natural disasters in china have caused 243,533 million disasters, 1583 deaths, 6017 million emergency relocation in 2014, and the direct economic losses have reached 337,38 billion yuan [1]. natural disasters not only cause a large number of casualties and economic losses, but also cause varying degrees of psychological trauma to the affected people, which may affect individual behavior in the short-term or even in the long-term [2]. many cases have shown that anxiety in disaster scenarios can urge the affected children to make some aggressive behaviors [3], these experiences of negative psychological behaviors will increase the difficulty of rescue and resettlement. if it is not alleviated in time and effectively, it will accumulate gradually, and easily turn into illegal behavior, or even evolve into group events [4] and reduce the stability and safety of society [5,6]. behavior analysis of the disaster victims plays an important role in crisis management, disaster emergency response and implementation of evacuation plan [7], and is also one of the core scientific issues in emergency management of public emergencies [8]. therefore, in order to rescue effectively and block the evolution and upgrading of individual events, and scientifically formulate rescue strategies after the disaster, it is necessary to deeply analyze the mutation mechanism of individual psychological behavior state of the affected people after natural disasters. many researchers have studied the psychology and behavior of the people affected by climate change and natural disasters. doherty and clayton [9] analyzed three effects of climate change on people’s psychology: direct psychological impact (asd or ptsd); indirect psychological impact (decreased wellbeing); social psychological impact (social conflict). norris et al. [10] analyzed more than 160 empirical studies on disasters from 1981 to 2001 and found that the occurrence of nature disasters would bring people some shortor long-term psychological problems, such as specific psychological injury, mental and physical disorders. aiming at whether there is a correlation between natural disasters and suicide behavior, kõlves et al. [11] has analyzed the existing empirical studies and found that different types of natural disasters have different effects on suicide rate, and suicide rate is mainly affected by economic conditions. hu et al. [12] believed that people had panic behavior in evacuation and temporary resettlement after the disaster, and panic behavior was infectious, which made more affected groups panic. helbing et al. [13] also simulated the irrational panic escape behavior of groups under the emergency. panic psychology and irrational behavior are easy to lead to stampede events under emergency events, which are the important factors to increase the loss of life and property [14]. in addition, some studies have proved that people have prosocial behaviors and show resilience after disasters. bonanno et al. [15] found that resilience is mainly related to income, social support and other factors of the affected people by questionnaire survey and multivariate analysis. van der linden [16], akerlof et al. [17] and other researchers also used empirical methods to explore the correlation between risk perception and social structure, economy and other control factors after natural disasters. these studies lay a theoretical and empirical foundation for exploring the mutation mechanism of the psychological and behavioral state of the victims. a r t i c l e i n f o article history received 06 january 2020 accepted 15 april 2020 keywords major natural disaster psychology behavior mutation mechanism a b s t r a c t aiming at the problem of victim’s rational and irrational behavior during recovery stage in major natural disaster, this paper analyzed the victim’s psychological and behavioral characteristics and influencing factors during recovery stage based on lots of literatures and an interview with victims. besides, a mechanism chart of the formation of the victim’s psychological behavior and a cusp catastrophe model are constructed. the catastrophe mechanism of the victim’s psychological behavior during recovery stage is simulated under the influence of the physical demand factors and the emotional needs factors. the two different functions of the two demand factors are analyzed. the model is partially verified by the investigation data of the disaster area. finally, the paper put forward some strategies to cope with victim’s psychology and behavior during recovery stage based on the cusp catastrophe model analysis. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: xuxh@csu.edu.cn https://doi.org/10.2991/jracr.k.200415.001 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:xuxh%40csu.edu.cn?subject= 28 x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 however, there are some deficiencies in the current researches on natural disasters and psychological behaviors of disaster victims. first, researchers only analyze whether natural disasters will cause people’s psychological and behavioral abnormalities, and they pay more attention to the results of people’s behavior after natural disasters, and lack researches of the process mechanism of psychological behavior changes of the victims which caused by natural disasters; second, the existing empirical methods assume that the control factors and the psychological behaviors of the victims are linear and continuous, and this assumption has limitations in the research of the real psychology of the victims. vallacher and nowak [18] analyzes that the social psychology is dynamic and complex, which is difficult and hard to describe with the traditional methods and empirical researches and needs to be further explained by the system theory. as an important branch of system theory, catastrophe theory can better analyze the system which is unknown by the internal function mechanism. catastrophe theory is a theory put forward by thom [19], a french mathematician, which studies the phenomenon of system’s jumping and discontinuity change caused by the continuous change of external control parameters in the system. zeeman [21] made further research on catastrophe theory and proposed zeeman mutation mechanism, and analyzed several characteristics of cusp mutation system: (1) the system has polymorphism, and with the change of control system parameters, the system will jump from one stable state to another; (2) the system is inaccessible, and there is no absolute stable state between different states of the system. in reality, it is inaccessible; (3) the system is divergent, two similar paths near the bifurcation set of control system change will make the system reach two completely different states; (4) the system is hysteretic, it will take some time for the system to recover to the original state after a disturbance change. catastrophe theory is a theory based on topology, bifurcation theory and stability theory. it directly deals with discontinuity without any special internal mechanism, which makes it particularly suitable for studying the system with unknown internal function and can effectively explain the “black box” problem. when the system has the above catastrophe characteristics, catastrophe theory can be used to explain [20]. zeeman [21] and stewart and peregoy [22] applied catastrophe theory to the study of psychology and elaborated some specific hypotheses of catastrophe theory to describe people’s attitude change. van der maas et al. [23] also applied catastrophe theory to the mutation mechanism analysis of people’s attitude change, and proved the feasibility of catastrophe theory model to describe attitude change with empirical data. scott [24] gave an effective explanation for the clinical psychological phenomena that cannot be explained by other theories in clinical psychology, such as schizophrenia, emotional disorders, aggression, etc. using catastrophe theory. flay [25] thinks that people’s psychological process and social behavior cannot be simply expressed by gradient function, which satisfy some characteristics of cusp or butterfly mutation model, and summarizes the cusp mutation model is more suitable for describing people’s behavior changes. catastrophe theory can not only study social science problems [26], but also analyze the behavior of employees in enterprise management. xu and bin [27] etc. use catastrophe theory to analyze the anti-production behavior of employees in enterprises, conflict behavior [28,29], team knowledge sharing behavior [30], sudden change of employees’ behavior [31], etc. yiu and cheung [32] analyzed the characteristics of employee conflict behavior in the construction industry and established a cusp catastrophe model. at the same time, the data showed that cusp catastrophe model was better than linear model and regression model to fit employee’s conflict behavior. chen [33] explained that the random cusp catastrophe model can well analyze the rational and irrational behaviors in social behaviors, and illustrated the application of the catastrophe model in social behaviors with examples. to sum up, catastrophe theory can well analyze the changes of human psychological attitude, sudden changes of behavior and other social psychological behavior phenomena, which provides a theoretical basis and reference for the study of the catastrophe mechanism of the psychological behavior of the victims after natural disasters. however, the psychological behavior state of disaster victims is very different from that of ordinary emergencies and behavior change of employees, and has its own unique influence evolution mechanism. at present, there are few literatures to analyze the mutation mechanism of psychological behavior state of disaster victims, which needs further in-depth analysis. therefore, based on the existing literatures and the investigations of the disaster area, this paper analyzes the characteristics and influencing factors of the psychological behavior state of the disaster victims, and based on this, constructs the catastrophe model of the disaster victims’ psychological behavior state, explains the sudden change of the psychological behavior state of the disaster victims by using the catastrophe theory, and deeply analyzes the process change of the psychological behavior of the disaster victims, in order to effectively provide the theoretical basis of implementing the psychological intervention and rescuing in time after the disaster. 2. analysis of the influencing factors and characteristics of the individual psychological behavior state of the victims 2.1. analysis of characteristics of the psychological and behavioral state of the victims in view of whether natural disasters will change the existing social and economic relations and lead to the abnormal behaviors of the affected people, castillo and carter [34] studied the altruistic behavior, trust and reciprocal behavior in 30 communities suffered from hurricane mitch in 1988 in honduras. it was found that the mean value and variance of the behaviors of the affected people changed nonlinearly with the magnitude of the disaster impact, and moderate impact of natural disaster will improve people’s cooperative and reciprocal and prosocial behavior, but some families suffer from economic pressure, such as the loss of shelter, the huge loss of living resources and so on, and in order to maintain their livelihood, they have uncooperative behavior, even conflict behavior. kan [35] believes that people will have different psychological characteristics after disasters, sometimes they will bravely take responsibility, but sometimes they will choose to withdraw and escape. the dynamic disaster management model proposed by john leach also analyzes some special behaviors of victims in the impact stage x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 29 of disaster. about 15% of these behaviors are emotional out of control and overactive behavior, 75% are tense behavior and lack of initiative behavior, 10% are calm behavior and behavior with certain leadership [36]. defu [37] analyzed the characteristics of farmers’ disaster psychology and behavior, and found that people’s psychological behavior has the characteristics of limited rationality and shortsightedness, and often shows some emotional behavior and extreme behavior in a short period of time. huagui [38] believed that the individual behavior reaction of the victims was the external expression of their psychology. according to the rational degree of the victims’ response to the disaster, the behaviors of the victims were divided into panic behavior and calm behavior. it was found that the behaviors of the victims were affected by the distance of psychological state, and the small change of the distance of state would lead to different behaviors, and further divided into six categories, such as figure 1. through the analysis above, we can see that the psychological behavior state of the victims after the natural disaster has the following four characteristics, as shown in table 1. it can be seen that the characteristics of the psychological and behavioral state of the victims in the natural disaster scenario are in line with zeeman’s mutation characteristics. the cusp catastrophe model can be used to describe the mutation mechanism of the psychological and behavioral state of the victims. 2.2. analysis on the influencing factors of individual psychological behavior of the victims people’s psychological behavior is affected by many factors under emergencies. duoyong [39] divides the individual fear behavior into four categories: event itself, society, individual and the factors related to individual and disaster event, and establishes the individual fear model, and analyzes that individual fear mainly comes from psychological expectation. qingguo and xiaoyi [40] believes that human behavior in emergencies is a function of psychology, surrounding atmosphere and information, and analyzed the influence of various factors on human behavior. some scholars also think that risk cognition is the key factor to determine people’s psychological and behavioral states after disasters. the level of risk cognition determines different psychological and behavioral states of the victims [41], and people’s risk cognition is determined by different factors, such as the influence on social trust, individual disaster experience, psychological distance and other variables [42–44]. however, behavioral science believes that people’s various behaviors are caused by certain motives, which arise from various needs of people’s bodies [45]. in order to satisfy various needs, people need to determine their own behavior goals, and then take a series of actions. huijuan et al. [46] conducted a study on people’s psychological problems after the tangshan earthquake through more than 2000 questionnaires, and found that people’s material needs were extremely intense after the disaster, it was because that natural disasters had greater destructive power on people’s living conditions than other emergencies, and survival instinct urged people to pay more attention to material needs. at the same time, it is found that people still have psychological needs to be comforted and concerned in a long time after the disaster. from july to august in 2013, the research group went to lushan county, the severely affected area of ya’an earthquake, for field investigation. the investigation time was three months after the ya’an earthquake. the victims’ mood was basically stable and in the post disaster recovery period. through interviews and individual questionnaires, it is concluded that people’s psychological distress was mainly caused by economic loss. although the government gave some subsidies, it was far from enough that can make up for the losses caused by the earthquake. although people’s emotions gradually return to normal after the disaster, we find that more than 80% of the victims still have anxiety, panic and other emotions through the post-traumatic stress disorder self-assessment table, and they cannot invest in the post disaster reconstruction work well. they are eager for emotional communication between people and have certain prosocial behavior. according to the survey data, figure 1 | victims’ behavioral response types. table 1 | characteristics of victims’ psychology and behavior under natural disasters scenarios behavior characteristics of victims specific description 1. bimodality after the disaster, people’s psychological behavior state is only rational and irrational, which is especially embodied in the rational behavior of prosocial behavior and reciprocal cooperation; the irrational behavior of conflict, distrust and panic, etc. 2. unreachability after the disaster, people’s behavior is either rational or irrational, there is no intermediate state, and the change of people’s behavior is nonlinear. 3. bifurcation when people’s behavior state is in critical equilibrium state, small change of state distance will lead to people’s different behaviors. 4. hysteresis it will take some time for people to change from irrational behavior to rational behavior after disaster. 30 x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 the needs of people in the post disaster recovery period are mainly two aspects: material and emotional needs. through the material needs to satisfy the food, clothing, housing, transportation and life safety after the disaster, through the emotional needs to satisfy their sense of destination, eliminate the loneliness and depression after the disaster. the satisfaction degree of these two needs determines the risk cognition level of the affected people, and then determines the degree of panic and the rational degree of behavior. risk cognition plays an intermediary role between needs and psychological behavior. according to the literature analysis and field survey summary above, the influencing factors of the psychological behavior of the victims under the natural disaster scenario are shown in table 2. 2.3. analysis on the formation mechanism of individual psychological behavior changes of victims in this paper, the individual psychological behavior of the victims is defined the psychological behavior of the victims in the recovery period after natural disasters. based on the literature analysis and field research above, it can be seen that the material and emotional demand factors in this period will affect people’s psychological cognitive evaluation, and then determine people’s behavior. the psychological behavior of the victims may suddenly change between panic and calm behavior. this sudden change is caused by the mutation of distance perception of people’s psychological state. the occurrence of natural disasters will break people’s existing psychological balance state, resulting in a state distance [37]. when the state distance exceeds a certain threshold, people’s psychological behavior will be in different states, resulting in a kind of jumping change. based on huagui’s psychological behavior framework, this paper puts forward a change mechanism model of the psychological behavior of victims under natural disasters, as shown in figure 2. according to the formation mechanism model, there are two kinds of relatively stable psychological behavior states of the victims after natural disasters: panic and calm behavior. panic behavior is a negative steady state. in this state, the disaster victims will have fear, quarrel with other disaster victims, conflict behavior, and even plunder materials and other risk behaviors that threaten social stability and order. however, when the psychological state distance of the victims gradually reduces to the psychological perception threshold, people’s behavior may suddenly turn into calm behavior, such as actively participating in the post disaster rescue and reconstruction work, helping other victims and other prosocial behaviors. table 2 | factors of victims’ psychology and behavior under natural disaster scenario influencing factors specific description references material demand factors to meet the basic needs of the affected people in terms of clothing, food, housing, transportation and medical and health care, mainly to maintain the basic life and life safety of the victims. research on the disaster area of lushan county in sichuan ya’an earthquake in 2013 [46]; tangshan earthquake related psychological problems [47]; “sangmei” typhoon psychological influencing factors [48]; connotation mode of disaster relief demand emotional demand factors comfort and psychological guidance from mental health personnel, volunteers, relatives and friends, which is a kind of psychological comfort demand. [49] risk cognition the process of processing, filtering and decoding signals related risk to form a self-understanding of risk and plays an important role in promoting people’s risk behaviors. [41–44,50,51] figure 2 | formation mechanism for individual victim’s psychological behavior changes during the recovery stage of natural disasters. x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 31 3. establishment of cusp catastrophe model of individual psychological behavior and simulation analysis of mutation mechanism 3.1. construction of cusp catastrophe model of individual psychological behavior of victims according to the formation mechanism of the individual psychological behavior state of the disaster victims, the individual psychological behavior state of the disaster victims can be regarded as the conjugates of material demand factors and emotional demand factors after the disaster. the individual psychological behavior x of the victims is regarded as the state variable, and the satisfaction degree u of the material demand factor and the satisfaction degree v of the emotional demand factor are regarded as the control variables to control the individual psychological behavior of the victims. the individual psychological behavior x is measured by the perception degree of psychological state, u, v, x ∈ [−1, 1]. v represents the potential function of the cusp catastrophe model of individual psychological state behavior of the victims [19], as shown in formula (1). v x u v x ux vx( , , ) = + +4 2 (1) the psychological behavior surface m of the victims is the set of equilibrium points of the catastrophe model, i.e., the equilibrium surface of the catastrophe model, as shown in formula (2). m v x u v x x ux v= ¶ ¶ = + + = ( , , ) 4 2 03 (2) the set of singularities n is the second derivative of potential function v, as shown in formula (3). n x u= + =12 2 02 (3) the catastrophe area b of the psychological behavior of the victims is the set of bifurcation points of the catastrophe model, which is the solution obtained by the simultaneous elimination of x by formulas (2) and (3), as shown in formula (4). b u v= + =8 27 03 2 (4) formula (2) is a cubic equation, which has either one real root or three real roots. the number of real roots is determined by the determination formula ∆: d = +8 27 3 2u v formula (2) is a equilibrium surface equation of individual psychological behavior state of the victims. assuming that the satisfaction degree of material and emotional needs and the perception degree of psychological state of the victims are the values between the intervals [−1, 1], the greater the satisfaction degree of material and emotional needs is, the smaller the perception degree of psychological state is. according to formula (2), matlab is used to simulate the influence of continuous changes of control variables between the intervals [−1, 1] on the state variables. random values of u and v in [−1, 1] interval is determined by rand function, a set of data is a set of u and v, i.e., (u, v), randomly composed of 1000 data. in order to satisfy the formula (2), the perception of mental state x also changes in turn. finally, the equilibrium surface of the cusp catastrophe model of the individual psychological state of the victims can be obtained as shown in figure 3. then, according to this method, use formula (4) to get the bifurcation set of cusp catastrophe of the individual psychological behavior of the victims, i.e., the projection of the fold surface in figure 3 on the control surface, as shown in figure 4. when ∆ > 0, formula (2) has only one real root. the continuous change of control variables u and v will cause the stable change of state variable x, and the psychological behavior of the victims will be in a stable state. when ∆ = 0, u and v satisfying formula (2) have the same real roots, even when u = v = 0, the three real roots are the same. at this time, the psychological behavior of the victims is in a critical stable state. when ∆ < 0, the three real roots of formula (2) are not the same. it can also be seen from figures 3 and 4 that the psychological behavior of the victims is figure 3 | the cusp catastrophe model equilibrium of individual victim’s psychological behavior. figure 4 | the bifurcation set of cusp catastrophe model of individual victim’s psychological behavior. 32 x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 unstable at this time, which is also the mutation of the psychological behavior of the victims at this time [52]. 3.2. mechanism analysis of individual psychological behavior mutation of victims according to the cusp catastrophe model equilibrium surface of individual victim’s psychological behavior in figure 3 and the bifurcation set of cusp catastrophe model of individual victim’s psychological behavior in figure 4, after being rotated and simplified, and the visual graphs of figures 3 and 4 are drawn with the software visio. figure 4 is the projection of the fold surface in figure 3 on the control surface, and the visual model of individual psychological behavior state evolution of the victims after the occurrence of natural disasters is obtained, as shown in figure 5. this can analyze the causes, process and degree of individual psychological behavior of the victims. the balance surface in figure 5 is divided into three parts: the upper, the middle and the lower leaf. the upper leaf represents the calm behavior of the victims, the lower leaf represents the panic behavior of the victims. the individual psychological behavior of the victims is relatively stable in the upper and lower leaves of the balance surface, and the middle leaf is an unstable state, i.e., the region where the individual psychological behavior of the victims suddenly changes. when the change curve of individual psychological behavior of the victims in figure 5 is a → b or d → e, the perception distance of psychological state of the victims ∆x gradually increases, but it has not exceeded the psychological perception threshold, and the degree of calmness of the victims gradually decreases. they become more and more irrational, which is still a gradual change, and no sudden change has occurred. when passing through points b and e respectively, the psychological behavior curve of the victims passes through bifurcation region b, which suddenly changes from calm to panic behavior. however, there are differences between the curves f1 and f2 projected to the set of control plane bifurcation points, which shows that the degree of panic caused by these two different behavior changes is different. when the state curve of psychological behavior of the victims is b → c, δf1 = x(ub, vb) − x(uc, vc), while when the state curve of psychological behavior of the victims is e → f, δf2 = x(ue, ve) − x(uf, vf), and it is easy to find that δf2 > δf1 through the bifurcation set of the control surface, which indicates that the harm degree of the curve of psychological behavior d → e → f is more serious. at this time, the psychological behavior of the victims may be looting materials, conflicting with others or even group affairs. when the psychological behavior curve is a → b → c, the victims may only panic and wait for rescue passively, which does not pose a great threat to social stability and order. through the analyses above, it can be seen that when the material and emotional needs of the victims are constantly changing, the psychological behavior of the victims will jump from calm to panic behavior, which poses a huge threat to the stability and order of the society, but the material and emotional needs factors have different effects on the psychological behavior of the victims. there are two situations to discuss: (1) case 1: when the material demand of the victims is greater than zero, i.e., when u > 0, 8u3 + 27v2 > 0, formula (4) has no solution, and the individual psychological behavior curve of the victims does not go through bifurcation set. with the continuous change of v, the psychological behavior state of the victims is only continuous change, and there is no jump phenomenon; (2) case 2: when the individual material demand of the victims is not more than zero, i.e., when u ≤ 0, 8u3 + 27v2 ≤ 0, the formula (4) has a solution. from figure 4 it can be seen that when v gradually changes from −1 to +1, the psychological behavior curve of the victims passes through the bifurcation set, and the psychological behavior state of the victims suddenly jumps. according to the catastrophe theory, the degree of material demand satisfaction u is the splitting factor, i.e., the change of u will split the balance surface m and produce different states, which determines whether the psychological behavior of the victims will suddenly change. as shown in figure 5, when the material demand satisfaction reaches a relatively high level, i.e., when u is close to 1, with the continuous change of emotional demand satisfaction, the individual psychological behavior curve cp of the victims changes continuously from the upper leaf of the balance surface to the lower leaf, and its projection on the control surface does not go through the bifurcation set, indicating that the calm level of the victims is becoming gradually lower, but it will not suddenly evolve into an irrational threat to social stability, which is going to take a while, while emotional demand factor v is a regular factor, i.e., when u > 0, the change of v only causes the smooth change of state variables. through figures 4 and 5, it can be seen that when the satisfaction degree of material demand u decreases to 0, the slight change of emotional demand factor v will produce different behavior paths n1 and n2. the n1 curve reaches the lower leaf of the balance surface through the bifurcation region, resulting in panic behavior, while the n2 curve reaches the upper leaf of the balance surface through the bifurcation region, resulting in calm behavior. 4. local verification of point mutation model of individual psychological behavior of victims under the natural disaster situation, the individual psychological behavior mutation of the victims is instantaneous and dynamic, and its mutation process is difficult to capture. it is difficult to figure 5 | the intuitive evolution model of individual victim’s psychological behavior of nature disaster during recovery stage. x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 33 describe the cusp mutation process of the individual psychological behavior of the victims with the actual data. this paper uses the method of questionnaire survey to locally verify the two control variable types of material demand and emotional demand, and illustrates the validity of the model. in this study, the disaster victims in lushan county were randomly sampled and investigated. a total of 405 questionnaires were issued and 306 effective questionnaires were recovered. the effective rate of the questionnaire was 75.5%. the psychological behaviors of the victims were measured by pcl-c, post-traumatic stress disorder (ptsd) can not only reflect the psychological state of the victims (such as fidgety and depression), but also reflect the behaviors of the victims [53]. the higher the score is, the more likely it is to represent ptsd. this scale has been used for ptsd measurement after disasters for many times, with good reliability and validity [54]. social support, social trust and social network are used to measure the relationship between the emotional needs of the victims and ptsd. likert 5-point scale is used in this part of the questionnaire. the larger the number is, the more emotional support is obtained. the lower the income is, the poorer the family is. thus the greater the social vulnerability is, which makes the economic endurance worse, impairing the ability to cope with the disaster, and rising the degree of material demand after the disaster [55]. therefore, the relationship between material demand and ptsd can be measured by the level of income. according to the average monthly income level of local disaster victims, it can be divided into five levels: 1 = 1000 yuan or less, 2 = 1001–2000 yuan, 3 = 2001–3000 yuan, 4 = 3001–4000 yuan, 5 = 4000 yuan or more. the higher the total score of ptsd is, the more obvious the symptoms of ptsd is. in order to describe the relationship between income, emotional support and ptsd, 306 victims were divided into five groups according to income level and emotional support with spss19.0, and the mean value of ptsd in each group was obtained, then the scatter diagram of income level and ptsd’s mean value was drawn, as shown in figure 6; the scatter diagram of social network and ptsd’s mean value was drawn, as shown in figure 7; the scatter diagram of social support and ptsd’s mean value was drawn, as shown in figure 8; the scatter diagram of social trust level and ptsd’s mean value, as shown in figure 9. it can be seen from figure 6 that with the increase of the monthly average income level, the ptsd’s mean value gradually decreases, but when the monthly average income reaches 3000 yuan, the ptsd’s mean value suddenly decreases, the psychological behavior of the victims has changed obviously. from figures 7–9, it can be seen that the average value of ptsd gradually decreases figure 6 | scatterplot of income level and the average of ptsd. figure 7 | scatterplot of social network level and the average of ptsd. figure 8 | scatterplot of social support level and the average of ptsd. figure 9 | scatterplot of social trust level and the average of ptsd. with the improvement of social network, social support and social trust, but the change of emotional support level and the average value of ptsd presents a linear relationship without sudden jumps, the mutation of the system happens only because of the change of material demand. this shows that it is basically in line with the actual situation to regard the material demand satisfaction as the splitting factor and the emotional demand satisfaction as the regular factor. this is also basically consistent with maslow’s theory of demand level: people’s needs are divided into different levels. only when the needs of the bottom are met, people will pay attention to the needs of the higher level. after the occurrence of natural disasters, the greatest threaten people are facing is the threat of survival. people’s most concerned needs are physiological needs and security needs. at the same time, people have social attributes, which need the communication between people to be met, especially in the post disaster psychological trauma period. 34 x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 5. coping strategies of individual psychological behaviors of victims two to three months after the occurrence of natural disasters, the disaster area began to recover slowly. the affected people’s emotions were relatively stable, sometimes they would be very rational and actively involved in the post disaster reconstruction work, but sometimes they would suddenly become irrational, have some conflicts with others, or even group affairs, which was not only related to personal characteristics, but also affected by the demand factors of people at that time. the psychological trauma of the victims is shown in the way of behavior under the influence of demand factors. through the analysis above, it can be seen that the needs of the victims should be fully met in this period, but the priority and degree of these needs are different. (1) establish a perfect distribution and transportation system of rescue materials to ensure that the victims receive rescue materials as soon as possible. through the analysis above, it can be seen that the satisfaction degree of material demand determines whether the psychological behavior of the victims will change suddenly. when the satisfaction degree of material demand u > 0, the psychological behavior of the victims will only change continuously, and will not change into panic behavior through the bifurcation set. however, when u gradually reaches 0, the psychological behavior curve of the victims will go through the bifurcation set, and the psychological behavior will jump. therefore, in the first time after the disaster, we should meet the material needs of the victims through various ways, especially for the low-income people, reduce their relative sense of deprivation in terms of materials, and improve the disaster response ability and economic endurance. (2) establish a lasting psychological intervention mechanism and provide a sound emotional support system. in the post disaster recovery period, people not only care about their own food, clothing, housing and transportation, but also have emotional needs which are to be cared and concerned. natural disasters bring huge economic losses and mental trauma to the victims, often lacking confidence and courage in reconstruction. at this time, the government and people from all walks of life need to provide certain emotional support to help the victims recover their confidence in reconstruction. through the above analysis, we can also see that when the material needs of the victims are gradually reduced to 0, the small changes in emotional needs can lead to completely different behavioral responses of the victims. therefore, in the post disaster recovery period, we should also attach great importance to the emotional needs of the victims, and make full use of various resources to help the victims recover social networking. 6. conclusion in the period of natural disaster recovery, the victims sometimes show a high degree of rationality and actively invest in the post disaster reconstruction work, but sometimes they suddenly change from calm to panic behavior, or even conflict behavior. in view of this problem, this paper analyzes the influencing factors of the individual psychological behavior state of the victims based on the literatures and the investigations of the disaster area, constructs the catastrophe model of the individual psychological behavior of the victims by using the catastrophe theory, depicts the intuitive model of the evolution of the individual psychological behavior state of the victims, probes into the catastrophe mechanism of the individual behavior of the victims. in addition, the model is partially verified by the actual survey data. through the analysis, it is found that the satisfaction degree of material needs is the decisive factor of the psychological behavior mutation of the victims, and the satisfaction degree of emotional needs is the non decisive factor of the psychological behavior mutation of the victims. however, when the satisfaction degree of material needs is low to zero, the slight change of emotional needs will make the victims produce different psychological behaviors. therefore, it is necessary to establish a perfect distribution system of rescue materials to meet the material needs of the victims at the first time, and at the same time, the emotional needs of the victims cannot be ignored. it is necessary for the government to establish a lasting psychological intervention mechanism to provide sound emotional support for the victims in order to prevent the irrational behaviors caused by the minor emotional changes of the victims. here, the following suggestions are put forward for relevant risk management professionals: in disaster prone areas, take families as a unit to investigate and count the mental health status of residents, such as the ability to resist stress, whether there are family members with severe ptsd, etc., and divide all families in disaster prone areas into “very healthy, healthy, general, serious and deadly serious” according to their mental health status, and archive them. in the same way, the economic income, social network, social trust and social support level are investigated one-by-one, and finally classified and archived. according to the conclusion of this study, the priority of material needs of the disaster victims is higher than their emotional needs. therefore, we decided that the weight of the economic income level, mental health status, emotional needs factors (i.e. social network, social trust and social support level) reduce inturn in order to score the family. among them, the weight of three influencing factors of emotional needs factors accounts for the same proportion. for example, 40% of the economic income level, 30% of the mental health status, 10% of the social network level, 10% of the social trust level and 10% of the social support level (the relevant risk management professionals can make a second decision in the future). the higher the score of a single family is, the higher the material demand is, and the lower the score is, the more emotional intervention the family needs. in this way, the government can help different families in different situations in the first time after the disaster, in order to strive for the stability and harmony of the whole society. in terms of resource allocation for disaster recovery, we can also see the level of material demand of each family in the database according to their scores, so as to determine the degree of subsidy to their economy. although this paper analyzes the mutation mechanism of the individual psychological behavior of the victims, the verification of the model has limitations. it needs to consider the interaction of individual characteristics (such as age, gender, education level, etc.), information dissemination, government psychological intervention, population density, social structure, disasters’ type and other factors on the evolution of the psychological behavior of the victims. at the same time, the research data in this paper is limited, and a large number of empirical data are still needed to further verify in the future. x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 35 conflicts of interest the authors declare they have no conflicts of interest. acknowledgment this research is supported by grants from the project of national nature science foundation of china (71671189, 71971217). references [1] ministry of civil affairs. the basic situation of nationwide natural disasters in 2014. available from: http://www.mca. gov.cn/article/zrzh/201412/index.htm# [20 january 2015] (in chinese). [2] cameron l, shah m. risk-taking behavior in the wake of natural disasters. national bureau of economic research (nber). j hum res 2015;50:484–515. [3] kunimatsu mm, marsee ma. examining the presence of anxiety in aggressive individuals: the illuminating role of fight-or-flight mechanisms. child youth care forum 2012;41:247–58. [4] zhengkui l, kankan w, li w. psychological and behavioral research after major disaster in china. adv psychol sci 2011;19:1091–8 (in chinese). [5] hu zh, sheu jb. post-disaster debris reverse logistics management under psychological cost minimization. transport res part b methodol 2013;55:118–41. [6] rennemo sj, rø kf, hvattum lm, tirado g. a three-stage stochastic facility routing model for disaster response planning. transport res part e logistics transport rev 2014;62:116–35. [7] chae j, thom d, jang y, kim sy, ertl t, ebert ds. public behavior response analysis in disaster events utilizing visual analytics of microblog data. comput graph 2014;38:51–60. [8] fan w. advisement and suggestion to scientific problems of emergency management for public incidents. bull natl nat sci found china 2007;21:71–6 (in chinese). [9] doherty tj, clayton s. the psychological impacts of global climate change. am psychol 2011;66:265–76. [10] norris fh, friedman mj, watson pj, byrne cm, diaz e, kaniasty k. 60,000 disaster victims speak: part i. an empirical review of the empirical literature, 1981–2001. psychiatry 2002;65:207–39. [11] kõlves k, kõlves ke, de leo d. natural disasters and suicidal behaviours: a systematic literature review. j affect disord 2013;146:1–14. [12] hu zh, sheu jb, xiao l. post-disaster evacuation and temporary resettlement considering panic and panic spread. transport res part b methodol 2014;69:112–32. [13] helbing d, farkas i, vicsek t. simulating dynamical features of escape panic. nature 2000;407:487–90. [14] huagui z. individual behavior in emergency decision-making. impact sci soc 2008:31–7 (in chinese). [15] bonanno ga, galea s, bucciarelli a, vlahov d. what predicts psychological resilience after disaster? the role of demographics, resources, and life stress. j consult clin psychol 2007;75:671–82. [16] van der linden s. the social-psychological determinants of climate change risk perceptions: towards a comprehensive model. j environ psychol 2015;41:112–24. [17] akerlof k, maibach ew, fitzgerald d, cedeno ay, neuman a. do people “personally experience” global warming, and if so how, and does it matter?. glob environ change 2013;23:81–91. [18] vallacher rr, nowak a. the emergence of dynamical social psychology. psychol inquiry 1997;8:73–99. [19] thom r. stabilité structurelle et morphogénèse. new york: benjamin; 1972. [20] bode s. catastrophe theory (translated by fuhua l.). shanghai: shanghai scientific and technological literature press; 1983 (in chinese). [21] zeeman ec. catastrophe theory. in: güttinger w, eikemeier h, editors. structural stability in physics. berlin, heidelberg: springer; 1979, pp. 12–22. [22] stewart in, peregoy pl. catastrophe theory modeling in psychology. psychol bull 1983;94:336–62. [23] van der maas hlj, kolstein r, van der pligt j. sudden transitions in attitudes. socio methods res 2003;32:125–52. [24] scott dw. catastrophe theory applications in clinical psychology: a review. curr psychol res rev 1985;4:69–86. [25] flay br. catastrophe theory in social psychology: some applications to attitudes and social behavior. behav sci 1978;23:335–50. [26] lu j, lianyu y. the application of primary catastrophe theory to social science. syst eng theor pract 2002;10:3–5 (in chinese). [27] xu z, bin h. research on staff counterproductive work behaviors based on catastrophe theory. j manage sci 2012;25:44–55 (in chinese). [28] exiang l, hong y, xingsan q. an application of catastrophe theory to employees’ behaviors of organizational change. sci technol prog policy 2007;23:181–3 (in chinese). [29] xu z, bin h. research on conflict behavior between enterprises and employees based on catastrophe theory and evolutionary game. oper res manage sci 2014;23:228–37 (in chinese). [30] yan x, bin h, yongqing y. a stochastic catastrophe model of knowledge sharing with in a team including simulations. oper res manage sci 2013;22:240–9 (in chinese). [31] hu b, xia n. cusp catastrophe model for sudden changes in a person’s behavior. inform sci 2015;294:489–512. [32] yiu ktw, cheung so. a catastrophe model of construction conflict behavior. build environ 2006;41:438–47. [33] chen dg, chen x, tang w. stochastic cusp catastrophe model and its sample size determination in social behavior research. in: proceedings of the 142nd apha annual meeting and exposition. new orleans, la, usa: apha; 2014. [34] castillo m, carter m. behavioral responses to natural disasters. 2011 (unpublished manuscript). [35] kan s. disaster psychology. sci press 2010:16–50 (in chinese). [36] vorst hcm. evacuation models and disaster psychology. proc eng 2010;3:15–21. [37] defu m. research on the disaster psychology and behavioral choices with bounded rationality of farmers. hubei soc sci 2010:76–8 (in chinese). [38] huagui z. analysis on victims’ panic behavior and influencing factors under emergency disaster situations. academia bimestris 2012:90–6 (in chinese). [39] duoyong s. study of individual and crowd behavior decision under paroxysmal social public crisis [d]. changsha: graduate school of national university of defense technology, 2005 (in chinese). [40] ma qg, wang xy. the mathematic description of the individual and crowd behavior under unconventional emergency. j indus eng manag 2009;23:126–30 (in chinese). http://www.mca.gov.cn/article/zrzh/201412/index.htm http://www.mca.gov.cn/article/zrzh/201412/index.htm https://doi.org/10.1007/s10566-012-9178-6 https://doi.org/10.1007/s10566-012-9178-6 https://doi.org/10.1007/s10566-012-9178-6 https://doi.org/10.1016/j.trb.2013.05.010 https://doi.org/10.1016/j.trb.2013.05.010 https://doi.org/10.1016/j.trb.2013.05.010 https://doi.org/10.1016/j.tre.2013.12.006 https://doi.org/10.1016/j.tre.2013.12.006 https://doi.org/10.1016/j.tre.2013.12.006 https://doi.org/10.1016/j.cag.2013.10.008 https://doi.org/10.1016/j.cag.2013.10.008 https://doi.org/10.1016/j.cag.2013.10.008 https://doi.org/10.1037/a0023141 https://doi.org/10.1037/a0023141 https://doi.org/10.1521/psyc.65.3.207.20173 https://doi.org/10.1521/psyc.65.3.207.20173 https://doi.org/10.1521/psyc.65.3.207.20173 https://doi.org/10.1016/j.jad.2012.07.037 https://doi.org/10.1016/j.jad.2012.07.037 https://doi.org/10.1016/j.jad.2012.07.037 https://doi.org/10.1016/j.trb.2014.08.004 https://doi.org/10.1016/j.trb.2014.08.004 https://doi.org/10.1016/j.trb.2014.08.004 https://doi.org/10.1038/35035023 https://doi.org/10.1038/35035023 https://doi.org/10.1037/0022-006x.75.5.671 https://doi.org/10.1037/0022-006x.75.5.671 https://doi.org/10.1037/0022-006x.75.5.671 https://doi.org/10.1037/0022-006x.75.5.671 https://doi.org/10.1016/j.jenvp.2014.11.012 https://doi.org/10.1016/j.jenvp.2014.11.012 https://doi.org/10.1016/j.jenvp.2014.11.012 https://doi.org/10.1016/j.gloenvcha.2012.07.006 https://doi.org/10.1016/j.gloenvcha.2012.07.006 https://doi.org/10.1016/j.gloenvcha.2012.07.006 https://doi.org/10.1207/s15327965pli0802_1 https://doi.org/10.1207/s15327965pli0802_1 https://doi.org/10.1007/978-3-642-67363-4_3 https://doi.org/10.1007/978-3-642-67363-4_3 https://doi.org/10.1007/978-3-642-67363-4_3 https://doi.org/10.1037/0033-2909.94.2.336 https://doi.org/10.1037/0033-2909.94.2.336 https://doi.org/10.1177/0049124103253773 https://doi.org/10.1177/0049124103253773 https://doi.org/10.1007/bf02686568 https://doi.org/10.1007/bf02686568 https://doi.org/10.1002/bs.3830230404 https://doi.org/10.1002/bs.3830230404 https://doi.org/10.1016/j.ins.2014.09.055 https://doi.org/10.1016/j.ins.2014.09.055 https://doi.org/10.1016/j.buildenv.2005.01.007 https://doi.org/10.1016/j.buildenv.2005.01.007 https://doi.org/10.1016/j.proeng.2010.07.004 https://doi.org/10.1016/j.proeng.2010.07.004 36 x. xu et al. / journal of risk analysis and crisis response 10(2) 27–36 [41] sheeran p, harris pr, epton t. does heightening risk appraisals change people’s intentions and behavior? a meta-analysis of experimental studies. psychol bull 2014;140:511–43. [42] milfont tl. the interplay between knowledge, perceived efficacy, and concern about global warming and climate change: a one‐ year longitudinal study. risk anal 2012;32:1003–20. [43] spence a, poortinga w, pidgeon n. the psychological distance of climate change. risk anal 2012;32:957–72. [44] wachinger g, renn o, begg c, kuhlicke c. the risk perception paradox—implications for governance and communication of natural hazards. risk anal 2013;33:1049–65. [45] sanduo z, chuanming c, lu minghong. management-principles and methods[m]. shaihai: fudan university press; 2009, pp. 69–80 (in chinese). [46] huijuan d, xiaojun l, manqing d, et al. study on problems related to earthquake disaster-caused [47] li-ping j, lan y, yu-ling w, yi-ting w, wei-qiong j. qualitative research of survivors’ cognition and psychological problems after saomai typhoon. chinese mental health journal 2008;929–33 (in chinese). [48] yanna s, wangyuhai, jianhui l. preliminary study of demand model of disaster relief and its [49] dynes rr, quarantelli el, kreps ga. a perspective on disaster planning[r]. ohio state univ columbus disaster research center; 1972. [50] naylor r, courtney p. exploring the social context of risk perception and behaviour: farmers’ response to bovine tuberculosis. geoforum 2014;57:48–56. [51] becker g, aerts jcjh, huitema d. influence of flood risk perception and other factors on risk‐reducing behaviour: a survey of municipalities along the rhine. j flood risk manag 2014;7:16–30. [52] 李智明,师恪.证券收益的尖点突变模型[j].新疆大学学报(自 然科学版) 2003;125–9. [53] yingjun x, guoyu x, yingqiang d, et al. prevalence of post traumatic stress disorder symptoms among middle school students after wenchuan earthquake. chinese mental health journal 2010;17–20 (in chinese). [54] weathers fw, litz bt, herman ds, et al. the ptsd checklist (pcl): reliability, validity, and diagnostic utility[c]//annual convention of the international society for traumatic stress studies. san antonio: international society for traumatic stress studies; 1993. [55] cutter sl, boruff bj, shirley wl. social vulnerability to environmental hazards. social science quarterly 2003;84:242–61. https://doi.org/10.1037/a0033065 https://doi.org/10.1037/a0033065 https://doi.org/10.1037/a0033065 https://doi.org/10.1111/j.1539-6924.2012.01800.x https://doi.org/10.1111/j.1539-6924.2012.01800.x https://doi.org/10.1111/j.1539-6924.2012.01800.x https://www.ncbi.nlm.nih.gov/pubmed/21992607 https://www.ncbi.nlm.nih.gov/pubmed/21992607 https://www.ncbi.nlm.nih.gov/pubmed/23278120 https://www.ncbi.nlm.nih.gov/pubmed/23278120 https://www.ncbi.nlm.nih.gov/pubmed/23278120 基于随机赋权法和云模型的黄河流域水安全风险评估 flood disaster risk assessment and spatial distribution characteristics along the yangtze river in anhui province xianfu cheng1,2, , honghu sun1,2, zhang yuan1,2, guanglai xu1 1. college of territorial resources and tourism, anhui normal university, wuhu 241003, china 2. anhui key laboratory of natural disaster process and prevention, wuhu 241003, china abstract using analytic hierarchy process and entropy method to determine the weight of index, fuzzy comprehensive evaluation model of risk assessment was constructed in this paper. flood risk index and risk rating of area along the yangtze river in anhui province were drawn by using gis spatial analysis technology. from disaster causing factors, inducing environment and bearing body, average maximum 3 day rainfall, rainstorm number, vegetation coverage, river network density, standard deviation of the elevation, population density, proportion of rural population, density of gdp, per capita net income of farmers, and cultivated land area ratio were selected as evaluation indicators. the results show that flood risk rating and risk index of area along the yangtze river in anhui province has a spatial distribution characteristics from south to north gradually reduced. keywords: flood disaster; risk assessment; comprehensive assessment; spatial distribution, anhui province 安徽沿江地区洪涝灾害风险评价与空间分布特征 程先富 1,2 孙鸿鹄 1,2 张媛 1,2 徐光来 1 1.安徽师范大学国土资源与旅游学院，安徽 芜湖 241003, 中国 2.安徽自然灾害过程与防控研究省级实验室，芜湖 241003, 中国 摘要：本文运用层次分析法和熵值法确定指标的权重，构建风险评价的模糊综合评价模型，并利用 gis 空 间分析技术绘制出安徽沿江地区洪灾风险指数图和风险等级图。我们从致灾因子、孕灾环境和承灾体三方 面选取平均最大 3 日降雨量、暴雨次数、植被覆盖率、河网密度、高程标准差、人口密度、乡村人口比重、 gdp 密度、农民人均纯收入、耕地面积比等 10 个评价指标。研究结果表明：安徽沿江地区洪灾风险等级和 风险指数具有自南向北逐渐降低的空间分布特征。 关键词：洪涝灾害，风险评价，综合评价，空间分布，安徽 1. 引言 随着全球气候变化和城市化进程的加快，在自然 和人为因素的相互作用下，暴雨、洪涝等气象灾害发 生的频率、强度越来越大，造成的经济财产的损失越 来越严重。洪涝灾害的风险研究受到许多学者的广泛 关注，并成为灾害科学研究的前沿课题 [1] 。美国、日 本等发达国家早在20 世纪50、60 年代就开展了洪涝 灾害风险研究，制作了国家级的洪涝灾害风险图 [2-5] 。 我国从20 世纪80 年代中期开始开展洪涝灾害风险研 究，并对一些蓄滞洪区、城镇、水库与流域进行洪涝 灾害风险图的绘制。 目前国内外对洪涝灾害风险评估开展了深入研 究。大多集中于风险概率的计算、风险指标体系的构 建与评估、风险建模与评估等三个方面 [6-8] 。洪涝灾害 风险评价常用的数学方法有 [9-13] ：统计分析方法、决策 分析法和综合评价法等。统计分析方法包括主成分分 析法、相关分析法、回归分析法和时间序列分析法， 其中回归分析法是比较常用的方法。这种方法的不足 之处在于当样本容量足够大时，统计结果才比较可靠。 决策分析法中最常用的是层次分析法，思路清晰且系 统性强，所需定量数据较少，对问题本质分析得较透 彻，颇具实用性，但具有主观性强的缺点。综合评价 法包括模糊数学、灰色系统和人工神经网络等。此外， 还有系统动力学方法、灾害风险指数法和基于扩散理 论的风险评价方法等。其中模糊综合评价法是采用模 journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 238-242 published by atlantis press copyright: the authors 238 willieb typewritten text received 20 january 2014 willieb typewritten text accepted 14 september 2014 willieb typewritten text willieb typewritten text x. f. cheng et al 糊数学的方法进行综合评价，克服了从多方面对事物 进行评价难免带有模糊性和主观性的缺点，使评价结 果尽量客观从而取得更好的实际效果。 安徽沿江地区是安徽省经济较发达的地区。在全 球气候变化、城市化和人口与经济快速集聚的背景下， 该地区洪涝灾害发生较为频繁，灾情日益严重。前人 在本区开展的研究主要集中在极端降水、径流、灾害 成因与规律等方面。如吴丹娃等 [14] 分析了 2010 年安徽 沿江地区一次暴雨的中尺度特征；张之丽 [15] 分析了安 徽沿江圩区灾害的成因；朱维民等 [16] 探讨了安徽沿江 地区旱涝规律及其未来趋势。对洪涝灾害风险研究较 少。张海玉等 [17] 应用模糊数学方法评价了安徽沿长江 地区洪涝灾害的经济易损性。本文根据巢湖流域的实 情，选取评价指标，建立评价指标体系和模糊数学综 合评价模型，基于综合评价模型和 gis 空间分析技术 对安徽沿江地区洪涝灾害风险进行评价，揭示其空间 分布特征，为沿江地区制定规划、灾害风险管理与可 持续发展提供科学依据。 2. 研究区域和数据来源 2.1 研究区概况 安徽沿江地区位于安徽省南部，处于 116°～ 119°e ，30°～32°n 之间，下辖马鞍山市、芜湖市、 铜陵市、池州市、安庆市、和巢湖市 6 个地级市 20 个 县，总面积 39025 km2，占全省总面积的 27.95%，2008 年人口 1671 万人，占全省总人口的 24.8%。属于亚热 带湿润季风气候，夏季高温多雨，冬季寒冷干燥。年 平均气温 16-17.3℃，常年无霜期达 220-250d，年平均 降水量 770-1700mm 之间，降水量季节分配不均，多 集中在 5-9 月，地形以丘陵山地和平原为主。 2.2 数据来源 本文的气象资料来源于安徽省 78 个气象站点， dem 数据来源于地球系统科学数据共享网 strm 系 统 90m 空间分辨率，遥感数据为安徽沿江地区 2008 年 8 月 tm 影像，经解译获得安徽沿江地区土地利用 图和植被覆盖率图。社会经济数据来自于安徽省统计 年鉴及沿江各县市的统计年鉴。 3. 研究方法 3.1 指标体系的建立 在分析安徽沿江地区地理特征的基础上，依据洪 灾风险的相关原理，遵循数据尽量空间化、技术手段 先进有效、指标体系可操作性强等原则，从致灾因子、 孕灾环境和承灾体三个层面选取平均最大 3 日降雨 量、暴雨次数、植被覆盖率、河网密度、高程标准差、 人口密度、乡村人口比重、gdp 密度、农民人均纯收 入、耕地面积比 10 个评价指标，构建安徽沿江地区洪 涝灾害风险评价指标体系。 3.2 指标体系权重的确定 分别用层次分析法和熵权法求得主观权重和客观 权重，然后对主观权重和客观权重运用最小信息熵法 得到综合权重。ahp 确定评价指标权重的步骤如下： 采用 1～9 及其倒数的标度方法，对两两元素相对重 要性程度进行量化。构造判断矩阵，计算准则权重。 根据各层次所有元素的重要性权重值进行层次单排序 与总排序；利用一致性指标(ci)、随机一致性指标(ri) 和一致性比率(cr)进行一致性检验。若 cr≤0.1，则 判断矩阵具有满意的一致性。通过计算得到致灾因子、 孕灾环境和承灾体构成的判断矩阵的最大特征根为 3.0078，ci=0.0048，ri=0.59，cr=0.0077，表明一致 性较好。特征向量（归一化后）即为主观权重向量。 3.2.1 求算评价指标的客观权重 运用熵权法求算熵值，其公式如下： j ij n i j ij j d d d d e ln 1 ∑ = −= （1） ∑ = = n i ijj dd 1 （2） ∑∑ = = = n i m j ij ij ij d d d 1 1 （3） 式中，i 为评价单元；j 为评价指标，ej 为第 j 个 评价指标的熵值； dij 为接近程度；dj、dij 为中间变量。 对于正向指标 }{max* ij i j rr = *j ij ij r r d = （4） 对于负向指标 }{min* ijij rr = ij j ij r r d * = （5） 如果 ej 越大，j 指标越重要，则 m e e jj ln = ∑ = = m j j j j e e 1 θ （6） 如果 ej 越小，j 指标越重要，则 m e e jj ln = ∑ = − − = m j j j j en e 1 1 θ （7） 式中，θj 为 j 指标的客观权重， ej 为中间变量。 3.2.2 综合权重的确定 根据最小信息熵原理，对主观权重和客观权重进 行运算从而得到综合权重。 ∑= 2 1 21 2 1 21 )*(/)*( jjjjj wwwww （8） 式中，wj、w1j、w2j 分别为 j 指标的综合权重、主 观权重和客观权重。 通过计算，平均最大 3 日降雨量、暴雨次数、植 被覆盖率、河网密度、高程标准差、人口密度、乡村 人口比重、gdp 密度、农民人均纯收入、耕地面积比 的综合权重分别为 0.3207,0.0406,0.0889,0.1401,0.1018, 0.0805,0.0647,0.0548,0.0599,0.0480。 published by atlantis press copyright: the authors 239 willieb typewritten text willieb typewritten text flood disaster risk assessment and spatial distribution characteristics 3.3 模糊评价模型的建立 选择标准差的平均值与均值的平均值中的最小值 作为指标数据模糊子集分级的间隔值，每个指标数据 都有 5 个间隔点（d1、d2、d3、d4、d5），指标数据 分级标准见表 1。植被覆盖率、高程标准差、农民人 均纯收入与洪涝灾害风险呈负相关关系，故对其取倒 数，其对应倒数与洪涝灾害风险呈正相关关系。 3.4 隶属函数的建立 根据研究区的实际情况，并参考相关的研究成果， 运用分段线性函数确定洪涝灾害各个风险等级的隶属 函数。将洪灾划分为高风险区、较高风险区、中等风 险区、较低风险区、低风险区五级。其中高风险区基 于升半梯形函数，低风险区基于降半梯形函数，其余 等级区均基于三角形函数。根据指标数据分级标准及 分段函数，本项目建立了洪灾风险统一的模糊子集。 rij 表示第 i 个指标第 j 个风险等级的模糊隶属度，gi 表 示第 i 个指标的数值，五个等级的隶属函数如下。 1 d1 ≤ gi ≤ d2 ri1= 12 2 dd gd i − − d1<gi<d2 （9） 0 gi ≥ d2 0 gi ≤ d1 或 gi ≥ d3 12 1 dd dgi − − d1<gi<d2 ri2= 1 gi=d2 （10） 23 3 dd gd i − − d2<gi<d3 0 gi ≤ d2 或 gi ≥ d4 23 2 dd dgi − − d2<gi<d3 ri3= 1 gi=d3 （11） 34 4 dd gd i − − d3<gi<d4 0 gi ≤ d3 或 gi ≥ d5 34 3 dd dgi − − d3<gi<d4 ri4= 1 gi=d4 (12) 45 5 dd gd i − − d4<gi<d5 0 gi ≤ d4 ri5= 45 4 dd dgi − − d4<gi<d5 （13） 1 gi ≥ d5 3.5 模糊综合评价模型的构建 根据隶属函数，计算出各指标对应各等级的隶属 度，构成 i×j 阶隶属关系矩阵 r。将上述计算得到的 综合权重向量 a 与模糊关系矩阵 r 进行模糊合成运 算，得到模糊综合评价结果向量 b。对结果向量 b 取 最大值 gridmax，则 gridmax 隶属于评判集中的某个等级， 就得到所需要模糊综合评价等级，即洪涝灾害风险等 级。再将向量 b 所对应等级的序加权求平均，得到 gridmean，即洪涝灾害风险指数。 gridmax = max（b） (14) gridmean= ∑∑ × maxmax gridjgrid (15) 表 1 指标数据分级标准 table 1 division standard of index data. 间隔值 d1 d2 d3 d4 d5 年均最大 3 日降雨量(mm) 15.2 50.0 65.2 80.4 95.6 110.8 暴雨次数 1.0 3.0 4.0 5.0 6.0 7.0 植被覆盖率(取倒数) 0.35 1.00 1.35 1.70 2.05 2.40 河网密度(m/m2) 0.020 0.018 0.038 0.058 0.078 0.098 高程标准差(m,取倒数) 0.38 0.27 0.65 1.03 1.41 1.79 人口密度(人/km2) 235.67 167.43 403.10 638.77 874.44 1110.12 乡村人口比重(%) 0.1072 0.4345 0.5416 0.6488 0.7559 0.9631 gdp 密度(gdp /km2) 531.75 180 711.75 1243.49 1775.24 2306.98 农民人均纯收入(元,取倒数) 0.00004 0.00016 0.00020 0.00024 0.00028 0.00032 耕地面积比(%) 6.45 6.15 12.60 19.05 25.50 31.95 published by atlantis press copyright: the authors 240 x. f. cheng et al 4. 结果与分析 4.1 安徽沿江地区洪涝灾害风险空间分布特征 基于 gis 空间分析技术，栅格大小为 100m× 100m，运用模糊综合评价方法进行评价，得到安徽沿 江地区洪涝灾害风险指数图（图 1），运用最大隶属度 法确定洪涝灾害风险等级（图 2）。 图 1 安徽省沿江地区洪涝灾害风险指数 fig.1 flood risk index along yangtze river in anhui province 图 2 安徽省沿江地区洪涝灾害风险等级 fig.2 flood risk rank along yangtze river in anhui province 从空间分布格局上来看，安徽沿江地区洪灾风险 等级和风险指数均有自南向北逐渐增加的趋势。高风 险区主要集中在北部和沿江干流两侧的区域，该区地 势低平、经济易损性相对较高、河网密度较大、耕地 面积比重大，遭受洪涝灾害威胁的可能性就相对较大。 低风险区主要集中分布于皖南山地区和西部部分地 区，该区地形起伏较大、植被覆盖状况良好、经济发 展水平相对较低、耕地面积比重小，遭受洪涝灾害威 胁的可能性相对较小。 从风险等级构成上（图 2）来看，安徽沿江地区 高风险区面积约 11225.3km2，占区域总面积的 28.76%， 较高风险区所占比例较小，为 8.92%，约 3481km2。中 风险区、较低风险区及低风险区所占面积比例依次为 23.01%、20.15%和 19.16%。 4.2 安徽沿江地区各县（市）洪涝灾害风险分布特征 从各个县（市）的风险等级构成来看，马鞍山市、 庐江县、铜陵市、巢湖市、岳西及潜山县的高风险区 占县市面积的 50%以上，政府应加强对洪涝灾害的预 警及抵御工作。池州市、东至县、石台县、青阳县的 高风险区面积比例均低于 10%，且低风险所占面积比 重较大，说明其遭受洪涝灾害威胁的可能性较小。怀 宁县、安庆市、宿松县、枞阳县、无为县的中风险区 比例超过 40%以上，且高风险区及低风险区比例和不 超过 30%，说明其遭受洪灾的可能性一般。 从各个县（市）的风险指数均值来看，区域内各 县（市）的平均风险指数差异不大，均处于中等风险 水平和较高风险水平。属于较高风险水平的县（市） 有庐江县、巢湖市、马鞍山市、桐城市、含山县、安 庆市、枞阳县、和县、怀宁县、当涂县、潜山县、岳 西县、望江、县、铜陵市、太湖县、无为县，平均风 险指数依次为 3.726、3.522、3.381、3.375、3.351、3.342、 3.323、3.294、3.214、3.200、3.182、3.134、3.117、 3.046、3.033、3.012；属于中等风险水平的有芜湖市、 石台县、宿松县、芜湖县、池州市、铜陵县、东至县、 南陵县、繁昌县、青阳县，平均风险指数依次为 2.988、 2.891、2.873、2.811、2.762、2.734、2.723、2.713、 2.676、2.612。 5 结论 运用层次分析法和熵权法求算综合权重，构建模 糊综合评价模型，并结合 gis 空间分析技术，绘制了 安徽沿江地区洪涝灾害风险指数图和风险等级图，对 安徽沿江地区洪涝灾害风险的空间分布特征得到了较 为全面的认识。应用 gis 空间分析技术和栅格计算方 法，大大地提高了分析的精度。如果进一步选取评价 指标，提高网格精度和社会经济数据空间展布精度， 安徽沿江地区洪涝灾害风险格局研究会更有意义。今 后应结合遥感等实时数据及 3s 技术，模拟人类活动 干扰下洪涝灾害风险发展过程，可视化表达洪涝灾害 灾害风险，实现洪涝灾害风险的动态评估。 published by atlantis press copyright: the authors 241 flood disaster risk assessment and spatial distribution characteristics 致谢 本研究得到了国家自然科学基金项目(41271516)资助。 参考文献 1. shi peijun. theory on disaster science and disaster dynamics[j]. journal of natural disasters,2002,11(3):1-9. 史培军 . 三论灾害研究的理论与实践 [j]. 自然灾害学 报,2002,11(3):1-9. 2. dilley m, chen r.s, deiehmann u, et al. natural disaster hotspots: a global risk analysis synthesis report[r]. washington d c: hazard management unit, world bank, 2005. 3. apel h, thieken a h, merz b, et al, flood risk assessment and associated uncertainty[j]. natural hazards and earth system sciences, 2004 (4):295-308. 4. maxx d, robert s c, uwe d, et al. natural disaster hotspots: a global risk analysis[r]. world bank,2005. 5. islam a e , elsayed h , mohammed e b, gis-modelling of the spatial variability of flash flood hazard in a bu dabbab catchment, red sea region, egypt[j]. the egyptian journal of remote sensing and space sciences,2010 (13):81-88. 6. emlyn hagen j f, shroder j, lu x x, et al, reverse engineered flood hazard mapping in afghanistan: a parsimonious flood map model for developing countries[j].quaternary international ,2010 (226):82-91. 7. zhang hui, zhang jiquan, han junshan. gis-based assessment and zoning of flood/waterlogging disaster risk: a case study on m iddle and lower reaches of liaohe river[j]. journal of natural disasters, 2005,14(6):141-146. 张会,张继全,韩俊山.基于 gis 技术的洪涝灾害风险评估 与区划研究—以辽河中下游地区为例[j].自然灾害学 报,2005,14(6):141-146. 8. li ning, zhai yaxin, wang wei, et al. a comprehensive method to evaluate flood disasters based on basyesian stochastic assessment method[j]. journal of beijing union university (natural sciences), 2011, 25(3):70-73. 李宁,翟亚欣,王威,等. 基于贝叶斯随机评价方法的洪水 灾情等级评价[j].北京联合大学学报(自然科学版), 2011, 25(3):70-73. 9. zou qiang, zhou jianzhong, zhou chao, et al. flood disaster risk analysis based on variable fuzzy set theory[j]. transactions of the chinese society of agricultural engineering (transactions of the csae), 2012, 28(5): 126-132. 邹强,周建中,周超,等. 基于可变模糊集理论的洪水灾害 风险分析[j]. 农业工程学报,2012,28(5):126-132. 10. zhang jingjing. risk assessment and spatial-temporal distribution characteristics of flood and drought disasters in henan province[j]. transactions of the chinese society of agricultural engineering (transactions of the csae), 2012, 28(18):98-106. 张竟竟.河南省农业水旱灾害风险评估与时空分布特征 [j].农业工程学报,2012,28(18):98-106. 11. liu xinli, shi peijun. the theory and application of model research of regional flood risk assessment[j]. journal of natural disasters, 2001,10(5):66-72. 刘新立,史培军.区域水灾风险评估模型研究的理论与实 践[j].自然灾害学报,2001,10(5):66-72. 12. jin juliang, zhang libing, wei yiming. application of genetic algorithm based on ideal interval method to evaluation of flood disaster loss[j]. scientia geographica sinica, 2004,24(5):586-590. 金菊良,张礼兵,魏一鸣.基于遗传算法的理想区间法在洪 水灾情评价中的应用[j].地理科学,2004,24(5):586-590. 13. huang chongfu, zhang junxiang, chen zhifen, et al. toward a new kind of natural disaster risk zoning map[j]. journal of natural disasters, 2004,13(2):9-15. 黄崇福,张俊香,陈志芬,等.自然灾害风险区划图的一个 潜在发展方向[j].自然灾害学报, 2004,13(2):9-15. 14. wu danwa,yan xiaojing. mesoscale ch aracteristics analysis of a heavy rainfall along the yangtze river regions in anhui in 2010[j].journal of anhui agri sci, 2011,39(5):2756-2759 吴丹娃,严小静.2010 年安徽沿江地区一次暴雨的中尺度 特征分析[j].安徽农业科学.2011,39(5):2756-2759. 15. zhang zhili. disaster analysis and management planning in polder area along the yangtze river in anhui province[j].journal of hydraulic engineering, 1981,3:56-61. 张之丽.安徽沿江圩区灾害成因分析及治理规划意见[j]. 水利学报,1981,3:56-61. 16. zhu weimin,huang shixiang,fan dongguang. law and trend of drought/flood in the region along the yangtzer river of anhui[j]. journal of anhui agricultural university, 1988(3):43-49. 朱维民,黄世祥,范东光.安徽沿江地区旱涝规律及其未来 趋势[j]. 安徽农学院学报,1988(3):43-49. 17. zhang haiyu, cheng xianfu, ma wu. fuzzy assessment of the vulnerability risk of flood disaster: a case study on area along the yangtze river in anhui province[j]. journal of catastrophology, 2010,25(1):30-34. 张海玉, 程先富,马武,洪涝灾害经济易损性模糊评价— 以安徽沿长江地区为例[j].灾害学. 2010, 25(1):30-34 published by atlantis press copyright: the authors 242 1. 引言 2. 研究区域和数据来源 3. 研究方法 3.1指标体系的建立 4. 结果与分析 5 结论 atlantis press journal style shipwreck: a crisis with challenging solutions nikolaos p. ventikos school of naval engineering & marine architecture, laboratory for maritime transport, maritime risk group national technical university of athens athens, zografou, 15780, greece email: niven@deslab.ntua.gr alexandros koimtzoglou school of naval engineering & marine architecture, laboratory for maritime transport, maritime risk group national technical university of athens athens, zografou, 15780, greece e-mail: alexkoim@gmail.com konstantinos louzis school of naval engineering & marine architecture, laboratory for maritime transport, maritime risk group national technical university of athens athens, zografou, 15780, greece e-mail: konstantinos.lusis@gmail.com abstract in this paper, we deal with the subject of shipwreck removal and the methods that are employed during removal operations. we use a case study to illustrate the stages of these operations as well as the parameters that affect the process. furthermore, we outline a number of different scenarios of the most common techniques in order to demonstrate which one is the better fit for the particular case. the purpose of this paper is to highlight the difficulties that arise in shipwreck removal operations and offer a structured methodology for the planning stage of such a process. keywords: shipwrecks, wreck removal operation, wreck removal factors. 1 introduction shipwrecks are a growing concern for maritime authorities and port states around the globe either because they inhibit navigation or they pose an environmental threat due to leaks of toxic substances such as oil. removing the wreck is the best solution to both problems, even if in some cases such an operation is not cost-effective. wreck removal is a field with high levels of uncertainty and a large number of associated risk factors. additionally, because each case is unique there is great difficulty in formalizing the process, which is journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 160-174 published by atlantis press copyright: the authors 160 willieb typewritten text received 26 may 2014 willieb typewritten text accepted 30 july 2014 the reason only general instructions can be provided. furthermore, every shipwreck is a challenge for the salvors due to the fact that many parameters, which are also interdependent on each other, should be taken into account. this paper attempts to highlight the main parameters of a wreck removal operation, as well as to present the challenges faced and the risks assumed by salvors during wreck removal operations, by examining an illustrative case study. another goal of this paper is the definition of a framework for the preliminary assessment of different wreck removal methods. 2 literature review the concern over the environmental impact of sunken shipwrecks has been sparked the last few years mainly due to observed oil leaks that have been connected to wrecks such as the ss jacob luckenbach1 and the uss mississinewa2. this has led to the identification of shipwreck locations and the development of dedicated databases with international geographical scope such as the work in ref. (3) as well regionally focused studies. such studies are the work in ref. (4) that focuses on the pacific region, the work in ref. (5) that focuses on the skagerrak region of sweden and ref. (6), where a database of shipwrecks in greece has been developed. the same studies have dealt with the environmental risks connected to shipwrecks, which stem mainly from the leak of residual oil, as well as with developing risk estimating methodologies in order to rank the impact of each shipwreck and aid in the decision of allocating resources. recent efforts in this field are focusing on developing an acceptable general risk framework for shipwrecks. such an effort is described in ref. (7), which also compares the existing risk estimating models for shipwrecks. removing the wreck is the most drastic but also most effective way of dealing with all associated risks, either navigational or environmental. the available literature on the subject of wreck removal offers a detailed description of the theoretical and practical tools used by the salvors in these kinds of operations. manuals such as ref. (8) and ref. (9) cover a wide array of subjects from elements of naval architecture for the salvage engineer and rigging systems to environmental forces and surveys of the shipwreck. however, the planning and wreck removal method selection has not been formalized beyond some simple and general guidelines as to how they might be implemented. 3 the environment of a wreck removal operation the weather, the wind force, the sea and the seabed are the factors that define the environment of a wreck removal operation. each one of them can determine the method to be applied and lead an operation to failure, or contribute to its abortion altogether. although the estimation of these factors is difficult, they must be taken into consideration to conclude the feasibility of the operation of interest. on the other hand, there are circumstances that some environmental factors can be beneficial for an operation, for example, the salvage of a stranded ship by taking advantage of the tide. 3.1 weather weather is a very important factor for a wreck removal operation, because it will define when or even if the operation will commence. choosing the proper time frame for the operation, e.g. the season of year, may reduce its overall expenses and duration. towards this goal, the improved accuracy of the weather forecasts is a valuable tool for the proper organization of the wreck removal operation. 3.2 wind force the force of the wind is an important environmental factor, because when the shipwreck has been lifted above the surface of the sea it has a high exposed wind profile area. moreover, the stability of the equipment used for this kind of operations such as derricks, barges and support ships can be greatly affected by the wind. therefore the need for appropriate anchorage systems arises. the accelerations induced by the wind increase the risk level both for the integrity of the used equipment and for the safety of the salvors at work. the only precaution against the wind is to know the operational limits of the equipment and when to halt the operation. published by atlantis press copyright: the authors 161 3.3 the sea the environment surrounding a shipwreck can affect a wreck removal operation through various parameters, the most important of which are listed below:  waves,  currents,  water resistance,  buoyancy. 3.3.1 waves the accelerations induced by the waves to the support ships working on the surface affect the shipwreck. therefore, depending on the magnitude of the waves an operation can be either extremely dangerous or even impossible. however, sometimes wreck removal operations must be conducted even under heavy weather, as was the case of the submarine kursk.10 for the reasons stated above, it is vital during the planning stage of the wreck removal operation to take into account the action of the waves, depending on the geographical location and the season of the year. the data can be easily provided by the various oceanographic institutes around the globe. 3.3.2 currents the effect of currents to a wreck removal operation depends on their intensity. strong sea currents, which are able to alter the morphology of the seabed, can change the position of the shipwreck, or even bury it. furthermore, currents affect the maneuverability and anchoring of the equipment used for the operation, as well as the ability to effectively control the wreck during the lifting stage. however, in some cases the current may work advantageously for the operation and unbury the shipwreck from the seabed. 3.3.3 water resistance every object that moves inside a fluid experiences resistance. the magnitude of this force depends mainly on the following factors11: a) the projected area to the direction of the movement and, b) the squared velocity of the object. the projected area of a shipwreck is normally a number with an order of magnitude in the thousands or tens of thousands square meters. however, the main factor of the resistance force is the velocity of the wreck while ascending to the surface. a commonly used lifting velocity for a wreck removal operation is around 2-3 m/h. however, sometimes the prevailing conditions dictate faster lifting, as was the case of the submarine kursk10 where the lifting velocity was 10 m/h, which is considered high. due to the fact that usually the lifting speeds are quite low, the effect of the water resistance in a wreck removal operation should be considered negligible. 3.3.4 buoyancy of plating a buoyancy force is applied to every submerged object because of the volume of water the object displaces. the magnitude of this force is not significant for a submerged shipwreck but it reduces the required lifting capacity. the buoyancy of the plating can be estimated by the following equation: a v g   . (1) where: a, is the buoyancy of the plating, in n, v, is the volume of the displaced water, in m3, ρ, is the density of sea water, in kg/m3, g, is the acceleration of gravity equal to 9.81 m/sec2. the main difficulty is the precise calculation of the displaced water volume (v). a sufficient estimation for v is to divide the lightship weight by the specific gravity of steel. 3.4 the seabed the nature of the seabed can affect a wreck removal operation due to the following perspectives: first the sediments settling with time on the hull of the wreck (and inside the wreck) and second the seabed reaction force, the overcoming of which needs the application of a breakout force. 3.4.1 seabed sediments every submerged object suffers from the sediments of the seabed settling on it, causing, among others, corrosion issues. usually, sediments consist of clay, mud and microorganisms. the accumulated sedimentation may increase significantly the weight of the shipwreck. published by atlantis press copyright: the authors 162 the best way to calculate this additional weight is to conduct an on-site survey and collect a sample from the sediments in order to determine the respective specific gravity. in cases when that is not possible, an acceptable estimation is 1,600 kg/m3.8 if the calculated additional weight of the sediments is high enough, then salvors have to remove an adequate amount, by using specialized pumps. the removal of the stern of msc napoli is an example that shows how sediments can affect such an operation. during this operation, the chains broke on two separate occasions because of the mud that had settled on the wreck.12 3.4.2 seabed reaction when the hull of the wreck has been embedded in the seabed, either due to the action of sea currents, or to its own weight, a reaction force is applied to the wreck. in order to unbury the wreck, a breakout force greater than the reaction of the seabed must be applied. however, in many cases the available lifting capacity is not high enough for immediate breakout and therefore it may take some time to overcome the reaction of the seabed. the exact amount of time is not fixed and depends on the parameters of each specific case. depending on the type of the seabed the magnitude of the required breakout force varies. this is the reason why it is important to determine the soil type of the bottom of the sea prior to setting up the wreck removal operation. ref. 9 offers a way of calculating the immediate breakout force as well as the time needed to detach the shipwreck from the seabed, when the lifting capacity is not high enough. the first step is to calculate the soil bearing strength (qu) of the seabed. the next step is to calculate the surface of the ship in contact with the seabed (a), which is the same as the wetted surface area of the ship, calculated for a draught equal to the penetration depth of the wreck. by estimating the soil bearing strength and the embedded surface of the wreck we are able to use the equation below to estimate the bearing capacity of the seabed: q uf a q  . (2) where: fq, is the bearing capacity of the seabed, in n, a, is the area of contact with the soil, in m2, qu, is the soil bearing strength, in pa. after estimating the soil bearing capacity, the immediate breakout force that is required to unbury the shipwreck must be calculated. this calculation is based on empirical equations and therefore the results are usually overestimated. the following equation is used for calculating the immediate breakout force (fib): 2.75( ) [1 0.97 ] d b ib q f f e    (3) where: fq, is the bearing capacity of the seabed, in n, b, is the equivalent breadth of the wreck, in m, d, is the equivalent embedment depth of the wreck, in m. the area of the ship, which is in contact with the seabed (a), is assumed to be a rectangular shape of the same area. the length of the equivalent rectangular (l) equals the ship’s length. the equivalent breadth (b) is calculated as a function of the area (a) and the length (l). for calculating the equivalent embedment depth (d) we need the volume of the seabed soil (v) that has been displaced by the wreck. this volume is equal to the volume of displaced water at a draught equal to the embedment depth. therefore, the equivalent depth d is equal to the ratio v/a. the immediate breakout force (fib) is the force needed to immediately detach the shipwreck from the seabed. however, in most cases the immediate breakout force is much higher than the available lifting capacity, which means that a smaller force (fap) will be applied to the shipwreck. the applied breakout force (fap) is calculated from the following formula, where negative are the forces with the same direction as weight and positive the ones with the opposite direction. 0.75ap ship sed plf lc w w b     . (4) where: fap, is the applied breakout force, in t, 0.75 is a safety factor reducing the available lifting capacity9, ` lc, is the lifting capacity, in t, wship, is the weight of the wreck, in t, wsed, is the weight of sediments, in t, bpl, is the plating buoyancy, in t. published by atlantis press copyright: the authors 163 as a result, the detachment process will take more time, which can be estimated by the following empirical equation: 4 2 ( ) ( )t t d b p   . (5) where: t, is the breakout time, in minutes, t, is the breakout parameter depending on the ratio fap/fib (figure 1), d, is the equivalent embedment depth of the wreck, in feet, b, is the equivalent breadth of the wreck, in feet, p, is the average breakout pressure applied to the seabed, in psf, calculated by the ratio fap/a. 4 shipwreck removal methods there are many methods used to remove a wreck, but the most commonly used are restoring buoyancy and lifting with mechanical means. the other available wreck removal methods are quite specialized and are only used on cases, where special circumstances mandate consideration of non-conventional solutions. due to the fact that every removal operation offers unique challenges to salvors, usually the best solution is to combine various methods in order to exploit the advantages of each of one. the selection of a wreck removal method for a particular case involves taking into account a multitude of parameters. the most important are environmental parameters, such as prevailing weather conditions and the type of the seabed, and the weight of the shipwreck, which will determine the required lifting capacity. 4.1 restoring buoyancy most or even all compartments of a sunken ship may have been flooded with water. this means that the wreck displaces less water resulting in a significant loss of buoyancy. the basic concept behind this method is to restore the lost buoyancy of the sunken ship in order to be able to float again back on the surface. to achieve this, the flooded compartments of the ship have to be sealed and the trapped water removed, these are the two stages of the method of the restoring of buoyancy. sealing cracks on the hull can be achieved by using various types of mainly steel patches or cofferdams. however, depending on the hull type and material, different types of patches can be used. some of the alternative materials for patches are: concrete, composite materials, and wood.13 after sealing the compartments, the next stage is to remove the trapped water. there are two ways of dewatering: (i) pumping and (ii) displacing the trapped water with compressed air. both methods require the use of pumps, although the type for each one of these ways differs. the use of compressed air requires better patching and is considered more time-consuming comparing to pumping out the water. moreover, the air pressure inside the compartment needs to be monitored constantly to avoid either an explosion, which will occur when the internal air pressure exceeds the external hydrostatic pressure, or an implosion, which occurs when the opposite happens. the air compression dewatering method is suitable for cargo oil tanks of tanker ships, as well as fuel and water ballast tanks, because these kinds of tanks are easier to seal. a variation of the buoyancy restoration method is induced buoyancy, which is mainly used for small ships (e.g. yachts etc.). this method uses equipment, such as balloons, pontoons and sometimes special foams14 for displacing trapped water. one of the main disadvantages of this method is the lack of control of the movement of the shipwreck while rising towards the surface. additionally, the amount of buoyancy that is restored is fig. 1: breakout time parameter as a function of relative force.9 published by atlantis press copyright: the authors 164 much smaller compared to conventional buoyancy restoring methods. 4.2 lifting by mechanical means the second commonly used method for wreck removal operations is lifting by mechanical means. in general, the employed mechanical means are derricks, sheer legs and various lifting mechanisms, which are fitted on barges. steel wires and chains are mainly used for attaching the lifting mechanism to the shipwreck. the restriction of this particular method is the lifting capacity of the employed equipment. a common occurrence is the inadequacy of the available lifting capacity to lift the shipwreck in one piece, which then results in the decision to cut the shipwreck into a suitable number of pieces. chains or special steel wires, with high levels of hardness similar to that of diamonds, are used for the cutting process. an essential part of such an operation is to determine the length of each piece into which the wreck will be cut. for example, cutting an engine room in the middle is not preferred because objects positioned therein such as the main engine and the propeller shaft are going to make the cutting process extremely difficult and slow, resulting in delaying the overall operation. alternatively, a quicker but also more dangerous way of cutting a wreck into pieces is by using explosives. this option was used for cutting the wreck of the container ship msc napoli into two pieces.15 the advantages of lifting with mechanical means are: briefer preparation time because there is no need for sealing various compartments and the improved controllability of the shipwreck while lifting it towards the surface. moreover, the depth of the shipwreck is not a significant restriction for the application of this method. 4.3 other ways of removal there are also other ways to remove a wreck but they are not commonly used mainly because of their ineffectiveness and the damages that can be brought on the marine environment. two methods, that raise environmental concerns, are burying or exploding a wreck, which are therefore used only in times of war because of their immediacy and speed. finally, a different method is the in situ scrapping of the wreck, which has been used in the removal of the stern section of the bulk carrier new carissa that ran aground on a beach near coos bay, oregon, united states.16 4.4 comparison of common methods restoring buoyancy is less expensive but a more time consuming method compared to lifting a shipwreck with mechanical means; this is due to the fact that simpler equipment is used. however, the main disadvantage of restoring buoyancy is the restriction on the depth of the shipwreck because of the high values of hydrostatic pressure. for this reason, it is only used in low depths, i.e. up to about 40 m. another concern is the pressure inside the sealed compartments, which needs to be constantly monitored as it changes with the changes in depth during the lifting process, in order to avoid explosions or implosions. however, restoring buoyancy is a method that causes minimal damage to the hull, which is the reason that this method is also used for salvaging a sunken ship. an example of salvaging by restoring buoyancy is the case of the dry-dock fdn1, which sank off port blair in the andaman islands.17 on the other hand, while lifting a shipwreck with mechanical means requires no sealing, which simplifies the whole operation, it is essential to conduct a detailed study for rigging the wreck. additional advantages of this method are improved controllability of the shipwreck, the lack of a need for monitoring the pressure inside the hull and higher lifting velocities, which can result in reduced durations for these operations. however, the high operating and renting cost as well as the availability of lifting equipment such as derricks and/or sheer legs are major concerns when lifting a shipwreck by mechanical means. 5 an illustrative case study the best way to clarify the information presented before in this paper is to examine an illustrative case study. the analysis will present the stages of a wreck removal operation and highlight the most important factors in a hands-on way of thinking. 5.1 defining the shipwreck the case study involves a passenger ship sinking near the shore due to contact with reefs which resulted cracks published by atlantis press copyright: the authors 165 in her hull (assuming that there are no human casualties from this marine accident). another assumption is that the wreck lies 135 m from the surface of the sea and the embedment depth in the seabed is 2 m equally across the length of the sunken vessel. the location of the shipwreck in relation to the local geomorphology and its distance from the shore is shown in figure 2. the seabed in this location is highly inclined, which can render the operation extremely dangerous, because in case of any miscalculation the ship will sink deeper dragging the lifting equipment with her (in case this type of salvaging effort is selected). additionally, the wreck has been lying on the seabed for a long time, which means that there will be a significant accumulation of sediments on the hull. the particulars of the shipwreck are shown in table 1. it is also assumed that the seabed around the shipwreck is viscous and consists of finely grained sand. finally, at the time of the accident the ship had a displacement equal to 12,076 t. table 1: the main particulars of the shipwreck. data value dimensions length overall loa 142.95 m breadth mld. b 24.70 m weight groups displacement δ 12,076 t deadweight dwt 2,139 t lightship ls 9,938 t volume gross tonnage grt 21152 gt occupants passengers 1,537 crew 300 an important part of the planning stage of a realworld wreck removal operation would be cross checking the initial data by conducting an on-site survey. this is necessary because initial data coming from reports are either not reliable enough or might have changed due to the action of environmental factors. for example, the way the wreck lies on the seabed might have been altered by the action of waves and currents. 5.2 calculating the ship weight determining the weight of the shipwreck is one of the most critical parts of the wreck removal operation, because it will help to decide on the possible methods to be used. for this, it is essential to know the exact displacement of the ship at the time of the accident. depending on the type of the ship there are certain weights that can be easily removed to decrease the weight of the shipwreck, which can sometimes prove to be cost beneficial, even though the associated initial cost might seem steep. this is due to the fact that lightering the shipwreck, prior to the removal operation, results e.g. in the use of lower capacity lifting equipment, which minimizes the total cost of the operation. the weights to be removed in the context of this case study are the following:  fuels,  occupants and lifesaving equipment,  deck outfitting. additionally, the buoyancy of the hull plating, when the ship is in the water, needs to be taken into account in determining the required lifting capacity. in particular, the buoyancy force is subtracted from the weight of the shipwreck. on the other hand, a sunken shipwreck has some added weights, which increase significantly the weight of the wreck and they should also be taken into account. these are the following:  seabed sediments,  water trapped in certain compartments of the hull. finally, when the wreck lies partially embedded in a viscous seabed, the force required to break out the hull, which equals the suction by the seabed, needs to be fig. 2: the location of the illustrative wreck. published by atlantis press copyright: the authors 166 factored into the estimation of the total force required to bring the shipwreck to the surface of the sea. 5.2.1 weight of fuels the first step of the operation is to locate and collect the remaining fuels so as to avert any possible (additional) pollution while removing the wreck. table 2: analysis of fuel weights. volume weight (m3) (t) h.f.o. 180 460.11 437.10 diesel oil 76.00 64.60 lube oil 36.00 32.40 total 572.11 534.10 the fuel weight at the time of the accident is documented in the report of the captain and is shown in table 2. 5.2.2 weight of occupants / lifesaving equipment due to the assumption that all the passengers and crew were evacuated, the total weight of the wreck will be decreased by the respective weights. moreover, the weight of the equipment used for the evacuation procedure should be subtracted as well. taking into account the report of the captain and the particulars of the ship, these weights are equal to:  135.00 t for occupants,  90.50 t for the used lifesaving equipment. therefore, the total weight to be subtracted from the weight of the wreck equals 225.50 t. 5.2.3 weight of deck outfitting deck outfittings include anchors and chains, which can be easily removed from the wreck and reduce the total weight of the shipwreck. from the particulars of the ship, deck outfittings are estimated to be 26.70 t in weight. 5.2.4 plating buoyancy as described earlier in this paper, the buoyancy of the hull plating will be estimated by calculating the displacement of the lightweight of the wreck. for this calculation, the average specific gravity of the material will be the specific gravity of steel, which is 7.689 t/m3, reduced to account for the lighter materials typically used in a cruise ship, such as aluminum for the superstructures. therefore the average specific gravity, in the context of this case study, was considered equal to 6 t/m3. subsequently, the volume of the water displaced by the lightweight of the wreck equals 1,656 m3. moreover, the specific gravity of the seawater at the location of the wreck is 1.028t/m3, taking into account information from an on-site survey substituting these values into eq. (1) the buoyancy of the hull plating is calculated: a = 16,708 kn = 1,703 t. 5.2.5 seabed sediments as also described previously in this paper, while a shipwreck rests on the seabed various microorganisms and sediments accumulate on the hull (and within its interior), significantly increasing the total weight of the wreck. it is difficult to analytically calculate the specific gravity of the sediments, without examining samples taken from the actual site. however, the specific gravity of the sediments can be approximately taken equal to 1,600 kg/m3.8 for the particular case study, the uncertainty concerning the extent of the coverage of the sediments is taken into account by calculating four different scenarios. each scenario assumes that the sediments take up a specific percentage of the ship’s volume. therefore, the scenarios are differentiated in the weight calculated for the sediments. the scenarios range from 5% to 20% sediment extent of the ship’s volume. percentages above 20% were considered as extremely overestimated and therefore unrealistic. the results for the weight of sediments for each scenario are shown in table 3. table 3: scenarios for the weight of sediments. percentage of ship’s volume 5% 10% 15% 20% volume (m3) 2,538.6 5,077.1 7,615.7 10,154.2 weight (t) 4,061.7 8,123.4 12,185.0 16,246.7 5.2.6 flooding water the weight of the flooding water will be calculated considering the volume of the wreck, which remains flooded after emerging from the surface of the sea. this published by atlantis press copyright: the authors 167 weight is added to the total weight of the shipwreck only when the wreck is out of the water, because at that stage it is no longer counteracted by the buoyancy. the exact calculation of the flooded volume of the ship is extremely difficult. therefore, five different scenarios have been developed to take into account the uncertainty surrounding the value of the flooded volume table 4 shows the results of the calculations of the weight of flooded water for the different scenarios. table 4: scenarios for the weight of flooded water. percentage of ship’s volume flooded 10% 20% 30% 40% 50% volume (m3) 5,077 10,154 15,231 20,308 25,386 weight (t) 5,221 10,442 15,662 20,883 26,104 5.2.7 suction of the seabed one of the initial assumptions of this case study is that the seabed in the vicinity of the illustrative wreck is viscous with finely grained sand and that the embedment depth of the ship is 2 m. in a real-world operation this data would have been collected during a thorough on-site shipwreck survey. consequently, the soil bearing strength (qu) of the seabed is calculated at 287.28 kpa.9 the next step is to calculate the surface of the ship in contact with the seabed (a), which is equal to the wetted surface area of the ship for a draught of 2 m. using the information contained in the stability booklet, the wetted surface area of the ship is found to be 1,145.80 m2. by substituting the soil bearing strength of the seabed and the wetted surface area of the ship in eq. (2) the bearing capacity of the seabed can be calculated. after estimating the soil bearing capacity, the force required for immediate breakout of the wreck from the seabed is calculated. for this purpose we need to calculate the equivalent breadth (b) and the embedment depth (d) of the wreck. for calculating the equivalent breadth (b), it is assumed that the area of the ship in contact with the seabed (a) is a rectangular shape with the same area and length equal to the length of the ship, namely l = 142.95 m. the equivalent breadth (b) is therefore calculated as a function of the area (a) and the length (l) and it equals b = 8.02 m. the embedment depth (d) is calculated as a function of the displaced volume of the seabed soil (v), which is equal to the volume of water the ship displaces at a draught equal to d. using the information contained in the stability booklet of the ship, the displacement volume for a 2 m draught is 3,871.65 m3. the embedment depth is therefore calculated as a function of the volume (v) and the area (a) and it equals d = 3.38 m. substituting the above in eq. (3) the immediate breakout force is fib = 229,001.9 kn or 23,343.70 t. 5.3 wreck removal approaches for the particular illustrative case study, the method of restoring buoyancy is not (practically) feasible due to the large depth (135 m) of the shipwreck and the size of the sunken ship. therefore, removing the wreck by mechanical means is deemed as the appropriate method to be employed. two approaches will be examined: (i) in which the wreck is removed in one piece and (ii) in which the wreck is cut into several pieces and then the pieces are removed separately. there are three key stages in the removal operation, namely: 1. breaking the wreck out of the seabed. 2. the time period it takes to lift the wreck from the seabed up to the surface of the sea. 3. when the wreck is completely out of the sea. in each of these stages the weight of the wreck is different due to the different forces that are applied. this fact affects the selection of the lifting equipment to be used in the operation. in particular, the required lifting capacity must be determined from the worst stage, in terms of wreck weight, which is when the wreck has emerged from the surface of the sea and the buoyancy has no effect. this point will be illustrated below. 5.3.1 lifting the wreck in one-piece initially, the required force must be calculated for each of the key stages of the wreck removal operation. table 5 shows the results from the respective calculations. the weight of the sediments has been calculated as extending in 20% of the ship’s volume (see table 3), published by atlantis press copyright: the authors 168 while the weight of the flooding water has been calculated as 50% of the ship’s volume (see table 4). for both of these parameters, the calculations presented here are the worst of the previously considered scenarios. however, the same calculations have been conducted for each of the scenarios. table 5: the weight of the shipwreck for each stage of the operation. force (t) stage 1 stage 2 stage 3 wreck 11,077.3 11,077.3 11,077.3 sediments 16,246.7 16,246.7 16,246.7 plating buoyancy -1,703.2 -1,703.2 required for immediate breakout 23,343.8 flooding water 26,103.6 total 48,964.6 25,620.8 53,427.6 the next step is the selection of the derricks or sheer legs, based on the required lifting capacity. for this approach it was concluded that four derricks or sheer legs of the highest lifting capacity would be needed. the cost of the operation is outside the scope of this paper and has not been considered as a parameter for the selection of the lifting equipment. the capacities and dimension particulars for each one are shown in table 6. it is noted that the total available lifting capacity (lc) is not enough to immediately break out the wreck from the seabed. this does not mean that the selected lifting equipment is not adequate for this operation, but rather that the completion of stage 1 will take more time. the respective calculations are presented below. table 6: the main particulars of the derricks used in the one-piece approach. lifting cap. length breadth draught (t) (m) (m) (m) thialf 14,200 201.6 88.4 11.8 – 31.6 sapiem 14,000 198.0 87.0 10.5 – 27.5 svanen 8,700 102.6 71.8 dcv hermond 8,165 154.0 86.0 11.5 – 28.2 total 45,065 below are the calculations for the time needed to break out the wreck from the seabed, as described in a previous section of this paper. the applied breakout force is calculated by eq. (4), substituting the available lifting capacity (table 6) and the respective weights (table 5) fap = 8,177.5 t = 80,225.7 kn = 18,035,537 p. the average breakout pressure applied to the sediment is calculated by taking into account the applied breakout force and the contact area of the wreck to the seabed (a). therefore, the result is p = 1,452.5 psf.  the breakout time parameter is t = 4.17·106,  the equivalent embedment depth is d = 3.38 m = 11.09 ft,  the equivalent breadth of the wreck is b = 8.02 m = 26.30 ft. inserting the previously mentioned numbers into eq. (5) the breakout time in minutes is tb = 62,781 min, which translates to about 1,046 hrs or 44 days. the calculated breakout time will increase the total time required, which in turn would increase the associated cost for the operation. additionally, longer operation duration might also pose problems connected to the availability of the required equipment. breakout time may be reduced by using equipment with higher lifting capacity. for the particular case however, the location of the wreck, as seen in figure 3 (set up of the employed equipment), does not allow for the use of equipment with larger dimensions. also, using larger equipment with higher operating costs may render the operation not feasible from a cost-benefit point of view. for stage 3 of the operation the required lifting capacity is much higher (58%) than the available, including the 25% safety factor. this indicates that while the wreck emerges from the sea surface, an appropriate amount of flooding water should be pumped out of the wreck. the problem with this approach lies in the synchronization of the derricks or the sheer legs in order to stabilize the wreck during the pumping operations. additional challenges with regards to the lifting of the shipwreck in one-piece are the following:  an extremely large barge is needed for the transportation of the wreck after the removal operation, published by atlantis press copyright: the authors 169  the wreck in one-piece will have a large wind profile area. consequently the prevailing wind conditions might greatly influence the stability of both the wreck and the derricks, which is a major risk factor to the operation in general,  the use of four or five high capacity derricks or sheer legs simultaneously will add greatly to the complexity and the cost of the setting up of the operation, which will be largely dependent on the equipment availability (this might cause an insuperable problem to the overall effort). for the reasons mentioned above, lifting the wreck in one-piece is considered as (practicably) a not feasible option and therefore, cutting the sunken vessel in multiple pieces will be further studied hereafter in this paper. 5.3.2 lifting the wreck in multiple pieces when considering the lifting of the sunken ship in multiple pieces, the wreck will be cut into several parts as it lies on the seabed, which can also be translated into the use of fewer and of lower capacity lifting equipment. the benefits of this approach include cutting down on operational costs, increasing flexibility with regards to the topological positioning in the vicinity above the shipwreck, and the overall simplification of the planning stage of the operation. the downside is the increase in the total time needed for the operation due to the cutting process. the details of this approach as applied to the particular illustrative case study are described below. hence, it comes out that one high lifting capacity sheer leg will be used; in particular, the selected one has a lifting capacity of 4,536 t. for the cutting process, the use of a steel wire rope is promoted, which is preferable against the cut with a steel chain, this way, the hull pieces are more cleanly cut and with a higher precision. furthermore, steel chains will be employed for securing and lifting each piece up to the surface. it is important that the steel chains are positioned on the appropriate longitudinal points of the hull, which are the locations of the web frames. additionally, compared to lifting the wreck in one piece, a significantly smaller barge for transporting the individual wreck pieces will be required. this is due to the fact that once each piece has been lifted it will be immediately transported for scrapping. the same procedure was followed in the removal operation of the car carrier tricolor18, which sank after colliding with the containership cariba (2002) in the english channel. because of the inherent uncertainty in determining the weight of the sediments and the flooding water in the wreck’s hull, as described in previous sections of this paper, four case scenarios will be examined (figure 4). scenario 1 includes the same weights as for the lifting of the wreck in one-piece and will be used as a reference scenario for the evaluation of the other developed scenarios. moreover, this scenario is considered to be the worst case because it overestimates the flooding water and weight of sediments, as it does not account for the outflow of water and reduction in sediments, which will occur during the lifting process of each hull piece. after defining the total weight of the wreck, the next step is to answer the question: in how many pieces should the wreck be cut? the maximum weight of each piece will be determined by the available lifting capacity, considering a 25% safety factor8. the longitudinal load diagram of the wreck will determine the maximum length of each piece when it will be above the surface of the sea, which is the worst stage of the removal operation in terms of the forces applied on the hull parts. this method is differentiated from the one considering the lifting of wreck in one piece with respect to the necessary stages prior to the lifting fig. 3: the location of the wreck and the setup of the four derricks and the barge. published by atlantis press copyright: the authors 170 process. the first stage is the cutting process, which is assumed to last about 120 h for each piece. the same amount of time was used for the cutting of the engine room of the car carrier tricolor18. this is followed by gradually applying the lifting load on the steel chains in order to avoid multiple material failures during the lifting process. typically the time needed to apply the full lifting load can be assumed to be about 1h for each piece, which is considered an adequate amount of time. the next stage is breaking out the pieces from the seabed. prior to commencing the lifting process and after the breakout, each piece needs to be secured and the steel chains surveyed. for this case study it was assumed that the securing stage will last about 1h for each piece, which is again considered an adequate amount of time. after the lifting arrangement has been stabilized, the lifting process towards the sea surface may begin. this approach is complemented by a preliminary structural strength analysis for each piece, taking into account the exact positions of the steel chains and the respective weight distribution. the accurate estimation of the weight distributions for sediments, buoyancy of plating, the breakout force and flooding water is extremely difficult. however, they can be approximated by trapezoidal distributions, based on the length of the parallel midbody of the shipwreck. the load diagrams for each stage of the removal operation are calculated as the sum of the individual weight distributions, as seen in figure 5 for the reference scenario. based on the available lifting capacity and the load diagrams, it was concluded that the wreck would be cut into at least 12 pieces. if the wreck was cut into less than 12 pieces, then the available lifting capacity would not be sufficient to lift each piece. fig. 5: scenario 1 distribution of time for each stage of the operation without the 30% increasing factor. for the estimation of the total required time of the removal operation it was assumed that the lifting speed equals 3 m/h, which can be considered as a typical one and was the lifting speed of the stern section of the containership msc napoli12. therefore lifting each piece will take about 750 min or 12.5 h, considering that the wreck lies in 135 m depth. the duration in days for fig. 4: weight scenarios differentiating sediments and flooded water. published by atlantis press copyright: the authors 171 each stage of the reference scenario is shown in figure 6. it is noteworthy that the total breakout time equals to 7.5% of the required time in the choice of lifting the wreck in one piece. this difference is due to the fact that the required breakout force is much smaller for each individual piece than for the whole ship. fig. 6: weight distributions of shipwreck above and below the sea surface. the vertical lines mark the cutting points of the wreck pieces for the reference scenario. the estimation of the total time required for the removal operation does not include an analytic calculation for the following:  preparation and setting-up,  deployment of steel cutting wires and lifting chains,  draining sea water from each piece while emerging from the sea surface,  positioning each piece on the barge. the required time periods for the above are not fixed and are affected by many contributing factors. therefore, for this approach, an increasing factor of 30% will be applied to the total calculated time for the operation. figure 7 depicts the total required time in relation to the number of wreck pieces for each implemented scenario. the operation in the context of scenario 1, where the wreck is cut into 12 pieces, would take about 92 days and is used (only) as reference. the minimum number as well as the length of the pieces for each scenario has been based on the respective load diagrams so as the weight of each piece does not exceed the available lifting capacity. for numbers of pieces greater than the minimum for each scenario, they have been assumed to be of equal length. for example, for scenario 4 the minimum number of pieces is 7 (figure 7) and thereon it has been assumed that the wreck is cut into 8, 9, 10, 11 and 12 pieces of equal length, considering scenarios 2, 3 and 4, it appears that the optimum number of pieces is 10. at this point, the number of pieces in scenario 2 has been determined from the load diagram and the length is not the same for each one, while in scenarios 3 and 4 the pieces are of equal length. this amounts to the operation taking longer in scenario 3, because the available lifting capacity is not being efficiently used, as some pieces may be much lighter than others. fig. 7: total time for each scenario as a function of the number of wreck pieces, including the 30% increasing factor. 5.4 discussion a defining factor for the feasibility of wreck removal operations is the total cost. in the preliminary planning stage it is difficult to estimate the cost due to the lack of publicly available information, both for the required equipment as well as for the services rendered by the involved parties. however, it is important to published by atlantis press copyright: the authors 172 complement the setting up of a wreck removal operation with a detailed cost-benefit analysis. the results presented in this paper are in accordance with real-world operations such as the removal of the car carrier tricolor, which was cut into 9 pieces18. the methodology of this paper suggests cutting the wreck of the case study into 10 pieces, which is the optimum number derived from the scenario based analysis. the difference between the two cases is due to the fact that the tricolor was removed almost immediately after sinking and consequently there had been a very small amount of sedimentation on the hull. on the other hand, the wreck of the case study is assumed to have been underwater for a long time, which means that the accumulated sedimentation has increased its weight considerably. furthermore, the methodology outlined in this paper is important in order to evaluate a number of alternative wreck removal methods and select the most feasible one by taking into account the following parameters:  environment of the wreck removal operation,  weight of the shipwreck, and  estimated total time. the methodology of this paper is a scenario based approach in order to incorporate the uncertainty element of the analysis. the underlying uncertainties during the planning phase of a wreck removal operation lie mainly in the calculation of the weight of the shipwreck, which will determine the required equipment. in particular, this paper developed scenarios for the weight of the sediments accumulated on the hull as well as the flooding water inside the wreck, the scenario based approach is necessary due to the difficulty in collecting reliable data from reports and even on-site surveys. using the illustrative case study, this paper outlines the key stages of the planning phase of a wreck removal operation. 6 conclusions a wreck removal operation is a complex subject that can be affected by multiple factors. one of the main defining factors is the prevailing weather conditions on the location of the shipwreck, which increases the uncertainty as it can delay, halt or even postpone the entire operation. the planning is also greatly affected by the availability of the needed equipment and by the running costs; these factors determine the employed method and the total required time, simultaneously. the mostly used wreck removal method is lifting by mechanical means. the advantages thereof include less preparation time, greater flexibility for the operation, which leaves room for technical innovation, and improved controllability of the wreck during the lifting stage. additionally, a wreck removal operation starts and ends with a thorough and detailed survey of the shipwreck in order to gather as much information as is practicably feasible to better define the problem and find the best solutions. a wreck removal operation is a crisis with challenging solutions because it cannot be standardized into a rigid frame. this is due to the fact that each shipwreck has unique characteristics and must be studied individually and with extreme caution, so as to obtain the best possible results in terms of efficiency, safety and environmental friendliness. references 1. g. g. mcgrath et al., the investigation to identify the ss jacob luckenbach – using technology to locate a hidden source of oil that cause years of impacts and the future implications of sunken shipwrecks, (international oil spill conference, 2003) 2. j. wilkins, u.s. navy salvage report uss mississinewa oil removal operations, (naval sea systems command, 2004) 3. j. michel et al., potentially polluting wrecks in marine waters, (international oil spill conference, 2005) 4. m. j. barrett, spatial tools and analysis of wwii shipwrecks, (master’s project, nicholas school of the environment, duke university, 2011) 5. i. hassellöv, pre-study of shipwreck assessment and remediation, (the alliance for global sustainability, göteborg, 2007) 6. n. p. ventikos, k. louzis and a. koimtzoglou, the shipwrecks in greece are going fuzzy: a study for the potential of oil pollution from shipwrecks in greek waters, (human and ecological risk assessment: an international journal, 2013) 7. h. landquist et al., evaluating the needs of risk assessment methods of potentially polluting shipwrecks, (journal of environmental management, 2013) 8. w. i. milwee, modern marine salvage, (cornell maritime press, maryland, 1996). 9. c. a. bartholomew and m. bert and r. hooper, (u.s. navy salvage engineer’s handbook 1st revision, (u.s navy sea systems command, 2008) 10. smit, the salvage of the kursk, (technical report) published by atlantis press copyright: the authors 173 11. v. streeter, e. wylie and k. w. bedford, fluid mechanics, 9th edn. (mcgraw – hill, 1998) 12. p. glerum, removal of napoli stern section, (the 12th annual salvage & wreck removal conference, london, 2009) 13. us navy, salvor’s handbook, (naval sea systems command, 2004) 14. a. winer and w. f. searle, plastic foams for marine salvage, (salvage engineer’s journal, august 1970), pp. 97-110 15. j. porter, removal men, (lloyd’s list australia – the daily commercial news, sep. 10, 2009) 16. d. parrott, removal of the new carissa stern section, (the 12th annual salvage & wreck removal conference, london, 2009) 17. wisjmuller salvage bv, mission: (re-)floating drydock, (svitzer-wisjmuller, dvd) 18. j. c. burvingt, the tricolor experience in the framework of the contingency plan mancheplan, (cherbourg armees: prefecture maritime de la manche et de la mer du nord) published by atlantis press copyright: the authors 174 the response characteristics of xilingol grassland to uneven distribution of_edit.pdf received 26 september 2016 accepted 11 october 2016 the response characteristics of xilingol grassland to uneven distribution of precipitation at temporal and spatial scale qiaofeng zhang1,2,3,4, hongbo yu1,3, guixiang liu2, yuhai bao1,3 1college of geographical science, inner mongolia normal university, hohhot 010022, china 2institute of grassland research of chinese academy of agricultural sciences, hohhot 010022, china 3inner mongolia key laboratory of remote sensing and geography information system, hohhot 010022, china 4inner mongolia laboratory of disaster prevention and reduction and ecological safety monitoring, hohhot 010022, china abstract precipitation is the most important source of water supply, which has important significance on the vegetation growth, also is an important factor affecting the drought severity in xilingol grassland. based on the precipitation of 15 weather stations and modis mod13a3 ndvi data, pearson correlation analysis was used to analysis the response characteristics of ndvi to precipitation at spatial and temporal scale. results showed that from one month view, vegetation growth was the highest correlation with the precipitation of this month from may to july, but august and september were the highest correlation with the precipitation in july. accumulated month view, the accumulated precipitation in the last three months was very important to the vegetation growth, and snowfall in the winter of previous year played an important role in may of next year. spatially, from one month view, erenhot and sonid youqi desert steppe region were the highest correlation with the precipitation in two months before, the others were the highest correlation with the precipitation in one month before. accumulated month view, erenhot and sonid youqi were the highest correlation with the accumulated precipitation in the last four months, the others were the highest correlation with the accumulated precipitation in the last three months. keywords: precipitation, ndvi, spatial and temporal distribution, correlation. 1,2,3,4, 1,3, 2, 1,3 1. 010022 2. 010010 3. 010022 4. 010022 : 15 modis mod13a3 ndvi pearson ndvi 5~7 ndvi 8 9 ndvi 7 3 5 2 1 4 3 ndvi journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 206-212 published by atlantis press copyright: the authors 206 q.f. zhang / the response characteristics of xilingol grassland 1. [1-3] [4-9] [7, 10, 11] [10, 12-14] [15] 20 50 [16, 17] [18, 19] ndvi ndvi ndvi 0.83 [20] 15 5~9 ndvi 10~12 5~9 5~9 ndvi 2. 2.1 111°09´ ~ 120°01´e 41°35´ ~ 46°46´n 200 ~ 350mm 150mm 400mm 6~8 70% 0 ~ 3 3.5 ~ 4m/s 24 ~ 28m/s 34m/s 2800h 3000h 1 1. fig.1 vegetation zone and weather station in the study area 2.2 2.2.1 15 1 1999~2014 nasa modis mod13a3 2000 ~ 2014 1km published by atlantis press copyright: the authors 207 q.f. zhang / the response characteristics of xilingol grassland 15 2000~2014 5~9 3×3 5~9 225 15×15×1 ndvi 15 1999~2014 ndvi 5~9 ndvi 10~12 15 2000~2014 5~9 3×3 75 1 ×15 ×5 1125 15 ×15 ×5 ndvi 15 1999~2014 ndvi 5~9 ndvi 2.2.2 pearson pearson r 1 2 2 x x y y r x x y y 1 1 r [-1 1] r<0 r=0 r>0 0<|r|<0.4 0.4 0.7r 0.7 1r ndvi pearson 3. 3.1 1 5 ndvi 5 4 10 0.45 0.42 0.42 5 ndvi 3 2 1 12 11 6 ndvi 6 4 10 0.47 0.45 0.43 7 ndvi 7 6 0.50 0.49 7 ndvi 3 2 1 12 11 8 ndvi 7 6 0.50 0.44 8 ndvi 7 9 ndvi 8 7 0.42 0.44 9 ndvi 7 published by atlantis press copyright: the authors 208 q.f. zhang / the response characteristics of xilingol grassland 1. ndvi 5~9 pearson table1 correlation coefficient between ndvi from may to september and precipitation of each month pearson ndvi 5 6 7 8 9 9 0.36578 8 0.3649 0.42381 7 0.49606 0.49619 0.43705 6 0.46948 0.48685 0.4419 0.39962 5 0.45178 0.36983 0.30096 0.24274 0.23455 4 0.41644 0.45427 0.38186 0.35761 0.35163 3 0.19392 0.23587 0.21549 0.26482 0.24823 2 0.24098 0.21147 0.22335 0.22501 0.19188 1 0.15233 0.14655 0.16068 0.14788 0.14105 12 0.21159 0.26173 0.28769 0.22474 0.19806 11 0.09282(p=0.1653) 0.19627 0.15535 0.15441 0.1512 10 0.42484 0.42556 0.38521 0.35878 0.34042 p p < 0.05 2 3 6~9 0.66 0.64 0.64 0.62 3 5 10 0.65 2. ndvi fig.2 pearson correlation coefficient between monthly ndvi and cumulative precipitation 3 0.58 5 5 6 7 8 9 7 7 7 8 9 4~10 3 2 1 12 11 4~10 11~3 6~9 3 published by atlantis press copyright: the authors 209 q.f. zhang / the response characteristics of xilingol grassland 3 5 10 3 5 5~9 4~7 5 3.2 5~9 ndvi 3 1~2 5 0 2 1 3. ndvi fig.3 correlation coefficient between mean ndvi of growing season and monthly precipitation 5~9 ndvi 4 ndvi 0 4 3 8 8 2 1 4 3 published by atlantis press copyright: the authors 210 q.f. zhang / the response characteristics of xilingol grassland 4. fig.4 pearson correlation coefficient between ndvi of growing seasonand cumulative precipitation 4. 1 5~7 8~9 7 3 5 2 2 1 4 3 2013bak05b01 ” caas-astip-igr2015-04 email liugx804@163.com. [1] . 1982~1999 . . 2007, 62(1): 41-51. [2] . 1982-1999 . . 2003, 58(1): 119-125. [3] chen x, hu b. spatial and temporal variation of phenological growing season and climate change impacts in temperate eastern china. global change biology. 2005, 11(7): 1118-1130. [4] . 1961—2009 . . 2011(01): 22-24. [5] . 1961—2010 . . 2012(03): 7-9. [6] . 1961-2010 . . 2015(03): 135-136. [7] . 50 . ( ). 2011(03): 157-160. [8] . . . 2011(10): 126-130. [9] . . . 2011, 31(24): 7511-7515. [10] . . . 2011(05): 789-794. published by atlantis press copyright: the authors 211 q.f. zhang / the response characteristics of xilingol grassland [11] . . . 2012(06): 24-27. [12] . . . 2010(05): 46-49. [13] . . ( ). 2011(03): 304-310. [14] . . . 2010(23): 6538-6545. [15] . . . 2015(05): 119-128. [16] . 10a . . 2014(05): 1-9. [17] . . . 2006(02): 268-277. [18] . . , 2009. [19] . . . 2011(05): 789-794. [20] . 2000-2010 . . 2014(06): 1194-1204. published by atlantis press copyright: the authors 212 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjdffile false /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /description << /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /dan <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> /deu <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> /esp <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> /fra <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /ptb <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> /suo <feff004b00e40079007400e40020006e00e40069007400e4002000610073006500740075006b007300690061002c0020006b0075006e0020006c0075006f00740020006c00e400680069006e006e00e4002000760061006100740069007600610061006e0020007000610069006e006100740075006b00730065006e002000760061006c006d0069007300740065006c00750074007900f6006800f6006e00200073006f00700069007600690061002000410064006f0062006500200050004400460020002d0064006f006b0075006d0065006e007400740065006a0061002e0020004c0075006f0064007500740020005000440046002d0064006f006b0075006d0065006e00740069007400200076006f0069006400610061006e0020006100760061007400610020004100630072006f0062006100740069006c006c00610020006a0061002000410064006f00620065002000520065006100640065007200200035002e0030003a006c006c00610020006a006100200075007500640065006d006d0069006c006c0061002e> /sve <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word re-non-diversifiable risk in investment portfolios an aid to investment decision making received 1 january 2015 accepted 5 february 2015 non-diversifiable risk in investment portfolios --an aid to investment decision making emma anyika department of accounting and finance, mount kenya p.o. box 342-01000 thika, kenya e-mail: mmnk55378@gmail.com abstract modeling non diversifiable risk in investment portfolios is undertaken in this paper together with redefinition of estimators of diversifiable risk and portfolio expected returns to reflect normal market conditions. garch (general auto regressive conditional heteroskedasticity) models are then used to make forecasts of given time series, from which future predictions of non diversifiable risk, diversifiable risk and portfolio expected returns are made. the required investment decisions are then made. in making investment decisions several factors are considered. these include profits, dividend yield, price earning ratios, and expected future performance of financial institutions. this paper has considered expected future performance of financial institutions. in particular the paper derives a method of determining non diversifiable risk in investment portfolios that enables investors and investment managers make viable investment decisions. this study is expected to improve the accuracy of predicting future expected performance of financial institutions. investment analysts can now rely on the predictions to make good investment decisions. non-diversifiable risk, diversifiable risk, garch, portfolio 1. introduction risk underlies most investment decisions. this is because it is not possible to anticipate the occurrence of possible future events with certainty and hence, making any correct predictions about the cash flow sequence is not possible. the simplest expression for risk in investment is the standard deviation of returns on investments. for single investments other risk determining expressions include coefficient of variation and the beta (b) factor method, wilkes (1996) and value at risk, vlaar (2000). when dealing with investment portfolios, return on investment is a weighted average of the expected returns of the individual investment. this alone will not allow one to make investment decisions since one needs to determine the best portfolio by looking at that which will give maximum returns at the lowest risk. this necessitates determination of portfolio risk, which is divided into non–diversifiable and diversifiable risk. diversifiable risk is that which the investor can eliminate if he held an efficient portfolio. the non– diversifiable risk on the other hand is that risk which still exists in all well diversified efficient portfolios. the investor therefore seeks to eliminate the diversifiable risk. nevertheless non–diversifiable risk permeates most areas of investment. these are particularly those characterized by unstable rates of returns due to the erratic nature of market forces. these include interest related portfolios and also exchange rates related ones. for new, short term, and long term players in investment markets there is need for proper determination of non diversifiable risk as a basis for investment decision making. this paper focuses on the problem of limited tools for estimation and prediction of non–diversifiable risk in investment portfolios that has lead to indecisiveness by investors and incorrect investment decisions. non–diversifiable risk is not normally determined since most estimators present are journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 31-46 published by atlantis press copyright: the authors 31 e. anyika. used to determine diversifiable risk which can on the onset eliminate some risk by combining several investment portfolios until the best one is realized i.e. one with the highest returns at the lowest risk. thus new and long–term investors who want to enter risky businesses, i.e. risk seekers, lack risk benchmarks on which to base their investment decisions. many investors congest short–term investment markets believing that they have high returns and are less risky. this eventually brings down the earnings from these investments. for example short–term interest earning government bills and bonds will have their interest earning rates decreased if we have many investors in the market. thus investors are left with only long term investments to consider. the situation could at least be made bearable if they knew precisely how much risk they are bound to encounter and how it will affect their rates of return. the aim of this research was to determine a method of estimating non–diversifiable risk to enable investment managers make viable investment decisions. this estimator is that which accounts for the normal market conditions. this will decrease losses made by financial organizations due to inability to predict non– diversifiable risk accurately and reduce drastically the large provisions set aside to hedge and manage risk. the financial environment is full of risk hedging techniques and risk management measures for systematic risk. these are mostly done on speculative terms jennifer (2003). there are no real concrete and long lasting solutions to the problems of non–diversifiable risk. most financial market players blame it on the volatility of financial markets, i.e. that one cannot determine and predict non–diversifiable risk since the forces determining it keep on fluctuating. the proper determination of non–diversifiable risk in this paper is expected to, prevent the making of wrong investment decisions by, for example, relying only on an investment managers belief or experience in risk management, boost the confidence of new, short term and long–term investors since they would know the level of risk to be encountered. it would also enable investors have a broad spectrum of portfolios to invest in since most of them just engage in short term investments where non–diversifiable risk can be speculated easily, facilitate development of suitable risk managing and hedging techniques to minimize non– diversifiable risk and set free colossal amounts of money from elaborate schemes hatched to hedge and manage portfolio risk thus enabling reinvestment to improve or increase the profitability of financial organizations. the uncertainty of non–diversifiable risk, i.e., the fact that it is believed not to be exactly known robert (1993), always makes investment markets so jittery such that any small force acting on these markets sends them tumbling. the proper determination of non– diversifiable risk will go a long way into stabilizing the investment markets. basic concepts applied include, diversifiable risk (also known as non –systematic or specific risk) which is the risk that can be eliminated by diversification. it is unique to the company or it’s industry and includes management competence and shifts in demand for company products, non–diversifiable risk (also known as systematic or market risk) which is related to fluctuation of the market as a whole and cannot be eliminated by diversification, portfolio expected returns a weighted average of the expected returns of each asset held in isolation and investment portfolio which is a combination of more than one investment with a belief that risk shared is risk less. 2. literature review originally portfolio expected return was the only criterion for making investment decisions. this has been lacking particularly in areas where the future of the expected returns was uncertain. thus other methods which incorporated determination of future uncertainties have to be sought. markowitz (1952) identified statistical measures of dispersion as measures of risk to be a way of indicating future uncertainties in investment decision making. later on he modified his risk estimator to include correlation coefficient. this was significant since for example if investments are perfectly negatively correlated i.e. their correlation coefficient is negative one, and then holding them in a portfolio greatly reduces their risk. this is synonymous to the idea of not carrying all your eggs in one basket, i.e. in different baskets risk is shared thus less. researchers then began to examine the implications of all investors using this approach. this led to the development of the capital asset pricing model (capm), sharpe, et al (1964), which published by atlantis press copyright: the authors 32 non-diversifiable risk in investment portfolios gave a simple relationship between expected returns and risk in a competitive market. it also provided further insight into the process of diversification. although there is a great deal of evidence which supports capm, there are major difficulties in testing the model. first it is stated in terms of investors’ expectations rather than historic returns. secondly the market portfolio (m) should include all risky investments, whereas most market indices contain only a sample of shares. furthermore research in the 1980’s mainly on u.s.a stocks revealed a number of stock market anomalies which are inconsistent with capm. these include “the small firm effect” and “calendar effects” dimson (1988). the model should therefore not be relied upon for more than general indications of the market pricing mechanism. currently, research in this area is centered on how to use the markowitz estimator of risk and the capm in investment decision making. less emphasis is being placed on further development of better risk estimators and models. stulz (1999) argues that total risk is often costly and discusses how taking total risk into account in capital budgeting is necessary to make capital budgeting and capital structure decisions consistent. harper (2003) discusses risk and returns in balanced portfolio and views the efficacy of cash and cash equivalents in investor portfolios. all the above researchers do not account for non–diversifiable risk as properly defined in their risk estimation thus rendering their estimators inaccurate. this paper seeks to remedy this by determining total risk which accounts for diversifiable and non–diversifiable risk according to its definition. 2.1. research methodology the population for this study was "turnover of shares in kenya shillings (ksh) on the nairobi stock exchange (nse) from the year 2009 to 2013”. nse is a kenyan capital market. currently there are fifty six financial institutions listed on the nse. every day the total turnover of shares in terms of volume and ksh. is recorded for each one of the fifty six financial institutions. each month a bulletin is released showing the ten or twenty leading companies in terms of turnover in ksh. these companies differ from one period to another. thus data from nse is first analyzed. returns for shares are derived by taking a base month and the weighted average share price of an investment stock for that month and considering it as a buying price subsequent share prices are taken as the selling prices and the returns are thus calculated. simple random sampling is then used to pick the companies with the best returns over the five year period. this entails ranking the companies in terms of one with the highest turnover in ksh for the five year period. it also involves ensuring that the companies selected are those that are making positive returns frequently. this is due to the fact that this study determines methods that will enable investors make the best investment decisions, thus picking a financial institution with the highest and most frequent positive returns is the first step towards making good investment decision. turnover from shares is derived by multiplying the weighted average share price per month by the turnover of shares in volume per month. simple random sampling is then used to pick a number of companies from those with the highest turnover over the five year period. assumptions made include the following; the sample data size is representative of the population, the values of weights range from negative infinity to positive infinity. it is important to note that the estimators derived and tested are used to predict future performance of certain financial institutions from the rest of the population. these estimators are expected to be used for prediction of future performance of financial institutions situated in any location, the only limitation to the use of the estimators being the accuracy of the forecasting techniques. s–plus version 2000 and mat lab version 6.1 software are used for mathematical and statistical calculations, problem solving, graphing and forecasting. 3. results 3.1. derivation of true functions of diversifiable and non–diversifiable risk let the markowitz portfolio expected returns and diversifiable variance be given by (1) and (2) respectively. see andrew (1995) and alexander (1974). published by atlantis press copyright: the authors 33 e. anyika.     1 )()( i iiu rexre (1)       j ijji ii iim xxxp  11 222 2 (2) non–diversifiable variance is computed as     1 222 i iiq xp  (3) where,  ix ,  2 0 i , and ix = weight of an investment i ir = return of investment i  ure = expected returns 2 i = variance of investment i ij = covariance of investments i and j and u, m & q are arbitrary symbols differentiating i & u expected returns and non–diversifiable & diversifiable variances respectively. to find the weight of investment i that will maximize expected returns and minimize total risk we apply the classical optimization method with no constraints see rao (1994). we thus differentiate the expression; jixxxrxere i j ijji i ii i iimu                    ,2)( 1 11 22 1 2  with respect to ix i.e.  2 2 1 ( ) ( ) 2 2 0, u m i i i j ij ji e r p e r x x i j x              (4) and differentiate 2 2 2 2 2 2 1 1 1 1 2 ,m q i i i j ij i i i i j i p p x x x x i j                   with respect to ix , i.e.  2 2 2 1 4 2 0, m q i i j ij ji p p x x i j x             (5) where   2u me r p = maximum returns (derived by subtracting diversifiable variance from expected returns) 2 2 m qp p = total variance (derived by adding diversifiable variance to non–diversifiable variance) note: i) the second derivative of (4) is equal to 22 i implying that ix obtained will always maximize returns. ii) the second derivative of (5) is equal to 24 i implying that ix obtained will always minimize risk. see rao (1994). equate (4) to (5) to get, the expected returns of investment i . 2 2 1 1 ( ) 2 2 4 2 ,i i i j ij i i j ij j j e r x x x x i j                2 1 6 4 ,i i i j ij j e r x x i j       2 1 1 ( ) 3 2 , 2 i i i j ij j e r x x i j       (6) hence for infinite investments, values of ix are given by the expression                                                    re re re x x x      2 1 2 12 1 2 1 2 2 221 112 2 1 32 232 223    (7) therefore, it follows from the derived values of ix in equation (7) and substituting them in the square root of equations (2) & (3) that the true function for portfolio diversifiable risk is given by 1 2 2 2 1 1 1 ,i i i j i j i i j x x x i j                  (8) and that for portfolio non-diversifiable risk is   2 1 1 22  i iix  (9) 3.2 empirical study using sampled data in this section, an empirical study is done to check if the true functions developed in the previous section work as estimators. i) the data used was turnover of shares for kenya commercial bank (kcb) given in appendix 5 and east african breweries limited given in appendix 6 first we normalize /standardize the data by taking logarithms and getting the first difference i.e. log published by atlantis press copyright: the authors 34 non-diversifiable risk in investment portfolios 1 1 2 t t t t y y y y          , whereby the sum of the ratios of the values of the difference between consecutive periods and the earlier of these periods will give the trend line, see figure 1 given bymurray in1961 . final values for finding the risks are given by the weights,                              142371.0 158996.0 )590005.0(3)167223.0(2 )167223.0(23621844.03 552.0 2 1 2 1 2 1 x x where, variance of kcb = 0.3621844 variance of eabl = 0.590005 covariance of kcb & eabl = 0.167223 expected returns of kcb = 0.158996 expected returns of eabl = 0.142371 therefore diversifiable risk of a portfolio of kcb & eabl = 0.082687821 or 8.268% and its non – diversifiable risk = 8.88%. ii) if we assume that 0ix implying that sales are always positive and 1 1   i ix implying perfect market conditions for the same data as in i) above (markowitz assumptions) then 2 2 2 2 2 1 1 2 2 1 2 122pp x x x x     where 2 pp = portfolio variance. to find the value of 1x that minimizes portfolio risk we differentiate 2pp with respect to 1x and equate it to zero 2 2 1 1 2 12 1 2 2 pp x x x         121 2 11 122  xx  = 0 12112 2 11 222  xx  = 0 0.724369 1x + 0.33444 0.33444 1x = 0 0.389929 1x = 0.33444 1x = 389929.0 33444.0 note: the negative numerator value 1x indicates short selling. (act of selling securities you do not own) since 0ix  ， 1x becomes 389929.0 33444.0 1 x = 0.85769 kcb turnover of shares 1.5 1 0.5 0 0.5 1 1.5 1 2 3 4 5 6 7 8 9 1 11 months t u rn o v e r series1 figure 1: a plot of normalized turnover per month. published by atlantis press copyright: the authors 35 e. anyika. since 1 1i i x    ， 1423053.01 12  xx and diversifiable risk = 85.769 % while non–diversifiable risk = 14.23 % iii) when we use returns of shares of kcb from appendix 1 and eabl (figure 2) from appendix 4 we are not making any of the above assumptions the final results are as follows, the weights 13 14 09 1 23 14 12 07 2 3.3922 10 3.2582 10 6.8657 10 7.20592 10 3.2582 10 1.8238 10 1.5087 10 x x                                  where 3 (variance of kcb) = 3.3922 1310 3 (variance of eabl) = 1.8238 1210 2 (covariance of kcb & eabl ) = 3.2582 1410 ½ ( value for expected returns of kcb) = -6.8657 0910 ½ ( value for expected returns of eabl) = -1.5087 0710 9026.198436,5220 21  xx therefore diversifiable risk = 5.310 and non– diversifiable risk = 17.215 % theorem 1: non-diversifiable risk,   1 22 2 i ix  of a given investment i remains unchanged for i =1, 2, 3, …,  , investment of a given portfolio. we make the following assumptions; i) ix    ii) for diversifiability, 2 20 i ix    and for non diversifiability, 2 2i ix z  , where  z0 . iii) sampled data is normally distributed and representative of the population. iv) variances of investments i 1, 2, 3, …,  . are uncorrelated for non–diversifiability. proof: from equation (5), the value of 1x that minimizes risk for a two sampled investment portfolios is given by, 2 2 2 1 1 2 12 1 2 1 1 2 12 2 1 1 2 12 2 1 1 2 12 2 2 1 1 ( ) 4 2 4 2 0 4 2 4 2 4 4 m np p x s x s x x s x s x s x s x s x s s s            2 12 1 2 1 1 2 x s x s   (10) for three investments the value of 1x that minimizes risk is given by, 2 2 2 1 1 2 21 3 31 1 ( ) 4 2 2 0m n p p x s x s x s x        normalised returns for eabl -0.000003 -0.000002 -0.000001 0 0.000001 0.000002 0.000003 0.000004 0.000005 1 2 3 4 5 6 7 8 9 10 11 months returns series1 figure 2: normalized / standardized returns of eabl (derived by taking the first differences of the returns of shares, then the reciprocal of these differences). published by atlantis press copyright: the authors 36 non-diversifiable risk in investment portfolios which gives 2 1 1 2 21 3 314 2 2x s x s x s   so that 3 312 21 1 2 2 1 1 1 1 2 2 x sx s x s s    (11) for four investments the value of 1x that minimizes risk is given by, 3 312 21 4 411 1 1 1 2 2 22 2 2 1 1 1 x sx s x s x s s s     (12) continuing in the same manner we see that for infinite investments the value of 1x that minimizes risk is given by, 3 312 21 4 41 11 1 1 1 1 2 2 2 22 2 2 2 1 1 1 1 ... x sx s x s x s x s s s s      (13) similarly 2 3, ,...,x x x can be obtained. taking expectations of equations (10), (11), (12), (13), to remove bias in the sample variance and considering the assumption (iv) we obtain           2 1 1 1 2 212 2 1 1 1 2 212 2 1 1 1 22 2 1 1 1 0 0 e x s e x s x e s x e s n x x n as n x               where n is the sample size and n>1. note:  ije  = 0 from assumption (v) . for equation (11), similarly for investments 4, 5, 6,…, , 2 0i ix   .thus non–diversifiable risk estimator (since we are using sampled data) is   122 2q i ip x  of investment i = 1, 2, …, , for a given portfolio remains unchanged. iv) analysis of returns of shares of three investments (see appendix 1, 2 and 3) gives the following results, 18 20 15 1 21 20 15 21 2 15 21 20 3 08 07 07 1.14589 10 2.82915 10 9.10521 10 15097 10 2.82915 10 9.10521 10 6.83994 10 9.10521 10 6.83994 10 4.85532 10 6.86872 10 1.50874 10 3.32536 10 x x x                                                         (14) where the right hand side is the product of the reciprocal of determinant of variance covariance matrix, inverse of variance covariance matrix and values representing means of the three investment returns. 719814553.0 581127467.1 484843067.3 3 2 1    x x x thus diversifiable risk of a portfolio of kcb, eabl &stan chart = 2.54382 0610 %, and its non– diversifiable risk = 8.93518 0610 %. for a two investment portfolio of kcb and eabl diversifiable risk for kcb is 0.01755 and that of eabl is 0.15460. for an investment portfolio of kcb, eabl and stanchart diversifiable risk for kcb is 2.15374 0910 , eabl is 8.71066 0810 and that of stan chart is 9.14787 1110 . clearly portfolio risk has been diversified but non– diversifiable risk for kcb and eabl is not the same as that one for two investments as per the definition of non–diversifiable risk and the findings of theorem one. this is also true for turnover of a three and four investment portfolio. thus there is some white noise in the data analyzed. we therefore remedy our estimator for non–diversifiable risk to include the random error. this is done by adding   122 1 ie i s    to the non– diversifiable risk estimator. we thus denote this model estimator as,           2 1 1 1 1 2 21 3 312 2 2 1 1 1 1 2 32 2 2 1 1 , 1 0 0 , 0 . e x s e x s e x s n x x x n as n x                 published by atlantis press copyright: the authors 37 e. anyika.     1 2 1 22 2 2 1 1 ig i i e i i p x s s        where ie is an independent random variable with mean zero and variance 2 1 eis n  (i.e. sample variance) theorem 2: the non–diversifiable risk model estimator gp is a consistent estimator of non– diversifiable risk qp . we make the following assumptions; i) ie is an independent random variable with mean zero i.e.   02  ie e  and variance 1 2 n ie  ii) variances of investments i 1, 2, 3, …,  . are uncorrelated. iii) sampled data is representative of the population and is normally distributed. where g & q are arbitrary symbols. proof: by definition we note that     1 1 2 2 2 2 2 1 1 ig i i e i i p x s s        thus     1 21 22 2 2 2 2 2 2 1 1 1 1 2 i ig i i i i e e i i i i s x s x s s s               (15) the expectation of equation (15) is     2 2 2 2 1 1 1 2 i ig i i i i e e i i i e p e x s e x s s e s                             hence  2 2 2 1 1 g i i i n e p x n             as n  2 2 2 1 ( 0)g i i i e p n x             note : since 2 is is the sample variance of sampled data of investment i which is normally distributed thus there is bias in the unmodified form of the sample variance. therefore   1 22 2 1 g i i q i p x        this proves that gp is an unbiased estimator of qp . from (15), observe that       11 22 1 1 2 2 2 2 2 2 2 2 2 1 1 1 4 4 4 4 4 1 1 1 2 2 2 1 1 var var var var 2 2 2 1 1 1 var i i i i g i i i i e e i i i i i i i e i i i i i e i i p x s x s s s x x n n n x                                                                                        1 1 2 2 4 1 4 4 4 4 4 2 2 2 4 1 1 1 1 1 1 2 1 2 2 var 1 1 i i i i e i i i i i e i i e e i i i i i i n x x x n n                                                                                   (16) note: in equation (16),   122 2i ix  is rewritten as    1 22 2 2 2 i i i ix x   and   1 22 ie  as    1 22 2 i ie e    to assist in simplification. also  2 42var 1 s n   as proved by tobago (2010) recalling that as n    g qe   , it follows from this result (i.e.  var 0g  ) that gp is a consistent estimator of non–diversifiable risk. 4. discussion 4.1. presentation of results 4.1.1 forecasts of time series in forecasting we used garch (generalized autoregressive conditional heteroskedastic) models. this is because the time series values have low valued correlations. in garch modeling the true garch (a, b) model parameters of the time series are entered. these parameters correspond to a given garch (a, b) model for the conditional variance f (t) and innovations y (t), sequences published by atlantis press copyright: the authors 38 non-diversifiable risk in investment portfolios                           2 2 2 2 2 1 2 2 ( ) f t = l t + s 1 * f t 1 + s * f t + s p * f t p + w 1 y t + w y t + w q y t q      for time steps t = 1,2, … n, where s = order of ar w = order of ma a & b = model order determined by the number of elements of s & w s, w = coefficient. t = the current time index. y (t) = square root f (t) l (t) where l (t) is an identical and independent sequence n (0, 1) and y and f are related. the forecasts for the investment time series from appendix 1 through to appendix 3 are given in table i and table ii. 4.1.2 prediction of expected returns diversifiable risk and non–diversifiable risk using the forecasts from section 4.1.1 above and equations 7, 8 and 9 the predictions for expected returns, diversifiable risk and non–diversifiable risk are made as table iii and table iv. 4.2. result discussion 4.2.1 derivation of forecasts in forecasting using garch models we let t be the current time index, {f (t)} be the return series of interest, {y (t)} the innovations noise process and l(t) be an identical and independent sequence. the input coefficient vectors ar (autoregressive) and ma (moving average) are specified exactly as they would table i: the table below represents forecasted returns of shares of the given companies in kenyan shillings ( ksh). company eabl icdc kcb stan chart 1 55660000 3881300 38460000 32670000 2 42710000 3590200 29520000 25070000 3 36010000 3320800 24880000 21130000 4 31720000 3071700 21920000 18620000 5 28680000 2841300 19810000 16840000 6 26370000 2628200 18230000 15480000 7 24550000 2431000 16960000 14400000 8 23050000 2248600 15930000 13540000 9 21810000 2080000 15070000 12790000 10 20740000 1923900 14330000 12180000 11 19820000 1779600 13690000 11630000 12 19000000 1646100 13130000 11160000 13 18290000 1522600 12640000 10730000 14 17650000 1408400 12200000 10360000 15 17070000 1302800 11790000 10020000 16 16540000 1205000 11430000 9710000 17 16060000 1146000 11100000 9420000 table ii: the table below represents forecasts of turnover of shares of the given companies in ksh. company eabl icdc kcb stan/ chart 1 105230000 8609700 406500000 69829000 2 99830000 7718800 385640000 66245000 3 94710000 6920100 365850000 62846000 4 89850000 6204100 347080000 59621000 5 85240000 5562100 329270000 56561000 6 80860000 4986600 312370000 53659000 published by atlantis press copyright: the authors 39 e. anyika. be read from the arma (autoregressive moving average) (r,m) model equation when solved for f (t): f (t) = l (t) + ar (1) f (t-1) +… + ar(r) f (t-r) + y (t) ma (1) y (t-1) +… + ma (m) y (t-m) note that the coefficients of f (t) and y (t) are assumed to be 1, and are not part of the ar / ma input vectors. for the following arma (2, 2) model, of icdc f (t) = 0.036y (t-1) 0.002 y (t-2) + y (t) + 0.402 y (t1) + 0.107 y (t-2) ar = [0.036 – 0.002] and ma = [0.402 – 0.107]. the first 20 weights of the infinite order ar approximation may be found as follows: zi = garchar ([0.036 – 0.002], [0.402 – 0.107], 20); where zi represents weights of the polynomials generated and garchar converts j–th lag of the return series and innovations processes f (t-j) and y (t-j), respectively. to maintain consistency, the j-th element of the truncated infinite–order autoregressive output vector, zi (j), is the coefficient of the j–th lag of the observed return series, f (t-j), in the infinite order representation of the input arma (r, m) process. the ar and ma input vectors differ from the corresponding ar and ma ‘polynomials’ formally presented in time series. to estimate, and make forecasts from the garch (a, b) parameters from the equation: y (t) = square root (f (t)*l (t)) where y (t) represents innovations noise process, you simulate (i.e. , to reverse–engineer the process for comparison) (see bollersler (1986), box, jenkins (1994), engle and robert (1982), hamilton 1994). figures 3 and 4 show that the forecasted time series is normally distributed thus there is no need for its standardization before making predictions. this also explains the downward trend of the forecasts i.e. to maintain the bell like shape of the curve indicative of normally distributed data. (note that the plots show values for half the distribution because both halves are identical). see lucey (2000). returns of shares are forecasted in the long term since these data were prepared for long term investors who are not ready to sell their shares in the near future. short term forecasts are meant for speculative buyers who are risk averse and can sell their shares anytime the prices go up. the estimators used to predict future expected returns, diversifiable risk and non– diversifiable risk are applicable within the normal ranges of negative infinity to positive infinity. this is reflected in tables iii and iv where values range from the tens to hundreds to thousands etc, compared to markowitz estimator of diversifiable risk whose weight values ix must be greater than or equal to zero i.e. 0ix . the sum of these weights for any number of investments in a portfolio must be equal to one i.e. table iii: seventeen month predictions of returns of shares of the given companies portfolio expected returns diversifiable risk non–diversifiable risk eabl & icdc 0.559 or 55.9 % 0.219 or 22 % 0.223 or 22.32 % icdc & stanchart 1.27 or 127 % 0.503 or 50.3 % 0.5104 or 51.04 % eabl & stanchart 1.14 or 114 % 0.478 or 47.8 % 0.478 or 47.84 % eabl, icdc & stanchart 1.34 or 134 % 0.532 or 53.2 % 0.552 or 55.21 % table iv: six month predictions of turnover of shares of the given companies portfolio expected returns diversifiable risk non–diversifiable risk eabl & icdc 1822 % 150.82 % 255 % icdc & stanchart 1821 % 150.81 % 262 % eabl & stanchart 2063 % 203 % 203.2 % eabl, icdc & stanchart 474 % 83.79 % 95.3 % published by atlantis press copyright: the authors 40 non-diversifiable risk in investment portfolios     1 1 i ix . the first assumption limits the data set being used since if it is one with negative returns the estimator collapses. this means that one assumes positive returns always and yet many a time financial institutions experience losses or negative sales. the second assumption implies perfect market conditions, i.e. the portfolio is efficient with no systematic risk present. under normal financial environments this is not true since fluctuations in returns do occur frequently. the development of an estimator for non– diversifiable risk in this study is an indication of the presence of market risk. for kr the return of k capm is given by          2 m f km k f m e r e r e r e r      where m= market portfolio ( )ke r = expected returns of the risky security k  fe r = risk free rate of return  me r = expected returns of the market portfolio m 2 m = variance of the market portfolio m km = covariance between the single risky security k and the market portfolio m this model works under the following assumptions, see andrew (1993): i) all investors are risk–averse and measure risk in terms of standard deviation of portfolio return (as for the markowitz model). ii) all investors have a common time horizon for investment decision making ( e.g. one month or two years ). iii) all investors have identical subjective estimates of future returns and risks for all securities. there exists a risk–free asset and all investors may borrow or lend unlimited amounts at the risk–free nominal rate of interest. iv) all securities are completely divisible, there are no transaction costs or differential taxes, and there are no restrictions on short–selling. information is freely and simultaneously available to all investors. although there is a great deal of evidence which supports capm, much courage is required to develop a model on the basis of these assumptions. many of them are clearly unrealistic, for example the assumption that there exists a risk–free asset and all investors may borrow or lend unlimited amounts at the risk–free nominal rate of interest. the model should therefore not be relied upon for more than general indications of the market pricing mechanism as seen in section 2.1. the estimators developed in this study have addressed the above short comings such that none of the above assumptions are made in deriving these estimators. furthermore the beta factor i.e. 2 km k m b    is an indicator of non–diversifiable risk. this is not a good indicator since from the definition of non–diversifiable risk, there should not be any correlation between the various investments, yet the expression km in the beta factor formulae represents covariance between single risky security k and the market portfolio m which counterfeits the definition of non–diversifiable risk, thus giving weight to the doubts cast on the capm practical applicability by researchers in the 1980’s. the predictions in tables iii and iv indicate that the higher the risks the higher the expected returns. this is expected since high risks imply venturing into unknown areas. thus one can only be motivated to do so if he/she expects big rewards for taking these risks. this is experienced in financial markets where monopolists venture into new business areas. if their ventures are successful they expect to reap maximum benefit since there is no competition from other financial institutions. if they fail in their ventures they stand to incur heavy losses since they will not be able to recover their base capital among other costs incurred. there would be no other financial institution trading in the same items thus they cannot sell their business to them or merge with these financial institutions in order to diversify their risk thus minimizing it and reaping the benefits of economies of scale. the predictions show a clear distinction between diversifiable risk and non–diversifiable risk, whereby non–diversifiable risk is always the higher of the two risks. table iv which has six month predictions of turnover of shares shows this observation clearly. unlike the data for returns of shares, the data for published by atlantis press copyright: the authors 41 e. anyika. turnover of shares is not derived using a common reference price. variances in the time series tend to be huge leading to higher risks. the length of time the forecasts postulate, i.e. six month have an impact on risk. whereby investors can sell their investments easily when the prices are high thus returns will be high. the high returns will thus be accompanied by high risks as experienced by monopolists. 5. conclusion and recommendations 5.1. conclusion in this study a non–diversifiable risk model estimator, diversifiable risk and expected return estimators have been developed. in doing so we have done away with the many theoretical like assumptions that are usually employed in similar studies as seen in section 4.3.1 the presence of white noise in non–diversifiable risk has been established which confirms the fact that such a risk is independent i.e. cannot be diversified. this paper confirms that the risks are high as it should be in the stock market due to the high volatility experienced. moreover, our study indicates that diversification reduces diversifiable risk but not non– diversifiable risk. investment analysts have relied on markowitz estimator of risk markowitz (1952) , the market model and the capm to analyze the effect of risk in investment decision making. these have had various shortcomings as evidenced by earlier researchers (section 2.0) and this study’s findings (section 4.3). development of risk estimators and expected return estimators that have addressed these shortcomings in this study is expected to improve the accuracy of predicting future expected performance of financial institutions. investment analysts can now rely on the predictions to make good investment decisions. financial institutions can now venture into the unknown future financial environment knowing what to expect. predictability of the future of financial institutions is expected to spill over to investment markets. overtime investment markets have been known to be unstable, particularly stock, credit, bond and foreign exchange markets. it is hoped that a systematic method will be established which will ensure frequent determination and publication of non–diversifiable risk for respective financial institutions, i.e., those on nse and other capital markets with a view to reducing their future uncertainties. stabilized investment markets mean increased investor confidence. we should therefore expect to see an influx of investors in the investment markets. a thriving investment market bolsters the economy. such that we have increased levels of foreign exchange which strengthens the value of the kenyan shilling. imports and exports then become cheap thus influencing the economy positively. new financial institutions will be started and the existing ones sustained. employment opportunities will be created leading to increased purchasing power, money supply and gross domestic product. all these improve the economy and eventually the way of life of individuals in this economy. lending and borrowing of funds is hoped to increase due to stable interest rates. small scale businesses will thrive, improving the standards of living of individuals in these business environments. determination of non–diversifiable risk and prediction of future performance of financial institutions does not eliminate risk. nevertheless their proper estimation enables financial analysts make accurate current and future business plans (paul 2003). expected future cash flows are made accurately by incorporating this determined risk in their discounting methods. correct capital investment decision making is facilitated. required levels of cash and cash equivalents will always be maintained leading to high profitability of the financial institutions. capital allocation will also be made accurately, since the optimum portfolio will be accurately identified. despite the good investment decision making and proper financial planning anticipated portfolio risk has to be addressed as an entity. if left unchecked it can cause unexpected losses. this is particularly for financial institutions which trade in interest related portfolios and foreign exchange trading portfolios. with the determination of total risk, risk management techniques are expected to be properly initiated and established, jennifer (2003). hedging schemes such as derivatives, options and futures are hoped to cost less, karithi (2003). savings are anticipated in risk management departments which should be used to increase the wealth of owners of the financial institutions. these will then motivate them to continually invest in them thus sustaining their growth and development. published by atlantis press copyright: the authors 42 non-diversifiable risk in investment portfolios having discussed the positive impact this study is expected to have on the general environment, unstable or risky environments should be a thing of the past, while stabilized environments should be the norm. 5.2. recommendations from the above conclusive results this paper recommends the use of all derived estimators in both capital allocation in investment portfolios and net present value investment decision making criterion. risk is not only experienced in financial environments. it is also encountered in such areas as medical practices, whereby there is some risk attached to various curative methods. notably are the radiotherapy and chemotherapy methods of curing cancer. the environment also experiences some health hazards due to environmental pollution or poor sanitation, i.e. environmental risk. it would be challenging to adapt the risk determining methods developed in this study in determining medical and environmental risks with a view to reducing them. references andrew, a., della, b., philip, b., and peter, e. (1995) investment mathematics and statistics, pg. 208–257. kluwer law international box, g., jenkins, g., reinsel, g. (1994) “time series analysis; forecasting and control” prentice hall bollersler, t. (1986) “generalized autoregressive conditional heteroskedasticity’’ journal of econometrics, vol. 31, pp 307–327 dimson, e. (1988) stock market anomalies, cambridge university press engle, r. (1982) “autoregressive conditional heteroskedasticity with estimates of the variance of united kingdom inflation “ econometrica, vol. 50, pp.987–1007. hamilton, j. (1994) ‘time series analysis’ princeton university press. harper, r. (2003) asset allocation, decoupling and the opportunity cost of cash, journal of portfolio management, vol. 29 issue 4, p25. jennifer, j. (2003) risk management at the world bank global liquidity portfolios, pp 11 17. karithi, m. (2003) management of foreign exchange risk, kasneb newsline pg 19-25 lintner, j. (1965) the valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets, review of economics and statistics, 47, pg 13–37. markowitz, h. (1952) portfolio selection, efficient diversification of investments, basil blackwell. mossin, j. (1966) equilibrium in a capital asset market, econometrica, 34, pg 768–783, october murray, r.(1961) theory and problems of statistics si units, pg. 231, mcgraw–hill book company nicholas, f. and laverne, s. (1989) quantitative forecasting methods, pg. 117 &118, pws-kent. mcclure, p. (2003) management’s discussion and analysis, financial statements and investment portfolios international finance corporation 2003 annual report volume 2 pg 11–15. macharia, p. (2001). nse, bulletin pg, 2–4 nse rao. s. s. (1994) optimization theory and application wiley eastern limited new age international p.gs. 37–40. raymond, b. (2000) financial management 6th edition pg, 123-134 sharpe, w. (1964) “capital asset prices: a model of market equilibrium under conditions of risk”, journal of finance, vol. 19, pp. 425-442. stulz, r. (1999). what’s wrong with modern capital budgeting? financial practice & education, vol. 9 issue 2, pp7, dp publications taboga, m. (2010) ‘lectures on probability and statistics’, http://www.statlect.com. vlaar, p. (2000) value at risk models for dutch bond portfolios. journal of banking and finance, 24: 1131–1154. wilkes, f., samuels, j., brayshew, r. (1996) management of company finance, pg 262–265. international thomson business press. published by atlantis press copyright: the authors 43 e. anyika. appendices appendix 1: appendix 2: table v: returns for kcb shares from january 2009 – december 2013 base price april 2009 in kenyan shillings (ksh) year month 2009 2010 2011 2012 2013 jan -254977217 -4909732 -48293377 -28599388 -176773446 feb -11858496 -18114517 -35537228 -86093014 -27392971 mar -20843003 -11145914 -20663935 -235975275 -235975275 apr -35789720 -20446319 -16526970 -249196506 -201979585 may -12178766 -72691432 -44283975 -12896377 -72325367 jun -31529069 -69476411 -12607319 -24309319 -3479830 jul -8193042 -15160040 -94872136 -28713343 -80326198 aug -8925414 -78123604 -73423079 -16496101 -30548808 sep -27908155 -35911665 -7891213 -48476514 -44356839 oct -6450228 -7641496 -23904507 -60275745 -31303040 nov -19916397 -19712658 -10867080 -140518771 -39819260 dec -6620513 -2055670 -4412974 -16975041 -61658147 table vi: returns for standard chartered shares from january 2009 – december 2013 base price april 2009 in ksh year month 2009 2010 2011 2012 2013 jan -10356582 2465938 529514 1111462 7044060 feb 114576 6572177 3748395 1485758 7846250 mar 328887 938150 4481874 393326 21271177 apr -260376 811697 6695297 -82382 21024330 may -951995 473787 910638 713692 40984696 jun 435601 -239449 2715433 1376041 53320718 jul 2712442 698182 3454344 2466946 14738214 aug 2979260 170156 2664586 3125082 45759692 sep 3627526 11828563 788886 789498 30389636 oct 2478381 6078253 2120709 2634216 250200283 nov 4948147 1699087 618373 4069588 83728614 dec 1474898 -832158 543333 6367901 15751289 stan chart is one of the three companies with the highest returns and least negative sales thus analyzed for investment purposes. published by atlantis press copyright: the authors 44 non-diversifiable risk in investment portfolios appendix 3: appendix 4: table vii: returns for icdc (industrial commercial development corporation) shares from january 2009 – december 2013 base price april 2009 1n ksh. year month 2009 2010 2011 2012 2013 jan 11935779 834948 3910778 -661320 -4407558 feb 418644 1115537 2724842 -379153 -1360183 mar 1346134 748676 3267626 -614656 -137580 apr 3239774 130815 486736 -2644600* 89436 may 971928 936465 3403992 -980303 1238786 jun 731335 -356349 9535859 -1628149 3385594 jul 850178 458268 185853 -10295746 -510603 aug 1439235 344772 961124 -32399844 15592897 sep 4163159 4570032 -652724 -236345 5216308 oct 2460198 945932 73658 -552725 6314293 nov 774004 2910810 -666640 -572520 3275379 dec 1459217 507676 -96840 -148838 13410155 * this value is calculated from table xv as 20.875 – 38.375 = 17.5 x 151120 = 2644600 table viii: returns for east african breweries shares from january 2009 – december 2013 base price april 2009 in ksh. year month 2009 2010 2011 2012 2013 jan -10106177 -173378 4433440 980372 22434220 feb 2715705 -1470663 17493087 1298971 49769005 mar -9821081 -830676 5217513 1314180 50841507 apr -283444 -914474 0* 856305 90741122 may 1748910 -2648273 2786088 3520085 137930125 jun 948375 -7199135 459345 2908110 199612113 jul 489416 1106750 676910 13447207 22265976 aug 16667673 -3985070 -756480 14598393 323570569 sep 4112532 1861435 322465 22556215 155506703 oct 386565 3300665 4528066 8768165 162636994 nov -321409 -9450660 1236078 29758680 177642340 dec 503414 1054648 4122929 60118940 82052882 * is a rare value which indicates that the share price of april 2011 is at par with that of april 2009. published by atlantis press copyright: the authors 45 e. anyika. appendix 5: appendix 6: table ix: turnover for kcb from shares for the years 2009 – 2013 in ksh year month 2009 2010 2011 2012 2013 jan 327531709 4107488 22019778 8954014 77310126 feb 57563116 18114517 16772678 25391488 11980046 mar 64039370 6958143 10979475 63794401 31424618 apr 6232688 14910085 32051257 67655160 148367720 may 518649381 43367048 21730863 3217234 3331709520 jun 43872364 43372557 13403376 4093669 72749417 jul 15061956 9443960 29123540 4765700 97403422 aug 10745499 43918134 20571842 2858406 54024740 sep 31055691 20150434 3966706 6897065 85512669 oct 6136675 4387974 6767486 10770750 80214040 nov 16424561 12306168 3605193 30860459 137787282 dec 5838395 12496811 1241589 4357636 79203149 all values are greater than zero indicating presence of trading for the shares of kcb within this period. table x: turnover for east african breweries from shares for the year 2009 – 2013 in ksh. year month 2009 2010 2011 2012 2013 jan 456462305 4629179 57827475* 15273156 60558630 feb 14000968 9845272 228170700 18433014 96149482 mar 108458889 32040360 55053761 11714976 86315305 apr 78230475 30749188 44099577 8507807 171045565 may 42119583 40502995 33655937 33067468 66678263 jun 14083369 39176690 5991450 18107100 295773078 jul 10172851 7581238 7888605 108469564 37608452 aug 121598251 59206758 3609040 93887704 429389988 sep 34900943 26414646 2949602 108146238 205652636 oct 3955849 46838001 30125497 36685194 201641710 nov 12397185 45087524 11018748 98166600 212222050 dec 20424215 26285067 71302424 153152160 233933799 this value is derived as follows: from appendix 10 the row for eabl we have 771033 x 75 = 57827475 published by atlantis press copyright: the authors 46 connection parameters of heavy-tailed operational risk measurement model and management model received 26 april 2016 accepted 2 july 2016 connection parameters of heavy-tailed operational risk measurement model and management model meiling he1, shutao qing2, jianming mo1, xiang gao3 1chinese finance research institute, southwestern university of finance and economics, chengdu 610074, china 2hunan provincial party school department of economics, changsha 410006, china 3school of international business administration, shanghai university of finance and economics, shanghai 200433, china abstract in order to connect the heavy-tailed operational risk measurement model with management model, a model identifying the crucial supervising parameters of operational risk is built after the heavy-tailed operational var’s sensitivity is theoretically researched by the elasticity analysis method. further, the analysis of model application is illustrated with a numerical example. the crucial supervising parameters connect the operational risk measurement model and management model, which make the operational risk management frameworks to be a complete system. and a dynamical supervising system of operational risk is established. this research in theory improves the application of loss distribution approach to the operational risk measurement and management. keywords: operational risk; supervising parameters; elasticity theory; operational var 1. introduction operational risk management is an integrated frame work which includes risk identification, measurement, analysis, monitoring, mitigation and economic capital allocation. in such management framework, how to combine risk management model with measurement model is a major issue. however, current operational risk mainly aims to calculate capital reserve precisely and monitor management effect by directly observing the variation of the capital reserve. thus, operational risk measurement and management are practically independent system, which leads to lack efficiency and target in proposing measures for risk management. generally speaking, the aim of risk measurement is not restricted to calculate regulatory capital, more importantly, to provide valid foundation for making management regulations, thereby reducing the whole risks. *national natural science foundation of china(71171167, 71671144, 71373213, 71301130, 71501117, 13yjc790024) however, it is relatively late for us to attach importance to operational risk in theory and practice field. it wasn’t until june 2004 that new basel accord was issued and since then operational risk has been included in the regulatory system. therefore, the framework of operational risk management is still in the process of improving and burgeoning. and how to find connection parameters to combine measurement model with management model is still a problem which has not been solved. existing literatures regard bayesian network as the best model to manage operational risk. the parameters of the nodes in bayesian net work represent the characteristic parameters of the loss distribution approach. the methodology of choosing network nodes is divided into three categories: (1) regard key risk indicators (e.g., employee turnover rates) as nodes; (2) consider loss frequency and loss severity caused by loss events as nodes; (3) define characteristic parameters of the loss distribution as nodes. actually, three types of journal of risk analysis and crisis response, vol. 6, no. 3 (october 2016), 122-134 published by atlantis press copyright: the authors 122 m.l. he et al. / heavy-tailed operational risk measurement model nodes reflect operational risk from diverse perspectives. while key risk indicators are able to reflect the change of the operational risk, they still cannot reflect the measurement changes in results. although the mathematical expectation of loss frequency and severity distribution can reflect measure changes directly, under the condition that loss severity distribution is heavy-tailed, loss severity distribution may not have mathematical expectation. moreover, some empirical researches have proved that loss severity distribution has significantly heavy-tail characteristic, meaning that under some loss severity distribution, internet nodes are unavailable. fortunately, no matter how loss distribution changes, characteristic parameters still exist and directly reflect the change result of measurement. consequently, in bayesian model, using key risk indicator or expectation of loss frequency and severity as nodes do have some drawbacks, therefore it is appropriate to take characteristic parameter as internet nodes. previous literatures have found that the object that the operational risk management target at is exactly the characteristic parameter of loss distribution. chapelle and grama et al (2008) found that varied operational risk management measures affect different loss distribution: some management measures(e.g., “dashboard”) solely influence loss frequency distribution, some management measures (e.g., “rapid reaction”) solely influence loss severity distribution, while some other management measures (e.g., “audit tracking”, “business line”) have both influence on loss frequency distribution and loss severity distribution and thereby achieving regulation deviation. further more, that means, disparate management measures have different impact on different characteristic parameters and operational risk management is essentially to govern characteristic parameter. in order to maximize management efficiency, the chief management objects should be the characteristic parameter which exerts enormous effect on operational var. these parameters can be identified only by measurement model. can we identify the key management parameters by measurement model? at present, financial institutes mainly use loss distribution approach to measure operational risk. under that method, operational var is formulated by loss severity distribution and loss frequency distribution. however, with lda, we can only calculate operational var by convolution calculation formula, sparse vector method and panjer recursion etc. therefore, analytical solution of the operational value at risk is non-existent, which results in two problems. for one thing, we do not know which parameters exert the biggest impact on operational var, so management target is uncertain. for another thing, when one parameter changes, we are unable to tell whether capital reserve estimate changes and the direction it changes to. consequently, we fail to evaluate our risk management measures that mean operational risk measurement cannot afford useful information for risk management so that blindness exists in management. the existing of analytical solution of operational var is a prerequisite for addressing the above problems. bocker and kläuppelberg (2005), bocker and sprittulla (2006) and bocker (2006) have found that, with lda, when it comes to the general distribution, the analytical solution to operational var is not existing. but as for the heavy-tailed risk measurement, the value of the tail does have analytical solution. operational var exactly satisfies the condition: new basel accord stipulated that, the regulatory capital for operational risk is calculated by var at a confidence level 99.9%. it is actually tail value under high confidence level as well as heavy-tailed distribution. therefore, there is an analytical solution to heavy-tailed operational var. the analytical solution of operational var combines characteristic parameters with operational var directly. and thereby it is likely to discuss the sensitivity of capital reserve estimate to changes in characteristic parameter and its change trend. by doing so, we can identify the parameter, which influences capital reserve most as the key management parameter of the elasticity, and furnish references for management measures. that is, the key management parameters are used to synthesis measurement model with management model. measurement model not only affords management object to management model and take the function of effect surveillance of management model, but also provides valid ground for management measures revision. hence, the key management parameters technically become connection parameters, published by atlantis press copyright: the authors 123 m.l. he et al. / heavy-tailed operational risk measurement model turning the framework of operational risk management into a cycling system. literature 10 manifests that, when we use generalized extreme value distributions to model severity distribution, some operational loss events compiles with weibull distribution. so our article discusses the combination parameter for measurement model and management model under that distribution. literature 16 finds that, if loss severity distribution follows weibull distribution, we can get the analytical solution of the operational var. moreover, this article investigates the confidence level and sensitivity of operational var using computer simulation. based on this, our paper assumes the loss severity distribution follows weibull distribution, and explores the key management parameters by analyzing the sensitivity of operational var to characteristic parameter. first, our paper derives the analytical solution of the operational var under high confidence level. we probe into the parameter elasticity of operational var and discrimination model is established by operational risk measurement model and management model. finally, we prove its validity by numerical analysis. operational risk is embedded in all kinds of financial institutions and economic activities. compared with conventional market risks and credit risks, the management of operational risk is harder. partially, it cannot be eliminated by selling positions. our article creatively comes up with a valuable model to manage operational risk. and by identifying and researching the key parameters of operational risk, financial institutions are able to curb operational risk effectively and manage operational risk in some sense. 2. measurement of operational var using lda to measure heavy-tailed operational risk , one can derive the analytical solution of the operational var( ( )opvar α ) as follows at high confidence level: 1 1( ) (1 ) ( )t α opvar f en t a −∆ − ≅ − ∆ (1) where t∆ is the target horizon, a is a confidence level close to 1, f(⋅) is cumulative function, )( ten ∆ is the expected loss frequency during t∆ . for convenience, in the following, we use μ to replace )( ten ∆ for simplification. based on the nature of cumulative distribution functions, equation (1) implies 1)1(10 ≤−−≤ μa and consequently 1)1( ≥− αμ . according to the numerical results for operational risk loss severity distribution in literature 10, we assume the loss severity follows a weibull distribution: ])(exp[1)( wξ w w θ x xf −−= 000 >>> ww ,θ,ξx (2) where x denotes loss severity, wθ and wξ are the shape and scale parameters of the weibull distribution respectively. plug equation (4) into equation (2) to obtain w α μ θopvar wwt ξa 1 δ )1 (ln)( − ≅ 000 ≥>> µ,,θξ ww (3) since 1 1 ≥ − α μ , then 0 1 ln ≥ − α μ , operational risk restores normalcy, 0)(δ ≥wtopvar a , equation(3) makes sense. when 1)1( =− αμ , operational risk is non-existent, 0)(δ =wtopvar a , which indicates that the financial institution in question has ceased business operations. only on the condition that 1)1( ≥− αμ , we can discuss the key management indicators for operational risk. from equation (3) we know that at some certain confidence level a , the estimated )(αopvar only depends on characteristic parameters ( wθ , wξ andμ). wθ and wξ are the shape and scale parameters of the weibull distribution respectively, while μ is the characteristic parameter of loss frequency distribution. different management measures have different impact on loss frequency distribution as well as severity distribution. to be specific, policy effect on characteristic parameters differs, so the influence on )(αopvar differs. consequently, the parameter which has greatest effect on )(αopvar is the key management indicator for operational risk management. naturally, it is the connection parameter of measurement model and management model as well. in the following, we will discuss how to identify these parameters. 3. connection parameters of measurement model and management model 3.1 theoretical model connection parameters of measurement model and management model are the key management parameters for operational risk. in order to manage risk comprehensively, we need to identify two kinds of key parameters: first, distinguish main influential factors of published by atlantis press copyright: the authors 124 m.l. he et al. / heavy-tailed operational risk measurement model operational risk. there is no doubt that these influential factors can provide references for authorities to draw on when setting rules. second, we need to distinguish the key parameters which affect operational var’s sensitivity enormously. by monitoring and predicting the possible change of these key parameters, we can come up with possible direction for regulators to revise their measurements. in the following, let us investigate it through the sensitivity analysis )(αopvar . for different characteristic parameters ( wθ , wξ and μ ), the numerical size and variation range differs greatly. the variation of )(αopvar (i.e., )(δ αopvar ) caused by absolute variation of wθ , wξ and μ ( wθ∆ 、 wξ∆ 、 μδ ),cannot fully reflect the influence of characteristic parameters to the sensitivity of )(αopvar s. only the percentage change in the ( )()(δ αopvarαopvar ) )(αopvar in response to one percent change in the characteristic parameter ( ww ξξδ 、 ww θθδ 、 μμδ )does accurately denotes the sensitivity of )(αopvar relating to wθ wξ and μ .according to elastic theory, elasticity can be quantified as the ratio of the percentage change in one variable to the percentage change in another variable, while the latter variable has influence on the former. therefore, we can use parameter elasticity of )(αopvar as a proxy of the )(αopvar sensitivity with respect to characteristic parameters. definition the elasticity of )(αopvar in response to wθ , wξ , μ ww ξξ ξξ αopvarαopvar e w w δ )()( lim 0 ∆ →∆ = ww θθ θθ αopvarαopvar e w w δ )()(δ lim 0δ → = μμ αopvarαopvar e μμ ∆ = → )()(δ lim 0δ plug equation (3) into the definition we obtain: 1= wθ e (4) 1 ) 1 (lnln − − −= w w α μ eξ ξ (5) 1) 1 ln( − − = α μ ξe wμ (6) equation (4) denotes that, scale parameter elasticity of )(αopvar is unit elasticity. i.e., )(αopvar always changes 1% while θ changes1%.equation(5)and(6) imply that shape parameter elasticity of )(αopvar and frequency parameter elasticity of )(αopvar have no relationship with scale parameter, which means scale parameter has no impact on the )(αopvar sensitivity of characteristic parameters of ξ and μ . but shape parameter and frequency parameter do have some influence. the reason for this is that the operational risk is influenced by both loss severity distribution and loss frequency distribution. in the loss severity distribution, shape parameter denotes the thickness and length of the tail, (i.e., it decides the tail risk). in the loss frequency distribution, operational risk is decided by loss frequency. scale parameters and frequency parameters respectively represent the influence of severity distribution and frequency distribution. if the frequency parameter is the key parameter, then )(αopvar is mostly influenced by loss frequency distribution. thus, identifying the key parameter for operational risk actually equals to selecting parameter from shape parameter and frequency parameter. furthermore, we can distinguish the key influential indicator. in the following, we list the general methodology to judge the key management parameters. before we come up with our propositions, we would better reclaim our assumptions: the financial institutions we concerned about use loss distribution approach to measure operational risk and do not cease business operations. loss severity distribution of operational risk complies with weibull distribution. proposition 1 under the previous assumption, we find following rules (i) 0>μe ; as 11 ln0 ≤ − < α μ , 0≥ wξ e ,as 1 1 ln > − α μ , 0< wξ e (ii) as 1→α , −∞→ wξ e , 0→μe (iii)as 1) 1 ln(ln 1 ln ≥ −− − α μ α μ μξ ee w ≥ , otherwise, μξ ee w < .(see proof in appendix a) from proposition 1(i), we know that two possible factors contribute to the risk magnifying: one is the increasing of frequency parameters, published by atlantis press copyright: the authors 125 m.l. he et al. / heavy-tailed operational risk measurement model that mean though loss severity is not huge, increasing loss frequency would lead to overall loss increase. the other is the variation of the shape parameter: when 1 1 ln > − α μ , as shape parameter increases, the tail of the operational risk is thinner, and operational risk increases. conversely, if shape parameter decreases, operational risk decreases. specifically, in order to reduce operational risk, the change direction of shape parameter hinges on different operational condition. hence, in the reality to manage operational risk, if shape parameter is the key management parameter, we should judge the case first. then we can fix the change direction of the parameter and make management measures to achieve it. further, according to proposition 1(ii), theoretically, when confidence level a approximates to 1(confidence level is very high), shape parameter would exert enormous influence on operational risk (tend to infinity), while frequency parameter has little influence on it (tend to zero) according to proposition 1(iii), the different influence two parameters exert on operational risk can be represented by the criterion ])1([lnln])1([ln αμαμ −− . set )]1(ln[ln)]1(ln[ αμαμe −−−=∆ . given 1>∆e , shape parameter’s influence is greater than frequency parameter. as a increases, e∆ increases and the diversity becomes larger. given 1<∆e , shape parameter’s influence is less than frequency parameter. as a increases, e∆ increases and the diversity becomes smaller. when a is fixed, discriminant is only decided by μ ,i.e., the diversity between wξ e and μe is decided by μ . set %9.99=a , figure 1 draws how discriminant changes against the frequency parameter μ . fig.1. the ratio of weξ to μe as 1 1 > − α μ , then 001.0>μ . therefore figure1 regards 001.0=μ as the starting point of abscissa. given 0=∆e , we can get 0027.0=μ , 0= wξ e , i.e., shape parameters has no influence on operational risk. given 1−=∆e , we can get 0058.0=μ , i.e., as shape parameter increases, operational risk decreases and shape parameter and frequency parameter have the same influence on it. therefore, there exist three regimes for the relationship between μ and e∆ . according to figure1: in section ① , given 0027.0001.0 << μ , then 0>∆e and 1<∆e we know 0> wξ e and μξ ee w < .as shape parameter (or frequency parameter) increases, operational risk increases, but frequency parameter exerts greater influence than shape parameter. under this condition, frequency parameter is the main published by atlantis press copyright: the authors 126 m.l. he et al. / heavy-tailed operational risk measurement model factors to affect operational risk. i.e., loss frequency distribution is the main management object in section ② , given 0058.00027.0 << μ , then 0<∆e , 1<∆e . we know 0< wξ e μξ ee w < . as shape parameter decreases (or frequency parameter increases), operational risk increases, but frequency parameter exerts greater influence than shape parameter. under this condition, frequency parameter is the main factors to affect operational risk. i.e., loss frequency distribution is the main management object. in section ③ , given 0058.0>μ then 0<∆e and 1>∆e we know 0<wξe and μξ ee w > . as shape parameter decreases (or frequency parameter increases), operational risk increases, but shape parameter exerts greater influence than frequency parameter. under this condition, shape parameter is the main factors to affect operational risk. i.e., loss severity distribution is the main management object. as we can see in the figure1, with μincreasing, the curve shifts to the right bottom of the picture and discriminant becomes larger, i.e., the difference between wξ e and μe becomes larger. therefore, shape parameter plays a significantly pivotal role in risk management. consequently, according to different operational risk condition, financial institutes are supposed to firstly judge different section of loss distribution and then find out the key management indicator by proposition 1.this proposition not only provides foundation for setting and revising management measures, but also can be used to check the effect of the measures. from equation(5)~(6),under the certain confidence level, the elasticity of )(αopvar changes as shape parameter and frequency parameter change, and therefore relative size of w eξ and μe varies. owing to this, the key management parameter of the operational risk can change from shape parameter to frequency parameter or change from frequency parameter to shape parameter. further on, we will go ahead to introduce proposition 2 to discuss the changing rule between w eξ and μe . proposition 2 given previous assumptions, following rules can be found. (i) the change direction of the operational risk sensitivity. given 1 1 ln ≥ − α μ , then 0≥ ∂ ∂ w ξ ξ e w . given 1 1 ln0 < − < α μ , then 0< ∂ ∂ w ξ ξ e w ; 0< ∂ ∂ μ e wξ ; 0< ∂ ∂ w μe ξ , 0< ∂ ∂ μ eμ ; (ii) the change degree of the operational risk sensitivity： ⑴ given 1) 1 ln(ln 1 ln ≥ −− − α μ α μ ξ μ w , then μ e ξ e ww ξ w ξ ∂ ∂ ≥ ∂ ∂ , otherwise, μ e ξ e ww ξ w ξ ∂ ∂ < ∂ ∂ ; ⑵ given 1 1 ln ≥ − α μ ξ μ w , then μ e ξ e μ w μ ∂ ∂ ≥ ∂ ∂ , otherwise, μ e ξ e μ w μ ∂ ∂ < ∂ ∂ ; ⑶ given 1) 1 ln(ln 1 ln≥ −− α μ α μ , then w μ w ξ ξ e ξ e w ∂ ∂ ≥ ∂ ∂ ,otherwise, w μ w ξ ξ e ξ e w ∂ ∂ < ∂ ∂ ; ⑷ given 1 1 ln ≥ − α μ ,then μ e μ e μξw ∂ ∂ ≥ ∂ ∂ , otherwise, μ e μ e μξw ∂ ∂ < ∂ ∂ .(see proof in appendix b) from proposition2(i), as frequency parameter increases, w eξ and μe decreases; as shape parameter increases, μe decreases, but w eξ is uncertain which depends on different operational risk state. specifically, given 1 1 ln ≥ − α μ , w eξ increases as frequency parameter published by atlantis press copyright: the authors 127 m.l. he et al. / heavy-tailed operational risk measurement model increases; given 1 1 ln0 < − < α μ , w eξ decreases. the different influence degree of shape parameter and frequency parameter to w eξ and μe can be determined by four determinants, and therefore we can identify the most important characteristic parameter. according to proposition 2(ii), let 1) 1 ln(ln 1 ln = −− α μ α μ ξ μ w , 1 1 ln = − α μ ξ μ w .when a is set at 99.9%, we can get figure2. the point on the curve l1 denotes wξ and μ have the same effect on w eξ . left region of l1 denotes wξ have less impact than μ , right region of l1 denotes conversely. the point on the curve l2 denotes wξ and μ have the same effect on μe . left region of l2 denotes wξ have less impact than μ , while right region of l1 denotes conversely. the curve l1 intersects the curve l2 at(0.0151,0.041). this point of intersection denotes wξ and μ have the same effect on w eξ and μe . obviously,l1 and l2 divide coordinate plane into four sections і、п、ш and ⅳ, different region denoting different state of the operational risk. likewise, according to proposition 2(ii) , let ) 1 ln(ln 1 ln1 α μ α μ e −− =∆ , α μ e − =∆ 1 ln2 when setting %9.99=a ,we can get figure3.curve 1l′ denotes the different degree which wξ exerts on w eξ and μe . 0058.0=μ , 11 =∆e imply the influence degree is the same. therefore, 1l′ divide coordinate plane into two different regions: 1 1e∆ > and 1 1e∆ < .curve 2l′ denotes the different degree which μ exerts on w eξ and μe . 0027.0=μ 12 =∆e imply the influence degree is the same. therefore 2l′ divides coordinate plane into two different regions: 2 1e∆ > and 2 1e∆ < . fig.2. different degree of influence 1 fig.3. different degree of influence 2 given 1 1 > − α μ , then 001.0>μ .the figure2 and figure 3 both regard 001.0=μ as the starting point of abscissa. key influential indicator for operational risk is different due to different position of wξ and μ . in region і, μ e ξ e ww ξ w ξ ∂ ∂ > ∂ ∂ and μ e ξ e μ w μ ∂ ∂ < ∂ ∂ manifest wξ is the key influential parameter for w eξ and μ is the key influential parameter for μe .since in region і, 0151.0>μ .also note published by atlantis press copyright: the authors 128 m.l. he et al. / heavy-tailed operational risk measurement model figure3: w μ w ξ ξ e ξ e w ∂ ∂ > ∂ ∂ and μ e μ e μξw ∂ ∂ > ∂ ∂ . consequently, we have w μμξ w ξ ξ ee μ e ξ e ww ∂ ∂ > ∂ ∂ > ∂ ∂ > ∂ ∂ µ . wξ have the biggest influence on w eξ . in region п, μ e ξ e ww ξ w ξ ∂ ∂ < ∂ ∂ and μ e ξ e μ w μ ∂ ∂ < ∂ ∂ manifest μ is the key influential indicator for w eξ and μe , but the influential degree is diverse. also note 2l′ in figure 3: suppose 0027.0001.0 << μ , we have μ e μ e μξw ∂ ∂ < ∂ ∂ . that is μ has little effect on w eξ than μe .as μ decreases, the diversity increases. given 0027.0>μ , if μ has the greater effect on w eξ than μe .as μ increases, the diversity increases. in region ⅲ , μ e ξ e ww ξ w ξ ∂ ∂ < ∂ ∂ and μ e ξ e μ w μ ∂ ∂ > ∂ ∂ manifest μ is the key influential indicator for μe and wξ is the key influential indicator for w eξ .since in region ⅲ we have 0151.0<μ .also note figure3: given 0027.0001.0 << μ , then w μ w ξ ξ e ξ e w ∂ ∂ < ∂ ∂ and μ e μ e μξw ∂ ∂ < ∂ ∂ . consequently we deduce w ξξμ w μ ξ e μ e μ e ξ e ww ∂ ∂ > ∂ ∂ > ∂ ∂ > ∂ ∂ ,i.e., wξ has the biggest influence on μe ;given 0058.00027.0 << μ , then w μ w ξ ξ e ξ e w ∂ ∂ < ∂ ∂ and μ e μ e μξw ∂ ∂ > ∂ ∂ μ e μ e μξw ∂ ∂ < ∂ ∂ .consequently we deduce ( )wξ μ μ w e e e or μ ξ μ ∂ ∂ ∂ > ∂ ∂ ∂ and ( )w wξ ξμ w w e ee or μ ξ ξ ∂ ∂∂ > ∂ ∂ ∂ . all conclusions manifest that the effect one parameter exerts on the elasticity of another parameter is always greater than the effect that one parameter exerts on its own. given 0151.00058.0 << μ , then w μ w ξ ξ e ξ e w ∂ ∂ > ∂ ∂ and μ e μ e μξw ∂ ∂ > ∂ ∂ .consequently, we deduce μ e ξ e ξ e μ e μ w μ w ξξ ww ∂ ∂ > ∂ ∂ > ∂ ∂ > ∂ ∂ ,i.e., μ has the greatest effect on w eξ . in region ⅳ , μ e ξ e ww ξ w ξ ∂ ∂ > ∂ ∂ and μ e ξ e μ w μ ∂ ∂ > ∂ ∂ manifest wξ is the key influential indicator for w eξ and μe , but the influence degree is diverse. note curve 1l′ in figure 3: given 0058.0001.0 << μ , then w μ w ξ ξ e ξ e w ∂ ∂ < ∂ ∂ . wξ has little effect on weξ than μe . as μ decreases, the diversity becomes larger. given w μ w ξ ξ e ξ e w ∂ ∂ > ∂ ∂ . wξ has greater effect on w eξ than μe . as μ increases, the diversity becomes larger. from the above analysis, if we know the position of the characteristic parameter, we can distinguish key influential factors for w eξ and μe according to proposition 2. normally we can predict and supervise the possible variation of the key influential parameters. discrimination model of parameter management for operational risk is constructed by proposition1 and proposition2. for one thing, the model can identify the key influential factor .for another thing, it can predict and published by atlantis press copyright: the authors 129 m.l. he et al. / heavy-tailed operational risk measurement model supervise the possible variation of the key influential parameters. 3.2 numerical analysis from the logic process for establishing theoretical model, we know that such model are satisfied under certain condition. we need to check whether the state of the operational risk matches condition before we apply it: first, operational loss severity distribution must be heavy-tailed distribution. when using gev to comply with loss severity distribution, it is eligible to choose weibull distribution. if the loss distribution is not heavy-tailed, we can not use the model. second, after calculating characteristic parameters for loss severity distribution and loss frequency distribution, we should judge whether 1)1( >− αμ is tenable or not. if the above inequality is satisfies, then we can apply the model, otherwise we cannot use it. up to now, we do not have professional open database for operational risk to research, due to the confidentiality of the operational loss. therefore, our data using in empirical analysis is obtained by previous literatures which provide the characteristic parameters for us to test the model. literature 10 classifies operational loss according to the standard set by the new basel: damage to physical assets (dpa),internal fraud(if), employment practices and workplace safety(epws),clients, products and business practices (cpbp), and execution fraud (ef), delivery and process management (edpm), and external fraud (ef).the literature uses weibull distribution to comply with operational loss severity distribution and poisson distribution to comply with operational loss frequency distribution. the obtained characteristic parameters for operational loss distribution are presented in chart 1 as follows. weibull distribution is heavy-tailed distribution. when setting %9.99=a , 1)1( >− αμ satisfies (as chart1 denotes). therefore, this case satisfies the condition that model requires and we can use this model to identify the key management parameters for bank. in the following, we set %9.99=a and use the data obtained from wellbull distribution in literature 10 to verify the theory model. first, test and verify proposition 1. according to empirical fitting results in literature 10, from proposition1 we can get table1. table 1. the sensitivity of )(αopvar business line μ wξ α μ −1 e∆ if 0.52 0.59 515.9 -11.44 dpa 0.3376 0.52 337.6 -10.26 epws 0.62 0.57 617.0 -11.95 cpbp 3.24 1.3e-6 3242.0 -16.89 edpm 9.85 3.47e-7 9853.5 -20.40 ef 141.26 0.82 141261.1 -29.33 according to table1, from proposition 1 we know: (1) the variation direction of the operational var. since 0<∆e , then 0< wξ e .specifically, as shape parameter increases, operational risk value decreases. (2) the key influential parameters for operational var. discriminant is much greater than1.consequently, shape parameter exerts much bigger influence on operational risk than frequency parameter. criterion discriminant e∆ denotes different influence degree of two parameters for operational risk. in six kinds of operational loss, discriminant of ef is the biggest one, denoting the diversity of influence degree between two parameters is the greatest. however, the discriminant of dpa is the minimum, denoting the diversity of influence degree between two parameters is smallest. that conclusion can be obtained from figure 1 straightforwardly: given e∆ is greatly than 1, published by atlantis press copyright: the authors 130 m.l. he et al. / heavy-tailed operational risk measurement model it is the loss type ef that locates in the bottom right of the figure, manifesting the diversity of the two characteristic parameter influential degree is the greatest for such loss distribution. it is the loss type dpa that locates in the top left of the figure, manifesting the diversity of the two characteristic parameter influential degree is the smallest. next, test and verify proposition2. from table 2 we know, table 2. the discriminant of sensitivity formula business line )1ln(ln1ln α μ α μ ξ μ w −− α μ ξ μ w −1 ln ) 1 ln(ln 1 lnα μ α μ −− α μ −1 ln if 10.0 5.46 11.44 6.25 dpa 6.70 3.78 10.26 5.82 epws 12.9 6.95 11.95 6.42 cpbp 4.21e+07 2.02e+07 16.89 8.08 edpm 5.79e+08 2.61e+08 20.40 9.20 ef 5.05e+03 2.04e+03 29.33 11.86 (1) the variation direction of operational risk sensitivity. since 1)]1/(ln[ >− αμ , then 0>∂∂ wξ ξe w . specifically, as shape parameter increases, operational risk sensitivity increases (2) the comparison of changing degree of operational risk sensitivity. all criterion determinants are greater than 1, we can get μ e ξ e ww ξ w ξ ∂ ∂ > ∂ ∂ and μ e ξ e μ w μ ∂ ∂ > ∂ ∂ . it is obvious that shape parameter is the key influential parameter for w eξ and μe .but the influence degree is different to a certain degree. specifically, since 11 >∆e w μ w ξ ξ e ξ e w ∂ ∂ > ∂ ∂ , we can deduce w μ w ξ ξ e ξ e w ∂ ∂ > ∂ ∂ .that is to say, shape parameter exerts greater influence on weξ than μe . what’s more, as frequency parameter increases, 1e∆ increases and diversity increases. that conclusion can also be obtained from figure 2 and figure 3: in figure2, the coordinate of characteristic parameter for loss distribution locates region iv. in figure 3, the coordinate of characteristic parameter for loss distribution locates where 0058.0>μ . consequently, shape parameter is the key influence parameter for operational risk sensitivity. although 0<∂∂ wμe ξ , we still have 0>∂∂ wξ ξe w . consequently, as shape parameter increases, operational risk sensitivity increases. from the above illustrations, we know shape parameter is the key influential parameter for operational risk as well as for operational risk sensitivity: as shape parameter increases, )(αopvar at risk decreases. and decreasing velocity increases. this reflects that operational risk is increasingly sensitive relating to the change of the shape parameters and shape parameter becomes the key operational risk management parameter. therefore, in this bank, in order to reduce operational risk, proper measures should be organized and revised to amplify shape parameter. specifically, management measures like speed up rapid reaction ability, business continuity are expected. in different business line, the influence degree of shape parameter is different. therefore, the intensity of management measures should be diverse owing to this diversity. to be specific, for business lines which have strong impact on operational risk, management measures should be reinforced extensively. otherwise, the intensity of management measures should be published by atlantis press copyright: the authors 131 m.l. he et al. / heavy-tailed operational risk measurement model little. the above examples illustrate that the discrimination model of key management parameters for operational risk is validated and effective. the establish of the discrimination model offers a methodology to identify crucial management parameters. it combines measurement model and management model and make it possible for the integration of the two models. then operational risk management frameworks to be a complete system. first, from measurement model, we identify key management parameters. next, input the parameters into management model as management object to come up with measures to manage risks. then, judge the validity of the management measures by discrimination model as the measures have been enforced for some time: if characteristic parameters of loss distribution changes, the management measures are effective, otherwise, the management measure are invalid. if management measure are effective, we go ahead and analyze whether we achieve our goal, where the error comes from and slightly adjust our management measures. if management measures are invalid, we should find out the reason and amend the management measures. finally, take revised measures into the management model again. it is likely to integrate measurement model with management model into an operational risk management framework, forming a dynamic system. therefore, we are able to manage and monitor operational risk dynamically and consequently effectively curb operational risk. 4. concluding remarks the existing empirical analysis manifests that operational risk is heavy-tailed. when using generalized extreme value distribution to fit loss severity distribution, weibull distribution is a good choice to comply with some kind of operational loss. based on this, after deducing analytical solution of heavy-tailed operational risk, our article theoretically analyzes sensitivity of )(αopvar and the discrimination model which combines measurement model and management model. moreover we use examples to illustrate the validity of the model. the establish of the model offers methodology for supervising the effect on the management measures. further, our theory model combines measurement model and management model, making operational risk management model into a complete framework and forming an dynamic management system of the operational risk. our research further improves the application of the loss distribution approach in operational risk measurement and management. reference r. kennet. how to introduce an effective operational risk management framework. in robert hübner (ed.) operational risk: firm-wide issues for financial institutions[m]. london: a division of risk waters group ltd., 2003:73-92. a. peccia. using operational risk models to manage operational risk. in c. alexander (ed.), operational risk: regulation, analysis and management[m]. london: pearson education limited,2003:262-284. basel committee on banking supervision. international convergence of capital measurement and capital standards: a revised framework[s], bank for international settlements, 2004.6. c. alexander. bayesian methods for measuring operational risk[r]. icma centre discussion papers in finance icma, henley business school, reading university. 2000. c. alexander. managing operational risks with bayesian networks. in c. alexander (ed.), operational risk: regulation, analysis and management[m]. london: pearson education limited, 2003:285-295. r. g. cowell, r. j.verrall and y. k. yoon. modelling operational risk with beyesian networks[j]. journal of risk and insurance, 2007,74(4):795-827. m. neil, n. e. fenton and m. tailor. using bayesian networks to model expected and unexpected operational losses[j]. risk analysis, 2005, 25(4): 1539-6924. l. dalla valle and p. giudicib. a bayesian approach to estimate the marginal loss distributions in operational risk management[j]. computational statistics & data analysis, 2008,52(6): 3107-3127. shi dao-ji. applied method of extremes statistics[m]. tianjin: tianjin science and technology press, published by atlantis press copyright: the authors 132 m.l. he et al. / heavy-tailed operational risk measurement model 2006. dionne and h. dahen. what about underevaluating operational value at risk in the banking sector?(c), the 6th annual premier global event on erm, 2008. a. chapelle, y. crama, g. hübner and j.-p. peters. practical methods for measuring and managing operational risk in the financial sector: a clinical study [j], journal of banking & finance. 2008,32(6):1049-1061. j.d. opdyke, datamineit, llc. estimating operational risk capital with greater accuracy, precision, and robustness[j].journal of operational risk , june 2014 ,9(4) k. bocker and c. kläuppelberg. operational var: a closed-form approximation [j], risk of london,2005,18(12):90-93. k. bocker and j. sprittulla. operational var: meaningful means[j]. risk of london,2006,19(12):96-98. k. bocker. operational risk analytical results when high-severity losses follow a generalized pareto distribution (gpd)[j]. risk of london, 2006,8(4):117-120. ames mark, schuermann til, scott. hal s. bank capital for operational risk: a tale of fragility and instability[j]. journal of risk management in financial institutions 2015,8(3) pp. 227-243(17) mo jian-ming and zhou zong-fang. confidence interval and sensitivity of the operational var in lda[j], systems engineering, 2007,25(10): 33-39. s é verine plunus, georges h ü bner, jean-philippe peters.measuring operational risk in financial institutions[j]. applied financial economics ,september2012.22(18):1553-1569 appendix： a：proof of proposition1： proof for(i),we know 0) 1 (ln 1 ≥ − w α μ ξ and 0>wξ .when 1 1 ln0 ≤ − < α μ , 0≥ wξ e ,when 1 1 ln > − α μ , 0< wξ e .since 0 1 ln ≥ − α μ an d 0>wξ ,then 0) 1 ln( 1 > − = − α μ ξe wμ . for(ii),from equation(5) we know, −∞= − −= − →→ ]) 1 (lnln[limlim 1 11 w w α μ eξ ξ aa .from equation(6) we know, 0) 1 ln(limlim 1 11 = − = − →→ α μ ξe wμ aa . for(iii), since ) 1 (lnln 1 ln ) 1 ln( ) 1 (lnln 1 1 α μ α μ α μ ξ α μ e e w μ ξ w w −− −= − − − = − −ξ .when 1)1ln(ln1ln ≥−− α μ α μ , μξ ee w ≥ , otherwise, μξ ee w < . b：proof of proposition2： proof firstly we prove proposition(i).from equation(5) we know that： ) 1 (lnln2 α μ ξ e w w ξw − = ∂ ∂ −ξ (7) 1) 1 ln( − − − ∂ ∂ = α μ μ μ e w ξw ξ (8) since 0>wξ , 01 ln ≥ − α μ ,and 0>μ ,from equation (7) we know,when 1 1 ln ≥ − α μ , 0≥ ∂ ∂ w ξ ξ e w .when 1 1 ln0 < − < α μ , 0< ∂ ∂ w ξ ξ e w .from equation(8)we know 0< ∂ ∂ μ e wξ . from equation(6) we have: published by atlantis press copyright: the authors 133 http://www.ingentaconnect.com/content/hsp/jrmfi http://www.ingentaconnect.com/content/hsp/jrmfi https://www.researchgate.net/profile/severine_plunus https://www.researchgate.net/profile/georges_huebner https://www.researchgate.net/researcher/2026659025_jean-philippe_peters https://www.researchgate.net/researcher/2026659025_jean-philippe_peters https://www.researchgate.net/journal/0960-3107_applied_financial_economics https://www.researchgate.net/journal/0960-3107_applied_financial_economics m.l. he et al. / heavy-tailed operational risk measurement model 12 ) 1 (ln −− −∂ ∂ −= α μ ξ e w w μ ξ (9) 211 ) 1 (ln −−− − −= ∂ ∂ α μ μξw μ μ e (10) since 0>wξ , 01 ln ≥ − α μ , 0>μ ,from equation (10) we know 0< ∂ ∂ w μe ξ . for proposition (ii), from ) 1 ln(ln 1 ln α μ α μ ξ μ μ ee w ξ w ξ ww −− −= ∂ ∂ ∂ ∂ ξ we know that given 1) 1 ln(ln 1 ln ≥ −− − α μ α μ ξ μ w , we have μ e ξ e ww ξ w ξ ∂ ∂ ≥ ∂ ∂ , otherwise, μ e ξ e ww ξ w ξ ∂ ∂ < ∂ ∂ . from α μ ξ μ μ e ξ e w μ w μ − = ∂ ∂ ∂ ∂ 1 ln we know that, since 1 1 ln ≥ − α μ ξ μ w , μ e ξ e μ w μ ∂ ∂ ≥ ∂ ∂ , otherwise, μ e ξ e μ w μ ∂ ∂ < ∂ ∂ . from ) 1 ln(ln 1 ln α μ α μ ξ ee w μ w ξw −− −= ∂ ∂ ∂ ∂ ξ we know that given 1) 1 ln(ln 1 ln≥ −− α μ α μ , w μ w ξ ξ e ξ e w ∂ ∂ ≥ ∂ ∂ , otherwise, w μ w ξ ξ e ξ e w ∂ ∂ < ∂ ∂ . from α μ μ e μ e μξw − = ∂ ∂ ∂ ∂ 1 ln we know that given 1 1 ln ≥ − α μ , μ e μ e μξw ∂ ∂ ≥ ∂ ∂ ,otherwise, μ e μ e μξw ∂ ∂ < ∂ ∂ . published by atlantis press copyright: the authors 134 1. introduction 2. measurement of operational var 3. connection parameters of measurement model and management model 3.1 theoretical model 3.2 numerical analysis 4. concluding remarks reference << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <feff04180437043f043e043b043704320430043904420435002004420435043704380020043d0430044104420440043e0439043a0438002c00200437043000200434043000200441044a0437043404300432043004420435002000410064006f00620065002000500044004600200434043e043a0443043c0435043d04420438002c0020043c0430043a04410438043c0430043b043d043e0020043f044004380433043e04340435043d04380020043704300020043204380441043e043a043e043a0430044704350441044204320435043d0020043f04350447043004420020043704300020043f044004350434043f0435044704300442043d04300020043f043e04340433043e0442043e0432043a0430002e002000200421044a04370434043004340435043d043804420435002000500044004600200434043e043a0443043c0435043d044204380020043c043e0433043004420020043404300020044104350020043e0442043204300440044f0442002004410020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200441043b0435043404320430044904380020043204350440044104380438002e> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <feff04180441043f043e043b044c04370443043904420435002004340430043d043d044b04350020043d0430044104420440043e0439043a043800200434043b044f00200441043e043704340430043d0438044f00200434043e043a0443043c0435043d0442043e0432002000410064006f006200650020005000440046002c0020043c0430043a04410438043c0430043b044c043d043e0020043f043e04340445043e0434044f04490438044500200434043b044f00200432044b0441043e043a043e043a0430044704350441044204320435043d043d043e0433043e00200434043e043f0435044704300442043d043e0433043e00200432044b0432043e04340430002e002000200421043e043704340430043d043d044b04350020005000440046002d0434043e043a0443043c0435043d0442044b0020043c043e0436043d043e0020043e0442043a0440044b043204300442044c002004410020043f043e043c043e0449044c044e0020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200431043e043b043504350020043f043e04370434043d043804450020043204350440044104380439002e> /sky <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> /slv <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> /suo <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> /sve <feff0041006e007600e4006e00640020006400650020006800e4007200200069006e0073007400e4006c006c006e0069006e006700610072006e00610020006f006d002000640075002000760069006c006c00200073006b006100700061002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400200073006f006d002000e400720020006c00e4006d0070006c0069006700610020006600f60072002000700072006500700072006500730073002d007500740073006b00720069006600740020006d006500640020006800f600670020006b00760061006c0069007400650074002e002000200053006b006100700061006400650020005000440046002d0064006f006b0075006d0065006e00740020006b0061006e002000f600700070006e00610073002000690020004100630072006f0062006100740020006f00630068002000410064006f00620065002000520065006100640065007200200035002e00300020006f00630068002000730065006e006100720065002e> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice 渤海“暖式切变线”引发强对流天气成因分析 received 7 july 2016 accepted 10 august 2016 a case study of the severe convective in bohai sea and the establishment of early warning index on the sightseeing boat yanjun zhou, chengyu yan, yanjiang li the ocean observatory of qinhuangdao, qinhuangdao 066000，china abstract using the synchronous data such as weather radar, oceanic wrf model etc., a severe convective weather process has been analyzed on august 31, 2015 in bohai sea area, china. the results show that the shear line moved eastward and enhanced at the 925hpa and 850hpa in bohai bay from 08:00 to 20:00, the index k was 35 °c, index si was -1.81 °c, the cape was 166 j/kg and the vertical wind shear was 16m•s-1 in the stratification curve at 08:00 a.m, because of this background, the severe convective weather occurred in tianjin. the shear line moved northward at the 925hpa of the lower lay from 20:00 p.m on august 31to 08:00 a.m on september 1, the convective instability energy increased, several severe convective storm cell of meso-γscale and meso-βscale near the shear line have been induced in the central area of bohai sea, the new round of the was caused by a series of effecting of radar echo. severe weather process cape was released by the strong vertical movement and water vapor convergence of the boundary layer, the cut-in of weaker dry and cold of δθse (850-500 hpa) corresponded with the happening and impact area of the convection system in north of liaodong bay. and this paper established the early warning model about the severe convective weather for the sightseeing boat in bohai sea. keywords: bohai sea, severe convection, ship early warning 渤海强对流天气个例与游船预警指标初建 周艳军, 燕成玉, 李延江 秦皇岛海洋气象台，秦皇岛 066000，中国 摘要: 利用多部天气雷达、海洋 wrf 模式等同步资料，对 2015 年 8 月 31 日渤海一 次强对流天气漏报过程进行综合分析，结果表明：08 时-20 时渤海湾一线 925-850 百帕切 变线东移增强，08 时临近探空 k 指数 35℃、si 指数-1.81℃及 cape 为 166j/kg，垂直风 切变 16m•s-1，导致天津一线的强对流天气发生。20-02 时其前部低层 925 百帕“切变线“北 抬，对流性不稳定能量增强，在渤海中部的切变线附近诱发多个中γ、β尺度强对流风暴 单体，在雷达回波“列车效应“下，造成了秦皇岛近海新一轮强对流天气过程。强垂直运动 和边界层水汽辐合触发 cape 的释放。在辽东湾北部 δθse(850-500hpa) 较弱干冷切入与 对流系统的发生、落区有一定对应关系。初步建立渤海旅游船舶强对流天气预警模型。 关键词: 渤海, 强对流, 船舶预警 在大气西风带环流背景下，大多数渤海强 对流天气系统是由陆地西部或北部生成后移 向海区的，入海后稳定少变，在东部或南部海 岸带重新发展。由渤海区域生成的强对流系统 journal of risk analysis and crisis response, vol. 6, no. 3 (october 2016), 156-162 published by atlantis press copyright: the authors 156 y.j. zhou et al. / a case study of the severe convective in bohai sea 并由东南方向移向西北方向的个例较少，是目 前海上船舶强对流预警技术难点问题之一。近 年来，环渤海地区广大科技工作者克服海洋气 象灾害实况资料匮乏的不利条件，进行了一系 列科研探索，卢焕珍等人借助塘沽天气雷达和 自动站对产生海岸带雷暴大风的对流风暴进 行个例分析，总结出雷暴大风雷达回波的主要 形态、大风来临前径向速度场辐合特征等；王 彦 [1] 贺靓 [2] 、郭庆利 [3] 孙素琴 [4] 应用卫星资料、 多普勒雷达、闪电等监测资料对近年来海岸带 -近海发生强对流天气个例进行分析，得出强 对流系统中地闪、云顶亮温与雷达回波关系及 “列车效应”引发短时暴雨成因等；郭树军 [5] 王伟 [6] 从利用气象灾害风险评估方法和不完 全信息下的冰雹风险评价研究，给出陆地灾 害预警指标。上述分析均以单部雷达监测分 析为主，对回波源、发生、发展及演变规律的 描述存在着一定的局限性，对该海区船舶预警 研究甚少。 2015 年 8 月 31 日 20 时-9 月 1 日 04 时， 渤海西海岸带天津至-秦皇岛渤地区出现了由 强对流天气引发的短时暴雨、大风及持续性雷 电天气，其中，大风区域出现在山海关海区， 风速最大为 21.8m·s-1；短时暴雨分别出现在 天津和秦皇岛，小时雨量最大达 40mm；雷电 过程持续时间达 6h 以上。较强的强对流天气 系统对区域经济造成较大影响，直接威胁旅游 船舶航行安全，尤其是 2015 年 6 月 1 日长江 监利水域“东方之星“强对流天气引发重大沉 船事故之后，2016 四川广元又发生强对流天 气引发游船翻沉重大事故，游船安全航行成为 社会焦点之一。因此，加快研究海上游船强对 流预警试验研究十分迫切。入选强对流天气个 例，匹配河北海事局海难事故资料、新建海岛 站、浮标、探空及 ec 细网格、海洋 wrf 模 式产品等同步资料，分析入海前后的强对流中 尺度系统空间结构特征、边界层辐合线演变及 落区对应关系；建立强对流天气概念模型，归 纳 6-8 月近海 1-3 小时预警指标。为渤海旅游 船舶强对流天气预警技术研究提供一些新的 思路。 1. 天气背景与实况 渤海位于中高纬度之间，受高层大气西风 带影响较大，大多数不同尺度的强对流天气系 统是由陆地移向海区的，并在不同的海岸带区 域获取新的能量，重新加强发展，影响新的海 岸带直至内陆区域，它的生、消及演变规律仅 是天气尺度背景下的一个中尺度天气过程。 1.1 天气背景 2015 年 8 月 31 日 08 时（cst）500hpa 高空图上环渤海上空为弱冷空气配合的减弱 低槽，对应 08 时（cst）低层 925hpa 高空图 （图 1.a），渤海中南部为暖空气配合的“人字 型”切变线，其中渤海湾至莱州湾一线温度场 有明显暖湿平流，这种“上冷下暖”垂直结构 有利于对流天气的发生发展，在低层切变线动 力、水汽耦合下，给未来强对流天气强度加强 提供了必要条件。20 时低层 925hpa 高空图上 （图 1.b），渤海中南部“人字型”切变线原地 旋转，前部暖式切变线北抬至渤海中部，导致 新一轮强对流天气生成发展，中尺度系统沿东 南向西北部海岸带延伸发展。从天气背景分析 明显看出：整个过程分为两个时段，即 31 日 08 时-20 时“人字型”切变线造成天津海岸带 强对流天气，31 日 20 时-9 月 1 日 02 时“人 字型”切变线前部暖式切变为秦皇岛海岸带强 对流天气主要影响系统。 图 1. (a) 2015 年 8 月 31 日 08 时（cst）; (b) 2015 年 8 月 31 日 20 时（cst）为 925hpa 风场（单 位 m••s-1）温度场（单位℃） published by atlantis press copyright: the authors 157 y.j. zhou et al. / a case study of the severe convective in bohai sea 1.2 强对流实况与探空 对应实况与探空可分为两个不同时空尺 度见（图 2）。第一阶段为 2015 年 8 月 31 日 08 时-20 时，天津至渤海湾出现了由强对流天 气引发的短时暴雨、持续性雷电天气，小时雨 量最大达 39mm；08 时临近探空（乐亭站）k 指数为 35℃、si 指数为-1.81℃及 cape 为 166j/kg，0-6km 垂直风切变 16m••s-1，达到一 般强对流系统发生、发展阈值或临界值。第二 阶段为 2015 年 8 月 31 日 20 时-9 月 1 日 02 时，秦皇岛-山海关出现海区强对流引发的短 时暴雨、大风及持续性雷电天气，大风区域出 现山海关海区，风速最大为 21.8m·s-1；雷电 过程持续时间达 6h 以上。20-02 时其前部低层 925-850 百帕切变线北抬，20 时 k 指数 32℃， si 指数为 1.36℃，cape 为 383 j/kg，垂直风 切变为 19m••s-1，对流性不稳定能量进一步增 强，海区冷暖空气交汇于渤海中部的切变线附 近，有利于秦皇岛海岸带新一轮强对流天气发 生。 2. 天气雷达监测-渤海中尺度雷达拼图 目前，在渤海海岸带新一代天气雷达有六 部，如上（图 2）秦皇岛、天津、沧州、烟台、 大连及营口等。单部雷达 0.5º仰角，理论上基 本反射率监测半径 230km，径向速度 115km， 由于天线高度不一致和地物遮挡，监测距离底 高在渤海中部区域大于 1.1km，实际综合监测 能力为 150km 左右。对强对流天气发生、发 展及演变规律研究有局限性，综合拼图与资料 共享的实现（互联网拼图 10 分钟一幅）提升 了中尺度系统的监测能力和技术水准。 2.1 天津一线强对流天气系统入海前后的 演变特征 对照上述同步实况及天气分析，第一轮强 对流发生时间段为 8 月 31 日 16-18 时（cst）， 见（图 3.a1、a2）在渤海湾一带生成的多单体 风暴稳定少动，雷达反射率因子强回波带水平 尺度 230×80km，强度 40-50dbz，移向 w→e 向，移速为 20→30km/h，对应天气短时强降 图 3. 2015 年 8 月 31 日 16 时-18 时（cst）六部天气雷达基本反射率拼图 图 2. 红色圆圈表示 14-20 时和 20-02 时（cst）强对流天气落区，△探空站位置。 published by atlantis press copyright: the authors 158 y.j. zhou et al. / a case study of the severe convective in bohai sea 水、雷电等；2 小时后东移至莱州湾维持发展， 强回波面积缩小，水平尺度 180×70km，强度 维持 40-50dbz，移向 wsw→ene 向，移速为 20→30km/h，对应天气短时强降水、雷电、大 风等；天津至渤海湾为一般阵雨天气。但从莱 州湾至渤海湾北部（图 3.a3）所谓“暖式切变 线”回波带已经初步形成，并缓慢向中部海区 推进。 2.2 秦皇岛一线强对流天气雷达回波“列 车效应” 初步生成的渤海南部中尺度强对流回波 带在 31 日 20 时-22 时继续北抬发展，见（图 4.b1）雷达反射率强回波带水平尺度 350× 80km，强度 40-55dbz，移向 ssw→nne 向， 移速为 30→40km/h，影响范围：莱州湾-唐山 海岸带，对应天气短时强降水、大风、雷电等； 在渤海中部的切变线附近诱发多个中γ、β尺 度强对流风暴单体（图 4 b2、b3），与主体回 波带移向成 90º交角，水平尺度 30×50km， 强度 50-55dbz，移向 se→nw 向，移速为 50 →60km/h，具有明显的向前线性传播特征，即 雷达回波“列车效应“。23 时-02 时渤海低层水 汽被暖式切变线西南气流源源不断地输送到 辐合线上空， 有利于中小尺度对流单体及其 次级垂直环流的维持、发展， 且这些次级环 流规则排列， 从而“列车效应”得以维持，造 成了秦皇岛近海新一轮强对流天气过程。 2.3 秦皇岛海区回波带消亡阶段演变特征 从（图 5.c1）中明显看出 03 时（cst） 中 部 海 区 回 波 带 出 现 断 裂 ， 强 度 下 降 至 30-40dbz，04 时趋于趋于消亡，为一般阵雨 天气。值得关注的是在渤海湾南部沧州海区有 新的强对流回波带生成，见（图 5.c2、c3） 水平尺度 100×40km，强度 45-50dbz，移向 n→s 向，移速为 35→40km/h；整体中尺度回 波具有螺旋状特点，相对秦皇岛海区处于消亡 阶段，进入渤海湾的后部下沉气流垂直于沧州 海海岸带，摩擦辐合是对流天气发生的主要动 力因素，由于尺度偏小，所以持续时间不足 2 小时。 3. 中尺度海洋气象 wrf 模式物理量分 析 雷达监测分析结论为暖式切变线回波带 北抬，导致秦皇岛海区第二次强对流天气发 生。实际预报是一次漏报过程，见下图（图 6 左）强对流落区（棕黄色）没有报出渤海中部 至秦皇岛海岸带（红色圆圈区域），强天气中 尺度环境场分析过程中（图 6 右），主要影响 系统移动方向为东移山东半岛一线（切变线为 蓝色点划线），绥中-秦皇岛-唐山一带没有给出 有利于强对流天气动力和水汽条件，同时说明 此次过程可预报性偏低，预报难度较大，常规 的 ec 和 t639 数值模式对海区中尺度强对流 预报能力有一定局限性。面对服务需求，如何 图 4.2015 年 8 月 31 日 20 时-22 时（cst）六部天气雷达基本反射率拼图 图 5. 2015 年 9 月 1 日 03 时-05 时（cst）六部天气雷达基本反射率拼图 published by atlantis press copyright: the authors 159 y.j. zhou et al. / a case study of the severe convective in bohai sea 降低海区强对流漏报？可否借鉴其他模式产 品来提升实际可预报性？通过 wrf 海洋气象 中尺度模式物理量场进行同步分析得到一些 有意义的结果。 3.1 海洋气象 wrf 模式 预报系统由中国科学院南海研究所提供 技术支持。采用 arw(the advanced research wrf)3.3 版本。按照 1:3 的比例，外区域采用 45km 的网格尺度，预报系统中区域采用 15km 的网格尺度，小区域采用 5km 的网格尺度。采 用二重嵌套，中区域为华北东部至黄渤海区 域，小区域为渤海。主要的参数化过程包含云 微物理过程、积云参数化、长波辐射、短波辐 射、边界层参数化、陆面过程参数化以及次网 格扩散等。 3.2 对流有效位能(cape)3 个时次对比分 析 提高对流天气临近预报准确率的关键问 题是了解大气的垂直稳定度、垂直风切变及水 汽条件。对流有效位能(cape)是评价大气对流 潜力标准的不稳定指数。海洋气象 wrf 中尺度 数值模式产品提供了高时空分辨率的大气稳 定度等物理量信息，考虑了大气层结分布，能 够反映大气的不稳定度，被广泛用于诊断强对 流天气发生的可能性。 从（图 7）对流有效位能(cape)08 时初始 场运行结果明显看出不如 14 时初始场运行结 果，输出的（图 8）cape 大值区域比较接近于 雷达实况拼图；cape 的 20 时初始场运行结果 （图 9）的大值区域演变趋势更接近雷达实况 拼图，但“列车效应”中、小尺度系统没有模 拟出来，强度变化略显偏弱，从理论上讲 20 点初始场同化资料后，应更接近强对流发生的 物理量场和环流场，可能与渤海实况探空资料 同化内插值有关，对于渤海西路、北路入海强 对流系统各家数值预报之间的预报误差较小， 对于南来中尺度强对流误差偏大，海区中、小 尺度系统移向与高层大气引导气流相反，且持 续时间短，就此次过程，多家数值预报预报系 统移向为东至东南方向，而 wrf 能预报出北抬 的趋势实属不易；综合分析表明：强对流发生 前对流有效位能(cape)明显增大，逐小时强对 流回波带演变（图 4）和 925 hpa 切变线（图 1.b）与 cape 高值区较一致；对比中央气象台 20 时指导预报订正预报时效可后延 6-8 小时， 预期带来的海区服务效果为正值。14 时初始 场出图时间在 20 时，为最新的强对流天气订 正 08 时预报结论依据之一。 图 7. 2015 年 8 月 31 日 08 时（cst）初始场 wrf 模式模拟渤海区域 9 月 1 日 01 时对流有 效位能(cape) 图 8.2015 年 8 月 31 日 14 时（cst）初始场 wrf 模式模拟渤海区域 9 月 1 日 01 时对流有 效位能（cape） 图 6.2015 年 8 月 31 日 20 时（cst）中央气象台强对流预报与中尺度环境场分析，红圈为漏报区域 published by atlantis press copyright: the authors 160 y.j. zhou et al. / a case study of the severe convective in bohai sea 图 9.2015 年 8 月 31 日 20 时（cst）初始场 wrf 模式模拟渤海区域 9 月 1 日 01 时对流有 效位能(cape) 3.3 假相当位温 另外一个值得关注的热力因子为假相当 位温，是一种守恒的量，可以利用它来分析大 气的热力性质。大尺度背景下，假相当位温等 值线密集区附近有利于涡旋的发展，把大值区 和能量场空间分布作为强降水发生的必要条 件。在中尺度系统分析中，作为饱和气层静力 稳定度的基本判据。如（图 10）δθse 整个 空间 1000-300 hpa 垂直值域分布为低层大， 中间层小，高层大有利于对流天气发生发展， 从 20 时至 02 时中尺度低值气团（280ºk）前 沿由 124ºe 向西推进到 123ºe 附近，高度下降， 在辽东湾东北 δθse(925-700hpa) 较弱干冷 切入，相对于 39ºn 以南海区有利对流系统的 发生，而相对 39ºn 以北区域及落区不利于发 生发展。 4. 渤海旅游船舶强对流天气预警指标初 建 4.1 ec、wrf 模式海区强对流产品应用技 术 利用 ec 细网格、本地海洋气象与环境 wrf 模式应用技术，典型个例综合分析，采用备份 服务器回放技术，1 小时×5 公里格点资料 (0-24 小时预报场)：①强对流天气物理量剖 面（39n º，116-126e º）②0-24 小时降水+10 米风场③1000 百帕+10 米相对湿度等并提供 最大海浪监测与预测信息，建立 0-24 小时潜 势预报和 1-3 小时海区强对流预警指标。 4.2 秦皇岛海区强对流预警模型 1 小时预报预警：雷达反射率因子+雷电 监测+海岛站+浮标 2 小时预报预警：雷达反射率因子+雷达 拼图+海岛站+浮标+探空 3 小时预报预警：雷达反射率因子+卫星 云图 swap+wrf3 小时物理量 0-24 小 时 潜 势 预 报 ： ec+wrf 物 理 量 +wrfcape+中央台 spc 4.3 强对流发布方式及应用 通过海事值班通讯 vts 平台（船舶交通管 理系统）及微信方式，实时发布预警信息至船 长和值班经理。内容格式：中国气象局预警信 号标准模板或图片格式。2016 年 7-8 月共发 布秦皇岛近海强对流天气预警 5 次，其中，7 月 20 日、25 日在 ais 系统下，预报员用卫星 电话或手机直接与船长通话，王子号、公主号 及长城号三艘游轮均提前 1 小时安全返航。 5. 结论 海洋强对流天气是目前渤海旅游船舶预 警技术难点之一。大多数渤海强对流天气系统 是大气西风带环流背景下，由陆地西部或北部 生成后移向海区的，直接由渤海南部区域生成 的强对流系统移向西北方向的个例甚少。分析 结论如下： （1）环渤海地区天气尺度系统的有效配 置为不同时段的中尺度对流系统发展提供了 环流 背景 条件 。08 时-20 时渤海 湾一 线 925-850 百帕切变线东移增强，08 时临近探空 k 指数 35℃、si 指数-1.81℃及 cape 为 166j/kg，垂直风切变 16m••s -1 ，导致天津一线 的强对流天气发生。20-02 时其前部低层 925 百帕“暖式切变线”北抬，对流性不稳定能量 图 10.2015 年 8 月 31 日 14 时（cst）初始场 wrf 模式 39ºn 假相当位温剖面 a.20 时 b.23 时 c.02 时 published by atlantis press copyright: the authors 161 y.j. zhou et al. / a case study of the severe convective in bohai sea 增强，在渤海中部的切变线附近诱发多个中 γ、β尺度强对流风暴单体，在雷达回波“列 车效应”下，造成了秦皇岛近海新一轮强对流 天气过程。 （2）在渤海湾一带生成的多单体风暴稳 定少动，15 时雷达反射率因子强回波带水平 尺度 230×80km，强度 40-50dbz，移向 w→e 向，移速为 20→30km/h，对应天气短时强降 水、雷电等；2 小时后东移莱州湾维持发展。 20 时渤海南部中尺度强对流回波带北抬发展， 强 回 波 带 水 平 尺 度 350 × 80km ， 强 度 40-55dbz，移向 ssw→nne 向，移速为 30→ 40km/h，在渤海中部的切变线附近诱发多个中 γ、β尺度强对流风暴单体与主体回波带移向 成 90 º 交角， 水 平尺度 30 ×50km ，强度 50-55dbz，移向 se→nw 向，移速为 50→ 60km/h，具有明显的向前传播线性特征，即雷 达回波“列车效应”。03 时（cst）中部海区 回波带出现断裂，强度下降至 30-40dbz，04 时趋于消亡。 （3）通过典型个例分析初步建立预警模 型。1-3 小时预报预警：雷达反射率因子+雷 电监测+海岛站+浮标；0-12 小时潜势预报： ec+wrf 物理量+wrfcape+中央台 spc。在 2016 年 7-8 月实际应用中取得令人满意的效果。 致谢 项目资助:环渤海区域科技协同创新基金项目 （ 黄 渤 海 雷 暴 大 风 监 测 及 预 警 指 标 研 究 qyxm201502）。 参考文献 [1]王彦.风锋雷达回波特征分析[j].气象， 2006，32(12):23-28. [2]贺靓，于超，吕新民.渤海中南部海区一次 雷暴大风过程分析.海洋预报，2011， 28(1):19-24. [3]郭庆利，薛波，党英娜.渤海海峡雷雨大 风的多普勒雷达回波特征.海洋预报， 2011，28(1):13-18. [4]孙树琴.一次由“列车效应“引发的梅雨 锋暴雨研究.高原气象,2015, 34(1): 190 201. [5]s.j. guo, the meteorological disaster risk assessment based on the diffusion mechanism. journal of risk analysis and crisis response, 2012, 2(2): 124-130. [6]w. wang, c. zhou. the research of a hail risk evaluation under imperfect information. journal of risk analysis and crisis response, 2011, 1(2): 133-141. published by atlantis press copyright: the authors 162 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%ba%d8%f6%a61 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%ba%d8%f6%a61 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%c2%c0%d0%c2%c3%f13 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%c2%c0%d0%c2%c3%f13 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%b9%f9%c7%ec%c0%fb1 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%b9%f9%c7%ec%c0%fb1 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%d1%a6%fe%9f%b2%a82 http://www.hyyb.org.cn/magazinemenu/index.aspx?typeid=1&key=%d1%a6%fe%9f%b2%a82 1. 天气背景与实况 1.1天气背景 1.2强对流实况与探空 2. 天气雷达监测-渤海中尺度雷达拼图 2.1天津一线强对流天气系统入海前后的演变特征 2.2秦皇岛一线强对流天气雷达回波“列车效应” 2.3秦皇岛海区回波带消亡阶段演变特征 3. 中尺度海洋气象wrf模式物理量分析 3.1海洋气象wrf模式 3.2对流有效位能(cape)3个时次对比分析 3.3假相当位温 4. 渤海旅游船舶强对流天气预警指标初建 4.1 ec、wrf模式海区强对流产品应用技术 4.2秦皇岛海区强对流预警模型 4.3强对流发布方式及应用 5. 结论 致谢 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <feff005500740069006c0069006300650020006500730074006100200063006f006e0066006900670075007200610063006900f3006e0020007000610072006100200063007200650061007200200064006f00630075006d0065006e0074006f00730020005000440046002000640065002000410064006f0062006500200061006400650063007500610064006f00730020007000610072006100200069006d0070007200650073006900f3006e0020007000720065002d0065006400690074006f007200690061006c00200064006500200061006c00740061002000630061006c0069006400610064002e002000530065002000700075006500640065006e00200061006200720069007200200064006f00630075006d0065006e0074006f00730020005000440046002000630072006500610064006f007300200063006f006e0020004100630072006f006200610074002c002000410064006f00620065002000520065006100640065007200200035002e003000200079002000760065007200730069006f006e0065007300200070006f00730074006500720069006f007200650073002e> /eti <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> /fra <feff005500740069006c006900730065007a00200063006500730020006f007000740069006f006e00730020006100660069006e00200064006500200063007200e900650072002000640065007300200064006f00630075006d0065006e00740073002000410064006f00620065002000500044004600200070006f0075007200200075006e00650020007100750061006c0069007400e90020006400270069006d007000720065007300730069006f006e00200070007200e9007000720065007300730065002e0020004c0065007300200064006f00630075006d0065006e00740073002000500044004600200063007200e900e90073002000700065007500760065006e0074002000ea0074007200650020006f007500760065007200740073002000640061006e00730020004100630072006f006200610074002c002000610069006e00730069002000710075002700410064006f00620065002000520065006100640065007200200035002e0030002000650074002000760065007200730069006f006e007300200075006c007400e90072006900650075007200650073002e> /gre <feff03a703c103b703c303b903bc03bf03c003bf03b903ae03c303c403b5002003b103c503c403ad03c2002003c403b903c2002003c103c503b803bc03af03c303b503b903c2002003b303b903b1002003bd03b1002003b403b703bc03b903bf03c503c103b303ae03c303b503c403b5002003ad03b303b303c103b103c603b1002000410064006f006200650020005000440046002003c003bf03c5002003b503af03bd03b103b9002003ba03b103c42019002003b503be03bf03c703ae03bd002003ba03b103c403ac03bb03bb03b703bb03b1002003b303b903b1002003c003c103bf002d03b503ba03c403c503c003c903c403b903ba03ad03c2002003b503c103b303b103c303af03b503c2002003c503c803b703bb03ae03c2002003c003bf03b903cc03c403b703c403b103c2002e0020002003a403b10020005000440046002003ad03b303b303c103b103c603b1002003c003bf03c5002003ad03c703b503c403b5002003b403b703bc03b903bf03c503c103b303ae03c303b503b9002003bc03c003bf03c103bf03cd03bd002003bd03b1002003b103bd03bf03b903c703c403bf03cd03bd002003bc03b5002003c403bf0020004100630072006f006200610074002c002003c403bf002000410064006f00620065002000520065006100640065007200200035002e0030002003ba03b103b9002003bc03b503c403b103b303b503bd03ad03c303c403b503c103b503c2002003b503ba03b403cc03c303b503b903c2002e> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /pol <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> /ptb <feff005500740069006c0069007a006500200065007300730061007300200063006f006e00660069006700750072006100e700f50065007300200064006500200066006f0072006d00610020006100200063007200690061007200200064006f00630075006d0065006e0074006f0073002000410064006f0062006500200050004400460020006d00610069007300200061006400650071007500610064006f00730020007000610072006100200070007200e9002d0069006d0070007200650073007300f50065007300200064006500200061006c007400610020007100750061006c00690064006100640065002e0020004f007300200064006f00630075006d0065006e0074006f00730020005000440046002000630072006900610064006f007300200070006f00640065006d0020007300650072002000610062006500720074006f007300200063006f006d0020006f0020004100630072006f006200610074002000650020006f002000410064006f00620065002000520065006100640065007200200035002e0030002000650020007600650072007300f50065007300200070006f00730074006500720069006f007200650073002e> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice introduction for volume 10, issue 2 this issue contains 6 papers. there are 4 contributions in english and 2 contributions in chinese with english abstracts. the papers can be divided into four topics: covid-19, disaster psychological crisis, risk analysis and risk assessment in this issue, two papers are in covid-19 topic. the first paper “analysis of death risk of covid-19 under incomplete information” by chongfu huang, suggest a hybrid model to analyze the death risk under incomplete information. the hybrid model would be supported by the internet of intelligences, being a platform interacting with infectious disease specialists and local doctors who fuse the evidences with the experience of the known infectious diseases and provide a series of judgments related to the death risk of a human population in a given period to covid-19. the paper gives a virtual case to show how to use the hybrid model. the second paper “research on the prevention and control of covid-19 risk in long-term care facilities based on prospect theory ” by fengying zhang, et al., attempts to explore the uncertain relationship between the identification of the risk of covid-19 and the management and care behavior of the elderly care institutions, aiming to achieve management optimization. there is one paper in disaster psychological crisis. the paper “mutation mechanism of victim’s psychological behavior state after major natural disasters” by xuanhua xu, et al., analyzed the victim’s psychological and behavioral characteristics and influencing factors during recovery stage based on lots of literatures and an interview with victims. besides, a mechanism chart of the formation of the victim’s psychological behavior and a cusp catastrophe model are constructed. the model is partially verified by the investigation data of the disaster area. finally, the paper put forward some strategies to cope with victim’s psychology and behavior during recovery stage based on the cusp catastrophe model analysis. there are two papers in risk analysis. the first paper “patterns of fatal explosion accidents in china” by siheng sun, et al., explores the patterns of fatal explosion accidents (fea). statistical analysis was made based on the year, month, time interval of day, day of week, province, type of accident, place of occurrence and severity of accident by using the database containing 1,835 fea that happened in china during 2001-2018. the second paper “an analysis method of black swan event based on impact-spread tree”by xinying liu & hang li, solved the problem of whether a method can be used to quantitatively determine whether an event can be called a black swan event before it is defined as a black swan event. this paper proposes a set of analysis methods to determine the black swan event by constructing an impact-spread tree. the case study shows that the method of impact-spread tree can effectively distinguish between black swan events and non-black swan events. there is one paper in risk assessment, “a sales forecasting model for the consumer goods with holiday effects” by mu zhang, et al., which proposes a sales forecasting model for the consumer goods with holiday effects by introducing the seasonal decomposition and arima model. firstly, a dummy variable model is constructed to test the holiday effects in consumer goods market. secondly, using the seasonal decomposition, the seasonal factor is separated from the original series, and the seasonally adjusted series is then obtained. through the arima model, a trend forecast to the seasonally adjusted series is further carried out. finally, according to the multiplicative model, refilling the trend forecast value with the seasonal factor, thus, the final sales forecast results of the consumer goods with holiday effects can be obtained. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, being technology editor, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/) editors-in-chief: prof. chongfu huang beijing normal university, no.19 xinjiekouwai street, beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina, regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response vol. 10(2); july (2020), p. i doi: https://doi.org/10.2991/jracr.k.200720.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:hchongfu@126.com mailto:gordon.huang@uregina.ca https://doi.org/10.2991/jracr.k.200720.001 https://www.atlantis-press.com/journals/jracr%20 research framework for emergency rescue of disasters and crises zhi zhang, jianbo you, qiaoying li, xuan du the corps college of the chinese armed police force, beijing 102202，china e-mail: wjzhangzhi@139.com the emergency rescue refers to all the artificial interventions that aim to reduce the occurrence of and damages by the emergencies directly or indirectly. its role is to act immediately to prevent damage to life, property or environment. emergency rescue is the critical and direct stage in the emergency management process. disasters and crises have the nature of the sudden urgency, complexity, severity, and decide the characteristics of emergency rescue operations are urgent, difficult, dangerous and hard. the essay comes up with research framework of professional rescue: the emergency organization and command of multi-forces on strategy-battle-tactics level, the risk analysis of multi-disasters events and the risk management and control of rescue, the standardization of rescue skills and equipments and professional training, and the information platform combining application system, database and the preliminary analysis of main sources. keywords: emergency rescue, research framework, command, risk management 1. introduction after september 11 event in 2001, research in emergency rescue has attracted significant attention. the goal of research is to reduce the damages caused by emergencies and to sustain public safety. after the event of fight against sars in 2003, china has started to build the national emergency management system in an all-round way, with “one plan and three mechanisms” (compilation and revise of emergency plan, enhancement the construction of emergency management system, mechanism, and law) as the core [1]. in the face of highly complex trend of emergency events in modern society after the wenchuan earthquake in sichuan province in southwest china in 2008, the training of emergency rescue stuff has been given the priority. a magnitude 9.0 earthquake struck off the northeastern coast of japan on march 11, 2011, in which 15,854 people were killed, and another 3271 people missing. the knock-on effect caused by the earthquake, tsunami and nuclear accident has increased the difficulty of the task of the emergency rescue. however, it is somehow still difficult to provide a general view on the core scientific problems in the emergency rescue research due to its complexity. one common understanding is that emergency rescue research should include risk analysis and assessment, emergency response, operations command, rescue team organization and coordination, and recovery, etc, based on the development and changes of disasters and crises. although there are large numbers of researchers and topics in emergency rescue, there still exist three questions: how to determine the research framework according to the characteristics of emergency rescue? how to combine emergency management and emergency rescue operations effectively? what are the main directions and their contents in the research framework? the emergency and its response and management have typical distinction of complex systems just like the coupled human and natural systems [2]. it is a multi-subsystem, multi-level, multi-function system and the systems and factors involved are very often uncertain, non-linear, dynamic and open. thus, neither reductionism nor system approach is sufficient [3]. the methodology and approach are lack and pressed for. in this paper, the authors attempt to provide a description of framework for emergency rescue researches. journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 173-177 published by atlantis press copyright: the authors 173 administrateur texte tapé à la machine received 27 may 2012 administrateur texte tapé à la machine accepted 12 october 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine z. zhang, et al 2. emergency rescue concepts and characteristics 2.1. emergency rescue concepts the trigger condition of emergency rescue action is the occurrence of disasters and crises. according to the response law on emergency events of people’s republic of china issued and put in force in 2007, “emergency event” refers to “the bursting-out natural disasters, accident, public health incident and social security events that cause or could cause severe social damage and need to deal with by taking emergency measures.” emergency rescue refers to all human intervention to cut down the emergency events and mitigate their damage in a broad sense, and protective, controllable and aid action to the individuals and key targets in a narrow sense. emergency rescue mainly aims at natural disasters, accidents and public health incidents while the handling of which share similarities as centralized and unified actions could be taken in the aspect of human rescue, event control, public equipment and basic livelihood safeguard. emergency danger-elimination work in emergency rescue is also known as rescue and relief work. emergency rescue is still a feeble and emerging interdisciplinary subject. the theoretical research on emergency rescue of emergency events which is pioneering, basic and strategic, not only involves the nurturing of emergency rescue commanding and technical personnel, but also bears great significance in promoting the building of national emergency system. 2.2. the characteristics of emergency rescue operations disasters and crises have the nature of the sudden urgency, complexity, severity, and decide the characteristics of emergency rescue operations are urgent, difficult, dangerous and hard. 2.2.1 the situation is urgent the emergency events’ occurrence and development mechanism is a nonlinear complex system, including a variety of natural and man-made factors intertwined. even if you get a certain prediction early warning information, but disaster events’ occurrence time, place, harm degree, consequences, are difficult to judge accurately. disaster events occurred in recent years, showing a mass trend, the secondary events makes disaster events and derivative frequent more complex, resulting in harm to the rapid expansion and spread. after the burst of the emergency event if you do not respond promptly to their losses, over time, will be expanded dramatically, the situation is extremely urgent. for example, in sichuan wenchuan earthquake emergency rescue operations in 2008, the tangjiashan quake lake was formed after a massive quake-triggered landslide from tangjiashan mountain blocked the tongkou river, which ran through the beichuan county, one of the worst-hit areas in the quake. the swollen lake is the largest of more than 30 quake lakes in sichuan following the may 12 quake, posing a threat to 1.3 million people downstream. more than 250,000 people in low-lying areas in mianyang have been relocated under a plan based on the assumption that a third of the lake volume breached its banks. the situation was urgent without delay. 2.2.2 the action is difficult during the emergency rescue process, the natural environment and the human environment are often subjected to varying degrees of damage. earthquakes, typhoons, floods, snowstorms and other major disasters lead a direct result: communication, transportation, electricity are basic interrupt even paralysis in the disaster area, seriously restricting the commencement of forces to organize and command and rescue operations. major emergencies have integrated regional damage effects, the variety of the affected object to harmful levels of uneven performance of different type of disaster damage. especially in the large-scale implementation of the rescue operation, the rescue force often dispersed configuration to multiple rescue operations of the operating area is facing some limitations. the technical equipment and professionals will be faced with the difficult to meet the real needs of limitations. meanwhile, due to the long duration of the emergency rescue operations, as long as a few months, in the case of damage to the original social logistics system is affected by many factors interfere, rescue teams continued to self-supply more difficult. 2.2.3 the risk is intertwined the emergency rescue is a high-risk action, with the continued development of the emergency events, the secondary events and the derivative events continue to emerge, such as storms attacks, building collapse, burst dam failure, epidemic raging in storms and floods. the rescue work is facing a huge threat. for example, in the earthquake, relief and rescue teams published by atlantis press copyright: the authors 174 research framework for emergency rescue faced three threats. first, the aftershocks threat, housing is in an unstable state after the earthquake, rescue operators facing with the threat of secondary collapse. the second threat is artificially collapsed. after buildings collapsed, although the collapsed body formed stable combination structure relatively, but the rescue operation is easy to destroy the original structure and loss of balance, trapped personnel and rescue workers face buried pressure is dangerous. the third threat is the continued events, the earthquake triggered fires, floods, explosions, toxic gas leak, a serious threat to the trapped workers, rescue also increase the risk. 2.2.4 the task is hard crisis event usually means facing serious political and international issues, if disposed of improperly, will evolve into a national disaster. in many countries, the army has the advantages of a high degree of combat readiness, well-organized, quick response and tenacious style, has become the choice for emergency rescue force, army image is an important part of national image building. the forest fires of russia in the summer of 2010 exposed the existence of the problem of the russian troops to participate in emergency rescue operations. russian president dmitry medvedev inspected the burned of a naval base in the moscow outskirts, quite ironically said: “the fire, i ask our armed forces to assume the important task of protecting people, unfortunately, some military departments cannot protect themselves”. affected by the fires, russian government’s support rate hit a low record. thus, emergency rescue operations have special political significance. an important indicator to examine emergency rescue operation in handling emergency events, especially in face of urgent, dangerous and difficult rescue task, is whether the response of professional forces is scientific and efficient. 3. framework of emergency rescue research the theoretic research framework on emergency rescue of emergency events includes: the organization and command of multi-forces of emergency rescue, the risk management and control of multi-disasters events, the standardization of rescue skills and equipments and professional training, and the information platform of emergency rescue (fig. 1. ). fig. 1. the framework of emergency rescue research 3.1. organization and command of multi-forces “9.11” event emergency rescue operation shows that coordination and communication were serious problems ， and improvisation was common[4]. in china, “5.12” earthquake emergency response actions sparked in-depth discussion for strategic delivery capability, professional force and so on. the emergency rescue is a critical stage of emergency management. its core is organization and published by atlantis press copyright: the authors 175 z. zhang, et al command, including three levels of the strategic, operational and tactical. strategic level includes emergency rescue system, mechanism, and law. its tasks are involved: research and judge the situation of emergency rescue in domestic and abroad emergency events; understand the frontier issues of emergency rescue, make the practical strategy for professional forces in emergency rescue; study the inner link among the leading system, coordination mechanism and relative regulations and laws, provide theoretical support in terms of perfecting system and mechanism for professional emergency rescue forces; investigate the key direction, basic layout and components in building emergency rescue forces, the vertical management and horizontal coordinative mechanism in rescue operation, as well as the involved rules, regulations and laws. operational level is joint command of multi-forces that includes: study the command structure and its components, levels, function and general conditions in on-spot joint command [5]; probe the coordination of professional forces and other forces, like public security and local governments on all-level, along with the organization and implementation of international cooperation in emergency rescue, in particular under the real situation of diversified emergency rescue forces. tactical level is on-spot command of emergency rescue. the following tasks should be done: study the on-spot command patterns based on the type of emergency events, software and hardware safeguard measures, the authorization and emergency decision on rescue operation, the quick and accurate delivery in special and sensitive regions, sealing and control of the spot, search and rescue of key targets, etc. in organization and command of multi-force emergency rescue, being systematic is the focus. 3.2. risk management of rescue operation the role of the emergency rescue is to act immediately to prevent damage to life, property or environment [6]. richard yarwood considers the role of the emergency services in controlling, mitigating and resolving risk [7]. the risk and evolution of emergency events involves: probe the evolutionary patterns of nurturing, happening, development and mutation of emergency events; get to know the types, tense, spatial and temporal distribution of the events by human intervention to stop the mutation for hazard, so as to provide direct scientific support to weaken the consequences. the acceptor here means the objects which receive the actions produced or released or carried by the emergencies. we need to study the acceptor hazard-affected carrier, functional medium of the harmonious development of human society and natural environment, refers to the action object of emergency events, as well as the protective target of emergency rescue operation. the damage of hazard-affected carrier could lead to the activation or sudden release of its disaster elements, thereby causing secondary and derivative disasters and forming the chain of emergency events [3]. the key goal should be decided by studying different types of hazard-affected carriers so as to take proper action to block its happening and to achieve effective rescue. security management and control of emergency rescue is involved: study the security controlling technique in emergency rescue operation, master scientific methods, moment and dimension to measure and assess operation cost; mitigate the possible damage to the protective target and force staff by optimizing human intervention. 3.3. equipments and skills equipment, technology and training are effective implementation of the basic condition of emergency rescue. the serialized system of emergency rescue equipments is necessary. the research content include the planning, design and application of serialized equipment for professional emergency rescue, command theory and method of equipment maintenance; research and develop new equipment as on-scene uav (unmanned aerial vehicle) monitoring system in line with the series standards of equipment configuration based on various emergency rescue tasks. on the basis of the equipment, study professional emergency rescue techniques of typical disasters as earthquake, geological incident, forest fire, flood and drought; learn the repairing technique of traffic facilities as highway, bridge, tunnel, airport and port, the maintenance technique of important highway and tunnels under special time and occasions, the levee engineering rescue technique, the dam reinforcement technique of reservoir and hydropower stations, the engineering danger-elimination technique of the dammed lake, the emergency repairing technique of converting station, transmission wire and oil and gas pipeline, as well as the rescue technique of helicopter. the retrospective review of all search and rescue (sar) missions within yosemite national park during the 10-year study period between january 1990 and december 1999 shows, helicopter was the published by atlantis press copyright: the authors 176 research framework for emergency rescue primary mode of transport in 28% of sar incidents [8]. to combine personnel, technology and equipment effectively, must be strengthened skill training of emergency rescue equipment. probe the training methods, management, safeguard, patterns in view of base, analog, internet and situation, as well as interregional and joint training. 3.4. information platform information is the basis for decision-making, the quality of information to determine the quality of the decision-making. disasters and crises tend to disrupt the existing social order, resulting in distortion and confusion, affecting the quality of decision-making. emergency rescue operations need to rely on the information management tools to ensure the smooth flow of information up and down. the fresh, accurate information is essential for effective response to large-scale emergencies. information, at the top command level, at the field command level, and at the individual emergency personnel level, often difficult to obtain, process, interpret, and use for decision making in practice. so，we must strengthen the research of information hardware and software. first ， study the functional need of information collection, perception, positioning, delivery, summarization, research, distribution, integration and sharing; research, develop and integrate systematic modules as gis (geographic information system), remote sensing and monitoring, professional model, analog emulation, command and control, aid decision making, comprehensive inquiry. second, database of emergency rescue cannot be ignored. the database consists of basic database, expert repository, case base of great emergency events, database of emergency events participated by professional rescue forces, model base of disaster prediction and pre-warning, scenario and plan base of emergency rescue operation. we believe the high-tech but not make a perfect fetish of the high-tech, relying too much on high-tech to miscarriage of justice emergency rescue situation. we must give play to the initiative, and constantly strengthen the application research. 4. concluding comments in organizing and commanding, should focus on and improve top design. the emphasis should be laid on the systematic connection of strategy-battle-tactics level to produce a resultant force. in addition, emergency rescue should be incorporated into the legal system, standardized and scientific track. in handling multiple types of emergency events, being scientific is the focus. the risk analysis and assessment in the whole process should be strengthened. the emphasis lays on the security management and control of responsive actions. in terms of equipments and skills of emergency rescue for professional teams, being professional is the focus. system and standard should be formed and implemented in specific training courses. as to the direction of informational construction of emergency rescue, being modernized is the focus. the specific functional need of application system and database should be clearly defined. reference [1] c. c. shan, on the construction of crisis management system with chinese characteristics. china executive leadership academy pudong, 5:12-18, 2008. [2] j.g. liu, t. dietz, s.r. carpenter, et al. complexity of coupled human and natural systems. ambio, 317: 1513-1516, 2007. [3] w. c. fan, y. liu, w. g. weng, triangular framework and "4+1" methodology for public security science and technology. science & technology review, 6:3, 2009. [4] b. dearstyne, the fdny on 9/11: information and decision making in crisis. government information quarterly, 24:29–46 ,2007. [5] z. zhang, s. b. zhong, h. y. yuan, the effective space theoretic model of the forest fire suppression and its actual battle analysis. journal of mountain science, 5:534-539, 2010. [6] civil contingencies secretariat, civil contingencies act 2004: a short guide. hmso, london. 2004. [7] r. yarwood, risk, rescue and emergency services: the changing spatialities of mountain rescue teams in england and wales. geoforum, 41:257–270, 2010. [8] k. eric, m.d. hung, a. david, et al. search and rescue in yosemite national park: a 10-year review. wilderness and environmental medicine, 18:111-116 ,2007. published by atlantis press copyright: the authors 177 microsoft word volume 13, issue 2-1 journal of risk analysis and crisis response, 2023, 13(2), 93-116 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i2.359 93 article integrated supply chain risk assessment methodology based on modified fmea huseyin s. kilic 1, sena k. canbakis 1, melike karabas 1, sedef koseoglu 1, ezgi unal 1 and zeynep t. kalender 1,2,* 1 industrial engineering department, marmara university, istanbul (34840), maltepe, turkey 2 department of enterprise management and economics, czech technical university in prague, prague (12135), czechia * correspondence: tugce.simsit@marmara.edu.tr / ztsimsit@hotmail.com received: february 15, 2023; accepted: april 17, 2023; published: june 30, 2023 abstract: supply chains have rapidly developed and become more complex to increase productivity, reduce costs and meet demand. moreover, globalization affected complexity and uncertainty, so many risks have arisen in surviving among rivals. today, supply chain management became more difficult for recognizing the details and controlling the process. so, recognition and assessment of the risks in advance became vital to provide the continuity of the working processes of the company and not cause big losses. this study aims to minimize the damage that may occur by providing the risk assessment. for this purpose, a risk assessment methodology in which risks are determined via literature review and expert views is proposed. afterwards, a new technique based on failure mode and effects analysis (fmea) is utilized. intuitionistic fuzzy analytic hierarchy process and weighted product are synergistically used within fmea. moreover, an application in a widely known fast-moving consumer goods company is performed to validate the proposed methodology. keywords: risk assessment; supply chain management; failure mode and effects analysis; intuitionistic fuzzy analytic hierarchy process; weighted product 1. introduction nowadays, businesses must be more competitive more than ever to survive in today’s global business circumstances in which underestimated risks gained more importance since they can affect success of business steps and cause major loss. in this regard, assessment of risks is vital not only estimate the potential failures and take precautions before occurrence but also eliminate the loss [1]. in today’s business environment, one of the main functions in every business which operates successfully is the supply chain management (scm). in literature, supply chain (sc) basically defined as a system that consists of the integration of all parts of the chain such as suppliers, manufacturers, distributors, retailers, and customers. having a multi-stakeholder structure brings with it the potential of many risk sources [2]. it is possible that these risks have many different root causes, such as constraints, quality, miscommunication or uncontrollable or unpredictable reasons [3]. however, whatever the root causes, the most important thing is to understand these risks and thus manage them. for a basic risk management, these potential risks should be identified, measured, and then prioritized and appropriate methods should be determined [4]. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 94 regarding the importance of risk management in scs, the objective of this study is to define the possible risk factors in scs and assess these risk factors via proposing a methodology. moreover, a case study is performed in a fast-moving consumer goods company to implement the proposed methodology which includes failure mode and effects analysis (fmea), intuitionistic fuzzy analytical hierarchy process (if-ahp) and weighted product (wp). 2. literature review in the literature, while defining risk, many different terminologies are used, which basically represent undesirable results such as danger, destruction, and loss etc. in other words, risk is described as the probability of a particular adverse event occurring during a stated period. some sources [2] define risk as the result of a particular challenge, while others [5] define it as an unclear accident that will affect the accomplishment of objectives in compliance. 2.1. risk identification one of the critical steps that have a direct impact on the success of scm is the identification of risks. risk identification, which consists of searching, determining, and defining potential risks, is also the initial step in every risk assessment model. the performance of scs can affect by those potential risks whether positive or negative ways [4]. moreover, risk identification helps managers to better understand the future uncertainties in the processes hence these uncertainties could be managed properly with taking into consideration of potential risks [5]. risk identification can proceed in many ways. some of the related methods are listed as follows [2]:  geo mapping/sc mapping: visual maps of sc structures, dependencies and handoffs. using a powerful search tool, organizations can quickly interrogate captured data to reveal information about the profile of their sc, and gain valuable insight into potential risks. some geo mapping/sc mapping methodologies are icam definition for function modeling (idef0), supply chain operation reference (scor) and value stream mapping [6].  looking at historical problems: the possibility of reappearing historical problems might be high. these problems may have appeared in the system itself or to others.  researching industry trends: other organizations might have already investigated relevant risks.  group of experts brainstorming: people who have knowledge of risk and experiences in sc come together and share their ideas. knowledge is generally embedded in people's minds, and access to knowledge includes brainstorming meetings with the experts.  assessment surveys: surveys are fast and effective ways to collect information and data about risks in the sc.  site visit: site visits to sc associates enable to collect “specific” information and data on risk.  literature review: literature review is one of the efficient ways for determining risks. a systematic literature review provides guidance on how the risks are identified in previous studies and the appropriateness of the identification method [7].  qualitative methods: it is commonly used for risk identification and risk examination steps. some qualitative methods are fmea, empirical analysis and process-performance modelling [4]. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 95 furthermore, palaniappan [2] states that, there are also several tools used in risk identification such as checklists, cause and effect diagrams, gantt charts etc. the main purpose of using risk checklists is to provide a quick reference list for risk experts to ensure that all appropriate activities related to risk assessment have been addressed. it can be used to assure that steps have completed the activities associated with conducting an effective risk assessment. however, cause-and-effect diagrams which are also called as fishbone diagrams, identify many possible causes for an effect or problem. it can be used to structure a brainstorming session and immediately sorts ideas into useful categories. gantt chart can also be used in risk identification to provide and create a critical path to take notice of before risks occur. 2.2. risk classification determining and utilizing the most effective techniques for clarifying the risks of the organization and classifying the identified risks are required for the risk assessment process. it is possible to use different information gathering and assessment techniques. however, the important point is the applicability by the organization and getting useful results. after selecting the risk identification techniques, the classification of risks with their sources should be performed. based on the literature review as summarized in table 1, it is seen that there is not only one right way to do this. decisions are made with organizational competencies and techniques that are appropriate to the requirements of risk assessment. in this framework, risk classification should be done in a specific hierarchical structure. in the literature, there are several risk categories. some of them are listed as follows:  demand risk: variance of demand for a product depends on several factors such as, economic downturn, changes in customer behavior, communication failing, customer becoming more powerful in the bargaining or demand getting changeable. diabat et al. [3] define sub-risks as labor strikes, absence of qualified worker, unexpected loss of demand because of economic downturn, changeable demand. demand risk includes new product introductions, insufficient manufacturing and handling capability, high levels of process variations, changeable technology [8]. according to li et al. [9] it also consists of sudden engineering changes, order cancellations, late orders, skewed order cycles, air traffic, issues at customs, weather conditions, receivables, demand uncertainty. there are many components of demand risk studied by the researchers. however, main categories can be defined as “unexpected or very volatile customer demand”, “deferment in delivery”, and “brand image risk” [1].  supply risk: nowadays, there is an increased dependence on suppliers and it constrains the company to deal with supply risk. this situation leads to the distribution of negative events that affect the ability of domestic firms to meet customer demands (both quantitatively and qualitatively) at expected costs and time or threats to customer life and safety. primarily supply risk occurs by the cause of variability connected with product and supplier. according to li et al. [9], supply risk includes supply suspension, material scarcity, limit of capacity, quality issues, and layer risk. however, manuj and mentzer [8] claim that it should be examined as a disturbance of supply, inventory, timetable, technology connection, price increase, quality issues, technology changeableness, product complication and frequency of material design changes. main categories can be defined as “product quality issues”, “complication of important materials”, “unexpected failure of suppliers”, and “unstable behavior in supplier” [1]. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 96  operation and process risks: processes are series of value-adding and administrative actions or steps taken in order to accomplish an appropriate end set in motion by the company. operation and process risks totally connected to interruption to those processes. this affects company’s domestic capability to manufacture and supply goods/services that arise from the result of a disruption in a central performing or processing capacity [2]. operation and process risks cause a focused organization and show up logistics and manufacturing progress in addition to administrative activities [10]. risks factors mainly include planning, scheduling, and forecasting errors [4].  production risk: the production companies might be led to incompetence in the sc from unsatisfactory performance. although companies implement massively in programs, such as total quality management, lean production and six sigma, to develop domestic quality and competence; their domestic actions are affected to problems which may cause variation in capability and quality. the important risk sources consist of “manufacturing unpredictability”, “changes in production process”, “insufficient production capacity” [1]. it includes processes, innovation capability, and technical capability [11].  environmental risk: risk may occur as a result of cooperation between the sc network and its environment. in 2007, khan and burnes [12] stated that economic, political and social developments and events are raising the risk of sc disturbance that is more complex due to the fact that scs become longer. punniyamoorthy et al. [1] defined some keywords for the environmental risk such as “policy”, “macroeconomic”, “social”, “natural and manmade disasters”. it is also defined as uncontrollable incident that is related with external and contains changeable actions that show up from the sc and environmental interactions. it contains natural disaster, terrorism and war, political unrest in the region, changeable regulatory, strikes and absence of qualified person [1, 3].  financial risk: sc finance maintains financial services to all accompanying businesses which include upstream and downstream industries. it consists of a lack of understanding of risk importance, complicated credit environment and the process and acceptance of possessions rights which are complex and confused [10]. cost risk can be examined under financial risk and it contains supplier’s selling price, value-added cost, logistics cost [11]. according to kara and fırat [13], financial risk also includes economic environment, cost/price risks, cash flow risks, and changes in currency, economic recession, stability of gdp, tax risk.  regulatory risk: regulatory changes, both foreign and domestic, create challenges to global scs. industry-specific changes within countries, such as traceability rules for perishables; changes in a country’s stance toward imports and exports, or participation in existing international institutions are also examined in regulation risk [10]. penalty which stems from breaking the law, the state that does not foresee any movements such as import bans or export restrictions can be studied under this title [13].  customer and market risks: market risk is the likelihood of an encounter loss due to causes which have effects works of the financial markets in which the financier participated. it is affected by salability, marketability, and growth [11]. besides, market and customer risks are high level of risk categories that can occur in sc. environmental, economic, political factors and global competition have effects in this risk [4]. according to kara and fırat [13], these risks huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 97 include quality issues in goods and services, financial power of customers, customer relations, reputation, brand and image risk of the company, and disruption of order system.  logistics risk: logistics risk might be classified as the possible disruptions to the process of products, knowledge, and fund. in sc literature, factors and results of unpredictability in transport operations have been mentioned less. the important logistics factors are determined by several researchers as follows: “storage problems”, “transporter financial power”, “transport network management” and “deferments in distribution” [1]. according to wang et al. [11], logistics risk can be studied in three steps as internal, packaging and external. it contains transport failures, adaptability of transportation, delivery failures, weather conditions, inadequate packaging, and failures of labor who work in transportation [13]. risk classification is one of the vital steps of risk management in scs. the underlying reason is that key risk factors should be identified for every risk that is going to be classified and taken precaution to ensure that the risks do not overlap with each other as much as possible. after categorizing the risks which are identified in the first step, risk assessment process begins as the third step. 2.3. risk assessment the determined risks are assessed via a suitable technique and their impacts are obtained. afterwards, companies try to minimize sc risk factors to maximize productivity [14]. also, risk assessment methods balance demand frequency, the likelihood of supply, allocation of effective resources, determination of decision analysis paths [1]. different approaches and methodologies are implemented for risk assessment in the literature. bow-tie analysis is one of the mostly used methods to reduce uncertainty and to position different types of risk in sc by using fault and event tree. event tree analysis is applied to determine starting event and if event is sufficient, procedures are implemented in system plan. the analysis also provides to calculate risk impacts. alternatively, the expert’s knowledge provides rating to calculate probability of risks [8]. liu et al. [9] applied influence diagram as an assessment tool. topology and function layer are considered and processes are defined as controllable and uncontrollable. thus, risk elements are determined and the relationship of risk elements is described to each other. chaudhuri et al. [14] claimed that there was no readily proper procedure in the literature for risk assessment. step by step approach was applied in the study then numerical and linguistic data were provided in determining vulnerable scores for various subsystems. they used fmea to prioritize and reduce risk factors so vulnerable factors were identified and subsystems which were the degree of supplier involvement, the complexity of the process, complexity of logistics and manufacturing capabilities were defined. these subsystems help companies to reveal the coordination between suppliers and mechanisms. in 2013, li et al. [9] developed extended risk matrix approach to create new risk assessment method by adding the “detectability and recoverability” dimensions to traditional risk matrix approach which only consist of two dimensions: severity and probability. researchers selected the probability, severity and detectability dimensions to analyze and to evaluate for each risk factor to the multinational it/electronics company. the scores of these dimensions were derived from survey data and in conclusion “disruption risk” was determined more dangerous than the others. punniyamoorthy et al. [1] applied cronbach’s α-value to evaluate accuracy of risk factors. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 98 confirmative factor analysis was used to formalize risk assessment tools. the aim of study was to create a 45-component tool with six main risk structures and to provide the development of tools for assessing the sc risks. aqlan and lam [4] applied modelling approaches for sc risk assessment. modelling approaches were constitutive, qualitative and hybrid models. these models were developed in manufacturing environment. researchers focused on high-end manufacturing which was part of a manufacturing environment and discovered many uncertainty factors in this area. uncertainty is a well-known issue which affects the analysis of real-life cases and to overcome the vagueness problems researchers used several approaches. in 2016, bocquillon and ekallam [10] used fuzzy inference system to calculate the risk scores with the aim of revealing important risks that have a strong effect. after the application of the risk assessment method, the products having the highest risks were determined. the methods which are used in literature for risk identification, classification and assessment are summarized in table 1. table 1. literature review for the used methodologies. authors risk identification methods risk classification methods risk assessment methods shenoi et al. [15]  literature review  demand side risk  supply side risk  logistic risk  regulatory, legal and bureaucratic risk  infrastructure risk  stock/data management risk  environment risk  financial risk  ipa (important performance analysis) rostamzadeh et al. [16]  literature review  interview with experts  interview with expert panel consist of managersresearch expertspractitioners  environmental risks  organizational risks  sustainable supply risks  sustainable production/manufacturer risks  sustainable distribution risks  sustainable recycling risks  information technology related risks  integrated mcdm approach based on the topsis and critic method qazi et al. [17]  fmea  inbound-outbound logistics  the operators at the manufacturer  the operations at the suppliers  the final customers  bayesian belief network (bbn) with rnelpm (risk network expected loss propagation measure for risk)  utc (upper tail contribution)  risk network expected loss (rnel) kırılmaz and erol [18]  literature review  examined with the managers  quality problems  inability t to adapt changes in customer demand  impact probability matrix huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 99  increasing raw material prices  bankruptcy of supplier  supplier capacity risk  machine breakdowns  delivery chain disruptions  malfunction of it systems  accident risk (e.g., fire)  industrial action risk (strike etc.)  transportation failure  import restrictions  terrorist attack  extreme weather conditions  increasing customs duty luthra et al. [19]  fmea  empirical analysis  processperformance modeling - ahp bocquillon and ekallam [10] -  process control risk  environment risk - özveri and kabak [20] --  fmea, ahp  ahp-promethee jaberidoost et al. [7]  literature review -  ahp  saw aqlan and lam [4]  survey  supplier risk  customer risk  process and control risk  technology risks  product risks  occupational risk  transportation risk  commodity risk  fuzzy set theory method  top-down and bottom-up analysis palaniappan [2]  looking at historical problems  researching industry trends  group of experts brainstorming  assessment surveys  site visit  risk checklist  cause-and-effect diagrams:  gantt charts  geo mapping/supply chain mapping  supply risk  demand risks  process and control risks  environmental risks - aslania et al. [21] --  a hybrid of fuzzy fmea-ahp punniyamoorthy et al. [1]  survey  supply risks  environmental risks  demand risks  manufacturing risks  logistics risks  information risks  confirmative factor analysis (cfa) and lisrel 8.8. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 100 soni and kodali [22] -  labor risks  market risks  transfer risks  financial risks  living conditions  natural condition  promethee  ahp chan and wang [23] --  fuzzy topsis  ahp li et al. [9]  survey data  3d graph representation  supply risks  demand risks  process risks  disruption risks  extended risk matrix approach maheswaran and loganathan [24] --  ahp  promethee  fmea huang and paradi [25]  modified multistage sequential dea technique  endogenous risk  exogenous risk - chuang [26]  service blueprint  service facility o sales floor facility o sales floor security o sales floor surroundings  prior-service o incoming goods/merchandise activity o warehousing and inventory activity  in-service o customer choice /purchase flow o cashier flow  post-service o post-sale activity o warranty  fmea chang and hsiao [27] -  financial  functional  physical  psychological  social  temporal  multiplication model and questionnaire 3. materials and method according to the literature review, it is seen that there are several approaches to analyze risks in scm. however, uncertainty remains one of the biggest problems that researchers face in real life. in this paper, three approaches are synergistically utilized for risk assessment process to overcome the main problems. these approaches provide the risk rankings according to their importance via calculating revised risk priority number (r-rpn) with using fmea, ifs-ahp and wp methods. proposed methodology of this study is demonstrated in figure 1. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 101 figure 1. framework of the proposed model. different from the existing literature, a new integrated approach including fmea, ifs-ahp and wp is generated. to eliminate the deficiency of integrated use of ahp and fmea in the existing literature, a new approach with wp is used by also considering vagueness via fuzzy approach. regarding the current integrated use of ahp and fmea, it is observed that the sum of weighted factors is used. however, multiplication of factors is necessary to fit the original form of fmea. to make this multiplication is valid; the importance weight should be exponentially indicated. at this point wp is a suitable technique. the rest of this study is organized as follows. initially, each method including fmea, ifs-ahp and wp are explained in details in sub-sections. the new approach using the mentioned methods to substitute the conventional rpn method is elaborated in detail. afterwards, the proposed methodology is applied in a food company. finally, results are presented. 3.1. failure mode effect analysis (fmea) fmea is a reliable method for failure analysis [28]. it is used as a tool of risk assessment [24] and prioritizes the risks [2]. fmea, which was designed in 1960s, is one of the main systematic techniques focuses on failure analysis [21]. after the methodology is developed it was improved by ford motor in 1977 [24]. in today’s business environment, it is considered as a very beneficial risk assessment tool and implemented in various fields [28]. fmea is very successful in detecting the potential failures in processes [20], therefore it enables to decrease the potential risks and to improve the quality, safety and reliability of a system [28]. in processes, root causes of failures can vary, have different impacts and raise from several reasons. in that case, each failure mode with high risk should be considered for evaluation regarding their risk and then they should be prioritized. a significant point of fmea is that it is a proactive approach [20] which aims to detect the potential failures for each subsystem [28] before they happen in order to prevent them. in prioritization of the risks, risk priority number (rpn) is used in fmea [2]. in order to calculate rpn score, three dimensions for each failure mode should be considered such as severity fmea expert views literature review determination of risk categories ifs-ahp: weight determination of severity, occurrence and detection rankings of sub-risks wp revised rpn calculations r isk a ssessm ent huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 102 (s) which indicates the effect of failure, occurrence (o) which is the frequency of failure, and detection (d) which is the probability of detecting the failure [20]. rpn can only be calculated after these values of the risk criteria determined with eq. 1 [24]. 𝑅𝑃𝑁 = 𝑆 ∗ 𝑂 ∗ 𝐷 (1) the values of s, o and d for each failure mode are obtained from the experts, according to their importance level rating scale which can be between 1 and 10 [24]. these rating scales are given in the table 2 and table 3. table 2. occurrence & severity rating scale [20]. rating description potential risk rate for occurrence definition for severity if set 𝝁 ν π 10 extremely high more than one per day risk causes a loss of a customer or an employee 1 0 0 9 very high once every 3-4 days risk causes a huge lateness on system 0.8 0.1 0.1 8 repeated failures once every week risk could cause a major dissatisfaction on customer side 0.62 0.18 0.2 7 high 6 moderately high once every three months risk can cause a minor effect on product/service performance 0.47 0.23 0.3 5 moderate 0.33 0.27 0.4 4 relatively low once per year risk causes a minor dissatisfaction or just annoys customer 0.22 0.28 0.5 3 low 0.13 0.27 0.6 2 remote once every 3-6 years a customer cannot distinguish that 0.06 0.23 0.7 1 nearly impossible no one remembers last risk occurrence risk has no impact on system 0.02 0.18 0.8 table 3. detection rating scale [20]. rating description definition for detection if set 𝝁 ν π 10 absolute uncertainty no design control or no chance of detection 1 0 0 9 very remote the risk can be detected with unfeasible inspection 0.8 0.1 0.1 8 remote 7 very low the risk can be detected only with manual inspection 0.62 0.18 0.2 6 low 0.47 0.23 0.3 5 moderate the risk can be detected by an un-automated process 0.33 0.27 0.4 4 moderately high there is 100% inspection or review of the process, but still un-automated 0.22 0.28 0.5 3 high 0.13 0.27 0.6 2 very high there is 100% automated inspection or review of the process 0.06 0.23 0.7 1 almost certain there is automated shut-offs to prevent risk 0.02 0.18 0.8 the traditional fmea is based on prioritizing the risks. the underlying reason of method's extensive use is the simple structure of the method which makes it preferable. however, this structure is again the main reason why the method has some shortcomings. chang et al. [29] stated that significant criticisms about using the fmea in a traditional way. some of these criticisms are duplication of rpn elements, measurement scale violations, problems in determining each dimensions’ weights and drawback of determining the subjective structure with current rpn scale. these issues are stated in many researches and wu et al. [30] proposed a structure in which each dimension calculated differently since each of the three dimensions have different characteristics. in huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 103 2017, liu et al. [31] also emphasized the shortcomings of calculation of rpn and especially pointed out the problems about the weight of risk factors. huang et al. [25] searched the literature so that they summarized the major flaws. their analysis showed that major problems are about using crisp numbers in evaluation of risks, sensitive formulation of rpn to the variation and weighting of each dimension. they proposed a new form of fmea that uses linguistic distribution assessments on the failure modes. traditional calculation of fmea has been criticized for several reasons and becomes one of the mostly studied issues in literature. however, fmea has been proven to be one of the most important early preventative initiatives for companies and its popularity still increases. mostly, hybrid methods are generated to overcome the mentioned flaws. especially, to overcome problems in appropriate weight determination for each dimension, multi-criteria decision making (mcdm) techniques are used under the fuzzy environment. technique for order preference by similarity to ideal solution (topsis), vikor (vlsekriterijumska optimizacija i kompromisno resenje) methods under the fuzzy environment and preference ranking organization method for enrichment evaluation (promethee) method are the examples of hybrid models which are proposed in order to evaluate the dimensions in a more meaningful way [24]. in this study, relative importance among severity, occurrence and detection is calculated via using ifs-ahp method with the consideration of the consolidated opinions of four different managers and their teams as four different decision makers’ (dm) views of points. thus, not only the weighting determination problem is considered but also more suitable scale is used to analyze the uncertainty of dm’s opinions. 3.2. ifs-ahp in real-life decision-making process of a problem, dms are face to face with several circumstances such as time pressure, inefficient information etc. which are directly affect their evaluations. additionally in some cases dms ability or attention can be limited during the assessment procedure. these points make the information provided by dms insufficient and most of the time uncertain in many complex decision-making problems. in this regard, to overcome these uncertainties, fuzzy logic-based tools to describe uncertain decision-making information. in literature, traditional fuzzy logic theorems are developed by many researchers to better reflect and solve the uncertainty in the problem. in this sense, ifs theorems are generally used in approximate induction, sample definition and decision making in literature. xu and yager [32] point out that the ifs theory is useful in studies with uncertainty, and they have mentioned that there is a lot of work in the literature for further development of the theory. ifs are generally defined as a = 〈 x, μ (x), 𝑣 (x) : x ∈ x〉 (2) here, μ (x): x → [0,1] is defined membership and 𝑣 (x): x → [0,1] is defined non-membership function. these functions have a property of 0 ≤ μ (x) + 𝑣 (x) ≤ 1. in this case hesitation is defined with π (x) = 1 − μ (x) − 𝑣 (x) (3) the operations of addition ⊕ and multiplication ⊗ on ifs were defined by atanassov [33] as in eq. (4) and eq. (5). let a= 〈μ , 𝑣 〉 and b = 〈μ , 𝑣 〉 be ifs. then, 𝐴 ⊕ 𝐵 = 〈μ a + μ b − μ a . μ b , 𝑣a . 𝑣b〉 (4) 𝐴 ⊗ 𝐵 = 〈μ . μ , 𝑣 + 𝑣 − 𝑣 . 𝑣 〉 (5) huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 104 let 𝜆 is a real value (𝜆 > 0) and ⊘ represents the division operations, then other operational functions can be defined by dymova and sevastjanov [34] as 𝜆𝐴 = 1 − 1 − μ a 𝜆 , 𝑣a (6) 𝐴 = μ , 1 − (1 − 𝑣 ) (7) 𝐴 ⊘ 𝐵 = 〈𝑥, , 〉 𝑥 ∈ 𝑋 (8) with conditions a≤b, μ (𝑥) ≠ 0; 𝑣 (𝑥) ≠ 1 and μ . 𝑣 − μ . 𝑣 ≥ μ − μ . preliminaries of ifs sets are vital to better understand the ifs-ahp methodology steps which are given in detail in the following part. in this study, ahp is only used for the weight determination thus the explanations of the methodology consist of weight determination steps.  step 1. developing a hierarchical structure: objective of the problems is identified and criteria for each level are determined. then criteria are evaluated for all alternatives which are also defined in the hierarchical structure of the problem.  step 2. developing pairwise comparisons: evaluations of dms for each criteria and alternative are collected and pairwise comparison matrices are created based on the linguistic scale given in table 4. table 4. linguistic evaluation scale for ifs-ahp [35]. preference on pair-wise comparison ifns reciprocal ifns 𝝁 ν π 𝝁 ν π equally important (ei) 0.02 0.18 0.8 0.02 0.18 0.8 intermediate value (iv) 0.06 0.23 0.7 0.23 0.06 0.7 moderately more important (mmi) 0.13 0.27 0.6 0.27 0.13 0.6 intermediate value (iv) 0.22 0.28 0.5 0.28 0.22 0.5 strongly more important (smi) 0.33 0.27 0.4 0.27 0.33 0.4 intermediate value (iv) 0.47 0.23 0.3 0.23 0.47 0.3 very strong more important (vsmi) 0.62 0.18 0.2 0.18 0.62 0.2 intermediate value (iv) 0.8 0.1 0.1 0.1 0.8 0.1 extremely more important (emi) 1 0 0 0 1 0  step 3. calculation of dms’ weights: in mcdm methodologies, there can be several dms and importance of them can be different. in such cases, a linguistic term is assigned to the dm to rate then the weight of the dm is calculated by eq. (9). in table 5, the linguistic terms for dms’ ratings are presented. table 5. linguistic terms for decision makers [36]. linguistic terms dms’ ratings ifns 𝝁 ν π very important vi 0.8 0.1 0.1 important i 0.5 0.2 0.3 medium m 0.5 0.5 0 bad b 0.3 0.5 0.2 very bad vb 0.2 0.7 0.1 let’s assume that there are several dms in the problem and dl= 〈μ , v , π 〉 is an ifs for dm k. if there are m dms, to find the kth dm’s weight, eq. (9) is applied. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 105 𝜆 = ∑ 𝑤ℎ𝑒𝑟𝑒 𝜆 ∈ [0,1] 𝑎𝑛𝑑 ∑ 𝜆 = 1 (9)  step 4. determining if decision matrix: xu [36] defined an intuitionistic fuzzy weighted averaging (ifwa) operator to aggregate the dms’ opinions. in group-decision making problems, individual opinions must sum up to construct a decision matrix. let 𝑅( ) = (𝑟 ) is a pairwise comparison matrix developed in step 2 for each dm and 𝜆 = {𝜆 , 𝜆 , … 𝜆 } is the weight of dm found in step 3. then ifwa operator is calculated with 𝑅 = 𝑟 = 𝐼𝐹𝑊𝐴 𝑟 , 𝑟 , … 𝑟 = 𝜆 𝑟 ⊕ 𝜆 𝑟 ⊕ … .⊕ 𝜆 𝑟 = 1 − ∏ 1 − 𝜇 ( ) , ∏ 𝑣 ( ) , ∏ 1 − 𝜇 ( ) − ∏ 𝑣 ( ) (10)  step 5. calculation of ifs weights: in literature, there are several methods such as eigenvector calculations, arithmetic mean and geometric mean etc. to calculate weights. according to sadiq and tesfamariam [38] there is no significant difference among these methodologies. in their study, firstly they used geometric mean and then normalized it to find final weights with eq. (11) and eq. (12). let’s assume that ji represents the geometric mean operation and wi represents the weights (i=1 to n) as 𝐽 = (𝐽 ⊗ 𝐽 ⊗ … ⊗ 𝐽 ) / (11) 𝑤 = 𝐽 ⊗ (𝐽 ⊕ … ⊕ 𝐽 ) (12) 3.3. weighted product in problems where multiple criteria have different importance weights, weighted product (wp) is one of the easiest scoring methods preferred by many studies. within the method, the importance weight of each attribute (wj) becomes exponents and the score of an alternative (vi) is calculated as in eq. (13) [39]: 𝑉 = ∏ (𝑋 ) (13) where i represent alternative, j represent attribute, vi is the total score of alternative “i”, xij is the score of alternative “i” with respect to attribute “j” and wj is the importance weight of attribute “j”. the shortcomings of rpn calculations is mentioned several times in literature. a simple example generally to best way to better present the limitation of traditional calculation of fmea. assume that, in the first failure mode for a risk type, occurrence (o) is equal to 2, severity (s) is equal to 5 and detection (d) is equal to 7 and in the second mode “o” is equal to 7, “s” is equal to 5 and “d” is equal to 2. in such case, even though different scores are determined for different failure modes, overall rpn values for both failure modes are the same as equal to 70. this deficiency stems from assuming the severity, occurrence and detection’s importance as the same [20]. on the other hand, in real-world cases, the relative importance of each criterion is very important in real life cases. however, the given example clearly indicated that traditional calculation of rpn values make risk assessments deficient. in this paper, wp which is a single mcdm method is used for calculating rpn. in the text, rpn will be mentioned as revised risk priority number (r-rpn) to show the difference. wp model provides to obtain a consistent result by using the specified importance weight for the interested factor [40]. therefore, the new r-rpn is calculated via wp to eliminate the deficiency of the traditional approach of fmea with the formula: huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 106 revised − rpn = s(o)w(o) x s(d)w(d) x s(s)w(s) (14) where, w and s are the importance weight and score of each criterion for all sub-risks, respectively. relative weights for each criterion are calculated using ifs-ahp methodology and scores are determined based on the opinions of dms. these mentioned score values are also in a scale of ifs to overcome the hesitation of the dms. for each approach, the r-rpn is calculated for each failure mode. after calculating the r-rpn, it is possible to obtain ranking among the sub-risks based on the rules of ifs. in this study, traditional ranking methodology which will be mentioned in the application section is used for the final r-rpn values. 4. application in a food company the application of the proposed methodology is performed in a food company. firstly, a set of meetings were organized with the managers of demand, raw material planning, finance-operations and sc managers to specify main and sub-risk factors. sub-risks are explained in the following section according to the company’s operations. after specifying them, the integrated methodology is implemented and finally the rankings are obtained. 4.1. sub-risks identification the risk categories which are evaluated in surveys are demand risks, supply risks, operational risks, financial risks, and logistics risks. each of these risks and their sub-risks are explained in detail as follows:  demand risks: the sub-risks are defined via interview with the demand manager (manager 1) and literature review. these sub-risks are as follows: o unanticipated or volatile customer demand is a real problem for companies, from discrete manufacturing to process industries. many factors cause demand volatility including customer demand increase, product customization, technological developments, global competition, and upward supply variation. unanticipated or volatile customer demand is a challenge encountered by companies at all levels. dealing with volatile demand in an adequate method can bring about important advantage for a company, including minimum sc costs and enhanced customer service status. in addition, dealing with volatile demand efficiently might be a big advantage in competition. o equivalent products are the products that have the similar features with the existing product. these products can create risks since they can be easily preferred instead of the existing product. this risk directly affects the management of demand strategies. o change in market trend can result with huge variance between forecasted demand and actual demand. customers’ purchase habits or priorities can be affected by many factors in the market. o pricing strategy accounts for segments, payment capabilities, market conditions, competitor actions, trade margins and input costs, among others. price strategy is a very important element of demand risks, because if the company chooses wrong price strategy, it causes loss or less profit. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 107  supply risks: the sub-risks are defined via interview with the raw material planning manager (manager 2) and literature review. these sub-risks are as follows: o supplier’s inventory can cause the disruption on the customer’s production due to the insufficiency of alternative suppliers. however, this situation may have several reasons to occur. if the supplier becomes aware of the customer’s possible needs, they may take action to prevent this. o company’s inventory is related with the level of company’s selected goods. most of the raw materials are purchased before start to production. if their expiration dates are not too short, they should be stocked within the company. if these supplied stocks level is not managed properly, it can affect the production process negatively. o incapable supplier is one of the important sub-risks of supply risks. the supplier selection process has high importance on company’s strategies. the supplier’s capacity should be capable of satisfying the company’s demand. otherwise, this issue causes to lose revenue on both sides. the other point is that, the supplier must direct the own investment plans according to customers’ annual targets. the supplier’s facility conditions, number of employees etc. must meet the customer’s criteria, as well. o inconsistent forecast mostly occurs internally. most of the mentioned supply risks source from inconsistency between planned and actual. there are always uncontrollable factors such as disruption on market, changing climate etc. however, if there is a huge variance between the actual and forecasted values if there is no uncontrollable event, it can be surely said that there is a mistake on forecasting process. the forecast inconsistency may occur on supplier side or customer side.  operational risks: the sub-risks are defined via interview with finance-operations manager (manager 3) and literature review. these sub-risks are as follows: o inconsistent production rate can occur internally or externally. any situation of inconsistent production rate, which is on supplier or firm side, may affect all the business processes. there can be different kind of reasons, which are caused to inconsistency, but in studied case these are mostly machine malfunctions, maintenance on equipment. o database issues within the company happen internally. every company has a database or reporting system for each stage. these systems can be interrelated. for a sc process, the managers should be aware of the actual produced amount or sales trend of firm’s products by following from database system. o mismanaging the employees, which can occur at all level of employees, may affect all the business process. if it occurs on blue-collar employees, it may cause to disruption on production or logistics process. then, it causes to arise other types of risks, such as inconsistency on production rate. o inadequate or unqualified employees is related with human resources stage of business. it is impossible to carry out the business operation successfully with untrained employees or lack of human-power.  financial risks: the sub-risks are defined via interview with finance-operations manager (manager 3) and literature review. these sub-risks are as follows: huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 108 o the threat of economic crisis is an external and uncontrollable sub-risk factor. in unsteady markets, there is always a threat of economic crisis. this threat may affect the market’s consumption habits. also, the foreign-financed companies may be affected on internal side. o uncertainty in foreign exchange rate risk occurs if supplier uses a foreign currency, thus any fluctuation in a market may have a high impact. in some cases, the company must review on pricing regulations. or, the company must reorganize payment plan according to the value of local currency against to foreign currency. o cyber risks mean that any risk of financial disruption or loss happen from some sort of failure of its information technology system. any financial loss affects the whole stage of business, mostly sc system. o disagreement on financial aspects with supplier can change from company to company. the underlying reason is that each company may have a different type of internal management regulations. this difference may cause a conflict between firm and its suppliers. to explain, the payment schedule or billing process can be different and as a result, a disagreement may arise between the firms who are working together.  logistics risks: the sub-risks are defined via interview with the sc manager (manager 4) and literature review. these sub-risks are as follows: o inefficiency in transport infrastructure is one of the most important factors for a country's progress and companies’ logistics. transportation infrastructure improves the quality and safety of ability to move in the economy and upgrades economic growth. however, in some cases such as road construction etc., the logistics cost would be higher than estimated. o vehicle procurement problem occurs when the cooperated company's vehicles are inadequate or when there are not enough vehicles in hand. the transfer activity of the company is interrupted by the emergence of this risk. o track and trace of vehicles include the operation of deciding the current and previous locations (and other information) of different item or property. the company needs technological infrastructure to perform the process. if there is problem and risk in track and trace of vehicles, it causes delays in logistics process. o seasonal logistics risk depends on the demand changes among the seasons. as an example in summer, demand can be much more than winter, as a result, logistics activities become frequent with the increasing demand in the related time. 4.2. implementation of the proposed methodology on case study initially, managers in the case company were asked to meet with their teams and to present a consolidated evaluation data as a result of these interviews. in the following sections, where the application is explained step-by-step, the terminology "dm" reflects the aggregated opinion of the relevant manager and their team. this term was used to facilitate the follow-up of the application steps throughout the study. the aggregated data obtained as a result of interviews with each manager's team were used as an input and the weighted values of s, o and d are calculated via if-ahp. the weighted values are huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 109 shown as ws, wo, and wd respectively. thereafter, as the second step, the r-rpn of each risk subcriterion is calculated via modified fmea. ws, wo and wd are used as the inputs for modified fmea. 4.2.1. determination of criteria weights with ifs-ahp in determining the weights of s, o and d, ifs-ahp structure is used. the data is provided by managers which have different importance weights. the hierarchy of the evaluation process is presented in figure 2. figure 2. analytical hierarchy process structure of the case study. each managers’ evaluations for criterion using linguistic scale given in table 4 and their intuitionistic scores are presented in table 6. table 6. criteria evaluations provided by the decision makers. manager 1 (m1) -s o d -s o d s ei mmi emi s (0.02; 0.18) (0.13; 0.27) (1; 0) o 1/mmi ei iv o (0.27; 0.13) (0.02; 0.18) (0.06; 0.23) d 1/emi 1/iv ei d (0; 1) (0.23; 0.06) (0.02; 0.18) manager 2 (m2) -s o d -s o d s ei 1/mmi 1/emi s (0.02; 0.18) (0.27; 0.13) (0; 1) o mmi ei 1/vsmi o (0.13; 0.27) (0.02; 0.18) (0.18; 0.62) d emi vsmi ei d (1; 0) (0.62; 0.18) (0.02; 0.18) manager 3 (m3) -s o d -s o d s ei 1/iv smi s (0.02; 0.18) (0.28; 0.22) (0.33; 0.27) o iv ei emi o (0.22; 0.28) (0.02; 0.18) (1; 0) d 1/smi 1/emi ei d (0.27; 0.33) (0; 1) (0.02; 0.18) manager 4 (m4) -s o d -s o d s ei 1/mmi 1/vsmi s (0.02; 0.18) (0.27; 0.13) (0.18; 0.62) o mmi ei 1/smi o (0.13; 0.27) (0.02; 0.18) (0.27; 0.33) d vsmi smi ei d (0.62; 0.18) (0.33; 0.27) (0.02; 0.18) dms are rated based on the linguistic term scale which is presented in table 5. the evaluation of dms and their calculated weights, in which eq. (9) is used, are presented in table 7. after the determination of dms’ weights, fuzzy decision matrix is calculated by applying the ifwa operator. provided information is aggregated and final scores are presented as aggregated matrix in table 8. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 110 table 7. decision makers’ weights. manager linguistic terms 𝝁 ν π weights m1 very important 0.8 0.1 0.1 0.277 m2 very important 0.8 0.1 0.1 0.277 m3 important 0.5 0.2 0.3 0.223 m4 important 0.5 0.2 0.3 0.223 table 8. aggregated matrix. dimensions s o d 𝝁 ν π 𝝁 ν π 𝝁 ν π s 0.020 0.180 0.020 0.236 0.179 0.241 1.000 0.000 0.849 o 0.191 0.222 0.198 0.020 0.180 0.020 1.000 0.000 0.601 d 1.000 0.000 0.849 0.349 0.213 0.388 0.020 0.180 0.020 to compute the final if weights of each criterion, step 5 of the proposed ifs-ahp model is applied. based on the related calculations (eq. (11) and eq. (12)), final weight scores are found and these scores are presented in table 9. table 9. weight values of severity, occurrence and detectability. weights of dimensions 𝝁 ν π weight of severity (ws) 0.388167 0.523021 0.088812 weight of occurrence (wo) 0.361892 0.559930 0.078178 weight of detectability (wd) 0.442442 0.552132 0.005426 the scores of ws, wo and wd, determined if weights, are used to calculate r-rpn values. however, if scores should be converted to crisp numbers since wp method is used in calculating rrpn values. thillaigovindan et al. [41] stated that major problem in dealing with ifs is transforming them into a single numerical value although ifs can represent the dms’ opinions effectively. according to study, score functions can be considered as a tool for turning the intuitionistic sets to a single real number. in 2011, chen [42] made a comparative analysis for score functions in ifs and compared six different score functions then suggested an appropriate score function which is presented in eq. (15): 𝑆 𝑥 = 𝜇 − 𝑣 . 𝜋 (15) score function is evaluated as a proper function especially in a complex decision-making problem. in this study, eq. (15) is used to find single value for each criterion. after the equation is applied to the values given in table 9, final weights for ws, wo and wd are found as 0.342; 0.318 and 0.440, respectively. 4.2.2. calculation of r-rpns for each sub-risk the obtained values of ws, wo and wd are used in finding r-rpns. sub-risks are evaluated by experts from the case study company based on the scale given in table 2 and table 3. in table 10, these evaluations and their if scores are presented. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 111 table 10. evaluated sub-risks. attributes o cc u rr en ce s ev er it y d et ec ti on occurrence severity detection 𝝁 ν π 𝝁 ν π 𝝁 ν π supply risks supplier’s inventory 4 7 9 0.22 0.28 0.5 0.62 0.18 0.2 0.8 0.1 0.1 company’s inventory 2 8 1 0.06 0.23 0.7 0.62 0.18 0.2 0.02 0.18 0.8 incapable supplier 6 5 7 0.47 0.23 0.3 0.33 0.27 0.4 0.62 0.18 0.2 inconsistent forecast 10 5 10 1 0 0 0.33 0.27 0.4 1 0 0 financial risks uncertainty on foreign exchange rate 10 10 3 1 0 0 1 0 0 0.13 0.27 0.6 the threat of economic crisis 8 9 3 0.62 0.18 0.2 0.8 0.1 0.1 0.13 0.27 0.6 cyber risks 4 8 6 0.22 0.28 0.5 0.62 0.18 0.2 0.47 0.23 0.3 disagreement on financial aspects with supplier 5 5 6 0.33 0.27 0.4 0.33 0.27 0.4 0.47 0.23 0.3 operational risks inconsistent production rate 6 6 2 0.47 0.23 0.3 0.47 0.23 0.3 0.06 0.23 0.7 database issues within the company 7 7 3 0.62 0.18 0.2 0.62 0.18 0.2 0.13 0.27 0.6 mismanaging the employees 3 7 3 0.13 0.27 0.6 0.62 0.18 0.2 0.13 0.27 0.6 inadequate or unqualified employers 3 9 3 0.13 0.27 0.6 0.8 0.1 0.1 0.13 0.27 0.6 demand risks unanticipated or volatile customer demand 10 9 10 1 0 0 0.8 0.1 0.1 1 0 0 equivalent products 7 3 6 0.62 0.18 0.2 0.13 0.27 0.6 0.47 0.23 0.3 change in market trend 4 5 9 0.22 0.28 0.5 0.33 0.27 0.4 0.8 0.1 0.1 price strategy 2 8 1 0.06 0.23 0.7 0.62 0.18 0.2 0.02 0.18 0.8 logistic risks inefficiency in transport infrastructure (technically) 4 5 3 0.22 0.28 0.5 0.33 0.27 0.4 0.13 0.27 0.6 vehicle procurement problem 7 9 2 0.62 0.18 0.2 0.8 0.1 0.1 0.06 0.23 0.7 track and trace of vehicles 6 7 5 0.47 0,23 0.3 0.62 0.18 0.2 0.33 0.27 0.4 seasonal logistics 7 6 3 0.62 0.18 0.2 0.47 0.23 0.3 0.13 0.27 0.6 after the evaluations are collected from the experts for each sub-risk, r-rpn values are calculated via considering the weight values that are found with ifs-ahp and presented in table 11. 4.2.3. ranking sub-risks and results comparison of the if values is an important problem and several methodologies are generated and used in literature. however, score and accuracy functions are basically suggested for comparisons [37]. dymova et al. [34] defined the comparison of ifs based on the order relationships between any pair set and used the formulation given in eq. (16). let a and b be two ifs sets. suppose that s(x) =μ(x) +ν(x) represents the score function and that h(x) =μ(x)-ν(x) represents the accuracy function. comparison of the two ifs sets depend on the following calculations: 𝐼𝑓 𝑆(𝑎) > 𝑆(𝑏), 𝑡ℎ𝑒𝑛 𝑏 𝑖𝑠 𝑠𝑚𝑎𝑙𝑙𝑒𝑟 𝑡ℎ𝑎𝑛 𝑎; 𝐼𝑓 𝑆(𝑎) = 𝑆(𝑏), 𝑡ℎ𝑒𝑛 (1) 𝐼𝑓 𝐻(𝑎) = 𝐻(𝑏), 𝑡ℎ𝑒𝑛 𝑎 = 𝑏; (2) 𝐼𝑓 𝐻(𝑎) < 𝐻(𝑏), 𝑡ℎ𝑒𝑛 𝑎 𝑖𝑠 𝑠𝑚𝑎𝑙𝑙𝑒𝑟 𝑡ℎ𝑎𝑛 𝑏. (16) huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 112 table 11. r-rpn values based on proposed methodology. r-rpn values 𝛍 ν π supply risks supplier’s inventory 0.9946774 0.0002939 0.0050287 company’s inventory 0.9284840 0.0004365 0.0710795 incapable supplier 0.9871943 0.0006798 0.0121259 inconsistent forecast 1 0 0 financial risks uncertainty on foreign exchange rate 1 0 0 the threat of economic crisis 0.9938859 0.0002791 0.0058349 cyber risks 0.9839098 0.0007049 0.0153853 disagreement on financial aspects with supplier 0.9735362 0.001054 0.0254098 operational risks inconsistent production rate 0.9655471 0.0007396 0.0337133 database issues within the company 0.9874129 0.0005181 0.0120690 mismanaging the employees 0.9580722 0.0008057 0.0411221 inadequate or unqualified employers 0.979634 0.0004341 0.0199320 demand risks unanticipated or volatile customer demand 1 0 0 equivalent products 0.9796825 0.0006777 0.0196398 change in market trend 0.9887868 0.0004577 0.0107555 price strategy 0.9284840 0.0004365 0.0710795 logistic risks inefficiency in transport infrastructure (technically) 0.9289269 0.0013066 0.0697665 vehicle procurement problem 0.9926722 0.0002345 0.0070933 track and trace of vehicles 0.9876278 0.0006751 0.0116971 seasonal logistics 0.9809490 0.0006756 0.0183753 table 12. score and accuracy function values of sub-risks. attributes score function value accuracy function value supply risks supplier’s inventory 0.9943835 0.994971319 company’s inventory 0.9280474 0.928920529 incapable supplier 0.9865145 0.987874073 inconsistent forecast 1 1 financial risks uncertainty on foreign exchange rate 1 1 the threat of economic crisis 0.9936068 0.994165060 cyber risks 0.9832049 0.984614675 disagreement on financial aspects with supplier 0.9724822 0.974590173 operational risks inconsistent production rate 0.9648075 0.966286738 database issues within the company 0.9868948 0.987930972 mismanaging the employees 0.9572665 0.958877867 inadequate or unqualified employers 0.9791999 0,980068044 demand risks unanticipated or volatile customer demand 1 1 equivalent products 0.9790047 0.980360225 change in market trend 0.9883291 0.989244463 price strategy 0.9280474 0.928920529 logistic risks inefficiency in transport infrastructure (technically) 0.9276204 0.930233487 vehicle procurement problem 0.9924377 0.992906727 track and trace of vehicles 0.9869527 0.988302948 seasonal logistics 0.9802734 0.981624655 huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 113 firstly, the score and accuracy function values are calculated for each sub-risk based on the values presented in table 11. results are presented in table 12. according to the calculated score and accuracy function values, eq. (15) is implemented to each sub-risk and then final ranking is determined via comparing each sub-risk pair separately. in table 13, the final rankings of the mentioned sub-risks are presented. table 13. rankings of the sub-risks. attributes final rankings supply risks supplier’s inventory 2 company’s inventory 16 incapable supplier 8 inconsistent forecast 1 financial risks uncertainty on foreign exchange rate 1 the threat of economic crisis 3 cyber risks 9 disagreement on financial aspects with supplier 13 operational risks inconsistent production rate 14 database issues within the company 7 mismanaging the employees 15 inadequate or unqualified employers 11 demand risks unanticipated or volatile customer demand 1 equivalent products 12 change in market trend 5 price strategy 16 logistic risks inefficiency in transport infrastructure (technically) 17 vehicle procurement problem 4 track and trace of vehicles 6 seasonal logistics 10 regarding the final rankings shown in the table 13, the three most important risks are determined as “inconsistent forecast”, “uncertainty on foreign exchange rate” and “unanticipated or volatile customer demand”. hence, the required regulatory actions can be performed by obeying the obtained rankings. 5. conclusions in this paper, the sc risk factors, which have the highest risk rates, are aimed to be determined. initially, literature is reviewed and surveys are applied to reveal the possible risk factors. as an output of this part, five main risk factor categories including supply, demand, operational, financial and logistic risks are determined. moreover, to specify the sub-risks, data collection tools are used as meetings with experts and literature review. a couple of meetings are set with experts to observe the sub-risk factors in the case company in detail. besides risk identification and classification, risk assessment stage is also one of the most important parts in risk management process. various techniques are used at this stage and there is no one dominant technique. based on the importance of this stage, a hybrid model is proposed to huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 114 assess the sub-risks of the stated sc risk categories. in the generated hybrid model, ifs are used to evaluate the opinions of dms. it is a well-known fact that fmea is one of the most used methodologies and it is very popular among risk assessment studies. however, there are several deficiencies in this method. especially, the methodology causes a very important shortcoming in terms of weight calculations of severity, occurrence and detection. to eliminate this deficiency, wp is embedded to the method to provide a more accurate calculation basis. thus, an integrated model of fmea and if-ahp is proposed. by this way, revised rpn value is obtained for each sub-risk after the application of the questionnaire method on the managers for determining risk assessment criteria weights. as a result of the proposed methodology, the most important sub-risks have been decided as “inconsistent forecast”, “uncertainty on foreign exchange rate” and “unanticipated or volatile customer demand”. as an output of this paper, a comprehensive sc risk ranking report is provided for the company. the company can use that output, to take precautions or set their own risk priorities. hence, the main earning is basically to define the possible sc risks before they occur. also, this project’s output can be used by similar companies’ sc operations. for the further studies, the specified most important sub-risks can be analyzed deeply to define the source and eliminate the possible effects of these sub-risks in order to prevent any kind of losses. apart from this, different risk assessment methods can be applied in the same business unit and the results can be compared. moreover, the output of this study can be adapted to other industries. contributions: conceptualization, h.s.k.; methodology, h.s.k. and z.t.k.; formal analysis, z.t.k. and h.s.k.; investigation, s.k.c., m.k., s.k., e.u.; writing—original draft preparation, s.k.c., m.k., s.k., e.u.; writing— paper format preparation z.t.k.; writing—review and editing, h.s.k..; visualization, z.t.k. and h.s.k.; supervision, h.s.k. all authors have read and agreed to the published version of the manuscript. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] punniyamoorthy, m.; thamaraiselvan, n. and manikandan, l. assessment of supply chain risk: scale development and validation. benchmarking: an international journal, 2013, 20(1):79-105. doi: http://dx.doi.org/10.1108/14635771311299506. [2] palaniappan, p.k. risk assessment and management in supply chain. global journal of research in engineering, 2014, 14(g2):19-30. https://engineeringresearch.org/index.php/gjre/article/view/1130. [3] diabat, a.; govindan, k. and panicker, v.v. supply chain risk management and its mitigation in a food industry. international journal of production research, 2012, 50(11):3039-3050. doi: http://dx.doi.org/10.1080/00207543.2011.588619. [4] aqlan, f. and lam, s.s. a fuzzy-based integrated framework for supply chain risk assessment. international journal of production economics, 2015, 161:54-63. doi: http://dx.doi.org/10.1016/j.ijpe.2014.11.013. [5] tuncel, g. and alpan, g. risk assessment and management for supply chain networks: a case study. computers in industry, 2010, 61(3):250-259. doi: http://dx.doi.org/10.1016/j.compind.2009.09.008. [6] benjabutr, b. supply chain mapping: the how-to guide. supplychainopz, 2011. available online: https://www.supplychainopz.com/2011/06/supply-chain-mapping.html. (accessed on 12.04.2023) [7] jaberidoost, m.; olfat, l.; hosseini, a.; kebriaeezadeh, a.; abdollahi, m.; alaeddini, m. and dinarvand, r. pharmaceutical supply chain risk assessment in iran using analytic hierarchy process (ahp) and simple additive weighting (saw) methods. journal of pharmaceutical policy and practice, 2015, 8(1):9-18. doi: http://dx.doi.org/10.1186/s40545-015-0029-3. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 115 [8] manuj, i. and mentzer, j. t. global supply chain risk management. journal of business logistics, 2008, 29(1):133-155. doi: http://dx.doi.org/10.1002/j.2158-1592.2008.tb00072.x. [9] li, z.p.; yee, q.m.g.; tan, p.s. and lee, s.g. an extended risk matrix approach for supply chain risk assessment. in industrial engineering and engineering management (ieem), ieee international conference, 2013, (pp. 1699-1704). ieee. doi: http://dx.doi.org/10.1109/ieem.2013.6962700. [10] bocquillon, g. and ekallam, m. risk assessment and mitigation strategies: suppliers and retailers in the swedish organic food market. master thesis, jönköping university, international business school, international logistics and supply chain management, sweden, may 2016. [11] wang, x.; chan, h.k.; yee, r.w. and diaz-rainey, i. a two-stage fuzzy-ahp model for risk assessment of implementing green initiatives in the fashion supply chain. international journal of production economics, 2012, 135(2):595-606. doi: http://dx.doi.org/10.1016/j.ijpe.2011.03.021. [12] khan, o. and burnes, b. risk and supply chain management: creating a research agenda. the international journal of logistics management, 2007, 18(2):197-216. doi: http://dx.doi.org/10.1108/09574090710816931. [13] kara, m.e. and fırat, s.ü.o. supply chain risks: literature review and a new categorization. beykent üniversitesi fen ve mühendislik bilimleri dergisi, 2017, 10(1). https://dergipark.org.tr/en/pub/bujse/issue/30520/33016. [14] chaudhuri, a.; mohanty, b.k. and singh, k.n. supply chain risk assessment during new product development: a group decision-making approach using numeric and linguistic data. international journal of production research, 2013, 51(10):2790-2804. doi: http://dx.doi.org/10.1080/00207543.2012.654922. [15] shenoi, v.v.; dath, t.s.; rajendran, c. and shahabudeen, p. strategic action grids: a study on supply chain risk management in manufacturing industries in india. benchmarking: an international journal, 2018, 25(8):3045-3061. doi: http://dx.doi.org/10.1108/bij-11-2017-0321. [16] rostamzadeh, r.; ghorabaee, m.k.; govindan, k.; esmaeili, a. and nobar, h.b.k. evaluation of sustainable supply chain risk management using an integrated fuzzy topsis-critic approach. journal of cleaner production, 2018, 175:651-669. doi: http://dx.doi.org/10.1016/j.jclepro.2017.12.071. [17] qazi, a.; quigley, j.; dickson, a. and ekici, ş.ö. exploring dependency based probabilistic supply chain risk measures for prioritizing interdependent risks and strategies. european journal of operational research, 2017, 259(1):189-204. doi: https://doi.org/10.1016/j.ejor.2016.10.023. [18] kırılmaz, o. and erol, s. a proactive approach to supply chain risk management: shifting orders among suppliers to mitigate the supply side risks. journal of purchasing and supply management, 2017, 23(1):54-65. doi: http://dx.doi.org/10.1016/j.pursup.2016.04.002. [19] luthra, s.; garg, d. and haleem, a. the impacts of critical success factors for implementing green supply chain management towards sustainability: an empirical investigation of indian automobile industry. journal of cleaner production, 2016, 121:142-158. doi: http://dx.doi.org/10.1016/j.jclepro.2016.01.095. [20] özveri, o. and kabak, m. çok kriterli karar verme tekniklerinin hata modu ve etkileri analizinde kullanimi. muğla sitki koçman üniversitesi i̇ktisadi ve i̇dari bilimler fakültesi ekonomi ve yönetim araştirmalari dergisi, 2016, 4(2). [21] aslani, r.; feili, h. and javanshir, h. a hybrid of fuzzy fmea-ahp to determine factors affecting alternator failure causes. management science letters, 2014, 4(9):1981-1984. doi: http://dx.doi.org/10.5267/j.msl.2014.8.024. [22] soni, g. and kodali, r. a decision framework for assessment of risk associated with global supply chain. journal of modelling in management, 2013, 8(1):25-53. doi: http://dx.doi.org/10.1108/17465661311311969. [23] chan, h.k. and wang, x. fuzzy hierarchical model for risk assessment. principles, concepts, and practical applications. springer, london, uk, 2013. [24] maheswaran, k. and loganathan, t. a novel approach for prioritization of failure modes in fmea using mcdm. international journal of engineering research and applications, 2013, 3(4):733-739. issn: 2248-9622. [25] huang, j.; li, z. and liu, h.c. new approach for failure mode and effect analysis using linguistic distribution assessments and todim method. reliability engineering and system safety, 2017, 167:302–309. doi: http://dx.doi.org/10.1016/j.ress.2017.06.014. [26] chuang, p. t. combining service blueprint and fmea for service design. the service industries journal, 2007, 27(2):91-104. doi: http://dx.doi.org/10.1080/02642060601122587. [27] chang, h.s. and hsiao, h.l. examining the casual relationship among service recovery, perceived justice, perceived risk, and customer value in the hotel industry. the service industries journal, 2008, 28(4):513-528. doi: http://dx.doi.org/10.1080/02642060801917646. huseyin s. kilic, et al. / journal of risk analysis and crisis response, 2023, 13(2), 93-116 doi: https://doi.org/10.54560/jracr.v13i2.359 116 [28] su, c.t. and chou, c.j. a systematic methodology for the creation of six sigma projects: a case study of semiconductor foundry. expert systems with applications, 2008, 34(4):2693-2703. doi: http://dx.doi.org/10.1016/j.eswa.2007.05.014. [29] chang, k.h.; chang, y.c. and lai, p.t. applying the concept of exponential approach to enhance the assessment capability of fmea. journal of intelligent manufacturing, 2014, 25:1413–1427. doi: http://dx.doi.org/10.1007/s10845-013-0747-9. [30] wu, j.; tian j. and zhao t. failure mode prioritization by improved rpn calculation method. in reliability and maintainability symposium, 2014 (pp. 1-6). ieee. [31] liu, z.; lai, m.; zhou, t. and zhou, y. a supply chain risk assessment model based on multistage influence diagram: in service systems and service management, icsssm'09. 6th international conference, 2009 (pp. 72-75). ieee. [32] xu, z. and yager, r.r. dynamic intuitionistic fuzzy multi-attribute decision making. international journal of approximate reasoning, 2008, 48(1):246-262. doi: http://dx.doi.org/10.1016/j.ijar.2007.08.008. [33] atanassov, k.t. new operations defined over the intuitionistic fuzzy sets. fuzzy sets and systems, 1994, 61(2): 137-142. doi: http://dx.doi.org/10.1016/0165-0114(94)90229-1. [34] dymova, l. and sevastjanov, p. operations on intuitionistic fuzzy values in multiple criteria decision making. scientific research of the institute of mathematics and computer science, 2011, 10(1):41-48. http://www.amcm.pcz.pl/. [35] abdullah, l. and najib, l. a new preference scale mcdm method based on interval-valued intuitionistic fuzzy sets and the analytic hierarchy process. soft computing, 2016, 20:511-523. doi: https://doi.org/10.1007/s00500-014-1519-y. [36] tuğrul, f. personnel selection utilizing the decision-making mechanism created with the intuitionistic fuzzy topsis method. muğla journal of science and technology, 2022, 8(2):16-21. doi: http://dx.doi.org/10.22531/muglajsci.1158599. [37] xu, z. some similarity measures of intuitionistic fuzzy sets and their applications to multiple attribute decision making. fuzzy optimization and decision making, 2007, 6:109-121. doi: http://dx.doi.org/10.1007/s10700-007-9004-z. [38] sadiq, r. and tesfamariam, s. environmental decision-making under uncertainty using intuitionistic fuzzy analytic hierarchy process (if-ahp), stochastic environmental research and risk assessment, 2009, 23(1):7591. doi: http://dx.doi.org/10.1007/s00477-007-0197-z. [39] yoon, k.p. and hwang, c.l. multiple attribute decision making: an introduction. sage publication, london, uk, 1995. [40] zavadskas, e.k.; turskis, z.; antucheviciene, j. and zakarevicius, a. optimization of weighted aggregated sum product assessment. elektronika ir elektrotechnika, 2012, 122(6):3-6. doi: http://dx.doi.org/10.5755/j01.eee.122.6.1810. [41] thillaigovindan, n.; shanthi, s.a. and naidu, j.v. a better score function for multiple criteria decision making in fuzzy environment with criteria choice under risk. expert systems with applications, 2016, 59:7885. doi: http://dx.doi.org/10.1016/j.eswa.2016.04.023. [42] chen, t.y. a comparative analysis of score functions for multiple criteria decision making in intuitionistic fuzzy settings. information sciences, 2011, 181(17):3652-3676. doi: http://dx.doi.org/10.1016/j.ins.2011.04.030. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style a gamer’s nightmare: an analysis of the sony playstation hacking crisis bolanle a. olaniran department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, usa e-mail:b.olaniran@ttu.edu andrew potter department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, usa katy a. ross department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, usa brad johnson department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, usa abstract the sony playstation hacking crisis presents all too common personal data theft in the digital information age. the hacking necessitates the need for how such a crisis could be prevented in an attempt to safeguard customers’ personal information and ensure trust between client and vendor relationship. the research focuses on assessment of the sony playstation hacking using the anticipatory model of crisis management (amcm). using the amcm principles, it was found that sony corporation could have handled the crisis better. keywords: crisis management, crisis preparedness, anticipatory model, hacking. 1. introduction and rationale organizations inevitably experience crisis and whether or not the organization is prepared for a crisis determines some of the extent of the crisis at hand. scholars argue that a model is needed in order to help stop crises before they arise prompting the creation of the anticipatory model of crisis management. sony’s crisis in april of 2011 provides a significant example for studying the effects of an organization’s crisis to understand better the implications of taking certain actions to alleviate a crisis. sony experienced a security breach of its online service called the playstation network, and millions of customers had personal information stolen including credit card information. sony estimates the losses from the playstation network hacked at $171 million1. the purpose of this paper is to explore sony’s crisis through a framework of the anticipatory model of crisis management to highlight journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 151-159 published by atlantis press copyright: the authors 151 willieb typewritten text received 7 may 2014 willieb typewritten text accepted 30 july 2014 important implications for online service providers in the future. 2. case overview sony corporation is a company that produces several electronics. one of the more popular products produced by sony is the playstation gaming system. sony found itself in a colossal security breach. on april 20, 2011, sony executives started to investigate abnormal activity on the playstation network, which ultimately led to the theft of over 100 million playstation users’ personal information and for some, credit card information2-4. sony shut down the network the day after suspicious activity was detected and although sony released almost daily announcements concerning the system outage, the company waited almost a week (i.e., six days) after initial recognition to release an announcement of the hacking itself 3, 5-6. in the final analysis, sony is reported to have invested approximately $170 million to cover the expenses of caring for the consumers that had been affected, improving the network’s security and customer support, as well as the investigation into the hacking4. the next portion of this case study offers a brief overview of the anticipatory model of crisis management, which is used to examine the effectiveness of sony’s handling of the playstation hacking case. 3. anticipatory model of crisis management organizational crisis is defined as an unpredictable or a major threat that could have a negative effect on the credibility of the organization, the industry or its stakeholders7-9. in essence, crisis is characterize as an event that compromises one’s safety, customers, community, or threatens to destroy public trust in the organization, thus, damaging the company’s reputation10. hence an effective crisis management embodies a proactive approach that include prevention and especially at the pre-crisis phase7, 11-13. therefore, the anticipatory model of crisis management was created to meet this need. the anticipatory model of crisis management, otherwise referred to as the amcm, was originally developed to address crisis and crisis management at the front end rather than after the fact. the amcm contends that while one might not be able to prevent all crises from occurring, emphasis on preventing crisis from happening should still be a major priority. the central position of the anticipatory model is that significant attempts ought to be made to put in place programs that assess possible crisis triggering factors, such as human error and natural disaster among others, while putting in place appropriate plans to handle any crisis if and when they do occur. the original formulation of amcm was initially designed to address crisis emanating from organizational use of technology12. however, the amcm has been extended to other forms of crisis beyond technology and the new anticipatory orientation toward crisis management has moved the starting point for crisis evaluation13. the definition of crisis reflects the sense that the prevention of crisis not only safeguards the public’s health and safety, but also preserves the trust that the public has for organizations to prevent crises by ensuring their products are safe and that their business practices and communication with the public are honest while demonstrating good citizenry in the community in which these organizations exist or operate. with regard to public safety, the anticipatory model implies that best practices are maintained through competent communication within the organization and with the public as a whole. the basic premises and assumptions of the amcm consist of three main factors, namely expectations, enactment, and control. the expectation principle, speaks to the assumptions that people make about certain events12, 14-16. for example, expectations about the likelihood of a crisis happening would determine whether or not one made the provision to put in place a preventive action or countermeasure. however, it stands to reason that assumptions via expectation have the potential to bring about a self-fulfilling prophecy. for example, when organizational decision-makers assume that a particular technology is fail-safe, they err and might relax safety mechanisms and measures, such that additional counter measures become an afterthought and are never put in place to create necessary a buffer or redundant procedures 12, 14-17. regarding enactment, the assumption is that the very action that enables people and organizations can also cause destruction18. this idea pertains to the principles of enactment and expectations, which are germane to the anticipatory model12-15. enactment focuses on a process where a given action is brought about19. the notion of enactment was eventually extended to published by atlantis press copyright: the authors 152 consequences from those actions18. for example, failure to put in place a crisis plan might negatively impact the eventual or ensuing crisis management. with enactment conceived as a retrospective sense-making process, the model contends that the notion of “anticipation” (of crisis) in and of itself is an action, given that it determines the subsequent choices an organization makes based on available information. there is justification for this claim given the fact that decision-makers and especially organizational leaders often find themselves in situations where they have to envision opportunities, threats, and weaknesses in their environment and then take appropriate measures to safeguard their interests. therefore, the model asserts that decision-makers’ actions or inactions with anticipation would result in different outcomes. the third element in the model is the idea of control, which is the degree of power an organization has over events or crises. for example, if it is discovered that a shipment of medicine leaving a factory is tainted or deficient in anyway, control might refer to the company’s ability to stop the shipment before it hits pharmacy shelves. the control component intertwines with expectation and enactment to the extent that expectations influence enactments (decisions or actions) and actions exerts control over crisis situations. the same can be said for all the three major components of amcm that they are hierarchical rather than mutually exclusive. in essence, the model is complex rather than linear and takes into consideration that all the three components are intertwined and interdependent. in sum, crisis prevention requires a thorough reconnaissance of the complexity of relationships within (internal) and their environmental contexts (external). nevertheless, enactment and expectation must be present to facilitate an understanding of the process 12, 18. while enactment consists of specific actions, expectation about an object determines the type of action taken in the enactment process and provides organizations the needed control to handle a crisis. taken together these factors constitute the anticipatory process of crisis planning and crisis aftermath— where the occurrence of a crisis is foreseen and effort is made to avert or at minimum, reduce the impact of the catastrophe. as part of crisis planning and prevention, the issues management perspective and other crisis management literature have acknowledged the usefulness of amcm as a valuable tool in the crisis literature20-21. furthermore, beside amcm usage as an organizational communication tool in gaining and maintaining the public trust, it also serves as a key reminder that crisis prevention is critical and can make the difference in a matter of life and death for community members and other stakeholders. in the next section, the methods of this study are divulged. 4. methods in crisis management research, a common method used is case study. this project used a case study focusing on sony playstation. the researchers utilized and examined accessible news materials from media channels including news reports and stories. the news materials are analyzed in an attempt to track the chain of events in the issue of the sony playstation hacking crisis. the researchers used the tenets of the amcm model to assess decisions sony made during the course of the crisis. case studies involve the process of analyzing in depth, a particular event or phenomenon, such as the sony crisis, by examining detailed information surrounding the event22-24. the goals of this study were to uncover the exact missteps sony made in managing its crisis. thus, the series of decisions in the sony playstation hacking case were arranged on a timeline to better explore the case in its entirety. a timeline arrangement aids the researchers in tracking the steps and narrowing down the areas in which the organization made mistakes. additionally, the timeline technique affords a methodological approach that utilizes assumptions and ideas of the anticipatory model of crisis management (amcm), which were used to investigate and assess sony’s playstation hacking crisis communication and management. therefore, the study analyzed and evaluated accessible news materials through the lens of amcm. from the analysis, implications and limitations of the case study were offered. the following section provides an evaluation of the sony playstation hacking crisis. 5. analysis and evaluation of sony playstation hacking case there are four separate instances of how the tenets of the amcm apply to the ways in which sony handled the situation of the playstation network intrusion. first, sony failed to inform their customers about the breach until a week after the hackers infiltrated the network. published by atlantis press copyright: the authors 153 also, sony failed to inform the customers that credit card information have been stolen or compromised. instead, sony said that they did not believe financial information was stolen. second, sony did not immediately shut down the network when it knew of a possible security breach. third, sony inaccurately accused a hacker group without the proper information. fourth, sony gave a timeline for the network to be fully functional again, which it did not meet. all four of these components of the network crisis provide ample information for organizations to prepare better if they learn through the amcm. enactment and expectations all enlighten the first aspect of the sony playstation hacking crisis. expectations play a huge role in the first element of the sony crisis because consumers expect corporations to safeguard their credit card information when consumers are purchasing a product. however, sony did not meet the expectation principle because the credit card information was stolen from 12 million of the members2 and the hackers threatened to sell the information. expectation was also not met concerning the security breach because sony did not immediately inform its consumers that a security breach had occurred. sony waited one week after the initial breach to inform anyone outside of the organization about the breach. once it was known that the hackers stole credit card information during the breach, it means that there was an entire week where the information of millions of customers was in the hands of hackers and the consumers could not protect themselves. similar to the previous point, consumers expect a notification if there is even the slightest possibility their confidential information could be at risk. consumers’ expectations were not met when sony did not act immediately and prudently on the information it possesses. sony left the sony playstation network up and running while the crisis was ongoing, which affects all of the aspects of the amcm. sony had the control to make sure the security was the best available, consumers expect the best security, and sony could not act because of the lack of security measures, so all aspects of the amcm are present when evaluating sony’s lapse of action. similar to how sony did not inform players of the network being hacked, sony did not immediately close the network when the breach occurred because sony’s security could not detect the intrusion was occurring. sony waited until april 20th before acting on the information about network intrusion25. although, a company may be strategic in not alarming the public, but recent crises has shown for the most part that such a lack of notification is nothing more than mere incompetence26-27. furthermore, if sony had shut down the network immediately on the 17th, then few information would have been stolen by the hackers. the notification, would have also given affected customers the opportunity to take certain actions on their own (e.g., canceling credit cards). sony’s inability to act effectively affects both control of the crisis and expectations discussed in the amcm. consumers expect that a company would take all measures to stop a crisis from spiraling out of control. if the network was hacked, then consumers would expect sony to close any other possible ways the hackers could affect the network, which would probably entail shutting down the network. sony has direct control on whether or not the playstation network functions or not because sony owns the network. failing to act in a manner that is completely within a company’s direct control violates the vigilance test of the amcm. the notion of control also highlights an aspect of the crisis where consumer expectations were not met at the pre-crisis stage. sony may not be able to control whether or not hackers want to hack into a network. however, sony can control whether or not it has the best security in place for the network as highlighted by sony’s commitment to increasing security after the breach occurred25. sony’s lack of effective detection system compromises the security of the entire system. hackers continued to attack for three days while sony was oblivious to the attack. lulzsec, the group responsible for the intrusion, detailed its intentions for the attack as being simple, stating on june 2nd through a post on the pirate bay: “our goal here is not to come across as master hackers, hence what we’re about to reveal: sonypictures.com was owned by a very simple sql injection, one of the most primitive and common vulnerabilities, as we should all know by now. from a single injection, we accessed everything. why do you put such faith in a company that allows itself to become open to these simple attacks?”28. sony revamped its security scheme after the breach occurred, which implies that additional security existed in the first place25 but sony chose not to use the increased security for some reasons. consumers expect their information to be secure with the best sort of published by atlantis press copyright: the authors 154 encryption security especially if the information deals with finances. sony again violated those expectations by not adequately preparing for a possible crisis, which resulted in its 2011 crisis. sony’s third issue when using the amcm was that sony blamed the hacking group “anonymous” when the group had nothing to do with the 2011 breach29. instead of investigating the issue completely, sony decided to initially blame anonymous without the adequate information. sony had previously prosecuted george hotz, an anonymous hacker, for tampering with the playstation 3 to allow it to play unlicensed software, which hotz proceeded to inform other players how to do the same29. sony assumed and believed without credible information that anonymous perpetrated the attack because a text file titled “anonymous” with the contents reading “we are legion,” part of anonymous’ motto, was found in the sony servers after the intrusion. anonymous denied the claim by issuing the statement on may 4, 2011: "if you think anonymous placed the ‘file’ on the psn try this out. right click on your desktop, make a new text file, name it anonymous, and type in the text file, ‘we are legion.’ that done?”28. eventually, lulzsec, accepted the responsibility for the playstation network intrusion30. sony’s false statement implicates the notions of expectations and enactment. consumers expect that an organization knows what caused a crisis and if the organization does not know, then consumers do not want a corporation that falsely accuses individuals or organizations for the shortcomings of the corporation experiencing a crisis. falsely accusing anonymous further hurts sony’s crisis management because it looked as if sony did not know what was going on, which consumers expect of a multibillion dollar company. also, the fact that the crisis was kept secret for a week should have given sony ample time to investigate the problem. thus, sony’s behavior and actions did not meet the consumers’ expectation that sony should be able to provide them credible explanation about the crisis and in a timely manner. at the same time, falsely accusing another organization for the problem makes sony look as if it was not willing to accept its own responsibility for the crisis. sony tried to pin the crisis on a hacking group, which was fairly wellknown, and make the hacking group the scapegoat instead of taking responsibility for how its networks was compromised. the blame shifting and scapegoating strategies by sony not only violates expectations but also hurts sony in the eyes of its consumers. furthermore, sony set a timeline to restore the playstation network and did not meet the deadline. sony vowed to restore the network within a week’s time and did not meet its own expectation. first, this hurt the company’s consumers because consumers expect a technological company to understand how much work is needed to restore a network. instead, sony looked incompetent when it came to knowing how long it would take to restore the network, which did not help sony’s perception immediately after failing to stop a security breach on their network. second, control was affected by failing to meet the timeline because it is completely within the company’s power to meet its own deadlines. sony initially set the deadline at a week25, so sony had control as to when the network needed to be restored because it was sony, not the media or gamers, who had full control on how to handle consumer expectations. sony looked as if it did not have any clue regarding the functionality of its network, the security of the network, and capability of its technicians in repairing the network. subsequently, consumer expectations and hopes were further dashed due to the lack of control demonstrated by sony. finally, enacting the decision to restore the network appeared to be the right thing; however, the company should have put in place measures to meet the self-stipulated deadline. expectations were high and the reestablishing of the network was completely within the control of the company but the slow implementation of necessary protocol to meet the deadline did not bode well and hinders customers and other members of the public’s faith in sony and its crisis management plan. next, the implications of this case study with amcm on a general business psychology level are discussed. 6. implications crisis preparation without consideration of shareholder psychology in crisis response can create unintended and potentially costly consequences31. sony, in adhering to a traditional liability-reduction crisis management model, made this error. as a result what could have been a modest corporate public relations challenge evolved into a major company-wide crisis, eventually costing the company hundreds of millions of dollars. published by atlantis press copyright: the authors 155 manufacturers’ intensive promotion of internetbased gaming technology has resulted in a gaming community dedicated to cooperative online play (e.g. call of duty series, world of warcraft, etc.). in addition to individual gaming manufacturers’ marketing inferences attempting to imply a generally superior technology such as the online gaming community tends to consist of individuals in their late teens to midthirties32. however, this age range has little tolerance for unreliable technology. also, these individuals crave uninterrupted online access for a significant part of their social, educational, and/or family interaction regimes32. adding to this is a generally heightened emotional response to perceived injustices, both personal and social. not considering this volatile mix of dedication, reliance, technological expectation, and emotional reaction into an organization’s pre-crisis planning will likely yield catastrophic results. adherence to traditional procedure and reactive crisis response will no longer suffice in a world dominated by preference for instant-access to social and informational technologies. once the product of a simple checklist, crisis handling must now incorporate psychology and a presumption of instant and significant shareholder interaction via any number of electronic and social-media. corporations and crisis managers must also presume that their shareholder base is literally the entire world. at the very least it will be a significant community of like-minded individuals or entities, which communicate with a rapidity and volume unseen in human history. one component, which cannot be ignored, is the company’s own history and reputation within the industry, customer base, or community-online or otherwise. sony discovered this when its history became a distinct liability. its technological prowess and proclaimed commitment to dependability created a presumption of absolute reliability and trust in the gaming community. a boon for marketing, sales, and resulting corporate investors along with profitability created a presumption of infallibility and complete trustworthiness among consumers. nevertheless, when this expectation was challenged, the result was a perceived breach of faith with consumers that ensued immediate and costly backlash within the gaming community. as seen in the sony debacle, public relations departments and ample pre-crisis social considerations can play as much, or more, of a part in risk mitigation than an organization’s legal department. the shareholder base must not only be advised of actions, they must also perceive a corporate empathy and a sense of the corporation’s dedication to affected parties’ wellbeing. without this, actions shielding the corporation from legal liability on a particular issue will do little to alleviate the expense and public relations issues stemming from a likely flurry of nuisance litigation. meritless litigation, born of a sense of social justice or righteous indignation no less costly or resource intensive than more substantive legal challenges, and may well prove more expensive in the long-term. the issue of ongoing consumer and shareholder confidence is of great importance and should be considered a critical part of comprehensive pre-crisis planning. a few minutes on social media sites such as facebook give graphic illustrations of the tenacity of social memory, correct or otherwise, as it relates to corporate identities. unrelated issues from years past, sometimes decades, are seen circulating on a perpetual basis. it is common to see issues posted to facebook as current, vital, and deeply troubling only to be illustrated as revisions of urban legends and society-wide misinterpretations33. this is especially true of issues resulting from event associations with a significant customer vesting component, the emotional attachment creating a sense of injustice and giving rise to all manner of misperception and uninformed presumption. the 2011 sony crisis is a prime example as the online gaming community continues to view sony as somewhat untrustworthy. responses from various participants at sony’s online gaming forums34 indicate that some gamers remain wary of sony’s dedication to both game support and customer confidentiality. the sony playstation hacking case explicitly demonstrates how issue-based traditional crisis management strategies are no longer sufficient. these strategies presume an event-response relationship; using rigid protocols developed under centralized control structures and with corporate efficiency and liability mitigation the central concerns. this ignores one of the amcm pre-crisis planning basics, integrating fluidity and flexibility into the crisis response plan. a rigid plan cannot accommodate the randomness and emotional responses of human nature. thus, response schedules and logistics outside the scope and scale of the preplanned contingency plan render the plan all but useless. published by atlantis press copyright: the authors 156 given a rigid, centralized organizational philosophy, this resulting chaos can become overwhelming. notwithstanding, some of these challenges can be mitigated to a great degree by implementing amcm, using pre-crisis planning as a springboard for planning, testing, and revision in a beta environment. this allows identification and resolution of potential problem areas, including a more thorough training for crisis management teams. organizations, in general, can no longer afford traditional crisis planning methods. the traditional methods are inherently flawed when faced with modern technologies and societal expectations. in the age of instant global communication and community interaction via any number of social media outlets, stakeholders form opinions and initiate their responses faster than any traditional corporate crisis plan can accommodate. without a switch to amcm prevention centered mode or similar modality, crises that would have been quickly and simply dealt with in years past can, and often will, become insurmountable social and legal burdens. a failure to address human nature, especially in the areas of setting expectations and addressing a desire for regular, relevant information, yields an emotional gap which stakeholders will rush to fill with their own perceptions, presumptions, and conclusions, and which they then immediately begin communicating or sharing with others. the more emotional the stakeholder investment, the faster the information spreads. a popular idea, benevolent or otherwise, can grow to global scale in a matter of hours. this sea of social activity inevitably spawns issues which, regardless of accuracy, must be address by application of corporate resources. sometimes it requires significant resource investment, both in personnel and finances. if the application includes public image damage control, the investment will likely be accompanied by loss of revenue. in the sony play station hacking, the total investment and revenue loss amounted to almost $200 million, with an ongoing public relations challenge. it would have been far easier, and cheaper, to invest in pre-crisis planning model improvements. 7. limitations there are a few limitations to this case study. to start with, the present study employs a case study methodological approach. generalization for a case study is challenging and should be approached with caution23. future studies should be conducted to further analyze the detailed information in order to apply a general conclusion to a mass population. nonetheless, the analysis of the sony playstation hacking crisis provides valuable lessons to other companies that are at risk of hacking or theft of user information on what to do and what not to do when managing this kind of crisis. second, it is possible that a comparison of similar crises would yield more influential results. perhaps, by contrasting how sony has handled a crisis in the past and the playstation hacking crisis, a trend might emerge showing how sony handles crises in general. or juxtaposed, a compare/contrast method would yield information that proves sony took severe missteps for the playstation hacking crisis only. 8. conclusion sony made four primary mistakes when managing the 2011 hacking crisis. first, sony failed to inform its customers about the breach until a week after the incident and sony also failed to inform the customers that credit card information might have been stolen. second, sony did not act immediately to shut down the network. third, sony inaccurately accused a hacker group without the proper information. finally, sony gave a timeline for the network to be fully functional again, which it fails to meet. through the application of each of these missteps to the amcm, it is demonstrated how to prevent the same missteps from happening to another company. a proper pre-crisis communication management plan is integral to handling crises and thus, utilizing the amcm is one way of accomplishing this goal. implementation of the amcm as a pre-crisis focused strategy can increase consumer and shareholder confidence, along with its flexibility in addressing human nature, and consequently may help save the company’s reputation. references 1. m. hachman, sony playstation network hack nabbed personal info, maybe credit card information. pcmag.com (april 26, 2011). http://www.pcmag.comarticle2/0,2817,2384353,00.asp. 2. s. knafo, sony playstation network hack is just the beginning of giant data thefts: experts. huffington post, (2011, may 6). retrieved from published by atlantis press copyright: the authors 157 http://www.huffingtonpost.com/2011/05/06/playstationtheft-sony-hack_n_858355.html 3. j. tessler, sony explains playstation network hack to congress. huffington post, (2011, may 4). retrieved from http://www.huffingtonpost.com/2011/05/04/sonyplaystation-congress_n_857811.html 4. m. yamaguchi, sony playstation network hack to cost $170 million. huffington post, m. (2011, may 23). retrieved from http://www.huffingtonpost.com/2011/05/23/sonyplaystation-network-hack-cost_n_865432.html 5. d. goodin, user data stolen in sony playstation network hack attack. the register, (2011, april 26). retrieved from http://www.theregister.co.uk/2011/04/26/sony_playstatio n_network_security_breach/ 6. m. williams, playstation network hack timeline. pc world. (2011, may 1). retrieved from http://www.pcworld.com/article/226802/playstation_net work_hack_timeline.html 7. l. barton, crisis in organizations ii. (south-western college publishing, ohio, 2001). 8. w. coombs, ongoing crisis communication: planning, managing, and responding. (thousand oaks, ca: sage publications, 1999). 9. k. fern-banks, crisis communications: a casebook approach 2nd edn. (mahwah, nj: lawrence erlbaum, 2001). 10. c. pearson, s. misra, j. clair and i. mitroff, managing the unthinkable. organizational dynamics, 26(2), (1997) 51-64. 11. i. mitroff, crisis learning: the lessons of failure. the futurist, 36(5), (2002), 19-21. 12. b. olaniran, and d. williams, anticipatory model of crisis management: a vigilant response to technological crises. in handbook of public relations, eds. r. l. heath & g. vasquez (eds.) thousand oaks, ca: sage, 2001), pp. 487-500. 13. b. olaniran, and d. williams, the need for anticipatory perspective in crisis communication. in pre crisis planning, communication and management: preparing for the inevitable eds. b. olaniran, d. williams, and w. coombs (peter lang, ny, 2012), pp. 13-17. 14. b. olaniran, the role of perception in crisis management: a tale of two hurricanes. multicultural education, 15(2) (2007), 13-16. 15. b. olaniran and d. williams. (2004). burkian counternature and the vigilant response: an anticipatory model of crisis management and technology. in responding to crisis: a rhetorical approach to crisis communication, eds. d. millar and r. heath (lawrence erlbaum publishers, 2004), pp. 75-94. 16. b. olaniran, and d. williams, applying anticipatory and relational perspectives to the nigerian delta region oil crisis. public relations review, 34, (2008), 57-59. 17. j. scholl, d. williams, and b. olaniran, preparing for terrorism: a rationale for crisis communication center. in community preparedness and response to terrorism, eds. h. o'hair, r. heath, and g. ledlow, (praeger publishers, ct, 2005), pp. 243-268. 18. k. weick, enacted sensemaking in crisis situations. journal of management studies, 25, (1988), 305-317. 19. l. smircich, and c. stubbart, strategic management in an enacted world. academy of management review, 10, (1985), pp. 724-736. 20. r. heath, and j. sultan, pre-crisis management and communication: slippery steps or solid footing? in. pre crisis planning, communication and management: preparing for the inevitable, eds. b. olaniran, d. williams, and w. coombs (peter lang, ny: 2012), pp. 101-123. 21. t. jacques, issue management as strategic aspect of crisis prevention. in pre-crisis planning communication and management: preparing for the inevitable, eds. olaniran, b., williams, d. and coombs, w. (peter lang, ny, 2012), pp. 17-36. 22. j. creswell, research design: qualitative, quantitative, and mixed methods approaches (3rd ed.). (sage publications, ca, 2009). 23. j. maxwell, qualitative research design: an interactive approach. (3rd ed). (thousand oaks, ca: 2013). 24. r. yin, qualitative research from start to finish. (the guilford, ny: 2011). 25. p. seybold, update on playstation network and qriocity. playstation. blog, (2011, april 26). retrieved from: http://blog.us.playstation.com/2011/04/26/updateon-playstation-network-and-qriocity/ 26. b. a. olaniran and j. c. scholl. new england compounding center (necc) meningitis outbreak: a compounding public health crisis. international journal of risk assessment and crisis management, 4(2) (2014), 34-42. 27. b. olaniran, j. scholl, d. williams, and l. boyer, johnson and johnson phantom recall: a fall from grace or a re-visit of the ghost of the past. public relations review, 38, (2012), pp. 153-155. 28. a. martin, lulzsec’s sony hack really was as simple as it claimed. the atlantic wire. (2011, sep. 22). retrieved from: http://www.theatlanticwire.com/technology/2011/09/lulz secs-sony-hack-really-was-simple-it-claimed/42851/ 29. c. williams, playstation hack: sony blames anonymous hacktivists. the telegraph. (2011, may 5). retrieved from: http://www.telegraph.co.uk/technology/sony/8494177/pl aystation-hack-sony-blames-anonymoushacktivists.html 30. a. martin, sony blames anonymous for playstation hack. the atlantic wire (2011, may 4). retrieved from: http://www.theatlanticwire.com/business/2011/05/sonysays-it-found-anonymous-calling-card/37350/ 31. centers for disease control and prevention psychology of a crisis, crisis and emergency risk communication, (2002, september). retrieved from http://emergency.cdc.gov/cerc/pdf/cerc-sept02.pdf 32. entertainment software association essential facts about the computer and video game industry. entertainment software association, (2012). retrieved from http://www.theesa.com/facts/pdfs/esa_ef_2012.pdf published by atlantis press copyright: the authors 158 33. what’s new (ongoing), snopes, from http://wwww.snopes.com/info/whatsnew.asp 34. playstation 3 community forums (ongoing), sony corporation, retrieved from http://community.us.playstation.com/t5/playstation3/bd-p/22012 published by atlantis press copyright: the authors 159 1. introduction and rationale 2. case overview 4. methods 5. analysis and evaluation of sony playstation hacking case 6. implications 7. limitations 8. conclusion references 1. m. hachman, sony playstation network hack nabbed personal info, maybe credit card information. pcmag.com (april 26, 2011). http://www.pcmag.comarticle2/0,2817,2384353,00.asp. 2. s. knafo, sony playstation network hack is just the beginning of giant data thefts: experts. huffington post, (2011, may 6). retrieved from http://www.huffingtonpost.com/2011/05/06/playstation-theft-sony-hack_n_858355.html 5. d. goodin, user data stolen in sony playstation network hack attack. the register, (2011, april 26). retrieved from http://www.theregister.co.uk/2011/04/26/sony_playstation_network_security_breach/ 23. j. maxwell, qualitative research design: an interactive approach. (3rd ed). (thousand oaks, ca: 2013). 32. entertainment software association essential facts about the computer and video game industry. entertainment software association, (2012). retrieved from http://www.theesa.com/facts/pdfs/esa_ef_2012.pdf microsoft word a study on residents’ risk perception in abrupt geological hazard a study on residents’ risk perception in abrupt geological hazard* anping pan college of architectural and civil engineering, wenzhou university wenzhou, zhejiang 325035, p.r. china e-mail: pananping@sohu.com abstract in recent years, there has been an increased frequency and severity of natural disasters, such as typhoons, landslides, earthquakes, etc., now killing and injuring millions of people every year and causing mounting economic losses. the greater tragedy is that many of the losses due to disasters could have been averted. in order to reduce the abrupt geological hazards’ casualties effectively in remote mountain areas, local citizens’ attitudes of disaster prevention and precaution play an important role and cannot be ignored. this study conducted questionnaire survey to understand their disaster perception and actual evacuation behaviors, and aimed to discuss the special geographic environment and social structure of qingyuan county in east china's zhejiang province, investigated the outcrops in areas vulnerable to disasters, and conducts a survey on locals in shimuxia, pingtou, zhangcun, and tanggen villages. the researcher attempted to understand local people’s cognition of and reaction to landslides, mudflows, and other natural disasters as well as hazard perception and coping behaviors. the research results were expected to serve as a reference for the government or local authorities to prepare anti-disaster risk management. keywords: abrupt geological hazard; risk perception; mountainous area; coping behavior. * this material is based upon work funded by zhejiang provincial natural science foundation of china under grant no. y5110022 1. introduction along with global climate changes, environmental degradation and its own economic takeoff and population explosion, people and the environment are increasingly suffering from the effects of natural disasters. although there are some engineering methods to prevent natural disasters, there is no perfect method because natural phenomena often exceed the assumption. china is one of the most geological and meteorological disaster prone countries of the world. the abrupt geological disasters such as rock avalanches, landslide, mud-rock flows, etc. can be easily triggered by a variety of external stimulus, such as intense rainfall, earthquake shaking, water level change, storm waves or rapid stream erosion. the disasters lead to destroying roads, ruining residential area, blocking rivers and resulting in enormous property damage in terms of both direct and indirect costs each year. there is growing scientific evidence that risks due to geological disasters have become increasingly, especially in mountainous regions. because young people leave mountainous areas and go to towns, declining population and aging become serious problems in mountains. and destruction of mountain forests or inappropriate farming practices can accelerate erosion and expose land to the risk of landslides, floods and avalanches. when a disaster happens in such areas, the damage would be serious due to the barrier for delivering rescues resources and ineffectively dissemination of disaster related information. people would be besieged in the villages and couldn’t evacuate to safe places, and saving goods also couldn’t transport to the villages, rescue efforts would be hampered by bad weather, treacherous journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 44-55 published by atlantis press copyright: the authors 44 administrateur texte tapé à la machine received 15 february 2012 administrateur texte tapé à la machine accepted 12 march 2012 administrateur texte tapé à la machine pan anping mountain terrain. this implicates that prevention measures are barely carried out and people in emergency situations can’t be rescued due to insufficient capacities in basic rescue and to the absence of a mountain rescue system. mountainous areas in east china's zhejiang province typically have steep topographies, and are therefore vulnerable to typhoons, landslides and debris flows. for instance, in 2004, typhoon rananim caused countless casualties to the longxi township, yueqing city; and typhoon saomai in 2006 also caused numerous causalities and landslides. thereafter, residents in hillside areas began paying attention to natural disasters. in response to the public's growing concerns over landslides and debris flows, the national governments at all levels have launched many programs aimed at disaster mitigation. although great progresses have been reached in the field of abrupt geological hazard prediction and warning, many shortages still exist now, such as: the operation of excavation has not taken into consideration the willingness and behavior of the excavated. the lack of knowledge attributed from the studies referring to the recognition of risk, needs, and behavioral pattern of evacuees make the decisions made by authorities sometimes turn aside from the reality. under this circumstance, the government and disaster managers need to study the residents' population characteristics and their behavior in emergencies. when it comes to conceptualizing the idea of disaster prevention and rescue, people often emphasize on the rescuing and rebuilding after the disaster. experts on disaster relief have increasingly called for a greater emphasis on prevention as opposed to relief. the government as well as public bodies have also reinforced measures and policies to prevent disasters. in recent years, the chinese governments at all levels have made great efforts in undertaking disasterreduction projects, improving disaster early warning and emergency response, enhancing sci-tech support, strengthening personnel training and disaster reduction work in communities. disaster risk management needs to be motivated and based within governmental responsibilities, but its success cannot be accomplished without the benefits of widespread decision-making and the participation of many others. the public awareness of risk is therefore a necessary condition to engage in disaster risk reduction. people are more vulnerable when they are not aware of the hazards that pose a threat to their lives and assets. levels of risk awareness depend largely upon the quantity and quality of available information and on the difference in people’s perceptions of risk. risk perception among others is an important determinant of the behavior towards risks, e.g. for the decision to take preventive measures. if risk perception of people living in risk prone areas is known, effective information strategies on protective measures can be designed. since g.f. white's (1945) pioneering work, risk perception has been the central focus for many social scientists interested in natural hazard and disaster studies. the choice of adjustment basically depends on how people perceive threats and the associated risks for themselves. although risk perception is not sufficient for predicting successful evacuation, it is an important variable in determining the effectiveness of proposed evacuation projects. there is a diverse literature on risk perceptions and coping capacities that has significantly contributed to our understanding of how population’s act and cope faced with geological hazard-related risks in the last decades, for example, geographer r. kates (1971) first lists a number of factors that may affect risk perception and develop this into a human ecological model of human adjustment to natural hazards. first are the nature and features of the natural hazard involved including its magnitude, duration, frequency and temporal spacing. second are the frequency and intensity of personal experience of past, similar, events. finally, personality actors like fate control, different views of nature and the tolerance of dissonancecreating information are significant. and these factors are independent from the socio-economic environment. mileti (1993) underlines that people respond to a risk or hazard in ways consistent to their perception of that risk. it is their perception that influences behavior or action. lindell and perry (1993) argue that understanding public perception of natural hazards is necessary in order to impact hazard preparedness, and can be a problem because residents of at risk areas often have inaccurate beliefs about the hazard agent and its impacts, are unaware of available adjustments, and may have erroneous beliefs about the effectiveness of the adjustments of which they are aware. janis and mann (1977) prove that adaptive actions are motivated by published by atlantis press copyright: the authors 45 study on the residents’ awareness of the hazard, knowledge of how it can affect the community, and feelings of personal vulnerability to the potential consequences. in china, there are relatively few researches on perception of risk and short of thorough and systematic research at present. zhouqi et al. (2008) take shangbaiyun village, taibai county as example, and studies on the natural disaster perception of village in a mountainous area. sunlili et al. (2010) explore factors of influence on refuge behavior of mountainous area masses from typhoon disaster. hence, to know the hazard cognition of the residents in remote mountain area is conducive to the effective development of disaster reduction education. this paper will use methods from some disaster management theories and literatures to study the disaster management problems. residents’ attention to geological hazards, awareness of disaster prevention, disaster theoretical knowledge, knowledge of emergency behavior and views on disaster risk reduction education are studied through questionnaire investigation. the purpose of this study is to provide the authorities concerned with results and suggestions which can be helpful in the future planning. 2. study areas 2.1. qingyuan county background qingyuan county is located in southwest zhejiang province near the border between zhejiang and fujian. it lies between 27°25′ and 27°51′ north latitude and 118°50′ and 119°30′east longitude (fig.1). it faces longquan city and jingning county on the north. with an extensive hinterland in the rear, it shares borders with fujian province on three directions of east, south and west. the county is 67km across from east to west and about 49km from north to south, and about 1898 km2 in the area. qingyuan is a key forestry county with rich forest resources. the county is composed of 85.5% mountains, 1.2% surface water, 5.6% farmlands, and 7.7% roads, villages and towns. it has high, clustered mountains to the east, north and west, with the altitudes ranging from 330 m to 1800 m. the highest peak, mount baishanzu ranks second highest peak in zhejiang province with an altitude of 1,856.7 meters. it has a low-lying basin in its central-southern part, and capital city in its center. qingyuan county has seven towns, thirteen countryside areas, five residential areas and 344 administrative villages and the permanent population under its jurisdiction reached 203.6 thousand by the end of 2010. the county has a subtropical monsoon climate with four distinct seasons, with an annual mean temperature of 17.6°c, and the mean annual precipitation is about 1721.3mm. from may to november, known as the typhoon season, the rain is brought by the violent whirl wind and that consists about 60% of the total rainfall. it has a frostless period of 245 days per year. 2.2. current situation of geological disasters in qingyuan county qingyuan is a main geological hazard area in zhejiang province where geological environment is complex and geo-hazard occur continually. the types of geological disasters are many and the distribution scope of disasters is wide. the main ones are rock fall, landslide, debris flow, land collapse. in recent years, affected by extreme weather, earthquakes, engineering, construction and other factors, frequent geological disturbances caused serious damage to people's lives and property. for instance, on august 11, 2006, typhoons saomai made landfall with extremely strong wind and large amounts of rainfall, and therefore the rain-triggered debris flows hit shimuxia village of qingyuan county fig. 1. location of the case study area published by atlantis press copyright: the authors 46 pan anping burying about 20 people, leading to direct economic losses of more than 5 million yuan (fig.2). based on field survey, the distribution of geological hazards, geomorphology and geological structure, disasterpoint density, disaster risk as well as the relationship between geological disasters and human activities were precisely investigated, in 2011, the qingyuan government confirmed 168 geological hazard places as disasters monitoring area, which threaten the security of 17992 peoples and may bring about losses of nearing 100 million yuan (fig.3). 3. methodology people's beliefs about geological hazards and what can be done to manage their consequences can be formed and maintained in several ways. one relates to hazard experience. another involves informing people, usually via public hazard education programs, about the hazards they face and the measures and actions they can adopt to mitigate their risk. in order to study the perception of geological risks and its influencing factors, a survey was conducted among persons living in affected areas of qingyuan county. 3.1. questionnaire design and survey procedure the questionnaire is designed to obtain information on the public’s impressions of (1) what geological hazards (i.e. the mudflow or landslide) are and where they occur, (2) the threat from future geological hazards, and fig. 2. the scene of a debris flows at shimuxia village in qingyuan county on august 11, 2006 fig. 3. distribution of potential geological disaster points in qingyua county published by atlantis press copyright: the authors 47 study on the residents’ (3) how to respond in case of an emergency. the questionnaire had three categories of items: demographic background, risk perception and coping behaviors. some questions were designed to give information on demography and the socio-economic context in which respondents live (e.g., age group, gender, profession, etc.). some questions were designed to evaluate the respondent’s knowledge of geological hazards; understand the respondents level of and views on, hazard education; knowledge of any existing emergency plans; communication strategies; and general level of preparedness and responsibility in the event of a crisis. in some questions, respondents were allowed to select a single answer, while in others, they could select one or more alternatives (see section 4). in july 2011, we conducted a survey in which we interviewed a range of people to determine their general level of awareness and knowledge of geological hazards and risks in affected areas of qingyuan county. the research team was composed of 10 students trained in the technique of questionnaires and working under the close supervision of the author. we distributed the questionnaire to 300 inhabitants living in shimuxia, pingtou, zhangcun, and tanggen villages in the areas. a number of criteria were used to select respondents for the survey. the sample was stratified with respect to age, gender, education, profession and geological disasters experience. the local people responded based on their understanding which is very much related to natural disasters and disaster prevention. all respondents were required to complete the questionnaire on the spot, eliminating any chance to information from other people or sources. the overall response rate was 90%, of which 255 questionnaires were properly completed, for a completion rate of 85%. the main reason for refusal to participate seemed to be lack of available time (it took approximately 20 minutes to fill out the questionnaire). 3.2. sample characteristics the final sample consisted of 255 responds (145 females and 110 males). the age of the respondents ranged from 14 to 75 year, results for the total sample (255) were divided into three different age groups: 23(9.0%) participants were 14 to 17 years old (minors group), 197 (77.3%) were 18 to 65 years old (adults group), and 35 (13.7%) participants were older than 65(elderly group). with regard to the highest level of education completed, 36.5% (n=93) participants had not completed secondary education, while 46.3% (n=118) were junior high school graduates, 11.4% (n=29) were high school graduates, and 5.9% (n=15) had some university degree. most of respondents (78.4%, n=200) were farmers, the student population (which include high school, university and recent unemployed graduates) constituted 7.5% (n=19), the highly scholarized population (state officials, teachers, doctors, etc.) constituted 4.7% (n=12), and those in unskilled occupations (drivers, traders, etc.) constituted the remaining 9.4% (n=24). most of the respondents (n=211, 82.7%) reported at least one direct involvement of geological hazards (i.e. the mudflow or landslide) in the past, and their lives were temporarily influenced in a negative way by geological disasters; for instance, their children were unable to go to school because their schools were damaged by the landslides. 58.8% declared that their relatives were injured or their assets were damaged. 3.3. scoring rules we recoded these variables as follows:  socio-demographic and experiential characteristics: gender (male=1, female=0); age groups of participants (minors=0, adults=1, elderly=2); level of education (not completed secondary education =1, junior high school =2, high school =3, university degree=4); having suffered geological hazard in the past (yes=1, no=0); having received information about geological risks (yes=1, no= 0); having participated in disaster preparedness drill (yes=1, no=0).  respondent’s responses are presented based on the different aspects of risk perception tested. for the purpose of brevity, most of the results were summed as percentages. it should be noted that for some questions, responses may total more than 100% since respondents were permitted to select several answers. and the others, items were related to overall feelings of self-worth/self-acceptance. responses were measured on a likert five-point scale ranging from 1 to 5, with higher numbers indicating higher levels of likelihood, severity, worry, trust, etc.. and the format of a typical fivelevel likert item, ranging from ‘strongly agree’ published by atlantis press copyright: the authors 48 pan anping (score = 5) to ‘strongly disagree’ (score = 1). so the point 3 on the scale marked a neutral position. 3.4. questionnaire analysis the collected questionnaire forms were analyzed by using simple sorting procedures and basic mathematics and statistics methods. most of the calculations were done in the computer using spss 13 and microsoft excel software. statistical tests were conducted on a range of risk-related variables, with the significance level determined at p = 0.05. 4. results and discussions the results discussed in this section are based on the 255 questionnaires forms from the residents of sampling areas. the findings from the questionnaires were analyzed part by part. the detailed results are shown below followed by discussions. 4.1. hazard salience qingyuan county is stricken by a great variety of natural hazards; earthquakes, landslides, mudflows, floods, etc. how people respond to a natural disaster occurring or in preparation for another is often a function of their culturally-derived perception from previous training, education, and experiences. in order to determine the extent to which natural hazards are on the minds of the residents, one of the very first questions on the questionnaire asked them to list the likely natural hazards to affect their living villages (multiple choices). as shown in fig. 4, the results indicated that natural hazards were foremost in the minds of residents: 62.7% (n=160) indicated landslides; 59.2% (n=151) indicated mudflows; 53.3% (n=136) indicated torrents; 53.3% (n=100) indicated droughts. as for the other natural hazards, rock fall and earthquakes respectively represented 37.6% (n=96) and 29% (n =74). in studied locations, respondents were emphasizing their own experience compared to other risks or disaster they did not experience themselves, but they got to know through medias like television, etc. in the investigation, almost all respondents, more or less, knew something about natural hazard. but only thirtyone percent of them took the initiative to learn while the others just received the relative knowledge passively. and the level of people's knowledge about disasters reflects their abilities to deal with disaster information and further shapes the awareness and behaviors of disaster prevention and reduction. 4.2. geological hazard risk once a hazardous condition is recognized it must be evaluated to determine the threat or risk it presents. in order to stimulate the imagination of the interviewees and to highlight the lack of risk communication, the respondents were asked what they thought the geological hazards (i.e. the mudflows or landslides) could do to them around the region. the results (fig. 5) demonstrated that most of the population considers that their possessions and services would be affected (houses, schools, water supply, road, crops etc). respondents identified major impacts caused by the geological hazards as: destruction of farmland & forest (76.4%, n =195); damage to buildings (66.7%, n=170); destruction of major road (62.7%, n=160); loss fig. 5. respondent’s perceived effects which impacted on their lives from geological hazards fig. 4. respondent’s perception of natural hazards which commonly affect their villages published by atlantis press copyright: the authors 49 study on the residents’ of lives (54.9%, n=140); accidents due to panic (47.1%, n=120) and; pollution of water (39.2%, n =100). 4.3. hazard information effective disaster risk management depends upon a series of related actions and the means to engage the informed participation of all stakeholders. exchange of information and communication practices play key roles in the realization of these activities. several survey items were designed to assess the amount of information that residents have on the geological hazards they could face, and to determine the most common sources from which they had received this information. as the data in fig. 6 shows, 65.9% of participants said they had received “insufficient” or “very little” information about the effects of a potential occurrence of the hazards, with very few people saying that they felt they had received enough information. participants were then asked to indicate the sources from which they received the majority of their information about risks (table 1). the results showed the most common methods of receiving hazard information. the respondents tended to choose interesting and vivid ways to get disaster knowledge, and publicity and education of geological hazards knowledge played an important role in risk communication. it indicated that 70.6% (n=180) acquired the knowledge of geological hazards by the mass media (tv, broadcast, radio, newspaper, etc.); 60.7% (n=155) was informed by an elderly person in the village; with very few people (11.8%, n=30) acquired the information from internet. table 1 the origin that people acquire the information of the prevention of geological hazards items percentage （%） 1. pamphlets, leaflets, and posters on geological hazards which distributed by local government 50.2 2. disaster prevention education and emergency drill which be held in their villages 43.1 3.village cadre, part-time geological hazards monitor in the village 62.7 4. tv, broadcast, radio, newspaper, etc. 70.6 5. school education 23.5 6. acquaintance, especially elderly person in the village 60.7 7. research institute 19.6 8. self experience 27.5 9. internet 11.8 in the choice of platforms, the results indicated that mass media was the best means of informing the population of a looming crisis. it seemed that word of mouth also was a widespread means of communication of disaster information, while village cadre, geological hazards monitors and acquaintances were commonly mentioned sources. comparatively speaking, as an important place for popularizing the education, the school did not play an important role in disaster reduction education, the reason might be the lack of corresponding teachers in mountainous areas, and need to strengthen the faculty in the future. 4.4. protective behavior the public awareness of disasters is a state of mind after people’s processing information and knowledge of disasters and it will directly influence their attitudes and behavioral tendencies. and disaster preparedness is an essential element of the disaster management program. each type of geological disaster requires clean-up and recovery. the period after a disaster is often very difficult for families, at times as devastating as the disaster itself. families which are prepared ahead of time can reduce the fear, confusion and losses that come with disaster. they can be ready to evacuate their homes, know what to expect in public shelters and how to provide basic first aid. and be better able to cope with the disaster and recover from it more quickly. in order to evaluate respondent’s knowledge on preparing for hazards, we developed a number of questions which consisted of fourteen items (see table fig. 6. amount of information received regarding potential occurrence of geological hazards (%) published by atlantis press copyright: the authors 50 pan anping 2), inquiring about the adoption of different behaviors to reduce the impact of a crisis. the questionnaire was introduced with a sentence inviting respondents to think about a possible future geological hazard. respondents were asked to indicate whether or not they had adopted each of the presented behaviors. table 2 protective behaviors: items items percentage （%） 1. learn which disasters are possible where you live and how these disasters might affect your family. 30.6 2. keep a readily available list of emergency phone number(doctor, work, school, relatives) 33.3 3. learn about your village's warning signals, what they sound like, what they mean and what actions you should take when they are activated 23.1 4. attend a basic first-aid course 15.7 5. develop a family preparedness plan which be known to all family members. 24.3 6. ask someone (local government, relatives, etc.) information about what to do in case of emergency 17.3 7. review possible evacuation procedures with your family if you had to evacuate 46.3 8. store water in plastic containers such as soft drink bottles 32.5 9. store at least a 3-day supply of non-perishable food. 41.6 10. keep a working flashlight and extra batteries operated radio in a convenient place known to all family members 47.8 11. store important objects in a safe place 40 12. store at least one complete change of clothing and footwear per person 34.5 13. make some changes to home 14.9 14. purchase any kind of insurance against geological disasters 34.5 table 2 depicted the frequencies of specific protective behaviors adopted by participants in order to prepare for possible future geological hazards. the behavior that was more likely to be adopted by respondents was ‘‘keep a working flashlight and extra batteries operated radio in a convenient place known to all family members’’ (47.8%, n=122), followed by ‘‘review possible evacuation procedures with your family if you had to evacuate’’ (46.3%, n=118) and ‘‘store at least a 3-day supply of non-perishable food.’’ (41.6%, n=106). ‘‘make changes to the home’’ is the less diffused behavior, with only 14.9% of the respondents adopting this strategy. the mean number of adopted behaviors was 4.35 (std. deviation=2.61, min=0, max=10).about twenty percent of the respondents adopted seven or more than seven behaviors, and 32.5% of the participants adopted two or less behaviors. the results showed that the residents’ awareness of geological disaster risks was generally poor. even in villages where disasters had occurred relatively frequently in the past, the public had often failed to demand the most rudimentary protection. the development of increased public awareness about geological hazards and the understanding of disaster risks are vital elements in any comprehensive strategy for disaster reduction. public awareness should be conducted through all possible means, including in schools, in particular through the media and other official, public, professional and commercial means, at all levels of society. on exploring the number of adopted behaviors by age (fig. 7), chi-square values (x2 = 50.525, p = 0.002), emphasize the differences in preferred protective behaviors between different age groups. it was revealed that most of the younger generation was inadequately prepared for the impending disaster in normal. an important way to deal with them is to strengthen public education on disaster prevention, while the younger have no strong awareness of disaster prevention. 4.5. risk perception of geological hazard the risk perception category contained questions that addressed the characteristics of the geological disaster (i.e. the landslide). the items used in the questionnaire and their key terms are listed in table 3. its required respondents to make risk perception judgments regarding the likelihood that a geological hazard would fig. 7. the number of respondents adopted behaviors by different age groups. published by atlantis press copyright: the authors 51 study on the residents’ occur in their villages in future, the clearness of know the mitigation actions that their can adopt, the severity of consequences that such an geological hazard eruption would cause for their lives, how serious the effects of a geological hazard eruption might be for themselves and their families, and how much they worry about a potential geological hazard. all ratings were made on a 5 point scale, with higher numbers indicating greater likelihood, severity of consequences, and worry. the mean ratings of each item (with 95% confidence intervals) were also presented in table 3. as shown in table 3, on the first set of question, the mean score was 3.29 (s.d. =1.08) out of a total of 5 (1=likelihood very small; 5= likelihood very large). this indicated that villagers on average felt that a geological hazard would likely occur at their hometown in the following five years. the mean score of “know mitigation actions” was 2.70 (s.d. =1.38), which meant that villagers on average did not know the mitigation action clearly during a geological hazard eruption. in the third item, the mean score was 2.66 (s.d. =1.16), which showed that most of respondents did not think they had the capable of controlling the hazard to avoid a huge loss once a geological hazard occurred. in the fourth item, the mean score was 3.22 (s.d. =1.02), which indicated that it would threaten villagers’ life serious once a geological hazard occurred. in the fifth and sixth item, the mean score was 3.17 and 3.38 respective, which meant that if there was a geological hazard eruption, the villagers would be affected the quality of life and/or be brought financial loss serious. and to the last question, the mean score was 3.34 (s.d. table 3. risk perception of geological hazard and respondents' age: items key term description of scale age group mean standard deviation likelihood in the village in which you live, how likely is it that a geological hazard will occur in the following five years? all 3.29 1.08 14 to 17 years old 3.09 1.35 18 to 65 years old 3.35 1.06 older than 65 3.11 1.02 know mitigation actions if there is a geological hazard eruption, do you know the mitigation actions you can adopt clearly? all 2.70 1.38 14 to 17 years old 2.91 1.41 18 to 65 years old 2.64 1.35 older than 65 2.89 1.53 able to control if there is a geological hazard eruption, do you think that you are capable of controlling the hazard to avoid a huge loss? all 2.66 1.16 14 to 17 years old 3.00 1.21 18 to 65 years old 2.59 1.11 older than 65 2.83 1.22 threaten life if there is a geological hazard eruption, to what extent does the hazard threaten your life? all 3.22 1.02 14 to 17 years old 3.04 1.30 18 to 65 years old 3.26 .96 older than 65 3.14 1.19 affect life quality if there is a geological hazard eruption, to what extent does the hazard affect the quality of your life? all 3.17 .98 14 to 17 years old 2.87 1.10 18 to 65 years old 3.21 .96 older than 65 3.14 1.03 financial loss if there is a geological hazard eruption, to what extent does the hazard bring you financial loss? all 3.38 1.13 14 to 17 years old 3.30 1.06 18 to 65 years old 3.40 1.09 older than 65 3.29 1.41 dread in general, how afraid are you of a geological hazard? all 3.34 1.10 14 to 17 years old 2.83 1.23 18 to 65 years old 3.43 1.07 older than 65 3.20 1.13 note: ratings were made on a likert 5 point scale, with higher numbers indicating higher levels of likelihood, severity, worry and self-efficacy. published by atlantis press copyright: the authors 52 pan anping =1.10), which showed that villagers on average were afraid of a geological hazard in general. in order to determine whether age played a significant role in determining various aspects of risk perception, a series of comparisons were made among four groups (all, minors, adults, elderly group). the results of those analyses which were statistically significant also were presented in table 3. there were no clear patterns in the results of these analyses. 4.6. correlational analysis in order to determine whether the measures of sociodemographic and experiential characteristics used in the study were related to the residents’ perceptions of risk, a series of correlations were calculated among these measures (see table 4). as the data in table 4 illustrated, in general, the correlation matrix showed that when the values of know mitigation actions, and able to control increased, the values of “threaten life”, “affect life quality”, “financial loss”, and “dread” decreased. in addition, “know mitigation actions” and “able to control” were positively correlated with each other. secondly, as regards the relationship between education and the other study variables, we found that the value of “education” was positively correlated with the value of “know mitigation actions” and “able to control” (r=0.696 and 0.616 respective; p<0.01), and negatively correlated with the value of “likelihood”, “threaten life”, “affect life quality”, “financial loss” and “dread” (r=-0.544, -0.448, -0.372, -0.437 respective; p<0.01). finally, other significant correlations between the sociodemographic and experiential variables examined in the present study showed that having taken part in disaster drill activities and personal disaster experience was positively correlated with having received information about the geological hazards (r=0.610 and 0.468 respective, p<0.01). the above results indicate that introducing disaster risk reduction strategies through the educational system is one of the key successful interventions. disaster preparedness, prevention and response should be part of the general education curriculum. people in schools, villages and workplaces should be continuously informed and trained to cope with geological hazards. training session in risk reduction for geological hazards table 4. correlations between study variables (n=255). variable 1 2 3 4 5 6 7 8 9 10 11 12 13 socio-demographic and experiential characteristics 1. gender 1 2. age group -.086 1 3. education .187** -.103 1 4.personal disaster experience -.021 -.020 .101 1 5.received risk information -.022 -.051 .072 .610** 1 6.participate disaster drill .007 -.102 -.129* .223** .468** 1 risk perception of geological hazards 7.likelihood .066 -.011 -.544** -.098 -.016 .335** 1 8.know mitigation actions -.139* .009 .696** .097 .028 -.247** -.816 ** 1 9.able to control -.101 -.014 .616** .073 .019 -.274** -.882 ** .885 ** 1 10.threaten life .073 .011 -.448** .019 -.048 .209** .722 ** -.613 ** -.690 ** 1 11.affect life quality .036 .050 -.442** -.017 -.030 .233** .637 ** -.532 ** -.608 ** .744 ** 1 12.financial loss .060 -.011 -.372** .005 -.094 .193** .583 ** -.506 ** -.572 ** .744 ** .588 ** 1 13.dread .169** .052 -.437** -.009 -.012 .304** .746 ** -.662 ** -.721 ** .615 ** .494 ** .509 ** 1 ** correlation is significant at the 0.01 level (2-tailed). * correlation is significant at the 0.05 level (2-tailed). published by atlantis press copyright: the authors 53 study on the residents’ (i.e. the mudflow) complete with simulation exercises and concrete preparation actions should be conducted in mountainous areas. 5. conclusion the present study aimed to investigate disaster preparedness and risk perception among a group of people living in villages located in mountainous areas in qingyuan county, particularly exposed to the risk of abrupt geological hazards (i.e. the landslides). the results showed as following: (1) the natural hazards in qingyuan county were foremost in the minds of residents in the following descending order: landslides, mudflows, torrents, droughts, rock fall and earthquakes. (2) respondents identified major impacts caused by the geological hazards as descending order: destruction of farmland & forest, damage to buildings, destruction of major road, loss of lives, accidents due to panic and pollution of water. (3) most of participants had received “insufficient” or “very little” information about the effects of a potential occurrence of the hazards. mass media and word of mouth were the best means of informing the population of a looming crisis. (4) the mean number of adopted behaviors was 4.35 out of 14, which showed that the residents’ preparedness of geological disaster was poor. (5) villagers on average felt that a geological hazard would likely occur at their hometown in the following five years, and did not know the mitigation action clearly during a geological hazard eruption. they could not control the hazard to avoid a huge loss and would threaten villagers’ life serious once a geological hazard occurred. the villagers would be affected the quality of life and/or be brought financial loss serious, and were afraid of a geological hazard in general. there were no clear patterns in the results as age difference. (6) the variable “know mitigation actions” and “able to control” were positively correlated with “threaten life”, “affect life quality”, “financial loss”, and “dread decreased”. “know mitigation actions” was positively correlated with “able to control”. the variable “education” was positively correlated with “know mitigation actions” and “able to control”, and negatively correlated with “likelihood”, “threaten life”, “affect life quality”, “financial loss” and “dread”. the variable “participate disaster drill” and “personal disaster experience” was positively correlated with “received risk information”. acknowledgements this material is based upon work funded by zhejiang provincial natural science foundation of china under grant no. y5110022 references 1. china meteorological administration. yearbook of meteorological disaster in china (2005–2009). beijing: meteorological press, 2005–2009. (in chinese) 2. janis, i. and l. mann. (1977) decision making: a psychological analysis of conflict, choice and commitment. new york: free press. 3. slovic, p., ed. (2000). the perception of risk. london: earthscan. 4. chan, j. c. l. and k. s. liu, 2004: global warming and western north pacific typhoon activity from an observational perspective. j. climate, 17, pp. 4590– 4602. 5. f. barberi, m.s. davis b, r. isaia, r. nave, t. ricci. volcanic risk perception in the vesuvius population. journal of volcanology and geothermal research 172 (2008), pp. 244–258. 6. mileti, d. (1993) “communicating public earthquake risk information.” in prediction and perception of natural hazards, j. nemec, j. nigg, and f. siccardi (eds). boston: kluwer academic publishers. 7. lindell, m. and r. perry. (1993) “risk area residents’ changing perceptions of volcano hazard at mt. st. helens.” in prediction and perception of natural hazards, j. nemec, j. nigg, and f. siccardi (eds). boston: kluwer academic publishers. 8. kates, r.w., 1971. natural hazard in human ecological perspective: hypotheses and models. econ. geogr. 47 (3), pp.438–451. 9. lindell, m.k. and perry, r.w. household adjustment to earthquake hazard: a review of research, environment and behavior, 2000, 32 (4), pp. 461-501. 10. zhouqi, yuyaochuang. study on the natural disaster perception of village in a mountainous area—take shangbaiyun village, taibai county, as a example. journal of mountain science, 2008,26(5):571～576. (in chinese). 11. sun lili, chen ailian, wang xiangming. factors analysis of influence on refuge behavior against typhoon published by atlantis press copyright: the authors 54 pan anping disaster in mountainous area. journal of natural disasters, 2010, 19(6):165-170 (in chinese). published by atlantis press copyright: the authors 55 microsoft word the risk assessment of the fog disaster in beijing.doc the risk assessment of the fog disaster in beijing haibo hu1 jinjun pan2 1the institute of urban meteorology, cma, beijing 100089, china 2the beijing meteorological bureau, beijing 100089, china abstract: the fog probability in urban area is determined by the observatory data. taking the regular grid to be the assessed unit, the fragility exponential (fe) is computed upon the gird cell, in which the density of the road-net is stood for, and the fe can be modified according to the important facility distributed on the grid cell. the population density is utilized as the index of vulnerability. moreover, the probability, fragile, and vulnerability is composed by 5:2:1 to integrally measure the risk index of fog in urban area. the example use nearly 10 years observatory data of fog in beijing district to assess the disaster’s risk by spatial grid. the result indicates that the high risk area is consistent to the extending of highway and ring-roads, as parts of urban center area and airport, and so on. keywords: the fog in urban area; probability; fragility; vulnerability; the risk assessment 北京地区大雾灾害风险评估∗ 扈海波 1 潘进军 2 1 中国气象局北京城市气象研究所，北京 100089，中国 2 北京市气象局，北京 10089，中国 摘要: 选用大雾观测资料测算城市地区的雾灾危险性指数，以规则网格作为评估单 元，逐网格计算网格区域内的路网密度，以此作为雾灾的空间脆弱性指标，并针对 重点设施的分布情况对脆弱性指数进行空间叠加订正；选用网格内的人口密度作为 雾灾的易损性指标；危险性、脆弱性及易损性按 5：2：1 的分配比例综合测算雾灾 的风险指数。实例研究选用北京地区近 10 年大雾资料按空间网格化评估方法对大 ∗ 北京市科技计划项目“城市突发气象灾害风险防范与快速响应关键技术研究”（z090506016609001）、北京 市自然科学基金“空间改造 gp 适用与城市气象灾害评估的关键问题研究”（9102009）项目资助.扈海波, 男,1970 年 8 月出生,副研究员,博士,地图学与地理信息系统专业,主要研究方向为城市气象灾害、gis 理论及应 用. journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 59 雾灾害风险进行评估，结果表明：北京地区雾灾脆弱性指数的强弱分布与高速路及 环城路延伸方向一致，高速路段、环城路、市中心及机场等地段为雾灾的高风险区 域，北京东南部地区的雾灾风险也相对较高。 关键词: 城市大雾；危险性；脆弱性；易损性；风险评估. 1. 引言 城市是经济实体集中分布的地区，加 之城市交通网密集，工矿企业及各种 交通工具大量排放污染气体及尾废 气，致使空气中的烟尘污染物等类似 凝结核物质浓度加大。王继志等 （2002）认为随着城市的发展，城市 排放作用所产生的大气污染物在城市 及周边地区的聚集，加剧雾的生成， 城市雾的强度在逐渐加大 [2] 。“城市 雾”对城市居民生活质量和安全均带 来较大的影响，尤其对城市交通及居 民出行造成不利，甚至出现人员伤亡 事故。例如，2006 年 1 月在京沈高速 公路，就因大雾的原因发生一起 60 余 辆车追尾相撞 2 人死亡，10 余人受伤 的严重交通事故。“雾”被称为天气 杀手，是一种重要的城市气象灾害风 险源。 从现有文献资料来看，有关大雾 灾害风险评估的研究不多。kanplan 的 气象灾害风险评估三元组理论的基本 要求是“有那些不利的天气条件”、 “出现不利天气条件的可能性”及 “不利天气条件出现后所造成的影 响” [3] ，后两项表明一个概念——灾害 危险性评估，这是风险评估的核心。 一些文献采用多年气象资料分析特定 地区所属特定路段的雾的气候特征， 归纳分析大雾天气的年季、季节、昼 夜等变化特征 [4][5] ，这对揭示雾的规律 性认识是有意的，对风险评估也有借 鉴，但要揭示城市雾灾的真实风险仍 需在此基础上作更深一步的工作。城 市雾灾的风险评估除了要分析雾的气 候统计学特征及规律外，还应着重剖 析城市特殊功能及结构对大雾天气导 致灾难性事故的响应机制，比如，城 市交通路网密度、城市人口密度等指 标所呈现的城市承灾体脆弱性或易损 性分析。分析需要提取城市路网密 度、长度、道路路面空间占有率及城 市人口密度等指标，这些均可在 gis 的支持下，通过空间网格分割下的空 间叠置计算得到。 本文尝试在空间信息技术支持 下，提取一定大小的网格单元的城市 路网特征参数，按一定权重系数方程 换算路网密度条件下的雾灾脆弱性指 数，并叠加上基于 10 年（1996-2006） 统计资料得出的大雾的危险性及体现 城市人口密度的易损性指数，综合测 算城市雾灾的风险指数。 published by atlantis press copyright: the authors 60 2. 方法与步骤 2.1 危险性评估 图 1 城市路网与网格单元划分示意图 在研究区域范围内划分一定大小 的正方形网格单元（图 1）。选用多年 的大雾观测资料，分析大雾的年月际 的变化情况，并根据每个测站的年平 均大雾日数，做出大雾在城市地区的 年平均日数分布图，年平均日数值内 插到每个评估网格单元，每个网格单 元内的年平均日数与最大年平均日数 的比值作为网格区域内可产生雾灾的 危险性指数，归一化指数方程表示为 maxc c d i= 网格 （1） ic 为 该 雾网格内插后的年平均 日 数， maxc 为 雾所有网格的最大年平均 日数。 2.2 脆弱性评价 网格单元面与城市基础地理信息 底图上的交通线路作“空间交”计 算，得到每个网格单元内不同道路类 型的路线长度，以此核算路网密度。 鉴于城市道路类型不一样，其交通拥 堵状况不一，同等强度的大雾对不同 道路类型的交通造成的影响不一样， 也就是雾灾的响应程度不一样，因此 在空间交计算结束后，需要按不同道 路的类型及道路的长度核算“网格正 方形”内包含的路网信息，即可作为 初步的路网密度参数 ∑ = ×= n i ii wlr 1 网格 （2） il 为网格内道路类型 i 的长度，道路 类型可为“高速路、一级道路”等， iw 为道路类型 i 的权重参数。 最后将单网格路网统计参数 网格 r 进行归一化，即得到每个网格单元的 路网密度系数，这个系数为城市道路 系统响应雾灾的脆弱性指标值，不同 网格的脆弱性指标值的空间分布在 gis 中是一种空间图谱化结果（图 3）。 2.3 大雾易损性加权订正 雾灾的易损性与城市人口分布指 数（人口密度、人口数量）等相关， 甚至包括对大雾特别敏感的重要设 施，比如机场、港口、车站等的分 布。有关人口分布指数的网格化计算 方法已有成熟的空间离散化计算方 法，其应用方法可参考文献[6][7]，其 中人口分布指数的空间网格与脆弱性 指数的计算网格在空间地理坐标、尺 度大小上是一一对应的（图 3），但对 类似机场等重要设施响应雾灾害的易 损性指数订正时的网格计算，需要在 方法上做出微小的调整。文献[8][9]在 针对重点设施的易损性评价时，是依 据区域的统计方法来核算的，即某个 published by atlantis press copyright: the authors 61 区域内的重点设施越多，该区域的易 损性基数就越大。这种方法只适合区 域的易损性指数划分，而 1km*1km 小尺度网格区域与这类设施在空间上 可互为叠置（intersect），核算小 尺度网格在空间上包含或交叠的重点 设施，比如包含机场后的易损性指 数，完全可采用空间交运算算出待评 估网格与这类设施的空间相交面积， 然后乘上权重系数，即可得到网格区 域易损性对重点设施的订正结果，即 权重积网格与重点设施相交面网格 rmf ×= （3） 图 3 为脆弱性指数网格图的易损性订 正结果，可看出“机场”及附近区域 在订正后的脆弱性指数明显增强（图 3 的 a 部分）。 2.4 大雾灾害风险指数测算 大雾灾害风险指数的测算综合雾 灾的危险性、脆弱性指数及易损特征 参数，测算方法以网格评估单位进行 单一网格内各指数数值的综合叠加， 叠算公式如下 321 wfwrwdrtotal ×+×+×= 网格网格网格 （4） 有关 321 、w、ww 组合权重的确定， 多数文献采用的是经验估值方法、专 家咨询及打分的方法。这里“雾灾” 的 综 合 风 险 指 数 测 算 权 重 系 数 321 、w、ww 暂用经验估值，按 5： 2：1 的比例进行分配，估值依据如 下：雾灾的危害首先取决于大雾出现 的频率和强度等为表现的原生性致灾 因子的影响，即雾灾的危险性因素， 其权重比例定为 5/8，表明雾灾的风险 主要与大雾天气的出现概率及强度相 关； 2w 的权重比例为 2/8，也间接说 明雾灾风险与城市路网密集程度，尤 其与高速路的空间地理分布特征紧密 联系，取值表明二者之间经验化的关 联程度； 3w 为 1/8，体现了城市人口 等易损性因子对雾灾风险的响应程 度，另外，重点设施易损程度已通过 空间叠置计算在脆弱性指标中进行了 订正，暂不统计这部分易损性指标。 文献[8]在针对冰雹灾害的风险指数换 算时，采取的是 4：3：3 的比例进行 权重分配的，冰雹这种强致灾性天气 的灾害损失程度与下垫面地物分布所 体现的承灾体脆弱性及易损性的关联 程度更高，其承灾体种类更多更广， 而且只要下垫面存有承灾体，最终酿 成灾害的可能性均较大。雾灾的成灾 机制相对较弱的，它是以影响人的视 野导致判断失误，引发灾难性事故而 形成灾害的，因此，以路网密度、人 口密度及重点设施分布为体现的雾灾 承灾体脆弱性及易损性的权重系数不 益标识过高。 3. 方法与步骤 在 北 京地 区图 幅 范 围内 （ 东 经 115.390701º-117.515098º ， 北 纬 39.409352º -41.081614º ），共划分了 210 行、268 列网格，每个网格点大小 为东西跨 0.008 经距，南北跨 0.008 纬 距。 选用北京地区 19 个观测站近 10 年的大雾观测资料，以雾天出现的频 率及频次，测算北京地区雾灾危险性 指数，指数分布图经网格插值计算 后，将指数值赋值予网格单元，计算 出的北京地区雾灾危险性指数分布如 图 1 所示。 published by atlantis press copyright: the authors 62 按高速路、国道、省道、一级— — 四级公路等公路等级测算单位网格 内路网密度指数，以此作为雾灾的脆 弱性指标，不同道路类型选用不同的 权重系数，计算出的脆弱性指数如图 3，从图 3 中可见北京地区的雾灾脆弱 性指数强度分布与环城路、出京高速 公路（机场高速、八达岭、京津唐 等）的路线延伸方向一致，这与实际 最可能发生大雾灾害的路段分布情况 基本一致，二环、三环等环城路段的 脆弱性指数较高，与这些路段路网密 集、人多路杂的地物分布情况相吻 合。 图 2 北京地区大雾危险性指数分布图 图 3 北京地区雾灾承灾体脆弱性指数（订 正）网格分布图 通过与“机场”、“车站”等面 或面缓冲区域的叠加对脆弱性分布图 做易损性指数订正，订正结果见图 3 （a 处），从图中可见机场等重点设 施所处位置出现一高脆弱性区域，这 一订正对风险结果的正确性评价是必 要的。 图 4 为通过空间演算生成的北京 地区人口密度分布图，测算时将各网 格单元内人口密度值的归一化指数 （单网格内的人口密度除以最大人口 密度）作为易损性值，即暂将人口密 度作为衡量城市地区灾害易损程度的 指标。 图 4 北京地区人口密度分布图 图 5 北京地区雾灾综合风险指数融合结果 图 图 5 为在危险性、脆弱性及易损 性基础上按 5：2：1 的比值进行逐网 格单元的综合风险指数叠合计算的结 果分布图。从最终计算出的综合风险 评估结果可知，北京地区雾灾的高风 险区域分布在高速公路、环城路等车 辆流量大、车辆行驶速度较快的区 域。城中心近几年由于城市化的发 展，汽车尾气排放量加大等原因，其 雾灾的风险也出现增加的趋势，这里 路网密布，是雾灾的高风险区域，大 雾天气对这里稠密分布的市民的生活 a published by atlantis press copyright: the authors 63 及健康同样带来不容忽视的影响。北 京东南部地区的雾灾风险指数同样较 大，该地地势较低平，水汽充沛，相 对湿度较大，文献[10]认为此处还受烟 尘堆积及聚集的影响，大雾出现的频 次及强度相对较高，是雾灾的较高风 险区域。 4. 总结 （1）根据评估网格单元内的大雾 出现的频率、频次等指标核算雾灾的 危险性指数；对不同道路类型取不同 的权重系数，测算路网密度，以归一 化路网密度指数作为评估区域的雾灾 脆弱性指数，并针对重点实施的分布 对脆弱性指数进行订正。应用实例的 图示化评估结果表明北京地区脆弱性 指数的强弱分布与环城路及高速公路 延伸方向一致，结果与实际情况相吻 合。 （2）综合风险指数按 5：2：1 的 比例逐网格叠加上“雾灾”危险性、 承灾体脆弱性及易损性指数，指数比 例分配体现各种风险因子对“雾灾” 成灾机制的贡献大小。照此方法测算 的北京地区雾灾风险结果表明：高速 公路、环城路段及“路网密度大、人 口稠密”的城市中心地区为雾灾的高 风险区域；由于特殊的地理位置，北 京东南部地区的大雾出现的频次及强 度相对较高，是雾灾的较高风险区 域。 参考文献 [1]周淑贞,束炯.城市气候学.气象出版 社.1994. [2]王继志,徐祥德,杨元琴.北京城市能见度 及雾特征分析.应用气象学报,2002,13:160169. [3] stanley kaplan, b.john garrick. on the quantitative definition of risk. risk analysis, 1(1):p11-27. [4]何金梅,王冬梅,李晓霞.甘肃省高等级公 路沿线大雾天气气候特征及其预报服务.干 旱气象,2006,24(1):48-52. [5]张飒,冯建设.济青高速公路大雾天气气候 特征及其影响.气象,2005,31(2):70-89. [6]扈海波，王迎春，刘伟东.气象灾害事件 的数学形态学特征及空间表现. 应用气象学 报,2007,18(6):802:809. [7]扈海波,王迎春.基于数学形态学方法的统 计数值空间离散化图谱生成.计算机工 程,2007,33(21):9-11. [8]北京市气象局灾害风险评估小组.北京市 奥运期间突发气象灾害风险评估及对策研 究. 2007，北京市气象局. [9] b.john garrick. the use of risk assessment to evaluate waste disposal facilities in the united states of america. safety science, volume 40, issues 1-4, february-june 2002,p135-151. [10] 张光智,卞林根,王继志等.北京及周边地 区雾形成的边界层特征.中国科学 d 缉 地球 科学,2005,35(增 i):73-83. [11]wang jing-ai, su yun, et al. vulnerability identification and assessment of agriculture drought disaster in china. advances in earth science,2006,21(2):161～168. [12]郭虎,熊亚军,扈海波.北京市奥运期 间气象灾害风险承受与控制能力分析. 气象,2008,34(2):77-82. published by atlantis press copyright: the authors 64 research article 新型冠状病毒肺炎风险分析与危机管理研究应用 risk analysis and crisis management of novel coronavirus covid-19✩ weixi xu*, xuanhua xu central south university, changsha, hunan, china a r t i c l e i n f o article history received 9 march 2020 accepted 25 march 2020 keywords complementary judgment matrix big data decision-making maximum likelihood estimation infectious disease risk a b s t r a c t after the outbreak of new coronavirus pneumonia, all regions have responded to take measures. emergency decision-making plays an important role, directly related to the safety of the people, and more deeply will affect the future of the country. in this paper, we try to use the wisdom of large groups which are distributed in the social network to deal with the relevant indicators of the group big data decision-making model under the maximum likelihood estimation of the complementary judgment matrix, to establish the epidemic risk assessment system of infectious diseases, and to realize the group decision-making and risk analysis based on the big data. 摘要 新性冠状病毒肺炎疫情爆发后，各地纷纷响应采取措施。应急决策的制定举足轻重，直接关系到人民的安全，更深 层次会影响着国家的前途。本文试图利用汇聚分布在社会网络中的超大群体的智慧，通过建立互补判断矩阵的极大 似然估计下对群体大数据决策模型进行相关指标处理，建立传染病疫情风险评估体系，实现基于大数据的群体决策 与风险分析。 © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 8102191226@csu.edu.cn ✩项目来源：国家自然科学基金项目（71671189）；国家自然科学基金重点项目 (71790615, 71431006）；中南大学研究生自主探索创新项目（502221705）。作者简 介：徐未希（2000-）本科生，从事大数据决策理论与方法、信息系统与决策支持 系统、应急管理与决策、风险分析与管理等研究。 关键词 互补判断矩阵 大数据决策 极大似然估计 传染病风险 0. 引言 传染疫情的发生具有随机性、快速扩散性、衍生性等特征， 事件应急是一个多阶段交互过程、面临巨大的不确定性和风 险，应该充分利用汇聚分布在社会网络中的超大群体的智慧 要有效应对和化解，他们的行为大数据中蕴藏着丰富的高价 值知识，挖掘出来形成群智知识，通过一种超大群体智能协 作机制加以利用，以大幅提高事件应急的科学性和有效性。 突发疫情事件的爆发，会引起社会的广泛关注新闻媒体的大 量报道会引起群众们在微博等网络平台上纷纷发表自己的评 论，此即是超大群体智慧的体现，可以将之与专家给出意见 进行整和从而实现对大数据决策的实现。 本文意在构建群体应急决策模型，多从经典的决策方法出 发，加以优化改善以满足应急决策的对大数据时代下群众数 据的利用。建立基于互补判断矩阵的极大似然估计下群体大 数据决策模型，以保证决策质量。 在国外，xu和ren[1]根据负指数函数的前景理论将犹豫模糊 决策矩阵转化为犹豫模糊前景决策矩阵，并结合热力学中的 能量和熵来考虑决策值的数量和质量，提出了一种能够全面 反映应急决策过程的应急决策处理方法。xu等[2]针对重大突 发事件中公众对大群体应急决策属性缺乏关注的问题，提出 了一种对公共属性偏好进行数据挖掘的大群体应急风险决策 方法。li和sun等[3]利用物联网强大的数据采集和处理系统， 构建了应急 响应系统，将突发事件的群体决策问题转化为多 属性群体决策问题，实现共识最大化，有助于提高市政府或 政府应对紧急和严重事件的能力。xu等[4]针对具有未知阶段 权重的多阶段大群体应急决策问题和以区间数表示的偏好信 息，测度区间数相似性将每个阶段的偏好信息进行聚类，导 出相对熵优化模型来计算聚合和阶段权重以综合方案排名。 journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 77–81 doi: https://doi.org/10.2991/jracr.k.200421.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:8102191226%40csu.edu.cn?subject= https://doi.org/10.2991/jracr.k.200421.001 https://www.atlantis-press.com/journals/jracr 78 w. xu and x. xu / journal of risk analysis and crisis response 10(3) 77–81 wang等[5]针对人类在决策过程中面对风险和不确定性下的 心理行为往往被忽略的问题，提出了一种考虑专家在决策 过程中的心理行为的应急群决策方法，并与其他相关方法 进行了比较，说明了专家心理行为在应急群决策中的重要 性。zhang和wang等[6]建立了两个模型分别确定准则权重和 专家权重，将个人知识聚合成集体知识，以此新的决策支持 方法来生成决策知识。angel和jayaparvathy[7]利用随机petri 网建立了火灾疏散应急事故的分析模型，考虑了人为行为和 现实约束，结合实际数据做出应急决策。 而传染病疫情的发生，本文将在构筑模型之后进一步考虑 传染病发生后人们的心理特点及行为，建立传染病疫情社 会风险评估指标体系。该体系将从社会经济，心理因素与 社会稳定三个大方面进行综合评估，同样利用前面建立的 互补判断矩阵的极大似然估计下群体大数据决策模型对专 家和公众给予每项的评分进行数据处理，从而得出各项指 标的具体判断数值。从而，传染病疫情风险评估指标体系 正式建立。 1. 方法原理 1.1. 互补判断矩阵：在多属性决策中, 设 x = {x1, x2,... , xn}为方案集, 且记 n = {1, 2,... ,n}. 考虑专家对决策方案进 行两两比较, 并作出判断. (1)若专家按互反型标度[8]进行赋值, 给出互反判断矩阵 a = (aij)n×n, 它具有如下性质: aij> 0, aji = 1/aij, aii = 1, i, j ∈ n. 若 aij = aikakj, i, j,k ∈ n, 则称 a = (aij)n×n是完全一 致性互反判断矩阵. (2) 若专家按互补型标度[8]进行赋值, 给出互补判断矩阵 b = (bij)n×n, 它具有如下性质: bij > 0, b ij+ bji = 1, bii = 0.5, i, j ∈ n. 若 bikbkjbji = bkibjkbij, i, j,k ∈ n, 则称 b = (bij)n×n是完全一致性互补判断矩阵. 由互 反判断矩阵 a = (aij)n×n通过转换公式[8]bij = aij/(aij+ 1),i, j ∈ n, 得倒互补判断矩阵 b = (bij)n×n. 并且由互补判断矩阵 b = (bij)n×n通过转换公式[8]aij = bij/(1bij), i, j ∈ n, 也可 得互反判断矩阵 a = (aij)n×n. 易证下列定理成立. 定理 1.1　设 a = (aij)n×n是互反判断矩阵, 则通过转换公式 bij = 1/(1+ aji),　i, j ∈ n (1) 可得互补判断矩阵 b = (bij)n×n. 定理 1.2　设 b = (bij)n×n是互补判断矩阵, 则通过转换公式 aij = bij/bji,　i, j ∈ n (2) 可得互反判断矩阵 a = (aij)n×n. 定理 1.3　若 a = (aij)n×n是完全一致性互反判断矩阵, 则通 过 (1)式转换而得到的判断矩阵 b = (bij)n×n是完全一致性互 补判断矩阵. 定理 1.4　若 b = (bij)n×n是完全一致性互补判断矩阵, 则通 过 (2)式转换而得到的判断矩阵 a = (aij)n×n是完全一致性互 反判断矩阵. 定义 1.1设 b = (bij)n×n是互补判断矩阵, 称 a = (aij)n×n为 b 的转换矩阵, 其中 aij = bij/bji,i, j ∈ n. 由于完全一致性互补判断矩阵 b的转换矩阵 a = (aij)n×n为完 全一致性互反矩阵, 其中 aij = vi/vj, i, j ∈ n, 则特征根方程转化为: av nv= (1) 然而, 由于客观事物的复杂性及决策者知识水平的影响, 在 实际决策时决策者所做出的判断往往是非一致性的, 故 (1) 式一般也是不成立的. 因此, 我们可以用下列特征值问题来近似取代 (1)式: av v= l max (2) 式中, lmax为互反判断矩阵 a 的最大特征值, v为 a的最大 特征值所对应的特征向量, 归一化后就是 a 的排序向量, 显 然, 它也是互补判断矩阵 b 的排序向量. 我们称由此导出的 互补判断矩阵 b 排序向量的 方法为互补判断矩阵排序的特 征向量法, 简记为 cem. 易知下列结论成立. 定理 1. 5设互补判断矩阵 b = (bij)n×n, v = (v1, v2,..., vn)t, 是互补判断矩阵 b 的排序向量 (由 cem 法求得), 若对任 意 k有 bik ≥ bjk (bik ≤ bjk), 则 vi ≥ vj (vi ≤ vj); 且当前者所 有的等式成立时,有 vi = vj. 为了求得互补判断矩阵 b的排 序向量 v = (v1, v2,..., vn) t (即求解(2)式), 我们给出下 列迭代算法: 1) 对于给定的互补判断矩阵 b = (bij)n×n, 通过转换公式 aij = bij/bji, i, j ∈ n,得到相应的转换矩阵 a = (aij)n×n. 2) 任取初始正向量 v (0) = (v1 (0), v2 (0),..., vn (0))t, 给定迭代精度x. 置 k = 0. 3) 计算, d0 = maxi{vi (=0)}, v (0) = v (0)/d0. 4) 迭代计算 v(k + 1) = a v (k),dk+ 1 = max i {xi (k + 1)}, n (k+ 1) = v(k + 1)/dk+ 1. 5) 若|dk+ 1dk|<x, 则进行下一步; 否则, 置 k = k + 1, 转 4). 6) 将 v (k + 1)归一化, 即 v v k v k i n i= + + = å( ) \ ( ).1 1 1 7) v 为转换矩阵 a的排序向量, 也即为互补判断矩阵 b的 排序向量. 8) 结束. 1.2. 极大似然估计 由于正态分布的样本集都是随机抽取的相互独立的样本，由 此，可以只考虑一个样本集 d，通过估计参数向量 q，进行 联合密 度函数 p(d|q)，记已知样本集 d 公式如公式(3)所 示，联合概率 密度函数 p(d|q)称为相对于 d的θ的似然函数 如公式(4)所示。 d x x x x n={ }1 2 3, , ... (3) 1 1 2 3 4 ( ) ( | ) ( , , ... | )q q q q= = = = õp d p x x x x p d|n i n ( ) (4) w. xu and x. xu / journal of risk analysis and crisis response 10(3) 77–81 79 对公式(2)进行求导，并求出在参数空间中能使似然函数 l（q）最大的值记作q�， 为q 的极大似然估计量，如公式(5) 所示。 q q q� = ¼ = = õd x x x x d i n ( , , , , \ )1 2 3 1 n p |( ) (5) 1.3 正态分布的极大似然估计 所抽取的正态分布数服从于正态分布 n，其最大似然函数为 公式(6) l u e e i n x n x u i n i ( , ) ( ) ( ) ( ) s ps ps m s s2 1 2 2 2 1 21 2 2 2 2 2 1 2 = = å = õ = (6) 对公式(6)进行两边取对数，得到公式(7). lnl u n ln n x u i n i( , ) ( ) ( ) ( )s p s s 2 2 2 1 2 2 2 2 1 2 = = åln (7) 对(7)公式进行求导，得到(8),(9)： a s a s lnl u u x u i n i ( , ) ( ) 2 2 1 1 = = å (8) a s as s s ln( , ), ( ) u n x x i n i 2 2 2 4 1 2 2 1 2 = + = å (9) 联合公式(8)、(9)，令(8)、(9)等于 0，得出唯一值，其方 程分别为 (10)、(11)。 u x n x i n i= = = å 1 1 (10) s = = å 1 1 2 n x x i n i( ) (11) 此极大似然估计相关方程有且仅有唯一解（u,s ），且其为 此似然方程的最大值点，因为|u→¥|或 s 2→¥或 0 时，非负 函数 l（u,s 2）→0，因此，u 和 s 2 的极大似然估计近似点 为（u,s 2）。 极大似然估计是建立在极大似然原理基础之上的一种经验证切 实可行的统计方法，是概率论在统计学中最为常见的一种应用 [9]。 极大似然估计提供了一种给定观察大量数据来评估模 型参数的方法较适宜用于大数据样本处理。 经过重复多次实 验，观察其结果，利用实验所得结果某个参数值进行拟合出 使样本出现概率最大化的估值，即被称为极大似然估计[10]。 2. 建立基于互补判断矩阵的极大似然估计下群 体大数据决策模型 2.1. 关键词提取技术 特大突发事件一旦爆发, 微博平台形成实时热点话题, 代表 公众观点的微博文本极速增长形成大数据文 本流. 对每条 博文进行关键词提取可了解公众对事件的关注主题. tf-idf 是数据挖掘领域广泛使用的关键词提取技术, 通过考虑词频 (tf)与逆文档频率(idf)确定关键词权重, 评估词条在文档 集合中的重要性.[11] 对某特大突发事件, 设相关的微博文本量为m的文本集合 d = {d1,d2,...,dm}. 对每一条博文进行分 词、清洗、词 性标注与实体词识别, 得到由n个实体词组成的集合di = {w i1,w i2,...,w in}, 其中w ij为第i个文 本中的第j个实体词, 词条文本矩阵为 d d d d w m n n m m mn w w w w w w w w w = é ë ê ê ê ê ù û ú ú ú ú 1 2 11 12 1 21 22 2 1 2 � � � � � � � � . 定义 2.1 td-idf关键词提取函数为 w d d tf d idf d( , , ) ( , ) ( , ),w w w= ´ (1) idf d( , ) ,w = æ è ç ö ø ÷1g n dfw (2) 其中词频tf(w ,d)为词条w 在文本d中出现的频率, idf(w ,d) 为逆文档频率, n表示文本集合d中短文本总数, dfw 表示文 本集合中出现词条w 的文本数. 2.2. 基于互补判断矩阵的极大似然估计下群体 大数据决策模型模型构建 step 1: 针对某特大突发事件, 采用tf-idf技术对微博平台 上公众发布的文本大数据进行关键词提取, 挖掘与分析公众 关注主题, 获得方案评估准则c = {c1,c2,...,cn}及其权重 w = (w1,w2,...,wn)t.决策专家根据掌握的信息, 以互反 型标度对方案各属性进行判断.根据点估计的优良型准则进行 多次打分评估. step 2: 每一个决策者m(m = 1,2,...,m)在每一个属性 n(n = 1,2,...,n)下提供两两比较的方案并形成相应的互补 判断矩阵. step 3: 将收集到的所有数据运用matlab进行正态分布拟合， 根据不同权重算出每个方案下各项指标的评分融合成为一个 矩阵. step 4: 利用 cem 排序法求解, 则得互补判断矩阵的排序向 量，并最终得到方案排序. 3. 算例分析 针对武汉肺炎疫情发生，在微博上各民众的意见以及一百名 专家的意见作为大数据样本，假设一共有4种方案供选择，下 面运用此模型进行群体智能决策. 四种方案假设如下 x1: 派遣大量武警官兵与医疗救援队迅速进入病源区, 对民进 行救助并广泛组织群众进行及时就医隔离. 同时, 调派武警 官兵进行道路严守控制人来人往，并及时支援救援物资，国 80 w. xu and x. xu / journal of risk analysis and crisis response 10(3) 77–81 内其余核心医疗力量应迅速进行支援到达病源区域再进行分 区域调配，确保每一个病源区域都有较为充足的医疗资源。 x2: 组织部分医疗救援队与突击应急小分队一起, 对重症病 源区域迅速进行救助, 同时, 调派武警官兵严格控制人流来 往, 调派外界救援物资, 待突击小队反馈有效信息后, 进一 步委派分配国内其余医疗力量进行支援。 x3: 成立突击应急小分队, 深入病源区域, 了解具体情况, 并指导人民进行自我隔离, 同时, 调派武警官兵严格控制人 流来往, 调派外界救援物资, 待突击小队反馈有效信息后, 进一步再部署医疗救援队，进行核心病源区域支援。 x4：派遣直升机对灾区进行物资空投，通过无人机空中观 察，得到的情况供远程指挥部评估，进一步再部署医疗救援 队和武警官兵，前往核心病源区域支援。 针对以上四个方案, 大群体应急决策专家确定“成本效益” 、“方案时效性”、“症状轻重控制水平”“工作人员传染 风险”为评价方案的四个属性, step 1: 经过tf-idf技术对微博平台上公众发布的文本大数 据进行关键词提取，进行决策群体聚类的大数据处理近似,确 定四个属性的权重为: w = {0.1,0.25，0.3，0.35}. step 2: 将提取数据中民众和专家的数据观点对四个方案每一 项评分，构成数量足够多的(4×4)互补判断矩阵ai1j1，ai2j2... ajnin。 step 3: 将收集到的所有矩阵运用matlab进行正态分布拟合， 根据不同权重算出每个方案下各项指标的评分融合成为一个 矩阵，现举例矩阵为 b = é ë ê ê ê ê ù û 0 5 0 7 0 6 0 8 0 3 0 5 0 4 0 6 0 4 0 6 0 5 0 7 0 2 0 4 0 3 0 5 . . . . . . . . . . . . . . . . úú ú ú ú step 4：利用 cem 排序法求解, 则得互补判断矩阵b的排序 向量为 v = (0.4303, 0.1799, 0.2748, 0.1150)t, 于是求得的 方案排序均为 x1> x3> x2> x4, 最优方案为 x1。 4. 建立传染病疫情风险评估体系 4.1. 传染病疫情社会风险构成 基于上述互补判断矩阵的极大似然估计下群体大数据决策模 型，探讨了对大数据的决策方案，下面将运用此模型，进行 对风险指标的相关评估。 传染病疫情发生后，我们主要关心四个方面，政府效力，社 会经济，心理因素以及社会稳定，其中，政府效力是最重要 的一个方面，直接关系着上层决策制定的执行力与公信力， 居于首要考虑位置。 而心理因素指的是疫情使人们产生恐惧、害怕等心理，也会 影响个体行为，如医护人员与普通人员都会有不同的焦虑 等，传染病造成的个体的这些心理及行为，从整个社会角度 来看，属于社会心理风险范畴。 传染病发生后，需要对传染病进行预防和救治，可能导致一 些必备药品、物品短缺，供不应求，从而造成市场供求失 调，物价上涨，将对整个社会经济产生极大影响，即传染病 社会经济风险。 最后，传染病发生后，人们的各种恐慌心理所导致的行为， 以及物价的虚高与人群恐惧，势必会对社会秩序和治安造成 一定的影响，如违法犯罪行为增加，群体性事件，而医疗指 标安全也是因素很重要的一方面，这些共同构成了社会稳 定，造成严重的社会稳定风险。 4.2. 传染病疫情社会风险指标体系建立 本指标体系将基于xu现有风险指标体系[12]，于层面和详细 内容进一步完善相关指标，根据上文建立的互补判断矩阵的 极大似然估计下群体大数据决策依据，将微博上各民众的意 见以及一百名专家，再次对各项指标进行打分，作为大数据 样本，依据各项评估水平建立模型如下： 4.3. 传染病社会风险评估 传染病疫情暴发后，唯有准确地评估各层面的风险，才能更 好地应对疫情各方面的社会心理等方面变化。评估指标采用 的评估方法极为重要，本文在构建了传染病疫情政府效力， 心理因素，社会经济，社会秩序风险指标后，再将上文建立 的互补判断矩阵的极大似然估计下群体大数据决策模型进一 步运用。 社会风险评 估体系 一级指标 二级指标 三级指标 四级指标 政府效力 有效法措 实行 相关数据 遵守率 打击谣言力度 管理职责 领导公信力 通告宣传率 人民信任度 社会经济 个体指标 个人财富 平均收入 平均损失 个人职务 平均上班率 市场指标 物品供应 物价水平 生活用品供应 必备药品供应 宏观经济 失业率 通货膨胀率 心理因素 人民心理 患者情绪 负面心理人数比例 积极心理人数比例 未感人群 认知程度 恐慌程度 医护人员 心理 工作时间及 强度影响 平均工作时间 平均照顾病人个数 工作风险 工作感染率 社会稳定 医学指标 安全 症状数据 传染率，病死率 重轻症比例 传染地区 分布情况 区域致病率 保障以及 制度 治安情况 犯罪率，破案率 警察人数 社会保障 政府药物发放以及 相关补贴 商场正常营业率以 及物资供给情况 w. xu and x. xu / journal of risk analysis and crisis response 10(3) 77–81 81 具体步骤如下： 步骤一：民众，专家打分法，采用 1–9 标度法对每一指标相 对于上级指标的重要程度进行打分，再通过几何平均法求出 各层指标的权重。 步骤二：根据实际情况，将每一个决策者m(m = 1,2,…,m) 在每一个属性n(n = 1,2,…,n)下提供的数据评分结合并形成 相应的互补判断矩阵. 步骤三：根据求出的各层指标权重，对无量纲化指标值逐层 加权计算，最后得出一个总的社会风险值。 步骤四：对数据模型进行一致性检验，选取满足在区间中的 预估值进一步拟合。 步骤五：再次通过模型计算出各方面风险评估值后成的向量 组b t=（x1，x2...xn） 然而结合实际，对于有多个决策者参与的具有风险的应急决 策，若不筛选分类大群体进行偏好，大群体中存在的风险喜 好者或风险厌恶者会给决策带来极大风险，导致最终决策结 果失去信服力。[13]故此处应考虑进行修正，考虑决策者风 险偏好对语义值的影响。 步骤六：由于风险偏好值所属不同决策者的差异性，采取层 次聚类法分离出风险喜好者、风险中立者和风险厌恶者三个 组别。之后选择所有风险中立者作为应急决策群体，接着利 用不同决策者风险效用值相似度进行聚类，结合以下风险偏 好度测算公式得出应急决策群体的组成结构（其中决策者的 风险偏好值是将决策者的预案 风险效用值 g (ai) 和预案偏 好值 v(ai) 组合而得），剔除掉风险极度偏好者后得到打分 数值向量组b *，将b *和b t进行比较，若而二者接近则予以信 任，若偏差较大则考虑剔除风险极端偏好者数据重新代入步 骤二往复计算。 r( ) ( ) ( ) a a a v v i i ii p= =å 1 (1) d g r( ) ( ) ( )e a ai i ii p = ´ =å 1 (2) 通过各项风险值，检疫工作人员将通过与之前的重大疫情风 险值发生比较，在实际工作中科学、有效地应对不同风险等 级，可以通过不同途径对风险管理措施不断完善。 5. 结束语 本文针对大群体应急决策过程中庞大的数据网络，从矩阵采 样数据思想的角度出发，提出了一种社会网络环境下基于互 补判断矩阵的极大似然估计下群体大数据决策模型.一方面， 该模型考虑了疫情下大群体应急决策基于社会网络的广泛 性，通过数据采集技术和极大似然估计，结合属性权重和排 序向量的选取，整合成一个矩阵，在降低了决策复杂度的同 时也使得各方面数据权重的确定更为客观.另一方面，模型应 用于后面传染病风险评估模型的数据处理，更方便整合数据 以便呈现出各指标的风险标度。 本文提出的模型具有以下一些特点: 1) 将基于方案的互补矩阵判断方法和极大似然估计相结合, 考虑到不同属性下决策者有不同的权重, 使用几何平均 法计算。 2) 通过一个迭代算子, 可以使得决策群体自动达到群一致 性水平, 从而得到每个属性下群体的矩阵数据类型, 并 再次利用此模型求解每个属性的权重; 3) 通过排序向量求出的方案比较, 保证了决策结论的客观 性. 另外,不同方案的不同属性的差异性导致计算可能有 一定偏差，将考虑怎样进行复合矩阵不同数据类型的聚 合以及偏差修正，在此基础上进行进一步的分析研究. 由于互联网上数据庞大，从中筛选出有用的数据并进行整 合，以及提供给专家进行决策的必要性极大，更好地体现 了群体智慧，本文运用关键词提取技术，建立了互补判断 矩阵的极大似然估计下群体大数据决策模型，进一步地， 为了更好的对传染病造成的各种、社会风险进行评估，本 文从政府效力心理因素、社会经济以及社会秩序稳定四个 层面进行分析，详细构建了每一个层面下的传染病疫情社 会风险指标体系，并通过前文建立的决策模型进行数据处 理和往年数据的比较，为政府做出不同的级别疫情响应政 策以供参考。 参考文献 [1] ren p, xu z, hao z. hesitant fuzzy thermodynamic method for emergency decision making based on prospect theory. ieee trans cybern 2017;47:2531–43. [2] xu x, yin x, chen x. a large-group emergency risk decision method based on data mining of public attribute preferences. knowl based syst 2019;163:495–509. [3] li n, sun m, bi z, su z, wang c. a new methodology to support group decision-making for iot-based emergency response systems. inform syst front 2014;16:953–77. [4] cai c, xu x, wang p, chen x. a multi-stage conflict style large group emergency decision-making method. soft comput 2017;21:5765–78. [5] wang l, wang ym, martínez l. a group decision method based on prospect theory for emergency situations. inform sci 2017;418–419:119–35. [6] zhang l, wang y, zhao x. a new emergency decision support methodology based on multi-source knowledge in 2-tuple linguistic model. knowl based syst 2018;144:77–87. [7] sheeba aa, jayaparvathy r. performance modeling of an intelligent emergency evacuation system in buildings on accidental fire occurrence. saf sci 2019;112:196–205. [8] 徐泽水. ahp中两类标度的关系研究. 系统工程理论与实践 1999;19:97–101. [9] little rja, donald br. maximum likelihood for general patterns of missing data: introduction and theory with ignorable nonresponse. in: statistical analysis with missing data. 3rd ed. john wiley & sons, inc.; 2009. [10] açıkgöz i. parameter estimation with profile likelihood method and penalized em algorithm in normal mixture distributions. j stat manage syst 2018;21:1211–28. [11] salton g, buckley c. term-weighting approaches in automatic text retrieval. inform process manage 1988;24:513–23. [12] 徐选华. 传染病疫情社会风险评估指标体系研究[j].第八届 (2013）中国管理学年会——管理与决策科学分会场论文 集,2013.11.08. [13] 徐选华,杨玉珊,陈晓红.基于决策者风险偏好大数据分析的 大群体应急决策方法.运筹与管理,2019;28:1–10. https://doi.org/10.1109/tcyb.2016.2638498 https://doi.org/10.1109/tcyb.2016.2638498 https://doi.org/10.1109/tcyb.2016.2638498 https://doi.org/10.1016/j.knosys.2018.09.010 https://doi.org/10.1016/j.knosys.2018.09.010 https://doi.org/10.1016/j.knosys.2018.09.010 https://doi.org/10.1007/s10796-013-9407-z https://doi.org/10.1007/s10796-013-9407-z https://doi.org/10.1007/s10796-013-9407-z https://doi.org/10.1007/s00500-016-2155-5 https://doi.org/10.1007/s00500-016-2155-5 https://doi.org/10.1007/s00500-016-2155-5 https://doi.org/10.1016/j.ins.2017.07.037 https://doi.org/10.1016/j.ins.2017.07.037 https://doi.org/10.1016/j.ins.2017.07.037 https://doi.org/10.1016/j.knosys.2017.12.026 https://doi.org/10.1016/j.knosys.2017.12.026 https://doi.org/10.1016/j.knosys.2017.12.026 https://doi.org/10.1016/j.ssci.2018.10.027 https://doi.org/10.1016/j.ssci.2018.10.027 https://doi.org/10.1016/j.ssci.2018.10.027 https://doi.org/10.1080/09720510.2018.1496520 https://doi.org/10.1080/09720510.2018.1496520 https://doi.org/10.1080/09720510.2018.1496520 https://doi.org/10.1016/0306-4573(88)90021-0 https://doi.org/10.1016/0306-4573(88)90021-0 atlantis press journal style received 26 july 2015 accepted 18 august 2015 *corresponding author: fangchao@tsinghua.edu.cn crisis communication about nuclear accidents with psychological approaches yanran yang, lina jin, jinbin li, chao fang * institute of nuclear and new energy technology, collaborative innovation, center of advanced nuclear energy technology, key laboratory of advanced, reactor engineering and safety of ministry of education, tsinghua, university, beijing 100084, china abstract different from other natural disasters and health emergencies, nuclear accident is a kind of special crisis of organizational crises, which tends to generate public and media interests and criticism more easily. in this paper, we reviewed literatures on risk and crisis communication and analyzed the cases of risk and crisis communication after three famous nuclear accidents (chernobyl, three mile island and fukushima) and studied the public risk and crisis communication after these accidents with psychological methods, including (1) discussing the source of nuclear panic with needs theory and cognitive theory and (2) studying the irrational behaviors in nuclear accidents with psychoanalytic theory. after the above analysis, some methods of public risk and crisis communication about nuclear accidents are given. furthermore, it shows that new media such as wechat and microblog play a very important role in this process. keywords: risk communication, nuclear accidents, needs theory, new media 1. introduction risk and crisis communication is of great essence to public affairs. when health emergencies, natural disasters or other big emergencies happen, effective risk and crisis communication can help to reduce the fears and panic among public and inform the mass with necessary solutions like evacuation recommendation to protect the public from further damage. after the sars and hini, chinese government has formed a rather mature and complete working system of solving the health emergencies including the information disclosure of the outbreak of serious infectious diseases on website, which is significant for the social stability. but until now, chinese government lacks welldeveloped risk and crisis communication methods about nuclear accident because there was no terrible nuclear accident yet in china so this work has not been on the agenda (chen, 2012). however, as the biggest country of developing nuclear energy in the world (50% nuclear power plants under construction in the world are in china until 2014), the related risk and crisis communication about nuclear accident should not be neglected by the government and academic (dinget al. 2014, covello et al. 2001). on the other hand, different from the natural disasters and health emergencies, the nuclear accident is a kind of special crisis of organizational crises, which tends to generate public and media interest and criticism more easily. if not being handled appropriately, one failure of crisis communication after nuclear accident will certainly affect public’s confidence in the organization and further restrict the development of nuclear energy. therefore, an effective working model for the risk and crisis communication about nuclear accident should be built in china with taking precautions, especially after fukushima nuclear accident. in this paper, we reviewed the cases of risk and crisis communication after three famous nuclear accident (chernobyl, three mile island and fukushima) and studied the public risk and crisis communication after nuclear accident with psychological methods, including (1) discussing the source of nuclear panic with needs theory and cognitive theory and (2) studying the journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 169-177 published by atlantis press copyright: the authors 169 irrational behaviors in nuclear accidents with psychoanalytic theory. after the above analysis, some methods of public risk and crisis communication about nuclear accident are given in the final. 2. literature review and analysis on crisis communication after nuclear accidents 2.1. researches on risk and crisis communication after nuclear accidents for years, risk and crisis communication after nuclear accidents have attracted a lot of academic attention from various fields, communication, public administration, and psychology, etc. especially after the three most famous accidents, chernobyl, three mile island and the fukushima accidents, scholars spent a lot of efforts on analyzing the communicational behavior of mass media, utility companies, local governments and central government officials, trying to figure out the whole picture of information transmission after the great accidents. harry otway and other four scholars (1988) did a research on risk communication in europe after chernobyl using a media analysis of seven countries, analyzing their media coverage of the chernobyl emergency to identify common communication problems after this crisis. donald j. zeigler and james johnson (1989) analyze the information planning after chernobyl and three mile island and concludes that informational context of an accident will not be the main determiner of human behavior in future nuclear emergencies. rama hoetzlein (2012) drew an information map for the visual communication after the fukushima accidents, to providing more vivid and obvious information to lay people. in cassandra koerner ’s research (2014), she studies how media coverage of accidents at chernobyl, three mile island, and fukushima overwhelmed scientific claims of nuclear safety and security during the production process, trying to bridge the information gap between scientists, the public, and policymakers, by analyzing the titles of newspaper coverage. these researches all use the text material or other media coverage material to analyze the effect of such crisis communication and provide a better method based on their findings. other researches analyzes the different perception of risks of lay people and experts in order to provide better and more specific information in later crisis communication. in skarlatidou, cheng and haklay’s research (2012), they identify that communicating risks is a complex task as there may be significant differences between the risks perceptions of experts and the public. other similar researches did by tanja perko and other fellow scholars (2012, 2014) indicated the nuclear risk communications in different countries and figure out the difference knowledge and risk perception of experts and lay people, providing a better way to inform the public. all the researches mentioned above, focus on the nuclear risk and crisis communication. those focused on the crisis communication after the accidents, mainly use the media coverage material after a specific accident as the sample. by analyzing their communicational influence, they try to build a more effective way to inform the public after a nuclear power plant accident. while the risk communication in normal days are not designed for one specific nuclear emergency but for the public acceptance of nuclear. the risk communication of nuclear power plant, nuclear waste issue and even nuclear industry itself are aimed at a better scientific education on nuclear and radiation. a good risk communication during normal days can enable lay people to better understand the concept of radiation and how to behave against a nuclear emergency, which may make it easier for the crisis communication after the accident. therefore, the risk and crisis communication after the nuclear accident have two aims in total. on the one hand, correct information about the accident should be disclosed to the public to calm down the public and relieve the public panic also evacuation command should be directly sent to the people within the evacuation scope—that is the short-term aim of crisis communication after an emergency. on the other hand, a nuclear power plant accident may alter the public attitude towards nuclear industry. american people’s attitudes after tmi and japanese people’s attitudes after fukushima are typical examples of such attitudes shift. under such circumstances, crisis communication after nuclear accident has to have a long-term aim, which is reestablish the public trust and acceptance of nuclear industry. as donald j. zeigler and james johnson (1989) give it in their study, nuclear risk is not only the consequences of a nuclear accident but also the probabilities of next nuclear accident. published by atlantis press copyright: the authors 170 2.2. the case of chernobyl nuclear accident on 26 april 1986, an explosion happened at the chernobyl nuclear power plant, which was under the jurisdiction of the ussr at that time. this accident was considered as the worst nuclear accident up till now and was evaluated as a level 7 event according to the international nuclear event scale. however, the local government of the city near the plant, pripyat, did not inform the residents at the very beginning after the explosion happened. people of the ussr first heard this news from tv news 2 days later, on 28 april. there was a 20 second announcement saying that there had been an accident at the chernobyl plant and one of the nuclear reactors was damaged. reasons, casualties, effects and radiation levels were said being investigated in this short announcement. before this announcement, it is the swedish detective result that first found the airborne radioactivity which indicating a nuclear accident happened. only after that, did the ussr government admitted the nuclear accident. therefore, the ussr government went against the time first principle in risk communication by concealing the truth to both the domestic citizens and the international neighbors (otway, harry. et al.1988). as for the countries in the neighborhood of chernobyl, most of the authorities in europe that are responsible for risk management of radiation emergencies did not anticipate this accident to be so severe. the emergency plans they set before could not satisfy the need of risk communication after chernobyl accident. many countries encountered the problem that they even did not have clear information about what was happening in their own countries. the quantitative information provided by monitoring stations was different within the same countries due to the inconsistent units they used. besides the confusion caused by radiation report, the special information telephones were so overloaded by the calls that the callers could not get through. one country in europe even published a wrong number of the information telephones. as a result people with special needs and questions could not got the right information they want, which leads to the doubt of the credibility of the government and in turn made the risk communication even more difficult. 2.3. the case of three mile island nuclear accident on 28 march 1979, several water pumps in the nuclear plant of three mile island stopped working around 4 a.m. the breakdown of water pumps later made the core become overheat and the fuels began to melt. at around 7 o’ clock, the station manager gary miller announced that a general emergency had happened and it might cause radiological results and informed the local government. in the morning of that day, the nuclear regulatory commission (nrc) was called to help to deal with the accident. later around 10 o’ clock, the lieutenant governor of pennsylvania state, william scranton ⅲ held a press conference. he said that according to the metropolitan edison company, there had been an incident at three mile island unit 2 and everything was under control and there was no danger to public health and safety with only a small radiation to the environment. during the press conference, the officials did not anticipate a need to evacuate. then on 29 march, the met ed held their first press conference, jack herbein and walter creitz shared some new information about the degree of radiation, which had some conflicts with the information said in the press conference before. then distrust from media had been accumulated. even worse, jack herbein said in the press conference on 30 march that he did not know why they needed to tell each and everything they do to the press. the communication became better and smoother after the harold denton’s coming, the personal advisor of president jimmy carter. after he came to harrisburg, they held a joint press conference where the governor thornburg and harold denton tried to ease public fears. later on 1 april, president jimmy carter and his wife came to pay a personal visit to the three mile island and then gave a speech to public. the end of this accident was marked by a press conference held by governor thornburg on 6 april. he held this conference at the site of three mile island plant and said that it would be the last televised address. during the whole process of risk communication, lots of press conferences were used to deliver information and ease public fears. also, a 24-hour citizen’s information center was set up to answer people’s questions. though some doubt were cast on the metropolitan edison company, most people kept calm attitude during the whole process even when some of them were being published by atlantis press copyright: the authors 171 evacuated, which showed the efficiency of risk communication of the government. 2.4. the case of fukushima nuclear accident on 11 march 2011, a great earthquake of 9.0 magnitude happened in the afternoon. then a serious of tsunamis struck the fukushima coast, causing the fukushima daiichi reactors to shut down. at the night of 11 march, japanese government announced an evacuation order to those who lived within a radius of 3 kilometers from the fukushima ⅰ station. the next day, there was a hydrogen gas explosion in the unit 1 at 15:36, which led to significant radiation releases. after the explosion, the first national press conference was held at around 17:48 in tokyo. nuclear safety commission held the conference and the chief cabinet secretary yukio edano spoke on behalf of the government. he told the public about the existence of explosion and admitted that the causes were still under investigation. in his speech, he gave an evacuation order to the residents who lived with a radius of 3 kilometers from the fukushima ⅱ station. he asked the public to keep calm and avoid going out and do not listen to or spread rumors. though this press conference provided some basic information about what had happened, the whole risk communication process of this fukushima daiichi accident dissatisfied most japanese people. there were a lot of complaints about the delays in releasing radiation data and other important information. the japanese central government as well as tokyo electric power company (tepco) did not cooperate well on the risk communication that they even provided contradictory information to public. furthermore, the conflicts between the central government and tepco built up an information block that they did not do their utmost to share and utilize the information. these conflicts left a negative impression on the public so that they trusted neither of them. this collapse of trust directly led to the mass objection towards nuclear energy after the fukushima accident. 2.5. analysis of communication methods after three nuclear accidents all the risk communication mentioned above shared some flaws in information releasing, public communication and cooperation between government and nuclear plant runners. harry otway and other four scholars (1988) once claimed in their paper in 1988 that modern governments can communicate effectively with the public only through the channel of mass media, especially in emergencies where people must be informed immediately of the developments of the accidents and guided to behave correctly under such circumstances (3). however, as the development of social media and other new media channels, it is much more difficult for government to inform citizens the correct information through mass media. modern governments have to be aware of noises of misinformation, rumors and other distractions during crisis communication nowadays. in the new environment of crisis communication, there are three questions that should be defined before the communication—what is the correct information, what is the information that citizens need and want to know, and what is the best way to transmit such information. as hoetzlein (2012) points out, our primary source for information in times of emergency is the news media. yet, due to physical circumstances resulting from the crisis, information could be scarce or unreliable. one of the most controversial aspects of accidents is the confusing information on the one hand and a lack of information on the other, which should be responsible for accentuating public perceptions of risk and aggravating efforts in crisis managements (donald zeigler & johnson, 1989). natural disaster, such as floodwaters, approaching storms, earthquakes are phenomena that could be gauged by direct sensory inspection. unlike these disasters, nuclear disaster makes it inaccessible for news media and public to get the primary sources of information without the help from technical experts who can read and interpret instrument panels and computer displays. thus, after nuclear disaster, secondary sources of information replace primary ones. that could probably explain why it is so important and urgent for government and utility company to give the correct information right after the accident to public, for otherwise citizens may think they are already affected without even noticing or feeling the radiation. however, due to the complexity of nuclear accident, it is hard to get the “correct” information in the first place, especially when the utility companies conceal some information. such problems can be easily identified in the three cases. the governments and nuclear plant runners either delayed the release of information or gave the wrong one, which all led to the published by atlantis press copyright: the authors 172 doubt from public and collapse of public trust. therefore, a very basic requirement for risk and crisis communication after the nuclear accidents must be the publication of correct information (foa et al. 1986). after knowing what kind of information we have at hand, it is also very important to know what the public needs to know. for nuclear incidents change citizen perceptions toward the safety of nuclear power generation worldwide and may cause sharp declines in the number of individuals who support the development in this field (cassandra koerner, 2012). on determining what should be passed to citizens, government should take the public acceptance and perception of risks into consideration, which calls for more researches before the accidents, like what tanja perko and other scholars have done. also, by doing psychoanalytic researches of public, it is also possible to know what is their urgent need after an accident, which will be further discussed in this paper. last but not least, via what channels and in what forms to transmit the information are also big problems for crisis communication. in different times, choices have to be made according to its social circumstances. in the time of three mile island and chernobyl, special telephone line served as the most efficient way for normal people to solve their problems. when it came to the time of fukushima, website and sns (social networking system) became the most convenient way that normal people rely on to get the specific answer. thus, these channels for mass communication should be pay attention to as well. this field has already attracted a lot of academic attention to analyze the communicational effects of different methods. this paper will also shed a light on the communicational effect of social media in china. 3. the analysis of public nuclear panic with psychological method 3.1. the analysis of public nuclear panic with needs theory and cognitive theory the cause of public nuclear panic is important in risk communication because it shows what should be communicated with public. the maslow's needs theory (maslow, 1943) could give us the answer. in this theory, human motivations are divided into five aspects: physiological, safety, belongingness and love, esteem, and self-actualization (gerriget al. 2007). these five levels are often described like a pyramid, with fundamental needs at the bottom and self-actualization at the top (fig.1). physiological needs including food, water, house and so on, are the fundamental needs in daily life and these basic needs play dominant roles and once it cannot be satisfied, a strong motivation will be aroused to fulfill the demands. safety needs are also essential needs for human beings, which mean that people prefer the world to be safe, predictable, ordered and organized, while all unexpected and dangerous events would bring panic to public. in this sense, both of physiological needs and safety needs should be viewed as fundamental needs and motivations for human beings. fig.1. maslow's hierarchy of needs with this theory, the public panic after nuclear accident originates from the destroying of physiological and safety needs of “pyramid of hierarchy”. when nuclear accident happened and radioactive substance release, the radioactive substance could affect the safety of food, air and water in some ways and making public worry about the basic living. as mentioned above, physiological and safety needs are the base of the pyramid and once they are unsatisfied, the whole pyramid seems unstable and will probably collapse— arousing nerve, anxiety and irrational behaviors. under this circumstance, people feel that their lives are put under great threat and survival become a big problem, which drive them to do all they could to support themselves with a basic living system. such powerful and primitive instincts are actually subliminal thoughts that would control people’s mind, which also push all other higher-level issues into secondary importance. on the other hand, since the radiation cannot be seen or touched, people do not know whether they are under exposure unless using prospecting instruments (golay, m.w. 2001). as a result, people have nothing to do to self actulization esteem belongingness and love safety needs physiological needs published by atlantis press copyright: the authors 173 keep away from radiation and it is easy to make people consider that the contaminated air, water and food are long-time physiological threats to daily life though technically the radiation is not serious and can’t bring any physical and physiological damages. cognitive theory could also be used in explaining the panic to nuclear accident (gray, 1998). it is known that new fear-related information will evoke the fear memory and the brain will integrate them into it. as long as the schema of fear is activated, people will use coping pattern they learned in the past to deal with the new situation. for the general public, nuclear energy has a notorious reputation and has already taken root in the fear structure. disasters like the nuclear bomb and chernobyl accident left horrible impressions and any cue related to these catastrophes will recall them. in other words, what the public fear is not only the new situation alone, but the combination of the new accident and previous memories of disasters. for instance, although the fukushima accident has little similarity with chernobyl, people treat it as seriously as chernobyl because former accidents are in the deep mind and people retrieve them almost unconsciously. this fact above mentioned could also be explained with the representativeness, availability and anchoring/adjusting heuristics employed in cognitive theory. firstly, the representativeness heuristic focuses on similarity and resemblances and people with this mind tend to compare the new stimulus with prior events in memory and see if the new one resembles the old one and can represent it. in fukushima accident, most people considered the chemical explosion of the plant represents the nuclear burst because the scene of chemical explosion is similar to the atomic burst—fire, smoke, slam-bang and all imaginary factors in an explosion are equipped. as a result, without knowing the fact, people consider there is a nuclear burst and the damage to their daily life is huge, which is a misunderstanding. secondly, the availability heuristic refers to that people assess an event with instances or occurrences in the mind directly and if instances of a certain event can be recalled, people tend to believe that the new event is just like the old instances. the hiroshima nuclear bomb and chernobyl nuclear accident still dwell firmly in people’s memory and make people believe that all the nuclear accidents are as serious as them. last but not least, anchoring/adjusting heuristics also plays a very important role. people like to estimate by adjusting from an initial value and first impressions are most lasting, which can bias people’s evaluation of an incident. for instance, the hiroshima nuclear bomb brought more than 100 thousand deaths and injuries while rumor said that the chernobyl made more than 40 million victims affected. therefore, people anticipate that a nuclear accident anchors a massive casualty at the very first beginning. although other nuclear accident is far less severe than them, a high anchoring number make people overvalue the consequence (chen et al. 2012). in a word, with the experience of former serious nuclear accidents in mind, public believe that any nuclear accident may harm one’s health severely and the peace and harmony of society will also be broken. on the other hand, unpredictable and chaos world also are imagined after nuclear accident due to the memory of nuclear blasting in hiroshima and nagasaki. as a result, people will overrate the severity of nuclear accidents and overvalue the consequence under biased perception (koerner, 2014). furthermore, the pessimistic prediction can bring anxiety and anxiety brings more negative thoughts, just like a vicious circle often observed in risk science. 3.2. the analysis of irrational behaviors in nuclear accident with psychoanalytic theory the public irrational behavior is an essential part in the study of crisis communication and the psychoanalytic theory could be employed in this issue. sigmund freud proposed a series of psychoanalytic theories and the production of ego defense mechanisms when the conflict happens among identification (id), ego and superego are the main part of his theoretical framework(peng, 2004). anxiety breaks out when conflicts emerge and ego will employ defense mechanisms to protect the individual if the anxiety becomes overwhelming. defense mechanisms transfer the id impulses to acceptable forms to defend the conflicts, thus helping one to maintain in a favorable self-image and social acceptance. in other words, the defense mechanism is a method that people protect themselves when in a threating environment. no matter this method is really effective or not, it makes sense as long as people think it is helpful. that is because people like to deal with crisis actively other than passively enduring them and it actually help people stay in a positive mood(poumadère et al. 2013). displacement is such a typical kind of defense mechanism that discharges nervous feelings on nonthreatening target, which separates emotion from the original impulse in order to avoid dealing directly with what is harmful. for instance, there is salt-buying panic in china during the period of fukushima nuclear accident and with the defense mechanism— displacement, this phenomenon is easy to understand. radioactive substance could not even be seen or felt, which makes people unable to take any efforts to deal with them. therefore, people need a more concrete thing to outlet their strong emotion. salt is then chosen as a displacement—it is common, necessary, easy to published by atlantis press copyright: the authors 174 obtain and furthermore, it sounds like having a strong relationship with radiation protection because it contains iodine, which heard just like the iodine plates provided by government for avoiding the radioactive iodine i-131 gather in thyroid. with the above information, people establish the logical relationship between salt and radiation protection. on the other hand, rumors about sea water being polluted and salt with radioactive substance spread out to market in the future also make people begin to hoarding a great amount of salt in order to ensure the supply of daily salt in the family. through such action, anxiety is released and people feel more comfortable with the situation. salt works as a displacement of the nuclear panic and buying salt means avoiding nuclear matters under this defense mechanism. fortunately, chinese government release out salt stored in the stock to meet market needs and such method releases the anxiety of people and furthermore safeguards the social stability to a great extent. in this sense, the irrational behavior is the reflection of public nervous condition and could not be prohibited by the government. what the government should pay attention to is the reason and origin of the public irrational behavior and providing appropriate response methods, which means trying to properly channel public sentiments. 4. effective risk and crisis communication about nuclear accident with the statements mentioned above, the detail suggestions of crisis communication about nuclear accident are given as follow: (1) giving proper and instant information. basic information of the crisis needs to be launched. if there is a nuclear accident, the basic 5ws (what, when, where, who, how) need to be launched at the very first time(shao, 2008). then, the solutions such as evacuation notification, necessary actions to reduce the harm of nuclear radiation, symptoms and medical treatments or prohibition of eating local food in the radiation area need to be informed. it should be informed to the victims as early as possible and then to the potential victims to help them evacuate. due to the difficulty and longtime of solving plant explosion, crisis communication has to last even longer. post crisis stage also plays important role in the whole process. especially when the condition of plant is being taken good care of, the focus of public will tend to criticize and question the cause of the crisis, who should take the blame and responsibility and who should pay for the lost. during this stage, a press conference is very necessary to explain these questions and even a letter of apology on newspaper can be needed if the organization wants to show sincere regret. (2) improving the interactions between nuclear power and public. the reason why public is fear of nuclear power is that they never get access to it except those who live near the nuclear plant. as long as they get familiar with nuclear power, they will have a better understanding of it. therefore, it is helpful to invite public to visit the nuclear plants and companies in order to shorten the physical and mental distance between public and nuclear power. at the same time, nuclear scientists and professional staffs should make speech on their personal experience with nuclear power to help public eliminate the panic of nuclear power. meanwhile, the government should take the leading position in eliminating panic and do their best to convince the safety to public (shen et al 2013). what the public care most is the attitude of government and in this sense, the demonstration actions of government are better than any other homiletic sentences. (3) making experts closer to public. experts are rich in professional knowledge in this field compared to the lay people who do not have an adequate background. it has been found that due to the difference in role and knowledge, experts develop distinct cognition pattern when evaluate risks. with an overall view, experts will have a solid consideration and think more thoroughly whereas the general public’s incomplete thoughts may lead to irrationality. the communication between experts and the public is not on the same level, and furthermore, they even do not reach agreements on several basic concepts. therefore, to make the public accept professional advice, narrowing the gap in cognition and common premise should be set. the nonspecialist descriptions and explanations are necessary in the experts’ talk about nuclear accident and furthermore, the gentle mode, fluent expression and creditable sentences are also easy to be accepted by the public. it is not wise to invite a pedant to give comments on nuclear accident with lots of technical terms but with poor presentation skills through social media, which has been proved by amount of instances. (4) the government should communicate with public in modest manners. the communication between the public and government is an interactive process, but in fact, they are not equal in status. the public always acts as the passive side and only receives information published by atlantis press copyright: the authors 175 from the government. while the government always takes the dad pattern (decide, announce, defend) to the public. whether the government can treat the public as “partners” is vital to the efficacy and validity of communication. any policy of obscurantism will definitely fail unless the government can show enough respect to the public or gain trust and support from the community. the government should believe that a majority of people have the ability to understand scientific knowledge if the manner is appropriate. efforts should be made to build up a multiple communication network where authority, experts, media and public can be integrated organically and have a good interaction. one thing needs to mention here is that when refer to crisis communication, people will always connect it with mass communication channels such as newspaper, radio and tv programs, etc. (vaillant, 1997). however, we also have public communication and group communication (xie et al. 2003). though mass communication is easier to be used by organizations and government, the effect of public communication and group communication cannot be ignored. if not, they can serve as good methods for rumors spreading. also, a recent phenomenon in china needs to be paid special attention as well. that is the rise of opinion leaders on weibo and wechat. their existence forms the two-step flow of communication, which means the information comes from mass media, first goes to the opinion leader and then it flows to the flowers of opinion leader. if the official account of the organization posts an explanation of plant explosion on weibo but one of the opinion leader shares it with a comment of “untrue excuse” then many of his or her followers will follow his opinion even though they do not check whether it is true or not. this phenomenon may have great influence on the effect of crisis communication thus can be a main topic of our study to work on(esperanza et al. 2012.). we study the characteristics of information spreading with information tracing methods, which is a popular way to study the communication in the sense of big data analysis. the data is collected from the preset reprinting counter embedded in the website and if this news is reprint by other website or weibo, the counter point will be produced, whose trajectories could also be shown. the data is processed with ontospace software and the entropy and complexity could be obtained. fig.2 shows that the spreading of information “fukushima accident have no influence to chinese coastal fishery” in traditional websites and weibo after fukushima accident. the original point is the first launch of this information and other points means the reprinting by a secondary site. the connecting between two points represents a spreading channel. it is found that generally speaking, the times of information reprinting are no more than fifty in the websites but the reprinting times are more than one hundred in weibo. on the other side, it is also shown that the cascade reprinting exists in the case of weibo and the numbers of nodes are more than ten, compared with the case of traditional websites. with detailed statistical analysis, the entropy and complexity of these two net diagrams are given. the entropy and complexity are the scale of the spreading effect in communication and if the value of a spreading mode is larger than other modes, it means this mode is better in the sense of communication and feasible for the public to obtain. from table 1, it is obvious that the effect of weibo is better than traditional websites, which demonstrate that using new media is necessary and essential for the future communication of nuclear accident. fig.2. the spreading channel of same information in traditional website (a) and weibo (b). table 1 the entropy and complexity of fig.2 (a) and (b), the value of fig.2 (b)’s is normalized to 100 traditional websites weibo entropy 59 100 complexity 52 100 5. conclusion in this article, we reviewed the cases of risk and crisis communication after three famous nuclear accidents (chernobyl, three mile island and fukushima) and studied the public risk and crisis communication after nuclear accident with psychological methods. it is shown that the panic to nuclear accident is due to the worry of inaccessibility to food, air, water and other published by atlantis press copyright: the authors 176 basic physiological needs as well as the uncertainty of safety, so it is not easy to find path to eliminate public aversion about it. to the irrational behaviors of public, it is necessary to give channel to release these modes of people. according to the results, the government and experts who are in charge of risk and crisis communication should think in the public’s shoes and concern what the public concern and try to explain in a clear and easy way to make it understandable. notably, the new media such as wechat and microblog play very important role in the risk and crisis communication about nuclear accident nowadays, which shows much larger spreading range than traditional media. on the other hand, improving the public acceptance to nuclear energy is also helpful in the future risk and crisis communication, which is another way help to improve the communication work. acknowledgements this work is supported by the independent research and development fund of tsinghua university (grant no. 20121087948). references y. chen, the nature event's probabilistic risk assessment of nuclear power plants with improvements, journal of risk analysis and crisis response. 2(2) (2012) 139-145. y. chen, et al, the nature event's probabilistic risk assessment of nuclear power plants with improvements, journal of risk analysis and crisis response. 2(2) (2012) 139-145. v.t. covello, et al, risk communication, the west nile virus epidemic, and bioterrorism: responding to the communication challenges posed by the intentional or unintentional release of a pathogen in an urban setting, journal of urban health: bulletin of the new york academy of medicine. 78(2) (2001) 382-391. x.w. ding, et al, a two-step water-management approach for nuclear power plants in inland china, journal of risk analysis and crisis response. 4(4) (2014) 184-202. l. v. esperanza, et al, volcanic risk perception, locus of control, stress and coping responses of people living near the popocatépetl volcano in mexico, journal of risk analysis and crisis response. 2(1) (2012) 3-12. e.b. foa, et al, emotional processing of fear: exposure to corrective information, psychological bulletin 99(1) (1986) 20-35. r.j. gerrig, et al, psychology and life (18th edition) (allyn & bacon, boston, ma, 2007). m.w. golay, on social acceptance of nuclear power. new energy technologies: a policy framework for micronuclear technology. (2001) p. gray, improving emf risk communication and management: the need for analysis and deliberation, emf risk perception and communication, in proceedings international seminar on emf risk perception and communication, (ottawa, ontario, 1998) pp. 51-68. r.c. hoetzlein, visual communication in times of crisis: the fukushima nuclear accident, leonardo. 45(2) (2012)113-118. c. koerner, media, fear, and nuclear energy: a case study, the social science journal. 51 (2014) 240-249. a.h. maslow, a theory of human motivation, psychological review. 50(4) (1943) 370-396. h. otway, et al, risk communication in europe after chernobyl: a media analysis of seven countries, organization environment. 2(1988) 3-15. d. peng, general psychology, (beijing normal university press, beijing, 2004). m. poumadère, et al, risk assessment of radio frequencies and public information, journal of risk analysis and crisis response. 3(1) (2013) 3-12. x. shao, experimental study of attentional bias to threat information in trait anxious individuals. ph.d. thesis (tianjin normal university, tianjin, 2008) k.h. shen, et al. the study of panic to nuclear energy on psychological and sociological issues, icone21, (chengdu, china, 2013). a. skarlatidou, et al, what do lay people want to know about the disposal of nuclear waste? a mental model approach to the design and development of an online risk communication, risk analysis. 32(9) (2012)14961511. p. tanja, et al, is knowledge important? empirical research on nuclear risk communication in tow countries, health physics society. 102(6) (2012) 614-625. p. tanja, radiation risk perception: a discrepancy between the experts and the general population, journal of environmental radioactivity. 133 (2014) 86-91. g.e. vaillant, adaptation to life (boston, ma, 1997). x. xie, et al, risk communication and public rationality, advances in psychological science. 11(4) (2003) 375-81. d.j. zeigler, & j.h. johnson, information planning and the geographical context of radiological emergency management, the royal geographical society. 14(3) (1989) 350-363. published by atlantis press copyright: the authors 177 microsoft word re jracr-1_1_art5.doc a discrete model of the expected loss for catastrophe insurance in natural disasters chongfu huang1,2 xing shi2.3 1state key laboratory of earth surface processes and resources ecology (beijing normal university) beijing 100875, china, hchongfu@bnu.edu.cn 2bank of china insurance, beijing 100818, china 3academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education the peoples' republic of china, beijing normal university, beijing 100875, china abstract the traditional actuarial model of catastrophe insurance is too theoretical. in many cases, it is of no avail. in this paper, we suggest a discrete model of catastrophe insurance to calculate the expected losses. we successfully applied to calculate the premium rate of the typhoon in fujian province, china. in this model, we use the conditional probability distribution of catastrophic events given a lower-catastrophe and the probability distribution of the catastrophe times to express randomness of catastrophe, use the matrix showing the damage ratio of a single insured object and the matrix showing the insurance loss to express the relationship between catastrophe and economic loss, and use the discrete probability distribution and catastrophe losses and above the relationship to calculate the expected loss. when we use the model to calculate the premium rate, we consider adjustment factors and insurance amount, and partly apply the general actuarial model. research shows that the model versatility, easy operation, the results more reliable. keywords: typhoon, insurance, actuarial, fujian, house. 自然灾害巨灾保险中估计期望损失的一个离散化模型 黄崇福 1,3 石兴 2,3 1.北京师范大学地表过程与资源生态国家重点实验室，北京 100875 2. 中银保险有限公司，北京 100818 3. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875 摘要：传统的巨灾保险费率精算模型过于理论化，许多时候无法使用。本文建议了一个计算期望损失的 巨灾保险离散化模型，并成功应用于中国福建省的台风保险费率。在该模型中，我们用大于等于某级巨灾的巨 灾条件概率分布和它们出现次数的概率分布来表达巨灾的随机性，以巨灾中单体标的破坏比矩阵和破坏程度与 保险损失矩阵来表达巨灾与经济损失的关系，并用离散概率分布和巨灾与经济损失的离散关系来计算期望损失 值。在用该模型计算费率， 我们考虑了费率调整因子和保险金额，并部分用到了通用精算模型等。研究表明， 该模型通用性强，易操作，结果较为可靠。 关键词：台风，保险，精算，福建，住宅 1. 引言 自然灾害巨灾保险是以某一保险条款为基础， 以触发约定标准的巨灾风险为主要保险责任，在约 定的自然灾害巨灾保险区划范围内，对特定的保险 标的，依据法规，建立巨灾风险共保体，并对巨灾 保险所做出的制度性安排。 巨灾的内涵是指某一次灾害发生后，发生地已 无力控制灾害所造成的破坏，必须借助外部力量进 行处置。例如，风速达到每秒 20.8－24.4 米的 9 级 以上的台风，会形成巨灾。 巨灾风险，必须进行共保。亦即是在确定的巨 灾保险区划内，所有利益攸关方组成命运共同体。 通常的做法是，政府财政发起出资组建共保体，成 为 后的财务风险承担者。国内所有保险人，乃至 再保险人出资参加共保体，并组成管理层受托经营 管理。被保险人参与分担风险，比例自保一部分。 这种由政府、保险人和被保险人组成的新型共保 journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 48 体，使再大的巨灾风险也得以 为广泛地分担，有 助于实现巨灾风险的可保性。 巨灾保险能否实施，依赖于巨灾保险费率精算 模型是否有效。 2. 传统巨灾保险期望损失模型 传统的巨灾保险费率精算由3部分组成：巨灾事 件分析、经济损失估计和计算保险人损失期望值。 2. 1 巨灾事件分析 某一地区内，一定时间内发生某种巨灾事件主 要取决于自然灾害规律的作用和周边地区自然环 境，其一次损失程度的大小主要取决于自然灾害类 型及其风险强度和发生该自然灾害事件所在地区域 的建筑物分布和社会经济状况。在一定时间内，一 个地域内的自然环境及其建筑物分布具有一定的稳 定性，因此，我们可用一个地区的位置来代表它 们，设为 l 。在保险实践中，有的用邮编来表示地 区的划分，有的用经纬度来划分地区。 地点 l 发生某一程度的巨灾事件具有一定的随 机性，为了研究巨灾事件对该地点造成的社会损 失，选取与该巨灾事件导致的经济损失程度具有良 好相关性的物理特性的组合，即描叙巨灾事件强度 的变量的组合，记作 lx ，它 是一个随机变量族， 设其取值空间为 lω 。如果把一次巨灾风险所带来 的衍生灾害也算在同一次的话，我们可以近似假设 每次灾害的 lx 是独立同分布的。根据自然灾害发 生的机制，结合地点 l 及周围的自然环境特性，参 考历史数据和经验，我们可以设定 lx 的概率密度函 数为 )( xpl ，x是 lω 中的元素。 同样，地点 l 在时间 t （以年为单位）内发生该 巨灾事件的次数为随机过程，通常假设为由式（1） 所示的泊松过程 )(tn l 。它描述了对任意固定的时 间段t，在这一时间段内随机事件发生k次的概率。 参数 lλ 为泊松过程的强度。在t时间段的期望值由式 （2）计算。特别地，在时间段为一年的情况下，由 式（3）知，在地点 所面临某一巨灾风险可能发生 的期望次数正好为 lλ 。 k2,1,0, ! )( ))(( === − ke k t ktnp t k l λλ (1) ttne ll ×= λ))(( (2) llne λ=))1(( （3) 2.2 经济损失估计 巨灾事件发生时，影响经济损失程度大小的因 素虽然较多，归纳起来有四大类：该巨灾事件的物 理性质，主要是其强度信息；风险暴露单位数量价 值及其分布；建筑物的抗灾能力，即建筑物抗灾级 别和结构特性；以及社会减灾能力。显然前两者与 经济损失是正向关系，后两者与经济损失呈反向关 系。 根据历史数据和经验，我们可以采用模型来预 测某一灾害发生时的损失。对于地点 l ，我们假设 单次巨灾事件所造成的经济损失为： ),,,( lllll dmsxgy = (4) 其中， ls 、 lm 、 ld 分别表示地点 l 的抗灾能力、 经济价值分布和社会减灾能力信息。 2.3 计算保险人损失期望值 当地点 l 的经济损失确定了后，保险人的损失 主要受以下因素的影响：保险责任范围、保险金额 的共保比例、免赔额等保险条件。为了简单起见， 我们用 li 来表示地点 l 的保险相关情况。 设 lz 表示地点 l 发生一次巨灾事件所造成保险 人的损失，根据 ly 和 li ，通常作如下假设： )),,,,((),( lllllll idmsxghiyhz == (5) 则地点 l 面临单次巨灾事件时，保险人的期望损失 为： dxxpidmsxghze llllll l )()),,,,(()( ×= ∫ ω (6) 设 )(tvl 表示地点 l 在时间 t 内由单一巨灾事件 所造成保险人的总损失，则 ∑ = = )( 1 )( tn k ll l ztv (7) 这里的 k 是指巨灾事件发生的次数。 同理，地点 l 在时间 t 内由单一巨灾事件造成 保险人的总期望损失为： )())(())(( lll zetnetve ×= dxxpidmsxght llllll l )()),,,,(( ×××= ∫ ω λ (8) 由此可得地点 l 在一年内由该巨灾事件造成保险人 的总期望损失e为[1]： dxxpidmsxghe llllll l )()),,,,(( ××= ∫ ω λ (9) published by atlantis press copyright: the authors 49 显然，式（9）是一个理论模型，函数表达式g和h 均不易确定。 3. 离散化的巨灾保险期望损失模型 由于传统模型中一系列函数的表达式不易确 定，用少量巨灾样本估计出的有关参数其可靠性也 不高。所以，在许多情况下，传统模型的理论意义 大于实际意义。本文根据巨灾风险的特点，提出离 散化的巨灾保险期望损失模型，使理论模型过渡到 实用模型。 离散模型涉及两个主要的定义域，一个是巨灾 等级的定义域；另一个是财产破坏程度的定义域。 例如，大于9级的台风，其风力等级查表1可 知。而住宅破坏程度的定义域，可由表2给出。台风 的定义域记为w（用“wind”的第一个字母），破 坏定义域记为d（用“damage的第一个字母）。它 们可分别用式（10）和（11）表之。 },,,,{ 54321 wwwwww = ={9 级风,10 级风,11 级风,12 级风,13-17 级风} (10) },,,,{ 54321 dddddd = ={基本完好,轻微破坏,中等破坏,严重破坏,完全破坏} (11) 表1. 风力等级表 风速 风力级数 名称 陆地地面征象 （米/秒） （公里/小时） 0 静稳 静，烟直上 0-0.2 ＜1 1 软风 烟能表示风向，但风向标不能动 0.3-1.5 1-5 2 轻风 人面感觉有风，树叶微响，风向标能转动 1.6-3.3 6-11 3 微风 树叶及微枝摇动不息，旌旗展开 3.4-5.4 12-19 4 和风 能吹起地面灰尘和纸张，树的小枝摇动 5.5-7.9 20-28 5 清劲风 有叶的小树摇摆，内陆的水面有小波 8.0-10.7 29-38 6 强风 大树枝摇动，电线呼呼有声，举伞困难 10.8-13.8 39-49 7 疾风 全树摇动，迎风步行感觉不便 13.9-17.1 50-61 8 大风 微枝折毁，人行向前感觉阻力甚大 17.2-20.7 62-74 9 烈风 建筑物有小损（烟囱顶部及平屋摇动） 20.8-24.4 75-88 10 狂风 陆上少见，见时可使树木拔起或使建筑物损坏严重 24.5-28.4 89-102 11 暴风 陆上很少见，有则必有广泛损坏 28.5-32.6 103-117 12 飓风 陆上绝少见，摧毁力极大 32.7-36.9 118-133 13 37.0-41.4 134-149 14 41.5-46.1 150-166 15 46.2-50.9 167-183 16 51.0-56.0 184-201 17 56.1-61.2 202-220 资料来源：国家标准gb/t 19201-2006 published by atlantis press copyright: the authors 50 3.1 巨灾事件离散模型 设在某地发生一次大于等于m1巨灾的条件下， 这次巨灾是mi级的概率是一个离散关系，如表3所 示，记为p1(mi)。注意， p1(mi)是一个条件概率分 布，而不是某时段内发生mi级巨灾的概率分布。 表3. 发生mi级巨灾的概率分布p1(mi) mi级巨灾 m1 m2 … mi … mn 概率p1(mi) p1 p2 … pi … pn 再设某时段t内出现m1级以上巨灾的次数为k的 概率分布是一个离散关系，如表4所示，记为p2(k)。 表4. 时段t内m1级以上巨灾发生k次的概率分布p2(k) 时段t内次数k 0 1 2 … k … n 发生概率p2(k) f0 f1 f2 … fk … fn 3.2 经济损失离散模型 经济损失与不同巨灾下保险区划内单体标的破 坏比例有关。设mi级巨灾发生时，破坏程度为dj的单 体标的在保险区划内所占比例为aij。该离散关系可 以由表5示之。 由于该表不能直接反映单体标的因巨灾所蒙受 的经济损失状况。为此需要建立在不同破坏程度 下，经济损失程度比矩阵，它反映的是破坏程度为 dj时，从统计意义上来说，单体标的经济损失的情 况。自然，破坏程度越严重，经济损失所占该单体 标的原有价值的比例就越大。我们将此程度比记为 bj。该离散关系可以由表6示之。 表6反映的是破坏程度与经济损失程度的自然关 系。由于一次巨灾事件保险损失肯定小于或等于经 济损失，况且，保险赔偿金额需要考虑保险责任、 赔偿责任限制、免赔额等诸多因素，所以，保险精 算使用的，是在表6基础上结合保险实践给出的，我 们用表7表示单体标的保险损失程度。 表5. 巨灾中单体标的破坏比矩阵（％） 破坏 巨灾 d1 d2 … dj … dt m1 a11 a12 … a1j … a1q m2 a21 a22 … a2j … a2q … … … … … … .. mi ai1 ai2 … aij … aiq … … … … … … .. mn an1 an2 … anj … anq 表6. 保险区划内巨灾经济损失程度矩阵表（％） 破坏程度 d1 d2 … dj … dq 经济损失 b1 b2 .. bj … bq 表7. 保险区划内单体标的保险损失程度矩阵表(%) 破坏程度 d1 d2 … dj … dq 破坏程度 保险损失 c1 c2 .. cj … cq 保险损失 3.3 计算期望损失值的离散模型 假定发生了mi级巨灾，则依表5知，破坏程度为 dj的单体标的在所有单体标的中将占aij的比例。而由 表7知，这种破坏程度的住宅，会产生量值为cj的保 险损失比例。于是，mi级巨灾产生dj破坏程度的保险 损失程度比为 jijij cad ×= ，mi级巨灾对所有单体标 表 2. 住宅破坏程度表 名称 破坏情况描述 基本完好 建筑物承重和非重承构件完好，或个别非承重构件轻微损坏，不加修理可继续使 用。 轻微破坏 个别承重构件出现可见裂缝，非承重构件有明显裂缝，不需要修理或稍加修理即可 继续使用。 中等破坏 多数承重构件出现轻微裂缝，部分有明显裂缝，个别非承重构件破坏严重，需要一 般修复后可继续使用的。 严重破坏 多数承重构件破坏较严重，或有局部倒塌，需要大修，个别建筑修复困难。 完全破坏 多数承重构件严重破坏，结构濒于崩溃或已倒毁，已无修复可能。 published by atlantis press copyright: the authors 51 的可能造成的保险损失程度比di由式（12）计算， 这些数值可构成表8所示的保险区划内单体标的巨灾 台风保险损失程度比矩阵。 niddddd iqijiii ,...,2,1,21 =+++++= ll (12) 表8. 保险区划内单体标的巨灾保险损失程度比 矩阵（％） 破 坏 巨灾 d1 d2 … dj … dq 合 计 m1 d11 d12 … d1j … d1q d1 m2 d21 d22 … d2j … d2q d2 … … … … … … .. mi di1 di2 … dij … diq di … … … … … … .. mn dn1 dn2 … dnj … dnq dn 当mi级巨灾发生时，假定其在保险区划内的影 响范围所占区划面积的比为si，则产生的保险损失 程度之比hi由式（13）计算。 iii sdh ×= , i=1,2,..,n (13) 为简单起见，我们用表3给出的概率分布，依式 （14）计算出时段t内m1级以上巨灾发生的期望次数 k 作为各级巨灾的t内发生次数使用，即， n n k k fnffffkk )1(32)1( 432 1 −++++=−= ∑ = l (14) 按简单的累加原则，mi级巨灾在年内给保险区 造成的保险损失程度之比 可由式（15）计算。 ksdh ii k i ××= )( , i=1,2,..,n (15) 考虑mi级巨灾发生的概率，依表2，其t内mi级 巨灾造成的保险损失率mi可由式（16）计算。 )(kiii hpm = (16) 于是，根据随机变量期望值的定义，t时段内巨灾标 的保险期望损失率m可由式（17）进行计算。 )( 1 k i n i i hpm ∑ = = (17) 综合上面的相关计算结果，我们可以得出保险 期望损失率测算表，如表9所示。 表9 保险区划内m1级以上巨灾t时段内保险标的 期望损失率测算表（％） 序号 巨灾等级 m1 m2 … mn 合计 1 发生概率 p1 p2 … pn 1 2 损失程度比 d1 d2 … dn 100% 3 影响区域 s1 s2 … sn 4 期望次数 k k … k 5 期望损失率 m1 m2 … mn m 由表3-表8和式（12）-（17）给出的模型就称为 巨灾保险期望损失离散模型（简称“期望损失离散 模型”），表9中的 后一行数据，就是这一该模型 计算出的结果。依此进行的巨灾保险费率的厘定， 则需根据保险业自身的许多实际情况进行考虑，我 们将在下面的案例分析中加以讨论。 4. 期望损失离散模型在精算中的应用 本文将期望损失离散模型用于福建住宅台风保险 的精算，给出保险区划梯度费率表。 4.1 研究区概述 福建省位于中国东南沿海，全省土地面积为 12.14 万平方千米，约占全国土地总面积的 1.3%。 境内峰岭耸峙，丘陵连绵，河谷、盆地穿插其间， 山地、丘陵占全省总面积的 80%以上。 全省陆地海岸线总长 3000 多公里。由于靠近世 界上 大的台风源地，福建省是中国台风灾害 严 重的省份之一。相关资料表明，近 30 年来，平均每 年有 1.5 个台风登陆福建, 4 个台风影响福建, 每 年造成近百人死亡和亿元的直接经济损失, 台风灾 害是福建所有自然灾害中危害 大、损失 严重的, 已经严重影响福建省经济可持续发展，当地民众对 台风保险有较高的需求。 4.2 福建住宅台风保险期望损失计算 2009 年 6 月，本文第二作者对福建省民政厅进 行了调研，并收集了 2007 和 2008 年两年台风住宅 损失的相关数据资料。为简化相关的计算，我们作 如下相关假设： （一）住宅标的的抗灾能力主要体现在其建筑物 的结构性能方面，针对台风巨灾，以抗灾能力的强 published by atlantis press copyright: the authors 52 度排列依次分为钢结构、钢混、砖混、砖木和木结 构。台风主要发生在我国东南沿海地区，城乡差异 相对较小。在福建省，地级城市市区住宅主要是砖 混结构和钢混结构，城镇和农村地区 95％以上住宅 是砖混结构，两种结构有类似之处，为简化说明应 用方法，故以砖混结构代替所有住宅（福建省民政 厅所提供的资料也没有细分住宅结构），以下将单 体砖混结构住宅简称单体住宅； （二）基于福建省绝大部分住宅在地级市和农村 地区，经验估计每套住房的平均价 30 万元； （三）假设被保险人分担巨灾风险的比例为 20%，即保险人承担每套单体住宅 80%的经济损失。 每栋住宅都是足额办理承保手续的。 在财产保险精算中，易损度是一个重要的概 念。针对巨灾风险，我们给易损度的定义如下：在 确定巨灾保险区划不同地域内，以确定的时间段 （通常 近三年），对某一巨灾风险每次事件所造 成某类保险标的的损失金额进行年度累计，然后与 该地域内的保险标的价值累计相比，所得的比值就 是巨灾保险区划内不同地域同类保险标的的年度风 险易损度。 按照易损度大小来划分风险的强度就是风险梯 度。在巨灾保险区划内不同地区的同类保险标的的 易损度所代表的风险强度构成了巨灾保险区划。 以住宅为保险对象，以福建省各地级市为风险 梯度区域，根据所收集的 2007、2008 二年的台风住 宅损失相关数据资料，平均计算各地级市的台风住 宅易损度，由大到小排列如下：南平：0.007024， 三 明 ： 0.004702 ， 宁 德 : 0.002323 ， 龙 岩 ： 0.000770，泉州：0.000413，莆田：0.000285，福 州 ： 0.000262 ， 漳 州 ： 0.000149 ， 厦 门 ： 0.000056。据此，我们得到的福建住宅台风巨灾易 损度区划，见图 1。 南平市 .7024% 三明市 .4702% 龙岩市 .0770% 宁德市 .2323% 漳州市 .0149% 福州市 .0262% 泉州市 .0413% 莆田市 .0285% 厦门市 .0056% 福建省住宅易损度 .000056 .000057 .000149 .000150 .000262 .000263 .000285 .000286 .000413 .000414 .000770 .000771 .002323 .002324 .004702 .004703 .007024 图 1 中国福建省住宅台风巨灾保险区划图 published by atlantis press copyright: the authors 53 根据 2001-2007 年的《热带气旋年鉴》，在福建 住宅台风巨灾保险区划内，我们计算出，大于等于 9 级的台风中，台风为 wi 级的频率分布为表 10，而 9 级以上台风出现不同次数的频率分布为表 11。 表 10. 福建省 9 级以上台风不同等级发生频率的分 布情况 台风强度 wi 9 10 11 12 12 级以上 发生频率 0.3 0.35 0.25 0.09 0.01 根据 2001-2007 年的《热带气旋年鉴》，在福建 住宅台风巨灾保险区划内，我们计算出，大于等于 9 级的台风中，台风为 wi 级的频率分布为表 10，而 9 级以上台风出现不同次数的频率分布为表 11。 表 11. 福建省一年内出现 9 级以上台风次数 k 的频 率分布情况 k 0 1 2 3 4 5 6 ≥ 7 频率 0 0.07 0.35 0.39 0.12 0.06 0.01 0 我们分别用表 10 和表 11 中的数值作为表 3 和表 4 中相应的概率估计值来使用。于是，用式(14)可 计算出年内 9 级以上台风发生的期望次数 k ， 2.7801.0606.05 12.0439.0335.0207.0 )1( 8 1 =×+×+ ×+×+×+= −= ∑ =k kfkk 根据调研材料，我们整理出福建省台风巨灾保 险区划内单体住宅巨灾台风破坏比矩阵表（表 12），以及福建省保险区划内巨灾台风经济损失程 度矩阵表（表 13），它们是期望损失离散模型中的 表 5 和表 6 的具体体现。表 12 反映了巨灾风险的不 同级别、损失破坏程度及其分布概率等因素，某种 程度上也反映了单体住宅在不同的台风强度下的抗 灾能力。表 13 中的经济损失程度是指单体住宅遭受 巨灾风险后，在不同破坏程度下，可能的经济损失 占其整体价值之平均比例。例如，由该表知，在轻 微破坏情况下，一个单体住宅将遭受的经济损失程 度为 25％。 保险赔偿金额需要考虑保险责任、赔偿责任限 制、免赔额等诸多因素，涉及较多的主观判断，很 难在保险损失与经济损失之间建立一个非常精准的 函数关系。然而，根据经验数据，我们假设自保比 例为 20％以上为，从而可以统计一个平均线性比例 关系，如果采用比例免赔额，则存在一个比例关 系。为简化起见，采用比例系数为 r=20％，则可将 表 13 转化成表 14 的单体住宅的保险损失程度矩阵 表，它是期望损失离散模型中表 7 的具体体现。例 如，轻微破坏的保险损失程度是 c2= b2×r＝25％×80 ％＝20％。 将表 12 的各项数据按行乘以表 14 的各项数 据，即可得到单体住宅一次 9 级以上台风对住宅的 保险期望损失程度比矩阵，如表 15 所示，它是期望 损失离散模型中表 8 的在实例中的具体数值。 例如，9 级台风时，轻微破坏的住宅，会给保 险公司因这些住宅投保带来标的保险金额的损失比 为 d12＝a12×c2＝0.12％×20％＝0.0240％。 根据台风的结构、能量和热带气旋风场的‘三 圈’结构，每个台风基本呈椭圆形形状。一般说 来，从台风中心（风眼）至云墙（眼壁）的半径有 60 公里至 100 公里不等，假设取中间值为 80 公 表 12. 福建省台风巨灾保险区划内单体住宅巨灾台风破坏比矩阵（％） 破坏 台风 基本完好 轻微破坏 中等破坏 严重破坏 完全破坏 合计 9 级 99.8％ 0.12％ 0.08％ 0％ 0％ 100％ 10 级 99.617％ 0.342％ 0.0355％ 0.0053％ 0％ 100％ 11 级 99.5331％ 0.3964％ 0.0468％ 0.0218％ 0.0019％ 100％ 12 级 99.3975％ 0.5277％ 0.0559％ 0.0134％ 0.0055％ 100％ 12 级以上 97.1069％ 1.8442％ 0.8475％ 0.1522％ 0.0492％ 100％ published by atlantis press copyright: the authors 54 里。虽然台风影响的范围方圆直径约为 1000 公里左 右，但根据历史数据，在这个椭圆形半径为 80 公里 的环带内，暴风和暴雨强度 大，对地面财产和人 员造成大面积、大范围、大量保险标的重大损失和 伤亡可能性 高，对其他地方的影响是较小的。假 设台风在福建中部登陆，由东南向西北移动，横穿 整个福建省，且以台风中心所作的四个象限内都遭 遇大风暴雨，那么我们可以将其近似看作一个宽 160 公里，长 480 公里（福建东西 大间距约 480 公里）的长方形。则一次 9 级以上台风，在上述移 动路径下，对福建 为完整的、且 大范围的破坏 性台风灾害所致的遭灾面积为 7.68 万平方公里，约 占福建省陆地面积 12.4 万平方公里的 61.93％（由 于考虑的是巨灾台风对住宅的影响，故只考虑陆地 面积）。 巨灾保险区划一般较大，不同的巨灾风险、触 发的强度、发生位置、次生灾害、影响时间等因 素，对整个巨灾保险区划内的保险标的影响范围和 程度差异是较大的。就福建住宅台风巨灾保险区划 来说，一个巨灾台风也不可能对福建整个保险区划 都有破坏性影响，其影响范围和程度主要看生成时 间（是否与大潮汛、月盈月亏）、登陆地点、台风 强度、发展过程、行进路线、移动速度、影响时间 等因素来决定。如 2008 年的海鸥台风虽然达到 10 级，且在福建霞浦县长春镇登陆，但仅仅掠过福建 的北部（非常临近浙江的南部地区），随后就移向 东海，影响路径和事件十分短暂，所以该台风对福 建的影响的范围很微小。 基于以上分析，我们只能凭经验假设一次 9 级 以上台风对福建住宅台风巨灾保险区划的平均影响 范围取中间值为 30％，即，表 9 中的 si 均取值为 30 ％。 使用式（13）-（15）和表 10，表 15 提供的 数据，以及上面给出的 k 值和平均影响范围取中间 值 si，我可以得到表 16 所示的保险期望损失率测算 表。 例如，9 级台风时期望损失率计算出下： %0148.0 78.2%30%0592.03.0 111 )( 111 = ×××= ×××= = ksdp hpm k 4.3 费率计算公式 表 13. 福建省保险区划内巨灾台风经济损失程度矩阵表（％） 破坏程度 基本完好 轻微破坏 中等破坏 严重破坏 完全破坏 经济损失程度 0％ 25％ 55％ 85％ 95％ 表 14. 福建省保险区划内单体住宅巨灾台风保险损失程度矩阵表（％） 破坏程度 基本完好 轻微破坏 中等破坏 严重破坏 完全破坏 保险损失程度 0％ 20％ 44％ 68％ 76％ 表 15. 福建保险区划内单体住宅巨灾台风保险损失程度比矩阵（％） 损坏 台风 基本完好 轻微破坏 中等破坏 严重破坏 完全破坏 合计 9 级 0％ 0.0240％ 0.0352％ 0％ 0％ 0.0592％ 10 级 0％ 0.0684％ 0.0156％ 0.0036％ 0％ 0.0876％ 11 级 0％ 0.0793％ 0.0206% 0.0148% 0.0014% 0.1161% 12 级 0% 0.1055% 0.0246% 0.0091% 0.0042% 0.1434% 12 级以上 0% 0.3688% 0.3729% 0.1053％ 0.0374％ 0.8826％ published by atlantis press copyright: the authors 55 i．保额统计 保额统计有两种方法，一是在巨灾保险区划内 对每一住宅一一统计加总，二是按照抽样统计，得 出一套住宅平均价，以此计算。显然第二种方法较 好。巨灾保险因实施强制保险，所以统计在内所有 住宅必须参加保险。我们使用 2007 年和 2008 年福 建省统计年鉴，经研究和整理, 得福建省住宅保额 统计如表 17。 表 17. 福建台风巨灾保险区划内各地区保额统计表 地区 户数 （万户） 单价 （万元） 价值 （亿元） 福州 195 30.00 5,850.00 厦门 47 30.00 1,410.00 宁德 94 30.00 2,820.00 莆田 89 30.00 2,670.00 泉州 177 30.00 5,310.00 漳州 128 30.00 3,840.00 龙岩 82 30.00 2,460.00 三明 75 30.00 2,250.00 南平 87 30.00 2,610.00 合计 974 30.00 29,220.00 ii．费率调整因子的假定 根据巨灾保险费率组成和巨灾保险费率精算模 型，结合巨灾保险的特点和保险公司一般的营运成 本经验数据，我们假设运营成本附加因子为 20％； 被保险人因参与风险分担，自保 20％，且设有免赔 额，故其防灾防损意识加强，应该设置诱导性费 率，我们假设费率折扣因子为 5％；安全性附加因 子纯粹是一个主观的经验假设，为 10％。具体费率 调整因子假设如表 18 所示。 表 18. 福建住宅台风巨灾保险费率调整因子表 调整因 子名称 运营成本 附加因子 安全性附 加因子 费率折 扣因子 调整比 值 20.0% 10.0% 5.0% iii．保险区划内不同区域的梯度费率系数 在一个巨灾保险区划内的所有地区所遭受侵袭 和影响的台风次数应该是均等的，但是不同地区的 灾害风险强度是不一的，可以设定一个地级市为一 个风险梯度。根据相关资料，为简化说明，这里不 妨将风险强度差不多的地级市归并为一个风险梯度 地区，合并为 4 个。考虑漳州的实际情况和两年数 据资料的局限性，对此作了调整。于是，我们得到 福建省的风险梯度划分如表 19，假设福建省台风住 宅易损度 低的福州和厦门地区为 1，那么按照易 损度的数值，简单算术平均推算出各地的实际梯度 费率系数。 表 19. 福建住宅台风巨灾保险区划内风险梯度系数 所属范围（代码） 梯度系数 保额 （亿元） 福州、厦门（t1） 1 7,260 泉州、漳州、莆田（t2） 1.7736 11,820 龙岩、宁德（t3） 9.7233 5,280 南平、三明（t4） 36.8742 4,860 表 16. 福建省住宅台风巨灾保险区划内 9 级以上台风年度保险期望损失率测算表（％） 序号 台风等级 9 10 11 12 12 级以上 合计 （1） 发生概率 0.3 0.35 0.25 0.09 0.01 1 （2） 损失程度比 0.0592% 0.0876% 0.1161% 0.1434% 0.8826% 100% （3） 影响区域 30％ 30% 30% 30% 30% （4） 期望次数 2.78 2.78 2.78 2.78 2.78 （5） 期望损失率 0.0148% 0.0256% 0.0242% 0.0108% 0.0074% 0.0827% published by atlantis press copyright: the authors 56 iv．费率计算公式 （1） 巨灾保险费率的构成 巨灾保险费率由基本费、梯度费率和安全费率 构成。 基本费率又称纯风险基准费率。世界上没有统 一的标准。美国将百年一遇的洪水作为基准洪水。 根据我国社会经济变化转快的特点，以 3-5 年历史 数据统计所得的平均易损度为我国巨灾风险的基本 费率较为合适。 梯度费率反映的是不同地区所面临某类巨灾风 险的平均强度。巨灾保险区划内不同地区所面临的 同一巨灾风险的强度是不一的，梯度费率可根据经 验数据，诸如灾害风险强度、灾损率或以上述平均 易损度来确定。 不同地区的实际费率是在基本费率的基础上 乘以梯度费率系数而得。 在计算巨灾保险费率时考虑一定的安全系数， 涉及的费率称为安全费率，其与保险标的本身的抗 灾能力、区域防灾工程规划等密切相关。安全费率 的出现，使巨灾保险费率的精算有别于商业保险费 率精算。 （2） 调整系数 经营成本和相关税收数:由于巨灾保险的公共产 品属性，具有一定的公益性，所以商业的逐利性要 少考虑 [2] 。巨灾保险主要经营成本是代办费用、管 理费用、公估费用等，或者一定的微利。日本地震 保险费率也不考虑承保人盈利，这反映了地震保险 的公益本质。 折扣系数:保险人对以下四种情况给予被保险人 一定的折扣系数，一是为了鼓励防灾防损，保险人 设计具有诱导性的防灾防损差别费率。二是如果有 新的防灾防损设施投入使用，在新的保险年度厘定 费率时就要对折扣系数加以调整。三是国家关于建 筑规范、抗灾级别的提高而设置的折扣系数。有的 国家还设有建筑年限折扣系数等。总折扣系数一般 在 10％以内浮动。 （3） 费率调整因子 根据前述巨灾保险费率的构成和调整系数易 知，建立巨灾保险费率精算模型需要设立如下调整 因子： 运营成本附加因子：设第 i 种巨灾保险保险人的 运营管理成本（包括管理费用、佣金、税费等）比 率为 1,i θ 。 安全性附加因子：设第 i 种巨灾保险保险人的风 险波动安全和自然灾害风险发展趋势增强等因素的 附加率为 2,iθ 。 折扣因子：根据前述费率折扣系数所考虑的因 素，设第 i 种巨灾保险的实际费率折扣系数为 3,iθ 。 梯度因子：设第 i 种巨灾保险在整个保险区划内 一年的基本费率为 iα ，设第 i 种巨灾保险在地点 jil , 内一年的区域费率为 ji,α ，设第 i 种巨灾保险在地点 jil , 的梯度费率系数为 jit , ，即 jiiji t ,, ×= αα （18） （4）保险金额 在给定的保险区划内，我们根据民政部门、公 安局部门、邮政编码、投保单等相关数据资料，可 以轻而易举地统计得到在整个保险区划内各个保险 标的保险金额的合计，假设为 jib , 。 （5）计算公式 根据巨灾保险期望损失的一般模型（公式 （9））考虑上述的调整因子和调整系数，可得第 i 种巨灾保险平均费率的一般通用精算模型： ∑ ∑ ∫ ∑ −−×+×× ×× = −−×+×× = ω j iiijiji j jijijijijiiji j iiijiji i i tb dxxpidmsxh tb e ji )1()1()( ))(),,,,(( )1()1()( 2,1,3,,, ,,,,,, 2,1,3,,, , θθθ λ θθθ α （19） 4.4 费率计算演示 在本例中，费率计算公式具体化为： ∑ ∑ ∑ ≤≤ ≤≤ ≤≤ −−×+×× × = −−×+×× = mj jj mj j mj jj tq mq tq e 1 213 1 1 213 )1()1()( )( )1()1()( θθθ θθθ α （20） (1) 纯风险平均费率计算 根据式（20）、表 16、18 和 19，可以得到巨 灾保险的纯风险平均费率（即风险梯度 t1，参照系 数为 1 的地区住宅台风巨灾保险费率），具体计算 如下： 分子= e（总损失） published by atlantis press copyright: the authors 57 =总计保额×9 级以上台风年度保险期望损失率 =（7,260+11,820+5,280+4,860）×0.0827% =24.16494（亿元） 分母=（∑保额×梯度系数）×（1+费率折扣因子） ×（1-运营成本附加因子-安全性附加因子） =（7,260×1+11,820×1.7736+5,280×9.7233 +4,860×36.8742）×（1+5.0%） ×（1-20.0%-10.0%） =190，197.1（亿元） 综上，得： 纯风险平均费率=24.16494÷190197.1=0.0127% (2) 梯度纯风险费率计算 根据梯度费率定义，见式（18），我们计算出 各风险梯度区域的标准费率，如表 20 所示。 表 20 福建住宅台风巨灾保险区划梯度纯风险费率 测算表 (3) 梯度标准费率计算 根据以上测算，南平和三明地区一套价值为 30 万保额的单体住宅将缴纳 1,404.9 元保费（300,000 ×0.4683%），而在福州、厦门地区一套 30 万住宅 的保费为 38.1 元， 一套 80 万的住宅，缴纳保费为 101.6 元，如果一套价值为 200 万元的住宅，缴纳 保费为 254 元，对比说明两者差距是很大的。这就 印证了前述风险越高的地区，往往是农村地区，购 买力较弱，超出普通老百姓的承受能力，将给巨灾 保险方案的实施带来很多不良影响。基于前述的巨 灾保险准共产品属性及其在费率精算中的体现，有 必要对梯度纯风险费率进行调整，具体方法有二： 第一，基于实际梯度系数，结合上述考虑，我 们需要重新主观选定梯度系数，计算调整后的梯度 标准费率，这一费率也是对外公布的承保费率。如 表 21 所示。 表 20 与表 21 对比，风险梯度 1 与 4 的费率差 距由原先的 36.87 倍，缩小至 6 倍。保费支出得到了 有效均衡。 表 21 福建住宅台风巨灾保险区划梯度费率表 风险 梯度 所属范围 实际梯 度系数 选定梯 度系数 梯度标 准费率 1 福州、厦门 1 1 0.0446% 2 泉州、漳州、 莆田 1.7736 1.5 0.0669% 3 龙岩、宁德 9.7233 3 0.1338% 4 南平、三明 36.8742 6 0.2676% 第二，对龙岩、宁德、南平和三明等农村地 区，可根据物价指数，消费水平和人均收入，政府 对被强制要求购买住宅台风保险的被保险人，实行 财政补贴、所得税退税或营业税减免等措施，以提 高消费者的购买能力，从而降低被保险人购买巨灾 保险所支付的保费，达到相对于购买群体消费能力 能承受的程度。这样被保险人实际支付的保费还会 下降。如何优惠，主要由巨灾保险制度决定。 基于上述数据资料和应用方法所计算的福建住 宅台风巨灾保险费率，对被保险人来说应该是完全 可以接受的，对巨灾保险共同体来说，也能够建立 巨灾保险基金，这两方面说明费率测算是切合实际 的。 5. 结论与讨论 由于中国社会发展很快，相适应的巨灾保险资 料并不多，国外大型保险公司和机构开发的巨灾保 险模型很难在中国投入使用。从理论上讲，巨灾保 险费率精算并不难，但精算的精度不仅在于模型自 身的完备性，更在于模型的可用性和相关数据资料 是否能够支持这样的模型。显然，一个既能支持精 算，又能得数据资料支持的模型成为在中国进行巨 灾保险费率精算的首选。本文建议的离散模型，有 效使用时间系列不长的相关资料实现了对福建省台 风巨灾保险费率的精算。经福建省一些资深保险专 家的经验判断，认为给出的结果较为可靠。 参考文献 [1] 石兴，自然灾害风险可保性理论及其应用研究，北京师 范大学博士学位论文，2009。 s. xing, insurability study and its application on natural disasters, phd thesis, beijing normal university, 2009. [2] 张兆本，公共经济学, 北京：人民出版社，2005 z. b. zhang, public economics, beijing: the people's press, 2005 风险 梯度 所属范围 实际梯 度系数 梯度纯 风险费率 1 福州、厦门 1 0.0127% 2 泉州、漳州、莆田 1.7736 0.0225% 3 龙岩、宁德 9.7233 0.1235% 4 南平、三明 36.8742 0.4683% published by atlantis press copyright: the authors 58 paper template for pre-olympic congress received 30 march 2015 accepted 12 may 2015 a study of the role of customs in global supply chain management and trade security based on the authorized economic operator system liwen chen, yongfei ma school of economics and management, hebei university of technology, tianjin 300401, china e-mail: lwchen@hebut.edu.cn, mayongfei@customs.gov.cn abstract the core aims of authorised economic operator (aeo) are to establish an international corporate identity system and encourage supply chain enterprises to maintain high degrees of safety and security in their international trading. this paper conducts an in-depth study of customs management thinking regarding the aeo system and actively explores its implementation and integration with international customs standards common practice and fulfilling the requirements of any international obligations. the government will need to work with enterprises in developing basic conditions on aeo certification, recognising and dealing with problems as they arise, while seeking to maintain the facilitation of customs clearance for excising lawabiding and sustainable trade routes and ensuring the security. keywords: trade facilitation, trade security, aeo system, cooperation, customs external auditing, risk management according to economic historian angus madison, for nearly five hundred years, most of the achievements of the global economy development were the reduction communication and transportation costs associated with economic liberalization. and declining transportation and communication costs will continue to promote global economic integration. international logistics industry will help national economic development, and improve the competitiveness of national life enhancement by having an important impact. customs entry and exit as the national regulatory authority, charged with the sacred duty of checks and services is an important part of modern logistics management chain, and its management practices process the most direct and close contact in the logistics. at the same time, along with the growing international trade and more severe trade frictions, world customs organization (wco) is being focusing on the relationship between customs and business, which showed significant shift the bottom line from the perspective of trade promotion to the development of the enterprise itself, from a focus on customs duty between multinational, transnational responsibility, to tend between the business community to simplify customs clearance procedures from the goal, turning the entire supply chain to achieve facilitation, quietly changing international customs functions to enable customs to rise to partnerships with the business community of a “strategic” height (world customs organization, 2015). with a fast increase in the import and export trade volume and the existence of abundant processing trade and tax reduction and exemption, illegal activities still exist in the import and export channels. there is no actual change of the integrity status of the society. the growing inconsistencies between the strict execution and the high-efficient operation as well as the increasing trade volume and the limited management resources become more and more severe. and it is a critical problem confronting the customs, which is building a modern customs system with chinese characteristics. it is also the case with the international customs. the traditional way of relying on personnel input to improve the customs control quality can not suit the new situation. since 1990s, the world customs have surged up the new wave of reform and modernization with regard to adjusting the management functions, reforming the management system, innovating the management technologies and implementations and shifting the management philosophy so as to adapt to the changing of current situation. 1. aeo system---the base of the international trade facilitation and security trade facilitation and security have always been the two major themes for the reform and development of the world customs. in june 2005, china customs signed “framework of standards to secure and facilitate global trade”, which is advocated by wco. in order to promote the effective implementation of the relevant standards of ‘customs-to-business partnerships’, the wco framework introduces the concept of authorized economic operator, known as aeo (cp3 group, 2006) . any enterprise involved in the international trade, which is willing to join and proved to be in compliance with the safety standards and requirements of the international supply chain, could be authorized by the customs as an aeo, so as to be granted the trade facilitation privileges. journal of risk analysis and crisis response, vol. 5, no. 2 (july 2015), 87-92 published by atlantis press copyright: the authors 87 the rate of the smuggling cases keeps high in the processing trade in those customs surveillance zones located in many parts of the world, which are not connected to the e-control networks. there is a stable ratio of smuggling cases in processing trade to those in commercial frauds. therefore with the help of e-control networks, the customs management procedures are being simplified to improve the clearance efficiency. in the meantime, the security and authenticity of the data exchanged among the regional customs in processing trade are ensured. the essence of modern customs system is to build a both benefited co-operative customs-to-business partnership. customs authority, as a government administration, needs to build a new management mode and a set of operational mechanism to cultivate and maintain the relationship with their business partners. the co-operations are based on the promissory obligations between the customs and businesses, with the aim that the business partners can abide by the rules while the customs should administrate according to national law and regulations. with the principle of “facilitation for the law-abiding enterprises and punishment for any dishonest ones”, the customs may guide the enterprises to be honest and law-abiding so as to standardize the import and export operation. to strengthen the authorization of aeo is the base for realizing the security and facilitation of international trade (world customs organization, 2008) . the aeo program under the ‘framework of standards to secure and facilitate global trade’ promoted by wco requires the customs to grant the authentication to the following enterprises joining in the international supply chain, including: manufacturers, importers, exporters, brokers, carriers, consolidators, intermediaries, ports, airports, terminal operators, integrated operators, warehouses and distributors according to the wco standards or equivalent supply chain security standards. an international corporate identity system should be established, covering the private businesses which offer a high degree of security guarantees in respect of their role in the supply chain. these business partners should receive tangible benefits from such partnerships in the form of expedited processing and other measures. the implementation of the aeo program secures the successful realization of the facilitation and security in international trade (u.s. customs and border protection, 2006) . in the definition of the wco customs-business partnership, you must attach importance to such partnerships (partnership), security (security), authorized certification (authorization), technology (technology), communication (communication), convenience (facilitation) these factors the key role, especially in the 21st century global advocacy and research to implement the customs network, will provide a great convenience for the customs and generally established business partnerships, but will also enable efficient international interoperability become an urgent requirement for the effective realization global supply chain security and trade facilitation to pave the way (u.s. customs and border protection, 2008). trade frictions and protectionism in international trade not only exacerbated the decline in global trade, but also the associated adverse impact on the supply chain quickly spread to the entire international trade. in practice, a variety of measures were established to promote trade facilitation, mostly reflected in the program and through trade facilitation, coordination of applicable laws and regulations, standardization and improvement of infrastructure such as international trade and create a simplified, coordinated transparent and predictable environment. therefore, the content of trade facilitation involves very extensive, including almost all aspects of business processes, including customs and cross-border system is the core of the problem, in addition to including transport, licensing, quarantine, electronic data transfer, payment, insurance and other financial requirements, business information and other aspects. in world bank “reform in the regular procedure of import and export”, the report states that “trade facilitation program, the problem is a national need for a comprehensive legal and institutional reforms, rather than the customs of a thing.” for a long time, international trade security is simply understood as to reduce trade frictions, eliminate trade barriers and trade liberalization. after the 9.11, the united states developed countries and regions, such as europe increasingly paying attention to the safety of international logistics. in response to developments, customs and relevant regional organizations urgent desire to implement a cooperation in the international” safeguards both convenient and unobstructed flow of goods, and can effectively achieve a ‘safe’ as the core of the customs management functions “of the new system.”kyoto conventio“early on partnership between customs and business are described, effective communication between the customs and the business community , in consultation and cooperation with the perfect balance is critical to achieving effective regulation and convenience in this process customs authorities will be law-abiding traders as partners. ” 2005, “ the partnership between customs and the business community ” as a “global framework of standards to secure and facilitate trade,” one of the two pillars, the “standard framework” for customs partnercertified operators (aeo) were clearly defined, provides enterprises become aeo must meet minimum standards and benefits obtained. in 2008, the “21st century published by atlantis press copyright: the authors 88 customs” documents made it clear that in the 21st century customs should establish strategic partnerships with trusted operators, the “mutual benefit and win-win” as its ultimate goal (tweddle, 2007). with the full launch of the doha round, wto started to promote cooperation between the wco, customs and the business community of positive interaction will become an integral part of the legal framework for international trade content. in order to improve the overall business management of the customs performance targets, actively carry out routine inspection, special inspection and verification inspection, the practical realization of customs supervision and space “front push shift” (gutiérrez et al., 2013). 2. customs risk management---the key to ensure trade facilitation and security gradually, wco realizes the basic factors to ensure security and facilitation in the global trade lie in the businesses as the risk in the enterprises is also the inherent trouble for customs management. for this problem, customs administration may avoid, transfer, accept yet take it under control or conduct some other risk-control policies. the adoption of these strategies is based on the customs’ information about the business. aeo system derives in this sense. it focuses not only on the importing or exporting goods, but also on the business internal risks involved in the trading activities. the management techniques, business structure, information flow, globalization, etc. may directly affect the inherent risks in the business. if the authorization limits to the traditional way of inspection into the business accounts related to import and export activities, but not considers the overall risk management, this may increase rather than control the risk effectively. the above mentioned example---e-control network of processing trade is a case in point. with the wide use of the existing management, business structure, process design and information technology in the modern business community, the traditional mode of customs external auditing may be unable to conduct the inspection on the revised data by the business. it is comparatively easier for the old approaches of auditing to check the small-medium-sized, off-line enterprises as they are not very much complicated, that’s why one can track down such cases in this field. the traditional way of customs auditing lacks effectiveness in identifying the inherent risks in the business. the adoption of the risk management mode will be helpful to the effective administration of the customs external auditing. the key to the implementation of aeo system is the application of customs external auditing based on risk management (fletcher, 2007). 2.1 conditions and requirements for the authorized economic operator authorization wco member administrations and businesses acknowledge the importance to secure the supply chain and facilitate the cross-border movement of goods. both the customs and the enterprises realize the enhancement of one aspect may lead to the effective function of the other. the authorized economic operators in the international supply chain should adopt the selfassessment system measured to the pre-determined security standards and best practices to ensure that their internal policies and procedures provide adequate safeguards against the compromise of their goods and containers until they have been released from customs control at destination. the customs-to-business partnership program, based on the trading mode of aeo, allows the enterprises to make the secure plans flexibly in view of their own characteristics. the customs administration and the aeo may both realize and jointly set out secure and cooperative measures, which should be based on the commercial mode of aeo and in written form and by testified procedures, reflect the willingness of aeo business partners, including manufacturers, product suppliers and vendors, to obey the security criteria set out in the wco framework. based on the risk-analysis, customs administration should inspect the business flow and security measures regularly according to the security procedures of the agreement made with the operators involved. integrate different statistic data from the executing units, help enterprises fill in the charts and collect valuable information, get to know company’s self-control system. to be specific, there are thirteen conditions and requirements for customs and businesses in the “authorized economic operator guidelines”. they are: (1) demonstrated compliance with customs requirements: customs shall take into account the demonstrated compliance history of a prospective aeo when considering the request for aeo status; (2) satisfactory system for management of commercial records: the aeo shall maintain timely, accurate, complete and verifiable records relating to import and export. maintenance of verifiable commercial records is an essential element in the security of the international trade supply chain; (3) financial viability: financial viability of the aeo is an important indicator of an ability to maintain and improve upon measures to secure the supply chain; (4) consultation, co-operation and communication: customs, other competent authorities and the aeo, at all levels, international, national and local, should consult regularly on matters of mutual interest, including supply chain security and facilitation measures, in a manner which will not jeopardize enforcement activities. the results of this consultation should contribute to customs development and maintenance of its risk management strategy; (5) published by atlantis press copyright: the authors 89 education, training and awareness: customs and aeos shall develop mechanisms for the education and training of personnel regarding security policies, recognition of deviations from those policies and understanding what actions must be taken in response to security lapses; (6) information exchange, access and confidentiality: customs and aeos, as part of an overall comprehensive strategy to secure sensitive information, shall develop or enhance the means by which entrusted information is protected against misuse and unauthorized alteration; (7) cargo security: customs and aeos shall establish and/or bolster measures to ensure that the integrity of cargo is maintained and that access controls are at the highest appropriate level, as well as establishing routine procedures that contribute to the security of cargo; (8) conveyance security: customs and aeos shall jointly work toward the establishment of effective control regimes, where not already provided for by other national or international regulatory mandate, to ensure that transport conveyances are capable of being effectively secured and maintained; (9) premises security: customs, after taking into account the views of aeos and their necessary compliance with mandatory international standards, shall establish the requirements for the implementation of meaningful customs-specific security enhancement protocols that secure buildings, as well as ensure the monitoring and controlling of exterior and interior perimeters; (10) personnel security: customs and aeos shall, based on their authorities and competencies, screen the background of prospective employees to the extent legally possible. in addition, they shall prohibit unauthorized access to facilities, transport conveyances, loading docks and cargo areas that may reasonably affect the security of those areas in the supply chain under their responsibility; (11) training partner security: customs shall establish aeo requirements and mechanism whereby the security of the global supply chain can be bolstered through the commitment of trading partners to voluntarily increase their security measures, as may be more fully set forth in supplemental national criteria; (12) crisis management and incident recovery: in order to minimize the impact of a disaster or terrorist incident, crisis management and recovery procedures should include advance planning and establishment of processes to operate in such extraordinary circumstances; (13) measurement, analysis and improvement: the aeo and customs should plan and implement monitoring, measurement, analysis and improvement processes. 2.2 application of risk management based on aeo system the traditional approach of customs external auditing includes concrete approaches as checking materials, examining goods, etc. in order to define and seize any illegal cases, which may finally lead to punishment or rectification of the enterprises concerned. nevertheless, the customs control aims at systematic supervision over companies’ operations from different angles but not at investigation into possible illegal cases (grainger, 2012). the application of the aeo system requires the customs administration to conduct the auditing based on risk assessment so as to evaluate and testify the enterprise’s legal status, to standardize its internal management and to enhance the sense of lawabiding and self-discipline. so, it is necessary to put the customs external auditing into action (haughton, 2011) . based on the acknowledgement of the business operations and the risks involved, this auditing evaluates overall risk-management of the companies comprehensively and provides more safeguards for the risk-control campaign conducted by customs administration. it limits the monitoring scope and focuses on high risk fields. furthermore, compliance testing and substance testing may be applied better. depending on the automated management system, this mode gets into different layers of businesses and tries to find out weaknesses and risks by way of examining the internal control system. then, it may further audit and check the truthfulness and legitimacy of the import and export operation accordingly. it is effective to enable the auditing officials to detect problems or vulnerability and identify legal status of the businesses correctly with less endeavor and shorter period of time (cp3 group, 2005). the main structure is below: fig. 1: the structure of the risk management according to the chart, customs auditing officials should acknowledge the internal control management of the personnel involved in the import and export activities, including the essential situation of the business, flow of the shipments, capital, information and manpower, etc. collect relevant internal control materials and conduct compliance testing one item after another to examine and prove the system’s integrity, realize andidentify the business operation risk evaluate business risk control determine residual audit risk residual audit risk management reduce risk enhance value published by atlantis press copyright: the authors 90 credibility, enforcement and validity. in the event of obvious weaknesses and risks of the system, the substance testing will be necessary for some selected items in order to detect if such vulnerability may lead to any illegal cases against national law and regulations. after the testing, it is time to evaluate the risk control, determine residual audit risk, and take different measures accordingly. if everything under auditing is ok without evident weakness upon the completion of the checking, we can draw the conclusion of “no problem”, and finish the auditing task. if there is any trouble, the government shall ask the business community involved to make some improvement and rectification. the companies won’t get through the checking until they meet the necessary requirements of the customs management. in case of any illegal cases against national law and regulations, a routine auditing will be adopted in time to conduct more items of substance testing in the light of risks and weaknesses. if there is any tax problem concerning the companies during the checking, we shall differentiate the situation respectively. with regard to overdue taxes they have to pay, they may get a pass. concerning a supplementary tax imposed on the business, they can’t get the pass. if there exist possible smuggling or other illegal cases against national law or regulations, the company won’t be through the auditing at all (rodwell et al., 2012） in different stages of the auditing, we should focus on different parts by way of appropriate means of inspection according to actual needs. in the first stage, we may make use of the approaches such as “business analysis framework”, “business risk model”, “business information flow”, “business performance review”, “business risk management process”, etc. in the second stage, we can adopt methods like “risk control process framework”, “risk control standard matrix”, “information credibility testing”, “risk control identification”, “risk control evaluation”, “risk control testing”, “control reliance decision tree”, etc. during the third stage---the period of determining residual audit risk，we should pay attention to the fact that if there is any risk which hasn’t reduced to the degree of the acceptable level. if there exists the possibility, examine if some of the checking procedures are not effectively used. in the final stage, we think about proper suggestions to the enterprises concerned for the improvement of their risk control process. therefore, the business can better satisfy the conditions and requirements of aeo system. 3. conclusion the new system of aeo aims at the establishment of a new partnership between customs and business, in order to ensure the security and facilitation in the global trade. yet, customs and businesses may have different benefits and sometimes they are conflicting. customs insist on more tariff, “collect all the duties that should be done accordingly” whereas the businesses hold the opinion of the fewer the better. the key to the issue is the credibility, qualification, law-abiding of the business and the fair, polite law-enforcing of the customs. any lapse of either part may affect and damage the partnership. the customs external auditing on the enterprises is to promote a harmonious co--operation and mutual development of customs building. aeo program needs not only customs internal information, but also other social support and feedback, such as the information relevant to the enterprises and commodities from the bureau of industry and commerce, administration of taxation, administration of foreign exchange control, banks, etc. we shall set up a good relationship with different departments. customs administration will contact and audit the businesses which may apply for aeo system. customs may set a model to demonstrate the capability to execute in accordance with law and to enforce the law for the people. it is also a significant channel to prevent possible illegal offences and create a better situation for import and export trade. through aeo program, business can strengthen a standard management system; enhance a sense of law abidance and self-discipline. a new partnership, aiming at winwin co-operation between customs and business will be formed, which may benefit customs development. in this respect, customs external auditing will shoulder heavier responsibility and have an ever larger arena through the aeo system. in a long run, the concern over trade security should be highlighted. the scope of the customs service should be extended to the main players in the international supply chain. the needs of smes should be addressed. risk management should be further incorporated into the aeo system. the existing preferential or facilitative measures should be integrated and streamlined. all the aeos can enjoy the uniform treatments nationwide and worldwide. references world customs organization, 2015. the framework of standards to secure and facilitate global trade. world customs organization, 2008. authorized economic operator guidelines. u.s. customs and border protection, 2006. container security initiative, 2006-2011 strategic plan, us (available at: http://www.cbp.gov). u.s. customs and border protection, 2008. customs trade partnership against terrorism, (available at: http://www.cbp.gov/xp/cgov/trade/cargo_security/ctpat/). cp3 group, 2005. benefits from implementation of the wco framework of standards to secure and facilitate global trade. (available at: http://www.cp3group.com/attachments /wco_benefits.pdf) cp3 group, 2006. aeo guidelines, (available at: http://www. cp3group.com/attachments/aeo%20guidelines.pdf) published by atlantis press copyright: the authors 91 fletcher, t., authorised economic operator (aeo) programs: ibm’s perspective, world customs journal. 1(2) (2007) 6166. tweddle, d., logistics，security and compliance: the part to be played by authorized economic operators (aeos) and data management，world customs journal. 1(2) (2007) 101-105 grainger, a., supply chain security: adding to a complex operational and institutional environment, world customs journal. 1(2) (2012) 17-30. gutiérrez, x. and hintsa, j. and wieser, p. and hameri, a-p, voluntary supply chain security program impact: an empirical study with basic member companies, world customs journal. 1(2) (2013) 31-48. haughton, m.a., examining the business case for shipper participation in canada-usa trade security programmes, international journal of logistics research and applications. 10(4) (2011) 315-331. rodwell, s., van eeckhout, p. reid，a. and walendowski, j., 2012. study: effects of counterfeiting on eu smes and a review of various public and private ipr enforcement initiatives and resources, (available at: http://ec.europa.eu/ enterprise/enterprise_policy/industry/doc/counterfeiting_m ain%20r eport_final.pdf) published by atlantis press copyright: the authors 92 1. aeo system---the base of the international trade facilitation and security 2. customs risk management---the key to ensure trade facilitation and security 2.1 conditions and requirements for the authorized economic operator authorization 3. conclusion references research article anchor-based goals and personality effects on hazard identification in risk assessment piers fleming*, , harry england school of psychology, university of east anglia, centre for behavioural and economic social science, norwich, nr4 7tj, uk 1. introduction risk assessment involves a structured identification of potential hazards and the processes and outcomes associated with them. it is typically carried out in a professional capacity as a legal or operational requirement. identifying hazards is the first step to mitigating, controlling or eliminating the likelihood and severity of accidents and thereby save lives, protect people and maintain organisational operations. hazard identification is also a legal requirement for uk employers as part of employee protection [1]. failure to identify hazards prevents subsequent mitigation strategies from being developed and then the likelihood and/or severity of negative outcomes cannot be limited. for example, the failure to recognise covid-19 risk at mass gatherings between 8 and 13 march 2020 in the uk led to tens of thousands gathering in crowds at football matches and a horse racing festival with no mitigation strategies; these gatherings were followed by increased viral infection rates [2]. risk assessment and hazard identification are vital to mitigate potential hazards. hazard identification is carried out in all types of organisation, big and small, but the focus is typically on safety-critical industries and processes. expert risk assessors and sophisticated hazard identification procedures are used when the consequences of failed risk assessment are likely to lead to loss of life and/or serious harm, e.g. commercial aviation or pharmaceuticals. however, even small organisations without obvious safety concerns must still conduct risk assessments with hazard identification to protect people. employees and the public might still be at risk from a small business which has no expert risk assessment capabilities. for example, falling, colliding and fire hazards are ubiquitous. the great fire of london, which devastated much of london in 1666, was caused by a fire in a bakery. the uk health and safety executive offers guidance for all sizes of uk organisations and recognises identifying hazards as the first of three minimal steps, prior to assessing likelihood and severity, and then managing the identified hazards [3]. in smaller organisations without obvious safety concerns, hazard identification and risk assessment is likely to be conducted by a non-expert without an established procedure. these hazard identifications often occur across multiple different contexts, even within the same organisation, in micro risk-assessments. these non expert hazard identifications and risk assessments are relied upon to protect people. however, hazard identification is not straightforward and variability in hazard identification is common. existing evidence from the empirical literature shows a considerable range in people’s ability to identify hazards and one possible explanation is experience. in documented risk assessments where hazard identification techniques were used, on average, only 66–90% of hazards were identified across different types of construction projects [4]. a study of australian miners found the majority of those surveyed failed to identify common hazards (such as slips, falls and manual handling) [5]. even experts are unlikely to have sufficient knowledge and experience to identify all hazards [4]. however, experience, while not sufficient on its own, does appear to increase hazard identification. parents who spent more time with toddlers identified fewer than half the childcare hazards presented but still identified more hazards then healthcare and day-care workers [6]. a r t i c l e i n f o article history received 17 july 2020 accepted 11 october 2020 keywords goal target anchor hazard risk identification assessment a b s t r a c t hazard identification is a crucial first step in risk assessment. there are many cases in which hazard identification is carried out by non-experts. one concern is that valid hazards are overlooked and so not considered for mitigation or prevention. this study examined whether a goal-setting anchor could encourage the identification of more hazards and so reduce the likelihood that they are overlooked. seventy-two participants were recruited to an online study to identify hazards in four vignettes. the participants were randomly allocated to a high or low anchor condition in which they were told that experts typically identify at least two or at least eight hazards. participants also completed a five-factor personality measure. it was found that, compared to the low anchor, the high anchor increased word count, time on task and number of hazards identified. the effect of the anchor on hazards identified was robust even taking into account personality, time on task and word count. conscientiousness was also associated with identifying more hazards. overall, the use of anchors to set goals for hazard identification offers a low-cost intervention to improve risk assessment for non-experts. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: p.fleming@uea.ac.uk journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 113–118 doi: https://doi.org/10.2991/jracr.k.201014.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0003-3066-5313 http://creativecommons.org/licenses/by-nc/4.0/ mailto:p.fleming%40uea.ac.uk?subject= https://doi.org/10.2991/jracr.k.201014.001 https://www.atlantis-press.com/journals/jracr 114 p. fleming and h. england / journal of risk analysis and crisis response 10(3) 113–118 older and more experienced workers both have fewer accidents [7,8]. the least experienced workers identified the fewest hazards in a hazard identification task [5]. although experience offers a partial explanation, other factors must contribute to individual differences in hazard identification. there is limited evidence regarding the influence of personality on hazard identification but there is a wider literature regarding personality, risk-taking and accidents; both literatures suggest a role for attention and personality. risk-taking, personality and accident research finds that extraversion is associated with more traffic accidents; agreeableness and conscientiousness are associated with fewer accidents in general [9]. these associations have been explained as a result of vigilance, care and arousal associated with both personality traits and accidents. actual hazard identification evidence confirms greater visual fixation, and therefore attention, with conscientiousness and less with extraversion [10]. it may be that personality differences in hazard identification are partly explained by individual differences in attention. attention has been directly linked to hazard identification in further empirical evidence. auditory attention is associated with reduced vehicle accident involvement [11]. eye movements are associated with visual hazard identification [12]. distracted workers recognised fewer hazards [13]. we also know that motivation and attention are linked [14]. supervisory feedback is effective in adjusting attention and safety behaviour [15–17]. therefore, motivation and conscious control offer an opportunity to intervene and increase hazard identification. one promising avenue for an experimental intervention on hazard identification is via anchor-based goal-setting. we know that vigilance can be improved by goal-setting (via feedback on expected and attained performance) [18]. setting challenging goals compared to easy or no goals is a highly robust method to motivate attention, effort and performance [19]. even very high, arbitrary goals set by anchors are able to increase self-generated goals and maintain commitment and performance [20]. in one study participants were asked to generate ideas either to “come up with as many uses as you can” or “set a challenging and specific goal for the number of uses you will generate during this next period; for example, 10/120/240”. performance with higher anchor-based goals was much higher than in the low or no anchor condition [20]. it is this conceptual framework which we have adopted in this study. challenging goals should motivate greater attention, effort and performance in the domain of risk assessment, therefore hazard identification should also improve with high compared to low anchors. hazard identification may be open to manipulation using goal setting anchors; it may also be influenced by personality factors. a vignette task for hazard identification offers the opportunity to examine people’s ability to make risk assessments. this is crucial to understand as a first step in risk assessment. it is also important to consider non-expert risk-assessors because in the case of small businesses in non-risk-critical industries or micro risk assessments (such as student internships) hazard identification will, in all likelihood, be carried out by non-experts. in previous goal-setting tasks and in risk assessments the number of valid hazards identified offers a measure of performance. we can also consider time spent and words written as supplementary measures of task performance and engagement. if goal-setting is successful in affecting participants hazard identification performance this has a number of potential benefits. primarily as a training tool – participants could be made aware that simply by looking for more hazards they would be likely to be successful. it is possible to introduce unreasonably high anchors non-directively to motivate greater performance [20]; however, great caution should be used in using anchors in a genuine risk assessment – too low and genuine hazards might be missed, and too high and the risk assessment may appear untrustworthily incomplete. nevertheless, if this cheap and powerful technique is effective it could be vitally useful in improving risk assessments and training. the following objectives will be considered: the effect of high or low goal-setting anchors on number of hazards identified in a risk assessment, time spent on a risk assessment and words written in a risk assessment; the effect of personality traits of conscientiousness, agreeableness, extraversion, neuroticism and openness to experience on number of hazards identified. we would expect a higher anchor to increase time spent on task, words written and hazards identified. we would expect conscientiousness and possibly agreeableness to be associated with more hazards identified and extraversion might be associated with fewer hazards identified. 2. materials and methods 2.1. participants a sample of uk undergraduate students were invited to take part in the study via either a social media advert or through university participant recruitment software offering course credit. the final sample consisted of 72 participants. seventy-eight participants completed the survey; one participant withdrew, two failed the attention screen question and one was excluded for completing the survey in under one third of the median time to complete (median completion time for the whole survey was 21 min). two further participants were excluded as outliers (± 3 sd) in time to complete – both took well over 2 h to complete the risk assessments alone – more than double the time of the next nearest participant. the final sample was 77.8% female with an average age of 23.3 years (sd = 9.34). 2.2. materials and procedure participants were directed to online consent and study information. they then completed four risk assessment vignettes in random order. each vignette described a different workplace role – mechanic, hairdresser, nightclub security and newsagent. participants were asked to identify hazards within the role using an unlimited free text description, based upon the vignette. the vignettes were based on examples provided by the uk health and safety executive [3]. at the end of each vignette participants were asked: “what risks/hazards can you identify to the role holder and public in the vignette above? experts usually find at least [2/8] hazards.” half of the participants were given the low anchor value (two hazards) for all vignettes and half were given the high anchor value (eight hazards) for all vignettes. participants were subsequently asked to complete a 44-item 5-factor personality scale – the big five inventory [21] on a 5-point likert-type scale from disagree p. fleming and h. england / journal of risk analysis and crisis response 10(3) 113–118 115 strongly to agree strongly. the scales measure extraversion, e.g. “is talkative” (a = 0.86); agreeableness, e.g. “likes to cooperate with others” (a = 0.82); conscientiousness, e.g. “does a thorough job” (a = 0.78); neuroticism, e.g. “worries a lot” (a = 0.83) and openness, e.g. “is inventive” (a = 0.74). all items were presented in a random order within the scale. the final questions asked about demographics and an attention screen: “how many risk assessment vignettes were there in this survey?”. participant risk assessment responses were coded by two independent raters using the same coding scheme to count the number of valid hazards identified by each participant. the coding scheme was based upon an initial list of valid hazards devised during construction of the vignettes and updated iteratively to include any hazards identified by participants which were not on the initial list and deemed valid by the raters. each vignette included between 9 and 13 valid hazards to be identified (see table 1). all 72 participants’ responses were coded according to 46 hazard categories across the four vignettes leading to 3312 coded items. the two raters coded the vignettes, blind to the experimental condition codes, with an initial 3044 agreements and 268 disagreements (i.e. 91.9% agreement). the kappa for the two-raters pre-agreement hazards identified inter-rater reliability was κ = 0.81, p < 0.001, which indicates substantial agreement [22].1 a final inspection of all cases of rater-disagreement was conducted and a final coding was agreed between raters for all cases. this coding was aggregated for each participant to give a score for total hazards identified across all vignettes, which was used in the analysis. 3. results in addition to the total number of hazards correctly identified by each participant (as coded by two independent raters), two other measures of participant behaviour on the risk assessment task were calculated: (1) the total word count for the participant-response descriptions of the potential hazards and (2) the total time on task taken to identify hazards on the four risk assessment vignettes. a mean of 13.3 hazards (sd = 5.4) were identified by each participant across the four vignettes. total word count (mdn = 196 words, inter quartile range (iqr) = 109–257 words) and total time on task (mdn = 19.9 min, iqr = 11.0–17.4 min) were positively skewed. table 1 | averages and variability across the anchor conditions for hazards identified for each of the four vignettes vignette (total possible hazards) condition n mean (sd) min max t newsagent (12) low anchor 35 2.1 (1.40) 0 5 2.53* high anchor 37 3.0 (1.55) 0 6 – nightclub (13) low anchor 35 2.8 (1.64) 0 6 4.01** high anchor 37 4.4 (1.64) 2 8 – hairdresser (9) low anchor 35 2.6 (1.22) 0 5 4.50** high anchor 37 3.8 (1.17) 2 6 – mechanic (12) low anchor 35 3.5 (1.60) 1 7 1.91 high anchor 37 4.3 (1.88) 0 9 – *p < 0.05, **p < 0.001. table 2 | averages and variability across the anchor conditions for outcomes of the hazard identification task n hazards identified word count time on task mean (sd) mdn (iqr) mdn (iqr) low anchor 35 11.0 (4.87) 154 (90–233) 15.2 (7.7–21.6) high anchor 37 15.5* (5.05) 228† (157–296) 17.4‡ (12.7–29.1) *t(70) = 3.77, p < 0.001; †u = 415, z = −2.62, p = 0.009; ‡u = 510, z = −1.55, p = 0.121. 1re-analysis of the results using the rater 1 or 2 coding does not substantially affect the results obtained. 2the case produced a mahalanobis distance score of 26.7 and dfbeta for neuroticism = 1.3 which provide grounds for further consideration [23]. the regression analysis was run with and without the influential case. the coefficient for neuroticism was significant only when the influential case was included, it became non-significant when the influential case was omitted. the effect of the anchor condition (low anchor = at least two hazards versus high anchor = at least eight hazards) was tested on hazards identified by vignette (table 1), total hazards identified, total word count and total time on task (table 2). we can see that there was a trend toward greater hazard identification in the high anchor condition for each of the four vignettes (table 1). this difference was significant for three of the four vignettes. while some participants identified no hazards, only one found more than eight and no participants identified all of the possible hazards. there were significant effects of anchor condition in which total hazards identified and word count were higher with the high anchor, but there was no effect on time on task – see table 2. there was no significant difference between self-reported personality scores across anchor conditions (t < 1.5). the associations between the personality factors and the hazard identification tasks were examined using spearman rho correlations, see table 3. conscientiousness was associated with time on task, word count and hazards identified. word count was highly correlated with time on task, and hazards identified; hazards identified and time on task were moderately correlated. anchor condition was associated with word count and hazards identified. the pattern across the individual vignettes was similar to the total hazards identified. in tests of regression assumptions one influential case was found.2 the regression analysis reported below omits the influential case (see table s1 for the regression analysis including the influential case). the preceding analyses were not substantially affected by inclusion/exclusion of the influential case and are reported with the case included. the initial model was run with the five personality factors and the anchor manipulation (coded as 0 and 1 for low and high respectively), see table 4. conscientiousness was associated with significantly more hazards identified as was the high anchor condition. a second model was run including time on task and the total word count used in addressing the task; these significantly improved the regression model. in the second model word count and high anchor condition were associated with more hazards identified; conscientiousness was no longer predictive of hazard identification. 4. discussion anchor-based goal-setting was effective in increasing the number of hazards identified in a range of vignettes. when participants 116 p. fleming and h. england / journal of risk analysis and crisis response 10(3) 113–118 table 4 | regression analysis predicting total hazards identified from personality factors, anchor condition (low or high), time and words used on task model variables b 95% ci se beta t p f r-squared r-squared change 1 constant −9.726 (−24.246, 4.794) 7.268 – −1.338 0.186 4.321 0.288* – extraversion 0.019 (−0.182, 0.220) 0.101 0.022 0.185 0.854 – – – agreeableness 0.065 (−0.126, 0.257) 0.096 0.077 0.683 0.497 – – – conscientiousness 0.284 (0.063, 0.505) 0.111 0.290 2.569 0.013 – – – neuroticism 0.116 (−0.109, 0.341) 0.113 0.123 1.026 0.309 – – – openness 0.176 (−0.027, 0.380) 0.102 0.191 1.728 0.089 – – – anchor condition(0 = low, 1 = high) 4.929 (2.578, 7.280) 1.177 0.454 4.188 <0.001 – – – 2 constant −5.337 (−17.086, 6.412) 5.878 – −0.908 0.367 9.709 0.556** 0.268** extraversion 0.025 (−0.138, 0.189) 0.082 0.030 0.309 0.759 – – – agreeableness −0.009 (−0.165, 0.147) 0.078 −0.010 −0.111 0.912 – – – conscientiousness 0.153 (−0.031, 0.337) 0.092 0.156 1.663 0.101 – – – neuroticism 0.074 (−0.109, 0.257) 0.092 0.079 0.805 0.424 – – – openness 0.146 (−0.022, 0.315) 0.084 0.159 1.733 0.088 – – – anchor condition(0 = low, 1 = high) 3.236 (1.241, 5.230) 0.998 0.298 3.243 0.002 – – – time on task −0.020 (−0.108, 0.068) 0.044 -0.048 −0.453 0.652 – – – word count 0.027 (0.017, 0.037) 0.005 0.587 5.474 <0.001 – – – *p < 0.01, **p < 0.001. table 3 | non-parametric, spearman’s rho correlations between personality factors and hazard identification tasks 1 e 2 a 3 c 4 n 5 o 6 ac 7 tot 8 wc 1 extraversion – – – – – – – – 2 agreeableness 0.219 – – – – – – – 3 conscientiousness 0.157 0.160 – – – – – – 4 neuroticism −0.408** −0.292* −0.156 – – – – – 5 openness 0.133 0.117 −0.150 −0.090 – – – – 6 anchor condition −0.157 −0.027 −0.094 0.131 −0.045 – – – 7 time on task −0.102 −0.002 0.262* 0.009 0.107 0.184 – – 8 word count 0.004 0.124 0.316** 0.009 –0.075 0.311** 0.599** – 9 total hazards identified 0.008 0.075 0.256* 0.088 0.046 0.373* 0.406** 0.672** 10 newsagent hazards −0.047 0.068 0.235* 0.150 0.102 0.281* 0.267* 0.464** 11 nightclub hazards 0.000 −0.012 0.297* 0.043 −0.056 0.424** 0.392** 0.628** 12 hairdresser hazards 0.034 0.161 0.143 0.094 0.066 0.459** 0.412** 0.608** 13 mechanic hazards 0.050 0.118 0.090 0.002 0.086 0.216 0.286* 0.584** *p < 0.05, **p < 0.01. were asked to make risk assessments they wrote more and identified more hazards when told that experts find at least eight rather than at least two hazards. unsurprisingly, word counts, time on task and hazard identification were all positively associated with each other. greater conscientiousness was associated with greater hazard identification and a higher word count in the risk assessment. the effect of the anchor was additional to the effect of word count, time on task and personality. the use of anchors to set goals appears to be a simple and effective manipulation to improve the hazard identification stage of risk assessment. in this study the higher anchor caused more words to be written in the attempt to identify hazards. however, the effect of the anchor on hazard identification remained even over and above the effect of word count. not only did the anchor appear to increase participant effort in writing more words to try to identify hazards, but more hazards were identified even when the effect of number of words was accounted for in the regression. it may be that people do not know how many hazards to expect to identify and that the anchoring cue provides a goal to aim for. we know that challenging goals increase performance [19,20]. in this study without a baseline condition we cannot be sure if a target of at least two hazards reduced performance or if at least eight hazards increased performance. only three participants identified eight hazards (i.e. met the target) and only one participant exceeded the target by identifying nine hazards (all in the high anchor condition) and it therefore seems unlikely that the high anchor provided a ceiling to hazard identification. however, a higher more aspirational high anchor might be appropriate in future research. a pilot of four participants prior to the study indicated that participants identified an average of 2.5 hazards per vignette when asked to complete the task without being exposed to an anchoring figure. by comparison, participants in the low anchor condition identified an average of 2.75 hazards per vignette, whilst those in the high anchor condition identified an average of 3.88 hazards per vignette. further research should examine the effect of different levels of anchors compared to a baseline with no anchors. personality did have some influence on hazard identification. the effect of conscientiousness on hazard identification is consistent with the research suggesting that greater conscientiousness is associated with fewer accidents. this might be attributed to greater effort, attention or motivation on the task. this is also consistent with the finding that conscientiousness was associated with word p. fleming and h. england / journal of risk analysis and crisis response 10(3) 113–118 117 count and time on task. although hazard identification is not the only reason conscientiousness might be associated with fewer accidents – hazard awareness is one very important contributor to accident avoidance (in addition to care around known hazards). it is also noteworthy that higher word count, potentially indicative of more effort, did not produce more irrelevant or inaccurate hazard suggestions – as evidenced by the highly positive correlation between word count and hazards identified. this suggests that any technique which might increase effort would be useful in improving hazard identification in risk assessments. further research might investigate the association of effort with conscientiousness by looking closely at information search and anchored goal-setting and conscientiousness as part of a hazard identification task. there are a range of possible applications of this finding. it would be useful to habitually inform trainee risk assessors that if they aim to identify a very high number, for example 25 hazards, they are more likely to get a comprehensive list. further testing could confirm if this is effective for even extremely unlikely anchors, for example 1000 hazards. prior research would suggest that even extreme goal-setting anchors are effective [20]. if extreme anchors are effective it is possible that this approach could be extended beyond training to actual risk assessments without the concern that too low an anchor would cause hazards to be missed. it is also possible that this approach might influence expert risk assessors and this could also be examined in future research. there are some limitations to this study, for example, the sample is relatively homogenous with most participants being young with limited work experience. the participants are not experts and are being asked to identify hazards from settings with which they may be quite unfamiliar. the vignettes and hazard classifications were developed for this study and the online nature of the task led to noisy data – particularly for the time on task measure. there was no control on participants who may have been distracted while completing the study. however, the results show a clear pattern despite this noise, and the analysis across vignettes and typical time on task are largely as expected. furthermore, future studies might consider other relevant factors in hazard identification – the effect here was primarily driven by anchor-based goal-setting and it may be that constructs such as self-efficacy are useful in understanding differences in goal-setting [24]. overall, this paper demonstrates the usefulness of anchors in hazard identification. important risk assessments are often carried out by non-experts. the first, crucial, step of risk assessment is hazard identification. this study demonstrates that hazard identification is much improved by high-anchor goals. appropriate bodies should consider using anchors to prompt increased hazard identification performance as a low-cost intervention. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution he designed the study, collected the data and co-wrote the report; pf supervised the project, analysed the data and co-wrote the report. supplementary material supplementary data related to this article can be found at https:// doi.org/10.2991/jracr.k.201014.001. references [1] hmso. the management of health and safety at work regulations. uk statutory instruments. 1999. available from: http://www.legislation.gov.uk/uksi/1999/3242/regulation/3/ made. [2] nuki p. revealed: the three uk sporting events that may have led to a coronavirus death spike. the telegraph [internet]. 2020. available from: https://www.telegraph.co.uk/global-health/ science-and-disease/revealed-three-uk-sporting-events-mayhave-led-coronavirus-death/. [3] health and safety executive. managing risks and risk assessment at work [internet]. available from: https://www.hse.gov. uk/simple-health-safety/risk/steps-needed-to-manage-risk.htm (accessed march 18, 2020). [4] carter g, smith sd. safety hazard identification on construction projects. j constr eng manage 2006;132:197–205. [5] bahn st. workplace hazard identification: what do people know and how is it done?. in: proceedings of annual conference of the association of industrial relations academics australia and new zealand, gold coast (queensland): griffith university; 2012, pp. 1–9. [6] gaines j, schwebel dc. recognition of home injury risks by novice parents of toddlers. accid anal prev 2009;41:1070–4. [7] butler mc, jones ap. perceived leader behavior, individual characteristics, and injury occurrence in hazardous work environments. j appl psychol 1979;64:299–304. [8] furnham a. the psychology of behaviour at work: the individual in the organization. 2nd ed. hove (uk): psychology press; 2012. [9] clarke s, robertson it. a meta-analytic review of the big five personality factors and accident involvement in occupational and non-occupational settings. j occup organ psychol 2005;78: 355–76. [10] hasanzadeh s, dao b, esmaeili b, dodd md. role of personality in construction safety: investigating the relationships between personality, attentional failure, and hazard identification under fall-hazard conditions. j constr eng manage 2019;145:04019052. [11] arthur w, barret gv, alexander ra. prediction of vehicular accident involvement: a meta-analysis. human perform 1991;4: 89–105. [12] hasanzadeh s, esmaeili b, dodd md. impact of construction workers’ hazard identification skills on their visual attention. j constr eng manage 2017;143:04017070. [13] namian m, albert a, feng j. effect of distraction on hazard recognition and safety risk perception. j constr eng manage 2018;144:04018008. [14] engelmann jb, pessoa l. motivation sharpens exogenous spatial attention. motivation sci 2014;1:64–72. [15] barkan r. using a signal detection safety model to simulate managerial expectations and supervisory feedback. organ behav human decision process 2002;89:1005–31. [16] bruce mcafee r, winn ar. the use of incentives/feedback to enhance work place safety: a critique of the literature. j saf res 1989;20:7–19. https://doi.org/10.2991/jracr.k.201014.001 https://doi.org/10.2991/jracr.k.201014.001 http://www.legislation.gov.uk/uksi/1999/3242/regulation/3/made http://www.legislation.gov.uk/uksi/1999/3242/regulation/3/made https://www.telegraph.co.uk/global-health/science-and-disease/revealed-three-uk-sporting-events-may-have-led-coronavirus-death/ https://www.telegraph.co.uk/global-health/science-and-disease/revealed-three-uk-sporting-events-may-have-led-coronavirus-death/ https://www.telegraph.co.uk/global-health/science-and-disease/revealed-three-uk-sporting-events-may-have-led-coronavirus-death/ https://www.hse.gov.uk/simple-health-safety/risk/steps-needed-to-manage-risk.htm https://www.hse.gov.uk/simple-health-safety/risk/steps-needed-to-manage-risk.htm https://doi.org/10.1061/%28asce%290733-9364%282006%29132:2%28197%29 https://doi.org/10.1061/%28asce%290733-9364%282006%29132:2%28197%29 https://doi.org/10.1016/j.aap.2009.06.010 https://doi.org/10.1016/j.aap.2009.06.010 https://doi.org/10.1037/0021-9010.64.3.299 https://doi.org/10.1037/0021-9010.64.3.299 https://doi.org/10.1037/0021-9010.64.3.299 https://doi.org/10.1348/096317905x26183 https://doi.org/10.1348/096317905x26183 https://doi.org/10.1348/096317905x26183 https://doi.org/10.1348/096317905x26183 https://doi.org/10.1061/%28asce%29co.1943-7862.0001673 https://doi.org/10.1061/%28asce%29co.1943-7862.0001673 https://doi.org/10.1061/%28asce%29co.1943-7862.0001673 https://doi.org/10.1061/%28asce%29co.1943-7862.0001673 https://doi.org/10.1207/s15327043hup0402_1 https://doi.org/10.1207/s15327043hup0402_1 https://doi.org/10.1207/s15327043hup0402_1 https://doi.org/10.1061/%28asce%29co.1943-7862.0001373 https://doi.org/10.1061/%28asce%29co.1943-7862.0001373 https://doi.org/10.1061/%28asce%29co.1943-7862.0001373 https://doi.org/10.1061/%28asce%29co.1943-7862.0001459 https://doi.org/10.1061/%28asce%29co.1943-7862.0001459 https://doi.org/10.1061/%28asce%29co.1943-7862.0001459 https://doi.org/10.1037/2333-8113.1.s.64 https://doi.org/10.1037/2333-8113.1.s.64 https://doi.org/10.1016/s0749-5978%2802%2900007-9 https://doi.org/10.1016/s0749-5978%2802%2900007-9 https://doi.org/10.1016/s0749-5978%2802%2900007-9 https://doi.org/10.1016/0022-4375%2889%2990003-0 https://doi.org/10.1016/0022-4375%2889%2990003-0 https://doi.org/10.1016/0022-4375%2889%2990003-0 118 p. fleming and h. england / journal of risk analysis and crisis response 10(3) 113–118 [17] zohar d. a group-level model of safety climate: testing the effect of group climate on microaccidents in manufacturing jobs. j appl psychol 2000;85:587–96. [18] warm js, riechmann sw, grasha af, seibel b. motivation in vigilance: a test of the goal-setting hypothesis of the effectiveness of knowledge of results. bull psychon soc 1973;1:291–2. [19] locke ea, shaw kn, saari lm, latham gp. goal setting and task performance: 1969–1980. psychol bull 1981;90:125–52. [20] hinsz vb, kalnbach lr, lorentz nr. using judgmental anchors to establish challenging self-set goals without jeopardizing commitment. organ behav human decision process 1997;71:287–308. [21] john op, srivastava s. the big five trait taxonomy: history, measurement, and theoretical perspectives. in: pervin la, john op, editors. handbook of personality: theory and research, vol. 2. new york: guilford press; 1999, pp. 102–38. [22] landis jr, koch gg. the measurement of observer agreement for categorical data. biometrics 1977;33:159–74. [23] field a. discovering statistics using ibm spss statistics. 5th ed. london (uk): sage publications; 2018. [24] zimmerman bj, bandura a, martinez-pons m. self-motivation for academic attainment: the role of self-efficacy beliefs and personal goal setting. am educ res j 1992;29:663–76. https://doi.org/10.1037/0021-9010.85.4.587 https://doi.org/10.1037/0021-9010.85.4.587 https://doi.org/10.1037/0021-9010.85.4.587 https://doi.org/10.3758/bf03334347 https://doi.org/10.3758/bf03334347 https://doi.org/10.3758/bf03334347 https://doi.org/10.1037/0033-2909.90.1.125 https://doi.org/10.1037/0033-2909.90.1.125 https://doi.org/10.1006/obhd.1997.2723 https://doi.org/10.1006/obhd.1997.2723 https://doi.org/10.1006/obhd.1997.2723 https://doi.org/10.2307/2529310 https://doi.org/10.2307/2529310 https://doi.org/10.3102/00028312029003663 https://doi.org/10.3102/00028312029003663 https://doi.org/10.3102/00028312029003663 atlantis press journal style emerging risks: methodology, classification and policy implications ortwin renn stuttgart research center for interdisciplinary risk and innovation studies, stuttgart university seidenstr. 36, d-70174 stuttgart, germany e-mail: ortwin.renn@sowi.uni-stuttgart.de abstract good risk governance seems to rest on the three components: knowledge, legally prescribed procedures and social values. a ll three components are of particular importance for assessing and managing emerging risks, which are characterized by a lack of knowledge about the likelihood and magnitude of potential positive and negative consequences. this paper reports first about a protocol of how to govern emerging risks and then analyses the patterns of risks that would fall under the emerging risk category. six patterns were identified: system breakdown risks, amplifier risks, highly volatile and pervasive risks, psychosomatic risks, social risks, and knowledge management risks. for each of these risk patterns, the main characteristics and the policy implications are described and analyzed. keywords: emerging risk, risk governance, international risk governance council (irgc), risk management protocol, risk patterns, risk-policies 1. introduction emerging risks are the result of new or future threats, where there is low level of knowledge (or no knowledge at all) about the potential losses as well as the probability distribution of their occurrence (oecd, 2003). in some cases, this knowledge may be contested among actors with different interests or viewpoints. conventional approaches to projecting loss size, relative frequencies or probability distributions over time or severity of consequences are consequently most often ineffective if applied to emerging risks. attempts to assess emerging risks with technical or scientific instruments may prove inadequate as scientific understanding of emerging risks can change rapidly. therefore, adaptability and flexibility are vital to manage emerging risks in terms of individual, social, and economic impacts (kocharov, 2010). the international risk governance council (irgc) defines emerging risks as those types of risks, which are new, or familiar risks in new or unfamiliar conditions (irgc, 2009). e merging risks refer to threats that are perceived to be potentially significant but which may not be fully understood or assessed, thus not allowing risk management options to be developed with confidence. some emerging risks do not prove to be as significant as originally feared but others might prove to be worse than expected, with a high potential for loss of safety margins. emerging risks can be further broken down into three distinct, but overlapping categories, based on their main characteristic at a certain stage. these categories are described as follows (mazri and florin, 2014): • emerging technologies with emerging risk profile based on high uncertainty and lack of knowledge about potential impacts and interactions with the affected risk absorbing systems. nanotechnology and many other new technological developments would fit that description. • emerging technological systems with emerging interactions and systemic dependencies. the main issue here is the not the risk of the technologies journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 114-132 published by atlantis press copyright: the authors 114 willieb typewritten text received 10 august 2014 willieb typewritten text accepted 15 august 2014 willieb typewritten text (that may be known or well estimated) but the interactions of these risk (and also benefits) with other types of risks or activities that could lead to non-linear impacts and surprises. examples here are biofuels, climate change issues, all kind of it related risks, and critical infrastructure issues. • established technologies in a new emerging context or environment: the main problem here is that familiar technologies are operated in a new context or in different organizational settings that may change both the probability as well as the magnitude of potential impacts. one could also include here risks driven by complacency and overconfidence on one’s own ability to cope with sudden crisis. examples here are established technologies as a target for terrorist attacks, technologies in context of social or societal crisis (like civil unrest); new stress situation due to organizational changes such as mergers or just-intime production or the shift from public to private services. new assessment and management methods and procedures are particularly needed for emerging risks. they are at the crossroads between natural events (partially altered and amplified by human action such as the emission of greenhouse gases), economic, social and technological developments and policy driven actions, both at the domestic and the international level. emerging risks also face specific problems with respect to public perception and risk governance structures. risk management and governance processes dealing with emerging risks suffer from a loss of prior knowledge and frequently public plausibility. this paper has made the attempt to identify and classify some of the most likely candidates for emerging risk on all three levels described above. section 2 will describe the environmental and social context in which new emerging threats may arise. section 3 will introduce a protocol that the irgc has developed for identifying and managing emerging risks. this protocol will be applied to a set of global risk scenarios developed by leading institutions such as the world economic forum. six risk clusters emerged from this analysis as likely candidates for new emerging threats. each of these clusters is briefly described in section 4. the conclusions summarize the results and provide an outlook for future challenges. 2. the context for emerging risks the context for emerging developments and threats is characterized by increased vulnerabilities and interconnections between geographic areas as well as functional dependencies between the various sectors of society such as the physical world, the economy, the social relationships and the political cultures (stern, 2013). the potential for new risks to involve is likely to become augmented because of: increase of population and population density1; increase of population exposed to natural hazards and technological risks (dramatic increase in losses due to natural disasters over the last four decades, during the last decade natural hazard disasters have resulted annually in some 79,000 fatalities, with 200 million people affected) 2; increased use of hazard-prone land for productive purposes (for example 40 of the 50 fastest growing urban centres in the world are located in earthquake-endangered areas) (randall et al. 1996); increased interdependencies between natural, technical, social, and cultural hazards (fischerkowalski and amann, 2001); expected increase of hazard intensity due to climate change and other human interventions into geochemical cycles (rosa and dietz, 2010); growing diversity with respect to lifestyles and subcultures within societies (beck, 1992; beck, 2000). at a time, when the disaster potential increases, the coping mechanisms of many societies appear to become less effective. vulnerability is likely to increase due to (rosa et al. 2004): speed of urbanization (probably two thirds of the world population will live in cities after 2020) (jones and w.a. kandel, 1992); insufficient speed in building infrastructure to cope with it 3; coupling of independent risk sources (perrow, 1984) 1 according to the medium variant of the 2010 revision of world population prospects, the world population is expected to increase from 6.9 billion in mid-2011 to 9.3 billion in 2050 and to reach 10.1 billion by 2100. http://esa.un.org/wpp/analytical-figures/htm/fig_1.htm (accessed: july 10, 2014) 2 munich reinsurance (2013): natcatservice, munich-re: munich. see also: oecd (2003), p. 14. 3 oecd (2003), 44ff published by atlantis press copyright: the authors 115 interaction of natural disasters with chemical, technological, lifestyle, and social risks4; increase of mobility, migration and cultural derooting (black et al. 2013) loss of traditional management capabilities5; increase of social pressure and conflicts (iadociola and shupe, 2013); lack of capacity for mitigation and contingency management, etc 6. given these new challenges the world needs a concerted effort to deal with emerging risks. in particular new methodological as well as institutional solutions involving the different levels of risk governance at the local, national, international and global level are required as a means to provide adequate tools for limiting and managing the new emerging threats that will be embedded in the context of the global changes and challenges listed above (renn, 2008). 3. the irgc protocol for classifying emerging risk clusters what are eligible criteria to characterize risks that are not yet visible on the monitoring screen of risk management institutions? many risk management and governance institutions have been working on a methodology or a protocol to identify emerging risk or to propose a methodology of how to identify these risks. for our analysis i will use an approach that has been recently advocated by the international risk governance council (irgc) 7 . the protocol for identifying emerging risks is based on five steps, shown in fig. 1. 4 world economic forum (2013): global risks 2013. 8. edition. wec: geneva, pp. 63f. http://www3.weforum. org/docs/wef_globalrisks_report_2013.pdf (access: august 10) 5 wissenschaftlicher beirat der bundesregierung globale umweltveränderungen (wbgu) (2000): world in transition: strategies for managing global environmental risks. springer: berlin. 6 international federation of the red cross and red crescent societies (2013): world disasters report 2013. international federation: geneva. 7 the following paragraphs are partially taken from: irgc (2010): the emergence of risk. contributing factors. irgc: geneva and irgc (2014): emerging risk governance protocol. extended version. written by c. mazri and m.-v. florin. irgc: lausanne step 1: make sense of the presence and explore the future this first step of the irgc protocol aims to describe the current and possible future contexts, in which risks and opportunities may develop. it provides an understanding of the whole process with the required inputs from the environment regarding potential existing or emerging threats and opportunities. this step aims to continuously screen and make sense of the context, in which risk management takes place, and to feed the whole emerging risk governance process with reliable and updated inputs. the main goal is early warning or identification of potential threats to the organization. step 2: developing scenarios based on models and narratives at this stage, risk management agencies or organizations have identified threats and opportunities that deserve further analysis for exploring their impacts on risk governance in the future. by adopting a forwardlooking approach, this second step will rely on scenarios and narratives to describe how these threats and opportunities may become risks that require management or competitive advantages that can be seized. developing scenarios implies a p rocess of looking to the future and developing possible stories about how the future may unfold. scenarios represent combinations of formal models and plausible narratives (helgeson et al. 2013). depending on the topic and the state-of-the-art of available knowledge, scenarios differ in the composition of formal modeling and storytelling. some scenarios pursue well-known causal or functional relationships and vary only in their assumptions. in contrast, others come close to science fiction stories developing imaginative futures based on basic knowledge, formal logic and plausibility. almost all scenarios include methods for involving multiple actors and perspectives in describing what the future could look like. this is an open process and the resulting output represents the views of the participants. scenarios are not predicting the future. in the specific context of emerging risk governance, scenarios are rather meant for structuring and organizing the deep uncertainties decision makers are confronted with when anticipating the possible future threats and opportunities. they are constructed with the goal to enhance the ability to deal with the inherently uncertain and complex character of different and heterogeneous published by atlantis press copyright: the authors 116 social and cultural environments (healey and hodgkinson, 2012). the benefits of these approaches range from considering uncertainty in strategic decision making to organizational learning and the construction of common understanding among participants. the objectives of this second step are therefore to: • develop as detailed as possible scenarios exploring the possibility that threats and opportunities will affect risk management in the future. • provide evidence and support for future decisions about the threats and opportunities identified in the first step. • continuously update these scenarios according to changes observed in the environment and knowledge evolution. step 3: strategic decision making an important component of effective emerging risk governance is making the right decision at the right time. this exercise implies linking decision science with traditional risk governance, especially in the face of situations that are new, complex, uncertain or ambiguous. the purpose of the third phase of the protocol is to exploit all the scanning, sense-making and foresight capabilities deployed in the previous phases to fig. 1. irgc emerging risk governance protocol. 2. developing scenarios based on models and narratives 3. taking decisions at a strategic level option 2: developing precautionary approaches option 3: reducing vulnerability option 4: modifying the organisation’s risk appetite in line with risks option 5: using conventional risk governance instruments option 6: doing nothing 4. implementing strategies 5. reviewing risks and opportunities development 1. making sense of the present and exploring the future risk manager: emerging risk cycle communication option 1: acting on shaping context conditions published by atlantis press copyright: the authors 117 generate and select proactive strategies in dealing with emerging risks. practically, this implies to: • explore and identify opportunities for decisions, according to scenarios developed at step 2. • explore and describe the set of possible decision options. a t this stage, no option should be excluded. • identify windows of opportunity within each scenario that provide the most adequate conditions for initiating effective interventions. • identify thresholds of irreversibility after which interventions would no longer be effective. • define criteria for each intervention point for selecting, among the various options identified, those that align with the decision maker’s strategy for dealing with emerging risks. this is done according to the selected decision-making style and appetite for risk. • decide which emerging risk management option(s) will be implemented. • communicate about this process in a t ransparent manner. step 4: strategic option selection the irgc protocol discusses six different strategic options that may result from a thorough analysis in the third step: a) acting on context-shaping conditions: the first strategy is to change, modify or manipulate the context conditions that could lead to an emerging threat. one could for example decouple mutually dependent risk sources, remove risk amplifiers from the local risk source or train people to avoid specific situations that could lead to emerging risks. if context variables that matter are known and also changeable by the risk decision maker, this is the preferred option for risk management as it keeps the opportunities alive and prevents the risk to emerge. such ideal situations are rarely given but one can at least search for strategies to make context conditions more risk averse. b) developing precautionary approaches: aiming to avoiding the risk –or at least avoiding the occurrence of any irreversible consequence of the riskcould represent a valuable decision option in cases where a risk evaluation leads to high fears of unacceptable consequences. within this category of options, consideration and work should accordingly be dedicated to identify possibilities of actions where opportunities are maintained but the likely consequences of risk exposure are limited and at least partially reversible. of course, this subtle balance cannot be maintained forever and it is important to identify the point where actions that are more decisive may need to be taken. c) reducing vulnerability: if the decision making body cannot identify any opportunity for action on the sequence of events described in the scenario or if this intervention is considered inappropriately or too costly, reduction of exposure or vulnerability to the risk agent can be a strategic option. to do so, several possibilities can be considered: • reducing sensitivity to the risk by developing redundancies, improve personnel training or readjust protection capabilities. for instance, in the case of extreme events from natural hazards such as hurricanes, flooding, heat waves, or pandemics, one can base the protection strategy on t he worst-case scenario in order to reduce the vulnerability to the potentially affected population. however, this comes at a cost which may not be affordable. • building resilience and adaptation abilities to prepare for unexpected events. this strategy implies going beyond the anticipated worstcase scenarios to ensure that organisations will be able to rebound after any shock, adapt to any new circumstances and recover from unknown types or levels of stress. decentralizing decisions, enabling self organisation and social networking, and promoting diversity are examples of approaches allowing for resilience development. d) modifying the decision maker’s risk appetite in line with a new risk: the 2008 financial crisis revealed a significant imbalance between the risks some financial institutions were willing to take (risk appetite) and their risk taking capacities (the maximum amount of risk they can sustain, or their tolerance to risk) (barfield, 2007). among other consequences, this has demonstrated the need for increased transparency in financial governance and published by atlantis press copyright: the authors 118 for better consideration of risk appetite and risk taking capacity. e) resort to conventional risk management: this type of option for emerging risk management is partcularely relevant in cases where known risks develop in new or changing context conditions. it this case, it is reasonable to consider risk management options for familiar (as opposed to emerging) risks. t he risk mechanisms are well known but the governance would need to be revised in t he light of the new contextual developments. the most challenging part when dealing with these risks is therefore the awareness of the contextual evolution.this option would apply, for example, if malaria would develop in europe (or in general to higher latitudes), as a consequence of climate change. f) doing nothing: if, at the end of step 3, the decision makers conclude that the scenario/ narrative of a certain potential threat (identified at step 1) does not lead to anticipation that the threat will develop as a risk to the potentially affected popukation, then they can propose to “do nothing”. this means that the threat is not considered by the decision makers as plausible enough, or its impact is considered as negligible, to trigger any action at the present time. this does not prevent risk managers from doing more monitoring, as things may change over time. the six types of options described above are suggestions and starting points to trigger reflection and creativity in generating strategies. enhanced creativity and imagination in this exercise is likely to provide decision makers with more satisfactory strategies in the third and fourth step. step 5: reviewing risk development and decision after the decisions have been taken (step 3) and implemented (step 4), it is necessary to monitor how threats and opportunities unfold and review the relevance of decisions made. the monitoring of consequences of decisions taken and their systematic review is a k ey step in each governance process. it provides decision makers with the ability to revise their decisions if necessary and to adapt them as knowledge becomes available and as the context changes. in the context of emerging risk governance, both monitoring and reviewing become even more crucial as the whole protocol is expected to provide the decision making body with proactive capabilities. the final objectives of step 5 are therefore to: • demonstrate that the decision-making body is performing effectively in dealing with opportunities and risks that emerge from changing and evolving contexts. by pursuing a pathway of adaptive mangement, the decision makers learn step by step how to implement the decision and to monitor its intended and unintended consequences • provide valuable feedback loops to reiterate steps 1 and 2 of the process. • provide insights on the protocol’s strengths and weaknesses (engaging in a reflexive process). the protocol is a systematic and replicable framework with the aim to support decision makers in defining and revising strategies for emerging risk governance. it aims to cover all stages of risk emergence, from early signals of emergence of a r isk until the risk materializes or disappears. it promotes and organizes the development of a s et of capabilities required to deal with complex, evolving and uncertain environments: detect and interpret early the seeds of change that may create risks and/or opportunities in the present or future. help decision makers imagine the future contexts with which they may be confronted if no actions are taken. through communication, foster creativity and transparency when investigating possible decision options and identifying appropriate timing of intervention. constantly monitor contextual changes and adapt the organization’s pathway accordingly. there are challenges associated with these tasks. overcoming them is not only a matter of models and tools. it is also about creating organizational dynamics, building leadership at the senior level and overcoming cognitive limitations when confronted with deep uncertainties. accordingly, a cr oss-functional coordination is crucial for steering the required resources and expertise towards the same direction, triggering and facilitating interactions and finally ensuring scientific soundness of the process. published by atlantis press copyright: the authors 119 4. a systematic review of emerging risk clusters on the basis of the irgc emerging risk protocol the author initiated a systematic inquiry into a set of potential candidates for emerging threats that matched one of the three emerging risk categories and that had been identified and explored by various teams of experts and practioners. the scenarios selected and analyzed for this exercise were taken from the oecd, the world economic forum, swiss re, the bertelsmann foundation and risk worlds 8. these scenarios match the criteria developed for the first two steps of the irgc protocol, i.e. context analysis (step 1) and narrative scenario writing (step 2). the author added some policy implications (step 3-5) at the end of each subsection. the summary categories used in this article to pool risks into clusters are created by the author and are meant to reflect and summarize the outcomes of the different scenario exercises. methodologically, this could be classified as a n arrative meta-analysis (koricheva et al. 2013). 4.1. type 1: system breakdown risks (including cyber-risks) system breakdown risks are characterized by a break in a causal hazard chain, and this can occur within financial systems as experienced in the 2008 crisis, communication network systems or natural hazard response or relief systems. the threat of system breakdown is a feature of an interconnected world and it exists at many levels, from local to global. the main cause for system breakdown is the lack of governance capability and coherence in the face of 8 how the emerging risk scenarios have been identified, analyzed and summarizes, has been described in detail in: renn (2014). the sources for the scenarios are: oecd (2003): emerging risks in the 21st century. an agenda for action. oecd: paris; oecd (2012): global shocks: improving risk governance. oecd: paris; löfstedt, r. (ed.) (2003): riskworld. risk research, 6 (1–4), 289–619; world economic forum (2013): global risks 2013. wec: geneva; swiss re (2012): identifying emerging risk zuerich http://www.swissre.com/rethinking/emerging_risks/ emerging_riskhtml (access: july, 19, 2014); arpe, j.; glockner, h.; hauschild, h.; petersen, t.; schaich, a. und volkmann, t. (2012): die ökonomischen risiken der globalisierung. ergebnisse einer expertenund einer bürgerbefragung. global choices, 1. bertelsmann stiftung: osnabrück, germany. interconnectedness of localities and functional systems9. the coping mechanisms in place to manage these risks vary across countries and at global level. system breakdown risks are not severely affected by societal risk perception or cultural views it is the “visible” breakdown risks that have to be addressed and managed. system breakdown risks are promoted by: interconnectivity of hazardous systems such as natural and technological (water dam built in an earthquake prone zone); increase of vulnerabilities by constructing a multitude of low-probability, high-consequence technological devices (such as a liquid gas terminal next to a chemical factory); transfer of hazardous technologies in politically unstable or potentially violent societies; increased mobility of humans through migration and tourism (the world tourism organization counts on 1602 million tourist arrivals in 2020)10; and move into impact dimensions that show non-linear patterns. the last point deserves some more explanation. all cause-effect functions are linear if the delta increase is purely incremental and if the present location of the system on the cause-effect function is distant from possible non-linear thresholds (ulanowicz, 2013). both conditions may be jeopardized in the face of more recent developments. first, many technological, economic and social changes occur more rapidly than in the past making larger than incremental steps more probable. second, human activities have the potential to affect the ecological, economic and social fabric of human existence more profoundly than in the past. many analysts have called this the “intensity factor”(gheorghe and vamanu, 2004). the intensity of impacts has grown and, along with the increased influence on the existing systems, the likelihood that we are approaching non-linear thresholds increases as well. there are three prominent examples in this field: the first one is infectious disease, the second one nutrition, and the third infrastructure failure. the first two threats 9 governance deficits were mentioned as the main cause of emerging risks in all of the analyzed documents. an early version of this problem has been described in castells, m. (1996): the rise of the network society. blackwell: oxford. 10 oecd (2003), p. 24. published by atlantis press copyright: the authors 120 are related to a basic evolutionary characteristic: the more “food” is offered on nature’s market the more incentives are mobilized to use it as a new supply for other “hungry” organisms. this characteristic holds true in terms of volume of mutations as well as speed of evolution. as the number of humans approach 6.7 billion, the opportunity to serve as a food supply invites feeding organisms to utilize this cornucopian reservoir. in terms of evolutionary flexibility and mutational speed, this threat may be confined to microbes, in particular bacteria and viruses (gaze et al. 2013; håberg, 2013). they have extreme adaptations capabilities and their mutation rate is fast and effective. major sources for new infections may be the transfer of animal viruses onto humans as well as specialised bacteria that develop resistance against antibiotics. the spread of these newly developed microbes is facilitated and promoted through mobility (garrett, 1995). the centre of disease control in atlanta discovers a new virus or strain of bacteria almost every month (cdc, 2000). so far, modern pharmaceutical technology has been able to combat most of the newly evolved infections. yet the example of aids, and more recently ebola, exemplifies that a single failure in this attempt can cost millions of lives. furthermore, global environmental change can transport infectious diseases in areas that have not been accustomed to this new threat before (lipp et al. 2002). the second example is food supply for human consumption. the number of people is, of course, related to the volume of food produced to feed these people. such a r eservoir of nutrients attracts other organisms that would like to feed on them (mahul and gohin, 1999) 11 . again microbes are here of most eminent importance, but fungi and insects may also play a role. this threat is amplified by the continuous trend towards less and less plant species used for human consumption12. at this point in time, almost 80% of all serial food is confined to seven different plants. wheat and rice alone account for almost half of the world’s serial production. if anyone of these central serial species became infected by a new type of disease for which human ingenuity had no immediate cure 11 see also: wissenschaftlicher beirat der bundesregierung globale umweltveränderungen (wbgu) (2000): world in transition: strategies for managing global environmental risks. springer : berlin, pp. 81ff. 12 wissenschaftlicher beirat der bundesregierung globale umweltveränderungen (wbgu) (2000), pp. 93ff available the world would face a s erious famine with catastrophic results. the third example relates to technical infrastructure (kröger, 2008; llanso, 2013)13. human activities have become more and more interconnected and mutually dependent. such networks are based on critical infrastructures such as transportation means, communication routes, energy supply and, most importantly, information and communication services. many economic, political and social transactions rely on the availability and functionability of the existing infrastructure. as soon as infrastructural services are disturbed or malfunctioning, the original activity is likely to collapse and may infect other systems that rely on the well-functioning of the affected system. the most prominent example here is the emergence of more and more cyber-risks, including cyber-terrorism but also risks to privacy and civil liberties (neitzke, 2008). modern societies are characterized by an increasing vulnerability that will make system breakouts more probable, and awareness of the counter-movements has increased since the 11th september attack. there is now a competition race to improve resilience of all kinds of systems (hémond and robert, 2012; gallopin, 2006)14. for example the us defence department has protected its system by creating isolated islands where sensitive information is physically disconnected from the system rather than using firewalls for protection. this lack of connectivity could create their own risks, as the new system will only work if the division line between systems is self-reliant, and where there is problem of vital information not being available on a bounded system. what kind of policy options (step 4 and 5) would be available and recommendable for this emerging risk? the ideal societal response to these types of risks would be a g lobal risk management board, at the oecd or global level. due to the interconnected nature of the 13 see also: the white house (2013). critical infrastructure security and resilience. presidential policy directive/ppd-21, february 12, 2013. archived by webcite® at: http://www.webcitation.org/6q3ivri up. (access: august 4, 2014) 14 in particular to the threat of terrorism: committee on science and technology for countering terrorism (2002): making the nation safer: the role of science and technology in countering terrorism. national research council: national academies press: washington d.c. published by atlantis press copyright: the authors 121 http://www.webcitation.org/6q3ivri%20up http://www.webcitation.org/6q3ivri%20up risk, a national or trans-national forum (such as the european union) would be insufficient. there are a number of major constraints to achieving such a global risk management board. every system invites freeriding, and a global system that managed and enforced standards would also threaten the advantages for freeriders (dietz et al. 2003; falk et al. 2010). similarly, there are always winners when a s ystem collapses, so there would be countries or institutions that stood to gain by not participating or obstructing the creation of a comprehensive response to these system breakdown risks. therefore, the creation of an international risk governance council such as the irgc has been particularly relevant for this new cluster of risks. the implications for risk management are substantial. the potential costs of such disasters are beyond the scope of what even reinsures are able to pay. around 13 million people die every year from infectious diseases, primarily measles, pneumonia, cholera, aids, tuberculosis and malaria15. failures of infrastructure, in particularly dams and high-rises, cost thousands of lives each year. some analysts estimated that a total collapse of the three-gorges-dam presently built in china could kill as many as 1.3 mio people (wbgu, 1999). a series of risk assessments over the world demonstrates increasing rather than decreasing catastrophic potential over the last two decades (barrett, 2014). the probability for such damages to occur is difficult to calculate. this is particularly true for new health threats since mutations occur randomly and, in spite of an increased speed of mutations in the face of attractive food supplies, the most adaptive candidate may come up sooner or later (truly stochastic event). even calculating failure probabilities for critical infrastructure has become increasingly difficult due to the dependencies of failures from operational error or even sabotage and terrorism (aven and renn, 2009). this makes it hard for risk management institutions such as insurers or regulatory agencies to deal with breakdown risk. in addition, such risks may percolate though almost all types of conventional risk governance policies, starting with individual health insurance to liability legislation and social security systems. the best what risk management agencies can do is to pursue the resilience-based strategy. i.e. to limit the catastrophic potential independent of the low probability of a 15 oecd (2003), p.15 catastrophic release and to strive for decoupling of interconnected risk-inducing activities. 4.2. type 2: amplifier risks amplifier risks originate as physical risks, which then extend into the financial, political and psychological worlds. the effects of these risks might not have as high a physical impact as other risks such as breakdown risks, but they share the attribute of having a highly symbolic value. these risks tend to become amplified in the form described by the theory of social amplification of risks (kasperson et al. , 1998). the physical impact may be negligible but the quality or the circumstances of these impacts amplify these consequences in the perception of the observers and move them into behavioural activities that pose serious damage to society as a whole. amplifier risks are usually characterized by high complexity, uncertainty and ambiguity (renn, 2011). complexity refers to the difficulty of identifying and quantifying causal links between a multitude of potential candidates and specific adverse effects (marshall, 2013). the nature of this difficulty may be traced back to interactive effects among these candidates (synergism and antagonisms), long delay periods between cause and effect, inter-individual variation, intervening variables, and others. it is precisely these complexities that make sophisticated scientific investigations necessary since the cause-effect relationship is neither obvious nor directly observable. complexity requires systemic assessment procedures and the incorporation of new mathematical tools such as non-linear regression and fuzzy set theory. uncertainty is different from complexity. it comprises different and distinct components. they all have one feature in common: uncertainty reduces the strength of confidence in the estimated cause and effect chain. if complexity cannot be resolved by scientific methods, uncertainty increases. even simple relationships, however, may be associated with high uncertainty if either the knowledge base is missing or the effect is stochastic by its own nature. the last term in this context is ambiguity or ambivalence (renn and klinke, 2012). this term denotes the variability of (legitimate) interpretations based on identical observations or data assessments. most of the scientific disputes in the fields of risk analysis and management do nor refer to differences in published by atlantis press copyright: the authors 122 methodology, measurements or dose-response functions, but to the question of what all this means for human health and environmental protection. emission data is hardly disputed. most experts debate, however, whether an emission of x constitutes a s erious threat to the environment or to human health. another question in this context is: should regulation be confined to avoid significant health effects or should it be expanded to any measurable effect that could cause some still unknown damage (cf. the discussion on electromagnetic fields, emf)? again high complexity and uncertainty favour the emergence of ambiguity, but there are also quite a few simple and almost certain risks that can cause controversy and thus ambiguity. the main feature of amplifier risks is that high complexity, uncertainty and ambiguity are all associated with risks of this type. the causal chain is highly complex encompassing different types of consequences ranging from loss of human lives to financial losses. second, the risks are linked to high uncertainty, even indeterminacy because they are based on non-linear relationships. thirdly, the consequences are evaluated differently depending on social position, status and economic well-being. these damages include consumer boycotts, stock devaluation, removal of trust and confidence in risk management agencies and others. until now, risk management agencies have not adequately addressed these new amplifier risks. what is clearly needed, is a holistic and systemic concept of how to characterize, assess, and evaluate these risks. in addition, risk managers need to clarify what type of procedure is demanded for obtaining effective, efficient and politically feasible risk reduction results for coping with this risk cluster. 'mad cow disease' or bse appears to be the perfect example of an amplifier risk (dressel, 2000). the bse crisis occurred on the time between 1985 and 1996, the year when bse was discovered and the year when the transmission to humans became the most likely explanation for the occurrence of a n ew variant creutzfeldt-jakob disease. although the physical risk to human health turned out to be relatively small (in all of europe less than 150 cases of cjd), the repercussions on financial, economic, political and social affairs was tremendous. the cumulative expenditure by uk agricultural departments in response to the bse crisis, from 1996 to 2001, was estimated to have reached gbp 4.2 billion. the bulk of the expenditure has been given for compensation of commercial enterprises, especially compensation to farmers for the removal of cattle over 30 months of age from the human food chain and support to the slaughtering and rendering industries. other costs since 1996 that have been borne by public expenditure include the public inquiry into bse, at an estimated gbp 25 million (bse inquiry, 2001). the costs to the private sector of bse have also been considerable. the ban on british beef exports in march 1996 led to the complete loss of a trade worth gbp 700 million per year. in the first 12 months since march 1996, the total value of the market for uk produced beef fell by an estimated 36% in real terms (a combination of loss of exports and the drop in domestic demand), amounting to an estimated loss of value added to the uk economy of gbp 1.15 billion. in addition to these economic problems, consumers lost faith in the regulatory system, political officials were forced out of office and all over europe new food agencies were founded or old ones restructured. one of the initiating events for the disaster to happen was the decision to lower combustion temperature for animal feed. this decision was based on a cost-benefit analysis, which demonstrated a gain of max. gbp 30 million for doing so. at the time several scientists warned the british agricultural ministry to use more caution in this field as cows may be more sensitive to pathogens in animal feed since being vegetarians they have not developed any resilience against animal-borne diseases. it is certainly not proven that lowering the temperature has been the main reason for the spread of bse, yet the case demonstrates the narrow focus under which such a decision was made. the consequences that followed have changed the regulatory landscape in europe and cost taxpayers multi-million losses. retroactively, everyone knows better, but if a council of interdisciplinary scientists had been given the opportunity to review the situation, one could at least be more confident that the risks would have been appraised in a more competent manner. how can one deal with these amplifier risks? an idealized societal response to these risks would be the formation of cross-disciplinary risk management agencies, perhaps situated within existing institutions. in the terminology of the irgc strategies this would be equivalent to reducing vulnerabilities. such agencies would be required to link the physical, financial and political (governance) links between the risks. the published by atlantis press copyright: the authors 123 symbolic aspects cannot be addressed in isolation from the links, but without the symbolic assessment the links would pass unnoticed. institutional fragmentation and disciplinary thinking pose constraints to this type of societal risk management structure. in addition, the rationality of decision-makers and risk managers tends to underestimate the symbolic value of certain risks. 4.3. type 3: highly volatile and pervasive risks these risks cover those human activities that promote rapid environmental or cultural changes without proper knowledge of the consequences. trial and error, the usual method of testing innovations, may be unacceptable because the error is so costly in terms of human lives or money that nobody is prepared to pay for them. in addition, because of the complex nature of these risks, outcomes can often not be associated with direct causes. therefore, impacts are simulated with computer models as a m eans to avoid the real trial and error experience. the validity of such modeling efforts, however, remains contested (wildavsky, 1979). such risks are inherently both uncertain and unpredictable. all systems can cope with change if it is gradual enough to allow for adaptability, but the issue with these types of risks is the speed of change. there are many cases where a pervasive risk has been introduced, which has created profound changes that have not been good for humans. the best example is global change (national research council, 2010). this problem is normally framed as an environmental threat. rarely have risk assessment methods been used for dealing with this issue. yet there is an increasing amount of evidence that many natural disasters are being caused or promoted by global change. although it is too early to attribute the increasing losses of natural disasters to global climate change, it is an experiment with uncertain but potential catastrophic outcomes for human civilization (helmer and hillhorst, 2006). climate change is only the spearhead of a class of risk that can be characterized as global interventions of humans in natural cycles. the problem here is that the resulting effects are still widely uncertain but there is also hardly any escape for any society to avoid being affected when the worst case is going to happen. 16 16 renn (2014), pp.367ff. a second example is related to basic innovations such as nano-technology (shatkin, 2013). it is a truism that the risks and benefits of new technologies are uncertain as long as the scope of impacts is still unclear. problems occur, however, if, as stated above, the usual path of trial and error cannot be followed because the error is beyond what societies are willing to pay in terms of a r isk premium. this price is determined by two components: the preferences of (affluent) societies to tolerate new unknown risks (aversion factor) and the potential intensity of impacts that might be expected from these new technologies. with respect to nano-technologies analysts such as bill joy have warned the public that the impact of these new technologies in terms of social costs will outweigh the potential benefits by far (joy, 2000). innovation may become impeded from two sides: the high risk aversion factor of most people in industrial societies and the high impact factor for risks across the board of dimensions (health, environment, political, psychological) associated with new technologies. in response to this type of emerging risk the irgc protocol would probably recommend to pursue the precautionary concept. an ideal institutional management mechanism would be based understanding on the precautionary approach as part of the regulatory regime, not only with regard to the effects, i.e. the known impacts, but also focusing on the process, i.e. the speed of change (klinke and renn, 2001). it is not always possible to have control over this speed, and it would take a very courageous effort to pay special attention to the speeding up processes, even where the negative impacts are not noticeable. the precautionary principle, if well understood, provides an institutional means for increasing reversibility, if the processes are found to be more negative than initially perceived (bennet, 2000). such an approach is the more necessary the more ubiquitous, persistent and accumulative a r isk appears to be. ubiquity, persistence and accumulations are indicators for irreversibility, and should the alarm be sounded, society has to ensure controllability and reversibility in the face of risk. for example, the european regime for nanoparticles in food and cosmetics has not banned their use, but has tried to ensure that a gradual application in the market will allow the possibility of retreat should the risks turn out to be damaging (grobe et al. 2008). published by atlantis press copyright: the authors 124 simultaneously, they have introduced financial subsidies to develop new detection and risk assessment methods. a major constraint with the adoption of the precautionary principle is the potential for arbitrary judgments (charnley and elliott, 2000). there is a dilemma between the conflicting requirements for solutions fast enough to be innovative and slow enough to be reversible, in the face of scientific uncertainty. there are no simple negative/affirmative decisions, but a need to develop a set of good criteria for judging the right balance and speed of change. at global level, there is a growing scientific movement pressing for precaution. there is now a strong interest to find better regulatory regimes to balance speed of change with the necessity of innovation. high uncertainty risks demand management efforts based on market incentives. private insurance companies may play a s ignificant role here. in addition to judging risks according to the usual parameters damage potential and probability, other criteria such as ubiquity of hazard, persistence, and reversibility need to be considered (renn and sellke, 2001). such a multicriteria-analysis can be modeled in principle, yet all insurance companies will have problems with risks where the uncertainty analysis does not provide reliable probability estimates. 4.4. type 4: psychosomatic risks the nature of these types of risks is both ubiquitous and chronic. psychosomatic risks occur at both personal and institutional levels. they can be identified in the form of some internal or mental health, often associated with depression (kessler at al. 2003). psychosomatic risk stems from the lack of orientation within a modernized society one that struggles with proving convincing concepts for sense or meaning (alexander, 2013). the issue here is how to find a way out of the current fashion of moral relativism and pluralism of values and worldviews. at an institutional level, psychosomatic risks are related to performance and lead to an inconsistency in both institutional values and behavior. table 1 provides a list of major risks that the harvard school of public health has identified as main causes of death in the year 1990 and a p rediction of these main causes of death or chronic diseases for the year 2020 (klinke and renn, 1999). one can clearly see that the authors believe that infectious diseases will be of minor importance in the year 2020 (which according to our analysis is rather doubtful) while lifestyle risks and accidents will occupy top positions in the list of the most likely causes of death or chronic diseases. the 2020 pandora box of risks contains familiar threats such as heart attacks and cancer but also less familiar risks such as depressions and brain disorders. the list demonstrates that the balance between mind and soul, a vital condition for individual well-being, is in jeopardy. one of the main challenges in this respect will be the reconciliation between the material opportunities and the psychological and spiritual needs of humans. table 1. likely developments of risks. hazard statistics: causes of death and chronic diseases 1990 2020 infections heart attacks diarrhea depressions death at birth accidents depressions brain disorders heart attack chronic bronchitis cancer cancer the risks associated with electromagnetic fields can also be placed in this context (wiedemann and schütz, 2005). it is not a question of physical malfunctions that underlie objective verifiable data, but it concerns subjective perceptions or subjective feelings impeding on efficiency, which can lead to psychosomatic malfunctions. the decisive aspect is the question on the subjective risk perception of the affected people. the risks associated with electromagnetic fields have become a p opular subject of the media coverage and move into the centre of public debate. electromagnetic fields are perceived as risks due to the range of results from professional as well as from allegedly professional risk studies. the vast majority of serious scientific researches come to the conclusion that electromagnetic fields generate effects, which are usually not hazardous for the vast majority of human beings, i.e. neither epidemiological nor toxicological significant changes are noticeable (kostoff and lau, 2013). the risk debate about electromagnetic fields also reflects a d iscussion among experts where trust and reliability are challenged (renn, 2006). recognized scientists often maintain antagonistic standpoints leading to a situation that scientific reliable knowledge is doubted by the public. every expert‘s opinion provokes counter expertise. especially, thresholds are published by atlantis press copyright: the authors 125 subjects of public distrust. the degree of uncertainty among experts is interpreted as ignorance or as bribery (schütz and wiedemann, 2005). psychosomatic risks exhibit cultural variations, and the challenge is to find a set of universal values that overcome these differences. where there are none, or alternately, a l ot of collective values that need to be bridged there is a need to invent these universal values. the constraints for achieving a s ocietal response to these psychosomatic risks are the diversity of humans across the globe, and the plurality of views they hold. the increasing divide between the rich and poor makes the search more difficult and reduces the credibility of this mission. there will be a rise in the number of psychosomatic responses, and there is no clear countertrend or civil force that would drive towards this to this. this type of risk requires real action, as it decreases productivity and can also decrease trust in political institutions. in contrast to somatic illnesses that are of a t emporary nature, psychosomatic illnesses are chronic. individuals who suffer from this trend tend to isolate themselves from society, and become more of a drain than an asset to society. in terms of the irgc risk strategies, the resilience and adaptation model may be the best to pursue for this category of emerging risks. one needs to strengthen the resilience of people to find their own identity in a world of plural identities and of an inexhaustible appetite of the intellectual elite to deconstruct existing worldviews 17. at the same time, societal institutions can help individuals to develop better adaptive capacities to cope with orientation crisis and develop more resistance against self-doubts. 4.5. type 5: social risks to society social risks are understood in this context as consequences of social actions that impact human health and the environment or their perception (excluding genuine social risks such as lack of education or shelter). most prominent risks here are violence, crime, terrorism and sabotage (findlay, 2013; zimmermann, 2011). these risks have become particularly prominent 17 “modernity cannot survive in a purely objective and rational form. it needs romance, a connection to the inner life. modernity must be connected to subjectivity if the modern self is to be repaired and thrive”. from: alexander (2013), p. 146, since the attack on the world trade center. many analysts believe that, apart from infectious diseases, most lives are lost due to human induced violence (ca. 13 million fatalities annually due to infectious diseases; 79,000 fatalities annually due to natural hazards; ca. 8000 fatalities due to technological hazards; between 25 million each year thorough violence) 18 . human violence may become even a more prominent risk in modern interconnected societies since the infrastructure offers more opportunities for inducing break-down damages with relatively little effort. large water reservoirs for drinking water, high buildings, locally concentrated chemical facilities; mass-flow transport lines, large-scale technologies and many other new developments increase the overall vulnerability of societies. such vulnerabilities invite potential terrorists to take advantage of this situation19. the probability of terrorist attacks is likely to increase due to (moghaddam, 2010): the widening of the gap between the 10% richest and 10% poorest countries (same is true on the individual basis) 20 the political tensions due to hegemonic power struggles after the cold war; increase of fundamentalist positions in the world; dissatisfaction of particular groups with globalization processes and lack of social integration for a growing amount of displaced intellectuals in many countries. the attack on the world trade center has already been a major challenge to risk management institutions, in particular re-insurance companies (hartwig, 2002). most of the targets that are attractive to terrorists represent “accomplishments” of capitalist economies. in 18 data on natural, technological and health hazards in. oecd: (2003), p. 14-15; data on violence in: l. richardson (1990): statistics of deadly quarrels. boxwood press: pittsburg; and p. swabury and a. codevilla (1989): war: ends and means. basic books: new york 19 oecd: (2003), p. 209. 20 the institute for policy studies tracks the gap between ceo’s and american workers. the average ratio between salaries for a ch ief executive and american worker ranges around 350 to one. the avaerage pay in the united states is about $ 20 per hour. the ceo of wal mart earns $ 30-million a year, which is $ 15,000 per hour: about the wages of a year-round minimumwage worker. taken from: http://chronicle.com/ blogs/brainstorm/economists-fail-to-justify-obscene-ceo -pay/25500, (access: july 15, 2014). published by atlantis press copyright: the authors 126 most cases these targets are well-insured. the premiums, however, were calculated on the assumption of a low probability of terrorist attacks, if this had been included at all. other human-induced risks may get overlooked under the deep shock released by the recent terrorists attacks. yet, civil war, crime, and anomy (absence of legal enforcement) cost many more lives than terrorism (lozano, 2013). most of these risks are not addressed by any risk governance regimes. yet the impacts of violent behavior produce many indirect effects such as an increase in infectious diseases, famines, technological failures and many others. there is a second class of social risks that are related to changes in values, perceptions and attitudes. these risks are related to social definitions of what is framed as damage, causing agent or blame. for example, the term “human health” has been gradually expanded to include aspects of well-being and self-reliance (emson, 1987). medical treatment is extended to those who are not ill but feel deprived in one way or the other. lifestyle pharmaceuticals as well as wellness treatments are telling examples of such a new mental image of health. obviously changes in what is included in health care have major ramification on risk management and its social cost. changes in values also have repercussions on issues such as burden of proof (in tort cases), imposition of strict liability or degree of evidence for making someone accountable for any damage. even minor changes in the liability system can trigger large changes in damage claims. with respect to social value changes, it is difficult to predict the most likely developments of the future. but a few trends seems to prevail (wilkinson, 2013): affluent societies become more preoccupied with the notion of health and well-being. this will be reinforced by the trend of aging. the share of the population aged above 60 i s expected to reach 33% in 2050 in developed countries, compared to 19% in 2000 21 . at the same time, new technologies and pharmaceutical developments will enable the medical staff to tailor the treatment to individual needs of patients (hayes et al. 2014). the costs for such treatments will increase exponentially. it is still unclear how 21 oecd (2003), p. 18. social security systems will be able to handle such cost increases. modern science provides improved and more detailed models of complex cause-effect relationships. these new insights will help to identify causes for diseases or other type of damages that are presently associated with random events or background noise. this may lead to the emergence of surprises and new insights that may force risk managers to change their activities. moreover such surprises trigger losses of public trust and confidence in risk management institutions. legal reform in developed countries tends to emphasize the protection of the weaker part in a lawsuit. in particular legal systems with a strong jury influence make often economically powerful actors more accountable for their actions even if the evidence is less compelling. compensation is also likely to be awarded if the “victim” appears to be less informed than necessary in order to make prudent decisions on consumption. risk producers will be obliged to assure informed consent by those who use their products or are exposed to risks. in contrast to earlier times, it will make hardly any difference whether consumption or exposure have been voluntary or not. the main issue will be access to full information and an active effort by the risk producer for making all potential hazards known to the potential victims (odugbemi, 2008). taking all three trends together risk management institutions (public and private) may face an accelerated challenge. the social definition of what is regarded as harm includes more and more factors that were formerly interpreted as misgivings. in addition, science provides better models for new causal relationships and the legal system relaxes its provisions for what is called “hard evidence”. all three trends reinforce each other and will most likely lead to more developments framed as risks that, in turn, will demand collective risk reduction efforts. this trend may be counteracted, however, if a serious political movement towards more personal accountability of the consumer (individual and social) were to be launched. what can be done to develop a suitable strategy for dealing with social risks? may be a v ersion of the strategy “modifying the risk appetite in line with a new published by atlantis press copyright: the authors 127 risk” is appropriate. this strategy would, suggest that society places more responsibility on the individual and his or her own network to deal with social risks and to find an appropriate balance between opportunities given by plurality, increased options and individuality and the risks that go along with the loss of unambiguous values and group identity. a necessary condition for this is equal access to these opportunities, i.e. a co llective action to overcome the growing gap in the distribution of career, education and living opportunities (beck, 2013). 4.6. type 6: knowledge management risks (promoter) knowledge management risks refer to the quality and availability of information necessary to manage or reduce a g iven risk. although it may pose a risk in itself, it serves in most cases as a p romoter or modifier of other risk types. if a high quality knowledge management system is in place, the likelihood that any of the other risks become much more serious than anticipated, decreases considerably. it is closely related to system breakdown and amplifier risks. knowledge management describes the process of sorting through the high amount of information available and finding the right knowledge for the right purpose. this has always posed problems for society, but it now presents substantial risks to society, due to the high volumes of stored, possessed and generalized knowledge available (hayward and personick, 19981999). knowledge management risks could be a cau sal factor in the collapse of a particular system, or a mitigating factor in disaster management. one aspect of this risk is the challenge of dealing with uncertainty, and the need to select the right knowledge. a critical issue is acknowledgement of the boundaries of knowledge or ignorance, and the issues of framing, but achieving this is complicated by the abundance of information about certain aspects of knowledge, which can make these boundaries more difficult to distinguish (siegrist and cvetkovich, 2000). knowledge can be of a national, supranational or localized in character, and knowledge management needs to incorporate all these levels. for example, many of the agenda 21 pr ocesses on community collapsed because no universal criteria could be found to measure the sustainability of a particular community (bebbington and gray, 2001). societal management of knowledge management risks would have to include a shift in emphasis by research councils and sponsors to the identification and classification of information that is at present unavailable, and an expansion of knowledge outside the realm of the classic science fields. for example, decision-makers should be aware of the degree of uncertainty with regard to the consequences of nanoparticles in food and cosmetics. there is also a need for a better information brokering system, which cannot be fulfilled by the traditional universities. the development towards open access to all kind of knowledge pools leads to the need of having global knowledge brokers (not only providers) that help people as well as institutions to navigate through the jungle of knowledge claims. the constraints on the management of these risks are the difficulties in establishing the criteria for quality control of knowledge, and the degrees of freedom required for enlarging the knowledge base. there are many discussions on the virtual information transfer systems, but it is difficult to decide on the parameters needed to meet this function. one of the major factors in remaining competitive in a global market is the ability of global players to have fast and reliable information retrieval systems at hand. this is particularly true for risk management institutions. knowledge management risks serve more as promoters or attenuators of risks that constitute risks themselves. the irgc protocol has been designed to provide a orientation for selecting and processing knowledge. s uch orientations for knowledge management are needed for all the risk types that were discussed above. the better and the more accurate the knowledge basis the better the conditions for choosing the most appropriate risk management strategy. 5. conclusions to establish a framework for good governance on emerging risks, a stringent, logically well-structured and promising decision-making process is required. risk managers need substantive but flexible guidelines that are globally applicable to manage emerging risks. good governance seems to rest on t he three components: knowledge, legally prescribed procedures and social values. it has to reflect specific functions, from early warning (radar function), over new assessment and management tools leading to improved methods of published by atlantis press copyright: the authors 128 effective risk communication and participation. criteria of good governance have been discussed in many contexts (hubbard, 1999; oecd, 1995). they need to be transferred to risk-related issues and operationalized so that best practices can be identified and recommended. central items to be addressed are sound scientific expertise, burden of proof, consistency and coherence, non-discrimination, proportionality and costbenefit examination. in addition, governance structures should reflect criteria such as transparency, openness, accountability, effectiveness, and mediation of different/conflicting interests. the irgc emerging risk protocol is a first attempt to provide such an orientation for dealing with emerging threats. it has not been finalized and is under review by scholars and practioners from many countries and risk domains. yet a first application of the protocol in this article shows that it has the potential to identify emerging risk areas and suggest strategies to deal with many of the complex and ambiguous challenges posed by emerging risks worldwide. the promises of new developments and technological breakthroughs need to be balanced against the potential evils that the opening of pandora‘s box may entail. this balance is not easy to find as opportunities and risks are emerged in a cloud of uncertainty and ambiguity. the dual nature of risk as a potential for technological progress and as a s ocial threat demands a d ual strategy for risk management. it will be one of the most challenging tasks of the world risk management institutions to investigate and propose more effective, efficient and reliable methods of emerging risk assessment and risk management while, at the same time, ensure the path towards new innovations and technical breakthroughs. the late aaron wildavsky (1979) paraphrased this problem by giving one of his seminal articles the provocative title: “no risk is the highest risk of all”. poor and inadequate risk management, however, appears to be the second highest risk and a suitable candidate for disaster. the irgc remerging risk protocol might help risk managers to acquire necessary knowledge, articulate the insightful scenarios and develop the most appropriate risk management strategies for finding the right balance between opportunities and risks. references j. c. alexander, the dark side of modernity (polity, cambridge, 2013), p.144. t. aven and o. renn, the role of quantitative risk assessments for characterizing risk and uncertainty and delineating appropriate risk management options, with special emphasis on t errorism risk, risk analysis 29 (4) (2009), 587-600 r. barfield, risk appetite-how hungry are you? pricewaterhouse-coopers journal, special risk management edition, (pwc, london, 2007), http://docsfiles.com/pdf_ appetite.html (access: august 2, 2014). a. m. barrett, analyzing current and future catastrophic risks from emerging-threat technologies, current research project narratives, online publication, 63 (2014), http://research.create.usc.edu/current_synopses/63 (access august 5, 2014). j. bebbington and r. gray, an account of sustainability: failure, success and a r econceptualization, critical perspectives on accounting 12 (5) (2001) 557-588. u. beck. risk society: towards a new modernity. (sage, london, 1992). u. beck, risk society revisited: theory, politics and research programmes, in: the risk society and beyond. critical issues for social theory, eds. b. adam, u. beck and j. van loon (sage, thousand oaks, usa, 2000), pp. 211229. u. beck, why “class” is too soft a category to capture the explosiveness of social inequality at the beginning of the twenty-first century, britisch journal of sociology 64 (1) (2013) 63–73. p.g. bennet, applying the precautionary principle: a conceptual framework, in: foresight and precaution, volume 1, eds. m.p. cottam, d.w. harvey; r. p. paper and j. tait (a.a. balkema, rotterdam and brookfield, 2000), pp. 223-227. r. black, n. w. arnell, w. n. adger, d. thomas and a. geddes, migration, immobility and displacement outcomes following extreme events, environmental science & policy 27 (2013) 32-43. bse inquiry, executive summary of the report of the inquiry, (2001), available online at http://www.bseinquiry.gov.uk/ report/volume1/toc.htm (access: july 25, 2014). center for disease control and prevention (cdc): pandemic influenza: a planning guide for state and local officials, document 2.1. cdc: atlanta, 2000. g. charnley and e.d. elliott, risk versus precaution: a false dichotomy, in: foresight and precaution, volume 1, eds. m.p. cottam, d.w. harvey; r. p. paper and j. tait (a.a. balkema, rotterdam and brookfield, 2000), pp. 209-212. t. dietz, e. ostrom and p. stern, the struggle to govern the common, science 302 (2003) 1907–1912. k. dressel, the cultural politics of science and decisionmaking, an anglo-german comparison of risk political cultures: the bse case, (2000), available online at http://bse.airtime.co.uk/dressel.htm (access: july 30, 2014). h. e. emson, health, disease and illness: matters for definition, cmaj 136 (1987) 811-813. published by atlantis press copyright: the authors 129 a. falk, e. fehr and u. fischbacher, appropriating the commons: a theoretical explanation, in: the drama of the commons, eds. t. dietz, n. dolšak, e. ostrom, p. stern, s. stonich and e. weber (national academy press, washington, dc, 2002), pp. 157–191. m. j. findlay, governing through globalised crime (willan, cullumpton, uk, 2013). m. fischer-kowalski and c. amann, beyond ipat and kuznets curves: globalization as a vital factor in analysing the environmental impact of socio-economic metabolism, population and environment, 23 (1) (2001), 7-47. g. c. gallopin, linkages between vulnerability, resilience, and adaptive capacity, global environmental change 16 (2006) 293–303. l. garrett, the coming plague: newly emerging diseases in a world out of balance (penguin books, san francisco, 1995) w. h. gaze, s. m. krone, d. g. larsson, x .z. li, j. a. robinson, p. simonet and y. g. zhu, influence of humans on evolution and mobilization of environmental antibiotic resistome. emerging infectious diseases, 19 (7) (2013), doi: 10.3201/eid1907.120871. a. v. gheorghe and d. v. vamanu, complexity induced vulnerability, international journal of critical infrastructures 1(1) (2004) 76–84. a. grobe, o. renn and a. jaeger, risk governance of nanotechnology applications in food and cosmetics, irgc, geneva, 2008. s. e. håberg, l. trogstad, n. gunnes, a. j. wilcox, h. k. gjessing, s. o. samuelsen and c. stoltenberg, risk of fetal death after pandemic influenza virus infection or vaccination, new england journal of medicine 368 (4) (2013) 333-340. r. hartwig, one hundred minutes of terror that changed the global insurance industry forever. public policy paper. insurance information institute: new york. (2002). d. f. hayes, h. s. markus, r. d. leslie and e. j.topol, personalized medicine: risk prediction, targeted therapies and mobile health technology, bmc medicine 12 (1) (2014) 37-46. g. a. hayward and s.d. personick, protecting the infrastructures of the information age, telecordia exchange magazine 23 (1998-1999) 23-42. m. p. healey and g. p. hodgkinson, troubling futures: scenarios and scenario planning for organizational decision making, in: the oxford handbook of organizational decision, eds. d. g. p. hodgkinson and w. h. strabuck (oxford university press, oxford, 2012), pp. 565-587. j. helgeson, s. van der linden and i. chabay, the role of knowledge, learning and mental models in public perceptions of climate change related risks, in: learning for sustainability in times of accelerating change, eds. d. a. wals and p. b. corcoran (academic publishers, wageningen, 2013), pp. 329-346. m. helmer and d. hillhorst, natural disaster and climate change, disaster (special issue on climate change and disasters), 30 (1) (2006) 1–4. y. hémond, and b. robert, evaluation of state of resilience for a cr itical infrastructure in a co ntext of interdependencies, international journal of critical infrastructures 8 (2/3) (2012), 1-18. r ． hubbard, criteria of good governance, optimum, the journal of public sector management 30(2) (1999) 37-50. p. iadociola and a. shupe, violence, inequality and human freedom, third edition (rowman & littlefield, lanham, usa, 2013), pp. 451ff. irgc, emerging risks. sources, drivers and governance issues. irgc: geneva, 2009. b. g. jones and w.a. kandel, population growth, urbanization, disaster risk and vulnerability in metropolitan areas: a co nceptual framework, in: environmental management and urban vulnerability, discussion paper no. 168, a. kreimer and m. munasinghe (the world bank, washington d.c. ,1992). b. joy, why the future doesn’t need us, wired (april 8, 2000), 2. r. e. kasperson, o. renn, p. slovic, h. s brown, j. emel, r. goble, j. x. kasperson and s. ratick, the social amplification of risk. a conceptual framework, in: the earthscan reader in risk & modern society, eds. r. löfstedt and l. frewer (earthscan, london, 1998), pp. 149-162. r. c. kessler, p. berglund, o. demler, r. jin, d. koretz, k. r. merikangas and et al, the epidemiology of major depressive disorder: results from the national morbidity survey replication (ncs-r). journal of the american medical association 289 (2003) 3095–3105. a. klinke and o. renn, prometheus unbound. challenges of risk evaluation, risk classification, and risk management, working report no-.153, center of technology assessment in baden-württemberg, stuttgart, (1999), p. 8. a. klinke and o. renn, precautionary principle and discursive strategies: classifying and managing risks, journal of risk research 4(2) (2001) 159-174. a. kocharov, efsa’s identification of emerging risks, european food and feed law review, 3 (2010) 144-155. j. koricheva and j. gurevitch, place of meta-analysis among other methods of research synthesis, in: handbook of meta-analysis in ecology and évolution, eds. j. koricheva, j. gurevitch and k. mengersen (princeton university press, princeton, 2013), pp. 3-13. r. n. kostoff and c. g. lau, combined biological and health effects of electromagnetic fields and other agents in the published literature, technological forecasting and social change 80 (7) (2013) 1331-1349. w. kröger, critical infrastructures at risk: a need for a new conceptual approach and extended analytical tools, reliability engineering and system safety 93 (2008) 17811787 published by atlantis press copyright: the authors 130 e. k. lipp, a. huq and r. colwell, effects of global climate on infectious disease: the cholera model, clinical microbiology review 15(4) (2002) 757-770. t. llanso, g. tally, m. silberglitt and t. anderson, missionbased analysis for assessing cyber risk in critical infrastructure systems, in: critical infrastructure protection vii, volume 417, eds. j. butts and s. shenoi (springer, berlin and heidelberg, 2013), pp 201-214. r. lozano, m. naghavi, k. foreman, s. lim, k. shibuya, v. aboyans and m.cross, global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the global burden of disease study 2010. the lancet 380 (9859) (2013) 2095-2128. o. mahul and a. gohin, irreversible decision making in contagious animal disease control under uncertainty: an illustration using fmd in brittany, european review of agricultural economics 26 (1999) 39-58. g. r. marshall, transaction costs, collective action and adaptation in managing complex social–ecological systems, ecological economics 88 (2013) 185-194. c. mazri and m.-v. florin, emerging risk governance protocol, summary report, irgc: lausanne, 2014. f. m. moghaddam, the new global insecurity: how terrorism, environmental collapse, economic inequalities, and resource shortages are changing our world (praeger, santa barbara, 2010). national research council, advancing the science of climate change (national academies press, washington, d. c. 2010), pp. 102–105. h.-p. neitzke, m. calmbach, d. behrendt, kleinhückelkotten, e. wegner, and c. wippermann, risks of ubiquitous information and communication technologies, gaiaecological perspectives for science and society 17 (4) (2008), 362–369. h. odugbemi, public opinion, the public sphere, and quality of governance: an exploration, in: governance reform under real-world condition citizens, stakeholders, and voic, eds. h. odugbemi and t. jacobson (the world bank, washington, d. c., 2008), pp. 15–38. oecd, participatory development and good governance, oecd: paris, 1995. oecd, emerging risks in the 21st century: an agenda for action. oecd: paris, 2003. c. perrow, normal accidents (basic books, new york, 1984). j. b. randall, d. l. turcotte and w. klein, reduction and predictability of natural disasters. vol. xxv. santa-feeinstitute studies in the sciences of complexity. addisonwesley: new york. (1996) o. renn, risk communication insights and experiences in the emf debate, in: mobile emf and communication international perspectives, ed. m. nishizawa (national institute for earth science and disaster, tokyo and tsukuba, 2006), pp. 87-101. o. renn, risk governance. how to cope with uncertainty in a complex world (earthscan, london, 2008) o.renn, das risikoparadox. warum wir uns vor dem falschen fürchten (fischer, frankfurt am main, germany, 2014) o. renn and a. klinke, complexity, uncertainty and ambiguity in inclusive risk governance, in: risk and social theory in environmental management, eds. t.g. measham and s. lockie (csiro publishing, melbourne, australia, 2012), pp. 59-76. o. renn, a. klinke and m. van asselt, coping with complexity, uncertainty and ambiguity in risk governance: a synthesis, ambio, 40 (2) (2011) 231-246. o. renn and p. sellke, risk, society and policy making: risk governance in a co mplex world, international journal of performability engineering, 7 (4) (2001) 349-366. e. a. rosa and t. dietz, global transformations: passage to a new ecological era, in: human footprints on the global environment, eds. e.a. rosa, a. diekmann, t. dietz and c.c. jaeger (mit press, cambridge, usa, 2010), pp. 145. e. a. rosa, r. york and t. dietz, tracking the anthropogenic drivers of ecological impacts. ambio: a journal of the human environment, 33(8) (2004) 509-512. h. schütz and p. m. wiedemann, how to deal with dissent among experts. risk evaluation of emf in a scientific dialogue, journal of risk research 8 (6) (2005) 531 – 545. j. a. shatkin, defining risk assessment and how it is used for environmental protection, and its potential risk for managing nanotechnology risk, in: nantechnology. health and environmental risk, ed. j. a. shatkin (crc press, boca raton, usa, 2013), pp. 54. m. siegrist and g. cvetkovich, perception of hazards: the role of social trust and knowledge, risk analysis 20 (2000) 713–719. p. c. stern, design principles for governing risks from emerging technologies, in: structural human ecology. new essays in risk, energy and sustainability, eds. t. dietz and a. jorgenson (wsu press, pullman, washington, 2013), pp. 91-120. r. e. ulanowicz, circumscribed complexity in ecological networks, in: advances in network complexity, eds. m. dehner, a, mowshowitz and f. emmert-streib (wiley onlie library, doi: 10.1002/9783527670468.ch11, 2013), pp. 249-258. p. m. wiedemann and h. schütz, the precautionary principle and risk perception: experimental studies in the emf area, environmental health perspectives 113 (4) (2005) 402 – 405. a. wildavsky, views: no risk is the highest risk of all: a leading political scientist postulates that an overcautious attitude toward new technological developments may paralyze scientific endeavor and end up leaving us less safe than we were before, american scientist 3 (1979) 3237. a. wilkinson, elahi and e. eidinow, riskworld scenario, risk research, 6 (1–4) (2013) 297–334. wissenschaftlicher beirat der bundesregierung globale umweltveränderungen (wbgu), world in transition: published by atlantis press copyright: the authors 131 water resource management, annual report of 1998 (springer, heidelberg, 1999). e. zimmermann, globalisierung und t errorismus, in: globalisierung im fokus von politik, wirtschaft, gesellschaft, eds. t. mayer, r. meyer, l. miliopoulos, h. p. ohl and e. weede (vs verlag für sozialwissenschaften, wiesbaden, germany, 2011), pp. 183–198. published by atlantis press copyright: the authors 132 1. introduction 2. the context for emerging risks 3. the irgc protocol for classifying emerging risk clusters step 1: make sense of the presence and explore the future step 2: developing scenarios based on models and narratives step 3: strategic decision making step 4: strategic option selection step 5: reviewing risk development and decision 4. a systematic review of emerging risk clusters 4.1. type 1: system breakdown risks (including cyber-risks) 4.2. type 2: amplifier risks 4.3. type 3: highly volatile and pervasive risks 4.4. type 4: psychosomatic risks 4.5. type 5: social risks to society 4.6. type 6: knowledge management risks (promoter) 5. conclusions references atlantis press journal style received 18 february 2016 accepted 28 april 2016 overview on the developments and applications of hesitant fuzzy sets: an uncertain decision making tool bin zhu school of economics and management, tsinghua university, beijing 100084, china e-mail: binzhu@263.net zeshui xu business school, sichuan university, chengdu, sichuan 610064, china e-mail: xuzeshui@263.net abstract recently, hesitant fuzzy sets perform as an emerging tool in decision making under uncertainty. this paper reviews the main relevant papers, which are published in international academic journals from 2010 to 2016, relating to the interpretations, modellings, and applications of hesitant fuzzy sets. moreover, some possible future research directions are put forward. keywords: hesitant fuzzy set, decision making, aggregation, measure, clustering 1. introduction to deal with vague or imprecise phenomena in decision making, zadeh (1965) developed the fuzzy sets (z-fss) that have wide applications in various fields (zimmermann, 2001). as a new extension, torra (2010a) proposed the concept of hesitant fuzzy sets (hfss) to enhance the modelling abilities of z-fss. compared with other extensions of z-fss, such as atanassov’s intuitionistic fuzzy sets (atanassov, 1986) (a-ifss), type-2 fuzzy sets (dubois, 1980; zadeh, 1975) (t2fss) and fuzzy multisets (yager, 1986) (fmss), the hfs defines the membership degree of an element associated with several possible values. as introduced by torra (2010a), the difficulty of establishing the membership degree is not because we have a margin of error (as in a-ifss), or some possibility distribution (as in t2fss) on the possible values, but because we have a set of possible values. the typical application of hfss can arise in uncertain decision making, where the decision makers (dms) experience hesitancy in providing their judgments. ever since torra (2010a) introduced hfss, this concept has become a hot topic and received more and more attentions recent years. numerous relevant papers have published in international journals from 2010 to 2016. we now roughly group these papers, and then give a brief review of them. 2. hesitant fuzzy set and several extensions 2.1. hesitant fuzzy set torra (2010a) defined the concept of hfs as follows: definition 1 (torra, 2010a). let x be a fixed set, then a hfs on x is in terms of a function h that returns a subset of [0,1] . hfss have close relationship with the existing fuzzy sets, including z-fss, a-ifss, t2fss and fmss. specifically, hfss are ifss when hfss are nonempty closed intervals, and hfss can be represented as fmss or t2fss. journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 67-75 published by atlantis press copyright: the authors 67 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets torra (2010a) considered that a typical hfs is a finite subset of [0,1] . in particular, the typical hfs ( )h x represents the possible membership values of the set at x . based on this assumption, torra (2010a) developed some basic operations of hfss, such as the complement, the union, the intersection and the envelope of hfss. to further apply hfss to decision making under uncertainty, xia and xu (2011) restated the concept of hfss and defined hesitant fuzzy elements (hfes) which are convenient to be used to present the decision makers’ (dms) preferences in decision making. definition 2 (xia and xu, 2011). let x be a fixed set, a hfs on x is in terms of a function that when applied to x returns a subset of [0,1] , which can be represented as the following mathematical symbol: { , ( ) | }ee x h x x x= < > ∈ (1) where ( )eh x is a set of some values in [0,1] , denoting the possible membership values of the element x x∈ to the set e . for convenience, ( )eh h x= is called a hfe. for three hfes h , 1h and 2h , xia and xu (2011) developed some new operations: 1) { }hh λ λ γ γ∈=  ; 2) {1 (1 ) }hh λ γλ γ∈= − − ; 3) 1 1 2 21 2 , 1 2 1 2 { }h hh h γ γ γ γ γ γ∈ ∈⊕ = + − ; 4) 1 1 2 21 2 , 1 2 { }h hh h γ γ γ γ∈ ∈⊗ =  . then zhu et al. (2012b) developed two relationships for the operations: 1) 1 2 1 2( )h h h hλ λ λ⊕ = ⊕ ; 2) 1 2 1 2( )h h h h λ λ λ⊗ = ⊗ . furthermore, verma and sharma (2013) systematically investigated the properties of the operations of hfes, which is helpful for the development of hfss. to make hfss more important and applicable, verma and sharma (2013) then proposed four new operations of hfss and studied their properties. liao and xu (2014b) recently developed the subtraction and division operations over hfss, which is important in forming the integral theoretical framework of hfss. to compare hfes, xia and xu (2011) proposed a score function defined as follows: definition 3 (xia and xu, 2011). for a hfe h , ( ) 1/ h s h h γ γ ∈ = ∑# is called the score function of h , where h# is the number of the elements in h . for two hfes 1h and 2h , if 1 2( ) ( )s h s h> , then 1 2h h> ; if 1 2( ) ( )s h s h= , then 1 2h h= . based on the assumption that two hfes have the same number of elements and the elements are arranged in an increasing order, farhadinia (2013b) then developed a new score function that meets some properties shown in definition 4. definition 4 (farhadinia, 2013b). let # 1{ } { } h h j jh γ γ γ∈ == = be a hfe, where # h returns the number of values in h . a score function s of a hfe h is defined by # #1 # 1 1 2 ( ) # (# 1) h j hj jh j j j s h j h hj γ γ= = = = = + ∑ ∑ ∑ (2) the score functions are used to aggregate the elements in the hfe. xia and xu (2011)’s score function pays attention to the overall arguments fairly, whereas farhadinia (2013b)’s score function assigns more weights to the arguments whose values are high. for example, for two hfes 1 {0.2, 0.3, 0.7}h = and 2 {0.1, 0.4, 0.7}h = , according to definition 3, we have 1 2( ) ( ) 0.4s h s h= = , so 1 2h h= . however, according to definition 4, we get 1( ) 0.48s h = and 2( ) 0.5s h = , such that 1 2h h< . on the other hand, the ordered weighted averaging (owa) aggregation operators (yager, 1988) can also be applied to aggregate the elements in the hfe. in addition, the score functions are special cases of the owa aggregation operators. with different owa aggregation operators (yager, 1993), hfes can be compared relying on different decision environment. 2.2. dual hesitant fuzzy set zhu et al. (2012a) considered the situation where there are several possible values both for the membership degree and the non-membership degree, then defined the dual hesitant fuzzy set (dhfs) in terms of two functions that return two sets of possible values indicating the membership degree and non-membership degree respectively for each element in the domain. as a more comprehensive fuzzy set, dhfss can encompass published by atlantis press copyright: the authors 68 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets a-fss, a-ifss, hfss, and fmss as special cases with specific conditions. to apply dhfss to decision making under uncertainty, dual hesitant fuzzy elements (dhfes) (zhu et al., 2012a) were developed to present the dms’ preferences. hffes are more comprehensive than hfes due to its consideration of both membership and nonmembership. for example, if 0.1 and 0.2 are considered as the possible values for the membership degree, 0.3, 0.4 and 0.5 are considered as the possible values for the non-membership degree, then a dhfe indicating such a case can be denoted by {{0.1, 0.2},d = {0.3, 0.4, 0.5}} . furthermore, to develop basic operations and aggregation operations of dhfss, zhu et al. (2012a) developed an extension principle based on the ordered modular average (oma) (mesiar and mesiarovázemánková, 2011). however, since dhfss can reduce to some existing fuzzy sets with specific conditions, zhu and xu (2014) further defined the typical dhfs (t-dhf) to distinguish dhfss from a-fss, a-ifss and hfss. they also defined typical dhfes (t-dhfes) so as to apply t-dhfss to decision making. basic operations and their properties of t-dhfes were studied extensively. due to the advantages of dhfss in presenting hesitant fuzzy information, chen et al. (2014) developed the correlation and correlation coefficient of dhfss, then investigated the pattern recognition problems about the classification of metal materials based on dual hesitant fuzzy information. ye (2013) proposed a new correlation coefficient between dhfss as a new extension of the existing correlation coefficients. wang et al. (2015) developed some dual hesitant fuzzy aggregation operators, and then used them to multicriteria decision making problems. 2.3. extended hesitant fuzzy set (ehfs) torra (2010b) gave an example to show the application of hfss in decision making: two dms discuss the membership degree of x into a , one wants to assign 0.5 and the other 0.6 , which can be denoted by a hfe, {0.5, 0.6}h = . however, if the two dms both assign the value 0.5 , we can only save one value in the h and loss the other one, which appears to be an information loss problem of hfss. since the dms often have different importances in decision making, the loss of information provided by the important dms may lead to ineffective results. moreover, the preferences in h can not be distinguished to the two dms, respectively. thus hfes are often used to present preferences provided by the anonymous dms whose weights are the same. clearly, different dms can not be identified in such case. a similar assumption, that is the dms providing their preferences anonymous, is also in xu and xia (2011b) in developing distance measures of hfss. in addtion, dhfss also have this problem. to overcome the information loss problem and identify different dms in decision making, zhu and xu (2016) proposed the concept of extended hesitant fuzzy set (ehfs), where the values provided by the dms are collected by several possible value-groups. for example, if one dm assigns 0.5 , another assigns 0.5 or 0.6 to the membership degree of x into a , then two possible value-groups are (0.5, 0.5) and (0.5, 0.6) , where the first component of the value-group is provided by the first dm, and the second provided the other. in such a case, all the preferences provided by the dms are saved and distinguished clearly. zhu and xu (2016) further developed extended hesitant fuzzy elements (ehfes), and some basic operations of ehfes. so the case mentioned above can be denoted by an ehfe as {(0.5, 0.5), (0.5, 0.6)}h = . obviously, ehfss are defined as the cartesian product of hfss. this definition implies that hfss can be used to construct ehfss. on the contrary, ehfss can also reduce to hfss. as a special case of ehfss, all the operations of hfss are consistent with the operations of ehfss. zhu and xu (2016) developed a concept of reduced ehfes to investigate the relationship between hfss and ehfss. the relationships among ehfss, a-ifss, ehfss and dhfs have been studied. to show the advantages of ehfss in overcoming the drawbacks of hfss, zhu and xu (2016) developed some distance measures of ehfes, and then gave a comparative example in dealing with a multi-criteria decision making problem. these results show that the existing hesitant distance measures can be considered as a particular case of the extended hesitant distance measures. 2.4. interval-valued hesitant fuzzy set and generalized hesitant fuzzy set chen et al. (2013) originally developed the concept of interval-valued hesitant fuzzy set (ivhfs) in which the membership degree of an element to a given set is not published by atlantis press copyright: the authors 69 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets precisely defined, but by several possible interval values. similar to the definition of hfes, the basic elements of ivhfss are called interval-valued hesitant fuzzy elements (ivhfes). for example, suppose two possible interval values, that are [0.3, 0.4] and [0.5, 0.6] , indicating the membership degree of an element, they can be presented by an ivhfe denoted by {[0.3, 0.4],[0.5, 0.6]}h = . chen et al. (2012) further defined some basic operations of ivhfes, developed some interval-valued hesitant fuzzy aggregation operators, proposed intervalvalued hesitant fuzzy preference relations to describe hesitant uncertain information in group decision making, and provided numerical examples to show the applications of ivhfss in the decision making problems. based on ivhfss, bai (2013) developed some distance and similarity measures, wei and zhao (2013) developed some induced interval-valued hesitant fuzzy aggregation operators. with interval-valued hesitant fuzzy information, wei et al. (2014d) investigated multi-criteria decision making problems. moreover, wei (2013) defined a concept called the hesitant interval-valued fuzzy set which is exactly the same as the ivhfs (chen et al., 2013), and then introduced some aggregation operations of hesitant interval-valued fuzzy sets (wei et al., 2014d) which are also the same as the interval-valued hesitant fuzzy aggregation operators developed by chen et al. (2012). recently, chen and xu (2014) derived the properties and relationships of basic operations on ivhfss, chen and xu (2014) developed some new operations based on archimedean t-norm and t-conorm. qian et al. (2013) developed a concept of generalized hesitant fuzzy set (ghfs) which consists of generalized hesitant fuzzy elements (ghfes). for example, a decision group consisting of several dms provides some possible values for the membership degree of an element. some dms provide 0.4, some argue between 0.45 and 0.55, and the others insist on at least 0.6, then the membership degree can be presented by three intuitionistic fuzzy values (xu, 2007), which can be denoted by 1 (0.4, 0.6)α = , 2 (0.45, 0.45)α = and 3 (0.6, 0)α = . so we can get a corresponding ghfe denoted by (0.4, 0.6)gh =  (0.45, 0.45) (0.6, 0) . by the envelope operation of intuitionistic fuzzy values, the three intuitionistic fuzzy values can be transformed into three intervals as 1 [0.4, 0.4]α′ = , 2 [0.45, 0.55]α′ = and 3 [0.6,1]α′ = , respectively. thus, gh can also be represented by an ivhfe, denoted by {[0.4, 0.4],h′ = [0.45, 0.55], [0.6,1]} . furthermore, the basic operations of ghfss defined by qian et al. (2013) are consistent with the basic operations of ivhfss (chen et al., 2012). clearly, ivhfss and ghfss are two equivalent extensions of hfss connected by the envelope operation. the motivation of proposing ghfss and ivhfss is to allow the membership degree represented by a margin of error. on the other hand, according to definition 1, ivhfss and gfhss are two special cases of hfss. all the related operations of ivhfss and ghfss are consistent with the corresponding operations of hfss. 3. measures and clustering 3.1. distance and similarity measures distance and similarity measures are important in decision making. motivated by some famous distance and similarity measures, xu and xia (2011b) originally developed some distance and similarity measures of hfss, such as the hesitant normalized hamming distance, the hesitant normalized euclidean distance, the generalized hesitant normalized distance and the hesitant normalized hausdorff distance. moreover, the computation of distance similarity measures require that the hfes should have the same number of elements. since the dms often provided different number of possible values for the membership degree in most cases, xu and xia (2011b) proposed an optimistic and a pessimistic rules to extend hfes to make them have the same length. according to the generalized hesitant fuzzy weighted distance measure and the generalized hesitant fuzzy ordered weighted distance measure developed by xu and xia (2011b), peng et al. (2013) further developed a generalized hesitant fuzzy synergetic weighted distance measure, and then investigated some properties and special cases. recently, zhang and xu (2015b) presented a signed distance-based method to compare hfes by considering a proposed hesitancy index. published by atlantis press copyright: the authors 70 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets 3.2. correlation, entropy and cross-entropy measures correlation can reflect a linear relationship between two variables it is an important measure in data analysis, medical diagnosis, pattern recognition, and particularly decision making. based on an assumption that the hfes have the same number of elements, and the values in each hfe are arranged in an increasing order, xu and xia (2011a) defined some correlation measures on hfes, chen et al. (2013) defined some correlation coefficient formulas for hhfs. xu and xia (2012) defined the entropy and crossentropy measures for hesitant fuzzy information, and then discussed their properties. farhadinia (2013a) investigated the relationship between the entropy, the similarity measure and the distance measure of hfss. farhadinia (2013a) introduced some new formulas for the entropy and the similarity measure of hfss to provide a transformation from entropy measures to similarity measures. 3.3. clustering under hesitant fuzzy environment, zhang and xu (2015c) proposed a clustering algorithm that has advantages in computation based on boole matrices and similarity measures of hfss. zhang and xu (2015a) proposed a hesitant fuzzy agglomerative hierarchical clustering algorithm for hfss, and gave some comparisons between this algorithm and the intuitionistic fuzzy hierarchical clustering algorithm proposed by xu (2009). chen et al. (2013) developed some correlation coefficient formulas, and then apply them to clustering analysis. farhadinia (2013a) developed two clustering algorithms in which new indices of similarity measures for hfss are applied in data analysis and classification. 4. aggregation operators in decision making, the aggregation operator is the most used technique to obtain the overall performance of alternatives by aggregating individual preference information. under intuitionistic fuzzy environment, xu (2007); xu and yager (2006) developed some intuitionistic fuzzy aggregation operators. motivated by these operators, xia and xu (2011) developed a series of aggregation operators under hesitant fuzzy environment. based on quasi-arithmetic means (hardy et al., 1934), xia et al. (2013) further defined some quasi-hesitant fuzzy aggregation operators which can be considered as the generations of their previous work (xia and xu, 2011). based on these proposed aggregation operators, liao and xu (2014a) defined some hesitant fuzzy hybrid weighted aggregation operators, and then applied them to the multi-criteria decision making problems. to capture the interrelationship of hesitant fuzzy information, zhu and xu (2013b) developed hesitant fuzzy bonferroni means, and zhu et al. (2012b) developed hesitant fuzzy geometric bonferroni means. in aggregating hesitant fuzzy information, hesitant fuzzy bonferroni means can consider the whole arguments fairly, while hesitant fuzzy geometric bonferroni means pay attention to some arguments whose values are too high or too low. motivated by hesitant fuzzy bonferroni means, yu et al. (2012) further defined generalized hesitant fuzzy bonferroni means and applied them to decision making. besides bonferroni means (bonferroni, 1950), power average operators (yager, 2001) can also reflect the interrelationship of arguments. under hesitant fuzzy environment, zhang (2013) defined a wide range of hesitant fuzzy power average aggregation operators, liu and sun (2013) developed some generalized hesitant fuzzy power average aggregation operators. considering that the criteria often have different priority levels in multi-criteria decision making, wei (2012); yu (2012) developed some hesitant fuzzy prioritized operators. moreover, wei et al. (2014c) developed a series of hesitant triangular fuzzy aggregation operators. 5. hesitant fuzzy set under linguistic environment the advantages of hfss of handling the imprecise whereby two or more sources of vagueness appearing simultaneously can also be taken into account under the linguistic environment. to improve the modelling and computational abilities of linguistic approaches, rodriguez et al. (2012) developed the hesitant fuzzy linguistic term set (hflts) which enables the dms hesitate about several possible values to assess a linguistic variable. a context-free grammar is developed to generate rich linguistic expressions that can be represented by means of hflts. basic operations of hfltss and their properties are investigated in detail. published by atlantis press copyright: the authors 71 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets particularly, they defined the envelope operation of hfltss, which is a linguistic interval whose limits are obtained by means of upper bound and lower bound. actually, the envelop of hfltss is the uncertain linguistic variable that was defined by xu (2004) in 2004. to show the advantages of hfltss in decision making, they also presented a multi-criteria linguistic decision making model, in which the dms provide their assessments by using linguistic expressions based on comparative terms. later, to overcome the limitations of the existing linguistic models that use single-valued and predefined terms, rodríguez et al. (2013) further developed a new linguistic group decision model which is capable of dealing with comparative linguistic expressions based on context-free grammars and hfltss. the linguistic group decision model facilitates the elicitation of linguistic information. to further present linguistic information by hfltss, zhu and xu (2013a) developed the hesitant fuzzy linguistic preference relation (hflpr) as a basic decision making tool under hesitant fuzzy linguistic environment. in order to derive meaningful results in decision making by hflprs, zhu and xu (2013a) developed the consistency measures of hflprs to ensure that the dms are neither being random nor illogical for their preference information, and then established the consistency thresholds of hflprs to measure weather a hflpr is of acceptable consistency. with respect to the inconsistent hflprs, two optimization methods were also proposed by zhu and xu (2013a) to improve their consistency until they are acceptable. moreover, to compare hfltss better, wei et al. (2014a) gave two possibility degree formulas, and lee and chen (2013) developed the concept of likelihoodbased comparison relations. to aggregate hesitant fuzzy linguistic information, wei et al. (2014a); zhang and wu (2014) proposed some hesitant fuzzy linguistic aggregation operators. for the applications of hfltss in decision making, lin et al. (2014) developed some models for selecting an erp system. 6. applications of hfss to apply hfss to decision making under uncertainty, zhu and xu (2013c) introduced the concept of hesitant fuzzy preference relation (hfpr) to present hesitant fuzzy information, investigated the relationship between hfprs and fuzzy preference relations (tanino, 1984), and developed two regression methods to transform hfprs into fuzzy preference relations. to extend the traditional multiplicative preference relations (saaty, 1980) (known as comparison matrixes in analytic hierarchy process), xia and xu (2013) developed the hesitant multiplicative preference relation (hmpr), and then investigated the group decision making problems. under hesitant fuzzy environment, some new decision making methods have been developed. liao and xu (2013) proposed the hesitant fuzzy-vikor method based on several developed measures, such as the hesitant normalized manhattan -metric,l the hesitant fuzzy group utility measure, the hesitant fuzzy individual regret measure and the hesitant fuzzy compromise measure. ma et al. (2014) developed a hesitant fuzzy superiority and inferiority ranking method, and then applied it to deal with the multicriteria group decision making problems. based on the presentation of the dms’ preferences in hesitant fuzzy information, zhang and xu (2014) proposed an interval programming method; zhang and wei (2013) extended the vikor method and the topsis method to solve the group decision making problems; xu and zhang (2013) developed a maximizing deviation method; wei et al. (2014b) investigated the multi-criteria decision making problems with incomplete weight information; zhang and wei (2014) extended the vikor method to deal with a correlative multi-criteria decision making problem; feng et al. (2014) investigated the multi-criteria decision making problems for social network analysis by the topsis method. xu (2014) systematically summarized the theory of hfss, and showed the applications of hfss in practice. recently, zhu et al. (2015) developed generalized analytical network process (g-anp) that takes hesitant fuzzy preferences into account. g-anp eliminates the drawbacks of the fuzzy analytical network process (fanp) (mikhailov and singh madan, 2003) that cannot take consistency into account, and allows multiple forms of preferences, such as crisp (fuzzy) preferences, interval (interval fuzzy) preferences, hesitant (hesitant fuzzy) preferences and stochastic (stochastic fuzzy) preferences. for application, zhu et al. (2015) applied the g-anp to a real risk management problem. moreover, with the idea of hesitancy in decision making, zhu et al. (2016) developed the hesitant published by atlantis press copyright: the authors 72 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets analytical hierarchy process and then applied it to evaluate the strategic positions of islands and reefs. 7. conclusions in this paper, we have reviewed the developments and applications of hesitant fuzzy sets (hfss). hfss have the advantage in representing the decision makers’ (dms) preferences characterized by the hesitance whereby two or more possible values can be considered simultaneously for the membership. we have introduced dual hesitant fuzzy sets (dhfss), extended hesitant fuzzy sets (ehfss), interval-valued hesitant fuzzy sets (ivhfss) and generalized hesitant fuzzy sets (ghfss), and show their relationship with hfss. furthermore, we have divided the relevant papers into four groups: the studies on measures and clustering, aggregation operators, hfss under the linguistic environment, and decision making methods. the hfs is an emerging uncertain decision making tool with numerous studies in recent years. however, how to use hfss to deal with real-life decision maker problems such as risk decision making, emergency management and big data-driven decision making, and how to apply hfss to other fields, such as information retrieval and medical diagnosis, may be the highlights for further research. references krassimir atanassov. intuitionistic fuzzy sets. fuzzy sets and systems, 20(1) (1986). 87-96. zhiyong bai. distance similarity measures for interval-valued hesitant fuzzy sets and their application in multicriteria decision making. journal of decision systems, 22(3) (2013). 190-201. carlo bonferroni. sulle medie multiple di potenze. bollettino dell'unione matematica italiana, 5(3-4) (1950). 267-270. na chen, and z. s. xu. properties of interval-valued hesitant fuzzy sets. journal of intelligent and fuzzy systems, 27(1) (2014). 143-158. na chen, z. s. xu, and m. m. xia. interval-valued hesitant preference relations and their applications to group decision making. knowledge-based systems, 37(2012). 528-540. na chen, z. s. xu, and m. m. xia. correlation coefficients of hesitant fuzzy sets and their applications to clustering analysis. applied mathematical modelling, 37(4) (2013). 2197-2211. yuanfang chen, xiaodong peng, guohua guan, and huade jiang. approaches to multiple attribute decision making based on the correlation coefficient with dual hesitant fuzzy information. journal of intelligent and fuzzy systems, 26(5) (2014). 2547-2556. didier j dubois. (1980). fuzzy sets and systems: theory and applications (vol. 144): academic press. bahram farhadinia. information measures for hesitant fuzzy sets and interval-valued hesitant fuzzy sets. information sciences, 240(2013a). 129-144. bahram farhadinia. a novel method of ranking hesitant fuzzy values for multiple attribute decision-making problems. international journal of intelligent systems, 28(2013b). 752-767. xie feng, wanli zuo, junhua wang, and lizhou feng. topsis method for hesitant fuzzy multiple attribute decision making. journal of intelligent and fuzzy systems, 26(5) (2014). 2263-2269. gh hardy, je littlewood, and g pólya, inequalities (1934): cambridge university press, cambridge. liwei lee, and shyiming chen. (2013). fuzzy decision making based on hesitant fuzzy linguistic term sets intelligent information and database systems (pp. 21-30): springer. huchang liao, and zeshui xu. a vikor-based method for hesitant fuzzy multi-criteria decision making. fuzzy optimization and decision making(2013). 1-20. huchang liao, and zeshui xu. some new hybrid weighted aggregation operators under hesitant fuzzy multi-criteria decision making environment. journal of intelligent and fuzzy systems, 26(4) (2014a). 1601-1617. huchang liao, and zeshui xu. subtraction and division operations over hesitant fuzzy sets. journal of intelligent and fuzzy systems, 27(2014b). 65-72. rui lin, xiaofei zhao, and gangxi wei. models for selecting an erp system with hesitant fuzzy linguistic information. journal of intelligent and fuzzy systems, 26(5) (2014). 2155-2165. jing liu, and min sun. generalized power average operator of hesitant fuzzy numbers and its application in multiple attribute decision making. journal of computational information systems, 9(8) (2013). 30513058. zujun ma, nian zhang, and ying dai. a novel sir method for multiple attributes group decision making problem under hesitant fuzzy environment. journal of intelligent and fuzzy systems, 26(5) (2014). 2119-2130. radko mesiar, and andrea mesiarová-zemánková. the ordered modular averages. fuzzy systems, ieee transactions on, 19(1) (2011). 42-50. ludmil mikhailov, and g singh madan. fuzzy analytic network process and its application to the development of decision support systems. systems, man, and cybernetics, part c: applications and reviews, ieee transactions on, 33(1) (2003). 33-41. dinghong peng, chang-yuan gao, and zhi-fang gao. generalized hesitant fuzzy synergetic weighted distance measures and their application to multiple criteria decision-making. applied mathematical modelling, 37(2013). 5837-5850. gang qian, hai wang, and xiangqian feng. generalized of hesitant fuzzy sets and their application in decision support system. knowledge-based systems, 37(2013). 357-365. rosa m rodríguez, luis martı́nez, and francisco herrera. a group decision making model dealing with comparative published by atlantis press copyright: the authors 73 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets linguistic expressions based on hesitant fuzzy linguistic term sets. information sciences, 241(2013). 28-42. rosa m rodriguez, luis martinez, and francisco herrera. hesitant fuzzy linguistic term sets for decision making. fuzzy systems, ieee transactions on, 20(1) (2012). 109119. thoms saaty. (1980). the analytic hierarchy process: mcgraw-hill, new york. tetsuzo tanino. fuzzy preference orderings in group decision making. fuzzy sets and systems, 12(2) (1984). 117-131. vicenç torra. hesitant fuzzy sets. international journal of intelligent systems, 25(6) (2010a). 529-539. vicenç torra. hesitant fuzzy sets. international journal of intelligent systems, 25(1) (2010b). 529-539. rajkumar verma, and bhu dev sharma. new operations over hesitant fuzzy sets. fuzzy information and engineering, 2(2013). 1-18. hongjun wang, xiaofei zhao, and gangwen wei. dual hesitant fuzzy aggregation operators in multiple attribute decision making. journal of intelligent and fuzzy systems, 26(5) (2015). 2281-2290. cuiping wei, na zhao, and xijin tang. operators and comparisons of hesitant fuzzy linguistic term sets. fuzzy systems, ieee transactions on, 22(3) (2014a). 575585. guiwu wei. hesitant fuzzy prioritized operators and their application to multiple attribute decision making. knowledge-based systems, 31(2012). 176-182. guiwu wei. some hesitant interval-valued fuzzy aggregation operators and their applications to multiple attribute decision making. knowledge-based systems, 46(2013). 43-53. guiwu wei, hongjun wang, xiaofei zhao, and rui lin. approaches to hesitant fuzzy multiple attribute decision making with incomplete weight information. journal of intelligent and fuzzy systems, 26(1) (2014b). 259-266. guiwu wei, hongjun wang, xiaofei zhao, and rui lin. hesitant triangular fuzzy information aggregation in multiple attribute decision making. journal of intelligent and fuzzy systems, 41(4) (2014c). 1086-1094. guiwu wei, and xiaofei zhao. induced hesitant intervalvalued fuzzy einstein aggregation operators and their application to multiple attribute decision making. journal of intelligent and fuzzy systems, 24(4) (2013). 789-803. guiwu wei, xiaofei zhao, rui lin, and hongjun wang. models for hesitant interval-valued fuzzy multiple attribute decision making based on the correlation coefficient with incomplete weight information. journal of intelligent and fuzzy systems, 26(4) (2014d). 1631-1644. meimei xia, and zeshui xu. hesitant fuzzy information aggregation in decision making. international journal of approximate reasoning, 52(3) (2011). 395-407. meimei xia, and zeshui xu. managing hesitant information in gdm problems under fuzzy and multiplicative preference relations. international journal of uncertainty, fuzziness and knowledge-based systems, 21(06) (2013). 865-897. meimei xia, zeshui xu, and na chen. some hesitant fuzzy aggregation operators with their application in group decision making. group decision and negotiation, 22(2) (2013). 259-279. zeshui xu, and meimei xia. on distance and correlation measures of hesitant fuzzy information. international journal of intelligent systems, 26(5) (2011a). 410-425. zeshui xu. uncertain linguistic aggregation operators based approach to multiple attribute group decision making under uncertain linguistic environment. information sciences, 168(1) (2004). 171-184. zeshui xu. intuitionistic fuzzy aggregation operators. fuzzy systems, ieee transactions on, 15(6) (2007). 1179-1187. zeshui xu. intuitionistic fuzzy hierarchical clustering algorithms. journal of systems engineering and electronics, 20(1) (2009). 90-97. zeshui xu. (2014). hesitant fuzzy sets theory (vol. 314): springer. zeshui xu, and meimei xia. distance and similarity measures for hesitant fuzzy sets. information sciences, 181(11) (2011b). 2128-2138. zeshui xu, and meimei xia. hesitant fuzzy entropy and crossentropy and their use in multiattribute decision-making. international journal of intelligent systems, 27(9) (2012). 799-822. zeshui xu, and ronald r yager. some geometric aggregation operators based on intuitionistic fuzzy sets. international journal of general systems, 35(4) (2006). 417-433. zeshui xu, and xiaolu zhang. hesitant fuzzy multi-attribute decision making based on topsis with incomplete weight information. knowledge-based systems, 52(2013). 53-64. ronald r yager. on the theory of bags. international journal of general system, 13(1) (1986). 23-37. ronald r yager. on ordered weighted averaging aggregation operators in multicriteria decisionmaking. systems, man and cybernetics, ieee transactions on, 18(1) (1988). 183-190. ronald r yager. families of owa operators. fuzzy sets and systems, 59(2) (1993). 125-148. ronald r yager. the power average operator. systems, man and cybernetics, part a: systems and humans, ieee transactions on, 31(6) (2001). 724-731. jun ye. correlation coefficient of dual hesitant fuzzy sets and its application to multiple attribute decision making. applied mathematical modelling, 38(2) (2013). 659-666. dejian yu. hesitant fuzzy prioritized operators and their application in multi-criteria group decision making. african journal of business management, 6(33) (2012). 9370-9380. dejian yu, yingyu wu, and wei zhou. generalized hesitant fuzzy bonferroni mean and its application in multi-criteria group decision making. journal of information & computational science, 9(2) (2012). 267-274. lotfi a. zadeh. fuzzy sets. information and control 8(1965). 338-353. lotfi a. zadeh. the concept of a linguistic variable and its application to approximate reasoning—i. information sciences, 8(3) (1975). 199-249. nian zhang, and guiwu wei. extension of vikor method for decision making problem based on hesitant fuzzy set. applied mathematical modelling, 37(2013). 4938-4947. nian zhang, and guiwu wei. a multiple criteria hesitant fuzzy decision making with shapley value-based vikor method. journal of intelligent and fuzzy systems, 26(2) (2014). 1065-1075. published by atlantis press copyright: the authors 74 b. zhu, z.s. xu / overview on the developments and applications of hesitant fuzzy sets xiaolu zhang, and z.s. xu. hesitant fuzzy agglomerative hierarchical clustering algorithms. international journal of systems science, 46(3) (2015a). 562-576. xiaolu zhang, and zeshui xu. interval programming method for hesitant fuzzy multi-attribute group decision making with incomplete preference over alternatives. computers & industrial engineering, 75(2014). 217-229. xiaolu zhang, and zeshui xu. hesitant fuzzy qualiflex approach with a signed distance-based comparison method for multiple criteria decision analysis. expert systems with applications, 42(2) (2015b). 873-884. xiaolu zhang, and zeshui xu. novel distance and similarity measures on hesitant fuzzy sets with applications to clustering analysis. journal of intelligent & fuzzy systems, 28(5) (2015c). 2279-2296. zhiming zhang. hesitant fuzzy power aggregation operators and their application to multiple attribute group decision making. information sciences, 234(2013). 150-181. zhiming zhang, and chong wu. hesitant fuzzy linguistic aggregation operators and their applications to multiple attribute group decision making. journal of intelligent and fuzzy systems, 26(5) (2014). 2185-2202. bin zhu, and z. s. xu. consistency measures for hesitant fuzzy linguistic preference relations. fuzzy systems, ieee transactions on, 22(1) (2013a). 35-45. bin zhu, and z. s. xu. hesitant fuzzy bonferroni means for multi-criteria decision making. journal of the operational research society, 64(12) (2013b). 1831-1840. bin zhu, and zeshui xu. regression methods for hesitant fuzzy preference relations. technological and economic development of economy, 19(sup1) (2013c). s214-s227. bin zhu, and zeshui xu. some results for dual hesitant fuzzy sets. journal of intelligent and fuzzy systems, 26(2014). 1657-1668. bin zhu, and zeshui xu. extended hesitant fuzzy sets. technological and economic development of economy, 22(1) (2016). 100-121. bin zhu, zeshui xu, and meimei xia. dual hesitant fuzzy sets. journal of applied mathematics, 2012(2012a). 1-13. bin zhu, zeshui xu, and meimei xia. hesitant fuzzy geometric bonferroni means. information sciences, 205(2012b). 72-85. bin zhu, zeshui xu, ren zhang, and mei hong. generalized analytic network process. european journal of operational research, 244(1) (2015). 277-288. bin zhu, zeshui xu, ren zhang, and mei hong. hesitant analytic hierarchy process. european journal of operational research, 250(2) (2016). 602-614. hans jürgen zimmermann. (2001). fuzzy set theory-and its applications: springer. published by atlantis press copyright: the authors 75 1. introduction 2. hesitant fuzzy set and several extensions 2.1. hesitant fuzzy set 2.2. dual hesitant fuzzy set 2.3. extended hesitant fuzzy set (ehfs) 2.4. interval-valued hesitant fuzzy set and generalized hesitant fuzzy set 3. measures and clustering 3.1. distance and similarity measures 3.2. correlation, entropy and cross-entropy measures 3.3. clustering 4. aggregation operators 5. hesitant fuzzy set under linguistic environment 6. applications of hfss 7. conclusions references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <feff005500740069006c0069007a00610163006900200061006300650073007400650020007300650074010300720069002000700065006e007400720075002000610020006300720065006100200064006f00630075006d0065006e00740065002000410064006f006200650020005000440046002000610064006500630076006100740065002000700065006e0074007200750020007400690070010300720069007200650061002000700072006500700072006500730073002000640065002000630061006c006900740061007400650020007300750070006500720069006f006100720103002e002000200044006f00630075006d0065006e00740065006c00650020005000440046002000630072006500610074006500200070006f00740020006600690020006400650073006300680069007300650020006300750020004100630072006f006200610074002c002000410064006f00620065002000520065006100640065007200200035002e00300020015f00690020007600650072007300690075006e0069006c006500200075006c0074006500720069006f006100720065002e> /rus <feff04180441043f043e043b044c04370443043904420435002004340430043d043d044b04350020043d0430044104420440043e0439043a043800200434043b044f00200441043e043704340430043d0438044f00200434043e043a0443043c0435043d0442043e0432002000410064006f006200650020005000440046002c0020043c0430043a04410438043c0430043b044c043d043e0020043f043e04340445043e0434044f04490438044500200434043b044f00200432044b0441043e043a043e043a0430044704350441044204320435043d043d043e0433043e00200434043e043f0435044704300442043d043e0433043e00200432044b0432043e04340430002e002000200421043e043704340430043d043d044b04350020005000440046002d0434043e043a0443043c0435043d0442044b0020043c043e0436043d043e0020043e0442043a0440044b043204300442044c002004410020043f043e043c043e0449044c044e0020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200431043e043b043504350020043f043e04370434043d043804450020043204350440044104380439002e> /sky <feff0054006900650074006f0020006e006100730074006100760065006e0069006100200070006f0075017e0069007400650020006e00610020007600790074007600e100720061006e0069006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b0074006f007200e90020007300610020006e0061006a006c0065007001610069006500200068006f0064006900610020006e00610020006b00760061006c00690074006e00fa00200074006c0061010d00200061002000700072006500700072006500730073002e00200056007900740076006f00720065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f00740076006f00720069016500200076002000700072006f006700720061006d006f006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076016100ed00630068002e> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice received 8 january 2016 accepted 27 february 2016 research on the performance evaluation of government venture capital fund based on factor analysis and dea model hong-mei zhang1,2,3, ya-zhong chen1,2,3,zhe wang1,2,3 1guizhou university of finance and economics, guizhou institution for technology innovation & entrepreneurship investment, guiyang guizhou 550025, china 2guizhou university of finance and economics, guizhou institute for urban economics and development, guiyang guizhou 550025,china 3guizhou university of finance and economics, school of finance, guiyang guizhou 550025, china abstract according to the existing domestic and foreign research results as well as the characteristics of the guide fund, this paper constructs the index system of guide fund performance evaluation, selects the data of the part of the enterprises shared by the government guide venture capital fund and uses the factor analysis method to optimize the evaluation model, then uses empirical analysis based on dea-bcc model. the result shows that the optimized evaluation model can fully evaluate the performance of the guide fund. at the same time, empirical analysis based on dea-bcc model shows that the results can objectively reflect the operating conditions of companies accepted investment, which provides the objective reference for government venture capital fund and companies accepted investment. key words: government venture capital fund, performance evaluation index system, factor analysis method, dea method 1. introduction at present, with the continuous improvement of financial system and technology in china's financial industry, the transaction cost and opportunity cost of financial products in the field of financial market are gradually reducing, and the financial industry in china is entering a rapid development and constantly growing. among the many financial products in the financial market, all kinds of fund products with their high efficiency fast pace of the characteristics, as one of the important pillars of the financial market, have been the rapid development of universal. among them, the government venture capital fund as one of the policy funds, has played a positive role in the early stage of small and medium enterprises, and the government's macroeconomic regulation and control to focus on the key development areas plays a role in the regulation of financial markets and stimulate market activity. in “guidance on the establishment and operation of the guide fund for venture capital”[1], though the proposed "will guide the fund into the public finance evaluation system", the integrity is insufficient, in terms of chinese current situation and financial market system, in that the development of our country public finance expenditure performance evaluation system is late. so in the evaluation of guiding fund performance, building a reasonable and effective performance evaluation system of venture investment guide fund can help the government guide fund of venture capital in our country to develop faster. at present, the research on the performance evaluation of the guidance fund in foreign countries has made obvious achievements. since many foreign guide funds have been quite successful, the foreign scholars are more likely to carry out performance evaluation in the relevant research. lerner (1999)[2] took the small business investment corporation of the united states as an example, and studied its ability to absorb social capital performance evaluation, in order to test the effectiveness of the actual application of sbic program. "compared to similar corporations, supported companies have been more rapid growth in the past ten years, and more likely to attract venture capital". lerner's view was that the sbic program provides an effective help for the venture capital fund to absorb the introduction of venture capital fund in the government's support. however, once government as a venture capital investment body, its ability is often not enough to assume the role of selection, evaluation of investment opportunities. david b.audretsch et al. (2002)[3] used qualitative research methods and comprehensive innovation, commercialization and entrepreneurial behavior and other aspects to study the journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 15-20 published by atlantis press copyright: the authors 15 hong-mei zhang et al. / performance evaluation of government venture capital fund performance evaluation of sbic. the results show that sbic has a promoting effect on social production technology innovation and scientific and technological achievements. jonathan g.s.koppel (2008)[4] also assessed the sbic, which was evaluated from the point of view of government management. it is believed that the relationship between the supervision and management of venture capital fund should not be presented as subordinate management relationship, which was due to the united states international development council (aid) poor supervision. so we put forward suggestions: we should pay attention to the design of government guidance fund management framework. most of the studies in foreign countries are based on the analysis of the data samples of a specific fund to draw the overall performance. the selected variables can not be integrated into the system, and in the construction of evaluation index system, the guiding fund is not located in the public financial evaluation index system. what’s more, most of the studies have neglected the importance of theoretical research. during the ten years since the birth of china's guide fund, most scholars have gradually started to study the performance evaluation of the fund, and mostly focused on the construction of evaluation index system. guiying liu (2008)[5] comprehensively used factor analysis method to classify and select indicators to construct evaluation index system, and used the dea model to evaluate the performance of china's open funds. she improved the evaluation index system with innovative research methods, and put forward the factor analysis method, which can make the dea model analysis results more clearly and effectively. hongjiang li and xiaoyan bao (2011)[6] proposed that, in the analysis of the venture investment guide fund performance evaluation index system, index system should include five indicators, which reflect policy leverage, industry oriented and risk control, the government support and fund increment, thus to comprehensively and systematically reflect the focus of government policy. bo gui (2009)[7] used factor analysis and dea model to evaluate the competitiveness of the 16 listed banks, and the results showed that the indexes of the competitiveness of listed banks in china were the innovation, scale and profitability of the bank. yu yu (2013)[8] viewed the policy objectives, policy effects and operational capacity as the main factors to analyze the impact of the government venture capital to guide the performance of the fund, and proposed improving suggestions for the problems in the process of assessment, so as to pave the way for building evaluation index system. zhipeng qin (2014)[9] constructed the performance evaluation index system, which accords with the current situation in china, with the aid of economic history analysis method, mathematical economics method and econometric model method is based on the characteristics of china's venture capital fund. according to the conclusions, there are problems such as the contradiction between the regional distribution of the fund and the adjustment of regional economy in china. research on performance evaluation of venture capital fund in china is still in the exploratory stage. at present, it has not been able to meet the regulatory requirements of the index system and theoretical methods in the "guidance". with the theoretical research gradually deepening, the evaluation index system established by the domestic scholars tends to combine policy leverage and industry oriented and other indicators to put forward the innovative evaluation method and index system. this paper, based on the previous research achievements, will try to use factor analysis and dea method to study the performance evaluation of venture investment guide fund. the results show that using this method can be more effectively evaluate the performance of the guide fund, and provide objective evaluation conclusion for the venture investment guide fund and the investment enterprise development, which has a certain reference value. 2. evaluation index and sample data for the analysis of the venture investment guide fund performance, this paper synthesizes the domestic and foreign existing research results and the guide fund characteristics, and analyzes guide fund policy goals, economic benefits and risks of three factors specifically. in accordance with the scientific, reliable and objective principles and considering the efficiency of fund and risk etc. factors, the selected indicators are as follows: the fund investment scale, leverage effect, business profits, income growth rate, investment risk and profit growth rate. on the policy objectives of the investment guide fund, the purpose of the government to guide the establishment of the fund is to attract social capital by way of increasing treasury investment to support industries. on the one hand, it save treasury expenses, on one hand, play a role in amplifying the leverage effect. investment scale is the asset values of the investment guide fund, which reflects the situation of enterprises receiving guidance fund investment at the beginning of the period. leverage effect index shows the multiples of the guide fund drive social capital, which can measure the policy effect and risk. these two indicators are inputs of the venture investment guide fund, which reflects the government's policy guidance. also, the guide fund acts as a financial product, which plays the function of financial funds to control the market through investment funds to enterprises to produce economic benefit, and the fund needs to keep its value for the long-term operation. hedging means is that the investment benefits through the maintenance of capital circulation, so the economic published by atlantis press copyright: the authors 16 hong-mei zhang et al. / performance evaluation of government venture capital fund benefit of the fund is one of the factors to assess the performance of the fund. the operating profit and total profit indicators reflect the profit level of the enterprises receiving the guide fund investment at the end of the term, and the profit growth rate is the total profit of the enterprises at the end of the period, reflecting the business situation. venture investment guide fund as a kind of venture capital, of which the operation risk is an important evaluation index. this paper selects β value as risk metrics, and at the same time, enterprise profit growth rate reflects the business situation to accept the investment and the management ability, which is the risk assessment of the selection of the guide fund investment. this paper selects 10 companies to analyze the data in 2013. the investment scale and the leverage effect data are from the ministry of science and technology of the people's republic of china, the company's operating profit, total profit and profit growth rate data from its financial statements, investment risk data are from the enterprise beta. on the basis of the above six indicators, factor analysis method is used to screen, identify the representative factors and use the dea model for empirical analysis to conduct empirical research on the performance of the guide fund. the indicator system and data are shown in table 1, using the serial number for each enterprise data number from 1-10: table 1. government venture capital fund performance evaluation index and data fund order guiding fund shares enterprise investment scale leverage operate profit investment risk total profit profit growth rate 1 datang telecom polytron technologies co., ltd 56200270.2 5 155615858.74 1.03 265251721.2 25% 2 shanghai kangda chemical new materials co., ltd 6466791.5 5 42781992.71 1.02 46653568.82 22.73% 3 hubei dinglong chemical co., ltd 48899604.2 4 88726258.07 0.72 102519455.2 27.31% 4 luoyang long hua heat transfer polytron tec. co., ltd 30109634.1 4 67872107.25 0.99 94334990 15.01% 5 jiangsu wehrle environmental protection polytron technologies co., ltd 18045531.2 4 23646471.65 1.2 31754245.25 -60.15% 6 qingdao eastsoft polytron technologies co., ltd 15572302.4 4.23 209490271.66 1.25 273553112.9 -12.93% 7 changzhou almaden co., ltd 305099598.5 18.5 70125014.23 0.90 71050901.91 -18.35% 8 jiangsu xiupiang glasswork co., ltd 31298193.5 18.5 36830152.96 1.14 36717814.34 -38.56% 9 yangzhou yang jie electronic polytron technologies co., ltd 19303706.9 18.5 104710351.57 0.29 120723441.7 38.13% 10 jiangsu yawei machine tool co., ltd 37612196.6 18.5 81786430.68 1.28 94209085.07 3.41% 3. construction of index system based on factor analysis factor analysis is the use of statistical methods. under the premise of maintaining the stability of the index system, representative factors are selected among numerous factors, and according to the index data, the existing index is represented as a linear combination of the representative indexes, so as to reduce the dimension of the existing index system. according to the analysis results, if the cumulative variance contribution rate of the common factor is greater than 85%, it is indicated that the common factor can effectively guarantee the index system without losing information after reducing the dimension[10]. the treatment process is: first, the initial factor model is established, or: ( )mnfafafax ininii1i ≤++++= ε...221 (1) among them, there are total m indicators, 𝐹𝑗as the common factor, 𝑎𝑗𝑗 as the factor component which represents the correlation between 𝑋𝑗 and the various factors, 𝜀𝑗 as the error. in order to explain the meaning of the common factor, the factor load is needed to rotate: ( )mnfbfbfbx ininiii ≤++++= ε...2211 (2) among them, 𝑏𝑗𝑗 represents the factor component after the rotation. finally, each common factor is calculated to determine its score, and the factor scores are calculated, respectively: published by atlantis press copyright: the authors 17 hong-mei zhang et al. / performance evaluation of government venture capital fund factor score: mimiii xxxf βββ +++= ...2211 (3) comprehensive score: mimiii fffc λλλ +++= ...2211 (4) 𝐶𝑖 as the comprehensive score， 𝜆𝑖 as the factor score. before the factor analysis, the evaluation index system and data need kmo and bartlett ball type test, in order to test whether it is suitable for factor analysis. after spss analysis, the results as shown in table 2: table 2. kmo and bartlett test kaiser-meyer-olkin metric 0.502 bartlett ball type test approximate chi-square 25.226 df 15 sig. 0.047 visibly, kmo value is 0.502 > 0.5, and the p value of the spherical test, that is, sig value is 0.047 < 0.05, says that the ball hypothesis is rejected, that is, the value of each index in the index system is correlated, which is consistent with the factor analysis of the data requirements. therefore, the evaluation index system constructed in this paper is suitable for factor analysis. next, process the factor analysis of the 6 indicators data through spss 19.0, analysis results as shown in table 3: table 3. variance analysis of common factor factor eigenvalue variance ratio (%) cumulative contribution rate (%) 1 2.353 39.217 39.217 2 1.615 26.925 66.142 3 1.178 19.631 85.775 according to the results of factor analysis, the common factor 1, 2, 3 can better reflect the index information, the eigenvalue is above 1, and the cumulative contribution rate of the three common factors is more than 85%, that is, the common factors meet the requirements. and then according to the analysis of the gravel map, figure 1 can be shown: visibly, since the component 4 begins to show the slope change, the linear slope stabilizes after the component 4. in the gravel map, the straight line slope turns at the point of the component 4, so this paper determines the number of common factor is 3. then, the property of each common factor is determined by the analysis of the rotational component matrix, as shown in table 4: fig. 1 factor analysis of the gravel map table 4. rotational components table index common factor common factor 1 common factor 2 common factor 3 investment scale 0.042 -0.088 0.875 leverage -0.247 0.143 0.779 operate profit 0.978 0.096 -0.056 investment risk 0.095 -0.910 -0.191 total profit 0.979 0.057 -0.139 profit growth rate 0.293 0.863 -0.169 from the rotation table , common factor 1 has a higher degree of interpretation of the operating profit and total profit indicators, so its definition can be defined as the output factor. common factor 2 has a higher degree of interpretation of the investment risk and profit growth rate indicators, and the profit growth rate is the profit growth of the enterprise during the period of investment, reflecting the profitability of the enterprise and the management ability of the enterprise managers, and its value will directly affect the performance of venture capital fund; common factor 3 has a higher degree of interpretation of the investment scale and leverage indicators, so its definition can be defined as the input factor. then calculate the common factor score coefficient, the results can be obtained as shown in table 5: published by atlantis press copyright: the authors 18 hong-mei zhang et al. / performance evaluation of government venture capital fund table 5. common factor score coefficient fund order output factor risk factor input factor 1 1.50553 0.14568 -0.27165 2 -0.80061 0.36707 -0.94259 3 -0.10727 0.91508 -0.52582 4 -0.28986 0.21979 -0.73344 5 -1.0808 -1.36655 -0.70833 6 1.97371 -0.90715 -0.43957 7 -0.02193 -0.31143 2.46606 8 -0.95242 -0.76322 0.39084 9 -0.10972 2.10381 0.41313 10 -0.11664 -0.40307 0.35137 4. empirical analysis based on dea method according to the above analysis, the paper finally identifies three factors, which are input factor, output factor and risk factor. when using the dea model to analyze the data validity requirements, we need to change the negative value of the common factor score coefficient table. this paper draws lessons from “performance evaluation of open end funds in china based on factor analysis and dea method” [5], the conversion formula as shown below: )5()min/(max)min(9.01.0 jijiijij cc −−+= among them, 𝑚𝑎𝑚𝑖𝑗 represents the maximum value of the j item data, and min𝑗 represents the minimum value of the j item data. after conversion, data can be positive and will not affect the results of the analysis of the dea model. as shown in table 6: table 6. common factor score coefficient after conversion fund order input factor output factor risk factor 1 0.277151072 0.862052506 0.49218035 2 0.1 0.18255694 0.549595431 3 0.210041512 0.38684699 0.691715845 4 0.155222742 0.333047526 0.51139997 5 0.16185264 0.1 0.1 6 0.232814457 1 0.219140377 7 1 0.411992103 0.373633859 8 0.452071055 0.137826689 0.256467052 9 0.457956376 0.386125107 1 10 0.441649627 0.384086155 0.349868025 according to the data in table 5, this paper uses deap version 2.1 software to calculate the dea-bcc performance evaluation results of the 10 funds, with input factors as input indicators, output factors and risk factors as output indicators, results as shown in table 7. table 7. securities investment fund performance dea evaluation value ` synthetical efficiency pure technical efficiency scale efficiency return of scale 1 0.840 1.000 0.840 drs 2 1.000 1.000 1.000 3 0.755 1.000 0.755 drs 4 0.818 0.849 0.964 drs 5 0.198 0.618 0.321 irs 6 1.000 1.000 1.000 7 0.128 0.137 0.932 irs 8 0.127 0.221 0.576 irs 9 0.422 1.000 0.422 drs 10 0.270 0.301 0.900 irs published by atlantis press copyright: the authors 19 hong-mei zhang et al. / performance evaluation of government venture capital fund (1). return of scale according to the results of dea analysis, there are 4 funds in the 10 securities investment funds which are in the state of diminishing returns to scale, that is, the problem of output redundancy, indicating that the enterprises should appropriately reduce the input size so as to keep the input and output in a more efficient state. there are 4 funds in the state of increasing returns to scale. that is, owing to less investment by the enterprises, the production scale is not up to the best state, so it should appropriately increase the amount of input. the return of scale of other two funds is unchanged, which shows that the enterprise has been in the production and operation of the maximum efficiency of input and output efficiency. (2). synthetical efficiency observing the efficiency of the 4 funds with diminishing returns to scale, as can be known, its synthetical efficiency is lower than 1, and the reason that the efficiency of the fund 1, 3, 9 is low is the lower scale efficiency, whose pure technical efficiency is 1, that is, the enterprises need to increase capital investment; fund 4 with diminishing returns to scale is caused by two factors. in addition, the synthetical efficiency of the 4 funds of increasing returns to scale is low, and the reason is that both pure technical efficiency and scale efficiency are low. that is, the enterprises need to adjust their production scale as well as improve the management level of the company. in conclusion, the 8 fund invested enterprises still have space to improve, and need to adjust their own production to improve the comprehensive efficiency. (3). pure technical efficiency in these 10 funds, the pure technical efficiency of 5 funds is less than 1, respectively, 4, 5, 7, 8, 10, which shows the corresponding invested companies should improve the management level of technology to promote their pure technical efficiency and improve the synthetical efficiency. the pure technical efficiency of the fund 1, 2, 3, 6, 9 is 1, indicating that for the corporations of the investment of these funds, the use of resources is efficient in the current level of investment in the technology. (4). scale efficiency the scale efficiency of 8 funds is less than 1, respectively, 1, 3, 4, 6, 7, 8, 9, 10, indicating that the enterprises of the investment of the 8 funds are not currently effective, but also there is a certain improvement in space. 5. concluding remarks based on the reference to relevant literatures of the domestic and foreign scholars about the performance evaluation of the government venture capital fund, this paper chooses 6 indexes used to fully reflect the performance of guide funds and constructs the index system of guide fund performance evaluation, combined with actual situation of our country and the characteristics of guide funds. after collecting the data of 10 enterprises shared by the government guide venture capital fund, we uses the factor analysis method to optimize the evaluation model to the one including input index and output index and risk index, a total of three common factors, and then uses empirical analysis based on dea-bcc model. the result shows that most of the current guide funds are insufficient in the aspect of the production efficiency. at the same time, empirical analysis based on deabcc model shows that the results can objectively reflect the operating conditions of companies accepted investment, which provides the objective reference for government venture capital fund and companies accepted investment. references [1] ministry of commerce, ministry of finance, national development and reform commission, guidance on the establishment and operation of the guide fund for venture capital, 2008. [2] lerner, joshua. the government as venture capitalist: the long-run impact of the sbir program[j].journal of business,1999,(72):285-318. [3] david b.audretsch, albert n.link, john t.scott. public/private technology partnership: evaluating sbir-support research[j]. research policy, 2002,31(1):145-158. [4] jonathan g.s.koppel, the challenge of administration by regulation :preliminary findi ngs regarding the u.s. goverment’s venture capital funds. journal of public administration research an theory, 2008. [5] guiying liu, performance evaluation of open end funds in china based on factor analysis and dea method[d], central south university, changsha, 2008. [6] hongjiang li, xiaoyan bao, research on the performance evaluation of government oriented venture capital fund[j], business studies, 2011,(6):112-116. [7] bo gui, research on the competitiveness of listed banks based on dea and factor analysis[d], shandong university, jinan, 2009. [8] yu yu, research on the performance evaluation of government guided fund[d], tianjin university, tianjin, 2013. [9] zhipeng qin, research on the performance index system of venture capital guiding fund of china's strategic emerging industry[d], university of international business and economics, beijing, 2014. [10] huili hao, patrick long, scott curtis, attitudes of property owners to climate change considerations and their effects on future property values in coastal communities[j], jracr(journal of risk analysis and crisis response), 2012,(12):185-191. published by atlantis press copyright: the authors 20 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <feff004b0069007600e1006c00f30020006d0069006e0151007300e9006701710020006e0079006f006d00640061006900200065006c0151006b00e90073007a00ed007401510020006e0079006f006d00740061007400e100730068006f007a0020006c006500670069006e006b00e1006200620020006d0065006700660065006c0065006c0151002000410064006f00620065002000500044004600200064006f006b0075006d0065006e00740075006d006f006b0061007400200065007a0065006b006b0065006c0020006100200062006500e1006c006c00ed007400e10073006f006b006b0061006c0020006b00e90073007a00ed0074006800650074002e0020002000410020006c00e90074007200650068006f007a006f00740074002000500044004600200064006f006b0075006d0065006e00740075006d006f006b00200061007a0020004100630072006f006200610074002000e9007300200061007a002000410064006f00620065002000520065006100640065007200200035002e0030002c0020007600610067007900200061007a002000610074007400f3006c0020006b00e9007301510062006200690020007600650072007a006900f3006b006b0061006c0020006e00790069007400680061007400f3006b0020006d00650067002e> /ita <feff005500740069006c0069007a007a006100720065002000710075006500730074006500200069006d0070006f007300740061007a0069006f006e00690020007000650072002000630072006500610072006500200064006f00630075006d0065006e00740069002000410064006f00620065002000500044004600200070006900f900200061006400610074007400690020006100200075006e00610020007000720065007300740061006d0070006100200064006900200061006c007400610020007100750061006c0069007400e0002e0020004900200064006f00630075006d0065006e007400690020005000440046002000630072006500610074006900200070006f00730073006f006e006f0020006500730073006500720065002000610070006500720074006900200063006f006e0020004100630072006f00620061007400200065002000410064006f00620065002000520065006100640065007200200035002e003000200065002000760065007200730069006f006e006900200073007500630063006500730073006900760065002e> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style received 10 june 2015 accepted 6 august 2015 spatial effect on public risk perception of natural disaster: a comparative study in east asia zhongyu he department of urban planning and design, nanjing university, 22 hankou road nanjing, jiangsu 210093, china e-mail: hezy@nju.edu.cn guofang zhai department of urban planning and design, nanjing university, 22 hankou road nanjing, jiangsu 210093, china e-mail: guofang_zhai@nju.edu.cn abstract this paper explores the effect of geographical location on public risk perception of natural disasters. by conducting an identical questionnaire survey across three east asia countries (china, japan and south korea), the paper finds out that different country has its unique structure of risk perception. generally, the risk perception of sample residents in japan weakens as the distance from the risk source increases, which indicates japanese people’s risk perception reflects the actual risk probability. on the other hand, korean partially and chinese hardly perceive the probability of existing risks. the findings of this paper imply that risk perception to certain extent relates with existing risk, however, the latter is not always correctly perceived. economic development, socio-political system, historical and cultural backgrounds will affect public risk perceptions. keywords: spatial effect, risk perception, natural disaster, east asia, closeness to risk 1. introduction risk perception is defined as the personal belief that one can be potentially harmed or to be exposed to the risk. in a risk society claimed by ulrich beck (1986), the risk perception of the public plays an increasingly important role in shaping environmental policy and the risk management and response systems (slovic, 2000). the beginning of risk perception research can be traced to the nuclear debate of the 1960s, yet today it still remains as a phenomenon in search of an explanation (sjoberg, 2000). of course, real risk is undoubtedly a primary factor that affects risk perception, as illustrated by some well sited papers (von winterfeldt, john & borcherding, 1981; lichtenstein et al., 1978). however, numerous empirical studies have reported that risk is perceived quite differently even in the case where the statistical probability of a certain risk is identical. some scholars have examined the relationship between perceived risks and the type of disaster event and reported a significant result (ho, brossard & scheufele, 2008). others tested how various individual-level characters like worldviews, education levels and religious preferences may influence attitudes and perceptions about risk issues and risk events (kahann et al., 2006). on the other hand, a large body of literature focuses on how varied risk communication results in the disparity of risk perception under the cognitive-experiential theory or fuzzy trace theory (epstein, 1994; reyna & brainerd, 1991). some specific discussion includes the framing effect (tversky journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 161-168 published by atlantis press copyright: the authors 161 z. he, g. zhai & kahneman, 1981), ratio bias (yamagishi, 1997) and frequency effect (slovic, monahan & macgregor, 2000). risk perception is also believed to be affected by risk target, namely, people do not make the same estimate when they rate the risk to themselves, to their family, or to people in general (sjoberg, 2000). some recent studies have started to explore spatial issues related with risk perception. teigen (2005) suggests that perceived closeness, whether physical or psychological, can influence perceptions of risk, while some others argue that psychological proximity to hazard has a greater influence (brown et al., 1992). in another study, researchers find out that american citizens’ risk judgment towards terrorist attack is positively correlated with their geographical distance to the world trade center (fischhoff et al., 2003). meanwhile, a spatial mismatch between real risks and perceived risks by the public is reported as well, for example, brody et al. (2004)’s empirical study shows that people’s perceptions of air quality in texas, u.s.a. are not significantly correlated with scientifically measured pollution level in air monitoring stations. therefore, as is pointed out, the spatial effect on environmental risk perception is still in lack of scientific and systematic study (gattig & hendrickx, 2007). 2. geographical location and natural disaster perception among the various environment, health and safety risks, natural disaster is the one whose risk probability is closely related with spatial location. for example, it is self-evident that the closer to a river, the more likely for a house to be suffered during a flood (fig. 1), and similarly, residents who live near coastal areas have a bigger chance to be affected than those who live in the inner land during a tsunami. by conducting an identical questionnaire survey in three east asia countries-japan, china and south korea, this paper aims to provide some explanation of the effect of geographical location on public risk perception towards natural disasters. to our best knowledge, it is the first international comparative research focusing on the spatial effect on risk perception and therefore will contribute to the quite limited studies on this topic. especially, the following research questions are raised: 1) how does the risk perception of the public vary as the distance to natural disasters changes? 2) does the geographical location affects public risk perception in a similar or dissimilar way in the three countries? 3) what individual-level and country-specific factors account for the public risk perception? located in east asia, the three target countries share some common historic backgrounds as they are all deeply influenced by the ancient chinese civilization, at the same time they differ greatly in their economic development (table 1.), social and political system as well as natural conditions. table 1. main indicators in china, south korea and japan in the survey period (as of 2006) china korea japan area (100,000 km2) 960 9.93 37.79 population (million) 1316 48.3 127.8 nominal gdp (billion $) 2278 788 4554 nominal gdp per capita ($) 1732 16471 35650 real gdp growth rate (%) 10.2 4.0 3.1 fig. 1. forecasted flooded area by shonai river in nagoya city, japan (source: chubu construction bureau, mlit, japan, http://www.cbr.mlit.go.jp/shonai/sitemap/ 2015-06-01). this hazard map shows that the potential flooded depth of certain area is positively related with the proximity to the river. published by atlantis press copyright: the authors 162 risk perception’s spatial effect 3. survey and hypothesis the survey was carried out in 2006, around 3000 questionnaires were sent to residents living in three coastal cities with potential risk of tsunami-yokohama, tianjin, busan respectively. the initial questionnaire was designed in japanese and then translated into chinese and korean by native speakers who are fluent in japanese. the surveys were modeled after the total survey design (tsd) method, which attempts to achieve an optimum balance across all areas of effort. in the end, 2278 questionnaires were returned with an average response rate of 73.7% (table 2.). risk perception was measured through the survey by asking the question: “using a scale from 0 to 10 where 0 is the least and 10 is the most, how strongly do you feel about the following risks?” then 29 daily risks including natural disasters, epidemic diseases and incidents appeared in the questionnaire. individual characters of the respondent such as sex, age, income and education level are recorded. the respondents were also asked to provide the direct distance from their home to the coast line. as a comparison group, an earlier survey conducted in japan’s nagoya city and toki city in 2004 with a sample size of 428 is used together in the following analysis. the main difference between this survey and the cross-country one is that in the nagoya-toki survey, the target sample residents are living along the watershed of shonai river in the inner land rather than coastal area, which indicates they face a potential risk of flooding rather than tsunami. for the purpose of this paper, 3 out of the 29 risks are selected for further analysis, namely tsunami, flooding and global warming. tsunami is a major threat to coastal residents while flooding is the most frequent natural disaster occurred in watershed areas; the risk of both disasters decreases as the distance to coast line or river bank increases. on the other hand, global warming is a universal natural disaster producing damages to all the human beings with no regard to geographical locations. therefore, suppose people make rational decisions, the following hypothesis can be made: hypothesis 1: if the risk of global warming is correctly perceived, then people’s risk perception table 2. survey details japan_1 japan_2 china south korea survey time march 24 to april 16, 2004 april 14 to may 14, 2006 mid. nov. to mid. dec., 2006 nov.1 to dec.11, 2006 focused participants watershed residents coastal residents coastal residents coastal residents sampling method random sampling from telephone directory random sampling from telephone directory random sampling from school list on-site delivery distributed samples 1000 1000 1000 1090 validly samples 962 835 1000 1090 returned samples 428 450 963 865 survey process delivering survey booklets to sampled participants along with reply postcards sending reminder postcards to remind returning the reply postcards collecting the booklets delivering survey booklets to sampled participants along with reply postcards sending reminder postcards to remind returning the reply postcards collecting the booklets delivering survey booklets to sampled schools sampled schools distributed booklets to students students took booklets back and their parents answered them. collecting the booklets delivering the survey booklets to surveyors. surveyors distributed booklets to participants. collecting the booklets published by atlantis press copyright: the authors 163 z. he, g. zhai should bear no relationship with their spatial location from coast line or river bank. hypothesis 2: if the risk of tsunami is correctly perceived, then for residents living in coastal areas, their risk perception will decrease as they live further away from coast line while for residents living in watershed areas, their risk perception should bear no relationship with their residence location. hypothesis 3: if the risk of flooding is correctly perceived, then for residents living in watershed areas, their risk perception will decrease as they stay further away from river bank while for residents living in coastal areas, their risk perception should bear no relationship with their residence location. section 4 and section 5 will examine if these hypothesis can be validated or not. 4. three tests of spatial effect on risk perceptions test 1: first, the relationship between the public risk perception towards global warming and distance from coast line in the three countries are examined. the distance is divided into 6 groups: those who live less than 100m from the coast line, between 100m and 500m, between 500m and 1km, between 1km and 2km, between 2km and 5km, and those who live further than 5km from the coast line. respondents who state they are not sure about the distance are excluded from the analysis; this is why the number of observations is less than the number of total valid samples. as fig. 2 shows, japanese have the strongest risk perception in the three countries, and koreans’ risk perception is stronger than that of chinese respondents’. the perceived risks are not related with geographical locations in all the three groups, thus hypothesis 1 cannot be denied. however, we are not sure if this result is due to correct risk perception or merely a coincidence. therefore, it is necessary to do some further test. test 2: in the next step, a similar examination about tsunami risk is carried out (fig. 3), and an interesting pattern is found out: basically, japanese respondents correctly judge the real risks because their risk perception score lowers as the distance from coast line increases, and an f-test shows that the difference between different distance groups is statistically significant (p-value =0.001). the risk perception of korean respondents’ is partially correct, for a general decreasing tendency can be observed. on the other hand, the risk perception of the chinese respondents’ can hardly reflect existing risk in reality. the result indicates that for chinese and korean group, the observation in test 1 is a coincidence, and as a matter of fact, neither of the hypothesis 1 and hypothesis 2 should be accepted. however, for the japanese group, the two hypotheses seem to be true. to confirm the conclusion, another test is done as follows. fig. 2. global warming risk perception with regard to distance from coast line (no. of observations: japan=395, china=535, korea=658) fig. 3. tsunami risk perception with regard to distance from coast line (no. of observations: japan=395, china=538, korea=658) published by atlantis press copyright: the authors 164 risk perception’s spatial effect test 3: in this final test, we compare the two japanese groups (one from coastal area and the other from watershed area) and check how the three risks (global warming, tsunami and flooding) are perceived in the two groups. as shown in fig. 4 and fig. 5, the score of global warming shows no relationship with distance from coast line or river bank in both groups. for the coastal area group, the score of tsunami risk has an obvious positive correlation with proximity to coast line (sig. =0.001 in f-test) while the score of flooding indicates a disconnection between distance and risk perception (sig. =0.074 in f-test). at the same time, for the watershed area group, an opposite pattern is found out (p-values in f-tests checking difference between different distance groups are 0.508 for tsunami and 0.000 for flooding), expect the last group that lives more than 5km away from river bank (as the red circle shows in fig. 5). a further look upon this group reveals that the number of samples in this group is merely 10; therefore, it is reasonable to attribute the unexpected score of this group to the inadequate sample size. if this last group is dropped out, then the flooding risk perception will present a perfect pattern that goes down with the distance from home to the river bank increases. the results in test 3 are totally in line with the three hypotheses made earlier. so far, it is quite convincing to reach the conclusion that geographical location affects japanese people’s risk perception towards natural disaster in a dramatic way and in other words, japanese people always perceive natural disasters in a more sensitive and correct way than south korean and chinese people. 5. regression analysis analysis in section 4 reveals an interesting finding and confirms some of the concern of the paper. to fully understand factors that impact public risk perception, especially the separated influence from individual-level characteristics and spatial effect in different countries, a multivariate regression analysis is done in spss. three models are established, and the regression result is shown in table 3, the dependent variable is the score of perceived risk towards tsunami, and it is dealt as a consecutive variable ranging from 0 to 10. independent variables include age, gender, family’s annual income, education level of the respondent, and distance from home to coast line. age is divided into 7 groups with an interval of 10 years. family’s annual income is set based on the actual income level in each country and is treated as a consecutive variable with higher value indicating a higher income. for distance, again those who answered they were not sure about the distances from their home to the coast line are excluded from the regression. generally speaking, adjusted r2 of all the three models are quite small and therefore the goodness of fit of the models are not satisfying, which implies that individual characteristics and spatial location are not major factors that affect risk perception. sex and education level in all three countries are not statistically related with tsunami risk perceptions. age and income have no impact on risk perception for chinese and japanese respondents, neither. however, for fig. 4. risk perception of the japanese coastal area group (no. of observations: tsunami=395, flooding=395, global warming= 395) fig. 5. risk perception of the japanese watershed area group (no. of observations: tsunami=339, flooding=351, global warming= 353). published by atlantis press copyright: the authors 165 z. he, g. zhai respondents from south korea, older people and people with higher income show a stronger feeling about potential tsunami hazard. as for the spatial effect, the regression results confirm the above conclusions from section 4: for the survey in japan, all the spatial variables are significant in the model. for example, compare with residents living within 100m from the coast line, those living at a distance of 100m to 500m from coast line will have a lower score of about 1.2 in their feeling about potential risk of tsunami, and those living more than 5km from the coast line will have a lower score of about 3.2. for the survey in south korea, only the last spatial variable is statistically significant, even though the directions of all the other spatial variables are negative (-) in the model. this indicates that south korean residents living further from coast line tend to think the tsunami risk is lower than those living closer from coast line; however, they are not as sensitive as their japanese counterparts, the risk perceptions are statistically different only when the disparity of proximity to coast lines is large enough (<0.1m and >5km). for the survey in china, no pattern can be observed. the direction of the spatial variables is not identical, and none of the variables are significant. 6. discussion and conclusion by conducting an identical questionnaire survey across three east asia countries (china, japan and south korea) with a sample size of more than 2,200, this paper investigate the spatial effect on public risk perception towards natural disaster. a second concern of the paper is how individual characters influence the risk judgment. the relationship between geographic location and risk perception is a rarely discussed topic in existing literature; the comparative study between the three countries shows that different country has different structures of risk perception. and generally speaking, japanese basically, south koreans partially and chinese hardly perceive the existence of real risks. for natural disasters whose influence is identical across space (e.g. global warming), japanese people’s risk perception is not related with their geographical location; while for table 3. regression results independent variable japan china south korea sex (male=0, female=1) .013 -.049 -.040 age (1=10-19years old, 2=2029years old, …, 7=older than 70 years) -.133 .002 .499*** family annual income .027 .012 .119*** education (ref.=middle school) high school vocational school college postgraduate -.385 -.017 -.606 -.806 -.368 -.905** .046 -1.342 1.410 2.133** .834 1.369 distance (ref.=<0.1km) 0.1-0.5km 0.5-1km 1-2km 2-5km >5km -1.242* -1.630** -1.337* -2.220*** -3.175*** -.538 -.521 1.043 .354 .678 -.843 -.842 -1.037 -1.179 -2.388** constant 5.536*** 2.484*** 2.970** no. of observations adjusted r 2 395 0.019 538 0.005 658 0.140 ***p-value <0.01 **p-value<0.05 *p-value<0.1 published by atlantis press copyright: the authors 166 risk perception’s spatial effect natural disasters whose risk is closely related with proximity to the risk source (e.g. flooding, tsunami), japanese people’s risk perception is weakened as their distance from the risk source increases. this finding supports previous research (sjoberg, 2000; von winterfeldt, john & borcherding, 1981; lichtenstein et al., 1978) that real risk does affect risk perception, however, the mixed results from the three countries imply that real risks are not always correctly perceived, this is also confirmed by other papers (brody, peck & highfield, 2004). some reasons may account for this difference in the capability of risk judgment: first, based on the results of this paper, a country’s economic development is one factor that correlates with public risk perception, even though it is not necessarily to be a causal relationship. according to the hierarchy of needs theory by maslow, physiological needs is the most fundamental level of needs for human beings; safety needs is on a higher hierarchy and only with people’s physical needs relatively satisfied, does the individual’s safety needs take precedence and dominate behavior. in a less developed economy, its citizens will consider more about physical requirements for human survival like food, clothing, shelter, etc. while the more developed an economy is, the more likely physiological needs will be satisfied and therefore in a developed economy, people will concern more on safety issues like the potential risks of natural disasters. this may explain the different levels of risk judging performances in china-a developing country, south korea-a new industrialized country and japan-a highly developed country. second, socio-political system also exerts an impact on public risk perception. previous literature has found that trust in government can influence people’s perceptions and reactions to issues of food safety (houghton et al., 2006; kuttschreuter, 2006), and this seems be true for natural disasters too. japan is well known for its well-designed risk management system and legislation; japanese government is also among the top countries in government transparency ranking. both the central and local governments in japan have been making great efforts in risk education and risk communication with the public. these measures largely improve the public understanding towards natural disasters and help the public build a correct perception towards these risks. on the other hand, korean and chinese governments still have some space to improve in their emergency management and risk information disclosure. the result in japan’s survey not only implies that our respondents have a clear risk judgment, but also indicates they have an identical judging criterion. and this cannot be achieved without the effective risk education from the government. third, historic and cultural backgrounds will influence risk perception in one society as well. due to its special geographical location, japan has been suffering from the most disadvantageous natural conditions in the world. frequently occurred disasters have shaped the country’s unique cultural character, namely a crisis consciousness embedded in daily behavior (gao, 2003). as a result, japanese people are always prepared to various natural disasters and show more interests in accessing risk information. we believe such a personal concern will lead to a better risk understanding and to some extent explain the better performance of the japanese respondents in the survey. existing literature has pointed out that personal disaster experience also affects individual’s risk perception. however, since no tsunami has occurred in the three sample cities in recent history, the factor of personal experience can be excluded in explaining the difference in risk perceptions in the three countries. this paper also finds out that individual level characters, especially demographic factors, are not closely related with risk perception. this is not difficult to understand, for risk perception is a highly subjective judgment and may not be affected by a person’s physical condition such as income or education. besides, existing literature argues that demographic factors have different powers in explaining risk perception towards different hazard events (cummings, berube & lavelle, 2013). for example, education level is statistically significant for perceived emerging technological risks but not for manufacturing risks (cummings, berube & lavelle, 2013). meanwhile, some studies show that some other individual characters like worldview, religious belief, ideology and political identification significantly affect risk perception (brody, peck and highfield, 2004; cummings, erube & lavelle, 2013). since all these factors belong to the psychological dimension as same as risk perception, their findings are quite reasonable. also, these papers reveal that preferred media for information seeking somewhat accounts for public risk perception; due to the limited published by atlantis press copyright: the authors 167 z. he, g. zhai space in the questionnaire, this factor is not tested in our survey. risk geography is a relatively new branch in geography discipline. to help the public establish a correct risk perception with regard to space and location is quite important in a society with increasing uncertainty and hazard, since risk perception is closely related with the public’s response to the risk, and therefore determine whether the social and economic resource will be allocated effectively. when there is a mismatch between the real risk and perceived risk, the value of land and space will be distorted and economic loss will occur for the whole society. the current paper is a preliminary study to explore the spatial aspects of risk perception, future study can further examine if the explanations provided in this paper can be validated by conducting more elaborate experiments. future research can also thoroughly check the influence of culture by doing comparative studies between the western and eastern countries. what is more, it is important for policy makers to find out how to improve the public’s spatial consciousness towards natural disaster risks through effective risk communications. acknowledgements this research was co-supported by the national program on key basic research project of china (973) under grant no. 2010cb428506 and project of national natural science foundation of china under grant no. 51308280. references u. beck, risk society: towards a new modernity (sage, new deli, 1986) p. slovic, perception of risk. in the perceptions of risk, eds. p. slovic (earthscan publications, london, 2000), pp. 220231. l. sjoberg, factors in risk perception, risk analysis. 20(1) (2000) 1-11. d. von winterfeldt, r. s. john, and k. borcherding, cognitive components of risk ratings, risk analysis 1(2) (1981) 277–287. s. lichtenstein et al., judged frequency of lethal events, journal of experimental psychology: human learning and memory 4 (1978) 551–578. s. ho, d. brossard, and d. scheufele. effects of value predispositions, mass media use, and knowledge on public attitudes: toward embryonic stem cell research, international journal on public opinion research 20 (2) (2008) 171–192. d. kahan, p. slovic, d. braman, and j. gastil, fear of democracy: a cultural evaluation of sunstein on risk, harvard law review 119 (2006) 1071–1109. s. epstein, integration of the cognitive and psychodynamic unconscious, american psychologist 49 (1994) 709−724. v.f. reyna and c.j. brainerd, fuzzy-trace theory and framing effects in choice, gist extraction, truncation, and conversion, journal of behavioral decision making 4 (1991) 249−26. a. tversky and d. kahneman, the framing of decisions and the psychology of choice, science 211 (1981) 453–458. k. yamagishi, when a 12.86% mortality is more dangerous than 24.14%: implications for risk communication, applied cognitive psychology 11 (1997) 495–506. p. slovic, j. monahan and d.g. macgregor, violence risk assessment and risk communication: the effects of using actual cases, providing instructions, and employing probability vs. frequency formats, law and human behavior 24 (2000) 271−296. k.h. teigen, the proximity heuristic in judgments of accident probabilities, british journal of psychology 96 (2005) 423−440. j. d. brown, n. j. novick, k. a. lord, and j. m. richards, when gulliver travels: social context, psychological closeness, and self-appraisals, journal of personality and social psychology 62 (1992) 717−727. b. fischhoff, r. m. gonzalez, d. a. small and j. s. lerner, judged terror risk and proximity to the world trade center, journal of risk and uncertainty 26 (2003) 137−151. s.d. brody, b.m. peck and w.e. highfield, examing localized patterns of air quality perception in texas: a spatial and statistical analysis. risk analysis 24 (6) (2004) 1561-1574 a. gattig and l. hendrickx, judgmental discounting and environmental risk perception: dimensional similarities, domain differences, and implications for sustainability. journal of social issues 63 (1) (2007) 21-39 j. r. houghton, e. van kleef, g. rowe and l. j. frewer, consumer perceptions of food risk management practices: a cross-cultural study. health, risk and society 8 (2) (2006) 165−183. m. kuttschreuter, psychological determinants of reactions to food risk messages. risk analysis 26 (4) (2006) 1045−1057. f. gao, consciousness of crisis and characteristics of japanese culture, japanese study forum 1 (2003) 30-33. (in chinese) c. l. cummings, d. m. berube and m. e. lavelle, influences of individual-level characteristics on risk perceptions to various categories of environmental health and safety risks, journal of risk research, 16 (10) (2013) 1277-1295. published by atlantis press copyright: the authors 168 microsoft word perception of risks in nanotechnology_ determining key aspects in chile perception of risks in nanotechnology: determining key aspects in chile j. pozo instituto de ciencias básicas. universidad diego portales. ejército 441 santiago, chile. e-mail: julio.pozo@udp.cl www.udp.cl a. schmessane instituto de ciencias básicas. universidad diego portales. ejército 441 santiago, chile. e-mail: ann_schme@yahoo.es www.udp.cl a. pozo universidad diego portales santiago, chile. e-mail: alejandra_pozo@yahoo.com c. pozo universidad diego portales santiago, chile. e-mail: caritop80@yahoo.com abstract the purpose of this study is to provide preliminary base-line data on public perceptions about the risks associated with nanotechnology. we analyzed the current development of nanotechnology in chile and the risks associated with the incorporation of this new technology over time, considering the particular case of chile and its position in science and technology. a survey was applied to determine perceptions about the most prominent risks. keywords: science, nanotechnology, nanomaterials, technology, legislation, risks 1. introduction the size of nanoparticles represents one of the risks associated with nanotechnology, considering that the smaller a particle is, the greater its reactivity. thus, we can infer that a particle that is inert at the micro or macro scale may be harmful at the nano scale. the possible risks associated with nanotechnology include nanoparticles penetrating the skin and entering the bloodstream and the interaction and effect of nanoparticles on ecosystems. finally, there are economic effects associated with the proliferation of products in the market.1, 2 nanoparticles in contact with living tissue can cause the appearance of free radicals and cancer. some studies have dealt with the potentially harmful consequences of nanoparticles to living beings and their rise through the food chain.3, 4 currently, scientific research and technological problems related to nanotechnology deserve great interest from both theoretical and experimental journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 34-43 published by atlantis press copyright: the authors 34 administrateur texte tapé à la machine received 18 april 2011; accepted 12 march 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine j. pozo, a. schmessane, a. pozo, c. pozo points of view. this is because the results that are being generated by describing their behavior try to solve many problems currently facing humanity, which to date have remained unsolved. the objective of this study is to generate preliminary base-line information on public perceptions about the risks associated with nanotechnology. 1.1 theoretical analysis all changes in technology have associated economic changes. from the point of view of the economic importance of nanotechnology, nasa and the national science foundation have considered that the development of nanotechnology is of strategic importance. in the view of many people and institutions around the world, nanotechnology is producing a new industrial revolution. the financial and economic risks that nasa wants to avoid are evidenced by the fact it did not disclose the results of around 150 patent applications in the field of nanotechnology.5 some first world economies have made significant investments in research, development and innovation (r & d & i) in nanotechnology. therefore, it is expected that nanotechnology will have strong implications in future global economic trends, which may impact on almost all activities of our society. 1.1.1 nanotechnology, economics and global competitiveness it has been noted that in the very near future, the development of nanotechnology will revolutionize science, technology and human society. nanotechnology has the potential to transform medicine, biotechnology, agriculture, manufacturing, material sciences, information technology, and telecommunications, among other fields. the world is entering an era driven by accelerated technology projects and business ventures that generate significant economic value. we are in the middle of great changes in economic systems led by the rapid exponential growth of new technologies. nanotechnology should be understood in this context, as such, it is a continuation in the acceleration of advanced technology and perhaps most importantly, it can have the capacity to transform the future global economy. ten years ago, america appeared to be the only country concerned with developing nanotechnology. today, several countries have joined in this pursuit, such as china, india, japan, australia, israel, korea, and most recently russia. all of them are becoming aware of the potential advantages of “betting” on this sector, while europe seems to be more cautious.6 significant advances in the field of nanotechnology and nanoscience have been achieved in the last two years, to the point that no one doubts that this field is set to become the ultimate strategic sector in the most advanced economies worldwide. the evolution and growth of technology has become a cornerstone for the promotion of competitive innovation, the emergence of new businesses and prospects of economic progress for systems that strive to be players in the era of globalization and knowledge. while nanotechnological advances take place, economists and specialists are looking ahead to identify the impacts of the evolution and development of current technology both in economics and in business. nanotechnology is critical because it can provide the necessary tools to facilitate a reorganization of society, business and the structure of our economies.5, 6 1.1.2 nanotechnology as a strategic sector in the international arena nanotechnology projects can be divided into three phases6. nanotechnology is currently at the stage of research and the development of scientific knowledge, and is only beginning to have applications. during the next five years, many applications that are in the beginning stages are expected to be further developed. in the following ten years, nanotechnology will be consolidated as an industry, and consumers will be enjoying a wide range of products using nanotechnology. 1.2 overview the mechanical construction of molecules was first suggested by richard feynman, who is now considered the founder of nanotechnology. in an after-dinner speech entitled “there’s plenty of room at the bottom” at the california institute of technology in 1959, and later published in 19607, feynman proposed an alternative approach to miniaturization that would use mapublished by atlantis press copyright: the authors 35 perception of risks in nanotechnology: determining key aspects in chile chines to build smaller machines, which could then build still smaller machines, and so forth. according to feynman “the principle of physics, as far as i can see, do not speak against the possibility of maneuvering things atom by atoms. it is not an attempt to violate any laws; it is something, in principle, that can be done; but, in practice, it has not been done because we are too big”. the prefix nano refers to one-billionth of something, for example a longitude of 1 nanometer (1 nm= 10(-9) m). the word nanotechnology is used extensively in relation to science and technologies at the nanoscale, an extremely small scale at which molecular structures and their atoms are manipulated, with the possibility of making building-materials and machinery from atomic and molecular arrays. currently, there are super microscopes that can move individual atoms using a probe approach, such as the scanning tunneling microscopy (stm), which is capable of creating moving images using electrically conductive needles on the atoms in a conductive surface. the stm was invented by binnig and rohrer at the ibm laboratory in switzerland in 1982. in 1986, the inventors were awarded the nobel prize, a clear confirmation of the importance of stm. the stm maps topography by measuring the tunnel current between an atomically sharp tip and a metallic sample. another breakthrough came in 1986 with the invention of the atomic force microscope (afm) by binnig, quate and gerber. the afm measures the force between a tip and a sample. in contrast to the stm, insulated samples can also be imaged. the afm is suitable for different types of materials. the technique presented by these microscopes is very versatile and of great interest in the characterization of devices in the field of nanoscience and nanotechnology. in this context it has been possible to characterize the morphology of nonporous surfaces grown by electrodeposition, mechanical properties of metal surfaces in addition to significant magnetic properties. 1.2.1 definition of nanotechnology nanotechnology systematically organizes and manipulates properties and behaviors of matter at the atomic and molecular levels in order to create functional devices, materials and systems on a nanometer scale. in this context, the most frequently asked question is: what is nanotechnology? the us national science and technology council defines nanotechnology as: “the essence of nanotechnology is the ability to work at the molecular level, atom by atom, to create large structures with fundamentally new molecular organization. the aim is to exploit these properties by gaining control of structures and devices at atomic, molecular, and supramolecular levels and to learn to efficiently manufacture and use these devices”. 1.2.2 nanomaterials nanomaterials are single or multi-phase polycrystalline structures with a typical size of 1 to 100 nanometers (nm) in at least one dimension. the focus of nanomaterials is a bottom-up approach to structures and functional effects so that the building blocks of materials are designed and assembled in a controlled manner. some examples are nanoclays that reinforce plastics or carbon nanotubes to add conductivity to various materials such as zinc oxide and cerium oxide. zinc oxide is ideal for use in cosmetics and industrial application of radiation absorbing transparent ultraviolet light (uv), and antibacterial functions. cerium is ideal for use in catalytic engines fuel cells and transparent uv absorbers. in these nanomaterials, most atoms are located on the surface of the particles, augmenting or enhancing chemical, mechanical, optical and magnetic properties. applications of nanomaterials include the generation of computer chips, phosphors for high-definition tv, highpower magnets, high-sensitivity sensors, more durable satellites, more durable medical implants and others.8 1.2.3 carbon nanotubes computer companies with large laboratories, such as ibm or hp, have research programs in nanotechnology. the interest is to obtain smaller electronic chips, so that when silicon technology reaches its limit, it will probably be replaced by a new concept which is called carbon nanotubes. nanotubes are composed entirely of carbon materials where the basic unit is a plane of graphite rolled into a cylinder, forming tubes whose diameter is on the order published by atlantis press copyright: the authors 36 j. pozo, a. schmessane, a. pozo, c. pozo of a few nanometers. these single-walled tubes have properties that are stronger than steel. 1.2.4 the graphene this is a sheet of carbon one atom thick, which was first made in a laboratory in 2004. graphite, which is the material commonly used in pencils, is made up of many layers of graphene. it should be noted that the conductive properties of graphene are highly efficient and can be used in nanoelectronics. in the form of a long ribbon of dimensions on the order of nanometers, graphene has unique electrical properties and behaves as a metal or semiconductor. graphene has a structure in which carbon atoms in flat sheets are joined, forming a hexagonal honeycomb, with one atom at each vertex. graphite is obtained by placing many sheets on top of each other. then if a portion of one of these films is wound in the form of a sphere, like a football, fullerenes are obtained (molecules of recent interest). if the structure related to the fullerenes is rolled into a cylinder, it will form a carbon nanotube. within this field, a graphene honeycomb corresponds to a single extended two dimensional flat structure. 1.2.5 dry nanotechnology dry nanotechnology deals with the manufacture of structures in carbon (carbon nanotubes), silicon, inorganic materials, metals and semiconductors, and is used in electronics, magnetism and various optical devices, in addition to being controlled by selfassembled computers. 1.2.6 wet nanotechnology this technology is based on biological systems in wet environments, considering genetic material, membranes, enzymes and other cellular components. in brief, we can say that wet nanotechnology is used in the context of biological cell division and dna. 2. frame of reference of nanotechnology the frame of reference of nanotechnology is understood as the context in which it develops and is implemented. this context is composed of interest groups. the interest group is understood as all those companies or people that participate in processes associated with the development of this technology. these participants can be classified into 6 groups: the state, the society, consumers, universities, engineers and scientists, and businesses. all these groups can express their opinions regarding this technology, besides developing it or simply being affected by the benefits or risks of nanotechnology. particularly, we should understand that this tool functions and operates in a context and the possible risks or benefits depend on the use that these groups give to it. the role of the interest groups in the development of nanotechnology: the state and businesses. they have the obligation to regulate and control the use of nanotechnology, as well as rewarding the good use, since this encourages the adequate utilization of this technology. engineers, scientists and universities should educate, conduct outreach, and invest in nanotechnology to develop awareness in the population and among students. the society and consumers are the recipients of this technology. the contribution of these groups is not only to observe the evolution of and enjoy the possible benefits that this technology, but also to be informed and express their opinions on these themes.9, 10 2.1 implementation process of nanotechnology the implementation of nanotechnology is composed of 3 stages: the initial phase that aims to evaluate the actual versus the expected, in order to analyze the differences between what is and what is desired in nanotechnology. the second stage is to coordinate planning and the possible mass evolution of this technology in several years. the development stage involves all the processes to implement nanotechnology. these include regulation and legislation, investment strategies. the third and final stage is related to results because it is here where we can see and measure whether the technological benefits outweigh the risks. published by atlantis press copyright: the authors 37 perception of risks in nanotechnology: determining key aspects in chile fig. 1. schematic diagram of the interest groups that make up the frame of reference for nanotechnology. 3. responsible nanoscience and its meaning some background: in 2008, the european commission recommended a code of conduct for responsible nanosciences and nanotechnologies research. this is one of a growing number of codes of conduct being proposed for nanotechnology. in the case of the european union (eu), the main goals are commendable. it seeks a relationship among everyone involved in academic research and industry. the uk government recently issued the universal ethical code for scientists, which considered the following general principle “ensure that your work is lawful and justified”, and one injunction to “minimize and justify any adverse effect your work may have on people, animals and the natural environment”.11 3.1 responsible nanotechnology (rn) responsible nanotechnology is a relatively new concept that refers to a responsible management of potential risks, and the promotion of the benefits on behalf of the human kind. the potential of the nanotechnology from a social point of view is immense, as are the potential risks due to irresponsible use or management. the academic community that defends the concept of responsible nanotechnology pursues a vision of the world in which molecular production is utilized for beneficial and productive purposes, and in which evil uses of their potential is prevented by an efficient management of the technology. in the united states, the center for responsible nanotechnology (crn) promotes debate related to the responsible management of nanotechnology. to offer complete and precise information, clear explanations and feasible proposals, crn studies all aspects of nanotechnology and its impact. this includes technological, social, military, economic, and political sides. this center presents the results of its studies to specialized public, but also to the general public, and tries to present the information in the most effective way possible. the objective of crn is to investigate the ethical, legal and social implications of molecular production and to educate the people so they influence its development and applications.12 4. some risks associated with nanotechnology it is important to note that in october 2004 the uk health and safety executive estimated that more than 10,000 workers dealing with nanoparticles could be exposed to some risks, in addition to not using effective methods of protection to avoid ingestion and inhalation of or dermal exposure to nanoparticles in the production process. in 2005, the u.s. national institute of occupational safety and health reported that significant damage was found in the dna of the heart and arteries of mice exposed to carbon nanotubes. in the same year, a nasa study reported that the injection of commercially available carbon nanotubes (equivalent to 17 days of exposure for a worker) caused significant damage in the lungs of rats. researchers at the university of rochester reported that rabbits subjected to inhalation of carbon nanospheres showed increased susceptibility to blood clots. at the meeting of the american chemical association, a report13 was presented that showed that carbon nanoparticles dissolve in water and that even very small concentrations are toxic to soil bacteria. this caused alarm about possible interactions with natural ecosystems. in 2003, a study published in nature14 showed that nanoparticles can be absorbed by earthworms and other soil organisms, with the possibility of moving up the food chain. therefore, there is concern about the risks of nanoparticles in humans, but these risks will only be known over time. to date no real impact is known, othpublished by atlantis press copyright: the authors 38 j. pozo, a. schmessane, a. pozo, c. pozo er than several studies that suggest certain dangers. these risks are perhaps no greater than those of existing industries15. 4.1 risk factors in the context of the responsible nanotechnology (rn) and to facilitate its comprehension as a model to minimize the risks associated with nanotechnology in chile16, we developed a model that could be helpful for other countries as well. a mathematical relationship that incorporates the most prominent factors to consider is proposed in this work. this relationship takes into consideration the following fundamental aspects:  legislation (l)  funding for research (fr)  sanctions for undue use (suu)  state control (sc)  good practice rewards (gpr) as such, the model is represented by the following equation17: ( ) ( ) ( ) ( ) ( ) l sr puu sc gpr rn l w sr w puu w sc w gpr w = + + + + (1) where: wl, wsr, wsuu, wsc and wgpr are their respective weights. why do we consider these aspects and not others? the answer is simple, because investment in this field in chile by the private sector is low. thus, public funding of research is very important. technology in chile is based principally on imports, so state control is important. the question is the following: how to control the revenues for safe technologies in the country? we think the above can be obtained by differential taxes and funding. 5. nanotechnology in chile to contextualize the situation of nanotechnology in chile, it is important to consider current legislation or regulations as a reference framework, that is to say, a methodology that describes and discriminates between the correct and the incorrect. it would be desirable that regulation for the use of the nanotechnology is as detailed as possible, without leaving room for different interpretations, or in other words, that it has a clear meaning that allows for common understanding. government funding will also be important because it will allow for the development of new topics in research and development and innovation (r&d&i) in nanotechnology, which is lacking in chile and is of an immediate need.18 chile must identify nanotechnology as a strategic sector of development, and must consider a number of initiatives in order to maximize profits, while minimizing their risk. others studies19, 20 have shown the importance of public perceptions for the successful development and implementation of new technologies. perceptions can modify the direction, and behavior of scientific activity in many fields of research and technologic development. a relevant example is nuclear energy. to identify public concerns regarding nanotechnology, we conducted a survey to detect perceptions and knowledge about the risks and benefits of nanotechnology. this survey included questions about its regulation, and the possible impact and consequences of this technology in chile. because there are no previous reports regarding chilean public opinion on this matter, our first approximations were based on surveys directed to two discrete groups; the general public, and university students and professors (academic world). the intention is to find differences between the two groups, because science and technology at the nanoscale is studied in chile only in the academic world by research groups in few universities. the issues contemplated in the survey are the ingredients for responsible nanotechnology in chile, the proportion of them in the equation respond principally to the cultural framework in chile and how chilean society responds. published by atlantis press copyright: the authors 39 perception of risks in nanotechnology: determining key aspects in chile 5.1 chile as a competitive country in 2020 chileans should be aware of what is happening with new technologies to be better prepared for future challenges. with regard to nanotechnology, chile was selected by the world bank as pioneer by the millennium science initiative, which has been funding some projects in nanotechnology. 5.2 human resources in chile and s & t policies human resources are one of the predominant factors in the scientific and technological development of a country. a poor educational policy or its delay may result in the loss of international competitiveness. 6. opinion survey to evaluate the status of nanotechnology in chile, we conducted an opinion survey. detailed information on the specific questions and the complete results can be found in21. the survey was applied to different representative sectors of chilean society represented by 200 people divided into two groups of 100 each. the first group included the general public (private and public sector), and the second consisted of university students and professors. the survey had 15 questions, grouped as follows: 4 questions related to perceptions about nanotechnology, 3 related to the respondent’s knowledge of nanotechnology, 2 about personal opinions, 3 about risks associated with nanotechnology, and the last 3 about responsibility at the country level. 6.1 survey results there was no difference between the two groups, which allowed us to emphasize the following aspects that are prominent21:  perception although the majority of respondents had heard about nanotechnology and respondents were interested in knowing more about it21, they considered that the level of development and investments made in chile in this field are low (fig. 2). fig. 2. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question.  knowledge of its applications in relation to the applications of nanotechnology, close to 50% of respondents thought that the applications are in informatics, 24% in medicine, 20% in development of products and materials and 8% in foods (fig. 3). most of the respondents also thought that the level of knowledge about the theme depends more on motivation and personal interest than educational level and socioeconomic status.21   fig. 3. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question.  opinion almost all respondents stated that no regulations exist in chile (fig. 4). 70% stated that public opinion should be published by atlantis press copyright: the authors 40 j. pozo, a. schmessane, a. pozo, c. pozo considered in developing nanotechnology in chile (fig. 5). fig. 4. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. fig. 5. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question.  associated risks regarding potential risks, 55% thought that nanotechnology can harm both people’s health and the environment, with about 30% thinking that it can be harmful for peoples’ health and 15% concerned about environmental risks (fig. 6). nevertheless, 90% of respondents in both sectors stated that they did not know what risks may be associated with handling nanomaterials (fig. 7). besides, more than 90% of respondents considered that nanotechnology may improve global competitiveness (fig. 8). fig. 6. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. fig. 7. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. fig. 8. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. published by atlantis press copyright: the authors 41 perception of risks in nanotechnology: determining key aspects in chile  responsibility at the country level on the other hand, close to 90% of respondents considered that the development of nanotechnology is a pending challenge for chile (fig. 9). regarding who should finance research, 44% identified the private sector, 42% the state and 14% universities (fig. 10). fig. 9. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. fig. 10. a sequence of responses that involves percentages versus number of people for the two groups (general public and students/professors) in connection with the question. 7. conclusions the two groups (general public, and university students and professors) tended to give similar answers to all the questions. it is clear that the discussion in chile on nanotechnology is incipient and that the same level of information is available to everyone. based on this survey, it should be the responsibility of the chilean government to propose legislation in science and related technology in the context of research and development and innovation (r&d&i) that establish the norms to regulate the development of nanotechnology in chile. funding for research in nanotechnology should be provided by both the government and large chilean businesses. additionally, it is necessary to emphasize that the universities and institutes should be responsible to for developing adequately trained engineers and scientists with the ethical disposition to face the challenges of the future in this matter.5 based on the information and the results generated in this work, and with the help of decision tree 22, it is possible to assign relative weight factors (%) to each parameter in the equation (1). given that chile does not have legislation yet (l), and sanctions for the undue use (suu) to start is also important, both parameters are weighted with 30% followed by funding for research (rf) with 20%, state control (sc) and good practice rewards (gpr) both with 10% each. thus, the suggested model is as follows: (30%) (20%) (30%) (10%) (10%) rn l sr puu sc gpr = + + + + (2) this equation may be useful to acquire preliminary quantifiable baseline information. as an emerging field, the development of nanotechnology, by both the private and public sectors, should be undertaken with consideration of social responsibility. another aspect can also be considered in our model as the self-regulation by the society and the responsibility of the private sector. nevertheless the last aspect in our model gpr is zero, because the development of science and technology in chile is low. we think that any further development of this technology in chile should involve adequate legislation. we conclude that the main risks associated with nanotechnology relate to the lack of legislation and/or regulations in chile in the area of science and technology, published by atlantis press copyright: the authors 42 j. pozo, a. schmessane, a. pozo, c. pozo which results in a lack of leadership, and therefore a deficit in the construction of knowledgement in term of what are the determining factors in global economic dynamics. there are few specialists in chile to address the issue. finally, there should be a communication strategy to educate citizens, which is of vital importance to establish public policies and regulations on the subject. references 1. m. siegrist, c. keller, h. kastenholz, s.frey and a. wiek laypople’s and experts’ perception of nanotechnology hazards, (risk analysis, vol. 27, no 1, pp. 59-69, 2007). 2. theories and applications of risk analysis and crisis response the first international conference on risk analysis and crisis response, (september 2526, 2007, shanghai, china). http://racr.shmtu.edu.cn/en_index.asp 3. chi-fai chau, shiuan-huei wu and gow-chin yen the development of regulations for food nanotechnology (trends in food science & technology 18, 269-280, 2007). 4. k. thomas, p. sayre research strategies for safety evaluation of nanomaterials, parts i: evaluating the human health implications of exposure to nanoscale materials (toxicological science 87 (2) pp, 316-321, 2005). 5. j. pozo, a. schmessane, g. estay los ingenieros chilenos la nanotecnología y las estrategias económicas a futuro (actas cd xxiv congreso chileno de educación en ingeniería, sochedi 2010). 6. a. pedreño muñoz nanotecnología y economía: estrategias de futuro (instituto de economía internacional. universidad de alicante. url) http://iei.ua.es/nanotecnologia/nanotecnologia-yeconomia-estrategias-de-futuro 7. r. p. feynman, (engineering and sciences vol. 23, 22-36, 1960). 8. azom.com nanomaterials and their applications (nov. 23, 2001). http://www.azom.com/article.aspx?articleid=106 6 9. igor linkov et. al. emerging methods and tools for environmental risk assessment (journal of nanoparticle research 11: 513-527, 2009). 10. management principles for nanotechnology (nanoethics 2:43–60, 2008). 11. r. jones are you a responsible nanoscientist? (nature nanotechnology 4, 336 2009)doi:10.1038/nnano.2009.127 12. center for responsible nanotechnology 20012008 http://www.crnano.org/ 13. a. johansen et al. effects of c60 fullerene nanoparticles on soil bacteria and protozoans (article first published online: 9 dec 2009). doi: 10.1897/07-375.1 14. g. brumfiel, nanotechnology: a little knowledge (nature, vol. 424 pp. 246-248, 17 july 2003). 15. d. tomanek, the nanotube site 23 nov. 2005. http://www.pa.msu.edu/cmp/csc/nanotube.html 16. j. pozo and a. schmessane nanomaterials, the risks and benefit for the society and the environment (proceedings i latin american congress sra-la 2010, page 33). 17. parlamento europeo, comisión de de industria investigación y energía. sobre nanociencias y nanotecnologías: un plan de acción para europa 2005-2009. http://www.europarl.europa.eu/sides/getdoc.do?pu bref=-//ep//text+report+a6-20060216+0+doc+xml+v0//es 18. m. c. roco international perspective on government nanotechnology funding in 2005 (journal of nanoparticle research, 7: pp. 707-712, 2005). 19. m. cobb and j. macoubre public perceptions about nanotechnology: risks, benefits and trust (journal of nanoparticle research 6: 395-405, 2004). 20. s. currall, e.king, n. lane, j. madera and s. turner what drives public acceptance of nanotechnology (nature nanothecnology vol i december 2006). 21. j. pozo a. schmessane, a. pozo y c. pozo. nanotechnology survey chile 2011. http://nanotecnologiasurvey.blogspot.com/ 22. h. arsham herramientas para el análisis de decisión: análisis de decisiones riesgosas. http://home.ubalt.edu/ntsbarsh/opre640s/spanishp. htm published by atlantis press copyright: the authors 43 received 11 december 2015 accepted 15 february 2016 economic fluctuations and insurance industry crisis: the history, status quo and the future zhenhua guo school of finance, shanghai university of international business and economics, shanghai 201620, china abstract the healthy development of insurance industry influenced the stability of country's financial system. in recent years, the insurance industry has developed rapidly in china, but also hid the huge risks in it. as the economic growth rate slow down, china’s insurance industry may also encounter with crisis that japanese and taiwan has faced. in this paper, by analyzing the reasons and mechanisms of insurance industry crisis in japan and taiwan, found the insurance industry crisis closely related to the economic fluctuation, and provided some warnings and references to the insurance industry of china. keywords: economic fluctuations, insurance industry crisis，interest margin loss 经济波动与保险业危机：历史、现状与未来 郭振华 上海对外经贸大学金融学院，上海，201620 摘要：保险业的健康发展关乎国家金融体系稳定，近些年来，我国保险业发展迅速，取得了骄人的成绩， 但在成绩的背后也隐藏了巨大的风险，随着经济增长减速，我国保险业也可能遭遇日本，台湾曾经遭遇过的保险 业危机。本文通过梳理日本、台湾经济过的保险业危机，探讨其深层次原因与机制，发现其与经济波动的密切联 系，为我国保险业提供一些警示与借鉴。 关键字：经济波动、保险业危机、利差损 1. 中国保险业发展概况及保险业特征 1.1 中国保险业发展简述 自 1980 年国内保险公司复业以来，我国保险业 发展迅速，从一个基础薄弱的行业逐步成长为一个 关系国计民生的重要行业。2014 年财险保费收入达 到 7544.4 亿，相比 2002 年，增长了近 10 倍；寿险 保费收入达到 12690.28 亿，是 2002 年寿险保费收入 的 5.58 倍。截至 2014 年底，财险业资产规模达到 14061.48 亿元，相比 2002 年增长了 14.9 倍；寿险业 资产规模达到 82487.2 亿元，相比 2002 年增长了 16 倍。我国财产保险业和寿险业保费和资产增长态势 见图 1 和图 2。 1.2 我国保险业特征 保险业有一些特有的特征，这些特征也决定了 保险业的风险特点： 数据来源：wind 数据整理 图 1 中国财产保险业保费收入与资产规模 journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 38-46 published by atlantis press copyright: the authors 38 zhenhua guo / economic fluctuations and insurance industry crisis 1）财险负债期限短、寿险负债期限特长 保险是负债经营，由于财险产品和寿险产品在 保险期限上的巨大差异，导致财险业和寿险业的负 债期限有着天壤之别。我国财产保险公司经营财产 保险业务、短期意外险和短期健康险业务，承保期 限一般都在一年之内，导致财险公司的负债期限短。 寿险公司经营人寿保险、健康保险、年金保险和意 外伤害保险，承保期限多在 5 年以上，很多产品的 保险期限都是几十年甚至终身，导致寿险公司负债 期限特别长。 正是由于财险业负债期限短，寿险业负债期限 长，导致寿险业的“资产规模/保费收入”远超财险 业，我国寿险业 2014 年底的资产规模约为当年保费 收入的 7 倍，而财险业 2014 年底的资产规模约为保 费收入的 2 倍。 2）保险业保费增长速度通常远超 gdp 增速 世界各国的发展经验都表明，在人均 gdp 达到 一定规模（大约 2500 美元）后，保险业保费增长速 度就会以远超 gdp 增速的速度快速发展，我国也是 如此，借势于中国经济的快速发展与居民财富的大 量积累，过去 30 多年，我国保费的年平均增长速度 达到 20%以上。 3）我国保险业尚未经历过大的危机 我国保险业一路快速发展，资产规模与保费收 入不断增长，但不可忽视的是，我国保险行业长期 以来都处在保监会等国家监管机构的严格监管下， 尚未遭遇到过特别大的危机，无论是与国外同行相 比，还是与我国银行业相比，我国保险业在风险控 制方面还有许多不足。 2． 日本经济危机与日本保险业危机 1990 年代，由于日本经济的大幅波动，日本保 险业经历了一次大的危机。首先分析日本保险业的 资产负债表的潜在风险，然后分析 1990 年代的日本 经济危机对保险业资产负债表的影响，观察经济波 动对保险业的冲击。 2.1 保险业的资产负债表及其潜在风险 由于资料收集困难，无法收集到 1990 年左右日 本保险业的资产负债表，表 1、2 列出了 2000 年时 日本寿险业和非寿险业的资产负债表（参见附件表 1、表 2）。 可以看到，从资产端来看，日本保险业持有大 量金融资产和房地产，而这些资产的价值与宏观经 济状况息息相关，如果遇到经济危机，保险公司持 有的公司债、公司贷款可能面临违约，股市大跌会 迅速降低保险公司持有的股票价值，房价和土地价 格大跌则会迅速降低保险公司持有的房地产的价 值。从负债端来看，如果遇到经济危机，可能会出 现长期的市场利率下行，而长期利率下行会降低保 险负债的准备金评估利率，从而大幅抬升保险业的 准备金负债规模。当然，从有利的方面来看，市场 利率下行也会提升保险公司的持有的债券的价值。 总体来看，经济危机可能严重打击保险业，尤 其是杠杆更大的寿险业的资产负债表，促使保险业 资产缩水、负债增加，进而造成净资产下降甚至成 为负值，偿付能力严重恶化，甚至出现资金链断裂 导致破产倒闭。 2.2 1990年代的日本经济危机 1980 年代，日本出口大幅增长，经济快速增长， 对美形成巨额顺差，1984 年底日美顺差超过 1000 亿 美元。1985 年 9 月，美英法德日签订“广场协议”， 迫使日元升值，1988 年，日元兑美元从 238:1 升值 为 128:1。 日元升值造成的后果包括：1）出口迅速下降， 带动经济增速下降，日本采取货币宽松政策，大幅 数据来源：wind 数据整理 图 2 中国寿险险业保费收入与资产规模 published by atlantis press copyright: the authors 39 zhenhua guo / economic fluctuations and insurance industry crisis 下调利率，大量资金进入房地产和股市，日本股价、 房价高涨；2）日元升值后，日本企业大举海外投资、 购买美国资产； 3）广场协议同意美元贬值后，在 日元升值过程中，大量国际资本进入日本，购买日 本房地产，进一步促使房价高涨。结果，1985 年后 的 5 年间，日本股市市值增加 3 倍，房地产价格上 涨 2.4 倍。 4）1991 年，国际资本获利撤离，泡沫 破裂，股市、房市大跌，经济增速大幅下降。之后 的 1997 亚洲金融危机又使日本经济雪上加霜。 如图 3 所示，日本经济在经历 1980 年代的高速 增长后迅速下滑，从 1992 年开始，经济增速一直在 零增长附近徘徊，走入了失去的 25 年。在日元贬值、 日本经济崩盘后，日本的股市、住宅价格、土地价 格等无一例外地在同一时间段大幅下跌。图 4 显示 了日元汇率贬值情况，图 5 显示了日本股市日经指 数变化情况，图 6 显示了日本城市住宅价格指数变 化情况，图 7 显示了日本土地价格指数变动情况。 数据来源：wind 数据整理 图 3 日本 gdp 增长率（%）变化 数据来源：wind 数据整理 图 4 日本外汇汇率:美元兑日元 数据来源：百度网络 图 5 日本股市日经指数 数据来源：wind 数据整理 图 6 日本城市住宅价格指数 数据来源：wind 数据整理 图 7 日本土地价格波动 published by atlantis press copyright: the authors 40 zhenhua guo / economic fluctuations and insurance industry crisis 2.3 经济危机带来保险业危机 首先看对日本保险业资产端的影响：1990 年代 日本经济接近零增长，日本的股市、住宅价格、土 地价格等无一例外地在同一时间大幅下跌，大量企 业经营状况恶化。结果，保险公司持有的公司债、 公司贷款出现违约或形成不良资产，保险公司持有 的股票价值大幅缩水，保险公司持有的房地产的价 值也大幅缩水。但也有有利因素，即利率下行造成 保险业持有的固定收益类资产价值上升。总体来看， 日本保险业的资产规模有所下降，但规模不太大。 再看对负债端的影响：经济危机发生后，日本 采取了货币宽松政策，试图刺激经济上行，但经济 还是由于人口老龄化、技术创新瓶颈等原因持续低 迷，造成了市场利率的长期低位运行。而长期利率 下行会给寿险业带来严重的利差损问题，或者称为 准备金评估利率降低造成的负债大幅增加问题。由 此，日本寿险业的负债规模由于利率下行而大幅升 高。 资产略微缩水和负债大幅增加，使日本寿险业 的净资产大幅缩水，如表 1 所示，到 2000 年时，日 本寿险业的净资产居然只有总资产的 3.2%。而在健 康运行状况下，寿险业的净资产大致在总资产的 8% 左右。其中的一些寿险公司已经资不抵债，面临破 产。 1997 年,日本二战后倒闭了首家保险公司—— 日产生命保险公司，随后，东邦生命保险公司、第 百生命保险、第一火灾海上保险、大正生命保险、 千代田生命保险、协荣生命保险、东京生命保险、 大成火灾等多家保险公司相继破产，引起了日本金 融界的强烈震动。表 3 概述了 2000 年左右日本寿险 业倒闭的 7 家寿险公司的基本情况（参见附件表 3）。 有些意外的是，同期还有少量财险公司也破产 了，如第一火灾海上保险和大成火灾保险。 意外的 原因是因为财险公司通常自只做短期保险业务，不 存在承诺客户保证收益率的问题。表 4 是第一火灾 海上保险的发展历史（参见附件表 4）。 2.4 剖析日本保险业利差损问题 上述研究发现，经济危机对保险业的打击有资 产端的影响，更有负债端的影响，集中体现为保险 业的巨额利差损问题。 日本的经济在二战后开始高速增长，每年增长 高达 10%以上，经济起飞和高速增长阶段必然面临资 金短缺，所以市场利率较高，因此这时期内日本保 险业销售的寿险产品大多利率很高，高利率为保险 公司带来大量保费的同时，也为日后的危机埋下了 隐忧。到了 90 年代，经济增长乏力，gdp 增速甚至 出现了负增长，资金不再短缺，市场利率接近零利 率，寿险业产生了严重的利差损问题。 图 8 为日本利率的历史变化情况，表 5 为不同 年份日本寿险业的预定利率（1952—2002）（参见附 数据来源：wind 数据整理 图 8 日本存贷款利率变化（1960-2014） 数据来源：周国端.保险财务管理：理论、实务、 案例，中信出版社，2015 图 9 日本寿险业资产投资回报率变化 published by atlantis press copyright: the authors 41 zhenhua guo / economic fluctuations and insurance industry crisis 件表 5），图 9 为日本寿险业资产投资回报率的变化 情况（1991-2002）。由于寿险产品是长期产品，长 期意味着承诺客户的固定回报率是长期固定的，而 在保险期限内利率逐渐下行造成寿险公司资产投资 回报率下降后，必然会给寿险业带来利差损问题。 前期保证利率产品销售越多，后期利率下行程度越 严重，后期利率下行或低位运行时间越长，利差损 规模就会越大。 日本寿险业的利差损问题实际在 1980 年代利率 下行后就出现了。为了弥补利差损，1980 年代，保 险公司大量投资于回报率高的股市和房地产，在 20 世纪 80 年代末 90 年代初，由于日本实体经济不振， 在宽松的货币条件下，大量资本流入股市，房地产， 使得价格远远高出了合理价位。广场协定后，股市 市值以每年 30%速度增加，土地价格每年上涨 15%， 而同时期日本的名义 gdp 增长只有 5%左右，泡沫越 来越大，股价不断上涨，房地产价格也不断走强， 保险公司便寄希望于依靠股市和房地产的投资弥补 利差损，但 1991 年国际资本流出后，随之泡沫破裂， 股价一落千丈，日经指数直到今天也没有再达到过 当初的高位。房地产价格也发生崩盘，2005 年房价 竟然与 20 年前的房价基本一致。随着股市、房价的 暴跌，保险业不仅没有达到弥补利差损的目的，反 而使得资产进一步贬值，引起更大的危机。 此外，第一火灾海上保险公司是一家相互制保 险公司，历史悠久，本不应该出现利差损问题。但 是，如表 4 所示，该公司为了实现迅速增长，从 1963 年开始销售长期的储蓄性火灾保险作为主打产品， 并承诺了高额的利率，这就为日后的破产埋下了隐 忧。伴随着日本高速的发展，资产也不断增加，到 1992 年，资产达到 1000 亿日元，但其保费收入中， 两款高回报率承诺的长期储蓄产品占到 80%，到 2000 年，利差损有 120 亿日元之巨，在金融监管厅下令 停止一部分业务后，资产持续缩水，负债不断加大， 到破产时，已经资不抵债，负债 12706 亿日元，股 东权益为-1246 亿日元。 可见，无论是寿险公司还是财险公司，利差损 问题是日本保险业危机的主要原因，但根源是宏观 经济的大幅波动及保险业的错误预期。 3. 台湾经济低迷与台湾保险业危机 3.1 台湾2000年代的经济低迷 作为亚洲四小龙之一的台湾在战后发展迅猛， 出口导向型的政策与日本极为相似，gdp 增长率在 20 世纪 60 年代平均为 10%，70 年代平均为 9.4%， 80 年代平均为 8.1%，90 年代平均为 6.3%。台湾 gdp 增长率变动情况见图 10。 21 世纪初，台湾经济增长率下降，出现 2 次负 增长。2001 年，美国互联网泡沫破裂，全球经济开 始走缓，以出口电子产品为主的台湾经济受到波及， 出口负增长 17.1％，电子产业为龙头的岛内工业衰 退，产业大量转移到其他地区。经济下行的同时， 利率也开始不断走低，金融机构普遍遭遇困境，据 当时穆迪估计，台湾银行实际呆坏帐率应高于 15％， 总呆坏帐金额可能高达 600 亿美元，占 gdp 的 20％。 台湾利率变化情况见图 11。 数据来源：wind 数据整理 图 10 台湾 gdp 增长率变化（不变价） 数据来源：wind 数据整理 图 11 台湾存贷款利率变化 published by atlantis press copyright: the authors 42 zhenhua guo / economic fluctuations and insurance industry crisis 3.2 经济低迷对寿险业的影响 与日本情况类似，台湾寿险业在经济增速较快、 利率水平较高的时期销售了大量高预定利率保单， 然后，在经济下滑的 2000 年代，在市场利率不断走 低的情况下，台湾保险业开始出现利差损问题。台 湾寿险业新保单平均预定利率与保险资产收益率如 图 12 所示，可以看到，与市场利率变化水平基本一 致，台湾寿险业资产投资收益率从 1990 年代中期开 始逐渐下滑，由此一定会造成长期固定回报类保单 的利差损问题，1990 年代中期，台湾寿险公司预定 利率保单的承诺利率高达 7.8%，而 2004 年寿险业的 投资回报率在 4%左右，利差损在 4%左右。 与日本保险业危机时不同，在台湾此次危机中 有大量的外资保险公司出逃。按照《国际会计准则 第 4 号》的规定，欧洲寿险公司财务报表必须采用 公允价值计量。于是，在台湾经营的欧洲寿险公司 必须为利差损增提大量准备金，导致财务状况恶化， 必须增资补充资本。但是，在 2008 金融危机的打击 下，其母公司也已深陷困境，无法增资。于是，大 量外资保险公司从台湾撤走了。2009 年 2 月，ing 以 195 亿元新台币将台湾 ing 安泰人寿出售给了富 邦金控；2009 年 6 月，英国保诚以 1 元新台币将台 湾保诚出售给了中国人寿；2009 年 8 月，aegon（荷 兰全球人寿）集团以 31 亿元新台币将其台湾子公司 出售。成交价都远低于实际价值。 台湾当局为了缓解台湾当地保险业的财务困 境，消除利差损问题，允许保险公司开展 6 年期储 蓄业务，由于寿险公司能够提供高于银行存款利率 的投资收益率，寿险公司通过大力发展 6 年期左右 的储蓄产品，吸引到了大量的资金，被称为“银行 存款大搬家”。 4.中国经济波动与保险业风险：分析与建议 我国的保险业虽说已经取得了巨大的成功，但 是发展时间短，尚未经历过大危机的洗礼，使得我 国保险业可能会忽视种种风险。从经济增长速度和 市场利率变化来看，我国也极可能面临经济走缓， 利率低位运行的境地，我国保险业应该早做准备， 有效应对可能出现的低增长、低利率时代带来的冲 击。 4.1 我国保险业1990年代末的利差损问题 我国保险业从上世纪 80 年代开始复业，经过 30 多年的发展，取得了堪称世界瞩目的成绩。但发展 并非一帆风顺，在上世纪 90 年代，从 1992 年开始， 中国经济进入快速增长通道，银行存贷款利率很高， 保险公司据此错误预测了未来利率水平，销售了大 量长期保证高利率的保险产品，最高承诺利率高达 复利 8.8%，但随后银行存贷款利率不断下降，保险 公司的投资收益率也随之下降，寿险公司在 2000 年 之后的投资收益率平均水平在 5%左右，导致了利差 数据来源：wind 数据整理 图 13 中国 gdp 指数 数据来源：wind 数据整理 图 12 台湾寿险业新保单预定利率与 保险资产收益率 published by atlantis press copyright: the authors 43 zhenhua guo / economic fluctuations and insurance industry crisis 损问题。图 13 是中国 gdp 增长变化情况，图 14 是 中国银行存贷款利率的变化情况，图 15 是中国最大 寿险公司——中国人寿 2003 年以来的资产投资收益 率变化情况。 从 1999 年开始，保监会将寿险产品预定利率一 直维持在 2.5%的水平，部分意图就是为了消化 1990 年代末高利率保单的利差损。 4.2 我国保险业潜在的利差损问题 近些年来，全球经济不断减缓，我国经济增长 减速严重，市场利率逐渐下行，低利率时代渐行渐 近。 我国保险业利率设定问题上逐步向市场发展， 将决定权交由市场。2003 年普通寿险预定利率放开， 2015 年万能险、分红险保底利率上限放开，随着利 率的放开，势必会引发保险业的利率竞争，导致寿 险业承诺利率水平上升。例如，2014 年的开门红产 品，多数保险公司主打的年金类产品，采用了保监 会规定的最高利率上限 4.025%。但随着经济由高速 增长转向低速增长，市场利率水平必然会由高到低 发展，长期高利率产品几乎必然会产生利差损。 从我国保险公司经营范围来看，主要风险在寿 险业，因为寿险业经营长期保险，负债期限长。但 并非财险公司就没有利差损的风险，因为许多财险 公司也都推出了“寿险型”的保险产品，而这些产 品将会像寿险公司产品一样遭受利差损的威胁，所 以，不得不警惕“寿险型”财险公司的利差损风险。 4.3 对中国保险业建议 针对保险公司可能面临的风险，结合日本、台 湾保险业发展的相关经验，我们提出以下建议： 首先，高利率保险产品要短期化。由于未来利 率很难预测，保险公司应尽量避免开发销售长期保 证高利率产品。在市场利率水平较高时，如果希望 通过高保证利率产品占领市场，应该根据对宏观经 济、金融市场的预测来推出短期化的高收益产品， 因为高利率往往是不可持续的。 其次，对于长期保险产品，一定要采取低预定 利率。如前所述，长期高保证利率产品几几乎必然 会面临利差损问题，所以，如果要开发销售长期保 险产品，一定要将其设计成保证利率较低的、浮动 利率的分红型或万能型产品。 最后，虽然寿险业风险远大于财险业，但也要 警惕“寿险型”财险公司的利差损风险。 参考文献 [1].胡文富，日本人寿保险市场概况.保险研究，2001,(1):47. [2]孙立娟，梁斯，朱向辉，非理性繁荣:日本保险业危机回 顾及对我国的启示.南方金融，2014,(9):75-79. [3]水口啓子 .変わる生保 消える生保 :よくわかる「生 保」大選別時代の新基準，東洋経済新報社，2002. [4]池晶，论日本保险业的危机、对策及启示.东北亚论坛， 2001,(4):69-72. 数据来源：wind 数据整理 图 14 中国银行部门存贷款利率 数据来源：wind 数据整理 图 15 中国人寿的投资收益率变化 published by atlantis press copyright: the authors 44 zhenhua guo / economic fluctuations and insurance industry crisis [5]冯玉梅，日本保险业危机与中日保险业对比.现代日本经 济，200l,(3):16-21. [6]于超银，日本保险公司破产对我国保险业风险防范的启 示. 现代日本经济，2001,(3):26-31. [7] 胡 坤 ， 日 本 保 险 业 的 危 机 与 启 示 . 金 融 研 究 ， 2000,(11):100-108. [8]吉玉荣，张爱红，张维，日本保险业危机对我国保险监 管的启示. 南京审计学院学报，2007,(2):36-39. [9]周国端，保险财务管理：理论、实务、案例. 中信出版 社, 2015 [10] 孟昭亿，保险资金运用国际比较. 中国金融出版社， 2005 （本文受国家自然科学基金面上项目“小概率大损 失风险投保不足的实验研究”（批准号：71173144） 的资助。） 附件： 表 1 日本寿险业资产负债表（2000 年） 资产： 负债： 国内有价证券： 57.6% 政府债券 16.6% 公司债券 9.3% 股票 15.4% 贷款： 26.1% 政府贷款： 2.5% 公司贷款： 17.2% 现金+命令贷款： 5.3% 房地产： 4.3% 国外贷款： 2.1% 准备金： 90% 价格波动储备： 0.4% 其他负债： 9.6% 股东权益： 总资产的 3.2% 数据来源：孟昭亿，《保险资金运用国际比较》，中国金融出版社，2005 表 2 日本非寿险业资产负债表（2000 年） 资产： 负债： 存款： 6% 命令贷款： 1.7% 证券： 62.3% 日本中央政府债券 5.4% 地方政府债券 4.0% 公司债券 13.4% 股票 26.8% 国外证券 11.4% 其他证券 1.3% 贷款： 11.6% 房地产： 4.7% 其他资产： 13.7% 准备金： 83.7% 其他负债： 16.3% 股东权益： 总资产的 18.2% 数据来源：孟昭亿，《保险资金运用国际比较》，中国金融出版社，2005 published by atlantis press copyright: the authors 45 zhenhua guo / economic fluctuations and insurance industry crisis 表 3 日本寿险公司破产概况 单位：亿日元 破产公司 日产生 命 东邦生命 第百生命 大正生命 千代田生 命 协荣生命 东京生命 成立时间 1909 1898 1915 1914 1904 1935 1895 破产时间 1997.4 1999.6 2000.5 2000.8 2000.10 2000.10 2001.3 破产前资产总额 20.609 28.046 21.885 2.044 35.019 46.099 10.150 破产后资不抵债额 3.000 6.500 3.200 365 5950 6895 325 破产前预定利率 3.75-5.5 0 4.79 4.46 4.05 3.70 4.00 4.20 破产后预定利率 2.70 1.00 1.00 1.00 1.50 1.75 2.6 数据来源：水口啓子 .変わる生保 消える生保 :よくわかる「生保」大選別時代の新基準 [m].東洋経済新報 社，2002 表 4 第一火灾海上保险公司发展历史及破产情况 1949 年 第一火灾海上保险公司成立，属于相互制公司 1963 年 销售储蓄型火灾保险，保险期限多为 10 年，承诺高预定利率，为日本财产保险业界第一 款这样的产品； 该产品一直作为主打产品，引入营销员制度 1969 年 销售储蓄型分红意外伤害保险，保险期限 10 年甚至以上。 1972 年 总资产达 500 亿日元 1987 年 总资产达 5000 亿日元 1992 年 总资产达 10000 亿日元，两款储蓄型产品保费占比近 80% 1999 年 与协荣生命进行资本和业务合作：基金（相当于资本金）由 35 亿日元增至 416 亿日元。 1999 年 3 月 风险保费收入 597 亿日元，排名 17 位；但其总资产却排名第 8 位。 2000 年 利差损达 120 亿日元，金融监管厅下令停止一部分业务 2000 年 3 月 宣布破产。破产时公司：资产 11461 亿日元，负债 12706 亿日元，股东权益-1246 亿日元 2001 年 所有保险合同转移给财产保险契约者保护机构 数据来源：孟昭亿，《保险资金运用国际比较》，中国金融出版社，2005 表 5 不同年份日本寿险业的预定利率（1952—2002） 单位： % 保险期限 1952 1976 1981 1984 1990 1993 1994 1996 1999 2001 2002 <10 年 4.0 5.5 6.0 6.3 5.8 4.8 3.8 2.8 2.0 1.5 1.5−0.75 10-20 年 5.0 5.5 6.0 5.5 >20 年 5.0 5.5 数据来源：2003 年 2 月《周刊朝日》，申万研究 published by atlantis press copyright: the authors 46 1. 中国保险业发展概况及保险业特征 2． 日本经济危机与日本保险业危机 3. 台湾经济低迷与台湾保险业危机 4.中国经济波动与保险业风险：分析与建议 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /usedeviceindependentcolorforimages /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <feff004b00e40079007400e40020006e00e40069007400e4002000610073006500740075006b007300690061002c0020006b0075006e0020006c0075006f00740020006c00e400680069006e006e00e4002000760061006100740069007600610061006e0020007000610069006e006100740075006b00730065006e002000760061006c006d0069007300740065006c00750074007900f6006800f6006e00200073006f00700069007600690061002000410064006f0062006500200050004400460020002d0064006f006b0075006d0065006e007400740065006a0061002e0020004c0075006f0064007500740020005000440046002d0064006f006b0075006d0065006e00740069007400200076006f0069006400610061006e0020006100760061007400610020004100630072006f0062006100740069006c006c00610020006a0061002000410064006f00620065002000520065006100640065007200200035002e0030003a006c006c00610020006a006100200075007500640065006d006d0069006c006c0061002e> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice research article 基于前景理论的养老机构新型冠状病毒肺炎风险防控研究 research on the prevention and control of covid-19 risk in long-term care facilities based on prospect theory fengying zhang1, jianhui kong2,*, xiaofeng xie1, , kai xu3, zongfang zhou4 1west china school of nursing/west china hospital, sichuan university, chengdu 610041, china 2law school, southwest minzu university, chengdu 610041, china 3school of business, chengdu university, chengdu 610106, china1 4school of management and economics, university of electronic science and technology of china, chengdu 610054, china a r t i c l e i n f o article history received 06 april 2020 accepted 09 may 2020 keywords covid-19 risk assessment risk occurrence mechanism risk early warning mechanism long-term care facilities keywords covid-19 风险评估 风险发生机制 风险预警机制 养老机构 a b s t r a c t in the current complex and uncertainty of the risk of covid-19 in long-term care facilities, the elderly in the long-term care facilities are at the highest risk of being affected by covid-19. there is still uncertain relationship between the identification of this risk and the care management, caring behavior, cannot achieve management optimization. matrix analysis the change relation between the conditions for the covid-19 risk identification and risk behavior. the results show that there are n equilibrium points for the risk management of covid-19 in long-term care facilities. the complexity of the risk of covid-19 in long-term care facilities and the cognitive deviations caused by the limited rationality of managers and caregivers make the covid-19, the risk early warning system is not easy to converge to a stable point in the real environment, so as to achieve the best management status of the covid-19 risk in long-term care facilities. nursing staff behavior, risk management cost, risk loss and other parameters have different risk management effects for the covid-19. when the selected parameter is adjusted through the threshold, reflects the stability and selection tendency of the covid-19 risk management strategy. promote institutional managers and caregivers to adopt active risk management strategies: strengthen the awareness of institutional managers and caregivers about the complexity and uncertainty of risks; optimize the covid-19 risk management system, process and procedures; strengthen the research on risk accidents, serious consequences and effective prevention of covid-19; strengthen the review and supervision of the risk management of covid-19 and improve its value perception of the effective prevention of the risk of covid-19. 摘要 在当前养老机构新型冠状病毒肺炎风险复杂性与不确定性情况下,养老机构内年长者处于被感染的最高风险之中。探 索该风险的识别与养老机构管理及照护行为之间存在的不确定关系, 以期实现管理最优化。矩阵分析不同条件对新型 冠状病毒肺炎风险识别与风险行为转变间的关系。研究结果表明：养老机构新型冠状病毒肺炎风险预警管理存多个均 衡点,养老机构新型冠状病毒肺炎风险的复杂性和管理人员及照护人员的有限理性产生的认知偏差,使新型冠病毒肺炎 风险预警系统在现实环境中不易于收敛在稳定点,以达到养老机构新型冠状病毒肺炎风险最佳的管理状态。管理人员 行为、照护人员行为、风险管理成本、风险损失等参数对新型冠状病毒肺炎风险管理结构产生不同程度的影响,当选 择的参数调整通过临界值时反映出能影响新型冠状病毒肺炎风险管理策略的稳定性及选择倾向。促使养老机构管理者 和照护人员采取积极风险管理策略：加强养老机构管理者和照护人员对风险复杂性和不确定性的认知。优化新型 冠状病毒肺炎风险管理制度、流程和程序。加强养老机构新型冠状病毒肺炎风险发生、后果严重影响及有效防范的 研究。强化新型冠状病毒肺炎风险管理工作审查和监督,提高其对有效开展新型冠状病毒肺炎风险防范的价值感知。 © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author *通讯作者：孔建会,研究方向：风险管理。email: john@swun.edu.cn 作者简 介:张凤英,博士,副教授。研究方向：风险管理、护理管理及老年护理。email: zhangfengying@scu.edu.cn 1基金项目：四川省科技厅项目(2018zr0201)；国家自然科学基金项目(71871447) 1. 引言 高亮部分修改为：据世界卫生组织（who）数据显示，截至 2020年7月22日，全球168个国家和地区报告确诊新型冠状病 毒肺炎（以下简称covid-19）病例达1476.5256万，累计死 亡超过61万人 [1]。目前新型冠状病毒肺炎已知传播途径存 在人传染人,主要为飞沫、接触传播,covid-19的基本再生数 在2.8～3.3 之间,属于中高度传染性的传播疾病 [2]。who总 干事在1月30日宣布新型冠状病毒肺炎疫情构成“国际关注的 新型冠状病毒肺炎”,3月11日宣布新型冠状病毒为世界大流行 (pandemic)。长者和患有哮喘、 糖尿病、 心脏病等基础疾病 的人感染病毒的风险可能增加,长者最易感新冠病毒,也最容易 发展为重症的人群 [3]。许多国家在新型冠状病毒肺炎疫情 下,养老机构都有集聚发生感染的个案报道。新冠肺炎疫情期 间,养老机构及时合理的疫情管理对有效预防、及时控制和消 journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 37–42 doi: https://doi.org/10.2991/jracr.k.200714.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-4640-9800 http://creativecommons.org/licenses/by-nc/4.0/ mailto:zhangfengying%40scu.edu.cn?subject= https://doi.org/10.2991/jracr.k.200714.002 https://www.atlantis-press.com/journals/jracr%20 38 f. zhang et al. / journal of risk analysis and crisis response 10(2) 37–42 除传染病在养老机构发生、流行,保障长者的身心健康具有重 要意义。 1.1. 中国养老机构现状及分析 根据国家统计局数据显示,截止2016年底,中国60岁及以上老年 人口2.31亿,占总人口16.7%。截至2017年9月,中国养老机构总 数超过14.46万家,居住着老、弱、病、残、失能、半失能、高 龄老人,具有人员高度聚集、社会联系广泛及脆弱人群集中的 特点。养老机构在控制新型冠状病毒肺炎传播和保证长者身 心健康中肩负重大责任。 中国疾病预防控制中心发布的新型冠状病毒肺炎流行病学分 析,发现确诊病例年龄在30～79岁(86.6%),其中60岁以上的长者 病例数占31.2%(武汉为44.1%),整体粗病死率为2.3%,在≥80岁年 龄组的粗病死率最高为14.8%,长者可能面临更严重的被新型冠 状病毒感染的风险。意大利卫生部发布的分析报告显示新冠 肺炎患者死亡的平均年龄为81岁,80～89岁之间的死亡病例占 总死亡人数的42.2%,70～79岁之间的死亡病例占总死亡人数的 32%,50～59岁之间的死亡病例占总死亡人数的2.8%,90岁以上的 死亡病例占总死亡人数的14.1%,年龄越大,感染新冠肺炎之后 治疗难度越大。结合养老机构现状,长者常有合并基础病的情 况,包括高血压、糖尿病、心血管疾病和呼吸道传染病的长者 占有不同比例,有合并症的长者易于感染新型冠状病毒肺炎,且 病死率明显更高。 养老机构工作人员短缺严重,不仅在民营机构存在,公办机构或 医养结合的养老机构内也有类似问题。养老服务比其他工作 要更为繁杂,是一个服务群体庞大,需求比较多样,而且服务周 期较长。中国人力资源和社会保障部、民政部联合颁布《养 老护理员国家职业技能标准(2019年版)》,将从业人员的标准 从2011年版的“普通受教育程度”从“初中毕业”调整为“无学历 要求”,以吸纳更多人从事养老护理工作。对于低学历的从业者 群体,面对高风险的长者群体,防范和化解养老机构新型冠状病 毒肺炎的群发风险,存在相应不确定性。 养老机构总数庞大、长者普遍易感,专业养老照护队伍不足, 服务规范化水平偏低等问题突出,其新型冠状病毒肺炎风险 具有混沌特征。维护长者安全,提高养老机构管理质量是养 老行业关注的重要课题,护理安全更是养老机构管理的关键 所在。 1.2. 新型冠状病毒肺炎风险特征 风险是客观存在的事物可能结果的不确定性,是不利事件发生 的机会 [4]。20世纪80年代末至90年代初欧美管理者将风险 管理引入到医院管理中,形成医疗风险管理 [5]。新型冠状病 毒肺炎具有突发性、公共性、严重性、紧迫性、复杂性和易 变性的特点,且风险水平高、后果严重,并存在疫情下的养老机 构工作开展的各环节。实际上,在新型冠状病毒肺炎疫情下,只 要防控不当都可能发生新型冠状病毒肺炎风险及集聚发生的 风险,包括社区医院、诊所、家庭护理场所、养老院和敬老院 等。本文所称的新型冠状病毒肺炎风险是指因防控不当引起 养老机构内长者、照护人员客观存在的、潜在的新型冠状病 毒肺炎发生的疏忽、失误或过失,可能会导致长者患病、损害 或死亡事件,以及可能严重影响养老机构运行的事件,可由机构 管理、照护人员行为及环境设施等因素引起。 2. 养老机构新型冠状病毒肺炎风险防控研究 新型冠状病毒肺炎风险管理的首要任务是对高危风险的识别, 并组织消除或减少新型冠状病毒肺炎风险事件的发生,以最低 成本实现最大安全保障是最科学的风险管理方法 [6]。jones 等 [7]利用风险系统评估表,对存在高风险因素的老人在其床 头挂警示牌,给予安全处置。将当前存在的管理状况、风险因 素等数据全部录入计算机,建立养老机构最小数据集 [8],从而 为管理者提供信息和决策支持,使新型冠状病毒肺炎风险管理 更有效率。医疗失效模式与效应分析(failure mode effects analysis, fmea)可有效地在不利事件发生前觉察潜在错误,保 障患者安全,降低危险发生率 [9]。 新型冠状病毒肺炎风险管理重点是提高养老机构照护人员风 险意识和应对能力。leland等 [10]用mds 2.0评估量表对长者 护理风险发生与机构特征,包括管理流程、人力配置及床位数 等进行相关性分析以证明该评价体系的有效性。通过风险管 理,使养老机构新型冠状病毒肺炎风险在不同程度上得到转 化、分解、控制和有效控制。 2.1. 养老机构内新型冠状病毒肺炎风险发生与照 护人员间关系 养老机构是为长者提供生活照料及老年护理等综合服务的 场所,主要服务对象有失能、半失能老人 [11]。风险因素的 识别和确立有助于分析风险成因和机理,从源头上进行风险 管理 [12]。养老机构照护人员配备严重不足,民办养老机构 的养护比保持在1：15～1：20的范围内,远低于国际养老院1 ：4.5的标准 [13]。养老机构照护人员以小学文化程度居多, 知识面窄 [14],持证率仅在10%左右 [15],整体年龄偏大 [16] 。部分养老机构为节约开支,招聘人员少,待遇低,工作强度大 导致人员流失率增加 [17]。养老机构照护人员日常工作任 务重,重复性强,工作繁琐,压力大 [18]导致照护人员从业意愿 低,对工作满意度不高。中国专业化养老照护人员数量严重 不足,照护人员在非发达地区流失率达20% [19]。养老照护人 员的离职意愿与其工作满意度密切相关,满意度越低,离职意 愿越高 [20]。照护人员对工作不满意,很难正确对待长者的 护理需求。在新冠肺炎疫情影响下,面临新型冠状病毒肺炎 风险更加凸显,风险发生与照护人员行为间存在必然联系,将 面临更大的挑战。 2.2. 新养老机构内新型冠状病毒肺炎风险行为的 发生机理研究 风险决策行为是面向未来,而不是面向现在和过去,多数决策都 是在不确定条件下进行。20世纪以来风险管理决策领域由von neuman和morgenstern提出的期望效用理论认为期望效用值 可用备选方案结果出现的概率与该方案效用值的函数来表示 [21]。按照认知心理学的实验解析,当行为者的信息加工能力 具有局限性时,其判断和决策就会产生偏差 [22]。而前景理论 因其能够有效地解决期望理论和事件预测的不一致,从而成为 行为经济学的理论基石。基于此,本文引入前景理论对养老机 构新型冠状病毒肺炎风险管理主体的风险行为博弈演化过程 和机理进行分析,从风险感知理论角度诠释不同新型冠状病毒 肺炎风险主体行为倾向的原因和新型冠状病毒肺炎风险管理 f. zhang et al. / journal of risk analysis and crisis response 10(2) 37–42 39 图1 | 前景理论下价值函数图. 表1 | 基于前景理论的新型冠状病毒肺炎风险行为的主要影响因素及 行为选择 博弈双方 主要影响因素 因素c 1 因素c 2 因素c 3 自身相 对收益 新型冠状病 毒肺炎风险 关联程度 监管与 惩罚 照护人员 “规范”照护行为a1 r1+r1 q1 01 “非规范”照护行为a2 r1 q1 k1 参考标准 ta1 ta2 ta3 权重 wa1 wa2 wa3 机构管理者 “积极”管理行为b1 r2+r2 q2 02 “消极”管理行为b2 r2 q2 k2 参考标准 tb1 tb2 tb3 权重 wb1 wb2 wb3 表2 | 管理及照护行为博弈矩阵 照护人员(a) 机构管理者(b) 积极的新型冠状病毒 肺炎风险管理(b1) 消极的新型冠状病毒 肺炎风险管理(b2) “规范”照护行为(a1) –va –va+pb–kßl –vb –vb–k(1-ß) l “非规范”照护行为 (a2) –pa–gßl –ßl –vb+pa–g(1-ß)l –(1-ß)l 策略选择的条件,通过分析不同条件下风险管理行为与护理行 为相互关系及演化结果的影响,为养老机构新型冠状病毒肺炎 风险管理提供理论和决策支持。 2.3. 养老机构新型冠状病毒肺炎风险行为演化 模型 2.3.1. 模型假设 (1)养老机构新型冠状病毒肺炎风险管理博弈过程中存在养老 机构管理者和照护人员两类博弈群体,因养老机构相对于一般 的社区服务中心具有的不确定性和复杂性,管理者和养老机构 照护人员在新型冠状病毒肺炎风险感知与偏好、管理知识、 能力、角色以及对于新型冠状病毒肺炎风险信息的获取及分 析能力等方面存在差异,表现出有限理性特征。对于新型冠 状病毒肺炎风险管理策略等选择主要基于风险策略选择收益 及损失的心理预期和感知价值,而不是实际收益和损失。根据 kahneman和tversky的前景理论,该感知价值可用前景价值v衡 量,用价值函数v(x)和决策权重函数p(p)代替期望效用理论中 的效用函数u(x)和概率p。因此,前景理论的效用值可以表达为 e(a)=σp(pi)v(x),在所有备选方案中,e指最大的方案即为最佳 方案 [23]。 kahneman给出基于前景理论的价值函数v(x) ,如图1所示。 其中,原点是参考标准,x是养老机构相对的收益(或损失),v是 价值。从图中可以总结出前景理论的4个重要特点：①养老 机构更关心收益或损失的相对水平,因此,价值是关于收益(或 损失)相对变化x的函数；②养老机构对损失比对收益更敏感, 因此,价值函数在损失状态下的斜率比在收益状态斜率下更大, 损失时下降更快；③越接近参考标准,函数越敏感,即在参考标 准附近,随着养老机构收益(或损失)的相对变化,价值的变化更 大；④价值函数在收益状态(x>0)时是凹形,在损失状态(x<0) 是凸形,且在参考标准附近,敏感性特征越强。 根据前景理论,价值函数的计算公式如下 v x x x o x x ( ) , ( ) , = ³ < ì í ï îï a bl 0 (1) 其中,0<α、β<1,λ>1,α和β分别给出了价值函数的收益区域和损 失区域的凹凸程度,其反映了养老机构对待收益和损失的不同 风险态度：α和β越大,表明养老机构越倾向于冒险；λ表示养 老机构的损失规避程度：λ越大,表明养老机构对损失的规避 程度越大。 前景理论为不确定条件下的决策研究做出开创性的贡献,目前 广泛应用于各种决策分析中 [24,25]。在现实中,养老机构照 护人员往往不是完全理性的,容易受到周边环境和其心理因素 的影响。根据前景理论分析价值的载体是管理者及照护人员 的收益变化而不是最终状态,这种变化根据对参考标准的偏离 程度来定义,因此,管理者及照护人员常常以自己的视角或参考 标准来决定其行为。由于参考标准的不同,使得其预期的“收 益”或“损失”不同。 2.4. 基于前景理论的变量设定及矩阵分析 2.4.1 变量设定通 过上述分析,综合考虑影响护理行为的上述三类因素,即自身相 对收益况c1、新型冠状病毒肺炎风险的关联程度c2以及养老 机构的监管与惩罚c3,构建基于前景理论的新型冠状病毒肺炎 风险行为模型。影响新型冠状病毒肺炎风险行为的主要因素 及行为选择如下表所示。 其中,r1表示照护人员的收益,r1表示选择”规范”的照护行为获 得的额外奖金,ta1、ta2、ta3分别为三类因素在照护人员方所 选择的参考标准。r2表示机构管理者的收益,r2表示选择“积 极”管理行为获得的额外奖金,tb1、tb2、tb3分别为三类因素 在机构管理者方所选择的参考标准。 表2中,va、vb分别为照护人员和机构管理者所选择的“规范”照 护行为和积极开展新型冠状病毒肺炎风险管理活动所支付的 成本的感知价值,va,vb∈(0,∞)。pa、pb 分别为当仅有照护人 员或机构管理者积极开展新型冠状病毒肺炎风险管理时,机构 管理者或照护人由于“消极”管理行为及“坏”的照护行为所受 处罚的感知价值,pa,pb ∈(0,∞)。l为两方需要承担新型冠状病 毒肺炎风险事故发生时的风险损失感知价值,包括声誉损失、 经济损失等,l ∈(0,∞)。ß为新型冠状病毒肺炎风险损失在管 理者和照护人员间的分担比例,ß∈(0,1)。k、g分别为当仅有 40 f. zhang et al. / journal of risk analysis and crisis response 10(2) 37–42 机构管理者或照护人员“积极”开展新型冠状病毒肺炎风险管 理及“好”的照护行为时,两方所需承担的新型冠状病毒肺炎风 险损失的折扣系数,k、g∈(0,1)。x、y分别为机构管理者和照 护人员选择积极开展新型冠状病毒肺炎风险管理和规范的行 为的概率,x、y ∈(0,1)。 2.4.2. 矩阵分析 (1) 当机构管理者和照护人员选择“积极”的风险管理行为和“ 规范”的照护行为(b 1 、a 1 )时,两方分别需付出的新型冠状 病毒肺炎风险成本为va、vb。 (2) 当机构管理者和照护人员各自选择“积极”的风险管理行 为和“坏”的照护行为(b 1 、a 2 )时,机构管理者需要支付的新 型冠状病毒肺炎风险管理成本vb,照护人员因“坏”的照护 行为收到处罚pb。由此产生的新型冠状病毒肺炎风险l, 依据新型冠状病毒肺炎风险损失责任分担比例ß,以及当 仅有机构管理者积极开展新型冠状病毒肺炎风险管理的 风险损失折扣系数k,机构管理者承担损失为kßl,照护人员 承担损失为k(1-ß) l。 (3) 当机构管理者和照护人员各自选择“消极”的风险管理行 为和“规范”的照护行为(b 2 、a 1 )时,机构管理者因消极的照 护管理受到处罚pb。照护人员需要支付新型冠状病毒肺 炎风险成本va,产生的新型冠状病毒肺炎风险损失分担比 例ß和只有照护人员采取“规范”的护理行为的风险损失折 扣系数gßl,照护人员承担损失为g (1-ß)l。 (4) 当机构管理者和照护人员各自选择“消极”的风险管理行 为和”非规范”的照护行为(b 2 、a 2 )时,两方应承担的新型冠 状病毒肺炎风险损失感知成本l,依据新型冠状病毒肺炎 风险损失分担比例为ß,机构管理者承担损失为ßl,照护人 员承担损失为(1-ß)l。 3. 基于前景理论的风险控制策略 养老机构风险管理的最佳状态是机构管理者和照护人员双方 都充分重视并积极开展新型冠状病毒肺炎风险管理活动及开 展“规范”的护理行为,最大程度提高管理效率,降低新型冠状病 毒肺炎风险。根据均衡点稳定性分析可推知,行为转化系统要 实现新型冠状病毒肺炎风险发生概率的最小化及新型冠状病 毒肺炎风险损失最小目标应达到ca ≤ db + (1 − k)βl,ca ≤ da + gβl,cb ≤ da + (1 − g)(1 − β)l和cb ≤ db + k(1 − β)l这4 个条件时,系统将收敛于均衡点ｅ4(1,1),此时,机构管理者重视 和积极开展新型冠状病毒肺炎风险管理活动,照护人员积极采 取“规范”的护理行为,养老机构新型冠状病毒肺炎风险发生概 率降到最低,损失也随之降到最低。 ca的约束条件说明：通常情况下,医疗机构管理者积极开展新 型冠状病毒肺炎风险管理的成本,不能超过照护人员采取“非 规范”的照护行为给养老机构带来的对照护人员惩罚的感知价 值和新型冠状病毒肺炎风险损失感知价值,也不能超过照护人 员积极采用“规范”的行为开展护理活动所感知的处罚风险以 及风险损失。cb的约束条件说明,照护人员积极开展“规范”的 护理行为的成本应小于管理者消极开展新型冠状病毒肺炎风 险管理给照护人员带来的对养老机构惩罚的感知价值以及风 险损失感知价值,也应小于机构管理者积极开展新型冠状病毒 肺炎风险管理所感知的处罚风险以及风险损失。为了使得两 方以最大的概率收敛于均衡点ｅ4(1,1),需要尽量使得上述4个 条件得到满足。然而,风险感知差异和前景理论效应的存在可 能导致系统难以收敛于均衡点ｅ4(1,1)。 noel等 [26]研究表明,风险感知是许多健康理论的核心,与个 人行为之间呈相关关系。风险双因素理论认为风险感知包含 对决策后果发生的主观概率和错误决策后果严重性的价值判 断。风险感知水平越低,面对潜在风险威胁时,越易于倾向忽视 风险而未积极采取事先预防的办法使损失降低,反之,若风险感 知水平越高,面对潜在新型冠状病毒肺炎风险时,越易于选择积 极规避新型冠状病毒肺炎风险或采取事先预防新型冠状病毒 肺炎风险的措施从而降低其引起的损失。影响新型冠状病毒 肺炎风险感知的因素包括新型冠状病毒肺炎风险认知偏差、 风险可接受水平等。 实际新型冠状病毒肺炎风险管理成本小于感知的新型冠状病 毒肺炎风险管理成本,即υ(ca ≥ ca),υ(cb ≥ cb)。当机构管理者 或照护人员选择积极开展新型冠状病毒肺炎风险管理时,该行 为的发生概率p1=p2=1,根据假设1和前景理论,可知： c p c p c c c c a a a a a b = + = + = ³ = p u p u p u p u u p ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( 1 11 0 1 0 0 pp c p c c c b b b b 2 21 0 1 0 0 ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) u p u p u p u u + = + = ³ 其中,ca和cb分别为机构管理者和照护人员积极开展风险管理 的实际成本。 (2)机构管理者和照护人员在对策略选择价值进行判断时,容易 存在侥幸和乐观偏见。乐观偏见表现为一种非现实的乐观主 义,人们在考虑自身风险时更易于存在乐观偏见,认为自己比一 般人更少遭遇消极事件,比一般人更多遇见积极事件 [27],倾 向于事情向好的结果发展。一般而言,乐观偏见降低了管理者 和员工对风险的认知水平,从而低估新型冠状病毒肺炎风险的 存在,减少解决问题的努力,降低防范新型冠状病毒肺炎风险 的意愿。机构管理者和照护人员的乐观偏见使得双方在明知 消极风险管理可能带来风险损失的情况下,依然寄希望于偶然 不发生风险事故和损失,容易低估消极风险管理行为所可能导 致的风险事故发生的概率,以及处罚概率。并且,根据前景理 论,权重函数具有除了极小概率事件外,p(p1) ≤ p0,且当 dwi > 0 时,u w( )d i 具有凸函数的特征。因此可见：p(p3) ≤ p3,p(p4) ≤ p4,p(p5) ≤ p5,υ(da) < da,υ(db) < db,υ(l) < l。其中,p3为新型冠状病 毒肺炎风险事故发生的客观概率：p1、p3分别为机构管理者 和照护人员因新型冠状病毒肺炎风险发生被处罚的客观概 率；da、db分别为机构管理者和照护人员采取消极风险管理策 略收到的实际处罚成本；l为新型冠状病毒肺炎风险发生的实 际损失。由此可得： d p d p p d p d d p d a a a a b b = + = £ = + p u p u p u p u p ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( 4 4 4 4 3 1 0 1-= £ = + = p p d p d l p l p p l b b5 5 5 3 3 3 0 1 0 ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) u p u p u p u p u ££ p l3 (3)根据前景理论中提出的反射效应(reflection effect)人们 对于获得和损失的偏好是不对称的,而面对获得(或盈利)的前 景时,人们有风险规避(risk averse)的倾向；而面对可能损失 的前景时,人们有风险追求(risk seeking)的倾向。对于养老 机构新型冠状病毒肺炎风险管理决策者而言,当面对确定性损 失时,机构管理者和照护人员作为有限理性决策主体更倾向于 风险偏好,宁可放弃积极风险管理策略所发生的确定性成本支 出,而倾向于冒险选择消极风险管理策略,选择面对不确定的风 险事故损失和处罚,使得系统难以收敛于均衡点ｅ4(1,1)。因而 易于出现机构管理者和照护人员的新型冠状病毒肺炎风险成 f. zhang et al. / journal of risk analysis and crisis response 10(2) 37–42 41 本被高估,风险损失被低估,使得ca和cb偏大,而da、db和l偏小, 双方存在偏好冒险选择消极风险管理策略的倾向。现实中机 构管理者和照护人员存在风险管理重视程度不足,采取消极风 险管理策略现象难以杜绝的原因。当养老机构满足以上条件 时,该系统将收敛在均衡点b(1,1),即机构管理者和照护人员都 重视并积极开展风险管理活动和采用“规范”的照护行为,从而 实现养老机构新型冠状病毒肺炎风险发生概率和新型冠状病 毒肺炎风险损失降到最小。 因此,促使机构管理者和照护人员采取积极风险管理策略以 及“规范”的护理行为在于改变双方对于新型冠状病毒肺炎风 险事件发生率、风险管理成本、内附损失等方面的认知偏 差。包括：加强机构管理者和照护人员对养老机构新型冠 状病毒肺炎风险复杂性和不确定性的认知,提升其新型冠状病 毒肺炎风险识别和评估能力,使得双方对风险管理成本感知 价值的主观判断更加合理与准确。通过优化新型冠状病毒 肺炎风险管理制度、流程和程序,降低新型冠状病毒肺炎风险 管理成本支出,提高新型冠状病毒肺炎风险管理效率。充分 了解养老机构新型冠状病毒肺炎风险事故所产生的家庭、社 会、经济等负面影响,加强养老机构新型冠状病毒肺炎风险事 故、后果严重影响及有效防范的研究,对养老机构新型冠状病 毒肺炎风险防范的重要价值和社会作用形成更全面和清晰的 认识。通过强化新型冠状病毒肺炎风险管理工作审查和监 督,对造成严重后果的新型冠状病毒肺炎风险消极管理行为,从 经济、声誉、资质等综合角度加强管理,提高其对有效开展新 型冠状病毒肺炎风险防范的价值感知。 conflicts of interest we declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, “基于前景理论的养老机 构新型冠状病毒肺炎风险防控研究”. 参考文献 [1] who. who coronavirus disease (covid-19) dashboard. geneva, switzerland: who; 2020. available from: https:// covid19.who.int. [2] 周涛, 刘权辉, 杨紫陌, 廖敬仪, 杨可心. 等.武汉新型冠状 病毒感染肺炎基本再生数的初步预测.中国循证医学杂志 2020;20:1–6. zhou t, liu q, yang z, liao j, yang k. preliminary prediction of novel coronavirus infected pneumonia with the number of regeneration. chinese j evid based med 2020;20:1–6. [3] who. statement of the second meeting of the international health regulations (2005) emergency committee on the 2019 new coronavirus outbreak. geneva, switzerland: who; 2020. available from: https://www.who.int/zh/news-room/ detail/30-01-2020-statement-on-the-second-meeting-of-the international-health-regulations-(2005)-emergency-committee regarding-the-outbreak-of-novel-coronavirus-(2019-ncov). [4] 唐方成. 新技术商业化的风险要素及其作用机理：基于社 会技术系统理论的实证研究. 系统工程理论与实践 2013;33: 622–31. tang f. risk factors of new technology commercialization and their mechanism of action: empirical research based on social technology system theory. syst eng theory pract 2013;33:622–31. [5] 戚妙芬, 周颖. 护理风险管理在急诊患者安全管理中的实施 效果评价[j]. 护士进修杂志 2014;3:33–6. qi m, zhou y. implementation effect evaluation of nursing risk management in emergency patient safety management. j nurs educ 2014;3:33–6. [6] 孙晓敏, 丁雅英, 汪虹, 等. 住院患者护理风险评估预警监控传 报系统的建立与应用效果评价. 中国护理管理 2017;17:378–82. sun x, ding y, wang h, et al. establishment and application effect evaluation of inpatient nursing risk assessment, early warning monitoring and communication system. nurs manage china 2017;17:378–82. [7] jones d, whitaker t. preventing falls in older people: assessment and interventions. nurs stand 2011;25:50–5. [8] saliba d, jones m, streim j, ouslander j, berlowitz d, buchanan j. overview of significant changes in the minimum data set for nursing homes version 3.0. j am med dir assoc 2012;13:595–601. [9] denny ds, allen dk, worthington n, gupta d. the use of failure mode and effect analysis in a radiation oncology setting: the cancer treatment centers of america experience. j healthc qual 2014;36:18–28. [10] leland ne, gozalo p, teno j, mor v. falls in newly admitted nursing home residents: a national study. j am geriatr soc 2012;60:939–45. [11] onder g, carpenter i, finne-soveri h, gindin j, frijters d, henrard jc, et al. assessment of nursing home residents in europe: the services and health for elderly in long term care (shelter) study. bmc health serv res 2012;12:5. [12] carrillo-menéndez s, suárez a. robust quantification of the exposure to operational risk: bringing economic sense to economic capital. comput oper res 2012;39:792–804. [13] chiang cy, hsu sj, chan ct. a resident’s behavior simulation model for nursing home healthcare services. biomed mater eng 2014;24:69–75. [14] 黄菲, 张会君, 尹姣. 辽宁省养老机构照护人员培训现状及需 求[j]. 中国老年学杂志 2012;32:570–3. huang f, zhang h, yin j. current situation and demand of nursing staff training in liaoning province. chinese j gerontol 2012;32:570–3. [15] 张增芳. 老龄化背景下机构养老的供需矛盾及发展思路—基 于西安市的数据分析[j]. 西北大学学报：哲学社会科学版 2012;42:35–9. zhang z. contradiction between supply and demand of institutional endowment and its development idea in the context of aging—based on data analysis of xi ‘an. j northwest univ phil soc sci ed 2012;42:35–9. [16] 郭红艳, 王黎, 彭嘉琳, 谢红. 等. 养老机构管理者对质量管理 体会的质性研究. 护理管理杂志 2013;13:707–9. guo h, wang l, peng j, xei h. qualitative research on the quality management experience of nursing home managers. j nurs manage 2013;13:707–9. [17] 李虹彦, 殷欣, 刘涛, 李君歆. 等. 社会养老机构中虐待老年人 问题的现状与思考[j]. 中国老年学杂志 2015;32:4846–8. li h, yin x, liu t, li j. current situation and consideration of maltreatment of the elderly in social endowment institutions. chinese j gerontol 2015;32:4846–8. https://www.who.int/zh/news-room/detail/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-%282005%29-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-%282019-ncov%29 https://www.who.int/zh/news-room/detail/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-%282005%29-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-%282019-ncov%29 https://www.who.int/zh/news-room/detail/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-%282005%29-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-%282019-ncov%29 https://www.who.int/zh/news-room/detail/30-01-2020-statement-on-the-second-meeting-of-the-international-health-regulations-%282005%29-emergency-committee-regarding-the-outbreak-of-novel-coronavirus-%282019-ncov%29 https://doi.org/10.7748/ns2011.08.25.52.50.c8674 https://doi.org/10.7748/ns2011.08.25.52.50.c8674 https://doi.org/10.1016/j.jamda.2012.06.001 https://doi.org/10.1016/j.jamda.2012.06.001 https://doi.org/10.1016/j.jamda.2012.06.001 https://doi.org/10.1016/j.jamda.2012.06.001 https://doi.org/10.1111/j.1945-1474.2011.00199.x https://doi.org/10.1111/j.1945-1474.2011.00199.x https://doi.org/10.1111/j.1945-1474.2011.00199.x https://doi.org/10.1111/j.1945-1474.2011.00199.x https://doi.org/10.1111/j.1532-5415.2012.03931.x https://doi.org/10.1111/j.1532-5415.2012.03931.x https://doi.org/10.1111/j.1532-5415.2012.03931.x https://doi.org/10.1186/1472-6963-12-5 https://doi.org/10.1186/1472-6963-12-5 https://doi.org/10.1186/1472-6963-12-5 https://doi.org/10.1186/1472-6963-12-5 https://doi.org/10.1016/j.cor.2010.10.001 https://doi.org/10.1016/j.cor.2010.10.001 https://doi.org/10.1016/j.cor.2010.10.001 https://pubmed.ncbi.nlm.nih.gov/24211884/ https://pubmed.ncbi.nlm.nih.gov/24211884/ https://pubmed.ncbi.nlm.nih.gov/24211884/ 42 f. zhang et al. / journal of risk analysis and crisis response 10(2) 37–42 [18] 袁群, 易霞, 张银华, 陈燕, 秦莉花. 等. 养老机构护理员工作压 力研究现状与进展[j]. 中国护理管理 2015;15:112–15. yuan q, yi x, zhang y, chen y, qin l. research status and progress on work stress of nursing assistants in nursing institutions. china nurs manage 2015;15:112–15. [19] 唐喻莹, 徐杉, 李娜, 孙鸿燕. 护士多点执业的利弊分析与建 议. 中华护理杂志 2017;52:119–22. tang y, xu s, li n, sun h. analysis and suggestions on multipoint practice of nurses. chinese j nurs 2017;52:119–22. [20] decker fh, harris-kojetin ld, bercovitz a. intrinsic job satisfaction, overall satisfaction, and intention to leave the job among nursing assistants in nursing homes. gerontologist 2013;49:596–610. [21] von neumann j, morgenstern o. theory of games and economic behavior. princeton, nj, usa: princeton university press; 1944, pp. 17–28. [22] 赵泽斌, 满庆鹏. 基于前景理论的重大基础设施工程风险管 理行为演化博弈分析. 系统管理学报 2018;27:110–17. zhao z, man q. evolutionary game analysis of risk management behavior in major infrastructure projects based on prospect theory. j syst manage 2018;27:110–17. [23] jou rc, chen kh. an application of cumulative prospect theory to freeway drivers’ route choice behaviors. transport res part a 2013;49:123–31. [24] wang l, zhang zx, wang ym. a prospect theory-based interval dynamic reference point method for emergency decision making. expert syst appl 2015;42:9379–88. [25] shi y, cui x, li d. discrete-time behavioral portfolio selection under cumulative prospect theory. j econ dyn control 2015;61:283–302. [26] brewer nt, chapman gb, gibbons fx, gerrard m, mccaul kd, weinstein nd. meta-analysis of the relationship between risk perception and health behavior: the example of vaccination. health psychol 2007;26:136–45. [27] harris ajl, hahn u. unrealistic optimism about future life events: a cautionary note. psychol rev 2011;118:135–54. https://doi.org/10.1093/geront/gnp051 https://doi.org/10.1093/geront/gnp051 https://doi.org/10.1093/geront/gnp051 https://doi.org/10.1093/geront/gnp051 https://doi.org/10.1016/j.eswa.2015.07.056 https://doi.org/10.1016/j.eswa.2015.07.056 https://doi.org/10.1016/j.eswa.2015.07.056 https://doi.org/10.1016/j.jedc.2015.10.002 https://doi.org/10.1016/j.jedc.2015.10.002 https://doi.org/10.1016/j.jedc.2015.10.002 https://pubmed.ncbi.nlm.nih.gov/17385964/ https://pubmed.ncbi.nlm.nih.gov/17385964/ https://pubmed.ncbi.nlm.nih.gov/17385964/ https://pubmed.ncbi.nlm.nih.gov/17385964/ https://doi.org/10.1037/a0020997 https://doi.org/10.1037/a0020997 mcsim-based occupational health risk assessment on benzene received 10 september 2016 accepted 20 april 2016 application of risk estimation of noise-induced hearing loss method in evaluations of occupational disease hazards in construction projects in china* minyan li, deyin huang†, qian zhang institute of occupational health, tianjin bohai chemical industry group co. ltd, tianjin 300051,china mao liu centre for urban public safety research, nankai university, tianjin 300071,china abstract to protect the workers exposed to industrial noise, prevent and control the occurrence of noise-induced hearing loss, and reduce the incidence of occupational noise-induced deafness, a method and its application example were presented to assess the risk of noise-induced hearing loss. based on the data collected by occupational hygienic investigations and measurements, the consequences and probabilities of noise-induced hearing loss could be predicted by using the iso 1999:2013(e) risk model when the appropriate frequency combinations of interest and fences have been chosen according to the need, and then the corresponding management measures should be taken. it shows that this method could be widely used in evaluations of occupational disease hazards in construction projects in china. it will provide the technical basis in aspect of hearing protection for enterprises, assessment organizations and administration to improve the occupational health management continuously, to give reasonable advices for enterprises to protect the workers’ hearing, and to supervise and inspect the industrial occupational health situations, respectively. keywords: hearing loss, noise, occupational disease, risk assessment, iso 1999 * this study was supported by municipal key science and technology support project of tianjin (no. 13zczdsy02300). † corresponding author: huang_deyin@126.com 1. introduction in china and abroad, guidelines and criterion have been drafted to evaluate the health risk of people exposed to the toxic chemical substances, while in respect to the health risk assessment of noise, a set of systemic acoustics criterion have been set by iso/tc 43, acoustics committee. a noise-induced health risk estimation method is introduced in iso 1999 in detail, and the health effects in this model only refers to hearing loss. iso (the international organization for standardization) published the first edition of iso 1999, i.e. iso 1999:1975(e) “acoustics – assessment of occupational noise exposure for hearing conservation purposes”, which gave the method to measure the noise and listed the risk of hearing impairment for conversation speech frequencies (500 hz、1000 hz、2000 hz) when people exposed to different noise levels for different durations. after that, iso 1999: 1990(e) was published to cancel and replace the first edition, and in china, the first standard to assess the noise-induced hearing loss gb/t 14366-93 “acousticsdetermination of occupational noise exposure and estimation of noise-induced hearing impairment” (mod iso 1999:1990(e)) was published. compared to the 1975 edition, hearing loss of different frequencies (500 hz、1000 hz、2000 hz、3000 hz、4000 hz、6000 hz) of both genders exposed to different noise levels for different durations and method to calculate the journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 76-84 published by atlantis press copyright: the authors 76 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations risk of noise-induced impairment was added. recently, the latest edition iso 1999: 2013(e) “acousticsestimation of noise-induced hearing loss” was published to cancel and replace the second one, while gb/t 14366-93 has not been updated yet. data of database b are added in the third edition compared to the second one with some details revised. an approval aq standard “risk management guidelines for noise occupational disease hazard” drafted by our team introducing the latest iso model to estimate noise-induced hearing loss will be published very soon. based on the iso 1999:2013(e), method to estimate the risk of noise-induced hearing loss is introduced and applied in the noise occupational disease hazard assessment. the frequency combinations and fences matching the present state of chinese assessment of occupational hazard are selected to estimate the noise-induced hearing loss to guide the employing unit to take corresponding management measures according to the different risk levels. 2. method 2.1. investigation on the noise occupational exposure situation the investigation on the noise occupational exposure situation includes production process, the fixed number of staff members of each workshop post and their tasks, the exposure situation of each post, the selection and use of hearing protectors, the draft and implement situation of hearing conservation program and occupational epidemiology data collection of similar enterprises in china, etc. according to the investigation results, the noise sources and workers affected could be identified to be used to analysis the occupational exposure characteristics. 2.2. evaluation of the occupational noise exposure the instruments used, the places to measure noise and the method to measure and calculate the sound levels all should be according to gbz/t 189.8. a-weighted noise exposure level normalized to a nominal 8 h working day or nominal week of 5 eight-hour working days could be used to determine whether workers need to wear hearing protectors or not. when the sound levels of the workplaces could not be measured, the occupational noise exposure levels could be forecast according to the measure data of similar enterprises in china or the past measure data of the workplaces of the employing unit itself. 2.3 risk assessment of noise-induced hearing loss the potential risk of occupational noise-induced hearing loss could be quantitatively expressed by the noise-induced permanent threshold shift (nipts) and the percentages of a population whose hearing threshold level associated with age and noise (htlan) and hearing threshold level associated with age (htla) exceed the fence. to assess the risk of hearing loss and to estimate the hearing loss of hearing threshold level associated with age and noise and noise-induced permanent threshold shift is the manner to evaluate the affect that the noise induced in the occupational population. after selecting the appropriate frequency combinations and fences, percentages in the population whose htlan and htla of the selected frequency combinations which exceed the fences could be calculated, so that the nipts and the percentages of a population whose htlan and htla exceeds the fence could be estimate. 2.3.1. select the appropriate frequency combinations and fences the hearing threshold level of the chosen frequency combinations could be used to evaluate the noise-induced hearing loss, so the first step is to select the frequencies of interest and the appropriate fence according to the objective of the noise occupational disease hazard risk assessment. (1) choose the average hearing threshold level of high frequencies (3000 hz、4000 hz and 6000hz) of any ear and the fence as 10 db, according to the “ high-frequency standard threshold shift (hsts)” defined in the “hearing protection specification for employee in industrial enterprises” published by ministry of health of the people's republic of china in 1999. the related noise occupational disease hazard evaluation results will be used to guide managers to take appropriate measures. the specification provides that the enterprises should take measures to protect the hearing of those who have hsts to prevent a further hearing loss. published by atlantis press copyright: the authors 77 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations (2) choose the average hearing threshold level of high frequencies (3000 hz、4000 hz and 6000hz) of both ears and the fence as 40 db, according to the precondition to be diagnosed as occupational noise deaf patients in accordance with the gbz 49-2014. the related noise occupational disease hazard evaluation results will be used to remind the managers to take appropriate measures before someone has been diagnosed as an occupational noise deaf patient. (3) choose the weighted average hearing threshold level of speech frequencies (500 hz、1000 hz and 2000 hz) and high frequency (4000 hz) of better ear, i.e. 1/3×[hl500 hz+hl1000 hz+hl2000 hz]×0.9+hl4000 hz×0.1, and the fence as 25 db, according to the diagnose results of the occupational noise deafness level according to the gbz 49-2014. the related noise occupational disease hazard evaluation results will be used to warn the managers that occupational noise deafness cases will occur if no measures have been taken. 2.3.2. hearing threshold level associated with age and noise (htlan) the hearing threshold level, in decibels, associated with age and noise (htlan), h’, of a noise-exposed population is calculated, for the purposes of this iso model , by using formula (1): = 120 h n h h n × + −’ (1) where h’ is the hearing threshold level, expressed in decibels, associated with age and noise (htlan); h is the hearing threshold level, expressed in decibels, associated with age (htla); n is the actual or potential noise-induced permanent threshold shift (nipts), expressed in decibels. this formula is applicable only to corresponding percentage values of h’, h, and n. note that the relationship expressed in formula (1) is an approximation to the biological events and is considered accurate enough for the purposes of this iso model. the term (h × n)/120 starts to significantly modify the result only when h + n is more than approximately 40 db. 2.3.3. databases for hearing threshold levels associated with age (htla) 2.3.3.1. choice of database the hearing of a non-noise-exposed population as a function of age depends on the degree to which other factors besides aging are inadvertently included; diseases, history of ototoxic drugs, and unknown noise exposure of occupational or non-occupational origin may modify the htla. different approaches to screening such data have been used and the selection of the most appropriate database depends on the purpose of the application. two databases (databases a and b) to be used for htla should be according to the specification in iso 1999:2013(e). 2.3.3.2. database a database a derives from otologically normal persons. the statistical distributions of the thresholds of such highly screened populations have been standardized in gb/t 7582 separately for male and female populations. there are two ways to acquired database a: (1) calculating database a by formulae specified in iso 1999:2013(e), and (2) the look-up table method by using the tables of database a provided in the iso model. the formulae applicable for the hearing threshold level, h, as a function of age, y (years), for the various ranges of the percentage, q, could be expressed as: for q = 50%: ( ) 2 md,y md,1818h a y h= − + (2) for 5% ≤ q < 50%: md,yq uh h ks= + (3) for 50% < q ≤ 95%: md,yq lh h ks= − (4) where, su is the standard deviation of the upper half of the distribution; published by atlantis press copyright: the authors 78 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations sl is the standard deviation of the lower half of the distribution; hmd,18 is the median value of the hearing threshold of otologically normal persons of the same sex aged 18 years, which for practical purposes is taken as zero. hence, hq has been termed the hearing threshold level associated with age. here, the values for the multiplier k corresponding to the normal (gaussian) distribution, the values for the coefficient could be looked up in the tables of iso 1999: 2013(e), and the parameters su and sl could be calculated by formulae given in this iso standard. 2.3.3.3. database b three examples of database b are presented in iso 1999: 2013(e) for an unscreened population (males and females). these examples are compiled from representative data from three industrialized countries: sweden (b.2), norway (b.3), and the united states (b.4). b.2 and b.3 represent populations who have not been exposed to occupational noise, while subjects with occupational noise exposure are included in b.4. 2.3.4. noise-induced permanent threshold shift 2.3.4.1. calculation of n50 the median potential nipts values to be used are functions of audiometric frequency, the exposure duration, the ratio t/t0, and the noise exposure level normalized to a nominal 8 h working day, lex,8h, and 5 days per week, averaged over the exposure duration t. for exposure durations between 10 years and 40 years, the median potential nipts values, n50, in decibels, are given for both genders by formula (5): [ ] 250 0 ex,8h 0lg( / ) ( )n u v t t l l= + − (5) where lex,8h is the noise exposure level normalized to a nominal 8 h working day, expressed in decibels; l0 is the sound pressure level, defined as a function of frequency, expressed in decibels, below which the effect on hearing is negligible; t is the exposure duration, expressed in years; t0 is 1 year; u, v and l0 are given as a function of frequency in table 1 in iso 1999:2013(e). for exposure durations less than 10 years, n shall be extrapolated from the value of n50 for 10 years using formula (6): 50, 10 50, 10 lg( 1) lg(11)t t t n n< = + = (6) formula (6) is valid for exposure durations between 1 year and 10 years. 2.3.4.2. statistical distribution of noise-induced permanent threshold shift, n for the purposes of iso 1999:2013(e), the statistical distribution of n is approximated by two different halves of two normal (gaussian) distributions. the upper half, for the percentage with hearing worse than the median, is characterized by the parameter du; the lower half has a smaller dispersion characterized by the parameter percentage dl. for a percentage of, q, of the population such that 5% ≤ q < 50%, the nipts is given by formula (7): 50q un n kd= + (7) for a percentage of, q, of the population such that 50% < q ≤ 95%, the nipts is given by formula (8): 50q ln n kd= − (8) du and dl shall be calculated in accordance with corresponding specification in iso 1999:2013(e). besides, the three particular values n10, n50, n90 can also be looked up from the table provided in the iso model. 3. noise-induced risk management based on the assessment results of risk of hearing loss due to noise exposure, the key of risk management is to reduce n. for the workers who have hsts, managers should take measures to prevent a further hearing loss; for those who meet the precondition to be diagnosed as occupational noise deaf patients, they should be considered to be diagnosed and classified according to their situation; and for those already have been diagnosed as occupational noise deaf patients, they should be transferred off the noise workplaces. published by atlantis press copyright: the authors 79 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations 4. application example analysis there are 36 winding machines in winding workshop of polymers section of cell 2 department in the battery assembly enterprise evaluated, the occupational hazard of which is identified and analyzed in a status quo assessment. 104 workers in this workshop exposed to noise when working in this area. 4.1. objects 4.1.1. principles to choose objects the population selected in the application example analysis should meet the requirements as follows: have (1) no prior employment (with a different employer) in a high noise environment; (2) no ear disease or history of ear disease; (3) no family history of hearing loss; (4) no history of diabetes; (5) never use ototoxic drugs; (6) no non-occupational noise exposure; (7) no other hearing loss risk factors. 4.1.2. objects to assess choose workers exposed to noise of this workshop as objects whose risk of noise-induced hearing loss will be assessed. they are mostly male between 20 and 30 years old. the fixed number of staff members of the workshop post and their tasks, and the exposure situation of each post are given in table 1. table 1. the fixed number of staff members of the workshop post and their tasks, and the exposure situation of each post posts number of staff members per shift shifts total number of staff members tasks and operate time operators a 10 three shifts with three changing from monday to thursday and three shifts with two changing from friday to sunday 30 they operate on the no.7, no.8, no.11, and no.13 winding machines and work for 7 h per shift from monday to thursday and for 10 h per shift from friday to sunday. operators b 9 four shifts with three changing 36 they operate on the no.9, no.10, and no.12 winding machines and work for 7 h per shift for 5 days per week. monitors a 2 three shifts with three changing from monday to thursday and three shifts with two changing from friday to sunday 6 they check on the no.7, no.8, no.11, and no.13 winding machines and work for 7 h per shift from monday to thursday and for 10 h per shift from friday to sunday. monitors b 2 four shifts with three changing 8 they check on the no.9, no.10, and no.12 winding machines and work for 7 h per shift for 5 days per week. precision monitors a 2 three shifts with three changing from monday to thursday and three shifts with two changing from friday to sunday 6 they check on the no.7, no.8, no.11, and no.13 lines and work for 7 h per shift from monday to thursday and for 10 h per shift from friday to sunday. precision monitors b 2 four shifts with three changing 8 they check on the no.9, no.10, and no.12 lines and work for 7 h per shift for 5 days per week. 4.2. method 4.2.1. noise occupational exposure evaluation the noise exposure levels of workers in this workshop are given in table 2. based on the measurement results, the calculated lex,w of workers in this workshop is between 85.2 db and 88.0 db, with the average of 86.9 db. 4.2.2. risk of hearing loss due to noise assume that the male workers exposed to the industrial noises of this workshop without any personnel hearing protectors from 20 years old, and the noise exposure level lex,w=86.9 db. predict the hsts and occupational noise deafness risks of population when they are 30, 40, 50 or 60 years old. a microsoft® excel worksheet was compiled based on the method given before and the information above was input. published by atlantis press copyright: the authors 80 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations 4.3. results choose database a to calculate the hearing threshold level associated with age, and use the compiled excel worksheet to predict the hsts and occupational noise deafness risk for these male workers if they do not wear any hearing protectors. (1) risk of hsts the predicted risks of hsts are given in table 3 and the risk assessment results are expressed in fig. 1. the results show that h and h’ both increase with the duration and n presents an increasing trend on the condition of fixed initial exposure age and exposure noise level. however, the risk of hsts due to noise decreases after a number of years of exposure. this is an inherent disadvantage of the concept. it should not be interpreted as if the harmful effects of noise cease to exist. the explanation is that people who have crossed the fence because of age-related threshold shifts are no longer eligible for a risk of hearing loss due to noise fig.1. assessment results of hsts risk for male workers of the workshop at the age of 60. in fig 1, we have: ① risk of hearing loss due to age and noise exposure, 90.8% % (point x); ② risk of hearing loss due to noise exposure, 6.1% % (difference between point x and point y); table 2. a-weighted noise exposure level normalized to a nominal week of 5 eight-hour working days (lex,w) of workers of winding workshop of polymers section of cell 2 department. workshop posts lex,w(db) winding workshop of polymers section monitor of line 8 87.4 monitor of line 12 86.7 precision monitor of line 9 and line 10 85.6 precision monitor of line 7 and line 11 87.4 precision monitor of line 8 87.6 precision monitor of line 12 86.7 precision monitor of line 9 and line 10 86.0 precision monitor of line 7 and line 11 87.6 operator a of line 7 and line 11 87.6 operator b of line 7 and line 11 87.6 operator d of line 7 and line 11 87.9 operator d of line 7 and line 11 86.9 operator e of line 7 and line 11 87.8 operator a of line 9 and line 10 85.8 operator b of line 9 and line 10 85.6 operator d of line 9 and line 10 85.2 operator d of line 9 and line 10 86.3 operator e of line 9 and line 10 85.7 operator of line 13 87.6 operator a of line 8 87.2 operator b of line 8 87.4 operator c of line 8 87.4 operator d of line 8 88.0 operator a of line 12 86.7 operator b of line 12 85.9 operator c of line 12 86.6 operator d of line 12 86.9 table 3. predicted risk of hsts for the male workers in this workshop. age (years) duration (years) h'10 (db) h'50 (db) h'90 (db) h10 (db) h50 (db) h90 (db) n10 (db) n50 (db) n90 (db) risk of hearing loss due to age and noise exposure (%) non-noise exposed population risk of hearing loss due to age (%) risk of hearing loss due to age and noise exposure (%) 30 10 21.5 7.5 -4.5 14.4 2.2 -7.6 7.1 5.3 3.0 40.9 20.5 20.4 40 20 30.2 13.3 -0.5 22.5 7.3 -4.8 7.7 5.9 4.2 62.1 41.1 21.0 50 30 42.6 21.5 4.5 35.3 15.5 -0.3 7.3 6.0 4.8 80.8 67.3 13.5 55 35 50.3 26.6 7.5 43.5 20.8 2.5 6.8 5.9 4.9 86.8 77.6 9.2 60 40 59.1 32.4 10.8 52.9 26.8 5.8 6.1 5.7 5.0 90.8 84.7 6.1 published by atlantis press copyright: the authors 81 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations ③ htla of noise-exposed population; ④ nipts, 10 %; ⑤ assumed fence, 10 db; ⑥ non-noiseexposed population risk of hearing loss due to age, 84.7% % (point y); ⑦ nipts, 50 %; ⑧ htla of non-noise-exposed population; ⑨ nipts, 90 %. (2) risk of occupational noise deafness the predicted risks of occupational noise deafness are given in table 4 and the risk assessment results are expressed in fig. 2. fig.2. assessment results of risk of occupational noise deafness for male workers of the workshop at the age of 60. in fig 2, we have: ① risk of hearing loss due to age and noise exposure, 13.0% % (point x); ② risk of hearing loss due to noise exposure, 2.9% % (difference between point x and point y); ③ htla of noise-exposed population; ④ nipts, 10 %; ⑤ assumed fence, 10 db; ⑥ non-noiseexposed population risk of hearing loss due to age, 10.1% % (point y); ⑦ nipts, 50 %; ⑧ htla of non-noise-exposed population; ⑨ nipts, 90 %. the results show that h and h’ both increase with the duration and n presents an increasing trend on the condition of fixed initial exposure age and exposure noise level. tails of the statistical distributions for 0% < q < 5% and for 95% < q < 100% are unreliable and should not be estimated since few experimental data exist to validate these ranges. (3) hearing loss of different frequencies it shows that the hearing threshold level of 4000 hz is the maximum of h’50 and n50, the curve of audiogram shows a v shape, and h’50 and n50 of every frequency both increase with the duration on the condition of fixed initial exposure age and exposure noise level (table 5, fig. 3 and table 6, fig. 4 ). table 5. predicted results of h’50 of each frequency for the male workers of the workshop age (years) duration (years) h’50 (db) 500 hz 1000 hz 2000 hz 3000 hz 4000 hz 6000 hz 30 10 0.5 0.6 2.0 6.4 9.2 6.7 40 20 1.7 1.9 5.3 11.2 15.4 13.3 50 30 3.6 4.1 9.5 17.7 24.0 22.9 55 35 4.8 5.5 12.1 21.6 29.3 29.0 60 40 6.2 7.1 14.9 26.1 35.3 35.9 table 4. predicted risk of occupational noise deafness for the male workers in this workshop age (years) duration (years) h'10 (db) h'50 (db) h'90 (db) h10 (db) h50 (db) h90 (db) n10 (db) n50 (db) n90 (db) risk of hearing loss due to age and noise exposure (%) non-noise exposed population risk of hearing loss due to age (%) risk of hearing loss due to age and noise exposure (%) 30 10 11.5 1.8 -5.9 9.9 0.9 -6.4 1.6 1.0 0.5 0-5.0 0-5.0 0-5.0 40 20 15.0 4.2 -4.4 13.1 2.9 -5.3 1.9 1.3 0.9 0-5.0 0-5.0 0-5.0 50 30 20.0 7.6 -2.4 18.1 6.1 -3.5 1.9 1.5 1.1 0-5.0 0-5.0 0-5.0 55 35 23.2 9.6 -1.2 21.3 8.1 -2.4 1.8 1.5 1.2 7.3 5.1 2.2 60 40 26.8 12.0 0.1 25.0 10.5 -1.1 1.7 1.5 1.2 13.0 10.1 2.9 published by atlantis press copyright: the authors 82 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations table 6. predicted results of n50 of each frequency for the male workers of the workshop age (years) duration (years) n50（not be modified）(db) 500 hz 1000 hz 2000 hz 3000 hz 4000 hz 6000 hz 30 10 0 0 1.0 4.8 7.1 4.2 40 20 0 0 1.9 5.9 8.1 4.9 50 30 0 0 2.5 6.5 8.8 5.3 55 35 0 0 2.7 6.8 9.0 5.5 60 40 0 0 2.9 7.0 9.2 5.6 0 1000 2000 3000 4000 5000 6000 7000 10 9 8 7 6 5 4 3 2 1 0 n 50 (d b ) frequency (hz) age=30,duration=10 age=40,duration=20 age=50,duration=30 age=55,duration=35 age=60,duration=40 fig.4. predicted results of n50 of each frequency for the male workers of the workshop. 5. conclusions and discussions it should be noted that all sound pressure levels do not consider the effect of hearing protectors which would reduce effective exposure levels. hearing protectors could be worn to reduce the effective exposure levels to protect the hearing of workers exposed to industrial noise. database a is chosen in this application example analysis. if database b is needed, only sweden, norway and the united states data are available, this may be lack of convincing when they are used to assess the risk of hearing loss of chinese population. in further research, a hearing threshold level database of chinese population could be compiled following proper principles reference to the levels of the three countries of database b in iso 1999:2013(e). besides, it is emphasized that for practical situations the accuracy of the prediction of the hearing threshold level of a noise-exposed population will largely be a function of the accuracy of the selected database for htla, so the results are only estimated values in statistics. this iso model is based on statistical data and therefore cannot be applied to the prediction or assessment of the hearing loss of individual persons except in terms of statistical probabilities. the frequencies and fences are chosen based on the hsts defined in the “hearing protection specification for employee in industrial enterprises” and the precondition and definition of occupational noise deafness level according to the gbz 49-2014. the choice of frequencies and fences, though determined primarily by medico-legal considerations, can also be influenced by economic and ethical considerations. this method to assess the noise occupational exposure risk could be used to quantitatively estimate the risk of hearing loss due to noise, guide the enterprises to take effective risk management measures to protect the workers’ health according to the different risk levels estimated, such as making a hearing conservation pro0 1000 2000 3000 4000 5000 6000 7000 35 30 25 20 15 10 5 0 h ' 50 (d b ) frequency (hz) age=30,duration=10 age=40,duration=10 age=50,duration=10 age=55,duration=10 age=60,duration=10 fig.3. predicted results of h’ 50 of each frequency for the male workers of the workshop. published by atlantis press copyright: the authors 83 m. li et al. / application of risk estimation of noise-induced hearing loss method in evaluations gram and implementing the occupational exposure evaluation, engineering control, organization and management, selection and use of hearing protectors, occupational health surveillance, warning for occupational hazards, training and records management, effectiveness evaluation of hearing conservation program, etc. acknowledgements this study was supported by municipal key science and technology support project of tianjin (no. 13zczdsy02300). references [1] iso 1990:1975 (e) acoustics – assessment of occupational noise exposure for hearing conservation purposes [2] iso 1999:1990(e) acousticsdetermination of occupational noise exposure and estimation of noise-induced hearing impairment [3] iso 1999:2013(e) acoustics estimation of noise-induced hearing loss [4] gbz 49-2014 diagnosis of occupational noise-induced deafness [5] robert i davis, wei qiu, nicholas j heyer. the use of the kurtosis metric in the evaluation of occupational hearing loss in workers in china: implications for hearing risk assessment. noise &health, 2012, 14(61) : 330~34 published by atlantis press copyright: the authors 84 1. introduction 2. method 2.1. investigation on the noise occupational exposure situation 2.2. evaluation of the occupational noise exposure 2.3 risk assessment of noise-induced hearing loss 2.3.1. select the appropriate frequency combinations and fences 2.3.2. hearing threshold level associated with age and noise (htlan) 2.3.3. databases for hearing threshold levels associated with age (htla) 2.3.3.1. choice of database 2.3.3.2. database a 2.3.3.3. database b 2.3.4. noise-induced permanent threshold shift 2.3.4.1. calculation of n50 2.3.4.2. statistical distribution of noise-induced permanent threshold shift, n 3. noise-induced risk management 4. application example analysis 4.1. objects 4.1.1. principles to choose objects 4.1.2. objects to assess 4.2. method 4.2.1. noise occupational exposure evaluation 4.2.2. risk of hearing loss due to noise 4.3. results 5. conclusions and discussions acknowledgements references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <feff04180437043f043e043b043704320430043904420435002004420435043704380020043d0430044104420440043e0439043a0438002c00200437043000200434043000200441044a0437043404300432043004420435002000410064006f00620065002000500044004600200434043e043a0443043c0435043d04420438002c0020043c0430043a04410438043c0430043b043d043e0020043f044004380433043e04340435043d04380020043704300020043204380441043e043a043e043a0430044704350441044204320435043d0020043f04350447043004420020043704300020043f044004350434043f0435044704300442043d04300020043f043e04340433043e0442043e0432043a0430002e002000200421044a04370434043004340435043d043804420435002000500044004600200434043e043a0443043c0435043d044204380020043c043e0433043004420020043404300020044104350020043e0442043204300440044f0442002004410020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200441043b0435043404320430044904380020043204350440044104380438002e> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <feff004200720075006700200069006e0064007300740069006c006c0069006e006700650072006e0065002000740069006c0020006100740020006f007000720065007400740065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400650072002c0020006400650072002000620065006400730074002000650067006e006500720020007300690067002000740069006c002000700072006500700072006500730073002d007500640073006b007200690076006e0069006e00670020006100660020006800f8006a0020006b00760061006c0069007400650074002e0020004400650020006f007000720065007400740065006400650020005000440046002d0064006f006b0075006d0065006e0074006500720020006b0061006e002000e50062006e00650073002000690020004100630072006f00620061007400200065006c006c006500720020004100630072006f006200610074002000520065006100640065007200200035002e00300020006f00670020006e0079006500720065002e> /deu <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> /esp <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> /eti <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> /fra <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> /gre <feff03a703c103b703c303b903bc03bf03c003bf03b903ae03c303c403b5002003b103c503c403ad03c2002003c403b903c2002003c103c503b803bc03af03c303b503b903c2002003b303b903b1002003bd03b1002003b403b703bc03b903bf03c503c103b303ae03c303b503c403b5002003ad03b303b303c103b103c603b1002000410064006f006200650020005000440046002003c003bf03c5002003b503af03bd03b103b9002003ba03b103c42019002003b503be03bf03c703ae03bd002003ba03b103c403ac03bb03bb03b703bb03b1002003b303b903b1002003c003c103bf002d03b503ba03c403c503c003c903c403b903ba03ad03c2002003b503c103b303b103c303af03b503c2002003c503c803b703bb03ae03c2002003c003bf03b903cc03c403b703c403b103c2002e0020002003a403b10020005000440046002003ad03b303b303c103b103c603b1002003c003bf03c5002003ad03c703b503c403b5002003b403b703bc03b903bf03c503c103b303ae03c303b503b9002003bc03c003bf03c103bf03cd03bd002003bd03b1002003b103bd03bf03b903c703c403bf03cd03bd002003bc03b5002003c403bf0020004100630072006f006200610074002c002003c403bf002000410064006f00620065002000520065006100640065007200200035002e0030002003ba03b103b9002003bc03b503c403b103b303b503bd03ad03c303c403b503c103b503c2002003b503ba03b403cc03c303b503b903c2002e> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <feff0041006e007600e4006e00640020006400650020006800e4007200200069006e0073007400e4006c006c006e0069006e006700610072006e00610020006f006d002000640075002000760069006c006c00200073006b006100700061002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400200073006f006d002000e400720020006c00e4006d0070006c0069006700610020006600f60072002000700072006500700072006500730073002d007500740073006b00720069006600740020006d006500640020006800f600670020006b00760061006c0069007400650074002e002000200053006b006100700061006400650020005000440046002d0064006f006b0075006d0065006e00740020006b0061006e002000f600700070006e00610073002000690020004100630072006f0062006100740020006f00630068002000410064006f00620065002000520065006100640065007200200035002e00300020006f00630068002000730065006e006100720065002e> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice received 10 september 2015 accepted 20 october 2015 基金资助：国家科技支撑计划（2012baj01b01） 作者简介：张伟（1987-），男，河南商丘人，硕士，工程师。e-mail:weizhang_66@163.com，mobile:86-18510115600 research on urban waterlogging disaster risk assessment based on arcgis and mike flood — a case study on shijiazhuang wei zhang*, yingjun hu, jiazhuo wang, chunyang zhang china academy of urban planning & design, beijing 100037, china e-mail: weizhang_66@163.com abstract urban flood risk assessment supported by accurate and precise modeling could make great contributions to urban flood management. in this paper, an urban drainage system model of shijiazhuang city was built based on the mike21, mike urban and mike 11 modules in mike flood platform in order to provide a storm sewer drainage capacity and waterlogging risk evaluation. the results show that in shijiazhuang, 64.5% of storm sewer drainage criteria is less than 1-year return period storm, and only 28.9% of storm sewer can meet the drainage criteria of 2-year return period storm. the waterlogging risk area grows from 3.2, 15.1 to 18.5 square kilometer when the return period is 10, 20 and 50 years. both the current condition and the drainage system model database indicate the reasons of the drainage issue in shijiazhuang which can be summarized as the rapid urbanization, the low capacity of pipe network and pumping stations, the low construction standard, the lag between the sewer system construction and the demand, the lack of drainage outlet, and the imperfect management system. keywords: waterlogging disaster risk assessment; mike flood; arcgis; shijiazhuang city 基于 arcgis 和 mike flood 的城市内涝风险评估研究 ——以石家庄市为例 张伟*，胡应均，王家卓，张春洋 中国城市规划设计研究院，北京 100037 摘要：本研究基于 mike flood 平台，辅以 arcgis 插值运算、拓扑检查、栅格计算、统计分析等功能， 分别通过 mike21、mike urban、mike 11 模块建立二维地表漫流模型、雨水管网模型和河道模型。通 过模型耦合运算，对石家庄市雨水管网排水能力和内涝风险进行了科学评估，评估结果表明石家庄市雨水 管网排水能力小于 1 年一遇的比例为 64.5%，满足 2 年一遇排水能力的管网比例仅为 28.9%；石家庄市内涝 风险区基于积水时间和积水深度两个要素划分为低风险区、中风险区和高风险区三种类型，随着暴雨重现 期的增大，内涝风险区域逐渐增加，经统计在重现期为 10 年、20 年和 50 的降雨事件下，中风险等级以上 的区域面积分别为 3.2、15.1 和 18.5km2。结合石家庄市现状情况及排水模型数据库分析，产生内涝的原因 主要包括城市硬化面积大，管网及泵站设施排水能力不足，建设标准低，城市排水水系建设滞后，雨水缺 乏退水出路，管理体制不完善。 关键词：内涝风险评估；mike flood；arcgis；石家庄市 journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 226-233 published by atlantis press copyright: the authors 226 1 引言 随着我国城镇化快速发展，由于城市开发强度过 高，大量硬质铺装，改变了原有的自然生态本底和水 文特征。城市每逢遭遇高强度降雨时，大量降雨在短 时间内形成地表径流，大大增加了城市排水压力，经 常出现“城内看海”现象，“逢暴雨必涝”已成为中国城 市的真实写照。2010 年住房和城乡建设部对全国 351 个城市的抽样调查显示，仅 2008-2010 年就有 62%的 城市发生过不同程度的暴雨内涝 [1] 。2013 年 3 月 25 日， 国务院办公厅正式发布了《国务院办公厅关于做好城 市排水防涝设施建设工作的通知（国办发[2013]23 号）》；2013 年 6 月住房和城乡建设部颁布了《城市排 水（雨水）防涝综合规划编制大纲》（以下简称“大纲”）。 目前，城市内涝风险评估应用较多的方法有：历史灾 情数理统计法、指标体系法和水文水力学模型与仿真 模拟法。周魁一提出“历史模型”的概念，论述了基于 历史数据的历史模型方法和在灾害问题中的应用 [2] ； hans de model 利用荷兰 1990-2000 年以及未来 100 年 规划的空间地理信息分析了城市化对洪水产生的影响 [3] ；北美学者利用指标体系从国家、市级尺度对洪水 灾害风险进行了区划和评估 [4-7] ；杜鹃等从孕灾环境的 自然属性、承灾体的社会环境以及致灾因子的特点出 发，构建了洪水灾害综合风险评估指标体系，并将理 论成果运用在湘江流域风险评估中 [8] 。运用模型对城 市内涝风险进行评估在国内起步相对较晚：1970 年以 前主要运用基于物理机制的经验方程满足城市径流计 算的需要；1970-1990 年相继将一些新的理论和方法引 入城市雨洪模型之中，使得雨洪模型应用于城市管网 汇流以及水质模型 [9-10] ；2000 年以后，模型与 gis、 rs 等结合广泛，开始研究城市防洪调度以及内涝积水 的模拟与仿真 [11] 。大纲提出“在排水防涝设施普查的 基础上，推荐使用水力模型对城市现有雨水管网和泵 站等设施进行评估，分析实际排水能力”，“推荐使用 水力模型进行城市内涝风险评估”。目前，国外应用水 力模型进行城市内涝风险评估及排水系统相关规划设 计已较成熟 [12-14] ，国内应用模型辅助进行方案设计尚 处于研究阶段 [15-17] 。国际上用于评估城市内涝风险的 水力学模型有丹麦水利研究院（dhi）开发的 mike 系 列 模 型 [18-19,23] ， 英 国 hr wallingford 公 司 的 infoworks 系列模型[14,20-22]，美国环境保护署（epa） 开发的 swmm 模型[23-24]，欧洲委员会联合研究中心开 发的 lisflood 模型[25-26]等。其中，mike 系列模型 功能全面，运算能力强，并集成了 esri 公司的地理信 息系统技术，广泛应用于国内洪涝灾害风险评估中。 本研究采用 mike flood 水力模型，结合 arcgis 水文分析，以石家庄市中心城区为例，探讨模型在特 大城市尺度上内涝风险评估的应用，为城市排水防涝 决策提供技术支撑。 2 研究区概况 本研究区域包括石家庄市中心城区，东部良村开 发区和化工基地两个片区，具体范围为：北至滹沱河、 西三环、南到南三环路，东至规划铁路东南环线，总 面积约 483.0 km2。 2.1 自然地理 研究区西依太行山，北濒滹沱河，东南部为辽阔 的河北平原。地势较为平坦，自然地形由西北向东南 倾斜，平均坡度约为 1‰，海拔标高在 52-95m。 2.2 河流水系 研究区现有明渠：石津北干渠、石津南支渠、四 支渠、五支渠、桥西明渠、元村明渠、南栗明渠、东 明渠、总退水渠、东环水系、南环水系、引水中线和 汪洋沟。其中石津南支渠、五支渠、桥西明渠、元村 明渠、南栗明渠、东明渠、总退水渠、南环水系和引 水中线 9 条明渠雨季排涝、平时蓄水美化环境；四支 渠沿途用地基本由农业用地改为城市建设用地，因此 现状四支渠基本丧失排水功能；石津北干渠同时承担 农业灌溉和南水北调中线输水任务。 3 模型构建 3.1 城市二维地表漫流模型 （1）地形模型 为进行城市内涝积水情况分析，首先需建立区域 地形模型。本模型源数据采用 1:500 地形图中的等高 线和等高点要素，通过 arcgis 克里金法插值分析， 建立石家庄市数字高程模型(dem),然后将 dem 通过 数据转换，转入 mike21 模型，建立可用于模型计算 的标准文件（文件格式为.dfs2）。 （2）城市下垫面模型 城市降雨地表径流除了与降雨强度有关，还受地 表下垫面影响。本研究采用实地踏勘、大比例尺地形 图（1:500）、高分辨率遥感影像（2013 年 4 月）相结 合的方式，对研究范围内下垫面进行解析，通过综合 分析将石家庄市中心城区、良村开发区和化工基地的 下垫面划分为河流水体、绿地、道路、建设用地与裸 地等五大类。 published by atlantis press copyright: the authors 227 图 1 石家庄市下垫面类型分布图 3.2 雨水管网模型 （1）雨水管网 结合市政管网普查数据，以及近几年新建雨水管 网竣工图，将雨水管网 cad 数据导入 arcgis 数据库， 经拓扑检查后将 arcgis 数据库导入 mike urban 模型中，构建 mike urban 管网模型。 （2）雨水泵站 石家庄市地形受铁路分割较为严重，穿过铁路的 路段基本以地道桥形式存在。目前石家庄市共有排水 泵站 37 座，其中地道桥泵站 36 座，雨污合建泵站 1 座。 图 2 石家庄市雨水管网模型 3.3 河流水系模型 通过 mike11 模型对石家庄市排水河道模型进行 构建，构建要素包括河道、橡胶坝、闸门等，要素属 性主要有河道断面特征，高程信息，橡胶坝及闸门调 度规则等。 published by atlantis press copyright: the authors 228 图 3 石家庄河道模型 3.4 模型耦合 本研究通过 mike21 和 arcgis 建立二维地表模 型；通过 mike urban 和 arcgis 建立雨水管网模 型；通过 mike11 建立河道模型。城市雨水管网排水 能力评估以及内涝风险评估，是在三个模型的基础上， 通过 mike flood 平台进行耦合计算，并通过相关统 计得到的结果，mike flood 城市内涝风险评估模型 构架见图 4。mike flood 模型在耦合运算中，mike urban 模块和 mike21 的水力信息交互是双向的。 通过 mike flood 这种实时的全面的交互式模拟能 够准确呈现地表径流形成的过程，管网溢流的过程， 以及地面积水的过程，能够比较客观、科学的评估城 市内涝风险。 3.5 模型参数 本研究水文模型选用时间~面积（t-a）法作为降 雨径流模型，该模型适用于高度城市化地区。t-a 模 型中主要涉及以下参数： （1）初期损失 初期损失表征降雨开始时，由于地表土壤的蓄存、 植物的吸收、低洼处的截留等造成的没有生成地表径 流的那部分水量。根据研究区现状建设情况，本研究 对雨水系统的初期损失拟采用 5mm。 （2）积水时间 图 4 城市内涝风险评估模型构架图 集水时间是指从降雨落地到流入临近雨水口所经 历的时间。一般来讲，不透水地面的集水时间短，而 绿地等自然地表的集水时间相对较长。根据研究区地 块集水区的大小和形状，相应地选择 5-10 min 作为其 集水时间。 （3）不透水率 不透水率指区域内集水区面积的不透水面积比。 不透水率愈大，相应地，管道中的水流量愈大，产生 内涝灾害的风险从而进一步增加。本研究对不同下垫 面不透水率取值见表 1。 published by atlantis press copyright: the authors 229 表 1 不同下垫面类型不透水率取值表 下垫面类型 不透水率 绿地 15% 道路 85% 建设用地 70% 裸地 35% （4）水文衰减系数：该系数控制透水区的降雨实 际进入河道的量，即径流在汇流过程中的沿途损失量， 且考虑降雨因蒸发、渗透、蓄水等因素造成的水量损 失。根据模型经验，水文衰减系数取 0.9。 3.6 降雨事件 本研究降雨雨型通过对石家庄市 1961-2013 共 53 年降雨资料统计分析，得出石家庄市不同重现期下短 历时降雨强度曲线见图 5。 图 5 石家庄市不同重现期下 2h 设计降雨 4 结果与分析 4.1 城市雨水管网排水能力评估 基于 mike urban 模型，石家庄市雨水管网分 别采用 1 年一遇、2 年一遇、3 年一遇和 5 年一遇降雨 进行评估，通过管网模型综合运算，采用 linkflood 方 法对管网进行评估。 linkflood 评估即以管线压力水头与对应地表高程 值之间的关系进行管网排水能力评估。具体的来说， 当管线压力水头大于对应地表高程值，说明管线排水 能力不足，反之则说明管网排水能力满足相应排水标 准。 经过统计分析，石家庄市雨水管网排水能力小于 1 年一遇的比例为 64.5%，仅 28.9%比例的雨水管网排 水能力大于 2 年一遇。石家庄市现状管网排水能力评 估结果具体见图 6 和表 2。 图 6 石家庄市现状管网排水能力评估图 表 2 石家庄市现状排水能力评估统计表 经评估排 水能力小 于 1 年一遇 的管网 (km) 经评估排水能 力 1-2 年一遇的 管网（包括 1 不 包括 2, km） 经评估排水能 力 2-3 年一遇的 管网（包括 2 不 包括 3, km） 经评估排水能 力 3-5 年一遇的 管网（包括 3 不 包括 5, km） 经评估排水 能力大于等 于于 5 年一 遇的管网 (km) 500.8 51.0 23.0 25.6 175.6 4.2 城市内涝风险评估 基于 mike flood 耦合模拟结果，在不同暴雨重 现期降雨事件下，结合石家庄市地表积水时间和积水 深度，采用双因子评估方法，本文建立了石家庄市内 涝风险矩阵表。总体将内涝风险划分为三类：内涝低 风险区、内涝中风险区和内涝高风险区。 表 3 内涝风险矩阵表 0-30min 30-60min 60-120min 0.15-0.25m 内涝低风险区 内涝中风险区 内涝中风险区 0.25-0.5m 内涝中风险区 内涝高风险区 内涝高风险区 0.5<m 内涝高风险区 内涝高风险区 内涝高风险区 利用 arcgis 软件对模拟结果进行统计，通过对 每个区域内涝积水时间和内涝积水深度进行分析，绘 制出石家庄市 10 年一遇、20 年一遇和 50 年一遇 2h 内涝风险区划图，分别见图 7、图 8 和图 9。 0 5 10 15 20 25 0 10 20 30 40 50 60 70 80 90 100 110 120 降 雨 强 度 （ m m ） 时间（min） p=1a p=2a p=3a p=4a p=5a p=6a p=7a published by atlantis press copyright: the authors 230 图 7 10 年一遇内涝风险区分布示意图 图 8 20 年一遇内涝风险区分布示意图 图 9 50 年一遇内涝风险区分布示意图 由内涝风险区划图可知，石家庄市内涝风险高风 险区主要集中在和平西路与京广铁路交叉区域，东部 环山湖附近区域，其余高风险区主要分布在地道桥区 域。 根据内涝风险区划图，利用 arcgis 栅格计算工 具，统计出不同暴雨重现期下各等级风险区面积，见 表 4。由表 4 可知，随着重现期的增大，内涝风险区 逐渐增加，在重现期为 10 年、20 年和 50 的降雨事件 下，中风险等级以上的区域面积分别为 3.2、15.1 和 18.5km2。 表 4 不同暴雨重现期下内涝风险区面积统计表 内涝低风险区 （km2） 内涝中风险区 （km2） 内涝高风险区 （km2） 10 年一遇 0.9 2.0 1.2 20 年一遇 3.6 8.6 6.5 50 年一遇 3.8 10.1 8.4 4.3 城市内涝原因解析 （1）石家庄市城市硬化面积大，绿地空间少。石 家庄市中心城区二环内总面积 106km2,将二环内城市 用地按照建设用地、道路用地、绿地、水系和裸地五 种类型划分，经过统计建设用地和道路用地比例超过 了 80%，绿地仅为 16%左右。每逢遭遇高强度降雨时， 城市形成大量地表径流，短时间内径流无法排出是形 成城市内涝的直接原因。 （2）雨水管网及泵站排水能力不足，建设标准低， 时间久，汇水分区大。石家庄市现状雨水管道重现期 一般取 1 年，但是在建设中由于城建资金较紧张，部 分雨水管道没有按照规划标准形成，并且在建设过程 中存在错接、乱接现象。石家庄市现状中华大街、平 安大街、建设大街等南北向主干管所承担的雨水汇水 区面积约 3-4km2，而主干管管径多在 2.5m 以下。如果 按照汇水面积 3km2，管网坡度 0.8‰，设计标准 2 年 一遇估算，雨水主管网最大尺寸约 3.6×2.8m 的方涵。 因此现状雨水主干管由于汇水分区大，管径小，再加 上汛期降雨时间短、降雨强度大等特点，排水能力不 足，易引发城市内涝。 （3）城市发展迅速，水系建设滞后，城市排水通 道不足，雨水缺乏退水出路。石家庄市现状老城区基 本在城市二环路以内，为了满足城市排水需要，城内 先后修建了五支渠、桥西明渠、元村明渠、南栗明渠、 东明渠、总退水渠等内河水系，以满足城市排水需求。 随着城市规模的扩大，目前西二环以西区域，东部良 村开发区和化工基地已规划为城市建设区，而在这些 区域城市水系建设严重滞后。西二环以西区域，历史 上存在四支渠排水通道，而现状四支渠路由已逐渐被 填埋、盖板，部分渠段上方已建有大量住宅，完全丧 失了排水能力；东部开发区目前排水主要通过管网排 入环山湖，而环山湖为一封闭水域，下游没有退水出 路，雨水只能通过入渗、蒸发消耗处理，环山湖的容 量完全满足不了开发区的排水需求。 （4）管理体制不完善。石家庄市雨水管理存在“九 龙治水”现象，老城区排水设施就有排水处、二环路管 理处、园林局、交通局、太平河管理处、裕华区、市 场管理、小区物业以及城中村等多家单位管理，都在 自己的管辖范围内行使权利，各个部门都有自己的利 published by atlantis press copyright: the authors 231 益，各自为政，不能从全局出发处理问题。以石家庄 市民心河为例，民心河现状承担着市区景观和排水功 能，景观功能归属民心河水系管理处管辖，排水功能 归属排水管理处管辖，而民心河景观水体水源均来源 与水库调水。由于初期雨水污染问题，每逢降雨为了 保障景观水体水质，民心河上的雨水闸门基本均处于 关闭状态，排水出路被封，管网排水不畅，易引发城 市内涝问题。 5 结论 （1）本研究首次运用数学模型模拟石家庄市城市 排水防涝系统，对于城市基础设施的精细化建设与管 理具有重要意义。通过 mike 21、mike urban、 mike 11 模块建立二维地表漫流模型、雨水管网模型 和河道模型，通过耦合运算对雨水管网排水能力和内 涝风险进行了科学评估，为城市雨水管网管理和内涝 风险控制提供技术支撑。 （2）对石家庄市雨水管网进行了 1 年一遇、2 年 一遇、3 年一遇和 5 年一遇的排水能力评估。采用 link flood 方法对雨水管网排水能力进行评估，石家庄市雨 水管网排水能力小于 1 年一遇的比例为 64.5%，满足 2 年一遇排水能力的管网比例仅为 28.9%。 （3）对石家庄市进行了 10 年一遇、20 年一遇和 50 年一遇的内涝风险评估。采用积水时间和积水深度 双因子对石家庄市内涝风险进行评估，将石家庄市内 涝风险划分为低风险区、中风险区和高风险区，随着 暴雨重现期的增大，内涝风险逐渐增加，在重现期为 10 年、20 年和 50 的降雨事件下，中风险等级以上的 区域面积分别为 3.2、15.1 和 18.5 km2。石家庄市内涝 风险区主要集中在和平西路与京广铁路交叉区域，东 部环山湖附近区域及地道桥区域。 （4）石家庄市产生内涝原因多种。城市硬化面积 大，绿地空间少，易形成地表径流；雨水管网及泵站 设施排水能力不足，建设标准低，雨水主干管管径与 汇水区规模不匹配；城市水系建设滞后城市发展速度， 排水通道不足，雨水缺乏退水出路；此外，城市雨水 管理体制不完善，“九龙治水”现象普遍。 （5）针对石家庄市内涝风险情况建议采取以下措 施进行缓解：新建水系，确保排水出路，对于石家庄 东部地区尤其是京港澳高速以西的地区，规划新建东 部退水明渠，保障东部排水出路；恢复四支渠，优化 排水分区，提高管网建设标准，尽量减少五支渠排水 压力，对于汇水分区过大雨水主干管，采用增加主干 系统等方式进行改造，原则上一个子排水分区的面积 在 2-5 平方公里左右；科学管理，统筹协作，明确相 关管理部门职责，科学解决水系景观与排水功能之间 的矛盾，加强民心河闸门汛期调度。 参考文献 [1] 章林伟. 海绵城市建设概论. 给水排水,2015,06:1-7. zhang linwei. the conspectus of sponge city construction. water & wastewater engineering, 2015,06:1-7. [2] 周 魁 一 . “ 历 史 模 型 ” 与 灾 害 研 究 . 自 然 灾 害 学 报,2002,11(1):10-14. zhou kuiyi. historical model and method of disaster research. journal of natural disasters, 2002,11(1):10-14. [3] hans de moel, jeroen c, aerts j h,et al.development of flood exposure in the netherlands during the 20th and 21st century. global environmental change, 2011, 21:620-627. [4] cutter s l, boruff b j, shirley w l. social vulnerability to environmental hazards. social science quarterly, 2003, 84:242-261. [5] cutter s l, mitchell j t, scott m s. revealing the vulnerability of people and places: a case study of georgetown county, south carolina. annals of the association of american geographers, 2000, 90:713-737. [6] [6] clark g, moser s, ratick s,et al． assessing the vulnerability of coastal communities to extreme storms: the case of revere, ma, usa. mitigation and adaptation strategies for global change, 1998(3):59-82. [7] rygel l, o’sullivan d, yarnal b. a method for constructing a social vulnerability index: an application to hurricane storm surges in a developed country. mitigation and adaptation strategies for global change, 2006,11:741-764. [8] 杜鹃,何飞,史培军.湘江流域洪水灾害综合风险评估. 自然灾害学报,2006,15(6):38-44. du juan, he fei, shi peijun. integrated flood risk assessment of xiangjiang river basin in china. journal of natural disasters, 2006,15(6):38-44. [9] 岑 国 平 . 城 市 雨 水 径 流 计 算 模 型 ［ j ］ . 水 利 学 报,1990(10):68-75. cen guoping. a model to simulate storm water runoff in urban area. journal of hydraulic engineering, 1990(10):68-75. [10] 周玉文,赵洪宾.城市雨水径流模型研究［j］.中国给水 published by atlantis press copyright: the authors 232 排水,1997,13(4):4-6. zhou yuwen, zhao hongbin. theses and researches a study on urban storm runoff model. china water and wastewater, 1997,13(4):4-6. [11] 仇劲卫,李娜,程晓陶等.天津城区暴雨沥涝仿真模拟系 统［j］.水利学报,2000 (11):34-42. qiu jinwei, li na, cheng xiaotao, et al. the simulation system for heavy rainfall in tianjin city. journal of hydraulic engineering, 2000 (11):34-42. [12] artina s, bolognesi a, liserra t, et al. simulation of a storm sewer network in industrial area: comparison between models calibrated through experimental data. environ model software, 2007, 22(8):1221-1228. [13] samantaray d, chatterjee c, singh r, et al. flood risk modeling for optimal rice planning for delta region of mahanadi river basin in india. natural hazards, 2015, 76(1):347-372. [14] rubinato m, shucksmith j, saul a j, et al. comparison between infoworks hydraulic results and a physical model of an urban drainage system.. water science & technology a journal of the international association on water pollution research, 2013, 68(2):372-379. [15] 谢映霞. 城市排水与内涝灾害防治规划相关问题研究. 中国给水排水, 2013, (17):105-108. xie yingxia. urban drainage and waterlogging disaster prevention planning. china water & wastewater, 2013, (17):105-108. [16] 谢莹莹. 城市排水管网系统模拟方法和应用[d]. 同济 大学, 2007. xie yingying. the method and application of urban sewer system simulation [d]. tongji university, 2007. [17] 王乾勋, 赵树旗, 周玉文等. 基于建模技术对城市排水 防涝规划方案的探讨——以深圳市沙头角片区为例. 给水排水, 2015, (3):34-38. wang qianxun, zhao shuqi, zhou yuwen et al. probe into urban drainage and waterlogging prevention plan based on modeling-case study on shataukok district of shenzhen. water & wastewater engineering, 2015, (3):34-38. [18] 姚双龙. 基于 mike flood 的城市排水系统模拟方法 研究[d]. 北京工业大学, 2012. yao shuanglong. urban drainage system modeling research based on mike flood [d]. beijing university of technology,2012. [19] 初祁,彭定志,徐宗学,孟德娟,甄婷婷,姜其贵. 基于 mike 11 和 mike 21 的城市暴雨洪涝灾害风险分析. 北京师范大学学报(自然科学版),2014,05:446-451. chu qi, peng dingzhi, xu zongxue, meng dejuan, zhen tingting, jiang qigui. risk analysis of urban flooding by using mike 11 and mike 21. journal of beijing normal university (natural science), 2014,05:446-451. [20] liang x, wang p, lv y, et al. impact simulation of inland water level on drainage capacity of pipeline network. urban roads bridges & flood control, 2014. [21] wang j, zhang c, rong h, et al. application of hydraulic model in planning design of drainage system. environmental science & technology, 2014. [22] 刘志生 , 李晶 , 崔凯 , 张莉 , 任晓强 . 基于 arcgis 与 infoworks 的城市排水系统模拟研究 . 中国给水排 水,2013,21:144-147. liu zhisheng, li jing, cui kai, zhang li, ren xiaoqiang. simulation of urban drainage system based on arcgis and infoworks. china water & wastewater, 2013,21:144-147. [23] luo f l, yuan y, fan j b. application of swmm and mike11 coupled model in urban tidal river network. china rural water & hydropower, 2013. [24] ma j, li j, xu y, et al. solution to the problem on urban drainage system overflow in rainy season with storm water management model(swmm). water purification technology, 2012. [25] florian pappenberger, keith beven, ad de roo, et al. uncertainty analysis of the rainfall runoff model lisflood within the generalized likelihood uncertainty estimation (glue). international journal of river basin management, 2004, 2(2):123-133. [26] van d k j m, younis j, de roo a p j. lisflood: a gis-based distributed model for river basin scale water balance and flood simulation.. international journal of geographical information science, 2010, 24(2):189-212. published by atlantis press copyright: the authors 233 1 引言 2 研究区概况 2.1 自然地理 2.2 河流水系 3 模型构建 3.1 城市二维地表漫流模型 3.2 雨水管网模型 3.3 河流水系模型 3.4 模型耦合 3.5 模型参数 3.6 降雨事件 4 结果与分析 4.1 城市雨水管网排水能力评估 4.2 城市内涝风险评估 4.3 城市内涝原因解析 5 结论 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word volume 13, issue 2-4 journal of risk analysis and crisis response, 2023, 13(2), 148-160 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i2.362 148 article a study on the impact of fiscal pressure on the deviation of local government budget revenues chen ye 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: 1443795359@qq.com; tel.: +86-0851-18005159270 received: march 13, 2023; accepted: may 31, 2023; published: june 30, 2023 abstract: this paper examines the impact of fiscal pressure on the degree of deviation from budgetary revenue based on provincial panel data from 2011 to 2020. the results show that fiscal pressure has a positive impact on the deviation of local government budget revenues, with the impact of fiscal pressure on the deviation of non-tax revenues significantly higher than that of tax revenues; the positive impact of fiscal pressure on the deviation of local government budget revenues is more obvious in regions with higher reliance on land finance; the proportion of tax revenues in fiscal revenues negatively moderates the impact of fiscal pressure on the deviation of local government budget revenues. the impact of fiscal pressure on the degree of deviation of local government budgetary revenue is negatively moderated by the share of tax revenue in fiscal revenue. keywords: budget revenue deviation; fiscal pressure; modern budgetary system 1. introduction since 2020, due to the dual impact of downward pressure on the economy and the intensification of policies to reduce taxes and fees, the nation's fiscal balance has become "tight", and the pressure on local fiscal balance has gradually increased. in addition, the impact of the new pneumonia epidemic has further reduced fiscal revenues, and the contradiction between local government revenues and expenditures has become more prominent. this, coupled with insufficient budget coordination and management, the government has not yet firmly established a sense of living a tight life, the budget is not binding enough, the efficiency of resource allocation and use needs to be improved, and the budget is still not open and transparent enough, has affected the coordination and sustainability of financial resources. it is crucial for local governments to accurately forecast economic growth trends and prepare budgets in a scientific manner, while coordinating the balance between fiscal revenue and governance objectives. this is a huge test for local governments' budgetary capacity. as an important tool of financial management, the budget is a central reflection of the government's activities, and the country's ability to govern depends to a large extent on its budgetary capacity. the report of the 19th party congress clearly points out the establishment of a "modern budget system that is comprehensive, standardized, transparent, scientific and strong in restraint", and the report of the 20th party congress also mentions "improving the modern budget system". the modern budget system requires that budget preparation and budget execution should be consistent, chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 149 but the reality is that since the reform of the tax sharing system, the gap between the chinese government's budget and final revenue has been gradually widening, and this gap is reflected in the deviation between the budget voted by the local people's congresses at the beginning of the year and the final budget executed by local governments at the end of the year. this deviation is referred to as the budget deviation and is mainly based on the over-recovery of revenues. based on the above realistic background, it is of certain theoretical value and practical significance to study the impact of financial pressure on the deviation of income from the budget and accounts. 2. review of the literature 2.1. studies related to financial stress research on the measurement of fiscal stress: most scholars use the fiscal gap to measure fiscal stress, such as those who use the ratio of general budget gap to regional gdp (wang zhuhua et al., 2017) [1], some use the ratio of general budget gap to general budget revenue to measure (shouwei qi et al., 2020) [2], and a few scholars measure local governments' fiscal pressure (zhu jun, 2018) [3]. in addition to this, some scholars use fiscal stress shocks to measure, for example, some scholars use the abolition of agricultural tax reform to measure the change in fiscal stress (yu jingwen et al., 2018) [4] , or the income tax sharing reform as a quasi-natural experiment, using the relative loss of local government income tax before and after the reform as a measure of the amount of change in fiscal stress (xu chao et al., 2020) [5]. research on the impact of fiscal pressure on local government behavior: most of the current literature examining local government behavior under fiscal pressure focuses on local governments' strategies to increase revenue and reduce pressure. fiscal pressure usually induces local governments to increase tax effort and strengthen tax regulation (chen, 2016; chen, 2017; gao, zhengbin et al. 2019; zhao, y. et al. 2019; li, guangzhong et al. 2020) [6], while fiscal pressure also intensifies local governments' debt raising behavior, which leads to higher local debt risk (hong, yuan et al. 2018; mao, jie et al. 2020) [7]. fiscal pressure also induces local governments' preference for "land finance" (luo biliang 2010; wang jian et al. 2019; huang lingxiang et al. 2020; huang siming et al. 2020) [8]. 2.2. studies related to budget deviations research on the impact of fiscal transparency on budget deviations: scholars have shown that increased fiscal transparency significantly reduces deviations in local budget revenues and increases deviations in local budget expenditures, and that it has a significantly stronger inhibitory effect on deviations in non-tax revenues than tax revenues (xiao, peng and fan, 2021) [9]; some scholars have assessed the impact of the new budget law on deviations in local government budgets and found that some scholars assessed the impact of the new budget law on the deviation of the budget of local governments, and found that the new budget law reduces the deviation of budget revenues and expenditures by strengthening the scientific nature of revenue budgeting and strengthening the constraints of budget execution (li and liu, 2020) [10], while some scholars further investigated whether there are differences in the impact of the new budget law on the deviation of the budget under different budget environments. some scholars have shown that state audit supervision enhances fiscal transparency through hard constraints and strong accountability mechanisms, thus chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 150 affecting the deviation of fiscal expenditure from the budget and that fiscal transparency has a mediating effect on the deviation of expenditure from the budget forecast (yang zaiting and wang jinxiu, 2020) [11]. research on the influence of political incentives on budget deviations: scholars have shown that local officials are strongly incentivized by finance and promotion and may achieve gdp over planned growth, but there is a positive correlation between gdp over planned growth and budget deviations, while in the budget execution process taxation departments will increase their taxation efforts under the pressure of taxation task assessment, thus achieving fiscal over-collection (feng hui and shen zhaozhang, 2015) [12], on the basis of which some scholars found that local governments would strategically underestimate the planned gdp used in budget preparation due to political incentives, while continuously over-recovering in the budget year to stimulate local economic development, resulting in high actual gdp in the final accounts, making a positive correlation between overplanned gdp growth rate and non-tax revenue pre-final deviation (ma haitao et al., 2017) [13]; there are also scholars point out that the relationship between tax planning and budget deviation is not a simple linear one, but an inverted u-shaped one, with tax planning exceeding a certain standard creating a reverse incentive for tax authorities (feng hui and shen zhaozhang, 2015) [14]. 2.3. study on the impact of financial pressure on budget deviations it has been shown in the literature that since the tax sharing reform, fiscal power has gradually been concentrated to the central government, while the service power has gradually shifted down to the local level, and local governments will try to improve the efficiency of tax collection and management to ensure fiscal revenue under fiscal pressure, with fiscal over-revenue positively correlated with the hand captured by the central government (feng hui and shen zhaozhang, 2015) [12]. under a fiscal decentralization system, vertical fiscal imbalances of local governments have a positive effect on the degree of local revenue deviation, while promotion incentives amplify the extent of this effect (xi yu and sun yudong, 2021) [15]. if fiscal revenues are subdivided into tax and non-tax revenues, the fiscal expenditure gap rate is similarly positively related to the degree of non-tax revenue budget deviation (ma et al., 2017) [13]. in addition to the above influencing factors, fiscal competition will force local governments to undergo greater fiscal pressure, and local governments may use land concessions to pursue fiscal over-recovery and adjust their beginning budgets, resulting in a negative correlation between fiscal pressure and budget deviation within a budget year, but expenditure pressure across budget years does not involve budget adjustments, hence a positive correlation between the two (feng hui, 2017) [16]. in summary, the existing literature is rich in theoretical studies and attribution analysis on fiscal pressure and budget deviations, but there are fewer empirical studies on deviations in local government budget revenues from the perspective of fiscal pressure. secondly, we investigate the different effects of fiscal pressure on the deviation of tax revenue and non-tax revenue, as well as the effects on the deviation of budget revenue under different degrees of dependence on land finance; thirdly, we find that optimizing the structure of fiscal revenue can help local governments to correct the mechanism of budget deviation. 3. theoretical analysis and research hypothesis chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 151 since the 1994 tax-sharing reform, the central government has gradually established its central position in the allocation of financial resources with the continuous upward shift of financial powers and the gradual devolution of ministry powers, and the central government's ability to macro-control the economy has been enhanced. this reform has led to a year-on-year increase in the proportion of the central government's revenue in the total national fiscal revenue, but the corresponding ministry powers are still concentrated in local governments, which makes local governments bear an excessively heavy expenditure responsibility and triggers enormous financial pressure. local governments, as suppliers of public goods, are also responsible for economic development. faced with enormous financial pressure, local governments are naturally motivated to increase their revenues to meet their expenditure needs. in order to attract capital inflows and promote regional economic growth, local governments may increase their fiscal revenues through land concessions, which, unlike tax revenues, lack statutory bases and constraints and are less difficult to realize. in addition, after the financial crisis in 2008, there was a proliferation of financing platforms set up by local governments, through which local governments issued urban investment bonds to raise largescale debts. at the same time, local governments enjoy considerable discretionary power over nontax revenues, and the "high degree of freedom" in the collection and management of such revenues is a powerful incentive for local governments to pursue extra-budgetary revenues. prior to the enactment of the new budget law, local governments enjoyed discretionary power over overrecovery of revenue, which was outside the supervision of the national people's congress and could easily exacerbate deviations from the budget. the above-mentioned behaviors of local governments in the face of financial pressure have further increased the extent of budget deviations. this has led to a lack of supervision and restraint in budget execution by local governments, resulting in widespread budget deviations. accordingly, this paper proposes the hypothesis that fiscal pressure exacerbates the extent of deviations in local government budget revenues. 4. model, data and variable selection 4.1. model setting this paper uses chinese provincial panel data for the period 2011-2020, and the following model is constructed to explore the relationship between fiscal pressure and deviations from budgetary revenue. budget = α + α fispressure + βx + μ + δ + ε (1) where budget denotes the fiscal budget revenue deviation in province i in year t; fispressure denotes local government fiscal pressure; x is a control variable; μ is an individual effect; δ is a time effect; and ε is a random disturbance term. the issue of endogeneity is an important issue that must be considered in the empirical model. based on the hausman test results, this paper uses a panel fixed effects model, which can overcome the omitted variable bias to a certain extent. however, using the fixed effects model alone is not sufficient, and this paper further uses a systematic gmm model to address the endogeneity problem. 4.2. selection of variables 4.2.1. explained and explanatory variables chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 152 the core explanatory variable in this paper is fispressure, which is commonly used in academia to measure the fiscal pressure faced by local governments, either by the fiscal gap or by fiscal dependency. the explanatory variable budget is the difference between the final revenue and the budgeted revenue. it is worth noting that the budgeted revenue is the adjusted budgeted revenue of the local government, not the budgeted revenue that was approved by the national people's congress at the beginning of the year. see table 1 for the calculation formula. 4.2.2. other variables in addition to fiscal pressure, there are several factors that can affect the extent of deviation from the budget. the control variables selected in this paper include: regional gdp per capita (lpgdp), unemployment rate (unemployment), urbanization rate (urban) and the timing of the enactment of the new budget law. in addition, the paper also chooses to use land transfer revenue (ltf) and fiscal revenue structure (structure) as the threshold and moderating variables respectively. the names and meanings of the variables are shown in table 1. table 1. name and meaning of each variable. name of the variable implication symbols deviation of income from the financial budget degree of deviation from budgeted revenue = (|fiscal final revenue fiscal budget revenue|) / fiscal budget revenue budget financial pressures fiscal pressure = (general public budget expenditure general public budget revenue) / general public budget revenue fispressure regional gdp per capita natural logarithm of regional gdp per capita lpgdp unemployment rate -unemployment urbanization rate urbanization rate = year-end urban population/total population urban when the new budget law was enacted 0 prior to 2015, 1 in 2015 and thereafter budget law revenue from land concessions land premium income/general public accounts income llr structure of fiscal revenue local government tax revenue / local government revenue structure 4.3. data sources in view of the uniformity of data caliber and availability, this paper selects panel data of 31 chinese provinces from 2011 to 2020 to empirically analyze the impact of fiscal pressure on the deviation of budgetary revenue. data on the main variables are obtained from the china statistical yearbook and the china finance yearbook, the moderating variables from the china fiscal transparency assessment, and the threshold variables from the china land and resources yearbook. 4.4. descriptive statistics table 2 shows the descriptive statistics for each variable for the years 2011-2020.the data of each variable are normally distributed. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 153 table 2. descriptive statistics. name of the variable observations average standard deviation minimum value maximum value deviation of income from the financial budget 310 0.03 0.06 0.00 0.48 financial pressures 310 0.49 0.20 0.10 0.93 regional gdp per capita 310 10.78 0.45 9.69 12.01 unemployment rate 310 3.24 0.64 1.20 4.61 urbanization rate 310 58.05 13.14 22.81 89.60 when the new budget law was enacted 310 0.60 0.49 0.00 1.00 revenue from land concessions 310 0.45 0.23 0.00 1.18 structure of fiscal revenue 310 0.74 0.08 0.60 0.96 note: the vif values of each variable are less than 10, and there is no multicollinearity problem. 5. analysis of empirical results 5.1. baseline regression results table 3. regression results of the effect of financial pressure on the deviation of income from the budget and accounts. name of the variable （1） （2） （3） （4） fispressure 0.2888** (0.1279) 0.4207*** (0.1332) 0.4135*** (0.1356) 0.4192*** (0.1343) lpgdp - -0.0066 (0.0505) - -0.0119 (0.0542) unemployment - -0.0056 (0.0092) - -0.0059 (0.0093) urban - -0.0042* (0.0024) - -0.0045* (0.0024) budget law -- -0.0044*** (0.0010) 0.0062 (0.0068) constant term -0.0939 (0.0624) 0.1739 (0.4375) 0.0981 (0.0656) 0.2479 (0.4865) number of observations 310 310 310 310 adjusted r2 0.0217 0.1191 0.1238 0.1172 individual fixed effects control control control control time fixed effects control control control control note: robust standard errors in brackets; ***, ** and * represent significant at the 1%, 5% and 10% levels respectively. column (1) of table 3 controls for year and region fixed effects only, and the coefficient of fispressure is 0.2888, which passes the 1% significance test, indicating that fiscal pressure significantly increases the degree of deviation from pre-final revenue. in order to eliminate the influence of other possible factors on the results, columns (2)-(4) gradually add control variables to the first column. column (2), after adding the control variable of economic and social development, the coefficient of fispressure is 0.4207, which passes the 1% significance test, and column (3), after adding only the control variable of when the new budget law was issued. the coefficient of fispressure is 0.4135 at the 1% level of significance after adding all control variables in column (3), and 0.4192 at the 1% level of significance after adding all control variables in column (4). the results chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 154 of the benchmark regression also suggest that local governments have a "soft constraint" on budget execution under china's fiscal decentralization system. 5.2. robustness tests 5.2.1. endogenous problems considering that local government budgeting in china uses the base method and is often characterized by incremental budgeting in the process of budgeting, resulting in the current period budgeting being highly susceptible to the influence of the previous period's final accounts (feng, 2017) [14], this paper uses the firstand second-order lagged terms of the explanatory variables as instrumental variables and uses a systematic gmm model for estimation to mitigate the estimation bias caused by endogeneity . compared to differential gmm, systematic gmm can better handle the endogeneity of panel data and is also more suitable for small sample data. to avoid bias in the estimation due to excessive use of instrumental variables, the lag and collapse options are chosen to be added to the econometric model. budget = α + α budget + α budget + α fispressure + βx + μ + δ + ε (2) the regression results, as shown in table 4, show that the deviation of pre-final revenue in the previous period has a significant positive effect on the deviation of pre-final revenue in the current period, demonstrating the asymptotic nature of local government revenue budgeting (liu, yuan and li, 2022) [17]. the coefficients of the core explanatory variables remain significantly positive, further validating the robustness of the baseline regression model. the results of the ar(1) and ar(2) tests show that there is first-order serial autocorrelation but not second-order serial autocorrelation in the series of the model error terms. from the hansen test results, the instrumental variables selected in this paper are appropriate and there is no over-identification problem in the model, and the diff-inhansen test also indicates that the instrumental variables are exogenous table 4. endogeneity test. name of the variable sys-gmm estimates l1. budget 0.1645*** (0.0533) l2. budget 0.0303 (0.0405) fispressure 0.1108** (0.0481) ar(1) 0.0070 ar(2) 0.5300 hansen test 0.9330 diff-in-hansen test 0.9360 control variables yes individual fixed effects control time fixed effects control number of observations 248 note: robust standard errors in parentheses; ***, **, * represent significant at the 1%, 5% and 10% levels respectively; hansen's test and ar(1) and ar(2) are reported as statistical p-values. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 155 5.2.2. different measures of financial stress at present, it is generally accepted in china's academic circles that fiscal pressure refers to the gap between fiscal revenue and expenditure, and fiscal pressure is usually measured by the scale of fiscal expenditure exceeding fiscal revenue, but in practice, scholars will adopt different measurement methods according to their own research perspectives, so there is no completely uniform standard. theoretically, the fiscal pressure of local governments can be measured by the degree of dependence on the central government's transfer payments on the one hand, while on the other hand, the local governments' own fiscal revenue and expenditure gap can also reflect the fiscal pressure situation. based on this, this paper draws on the method of wang jhuhua et al. (2017) [1] and uses the on-budget fiscal revenue and expenditure gap/gdp (deficit) method to re-measure the fiscal pressure of local governments in order to test the robustness of the baseline regression results. the regression results are shown in table 5. the coefficient of the core explanatory variable deficit remains significantly positive, indicating that fiscal pressure can significantly increase the degree of fiscal budget revenue deviation, once again validating the robustness of the benchmark regression results. table 5. tests for substitution of explanatory variables. name of the variable deviation of income from pre-final accounts deficit 0.1255*** (0.0208) control variables yes individual fixed effects control time fixed effects control number of observations 310 r2 0.0508 5.2.3. other robustness tests table 6. other robustness tests. name of the variable considering the impact of the new crown pneumonia outbreak in 2020 exclusion of municipalities from the sample fispressure 0.4409*** (0.1522) 0.3545*** (0.0986) control variables yes yes individual fixed effects control control time fixed effects control control number of observations 279 270 adjusted r2 0.1176 0.1024 note: ***, **, * indicate significant at the 1%, 5% and 10% levels respectively. first, excluding the impact of the new crown pneumonia epidemic in 2020, which suddenly broke out on a large scale in 2020 and had a significant impact on the production and livelihood of the regions, local governments may not be able to adjust their budgets in a timely and effective manner in the face of the sudden epidemic, considering the uncertainty of the impact of the epidemic on economic activities. in this paper, we exclude the sample data of 2020 and re-run the baseline regression, the results are shown in column (1) of table 6. fiscal pressure still has a significant positive effect on the deviation of budget revenue, indicating that the sample selection does not affect the chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 156 research results of this paper. secondly, the influence of municipalities directly under the central government is considered. considering the special administrative status of the municipality, which may have an impact on the benchmark regression, this paper excludes the municipality from the sample and re-runs the benchmark regression. the results are shown in column (2) of table 6, and the coefficient of fispressure is still significantly positive after excluding part of the sample, indicating that the findings of this paper are robust. 5.3. heterogeneity analysis 5.3.1. deviation of income from pre-final accounts when local governments are under financial pressure, they tend to increase non-tax revenue to relieve the pressure. the reason is that if local governments simply increase tax revenue, it will lead to a higher tax burden in the region and thus discourage the inflow of capital, which is not conducive to long-term economic development. in addition, because non-tax revenues are not subject to the same statutory requirements as tax revenues, local governments have considerable autonomy in their collection and management, and thus have a greater incentive to deviate from the budgeted revenues. the results of this paper are shown in table 7, which shows that for every unit increase in fiscal pressure, the deviation from the budget of tax revenue increases by 11.24%, while that of non-tax revenue increases by 21.49%, with the degree of deviation from the budget of non-tax revenue being significantly higher than that of tax revenue. this also confirms the tendency of local governments to behave in the face of fiscal pressure as explained in the previous section. table 7. heterogeneity analysis of tax revenue and non-tax revenue. name of the variable tax revenue non-tax revenue fispressure 0.1124*** (0.0426) 0.2149* (0.1217) control variables yes yes individual fixed effects control control time fixed effects control control number of observations 310 310 r2 0.0312 0.0159 note: ***, **, * indicate significant at the 1%, 5% and 10% levels respectively. in order to further explore the impact of financial pressure on positive and negative deviations of budgetary revenue, this paper considers deviations of budgetary revenue greater than zero as positive deviations and deviations of budgetary revenue less than zero as negative deviations, and thus identifies the overand under-recovery behavior in fiscal revenue. table 8. heterogeneity test for over-collection and short-collection behavior. name of the variable overcharging short collection behaviour fispressure 0.3427* (0.1796) 0.9799** (0.4244) control variables yes yes individual fixed effects control control time fixed effects control control number of observations 243 67 adjusted r2 0.1146 0.4139 note: ***, **, * indicate significant at the 1%, 5% and 10% levels respectively. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 157 the empirical results are shown in table 8. the coefficients of fispressure in columns (1) and (2) are both significantly positive, implying that fiscal pressure triggers over-revenue behavior of most local governments while also inducing short-revenue behavior of some local governments. it is worth noting that the degree of impact on short-revenue behavior of local governments is greater than that on over-revenue behavior, possibly because when facing fiscal pressure, due to the greater rigidity of fiscal expenditure local governments may take into account the growth in expenditure when preparing their revenue budgets and thus increase their revenue budgets, but the actual revenue in the following year often fails to meet expectations, instead increasing the degree of deviation from the budget and accounts. 5.3.2. land finance since the reform of the taxation system, there has been a serious mismatch between the central government and local governments in terms of financial and administrative powers, and the budgetary pressure faced by local governments has been increasing. as a result, land revenue has become the main source of non-tax revenue for local governments in recent years, while this part of the funds is at the "free disposal" of local governments, and governments at all levels naturally favor land finance, but this has to a certain extent increased the degree of deviation from the budget revenue. the high dependence on land finance means that local governments tend to become inertially dependent on land finance, and given the switching costs associated with adjusting the revenue structure, local governments tend to draw more heavily on land finance when they are under financial pressure, as it is easier to obtain revenue through the sale of land, but this practice exacerbates the degree of deviation from budget revenue. based on this, this paper uses land concession revenue as the threshold variable to test the threshold effect on the deviation of pre-determined revenue, and the results are shown in table 9. from the f-statistic and p-value, we can see that the single threshold passes the 10% significance test, and the threshold value of land concession revenue is 0.3122. the effect of fiscal pressure on the deviation of pre-determined revenue differs at different levels of land concession revenue: compared to the effect of fiscal pressure on the deviation of budget revenue differs across different levels of land concession revenue: compared to regions with lower land concession revenue, regions with higher land concession revenue have a higher degree of influence on the deviation of budget revenue. table 9. threshold effect test. threshold variable：revenue from land concessions（llr） -models f-value p-value number of bs threshold values 1% 5% 10% deviation of income from the pre-account single threshold 11.69* 0.0967 300 21.5707 14.4066 11.6555 double threshold 3.82 0.3900 300 13.3301 9.6961 7.7646 three-fold threshold 0.48 0.9467 300 17.7591 11.9920 8.5288 note: * indicates significant at the 10% level. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 158 6. further analysis the previous empirical analysis has shown that fiscal pressure has increased the degree of deviation from the budget, so how to alleviate the fiscal pressure of local governments to reduce the degree of deviation from the budget. theoretically, local governments have not yet established a main tax system, and the structure of the local tax system still needs to be improved, so it is difficult to maintain the stability and sustainability of tax revenues. this section examines the adjustment mechanism of fiscal pressure affecting the deviation of budgetary revenue from the perspective of fiscal revenue structure. the results are shown in table 10, where the coefficients of the fispressure and structure crossmultiplier variables are significantly negative, indicating that the optimization of local government revenue structure can effectively reduce the degree of deviation from the budget revenue, and that, due to the higher degree of transparency as well as standardization of tax revenue, when tax revenue accounts for a larger proportion of fiscal revenue, local governments have less room to draw on nontax revenue, thus this effectively reduces budget deviations. the increase in the share of tax revenue in fiscal revenue not only ensures the quality of fiscal revenue, but also corrects the bias of local governments to achieve fiscal overspending through non-tax revenue. table 10. tests for moderating effects. name of the variable deviation of income from the pre-account fispressure 1.3049*** (0.2846) structure 0.8100*** (0.1427) fispressure * structure -1.6789*** (0.3510) control variables yes individual fixed effects control time fixed effects control number of observations 310 adjusted r2 0.2969 note: ***, **, * indicate significant at the 1%, 5% and 10% levels respectively. 7. conclusions and policy recommendations this paper finds the influence of fiscal pressure on the deviation degree of local government budget revenue through empirical analysis, and draws the following conclusions: first, fiscal pressure shows a positive influence on the deviation degree of local government budget revenue, among which the influence of fiscal pressure on the deviation degree of non-tax revenue budget is significantly higher than that of tax revenue, probably because it is a common practice for local governments to expand non-tax revenue when facing fiscal pressure; second, the positive influence of fiscal pressure on the deviation degree of local government budget revenue is more obvious in the positive effect of fiscal pressure on the degree of deviation of local government's budget revenue is more obvious in areas with higher land grant revenue, further illustrating the dependence of local governments on land finance; thirdly, the proportion of tax revenue to fiscal revenue will negatively adjust the effect of fiscal pressure on the degree of deviation of local government's budget revenue, and the optimization of fiscal revenue structure can effectively alleviate the degree of deviation of budget revenue. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 159 the findings of this paper have certain practical significance in further reducing the degree of revenue deviation from the budget and accounts, improving the scientific nature of revenue budgeting and effectively constraining budget execution. first, the financial and administrative powers between the central government and local governments should be clarified. we can try to build a local tax system with property tax as the main body, give local governments more autonomy in real estate tax collection by "fully authorizing" them, actively and steadily promote real estate tax reform and improve the taxation system of grassroots governments, and combine the "fee-to-tax" reform with the "fee-to-tax" reform by clearing fees and establishing taxes. the reform should be combined with the "fee to tax" reform to expand the scope of local taxes by clearing fees and establishing taxes, helping local governments to obtain stable tax revenues, appropriately decentralizing financial powers while appropriately delegating ministerial powers, and establishing a decentralization structure that matches financial and ministerial powers. second, enhance the binding force of budget execution. focus on enhancing the status of the budget department in the finance sector, gradually improve the standards and norms of budget preparation, while focusing on the scientific and seriousness of budget preparation, so that every expenditure has to have a budget in the first place, and strictly prohibit over-budget and no-budget arrangements for expenditure; attempts can be made to add budget execution efficiency to the appraisal system of local government officials to help local governments improve their budget execution. third, strengthen the supervision of budget management. strengthen the npc's budget review and supervision function, establish an all-round and whole-process supervision mechanism, strictly audit budget adjustments, not adopt any unscientific and unreasonable budget adjustments, strictly implement the budget approved by the npc, and once the budget is approved, no adjustment is allowed except for amendments through legal procedures, improve the audit supervision and audit scope of the budget, further improve financial transparency, and give full play to social supervision in local the role of social supervision in local government budget management is further enhanced, and power is fully exposed to the sun. funding: this research received no external funding. conflicts of interest: the author declares no conflict of interest. references [1] wang zhuhua. fiscal pressure, government spending competition and local government debt an analysis based on a spatial econometric model[j]. review of economics and management, 2017,33(05):74-82. doi: https://doi.org/10.13962/j.cnki.37-1486/f.2017.05.010. [2] shouwei qi, mao hui, zhang jidong. fiscal pressure, economic stimulus and attracting investment with land empirical evidence based on micro data of industrial land[j]. finance and trade economics, 2020,41(04):20-34. doi: https://doi.org/10.19795/j.cnki.cn11-1166/f.20200409.005. [3] zhu jun, xu zhiwei. fiscal decentralization, inter-regional competition and economic volatility in china[j]. economic research, 2018,53(01):21-34. [4] yu jingwen, chen xiaoguang, gong liutang. how does financial pressure affect the supply of public services by county governments? [j]. financial research, 2018(01):18-35. [5] xu chao,pang yumeng,liu di. local fiscal pressure and government expenditure efficiency a quasinatural experimental analysis based on income tax sharing reform[j]. economic research, 2020,55(06):138154. [6] chen xiaoguang. fiscal pressure, tax collection and regional inequality[j]. china social science, 2016(04):5370+206. chen ye / journal of risk analysis and crisis response, 2023, 13(2), 148-160 doi: https://doi.org/10.54560/jracr.v13i2.362 160 [7] xu junwei, mao jie, guan xinghua. local government hidden debt reconceptualization based on the precise definition of financing platform companies and the perspective of financial potentials[j]. management world, 2020,36(09):37-59. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2020.0136. [8] huang s. m., wang q. fiscal pressure, land finance and land supply strategies of prefecture-level municipal governments [j]. contemporary finance and economics, 2020(12):26-38. doi: https://doi.org/10.13676/j.cnki.cn36-1030/f.2020.12.004. [9] xiao peng, fan rong. analysis of the impact of local fiscal transparency on the deviation of fiscal budget and final accounts[j]. journal of central university of finance and economics, 2021(03):3-14. doi: https://doi.org/10.19681/j.cnki.jcufe.2021.03.002. [10] liu yuan, li jianjun. governance effects of social supervision on deviations from local government budgets[j]. local finance research, 2022(03):13-24. [11] yang zaiting, wang jinxiu. state audit supervision, fiscal transparency and local budget deviation[j]. modern economic inquiry, 2020(02):33-40. doi: https://doi.org/10.13891/j.cnki.mer.2020.02.005. [12] feng hui, shen zhaozhang. promotion incentives, grasping hands, and local fiscal over-recovery[j]. contemporary finance and economics, 2015(06):35-44. doi: https://doi.org/10.13676/j.cnki.cn361030/f.2015.06.005. [13] ma haitao, bai yanfeng, kuang xingxing. political incentives, lack of financial resources and deviation of local non-tax revenue budget an analysis based on inter-provincial panel data[j]. local finance research, 2017(01):21-29. [14] feng hui, shen zhaozhang. political incentives, taxation plans and local fiscal revenue budget deviations-an analysis based on inter-provincial dynamic panel data model[j]. journal of yunnan university of finance and economics, 2015,31(03):27-39. doi: https://doi.org/10.16537/j.cnki.jynufe.000004. [15] xi yu, sun yudong. economic development, vertical fiscal imbalance and budgetary revenue deviation an empirical analysis based on threshold regression of provincial panel data in china from 2007-2017[j]. finance and trade research, 2021,32(06):1-11. doi: https://doi.org/10.19337/j.cnki.34-1093/f.2021.06.001. [16] feng hui. local government competition, fiscal pressure and the science of local budgeting an analysis based on provincial panel data[j]. contemporary finance and economics, 2017(05):34-43. doi: https://doi.org/10.13676/j.cnki.cn36-1030/f.2017.05.004. [17] li jianjun, liu yuan. can the new budget law reduce deviations from local government budgets and final accounts? --evidence from municipalities and states in sichuan province[j]. fiscal research, 2020(07):39-52. doi: https://doi.org/10.19477/j.cnki.11-1077/f.2020.07.004. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style research progress on risk acceptance criteria of long-distance oil and gas transportation pipeline wu zongzhi china academy of safety science and technology beijing 100012, china zhang shengzhu* china academy of safety science and technology beijing 100012, china e-mail: zhangshengzh5168@163.com abstract for acceptable risk and risk acceptance criteria, concepts and definition methods were proposed. determination principles and application status including alarp were investigated. differences between long-distance oil/gas transportation pipeline and other industrial facilities in risk tendency, degree of risk control, environment pollution risk, and specific risk in pipeline shutdown were analyzed. determination methods and standards of acceptable criteria of individual, social, property and environment risks were discussed. individual and social risk standards of the pipeline were proposed according to china’s actualities. keywords: long-distance oil and gas transportation pipeline, acceptable risk, acceptance criteria, standards 油气长输管道风险接受准则研究进展 吴宗之 张圣柱* 中国安全生产科学研究院，北京 100012 摘要：在对可接受风险和风险接受准则的研究进展进行讨论的基础上，归纳了其概念和界定方法。探讨了 alarp 原则等风险可接受准则的确定原则，以及各原则的应用现状。分析了油气长输管道与其他工业设 施在风险倾向、风险控制严格程度、环境污染风险，以及管道特有的停输风险等方面存在的差异。讨论了 个人风险、社会风险、财产风险和环境风险的可接受准则确定方法及标准。结合中国实际情况，提出了油 气长输管道的个人风险和社会风险标准。 关键词：油气长输管道，可接受风险，接受准则，标准 * corresponding author journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 188-194 published by atlantis press copyright: the authors 188 administrateur texte tapé à la machine received 14 november 2011 administrateur texte tapé à la machine accepted 10 february 2012 administrateur texte tapé à la machine risk acceptance criteria of long-distance pipeline 1. 引言 管道运输具有高效率、低成本且不易受运输周围环 境影响等优势，已成为油气运输的首选方案。油品 和天然气具有易燃、易爆等特性，并且管道储运工 艺复杂，一旦出现故障容易引发火灾、爆炸等事 故，导致严重人员伤亡、经济损失和环境污染。如 1999 年 6 月 10 日美国华盛顿州汽油管道爆炸事故造 成 3 人死亡、8 人受伤、1 处住宅及 1 座水处理厂受 到严重破坏，2000 年 8 月 19 日美国新墨西哥州天然 气管道爆炸事故造成 12 人死亡，2010 年 5 月 2 日山 东胶州原油管道爆裂事故导致数百吨原油泄漏， 2011 年 5 月 31 日陕西铜川天然气管道闪爆事故迫使 上万人被紧急疏散等。尽管事故原因不同，但这一 系列事故却反映出管道风险未能被有效控制的问 题。 自 20 世纪 70 年代起，风险评价被广泛应用于油 气长输管道系统中，通过估算长输管道系统的风险 值，并以既定的风险可接受水平作为评价依据，判 断风险值是否可以接受，以此确定管道的安全水 平。确定合理的风险可接受水平是风险评价的关 键，风险可接受水平过高会增加企业和社会成本， 过低则会导致风险被忽视，给管道安全运行埋下隐 患。对于油气长输管道，赵忠刚 1, 2 、秦岭 3 、梁运 枚 4 等从风险接受准则及其确定方法，管道事故个 人可接受风险、社会可接受风险和环境可接受风险 等方面开展了研究，但至今仍没有专门用于评价油 气管道事故可接受风险的统一标准，另一方面，风 险可接受水平的确定是基于历史统计数据之上的， 而我国对于长输管道事故的统计资料有限，亟待开 展进一步研究。 目前我国油气长输干线管道已达 8 万公里，事故 时有发生。随着境外油气资源的进一步落实，西气 东输三线、四线等的构思与建设，以及北油南运、 川气东送等规划，未来几年将建设一批长输管道， 到“十二五”规划期末全国长输油气管道总里程将 超过 10 万公里 5 。结合我国油气长输管道实际情 况，借鉴相关国家成熟的风险可接受水平确定方 法，探讨适用于我国的标准，对于新建管道选线和 在役管道风险评价，保障管道安全运行具有重要意 义。 2. 可接受风险与风险接受准则 20 世纪 60 年底末，在核能、化工等领域相关学者开 始对风险的可接受性进行讨论。1968 年 chauncey starr 6 提出了“怎样安全才够安全”(how safe is safe enough)这一问题，并根据“揭示优先法”得出了不 同风险的社会可接受性度量。1976 年 lowrance 7 在 其 所 著 的 《 of acceptable risk: science and the determination of safety》一书中，提出了“只有认为 一个事物风险可接受时，它才是安全的”这一观 点。1981 年 fischhoff 8 在所著的《acceptable risk》 一书中对可接受风险进行了深入探讨，认为可接受 风险问题是一个决策问题，其可接受程度与对风险 的认识程度、甘愿冒险的程度、风险的可控程度、 灾害是否具有毁灭性以及恐惧心理等因素相关。 2001 年 marszal 9 提出可接受风险是道德、法律和经 济上的一种平衡。英国健康和安全委员会(hse)对可 接受风险的定义为任何可能会被风险影响的人，为 了生活或工作的目的，如果风险控制机制不变，准 备接受的风险即为可接受风险。2000 年岑慧贤 10 对 可接受风险的概念及其界定方法进行了研究，认为 其是社会公众根据主观愿望对风险水平的接受程 度。 1988 年英国健康和安全委员会(hse)发表了核电 站可容忍风险标准的研究报告《核电站的容许风 险》，报告中指出一条“地方不容许线”，在该线 以上的风险被认为是不容许的风险，在同年的《可 容忍风险》报告中，hse 认为低于 10 -6 的死亡概率 是可以忽略的。挪威石油理事会将“可接受性准 则”定义为“用于表达 工程的风险可接受性水 平”，它认为其“指导方针”是“可接受性准则陈 述了人员伤亡、环境损害和财产、商业利益受损的 风险的特点”。风险接受准则可定义为在规定的时 间或系统的某一行为阶段内可接受的风险等级。它 是进行风险评价和采取风险控制措施的依据，风险 接受准则可通过定量和定性的方式进行表示。 近年来，国内一些学者对核电工程、建筑工程、 自然灾害等领域的可接受风险进行了研究并取得了 积极成果 11, 12, 13 。对于油气长输管道可接受风险及 风险接受准则，我国学者主要是在研究国外相关领 域可接受风险确定依据和方法的基础上，探讨了适 用于我国的风险接受准则，对于可接受风险的实质 和符合我国实际的风险接受准则，仍需开展深入研 究。 3. 风险可接受准则的确定原则 可接受风险水平过高会使风险被忽视，导致风险控 制措施不能有效实施。可接受风险水平也不是越小 越好，在各领域中尤其是工程领域，采取风险控制 措施就意味着人员和资金的投入，势必会使整个系 统的成本增加，而且当风险降低到一定程度时，风 险的进一步降低往往会以成本呈指数增加为代价。 应将风险确定在一个与经济社会发展相协调，合 published by atlantis press copyright: the authors 189 wu zongzhi, zhang shengzhu 理、可行的水平。风险可接受准则的制定需要满足 工程中的实际需要，符合公认的行为标准，并能从 自身活动和相关事故中得到经验 14, 15 。目前风险可 接受准则有以下几种： 3.1. alarp (as low as reasonably practice)原 则。 该原则通过容许上限和容许下限将风险划分为 3 个 区域(如图 1 所示)：风险水平超过容许上限即为不可 接受风险，除特殊情况外该风险无论如何都不能被 接受，相关活动必须停止；风险水平低于容许下限 则风险被认为是可以接受的，无需采取风险控制措 施；风险水平处于容许上限和容许下限之间时，可 采取合理可行的控制措施尽可能的来降低风险 16, 17 。 图 1 alarp 原则 3.2. gamab (globalement au moins aussi bon) 原则。 该原则指出新系统的风险与已经接受的现存系统的 风险相比较，新系统的风险水平至少要与现存系统 的风险水平大致相当，也称为比较原则。 3.3. mem (minimum endogenous mortality)原 则。 该原则指出新活动所带来的危险不应比人们在日常 生活中所接触的其它活动的风险有明显增加，一般 认为，新活动所带来的风险增加量不应超过 1%。 目前国内外普遍采用 alarp 原则，即最低合理 可行原则，对于处于不可接受区域的风险，必须停 止相关活动或采取措施使风险降低到 alarp 区 域，对于处于 alarp 区域的风险，可采取合理可 行措施使之尽可能降低。 4. 风险可接受准则的确定方法及标准 油气长输管道具有距离长、分布范围广的特点，在 整条管道长度上具有不同的风险倾向，而其他工业 设施主要集中在一个区域，风险差异较小；油气长 输管道在选线过程中，迫于周围环境实际情况，极 小部分管道可能会靠近中高密度人员场所或高敏感 区域，而其他新建工业设施在选址过程中对于风险 的控制则更加严格；油气长输管道系统故障容易导 致泄漏、火灾、爆炸等事故，与其他工业设施事故 一样，会导致人员伤亡和经济损失，但由于长输管 道多采用高压力、大管径管道进行输送，一旦发生 事故容易造成油品和天然气的大量泄漏，环境污染 更为严重。另一方面，其他工业设施仅是生产一种 商品或提供一项服务，而油气长输管道却是社会经 济和人民生活命脉，一旦发生较长时间停输将会导 致严重后果，公众对停输风险厌恶度高。下面对油 气长输管道风险的生命风险(个人风险和社会风险)、 财产风险和环境风险的可接受准则确定方法及标准 进行探讨。 4.1. 生命风险可接受准则确定方法及标准 4.1.1. 个人风险 个人风险是指既定区域内某一固定位置上的人员个 体死亡概率，与人员的数量和人员是否处于此位置 无关，只与该位置的空间坐标有关。目前个人风险 可用 afr、ai、风险矩阵等方法进行确定。 afr (annual fatality risk)即年死亡风险，指一 个人在一年内的死亡概率，其计算一般通过划分网 格的方法来实现，需要根据危险源的全部事故情 景、所有的天气等级、气象状况进行综合计算并累 加得到。 ai (aggregated inidcator)即聚合指数，指单位国 民生产总值的平均死亡率，对于油气长输管道可通 过式(1)进行定义： p p p ai n gnp  (1) 式中：np—油气长输管道行业死亡人数； gnpp—油气长输管道行业所占国民生产总 值。 风险矩阵通过将事故发生的可能性和严重度相叠 加得到风险等级 18 ，如表 1 所示： published by atlantis press copyright: the authors 190 risk acceptance criteria of long-distance pipeline 表 1 风险评价矩阵 事故后果 事故可能性 风险程度 ⅴ (可忽略的) ⅳ (轻度的) ⅲ (中度的) ⅱ (严重的) ⅰ (灾难的) a(频繁) m m h h h b(很可能) s m m h h c(有时) l s m h h d(极少) l l s m h e(不可能) l l s m m 其中：h 类为“不可接受”风险；m 类为“不希望”的风险；s 类为“有条件的可接受”的风险；l 类为 “可接受”的风险。 目前个人风险多用 afr 表示，英国、荷兰、澳 大利亚等国家将个人死亡率低于 10 -6 每年作为可接 受风险标准，高于 10 -5 作为不可接受标准，10 -5 至 10 -6 之间的风险需加强监控和整改 19 。我国尚无统一 标准，国家“十五”科技攻关项目《城市公共安全 规划技术、方法和程序研究》成果 20, 21 ，所建议的 个人风险容许标准如表 2 所示。 参照国外研究机构的建议标准和《危险化学品重 大危险源监督管理暂行规定》，结合油气长输管道 的特点，本文提出了油气长输管道的个人风险标 准，如表 3 所示。 表 2 建议的个人风险容许标准 应用对象 典型对象 最大可容许风险/ 每年 标准说明 高敏感或高密度 场所 党政机关、军事禁区、军事管理区、 古迹、学校、医院、敬老院、居民 区、大型体育场馆、大型商场、影剧 院、大型宾馆饭店等 1×10 -6 在高敏感或高密度场所不接受 1×10 -6 的个人风险。1×10-6 每年的 个人风险等值线不应进入该区 域。 中密度场所 零星居民、办公场所、劳动密集型工 厂、小型商场(商店)、小型体育及文化 娱乐场所等 1×10 -5 1×10 -5 每年的个人风险等值线不 应进入该区域。 低密度场所 技术密集型工厂、公园、广场 1×10 -4 1×10 -4 每年的个人风险等值线不 应进入该区域。 企业内部 企业内部 1×10 -3 厂区内不应出现 1×10-3 每年的个 人风险等值线。 表 3 建议的油气长输管道个人风险标准 对象类型 典型对象 最大可容许风险 /每年 1. 高敏感场所 2. 重要目标 3. 特殊高密度场所 1. 如学校、医院、幼儿园、养老院等 2. 如党政机关、军事管理区、文物保护单位等 3. 如大型体育场、大型交通枢纽等 310 -7 1. 居住类高密度场所 2. 公众聚集类高密度场所 1. 如居民区、宾馆、度假村等 2. 如办公场所、商场、饭店、娱乐场所等 110 -6 中密度场所 如零星居民、办公场所、劳动密集型工厂、小型商场 (商店)、小型体育及文化娱乐场所等 5×10 -6 低密度场所 如技术密集型工厂、公园、广场等 5×10 -5 published by atlantis press copyright: the authors 191 wu zongzhi, zhang shengzhu 4.1.2. 社会风险 社会风险是跟人员密度相关的函数，指的是能够引 起大于等于 n 人死亡的事故累积频率 f，代表单位 时间(年)内的死亡人数，与空间位置无关，只受环境 人口密度的影响。目前社会风险可用 fn 曲线、 pll、viih 等方法进行确定。 fn 曲线是以死亡人数(n)对应各种事件后果发生 频率累加值(f)作图的分布图形。fn 曲线将整个风 险区域划分为三部分，包括不可容许区、alarp 区 和可容许区，根据实际得到危险源的社会风险曲线 落入的区域，采取不同的风险控制措施。 pll (potential loss of life)即潜在死亡人数，表 示一个区域内可能因事故导致的全部死亡人数，可 通过式(2)进行定义： 1 n i i i pll ir d     (2) 式中：iri—第 i 个网格的个人风险； di—第 i 个网格的人口数量； n—网格的总数目。 viih (value injuries and ill health )即受伤和不健 康值，它假设一个人的死亡与一定数量的人员受伤 或健康损害相当，从而综合考虑事故造成的人员受 伤和健康损害等影响，解决了人员死亡以外的风险 量化问题。 目前社会风险多用 fn 曲线表示，荷兰、英国、 香港等国家和地区的社会风险推荐标准如图 2 所 示，其中，荷兰风险标准最为严格，英国风险标准 主要针对现有危险设施，标准相对较低，香港风险 标准介于两者之间。 累 积 频 率 f (次 / 年 ) 死亡人数n/个 荷兰 香港英国 1e-8 1e-9 1e-7 1e-5 1e-6 1e-4 1000100101 1e-3 0.01 0.10 1.00 图 2 社会风险可接受标准 油气长输管道路线长、范围广，在管道新建过程 中，征地是最重要的问题，标准过高会增加土地的 占用量，致使建设成本过高，不利于长输管道的发 展。另一方面，与发达国家相比，中国在油气长输 管道的技术、设备和安全管理等方面还存在差距， 达不到其安全运行水平，需要逐步完善、提高。本 文建议采用香港的社会风险标准作为油气长输管道 的可容许社会风险标准(如图 2 所示)。 4.2. 财产风险可接受准则确定方法及标准 财产风险一般指整条管道或特定一段的总风险。类 似于 fn 曲线，财产风险可接受准则可通过 pl 曲线 进行表示。p 为财产损失累积频率，l 为财产损失 值，pl 曲线将整个区域划分为两部分，曲线以上为 不可接受风险区，应采取措施进行控制，以下为可 接受风险区，可不采取控制措施。秦岭等 3 提出的 财产风险可接受标准如图 3 所示，各概率的意义如 表 4 所示。 l p 50 1e-2 财 产 损 失 概 率 财产损失值(%) 1e-5 1e-6 1e-4 100151 1e-3 图 3 财产风险可接受标准 表 4 各概率所代表意义 概率等级 意义 10 -6 事故后果严重，且无事先征兆 10 -5 后果严重但事故发生前发出警告 10 -4 事故后果严重，人员能够逃离，财产损 失率约为 50% 10 -3 后果不严重，财产损失率约为 10% 10 -2 损失很小，如 1% 4.3. 环境风险可接受准则确定方法及标准 环境破坏主要是因油品和气体泄漏造成的。可通过 环境破坏程度和可接受的频率确定环境风险可接受 published by atlantis press copyright: the authors 192 risk acceptance criteria of long-distance pipeline 准则，秦岭等 3 提出的环境风险可接受标准如表 5 所 示。 表 5 环境风险可接受标准 环境破坏等级 平均恢复期(a) 可接受频率 较小 0.5 1×10 -1 中等 2 2.5×10 -2 重大 5 1×10 -2 严重 20 2.5×10 -3 现阶段，风险可接受准则的确定主要是基于定量 风险计算之上的，个人风险多用 afr 表示，社会风 险多用 fn 曲线表示，财产风险可用 pl 曲线表示， 环境风险可用可接受频率与平均恢复期进行表示， 既有通用标准，也有仅用于油气长输管道的标准。 标准的合理性与基础数据的丰富程度和可信性密 切相关，目前，标准主要来源于西方发达国家，是 在对大量真实数据进行统计分析基础上确定的，在 这些国家已被证明具有可操作性。与西方发达国家 相比，我国在经济、社会发展状况以及自然环境方 面存在较大差异，另一方面，我国对相关数据的统 计滞后，且完整性较低，仍需对风险可接受准则的 确定方法和适用于我国的标准进行研究。 在借鉴工业设施风险标准的基础上，本文提出了 油气长输管道的个人风险标准和社会风险标准与学 者共同探讨。公众对停输风险的厌恶度高，且油气 长输管道泄漏事故造成的环境污染一般较为严重， 有待根据我国实际情况确定合理的停输和环境风险 标准。 5. 建议 风险接受准则研究是油气长输管道风险研究的关键 方面，合理的风险标准是指导油气长输管道系统风 险控制的重要依据。针对存在的问题，对今后的油 气长输管道风险接受准则研究提出以下建议： (1) 目前我国在用的相关风险接受准则及标准均 来源于西方发达国家，准则的表述形式和标准的高 低是否适用于我国仍需讨论。建议政府安全生产主 管部门在研究的基础上提出我国的油气长输管道生 命、环境等风险标准。 (2) 政府安全生产主管部门和管道运营商共同建 立管道事故数据库，详细记录我国境内不同区域的 管道失效数据，确保所记录数据的规范性和通用 性，为建立在概率统计基础上的风险接受准则的合 理确定提供依据。 (3) 受技术水平和管理水平的影响，油气长输管 道风险可接受准则是动态变化的，需要根据实际情 况对标准适时做出调整，我国幅员辽阔、管道所经 过区域地质条件复杂多样，可探讨适用于不同地段 的差异性标准。 致谢 文章得到了国家“863”计划项目(2009aa11z211) 的资助。 参考文献 1. z. z. gang, y. a. lin, z. x. fen, research progress of the acceptability of the risk assessment of oil and gas pipelines, technology supervision in petroleum industry 5 (2005) 94–98. 赵忠刚, 姚安林, 赵学芬, 油气管道可接受性风险评估 的研究进展[j], 石油工业技术监督, 2005, 5: 94-98. 2. z. z. gang, y. a. lin, l. y. lv, l. d. quan, defining acceptable risk standard values of oil and gas pipeline, journal of southwest petroleum university(science & technology edition) 30(2) (2008) 147–150. 赵忠刚, 姚安林, 李又绿, 李大全, 油气管道可接受风险 标准值的界定研究[j], 西南石油大学学报(自然科学 版), 2008, 30(2): 147-150. 3. q. ling, c. l. qiong, c. kang, acceptance criterion on risk in oil and gas pipeline, natural gas and oil 25(2) (2007) 15–17. 秦岭, 陈利琼, 陈康, 油气管道风险接受准则[j], 天然气 与石油, 2007, 25(2): 15-17. 4. l. y. mei, l. hui, y. wei, study progress in risk acceptance criteria of long-distance natural gas transmission pipeline, oil & gas storage and transportation 28(10) (2009) 6–9. 梁运枚, 骆晖, 杨伟, 天然气长输管道风险接受准则的 研究进展[j], 油气储运, 2009, 28(10): 6-9. 5. 华建敏. 十二五末我长输油气管道总里程超 10 万公里 [eb/ol]. 2010-09-25. http://news.163.com/10/0925/20/6hf1q16500014jb6.ht ml 6. s. chauncey, social benefit versus technological risk, science, 165(3899) (1969) 1232–1238. 7. lowrance (eds.), of acceptable risk: science and the determination of safety. (william kaufmann inc., california, 1976) 8. fischhoff b, lichtenstein s, slovic p, et al. (eds.), acceptable risk. (cambridge university press, new york, 1981) 9. marszal e m, tolerable risk guidelines, isa transactions, 40(4) (2001) 391-399 10. c. h. xian, f. h. yang, w. q. he, concept of acceptable risk, chongqing environment science 22(3) (2000) 18– 19. published by atlantis press copyright: the authors 193 wu zongzhi, zhang shengzhu 岑慧贤, 房怀阳, 吴群河, 可接受风险的界定方法探讨 [j], 重庆环境科学, 2000, 22(3): 18-19. 11. s. z. gang, z. z. yi, x. lan, study of risk acceptance of nuclear power, chinese journal of nuclear science and engineering 22(3) (2002) 193–198. 时振刚, 张作义, 薛澜, 核能风险接受性研究[j], 核科学 与工程, 2002, 22(3): 193-198. 12. g. z. lin, j. z. ke, l. x. hui, study on the risk acceptance criteria in construction, j. xi’an univ. of arch. & tech.(natural science edition) 40(1) (2008) 76–79. 郭章林, 贾增科, 李晓慧, 建筑工程施工风险接受准则 研究[j], 西安建筑科技大学学报(自然科学版), 2008, 40(1): 76-79. 13. s. z. hai, l. x. lin, acceptable risk and disaster research, progress in geography 29(1) (2010) 23–30. 尚志海, 刘希林, 可接受风险与灾害研究[j], 地理科学 进展, 2010, 29(1): 23-30. 14. l. b. yan, study on criteria of acceptable risk, master thesis, zhenjiang: jiangsu university, (2010), pp.5-23. 李宝岩, 可接受风险标准研究, 江苏大学硕士学位论文 [d], 江苏大学, 2010, 5-23. 15. y. fei, t. tao, research and development of safety technology in signaling system of rail transit, china safety science journal 15(6) (2005) 94–99. 燕飞, 唐涛. 轨道交通信号系统安全技术的发展和研究 现状[j]. 中国安全科学学报, 2005, 15(6): 94-99 16. en50126, railway applications-the specification and demonstration of reliability, availability, maintainability and safety (rams) [s], 1999 17. qingdao safety engineering research institute of china petroleum & chemical corporation (eds.), a guidance for quantitative risk assessment in the petrochemical plant. (china petrochemical press, beijing, 2007) 中国石油化工股份有限公司青岛安全工程研究院, 石 化装置定量风险评估指南[m], 北京: 中国石化出版社, 2007. 18. w. z. zhi, a preliminary discussion on establishment of the system of industrial accident risk management in china, china safety science journal 15(s) (1995) 183– 186. 吴宗之, 建立我国工业事故风险管理制度的探讨[j], 中 国安全科学学报, 1995, 5(s): 183-186. 19. w. z. zhi, study on methods and contents for land use safety planning, journal of safety and environment 4(6) (2004) 86–90. 吴宗之, 城市土地使用安全规划的方法与内容探讨[j], 安全与环境学报, 2004, 4(6): 86-90. 20. china academy of safety science and technology, urban public safety planning techniques, methods and procedures, national scientific and technological research report of the tenth five-year of china, 2004. 中国安全生产科学研究院, 城市公共安全规划技术、 方法与程序研究[r], “十五”国家科技攻关课题研究 报告, 2004. 21. w. z. zhi, d. y. quan, w. l. jun, et al, quantitative area risk assessment method and its application in land use safety planning for major hazard installation, engineering science 8(4) (2006) 46–49. 吴宗之, 多英全, 魏利军等, 区域定量风险评价方法及 其在城市重大危险源安全规划中的应用[j], 中国工程 科学, 2006, 8(4): 46-49. published by atlantis press copyright: the authors 194 atlantis press journal style tiems drevs initiative: disaster resilience establishment for vulnerable societies k. harald drager the international emergency management society tiems østvangveien 29, 0588 oslo, norway e-mail: khdrager@online.no thomas v. robertson thinking teams 13220 nw 33rd ave, vancouver, wa 98685, usa e-mail: tvrobertson@yahoo.com abstract due to changing climate, population growth, and other factors the loss of life and damage to physical property has increased significantly in developing countries. the effects of these disasters can cause long term impacts on productivity and growth throughout the world. although it is in the international community’s humanitarian and economic interests to provide aid to vulnerable populations, attempts to do this are often hampered by the challenges of coordination across cultures and organizations. it is estimated that $1 spent on preventative measures saves up to $7 in emergency response and rehabilitation; however coordination can be particularly difficult without the pressure of on-going emergency. global preparedness can be improved by: (1) establishing a global pool of emergency management expertise, with shared frames of reference (terminology, knowledge base, models, frameworks, methodologies) and shared best practices; and (2) creating a foundation for better global collaboration by developing and testing models of local participatory governance in international collaboration for disaster preparedness. keywords: tiems, disaster risk reduction, resilience, vulnerable, emergency management, global preparedness 1. world disasters in 2013 from world disasters report 2013 (international federation of red cross and red crescent societies, 2013) (international federation of red cross and red crescent societies, 2013), an overview of disasters in 2012 is shown in figures 1 and 2 below. the following table 1 summarizes the disaster picture and consequences for the period 2003 – 2012. floods and transports are the most severe natural and technological disasters respectively, in the period 2003 – 2012. while a positive downward trend is seen for transport accidents in this period, the flood disasters seem to be steady but fluctuating over the period. most disasters happened in asia during the period, with a slowly decreasing trend in asia and africa. asia also had the most deaths after disasters, 54% of the world total in 2012, while africa had 22% in 2012. deaths in asia have been fluctuating over the period 2003 – 2012, while it has been quite steady in africa. journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 133-140 published by atlantis press copyright: the authors 133 willieb typewritten text received 25 may 2014 willieb typewritten text accepted 20 june 2014 willieb typewritten text the population in asia has been most affected, with 65% of the area’s population impacted by disasters in 2012, with a steady trend in the period 2003 – 2012. 31% of africa’s population was affected in 2012, which also was a steady trend in the period 2003 – 2012. the cost of damage was highest in americas with 66% of the world total in 2012, with 18% in asia in 2012. the high americas total is mostly due the heavy costs after the hurricane sandy. the overall picture is that floods are the number one disaster, and asia is the continent most affected by disasters. these figures and conclusions should be very clear indicators of how and where to prioritize activities for disaster risk reduction. the fact that the population of the globe is rapidly growing concomitantly with the ever-increasing impact of natural and man-made disasters is not coincidental. civil strife, stress on the environment, immense demand for energy, and rapid rise of economies in developing countries are the result of this convergence. it is not surprising, then, that the loss of life and damage to physical property has increased so significantly in developing countries (scholtens, 2008, bandyk, 2010). 2. observations from a recent disaster at the end of 2013 we were shocked by the devastating consequences of the typhoon haiyan that struck the philippines on the 8th of november 2013. the terrible death toll increased each day and reached more than 6000 and more than one million people were displaced and in desperate need of help and support. the media channels brought us shocking pictures every day, so the tragedy was exposed in all its cruelty in our homes. a glimpse into the vulnerability of the philippines is provided in recent personal correspondence from ricardo q. cabugao, jr., chairperson of the information technology education department of batanes state college, basco, philippines. mr. cabugao, who is also a part of the school of disaster risk reduction and management council. writes that batanes state college, located on a small island in the northernmost tip of the philippines, is prone to natural disasters including earthquakes, typhoons, landslides, tsunamis, and tidal waves. for example, on september 21, 2013, the island was struck by super typhoon odette (usagi), leaving damaged roads, bridges, and lines of communication. natural disasters have displaced the ivatan indigenous people on the island, forcing them to relocate to mindanao and visaya. typhoon haiyan brought an outpouring of international aid. but what about helping the philippines improve their resilience through preparedness programs before the disasters strike? typhoons come every year, maybe not as powerful as haiyan, but the need for shelters and early warning must have been evident to all for a long time before the recent tragedy. if we also add earthquakes and landslides and other disasters hitting the philippines from time to time, the question is, could international table 1. disaster characteristics for the period 2003 2012. disaster charac-teristics most severe second worst natural disaster type floods, 39*% (2012); steady/fluctuating 2003-2012 wind storms, 25*%; steady/fluctuating 2003-2012 technical disaster type transport accidents, 72*%; decreasing 2003-2012 industrial accidents, 13*%; decreasing 2003-2012 disaster location asia, 38*% (2012) slowly decreasing 2003-2012 africa, 22*% (2012) slowly decreasing 2003-2012 disaster deaths asia, 54*% (2012 down) fluctuating 2003-2012 africa, 19*% (2012) steady 2003-2012 affected population asia, 65**% (2012 down) steady 2003-2012 africa, 31**% (2012 up) steady 2003-2012 cost of damage americas, 66%* (2012 up) fluctuating 2003-2012 asia, 18%* (2012 down) fluctuating 2003-2012 * percent of world total ** percent of country population published by atlantis press copyright: the authors 134 aid to improve early warning and preparedness do a better job of saving lives and property than aid provided only after disasters such as haiyan strike? perhaps our global community can learn from this recent tragedy, to invest more help and assistance in those areas we know will be hit, before disasters strike. 3. what do vulnerable nations need? disasters such as those experienced in the philippines lead to the following conclusions: 1. building disaster preparedness will reduce the devastating consequences of natural disasters in vulnerable countries 2. vulnerable countries with limited resources need to get professional help with their disaster preparedness 3. disaster preparedness should be based on risk assessments identifying potential threats and means to deal with these threats 4. the goal must be to provide help building up local expertise to “help the country to help themselves in a critical situation” 5. education and training of locals, offering courses on all levels from universities to primary schools, are the key to create a risk management culture in the vulnerable countries 6. media could play a role putting a focus on the preparedness challenges in vulnerable countries, and stimulating the willingness of the international community to donate financially to build up preparedness and create resilient societies 7. an international organization is needed to be the central coordinating organization of this effort. today the united nations is probably the best international organization to take on this task, but other international organizations should also be considered 8. a cooperation model needs to be established to allow all qualified stakeholders in the global emergency management sector to participate and contribute to building resilient societies 9. the receiving countries must likewise be helped to build local expertise able to manage the local situation and manage the preparedness projects 10. there must be a focus on reducing aid administrative costs, so most of the means collected goes to the activity planned 11. control mechanisms for disaster preparedness projects should be established that measures effectiveness and economics, and insures projects are managed according to schedule and budget. disaster prevention and preparedness require a wide range of measures, both long and short term, aimed at saving lives and limiting the amount of damage caused. particularly, confronting recurrent calamities is a stupendous task, with attendant regulatory, institutional, resource and technological and managerial prerequisites (poudyal chhetri, 2001). some of the more pertinent challenges that need to be addressed are highlighted below:  it is imperative that each country take into account the unprecedented local impacts of global warming, green house effects and climate change when planning a multi-hazard early warning system or network, and in formulating and implementing disaster management policies.  emergency managers must be educated and trained in core competencies related to coordinating response to large, complex disasters. there must be a focus on mentoring leadership and working with teams on an inter-sectoral, interdisciplinary, international basis.  there needs to be a comprehensive and unified disaster management policy to reduce the social and economic costs to the community caused by disasters. the policy should properly address problems in urban, suburban, and rural areas.  the link between environment and infrastructure in the region needs to be understood and accounted for in disaster planning  experiences, ideas, and strategies should be shared with all stakeholders, as part of an inclusive planning and execution process. disaster prone countries need to formulate and implement strategies, plans of action, and programs for disaster risk reduction. they should develop their institutional and technical capabilities in order to cope with disasters. preparedness and preventive measures are highly desirable to reduce the disaster losses. of course, effective, efficient response and rehabilitation capabilities/mechanisms are also important to minimize and redress disaster losses and damages. it is estimated that in 2013, over $69 billion was spent worldwide on emergency and incident management (research and markets, 2014). despite the need, or perhaps because published by atlantis press copyright: the authors 135 of it, there is great variability concerning the level of cooperation and coordination that exists at all levels. cooperation and coordination varies from community level to local, village, municipality, district/province and region. at the international level (kahn, 2005) it also varies from regional to global. just as the number and severity of disasters are increasing across the globe, so are social and economic costs of disasters and global humanitarian challenges increasing in scope and complexity. the need and importance of cooperation and coordination among the national/international organizations and countries is pertinent particularly in the development of humanitarian assistance. it is vital that those in charge of complex and multisectoral/multinational emergency operations be trained and educated in the basic competencies of emergency management. it is clear that as the largest countries struggle with the pressing need for disaster management, there has been an increasing interest in structured and consistent disaster management education of first responders and emergency managers. developing nations have a critical need for structured learning programs. in the united states, although there has been great activity in creating academic programs at all levels, there is much work on standardization yet to be done. an international perspective must be taken on what the most important knowledge and skills would be for basic training in emergency management. emergency managers must be educated and trained in the management skills of leadership, teams and team building, and conflict resolution. they must also have great respect for the experiential knowledge possessed by all nations as well as by the indigenous peoples of the world. disasters must also become a core international development issue. integrating disaster risk reduction into development policy, programs and practice is of great importance. it needs to be ensured that disaster risk reduction is included in poverty reduction strategy papers and development plans and programs. 1 to give an example of the benefits of disaster risk reduction approach, the world bank recently estimated that, on an average, countries can save $7 in disaster recovery costs for every $1 spent on risk reduction measures (jaurequi, sholk, radday, & stanzler, 2011). on the other hand, better systems for the collection, 1 sahana. (n.d.). retrieved from http://sahanafoundation.org analysis and dissemination of disaster impacts are important. by establishing performance targets linked to climate change, health, and livelihoods, progress can be assessed, and disaster risk reduction can be better integrated into both humanitarian and development efforts. 4. international response to disasters this situation calls for attention to the need for solid cooperation and coordination among government authorities, international organizations and ngos for effective disaster management (hagen, 2012). cooperation is considered by many as the best way to marshal the diverse players in pursuit of their common humanitarian goals. 2 in addition, since the impact of a disaster in a particular country or region can have a multitude of effects in other countries or regions, cooperation and coordination are needed to effectively accommodate the interests of diverse stakeholders. although a disaster may cause physical effects in a single nation, it produces ripples that are felt in neighboring nations and can cause long term impacts on productivity, growth and the economies throughout the world. in addition to the inter-country cooperation and coordination among disaster management stakeholders, there is a growing need for regional, international and global cooperation and coordination among the countries that have common problems, as a country's disaster situation may affect many other nations. for example, an earthquake disaster can have wide geographical coverage. where much of the region's transport and communications infrastructure crosses several borders, the need for regional-level services is obvious. similarly, failure to apply disaster risk management in any one country can affect such infrastructure and, thus, the region as a whole. the gisjlegen disaster in belgium is an example of international cooperation and optimized coordination in the area near the border of belgium and france3. an accident at a pumping station in belgium, very near the french border, resulted in many burn casualties, exceeding the capacity of local hospitals. through effective international cooperation, special burn 2 prc senior officials, regional cooperation of disaster management and preparedness (urumqi, china, 2006). 3 steiner, n. (2013). unpublished communication. published by atlantis press copyright: the authors 136 hospitals in the north of france were used to treat victims. a potentially complex international operation hindered by geographic boundaries, sectors, and jurisdictions was carried out successfully. regional and international cooperation is essential not only to cope with the impacts of disaster but also to help ensure that the region sustains economic growth. moreover, enhanced regional and international cooperation and coordination in disaster risk management offers attractive opportunities to provide resources. disaster preparedness has remained inadequate in many countries where lack of nationallevel updated disaster legislation hinders national focal agencies in inter-institutional and intergovernmental coordination. many aspects of disaster management are transborder. as a result, those countries could be economically and socially affected due to the serious natural disasters in a neighboring country. a neighbor may be able to provide valuable help in evaluating risk; mitigating, forecasting, developing and communicating early warnings; and responding to disasters. there is incredible value in utilizing international and indigenous knowledge systems when facing the challenges of reconstruction following disaster (hagen, 2010, hagen & hagen, in press). regional planning and joint efforts are needed to solve disaster problems, because disasters involve uncertainty and they are major concerns for all countries. countries that have common goals and interests should share disaster data and information so as to reduce the impact of potential disasters. international cooperation is critical to find more effective ways to forecast risk, to better manage the response and to develop organizational resilience to interruption and different types of crisis and disasters (poudyal chhetri, 1999). beyond local trans-border and joint economic interests, there is main theme of this paper, humanitarian aid to vulnerable nations. the international community has certainly demonstrated its willingness to help and support the victims of catastrophes. for example, many of us living in peaceful environments far away from the tragedy in the philippines queued up to offer support. this is good, and it shows that we react and help when our “brothers and sisters” in this world are in desperate need. international cooperation in disaster response is facilitated by organizations throughout the world several united nations elements participate, including:  united nations office for the coordination of humanitarian affairs (un-ocha)  united nations human settlements program (unhabitat)  united nations disaster relief organization (undro), united nations international strategy for disaster reduction (unisdr)  world meteorological organization (wmo)  united nations environmental program (unep). nato and eu also play an important role in this area. including the euro-atlantic disaster response coordination centre (nato-eadrcc) and the emergency response coordination centre (eu-ercc). (ministry of home affairs, 1998). eadrcc is a “24/7” focal point for coordinating disaster relief efforts among nato members and partner countries. ercc (formerly mic), based at the european commission in brussels, is accessible 24/7 and can spring into action immediately when it receives a call for assistance. the ercc works in close cooperation with national crisis centers throughout the 32 countries participating in the community mechanism for civil protection (which includes the eu 28, the former yugoslav republic of macedonia, iceland, liechtenstein and norway). the world bank, in association with the global facility for disaster reduction and recovery, has provided technical and financial support to over twenty countries seeking to improve disaster risk management. for example, their mozambique water resources development project is a six year project that includes goals for improved flood management and mitigation. moreover, there are a number of international ngo’s and other international organizations which can offer valuable help and expertise supporting international collaboration. the international emergency management society (tiems), described below, is an example of an ngo working as part of the international community to achieve better global disaster preparedness. 5. barriers to international response changing climate, population growth, and other factors conspire to make vulnerable world populations increasingly at risk of loss of life and property due to disasters. the international community is more than published by atlantis press copyright: the authors 137 willing to help these populations, as evidenced by the outpouring of aid after disaster strikes. the benefits of international cooperation and coordination are very clear. however, attempts to function as a global community are often hampered by the challenges of coordination across cultures and organizations. coordination can be particularly difficult without the pressure of on-going emergency, making efforts at global preparedness even more difficult. clearly, there is plenty of room for improvement in global preparedness, through measures such as the following:  reinforce links between preparedness and response – incorporate learning from disasters into preparedness, and allocate sufficient emergency budgets to preparedness.  identify and remove stovepipes that hinder cooperation and coordination within the humanitarian community (un agencies, ngos, ingos, academic, corporate sectors etc.)  exploit technologies that support collaboration, such as the disaster management information system (dmis)1 software that has been deployed in response to disasters around the world, to share experiences, exchange views and learn lessons. also, it is well known that public information, education and disaster awareness are crucial. therefore, it is highly essential to raise disaster awareness and educate vulnerable populations on disaster impact and response. there must be respect for, and inclusion and incorporation of, local and indigenous knowledge. what stands in the way of the international collaboration that would be so helpful in implementing measures such as these, with clear benefits? two fundamental barriers have been identified (jaurequi, sholk, radday, & stanzler, 2011): (1) insufficient coordination among actors due to divergent priorities and values, limitations to mandates, unaligned information collection and sharing, and inexperienced staff and organizations (2) limited involvement of affected populations and governments in response. the international emergency management society (tiems) has undertaken an initiative to address these two fundamental barriers. we call the initiative disaster resilience establishment in vulnerable societies (drevs). (drager & robertson, global response for capacity building of disaster preparedness a tiems initiative, 2014) we describe tiems and our initiative below. 6. tiems tiems was founded in 1993 in washington dc, usa, and is today registered as an international, independent, non-profit ngo in belgium. tiems is an international network of users, planners, researchers, industry, managers, response personnel, practitioners, social scientists, and other interested parties and individuals concerned with emergency and disaster management. tiems provides a platform for all stakeholders within the global emergency and disaster management community to meet, network and learn about new technical and operational methodologies. it also aims to exchange experience with good industry practices including the best from different risk cultures. the belief is that this will influence policy makers worldwide to improve global cooperation and to establish global standards within emergency and disaster management. tiems stimulates the exchange of information regarding the use of innovative methods and technologies within emergency and disaster management. in this way, it will improve society’s ability to avoid, mitigate, respond to, and speedily recover from natural and technological disasters. tiems is building a network of experts through local tiems chapters all over the world, with the aim of “think globally and act locally”. tiems chapters are self governed entities within tiems framework. today chapters are established in italy, iraq, romania, be/ne/lux, india, finland, middle east and north africa (mena), japan, korea and china. dialogue is also opened with experts in more countries, which see the benefits of tiems international expert network where partnership, education and research in disaster resilience is the focus in local activity, and where culture differences is put on the education and research agenda. tiems chapters play the main role as hosts of tiems international events, and the tiems japan chapter will be the host of tiems next annual conference in 2014, in niigata, japan on 21 – 23 october, with the support of the governor of niigata. the date, 23rd of october, coincides with the anniversary date of the big 2014 niigata earthquake. published by atlantis press copyright: the authors 138 one of the main topics in japan will be focused on experiences with past disasters in the niigata area. tiems believes in a global dialogue to learn from the exchange of expert information and experience between all stakeholders involved in emergency and disaster management. last year tiems arranged 10 conferences and workshops around the world, in japan, usa, iraq, finland, china, france and germany. they focused on important and timely topics in emergency and disaster management. in 2014, tiems has so far planned 13 conferences and workshops in the usa, japan, iraq, china and finland. we expect additional 2014 workshops and conferences will be arranged by other tiems chapters. tiems also initiates and takes part in research and development projects that aim at developing and/or improving methods and technologies in emergency and disaster management. tiems is also developing an international education training and certification program in emergency and disaster management. the international education program is composed of tiems chapter training and tiems qiedm (qualifications in international emergency and disaster management) certification. more details are found on tiems website: www.tiems.org. the tiems network constitutes a large international, multidisciplinary group of experts, with different educational backgrounds and various experiences in the field of emergency and disaster management. they represent a unique source of expertise and ideas, which are important for creating resilient societies. tiems’s latest initiative, which was launched by the tiems china chapter and discussed during the tiems annual conference in france, is to establish tiems task force groups. each tiems task force group would comprise qualified tiems scientists in different fields and from different cultures. these task groups could cooperate with unocha, and/or other international organizations and/or with local emergency management government agencies, and directly join in emergency management operations when they occur. during the tiems china chapter symposium on emergency medical care in guangzhou, china in 2013, tiems emergency medicine and tiems education, training and certification task force groups were established. 7. disaster resilience establishment in vulnerable societies tiems believes global emergency preparedness can be improved, and losses in particularly vulnerable areas reduced, through: (a) establishment and dissemination of a standardized base of emergency management knowledge, inclusive of and adaptable to local knowledge and conditions; (b) greater emphasis by the international aid community on preparedness versus response in vulnerable regions. tiems has established an initiative to catalyze these improvements disaster resilience establishment in vulnerable societies (drevs). the initiative has two thrusts: the first is the tiems qualifications in international emergency and disaster management (qiedm) certification program (www.tiems.org). the goal of this program is to spread basic emergency management expertise and best practices worldwide, using a core knowledge base that will facilitate crossorganizational collaboration. beyond this standard core, the program engages local experts to incorporate critical region and culture-specific knowledge and practice. to date, tiems has developed educational materials, established a pool of local expert teachers, and run pilot courses. we are currently identifying other sources of educational materials, and exploring collaboration with related trainings and certifications. the second thrust seeks to enhance disaster resilience in vulnerable societies, by exploring models of participatory governance in the application of international aid. tiems is currently working with the european union to develop these models, under the action plan on science in society issues in epidemics and total pandemics (asset) program. the philippines is a prime candidate for a drevs pilot. details on this project are forthcoming, and organizations and individuals interested in participating are encouraged to contact the authors. while coordinating international emergency response will always be a complex organizational and political endeavor, tiems believes its exceptional network of qualified and experienced international experts is in a unique position to assist in the creation of more resilient societies worldwide. we believe our resources are particularly well-suited to enhancing published by atlantis press copyright: the authors 139 disaster preparedness in vulnerable countries, by helping build international alliances and increasing global expertise in emergency management. references m. bandyk, why natural disasters are more expensive but less deadly. us news and world report (2010, march 24). k. h. drager and t. robertson, global response for capacity building of disaster preparedness a tiems initiative, 10th international conference of the international institute for infrastructure resilience and reconstruction (purdue university, west lafayette, 2014). j. c. hagan, cross-cultural aspects of disaster reconstruction: importance of lessons learned from indigenous cultures, proceedings of the international symposium on "5.12" massive wenchuan earthquake reconstruction & catastrophic disaster management (chengdu, 2010). j. c. hagan and s. p. hagan, the immediate post-disaster reconstruction phase: alternate care settings and vulnerable populations. in cabi, disaster management: medical preparedness, response, and homeland security. (in press) j. c. hagen, disaster management core competencies in global disaster management. proceedings of the tiems workshop resilient asia: implementing international emergency management system (tokyo, japan. 2012). international federation of red cross and red crescent societies, world disasters report 2013 (geneva, switzerland, 2013). c. jaurequi, j. sholk, a. radday and c. stanzler, international disater response (social impact research, boston, 2011). m. e. kahn, the death toll from natural disasters, the review of economics and statistics 87 (2005) 271-284. ministry of home affairs, problems of disaster management in nepal and how to solve them (ministry of home affairs, kathmandu, nepal, 1998). m. b. poudyal chhetri, mitigation and management of floods in nepal (ministry of home affairs, kathmandu, nepal, 2001). m. b. pudyal chhetri, disaster management in nepal: problems and solutions, in: j. e. ingleton ed, natural disaster management (tudor rose holdings limited, leicester, uk, 1999). research and markets, worldwide emergency management and incident market forecasts (dublin, ireland, 2014). a. scholtens, controlled collaboration in disaster and crisis management in the netherlands, history and practice of an overestimated and underestimated concept, journal of contingencies and crisis management 16 (2008) 195-207. published by atlantis press copyright: the authors 140 1. world disasters in 2013 2. observations from a recent disaster 3. what do vulnerable nations need? 4. international response to disasters 5. barriers to international response 6. tiems 7. disaster resilience establishment in vulnerable societies references research article 基于百度指数的战略性新兴产业上市公司股权融资效率影响因 素研究 influencing factors of equity financing efficiency of the listed companies in strategic emerging industry based on baidu index zhi-yuan lü1, mu zhang1,2,* 1school of big data application and economics, guizhou university of finance and economics, guiyang 550025, china 2guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang 550025, china 摘要 为进一步提高战略性新兴产业上市公司股权融资效率，本文选取2014-2018年208家战略性新兴产业上市公司相关数 据，通过dea-malmquist模型计算出规模效率变动、纯技术效率变动和全要素生产率变动并将三者作为股权融资效率 的代表变量，同时运用tobit模型对所选取的资产负债率、应收账款周转率和子产业网络关注度等11个变量的年增长 率进行分析，找出其中影响显著的变量。研究结果表明：在相关的年增长率变量中，子产业网络关注度对战略性新 兴产业上市公司股权融资规模效率变动、纯技术效率变动和全要素生产率变动都有显著影响，其中对规模效率变动 的影响是负向的，但资产负债率、资产报酬率、总资产、无形资产、日收益波动率和前三名高管薪酬对三者都无显 著影响；应收账款周转率、净利润增长率只对战略性新兴产业上市公司股权融资规模效率变动和全要素生产率变动 有显著正向影响，而股权集中度只对二者有显著负向影响；综合杠杆和资产报酬率这两个变量前者只对全要素生产 率变动有显著负向影响，后者只对纯技术效率变动产生显著正向影响。得出结论为：对于战略性新兴产业上市公司 股权融资效率的影响因素中除常见的财务指标和非财务指标外，百度指数大数据指标作用凸显。. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). a r t i c l e i n f o article history received 21 august 2020 accepted 22 october 2020 keywords strategic emerging industries listed companies equity financing efficiency influencing factors baidu index tobit model a b s t r a c t in order to further improve the equity financing efficiency of listed companies in strategic emerging industries, this paper selects relevant data of 208 listed companies in strategic emerging industries from 2014 to 2018, and calculates the changes in scale efficiency, pure technical efficiency and total factor productivity through the data envelopment analysis (dea)-malmquist model. taking the three as representative variables of equity financing efficiency, at the same time, using the tobit model to analyze the annual growth rate of selected 11 variables including asset-liability ratio, accounts receivable turnover rate and subindustry network attention, and find out which variables that have a significant impact. the research results show that among the relevant annual growth rate variables, the degree of attention of sub-industry networks has a significant impact on the changes in the scale efficiency of equity financing of listed companies in strategic emerging industries, the changes in pure technical efficiency, and the changes in total factor productivity, among which the changes in scale efficiency the impact of is negative, but asset-liability ratio, return on assets, total assets, intangible assets, daily return volatility and top three executive compensation have no significant impact on the three; accounts receivable turnover rate, net profit. the growth rate only has a significant positive effect on the changes in the scale and efficiency of equity financing of listed companies in strategic emerging industries and the changes in total factor productivity, while the concentration of equity only has a significant negative effect on both; the two variables of comprehensive leverage and return on assets. the former only has a significant negative impact on changes in total factor productivity, while the latter only has a significant positive impact on changes in pure technical efficiency. the conclusion is: in addition to the common financial indicators and non-financial indicators, the big data indicators of baidu index have a prominent role in the factors affecting the equity financing efficiency of listed companies in strategic emerging industries. journal of risk analysis and crisis response vol. 10(4); november (2020), pp. 130–137 doi: https://doi.org/10.2991/jracr.k.201028.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr 关键词 战略性新兴产业 上市公司 股权融资效率 影响因素 百度指数 tobit模型 中图分类号 f276.6 文献标识码 a *corresponding author. email: rim_007@163.com http://creativecommons.org/licenses/by-nc/4.0/ https://doi.org/10.2991/jracr.k.201028.001 https://www.atlantis-press.com/journals/jracr mailto:rim_007%40163.com?subject= z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 131 1. 引言 战略性新兴产业是以重大技术突破和重大发展需求为基础， 对经济社会全局和长远发展具有重大引领带动作用，知识技 术密集、物质资源消耗少、成长潜力大、综合效益好的产 业。加快培育和发展战略性新兴产业是全面建设小康社会、 实现可持续发展的必然选择、是推进产业结构升级、加快经 济发展方式转变的重大举措、是构建国际竞争新优势、掌 握发展主动权的迫切需要，对推进我国现代化建设具有重要 战略意义。加快培育和发展战略性新兴产业具备诸多有利条 件，也面临严峻挑战。经过改革开放30多年的快速发展，我 国综合国力明显增强，科技水平不断提高，建立了较为完备 的产业体系，特别是高技术产业快速发展，规模跻身世界前 列，为战略性新兴产业加快发展奠定了较好的基础。同时， 也面临着企业技术创新能力不强，掌握的关键核心技术少， 支持创新创业的投融资和财税政策、体制机制不完善等突出 问题。回顾世界经济强国中相关产业的发展历程，其中资本 市场直接融资为促进其发展起至关重要的作用。同样，股权 融资对于中国战略性新兴产业上市公司发展的重要性也是毋 庸置疑的。对此，为更好的进行股权融资首要面临的问题是 如何提高股权融资的效率。股权融资效率是指公司在股权融 资的财务活动中所实现的效能，融资效率提高的同时可以有 效的减少相应的融资成本，更有助于公司长远的发展，所以 找到具体的、显著的影响因素对于提高股权融资效率有着关 键的作用。本文研究的目的就是为了寻找对战略性新兴产业 上市公司股权融资效率有显著影响的因素，为相关产业的发 展提供一个途径。 本文将国内外学者关于股权融资效率影响因素的研究方法以 及所选取变量做以下总结：第一，研究方法。在现有的研究 中，国内外学者研究股权融资效率影响因素的整体思路是先 以数据包络分析法（dea）分析股权融资效率，然后进一步研 究相关产业股权融资效率的影响因素。对于高新技术产业上 市公司，肖雅（2018）[1]运用dea模型对其股权融资效率进 行分析然后对影响因素进行研究，为提高相关产业上市公司 的股权融资效率提供了方案；对于中小企业，张征超（2016） [2]主要是通过dea分析法对股权融资效率进行测评进而研究 股权融资效率的影响因素，为解决中小企业融资问题提供了方 案；对于基建以及文化服务产业，朱萌（2018）[3]同样采用 dea模型对相关产业的股权融资效率进行测评并进一步研究股 权融资效率的相关影响因素；在人工智能领域，刘超（2019） [4]采用dea模型分析相关上市公司的股权融资效率并对其影 响因素进行探究。除上述情况外还有一些角度较为新颖的研 究，如kaffash和marra（2017）[5]从金融领域特性的角度研究 相关产业股权融资效率，对提高股权融资效率提供一个新的 思路；在对战略性新兴产业的创新发展效率的研究中，刘春 姣（2019）[6]使用malmquist模型对创新效率进行分析；金辉 （2018）[7]的股权融资效率测评方法与常见的dea模型分析 不同，其方法是将融资效率视为投资报酬率与资本成本率的 比值，为股权融资效率的测度提供新的角度。 第二，股权融资效率影响因素。在有关股权融资效率影响因 素的研究中，相关文献选取的变量类型有财务型，如毕超 （2019）[8]在研究对新三板上市公司融资效率影响因素时选 取了资产、负债、营收和费用等方面的变量。张征超（2016） [2]在研究创业板上市公司股权融资效率影响因素时,  选取 成本费用等变量进行分析。在相关影响因素的研究中还有企 业经营类型的变量，如romano（2001）[9]发现企业的融资 效率受到经营目标、资本结构、管理模式、生命周期的显著 影响。eisdorferetal（2013）[10]研究资本结构等指标与资本 融资效率之间的影响。还有融资角度类型的变量，如almeida 和  wolfenzo（2005）[11]在研究发现，企业外源融资需求增 加会导致重置资本的发生，企业可以将资本从产出效率较低 的项目转移至效率较高的项目，进而提升融资的效率。张玲 华（2018）[12]是从企业融资的方式，即内源融资以及外源 融资，研究不同的融资方式对中小企业的融资效率的影响。 还有将股价作为影响因素的研究，如朱文莉（2018）[13]分 析交易日中股价的走势来研究其对上市企业融资效率的影 响。 综上所述，现有的文献对于股权融资效率影响因素的研究所 选取的变量范围不够广泛，往往只涉及较少的财务变量，并 且少有文献考虑百度搜索指数对于战略性新兴产业股权融资 效率的影响，所以本文考虑到扩大财务变量数量并结合大数 据变量，将选取应收账款周转率、资产负债率以及百度搜索 指数等11个变量进行研究，对股权融资效率评测方面选择 dea-malmquist模型分析后的结果代表股权融资效率，在此 基础上，通过受限因变量模型（tobit模型）对股权融资影响 因素进行分析，找出影响显著的变量。 2. 变量选取及原假设 本文结合国内外文献并考虑全面性、客观性以及各方面数据 的完整性来选取变量。被解释变量以三种融资效率变动为标 的，选取了规模效率变动、纯技术效率变动、全要素生产率 变动从三个层面分析股权融资效率。解释变量按照财务因素 和非财务因素两个方面选取，为对应被解释变量的变动，财 务因素中选取了应收账款周转率年增长率、总资产报酬率年 增长率、资产负债率年增长率和净利润增长率年增长率，借 以反映公司营运方面、未来发展、偿还债务以及自身盈利的 能力。非财务因素中选取了无形资产年增长率、股权集中度 年增长率、总资产年增长率、股票日收益波动率年增长率、 高管前三名薪酬总额年增长率、综合杠杆年增长率和子产业 的网络关注度年增长率，来体现上市企业规模大小、研发能 力、收益的稳定性、杠杆水平、投资者对战略性新兴产业七 个子产业的关注程度、股权融资结构以及公司治理结构等方 面的能力。综上做出如下假设： h1：子产业的网络关注度会影响战略性新兴产业上市公司股 权融资效率。 百度搜索指数是以网民在百度的搜索量为数据基础，以关键 词为统计对象，科学分析并计算出各个关键词在百度网页搜 索中搜索频次的加权和。根据搜索来源的不同，百度搜索 指数分为pc搜索指数和移动搜索指数（http://index.baidu. com)。大数据百度pc搜索指数、大数据百度移动搜索指数是 以“大数据”为关键词的百度搜索指数，反映了某一地区大 数据产业的影响力。本文的百度搜索指数以战略性新兴产业7 个子产业，即节能环保、信息、生物、高端装备制造、新能 源、新材料、新能源汽车为搜索标的，借以反应7个子产业的 影响力以及被关注热度，从关注度上研究其对股权融资效率 的影响。由于采用百度搜索指数研究其对股权融资效率的文 章比较少，所以在本文中认为该指标会对战略性新兴产业上 市公司股权融资效率产生影响。综上做出假设h1。 h2：应收账款周转率与战略性新兴产业上市公司股权融资效 率之间呈正相关关系。 应收账款周转率是应收账款与收入之比，反应应收账款的回 款速度。现金流对于企业而言是核心支撑，应收账款周转越 快，企业收回款项的压力越小，现金流越健康，若应收账款 一直无法回收，企业现金流压力大，导致经营发展受到阻 132 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 碍，势必会对股权融资效率产生一定的影响。同时国内学者 对于应收账款的相关研究中也表示出该指标的重要性，如时 艳娇（2010）[14]提出了应收账款的回款速度会影响企业的 竞争力。李高雅（2016）[15]通过对我国上市公司应收账款 与利润相关性进行实证分析后发现应收账款对于上市公司利 润有较强的影响。综上做出假设h2。 h3：净利润增长率与战略性新兴产业上市公司股权融资效率 之间呈正相关关系。 对于财务指标中的净利润增长率，该指标反映了企业的利润 增长状况。该指标是投资时的重要参考指标。通常该指标越 大，企业越具有成长性，经营水平越高，投资者对企业未来 的发展具有良好的预期，企业的股权融资效率会相应提高。 国内学者对净利润增长率指标的研究中同样表明了这种趋 势，如阮孝麟（2013）[16]研究创业板上市公司的融资效率 时发现净利润增长率对股权融资效率有正向影响，并且对于 处于成长中的企业，其净利润增长率表现不稳定，但总体趋 势向上；发展成熟后，净利润增长率会趋于稳定，波动幅度 逐渐缩小；企业步入衰退期后，会经常出现净利润增长率为 负值的情况。综上做出假设h3。 h4：资产报酬率与战略性新兴产业上市公司股权融资效率之 间呈正相关关系。 资产报酬率是企业利润与资产的比值，是衡量企业获利能力 的指标，资产报酬率越高说明该企业的获取利润的能力越 高，资产报酬率过低则反应企业在经营方面可能出现了问 题。同时国内学者在相关研究中也声明了该指标的重要性， 如蒋坛军（2017)[17]提出总资产报酬率是影响企业发展的关 键因素之一。段翀（2014）[18]实证研究后发现，资产负债 率、资产报酬率和应收账款周转率会影响企业的信用风险。 综上本文认为资产报酬率会提高战略性新兴产业股权融资效 率，因此做出假设h4。 h5：股权集中度与战略性新兴产业上市公司股权融资效率呈 负相关关系。 股权集中度是来体现企业股权分散化程度的指标。当股权越集 中，对企业而言其经营决策将会更加注重股东收益最大化，整 体战略实施更加有效率，但同样可以分析的是，股权若过于集 中将不利于员工的发展，企业应对风险的能力将会降低。在国 内学者关于股权集中度的研究中也同样表示出股权集中度为融 资效率带来的负面效应，如王元芳(2016)[19]研究发现随着股 权过于集中化融资效率会随之降低。综上做出假设h5。 h6：综合杠杆与战略性新兴产业上市公司股权融资效率之间 呈负相关关系。 综合杠杆指的是财务杠杆和经营杠杆的乘积，本身反映每一 单位销售量的变化对应每股收益的变化。国内学者在相关的 研究中也注重综合杠杆这个指标，赵红英（2012）[20]在财 务杠杆理论的基础上提出应用综合杠杆法来对企业折现问题 进行分析。虽然毕超（2019）[8]研究综合杠杆对新三板企业 融资效率的影响，结果并不显著，但是本文认为，综合杠杆 过高会面临更高的风险，不利用企业的发展。结合战略性新 兴产业上市公司特点，综上做出假设h6。 3. 模型构建与数据来源 3.1. 数据来源与描述性统计 本文数据来源于锐思数据库、国泰安数据库和百度指数大 数据平台，并参考黄亚茜（2019）[21]关于战略性新兴 产业相关上市公司选取方法，按照战略性新兴产业的节能 环保、新一代信息技术、生物医药、高端装备制造、新能 源、新材料、新能源汽车这七大子产业选择具有代表性的 208家上市公司。为保证样本量的充足性，选取数据的时 间跨度为2014年—2018年。从这208家公司2014年—2018 年五年的年度财报中选取应收账款周转率、资产报酬率、 资产负债率、净利润增长率、总资产、无形资产、股权集 中度、日收益波动率、高管前三名薪酬总额以及综合杠杆 这十个变量。通过百度指数大数据平台，按照战略性新兴 产业七大子产业五年的搜索量划分到每一类公司的方法确 定相应的子产业网络关注度。在汇总的数据中，对这十一 个变量中少数缺失的数据进行平均化处理。对于规模效率 变动、纯技术效率变动以及全要素生产率变动这三个被解 释变量是以208家代表性上市公司2014-2018年的年度财报 中的实收资本、股权集中度、资产负债率为输入变量，净 资产收益率、主营业务收入和托宾q值为输出变量，并根 据dea-malmquist模型测算方法（马军伟，2013）[22]由 deap软件输出，由于五年的数据变为了四年所以为对应相 关的变化将选取的11个变量通过公式（1）agxi  =  (xt xt-1)/xt-1分别求其年增长率agxi得到最终的解释变量，其 中i = 1 - 11，t = 2015, 2016, 2017, 2018；xt和xt-1分 别代表相关变量对应年份的取值。 从表2中可以看出，应收账款周转率、总资产等变量的最大 值和最小值差距很大，应收账款最小值0.346次，最大值高 达734.918次，总资产最小值为308万元、最大值达到7830亿 元。数据表明在208家上市公司样本中既有资金流动快、偿债 能力强体量大的企业又有收账速度相对慢、股权相对分散、 利润增长走下坡路的企业。从子产业网络关注度的描述性统 计中可以看出，在2014—2018这五年内关于战略性新兴产业 相关子产业的搜索量最大值达到306.564万次，均值为38.604 万次，说明人们对于该产业的关注程度非常高，也表明了相 关产业的重要性以及研究的必要性。战略性新兴产业涉及七 大子产业，各类公司的方向、体量、利润率、偿债能力等方 面的不同是客观存在的。 3.2. 模型设定 tobit回归是受限因变量回归的一种，其概念最早由诺贝尔经 济学家获得者james tobin提出。受限因变量是指虽然因变量 的取值是连续的随机的，但是其本身还是在一定的区间内变 化。经典的tobit 模型是james tobin在分析家庭耐用品的支出 情况时对probit  回归进行的一种推广，其后又被拓展成多种 表1 | 变量汇总 变量 代码 规模效率变动 cy1 纯技术效率变动 cy2 全要素生产率变动 cy3 应收账款周转率年增长率 agx1 资产报酬率年增长率 agx2 资产负债率年增长率 agx3 净利润增长率年增长率 agx4 总资产年增长率 agx5 无形资产年增长率 agx6 股权集中度年增长率 agx7 日收益波动率年增长率 agx8 高管前三名薪酬总额年增长率 agx9 综合杠杆年增长率 agx10 子产业网络关注度年增长率 agx11 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 133 情况如周华林（2012）  将”如周华林（2012）[21]”删除， 这句话多余了。  [21]，amemiya将其归纳为ⅰ型到ⅴ型tobit 模型。标准的ⅰ型tobit模型如下： y x y y if y y if y i i i i i i i i * * * * * = + = > = £ b m 0 0 0    （2） 在上式中， yi * 是被解释变量，大于0时被观察到，取值为yi， 小于等于0时在0处截尾，xi是自变量向量，β是系数向量，误 差项μi独立且服从正态分布μi: n(0, σ 2)。综上并结合标准型 tobit回归模型确定现有变量的计量方程如下： cy agx agx agx agx agx agx agx j * = + + + + + + + + b b b b b b b b b 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 88 8 9 9 10 10 11 11 agx agx agx agx cy cy if i j j j cy + + + + = > b b b m * * 00 0 0 cycy ifj j * *= ≤  （3） 其中j = 1, 2, 3，β 1 β 11 为解释系数，μi为误差项，β0为常 数项；cy 1 为规模效率变动，cy 2 为纯技术效率变动，cy 3 为 全要素生产率变动，即从三个方向来考察战略性新兴产业上 市公司股权融资效率；agx 1 = agx 11 按照顺序分别为应收账 款周转率年增长率、资产报酬率年增率、资产负债率年增长 率、净利润增长率年增长率、总资产年增长率、无形资产年 增长率、股权集中度年增长率、日收益波动率年增长率、高 管前三名薪酬总额年增长率、综合杠杆年增长率和子产业网 络关注度年增长率。 4. 实证分析 4.1. hausman检验 hausman检验是在使用面板数据时，对选取随机效应模 型(random-effects model)还是固定效应模型(fixed-effects model)进行数据分析判断的一种确定方法。由于本文所 使用的是面板数据，所以首先对这些数据进行hausman检 验。hausman统计量为： 表2 | 相关变量描述性统计 变量（单位） 样本数 最小值 最大值 平均值 标准差 规模效率变动 832 0.295 1.681 0.959 0.124 纯技术效率变动 832 0.893 1.463 1.031 0.070 全要素生产率变动 832 0.282 2.165 0.956 0.112 应收账款周转率(次) 1040 0.346 734.918 10.685 42.584 资产报酬率(%) 1040 –54.951 93.374 6.452 6.545 资产负债率(%) 1040 3.500 95.260 41.800 18.774 净利润增长率(%) 1040 –12678.700 3519.068 0.958 631.061 无形资产(千万) 1040 0.003 1400.000 47.200 114.000 股权集中度(%) 1040 2.520 94.300 33.033 13.766 总资产(千万) 1040 0.308 78300.000 1260.000 4450.000 日收益波动率(%) 1040 0.004 209.497 3.270 6.496 高管前三名薪酬总额(十万) 1040 2.243 247.000 22.694 18.710 综合杠杆 1040 –44.908 125.547 2.593 5.832 子产业网络关注度(万次) 1040 3.249 306.564 38.604 63.000 h b b var b var b b b k= -( ) ( )( )éë ùû -( ) ( ) -' ~ 1 2c    （4） 可以看出hausman统计量是服从自由度为k的c2分布。从统 计结果中的p值可以判断出两种模型的选取，即当p值小于 显著性水平时，就拒绝原假设应选择固定效应模型，当p值 大于显著性水平时，就接受原假设应选择随机效应模型。 在规模效率变动层面对所选取数据进行hausman检验结果如 下表3： 分析该检验结果可知，p值为0.2926>0.1，即该检验在10% 显著性水平不显著，接受了原假设，认为是随机效应，所以 相比于固定效应模型在该层面应该选择随机效应模型进行分 析。在纯技术效率变动层面对所选取数据进行hausman检验 结果如下表4： 分析该检验结果可知，p值为  0.0382<0.05，即该检验在5% 显著性水平上显著，即拒绝了原假设，认为是固定效应，所 以在该层面应该选择固定效应模型进行分析更合理。在全 要素生产率变动层面对所选取数据进行hausman检验结果如 下表5： 分析该检验结果可知，p值为0.0014<0.1，即该检验在1%的 显著性水平上显著，检验结果接受了原假设，认为是固定效 应，所以应该选择固定效应模型进行分析更合适。 4.2. tobit回归结果与分析 根据相关数据，利用stata计量软件15.1版本进行tobit模型分 析，对战略性新兴产业股权融资效率影响因素分析结果如下： （1）规模效率变动回归结果分析 由表6规模效率变动回归结果可以看出，子产业的网络关注度 年增长率对战略性新兴产业上市公司股权融资规模效率变动 呈负相关关系。相关系数为-0.0800，且相关性在1%水平上显 著。子产业网络关注度上升的因素一般分为两种：积极因素 和消极因素。虽然战略性新兴产业上市公司受到二者的综合 影响，但是在所选择的时间周期内规模效率层面股权融资效 率整体受到的影响偏消极，因此降低了股权融资规模效率。 该结果接受原假设h1：子产业的网络关注度会影响战略性新 兴产业上市公司股权融资效率。 134 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 表3 | 规模效率变动hausman检验结果 coefficients (b) (b) (b–b) sqrt(diag (v_b–v_b)) fe re difference s.e. agx1 0.0508 0.0512 –0.0004 0.0052 agx2 –0.0009 –0.0004 –0.0005 0.0011 agx3 –0.0084 –0.0073 –0.0011 0.0050 agx4 0.0001 0.0001 0.0000 0.0000 agx5 –0.0001 –0.0001 0.0000 0.0001 agx6 –0.0480 –0.0495 0.0015 0.0063 agx7 0.0002 0.0002 0.0000 0.0002 agx8 –0.0001 0.0000 –0.0001 0.0001 agx9 –0.0050 –0.0020 –0.0030 0.0022 agx10 –0.0028 –0.0023 –0.0005 0.0013 agx11 –0.0895 –0.0800 –0.0095 0.0035 _cons 0.9729 0.9709 0.0020 0.0008 b = consistent under ho and ha; obtained from xtreg, b = inconsistent under ha, efficient under ho, obtained from xtreg, test: ho: difference in coefficients not systematic, chi2(12) = (b–b)’[(v_b–v_b)^(–1)](b–b) = 13.01, prob>chi2 = 0.2926. 表4 | 纯技术效率变动hausman检验结果 coefficients (b) (b) (b–b) sqrt(diag (v_b–v_b)) fe re difference s.e. agx1 0.0027 0.0043 –0.0017 0.0028 agx2 0.0015 0.0022 –0.0007 0.0006 agx3 0.0081 0.0068 0.0013 0.0027 agx4 0.0000 0.0000 0.0000 0.0000 agx5 –0.0002 –0.0001 –0.0001 0.0000 agx6 0.0045 0.0021 0.0024 0.0035 agx7 –0.0003 –0.0003 0.0000 0.0001 agx8 0.0000 0.0000 0.0000 0.0000 agx9 0.0018 0.0011 0.0007 0.0012 agx10 0.0008 0.0011 –0.0003 0.0007 agx11 0.0881 0.0805 0.0076 0.0019 _cons 1.0163 1.0176 –0.0013 0.0004 b = consistent under ho and ha; obtained from xtreg, b = inconsistent under ha, efficient under ho, obtained from xtreg, test: ho: difference in coefficients not systematic, chi2(12) = (b–b)’[(v_b–v_b)^(–1)](b–b) = 20.56, prob>chi2 = 0.0382. 表5 | 全要素生产率变动hausman检验结果 coefficients (b) (b) (b–b) sqrt(diag (v_b–v_b)) fe re difference s.e. agx1 0.0549 0.0569 –0.0020 0.0048 agx2 –0.0001 0.0007 –0.0008 0.0010 agx3 –0.0114 –0.0075 –0.0039 0.0046 agx4 0.0001 0.0000 0.0000 0.0000 agx5 –0.0002 –0.0001 –0.0001 0.0001 agx6 –0.0491 –0.0495 0.0004 0.0058 agx7 0.0005 –0.0003 0.0008 0.0002 agx8 0.0000 0.0000 0.0000 0.0001 agx9 0.0000 –0.0008 0.0007 0.0020 agx10 –0.0043 –0.0036 –0.0007 0.0012 agx11 0.0189 0.0210 –0.0021 0.0032 _cons 0.9528 0.9527 0.0001 0.0007 b = consistent under ho and ha; obtained from xtreg, b = inconsistent under ha, efficient under ho, obtained from xtreg, test: ho: difference in coefficients not systematic, chi2(12) = (b–b)’[(v_b–v_b)^(–1)](b–b) = 30.28, prob>chi2 = 0.0014. 表6 | 规模效率变动回归结果 cy1 coef. std. err. z p > z agx1 0.0512 *** 0.0099 5.1600 0.0000 agx2 –0.0004 0.0019 –0.2100 0.8340 agx3 –0.0073 0.0096 –0.7600 0.4460 agx4 0.0001 *** 0.0000 3.9900 0.0000 agx5 –0.0001 0.0001 –0.6800 0.4980 agx6 –0.0495 *** 0.0120 –4.1100 0.0000 agx7 0.0002 0.0004 0.4900 0.6250 agx8 0.0000 0.0001 0.0200 0.9850 agx9 –0.0020 0.0039 –0.5200 0.6010 agx10 –0.0023 0.0023 –1.0200 0.3070 agx11 –0.0800 *** 0.0116 –6.8700 0.0000 _cons 0.9709 0.0046 212.8900 0.0000 注：***p < 0.01, **p < 0.05, *p < 0.1. 应收账款周转率年增长率与战略性新兴产业股权融资在规模 效率变动方面呈现正相关关系，相关系数为0.0512，且相关 性在1%水平上显著。表明在其他变量不变的情况下，战略性 新兴产业上市公司应收账款周转率对股权融资的规模效率产 生较大的正向影响。应收账款周转率越大，表明随着企业回 款周期的缩小，企业现金流更加健康，股权融资规模效率也 相应提高。因此该结果接受原假设h2：应收账款周转率与战 略性新兴产业上市公司股权融资效率之间呈正相关关系。 净利润增长率对应的年增长率变化与战略性新兴产业股权融 资在规模效率变动方面呈现正相关关系，相关系数为0.0001， 且相关性在1%水平上显著。随着净利润增长率的提高，表明 企业的盈利水平越高，企业的经营成果更加有效，对股权融 资有很好的促进作用，融资规模效率随之提高。因此该结果 接受原假设h3:净利润增长率与战略性新兴产业上市公司股 权融资效率之间呈正相关关系。 股权集中度年增长率对战略性新兴产业上市公司股权融资规 模效率变动呈负相关关系。相关系数为-0.0495，且相关性 在1%水平上显著。在一定范围内，股权集中度越高，企业的 经营决策权利将越集中，企业未来的走向很可能受到个人因 素的影响越大，不利于企业的发展，股权融资效率也相应降 低。股权集中度相对越低，企业在战略选择和应对风险方面 会有更加综合的考量，企业未来的发展将会更好，股权融资 效率相应提高。该结果接受原假设h5：股权集中度与战略性 新兴产业上市公司股权融资效率呈负相关关系。 规模效率变动层面的非显著变量，如资产负债率、资产报酬 率、无形资产、总资产、日收益波动率、前三名高管薪酬和 综合杠杆，其中资产报酬率和综合杠杆回归结果拒绝了原假 设，可能原因有以下几个方面：第一，在所选取的时间周期 内这些变量的波动幅度过小，对股权融资规模效率并无明显 的影响。如资产负债率、无形资产、总资产和前三名高管薪 酬。第二，有可能是公司在这个方面的承受能力整体较强， 如综合杠杆。第三，有可能是股权融资规模效率对其敏感程 度较差，如资产报酬率。 (2) 纯技术效率变动回归结果分析 由表7纯技术效率变动回归结果可以看出，子产业的网络关 注度年增长率对战略性新兴产业上市公司股权融资纯技术效 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 135 率变动呈正相关关系。相关系数为0.0805，且相关性在1%水 平上显著。表明在其他变量不变的情况下，战略性新兴产业 上市公司子产业网络关注度对股权融资纯技术效率产生较大 的正向影响。子产业网络关注度上升，股权融资规模效率下 降，对公司而言为此会做出相应的措施应对风险，从而提升 股权融资纯技术效率以提升总体融资效率水平。所以该结果 接受原假设h1：子产业的网络关注度会影响战略性新兴产业 上市公司股权融资效率。 资产报酬率年增长率与战略性新兴产业股权融资在纯技术效 率变动方面呈现正相关关系，相关系数为0.0022，且相关性 在1%水平上显著。表明在其他变量不变的情况下，战略性新 兴产业上市公司资产报酬率对股权融资纯技术效率产生较大 的正向影响。资产报酬率越大表明公司的获利能力在增长， 盈利水平提高，表明企业经营管理较好，从而提高了股权融 资效率。因此该结果接受原假设h4：资产报酬率与战略性新 兴产业上市公司股权融资效率之间呈正相关关系。 纯技术效率变动层面的非显著变量，如应收账款周转率、净 利润增长率、资产负债率、无形资产、总资产、日收益波动 率、股权集中度、前三名高管薪酬和综合杠杆，其中应收 账款、净利润增长率、股权集中度和综合杠杆拒绝了原假 设，可能原因有以下几个方面：第一，在所选取的时间周期 内这些变量的波动幅度过小，对股权融资纯技术效率并无明 显的影响。如资产负债率、无形资产、总资产和前三名高管 薪酬。第二，有可能是公司在这个方面的抗风险能力整体较 强，如综合杠杆。第三，有可能是股权融资技术效率对其敏 感程度较差，如资产报酬率、应收账款周转率、股权集中 度、净利润增长率。 （3）全要素生产率变动回归结果分析 由表8全要素生产率变动回归结果可以看出，子产业的网络关 注度年增长率对战略性新兴产业上市公司股权融资在全要素 生产率变动方面呈正相关关系。相关系数为0.02101，且相关 性在10%统计水平上显著。表明在其他变量不变的情况下，战 略性新兴产业上市公司子产业网络关注度对股权融资全要素 生产率产生较大的正向影响。说明在所选取的时间周期内， 战略性新兴产业子产业的网络关注度整体上偏向于积极因 素，积极因素多，表明公司营运能力更强，投资者对未来企 业发展将有更好的预期，投资人的投资意愿会更加强烈，促 进公司股权融资效率，从而提高股权融资全要素生产率。所 以该结果接受原假设h1：子产业的网络关注度会影响战略性 新兴产业上市公司股权融资效率。 应收账款周转率年增长率与战略性新兴产业上市公司股权融 资在全要素生产率变动方面呈现正相关关系，相关系数为 0.05694，且相关性在1%统计水平上显著。应收账款周转率 越大，表明随着企业回款周期越小，企业现金流更加健康， 企业的未来的经营状况和发展也会越好，这不仅对股权融资 规模效率产生正向作用同时对整体上股权融资效率有促进作 用。因此该结果接受原假设h2：应收账款周转率与战略性新 兴产业上市公司股权融资效率之间呈正相关关系。 净利润增长率对应的年增长率变化与战略性新兴产业上市公 司股权融资在全要素生产率变动方面呈现正相关关系，相关 系数为0.0001，且相关性在5%统计水平上显著。净利润增长 率提高，代表公司的盈利水平越高，其发展潜力也就越大。 一般情况下，利润率高的公司更容易受到投资人的偏好，往 往也能够以一个较低的成本融资，从而提高了融资效率。因 此该结果接受原假设h3:净利润增长率与战略性新兴产业上 市公司股权融资效率之间呈正相关关系。 股权集中度年增长率与战略性新兴产业上市公司股权融 资在全要素生产率变动方面呈现负相关关系。相关系数 为-0.04952，且相关性在1%统计水平上显著。表示在其他变 量保持不变的情况下，战略性新兴产业上市公司股权集中度 对股权融资全要素生产率有显著的负向作用。股权集中度越 高，企业经营决策权力的掌控程度越高，在各种方面都会受 到个人或者少数人的权利干预的概率会越大，越不利于企业 综合发展，企业经营水平随之下降，进而降低股权融资效 率。因此该结果接受原假设h5：股权集中度与战略性新兴产 业上市公司股权融资效率呈负相关关系。 综合杠杆年增长率对战略性新兴产业上市公司股权融资在全 要素生产率变动方面呈现负相关关系。相关系数为-0.00395， 且在10%统计水平上显著。表示在其他变量保持不变的情况 下，战略性新兴产业上市公司综合杠杆对股权融资全要素生 产率有显著的负向作用。综合杠杆越高，企业面临的财务风 险和经营风险将会越高，风险越高往往对融资而言将会越困 难，融资效率大打折扣。因此该结果接受原假设h6：综合杠 杆与战略性新兴产业上市公司股权融资效率之间呈负相关关 系。 全要素生产率变动层面的非显著变量，如资产负债率、资产 报酬率、无形资产、总资产、日收益波动率、前三名高管薪 酬，其中资产报酬率拒绝了原假设，原因有以下几个方面： 第一，在所选取的时间周期内这些变量的波动幅度过小，对 股权融资全要素生产率变动并无明显的影响。如资产负债 表7 | 纯技术效率变动回归结果 cy2 coef. std. err. z p > z agx1 0.0043 0.0054 0.8000 0.4250 agx2 0.0022 ** 0.0010 2.1300 0.0340 agx3 0.0068 0.0053 1.2900 0.1980 agx4 0.0000 0.0000 –0.2800 0.7780 agx5 –0.0001 0.0001 –1.1200 0.2610 agx6 0.0021 0.0066 0.3200 0.7520 agx7 –0.0003 0.0002 –1.3600 0.1740 agx8 0.0000 0.0001 0.1400 0.8900 agx9 0.0011 0.0021 0.5100 0.6110 agx10 0.0011 0.0012 0.8900 0.3730 agx11 0.0805 *** 0.0064 12.6300 0.0000 _cons 1.0176 0.0025 407.6100 0.0000 注：***p < 0.01, **p < 0.05, *p < 0.1. 表8 | 全要素生产率变动回归结果 cy3 coef. std. err. z p > z agx1 0.05694 *** 0.00911 6.25000 0.00000 agx2 0.00068 0.00175 0.39000 0.69800 agx3 –0.00746 0.00884 –0.84000 0.39900 agx4 0.00005 ** 0.00002 2.52000 0.01200 agx5 –0.00015 0.00012 –1.27000 0.20300 agx6 –0.04952 *** 0.01107 –4.47000 0.00000 agx7 –0.00026 0.00038 –0.68000 0.49600 agx8 0.00004 0.00010 0.34000 0.73200 agx9 –0.00076 0.00356 –0.22000 0.83000 agx10 –0.00359 * 0.00209 –1.72000 0.08500 agx11 0.02101 * 0.01071 1.96000 0.05000 _cons 0.95274 0.00419 227.16000 0.00000 注：***p < 0.01, **p < 0.05, *p < 0.1. 136 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 率、无形资产、总资产和前三名高管薪酬。第二，有可能是 股权融资全要素生产率变动对其敏感程度较差，如资产报酬 率、日收益波动率。 为进一步验证子产业网络关注度对战略性新兴产业股权融资 效率的影响，将子产业网络关注度作为单独指标对纯技术效 率变动、规模效率变动和全要素生产率变动进行tobit回归， 回归结果如下表9： 由表9可知，在对子产业网络关注度的单独回归中，其对规模 效率变动、纯技术效率变动的相关性在1%统计水平上显著， 对全要素生产率变动的相关性在10%统计水平上显著。表明子 产业网络关注度对于战略性新兴产业股权融资效率的变动具 有一定的作用。 5. 结论 整体而言，战略性新兴产业股权融资规模效率变动、纯技术 效率变动和全要素生产率变动的影响因素各有不同，其中 规模效率变动受到显著影响的年增长率变量有应收账款周 转率、净利润增长率、股权集中度和子产业网络关注度。纯 技术效率变动受到显著影响的年增长率变量有资产报酬率和 子产业的网络关注度。全要素生产率变动受到显著影响的 年增长率变量有应收账款周转率、净利润增长率、股权集中 度、综合杠杆和子产业网络关注度。实证结果表明，子产 业网络关注度对战略性新兴产业股权融资规模效率变动、纯 技术效率变动和全要素生产率变动都具有一定的显著性。 相对于其他影响显著的年增长率变量，如股权集中度、净利 润增长率等，子产业网络关注度对股权融资效率的影响更为 广泛。 战略性新兴产业上市公司为提高股权融资效率，不论是从规 模效率变动、或全要素生产率变动入手，在都应该注重应收 账款周转率、净利润增长率和股权集中度这三个变量，同时 全要素生产率层面还应该注重综合杠杆这个变量；应根据相 关产业的情况降低资金回收时间，提高资金回收效率，还应 该注重企业营业水平，提高净利润；适当降低股权集中程 度，释放更多的决策权力，让企业战略运营更加综合化、全 面化；对全要素生产率而言还应该降低公司的经营风险和财 务风险，降低杠杆水平，提高企业的发展潜力。针对纯技术 效率变动公司应该注意资产报酬率这个变量；应该增加企业 收入，降低相关成本，节约资金，提高资产的利用效率。此 外，股权融资规模效率变动、纯技术效率变动和全要素生产 率变动三者应该同样注重子产业网络关注度这个变量；注重 网络中的曝光度，增加积极因素，减少负面消息。 表9 | 子产业网络关注度单独回归结果 变量 回归ⅰ 回归ⅱ 回归ⅲ cy1 cy2 cy3 agx11 –0.07829 *** 0.07860*** 0.02128* (0.00000) (0.00000) (0.05400) _cons 0.97131 1.01830 0.95281 (0.00000) (0.00000) (0.00000) 注：***p < 0.01, **p < 0.05, *p < 0.1, 括号中为p值。 acknowledgments to the regional project of national natural science foundation of china (71861003) and the innovative exploration and new academic seedlings project of guizhou university of finance and economics (guizhou-science cooperation platform talents [2018] 5774-016) for their support. 参考文献 [1] 肖雅,郭晓顺.新三板高新技术企业股权融资效率评价[j].财 会月刊, 2018;57–61. xiao y, guo xs. evaluation on the efficiency of equity financing of high-tech enterprises on the new third board. finan account monthly 2018;57–61. [2] 张征超,张婷婷.创业板上市公司股权融资效率测评[j].财会 月刊 2016;55–8. zhang zc, zhang tt. equity financing efficiency evaluation of gem listed companies. finan account monthly 2016;55–8. [3] 朱萌.ppp项目上市公司股权融资效率的评价及影响因素研 究[j].经济论坛 2018;67–74. zhu m. research on the evaluation and influencing factors of equity financing efficiency of listed companies in ppp projects. econ forum 2018;67–74. [4] 刘超,傅若瑜,李佳慧,周文文.基于dea-tobit方法的人工智 能行业上市公司融资效率研究[j].运筹与管理  2019;28: 144–52. liu c, fu ry, li jh, zhou ww. research into financing efficiency of artificial intelligence industry based on dea-tobit method. oper res manage sci 2019;28:144–52. [5] kaffash s， marra m． data envelopment analysis in financial services: a citations network analysis of banks， insurance companies and money market funds． ann oper res 2017;253:307–44． [6] 刘春姣.基于malmquist模型的战略性新兴产业创新效率实 证分析[j].统计与决策, 2019;35:147–9. liu cj. an empirical analysis on innovation efficiency of strategic emerging industries based on malmquist model. stat decision 2019;35:147–9. [7] 金辉,黄珏.我国新三板企业融资效率影响因素分析[j].杭州 电子科技大学学报(社会科学版)2018;14:24–9. jin h, huang y. an analysis on influencing factors of financing efficiency of neeq enterprises. j hangzhou dianzi univ (soc sci) 2018;14:24–9. [8] 毕超,李逍雯.新三板企业股权融资效率及影响因素研究 [j].财会通讯 2019;31–4. bi c, li xw. research on equity financing efficiency and influencing factors of new third board enterprises. commun finan account 2019;31–4. [9] romano ca， tanewski ga， smyrnios kx．capital structure decision making: a model for family business. j bus ventur 2001;16;285–310． [10] eisdorfer a， giaccotto c， white r．capital structure， executive compensation， and investment efficiency． j bank finan 2013;37:549–62． https://doi.org/10.1007/s10479-016-2294-1 https://doi.org/10.1007/s10479-016-2294-1 https://doi.org/10.1007/s10479-016-2294-1 https://doi.org/10.1007/s10479-016-2294-1 https://doi.org/10.1016/s0883-9026%2899%2900053-1 https://doi.org/10.1016/s0883-9026%2899%2900053-1 https://doi.org/10.1016/s0883-9026%2899%2900053-1 https://doi.org/10.1016/j.jbankfin.2012.09.011 https://doi.org/10.1016/j.jbankfin.2012.09.011 https://doi.org/10.1016/j.jbankfin.2012.09.011 z.-y. lü and m. zhang / journal of risk analysis and crisis response 10(4) 130–137 137 [11] almeida h， wolfenzon d． the effect of external finance on the equilibrium allocation of capital. j finan econ 2005;75:133–64． [12] 张玲华.中小企业融资方式对融资效率的影响分析[j].科技 视界 2018;203–4. zhang lh. analysis of the influence of smes’ financing methods on financing efficiency. sci technol view 2018;203–4. [13] 朱文莉,王奥运.社会责任、所有权性质与现金持有价值—— 基于a股市场的经验证据[j].财会月刊 2018;30–40. zhu wl, wang ay. social responsibility, ownership nature and cash holding value: empirical evidence from a-share market. finan account monthly 2018;30–40. [14] 时艳娇.制造型企业应收账款影响因素的实证研究[d].湖南 大学中国湖北省武汉市武昌区八一路299号, 2010. shi yj. an empirical study on the influencing factors of accounts receivable in manufacturing enterprises. hunan university, no. 299 bayi road, wuchang district, wuhan city, hubei province, china, 2010. [15] 李高雅.我国上市公司应收账款与利润质量的相关性研究 [j].商业会计 2016;76–8. li gy. research on the correlation between accounts receivable and profit quality of listed companies in china. commercial account 2016;76–8. [16] 阮孝麟.创业板上市公司融资效率研究[d].贵州财经大学, 中国贵州省贵阳市花溪区花溪大学城, 2013. ruan xl. research on the financing efficiency of listed companies on the growth enterprise market. guizhou university of finance and economics, huaxi university town, huaxi district, guiyang city, guizhou province, china, 2013. [17] 蒋坛军.“总资产报酬率”为何重要？[n]. 企业家日报, 2017– 04–17(003). jiang tj. why is “return on total assets” important?. entrepreneur daily, 2017–04–17(003). [18] 段翀,刘忻梅.基于ccsd模型的上市企业信用风险评价研究 [j].征信 2014;32:14–18+52. duan c, liu xm. research on credit risk evaluation of listed enterprises based on ccsd model. credit ref 2014;32:14–18+52. [19] 王元芳.股权结构、机构投资者与融资效率关系研究[d].山 东财经大学, 中国山东省济南市二环东路7366号, 2016. wang yf. research on the relationship between equity structure, institutional investors and financing efficiency. shandong university of finance and economics, no. 7366, east 2nd ring road, jinan, shandong province, china, 2016. [20] 赵 红 英 . 论 特 定 企 业 折 现 率 的 确 定 [ j ]. 现 代 商 贸 工 业 2012;24:113–14. zhao hy. on the determination of discount rate of specific enterprises. modern business trade ind 2012;24:113–14. [21] 黄亚茜.战略性新兴产业上市公司股权融资效率评价及影响 因素研究[d].贵州财经大学, 中国贵州省贵阳市花溪区花溪 大学城 2019. huang yx. research on evaluation and the influencing factors of equity financing efficiency of listed companies in strategic emerging industries. guizhou university of finance and economics, huaxi university town, huaxi district, guiyang city, guizhou province, china, 2019. [22] 马军伟.基于malmquist模型的战略性新兴产业金融支持效 率研究[j].西安财经学院学报 2013;26:11–15. ma jw. efficiency of financial support for strategic emerging industries based on malmquist model. j xi‘an inst finan econ 2013;26:11–15. https://doi.org/10.1016/j.jfineco.2004.06.001 https://doi.org/10.1016/j.jfineco.2004.06.001 https://doi.org/10.1016/j.jfineco.2004.06.001 microsoft word quantum-behaved particle swarm optimization with nm method_n received 6 december 2014 accepted 8 february 2015 quantum-behaved particle swarm optimization with nelder-mead simplex search method weiquan yao department of basic course, guangzhou city polytechnic, 248 guangyuan zhong road guangzhou, 510405, china e-mail: yaowq6916@126.com abstract this paper proposes a novel hybrid algorithm based on quantum-behaved particle swarm optimization (qpso) algorithm and nelder-mead (nm) simplex search method for continuous optimization problems, abbreviated as qpso-nm. this hybrid algorithm is very easy to be implemented since it does not require continuity and differentiability of objective functions, and it also combines powerful global search ability of qpso with precise local search of nm simplex method. in a suite of the first 10 test functions taken from cec2005, qpso-nm algorithm is compared with other four popular competitors and six special algorithms that are dedicated to solve cec2005 test function suite. it is showed by the computational results that qpso-nm outperforms other algorithms in terms of both convergence rate and solution accuracy. the proposed algorithm is extremely effective and efficient at locating optimal solutions for continues optimization. keywords: swarm optimization,nelder -mead simplex method,hybrid algorithm, continuous optimization 1. introduction particle swarm optimization (pso) is a swarm intelligence based algorithm which stems from the study of artificial lives and foraging behaviors of biological populations such as bird and fish schools [1,2]. in pso, each particle represents a potential solution to the optimization problem, and flies around the search space with a certain speed which is continually updated according to flight experience of particle itself and that of others. it is also apparent, however, that pso may easily be stuck in a local optimum [2]. to improve it, sun et al. proposed a new pso model in 2004, which was based on delta potential well and believed that particles had quantum behaviors. from this new model, a quantum-behaved particle swarm optimization (qpso) algorithm was proposed [3]. in quantum spaces, particles are able to search in the whole feasible solution spaces. therefore, the global search ability of qpso may be much better than that of the standard pso. unlike pso, qpso has no velocity vector and fewer parameters to adjust, and it is much easier for implementations. but, qpso still has the probability of being trapped in the local optimum, and it is not good at exploitation [4]. nelder-mead simplex method [5] is a simple direct search technique and it does not require any gradient information, so it was widely used in solving unconstrained function optimization [6]. this paper tries to integrate nm simplex method into qpso, and proposes hybrid qpso-nm algorithm so as to improve the local search ability of qpso. the rest of this paper is organized as follows: section 2 gives some preliminaries on qpso and nm simplex method. section 3 describes the flowchart of the hybrid algorithm and section 4 presents experimental results and analysis. finally, major results of this paper are summarized in section 5. journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 47-53 published by atlantis press copyright: the authors 47 w.q. yao 2. preliminaries 2.1. qpso algorithm qpso is a variant of pso which uses wave function ),( tx to describe the status of particles. by solving schrodinger equation, the probability density function of particles appearing in a certain point in the space could be obtained. then use monte carlo stochastic simulation, we may get position equation of particle i at t+1 th iteration, which is formulated as follows [2,3]. ] )( ln[)()()( , ,,, 1 1 2 1 1   tu tltptx ji jijiji , (1) where ),(:)( )()]([)()()( ,, 10 1 ut tgttpttp j jjjijji    (2) )(tpi is called local attractor point of particle i; )(, tp ji and )(tg j are jth dimension of pbest position of particle i and gbest position of the whole swarm respectively. )(, 1tu ji is a uniformly distributed random number over (0,1). )(, tl ji is calculated by the following formula. |)()(|)( ,, txtctl jijji  2 , (3) where )(tc is the mbest position with coordinates being the following. ).)(, ,)(,)(( )())(),(),(()( , ,,         m i di m i i m i i m i id tp m tp m tp m tp m tctctctc 1 1 21 1 121 1 11 1   ( 4) thus, the position updating equation of particle i in qpso is ]. )( ln[|)( )(|)()( , , ,, 1 1 1    tu tx tctptx ji ji jjiji  (5) in general, we call pso using equation (5) to update particles’ positions a quantum-behaved particle swarm optimization (qpso) algorithm.  in (5) is known as contraction – expansion (ce) coefficient, the only parameter to be tuned by the algorithm, which is used to control convergence rate. in [8], it is suggested that the value of  could linearly decrease from 1 to 0 during the search process, and that this would be a simple but effective way. so, computational formula of  is 0 1 0( ) ( ) /t t t        , (6) where 1 and 0 are initial and final values of  respectively. often we set 1 01, 0.5   . 2.2. nelder-mead simplex method nelder and mead proposed a simplex search method, which is a local search algorithm designed for unconstrained optimization without using any gradient information [7]. for a function minimizing problem with d variables, nm method continuously updates “simplex” through iterations after comparing objective values at d+1 vertices followed by replacing the worst vertex with a new one that is generated by using four basic procedures: reflection, expansion, contraction and shrinkage. through these procedures, the simplex can successfully improve itself and get closer to the optimum. 3. qpso-nm algorithm the local nm search method is integrated into the main qpso, and critical steps of qpso-nm algorithm are illustrated below: (1) initialization of nm “simplex”. since nm method uses d+1 vertices for d dimensional optimization problems, d randomly generated positions in the search range, together with the gbest position are used to form the initial nm “simplex”. (2) a parameter named local search period (l) is used to control the running of nm search method, that is to say, if the current number of iterations is a multiple of l, then the main algorithm will turn to the local search. (3) max number of function evaluations for nm simplex method (l_maxfes) is used to determine the running time of the local search. suppose f is a function to be minimized, the pseudocode of qpso-nm may be depicted as follows. initialize position vectors of each particle and evaluate them determine pbest and gbest positions initializing nm “simplex” published by atlantis press copyright: the authors 48 quantum-behaved particle swarm optimization gen = 1 while terminal condition is not met compute mbest position c using (3) compute the value of  using (5) for i=1:m for j = 1:d  = rand(0,1) pi,j =  * pi,j + (1- )*gj u = rand(0,1) if rand(0,1) < 0.5 xi,j = pi,j +  *| cj-xi,j|*ln(1/u) else xi,j = pi,j  *| cjxi,j|*ln(1/u) end end evaluated the new position update pbest of the i th positon end update gbest postion if mod (gen,l) ==0 conduct a local search using nm simplex search method and update fes correspondingly find the best position after nm simplex search, denoted as nmbest。 and use it to update gbest accordingly. end gen = gen + 1 end 4. computational results and analysis now we use qpso-nm algorithm to solve the first 10 test functions from cec2005 (see table 1). the experiment is conducted on the 10-d problems with the terminal criterion max_fes setting to 1e+5. the value of parameter l is set to 50, and that of l_maxfes is set to 500. to conduct fair comparisons between our hybrid qpso-nm and other algorithms, the number of max function evaluations (max_fes) is used as a time measure, i.e. all algorithms will be terminated once fes reaches to max_fes. and the best function error value is employed as evaluation criterion for each algorithm. the convergence curves of spso [9], clpso [10], qpso [3], dms-pso [11] and qpso-nm on each test function are plotted in figure 1 . from figure 1, it could be seen that convergence curves of qpso-nm (the red solid lines) can approach the true optimal value faster than the rest algorithms on functions f1, f2, f3, f4, f6, f8 and f10. while for functions f5 and f9, dms-pso is better than qpso-nm at first phase of the evolution, but the difference becomes very tiny at last. for function f7, the convergence speed of qpso-nm is slower than spso at first, but eventually the former outperforms the latter in terms of the quality of solutions. it is shown by the final tendency of convergence curves that the red solid lines always locates in the lowest place, which means a higher solution accuracy of qpso-nm than other competitors. all in all, qpso-nm is superior to other four algorithms in terms of both convergence speed and solution accuracy. table 1. the first 10 test functions from cec2005 func no. function name bounds minimum f1 shifted sphere function [-100,100] -450 f2 shifted schwefel's problem 1.2 [-100,100] -450 f3 shifted rotated high conditioned elliptic function [-100,100] -450 f4 shifted schwefel's problem 1.2 with noise in fitness [-100,100] -450 f5 schwefel's problem 2.6 with global optimum on bounds [-100,100] -310 f6 shifted rosenbrock's function [-100,100] 390 f7 shifted rotated griewank's function without bounds initial bounds [0, 600] -180 f8 shifted rotated ackley's function with global optimum on bounds [-32,32] -140 f9 shifted rastrigin's function [-5,5] -330 f10 shifted rotated rastrigin's function [-5,5] -330 the source code of these test functions is available on p.n.suganthan’s website:http://www.ntu.edu.sg/home/epnsugan/ for each test function, the qpso-nm is run 25 times. the minimum (min), maximum (max), mean value (mean) and the standard deviation (stdev) of the best function error values are presented in table 2. from table 2, it is found that the mean error values are zero when solving functions f1, f2, f4 and f9, i n d i c a t i n g t h a t q p s o n m i s v e ry e f f e c t i v e i n optimizing the above functions. for the purpose of comparison, the mean of the best function error values of qpso-nm together with spso,clpso,qpso, published by atlantis press copyright: the authors 49 w.q. yao 0 2 4 6 8 10 x 10 4 0 1 2 3 4 5 fes lo g 1 0 (e rr o r+ 1 ) f1 spso clpso qpso dms-pso qpso-nm clpso qpso qpso-nm spso dms-pso 0 2 4 6 8 10 x 10 4 0 1 2 3 4 5 fes lo g 1 0 (e rr o r+ 1 ) f2 spso clpso qpso dms-pso qpso-nmspso qpso-nm qpso dms-pso clpso 0 2 4 6 8 10 x 10 4 0 2 4 6 8 10 fes lo g 1 0 (e rr o r+ 1 ) f3 spso clpso qpso dms-pso qpso-nm spso clpso dms-pso qpso qpso-nm 0 2 4 6 8 10 x 10 4 0 1 2 3 4 5 fes lo g 1 0 (e rr o r+ 1 ) f4 spso clpso qpso dms-pso qpso-nm spso dms-pso qpso-nm qpso clpso 0 2 4 6 8 10 x 10 4 0 1 2 3 4 5 fes lo g 1 0 (e rr o r+ 1 ) f5 spso clpso qpso dms-pso qpso-nm qpso-nm spso clpso qpso dms-spo 0 2 4 6 8 10 x 10 4 0 2 4 6 8 10 12 fes lo g 1 0 (e rr o r+ 1 ) f6 spso clpso qpso dms-pso qpso-nm clpso spso qpso dms-pso qpso-nm 0 2 4 6 8 10 x 10 4 0 1 2 3 4 fes lo g 1 0 (e rr o r+ 1 ) f7 spso clpso qpso dms-pso qpso-nm clpso spso dms-pso qpso-nm qpso 0 2 4 6 8 10 x 10 4 1.32 1.325 1.33 1.335 1.34 1.345 1.35 1.355 fes lo g 1 0 (e rr o r+ 1 ) f8 spso clpso qpso dms-pso qpso-nm qpso clpso spso dms-pso qpso-nm 0 2 4 6 8 10 x 10 4 0 0.5 1 1.5 2 2.5 fes lo g 1 0 (e rr o r+ 1 ) f9 spso clpso qpso dms-spo qpso-nm spso qpso clpsoqpso-nm dms-pso 0 2 4 6 8 10 x 10 4 0.5 1 1.5 2 2.5 fes lo g 1 0 (e rr o r+ 1 ) f10 spso clpso qpso dms-pso qpso-nm dms-pso qpso clpso spso qpso-nm fig. 1. convergence curves of spso, clpso, qpso, dms-pso and qpso-nm on each test function. the abscissa is fes, and the ordinate is log10(error+ 1), where error is the best function error value. dms-pso, fea[12], hybrid rcga[13], pso variant[14],de[15] ， es [16] and rcma[17], are listed in table 3 and table 4, where the mean error values of fea, hybrid rcga,pso variant, de, es and rcma are directly taken from corresponding literatures. table 2. the min, max, mean and stdev of the best function error values of qpso-nm f1 f2 f3 f4 f5 min 0.00000e+00 0.00000e+00 1.59730e-11 0.00000e+00 1.36993e-10 max 0.00000e+00 0.00000e+00 2.43565e-06 0.00000e+00 6.04978e-07 mean 0.00000e+00 0.00000e+00 1.00389e-07 0.00000e+00 4.38655e-08 stdev 0.00000e+00 0.00000e+00 4.86528e-07 0.00000e+00 1.21641e-07 f6 f7 f8 f9 f10 min 0.00000e+00 9.85728e-03 2.00002e+01 0.00000e+00 9.94959e-01 max 2.76780e+00 1.08312e-01 2.01894e+01 0.00000e+00 8.95463e+00 mean 1.32368e+00 5.94651e-02 2.00561e+01 0.00000e+00 5.15519e+00 stdev 7.45577e-01 2.84646e-02 4.86346e-02 0.00000e+00 1.82574e+00 table 3. the comparison of the mean best function error values for functions f1-f5 f1 f2 f3 f4 f5 spso 1.05387e+03 6.35766e+02 1.58418e+06 1.22268e+03 1.91479e+03 clpso 0.00000e+00 3.68345e-13 6.94785e+04 4.09273e-13 5.21414e+02 qpso 0.00000e+00 1.02318e-12 1.47868e+05 1.87933e-09 3.24642e+02 dmspso 0.00000e+00 2.69515e-07 1.37347e+05 2.24495e-04 9.37320e-09 fea 7.89310e-09 9.15620e-01 6.45730e+05 1.34520e+02 3.51070e+02 hybrid rcga 8.34460e-09 8.20750e-09 5.70530e+02 8.31770e-09 8.93750e-09 pso variant 0.00000e+00 0.00000e+00 2.29034e+05 0.00000e+00 0.00000e+00 de 0.00000e+00 5.70000e-02 2.09000e+05 6.19000e-01 1.21796e+02 es 8.16020e-09 2.90000e-06 3.52170e+05 4.13570e+03 1.36820e+03 rcma 9.86950e-09 9.93600e-09 4.77086e+04 1.99669e-08 2.12417e-02 qpsonm 0.00000e+00 0.00000e+00 1.00389e-07 0.00000e+00 4.38655e-08 published by atlantis press copyright: the authors 50 quantum-behaved particle swarm optimization the five functions in table 3 are unimodal functions. f1, f2 and f3 have different condition numbers which make f3 to be harder than f2 and f2 to be harder than f1. for f1, we could observe, from the results , that the mean best function error values of qpso-nm, clpso,qpso,dms-pso,pso variant and de are all equal to zero, which indicates that this test function could be effectively optimized by these algorithm. for f2 and f4, qpso-nm and pso variant achieve a better result. f3 is a very hard function to be optimized, and qpso-nm obtains a significant better result than other algorithms of which the order of magnitude reaches to 10-7 clearly lower than its counterparts. f5 is schwefel’s problem 2.6 with global optimum on bounds, for this function the performance of qpso-nm is worse than that of pso variant, hybrid rcga and dms-pso. based on the analysis above, we can conclude that qpso-nm and dms-pso have better overall performances for functions f1-f5. furthermore, qpso-nm evidently surpasses pso variant when solving f3. therefore, qpso-nm is deemed to be the most competitive algorithm among these eleven algorithms. table 4. the comparison of the mean best function error values for functions f6-f10 f6 f7 f8 f9 f10 spso 9.27075e+07 3.19381e-01 2.03047e+01 1.75415e+01 3.15131e+01 clpso 2.40375e+00 1.41874e-01 2.03577e+01 9.15362e-01 6.68042e+00 qpso 1.94224e+01 2.79719e-01 2.03674e+01 1.08812e+00 9.67376e+00 dms-pso 4.21388e+00 7.05375e-02 2.01527e+01 0.00000e+00 5.51815e+00 fea 3.21610e+03 1.83840e+00 2.01430e+01 7.91790e-09 2.34010e+01 hybrid rcga 8.87370e-09 1.17230e-02 2.03540e+01 1.15420e+00 4.97480e+00 pso variant 1.55292e+01 1.71838e-01 2.03034e+01 2.78588e+00 1.85821e+01 de 3.53800e+00 1.57751e+02 2.04440e+01 4.54000e-01 5.11460e+01 es 7.49040e+01 1.18260e+00 2.03680e+01 4.48520e+01 1.03020e+02 rcma 1.48962e+00 1.97139e-01 2.01923e+01 4.37855e-01 5.64289e+00 qpso-nm 1.32368e+00 5.94651e-02 2.00561e+01 0.00000e+00 5.15519e+00 functions f5-f10 are multimodal functions. qpsonm achieves the best result for f8, and it (along with dms-pso) also performs the best when optimizing f9. for f6, f7 and f10, hybrid rcga is the best one followed by qpso-nm: the differences between the above two algorithms are small for f7 and f10, while hybrid rcga obviously outperforms qpso-nm for f6. we may carefully come to the conclusion that qpso-nm and hybrid rcga can achieve better effects than other algorithms in solving functions f6f10. for each function, rank all algorithms according to their mean function error values from small to large, and calculate total and final ranks for each algorithm shown in table 5. it can be clearly seen from table 5 that the top four algorithms are qpso-nm, dms-pso, hybrid rcga and pso variant, respectively. furthermore, it is also found from table 3 and table 4 that: pso variant is very effective on unimodal functions f1-f5 but fails to solve multimodal functions f6-f10; hybrid rcga, however, can nicely minimize multimodal functions f6f10 but is inferior to qpso-nm on unimodal functions f1-f5. this phenomenon may be explained by the “no free lunch” theorem [18], and it says “any elevated performance over one class of problems is offset by performance over another class,” indicating that there is no algorithms perform the best over any class of problems, i.e. each algorithm has its corresponding application range. however, qpso-nm has an overall better performance on both unimodal and multimodal functions compared with pso variant and hybrid rcga, which could be explained by a finding in [19] stating “the hybrid of algorithms may be an effective measure to expand algorithms’ application range and improve their performances.” thus, qpso-nm is the most competitive algorithm in solving the first 10 test functions from cec2005 with an expended application range achieved by a successful hybrid of qpso and nm simplex method. table 5. ranks of all algorithms f1 f2 f3 f4 f5 f6 f7 f8 f9 f10 total ranks final ranks spso 11 11 11 10 11 11 8 6 10 9 98 11 clpso 1 3 4 3 9 4 4 8 6 5 47 5 qpso 1 4 6 4 7 8 7 9 7 6 59 7 dms-pso 1 7 5 7 3 6 3 3 1 3 39 2 fea 7 10 10 9 8 10 10 2 3 8 77 9 hybrid rcga 9 5 2 5 2 1 1 7 8 1 41 3 pso variant 1 1 8 1 1 7 5 5 9 7 45 4 de 1 9 7 8 6 5 11 11 5 10 73 8 es 8 8 9 11 10 9 9 10 11 11 96 10 rcma 10 6 3 6 5 3 6 4 4 4 51 6 qpso-nm 1 1 1 1 4 2 2 1 1 2 16 1 published by atlantis press copyright: the authors 51 w. q.yao 5. based on the pso of enterprise credit risk assessment model of bp neural network research (1)b p neural network parameter settings. for the number of input layer nodes, because the network is the enterprise credit risk assessment index of the input sample data, therefore, the number of input layer nodes by pca transformation after the extraction of characteristic data, the number of decisions. determine the number of input layer node model for 14. for the number of output layer nodes, as a result of enterprise credit risk assessment for yes or no only two kind of criterion, as a result, the output layer node number 1. for the determination of the number of hidden layer nodes, this model adopts the method of variable number in hidden layer. namely the start into the more hidden layer nodes, according to the empirical formula to determine the hidden node number is 6. as you progress through the training process, and gradually trim off doesn't work node in the network training. for each neuron, if in the process of training, the connection power is less than the weight of the absolute value of the numerical convergence factor, and then assign the weights of zero. if all of the neurons connection weights are zero, this god by nodes are removed. this paper using b p algorithm for neural network training, factor weight to 0. 001, mistakenly sent to receive the folding for child is 0. 00001. for the network weights, we using pso algorithm to optimize .the weight is decided by the parameters of the particles of the network. [20] (2) the design of fitness function, the size of the fitness available under type said: the yrj is the expected output node, drj is a node of the actual output. (3) the pso parameters settings particle dimension: is determined by the network connection weights, namely by the input node output nodes and the number of hidden layer nodes. suppose i, h, o, respectively of input layer, hidden layer and output layer neuron number, the dimension of each particle size d can be expressed as: d = (i3h) + 3 o (h) + h + o. therefore, this paper models the dimensions of the particle size for d = 109. the number of particles, on the basis of experience for general problem, particle number take 10 can obtain good effect. in order to achieve better training effect, in this paper, the particle number 15. vmax: maximum speed, decided to particles in a loop of the largest mobile distance, usually set the width of the range of particle, this paper set the particle belongs to [25, 5], then the size of thevmax is 10. learning factor: cl and c2 is usually equal to 2, but there are other values in the literature, but its range between 0 and 4. this article take the c l = c2 = 2. stop condition: maximum cycle number and the minimum error. in this paper, the minimum error in the model set for a classification error, the largest circulation set for 2000. (4) initialization: randomly generated m a particle, the particle's position and velocity range control between the (0, 1), the dimensions of the particle's position and velocity vector is equal to d. sets the pbest of each particle to the initial position, the best value for the gbest pbest. (5) the fitness evaluation: the population of each individual fitness evaluation. if is better than that of p best pbest is when in position. if all particles is superior in gbe st, gbe st is replaced. (6) the position and velocity update: update position and velocity of each particle. (7) algorithm termination: if meet the precision requirement or the whole evolution has reached the maximum number of iterations (to 2000), an algorithm and record the current best individual in the whole group. otherwise, go to step (5). in matlab 2012 implemented in the model proposed in this paper. based on credit risk evaluation index system, we adopt 15 projects related data as sample, 12 of them as the training sample, 3) as a test sample, the data after normalization, to analyze the main points and processing, as the input data into a network, the network training and testing respectively. as you can see, in this paper, the constructed model prediction results and the actual value of the original data error is very small 6. conclusions this paper studies the hybrid algorithm that integrates qpso and nelder -mead (nm) simplex search method. qpso has powerful global search abilities, but its local search abilities are not so good. nelder-mead simplex search method is a popular and simple direct search technique for unconstrained function optimization without using gradient information. in order to combine  2 2 rj rj r j y d e    published by atlantis press copyright: the authors 52 quantum-behaved particle swarm optimization powerful global search ability of qpso with precise local search of nm simplex method, this paper studies the hybrid of the above two algorithms and proposes qpso-nm. in a suite of the first 10 test functions taken from cec2005, qpso-nm algorithm is compared with other four popular competitors and six special algorithms that are dedicated to solve cec2005 test function suite. it is showed by the computational results that qpso-nm outperforms other algorithms in terms of both convergence rate and solution accuracy. the experimental results show that the proposed algorithm is extremely effective and efficient at locating optimal solutions for continues optimization. 7. references [1] kennedy j ， eberhart r c ． particle swarm optimization[c] ． proceedings of ieee international conference on neural networks,1999:1942-1948 [2] j. sun, research on quantum-behaved particle swarm optimization algorithm, doctoral dissertation, jiangnan university,2009. [3] j. sun, b. feng, w. xu, particle swarm optimization with particles having quantum behavior, in: ieee congress on evolutionary computation, 2004, pp.325–331. [4] j. sun, w. fang, v. palade, x.j. wu, w.b. xu, quantum-behaved particle swarm optimization with gaussian distributed local attractor point, applied mathematics and computation 218 (2011) 3763–3775. [5] j. a. nelder and r. mead, a simplex method for function minimization, the computer journal, 7 (1965) 308-313. [6] d. h. chen, z. saleem, and d. w. grace, a new simplex procedure for function minimization,international journal of modeling and simulation, 6 (1986) 81-85 [7] j.a. nelder, r. mead, a simplex method for function minimization, computer journal, 7 (1965) 308–313. [8] j. sun, w. fang, x.j. wu, v. palade, w.b. xu, quantum-behaved particle swarm optimization: analysis of the individual particle’s behavior and parameter selection, evolutionary computation, doi:10.1162/evco_a_00049. [9] y. shi, r. eberhart, a modified particle swarm optimizer, in: proceedings of the ieee international conference on evolutionary computation, piscataway, nj: ieee press, 1998, pp. 69-73. [10] j. j. liang, a. k. qin, p. n. suganthan, s. baskar, comprehensive learning particle swarm optimizer for global optimization of multimodal functions, ieee congress on evolutionary computation,2006,vol.10, 281-295. [11] j. j. liang and p. n. suganthan, dynamic multiswarm particle swarm optimizer, in: proceedings of 2005 ieee swarm intelligence symposium, 2005, pp. 124-129. [12] s. alonso, j. jimenez, h. carmona b. galvan, g. winter, performance of a flexible evolutionary algorithm, in: special session on real-parameter optimization at cec-05, edinburgh, uk, 2005. [13] c. garcía-martínez, m. lozano, hybrid realcoded genetic algorithms with female and male differentiation, in: special session on realparameter optimization at cec-05, edinburgh, uk, 2005. [14] m. f. tasgetiren, y.c. liang, g. gencyilmaz, i. eker, global optimization of continuous functions using particle swarm optimization, in: special session on real-parameter optimization at cec05, edinburgh, uk, 2005. [15] l. t. bui, y. shan, f. qi, and h. a. abbass, comparing two versions of differential evolution in real parameter optimization, in: special session on real-parameter optimization at cec-05, edinburgh, uk, 2005. [16] l. costa, a parameter-less evolution strategy for global optimization, in: special session on realparameter optimization at cec-05, edinburgh, uk, 2005. [17] d. molina, f. herrera, m. lozano, adaptive local search parameters for real-coded memetic algorithms, in: special session on real-parameter optimization at cec-05, edinburgh, uk, 2005. [18] d. h. wolpert , w. g. macready, no free lunch theorems for optimization, ieee transactions on evolutionary computation ,1997, vol. 1, pp. 67–82. [19] l. wang, intelligent optimization algorithm with applications, beijing: tsinghua university press, 2001,119-121. [20] shi y, ebe rha rt r.a modified paretic leeward optimizer [ c ] / / i eee world congress on compu tational intelligence, 1998: 69273 . published by atlantis press copyright: the authors 53 review risk forecasting in the light of big data roman kernchen*, eyvor institute, roedingsmarkt 20, hamburg 20459, germany 1. introduction society is increasingly connected through networks that link the world around us and facilitate a global exchange of people, commodities, capital, knowledge and ideas, create numerous exciting opportunities, new services and advantages for humanity. yet concurrently, the underpinning networks have provided routes by which potentially dangerous and harmful incidents can propagate quickly and worldwide. this complexity poses a considerable challenge for risk analysis and forecasting. conventional methods of risk analysis tend to underestimate the probability and impact of risks (e.g. pandemics, financial collapses, terrorist attacks), as sometimes the existence of independent observations is wrongly assumed and cascading errors that can occur in complex systems are not considered. big data, machine learning, and predictive analytics offer new opportunities for understanding and managing risks in complex environments and there is a widespread belief that big data can aid in improving risk forecasts [1–3]. many institutions are already adopting innovative ways to use big data analytical methods to improve their risk assessment processes and to predict risks from economic, social, or environmental data [4–6]. this a r t i c l e i n f o article history received 06 july 2020 accepted 29 december 2020 keywords big data risk forecasting systemic risk predictive analytics machine learning security risk sustainable development risk financial risk jel classification c53 c55 f52 f47 f37 g32 h56 a b s t r a c t life in modern society is increasingly connected by networks that link the world around us and create numerous exciting opportunities, new services and advantages for humanity. yet concurrently, these underpinning networks have provided routes by which potentially dangerous and harmful incidents can propagate quickly and worldwide. this complexity poses a considerable challenge for risk analysis and forecasting. conventional methods of risk analysis tend to underestimate the probability and impact of risks (e.g. pandemics, financial collapses, terrorist attacks), as sometimes the existence of independent observations is wrongly assumed and cascading errors that can occur in complex systems are not considered. the purpose of this article is to assess critically the potential of big data to profoundly change the current capability for risk forecasting in diverse areas and the assertion that big data leads to better risk predictions. in particular, the focus is on big data implications for risk forecasting in the areas of economic and financial risks, environmental and sustainable development risks, and public and national security risks. the article concludes that big data and predictive analytics offer substantial opportunities for improving risk forecasting but may not replace the significance of appropriate assumptions, adequate data quality and continuous validation. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *email: kernchen@eyvor.org journal of risk analysis and crisis response vol. 10(4); december (2020), pp. 160–167 doi: https://doi.org/10.2991/jracr.k.201230.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr article will discuss the application of big data and predictive analytics for risk forecasting in general and in the specific context of economic and financial risks, environmental and sustainable development risks, and public and national security risks. the article will begin by explaining the concepts of risks and big data and will then address the potential promises and pitfalls of using big data for risk analysis and forecasting. following this, common types of analytical methods for big data risk forecasting are identified and a general framework for their use is outlined. the last section of this article discusses the application of big data analytics in the abovementioned risk domains. risks exist in every area of life, and in both the business world and the public administration, decisions have to be made constantly, the outcome of which is uncertain. understanding the uncertainty, however, may help us make better decisions. a risk in this context can be defined as a random event that may possibly occur and, if it does occur, would have a negative impact on the objectives of the entity [7]. different disciplines have different ways of classifying risks. a common classification consists of the division into the three categories of: ‘known knowns’, ‘known unknowns’, and ‘unknown unknowns’, corresponding to different levels of uncertainty [8]. the term “unknown unknowns” was brought to the fore by us secretary of defense donald rumsfeld at a press conference http://orcid.org/0000-0002-4082-3023 mailto: kernchen@eyvor.org https://www.atlantis-press.com/journals/jracr r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 161 in 2002 when he addressed the lack of evidence for a link between the iraqi government and the supply of weapons of mass destruction to terrorist groups. the term has, however, been used long before this [9]. in a risk context, the idea of unknown unknowns intuitively grasps the fact that the events actually occurring are not covered by the events identified in the risk assessment. the risks that threaten modern societies compose a complex network that often underlies crisis events [10]. so far, however, little is known about how risk materializations in different domains influence each other. modern societies are highly dependent on the reliable functioning of systems that are interconnected in explicit or implicit ways [11]. whilst the increase in interconnection between various infrastructural systems can lead to a higher level of service efficiency, it also makes the involved systems susceptible as a whole to cascading failures [2,12]. such cases of cascades of failures have been studied in model networks in general and in the context of transport systems, in financial institutions and within ecological systems in particular. besides the risk of cascade failures occurring within a specific domain, there are other risks arising from the interconnection between systems in different domains. in fact, the key thesis behind many collapses of society in modern history is that there is a cascade of different risks that are unfolding, and hence there is a particular need to quantify the dynamics of largescale risk materialization that looms in this globally interconnected web. in this context, the predictive analysis of big data offers enormous opportunities for new insights, especially about networks, spatial and temporal dynamics, for the understanding of human systems on the systemic level and for the detection of interactions and nonlinearities in the relationships between variables [13]. 2. implications of big data for risk forecasting big data technology is seen as the digital-age equivalent of the telescope or microscope [14]. new repositories for powerful data are emerging from social media websites, search engines and other sources, which accumulate vast amounts of information every second. international data corporation predicts that the global datasphere will grow from 33 zettabytes (zb) in 2018 to 175 zb by 2025 [15]. while conceptions of big data have been around since the 1990s, the term only reached a higher level of relevance in technology and in the public perception during the last 15 years. the term big data is used to describe exceptionally large datasets that can only be analyzed by computation, either separately or in combination with other sets of data to uncover previously unrecognized patterns, trends and associations. the uniqueness of big data approaches lies in the fact that they offer the ability to collect and analyze data in a range and depth that would otherwise be difficult to achieve [16,17]. there are typically three characteristics associated with big data, often referred to as the three vs: volume, variety, and velocity [17–19]. in this context, (1) volume refers to the processing of ever larger amounts of data (e.g. terabyte, petabyte or larger); (2) variety refers to the diversity of the generated data, which may come from a variety of different sources and is generally one of three types: structured, semi structured and unstructured data. the variety in data types frequently requires distinct processing capabilities and specialist algorithms; (3) velocity refers to the speed at which data is generated, processed and analyzed. several countries and organizations have initiated various complex and large-scale projects to use big data, such as the european human brain project [20] or csiro’s australian square kilometre array pathfinder [21], demonstrating that big data is useful for more than just relatively small problems, viz. well-structured cases characterized by repeated evaluation of predictions [22]. furthermore, data mining and machine learning have emerged as powerful tools for big data forecasting to exploit the power of unstructured data and to extract new knowledge and identify significant patterns and correlations hidden in the data [23]. these two developments together have recently sparked an all-encompassing enthusiasm based on the idea that large data will lead to much better forecasts in all domains, from scientific discovery to medical, financial, commercial and political applications. recently, it even led to the idea that for the domain of predictive analysis, some kind of master algorithm could possibly be developed one day [24]. however, as some simple examples such as weather forecasting show, the best forecasts generally arise from a reasonable compromise between modelling and quantitative analysis, where conceptual insight counts as much as the amount of data [25]. the capability of making predictions from the data alone may be theoretically possible, but in practice it is often unachievable, even in the face of a clear situation where one has perfectly accurate information about the system. hence, big data does not make the importance of data quality and modelling assumptions obsolete. while big data can be crucial for understanding risks in interdependent systems, particularly in the field of disaster modelling, and can also improve the accuracy of traditional risk assessment techniques, risk analysis is clearly not just a computational challenge that can be solved by more data. ultimately, for an accurate prediction, it must be possible to understand the risk analysis in order to moderate the occurrence and impact of negative outcomes. the difficulties of using big data for risk analysis can be illustrated with the example of google flu trends (gft). the google flu tracking system is frequently cited as a negative example for the usage of big data forecasting, although the erroneous forecasts in this particular case were arguably caused by the algorithm dynamics that influenced google’s search algorithm and the changes that google itself had made to the search algorithm [26]. gft has persistently overestimated flu prevalence, predicting more than twice the percentage of consultations for influenza-like illnesses than the us centers for disease control and prevention, which have based their estimates on surveillance reporting from labs located throughout the united states. the mistakes made in the conceptual design of gft have often been described as big data hubris in a general criticism of the tendency to see big data not as a supplement but as a replacement for traditional data collections and their analysis. 3. types of analytical methods for big data risk forecasting the ability to conduct predictive analysis based on large volumes of data is one of the interesting opportunities arising from the spread of large data architectures. an increase in computing power and memory, along with improved algorithms and a better knowledge of their application now enable us to build models from very large data sets [23]. in essence, the techniques of predictive analysis, which attempt to discover patterns and capture relationships in 162 r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 data, can be divided into two groups: while some of the techniques are aimed at discovering historical patterns in the outcome variables and extrapolating them into the future, others attempt to capture and use the interdependencies between the outcome variables and the explanatory variables for predictions. among the techniques commonly used nowadays for predicting large amounts of data are logistic and linear regression models, decision trees, perceptrons, artificial neural networks, association rules, k-nearest neighbors, latent semantic analysis, naive bayes classifier and random forests [27]. further innovations in techniques for big data forecasting can be expected in the near future, with real-time analysis in particular, due to the growth of location-based social media and mobile applications, likely to become a promising field of research and development [17]. notwithstanding the hype surrounding the various methods for big data forecasting, the application of such analytics is still a labor intensive process [28]. recent solutions for the analysis of big data are often based on proprietary applications or on general purpose software systems. typically, organizations require significant efforts to adapt such solutions to their individual needs, which may include integrating different data sources and implementing the software on the organization’s hardware. today, a number of advanced technologies for analyzing large amounts of data (e.g. bigquery, hadoop, mapreduce, wibidata, and skytree) are commercially available that enable the generation of insights to improve organizational strategies and decision-making processes. the remainder of the article will focus on big data implications for risk forecasting in the areas of economic and financial risks, environmental and sustainable development risks, and public and national security risks. 4. public and national security risks forecasting intelligence agencies and law enforcement authorities around the world are confronted with an increasingly complex spectrum of threats from a variety of different origins and backgrounds: national governments, groups and individuals with widely differing motivations and a growing arsenal of approaches. terrorism, crime and other security challenges can only be adequately dealt with if reliable information is available in good time and successful operations must be able to use information from a variety of sources, such as human intelligence, signals intelligence and open source intelligence. big data promises to uncover hitherto undetected security-related patterns and to reveal unexpected hidden knowledge that may hold the key to the prevention of future crimes or acts of terrorism [5,29,30]. public authorities therefore have invested substantially in big data collection platforms and technologies for gathering and analyzing information [31]. by increasingly adopting data processing methodologies for forecasting purposes, security experts have been influenced by similar big-data applications in the corporate world. in intelligence, counterterrorism, policing and peacekeeping, operations have been transformed by the capabilities of big data and predictive analysis to detect unexpected security-related patterns and identify potential threats [32,33]. predictive analysis with big data offers security experts the promise of safeguarding the future by anticipating the next terrorist attack and detecting potential crimes before they are committed. consequently, predictive analysis is used for the purpose of forward-looking decision-making to respond to the growing range of security issues, from terrorism and crime to natural disasters and poverty [34–36]. the shift from past data analysis to the prediction of future events is a central claim of big data analytics. this increased focus on the future will presumably reinforce the division between surveillance on the one hand and case history on the other, and the intense pursuit of pattern discovery will probably warrant a significant expansion of data access [5,37]. in the context of policing, the forecasting promise of big data has received considerable public recognition, making predictive policing one of the latest expressions of a “big data revolution” for security operations [34]. predictive policing, defined as the application of analytical techniques, and in particular quantitative techniques, to help in the identification of promising targets for police interventions and to prevent or solve crimes, can offer a number of distinct advantages to law enforcement agencies [38]. police authorities in the united states and europe have recently purchased commercially available software (e.g. predpol, keycrime) for predicting crimes [25]. the integration of artificial intelligence (ai), machine learning and large databases in counterterrorism and crime-fighting has been applied to a variety of intelligence and operational missions, such as the determination of the structure of terrorist networks and criminal organizations, the localization of high value targets, etc. [29]. the use of ai and big data analytics for intelligence objectives is a novel approach to information extraction that transforms each stage of the intelligence process, starting with the gathering and analysis of information, through to the development of the intelligence picture and the conversion of the information into operational measures [39]. the use of forecasting with big data in the intelligence and counter terrorism arena has, not least because of the edward snowdon revelations, sparked intense controversy between advocates of the pursuit and expansion of the deployment of this technology and its critics. advocates of the use of big data and ai in the intelligence domain argue that their effectiveness in this area has long been proven and that many authorities around the world have used them and have experienced great success [32]. these proponents argue that almost everyone today has a digital footprint that can be traced and analyzed, and therefore much data can be collected through the use of mobile phones, computer systems, apps, social networks, electronic communications, and many other technologies [34]. arguments opposing the use of big data analysis can be divided into three categories: generic arguments reflecting concerns about the growing use of big data and ai and the impact of these developments on our modern society as a whole; use related claims that it is impossible to exploit big data in an effective way in terrorism and crime prevention; and ethical considerations, which suggest that the possibility of harm to innocent civilians caused by the use of big data prediction in intelligence and crime prevention should preclude the use of this technology [29]. the traditionally fragile equilibrium between effectiveness in the fight against crime and terrorism and the liberal democratic principles of society is becoming even more critical when ai and big data analytics-based countermeasures are employed. the integration of ai and big data for intelligence purposes has proven to be effective r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 163 in detecting terrorists and criminal activity and has helped to thwart terrorist attacks [29,40]. however, this technological approach is more advanced than the current regulatory framework and consequently poses a significant risk of violating human rights and privacy [41–43]. moral and ethical issues arising from the use of these technologies include, on the one hand, the fact that the rights of the individual, in particular the privacy of citizens, freedom of speech and opinion, could be violated. if the government can classify certain utterances as suspicious and if it is able to monitor statements made by all citizens on social networks as well as their daily behavior as expressed in various databases, it should be considered that the government can misuse this information. on the other hand, the inherent error margin of this intelligence gathering approach is liable to subvert the legal rights of suspected persons to conduct fair investigations and legal proceedings, causing them irreparable harm and possibly even putting their lives at risk. screening for perpetrators is a difficult methodological problem, whether it is manual profiling by experts or ai and big data-based automatic profiling of individuals. the problem arises from the low prevalence of perpetrators in the population, known as “finding the needle in the haystack”, which can lead to a high rate of false positive results [44]. 4.1. big data for predicting risks of political instability forecasting with big data has become increasingly important, not only in the areas of counterterrorism and crime-fighting, as described above, but also in predicting risks of political instability. companies, governments and international organizations have good reasons to make efforts to anticipate such risks. businesses want to know the risks of investing in volatile sectors, while governments must focus their policies and foreign aid on alleviating human suffering and economic collapse [45]. the capability to successfully predict political instability – from the risk of societal unrest, riots and protests to civil wars and interstate wars, violent coups, genocides or state collapse – would profoundly alter the ability of nations to proactively respond to global instability and intervene before unrest escalates into conflict, or to provide the means to improve preventive measures. one of the most important means of feeding modelling approaches in the field of conflict forecasting is the event database. political event data are recordings of interactions between political actors using a shared set of codes for both actors and actions, which allow for an aggregated analysis of political behavior [4]. databases of such events are essentially recordings of both material interactions between political entities and verbal expressions, which are used to search for temporal patterns in the chronological order and intensity of the records [46]. essentially, conflict forecasting approaches have assumed three different forms: (1) individual experts who summarize available information and form a judgement, (2) collections of such experts who are brought together to form a consensus view, and (3) data-driven computer models that use patterns of past actions to predict future physical behavior. recent progress in computational methods, and in particular in text analysis, has enabled automation of the data-drive approach significantly, thus allowing far more comprehensive sources of real-time information to be analyzed, thereby facilitating the transition from structural to short-term registration of tensions and other conflict characteristics [47]. event databases like defense advanced research projects agency (darpa)’s integrated conflict early warning system [48] and the european media monitor [49] are currently among the most prominent of these automated and integrated systems to monitor, assess, and forecast national, sub-national, and internal crises. both compile extensive real-time archives of incidents of physical disturbances in countries of interest and compile regular reports that summarize the main emerging spatiotemporal dynamics. machine-readable documents on political issues, including press releases, announcements, presentations, press briefings, and intelligence reports, the volume of which is constantly increasing, have become the basis of many policy analyses [4]. the expansion of machine-readable content is being driven by the increasing online publication of established media sources such as agence francepresse and reuters, the growth of a number of locally based agencies that regularly report in english and publish these reports online, and the advent of international media sources including bbc monitoring, al-jazeera, xinhuanet and allafrica. the constantly rising availability of these types of documents offers both opportunities and challenges for risk forecasting of political instability [4,47,50]. while these changes in the global news media industry, as well as worldwide internet accessibility, provide significant potential for novel applications, the inherent process of generating event data, which is not static but rather very dynamic, also creates a number of technical challenges and requires frequent, if not continuous, validation activities from users. an advantage of machine-encoded event data as a viable alternative to human experts is the significantly lower cost of machine coding, for which there is virtually no marginal costs in a well-developed system if the texts are collected automatically. additionally, it is much easier and more cost-effective to modify an existing machine-coding solution than to instruct human experts in an existing protocol [4]. in addition to news, information from social media is also attracting increasing attention, the advantage of which is both the considerable amount of data and the very nature of turning every user into a potential reporter, thus potentially extending the range of events covered considerably and facilitating real-time coverage of a wide range of events [47,51]. however, their use poses considerable challenges, as social media diffuses information and mobilizes people, and most content, such as twitter content, is either not relevant for the purposes of conflict prediction or has been deliberately implanted with false information for manipulation [47,52]. 5. environmental and sustainable development risks forecasting environmental risk forecasting contains many inherent uncertainties due to factors that go beyond the scope of ecology (e.g. demographic change, climate change, governance measures), unknown reactions in interconnected socio-ecological systems and unforeseeable human actions. the importance of estimating future environmental risks has increased in recent years as the pace of change processes has been accelerating and the level of uncertainties has risen [53]. with big data analytics tools, there is now a variety of novel ways to capture complex ecosystem interrelationships and provide the information needed to conduct environmental risk 164 r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 forecasting on a larger scale [54]. the following sections present a brief summary of the impact of big data analytics in the forecasting of climate risks and sustainable development risks. 5.1. big data impact on climate risk forecasting climate change is a major challenge for society, particularly in terms of its capacity to take individual and collective decisions that will enable appropriate responses to address it [55]. in many respects, it differs from other environmental problems facing modern civilization in its time scale and in its complex relationship between human activities, the embedded societal structures and interactions that are emerging between different environmental systems [56]. climate change is leading to cascade-like risks in technical installations, ecological systems, the economy, and society, all of which are often interlinked and create the conditions for irreversible and unwanted exceeding of threshold values at various levels [57]. forecasting climate risks across sectors and in a way that is meaningful to decision-makers thus represents a major scientific challenge. big data analytics is seen as very promising in terms of predicting risks associated with climate change [22,58,59]. the intergovernmental panel on climate change has described the benefits of a risk-based approach to better understand both the dynamic interactions of spatial and temporal determinants leading to specific impacts of climate change, and the role of adaptation initiatives in managing corresponding risks [55]. the most fundamental components for the analysis and prediction of climate risks are verifiable, up-to-date and comparable data and relevant modelling. conventional approaches to risk forecasting and assessment are challenged by the substantial temporal and geospatial dynamics of climate change, by the enhancement of risks of certain societal settings, and by the interaction of several risk factors. however, today big data from climate model simulations is increasingly being used to predict future trends in climate change and to assess the associated risks [59,60]. there are various ways in which big data elements could contribute to improving the modelling of climate risks and impacts. new forms of data could be useful for the calibration of risk models and crowdsourcing and crowdsensing data collected for a specific purpose could be useful, as the assumption of constancy can be justified by reference to the user base [22]. to manage their responses, stakeholders and policy makers need to forecast the potential local risk impacts of climate change at the county-to-city level. part of this information could be derived by combining fine-grained climate risk assessments with ai-based big data analytics of weather extremes, property damage, health impairment and other variables [61]. a number of big data analytics-based tools for screening climate risks are currently being developed, including the world bank’s “climate and disaster risk screening tool” [62], and many institutions are using them to better understand climate risk in their decision-making. the primary challenge in forecasting the risks of global climate change is clearly the complexity and myriad of interacting factors. each incremental change in greenhouse emissions and temperature gives rise to different responses in climatological, ecological, hydrological and other biophysical systems, varying from short term impacts on primary productivity to longer term effects such as rising sea levels, degradation or land formation, whereby the coupling of systems can lead to reactions that affect other systems, including feedback effects on climate [56]. some recent studies emphasize the changing nature of the three components of risk (hazard, exposure and vulnerability) and point to the need for the development of coherent guidance on strategies and methodologies that better account for the dynamic nature of the individual risk components and their interaction [63]. this is particularly important since climate risks are not only a function of physical processes and shifting characteristics of climate systems but are also shaped by complex interactions with socio-economic drivers that can change and evolve within the macroscale conditions and may also change norms and values. with the driving forces and physical consequences of climate change being better known, researchers are increasingly turning their focus on analyzing these socio-economic drivers of climate change. big-data elements could become useful in this research field, as there are no well-proven universal theoretical concepts for such target systems [22]. 5.2. risk indicators for sustainable development risks forecasting big data for development, which identifies the potential of big data analytics to produce practical information that can be used to improve global development, e.g. by analyzing and forecasting sustainable development risks, could complement established methods of handling sustainable development data by opening up new perspectives on problems and deepening and accelerating analysis [64,65]. there is a growing public understanding of global risks associated with environmental disasters, pollution, land degradation, poverty, food security, migration flows, and levels of violence and conflict. to quantify these risks and to support governance worldwide, a plethora of performance indicators and databases has been introduced in recent times [66]. an important driving force behind this development is related to the monitoring of the 17 sustainable development goals (sdgs) and the united nations’ call for a “data revolution for sustainable development” [67,68]. the risks to sustainable development are manifold and are shaped by the interactions between a variety of socio-economic factors and the changing physical environment. sdgs represent an unprecedented effort towards global sustainable development, the complexity of which makes the use of relevant risk indicators reasonable. currently, a multitude of organizations such as national and international institutions, universities, think tanks, investment organizations, journals, and reinsurance companies are publishing global risk indicators [66]. such indicators are crucial to many sustainability initiatives because they are a useful tool for generating knowledge on complex issues, facilitating informed decision making and allow for effective communication between experts and non experts as well as invaluable awareness raising on specific issues [69]. for the compilation of risk indicators, data quality is of great importance, especially considering the fact that this data can significantly influence the outcome of policies. big data is expected to have the potential to support addressing development challenges and meeting the requirements for the development of sdg risk indicators [70]. although risk indicators have been used in the past to forecast risks to socio-economic developments, in some cases the amount of data available allows big data to provide better predictions and improve forecasting accuracy by complementing existing statistical series with more granular, higher-frequency data. r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 165 6. economic and financial risks forecasting technological advances have transformed economies and financial markets into increasingly more complex and dynamic systems, with market participants becoming ever more interconnected, transactions managed on timescales in the sub-millisecond range and masses of data generated, stored and processed [12,71]. these developments present new challenges and opportunities with regard to the understanding and management of risks in financial systems [72,73]. part of this is the question of how we can leverage big data to create more accurate analytical methods for forecasting bubbles and collapses in economic and financial systems. the global financial crisis of 2008 has prompted a large number of regulatory changes, but little progress has been made in providing early information on the vulnerabilities and risks of banks [74]. therefore, building accurate scenario models of future financial risks and vulnerabilities is essential for economists, business leaders, private and institutional investors and policy makers to be able to make a realistic assessment of future economic developments and their implications, and to be prepared to respond appropriately. as analysis of the 2008 financial crisis has suggested, conventional models based on standard economic and financial estimates were not entirely satisfactory [6]. forecasting banking distress is an important issue, with many efforts focused on identifying risk accumulation at an early stage, currently often using aggregate accounting data to measure imbalances. however, despite their rich information content, accounting data pose great challenges due to the low reporting frequency and long publication cycles. a key problem with traditional models is that they are usually unable to take into account the psychological responses of financial investors to certain incidents and disclosures. market participants make predictions about the future development of share prices and other investment opportunities, which could be influenced by their mood or by inconsistent expectations, and the neglect of the entirety of these emotional behaviors could greatly distort the predictions of the models and lead to significant negative consequences, up to system-wide financial crises. most recent research that applies a form of text-based sentiment analysis to investigate the state of the economy or financial markets is using either news or social media generated data. the spread of the internet and social media has created a huge amount of new data containing potentially revealing information about the sentiments, opinions, expectations and fears of its users. a better understanding of behavior in financial markets is expected to provide a more solid basis for political and economic decision-making and support risk management strategies [3,6,75]. so far, however, the analysis of social media data has mostly provided short-term indications that are of limited use for fundamental analysis. such evidence is most useful for explaining events in retrospect rather than for making predictions. nevertheless, social media analysis is an area that is considered to have great potential for future exploration and research [1]. an algorithmic analysis of sentiment trends in large volumes of financial news documents was used, for instance, by nyman et al. to assess how narratives and moods play a role in influencing developments in the financial system [3,76]. according to nyman et al., in their study, changes in emotional content in market narratives are highly correlated across data sources. many financial institutions are currently exploring innovative ways of using big data analysis to improve their internal risk assessment systems [77]. the development of better models identifying highrisk areas could improve the tools available to regulators for early detection of potential financial crises. an effort that would serve all stakeholders and could significantly increase efficiency would be to develop innovative ways to facilitate the management and exchange of data within the financial industry and with both academic researchers and national regulatory authorities. a corresponding research and development approach based on blockchain technology, where data is encrypted by breaking it down into blocks that are distributed to computer nodes, is currently being pursued by massachusetts institute of technology (mit) researchers [1]. such a solution could have the potential to become a trusted tool for financial industry stakeholders for establishing new aggregated risk metrics and for understanding systemic risks better, because if financial institutions are confident, they can exchange data without revealing proprietary secrets. 7. conclusion big data offers substantial opportunities for improving risk forecasting, but may not replace the significance of appropriate assumptions, adequate data quality and continuous validation [2,78,79]. although there are different understandings as to whether or not the main methods of risk analysis for large amounts of data are similar to conventional methods, it is widely considered that the availability of big data allows novel risk analysis. big data and predictive analytics cannot provide a sure-fire method for identifying all critical problems before they occur, but big data-driven development of more accurate early-warning indicators and ways to monitor patterns, are approaches that are more likely to avert an imminent risk and are a worthwhile effort for a range of different fields of application [1,10,80]. however, in order to derive the most benefit from it, the resulting advances at the interface of machine learning, statistics, and ai must be linked to an appropriate methodological basis, especially with regard to the possible ethical, social and legal consequences of a possible misuse of the application of such methods [5]. conflicts of interest the author declares no conflicts of interest. references [1] dahleh m, ozdaglar a, lo aw, bruce e, wilbur j. workshop on data, analytics, and risk in finance summary report. cambridge (ma): mit institute for data, systems, and society; 2016, pp. 1–10. [2] bakdash j, marusich l. risk analysis in big data. 2015. available from: ssrn 2641726. [3] nyman r, kapadia s, tuckett d, gregory d, ormerod p, smith r. news and narratives in financial systems: exploiting big data for systemic risk assessment. bank of england, staff working paper no. 704, 2018. 166 r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 [4] beieler j, brandt pt, halterman a, schordt pa, simpson em. 3 generating political event data in near real time: opportunities and challenges. in: alvarez r, editor. computational social science: discovery and prediction. cambridge: cambridge university press; 2016, pp. 98–120. [5] aradau c, blanke t. politics of prediction: security and the time/ space of governmentality in the age of big data. eur j soc theory 2017;20:373–91. [6] audrino f. financial risk forecasting in the era of big data: the role of investors’ sentiment and attention. 2019. available from: https:// www.openaccessgovernment.org/financial-risk-forecasting/65964/. [7] vose d. risk analysis: a quantitative guide. chichester: john wiley & sons; 2008, pp. 1–729. [8] jorion p. risk management lessons from the credit crisis. eur finan manage 2009;15:923–33. [9] aven t. on the meaning of a black swan in a risk context. saf sci 2013;57:44–51. [10] helbing d. globally networked risks and how to respond. nature 2013;497:51–9. [11] centeno ma, nag m, patterson ts, shaver a, jason windawi a. the emergence of global systemic risk. annu rev sociol 2015;41:65–85. [12] renn o. systemic risks: the new kid on the block. environ sci policy sustain dev 2016;58:26–36. [13] cerchiello p, giudici p. big data analysis for financial risk management. j big data 2016;3:18. [14] lohr s. dataism: the revolution transforming decision making, consumer behavior, and almost everything else. new york: harper business; 2015. [15] reinsel d, gantz j, rydning j. data age 2025: the evolution of data to life-critical. don’t focus on big data: focus on the data that’s big. framingham, ma, usa: idc whitepaper us44413318, international data corporation; 2017, pp. 2–24. [16] manyika j, chui m, brown b, bughin j, dobbs r, roxburgh c, et al. big data: the next frontier for innovation, competition, and productivity. technical report, mckinsey global institute, 2011. [17] gandomi a, haider m. beyond the hype: big data concepts, methods, and analytics. int j inform manage 2015;35:137–44. [18] chen m, mao s, liu y. big data: a survey. mobile netw appl 2014;19:171–209. [19] mcafee a, brynjolfsson e. big data: the management revolution. harv bus rev 2012;90:60–6, 68, 128. [20] human brain project. 2020. available from: https://www. humanbrainproject.eu/. [21] commonwealth scientific and industrial research organisation (csiro). australian square kilometre array pathfinder (askap). 2020. available from: https://www.atnf.csiro.au/projects/askap/ index.html. [22] knüsel b, zumwald m, baumberger c, hadorn gh, fischer em, bresch dn, et al. applying big data beyond small problems in climate research. nat climate change 2019;9:196–202. [23] junqué de fortuny e, martens d, provost f. predictive modeling with big data: is bigger really better? big data 2013;1:215–26. [24] domingos p. the master algorithm: how the quest for the ultimate learning machine will remake our world. new york: basic books; 2015. [25] hosni h, vulpiani a. forecasting in light of big data. philos technol 2018;31:557–69. [26] lazer d, kennedy r, king g, vespignani a. big data. the parable of google flu: traps in big data analysis. science 2014;343:1203–5. [27] mackenzie a. the production of prediction: what does machine learning want? eur j cult stud 2015;18:429–45. [28] sivarajah u, kamal mm, irani z, weerakkody v. critical analysis of big data challenges and analytical methods. j bus res 2017;70:263–86. [29] ganor b. artificial or human: a new era of counterterrorism intelligence? stud conflict terror 2019;1–20. [30] staniforth a, akhgar b. chapter 3 harnessing the power of big data to counter international terrorism. in: akhgar b, editor. application of big data for national security. oxford, uk: butterworth-heinemann; 2015, pp. 23–38. [31] hollin r. chapter 2 drilling into the big data gold mine: data fusion and high-performance analytics for intelligence professionals. in: akhgar b, editor. application of big data for national security. oxford, uk: butterworth-heinemann; 2015, pp. 14–20. [32] cheng t, bowers k, longley p, shawe-taylor j, davies t, rosser g, et al. cpc: crime, policing and citizenship–intelligent policing and big data. london: ucl spacetime lab; 2016. [33] read r, taithe b, mac ginty r. data hubris? humanitarian information systems and the mirage of technology. third world quart 2016;37:1314–31. [34] flyverbom m, hansen hk. policing and anticipatory transparency: on digital transformations, proactive governance and logics of temporality. in: august v, osrecki f, editors. der transparenzimperativ. wiesbaden: springer vs; 2019, pp. 171–86. [35] amoore l. the politics of possibility: risk and security beyond probability. durham: duke university press; 2013, pp. 1–232. [36] hansen hk. numerical operations, transparency illusions and the datafication of governance. eur j soc theory 2015;18:203–20. [37] lyon d. surveillance, snowden, and big data: capacities, consequences, critique. big data soc 2014;1:2053951714541861. [38] perry wl, mcinnis b, price cc, smith sc, hollywood js. predictive policing: the role of crime forecasting in law enforcement operations, santa monica, ca: rand corporation; 2013. [39] kernchen r. coping with complexity in biological threat assessment. 2020, p. 12. available from: ssrn 3634621. [40] fox b, reid ja, masys aj. science informed policing. cham, switzerland: springer; 2020. [41] walsh pf. intelligence leadership and governance: building effective intelligence communities in the 21st century. london: routledge; 2020, p. 214. [42] mckendrick k. artificial intelligence prediction and counterterrorism. london: the royal institute of international affairs chatham house; 2019. [43] hasisi b, perry s, wolfowicz m. counter-terrorism effectiveness and human rights in israel. in: shor e, hoadley s, editors. international human rights and counter-terrorism. singapore: springer; 2019, pp. 1–21. [44] neuman y, cohen y, neuman y. how to (better) find a perpetrator in a haystack. j big data 2019;6:9. [45] hendrix cs. keeping up with the future: upgrading forecasts of political instability and geopolitical risk. policy briefs pb19-10, peterson institute for international economics; 2019. [46] leetaru k. can we forecast conflict? a framework for forecasting global human societal behavior using latent narrative indicators. urbana, illinois, usa: university of illinois at urbanachampaign; 2016. [47] chadefaux t. conflict forecasting and its limits. data sci 2017;1:7–17. [48] lockheed martin. integrated conflict early warning system (icews). 2020. available from: https://www.lockheedmartin. com/en-us/capabilities/research-labs/advanced-technology-labs/ icews.html. https://doi.org/10.1017/cbo9781316257340.005 https://doi.org/10.1017/cbo9781316257340.005 https://doi.org/10.1017/cbo9781316257340.005 https://doi.org/10.1017/cbo9781316257340.005 https://doi.org/10.1017/cbo9781316257340.005 https://doi.org/10.1177/1368431016667623 https://doi.org/10.1177/1368431016667623 https://doi.org/10.1177/1368431016667623 https://www.openaccessgovernment.org/financial-risk-forecasting/65964/ https://www.openaccessgovernment.org/financial-risk-forecasting/65964/ https://doi.org/10.1111/j.1468-036x.2009.00507.x https://doi.org/10.1111/j.1468-036x.2009.00507.x https://doi.org/10.1016/j.ssci.2013.01.016 https://doi.org/10.1016/j.ssci.2013.01.016 https://doi.org/10.1038/nature12047 https://doi.org/10.1038/nature12047 https://doi.org/10.1146/annurev-soc-073014-112317 https://doi.org/10.1146/annurev-soc-073014-112317 https://doi.org/10.1080/00139157.2016.1134019 https://doi.org/10.1080/00139157.2016.1134019 https://doi.org/10.1186/s40537-016-0053-4 https://doi.org/10.1186/s40537-016-0053-4 https://doi.org/10.1016/j.ijinfomgt.2014.10.007 https://doi.org/10.1016/j.ijinfomgt.2014.10.007 https://pubmed.ncbi.nlm.nih.gov/23074865/ https://pubmed.ncbi.nlm.nih.gov/23074865/ https://www.humanbrainproject.eu/ https://www.humanbrainproject.eu/ https://www.atnf.csiro.au/projects/askap/index.html https://www.atnf.csiro.au/projects/askap/index.html https://doi.org/10.1038/s41558-019-0404-1 https://doi.org/10.1038/s41558-019-0404-1 https://doi.org/10.1038/s41558-019-0404-1 https://doi.org/10.1089/big.2013.0037 https://doi.org/10.1089/big.2013.0037 https://doi.org/10.1007/s13347-017-0265-3 https://doi.org/10.1007/s13347-017-0265-3 https://doi.org/10.1126/science.1248506 https://doi.org/10.1126/science.1248506 https://doi.org/10.1177/1367549415577384 https://doi.org/10.1177/1367549415577384 https://doi.org/10.1016/j.jbusres.2016.08.001 https://doi.org/10.1016/j.jbusres.2016.08.001 https://doi.org/10.1016/j.jbusres.2016.08.001 https://doi.org/10.1080/1057610x.2019.1568815 https://doi.org/10.1080/1057610x.2019.1568815 https://doi.org/10.1080/01436597.2015.1136208 https://doi.org/10.1080/01436597.2015.1136208 https://doi.org/10.1080/01436597.2015.1136208 https://doi.org/10.1007/978-3-658-22294-9_7 https://doi.org/10.1007/978-3-658-22294-9_7 https://doi.org/10.1007/978-3-658-22294-9_7 https://doi.org/10.1007/978-3-658-22294-9_7 https://doi.org/10.1177/1368431014555260 https://doi.org/10.1177/1368431014555260 https://doi.org/10.1177/2053951714541861 https://doi.org/10.1177/2053951714541861 https://doi.org/10.1007/978-981-10-3894-5_22-1 https://doi.org/10.1007/978-981-10-3894-5_22-1 https://doi.org/10.1007/978-981-10-3894-5_22-1 https://doi.org/10.1007/978-981-10-3894-5_22-1 https://doi.org/10.1186/s40537-019-0172-9 https://doi.org/10.1186/s40537-019-0172-9 https://doi.org/10.3233/ds-170002 https://doi.org/10.3233/ds-170002 https://www.lockheedmartin.com/en-us/capabilities/research-labs/advanced-technology-labs/icews.html https://www.lockheedmartin.com/en-us/capabilities/research-labs/advanced-technology-labs/icews.html https://www.lockheedmartin.com/en-us/capabilities/research-labs/advanced-technology-labs/icews.html r. kernchen / journal of risk analysis and crisis response 10(4) 160–167 167 [49] european commission. european media monitor. 2020. available from: https://emm.newsbrief.eu/overview.html. [50] schrodt pa. comparing methods for generating large scale political event data sets. text as data meetings. new york: new york university; 2015. [51] donnay k. big data for monitoring political instability. international development policy. revue internationale de politique de développement, 2017, p. 8. [52] suh jh. forecasting the daily outbreak of topic-level political risk from social media using hidden markov model-based techniques. technol forecast soc change 2015;94:115–32. [53] bengston dn, kubik gh, bishop pc. strengthening environmental foresight: potential contributions of futures research. ecol soc 2012;17:10. [54] van den brink pj, choung cb, landis w, mayer-pinto m, pettigrove v, et al. new approaches to the ecological risk assessment of multiple stressors. mar freshwater res 2016;67:429–39. [55] masson-delmotte v, zhai p, pörtner h, roberts d, skea j, shukla pr, et al. global warming of 1.5°c. an ipcc special report on the impacts of global warming of 1.5°c above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty. geneva, switzerland: intergovernmental panel on climate change; 2018, p. 630. [56] neil adger w, brown i, surminski s. advances in risk assessment for climate change adaptation policy. london: the royal society publishing london; 2018. [57] king d, schrag d, dadi z, ye q, ghosh a. climate change: a risk assessment. in: hynard j, rodger t, editors. london, beijing, delhi, cambridge (ma): centre for science and policy; 2017, pp. 1–79. [58] kennel cf, briggs s, victor dg. making climate science more relevant. science 2016;354:421–2. [59] zhang z, li j. big data mining for climate change. amsterdam, oxford, cambridge (ma): elsevier; 2019, pp. 1–346. [60] ganguly ar, kodra e, bhatia u, warner me, duffy k, banerjee a, et al. data-driven solutions. climate 2020: degrees of devastation. united nations association, 2018, pp. 82–5. available from: www.una.org.uk/climate-2020-degrees-devastation. [61] xu y, ramanathan v, victor dg. global warming will happen faster than we think. nature 2018;564:30–2. [62] world bank group. climate & disaster risk screening tool. 2019. available from: https://olc.worldbank.org/content/climate disaster-risk-screening-tool. [63] viner d, ekstrom m, hulbert m, warner nk, wrefrod a, zommers z. understanding the dynamic nature of risk in climate change assessments—a new starting point for discussion. atmos sci lett 2020;21:e958. [64] hilbert m. big data for development: a review of promises and challenges. dev policy rev 2016;34:135–74. [65] flyverbom m, madsen ak, rasche a. how big data reshapes knowledge for international development–a governmentality perspective. the 32nd egos colloquium 2016: organizing in the shadow of power. napoli, italy: copenhagen business school, centre for corporate social responsibility (cbscsr); 2016, p. 35. [66] visser h, de bruin s, martens a, knoop j, ligtvoet w. what users of global risk indicators should know. glob environ change 2020;62:102068. [67] un ieag. a world that counts–mobilising the data revolution for sustainable development. new york: un secretary-general’s independent expert advisory group on a data revolution for sustainable development; 2014, pp. 1–32. [68] sachs j, kroll c, schmidt-traub g, lafortune g, fuller g. sustainable development report 2019. new york, ny, usa: bertelsmann stiftung and sustainable development solutions network (sdsn); 2019. [69] mair s, jones a, ward j, christie i, druckman a, lyon f. a critical review of the role of indicators in implementing the sustainable development goals. in: leal filho w, editor. handbook of sustainability science and research. world sustainability series. cham, springer; 2018, pp. 41–56. [70] hammer c, kostroch mdc, quiros mg. big data: potential, challenges and statistical implications. washington, dc: international monetary fund; 2017. [71] oecd and international institute for applied systems analysis. systemic thinking for policy making: the potential of systems analysis for addressing global policy challenges in the 21st century. in: hynes w, lees m, müller jm, editors. new approaches to economic challenges. paris: oecd publishing; 2020, pp. 1–174. [72] flood md, jagadish h, raschid l. big data challenges and opportunities in financial stability monitoring. finan stab rev 2016;20:129–42. [73] cincotti s, sornette d, treleaven p, battiston s, caldarelli g, hommes c, et al. an economic and financial exploratory. eur phys j spec top 2012;214:361–400. [74] rönnqvist s, sarlin p. bank distress in the news: describing events through deep learning. neurocomputing 2017;264:57–70. [75] kim a, yang y, lessmann s, ma t, sung mc, johnson jev. can deep learning predict risky retail investors? a case study in financial risk behavior forecasting. eur j oper res 2020;283:217–34. [76] bholat psd, hansen s, schonhardt-bailey c, santos p. text mining for central banks. available from: ssrn 2624811. [77] liang y, quan d, wang f, jai x, li m, li t. financial big data analysis and early warning platform: a case study. ieee access 2020;8:36515–26. [78] buchanan m. the limits of machine prediction. nat phys 2019;15:304. [79] hassani h, silva es. forecasting with big data: a review. ann data sci 2015;2:5–19. [80] choi tm, lambert jh. advances in risk analysis with big data. risk anal 2017;37:1435–42. https://emm.newsbrief.eu/overview.html https://doi.org/10.4000/poldev.2468 https://doi.org/10.4000/poldev.2468 https://doi.org/10.4000/poldev.2468 https://doi.org/10.1016/j.techfore.2014.08.014 https://doi.org/10.1016/j.techfore.2014.08.014 https://doi.org/10.1016/j.techfore.2014.08.014 http://dx.doi.org/10.5751/es-04794-170210 http://dx.doi.org/10.5751/es-04794-170210 http://dx.doi.org/10.5751/es-04794-170210 https://doi.org/10.1071/mf15111 https://doi.org/10.1071/mf15111 https://doi.org/10.1071/mf15111 https://doi.org/10.1071/mf15111 https://doi.org/10.1098/rsta.2018.0106 https://doi.org/10.1098/rsta.2018.0106 https://doi.org/10.1098/rsta.2018.0106 https://doi.org/10.1126/science.aag3248 https://doi.org/10.1126/science.aag3248 https://doi.org/10.1038/d41586-018-07586-5 https://doi.org/10.1038/d41586-018-07586-5 https://doi.org/10.1002/asl.958 https://doi.org/10.1002/asl.958 https://doi.org/10.1002/asl.958 https://doi.org/10.1002/asl.958 https://doi.org/10.1111/dpr.12142 https://doi.org/10.1111/dpr.12142 https://doi.org/10.1016/j.gloenvcha.2020.102068 https://doi.org/10.1016/j.gloenvcha.2020.102068 https://doi.org/10.1016/j.gloenvcha.2020.102068 https://doi.org/10.1007/978-3-319-63007-6_3 https://doi.org/10.1007/978-3-319-63007-6_3 https://doi.org/10.1007/978-3-319-63007-6_3 https://doi.org/10.1007/978-3-319-63007-6_3 https://doi.org/10.1007/978-3-319-63007-6_3 https://doi.org/10.1140/epjst/e2012-01699-6 https://doi.org/10.1140/epjst/e2012-01699-6 https://doi.org/10.1140/epjst/e2012-01699-6 https://doi.org/10.1016/j.neucom.2016.12.110 https://doi.org/10.1016/j.neucom.2016.12.110 https://doi.org/10.1016/j.ejor.2019.11.007 https://doi.org/10.1016/j.ejor.2019.11.007 https://doi.org/10.1016/j.ejor.2019.11.007 https://doi.org/10.1016/j.ejor.2019.11.007 https://doi.org/10.2139/ssrn.2624811 https://doi.org/10.2139/ssrn.2624811 https://doi.org/10.1109/access.2020.2969039 https://doi.org/10.1109/access.2020.2969039 https://doi.org/10.1109/access.2020.2969039 https://doi.org/10.1038/s41567-019-0489-5 https://doi.org/10.1038/s41567-019-0489-5 https://doi.org/10.1007/s40745-015-0029-9 https://doi.org/10.1007/s40745-015-0029-9 https://doi.org/10.1111/risa.12859 https://doi.org/10.1111/risa.12859 microsoft word re-multiple internet of intelligences for risk analysis.doc multiple internet of intelligences for risk analysis* chongfu huang1,2 1. academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china 2. beijing cazl technology service co., ltd., beijing 100088, china e-mail: hchongfu@bnu.edu.cn received 24 april 2014 accepted 4 june 2014 abstract in this study, we define a multiple internet of intelligences (m-ioi) as the processing of homological information in layers. if agents in an m-ioi not only provide information in response to a question asked by a customer but also review information from other agents and summarize it, we refer to this as a summarizing m-ioi. the fuzzy mathematics method of normal diffusion is suggested to transform the summaries into fuzzy sets so that a satisfactory answer to the question is given. a summarizing m-ioi is used in a case study of typhoon dynamic risk in wenzhou, china, where an insurance company wants to know whether the level of the risk will increase significantly. the effective knowledge in a summarizing m-ioi is measured to evaluate the quality of the answer. we also discuss the relation between iois and a global brain. keywords: internet of intelligences; risk analysis; fuzzy set; normal diffusion; typhoon dynamic risk 1. introduction *project supported by the national basic research program of china (973 program) (no.2012cb955402), and partly supported by the beijing cazl technology service co., ltd.. an internet of intelligences (ioi) is a new website tool that collects and integrates messages, experience, knowledge, and judgment from intelligent agents to provide answers to questions from users (huang, 2011). unlike google and quora (gannes, 2010), an ioi uses models to process the information from various agents to produce satisfactory answers. a redesigned google website could directly process “millions of different fact-seeking searches” and provide short answers at part of its results. “google questions and answers” would allow users to collaboratively find satisfactory answers to their questions through the web. quora considers the contributions it inspires a sort of “inverse blogging.” if someone asks a question, it is because he or she wants an answer. google is a web search engine that allows users find other sites on the web based on keyword searches. any answer suggested by google must be supported by the sheer volume of web pages. quora is a dialogue platform in which a user can ask a question on the quora website and quickly receive answers from the network members who possess the relevant knowledge (badilescu-buga, 2013). neither google nor quora fully utilizes the pool of opinions on the internet to answer questions. the main reason is that, although semantic analysis appears to be theoretically powerful on the web, it cannot process the missing structured data (habernal and konopik, 2013) to provide a satisfactory answer. the concept of an ioi arose from online services for customers who have questions regarding some element of risk. there are always others who have experience, knowledge, and judgment concerning these questions (huang, 2011). an ioi is an internet-based service in which people employed by the service act as intelligent agents, and models are used to process the information provided by the agents. journal of risk analysis and crisis response, vol. 4, no. 2 (june 2014), 61-71 published by atlantis press copyright: the authors 61 c.f. huang the first ioi was constructed to serve applicants to universities based on their entrance exam scores. the ioi collects and processes suggestions from college students and high school teachers. this system can help applicants choose among colleges and increase the probability of their being accepted to their first choice of schools, which is the greatest concern of their families. another ioi has been used to survey the demand for typhoon insurance from the aquaculture industry in wenzhou, china (ai, 2014). the third ioi was designed to assist in writing proposals for grants. this ioi increases the chances of success for applicants facing a highly competitive environment and asymmetric information. in practice, it is very difficult to build a complex mathematical model to process natural language information in an ioi that is both soft and flexible. to overcome this difficulty in this study, we propose a multiple ioi in which the agents not only provide information but also acquire knowledge by summarizing information. this study is organized as follows. in section 2, we define an ioi. in section 3, we define a multiple ioi with layers. in section 4, we suggest a multiple ioi to summarize information. then, in section 5 we employ the normal diffusion technique to integrate the summaries for risk analysis. section 6 presents an application of a multiple ioi in analyzing typhoon dynamic risk. section 7 presents a model to measure the effective knowledge in an ioi and the quality of service. conclusions are provided in section 8. 2. internet of intelligences the concept of an internet of intelligences (ioi) is based on the observation that everyone can act as an intelligent agent, having relatively more knowledge than others on some topic that may be used to help others to solve some problem through the internet. for example, a tourist can answer questions regarding travel information (e.g., safety, discounts, or fraud) based on his/her experience. an ioi is intended to offer satisfying solutions for customers through human agents that apply their experience, knowledge and judgment to solve the customer’s problems. to define an ioi, we first define the following terms: agent, network and model. an individual who can provide others with experience, knowledge, and judgment to solve problems is called an intelligent agent or simply an agent. a computer system through which the agents can serve customers is called a network. the internet is the most convenient network. in the remainder of the paper, unless stated otherwise, “network” refers to the internet. a scientific model is a representation of an object or a system. any mathematical expression describing relationships among variables, any mathematical operation for processing information and any human brain paradigm for analyzing questions can be called a model. in the remainder of the paper, unless stated otherwise, “model” refers to a mathematical model or a human brain paradigm for processing information. for example, the linear regression method is a mathematical model. the set of rules used by an editor-in-chief of a journal is a human brain paradigm for processing the information in the comments from reviewers on papers submitted to the journal. slightly revising the definition given by huang (huang, 2011), we formally present a definition of an internet of intelligences (ioi): definition 1. let a be a set of agents, let n be a network used by a, and let m be a model to process information provided by a. a triple <a, n, m> is called an internet of intelligences, denoted as f. an ioi is a network that provides knowledge products, where more than one intelligent agent is connected by a computer network and the agents’ experiences, knowledge, judgment are collected and integrated by some model. if the model of an ioi is overly coarse, its product will be of low quality. to evaluate an ioi f, we suppose that there is a tool t that can measure the intelligence level of ai in a={a1, a2,..., an}, i.e., qi = t(ai). in addition, we assume that z can measure the intelligence level of f, i.e., q = z(f). when q>max{q1, q2,..., qn}, the ioi is called a positive ioi. slightly revising the definition given by huang (huang, 2011), we have the following definition: definition 2. let f=<a, n, m> be an internet of intelligences with an intelligence level q. let the highest intelligence level of the individuals in a be q. if q>q, f is called a positive internet of intelligences, denoted as f↑. published by atlantis press copyright: the authors 62 multiple internet of intelligences for risk analysis for example, let a={john, mary, smith} be a team participating in a trivia quiz game. in this competition, all players must wait until the host reads a question, after which a light is turned on as a “ready” signal. the players press a button when they know the answer, and the first player to press their button wins the chance to respond. john, mary and smith use an ioi to compete, but other teams do not. that is, an ioi will compete with the other teams. in this case, the score is used to measure the intelligence level of a player. we suppose that the scores of john, mary and smith in a past competition are q1, q2, and q3, respectively. the team's score is q1+q2+q3. without loss of generality, we assume q1>q2, q3. when the team competes again using the ioi, the team's score is s. the intelligence level of the ioi is defined as q = s/3. if q>q1, the ioi is positive. a simple ioi is illustrated in fig. 1, and its topology is shown in fig. 2. the architecture of an ioi serving customers is illustrated in fig. 3. the most important ingredient for constructing a positive ioi is to have a powerful model m to extract and summarize knowledge from the information provided by the agents and then to construct knowledge products. strictly speaking, it is the task of artificial intelligence to find m. currently, ibm’s “watson”, a supercomputer system capable of answering questions posed in natural language (ferrucci et al., 2013), has the highest level of artificial intelligence. in a recent competition, watson had access to 200 million pages of structured and unstructured content requiring four terabytes of disk storage. this content included the full text of wikipedia, although the system was not connected to the internet during the game. the algorithms used in “watson” are very complex, and the system is supported by an extremely large database. to alleviate the difficulty of discovering a complex model, in the next section we propose a multiple ioi to process information. 3. multiple internet of intelligences in practice, the information in an ioi might be heterogeneous, gathered from multiple sources such as natural language, databases and images. it is very agents customer’s questions databases online documents s m answers fig. 3. an internet of intelligences provides customers with answers to questions. unlike other q&a systems, an ioi uses information from agents and applies models to process the information, where s is a network server and m is a mathematical model. c1 c2 c3 a1 a2 a3 s(network server) m(mathematical model) fig. 1. a simple internet of intelligences composed of a network server s, a mathematical model m, three computers c1, c2, and c3 and three agents a1, a2, and a3. a1 s m a2 a3 fig. 2. topology of the example internet of intelligences composed of three nodes, where s is the network server, m is a mathematical model and a1, a2, and a3 are agents. published by atlantis press copyright: the authors 63 c.f. huang difficult to construct a complex mathematical model to process such information. to alleviate this difficulty, the process is divided into several layers. by analogy, files in the graphics editing software photoshop can be composed of layers containing any manner of fills, strokes, images, fonts and miscellaneous objects to enable easy editing. it would be easier for an ioi to process homological information using the same approach. the homological information includes the experiences of agents, databases, monitoring systems, judgment, and reasoning that can be combined and modified using mathematical models. when the number of the layers in an ioi is greater than one, the ioi is called a multiple internet of intelligences (m-ioi). formally, we give the following two definitions: definition 3. let f=<a, n, m> be an internet of intelligences. a process for processing homological information in f is called a layer of f. definition 4. let f=<a, n, m> be an internet of intelligences. if the number of layers of f is greater than one, f is called a multiple internet of intelligences. an m-ioi consisting of three layers is illustrated in fig. 4. the information in each layer is homological; i.e., it is of the same type. three separate models are used to process the individual layers. lastly, a more complex model is used to integrate the results from the models. in an ioi, a model can be a scientific model, a human brain paradigm or a special tool. we use me, mo, bp, ot to denote mathematical expressions, mathematical operations, human brain paradigms and other tools, respectively. then, a model m in an ioi must be an element or a combination of elements of the set m in eq. (1). { , , , }me mo bp ot=m (1) the most difficult task is to process online information contained in natural language expressions; even the processing only summarizes the information. because humans are better than any mathematical model at understanding and summarizing information in natural language expressions, we employ a human brain paradigm for summarizing information to create an mioi. 4. a multiple internet of intelligences for summarizing the simplest task for model 1 in fig. 4 is to summarize the information provided by agents responding to questions that are asked by a customer. we denote a question as h and the response of agent a as e=a(h). if there are n agents a1, a2, …, an responding to a question h, the ioi will have a set e of n responses, as shown in eq. (2). e={e1, e2, …, en} (2) the function of a “summarizing” model is to process e and output a summary. if e1, e2, …, en are data, it is easy to find a statistical model to summarize them. however, if the responses are expressed in natural language, it is very difficult for a mathematical model to summarize the information. in this case, we invoke the agents to summarize. we do not know how an agent summarizes e, but we know that he/she is able to do so. the means for processing the natural language information is a human brain paradigm. without loss of generality, we assume that an agent a summarizes e and produces a summary ea. for simplicity, the summary ia e given by agent ai is written as ei. in general, the summaries given by the agents will be different. if the summaries are expressed in natural language, it is difficult to find a mathematical model that will provide an integrated answer the customer’s a b d c online layer document layer database layer model 3 model 2 model 1 customer’s questions s m answers fig. 4. a multiple internet of intelligences consists of three layers to process online information, documents and databases separately, where a, b, c, d,… are agents. published by atlantis press copyright: the authors 64 multiple internet of intelligences for risk analysis question. to obtain structured data, “summarizing” must be performed according to some rules. a question arises if and only if there are uncertainties in a customer’s mind. for example, if a businessman has to choose among four cities in a district to invest in a flower shop, his question is h: which cite will give me the largest profit? (3) the profit depends on the drought risk and the number of customers (huang and inoue, 2007). the uncertainties in h are “the drought risks in each of the four cities” and “the number of customers in each of the four cities”. in some sense, answering a question reduces or describes the uncertainties. this is the role of risk analysis. assuming that a question must be related to some uncertainties, an agent can summarize e1, e2, …, en by completing a form (e.g., a questionnaire) and thus reduce the uncertainties. then, the sets e1, e2, …, en will be quantitative. that is, a set of responses e to a question h can be quantified using summaries. the summaries provide data samples that can be processed by a mathematical model to produce a satisfactory answer. figure 5 shows the summarizing m-ioi. in general, a customer purchases a service for the lowest possible price. the service is expected to be performed by some deadline. this expectation implies that the number of agents responding a question will be limited. when the size of a sample is small (below 30), we recommend the normal diffusion technique to process the samples. 5. normal diffusion technique for risk assessment the concept of information diffusion (huang, 1997) was first introduced in function learning from a small sample of data (huang and moraga, 2004). the approximate reasoning of information diffusion was used to estimate probabilities and fuzzy relationships from scant, incomplete data for grassland wildfires (liu et al., 2010). the interior-outer-set model, which is based on information diffusion theory, can be used to calculate a possibility–probability distribution from a small sample. this model provided better multi-valued results for flood risk management (zou et al., 2012). information diffusion theory was used to evaluate accident rates in dangerous chemical transportation and analyze the consequences of such accidents with gis simulation technology (zhang and zhao, 2007). the use of information diffusion for fuzzy mathematics can be illustrated as follows (huang, 2002). let { 1 2 }ix x | i , , ,m= = be a given sample and let }{uu = be its universe. the function in eq. (4) is called a normal diffusion function. .,], 2 )( exp[),( 2 2 uuxx h ux ux ∈∈ − −=μ (4) the diffusion coefficient h can be calculated using eq. (5) (huang, 2012) 0.8146 ( ), 5; 0.5690 ( ), 6; 0.4560 ( ), 7; 0.3860 ( ), 8; 0.3362 ( ), 9; 0.2986 ( ), 10; 2.6851( )/( 1) b a m b a m b a m h b a m b a m b a m b a m − = − = − = = − = − = − = − − , 11.m ⎧ ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎪ ≥⎩ (5) where 11 max { } and min { }i ii mi mb x a x .≤ ≤≤ ≤= = a b d c online layer samples layer customer’s questions s mo answers summarizing (bp) uncertainty 1 uncertainty 2 uncertainty 3 º º fig. 5. a summarizing m-ioi consists of two layers. a human brain paradigm (bp) titled “summarizing” is used to process online information and output data samples. a mathematical operation (mo) is used to process the samples. here, s is a network server, and a, b, c, d,… are agents. published by atlantis press copyright: the authors 65 c.f. huang if ix x∈ is a piece of information describing a risk, we can use the normal diffusion function to process x for risk assessment. specifically, when 1 2 mx , x , , x are m judgments, we can obtain a comprehensive judgment through the following steps: step 1: using eq. (5), calculate a diffusion coefficient h. step 2: using eq. (4) , {1, 2, , },i i m∀ ∈ transform a sample point xi into a fuzzy set with the membership function in eq. (6) ( ) ( ) exp[ ], i 2 i x 2 x u u u u . 2h μ − = − ∈ (6) step 3: sum over all of the membership functions 1 ( ),x uμ 2 ( ), , ( )mx xu uμ μ to obtain the function 2 2 1 ( ) ( ) exp[ ], . 2 m i i x u f u u u h= − = − ∈∑ (7) step 4: find the absolute maximum (u0, f(u0)) of f(u) on u using software such as mathematica or matlab the critical point u0 is the comprehensive judgment we seek. the critical point u0 is more robust than the simple average given in eq. (8); i.e., when m is small, a larger individual deviation will not significantly change the result. 1 2 1 ( )mx x x xm = + + + (8) for example, assume the universe u = [0, 1], and let the judgments be 1 2 3 4 50.3, 0.4x x x x x= = = = = . we have h = 0.8146(0.4-0.3) = 0.08146, 2h2=0.01327 2 2(0.3 ) (0.4 ) ( ) 4 exp[ ]+ exp[ ], . 0.01327 0.01327 u u f u u u − − = − − ∈ the absolute maximum of )(uf (see fig. 6(a)) is at (0.3124, 4.515), so the critical point is u0=0.3124. the simple average of x1, x2, …, x5 is 1 (0.3 0.3 0.3 0.3 0.4) 0.32. 5 x = + + + + = if the judgment of agent 5 differs more from the others, e.g., 5 0.8,x′ = then the absolute maximum of )(uf ′ (see fig. 6(b)) will be at 0( , ) (0.3, 4).u v′ ′ = the simple average is 0.4.x′ = we have 0 0| | 0.0124 0.08 | | .u u x x′ ′− = < = − this example demonstrates that the comprehensive judgment u0 is more robust than the simple average x . in gymnastics competitions, the highest and lowest scores for an athlete are discarded to avoid biased judging. in risk analysis, however, any information is important, particularly when the sample size is small. in this case, the normal diffusion technique can provide a comprehensive judgment that is robust (no one individual can grossly affect a result with a significantly different input). 6. a case study in typhoon dynamic risk china is a country that has suffered devastation from numerous typhoons in its long history. on average, 7.2 typhoons affect china every year. most of these typhoons occur on the southeastern coast of china. however, the typhoon risk is changing because of socio-economic development and population growth. one of the most important issues in planning for typhoon-related damage is to assess typhoon dynamic risk. 6.1. question and responses suppose that a new insurance company is considering selling policies to clients in wenzhou, a coastal city in zhejiang province with nearly 60,000 residents. the company’s profits will be sensitive to the risk of typhoon-related damage. if the level of risk is expected to increase in the next year, the company will not offer 0.2 0.4 0.6 0.8 1 1 2 3 4 0.2 0.4 0.6 0.8 1 1 2 3 4 (a) 1 2 3 4 5 0.3, 0.4 x x x x x = = = = = (b) 1 2 3 4 5 0.3, 0.8 x x x x x = = = = ′ = fig. 6. the normal diffusion technique can robustly give a comprehensive judgment: a critical point. a larger individual deviation can substantially change a function, e.g., (a) versus (b), but the critical points for the absolute maximums, (0.3124, 4.515) and (0.3, 4), are not substantially different. uu v v ( )f u ( )f u′ published by atlantis press copyright: the authors 66 multiple internet of intelligences for risk analysis coverage to clients there, but if the risk is expected to decrease, the company will offer coverage. in general, the company will conduct research or fund an external organization to study this problem. as a result, data on typhoons and resulting disasters will be collected and studied. in addition, reports published by the ipcc and the development plans for wenzhou must be reviewed. this research will require a substantial amount of time and money. in contrast, a multiple ioi supported by climatologists, meteorologists and government offices could accomplish the task more quickly and at a lower cost. suppose that the company issues the following question: our company provides insurance, and we are considering whether to sell policies in wenzhou, where the risk of loss from typhoons is high. if the level of risk is likely to increase next year, possibly due to global warming, we will not sell policies there. if the level of risk is likely to decrease, possibly due to improvements in drainage systems, then we will sell policies there next year. we are unable to decide. please advise us based on inputs from experienced professionals. we assume that the company will pay 6000 rmb to a service with an ioi, and each agent for the ioi will receive 1000 rmb if he/she responds to the question with useful information. furthermore, we assume that 5 agents respond; table 1 lists the five responses. 6.2. quantitative summarizing when an ioi is employed to collect responses, we are confronted with the following problems: (1) some agents may maliciously provide false information in an attempt to discredit the online service. (2) agents may provide inconsistent information as a result of differing perspectives. (3) the information may be incomplete because time and the budget are limited. the ioi will filter false information and substantially inconsistent information. the information remaining is called the effective information. we assume that the responses in table 1, which are denoted as e1, e2, …, e5, contain useful information. for example, e1= “current information shows that global warming will not significantly affect typhoon activity”. let agent i, denoted as ai, summarize all of the responses and give a summary ei, as shown in eq. (9). it should be noted that sumi will differ from sum j because of the different backgrounds of ai and aj. 1 2 5sum { , , , }i ie e e e= (9) to obtain structured data that will be analyzed using the normal diffusion technique, the agents are required to summarize the responses using table 2, which is designed to reduce the five uncertainties in the question asked by the customer. table 1. responses from 5 agents to the question of whether the risk of typhoon-related losses in wenzhou is likely to increase. agent discipline response agent 1 climatologist current information shows that global warming will not significantly affect typhoon activity. agent 2 meteorologist the records of typhoons that have affected wenzhou show that the frequency of typhoons has not significantly changed in recent years, but the intensity of the storms has increased. agent 3 civil affairs bureau staff in wenzhou, typhoon-related losses have increased approximately 5% per year in recent years. agent 4 statistics bureau staff in wenzhou the annual growth of wenzhou's gross domestic product (gdp) is approximately 9%. typhoon-related losses show a decreasing trend. agent 5 water affairs bureau staff in wenzhou the government has been investing heavily to improve drainage facilities. in the next five years, the standard to prevent a flood will be raised from a return period of 30 years to 50 years. table 2. questions for agents to quantitatively summarize responses. 1. will global warming affect the typhoon risk in wenzhou? a. no information b. not at all c. slightly d. moderately e. significantly f. drastically 2. will drainage facilities be improved? a. no information b. not at all c. slightly d. moderately e. significantly f. drastically 3. in recent years, how did typhoon-related losses compare to gdp growth? a. no information b. much lower c. slightly lower d. approximately the same e. slightly higher f. much higher 4. is the typhoon risk decreasing? a. no information b. no c. slightly d. moderately e. significantly f. drastically 5. what should the insurance company do next year? a. no information b. do not sell policies c. more data required d. further analysis required e. prepare to sell policies f. sell policies published by atlantis press copyright: the authors 67 c.f. huang it is simple to transform a summary e into a fivedimensional vector with the mapping function in eq. (10). a 1 b 0.00 c 0.25 d 0.50 e 0.75 f 1 → −⎧ ⎪ →⎪ ⎪ →⎪ ⎨ →⎪ ⎪ → ⎪ →⎪⎩ (10) let rij be the option of the ith agent to the jth question. for example, agent 1 chooses options c, e, c, a, and e, respectively, to the five questions. the summary e1 is thus the vector: 11 12 13 14 15( , , , , ) (0.25r r r r r = , 0.75, 0.25, 1, 0.75)− . in this case, the matrix in eq. (11) shows all of the options of the agents for the five questions. 11 12 13 14 15 21 22 23 24 25 31 32 33 34 35 41 42 43 44 45 51 52 53 54 55 0.25 0.75 0.25 1 0.75 1 0.75 0.5 0.25 0.75 (11)0.75 1 1 0.75 0.25 0.25 0.75 0.25 1 0.75 0 0.75 0.25 0.5 0.75 r r r r r r r r r r r r r r r r r r r r r r r r r r −⎛ ⎞ ⎛ ⎞ ⎜ ⎟ ⎜ ⎟−⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟= = − ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟ ⎝ ⎠⎝ ⎠ 6.3. processing summaries with the normal diffusion technique the elements in the first column of r in eq. (11) are the responses to “will global warming affect the typhoon risk in wenzhou? ” a comprehensive judgment can be given by the normal diffusion technique with the sample 1 11 21 31 41 51{ , , , , } {0.25, 1, 0.75, 0.25, 0}x r r r r r= = − where “ 21 1r = − ” means in summary e2, agent 2 did not answer question 1. the available set of answers to the questions is 1 11 31 41 51{ , , , } {0.25, 0.75, 0.25, 0}x r r r r′ = = using eq. (5), we obtain the diffusion coefficient 0.61095.0)-.750( 0.8146 ==h then, 11r = 0.25 can be transformed into a fuzzy set as shown in fig. 7. similarly, we transform other elements in 1x ′ into fuzzy sets. summing the four membership functions, we have the function 2 2 21.33955(0.25 ) 1.33955(0.75 ) 1.33955( ) 2 u u uf u e e e− − − − −= + + the absolute maximum of the function is at (0.2925, 3.642). therefore, the comprehensive judgment on question 1 is u1=0.292506. this value is closest to option c, namely, the answer is the following: global warming will affect the typhoon risk in wenzhou slightly. furthermore, the other columns lead to the following answers: • drainage facilities will be improved significantly. • in recent years, typhoon-related losses were slightly lower than gdp growth. • the risk typhoon will significantly decrease. • prepare to sell policies next year. 7. measuring the effective knowledge when we use a statistical model to analyze a relationship between two variables, the larger the sample is, the more accurate the result; and the better the statistical model is, the more reliable the result. similarly, the quality of service provided by an mioi is determined by its effective knowledge and its model m. the more effective knowledge is, the more credible the answer given by the ioi; and the better the model m is, the more reliable the answer. in an m-ioi, the set e of effective information shown in eq. (2) will be provided by agents using their experience and judgment. for e, we assume m agents provide m summaries e1, e2, …, em, which constitute a summary set s shown in eq. (12). 1 2{ , , , }ms e e e= (12) when m=n, we suggest the following procedure to measure the effective knowledge in an m-ioi. first, we assume that a customer will pay an amount of money g for a service using an ioi, and any agent will receive compensation g if he/she provides useful μ -2 -1 1 2 0.2 0.4 0.6 0.8 1 fig. 7. a comment quantified as r11=0.25 is transformed into a fuzzy set using the normal diffusion function. u ) 61095.02 )25.0( exp()( 2 2 11 × − −= u urμ published by atlantis press copyright: the authors 68 multiple internet of intelligences for risk analysis information, where g<g. that is, n=int(g/g) is the expected number of agents employed by the service at one time, where “int” means to round to the next lower integer. second, we assume that n agents provide useful information in answer to the customer’s question. the ratio of n to n, i.e., /n n , is called the basis of the effective knowledge, denoted as b. according to the previous assumption, the system will stop receiving information when n is equal to n; i.e., the system receives information while 0 1b≤ ≤ . any question asked by a customer must include some uncertainties. the task of the agents who provide information in an ioi is to reduce the uncertainties. the number of uncertainties in the question is denoted as w. if summary ei reduces w uncertainties, w is the dimension of ei. the average dimension of the summaries is w . the ratio of w to w is called the dimension of the effective knowledge, denoted as d; i.e., /d w w= , and 0 1d≤ ≤ the effective knowledge from an m-ioi can be measured using eq. (13), where c is called the measure of effective knowledge. the closer c is to 1, the more credible the answer given by the m-ioi. int( / ) n w n w c b d n w g g w = × = × = × (13) we can use eq. (13) to measure the effective knowledge of dynamic risk in the case of typhoons. we require the values of the variables in equation (13), i.e., n, g, g, w and w. in our case, the company will pay 6000 rmb for the service, and each agent will receive 1000 rmb. therefore, g=6000, g=1000, 5 agents provide effective information and n =5. “summarizing” is achieved by obtaining answers to the five questions in table 2 and reduces the five uncertainties with the effective information given by the agents. therefore, w=5. option “a” means the agent did not reduce the corresponding uncertainty. an “a” will reduce by one the dimension of ei. for example, because there is an “a” in e1, i.e., 14 1r = − , the dimension of summary e1 is w1=5-1=4. similarly, w2=4, w3=4, w4=5 and w5=5. the average dimension of the summaries e1, e2, …, e5 is 4 4 4 5 5 4.4 5 w + + + + = = the measure of effective knowledge is the following: 5 4.4 0.733 int( / ) int(6000 / 1000) 5 n w c q q w = × = × = the value 0.733 means the final judgment (shown at the end of section 7) on typhoon risk in wenzhou is not absolutely certain, but the judgment is nevertheless valuable for the insurance company when considering whether to insure clients in the region. 8. conclusions and discussion the new online tool, the internet of intelligences (ioi), answers questions from customers based on the experience, knowledge, and judgment of agents responding to the questions. unlike other question-andanswer systems such as google and quora, an ioi applies models to process the information from multiple agents to arrive at satisfactory answers. an ioi is not a social networking site but a brain trust. an ioi is defined as a triple <a, n, m>, denoted as f, consisting of a set a of agents, a network n used by a, and a model m to process information provided by a. when the intelligence level of f is greater than the highest intelligence level of the individuals in a, f is called a positive ioi, denoted as f↑. in practice, the information in an ioi might be heterogeneous, obtained from multiple sources. we define a multiple internet of intelligences (m-ioi) as the processing of homological information in layers, which allows the use of simpler models. in a summarizing mioi, the agents are required to both provide and process the information. the small samples from summaries are optimally processed using the normal diffusion model to generate an integrated answer that is more robust. an example of typhoon-related insurance risk shows that an m-ioi can predict for the customer, an insurance company, whether the level of risk will increase significantly. the methods reported in this study were applied to an ioi assessing the demand for typhoon insurance from the aquaculture industry in wenzhou, china. readers interested in this online system may refer to http://www.cazl.cn/ioi4ndra. the internet was based on the idea that there would be multiple independent networks of rather arbitrary design. as we now know, the internet embodies a key underlying technical idea, namely that of open published by atlantis press copyright: the authors 69 c.f. huang architecture networking. the predecessor to the internet, the arpanet, which was developed in the early 1970s by the u.s. department of defense (tanenbaum, 1996), introduced several major innovations: email, the electronic text messages exchanged by users; a remote connection service for controlling a computer; and the file transfer protocol (ftp), which allows information to be sent from one computer to another in bulk. in the early 1990s, the world wide web (www) was created based on the internet, a large and global tcp/ip network. arpanet provided the starting point for organizations to connect their computers. the web allowed the enormous growth of all types of "people-topeople" traffic. the evolution of the internet is leading to a digital translation of the physical world, enabling the latter to be permanently connected. this is what is called the internet of things. the next step is the internet of intelligences. from the arpanet to iois, the development of internet routing is shown in fig. 8. as internet speeds increase, it will be possible to share intelligences on a platform. the platform will be a part of the global brain, defined as a network that is an immensely complex, self-organizing system. a global brain not only processes information but can also play the role of a brain: making decisions, solving problems, learning new connections, and discovering new ideas. no individual, organization or machine is in control of this system; its knowledge and intelligence are distributed over all of its components (heylighen, 2011). a global brain may be able to solve current and emerging global problems that have eluded more traditional approaches. at the same time, it will create new technological and social challenges that are as yet difficult to imagine. an ioi, as a part of a global brain, might offer a viable platform to solve extremely complex problems online through intelligent agents. a global brain supported by many iois is shown in fig. 9. there is no doubt that existing mathematical tools will not be sufficient to describe the future global brain. more powerful mathematical tools beyond differential calculus, probability theory and fractal theory will be required. fuzzy logic (zadeh, 1965, 1975, 1996) could be developed to support utilities to process flexible information expressed in natural language. in the future, a global brain could provide timely, effective and lowcost analysis of risks, making the world safer. references f. l. ai, a research on building and applying the risk analysis of natural disaster of internet of intelligences service platform, phd thesis, beijing normal university, beijing, 2014. e. badilescu-buga, knowledge behaviour and social adoption of innovation, information processing and management 49(4) (2013) 902–911. d. ferrucci, a. levas, s. bagchi, d. gondek and e. mueller, watson: beyond jeopardy! artificial intelligence (199– 200) (2013) 93–105. l. gannes, how quora is trying to build an ideal society (2010), http:// gigaom.com/2010/07/26/how-quora-istrying-to-build-an-ideal-society/. i. habernal, m. konopik, swsnl: semantic web search using natural language, expert systems with applications 40(9) (2013) 3649-3664. f. heylighen, conceptions of a global brain: an historical review, in evolution: cosmic, biological, and social, eds. c1 c2 c3 a1 a2 a3 s(network server) m(information processing model) c1 c2 c3 a1 a2 a3 s(network server) m(information processing model) c1 c2 c3 a1 a2 a3 s(network server) m(information processing model) c1 c2 c3 a1 a2 a3 s(network server) m(information processing model) fig. 9. in the future, a global brain would be supported by many internets of intelligences to solve extremely complex problems online using intelligent agents, where each person plays a role as a brain cell. internet of individuals social tech web 2.0 internet of things pervasive computing web 3.0 internet of intelligences augmented thinking web 4.0 internet of organizations ecommerce web 1.0 fig. 8. the development of internet routing: from web 1.0 to web 4.0. an internet of intelligences is the future of the internet. currently, we are in the process of transition from web 2.0 to web 3.0. published by atlantis press copyright: the authors 70 multiple internet of intelligences for risk analysis l. e. grinin, r. l. carneiro, a. v. korotayev and f. spier (uchitel publishing, volgograd, russia, 2011) pp.274-289. c. f. huang, principle of information diffusion, fuzzy sets and systems 91(1) (1997) 69–90. c. f. huang, information diffusion techniques and small sample problem, international journal of information technology and decision making 1(2) (2002) 229-249. c. f. huang, internet of intelligences in risk analysis for online services, journal of risk analysis and crisis response 1(2) (2011) 110-117. c. f. huang, risk analysis and management of natural disaster (science press, beijing, 2012). (in chinese) c. f. huang and h. inoue, soft risk maps of natural disasters and their applications to decision-making, information sciences 177(7) (2007) 1583–1592. c. f. huang and c. moraga, a diffusion-neural-network for learning from small samples, international journal of approximate reasoning 35(2) (2004) 137–161. x. p. liu, j. q. zhang, w. y. cai and z. j. tong, information diffusion-based spatio-temporal risk analysis of grassland fire disaster in northern china, knowledge-based systems 23(1) (2010) 53-60. a. s. tanenbaum, computer networks (prentice hall, englewood, nj, 1996). l. a. zadeh, fuzzy sets, information and control 8(3) (1965) 338-353. l. a. zadeh, the concept of a linguistic variable and its application to approximate reasoning i, information sciences 8(3) (1975) 199-249. l. a. zadeh, fuzzy logic = computing with words, ieee trans. fuzzy systems 4(2) (1996) 103-111. j. h. zhang and l. j. zhao, risk analysis of dangerous chemicals transportation, systems engineering theory & practice 27(12) (2007) 117–122. q. zou, j. z. zhou, c. zhou, j. guo, w. p. deng, m. q. yang and l. liao, fuzzy risk analysis of flood disasters based on diffused-interior-outer-set model, expert systems with applications 39(6) (2012) 6213-6220. published by atlantis press copyright: the authors 71 microsoft word volume 12, issue 4-1 journal of risk analysis and crisis response, 2022, 12(4), 155-169 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i4.341 155 article digital finance, fiscal transparency and government debt risk hai-feng zha 1 and wei li 2,* 1 graduate school, anhui university of finance and economics, bengbu (233030), anhui, china 2 school of finance, anhui university of finance and economics, bengbu (233030), anhui, china * correspondence: liweiaufe@163.com received: october 25, 2022; accepted: december 28, 2022; published: december 31, 2022 abstract: the expansion of the breadth of coverage and depth of use of digital finance greatly improves the efficiency of financial services to the real economy, and has an important impact on the economic development of local governments in china. this paper takes the provincial local government debt risk data from 2011 to 2020 as the research sample to deeply investigate the effect and mechanism of digital finance on local government debt risk. the results show that, firstly, digital finance can effectively suppress local government debt risk, and a significant long-term suppressive effect is found by quantile model test. secondly, in the mechanism of action test, digital finance effectively alleviates local government debt risk through the path of local government fiscal transparency, and has played a significant regulating effect. finally, based on different regions and development levels, the effect of digital finance on local government debt risk is found to be more significant in eastern regions and regions with high urbanization levels. the findings of this paper provide some theoretical value and practical significance for preventing the occurrence of local government debt risk and regional systemic risk. keywords: digital finance; fiscal transparency; government debt risk; mediating effect; moderating effect 1. introduction while local governments have played an important role in china's economic development by leaps and bounds over the past 40 years since the reform and opening-up, they have also caused a rapid expansion in the scale of local government debt, which has accumulated hidden risks for local governments. as of the end of december 2021, china's local government debt balance was as high as 3,047 billion yuan, which poses a huge challenge to local government fiscal revenues and expenditures, easily causing an imbalance between government fiscal revenues and expenditures and further intensifying the scale of local government debt and risk accumulation. local governments are burdened with basic public service functions and economic development tasks, and the demand for funds for projects such as urban shantytown renovation, infrastructure construction and industrial structure upgrading forces, which make local governments to continue to expand the scale of debt. and debt risk gradually becomes an important resistance in the process of economic transformation. with the change of china's economy from high-speed growth to high-quality stage and the uncertainty of macroeconomic policy environment, the problem of local government debt risk has become increasingly prominent. the report of the 19th party congress even points out that hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 156 it is necessary to deepen the reform of the financial system and resolutely fight the hard battle to prevent and resolve major risks. therefore, the prevention and resolution of local government debt risk has become an important topic of research for many scholars. local governments must have their own reasons for raising debt, mainly in the following three aspects: first, to achieve local economic development. in order to achieve urbanization, industrialization and the important responsibilities of local economic development, local governments raise funds through participation in ppp projects, government public service purchases, etc. in addition to increased fiscal spending and other reasons, local governments continue to expand external channels to raise funds, the scale of government debt continues to expand, thus triggering a sharp rise in the risk of local government debt [1, 2]. second, it is influenced by the reform of fiscal decentralization system. there is a significant positive correlation between fiscal decentralization and local government debt risk. the reform of fiscal decentralization system has weakened the financial power functions of local governments, increased fiscal expenditures for transactional functions far beyond the government's own financial affordability, and the pressure of government budget constraints has further stimulated their need to achieve local development by raising debt [3, 4] . on the other hand, china's fiscal transfer payment system makes local governments believe that the central government will be willing to pay for their debts if they are unable to do so. at the same time, with the implicit guarantee of the central government, commercial banks and other financial institutions are willing to provide more financial support for local governments, increasing the enthusiasm of local governments in lending behavior [5]. third, the influence of the political system. due to the pressure of their own performance during their tenure, the stimulation of promotion mechanism and the competition of debt between governments, all stimulate the expansion of local government debt scale, laying hidden dangers for local government debt risks, which may eventually induce the occurrence of systemic financial risks [6–8]. with the development of internet technology, digital finance, a product of the combination of finance and technology, has come into being. digital finance has greatly subverted the traditional financial industry development model, relying on big data, blockchain and other technologies to open new paths for the realization of financial services to the real economy. on the one hand, it breaks the space limitation and reduces information asymmetry. unlike traditional offline financial services, digital finance breaks spatial constraints and uses big data technologies such as the internet to integrate customer information and promote direct transactions between the supply side and the demand side [9]. at the same time, with the support of internet technology, relevant information collection and communication barriers are effectively enhanced to alleviate the information asymmetry problem of traditional financial service models and improve risk identification and control capabilities [10, 11]. on the other hand, the efficiency of financial services to the real economy is improved and the financial environment is improved. the policy and targeting characteristics of digital finance have an important guiding role in promoting the "de-funding" of finance, optimizing the financial service environment and innovating financial products, which greatly improves the mode of obtaining financing through banks and other financial institutions, eases financing constraints. it is of great significance to improve the efficiency of financial services for the real economy and achieve optimal allocation of capital [12, 13]. in summary, it is important for local governments to ensure the circulation of local financial resources and maintain the normal operation of the local economy by means of indebtedness. hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 157 however, excessive government indebtedness may lay hidden dangers for the stability and development of local market economy and financial market. at the same time, the development of digital finance has an important role in improving the financial environment, solving the problem of "difficult and expensive financing" for enterprises, improving the efficiency of capital allocation and reducing transaction costs. as an important subject in the financial market, can the local government debt crisis be improved under the development of digital finance? therefore, this paper incorporates digital finance into the debt risk framework and tries to clarify the effects and mechanisms of digital finance on local government debt risk. this paper specifically expands on the following two aspects: first, it studies whether digital finance plays a significant role in local government debt risk from the perspective of digital finance, which has important academic value and practical significance for financial risk management and stabilization of financial markets. second, we dig deeper into the potential mechanism of action between digital finance and local government debt risk, which provides important solution ideas for local governments to alleviate financial imbalance and prevent debt risk. 2. theoretical analysis and research hypothesis the effective combination of internet technology and finance drives the rapid development of digital finance in china. due to the development and popularity of digital technologies such as cloud computing and mobile payment in china, the expansion of digital finance inclusiveness and practicality is promoted, which greatly boosts the development of digital finance. from the perspective of inclusiveness, digital finance combines technologies such as big data and artificial intelligence to establish a bridge between market players and investment players, and the low-cost access model can expand the breadth of digital finance coverage and help alleviate problems such as information transparency and asymmetry. from the practical point of view, the use of digital finance technology can accelerate the construction of the link between china's multi-level capital market, provide a more effective way for risk management, and achieve the goal of "de-focusing" financial services to the real economy. therefore, this paper deeply analyzes the effect and mechanism between digital finance and local government debt risk, which is of great practical significance to strengthen the combination of digital and finance and prevent financial risks. 2.1. digital finance and local government debt risk digital finance makes full use of technologies such as big data, blockchain and cloud computing to break the traditional risk regulation model and play an important role in potential risk identification and warning, the characteristics of digital finance are gradually coming to the fore. on the one hand, digital finance improves information asymmetry, promotes the increase of financial revenue and reduces the occurrence of risks, and the multi-scene applications it creates help to spread the depth of use and coverage of information through various channels, greatly reducing the phenomenon of information opacity, making any transaction "nowhere to hide" and "traceable". local governments do not need to rely on the traditional model for debt raising, and make full use of digital financial technology to scientifically design the price and maturity structure of local government bonds, promote local bond market reform, effectively reduce the risk of local government debt default, and stabilize the financial market order. at the same time, local governments can crack down on tax evasion and financial money laundering through digital hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 158 financial technology, which enhances the efficiency of local government fund allocation and the increase of local fiscal tax revenue, and reduces the probability of financing through external channels. in addition, local governments can solve the problems of "scale mismatch" and "maturity mismatch" through internet technology, provide scientific and reasonable arrangements for local government debt financing, maximize the efficiency of capital utilization, and reduce the possibility of debt risks. overall, local government debt risk has a negative guiding effect on the generation and outbreak of regional systemic risks, jeopardizing the stability of financial markets and sustainable and healthy economic development, and how to prevent and resolve local government debt risk has become an important dilemma for governments at all levels. in response to the emerged problems, digital finance provides feasible ideas for solving local government debt risks and promotes financial stability to a certain extent. based on the above research analysis, this paper proposes hypothesis one. h1: digital finance has a significant dampening effect on local government debt risk. 2.2. influence mechanism: improving financial transparency the cost effect, coverage effect and usage effect of digital finance is not only an important means to promote the high-quality development of china's economy, but also an important path to address the development of financial services for the real economy. in addition to promoting inclusive economic growth, the development of digital finance also has an important impact utility on fiscal behavior [14] and a significant differential promotion effect on local tax revenue, which can significantly increase the positive stimulation effect of digital finance on local tax revenue through coverage breadth. meanwhile, with the expansion of local government debt, understanding the financial situation of local governments has become an urgent task. in 2014, the state council pointed out that it is necessary to build a standardized and transparent budget system, prevent and resolve fiscal risks, and regularly disclose clear information about government bonds and project investments to the society. it is thus clear that local financial transparency also affects debt risk, and digital finance has significant advantages in addressing the information asymmetry between the government and the public, greatly alleviating the need to improve the barrier between the government and the public and greatly increasing government financial transparency through integration of government debt situation, investment situation, and financial flows. some scholars study local government management from the perspective of fiscal transparency and find that increased fiscal transparency significantly enhances information sensitivity, which has a debilitating effect on the size of local debt [15]. others believe that digital finance affects the scale and cost effect of local government financing through two ways: fiscal transparency and financial resource allocation efficiency [11] . based on this, this paper proposes hypothesis two. h2: digital finance can improve the financial transparency of local governments to curb the risk of local government debt. 3. study design 3.1. model construction 3.1.1. baseline regression model hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 159 based on the above theoretical analysis and research hypotheses, this paper constructs the baseline regression model:    5 2 ,,,10,lgd i yearprotjitjtj controldifrisk  (1) where lgdrisk , is the explanatory variable, which indicates the debt risk status of province or city j at time t. and dif , is the core explanatory variable, indicating the digital financial level of province or city j at time t; the control variables mainly include the level of economic development, industrial structure, urbanization level, fiscal deficit rate, urban unemployment rate, marketization index, and macroprudential regulation index. ε , is the random error term; τ , denotes controlling for individual spatial and time fixed effects. 3.1.2. quantile regression model to further investigate the differential impact of digital finance on local government debt risk, this paper constructs a panel quantile regression model for estimation tests in order to estimate the dynamic evolution characteristics of digital finance more accurately on government debt risk at different quantile points, with the following model settings: q,,,q )()](lg[ difqdifqdrisky tjtjtj  (2) in equation (2), the q , (dif) denotes the core influencing factors and other influencing factors of local government debt risk. y lgdrisk , q , (dif) denotes the values of local government debt risk at different quantile points q when other influencing factors are established. θ denotes the regression coefficients of the influenced explanatory variables at different quartiles q. also, in order to accurately estimate the regression coefficients θ that solves the minimization problem, the estimated model is shown in equation (3). qtjtj n difqdriskqtjtj n drisk difqdriskqdifqdriskq qtjtjqdiftjqtj   )(lg)1()(lgmin ,,)(lg,,lg ,,)(,,    (3) where n denotes the sample size and the meaning of the variable is consistent with the above. also, the different quantile points selected are 10%, 25%, 50%, 75% and 90% of the five quantile points to verify the differential evolutionary characteristics of digital finance on local government debt risk. 3.2. variable selection and measurement 3.2.1. explained variable: local government debt risk the explanatory variable in this paper is local government debt risk. and referring to the studies of related scholars [16] , the kmv model is used to measure the risk of local government debt in china. this paper uses local government revenue f replaces the value of enterprise assets, and local governments should repay their debts b , from which the default distance of local governments in china can be deduced. local government revenues f obeys a random distribution of f = f(x ) , when the local government is unable to pay its debts, i.e., f < b , then a default will result. at this point, the probability of default of the local government is: )]([])([p)( 1 tttttt bfxpbxfbfpp  (4) hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 160 when x ~n(0,1), equation (4) can become: )]([)]([ 11 ttt bfnbfxpp   (5) assume that local government revenues f obeys the stochastic process: df = gf dt + f dx , t = 0, f = f, then the local government revenue is: ]) 2 1 [(exp 2 tt xttgff   (6) tftgtffe t 22 )var(ln, 2 1 ln)ln(   (7) based on the above assumptions, the growth rate g and volatility of local government revenues σ can be measured:            1 1 1 1 211 )ln 1 1 (ln 2 1 n t n t t t t t f f nf f n  (8) 21 1 1 2 1 ln 1 1 g       n t t t f f n (9) according to equations (8) and (9), the local government default distance dd can be obtained. tt ttg b f dd t t      ))( 2 1 (ln 2 (10) in this paper, default distance is used as a proxy variable for local government debt risk, and the larger the value of default distance, the lower the probability of default risk for local governments. the default risk of local governments in china is measured according to equation (10). 3.2.2. core explanatory variables: digital finance this paper adopts the digital inclusive finance index compiled by the digital finance research center of peking university as a proxy variable for digital finance. this index includes three firstlevel indices in addition to the total index, which are breadth of coverage, depth of use, and degree of digital support services. therefore, this paper mainly selects the total digital finance index, the breadth of coverage and the depth of use for the empirical study of this paper. 3.2.3. mediating variable: fiscal transparency the development of digital finance strongly improves the ability of information collection and processing, greatly enhances the transparency of local government finance, and has an important impact on the prevention and control of local government debt default. therefore, it is important to study the mechanism of the role between digital finance and local government debt risk. in this paper, we select tsinghua university's research report on financial transparency of chinese urban governments to measure the financial transparency status of each provincial local government, and process it by weighting and other methods as the final proxy variable. 3.2.4. control variables hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 161 1) level of economic development: the level of economic development of a region will have an important impact on the default status of government debt in that region, and this paper uses gdp per capita to measure the level of economic development of local governments. 2) industrial structure: the main purpose of local governments to raise funds is to achieve long-term development of the local economy, and the shape of regional industrial structure also has a stronger impact on the scale of local government debt and the probability of default. therefore, this paper measures the industrial structure status by the share of secondary industry in gdp. 3) level of urbanization: the level of urbanization of local governments can expand the size of government debt, which may exacerbate the probability of government default. therefore, this paper measures the level of urbanization of a place by the percentage of urban population. 4) fiscal deficit ratio: the fiscal deficit ratio is an accurate reflection of the local government debt situation, and to a certain extent, measures the local governments fiscal revenue and expenditure gap. therefore, this paper reflects the fiscal deficit rate of local governments by the measure of fiscal expenditure minus fiscal revenue divided by fiscal revenue. 5) urban unemployment rate: the proportion of urban unemployment to the sum of urban employment and unemployment is selected to measure. 6) marketization index: the index was selected according to the china marketization index report based on the measurement of their environmental system. 7) macro-prudential regulation: as local governments obtain funds through financing platforms and bond issuance to promote local economic development, the central government may take corresponding regulatory measures to prevent potential default risks. therefore, this paper refers to relevant studies [17] and selects the macro-prudential supervision system of china published by the international monetary fund to measure the macro-prudential supervision index. table 1. results of descriptive statistics for each variable. variable name variable definition observations average value standard deviation minimum value maximum value lgdrisk local government debt risk 310 3.295 3.998 -7.447 14.812 dif digital finance 310 5.219 0.668 2.909 6.068 fsp financial transparency 310 3.603 0.433 2.639 4.573 edl economic development level 310 9.784 0.884 7.223 11.619 est industry structure 310 0.413 0.083 0.158 0.620 ubl level of urbanization 310 0.588 0.124 0.350 0.938 czl fiscal deficit rate 310 1.536 1.699 -0.602 10.800 syl urban unemployment rate 310 3.255 0.637 1.210 4.610 sch marketability index 310 7.941 1.892 3.359 11.934 regu macroprudential regulation 310 3.800 1.539 2.000 7.000 hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 162 3.3. data sources and descriptive statistics this paper focuses on the relationship between digital finance and local government debt risk, using sample data from 31 provinces and cities from 2011 to 2020, with data mainly from the china statistical yearbook and wind database. the results of descriptive statistics of each variable in this paper are shown in table 1. 4. empirical results and analysis 4.1. benchmark test results 4.1.1. baseline regression effect and dynamic effect table 2. estimated results of baseline regression and dynamic effects of digital finance affecting local government debt risk. variables lgdrisk: government debt risk baseline return dynamic effects lagging 1 period lagging 2 periods lagging 3 periods (1) (2) (3) (4) dif -1.036 *** (-3.14) l1.dif -1.131*** (-3.58) l2.dif -1.118 * (-2.95) l3.dif -1.494*** (-4.33) edl -3.494*** (-3.41) -2.199** (-1.99) -2.048* (-1.89) 0.301 (0.22) est 3.217*** (6.00) 2.145*** (5.06) 1.811*** (3.94) 1.097*** (2.95) ubl -0.469** (-2.01) -2.138*** (-3.33) -4.910*** (-4.70) -0.575 (-1.07) czl 0.382** (2.24) 0.357** (2.16) 0.336** (2.06) 0.932*** (4.30) syl 0.870*** (4.13) 0.913*** (4.17) 0.885*** (3.86) 0.714*** (2.83) sch -0.346** (-2.10) -0.212* (-1.66) -0.179 (-1.60) -0.163 (-1.55) regu -0.152* (-1.90) -0.166** (-2.07) -0.292*** (-3.09) -0.101 (-1.01) constant 8.702*** (7.89) 7.140*** (6.94) 3.069 (1.30) -5.6748 (-1.05) time/area fixed effects y y y y n 310 270 240 210 adjusted r2 0.0581 0.0473 0.2792 0.3299 note: t-values are in parentheses, and "***, **, **" indicate significance levels at 1%, 5%, and 10%, respectively. table 2 mainly shows the results of the baseline regression and dynamic effect estimation for the impact of digital finance on local government debt risk. column (1) of table 2 shows the results of the benchmark regression, which shows that the estimated coefficient of digital finance is -1.036 and passes the 1% significance test. this result indicates that digital finance has a significant inhibitory hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 163 effect on local government debt risk, and hypothesis 1 is initially tested. on the one hand, digital finance breaks the traditional financing model of local governments, promotes the development of diversity of financing methods and channels, realizes real-time monitoring of local government fund flows through technical means, reduces the possibility of local governments hiding negative news, releases favorable information to the capital market and investment subjects, and helps reduce the possibility of local government debt risk. on the other hand, the application of financial technology can help local governments innovate financing tools and rationalize the scale and maturity structure of local bonds, mitigate the consequences caused by the imbalance of local government finance, and realize the optimal allocation of resources. is the negative shock effect of digital finance on local government debt risk a short-term choice or a long-term driver? this paper further investigates the variability of the long-term and short-term effects of digital finance on local government debt risk by referring to the dynamic effects test, and the estimated results are shown in columns (2) to (4) of table 2. it is found that the estimated coefficients of digital finance pass the 1% significance test, which indicates that the inhibitory effect of digital finance is not a "fleeting" effect, but has a significant long-term effect, which also supports the correctness of hypothesis 1. regarding the estimation results of the control variables, the level of economic development and local government debt risk show a negative effect, and the improvement of economic development level contributes to the control of local government debt risks. while the change in industrial structure promotes the increase in local government debt risk, which is mainly due to the fact that the upgrading and transformation of economic structure imposes a huge burden on local finances and requires external funding to achieve this goal, causing the expansion of debt scale and further stimulating the accumulation of debt risk. there is a significant positive effect between urban unemployment rate and government debt risk, and the high unemployment rate increases the government's expenditure on livelihood protection, which increases the financial pressure on local governments. meanwhile, the level of urbanization, the level of marketization and external macro regulation play a negative impact on debt risk, which fully indicates that the improvement of urbanization level, marketization level and macro prudential regulation are beneficial to the suppression of local government debt risk. 4.1.2. quantile model test table 3 mainly examines the impact effect of digital finance at different quartiles, and found that the inhibitory effect of digital finance did not find significant variation, and this inhibitory effect showed a trend of strengthening. the explanation for this is that, on the one hand, the development of digital finance promotes the economic growth and marketization level of local governments, and the demand for funds is not too urgent, and it also reduces the financing cost and suppresses the government debt risk. on the other hand, the development of digital finance greatly enhances local government information disclosure, improves information liquidity and accuracy, reduces the cost of information access for the public, and alleviates information asymmetry, which ultimately affects local government debt risk. therefore, hypothesis 1 is further verified. hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 164 table 3. results of the quantile test for the impact of digital finance on government debt risk. variables lgdrisk: government debt risk q (1) q=0.1 (2) q=0.25 (3) q=0.5 (4) q=0.75 (5) q=0.9 dif -2.501*** (-3.40) -2.857** (-4.98) -3.535*** (-7.79) -4.517*** (-7.98) -5.672*** (-8.46) edl -1.383** (-2.33) -1.349** (-2.08) -1.430** (-2.39) 0.547 (1.20) 0.584 (1.08) est 2.460** (2.12) 2.015* (1.92) 1.356* (1.78) -8.774** (-2.32) -9.640** (-2.15) ubl -10.970*** (-2.66) -12.235*** (-3.81) -16.618*** (-6.54) -11.622*** (-3.67) -6.826* (-1.82) czl 0.618*** (2.47) 0.164 (0.84) 0.242* (1.67) 0.131 (0.68) 0.307** (1.99) syl 1.710*** (4.31) 1.619*** (5.24) 1.922*** (7.87) 1.779*** (5.84) 1.524*** (4.22) sch -2.137*** (-4.05) -0.455 (-1.11) -0.568* (-1.75) -0.716* (-1.77) -0.889* (-1.85) regu -0.218** (-1.86) -0.392** (-1.98) -0.368** (-2.36) -0.357* (-1.83) -0.445* (-1.93) constant 4.337*** (3.79) 5.248*** (3.05) 6.133*** (5.60) 5.417*** (4.75) 5.217*** (5.35) time/area fixed effects yes yes yes yes yes n 310 310 310 310 310 adjusted r2 0.3892 0.3492 0.3375 0.3682 0.3908 note: t-values are in parentheses, and "***, **, **" indicate significance levels at 1%, 5%, and 10%, respectively. 4.2. robustness test and endogeneity treatment 4.2.1. robustness test in this paper, we downscaled the total digital finance index and used digital finance coverage breadth and depth of use to replace the total digital finance index for robustness testing, and the specific estimation results are shown in columns (1) and (2) of table 4. it can be found that the robustness test estimation results are basically consistent with the baseline regression estimation results, and digital finance still has a significant inhibitory effect on local government debt risk, which fully indicates that the research findings of this paper are robust and reliable. 4.2.2. endogenous processing in order to better deal with the possible endogeneity problem, this paper applies the systematic gmm method and instrumental variables method to further verify the robustness of the estimation results, which are shown in table 4. in the systematic gmm test, the significance and direction of the effects of digital finance and local government debt risk do not change significantly. in the instrumental variable method test, this paper selects the number of internet broadband access users as an instrumental variable, mainly because the number of internet broadband access users reflects the degree of using internet technology to obtain information and the popularity of internet technology from the side, which has an important influence on the development of digital finance, but there is a weak connection or even unrelated, satisfying the condition of exogeneity of instrumental variables. it can be found that in column (4), the number of internet broadband access users has a significant effect on digital finance, and the f-value in the first stage is 37.31, which is hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 165 greater than 10, and the instrumental variables in this paper are valid. in the second stage estimation results in column (5), the estimated coefficient of digital finance is consistent with the baseline regression estimation results, and there is no significant variation, and it passes the 5% significance verification. therefore, the regression estimation results of this paper are robust. table 4. robustness test and endogeneity treatment empirical results. variables robustness tests endogenous processing digital finance index downscaling system gmm method instrumental variables method (1) breadth lgdrisk (2) depth lgdrisk (3) lgdrisk (4) first stage dif (5) second stage lgdrisk iba 0.381*** (4.33) dif -0.658** (-2.41) -0.941*** (-2.84) -2.003** (-2.41) -1.407*** (-3.15) edl -3.952*** (-4.09) -3.698*** (-3.79) 0.180 (0.49) 1.002*** (6.58) -0.133 (0.37) est 2.164*** (6.85) 2.036*** (6.26) 2.265*** (6.32) -3.703*** (-6.76) 2.811** (5.07) ubl -0.934** (-2.16) -0.448* (-1.88) 0.979 (1.25) -0.072 (-0.07) -0.929*** (-3.95) czl -0.355 (-1.15) -0.325 (-1.07) 0.085 (0.82) -0.159*** (-3.42) 0.067 (0.48) syl 0.868*** (4.11) 0.876*** (4.16) 0.762** (2.02) -0.124*** (-2.62) 0.834*** (2.46) sch -0.353** (-2.12) -0.309** (-2.00) -0.329*** (-2.95) 0.045 (1.39) 2.179*** (9.62) regu -0.120 (-1.59) -0.104 (-1.50) -0.220*** (-2.69) -0.121*** (-12.45) -0.941*** (-4.81) constant 29.607*** (3.76) 29.815*** (3.85) 7.802 (1.09) -4.615*** (-3.88) 33.084*** (6.71) time/area fixed effects yes yes yes yes yes n 310 310 310 310 310 adjusted r2 0.0303 0.0344 --0.8940 0.6940 phase 1 f-value 37.31 cragg-donald wald values 39.33 note: t-values are in parentheses, and "***, **, **" indicate significance levels at 1%, 5%, and 10%, respectively. in the systematic gmm estimation, ar(1) is 0.015, ar(2) is 0.303, and sargan and hansen are 0.327 and 0.228, respectively. in the instrumental variables test, the f-value in the first stage is 37.31, and in the second stage the kleibergen-paap rk lm statistic and hansen j statistic had p-values of 0.000 and 0.432, respectively. 4.3. mechanism of action test after verifying the direct transmission mechanism between digital finance and local government debt risk, it is necessary to consider through what channels digital finance may affect local government debt risk? according to the theoretical analysis and research hypothesis of this paper, this paper takes fiscal transparency as an important intermediate variable to test the effect of the mechanism of digital finance affecting local government debt risk, and the specific econometric model is shown in equations (11) and (12). hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 166 yearprotjitjtj controldiffsp ,.,10,    (11)    6 3 ,,tj,2,10, fsplgd i yearprotjitjtj controldifrisk  (12) table 5 mainly describes the mechanism of the effect of digital finance on the risk of local government debt. in the test of mediating effect, the effect of influence between digital finance and fiscal transparency is mainly verified, and the results show that the regression coefficient of digital finance is 0.728, which passes the 1% significance verification. it indicates that digital finance has a significant positive influence on fiscal transparency, and digital finance helps to improve the fiscal transparency of local governments, which means that fiscal transparency is an important digital finance influence on local government debt risk mediating factor, and hypothesis 2 was initially verified. table 5. estimation results of the test of the mechanism of action of digital finance affecting local government debt risk. variables intermediary effects moderating effects (1) fsp (2) lgdrisk (3) lgdrisk dif 0.327*** (6.53) -1.910* (-1.71) -2.656*** (-1.88) fsp -0.287*** (-2.86) -0.674*** (-3.11) dif×fsp 0.553** (2.2) edl 0.123** (2.02) -1.025* (-1.87) -4.063*** (-3.65) est -2.125*** (-4.16) 16.743*** (4.64) 26.069*** (5.74) ubl -0.326 (-0.78) -6.323 (-1.54) -0.222 (-0.04) czl 0.031 (1.25) -0.716*** (-3.44) -0.333 (-1.03) syl 0.230*** (2.72) 1.061*** (5.24) 0.878** (4.12) sch 0.073 (1.61) -0.637** (-2.15) -0.320 (-1.02) regu -0.235** (-2.44) -0.221*** (-2.74) -0.157* (-1.96) constant 1.162** (1.99) 2.720*** (3.39) 2.577*** (3.21) time/area fixed effects yes yes yes n 310 310 310 adjustedr 0.0623 0.3206 0.0394 note: t-values are in parentheses, and "***, **, **" indicate significance levels at 1%, 5%, and 10%, respectively. according to the mediating effect model action, if ω insignificant and ω is significant, it indicates that digital finance affects local government debt risk exclusively through the pathway of fiscal transparency; if ω and ω both pass the significance test, it implies that fiscal transparency plays only a partial mediating effect. therefore, from column (2) of table 5, it can be found that the regression coefficients of both digital finance and fiscal transparency pass the significance verification, hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 167 which also indicates that fiscal transparency plays an important partial mediating effect in the process of digital finance affecting local government debt risk, and hypothesis 2 is fully verified. after considering the mediating effect, it is also important to consider whether fiscal transparency plays a moderating effect is also a key concern of the mechanism of action test. therefore, the interaction term of digital finance and fiscal transparency is added to the benchmark model, and the specific test results are shown in column (3) of table 5. it can be found that fiscal transparency has a significant moderating effect on digital finance to suppress local government debt risk, and fiscal transparency further releases the suppressive effect of digital finance on local government debt risk, and hypothesis 2 is further verified from the perspective of moderating effect. 4.4. heterogeneity test table 6. heterogeneity test results. variables lgdrisk: government debt risk regional heterogeneity developmental level heterogeneity (1) eastern (2) midwest (3) high level of urbanization (4) low level of urbanization dif -1.778** (-2.38) -0.730 (-0.53) -2.032*** (-2.91) -0.875 (-1.05) edl -3.559** (-2.31) -2.663 (-1.01) -0.601 (-0.74) -7.111*** (-3.22) est 4.585*** (3.48) 4.017*** (2.89) 4.447*** (2.78) 2.649 (1.29) ubl -1.220* (-1.93) 1.150 (0.17) -1.690*** (-2.88) -0.561 (-1.17) czl 0.833** (2.32) -0.427* (-1.82) 1.652*** (3.29) 0.115 (0.35) syl 0.697** (2.15) 0.593* (2.16) 0.486 (1.48) 1.230*** (4.05) sch -1.561** (2.55) 0.447 (1.01) -0.085 (-0.18) -0.873 ** (-2.17) regu -0.228* (-1.88) 1.127 (1.15) -0.269** (-2.29) -0.161* (-1.64) constant 2.200*** (3.35) -4.898 (-0.21) 2.975*** (4.13) 2.649*** (3.51) time/area fixed effects yes yes yes yes n 110 200 140 170 adjusted r2 0.0028 0.5470 0.0209 0.2785 note: t-values are in parentheses, and "***, **, **" indicate significance levels at 1%, 5%, and 10%, respectively. after the baseline regression and quantile regression tests, this paper divides the research sample according to regions and development levels to further investigate whether the effect of digital finance on local government debt risk varies according to regional differences and development levels, and the results are shown in table 6. in the regional heterogeneity test, it can be found that the baseline regression coefficient of digital finance passes the 1% significance test in the eastern region and has a significant inhibitory effect on local government debt risk, while it is not significant in the central and western regions. the explanation for this conclusion: digital finance is a combination of technology and finance, which has high requirements for technology level and environment. for the eastern region, with many high-tech enterprises and technical talents and financial support, the development level of digital finance is significantly higher than that of the hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 168 central and western regions, which has a significant role in promoting the application and development of digital finance and has a more significant impact on local government debt risk. the impact on local government debt risk is more significant. for the central and western regions, there is an obvious gap between the level of economic development and the eastern region, and more consideration is given to improving the regional economic level, which is not conducive to the development of digital finance. therefore, digital finance in the developed eastern region is more likely to inhibit the expansion of local government debt risk compared to the central and western regions. in the test of development level heterogeneity, it can also be found that the estimated coefficients of digital finance at different urbanization levels are -1.371 and -0.141, respectively. however, the negative effect of digital finance on local government debt risk is more significant in regions with high urbanization levels. compared with regions with low urbanization levels, regions with high urbanization levels consider more about achieving high-quality economic development and pay more attention to the application and development of high technology levels, which creates a favorable atmosphere for the development of digital finance. 5. conclusions along with the development of modern information technology in china, digital finance breaks through the traditional financial industry confinement and has an important role in china's macro economy and micro subjects. as a product of finance and technology, digital finance provides more efficient and convenient services for local governments' capital circulation, and offers a variety of possibilities in terms of financing subjects and path choices. of course, the development of digital finance also greatly creates a favorable atmosphere for economic structural transformation and the construction of a double-cycle economic development pattern. on the other hand, the development of digital finance also has an important impact on the improvement of local government governance, effectively grasping the initiative of risk occurrence, balancing the relationship between government debt risk and economic development, and providing an important guarantee for promoting highquality sustainable and healthy economic development. according to the relevant theoretical analysis, this paper conducts an in-depth study on digital finance and local government debt risk, and empirically examines the effect and mechanism of digital finance on local government debt risk using national provincial panel data from 2011 to 2020 as the research sample, and the research results show that: first, digital finance has a significant inhibitory effect on local government debt risk. as the risk of local government debt rises, the inhibitory effect of digital finance presents long-term characteristics. second, the results of intermediary and moderating effects show that fiscal transparency is an effective way for digital finance to reduce the risk of local government debt. fiscal transparency can further release the negative impact effect of digital finance on local government debt risk. third, after distinguishing the geographical and urbanization levels, the inhibitory effect of digital finance on local government debt risk is more significant in eastern regions and regions with high urbanization levels. funding: this research was funded by the philosophy and social science planning project of anhui province (national social science fund incubation project), grant number ahskf2021d07. hai-feng zha and wei li / journal of risk analysis and crisis response, 2022, 12(4), 155-169 doi: https://doi.org/10.54560/jracr.v12i4.341 169 conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] wu j, wu y, wang b. local government debt, factor misallocation and regional economic performance in china. china world econ. 2018; 26(4):82-105. doi: https://doi.org/10.1111/cwe.12250. [2] xiang h, wu j, xie j. does local debt affect economic fluctuation. china ind econ. published online 2017:1-14. doi: https://doi.org/10.19581/j.cnki.ciejournal.2017.01.003. [3] fan g, lv y. fiscal prudence and growth sustainability: an analysis of china’s public debts. asian econ policy rev. 2012; 7(2):202-220. doi: https://doi.org/10.1111/j.1748-3131.2012.01234.x. [4] zou f. analysis and research on local government debt problems in china. mod econ. 2019; 10(01):188197. doi: https://doi.org/10.4236/me.2019.101013. [5] ouyang ay, li r. fiscal decentralization and the default risk of chinese local government debts. contemp econ policy. 2021; 39(3):641-667. doi: https://doi.org/10.1111/coep.12531. [6] zhang l, han r, zhang j, li l, zhang d. land-leasing behavior, local officials’ promotions, and chinese cities’ debt risks. int j strateg prop manag. 2021; 25(6):485-496. doi: https://doi.org/10.3846/ijspm.2021.15654. [7] yu c, hou l, lyu y, zhang q. political competition, spatial interactions, and default risk of local government debts in china. pap reg sci. 2022; 101(3):717-743. doi: https://doi.org/10.1111/pirs.12668. [8] huang s. the unintended consequence of local government debt: evidence from stock price crash risk. financ res lett. 2022; 50(july):103296. doi: https://doi.org/10.1016/j.frl.2022.103296. [9] huang h. the formation and challenges of digital financial ecosystem -experience from china. economist. 2018; 4:80-85. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2018.04.011. [10] wang j, zhu w. can the development of digital finance correct the ineffecient investment of enterprises. financ econ. 2020; (03):20-31. [11] hou s, song l. the impact of digital finance on local government debt financing. public financ res. 2020; (09):52-64. doi: https://doi.org/10.19477/j.cnki.11-1077/f.2020.09.005. [12] yanping p, xiao c. financial development, marketization and capital allocation efficiency of the service industry. 2019; (06):43-52. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2014.06.001. [13] yanan w, xin y, lin x. can digital finance boost the real economy. 2020; (03):1-13. [14] zhang x, wang g, zhang j, he z. digital economy, financial inclusion, and inclusive growth. econ res j. 2019; 25(8):8-11. [15] zhang p, ma w. local public debt management from the perspective of fiscal transparency. economist. published online 2020:80-89. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2020.08.009. [16] zhang y, shi b. non-tradable shares pricing and optimal default point based on hybrid kmv models: evidence from china. knowledge-based syst. 2016; 110:202-209. doi: https://doi.org/10.1016/j.knosys.2016.07.028. [17] ruan s, zha h, li w, chen x. double pillar regulation and systemic financial risk. econ probl. published online 2020:33-40. doi: https://doi.org/10.16011/j.cnki.jjwt.2020.11.005. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style an empirical research on the relationship between property insurance premiums and macroeconomic variables based on ardl model* guiyun you school of insurance and economics, university of international business and economics beijing 100029, china e-mail: youguiyun0330@126.com shanshan cao, jing feng, shu yu school of economics, ocean university of china qingdao 266100, china e-mail: caoshanshan1106@163.com smsrfj@126.com ys710639470@126.com abstract given that most of property insurance policies are one-year contracts and have a high renewal, this paper establishes auto-regressive distributed lag model (ardl) which considers adding lags of the dependent variable and/or lags of some independent variables. based on the data of insurance premiums in china, gross domestic product (gdp), consumer price index (cpi) and fixed-asset investment during the period from 1980 to 2012, this paper analyzes the long-term and short-term relationships between them with method of the ardl bounds testing approach. the results indicate that gdp is the major factor driving the growth of property insurance premiums in china; fixed-asset investment has significant impact on chinese property insurance premiums, and they show the conspicuous negative correlation; moreover, cpi has little effect on the premium income. keywords: insurance premiums, macroeconomic variables, ardl model *project supported by chinese ministry of education(11yja790193). 1. introduction in this paper, data of insurance premiums in china, gross domestic product (gdp), consumer price index (cpi) and fixed-asset investment during the period from 1980 to 2012 are utilized in order to establish ardl to analyze the long-term and short-term relationships between them. the outcome indicates that gdp is the factor which influences property insurance mostly. numerous scholars consider that economic growth is the main factor affecting premiums. macroeconomic variables, which include gdp, cpi as well as fixedasset investment, may reflect the economic development to some extent. moreover, several foreign scholars studied the relationships between premiums and some macroeconomic variables. specifically, outreville (1990) [1] analyzed the panel data from 55 developing countries. the result indicated that the economic growth had a positive impact on property insurance premiums. journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 175-181 published by atlantis press copyright: the authors 175 mailto:lvfangzhao@hotmail.com willieb typewritten text received 12 june 2014 willieb typewritten text accepted 12 august 2014 willieb typewritten text ward and zurbruegg (2000) [2] studied insurance market of 9 member states in the organization for economic cooperation and development. however, the results of their research were slightly different from outreville’s(1990). in their view, the relationship between economic growth and the development of insurance industry varied among countries. in details, they believed that the improvement of insurance industry could promote the economic growth in some countries; however, the opposite happens in others. besides, the research by .kugler and ofoghi’s (2005) [3] was different from outreville’s(1990) as well as ward and zurbruegg’s (2000). in their research, aggregate data was replaced by component metrics data to study the relationship between insurance industry and economic growth in uk. arena (2008) [4] also differed from the previous scholars in method of research. he analyzed the data of 56 countries from 1976 to 2004 with the method of gmm estimation model of dynamic panel, and then he concluded that there existed two-way causality relationship between premium and gdp. differing from the scholars mentioned above, after reviewing 85 papers related to the relationship between development of insurance industry and economic growth, outreville (2011) [5]pointed out that the insurance demand was affected by exogenous variables. as can be seen from the above-mentioned foreign scholars’ researches, the regression analysis was widely adopted and gdp was the only economic variable selected which affects the insurance industry. chinese scholars also carried out some researches on the relationship between premiums and macroeconomy. specially, linear regression was utilized by some researchers. for example, lin baoqing, hong xixi and wu jiangming (2004)[6] thought that economic growth may has diverse effects on premium income in different insurance market structure. they selected twostage inter-provincial panel data to study the development of insurance industry in china during the two-stage period which were oligopoly highly monopolistic stage from 1988 to 1992 and monopolistic competition stage from 1997 to 2002 respectively. the results showed that there existed highly stable positive relationship between variables no matter which stage we were in. however, to some extent, the coefficients may be different. the former study mainly analyzed the relationship between economic development and insurance industry in different time periods. while wu xiangyou (2009) [7] established a multiple linear regression model using the data of 30 provinces in china from the period of 1997 to 2007. he found that the economic development was the fundamental driving force of the advance of insurance industry. moreover, the distinctions in the level of economic growth had led to deviations in the level of advance in insurance industry. be similar to lin baoqing (2004) and wu xiangyou (2009), zhao hongmei and su huijuan (2013)[8] also used inter-provincial panel data. nevertheless, their study was based on the developing individual (time) fixed effect model. the different intercept terms of the model were used as the base of calculating the weights. as a result, the panel data was adjusted according to the regional weight as well as time weight. after that, they used the adjusted panel data to finish the stepwise regression, and then they analyzed the influencing factors for the demand of property insurance in china. the results showed that the influencing factors for the demand of property insurance contained disposable income per person, fixed-asset investment and population considering the regional differences. however, the influencing factors became fixed-asset investment in consideration of the time differences. some scholars, including qian zhen (2008)[9] and zhang chunhai (2010)[10], used vector auto-regression (var) model. qian added impulse response function to var model, and she selected more indexes, which included gdp, urban and rural resident savings as well as disposable income per person, to explore the relationships among economic growth, savings, consumption and insurance development. eventually she found that there existed dynamic co-integrated relationship. besides, zhang selected time series data of gdp and premiums which covered the period from 1985 to 2009 to analyze the correlative mechanism of economic growth and premium with var model. they believed that gdp growth played an important role on the increase of premium. other scholars established a combinational model. for example, rao weinan (2011)[11] analyzed the relationship between premium and economic development of china in terms of the insurance penetration. moreover, he constructed a portfolio mode on the base of quarterly data, and ultimately he achieved that the elastic coefficient of the national premium income gross to gdp was 1.102[11]. published by atlantis press copyright: the authors 176 without considering the autocorrelation or hysteresis quality of dependent variables, linear regression and var were used as the main research methods in the previous studies. in this paper, considering the characteristics of the property insurance that the policies are one-year period and high renewal, the model of ardl, which adds lags of the dependent variable and/or lags of some independent variables, is constructed. additionally, this paper carries out the bound test in co-integration test when time series data are not more than integrated of 1. index selection also differs from the previous studies. previous researches usually just took gdp as an indicator of economic growth. however, in order to comprehensively reflect macroeconomic conditions, this paper also adds cpi and fixed-asset investment for empirical analysis. 2. model and methods ardl is a relatively innovative dynamic co-integration test model, which was first used by charemza and deadman (1992) [12]. besides, this model was gradually improved by pesaran and smith (1998), pesaran and shin (1999) [13], pesaran (2001) [14] and so on. form of the model is 0 1 2 1 0 n n t i t i j t j t i j y y xβ β β ε− − = = = + + +∑ ∑ (1) when using dynamic test model to determine the long-term or short-term coefficients between the variables, the premise is the existence of co-integration relationship between variables. therefore, at first this paper applies bound test method to determine whether there is a co-integration relationship between the variables. there are two steps as followed: firstly, unconstrained error correction model (uecm) is constructed: 0 1 2 1 0 1 1 2 1 ty n n i t i j t j i j t t t y x y e ∆ α α ∆ α ∆ λ λ ε − − = = − − = + + + + + ∑ ∑ (2) among them, 0α is constant, i1α and j2α are shortterm dynamic coefficients, 1λ and 2λ are long-term dynamic coefficients, tε is white-noise process. afterwards, the optimal lag order of each differential item in formula (2) is determined respectively by the aic or sbc information criterion. and then its f statistics as well as joint significant test are carried out based on the following assumptions. 0 1 2 1 1 2 : 0 : 0 0 nul l hypot hesi s al t er nat i ve hypot hesi s or h h λ λ λ λ = =  ≠ ≠ since the asymptotic distribution of f statistic is nonstandard, pesaran (2001) has calculated the two different thresholds of f statistic corresponding different number regression item. if the calculated f statistic is larger than the upper bound of the threshold, the null hypothesis will be rejected. namely, there exist longterm relationships between variables. in additional, if the calculated f statistic is less than the lower bound of the threshold, the null hypothesis will be accepted. namely, there do not exist long-term relationships between variables. moreover, if the f statistic falls between two thresholds, the relationship between time series without unit root test cannot be judged. if all variables are integrated of 1, the null hypothesis will also be rejected. namely, there exist long-term relationships between variables. however, if all variables are integrated of 0, the null hypothesis will be accepted. namely, there is not exist long-term relationships between variables. secondly, on the premise of the existence of a longterm relationship, this paper applies ardl model to estimate the long-term relationship between coefficients of variables. compared with traditional eg two-step method and johansen method, ardl bounds testing method has the following advantages: firstly, regardless of the time series are i(1) or i(0) at the same time, this method can be used to test the long-term relationships between the variables. secondly, this method is the most efficient co-integration technique when it comes to smaller sample data. further, this approach corrects for possible endogeneity of explanatory variables effectively. 3. empirical study 3.1. data considering data availability, the paper sets sample from the data of property insurance premiums, gdp, cpi and fixed-asset investment during the period from 1980 to 2012 to study the relationships between property insurance premiums and macroeconomic variables. the data of property insurance premiums from 1980 to 1998 originates from property insurance [15], while others come from insurance association of china. published by atlantis press copyright: the authors 177 besides, units are one hundred million yuan. the data of gdp, fixed-asset investment, consumer price index comes from the national bureau of statistics. the units of two former are one hundred million yuan, the latter chooses 1978 as the base period. in order to eliminate the effects of heteroskedasticity, the paper takes the natural logarithm on property insurance premiums, gdp, and fixed asset investment. the actual value of property insurance premiums income, gdp, and fixed assets investments are replaced by p, gdp, and fai respectively. and their logarithm value is replaced by lp, lgdp, and lfai respectively. in additional, cpi is also represented by cpi. 3.2. unit root test ardl bounds testing has certain prerequisites that each sequence is less than integrated of 1. as a result, stationary of each sequence is needed to be tested to determine the order of integration before co-integration test. this article uses eviews6.0 for unit root test and results are shown in table 1. as can be seen, the original sequence of lp, lgdp, and cpi are non-stationary at 10% significance level. via differencing, the differential results of variable lp (dlp) is stationary at the 1% significance level. therefore, lp is i(1).the differential results of variable lgdp (dlgdp) is stationary at the 5% significance level according to adf test, while it is stationary at 10% significance level according to pp test. consequently, lgdp is i(1). the differential results of variable cpi (dcpi) is stationary at the 5% significance level according to adf test, while it is stationary at 10% significance level according to pp test, so lgdp is i(1). the variable lfai is stationary at the 5% significance level according to adf test, while it is nonstationary according to pp test, and its differential results (dlfai) is stationary at the 5% significance level according to adf test or pp test, so dlfai is i(0)or i(1).subsequently, model variables are i(0) or i(1), which qualify prerequisite of ardl bounds testing, so that the ardl model can be established for co-integration test. 3.3. ardl bound testing in order to test the long-term relationships among property insurance premiums, gdp, cpi and fixed-asset investment variables, the optimal lag order of each differential item in formula (2) should be determine by the aic or sbc information criterion firstly. taking the sample size and the lag phase into account, it is better to choose the maximum lag order of the differential variable which is no more than 4. besides, this paper uses microfit4.0 to test two cases whether or not to consider trend. the results are presented in table 2. as shown from table 2, whether containing trend term or not, when n equal to 2, aic and sbc are the most significant. therefore, the best lag order is 2. next, this paper carries out bound margin cointegration test on models with trend or not. when trend table 1 adf and pp unit root test results for each time series. variables adf test form adf test results pp test form pp test results integration lp c,t,0 -2.7878 c,t,4 -2.7156 i(1) dlp c,0,0 -4.9753*** c,0,4 -5.0102*** lgdp c,t,4 -1.1978 c,t,2 -1.6037 i(1) dlgdp c,0,3 -3.5658** c,0,0 -2.7221* lfai c,t,1 -3.8050** c,t,1 -2.2631 i(0) or i(1) dlfai c,0,0 -3.4948** c,0,3 -3.5564** cpi c,t,1 -2.7889 c,t,3 -1.8746 i(1) dcpi c,0,1 -3.3889** c,0,3 -2.6417* note: d represents the first difference; (c, t, n) denotes the unit root test model intercept, time trend and lag order. ***, **, * represent significance at 1%, 5% and 10% levels, respectively; determine lag order by the sic criteria. table 2.aic and sbc information criterion of ardl-ecm model. lag length without trend with trend aic sbc aic sbc 1 18.8805 15.2955 20.1213 15.8194 2 16.0898 9.7844 16.4845 9.4785 3 25.6562 16.7688 24.7097 15.1386 4 26.0334 14.7097 28.4330 16.4432 published by atlantis press copyright: the authors 178 term is excluded, f (lp | lgdp, lfai, cpi) = 10.8455, corresponding p is 0.000, the null hypothesis is rejected. namely, lp and lgdp, lfai, cpi have long-term stable relationships. when trend item is contained, f (lp | lgdp, lfai, cpi) = 8.9462, corresponding p is 0.001, the null hypothesis is also rejected. namely, lp and lgdp, lfai, and cpi have long-term stable relationships. therefore, whether containing trend item or not, there exist long-run equilibrium relationships among the four variables. 3.4. ardl model estimating this paper estimates two cases of ardl model form that considering trend or not. estimation results are shown in table 3. the results can be seen from the table above. whether containing trend item or not, model form is ardl (2,2,2,0) by aic criteria, while it is ardl (2,0, 1,0) by sbc criteria. the above table shows that model error is small by aic criteria, so the model form is ardl (2,2,2,0). table 4 shows estimated coefficients of the model. as can be seen from table 4, when trend term is contained, the trend term in ardl (2,2,2,0) model is not significant. however, when trend term is excluded, the coefficients of model are more significant. therefore, ardl (2,2,2,0) model excluding trend term has priority to estimate the long-term coefficients between the variables and the corresponding error correction model. the table 5 illustrates the long-term relationships, and the equation is: 16.6573 3.0067 0.00628 0.84109 lp lgdp cpi lfai = − + − − (3) in the long run, there is a balanced relationship between property insurance premiums as well as gdp, cpi, and fixed asset investment. the elasticity of gdp, table 3 ardl model form. without trend with trend aic sbc aic sbc model form ardl(2,2,2,0) ardl(2,0,1,0) ardl(2,2,2,0) ardl(2,0,1,0) standard error 0.66877 0.071797 0.065381 0.071944 table 4 ardl (2,2,2,0) estimated coefficients. with trend without trend variables coefficients t statistic p variables coefficients t statistic p lp(-1) 0.40599 2.1121 0.047 lp(-1) 0.44059 2.2595 0.035 lp(-2) 0.25235 1.6529 0.114 lp(-2) 0.32683 2.2318 0.037 lgdp 1.7066 2.6191 0.016 lgdp 1.2535 2.1648 0.042 lgdp(-1) -2.0858 -2.4478 0.024 lgdp(-1) -2.2064 -2.5444 0.019 lgdp(-2) 1.5268 2.6809 0.014 lgdp(-2) 1.1997 2.2568 0.035 cpi -0.00111 -0.43971 0.665 cpi 0.000826 0.38220 0.206 cpi(-1) 0.000517 0.15920 0.875 cpi(-1) 0.000345 0.10371 0.918 cpi(-2) 0.00335 -1.9174 0.070 cpi(-2) -0.0029 -1.6507 0.114 lfai 0.97840 0.55813 0.583 lfai 0.16181 0.93453 0.361 c -8.7392 -1.7372 0.098 c -2.1842 -1.1391 0.267 t -0.06818 -1.4042 0.176 r-squared 0.9991 r-squared 0.9990 dw-statistic 2.5191 dw-statistic 2.5772 table 5 ardl (2,2,2,0) model estimation results of long-term relationships. variables coefficients standard error t statistic p lgdp 3.0067 0.70094 4.2896 0.000 0.008 cpi -0.00628 0.002215 -2.8336 lfai -0.84109 0.42668 -1.9712 0.058 0.000 c -16.6573 2.9942 -5.5632 published by atlantis press copyright: the authors 179 cpi, and fixed-asset investment are 3.0067, -0.00628, 0.84109 respectively. specifically, when gdp, cpi and fixed asset investment change by 1%, the property insurance premiums will change by 3.0067%, 0.00628%, -0.84109% correspondingly. this result shows that gdp is the major factor driving the growth of property insurance premium income in china. moreover, fixed-asset investment has a significant impact on chinese property insurance premium income, and they show the negative correlation. lastly, cpi almost has no effect on the premium income. table 6 shows the error correction model is 1 1 1 2.1842 0.32683 1.2535 1.1997 0.000825 0.002903 0.16181 0.23258 ( 1) t t t lp lp lgdp lgdp cpi cpi lfai emc − − − ∆ = − − ∆ + ∆ − ∆ + ∆ + ∆ + ∆ − − (4) equation (4) gives the short-term dynamic relationships among property insurance premiums as well as gdp, cpi, and fixed-asset investment. the coefficient of emc (-1) is negative and significant at the 5% level, indicating that when the short-term fluctuations deviate from the long-term equilibrium, the adjustment intensity of model modifying from nonequilibrium to equilibrium is -0.23258. 4. conclusions in this paper, with the related data from 1980 to 2012, ardl bound testing method is established to study the long-term relationship between property insurance premiums as well as gdp, cpi, and fixed-asset investment. besides, this paper uses error correction model to analyze long-term impact of above factors. as a consequence, several conclusions can be carried out through the research. firstly, gdp has a positive and marked impact on property insurance premiums. therefore, gdp is the most important factor which affects premiums. in details, when gdp grows by 1%, property insurance premiums will be increased by 3.0067%. secondly, cpi and fixed-asset investment have negative impacts on property insurance premiums in china, and the impact of cpi on property insurance premiums is much less than the influence of fixed-asset investment. economically, gdp growth reflects the improvement of living standards which means that people have the ability to buy more insurance products, promoting the growth of premiums. however, cpi growth means serious inflation and rising prices, which induces the falling purchasing power of buying insurance products. moreover, fixed-asset investment and insurance products are both an investment option for residents when income is certain, the growth of fixed asset investment will have a negative impact on premiums. references [1] j. outreville, the economic significance of insurance markets in developing countries, journal of risk and insurance, 57(3)(1990)487-498. [2] d. ward, r. zurbruegg, does insurance promote economic growth? evidence from oecd countries, journal of risk and insurance,67(4)(2000)489-506. [3] m. kugler, r. ofoghi, does insurance promote economic growth? evidence from the uk, money macro and finance research group in its series money macro and finance (mmf) research group conference, (2005)8. [4] m. arena, does insurance market activity promote economic growth? a cross-country study for industrialized and developing countries, journal of risk and insurance,75(4)(2008)921-946. [5] j. outreville, the relationship between insurance growth and economic development: 85 empirical papers for are view of the literature, international centre for economic research working papers,(2011). [6] baoqing lin, xixi hong, jiangming wu, an empirical analysis about demand income elastic coefficient of table 6 ecm model estimation results of ardl (2,2,2,0). variables coefficients standard error t statistic p dlp1 -0.32683 0.14644 -2.2318 0.036 dlgdp 1.2535 0.57905 2.1648 0.041 dlgdp1 -1.1997 0.53159 -2.2568 0.034 dcpi 0.000825 0.0021607 0.38220 0.706 dcpi1 0.002903 0.0017588 1.6507 0.112 dlfai 0.16181 0.17315 0.93453 0.360 dc -2.1842 1.9174 -1.1391 0.266 0.032 ecm(-1) -0.23258 0.10190 -2.2824 published by atlantis press copyright: the authors 180 property insurance in china, journal of financial research, 7(2004)90-99. [7] xiangyou wu, an empirical study on the unbalance development of china’s insurance industry across provinces: 1997-2007, the journal of quantitative & technical economics, 6(2009)99-113. [8] hongmei zhao, huijuan su, a research on influencing factors for property insurance demands in china: based on region-weighted and time-weighted panel data, insurance studies, 2(2013)38-44. [9] zhen qian, the long term interactive relationship of economic growth 、 residents consumption and insurance development: based on var model and impulse responses function. statistics & information forum, 7(2008)50-54. [10] jian sun, and chunhai zhang, study the mechanism of the entire insurance industry and economic growth based on var model, insurance studies, 9(2010)35-39. [11] weinan rao, research on the relationship between the total premium income of china's insurance industry and gdp, economic research, 4(2011)45-48. [12] w. w. charemza, d. f. deadman, new direction in econometric practice(edward elgar, uk1992). [13] m. h. pesaran, y. shin, an auto-regressive distributed lag modeling approach to co-integration analysis(cambridge university press, cambridge1999). [14] m. h. pesaran, y. shin, r. j. smith, bounds testing approaches to the analysis of level relationships, journal of applied econometric, 16(2001)289-32. [15] lin qiao, xujin wang, property insurance(china renmin university press, beijing 2003). published by atlantis press copyright: the authors 181 1. introduction 2. model and methods 3. empirical study 3.1. data 3.2. unit root test 3.3. ardl bound testing 3.4. ardl model estimating 4. conclusions references atlantis press journal style study and application on risk assessment method of coal worker pneumoconiosis based on logistic regression model * qian zhang, deyin huang, minyan li institute of occupational health, tianjin bohai chemical industry group co. ltd, tianjin 300051, china received august 18, 2018 accepted september 8, 2018 abstract to research the risk assessment method of coal-worker's pneumoconiosis caused by coal dust exposure, predict the incidence of coal-worker's pneumoconiosis, and provide technical basis for occupational risk management. coal-workers of a coal transportation workshop of a heat power station were chosen as objectives. the incidence of coal-worker's pneumoconiosis was predicted by logistic non-linear regression model using the concentrations of the respirable coal dust that the workers exposed to and their exposure durations. furthermore, a practical table of the relationship between the incidence probability of coal-worker's pneumoconiosis and exposure situation was created by using r programming language. it was found that if these coal workers exposed to the respirable coal dust of the current concentration for 5 or 10 years, the risk of coal-worker's pneumoconiosis would be between 10 -5 and 10 -6 , and for 20 or 30 years, it would be between 10 -3 and 10 -2 . assuming that they exposed to coal dust of this concentration for 30 years, the risk would exceed the acceptable risk level and measures should be taken. in this study, a practical quantitative approach was proposed to assess the risk of coal-worker's pneumoconiosis caused by coal dust exposure. based on the results, it showed that as the risk management target, the concentration and exposure duration should be well controlled to reduce the risk of coal-worker's pneumoconiosis. this method could be applied in evaluations of occupational disease hazard in construction projects and help to control and manage the risk of coal-worker's pneumoconiosis. keywords: coal dust; occupational exposure; pneumoconiosis; risk assessment; logistic regression model * corresponding author : deyin huang (1962-), female, chief physician, research direction is occupational health and risk analysis. this study was supported by municipal key science and technology support project of tianjin (no. 13zczdsy02300) and projects funded by the national health service (no. 1311400010903). 1. introduction pneumoconiosis is one of the legal occupational diseases in china, which is a systemic disease mainly caused by diffuse pulmonary fibrosis because of long-term inhalation and deposition in the lung of productive dust in occupational activities. pneumoconiosis is still the most prevalent occupational disease in china, and the annual number of reported pneumoconiosis accounts for more than 80% of the total number of reported occupational diseases. according to 2015-2016 published national occupational disease reports, a total of 29,180 occupational diseases were reported in 2015 and 26,081 new cases of occupational pneumoconiosis were reported, among which 93.92% were coal worker pneumoconiosis and silicosis. a total of 31,789 occupational diseases were reported in 2016, and 27,992 new cases of occupational pneumoconiosis were reported, an increase of 1911 compared with 2015, among which 95.49% were coal worker pneumoconiosis and silicosis. therefore, it is urgently needed to carry out the research and application of occupational exposure risk assessment technology for productive dust such as coal dust, and formulate the best prevention and control measures of pneumoconiosis through risk assessment, so as to provide scientific and technological support for the 157 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 implementation of special management of productive dust operations and prevent the occurrence of pneumoconiosis. this paper introduced the method of coal dust occupational exposure risk assessment based on the example of a large petrochemical enterprise thermal power plant coal transportation workshop. 2. object and method 2.1 object the coal transporters in the coal transportation workshop of the thermal power department of a large petrochemical company were mainly responsible for the discharge, storage and loading of fuel coal. the workers have the characteristics of long dust exposure and high exposure level, and a new case of pneumoconiosis has been reported. therefore, the company's thermal power plant coal transportation workshop was selected as the research object to evaluate the risk of occupational exposure to coal dust in the coal transportation workshop. 2.2 method the risk assessment of occupational exposure to coal dust was carried out according to the four-step method of epa health risk assessment, which was widely used in the world, namely risk identification, exposure assessment, dose-response assessment and risk characterization [1-3] . the procedures and contents of risk assessment methods for pneumoconiosis caused by occupational exposure to coal dust were as follows: （1）risk identification in order to identify occupational exposure and potential risk of coal dust in the workplace, relevant data was collected, including occupational history, number of exposed population, gender, age distribution, exposure mode, exposure time, worker protection conditions, free silica content of coal dust, epidemiological data of workers exposed of coal dust. （2）exposure assessment the respirable dust of coal dust was collected, and the scene sampling and laboratory testing were conducted in accordance with current standards related to occupational health. （3）dose-response assessment the logistic model was selected as the dose-response model of coal worker's pneumoconiosis. population epidemiological data were used to determine the model parameters, and a dose-response model was obtained to evaluate coal worker's pneumoconiosis caused by respiratory coal dust exposure. non-linear regression of spss 19.0.0 was used to fit the dose-response model parameters of coal worker pneumoconiosis caused by occupational exposure to respirable coal dust. （4）risk characterization the dose-response model was used to calculate the risk of coal worker pneumoconiosis with different working age. the risk of coal worker pneumoconiosis caused by the long-term work of coal dust exposed population in this environment was characterized. the relationship between respiratory coal dust concentration, dust duration and incidence of pneumoconiosis was given by computer r language. the technical route was shown in the figure below. figure 1 technical route of occupational disease risk assessment for coal transporters in thermal power plants 3. results 3.1 occupational hygiene survey of coal transportation workshop the technology operation of coal transportation workshop includes train unloading, coal loading and belt transportation. storage technology: transportation coal by train screw coal unloader coal yard. transportation technology: ring type coal feeder vibrating feeder belt screen crushing belt coal bin. the coal transportation workshop was equipped with comprehensive ventilation facilities, with a total of 93 dust removal facilities, and workers were provided with dustproof masks. risk identification occupational hygiene survey occupational epidemiology data collection occupational exposure assessment determination of dose-response model parameters dose-response assessment logistic model chart of relationship between respiratory coal dust concentration, dust duration and incidence of pneumoconiosis computer r language calculate directly according to the formula of dose-response model risk characterization 158 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 3.2 determination of coal dust exposure level in workplace field sampling and laboratory testing were conducted in accordance with gbz159 specifications of air sampling for hazardous substances monitoring in the workplace, gbz/t192.1 determination of dust in the air of workplace-part 1: total dust concentration and gbz/t192.2 determination of dust in the air of workplace-part 2: respirable dust concentration. according to the exposure data obtained from the occupational health survey, and based on the monitoring data of environmental concentration in the field, the permissible concentration-time weighted average of 8 hours of respiratory coal dust was calculated to evaluate the risk of pneumoconiosis caused by coal dust. the test results of coal dust concentration were shown in table 1. table1 the detection results of coal dust in the air of workplace position coal pipe pump inspector 2#,3# belt inspector 4# belt inspector results twa （mg/m3） 0.9 2.7 4.5 3.3 dose response relationship of coal worker pneumoconiosis based on logistic model （1）logistic regression model the logistic model was selected as the dose-response model of coal worker pneumoconiosis. as a statistical method, the logistic regression model could objectively and truly present the internal relationship between risk factors and disease risk. basic principles of the logistic model: a series of case data was used to fit the logistic regression model, and the model was used to present the relationship between the probability of taking certain values of independent variables and dependent variables. at the same time, all independent variables other than a specific independent variable were controlled to reflect the magnitude of the deterministic effect of that specific independent variable on the probability of taking a specific value of the dependent variable. the risk factors affecting the occupational exposure to coal dust were taken as independent variables and the prevalence rate of coal worker pneumoconiosis as dependent variables. multiple logistic regression analysis was performed to fit the logistic regression model. the following formula shows the relationship between the probability of a disease p and the daily dust concentration a and the working age of the dust exposure b: where: p—the incidence rate of pneumoconiosis of coal workers with different dust concentration and working age; βi0—the constant term of logistic regression analysis; βi1—the regression coefficient of logistic regression analysis; a—daily dust concentration（mg/m 3 ）; b—the working age of the dust exposure（year）. （2）the parameter fitting of the logistic model data of pneumoconiosis epidemiology of coal workers in typical domestic industries were collected. according to the data of occupational exposure level of coal dust and incidence of pneumoconiosis in each investigation instance, the parameter values in the logistic model were fitted. this study mainly collected and adopted the following epidemiological investigation data of coal worker pneumoconiosis [4-9] , as shown in table 2. table 2 epidemiological survey results of coal workers with pneumoconiosis region industry type of work cumulative exposure to dust （respiratory coal dust） （mg/m3·year） cumulative incidence a coal mine in hebei coal mine coal digger 8.73 0 26.18 0 43.63 0.0008 61.08 0.002 78.53 0.0036 95.99 0.0125 113.44 0.0156 130.89 0.0207 148.34 0.031 165.79 0.0431 183.25 0.0573 200.70 0.0787 218.15 0.0974 235.60 0.1434 253.05 0.1918 270.51 0.2402 287.96 0.288 305.41 0.3165 322.86 0.3592 340.31 0.3592 inner mongolia coal mine equipment operator, belt driver, etc 61.08 0.0194 78.53 0.0441 the yangtze river north shore port transportation stevedore 52.36 0.1781 shanxi coal mine coal mine workers 16.29 0.01 13.04 0.005 159 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 based on the dose-response relationship between occupational exposure accumulation of respiratory coal dust and incidence of pneumoconiosis in table 2, spss 19.0.0 logistic nonlinear regression was used to fit the collected data of pneumoconiosis epidemiology. the parameter values in the logistic model were obtained: βi0=-20.890, βi1=3.508, r 2 =0.898. by testing, it could be considered that the product of dust concentration and working age of the dust exposure, namely, the accumulated dust quantity was related to the incidence rate. the dose-response model of coal worker's pneumoconiosis is: where: a—respirable coal dust concentration, namely permissible concentration-time weighted average of 8 hours（mg/m 3 ） b—the working age of the dust exposure（year） 3.4 risk characterization results （1）direct calculation using the logistic regression model established in the dose-response assessment, the actual exposure level data of the workers were substituted into the dose-response model established above to calculate the risk value of coal worker pneumoconiosis. the coal dust exposure at each position and the predicted risk of coal worker pneumoconiosis were shown in table 3. table 3 coal dust exposure and risk of coal worker pneumoconiosis at each position position concentration of respirable dust（twa） mg/m 3 risk of coal worker pneumoconiosis with different working ages 5 years 10 years 20 years 30 years coal pipe pump inspector 0.9 1.66×10 -7 1.88×10 -6 2.14×10 -5 8.89×10 -5 2#,3# belt inspector 2.7 7.81×10 -6 8.89×10 -5 1.01×10 -3 4.18×10 -3 4# belt inspector 4.5 4.69×10 -5 5.33×10 -4 6.03×10 -3 2.45×10 -2 （ 2 ） relationship between respiratory coal dust concentration, dust duration and incidence of pneumoconiosis based on computer r language in order to improve the practicability of the risk assessment method in this study and facilitate the calculation of the risk of pneumoconiosis caused by occupational exposure to respirable coal dust, a zoom table was given for the relationship between the dust permissible concentration-time weighted average of 8 hours, the working age of the dust exposure and the incidence rate of pneumoconiosis of coal workers. the incidence of pneumoconiosis of coal workers corresponding to different dust concentration and 5, 10, 20 and 30 years of working age could be found in the table. taking the 20-year working age of coal conveyors as an example, the risk prediction of coal worker pneumoconiosis exposed to different respirable coal dust concentration was shown in table 4. table 4 prediction of the risk of coal worker pneumoconiosis with 20 years of dust exposure working age under different concentrations of respirable coal dust concentration (mg/m 3 ) risk concentration (mg/m 3 ) risk concentration (mg/m 3 ) risk concentration (mg/m 3 ) risk concentration (mg/m 3 ) risk 0.5 2.73×10-6 10.5 1.06×10-1 20.5 5.53×10-1 30.5 8.33×10-1 40.5 9.31×10-1 1 3.10×10-5 11 1.22×10-1 21 5.74×10-1 31 8.41×10-1 41 9.34×10-1 1.5 1.29×10-4 11.5 1.40×10-1 21.5 5.94×10-1 31.5 8.48×10-1 41.5 9.36×10-1 2 3.53×10-4 12 1.59×10-1 22 6.14×10-1 32 8.55×10-1 42 9.39×10-1 2.5 7.71×10-4 12.5 1.79×10-1 22.5 6.32×10-1 32.5 8.62×10-1 42.5 9.41×10-1 3 1.46×10-3 13 2.01×10-1 23 6.50×10-1 33 8.68×10-1 43 9.43×10-1 3.5 2.51×10-3 13.5 2.23×10-1 23.5 6.67×10-1 33.5 8.74×10-1 43.5 9.46×10-1 4 4.00×10-3 14 2.45×10-1 24 6.83×10-1 34 8.80×10-1 44 9.48×10-1 4.5 6.03×10-3 14.5 2.69×10-1 24.5 6.98×10-1 34.5 8.85×10-1 44.5 9.49×10-1 5 8.71×10-3 15 2.93×10-1 25 7.13×10-1 35 8.90×10-1 45 9.51×10-1 5.5 1.21×10-2 15.5 3.17×10-1 25.5 7.27×10-1 35.5 8.95×10-1 45.5 9.53×10-1 6 1.64×10-2 16 3.42×10-1 26 7.40×10-1 36 8.99×10-1 46 9.55×10-1 6.5 2.16×10-2 16.5 3.67×10-1 26.5 7.53×10-1 36.5 9.04×10-1 46.5 9.56×10-1 7 2.78×10-2 17 3.91×10-1 27 7.65×10-1 37 9.08×10-1 47 9.58×10-1 7.5 3.51×10-2 17.5 4.16×10-1 27.5 7.76×10-1 37.5 9.12×10-1 47.5 9.59×10-1 8 4.37×10-2 18 4.40×10-1 28 7.87×10-1 38 9.15×10-1 48 9.61×10-1 8.5 5.35×10-2 18.5 4.64×10-1 28.5 7.97×10-1 38.5 9.19×10-1 48.5 9.62×10-1 9 6.46×10-2 19 4.87×10-1 29 8.07×10-1 39 9.22×10-1 49 9.63×10-1 9.5 7.70×10-2 19.5 5.10×10-1 29.5 8.16×10-1 39.5 9.25×10-1 49.5 9.65×10-1 10 9.08×10-2 20 5.32×10-1 30 8.25×10-1 40 9.28×10-1 50 9.66×10-1 160 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 in addition, the relationship between pneumoconiosis incidence and dust concentration of coal workers with different working age could be plotted according to the risk prediction zoom table. take the 20-year working age of coal conveyer for example, see figure 2. figure 2 relation between incidence of pneumoconiosis and dust concentration in coal workers with 20 years of working age 4. risk control countermeasures and suggestions as can be seen from the results of risk characterization, based on the working age of 30 years, it was assumed that workers do not wear dust masks, and that the risk of exposure to coal dust was higher than 1 ×10 -4 level for belt inspectors 2#,3# and 4#, and measures should be taken to reduce the risk. in view of the high risk of occupational hazards existing in the occupational exposure of coal dust, the engineering technical control measures should be firstly improved according to the priority of risk, and the following problems existing in the current occupational-disease-prevention facilities should be improved. (1) the excessive coal dust in the existing post mainly refers to the loose coal dust when the belt machine drops the coal. it was suggested to consider setting a better sealed dust cover at the coal drop place, and to form a micro negative pressure inside the cover when the coal falls, so as to reduce the loose coal dust. carefully analyze the reason why existing dust removal system cannot effectively prevent dust, further clarify the form, location and quantity of the dust removal system, and ensure that the coal dust concentration must measure up to the standards. (2) reduce the transfer drop of coal as much as possible. it must be ensure that the automatic dust remover and spray dust removal facilities set at each transfer point of belt conveyor can meet the requirements of dust prevention with the long-term use. (3) strengthen the tightness of the pulverizing system of the workshop, prevent the leakage of coal powder and reduce the impact on the working environment. (4) strengthen wet operation with process permission. the atomizing sprinkler facilities , corresponding supporting water pipes and gathering flume should be added to ensure the implementation of wet operation and reduce dust. (5) the wind speed of the dust collector should be increased, in order to improve the efficiency of dust collection and make the concentration of coal dust in the air of workplace measure up to the standards. in addition, strict occupational health management measures should be adopted to strengthen personal protection of workers, equip workers with dust masks, and guide and supervise workers to wear them correctly; strengthen the training and education for operators to use and maintain respiratory protective equipment, and require operators to strictly observe operating procedures; strengthen the daily maintenance and overhaul of the protection facilities to keep them in normal operation and good ventilation and dust removal effect. regular occupational health inspections and routine monitoring of coal dust were needed to control and reduce risks. 5. conclusions and discussions by using logistic regression model of the nonlinear regression analysis, this paper set up the occupational exposure risk assessment method of coal dust on the basis of conventional epidemiological data, through the main factors of coal workers pneumoconiosis respirable coal dust concentration and the working age of the dust exposure. this method could predict the coal workers pneumoconiosis incidence probability, in order to evaluate occupational disease hazard risk of coal workplace. risk characterization adopted computer r language to conduct a visual study on the relationship respiratory coal dust concentration, dust duration and incidence of pneumoconiosis, and conduct a zoom table of the results of risk characterization, which was convenient for occupational health workers to use and consult, so that the research results have strong practicability and operability. 161 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 by case study, it was proved that the risk assessment method of coal dust occupational exposure proposed in this study could be applied to the occupational hazard assessment of construction projects, coal dust occupational exposure assessment and risk analysis, as well as the daily risk management of occupational hazards of employers and the management of high-risk occupational diseases such as coal dust. this was of great significance to optimize working conditions, prevent and control occupational diseases and reduce the risk of occupational diseases. references [1] li min-yan, huang de-yin, liu mao. mcsim-based occupational health risk assessment on benzene[j]. journal of risk analysis and crisis response, 2013, 3(3):135-145. [2] zhang qian, huang de-yin, liu mao. internal exposure simulation based on exposure related dose estimating model[j]. journal of risk analysis and crisis response, 2013,3(4):175-184. [3] zhang q, huang d y, liu m. study on risk evaluation based on occupational exposure evaluation and carcinogenic risk simulation[j]. journal of risk analysis and crisis response, 2014, 4(4):228-232. [4] wang y j, zhang j, yin x y. development trend of incidence of pneumoconiosis in tianjin[j]. chinese journal of industrial hygiene and occupational diseases, 2001,1:70-71. [5] hua zheng-bin, sun zhi-qian, qin tian-bang. dose-response relationship between cumulative dust exposure and pneumoconiosis risk among workers exposed to dust in a coal mine in hebei[j]. journal of labour medicine, 2015,32(1):1-6. [6] wang xin-ping, liu tie-min. study on the dose-response relationship between respirable coal dust and pneumoconiosis[j]. journal of safety science and technology, 2007,3(2):70-73. [7] cai li-qun wei zhe. study on the developmental regularity of coal workers' pneumoconiosis using logistic regression model[j]. chinese journal of industrial hygiene and occupational diseases, 1990,8(4):223-226. [8] pei jun-xue, li wan-peng. the relationship between coal dust exposure and pneumoconiosis [j]. inner mongolia coal economy, 2001,4:12-14. [9] hu jian-an, chen an-chao, li pei-shan, et al. dose-response relationship between respirable dust and pneumoconiosis in coal mine[j]. chinese journal of industrial medicine, 1992,5(2):66-69 162 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 157-162 received 11 april 2015 accepted 24 may 2015 disaster risk research literature on statistics analysis in china journal net shuzhen li, alateng tuya* college of geographical science, inner mongolia normal university，hohhot 010022, china, e-mail：920194467@qq.com, altty@imnu.edu.cn abstract disaster risk literature collect from cnki, analysis of the development of china journal net’s literature on disaster risk. according to the literature was published in the year, subject, author, literature sources, databases, was cited and downloaded several aspects, such as for quantitative analysis. statistical analysis of the literature on disaster risk can draw the following result: risk of disaster in china started late until the beginning of 2007, the increasing number of documents; domestic disaster risk leading university research station in nanjing information engineering university, the leading journal is disaster reduction in china; domestic cooperation on disaster risk degree of harmony with the rate of increase in value more, gradually in-depth research; core journal literature citations high, indicating that field has a central effect, receive attention in academia. keywords: disaster risk; bibliometrics method; cnki 中国期刊网灾害风险文献统计分析 李淑珍，阿拉腾图娅 1 内蒙古师范大学地理科学学院，内蒙古 呼和浩特 010022，中国 摘要：从中国知网搜集灾害风险研究的文献，分析国内期刊网灾害风险研究文献的发展。使用文献计量法， 对检索到的灾害风险文献，按文献发表年份、题目、作者、文献来源、数据库、文献被引及下载频数等方 面来进行量化分析。统计分析灾害风险研究文献可以得出以下结果：灾害风险研究在国内起步晚，直到 2007 年开始文献数量迅速增多；国内对灾害风险研究站领先地位的大学是南京信息工程大学，期刊是中国减灾； 国内对灾害风险研究作者合作度和合著率数值越来越增高，可知该研究领域发表文献质量越来越提高、研 究内容逐渐深入；核心期刊文献被引频次高，表示该领域中具有核心效应，在学术界得到重视。 关键词：灾害风险；文献计量法；中国知识资源总库 项目基金：国家自然基金项目（41261099）资助 作者简介：李淑珍（1991—），女，内蒙古人，硕士研究生，研究方向：自然地理学、自然灾害研究 *通信作者：阿拉腾图娅（1969—），女，教授，从事资源环境遥感研究 journal of risk analysis and crisis response, vol. 5, no. 2 (july 2015), 129-140 published by atlantis press copyright: the authors 129 mailto:altty@imnu.edu.cn 前言： 由于现代社会人口不断增加、物质财富不断增多， 灾害的后果日益严重，使我国涌出大量的研究员对灾 害风险进行研究，灾害风险研究文献数量不断增多。 科学文献的数量是衡量科学知识的重要尺度之一，科 学知识量的增长及其规律与科学文献的增长及其规律 是紧密联系的，灾害风险文献的数量变化可以反映该 科学领域知识量的变化情况。文献计量方法可以分析 某一个科学的产生、发展、分化和相互渗透的动向； 可以确定重点文献、核心作者及核心期刊，为文献报 道和指导读者重点阅读提供最佳方案；可以了解研究 科学家的分布，评价作者名望和识别科学人才。 文献计量学的发展大致分为三个阶段，萌芽阶段 （1917-1933 年）、尊定时期（1934-1960 年）、全面发 展与分化时期（1960 年至今）[1]。20 世纪初叶，人们 已经开始对文献进行定量化研究，但当时文献计量法 并没有作为一门独立的学科而存在，它源于社会科学 借用自然科学研究的方法 [2] ，经过数十年的发展形成 一个独特的研究方法体系。1969 年英国著名情报学家 阿伦· 普里查德（alan pritchard ）首次提出用术语 “bibliometrics”取代“统计目录学”，这一术语的出现标 志着文献计量学的正式诞生 [3] 。直到 80 年代，随着量 化方法在科研评价中的应用，科研评价也经历了从专 家定性评价到定量评价相结合的过程，文献计量方法 在科研评价中的作用逐渐受到重视，它是图书馆学情 报学等领域的方法论体系中的重要组成。 文献计量法是主要由文献计量特征、文献统计分 析的方法、基本流程与文献计量规律几个部分构成 [4、 5] ，以信息的外部特征为研究对象、以定量分析为出发 点，在科学研究的诸多领域中解决了许多实际问题， 得到了社会的普遍认可。近年来，我国发生的灾害的 数量与造成的损失急剧增加，灾害已经是阻碍社会和 经济稳定发展的一个主要因素,同时研究人员对灾害 风险的研究也逐渐深入，为了使灾害风险研究人员迅 速掌握该领域的研究动态，笔者用文献计量法对国内 期刊网灾害风险文献进行统计分析，为进一步研究灾 害风险提供基本数据。通过文献统计分析，了解国内 期刊网灾害风险研究的演进历程，影响力较高的研究 员，已经扩展的范围及发展趁势。 1. 研究方法 1.1. 获取文献资源 文献检索时间是 2014 年 11 月，以中国知识资源 总库（中国知网 cnki）为文献信息获取平台，检索类 型为以“主题”为检索项，以“灾害风险”为检索词，共 检索到 3912 篇文献。其中，期刊文献有 2739 篇，硕 士论文有 568 篇，博士论文有 179 篇，会议论文有 416 篇，学术辑刊文献有 10 篇。且索取每篇文献的题名、 作者、来源、发表时间、数据库、被引及下载次数等 相关信息。再将搜索到的文献手工复制到 excel 表格 中，并进行整理，这种方式虽然繁琐，但收集的文献 比较全面。 1.2. 数据库的选取原因 中国期刊网即有优势也有弊端，本文章主要依靠 它的优势来进行撰写的。首先，与国际三大检索系统 《科学引文索引》—sci、《工程索引》—ei、《科技会 议录索引》—istp 比较可以知道，虽然很多国家把三 大检索系统作为评价高等院校、科研机构的科研和学 术水平的标准，但它们不收录中文期刊。中国期刊网 的大量汉语文献，为中国众多的学者提供知识互动平 台，为国内学术领域的发展提供了基础。其次，与国 内其它数据库检索功能相比的话，中国期刊网也有着 自己不可代替的一面。比如，中国期刊网提供的检索 字段比维普和万方多，它除了提供一些必须的检索字 段以外还提供了参考文献、全文、智能检索、中途分 类号、issbn 等字段，便于更准确的检索到所需要的 文献，并且中国期刊网还能进行引文查询等优势。 published by atlantis press copyright: the authors 130 1.3. 文献数量分析与统计法 对检索到的 3912 篇文献，按学术期刊、硕士论 文、博士论文、会议论文和学术辑刊统计分类。根据 《2014 年北大核心期刊目录》将学术期刊分成普通期 刊与核心期刊。将检索到灾害风险文献采用文献计量 分析法，利用 excel 2010、origin 8.0 等进行统计分析。 按研究需求统计文献数量的年际分布、各期刊的载文 量、各大学发表的硕博士论文数量、核心作者群、被 引及被下载次数等。统计核心作者群时，根据文献计 量学的洛特卡定律 [6] 以第一作者为统计对象，根据普 赖斯（d.price）提出的计算公式确定核心作者群数量 [7] ： m=0.749（nmax） 1/2 (1) 公式中 m 为论文篇数，nmax 为所有统计文献里发表文 献数量最多的作者的总发表文献数量，即发表文献数 量 m 以上的作者称为核心作者群。 统计文献被引用速度时利用加菲尔德（e garfield） 提出的即年指标测度。计算公式: = 当年发表的文献被引用当年文献的篇数 即年指标 当年文献总数 (2) 2. 统计分析结果 2.1. 灾害风险研究文献数量的阶段分析 文献量是指某一学科研究者在某一段时间内所发 表的论文数量的多少。文献量的多少从一定程度上反 映着该学科的发展水平，因此统计分析中国期刊网灾 害风险研究文献的数量来衡量该研究领域的发展动 态。统计 1983—2014 年的中国期刊网灾害风险文献数 量变化，了解我国灾害风险研究领域的发展演进。从 图 1 可以看出，我国期刊网灾害风险的研究可大致分 为初期、发展、兴旺阶段。 1983—2000 年为初期阶段，文献量极少，年均文 献数量为 7.2 篇。其中 1983—1986 年均只发表 1 篇文 献，且未检索到 1988、1989 年的相关文献。最早的一 篇文献是多萝西·内尔金发表在《科学对社会的影响》 上的《谈风险的社会和政治认可》，它是在中国期刊网 上首次发表的灾害风险研究文献。在初期阶段研究文 献主要以风险、灾害及自然灾害等的理论概述为主， 但未形成一个众多研究者认同的统一理论概念。被后 者参考引用的次数最多的是黄崇福发表在核心期刊 《自然灾害学报》上的《自然灾害风险分析的基本原 理》和《以历史灾情资料为依据的农业自然灾害风险 评估方法》。 2000 年开始全球范围内广泛开展“国际减灾战略” （international strategy for disaster reduction, isdr）， 整体上改变了前者的减灾观点，从灾后的反应转变为 灾 前 防 御 ， 推 动 了 学 者 们 对 灾 害 风 险 的 研 究 。 2001—2006 年的发展阶段。此阶段文献量缓慢波动上 升，到 2006 年达到 140 篇，发展阶段年均文献量为 76 篇，文献研究方向逐渐多样化，内容从单一的理论 研究推广到多种方法研究、及广泛到不同区域尺度上 研究。 由于 2007 年发布《国家综合减灾“十一五”规划》 和 2008 年百年奥运会的到来对中国灾害风险研究有 图 1 灾害风险文献数量年际变化 fig.1. annual variations of the published amounts of articles involved the disaster risk published by atlantis press copyright: the authors 131 着很大的推动作用， 2007—2013 年为兴旺阶段。从 2007 年开始中国期刊网灾害风险文献年均数量急速增 长，发展速度迅速上升， 2013 年文献数量达到最多 653 篇，兴旺阶段年均文献数量为 415.7 篇。2014 年 文献数量为 393 篇，因为未能检索到全年的文献数量， 所以在文献数量年即变化图上有下滑现象。国内经过 数十年的灾害风险研究中，研究者们对风险、灾害风 险及自然灾害风险等名词有了较统一的认识。比如， 自然灾害风险的统一概念逐渐浮现，它是由自然事件 或力量为主因导致的未来不利事件情景 [8] 。 初步阶段文献数量少，研究方向也不多，经过发 展阶段和兴旺阶段之后，文献数量不断增加，研究方 向也逐渐增多。总的来说，1983—2013 年间中国期刊 网灾害风险研究文献数量的不断增加表示我国对灾害 风险领域的研究呈现越来越强劲的势头。 2.2. 按文献类型统计分析 2.2.1 硕博士学位论文统计分析 灾害风险研究领域的发展给国内众多硕博士带 来了很多科研机遇，硕博士学位论文发表数量可以反 映灾害风险分析领域的发展现状。统计 1983—2014 年 在中国期刊网上发表的灾害风险硕博士学位论文，最 早是 1992 年北京师范大学的黄崇福的博士论文，题名 为《信心扩散原理与计算机思维及其在地震工程中的 利用》。硕士论文最早是 1999 年大连理工大学的郝军， 文献名为《石化企业投资、灾害风险与安全效益研究》。 表 1.发表论文数量 10 篇或以上的大学 table 1. ten or above ten papers were publishes in the universities 排序 学位授予单位 硕士论文（篇） 博士论文（篇） 总数（篇） 1 南京信息工程大学 44 6 50 2 华东师范大学 18 11 29 3 兰州大学 25 3 28 4 成都理工大学 18 9 27 5 长安大学 20 7 27 6 北京师范大学 12 10 22 7 中国地质大学（北京） 16 5 21 8 西南财经大学 19 1 20 9 中国海洋大学 11 8 19 10 中南大学 13 3 16 11 西南大学 12 3 15 12 重庆大学 10 4 14 13 大连理工大学 9 4 13 14 东北师范大学 12 1 13 15 上海师范大学 10 2 12 16 华中农业大学 6 5 11 17 天津大学 10 1 11 18 中国农业科学院 5 5 10 19 浙江大学 6 4 10 20 首都师范大学 9 1 10 published by atlantis press copyright: the authors 132 之后各大学陆续发表，直到 2014 年灾害风险硕士博士 论文文献总数量达到 747 篇。 在 3912 篇灾害风险文献中硕博士论文有 747 篇， 约占总文献的 19％。其中硕士论有 568 篇、博士论文 有 179 篇，比例为约 3﹕1，分布于 169 所学校。根据 《学位授予单位》统计得出发表硕博士论文在 10 篇或 以上的学校有 20 所，文献发表数量共 378 篇，占总硕 博士论文的 51％。列出这些大学的名称和发表文献数 量见表 1。 南京信息工程大学以 50 篇的总发量在中国期刊 网硕博士论文中处在领先地位，发文献量约占总量 6.7％，在所有大学中发表硕士论文量为最多共 44 篇， 博士论文为 6 篇。另外发文较多的有华东师范大学共 29 篇，其中硕士论文有 18 篇，博士论文有 11 篇，所 有大学中发表博士论的文数量为最多。其次，兰州大 学有 28 篇、成都理工大学和长安大学有 27 篇、北京 师范大学有 22 篇、中国地质大学（北京）有 21 篇、 西南财经大学有 20 篇。还有中国海洋大学、中南大学、 西南大学、重庆大学、东北师范大学、大连理工大学、 上海理工大学、天津大学、华中农业大学、首都师范 表 2. 期刊刊名及载文量 table 2. name of periodicals and the amount of downloading 排序 普通期刊名 载文量（篇） 核心期刊名 载文量（篇） 1 中国减灾 297 自然灾害学报 142 2 中国应急管理 33 灾害学 94 3 中国农学通报 29 中国安全科学学报 23 4 安徽农业科学 28 地理科学进展 21 5 中国地质灾害与防治学报 28 地质通报 19 6 城市与减灾 24 中国农业气象 17 7 防灾科技学院学报 23 北京师范大学学报 15 8 贵州气象 20 资源科学 14 9 中国防汛干旱 17 地理科学 14 10 防灾博览 16 地理研究 13 11 工程地质学报 16 气象 13 12 地质灾害与环境保护 14 中国人口.资源环境 13 13 气象科技 13 气候与环境研究 12 14 气象研究与应用 13 山地学报 12 15 中国民政 13 地球科学进展 11 16 中国水利 12 水文地质工程地质 11 17 海洋开发与管理 11 18 气象与减灾研究 11 19 现代农业科技 11 20 气象与环境科学 10 published by atlantis press copyright: the authors 133 大学、浙江大学及中国农业科学院等大学的文献量均 少于 20 篇。 其余 149 所学校的发表文献量有 369 篇，占总硕 博士总论文 49％。其中 67 所大学只有发表 1 篇论文、 30 所大学发表 2 篇论文、18 所大学发表 3 篇论文、11 所大学发表 4 篇、7 所大学 5 篇论文、8 所大学 6 篇论 文、5 所大学 7 篇论文、3 所大学发表 9 篇论文。 2.2.2 统计分析期刊文献 在学术界里核心期刊和普通期刊的作用和意义有 很大的区别，核心期刊是学术期刊质量评判和论文学 术水平鉴定的主要依据，核心期刊文献是热点研究问 题，因此按《2014 年北大核心期刊目录》统计分类 2739 篇灾害风险期刊文献，其中核心期刊文献有 1017 篇， 普通期刊文献有 1722 篇。图 2 是 1983—2014 年间的 中国期刊网期刊文献分布。 1993、2005 年普通期刊与核心期刊的文献发表数 量相等，在 1999、2001、2003、2005 年核心期刊上的 文献发表数量较多，但相差不大，见图 2。从 2007 年 开始普通期刊文献发表数量急速增长，年均文献发表 量远超核心期刊文献发表量。普通期刊从 2008 年开始 年文献发表数量达到 100 篇以上，核心期刊从 2011 年 开始年文献发表量达到 100 篇以上。2013 年是期刊文 献达到顶峰时期，普通期刊文献在达到 328 篇，核心 期刊文献达到 157 篇。最早发表是 1983 年在《科学对 社会的影响》上的普通期刊文献，核心期刊文献最早 发表是 1987 年在《金融与经济》上。总的来说，2007 年开始期刊文献增加速度比较明显，核心期刊文献数 量也随之增加。可见随着国家建设对灾害风险研究的 需求不断增加，在众多地理学界学者、专家的支持和 培育下，高质量的论文越来越多，灾害风险得到逐步 发展和进步。 核心期刊文献年发表数量增加，但远远比不过普 通期刊的数量增加速度。有两个原因，一个原因是在 核心期刊上发表的论文档次需求比较高，所以文献发 表难度较高。还有一个原因是核心期刊种类比普通期 刊种类少。统计 1983—2014 年间的中国期刊网总期刊 种类数量及每个期刊的载文量可知，2739 篇期刊文献 共有 953 种期刊上发表，其中普通期刊有 672 个，载 文量有 1722 篇。核心期刊有 281 个，载文量有 1017 篇。 比较载文量在 10 或以上的核心期刊与普通期刊， 见表 2。核心期刊有 16 个，载文量为 444 篇，占核心 期刊文献的 44％。普通期刊有 20 个，载文量有 639 篇，占普通期刊文献的 37％。中国减灾以 297 篇的载 文量在中国期刊网灾害风险研究领域中占领先位置， 图 2. 灾害风险期刊文献数量的年际分布 fig.2. number of annual distribution of disaster risk journal literature 表 3. 对比核心期刊与普通期刊被下载次数 table 3. compare of core periodicals and common periodicals in the amount of downloading 被下载次数 3000~2500 2500~2000 2000~1500 1500~1000 1000~500 500~0 0 核心期刊 2 3 13 32 137 815 15 普通期刊 0 0 1 1 43 1606 71 published by atlantis press copyright: the authors 134 其次有，自然灾害学报、灾害学报。普通期刊里，中 国减灾载文量最多，有 297 篇，占普通期刊文献的 17.2％，其次是中国应急管理、中国农学通报等。核 心期刊里，自然灾害学报的载文量最多，有 142 篇， 占核心期刊文献的 14％、其次是灾害学、中国安全科 学学报等。 核心期刊作为重要期刊，得到很多研究员的关注， 统计 1983—2014 年间的中国期刊网期刊文献的被下 载次数，见表 3。统计比较中国期刊网灾害风险核心 期刊文献与普通期刊文献的被下载次数，被下载次数 越的文献表示在灾害风险研究领域中越受关注，影响 力越大。从表 3 可以看出，被下载 2000 次以上的核心 表 4. 灾害风险总文献核心作者及文献篇数 table 4. amount of core authors and literatures in the whole literature about disaster risk 人数 文献第一作者 篇数 1 金磊 50 1 史培军 33 1 徐娜 19 1 黄崇福 17 2 肖志强、扈海波 13 2 张继权、刘希林 11 2 尚志海、刘传正 9 5 周萍、赵思建、殷杰、王绍玉、刘兰芳 7 10 张俊香、张春山、尹占娥、薛晔、唐川、孙筠、罗培、刘耀龙、冯鸽、崔鹏 6 24 朱良峰、朱传林、郑慧、赵州、张文宗、张庆阳、叶欣梁、王志涛、王禹、王东明、 唐亚明、任鲁川、彭贵芬、潘东华、娄伟平、刘敏、刘丽、刘家福、李瑞昌、李红 英、黄慧、胡宝清、杜军、丁文广 5 表 5. 灾害风险核心期刊的核心作者群及篇数 table 5. amount of core authors and articles in the core periodicals about disaster risk 人数 文献第一作者 篇数 1 史培军 20 1 黄崇福 14 1 刘希林 9 1 金磊 8 3 张继权、刘传正、扈海波 6 1 李红英 5 10 赵思建、殷杰、薛晔、吴益平、王志涛、苏筠、尚志海、罗培、刘丽、崔鹏 4 25 周洪建、张俊香、张竟竟、张春山、殷坤龙、谢全敏、唐亚明、唐川、孙才志、石菊 松、任鲁川、潘东华、倪长健、刘新立、李谢辉、金菊良、金江军、蒋卫国、胡培培、 韩用顺、葛怡、杜鹏、丁文广、陈香、陈鹏 3 published by atlantis press copyright: the authors 135 期刊有 5 篇，普通期刊未被下载过。被下载 2000~1500、 1500~1000、普通期刊未被下载过。被下载 2000~1500、 1500~1000、1000~500 次的核心期刊依次有 13、32、 137 篇，普通期刊依次有 1、1、43 篇。表示灾害风险 论文的核心期刊文献的被下载次数比较多，学术界、 社会的关注度高。被下载 500 次以下的核心期刊文献 与普通期刊文献均集中在近几年，表示文献被下载具 有时间滞后特点。 2.3. 对文献作者进行统计分析 2.3.1．统计文献的核心作者群 作者是期刊论文的重要外部特征之一，是决定期 刊论文质量的关键；而核心作者更是期刊质量稳固和 继续发展的坚实基础 [9] 。统计比较中国期刊网灾害风 险总文献核心作者群和核心期刊文献核心作者群，从 灾害风险总文献中统计出共有 2584 位第一作者，核心 期刊文献中统计出共有 757 位第一作者。根据文献计 量学的洛特卡定律以第一作者为统计对象，把 nmax 总 =50 和 nmax 核=20 代入式（1）中，以取整的原则得出 m 总=5 和 m 总=3，即发表文献在 5、3 篇以上的作者为 中国期刊网灾害风险总文献与核心期刊文献的核心作 者群，见表 4 和表 5。 比较表 4 和表 5 看出，灾害风险总文献核心作者 共有 49 位，核心期刊核心作者共有 43 位，其中史培 军、黄崇福、金磊、扈海波、张继权、刘希林等 24 位 作者相互重叠，并且这些作者的文献发表数量均在领 先位置，表示核心期刊核心作者群也是总文献的核心 作者群，核心期刊及核心期刊作者在该研究领域中具 有核心效应。 灾害风险研究总文献核心作者群的文献总发数量 400 篇，占总文献 10％。核心期刊核心作者群的文献 总发数量 189 篇，占核心期刊总文献 19％。这与普赖 斯定律 50％相差太远，表示核心作者发表的文献量极 少，在文献计量角度上核心作者群还未形成，需要更 多的研究人员来领头研究灾害风险研究。 2.3.2．作者合作度与合著率 表 6. 作者发文量与合作度 table 6. amount of publishment by the author and cooperation 年份 发文作者与论文篇数（人/篇） 人次 合作度 合著率％ 1 2 3 4 ≥5 合计 1983—1986 2 2 4 8 4 50 1987—1990 3 3 6 12 1.5 50 1991—1994 11 9 2 22 35 1.6 50 1995—1998 34 13 8 2 1 58 97 1.7 41 1999—2002 46 21 12 8 8 95 197 2.1 51.6 2003—2006 157 152 69 33 44 455 1026 2.3 65.5 2007—2010 515 157 154 125 104 1055 2378 2.6 51.2 2010—2014 826 319 307 220 272 1944 4784 2.5 57.5 平均 2.3 52.1 合计 1594 674 554 388 429 3639 8537 比例 43.8 18.5 15.2 10.7 11.8 100 published by atlantis press copyright: the authors 136 随着对合作者问题研究的逐渐深入，规定了合作 度和合著率两个计量指标来反映某科学领域内作者合 作智能的发挥程度，数值越高，合作智能发挥越充分， 表示研究越深入、文献质量越高。美国学者 d.deb beaver 等在《科学合作研究》这篇论文上通过科学论 文合著的文献计量学方法论证了科学合作研究的客观 存在 [10、11] 。文献作者合作度是指在某一确定时间内， 每篇文献的平均作者数 [12] 。文献合著率是指在某一确 定时间内，合著文献在总文献中的百分比量。 分析中国期刊网灾害风险文献在 1984—2014 年 间作者论文量、合作度和合著率的变化可以总结出以 下几个结果：见表 6 可知，单个作者的发文量为 1594 篇，占 43.8％，两个人及以上作者的发文量有 2045 篇， 占 56.2％，合著的文献数量比单个作者文献数量较多。 说明随着现代科学的发展，灾害风险研究的各种现代 化手段的应用、及学科间的交叉渗透，文献作者开始 向群体联合研究方向发展；平均合作度是 2.3，在 1987—1990 年为最少，有 1.5，之后期间数值一直提 高，直到 2010—2014 年，达到 2.3。平均合著率是 52.1％，除了在 1995—1998 年间为最少，有 41％，其 他期间均在 50％以上，特别是在 2003—2006 年间达 到最高，有 65.5％。表示中国期刊网灾害风险文献的 研究水平逐渐提高，难度逐渐增大、深度逐渐增加； 从合作度和合著率的数量上来看，中国期刊网灾害风 险文献的作者合作成都还未达到很高的水平，需要继 续增加作者合作成度来进一步完善文献质量。 2.4 文献利用速度分析 文献被引用的频次是衡量该科学领域交流程度和 作者吸收同领域研究进展的信息程度，它是科技文献 的重要组成部分，也是衡量文献质量的一个重要指标。 从布拉德福定律（bradford law）可知，某一科学中的 多数论文及研究成果集中于该学科中少数的核心期刊 中 [13] 。在中国期刊网上，统计 1983—2014 年间的核心 期刊、普通期刊，博士论文、硕士论文及会议论文等 文献的被引用频次。表 7 中表示，被引用文献在总文 献中的百分比、平均每篇文献的被引次数及最多被引 次数，表示各类文献被引差距及各类文献的平均被引 次数。 从表 7 可知，核心期刊的被引用最多，1017 篇核 心期刊的 72％曾经被引过，平均每篇核心期刊被引用 14.1 次。其次有，169 篇博士论文的 70％曾经被引过， 平均每篇博士论文被引用 7.6 次。再次为硕士论文， 普通期刊，会议论文等（学术辑刊数量仅有 10 篇，所 以被忽略）。被引次数在 200 次以上有 6 篇文献均为核 心期刊。有史培军的 《三论灾害研究的理论与实践》 被引 508 次；第一作者为周成虎的《基于 gis 的洪水 灾害风险区划研究》被引 315 次；史培军的《四论灾 表 7. 文献被引比例 table 7. percent of quotation by other literatures 文献类型 被引篇数／文献数量 被引总次数／文献数量 最多被引次数 核心期刊 72％ 14.1 508 博士论文 70％ 7.6 74 硕士论文 44％ 1.8 20 普通期刊 38％ 2.9 194 会议论文 10％ 0.2 8 published by atlantis press copyright: the authors 137 害系统研究的理论与实践》被引 272 次；黄崇福的《自 然灾害风险分析的基本原理》被引 250 次；第一作者 为殷坤龙的《滑坡灾害空间区划及 gis 应用研究》被 引 246 次；第一作者是黄崇福的《以历史灾情资料为 依据的农业自然灾害风险评估方法》被引 215 次。其 中，被引指标在微观评价中出现负面现象，微观层次 评价是指对研究人员个人和期刊的评价 [14] 。也有学者 为了提高自己论文的档次而盲目的引用名人的文献。 文献被引用的速度或时差是评价一个科学领域跟 踪最新研究动态的一个重要指标，利用加菲尔德（e garfield）提出的即年指标对灾害风险文献利用速度进 行测度。根据公式 2 计算 1983—2014 年灾害风险文献 的即年指标，指标变化趁势如图 3 所示，图中直线中 断表示某些年间未检索到相关文献。 在 1983~1991 年间，由于文献数量过少，灾害风 险总文献即年指标波动大，接近于 1 或 0，走向两个 极端，没有研究意义。1992—2011 年，灾害风险总文 献从 10 篇左右增加到 500 篇左右，随着总文献数量增 大被引用的文献数量也增加，总文献平均即年指标为 0.68；核心期刊平均即年指标是 0.87，其中未检索到 2000 年的核心期刊。2012—2014 年文献总发表数量增 加到顶端，但从图 3 上看，即年指标均不到 0.5；2013、 2014 年核心期刊的即年指标也不到 0.5，因为文献被 引用具有时间滞后特点。总的来说，灾害风险文献中 核心期刊文献的被引用比较频繁，核心期刊文献具有 核心效应，核心期刊理论所使用的引用指标在一定程 度上也满足了研究人员对学术水平进行评价的需求。 相反，低水平、低质量的科研论文，是不大可能甚至 不会被引用的。 3 结论 基于 cnki 文献数据库的文献统计分析，分析 1983—2014 年的灾害风险研究的文献产出、研究的主 要领域和学科的主要特点等，可以得出以下结论： （1）中国期刊网灾害风险文献发展经过初步阶段 （1983—2000 年）、发展阶段（2001—2006 年）及兴 旺阶段（2007—2013 年）。从文献数量增加速度上看， 在 2007 年以前处于缓慢上升阶段，之后开始快速上 升，直到 2013 年文献发表数量达到顶峰值；从研究内 容上来看，起初发表文献方向单一，随着灾害风险研 究领域的发展，研究内容变得更加多样、全面、成熟。 总之它是一门新兴学科，有现越来越强劲的发展势头。 （2）统计硕博士论文得知，南京信息工程大学论 文总发表数量占领先地位，其中硕士论文数量在总硕 士论文中最多。华东师范大学的总发表数量文排第二， 其中博士论文是总博士论文中的最多。硕博士论文发 表比率为 3:1，国内博士人数少，对博士论文质量的要 求高，发表难度较大。统计期刊文献得知，中国减灾 的载文量在灾害风险研究领域中排第一，其次是核心 期刊自然灾害学报的载文量最多。我国核心期刊数量 比普通期刊少，但核心期刊文献质量高，具有核心作 用，被下载次数多。核心期刊近几年的平均被下载次 数较少，表示文献被下载具有时间滞后特点。 （3）统计中国期刊网灾害风险文献核心作者群可 知，核心期刊文献的核心作者群在整个研究领域中具 有核心效应，关注度高。但用普赖斯定律计算可知， 核心作者发表的文献量少，在文献计量角度上核心作 图 3. 1983—2014 年即年指标变化趁势 fig. 3. a few years from 1983 to 2014 index trend published by atlantis press copyright: the authors 138 者群还未形成，需要更多的研究人员来领头研究灾害 风险研究。灾害风险研究领域的作者合作度和合著率 逐渐增加，表示文献作者开始向群体联合研究方向发 展，有利于产出高质量的文献。 （4）统计中国期刊网文献被引频次可知，灾害风 险文献中核心期刊文献的被引用率最高，核心期刊文 献质量高，也是研究领域的热点问题。其次依次是博 士论文、硕士论文、普通期刊、会议论。根据际年指 标测度可知，核心期刊即年指标数值比总文献几年指 标高，均在 0.5 以上。但在 2012—2014 年文献即年指 标均不到 0.5，因为文献被引用具有时间滞后特点，文 献被发表之后的短期间里不可能多次被引用。 参考文献： [1] 赵蓉英，徐丽敏.文献计量学发展演进与研究前沿 的 知 识 图 谱 探 析 [j]. 中 国 图 书 馆 学 报,2010,36(189):061-064 [2] zhao rongying, xu limin. the knowledge map of the evolution and research frontiers of the bibliometrics[j]. journal of library science in china, 2010, 36(189) :061-064 [3] 王曰芬，文献计量法与内容分析法的综合研究[d]. 南京理工大学，2007:1-47 wang yuefeng, a synthetic research of bibliometric method and content analysis method[d]. nanjing university of science and technology, 2007: 1-47 [4] 邱均平.文献计量学[m].北京：科学技术文献出版 社，1988:1-40 qiu junping. bibliometric[m]. beijing: science and technology literature press,1998:1-40 [5] 罗武胜,文献计量学概论[m].广州：中山大学出版 社，1994:2-36 luo wusheng. introduction to literature metrology[m]. guangzhou: sun yat-sen university press, 1994: 2-36 [6] élaine gauthier. bibliometric analysis of scientific and technological research: a user,s guide to methodology[c]. science and technology redesign project statistics canada,1998. [7] lotka, a. j. the frequency distribution of scientific productivity, journal of the washington academy of science, 1962, 16(12):317-323 [8] 丁学东.文献计量学基础[m].北京：北京大学出版 社,1993：204-232 ding xuedong. the foundation of the literature metrology[m]. beijing: peking university press,1993：204-232 [9] 黄崇福.自然灾害风险分析与管理[m].北京：科学 出版社，2012：43 chongfu huang. analysis and management of natural disaster risk[m].beijing: science press, 2012: 43 [10] 廉清，《图书情报工作》核心作者群分析研究[j]. 现代情报，2004，（11）:55 lian qing,the analysis and research of the major authors in 《library and information service》 [j].journal of modern information，2004,（11）:55 [11] 金炬，武夷山，梁战平.国际科技合作文献计量学 研究综述——《科技计量学》（scientometrics）期 刊相关轮综述[j].图书馆情报工作，2007，52（3）： 64 jinju, wu yishan, liang zhangping. a summary of bibliometric study on international science and technology cooperation[j]. library and information service, 2007, 52（3）：64 [12] beaver d deb, rosen r. studies in scientific collaboration: partⅰ: the professional origins of scientific co-authorship: scientometrics, 1978, 1(1): 65-84 published by atlantis press copyright: the authors 139 [13] 安秀芬. 《图书情报工作工作》论文作者群的统 计分析[j].图书馆情报，1994,（5）:29-30 an xiufen. the statistic analysis of the authors in 《library and information service》[j]. library and information, 1994, (5): 29-30 [14] 杨廷郊，马费城.布拉德福定律的理论发展[j].情报 学刊，1981，（4）:61-62 yang tingjiao. bradford law’s theory development. journal of information science[j],1981,（4）:61-62 [15] 武夷山,梁立明.采用文献计量学指标进行科研绩 效量化评价应注意的几个问题[j].分析与评论， 2001,12(2)：110 wu yishan, liang liming. some problems about should pay attention to use the bibliometric index to proceed scientific research performance quantitative evaluation[j]. analysis and comment, 2001,12(2):110 published by atlantis press copyright: the authors 140 1. 研究方法 2. 统计分析结果 3结论 characteristic analysis of ecological water use in xilin river basin based on remote sensing hongbo yu 1 , jie chen 2,* , qiaofeng zhang 1 1. college of geograghy science, inner mongolia normal university, huhhot 010022, china 2. inner mongolia radio and tv station, huhhot 010058, china received october 27, 2018 accepted november 15, 2018 abstract our study area is the xilin river basin. remote sensing approach was used to retrieve ecological water use by using modis images from apr. to sept. of 2000、2007、20102014, and using auxiliary environmental data from the same time periods. the results showed the spatial distribution of ecological water use was consistent with the land surface condition. the ecological water use had the trend of strip reduction from the upstream to the middle and lower reaches. the annual variation was extremely related to the precipitation. large amount of precipitation in the year, ecological water use was larger, and vice versa. the areas with higher ecological water use were forest land and farmland, followed by unutilized land. grassland and construction land were lower. the order of ecological water use of different land use type was basically: forest land> farmland>unutilized land>construction land>grassland. in the year of abundant water, the time of the maximum value of ecological water use appeared in june-july with the time of precipitation. in dry year, the time of the maximum value had no rule. the single factor correlation analysis of the monthly ecological water use and monthly meteorological factors showed that the most relevant meteorological factor is precipitation. the correlation coefficient is 0.776. keywords: xilin river basin, ecological water use, remote sensing, land use 基于遥感的锡林河流域生态用水特征分析 于红博 1 ，陈 杰 2,* ，张巧凤 1 1. 内蒙古师范大学地理科学学院，呼和浩特 010022 2. 内蒙古广播电视台，呼和浩特 010058 摘 要：本研究选择锡林河流域为研究区，利用遥感方法对2000年、2007年、2010年-2014年每年4-9月的 modis影像数据进行处理，结合同期气象资料以及实测数据估算流域生长季生态用水量。结果表明，流域 生态用水量空间分布与地表状况比较吻合，生态用水量基本均为由上游到中下游呈条带状减少的趋势，年 际变化与降水量极为相关，降水量大的年份，生态用水量大，生态用水量大的区域所占面积越大，包括流 域大部分区域，反之亦然。生态用水较高的地类为林地和耕地，其次为未利用地，草地和建设用地生态用 水较低，基本为林地>耕地>未利用地>建设用地>草地。在降水充沛的年份，生态用水最大值主要出现在雨 量充沛的6-7月份，在降水欠缺的年份，最大值出现的时间没有规律。将月生态用水量与月气象数据进行了 单因子相关分析，表明降水是影响生态用水量大小的主要限制因子，相关系数为0.776。 关键词：锡林河流域；生态用水；遥感；土地利用 *通讯作者：1106275721@qq.com 216 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 1 引言 锡林河流域位于内蒙古高原中东部，在内 蒙古锡林郭勒盟和赤峰市克什克腾旗境内，东 缘属于大兴安岭西麓低山丘陵区。该流域内的 天然草地是内蒙古高原的典型代表，是京津冀 等地区重要的绿色屏障，该流域拥有目前我国 最大的草原与草甸生态系统类型的自然保护 区，对草原生物多样性保护起到重要作用，在 全球变化动态监测等方面占有重要的空间位 置和明显的国际影响 [1] 。该流域属于干旱半干 旱地区，降水量小，蒸散量大，水量收支不平 衡。特别是近年来，煤矿开采频繁，城镇化发 展迅速，工业用水增加，人类在将水资源的使 用权优先地赋予给灌溉农业、工业和居民生活 的同时，挤占和剥夺了生态系统用水，水资源 供需矛盾十分严峻。如何保护生态环境和合理 配置水资源是一个备受关注并需亟待解决的 问题，生态用水研究受到了专家和学者的广泛 关注 [2-5] 。 “生态用水”一词由汤奇成于 1989 年提 出，随着人们对生态环境的日益重视，生态用 水问题已引起越来越广泛的关注，成为各级领 导、专家学者讨论和研究的热点。“生态用 水”其内涵宽泛，尚未有明确统一的定义。许 多研究者根据不同的理解，对生态用水概念作 出多种不同的界定，其中将蒸散量看作是生态 用水占有相当比例。宋炳煜先生于 2003 年指 出生态用水是维持各类生态系统正常发育与 相对稳定所必需消耗和现存的水分，认为草原 植被的生态用水量主要体现在维持草原植被 正常发育与相对稳定所消耗(植物蒸腾和土壤 蒸发)的水和在现状环境条件下土壤所贮存的 水量，因此其数量等于植物蒸腾量以及土壤蒸 发量（即植被蒸散量）与土壤贮水量之和 [6] ， 本研究采用了该观点。 本研究选择锡林河流域为研究区，运用遥 感方法，结合气象资料和实测数据，对 2000 年、2007 年、2010 年-2014 年每年 4-9 月的 modis 影像进行处理，估算出该流域的生长 季生态用水量，来探究锡林河流域生态用水量 的时空变化特征、不同土地利用类型生态用水 特性以及自然影响因子。本研究对该流域植被 耗水量和生态用水量有定量的了解，对科学有 效地利用干旱地区的有限水资源，为生态环境 发展的可持续性以及水资源的可持续利用具 有重要的现实意义。 2 研究区概况 锡林河流域地势由东向西逐渐降低，流域 总面积约 10786km 2 ，总长 175km。该流域地 貌具有明显的分区性，锡林河以南为多级玄武 岩台地，锡林河中下游是以低山丘陵与高平原 相间分布为特征的内蒙古高原的一部分。该区 的土壤具有明显的地带性，即由东南向西北有 规律地分布着黑钙土、暗栗钙土和淡栗钙土。 气候属大陆性温带半干旱气候，冬季寒冷干 燥，夏季温暖湿润，根据多年气象资料，锡林 河内从东南向西北降水量逐渐递减，东南部的 年降水量为 400mm 左右，西北部为 250mm 左 右，年平均气温则从东南向西北逐渐增加。 锡林河流域地带性植被的基本类型是草 原(草甸草原、典型草原)，约占植被总面积的 85%；锡林河上游地势较高的三级熔岩台地 上，代表群系为贝加尔针茅(stipa baicalensis) 草原和线叶菊(filifoliusibiricum)草原，中游暗 栗 钙 土 亚 带 的 代 表 群 系 为 羊 草 (leymus chinensis)草原和大针茅(stipa grandis)草原，下 游淡栗钙土亚带的代表群系为克氏针茅(stpa krylovii)草原和冷蒿(artemisia frigida)草原；在 锡林河流经区域的地段上形成了湿地植被，此 外，在锡林河流域还有浑善达克沙地东北端榆 树疏林、灌丛、草本镶嵌分布所形成的沙生植 被 [1][7] 。 图 1 锡林河流域空间位置图 fig.1. spatial location map of xilin river basin 3 数据来源 3.1 气象数据 由中国气象科学数据共享服务网提供的 2000 年、2007 年、2010 年-2014 年每年 3 月 5 日-10 月 15 日逐日气象数据，包括大气相对湿 度、平均气温、实际水气压、2m 高度处风速、 降水、实际日照时数。气象数据包括研究区锡 林浩特市及周边阿巴嘎旗、克什克腾旗、东乌 珠穆沁旗、西乌珠穆沁旗、化德、多伦县共 7 个站点的站点数据。 以上气象数据均利用 kriging 方法进行插 值，得到面数据。 217 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 3.2 遥感数据 遥感影像资料为美国 nasa 提供的 2000 年、2007 年、2010 年-2014 年每年 3 月 5 日-10 月 15 日锡林河流域 16 天合成 modis 卫星数 据，是经过验证、modis 科学组推荐使用的 l3 级产品。包括地表温度产品 mod11a2，地 表反射率和太阳天顶角产品 mod09a1，地表 窄波段反照率产品 mcd43a3，植被指数和太 阳天顶角产品 mod13a1，版本均为 5.0。经 过处理，数据分辨率统一至 500m，投影为 albers conical equal area。 3.3 实测数据 野外实测数据为 2013-2014 年 4-9 月份 0-0.5m 深度土壤含水量数据，每个样地三次重 复，取平均值，在研究区内共设有 30 个样地。 3.4 数据处理软件 用 mrt 软件对 modis 影像进行投影转 换，用 pci、envi 和 arcgis 软件对 modis 影 像进行处理、运算、分析和制图，用 visual foxpro 6.0 进行部分程序的编写。 4 理论及研究方法 根据宋炳煜先生的观点，生态用水量为： 生态用水量=植被蒸散量+土壤贮水量。 流域蒸散量主要基于地表能量平衡原理 估算 8 ： hgrle n  (1) 式中，rn 为地表净辐射通量，w/m 2 ，g 为土 壤热通量，w/m 2 ，h 为显热通量，w/m 2 ，le 为潜热通量，w/m 2 。 首先估算地表特征参数各分量(归一化植 被指数，植被盖度，地表比辐射率，地表反照 率)，然后得到地表能量平衡各分量(净辐射通 量，土壤热通量，显热通量，潜热通量)，由 此算出瞬时蒸散量，通过积分运算，时间尺度 扩大到日蒸散量，根据包括该月中日子的遥感 结果按所占日数进行了加权求和得到月蒸散 量。具体方法及公式请参见参考文献[8]。 土壤贮水量根据地表含水量指数 swci 进行运算 [9] ： 76 76 bb bb swci    (2) 式中，b6、b7 分别为 modis 数据的第 6、第 7 波段反射率值。该模型从水的吸收对植被和 土壤反射率波谱的综合影响入手，直接获取其 地表水分含量指标。计算得到的值用野外实测 数据(0.5m 深度内的土壤含水量)进行了修正。 同时根据包括该月中日子的遥感结果按所占 日数进行了加权平均得到月土壤贮水量。 由于本文采用地表能量平衡方法和地表 含水量指数法估算流域蒸散量和土壤贮水，该 方法并不适用于水体，而且所用 modis 产品 的分辨率为 500m，该流域水体面积小且河流 窄，在影像图上基本没有显现，综上两点，本 文所研究区域不包括水体。 5 蒸散量结果检验 利用 fao 推荐的估算参考作物蒸散的 penman-monteith(简称p-m)公式和作物系数对 2012年3月29日-10月6日每8天的日均蒸散量 (共计23个时间段)进行了计算，fao推荐的估 算参考作物蒸散的p-m公式为 [10] )34.01( )( 16.273 900 )(408.0 2 2 0 u eeu t gr et ds d n        (3) 式中，et0 为参考作物蒸散量，mm/d；rn 为作 物表面的净辐射通量，mj/(m 2 d)；g 为土壤热 通量， mj/(m 2 d) ；  为饱和水汽压斜率， kpa/℃； s e 为饱和水汽压，kpa； d e 为实际 水汽压，kpa； 为干湿球常数，kpa/℃；u2 为 2m 高度处的风速，m/s；td 为日均温，℃。 通过参考作物蒸散量et0和作物系数kc可 以确定某种作物的实际蒸散量，即 et=kc·et0 (4) 式中，et 为作物实际蒸散量，mm/d。 参考 fao56 文件，作物系数取值如下： 4-5 月份为 0.8，6-7 月份为 0.95，8-9 月份为 0.85。同时对遥感反演的值进行了流域平均， 二 者 进 行 对 比 ( 图 2) ， 平 均 相 对 误 差 为 16.678%，在误差允许范围之内。因此，说明 该遥感方法有一定的可用性。 图 2 p-m 估算蒸散量与遥感反演蒸散量对比图 fig.2. contrast diagram of p-m evapotranspiration and evapotranspiration of remote sensing retrieval 218 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 6 结果分析 6.1 锡林河流域生态用水时空分析 为便于比较，采用统一的分类阈值，制作 了 2000 年、2007 年、2010-2014 年每年生长 季(4-9 月份)流域总生态用水量(图 3)。由于植 被类型、自然条件的差异，锡林河流域各地区 生态用水量差异较大，其空间分布与地表状况 比较吻合，由上游到中下游逐渐减小，与植被 覆盖区域、气候条件以及土壤的地带性分布基 本一致，由东南到西北呈条带状减少趋势。河 流及两侧的低湿地植被、上游地区的草甸草原 生态用水量较大；林地、耕地生态用水量大于 草地和建设用地；植被密集地带大于稀疏植被 的生态用水量。 图4为生长季生态用水和降水随时间的变 化曲线图。由图可知，降水充沛年份生长季的 生态用水明显高于降水稀少年份生长季的生 态用水。生态用水量由蒸散量和土壤贮水量组 成，因此与降水量极为相关。由图3和图4可知， 降水量大的年份，生态用水量大，生态用水量 大的区域所占面积越大，包括流域大部分区 域，降水量小的年份，生态用水量也小，生态 用水量大的区域仅出现在上游地区及河流流 经地段。图5为七年生长季生态用水的直方图， 直方图的峰值及主值区间的变化与降水多寡 的变化相一致，在降水充沛的年份，峰值与主 值区间均向横轴右侧移动。锡林河流域日照和 热量充足，能够满足植被生长的光热条件，因 此降水量多寡是影响植被长势好坏的主要原 因，也是限制生态用水大小的主要气候因子。 2000 年 2007 年 2010 年 2011 年 2012 年 2013 年 2014 年 图 3 生长季锡林河流域 modis 影像反演的月生态用水量(mm/month) fig.3. ecological water use retrieved from modis image of xilin river basin in growing season 图 4 生长季生态用水和降水随时间的变化曲线图 fig.4. ecological water use and precipitation change curve of xilin river basin in growing season 图 5 生长季锡林河流域总生态用水量直方图 fig.5. histogram of ecological water use in xilin river basin in growing season 219 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 6.2 不同土地利用类型的生态用水分析 研究区分为 6 个地类：草地、耕地、林地、 未利用地、建设用地和水体。如前所述，本研 究不包括水体。 图 6 为生长季 4-9 月份锡林河流域不同土 地利用类型的生态用水。由图可知，生态用水 较高的地类为林地和耕地，其次为未利用地， 草地和建设用地生态用水较低，基本为林地> 耕地>未利用地>建设用地>草地。林地和耕地 为密集型植被，林分内土壤贮水量高，耕地由 于灌溉用水的补给，土壤贮水量也较高，此外 这两种地类植被覆盖度高，叶面积指数大，植 物蒸腾和土壤蒸发较大；未利用地主要为流域 两侧的低湿地植被，由于临近河流，土壤水分 较充足，植被蒸散量也较大；草地的生态用水 较低，甚至低于建设用地，主要原因一是城镇 分布有不少的绿化面积，种植有大量的灌溉树 种和观赏植物，蒸腾和蒸发较大，二是整个流 域大部分为草地所覆盖，可是草地的类型和覆 盖度是不一样的，虽然上游草地植被覆盖度 高，蒸散量较大，土壤水分较充足，但所占面 积小，中下游草地植被稀疏，蒸散量小，特别 是在降水亏缺的年份，出现干旱状态，所以平 均下来草地的生态用水较低，但在降水充沛的 年份，由于降水补给充足，会出现草地的生态 用水与建设用地的生态用水相当的情况。 图 6 锡林河流域生长季不同土地利用类型的生态 用水(mm) fig.6. ecological water use of different land use type in xilin river basin in growing season 图7为锡林河流域生长季不同土地利用类 型的月生态用水。在降水充沛的年份，有充足 的水分对土壤水进行补给，土壤含水量较高， 同样有充足的水分可供蒸发蒸腾，蒸散量较 大，因此各地类生态用水量均较大，最大值主 要出现在雨量充沛的6-7月份。在降水欠缺的 年份，土壤含水量低，也没有充足的水分可供 蒸散，蒸散量在各个月均相差不大，因此各地 类生态用水量较小，变化也较为平缓，最大值 出现的时间没有规律。 图 7 锡林河流域生长季不同土地利用类型的月生态用水(mm) fig.7. monthly ecological water use of different land use type in xilin river basin in growing season 220 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 6.3 锡林河流域生态用水与气象因子的关 系 图 8 为月生态用水与月降水量散点图，它 们之间有较好的线性相关性： 月生态用水(mm)=0.759×月降水(mm)+110.3 (r 2 =0.601) 图 8 生态用水与降水散点图 fig. 8. scatter diagram of ecological water use and precipitation 七年的月生态用水量与月日均气温、风 速、大气相对湿度、水汽压和月降水量的单因 子相关分析如表 1 所示，结果表明，生态用水 量与降水量的相关性最大，为 p<0.1 显著正相 关，其次为水汽压和气温，均为 p<0.5 显著正 相关，与风速和大气相对湿度的相关性较小， 均为正相关。由此可知，水分是影响生态用水 量大小的主要限制因子，由于气温季节变化明 显，因此气温也是影响生态用水量的重要因 子。 表 1 4-9 月份锡林河流域月生态用水量与气象 因子之间相关关系 table 1. correlation between monthly ecological water use and meteorological factors of xilin river basin from apr. to sept. 指标 气温 风速 大气相 对湿度 水汽 压 降水 量 相关系数 0.336 0.082 0.253 0.380 0.776 7 结论 本研究选择锡林河流域为研究区，对 modis影像进行处理及运算，运用遥感方法 反演出2000年、2007年、2010-2014年每年4-9 月份生长季的生态用水量。经过分析，结论如 下： 流域生态用水量空间分布与地表状况比 较吻合，生态用水量基本均为由上游到中下游 呈条带状减少的趋势。河流及两侧的低湿地植 被、上游地区的草甸草原生态用水量较大；林 地、耕地生态用水量大于草地和建设用地；植 被密集地带大于稀疏植被的生态用水量。年际 变化与降水量极为相关，降水量大的年份，生 态用水量大，生态用水量大的区域所占面积越 大，包括流域大部分区域，反之亦然。 生态用水较高的地类为林地和耕地，其次 为未利用地，草地和建设用地生态用水较低， 基本为林地>耕地>未利用地>建设用地>草 地。 在降水充沛的年份，生态用水最大值主要 出现在雨量充沛的6-7月份，在降水欠缺的年 份，最大值出现的时间没有规律。 将月生态用水量与月气象数据进行了单 因子相关分析，表明降水是影响生态用水量大 小的主要限制因子，相关系数为0.776，它们 之间具有较好的线性相关。 致谢 国家自然科学基金项目(41661009)；内蒙古自治区 自然科学基金（2017ms0408）；内蒙古自治区高 等学校科研项目（njzy17047） 参考文献 [1] p. hao and l. zhen qing, evaluation of grassland ecosystem services in xilin river basin, acta prataculturae sinica. 16(4) (2007) 107-115. 彭皓,李镇清,锡林河流域天然草地生态系统服 务价值评价,草业学报.16(4)(2007)107-115. [2] w. yu juan, w. shu dong, y. sheng tian, et al, dynamic simulation of vegetation eco-water of the yellow river basin, journal of natural resources. 29(3) (2014) 431 440. 王玉娟,王树东,杨胜天,等,黄河流域植被生态 用水过程动态模拟,自然资源学报. 29(3) (2014) 431-440. [3] h. ge sheng, on improvement of ensuring system concerning eco-environmental water use in china, hebei law science. 34(7) (2016) 16-29. 胡德胜,论我国生态用水保障制度的完善,河北 法学. 34(7)(2016)16-29. [4] p. xingyao, l. honglu, l. fahu, et al, ecological water requirement of irrigated region in tongzhou district of beijing based on gis technique, transactions of the chinese society of agricultural engineering. 23(2) (2007) 42-47. 潘兴瑶,刘洪禄,李法虎,等,基于 gis 技术的北京 通 州 区 灌 区 生 态 需 水 研 究 , 农 业 工 程 学 报.23(2)(2007) 42-47. [5] c. chao, p. xuebiao, z. lizhen, et al, impact of climate change on cotton production and water consumption in shiyang river basin, transactions of the chinese society of agricultural engineering. 27(1) (2011) 57-65. 陈超,潘学标,张立祯,等,气候变化对石羊河流 域 棉 花 生 产 和 耗 水 的 影 响 , 农 业 工 程 学 报.27(1)(2011) 57-65. 221 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 [6] s. bing yu and y. jie, discussion on ecological use of water research, journal of natural resources. 18(5) (2003) 617 -625. 宋炳煜,杨劼,关于生态用水研究的讨论,自然资 源学报.18(5)(2003)617-625. [7] g. xiao he, h. chun yang, p. yao zhong, et al, optimizing management on degraded grassland in xilin river basin based on ecological risk assessment, chinese journal of applied ecology. 18(5) (2007) 968-976. 顾晓鹤,何春阳,潘耀忠,等,.基于生态风险评估 的锡林河流域退化草地优化管理,应用生态学 报.18(5) (2007)968-976. [8] y. hong bo, b. yu hai, l. he ping, et al, remote sensing retrieval of evapotranspiration in xilin river watershed, research of soil and water conservation. 21(1) (2014) 224-228. 于红博,包玉海,李和平,等,锡林河流域蒸散量 遥感反演,水土保持研究.21(1)(2014)224-228. [9] d. xiao, w. shi xin, z. yi, et al, a digital image watermarking algorithm based on quadtree and error correcting code, geomatics and information science of wuhan university. 32(3) (2007) 205-207, 211. 杜晓,王世新,周艺,等,一种新的基于 modis 的 地表含水量模型构造与验证,武汉大学学报(信 息科学版).32(3)(2007)205-207,211. [10] y. hong bo, evapotranspiration estimated by a scaling model in the ordos plateau, china, phd thesis, huhhot: inner mongolia university, (2009), pp.26-28. 于红博,黄土丘陵沟壑区植物蒸腾和植被蒸散 估算尺度转换模型研究,内蒙古大学博士学位 论文, 内蒙古大学,2009,26-28. 222 ___________________________________________________________________________________________________________journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 216 -222 eleuteri&arduino_finale individual, community and societal effects of the global financial crisis: a bioecological model francesca eleuteri department of human sciences, lumsa university, 101, piazza delle vaschette rome, 00193, italy e-mail: f.eleuteri@lumsastud.it lisa s. arduino*° *department of human sciences, lumsa university, 101, piazza delle vaschette rome, 00193, italy °institute of cognitive sciences and technologies, national research council, 44, via s. martino della battaglia rome, 00185, italy e-mail: l.arduino@lumsa.it abstract prior research suggests that the current economic crisis has negative impacts on population mental health, while some studies show that the recession also causes positive effects. this paper aims to understand the individual, community and societal effects of the global financial crisis by examining several studies using the bioecological model. the analysis suggests that coping methods for the global financial crisis are highly associated with: social policies and labour market programmes, social support, employment status and individual vulnerability. keywords: economic downturn, economic recession, mental health, suicides, unemployment 1. introduction starting in december 2007, a global financial crisis spread around the world beginning with the housing bubble in the united states. occurring periodically in residential markets, this was characterized by rapid increases in valuations of housing until home prices reach unsustainable levels and then decline. in september 2008, the official application for bankruptcy of lehman brothers in the united states exacerbated the crisis and caused devastating effects on the global markets, bonded together by globalisation. among these main causes of the crisis, high raw material prices such as oil, a world food crisis and high inflation in several international countries threated recession in all parts of the world. finally, a credit crisis with consequent trust collapsed global markets. both western and eastern economies faced recession and the unemployment rate rose in almost every country, even in the wealthiest (f.j. tsai & c.c. chan, 2011). recession is defined as a significant decline in economic activity spreading across the economy, lasting more than a few months, normally visible in production, employment, real income and other indicators (business cycle dating committee of the national bureau of economic research, 2008). although the crisis started in the high-income countries, hitting the real economy with great force, low and middle-income countries were at risk too (a. jack, 2009). this large-scale event effected population on several levels (j.m. kirigia et al., journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 203-213 published by atlantis press copyright: the authors 203 willieb typewritten text received 29 august 2014 willieb typewritten text accepted 20 october 2014 willieb typewritten text willieb typewritten text f. eleuteri, l.s. arduino 2011; m. suhrcke & d. stuckler, 2012) including health. according to studies on economic downturn and health effects, since recession causes unemployment, the adoption of less healthy lifestyles and poor disease management occurs (b. abel-smith, 1986; r. catalano, 1991; r. catalano & b. bellows, 2005; a.e. clark & a.j. oswald, 1994; f. mckee-ryan, z.l. song & c.r. wanberg, 2005; g.c. murphy & athanasou j.a., 1999; d. stuckler et al, 2009). moreover, road-traffic injuries and alcohol-related deaths may decrease during an economic turndowns (u. gerdtham & c.j. ruhm, 2006; c.j. ruhm, 2000; j.a. tapia-granados & a. lonides, 2008; m. suhrcke & d. stuckler, 2012). however, positive effects in the life of an individual can also be brought on by recession: according to these studies, people may engage in more health-promoting activities. by examining evidence from various contradictory approaches, this paper aims to understand the individual, community and societal effects of the global financial crisis. primarily through exploring coping methods and resilience factors, an understanding of how some individuals better cope in economic downturn versus others will be brought forth. to identify the consequences of recession on individuals a comprehensive system approach that can capture complex multisystemic outcomes of a large-scale event is needed. this approach contrasts sharply with the literature on economic crisis, which focuses mainly on sociological effects without considering biological, individual and macrosystemic effects (d. stuckler et al., 2009). therefore, we believe that the ecosystemic model proposed by bronfenbrenner (1979) and then expanded to include biological features (u. bronfenbrenner & s.j. ceci, 1994), is the comprehensive system needed to analyse the individual, family, community, and societal levels of an economic downturn. the model has been applied to several empirical studies examining whether under certain circumstances, the exposure to this event might promote or support resilience. economic downturns are large-scale events which involve, with great force, the whole community. discussion on strengths and weaknesses of research through theoretical perspectives offer a hypothesis explaining a possible link between mental health and economical downturn. however, the empirical literature must evaluate whether such a link exists. the bioecological model will therefore be applied to such an evaluation in order to understand which hypotheses the data will support better. 2. the bronfenbrenner’s bioecological model ecological models represent an evolving and impressive theoretical body, which aims to explain processes and features underlying human development in the environments in which human beings actually live (u. bronfenbrenner, 1994). bronfenbrenner is one of the most active author in this domain who analysed and integrated data collected by researchers in different disciplines over many decades. throughout the years the model was revised and updated, but bronfenbrenner launched the first ecological paradigm in the 1970s (u. bronfenbrenner 1974, 1976, 1977, 1979). this was in response to the restricted theories of developmental psychology, which did not promote studies of children in real-life settings but only in “strange situations with strange adults for the briefest possible periods of time” (u. bronfenbrenner, 1977, p. 513). the bioecological model states that, since the embryological phase, genetic material does not determine definitive characteristics of humans but rather interacts with the environment in determining developmental issues (u. bronfenbrenner & s.j. ceci, 1994). this emphasizes the complex interrelation between genetics and environment. in the earliest phase of human’s life, the psychological processes are not complete because they need psychological content present in the environment. hence, development is determined by bidirectional dynamics, which are already manifestations of attention, action and response. thus, the transformation from genotypes to phenotypes is occurring in two-way processes over time. the bioecological model is defined by three main propositions: (i) proximal processes are formed of interactive communication occurred over extended period of time between a person, an active player and the immediate environment; (ii) the features of the proximal processes change constantly together with person and environment compel humans to adapt to its conditions; (iii) proximal processes are fundamental for actualizing genetic potential for effective psychological development. published by atlantis press copyright: the authors 204 effects of the financial crisis the studies of bronfenbrenner & ceci (1994) aim to explain how genotypes change into phenotypes. they elaborate three hypotheses which provide an insight into the dynamics between genetics and environment which are responsible of human development. the first hypothesis states that proximal processes, by improving developmental functioning, increased the individual differences attributable to actualized genetic potential. this has a direct outcome in terms of heritability (h2): it will be higher when proximal processes are strong and lower when such processes are weak. according to the second hypothesis proximal processes realise genetic potentials both for enhancing functional skills and for decreasing degrees of dysfunction. finally, the last hypothesis states that if individuals live in environments which guarantee developmental resources and encourage engagement in proximal processes, then the capacity of these processes to realize genetic potentials will be greater for those living in more disadvantaged settings (u. bronfenbrenner & s.j. ceci, 1994). for testing these hypotheses, bronfenbrenner & ceci (1994) analysed the literature on genetic inheritance and confirmed that no studies were found on genetic inheritance in contrasting environments that also contained data on proximal processes. furthermore, confirmatory results of these hypotheses suggest that many participants may have hidden capabilities beyond those already shown, and these unvoiced potentials can be recognized through appropriate social policies. therefore, bronfenbrenner (1977) conceptualizes environments as contexts of development, being inspired by lewin’s theory of psychological fields (1917, 1931, 1935): he stated that the ecological environment is characterized by several interrelated structures, each inside the other and moving from the centre to the outside they can be defined as described below. 2.1. microsystems according to the bronfenbrenner’s theory a microsystem consists of series of activities, social roles, and interpersonal relations experienced by the developing human being in a given face-to-face setting with particular physical and material features, and containing other persons with distinctive characteristics of temperament, personality, and systems of belief (u. bronfenbrenner, 1994). the microsystem consists of stable characteristics of the individual, such as selfesteem, previous knowledge, intelligence, developmental level, perception of stress and general health status (m.c. logsdon & s. gennaro, 2005). the typical settings of the microsystems are family, school, peer group and work place. another example could be the religious setting (u. härkönen, 2007; h. penn 2005, 45). proximal processes are more active in sustaining development within the immediate environment of the microsystem (u. bronfenbrenner, 1994). 2.2. mesosystems ‘mesosystem’, offered by bronfenbrenner’s theory (1994), comprises the linkages and processes taking place between two or more settings containing the developing person. therefore a mesosystem is a system of multiple microsystems, including interactions, such as family and peer relationships, or family and neighbors. 2.3. exosystems the exosystem encompasses the linkage and processes taking place between two or more settings, at least one of which does not ordinarily contain the developing person, but in which events occur within the immediate settings that does contain that person (e.g. for a child, the relation between the home and the parent’s work place; for a parent, the relations between the school and the neighborhood group) (u. bronfenbrenner 1989, 227). 2.4. macrosystems the concept of macrosystem has changed due to influences by the vygotski´s theory on bronfenbrenner’s work (u. härkönen, 2007). the current definition (u. bronfenbrenner, 1989, p. 228) runs as follows and the addition is italicized: the macrosystem consists of the overarching pattern of micro-, meso-, and exosystems characteristic of a given culture, subculture, or other broader social context, with particular reference to the developmentally-instigative belief systems, resources, hazards, life styles, opportunity structures, life course options, and patterns of social interchange that are embedded in each of these systems. therefore, the macrosystem comprises the cultural context and includes societal norms, published by atlantis press copyright: the authors 205 f. eleuteri, l.s. arduino sociopolitical factors, cultural subsystem norms, governmental systems, economic factors and the environmental effects of the event. 2.5. chronosystems the chronosystem and its relation to the other fields was developed only later. the chronosystem is a description of the evolution or stream of development of the external systems in time. the chronosystem models can cover either a short or long period of time (u. bronfenbrenner, 1989). bronfenbrenner’s theory has been applied over the years to several research fields, mainly in the developmental and educational psychology, in order to explain child development (u. härkönen, 1991, 1996; h. penn, 2005). the model has proven to be very useful in analysing also the effects of mass trauma caused by catastrophic events (m.a. hoffman and t. kruczek, 2011), where interventions at multiple levels are desiderable. 3. financial crisis and mental health: previous studies and theoretical models several studies regarding the relation between economic downturns and mental health outcomes focus on macrolevel social aspects. in 1929 after world war ii, the u.s. faced the great depression after the fall in stock prices that began around september the 4th, and became worldwide news with the stock market crash of october the 29th of the same year. in most countries, the effect was felt starting in 1930 lasting until the late 1930s or middle 1940s. it was the longest, most widespread, and deepest depression of the 20th century. durckheim (1951) first hypothesized that economic downturns increased psychiatric pathology potentially leading to suicide (a. horwitz, 1984) and that macroeconomic factors could have consequential psychological effects on individuals (z. zivin et al., 2010). afterwards, brenner’s empirical studies confirmed that economic downturns increase mental hospitalization (m. brenner, 1973); negative mental health outcomes were experienced by individuals who had not lost employment but who worked in organizations that lost many employees because of fear and stress (m. brenner, 1990). brenner’s economic change model of pathology is a theoretical model which account for complex relationships between the economic environment and the changes in physical, social and mental well-being (m. brenner, 1987). according to brenner’s model, economic instability is one of the 5 elements to influence mental health outcomes, along with long-term economic growth, economic inequality, adaptational error related to economic growth and random shocks. economic instability includes recession and structural economic changes that are damaging to particular subgroups of society: high unemployment rates and business failures, declines in work force participation, and declining returns to investment (m. brenner, 1987). long-term trend, cyclical change and lag period are three critical factors implied in brenner’s theory. 'longterm trend' refers to lengthy, relatively even process of growth that is healthy for the economy. the cyclical change is the semi-periodic variation of several factors of national economic activity (employment, investment and consumption patterns) and cyclical change is one of the two elements of economic instability. short-term cyclical changes usually interact with longer-term or structural changes. the lag period refers to the time range during which the effects could take place. brenner (1987) stated that the negative effects on the population's health could arise within a 6-year period, including and following recession, and that they will have at least 2 peaks. in the first year of recession, the first peak is characterized by severe psychological reactions to economic downturns; general increases of suicide rate, mortality of highly vulnerable groups of the population, and more criminal acts occur. the second peak involves a decline in nutrition and medical care utilization affecting infant mortality and mortality among the chronical ill. brenner (1987) considers three additional indicators referring to some important behavioural risk factors: alcohol consumption, cigarette smoking and fat consumption. this phase lasts up to three years following recession and it presents negative effects on physical, mental and social health. catalano & dooley (1977) used ‘provocation’ and ‘uncovering’ hypotheses to account for the fact that macroeconomic indicators are directly related to the use of mental health facilities. individuals and society decrease tolerance for disordered behaviour when there is economic instability, hence ‘uncovering’ behaviour resulting in more manifest psychiatric disorder observed by health providers. furthermore, economic downturns can cause new disorders by eliciting or ‘provoking’ published by atlantis press copyright: the authors 206 effects of the financial crisis maladaptive behaviour. catalano et al. (1985) proposed another relevant theory, the ‘shift’ hypothesis. this model suggests that economic contractions cause less private treatment in certain populations, and an increased use of public mental health facilities. the relationship between economic downturns and negative mental health effects has also been explained by other factors. stress, caused by economic concerns, could determine an increase in drug and alcohol use, which can result in poorer mental health (z. zivin et al, 2010). higginbottom et al. (1993) hypothesized that the financial strain, that results from losing a job, can affect marital life and satisfaction and further cause mental distress. according to price et al. (2002), facing and dealing with the financial crisis can produce loss of personal control which further lead to depression, impaired functioning and poor health (r.h. price et al., 2002). financial strain can undermine social support determining negative relationships with family (k.d. lincoln et al., 2005). finally, workplaces form one of the main situations where it is possible to find social support; therefore, unemployment can lead to psychological distress (t. atkinson et al., 1986). these theoretical perspectives and existing research have offered reasons why there may be a link between economic downturns and population mental health, but only the empirical literature can evaluate whether such a link exists. it is fundamental to understand what kind of measurements researchers uses to examine the relationship between economic downturns and mental health outcomes. studies usually use the unemployment rate because it is accessible and may provide a general overview of the well being of the economy (d. dooley & r. catalano, 1984; z. zivin et al., 2010). empirical research studies analyse three primary mental health indicators: mental disorders, admissions to mental health facilities and suicide. these studies examined differences in rates of admissions based on a variety of factors, including sex, age, socio-economic status, educational attainment and prior mental health status. findings differ by demographic characteristics (such as sex and age) as well as occupational status. however, most research on the impact of unemployment on mental health focuses on suicide. 4. application of the bioecological model to the global financial crises the multi-system framework proposed by bronfenbrenner (1994) facilitates the identification of the effects related to economic downturns and allows an interpretation on the different level of the system. several studies state that economic downturns could have negative effects on population mental health (d. stuckler et al., 2009; a. kentikelenis et al., 2011; m. brenner, 1987) whereas other point out that financial crisis may produce in some cases positive outcomes (c.j. ruhm, 2000). the studies analysed are based primarily on a shared western socio-economic system because, even if the effects of the recession appeared in the eastern countries too, several cultural biases could arise. the first selected study investigates mortality rates over the past three decades and identified how governments might reduce adverse effects (d. stuckler et al., 2009). stuckler and et al. (2009), in order to examine the associations between changes in employment and mortality, use multivariate regression, correcting for population ageing, past mortality, employment trends and country-specific differences in health-care infrastructure. moreover, they investigate different associations using several types of government expenditure for 26 european union countries between 1970 and 2007. the conclusions of the study state that societal and cultural subsystem norms can either mitigate or hinder adaptive coping and response; indeed adverse health effects on suicides are mitigated when investments in active labour market programmes are high (>$190 per head). according to stuckler et al. (2009), weaker labour market protections in the central and eastern european countries have made their populations very exposed to the potential for negative health effects when unemployment rates sharply rise. in addition, short-term effects of recession include an increase of premature deaths associated with intentional violence, as well as crime rates (m. brenner, 1987; a. kentikelenis et al, 2011; d. stuckler et al, 2009). researchers observed an high correlations between economic depression with suicides and alcohol-related deaths especially in people of working age. in addition, social inequalities widened, but in the meantime, reductions in road-traffic fatalities have been observed (d. stuckler et al., 2009). within the mesosystem, during the financial crisis, researchers published by atlantis press copyright: the authors 207 f. eleuteri, l.s. arduino observed increased risk in mortality of people with low social support, as in the communist countries of central europe in the 1980s, when the rise in mortality was greatest in unmarried men. within the microsystem, rapid and large rises in unemployment were associated with short-term rises in suicides in working-age men and women, in homicides and in mental distress. moreover, nutrition diet became less healthy during recession. the second study investigates the relationship between economic conditions and health, specifically how health responds to transitory changes in economic conditions (c.j. ruhm, 2000). fixed-effect models are estimated using longitudinal data for the 1972–1991 period, with health inferred by total and age-specific mortality rates and 10 particular causes of death. the unit of observation is the state, and most of the analysis focuses on within-state variations in unemployment and personal incomes; limited attention is focused on changes in national unemployment rates. in addition, microdata for 1987–1995 from the behavioral risk factor surveillance system (brfss) are used to examine how risky behaviours and time-intensive health investments in physical activity, diet, and preventive medical care vary with the status of the economy. state fixed-effects are again checked for, as are a variety of demographic characteristics and general time effects. ruhm (2000) found that employment rates are negatively and significantly related to total mortality and 8 of the 10 specific causes or fatalities, with suicides representing an important exception. moreover, higher joblessness is associated with reduced smoking and obesity, increased physical activity, and improved diet. another representative study describes changes in health and health care in greece (a. kentikelenis et al., 2011). the method includes analysis of data from the eu statistics on income and living conditions, which provide comparable cross-sectional and longitudinal information on social and economic characteristics and living conditions throughout the eu. researchers recruited in greece samples of 12,346 and 15,045 respondents in 2007 and 2009, respectively, by use of consistent methods, of which a total of 26,489 had complete sociodemographic data. they drew reports from medical research institutes, health prefectures, and non-governmental organisations (ngos). these reports include epidemiological indicators, data on hospital admissions, and reports on mental health problems and the status of vulnerable groups. kentikelenis et al. (2011) found a decrease in seeking medical care even necessary because of excessive waiting times and travel distances and reductions in accessing outpatient treatments. moreover, drug users drop drugrehabilitation programmes and individuals committed deliberate self-infection to obtain access to benefits of €700 per month. within the mesosystem, infections rise due to shared facilities among intravenous drug users and at the microsystem level, suicides increased (+40% in the first half of 2011). kentikelenis et al. (2011) observed an increase in hiv infections and heroin use, as well an increase in prostitution and unsafe sex. in addition, their findings report reductions in alcohol consumption and drunk-driving. the inability to repay high levels of personal debt can be contributed as a key factor. the fourth study analyses the association between unemployment and suicide during different levels of national unemployment adjusting for several factors that explain or mediate the relationship (n. maki & p. martikainen, 2010). their data comprised of annual population-register and death-register information on 25 to 64 year old finns at the beginning of each year in the period 1988 to 2003, thus, forming 16 separate followup cohorts. experience of unemployment was measured at baseline and during the previous year for each cohort. suicide was followed for 12 months after each baseline. the method includes comprehensive population registration data with a large random sample of suicides. there was practically no loss of follow-up, and because the variables used in this study were obtained from registers, there was no self-report bias. the categorization of a death as a suicide is made by the forensic examiner; the medico-legal autopsies are carried out by a small group of well-trained pathologists and 97.9% of all violent and accidental deaths are subject to this kind of autopsy. this study observed high excess suicide mortality among those who had experienced any length of unemployment. adjustment for social class and living arrangements decreased the age-adjusted relative differences only a little, but income decreased the association notably especially among men. however, a more detailed understanding of the pathways between unemployment and suicide and of the mediating effect of income, further work should include additional published by atlantis press copyright: the authors 208 effects of the financial crisis confounders such as home ownership or other wealth indicators. workers in a precarious situation, such as unstable employment, had a lower suicide risk during times of high general unemployment, but among the long-term unemployed, the risk either remained quite similar between periods (among the men) or was related inconsistently to general unemployment (among women). this alteration in relative suicide risk by national unemployment level is in agreement with a comparative study of 26 european union countries that found substantial variation between populations in how sensitive suicide mortality was to economic crises. according to the authors (n. maki & p. martikainen, 2010), the explanation for the inverse association between national unemployment levels and suicide mortality rates appearing in finland and sweden includes governmental commitment to social support during times of economic stagnation. this was also explained as the counter-cyclical variation in one of the more proximate determinants of suicide, namely alcohol consumption. maki & martikainen (2010) found that large rises in unemployment were associated with shortterm rises in suicides in working-age men and women, in homicides and in mental distress. research also found that long-term unemployment seems to have a causal effect on the risk of suicide, which among men in particular may be partly mediated by having a low income. the final study analyses the post-world war ii changes in mortality rates in relation to deleterious economic changes, especially unemployment, business failure rates, and declines in real per capita income (m. brenner, 1987). the findings reported are based on 127 multiple regression time-series equations which indicated recession has negative effects on two types of populations: individuals who lose employment and individuals who have shortor medium-term work contracts in firms which lose great numbers of employees derived from the economic crisis. for the second type of individuals, work becomes a stress experience for several reasons. the reasons are a set of factors that brenner (1987) divides into 4 key points, well known in the work-stress literature: i) workers’ error tolerance is reduced causing a major mental effort, anxiety increases over work performances and major responsibility; ii) loss of autonomy and self-regulation; iii) demand for higher levels of productivity, involving quantitative increases in workload; iv) possible decreases in wages and/or lack of career advancement. brenner’s (1987) research found increases in total mortality for virtually all age groups, in both sexes, for major causes of death and causes due to psychopathological conditions. workers' social support networks are known to be important modifiers in the stress-illness process. within the mesosystem, the work environment is one of the most affected and findings show increased work stress such as a smaller margin for error, anxiety over work performance, an increased level of responsibility, closer supervision by management, quantitative increases in workload, lack of career advancement. finally, the economic depression causes disruption of workers' social support networks due to damaged social relationships. at the microsystem level, according to brenner (1987), behavioural risk factors during economic downturns are alcohol consumption, cigarette smoking, and fat consumption. 5. limitations of the selected studies stuckler et al. (2009) has several limitations: they focused for instance on the overall experience of populations. these effects within populations could count each other out to yield no overall population average effect of recession on all-cause mortality rates. secondly, they investigated only mortality rates, which is a measure unrepresentative of the population health. the models proposed by stuckler et al. (2009) almost certainly underestimate the full effects of recession on health. thirdly, the study did not include data regarding social protections although they may vary considerably between countries, in particular in central and eastern european countries, which had high unemployment rates and weak social protections. moreover, they could not measure job insecurity for eu countries over the full study period. finally, as other studies regarding the same topic, they investigated only the short-term consequences of economic downturns. however, their measures of rises in unemployment are signs of the economic stress and uncertainty faced by the population, and not only by people who lose their jobs. the limitations of the brenner (1987) model are related to the lack of absolute evidence, which explains the causal relations, although considerable statistical support is offered for economic instability and its relation to adverse change in national health and wellbeing. maki & martikainen (2010) had limitations in published by atlantis press copyright: the authors 209 f. eleuteri, l.s. arduino their determining of employment status at the time of death as it was only examined at the turn of each year. it is possible that redundancy could affect health in a manner similar to other stressful life events producing higher excess mortality soon after the event and a weakening effect over time. there are substantial differences in study quality, yet these fails to fully explain the heterogeneity in findings. heterogeneity based on elements such as populations, methods for measuring employment, and reporting of suicide likely accounts for some of the observed differences in findings across studies. the research faces measurement problems associated with examining the relationship between economic downturns and population mental health for several reasons: potential unmeasured variables/data missing, level of analysis, use of aggregate economic measures which may not sufficiently capture an individual community’s experiences. the selected studies may be affected by the ecological fallacy (d.a. freedman, 1999). for example, errors that may arise from attributing population-level phenomena to individual experiences may prove to be a concern in this type of research. comparability of research standards are not sufficiently guaranteed because studies cover a long period. moreover, the selected empirical research could not be representative of the specific country or only few countries are studied in the current research. concerns about the limitations of the investigation also arise because research investigates the short-term consequences of economic turmoil (i.e., < 3 years). some effects of the great depression seem indeed manifested 5–7 years after the bank crises of the late 1920s and early 1930s (d. stuckler et al., 2009). the short and long term effects of a crisis differ. for instance, risks of cancer from a rise in tobacco use as a coping method to stress would require decades to manifest as lung cancer (z. zivin et al, 2010). it is important to relate the economic consequences of recession to specific behavioural changes and the temporality of their effects on specific health outcomes. as already stated, some effects will appear in coming years following the current crisis, so that it is possible to consider only older studies based on previous economic downturns, such as in brenner’s (1987) research. finally, investigations were limited to high-income countries with well-developed social safety systems that provide formal insurance mechanisms to populations and a good level of wealth that allows people to react to income shocks. the effects of the current crisis for poor countries are likely to differ in both nature and scale. in poor countries, the population lives in or close to abject poverty, so any aggregate income shock is likely to push many people below subsistence levels. an additional decrease in available resources and hence in consumption in an under-nutrition setting seems very likely to be health and possibly life-threatening in poor countries (z. zivin et al., 2010). 6. conclusions previous research suggests that economic crises have the greatest potential adverse effects on health when economic changes are rapid, social protection and cohesion are weak and drugs and alcohol are widely available (m. suhrcke & d. stuckler, 2012). moreover, there is a significant relationship between economic crises and psychopathology including suicide, help seeking for mental health problems, onset or exacerbation of mood disorders and distress. the poor, less educated, and unemployed populations are the most affected during a recession (z. zivin et al., 2010). however, some studies, such as ruhm’s (2000) research, document a strong inverse relationship between macroeconomic conditions and health. in understanding the individual, community and societal effects of the global financial crisis we examined the evidences from several approaches, the coping methods and resilience factors which allow some individuals to better cope with economic downturn. empirical studies suggest that coping methods for global financial crisis are related to: social policies and labour market programmes (macrosystem); social support and employment status (mesosystem); individual vulnerability (microsystem). the macrosystem level supportive programs promoted by governments are needed to better protect populations and provide for specific measures which help unemployed people and individuals who lose their jobs could mitigate health effects on suicides. high investments in active labour market programmes (> $190 per head) could have positive effects in reducing the suicides rates (d. stuckler et al., 2009). government intervention focused on supportive programs, which help people to find jobs, can highly influence people’s approach to the financial downturns. in this perspective, published by atlantis press copyright: the authors 210 effects of the financial crisis people who react to financial strain in a more positive way or in a proactive way differ from those who negatively cope with crisis regarding environmental factors. within the mesosystem, people who have social support such as family or working environment can positively cope, whereas people with low social support have greater risk in mortality (e.g,. in the communist countries of central europe in the 1980s, when the rise in mortality was greatest in unmarried men) (d. stuckler et al., 2009). this conclusion highlights people’s needs to work in order to have better coping methods facing the economic crisis. the mesosystem is strictly connected to the macrosystem as working policies designed by governments and institutions determine working environment. within the microsystem, analysing the impact of economic change on health, it is crucial to identify the population at risk (m. brenner, 1987). as described, rapid and large rises in unemployment were associated with short-term rises in suicides in working-age men and women (n. maki & p. martikainen 2010; d. stuckler et al., 2009). behavioural risk factors during economic downturns are usually alcohol consumption, cigarette smoking, and fat consumption (m. brenner, 1987). these findings suggest that individual vulnerability factors such as low resilience and genetic predisposition (e.g., to addiction or suicides) could determine the different coping methods facing the financial crisis (b. bondy et al., 2006; m.a. enoch & d. goldman, 1999; m. kreek et al., 2005). according to rutter (2006), resilience can be defined as reduced vulnerability to environmental risk experiences, the overcoming of stress or adversity, or a relatively good outcome despite risk experiences (m. rutter, 2006). the author suggests that resilience is an interactive concept in which the presence of resilience has to be inferred from individual variations in outcome among individuals who have experienced significant major stress or adversity. findings showed that intermittent exposure to brief periods of stress, far from being damaging, increases resistance to later stresses (m. rutter, 2012). therefore, individuals who in the past already experienced economic downturns are more likely to cope better during the crisis. many studies have focused on suicide rates, and future research may focus on measures of psychopathology, as suicide is an extreme outcome and rare event (z. zivin et al., 2010). it would be extremely interesting to investigate recent empirical research on psychiatric disorders. moreover, there is relatively little discussion in the current research of policy approaches to address the negative impact of economic crises on mental health. the few studies that addressed policy questions suggested that the expansion of social protection programs such as labour, force expansion programs, social support systems; furthermore, access to health care and health insurance are needed (d. stuckler et al., 2009; z. zivin et al., 2010). while governments are not inclined to spend money during a recession, expanded societal protection plans seems to decrease the negative effects of economic downturns on population mental health. empirical works on the relationship between economic decline and mental health indicate that health systems and policymakers should consider the health and social impact of economic downturns and develop policies and programs accordingly (z. zivin et al., 2010). based on our review, the multi-system model suggests that policymakers should consider the macrosystem and the mesosystem as the starting point of the policies elaborated on to deal with the current economic and financial crisis. governments should make investments in labour market programs that could help people who lose their job to quickly find another one. specific supportive programs for vulnerable individuals who present risk factors, for instance for substance abuse, suicides and depression, should be improved. for future studies, it will be necessary to focus on the effects of recession on health, analysing different kind of measurements, as the current research mainly investigates mortality rates, which is clearly an incomplete measure of population health. moreover, future research should investigate the long-term consequences of economic crises, as current data show only short-term effects of economic turmoil. to conclude, it has to be taken into consideration that fear and anxiety due to the present crisis could be particularly long-lasting; even when the market recovers, people’s stresses and associated behaviours (such as health-care seeking or alterations to healthsystem budgets) might not (d. stuckler et al., 2009). references published by atlantis press copyright: the authors 211 f. eleuteri, l.s. arduino f.j. tsai and c.c. chan, the impact of the 2008 financial crisis on psychological work stress among financial workers and lawyers, int arch occup environ health, 84, (2011) 445-452. a. jack, how will the financial crisis affect health? financial times, 338, (2009) 858-861. j.m. kirigia, b.m. nganda, c.n. mwikisa and b. cardoso, effects of global financial crisis on funding for health development in nineteen countries of the who african region, bmc int health hum rights, 11, (2011) 1-10. m. suhrcke and d. stuckler, will the recession be bad for our health? it depends, social science & medicine, 74, (2012) 647-653. b. abel-smith, the world economic crisis. part 1: repercussions on health, health policy plan, 1, (1986) 202–213. r. catalano, the health-effects of economic insecurity, american journal of public health, 81, (1991) 1148– 1152. r. catalano and b. bellows, if economic expansion threatens public health, should epidemiologists recommend recession? international journal of epidemiology, 34, (2005) 1212-1213. a.e. clark and a.j. oswald, unhappiness and unemployment, economic journal, 104, (1994) 648-659. f. mckee-ryan, z.l. song and c.r. wanberg, psychological and physical well-being during unemployment: a metaanalytic study, journal of applied psychology, 90, (2005) 53-76. g.c. murphy and j.a. athanasou, the effect of unemployment on mental health, journal of occupational & organizational psychology, 72, (1999) 83-99. d. stuckler, s. basu, m. suhrcke, a. coutts and m. mckee, the public health effect of economic crises and alternative policy responses in europe: an empirical analysis, the lancet, 374, (2009) 315–323. u. gerdtham and c.j. ruhm, deaths rise in good economic times: evidence from the oecd, economics and human biology, 4, (2006) 298-316. c.j. ruhm, are recessions good for your health? the quarterly journal of economics, 115, (2000) 617-650. j.a. tapia granados and a. lonides, macroeconomic fluctuations and mortality in postwar japan, demography, 45, (2008) 323-343. u. bronfenbrenner, the ecology of human development: experiments by nature and design. (harvard university press, cambridge, 1979). u. bronfenbrenner, and s.j. ceci, nature-nurture reconceptualized in developmental perspective: a bioecological model, psychological review, 101, (1994) 568-586. u. bronfenbrenner, ecological models of human development, in international encyclopedia of education, vol.3 2nd, (ed. elsevier, oxford, 1994). u. bronfenbrenner, developmental research, public policy, and the ecology of childhood, child dev., 45, (1974) 1-5. u. bronfenbrenner, the experimental ecology of education, teach. coll. rec., 78, (1976) 157-204. u. bronfenbrenner, towards an experimental ecology of human development, am.psychol., 32, (1977) 515-31. k. lewin, kriegslandschaft, zeitschrift für angewandte psychologie, 12, (1917) 440-47. k. lewin, environmental forces in child behavior and development, in a handbook of child psychology, murchison, c. (clarck university press, worcester, 1931). k. lewin, a dynamic theory of personality (mcgraw-hill, new york, 1935). m.c. logsdon and s. gennaro, bioecological model for guiding social support research and interventions with pregnant adolescents, issues in mental health nursing, 26, (2005) 327-339. u. härkönen, the bronfenbrenner ecological systems theory of human development, scientific articles of v international conference person.color.nature.music. (2007). h. penn, understanding early childhood education. issues and controversies (bell & bain ltd., glasgow, 2005). u. bronfenbrenner, ecological systems theory. annals of child development., 6, (1989) 187-249. m.a. hoffman and t. kruczek, a bioecological model of mass trauma: individual, community, and societal effects, the counseling psychologist, 20, (2011) 1–41. e. durkheim, suicide: a study in sociology (the free press, new york, 1951). a. horwitz, the economy and social pathology, annual review of sociology, 10, (1984) 95–119. z. zivin, m. paczkowski and s. galea, economic downturns and population mental health: research findings, gaps, challenges and priorities, psychological medicine, 41, (2011) 1343-1348. m. brenner, mental illness and the economy (harvard university press, cambridge, ma., 1973). m. brenner, influence of the economy on mental health and psychophysiological illness: international perspective, community mental health in new zealand, 5, (1990) 2– 10. m. brenner, relation of economic change to swedish health and social well-being, 1950–1980, social science and medicine, 25, (1987) 183–195. r. catalano and c.d. dooley, economic predictors of depressed mood and stressful life events in a metropolitan community, journal of health and social behavior, 18, (1977) 292–307. r.a. catalano, d. dooley and r.l. jackson, economic antecedents of help seeking: reformulation of time-series tests, journal of health and social behavior, 26, (1985) 141–152. s.f. higginbottom, j. barling and e.k. kelloway, linking retirement experiences and marital satisfaction: a mediational model, psychology and aging, 8, (1993) 508– 516. r.h. price, j.n. choi and a.d. vinokur, links in the chain of adversity following job loss: how financial strain and loss published by atlantis press copyright: the authors 212 effects of the financial crisis of personal control lead to depression, impaired functioning, and poor health, journal of occupational health psychology, 7, (2002) 302–312. k.d. lincoln, l.m. chatters and r.j. taylor social support, traumatic events, and depressive symptoms among african americans, journal of marriage and the family, 67, (2005) 754–766. t. atkinson, r. liem and j.h. liem, the social costs of unemployment – implications for social support, journal of health and social behavior, 27, (1986) 317–331. d. dooley and r. catalano, the epidemiology of economic stress, american journal of community psychology, 12, (1984) 387–409. a. kentikelenis, m. karanikolos, i. papanicolas, s. basu, m. mckee and d. stuckler, health effects of financial crisis: omens of a greek tragedy, the lancet, 378, (2011) 14571458. n. maki and p. martikainen, a register-based study on excess suicide mortality among unemployed men and women during different levels of unemployment in finland, j epidemiol community health, 66, (2010) 302-307. d.a. freedman, ecological inference and the ecological fallacy, international encyclopedia of the social & behavioral sciences, 549, (1999) 1-7. b. bondy, a. buettner and p. zill, genetics of suicide, molecular psychiatry, 11, (2006) 336–351. m.a. enoch and d. goldman, genetics of alcoholism and substance abuse, psychiatric clinics of north america, 22, (1999) 289-299. m. kreek, d.a. nielsen, e.r. butelman and k. laforge, genetic influences on impulsivity, risk taking, stress responsivity and vulnerability to drug abuse and addiction, nature neuroscience, 8, (2005) 1450-1457. m. rutter, implications of resilience concepts for scientific understanding, annals of the new york academy of sciences, 1094, (2006) 1–12. m. rutter, resilience as a dynamic concept, development and psychopathology, 24, (2012) 335–344. published by atlantis press copyright: the authors 213 microsoft word banking balance sheet channel of systemic risk received 24 january 2015 accepted 20 february 2015 banking balance sheet channel of systemic risk vincent b.y. gan putra business school,universiti putra malaysia, 43400 upm serdang, selangor, malaysia e-mail: gan.phd11@grad.putrabs.edu.my abstract the role of the banking balance sheet as the source and transmitter of systemic risk is explored. we find that the key balance sheet channels of systemic risk are; (i) bank capital structure choice, (ii) interconnectedness and interdependencies among firms, (iii) correlations of asset composition and returns and; (iv) behavioral determinants that affect the choices of bank managers. furthermore, we argue that the source of systemic risk lies with the endogenous risk of the banking balance sheet. we discuss the challenges in managing and measuring endogenous and systemic risk. considering the strong evidence that book values of leverage are key state variables, we suggest new methods to manage and measure systemic risk. keywords: banks, balance sheet, systemic risk, measuring risk 1. introduction systemic risk is an elusive notion. there is a lack of consensus as to what defines systemic risk and how to measure and manage it (de bandt & hartmann, 2000; biasis, flood, lo, & valavanis, 2012; galati & moessner, 2012). by convention, systemic risk is thought to be an external negative shock or system architecture failure (hellwig, 2009; bianchi, 2011). scholes (1996) attributes systemic risk to four causes; (i) behavioral as seen in panic runs; (ii) structural failure of the financial infrastructure; (iii) innovation risk from new financial engineering products and; (iv) dislocations in markets due to unanticipated changes in regulations. rarely is it mentioned that systemic risk is endogenous in nature to the banking balance sheet. most of the discussion in the literature mostly focuses on the role of basel capital requirements in worsening procyclicality but ignore the more fundamental role of the banking balance sheet (drumond, 2009). in this paper, we explore this endogenous paradigm in greater detail to uncover how the balance sheet is related to systemic risk. the aggregate destructive effects of systemic risk on the real economy and social welfare in a financial crisis is immense. * the financial and social costs of the global financial crisis of 2007-2009 were unprecedented. acharya, schnabl and suarez (2013) estimate the losses incurred by asset backed commercial papers conduits alone at $68 billion and $204 billion calculated at a conservative loss rate of 5% and 15% respectively and investors lost an estimated $1.8 billion and $5.2 billion respectively. † the imf (2008a and 2008b) estimates aggregate losses of $945 billion, $565 billion in real estate loans, and $494 billion in other securities in the april of 2008 but revised the figures upwards to $1.4 trillion for aggregate losses, $750 billion in real estate lendings and $650 billion in other securities by october 2008. total bank writeoffs were *see allen, babus, & carlettti (2009) and laeven (2011) for surveys on financial and banking crises † the literature on the international propagation of financial shocks provides rich evidence on the effects of systemic risk to the real economy. jotikashitra, lundblad & ramadorai (2012) show how fire sale shocks from developed economies can affect stock prices in emerging markets. cetorelli & goldberg (2012) show that internal capital markets can act as a channel for the international propagation of domestic liquidity shocks to international markets. schnabl (2012) studying loans to firms and khwaja & mian (2008) who studied bank deposits show how liquidity shocks can transmit from banks to firms where firms face a credit crunch and this increases the likelihood of financial distress for firms in the real economy journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 16-30 published by atlantis press copyright: the authors 16 v.b.y. gan estimated to be around $580 billion globally by september 2007 if the figures from the imf are to be taken at face value. these enormous costs to society are enough to warrant more effort be applied in designing a workable and reliable systemic risk metric. a systemic risk measure is a crucial tool in two important contexts; prudential policy and regulation as a part of risk management and financial intermediation. specifically, regulations and policies are created and evaluated to support systemic financial stability, limiting the risks and costs arising from a crisis and also to provide the tools for crisis resolution (laeven, 2011; galati & moessner, 2012). there is no question concerning the importance of systemic risk measures. nevertheless, despite the rich literature on this subject, a consensus on how best to measure systemic risk is still unobtainable. some authors such as danielson (2002) argue that due to current limitations on technology, systemic risk modeling is a futile venture as it is simply too unreliable, poorly defined and deeply affected by complicated moral hazard issues. others argue that risk metrics themselves aggravate the very crisis it is supposed to prevent. for instance galati & moessner (2012) argue that the crosssectional (distribution of risk) and timeseries (procyclical) dimensions of systemic risk could also make any systemic risk metric potentially unreliable and inconsistent. others have highlighted how model risk and other uncertainties in measurement contribute to the unreliability of systemic risk metrics (danielsson, 2002 and 2008; hansen & sargent, 2011). financial variables are also known to be nonlinear, nongaussian and non-stationary in nature which is at odds with the assumptions of most systemic risk metrics in the course of measurement (bougen & drury, 1975). the purpose of this paper is twofold. the first is to analyze the role of the banking balance sheet as a channel for systemic risk. we present theoretical findings and empirical evidence from measuring systemic risk to argue for an endogenous origin of systemic risk from within the balance sheet itself. we survey the literature and present a synthesis to argue the case that the balance sheet is in fact the main source and transmitter of systemic risk. second, we examine new ways to manage and measure both endogenous and systemic risk. we contribute by making the case that the banking balance sheet is the source and transmitters of systemic risk by virtue of endogenous risk. additionally, we provide arguments and evidence to consider a serious rethinking in how risk is measured by showing that book values instead of market values are the relevant state variable with asset pricing consequences. this leads to our call for a back to basics research to design and construct new metrics that can accommodate all the characteristics of the banking balance sheet in its natural state. this paper is organized as follows; section 2 discusses the balance sheet channel of systemic risk; section 3 discusses how systemic risk is endogenous in the balance sheet; section 4 discusses the management and measurement endogenous risk and section 5 concludes. 2. the balance sheet channel of systemic risk in this section we attempt to answer the question, “what role does the banking balance sheet play in the creation and transmission of systemic risk?” we answer this question by reviewing the insights and empirical evidence drawn from important recent theoretical and empirical works relating to systemic risk measurement. broadly, from synthesizing the literature, we find that the key balance sheet channels of systemic risk are; (i) bank capital structure choice, (ii) interconnectedness and interdependencies among firms, (iii) correlations of asset composition and returns and; (iv) behavioral determinants that affect the choices of bank managers. 2.1. bank capital structure choice major headings should be typeset in boldface with the first letter of important words capitalized. 2.1.1. balancing between capital sufficiency liquidity creation the first and most common class of systemic risk theories relate to banking: addressing the issue of single bank instabilities within the fractional reserve system. the banking balance sheet plays a critical role in providing liquidity to the real economy but is exposed to serious maturity mismatches in assets and liabilities simply because banks delicately balance between borrowing short and lending long. diamond & dybvig (1983) in their seminal paper on bank runs posit that banks are providers of insurance for published by atlantis press copyright: the authors 17 banking balance sheet channel depositors against liquidity shocks. a bank run is seen as a self-fulfilling prophecy triggered by the fear of early withdrawals by a sufficiently large number of depositors. other authors have focused on the informational impact of the viability of bank investments as a trigger to depositor runs. diamond & rajan (2000) study the optimal bank capital structure and its role in liquidity creation. in their model, they show that under uncertainty which increases deposit fragility to runs, bank capital which reduces the probability of financial distress to the bank also reduces liquidity creation and the amount the bank can induce borrowers to pay. this optimal balancing act of liquidity creation, cost of distress and the ability to force borrowers to repay is a consequent trade off effect of the optimal bank capital structure which the authors propose as the explanation of bank capital decline over the last two hundred years. diamond & rajan (2001) then further study the beneficial role demandable debt. they propose in an incentive framework without asymmetric information and loan liquidation costs that deposit contracts commits banks to liquidity creation by satisfying depositors’ withdrawals needs while simultaneously shielding long term borrowers from liquidity shocks despite having relationship related power in loan collection. therefore, in their framework liability fragility is a necessary condition for the efficient provision of credit in the economy. based on this model, diamond & rajan (2005) further argue that bank failures can trigger and propagate a systemic crisis even in the absence of a panic driven run. this is due to the inherent structure of banks that finance illiquid assets with demandable claims. in essence, illiquidity stems from the bank’s asset side of the balance sheet. when a failure occurs, the common liquidity pool is shrunk creating or worsening aggregate shortages leading to a contagion of failures across the system. while past theories focused on deposits, the 2008 crisis instead saw a run on the “shadow banking” sector where the withdrawal of funds were done by financial institutions, and the assets held were asset backed securities (most common were asset backed commercial papers, mortgage securities, collateralized debt obligations and credit default swaps) which were traded on markets. uhlig (2008) updates the past theories of bank runs by proposing a model of a systemic bank run that incorporates the salient characteristics of the 2008 global financial crisis which was fundamentally different from the models proposed by past authors. his model shows that the 2008 crisis is best explained by uncertainty-averse investors who fear being saddled with the worst asset among a diverse portfolio and is therefore not willing to bid more than the lowest price for securities now being sold by the distressed core financial institutions. the larger the market share of these distressed core banks, the larger the additional liquidity is needed; increasing the likelihood of a wide spread run on core banks. a common theme of these studies is the effect of illiquidity arising from bank runs on deposits as a manifestation of systemic risk being the trigger to a systemic crisis. drehmann & nikolaou (2012) explore the issue of liquidity in banking and its relation to provisions of liquidity by central banks. the authors construct a measure of systemic risk based on the bidding of funds at central bank auctions by banks. a bank’s bid for funds reveals its funding liquidity risk. they construct a measure of funding liquidity risk as the sum of the premium banks are willing to pay above the expected marginal rate times the volume bid, normalized by the expected amount of money supplied by the central bank. this measure can be interpreted as the weighted average insurance premium against funding liquidity risk. using a unique and confidential data set of all bids in all auctions by european banks conducted at the european central bank (ecb), they find that funding liquidity risk spiked around key events of crises. typeset subsubheadings in medium face italic and capitalize the first letter of the first word only. section numbers to be in roman 2.1.2 the role of leverage and other short term liabilities conventional wisdom has it that interest rates are the most important macroeconomic state variable that affects economy wide prices and therefore stability (woodford, 2003). however, the events of the 2008 crisis has forced a crucial rethinking about interest rates and leverage. the renewed interest in the procyclicality of leverage and obviously the balance sheet has brought the financial instability hypothesis proposed by minsky (1977 and 1992) back on stage. minsky’s theory describes periods of calm and turbulence with declining risk aversion as prices and debt rises. the minsky moment is when the boom soon becomes bust. this instability is inherent within capitalist economies which seems to eerily describe the 2008 global crisis published by atlantis press copyright: the authors 18 v.b.y. gan and the ongoing european debt crisis. keen (1995 and 2013) formalizes the theory and shows empirically how financially driven business cycles can lead to a debt deflation crisis as the seeds of the crisis is sown when the debt to equity ratio rises while prices and interest rates rise while liquidity becomes scarce. the relationship of liabilities, capital and assets to systemic risk was further explored by geanakoplos (2010). the author points out that the collateral rate (leverage) is an equilibrium variable separate from interest rates. major movements in the collateral rate are the result of the leverage cycle and hence can be an indicator of systemic risk. he refers to the phenomenon of the leverage cycle where despite being highly in debt, people and firms can purchase many assets with little money forward and the other end of the cycle where asset purchases can be fully or almost fully funded when leverage is very low. when leverage is low, the increase in purchases drive prices upwards as credit is easily available but prices plummet when credit becomes more constrained. this cycle ends when bad news emerges and creates uncertainty, a sharp rise in collateral rate, and when leveraged people and firms suffer losses and bankruptcies. these factors reinforce each other in a vicious feedback loop which was the prime character of the 2008 crisis originating from the real estate market financed with subprime loans. empirically, adrian & brunnermeier (2010) propose to measure systemic risk by proposing the conditional value-at-risk (covar) of the financial system, conditional on the distress of individual financial institutions. they found that certain characteristics of the firm were good predictors of systemic risk; higher leverage (total book assets/ total book equity ratio), higher degree of maturity mismatch between liabilities and assets and larger size as measured by total assets. lopezespinosa, moreno, rubia, & valderrama, (2012) found that short term wholesale funding to be the most significant balance sheet contributor to systemic risk utilizing the covar methodology. the authors suggested that an optimal capital buffer structure be designed by weighing the relative importance of systemically important factors of banking systemic risk which are bank size, interconnectedness, substitutability, global activity and complexity against bank capital requirements. applying the covar methodology to stress testing, maino & tintchev (2012) further expand stress testing of individual countries to costress testing related financial institutions. the authors model bank capital asset ratios ( total capital/ risk weighted assets) which are used as regulatory capital requirements in basel ii as a function of future losses and credit growth using a generalized method of moments to calibrate adverse shocks to credit quality (represented by nonperforming loans) and credit growth. their proposed measure of systemic risk: the costress, mirrors the covar measure described earlier and captures the tail risk co movements among banks in the system. they define this measure as the level of banking stress conditional on the distress of individual banks. this allows the marginal contribution of an individual firm to system wide risk to be calculated in a procedure similar to covar. the key finding from empirical analysis is that credit risk is a major systemic vulnerability. banks with weak capital buffers and a high proportion of non performing loans were vulnerable to moderate credit quality shocks and therefore very vulnerable to insolvency. while adrian and brunnermeier’s (2010) covar focuses on measuring system wide risks, acharya et al (2010) propose the systemic expected shortfall (ses) to measure an institutions’ contribution to systemic risk. ses measures the likelihood that an institution will be undercapitalized in the event the whole system is undercapitalized as well. further analysis on the levels of systemic risk of financial firms found that surprisingly, insurance firms contribute the least systemic risk. securities brokers were found to be the riskiest and leverage to be key driver of systemic risk. 2.1.3 bank size traditionally it has been viewed that banks that are larger and more international in outlook are more stable and less likely to face insolvency (vennet, 1996). empirical evidence provided by amihud, delong & saunders (2002) showed no evidence that cross border banking mergers contributed to high systemic insolvency risk to either parties of the acquisition. however, recent evidence from systemic risk measurement has shown otherwise. bank size as measured by book assets were consistently found to be important predictors and channels for systemic risk. recently, vallascas & keasey (2012), measured systemic via two measures; the distance to default beta (βdd) and coexceedances. by modeling bank equity as a call options on the market value of assets, the authors use the distance to default which is the number of standard deviations t the market value of assets are above the default threshold where the market value of assets is lower than the book value of liabilities to derive βdd which is estimated by median regression. βdd measures the sensitivity of a bank’s default risk in relation to changes in system wide risk. larger values indicate higher exposures to systemic shocks. they show that bank size as measured by total assets, the share of noninterest income and the growth of earning assets are also key determinants of the bank’s exposure to systemic risk. interestingly, they show that bank size is a relative risk determinant, with smaller economies published by atlantis press copyright: the authors 19 banking balance sheet channel being better off than with small banks rather than large banks. krause & giansante (2012) extend network analysis of banks even further by incorporating the structure of the balance sheet in constructing a network of interbank loans and simulate how a contagion can spread from one balance sheet to the other. the authors measure contagion as the fraction of failing banks in their simulation. they trigger an exogenous failure and track the spread of this failure within the system. in their model, a bank’s balance sheet is characterized by cash reserves, loans to customers, and interbank loans on the asset side and deposits interbank loans and equity on the liability side. they model the default of interbank loans and track how it affects the liability side of lending banks. if the amount repaid by the defaulting party creates losses that exceed the lending party’s equity, the lending party will have to liquidate its assets to increase equity. the defaulting party will also call in loans borrowed out to other banks who at the same time suffered losses from the defaulting party’s inability to repay loans in full. subsequently, the lending party will have liquidated its assets even further to repay its interbank loans. a full blown contagion is now present and both banks are now in financial failure. the key findings of the simulation point to bank size being the prime factor determining the occurrence of contagion in the system. but the extent of which contagion spreads is determined by the network structure of interbank loans which measure the degree of interconnectedness among banks. 2.1.4 the role of nontraditional banking activities and off balance sheet items with banking regulation being relaxed, for instance, the repeal of the glasssteagal act of 1932 which allowed commercial banks to venture into investment activities, the demand for securitized assets boomed. securitization allowed banks to maximize profits and leverage while maintaining required capital ratios constant through the creation of off balance sheet special purpose vehicles. essentially engaging in regulatory arbitrage. acharya et al (2013) show empirically how banks were engaged in this behavior but were concentrating risks on their balance sheets instead of efficiently distributing it across the economy as per their traditional role in financial intermediation. this finally led to a run in the shadow banking sector. gorton & metrick (2008) also show that the global financial crisis can be traced to its beginnings in the august of 2007 with the run on repos in the shadow banking sector which finally led to the beginning of the global crisis with the demise of lehman brothers in september 2008. deyoung & torna (2013) provide evidence that while the wide range of nontraditional activities deserves attention with special caution be paid to activities like venture capital, investment banking and asset securitization. further investigation by papanikalaou & wolff (2013) found that off balance sheet leverage to be especially pernicious and further suggest that investment banking activities be again off limits to commercial banks. de jonghe (2010) measures systemic risk by measuring the tailβ which indicates the probability of an extreme decline in a bank’s stock price conditional on a crash of the banking index. key findings show the degree of noninterest income can increase the tailβ and therefore providing evidence that the increase in noninterest income increases systemic risk in agreement with other studies such as brunnermeier, dong & palia (2012). additionally, the study provides evidence that smaller and better capitalized banks are able to cope better with extreme shocks. 2.1.5 the choice between debt or equity and systemic risk equity is the risk capital of a firm, contributing to stability and solvency. debt is used as leverage to increase earnings by contributing more capital to finance assets. the choice between debt and equity and the optimal mixture of both is one of the most important managerial decisions. it not only affects firm value, future investments but also the systemic risk that arises from the balance sheet. the balance sheet itself can become a source and mechanism of financial contagion that may lead to a system wide collapse (krause & giansante, 2012). changes in leverage can increase risk and therefore signal negative market information with adverse effects to firm value especially when the firm is overladen with debt; suffering from debt overhang. the debt overhang theory of myers (1977) predicts that higher leverage increases the probability of the firm underinvesting which affects future earnings and result in lower stock prices. hence, an increase in leverage increases the risk of default which increases systemic risk as a whole. dimitrov & jain (2008) provide empirical evidence that increases in leverage is negatively related to deteriorating firm performance which impacts future stock prices negatively. the choice between the amount of debt and equity is clearly a source of risk. this risk emanates from the balance sheet and increases the default risk of firms as leverage increases. default by systemically important firms or institutions can then cause risk to become systemic, threatening the stability of an entire system. the importance of debt overhang was clearly illustrated in the global financial crisis when governments published by atlantis press copyright: the authors 20 v.b.y. gan struggled to make the right decisions on whether to use asset purchase or equity interventions to efficiently recapitalized highly leveraged banks that failed (philippon & schnabl, 2013). 2.2. the interconnectedness and interdependencies of firms the arguments for bank mergers to create more interconnected banks are similar to arguments for larger banks. however, recent evidence show that larger and more interconnected banks tend to concentrate systemic risk on their balance sheets instead of transferring it efficiently across the economy. for instance, nier, yang, yorulmazer & alentorn (2007) find that better capitalized banks support a more resilient system against contagion, and that small increases in interbank connectivity increases the contagion effect but only up to a certain threshold where connectivity actually increases system resilience. the larger the size of interbank exposures, the larger the risk of external shocks, and the more concentrated the system is, the levels of systemic risk is also higher. billio, lo, getmansky, & pelizzon (2012) propose two econometric measures of systemic risk that measures the degree of interconnectedness in a sample of monthly returns of hedge funds, banks, brokers, and insurance firms. the authors construct indexes to proxy for monthly returns of the four different financial sectors. they use principle component analysis to decompose the covariance matrix of the four returns index to capture any changes in commonality in asset returns. the second measure proposed is a granger causality network to measure the direction of the relationship between institutions. empirical findings reveal that systemic risk has been increasing since 1994, peaking in 1998, and another peak in 2008. this is consistent with the increasing interconnectivity among financial institutions. 2.3. correlations and commonality of asset returns acharya (2009) defines systemic risk as the risk of joint failures caused by the correlation of asset returns of bank balance sheets. systemic risk arises from the preferences banks have for highly correlated asset returns which manifest as aggregate risk. the author proposes a theory of systemic risk where banks have a systemic risk shifting incentive that depends on the health of other banks as failure acts as a negative externality. the important implication from this is that banks have the incentive to survive or fail together rather than using profits from healthy banks to bail out or subsidize the failure of others. this has important policy implications as regulatory mechanisms such as bank closure or capital adequacy may in fact aggravate systemic risk levels. ibragimov, jaffee & walden (2011) further explore this from a different angle. while intermediaries in acharya (2009) are exposed to correlated investments, they do not diversify or trade these risks as a part of risk sharing throughout the financial system, ibragimov et al (2011) models the risk portfolios of different intermediaries as independent and risk is traded or diversified. systemic risk arises from the trading of risk which leads to the interdependence of intermediaries created from institutions basically holding the same diversified portfolio. this diversification creates a negative externality that can shock all institutions simultaneously. de nicolo & kwast (2002) argue that firm interdependency as measured by the correlation of stock returns can provide an indicator of systemic risk potential. they measure total interdependency by the correlations of percentage changes in stock prices. an increase in stock returns therefore signifies an increase in the potential that a shock may turn systemic. not surprisingly, the authors find that there is a significant positive upward trend in stock return correlations which echoes the view that systemic risk has been rising during the period of study. they proceed to study the effect of firm consolidation on systemic risk and found evidence that consolidation increases interdependency and may increase systemic risk although other unidentified factors may explain the correlations. extending the work mentioned above, patro, qi & sun (2012) also study the stock return correlations as a measure of systemic risk. the authors extend earlier works by showing the difference between correlation patterns of financial and nonfinancial firms and also the correlations of defaults. the authors differentiate between systematic and idiosyncratic risks of stock returns. they show that stock return correlations increase when the correlations of idiosyncratic risk among banks also increase. therefore, systemic risk increases. their findings reveal that sharp correlation movements coincide or tend to predict significant financial events, especially during the recent global financial crisis. 2.4. behavioral channels and procyclicality the banking balance sheet acts as a transmission channel for systemic risk through the very actions of its own managers. the profit maximizing behaviors of managers can be shown to enhance procyclicality. typically, managers have the incentive to increase the size of the balance sheet. consider acharya & naqvi (2012) who propose a model of liquidity abundance at banks that aggravates the risk taking moral hazard published by atlantis press copyright: the authors 21 banking balance sheet channel which in turn leads to excessive lending and asset price bubbles as a result of excessive liquidity in the real economy. the authors propose a novel model that focuses on the actions of bank loan officers and how they are compensated. bank loan officers as the risk taker on behalf of the bank are compensated based on the volume of loans that are taken out by consumers. this is meant to induce greater effort by the loan officer against agency problems. however, the practice induces excessive risk taking as the loan officers are only penalized after the loans have been given out if the bank suffers a sufficiently high liquidity shortfall. when the general macro-environment becomes riskier, investors tend to hold on to more deposits and the banks become flushed with excess liquidity. this induces the bank to commit to excessive credit volume and the price of assets increases. shleifer & vishny (2010) within the context of originating and distributing securities by banks in financial markets propose that systemic risk is created due to the profit maximizing behavior of banks catering to investor demand during good times which lead to balance sheet expansion. this profitable expansion during good economic periods causes instability. these banks will have to liquidate their portfolio holding as fire sale prices which are below fundamental values in bad times lead to major downward revisions of security prices in a downward spiral. 3. balance sheet risk endogeneity and systemic risk we have thus far explored the literature and synthesized the main channels which the banking balance sheet channels systemic risk by providing theoretical arguments and empirical evidence from many endeavors of systemic risk measurement by many researchers. we now further argue that the main source of systemic lies endogenously within the procyclical nature of the banking balance sheet. in a stylized balance sheet, assets are financed by liabilities and equity. equity as the risk capital and liabilities posing liquidity and insolvency risks to the firm. hence, assets are practically financed by risk. the interplay or entanglement of the relationship between assets, liabilities and equity creates systemic endogenously from the balance sheet where each element carries its own risk. systemic arises and propagates through the balance sheet due to the procyclical nature of the balance sheet being tied to the business cycle. this is due to the procyclical nature of leverage and capital (brunnermeier & pedersen, 2009; drumond, 2009). in turn leverage and capital is tied to the expansion in good times and contraction of the balance sheet in bad times. figure 1 illustrates the expansion and contraction process. banks tend to decrease lending during recessions due to concerns about increased credit and default risk of borrowers which in turn exacerbates the bearish conditions of the economy leading to a credit crunch. during this period of downswing, balance sheets of banks and firms contract. banks and firms that have insufficient capital and cash will become illiquid as lenders scramble to call in loans and cease further lending. the risk of insolvency increase as firms struggle to generate sufficient earnings to finance long term survival. the risk of systemic insolvency and final collapse increases greatly with the downswing of the business cycle. systemic risk is therefore inherent in the balance sheet which is procyclical with the business cycle. reduced lending causes firms that are credit constrained to decrease economic activity and future investments while hoping to increase it when the economy recovers. this gush of optimism, cash, spending, lending and borrowing during recovery when the business cycle is on the upswing can potentially lead to an overheating economy. published by atlantis press copyright: the authors 22 v.b.y. gan endogenous risk as defined by benink, danielsson, & jonsson (2008: 91) is the phenomenon where there is a critical mass of agents who adjust their balance sheets in concert and aggregately exert significant asset price changes. endogenous risk exists in the banking balance sheet as a result of a feedback loop between rational agents and future asset prices. danielsson, shin & zigrand (2011) discuss how market participants resonate with the expectation of future asset prices and make concerted changes to their balance sheet that bring about realized volatility. cont & wagalath (2013) show the occurrence of fire sales as the manifestation of this negative feedback loop. this is somewhat similar to the shadow bank runs described by uhlig (2008) where all agents simultaneous discard assets at depressed prices to maintain capital ratio requirements. 4. potential management of endogenous risk and new methods in systemic risk measurement 4.1. how to better manage endogenous and systemic risk of the banking balance sheet? a way forward to manage these risks would be to design prudential policy and regulatory tools to address the two core dimensions of systemic risk; the time series dimension (procyclicality of risk) and the cross sectional dimension of risk. drumond (2009) and galati & moessner (2012) discuss how bank capital regulatory requirements such as those required by basel ii can actually accentuate or limit the procyclical nature of banking and its subsequent effect on the evolution of systemic risk. the basel ii capital adequacy ratios (car) were created to tie capital to the riskiness of underlying assets. the rationale being that capital is made sufficient to cover for asset shortfalls. the prime car is the tier 1 ratio defined as the sum of common stock, retained earnings, and preferred stock to risk weighted assets must be equal or more than 8%. however, cars prove to be the accentuating factor of procyclicality as an increase in car as required during crisis leads to a decline in lending which contracts the asset side of the balance sheet and ultimately leading to a credit crunch in the economy. therefore, to address the procyclical nature of systemic risk, regulatory instruments such as capital requirements that may actually increase systemic risk levels can be countered with countercyclical capital chargers (kashyap & stein, 2004). acharya et al (2010) even suggest that regulators should consider a taxation scheme to align the incentives of banks towards reducing their contribution to systemic risk. this tax will vary with the cost of raising capital and the cost of a government bailout. the idea is to encourage banks to seek equity funding and not depending on the government or central banks for last resort bailouts. fig. 1. expansion and contraction of the balance sheet. published by atlantis press copyright: the authors 23 banking balance sheet channel the general conclusion is that regulation on capital requirements and credit growth can prevent crises by improving the quality of a bank’s capital (laeven, 2011). lopezespinosa et al (2012) suggested that an optimal capital buffer structure be designed by weighing the relative importance of systemically important factors of banking systemic risk which are bank size, interconnectedness, substitutability, global activity and complexity against bank capital requirements. danielson (2002), however, holds a contrasting view. he argues that due to current limitations on technology at that time, systemic risk modeling is a futile venture as it is simply too unreliable, poorly defined and deeply affected by complicated moral hazard issues that is better to abolish minimum capital requirements and instead requiring banks to purchase insurance to cross insure each other against crisis which is consistent with acharya (2009)’s theory that banks unlike other industries have the incentive to fail or survive together rather than buying out failed competitors; banks have such correlated exposures and investments that no bank wants to use their internal profits to subsidize the failure of others. the focus on market values of equity or net worth as a key state variable alongside interest rates has a long and established tradition with the works of bernanke & gertler (1989) and kiyotaki & moore (1997). this mold was broken by geanakoplos (2010) and geanakoplos and fostel (2012) who have emphasized that leverage driven by collateral rates (haircuts) as being the key state variable in the global financial crisis. this indicates that a refocus towards book values of equity and leverage should be considered seriously. a key step to limiting balance sheet size through a leverage ratio backstop will be implemented in basel iii (bcbs, 2010: 60-61). however, its actual effectiveness is still speculative as the leverage ratio is not treated as a risk measure but only a backstop. we thus argue that insolvency risk using book value be measured as either the leverage ratio (total assets/ total equity) or the gearing ratio (total debt/ total equity). adrian, moench & shin (2013) provide strong empirical evidence that book leverage is in fact a key and relevant state variable. giving much credence to geanakoplos’s theory of the leverage cycle. furthermore, jarrow (2013) demonstrates that the theoretical foundations for the gearing ratio actually measures and controls for the similar insolvency risk as value-at-risk (var) but in a much simpler and easier implementation. bichsel & blum (2005) even provide evidence that the ‘crude’ leverage ratio can address the risk sensitiveness weakness of var.shleifer & vishny (2010) show that in originating and distributing securities, the profit maximizing behavior of banks give rise to systemic risk. this is because the risks of the assets involved were neglected and out of mind (gennaioli, shleifer & vishny, 2012). gennaioli, shleifer & vishny (2013) propose that to prevent risks from being neglected, balance sheet leverage control should be considered seriously by regulators. others like rochet (1992), and blum (2008) have argued that a leverage ratio acting as a capital floor is required for bank stability. these authors provide strong arguments to return to the basics of financial ratios as the first basis for risk measurement which should be considered seriously to manage endogenous risk especially by limiting the size and exposures of the banking balance sheet to systemic risk. 4.2. difficulties in measuring endogenous risk from balance sheet data the easiest and most direct method to measure risk from balance sheet data is to use ratios. despite the large family of sophisticated statistical methods of measurement, benos & papanastasopoulos (2007) show that financial ratios are still relevant today as they contain significant and incremental information. they show that a hybrid model of ratios and contingent claims outperforms the merton (1974) credit risk model and the zscore in bankruptcy prediction. financial ratios and ratings are still used widely to measure certain risks inherent in the balance sheet: liquidity risk, credit risk, leverage risk, and insolvency risk among the more common risks. the most common ratios used to assess financial condition are the leverage ratio, gearing ratio and the quick ratio or the current ratio. the inherent nature of risk in the balance sheet was formally exploited by altman (1968) in a seminal work that used ratios to predict corporate bankruptcy. unfortunately, financial ratios are characterized by a lack of normality (horrigan, 1965). the components of ratios themselves were found to be generally non proportionate and nonlinear (sudarsanam & taffler, 1995). barnes (1982) showed that the non proportionate nature of ratios leads to skewed and non normal distribution. in fact, financial ratios have been found to be heavy tailed with infinite variances related to powerlaw distributions or also known as the asymmetric levy probability density function (podobnik, valentincic, horvatic & stanley, 2011). in general, ratio analysis relies on measures of central tendencies by using the mean or median as the benchmark. however, there are no parametric tests to compare the medians of two samples. hence, relying of the testing of means, it is common to truncate outliers for analysis (bougen & drury, 1975; deakin, 1976). commonsize ratio analysis assumes that by ratio analysis the differences across firms and industries especially for size can be controlled. however, the published by atlantis press copyright: the authors 24 v.b.y. gan asymmetric and disproportionate functional form negates this size control motive as ratios are not stable across size. this is due to the denominator having a disproportionate effect on the final ratio. any equal changes in the numerator and the denominator will not lead to equally proportionate changes in the ratio. to achieve size control, the numerator must be in strict proportionality with the denominator. typically, values of ratios considered to be outliers are trimmed or windsorized. these outliers usually occur when the denominator is close to zero: when earnings are almost nil due to bankruptcy or market value of equity when stock prices crash. the use of trimming and windsorizing is motivated to control for industry wide factors which is similar in motive to control for size as industry factors can also range from the very large to the very small. however, to achieve control, assumptions of strict proportionality must be adhered to, where λ is constant no matter the change in the values of y and x:  x y or y= λx (1) however, in practice, strict proportionality is frequently violated by the presence of an intercept, error term and the dependence of y on other variables and the non linearity of the relationship. for example, in the presence of an intercept: y= λx + k or x k x y   (2) this deviation occurs when there are very small firms, the value of x will be small enough for the intercept, k, to impact the relationship between x and y but with a bias toward x which in this case are the small firms. hence, strict proportionality is not followed and results of any analysis could be bias and skewed (lev & sunder, 1979). this is because ratios are similar to scales but with arbitrary proportions. when a ratio is then used for scaling purposes to show how variables change in relation with each other, the ratio of x y should be constant, linear, with a zero intercept and scale invariant so that relative changes are similar at all scales. for example, in the cross section, if the value of debt is larger in one firm, then equity should be larger as well, ceteris paribus. similarly in the time series, if debt should grow by 5% in the preceding year, then equity should grow by 5% this year, ceteris paribus. hence, without scale invariance, a reliable conclusion cannot be reached. 4.3. potential new methods to measure systemic risk an alternative but growing class of literature on the measurement of systemic risk is derived from the application of theories and methods from statistical physics in economic and financial phenomenon. mantegna & stanley (2000) and sornette (2003) crystalizes much of the thrust and reasoning of how principles and methods from physics can be applied in a rich variety of financial systems. an interesting feature of these methods is that they attempt to characterize the financial system as a whole complex adaptive system without being bounded to gaussian and stationarity assumptions. these methods make no distributional assumptions and hence do not require trimming, truncating or windsorizing of outliers. as argued by stanley, gabaix, gopikrishnan, & plerou (2007), outliers do not exist. these methods may prove immensely useful in using balance sheet risk ratios while overcoming the problems mentioned above. in a series of papers, wang, yamasaki, havlin, & stanley (2006), wang, yamasaki, havlin, & stanley (2008), wang, shieh, havlin & stanley (2009a) and wang, yamasaki, havlin & stanley (2009b) study the properties of stock returns using high frequency data of all stock returns for all securities listed on the new york stock exchange (nyse). the key difference from conventional finance studies on stock returns is that these papers atomize stock returns to the smallest unit of measurement possible; they use a sampling time of just one minute. however, wang et al (2009a) study overnight and daytime returns and hence use daily opening and closing prices for all stocks listed on the nyse on december 31, 2007. the key finding in studying stock price volatility is that follows a power law distribution in its tails and return intervals display scaling and memory of past movements. analogous to climate and earthquake data, systemic risk as the occurrence of a rare event lying in the heavy tails can be quantified from power law distributions. in fact, power law distributions seem to be a natural order. podobnik et al (2011) show that the distribution of financial ratios and even altman’s (1968) z-score is characterized by power laws and scaling. gabaix et al (2006) further explains that even the distribution of city sizes (described by zipf’s law which is also a power law), stock returns, trading volume, price impact and the size of large institutional investors can all be described by a natural power law. this suggests that future development of endogenous and systemic risk metrics should be based on the natural distribution of balance sheet variables at book values. caetano & yoneyama (2009) propose a novel method to detect the occurrence of an imminent stock published by atlantis press copyright: the authors 25 banking balance sheet channel market drawdown. using a wavelet decomposition method to detect abrupt changes in a time series of stock market indices; the hong kong hang seng and the brazilian ibovespa covering the pre and post-crash of 1929 and the recent 2008 market crashes, the behavior of wavelet coefficients was found to be provide useful insights on the probability of a future drawdown. aggregating the information provided by the coefficients to create an index which showed good capabilities of monitoring crashes and drawdowns. caetano & yoneyama (2011) yet again propose a novel measure of systemic risk by modeling the hong kong hang seng, u.s. dow jones and the brazilian ibovespa index from 1993 to 2007 as a catalytic chemical reaction. the measure of risk is the degree of influence of one index on the other. the authors show how a strong market represented by the dow jones as the reagent with high concentration in a chemical process can influence the behavior of lesser markets represented by the hang seng and ibovespa. performing 200 monte carlo simulations, they calculate the var for each market and show that the larger market does significantly influence the dynamics of smaller markets. the efficient market hypothesis (emh) has been much criticized for its unrealistic assumptions. the fractional market hypothesis (fmh) was then put forward to address these deficiencies; it assumes non gaussian statistics, it is a non-stationary process, it allows historical correlations, it allows scaling and its fundamental continuous unstable nature at any scale. blackledge (2010) propose the use of a non-stationary fractional dynamic stochastic model of economic signals to assess systemic risk. he proposes to use the time varying fourier dimension of the fractional diffusion equation as the measure of market volatility. as the value of the fourier dimension continues to increase in a time series, the probability of volatile market behavior increases. performing a case study on the subprime credit default swap abx index from july 24, 2006 to april 2, 2009, the model is able to show that the index exhibits a clear phase transition period or criticality which preceded the crash of 2008. alternative representations of price bubbles were proposed by jiang, zhou, sornette, woodard, bastiaensen & cauwels (2010). studying market crashes as analogous to earthquakes, the authors posit that similar to other large complex dynamic and nonlinear systems, stock market crashes are caused by the slow accumulation of long range correlations that lead to a collapse in one critical moment. the challenge is to capture this self -critical instant and describe its behavior before and after the crash. building on previous works; sornette & johansen (1997) and johansen, ledoit & sornette (2000) among others, they propose a logperiodic power law (lppl) model that models volatility as oscillations of the system. the lppl specification is given by:    ln cos lnm mc c cr t a b t t c t t t t        (3) where tc is the most probable time of crash, r the stock returns and m,  ,  , a, b and c are parameters to be determined by sum of squares minimization.the measure of a bubble is the faster than exponential rate of increase in asset prices driven by accelerating oscillations. the model is designed to capture the many human facets in crashes; the positive feedback loop of higher return expectations of participants and the negative feedback loop of crash expectations. testing the model on the shanghai and shenzhen stock index between may 2005 and july 2009, the model showed how the stock market evolved into critical states twice from the middle of 2005 and november 2008 and was able to predict the bursting of the price bubble twice in october 2007 and august 2009. interestingly, gnacinski & makowiec (2004) showed that in addition to two types of known bubbles already described by earlier works above; the first type is where log period oscillations drive the bubble and ends in a crash and the second type where oscillations appear as an antibubble after the crash to further push prices down. there is a third bubble called the inverted bubble where after draw downs had occurred, extraordinary draw-ups occurred after the logperiodic behavior had ended. a promising area we propose is the study of bubbles. usually price bubbles are studied. we would suggest that the size of the balance sheet be measured as a bubble. a faster than exponential rate of balance sheet size expansion based on book values can be a promising measure of endogenous risk and thus reflect the magnitude of systemic risk. the lppl methodology discussed above could yield a potential method to measure both endogenous and systemic risk as the lppl specification takes into account rational expectations, herding potential, and process of bifurcations and phase transitions. yan, woodard & sornette (2012) use repo data to study the behavior of leverage bubbles with this method. further research down this methodological line with book value balance sheet data could be promising indeed. 5. concluding remarks in this paper, we explored the possible channels in which the banking balance sheet transmits and serves as a reservoir of systemic risk. we achieve this by integrating the theoretical literature and the systemic risk measurement literature. our synthesis of theory and empirical evidence from measuring systemic risk reveal published by atlantis press copyright: the authors 26 v.b.y. gan an endogenous origin of systemic risk within the banking balance sheet. conceptually this risk is known as endogenous risk. this paper shows that this is the true origin of systemic risk where aggregate adjustment by market agents to their balance sheet affects real prices and realized volatility. overall, we find that the literature identifies four banking balance sheet channels of systemic risk: (i) bank capital structure choice, (ii) interconnectedness and interdependencies among firms, (iii) correlations of asset composition and returns and; (iv) behavioral determinants that affect the choices of bank managers. furthermore, we find that the root source of these channels is the endogenous risk and procyclicality of the balance sheet. these four channels can be framed within the context of endogenous risk. as individual economic agents respond to the external and internal banking ecosystem, they make choices in choosing the level of equity and debt to finance the acquisition of assets with the accounting identity as the ultimate binding constraint. banks tend to finance similar assets as they have strong incentives to thrive or fail together. thus, they make much correlated decisions leading to a very high degree of interconnectedness and correlated asset structures. aided by incentives at the ground level, managers and other officers have strong incentives to make easy loans and neglect risks. on the aggregate, when all banks act similarly simultaneously, it creates endogenous risk which exacerbates procyclical tendencies. when forced to deleverage quickly, these aggregate balance sheet adjustments create a credit crunch sending a strong shock to the real economy and asset prices as banks race to the bottom to shed assets on their balance sheet in a fire sale. in our integrating review, we find a strong case to return to a back to basics approach of relying more on book value balance sheet data. geanakoplos (2010) and geanakoplos & fostel (2012) provide the theoretical foundations for leverage to be considered a key state variable. adrian et al (2013) even provide strong evidence that the leverage ratio as measured bytotal assets/ total book equity is in fact the most relevant state variable for asset pricing, not market equity prices. others like adrian & brunnermeier (2010), vallascas & keasey (2012), dejonghe (2010) have also provided evidence that book leverage were key determinants of systemic risk. this calls for new approaches to management and measurement of both endogenous and systemic risk. finally, we present some new approaches to systemic risk measurement. these new approaches suggest that any new metric based on the natural distribution of book value balance sheet variables to measure and manage endogenous and systemic risk should be robust to non-normality, non-linearities, nonstationarity, autocorrelation, and model risk. it should be dynamic to reflect how decisions regarding risks are made in real time and be invariant to scale and time while possessing the scaling property that allows for long range memory. scaling is an intuitive property simply because people make decisions that affect the behavior of prices and risk metrics. in real life, past performance is salient to the decision maker and leads to mental heuristics biases that have a strong bearing for decision making under risky conditions (bordalo, 2012). an even further rethinking of the definition of risk should be considered. since risk is “fuzzy” in nature, the use of fuzzy sets may be fruitful. in particular, a rethinking of risk proposed by huang & duan (2008) suggests that we should view risk not just defined by loss but by a larger context of an adverse event. this we leave to future research. 6. acknowledgements vincent gan thanks the ministry of higher education, malaysia for partial financial support during the course of his phd candidature at putra business school, universiti putra malaysia. references acharya, v. v. (2009). a theory of systemic risk and the design of prudential bank regulation. journal of financial stability, 5, 224-255. acharya, v. v., schnabl, p., & suarez, g. (2013). securitization without risk transfer. journal of financial economics, 107(3), 515-536. acharya, v., & naqvi, h. (2012). the seeds of a crisis: a theory of bank liquidity and risktaking over the business cycle. journal of financial economics, http://dx.doi.org/10.1016/j.jfineco.2012.05.014. acharya, v., pedersen, l., philippon, t., & richardson, m. (2010). measuring systemic risk. working paper, new york university. adrian, t., & brunnermeier, m. k. (2010). covar. staff report 348,federal reserve bank of new york. adrian, t., moench, e., & shin, h. s. (2013). leverage asset pricing. working paper, princeton university. allen, f., babus, a., & carletti, e. (2009). financial crises: theory and evidence. annual review of financial economics, 1, 97-116. altman, e. i. (1968). financial ratios, discriminant analysis and the prediction of corporate bankruptcy. journal of finance, 23(4), 589-609. amihud, y., delong, g. l., & saunders, a. (2002). the effects of cross-border bank mergers on bank risk and value. journal of international money and finance, 21(6), 857-877. published by atlantis press copyright: the authors 27 banking balance sheet channel barnes, p. (1982). methodological implications of non normally distributed financial ratios. journal of business finance & accounting, 9(1), 51-62. bcbs. (2010). basel iii: international framework for liquidity risk measurement, standards, and monitoring. basel committee on banking supervision. bank for international settlements (bis). benink, h., danielsson, j., & jonsson, a. (2008). on the role regulatory banking capital. financial markets, instituitions, and instruments, 17(1), 85-96. benos, a., & papanastasopoulos, g. (2007). extending the merton model: a hybrid approach to assesing credit quality. mathematical and computer modelling, 46, 4768. bernanke, b., & gertler, m. (1989). agency costs, net worth, and business fluctuations. american economic review, 79(1), 14-31. bianchi, j. (2011). overborrowing and systemic externalities in the business cycle. american economic review, 101(7), 3400-3426. biasis, d., flood, m., lo, a. w., & valavanis, s. (2012). a survey of systemic risk analytics. annual review of financial economics, 4, 255-296. bichsel, r., & blum, j. (2005). capital regulation of banks: where do we stand and where are we going? swiss national bank. quarterly bulettin, 23, 42-51. billio, m., getmansky, m., lo, a. w., & pelizzon, l. (2012). econometric measures of connectedness and systemic risk in the finance and insurance sectors. journal of financial economics, 104, 535-559. blackledge, j. (2010). systemic risk assessment using a non stationary fractional dynamic stochastic model for the analysis of economic signals. isast transactions on computers and intelligent systems, 2(1), 76-94. blum, j. (2008). why 'basel ii' may need a leverage ratio restriction. journal of banking & finance, 32, 1699-1707. bordalo, p., gennaioli, n., & shleifer, a. (2012). salience theory of choice under risk. quarterly journal of economics,doi:10.1093/qje/qjs018, 1243–1285. bougen, p. d., & drury, j. c. (1975). u.k. statistical distributions of financial ratios. journal of business finance & accounting, 7(1), 39-47. brunnermeier, m. k., & pedersen, l. h. (2009). market liquidity and funding liquidity. review of financial studies, 22(6), 2201-2238. brunnermeier, m. k., dong, g., & palia, d. (2012). bank's noninterest income and systemic risks. princeton university working paper. caetano, m. a. l., & yoneyama, t. (2009). a new indicator of imminent occurence of dradown in the stock market. physica a, 388, 3563-3571. caetano, m. a. l., & yoneyama, t. (2011). a model for the evaluation of systemic risk in stock markets. physica a, 390, 2368-2374. cetorelli, n., & goldberg, l. s. (2012). banking globalization and monetary transmission. journal of finance, 67(6), 1811-1843. cont, r., & wagalath, l. (2013). fire sales forensics: measuring endogenous risk. working paper no. 2013fin-01,ieseg school of management. danielsson, j. (2002). the emperor has no clothes: limits to risk modelling. journal of banking & finance, 26, 12731296. danielsson, j. (2008). blame the models. journal of financial stability, 4, 321-328. danielsson, j., shin, h. s., & zigrand, j.-p. (2011). endogenous and systemic risk. working paper, london school of economics, retrieved from www.riskresearch.org, accessed on 17 june 2012. de bandt, o., & hartmann, p. (2000). systemic risk: a survey. working paper no.35, european central bank, working paper series. de jonghe, o. (2010). back to basics in banking? a micro analysis of banking system stability. journal of financial intermediation, 19, 387-419. de nicolo, g., & kwast, m. l. (2002). systemic risk and the financial consolidation: are they related? journal of banking & finance, 26, 861-880. deakin, e. b. (1976). distributions of financial accounting ratios: some empirical evidence. the accounting review, 51(1), 90-96. deyoung, r., & torna, g. (2013). nontraditional banking activities and bank failures during the financial crisis. journal of financial intermediation, 22, 397-421. diamond, d. w., & dybvig, p. h. (1983). bank runs, deposit insurance and liquidity. journal of political economy, 91(3), 401-419. diamond, d. w., & rajan, r. g. (2000). a theory of bank capital. journal of finance, 55(6), 2431-2465. diamond, d. w., & rajan, r. g. (2001). liquidity risk, liquidity creation, and financial fragility: a theory of banking. journal of political economy, 109(2), 287-327. diamond, d. w., & rajan, r. g. (2005). liquidity shortages and banking crises. journal of finance, 60(2), 615-647. dimitrov, v., & jain, p. c. (2008). the value relevance of changes in financial leverage beyond growth in assets and gaap earnings. journal of accounting, auditing & finance, 23(2), 191-222. drehmann, m., & nikolaou, k. (2012). funding liquidity risk: definition and measurement. journal of banking and finance, doi: 10.1016/j.jbankfin.2012.01.002. drumond, i. (2009). bank capital requirements, business cycle fluctuations and the basel accords: a synthesis. journal of economic surveys, 23(5), 798-830. galati, g., & moessner, r. (2012). macroprudential policya literature review. journal of economic surveys, doi: 10.1111/j.1467-6419.2012.00729.x. geanakoplos, j. (2010). the leverage cycle. cowles foundation discussion paper no.1715r. gennaioli, n., shleifer, a., & vishny, r. (2012). neglected risks, financial innovation, and financial fragility. journal of financial economics, 104, 452-468. published by atlantis press copyright: the authors 28 v.b.y. gan gennaioli, n., shleifer, a., & vishny, r. w. (2013). a model of shadow banking. journal of finance, 68(4), 13311363. gnacinski, p., & makowiec, d. (2004). another type of log periodic oscillations on polish stock market. physica a, 344, 322-325. gorton, g., & metrick, a. (2012). securitized banking and the run on the repo. journal of financial economics, 104, 425-451. hansen, l. p., & sargent, t. j. (2011). robustness and ambiguity in continuous time. journal of economic theory, 146, 1195–1223. hellwig, m. f. (2009). systemic risk in the financial sector: an analysis of the subprime-mortgage financial crisis. de economist, 157(2), 129-207. horrigan, j. o. (1965). some empirical bases of financial ratio analysis. the accounting review, 40(3), 558-568. huang, c., & ruan, d. (2008).fuzzy risks and and an updating algorithm with new observations. risk analysis, 28(3),681-694. ibragimov, r., jaffee, d., & walden, j. (2011). diversification disasters. journal of financial economics, 99, 333-348. international monetary fund. (2008a). containing systemic risks and restoring financial soundness, global financial stability report. april 2008, http://www.imf.org/external/pubs/ft/gfsr/2008/01/inde x.htm. international monetary fund. (2008b). financial stress and deleveraging: macro-financial implications and policy, global financial stability report. october 2008, http://www.imf.org/external/pubs/ft/gfsr/2008/02/inde x.htm. jarrow, r. (2013). a leverage ratio rule for capital adequacy. journal of banking & finance, 37, 973-976. jiang, z. q., zhou, w. x., sornette, d., woodard, r., bastiaensen, k., & cauwels, p. (2010). bubble diagnosis and the prediction of the 2005-2007 and 2008-2009 chinese stock market bubbles. journal of economic behavior & organization, 74, 149-162. johansen, a., ledoit, o., & sornette, d. (2000). crashes as critical points. international journal of theoretical and applied finance, 3(2), 219-255. jotikasthira, c., lundblad, c., & ramadorai, t. (2012). asset fire sales and purchases and the international transmissions of funding shocks. journal of finance, 67(6), 2015-2050. kashyap, a., & stein, j. (2004). cyclical implications of the basel-ii capital standards. federal reserve bank of chicago economic perspective, 28 (q1), 18-31. keen, s. (1995). finance and economic breakdown: modelling minsky's "financial instability hypothesis". journal of post keynesian economics, 17(4), 607-635. keen, s. (2013). a monetary minsky model of the great moderation and the great recession. journal of economic behavior and organization, 86, 221-235. khwaja, a. i., & mian, a. (2008). tracing the impact of bank liquidity shocks: evidence from an emerging market. the american economic review, 98(4), 1413-1442. kiyotaki, n., & moore, j. (1997). credit cycles. journal of political economy, 105(2), 211-248. krause, a., & giansante, s. (2012). interbank lending and the spread of bank failures: a network model of systemic risk. journal of economic behavior & organization, http://dx.doi.org/10.1016/j.jebo.2012.05.015. laeven, l. (2011). banking crises: a review. annual review of financial economics, 3, 4.1-4.24. lev, b., & sunder, s. (1979). methodological issues in the use of financial ratios. journal of accounting and economics, 1, 187-210. lopez-espinosa, g., moreno, a., rubia, a., & valderrama, l. (2012). shortterm wholseale funding and systemic risk: a global covar approach. journal of banking & finance, http://dx.doi.org/10.1016/j.jbankfin.2012.04.020. maino, r., & tintchev, k. (2012). from stress to costress: stress testing interconnected banking systems. imf working paper, wp/12/53. mantegna, r. n., & stanley, h. e. (2000). an introduction to econophysics: correlations and complexity in finance. cambridge, united kingdom: cambridge university press. merton, r. (1974). on the pricing of corporate debt: the risk structure of interest rates. journal of finance, 29, 449470. minsky, h. p. (1977). the financial instability hypothesis: an interpretation of keynes and an alternative to"standard" theory. nebraska journal of economics and business, 16(1), 5-16. minsky, h. p. (1992). the financial instability hypothesis. working paper no.74, the jerome levy economics institute of bard college. myers, s. c. (1977). determinants of corporate borrowing. journal of financial economics, 5, 147-175. nier, e., yang, j., yorulmazer, t., & alentorn, a. (2007). network models and financial stability. journal of economic dynamics & control, 31, 2033-2060. papanikolaou, n. i., & wolff, c. c. (2013). the role of on and offbalance-sheet leverage of banks in the late 2000s crisis. journal of financial stability. doi:http://dx.doi.org/doi:10.1016/j.jfs.2013.12.003 patro, d. k., qi, m., & sun, x. (2012). a simple indicator of systemic risk. journal of financial stability, doi:10.1016/j.jfs.2012.03.002. philippon, t., & schnabl, p. (2013). efficient recapitalization. journal of finance, 68(1), 1-42. podobnik, b., valentincic, a., horvatic, d., & stanley, h. e. (2011). asymmetric levy flight in financial ratios. pnas, 108(44), 17883-17888. rochet, j.-c. (1992). capital requirements and the behavior of commercial banks. european economic review, 36, 1137-1178. published by atlantis press copyright: the authors 29 banking balance sheet channel schnabl, p. (2012). the international transmission of bank liquidity shocks: evidence from an emerging market. journal of finance, 67(3), 897-932. scholes, m. (1996). global financial markets, derivative securities, and systemic risks. journal of risk and uncertainty, 12, 271-286. shleifer, a., & vishny, r. w. (2010). unstable banking. journal of financial economics, 97, 306-318. sornette, d. (2003). why stock markets crash: critical events in complex financial systems. new jersey: princeton university press. sornette, d., & johansen, a. (1997). large financial crashes. physica a, 245, 411-422. stanley, h. e., gabaix, x., gopikrishnan, p., & plerou, v. (2007). economic fluctuations and statistical physics: quantifying extremely rare and less rare events in finance. physica a, 382, 286-301. sudarsanam, p. s., & taffler, r. j. (1995). finanical ratio proportionality and inter-temporal stability: an empirical analysis. journal of banking & finance, 19(1), 45-60. uhlig, h. (2010). a model of a systemic bank run. journal of monetary economics, 57, 78-96. vallascas, f., & keasey, k. (2012). bank relience to systemic shocks and the stability of banking systems: small is beautiful. journal of international money and finance, doi:10.1016/j.jimonfin.2012.03.011. vennet, r. v. (1996). the effect of mergers and acquisitions on the efficiency and profitability of ec credit institutions. journal of banking & finance, 20(9), 1531-1558. wang, f., shieh, s. j., havlin, s., & stanley, h. e. (2009a). statistical analysis of the overnight and daytime return. physical review e, 79(056109). wang, f., yamasaki, k., havlin, s., & stanley, h. e. (2006). scaling and memory of intraday volatility return intervals in stock markets. physical review e, 73(026117). wang, f., yamasaki, k., havlin, s., & stanley, h. e. (2008). indication of multiscaling in the volatility return intervals of stock markets. physical review e, 77(016109). wang, f., yamasaki, k., havlin, s., & stanley, h. e. (2009b). multifactor analysis of multiscaling in volatily return intervals. physical review e, 79(016103). woodford, m. (2003). interest and prices:foundations of a theory of monetary policy. princeton university press. yan, w., woodard, r., & sornette, d. (2012). leverage bubble. physica a, 391, 180-186. published by atlantis press copyright: the authors 30 基于随机赋权法和云模型的黄河流域水安全风险评估 received 30 june 2015 accepted 5 august 2015 a study on spatial-temporal rainstorm risk at civil airports in china xiaomei guo 1,2 , xiaobing hu 1,2 , hang li 1,2 , zhen xu 1 1academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china 2 state key laboratory of earth surface processes and resource, beijing normal university, beijing 100875, china abstract based on daily precipitation data of 174 meteorological stations near civil airports in china, flight data over the same period (1994-2013) and other relevant airport data, this paper carries out a study on risk analysis of rainstorm at chinese civil airports. we use the platform of arcgis, analyze the four essential factors of disasters, i.e., rainstorm hazard, vulnerability, exposure and the capability of disaster prevention and mitigation, combine the traditional model of meteorological disaster risk assessment and characteristics of chinese civil aviation, and then develop monthly rainstorm disaster risk zoning maps for civil airports in china. the results can provide useful information to daily operations of civil aviation of china, as well as to planning and management of civil aviation systems, in order to better mitigate the impact of heavy rains on daily functions of civil aviation of china. keywords: civil airport; rainstorm disaster; risk assessment; risk zoning; spatial-temporal analysis 中国机场暴雨灾害风险时空分析 郭晓梅 1,2，胡小兵 1,2，李航 1,2，许珍 1 1. 减灾与应急管理研究院，北京师范大学，北京 100875, 中国 2. 地表过程与资源生态国家重点实验室，北京师范大学，北京 100875, 中国 摘要：基于机场附近 174 个气象站点 1994-2013 年的逐日降水数据、同期实时航班数据和机场吞吐量等相 关资料，本文展开中国 174 个机场暴雨灾害风险的初步研究。在 arcgis 平台上，对中国机场的暴雨危险性、 机场脆弱性、暴露性和防灾减灾能力等 4 方面进行分析，结合传统气象灾害风险评估模型和中国民用航空 特点，得到中国机场不同月份暴雨灾害风险等级区划图。结果可为中国民航日常运营提供依据，以指导民 航的规划和建设运营，最大限度降低暴雨带来的灾害损失。 关键词：中国机场; 暴雨灾害; 风险评估; 风险区划; 时空分析 1. 引言 在全球气候变化的大背景下，气象灾害风险大  大增加。ipcc 指出，全球自然灾害中 80%以上是气 象灾害或者其衍生灾害 [1] 。我国近 20 年（1994-2013 年）气象灾害灾情分析表明，死亡人口和直接经济 损失占所有自然灾害的比例多年平均值分别为 55%  作者简介：郭晓梅（1990-），女，硕士，研究方向为自然 灾害. e-mail：201321480021@bnu.edu.cn 和 87% [2] 。全球灾害数据库（em-dat）统计表明， 降水，洪涝，台风等灾害是中国主要的气象灾害， 而且呈上升趋势。随着中国经济快速发展，民用航 空运输业也日新月异，根据 2013 年中国民航统计公 报显示：2013 年全行业完成运输总周转量 671.72 亿吨公里，比增长 10.9%；全国民航运输机场完成 旅客吞吐量 7.54 亿人次，固定资产投资总额达 1452.2 亿元。气象灾害已经成为影响民航可持续发 展的重要因素，暴雨，台风，雪灾，沙尘暴等灾害 journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 188-198 published by atlantis press copyright: the authors 188 x.m. guo et al 天气都成为中国民航主要的致灾因子，结果轻则是 直接导致机场和航路关闭，并可能造成机场和航路 导航设备的损坏；严重则造成大面积航班延误，甚 至引发灾难性的航空事故。例如，2008 年我国南方 遭遇百年罕见的雪灾，造成全国共 30 多个机场间 断性关闭，航班大面积延误、停飞，导致近千万旅 客滞留机场；2000 年武汉空难，机组、旅客共遇难 51 人，原因是飞机在盘旋过程中可能遭遇雷击或者 风切变；2002 年 4 月 15 日，中国国航由于大暴雨， 加上操作不当，发生“金釜山空难”，遇难 128 人。 我国灾害性天气具有全天候、全年性和突发性等特 征，而且地方性特点明显 [3-4] 。春季多风、西北地区 常有扬沙，夏季东部地区多台风和雷暴，秋季有雷 暴，冬季北部和西部多大风和降雪，其中大风、大 雾、降雪、扬沙、雷暴等恶劣天气条件严重影响航 班的准点率和飞机飞行安全。根据美国航空数据网 站 flightstats 公布的数据 [5] 2014 年 10 月份中 国的航空公司的准点率最高的是中国国航，准点率 73.3%，南方航空，东方航空的准点率则更低，为 68%左右。根据中国民航局统计数据显示，2013 年 全国航空公司共执行航班 278.0 万班次，不正常航 班占 76.9 万班次，其中天气原因 21.8%。航班延误， 取消等不正常航班给旅客出行等带来重大影响， 2013 年收到的旅客投诉同比 2012 年增加 13.66%， 甚至由于航班延误而引发旅客打砸机场的事件。气 象灾害已对民航的发展构成严重威胁。民航气象是 航空业正常运行的保障 [6] ，全面地、系统地进行民 航气象灾害风险研究对我国民用航空业的可持续发 展及安全运营具有巨大的战略意义。 灾害学作为一门交叉学科， 20 世纪伴随着保 险业发展在一些发达国家迅速发展。直到 60 年代， 自然灾害的研究仍然主要限于灾害机理以及灾害预 测，侧重于灾害形成条件的调查和活动过程的考察。 一些较发达国家在 70 年代开始进行灾害风险的评 估工作。美国内务部的调查所等相关部门对加利福 尼亚、旧金山等地区的滑坡、地震等 10 种自然灾害 进行了风险评价与损失评估，其中包括飓风、龙卷 风、风暴潮等气象灾害风险评估 [7] 。lavalle(2005) 提出 jrc 风险评估法，对欧洲 10 个国家进行了天 气灾害的综合风险评估 [8] 。我国的自然灾害评价工 作起步较晚，90 年代后，台风、沙尘暴等气象灾害 的风险评估才开始受到较多重视。气象灾害风险研 究则在某些具体行业中的开展了应用研究。在对天 气、气候变化最为敏感的农业研究的文献较多，如： 李世奎等较早提出了较完善的农业气象灾害风险区 划指标体系和农业气象灾害风险评价模型 [9-10] ；高 晓蓉等基于气象灾害四要素理论，对东北地区玉米 的主要气象灾害风险评价模型进行了研究 [11] ；于飞 等基于信息扩散理论对贵州夏旱、倒春寒等八种主 要气象灾害进行不确定风险分析 [12] 。基础设施中与 气象密切相关的行业的气象灾害风险评价也有了初 步的探索。如：武永峰等基于地 gis 技术和灰色综 合评价法，建立了公路交通暴雨风险评价模型，对 湖南省公路的暴雨风险进行了定量评估 [13] ；付桂琴、 张文宗根据 1983-2008 年河北气象灾害灾情数据， 将河北电网气象灾害风险分高、中、低三个等级 [14] 。 天气原因的航班延误、暴雨灾害对民航的重大影响 引起人们的广泛关注，但民航灾害风险相关的研究 仍然较少。janic(2000)以火山灰为例 [15] ，对欧洲民 航进行了风险安全评估；李铭等基于民航雷达站的 雷电定位数据，结合地形地貌特征进行了雷击风险 区划 [16] ；张甦等 [17] 分析了民航广汉机场的气象能见 度特征。民用航空业的气象灾害风险研究，目前仅 限于民航灾害风险分析和气象预报预警阶段 [18-20] ， 民航气象灾害风险评价理论和方法模型仍在探索中。 传统的灾害学研究比较偏重于巨灾大灾对物理 承灾体造成的损失，例如，地震台风造成机场的基 础设施受到破坏。这种偏重于巨灾大灾和物理损失 的灾害研究思路对民航业的安全建设当然非常重要， 但却不易受到民航相关人员的认可和重视，因为民 航工作人员和旅客所重点关注的为更经常发生的 （相对于巨灾大灾）恶劣天气事件对民航运输服务 功能的影响。例如，硬件条件差的地方小机场，虽 然对巨灾大灾的脆弱性很高，但其服务功能方面的 脆弱性未必就高（因为旅客流量小，很容易疏导， 从而暴露度小）；而硬件条件最好的国际枢纽大机场， 虽然对巨灾大灾的物理抵抗能力很强，可是一场暴 雨却可能导致大规模的服务功能混乱。因此，要开 展能得到民航行业认可的灾害风险研究，就必须转 变思路，从传统的偏重于巨灾大灾和物理损失，到 强调和研究日常性恶劣天气事件所造成的服务功能 损失。如何才是好的转变思路的研究方法，本文就 是要进行这样的尝试。本文在现有气象灾害风险评 价与灾害综合风险管理相关理论研究的基础上，充 分考虑民用航空的行业特点，对我国主要机场的暴 雨灾害展开定量的、偏重于功能损失的风险评估。 在全球民航快速发展，气候变化加剧的复杂背景下， 本文以暴雨灾害为例，对民航气象灾害综合风险评 估模型进行初步探索，充分考虑机场服务功能脆弱 published by atlantis press copyright: the authors 189 性、暴露度和防灾减灾能力。风险值与气象指标相 比更加直观且更具建设性，可以为民航相关行业决 策和民众出行提供依据。风险模型随时利用机场天 气情况计算出不同区域的灾害风险值，结合当今发 达的网络平台，实时更新；而中长期的气象灾害风 险等级可以为机场建设标准提供参考依据。航空公 司可以根据短期的气象灾害风险区划慎重地调整和 安排不同地区的飞机类型和数目，以降低安全隐患 和节省资金。游客也可以根据风险等级选择出行与 否，或者是否买保险等。文章第一部分为引言；第 二部分介绍了使用的资料和方法；第三部分，以月 为单位，分析与计算得到不同月份中国机场的暴雨 危险性、机场暴露度、脆弱性和防灾减灾能力，并 且基于arcgis平台，对暴雨灾害四要素进行区划； 第四部分，结合传统气象灾害风险评估模型计算得 到中国174个机场的暴雨灾害综合风险值和区划图； 第五、六部分进行了总结和讨论。 2. 资料和方法 2.1. 资料 1) 灾情资料：根据《中国气象灾害大典：综合卷》 收集了全国 2005-年暴雨、洪涝、沙尘暴等气象 灾害。 2) 暴雨资料：选取了全国 174 个机场附近的气象 站 1994-2013 年 20-20 时日值降水数据。 3) 实时航班数据： 2013 年全国机场航班实时数 据。 4) 机场资料：2013 年 174 个机场的飞行区等级数 据、抗灾投入、机场货邮吞吐量、旅客吞吐量 等 。 中 国 民 用 航 空 局 [21] http://www.caac.gov.cn/h1/h2/。 2.2. 研究方法说明 1）自然断点分级法：（natural breaks classification） 用统计公式来确定属性值的自然聚类，公式的 功能就是减少同一级的差异，增加级间的差异， 其公式为： ssdi-j   j k 1 ( a[k] i-j ) 2 , 1 i j n ） (1) 式中，a 为一个数组（数组长度为 n）； 为每 个等级中的平均值。 2）层次分析法 [22] ：（analytic hierarchy process，简 称：ahp）该方法是对一些较为复杂又难以 完全定量分析的问题做出决策的简易方法， 主要包括以下四个步骤： ① 指标体系的层次结构建立，根据灾害系统等 相关理论将指标层次化，一般分为目标层、 标准层和指标层。 ② 构造判断矩阵，根据专家调查结果，评定该 层次中各有关元素的相对重要程度。本研究 结合“中国机场暴雨灾害风险的指标体系”设 计了“专家打分表”，采用问卷调查的方式对民 航、气象、灾害等领域的专家及研究人员等 进行问卷调查。问卷发放 60 份，回收 58 份， 回收率 96.7%，有效问卷 56 份，有效率 96.6%。 采用比例最大原则确定为最终专家打分结果。 ③ 求特征向量和特征根，用矩阵向量法计算判 断矩阵的最大特征根 max ，获得每层次的各 指标的权重值。 ④ 一致性检验，通过一致性比例公式：cr=ci/ri 来检验判断矩阵的权重分配是否合理。其中， ci 为一致性指标，ci=( max -n)/(n-1)；ri 为 随机一致性指标。当 cr<0.10 时，认为判断 矩阵具有一致性，否则，需对判断矩阵进行 调整。 3） 综合指标法：把原始指标数据构造成一组综合 性指标 x， 1 , 1, 2, , n i i i x x w i n    （2） 其中，x 为综合指标；xi 为各指标的原始值； wi 为各指标由层次分析法得到的相应权重值。 3. 模型建立 3.1. 暴雨灾害风险评估体系的指标构建与计算 根据史培军对灾害系统论 [23] 的总结，灾害包括 致灾因子，承灾体和孕灾环境三要素，而社会经济 环境是孕灾环境的重要组成部分。张继权等 [24] （2007） 在气象灾害的研究中提出气象灾害风险形成除了致 灾因子危险性、承灾体的脆弱性和暴露性外，防灾 减灾能力也是制约和影响气象灾害风险的因素。防 灾减灾能力表示某一区域遭受灾害破坏后，在一定 时期内能够从气象灾害中恢复的程度，包括：应急 管理能力、减灾投入、资源准备等。防灾减灾能力 越高，可能灾害损失越小，气象灾害风险就越小。 published by atlantis press copyright: the authors 190 http://www.caac.gov.cn/h1/h2/ x.m. guo et al 风险具有不确定性，对于民航一个具体行业， 灾害风险涉及影响因素也比较多。本文根据灾害系 统论和相关灾害风险理论，以致灾因子、承灾体和 孕灾环境三个方面为基础（如图 1），选取了致灾因 子、暴露性、脆弱性和防灾减灾能力四个指标构建 中国民航暴雨灾害风险评估指标体系。 致灾因子(r)：1994-2013 年全国机场附近的气 象站点日降水量 50mm 记为一次暴雨过程。暴雨灾 害的发生与降水的强度和频次有重大关系，为了较 科学地定量地分析中国机场致灾因子的危险性强度， 本文选取暴雨频次和暴雨强度为主要评价指标 [25] ， 统计机场附近各气象站每年的日暴雨过程降水量与 日数，采用百分位数确定降水对暴雨灾害形成的影 响程度（即第 60、80、90、95、98 百分位数作为划 分降水量值的位置指标），将其划分为 1~5 个等级 （如表 1）。等级越高，机场暴雨灾害危险性越高。 将暴雨强度 1、2、3、4、5 级作为暴雨致灾因子的 权重，导入到 arcgis，应用空间分析模块中提供的 反距离加权工具（idw）进行空间插值，得到机场 暴雨灾害危险等级区划图。 暴露程度(e)：是指可能受到暴雨威胁的所有人 和财产，如：人员、生命线等。本文用中国民用航 空局提供的 2013 年各个机场的旅客吞吐量、货邮吞 吐量、飞机起降次数等可能受到的暴雨影响程度来 度量暴露性，运用层次分析法，指标两两对比基于 民航行业的专业人士问卷调查结果，计算得到权重 分别为 0.454、0.225 和 0.321。并利用自然断点分级 法将承灾体的暴露性指数分 5 个等级，并基于 arcgis 绘制出中国暴雨暴露性区划图。 脆弱性(v)：1994-2013 年间机场在暴雨降水影 响下关闭时间 t 与该降水时间 t 的关系 图 1. 中国机场暴雨灾害风险评估概念 表 1 暴雨过程强度等级标准 天数 1 级 （60--80%） 2 级 （80--90%） 3 级 （90--95%） 4 级 （95--98%） 5 级 （98%以上） 样本长 (/d) 暴雨值 (mm) 1 50.5≤r<71.0 71.0≤r<90.0 90.0≤r<114.5 114.5≤r<129.0 r≥129.0 122 50 2 66.3≤r<80.4 80.4≤r<106.3 106.3≤r<125.7 125.7≤r<130.4 r≥130.4 61 50 3 72.9≤r<90.5 90.5≤r<119.3 119.3≤r<133.8 133.8≤r<147.1 r≥147.1 53 50 4 82.1≤r<111.2 111.2≤r<121.0 121.0≤r<154.7 154.7≤r<168.4 r≥168.4 26 50 5 104.5≤r<120.2 120.2≤r<140.5 140.5≤r<162.1 162.1≤r<196.0 r≥196.0 23 50 6 112.1≤r<130.1 130.1≤r<153.5 153.5≤r<190.2 190.2≤r<213.7 r≥213.7 16 50 ≥7 125.7≤r<147.1 147.1≤r<167.4 167.4≤r<203.8 203.8≤r<221.7 r≥221.7 12 50 published by atlantis press copyright: the authors 191 , i i t v v v t l   (3) 式中，t 是某机场有暴雨当天的计划航班总数，ti 为某机场受当天暴雨影响的非正常航班数（取消、 返航、延误和迫降等），v 为非正常航班所占比例， li 为暴雨的时长，v 为某机场的暴雨脆弱性。计算 所得结果用自然断点分级法将承灾体的脆弱性指数 分 5 个等级，并基于 arcgis 绘制出中国暴雨脆弱性 区划图。 防灾减灾能力(c)：机场飞行区是飞机地面活动 及停放提供适应飞机特性要求和保证运行安全的构 筑物的统称，包括：跑道及升降带、滑行道、停机 坪、地面标志、灯光助航设施及排水系统等，能较 好的表征一个机场对灾害性天气的防灾减灾能力。 本文把机场的飞行区划等级 3c、4c、4d、4e、4f 划分为 1、2、3、4、5 共五级。飞行区划等级与机 场建筑占地面积 s，分别用 0.5 的权重，计算出中国 174 个机场的防灾减灾能力值。基于 arcgis 绘制出 机场防灾减灾能力区划图。 3.2. 暴雨灾害风险区划与权重确定 自然灾害风险评估 [10~12] ，一般采用的灾害评估 模型为: vhr  (4) 在以上因子定量分析评价基础上，暴雨灾害风 险指数计算公式如下： )5( fvehr  (5) )5(** fvehr wwww fveh  (6) wa i i i h * 5 1    (7) wa i i i e * 3 1    (8) 式中，r 表示机场暴雨灾害风险指数， r 值越大， 风险程度越大。 h，e，v ，f 分别表示暴雨危险 性、机场的暴露性、机场脆弱性、防灾减灾能力， w 为个评价指标的权重系数，a 为各评价指标的量 化值，i 为评价因子系数。h、e、v、f 各评价因子 的权重分别为 wh 、 we 、 wv 、 wf ，通过层次分析法 计算获得，两两对比的指标由专家打分获得。 判断矩阵每一列归一化： , , 1, 2, ij mij kj k=1 = i j m a a a    （ ） （9） 每一列判断矩阵相加： a i=  m j a 1 ij （i,j=1,2,…m） （10） 对向量 a =( a 1, a 2,… a m)作归一化处理：  , , 1, 2,i mi j j i j m a w a     (11) 式中，a 为评价指标量化值，i、j 为相应系数，w 为相应的权重。其中，计算得到判断矩阵的最大特 征根为 4.017，一致性检验得 cr=0.067<0.1。依次 所得到的 w1, w2,…, wm 即为所求各评价指标的权 重。通过上述方法，计算各评价因子的指标权重结 果如表 2。 最后利用 arcgis 中将机场暴雨灾害风险划分 五个等级（高、较高、中等、较低，低），并且用 arcgis 空间插值法绘制出中国机场暴雨灾害风险 区划等级图。 表 2 中国机场暴雨风险权重 目标层 标准层 指标层 中 国 机 场 暴 雨 风 险 危险性（0.425） 1 级暴雨强度频次（0.067） 2 级暴雨强度频次（0.133） 3 级暴雨强度频次（0.200） 4 级暴雨强度频次（0.267） 5 级暴雨强度频次（0.333） 暴露性（0.295） 货邮吞吐量（0.225） 旅客吞吐量（0.454） 航班数（0.321） 脆弱性（0.147） 机场脆弱性 防灾减灾能力（0.133） 机场飞行等级 published by atlantis press copyright: the authors 192 x.m. guo et al 4. 结果与分析 4.1. 暴雨致灾因子危险性分析 分析中国综合暴雨强度频次分布图，可知：各 级暴雨频次最大地区 6 月份主要在华南一带如：白 云机场等，七月、八月随着副热带高压的移动，中 国雨带也随着移动，图 7 暴雨危险性变化与这一现 相似。七月末、八月初，华北、华中地区暴雨比较 频繁，机场的暴雨危险性也相应增加。 4.2. 机场暴露性分析 如图 2，中国机场 2013 年平均暴露程度图，旅 客吞吐量、货邮吞吐量和飞机起降次数较多的机场 如：首都国际机场、浦东机场、虹桥机场、白云机 场、咸阳机场、长水国际机场等机场，在灾害下的 暴露度异常大，如：长水国际机场的暴露度是全国 机场平均暴露度的 6.7 倍，首都机场的暴露度是全 国机场平均暴露度的 18 倍。暴雨发生时，机场航班 多，与其他机场的联系密切，影响范围也比小机场 大。从时间上分析，中国机场月平均暴露性也有较 明显的季节差异，从 7 月到 10 月，中国机场的暴露 性比其它月份较高，8 月的暴露度最高，如图 3。 4.3. 机场暴雨脆弱性分析 如图 4,机场暴雨脆弱性等级分布，达到五级的 机场一共有 5 个：广西的福成机场、湖南的永州机 场、云南的阿里机场、黑龙江的伊春机场和首都国 际机场。脆弱性为 4 级的机场较多,除周水子机场、 榆阳机场等几个机场外，主要都分布在长江流域， 如：荷花机场、九江机场。 图 3 中国机场月平均暴露度 图 2 中国机场暴露性等级分布图 published by atlantis press copyright: the authors 193 4.4. 机场防灾减灾能力分析 由机场的排水系统，跑道等作为机场等级评价 标准，机场等级越高对不同程度的暴雨有较高的应 对能力。比如：增加跑道长度有利于在降落时气象 条件不佳、刹车反推失效或错过最佳接地点的情况 下避免冲出跑道，亦有利于在紧急中断起飞的情况 下利用剩余跑道长度减速刹车。如图 5，我国达到 5 图 5 中国机场防灾减灾能力等级分布图 图 4 中国机场脆弱性等级分布图 published by atlantis press copyright: the authors 194 级（4f 等级）的机场有 11 个,有较强的防灾减灾能 力。如：首都国际机场、浦东国际机场、白云国际 机场、宝安国际机场、萧山国际机场、长水国际机 场、天河国际机场、双流国际机场、咸阳国际机场、 滨海国际机场、禄口国际机场、两江国际机场。机 场等级为 4 级的机场一共有地窝堡机场、龙嘉机场、 长乐机场等 20 多个机场。 4.5. 中国机场暴雨灾害风险等级分析 4.5.1. 中国机场暴雨灾害风险分析 如图 6，暴雨危险性与暴雨灾害风险呈现非常 明显的正相关关系，中国民航暴雨高风险主要由于 致灾因子（暴雨）高危险性起着主要作用，但这种 趋势也不是绝对的，如：永州机场、金湾机场、栎 社机场、井冈山机场等，在暴雨月均危险性较低的 情况下，暴雨灾害风险仍然相对较高。结合图 3-5 分析可得，首都国际机场、白云机场、浦东国际机 场等较大型机场除了致灾因子外，机场的暴露性起 着很重要的作用；永州机场、栎社机场、井冈山机 场等脆弱性等级较高，使得整体灾害风险值较大； 而三义机场、南充机场、姜营机场等防灾减灾能力 较低，脆弱性等级高，即使暴雨危险性较低，其暴 雨灾害综合风险等级同样很高。 4.5.2. 中国机场暴雨灾害风险的月变化 如图 7，每年的 10 月至次年的 4 月均为全国性的暴 雨灾害低危险性时段，同样也是全国机场暴雨低风 险时段。5 月到 9 月暴雨灾害高风险中国机场分布 较广，且高风险分布随着季节变化明显。 4 月开始中国南部个别机场（如：佛山机场、 宝安机场、井冈山机场等）随着暴雨危险性上升， 成为暴雨灾害高风险机场。5 月份中国东南部地区 和东部地区的机场暴雨灾害风险所有上升，首都国 际机场由于暴露度有所上升，暴雨灾害风险较高； 周水子机场的脆弱性较高，同样在暴雨灾害风险等 级区划中突显出来。但是中国西南，西部和北部地 区的机场暴雨灾害风险普遍较低。 6 月份中国大部分地区进入雨季，全国暴雨危 险性上升，暴雨灾害风险也全国性上升。中国南部、 东部、东北部地区机场的暴雨灾害风险水平较高， 暴雨灾害风险较高的机场有 18 个，占全国所有机场 的 10.06%；而暴雨灾害高风险机场有 21 个，占全 国机场的 12.42%。另外，遥墙机场和周水子机场两 个机场虽然暴雨危险性较低，但脆弱性较大，并且 防灾减灾能力较弱 ，同样成为暴雨灾害高风险机场。 7 月中国机场暴雨灾害高风险区继续向北推进， 范围扩大到最大。暴雨灾害较高风险机场共有 23 个， 占全国机场数 13.02%；暴雨灾害高风险机场共 28 个，占全国机场数 16.57%。暴雨危险性随中国雨带 向北移动，长江以北地区暴雨风险上升，但是江淮 地区暴雨危险性降低，部分机场的暴雨风险也明显 降低，如：永州机场、邵通机场和普者黑机场等。8 月份这种趋势更加明显，但是总体上暴雨灾害高风 险机场数量与 7 月份相比有所减少。 9 月开始中国雨带随副热带高压南撤，暴雨危 险性也迅速降低，暴雨灾害高风险机场主要分布在 东南沿海地区和一些暴露性较大的大机场，如：流 亭机场、浦东国际机场、义乌国际机场、永强机场， 佛山机场、美兰机场、首都国际机场和遥墙机场。 从机场角度分析，部分机场的暴雨灾害风险时段性 较强，主要集中在 6 至 9 月份，但灾害风险大，如： 首都国际机场、浦东国际机场和双流机场等；部分 机场灾害风险不大，但是全年都有暴雨灾害风险存 在，如：梧州机场、右江机场和梅州机场等。 图 6 中国机场月平均暴雨灾害风险与危险性对比 published by atlantis press copyright: the authors 195 http://baike.baidu.com/view/376985.htm http://baike.baidu.com/view/176714.htm http://baike.baidu.com/view/376992.htm http://baike.baidu.com/view/376992.htm http://baike.baidu.com/view/120522.htm http://baike.baidu.com/view/346835.htm http://baike.baidu.com/view/6238016.htm http://baike.baidu.com/view/6238016.htm http://baike.baidu.com/view/196559.htm http://baike.baidu.com/view/56450.htm http://baike.baidu.com/view/194581.htm http://baike.baidu.com/view/76193.htm http://baike.baidu.com/view/194601.htm http://baike.baidu.com/view/346885.htm 图 7 中国机场暴雨灾害风险区划图 5. 结论 从风险区划图上看， 中国机场暴雨灾害高风险 分布与中国雨带南北移动有较明显的一致性，而且 暴雨灾害高风险机场主要分布在东南沿海地区，东 部、南部地区和北部个别机场；而西部、西北部地 区的机场暴雨灾害风险较低。暴雨灾害风险评价适 当考虑了机场暴露度、脆弱性和防灾减灾能力突显 出了暴雨灾害发生过程中不同机场的不足之处。机 场可以根据本身暴雨灾害风险的时段特性采取相应 对策。分析中国 2009-2013 年前十的机场重大延误 事件，其中五件由暴雨引起，或者与其相关。2012 年 7 月 21 日首都国际机场由于大暴雨导致大面积延 误，全天取消航班超过 500 架次，滞留机场的旅客 一度高度 8 万人；2010 年 7 月 16 日由于暴雨双流 机场两度陷入瘫痪状态，上万人滞留机场，同时导 致黄龙机场、咸阳机场，中甸机场等周边机场出现 较大面积延误。通过上述的中国机场暴雨风险等级 图与近年实际机场延误状况比较，可以看出，夏季 暴雨风险指数较高的机场：首都国际机场、三义机 场、咸阳机场等机场的实际灾情亦最为严重。说明 本研究民航暴雨风险评价结果，有一定的可信度。 published by atlantis press copyright: the authors 196 x.m. guo et al 6. 讨论 系统、全面、深入地研究暴雨灾害对我国民用 航空安全的影响以及相关的防御规划和措施，对我 国民用航空业的可持续发展及安全运营具有巨大的 战略意义。其中，研究我国民用航空系统所面临的 天气灾害风险的等级和区划，对民航安全能力建设 具有重要的作用。只有在对暴雨灾害风险进行科学 地、系统地分级和区划以后，才能为提高民航系统 防御和应对灾害性天气的能力而做出合理而有效的 规划，提高民航系统运营的安全性。 本文对中国民航暴雨灾害风险的评估方法展开 了初步探索研究，与传统灾害风险评估的区别在于： 风险评价中的脆弱性主要指机场的功能脆弱性，并 不是机场相关建筑设施的物理脆弱性；所评估的内 容是机场的服务功能的在暴雨灾害中的损失风险。 对于民航而言，虽然，如地震等巨灾发生时损失可 能非常大，但是发生的概率非常小。相比而言，在 日常运营过程中，频发的气象灾害对机场的影响受 民航相关人员的关注。文章并没有深入挖掘风险指 数与实际灾情的差异，对于个别机场暴雨灾害风险 计算结果和实际灾情有所出入，其原因涉及以下几 个问题：1 民航暴雨灾害风险评价方法仍然在探索 中，其模型和评价所涉及的指标需要进一步的改进 完善，以找出更能体现暴雨灾害危险性，民航暴露 性和防灾减灾能力的指标；2 收集的风险评估相关 数据的准确性需要进一步确认；3 另外，民航脆弱 性评估也有待进一步研究。民航是一个以机场为节 点、航班为链接而相互联系的网络系统，某一区域 的机场受到灾害的影响后，该区域以外的其它机场 也可能会由于航班链接而受到间接影响，如：2013 年 7 月北京东部地区的暴雨直接造成首都国际机场 大量航班延误取消，其间接影响是：武宿机场、新 郑机场等几十个机场的延误情况也明显加重。因而 用传统脆弱性评价法 [26-27] 不能有效地评估民航系统 的脆弱性，因此，将来在评估民航系统脆弱性时需 要充分利用学科综合特性，借鉴研究网络系统的相 关理论和方法。 致谢 本项研究得到了下列基金项目的资助：国家重点 基 础 研 究 发 展 计 划 （ 973 ） 项 目 ( 项 目 号 ： 2012cb955404)；国家中央高校基础研究基金(项目号： 256-105570gk) ；北京师范大学地表过程与资源生态 国 家 重 点 实 验 室 自 由 探 索 项 目 基 金 （ 项 目 号 ： 2015-zy-05）。 参考文献 [1] ipcc. 2011. summary for policymakers. in: intergovernmental panel on climate change special report on managing the risk of extreme events and disasters to advance climate change adaptation. cambridge university press, cambridge, united kingdom and new york, ny, usa. [2] 吴吉东,傅宇,张洁,李宁. 1949—2013 年中国气象灾害 灾情变化趋势分析. 自然资源学报,2014,09:1520-1530. [3] 中国气象局． 中国气象灾害年鉴( 2011)． 北京: 气象 出版社，2012: 172－177. [4] 袁艺,马玉玲. 近 30 年我国自然灾害灾情时间分布特征 分析. 灾害学,2011,03:65-68+82. [5] http://www.flightstats.com. [6] 何勇,何春林. 民航气象预警平台的设计与实现. 空中 交通管理,2008,06:23-25+22. [7] brabb e e, pampeyan e h, bonilla m g landslide susceptibility in san mateo county, california.us geological survey miscellaneous field studies map mf-360,scale 1: 62500,1972. [8] lavalle c, de roo a, rarredo j, etal. towards an european integrated map of risk from weather driven events: a contribution to the evaluation of territorial cohesion in europe [r]. technical eur reports, eur, 2005. [9] 李世奎,霍治国,王素艳,刘荣花,盛绍学,刘锦銮,马树庆, 薛昌颖. 农业气象灾害风险评估体系及模型研究. 自 然灾害学报,2004,01:77-87. [10] 霍治国,李世奎,王素艳,刘锦銮,薛昌颖. 主要农业气象 灾 害 风 险 评 估 技 术 及 其 应 用 研 究 . 自 然 资 源 学 报,2003,06:692-703. [11] 高晓蓉，王春乙，张继权，温煦. 东北地区玉米主要气 象 灾 害 风 险 评 价 模 型 研 究 . 中 国 农 业 科 学.2014,47(2):4257-4268. [12] 于飞,谷晓平,罗宇翔,郑小波. 贵州农业气象灾害综合 风险评价与区划. 中国农业气象,2009,02:267-270. [13] 武永峰,张勇,陈鲜艳,尚赞娣,刘布春. 湖南省公路交通 暴雨风险评价研究. 自然灾害学报,2011,05:148-154. [14] 付桂琴,张文宗. 河北省电网气象灾害风险评价. 中国 气象学会 . 创新驱动发展 提高气象灾害防御能力 ——s3 第三届气象服务发展论坛——公众、专业气象 预报服务技术与应用.中国气象学会: 2013: 6. [15] milan janic. an assessment of risk and safety in civil aviation . journal of air transport management. 6 (2000): 43-50. [16] 李铭, 王淑一, 张宇峰. 民航雷达站雷电灾害风险评估. 黑龙江气象,2012,03:39-40. published by atlantis press copyright: the authors 197 http://www.flightstats.com/ [17] 张甦, 郝丽萍, 李子良, 梁燕芳. 民航广汉机场气象能 见度的周期性变化特征的初步分析 . 高原山地气象研 究,2009,03:69-72. [18] 王立倩, 王春玲, 周杰群. 构建民航气象数据库服务系 统运维自动化平台. 气象通信与信息技术委员会、国家 气象信息中心.第 31 届中国气象学会年会 s13 气象通 信与信息技术应用实践与新技术探索.气象通信与信息 技术委员会、国家气象信息中心, 2014: 5. [19] 魏学军,姜桓,王贵明,孙建睿. 雷暴对航空飞行的灾害 分析. 内蒙古气象,2013,04:42-44. [20] 樊运晓,罗云,陈庆寿. 区域承灾体脆弱性综合评价指标 权重的确定. 灾害学,2001,01:86-88. [21] 中国民用航空局 http: //www.caac.gov.cn/h1/h2/. [22] 刘刚,基于层次分析法的社区灾害风险脆弱性评价.兰 州大学学报，2013,41( 4)：102-108 [23] 史培军. 再论灾害研究的理论与实践. 自然灾害学 报,1996,04:8-19. [24] 张继权，李宁等。主要气象灾害风险评价与管理数量 化方法及其应用. 北京师范大学出版社. 北京.2007. [25] cannon blaikie, davis i p t and wisner b. at risk: natural hazards. people’s vulnerability, and disasters . london: rout ledge, 1994: 141－156. [26] burton i, kates r w white g f. the environmental hazard. oxford univ. press, oxford, 1978. [27] blarkre p, cannon t, davis i, et al. at risk: natural hazards, peoples vulnerability, and disasters. london; rout ledge, 1994. published by atlantis press copyright: the authors 198 http://www.caac.gov.cn/h1/h2/ 2 journal of risk analysis and crisis response vol. 9(1), march (2019), pp. 2–10 doi: https://doi.org/10.2991/jracr.b.190328.001; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). geospatial information diffusion technology supporting by background data chongfu huang1,2,3 1key laboratory of environmental change and natural disaster, ministry of education, beijing normal university, beijing 100875, china 2state key laboratory of earth surface processes and resource ecology, beijing normal university, beijing 100875, china 3faculty of geographical science, academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china received december 20, 2018 accepted january 10, 2019 abstract in this paper, we express the initial concept of geospatial information diffusion supporting by background data, which plays a role as a bridge to diffuse the information carried by the observations, obtained from observed units, to gap units. the self-learning discrete regression, based on the multivariate normal diffusion, is suggested to supplement incomplete geospatial data to be complete. the suggested method has obvious advantages over the geographic weighted regression and the artificial neural network for inferring the observations in gap units keywords: geographic unit, background data, information diffusion, normal diffusion, self-learning discrete regression 借助背景数据的地理空间信息扩散技术 * 黄崇福 1,2,3 1. 北京师范大学，环境演变与自然灾害教育部重点实验室，北京 100875，中国 2. 北京师范大学，地表过程与资源生态国家重点实验室，北京 100875，中国 3. 北京师范大学，地理科学学部，减灾与应急管理研究院，北京 100875，中国 摘要:本文提出了以背景数据为桥梁，将已观测数据携带的信息，扩散到空白单元的信息扩散基本思想，并 给出了基于多元正态扩散的自学习离散回归模型，从而实现了将地理空间上的不完整数据补充为完整数据。本 文的方法，在空白地理单元数据的推测方面，比地理加权回归和人工神经网络，有明显的优势。 关键词:地理单元；背景数据；信息扩散；正态扩散；自学习离散回归 1. 引言 在全球环境变化驱动下，人们对陆地表层系统中 各种现象的研究，须使用更精细的观测数据测度更具  *本研究由国家重点研发计划（编号：2017yfc1502902） 和国家自然科学基金项目（编号：41671502）资助。 体的研究对象 [1] 。然而，由于资源或时间的制约，为 研究某一问题，人们获得的数据，常常不完整。例如， 2008 年汶川大地震[2]发生一天后，救援人员对通信中 断的受灾地区，仍然不知其灾情如何。灾情信息的不 完整，严重制约着灾害救助的时效性与精准性。为解 决此问题，人们统计回归历史灾害数据，得出经验公 https://doi.org/10.2991/jracr.b.190328.001 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 2 3 geospatial information diffusion technology supporting by background data chongfu huang1,2,3 1key laboratory of environmental change and natural disaster, ministry of education, beijing normal university, beijing 100875, china 2state key laboratory of earth surface processes and resource ecology, beijing normal university, beijing 100875, china 3faculty of geographical science, academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china received december 20, 2018 accepted january 10, 2019 abstract in this paper, we express the initial concept of geospatial information diffusion supporting by background data, which plays a role as a bridge to diffuse the information carried by the observations, obtained from observed units, to gap units. the self-learning discrete regression, based on the multivariate normal diffusion, is suggested to supplement incomplete geospatial data to be complete. the suggested method has obvious advantages over the geographic weighted regression and the artificial neural network for inferring the observations in gap units keywords: geographic unit, background data, information diffusion, normal diffusion, self-learning discrete regression 借助背景数据的地理空间信息扩散技术 * 黄崇福 1,2,3 1. 北京师范大学，环境演变与自然灾害教育部重点实验室，北京 100875，中国 2. 北京师范大学，地表过程与资源生态国家重点实验室，北京 100875，中国 3. 北京师范大学，地理科学学部，减灾与应急管理研究院，北京 100875，中国 摘要:本文提出了以背景数据为桥梁，将已观测数据携带的信息，扩散到空白单元的信息扩散基本思想，并 给出了基于多元正态扩散的自学习离散回归模型，从而实现了将地理空间上的不完整数据补充为完整数据。本 文的方法，在空白地理单元数据的推测方面，比地理加权回归和人工神经网络，有明显的优势。 关键词:地理单元；背景数据；信息扩散；正态扩散；自学习离散回归 1. 引言 在全球环境变化驱动下，人们对陆地表层系统中 各种现象的研究，须使用更精细的观测数据测度更具  *本研究由国家重点研发计划（编号：2017yfc1502902） 和国家自然科学基金项目（编号：41671502）资助。 体的研究对象 [1] 。然而，由于资源或时间的制约，为 研究某一问题，人们获得的数据，常常不完整。例如， 2008 年汶川大地震[2]发生一天后，救援人员对通信中 断的受灾地区，仍然不知其灾情如何。灾情信息的不 完整，严重制约着灾害救助的时效性与精准性。为解 决此问题，人们统计回归历史灾害数据，得出经验公 式，一旦发生破坏性地震，就可根据震级对灾情进行 粗估，也称为快速评估 [3] 。这种经验性的“隔空判 灾”，能将灾情的估计精度控制在数量级误差之内， 就已经很不错。精细一点，人们可借助地理信息系统 （gis）中使用的数学插值模型[4]，依据在一些点上 采集到的灾情，推测其它点上的灾情。问题是，数学 插值要求数据具有连续性，而除了温度场以外，大多 数地表数据都不满足连续性这一条件。因此，数学插 值的结果，往往不尽人意。 由于概率空间中的信息扩散方法 [5] ，能在小样本 条件下显著提高概率分布的估计精度，文[6]尝试着 将此方法发展到地理空间上，将不完整数据补充为完 整数据。本文继续此项工作，完善重要概念的定义， 规范化如何借助背景数据进行地理空间信息扩散。 2. 概率空间中的信息扩散方法 让我们用一个简单的例子来说明信息扩散的初衷。 当我们对所研究系统进行一次观测时，假定我们得到 了一个数据。我们可以将此数据看作是天上掉下来的 一滴水(图 1(a))，这滴水落在地面的什么位置，有一 定的随机性。只有统计较多的水滴，我们才可能知道 一段时间内落下水滴在地面的分布。但如果统计水滴 落下后的影响面积(图 1(b))，统计较少的水滴，就能 知道分布。 这种集值化后能提高概率分布估计精度的现象， 被称为信息扩散原理 [7] ：设 x 是从概率密度函数为 p(x)的总体中随机抽取的一个样本， )(ˆ xp 是用 x 对 p(x)的一个非扩散估计。如果 x 不完备，则一定存在 一个基于 x 的扩散估计 )(~ xp ，使得 )(~ xp 比 )(ˆ xp 更精 确。 估计概率密度函数时，将 x 中各样本点 x 携带的 信息在观测空间中的扩散，其实是一种在概率空间中 的扩散。本文将传统的信息扩散方法，称为“概率空 间中的信息扩散方法”。最常用的有线性信息分配法 和正态信息扩散法 [8] 。用正态扩散函数  将样本点 x 的信息在概率空间中进行扩散，如图 2 所示。 目前，概率空间中信息扩散理论的基础比较稳固， 应用涉及面较广，尤其在风险分析方面。例如，信息 扩散方法被用于分析太湖蓝藻暴发风险，为政府决策 提供了重要依据 [9] ；用于研究美国东海岸飓风风险， 计算出保守风险值和冒险风险值 [10] ；用于研究中国最 近 20 年的水灾数据，计算出不同可能性的多值风险 (a) 观测值是 x (b) x 被变成一个模糊集 a 图 1. 信息扩散类似于水滴影响面积。统计少量水滴的影响面积，就能估计水滴的分布。 图 2. 用正态扩散函数将样本点信息在概率空间中进行扩散 概率空间  一个点x  一个模糊集a ωuxx h ux h uxa    ,], 2 )( exp[ 2 1 ),(: 2 2   概率空间 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 4 [11] ；用于研究中国北方草原火灾风险，为畜牧业生产 制定补偿计划提供了依据 [12] ；用于研究洞庭湖区洪水、 干旱、虫害和鼠害，绘制出粮食生产的自然灾害风险 图 [13] ；用正态扩散处理我国水利部数据库中的水灾数 据，生成可变模糊集，拟合出了水灾风险曲线 [14] ；用 于北塞浦路斯旅游保险气候指数计算，绘制了风险图 [15] ；将信息扩散模型嵌入土壤流失方程，评估环渤海 地区不同降雨情况下的土壤侵蚀风险 [16] ；用于分析液 化天然气站过去 12 个月的运行数据，能及时发现潜 在的风险 [17] ；用于研究草原生物灾害，绘制出我国北 方 10 省区的风险图 [18] ；用于研究农业保险业务数据， 计算不同损失的可能性，为政府财政支持农险提供依 据 [19] ；用于广东省北江、西江和绥江三江汇流区的洪 水风险评估，能根据洪峰水位推断水灾面积的几种可 能性 [20] ；用于广州南沙区黄阁镇和南沙镇的区域环境 风险评估，帮助当地政府优化工业区布局，建立风险 防范管理程序 [21] 。 文[22]用解析几何学的方式证明，信息扩散原理， 不仅在概率空间中成立，而且在几何空间中也成立 （见图 3）。这就意味着，我们可以将信息扩散技术， 拓展到在地理空间上去，以填补地理单元上的数据空 白，使不完整的空间数据集，变为完整的数据集。 3. 地理单元上的不完整数据 当我们对一个研究区域上的某种地表现象展开 研究时，该区域中各地理单元上的地理位置、人口 等信息很容易得到，但有些信息却不易得到。例如， 当我们对某县的自然灾害风险进行研究时，一些乡 镇的历史灾害资料就难以收集到。又例如，当重大 自然灾害发生后，灾害管理部门很难在 1 小内得到 灾区所有乡镇的灾情信息。依据不完整灾情制定的 应急救灾方案，直接影响灾害救助效果。 设 g 是一个研究区域，其上的某一现象 f，例 如地震风险，是被研究的对象。不失一般性，假定 g 由 n 个地理单元 g1, g2, ..., gn 组成，即， g={ g1, g2, ..., gn } (1) 更进一步地，我们假设，通过观测 g 上所有的地 理单元与 f 有关的数据（或向量），可以识别出现 象 f。对地理单元 gi 进行观测得到的数据，记为 wi。 例如，当我们研究由 129 个县级地理单位组成 的云南省“地震风险”（earthquake risk）现象时， 该被研究的现象可写为 f=erisk。研究区域 g 是云南 省（yunnan），记为： gyunnan ={g1, g2, ..., g129} 式中，g1=昆明市五华区，g2=昆明市盘龙区，...,g129= 临沧市双江拉祜族佤族布朗族傣族自治县。 为了识别 gyunnan 上的 erisk，我们必须(观测)知道 每个单元的“地震危险性 [23] ”和“地震易损性 [24] ”。 例如，对玉溪市通海县，即地理单元 g27，我们必须 知道下面向量中每个分量的具体数值： w27=（通海县地震危险性，通海县地震易损性） 如果我们能获得研究区域 g 上识别 f 所需的所 有地理单元上的值，则称该研究区域上的数据对于 识别 f 是完整的，否则称 w 为不完整。据此，本文 给出“数据不完整”的形式化定义如下： 定义 1：设 g 由地理单元 g1, g2, ..., gn 组成，对 gi 的观测记为 wi, i=1, 2, …, n。假定 g 上的地表现象 f 可用这些地理单元上的观测数据集合 图 3. 信息扩散原理的解析几何表述。(a) 用算子和数据集 a 估计 x 和 y 间的关系 r（真实关系，假定为右侧曲 线），得一个估计 r(, a) （假定为左侧曲线）的所有元素（r(, a)中的点）的隶属度值是 1。（b）将 a 中的所 有数据用信息扩散函数变为模糊集，生成数据集 a 。用算子和数据集 a ，得对 r 的另一个估计 r(, a )，其所 有元素都是拟三角形模糊数。（c）r(, a )在 oy 的投影。rkp 表示 r(, a )的水平集k,p。一定存在 rkp 上的一个 点 p(xp，yp)，使 r 和 p 间的距离小于 r 和 k 间的距离。 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 4 5 [11] ；用于研究中国北方草原火灾风险，为畜牧业生产 制定补偿计划提供了依据 [12] ；用于研究洞庭湖区洪水、 干旱、虫害和鼠害，绘制出粮食生产的自然灾害风险 图 [13] ；用正态扩散处理我国水利部数据库中的水灾数 据，生成可变模糊集，拟合出了水灾风险曲线 [14] ；用 于北塞浦路斯旅游保险气候指数计算，绘制了风险图 [15] ；将信息扩散模型嵌入土壤流失方程，评估环渤海 地区不同降雨情况下的土壤侵蚀风险 [16] ；用于分析液 化天然气站过去 12 个月的运行数据，能及时发现潜 在的风险 [17] ；用于研究草原生物灾害，绘制出我国北 方 10 省区的风险图 [18] ；用于研究农业保险业务数据， 计算不同损失的可能性，为政府财政支持农险提供依 据 [19] ；用于广东省北江、西江和绥江三江汇流区的洪 水风险评估，能根据洪峰水位推断水灾面积的几种可 能性 [20] ；用于广州南沙区黄阁镇和南沙镇的区域环境 风险评估，帮助当地政府优化工业区布局，建立风险 防范管理程序 [21] 。 文[22]用解析几何学的方式证明，信息扩散原理， 不仅在概率空间中成立，而且在几何空间中也成立 （见图 3）。这就意味着，我们可以将信息扩散技术， 拓展到在地理空间上去，以填补地理单元上的数据空 白，使不完整的空间数据集，变为完整的数据集。 3. 地理单元上的不完整数据 当我们对一个研究区域上的某种地表现象展开 研究时，该区域中各地理单元上的地理位置、人口 等信息很容易得到，但有些信息却不易得到。例如， 当我们对某县的自然灾害风险进行研究时，一些乡 镇的历史灾害资料就难以收集到。又例如，当重大 自然灾害发生后，灾害管理部门很难在 1 小内得到 灾区所有乡镇的灾情信息。依据不完整灾情制定的 应急救灾方案，直接影响灾害救助效果。 设 g 是一个研究区域，其上的某一现象 f，例 如地震风险，是被研究的对象。不失一般性，假定 g 由 n 个地理单元 g1, g2, ..., gn 组成，即， g={ g1, g2, ..., gn } (1) 更进一步地，我们假设，通过观测 g 上所有的地 理单元与 f 有关的数据（或向量），可以识别出现 象 f。对地理单元 gi 进行观测得到的数据，记为 wi。 例如，当我们研究由 129 个县级地理单位组成 的云南省“地震风险”（earthquake risk）现象时， 该被研究的现象可写为 f=erisk。研究区域 g 是云南 省（yunnan），记为： gyunnan ={g1, g2, ..., g129} 式中，g1=昆明市五华区，g2=昆明市盘龙区，...,g129= 临沧市双江拉祜族佤族布朗族傣族自治县。 为了识别 gyunnan 上的 erisk，我们必须(观测)知道 每个单元的“地震危险性 [23] ”和“地震易损性 [24] ”。 例如，对玉溪市通海县，即地理单元 g27，我们必须 知道下面向量中每个分量的具体数值： w27=（通海县地震危险性，通海县地震易损性） 如果我们能获得研究区域 g 上识别 f 所需的所 有地理单元上的值，则称该研究区域上的数据对于 识别 f 是完整的，否则称 w 为不完整。据此，本文 给出“数据不完整”的形式化定义如下： 定义 1：设 g 由地理单元 g1, g2, ..., gn 组成，对 gi 的观测记为 wi, i=1, 2, …, n。假定 g 上的地表现象 f 可用这些地理单元上的观测数据集合 图 3. 信息扩散原理的解析几何表述。(a) 用算子和数据集 a 估计 x 和 y 间的关系 r（真实关系，假定为右侧曲 线），得一个估计 r(, a) （假定为左侧曲线）的所有元素（r(, a)中的点）的隶属度值是 1。（b）将 a 中的所 有数据用信息扩散函数变为模糊集，生成数据集 a 。用算子和数据集 a ，得对 r 的另一个估计 r(, a )，其所 有元素都是拟三角形模糊数。（c）r(, a )在 oy 的投影。rkp 表示 r(, a )的水平集k,p。一定存在 rkp 上的一个 点 p(xp，yp)，使 r 和 p 间的距离小于 r 和 k 间的距离。 w={w1, w2, …, wn}, (2) 来识别。当 w 的所有元素均被赋值时，称 w 是在 g 上识别 f 的完整数据，否则称为不完整数据。 例如，图 4 的研究区域是一次大洪水造成的灾区， 一些地理单元上的死亡人数、重伤人数和灾民人数没 有统计上来，灾区的灾情数据不完整。此时，我们难 以识别出研究区域上的地表现象“灾情”。 显然，数据完整与否，与所研究的地表现象和 区域大小有关。越是复杂或精细的地表现象，越是 容易出现数据不完整；研究区域涉及的地理单元越 多，要获得完整的数据越是困难。 4. 地理空间上的信息扩散 当研究区域上的数据不完整时，有两条途径或许 能补齐数据，使之完整。一条途径是追加调研；另一 条途径是用数学模型进行插值。由于时间或成本的制 约，追加调研的途径常常走不通；由于有一定跨度的 相邻地理单元上的大多数地理特征的属性值并不连续， 插值法生成的数据，并无价值。 人们自然想到，是否可以用概率空间中的信息扩 散方法，将地理空间上已经得到的数据，扩散到缺少 数据的地理单元上，得到虽然与客观数据相比会有所 偏差，但却完整且有价值的数据。 显然，概率空间中的信息扩散方法，并不能直接 用于地理空间上。事实上，概率空间中的信息扩散， 是从样本 x 到到以样本空间 ω 为论域的模糊幂集 )(ωp 上的一个数学映射： ωuux ωx x   ),( )( :    p (3) 式中的称为信息扩散函数。式(4)是人们常用的正态 信息扩散函数，将样本点 x 携带的量值为 1 的信息， 由扩散系数 h 控制，按正态分布的形式，在样本空间 ω 中进行扩散。 ωuxx h ux h ux    ,], 2 )( exp[ 2 1 ),( 2 2   (4) 只有当 x 是小样本（少于 30 个样本点），不足 以用其估计其所来自总体的分布时，为提高估计精度， 对 x 进行扩散处理，才有意义。样本空间 ω 中的点 u 从样本点 x 能扩散得到多少信息量，由 u 与 x 之间的 距离决定。 式(3)中，被扩散的 x，本身就是概率空间中的一 个点，所以称为概率空间“中”的信息扩散。当我们 将式(2)中对地理单元 g 的观测值（或向量）w 视为一 个随机抽样时，当然也可以在 w 的概率空间中进行信 息扩散，但当 w 不是经纬度时，w 就不是地理空间上 的点，w 携带信息就不是在地理空间上被扩散。 我们需要通过某种媒介的帮助，才能将 w 携带的 信息，在地理空间上进行信息扩散。为此，我们界定 三个基本的概念：“空白单元”、“媒介”和“背景 数据”。 定义 2：设 g 和 o 是研究区域 g 中的两个地理单 元。如果在识别 g 上的现象 f 时，g 被观测并被赋值， 而 o 没有，则对于识别 f 而言，称 g 是一个被观测单 元，o 是一个空白单元。 图 4. 研究区域上的数据不完整，难以对总体“灾情”做出判断。 死亡：0人 重伤：2人 灾民：500人 损死亡：0人 重伤：5人 灾 民 ： 2000 人 死亡：未知 重伤：未知 灾民：未知 死亡：1人 重伤：20人 灾民：5000人 。。。 。。。 。。。 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 6 例如，在洪水灾区，对识别总体灾情而言，已经 被调查过灾情并获得数据的地理单元，是被观测单元； 没有被调查过灾情的地理单元，是空白单元。 对地理单元 g 的观测值（或向量）w 称为一个已 观测数据；对空白单元 o 的相应值，称为一个待观测 数据。 定义 3：设 o 是一个空白单元。如果能用数据， 依据一些已观测数据，对 o 赋值，则称是对空白单 元进行估值的媒介。 例如，根据 tobler 的第一地理定律，“每一事物 都与其他事物有关。与近处事物的关系，比与遥远的 事物关系密切” [25] ，gis 分析师经常使用反距离加权 （idw）插值来填补数据空白[26]。在 idw 法中，xo 为点 o 处的估计值，xi 为已知点 i 处的 x 值，di 为点 i 和点 o 之间的距离。借助于一系列距离 d1，d2，...， dn，使用一系列已知值 x1，x2，...，xn 估计 xo。idw 法中的 d1，d2，...，dn 就是媒介。 什么样的媒介，能帮助我们实现信息在地理空间 上的扩散呢，我们自然想到了地理特征的属性值。 在认识论中，事物本身具备的性质，称为特性， 通常是内在的，独有的，例如，熔点是晶体的特性。 人们根据某种事物所共有的特性抽象出的某一概念， 称为特征，一般从表面上就看得出来，例如，生命是 生物的特征。 地球自然创造的地理特征，称为自然地理特征， 例如，气候、地形、土壤、水文、日照时间、无霜期 等。人类社会创造的地理特征，称为人文地理特征， 例如，人口、交通、经济、宗教、国家政策等。 量化描述地理特征的数值，称为地理特征的属性 值，例如，100 万是“人口”特征的一个属性值。所 要讨论属性值涉及的范围，叫做该属性值的论域。下 面，我们用地理特征的属性值来定义“背景数据”。 定义 4：设 g1, g2, ..., gn 是 n 个被观测单元，o 是 一个空白单元，记 g={ g1, g2, ..., gn, o}. (5) 设 z 是多个特征的属性值向量， }{ 21 ogggg z,z,,z,zz n  (6) 是地理特征属性向量的集合。如果 zg 中的数据是媒 介，称 zg 为背景数据集，简称背景数据。 例如，用“人口”、“人均 gdp”和“相对暴 露度”等数据，依据被观测单元的灾情，我们能对空 白单元的灾情，进行估值。此时，“人口”、“人均 gdp”和“相对暴露度”等就是背景数据。 依据定义 4 可知，一个属性值的数据是否为另一 个属性值的背景数据，由两个因素决定，一是能否发 挥制约作用；二是能否容易获得。当我们研究一个地 区的洪水灾情时，各地理单元上的水灾程度，受地理 单元的位置、人口和经济发展程度制约，而且这些信 息很容易得到。地理位置，常常决定了地理单元暴露 于洪水的程度。河流流经区域较大的地理单元，其暴 露度较高。“相对暴露度”是由地理位置转化而来的， 具有地理位置的信息。 至此，我们可以提出借助背景数据在地理空间上 进行信息扩散的形式化定义如下： 定义 5：设 w 是用于识别区域 g 上的现象 f 的 不完整数据集，zg 是背景数据。如果模型 能用 zg 使 w 变成完整数据，则称 用 zg 对 w 的信息在 g 上 进行了扩散。 显然，任何内插算法都不能在地理空间上进行信 息扩散，因为背景数据在内插法中发挥不了作用。 brunsdon 等人提出的地理加权回归方法(gwr)[27]，本 质上是一个地理空间上的信息扩散模型。 gwr 是一种空间预测模型[28]，基本表述式为： )()()()( 10 v,uxv,ubv,ubv,uy  (7) 式中，y 表示服从高斯分布的因变量，x 是自变量，u 和 v 是数据的坐标，b0 是截距系数，b1 是斜率系数， 是随机误差项。多元 gwr 用式(8)表之。. i m k ikiikiii xvuvuy    1 0 ),(),( (8) 式中，yi 表示在位置 i 处的因变量， )(0 ii v,u 是在 i 处的截距系数，xik 是在位置 i 处的第 k 个自变量的值， )( iik v,u 是第 k 个自变量的局部回归系数。在 gwr 模型中， )( ii v,u 表示笛卡尔点坐标， i 为随机误差 项。 事实上，gwr 模型仍是一个线性回归模型，通 常用最小二乘法来估计模型中的系数。gwr 当然也 可拓展为广义的线性模型，例如 logistic 回归和 poisson 回归[29]。如果我们清楚地知道什么类型的函 数可以表达 y 和 x 之间的关系，无论它是多么的高度 非线性或非单调，我们都可以用此函数来改造 gwr 模型，达到最佳的回归效果。然而，问题在于特别 难以确定哪种非线性函数适合于表达特定的地球表 面现象。逃避这一问题的最常见做法，是想当然地 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 6 7 例如，在洪水灾区，对识别总体灾情而言，已经 被调查过灾情并获得数据的地理单元，是被观测单元； 没有被调查过灾情的地理单元，是空白单元。 对地理单元 g 的观测值（或向量）w 称为一个已 观测数据；对空白单元 o 的相应值，称为一个待观测 数据。 定义 3：设 o 是一个空白单元。如果能用数据， 依据一些已观测数据，对 o 赋值，则称是对空白单 元进行估值的媒介。 例如，根据 tobler 的第一地理定律，“每一事物 都与其他事物有关。与近处事物的关系，比与遥远的 事物关系密切” [25] ，gis 分析师经常使用反距离加权 （idw）插值来填补数据空白[26]。在 idw 法中，xo 为点 o 处的估计值，xi 为已知点 i 处的 x 值，di 为点 i 和点 o 之间的距离。借助于一系列距离 d1，d2，...， dn，使用一系列已知值 x1，x2，...，xn 估计 xo。idw 法中的 d1，d2，...，dn 就是媒介。 什么样的媒介，能帮助我们实现信息在地理空间 上的扩散呢，我们自然想到了地理特征的属性值。 在认识论中，事物本身具备的性质，称为特性， 通常是内在的，独有的，例如，熔点是晶体的特性。 人们根据某种事物所共有的特性抽象出的某一概念， 称为特征，一般从表面上就看得出来，例如，生命是 生物的特征。 地球自然创造的地理特征，称为自然地理特征， 例如，气候、地形、土壤、水文、日照时间、无霜期 等。人类社会创造的地理特征，称为人文地理特征， 例如，人口、交通、经济、宗教、国家政策等。 量化描述地理特征的数值，称为地理特征的属性 值，例如，100 万是“人口”特征的一个属性值。所 要讨论属性值涉及的范围，叫做该属性值的论域。下 面，我们用地理特征的属性值来定义“背景数据”。 定义 4：设 g1, g2, ..., gn 是 n 个被观测单元，o 是 一个空白单元，记 g={ g1, g2, ..., gn, o}. (5) 设 z 是多个特征的属性值向量， }{ 21 ogggg z,z,,z,zz n  (6) 是地理特征属性向量的集合。如果 zg 中的数据是媒 介，称 zg 为背景数据集，简称背景数据。 例如，用“人口”、“人均 gdp”和“相对暴 露度”等数据，依据被观测单元的灾情，我们能对空 白单元的灾情，进行估值。此时，“人口”、“人均 gdp”和“相对暴露度”等就是背景数据。 依据定义 4 可知，一个属性值的数据是否为另一 个属性值的背景数据，由两个因素决定，一是能否发 挥制约作用；二是能否容易获得。当我们研究一个地 区的洪水灾情时，各地理单元上的水灾程度，受地理 单元的位置、人口和经济发展程度制约，而且这些信 息很容易得到。地理位置，常常决定了地理单元暴露 于洪水的程度。河流流经区域较大的地理单元，其暴 露度较高。“相对暴露度”是由地理位置转化而来的， 具有地理位置的信息。 至此，我们可以提出借助背景数据在地理空间上 进行信息扩散的形式化定义如下： 定义 5：设 w 是用于识别区域 g 上的现象 f 的 不完整数据集，zg 是背景数据。如果模型 能用 zg 使 w 变成完整数据，则称 用 zg 对 w 的信息在 g 上 进行了扩散。 显然，任何内插算法都不能在地理空间上进行信 息扩散，因为背景数据在内插法中发挥不了作用。 brunsdon 等人提出的地理加权回归方法(gwr)[27]，本 质上是一个地理空间上的信息扩散模型。 gwr 是一种空间预测模型[28]，基本表述式为： )()()()( 10 v,uxv,ubv,ubv,uy  (7) 式中，y 表示服从高斯分布的因变量，x 是自变量，u 和 v 是数据的坐标，b0 是截距系数，b1 是斜率系数， 是随机误差项。多元 gwr 用式(8)表之。. i m k ikiikiii xvuvuy    1 0 ),(),( (8) 式中，yi 表示在位置 i 处的因变量， )(0 ii v,u 是在 i 处的截距系数，xik 是在位置 i 处的第 k 个自变量的值， )( iik v,u 是第 k 个自变量的局部回归系数。在 gwr 模型中， )( ii v,u 表示笛卡尔点坐标， i 为随机误差 项。 事实上，gwr 模型仍是一个线性回归模型，通 常用最小二乘法来估计模型中的系数。gwr 当然也 可拓展为广义的线性模型，例如 logistic 回归和 poisson 回归[29]。如果我们清楚地知道什么类型的函 数可以表达 y 和 x 之间的关系，无论它是多么的高度 非线性或非单调，我们都可以用此函数来改造 gwr 模型，达到最佳的回归效果。然而，问题在于特别 难以确定哪种非线性函数适合于表达特定的地球表 面现象。逃避这一问题的最常见做法，是想当然地 人为假设出某种函数；略为装饰一下的做法是追加 统计检验。 为解决不依靠假设函数而识别非线性关系的问 题，人们提出了人工神经网络（ann）模型。由于 训练后的 ann，能够担当定义５中模型  的角色， ann 可以作为一个地理空间上的信息扩散模型使用。 理论上，只要有足够多的隐含层神经元，ann 能以 任意精度逼近任何一个连续函数 [30] 。现实中，由于 ann 是在计算机上学习训练样本，因此任意精度逼 近的事实并不会出现 [31] 。更致命的是，如果训练样 本由于随机干扰等而出现矛盾冲突，则 ann 不收敛 [31] 。 为了填补地理空间上的数据空白，用 gwr 模型 或 ann 模型，在背景数据的帮助下，我们都可以将 已观测数据的信息，扩散到空白单元中去，将地理 空间上的不完整数据，变为完整数据。但是，当遇 到非线性问题或矛盾样本时，由这两种模型推测出 的空白单元中的待观测数据，可能与客观数据相差 过大，失去使用价值。为解决这一问题，我们建议 用基于信息扩散理论的自学习离散回归(sldr)模型[7] 来实现地理空间上的信息扩散。 5. 自学习离散回归模型 不失一般性，假设研究区域 g 由 n-q 个被观测单 元 g1, g2, ..., gn-q，和 q 个空白单元 gn-q+1, ..., gn 组成，即， g={ g1, g2, ..., gn-q, gn-q+1, ..., gn} (9) 更进一步，假设 t 个地理特征的属性值是背景数 据，地理单位 gi 第 j 个特征的属性值是 zij。于是，关 于 g 上的信息可由如表 1 示之。 表 1. 研究区域 g 上的观测值和背景数据 地理单元 背景数据 观测值 g1 z11 z12 ... z1t w1 g2 z21 z22 ... z2t w2 ... ... ... gn-q zn-q1 zn-q2 ... zn-qt wn-q gn-q+1 z n-q+1,1 z n-q+1,2 ...z n-q+1,t unknown ... ... ... gn zn1 zn2 ... znt unknown 统计回归的经验告诉我们，除非用于回归的样 本，其容量 n-q30(t+1)，或者我们知道背景数据和 观测值的因果关系函数类型（例如，线性函数或指 数函数），否则 gwr 模型填补出来的数据不可信； 训练 ann 的经验告诉我们，除非用于训练的样本， 即表１中非 unknown 行构成的样本，内中没有矛盾 冲突，否则 ann 模型在不能收敛情况下填补出来的 数据也不可信。 sldr 模型，则能解决 gwr 和 ann 面临的上 述问题。该模型由“构建关系矩阵”和“用背景数 据推测待观测数据”两部分构成。 5.1.用背景数据和已观测数据构建关系矩阵 设 x 是定义在集合上的一个变量。如果我们在 讨论 x 时所关注的是上的一个子集 u，则称 u 是 x 的论域。当 u 用于接收观测值所扩散的信息时，称 u 是监控空间。 设uj， j=1,2,...,t，是用于扩散背景数据中第j个 地理特征属性值的监控空间，而ut+1是用于扩散已观 测数据的监视空间。令=t+1，则背景数据监控空间 和已观测数据监视空间构成了一个维监控空间： .21 uuu   (10) 式中， .,,2,1},,,,{ 21   juuuu jjmjjj 令=n-q。从表 1 中，我们得到容量为的维样 本 x： },,2,1|),,,,{( 121     ixxxxx iiii (11) 式中， .,,2,1,,,,, 12211     iwxzxzxzx iiitiiiii 对于维样本点， ,),,,( 21 xxxx iiii  x (12) 和维监控点， ,),,,( 2121 21   uuuuuu kkk  u (13) （此处 ,,2,1},,,2,1{   jmk jj ），我们用式(14) 的维正态扩散公式，将 x 的信息扩散到 .u .] 2 )( exp[),( 1 2 2       j j jkij i h ux jux (14) 式中的扩散系数 hj, j=1,2,..., 根据表１中的背景数据 和已观测数据，分别用式(15)进行计算。 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 8                    11. 1),)/(2.6851( 10; ),( 0.2986 9; ),( 0.3362 8; ),( 0.3860 7; ),( 0.4560 6; ),( 0.5690 5; ),( 0.8146        ab ab ab ab ab ab ab h j (15) 式中， }.{min},{max 11 ij i ij i xaxb    令         1 1 2 2 ]. 2 )( exp[ 21 i j j jkij kkk h ux q j (16) 我们获得了一个 uuu  21 上的，关于 x 的信息 矩阵，如式(17)所示:  mmmmkkkkq   121121 }{ q (17)  },,2,1{  mk  ，令 },{max 121 11 1    kkkk mj mk k qs jj      (18) 和 . s q r k kkkk kkkk    121 121      (19) 我们可构造出表 1 中背景数据和观察之间的关系矩 阵，记为：  mmmmkkkk rr   121121 }{  (20) 5.2.用背景数据推测待观测数据 设 ),,,( 21 tzzz z 为表 1 中空白单元的背景数据， 且 .),,,( 1211211 121      uuuuuu kkk u 我们可以用式(21)的-1 维正态扩散公式将此 z 变为论 域 121  uuu  上的一个模糊集，并取其隶属度 最大值用式(22)进行归一化。 . h uz , j j jkj j      1 1 2 2 1 ] 2 )( exp[)(   uz (21)                        1 1 2 2 11 1 ]. 2 )( exp[ },{max , 121 121 121 121       j j jkj kkk kkk j mk kkk kkk h uz q qs s q a j jj     (22) 此模糊集记为 a ~ ，其各隶属度值是： .1,,2,1,,,2,1, 121      jmka jjkkk 对于模糊输入 a ~ ，使用近似推理公式（23）， 我们可以得到一个具有隶属函数 )(   kb u 的模糊输出 b ~ 。 }.min{max)( 121121 11 1    kkkkkkk j mk kb r,au jj      (23) 最后，使用式（24）的重心方法，我们获得了一 个分明值 w： . )( )( 1 1                m k kb m k kkb u uu w (24) 由上述公式(14)-(24)组成的模型，被称为自学习 离散回归（sldr）模型。 一个关于洪水灾区中空白单元损失数据推测的 研究表明，相比地理加权回归法，本文提出的自学 习离散回归信息扩散法，能减少大约 60％的误差。 6. 结论与讨论 面对地理学中的数据不完整问题，人们对插值和 地理加权回归（gwr）等方法，进行了大量卓有成 效的研究，解决了一系列现实问题。但当地理单元较 大时，大多数地理特征的数属性值在相邻单元上并不 连续，插值法对变量连续性的基本要求得不到满足， 插值得到的数据，没有意义。而当所研究的地理学现 象过于复杂时，任何假设出来的，描述此现象的函数 类型，都无法证实。基于不能证实的假设函数进行的 地理加权回归，统计结果的可靠性，疑问重重。 由于将观测值变模糊集的信息扩散方法，能部分 弥补小样本缺陷，提高系统识别精度，似乎我们可以 将信息扩散方法用于相邻地理单元上的信息扩散，由 几个相邻地理单元上的已知数据，推导出空白单元上 的数据，使研究区域的不完整数据，变成完整数据。 遗憾的是，这条路不通，因为传统的信息扩散模型， journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 8 9 是在所处理观测值的论域上，将观测值变模糊集。当 观测值是随机样本点时，此论域，是随机事基本空间。 因此，传统的信息扩散，是在概率空间中进行，而不 是在地理空间上进行。 在传统信息扩散理论框架之内，本文提出了，以 背景数据为桥梁，将已观测数据携带的信息，扩散到 空白单元的方法，称为借助背景数据的地理空间信息 扩散技术，并给出了自学习离散回归（sldr）模型。 从原理上讲，sldr 与 gwr 及人工神经网络 （ann）一样，都属于统计回归方法，但 sldr 不 仅能统计回归出非线性关系，而且容忍观测数据间存 在矛盾，而不会出现 ann 的收敛问题。 案例研究表明，sldr 的估计比地理加权回归法 减少了大约 60％误差，优势明显。 概率空间中的信息扩散模型，自然要比地理空间 上的信息扩散模型简单，因为前者是无约束扩散，后 者是有约束。随机样本在扩散模型中不受样本以外的 因素影响，地理数据在扩散模型中则受到背景数据的 约束。少数几个背景数据支持下的信息扩散，其结果 与大量背景数据时的肯定不同；制约程度较高的背景 数据支持下的信息扩散，其修补得到的完整数据的质 量，肯定高于制约程度较低的质量。 地理空间上的信息扩散的研究，为信息扩散理论 和方法的发展，提供了巨大的空间。 背景数据支持下的，地理空间上的信息扩散，本 质上是某种“联想”。人类通过联想对事物进行判断， 成功与否，既与经验和知识有关，也与联想者的智能 水平有关。地理空间上信息扩散中用到的背景数据和 给定的不完整数据，相当于联想者的经验和知识；扩 散模型的好坏，相当于联想者智能水平的高低。 在人工智能席卷全球的今天，地理空间上信息扩 散理论和方法的研究，必定能在智慧地学的建设中发 挥重要作用。 参考文献 [1] 傅伯杰. 新时代自然地理学发展的思考. 地理科学进展 , 2018, 37 (1) :1-7. [2] 张勇, 许力生, 陈运泰. 2008 年汶川大地震震源机制的时 空变化. 2009, 52 (2): 379-389. [3] 王晓青, 丁香, 王龙等. 四川汶川 8 级大地震灾害损失快 速评估研究. 地震学报, 2009,31(2):205-211. [4] eldrandaly k a，abu-zaid m s. (2011). comparison of six gis-based spatial interpolation methods for estimating air temperature in western saudi arabia. journal of environmental informatics, 2011, 18(1), 38–45. [5] huang c f. principle of information diffusion. fuzzy sets and systems, 1997, 91(1): 69-90. [6] 黄崇福. 地理空间上的信息扩散及其在风险分析中的应 用. 一带一路背景下的风险分析和危机反应---中国灾害 防御协会风险分析专业委员会第八届年会论文集(西安， 2018 年 10 月 20-21 日), 1-7. [7] huang c f, shi y. towards efficient fuzzy information processing: using the principle of information diffusion, heidelberg, germany: physicaverlag (springer), 2002. [8] huang c h, zong t, chen z f. four models to calculate a fuzzy probability distribution with a small sample. international journal of information technology & decision making, 2007, 6(4): 611–623 [9] chen q, rui h, li w, zhang y. analysis of algal bloom risk with uncertainties in lakes by integrating self-organizing map and fuzzy information theory. science of the total environment, 2014, 482–483(1): 318–324. [10] feng l, luo g, application of possibility-probability distribution in risk analysis of landfall hurricane—a case study along the east coast of the united states. applied soft computing, 2011, 11 (8): 4563–4568. [11] zou q, zhou j, zhou c, guo j, deng w, yang m, liao l. fuzzy risk analysis of flood disasters based on diffusedinterior-outer-set model. expert systems with applications, 2012, 39 (6): 6213–6220. [12] liu x, zhang j, cai w, tong z. information diffusionbased spatio-temporal risk analysis of grassland fire disaster in northern china. knowledge-based systems, 2010, 23 (1): 53–60. [13] zhong l, liu l, liu y. natural disaster risk assessment of grain production in dongting lake area, china. agriculture and agricultural science procedia, 2010, 1(1): 24-32. [14] li q, zhou j, liu d, jiang x. research on flood risk analysis and evaluation method based on variable fuzzy sets and information diffusion. safety science, 2012, 50(1): 12751283. [15] olya h g t, alipour h. risk assessment of precipitation and the tourism climate index. tourism management, 2015, 50: 73-80. [16] xu l, xu x, meng x. risk assessment of soil erosion in different rainfall scenarios by rusle model coupled with information diffusion model: a case study of bohai rim, china. catena, 2012, 100(2): 74-82. [17] chu y y, dong w l, li y, liang d. risk prediction model of lng terminal station based on information diffusion theory. procedia engineering, 2013, 52: 60-66. [18] hao l, yang l, gao j m. the application of information diffusion technique in probabilistic analysis to grassland biological disasters risk. ecological modelling, 2014, 272: 264-270. [19] xing l, lu k. the importance of public-private partnerships in agricultural insurance in china: based on analysis for beijing. agriculture and agricultural science procedia, 2010, 1(1): 241-250. [20] zou q, zhou j, zhou c, song l, guo j, liu y. the practical research on flood risk analysis based on iiosm and fuzzy α-cut technique. applied mathematical modelling, 2012, 36(7): 3271-3282. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 2–10 10 [21] xu l, liu g. the study of a method of regional environmental risk assessment. journal of environmental management, 2009, 90(11): 3290-3296. [22] mako z. approximation with diffusion-neural-network. proceedings of the 6th international symposium of hungarian researchers on computational intelligence (november 18-19, 2005, budapest), 2005, pp. 589-600. [23] yan j p, li s s, bai j, liu x y. the spatial symmetry axis of earthquake hazard in china. journal of risk analysis and crisis response, 2013, 3(1): 59-64. [24] cartier s, vallette c, mediene h. urban seismomorphoses seismic vulnerabilities, an embarrassing legacy. journal of risk analysis and crisis response, 2012, 2(2): 96-106. [25] tobler w r. a computer movie simulating urban growth in the detroit region. economic geography, 1970, 46:234-240. [26] eldrandaly k a, abu-zaid m s. comparison of six gisbased spatial interpolation methods for estimating air temperature in western saudi arabia. journal of environmental informatics, 2011, 18(1): 38–45. [27] brunsdon c, fotheringham a s, charlton m e. geographically weighted regression: a method for exploring spatial nonstationarity. geographical analysis, 1996, 28(4): 281298. [28] lieske d j, bender d j. a robust test of spatial predictive models: geographic cross-validation. journal of environmental informatics, 2011, 17(2): 91-101. [29] fotheringham a s, brunsdon c, charlton m. geographically weighted regression: the analysis of spatially varying relationships, chichester, usa: john wiley & sons, 2002. [30] hornik k, stinchcombe m, white h. multilayer feedforward networks are universal approximators. neural networks, 1989, 2 (5): 359–366. [31] wray j, green, g g r. neural networks, approximation theory, and finite precision computation. neural networks, 1995. 8 (1): 31–37. [32] huang c f, moraga c. a diffusion-neural-network for learning from small samples. international journal of approximate reasoning, 2004, 35: 37–161. research article five supplementary therapies for covid-19 li-yuan liu* college of life science, beijing normal university, china 1. introducton covid-19 is sars-like infectious disease caused by a new type of coronavirus, named sars-cov2 or 2019-ncov. this new viral pneumonia began to outbreak in wuhan, china, in december 2019, and is spreading all over the world in the present after 3 months [1–5]. there have non-specific medications, resulting of a longer period of treatment and higher fatality rate. therefore, combining therapies, alternative medicines and supplementary methods are necessary and helpful. covid-19 is characteristic of both damage of the respiratory system and the immune system or both similarities of sars and aids. the sars-like symptoms result of pneumonia with fever, cough and dyspnea [1–4]. the aids-like symptoms result in damage of the lymphatic system and include non-reaction, non fever and significant decrease of numbers of lymphocytes and platelets in the blood [3,6]. recognition of the latter is great important for clinical treatment and judge of prognosis. for such reason of syndrome, we more agree the name of covid, but not the cov-pneumonia. 2019-covsyndorme or sars-ii is preferable than all present names. 2. face mask is not adequate and sputum suction is useful for a patient face masks with a special air filter designed to protect from tiny airborne particles. when a person of covid-19 coughs/exhales, the internal virus usually spread to environment through small droplets from the nose/mouth. in the present, there is not any specific antiviral drug to kill the virus reproduced in the body. therefore, for a patient, inhaling viruses are harmful and exhaling viruses are favorable. wearing a mask every 24 h must increase respiratory resistance obviously and block to exhale viruses completely, which is injurious to the patient, especially with dyspnea in the middle and severe cases. thus, avoiding wearing a mask is considerable benefit for the treatment of patients. recently, author of this paper have designed and made a new apparatus. the exhaled airs from patients are collected by a big mask and sent into the apparatus, in which any viruses and bacteria can be completely killed by more than 150°c high temperature. therefore, the patients without wearing the face mask can breathe freely, even more save effort, but do not worry to spread viruses to around environment and other people. let patients cough up sputum is not only for keeping the airway open, but also for discharging viruses. note, do not inhibit cough, but stimulate to cough. some herbs in chinese traditional medicine have such action. one main element of herbs is chinese ephedra that may be replaceable by ephedrine as a western drug. sputum suction is useful for clearing secretions and viruses, and keeps the airway open. for severe patients and near-death patients, suction can save lives! as a fact, when conducted sputum suction through inserting pipe with machine in wuhan, the died numbers half decreased from that day (february 23). 3. semireclining posture is effective for respiration and inflammation semireclining posture is usually used instead of lying supine in patients with heart failure for lightening the load of the heart a r t i c l e i n f o article history received 23 april 2020 accepted 09 october 2020 keywords covid-19 therapy sputum suction semireclining posture coldrex vitamin b2 so2 aspirin dexamethasone a b s t r a c t the covid-19 is characteristic of both damage of the respiratory system and the immune system or both similarities of sars and aids. these features decide difficulty of specific treatment. the worldwide transmission and higher fatality rate call for new skills for prevention and treatment. here, we recommend five supplementary therapies for covid-19. (1) discharging viruses from lungs by exhaling, coughing and suction are firstly emphasized. (2) semireclining posture instead of lying supine for lightening hyperemia and edema of inflammation and dyspnea. (3) non-steroidal anti-inflammatory coldrex, such as tylenol and paracetamol, can be safely used for controlling inflammation. (4) vitamin b2 must be used for repairing of mucous membrane of the respiratory tract and as antioxidants. (5) it is most emphasized that sulfur dioxide (so2) produced from burning sulfur and explosion of firecrackers has a potent antimicrobial effect for air disinfectant and killing viruses in the lungs. in addition to the general therapies, all these methods or therapies can be simply and effectively supplied to patients for discharging viruses, inhibiting viruses and killing viruses, and for recovery from damages. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *email: liu.liyuan@bnu.edu.cn journal of risk analysis and crisis response vol. 10(3); october (2020), pp. 97–100 doi: https://doi.org/10.2991/jracr.k.201022.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:liu.liyuan%40bnu.edu.cn?subject= https://doi.org/10.2991/jracr.k.201022.001 https://www.atlantis-press.com/journals/jracr 98 l.-y. liu / journal of risk analysis and crisis response 10(3) 97–100 through decreasing returned blood volume from the periphery to the heart. for pneumonia patients, semireclining posture can lighten hyperemia and edema of inflammation by decreasing returned blood volume too. also, semireclining posture can relieve constriction of the chest wall and make the rib and diaphragm move more freely, which let the breath become easily and effectively. both two actions are helpful for the treatment of pneumonia. actually, however, the severe patients always lie supine in the bed with difficulty breathing day and night. as used results, when the supine patients sit up, their dyspnea improved promptly. 4. coldrex for anti-inflammation like the sars, the main fatality of the damage in covid-19 resulted from the overwhelming chemical counterattack from the immune system, with excessive activated lymphocytes. this cytokine storm, as it is called, is intended by body to kill the virus; but in a weakened state of patients, it can cause inflammation, leaky blood vessels and even pneumonia, a hallmark of sars and covid-19. they are medications that effectively fight inflammation in the body. steroids and non-steroidal anti-inflammatory drugs (nsaids) work by reducing inflammation, swelling, and mucus production in the airways of a person. steroids keep the immune system from making substances that cause inflammation and slow or stop the immune system processes that trigger inflammation. steroids suppress inflammation and reduce the signs and symptoms of inflammatory conditions, which can help control conditions in which the immune system mistakenly attacks its own tissues. but suppressing of immune system inhibits the normal function that producing anti-body and sensitized t-cells to kill the virus, which is a sharp contradiction. more seriously, covid-19 is not only like the sars but also like the aids in that the immune system is attacked and damaged. therefore, steroids should be avoided to be used as much as possible. for these reasons, the nsaids are firstly considered instead of steroids to reduce inflammation in the lungs. all nsaids work by reducing the levels of prostaglandins, chemicals that are released when there is inflammation and injury, and that cause pain and fever. nsaids block the enzyme (cyclooxygenase) that makes prostaglandins, resulting in lower concentrations of prostaglandins. as a consequence, inflammation, pain, and fever are reduced. in the nsaid, aspirin is a classic drug. paracetamol is a viable alternative to the nsaids, especially because of the low incidence of adverse effects. paracetamol also specifically inhibits concentration of prostaglandins [7]. we early and firstly defined covid-19 as a cold, a severe influenza, and suggested to use coldrex, tylenol, for anti-allergy and anti-inflammatory. the nsaids and paracetamol are widely used main elements of coldrex. therefore, coldrex is good selection for reducing the inflammation and fever of covid-19. actually, some people treated themselves as the common cold and taken orally coldrex when felt fever. after recovered, they were checked to be positive sars-cov2. thus, nsaids can be widely and safely used in all patients without worrying any steroids-like adverse effect. tylenol (acetaminophen), paracetamol or their compound preparation is firstly selected coldrex. 5. vitamin b2 for repairing damage in the mucous membrane of respiratory tract vitamin b2 (v-b2), also called riboflavin, is a water-soluble vitamin. water-soluble vitamins are stored in the body in very limited amounts and are excreted through the urine. therefore, daily supplement is good. riboflavin is required to release energy from protein, carbohydrate and fat. it is also involved in the transport and metabolism of iron in the body and is needed for the normal structure and function of mucous membranes and skin. v-b2 health benefits include maintaining energy levels, protect the digestive tract, offers a powerful punch of antioxidants, and preventing aids/hiv. riboflavin deficiency cause dermatoses and mucous membrane lesions, such as in the mouth, tongue, eyes and skin. the symptoms include: stomatitis (oral ulceration), glossitis, reddening of the eyes and dermatitis [8,9]. for protecting the mucous membrane and repairing the damage of the respiratory tract in the inflammation, conventional applications of v-b2 (30–60 mg/day) must be helpful for cooperating roles of treatment. as ever used results, over 30 children with pneumonia had been quickly cured. also, vitamin c must be conventional used every day. 6. sulfur dioxide (so 2 ) for disinfectant and killing viruses sulfur dioxide (so₂) is a colorless gas with an irritating, pungent odor and toxic in large amounts. exposure to so2 may cause irritation to the eyes, nose, and throat. although its chief uses are in the preparation of sulfuric acid, sulfur trioxide, and sulfites, so2 also is used as a disinfectant, an insecticide, a reducing agent, a bleach, and a food preservative. 6.1. endogenous so 2 and the function so2 is endogenously generated through the aspartate aminotransferase pathway. the role of so2 in mammalian biology is not yet well understood. new studies have suggested various potential mechanisms. so2 is a novel endogenous gaseous signaling molecule, which plays a significant physiological role in regulating cardiac and blood vessel function, and aberrant or deficient so2 metabolism can contribute to several different cardiovascular diseases [10,11]. it was shown that endogenous so2 plays a role in diminishing an experimental lung damage caused by oleic acid: lowered lipid peroxidation, free radical formation, oxidative stress and inflammation [12]. 6.2. antimicrobial role of so 2 so2 is a broad-spectrum antimicrobial agent that has an inhibitory effect on a wide variety of microorganisms. moreover, inhalation l.-y. liu / journal of risk analysis and crisis response 10(3) 97–100 99 of so2 can inhibit influenza virus, etc. [13–15]. it has been understood since the early 1900s that only the free forms of so2 (and not the bound) have an antimicrobial effect. it was further discovered in the 1960s that molecular so2 was several hundred times more effective than bisulfite. the mechanism for antimicrobial effect of so2 works by the so2 entering the microbe and disrupting the activity of the enzymes and proteins of the cell. since only the molecular form of so2 can enter through the cell membrane, it is the concentration of molecular so2 that controls microbial growth [16]. in the 1911, severe plague occurred in the northeast of china. in a town, one quarter residents had died. the government encouraged residents to set off firecrackers in the chinese new year. after then, the death number decreased rapidly to zero. recent years, a lot of farmers in china claimed to overcome african swine fever and bird flu after setting off firecrackers in the hogcote or chicken coop. firecracker explosion produces so2, no2 and carbon particles. both so2 and no2 can kill bacteria and viruses, and carbon particles can absorb microorganisms [17]. the sars-cov2 spread through air. but most disinfectants are liquid sprayed to surface of objects, which are not efficient to kill viruses floating in the air. the aerosol transmission is the route of infection for most unknown source. therefore, indoor/outdoor air disinfection is indispensable for blocking viral transmission. in the present, so2 is almost the only adapted air disinfectant, and also the cheapest product and the simplest operation. in the same time, inhaling so2 can kill viruses in the lung cavity and lung tissues. so2 can be produced by burning sulfur (figure 1) and setting off firecrackers. we suggest to set off firecrackers in the outdoor and burn sulfur in the room. in the usage of the first time, people leave one room: closing the door and window; burning sulfur 1 g/10 m2; keeping closing more than 3 h. after then, sulfur is burnt 0.2 g/10 m2 every day and people can remain in the room at the same time. patients can stay in the sickroom with lower concentration of so2 all the time. when without specific medications and with numerous infected people, so2 should become one important killer of viruses, especially in the backward countries and regions. 7. discussion and summary currently, there is no vaccine to prevent the spread of coronavirus and no specific medications to kill the coronavirus. for treatment of covid-19, supportive treatments include supplementary liquids, reducing fever and supplemental oxygen and respirator. treatment options are currently being investigated around the world. there is some evidences that certain medications may have the potential to be effective with regard to preventing illness or treating the symptoms of covid-19. remdesivir is hotly trialed and recommended [18]; a combination of seemed un-related plaquenil and azithromycin appeared good effect to clean the coronavirus in the body [19]; and a combination of two anti-hiv medications, lopinavir and ritonavir, had a significant reduction in the levels of the coronavirus. a uk trial of the steroid dexamethasone (6 mg/day, 10 days) was confirmed its life-saving benefits for covid-19 patients on ventilators [20]. among patients on ventilators, the rate of death for patients on the drug was 29.3% compared to 41.4% on those without. however, among the group who were not receiving any oxygenation at the time the trial started, 17.4% on the steroid died compared to 14% who did not receive it—suggesting the drug increased their mortality risk. i am surprised to see these results because it is general knowledge to inject large amount of dexamethasone (e.g. 5–10 mg) once in a while for severe patients with sars, covid-19 and other infections. apparently, it is never to be applied to mild patients. as we emphasize above that steroids should be avoided, and here suggest combination of oral coldrex and injection of large dose of steroids for severe patients once in a while. more exciting progress is anticoagulation (e.g. aspirin) in part because the virus can cause severe blood clots. among coronavirus patients who were placed on ventilators, those who received blood thinners died less often than those who did not receive blood thinners [21]. in addition to the above “standard” and new therapies, we recommend those five methods or therapies that can be effectively supplied to patients for discharging viruses, inhibiting viruses and killing viruses, and for recovery from damages. after all, we have to escape and evacuate for physical quarantine and we must insist waiting for coming of hot summer, for making vaccines and for discovering specific medications in the spring 2020 (when write this paper). actually, the past summer helped to control the epidemic in china and many other counties effectively, and keep steady in america and a few other countries (when revise this paper). world health organization and many other experts warned again and again that a second wave of epidemic will come at the autumn and winter. the vaccine is on the test-way yet and non-specific drug was found in the past three-quarters. thus then, these therapies can continue to help treatment on covid-19. conflicts of interest the author declares no conflicts of interest.figure 1 | burning medical sublimation sulfur in the bottom of a bowl. 100 l.-y. liu / journal of risk analysis and crisis response 10(3) 97–100 references [1] wu f, zhao s, yu b, chen ym, wang w, song zg, et al. a new coronavirus associated with human respiratory disease in china. nature 2020;579:265–9. [2] li q, guan x, wu p, wang x, zhou l, tong y, et al. early transmission dynamics in wuhan, china, of novel coronavirus infected pneumonia. n engl j med 2020;382:1199–207. [3] guan wj, ni zy, hu y, liang wh, ou cq, he jx, et al. clinical characteristics of 2019 novel coronavirus infection in china. n engl j med 2020;382:1708–20. [4] huang c, wang y, li x, ren l, zhao j, hu y, et al. clinical features of patients infected with 2019 novel coronavirus in wuhan, china. lancet 2020;395:497–506. [5] cohen j, kupferschmidt k. strategies shift as coronavirus pandemic looms. science 2020;367:962–63. [6] xu z, shi l, wang y, zhang j, huang l, zhang c, et al. pathological findings of covid-19 associated with acute respiratory distress syndrome. lancet respir med 2020;8:420–2. [7] hyllested m, jones s, pedersen jl, kehlet h. comparative effect of paracetamol, nsaids or their combination in postoperative pain management: a qualitative review. br j anaesth 2002;88:199–214. [8] roe da. riboflavin deficiency: mucocutaneous signs of acute and chronic deficiency. semin dermatol 1991;10:293–5. [9] barthelemy h, chouvet b, cambazard f. skin and mucosal manifestations in vitamin deficiency. j am acad dermatol 1986;15:1263–74. [10] chen ss, tang cs, jin hf, du jb. sulfur dioxide acts as a novel endogenous gaseous signaling molecule in the cardiovascular system. chin med j 2011;124:1901–5. [11] liu d, jin h, tang c, du j. sulfur dioxide: a novel gaseous signal in the regulation of cardiovascular functions. mini rev med chem 2010;10:1039–45. [12] chen s, zheng s, liu z, tang c, zhao b, du j, et al. endogenous sulfur dioxide protects against oleic acid-induced acute lung injury in association with inhibition of oxidative stress in rats. lab invest 2015;95:142–56. [13] fairchild ga. effects of ozone and sulfur dioxide on virus growth in mice. arch environ health 1977;32:28–33. [14] fairchild ga, roan j, mccarroll j. atmospheric pollutants and the pathogenesis. arch environ health 1972;25:174–82. [15] wakabayashi m, bang bg, bang fb. mucociliary tran sport in chickens infected with newcastle disease virus and exposed to sulfur dioxide. arch environ health 1977;32:101–8. [16] henderson p. sulfur dioxide: science behind this anti-microbial, anti-oxidant, wine additive. pract winery vineyard j 2009. available from: https://beerbrew.com/words/wp-content/uploads/2014/05/ so2science.pdf. [17] buckley rd, loosli cg. effects of nitrogen dioxide inhalation on germfree mouse lung. arch environ health 1969;18: 588–95. [18] wang m, cao r, zhang l, yang x, liu j, xu m, et al. remdesivir and chloroquine effectively inhibit the recently emerged novel coronavirus (2019-ncov) in vitro. cell res 2020;30:269–71. [19] gautret p, lagier jc, parola p, hoang vt, meddeb l, mailhe m, et al. hydroxychloroquine and azithromycin as a treatment of covid-19: results of an open-label non-randomized clinical trial. int j antimicrob agents 2020;56:105949. [20] recovery collaborative group, horby p, lim ws, emberson jr, mafham m, bell jl, et al. dexamethasone in hospitalized patients with covid-19 — preliminary report. n engl j med 2020:nejmoa2021436. [21] paranjpe i, fuster v, lala a, russak aj, glicksberg bs, levin ma, et al. association of treatment dose anticoagulation with in hospital survival among hospitalized patients with covid-19. j am coll cardiol 2020;76:122–4. https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1038/s41586-020-2008-3 https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1056/nejmoa2001316 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1056/nejmoa2002032 https://doi.org/10.1016/s0140-6736(20)30183-5 https://doi.org/10.1016/s0140-6736(20)30183-5 https://doi.org/10.1016/s0140-6736(20)30183-5 https://doi.org/10.1126/science.367.6481.962 https://doi.org/10.1126/science.367.6481.962 https://doi.org/10.1016/s2213-2600(20)30076-x https://doi.org/10.1016/s2213-2600(20)30076-x https://doi.org/10.1016/s2213-2600(20)30076-x https://doi.org/10.1093/bja/88.2.199 https://doi.org/10.1093/bja/88.2.199 https://doi.org/10.1093/bja/88.2.199 https://pubmed.ncbi.nlm.nih.gov/1764356/ https://pubmed.ncbi.nlm.nih.gov/1764356/ https://doi.org/10.1016/s0190-9622(86)70301-0 https://doi.org/10.1016/s0190-9622(86)70301-0 https://doi.org/10.1016/s0190-9622(86)70301-0 https://doi.org/10.3760/cma.j.issn.0366-6999.2011.12.024 https://doi.org/10.3760/cma.j.issn.0366-6999.2011.12.024 https://doi.org/10.3760/cma.j.issn.0366-6999.2011.12.024 https://doi.org/10.2174/1389557511009011039 https://doi.org/10.2174/1389557511009011039 https://doi.org/10.2174/1389557511009011039 https://doi.org/10.1038/labinvest.2014.147 https://doi.org/10.1038/labinvest.2014.147 https://doi.org/10.1038/labinvest.2014.147 https://doi.org/10.1038/labinvest.2014.147 https://doi.org/10.1080/00039896.1977.10667249 https://doi.org/10.1080/00039896.1977.10667249 https://doi.org/10.1080/00039896.1972.10666157 https://doi.org/10.1080/00039896.1972.10666157 https://doi.org/10.1080/00039896.1977.10667263 https://doi.org/10.1080/00039896.1977.10667263 https://doi.org/10.1080/00039896.1977.10667263 https://beerbrew.com/words/wp-content/uploads/2014/05/so2science.pdf https://beerbrew.com/words/wp-content/uploads/2014/05/so2science.pdf https://doi.org/10.1080/00039896.1969.10665457 https://doi.org/10.1080/00039896.1969.10665457 https://doi.org/10.1080/00039896.1969.10665457 https://doi.org/10.1038/s41422-020-0282-0 https://doi.org/10.1038/s41422-020-0282-0 https://doi.org/10.1038/s41422-020-0282-0 https://doi.org/10.1016/j.ijantimicag.2020.105949 https://doi.org/10.1016/j.ijantimicag.2020.105949 https://doi.org/10.1016/j.ijantimicag.2020.105949 https://doi.org/10.1016/j.ijantimicag.2020.105949 https://doi.org/10.1056/nejmoa2021436 https://doi.org/10.1056/nejmoa2021436 https://doi.org/10.1056/nejmoa2021436 https://doi.org/10.1056/nejmoa2021436 https://doi.org/10.1016/j.jacc.2020.05.001 https://doi.org/10.1016/j.jacc.2020.05.001 https://doi.org/10.1016/j.jacc.2020.05.001 https://doi.org/10.1016/j.jacc.2020.05.001 microsoft word a gis-based micro-simulation queue model for vehicle evacuation.doc a gis-based micro-simulation queue model for vehicle evacuation jian ma department of civil and architectural engineering, city university of hong kong, tat chee avenue, kowloon tong, hong kong. school of transportation and logistics, southwest jiaotong university, chengdu, china shaobo liu department of civil and architectural engineering, city university of hong kong, tat chee avenue, kowloon tong, hong kong. weili wang department of civil and architectural engineering, city university of hong kong, tat chee avenue, kowloon tong, hong kong. peng lin aecom (hong kong) ltd., new town plaza, shatin, hong kong siuming lo* department of civil and architectural engineering, city university of hong kong, tat chee avenue, kowloon tong, hong kong. abstract in urban areas, people’s life may be threatened by disasters such as earthquakes. to reduce the life risks, efficient responses including evacuation are of critical importance. however, a drill evacuation appears impossible; simulation is thus used to examine the effectiveness of an evacuation plan. considering the capacity of the routes, a gis-simqueue vehicle evacuation model is integrated with a geographic information system. simulation results indicate that the model can reproduce vehicle evacuation features, which are in agreement with empirical ones. keywords: evacuation model, simulation, gis. *bcsmli@cityu.edu.hk 1. background in facing a major event, such as the world expo or the olympic games, a densely populated city may need to establish a crisis response strategy to manage the vehicular traffic and crowd movement under disastrous situations, such as earthquake, nuclear plant accident, floods, tsunami, wildfire, tornadoes or leakage of toxic gases. we understand that absolute prevention of disasters and controlling their spread maybe impossible. thus evacuation of people from the hazardous region(s) is per se a way to reduce the life risks of disasters. inefficient evacuation may cause life lost, e.g., in 1984, thousands were killed in a poisonous gas leak at a union carbide corporation pesticide plant in bhopal, india. while in contrast, the mississauga evacuation in 1979 is an example of successfully response to an emergency. as a consequence, how to perform safe and efficient evacuation becomes a major concern of government, city and building planners, fire fighting and rescue officers, insurers, administrators as well as the people themselves. it is noticed that in 1999, due to the hurricane floyd, the largest evacuation in us history has forced more than 2.6 million residents in florida, north carolina and journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 178-187 published by atlantis press copyright: the authors 178 administrateur texte tapé à la machine received 14 july 2012; accepted 21 september 2012 administrateur texte tapé à la machine j. ma et al. south carolina to move out of their homes. for this kind of large scale evacuation, the residents need to be evacuated by vehicles. the burst flow of vehicles may trigger serious traffic jam. thus, the government should establish adequate emergency response plan to ensure the public safety. to archive this aim, the authorities need to acquire honorable understanding of the evacuation pattern of the vehicles. they can as a result design the spatial setting of buildings with plenty of routes and shelters, and to establish effective management strategies to control the crowd and vehicle traffic1-3. it is noticed that arranging large scale drill exercises seems impossible to evaluate the evacuation efficiency of a large region. thus, most of the previous studies and evaluation methods are qualitative. the effects of different management strategies are difficult to assess quantitatively. however, with the recent advancement of digital computers, simulating the vehicle evacuation process becomes more and more popular and important. some computer programs, such as the netvac4, massvac5,6, rems7, and cemps8,9, paramics10 have been developed. individual behaviors are now attracting researches’ attention, e.g. stern11,12 has studied regional evacuation problem considering the impact of human behavior. although evacuating the people of a whole city is rare, it is not unexpected to evacuate a region in a city in particular facing disastrous events, such as leakage of toxic gases from a chemical plant. if densely populated buildings surround the plant, which is not uncommon in many asian countries, the clearance process will involve the evacuation of vehicles and pedestrians. simulating the evacuation process will as a consequence require understanding of the movement pattern of each individual13 and will provide the crisis response fundamental evaluation of the evacuation efficiency13-15. more recently, a large number of evacuation studies are conducted using well-established dynamic traffic simulation models. these models, including both microscopic models, such as paramics10,14, corsim16, integration17, and macroscopic models, such as dynasmart18, dynamit19, transcad20, and indy21 were developed for regular day-to-day traffic applications. researches begin to pay attention to emergency evacuations, e.g., a strategic dynamic traffic assignment model for hurricane evacuation was developed by brown22. in a number of studies using microscopic models, model parameters describing driving behaviors (such as headway, acceleration, reaction time) have been adjusted for the case of emergency evacuation13,23. summarizing the current studies we can find that most evacuation models developed in the recent decades are “mass” evacuation models which adopt the network flow theory to model the vehicle flow. although this method has the benefit of supporting real time application and simulation of city scale evacuation process, they cannot record the movement pattern of an individual. as a consequence, a microscopic simulation model, which is capable of tracking detailed movement of each vehicle or each pedestrian, is necessary. in this kind of models, real-life influential factors such as the gender difference13, route choice behavior15, the spillback arising from traffic congestion and the breakdowns of vehicle can all be took into account. this article as a result presents a microscopic simulation model, gissimqueue, which is integrated with the geographic information system (gis). this gis-based model attempts to model the evacuation of each vehicle. the current application of gis technology ranges from providing maps to managing sophisticated topological databases to multi-objective decision making as well as intelligent vehicle highway system (ivhs). it can be employed to evaluate, display, send and receive information for the crisis response. information stored in the database includes available routes, state highways, county highways, municipal highways, the distribution of population, cars per county, the position of special facilities such as fire and police stations, rescue centers, and emergency shelters. such information is rudimentary for emergency evacuation planning24-26. in addition, gis can also provide vivid pictures, which can dynamically represent the evolving of the hazard. it should be noticed that gis alone is unable to dynamically model spatial information, it requires simulation model to predict the process. gis database serves as the inputs for the simulation model. the simulation results can provide dynamic information for managing traffic during emergency evacuation. integrating gis with traffic simulation model has been established in many models. for example, silva8,9 has developed a spatial decision support system, which integrates the simulation model with gis for emergency published by atlantis press copyright: the authors 179 a gis-based micro-simulation queue model for vehicle evacuation planning. it adopts the spatial data structure and network modeling algorithm of gis for the simulation model to predict traffic flow through the network under various scenarios. cova and church27,28 adopted the concept of emergency planning zone for modeling community evacuation on a gis platform. the proposed gis-simqueue is established by using c++ on the basis of two gis modules namely netengine and mapobjects. these two modules are provided by environmental system research institute inc. (esri). netengine is a suit of dynamic link libraries (dlls) which is designed to facilitate advanced network analysis. it provides ready-to-use algorithms such as the shortest path algorithm. mapobjects comprises an activex control (ocx) called the map control and a set of activex automation objects to facilitate the management of spatial data and interactive operation. 2. the simulation and gis network in the simulation model, the road network of a region is described in terms of links and nodes. links represent the road and nodes represent the intersections of the roads. a link represents a one-way road, whereas a twoway road is represented by two links separately. every node and link has its unique number to be identified, and also every node stores two link lists, which record the inbound and outbound links. accordingly, it is efficient to traverse though the whole network and to facilitate the route choice at each node (intersection). however, the netengine adopts a different way to describe the road network. in netengine, three elements namely junctions, edges and turns are used to represent the intersections, edges, and the accessibility between intersection and edge. to make full use of the algorithms of netengine, a mapping relation of these two data structures (the simulation model and the gis) has be established. the junction is compatible with node defined in the simulation model. however, the edge behaves differently from link in that one edge can represent the arc being traversed in from–to direction as well as traversed in to–from direction. an edge of two directions, with a common id number, is discriminated by another id number the layer id number. in this circumstance, in order to obtain a one-way link in the simulation model from the netengine, two id numbers is needed to convey to the netengine, and vice versa. 3. a micro-simulation queue model queue model in its simple form assumes that service can be provided at a certain rate and request for service comes with another rate. customers are allowed to wait in line and leave by simple fifo (first-in first-out) rule. if the rate of request for service exceeds the service rate, a queue builds up. gawron29 has introduced a queue model, which is different from traditional queue model in that the number of vehicles leaving a link in a simulation interval is constrained by the section capacity ca of the link as well as by its holding capacity (storage). a road can accommodate a maximum number of vehicles: 5.7/max lanenlengthn ×= (1) where length is the road length, and nlane is the number of lane of road and 7.5 m is taken as a standard space occupied by a vehicle under congestion proposed by nagel30. then the simulation queue model can be described as: at each time step, all vehicles on link move at free speed, and a certain number of vehicles which have arrived at the end of link may leave the link and join to the next link. however, the number of vehicles left is constrained by the section capacity of the link as well as by the storage of destination link. if the destination is full, no vehicle can leave the current link. simon31 has described the procedure as shown in fig. 1. fig. 1. simulation algorithm. published by atlantis press copyright: the authors 180 j. ma et al. this approach considers the delay due to the constraint of section capacity. however, in this approach32 the links are always selected in the same sequence. this may give some links a higher priority than others under congested conditions. simon31 tried to rectify this shortcoming by randomizing the link sequence, while cetin32 proposed to select the sequence of links in accordance with their through capacity. the higher through capacity will be the earlier sequence of being selected. however, an earlier selection cannot ensure an earlier movement. taking the street intersection in fig. 2 as an example, links are selected sequentially in the order of 1, 2, 3, 4. we assume that vehicles on links 1, 2 and 4 are going to move into link 3, which can accommodate a maximum of 3 vehicles at the same time. at step t, 2 vehicles are already waiting on link 3 and only one free space is available. link 1 is selected first and one vehicle on it can enter link 3. link 2 is then selected, but there will be no available space for vehicles on link 2 to enter. subsequently, link 3 is selected, and vehicles on it move ahead, which results in another available free space. accordingly, link 4, the last one being selected, can allow the movement of its vehicles. we can see that link 2 is selected earlier than that of link 4, but link 4 has a higher chance of movement. that means the earlier selected link is not necessary to have a higher chance of movement. if link 3 is totally occupied at step t, then the last selection link 4 has the highest chance of movement. we can as a result conclude that the most significant factor restricting movement of a vehicle is whether its destination link has free space, or whether its destination link has been selected in advance. when a vehicle moves from its origin segment to a new one, it is assumed that all vehicles on the new segment have already been processed. this assumption may not always be satisfied. under this situation, a vehicle will temporarily be stored on a virtual link. cetin32 utilized the conception of virtual link dealing with node pass process. once the vehicle’s destination is processed, it will be “released” from the virtual link to the destination. it should be noticed that sometimes the vehicles on the virtual link cannot move at all owing to the fact that the destination link is full or owing to the constraint of approach capacity (cci). this capacity limits the numbers of cars that can actually pass the intersection, i.e., the node. the capacity value of an intersection can be determined by highway capacity manual (hcm). netvac4 modeled the intersection by introducing a node pass which calculates how many vehicles can be removed from each of the links entering a particular inbound links constrained by approach capacity. in our gis-simqueue model, a scanning process, which is similar to a node pass method used in netvac and tevacs, is adopted to model the complicated flow pattern of intersection. the scanning process can be described as follows: in a time interval dt, the number of vehicles intend to move from link i to link j (nij) is calculated. nij is first constrained by the section capacity clink of link i. if a link has reached its section capacity clink, no more vehicle than linkc dt× is allowed to add to nij. then the totally number of vehicles intending to enter link j are computed as j ij i n n= ∑ . the number of vehicles can be received by link j in an interval are restricted by its approach capacity ca as well as by the way of traffic control. two kinds of traffic control are considered here, namely, signalized intersection and unsignalized intersection (including primary priority in which only primary priority inbound links are considered while secondary priority links are ignored). to further illustrate this process, we further discuss the unsignalized intersection. we compute an equivalent green split of a cycle time for all primary links. for the j-th incoming link, the green split gj is given by: , /j lane j k lane k k n n g n n = ∑ (2) where k is the number of inbound links at current node. k k n∑ represents the total number of vehicles moving fig. 2. example of intersection passing. published by atlantis press copyright: the authors 181 a gis-based micro-simulation queue model for vehicle evacuation through the node to enter their destination links. lanen represents the number of lanes in the j-th incoming link. thus the approach capacity of link j is calculated as: j j jca g c= (3) accordingly, in a time interval of dt, the total number of vehicles entering into link j cannot exceed caj dt× , and the number of vehicles allowed from the i-th link to the j-th link is restricted by mij, ( )jij ij j ca m n n = (4) the above formula is first proposed by sheffi4 and adopted again by han33 in which the number of vehicles allowed to move merely depends on the length of the waiting queue (the number of vehicles intending to move) and the vehicles’ waiting time has not been considered. however, the waiting time plays an important role in affecting the overall flow features and thus should be considered31. under this circumstance, mij can be rewritten as follows: ( )jij ij i kj k k ca m n t n t = × × ×∑ (5) where ti is the total waiting time of all vehicles in the ith link. by doing so, this procedure can process not only the case that several links converge at one link but also the case that one link diverges into several links. the above process merely scans the dynamic capacity of a link and ignores its static capacity. it was noticed that the static capacity may have considerable influence on the node pass, thus we will discuss it in the following section. the static capacity is introduced to determine how many vehicles can be allowed to move due to the limitation of the static storage capacity of the destination link. this procedure is accomplished by another scanning process. as pointing out earlier, a road can accommodate nmax=length × nlane/7.5 vehicles at most at a time. accordingly, the available free space at step t in an inbound link j is restricted by the number of vehicles on it at step t-1 denoted as nvj(t-1). the available free space in j-th link at simulation step t, nsj(t) is expressed as: max,( )= ( 1)j j jns t n nv t− − (6) the number of vehicles allowed from i-th link to j-th link at k simulation step is further restricted by: ( ) ( ) ( )jij ij i kj k k ns t ns t n t n t = × × ×∑ (7) according to eqs.(5) and (7), the number of vehicles are restricted by both mij and nsij. so the allowance number of vehicles from i-th link to j-th link at simulation step k will be finally restricted by: [ , ( )] ( ) ( )j jij ij i kj k k min ca ns t mns t n t n t = × × ×∑ (8) the algorithm of the two scanning processes can be implemented as the flow chart shown in fig. 3. precisely, vehicles can still leave in this time step according to the approach capacity31, i.e., when one of the following items can be fulfilled, the vehicles can leave: (i) passn <int( linkc ), (ii) passn ==int( linkc ) && (rand( )/rand_max > ( linkc -int( linkc ))). here, rand( )/rand_max means a random number in between 0 and 1. published by atlantis press copyright: the authors 182 j. ma et al. it should be noted that the proposed gis-simqueue model seems like but actually is different from the daganzo’s cell transmission model (ctm)34,35. in the gis-simqueue model, the movements of the vehicles are restricted by the capacity of the link as well as the node, which is similar to the ctm. however, vehicles are treated individually but not as a whole in the gissimqueue model. 4. the gis-based evacuation simulation in the aforesaid queue model, vehicles will move ahead according to a pre-defined route, which depends on the choice of destinations according to the following principles36,37: • nearest emergency shelter (evacuees will move to the nearest shelters), • pre-determined shelter (evacuees will move to the pre-allocated shelters), • dynamic traffic network condition (evacuees will move depending on the traffic condition). all the principles are aiming at evacuating the residents as soon as possible and as fast as possible. thus in our simulation model, an allocation function of the netengine is adopted to allocate the nearest emergency center for each vehicle on the link in the network by minimizing the travel cost. such function can also be used to locate the nearest emergency shelter or service station for each street in a city. having determined the destination, we can establish the original-destination (od) matrix, and the choice of route can be formulated by the traffic assignment methods. in our model, we adopted a simulation-based dynamic traffic assignment (dta)29 method to assign the vehicles to the road network. fig. 4 outlines the components of the gis-simqueue model. at each time step of the simulation, vehicles on links are removed from or are added onto the links, thus the cost of a link varies at every step. this is due to the fig. 3. algorithm of the two scanning processes. fig. 4. components of gis-simqueue model. fig. 5. schematic of the urban freeway network. published by atlantis press copyright: the authors 183 a gis-based micro-simulation queue model for vehicle evacuation reason that the cost is related to the link length and number of vehicles on it. updating this very attribute in gis can reflect the dynamical characteristics of vehicle flow, so all vehicles can get the updating information of all network in real time, which is paramount during emergency evacuation. 5. simulation results of queue model for the reason that few data are available concerning traffic condition during large scale evacuation, here in this section we firstly validate the model by comparing the calculated evacuation times with those observed scenarios that are similar to evacuations, e.g., data on congested urban freeway, which is referenced from urbanik38. urbanik compared the collected field data with the simulation results of i-dynev. the studied area is represented schematically in fig. 5. for the sake of the completeness, we briefly introduce the field data. these data were collected on a 2.9-mile section of interstates 35 in travis, north of austin, texas. the site was a fourlane freeway (two lanes in each direction) in rolling terrain with 12ft lanes, a 3ft paved left shoulder, and a 10ft paved right shoulder. the section capacity was 2040, which was equivalent to discharge headway of 3600/2040=1.76 seconds. table 1. characteristics of road network. roadway segment length/ft free speed number of lanes 1 2000 60 2 2 9620 60 2 3 3710 60 2 4 500 45 1 5 500 45 1 traffic volumes were observed at the starting point, end point and each ramp along the transportation network at 5 minutes intervals, and were available for 25 time periods (from 0 to 120). the input rate of vehicle to the network and their exit rate are shown in fig. 6. from this figure we can see that the highest inputting rate happens between 40-60 intervals. it is worthwhile to notice that no congestion-induced capacity reduction has happened even stop-and-go has been observed. the queuing processes are recorded on the main roads and on-ramp roads. the rate of vehicles exiting the transportation network as a function of time is a critical parameter describing the traffic condition during evacuation process. the traffic volumes on the network at initial time are set to zero, and then are added into the network according to its rate of input. the rate of vehicles leaving the network computed by gawron’s queue model is in good agreement with the observed data (fig.6 refers). the two sets of data are then compared with the output of the gis-simqueue. as we can find, the trend of these data goes well with the empirical observed one. the maximum difference between them is about 100 vehicles, which we believe mainly results from the difference of the initial conditions, i.e., we do not have the initial distribution of the vehicles on the road. the comparison shows that queue simulation model can effectively model the dynamic aspects of rush hour traffic with no congestion induced capacity reduction. however, whether such models are suitable for simulating traffic flow in congestion-induce capacity reduction scenarios should be further studied. sensitivity of the simulation time interval, dt, is another issue that should be considered. two simulation intervals, 1 and 100-second, are examined. sheffi4 has pointed out that simulation interval should be governed by the static capacity of link nmax in the network. for example, if the j-th link can merely accommodate nmax vehicles at most at the same time, then the maximum number of vehicles leaving the j-th link in a time step cannot exceed nsites vehicles. similarly, the maximum number of vehicles that can enter into the j-th link cannot exceed its static capacity. also, the simulation 0 50 100 150 200 250 300 350 400 450 500 0 20 40 60 80 100 120 140 time( 5 minutes) t h e n u m b e r o f v e h i c l e s the observed input rate the observed exit rate gawron queue model gis-simqueue fig. 6. a comparison of model results with observed data. published by atlantis press copyright: the authors 184 j. ma et al. interval cannot greater than the free flow travel time t0=l/v0. the maximum simulation interval is determined as follows: max[ , , / ]sites o link j n n dt min l v c c ≤ (9) the maximum simulation interval is about 13-second (500 × 0.3048/12.5). this value is merely a theoretical value. fig. 7 illustrates that the outputs given by the simulation using a time interval of 100-second are agree well with the observed data. 5.1. an example in order to verify the credibility of the output of the gis-simqueue model, the output of the model is further compared with another established model, the tevacs developed by han33. a hypothetical network as fig. 8 once used by han33 is selected for the comparison. the network comprises 56 links and 24 nodes. the number of public emergency shelters is 4 and the total number of vehicles required to be evacuated is 7000. all links in the network are bi-direction, that is to say, vehicles can move in a from-to direction or in a to-from direction. each direction owns 3 lanes. the length of links is 500m long. the section capacity is 1500, the free speed is 50km/h and the jam density is 200pcu/km. since the initial distribution of vehicles is not clearly stated, we then uniformly distribute the vehicles on all the links. the average number of vehicles of a link are 7000/48 (the total number of links taking part in the assignment is 48) and they are evenly added to network within 100-second for simplicity. table 2 illustrates the comparison of the outputs of gawron’s queue model, tevacs and gis-simqueue model. two time intervals are selected for the simulation and the results show that the clearance time computed by gawron’s model is smaller than that of the output of tevacs and gis-simqueue. this is due to the reason that gawron’s queue model has disregarded the constraint of the phase-approaching capacity. whereas, the gis-simqueue includes the phaseapproaching capacity that is necessary to depict the real traffic characteristic. the outputs given by both gissimqueue and tevacs are in good agreement. table 2. the simulation results. length of simulation interval/second gissimqueue/ min gawron’s model/min tevacs/ min 1 29.6 25.7 * 100 31.7 26.6 32.00 6. concluding remarks with the rapid urbanization process in many asian cities, the provision of evacuation planning is now becoming the last line of protecting densely populated areas from disastrous events. the governments as well as researchers are paying more and more attention to evacuation planning. as a consequence, simulation model is considered as the tool that can assist to establish evacuation planning. fig. 7. comparison of the simulated results with different dt and the observed data. fig. 8. hypothetical road network.33 published by atlantis press copyright: the authors 185 a gis-based micro-simulation queue model for vehicle evacuation a gis based micro-simulation queue model for vehicle evacuation is established in this paper. the approach of two scanning processes is proposed to model of the congestion and spill-back features of vehicle evacuation at street intersection. the proposed gis-simqueue model is a microscopic traffic simulation model that models each vehicle. thus different individual behavior, and current traffic simulation techniques as well, can all be easily took into account. what is more, a gis component, netengine, which can provide all kinds of network algorithms needed for evacuation plan, is integrated in our model for route selection and resource allocation. as a consequence, the proposed model can make full use of current gis technology to improve its computational efficiency. this paper is our initial work and only the vehicle evacuation is concerned. further study on the influence of pedestrian on traffic is in progress. acknowledgements the work described in this paper was fully supported by the grants from the research grant council of the hong kong administrative unit, china [project no. cityu118708]. references 1. w. ng and t. s. waller, reliable evacuation planning via demand inflation and supply deflation, transportation research part e, 46(6) (2010): 1086–1094. 2. l. c. richard and m. s. ryan, modeling small area evacuation: can existing transportation infrastructure impede public safety? california department of transportation task order 3021, final report, http://www.ncgia.ucsb.edu/vital/research/pubs/200204evacuation.pdf (2002) 3. a. pel, m. bliemer and s. hoogendoorn, a review on travel behaviour modeling in dynamic traffic simulation models for evacuations, transportation, 39 (2012):97123. 4. y. sheffi, h. mahmassani and w. b. powell, a transportation network evacuation model, transportation research part a, 16(3) (1982): 209-218. 5. a. g. hobeikaand and b. jamei, massvac: a model for calculating evacuation times under natural disasters, emergency planning simulations series, 15(1985):23-28. 6. a. g. hobeikaand and c. kim, comparison of traffic assignments in evacuation modeling, ieee transactions on engineering management, 45(2) (1998): 192-198. 7. s. tufekciand and m. t. kisko, regional evacuation modeling system (rems): a decision support system for emergency area evacuations, computers industrial engineering, 21(1-4) (1991): 89-93. 8. f. n. de silva and r. w. eglese, integrating simulation modeling and gis: spatial decision support systems for evacuation planning, journal of the operational research society, 41(4) (2000): 423-430. 9. f. n. de silva, providing spatial decision support for evacuation planning: a challenge technology, disaster prevention and management, 10(1) (2001):11-20. 10. r. l. church and t. j. cova, mapping evacuation risk on transportation networks using a spatial optimization mode, transportation research part c, 8(1-6)(2000): 321-336. 11. e. stern and z. sinuany-stern, a behavioral-based simulation model for urban evacuation, papers of the regional science association, 66(1989): 87-103. 12. z. sinuany-stern and e. stern, simulating the evacuation of a small city: the effects of traffic factors, socioeconomic planning sciences, 27(1993): 97-108. 13. p. jenkins, j. l. renne and j. kiefer, gender differences in self-reported evacuation experiences: analysis of the city assisted evacuation program during hurricane gustav. transportation research board special report 2(2011):127-133. 14. t. j. cova and j. p. johnson, microsimulation of neighborhood evacuations in the urban-wildland interface, environment and planning a, 34(2002):2211-2229. 15. x. chen and f. b. zhan, agent-based modelling and simulation of urban evacuation: relative effectiveness of simultaneous and staged evacuation strategies, journal of the operational research society, 59(2006):25-33. 16. b. williams, a. tagliaferri, s. meinhold, j. hummer and n. rouphail, simulation and analysis of freeway lane reversal for coastal hurricane evacuation, journal of urban planning and development, 133(1) (2007):61–72. 17. s. w. mitchell and e. radwan , heuristic prioritization of emergency evacuation staging to reduce clearance time, in: proceedings of the 85th annual meeting transportation research board, washington, dc, usa, (2006), pp.219-228. 18. p. murray-tuite, perspectives for network management in response to unplanned disruptions, journal of urban planning and development, 133(1) (2007):9–17. 19. r. balakrishna, y. wen, m. ben-akiva and c. antoniou, simulation-based framework for transportation network management for emergencies, transportation research record, 2041(2008):80–88. 20. h. wang, s. andrews, n. daiheng and j. collura, scenario-based analysis of transportation impacts in case of dam failure flood evacuation in franklin county, massachusetts. in: proceedings of the 89th annual meeting of the transportation research board, washington, dc, usa. (2010) published by atlantis press copyright: the authors 186 j. ma et al. 21. g. klunder, s. terbruggen, j. mak and b. immers, large-scale evacuation of the randstand: evacuation simulations with the dynamic traffic assignment model indy, in: proceedings of the 1st international conference on evacuation modeling and management, the hague, the netherlands (2009). 22. c. brown, w. white, c. van slyke and j. d. benson, development of a strategic hurricane evacuation-dynamic traffic assignment model for the houston, texas, region, transportation research record, 2137(2009): 46–53. 23. h. tu, g. tamminga, h. drolenga, j. de wit and w. van der berg, evacuation plan of the city of almere: simulating the impact of driving behaviors on evacuation clearance time, in: proceedings of the 1st international conference on evacuation modeling and management, s. hoogendoorn , a. pel, m. taylor and h. mahmassani, (eds.), elsevier, amsterdam (2010), pp. 67–75. 24. m. lepofsky, m. abkowitz and p. cheng, transportation hazard analysis in integrated gis environment, journal of transportation engineering, 119(2) (1993): 239-254. 25. m. lindell, j. kang and c. prater, the logistics of household hurricane evacuation, natural hazards, 58(2011):1093-1109. 26. j. li, b. zhang, w. liu and z. tan, research on orems-based large-scale emergency evacuation using vehicles, process safety and environmental protection, 89(2011):300-309. 27. t. j. cova and r. l. church, modeling community evacuation vulnerability using gis, international journal of geographic information science, 11(1997): 763-784. 28. t. j. cova and j. p. johnson, a network flow model for lane-based evacuation routing, transportation research part a, 37(2003):579-604. 29. c. gawron, simulation-based traffic assignment. phd thesis, university of cologne, germany, 1998. 30. k. nagel, multi-modal traffic in transims. in: pedestrian and evacuation dynamics, springer, berlin (2001), pp.161–172. 31. p. m. simon and k. nagel, simple queuing model applied to the city of portland, international journal of modern physics c, 10(5) (1999):941-960. 32. n. cetin, k. nagel, b. raney and a. voellmy, large scale multi-agent transportation simulations, computer physics communications, 147(2002):559–564. 33. a. f. han, tevacs: decision support system for evacuation planning in taiwan, journal of transportation engineering, 116(6) (1990): 821-830. 34. c. f. daganzo, the cell transmission model: a simple dynamic representation of highway traffic, transportation research part b, 28 (4) (1994): 269-287. 35. c. f. daganzo, the cell transmission model part ii: network traffic, transportation research part b, 29 (2) (1995):79-93. 36. f. southworth, regional evacuation modeling: a stateof-the-art review, oak ridge national laboratory ornl-11740, tennessee (1991). 37. h. d. sherali, a location-allocation model and algorithm for evacuation planning under hurricane/flood conditions, transportation research b, 25(6) (1991): 439-452. 38. t. urbanik, m. p. moeller and k. barnes, benchmark study of the i-dynev evacuation time estimate computer code, nureg/cr-4873,pnl-6171, pacific northwest laboratory, richland, washington (1988). http://www.ocipep.gc.ca/research/scie_tech/emerman/dra ft_guid/index_e.asp. published by atlantis press copyright: the authors 187 atlantis press journal style received 1 july 2015 accepted 25 january 2016 eutrophication hazard evaluation using copula-cloud dengfeng liu, dong wang*, yuankun wang key laboratory of surficial geochemistry, ministry of education; department of hydro-sciences, school of earth sciences and engineering, nanjing university nanjing, 210023, china abstract a copula-cloud (c-c) coupled framework is proposed with respect to the uncertainty and multi-criteria mapping in water environmental hazard evaluation, with introducing copulas and cloud model theory. in the c-c framework, cloud models are firstly generated with given risk criteria which quantify the concept of risk from each dimension, before that copulas are estimated with the dependence of observed water quality data. the multi-dimensional copula-cloud model can be finally constructed with connecting all clouds by copulas, which can model the randomness, fuzziness and dependence of each risk criterion in assessment. the c-c technique have been applied on cases of representative lakes or reservoirs in china. results in comparison with relevant proposed methods verify the completeness and effectiveness of the technique. as a beneficial exploration on clouds and copulas, the proposed c-c technique provide an innovative approach in relevant risk assessment. keywords: risk assessment, copula, cloud model, eutrophication hazard; water environment copula-云在富营养化风险评价中的应用 刘登峰，王栋 * ，王远坤 表生地球化学教育部重点实验室，南京大学地球科学与工程学院水科学系，南京，210046 摘要：针对现有水环境风险评价模型中概念不确定性和多指标映射问题，引入云模型和 copula 函数，提出 了一种耦合 copula 和云的风险评价模型(c-c)。在 c-c 方法框架下，首先根据富营养化指标量化体系通过 云发生器生成相应云模型，再由各类指标下实测数据的相依特性构建 copula 多元映射模型，最后由 copula 联接各个云得到多维 copula-云评价模型。该模型能够同时刻画富营养化风险评价中的模糊性、随机性以及 各风险分量间的相依性。将 c-c 模型应用于我国代表性湖库的富营养化风险评价中，并将评价结果与多种 现有方法进行对比。分析表明 c-c 模型考虑因素全面，评价结果准确可靠，是云模型和 copula 多元分析理 论在风险评价领域的有益探索。 关键词：风险评价；copula；云模型；富营养化 1. 引言 在环境问题日益严重的大背景下，以水体富营 养化为代表的水环境风险日渐显现，严重威胁了居 民的生产生活安全。oecd 将富营养化定义为“水 体营养盐增加引起的一系列征兆变化，其中藻类和 大型植物生产力的增加、水质恶化和其他征兆变化 破坏了水的利用” [1] 。对以富营养化为代表的水环 境风险的科学评价是控制水体污染，保证区域水环 境安全的重要前提。 * corresponding author：wangdong@nju.edu.cn 富营养化风险评价，即通过既定风险指标体系 确定水体的富营养化状态属性，是一个涉及定性定 量转化的多变量综合决策过程。近年来的研究从不 同角度入手，提出的评价模型也各有特点。如基于 weber-fechner 定律、突变模型、集对分析与可变模 糊集等方法 [2-5] 。丁昊和王栋考虑了评价过程中模糊随机性二元共存的特点，引入云模型理论用于富营 养化评价 [6] ；刘登峰等进一步总结了富营养化评价 的 3 个重要问题，并以此提出了一种熵-云耦合评价 模型 [7]。已有的混合熵权方法虽有利于使权重分配 更趋合理，但也无法排除其存在更大不确定的可能 性。而权重本质上作为多指标映射的一种线性实现 journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 10-14 published by atlantis press copyright: the authors 10 dengfeng liu et.al. / eutrophication hazard evaluation using copula-cloud 方式，在原理上具有一定局限性，亟待进一步研究 完善。 copula 函数是一类连接多元随机变量的边缘分 布以得到其联合分布的联结函数，由于该函数形式 灵活且构造简单，因此被逐渐广泛应用于水资源多 元分析领域 [8-10] 。copula 函数能够捕捉和刻画多元 变量间的相依结构 [11-12] ，实质上是一种基于变量相 依特性的多元映射方法。张翔等利用 copula 函数构 建了水质水量的多维联合分布 [13] ；陈晶等进一步探 讨了 copula 函数在水环境评价中的合理性和可靠性 [14] 。本文针对现有云评价模型中结构的复杂性以及 定权问题，简化了云评价模型的构造，从各指标间 的相依性入手引入 copula 函数，提出了一种耦合 copula 和云的风险评价模型（c-c）。首先根据风险 指标体系通过云发生器生成若干云模型，再通过各 类指标下实测数据的相依性构造 copula 函数，最后 由 copula 函数联接云得到多维 copula-云模型。将 cc 模型应用于我国代表性湖库的富营养化风险评价 中，并将评价结果与多种现有方法对比，以验证模 型的可靠性。 2. copula-云风险评价模型(c-c) 2.1. 云模型 云模型是李德毅提出的一种定性概念与其定量 表示之间的不确定性转换模型 [15] 。云模型已被证明 具有应用的普适性，并开始应用于风险评价领域 [16] 。此外，云模型在富营养化评价中的应用原理见 文献[6-7]，此处不作赘述。 文献[7]所提熵-云模型采用一个指标下每个等级 对应一个云模型的研建方式，即一个富营养化熵-云 评价模型由 30 组云模型构成，这种模型结构略显复 杂，不仅增加了建模的难度，也增加了评价的工作 量。本文提出了一种简化的云模型研建方式的方 法，即通过半云发生器 [17] 生成一个云模型以表征一 个指标下所有等级的风险程度。在该建模思路下只 需要针对每个评价指标建立 5 组云模型，这样较大 程度上减少了原有方法的工作量，是对原有云评价 方法的改进。 2.1.1 参数确定 云模型用参数组(ex, en, he)表达某一个指标下 的风险程度。各参数由式(1)得到[7-8]： min max max min ( ) / 2 ( ) / 6 ex b b en b b he k en = +  = −  = ⋅ (1) maxb , minb 为描述某一风险等级定量区间的上下确 界；k 取 0.1[7]。 参考文献[18]建议的指标体系确定各指标下的云 模型参数（见表 1）。 表 1 各风险评价指标下云模型参数(ex, en, he)表 风险指标 云模型参数 ex en he chl-a 75 25.00 2.50 tp 250 83.33 8.33 tn 2500 833.33 83.33 cod 14 4.67 0.93 sd 0 5.00 1.00 2.1.2 模型生成 对于本例中的各风险指标变量，采用半云发生 器生成其相应云模型： ① 生成以 en 为期望值、he 为标准差的一个正 态随机数 en’； ② 生成以 ex 为期望值、en’的绝对值为标准 差的一个正态随机数 x。x 称为富营养化风险论域内 的一个云滴； ③ 计算，令 2 2 ( ) 2( ')e ,( ) 1, x ex en x enu u x x en − −  <= =   ≥ (上升云 [20] ) (2) 或 2 2 ( ) 2( ')e ,( ) 1, x ex en x enu u x x en − −  >= =   ≤ (下降云) (3) u 为 x 隶属于富营养化概念的确定度，以量化 富营养化风险程度； ④ 重复(1)-(3)，直至产生 n 个云滴为止。 2.2. 多维 copula-云模型 copula 函数 ( ,..., )c ⋅ ⋅ 是定义在[0,1]d 上的具有 零基面且 d 维递增的分布函数。根据 sklar 定理 [21] ，令 ( ,..., )h ⋅ ⋅ 为具有边缘分布 1 ( ),..., ( )du u⋅ ⋅ 的 d 维联合分布函数，那么存在一个 copula 函数 ( ,..., )c ⋅ ⋅ ，满足： 1( ,..., ) ( ,..., )dh c u u⋅ ⋅ = (4) published by atlantis press copyright: the authors 11 dengfeng liu et.al. / eutrophication hazard evaluation using copula-cloud 若 1 ( ),..., ( )du u⋅ ⋅ 连续，则 ( ,..., )c ⋅ ⋅ 唯一确定。 常 用 的 copula 函 数 包 括 多 元 gaussian copula、多元 t-copula、archimedean 型 copula、 极值 copula 等，关于 copula 理论的详细介绍参见 [22-23]，此处不做赘述。在各类 copula 函数中， 多元 gaussian copula 因其能够描述多元正态变量 间的相依性，且在变量维数大于 3 时效果优良，故 广泛应用于各类 gaussian 过程或近似 gaussian 过 程的多元建模中 [24] 。令 ρ 为对角元素为 1 的 d 维对 称正定矩阵，且 ρφ 是以 ρ 为相关系数矩阵的标准 正态分布，则相应 d 维 gaussian copula 函数为： 1 1 1 1( ,..., ; ) [ ( ),..., ( )]d dc u u u uρ ρ ρρ − −= φ φ φ (5) 本文采用 gaussian copula 构造多维 copula云模型,从而实现多指标到风险等级的映射。富营养 化指标体系下包含 5 评价指标，因此维数 d=5。首 先通过半云发生器生成 5 个指标维度上的云模型 1 5( ),..., ( )u u⋅ ⋅ ，再计算水质数据的相关系数矩阵 ρ 并得到 gaussian copula 函数 c，最终通过 c 联 结 5 个风险指标下的云得到 5 维 copula-云评价模 型 1 5( ,..., )c u u 。 2.3. c-c 模型评价 c-c 建模的基本框架是：首先根据既定富营养 化风险指标体系生成各个指标下云模型，再通过各 实测水质数据的相依性构造 gaussian copula 函 数 ， 最 后 由 copula 函 数 联 接 各 个 云 得 到 多 维 copula-云模型 c-c 模型构建与评价的具体实现步骤如下： 步骤1 确定富营养化指标体系。本文参考文献 [19] 中建议 的 富营 养 化 指标 体 系 ，选 取 叶 绿素 a(chl-a) 、 总 磷 (tp) 、 总 氮 (tn) 、 化 学 需 氧 量 (cod)、透明度(sd)作为5个风险评价指标，并将各 指标按富营养化风险由低到高划分为6个等级（i, ii, iii, iv, v, vi）。 步骤2 生成各风险指标下的云模型。根据云模 型参数计算公式（式（1）），由富营养化等级限值 确定云模型参数 ( , , )ex en he （表1），由半云发 生器（式（2）、（3））得到各指标下的半云模型 ( )chl au − ⋅ 、 ( )tpu ⋅ 、 ( )tnu ⋅ 、 ( )codu ⋅ 、 ( )sdu ⋅ 。 步骤3 构建多维copula-云模型。根据水质数据 的相关系数矩阵得到相应gaussian copula函数 c （式（5）），联结各指标下的云模型得到多维 copula-云模型 ( , , , , )chl a tp tn cod sdc u u u u u− 。 步骤4 富营养化风险评价。将水体水质实测值 1 5( ,..., )x x x= 代入多维copula-云模型得到确定度 1 2 3 4 5[ ( ), ( ), ( ), ( ), ( )]chl a tp tn cod sdce c u x u x u x u x u x−= 。通过比对ce与各等级所对应的标准限值确定最终 富营养化风险等级lce，评价结束。 3. 实例应用 3.1. 评价结果 以我国 12 处代表性湖库为例 [19] ，验证 c-c 富营 养化风险评价模型的适用性。将水质实测值带入各 风险指标维度上的云模型，得到湖库在各风险指标 维度上的确定度分布（见图 3）。 图 3 湖库各风险指标确定度雷达图邛海 （左上至右下依此：洱海，博斯腾湖，于桥水库，磁湖， 巢湖，甘棠湖，蘑菇湖，西湖，玄武湖，墨水湖，东山 湖） 运用 c-c 模型进行评价时，首先将各风险指标 的等级限值代入多维 copula-云模型，得到不同风 险等级下确定度的标准限值；其次，将各水体实测 published by atlantis press copyright: the authors 12 dengfeng liu et.al. / eutrophication hazard evaluation using copula-cloud 值代入多维 copula-云模型得到富营养化确定度 ce，结果详见表 2。 表 2 各湖库富营养化确定度及风险等级 湖库 uchl-a utp utn ucod usd ce lce 邛海 0.0123 0.3546 0.0431 0.0266 0.8373 0.0123 iii 洱海 0.0184 0.0229 0.0207 0.0751 0.8912 0.0184 iii 博斯腾湖 0.0196 0.0561 0.1850 0.1845 0.9583 0.0196 iv 于桥水库 0.0629 0.0270 0.2066 0.1663 0.9734 0.0629 v 磁湖 0.0582 0.1476 0.5150 0.1205 0.9916 0.0582 v 巢湖 0.0538 0.4184 0.9626 0.1174 0.9985 0.0538 v 甘棠湖 1.0000 0.3859 0.9109 0.3205 0.9974 1.0000 v 蘑菇湖 1.0000 1.0000 1.0000 1.0000 0.9952 1.0000 vi 西湖 1.0000 0.3923 0.9489 0.7153 0.9973 1.0000 v 玄武湖 1.0000 1.0000 1.0000 0.5452 0.9981 1.0000 vi 墨水湖 1.0000 1.0000 1.0000 0.9963 0.9996 1.0000 vi 东山湖 1.0000 1.0000 1.0000 1.0000 0.9986 1.0000 vi (注：各等级所对应确定度标准限值为i[0, 0.0026);ii[0.0026, 0.0029); iii[0.0029, 0.0063); iv[0.0063, 0.0162); v[0.0162, 0.6570); vi[0.6570, 1.0000]) 3.2. 结果对比 为进一步验证评价结果的有效性，选取单一指标 法、文献[19]、云模型法[7]等现有评价方法进行对 比，各方法的评价结果见表 3。 表 3 水体富营养化风险评价结果对比 湖库 单一指标 文献[19] 云模型法 c-c 模型 邛海 v iii iv iii 洱海 iv iii iii iii 博斯腾湖 v iii iv iv 于桥水库 v iv iv v 磁湖 v v v v 巢湖 vi v v v 西湖 vi v v v 甘棠湖 vi v v v 蘑菇湖 vi v vi vi 玄武湖 vi vi vi vi 墨水湖 vi vi vi vi 东山湖 vi vi vi vi 由表 3，将 c-c 模型评价结果与几种参照方法进 行对比分析，在所有的 12 个评价实例中，c-c 模型 的评价等级均是各种方法结果的众数。这表明 c-c 模型最能代表各方法的评价结果，即评价结果具有 较强的代表性。此外，评价等级反映出水体富营养 化风险的高低。由表中可得，虽然在某些评价例中 各方法的评价结果有所出入，但整体上能够反映出 各湖库水体的富营养化情况，即由上至下富营养化 风险逐渐加剧，这与实际水质景观和用水情况相一 致 [19] 。 4. 结论 本文针对风险评价中的不确定性（随机性、模 糊性）以及多指标映射问题，引入云模型和 copula 多元分析理论，提出了一种简化的 c-hc 富营养化 评价模型。首先，云模型刻画了富营养化评价中的 模糊性和随机性，半云发生器的引入使得一个指标 下的富营养化概念只需要一个云模型与之对应，简 化了云模型的研建；其次，copula 函数刻画了不同 评价指标间的相依特性，通过 copula 函数连接各指 标下的云模型即可得到多维 copula-云评价模型。实 例证明 c-hc 模型评价结果直观有效，该模型亦可 应用于其他相关风险评价领域。 copula 函数理论与云模型的耦合是一个崭新的 研究方 向， 本文 以 gaussian copula 为例 探讨了 copula-云耦合建模在水体富营养化风险评价中的适 用性。对于 copula 函数类型、云模型映射模式等方 面的研究值得进一步开展。 致谢 资助项目：教育部博士点基金(20120091110026, 20100091120059)、江苏省教育厅青蓝工程。 published by atlantis press copyright: the authors 13 dengfeng liu et.al. / eutrophication hazard evaluation using copula-cloud 参考文献 1. eutrophication of waters: monitoring, assessment and control. organization for economic co-operation and development, 1982. 2. zhang b., liu j. l. evaluation method for lake eutrophication influence and public satisfaction. advances in water science, 2009, 20(5): 695-700. 张宝, 刘静玲. 湖泊富营养化影响与公众满意度评价方 法. 水科学进展, 2009, 20(5): 695-700. 3. li z. y., guo c., wang j. y., et al. generalized expression of potential function in catastrophe model and its application in eutrophication evaluation. advances in water science, 2010, 21(1): 101-106. 李祚泳, 郭淳, 汪嘉杨, 等. 突变模型势函数的一般表示 式及用于富营养化评价. 水科学进展, 2010, 21(1): 101106. 4. li w. j., qiu l., chen x. n., et al. assessment model for river ecology health based on set pair analysis and variable fuzzy set. journal of hydraulic engineering. 2011, 42(7): 775-782. 李文君, 邱林, 陈晓楠, 等. 基于集对分析与可变模糊集 的河流生态健康评价模型. 水利学报, 2011, 42(7): 775782. 5. yin x., li r. z., yang j. w., et al., eutrophication evaluation model for lake and reservoirs based on extended-blind number. acta scientiae circumstantiae. 2014, 34(4): 1045-1053. 尹星, 李如忠, 杨继伟等. 基于延拓盲数的湖库水体富 营养化评价模型. 环境科学学报, 2014, 34(4): 10451053. 6. ding h., wang d. the evaluation method of water eutrophication based on cloud model. acta scientiae circumstantiae. 2013, 33(1): 251-257. 丁昊, 王栋. 基于云模型的水体富营养化程度评价方法. 环境科学学报, 2013, 33(1): 251-257. 7. liu d. f., wang d., ding h., et al. eutrophication assessment by entropy-cloud model. journal of hydraulic engineering, 2014, 45(10): 89-98. 刘登峰, 王栋, 丁昊, 等. 水体富营养化评价的熵-云耦 合模型. 水利学报, 2014, 45(10): 89-98. 8. favre a c, el adlouni s, perreault l, et al. multivariate hydrological frequency analysis using copulas. water resources research, 2004, 40(1). 9. genest c, favre a c, béliveau j, et al. metaelliptical copulas and their use in frequency analysis of multivariate hydrological data. water resources research, 2007, 43(9). 10. salvadori g, de michele c. multivariate multiparameter extreme value models and return periods: a copula approach. water resources research, 2010, 46(10). 11. vandenberghe s, verhoest n e c, de baets b. fitting bivariate copulas to the dependence structure between storm characteristics: a detailed analysis based on 105 year 10 min rainfall. water resources research, 2010, 46(1). 12. hao z, singh v p. modeling multisite streamflow dependence with maximum entropy copula. water resources research, 2013, 49(10): 7139-7143. 13. zhang x., ran q. x., xia j., et al., jointed distribution function of water quality and water quantity based on copula. journal of hydraulic engineering. 2011, 42(4): 483-489. 张翔, 冉啟香, 夏军, 等. 基于 copula 函数的水量水质 联合分布函数. 水利学报, 2011, 42(4): 483-489. 14. chen j., wang w. s., li y. q., copula assessment method and its application for eutrophication of lake water quality assessment. journal of sichuan university (engineering science edition) , 2011, 1. 陈晶, 王文圣, 李跃清. copula 评价法及其在湖泊水质 富营养化评价中的应用. 四川大学学报 (工程科学版), 2011, 1. 15. li d, liu c, gan w. a new cognitive model: cloud model. international journal of intelligent systems, 2009, 24(3): 357-375. 16. liu d, wang d, wu j, et al. a risk assessment method based on rbf artificial neural network-cloud model for urban water hazard. journal of intelligent and fuzzy systems, 2013. 27(5): 2409-2416. 17. li d.y., du y. artificial intelligence with uncertainty. chapman & hall/crc, 2007. 18. shu j. h. the assessment methodology for eutrophication level of lakes in china. environmental pollution & control, 1990, 12(5): 2-7. 舒金华. 我国湖泊富营养化程度评价方法的探讨. 环境 污染与防治, 1990, 12(5): 2-7. 19. liu d. f., wang d., ding h., et al. gamma-cloud for eutrophication evaluation. journal of china hydrology. 2014, 36(1):8-11. 刘登峰, 王栋, 丁昊, 等. 一种基于 gamma-云的富营养 化评价方法及其应用. 水文, 2014, 34(5): 45-50. 20. sklar a. fonctions de repartition à n dimensions et leurs marges. publication de l’institut de statistique de l’université de paris, 1959, 8:229-231. 21. joe, h., dependence modeling with copulas. crc monographs on statistics & applied probability. chapman & hall, london, 2014. 22. nelsen, r. b., an introduction to copulas. springer, 1999. 23. renard b, lang m. use of a gaussian copula for multivariate extreme value analysis: some case studies in hydrology. advances in water resources, 2007, 30(4): 897-912. published by atlantis press copyright: the authors 14 1. 引言0f( 2. copula-云风险评价模型(c-c) 2.1. 云模型 2.2. 多维copula-云模型 2.3. c-c模型评价 3. 实例应用 3.1. 评价结果 3.2. 结果对比 4. 结论 致谢 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /usedeviceindependentcolorforimages /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <feff04180437043f043e043b043704320430043904420435002004420435043704380020043d0430044104420440043e0439043a0438002c00200437043000200434043000200441044a0437043404300432043004420435002000410064006f00620065002000500044004600200434043e043a0443043c0435043d04420438002c0020043c0430043a04410438043c0430043b043d043e0020043f044004380433043e04340435043d04380020043704300020043204380441043e043a043e043a0430044704350441044204320435043d0020043f04350447043004420020043704300020043f044004350434043f0435044704300442043d04300020043f043e04340433043e0442043e0432043a0430002e002000200421044a04370434043004340435043d043804420435002000500044004600200434043e043a0443043c0435043d044204380020043c043e0433043004420020043404300020044104350020043e0442043204300440044f0442002004410020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200441043b0435043404320430044904380020043204350440044104380438002e> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <feff005500740069006c006900730065007a00200063006500730020006f007000740069006f006e00730020006100660069006e00200064006500200063007200e900650072002000640065007300200064006f00630075006d0065006e00740073002000410064006f00620065002000500044004600200070006f0075007200200075006e00650020007100750061006c0069007400e90020006400270069006d007000720065007300730069006f006e00200070007200e9007000720065007300730065002e0020004c0065007300200064006f00630075006d0065006e00740073002000500044004600200063007200e900e90073002000700065007500760065006e0074002000ea0074007200650020006f007500760065007200740073002000640061006e00730020004100630072006f006200610074002c002000610069006e00730069002000710075002700410064006f00620065002000520065006100640065007200200035002e0030002000650074002000760065007200730069006f006e007300200075006c007400e90072006900650075007200650073002e> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <feff004200720075006b00200064006900730073006500200069006e006e007300740069006c006c0069006e00670065006e0065002000740069006c002000e50020006f0070007000720065007400740065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e00740065007200200073006f006d00200065007200200062006500730074002000650067006e0065007400200066006f00720020006600f80072007400720079006b006b0073007500740073006b00720069006600740020006100760020006800f800790020006b00760061006c0069007400650074002e0020005000440046002d0064006f006b0075006d0065006e00740065006e00650020006b0061006e002000e50070006e00650073002000690020004100630072006f00620061007400200065006c006c00650072002000410064006f00620065002000520065006100640065007200200035002e003000200065006c006c00650072002000730065006e006500720065002e> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <feff0054006900650074006f0020006e006100730074006100760065006e0069006100200070006f0075017e0069007400650020006e00610020007600790074007600e100720061006e0069006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b0074006f007200e90020007300610020006e0061006a006c0065007001610069006500200068006f0064006900610020006e00610020006b00760061006c00690074006e00fa00200074006c0061010d00200061002000700072006500700072006500730073002e00200056007900740076006f00720065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f00740076006f00720069016500200076002000700072006f006700720061006d006f006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076016100ed00630068002e> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word jracr_ayyub-kearney2011-11-15 towards the development of regional risk profiles and adaptation measures for sea level rise bilal m. ayyub, phd, pe professor and director, center for technology and systems management, department of civil and environmental engineering, university of maryland college park, md 20742, usa 301-405-1956 (tel), ba@umd.edu michael kearney, phd professor, department of geography, university of maryland college park, md 20742, usa kearneym@geog.umd.edu abstract risks from future sea level rise entail significant uncertainties concerning overall potential impacts, the specific threats faced by particular areas and what benefit or costs are associated with strategies for addressing such risks. the proposed risk quantification and management framework is consistent with quantitative risk analysis practices in order to enable decision making in a multi-hazard framework. quantifying risk enables the examination of adaptation measures requiring basic physical inputs that can underpin viable engineering solutions for sustaining coastal infrastructure. keywords: coastal engineering, coastal infrastructure, climate change, extreme event, risk, sea level rise. 1. background by the end of this century, some estimates suggest at least 100 million people worldwide will be affected by rising sea levels. this number, large as it may be, hinges on the relatively conservative upper end of scenarios for future sea level rise of the fourth assessment of the intergovernmental panel on climate changes (ipcc 2007). among many climate scientists there exists considerable disquiet that this top end estimate could prove too low, as the contribution from polar ice melting still remains highly uncertain. the resulting impacts on global sea levels could be a rise on the order of 19.6 feet. an increase in the global trend is likely, and this increase will be of the order of two to two-anda-half times what occurred in the 20th century, historically a period of the highest rate of sea level rise in the last thousand years (kearney 2008). the 2008 hurricane ike, a category 5 storm, hit the east texas coast in one of the most populated and commercially important metropolitan areas in the gulf of mexico, galveston bay and galveston of the houston metropolitan region. among the many concerns about the hazards the storm posed to people and structures, was the fate of galveston island and the city of galveston. the catastrophe of the 1900 hurricane still resonates to this day in the area. as it turned out, like had happened many times since the seawall had been constructed in 1903, damage to galveston was largely confined to wind and flood damage; the horrendous destruction wrought by huge storm waves in the 1900 hurricane was largely avoided, with only fairly local damage by wave overtopping of the sea wall. it is difficult to imagine that engineers who designed the galveston sea wall in 1903 had any idea that it would still be serving its purpose more than a century later, even though the limited wave overtopping during ike journal of risk analysis and crisis response, vol. 1, no. 1 (july 2011), 75-89 published by atlantis press copyright: the authors 75 ayyub and kearney suggests that the freeboard of the structure is no longer sufficient. the state of coastal science and engineering back then, regarding wave generation and dynamics, the loads that waves could impose on coastal structures, and the relations of shore processes to sea level rise, even they realized sea levels were rising, was primitive. the march of coastal science and engineering since has been impressive. nevertheless, in application there still much left to be done, particularly when tied to maintaining present and future infrastructure in an era where sea levels could rise faster than at any time since the advent of instrumental records; hence, no detailed analogues. even knowing where the future shoreline might be in many cases, in lieu of simple submergence, remains problematical. the challenge of such a sea level rise is indeed formidable, and requires immediate attention in order to examine associated risks and to assess the socioeconomic impacts for the purpose of developing appropriate long-term measures and mitigation strategies. the impacts on other parts of the globe such as southern asia can be total devastation for particular countries. one of the important economic consequences of sea level rise that merits immediate attention is the impact on ports, shipyards, naval installations, and the transportation arteries that support them. as an example, in the chesapeake bay the port of baltimore has experienced in recent years a 28% growth in foreign cargo, amounting to 32 million tons in 2004. the port is directly responsible for 19,000 direct jobs ($2.4 billion in personal wages and salary), $2 billion in business revenue, and generates $278 million in state, county and municipal taxes (state of maryland governor’s office 2006). the total economic impact is well beyond these estimates. comparable figures are available for the port of norfolk and portsmouth in virginia, plus with the nation’s largest naval installation, the added impact on national security and the ability to project national power to areas across the world. 2. sea-level rise and associated risks fairbridge (1950, 1958, 1960, 1961) documented that the ocean levels rose and fell over long time scales producing what has become known as the fairbridge curve of the holocene eustatic fluctuations based on detailed observations off western australia and afterwards from elsewhere in the world. he formulated the hypothesis that sea levels had been rising for the last 16,000 years and that the rise showed regular periodic oscillations of rise and fall over this period with oscillations continued throughout the last 6,000 years to the present time, but with diminishing amplitude. the oscillations include a relatively short periodicity component of relatively rapid rises and falls of up to four meters, although up to three meters is more common, taking place over periods of no more than 10 or 20 years. this short-periodicity component would now have catastrophic consequences for the world. over the next 100 years and possibly within our lifetime such an occurrence is likely. the periodicities are revealed in a rich variety of sources, including: geology; geomorphology; glaciations; sediments; sand dunes; beach rock; the circulation of the ocean; geomagnetic records; and the records of the isotopes of carbon, oxygen, beryllium, chlorine and hydrogen in tree rings, ice cores, biota, rocks, air and water (mackey 2007; finkl 1995 and 2005). changes in the average sea level involve several primary categories of variables that are interdependent with nonlinear associations: (1) steric (volume) due to changes in temperature (thermosteric) and salinity (halosteric) levels of oceans; (2) worldwide carbon inventory; (3) the shape of the basins that contain the oceans; (4) the mass of water in these basins from melting of glaciers; and (5) local variations in land adjacent to the ocean basins. global warming causes the oceans to warm up; this in turn causes thermal expansion of the oceans leading to rising sea level. global warming also causes the poles to warm up leading to the melting of land-based ice sheets, glaciers, and ice caps. for example, most of the eastern and western united states coastlines are observing a steady rise. the gulf coast is observing a more concerning steady sea-level rise rate; whereas some locations in alaska are actually observing a fall in sea level. this fall is due to uplifting of land due to tectonic plates, i.e., the uplifting rate is greater than the sea level rise rate, making it appear as if sea levels were dropping when in fact land is moving more rapidly upwards. in addition to volumetric expansion of oceans and melting of ice sheets, ocean salinity can cause oceans to expand or contract, changing sea level both locally and globally. in published by atlantis press copyright: the authors 76 towards the development of regional risk profiles and adaptation measures for sea level rise simple terms, ocean salinity is primarily caused by the amount of carbon present in water. human beings are redistributing carbon around the globe. oceans absorb this “carbon” and become more saline, increasing their capacity to store heat and therefore expanding further. carbon concentrations are greatest in the north atlantic ocean where industrialized nations are located (ccsp 2009; dean 1987; ipcc 2007). these variables can form a basis for defining scenarios as recommended by the ipcc (2007) with associated probabilities. defining risk as the potential of losses for a system resulting from an uncertain exposure to a hazard or as a result of an uncertain event (ayyub 2003) offers a basis for risk quantification for identified risk events or event scenarios and associated rates, system vulnerabilities and potential consequences. this definition offers a basis to quantify risk as the rate (measured in events per unit time, such as a year) or probability that lives, economic, environmental, and social/cultural losses will occur due to an event including the non-performance of an engineered system or component. the nonperformance of the system or component can be quantified as the probability that specific loads (or demands) exceed respective strengths (or capacities) causing the system or component to fail, and losses are defined as the adverse impacts of that failure if it occurs. risk can be viewed to be a multi-dimensional quantity that includes event-occurrence rate (or probability), event-occurrence consequences, consequence significance, and the population at risk; however, it is commonly measured as a pair of the rate (or probability) of occurrence of an event, and the outcomes or consequences associated with the event’s occurrence that account for system weakness, i.e., vulnerabilities. another common representation of risk is in the form of an exceedance rate (or exceedance probability) function of consequences. in a simplified notional (or cartesian) product, it is commonly expressed as: risk=event rate  vulnerability  consequence (1) this equation not only defines risk but also offers strategies to control or manage risk: by making the system more reliable through vulnerability reduction or by reducing the potential losses resulting from a failure or impacting event rates. the probability of failure part of the equation can be influenced by engineers by strengthening of existing structures or by adding additional protection; however the consequence part is highly dependent upon the actions and decisions made by residents, government and local officials, including land-use changes, protection measures of coastal areas, response and population relocation plans and practices. event rates can be impacted by policies relating to global warming and carbon reduction as examples. in densely populated areas, simply increasing the reliability of a protection system may not reduce risks to acceptable levels and increasing consequences through continued development of flooding-prone areas can offset any risk reductions. 3. primary physical processes and considerations 3.1. shore erosion and retreat there is no more fundamental and crucial information in assessing risk and sustainability of developed coasts from sea level rise than to be able to forecast the position of shoreline. flooding and wave damage from storms, risks to life and limb, and even assurance that structures or development constructed today will meet their designed amortization schedules, all either increase (in the case of the first two) or decrease (in the case of the last) with proximity to the shoreline. for the last thirty years, immense effort has been devoted to deriving estimates of shore erosion or retreat rates. because obtaining rates from in situ shore profiles is time consuming and likely to be flawed by being too short in record and site specific, historical maps and aerial photography have been the principal means of obtaining long term and synoptic erosion and/or retreat rates. this has been both good and bad. good, because the data are probably reliable up to the date compiled (and probably for the immediate future); however, bad because there is no way beyond linear extrapolation to forecast future trends from such information. evidence supports that the late 20th sea level record documents considerable, even increasing, inter-decadal variability. the 1990s and early 2000s are best the example of this, with an acceleration (in the u.s. middle atlantic coast, exceeding >1 cm/yr) that was the largest of the last half century, followed by a deceleration with dramatic intraannual low stands in sea level. if even a reasonable correlation between sea level rise and shore erosion/retreat could be determined by regression – and published by atlantis press copyright: the authors 77 ayyub and kearney then probably only on sandy beaches as discussed below – it would be credulous to believe that such a relationship might be meaningfully extrapolated for predicting future trends decades away. bruun (1962) sometime ago proposed a now a wellpracticed relationship between the amount of sea level erosion and sea level rise. this two dimensional model indicates that the r, the amount of shoreline retreat is a function of the rise in sea level, s, the cross-shore width of the active profile, l, the depth of closure, h, and b, the elevation of the dune crest or cliff, i.e., the landward limit of sediment transport. in simple algebraic form, the model is as follows: bh ls r   (2) the chief difficulty in applying this model is determining the depth of closure, or the depth of the seaward of the active beach profile. a principal recurring theoretical objection to bruun’s rule is that it is predicated on an equilibrated beach profile. bruun original postulate for the existence of equilibrium profile was based on analyses of beach profiles in monterey, california, and denmark (bruun 1954). bruun found that a power function provided the best mathematical description of profile plan as follows: 3/2)( yayh  (3) where y is the distance in meters in the shore normal direction, h is the depth in meters, and a is a profile scaling factor related to sediment size (dean 2002). bruun assumed that the profile was in equilibrium, an assumption later validated by dean (1977). in the decades since dean’s validation, there have been variations offered to improve bruun’s original model for specific conditions, particularly for gravity forcing in the upper profile (komar 1998), but the basic power relationship holds. only in the instance of severe coastal storms tracking close to shore, which can produce high, short period waves of great erosive power, moving sediment so far offshore that it may take decades for the summer long period swell to return it (zhang et al. 2004), is the assumption of an equilibrium profile tacitly inviolate, and thus excluding a necessary theoretical underpinning of the model. to be sure, it also is open to question whether the bruun rule would still apply if global sea level rise was to accelerate toward the top of the envelope of the ipcc’s estimates. with such a rapid rise, far beyond the global sea level trends current when bruun and others made their observations, the likelihood of an equilibrium existing in beach profiles may be moot, if for no more fundamental reason than shoreline retreat would shift largely to coastal submergence rather erosion. in summary, then, usefulness of the bruun rule lies not so much in ability to predict shoreline retreat during a period of perturbation, i.e., during sea level rise, but rather as a scenario-building tool for predicting how in relaxation beaches adopt a new equilibrium profile, during which the shoreline assumes a new landward position. it is limited to sandy beaches – not mud beaches, marsh shorelines, and the like – comprising geometrically simple coasts. bruun never argued for universality of his model. recent work (zhang et al. 2004), refining the multiplier effect of the shoreline retreat vis-à-vis sea level rise of the brunn rule for u.s. atlantic coast barrier beaches, can provide a reasonable forecast of where shorelines might be with a certain rise in sea level. for non-sandy beaches and shorelines, especially along irregular coasts, the modeling of future shoreline position with sea level rise currently lacks any physical foundation. rosen (1980) tried to adapt the bruun rule for chesapeake bay as an example, an estuarine system of over 9,600 km of shoreline, and predictably had very limited success. apart from the fact that are few sandy beaches in the chesapeake, with mud or marsh shorelines predominating, the absence of ocean swell waves means that the efficiency of the comparatively small storm waves is heightened since sediment eroded during winter stays offshore unlike the open coast where long period swell out of the southeast moves much of it back on shore in summer. 3.2. waves: the force of coastal change rising sea levels will not only erode shorelines and cause coastal submergence, exposing infrastructure to inundation, but also bring the power of waves closer to structures not previously within the zone of potential wave attack, but impose greater static and dynamic loads on existing marine facilities. long shore transport published by atlantis press copyright: the authors 78 towards the development of regional risk profiles and adaptation measures for sea level rise rates also are a function of incident waves and their height. the actual influence of sea level rise on wave dynamics as they might affect coasts may be most pronounced in protected shallow coasts like bays and estuaries. if one assumes that open coast shore profiles will equilibrate as sea levels rise, then the near-shore depth parameter should effectively remain unchanged. open coasts will not be immune from the general effects of global change, however; already there is evidence that significant wave heights (hs) from intense coastal storms are increasing in both the north pacific and north atlantic (ruggiero et al. 2010). nevertheless, because the mud or marsh shores of estuarine coasts probably do not maintain a nominal equilibrium profile, at least in the sense as documented by bruun and others, then a deeper shore profile in a shallow bay could lead to potentially larger waves from the same wind field. in chesapeake bay, with an average depth of 4.5 – 6.1 m (15 – 20 ft), a rise in sea level rise by 2100 ad at the upper envelope of the ipcc ar4 predictions (~60 cm) would be a proportionately significant increase in water depth (~13%), with the potential for proportionately much larger waves. because wave power varies by the square of the wave height, the implications of deeper water in shallow protected coasts becomes clear. as an example, consider the equation for maximum drag force (fd) exerted by waves on cylindrical object like a piling: khcf dd 22/1  (4) where the principal parameters are: c, the diameter of the cylinder; cd, the coefficient for the wave drag exerted on it; h, the wave height; and k, a coefficient. this formula is predicated for shallow water situations using solitary wave theory, and does not assume oscillatory conditions of airy wave theory (goda 2007). for vertical structures like sea walls, bulkheads, or breakwaters, accurate wave height predictions not only are necessary for estimating hydrodynamic loads but also designing freeboard limits with acceptably low probabilities of wave exceedance – e.g., 0.001 for a critical facility or buildings with high occupancy rates. moreover, the breaking wave force (fb) exerted against vertical structure is again related to the square of the stillwater depth (d): 22 91.11.1 ddcf pb   (5) where the other parameters are: cp, the dynamic pressure coefficient; and γ, the specific gravity of water. even wave runup, an important consideration in the design of inclined sea walls, is a function of the significant wave height (hs). the relationship is also dependent on wave incidence, the surf-similarity parameter (ξ), slope angle, and material permeability. however, because runup is critical to the determination of sea wall overtopping and the vulnerability of landward structures to flooding and even wave damage, again the influence of sea level on significant wave height cannot be discounted. 3.3. sea level rise and the tidal frame the literature on the effects of global warming and accelerated rates of sea level rise is replete with descriptions of increased flooding risk and loss of life, damage to coastal structures, and overall for potential social disruption. the threat of even more powerful waves is also considerable; however, often missing in such hazard forecasts is any mention of tides, other their capacity to facilitate the damage done by storm surge and waves to communities if storms hit the coast at high tide, especially a perigean spring tide. this is not wholly surprising as tides, for many coasts, except those at the upper end of the mesotidal range or higher, are often secondary to waves in coastal evolution. for example, in a microtidal coast like chesapeake bay where mean tidal range is in the middle and upper part of the estuary is 0.3 m or less, tidal velocities are low. in baltimore harbor as an example, mean tidal velocities average about 0.8 kts (~1.5 km hr-1) according to noaa (1999). such low velocities have very little erosive potential, especially for the estuarine mud that characterize much of the bay as shields function makes clear; but this could change if the tidal frame is increased as a result of sea level rise. definitive studies of tidal amplification with rising sea levels are few, and where they have been investigated it is not clear whether human activities (e.g., port construction changing harbor hydrography) contemporary with sea level rise were more a factor published by atlantis press copyright: the authors 79 ayyub and kearney than changing water levels. a recent study (jay, 2009) along the u.s pacific coast down into mexico showed that tides have been increasing at a rate of 2.2% per century. in astoria, oregon, tides were increasing at the highest rate found for the study area, about 25 cm per century. the study concluded the effects of sea level rise on tidal amplification would result in greater rates of shore erosion. it is likely the impact on shore erosion will be most pronounced for mesotidal (2-4 m mean tidal range) and macrotidal (> 4 m mean tidal range) coasts, where coarser sediment entrainment and transport by the tides already occurs. one specific result of increasing tidal velocities could be greater scour of bridge and pier pilings. most studies of scour around pilings have focused on unidirectional flow conditions typical of rivers (richardson and davis 2001), rather than bidirectional flow that occurs in tidal conditions. some recent studies (e.g., escarameia 1998; vasquez and walsh 2009) however have examined scour under tidal conditions, with respect to flood duration, the effect of reversal on flow direction, etc. a growing literature (stevenson, et al. 1988) has documented that sinusoidal tides seldom characterize coastal areas, and more commonly tidal cycles display strong differences between time and peak velocities with respect to flood and ebb tides (termed time velocity asymmetry) according to postma (1961). this phenomenon where peak velocities typify one part of the tidal cycle certainly can affect the nature of scour on the pilings. in many areas, especially estuarine channels, there is evidence that ebb domination eventually emerges over the cycle of channel development (stevenson, et al 1988). how sea level rise will affect such channels, particularly as modified by shore protection features, has yet to receive close examination. ultimately, the major limitation in forecasting whether rising sea levels will amplify the tidal frame and increase tidal velocities and, in addition, whether existing or future shore protection features could contribute to this trend, is a lack of specific information of the relations of sea level rise to tidal dynamics for many developed coasts. this information includes temporal trends in sea level rise vis-à-vis changes in tidal amplitude, and adequate baseline data on port hydrography, beyond current velocities, where dredging and new construction may be enlarging or contracting the area flooded. the approaches to the port of baltimore, for example, are dredged often enough, the hydrographic characteristics could vary substantially from one year to the next. such changes also affect wave characteristics. 4. developing a regional risk profile 4.1. a risk methodology probabilistic risk analysis as described by ayyub (2003), kumamoto and henley (1996), and modarres, et al. (1999) can be used to develop the overall risk analysis methodology suitable for quantifying and managing risks associated with sea-level rise. risk assessment is a systematic process for quantifying and describing the nature, likelihood and magnitude of risk associated with some substance, situation, action or event, including consideration of relevant uncertainties (ayyub 2003). its objective is to provide, to the maximum extent practical, a scientific basis for answering the following questions (adapted after kaplan and garrick 1981):  what could happen?  how can it happen?  how likely is it to happen?  what are the consequences if it happens and associated uncertainties?  what can be done to reduce the risks in a cost effective manner?  what effects would these risk management decisions have on subsequent risks and options? in an all-hazard context, risk analysis answers these questions by defining an exhaustive set of hazard or threat scenarios, assessing the likelihoods, vulnerabilities, and consequences reflecting existing threat or hazard reduction countermeasures, vulnerability reduction actions, and consequence mitigation actions. the combination of these three fundamental elements (hazard or threat, vulnerability, and consequence) gives the familiar expression for risk, r, as provided in eq. 1. the process of risk management entails identifying actions, including countermeasures, planning options, land-use changes, consequence mitigation strategies, etc. aimed at reducing or minimizing these risks in an published by atlantis press copyright: the authors 80 towards the development of regional risk profiles and adaptation measures for sea level rise efficient and cost-effective manner with limited impact on future options. the selection of risk reduction alternatives depends on two factors – their cost to implement and relative cost-effectiveness. a common measure of cost-effectiveness for a given investment alternative is its benefit-to-cost ratio. in general, the computation of defensible benefit-to-cost ratios requires consideration of all aspects of risk, including consequence (economic loss, public health and safety, etc.), vulnerability (security and physical), and threat likelihood within a unified probabilistic framework. the rationale behind this assertion is that a probabilistic paradigm permits rational and coherent comparisons among decision alternatives that affect multiple assets to determine the most cost-effective risk reduction strategies. furthermore, knowledge of the quantitative risks under various investment alternatives facilitates a rational comparison with other societal risks (such as fire, earthquake, disease, flood and other natural hazards) to assist in establishing acceptable risk levels and achieve all-hazard risk reduction objectives (ayyub et al. 2007; mcgill, et al. 2007). ayyub et al. (2007) developed an approach called the critical asset and portfolio risk analysis (capra) methodology. in general, capra is a fivephase process. capra consists of several steps as shown in figure 1 and discussed below:  scenario identification: this step characterizes the functions (or missions) applicable to an asset, portfolio, and region and identifies hazard and threat scenarios that could cause significant regional losses should they occur. for natural hazards, this phase considers the estimated annual rate of occurrence, and screens out infrequent scenarios. the outcome of this phase is a complete set of hazard and threat scenarios that are relevant to the region under study.  hazard likelihood assessment: this step produces estimates of the annual rate of occurrence for each threat or hazard scenario including the time-variant hazard profile associated with sea-level rise for a region. for natural hazards, the results from this phase yield an annual rate or probability of occurrence for a hazard affecting the asset or a region and the intensity of the hazard as a function of time.  vulnerability assessment: this step estimates the effectiveness of measures to protect, reduce hazard intensity, detect, delay, respond to, and eliminate a hazard that might cause harm to a region. this phase provides estimates of the probability of success for each hazard scenario, and, if combined with estimated losses, yields an estimate of conditional risk.   fig. 1. the critical asset and portfolio risk analysis (capra) methodology. published by atlantis press copyright: the authors 81 ayyub and kearney  consequence and criticality assessment: this step estimates the loss potential for each scenario identified for the region by considering the maximum credible loss, fragility of the target elements, effectiveness of mitigation strategies, and effectiveness of consequence-mitigation measures to respond to and recover. the results of this phase provide estimates of potential loss for each hazard and threat scenario, which are used to screen scenarios and determine those that warrant further analysis.  benefit-cost analysis: this step assesses the costeffectiveness of proposed countermeasures and consequence mitigation strategies produced from the developing of strategy tables. the results from this phase provide benefit-to-cost ratios for each proposed risk reduction alternative, which are used to inform resource allocation decisions. risk associated with sea-level rise is quantified using a regional sea-level rise (s), probability distribution fs at time t, scenarios of underlying variables (i) defining s, and respective probabilities pi, regional storm rate () that is dependent on s and i, scenarios of underlying variables (j) defining j, and respective probabilities qj, and the conditional probability p(c>c) with which a consequence valuation (c) exceeds different levels (c) for i, j and coastal state at time t. a loss-exceedance probability at time t can be expressed as follows (ayyub, et al 2011):                  i s j jijsi dsjiccpqfptccp ),|();(  (6) where fs is the probability density function of sea level (s) at time t; pi is the probability of a scenario of underlying variables (i) defining s; is the regional storm rate that is dependent on s and i; qj is the probability of a scenario of underlying variables (j) defining ; and p(c>c|i, j) is the probability that the consequence c exceeds c under a state defined by the pair (i, j) and the corresponding state of the coast at time t. summations are over all scenario types i and j using a suitable discretization. the increased storm activities would include increased water depth, tidal effects, shoreline changes, if any, and wave run-up. this model is consistent with recently developed and used risk model for natural hazards, such as the risk models for developing protection strategies of hurricane-prone regions (ayyub, et al. 2009a and 2009b, usace 2006). figure 2 defines a logic and computational flow diagram for the proposed risk methodology for sea-level rise at a particular region starting with hazard identification and definition, followed by inventory definition to estimate losses based on inundation mapping, and finally constructing risk profiles and estimating associated uncertainty. figure 3 provides the corresponding probability and risk tree based on the discretization of the underlying variables and system states according to eq. 6. one of the objectives of the risk analysis is to quantitatively assess the uncertainties associated with fig. 2. a risk methodology for seal-level rise at a particular region. published by atlantis press copyright: the authors 82 towards the development of regional risk profiles and adaptation measures for sea level rise resulting risk profiles. a generalized treatment of uncertainty is available as provided by ayyub and klir (2006); however the methodology proposed in this paper utilizes a simplified treatment that is familiar to practitioners in which two fundamentally different sources of uncertainty affecting an estimated risk profile are considered. the first is attributed to the inherent randomness of events in nature. these events are predicted in terms of their likelihood of occurring (e.g., the chance of storm occurrence). this source of uncertainty is known as aleatory uncertainty and is, in principle, irreducible within present and foreseeable state of knowledge. the second source of uncertainty is attributed to our lack of knowledge or data. for example, the ability to determine the likelihood of an event (i.e., its rate of occurrence) requires that certain data be available. depending on the volume of data that is available, the accuracy of the estimate of the rate of occurrence will vary. if limited data are available, the estimated rate may be quite uncertain (i.e., with a wide interval for a prescribed confidence level). a second type of knowledge uncertainty is attributed to our lack of understanding (e.g., knowledge) about the physical processes that must be modeled (e.g., the meteorological processes that generate hurricane events). often scientists and engineers have interpretations of existing data and models of physical processes of interest that competing in the sense they lead to different results, while at the same time are consistent with observations. in these instances expert evaluations are often required to assess the current state of knowledge and to quantitatively evaluate the level of uncertainty. these sources of uncertainty are referred to as epistemic (knowledge-based) uncertainty. the distinction between what is aleatory and what is epistemic uncertainty can often seem arbitrary. for example, the distinction depends on the models that are used in a particular analysis. in addition, their estimates can change in time. nonetheless, making a distinction between the sources of uncertainty in a logical manner helps ensure that all uncertainties are quantified and those that can be reduced with additional data or knowledge are identified. in principle, epistemic uncertainties are reducible with the collection of additional data or the use/development of improved models. however, in a given project, it is typically not possible to reduce these uncertainties. it should be noted that epistemic uncertainties in each part of the analysis lead to uncertainty in the final risk results. propagating the uncertainties of the individual parts of the analysis through to the final result produces a probability distribution on the risk profile as provided in figure 2. 4.2. hazard analysis this section provides a preliminary demonstration of the proposed methodology using publically available information on the city of baltimore (ayyub et al. 2011). the first step is to estimate the sea level rise as a function of time. data obtained from the national oceanic and atmospheric administration (noaa) website was used as lower bound on the estimates. the data includes current sea level-rise trends for the city of baltimore recorded for over 100 years at baltimore (noaa station # 8574680) and its record goes back to the year 1902. figure 4 shows current sea level trends for baltimore along with a trend line. this trend line is used to estimate future sea level elevations using a linear trend for the purpose of demonstration, and it is specific for the station location. the resulting linear trend is sea level = 0.0031y 5.8699 (7) where y is the year, such as 1992. using this model, the predictions of table 1 can be obtained for 200 years. it should be noted that the results displayed in this table, are solely based on current sea level rise trends and do not include predictions made by the ipcc (2007) based     fig. 3. a probability and risk tree for seal-level rise at a particular region. published by atlantis press copyright: the authors 83 ayyub and kearney on scenario a1b (see figure 5) defining economic, energy and population trends, and not accounting for other effects reported in other studies, such as the melting of any ice masses (vermeera and rahmstorf 2009) where the following model was proposed: dt dt btta dt tdh  )( )( 0 (8) where h = sea level as a function of time t; t = temperature above the baseline temperature t0 at which sea level is in equilibrium with climate. the first term in this equation models the long-term trend, and the second term accounts for the short-term effect since some components of sea level adjust quickly to temperature changes, e.g., the heat content of the oceanic surface mixed layer. the temperature t requires some time to achieve its full effect on the sea-level rise, called the time lag , i.e., t should be the temperature value at  + t. the model parameters a, b,  and t0 can be empirically estimated from data. using a1b ipcc scenario with temperature range above 1980-2000 temperature of 2.3 to 4.3oc, the sea-level rise above the 1990 level in the year 2100 is estimated to have an average of 124 cm and within the range 97 to 156 cm; about 3-orders of magnitude of the ipcc predictions. table 1. sea level prediction for the city of baltimore (for illustration purposes only). year mean sea level (m) sea-level rise, m (ft) 2000 0.3301 0 (0) 2010 0.3611 0.031 (0.102) 2020 0.3921 0.062 (0.203) 2030 0.4231 0.093 (0.305) 2040 0.4541 0.124 (0.407) 2050 0.4851 0.155 (0.509) 2060 0.5161 0.186 (0.610) 2070 0.5471 0.217 (0.712) 2080 0.5781 0.248 (0.814) 2090 0.6091 0.279 (0.915) 2100 0.6401 0.310 (1.017) 2110 0.6711 0.341 (1.112) 2120 0.7021 0.372 (1.220) 2130 0.7331 0.403 (1.322) 2140 0.7641 0.434 (1.424) 2150 0.7951 0.465 (1.526) 2160 0.8261 0.496 (1.627) 2170 0.8571 0.527 (1.729) 2180 0.8881 0.558 (1.831) 2190 0.9191 0.589 (1.932) 2200 0.9501 0.620 (2.034) 4.3. land, asset and resource inventory much information is available in geographical information system (gis) format about the city of baltimore, thus it is used for demonstration purposes. in an effort to most efficiently and effectively analyze this information and how it can be potentially affected by sea-level rise, it is important to first indentify what type of information is needed to sufficiently capture the key assets to define the consequences of greatest concern. the key assets of concern identified in this case study include the following main categories (ayyub et al. 2011): 1. people. three methods were identified to estimate the population affected: one method uses the census data, and the other two use real property data.  2000 census tract data and population counts for each tract – for each census that the united states performs every 10 years population counts are tallied and divided into small areas within each county called tracts. free gis data files are published for public use that include the size of the tract and the number of people. an approach of counting populations affected is by using this population data to compute population density based on census tracks and multiply the population density by the affected areas. other approaches are offered below.  real property “dwelling units” data – for each property listed in baltimore city information on the number of dwelling units is available. totaling all the affected dwelling units within the area of concern then multiplying this by an average number of people per household is another possible population count method.  real property “zoning code” data – each property also has a code associated with it that corresponds to its intended use (such as commercial, residential or industrial are the three main categories). for each residential zoning code, minimum and maximum numbers of units per acre allowed by city of baltimore laws are provided. a third method of counting people affected is to estimate the number of dwellings based on the area of the property and the zoning code, then multiply this number by the estimated number of people per dwelling. the first method is used for the case study of the city of baltimore. the last two methods may be used; however, they involve uncertainty due to reliance on several assumptions. published by atlantis press copyright: the authors 84 towards the development of regional risk profiles and adaptation measures for sea level rise 2. land and environment. inundation of land and its impact on the environment is another primary loss component. by estimating areas of inundation and landuse types, environmental impacts can be assessed to facilitate a proper valuation. 3. property. of greatest importance herein is the property loss (using valuations in dollars) that could result from a sea-level rise. the real property dataset provides a wide variety of information about each property. of greatest importance is the approximate cost of the property in dollars. while this may not perfectly represent the cost of the buildings on the property it has a strong enough correlation to make the assumption that high property values typically indicate that the building(s) located on that property also have higher value and thus high cost of replacement if damaged. in addition, values for the approximate square footage can be obtained that can be used with valuation unit prices to estimate replacement and content values. using zoning classification would offer the means to estimate residential, commercial, industrial, and other property values. 4. roadways and railways. the length of roadways and railways within inundation areas can be estimated with respective daily traffic volumes and movement of goods. the city of baltimore provided many files including all roads (small alleyways to large highways), and railways. these two variables would offer strong bases for valuation. 5. other specific assets. there are many other buildings and other structures throughout the city along the coastal lines, many of which would be of particular importance to know about for consequence estimation, such as:  ports and shipyards  manufacturing plants  water intakes  government structures  stadiums  religious institutions  commercial and retail structures  historical and cultural landmarks  government structures  schools  assisted living, nursing homes  hotels the properties of each of these specific assets include the relevant information that would enable consequence estimation including the approximate number of people that could be affected. 4.4. inundation mapping and risk profile assuming that the city of baltimore does not have a coastal protection system in place, and therefore is vulnerable to sea-level rise, the development of inundation maps requires topographical maps for the city of baltimore. the sea-level rise trends of table 1 were used in combination with the topographical data to define the inundated areas. using an inventory summary of affected land, assets and resources with hypothetical valuations, and loss intensity, i.e., a risk profile, can be produced as provided by ayyub et al. (2011). figure 6 shows the trends of the inventory components affected by inundation. ‐0.3 ‐0.2 ‐0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 m e a n  s e a  l e v e l ( m ) year fig. 4. mean sea level trend for baltimore, md. published by atlantis press copyright: the authors 85 ayyub and kearney 5. developing a regional risk profile the risk management phase assesses the costeffectiveness of proposed countermeasures and consequence mitigation strategies for reducing the risk associated with an asset or portfolio of assets or a region. in the context of sea-level rise, countermeasures aim to reduce vulnerabilities of coastal lines, property and asset exposure, impact on resources and populations, and land use changes. consequence mitigation strategies aim to reduce the potential consequences given the occurrence of a successful scenario. risk management entails decision analysis for a cost-effective reduction of risk given finite available resources. the benefit of a risk mitigation action can be assessed as the difference between the risk before and after implementation (ayyub 2003): benefit = unmitigated risk – mitigated risk (9) the benefit-to-cost ratio can be calculated as: benefit-to-cost ratio (b/c) = benefit/cost (10) where ratios greater than one are desirable. the cost in eq. 10 is the cost to implement and sustain the risk mitigation action. in general, larger benefit-to-cost ratios indicate better risk mitigation actions from a costeffectiveness standpoint. however, selection of the optimal risk mitigation action must also consider the cost to implement relative to available resources as well as whether the strategy achieves risk reduction objectives. the probability that a favorable benefit-to-cost ratio will be realized can be represented as:  0costbenefit11 cost benefit        pp (11) the model in eq. 9 is analogous to the familiar stressstrength model used in reliability engineering (ayyub 2003, modarres, et al. 1999). in general, both benefit and cost in eq. 5 are random variables that can assume any parametric distribution. with knowledge of these distributions, the probability of realizing a favorable benefit-to-cost ratio can be computed using techniques such as the second order reliability model (ayyub 2003). 6. natural or soft engineering solutions to sea level rise the use of natural systems for mitigation of coastal erosion, flooding hazards, and the overall coastal vulnerability to accelerate sea level rise is gaining in popularity. the most often cited natural solutions to moderating the extent and power of storm surges during hurricanes are coastal wetlands, particularly marshes. the general rule of thumb is that surge height will be decreased for each linear 2.5 km of marshes, which unfortunately lacks rigorous testing. it was hoped during hurricane katrina, whose track transverse a considerable distance of coastal marshes along the louisiana coast, that the storm surge would have been appreciably diminished, though there have been no concrete data yet produced that has determined definitively the degree of surge dampening that could be attributed to the marshes. in the event, whatever dampening of surge elevations the marshes caused, the track of the storm flooded new orleans after it moved to the northwest of the city. fig. 5. the ipcc storylines, i.e., scenarios (adapted from ipcc 2007). published by atlantis press copyright: the authors 86 towards the development of regional risk profiles and adaptation measures for sea level rise several problems make assessing the possible degree of coastal flood protection that might afforded by marshes and other coastal wetlands (like swamps) problematical. marsh canopies differ in height and structure, biomass, and the extent of coverage of intact marshes themselves can vary widely, especially in an area of active marsh loss (e.g., louisiana). an additional complication occurs when winds from a storm (i.e., hurricane) where wind fields can affect coasts long before the arrival of the storm surge. resio and westerink (2008) theorize that in these circumstances sustained winds from exceptionally strong hurricanes can blow down grasses and other marsh plants (known as lodging), effectively negating the ability of marshes to dampen surge levels. some evidence suggests that this is indeed what happened during hurricane katrina (resio and westerink 2008). some recent findings link excess anthropogenic nutrient inputs in marshes along the u.s. atlantic and gulf coasts to organic matter decomposition and root degradation (swarzenski et al. 2008). thus, poor and shallow rooting as well as the lack of structural strength of the root mat can greatly increase the likelihood of plant lodging in marshes with even low intensity hurricanes (category 1 or 2). in fact, the general appearance of such marshes can be deceptive with regard to their potential for storm surge and wave dampening. turner et al. (2004) found a strong bias to lush above-ground biomass while the rooting is poorly developed with low shear strength. it should not be construed, however, that natural solutions have no real contribution to make in planning for coastal protection in an era of rapid sea level rise. rather the appropriate conclusion is that wetlands in particular clearly enhance coastal sustainability especially with regard to ecosystem services; nevertheless, their role in coastal protection is likely to be secondary to more traditional structural measures. moreover, with rates of loss of coastal marshes likely to increase dramatically with accelerated sea level rise, marsh survival is its own unique problem. 7. concluding remarks and research needs quantifying risk using a probabilistic framework produces hazard (elevation) and loss-exceedance probability curves based on a spectrum of sea-level rise scenarios according the mean sea level as a function of time and increased storm rates with associated surges, waves and precipitation with uncertainty quantification. the methodology provides a process for evaluating the loss potential for a region covering land-use changes, population affected, and property at risk by considering the topography and asset inventory for the region. the quantification of risk will enable decision makers to consider various alternatives to manage risk through setting appropriate policy relating to land use, land-use changes, infrastructure planning, building requirements and permits, water resource planning, and the enhancement of consequence mitigation measures. this preliminary, conceptual framework for quantifying risks associated with sea-level rise requires refinement and development of computational details. moreover the state of the inventory requires further developing by focusing on the coastal areas. the inventory used in this paper is developed for rail safety studies and is incomplete and/or inaccurate along the coastal lines. the increase in storm activity with wave run-up intensity escalation due to the rising sea level requires further investigation. the impacts of such increased activities at coastal lines would lead to interdependence with land-use and human-activity changes. 0 2000 4000 6000 8000 10000 12000 14000 2000 2050 2100 2150 2200 # p e o p le year 3x ipcc ipcc 0 1000 2000 3000 4000 5000 6000 7000 2000 2050 2100 2150 2200 # p ro p e rt ie s year 3x ipcc ipcc 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 2000 2050 2100 2150 2200 p ro p . v a lu e ( b ill io n $ ) year 3x ipcc ipcc 0 20 40 60 80 100 120 140 2000 2050 2100 2150 2200 r o a d w a y s ( m i) year 3x ipcc ipcc 0 1 2 3 4 5 6 7 8 9 2000 2050 2100 2150 2200 in u n d a ti o n a re a ( s q m i) year 3x ipcc ipcc fig. 6. illustrative inundation inventory components necessary for quantifying risk profiles for the city of baltimore. published by atlantis press copyright: the authors 87 ayyub and kearney models for global sea level rise in the fourth assessment of ipcc indicate that a dramatic rise in the sea level trend is only a few decades away. the threat posed by this rise to the world’s coastal infrastructure is probably without historical parallel; nevertheless, meeting it will require that science and engineering provide the necessary information on future coastal dynamics and risk that will affect structures and the social – economic and life – sustaining services they provide. fulfilling this mandate will not be easy, as the science for future sea level change is still not settled and the requisite data for developing engineering solutions even for specific localities are incomplete or absent. however, none of this will forestall calls for the engineering community to provide a “fix” regardless of the lack of precedent or inadequate understanding. one emerging consensus about future sea level rise is that, whatever its eventual magnitude (ignoring the calamity of an extensive polar melt down), rapid change (acceleration) is only decades away (ipcc 2007). mobilizing the resources to address the vulnerability of existing coastal infrastructure and what may be done to lessen it (retrofitting, replacement or even redevelopment) is thus timely. for the assessments of structural capability of essential infrastructure to even overall risk to port operations from sea level rise, basic coastal science and engineering information on the following would seem a minimal requirement:  future shoreline position  storm wave height from evaluation of freeboard  wave power  changes in tidal hydrography determinations of all these essential factors in considering future sea level rise poses challenges; however it is possible to develop scenarios, if not exactly robust estimates, whereby planning can occur. acknowledgements the authors acknowledge the financial support of the following units of the university of maryland at college park: college of computer, mathematical and physical sciences, a. james clark school of engineering, college of college of behavioral and social sciences, department of geography, department of civil & environmental engineering, and center for technology & systems management, and the council on disaster risk management of the american society of civil engineers. also, they acknowledge the assistance of cy. chang, p. broqueres, k. markham, j. ramirez, and a. ritter. references 1. ayyub, b. m. 2003. risk analysis in engineering and economics, chapman & hall/crc press, fl. 2. ayyub, b. m., and klir, g. j., 2006. uncertainty modeling and analysis in engineering and the sciences, chapman & hall/crc, press boca raton, fl. 3. ayyub, b. m., foster, j., mcgill, w. l., 2009a. risk analysis of a protected hurricane-prone region i: model development, asce natural hazards review, 10:2(38), 38-53. 4. ayyub, b. m., mcgill, w. l., foster, j., jones, h. w., 2009b, risk analysis of a protected hurricane-prone region ii: computations and illustrations, asce natural hazards review, 10:2(38), 54-67. 5. ayyub, b. m., mcgill, w. l., kaminskiy, m., 2007, critical asset and portfolio risk analysis for homeland security: an all-hazards framework, risk analysis journal, 27(3), 789-801. 6. ayyub, b. m., ramirez, j., markham, k., and broqueres, p., 2011 (in press), quantifying regional risk profiles attributable to sea-level rise, asce monograph, reston, va 7. bruun, p. 1954. coast erosion and the development of beach profiles. beach erosion board technical memorandum. no. 44. u.s. army engineer waterways experiment station, vicksburg, ms. 8. bruun, p., 1962. sea level rise as cause of shore erosion. american society of civil engineering proceedings, journal waterways harbors division 88: 117-130. 9. ccsp, 2009. coastal sensitivity to sea-level rise: a focus on the mid-atlantic region. a report by the u.s. climate change science program and the subcommittee on global change research, u.s. environmental protection agency, washington, dc. 10. dean, r. g. 1977. equilibrium beach profiles – u.s. atlantic and gulf coasts. ocean engineering report no. 12, university of delaware, newark, pp. 1-45. 11. dean, r. g., 2002. closure depth considerations along the florida shoreline. ufl/coel – 2002/009, department of civil and coastal engineering, university of florida, gainesville, florida. 12. escarameia, m., 1998. laboratory investigation of scour around large structures in tidal waters. conference on basics of sediment transport and scouring. hr wallingford (http://kfki.baw.de/conferences/iche/1998cottbus/55.pdf). 13. fairbridge, r. w., 1950. the geology and geomorphology of point peron, western australia, j. of the royal society of western australia, 33, 1-43. 14. fairbridge, r. w., 1958. dating the latest movements in the quaternary sea level. new york academy of science transactions, 20 471-482. published by atlantis press copyright: the authors 88 towards the development of regional risk profiles and adaptation measures for sea level rise 15. fairbridge, r. w., 1960. the changing level of the sea, scientific american, 202 (5), 70-79. 16. fairbridge, r. w., 1961. eustatic changes in sea-level, in l. h. ahrens, k. rankama, f. press and s. k. runcorn (eds), physics and chemistry of the earth, vol. 4, london: pergamon press, 99-185. 17. finkl, c w., jr., (ed.), 2005. the sun, earth and moon in honor of rhodes w. fairbridge. j. of coastal research, special issue no. 42. 18. finkl, c. w., jr., (ed.), 1995. holocene cycles: climate, sea levels, and sedimentation. a jubilee volume in celebration of the 80th birthday of rhodes w. fairbridge. j. of coastal research, special issue no. 17. 19. goda, y., and kudaka, m., 2007, “on the role of spectral width and shape parameters in control of individual wave height distribution,” coastal engineering j., 49(3), 311335. 20. ipcc, 2007. climate change 2007: synthesis report. contribution of working groups i, ii and iii to the fourth assessment report of the intergovernmental panel on climate change, geneva, switzerland. 21. jay, d. 2009. evolution of tidal amplitude in the eastern pacific ocean. geophysical research letters 36: l06604l, doi: 10.1029/2008gl/036796. 22. kaplan, s., and garrick, j. b., 1981, “on the quantitative definition of risk,” risk analysis journal, 1(1), 11-27. 23. kearney, m. s. 2008. the potential for significant impacts on chesapeake bay: sea level impacts and ecology. in m. mccracken (ed.), the likelihood and character of large and disruptive climate change. london: earthscan, 85-100. 24. komar, p. 1998. beach processes and sedimentation. second edition. prentice hall, nj. 25. kumamoto, h., and henley, e.j., 1996, probabilistic risk assessment and management for engineers and scientists, second edition, ieee press, new york. 26. mackey, r., 2007. rhodes fairbridge and the idea that the solar system regulates the earth’s climate, j. of coastal research, special issue no. 50. 27. mcgill, w. l., ayyub, b. m., kaminskiy, m., 2007, a quantitative asset-level risk assessment and management framework for critical asset protection, risk analysis international journal, society for risk analysis, 27(5), 2007. 28. modarres, m., kaminskiy, m., and krivstov, v., 1999. reliability engineering and risk analysis: a practical guide, marcel decker inc., new york, ny. 29. noaa. 1999. assessment of the national ocean service’s tidal current program. noaa technical report nos co-op 022. u.s department of commerce, silver spring, md. 30. postma, h. 1961. suspended matter and secchi disk visibility in coastal waters. netherlands journal of sea research 1: 359-390. 31. resio, d.t., and westerink, j.j., 2008. modeling of storm surges. physics today 61: 33-39. 32. richardson, e. and davis, s.. 2001. evaluating scour at bridges: fourth edition. national highway institute, federal highway administration, hydraulic engineering circular no. 18. 33. rosen, p., 1978. a regional test of the bruun rule on shoreline erosion. marine geology 26: m7-m16. 34. ruggiero, p., komar, p. d., and allen, j. c., 2010. increasing wave height and extreme value predictions: the wave climate of the pacific northwest. coastal engineering 57: 539-552. 35. stevenson, j.c., ward, l.g., and kearney, m.s., 1988. sediment transport and trapping in marsh systems: implications of tidal flux studies. marine geology 80: 37-59. 36. swarzenski, c.m., doyle, t.w., fry, b., and hargis, t. g., 2008, “biogeochemical response of organic-rich freshwater marshes in the louisiana delta plain to chronic river water influx. biogeochemistry 90: 49–63. 37. turner, r. e., swenson, e. m., milanj, c. s., lee, m., and oswald, t. a., 2004. belowground biomass in healthy and impaired salt marshes. ecological research 19: 29–35. 38. usace, 2006. interagency performance evaluation task force draft report on “performance evaluation of the new orleans and southeast louisiana hurricane protection system,” draft volume viii – engineering and operational risk and reliability analysis, usace, washington, dc. https://ipet.wes.army.mil 39. vasquez, j. a. and walsh, b. w., 2009. cfd simulation of local scour in complex piers under tidal flow. proceedings of the 33rd iahr congress: water engineering for a sustainable environment, international association of hydraulic engineering & research (iahr), pp. 913-920. 40. vermeera, m., and rahmstorf, s., 2009, global sea level linked to global temperature, www.pnas.org/ cgi/doi/10.1073/pnas. 41. zhang, k., douglas, b.c., and leatherman, s.p., 2004. global warming and long-term sandy beach erosion. climatic change 64, 41-58. published by atlantis press copyright: the authors 89 atlantis press journal style received 17 june 2015 accepted 28 august 2015 dealing with emergencies: the case of a heavy disruption of the mexico city metro system diego padilla-pérez, jaime santos-reyes saracs research group, systems engineering department, sepi-esime zac, ipn, u.p. "adolfo lopez mateos", edif. 5, 2do. piso, mexico city, c.p. 07738, mexico e-mail: jrsantosr@hotmail.com samuel olmos-peña centro universitario uaem valle chalco, hermenegildo galeana no. 3, col. ma. isabel chalco edo. de méxico, c.p. 56616, mexico e-mail: samuelop@gmail.com abstract the paper presents the results of a forecasting model associated with the affluence of users of the metro line-b of mexico city's metro system. it also presents in a way a retrospective analysis of the metro incident that occurred on september, 2011, in the same metro line; the incident affected seven metro stations and about 17 thousand commuters. the approach has been the use of artificial neural networks (ann). the main conclusions may be summarized as follows: (i) the metro incident has illustrated the fact that different modes of urban transport are highly interdependent; (ii) the proposed ann model has the potentiality to be used to forecasting the affluence of users for any metro line for the case of mexico city's metro system; (iii) the above (ii) can be used as input to the decision process in order to implement the required number of coaches to assist the affected commuters; (iv) both (ii) and (iii) should be part of an emergency response plan to mitigate the impact of cascading failures due to interdependencies amongst the different modes of urban transport. keywords: ann, emergency response, metro system, megacity. 1. introduction megacities have become a global phenomena that has dispersed around the planet (a megacity has been defined as an urban area with more than 10 million residents) (kotkin, et al., 2014). in the early 80s there were only three megacities; i.e., tokyo, new york and mexico city. it is believed that there are about 29 such cities as in 2014 and account for roughly 13% of the world’s urban population (kotkin, et al., 2014). however, one of the biggest challenges facing megacities lies in lagging infrastructure; i.e., these cities continue to add population, without the infrastructure that paralleled the growth (kotkin, et al., 2014). an example of a lack of infrastructure is related to transportation (kotkin, et al., 2014). moreover, in the ultra dense environment of developing country megacities, traffic congestion is also worsening and effectively affecting the urban mobility (heinrichs, et al., 2012; kotkin, et al., 2014). for example, nearly half of mumbai (india) commuters spend at least one or two hours to get to work, far more than workers in smaller rivals such as chennai, or hyderabad (india). on the other hand, about 50% of formal sector workers in mumbai expressed the desire to move elsewhere, in part to escape train or car commute; only a third of workers in other cities expressed this sentiment (kotkin, et al., 2014). in the literature review, on the other hand, it has been found that infrastructures are highly journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 142-151 published by atlantis press copyright: the authors 142 interdependent and present a great challenge to society today concerning how they are to be managed so as to produce an acceptable risk is a question which has come to the fore in dramatic ways in recent years (gómez, 2011; gutierrez, 2014; kroger, 2008; zhang, et al., 2012; lee, et al., 2003; alam, 2012; ap, 2014; hayward, 211; santos-reyes, et al., 2015). a vast amount of research has been published on the subject (balakrishnan, et al., 1998; haimes, et al, 1998; rinaldi, 2001; amin, 2001; lee, et al., 2003; cagno, et al., 2009; rosato, et al., 2008; panzieri & setola 2008; baiardi & telemon 2008). overall, it may be argued that metro underground systems are highly efficient and reliable; e.g., the mexico city metro carries more than 1,600 million users per year, equivalent to 5.1 million users on weekdays (stcm, 2014). in addition, its efficient operation contributes to the economic and social benefits, and to the reduction of environmental pollution. on the other hand, it has been found that given the nature of interdependencies amongst these systems, when a failure occurs, usually cause cascading failures affecting, inter alia, other systems; e.g. other modes of transportation. the following two examples illustrate this: (i) on september 11, 2011, there was an electrical fault in the metro line-b of mexico city's metro underground system; it affected seven metro stations and about 17 thousand commuters. it is believed that the users had to leave the underground metro and find alternative modes of transportation to reach their final destinations (gómez, 2011). (ii) on november 14, 2014, nearly 300 thousand commuters were affected by the collapse of the metro system in the capital city of santiago, chile (gutierrez, 2014). a power failure occurred in three of the five metro lines. the accident did not allow the commuters to reach to their destinations and caused a huge traffic chaos on the streets of the capital city. thousands of commuters took their vehicles to work, which led to a collapse in most of the capital's arteries, while pedestrians were unable to use the public transport and most waited between one and two hours without getting a bus; taxis could not cope with the amount of commuters either (ap, 2014; gutierrez, 2014). the paper attempts to answer the following questions: why did the 2011 metro disruption cause such a chaos in the city? were there enough other modes of transport (e.g. coaches) to cope with the disruption? given this, it is clear that a better understanding of the metro transport system under different scenarios may become a necessary step forward towards an efficient and reliable urban transport system. in the literature search, however, there is very little attention given to this. the paper addresses this very issue by analyzing an scenario regarding a disruption of the metro line-b of mexico city's metro underground system. the approach has been the application of an artificial neural network (ann). the paper gives an account of the key findings of the ongoing research project. 2. emergency response system in place in case of a disruption of the metro system 2.1. the mexico city's metro and the affected metro line-b the mexico city metro underground system carried an estimated 1.6 billion passengers in 2012 (stcm, 2014). the metro system is being regarded as the second largest metro system in north america after the new york city metro. fig. 1 shows the metro network and the affected metro line-b on september 13, 2011. the affected stations (i.e., stn-b7 to stn-b13) were the ones shown in red colour in the figure (gómez, 2011). fig. 2 shows the number of trains versus the users demand for this particular metro line-b (i.e. 05 hrs 24 hrs). it can be seen that at peak hours (i.e. 07:00 09:00 hrs and from 18:00 21:00 hrs) the highest number of trains required (about 38) for coping with the demand of users (stcm, 2014). the data from figs 1&2 have been used to the analysis being reported in section 3. 2.2. the emergency response in case of a disruption in the metro system in mexico city, one of the existing emergency plans, is the use of the coach 'passenger transportation network'(known as 'rtp'). that is, these transport systems are implemented, either when the metro underground system undergoes maintenance, or when accidents occurring in the system (rtp, 2015). table 1 published by atlantis press copyright: the authors 143 shows the fleet available in the city for the case of an emergency. fig.3, on the other hand, shows the map where the eight parking stations of the coaches are located (these are shown as 'stars' like symbols and in red colour). table 1. types of coaches available during an emergency (rtp, 2015). type of coach capacity (number of people) fleet type-i 100 934 type-ii 90 210 type-iii 70 146 type-iv 240 51 type-v 160 12 type-vi 53 105 total 1458 3. materials & methods an artificial neural network (ann) is a mathematical model that can be implemented as a software simulation that tries to simulate two essential properties of the human brain in relation with its high capabilities of parallel information processing (safi & bouroumi, 2013). technically speaking, the conception of a neural solution to a practical problem requires three main steps (safi & bouroumi, 2013). the first step is the choice of a suitable architecture for the ann; i.e., the number of neurons to use and a suitable way for connecting them in order to form the whole network. the second step is the choice of a suitable algorithm for training the network, i.e., a method for determining the best possible value for each synaptic weight modeling the physical connection between two neurons. the third step is the fig. 1. the affected metro line-b of the mexico city's metro network. (adapted from stcm, (2014)). fig. 2. number of required trains in relation to the demand of users (stcm, 2014). published by atlantis press copyright: the authors 144 choice or the collection of a good set x of sample examples, i.e., the learning database which will serve as input data for the learning algorithm or training algorithm (safi & bouroumi, 2013). fig. 4 shows, on the other hand, that each signal is weighted by a weight, they add up when the activation function is applied in order to obtain the corresponding output. fig. 4. a neuron (paz, 2008). an ann is a model consisting of a vector p representing inputs, a weight vector w, representing the weight of the connections that amplifies or attenuates the input p, an activation function f and a scalar a, which represents the output, and is given by the following expression (paz, 2008): (1) where, (2) the subscripts of the weight matrix w, represents the terms involved in the connection, the first subscript represents the destination neuron and the second, is the source of the signal fed to the neuron. for example, w1r subscripts indicate that this weight is the connection from the r-th input to the first neuron. fig. 5 shows an elman architecture network; this architecture network is commonly associated with a two-layer network with a feedback from the output of the first layer at the entrance of the first layer. this connection allows the elman network to detect and generate patterns that vary over time. from the above figure, it can be seen r inputs, s 1 neurons in the first layer, s 2 neurons in the second layer; the output of layer-one is the input of layer-two and feeds back to the layer-one, so that layer-two can be seen as a network of layer r=s 1 inputs, s=s 2 neurons fig. 3. location of the eight coach stations in the capital city (rtp, 2015). published by atlantis press copyright: the authors 145 and a weight matrix w 2 with dimensions s 2 x s 1 ; the entry of the second layer is a 1 and the output is a 2 . the activation function f takes the total input sk. the activation is given by, (3) the hyperbolic tangent function ('tansig') in the range -1 to 1 for the hidden layers is given by: (4) the identity function ('purelin') to the output layer is being defined as: y = x (5) therefore, the output of each layer is given by the following: hidden layer, a1(k) = tansig ((iw1,1p + lw1,1ªkk-1) + b1) output layer, a2(k) = purelin ((lw2,1a1(k) + b2) the ann model proposed in this study has been the retro-propagation elman architecture and it shown in fig. 6. overall, the proposed ann model contains an input neuron, three hidden layers and one output neuron. each of the hidden layer contains 18 neurons, as a hyperbolic tangent sigmoid transfer function. the output layer has one neuron and an identity function as transfer function. the number of users in the metro line-b has been considered as the input neuron; the output neuron, on the other hand, has been considered as the total number of trains needed to transport the users (fig. 2). fig. 6. elman architecture retro-propagation network for the present case. fig. 5. elman architecture network (paz, 2008). published by atlantis press copyright: the authors 146 4. results & discussion 4.1. forecasting of the affluence of users in the metro line-b table 2 shows the results of the forecasting of the affluence of commuters in the metro line-b and for august 2012. it can been that the forecasting of the number of users is given per day of the month being considered in the analysis. once the forecasting has been done, the next question considered was the following: how many coaches may be required to transport commuters in case of a failure of the metro line-b? table 2. forecasting of the affluence of commuters in the metro line-b for august 2012. day no. users day no. users 1 923592 17 986990 2 883815 18 892790 3 966931 19 715361 4 726103 20 1011153 5 710545 21 906211 6 1016453 22 741689 7 950034 23 759497 8 920144 24 961916 9 868086 25 726986 10 985053 26 724792 11 842271 27 1018134 12 714392 28 984340 13 1026027 29 1001494 14 1010184 30 1023320 15 974482 31 982659 16 912878 total 27868323 in order to be able to answer the above question, it was necessary to determine the number of users per hour and for each of the metro stations of the line-b (fig. 1). in order to achieve this, two sets of data were used: first, the affluence of users in the metro line from april to june 2012 (stcm, 2014); second, the data associated with the number of trains running every hour; i.e., from 05:00 hrs to 24:00 hrs (fig. 2). (this data has been provided by the metro system organization (stcm, 2014)). table 3 shows the results for each hour and for each station of the metro line under study. from table 3, it can be seen, for example, that between 07:00 to 10:00 hrs, the affluence of users has been estimated as 8,056; i.e., the amount of users per hour at station 1 (fig. 1). regarding the number of coaches needed in case of a failure in the metro line-b; this was obtained straightforward by dividing the data shown in table 4 with those shown in table 1. based on this consideration, the calculated number of coaches needed when a failure occurs and at a rush hour (i.e., 08:00 hrs), are shown in table 4. the calculated number of coaches needed in case of an emergency shown in table 4 considers only three types of coaches; i.e., types i, ii, and iii (table 1). this is because they are the ones that are explicitly used in case of an emergency of the metro system (stcm, 2014). 4.2. number of coaches needed during the 2011 metro disruption the model and the results presented in the previous sections helped to assess the number of coaches needed during the emergency situation on 13 september 2011 (the incident occurred at a peak hour; i.e., 08:00 hrs). an electrical fault affected stations 7 to 13, as shown in fig. 1. as motioned in the introduction section, thousands of commuters were affected by the incident. according to the analysis, the required number and type of coaches (table 1) are the following:  160 coaches type-i;  178 coaches type-ii; and  229 coaches type-iii. finally, the following question has been addressed: how long does it take the coaches to get to the stations7-13? table 5 shows the results of the analysis. (it should be emphasised that the analysis has been conducted by using the information shown in fig. 3). 4.3. discussion 4.3.1. the proposed ann model inspired by biological neural networks, anns are massively parallel computing systems consisting of an extremely large number of simple processors with many interconnections. ann models attempt to use some “organizational” principles believed to be used in the human (jain, et al., 1996). ann have been applied to a variety of problems ranging from pattern classification, clustering/categorization, function approximation, published by atlantis press copyright: the authors 147 table 4. number of coaches required at peak hour (08:00) in each of the stations of the metro line-b (table 1). no. station type i type ii type iii 1 81 90 115 2 12 14 17 3 25 28 36 4 17 19 24 5 9 10 12 6 12 13 17 7 25 28 35 8 19 21 27 9 21 24 30 10 23 25 33 11 16 18 23 12 24 26 34 13 32 36 46 14 16 18 23 15 18 21 26 16 20 22 28 17 17 19 25 18 19 21 27 19 15 16 21 20 15 16 21 21 16 18 23 22 17 19 25 23 23 25 33 24 73 81 104 prediction/forecasting, optimization, control (jain, et al., 1996). in the present case, an ann model has been put forward to forecast the number of users of a particular metro line of the metro underground system of mexico city. the model has been based on elman's architecture retro-propagation network (figs. 5&6). the proposed model contains an input neuron, three hidden layers and one output neuron (fig. 6). each of the hidden layer contains 18 neurons and an hyperbolic tangent sigmoid transfer function has been considered. the output layer, on the other hand, contains one neuron and an identity function as a transfer function. the number of users in the metro line-b has been considered as the input neuron; the output neuron has been considered as the total number of trains needed to transport the users demand (fig. 2). the software matlab (r2009a version) has been used to train the ann model; e.g., 'trainscg', the step conjugate gradient with a value of 1x10 -18 and with 33000 iterations has been considered to train the model. moreover, the users data from 2000 to 2011 has been used to train the proposed model. finally, the forecasting of the affluence of users for the metro line-b has been done for august 2012. table 3. affluence of commuters in the metro line-b from 05:00 24:00 hrs. station no. 05:00 hrs 06:00 hrs 07:00 10:00 hrs 11:00 17:00 hrs 18:00 21:00 hrs 22:00 hrs 23:00 hrs 24:00 hrs 1 424 5300 8056 7632 8056 4028 2756 1696 2 64 801 1218 1153 1218 609 416 257 3 131 1637 2448 2357 2448 1244 852 524 4 88 1101 1672 1585 1672 837 572 352 5 46 575 874 827 874 437 299 184 6 63 781 1187 1124 1187 594 406 250 7 131 1628 2476 2346 2476 1238 847 521 8 100 1251 1901 1801 1901 951 651 401 9 112 1400 2128 2016 2128 1064 728 448 10 120 1510 2294 2174 2294 1148 785 483 11 86 1069 1626 1540 1626 812 557 342 12 125 1560 2372 2247 2372 1186 811 500 13 170 2120 3222 3052 3222 1612 1103 679 14 83 1036 1575 1493 1575 788 539 331 15 97 1216 1849 1751 1849 924 632 389 16 105 1307 1987 1882 1987 1530 680 418 17 92 1150 1749 1657 1749 875 599 368 18 98 1225 1862 1764 1862 931 637 392 19 77 971 1476 1399 1476 738 505 311 20 77 973 1479 1402 1479 740 506 312 21 85 1066 1621 1536 1621 810 555 341 22 92 1140 1732 1642 1732 867 593 365 23 120 1501 2281 2161 2281 1140 781 480 24 382 4779 7264 6881 7264 3585 2485 1529 published by atlantis press copyright: the authors 148 one way to validate the forecast of the affluence of users of the metro line-b, is by applying the 'goodness of fit'. this indicates, for example, the discrepancy between the actual number of users of the metro line and the forecasting with the ann model. table 6 summarizes the results of the validation process of the proposed model. table 6. summary of the validation of the proposed model. mean error (me) -20170.22 mean absolute error (mae) 85262.0828 sum of squared error (sse) 3.8128x10 11 mean square error (mse) 1.23x10 10 standard deviation error (sde) 112736.7418 bias error estimation (u m ) 0.0107837 model variability (u s ) 0.0302528 error remnant (u c ) 0.0509638 u of theil 0.00363271 overall, it can be argued that the proposed model may be considered good enough to forecast the affluence of users such as the present case study. for example, the model variability (u s ) shows the ability to replicate the degree of variability of the forecasting with respect to the number of users of the metro line being considered. the degree of variability found is 0.0302528 and it is in the range between -1 to 0 and 0 to 1, being zero the maximum variability. 4.3.2. the emergency response system one of the biggest challenges that megacities face is the problem associated with transport and traffic congestion (kotkin, et al., 2014; heinrichs, et al., 2012). one the one hand, urban mobility has been the subject of importance in recent years (cec, 2007; heinrichs, et al., 2012). that is, major cities worldwide have implemented policies aiming at enhancing mobility while at the same time reducing, among other things, congestion, accidents and pollution (cec, 2007). on the other hand, there has been very little done on how to prevent, for example, the disruption of urban mobility (santos-reyes, et al., 2014, 2015). traffic congestion may be regarded as a condition on road networks that is characterized by slower speeds, longer travel times, and increased vehicular queuing. congestion in the eu is often located in and around urban areas and costs nearly 100 billion euro, or 1% of the eu's gdp, annually (cec, 2007). there are a number of specific circumstances which cause or aggravate congestion; e.g. traffic accidents. traffic research still cannot fully predict under which conditions a traffic congestion may suddenly occur. it has been found that accidents may cause (and have caused) cascading failures which then spread out and create a sustained traffic jam; for example, the failure of the metro line-b that occurred on september 11, 2011 in mexico city (gomez, 2011). the above example illustrates the effects of cascading failure. studies such as the present case, may help to gain a better understanding of cascading failure. for example, the proposed model may help to better prepared for an emergency in case of a disruption of the table 5. distance and the time from the parking stations to the metro stations 7 and 13 (figs 1&3). parking station no. fleet station-7 distance (km) station-7 time station-13 distance (km) station-13 time 1 170 43.5 1 hr 5 min 2 hrs 50 min 38.2 55 min 1hr 30 min 2 168 28.1 45 min 2hrs 20 min 20.3 35 min 55 min 3 168 21.6 28 min 1 hr 40 min 15 28 min 1hr 4 153 17.4 18 55 min 21.8 40 min 1hr. 5 183 14.6 22 min 1 hr 20 min 11.3 30 min 1hr 6 233 9.8 18 40 min 11.8 18 35 min 7 221 7.6 20 35 min 12.9 30 50 min 8 162 14 22 40min 15.1 35 min 1hr 10 min published by atlantis press copyright: the authors 149 operation of any metro line (tables 2-4). by forecasting the affluence of users at any given time may help to take better decision, for example, on the number of coaches to take the affected users to their destinations (tables 4,5). moreover, it can be argued that by conducting studies such as this, it may help to eliminate or mitigate the impact of the following: a) wasting time of motorists and passengers; b) delays, which may result in late arrival for employment, meetings, and education, resulting in lost business, disciplinary action or other personal losses; c) wasted fuel increasing air pollution and co2 emissions owing to increased idling, acceleration and braking; d). the probability of collisions due to tight spacing and constant stoppingand-going. 4.3.3. considerations and limitations last but not least, it is important to mention the limitations of the application of the proposed ann model. that is, the data used in the analysis has been obtained from the organization running the metro system (stcm, 2014). in other words, it is unknown whether the quality of data has been assured. another limitation of the model is that it does not consider the road traffic. effectively, this can (and should) be considered explicitly by the model; this may contribute to further enhance its applicability. 5. conclusions the paper has presented the results of a forecasting model associated with the affluence of users of the metro line-b of the mexico city's metro. it also has presented in a way a retrospective analysis of the metro incident that occurred on september, 2011, in the same metro line. the approach has been the use of artificial neural networks (ann). the main conclusions may be summarized as follows: (i) the two examples of the disruption of the metro systems mentioned in the introduction section have shown the effects of cascading failure due to the interdependencies amongst the modes of urban transport. moreover, the consequence of this (i.e., traffic congestion) shows that lack of a coherent and effective emergency response plan to cope with cascading failure. (ii) the proposed ann model has the potentiality to be used to forecasting the affluence of users for any metro line for the case of the mexico city's metro system. (iii) the above (i) can be used as input to the decision process in order to implement the required number of coaches to assist the affected commuters. (iv) both (ii) and (iii) should be part of an emergency response plan to mitigate the impact of cascading failure due to interdependencies amongst the different modes of urban transport. more research is needed to explore other scenarios for analysis. moreover, data associated with mexico city's road traffic is badly needed to enhance the applicability of the proposed model. acknowledgements this project was funded under the following grant: sipipn: no. 20151117. references alam, a., crisis transmission: global financial crisis. journal of risk analysis and crisis response, 2(3) (2012) 157-165. amin, m., toward self-healing energy infrastructure systems, ieee computer applications in power, 14 (2001) 20-28. ap. 2014. power failure halts main subway line in chile's capital, causing chaos in evening rush hour. retrieved from: http://www.foxnews.com/world/2014/08/20/powerfailure-halts-main-subway-line-in-chile-capital-causingchaos-in-evening/ baiardi, f., and telmon, c., risk management of an information infrastructure: a framework based upon security dependencies. international journal system of systems engineering, 1(1/2) (2008) 237-256. balakrishnan, a., magnanti t. l., and mirchandani, p., designing hierarchical survivable networks. operations research, 46 (1998) 116-136. cagno, e., de ambroggi, m., grande, o., and trucco, p., risk analysis of underground infrastructures in urban areas: time-dependent interoperability analysis. (reliability, risk and safety: theory and applications, in: bris, guedes soares, martorell, editors. esrel conference, 2009). cec (commission of the european community)., towards a new culture for urban mobility. (cee, sec(2007) 1209, brussels). gómez, l. 2011. falla eléctrica en el metro retrasa a 17 mil pasajeros, la jornada, 15 de septiembre del 2011, p.23. available at: http://www.jornada.unam.mx/2011/ 09/13/capital/ 035n3cap (accessed date: 10 june 2015). published by atlantis press copyright: the authors 150 gutiérrez, e. 2014. colapsa metro de la capital chilena por falla eléctrica. la jornada, 14 november 2014, accessed at: http://www.jornada.unam.mx/ultimas/2014/11/14/colapsametro-de-la-capital-chilena-por-falla-electrica-9762.html (accessed date: 14 nov 2014). haimes, y. y., matalas, n. c., lambert, j. h., jackson, b. a., and fellows, j.f.r., reducing vulnerability of water supply systems to attack, journal of infrastructure systems, 4 (1998) 164-177. hayward, k. 2011. flying through an era of volcanic ash. a discussion paper by the royal aeronautical society (2011) 1-7. heinrichs, d., krellenberg, k., hansjürgens, b., martínez, f., risk habitat megacity. (springer, 2012). jain, a.k., mao, j., mohiuddin, k.m., artificial neural networks: a tutorial. ieee computers, (1996) 31-44. kotkin, j., cox, w., modarres, a., renn, a.m., the problem with megacities. (chapman university press. usa, 2014). kroger, w., critical infrastructures at risk: a need for a new conceptual approach and extended analytical tools. reliability engineering & system safety, 93 (2008) 17811787. lee, e. e., mendonca, d., mitchell, j. e., and wallace, w. a., restoration of services in interdependent infrastructure systems: a network flows approach, (technical report 3803-507, decision sciences and engineering systems, rensselaer polytechnic institute, troy, ny, us, 2003) panzieri, s., and setola, r., failures propagation in critical interdependent infrastructures. international journal modelling, identification and control, 3(1) (2008) 69-78. paz, p., modelos de pronósticos y series de tiempo (ipn, méxico, 2008). rinaldi, s.m., peerenboom, j. p., and kelly, t.k., identifying, understanding, and analyzing critical infrastructure interdependencies, ieee control systems magazine, 21 (2001) 11-25. rosato, v., issacharoff, l., tiriticco, f., meloni, s., de porcellinis, s., and setola, r., modelling interdependent infrastructures using interacting dynamical models. international journal critical infrastructures, 4(1/2) (2008) 63-79. rtp.2015. red de transporte de pasajeros. available at: http://www.rtp.gob.mx/servicios.html (accessed date: 10 june 2015) safi, y., & bouroumi, a., prediction of forest fires using artificial neural networks. applied mathematical sciences, 7 (6) (2013) 271 286. santos-reyes, j. & ávalos-bravo, v., a preliminary analysis of two bus rapid transit accidents of mexico city's transport system. procedia engineering, 84 (2014) 624633. santos-reyes, j., padilla-pérez, d., & beard, a.n., modeling critical infrastructure interdependency: the case of the mexico city metro transport system, human and ecological risk assessment: an international journal, 21(5) (2015) 1428-1444. stcm. 2014. metro de la ciudad de méxico, afluencia por tipo de acceso enero – marzo 2012, 5 de abril 2012, http://www.metro.df.gob.mx/operacion/ afluacceso.html zhang, z., you, j., li, q., du, x. (2012). research framework for emergency rescue of disasters and crises. journal of risk analysis and crisis response, 2(3) (2012) 173-177. published by atlantis press copyright: the authors 151 microsoft word assessment method of emergency preparedness system vulnerability based on the complex network theory1.doc assessment method of emergency preparedness system vulnerability based on the complex network theory tianhan jiang china academy of safety science and technology, anquan plaza, 32 beiyuan road, chaoyang district, beijing100012,china. e-mail:jiangth@chinasafety.ac.cn a novel assessment method of emergency preparedness system vulnerability (epsv) is established to evaluate the emergency system construction. the assessment index system and its complex network were constructed for epsv. a simple weighted sum method was used to score the indices. and then the epsv was calculated based on the complex network theory. finally, this model was applied to assessment epsv of 3 regional governments. the results show that the emergency preparedness complex network can characterize the relationship between the components of an emergency preparedness system. this model can assessment the weak links and structural faults of regional emergency preparedness. the emergency preparedness complex network is rational, and the method is effective and feasible. keywords: emergency management, emergency incident, emergency preparedness, system vulnerability 基于复杂网络的应急准备系统脆弱性评估方法 江田汉 中国安全生产科学研究院，北京 100012，中国 摘要：建立突发事件应急准备系统脆弱性评估方法，为应急体系建设提供评估工具。选择突发事件应急准 备脆弱性评估指标，建立应急准备系统脆弱性复杂网络，采用简单加权求和方法计算指标得分，最后基于 复杂网络计算突发事件应急准备系统脆弱性，并将该评估指标和方法应用于 3 个实例。结果表明该应急准 备复杂网络较好地描述了突发事件应急准备体系各组成部分之间的关联和耦合关系。该方法可查找突发事 件应急准备能力的薄弱和不足之处，评估应急准备系统结构性的缺陷。该应急准备系统脆弱性复杂网络较 合理、评估方法可行。 关键词：应急管理，突发事件，应急准备，系统脆弱性 1. 引言 应急准备过去被作为应急管理的四个阶段—减灾、 准备、响应、恢复的一个组成部分。“9·11”事件 之后，突发事件应急管理由以应急处置为主向应急 准备为核心转变 [1] 。应急准备被定义为“一系列事 先精心设计的关键任务和行动，以建立、保持和改 进国内各类事件的预防、保护、响应、恢复所必须 的操作能力。应急准备是一个连续的过程，包含各 级政府的努力，政府、私人部门和非政府组织间的 协调行动，以识别危险、确定脆弱性和识别所需要 的资源” [1] 。国家应急准备的周期包括计划、组 织、装备和培训、演练、评估和改进等关键环节 [2,3] ，从而把应急准备提升为“预防、保护、响应、 恢复”之外的基础性、全过程的行动。因此，应急 准备是指为有效应对突发事件，提高应急管理能力 而采取的各种措施与行动的总称，包括意识、组 织、机制、预案、队伍、资源、培训演练等各种准 备 [4] 。卡特里娜飓风造成严重后果主要原因之一是 应急准备不足和应急管理系统脆弱性 [5] 。在应急准 journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 195-200 published by atlantis press copyright: the authors 195 administrateur texte tapé à la machine received 26 may 2012 administrateur texte tapé à la machine accepted 16 october 2012 administrateur texte tapé à la machine 备体系中加强对脆弱性管理已成为美国国家政策 [1,6,7] 。 脆弱性是面对灾害时唯一可以真正控制的因素 [8] 。 不同领域对“脆弱性”概念的理解存在差异，且由 于脆弱系统的复杂性，脆弱性评价研究尚未成熟 [9] 。目前，脆弱性概念主要侧重于“结果”或“脆 弱性的表现及其原因” [10] ，有风险-灾害（rh）模 式、压力释放（par）模式、政治经济模式、基于区 域的综合脆弱性模式和恢复力模式等 5 种经典脆弱 性分析模型 [11] 。脆弱性评价方法主要有综合指数 法、图层叠置法、脆弱性函数模型评价法、模糊物 元评价法和危险度分析法等 [10] 。应急体系脆弱性相 关研究也逐渐展开，如基于历史数据、指标体系和 实际调查的灾损率曲线的自然灾害脆弱性评估 [11] ， 在应急管理人力资源 [12] 、应急管理的法规及其相关 制度、个人、家庭、组织和社区等应对主体等在应 急体系中的作用 [13] 、应急管理脆弱性来源及其发生 模型 [8] ，这些研究多涉及应急准备的若干方面 [12,13] 或 侧重于脆弱性评估概念模型 [14,15] 。应急管理脆弱性 的研究应得到重视 [16] ，脆弱性分析应成为应急预案 必不可少的组成部分 [17] 。 笔者曾提出基于风险的应急准备脆弱性评估指标体 系和评估方法，用于查找区域突发事件应急准备工 作中薄弱环节和不足之处 [18] 。但该方法不能识别我 国应急准备体系结构性缺陷和制度化薄弱 [17] 。文 中，考虑突发事件应急准备脆弱性要素之间的关联 度和耦合性，基于复杂网络理论建立全风险的突发 事件应急准备系统脆弱性评估方法。 2. 评估方法 2.1. 评估指标 根据《突发事件应对法》、“一案三制”的应急管 理体系框架以及城市重大事故应急能力评估指标体 系[18-20]，经过多轮次的专家访谈，选择由 11 个一级 指标、45 个二级定性指标构成的突发事件应急准备 脆弱性评估指标体系，见表 1。特别指出的是，本 评估指标体系仅适用于省级地方人民政府，目的是 查找各地区应急准备工作中存在的问题和不足，持 改进各地区突发事件应急体系建设，不断提高各级 人民政府突发事件应对能力。这是针对各类突发事 件应急准备脆弱性的共性问题的综合评估，不适用 于具体某类突发事件的评估。 2.2. 计算方法 2.2.1. 应急准备脆弱性指标评分计算 采用正面描述的方法对每个指标分为优、良、中、 差 4 个级别，分别赋分值 0、1、2 和 3。 （1）二级指标评分计算 假设有 p 个评估人员独立评分，则每个二级指标有 p 个分值。该二级指标评分按如下规则计算， 表 1 突发事件应急准备脆弱性评估指标 一级指标 二级指标 一级指标 二级指标 1.应急预案制订与管理 1.1 应急预案制定小组 1.2 风险分析 1.3 应急能力评估 1.4 应急预案评审 1.5 预案的修订改进 1.6 应急预案体系 7.应急恢复 7.1 调查评估 7.2 恢复重建计划 7.3 总结报告 2.应急法制与制度化建设 2.1 应急法制 2.2 应急制度 2.3 应急体系建设规划 8.宣传、教育和培训 8.1 应急管理培训 8.2 应急管理宣传 8.3 应急知识教育 3.应急组织体系 3.1 领导机构 3.2 办事机构 3.3 工作机构 3.4 专家组 9.应急演练 9.1 演练规划 9.2 演练组织实施 9.3 演练总结评估与改进 4.风险评估体系建设 4.1 重大危险源 4.2 危险区域 4.3 隐患治理 4.4 综合减灾计划 4.5 风险评估 10.应急基础保障 10.1 应急指挥平台 10.2 应急物资 10.3 应急装备 10.4 应急资金 10.5 突发事件信息系统 10.6 应急通信 5.监测预警 5.1 突发事件监测 5.2 突发事件预警 5.3 突发事件信息报告 11.应急救援队伍 11.1 军队武警 11.2 综合性应急救援队伍 11.3 专业性应急救援队伍 11.4 基层综合性应急救援队伍 11.5 志愿者队伍 6.应急指挥协调 6.1 重大突发事件统一指挥 6.2 应急响应的部门协调 6.3 跨区的应急协调 6.4 应急救援人员的安全 published by atlantis press copyright: the authors 196 2 3 4 2 1 2 3 4 n n n s n n n n  ＋2 ＋3 ＋ ＋ ＋ (1) 其中，n1、n2、n3 和 n4 分别为将指标评为优、良、 中、差的专家数，p＝n1＋n2＋n3＋n4。 （2）一级指标评分计算 假设某一级指标分为 k 个二级指标，其得分记为 s21，s22，……，s2k（k 为整数），则该一级指标的 评分 s1 按如下规则计算： 1 2 2 1 k i i i s w s    (2) 式中，w2i 为二级指标权重，若没有确定，则 w2i 默认 为 1/k。 2.2.2. 应急准备系统脆弱性计算 将各级各个评估指标等效为节点，如果两指标之间 存在关联度和耦合关系，则在它们之间建立一条 边，并赋予合适的权重，并建立如下规则： （1）一级指标必与其二级指标有单向边相连； （2）一级指标可与其他二级指标无单向边相连； （3）二级指标与一级指标无单向边相连； （4）二级指标可与其他二级指标有单向边相连； （5）一级指标与其二级指标之间边的权重赋予平均 值； （6）各指标出度的权重赋予平均值，如果出度为 1，则权重为 1； 对评估指标进行因果分析 [21] ，结合专家的先验知 识，建立应急准备系统脆弱性网络如图 1。其中， 红色节点为一级指标，蓝色节点为二级指标。 按式（1）和式（2）计算各节点 i 的初始脆弱性， 记为 is ，记节点 i 指向节点 j 的权重为 ijw ，节点 i 的度为 id ，节点总数为 n ，则节点 i 脆弱性 iv 按 式（3）计算： j j 1, j n i ji i i i s s w v d      (3) 系统脆弱性 iv 按式（4）计算： j 1 j 1, j 1 ( ) n n i ji i i i n i i s s w v d          (4) 式中，符号意义同上。 文中各指标的权重均为默认值。 3. 应用实例 将评估模型应用于 a、b、c 三地，以检验该评估指 标体系和评估方法，其评估过程如下：（1）组织 a 政管委、安监局、消防支队、民政局、交通支队、 公安局、电力公司、地震局、水务局、气象局、卫 生局等 12 个部门推荐的 15 位专家组成评估小组， 结合 a 的实际情况，对 a 应急准备能力进行评估； （2）组织 b 政府办公室、安全生产监督管理局、消 防处、民政局、交通管理处、治安支队等部门的 9 位专家，结合 b 的实际情况，对 b 应急准备能力进 行实际评估；（3）组织 c 应急联动中心、安全生产 监督管理局、公安局、交警支队、地震局、民政 图 1 应急准备系统脆弱性网络 published by atlantis press copyright: the authors 197 局、卫生局、某公司等 11 位专家组成评估小组，结 合 c 的实际情况，对 c 应急准备能力进行客观评 估；上述专家评分均在显著性水平 α=0.05 下通过 了 kappa 一致性检验，具体方法见文献[22]。然后 将应急准备能力评分转换为应急准备脆弱性得分， 并作归一化处理，再分别计算各节点的脆弱性和系 统脆弱性。 3.1. 二级指标脆弱性分析 分析具体的二级指标可以详细地掌握突发事件应急 准备系统脆弱性突出之处。按式（1）计算 a、b、 c 三地应急准备一级指标的脆弱性，结果见图 2。a 地应急指挥平台建设情况，b 地的重大突发事件统 一指挥和军队武警协同情况较好，而 c 地志愿者队 伍建设较差，但应急预案制定、评审与修订改进工 作做得较好，应急管理培训和演练规划做得最好。 3.2. 一级指标脆弱性分析 按式（2）算 a、b、c 三地应急准备一级指标的脆弱 性，结果见图 3。 总体而言，a 地的 11 项一级指标脆弱性最大，c 地 最小，b 地介于两者之间。a、b、c 三地在应急预案 制定与管理、应急法制和制度化建设、应急组织体 系、风险评估体系建设等方面建设得较好，特别是 在应急组织体系、风险评估体系建设等方面。在监 测预警、应急指挥协调、应急恢复、宣传、教育和 培训、应急演练、应急救援队伍等方面的建设较 差。 3.3. 系统脆弱性分析 按式（3）和式（4）分别计算 a、b、c 三地突发事 件应急准备系统脆弱性，结果见图 4。a 地应急准备 系统脆弱性最大，b 地次之，c 地最小。由于 a 地在 监测预警、应急指挥协调、应急恢复、应急演练等 方面最显薄弱与不足，突发事件应急准备脆弱性的 多种影响因素在复杂的关联和耦合作用下，a 地应 急准备系统脆弱性最大，反映出该地突发事件应急 准备系统脆弱性的总体情况。 关于应急准备体系建设，笔者实地调研了 a 地应急 管理办公室、城市管理监督指挥中心、部分专项应 急指挥部和街乡镇基层组织，发现各部门与街乡镇 应急能力建设不平衡，布局不合理，出现能力缺失 或者重复建设等现象，如此必将造成突发事件应急 准备体系结构性的脆弱，北京“6·23”特大暴雨发 生时 a 地应急抢险救援效果较差就是例证。 4. 结论 （1）该方法较好地解决了突发事件应急准备系统脆 弱性评估问题，可为突发事件应急体系建设提供参 考 图 2 应急准备二级指标脆弱性 图 4 应急准备系统脆弱性 图 3 应急准备一级指标脆弱性 published by atlantis press copyright: the authors 198 （2）。该应急准备复杂网络较好地描述了应急准备 不同组成部分、不同层级之间的相互作用相互影响 的关系，能够从宏观和微观两个角度发现突发事件 应急准备能力的薄弱和不足之处。 应该指出的是该方法在指标以及指标间关联度和耦 合性的权重设置方面还待进一步完善。 致谢 本文受国家自然科学基金重大研究计划项目 （90924303、91024031）资助。 中国安全生产科学研究院刘铁民、李湖生、邓云 峰、姜传胜、王建光、王晶晶等同志参与了评估指 标体系与方法的讨论，并提出了建设性的建议。 参考文献 [1] u.s.department of homeland security.the national preparedness guidelines[r].[eb/ol].(2007-09-13)[201 0-04-05].http://www.dhs.gov/xlibrary/assets/national_ preparedness\linebreak_guidelines.pdf. [2] u.s. department of homeland security.the federal preparedness report[r].[eb/ol].(2009-01-26)[2010-04 -05].http://www.fas.org/irp/agency/dhs/fema/prep.pdf. [3] u.s. department of homeland security.national response framework[r].[eb/ol].(2008-01)[2010-04-0 5].http://www.fema.gov/pdf/emergency/nrf/nrf-core. pdf. [4] li husheng,liu tiemin.the research progresses and key scientific problems on emergency preparedness system [j].journal of safety science and technology, 2009,5(6):5-10. 李湖生.刘铁民.突发事件应急准备体系研究进展及 关键科学问题 [j]. 中国安全生产科学技术 ,2009,5 (6):5-10. [5] the white house.the federal response to hurricane katrina: lessons learned[r].[eb/ol].(2006-02-23) [2012-05-22].http://georgewbush-whitehouse.archives. gov/reports/katrina-lessons-learned/. [6] fema.comprehensive preparedness guide(cpg) 101 developing and maintaining emergency operations plans.[eb/ol].(2010-11)[2012-05-22].http://www. fema.gov/pdf/about/divisions/npd/cpg_101_v2.pdf [7] u.s. department of homeland security.the 2009 national infrastructure protection plan[r].[eb/ol]. (2009)[2012-05-22].http://www.dhs.gov/xlibrary/assets /nipp_plan.pdf. [8] mcentire d a.the status of emergency management theory-issues,barriers,and recommendations for improved scholarship[c].(2004-06-08)[2012-05-22]. http://www.training.fema.gov/emiweb/downloads/davi d%20mcentire%20-%20%20status%20of%20 emergency %20management%20theory.pdf. [9] rygel l,o'sullivan d,yarnal b.a method for constructing a social vulnerability index:an application to hurricane storm surges in a developed country[j]. mitigation and adaptation strategies for global change, 2006,11(3):741-764. [10] li he,zhang pingyu,cheng yeqing.concepts and assessment methods of vulnerability[j].progress in geography,2008,27(2):18-25. 李鹤,张平宇,程叶青.脆弱性的概念及其评价方法[j]. 地理科学进展,2008,27(2):18-25. [11] shi yong,xu shiyuan,shi chun,sun ali,zhao qing liang.progress in research on vulnerability of natural disasters[j]. journal of natural disasters,2011(2):131137. 石勇,许世远,石纯,孙阿丽,赵庆良.自然灾害脆弱性研 究进展[j].自然灾害学报,2011,20(2):131-137. [12] o'brien g.uk emergency preparedness:a step in the right direction?[j].journal of international affairs,2006, 59(2):63-86. [13] comfort l k.rethinking security:organizational fragility in extreme events[j].public administration review,2002, 62(s1):98-107. [14] liu tiemin.recognition of disaster causes-study of the vulnerability[j].journal of safety science and technology,2010,6(5):5-10. 刘铁民.事故灾难成因再认识——脆弱性研究[j].中 国安全生产科学技术,2010,6(5):5-10. [15] shi peijun.theory on disaster science and disaster dynamics[j].journal of natural disasters,2002,11(3):19. 史培军.三论灾害研究的理论与实践[j].自然灾害学 报,2002,11(3):1-9. [16] li husheng,jiang chuansheng,liu tiemin.the key scientific problems and research progress of major crisis emergencies response[j].journal of safety science and technology,2008,4(5):13-18. 李湖生,姜传胜,刘铁民.重大危机事件应急关键科学 问题及其研究进展[j].中国安全生产科学技术,2008, 4(5):13-18. [17] liu tiemin.the significance of emergency preparedness highlighted by yushu earthquake once again[j].journal of safety science and technology, 2010,6(2):5-7. 刘铁民.玉树地震灾害再次凸显应急准备重要性[j]. 中国安全生产科学技术,2010,6(2):5-7. [18] jiang tianhan,deng yunfeng,li husheng,liu tiemin,jiang chuansheng,wang jingjing.assessment method of regional emergency preparedness vulnerability[j].journal of northeastern university (natural science),2011,32(s2): 273-278. 江田汉,邓云峰,李湖生,刘铁民,姜传胜,王晶晶.区域 突发事件应急准备脆弱性评估方法[j].东北大学学报 （自然科学版）,2011,32(s2):273-278. [19] deng yunfeng,zheng shuangzhong,liu gongzhi,liu tiemin.study on city emergency capability assessment system[j].journal of safety science and technology, 2005,1(6):33-36. 邓云峰,郑双忠,刘功智,等.城市应急能力评估体系研 究[j].中国安全生产科学技术,2005,1(6):33-36. published by atlantis press copyright: the authors 199 [20] deng yunfeng,zheng shuangzhong,capability assessment for urban emergency incident readiness[j]. journal of safety science and technology,2006,2(2):913. 邓云峰,郑双忠.城市突发公共事件应急能力评估— —以南方某市为例[j].中国安全生产科学技术,2006,2 (2):9-13. [21] zheng shuangzhong,deng yunfeng,jiang tianhan. kappa analysis on assessment of urban emergency capability[j]. china safety science journal,2006,16(2): 69-72. 郑双忠 , 邓云峰 , 江田汉 . 城市应急能力评估体系 kappa 分析[j].中国安全科学学报,2006,16(2):69-72. published by atlantis press copyright: the authors 200 microsoft word systematic literature review on the promoting mechanism of high-level financial opening to innovative development of manufacturing industry journal of risk analysis and crisis response, 2023, 13(1), 80-92 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i1.357 80 review a systematic literature review of the empirical research on the promoting mechanism of high-level financial opening to innovative development of manufacturing industry jia-qi wu 1,2 and mu zhang 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: october 7, 2022; accepted: march 15, 2023; published: march 31, 2023 abstract: to further enrich the research on the promotion mechanism of high-level financial openness on the innovation of the manufacturing industry, this article summarizes the domestic and foreign research on the impact mechanism of financial openness. currently, the domestic and foreign studies are mostly based on the characteristics of different financial industries, including the role of opening-up on enterprise innovation and its influence on the bank industry; the effect of foreign direct equity participation on enterprise innovation and its influence on the securities industry; the impact of the host country on the innovation of insured companies and its influence on the insurance industry. future research directions may include deepening research on the transmission mechanism of high-level financial openness on manufacturing innovation and expanding research methods for the impact mechanism of high-level financial openness on manufacturing innovation development. keywords: high-level financial openness; innovative development of manufacturing industry; promotion mechanism; empirical analysis; literature review 1. introduction the manufacturing industry is an important foundation of the real economy, and a key link in the industrial chain to create value, absorb employment, and drive the development of other industries. in recent years, promoting the transformation and upgrading of the manufacturing industry, the high-quality development has become the focus of china’s economic development. in 2017, the people's bank of china jointly issued the guiding opinions on financial support for the construction of a strong manufacturing country, aiming to focus on strengthening financial support for technological innovation, transformation, and upgrading of the manufacturing industry. in 2019, medium and long-term loans and credit loans for manufacturing industry began to be included in the assessment in macro-prudential assessment. in 2020, the national development bank set up a special loan of 250 billion rmb to support the development of china's manufacturing industry towards a more intelligent, green, globalized industrial chain and other high-end direction [1]. after the 14th five year plan was proposed, all core technology industries have become main support jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 81 targets for banks. many banks have stepped up their credit policies to favor these technology enterprises. the registration-based ipo system has enabled more growing enterprises to obtain opportunities to go public. governments at all levels have issued bond issuance policies to meet the diversified financing needs of technological innovation enterprises. at the same time, they also guided funds to support key enterprises to develop emerging industries. since the reform and opening-up, china began to explore forward for financial opening-up, which can be roughly divided into three stages: the economic start-up period, the economic rise period, and the new period [2]. from 1978 to 2000, the financial liberalization during the economic start-up period. before the reform and opening-up, china implemented a unified system in the financial industry, and in 1978, the reform and opening-up officially kicked off the opening-up of china's finance [3]. during this period, china’s financial opening mainly involved two aspects: the opening of the financial industry and the reform of the foreign exchange system. in 2001-2016, the period of economic rise, the opening-up was also divided into two stages: in the early stage, we mainly relaxed the market entry restrictions for foreign financial institutions, to attract foreign capital to carry out business and accelerate the construction of china's financial market system; the later stage, mainly enabled the successive opening of two industries, banking and insurance. from 2017 to 2022, as china's position in the international financial market has gradually improved, the focus on finance in the new times can be concluded in four areas: opening-up the financial industry, marketizing the rmb exchange rate, internationalizing rmb, and opening the capital account. although the capital account opening has further expanded in this period, the government still followed a gradual, prudent, and controllable path to prevent the outbreak of a systemic financial crisis. the high level of financial openness for the manufacturing industry can improve the existing innovative capital investment, increase external financing channels, and alleviate financing constraints. yu bin [4] proposed that, due to the imperfect capital market system in china, the development of financial institutions serving small and medium-sized enterprises (smes), and the limited and narrow financing channels for enterprises faced many difficulties in transformation and upgrading. it caused the serious imbalance between the financial industry and the real economy. financial liberalization is imminent. based on empirical evidence of financial openness in other countries, scholars have analyzed the impact of liberalization on manufacturing in different financial industries. the financial services provided by the stock and bond markets have a greater role in promoting technological innovation in the manufacturing industry [5]. the improvement of the institutional level of the banking industry can increase the total amount of credit, providing more funds for enterprises to alleviate financing constraints, and promoting technological innovation activities of enterprises to a certain extent [6]. some scholars have also turned their attention to possible problems in the process of high-level opening-up [7-9]. financial openness also plays a positive role in promoting the reform of the supply-side structure of finance itself [10]. this paper will review domestic and foreign research on the impact of high-level financial openness on the development of the manufacturing industry innovation, and analyze the impact on the manufacturing industry from the perspective of high-level opening conditions in the banking, securities, and insurance industries. it is expected to further promote relevant empirical research. the remaining structure of this article is arranged as follows: part 2 introduces the empirical study on the promotion mechanism of high-level opening-up in the banking industry on the innovative jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 82 development of the manufacturing industry; part 3 introduces the research on the promotion mechanism of high-level openness of the securities industry to the innovative development of the manufacturing industry; part 4 introduces the empirical research on the promotion mechanism of high-level openness of the insurance industry to the innovative development of the manufacturing industry; part 5 is a brief review. 2. the empirical study on the promoting mechanism of high-level opening of banking industry on innovative development of manufacturing industry at present, domestic and foreign related research mainly focuses on the role of opening-up of the banking industry to firms and its mechanism examination. the direct effect of foreign bank entry, through increased competition among local banks, as well as the spillover effect it causes on enterprises, is concentrated on optimizing the allocation of credit funds and strengthening the role of debt governance. the former is manifested in easing the financing constraints of innovative enterprises [11]. while the latter is manifested in hardening the budget constraints of state-owned enterprises（soes）[12], encouraging them to engage more in technological innovation activities, and improve production efficiency [13, 14]. the empirical results of bai jun [15] show that the promotion effect of foreign bank entry on local enterprise innovation is realized through direct effects and spillover effects (increasing competition among local banks). the entry of foreign banks will enhance the efficiency of the host country's banking industry by intensifying market competition [16], and provide a better external financing environment for the innovation of host country enterprises [11]; foreign banks' entry enhance the ability of host country banks to evaluate enterprise innovation projects through technology spillovers [17], which can provide better financial services for enterprise innovation activities [18]. the empirical analysis by li junqing and xie fang [19] shows that the entry of foreign banks can significantly increase the r&d investment of host country enterprises. acharya and xu [20], blanco and wehrheim [21], zhu zhujun, et al. [22] analyzed the innovation effect of foreign investment opening in the upstream banking sector on downstream firms from the financing constraint channel. meanwhile, some scholars also take a trade perspective, along the classic competition and innovation research paradigm in industrial organization theory [23], analyzed the competitive effect, self-selection effect and market scale effect of foreign investment opening in upstream banking sector on firm innovation [24, 25]. in addition, the empirical results of zhu zhujun et al. [26] show that the opening-up of foreign capital in the banking industry has a significant positive impact on enterprise innovation, a significant positive impact on invention patents and utility model patents, and no significant impact on design patents. the opening-up of foreign capital in the banking industry has significantly improved indicators such as the number of patents cited, patent versatility, and originality. mechanistically, foreign investment opening in the banking sector has a significant positive effect on enterprise innovation through a positive cost-saving effect. the total effect has an inverted u-shaped relationship with the degree of foreign bank entry and remains in the positive effect range during the sample period. the mechanism test conducted by sheng bin and wang hao [27] shows that innovation promotion channels are an important mechanism for foreign banks to enter and affect the quality of enterprises' export products. the mechanism test conducted by yu xulan and ma hanjiang [28] found that foreign banks facilitate the absorption and transformation of international knowledge of chinese enterprises in the process of foreign trade and fdi introduction by providing convenience for cross-border information jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 83 communication, optimizing the credit term structure of enterprises, and improving the service level of the local banking industry, which contributes to the improvement of their independent innovation capabilities. 3. the empirical study on the promoting mechanism of high-level opening of securities industry on innovative development of manufacturing industry at present, domestic and foreign research mainly focuses on the role of capital market opening, foreign direct equity participation in enterprise innovation, and its mechanism testing. 3.1. capital market opening opening capital market can diversify investment risks, improve investment efficiency, and promote the expansion of corporate r&d investment; it can also enhance the level of corporate r&d investment by improving corporate governance and the effectiveness of external supervision [29]. capital market opening can remove the flow barriers in domestic and foreign markets, improve capital allocation and corporate financing constraints, and beneficially promote corporate innovation [30]. from the perspective of capital inflows, xie fang and guo na [31] have demonstrated that equity capital inflows significantly increase the r&d investment of host country enterprises, while debt capital inflows have a negative impact. the empirical evidence of moshirian et al. [32] suggests that relaxing financial restrictions, strengthening risk sharing between domestic and foreign investors, and improving corporate governance are channels to allow stock market liberalization to promote technological innovation. xie fang [33] empirically found that the liberalization of the stock market promotes enterprise innovation by easing financing constraints, enhancing risk sharing, and improving corporate governance. from the qualified foreign institutional investors (qfii) perspective, luong et al. [34] found that overseas institutional investors have a positive impact on enterprise innovation. bena et al. [35] empirically found that foreign institutional investors have significantly improved the company's innovation output by guiding company value investment through discipline and supervision measures. research by jiang wenjun [36] shows that qfii shareholding has a strong technology spillover effect, and institutional investors from countries with high innovation levels are more able to help enterprises enhance their innovation capabilities. the empirical results of jiang shuiquan et al. [37] show that the shareholding ratio of qfii is positively correlated with enterprise innovation investment, and the marginal effect of qfii shareholding on enterprise innovation investment is better. liu ye and meng hantong [38] found that both qfii shareholding and enhanced shareholding checks and balances can promote enterprise innovation investment; increased environmental uncertainty will inhibit the promotion of qfii to enterprise innovation investment; qfii increases enterprise innovation investment by easing enterprise financing constraints. from the perspective of the shanghai shenzhen hong kong stock connect, ma yanyan et al. [39] found that the land port connection is conducive to the promotion of enterprise research and development scale; the stimulation of the stock connect to r&d is mainly achieved by reducing credit dependence [40], improving the role of external supervision [41], and increasing the level of tfp. it can improve the level of r&d by solving the productivity paradox. feng ruoyang and wen jun [42] found that the stock connect system has improved the level of technological innovation of jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 84 soes; it can alleviate the financing constraints faced by soes, promote soes to increase their r&d investment and improve the level of technological innovation. it can provide more opportunities for long-term institutional investors to enter by improving the stock liquidity of soes, and improve the technological innovation level of soes. qi di [43] found that after the implementation of the stock connect, the target company significantly increased its r&d investment and patent output [44], improving enterprise innovation performance. zhu lin and yi zhihong [45] found that after the implementation of the stock connect trading system, the innovation level of the target enterprise significantly increased; managers' professional anxiety [46, 47] plays a mediating and moderating role. the information environment of listed companies has regulatory effects. huang jianqiao et al. [48] found that the stock connect significantly improved the level of enterprise innovation output, and this effect is more significant in enterprises with a high degree of separation of ownership, low shareholding ratio of institutional investors, poor information environment, and high financing needs. jia lihuan and xiao xiang [49] found that the stock connect can promote high-quality development of enterprises by reducing agency costs and increasing innovation investment. research by sun zeyu and qi baolei [50] shows that the stock connect system improves innovation performance by enhancing management's willingness to innovate and improving enterprise innovation capabilities. the empirical study by liu yang and zang rihong [51] shows that the implementation of the sh-hk stock connect trading system has a significant promoting effect on enterprise innovation, which is manifested in increasing the willingness of enterprises to innovate and improving the level of innovation input and output. wu yuxuan and dong li [52] demonstrated empirically that the stock connect policy can promote enterprise technological innovation by optimizing the level of corporate governance of the target enterprise and reducing the i and ii types of agency costs in corporate governance [53, 54]. based on the above research, li chengming et al. [55] found in their mechanism analysis that the stock connect can help industrial policies attract more enterprise r&d investment, and help improve the utilization efficiency of the investment. lv xiaojun et al. [56] found that the stock connect significantly promoted the quality of enterprise innovation, manifested by the gradual shift in the patent structure toward high-quality invention patents, as well as the increase in patent citation rates. 3.2. direct foreign equity participation yang dewei [57] empirically shows that foreign shareholding is significantly and negatively correlated with enterprise technological innovation. based on the research conducted by guadalupe et al. [58], boubakri et al. [59], and others, li wengui and yu minggui [60] argue that foreign shareholding can bring benefits such as improving corporate governance and technology spillovers for enterprises to carry out innovation activities, but the empirical test results are not significant. based on the research conducted by choi et al. [61] and luong et al. [34], zhong xi et al. [62] empirically demonstrated that foreign shareholding will promote firms’ innovation performance. internationalization strategy plays a partially mediating role. the gap in business expectation enhances the promoting effect of foreign ownership on enterprise innovation performance, while redundant resources weaken this promoting effect. di lingyu et al. [63] found that foreign shareholders' equity participation significantly improves the level of r&d investment in soes. when the source of foreign capital is relatively developed, and foreign shareholders take strategic cooperation as the main motivation for equity participation, their equity participation has a more jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 85 significant effect on improving the level of r&d investment in soes. 4. the empirical study on the promoting mechanism of high-level opening of insurance industry on innovative development of manufacturing industry at present, domestic and foreign research mainly focuses on the role of the host country's insurance industry in enterprise innovation and its mechanism testing. in terms of technology insurance, sun hongtao [64] argues that the support of the u.s. intellectual property insurance for technology innovation is demonstrated by the wide range of subjects protected by intellectual property insurance [65], spreading liability and promoting the progress of patented technologies, and protecting small companies from the oppressive litigation of large companies [66]. xu xiaohui [67] elaborated on the approach of how technology insurance reduces technology risks from aspects such as technology personal insurance, technology property insurance, and the synchronization of insurance and management. wang xianglan [68] clarified that technology insurance can effectively resolve and transfer the technological risks faced by enterprises in their independent innovation activities. ma yanxin [69] demonstrated that growing input on insurance has a significant promoting effect on the improvement of technological innovation capabilities of industrial enterprises. ge yu and wang yuxin [70] believe that technology insurance can spread the risks of scientific and technological innovation, compensate for the losses, accumulate funds, and supervise the risks of scientific and technological innovation. wang lei and gu mengdi [71] provided a systematic theoretical framework for research on technology insurance. tiller and bedigian [72], weixian xue et al. [73] have empirically studied the impact of technology insurance on innovation and profitability of technology companies. shen fei et al. [74] empirically demonstrated that patent enforcement insurance enhances enterprise technological innovation by reducing technology spillover losses [75] and external financing constraints [76]. research by qiu yangdong [77] shows that the incentive effect of the patent insurance pilot policy on enterprise innovation is reflected in encouraging enterprises to increase r&d investment. in terms of director and executive liability insurance, jensen [78] believes that director liability insurance will have a significant impact on directors' behavior and decision-making, thereby affecting enterprise resource allocation and enterprise innovation. d&o insurance improves corporate governance through both external oversight [79] and signaling [80, 81]. ling shixian and bai ruifeng [82] empirically tested the impact of directors' liability insurance on enterprise innovation and the path of its governance function. hu guoliu et al. [83] empirically demonstrated that purchasing d&o insurance can help strengthen the promotion of small and medium-sized investor protection on technological innovation of enterprises. fang junxiong and qin xuan [84] found a significant positive correlation between d&o insurance and company innovation. li conggang and xu rong [85] empirically found that the introduction of d&o insurance significantly improved the innovation output and efficiency of enterprises, consistent with the incentive effect hypothesis [86]. hu guoliu et al. [87] empirically examined the impact of d&o insurance on enterprise independent innovation from the perspective of manager risk tolerance [88]. xia tongshui and zang xiaoling [89] empirically investigated the impact of d&o insurance on firms' technological innovation and the moderating role of equity structure in it. zhai shuping et al. [90] explored the impact and mechanism of d&o insurance on enterprise innovation efficiency based on the incentive and supervision hypothesis [91] and the moral hazard hypothesis. ling shixian and liu ao [92] empirically examined the impact of jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 86 d&o insurance on enterprise innovation, and examined the interaction between management compensation incentives and directors' liability insurance. research by wang et al. [93], zhang zenglian and xu fangyuan [94] shows that d&o insurance promotes high-quality development of enterprises by reducing agency costs and improving innovation capabilities. fei shen et al. [95] demonstrated that d&o insurance affects enterprise innovation through management risk appetite, management level, and incentive mechanisms. zhao guoyu and liang huiping [96] explored the role and path of director liability insurance in enterprise innovation from the perspective of external investors. the empirical study by zhou donghua et al. [97] shows that the purchase of directors' liability insurance can significantly improve the level of enterprise risk taking [98] and enhance the innovation ability of enterprises. in addition, xiao xiaohong and pan ye [99] examined the impact of d&o insurance on green innovation. gao kai et al. [100] studied the relationship between d&o insurance and enterprise green innovation from the perspective of executive decision making and behavior. in terms of environmental pollution liability insurance, ning jinhui et al. [101] demonstrated empirically that applying for environmental liability insurance [102] can promote the improvement of enterprise innovation capabilities, and it plays a positive role in enterprise innovation governance. environmental liability insurance improves the innovation ability of enterprises by easing financing constraints. environmental liability insurance has a more significant effect on promoting the innovation level of non-soes. in terms of insurance fund holdings, research by luong et al. [34] and liu dongjiao et al. [103], insurance fund shareholding promotes the efficiency of enterprise innovation by improving the quality of internal control. 5. brief review (1) in terms of empirical research on the promotion mechanism of high-level opening-up of the banking industry to the innovative development of the manufacturing industry, currently, relevant domestic and foreign research mainly focuses on the role of opening-up of the banking industry to foreign countries in enterprise innovation and its mechanism testing, among which, there are many mechanisms testing such as financing constraint effect, cost saving effect, trade promotion effect, and industry competition effect. however, it is rare to examine mechanisms such as the correlation effect of foreign banks and the network expansion effect of foreign banks. at the same time, existing studies have focused on simple mediating effect tests, while simple regulatory effect tests and threshold effect tests are relatively rare; the tests of complex mediating effects, complex regulatory effects, and complex threshold effects are relatively rare. in addition, the perspectives of heterogeneity testing are not rich enough. to sum up, empirical research on the promotion mechanism of high-level opening-up of the banking industry on the innovative development of the manufacturing industry at home and abroad needs to be further expanded. (2) in terms of empirical research on the driving mechanism of high-level openness in the securities industry for the innovative development of the manufacturing industry, at present, relevant studies at home and abroad mainly focus on the role of capital market opening and foreign direct equity participation on enterprise innovation along with its mechanism tests. among which, there are more tests on the mechanisms of financing constraint effect, technology spillover effect, improving corporate governance effect, while the tests on the mechanisms such as stock liquidity jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 87 effect, debt structure optimization effect, innovation willingness effect, internationalization strategy effect are less frequently examined. besides, simple intermediation effect tests are the focus of existing studies, while simple regulation effect tests and threshold effect tests are rare. complex intermediation effect, complex regulation effect and complex threshold effect tests are rare. in addition, the perspective of heterogeneity tests is not rich enough. in summary, the empirical research on the mechanism of promoting the development of manufacturing innovation by high level of opening-up of securities industry at home and abroad needs to be further expanded. (3) in terms of empirical research on the promotion mechanism of high-level opening-up of the insurance industry to the innovative development of the manufacturing industry, at present, the relevant domestic and foreign research mainly focuses on the role of the insurance industry in the host country in enterprise innovation and its mechanism testing. among them, there are more studies on the effects of technology insurance and director and executive liability insurance, while relatively few studies are on the effects of environmental pollution liability insurance and insurance fund shareholding. at present, there are studies at home and abroad that have rarely examined the mechanism of the impact of foreign insurance on enterprise innovation, especially these mechanisms such as the technology insurance personal insurance effect, the technology insurance property insurance effect, the technology insurance cooperative development effect, the director and executive liability insurance effect, the environmental pollution liability insurance effect, and the linkage effect of foreign insurance companies. at the same time, existing studies have focused on simple mediating effect tests, while simple regulatory effect tests and threshold effect tests are relatively rare; the tests of complex mediating effects, complex regulatory effects, and complex threshold effects are relatively rare; in addition, the perspectives of heterogeneity testing are not rich enough. to sum up, empirical research on the driving mechanism of high-level opening-up of the insurance industry on the innovative development of the manufacturing industry at home and abroad has just begun. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] fan y s, yin x n. research group of xilingguolemeng central sub-branch of the people's bank of china. research on the transmission mechanism of financial support promoting the manufacturing industry [j]. northern finance journal, 2021 (02): 85-90. doi: https://doi.org/10.16459/j.cnki.15-1370/f.2021.02.019. [2] bian w w. progress and challenges of china’s financial opening-up [j]. jiangsu commercial forum, 2023 (01): 86-90. doi: https://doi.org/10.13395/j. cnki. issn.1009-0061.2023.01.008. [3] guo w, si m h. logic mechanism, experience inspiration and future orientation of 70-year financial opening in new china [j]. world economy studies, 2019 (10): 15-26+88+134. [4] yu b. the important reason for the imbalance between finance and the real economy lies in the real economy [j]. china development observation, 2017 (05): 30-32. doi: https://doi.org/10.3969/j.issn.1673033x.2017.05.009. [5] brown j r, martinsson g, petersen b c. law, stock markets, and innovation. the journal of finance, vol.68, no.4, 2013, pp.1517-1549. doi: https://doi.org/10.1111/jofi.12040. [6] xie w m, fang h x. financial development, financing constrains and firms' r&d investment [j]. journal of financial research, 2011 (05): 171-183. jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 88 [7] liu f, zheng l s. a study on the high-level financial opening in china [j]. theoretical exploration, 2021, (03): 84-92. doi: https://doi.org/10.3969/j.issn.1004-4175.2021.03.013. [8] wang f h. building rules to adapt to the high-level opening of finance [j]. international finance, 2021 (02): 27-29. doi: https://doi.org/10.16474/j.cnki.1673-8489-2021.02.005. [9] zhang y y, lian p, wu g, zhao q m, wang x. the path and risk prevention of high-level financial opening [j]. china forex, 2022 (13): 18-25. doi: https://doi.org/10.13539/j.cnki.11-5475/f.2022.13.022. [10] zhou q l, zhu y j. realizing high-level opening of the financial industry to the outside world, promoting quality optimization of the financial supply system [j]. finance economy, 2019 (23): 13-15. doi: https://doi.org/10.14057/j.cnki.cn43-1156/f.2019.23.003. [11] cornaggia j, mao y, tian x, wolfed b. does banking competition affect innovation? [j]. journal of financial economics, 2015, 115(1): 189-209. doi: https://doi.org/10.1016/j.jfineco.2014.09.001. [12] tsai y j, chen y, lin c, hung j. the effect of banking system reform on investment-cash flow sensitivity: evidence from china [j]. journal of banking and finance, 2014, 46(01): 166-176. doi: https://doi.org/10.1016/j.jbankfin.2014.04.022. [13] bertrand m, schoar a, thesmar d. bank deregulation and industry structure: evidence from the french banking reforms of 1985 [j]. journal of finance, 2007, 62(02): 597-628. doi: https://doi.org/10.1111/j.15406261.2007.01218.x. [14] jian z, gan c h, yu d f. marketization of banking system, allocation of credit and industry restructuring. [j]. economic research journal, 2013, 48(05): 112-127. [15] bai j, meng q x, shen y y. does foreign bank entry affect domestic enterprise innovation? [j]. accounting research, 2018 (11): 50-55. doi: https://doi.org/10.3969/j.issn.1003-2886.2018.11.007. [16] claessens s, demirgüc-kunt a, huizinga h. how does foreign entry affect domestic banking markets？ [j]. journal of banking & finance, 2001, 25(05): 891-911. doi: https://doi.org/10.1016/s0378-4266(00)00102-3. [17] akbar y h, mcbride j b. multinational enterprise strategy, foreign direct investment and economic development: the case of the hungarian banking industry [j]. journal of world business, 2004, 39(01): 89105. doi: https://doi.org/10.1016/j.jwb.2003.08.001. [18] laeven l, levine r, michalopoulos s. financial innovation and endogenous growth [j]. journal of financial intermediation, 2015, 24(01): 1-24. doi: https://doi.org/10.1016/j.jfi.2014.04.001. [19] li j q, xie f. may foreign banks entry promote corporate innovation? ——analysis based on crosscountry evidence [j]. studies of international finance, 2020 (01): 54-64. [20] acharya v, xu z. financial dependence and innovation: the case of public versus private firms [j]. journal of financial economics, 2017, 124(02): 223-243. doi: https://doi.org/10.1016/j.jfineco.2016.02.010. [21] blanco i, wehrheim d. the bright side of financial derivatives: options trading and firm innovation [j]. journal of financial economics, 2017, 125(01): 99-119. doi: https://doi.org/10.1016/j.jfineco.2017.04.004. [22] zhu z j, huang x h, yu x. financial sector liberalization and promotion of chinese manufacturing firms' competitiveness [j]. the journal of quantitative & technical economics, 2018, 35 (03): 114-131. [23] aghion p, bloom n, blundell r, griffith r, howitt p. competition and innovation: an inverted-u relationship [j]. quarterly journal of economics, 2005, 120(02): 701-728. doi: https://doi.org/10.1093/qje/120.2.701. [24] impullitti g, licandro o. trade, firm selection and innovation: the competition channel [j]. economic journal, 2018, 128(608): 189-229. doi: https://doi.org/10.1111/ecoj.12466. [25] aghion p, bergeaud a, lequien m, melitz m j. the heterogeneous impact of market size on innovation: evidence from french firm-level exports [r]. harvard university working paper, 2019. doi: https://doi.org/10.1162/rest_a_01199. [26] zhu z j, chen h y, wang f. opening of banking industry and the solution to the innovation trap of chinese firms [j]. china industrial economics, 2020 (10): 175-192. doi: https://doi.org/10.3969/j.issn.1006480x.2020.10.010. [27] sheng b, wang h. financial openness, independent innovation, and firms’ export products quality: analysis on experience of foreign banks entering china [j]. finance and trade research, 2021, 32 (11): 1-15. [28] yu x l, ma h j. research on the absorption of international knowledge spillover and the improvement of innovation ability of foreign banks [j]. new finance, 2022 (02): 38-48. [29] eichengreen b, gullapalli r, panizza u. capital account liberalization, financial development and industry growth: a synthetic view [j]. journal of international money and finance, 2011, 30(06): 1090-1106. doi: https://doi.org/10.1016/j.jimonfin.2011.06.007. jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 89 [30] larrain m, stumpner s. capital account liberalization and aggregate productivity: the role of firm capital allocation [j]. the journal of finance, 2017, 72(04): 1825-1858. doi: https://doi.org/10.1111/jofi.12497. [31] xie f, guo n. the effect of capital account sub-items opening on corporate innovation? evidence form cross-country data [j]. journal of central south university of economics and law, 2019 (05): 124-133. [32] moshirian f, tian x, zhang b, et al. stock market liberalization and innovation [j]. journal of financial economics, 2021, 139(03): 985-1014. doi: https://doi.org/10.1016/j.jfineco.2020.08.018. [33] xie f. does stock market liberalization promote corporate innovation? cross-country evidence [j]. nankai economic stuies, 2022 (07): 145-161. [34] luong h, moshirian f, nguyen l, tian x, zhang b. how do foreign institutional investors enhance firm innovation? [j]. journal of financial and quantitative analysis, 2017, 52(04): 1449-1490. doi: https://doi.org/10.1017/s0022109017000497. [35] bena j, et al. are foreign investors locusts? the long-term effects of foreign institutional ownership [j]. journal of financial economics, 2017, 126(01): 122-146. doi: https://doi.org/10.1016/j.jfineco.2017.07.005. [36] jiang w j. research on technology spillover effect of qfii based on perspective of financial opening [j]. zhejiang finance, 2018 (03): 55-61. doi: https://doi.org/10.3969/j.issn.1674-5477.2018.02.005. [37] jiang s q, chen y y. institution environment, foreign equity participation and enterprise innovation: an empirical study based on qualified foreign institutional investors' shareholding [j]. friends of accounting, 2020 (21): 50-57. doi: https://doi.org/10.3969/j.issn.1004-5937.2020.21.008. [38] liu y, meng h t. research on the impact of qualified foreign institutional investors on enterprise innovation investment——based on the perspective of financing constraints and environmental uncertainty [j] industrial economic review, 2022, 13(02): 39-52. [39] ma y y, yu m m, cheng j j. does stock market liberalization cause firms’ r&d investment boom? —— evidence from the sample of mainland-hong kong stock connect program [j]. collected essays on finance and economics, 2019 (08): 39-52. [40] xu f. research on bank credit and enterprise innovation dilemma [j]. china industrial economics, 2019 (01): 119-136. [41] gillan s l, starks l t. corporate governance proposals and shareholder activism: the role of institutional investors [j]. journal of financial economics, 2000, 57(02): 275-305. doi: https://doi.org/10.1016/s0304405x(00)00058-1. [42] feng r y, wen j. does shanghai-hong kong stock connect enhance technological innovation of stateowned enterprises in china? [j]. industrial economics research, 2019 (04): 88-100. [43] di q. research on the sh-sz-hk stock connect mechanism and business innovation: empirical evidences based on multi-phase did [j]. contemporary finance & economics, 2020 (02): 76-88. [44] ang j s, cheng y, wu c. does enforcement of intellectual property rights matter in china? evidence from financing and investment choices in the high-tech industry [j]. the review of economics and statistics, 2014, 96(02): 332-348. doi: https://doi.org/10.1162/rest_a_00372. [45] lin z, yi z h. does the hong kong-shanghai stock connect spur corporate innovation? [j]. business and management journal, 2020, 42(02): 40-57. [46] wright p, et al. impact of corporate insider, blockholder, and institutional equity ownership on firm risk taking [j]. the academy of management journal, 1996, 39(02): 441-463. doi: https://doi.org/10.5465/256787. [47] manso g. motivating innovation [j]. the journal of finance, 2011, 66(05): 1823-1860. doi: https://doi.org/10.1111/j.1540-6261.2011.01688.x. [48] huang j q, wang y q, deng y l, chen y s. does stock market liberalization improve corporate innovation output? a quasi-experiment of china's mainland-hong kong stock connect [j]. accounting and economics research, 2020, 34(01): 21-37. [49] jia l h, xiao x. stock market liberalization and corporate high-quality development: from the perspective of agency cost and innovation incentive [j]. modern economic research, 2021 (12): 105-115+132. [50] sun z y, qi b l. capital market liberalization and corporate innovation——a qusai-natural experiment from “shanghai-hong kong” and “shenzhen-hong kong” stock connect [j]. science& technology progress and policy, 2021, 38(08): 109-117. [51] liu y, zang r h. capital market opening and enterprise innovation based on the empirical data of shanghai-hong kong stock connect [j]. journal of beijing technology and business university (social science edition), 2021, 36(03): 79-90. jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 90 [52] wu y h, dong l. impact of capital market opening on technological innovation of china's listed manufacturing enterprises: based on mediation effect of corporate governance [j]. science and technology management research, 2022, 42(12): 9-16. doi: https://doi.org/10.3969/j.issn.1000-7695.2022.12.002. [53] ferreira m a, matos p. the colors of investors’ money: the role of institutional investors around the world [j]. journal of financial economics, 2007, 88(03): 499-533. doi: https://doi.org/10.1016/j.jfineco.2007.07.003. [54] aggarwal r, et al. does governance travel around the world? evidence from institutional investors [j]. journal of financial economics, 2010, 100(01): 154-181. doi: https://doi.org/10.1016/j.jfineco.2010.10.018. [55] li c m, li y f, dong z y. capital market opening and industrial policy effectiveness: based on the perspective of enterprise innovation [j]. industrial economics research, 2022 (03): 1-14+40. [56] lv x j, hu h x, wang h j. capital market liberalization and enterprise innovation quality based on the perspective of “innovation trap” of chinese enterprises emphasizing quantity over quality [j]. journal of systems & management, 2002, 31(05): 1018-1027. [57] yang d w. empirical study on the impact of ownership structure on enterprise technological innovation: based on the analysis of listed companies on china's small and medium sized board [j]. public finance research, 2011 (08): 56-60. [58] guadalupe m, kuzmina o, thomas c. innovation and foreign ownership [j]. american economic review, 2012, 102: 3594-3627. [59] boubakri n, cosset j, saffar w. the role of state and foreign owners in corporate risk-taking: evidence from privatization [j]. journal of finance and economics, 2013, 108: 641-658. doi: https://doi.org/10.1016/j.jfineco.2012.12.007. [60] li w g, yu m h. ownership structure and enterprise innovation of privatized enterprises [j]. management world, 2015 (04): 112-125. [61] choi s b, lee s h, williams c. ownership and firm innovation in a transition economy: evidence from china [j]. research policy, 2011, 40(3): 441-452. doi: https://doi.org/10.1016/j.respol.2011.01.004. [62] xi z, song t b, chen w h, zhu z j. how foreign shareholdings influence innovation performance? - mediating role of internationalization strategy and moderating role of a firm’s behavior [j]. systems engineering, 2020, 38 (03): 112-122. [63] di l y, bu d l, shi x y. industrial policy coordination between central and local governments, foreign equity participation, and the r&d investment level of state-owned companies. [j]. industrial economics research, 2021 (05):83-96. [64] sun h t. intellectual property insurance in the united states: the amulet of enterprise technological innovation [j]. science & technology progress and policy, 2006 (09): 16-18. doi: https://doi.org/10.3969/j.issn.1001-7348.2006.09.005. [65] reyes j a. patents and insurance: who will pay for infringement? [j]. boston university journal of science and technology law, 1995, 3: 21. [66] stueber t j. insurance coverage for patent infringement [j]. william mitchell law review, 1991, 17: 10801082. [67] xu x h. problems and solutions of the scientific and technological innovation supporting the independent innovation of enterprises [j]. scientific management research, 2011, 29 (04): 109-112. [68] wang x l. research on countermeasures of technological insurance in support of independence innovations of enterprises——setting hebei province as an example [j]. technoeconomics & management research, 2011 (08): 58-61. doi: https://doi.org/10.3969/j.issn.1004-292x.2011.08.014. [69] ma y x. research on the impact of the insurance growth on the technological innovation of chinese industrial enterprises [j]. economy and management, 2011, 25 (10): 37-41. [70] ge y, wang y x. resolve the risk of high-tech enterprises in technological innovation with science technology insurance [j]. scientific management research, 2012, 30(01): 105-108. doi: https://doi.org/10.3969/j.issn.1004-115x.2012.01.026. [71] wang l, gu m d. research on the insurance support model for technological innovation in small and medium-sized enterprises: a theoretical framework [j]. modern management science, 2013 (07): 9-11+18. doi: https://doi.org/10.3969/j.issn.1007-368x.2013.07.003. [72] tiller s e, bedigian b. intellectual property and technological insurance coverage [j]. maryland bar journal, 2001, 34: 15-18. jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 91 [73] xue w x, liu q, liu j. study on the effect of technological insurance on technological enterprises' innovation and profit ability [j]. science & technology progress and policy, 2013, 30(24): 95-99. doi: https://doi.org/doi:10.6049/kjjbydc.2013090623. [74] shen f, zhou y, liu j f(a). patent enforcement insurance, corporate innovation and operating performance [j]. journal of industrial technological economics, 2021, 40(04): 119-128. doi: https://doi.org/10.3969/j.issn.1004-910x.2021.04.014. [75] pandit s, wasley c e, zach t. the effect of research and development (r&d) inputs and outputs on the relation between the uncertainty of future operating performance and r&d expenditures [j]. journal of accounting, auditing & finance, 2011, 26(01): 121-144. doi: https://doi.org/10.1177/0148558x1140058. [76] ciftci m, cready w m. scale effects of r&d as reflected in earnings and returns [j]. journal of accounting and economics, 2011, 52(01): 62-80. doi: https://doi.org/10.1016/j.jacceco.2011.02.003. [77] qiu y d. can patent insurance motivate enterprises to innovate? based on the perspective of extensive margin and intensive margin [j]. industrial economics research, 2022 (02): 114-127. [78] jensen m. the modern industrial revolution, exit and the failure of internal control systems [j]. the journal of finance, 1993, 48(03): 831-880. doi: https://doi.org/10.1111/j.1540-6261.1993.tb04022.x. [79] yuan r l, sun j, cao f. directors’ and officers’ liability insurance and stock price crash risk [j]. journal of corporate finance, 2016, 37: 173-192. doi: https://doi.org/10.1016/j.jcorpfin.2015.12.015. [80] gupta m, prakash p. information embedded in directors and officers insurance purchases [j]. the geneva papers on risk and insurance issues and practice, 2012, 37(03): 429-451. [81] boyer m m, stern l h. is corporate governance risk valued? evidence from directors' and officers' insurance [j]. journal of corporate finance, 2012, 18(2): 349-372. doi: https://doi.org/10.1016/j.jcorpfin.2011.12.005. [82] ling s x, bai r f. directors' liability insurance, board governance and corporate innovation [j]. science & technology progress and policy, 2018, 35(10): 100-106. [83] hu g l, zhao y, hu j. directors’ and officers’ liability insurance, tolerance of failure and enterprise independent innovation [j]. management world, 2019, 35(08): 121-135. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2019.0110. [84] fang j x, qin x. the relief of executive professional risk and improvement of innovation decision—— based on the evidence of d&o insurance [j]. insurance research, 2018 (11): 54-70. [85] li c g, xu r. d&o insurance, corporate governance and corporate innovation: evidence from chinese a-share listed firms [j]. financial regulation research, 2019 (06): 85-102. doi: https://doi.org/10.3969/j.issn.2095-3291.2019.06.006. [86] li j t, tang y. ceo hubris and firm risk taking in china: the moderating role of managerial discretion [j]. the academy of management journal, 2010, 53(01): 45-68. doi: https://doi.org/10.5465/amj.2010.48036912. [87] hu g l, zhang x, zeng c h. directors’ and officers’ liability insurance, ordinary investor protection and enterprise technological innovation [j]. soft science, 2018, 32(07): 89-92+138. [88] tian x, wand t y. tolerance for failure and corporate innovation [j]. review of financial studies, 2014, 27: 211-255. doi: https://doi.org/10.1093/rfs/hhr130. [89] xia t s, zang x l. director and executive liability insurance, equity structure, and enterprise technology innovation [j]. finance and accounting monthly, 2019 (22): 168-176. [90] zhai s p, zhang x l, wang m. directors’ and officers’ liability insurance and corporate innovation efficiency --“make the best use of the situation” or “aggravate a complicated situation”? [j]. journal of business economics, 2020 (04): 52-67. [91] o’sullivan n. insuring the agents: the role of directors' and officers' insurance in corporate governance [j]. the journal of risk and insurance, 1997, 64(03): 545-556. doi: https://doi.org/10.2307/253764. [92] ling s x, liu a. directors' and officers' liability insurance, management incentive and corporate innovation——an empirical study based on the data of a-share listed companies [j]. financial regulation research, 2020 (09): 50-65. [93] wang j, zhang j, huang h, et al. directors' and officers’ liability insurance and firm innovation [j]. economic modelling, 2020, 89: 414-426. doi: https://doi.org/10.1016/j.econmod.2019.11.011. [94] zhang z l, xu f y. directors' and executives' liability insurance and high-quality enterprise development: based on the perspective of agency cost and innovation incentive [j]. east china economic management, 2021, 35 (02): 78-86. jia-qi wu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 80-92 doi: https://doi.org/10.54560/jracr.v13i1.357 92 [95] shen f, zhou y, liu j f (b). does directors’ and officers’ liability insurance innovate or suppress enterprise innovation [j]. journal of technology economics, 2021, 40(05): 82-92. doi: https://doi.org/10.3969/j.issn.1002-980x.2021.05.009. [96] zhao g y, liang h p. does director and executive liability insurance promote enterprise innovation—— from the perspective of credit rent-seeking and financing constraints [j]. foreign economics and management, 2022, 44(03): 103-117. [97] zhou d h, luo s z, zhao y j. directors' and officers' liability insurance and enterprise innovation [j]. scientific research management, 2022, 43(04): 201-208. [98] core j e. on the corporate demand for directors' and officers' insurance [j]. journal of risk & insurance, 1997, 64(01): 63-87. doi: https://doi.org/10.2307/253912. [99] xiao x h, pan y. directors' and officers' liability insurance and enterprise green innovation: incentive tools or self-interested means? [j]. science & technology progress and policy, 2022, 39(13): 94-104. doi: https://doi.org/10.6049/kjjbydc.2020120115. [100] gao k, zhao h q, wang l. directors' and officers' liability insurance and green innovation of manufacturing enterprises: an analysis from the mediating effect of internal control [j]. east china economic management, 2022, 36(02): 119-128. [101] ning j h, jin y, zhang y b. environmental pollution liability insurance and enterprise innovation: promote or suppress? [j]. science & technology progress and policy, 2019, 36 (17): 90-97. doi: https://doi.org/10.6049/kjjbydc.2019010455. [102] misheva, irena. environmental insurance: a part of risk of management in enterprises utilizing hazardous production [j]. economics, 2016, 21(02): 60-84. [103] liu d j, liu k, zhuang p t. insurance funds shareholding and corporate innovation efficiency--empirical evidence from a-share listed companies [j]. collected essays on finance and economics, 2021 (08): 59-68. copyright © 2023 by the authors. this is an open-access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 常用qq:361839670 手机：15738553806 received 16 march 2016 accepted 2 may 2016 study on performance evaluation of government comprehensive supervision for safety production based on balanced score card --a case study in shandong province, china hao yu1,2 1 school of emergency management, henan polytechnic university, jiaozuo 454000, henan, china 2china university of geosciences (beijing), beijing 100083, china e-mail:haoyu_he@163.com abstract the government supervision is the key to guarantee work safety. in order to improve the level of safety production supervision of the government, a performance evaluation indicator system of safety production supervision within city and county two levels for the local government has been established according to the theory of key performance indicators. 4 first-level indicators and 40 secondary indicators have been designed. the weight of these indicators has been determined by using the analytic hierarchy process and delphi method. this study may contribute to providing management tools and ideas for the local government performance evaluation on the supervision safety production in china. keywords: local government, safety production supervision, performance evaluation, indicator system 1. introduction safety production is a sign of social civilization progress, and an important booster to promote the economic transformation and upgrading. in recent years, there are a large number of safety accidents in china. from the start of 2015, all kinds of safety accidents have occurred frequently, shanghai bund stampede is heart-wrenching, then harbin "1.2" fire accident, shaanxi xianyang "5.15" traffic accident, henan pingdingshan "5.25" fire accident, and "oriental star" passenger ship capsized accident is still a painful accident. the frequent occurrence of these accidents brings huge pain and sufferings to people, at the same time, the social economic construction and national development also suffer. to find an effective way to prevent these safety accidents, the government and all relevant personnel have made great efforts, especially in the regulation of work safety. the frequent safety accidents are related to less attention to safety production, lacking in the designated position and malfeasance behaviors of individual officials with the value orientation of local government centering on gdp. with the purpose of changing the status quo of safety production in china, the state council of china adopted various measures to improve the efficiency for regulation and administration on safety production each level of government, the introduction of the performance management is one of the important measures. an effective method of performance management is the implementation of performance assessment, namely test and evaluation are carried out to the supervision of relevant government departments by using targeted indicators or standards. and through the results of the evaluation, the shortages of government are reflected to guide the government supervision in the future. so, establishing a scientific and practical government performance evaluation system of safety production supervision is the long-term effective measure to improve the efficiency of safety production. as to the world recognized difficult problem of government performance evaluation, evaluation indicator system design is arguably one of the difficulties on account of the wide scope of the government activities, complex process and the complicated factors. although the domestic academic circles have a lot of research on journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 85-94 published by atlantis press copyright: the authors 85 h. yu / study on performance evaluation of government comprehensive supervision government performance evaluation, a set of widely accepted indicator system still does not appear, and practicality is poor. in terms of performance evaluation of government regulation and work safety, in recent years, some scholars and experts have made some relevant research on the performance evaluation of safety production supervision, and established relevant indicator systems, however, in view of the local government overall performance evaluation of safety production supervision, few scholars designed the concerned indicator system. therefore, this study intends to explore the local government performance evaluation system of safety production supervision, combined the predecessors' research results and practical investigation, a set of assessment system applicable to the city (county) level is developed, namely (city) county safety production supervision performance evaluation indicator system, and then the related processes for implementation evaluation is discussed. this study may offer scientific theory and method, reasonable and effective tools and methods for performance evaluation work for the government safety production comprehensive supervision, which can improve safety production supervision performance of the government and then improve the overall efficiency. in this study, on-the-spot investigation method is adopted, including interview and questionnaire survey. interview survey's goal is to define the job requirements, difficulties and problems of the government in the supervision of safety production and the survey can provide material and basis. this study uses interview method, according to the experiences of previous evaluation of relevant experts in tai’an city government; the filtered set indicator in the first round is amended, but only the secondary indicators change, the first-level indicators remain unchanged. in this case, the second round evaluation indicator of the performance evaluation of safety production supervision for local government is made sure. questionnaire survey method is to use special questionnaires, questionnaire of second round indicator will be distributed to the tai’an concerned municipal government departments, and eligible staff information, instructions on the use of the questionnaire and evaluation indicator are listed in the questionnaire. tai’an is a city of shandong province, china. simple words are used to explain the meaning of each indicator, and tai’an municipal government departments are organized to fill in the investigators, some advice is collected, and then the evaluation indicator system is determined, at last the reasonable feasibility is verified. 21 questionnaires are distributed to tai’an government, 15 questionnaires are distributed to tai’an administration of work safety, 15 questionnaires are distributed to each of the four subordinate areas and 85 effective questionnaires are taken back to determine the final result. 2. the framework of performance evaluation indicator system of work safety regulation for the local government 2.1. the strategy map of local government work safety regulation balanced score card (bsc) is proposed by robert s. kaplan and david p. norton in the united states in 1992, which is the latest and the most comprehensive in theory and method of enterprise performance management at present. this method decomposes the enterprise strategic target to layers, and finally into a variety of detailed, balanced performance evaluation system for each other, from the financial performance, business process, innovation and growth, customer satisfaction four dimensions, a benign pdca circulation is formed. this study tries to build a performance evaluation system of safety regulation for the local government under the framework of bsc (six factors and four dimensions), based on the smart principles of key performance indicators (kpi). four dimensions of safety regulation performance management for the local government are shown in fig. 1. a clear and complete safety strategy map is the beginning of sound cycle of safety regulation performance management for local government. according to the safety strategy map, this research designs indicators of safety production supervision performance for local government based on kpi theory. and smart principle is the kpi setting criterion, namely each selection for indicator of the system should follow the principle of smart principles, five basic principles are shown in table 1. published by atlantis press copyright: the authors 86 h. yu / study on performance evaluation of government comprehensive supervision fig. 1. four dimensions relationship of safety regulation performance management table 1. smart principles s（specific）principle all key performance indicators are to clearly describe the required action plans completed by the evaluation main body and object. m（measurable）principle key performance indicators should be quantified as much as possible; there should be quantitative data, such as quantity, quality, and time and so on, which can be measured objectively. a（attainable）principle two aspects: first, the quota is moderate and reasonable, which is on the premise of consensus between the superior and lower level, in government departments, and the government departments and enterprises can control the scope of the task; second, the target must be "after a certain effort" to implement, and not just repeat the target before. r（relevant）principle two aspects: first, superior objectives must be set before the lower targets, and they have consistency, avoid repeating or target fault; second, staff's kpi target is associated with individual main job duties within the team. t（time-bound）principle there is no difference between the targets without time limits and no targets. 2.2. performance evaluation indicator system of work safety regulation for the local government safety production supervision process is a complex multi-factor, multi-variable, multi-level system, with complicated process and related factors, and for a long time, it is always difficult to take quantitative analysis. therefore, comprehensive consideration of various factors must be in the process of selection for performance evaluation indicator of government safety production supervision because the selected indicators accuracy will directly affect the measurement result. not only that, the design of the indicator weight is also very important, if the weights are designed not reasonable, the assessment result is also biased. safety production law of the people’s republic of china (amendment) came into force on december 1 last year, of which a mechanism financial condition to succeed in financial aspect, the government should show the public management cost. supervision strategy target supervision evaluation target work safety vision and strategy customer service for the realization of the idea, the government should prove their management standards to the public. supervision strategy supervision evaluation target internal point of view to satisfy the social public, the government should regulate management process. supervision strategy target supervision evaluation target financial condition for the realization of the idea, measures should be taken by the government to implement the departments and staff learning and development. supervision strategy target supervision evaluation target published by atlantis press copyright: the authors 87 h. yu / study on performance evaluation of government comprehensive supervision of five parties including the responsibilities of business entities, participation of employees, governmental supervision, industry self regulation, and supervision from the general public shall be established, and also powers and duties the government safety administrative departments at all levels should conduct are defined. so, the establishment of performance evaluation indicator system of safety production supervision for local government must be closely around the regulation content. this study integrates performance, government performance and the government safety production supervision relevant research results both at home and abroad, combines with shandong province and tai’an city issued xxxx annual safety production, safety production target responsibility appraisal score detailed rules and other tai’an city government relevant documents, four first-level indicators to the performance measurement are determined, thus the first round of the evaluation indicator system is preliminarily determined. in addition, the government performance evaluation's goal is to complete an organic integration and equilibrium of government democratic administration and administrative efficiency, the government must pay attention to public participation, satisfaction with citizens and businesses and the opinions and suggestions should be reflected, and government behavior can be effectively supervised, to make the government in the process of resource management to achieve transparent and open. therefore, according to the actual situation of work safety regulation performance evaluation of the government, eligible persons, and the characteristics of public participation, under the condition of tai’an city government officials advising, secondary indicators are divided into the following three types of properties: (1) safety administration statistics confirm type: the assessment team conducts research, collects data for the head of the whole performance evaluation, then the head checks out the points; (2) assessment group evaluation type: different assessment groups are organized, do research and give a mark for different industries and departments, and the indicators are achieved by model calculation; (3) public comment type: enterprises and citizens are organized to fill in the questionnaire survey and give advice, to the end of the test, evaluation software will process the questionnaires to get the results of corresponding indicators. indicator system includes four first-level indicators: the construction of the supervision base, execution building, administrative efficiency and supervisory effect. (4) first-level indicators are divided into 40 secondary indicators. the design idea is: emphasis on micro aspect, on-site supervision function, highlighting the base and execution construction, attaching great importance to the administrative efficiency and the supervisory effect. the performance evaluation indicators of government safety production are shown in table 2. the comprehensive statistical results of indicators are shown in table 3. 3. design and determining weight of the evaluation indicator the performance evaluation indicator system of safety production supervision for local government includes two levels: first-level indicators and secondary indicators, in the process of design weights, the weight score of secondary indicators should be 100 points. the analytic hierarchy process (ahp) is chosen to determine the weights of first-level indicators. ahp method was put forward by american professor saaty, which is a practical method of making multiple criteria decision. ahp decomposes the complex problem to layers, forming a recursive class hierarchy; all the elements are compared in pairs to determine the relative importance of various factors in each hierarchy elements, and judgment matrix is built, the largest characteristic root and the corresponding eigenvector to the judgment matrix are calculated, then the factors’ weights are achieved, on the basis of this, combination weights of each factor are calculated. 3.1. the determination method of weight score for first-level indicator the process to determine the weights of first-level indicator according to the following steps to implement: a) four first-level indicators are compared in pairs; the design expert scoring tool table is shown in table 4. b) constructing the judgment matrix four first-level indicators are compared in pairs, and published by atlantis press copyright: the authors 88 h. yu / study on performance evaluation of government comprehensive supervision table 2. the performance evaluation indicator system of safety production supervision for local government first-level indicator secondary indicator indicator type a.the construction of the supervision base （19） a.1 organizational architecture safety administrative statistics confirm type a.2 special fund safety administrative statistics confirm type a.3 regulatory funds and equipment safety administrative statistics confirm type a.4 regulatory team construction safety administrative statistics confirm type a.5 performing duties assessment group evaluation type a.6 enterprise standardization safety administrative statistics confirm type a.7 occupational hazards declaration safety administrative statistics confirm type a.8 activities establishment situation safety administrative statistics confirm type b. executive force （53） b.1conducting annual activities situation safety administrative statistics confirm type b.2 hold meetings situation safety administrative statistics confirm type b.3 the responsibility system safety administrative statistics confirm type b.4 cadres performance evaluation safety administrative statistics confirm type b.5governance illegal rule violations assessment group evaluation type b.6 close enterprises situation assessment group evaluation type b.7major festivals and time arrangement safety administrative statistics confirm type b.8 special project rectification assessment group evaluation type b.9 work safety month safety administrative statistics confirm type b.10 safety culture propaganda safety administrative statistics confirm type b.11 employee training assessment group evaluation type b.12head,management personnel training assessment group evaluation type b.13emergency rescue command system safety administrative statistics confirm type b.14 contingency plan management safety administrative statistics confirm type b.15work safety emergency information construction and basic data updating safety administrative statistics confirm type b.16 law enforcement plan assessment group evaluation type b.17 dynamic regulation safety administrative statistics confirm type b.18supervising enterprises establishing rules assessment group evaluation type b.19supervising institutions and the staff setting of enterprises assessment group evaluation type b.20supervising enterprises safety costs assessment group evaluation type b.21yellow card warning safety administrative statistics confirm type c.administrative efficiency （8） c.1 work safety license assessment group evaluation type c.2 three simultaneous review assessment group evaluation type c.3hidden danger rectification operation mechanism assessment group evaluation type c.4 major hazards safety administrative statistics confirm type d. supervisory effect （20） d.1mortality per 100000 people in mining industry safety administrative statistics confirm type d.2mortality per one hundred million yuan safety administrative statistics confirm type d.3mortality per thousand vehicles safety administrative statistics confirm type d.4mortality per one million tons coal safety administrative statistics confirm type d.5listed supervisory completion safety administrative statistics confirm type d.6 public satisfaction public comment type d.7 enterprise satisfaction public comment type published by atlantis press copyright: the authors 89 h. yu / study on performance evaluation of government comprehensive supervision table 3. the comprehensive statistical results of performance evaluation indicator system of safety production supervision for local government table 4. contrast between four first-level indicators item the construction of the supervision base execution building administrative efficiency supervisory effect the construction of the supervision base execution building administrative efficiency supervisory effect the results are presented using 1~9 and reciprocal scaling comparison, 1~9 scales and its reciprocal are shown in table 5. tai’an government officials compare four first-level indicators in pairs according to table 5, and construct judgment matrix nnijaa ×= )( by statistical scores, as shown in table 6. table 5. ahp scaling method and its meaning scale 1 3 5 7 9 2,4,6,8 meaning the same important a little more important obviously important highly important absolutely important in the middle state of the adjacent judgment table 6. comparison results of indicator importance ratings a1 a2 … an a1 a11 a12 … a1n a2 a21 a22 … a2n … … … … … an an1 an2 … ann judgment matrix has the following properties: ① aij>0; ② aji=1/aij; ③ aii=1. c) evaluation the judgment matrix a is input to the matlab software, the largest eigenvalue of a and normalized feature vector are computed. d) consistency check the consistency indicator is calculated using the formula 1 max − − = n n ic λ , where n is the order number of the judgment matrix. the consistency ratio, r c c i i r = , among which, ri needs to select from table 7. the calculated rc should be below 0.1. indicator property first-level indicator a total of indicator attribute secondary indicator number safety administrative statistics confirm type assessment group evaluation type public comment type a.the construction of the supervision base 8 7 1 0 b.execution building 21 12 9 0 c.administrative efficiency 4 1 3 0 d. supervisory effect 7 5 0 2 total 40 25 13 2 published by atlantis press copyright: the authors 90 h. yu / study on performance evaluation of government comprehensive supervision table 7. the values table of mean random consistency indicator ir n 1 2 3 4 5 6 7 8 9 10 … ri 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.41 1.49 … when judgment matrix a meets the consistency, the corresponding eigenvector u is calculated. the vector is called weight vector after resulting from the standardization, and each component of weight vector is the relative importance of the first-level indicator, namely the weight coefficient. the weight vector of the relative importance of each component is the first-level indicator, namely the weight coefficient. the indicator weight coefficients obtained by analytic hierarchy process (ahp) are within the range of (0, 1), and the sum of the weight coefficient of every indicator is 1. for the convenience of calculation, weight coefficient would be expanded 100 times, and then the weight scores of each indicator are achieved. first-level indicator weights are acquired according to this method, as shown in table 8. table 8. design of evaluation indicator weights first-level indicator secondary indicator number weight average weight of secondary indicator the construction of the supervision base 8 19 2.375 execution building 21 53 2.524 administrative efficiency 4 8 2.000 supervisory effect 7 20 2.857 3.2. the determination method of weight score for secondary indicator the performance evaluation indicator system of safety production supervision for local government involves a lot of secondary indicators, it is difficult to make consistency check by using ahp, so delphi method is adopted to design the weights of secondary indicators. the detailed calculation steps are as follows: (a) divide the weigh scores of first-level indicators by the total number of the corresponding secondary indicators, and the average score of (b) each secondary indicator is calculated. (c) the retrieved effective questionnaires are designed in four important degree levels which are important, more important, and general and rejection, according to average score of the first step, four levels are set points respectively. when doing the questionnaires, experts give scores on the important degree. (d) the questionnaires are statistical computed, average experts’ score of each indicator is calculated, and then round to integer because the weight score must be an integer. (e) the weighs of indicators finally determined are tested, the sum of weight scores of secondary indicators included in the first-level indicator is equal to the first-level indicator weight, if it doesn’t meet the conditions, the corresponding adjustment is made. finally the performance evaluation indicator system and weight score of work safety regulation for local government are acquired as shown in table 9. 4. design of evaluation model administration of production safety supervision effectiveness evaluation system, a total of 100 points, among them, 19 points, executive ability construction execution 53 points, 8 points, administrative efficiency supervision effect is 20 points. according to the index of attributes, each relevant personnel or directly select points scored by the system statistics, or according to the evaluation criteria, in after the completion of the test statistics each secondary index score and is the evaluation scores. among them, xz for total score, xk as the first k index score： ∑ = = 40 1k kz xx published by atlantis press copyright: the authors 91 h. yu / study on performance evaluation of government comprehensive supervision table 9. the performance evaluation indicator system and weight score of work safety regulation first-level indicator secondary indicator weight score a. the construction of the supervision base （19） a.1 organizational architecture 2 a.2 special fund 2 a.3 regulatory funds and equipment 3 a.4 regulatory team construction 3 a.5 performing duties 3 a.6 enterprise standardization 3 a.7 occupational hazards declaration 2 a.8 activities establishment situation 1 b. execution building （53） b.1conducting annual activities situation 2 b.2 hold meetings situation 2 b.3 the responsibility system 4 b.4 cadres performance evaluation 2 b.5governance of illegal rule violations 3 b.6 close enterprises situation 3 b.7major festivals and time arrangement 3 b.8 special project rectification 3 b.9 work safety month 4 b.10 safety culture propaganda 2 b.11 employee training 2 b.12head,management personnel training 2 b.13emergency rescue command system 2 b.14 contingency plan management 2 b.15work safety emergency information construction and basic data updating 4 b.16 law enforcement plan 2 b.17 dynamic regulation 2 b.18supervising enterprises establishing rules 2 b.19 supervising institutions and the staff setting of enterprises 2 b.20 supervising enterprises safety costs 2 b.21 yellow card warning 3 c. administrative efficiency （8） c.1 work safety license 2 c.2 three simultaneous review 2 c.3hidden danger rectification operation mechanism 2 c.4 major hazards 2 d. supervisory effect （20） d.1mortality per 100000 people in mining industry 3 d.2 mortality per one hundred million yuan 3 d.3 mortality per thousand vehicles 3 d.4 mortality per one million tons coal 3 d.5 listed supervisory completion 3 d.6 public satisfaction 3 d.7 enterprise satisfaction 2 5. empirical research according to the above safety regulation performance evaluation index system and model, this study used questionnaire tai’an in shandong province city of production safety supervision department for the performance evaluation. assessment process includes: set up assessment team; to form a panel; determine the sampling personnel, training personnel; questionnaires; recycling questionnaire; the questionnaire on statistics and analysis. the poll gave out 78 copies of questionnaires, back to 65. test result is as the following table 9 and table 10. published by atlantis press copyright: the authors 92 h. yu / study on performance evaluation of government comprehensive supervision table 9. primary index scores first grade assessment indicator score full score safety supervision organs 16.8 19 construction of administrative capacity 29.8 53 executive efficiency 6.4 8 supervision efficiency 18.2 20 table 10. secondary index scores second level evaluation index score grade a．1 organizational architecture 3.00 excellent a．2 special fund 2.00 excellent a．3 funds and equipment 3.00 excellent a．4staff team-building 2.50 medium a．5due diligence system 2.00 good a．6 quality standard construction. 1.50 medium a．7occupational hazards report 1.00 medium a．8activities create 1.50 medium a．9conduct meetings, 1.00 poor a．10 accountability systems 3.00 good a．11crack down on speculation 2.00 medium a．12 close illegal enterprises 2.00 medium a．13 how to arrange major festivals and holidays 1.2 poor among them, system test total score 65 points, excellent index number to 10, good indicator number is 2, medium index number for 22, poor index number is 6. comprehensive assessment results for the medium. 6. conclusions performance evaluation of work safety regulation for local government can achieve the objective evaluation of the government safety regulation situation, the advantages and disadvantages of government work safety regulation can be found out, which can provide reference to government work safety supervision. it is conducive to improve work safety regulation performance of local government, and then enhance the overall operating efficiency of the government. this study constructs a performance evaluation system of work safety regulation for local government based on balanced score card method, designs a performance evaluation indicator system of work safety regulation for local government suitable for city and county level. the factors affecting work safety regulation performance of local government are analyzed systematically, according to the combination of qualitative and quantitative method, a multi-level and complex problem is made a systematic and hierarchical analysis. in practical application, the indicator system should be adapted to the changing of national policy and the government safety regulation level continuously. in addition, performance evaluation of work safety regulation for local government is meaningful with rewards and punishment; otherwise it can’t play an incentive role. references luo yun. the construction and optimization of work safety regulation mode in china. modern occupational safety, 2014(1): p36-38. luo yun, lixin, cui gang. the theory and method of government safety regulation performance evaluation. 2010 academic essays of china association of occupational safety and health: p16-25. chu dejiang. the present situation and foresight of the local government performance evaluation indicator system research. journal of beijing administrative college, 2008(1): p17-21. published by atlantis press copyright: the authors 93 h. yu / study on performance evaluation of government comprehensive supervision chen tianxiang, fu lin. research on the performance measurement of government's supervision on the coal mine safety production. journal of xiangtan university(philosophy and social sciences), 2009(5): p18-21. wang yonggang, lin zhixiao, chen daogang, zheng hongyun. research on the evaluation model of the performance of the civil aviation safety supervision and management bureau based on balance score card. safety and environmental engineering, 2013(4): p132-135. dong zhengliang, wang fangning, guo qiming etc. the construction of safety performance management system based on the balanced score card. industrial safety and environmental protection,2007(11): p57-59. shan guoqi,shen feng. empirical study on the performance evaluation system of enterprises based on balanced score card. science and technology management research, 2010(16): p243-246. sun yuqian. the method and system development for performance evaluation of government comprehensive supervision for safety production. china university of geosciences (beijing), 2015 published by atlantis press copyright: the authors 94 1. introduction 2. the framework of performance evaluation indicator system of work safety regulation for the local government 2.1. the strategy map of local government work safety regulation 2.2. performance evaluation indicator system of work safety regulation for the local government 3. design and determining weight of the evaluation indicator 3.1. the determination method of weight score for first-level indicator 3.2. the determination method of weight score for secondary indicator 4. design of evaluation model 5. empirical research 6. conclusions references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <feff005500740069006c0069006300650020006500730074006100200063006f006e0066006900670075007200610063006900f3006e0020007000610072006100200063007200650061007200200064006f00630075006d0065006e0074006f00730020005000440046002000640065002000410064006f0062006500200061006400650063007500610064006f00730020007000610072006100200069006d0070007200650073006900f3006e0020007000720065002d0065006400690074006f007200690061006c00200064006500200061006c00740061002000630061006c0069006400610064002e002000530065002000700075006500640065006e00200061006200720069007200200064006f00630075006d0065006e0074006f00730020005000440046002000630072006500610064006f007300200063006f006e0020004100630072006f006200610074002c002000410064006f00620065002000520065006100640065007200200035002e003000200079002000760065007200730069006f006e0065007300200070006f00730074006500720069006f007200650073002e> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <feff005500740069006c0069007a007a006100720065002000710075006500730074006500200069006d0070006f007300740061007a0069006f006e00690020007000650072002000630072006500610072006500200064006f00630075006d0065006e00740069002000410064006f00620065002000500044004600200070006900f900200061006400610074007400690020006100200075006e00610020007000720065007300740061006d0070006100200064006900200061006c007400610020007100750061006c0069007400e0002e0020004900200064006f00630075006d0065006e007400690020005000440046002000630072006500610074006900200070006f00730073006f006e006f0020006500730073006500720065002000610070006500720074006900200063006f006e0020004100630072006f00620061007400200065002000410064006f00620065002000520065006100640065007200200035002e003000200065002000760065007200730069006f006e006900200073007500630063006500730073006900760065002e> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <feff004e006100750064006f006b0069007400650020016100690075006f007300200070006100720061006d006500740072007500730020006e006f0072011700640061006d00690020006b0075007200740069002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b00750072006900650020006c0061006200690061007500730069006100690020007000720069007400610069006b007900740069002000610075006b01610074006f00730020006b006f006b007900620117007300200070006100720065006e006700740069006e00690061006d00200073007000610075007300640069006e0069006d00750069002e0020002000530075006b0075007200740069002000500044004600200064006f006b0075006d0065006e007400610069002000670061006c006900200062016b007400690020006100740069006400610072006f006d00690020004100630072006f006200610074002000690072002000410064006f00620065002000520065006100640065007200200035002e0030002000610072002000760117006c00650073006e0117006d00690073002000760065007200730069006a006f006d00690073002e> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <feff005400650020006e006100730074006100760069007400760065002000750070006f0072006100620069007400650020007a00610020007500730074007600610072006a0061006e006a006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b006900200073006f0020006e0061006a007000720069006d00650072006e0065006a016100690020007a00610020006b0061006b006f0076006f00730074006e006f0020007400690073006b0061006e006a00650020007300200070007200690070007200610076006f0020006e00610020007400690073006b002e00200020005500730074007600610072006a0065006e006500200064006f006b0075006d0065006e0074006500200050004400460020006a00650020006d006f0067006f010d00650020006f0064007000720065007400690020007a0020004100630072006f00620061007400200069006e002000410064006f00620065002000520065006100640065007200200035002e003000200069006e0020006e006f00760065006a01610069006d002e> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word research of city rainstorm waterlogging scene simulation -in daoli district of harbin city as an example received 3 september 2014 accepted 15 february 2015 research of city rainstorm waterlogging scene simulation -in daoli district of harbin city as an example peng chen1, jiquan zhang2, lingfeng zhang1, yingyue sun1 1. college of tourism and geographical sciences, jilin normal university, siping 1360003, china 2. school of environment, northeast normal university, changchun 130117, china     abstract as global warming urbanization process accelerated, urban waterlogging problem is increasingly serious, has become the research hotspot nowadays. city rainstorm waterlogging scene simulation is one of the important link in the research of the process in which residents shelter when city rainstorm waterlogging occurred. to develop urban rainstorm product scenario simulation study can guide us to determine the product elimination and influence range and location. this study using a one-dimensional and two-dimensional unsteady flow as the basic equation, with irregular grid as the basic frame, build urban rainstorm waterlogging deposition numerical simulation model, combining with the information diffusion theory calculation rainstorm probabilities in the study area, on this basis, set the study area rainstorm waterlogging situation, realized the different waterlogging scene simulation and visualization. the results show that the heavy rain that just sccumulate waterlogging disasters occurs once every five years in the study area, and depth of 0-0.47m,velocity between 0-0.43m/s.there is the biggest produce waterlogging which sccumulate with the heavy rain occurs once every 100 years, and the depth of 0-3m,velocity between 3-3.4m/s. keywords: harbin; rainstorm waterlogging disaster; scenario simulation 城市暴雨积涝情景模拟研究—以哈尔滨市道里区为例 陈鹏 1 ，张继权 2 ，张立峰 1 ，孙滢悦 1   1.  吉林师范大学旅游与地理科学学院，四平 136000, 中国 2.  东北师范大学环境学院，长春 130117, 中国 摘要：随着全球气候变暖、城市化进程加快，城市积涝问题日趋严重，已经成为当今研究热点。城市暴雨 积涝灾害居民避难迁安研究过程中，城市暴雨积涝情景模拟是其重要环节之一。开展城市暴雨积涝情景模 拟研究可以指导我们确定积涝影响范围及发生位置。本研究利用一、二维非恒定流为基本控制方程，以不 规则网格为基本骨架，构建城市暴雨积涝数值模拟模型，并对研究区降雨过程分析与强度计算，以此为基 础设定研究区暴雨积涝情景，实现了不同积涝情景模拟结果及可视化。研究结果表明当研究区发生 5 年一 遇的暴雨时区内刚好发生积涝灾害，且水深为 0-0.47m 之间，流速在 0-0.43m/s 之间；100 年一遇的暴雨时 区内积涝为最大，且水深在 0-3m 之间，流速在 0-3.4m/s 之间。 关键词：哈尔滨；暴雨积涝灾害；情景模拟 1. 引言 随着全球气候变暖和城市化进程不断加快，未来 城市暴雨积涝灾害发生频率将越来越高。从历史城市  暴雨积涝灾害统计可知，城市道路积水深度可达 1.5-3.0m，车辆、行人无法正常通行，地下设施被淹，                                                                 作者简介：陈鹏（1980‐），男，讲师，博士，从事自然灾害 风险评价与管理研究.e‐mail：pp11290@163.com  负一层住户生命受到严重威胁。从2008年到2013年全 国351个城市都发生了不同程度的积涝灾害，道路积水 深度在50cm以上的城市达到60%，积水超过半小时的 占到将近80%，其中城市积涝灾害较重的城市有大连、 北京、武汉、杭州、郑州、天津、哈尔滨等。因此， 开展城市暴雨积涝灾害情景模拟对城市居民避难迁安、 积涝灾害预防预警等具有重要意义[1]。近10多年来， journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 66-72 published by atlantis press copyright: the authors 66 p. chen et al. 从国际减灾战略到国内自然灾害预防预警等减灾工 作中，灾害的预防预警尤为重要；而预防预警的前 提就是灾害的情景模拟，同时也是城市预防自然灾 害，控制和降低自然灾害的重要基础性研究 [1] 。通 过搜集与整理现有城市暴雨积涝情景模拟研究发现， 对于积涝模型建立过程多数是以水力学方法、水动 力学方法构建，且模型多数未考虑地下排水管网或 是简化排水管网，在计算精度上有所欠缺，尤其是 对积水流速并未考虑 [2‐5] 。因此，本文针对城市地区 小尺度的暴雨积涝灾害为研究对象，结合哈尔滨市 道里区实证研究，提出了一套完整的城市暴雨积涝 情景模拟思路与方法，创建了城市暴雨积涝情景模 拟模型与范式，以充实、完善小尺度城市自然灾害 应急理论与方法，为我国制订城市暴雨积涝灾害应 急管理和规划提供依据。本文研究与当前国际上很 多大城市开展的基于小尺度的灾害应急管理趋势相 一致，也是确保大尺度城市灾害应急管理及评估精 度的前提和基础。  2. 研究区概况 2.1. 自然概况 研究区位于东经 125°42′―130°10′，北纬 44°04′―46°40′，黑龙江省省会，副省级城市。 市内总体地形为南高北低，西高东低。江南城区地 貌属于堆积和剥蚀堆积类型，内河切割松花江阶地 和漫滩地，形成了起伏变化的地貌。同时，区内由 于近些年超采地下水导致部分区域地面下降，道路 极易积水。市内多年平均降水量 545.7mm，流域内 持续时间长、雨量较大、影响范围广的大暴雨发生 时间多集中在 7、8 两个月份，出现次数占大暴雨总 数的 84-88%，可见 7、8 月份是暴雨频发期。汛期 6-9 月份降水量占全年降水量的 80%，其中 7、8 月 占 52%，此期间极易形成暴雨积涝灾害。 2.2. 研究区积涝灾害概况 近几年哈尔滨城市化进程较快，不论是新城区 还是老城区都存在着排水管道偏少，管渠断面过小， 截止今年，哈尔滨市排水管线共有 993 公里，排水 管网覆盖率只有 66%，还有 34%地区没有排水设施。 市内排水管网的排水能力为 117 m 3 /s，比中雨的排 水能力低 68m 3 。一遇中雨，每秒就会有 68m 3 雨水因 无法即排而停留在路面上，降雨超过半小时以上， 哈尔滨市一些街道就会积水。 2.3. 数据来源 本研究所需数据中气象资料来自黑龙江省气象 局多年降雨数据（1961-2009），空间数据采用高分 辨率快鸟影像（分辨率 0.61m），以及道里区排水管 网数据。所需城市下垫面基础信息从快鸟影像数据 中提取，其中包括道路信息、居民点信息、地面糙 率、面积等参数，所用数据均利用 arcgis 软件中的 空间分析工具进行分析和提取。 3. 城市暴雨积涝数值模拟 3.1. 研究区集水区确定 集水区是指将某一流域与另一流域分开的分水 岭，包括对一河流或湖泊供应水源的全部区域或地 区。而事实上因地质，地下水等诸多因素影响，无 法仅以简单的面积来确定，同时人为的改变，如城 市下垫面变为水泥路、柏油路等，其产汇流时间、 方向及下渗能力都有所改变。所以确定某一区域集 水区域时必须以数字高程模型（dem）为基准，利用 arcgis 软件中的水文分析功能模块生成水流流向 图。由于城市下垫面特点，本模型中对于水流形成 过程中只考虑水流受重力影响情况。根据 arcgis 软件提取结果把研究区分成 2645 个积水区域，并结 合道里区的行政边界所涉及到的各积水区域进行整 合。 3.2. 道里区降雨时程分析 一次降雨过程由于不同区域、不同降雨过程， 从而导致一次降雨过程的雨峰出现的时段不同。当 一次降雨过程中雨峰出现时，则是降雨量最多时刻， 此时刻对于城市积涝来说是最严重时段。所以，要 想了解城市暴雨积涝的形成过程，必须了解一次降 雨过程中的降雨时程分配。通过对研究区历史降雨 过程统计分析得出哈尔滨的降雨时程分配特点。依 据分析结果对研究区降雨过程作了五种雨型概化。 分别是雨峰 0.4 处、雨峰 0.6 处的雨峰靠前、雨峰 靠后、均匀雨峰、雨峰前置、雨峰后置（图 1）。 3.3. 排水分区及排水设施概化 1. 排水分区确定 排水分区是城市暴雨积涝模型中不可缺少的重 要组成部分，其中主要考虑排水地区的地形、水系、 水文地质、容泄区水位和行政区划等因素。其划分 published by atlantis press copyright: the authors 67 research of city rainstorm waterlogging scene simulation    原则：（1）排水分区划分应根据城市总体规划布局， 结合城市废水受纳体位置进行划分；（2）排污系统 需根据城市规划并且结合竖向规划及城市路网、污 水受纳体和污水处理厂分布位置进行合理布局；（3） 雨水系统应布局应根据城市已有规划及地形，结合 竖向规划和城市废水受纳体位置，按照流域划分及 布局；（4）雨污混排系统应综合雨水、污水、泵站、 污水系统布局的要求，并考虑截流干管（渠）和溢 流井位置进行合理布局。在哈尔滨城市排水规划中， 研究区分属于沿江排水分区与何家沟排水分区（图 2）。 图 1.研究区降雨时程分配 通过对哈尔滨市道里区以划分的排水分区分析 可知，已有的两个排水分区不能满足城市暴雨积涝 模型需要。现有的两个排水分区从模型精度上考虑 不能满足模型模拟的精度，为了提高模型模拟的精 度，对排水区进行了重新划分。结合上述排水分区 原则，根据管道走向以及排水口位置，概化多个排 水小区，以适应排水计算需要（图 3）。 图 2.道里区现状排水分区 图 3.道里区重分排水分区 2. 排水管网概化 城市中排水管网分布及其复杂、纵横交错，不 同城市排水管网建设进度也不一样。随着城市化进 程加快，排水管网建设远远不能满足城市发展需要。 一般的城市排水管网构成由排水管道、水井、明沟、 泵站等组成，其中雨水井承担接收地面来水。排水 管道中积水是由排水泵站、闸或淹没出流管道来实 现，城市的一级、二级河道、明沟和坑塘、湖泊接 纳泵站、闸门、或淹没出流管道的来水 [6] 。研究区 的城市排水图来自于哈尔滨城市排水总体规划（cad 格式）。通过 gis 的数据处理功能，对其格式、坐标 投影进行转换。处理成为与其它数据相同的格式 （shape 格式）。由于城市排水规划图只有干管信息， 所以本研究只考虑干管排水概化。 2. 城市下垫面概化与网格划分 城市下垫面地物概化过程非常复杂，对于连续 性阻水建筑物（如立铁路、房屋、交桥、堤防、高 于地面的街道）在概化时需进行特殊考虑。本研究 中并没有设置节点，同时对于城内的二级河流，也 并未概化成为特殊通道，而是按照其实际宽度概化 成尺度比较小的网格，分别在网格和通道属性中注 明 [7-9] 。其它设施如阻水建筑、排水设施等，也分别 在网格和通道属性中注明（图 4-图 5）。由于网格与 通道都是实体，都可以储存数据，并已建立拓扑关 系，研究区共划分网格 1724 个，通道 3572 条。在 对研究区下垫面概化时结合网格划分特点，将研究 区概化为含排水管网的不规则网格、含阻水建筑物 不规则网格、不含阻水建筑物不规则网格等。 图 4.  研究区下垫面概化 图 5.研究区网格和通道及其属性 published by atlantis press copyright: the authors 68 p. chen et al. 3.4. 城市暴雨积涝模型构建 1. 模型控制方程 目前国内外针对城市道路积水的数值模拟方法 较多，但大多数只是模拟出水深，对于流速、流向 并未计算 [10-13] 。当积涝灾害发生时，道路积水、流 速对居民水中行走都会产生影响，因此，需考虑水 深对居民出行影响程度，又兼顾考虑水流速度对居 民出行影响程度。研究中城市暴雨积涝数值模型以 二维非恒定流为基本骨架，以不规则网格概化地物， 同时结合排水管网中一维非恒定流综合考虑建立暴 雨积涝数值模型。具体由四个大的模块构成，分别 为产流模型、汇流模型、排水模型以及积涝模型（图 6）。具体公式如下： 二维非恒定流方程： 连续方程： 0 （1） 动量方程： √ 0 （2） √ 0 （3） 式中：h 为水深；h 为水位（h=h+z，z 为下垫 面高程）；q 为源汇项，在模型中代表有效降雨强度； m、n 分别为 x、y 方向上的单宽流量；u、v 分别为 流速在 x、y 方向的分量；n 为糙率系数；g 为重力 加速度；t 为水流在网格内流动时间。 一维非恒定流基本控制方程： （4） 式中：q 为截面流量；a 为计算断面的过水面积；sf 为摩阻坡降；t 为积水到达通道时间；l 为网格通道 长度。 宽顶堰溢流公式： 2 / （5） 式中：qj 为堰顶单宽流量；m 为宽顶堰溢流系数； qs 为淹没系数；hj 为堰顶水位。 流速计算公式： √ / （6） 式中：v 为水流速度；m、n 分别为 x、y 方向上的单 宽流量；h 为积水深度。 2. 城市道路积水流向提取 利用 arcgis9.3 软件的水文分析功能利用 dem 提取地表水流径流水流方向、汇流累积量、水流长 度、河流网络（包括河流网络的分级等）以及对研 究区的流域进行分割等。利用提取的基本水文因子， 结合 dem 可以再现水流的流动过程，最终完成水文 分析过程，并提取研究区各网格内水流方向(图 7)。 图 6.城市暴雨积涝模型构建流程 图 7.研究区水流方向 3. 积涝数值模拟模型构建 城市暴雨积涝数值模拟模型构建方法多数是利 用集水区的积水量，采用基于 gis 的“等体积法”， 模拟城市道路积水区域和淹没深度 [15、16] 。本模型在 吸取有限差分法和有限体积法的优点，在网格周边 计算流量，在网格形心处计算水位 [17-18] ，以提高模 型计算的精度。研究区不规则网格绘制过程是利用 arcgis 软件实现的，绘制后的不规则网格进行属性 赋值，主要包括高程、初始水深、不规则网格编号、 通道编号、下垫面糙率等。计算过程中以二维非恒 published by atlantis press copyright: the authors 69 research of city rainstorm waterlogging scene simulation    定流控制方程为基本方程，水流方向按照 x、y 方向 进行分解（图 8），通道作为各个网格水流交换的媒 介，网格内水深为下垫面积水深度。  综合以上城市暴雨积涝数值模拟过程，利用 arcgis 空间数据管理功能集成管理研究需要的空 间信息以及属性信息，并对研究区域进行水文分析、 城市现状分析。在此基础上划分网格、生成通道， 并通过空间关联，对网格和通道建立拓扑关系，通 过空间分析和统计将在模拟所需参数以属性方式分 别赋给通道和网格。通过格式转换，生成文本文件 提供给计算模块，计算模块以 vc++编程开发，计算 结果生成文本文件，通过属性关联提供给 arcgis 网格和通道图层，使得计算结果可以可视化（图 9）。   图 8.不规则网格模型 图 9.  城市暴雨积涝数值模拟可视化结果  4. 情景设定 情景模拟是通过人为假定的若干种情景，模拟 再现整个研究区内承灾体危险性变化情况，以便更 好地预测未来灾情情况。情景模拟能够获得关于被 测对象更加全面的信息，对未来的预测、预警、应 急等至关重要。其情景设定应具有针对性、直接性、 可信性。由于情景模拟与现实较接近，其结果可为 政府部门解决实际问题提供直观、可靠的数据信息。 情景模拟法研究过程中多数采用研究对象相关数据 为基础，提取相关参数，并以历史灾情数据分析结 果为基础设定不同情景，同时利用数学方法构建模 拟模型。 4.1. 研究区暴雨现状与强度计算 4. 哈尔滨市暴雨现状 根据国家气象局对暴雨的界定可以看出，日 降水量 50 即为暴雨；日降水量 100 为大 暴雨；日降水量 250 即为特大暴雨。市内全年 平均降水量 569.1 毫米，降水主要集中在 6-9 月， 夏季占全年降水量的 60%。四季分明，冬季 1 月平 均气温约零下 19 度；夏季 7 月的平均气温约 23 度。 哈尔滨市城区地势起伏，极易造成城市积涝，严重 影响城市交通和居民生活。依据哈尔滨市各个气象 台站 46 年的统计资料，哈尔滨市区每年发生 1.27 次暴雨，具体哈尔滨市暴雨统计如图 10。   图 10.哈尔滨市各月暴雨直方图   5. 城市降雨强度计算 降雨是积涝灾害的致灾源，同时也是本研究中 积涝数值模型中的主要输入参数。通过对 1961‐2010 年的历史降雨数据分析，并结合哈尔滨市降雨模型 中的暴雨强度公式，计算出哈尔滨市道里区降雨过 程中的任意时段内的平均降雨强度，作为积涝数值 模型中的参数输入。表达式为：  .                               (7)  式中：q 为暴雨强度；p 为暴雨重现期；t 为降水历 时。  4.2. 情景分析及设定 依据国家规定，城市排水管网的即排能力应为 每秒 185 立方米，这个排水能力可抵御中雨，即每 小时 25mm 的降水量。可哈尔滨市排水管网的即排能 力为每秒 117m 3 ，比中雨的即排能力每秒低 68m 3 。也 就是说如果哈尔滨降中雨的话，每秒就会有 68 m 3 的雨水因无法即排而停留在路面上；降雨超过半小 published by atlantis press copyright: the authors 70 p. chen et al. 时以上，哈尔滨市的一些街道就会积水。依据情景 设定原则，情景模拟式通过人为假定一系列情景， 模拟研究区整个评价区承灾体危险性变化情况，以 便预测将来灾情局势。但对本研究区，根据历史灾 情统计与分析结果得知：5 年一遇暴雨产生积涝概 率较低；100 年一遇暴雨发生概率较低，一旦发生 100 年一遇暴雨，则对研究区居民生命、财产安全 影响较大。所以，本文暴雨积涝情景设定为 5 年、 100 年一遇的暴雨概率所导致的积涝情景（图 11-12 图）。 图 11.    5 年、100 年一遇暴雨道路积水情景 图 12. 5 年、100 年一遇暴雨道路积水流速 5. 结论 本文以城市频发的暴雨积涝灾害为研究对象， 结合哈尔滨道里区为实证研究，提出了一套基于信 息扩散理论的小尺度的城市暴雨积涝数值模型构建 与情景模拟思路与方法。研究结果如下：  （1）以 gis 技术为基础数据平台，以一、二维 非恒定流为模型基本控制方程，以不规则网格为基 本骨架，构建了基于“三层空间（空中降雨‐下垫面 ‐排水管网）”的城市暴雨积涝数值模拟模型；  （2）以研究区暴雨强度计算结果为基础，作为 积涝数值模型的初始输入参数，并结合历史暴雨资 料分析结果发现：5 年一遇暴雨发生概率较高；100 年一遇暴雨发生概率较低，但 100 年一遇暴雨发生 对研究区影响较大； （3）依据上述两方面研究结果，结合 gis 技术 实现了不同情景下暴雨积涝情景模拟。模拟结果表 明当研究区发生 5 年一遇的暴雨时区内刚好发生积 涝灾害，且水深为 0‐0.47m 之间，流速在 0‐0.43m/s 之间；100 年一遇的暴雨时区内积涝为最大，且水 深在 0‐3m 之间，流速在 0‐3.4m/s 之间。  由于城市暴雨积涝灾害系统是个复杂系统，受 多变量因素所制约，导致在数值模型构建及模拟过 程中精度问题会随着地形变化而变化，因此，本文 采用的城市暴雨积涝数值模型构建过程与方法在哈 尔滨市道里区已得到验证，但移植到其他城市区域 需进行模型参数重新校正。  致谢 本项研究得到了下列基金项目的资助：国家自然 科学基金(41371495)；国家科技支撑计划课题(2013b ak05b01)；国家科技支撑计划课题(2013bak 05b02)； 吉林省科技厅项目(20150520081jh)；吉林省自然科学 基金项目(201215224)；四平市社会科学基金项目的编 号(201118)。 参考文献 [1] chen peng, zhang jiquan, jiang xinyu, liu xingpeng, baoyulong, sun yingyue. scenario simulation-based assessment of trip difficulty for urban residents under rainstorm waterlogging. international journal of environmental research and public health. 2012, 9(6): 2057-2074. [2] 朱强,孙敏,陈秀万,等.基于动态数字地形模型的流域产 汇流模拟方法术.北京大学学报(自然科学版), 2009, 45(5): 805-810. [3] 张萍萍,王丽,张宁.武汉市城市积涝预警系统及其仿真 模拟效果. 暴雨灾害, 2010, 29(1): 71-75. [4] 仇劲文,李娜,程晓陶,等.天津市城区暴雨沥涝仿真模拟 系统. 水利学报, 2000 (11), 34-42. [5] 解以扬,李人鸣,李培彦.城市暴雨内涝数学模型的研究 与应用. 水科学进展, 2005, 16(3) :384-390. [6] 尹占娥,许世远,殷杰,等.基于小尺度的城市暴雨内涝灾 害 情 景 模 拟 与 风 险 评 估 . 地 理 学 报 , 2010, 65(5):554-557. [7] theo g s, martin t, norman e. analysis and modeling of fooding in urban drainage systems. journal of hydrology, 2004, 299(3): 300-311. published by atlantis press copyright: the authors 71 research of city rainstorm waterlogging scene simulation    [8] black j, endreny t a. increasing stormwater outfall duration, magnitude, and volume through combined sewer separation. journal of hydrologic engineering, 2006,11(5):472-481. [9] zaghloul n a. flow simulation in circular pipes with variable roughness using swmm extran mode. journal of hydrographic engineering, 1998, 124(1):73-76. [10] 李娜,仇劲卫,程晓陶,等.天津市城区暴雨沥涝仿真模拟 系统的研究.自然灾灾害学报,2002.11(2):113-118. [11] 解以扬,李大鸣,沈树勤,等.“030704”南京市特大暴雨 内 涝 灾 害 的 仿 真 模 拟 . 长 江 科 学 学 院 报,2004,21(6):73-76. [12] 尹占娥,许世远,殷杰,等.基于小尺度的城市暴雨内涝灾 害情景模拟与风险评估.地理学报,2010,65(5):554-557. [13] 景垠娜,尹占娥,殷杰,等.基于 gis 的上海浦东新区暴雨 内涝灾害危险性分析.灾害学,2010,25(2):58-63. [14] 刘仁义,刘南.基于 gis 的复杂地形洪水淹没区计算方 法.地理学报,2001,56(1):1.6． [15] 刘仁义,刘南.基于 gis 的复杂地形洪水淹没区计算方 法.地理学报,2001,56(1):1.6． [16] 郭利华,龙毅．基于 dem 的洪水淹没分析.测绘通 报,2002,5(11):25.30． [17] 权瑞松,刘敏,等.基于简化内涝模型的上海城区内涝危 险性评价.人民长江,2010,41(2)：33-34. [18] 程晓陶,杨磊,陈喜军.分蓄洪区洪水演进模型.自然灾害 学报,1996,5(1):34-40. [19] 黄崇福.模糊信息优化处理技术及其应用].北京：北京 航空航天大学出版社,1995.     published by atlantis press copyright: the authors 72 the study of estimation on maize irrigation water requirement in the northwest of liaoning province xiaojing liu1* donglai ma2 jiquan zhang3 peng chen1 1. college of tourism and geographic science, jilin normal university, siping 136000, p.r. china 2. university of tsukuba graduate school of system and information engineering, tsukuba 3050006, japan 3. environment college, northeast normal university, changchun 130000, p.r. china received november 14, 2018 accepted january 3, 2019 abstract this paper took the northwest of liaoning province as a study area and maize as the research object. effective rainfall, crop water requirement and soil available water content at 6 growth stages of maize were calculated by using penmanmontes equation, effective coefficient method and cropland soil moisture index. then the estimation model of irrigation water requirement of maize at different growth stages of the northwest of liaoning province. taking 2006 as an example, the irrigation water requirement of maize at different growth stages was estimated and the spatial distribution maps were drawn by gis. the results show that overall the irrigation water requirement of maize in the western of the study area is more than that of in the northeastern. the irrigation water requirement of maize in heading stage has increased significantly because of the severe drought status. it can provide technical support for a region to relieve crop drought. keywords: maize, drought, irrigation water requirement, csmi 辽西北地区玉米灌溉需水量估算研究 刘晓静 1*，马东来 2，张继权 3，陈鹏 1 1. 吉林师范大学旅游与地理科学学院，四平 136000，中国 2. university of tsukuba graduate school of system and information engineering，tsukuba 3050006, japan 3. 东北师范大学环境学院，长春 130024，中国 摘要：本文以辽西北地区为研究区，以玉米为研究对象，通过彭曼—蒙特斯方程、有效利用系数 法和农田浅层土壤指数 csmi，利用玉米各生育阶段的有效降雨量、作物需水量和土壤有效水含 量，构建了玉米不同生育阶段的灌溉需水量模型。以 2006 年为例，计算了研究区玉米各生育阶 段的灌溉需水量，并借助 gis 技术绘制玉米各生育阶段的灌溉需水量空间分布图。结果表明，整 体来看，研究区西部玉米的灌溉需水量高于东北部的玉米；抽穗—乳熟期，辽西北各区域玉米灌 溉需水量显著增多，这除了与玉米在该阶段自身生长需大量水分有关外，还与该时期严重的干旱 状况有关。研究结果可为缓解农业作物干旱提供技术支撑。 * 通讯作者：刘晓静(1985-)，女，讲师，博士，主要从事自然灾害风险评价与管理方面的教学和科研工作。email： liuxj@jlnu.edu.cn 基金项目：国家自然科学基金项目“辽西北灌区玉米干旱灾害动态风险预警与玉米灌溉调控模拟研究（41501559）”、国 家自然科学基金项目“多灾种农业气象灾害综合风险动态评估研究——以吉林省中西部玉米产区为例（41571491）”共同 资助。 关键词：玉米；干旱；灌溉需水量；csmi 干旱灾害影响最直接的是农业。中国是一个幅员 辽阔、地形地貌多样的农业大国，因此也成为世界上 受干旱灾害影响最为严重、面积最广的国家之一。对 于旱作地区来说，农作物的生长发育主要依赖于自 然降水。为了减轻干旱对旱作农业造成的影响和危 害，保证农作物的正常生长，需要对农作物进行灌 溉[1-3]。因此，作物灌溉需水量的估算是防止作物干 旱，缓解作物干旱状况，保障粮食生产安全最直接、 有效的手段。 作物灌溉需水量主要由作物需水量、有效降雨量 及土壤有效水含量三个要素共同决定。基于此，众多 研究以全生育期为时间尺度计算了作物的灌溉需水量 [4-6,8]，但是这些研究并未考虑作物不同生育阶段对水 分的不同需求。因此，本文以辽西北地区为研究区，以 玉米为研究对象，以玉米各生育阶段为时间尺度，结 合农田浅层土壤湿度指数 csmi，构建辽西北地区玉米 各生育阶段的灌溉需水量估算模型，估算辽西北地区 玉米不同生育阶段的灌溉需水量，为区域快速制定作 物防旱抗旱决策提供技术支撑。 1 研究材料与方法 1.1 研究区概况及资料选取 辽西北地区地处辽西丘陵山地，是辽宁省玉米重 要产地，玉米播种面积占到了全省的 67%。同时，该 区域也是辽宁省干旱最频繁、严重的地区，素有“十年 九旱”之称。受大陆性季风气候影响，辽西北地区春旱 现象时常发生，伏旱对产量影响较重，如 2006 年辽西 北地区发生了自 1951 年以来特大伏旱，粮食减产 4.07109kg。辽西北地区主要的玉米品种为晚熟类型 [7-8]，根据研究区各农业气象站提供的多年玉米发育 期资料，参考联合国粮农组织 fao 划分作物发育期的 建议，将辽西北地区玉米生育期划分为：播种—出苗、 出苗—七叶、七叶—拔节、拔节—抽穗、抽穗—乳熟、 乳熟—成熟 6 个阶段。 研究数据包括辽西北地区的遥感数据、气象数据、 农作物生长数据、土壤含水量及特征数据、土地利用 类型图、水利工程与灌溉设施数据、地形图等。 1.2 研究方法 辽西北地区作物根系多数集中在 50cm 以上土层 中。农田浅层土壤湿度指数（cropland soil moisture index，csmi）是由归一化植被指数 ndvi 和地表含水 量指数 swci 共同组成的，能够对 0~50cm 土壤墒情 进行准确监测的植被指数[9-10]。csmi 的计算公式如下： ndvi swci csmi ndvi swci − = + （1） 作者在前期工作中利用 csmi 反演了辽西北玉米 生长季 0~50cm 深度土壤含水量，并对反演模型进行 了验证，详细的计算方法及验证过程请参见文献[11]， 反演结果如表 1 所示。农田浅层土壤湿度指数 csmi 与玉米不同生长阶段的土壤湿度之间存在正、负两种 相关关系，为了便于分析，对 csmi 指数进行标准化 处理，公式如下： minmax min' xx xx x ii − − = （2） minmax max' xx xx x ii − − = （3） 式中 x'i 为指标 xi 标准化后的值，xmax、xmin 分别为指标 xi 的最大值和最小值。对与玉米拔节后土壤湿度呈正 相关的 csmi 指数采用公式（2）进行标准化，将其标 准化为与该阶段土壤湿度呈负相关；为了保持量纲的 一致性，对与玉米播种-拔节期间土壤湿度呈负相关的 csmi 指数则采用公式（3）进行标准化，使其二者关 系仍然保持负相关。 表 1 csmi 与实测 0~50cm 平均土壤湿度线性回归方程 生育期 线性回归方程 播种—出苗 0.1476 0.2554y x= − + 出苗—七叶 0.1876 0.2795y x= − + 七叶—拔节 0.1985 0.2736y x= − + 拔节—抽穗 0.219 0.0678y x= + 抽穗—乳熟 0.2642 0.0374y x= + 乳熟—成熟 0.2328 0.0443y x= + 1.3 玉米生育阶段灌溉需水量估算模型的建立 作物一生所需的水分有两种，生理需水和生态需 水。生理需水是作物为满足其自身正常的生命活动需 要从环境中吸收的大量水分，是指在健康无病、养分 充足、土壤水分状况最佳等条件下，作物经过正常生 长发育，从播种到成熟收获所需要的水分。生态需水 是指作物在维持自身正常生长时环境所需要的水分， 包括棵间蒸发和渗漏等。作物的生理需水量和生态需 水量的总和称为作物需水量。而在作物生长发育过程 52 journal of risk analysis and crisis response vol. 9(1), march (2019), pp. 52–59 doi: https://doi.org/10.2991/jracr.b.190328.006; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). the study of estimation on maize irrigation water requirement in the northwest of liaoning province xiaojing liu1* donglai ma2 jiquan zhang3 peng chen1 1. college of tourism and geographic science, jilin normal university, siping 136000, p.r. china 2. university of tsukuba graduate school of system and information engineering, tsukuba 3050006, japan 3. environment college, northeast normal university, changchun 130000, p.r. china received november 14, 2018 accepted january 3, 2019 abstract this paper took the northwest of liaoning province as a study area and maize as the research object. effective rainfall, crop water requirement and soil available water content at 6 growth stages of maize were calculated by using penmanmontes equation, effective coefficient method and cropland soil moisture index. then the estimation model of irrigation water requirement of maize at different growth stages of the northwest of liaoning province. taking 2006 as an example, the irrigation water requirement of maize at different growth stages was estimated and the spatial distribution maps were drawn by gis. the results show that overall the irrigation water requirement of maize in the western of the study area is more than that of in the northeastern. the irrigation water requirement of maize in heading stage has increased significantly because of the severe drought status. it can provide technical support for a region to relieve crop drought. keywords: maize, drought, irrigation water requirement, csmi 辽西北地区玉米灌溉需水量估算研究 刘晓静 1*，马东来 2，张继权 3，陈鹏 1 1. 吉林师范大学旅游与地理科学学院，四平 136000，中国 2. university of tsukuba graduate school of system and information engineering，tsukuba 3050006, japan 3. 东北师范大学环境学院，长春 130024，中国 摘要：本文以辽西北地区为研究区，以玉米为研究对象，通过彭曼—蒙特斯方程、有效利用系数 法和农田浅层土壤指数 csmi，利用玉米各生育阶段的有效降雨量、作物需水量和土壤有效水含 量，构建了玉米不同生育阶段的灌溉需水量模型。以 2006 年为例，计算了研究区玉米各生育阶 段的灌溉需水量，并借助 gis 技术绘制玉米各生育阶段的灌溉需水量空间分布图。结果表明，整 体来看，研究区西部玉米的灌溉需水量高于东北部的玉米；抽穗—乳熟期，辽西北各区域玉米灌 溉需水量显著增多，这除了与玉米在该阶段自身生长需大量水分有关外，还与该时期严重的干旱 状况有关。研究结果可为缓解农业作物干旱提供技术支撑。 * 通讯作者：刘晓静(1985-)，女，讲师，博士，主要从事自然灾害风险评价与管理方面的教学和科研工作。email： liuxj@jlnu.edu.cn 基金项目：国家自然科学基金项目“辽西北灌区玉米干旱灾害动态风险预警与玉米灌溉调控模拟研究（41501559）”、国 家自然科学基金项目“多灾种农业气象灾害综合风险动态评估研究——以吉林省中西部玉米产区为例（41571491）”共同 资助。 关键词：玉米；干旱；灌溉需水量；csmi 干旱灾害影响最直接的是农业。中国是一个幅员 辽阔、地形地貌多样的农业大国，因此也成为世界上 受干旱灾害影响最为严重、面积最广的国家之一。对 于旱作地区来说，农作物的生长发育主要依赖于自 然降水。为了减轻干旱对旱作农业造成的影响和危 害，保证农作物的正常生长，需要对农作物进行灌 溉[1-3]。因此，作物灌溉需水量的估算是防止作物干 旱，缓解作物干旱状况，保障粮食生产安全最直接、 有效的手段。 作物灌溉需水量主要由作物需水量、有效降雨量 及土壤有效水含量三个要素共同决定。基于此，众多 研究以全生育期为时间尺度计算了作物的灌溉需水量 [4-6,8]，但是这些研究并未考虑作物不同生育阶段对水 分的不同需求。因此，本文以辽西北地区为研究区，以 玉米为研究对象，以玉米各生育阶段为时间尺度，结 合农田浅层土壤湿度指数 csmi，构建辽西北地区玉米 各生育阶段的灌溉需水量估算模型，估算辽西北地区 玉米不同生育阶段的灌溉需水量，为区域快速制定作 物防旱抗旱决策提供技术支撑。 1 研究材料与方法 1.1 研究区概况及资料选取 辽西北地区地处辽西丘陵山地，是辽宁省玉米重 要产地，玉米播种面积占到了全省的 67%。同时，该 区域也是辽宁省干旱最频繁、严重的地区，素有“十年 九旱”之称。受大陆性季风气候影响，辽西北地区春旱 现象时常发生，伏旱对产量影响较重，如 2006 年辽西 北地区发生了自 1951 年以来特大伏旱，粮食减产 4.07109kg。辽西北地区主要的玉米品种为晚熟类型 [7-8]，根据研究区各农业气象站提供的多年玉米发育 期资料，参考联合国粮农组织 fao 划分作物发育期的 建议，将辽西北地区玉米生育期划分为：播种—出苗、 出苗—七叶、七叶—拔节、拔节—抽穗、抽穗—乳熟、 乳熟—成熟 6 个阶段。 研究数据包括辽西北地区的遥感数据、气象数据、 农作物生长数据、土壤含水量及特征数据、土地利用 类型图、水利工程与灌溉设施数据、地形图等。 1.2 研究方法 辽西北地区作物根系多数集中在 50cm 以上土层 中。农田浅层土壤湿度指数（cropland soil moisture index，csmi）是由归一化植被指数 ndvi 和地表含水 量指数 swci 共同组成的，能够对 0~50cm 土壤墒情 进行准确监测的植被指数[9-10]。csmi 的计算公式如下： ndvi swci csmi ndvi swci − = + （1） 作者在前期工作中利用 csmi 反演了辽西北玉米 生长季 0~50cm 深度土壤含水量，并对反演模型进行 了验证，详细的计算方法及验证过程请参见文献[11]， 反演结果如表 1 所示。农田浅层土壤湿度指数 csmi 与玉米不同生长阶段的土壤湿度之间存在正、负两种 相关关系，为了便于分析，对 csmi 指数进行标准化 处理，公式如下： minmax min' xx xx x ii − − = （2） minmax max' xx xx x ii − − = （3） 式中 x'i 为指标 xi 标准化后的值，xmax、xmin 分别为指标 xi 的最大值和最小值。对与玉米拔节后土壤湿度呈正 相关的 csmi 指数采用公式（2）进行标准化，将其标 准化为与该阶段土壤湿度呈负相关；为了保持量纲的 一致性，对与玉米播种-拔节期间土壤湿度呈负相关的 csmi 指数则采用公式（3）进行标准化，使其二者关 系仍然保持负相关。 表 1 csmi 与实测 0~50cm 平均土壤湿度线性回归方程 生育期 线性回归方程 播种—出苗 0.1476 0.2554y x= − + 出苗—七叶 0.1876 0.2795y x= − + 七叶—拔节 0.1985 0.2736y x= − + 拔节—抽穗 0.219 0.0678y x= + 抽穗—乳熟 0.2642 0.0374y x= + 乳熟—成熟 0.2328 0.0443y x= + 1.3 玉米生育阶段灌溉需水量估算模型的建立 作物一生所需的水分有两种，生理需水和生态需 水。生理需水是作物为满足其自身正常的生命活动需 要从环境中吸收的大量水分，是指在健康无病、养分 充足、土壤水分状况最佳等条件下，作物经过正常生 长发育，从播种到成熟收获所需要的水分。生态需水 是指作物在维持自身正常生长时环境所需要的水分， 包括棵间蒸发和渗漏等。作物的生理需水量和生态需 水量的总和称为作物需水量。而在作物生长发育过程 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 52 53 中，水分数量特定的情况下，作物实际吸收到的水分 数量称为作物实际耗水量。当任一时段外部环境提供 的水量（有效降雨量与土壤有效水含量之和）大于等 于本时段作物的需水量时，则在该时段，作物实际耗 水量等于作物需水量；当任一时段外部环境提供的水 量（有效降雨量与土壤有效水含量之和）小于本时段 作物的需水量时，则在该时段，作物实际耗水量等于 降雨量与土壤有效水含量之和，作物需水量与作物实 际耗水量之间的差值即为作物灌溉需水量[12]。 在旱生作物的整个生育期期间，任一时段土壤计 划湿润层中需灌水量，取决于该时段的来水量和作物 需水量的多少。依据上述原理，构建辽西北地区玉米 各生育阶段的灌溉需水量模型，公式如下： ( )i ci i i iir et p we i= − + + （4） 其中，iri 为辽西北地区玉米在生育阶段 i 的灌溉需水 量，mm；etci 为该地区玉米 i 阶段需水量，mm；pi 为 该地区生育阶段 i 内的有效降雨量，mm；wei 为本生 育阶段土壤有效水含量，mm；ii 生育阶段 i 的计划灌 溉量，mm。 根据作物生长水分的来源，辽西北地区的农业生 产方式可以分为两种，灌溉型农业和雨养型农业。研 究以雨养玉米为研究对象。对于雨养型农业来说，作 物的生长发育完全依靠自然降水，因此灌溉量 i 为 0， 此时玉米生育阶段的灌溉需水量模型为： ( )i ci i iir et p we= − + （5） 当区域外部来水量大于玉米需水量时，则认为该 区域不需要灌溉；当区域外部环境无法为玉米提供充 足的需水量时，则此时玉米的缺水量即为区域需灌溉 的水量。因此，玉米在某一生育阶段的灌溉需水量应 为： 0 ( ) -( ) >( ) ci i i i ci i i ci i i et p we ir et p we et p we   +  =   + + ， ， （6） 1.4 玉米生育阶段灌溉需水量估算模型参数的确 定 1.4.1 生育阶段有效降雨量 一次降雨的总雨量分配主要有三个方式：深层渗 漏、径流和留存于土壤中。雨水中的一部分渗漏到作 物根区吸水层以下，一部分作为径流从土壤表面流走， 地表的径流水和深处的渗漏水都不能被作物利用，换 言之，这两部分的降雨量对于作物是无效的。剩余的 降雨量贮存在土壤根区中，并为作物生长提供水分， 这部分可被作物吸收利用的降雨量称为有效降雨量[13]。 目前，多采用降雨有效利用系数计算有效降雨量， 公式如（7）所示： e j j p p=  （7） 式中，pj 为某次降雨总量，mm；pe 为该次降雨的有效 降雨量，mm；αj 为该次降雨量的有效利用系数。依据 康绍忠，蔡焕杰（1996）的研究结果，α 的取值如公式 [14]： 0 5 0.9 5 50 0.75 50 j j j j p mm mm p mm p mm    =     ， ， ， （8） 通过累积逐日有效降雨量得到玉米各生育阶段的有效 降雨量，公式如（9）所示： , 1 j i n ei e j p p = =  （9） 其中，pei 为生育阶段 i 的有效降雨量，mm；i=1,2,…6, 分别表示玉米的 6 个生育阶段；pej,i 为生育阶段 i 内第 j 次降雨的有效降雨量，mm；j=1,2,…,n，表示该生育 阶段内降雨次数。 1.4.2 生育阶段需水量的计算 作物需水量的影响因素有很多，包括气象条件 （温度、日照、湿度、风速）、土壤水分含量、作物种 类及其生长发育阶段、土壤肥力、农业技术措施等。目 前，计算作物需水量最常用的方法是通过计算参照作 物的需水量来得到某种作物的需水量。研究表明，联 合国粮农组织 fao 推荐的单作物系数法在预测东北地 区作物需水量时效果最好[15]。因此，本文采用单作物 系数法计算作物的逐日蒸散量，即作物逐日需水量， 则作物生育阶段的需水量由其逐日需水量累积得到。 作物逐日需水量的计算公式如下： 0c cet k et= （10） 式中，etc 为逐日作物蒸散量，即逐日作物需水量，mm； kc 为作物系数；et0 为逐日作物参考蒸散量，mm。kc、 et0 分别采用联合国粮农组织 fao 推荐的修正公式和 彭曼蒙特斯公式进行计算[16,17]。 1.4.3 土壤有效水含量的计算 土壤中的水分并非所有都对作物有效。例如，土 壤吸湿水和膜状水就属于无效水，由于它们被土壤颗 粒紧密吸附，没有溶解能力，因此不能被植物利用。而 土壤中的重力水虽然可以被植物利用，但因其很快便 渗漏出根层，因此也不计入有效水。因此，一般说土壤 中可被植物吸收利用的有效水，指的就是毛管水。任 一时段某一深度的土壤有效水含量，可通过同时段同 深度土壤含水量与萎蔫系数（植物根系无法吸收到水 分而发生永久萎蔫时的土壤含水量）的差值求得，计 算公式如下： hj hj h we w f= − （11） 式中，wehj 为时段 j 土壤 h 层深度的有效水分含量， mm；whj 为时段 j 土壤层 h 深度的水分含量，mm；fh 为土壤 h 层深度的萎蔫系数，mm。 因此，0~50cm 土层的平均土壤有效水含量计算如 公式（12）： (50) (50) (50)j j we w f= − （12） 式中，we(50)j 为时段 j 的 0~50cm 土层的平均有效水分 含量，mm；w(50)j 为时段 j 的 0~50cm 土层平均土壤含 水量；f(50)为该土层深度的平均萎蔫系数。 结合土壤湿度的反演方程，上式可变换为： (50) (50) ( )j jwe f csmi f= − （13） 据此，玉米生育阶段的灌溉需水量模型又可变换为： (50) (50) (50) 0 ( ( ) ) -( ( ) ) >( ( ) ) ci i i i ci i i ci i i et p f csmi f ir et p f csmi f et p f csmi f   + −  =   + − + − ， ， （14） journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 54 中，水分数量特定的情况下，作物实际吸收到的水分 数量称为作物实际耗水量。当任一时段外部环境提供 的水量（有效降雨量与土壤有效水含量之和）大于等 于本时段作物的需水量时，则在该时段，作物实际耗 水量等于作物需水量；当任一时段外部环境提供的水 量（有效降雨量与土壤有效水含量之和）小于本时段 作物的需水量时，则在该时段，作物实际耗水量等于 降雨量与土壤有效水含量之和，作物需水量与作物实 际耗水量之间的差值即为作物灌溉需水量[12]。 在旱生作物的整个生育期期间，任一时段土壤计 划湿润层中需灌水量，取决于该时段的来水量和作物 需水量的多少。依据上述原理，构建辽西北地区玉米 各生育阶段的灌溉需水量模型，公式如下： ( )i ci i i iir et p we i= − + + （4） 其中，iri 为辽西北地区玉米在生育阶段 i 的灌溉需水 量，mm；etci 为该地区玉米 i 阶段需水量，mm；pi 为 该地区生育阶段 i 内的有效降雨量，mm；wei 为本生 育阶段土壤有效水含量，mm；ii 生育阶段 i 的计划灌 溉量，mm。 根据作物生长水分的来源，辽西北地区的农业生 产方式可以分为两种，灌溉型农业和雨养型农业。研 究以雨养玉米为研究对象。对于雨养型农业来说，作 物的生长发育完全依靠自然降水，因此灌溉量 i 为 0， 此时玉米生育阶段的灌溉需水量模型为： ( )i ci i iir et p we= − + （5） 当区域外部来水量大于玉米需水量时，则认为该 区域不需要灌溉；当区域外部环境无法为玉米提供充 足的需水量时，则此时玉米的缺水量即为区域需灌溉 的水量。因此，玉米在某一生育阶段的灌溉需水量应 为： 0 ( ) -( ) >( ) ci i i i ci i i ci i i et p we ir et p we et p we   +  =   + + ， ， （6） 1.4 玉米生育阶段灌溉需水量估算模型参数的确 定 1.4.1 生育阶段有效降雨量 一次降雨的总雨量分配主要有三个方式：深层渗 漏、径流和留存于土壤中。雨水中的一部分渗漏到作 物根区吸水层以下，一部分作为径流从土壤表面流走， 地表的径流水和深处的渗漏水都不能被作物利用，换 言之，这两部分的降雨量对于作物是无效的。剩余的 降雨量贮存在土壤根区中，并为作物生长提供水分， 这部分可被作物吸收利用的降雨量称为有效降雨量[13]。 目前，多采用降雨有效利用系数计算有效降雨量， 公式如（7）所示： e j j p p=  （7） 式中，pj 为某次降雨总量，mm；pe 为该次降雨的有效 降雨量，mm；αj 为该次降雨量的有效利用系数。依据 康绍忠，蔡焕杰（1996）的研究结果，α 的取值如公式 [14]： 0 5 0.9 5 50 0.75 50 j j j j p mm mm p mm p mm    =     ， ， ， （8） 通过累积逐日有效降雨量得到玉米各生育阶段的有效 降雨量，公式如（9）所示： , 1 j i n ei e j p p = =  （9） 其中，pei 为生育阶段 i 的有效降雨量，mm；i=1,2,…6, 分别表示玉米的 6 个生育阶段；pej,i 为生育阶段 i 内第 j 次降雨的有效降雨量，mm；j=1,2,…,n，表示该生育 阶段内降雨次数。 1.4.2 生育阶段需水量的计算 作物需水量的影响因素有很多，包括气象条件 （温度、日照、湿度、风速）、土壤水分含量、作物种 类及其生长发育阶段、土壤肥力、农业技术措施等。目 前，计算作物需水量最常用的方法是通过计算参照作 物的需水量来得到某种作物的需水量。研究表明，联 合国粮农组织 fao 推荐的单作物系数法在预测东北地 区作物需水量时效果最好[15]。因此，本文采用单作物 系数法计算作物的逐日蒸散量，即作物逐日需水量， 则作物生育阶段的需水量由其逐日需水量累积得到。 作物逐日需水量的计算公式如下： 0c cet k et= （10） 式中，etc 为逐日作物蒸散量，即逐日作物需水量，mm； kc 为作物系数；et0 为逐日作物参考蒸散量，mm。kc、 et0 分别采用联合国粮农组织 fao 推荐的修正公式和 彭曼蒙特斯公式进行计算[16,17]。 1.4.3 土壤有效水含量的计算 土壤中的水分并非所有都对作物有效。例如，土 壤吸湿水和膜状水就属于无效水，由于它们被土壤颗 粒紧密吸附，没有溶解能力，因此不能被植物利用。而 土壤中的重力水虽然可以被植物利用，但因其很快便 渗漏出根层，因此也不计入有效水。因此，一般说土壤 中可被植物吸收利用的有效水，指的就是毛管水。任 一时段某一深度的土壤有效水含量，可通过同时段同 深度土壤含水量与萎蔫系数（植物根系无法吸收到水 分而发生永久萎蔫时的土壤含水量）的差值求得，计 算公式如下： hj hj h we w f= − （11） 式中，wehj 为时段 j 土壤 h 层深度的有效水分含量， mm；whj 为时段 j 土壤层 h 深度的水分含量，mm；fh 为土壤 h 层深度的萎蔫系数，mm。 因此，0~50cm 土层的平均土壤有效水含量计算如 公式（12）： (50) (50) (50)j j we w f= − （12） 式中，we(50)j 为时段 j 的 0~50cm 土层的平均有效水分 含量，mm；w(50)j 为时段 j 的 0~50cm 土层平均土壤含 水量；f(50)为该土层深度的平均萎蔫系数。 结合土壤湿度的反演方程，上式可变换为： (50) (50) ( )j jwe f csmi f= − （13） 据此，玉米生育阶段的灌溉需水量模型又可变换为： (50) (50) (50) 0 ( ( ) ) -( ( ) ) >( ( ) ) ci i i i ci i i ci i i et p f csmi f ir et p f csmi f et p f csmi f   + −  =   + − + − ， ， （14） journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 54 55 2 结果与分析 分析历史资料可知，辽西北地区 2006 年发生了自 1951 年以来的特大伏旱。因此，以 2006 年为例，利用 辽西北地区 11 个农业气象站点 2006 年逐日气象要素， 分别计算各站点玉米逐日需水量和日有效降雨量，通 过累积确定玉米各生育阶段的玉米需水量和有效降雨 量，根据辽西北玉米灌溉需水量模型，确定各站点玉 米在不同生育阶段的灌溉需水量。采用反距离加权法 对各站点玉米不同生育阶段的灌溉需水量进行空间插 值，得到辽西北地区玉米播种-出苗-七叶（图１）、七 叶-拔节-抽穗（图２）和抽穗-乳熟-成熟（图３）等 6 个生育阶段的灌溉需水量空间分布图。 辽西北地区玉米在播种—出苗阶段灌溉需水量范 围约为 2-34mm，在该阶段，由于降雨量极少，整个辽 西北地区玉米干旱较为严重，辽西北地区玉米干旱强 图 1 辽西北地区玉米播种-出苗-七叶阶段的灌溉需水量空间分布 图２ 辽西北地区玉米七叶-拔节-抽穗阶段的灌溉需水量空间分布 图３ 辽西北地区玉米抽穗-乳熟-成熟阶段的灌溉需水量空间分布 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 56 2 结果与分析 分析历史资料可知，辽西北地区 2006 年发生了自 1951 年以来的特大伏旱。因此，以 2006 年为例，利用 辽西北地区 11 个农业气象站点 2006 年逐日气象要素， 分别计算各站点玉米逐日需水量和日有效降雨量，通 过累积确定玉米各生育阶段的玉米需水量和有效降雨 量，根据辽西北玉米灌溉需水量模型，确定各站点玉 米在不同生育阶段的灌溉需水量。采用反距离加权法 对各站点玉米不同生育阶段的灌溉需水量进行空间插 值，得到辽西北地区玉米播种-出苗-七叶（图１）、七 叶-拔节-抽穗（图２）和抽穗-乳熟-成熟（图３）等 6 个生育阶段的灌溉需水量空间分布图。 辽西北地区玉米在播种—出苗阶段灌溉需水量范 围约为 2-34mm，在该阶段，由于降雨量极少，整个辽 西北地区玉米干旱较为严重，辽西北地区玉米干旱强 图 1 辽西北地区玉米播种-出苗-七叶阶段的灌溉需水量空间分布 图２ 辽西北地区玉米七叶-拔节-抽穗阶段的灌溉需水量空间分布 图３ 辽西北地区玉米抽穗-乳熟-成熟阶段的灌溉需水量空间分布 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 56 57 度由北向南、从东至西增高，因此西部需灌水较多，而 东北部较少。出苗—七叶阶段，部分地区大面积降雨， 虽在一定程度上缓解了玉米旱情，但整体来看玉米的 干旱状况仍然由北向南、从东至西增高，且玉米在该 阶段开始光合作用，蒸腾蒸发量增大，因此该阶段的 玉米灌溉需水量较前一阶段也增多，约为 13-56mm， 阜新市需水量较少，其余各市都较多，灌溉需水量趋 势符合该阶段实际干旱分布状况。玉米在七叶-拔节阶 段，由于受上一阶段降雨影响，干旱状况继续有所缓 解，因此玉米灌溉需水量在该生育阶段与上一阶段相 比整体有所减少，约为 0-24mm，整体来看，仍然是东 北部偏少，西部偏多，其中沈阳市和铁岭市的灌溉需 水量显著下降，葫芦岛南部、锦州北部和中部地区玉 米灌溉需水量最少，朝阳市、阜新北部和锦州南部玉 米灌溉需水量最多。玉米进入拔节期，研究区东北部 降雨量较少，因此东北部灌溉需水量与上一阶段相比 明显增多，在此阶段玉米灌溉需水量大约 15-44mm， 其中朝阳、阜新西部干旱程度及状况最为严重，因此 需水最多。抽穗-乳熟阶段是玉米生长的关键期，也是 全生育期需水最多的时期。在该阶段玉米灌溉需水量 在 20-113mm，与上一生育阶段相比各区域玉米灌溉需 水量都显著增多，其中锦州市市辖区和沈阳市市辖区 灌溉需水量最少，阜新市需灌水最多，其次为朝阳市， 东部和南部较少。乳熟—成熟阶段，辽西北地区西部 和南部玉米干旱状况有所缓解，干旱较为严重的地区 主要分布于朝阳西部、沈阳北部和阜新。在该阶段玉 米灌溉需水量约为 30-71mm，朝阳西部和阜新市玉米 灌溉需水量最多；由于该阶段玉米需水较少且沈阳北 部在前几个生育阶段干旱并不严重，因此沈阳北部玉 米的灌溉需水量并不是最多；与上一生育阶段相比， 各地区的灌溉需水量均有所下降。 3 结论与讨论 灌溉是缓解作物干旱的最有效手段。本文以 2006 年为例，通过计算研究区玉米逐日需水量和日有效降 雨量，根据研究区玉米的生育阶段，累积获得玉米各 生育阶段的玉米需水量和有效降雨量，利用不同生育 阶段的农田浅层土壤湿度指数反演了玉米各生育阶段 的土壤含水量，构建了辽西北玉米各生育阶段灌溉需 水量估算模型，确定各站点玉米在不同生育阶段的灌 溉需水量。通过对典型案例年的分析发现，玉米在不 同的生育阶段对灌溉需水量的需求也不同。整体来看， 辽西北地区西部玉米的灌溉需水量高于东北部的玉米 灌溉需水量；出苗-七叶阶段，部分地区大面积降雨缓 解了玉米在该阶段及下一阶段的干旱状况，但研究区 玉米的灌溉需水量仍然呈现西部高于东北部；抽穗-乳 熟阶段，辽西北地区玉米灌溉需水量最多，这种需求 不仅与玉米自身生长对水分的要求不同有关，还与外 部来水量紧密相关。 由于土壤数据的限制，本文仅以 2006 年为例探讨 了研究区玉米不同生育阶段的灌溉需水量，时效性较 差。但是本文的重点在于玉米不同生育阶段灌溉需水 量估算模型的构建，所以研究成果是可以接受的。另 外，研究中发现，作物的生长变化可以在一定程度上 反应作物灌溉需水量的变化，这将是下一步要继续完 成的工作。 参考文献 1. 郑大玮,李茂松,霍治国.农业灾害与减灾对策.北京:中国 农业大学出版社,2013. 2. 史培军 . 中国自然灾害风险地图集 .北京 :科学出版 社,2012. 3. 肖俊夫.中国玉米生长期干旱与灌溉投入问题分析.灌 溉排水学报,2009,28(5):21-24. 4. daccache a.,knox j.w.,weatherhead e.k.,et al. implementing precision irrigation in a humid climaterecent experiences and on-going challenges. agricultural water management,2015,147:135-143. 5. 黄志刚, 肖烨, 张国, et al. 气候变化背景下松嫩平原玉 米灌溉需水量估算及预测. 生态学报, 2017, 37(7):23682381. 6. 牛纪苹, 粟晓玲, 唐泽军. 气候变化条件下石羊河流域 农业灌溉需水量的模拟与预测. 干旱地区农业研究, 2016, 34(1):206-212. 7. 郭建平.东北地区近46年玉米气候资源变化研究.中国 农业气象,2009,30(3):302-307. 8. 高晓容. 近50年东北玉米生育阶段需水量及旱涝时空 变化.农业工程学报,2010,12(28):102-109. 9. 杜晓,王世新,周艺.一种新的基于 modis 的地表含水量 模型构造与验证.武汉大学学报(信息科学版), 2007, 32(3): 205-207. 10. 张红卫,陈怀亮,申双和,等.基于表层水分含量指数的土 壤干旱遥感监测.遥感技术与应用,2008,23(6):624-628. 11. 刘晓静,马东来,张继权,等. 辽西北地区玉米干旱的识 别方法. 干旱区研究, 2018(3). 12. a.araya,leo stroosnijder. assessing drought risk and irrigation need in northern ethiopia. agriculture and forest meteorology,2011,151: 425-436. 13. c. brouwer,m. heibloem. fao:irrigation water management training manual no.3:irrigation water needs. rome,natural resources management and environment department,1986. 14. 康绍忠 ,蔡焕杰.农业水管理学 .北京:中国农业出版 社,1996. 15. 李彩霞,陈晓飞,韩国松,等.沈阳地区作物需水量的预测 研究.中国农村水利水电, 2007(5):61-67. 16. 刘钰, l.s.pereira.对 fao 推荐的作物系数计算方法的 验证.农业工程学报, 2000, 16(5):26-30. 17. r g allen, l s pereira, d raes, et al. crop evapotranspiration guidelines for computing crop water requirements. fao irrigation and drainage paper 56, 1998. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 58 度由北向南、从东至西增高，因此西部需灌水较多，而 东北部较少。出苗—七叶阶段，部分地区大面积降雨， 虽在一定程度上缓解了玉米旱情，但整体来看玉米的 干旱状况仍然由北向南、从东至西增高，且玉米在该 阶段开始光合作用，蒸腾蒸发量增大，因此该阶段的 玉米灌溉需水量较前一阶段也增多，约为 13-56mm， 阜新市需水量较少，其余各市都较多，灌溉需水量趋 势符合该阶段实际干旱分布状况。玉米在七叶-拔节阶 段，由于受上一阶段降雨影响，干旱状况继续有所缓 解，因此玉米灌溉需水量在该生育阶段与上一阶段相 比整体有所减少，约为 0-24mm，整体来看，仍然是东 北部偏少，西部偏多，其中沈阳市和铁岭市的灌溉需 水量显著下降，葫芦岛南部、锦州北部和中部地区玉 米灌溉需水量最少，朝阳市、阜新北部和锦州南部玉 米灌溉需水量最多。玉米进入拔节期，研究区东北部 降雨量较少，因此东北部灌溉需水量与上一阶段相比 明显增多，在此阶段玉米灌溉需水量大约 15-44mm， 其中朝阳、阜新西部干旱程度及状况最为严重，因此 需水最多。抽穗-乳熟阶段是玉米生长的关键期，也是 全生育期需水最多的时期。在该阶段玉米灌溉需水量 在 20-113mm，与上一生育阶段相比各区域玉米灌溉需 水量都显著增多，其中锦州市市辖区和沈阳市市辖区 灌溉需水量最少，阜新市需灌水最多，其次为朝阳市， 东部和南部较少。乳熟—成熟阶段，辽西北地区西部 和南部玉米干旱状况有所缓解，干旱较为严重的地区 主要分布于朝阳西部、沈阳北部和阜新。在该阶段玉 米灌溉需水量约为 30-71mm，朝阳西部和阜新市玉米 灌溉需水量最多；由于该阶段玉米需水较少且沈阳北 部在前几个生育阶段干旱并不严重，因此沈阳北部玉 米的灌溉需水量并不是最多；与上一生育阶段相比， 各地区的灌溉需水量均有所下降。 3 结论与讨论 灌溉是缓解作物干旱的最有效手段。本文以 2006 年为例，通过计算研究区玉米逐日需水量和日有效降 雨量，根据研究区玉米的生育阶段，累积获得玉米各 生育阶段的玉米需水量和有效降雨量，利用不同生育 阶段的农田浅层土壤湿度指数反演了玉米各生育阶段 的土壤含水量，构建了辽西北玉米各生育阶段灌溉需 水量估算模型，确定各站点玉米在不同生育阶段的灌 溉需水量。通过对典型案例年的分析发现，玉米在不 同的生育阶段对灌溉需水量的需求也不同。整体来看， 辽西北地区西部玉米的灌溉需水量高于东北部的玉米 灌溉需水量；出苗-七叶阶段，部分地区大面积降雨缓 解了玉米在该阶段及下一阶段的干旱状况，但研究区 玉米的灌溉需水量仍然呈现西部高于东北部；抽穗-乳 熟阶段，辽西北地区玉米灌溉需水量最多，这种需求 不仅与玉米自身生长对水分的要求不同有关，还与外 部来水量紧密相关。 由于土壤数据的限制，本文仅以 2006 年为例探讨 了研究区玉米不同生育阶段的灌溉需水量，时效性较 差。但是本文的重点在于玉米不同生育阶段灌溉需水 量估算模型的构建，所以研究成果是可以接受的。另 外，研究中发现，作物的生长变化可以在一定程度上 反应作物灌溉需水量的变化，这将是下一步要继续完 成的工作。 参考文献 1. 郑大玮,李茂松,霍治国.农业灾害与减灾对策.北京:中国 农业大学出版社,2013. 2. 史培军 . 中国自然灾害风险地图集 .北京 :科学出版 社,2012. 3. 肖俊夫.中国玉米生长期干旱与灌溉投入问题分析.灌 溉排水学报,2009,28(5):21-24. 4. daccache a.,knox j.w.,weatherhead e.k.,et al. implementing precision irrigation in a humid climaterecent experiences and on-going challenges. agricultural water management,2015,147:135-143. 5. 黄志刚, 肖烨, 张国, et al. 气候变化背景下松嫩平原玉 米灌溉需水量估算及预测. 生态学报, 2017, 37(7):23682381. 6. 牛纪苹, 粟晓玲, 唐泽军. 气候变化条件下石羊河流域 农业灌溉需水量的模拟与预测. 干旱地区农业研究, 2016, 34(1):206-212. 7. 郭建平.东北地区近46年玉米气候资源变化研究.中国 农业气象,2009,30(3):302-307. 8. 高晓容. 近50年东北玉米生育阶段需水量及旱涝时空 变化.农业工程学报,2010,12(28):102-109. 9. 杜晓,王世新,周艺.一种新的基于 modis 的地表含水量 模型构造与验证.武汉大学学报(信息科学版), 2007, 32(3): 205-207. 10. 张红卫,陈怀亮,申双和,等.基于表层水分含量指数的土 壤干旱遥感监测.遥感技术与应用,2008,23(6):624-628. 11. 刘晓静,马东来,张继权,等. 辽西北地区玉米干旱的识 别方法. 干旱区研究, 2018(3). 12. a.araya,leo stroosnijder. assessing drought risk and irrigation need in northern ethiopia. agriculture and forest meteorology,2011,151: 425-436. 13. c. brouwer,m. heibloem. fao:irrigation water management training manual no.3:irrigation water needs. rome,natural resources management and environment department,1986. 14. 康绍忠 ,蔡焕杰.农业水管理学 .北京:中国农业出版 社,1996. 15. 李彩霞,陈晓飞,韩国松,等.沈阳地区作物需水量的预测 研究.中国农村水利水电, 2007(5):61-67. 16. 刘钰, l.s.pereira.对 fao 推荐的作物系数计算方法的 验证.农业工程学报, 2000, 16(5):26-30. 17. r g allen, l s pereira, d raes, et al. crop evapotranspiration guidelines for computing crop water requirements. fao irrigation and drainage paper 56, 1998. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 52–59 58 59 atlantis press journal style received 30 march 2016 accepted 15 june 2016 identifying crisis threats: a partial synthesis of the literature on crisis threat assessment with relevance to public administrations christian kalbassi swiss graduate institute of public administration, faculty of law, criminal justice and public administration university of lausanne, ch-1015, lausanne, switzerland e-mail: christian.kalbassi@unil.ch abstract the crisis management literature offers a host of theoretical treatises that are relevant to crisis threat assessment in public administrations; however, these insights remain non-cumulative, non-consolidated, and unstructured in a field of individual contributions. using existing literature regarding crisis definitions, crisis classifications, and crisis threat assessment tools, this exploratory paper identifies crucial analytical approaches to crisis threat assessment with relevance to public administrations. a critical evaluation of the specific research reveals gaps in the theoretical knowledge base and identifies three conceptual requirements for promoting a profound theorization of the field: the provision of relevant crisis threat variables, the specification of these variables, and the consolidation of the existing crisis threat assessment literature. keywords: crisis threat identification, crisis threat assessment, crisis classification, analytical methods, analytical frameworks, public administration. 1. introduction with recent modernization processes (boin and 't hart, 2003) and the emergence of new global risks (beck, 2007), crises have become more complex. the new millennium brought with it ‘a series of new crisis challenges due to climate change and environmental issues, increasingly risk-prone vital networks, massive terrorist threats, huge increases in the cost of disasters due to global coupling and just-in-time production methods, global civil unrest, and financial and economic meltdowns’ (topper and lagadec, 2013: 5). to address these new circumstances, crisis management scholars require analytical methods and frameworks that are better adapted for crisis threats (boin and mcconnell, 2007; doktor, 2007; enander, lajksjö and tedfeldt, 2010; gundel, 2005; kouzmin, 2008; lalonde, 2007; quarantelli, 2001; robert and lajtha, 2002; rosenthal, boin and comfort, 2001; roux-dufort, 2007; topper and lagadec, 2013). however, the design of appropriate analytical methods and frameworks – which are hereinafter summed up as “analytical aids” – is challenging. a profound conviction led the field to the basic assumption that the essence of a crisis is a ‘wild and maverick reality, impossible to understand and grasp within frameworks shaped, built and stamped to contain stable and repeated phenomena’ (topper and lagadec, 2013: 8). consequently, the field suffers from methodological monolithism, caused by a strong focus on case studies and an underestimation of theorization attempts (roux-dufort, 2007). indeed, the assumption of non-generalizability of previous results may prevent or complicate the introduction of new and sustainable analytical aids. this raises the issue of what cross-case relevant analytical aids for identifying crisis threats with relevance to public administrations have already been implemented and what key crisis threat variables they refer to. journal of risk analysis and crisis response, vol. 6, no. 3 (october 2016), 110-121 published by atlantis press copyright: the authors 110 ch. kalbassi / identifying crisis threats in order to address this concern, this study provides an in-depth review and analysis of the literature on crisis threat assessment with relevance to public administrations, using three angles: crisis definitions, classifications, and assessment tools. the objective is to offer a foundation to develop new analytical aids in the field. the following research questions will be answered: a) what conceptual requirements would promote a profound theorization of the field of crisis threat assessment with relevance to public administrations? and b) what key crisis threat variables are relevant to the specific field? to answer these research questions, a thematic literature review has been performed. the purpose of such a review can be, among others, to discover important variables relevant to a topic and to distinguish what has been done from what needs to be done (cf. hart, 1998). over 200 relevant books and journal articles were studied, and 87 crisis threat variables with relevance to public administrations were identified. after carefully examining each item, they were reduced to 22 distinct variables. table 2 lists the identified variables. hereinafter, “crisis threats" refer to anything that would contribute to the tampering, destruction, or interruption of any service or item of value in a governmental jurisdiction (bayne, 2002). “crisis threat variables” are presented as factors that determine whether or not a situation develops into a certain type or intensity of a crisis. 2. literature review regardless of whether crises evolve abruptly or cumulatively (hwang and lichtenthal, 2000), the phases of crisis management can roughly be divided into ‘horizon scanning’, ‘early warning/preparation’, ‘containment’, ‘recovery/disaster’, and ‘evaluation’. in the first two phases, the crisis has not yet occurred and therefore preventive actions are the main concern. in phase three, the crisis hits an organization or a system and crisis mitigation measures are required, whereas in the fourth and fifth, a conclusion involving a recovery or catastrophic escalation of the situation gradually emerges (claeys and cauberghe, 2014; mitroff, 1988; sikich, 1995; töpfer, 1999). crisis threat assessment is the process through which a manager or leader makes sense of a crisis situation in order to develop an effective crisis plan (yin and jing, 2014). it starts in phase three of crisis management when, for example, a manager or leader realizes that a crisis is arising. when focusing on crisis threat assessment, the probabilities of crisis occurrence and prevention are not the main topics of interest, as would be the case in risk assessment (habegger, 2010). in this article, crisis threat assessment is not understood as part of risk assessment or the identification of possible threats before they occur, but rather a process that supports managers or leaders with the identification of more or less apparent threats that emanate from the crisis. at this point, the crisis is already evolving, and crisis threat patterns must be described for the derivation of appropriate crisis-response strategies. situation awareness processes – in other words, knowing what is going on around you (crichton, lauche and flin, 2005; endsley and garland, 2000) – become the focal point. situation awareness has been identified as a crucial component of the decisionmaking process (canon-bowers and bell, 1997); hence, making an accurate crisis threat assessment is the most critical aspect of a crisis response (shapiro and koocher, 1996). in particular, in combination with the quickly evolving nature of a crisis, inaccurate situational assessments can lead to a multidimensional crisis and an insidious threat pattern. therefore, identification of the key variables shaping the crisis is an important prerequisite for developing an effective crisis management plan (yin and jing, 2014). previous studies distinguish between the ‘objective’ and ‘subjective’ dimension of crisis assessment: the objective view describes crisis threats as existing in “real” form in the world. this means that the threats can be assessed by measureable variables with respect to the magnitude and severity of the consequences (gephart, van maanen and oberlechner, 2009). the subjective view addresses the crisis management process, including the individual judgments and perceptions of managers, and therefore represents the perspective of how people subjectively perceive crisis threats to themselves, groups, or organizations (boin, 2005; weick, 1993; yin and jing, 2014). this study addresses the objective view. processes associated with the crisis mitigation phase that result from managerial failures during crisis management (e.g., sheaffer, richardson and rosenblatt, 1998) and errors linked to media published by atlantis press copyright: the authors 111 ch. kalbassi / identifying crisis threats communication (pasquier and fivat, 2012) are thus not the primary focus of this paper. there are at least three distinguishable approaches in the crisis management literature that are germane to the objective view of crisis threat assessment: crisis definitions determine which potential events or circumstances can be examined, crisis classifications discuss patterns of crises in terms of pertinent crisis attributes, and the more practice-oriented crisis threat assessment tools provide crisis threat assessment frameworks to managers and leaders (see overview of the three approaches in table 1). 2.1. crisis definitions crisis definitions describe the primary underlying characteristics of crises. they examine crises from the viewpoint of the characteristics that constitute a crisis situation (hwang and lichtenthal, 2000) to distinguish them from other situations. despite diverse crisis definitions, most crisis researchers agree on the major characteristics of crises: potential threats, triggers, some form of incubation, uncertainty about the resolution of a crisis and its after-effects, time pressures, and potentially harmful impacts on individuals, organizations, or society as a whole (lalonde, 2011). summarized in one sentence, a crisis can be defined as ‘a serious threat to the basic structures or the fundamental values and norms of a social system, which – under time pressure and highly uncertain circumstances – necessitates making critical decisions’ (rosenthal, charles and 't hart, 1989: 10).1 the above definition applies to all types of disruption, including eco-threats, information technology (it) crashes, economic adversity, intrastate conflicts, prison riots, regional wars, exploding factories, and natural disasters (dayton, 2004). these disruptions can have immediate and measurable effects such as pollution, fire, and the cost of intervention, as well as indirect effects such as changes in laws and damage to reputations (jaques, gatot and wallemacq, 2007). the definition suggests an abrupt and severe disruption of the routine processes and institutional structures of a social system (boin, 2005), which can be 1 for a more detailed description of the criteria “threat”, “uncertainty” and “urgency”, see stern, e., 2014. from warning to sense-making: understanding, identifying and responding to strategic crises, oecd 2014. defined as a general precondition for a crisis. however, these disruptions are not easy to identify in practice, as they are often rooted in a variety of human, social, technical, economic, legal, and ethical dimensions, which in turn complicates the unified definition of a crisis (jaques, gatot and wallemacq, 2007). thus, the concept of crisis in the crisis management literature remains vague (roux-dufort, 2007; roux-dufort and lalonde, 2013; topper and lagadec, 2013); this vagueness can be explained by the fact that the identification and acceptance of a crisis is largely dependent on the perceptions of the relevant actors (boin, 2005; svedin, 2009; ’t hart and boin, 2001). consequently, a crisis definition cannot simply rely upon clearly specifiable indicators. this is the primary disadvantage of the concept of crisis: its main characteristics are, to a great extent, open to interpretation. major problems arise, for example, from the question of how “time pressure” should be measured: many crises such as the 2008 financial market crisis allow for a comparatively long response time in comparison to a crisis caused by a flood or a technological disaster. the decision of the swiss federal council to bailout the union bank of switzerland (ubs) with 66 billion swiss francs was made ten months after the serious liquidity problems were detected at the bank, nine months after the installation of a prestigious crisis management team with representatives at the highest level of state, and one month after lehman brothers declared bankruptcy. in all three cases (flooding, technological disaster, and financial market crisis), time pressure is high, but the period of action cannot explicitly be reduced to a specific time scale. it can be said that there is a lack of consensus as to what is a crisis (antonacopoulou and sheaffer, 2013). however, when a situation roughly satisfies the definition of a crisis (e.g., threat, time pressure, uncertainty, etc.), it is suitable for crisis threat assessment. whilst the suitability of a situation for assessment can be established by diagnostic aids (stern, 2009), the specific influence of crucial crisis threat variables must be based on other theoretical foundations. in order to obtain a more differentiated view of these variables, particular attention must be paid to crisis classification. published by atlantis press copyright: the authors 112 ch. kalbassi / identifying crisis threats 2.2. crisis classification crisis classification is a method for explaining crisis phenomena and the interdependence of specific crisis threat variables. gundel (2005: 106) emphasizes the importance of typologies for practical implications and improved understanding of crisis phenomena: ‘if one wants to know how different types of crises develop, what kinds of problems surround them and, most important, how they can be handled, (…) a classification would surely be helpful by identifying common traits of different crises.’ consequently, crisis management scholars have approached the topic via primary causes or primary traits (e.g., natural, technological, social, economic, political), one specific attribute (i.e., national vs. international, episodic vs. continuous, corporate vs. public, and normal vs. abnormal), segmentation (i.e., internal, external, technical, and organizational), time scaling (i.e., before, during, after), topical categories (e.g., product defects, international crises), and crisis components (cf. gundel, 2005; topper and lagadec, 2013). early forms of classification had a strong emphasis on the causes and triggers of crises. the distinction between ‘natural’ and ‘man-made’ disasters dominated the early years until the end of the 1980s (rosenthal and kouzmin, 1993). at that point, mitroff, pauchant and shrivastava (1988) came up with a more complex classification along internal-external and technicalsocial axes and suggested that organizations should consider forming dual crisis portfolios for preventive actions and crisis clusters. if organizations introduced such portfolios, it was argued, they could protect themselves with minimal coverage across the crisis cause clusters ‘external/internal threats’, ‘breaks’, ‘psychopathic threats’, and ‘external actions/threats’. rosenthal and kouzmin (1993) took up this common distinction form and extended it by creating subgroups of natural and man-made disasters by introducing the subclusters of mine disasters, oil spills, air disasters, crowd disasters, nuclear crises, terrorism, and chemical explosions. hwang and lichtenthal (2000) broke with the established tradition, arguing for a classification based on the ways in which a crisis develops instead of focussing on the causes. based on the theory of punctuated equilibria in biology, they present a distinction between abrupt and cumulative crises by focusing on dimensions such as ‘build up speed’, ‘predictability’, ‘specificity’, ‘recognition’, ‘trigger point’, ‘probability of occurrence’, and ‘misalignment with environment’. abrupt crises are relatively quick to emerge, hard to predict, clearly recognizable, triggered by a specific event, time-constant, and more aligned with the environment than cumulative crises, which are more likely to evolve gradually and are highly predictable, difficult to recognize, triggered by a threshold-limit, time-increasing, and in many aspects misaligned with the environment. with regards to the y2k-related it problems known as the ‘millennium bug’, quarantelli (2001) asserted that former typologies were outdated because they ignored relevant attributes such as social consensus, political will, and the influence of mass media. he noted that the y2k problem was comparably easy to solve because of the crisis-specific social consensus on what could be done, who would do it, and the resources that had to be provided. inspired by the attacks on the world trade centre in 2001, mitroff and alpaslan (2003) emphasized the need to distinguish ‘normal crises’ from ‘abnormal crises’. whereas normal accidents involve known types of personnel, as well as physical and economic crises, abnormal accidents represent criminal crises such as acts of terrorism, information crises, and reputation crises. gundel (2005) opted for a more methodological approach, claiming that the older typologies failed to meet mandatory requirements for a typology (i.e., encompassing all kinds of crises, considering related prevention and response measures for each class, and a manageable and exclusive number of subsets). with respect to these conditions, gundel identified the proactive criteria ‘predictability’ and the reactive ‘influence’ that determined four crisis clusters: conventional, unexpected, intractable, and fundamental crises. in fact, following gundel’s call for more sufficient typologies, a sudden historical caesura can be noted in the development of classifications with relevance to public administrations. recent publications largely dispense with crisis classification, and scholars have not replaced or extended it with more appropriate approaches. it seems that the expanding complexity of crises has made it increasingly difficult to introduce new crisis classifications. an almost confusing wealth of contributions to the understanding of individual published by atlantis press copyright: the authors 113 ch. kalbassi / identifying crisis threats phenomena of crises has become the focus of the recent literature, which can be divided into four dimensions that are relevant to crisis threat assessment in public administrations: crisis threats to a system/society, the quality of information for crisis threat assessment, political-administrative challenges, and the assertiveness of a crisis management team. 2 the first dimension describes threats emanating directly from the crisis, such as the novelty of a crisis (howitt and leonard, 2009), the human, environmental, economic and societal extent of crisis damage (hohl, brem and balmer, 2013), the reversibility of the damage (german advisory council on global change, 1998; la porte, 2007), the latency of the effects of the crisis (german advisory council on global change, 1998; la porte, 2007), the alignment of crisis effects with the environment (hwang and lichtenthal, 2000), the spatial impact (german advisory council on global change, 1998; international risk governance council, 2008), and the mobilization potential of a crisis (fink, 2000; german advisory council on global change, 1998). the second dimension outlines the quality of information available in the particular crisis, such as the completeness of knowledge about the causes and evolution of the crisis (ansell, boin and keller, 2010; la porte, 2007), the uncertainty over solutions (ansell et al. 2010), the access to information relevant for diagnosis and resolution, the consensus on the severity of the crisis among relevant actors, and the consensus on the utility/credibility of the available information (la porte, 2007). the third dimension determines the integration of other political-administrative capabilities in crisis mitigation. ansell et al. (2010) assert that crises have become increasingly ‘transboundary’ in nature in recent years, challenging political-administrative capabilities in terms of the number of governments involved, the extent of policy areas affected, and the past and future actors involved. a crisis is more difficult to respond to when it scores high on these three dimensions. the fourth and final dimension describes the assertiveness of the crisis manager or the crisis manager’s organization in relation to the emerging crisis. dimensions include whether the crisis management team shares responsibility for the crisis (brown and ki, 2013; coombs, 2007; fink, 2000; 2 for a summary overview of the four dimensions, see table 2. mcauley, duncan and russell, 1992), whether the crisis interferes with the normal operations or daily business of the crisis manager’s organization (fink, 2000), and whether there is the potential for contradictory strategies or meaning-making processes in the area affected (ansell, boin and keller, 2010; klinke and renn, 2001). in summary, crisis classifications provide many insights into crisis phenomena and are significantly more precise in describing crucial crisis threat variables than crisis definitions. the many individual contributions lead, in aggregate, to a differentiated view of the influence of specific crisis threat variables. however, non-cumulative and non-consolidated contributions make it difficult to deal with the variety of potentially relevant variables. in order to minimize this problem and to achieve a certain degree of applicability to crisis threat assessors, some crisis management scholars have developed crisis threat assessment tools, as discussed in the next section. 2.3. crisis threat assessment tools crisis threat assessment tools deliver a framework that is oriented towards practical responses and that should help crisis assessors identify potential crisis threats. turning the mountains of raw data (e.g., reports, rumours, and pictures) into a coherent picture of the situation is indeed a major challenge in itself (ansell, boin and keller, 2010). however, only a few noncumulative contributions that are relevant to public administrations can be identified in the existing literature. fink’s (2000) approach was intended to support managers and leaders with a crisis threat assessment tool that facilitated the entire process of crisis threat assessment. five variables (i.e., ‘intensity’, ‘media or government scrutiny’, ‘interference with normal operations of business’, ‘threat to public image’, and ‘impact on a company’s bottom line’) are scored on a scale from 1 to 10, which reveals how damaging a crisis may be if there is no crisis intervention; these figures can then be plotted on a crisis barometer that combines the axes ‘probability’ and ‘crisis impact value’. lewis’ (2006) ‘human impact assessment tool’ enables a planner to consider the social and emotional impacts of a crisis on the personnel in a crisis-affected organization. the factors ‘locus/area of impact’, published by atlantis press copyright: the authors 114 ch. kalbassi / identifying crisis threats ‘preparation/capacity to manage the incident’, ‘duration of the event’, ‘impact on operations’, ‘damage to premises’, ‘damage to reputation’, ‘injury and death to personnel’, ‘external damage and injury’, ‘impairment of personal property’, ‘impact on and damage to social structure’, and ‘concurrent stressors’ focus on more than the estimated stress level of the workforce. this tool requires rating each factor on a scale from 1 to 10 and defining an overall crisis-tolerance threshold. la porte (2007) focused on the properties of crises that have strong implications for organizational design, specifically when political and agency leaders consider enhancing the crisis management capacity of their institutions. the conditions ‘consensus on seriousness of the crisis’, ‘overall magnitude’, ‘speed of crisis unfolding’, ‘propagation of effects’, ‘perceived duration of effects’, ‘knowledge of causes and consequences’, ‘mix of information for diagnosis/remedy’, and ‘consensus on utility/credibility of information’ may present themselves in a variety of combinations, thereby multiplying the management challenges that can confront the institutions charged with responding to crises. in general, the higher the cumulative score for the anticipated crisis effects and the lower the confidence in the available information and causal knowledge about the particular crisis, the more likely it is that institutional disarray and systemic collapse will result. because the authors of these crisis assessment tools did not derive their variables in a retraceable manner from crisis management theory, it is difficult to determine the train of thought that led to the identification and selection of the variables. due to their intent to establish order out of chaos, the authors all chose heuristic methods to conceptualize their theoretical constructs. in summary, crisis threat assessment tools identify key crisis threat variables and operationalize them through scales and threshold values; however, the lack of a theoretical foundation, the relatively low number of contributions to the field, and the differing fields of application (e.g., human impact, organizational crises) make it difficult to use them as analytical aid in their current state. table 1 gives an overview of the three presented approaches: crisis definitions, crisis classifications, and crisis threat assessment tools. 3. synthesis: three conceptual requirements for promoting a fundamental theorization of the field in light of the research literature presented above, it is evident that the crisis management literature provides a host of contributions that are relevant to threat assessment by public administrations. however, these insights are scattered, vague in variable definition, and only applicable to particular fields in crisis management. this indicates a lack of theorization in the field of crisis threat assessment. in order to discuss this statement in more detail, three conceptual requirements for promoting a fundamental theorization of the field of crisis threat assessment with relevance to public administrations will be discussed: the provision of relevant crisis threat variables, the specification of these variables, and the consolidation of the existing literature on crisis threat assessment. 3.1. provision of relevant crisis threat variables today, crises can result from complex interlinked circumstances; thus, the causes are frequently not only unique but also difficult to identify, even in retrospect. in a crisis, it is often unclear which crisis attributes are dependent (i.e., influenced factors) and which are independent (i.e., influencing factors). one can even expect that factors will change direction several times during a crisis. intense media coverage, for example, can result from a lack of crisis communication or from high sensitivity in the population, politicians, or media representatives. therefore, it is not the crisis causes that should be the focus of crisis threat assessment, but rather the circumstances in which a crisis is embedded. every crisis has a social impact that transforms any type of crisis cause into a unique threat pattern for organizations or systems. earthquakes, floods, technological disasters, mass demonstrations, and cyber-attacks only cause crises in relation to sociallysignificant effects. the primary challenge is therefore to identify the variety of relevant crisis threat variables that determine the potential for a situation to develop into a crisis. in other words, the field of crisis management must shift its attention on crisis threat variables that explain the potential impact (e.g., the specific damage potential published by atlantis press copyright: the authors 115 and the management challenges that crises pose to organizations and systems) so as to move away from the causal perspective. recent crises have become increasingly transboundary and interconnected (boin and lagadec, 2000) and intensively scrutinized by the media (’t hart, heyse and boin, 2001), and thus crisis causes become extremely difficult to identify. the increase in dynamic interrelations in today’s world can be depicted by the significant historical increase in crisis damage costs: the average insured global cost of man-made disasters and natural disasters was about 6 billion usd/year in the 1970s and 1980s, whereas it amounted to about 60 billion usd in 2014 (swiss re, 2015). technological and residential development, economic interdependence, and population growth all make society more vulnerable to crises. in previous decades, for example, a natural disaster would potentially impact a few farmhouses, whereas today in the same location, a small city that has significant economic power and a high level of interconnectivity with other dependent social systems might be affected. more people would be involved because of the expansion of residential development, as would other governments, enterprises, and interest groups from geographically distant regions. these presence-oriented and future-thinking variables offer a more complete picture of the threat pattern and the potential for cascading effects emanating from crisis situations than the unilateral declaration of crisis causes. hence, the suggestion of possible interactions between a variety of key crisis threat variables exceeds the situational awareness potential of the causal perspective, which often focuses on rigid crisis classifications on the basis of one or a few crisis attributes that are viewed as ‘responsible’ for the manifestation of the crisis. moreover, a series of these key crisis threat variables can aggravate situations and processes from relatively normal table 1. overview of the three approaches: crisis definitions, classifications, and threat assessment tools. crisis definitions crisis classifications crisis threat assessment tools objective identification of commonalities among all crises categorization of crises through specific attributes assessment support for managers and leaders in crisis situations structural approach distinction of crises from other events or conditions using main crisis characteristics determination of crisis clusters through the influence of specific crisis attributes provision of relevant crisis variables for crisis threat assessment determination of key variables for crisis threat assessment low, due to highly interpretable variables low, due to scattered discussion of crisis phenomena high, due to the comprehensive presentation of relevant variables abundance of variables for crisis threat assessment low, due to the limited number of variables discussed high, due to the discussion of a wide range of crisis threat variables low, due to the limited number of variables discussed data collection method heuristic relatively heuristic but selectively validated by methodological standards such as surveys heuristic operationalization of variables none none yes, but no detailed explanation about the scaling method and scaling levels test by confirmatory methods none none none inclusion of new theoretical insights no significant modification for more than 10 years no significant modification since gundel’s classification in 2005 no theoretical debate at all published by atlantis press copyright: the authors 116 ch. kalbassi / identifying crisis threats disruptions to severe crises and thereby contribute to highly variable threat patterns and difficult-to predict crises. consequently, the provision of key crisis threat variables is fundamental for an analytical aid for crisis threat assessment. for example, a crisis threat assessment framework on key crisis threat variables could broaden an assessor’s perspective (cf. table 2) and help him or her to resist the temptation to make a hasty and undifferentiated decision through unilateral processes of situational awareness. 3.2. specification of variables the identification and consolidation of relevant key crisis threat variables alone is insufficient for the development of analytical aids. the field of crisis management has failed to systematically define and operationalize its fundamental crisis threat variables. due to the underspecification of its major concepts and variables, analytical results remain largely unverifiable by subsequent studies (schulman, 2011). as many crisis scholars still consider the uniqueness of a crisis to be a non-alterable concept (roux-dufort, 2007), they consequently dispense with the operationalization of variables for cross-case analysis. this is apparent in the lack of pattern-detecting quantitative cross-case studies currently available in the crisis threat assessment literature that is relevant to public administrations. the large majority of crisis threat treatises ignore quantitative variable testing. the rare quantitative studies in the field do not analyse a series of crises in a comparative manner, but rather cluster crisis threat variables from top manager surveys (e.g., mitroff et al., 1988; yin and jing, 2014). however, the specific variables (e.g., ‘limited time’ and ‘pressure from the public’) are not further specified. the very meaning of these variables is left to the reader. moreover, because organizations rarely have significant experience managing crises (james, wooten and dushek, 2011), the corresponding studies may reproduce or rearrange already known insights about threat variables without contributing to greater specification. one analysis that makes an explicit claim of crosscase pattern detection is the extensive quantitative study of brecher and wilkenfeld (2000), which was based on 412 international crises and 895 foreign policy crises. this study created a more critical view of world politics by focussing on hostile and military interactions between adversarial states, but it did not provide sufficient variables and variable specification for the generic approach proposed by this article. even gundel’s (2005) heuristically-derived but methodologically-sophisticated approach to classification criteria barely meets these requirements. the definition of the two classification criteria ‘predictability’ (i.e., ‘a crisis is predictable, if place, time or in particular the manner of its occurrence are knowable to at least a third competent party and the probability of occurrence is not to be neglected’, p. 109) and ‘influence’ (i.e., ‘a disaster or crisis can be influenced if responses to stem the tide or to reduce damages by antagonising the causes of a crisis are known and possible to execute’, p. 109) remain at the highly abstract and interpretable level, even though gundel’s approach was developed for theoretical and practical crisis threat assessment. despite his suggestion that a crisis can be knowable by expert witnesses or through a threshold value established through conventions or precedents, gundel essentially leaves crisis threat assessors to fend for themselves. specifications for the terms ‘place’, ‘time’, ‘manner of occurrence’, and ‘third competent party’ are not provided. the classification criteria ‘influence’, which represents response possibilities, is even more difficult to apply: the reactive countermeasures should have a sufficient effect, doing more good than harm, and they should exceed simple measures such as evacuation. the impressive methodological content of gundel’s contribution notwithstanding, it is evident that other crisis scholars run into trouble when attempting to critically receive, test, or even build upon such existing treatises. definitions like gundel’s provide valuable theoretical insights into relevant crisis threat variables, as mentioned in section 3.1, but remain inapplicable to a crisis threat assessment framework due to the vague nature of the variables presented. hence, crisis threat assessment research must confront this critical issue by considering and testing the specification possibilities for crisis threat variables. in particular, a complex differentiation of these variables that is modelled after the variety of human, social, technical, economic, legal, and ethical crisis dimensions (see section 2.1), as well as sector-specific circumstances needs to be undertaken. published by atlantis press copyright: the authors 117 ch. kalbassi / identifying crisis threats 3.3. consolidation of existing crisis threat assessment treatises given the current state of the literature, it is hardly feasible to create a comprehensive crisis threat assessment framework for public administrations. indeed it is questionable whether such a general orientation would be requested. for weick (2007), a diverse set of theories, models, and causal assumptions served as a hedge against surprise and analytical error in crisis situations. however, the consolidation or partial consolidation of existing crisis threat assessment treatises can bring the field closer to complexity-reducing analytical aids. crisis management is a young field, and all young fields are refined by a certain kind of decoupled experimentation (dezenhall, 2014); however, the multiplicity of today’s crisis threat variables requires an urgent debate about fundamental assumptions in the field. the crisis management literature lacks cumulative research and acknowledged standards that are reflected in methodological approaches. whereas the concept of crisis has been the subject of profound attention in longestablished disciplines such as history, economics, political science, psychology, medicine, and philosophy, it remains elusive in the field of organizational science and management (roux-dufort and lalonde, 2013). in other academic fields, standardization attempts have been intensively implemented: doctors profit from standard examination methods and general models detecting diseases, psychotherapists can choose between a host of methodological schools (such as psychoanalysis, behavioural psychology, and client-centred psychotherapy) in response to individual crises, and economists have defined a range of standard factors that can influence financial crises (such as the investment of currency reserves, changes in interest rates, price stability, and consumer behaviour). even the crisis management-related field of risk management provides an integral risk assessment framework involving a series of risk categories and a risk matrix with the axes ‘probability’ and ‘extent of damage’. thanks to the introduction of this framework, risk assessment has become a part of a rational decision-making process that allows the description and comparison of the costs and utilities of potential risk-mitigating options (habegger, 2010). the crisis management literature is a long way away from such a methodological and conceptual consensus. in order to mature as a scientific field and consolidate its previous findings, the field must present existing insights in such a way that they are easily accessible, which in turn would facilitate dialectical processes between crisis management scholars. this is the cornerstone of an academic field: systematizing its findings and establishing central concepts, categories, and measurement definitions (klaus and buhr, 1975). crisis management scholars must therefore properly define and operationalize their concepts in a universal or methodology-specific manner in order to make their concepts interesting for cumulative research, thereby promoting the development of crisis threat assessment theory. the assumption that every crisis is different and not understandable with tools and frameworks (topper and lagadec, 2013) must be critically challenged, as this is essentially the same as saying that doctors do not need standard examination methods because every patient is unique and will exhibit an individual disease pattern to which he or she will respond in a unique way. no analytical method or framework for crisis threat assessment should be declared sacrosanct or exclusively applicable for pattern detection, but if theoretical constructs are included in our deliberations, the field could prevent the loss of sophisticated data and analytical approaches in the current scattered (roux-dufort, 2007) and incoherent conceptual morass. 4. conclusion, limitations and future research this study sought to overcome a shortcoming in the research by consolidating previous literature that is relevant to crisis threat assessment by public administrations. in addition, and in answer to the first research question, section three showed that fragmentary, unspecific, and unconsolidated theoretical treatises have little sustainable effect on the specific research. in particular, the field is struggling to develop an effective body of analytical aids that are relevant to public administrations. this represents a profound lacuna in the crisis management literature, as the increasing variety of influencing variables in today’s world can easily overstrain a crisis threat assessor’s capabilities. as vertzberger (1990) noted a quarter-century ago, when the published by atlantis press copyright: the authors 118 ch. kalbassi / identifying crisis threats flood of information was not nearly as overwhelming as it is today, it is not the lack of information but the misperception of available information that regularly leads to inappropriate decisions. crisis management scholars can minimize such aberrations if they systematically define the key variables and fields of crisis threat assessment by analysing the influence of variables in terms of cross-case relevance. if we continue to argue that everything that occurs in crisis situations is dependent on the specific circumstances, then the field of crisis management risks losing its legitimacy as an academic field by ignoring scientific premises, as discussed above in section three. it must therefore establish its central concepts by constructing fieldappropriate analytical aids around key crisis threat variables. moreover, analytical frameworks must be refined for public administration departments/subsections with assigned variables for federal administrations, lower levels of government (ladner and soguel, 2015), different policy sectors, and industry-specific conditions. table 2 presents an example of how the field “public administration” could be outlined with regards to the literature examined in this article (cf. section 2). this variable overview, which is based on the literature reviewed in section two, may be adequate to answer the second research question. each variable’s influence must be specified, operationalized, and tested through confirmative analysis in future research. the advantage of the variable list is that it provides a more exhaustive view of relevant key crisis threat variables than previous approaches. for example, eleven variables that address assertiveness and the quality of information support the determination of potential “influences” (gundel, 2005) on a crisis. however, the framework presented here is insufficient for establishing an analytical aid, since it leaves the identification of causal linkages and the solution of the analytical puzzle to the crisis threat assessor. restricting the crisis threat assessment framework to public administrations – and the very field of crisis threat assessment itself – therefore only meet the analytical requirements when the influence of a variable or a variable combination is linked with a certain degree of validity or statistical significance. to this end, challenging and well-specified explanatory hypotheses on the linkages between key crisis threat variables or between key crisis threat variables and a certain outcome must be formulated and tested through exploratory and confirmatory cross-case analysis. because the under-specified concepts in the crisis management literature are unlikely to lead to powerful and testable hypotheses (schulman, 2011), we can scarcely verify or disprove the existing research. the scientific challenge for the field of crisis threat assessment is thus to move towards a fundamental theorization through the systematic cross-case study of key crisis threat variables in a variety of subfields. because this requires, among other things, the use of unfamiliar approaches, renewing the debate on adequate methods for crisis threat assessment is a promising future direction. table 2. potential crisis threat assessment framework for public administrations crisis threats information quality political-administrative challenges assertiveness of the crisis management team • novelty • latency • crisis development • spatial impact of the crisis • extent of damage • reversibility • mobilization potential • access to information • consensus on utility of information • consensus on crisis severity • knowledge of causes • uncertainty level of potential solutions • vertical political boundaries • horizontal political boundaries • policy areas • time demarcation • responsibility • degree of interference with normal operations • consensus on (political) goals • concurrent stressors • autonomy • availability of resources published by atlantis press copyright: the authors 119 ch. kalbassi / identifying crisis threats references ansell, c., boin, a. and keller, a., managing transboundary crises: identifying the building blocks of an effective response system, journal of contingencies and crisis management 18(4)(2010), 195–207. antonacopoulou, e.p. and sheaffer, z., learning in crisis: rethinking the relationship between organizational learning and crisis management, journal of management inquiry 23(1)(2013), 5–21. bayne, j., an overview of threat and risk assessment (sans institute, swansea uk, 2002). beck, u., weltrisikogesellschaft: auf der suche nach der verlorenen sicherheit 1st ed. (suhrkamp verlag, frankfurt am main, 2007). boin, a., from crisis to disaster: towards an integrative perspective, in what is a disaster?: new answers to old questions, eds. perry r. w. and quarantelli e. l. (xlibris corporation, 2005) pp. 153–172. boin, a. and ’t hart, p., public leadership in times of crisis: mission impossible?, public administration review 63(5)(2003), 544–553. boin, a. and lagadec, p., preparing for the future: critical challenges in crisis management, journal of contingencies and crisis management 8(4)(2000), 185– 191. boin, a. and mcconnell, a., preparing for critical infrastructure breakdowns: the limits of crisis management and the need for resilience, journal of contingencies and crisis management 15(1)(2007), 50–59. brecher, m. and wilkenfeld, j., a study of crisis (university of michigan press, ann arbor, 2000). brown, k.a. and ki, e.-j., developing a valid and reliable measure of organizational crisis responsibility, journalism & mass communication quarterly 90(2)(2013), 363–384. canon-bowers, j.a. and bell, h.h., training decision makers for complex environments: implications of the naturalistic decision making perspective, in naturalistic decision making, eds. zsambok c. e. and klein g. a. (l. erlbaum associates, mahwah, n.j., 1997) pp. 99–110. claeys, a.-s. and cauberghe, v., keeping control: the importance of nonverbal expressions of power by organizational spokespersons in times of crisis, journal of communication 64(6)(2014), 1160–1180. coombs, w.t., protecting organization reputations during a crisis: the development and application of situational crisis communication theory, corporate reputation review 10(3)(2007), 163–176. crichton, m.t., lauche, k. and flin, r., incident command skills in the management of an oil industry drilling incident: a case study, journal of contingencies and crisis management 13(3)(2005), 116–128. dayton, b.w., managing crises in the twenty-first century, international studies review 6(1)(2004), 165–194. dezenhall, e., glass jaw: a manifesto for defending fragile reputations in an age of instant scandal (twelve, new york, ny, 2014). doktor, c., krisenmanagement im wandel: grundlagen, konzeptionelle weiterentwicklung und aktuelle trends, in bulletin 2007 zur schweizerischen sicherheitspolitik, eds. wenger a. and mauer v. (center for security studies (eth zürich), zürich, 2007) pp. 11–39. enander, a., lajksjö, ö. and tedfeldt, e.-l., a tear in the social fabric: communities dealing with socially generated crises, journal of contingencies and crisis management 18(1)(2010), 39–48. endsley, m.r. and garland, d.j. eds., situation awareness analysis and measurement (crc press, mahwah, nj, 2000). fink, s., crisis management: planning for the inevitable rev. (iuniverse, lincoln, neb, 2000). gephart, r.p., van maanen, j. and oberlechner, t., organizations and risk in late modernity, organization studies 30(2-3)(2009), 141–155. german advisory council on global change, strategies for managing global environmental risks: annual report 1998 (springer, berlin, heidelberg, 1998). gundel, s., towards a new typology of crises, journal of contingencies and crisis management 13(3)(2005), 106– 115. habegger, b., bewertung von risiken (center for security studies (eth zürich), zürich, 2010). hart, c., doing a literature review: releasing the social science research imagination (sage publications, london, 1998). ’t hart, p. and boin, a., between crisis and normalcy: the long shadow of post-crisis politics, in managing crises: threats, dilemmas, opportunities, eds. rosenthal u.,boin a.,and comfort l. k. (charles c thomas, springfield, il, 2001) pp. 28–46. ’t hart, p., heyse, l. and boin, a., new trends in crisis management practice and crisis management research: setting the agenda, journal of contingencies and crisis management 9(4)(2001), 181–188. hohl, m., brem, s. and balmer, j., a method for risk analysis of disasters and emergencies in switzerland (swiss federal office for civil protection, bern, 2013). howitt, a.m. and leonard, h.b., the novelty of crises: how to prepare for the unprecedented, in the l.a. earthquake sourcebook, eds. lewis j. and ulin d. (designmatters at art center college of design, pasadena, ca, 1 february 2009) pp. 210–217. hwang, p. and lichtenthal, j.d., anatomy of organizational crises, journal of contingencies and crisis management 8(3)(2000), 129–140. international risk governance council, an introduction to the irgc risk governance framework (international risk governance council, geneva, 2008). james, e.h., wooten, l.p. and dushek, k., crisis management: informing a new leadership research agenda, the academy of management annals 5(1)(2011), 455–493. published by atlantis press copyright: the authors 120 ch. kalbassi / identifying crisis threats jaques, j.-m., gatot, l. and wallemacq, a., a cognitive approach to crisis management in organizations, in international handbook of organizational crisis management, eds. pearson c. m.,roux-dufort c.,and clair j. (sage publications, los angeles, 2007) pp. 161–192. klaus, g. and buhr, m., philosophisches wörterbuch (bibliographisches institut, leipzig, 1975). klinke, a. and renn, o., precautionary principle and discursive strategies: classifying and managing risks, journal of risk research 4(2)(2001), 159–173. kouzmin, a., crisis management in crisis?, administrative theory & praxis 30(2)(2008), 155–183. ladner, a. and soguel, n., managing the crises – how did local governments react to the financial crisis in 2008 and what explains the differences? the case of swiss municipalities, international review of administrative sciences 81(4)(2015), 752–772. lalonde, c., the potential contribution of the field of organizational development to crisis management, journal of contingencies and crisis management 15(2)(2007), 95–104. lalonde, c., managing crises through organisational development: a conceptual framework, disasters 35(2)(2011), 443–464. la porte, t.r., anticipating rude surprises, in communicable crises: prevention, response, and recovery in the global arena, ed. gibbons d. e. (iap, charlotte, nc, 2007) pp. 27–44. lewis, g.w., organizational crisis management: the human factor (auerbach publications, boca raton, 2006). mcauley, e., duncan, t.e. and russell, d.w., measuring causal attributions: the revised causal dimension scale (cdsii), personality and social psychology bulletin 18(5)(1992), 566–573. mitroff, i.i., crisis management: cutting through the confusion, sloan management review 29(2)(1988), 15–20. mitroff, i.i. and alpaslan, m.c., preparing for evil, harvard business review 81(4)(2003), 109–115. mitroff, i.i., pauchant, t.c. and shrivastava, p., the structure of man-made organizational crises: conceptual and empirical issues in the development of a general theory of crisis management, technological forecasting and social change 33(2)(1988), 83–107. pasquier, m. and fivat, e., crisis communication characteristics & errors: a case study, public communication review 2(1)(2012), 16–34. quarantelli, e.l., another selective look at future social crises: some aspects of which we can already see in the present, journal of contingencies and crisis management 9(4)(2001), 233–237. robert, b. and lajtha, c., a new approach to crisis management, journal of contingencies and crisis management 10(4)(2002), 181–191. rosenthal, u., boin, a. and comfort, l.k. eds., managing crises: threats, dilemmas, opportunities (c. c. thomas, springfield, 2001). rosenthal, u., charles, m.t. and ’t hart, p., coping with crises: the management of disasters, riots, and terrorism (c.c. thomas, springfield, il, 1989). rosenthal, u. and kouzmin, a., globalizing an agenda for contingencies and crisis management: an editorial statement, journal of contingencies and crisis management 1(1)(1993), 1–12. roux-dufort, c., is crisis management (only) a management of exceptions?, journal of contingencies and crisis management 15(2)(2007), 105–114. roux-dufort, c. and lalonde, c., editorial: exploring the theoretical foundations of crisis management, journal of contingencies and crisis management 21(1)(2013), 1–3. schulman, p.r., problems in the organization of organization theory: an essay in honour of todd laporte, journal of contingencies and crisis management 19(1)(2011), 43– 50. shapiro, d.e. and koocher, g.p., goals and practical considerations in outpatient medical crises intervention, professional psychology: research and practice 27(2)(1996), 109–120. sheaffer, z., richardson, b. and rosenblatt, z., earlywarning-signals management: a lesson from the barings crisis, journal of contingencies and crisis management 6(1)(1998), 1–22. sikich, g.w., emergency management planning handbook (mcgraw-hill inc.,us, new york, 1995). stern, e.k., crisis navigation: lessons from history for the crisis manager in chief, governance 22(2)(2009), 189– 202. svedin, l.m., organizational cooperation in crises (ashgate publishing limited, farnham, england, 2009). swiss re, sigma: natural catastrophes and man-made disasters in 2014 (swiss re (sigma), zurich, 2015). töpfer, a., plötzliche unternehmenskrisen gefahr oder chance?: grundlagen des krisenmanagement, praxisfälle, grundsätze zur krisenvorsorge (hermann luchterhand verlag, neuwied; kriftel, 1999). topper, b. and lagadec, p., fractal crises a new path for crisis theory and management, journal of contingencies and crisis management 21(1)(2013), 4–16. vertzberger, y.y.i., the world in their minds: information processing, cognition, and perception in foreign policy decisionmaking (stanford university press, stanford, 1990). weick, k.e., the collapse of sensemaking in organizations: the mann gulch disaster, administrative science quarterly 38(4)(1993), 628–652. weick, k.e., the generative properties of richness, academy of management journal 50(1)(2007), 14–19. yin, s.and jing, r., a schematic view of crisis threat assessment, journal of contingencies and crisis management 22(2)(2014), 97–107. published by atlantis press copyright: the authors 121 1. introduction 2. literature review 2.1. crisis definitions 2.2. crisis classification 2.3. crisis threat assessment tools 3. synthesis: three conceptual requirements for promoting a fundamental theorization of the field 3.1. provision of relevant crisis threat variables 3.2. specification of variables 3.3. consolidation of existing crisis threat assessment treatises 4. conclusion, limitations and future research << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <feff05d405e905ea05de05e905d5002005d105d405d205d305e805d505ea002005d005dc05d4002005db05d305d9002005dc05d905e605d505e8002005de05e105de05db05d9002000410064006f006200650020005000440046002005d405de05d505ea05d005de05d905dd002005dc05d405d305e405e105ea002005e705d305dd002d05d305e405d505e1002005d005d905db05d505ea05d905ea002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e05d005de05d905dd002005dc002d005000440046002f0058002d0033002c002005e205d905d905e005d5002005d105de05d305e805d905da002005dc05de05e905ea05de05e9002005e905dc0020004100630072006f006200610074002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /pol <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> /ptb <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> /rum <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> /rus <feff04180441043f043e043b044c04370443043904420435002004340430043d043d044b04350020043d0430044104420440043e0439043a043800200434043b044f00200441043e043704340430043d0438044f00200434043e043a0443043c0435043d0442043e0432002000410064006f006200650020005000440046002c0020043c0430043a04410438043c0430043b044c043d043e0020043f043e04340445043e0434044f04490438044500200434043b044f00200432044b0441043e043a043e043a0430044704350441044204320435043d043d043e0433043e00200434043e043f0435044704300442043d043e0433043e00200432044b0432043e04340430002e002000200421043e043704340430043d043d044b04350020005000440046002d0434043e043a0443043c0435043d0442044b0020043c043e0436043d043e0020043e0442043a0440044b043204300442044c002004410020043f043e043c043e0449044c044e0020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200431043e043b043504350020043f043e04370434043d043804450020043204350440044104380439002e> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word parts of speech of anquan.doc parts of speech of anquan ming xu1,2,*, zongzhi wu3, yun luo1,2 1school of engineering ＆ technology, china university of geosciences (beijing), beijing 100083, china 2 key laboratory on deep geodrilling technology, ministry of land and resources (china university of geosciences (beijing)), beijing 100083, china 3china academy of safety science and technology, beijing 100012, china *e-mail: xuming@cugb.edu.com abstract researches on connotation of anquan were research hot spots in safety discipline field and the starting points of development of the discipline. the presupposition that anquan is noun was the foundation of those studies. but is this hypothesis really exists? in this paper, research has shown that the part of speech of anquan is only adjectives rather than noun in modern chinese dictionaries. there are many sentences which anquan was noun in them and this is the phenomenon of nominalization of predicate element in chinese. anquan was used as noun widely in safety discipline field and that was an example of nominalization of adjectives. a recommendation on revising the parts of speech of anquan in the chinese dictionary was proposed, namely, increasing the interpretation of anquan in case of it is noun for the demand of application and research in domestic and facilitating international convenient communication and learning. keywords: anquan, parts of speech, safety science, nominalization “安全”的词性 许铭 1,2,* ，吴宗之 3, 罗云 1,2 1. 中国地质大学（北京）工程技术学院，北京，100083 2. 国土资源部深部地质钻探技术重点实验室（中国地质大学（北京））,北京 100083 3. 中国安全生产科学研究院,北京 100012） *e-mail: xuming@cugb.edu.com 摘要：“安全”的内涵研究是安全学科领域的热点，也是学科发展的出发点。这些研究都基 于一个前提：“安全”是名词。这个前提存在吗？本文研究表明，汉语词典中只有 “安全”是形容 词的释义，没有属于名词的释义。汉语中的确存在“安全”为名词的许多例句，这是汉语谓词性 成分名词化现象。安全学科中“安全”一词广泛作为名词使用，就是形容词名词化的例子。为了 满足国内应用和研究的需要，方便国际交流和学习，建议汉语词典中加注“安全”为名词的释义。 关键词：安全；词性；安全学科；名词化 journal of risk analysis and crisis response, vol. 4, no. 2 (june 2014), 108-112 published by atlantis press copyright: the authors 108 willieb typewritten text received 20 february 2014 willieb typewritten text accepted 19 may 2014 willieb typewritten text 1 引言 安全学科是我国发展最快的新兴学科之 一。“安全科学技术”在国家标准《学科分类与 代码》（gb/t13745-2009）中为一级学科（代 码 620）。“安全科学与工程”在教育部《普通高 等学校本科专业目录(2012 年)》中是工科门类 下的一级学科（代码 0829），在 2011 年国务院 学位办发布的《学位授予和人才培养学科目 录》中，“安全科学与工程”为一级学科（代码 0837）。“安全科学”在《中国图书馆分类法》（第 五版，2010 年出版）中为一级类目（x9）。但 作为一门年轻的学科，建立科学系统、完整严 密的知识体系仍然是安全学科紧迫而艰巨的 任务 [1,2] 。 “安全”一词是安全学科的核心术语，其 内涵和外延、自然属性和社会属性、组成要素 及其相互关系等一直是研究的热点 [3-9] ，也是 学科发展的基础。这些研究都基于一个前提： “安全”是名词。 查阅《现代汉语词典》（第五版） [10] ，“安 全”是形容词，不是名词，其解释为：形 没有 危险；不受威胁；不出事故：～操作|～地带| 注意交通～。这个解释也是国内安全科技领域 专业词典以及其他工具书的解释 [11-17] 。解释相 同，说明词性也相同，即安全的词性是形容词， 不是名词。目前，鲜见“安全”词性的研究文 献。 词性是词语重要的分类标准，与词语的含 义和用法密切相关，实际使用中不仅要语义明 确清晰，还应符合语法要求。术语是学科的基 础，专业科学技术名词术语的规范化，统一学 科的基本概念并加以准确的表述，是科学技术 工作的一项基本建设 [17] 。因此，在研究“安全” 内容的同时不能忽视其词性，研究其词性，说 明其具有名词的用途和理由，对于学科本身的 科学性和严谨性非常重要。 2 汉语中“安全”的词性 2.1 现代汉语 如前所述，“安全”的词性是形容词。形容 词是表示人或事物的性质或状态的词，如“高、 细、软、白、暖和、活泼”[10]。 “安全”是名词的说法最早来源于刘潜先 生 1985 年发表的《从劳动保护工作到安全科 学（之二）——关于创建安全科学的问题》[18]， 文中认为：“从‘安全’的词性来说是状态性的名 词，可以表示人类的身心所处的‘状态’，也可 以展开，表示人类身心安全状态所处的向顺向 或逆向转换的‘过程’。能够表达一个完整的事 物所存在的状态和自身变化的运动过程，因此 可以做为学科名称使用。”但是刘潜先生未详 细说明安全是名词的依据。 对照《现代汉语词典》（第五版）关于形 容词的解释，刘潜先生上面论述的：“‘安全’ 表示人类的身心所处的‘状态’……”，显然其当 属形容词，而非名词。 何为名词？“名词”：名 ①表示人或事物 名称的词，如“人、牛、水、友谊、团体、今 天、中间、北京、孔子”。②（~儿）术语或近 似术语的字眼（不限于语法上的名词）：化学 ~|新~儿。③表达三段论法结构中的概念的词 [10] 。 “安全”是安全学科的核心术语，从这个角 度，其必须是名词。如果先有安全科学，后来 才在这个学科中定义了一个叫“安全”的新概 念，那么其词性自然是名词。但是，安全一词 早在安全科学定名之前已广为人知；而且在安 全科学定名之后，并没有对安全的理解带来太 大的变化。 从辩证法的角度，只要定义了一个事物的 反面，也就定义了它的正面。“安全”的反面是 “危险”，如果“危险”是名词，则“安全”也是 名词。 《现代汉语词典》（第五版）解释“危险”： 形 有遭到损害或失败的可能：～期|～区|～标 志|山路又陡又窄，攀登的时候非常～[10]。 可见，“危险”的词性也是形容词。现代汉 语中找不到依据，古代汉语中安全是否有当名 词的用法呢？ 2.2 古代汉语 在古代汉语中，“安”和“全”常是分开来讲 的，连起来使用的不常见。《说文解字》 [19] 讲 “安：静也，从女在宀（宀：交覆深屋也）下。” “全：完也，纯玉曰全”。“危：在高而惧也”， 引申为“凡可惧之称”。“险”即“阻，难也”。《康 熙字典》 [19] 讲“安：危之对也”，“全：完也。” 归纳起来就是“无危则安”、“无缺即全”。 published by atlantis press copyright: the authors 109 可见古代“安”和“全”也是表示人或事物 的性质或状态的词，即形容词。 古籍中有关“安全”的用法，大致有两种词 性。 （1）形容词，指无危险、不受威胁、平 安、太平、安定等。 《易•系辞下》中：“是故君子安而不忘危， 存而不忘亡，治而不忘乱，是以身安而国家可 保也。”这里指平安、不受威胁等状态。 《国策•齐策六》中有：“今国已定，而社 稷已安矣。”以及（明）无名氏的《临潼斗宝》 第三折：“你那铺谋定计枉徒然，我救的这十 七国诸侯得安全。”都指国家太平、安定、不 受威胁等状态。 《百喻经·愿为王剃须喻》中：“昔者有王， 有一亲信，於军阵中，殁命救王，使得安全。” 以及宋范仲淹的《答赵元昊书》中：“有在大 王之国者，朝廷不戮其家，安全如故。”这里 都指性命安全无危险。 （2）动词，指保全、保护，即“使安全”。 如《晋书·慕容垂载记》：“孤受主上不世之 恩，故欲安全长乐公，使尽众赴京师，然后脩 復国家之业，与秦永为邻好。”《南史·陈纪下·后 主》：“隋文帝以陈氏子弟既多，恐京下为过， 皆分置诸州县，每岁赐以衣服以安全之。”宋 苏轼《徐州谢上表》：“察孤危之易毁，谅拙直 之无他，安全陋躯，畀付善地。”都是保全性 命、保护人身安全之意。 归纳起来，汉语中，安全的词性有两种： 形容词和动词。动词的用法是古汉语的特殊用 法，现在已不用了。 3 英文中“安全”的词性 人们认为安全学科的“安全”一词对应的 英文单词主要有 safety 和 security。《牛津高阶 英语词典(第 8 版)》[21]对这两个单词注解的词 性都是名词，其释义摘录如表 1 所示。 英语中，safety 常用于描述身体免遭伤害 或免受危险，security 常用于描述经济上或精 神上得到保护或感受到保护。 《大英百科全书》（在线版）（encyclopædia britannica online）解释“safety(condition）”为： 为了最小化或消除能造成身体受伤的危险条 件的那些活动(行为)。这个解释表明是名词。 表 1. safety 和 security 的释义 词性 safety security 不可数 名词 the state of being safe and protected from danger or harm. the activities involved in protecting a country , building or person against attack, danger etc. 不可数 名词 the state of not being dangerous. protection against sth bad that might happen in the future. 不可数 名词 a place where you are safe. the state of feeling happy and safe from danger or worry. 归纳起来，不论 safety 还是 security，它 们的词性为名词。 英语中“安全”为形容词时对应的单词为 “safe”，其释义摘录如下[21]： ①adj. protected from any danger or harm. ②adj. not likely to lead to any harm or danger. ③adj. not harmed, damaged, lost etc. 可见，其释义与汉语词典中的解释接近。 4 讨论 综上所述，国内词典没有“安全”是名词的 解释。《现代汉语词典》（第五版）中“安全”解 释对应的英语单词为 safe。那么汉语中“安全” 可能成为名词吗？ 其实许多学者已经关注并研究此类现象, 这类现象被称为：动词形容词的“名物化”或 “名词化” [22] 。 语言学家朱德熙先生认为，汉语谓词性成 分可以名词化 [23] 。形容词属于性质的范畴，汉 语的形容词分为性质形容词和状态形容词两 大类，都可以名词化。例如，汉语表性质的形 容词一般不能直接名词化并指称具有这种性 质的人，其中一部分只有加上某种限制之后， 才能具备这个功能。例如“糊涂”，加上一个 “老”，成为“老糊涂”，就名词化了，并指称 具有这种性质的人(你真是一个老糊涂)。 文献[24]指出，用事物的属性来指称事物 本身，是人类语言的共同性质。有些谓词性成 分不用加任何形式标记也可以名词化，并且名 published by atlantis press copyright: the authors 110 词化后也是转指与这个谓词性成分相关的施 事、受事、与事、工具等等。 日常生产生活中，常把“安全”作为名词使 用。例如： （1）这款新车安全吗？是问它的安全性 怎么样，指称车辆的安全性。 （2）锅炉的安全怎么样？是问锅炉的安 全状况怎么样，指称锅炉的安全状况。 （3）“张安全”、“李安全”就是指称姓张 的安全员或姓李的安全员。 （4）安全抓得好不好？指称的是安全工 作、安全活动、安全教育、安全管理等。 （5）我是学安全的。指称的是安全专业。 （6）主抓安全的领导。指称的是安全工 作。 （7）不懂安全。是说不懂安全知识、不 懂安全技术、不懂安全管理或不懂安全科学， 指称的是安全知识、安全技术、安全管理、安 全科学。 （8）大安全。相对于传统安全提出的整 体安全概念，指称整体安全体系等。 可见，安全一词的确可名词化，而且大量 存在作为名词使用的情况，这些用法表达的语 义清楚明了。 此外，“安全”一词也有用作副词的例子。 例如“飞机安全着陆”、“安全行车”、“安全驾 驶”等都是修饰动词的，这时其为副词。 鉴于上述现象及安全学科发展的需要，笔 者建议《现代汉语词典》等汉语词典应加注“安 全”是名词、副词的释义，一方面为满足国内 应用和研究的需要，另一方面也能和国际接 轨，方便交流和学习。具体释义如下： 安全：①形 没有危险；不受威胁；不出 事故：～地带|～通道。 ②名 安全学科的术语（safety）：身体免 受危险或免遭伤害的状态或条件：人身～|～工 作|注意交通～|～科学。 ③名 社会安全的术语（security）：保护 国家、建筑、个人免遭攻击等危险的而采取的 活动和行为；摆脱危险或烦恼而感到快乐和安 全的心理状态：国家～ | 社会～ |～防卫 | ～ 心理。 ④副 无失误；不出事故等意外情况：飞 机～着陆| ～行车 | ～驾驶。 5 结论 1）现代汉语词典中“安全”的词性为形容 词，没有名词、副词等词性的释义。与《现代 汉语词典》（第五版）中“安全”解释对应的单 词是 safe，而不是学界普遍使用的 safety 和 security。 2）汉语中形容词可以名词化，日常用语 中的确存在“安全”为名词的许多用法。 3）“安全”作为安全学科中的术语，已被 人们广泛接受和使用。为了满足国内应用和研 究的需要，方便国际交流和学习，建议汉语词 典中加注其为名词、副词的释义。 致谢 本 研 究 获 国 家 科 技 支 撑 计 划 项 目 （ 2012bak13b00 ）；国家软科学研究计划 (2010gxs5b169)；中央高校基本科研业务费 专项资金资助项目（2652012040）；国土资源 部深部地质钻探技术重点实验室开放式基金 项目（nlsd201208）资助。 参考文献 1. 吴宗之. 中国安全科学技术发展回顾与展望. 中国安全科学学报，2000,10（1）：1-5 wu zong-zhi. review and prospects of safety science and technology in china. china safety science journal, 2000,10（1）：1-5 2. 吴宗之. 安全科学与灾害学探讨. 灾害学， 1991,6（1）：80-83 wu zong-zhi. a preliminary study on safety science and catastrophology. journal of catastrophology, 1991,6（1）：80-83 3. 崔克清.安全本质的再认识与当代安全问题. 化工进展,1994(3):23-26 cui ke-qing. modern safety problems and recognition of safety essence. chemical progress, 1994(3):23-26 4. 赵云胜，罗中杰. 论安全科学的几个基本问题. 地质勘探安全,1994(4):34-38 zhao yun-sheng, luo zhong-jie. the basic questions on safety science. geological prospection safety, 1994(4):34-38 5. 徐德蜀. 树立大安全观，保持社会可持续发展. 科技潮，1997（9）：79 xu de-shu. seting up the big safety concept published by atlantis press copyright: the authors 111 and maintainning social sustainable development. the tide of science & technology, 1997(9):79 6. 张景林, 王桂吉. 安全的自然属性和社会属性. 华北工学院学报, 2001, 22(4): 272-276 zhang jing-lin, wang gui-ji. the natural characteristics and social characteristics of safety. journal of north china institute of technology, 2001, 22(4): 272-276 7. 郑贤斌，李自力. 安全的内涵和外延. 中国安 全科学学报, 2003,13(2):1-3 zheng xian-bin, li zi-li. the connotation and extension of safety. china safety science journal, 2003, 13(2):1-3 8. 吕保和. 安全科学若干基础问题研究. 中国工 程科学，2006,8（11）：57-61 lv bao-he. study on some basic problems of safety science. engineering science, 2006, 8 （11）: 57-61 9. 中国社会科学院语言研究所词典编辑室 编. 现代汉语词典（第五版）. 北京：商务印书 馆,2005 dictionary editorial office of language institute of chinese academy of social sciences. modern chinese dictionary (the fifth edition). beijing, commercial press, 2005 10. 《安全科学技术词典》编委会. 安全科学技术 词典. 北京：中国劳动出版社,1991 editorial committee for “dictionary of safety science and technology”. dictionary of safety science and technology. beijing, china labor press, 1991 11. 安全科学技术百科全书. 北京:中国劳动社会 保障出版社, 2003 safety encyclopedia of science and technology. beijing, china labor social security publishing house, 2003 12. 张清源. 现代汉语常用词词典. 成都: 四川人 民出版社, 1992 zhang qing-yuan. modern chinese common word dictionary. chengdu, sichuan people's publishing house, 1992 13. 莫衡. 当代汉语词典. 上海: 上海辞书出版社, 2001 mo heng. contemporary chinese dictionary. shanghai, shanghai lexicographic publishing house, 2001 14. 任超奇. 新华汉语词典.武汉: 崇文书局, 2006 ren chao-qi. xinhua chinese dictionary. wuhan, chongwen publishing company, 2006 15. 熊武一, 周家法. 军事大辞海•上. 北京: 长城 出版社, 2000 xiong wu-yi, zhou jiafa. the military big lexicon (a). beijing, the great wall press, 2000 16. 严济慈. 安全科学技术词典 序言. 安全科学 技术词典，北京：中国劳动出版社，1991 yan ji-ci. dictionary of safety science and technology-foreword. beijing, china labor press, 1991 17. 刘潜. 安全科学和学科的创立与实践[m]. 北 京：化学工业出版社，2010 liu qian. the establishment and practice of safety science and discipline. beijing, chemical industry press, 2010 18. （汉）许慎 撰，（宋）徐铉 校定. 说文解字. 北京：中华书局,2009 (han dynasty) xu shen. shuo wen jie zi. beijing, zhonghua book company, 2009 19. 中华书局编辑部. 康熙字典（检索本）. 北京： 中华书局,2010 editorial department of zhonghua book company. kangxi dictionary (retrieve edition). beijing, zhonghua book company, 2010 20. （英）霍恩比 著. 牛津高阶英语词典(第 8 版). 北京：商务印书馆,2012 s. hornby. oxford advanced learner’s dictionary (8th edition). beijing, commercial press, 2012 21. 胡裕树,范晓. 动词形容词的“名物化”和“名词 化”. 中国语文, 1994, 2:81-85 hu yu-shu, fan xiao. nominalization of verb and adjective. chinese language, 1994, 2:81-85 22. 朱德熙. 语法答问. 北京：商务印书馆，1985 zhu de-xi. questions and answers on grammar. beijing, commercial press, 1985 23. 姚振武. 汉语谓词性成分名词化的原因及规 律. 中国语文, 1996,1:31-39 yao zhen-wu. cause and law of nominalization of predicate element in chinese. chinese language, 1996,1:31-39 published by atlantis press copyright: the authors 112 research article an analysis method of black swan event based on impact-spread tree xinying liu1,3, hang li2,3, xiaobing hu2,3,* 1the flight technology college, civil aviation university of china, tianjin 300300, china 2college of electronic information and automation, civil aviation university of china, tianjin 300300, china 3china-france research center of applied mathematics for atm, civil aviation university of china, tianjin 300300, china 基于影响扩散树的黑天鹅事件分析方法 刘 馨营1,3, 李 航2,3, 胡 小兵2,3,* 1飞行技术学院 中国民航大学，天津 300300 2电子信息与自动化学院 中国民航大学，天津 300300 3中法联合空管应用数学研究中心 中国民航大学，天津 300300 article info article history received 11 may 2020 accepted 05 july 2020 keywords black swan effect impact analysis impact-spread tree judgment air crash 关键词 黑天鹅事件 事件分析 影响扩散树 事件判定 坠机 abstract the black swan event has the characteristics of rarity and serious consequences. at present, the analysis of black swan events is generally preconceived (that is, the event is assumed to be a black swan event, and then the influence of the event is discussed), or whether it is a black swan event is directly judged according to its characteristics. this method of direct judgment by definition is one-sided and inaccurate, which is not conducive to the development and maturity of research on black swan events. therefore, this paper solved the problem of whether a method can be used to quantitatively determine whether an event can be called a black swan event before it is defined as a black swan event. this paper proposes a set of analysis methods to determine the black swan event by constructing an impact-spread tree. this method first defines the type of event, analyzes the possible impact range of the event, and draws it into a tree structure. then, according to the development law of the event, analyzes the impact spread relationship between nodes in the tree structure, and establishes the influence diffusion tree of the event. finally, by analyzing the impact spread ratio of the event, determine whether it is a black swan event. the case study shows that the method of impactspread tree can effectively distinguish between black swan events and non-black swan events. 摘要 黑天鹅事件具有稀有性和后果严重性的特征。目前对于黑天鹅事件的分析普遍存在先入为主的问题（即，先假定事 件是黑天鹅事件，再讨论事件的影响），或是根据特征直接判断是否是黑天鹅事件。这种直接用定义去判断的方法 存在片面性和不准确性，不利于黑天鹅事件研究的发展成熟。因此，本文旨在解决事件发生之后，其尚未被定义为 黑天鹅事件之前，是否可以利用一种方法来定量分析判断其是否可以被称为黑天鹅事件这一问题。通过对影响扩散 树的构建，提出了一套判定黑天鹅事件的分析方法（即，先分析事件的影响，再据之判定是否为黑天鹅事件）。该 方法先明确事件的类型，分析事件可能的影响范围，并初步绘制成树形结构；之后，根据事件的发展规律，分析树 状结构中各节点间的影响扩散关系，建立事件的影响扩散树；最后，通过分析事件的影响扩散比例，判定它是否为 黑天鹅事件。实例研究表明：影响扩散树这一方法可以有效地区分黑天鹅事件与非黑天鹅事件，有助于黑天鹅事件 分析中的对于事件类型的合理判定。 © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: huxbtg@163.com 1. 引言 随着社会发展和自然环境的不断变化，各种不安全事件频 发，其中有一类事件，发生频率很低，但如果发生其影响将 广泛深远。学者们通常将这类事件称为黑天鹅事件。例如， 美国的911恐怖袭击事件，它不仅造成了大量的人员死亡，而 且对经济发展、社会稳定和国际声誉等都产生了巨大影响。 因此，非常有必要研究和分析黑天鹅事件。 目前，学者们对黑天鹅事件已经展开了相关研究。liargovas p [1]选取了2001至2005年间三个黑天鹅事件，利用事件分析法 分析了此类事件对希腊股市的冲击。杜利芳 [2]论述了黑天鹅 与供应链及供应链风险管理之间的关系；茹瑜 [3]根据2011年 1月到2012年12月期间遭受黑天鹅事件冲击的32家上市公司的 journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 64–68 doi: https://doi.org/10.2991/jracr.k.200708.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto: huxbtg@163.com https://doi.org/10.2991/jracr.k.200708.001 https://www.atlantis-press.com/journals/jracr%20 x. liu et al. / journal of risk analysis and crisis response 10(2) 64–68 65 图1 | 黑天鹅事件、非常规突发事件与突发事件在定义上的包含关系. 图2 | 黑天鹅事件、非常规突发事件与突发事件可能影响范围. 表1 | 黑天鹅事件定义 [4–13] 序号 关键词 来源 1 稀有性+冲击性+事后预测性 taleb，2007 2 不可能 paté-cornell，2011年 3 令人惊讶+极端 aven，2012 4 不可预测 masys，2012 5 不可预测+重大 牟善军，2013 6 不可能+存在性 徐会永，2015 7 令人惊讶 hajikazemi，2016 8 异常值+抓住情绪+广泛接受 mueller，2013 9 意外性+极端影响+突发 刘霞，2017 10 出乎意料+颠覆性 周敏凯，2017 财务数据，运用因子分析和logit回归法对影响其股价恢复 能力的因素进行了分析。总的来说，以往研究都是直接认定 或假定一个事件就是黑天鹅事件，即“先入为主”的认定方 法，再利用相关方法分析该事件的影响，缺乏判定黑天鹅事 件的专门方法。鉴于此，笔者在探讨了黑天鹅事件概念的基 础上，通过建立影响扩散树，提出一套区别于“先入为主” 的判定分析黑天鹅事件的方法。新方法先分析一个事件的影 响，再据之判定该事件是否为黑天鹅事件。 2. 黑天鹅事件 2.1. 黑天鹅事件的定义 黑天鹅这一概念最早应用于经济学领域，金融专家taleb在 《黑天鹅》一书中，对黑天鹅事件的特征及事例进行了具体 介绍 [4]。之后，黑天鹅这一概念逐渐被运用到其他领域，不 同学者对其定义进行探讨描述，常见的黑天鹅定义见表1。由 表1可知，虽然黑天鹅事件定义众多，但学者们基本认同黑天 鹅事件具有发生频率低和影响严重等特征。 2.2. 黑天鹅事件与相关概念的关系 为了判定和分析黑天鹅事件，有必要将黑天鹅事件与现有相 关概念进行区分。这里主要探讨黑天鹅事件与突发事件和非 常规突发事件概念的关系。 突发事件：指突然发生，造成或者可能造成严重社会危害， 需要采取应急处置措施予以应对的自然灾害、事故灾难、公 共卫生事件和社会安全事件 [14]。 非常规突发事件：指发生的前兆不明确，具有明显的复杂性 特征和潜在次生衍生危害，破坏性严重，采用常规管理方式 难以应对处置的突发事件 [15]。事实上,国外学者很早就开始 研究这类事件，并将其称之为极端事件。其中美国国家科学 基金项目的专项研究则从事件的性质对非常规突发事件进行 了定义 [16]。 根据这两类事件的定义，以及以往黑天鹅事件的研究，可以 从事件涵盖范围和事件影响领域两方面分析三者之间的关 系。就事件涵盖范围而言，根据定义不难发现，突发事件涵 盖的事件类型最多，包含非常规突发事件和黑天鹅事件。而 黑天鹅事件和非常规突发事件从定义和特征看十分相似，但 学者们在研究两类事件时还是有所区别。并非所有非常规突 发事件都需要从黑天鹅事件角度进行分析。例如，2004年巴 拉圭超市火灾事件，事件共274人遇难，数百人受伤。这件事 被当作非常规突发事件进行研究，但黑天鹅事件的研究中并 未提及[16]。由此可见，非常规突发事件包含黑天鹅事件。 三者定义上的包含关系见图1。就事件影响范围而言，显然， 突发事件的影响领域最小。进一步分析黑天鹅事件和非常规 突发事件相关研究，可以发现，虽然两类事件都会造成严重 影响，但影响范围有差异。黑天鹅事件更强调系统全局性的 危害，它可以看作非常规突发事件中的极端情形。三者影响 范围关系也可以用图2进行阐明。 3. 黑天鹅事件的判定方法 为解决黑天鹅事件判定分析中，普遍存在的“先入为主”的 问题，笔者在传统事故分析方法的基础上，提出了影响扩散 树的方法以判定和分析黑天鹅事件，包括3个步骤。 3.1. 构建影响扩散树 常见的应用于安全和风险领域的树形结构的方法主要是故障 树和事件树。故障树从顶事件追溯因果路径 [17]，而后者从 单个事件出发，探索事件之后的可能性。这里，借鉴“树” 的思想构建影响扩散树。影响扩散树旨在研究原事件对可能 相关领域造成的影响，包括各领域内部或之间影响的相互传 递和扩散。需要注意的是，不同类型事件，其影响范围领域 有差异。对于某一类型事件，其影响扩散树建立步骤如下： （1）明确事件类型，分析其可能的影响领域，即影响扩散树 的第二层。（2）根据事件发生发展的实际情况，细化其在各 个领域的影响，即影响扩散树的最底层，每个具体影响称为 一个节点；（3）分析各节点间影响扩散关系，详见2.2节， 并用特定箭头相连，最终完成影响扩散树构建，如图3所示。 3.2. 分析各节点间影响扩散关系 分析各节点间影响的传递和扩散情况，才有可能真实掌握事 件造成的全部影响。根据事件影响的具体演化过程，可以将 66 x. liu et al. / journal of risk analysis and crisis response 10(2) 64–68 演化构成分为三种：直链式、辐射式、网络式。直链式是指 r影响开始，逐步演化成r1、r2…rn，路径呈现直链式。辐射 式是指r影响开始，同步演化成r1、r2…rn，路径呈并行发散 式。网络式是指r影响开始，逐步演化成r1、r2…rn，但r1、 r2…rn又可以反馈给r事件，互相之间蔓延，最终形成网络的 结构。 通过对各节点之间的关系进行分析，找出影响的传播关系， 在树状结构图的基础上，用箭头进行表示。 3.3. 判定方法 有了影响扩散树，就可以根据一个事件的影响在影响扩散树 中的扩散情况来判断该事件是否可以称为黑天鹅事件，而不 是先入为主地直接假定该事件为黑天鹅事件。如何基于影响 扩散树来判定黑天鹅事件，一般是可以根据事件的类型、领 域等采用定制化的判定标注。本文采用一种简单的扩散比例 判断法，即，要判断一个事件是否是黑天鹅事件，先要根据 构建的影响扩散树，计算事件影响的扩散比例：有了影响扩 散树，就可以根据一个事件的影响在影响扩散树中的扩散情 况来判断该事件是否可以称为黑天鹅事件，而不是先入为主 地直接假定该事件为黑天鹅事件。如何基于影响扩散树来 判定黑天鹅事件，一般是可以根据事件的类型、领域等采用 定制化的判定标注。本文采用一种简单的扩散比例判断法， 即，要判断一个事件是否是黑天鹅事件，先要根据构建的影 响扩散树，计算事件影响的扩散比例： a i s s i( ) = （1） 式中： a为某一事件类型， a(i)为某一具体事件的影响扩 散比例，si为事件i的影响在影响扩散树中的扩散节点数总 和，s为a类事件可能造成的所有影响的总和，即影响扩散树 中所有的节点。 然后，确定给定类型事件影响扩散的门限比例r。它的确定方 法多种多样，这里简单通过调研或专家打分的方式确定。 基于以上两步，则可以判定某一事件i是否为黑天鹅事件：如 果a(i) ≥ r，则称该事件为黑天鹅事件。如前所述，实际中， 不同类型事件，其影响扩散有差异，相应的影响扩散树和门 限比例也可能不同。下面以民航坠机事件为例，介绍如何应 用影响扩散树方法判定某个事件是否为黑天鹅事件。 需要强调的是：基于公式（1）的扩散比例判断法只是基于影 响扩散树来判定黑天鹅事件的一种可行标准而已。结合事件 类型、领域等特点，完全可以采用其他的判定标准，比如， 不是简单地进行节点数的加和，而是采用较为复杂的加权求 和法（即，每个节点有各自的权重，事件在一个节点处的影 响有一个百分比的分值）来构造判定标准，如下： a i sj j j ( ) = å( ) å w w * （2） 其中ωj是影响扩散树中节点j的权重，sj是事件i在节点j处的影 响的百分比分值。如果事件i在节点j处没有影响，则sj = 0。显 而易见，节点权重设定和事件在节点处影响的百分比分值的 计算都是基于问题特点而高度定制化的技术。本文后面主要 采用基于公式（2）的扩散比例的判定标准，以演示基于影响 扩散树的黑天鹅事件判断分析方法的应用过程。 4. 案例研究 为了验证前面所提出方法的可行性和有效性，这里，以民航 坠机类型事件为例，并选取两个典型的具体事件，波音737 max 8事件和伊春空难事件，分析它们是否为黑天鹅事件。 4.1. 事件简介 波音737 max 8事件：2019年3月10日8点38分，埃塞俄比亚航 空公司波音737 max 8客机，从亚的斯亚贝巴飞往内罗毕。飞 机起飞6分钟后失联，随后被确认坠毁，机上157名乘客和机 组人员全部遇难。 伊春空难事件：2010年8月24日21时36分,河南航空公司(原鲲 鹏航空有限公司)b3130飞机在执行哈尔滨—伊春航班任务时, 在黑龙江省伊春市林都机场降落时失事。机上人员96人,54人 生还,42人遇难。 4.2. 基于影响扩散树的坠机事件分析 根据前面的介绍，首先建立坠机类型事件的影响扩散树。基 于历年来发生的坠机事件，对事件可能造成影响的领域进行 分类，并划分出可能造成的结果，构建出坠机事件的影响扩 散树，如图5。 下面，逐一分析两个民航坠机事件的影响扩散情况。 波音737 max 8空难事故：737 max 8空难事故的发生对很多 社会领域都造成了巨大的冲击。主要有：（1）首当其冲的 是民航业，两起事故导致737 max 8机型飞机在全球范围停 飞，这对拥有该机型飞机的国内外航空公司的运营力、信任 度都造成了影响。停飞同时还增加了航空公司的运营成本， 包括常规成本及天气成本等 [18]。此次停飞对福建航空、上 海航空这些737 max 8机型占机队总数比重较大的航空公司影 响更大。同时，受到影响的航空公司的股价也都出现了大幅 缩水；（2）737 max 8机型飞机制造商-波音公司的声誉、经 图3 | 影响扩散树示意图. 图4 | 典型的影响扩散方式. x. liu et al. / journal of risk analysis and crisis response 10(2) 64–68 67 图5 | 坠机类事件影响扩散树. 图6 | 波音737max8事件影响扩散树. 济及订单量因此都造成了影响，例如其股价在3个交易日内 跌去了11.02%，市值蒸发256.51亿美元。未来与欧洲空客公 司的竞争中也将处于劣势地位；（3）保险业也会受一定影 响，因为保险公司要对空难中投保的航空公司、乘客进行巨 额的经济赔偿；（4）对民众心理造成恐慌。很多民众因此 对民航安全性提出质疑，未来出行方式可能也会有所改变； （5）民航法规发生变化。737 max 8空难事件发生后，各国 民航部门相继出台的停飞波音737 max 8飞机的政策，我国民 用航空局是第一时间出台这一规定。（6）国际关系受到影 响。美国和欧盟因此事在贸易关税方面争锋相对，两方之间 贸易关税的博弈愈加激烈。根据相关报道及统计数据，结合 前面的坠机类事件影响扩散树，可以得到737 max 8空难事件 的影响扩散树，如图6所示。加灰色代表具体事件造成的影 响。 伊春空难事件：伊春空难的发生结束了2102天的中国航空业 的安全运行，对中国航空业敲响了安全的警钟。但就影响范 围来看，空难发生虽然引发了全国热议，但影响范围有限， 主要有：（1）首先是金融业，事故发生后航空业的股价全 线下挫，尤其是南方航空公司、东方航空公司以及海南航空 公司，股价都出现了大幅缩水，但其影响只限于国内的航空 股，同时国家在事件发生后的一系列措施加上各方的补救， 对于金融领域的影响很快就消失了；（2）伊春林都机场在事 件发生后，停航了一天，由于地处边陲，因此没有造成大面 积的人员滞留，并且此后飞机客座率也没有受到影响；（3） 对于失事飞机所属航空公司河南航空而言这是一次毁灭性的 打击，河南航空的所有航班停止运营。由于资不抵债，河南 航空于2011年进入破产重组程序，并且撤销了河南航空有限 公司企业名称登记。根据相关报道及统计数据，可以得到 如图7的伊春空难的影响扩散树。 根据坠机类事件的影响扩散树，第四层节点对于整个公司及 领域的影响权重通过专家打分法来确定。坠机事件的扩散 比例门限为80%，根据两个事件相应的影响扩散树，即图6和 图7，并运用公式（2）分别计算得到737max 8事件的扩散比例 为96.08%，伊春空难扩散比例为55.56%。可以得出，737max 8 事件是黑天鹅事件，而伊春空难不是。这与有关这两个事件 的事后影响报道，以及相关的统计数据定性分析得到的结论 一致，从而验证了所提出方法的有效性。 5. 结论与展望 区别于“先入为主”的黑天鹅事件判定分析方法（即，先假 定一个事件就是黑天鹅事件，然后再讨论该事件的影响）， 文本提出了一种基于影响扩散树判定分析黑天鹅事件的方 法。通过分析事件可能的影响范围，及各具体影响间的扩散 关系，建立事件的影响扩散树，再分析事件影响扩散比例， 68 x. liu et al. / journal of risk analysis and crisis response 10(2) 64–68 图7 | 伊春空难事件影响扩散树. 从而判定该事件是否为黑天鹅事件。案例研究结果与相关报 道和统计数据的定性分析结论一致，验证了所提出方法的有 效性。 本文主要是对黑天鹅事件是否可以用定量方法去分析判断的 一种初步探索和尝试，将来还可对所提出的方法进一步开展 深入改进和拓展研究。例如，利用专家打分法得到的权重存 在一定的主观性，可以研究采用一些客观指标（经济体量、 持续时间，等等）合成计算权重的可能性，并应用基于机器 学习的大数据技术。另外，对方法的进一步验证和应用需要 用更多的事件对比和数据支持，除了分析更多民航客机坠毁 事件，还要研究更多其他领域（公共卫生领域，金融领域， 等等）的事件，以便能够进一步改进和完善影响扩散树的方 法架构和细节。 参考文献 [1] liargovas p, repousis s. the impact of terrorism on greek banks’ stocks: an event study. int res j finance econ 2010;51:88–96. [2] 杜利芳. 基于黑天鹅理念的制造企业供应链风险管理研究 [d]. 太原理工大学, 2014. [3] 茹瑜,许华,吴东生.黑天鹅事件冲击下上市公司股价恢复能 力研究— —基于财务分析的视角[j]. 金融经济, 2013:93–6. [4] taleb nn. the black swan: the impact of the highly improbable. new york: random house trade paperbacks; 2007, pp. 3–4. [5] paté-cornell e. on black swans and perfect storms: risk analysis and management when statistics are not enough. risk anal 2012;32:1823–33. [6] aven t. risk, surprises and black swans: fundamental ideas and concepts in risk assessment and risk management. new york: crc press; 2014, pp. 10–17. [7] masys aj. black swans to grey swans: revealing the uncertainty. disaster prev manage 2012;21:320–35. [8] 牟善军.当心黑天鹅事件— —浅谈风险评估的局限性[j]. 安 全、健康和环境, 2013;13:30. [9] 徐会永.论科技期刊名称中“黑天鹅事件”的启示[j]. 编辑学 报, 2015;27:16–18. [10] hajikazemi s, ekambaram a, andersen b, zidane yj-t. the black swan – knowing the unknown in projects. proc soc behav sci 2016;226:184–92. [11] mueller j, stewart mg. the curse of the black swan. j risk res 2016;19:1319–30. [12] 刘霞,白敏.中国股市“黑天鹅”事件的形成演化机理及其 防控研究— —以中石化黄岛事件为例[j]. 安阳工学院学 报,2017;16:38–43. [13] 周敏凯.当前西方民主乱象的根源与未来走向 — —基于 英美“黑天鹅事件”的理论思考[j].人民论坛·学术前 沿,2017:32–40. [14] 全国人民代表大会常务委员会. 中华人民共和国突发事件应 对法[l]. 2007-08–30. [15] 韩智勇,翁文国,张维,等.重大研究计划“非常规突发事件 应急管理研究”的科学背景、目标与组织管理[j].中国科学 基金,2009;23:215–20. [16] 刘霞,严晓,刘世宏.非常规突发事件的性质和特征探析 [j].北京航空航天大学学报(社会科学版),2011;24:13–17. [17] 王燕,尹盼盼,沈梦露.基于事故树的铁路客运站火灾风险因 素分析[j].中国安全科学学报,2019;29:44–7. [18] 李航,郭晓梅,胡小兵.灾害性天气下航空公司天气成本测算 模型[j].中国安全科学学报,2019;29:7–12. https://doi.org/10.1111/j.1539-6924.2011.01787.x https://doi.org/10.1111/j.1539-6924.2011.01787.x https://doi.org/10.1111/j.1539-6924.2011.01787.x https://doi.org/10.1108/09653561211234507 https://doi.org/10.1108/09653561211234507 https://doi.org/10.1016/j.sbspro.2016.06.178 https://doi.org/10.1016/j.sbspro.2016.06.178 https://doi.org/10.1016/j.sbspro.2016.06.178 https://doi.org/10.1080/13669877.2016.1216007 https://doi.org/10.1080/13669877.2016.1216007 乡村旅游安全风险形成机理研究 research on mechanism of risk coupling for rural tourism safety jingfeng luo1,2 1college of tourism, huaqiao university, quanzhou 362021, china 2center for tourism safety& security research of china tourism academy, quanzhou 362021, china e-mail: luojingfeng2006@126.com abstract in view of that risk impact processes has not been described and analyzed for rural tourism safety accidents, the paper aims at working out analysis model of risk coupling for rural tourism safety accidents. first, based on the regional environmental risk system theory, the rural tourism safety system is analyzed. second, under the four dimension of man, machine, environment, and management, the diamond model is established for rural tourism safety risk. then, based on connotation of coupling, the definition and types of risk coupling was given for rural tourism safety accidents. finally, based on the periphery theory and trigger working principle, a risk coupling mechanism model of rural tourism safety was established according to the “rural tourism risk trigger”. the results may give countermeasures and suggestions to risk management of rural tourism safety. keywords: rural tourism, tourism safety, risk coupling, mechanism, trigger, periphery theory 乡村旅游安全风险耦合机理研究 罗景峰 1,2 1 华侨大学旅游学院，福建泉州 362021 2 中国旅游研究院旅游安全研究基地，福建泉州 362021 摘要：鉴于乡村旅游安全事故风险作用过程迄今未被描述和分析，提出乡村旅游安全事故风险耦合分析模型。 首先，运用区域环境风险系统理论，分析乡村旅游安全系统构成及其要素相互作用关系；其次，从人、物、环 境及管理的维度出发，建立乡村旅游安全风险因素钻石模型；然后，基于耦合的内涵，界定乡村旅游安全事故 风险耦合的含义及类型；最后，根据界壳与触发器的优势互补性，建立乡村旅游安全风险耦合机理模型，并通 过该模型描述了乡村旅游安全耦合风险的形成机理。研究结果可为乡村旅游安全风险管理提供可行对策、合理 建议。 关键词：乡村旅游，旅游安全，风险耦合，机理，触发器，界壳理论 罗景峰（1975-），男，蒙古族，辽宁阜新人，讲师，博士，国家注册安全工程师，中国职业安全健康协会会员，中国系统工程学会会员， 主要研究方向为旅游风险分析与安全评价。 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 45 received 26 december 2016 accepted 27 january 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cmfd&sfield=kw&skey=+regional+environmental+risks http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cjfq&sfield=kw&skey=types+of+risks 与蓬勃发展的乡村旅游业相比，乡村旅游安全风 险管理工作尚未受到应有的重视，这与“安全是旅游 的生命线”理念极不适应。5.1 凤凰吊桥倾斜、8.7 上 海市青浦区农家乐液化气泄漏爆炸等事故的多风险耦 合特性，更增加了乡村旅游安全风险防控的难度。从 预防为主的观点出发，分析乡村旅游安全风险形成机 理，厘清乡村旅游安全风险作用路径，对于有效防控 乡村旅游风险显得至关重要。目前，有关这一方面的 研究尚未见到相关报道。为此，在分析乡村旅游安全 风险因素并划分乡村旅游安全耦合风险类型的基础 上，将界壳理论与触发器工作原理有机结合，建立乡 村旅游安全风险耦合机理模型，以期为开展乡村旅游 安全风险管理提供参考和借鉴。 1. 乡村旅游安全系统分析 借鉴区域环境风险系统理论 [1] ，结合乡村旅游安 全实际情况，乡村旅游安全系统可由风险源、控制机 制和受体三个部分构成，如图 1 所示。风险源即为可 能产生危害乡村旅游主体(乡村旅游者)、乡村旅游客 体(乡村旅游资源、乡村社区)及乡村旅游媒体(乡村旅 危险性 风险源 控制机制 受 体 乡村旅游安全 危 害 受 体 承 受 风 险 因 素 的 危 害 易 损 性 影 响 控 制 机 制 保 护 受 体 防范风险因素的危害 影响控制机制 有效性 安 全 性 脆 弱 性 图 1. 乡村旅游安全系统。 游从业者、乡村旅游交通)的源头，它是乡村旅游风 险事件发生的先决条件；控制机制包括对风险源的控 制设施设备、管理规章制度等的控制，它是保障乡村 旅游安全的有效途径和必要手段；受体则指风险承受 者，如乡村旅游主体、乡村旅游客体以及乡村旅游媒 体，它是乡村旅游安全的研究对象。构成乡村旅游安 全系统的三个子系统间存在着相互联系、相互影响关 系，其风险源的“危险性”、控制机制的“有效性”、 受体的“安全性、易损性、脆弱性”三者的时空交叉 点共同决定了系统的即时安全状态。 由图 1 可知：风险源的存在是引发乡村旅游安全 事故的前提，其危险性大小决定对受体的危害程度， 其危险性受控效度直接影响受体的安全状况，同时， 风险源的动态可变性要求控制机制要根据风险源的 变化情况及时做出调整；控制机制通过对风险源的有 效控制实现对受体的保护作用的同时，还需根据受体 的特点、发展及变化态势进行适当调整，以实现对风 险源的动态可靠控制；受体作为乡村旅游安全的研究 对象，其安全问题呈现出复杂性、动态性及不确定性 等特点，一旦受体遭受风险源的攻击，将发生乡村旅 游安全事故，造成一定的人员伤亡或财产损失。 2. 乡村旅游安全风险耦合分析 2.1 乡村旅游安全风险影响因素 乡村旅游安全 [2] 是指乡村旅游现象中一切安全 现象的总称，既包括乡村旅游主体安全，也包括乡村 旅游客体安全，还包括乡村旅游媒体安全。根据危险 源理论 [3] ，无论是乡村旅游主体、客体或是媒体，其 安全风险影响因素均可概括为人的因素、物的因素、 环境因素以及管理因素。借鉴波特钻石模型，可将影 响乡村旅游安全风险的因素描述为乡村旅游安全风 险因素钻石模型，如图 2 所示。由图 2 可知，人的不 乡村旅游 安全风险 物 人 环境 缺 失 不安全状态 不安全行为 不 确 定 管理 图 2. 乡村旅游安全风险因素钻石模型。 安全行为、物的不安全状态、管理缺失以及环境的不 确定性等是导致乡村旅游安全风险产生的诱发条件， 并且人的因素、物的因素、环境因素及管理因素间存 在着相互联系、相互作用、相互耦合关系，一起乡村 旅游安全事故的发生不仅仅是人、物、环境及管理中 某一单因素作用的结果，更多是各种风险因素非线性 耦合作用的结果。 2.2 乡村旅游安全风险耦合与耦合风险 根据耦合的内涵 [4] ，乡村旅游安全风险耦合是指 乡村旅游活动过程中不同风险因素或风险因子之间 的相互依赖、相互影响的关系与程度。于是，乡村旅 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 46 人-物-环耦合 风险(a,b,c) 人因耦合风险(a) 物因耦合风险(b) 环境因素耦合风险(c) 管理因素耦合风险(d) 人-物-管耦合 风险(a,b,d) 人-环-管耦合 风险(a,c,d) 物-环-管耦合 风险(b,c,d) 人-物-环-管耦合 风险(a,b,c,d) 人-物耦合风 险(a,b) 人-环耦合风 险(a,c) 人-管耦合风 险(a,d) 物-环耦合风 险(b,c) 物-管耦合风 险(b,d) 管-环耦合风 险(c,d) 双因素耦 合风险 三因素耦 合风险 单因素耦 合风险 乡 村 旅 游 地 系 统 耦 合 风 险 全因素耦 合风险 图 3. 乡村旅游安全事故耦合风险类型。 游安全风险耦合方式可以概况为单因素风险耦合、双 因素风险耦合、三因素风险耦合以及全因素风险耦合 等四种。其中，双因素风险耦合、三因素风险耦合与 全因素风险耦合统称多因素风险耦合。风险耦合作用 的结果使得乡村旅游安全系统产生新的风险，这种新 的风险称为耦合风险，耦合风险类型如图 3 所示。 单因素耦合风险是指乡村旅游安全系统中人、 物、环境及管理各风险因素自身所包含的风险因子间 相互作用、相互影响所引致的风险，包括人因耦合风 险、物因耦合风险、环境因素耦合风险以及管理因素 耦合风险。例如，在导游或领队未向游客告知旅游目 的地居民的风俗禁忌的情况下，游客违反旅游目的地 居民的风俗习惯而引发的主客冲突事故，就是典型的 人-人风险因素产生的耦合风险。 双因素耦合风险是指乡村旅游安全系统中人、 物、环境及管理风险因素两两相互作用、相互影响所 引致的风险，包括人-物耦合风险、人-环耦合风险、 人-管耦合风险、物-环耦合风险、物-管耦合风险和管 -环耦合风险。例如，2013 年 5 月 1 日，由于吊桥拉 钩、拉索、受力柱断裂，加之日常管理维护缺失，导 致凤凰吊桥发生倾斜事故，造成 37 名游客落水(其中 2 人轻伤)的后果，就是一起典型的物-管耦合风险引 发的旅游安全事故。 三因素耦合风险是指乡村旅游安全系统中人、 物、环境及管理风险因素三三相互作用、相互影响所 引致的风险，包括人-物-环耦合风险、人-物-管耦合 风险、人-环-管耦合风险和物-环-管耦合风险。例如， 由于驾驶员酒驾、游客安全意识不足、逃生门加暗锁、 拼装车横行市场、验车不规范、法律不健全等原因共 同作用，导致台湾旅游大巴事故频发，自 2008 年至 今已造成大陆游客意外死伤共计 298 人，就是典型的 人-物-管理风险因素相互耦合产生的耦合风险。 全因素耦合风险是指乡村旅游安全系统中人、 物、环境及管理风险因素相互作用、相互影响所引致 的风险，即人-物-环-管耦合风险。例如，“6·1 东方之 星旅游客船倾覆事件”造成 442 人遇难，就是一起典 型的人(船长和当班大副对恶劣天气及其风险认知不 足等)-物(客轮经历三次改建、改造和技术变更，风压 稳性衡准数逐次下降，虽然符合规范要求，但不足以 抵抗所遭遇的极端恶劣天气等)-环境(“飑线伴有下击 暴流”带来的强风暴雨等)-管理(恶劣天气预警机制缺 失、旅游客运公司安全管理不到位、应急管理建设薄 弱等)耦合风险引致事故。 3. 乡村旅游安全风险耦合机理 3.1 界壳理论及乡村旅游安全界壳 我国学者曹鸿兴根据系统与环境的作用关系，开 创性地提出了界壳理论 [5](periphery theory)，指出“界 壳是处在系统外围能护卫系统且与环境进行交换的中 介体，它是系统的周界，是系统的一部分，又和环境 相毗邻”，其结构如图 4 所示。其中，界壁(w)起到卫 护系统本身的功能，界门(p)通过输入/输出功能可以实 现系统与环境之间进行物质、能量和信息交换。界壳 理论作为系统理论的新成员，为研究包括气象、生态、 经济、农业、安全等系统周界一般规律提供了崭新的 思路与方法 [5] 。 近年来，伴随我国安全生产形势的复杂化、风险 防控难度的加大，一些学者探索性地将界壳理论与安 全风险相关问题结合，进行了诸多有益探索与创新。 陈勇刚等 [6] 对安全及安全管理界壳进行了初步界定； 黄强等 [7] 提出了系统周界的观控模型，为水资源多维 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 47 界门(p) 界门(p) 界壁(w) 界壁(w) 系里(i) 输入/输出 输入/输出 系里：乡村旅游主体(旅游者)； 界壁：旅游者安全指数、旅游从业者安全指数、基础设施、交通 工具、农事体验设施设备、安全标志、安全机构设置和人员配置、 安全生产规章制度、安全监督检查、安全预警机制、安全法律法 规、旅游保险、应急救援、安全投入、安全宣教； 界门：主客冲突潜在指数以及界壁中的薄弱环节； 环境：自然灾害、社会治安以及环境卫生等。 »· ¾³ 图 4. 乡村旅游主体安全事故风险因素管控界壳。 »· ¾³ 界门(p) 界门(p) 界壁(w) 界壁(w) 系里(i) 输入/输出 输入/输出 系里：乡村旅游客体(乡村旅游资源、乡村社区)； 界壁：基础设施、安全设施、资源保护法律法规、资源保护投入、资 源开发利用条件、资源保险、资源容量预警机制、资源保护教育和宣 传、政府监管、社区利益分配公平程度、社区医疗卫生设施、突发事 件应急能力； 界门：旅游者不文明行为、旅游者碳足迹、生活垃圾和旅游活动污染、 不当的旅游开发、建设性破坏、主客冲突潜在指数以及界壁中的薄弱 环节； 环境：自然生态环境、人文社会环境、社区治安环境、环境卫生等。 图 5. 乡村旅游客体安全事故风险因素管控界壳。 »· ¾³ 界门(p) 界门(p) 界壁(w) 界壁(w) 系里(i) 输入/输出 输入/输出 系里：乡村旅游媒体(乡村旅游从业者、乡村旅游交通)； 界壁：旅游从业者安全指数、旅游者安全指数、交通工具状况、 交通安全设施标识、交通安全管理状况、安全宣教、事故救援保 障； 界门：主客冲突潜在指数、驾驶员情况以及界壁中薄弱环节； 环境：旅游从业者所处环境、天气状况、道路状况等。 图 6. 乡村旅游媒体安全事故风险因素管控界壳。 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 48 临界调控风险分析提供了新途径；郑怀昌等 [8] 利 用界壳理论分析了矿山系统安全风险产生机理，并提 出防控策略；刘朝峰等 [9] 提出了地震灾害综合防御界 壳的协同演化测度模型。综上，界壳理论在提升系统 风险防控能力、保障系统安全上具有很好的适用性。 于是，在对乡村旅游安全系统及其安全风险影响因素 分析的基础上 [2] ，根据界壳理论，构建乡村旅游主体 安全事故风险因素管控界壳、乡村旅游客体安全事故 风险因素管控界壳以及乡村旅游媒体安全事故风险因 素管控界壳，如图4、5、6所示。 3.2 乡村旅游安全风险耦合机理 分析乡村旅游安全风险耦合机理，有助于厘清乡 村旅游安全各风险因素耦合作用路径，为降低风险、 提高防控能力提供科学参考，同时也是建立有效的乡 村旅游安全保障体系的必要前提。基于触发器工作原 理的风险形成机理模型 [4,10] 可为乡村旅游安全风险耦 合机理研究提供思路和借鉴，但该模型亦存在两点不 足:①理论基础薄弱；②模型中阀门和滤波器仅具有抵 抗(被动的防御)、自修复、自组织等有限功能，缺乏 主动防御功能、缺乏系统与外界环境进行物质、能量 和信息交换机制等。而界壳 [5-9] 具有坚实的理论基础、 完善的系统卫护、交换及抵抗等功能，但却不具备耦 合功能。为此，将界壳功能与触发器的耦合振荡器功 能进行有机结合，建立一种优势互补的乡村旅游安全 风险耦合机理模型，如图 7 所示。 主体安全事故风险因素 客体安全事故风险因素 媒体安全事故风险因素 主体安全事故风险因素管控界壳 客体安全事故风险因素管控界壳 媒体安全事故风险因素管控界壳 耦合 振荡器 耦合前 卫护 交换 抵抗 失败 零耦合 弱耦合 强耦合 风险不变 风险减小 风险增大 产生新风险因素 耦合后 管控界壳群 图 7. 乡村旅游安全风险耦合机理模型。 由图 7 可知：当乡村旅游主体、客体及媒体安全 事故风险因素管控界壳的卫护、交换及抵抗均成功 时，各风险因素不会通过管控界壳而进入耦合振荡器 进行风险耦合，乡村旅游安全风险处于可控状态，不 会引发乡村旅游安全事故；当乡村旅游主体、客体及 媒体安全事故风险因素管控界壳中任一管控界壳的 卫护、交换及抵抗失败时，其对应风险因素会通过管 控界壳群而进入耦合振荡器进行风险耦合，产生耦合 风险，此时乡村旅游安全风险处于不可控状态，将会 引发乡村旅游安全事故。根据耦合振荡器对风险的耦 合程度，耦合风险包括零耦合、弱耦合和强耦合三种 类型 [4] 。零耦合是指风险通过耦合振荡器之后，风险 保持不变；弱耦合是指风险通过耦合振荡器之后，风 险减小或降低；强耦合是指风险通过耦合振荡器之 后，风险增大或产生了新的不可预知的风险。按照系 统安全的观点，既然风险是绝对的，则防控乡村旅游 安全风险就应该做到“避免强耦合、追求零耦合或弱 耦合，将风险保持到可控程度”。 4. 结论 乡村旅游安全是一个复杂系统工程问题，无论是 其受体构成要素的乡村旅游主体、客体及媒体，或是 导致乡村旅游安全事故发生人、物、环境及管理等风 险引致因素，均呈现出复杂性、不确定性和可变性等 特点，这更增加了乡村旅游安全风险防控的难度。分 析乡村旅游安全风险耦合机理，可以从风险产生的源 头和作用路径上为风险管控工作提供参考依据。本文 在分析乡村旅游安全系统及其风险因素的基础上，对 乡村旅游安全耦合风险类型进行了具体划分，并依据 触发器原理、界壳理论提出了乡村旅游安全风险耦合 机理模型，得到如下结论： (1)乡村旅游安全系统是风险源、控制机制、受体 及三者相互作用关系的集合，其风险源的“危险性”、 控制机制的“有效性”、受体的“安全性、易损性、脆 弱性”三者的时空交叉点共同决定了乡村旅游安全系 统的即时安全状态。 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 49 (2)一起乡村旅游安全事故的发生不仅仅是人、 物、环境及管理中某一单因素作用的结果，更多是各 种风险因素非线性耦合作用的结果。 (3)乡村旅游安全风险耦合方式可以概括为单因 素风险耦合、双因素风险耦合、三因素风险耦合以及 全因素风险耦合等四种类型，乡村旅游安全风险的耦 合产生乡村旅游安全耦合风险。 (4)所建立乡村旅游安全风险耦合模型，以风险 管控界壳取代触发器的阀门与过滤器的功能，并保留 触发器耦合振荡器的耦合功能，做到优势互补，无论 是在理论支撑或是在对风险耦合作用路径描述上，更 为科学合理。同时，该风险耦合机理模型也为如何管 控乡村旅游安全风险、降低乡村旅游安全事故率提供 对策参考，即可通过构造乡村旅游主体、客体及媒体 安全事故风险因素管控界壳以防止风险溢出并进一 步耦合产生耦合风险，另外，在此基础上还可构造乡 村旅游主体、客体及媒体安全保障界壳以卫护乡村旅 游主体、客体及媒体的安全。 致谢 本 文 得 到 福 建 省 社 科 规 划 一 般 项 目 (fj2015b217) ；福建省中青年教师教育科研项目 (jas150095)；中央高校基本科研业务费资助项目·华 侨 大 学 哲 学 社 会 科 学 青 年 学 者 成 长 工 程 项 目 (14skgc-qg16)的资助，在此表示衷心感谢。 参考文献 1. 毕军,杨洁,李其亮.区域环境风险分析和管理.北京:中国 环境科学出版社,2006. 2. 罗景峰.乡村旅游安全影响因素辨识研究.安徽农业大学 学报(社会科学版),2016,26(4):30-34. 3. 隋鹏程,陈宝智,隋旭. 安全原理 .北京: 化学工业出版 社,2005. 4. 张津嘉,许开立,王贝贝,等.瓦斯爆炸事故风险耦合演化机 理研究.中国安全科学学报,2016,26(3):81-85. 5. 曹鸿兴,封国林,蔡秀华,等.界壳论精要及其应用.北京:科 学出版社,2011. 6. 陈勇刚,田水承,李红霞.界壳论与矿山安全管理界壳基础 研究.中国矿业,2003,12(11):18-20. 7. 黄强,李勋贵,leon feng,等.系统周界的观控模型及其应 用.系统工程理论与实践,2005,25(3):101-106. 8. 郑怀昌,李明.界壳理论在采空区失稳判定与危害控制研 究中的应用探讨.黄金,2005,26(12):19-22. 9. 刘朝峰,苏经宇,王威,等.地震灾害综合防御界壳的协同演 化测度模型.系统工程理论与实践,2014,34(8):2186-2192. 10. 冯艳飞,陈媛.企业战略风险的形成机理研究.武汉理工大 学学报(信息与管理工程版),2008,30(5):816-819. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 45–50 ___________________________________________________________________________________________________________ 50 1. 乡村旅游安全系统分析 2. 乡村旅游安全风险耦合分析 3. 乡村旅游安全风险耦合机理 4. 结论 致谢 参考文献 atlantis press journal style study on the relationship between financial development and poverty alleviation in guizhou province based on kuznets curve jianchun yang school of business administration, guizhou university of finance & economics, guiyang 550025 huaxi zone, guiyang, china e-mail: 772957289@qq.com chaochao liang school of business administration, guizhou university of finance & economics, guiyang 550025 huaxi zone, guiyang, china e-mail: 416216094@qq.com received may 1, 2018 accepted august 19, 2018 abstract based on the construction of the nonlinear effect model of financial poverty alleviation, the paper used granger causality test and regression analysis to analyze the relationship between financial development and poverty alleviation in guizhou province. the results present that the relationship between financial development and poverty alleviation follows the kuznets inverted u-curve, which means that financial development, may have positive effects on poverty alleviation in the early stage and then have negative effects on poverty alleviation in later stage. based on this, in-depth analysis was conducted and development suggestions were put forward. keywords: kuznets curve, financial development, poverty reduction 1. introduction guizhou is the province with the largest number of poor people, the widest area of poverty and the highest intensity of poverty in china. since the reform and opening up, guizhou province has made significant progress in poverty alleviation. the number of poor people in guizhou has decreased from 15.87 million in 1978 to 2.36 million in 2007. the number of poor people in guizhou province has shrunk from 6.26 million in 2008 to 4210,000 in 2010 measured by the low-income line rather than the absolute poverty line which was no longer be officially adopted in the measurement of poverty since 2008. the number of poor people in guizhou province has decreased from 11.49 million in 2011 to 4.93 million in 2015 since the country have adopted the poverty line of 2,300 yuan per person in 2011. and the poverty rate of guizhou has reduced from 59.1% in 1978 to 6.5% in 2007. according to the low-income line, which was officially adopted to replace the absolute poverty line to measure the poverty in 2008, poverty rate of guizhou province has decreased from 17.4% in 2008 to 12.1% in 2010; the poverty rate in guizhou province has decreased from 33.1% in 2011 to 14.03% in 2015 based on the poverty line of 2,300 yuan per person adopted since 2011. however, guizhou province is still the main battlefield of poverty alleviation. under the new situation, precise poverty alleviation of guizhou province still faces many bottlenecks and constraints, including the backward economic development, the 151 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 fragile ecological environment, low quality of the population, the lag of public service, especially the financing constraint problem. therefore, it is of great practical significance to study the relationship between financial development and poverty alleviation in guizhou province. 2. literature review for a long time, domestic and foreign scholars have done a lot of research on the relationship between financial development and poverty but achieved no agreement on that due to the inconsistent result on analysis of the model, the index selection, measurement error, the heterogeneity of the data, and the related explanation variable. after making systematic review of existing research, we have found three main ideas in terms of the relationship between financial development and poverty. the first thinking is that financial development is positively correlated with poverty alleviation. through conducting analysis, jalilian & kirkpatrick (2002), akhter et al (2010) found a positive correlation between financial development and poverty alleviation and believed that financial development was beneficial to poverty alleviation. the second view is that financial development is not positively correlated with poverty alleviation. fowowe & abidoye (2013) believed that financial development poverty may not help to reduce the poverty. by conducting dynamic panel data analysis, seven & coskun (2016) explored whether the development of banks and stock markets was helpful to alleviate income inequality and poverty in emerging countries. the research results showed that though the financial development promoted economic growth, but it did not serve poor people in emerging countries, both banks and the stock market did not play an important role in poverty reduction. the third view hold is that there is a nonlinear relationship between financial development and poverty alleviation. greenwood & jovanovic (2009) suggested that the relationship between financial development and income distribution was based on the kuznets "inverted u" curve, so financial development might have indirect impact on the poverty reduction by influencing income distribution directly, which meant that the relationship between financial development and poverty alleviation may be demonstrated by using kuznets "inverted u" curve. cui yanjuan (2014) selected financial development scale and financial efficiency as the index of financial development and adopted panel data to examine the impact of financial development on poverty reduction, verifying the prediction of greenwood & jovanovic. the diversification of the research result on the relationship between financial development and poverty alleviation reports that there is room for further study about the topic. from the perspective of research content, there are many researches on the causal relationship between financial development and poverty alleviation but less discussion on the mechanism of the impact of financial development on poverty alleviation; from the perspective of research methods, most researches are carried out under the framework of linear model, and limited ones study the non-linear relationship. in terms of the research object, the research on regional financial poverty alleviation is very limited while the object of many studies in the financial poverty alleviation always is the country. therefore, this paper will probe into the non-linear relationship between financial development and poverty alleviation in guizhou province. 3. model construction and variable description 3.1. model building to examine the non-linear effect between financial development and poverty alleviation, this paper constructs the following analysis model. tttt finfinpov   2 210 (1) ttttt crrjgdpfinfinpov   43 2 210 ln (2) model (1) examines the direct non-linear effect of financial development on poverty alleviation which is mainly determined by poor people’s access and accessibility of financial product services. model (2) examines the indirect non-linear effect of financial development on poverty alleviation. indirect effect of financial development on poverty alleviation means that the financial development may influence poverty alleviation through promoting economic growth or income distribution which is beneficial to alleviating poverty. pov, rjgdp and crpov respectively represent 152 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 poverty alleviation, economic growth, income distribution. fin is used to represent variable of financial development which is measured from two perspectives including financial development scale (fir) and the financial development efficiency (fe). ε refers to the error term; α and β are the estimated coefficients; t represents the time of period. if β1﹥0 andβ2﹤0 are workable at the same time, the relationship between financial development and poverty alleviation follows inverted ushape. if β1﹤0 and β2﹥0 are established at the same time, u-shape relationship between financial development and poverty alleviation is reported exist. if β2=0, linear relationship between financial development and poverty alleviation is demonstrated. 3.2. variable declaration  poverty alleviation (pov). indexes which were used to measure poverty mainly include poverty rate and poverty index, sen index, fgt index, etc. considering the availability of data as well as the long research time span, this paper adopts the engel coefficient of rural households as indexes measuring poverty level. the lower the engel coefficient of rural households is, the higher the degree of poverty alleviation is, and vice versa.  financial development (fin). this paper measured the financial development from two perspectives including the financial development efficiency (fe) and the financial development scale (fir). the ratio of loan/savings deposit is adopted to measure the efficiency of financial development (fe). we organized the data of the deposit and loan of financial institutions and the gdp of the region then compiled them into financial related rate to measure the scale of financial development.  economic growth (rjgdp). this paper selected per capita gdp to measure the economic development of guizhou province. in order to reduce the influence caused by unusually large fluctuation of variable data and possible heteroscedasticity, we conducted a logarithm of this index.  income distribution (cr). in terms of income distribution, existing research always adopted theil index, gini coefficient and per capita income rate of urban and rural residents to measure it. in this paper, the urban-rural gap was measured by the per capita disposable income of urban residents and the per capita net income of rural residents. 4. econometric analysis this paper used eviews software to carry out econometric analysis. the data used in this paper was from guizhou statistical yearbook, and the data sampling period is from 1978 to 2016. 4.1. stationarity test. we adopted augmented unit root (adf), which was proposed by dickey and fuller, to examine the stationarity of time series data. the unit root test results were demonstrated in table 1(see table 1) and they presented that pov, fir, fe, rjgdp, cr under the first order difference were passed the significance level of 1% adf unit root test, reporting that first order difference sequence of every variable was stationary series. table 1 variable unit root test results variable test type（c， t，l） adf test value critical values at each level of significance the inspection results 1％ 5% 10％ dpov （c，t，0） －6.0926 －4.2268 －3.5366 －3.2003 smooth dfir （c，t，3） －4.3777 －4.2529 －3.5486 －3.2070 smooth dfe （c，t，0） －6.1837 －4.2268 －3.5366 －3.2003 smooth drjgdp （c，0，0） －3.7146 －3.6210 －2.9434 －2.6103 smooth dcr （c，t，0） －4.5656 －4.2268 －3.5366 －3.2003 smooth note: c: includes intercept term; t: includes trend term; l: lag order number; d plus variable represents the first difference. 153 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 4.2. granger causality test granger causality test results (see table 2) demonstrated that guizhou financial development scale, financial development efficiency and economic growth was the granger cause of poverty alleviation, and there was two-way granger causality between financial development scale and economic growth. it revealed that the financial development of guizhou had not only the direct promotion effect on poverty alleviation, but also had the indirect promotion effect on poverty alleviation by promoting economic growth and reducing poverty. at the same time, the results showed that there was no granger causality of statistical significance between income distribution and poverty alleviation, reporting that at present stage in guizhou, the indirect promoting effects of financial development on poverty alleviation by improving income distribution was not obvious. table 2 granger causality test results lag order null hypothesis f statistic p values lags:2 fir does not granger cause pov 3.3025 0.0497 pov does not granger cause fir 0.3827 0.6854 lags: 2 fe does not granger cause pov 3.5692 0.0400 pov does not granger cause fe 0.5469 0.5841 lags: 1 lnjgdp does not granger cause pov 4.3345 0.0447 pov does not granger cause lnjgdp 2.0377 0.1623 lags: 1 cr does not granger cause pov 1.1291 0.3359 pov does not granger cause cr 0.4217 0.6595 lags: 2 fir does not granger cause lnjgdp 2.6948 0.0829 lnjgdp does not granger cause fir 7.3300 0.0024 4.3. regression analysis according to the model (1) and model (2), we conducted regression analysis by involving indexes including financial development efficiency (fe) and financial development scale (fir) to measure the level of financial development and test the non-linear effect of financial development on poverty reduction in guizhou province. regression results were reported in table 3. regression analysis results showed that in the regression model (1) to (4), the respective coefficients of financial development scale (fir) and the financial development efficiency (fe) are β1﹥0，β2﹤0, which presented that regardless of from perspective of direct or indirect promoting effects, the reversed u-shape relation between financial development and poverty alleviation in guizhou has been found, verifying greenwood & jovanovic's point of view. based on the regression model (1) and regression model (2), matlab software was used to draw figures describing the non-liner relationship between financial development scale (fir) and poor reduction and the one of financial development efficiency (fe) and poor reduction (pov). from figure 1 and figure 2, we can see more clearly that financial development may have positive effects on poverty alleviation in the early stage and then have negative effects on poverty alleviation later. 154 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 table 3 the effects of guizhou financial development on poverty alleviation: regression analysis (explained variable: pov) variable the regression model（1） the regression model（2） the regression model（3） the regression model（4） constant 0.6413 *** （2.8249） -1.4503 *** （-7.8414） 0.7674 *** （10.4536） -1.0104 *** （-3.3117） fir 0.1172 *** （17.8113） 0.2755 *** （2.7737） fir² -0.0608 *** （-5.7782） -0.0809 *** （-4.1021） fe 3.6469 *** （9.8761） 2.9913 *** （6.1417） fe² -1.5379 *** （-8.5945） -1.2588 *** （-5.7284） lnrjgdp -0.0344 ** （-2.0620） -0.0219 ** （-2.4372） cr -0.0223 （-1.0093） 0.0282 *** （2.8676） r² 0.8896 0.9153 0.9039 0.9329 note: r² = goodness of fit; ***p<.01 and **p<.05; and the t-statistic of the estimated values of each parameter was presented in the bracket. fig. 1 "inverted u" diagram of pov and fir fig. 2 "inverted u" diagram of pov and fe 5. conclusions and discussions 5.1 conclusions the following conclusions can be obtained through metering analysis: (1) the non-linear relationship between the financial development and poverty alleviation in guizhou province follows the kuznets' inverted u-shaped curve. due to the low level of economic development, high poverty level and high poverty rate, guizhou province has been in the "poverty trap". thus, in the early stage of economic development, it is difficult for poor people to make benefit from financial development since they are limited by the high threshold of financial services. with the development of finance and economy in guizhou province, the threshold of financial services has been gradually lower, thus the financial development may gradually be conducive to the alleviation of poverty. (2) at present, the financial development of guizhou province plays an active role in promoting poverty alleviation. as the core of modern economy, finance plays an important role in capital formation. 155 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 finance acts as a mediator which can promote the transformation of savings to investment and increase the capital accumulation, thus prompting economy out of the "poverty trap" and reaching a higher level of equilibrium. 5.2 theoretical contributions based on the related index and data of financial development, poverty alleviation, economic growth and income distribution in guizhou province, this paper verified the direct and indirect non-linear relationship between financial development and poverty alleviation which follow the kuznets' inverted u-shaped curve. 5.3 suggestions to sum up, on the one hand, guizhou province should expand the financing scale for poverty-stricken areas and poor groups in the financial poverty alleviation work. in the development of modern economy and society, capital is a middle stage and one of key factors. for poverty-stricken areas, the need for fund is even more urgent. therefore, it is necessary to increase loans for poor areas and poor groups, encourage and guide various financial institutions to expand the agricultural loans amount in poverty-stricken areas, and promote the reduction of social financing costs. on the other hand, the degree of precision of financial poverty alleviation may be strengthening. the fund for poverty alleviation may be uniformly arranged and government should make target investment according to the cause and the type of poverty. government may concentrate to promoting infrastructure construction, investment for basic education and economic restructuring. acknowledgements the project is supported by the ministry of education research in the humanities and social sciences planning fund (no. 16yjazh069). references chengwei huang, tao ye. provincial model of poverty alleviation: a study on the precise poverty alleviation model of guizhou's precise poverty alleviation[m]. beijing: socialscience literature press, 2016. akhter s, liu y, daly k. cross country evidence on the linkages between financial development and poverty[j]. international journal of business and management, 2009,5(11):207-214. jalilian h, kirkpatrick. growth and poverty reduction in developing countries[j]. international journal of finance and economics, 2002,7(2):97-108. fowowe b, abidoye b. the effect of financial development on poverty and inequality in african countries[j]. manchester school, 2013,81(4):562-585. seven u, coskun y. does financial development reduce income inequality and poverty? evidence from emerging countries[j]. emerging markets review, 2016,26:34-63. jeremy greenwood, boyan jovanovic. financial development, growth, and distribution of income[j]. journal of political economy, 1990,98(5):1076-1107. yanjuan cui. financial development and poverty alleviation: path, effect and policy implications[m]. beijing: economic science press, 2014. jianchun yang, rui shi. comparison of dynamic effects of financial support for tourism industry development -taking guizhou and zhejiang provinces as examples[j]. social scientists, 2014,(6):88-92. rongrong shi et al. the threshold effect of financial poverty reduction and its empirical test -based on the research on the data of china's western provinces[j]. china soft science, 2013,(3):32-41. 156 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 151-156 atlantis press journal style fuzzy multicriteria risk assessment for hazardous waste management: the case of istanbul seda uğurlu istanbul technical university, industrial engineering department, 34367 macka, istanbul turkey sedayanik@itu.edu.tr cengiz kahraman * istanbul technical university, industrial engineering department, 34367 macka, istanbul turkey kahramanc@itu.edu.tr abstract a hazardous waste is waste that poses substantial or potential threats to public health or the environment. hazardous waste management is an area of the waste management field concerned with the proper containment, management, and disposal of wastes which could be considered hazardous. in this paper, we use a multicriteria evaluation method, namely vikor, for various hazardous waste treatment alternatives under fuzzy environment. these alternatives are physical treatment processes, chemical treatment processes, thermal treatment processes, and biological treatment processes. under the main criteria, economic, risks, and technical, we consider 11 sub-criteria. the fuzzy set theory let us use linguistic evaluations for the alternatives. keywords: fuzzy sets, vikor, multicriteria evaluation, waste management, hazardous waste, risk assessment * corresponding author. journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 29 javascript:popup_imp('/imp/compose.php',700,650,'to=sedayanik%40itu.edu.tr'); mailto:kahramanc@itu.edu.tr 1. introduction hazardous waste has a chemical composition or other properties that have to be managed in order to prevent its release into the environment that can result in illness, death or other harm to living organisms including humans. the word “waste” means that the substances of concern have no apparent value. and a “hazard” is expressed as the potential of an unacceptable outcome 1 . the identification of a waste as hazardous waste is the process to determine whether a waste poses a sufficient chemical or physical hazard to merit regulation. at first glance, one would imagine that distinguishing between hazardous and nonhazardous wastes is a simple matter of chemical and toxicological analysis. other factors must be considered, however, before evaluating the actual hazard posed by a waste’s chemical composition 2 . the hazardous waste has one or more hazard characteristics given in table 1. the "hazard" or toxicity (hazards to human health) of the waste is the most important factor in determining how a waste should be managed. hazardous waste is required to undergo a proper treatment to destroy or render them environmentally acceptable. handling and disposal of hazardous waste and attending requirements have been substantially expanded and subject not only to technology needs but to the hazardous waste regulatory environment. hazardous waste management alternatives such as landfilling of wastes are discouraged with the recent regulations whereas new alternative technologies are seek to destroy, stabilize, or treat hazardous wastes by changing their chemical, biological or physical characteristics. the environmental and public health concerns increasingly stringent regulatory climate and encourage the hazardous waste management as a cleanup business 3 . a hazardous waste management scheme must be justified in terms of scientific evidence, engineering designs and processes, technologic practicality, economic realities, ethical considerations, and local, state, and national regulations 1 . the regulations generally provide a control on the risks associated with hazardous waste which is defined as the likelihood that harm will occur when a receptor (e.g., human or a part of an ecosystem) is exposed to that hazard. hazardous waste risk range from toxic risks which the probability that a certain population will have an incidence of a particular illness to other risks such as ecosystem stress, loss of important habitats, decreases in the size of the population of sensitive species, public safety and public welfare hazards. the wide range of risks encountered requires specific solutions often at high costs. facilities that generate, transport, transfer, treat or dispose of hazardous wastes require substantial construction, operating costs and have become extremely expensive and complex. there are usually a range of treatment alternatives for various types of hazardous wastes and their components dependent upon physical and/or chemical properties 3 . a challenge of hazardous waste management is the selection of treatment alternative table 1. four types of characteristics of hazardous wastes. hazard criteria physical/chemical classes corrosivity able to destroy tissue by chemical reactions. acids, bases, and salts of strong acids and strong bases. they have a ph less than 2.0 or higher than 12.5. (e.g. rust removers, waste acid, alkaline cleaning fluids, and waste battery fluids) ignitability able to oxidize by burning. any substance that spontaneously combusts at 54.3 o c in air or at any temperature in water, or any strong oxidizer. (e.g. paint and coating wastes, some degreasers, and other solvents) reactivity able to react, detonate, or decompose explosively at environmental temperatures and pressures. a reaction usually requires a strong initiator such as an explosive, confined heat or explosive reactions with water. (e.g. wastes from cyanide-based plating operations, bleaches, waste oxidizers, and waste explosives) toxicity causing harm to organisms soon after or after a long period of exposure. toxic chemicals include pesticides, heavy metals, and mobile or volatile compounds that migrate readily. published by atlantis press copyright: the authors 30 seda uğurlu, cengiz kahraman considering the technical, economical and risk associated aspects. the aspects of the problem are usually represented in the form of multiple criteria which often express tradeoffs between the management objectives. in order to raise awareness of the tradeoffs between the objectives and compare the differences among the alternatives in an optimizing framework, hazardous waste management decision making may depend upon multiple criteria decision making (mcdm) models which promote participation and synthesis of a wide variety of information. hazardous waste management is a complex issue not only because of its broad scope but also because of the wide range of attributes to be assessed. specifically, risk assessment in hazardous waste management and the comparison of the estimated relative risks or impacts associated with each management alternative relies on attributes and parameters difficult to define with certainty. the components of the risk assessment and decision making process involve both quantitative and qualitative factors. these difficulties constitute a method such as fuzzy logic to be used for integrating the risk levels, uncertainty and valuation mathematically in decision making of hazardous waste management. in this paper, we aim to present a fuzzy vikor methodology to make a multicriteria selection among the treatment alternatives of hazardous waste management. the proposed methodology is applied to a problem case of istanbul to demonstrate the potential of the methodology. the rest of this paper is organized as follows. the regulations of hazardous waste management are reviewed in section 2. a summary of the literature on multicriteria decision making for hazardous waste management is given in section 3. section 4 includes a brief description of the treatment methods used in hazardous waste management. then the criteria for the selection of the treatment method are explained in section 5. the fuzzy vikor multicriteria decision making methodology is introduced in section 6. an application is presented in section 7. the obtained results and future research directions are discussed in section 8. 2. hazardous waste management the waste amount has increased dramatically after the industrial revolution with the growing consumption of goods. chemical accidents, spills of hazardous waste or illegal midnight dumping that contaminates property are familiar. yet, even when hazardous waste is managed or disposed of in a careful manner, it may still pose a serious threat to public health and the environment. for example, toxic hazardous wastes can leak from a poorly constructed or improperly maintained hazardous waste landfill. such waste contamination can pollute groundwater, the primary source of drinking water for half the nation 2 . in order to prevent from these severe and irreversible results of the mismanagement of hazardous waste, nations have developed comprehensive regulations and standards. the recognition of the need for regulations is not very old and the first release of the acts of hazardous waste management differs among nations. the first national directives were enacted in 1972 at germany, in 1976 at usa and in 1995 at turkey. the common aim of these acts is to create a framework for the management of hazardous waste safely from the moment it is generated; while it is transported, treated, or stored; until the moment it is disposed. the management systems of hazardous waste regulated by the acts mainly establish requirements for the following 2 :  procedures for hazardous waste identification and classification.  provisions for facilitating hazardous waste recycling and universal wastes safely  standards for hazardous waste generating facilities  regulations for waste reduction and minimization  regulations for hazardous waste transporters  standards and provisions for treatment, storage, and disposal facilities and units  restrictions on land disposal  standards on units conducting combustion of hazardous waste.  designing corrective action against any spills or releases into the environment  permitting, inspection and enforcement based on these regulations as listed above, hazardous waste management comprises a diverse range of activities including identification, reduction, recycling, segregation, modification, treatment and disposal. these activities can be classified in two folds. one fold is the activities aiming to produce less amount of waste such as published by atlantis press copyright: the authors 31 fuzzy multicriteria risk assessment reduction and recycling. yet, an amount of waste is generated that need to be dealt with the second fold of the activities, namely treatment or disposal. a general overview of the hazardous waste management is illustrated in figure 1. fig.1. hazardous waste management 4 . historically, disposal was presented as a form of treatment on the grounds that after disposal the characteristics of deposited wastes frequently changed as a result of degradation, a phenomenon that greatly increases the potential of hazards on human health and environmental pollution. thus, disposal methods such as land filling, dumping or spilling have been restricted with additional precautions designed to protect soil, water, and air resources. treatment of the hazardous waste is the ultimate aim to remove or modify the characteristics of a substance that render it hazardous and perhaps add a value to the waste. a challenge is to identify the hazardous waste treatment options and select the best option in view of the risks, effectiveness and costs associated these options. the hazardous waste risk management framework is represented as in figure 2. 3. literature review there has been a growing body of literature in the area of risk-based decision analysis on hazardous waste. combining the risk analysis with the variety of other aspects of hazardous waste management, many studies make use of decision analysis methodologies by which the technical issues and non-technical issues such as cost and effectiveness are considered. fig. 2. hazardous waste risk management framework a systematic and consistent decision making approach to dredging and disposal including contaminated sediment management have been developed in ref. 5. they have identified and evaluated environmentally acceptable and cost-effective disposal alternatives within an effective decision making framework for costs, risk reduction and potential beneficial uses of the disposal material. they have considered tree alternatives, deepwater confined disposal, near shore fill or capping and, upland disposal in the presented hypothetical disposal scenario and adopted a risk-cost trade off approach in making decisions. their approach entails the performance of sequential evaluations consisting of risk analysis, estimation of costs, integration of the results into a computational framework for trade-off analysis, and the application of decision analytical tools to build consensus among stakeholders and the general public in selecting a preferred alternative. the decision problems related to hazardous waste management system have been investigated by many approaches including mathematical models in the search for an optimal strategy. a multi-objective integer programming approach to select hazardous waste hazardous waste management options hazardous waste characterization cost& effectiveness risk characterization risk management decision published by atlantis press copyright: the authors 32 seda uğurlu, cengiz kahraman treatment and disposal facilities, and transportation routes have been employed in ref. 6 they searched the best alternatives for two objectives; minimization of total cost, which includes treatment and disposal costs (sum of the capital cost, operation and maintenance cost of the treatment and disposal facilities), and transportation cost as well as minimization of total risk, which includes waste treatment and disposal risk as well as risk involved in waste transportation. they aimed to find the optimal location of treatment and disposal facilities, waste quantities traveling on the transportation links and waste quantities being processed at the treatment or disposal facilities under the constraints such as waste-technology compatibility, allowable capacities for treatment and disposal technologies at specific sites, etc. finally, they presented an example problem to perform sensitivity analysis in order to study the effect of change in the preferences of the objectives in the utility function and the weightings given to the objectives. radioactive waste attracts a special consideration in hazardous waste literature. in this research stream, ref. 7 proposed a decision support system for the identification of optimal remedial strategies to restore water systems after accidental introduction of radioactive substances. this decision support system includes an evaluation module based on a multiattribute value model to rank alternatives and a module to perform multi-parametric sensitivity analyses, both with respect to weights and values, to allow us to gain insights into the problem. ref. 8 used this decision support system in their study searching for optimum remedial strategies for contaminated lakes. one can deduce from ref. 9 the use of multicriteria decision analysis with an outranking methodology, promethee. this tool allowed them to incorporate stakeholder values into the decision process for soliciting public participation and analyze technological hazardous waste management alternatives with the option to introduce new alternatives at any point during the analysis. the use of promethee was chosen also due its capability of handling semiquantitative scales (e.g., high, middle, low) compared to other optimization methods such as multiattribute utility theory (maut) or analytical process hierarchy (ahp). a review of existing decision-making approaches at hazardous waste management regulatory agencies in the united states and europe is presented in ref.10. the study synthesizes state-of-the-art research in multi criteria decision analysis (mcda) methods in this area. they also tested an mcda approach, promethee, for coupling expert judgment and stakeholder values in a hypothetical contaminated sediments management case study. the hazards of contaminated dredged material disposal and the associated risks and costs which are highly uncertain have been incorporated in the decision analysis using fuzzy set theory 11 . by this means, these uncertainties and the risk-cost trade-offs are explicitly considered selecting an appropriate management alternative in the proposed multicriteria decision framework. a crisp and a fuzzy approach which are applied in dam safety and nuclear industries for risk-based decision analysis are investigated in ref. 12. one of the approaches is the structured explicit decision analysis, in the case of dam safety, which develops consistency to a complex decision problem through simplification. the major decision objectives and goals, which are summarized in regulatory compliance, risk control, and cost-effectiveness, are used to structure the process. the structured process needs to: (a) satisfy regulatory standards, (b) identify actions and measures to control life-loss and economic risks, and (c) evaluate costs and cost-effectiveness to justify, rank, select, and prioritize actions. the second approach which is used in the case of the utility and nuclear industries is the implicit multicriteria decision analysis. the proposed decision process requires the identification of the decision criteria relevant to the problem in hand, and the specification of the relative importance of the criteria by means of value eliciting from managers or stakeholders. after that, decision alternatives are scored against the decision criteria and then ranked based on the multiattribute theory. this approach commonly employs ranking schemes as fuzzy logic and heuristic methods. one can infer from ref. 13 the use of the approaches investigated in their former paper for the management of contaminated ground water resources problem. they used health risk assessment and economic analysis through a multi-criteria decision analysis framework to select the best remedial alternative. the decision criteria set included maximum individual risk, expected individual risk, population risk, risk index, and cost per cancer case avoided or cost per life saved. three potential approaches were investigated for alternative published by atlantis press copyright: the authors 33 fuzzy multicriteria risk assessment ranking, a structured explicit decision analysis, a heuristic approach of importance of the order of criteria, and a fuzzy logic approach based on fuzzy dominance and similarity analysis. 4. hazardous waste treatment methods the objective of waste treatment is to modify the physical and/or chemical properties of the waste to make it suitable for safe disposal 4 . treatment methods are often used prior to ultimate disposal of the hazardous wastes. the treatment technologies convert the waste into a more innocuous form, or immobilize toxic components so that it will not migrate and present a hazard when released into the environment. treatment technologies are characterized by their effectiveness to treat specific waste types, their waste residue generation factor, the cost and the risk considerations associated with them 14 . the hazardous waste treatment processes may be classified in four main groups: (i) physical treatment processes (ii) chemical treatment processes (iii) thermal treatment processes (iv) biological treatment processes various methods are used under these three groups of treatment processes. some of them are listed in table 2. by physical treatment methods, a waste is often separated into its constituents. the principle of separation is related to a physical property such as particle size or relative density. the physical separation processes are often applied for the primary purpose of isolating components of a waste for recovery. in the context of final disposal, separation is useful to reduce the quantity or hazardousness of the residue and simplify the process. some of the separation methods such as filtration, centrifuging, sedimentation, etc. depend on different physical principles or different phase of wastes such as solid waste or liquid waste. encapsulation is also a physical treatment method by which hazardous wastes are converted into a form designed to prevent or significantly reduce release into the environment. hazardous materials are immobilized by stabilization and incorporation within a solid matrix such as cement concrete or proprietary organic polymers. it is commonly used prior to land filling. by chemical treatment methods, the chemical structure of the hazardous waste material is converted into substances which have no or less hazardous properties. the process may be described as the alteration of the chemical nature/structure of the hazardous constituents of a waste using chemical reactions. some of the chemical treatment processes and the common elements used in these processes are illustrated as follows: neutralization occurs by the reaction of the waste acid with an alkali, oxidation is started with common oxidizing substances such as hydrogen peroxide or calcium hypochlorite, reduction is used to convert inorganic substances to a less mobile and toxic form by the use of chemicals, hydrolysis is the decomposition of table 2. hazardous waste treatment methods 4. physical treatment processes chemical treatment processes biological treatment processes gas cleaning liquid solids separation removal of scientific components mechanical collection centrifugation adsorption adsorption activated sludge electrostatic precipitation clarification crystallization chemical oxidation aerobic lagoons fabric filter coagulation dialysis chemical precipitation spray irrigation wet scrubbing filtration distillation chemical reduction trickling filters activated carbon adsorption flocculation electrodialysis combination and addition waste stabilization ponds adsorption floating evaporation ion exchange foaming leaching neutralization sedimentation reverse osmosis pyrolysis thickening solvent extraction stripping published by atlantis press copyright: the authors 34 seda uğurlu, cengiz kahraman hazardous organic substances typically (e.g. decomposing certain pesticides with sodium hydroxide), ion exchange removes dissolved salts primarily inorganics from aqueous solutions, precipitation is mainly used to convert hazardous heavy metals to a less mobile, insoluble form. a commonly-used treatment method which may also be classified as a chemical process is combustion. thermal treatment which is generally named as combustion is the reaction of hazardous wastes with oxygen sometimes in the presence of an external energy source. if the waste material is completely organic in structure, it is, in theory, completely destructible using principles based in thermodynamics. energy may also be produced during the reaction and the heat may be recovered. however, in combustion potentially more risky off-gases containing chlorinated hydrocarbons and/or ashes containing heavy metals are produced 1 . by biological treatment processes, certain compounds of waste substances are broken down by miocroorganism based activities through metabolism to provide nutrient energy or through co-metabolism. effective processing depends on selecting a microorganism suitable for the species requiring treatment, and on carrying out the treatment under conditions which are suitable, including careful control of the nature and composition of the waste. generally, biological processes destruct the organic wastes in theory and have little effect on inorganic wastes including metals, although some absorption into a sludge phase may occur with some processes. another issue is that biological processes become less efficient as concentrations of treatable constituents diminish 15 . during all treatment processes, the input waste must be monitored and possibly controlled also to maintain environmental conditions that do not upset or destroy the microorganisms in the system. some of the monitoring and control requirements for each of the systems include temperature, ph, oxygen availability, additional food sources, etc. 1 . the various biological processes fall into two main categories; aerobic processes in which the presence of oxygen is necessary, and those operating in oxygen free conditions, known as anaerobic processes 15 . 5. risk assessment criteria it is essential to identify the decision objectives and the set of comprehensive and non-redundant criteria corresponding to the objectives in decision analysis. the immediate objectives in hazardous waste management typically include reducing the risks and minimizing the costs. technical efficiency may also be included in the objectives with an attention on the hazardous waste treatments. the criteria corresponding to the objectives have been specified as the capital costs, operations and maintenance costs, land costs, waste-treatment technology compatibility, efficiency of offsetting hazard potential, pollutant removal efficiency, human health risks, ecological risks and transportation risks. capital costs: this criterion includes the economical amount of the first investment for the facility and the technology. operation and maintenance costs: this criterion represents the total operation and annual costs per treated waste. land costs: this criterion takes into account the economic value of the land needed for the hazardous waste management system based on the size of the area and the placement of the land with respect to its distance to the city center. waste-treatment technology compatibility: this criterion refers to the consideration of compatibility between the technology and different waste types. efficiency of offsetting hazard potential: this criterion includes the degree waste management system eliminates the hazardousness of the waste. waste residue: this criterion represents the amount of waste generated as a result of the waste treatment. resource recovery: this criterion stands for the amount of value (e.g. output energy, recovered waste, etc.) that can be recovered from the waste. capacity: this criterion refers to amount of waste which the hazardous waste management system is able to treat. human health risks: this criterion consists of risks associated with the hazardous substances such as emission levels and heavy metal released to air, water or land affecting human health. ecological risks: this criterion infers to the risks associated with the hazardous substances such as emission levels and heavy metal released to air, water or land affecting the ecosystem and causing loss of important habitats, decreases in the size of the population of sensitive species. transportation risk: this criterion covers the total transport risk imposed on the public and the published by atlantis press copyright: the authors 35 fuzzy multicriteria risk assessment environment by the movement of the hazardous waste between locations such as generation, collection, storage, treatment and disposal locations. 6. fuzzy multicriteria assessment vikor initiated by ref.16, of which the compromise solution should have a maximum group utility (majority rule) and minimum individual regret of the opponent, is proposed to deal with multicriteria decision-making problems. vikor method is based on the compromise programming of mcdm. the method works on the principle that each alternative is evaluated by each criterion function; the compromise ranking is presented by comparing the degree of closeness to the ideal alternative 17 . in fuzzy vikor, it is suggested that decision makers use linguistic variables to evaluate the ratings of alternatives with respect to the criteria. fuzzy set theory uses approximate information and uncertainty to generate decisions. this makes it favorable for human reasoning, intuition, judgment, perception and preference which are generally vague and difficult to measure. since knowledge about real-world problems, which has imprecision in the variables and parameters, can be expressed by fuzzy sets, many engineering and decision problems have been simplified and processed using fuzzy techniques. a fuzzy approach has been applied to classical vikor to capture the imprecision in the evaluations of the decision makers. the use of fuzzy vikor is summarized in steps as follows 18,19 : step 1: a group of decision-makers (denoted with k) identifies the evaluation criteria (denoted with n) and the alternatives (denoted with m). step 2: appropriate linguistic variables for the weights of the criteria and alternative ratings are chosen. examples of linguistic variables corresponding to fuzzy scores are given in table 3 and 4. these triangular fuzzy numbers with their membership values are illustrated in figures 3 and 4. table 3. fuzzy evaluation scores for the weights linguistic variable fuzzy score very low (0.0 , 0.1 , 0.2) low (0.1 , 0.2 , 0.4) medium (0.3 , 0.5 , 0.7) high (0.6 , 0.8 , 0.9) very high (0.8 , 0.9 , 1.0) fig.3. the triangular fuzzy weights. table 4. fuzzy scores for the alternative ratings. linguistic variable fuzzy score very poor (0 , 0 , 1) poor (0 , 1 , 3) medium poor (1 , 3 , 5) fair (3 , 5 , 7) medium good (5 , 7 , 9) good (7 , 9 , 10) very good (9, 10 , 10) fig. 4. the triangular fuzzy scores for alternative ratings. step 3: the linguistic evaluations of the decision makers are collected to calculate the aggregated fuzzy weights of the criteria, and aggregated fuzzy rating of alternatives.  k jjjj www k w ~...~~ 1~ 21  0,00 0,20 0,40 0,60 0,80 1,00 0,00 0,25 0,50 0,75 1,00 very low low medium high very high 0,00 0,20 0,40 0,60 0,80 1,00 0,00 0,25 0,50 0,75 1,00 very poor poor medium poor fair medium good good very good (1) published by atlantis press copyright: the authors 36 seda uğurlu, cengiz kahraman step 4: fuzzy decision matrix is constructed for the implementation of vikor. a fuzzy multicriteria decision making problem can be expressed in matrix format as follows:   n wwww ~,,~,~ ~ 21  step 5: fuzzy best value (fbv; *~ j f ) and fuzzy worst value (fwv; j f ~ ) of each criterion function are determined. step 6: separation measures i s ~ and i r ~ are calculated. step 7: i q ~ values are calculated. such that, , ii rr ~ min ~*  , ii rr ~ max ~   *~ s is the maximum majority rule, and *~ r is the minimum individual regret of an opponent strategy. as well, v is introduced as weight of the strategy of the maximum group utility, usually set as v = 0.5. step 8: i q ~ values are defuzzified and the alternatives are ranked by the index i q ~ . various defuzzification strategies have been suggested in the literature. here we have chosen to use the graded mean integration approach is used 20 . according to the graded mean integration approach, for triangular fuzzy numbers, a fuzzy number c ~ = (c1, c2, c3) can be transformed into a crisp number by employing the below equation: step 9: the best alternative is determined as the one with the minimum i q . 7. application the city of istanbul and the area close to istanbul is the most industrialized area in turkey. as expected, this area also serves as an inhabitation area for the highest portion of the population in turkey. thus, the hazardous waste resulting from the industry and high population density make up a considerably high amount. however, the facilities for the treatment of hazardous waste are very limited or almost not available in this area. the limited number of hazardous waste treatment facilities is not enough to treat the amount and the types of the hazardous waste generated presently. in turkey, the industrial waste amount was stated to be 12,5 million tons, of which 1,14 million tons were in the category of hazardous waste in 2008 21 . the hazardous waste amount for the european part of istanbul and three neighboring cities, tekirdag, edirne and kirklareli was estimated to be 160.000 tons/year in 2007. this amount has been envisioned to increase to 180.000 tons/year in 2010 by ministry of environment in the action plan for 2008-2012. the distribution of the hazardous waste types for the european part of istanbul and the cities tekirdag, kırklareli and edirne is given in table 5. the hazardous waste generated in the asian part of istanbul and the neighboring city izmit which is also a highly industrialized area is mainly treated in the treatment facility located in izmit. this facility treats the hazardous waste by combustion with a burning capacity of 35.000 tons/year and storage capacity of 160.000 tons/year. in spite of the high amount of hazardous waste generated in istanbul, the capacity and the types of treatment technologies available in this area are very limited and scarce. hazardous waste treatment facilities are required in the areas such as istanbul which are  k ijijijij xxx k x ~...~~ 1~ 21               mnmm n xxx xxx xxx d ~~~ . ~~~ ~~~ ~ 11 212221 11211     bjxf ijij  ,~max ~* cjxf ijij  ,~min ~ ) ~~ (/)~ ~ (~ ~ * 1 * jj n j ijjji ffxfws     )~~/()~~(~max~ ** jjijjjji ffxfwr  ) ~~ /() ~~ )(1() ~~ /() ~~ ( ~ **** rrrrvssssvq iii   ii ss ~ min ~*  ii ss ~ max ~   6 4 321 ccc c   (2) (3) (4) (5) (6) published by atlantis press copyright: the authors 37 fuzzy multicriteria risk assessment highly industrialized with high density of population for both the minimization of the costs and stress on human health and ecology. there has been considerable attention on the need for additional treatment facilities and plans have been generated for fulfilling this need. however, it is important to decide which treatment technologies is adequate based on the waste types and quantities generated, the risks associated with these waste and the economical perspectives of the treatment technologies. we propose to use a mcdm, fuzzy vikor as presented in section 6, for the decision making of treatment methods for istanbul area. the treatment method alternatives for istanbul are identified as the ones explained in section 4. these alternatives include:  physical treatment (t1)  chemical treatment (t2)  thermal treatment (combustion) (t3)  biological treatment (t4)  no treatment (direct disposal, e.g. land filling) (t5) the objectives of the decision making for treatment methods are identified as minimization of costs and risks and maximization of technical efficiency. thus, the criteria corresponding to the economic objectives are determined as capital costs (c1), operation and maintenance costs (c2) and land costs (c3). the risk criteria are specified as human health risks (c4), ecological risks (c5) and transportation risks (c6). the technical criteria are represented by wastetreatment technology compatibility (c7), efficiency of offsetting hazard potential (c8), waste residue (c9), resource recovery (c10) and capacity (c11). these criteria were explained in detail in section 5 and listed in table 6. table 5. the distribution of the hazardous waste in the european part of istanbul, tekirdag, kırklareli, edirne cities amount (tons/year) waste types tekirdag kırklareli edirne istanbul total (4 cities) total (turkey) pesticide and agrochemical waste 58 45 34 582 719 5926 wood protectors 24 0 5 232 261 1780 leather industry waste 2884 312 552 9619 13367 31739 oil refinery waste 361 0 19 3033 3413 65631 acids and alkali waste 1970 75 1289 14538 17872 62477 hazardous inorganic chemical waste 1071 15 903 15003 16992 48646 hazardous organic chemical waste 7364 229 4380 54064 66037 212870 dye and glue production waste 1381 100 216 15583 17280 54719 hazardous waste of printing processes 144 5 26 2543 2718 4908 energy production hazardous waste 1568 0 3419 392 5379 32531 metal production hazardous waste 4221 3 741 28424 33389 114586 mineral and glass production hazardous waste 304 76 375 2157 2912 18983 hazardous waste of galvanizing processes 952 21 77 13858 14908 44763 non-halogen waste oils 3207 721 880 49758 54566 208344 waste oils with halogen 42 6 19 507 574 2466 waste oils with emulsion 967 279 291 17876 19413 80075 other waste oils 708 139 170 12331 13348 53266 solvent waste with halogen 222 20 30 3312 3584 12228 solvent waste without halogen 752 65 110 10211 11138 36059 contaminated packaging waste 1177 115 546 11689 13527 48235 used filters and absorbands 646 135 269 7319 8369 32859 used oil filters 35 10 12 572 629 2507 used brake oils and antifreeze 36 28 21 784 869 3971 waste batteries 471 324 262 9282 10339 47613 used catalysts 3 0 1 20 24 180 physiochemical treatment sludge 1434 63 177 26276 27950 99433 contaminated mercury waste 64 14 19 512 609 2398 contaminated wood 219 119 100 3772 4210 21176 total 32285 2919 14943 314249 364396 1350369 published by atlantis press copyright: the authors 38 seda uğurlu, cengiz kahraman after determining the evaluation criteria and the alternatives, the steps of the fuzzy vikor are employed. evaluations of three decision maker are used in the analysis. in order to determine the importance of each treatment decision criterion and the ratings of the alternatives with respect to the criteria, decision makers used the five point scale and the seven point scale given respectively in table 2 and 3. table 6. list of evaluation criteria. aspects criteria economic capital costs operation and maintenance costs land costs risks human health risks ecological risks transportation risks technical waste-treatment technology compatibility efficiency of offsetting hazard potential waste residue resource recovery capacity the evaluations of decision makers for the weight importance and the overall fuzzy weights of each criterion of which are calculated according to eq. (1) are given in table 7. table 7. evaluations of decision makers for criteria and the overall fuzzy weight of each criterion. dm1 dm2 dm3 overall weight c1 l vl m (0.13 , 0.26 , 0.43) c2 l l vl (0.06 , 0.16 , 0.33) c3 vl vl vl (0 , 0.1 , 0.2) c4 vh vh h (0.73 , 0.86 , 0.96) c5 m h h (0.5 , 0.7 , 0.83) c6 vh h h (0.66 , 0.83 , 0.93) c7 vh vh vh (0.8 , 0.9 , 1) c8 vh vh vh (0.8 , 0.9 , 1) c9 m h l (0.33 , 0.5 , 0.66) c10 m l m (0.23 , 0.4 , 0.6) c11 h h vh (0.66 , 0.83 , 0.93) next step is to gather the decision makers’ ratings for the treatment alternative with respect to each criterion. then, these ratings are aggregated into an overall rating according to eq. (2). using these overall ratings, the decision matrix is constructed as given in table 8. then, separation measures from the fuzzy best value i s ~ and the fuzzy worst value i r ~ are computed using eq. (3) and (4). these values are given in table 9. table 8. fuzzy evaluation matrix for the given treatment alternatives. t1 t2 t3 t4 t5 c1 (8.3 , 9.6 , 10) (6.3 , 8.3 , 9.6) (2.3 , 4.3 , 6.3) (3 , 5 , 7) (5.6 , 7.6 , 9.3) c2 (7 , 8.6 , 9.6) (5.6 , 7.6 , 9.3) (7.6 , 9 , 9.6) (0 , 0.3 , 1.6) (8.3 , 9.6 , 10) c3 (2 , 3.6 , 5.6) (3 , 5 , 7) (7.6 , 9.3 , 10) (4.3 , 6.3 , 8.3) (5 , 7 , 8.6) c4 (0.3 , 1.3 , 3) (5 , 7 , 8.6) (2.3 , 4.3 , 6.3) (8.3 , 9.6 , 10) (0 , 0 , 1) c5 (0 , 0.3 , 1.6) (4.3 , 6.3 , 8.3) (0 , 0.6 , 2.3) (8.3 , 9.6 , 10) (0 , 0.3 , 1.6) c6 (3 , 5 , 7) (0.3 , 1.3 , 3) (1.3 , 3 , 5) (6.3 , 8.3 , 9.6) (0 , 0.3 , 1.6) c7 (7.6 , 9.3 , 10) (4.3 , 6.3 , 8.3) (8.3 , 9.6 , 10) (0.3 , 1.6 , 3.6) (5.6 , 7.6 , 9) c8 (0.3 , 1.3 , 3) (5 , 7 , 8.6) (3 , 5 , 7) (5.6 , 7.6 , 9.3) (0 , 0 , 1) c9 (2.3 , 4.3 , 6.3) (5 , 7 , 8.6) (8.3 , 9.6 , 10) (3 , 5 , 7) (0 , 0 , 1) c10 (3.6 , 5.6 , 7.6) (2.3 , 4.3 , 6.3) (7.6 , 9.3 , 10) (5 , 7 , 8.6) (0 , 0.3 , 1.6) c11 (3 , 5 , 7) (5 , 7 , 8.6) (8.3 , 9.6 , 10) (0 , 1 , 3) (6.3 , 8.3 , 9.6) published by atlantis press copyright: the authors 39 fuzzy multicriteria risk assessment table 9. separation measures of treatment alternatives from the fuzzy best and fuzzy worst values. is ~ i r ~ t1 (3.16 , 3.56 , 3.69) (0.75 , 0.74 , 0.83) t2 (2.26 , 2.43 , 2.19) (0.62 , 0.72 , 0.76) t3 (2.05 , 2.28 , 2.41) (0.52 , 0.67 , 0.76) t4 (1.92 , 2.51 , 3.00) (0.8 , 0.9 , 1) t5 (3.73 , 4.67 , 5.33) (0.8 , 0.9 , 1) after that, *~ s ,  s ~ , *~ r and  r ~ are calculated in order to compute the i q ~ using eq. (5). the value of v in this equation is taken as 0.5. the last step is to defuzzify the i q ~ values using eq. (6) and then rank the alternatives based on the qi index values. table 10 gives the results of the fuzzy vikor analysis. table 10. fuzzy vikor analysis results for hazardous waste treatment method selection. i q ~ i q rank order t1 (0.38 , 0.41 , 0.75) 0.47 3 t2 (0 , 0.14 , 0.27) 0.14 2 t3 (0 , 0.035 , 0.036) 0.03 1 t4 (0.5 , 0.55 , 0.63) 0.55 4 t5 (1 , 1 , 1) 1.00 5 according to the crisp qi values, the ranking of the alternatives is found to be t3, t2, t1, t4 and t5. the best alternative is identified as t3 by which thermal treatment methodologies (combustion) was denoted. the second best treatment methodology is concluded as t2, which is representing the chemical treatment. and the ranking order follows as t1, physical treatment, t4 biological treatment and t5 no treatment but directly disposal. 8. conclusion in today’s industrialized world, hazardous waste is inevitable however it poses a high risk if it is not managed properly. hazardous waste risk range from toxic risks including the probability of a certain population have a particular illness to other risks such as ecosystem stress, loss of important habitats, public safety and public welfare hazards. thus, the ultimate goal to manage hazardous waste safely from the moment it is generated; while it is transported, treated, or stored; until the moment it is disposed with the minimum cost. turkey is a developing country with a high industry growth and high population. however, the attention given to hazardous waste management is not sufficient. the first hazardous waste management controlling directive was announced 25 years ago. and yet, hazardous waste management activities have not been organized effectively to fulfill the need for dealing with the generated waste types and amounts. therefore, we have proposed a fuzzy vikor multicriteria decision analysis for the selection of the most appropriate hazardous waste treatment methodology. then, an application was presented to show the potential of the proposed methodology for the case of istanbul. in the application problem was designed to select among the physical, chemical, thermal, biological treatment processes or a disposal option without any treatment process. the alternatives were evaluated with respect to eleven criteria. the results of the decision analysis which was based on the evaluations of three experts from istanbul suggest that combustion is the treatment alternative which fits istanbul’s hazardous waste management needs best. the treatment methods to be considered next in an order are concluded as chemical treatment, physical treatment and biological treatment. the last option for hazardous waste to be considered is found to be direct disposal without any treatment. in the future research, similar studies can be conducted for the selection of specialized treatment methodologies which are appropriate for specific waste groups. besides, the problems may be solved by different multi-criteria decision-making methodologies such as fuzzy promethee, fuzzy electre or fuzzy topsis. published by atlantis press copyright: the authors 40 seda uğurlu, cengiz kahraman references 1. vallero, d.a. and peirce, j.j., engineering the risks of hazardous wastes (butterworth–heinemann publications elsevier science, ma, 2003). 2. shammas n.k., characteristics of hazardous industrial waste, in handbook of advanced industrial and hazardous wastes treatment, eds. wang l.k., hung, y.t., shammas n.k. (crc press, florida, 2010), pp.485517. 3. cheremisinoff n.p. and cheremisinoff p.n., hazardous materials and waste management: a guide for the professional hazards manager (noyes publications, new jersey, 1995). 4. misra v., pandey s.d., hazardous waste, impact on health and environment for development of better waste management strategies in future in india, environment international, 31 (2005) 417– 431. 5. pavlou, s. p., & stansbury, j. s., risk-cost trade off considerations for contaminated sediment disposal, human and ecological risk assessment, 4(4) (1998) 991–1002. 6. nema a.k. and gupta, s.k., optimization of regional hazardous waste management systems: an improved formulation, waste management 19 (1999) 441-451. 7. insua, d. r., galleqo, e., mateos, a., and rios-insua, s., moira: a decision support system for decision making on aquatic ecosystems contaminated by radioactive fallout. annals of operations research 95(2000) 341– 364. 8. gallego, e., jim´enez, a., mateos, a., sazykina, t., & rios-insua, s., application of multiattribute analysis (maa) to search for optimum remedial strategies for contaminated lakes with the moira system. paper presented at the 11th annual meeting of the international radiation protection association (madrid, 23–28 may 2004). 9. rogers, s. h., seager,t. p. and gardner, k.h., combining expert judgement and stakeholder values with promethee: a case study in contaminated sediments management, in comparative risk assessment and environmental decision making, eds. i. linkov & a. bakr ramadan (kluwer academic publishers, boston, ma 2004), pp. 305–322. 10. i. linkov, f. k. satterstrom, g. kiker, t. p. seager, t. bridges, k. h. gardner, s. h. rogers, d. a. belluck, and a. meyer, multicriteria decision analysis: a comprehensive decision approach for management of contaminated sediments, risk analysis, 26(1) (2006) 6178. 11. stansbury, j., member, p. e., bogardi, i., & stakhiv, e. z., risk-cost optimization under uncertainty for dredged material disposal, j. of water resources planning and management, 125(6) (1999) 342–351. 12. i. khadam and j.j. kaluarachchi, applicability of riskbased management and the need for risk-based economic decision analysis at hazardous waste contaminated sites, environment international 29 (2003) 503– 519. 13. i. khadam and j.j. kaluarachchi, multi-criteria decision analysis with probabilistic risk assessment for the management of contaminated ground water, environmental impact assessment review 23 (2003) 683–721. 14. w.c. jr. blackman, basic hazardous waste management (crc press lewis publishers, florida, 1996). 15. secretariat of the basel convention (sbc), 1999, technical guidelines on hazardous wastes: physicochemical treatment biological treatment, basel convention series/sbc no: 99/007, geneva, switzerland, september 1999. 16. s. opricovic and g.h.tzeng, compromise solution by mcdm methods: a comparative analysis of vikor and topsis, european j. of operational research 156(2) (2004) 445-455. 17. wang t.c., liang j.l. and ho c.y., multi-criteria decision analysis by using fuzzy vikor, in proceedings book of international conference on service systems and service management (2006) pp.901 – 906. 18. kaya, t. and kahraman c., 2010, multicriteria renewable energy planning using an integrated fuzzy vikor & ahp methodology: the case of istanbul, energy, 35, 2517-2527. 19. kaya, t. and kahraman c., 2011, fuzzy multiple criteria forestry decision making based on an integrated vikor and ahp approach, expert systems with applications (article in publication). 20. d. yong, plant location selection based on fuzzy topsis, international j. of advanced manufacturing technologies, 28 (2006) 839-844. 21. tui̇k, statistics of manufacturing industry water, waste water and waste for 2008, tui̇k haber bülteni 32 (2010). published by atlantis press copyright: the authors 41 atlantis press journal style received 26 june 2015 accepted 10 november 2015 analysis of a risky two unit system under marked process incorporating two repairmen with vacations nidhi tiwari department of applied sciences, teerthanker mahaveer university, moradabad, u. p. 244001, india e-mail: nidhitiwari26@gmail.com s. b. singh department of mathematics, statistics and computer science, g. b. pant university of agriculture and technology, pantnagar, uttarakhand, 263145, india e-mail: drsurajbsingh@yahoo.com abstract in this paper the system considered consists of two subsystems a and b. subsystem a has only one unit whereas subsystem b consists of two homogeneous units b1 and b2. here b2 is in hot standby with b1. in the present study we have incorporated two repairmen namely supervisor and novice to repair the failed units of the system. the supervisor is always there whereas the novice remains in vacation. the system is analyzed under “preemptiveresume-repair discipline”. we have used supplementary variable technique, laplace transformation and copula to obtain various transition state probabilities, reliability, availability, cost analysis and the steady state behaviour of the system. to make a comparison between the measures obtained using two families of copulas gumbel-hougaard family and bivariate clayton, the model has been solved with the help of gumbel-hougaard family of copula and bivariate clayton copula both in two different cases. also a comparison between two types of repair policies “head-of-line-repair” and “preemptive-resume-repair” has been carried out. at last some numerical examples have been taken. keywords: reliability, marked process, cost analysis, gumbel-hougaard copula, bivariate clayton copula, supplementary variable technique. 1. introduction recent advances in science and technology have led to a rapid increase in the complexity of most engineered systems. these complex engineered systems are composed of many subsystems due to which it is very risky to use them and they are characterized by observable complex behaviour. data from these repairable systems generally contain more information than only failure time. to find the reliability of such complex repairable system going to be a central challenge for engineers, mathematicians and statisticians. marked process (lindqvist and bedford, 2004) can prove to be more useful technique to tackle this problem. it is a well known phenomenon that to increase the reliability and availability of the system a unit that has failed is repaired. the repair facility may have a single repairman or more than one repairman to repair the failed units. in last decades there has been a growing interest among researchers (gupta and agarwal, 1984; ram and singh, 2010) in the reliability analysis of complex system incorporating one repairman. of course there are researchers (garg and goel, 1985; gupta and sharma, 1993; goel and gupta, 1985) who modeled the complex system incorporating two repairmen or two repair facilities or two repair journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 200-214 published by atlantis press copyright: the authors 200 n. tiwari, s. b. singh distributions. but they did not include the vacations of repairman in their study. though there are good studies (linmin and jiandong, 2009; tuteja and malik, 1992) who have given attention to the vacations of repairman but their modeling includes only one repairman. furthermore a review of past work (kovalenko and smolich, 2001) shows that while analyzing complex systems there can be two situations: (i) repairman is a full time employee of the organization (ii) due to present day condition in mind managerial staff of a company has to face tough time which becomes especially aggravating during the current economic crisis leading to appointment of employees not on permanent basis. both these situations lead to vacations of repairman, who will be called for repairing as and when they are required. hence vacation of the repairman can also play key role in the analysis of repairable complex systems. this paper is an extension of the work done in our two previous papers (tiwari and singh, 2010; tiwari and singh, 2010) in which we have used “head-of-linerepair” and “preemptive-repeat-repair” policies to repair the failed units. with above facts in view the present paper examines and proposes an approach to deal with the complex system incorporating information gathered due to marked process having two repairmen with different skills and availability. the system under study consists of two subsystems a and b arranged in series configuration. a has only one unit while b consists of two identical units b1 and bb2 (demcese, 2009; kumar and singh, 2008). the failure rates of both the units of subsystem b are same. subsystem a has two states: good and failed, while b can be in any of the three states: good, degraded and failed. two types of failure viz. partial and catastrophic can occur in subsystem b. in case of partial failure subsystem b goes in the state of partial failure which brings the whole system to the state of reduced efficiency. while in case of catastrophic failure the subsystem b and hence the whole system fails completely. the system can further fail completely if (i) subsystem a fails or (ii) both the units of subsystem b fail. there are two repairmen namely supervisor and novice available to repair the failed units of the system. the novice is not a regular worker he remains in vacation and can be called for repairing as per the requirements. further, it is assumed that novice can repair only subsystem b and he is not good at repairing of subsystem a. also, whenever there is a failure in a unit of subsystem b, supervisor starts repairing it but if at the same time there occurs any failure in other unit of b then novice is called for repairing. the system is under “preemptive-resumerepair discipline” where subsystem a has given priority over subsystem b. the present system is under marked process so at any state more than one type of failure can take place. the two types of failures between two states can be calculated with the help of copula (nelson, 2006; yan jun, 2007). here in the present study it is assumed that the failure from state s0 to s2 follows two distributions, i.e. constant and logarithmic. system can be repaired in two ways from states s6/s2 to s0 as well. failure rates are assumed to be constant in general whereas the repairs done by supervisor and novice follow general and exponential distribution respectively. the research is part of larger effort to investigate the application of copula methodology for reliability analysis into the decision making process. the model has been solved by using two different types of copulas viz. gumbel-hougaard and bivariate clayton. the following characteristics of the system have been analyzed: 1. transition state probabilities of the system. 2. asymptotic behaviour of the system. 3. various measures such as reliability, availability, m.t.t.f. analysis and cost effectiveness of the system. the following comparative studies have also been made to test the applicability of repair policies and copulas during the analysis of model: (1) a comparison between availabilities of two different types of repair policies, namely “head-ofline-repair” and “preemptive-resume-repair” with the application of copula. (2) a comparison between the various measures of reliability when two different copulas, i.e. gumbelhougaard and bivariate clayton are applied. 1.1. copula a two dimensional copula is a function c: [0, 1] × [0, 1] → [0, 1] that satisfies following two properties. 1. boundary conditions: (a) for all t in [0, 1], c (t, 0) = c (0, t) = 0 (b) for all t in [0, 1], c (t, 1) = c (1, t) = t 2. rectangular inequality: if u1, u2, v1, v2 are in [0, 1] with u1 ≤ u2 and v1 ≤ v2, then c (u2, v2) c (u1, v2) c (u2, v1) + c (u1, v1) ≥0 published by atlantis press copyright: the authors 201 analysis of a risky two unit system under marked process incorporating two repairmen with vacations 1.1.1. bivariate clayton copula a two dimensional bivariate clayton copula is defined as θθθ θ /1 2121 )1(),( −−− −+= uuuuc the clayton copula is well defined for 0 < θ < ∞ and for θ→0 and θ→∞ it converges to the product copula and comonotonicity respectively. it is not symmetric and has lower tail dependence (no upper tail dependence). due to the property of lower tail dependence, the clayton copula is a possible candidate for model building specially in the financial context. 1.1.2. bivariate gumbel-hougaard family copula a two dimensional bivariate clayton copula is defined as ∞≤≤−+−−= θ θ θθ θ 1),))log()log((exp(),( 1 2121 uuuuc for θ = 1 the gumbel-hougaard copula models independence, for θ→∞ it converges to comonotonicity. it is not symmetric and has upper tail dependence (no lower tail dependence) also it has positive dependence. 1.1.3. applications of copula in the present study copula representations of multivariate distributions allow us to fit any marginal we like to different random variables, and these distributions might differ from one variable to another. in the present paper we have applied copula to find the joint probability distribution of repair/failure probabilities following different types of distribution. 1.2 repair policies the comparison between two repair policies has been made in the present study: “head-of linerepair” and “preemptive-resume-repair”. 1.2.1 head-of-line-repair in this repair policy the failed unit will be repair according to first come first get. in state s3 when the system is already in degraded state due to the failure in one of the unit of subsystem b if subsystem a also fails, it comes to state s4 which is a fully failed state. in head-of-line-repair policy firstly the repair of failed b unit will be completed after that failed a unit will be repair. 1.2.2 preemptive-resume-repair in this type of repair we give priority to one unit over other unit for repair. in the present model we have given priority to subsystem a over b. so when the system comes to state s4 in which one of subsystem b unit has already in repair in s3 state and now a unit also fails, according to this policy we have given priority to subsystem a over b so the repairman will start to repair subsystem a. after the repair of subsystem a he will start to repair subsystem b from the point where he has left it. 2. assumptions (1) initially the system is in good state. (2) subsystems a and b are connected in series. (3) subsystem a has constant failure rate and can be in two states: good and failed. (4) subsystem b has three states namely good, degraded and failed. (5) system is under marked process. (6) repairs done by the supervisor and novice follow general time distribution and exponential time distribution respectively. (7) there are two types of failures from state s0 to s2 one is constant and other is logarithmic. (8) there are two types of repair between s6/s2 to s0. (9) subsystem a can be repaired by supervisor only. (10) system is under preemptive resume repair discipline, i. e. the preempted b is taken back in repair facility after priority repair of a from the point where it has been left. (11) after repair the system is as good as new. (12) in case of catastrophic failure both the repairmen repair the system. (13) joint probability distribution of repair rates, where repair is done by supervisor and novice follows gumbel-hougaard family and bivariate clayton copula. 3. state specification table to mention the different states of the system, the block diagram of the system and transition state diagram of the system are given in next page. here we have used some notations as g = good state, f = failed state, d = degraded state, published by atlantis press copyright: the authors 202 n. tiwari, s. b. singh fr = failed under repair, dr = degraded state under repair, frv = failed state, one unit is under repair and novice is in vacation. table 1: state specification chart states state of subsystem a subsystem b: number of good units system state s0 g 2 g s1 f 2 fr s2 g 0 fr s3 g 1 dr s4 f 1 fr s5 g 0 frv s6 g 0 fr 4. notations aλ failure rate of subsystem a. cp λλ , failure rates of subsystem b corresponding to partial and catastrophic failures for both the units. )(riφ repair rate of supervisor. )(riψ repair rate of novice if i = a/b/c then r = x/y/z respectively. x elapsed repair time for the subsystem a. y, z elapsed repair times for the partial and catastrophic failures respectively for the subsystem b. η, u vacation rate and variable for vacation of repairman. pi (t) probability that the system is in si state at an instant t for i = 1, 2,…., 6. (s)pi laplace transform of pi (t). p4(y, t) probability density function that at time t the system is in failed state s4 and the system is under repair, elapsed repair time is y. e p (t) expected profit during the interval (0, t]. k1, k2 revenue per unit time and service cost per unit time respectively. fig. 1: block diagram of the system fig 2: transition state diagram (x)sη ∫− x 0 η(x)dx)η(x)exp( )(s xη laplace transform of (x)sη (x)sη = dx 0 x 0 η(x)dx)xη(x)exp(-s∫ ∞ ∫− if ),(1 yu pφ= )(2 yu pψ= then the expression for the joint probability according to: (i) gumbel-hougaard family of copula is given as ]}))(log())(log{(exp[),( /121 θθθθ ψφ yyuuc pp −+−−= (ii) bivariate clayton family of copula is given as θθθ θ ψφ /121 ])}(()}([{),( −−− += yyuuc pp 5. formulation of mathematical model by probability considerations and continuity arguments, we obtain the following set of integro-differential equations governing the behaviour of the system. ∫=+++ ∞ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ 0 10 ),()()(2 dxtxpxtpdt d apa φλλλ dytypydztzpz ),()( ),()( 6 00 2 ∫+∫+ ∞∞ ψψ ),()( 0 3∫+ ∞ dytypypφ (1) published by atlantis press copyright: the authors 203 analysis of a risky two unit system under marked process incorporating two repairmen with vacations 0),()( 1 =+∂ ∂+ ∂ ∂ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ txpx xt a φ (2) 0),()( 2 =+∂ ∂+ ∂ ∂ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ tzpz zt ψ (3) ),()( 3 typyyt ppa ⎥⎦ ⎤ ⎢ ⎣ ⎡ +++ ∂ ∂+ ∂ ∂ φλλ dxtyxpx a∫= ∞ 0 4 ),,()(φ (4) 0),,()( 4 =+∂ ∂+ ∂ ∂ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ tyxpx xt a φ (5) 0),,()( 5 =+∂ ∂+ ∂ ∂ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ tuypu yt η (6) 0),()( 6 =+∂ ∂+ ∂ ∂ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ typy yt ψ (7) 5.1 boundary conditions )(),0( 01 tptp aλ= (8) )(),0( 02 tptp λ= (9) )(2),0( 03 tptp pλ= (10) t)(y,),,0( 3a4 ptyp λ= (11) ),(),,0( 35 typtup pλ= (12) ),,()(),0( 56 tuyputp η= (13) 5.2 initial condition: 0)0(0 =p (14) and other state probabilities are zero at t = 0. the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. (i) when gumbel-hougaard copula is applied to analyse the model then in the equations (1), (3), (7) and (9), ψ(y), ψ(z), λ and ξ will have the following expressions: ]}))(log())(log{(exp[)( /1 θθθ φψψ yyy pp −+−−= θθθ φψψ /1}))(log())(log{(exp[)( zzz cc −+−−= θθθλλ /1})log(log()log{(exp[ zc −+−−= )()()( yyy p ψφξ += (ii) when bivariate-clayton copula is applied to analyse the model then in the equations (1), (3), (7) and (9), ψ(y), ψ(z), λ and ξ will have the following expressions: θθθ φψψ /1])}({)}([{)( −−− += yyy pp θθθ φψψ /1])}({)}([{)( −−− += zzz cc θθθλλ /1])(log)[( −−− += zc )()()( yyy p ψφξ += 6. solution of the model taking laplace transformation of equations (1-13) and using (14), we get [ ] ∫+=+++ ∞ 0 10 ),()(1)(2 dxsxpxsps apa φλλλ dysypydzszpz ),()( ),()( 6 0 2 0 ∫+∫+ ∞∞ ψψ ∫+ ∞ 0 3 ),()( dysypypφ (15) 0),()( 1 =+ ∂ ∂+ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ sxpx x s aφ (16) 0),()( 2 =+∂ ∂+ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ szpz z s ψ (17) ∫= +++ ∂ ∂+ ∞ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ 0 4 3 ),,()( ),()( dxsyxpx sypy y s a ppa φ φλλ (18) 0),,()( 4 =+∂ ∂+ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ syxpx x s aφ (19) 0),,()( 5 =+∂ ∂+ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ suypu y s η (20) 0),()]( 6 =+∂ ∂+ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ sypy y s ψ (21) 6.1 boundary conditions (s)),0( 01 psp aλ= (22) (s)),0( 02 psp λ= (23) )(2),0( 03 spsp pλ= (24) s)(y,),,0( 3a4 psyp λ= (25) ),(),,0( 35 sypsup pλ= (26) ),,()(),0( 56 suypusp η= (27) solving equations (15-21) and using equations (2227), one can get the following transition state probabilities published by atlantis press copyright: the authors 204 n. tiwari, s. b. singh where )(ssss aapa φλλλ −++=∗ (35) )(2)()(2 )(2)( 2 ussss ssssd pzpp aapa ηλλλ λλλλ ψφ φ −−∗− −+++= (36) ∫−+ ∗+−∫ ∞ y dyyyussy 00 })())(2(exp{)( ξηψ also up and down state probabilities of the system is given by )()()( 30up spspsp += )()(121 0 sp s ss p p ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ∗ ∗− += φλ (37) )()( )()()()( 65 421down spsp spspspsp ++ ++= )()(1 )(1 0a sps ss s ss za ⎥ ⎥ ⎦ ⎤ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ − + ⎢ ⎢ ⎣ ⎡ − = ψφ λλ 2 0 2 0 )(2 )( ))((1)(2 )()(1 )(12 p p p pa pa u uss ussssp sp s ss s ss λη η ηλ λλ φ φφ + +∗+ +∗+−+ ∗ ∗−×−+ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ )(2 ))(2(1)(0 ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ +∗+ +∗+− ×× uss ussssp η ηξ (38) also it is noticeable that /1)()( downup sspsp =+ (39) the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. (i) when gumbel-hougaard copula is applied to analyse the model then in the equations (15), (17), (21), (23), (30), (34) and (36), ψ(y), ψ(z), λ and ξ will have the following expressions: ]}))(log())(log{(exp[)( /1 θθθ φψψ yyy pp −+−−= θθθ φψψ /1}))(log())(log{(exp[)( zzz cc −+−−= θθθλλ /1})log(log()log{(exp[ zc −+−−= )()()( yyy p ψφξ += (ii) when bivariate-clayton copula is applied to analyse the model then in the equations (15), (17), (21), (23), (30), (34) and (36), ψ(y), ψ(z), λ and ξ will have the following expressions: θθθ φψψ /1])}({)}([{)( −−− += yyy pp θθθ φψψ /1])}({)}([{)( −−− += zzz cc θθθλλ /1])(log)[( −−− += zc )()()( yyy p ψφξ += 7. asymptotic behaviour of the system using able’s lemma )say()(lim)}({lim 0 ftfsfs ts == ∞→→ in equations (37) and (38) we get the following time independent probabilities )( 1)(0 sd sp = (28) )( )(1 )( 01 sps ss sp aa ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − = φλ (29) )( )(1 )( 02 sps ss sp z ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − = ψλ (30) )(1 )(2)( 03 ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎣ ⎡ ∗ ∗− = s ss spsp pp φλ (31) )()(1 )(1 2)( 0 4 sp s ss s ss sp p a pa ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ∗ ∗− − = φ φλλ (32) )( )( ))((12)( 025 sp uss ussssp pp ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ +∗+ +∗+−= η ηλ φ (33) )(2)( )(2 ))(2(1)( 0 2 6 spu uss ussssp pλη η ηξ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ +∗+ +∗+−= (34) published by atlantis press copyright: the authors 205 analysis of a risky two unit system under marked process incorporating two repairmen with vacations )0( 1)(121)(up ds sssp p p ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ∗∗ ∗∗− += φλ (40) [ ] ⎢ ⎢ ⎣ ⎡ ++= 2 )0( 1)( adown apaza mmm d sp φψφ λλλλ ⎥ ⎥ ⎦ ⎤ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ + +− + ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ∗∗ ∗∗ +∗∗ +∗∗−+ ∗∗ ∗∗− )( ))((1 2)( 2 2 2 )( ))((1)(1 us uss u pp pp us uss s ss η η λη ξ φφ λ η η (41) where )(lim)0( 0 sdd s→ = (42) ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ −= → s ssm a s a )(1lim 0 φ φ (43) ∗=∗∗ → ss s 0 lim (44) the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. (i) when gumbel-hougaard copula is applied to analyse the model then in the equations (40), (41) and (42), ψ(y), ψ(z), λ and ξ will have the following expressions: ]}))(log())(log{(exp[)( /1 θθθ φψψ yyy pp −+−−= θθθ φψψ /1}))(log())(log{(exp[)( zzz cc −+−−= θθθλλ /1})log(log()log{(exp[ zc −+−−= )()()( yyy p ψφξ += (ii) when bivariate-clayton copula is applied to analyse the model then in the equations (40), (41) and (42), ψ(y), ψ(z), λ and ξ will have the following expressions: θθφθψψ /1])}({)}([{)( −−+−= ypypy θθθ φψψ /1])}({)}([{)( −−− += zzz cc θθθλλ /1])(log)[( −−− += zc )()()( yyy p ψφξ += 8. particular cases (1) when catastrophic failure does not occur in the subsystem b. in this by putting 0=cλ in equations (37) and (38), we get )()()()()()( 65421down spspspspspsp ++++= ×−+−= ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ s sssp s ss a pa a )(12)()(1 0a φφ λλλ ×+ +∗+ +∗+− +∗ ∗− ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ × ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ )()(2 )( ))((1 )(2)()(1 0 2 0 2 0 spu uss usss spsp s ss p p p p λη η η λ φ φ )(2 ))(2(1 ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ +∗+ +∗+− uss usss η ηξ (46) where )( 1)( 1 0 sd sp = (47) yussyu ssssssd p ppaapa ))(2(exp{)()(2 )(2)(2)( 2 1 ηψηλ λλλλ φφ +∗+−∫×− 0 ∗−−++= ∞ ∫− y dyy })(ξ 0 (48) )(ssss aapa φλλλ −++=∗ (49) the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. (i) when gumbel-hougaard copula is applied to analyse the model then in the equations (46) and (48), ψ(y) and ξ will have the following expressions: ]}))(log())(log{(exp[)( /1 θθθ φψψ yyy pp −+−−= )()()( yyy p ψφξ += (ii) when bivariate-clayton copula is applied to analyse the model then in the equations (46), (48), ψ(y) and ξ will have the following expressions: θθθ φψψ /1])}({)}([{)( −−− += yyy pp )()()( yyy p ψφξ += (2) when repair follows exponential distribution. )()()( 30up spspsp += )()(121 0 sp s ss p p ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ∗ ∗− += φλ (45) published by atlantis press copyright: the authors 206 n. tiwari, s. b. singh in this case the result can be derived by putting (50) )( )()(, )( )()( ]}))(log())(log{(exp[ ]}))(log())(log{(exp[)( /1 /1 xs xss ys yss xzs xzss a a a p p p cc cc z φ φ φ φ φψ φψ φφ θθθ θθθ ψ + = + = −+−−+ −+−− = in equations (37) and (38), we get )(121 )()()( 0 30up sp s spspsp p p ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ +∗ += += φ λ (51) )()()()()()( 65421down spspspspspsp ++++= )(112 )(1 1 0 0 a sp ss sp ss p a pa za ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎭ ⎬ ⎫ ⎩ ⎨ ⎧ ⎥ ⎥ ⎦ ⎤ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ ⎪⎭ ⎪ ⎬ ⎫ ⎪⎩ ⎪ ⎨ ⎧ +∗ × + + + + ⎢ ⎢ ⎣ ⎡ + = φφ λλ ψ λ φ λ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ × ++∗+ + p p uss sp φη λ )( 1)(2 0 2 ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ++∗+ + ξη λη )(2 1)()(2 02 uss spu p (52) where )( 1)( 2 0 sd sp = (53) ×− +∗ − + − + −+++= )(22 )( )( 2)( 2 2 u szs z s ssd p p pp a a apa ηλ φ φλ ψ ψλ φ φλλλλ ∫−+ ∗+−∫ ∞ y dyyyussy 00 })())(2(exp{)( ξηψ (54) the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. (i) when gumbel-hougaard copula is applied to analyse the model then in the equations (52) and (54), ψ(y), ψ(z), λ and ξ will have the following expressions: ]}))(log())(log{(exp[)( /1 θθθ φψψ yyy pp −+−−= θθθ φψψ /1}))(log())(log{(exp[)( zzz cc −+−−= θθθλλ /1})log(log()log{(exp[ zc −+−−= )()()( yyy p ψφξ += (ii) when bivariate-clayton copula is applied to analyse the model then in the equations (52) and (54), ψ(y), ψ(z), λ and ξ will have the following expressions: θθθ φψψ /1])}({)}([{)( −−− += yyy pp θθθ φψψ /1])}({)}([{)( −−− += zzz cc θθθλλ /1])(log)[( −−− += zc )()()( yyy p ψφξ += 9. numerical computation 9.1 availability analysis let the repair follows exponential distribution. also let vacation rate be η(u) = 0.5, failure rates of subsystem a and b for partial and catastrophic failures be λa = 0.5, λp = 0.25, λc = 0.25, repair rates be фp = фc = фa = ψp =1, θ = 1 and x = y = z = 1. putting all these values in equation (37), using equation (50) and taking inverse laplace transformation, we get: (i) in case of gumbel-hougaard copula: pup=-0.03612485642 e (-2.510615836t) +0.3068736726 e(-1.557348320 t) +0.06456337718 e(-1.425285235t) -0.001878777739 e(-0.6777682208t)-0.005058707285 e(-0.6247959226t)+0.6716252918 e(-0.03751979833t) (55) table 2: time vs. availability for gumbel-hougaard copula time pup 0 1.000000 1 0.720476 2 0.638253 3 0.602853 4 0.578305 5 0.556385 6 0.536127 7 0.516421 8 0.497432 9 0.479132 10 0.464976 published by atlantis press copyright: the authors 207 analysis of a risky two unit system under marked process incorporating two repairmen with vacations fig. 3: time vs. availability for gumbel-hougaard copula (ii) in case of bivariate clayton copula: pup= 0.06406442012 e (-2.663714173 t) + 0.3687669327 e(-1.748168308 t) +0.007333724874 e(-1.4444003723 t) – 0.008933923194 e(-0.6562343416 t) 0.00857068425 e(-0.6223021709 t)+0.5773259140 e(-0.03224395049 t) (56) fig. 4: time vs. availability for bivariate-clayton copula now in equations (55) and (56) setting t = 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, one can obtain tables 2 and 3 respectively. also figs. 3 and 4 show the variation of availability with respect to time in these two cases respectively. 9.2 reliability analysis for reliability analysis of the present model let the repair follows exponential distribution and λa = 0.5, λp = 0.25, λc = 0.25, фp = фc = фa = ψp = 0, θ = 1 and x = y = z = 1. putting all these values in equation (37) and using (50), taking inverse laplace transform and varying time, one can obtain tables 4 and 5 and correspondingly figs. 5 and 6 with respect to gumbel-hougaard and bivariate clayton copula respectively. the tables and figs. are given below: fig. 5: time vs. reliability for gumbel-hougaard copula table 4: time vs. reliability for gumbel-hougaard copula time pup 0 1.000000 1 0.576854 2 0.310925 3 0.161011 4 0.081258 5 0.040298 6 0.019739 7 0.009583 8 0.004622 9 0.002218 10 0.001061 table 3: time vs. availability for bivariate-clayton copula time pup 0 1.000000 1 0.620174 2 0.548297 3 0.523589 4 0.506472 5 0.490712 6 0.475406 7 0.460479 8 0.445955 9 0.431851 10 0.418173 published by atlantis press copyright: the authors 208 n. tiwari, s. b. singh fig. 6: time vs. reliability for bivariate-clayton copula 9.3 m. t. t. f. analysis m. t. t. f. of the system can be obtained as m. t. t. f. = )(lim 0 sp up s→ let repairs follow exponential distribution, then considering following cases, we can obtain the m. t. t. f. of the system with respect to different parameters. this can be obtained by substituting under mentioned values in the equation (50). (a) setting фa = фc = фp = ψa = ψp = 0, λp = 0.15, λc =0.10, x = y = z =1, θ = 1 and varying λa as 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, one can obtain tables 6 and 7 which show variation of m. t. t. f. with respect to λa in the case of gumbel-hougaard copula and bivariate clayton copula respectively. (b) setting фa = фc = фp = ψa = ψp = 0, λa = 0.20, λc = 0.10, x = y = z =1, θ = 1 and varying λp as 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, one can obtain tables 8 and 9 which demonstrate variation of m. t. t. f. with respect to λp in the case of gumbelhougaard copula and bivariate clayton copula respectively. table 5: time vs. reliability for bivariate-clayton copula time pup 0 1.000000 1 0.442542 (c) setting фa = фc = фp = ψa = ψp = 0, λa = 0.20, λp = 0.15, x = y = z =1, θ = 1 and varying λc as 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, one can obtain tables 10 and 11 which show variation of m. t. t. f. with respect to λc in the case of gumbel-hougaard and bivariate clayton copula respectively. variation of m. t. t. f with respect to λa, λp and λc in the cases (a), (b) and (c) for gumbel-hougaard and bivariate clayton copula have been shown by the figs: 7, 8, 9, 10, 11 and 12 respectively. the figs. and tables are given below: 2 0.209180 3 0.092058 4 0.043364 5 0.020339 6 0.009569 7 0.004510 8 0.002127 9 0.001004 10 0.000474 table 6: λa vs. m.t.t.f. for gumbel-hougaard copula λ mttf a .10 4.789117552 .20 3.320033009 .30 2.527646829 .40 2.035166969 .50 1.700699662 .60 1.459283665 .70 1.277112835 .80 1.134912933 .90 1.020914718 fig. 7: λa vs. m.t.t.f. for gumbel-hougaard copula published by atlantis press copyright: the authors 209 analysis of a risky two unit system under marked process incorporating two repairmen with vacations table 7: λ fig. 8: λa vs. m.t.t.f. for bivariate-clayton copula fig. 9: λp vs. m.t.t.f. for gumbel-hougaard copula a vs. m.t.t.f. for bivariate-clayton copula λ mttf a .10 4.572318664 .20 3.195599546 .30 2.446819664 .40 1.978419064 .50 1.658659691 .60 1.426887824 .70 1.251383502 .80 1.113984201 .90 1.003557701 table 9: λ vs. m.t.t.f. for bivariate-clayton copula p λ mttf p .10 3.463877777 .20 2.936183598 .30 2.496719324 .40 2.158183142 .50 1.895608990 .60 1.687870466 .70 1.520117750 .80 1.382128630 .90 1.266778289 fig. 10: λp vs. m.t.t.f. for bivariate-clayton copula table 8: λ vs. m.t.t.f. for gumbel-hougaard copula p λ mttf p .10 3.628119357 .20 3.033176195 .30 2.560000544 .40 2.202556920 .50 1.928394467 .60 1.713062314 .70 1.540071314 .80 1.398319465 .90 1.280176690 published by atlantis press copyright: the authors 210 n. tiwari, s. b. singh fig. 11: λc vs. m.t.t.f. for gumbel-hougaard copula table 10: λ vs. m.t.t.f. for gumbel-hougaard copula c λ mttf c .10 3.320033009 .20 3.001444774 .30 2.738646057 .40 2.518162301 .50 2.330534884 .60 2.168928741 .70 2.028281721 .80 1.904764759 .90 1.795427972 fig. 12: λc vs. m.t.t.f. for bivariate-clayton copula 9.4 cost analysis letting vacation rate to be η(u) = 0.5, failure rates of subsystem a and b for partial and catastrophic failures be λa = 0.5, λp = 0.25, λc = 0.25, repair rates be фp = фc = фa = ψp =1 and x = y = z = 1. furthermore, if the repair follows exponential distribution then from equation (50), on putting all these values and taking inverse laplace transform one can obtain equations (55) and (56). if the service facility is always available, then expected profit during the interval (0, t] is given by ∫ −= t upp tkdttpkte 0 21 )()( where k1 and k2 are the revenue per unit time and service cost per unit time respectively, (i) in case of gumbel-hougaard copula: e p (t) = k1 [0.001438884273 e (-2.510615836 t)table 11: λ vs. m.t.t.f. for bivariate-clayton copula c -.1970488353e(-1.557348320 t)-0.04529856593 e(-1.425285235 t)+0.002772006242 e(-0.6777682208 t)+ λ mttf c .10 3.195599546 0.008096575381 e(-0.6247959226 t)-17.90055708 .20 2.917391153 e(-.035751979833 t)+18.11764706]-k2t (57) .30 2.743894466 .40 2.625355735 (ii) in case of bivariate clayton copula: e p (t) = k1 [-0.02405078622 e (-2.663714173 t) -.2109447534e(-1.748168308 t)-0.005078743744 .50 2.539232103 .60 2.473826713 e(-1.444003723 t)+0.01361392208 e(-0.6562343416 t)+ .70 2.422466162 0.01375066459 e(-0.6223021709 t)-17.90493737 .80 2.381065192 e(-.03224395049 t)+18.11764707]-k2t (58) .90 2.346982891 keeping k1=1 and varying k2 at 0.1,0.2, 0.3,0.4, 0.5 in equations (57) and (58), one can obtain tables 12 and 13 which is depicted by figs. 13 and 14 respectively. published by atlantis press copyright: the authors 211 analysis of a risky two unit system under marked process incorporating two repairmen with vacations fig. 13: time vs. expected profit for gumbel-hougaard copula table 12: time vs. expected profit for gumbel-hougaard copula fig. 14: time vs. expected profit for bivariate-clayton copula time ep(t) k2=0.1 k2=0.2 k2=0.3 k2=0.4 k2=0.5 0 0 0 0 0 0 1 0.730773 0.630773 0.530773 0.430773 0.330773 2 1.302943 1.102943 0.902943 0.702943 0.502943 3 1.838773 1.538773 1.238773 0.938773 0.638773 4 2.312025 1.912025 1.512025 1.112025 0.712025 5 2.779364 2.779364 1.779364 1.279364 0.779364 6 3.225671 2.625671 2.025671 1.425671 0.825671 7 3.651885 2.951885 2.251885 1.551885 0.851885 8 4.058754 3.258754 2.458754 1.658754 0.858754 9 4.446980 3.546980 2.646980 1.746980 0.846980 10 4.817240 3.817240 2.817240 1.817240 0.817240 table 13: time vs. expected profit for bivariate-clayton copula time ep(t) k2=0.1 k2=0.2 k2=0.3 k2=0.4 k2=0.5 0 0 0 0 0 0 1 0.655671 0.555671 0.455671 0.355671 0.255671 2 1.131750 0.931750 0.731750 0.531750 0.331750 3 1.566394 1.266394 0.966394 0.666394 0.366394 4 1.981209 1.581209 1.181209 0.781209 0.381209 5 2.379751 1.879751 1.379751 0.879751 0.379751 6 2.762780 2.162780 1.562780 0.962780 0.362780 7 3.130691 2.430691 1.730691 1.030691 0.330691 8 3.483874 2.683874 1.883874 1.083874 0.283874 9 3.822742 2.922742 2.022742 1.122742 0.222742 10 4.147719 3.147719 2.147719 1.147719 0.147719 published by atlantis press copyright: the authors 212 n. tiwari, s. b. singh 10. comparison of effect of head of line repair policy and preemptive-resume repair policy on the availability of the system let us take η(u) = 0.5, λa = 0.5, λp = 0.25 and λc = 0.25, фp = фc = фa = ψp =1, θ = 1 and x = y = z = 1. also let the repair follows exponential distribution. using all the above values in equation (50) and modelling the system for head of line repair policy the availability of the system is obtained as pup= -0.06521710750 e (-2.380129270 t) + 0.3245962455 e(-1.712176291 t)cos(0.3939583835 t) +0.1295409494 e(-1.172176291 t) sin(0.3939583835)+ 0.00001576068496 e(-1.083764259 t) +0.6101708864 e (-0.02508722247 t) also from equation (55) the expression for the availability of the system in the case of preemptive resume repair policy is obtained as pup=-0.03612485642 e (-2.510615836 t) + 0.3068736726 e(-1.557348320 t) +0.06456337718 e(-1.425285235 t) – 0.001878777739 e(-0.6777682208 t) -0.005058707285 e(-0.6247959226 t) +0.6716252918 e(-0.03751979833 t) which further yield the following table. table 14: availability vs time in two different repair disciplines fig. 15: availability vs time in two repair disciplines the table 14 and correspondingly fig. 15 represent the variation of availability with respect to time in two repair policies viz. “head of line repair policy” and “preemptive-resume repair policy”. we find that in both the repair policies it decreases as the time increases. it can easily be concluded that the availability in preemptive-resume repair discipline is higher than the head of line repair discipline. hence preemptive resume repair policy should be adopted for repair purposes to obtain optimum returns wherever applicable. 11. comparison of effect of gumbel-hougaard and bivariate clayton copula on different measures of the system from tables 2 and 3 one can easily conclude that the value of availability of the system in both the cases is initially same and decreases as time increases but availability in case of gumbel-hougaard copula is greater than that of bivariate clayton copula. further on comparing reliability of the system given in tables 4 and 5 we observe that reliability of the system also decreases with respect to time but again its value is high for gumbel-hougaard copula. a critical examination of tables 6, 7, 8, 9, 10 and 11 reveal that the m. t. t. f. in case of gumbel-hougaard copula varies from value 4.789-1.020, 3.628-1.280 and 3.320-1.795 with respect to λa, λp and λc respectively as time passes. while in case of bivariate clayton copula it changes from 4.572 to 1.003, from 3.463 to 1.266 and from 3.195 to 2.346 with respect to λa, λp and λc respectively with the increment in time. these tables reveal that initially in each case m. t. t. f. is higher for gumbel-hougaard copula in comparison to bivariate clayton copula. but its decrement is more rapid in case of gumbel-hougaard copula than in case of bivariate clayton copula. finally if we critically examine the expected profit of the system we find from tables 12 and 13 that the maximum and minimum values of expected profit in case of gumbel-hougaard copula are 4.8172 and 0.3307 respectively. while in case of bivariate clayton copula these values are 4.1477 and 0.1477 respectively. so we can say that the profit will be large if we use gumbel-hougaard copula in place of bivariate clayton copula. time dependent up and down state probabilities preemptiveresume repair head of line repair s. n o. p p pti me pup down up down 1 0 1.000000 0 1.000000 0 2 1 0.720476 0.279524 0.709318 0.290682 3 2 0.638253 0.361747 0.611637 0.388363 4 3 0.602853 0.397147 0.573202 0.426798 5 4 0.578305 0.421695 0.553094 0.446906 6 5 0.556635 0.443365 0.538220 0.461780 preemptive resume repair from the above study one can conclude that gumbel-hougaard copula is better than bivariate clayton copula in practical situations. published by atlantis press copyright: the authors 213 analysis of a risky two unit system under marked process incorporating two repairmen with vacations 12. discussion several conclusions may be drawn on the basis of work and results presented in this paper. (1) by the examination of figs. 3, 4, 5 and 6 one can say that when η(u) = 0.5, λa = 0.5, λp = 0.25 and λc = 0.25, then availability and reliability of the system decreases as the time increases. (2) from figs. 13 and 14 one can easily observe that increasing service cost leads decrement in expected profit. (3) the observation of figs. 7, 8, 9, 10, 11 and 12 reveal that m.t.t.f. of the complex system decreases as the value of λa, λp and λc increases. computation also shows that m.t.t.f. depends more on λa in comparison to λp and λc. (4) the comparison of two repair facilities shows that preemptive resume repair policy is better than head of line repair policy for the considered system. (5) comparison of two types of copula viz. gumbelhougaard copula and bivariate clayton copula reveal that gumbel-hougaard copula should be adopted instead of bivariate clayton copula for better results. references bo h lindqvist and tim bedford, the identifiability problem for repairable systems subject to competing risks, applied probability trust. 36(3) (2004) 774-790. p. p. gupta and s. c. agarwal, a parallel redundant complex system with two types of failure under preemptive-repeat repair discipline, microelectron. reliab. 24(3) (1984) 395399. m. ram and s. b. singh, availability, m.t.t.f. and cost analysis of complex system under preemptive repeat repair discipline using gumbel-hougaard family copula, international journal of quality and reliability management (ijqrm). 27(3) (2010) 376-395. r. c. garg and l. r. goel, cost analysis of a system with common cause failure and two types of repair facilities. microelectron. reliab. 25(2) (1985) 281-284. p. p. gupta and m. k. sharma, reliability and m.t.t.f evaluation of a two duplex-unit standby system with two types of repair, microelectron. reliab. 33(3) (1993) 291295. l. r. goel and r. gupta, profit analysis of a cold standby system with two repair distributions, microelectron. reliab. 25(3) (1985) 467-472. hu linmin and li jiandong, reliability analysis of a three unit system with vacation and priority, icic express letters. 3(2) (2009) 171-176. r. k. tuteja and s. c. malik, reliability and profit analysis of two single unit models with three modes and different repair policies of repairmen who appear and disappear randomly, microelectron. reliab, 32(3) (1992) 351-356. a. i. kovalenko and v.p. smolich, reliability analysis for a two element system attended by two repairmen, journal of mathematical sciences. 107(6) (2001) 4443-4447. n. tiwari, s. b. singh, m. ram, reliability analysis of a complex system, attended by two repairmen with vacation under marked process with the application of copula, international journal of reliability and applications. 11(2) (2010)107-122. n. tiwari, suraj bhan singh, analysis of a complex system modelled by a marked point process and assuming vacations for a repairman, economic quality control. 25 (2) (2010) 221–242. m. a. el-damcese, analysis of warm standby systems subject to common-cause failures with time varying failure and repair rates. applied mathematical sciences. 3 (18) (2009) 853-860. a. kumar and s. b. singh, reliability analysis of an n unit parallel standby system under imperfect switching using copula. computer modelling and new technologies. 12(1) (2008) 47-55. r. b. nelson, an introduction to copulas, 2nd edn (springer, newyork 2006). yan jun, enjoy the joy of copula. journal of statistical software. 21(4) (2007) 1-21. published by atlantis press copyright: the authors 214 1. introduction 5.2 initial condition: (14) and other state probabilities are zero at t = 0. the present model has been studied under two different copulas viz. gumbel-hougaard and bivariate clayton. research on the sme’s collateral credit rationing under loan risk compensation mechanism changbing yang 1,2,3 , mu zhang 3 , junmeng lu 1,2,3 1 guizhou institute for urban economics and development, guizhou university of finance and economics, guiyang guizhou 550025, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang guizhou 550025, china 3 school of finance, guizhou university of finance and economics, guiyang guizhou 550025, china e-mail: 1532433977@qq.com, rim_007@163.com, 1532433977@qq.com received december 25, 2017 accepted july 15, 2018 abstract this paper discusses the function of loan risk compensation in the collateral credit rationing model. according to the derivation of my model, we find when the bank have some requirements on interest and collateral, at the same time, we add the loan risk subsidy variable into the above model, loan risk subsidy can lower the bank's collateral requirements for the enterprise, therefore, we can conclude that loan risk compensation can effectively alleviate the credit rationing faced by enterprises. keywords: loan risk compensation, collateral credit rationing model, sme, government 1. lntroduction small and medium sized enterprises of science and technology refer to the small and medium-sized enterprises which rely on a certain number of scientific and technological personnel to engage in scientific and technological research and development activities, obtain their own intellectual property rights and transform them into high and new technology products or services to achieve sustainable development. technology-based smes are an important part of the national economic growth process, and play an extremely important role in the process of promoting economic efficiency and rapid growth. scientific and technological small and medium-sized enterprises are the main force of the national technology renewal and replacement. as premier li keqiang advocated in the report of the government authorities in 2015, "mass entrepreneurship and innovation", the continuous growth of scientific and technological small and medium-sized enterprises can not only promote the higher level of the country's overall technological innovation, but also bring to our country endless employment opportunities and relieve the pressure of employment. however, because the small and medium sized enterprises themselves have a small size and limited investors' registered capital, that is, their own money funds are relatively limited, but in order to maintain technical advantages and provide better customer experience in their own fields, enterprises often have a large amount of investment in the field of research and development. therefore, the enterprises often need to carry out foreign financing. however, because of the small size, the authenticity of the financial data, the low transparency of the financial data, the low downstream customers and the low bargaining power of the upstream suppliers, the smes are difficult to finance in the creditor's rights market and often suffer from the credit rationing of the related financial institutions such as the banks. the market 142 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 environment is difficult. the small and medium sized enterprises of science and technology are an important part of the national economic growth process, and play an important role in the process of promoting economic efficiency and rapid growth. however, there is no uniform standard for the definition of technology-based smes at home and abroad. because of the different national conditions and economic development stages, different countries have different definitions of the happy technology field, even if they are the same, there are also some differences in the definition of happy technology in different stages of economic development in a country. for example, in the early days of our country, under the living standard of the people, local governments are just pursuing the continuous growth of gdp, but there is not enough attention to the protection of the environment and resources. the state has not regarded this field as a happy technology field supported by the state. however, as the living standards of people are gradually increasing, people have put forward more and more demands on the environment of life. countries also regard environment and resources as a happy technology field supported by the state. the division method of scientific and technological small and medium-sized enterprises is different from the division method of the general scientific and technological enterprises, because the main business can be used as a scientific and technological enterprise as long as it can be used as a scientific and technological enterprise as long as it meets the state's newly issued "happy field of national key support". however, the scientific and technological smes are a subdivision of the scientific and technological enterprises. but it still cannot be divorced from the field of technological enterprises. in view of the above facts, in order to clarify the main body of this paper, the author will divide the scientific and technological enterprises and non scientific and technological enterprises from the macro level, which is divided according to the division method of "the technology field of state key support" published by the state. then, from the micro level, the author divides the small and medium-sized enterprises in science and technology in detail according to the newly published evaluation method of small and medium enterprises (2017) 115 of the state science and technology type. on the macro level, according to the new division of national key support technology, the small and medium sized enterprises of science and technology can be divided into eight areas: electronic information, biology and new medicine, aerospace, new materials, high-tech services, new energy sources and energy conservation, resources and environment, advanced manufacturing and automation. in the dry area, because of the space problem, we cannot further explain the above sub fields, and interested readers can search on the internet by themselves. from the micro level, this paper mainly adopted the definition of small and medium sized enterprises in science and technology based on the latest state science and technology based small and medium enterprise evaluation method (2017) 115. that is, the small and medium sized enterprises of science and technology refer to the scientific and technological research and development activities based on a certain number of scientific and technological personnel, and to obtain independent intellectual property rights and convert them into high and new technology products or products services, so as to achieve sustainable development of smes. specifically, only when companies meet the following requirements can they be regarded as a technological sme. (1) registered enterprises in china (excluding hk, macao and taiwan). (2) the total number of employees is no more than 500, the annual sales revenue is no more than 200 million yuan, and the total assets are no more than 200 million yuan. (3) products and services provided by enterprises are not prohibited, restricted or eliminated by the state. (4) the enterprises did not have major safety, major quality accidents, serious environmental violations and serious dishonesty in scientific research for the last year and in the year, and the enterprises were not listed in the list of abnormal business names and serious illegal and trustworthy enterprises. according to the evaluation index of high-tech smes, the score of enterprise income is no less than 60 points, and the score of scientific and technological personnel is not 0 points. 143 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 of course, in the specific implementation of the loan risk compensation fund for small and medium-sized enterprises, the division of science and technology based smes may be slightly different from the above methods, such as the detailed rules for the management of the fund pool management of the loan risk compensation for small and medium-sized enterprises in jiangsu province, jiangsu province (2012) 72) first according to "the printing and issuing of the small and medium-sized enterprises" the notification on the quasi regulation (2011) (no. 300) of the ministry of industry and credit union (no. 300) on the classification standards of small and medium enterprises, divided the enterprises into small and medium enterprises and non medium and small enterprises, and then give priority to the priority support for the small and medium-sized enterprises, high-tech enterprises and private scientific and technological enterprises identified by the relevant departments of the province, and the priority support for the invention patent and the provincial related departments identified by the state. the small and medium-sized enterprises of high and new technology products should be given priority to the enterprises that have been set up or served by the personnel related to the high level talent introduction and training plan of the provinces and provinces, and the small and micro enterprises with independent intellectual property rights are given priority. the loan risk compensation mechanism is an important financial means to encourage and guide banks to increase credit support for technological smes. in comparison with this article, in particular, the loan risk compensation mainly refers to the establishment of a special fund pool by the government, if the bank loans give the government designated small and medium sized enterprises a loss, then the commercial banks can obtain a certain proportion of the interest compensation from the capital pool, thus promoting and encouraging the commercial banks to further relax the loans to small and medium-sized enterprises of science and technology. limit. by observing and studying the loan risk compensation model in different regions, the author finds that the loan risk compensation modes in each region can be attributed to the following categories: the direct compensation model of loan risk: this means that if the bank and other financial institutions lend a loss to the small and medium sized enterprises designated by the government, the commercial banks can obtain a certain proportion of capital compensation from the pool of funds. in general, if the loan is lost, the bank can apply to the government to apply the maximum loss to the government to manage the loan risk compensation funds. the compensation rate is 70% of the principal. this way of loan risk compensation is the most commonly used compensation mode by the government, and it is also the main loan risk compensation mode in this paper. the area that adopts this mode mainly includes jiangsu, zhejiang, shanghai, fujian and so on. the following i will take the detailed rules for the management of the management of the loan risk compensation pool for the loan risk compensation of small and medium-sized enterprises of jiangsu financial institutions (2012) 72) as an example, briefly introducing the method of jiangsu province in the implementation of this policy. first, the capital pool is set up by the financial capital of the city and county, and the provincial finance is matched on a certain scale, among which the provincial finance has the capital pool set up at the provincial level. the maximum amount of supporting funds is not more than 10 million yuan; the capital pool set up for the county (city) is not more than 5 million yuan. secondly, at the beginning of each year, the local financial departments jointly recommend the scientific and technological loan projects with the departments of science and technology and other departments, and the cooperative banks will separate the loans; then, the loans are issued according to the benchmark interest rate published by the people's bank, and the highest proportion of the floating rate shall not exceed 30%; finally, after the confirmation of the loan loss, the conditions for the fund pool to be compensated and the proportion of the pool and bank responsibility share, the maximum proportion of the pool of funds is not more than 70%. in addition, the single household loan does not exceed 20% of the capital pool and is not more than 5 million yuan. the model of loan discount compensation: this model is mainly to apply the interest subsidy to the government when the enterprise is lending money from the bank, that is, the part that the 144 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 enterprise would have paid the interest to the bank is now paid by the government. in addition, the government will also set up a special risk compensation fund. when the enterprise is unable to repay the bank's loan, the bank will obtain the fund from the risk compensation pool. a certain proportion of the loan loss compensation, therefore, this kind of loan risk compensation method is a financial means formed by the two kinds of financial special funds, which are the discount funds and the loan risk compensation funds. this way is directly subsidizing enterprises, and enterprises can get the most real preferential from the government. according to the author's reference to the relevant literature and through the local government portal, the application of this model is relatively wide, but it is mainly in some developed provinces and cities in the east to implement relevant policies, while the underdeveloped areas and other underdeveloped regions in the central and western regions are relatively deficient in the related policy fields. for example, according to the author's internship in the pudong state capital committee of shanghai, the total amount of interest rates of the government in each district of shanghai is not more than 500 thousand yuan, and the time is not more than two years. the main purpose is to reduce the financing cost of the related enterprises and enhance the vitality of the enterprise. in addition, in 2015, dongguan also introduced the "trial measures for the management of credit risk compensation funds and financial discount funds for dongguan" (no. 25) (no. 2015). the details of some details of loan risk compensation funds and discount funds were specified. (1) in the term of loan, the longest term for the enterprise to obtain loans is three years, the shortest period is two years; (2) on the loan amount, the maximum loan amount that the enterprise can obtain from the bank is 15 million yuan; (3) in the compensation for the bank's loan risk, the risk compensation is limited to 10% of the total amount of the total amount of credit funds that the pilot bank actually issued. the total amount of the risk loss is lower than the limit, and the total amount of the risk compensation funds is compensated; the total amount of the risk loss exceeds the limit. it is undertaken by the pilot bank on its own. (4) in terms of discount rate, enterprises can get 1 million yuan discount subsidy, and the interest discount will be completed within two years. of course, the above is only an example of the implementation of the relevant provinces and municipalities on the policy. most provinces and cities will fully consider their own development in the formulation and implementation of policies, and will not completely copy the policies of other families. however, i browse the provinces and cities through the internet. related websites found that although the details of policy formulation, such as the amount and time limit, are different, the total number of them is different. the risk compensation model of loan guarantee agency: this model mainly refers to the risks faced by the government and the guarantee agency when the enterprise is unable to repay the loan. the guarantee agency here not only refers to the professional guarantee corporation, but also includes some financial institutions, such as insurance companies, financial services company, bank and other banks. the main role of the guarantee agency in the loan process is to provide a guarantee for the borrower's loan, that is, once the enterprise guarantees the risk that the enterprise can not repay the loan, the guarantee agency will have the joint liability, the guarantee agency has the risk of repayment of some loans for the enterprise, so that the bank credit risk and the government's compensation pressure can be alleviated at the same time, in addition to this. the insurance company will also earn a certain amount of guarantee. according to the author's understanding, the government used this loan risk compensation model relatively few, only part of guangdong and shanghai adopted this model, because the author only consulted a small number of relevant literature, so may omit the relevant government, so it does not exclude the related areas in addition to parts of shanghai and guangdong also adopted the related models. in 2010, the "notice on the pilot project to carry out the performance assurance insurance of the short term loan performance of technology-based smes" (shanghai co [2010] no. 34), as an example, shanghai has a special scientific and technological performance loan, which mainly includes the four parties, namely, the enterprise itself, the bank, the insurance company and the government, and the enterprises put forward the credit loan application to the relevant government departments and the government 145 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 departments. a dedicated team is dedicated to conducting a due diligence survey on the application enterprise, and then submitted to the decision committee by the completed team to form a written survey report. if the decision committee passes the application of the enterprise, the cooperative bank will make credit lending to the enterprise, that is, it does not require any collateral from the enterprise, but in this process the enterprise must pay a certain guarantee cost to the enterprise. insurance companies, but in order to reduce the financing costs of enterprises and encourage enterprises to repay, the government will have a special financial subsidy fund for the successful repayment of the enterprise, subsidize its 50% guarantee costs. of course, the mode of shanghai also has certain characteristics of second modes, but there are still great differences in essence. in general, the government will not use a single model, because the different loan risk compensation model has its own limitations and the scope of enterprise application. therefore, in general, two or three different pool of funds will be set up by local governments, and two to three different loan risk compensation modes are used in a mixed way. according to the author's internship experience in shanghai pudong sasac, the government of pudong new area has adopted at least two kinds of loan risk compensation models mentioned above, which are the direct loan risk compensation mode and the loan discount compensation model. the former is a technology type small and medium-sized enterprise which can not provide the sufficient collateral to provide the credit loan for obtaining banks. lower bank risks; the latter, in addition to providing credit loans for obtaining bank credit for small and medium sized enterprises that can not provide sufficient value collateral, provides interest subsidy for those capable of obtaining bank loans, thus reducing the operating costs of the enterprises and helping the enterprises to grow better. the above three loan risk compensation models have different characteristics. the first loan risk model is the most commonly used model of loan risk compensation by local governments. the most important feature of the loan risk compensation model is that it can provide mortgage for the small and medium sized enterprises which can not provide the foot value mortgage, and can effectively share the risk of the bank in the loan process. this model mainly deals with those small and medium sized enterprises which can not provide the guarantee of foot value, and can not obtain loans from the banks. it is mainly used in those primary and technological smes, which can be applied to those small and medium sized enterprises with a certain scale, but the effect may not be better. the second model of loan risk compensation, that is, the risk compensation model of the interest discount loan, is mainly applicable to those enterprises with certain scale, which can obtain certain loans from the banks with their own strength, but because the business income and profit rate are not so high, there is a great pressure on the repayment of interest on the interest rate. the risk compensation mode can effectively alleviate the embarrassing situation. the third mode is a bit similar to the first mode, but there are still some differences. the third mode has introduced the third party market organization except the borrower and the borrower the guarantee corporation. this model can effectively help the small and medium sized enterprises to borrow the difficulty of borrowing, and can reduce the pressure of the government's compensation, but it will increase the business operation process and delay the time for the enterprise to get the loan. it is convenient. there is not a surprising difference in the body, and the core is to promote more and more small and medium sized enterprises in the region to gain more capital, so as to gain a better space for growth. under the condition of information asymmetry, if the bank can only adjust the interest rate, that is, the two parties only discuss the level of the interest rate in the credit contract, and there is no demand for the size of the assets, the debt ratio, the collateral and so on. then the credit market will produce a balanced credit rationing even if the external variable is added to the loan risk compensation variable. in the above models, the credit rationing phenomenon will not disappear, but under certain conditions, it can reduce the number of firms that are subject to credit rationing. the main content of the discussion in this paper is that when banks require the interest rate and collateral in the credit contract at the same time, whether the credit rationing phenomenon in the credit market 146 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 will disappear or whether the number of enterprises that have credit rationing will decrease and the exogenous variables of loan risk compensation variables will have an impact on the lending of smes. in real life, in addition to regulating interest rate, banks can also have many other requirements for enterprises in the credit contract, such as assets scale, debt ratio, collateral and so on. as is known to all, mortgage loans are very popular in the modern credit market. in the course of the loan to the enterprise, collateral has also played a vital role. on the one hand, when an enterprise breaks a contract, banks can make up for their losses by selling collateral from enterprises, on the other hand, collateral can have constraints and incentives on the enterprises themselves, and constrain enterprises to dare to default, because once a breach of contract will lose the ownership of the collateral, it will also seriously damage the reputation of the enterprise itself, the damage of credit will undoubtedly increase the cost of foreign financing, especially the cost of credit. in order to complete the contract, the enterprise will work hard for the above two points. in view of the above reality, many scholars at home and abroad have a wide and in-depth study of this field, but because of the difference between the model design and the premise hypothesis, the final conclusions are quite different. through the reading and thinking of the related literature, the author finds that the above theory can be summed up in the following two viewpoints. one is represented by stiglitz and weiss, they think even if the collateral is added to the credit treaty, equilibrium credit rationing will still exist; the other is represented by bester they think if the banks are allowed to adjust interest rates and collateral requirements in the credit contracts at the same time, the credit rationing will not exist and will form a separation equilibrium. stiglitz and weiss (1981) [1] think there are two reasons that lead to a balanced credit rationing in the credit market even if collateral is used as a credit contract clause. for some later researchers, the claim that collateral can be used to distinguish the risk category of the borrower, stiglitz and weiss (1987, 1987, 1992) [2][3][4] have established some theoretical models in the later period and discussed the related problems. he thinks that there are differences between borrowers in many dimensions. it is difficult to distinguish effectively different risk types based on the only feature of collateral. wette (1983) [5] further examines the role of collateral in the balance of credit rationing on the basis of the s-w model. many scholars have given different views on the above views. bester (19851987) [6][7] believes that if the banks are allowed to adjust interest rates and collateral requirements in the competitive credit market, there will be no credit rationing in equilibrium. chan and kanatas (1985) [8] think that mortgage can play a signaling role, that is, it can indirectly alleviate the information asymmetry between borrowers and borrowers. besanko and thakor (1987) [9] assume that only adverse selection exists in the model. the role of moral hazard is zero. banks will consider various factors including interest rate, mainly collateral, in the credit contract. riley (1987) [10] believes that high quality borrowers have a way of signaling the difference between them and low quality borrowers. hung-jen wang (2000) [11] believes that security is positively related to credit risk. to sum up, the author finds that the research of credit rationing by domestic and foreign scholars is mainly focused on debating parties, without considering government intervention. the biggest difference between the author and the previous literature is that when considering both sides of the loan, the third party government loan risk compensation factor is introduced at the same time, and the influence of loan risk compensation factors on the enterprise credit rationing is investigated. the theoretical significance of this paper is to introduce the loan risk compensation variable into the credit rationing model under the condition of information asymmetry. on the one hand, it expands the equilibrium credit rationing theory of asymmetric information conditions; on the other hand, the loan risk compensation funds are mainly funded by the finance, so this paper also has certain significance for the expansion of the theory of fiscal expenditure. therefore, the theoretical significance of this paper is mainly reflected in the above two aspects. the expansion of the theory is mainly based on the above two aspects. 2. model hypothesis each enterprise has its own assets a and is evenly distributed between ],0[ a . 147 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 (1) according to the number of enterprises' own assets, the enterprises can be divided into two kinds of enterprises of different sizes. the first class enterprises j own assets lja , s ka are owned by second types of enterprises k , that is, the size of the first class enterprises is larger than that of the second kinds of enterprises. (2) the credit market is a competitive market, that is, in a balanced state; the bank's profit is zero. (3) both banks and borrowers are risk neutral. (4) the enterprise does not have any liquid assets. if an enterprise only makes loans to the bank, it can invest in technology development projects, and the fixed capital b of each project is fixed. (5) when an enterprise breaks a contract, a bank will bear the corresponding liquidation expenses when handling the assets it collaterals. (6) in order to facilitate the latter model deduction, it is assumed that the liquidation cost is a fixed proportion h of the collateral value , hc is to express the cost of liquidation, 10  h , c is to represent the value of the collateral, 0c , therefore, the value compensation that the bank can obtain after the liquidation of the collateral is ch)1(  , let ht 1 , so the value that the bank can obtain after clearing the collateral is tc , 10  t . (7) e indicates that business operators are in a state of effort to pay more than the operator is in a lazy state; hp indicates that the probability of success of technology development projects is in the state of enterprise managers' efforts; lp indicates the probability of success of technology development projects when managers are lazy, lh pp  , lh ppp  . when the technology development project is successful, the enterprise gains revenue r , and the profit is zero when it fails. there are only two possibilities for technological development projects: success and failure; if the enterprise invests in technology development projects, and managers strive to operate, the following inequality relations are established: rpberp lh  . (8) the cost of obtaining loanable funds of banks is b , that is, the interest rate of depositors in the bank deposit is b , and the profit per unit loan of banks is r , that is, the interest rate of banks is r , let )1( rbh  , )1( rbm  , h indicates that the loan firm returns the bank's principal and interest without breach of contract, and m indicates that the bank will pay the principal and interest of the depositor to obtain the loan. (9) the probability that a bank gives a loan to each applicant is equal. 3. establishment and analysis of credit rationing model a bank provides a loan contract },{ ii cr to the enterprise, },{ ii cr indicate a combination of loan interest rates r and collateral requirements c faced by the borrower. for enterprises, loan contracts must satisfy incentive compatibility constraints and participation constraints. for banks, loan contracts must make bank profits non negative. therefore, a feasible mortgage credit constraint can be expressed in the following formula. for the enterprise: enterprise incentive compatibility constraints: ))(1()())(1()( caphrapecaphrap llhh  (1) the following equation can be derived from the above inequalities: cperh  / (2) when the equal sign is established, the maximum loan interest rate that the enterprise is willing to accept under the incentive compatibility condition of the enterprise can be obtained. the constraints of enterprise participation: aecaphrap hh  ))(1()( (3) the following equation can be derived from the above inequalities: 0)1()(  ecphrp hh (4) for banks, banks only have the constraints of participation and do not have incentive compatibility constraints, that is, the relationship between bank constraints is as follows: 0)1)((  mpbtchp hh  (5)  shows the loan loss readiness rate (constant), and  shows the government's compensation rate for bank loans, ]1,0[ . a feasible constraint on a mortgage, that is, the minimum standard for the borrower to obtain a loan from a bank, which must be met by its own assets: ca  (6) 148 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 since the credit market has been assumed to be a competitive market, the banks are bound to constantly adjust r and c until their expected profit is zero. in equilibrium, they are all satisfied, any bank participation constraint. however, in order to determine the r and c values, we must analyze the constraint conditions of the enterprise, the liquidation of enterprises collateral bank must pay a certain cost, so the relative enterprises direct repayment, the collateral for the banks is certainly better than direct repayment enterprises and banks to maximize their profit, will minimize collateral requirements, increase loan repayment requirements, until it reaches the critical condition of enterprise incentive compatibility, namely when the loan interest rate of 2 type corresponding to an equality on the banks is the optimal interest rate. the equation of the 2 equation is established. at the same time, the expression will be brought into the 5 form after the establishment. at the same time, the equation of the 4.5 form is also established, so we can get the following formula: 0)1)((]/[ 00  mpbtccperp hh  (7) 0c expresses the collateral value of the bank's expected profit at zero, and we can get the following formula according to the 7 form: tpp bpperpm c hh hh )1( )1()/( 0     (8) because 0c , 0)1()/(  bpperpm hh  . according to the 8 form, we can deduce the following two relationships: 0)1( ])1([ )1(/ 2 0     h hh hhh p tpp bppeprpm dt dc  （9） 0 )1( )1(0     tpp bp d dc hh h   (10) by bringing 8 into the 2, the best repayments can be obtained by the bank: tpp bpperpm p e rh hh hh )1( )1()/( 0       (11) according to the discrepancy of the enterprise's own assets, there will be several following cases: when ijac 0 , then, all enterprises will be allocated, that is, no business can get a loan from the bank. when 0c is between i ja , part of the first class enterprises will be rationed, and the other enterprises will get bank loans. the second enterprises will all be rationed. when 0c is between s k a , some second types of enterprises will be rationed, and others can get bank loans. all enterprises can get loans. when 0ca s k  , no business will be subject to credit rationing. in the above assumptions, the author only assumes that banks know the distribution of their own assets, but it is not clear which category each firm belongs to. because this article is about the credit rationing situation of high-tech smes, the author divides second kinds of enterprises into technology smes and non technology smes, and assumes that banks know the specific classification of enterprises in the second categories. according to the characteristics of small and medium-sized enterprises own we know, smes must belong to second types of enterprises, on the one hand, it is the small and medium-sized enterprises, on the other hand it has most of the assets for patents and other intangible assets, cash intangible assets have serious difficulties, therefore, according to the above two reasons the scientific and technological small and medium-sized enterprises are classified into second categories. suppose that tt is the proportion of compensation that a commercial sme can get from its liquidation value when breaking a contract, and the corresponding value of its collateral is tc ; nt indicates that the proportion of compensation that a bank can get from its liquidation value when a non technology sme breaks its contract, the corresponding value of its collateral is nc . through the above analysis, it is known that if banks can provide different types of loan 149 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 contracts to different types of enterprises, loan contracts for non technology smes will be },{ nn cr , and loans for technology smes will be },{ tt cr , which is the best for banks. if the bank for all the two types of enterprises to provide credit contract with },{ nn cr , then the part of smes will be able to obtain loans, but this is not optimal for the banks, because this part of sme default probability is relatively large, and the value of the collateral is relatively small, it will cause the bank could not reach profits; if the bank for all the two types of enterprises to provide credit contract },{ tt cr , there must be a part of the non tech smes are difficult to meet the demands, which is part of the risk level and relative enterprises in small and medium sized enterprises is not high, the bank could have a profit in the enterprises, but as provided by the contract does not lead to its maximization. to sum up, the best case for banks is to provide classified contracts and to provide different types of contracts for different types of enterprises. 4. conclusion this paper discusses the function of loan risk compensation in the collateral credit rationing model. according to the derivation of my model, we find when the bank have some requirements on interest and collateral, at the same time, we add the loan risk subsidy variable into the above model, loan risk subsidy can lower the bank's collateral requirements for the enterprise, therefore, we can conclude that loan risk compensation can effectively alleviate the credit rationing faced by enterprises. references [1] stiglitz j, weiss a. credit rationing in markets with imperfect information[j]. american economic review, 1981,71(3):393-410. [2] stiglitz j, weiss a. credit rationing: reply[j]. the american economic review, 1987,77(1):228-231. [3] stiglitz j, weiss a. macroeconomic equilibrium and credit rationing[r]. nber working paper, 1987, n0. 2164. [4] stiglitz j, weiss a. asymmetric information in credit markets and its implications for macroeconomics[j]. oxford economic papers, 1992,44(2):694-724. [5] wette h. collateral and credit rationing in markets with imperfect information: note[j]. american economic review, 1983,73(3):442-445. [6] bester h. screening vs. rationing in credit markets with imperfect information[j]. american economic review, 1985,75(4):850-855. [7] bester h. the role of collateral in credit market with imperfect information[j]. european economic review,1987,31(2):887-899. [8] chan y, kanatas g. asymmetric valuations and the role of collateral in loan agreements[j]. journal of money credit and banking, 1985,17(1):84-95. [9] besanko d, thakor a v. collateral and rationing: sorting equilibrium in monopolistic and competitive credit markets[j]. international economic review, 1987,28(3):671-689. [10] riley j.g. credit rationing: a further remark[j]. the american economic review, 1987, 77(1):224-227. [11] huang-jen wang. symmetric information and credit rationing: graphical demonstration[j]. financial analysts journal, 2000,56(2):85-95. 150 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 142-150 study on the changes of industrial structure in sichuan province before and after the "5.12 wenchuan earthquake" dongdong zhu 1, xianhua wu 2*, peipei xue 3 1. school of economics and management，southeast university，nanjing 211189，china 2. school of economics and management，shanghai maritime university， shanghai 200135，china 3. school of management engineering，nanjing university of information science & technology， nanjing 210044，china received november 16, 2018 accepted january 5, 2019 abstract on the 10th anniversary of the "5.12 wenchuan earthquake" disaster, what changes have occurred in the industrial structure of sichuan province? it is an important topic for government, industry and the public. this paper analyzed the impact of wenchuan earthquake on various industries in sichuan province by using the input-output table, deviation share analysis and structural decomposition. the results show that: (1) the growth rate of various industries is higher than the national average. (2) the output induced by final demand in the region accounts for a large proportion of the total output, and the output induced by the export demand accounts for a relatively small proportion. among them, the technical coefficient and the final demand coefficient have a greater impact on the regional final demand-induced output, and the export coefficient significantly affects the export-induced output. (3) in the mid-disaster stage, the purchasing power coefficient promotes the increase of output of the primary industry and most of the secondary industries. the final demand coefficient promotes the increase of output of the secondary industry, and the technical coefficient contributes greatly to the increase of output of the tertiary industry. (4) in the post-disaster period, the purchasing power coefficient and technical coefficient promoted the growth of the secondary industry, but the impact on the primary industry was not significant; the final demand coefficient promoted the growth of the tertiary industry, and the technical coefficient was reversed. finally, the corresponding policy recommendations were put forward. this paper enriches the study of disaster economics and provides an empirical reference for emergency and aid construction after the disaster. keywords: wenchuan earthquake ， input-output table ， shift-share method ， structural decomposition method *corresponding author: wxhua_77@ nuist.edu.cn 基金项目：国家自然科学基金重大研发计划培育项目：“支持应急决策的气象灾害大数据融合的方法研究” (91546117)；国家自然科学基金面上项目：“支持应急联动政策设计的气象灾害间接经济损失评估的方法研 究”(71373131)；国家社科重大招标项目：“基于大数据融合的气象灾害应急管理研究”(16zda047)。 study on the changes of industrial structure in sichuan province before and after the "5.12 wenchuan earthquake" dongdong zhu 1, xianhua wu 2*, peipei xue 3 1. school of economics and management，southeast university，nanjing 211189，china 2. school of economics and management，shanghai maritime university， shanghai 200135，china 3. school of management engineering，nanjing university of information science & technology， nanjing 210044，china received november 16, 2018 accepted january 5, 2019 abstract on the 10th anniversary of the "5.12 wenchuan earthquake" disaster, what changes have occurred in the industrial structure of sichuan province? it is an important topic for government, industry and the public. this paper analyzed the impact of wenchuan earthquake on various industries in sichuan province by using the input-output table, deviation share analysis and structural decomposition. the results show that: (1) the growth rate of various industries is higher than the national average. (2) the output induced by final demand in the region accounts for a large proportion of the total output, and the output induced by the export demand accounts for a relatively small proportion. among them, the technical coefficient and the final demand coefficient have a greater impact on the regional final demand-induced output, and the export coefficient significantly affects the export-induced output. (3) in the mid-disaster stage, the purchasing power coefficient promotes the increase of output of the primary industry and most of the secondary industries. the final demand coefficient promotes the increase of output of the secondary industry, and the technical coefficient contributes greatly to the increase of output of the tertiary industry. (4) in the post-disaster period, the purchasing power coefficient and technical coefficient promoted the growth of the secondary industry, but the impact on the primary industry was not significant; the final demand coefficient promoted the growth of the tertiary industry, and the technical coefficient was reversed. finally, the corresponding policy recommendations were put forward. this paper enriches the study of disaster economics and provides an empirical reference for emergency and aid construction after the disaster. keywords: wenchuan earthquake ， input-output table ， shift-share method ， structural decomposition method *corresponding author: wxhua_77@ nuist.edu.cn 基金项目：国家自然科学基金重大研发计划培育项目：“支持应急决策的气象灾害大数据融合的方法研究” (91546117)；国家自然科学基金面上项目：“支持应急联动政策设计的气象灾害间接经济损失评估的方法研 究”(71373131)；国家社科重大招标项目：“基于大数据融合的气象灾害应急管理研究”(16zda047)。 journal of risk analysis and crisis response vol. 9(1), march (2019), pp. 11–19 doi: https://doi.org/10.2991/jracr.b.190328.002; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 的变化情况进行了详细的分析，并研究了灾 区产业结构的变化情况 10。akhtar and santos 利用动态非正常投入产出模型评估了飓风 灾害造成的各部门经济损失，并对各部门的 敏感性进行了排序 11。吴先华等分别采用非 正常投入产出模型和可计算一般均衡模型 研究了灾害对各产业的影响等 12,13。但目前 很少有文献研究灾区产业结构在灾害前后 的变化情况，且对产业结构变化的原因探究 也不够深入，难以把握灾害的影响范围和变 化规律。 基于以上考虑，本文以汶川地震灾害为 例，运用偏离份额分析法、结构分解法，从 动态角度分析汶川地震对四川省各产业的 影响。一方面丰富了灾害经济学相关的研究 文献，另一方面可为灾后重建政策的完善提 供实证参考，具有重要的理论和现实意义。 2. 数据和方法说明 2.1 方法说明 2.1.1. 偏离份额分析法 将区域某时期经济总量的变动分解为 地区增长份额（n）、产业结构偏离份额（p） 和区位特征偏离份额（r），通过这些指标评 估研究区域的经济结构和产业竞争优势。具 体过程如下。 首先，将区域经济产出（gi）在某个时 间段的动态变化分解为 3 个变量，分别为地 区增长份额（ni）、产业结构偏离份额（pi） 和区位特征偏离份额（ri）,具体公式为: 𝐺𝐺𝑖𝑖 = 𝑥𝑥𝑖𝑖1 − 𝑥𝑥𝑖𝑖0 = ∆𝑥𝑥𝑖𝑖 (1) 𝐺𝐺𝑖𝑖 = ∆𝑥𝑥𝑖𝑖 = 𝑁𝑁𝑖𝑖 + 𝑃𝑃𝑖𝑖 + 𝑅𝑅𝑖𝑖 (2) 其中，ni、pi、ri 为： 𝑁𝑁𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑋𝑋1/𝑋𝑋0 − 1) (3) 𝑃𝑃𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑋𝑋𝑖𝑖1/𝑋𝑋𝑖𝑖0 − 𝑋𝑋1/𝑋𝑋0) (4) 𝑅𝑅𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑥𝑥𝑖𝑖1/𝑥𝑥𝑖𝑖0 − 𝑋𝑋𝑖𝑖1/𝑋𝑋𝑖𝑖0) (5) 式中，𝑥𝑥𝑖𝑖1和𝑥𝑥𝑖𝑖0分别对应研究区域第 i 产业的 报告期和基期产出，𝑋𝑋1和𝑋𝑋0分别对应参照 区域报告期和基期的总产出，𝑋𝑋𝑖𝑖1和𝑋𝑋𝑖𝑖0分别 对应参照区域第 i 产业的报告期和基期产出。 ni、pi、ri 的经济意义如下：ni 为地区增长 份额，表示如果研究区域与参照区域各产业 的经济产出总值同比例增长时，到报告期实 现的增长量。对 ni 值的正负性进行研究， 从而得出研究地区与参照地区各产业的整 体增长速度差异。即如果 ni 值大于 0，那么 研究区域产业的实际增长水平快于参照地 区，反之则小于 0。pi 为产业结构偏离份额， 指因研究区域内某产业的比重与全国相应 产业的比重差异而引起该产业的增长相对 于全国平均值所产生的偏差，pi 的值越大， 说明产业结构对经济产出总量增长的贡献 越大。ri 为区位特征偏离份额，表示研究区 域各个产业的基期产值，按照研究地区第 i 产业实际增长率计算的增长额，与按照参照 地区第 i 产业增长率计算的增长额之间的差 额，即该地区与全国的同一产业因增长速度 不同而引起的偏差，反映了研究地区该产业 相对于全国的竞争力。ri 的值越大，表明该 产业具有相对竞争优势，对经济增长的作用 就越大。 2.1.2. 结构分解法 通过对投入产出表进行结构分解，能够 取得关于经济结构变化的信息。基本思路是 通过对多种因素的分解，如技术变化、需求 变化、贸易模式变化等，对产业结构变化进 行解释。我国的投入产出表是竞争型投入产 出模型，在假设进口产品与国内产品互为完 全替代品的情况下，投入产出表的恒等关系 可以改写成： 𝑥𝑥 + 𝑚𝑚 = 𝐴𝐴𝑥𝑥 + 𝑓𝑓 + 𝑒𝑒 (6) 式中，x表示总产出列向量，m为进口列 向量（包括本地区从国内其他省份的调入和 从国外的进口），a是直接消耗系数矩阵，f 表示最终产出列向量。e为出口列向量（包 括本地对国内其他省份的调出和对国外的 出口）。 定 义 进 口 系 数 为 对 角 矩 阵 𝑀𝑀𝑖𝑖 = “5.12 汶川地震”前后四川省产业结构的变化研究 朱冬冬 1，吴先华 2*，薛佩佩 3 1. 东南大学/经济管理学院，南京 211189，中国 2. 上海海事大学/经济管理学院，上海 200135，中国 3. 南京信息工程大学/管理工程学院，南京 210044，中国 摘要：在“5.12 汶川地震”灾害发生 10 周年之际，探讨四川省的产业结构发生了哪些变化？ 是政府、产业界和社会公众关注的重要话题。本文利用投入产出表，采用偏离份额分析法和 结构分解法，分析了汶川地震对四川省各产业的影响。结果表明：（1）各产业的增长速度高 于全国平均水平。（2）区域内最终需求诱发的产出在总产出中占比较大，出口需求诱发的产 出占比较小。其中，技术系数、最终需求系数对区域最终需求诱发产出的影响较大，出口系 数显著影响了出口需求诱发的产出。（3）灾中阶段，购买力系数促进了第一产业及大部分第 二产业产出的增加，最终需求系数促进了第二产业产出的增加，技术系数对第三产业产出的 增加贡献较大。（4）灾后阶段，购买力系数、技术系数促进了第二产业的增长，但对第一产 业的影响不显著；最终需求系数促进了第三产业的增长，技术系数则相反。最后提出了相应 的政策建议。本文丰富了灾害经济学研究，可为灾后的应急和援建工作提供实证参考。 关键词：汶川地震；投入产出表；偏离份额分析法；结构分解法 1. 引言 频繁发生的自然灾害对人类社会生活 造成了严重影响。其中，地震灾害由于其突 发性强、破坏性大、次生灾害多、灾后重建 周期长等特点，对人类社会造成的影响尤为 严重。20 世纪以来，中国共发生了近 800 次 6 级以上地震，死亡人数超过 59 万。2008 年 5 月 12 日，我国四川省汶川县发生 8.0 级地震，受灾范围包括四川、甘肃、陕西等 6 个省的 237 个县（市、区），面积达 10 万 多平方公里，造成 69227 人遇难、374643 人受伤、17923 人失踪，直接经济损失达 8451 亿。面对严峻灾情，中央政府提出“对口援 建”政策，采取“恢复与发展相融合”的整 体跨越式重建之路，使灾区经济在较短的时 间内得到恢复。但是，在灾害重建前后，四 川省的产业结构发生了哪些变化？哪些因 素促进了这种变化？鲜见相关研究。 近年来，许多学者研究了灾害对产业结 构的影响。如 skidmore and toya、cuaresma 和 cavallo et al.提出灾害会导致产业结构发 生改变 1-3。okuyama 认为，灾害不仅会导 致产业结构变化，而且会进一步影响受灾地 区经济的长期增长 4。但这种影响有多大？ 持续时间有多长？学者们的观点不一。如 fujiki and cheng 认为灾害的影响是短暂的 5。 chang 通过采用简单的指标衡量地震后神户 市的恢复情况。结果表明，灾后重建需要注 入 3-4 年的经济收益，且恢复后的经济水平 比灾前大约低 10% 6。dupont and noy 采用 计量经济模型，分析了 1995 年神户地震后 灾区经济的长期走势。研究显示，地震对该 地区的经济产生了持续不利的影响 7。李卫 江等从产业网络中灾害扩散的角度，对其造 成的经济损失类型及研究进展进行了系统 梳理，评估了产业空间网络风险 8。学者们 还研究了灾害对农业等具体产业的影响。如 klomp and hoogezand 认为自然灾害会提高 对农业贸易的管制水平 9。okuyama 利用神 户地区的投入产出表，对神户地震后各产业 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 12 的变化情况进行了详细的分析，并研究了灾 区产业结构的变化情况 10。akhtar and santos 利用动态非正常投入产出模型评估了飓风 灾害造成的各部门经济损失，并对各部门的 敏感性进行了排序 11。吴先华等分别采用非 正常投入产出模型和可计算一般均衡模型 研究了灾害对各产业的影响等 12,13。但目前 很少有文献研究灾区产业结构在灾害前后 的变化情况，且对产业结构变化的原因探究 也不够深入，难以把握灾害的影响范围和变 化规律。 基于以上考虑，本文以汶川地震灾害为 例，运用偏离份额分析法、结构分解法，从 动态角度分析汶川地震对四川省各产业的 影响。一方面丰富了灾害经济学相关的研究 文献，另一方面可为灾后重建政策的完善提 供实证参考，具有重要的理论和现实意义。 2. 数据和方法说明 2.1 方法说明 2.1.1. 偏离份额分析法 将区域某时期经济总量的变动分解为 地区增长份额（n）、产业结构偏离份额（p） 和区位特征偏离份额（r），通过这些指标评 估研究区域的经济结构和产业竞争优势。具 体过程如下。 首先，将区域经济产出（gi）在某个时 间段的动态变化分解为 3 个变量，分别为地 区增长份额（ni）、产业结构偏离份额（pi） 和区位特征偏离份额（ri）,具体公式为: 𝐺𝐺𝑖𝑖 = 𝑥𝑥𝑖𝑖1 − 𝑥𝑥𝑖𝑖0 = ∆𝑥𝑥𝑖𝑖 (1) 𝐺𝐺𝑖𝑖 = ∆𝑥𝑥𝑖𝑖 = 𝑁𝑁𝑖𝑖 + 𝑃𝑃𝑖𝑖 + 𝑅𝑅𝑖𝑖 (2) 其中，ni、pi、ri 为： 𝑁𝑁𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑋𝑋1/𝑋𝑋0 − 1) (3) 𝑃𝑃𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑋𝑋𝑖𝑖1/𝑋𝑋𝑖𝑖0 − 𝑋𝑋1/𝑋𝑋0) (4) 𝑅𝑅𝑖𝑖 = 𝑥𝑥𝑖𝑖0(𝑥𝑥𝑖𝑖1/𝑥𝑥𝑖𝑖0 − 𝑋𝑋𝑖𝑖1 /𝑋𝑋𝑖𝑖0) (5) 式中，𝑥𝑥𝑖𝑖1和𝑥𝑥𝑖𝑖0分别对应研究区域第 i 产业的 报告期和基期产出，𝑋𝑋1和𝑋𝑋0分别对应参照 区域报告期和基期的总产出，𝑋𝑋𝑖𝑖1和𝑋𝑋𝑖𝑖0分别 对应参照区域第 i 产业的报告期和基期产出。 ni、pi、ri 的经济意义如下：ni 为地区增长 份额，表示如果研究区域与参照区域各产业 的经济产出总值同比例增长时，到报告期实 现的增长量。对 ni 值的正负性进行研究， 从而得出研究地区与参照地区各产业的整 体增长速度差异。即如果 ni 值大于 0，那么 研究区域产业的实际增长水平快于参照地 区，反之则小于 0。pi 为产业结构偏离份额， 指因研究区域内某产业的比重与全国相应 产业的比重差异而引起该产业的增长相对 于全国平均值所产生的偏差，pi 的值越大， 说明产业结构对经济产出总量增长的贡献 越大。ri 为区位特征偏离份额，表示研究区 域各个产业的基期产值，按照研究地区第 i 产业实际增长率计算的增长额，与按照参照 地区第 i 产业增长率计算的增长额之间的差 额，即该地区与全国的同一产业因增长速度 不同而引起的偏差，反映了研究地区该产业 相对于全国的竞争力。ri 的值越大，表明该 产业具有相对竞争优势，对经济增长的作用 就越大。 2.1.2. 结构分解法 通过对投入产出表进行结构分解，能够 取得关于经济结构变化的信息。基本思路是 通过对多种因素的分解，如技术变化、需求 变化、贸易模式变化等，对产业结构变化进 行解释。我国的投入产出表是竞争型投入产 出模型，在假设进口产品与国内产品互为完 全替代品的情况下，投入产出表的恒等关系 可以改写成： 𝑥𝑥 + 𝑚𝑚 = 𝐴𝐴𝑥𝑥 + 𝑓𝑓 + 𝑒𝑒 (6) 式中，x表示总产出列向量，m为进口列 向量（包括本地区从国内其他省份的调入和 从国外的进口），a是直接消耗系数矩阵，f 表示最终产出列向量。e为出口列向量（包 括本地对国内其他省份的调出和对国外的 出口）。 定 义 进 口 系 数 为 对 角 矩 阵 𝑀𝑀𝑖𝑖 = “5.12 汶川地震”前后四川省产业结构的变化研究 朱冬冬 1，吴先华 2*，薛佩佩 3 1. 东南大学/经济管理学院，南京 211189，中国 2. 上海海事大学/经济管理学院，上海 200135，中国 3. 南京信息工程大学/管理工程学院，南京 210044，中国 摘要：在“5.12 汶川地震”灾害发生 10 周年之际，探讨四川省的产业结构发生了哪些变化？ 是政府、产业界和社会公众关注的重要话题。本文利用投入产出表，采用偏离份额分析法和 结构分解法，分析了汶川地震对四川省各产业的影响。结果表明：（1）各产业的增长速度高 于全国平均水平。（2）区域内最终需求诱发的产出在总产出中占比较大，出口需求诱发的产 出占比较小。其中，技术系数、最终需求系数对区域最终需求诱发产出的影响较大，出口系 数显著影响了出口需求诱发的产出。（3）灾中阶段，购买力系数促进了第一产业及大部分第 二产业产出的增加，最终需求系数促进了第二产业产出的增加，技术系数对第三产业产出的 增加贡献较大。（4）灾后阶段，购买力系数、技术系数促进了第二产业的增长，但对第一产 业的影响不显著；最终需求系数促进了第三产业的增长，技术系数则相反。最后提出了相应 的政策建议。本文丰富了灾害经济学研究，可为灾后的应急和援建工作提供实证参考。 关键词：汶川地震；投入产出表；偏离份额分析法；结构分解法 1. 引言 频繁发生的自然灾害对人类社会生活 造成了严重影响。其中，地震灾害由于其突 发性强、破坏性大、次生灾害多、灾后重建 周期长等特点，对人类社会造成的影响尤为 严重。20 世纪以来，中国共发生了近 800 次 6 级以上地震，死亡人数超过 59 万。2008 年 5 月 12 日，我国四川省汶川县发生 8.0 级地震，受灾范围包括四川、甘肃、陕西等 6 个省的 237 个县（市、区），面积达 10 万 多平方公里，造成 69227 人遇难、374643 人受伤、17923 人失踪，直接经济损失达 8451 亿。面对严峻灾情，中央政府提出“对口援 建”政策，采取“恢复与发展相融合”的整 体跨越式重建之路，使灾区经济在较短的时 间内得到恢复。但是，在灾害重建前后，四 川省的产业结构发生了哪些变化？哪些因 素促进了这种变化？鲜见相关研究。 近年来，许多学者研究了灾害对产业结 构的影响。如 skidmore and toya、cuaresma 和 cavallo et al.提出灾害会导致产业结构发 生改变 1-3。okuyama 认为，灾害不仅会导 致产业结构变化，而且会进一步影响受灾地 区经济的长期增长 4。但这种影响有多大？ 持续时间有多长？学者们的观点不一。如 fujiki and cheng 认为灾害的影响是短暂的 5。 chang 通过采用简单的指标衡量地震后神户 市的恢复情况。结果表明，灾后重建需要注 入 3-4 年的经济收益，且恢复后的经济水平 比灾前大约低 10% 6。dupont and noy 采用 计量经济模型，分析了 1995 年神户地震后 灾区经济的长期走势。研究显示，地震对该 地区的经济产生了持续不利的影响 7。李卫 江等从产业网络中灾害扩散的角度，对其造 成的经济损失类型及研究进展进行了系统 梳理，评估了产业空间网络风险 8。学者们 还研究了灾害对农业等具体产业的影响。如 klomp and hoogezand 认为自然灾害会提高 对农业贸易的管制水平 9。okuyama 利用神 户地区的投入产出表，对神户地震后各产业 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 12 13 前相比，大部分产业灾后增长较慢，而第三 产业则保持比灾前更高的速度增长。具体请 见图 1 所示。 图 1 四川省 2002-2007 年、2007-2010 年、2010-2012 年各产业增长速度对比 注：2002-2007 投入产出表中公共管理和社会组织业数值为 0，故未对该产业增速进行分析。 3.2. 2002-2012 年间各产业的偏离份额分析 从灾前阶段来看，各产业的地区增长份 额均大于 0，表明各产业的增速高于全国平 均水平。其中，第一产业具有明显的竞争优 势，第二产业对经济的贡献大于第三产业。 具体而言，农林牧渔业等产业的结构偏离份 额小于 0，表明这些产业对四川省经济增长 的促进作用不明显。化学工业等产业的结构 偏离份额大于 0，表明这些产业对四川省经 济增长有明显的促进作用。其中，非金属矿 物制品业和金属冶炼及压延加工业对经济 增长的贡献较大。此外，农林牧渔业等产业 的区位特征偏离份额大于 0，相对全国平均 水平，这些产业具有较好的竞争力。但是非 金属矿采选业等产业的区位特征偏离份额 小于 0，表明相对于全国而言，这些产业没 有明显的竞争优势。具体请见图 2。 图 2 四川省 2002-2007 年各产业产出增长水平的偏离份额 注：（1）2002-2007 投入产出表中公共管理和社会组织业数值为 0，故未对该产业产出增长水平的偏 离份额进行分析。（2）r(xi)：区位特征偏离份额，p(xi)：产业结构偏离份额，n(xi)：地区增长份额。 -0,2 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 增 长 率 产业编号 2002-2007 2007-2010 2010-2012 -4000000 -2000000 0 2000000 4000000 6000000 8000000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 单 位 ： 万 元 产业编号 r(xi) p(xi) n(xi) 𝑚𝑚𝑖𝑖 /(𝐴𝐴𝐴𝐴 + 𝑓𝑓)，其中mi为第i产业的进口列向 量。则(6)式可变为： (𝐼𝐼 − (𝐼𝐼 − 𝑀𝑀𝑖𝑖)𝐴𝐴)𝐴𝐴 = (𝐼𝐼 − 𝑀𝑀𝑖𝑖)𝑓𝑓 + 𝑒𝑒 (7) 定义𝐷𝐷 = 𝐼𝐼 − 𝑀𝑀𝑖𝑖为区域购买力系数，即 为满足每单位区域产品需求所消耗的区域 内资源，则(7)式可变为： 𝐴𝐴 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1(𝐷𝐷𝑓𝑓 + 𝑒𝑒) (8) 根据(8)式可将总产出列向量分成两部 分：区域内最终需求诱发的产出和出口需求 诱发的产出，分别用𝑋𝑋𝑟𝑟和𝑋𝑋𝑒𝑒表示，如下式所 示： 𝑋𝑋𝑟𝑟 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1𝐷𝐷𝑓𝑓 (9) 𝑋𝑋𝑒𝑒 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1𝑒𝑒 (10) 为进一步分析哪些因素对总产出的增 长变化产生影响，将𝑋𝑋𝑟𝑟和xe进行分解。定义 𝐹𝐹 = 𝐷𝐷𝑓𝑓，上标 1、0 分别表示报告期和基期： 𝑋𝑋𝑟𝑟1 𝑋𝑋𝑟𝑟0 = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 (11) 定义𝐿𝐿𝑟𝑟 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1，运用两极分解法， 综合基期和报告期的分解，(11)式可转换为： 𝑥𝑥𝑟𝑟1 𝑥𝑥𝑟𝑟0 = √(𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴1)−1𝐹𝐹1 × (𝐼𝐼−𝐷𝐷 1𝐴𝐴0 )−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 × (𝐼𝐼−𝐷𝐷 0𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0)−1𝐹𝐹1 × (𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷1𝐴𝐴0)−1𝐹𝐹0 × 𝐿𝐿𝑟𝑟 0𝐹𝐹1 𝐿𝐿𝑟𝑟0𝐹𝐹0 × 𝐿𝐿𝑟𝑟 1𝐹𝐹1 𝐿𝐿𝑟𝑟1𝐹𝐹0 (12) 定义𝐷𝐷1∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1 )−1 𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴1 )−1 𝐹𝐹1 ，𝐷𝐷0∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴0)−1 𝐹𝐹0 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 ， 𝐴𝐴1∗ = (𝐼𝐼−𝐷𝐷 0𝐴𝐴1 )−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹1 ，𝐴𝐴0∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1 )−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷1𝐴𝐴0 )−1𝐹𝐹0 ，𝐹𝐹1∗ = 𝐿𝐿𝑟𝑟1 𝐹𝐹1 𝐿𝐿𝑟𝑟1 𝐹𝐹0 ，𝐹𝐹0∗ = 𝐿𝐿𝑟𝑟 0 𝐹𝐹1 𝐿𝐿𝑟𝑟0 𝐹𝐹0 ，则(6)式可转化为： 𝑋𝑋𝑟𝑟1 𝑋𝑋𝑟𝑟0 √𝐷𝐷1∗ ∗ 𝐷𝐷0∗ ∗ 𝐴𝐴1∗ ∗ 𝐴𝐴0∗ ∗ 𝐹𝐹1∗ ∗ 𝐹𝐹0∗ (13) 利用两极分解法，如下所示： 𝑑𝑑𝐷𝐷 = 1 2 ∗ (∆𝐷𝐷𝑖𝑖𝑟𝑟 − ∆𝐷𝐷𝑖𝑖𝑁𝑁)(∆𝐴𝐴𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 + ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (14) 𝑑𝑑𝐹𝐹 = 1 2 ∗ (∆𝐹𝐹𝑖𝑖𝑟𝑟 − ∆𝐹𝐹𝑖𝑖𝑁𝑁)(∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 + ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑟𝑟 ) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (15) 𝑑𝑑𝐴𝐴 = 1 2 ∗ (∆𝐴𝐴𝑖𝑖𝑟𝑟 − ∆𝐴𝐴𝑖𝑖𝑁𝑁)(∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁 + ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 ) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (16) 则(5)式可转换为： 𝑅𝑅𝑖𝑖𝑟𝑟 = 𝑋𝑋𝑟𝑟𝑖𝑖 0 (∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐴𝐴𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 − ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) (17) 根据同样的转换方法也可以对 𝑋𝑋𝑒𝑒1 𝑋𝑋𝑒𝑒0 进行 拆分，得到下式： 𝑅𝑅𝑖𝑖𝑒𝑒 = 𝑋𝑋𝑒𝑒𝑖𝑖 0 (∆𝐷𝐷𝑖𝑖𝑒𝑒 ∗ ∆𝐴𝐴𝑖𝑖𝑒𝑒 ∗ ∆𝑒𝑒𝑖𝑖𝑒𝑒 − ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) (18) 式中，∆d、∆a、∆f、∆e 分别代表区域 购买力系数变化、技术系数变化、区域最终 需求系数变化以及出口系数变化。 2.2 数据说明 本文主要用到与投入产出表相关的数 据。由于 2002 年前后的中国和各省的投入 产出表的分类差异较大，故采用了 2002、 2007、2010 和 2012 年的投入产出表。2012 年投入产出表的产业分类是按照 2011 年的 《国民经济行业分类》（gb/t4754-2011）进 行的，这与 2002、2007 和 2010 年采用的分 类标准略有不同。因此本文对一些产业进行 了归类和合并，如将通用、专用设备制造业、 仪器仪表及文化办公用机械制造业合并为 一个产业，将造纸印刷和文教体育用品、其 他制造产品和废品废料合并为一个产业，科 学研究和技术服务合并为一个产业等。 3. 实证分析 3.1 2002-2012 年间不同阶段各产业增长速 度对比 将各产业总产出作为衡量经济增长的 指标，2002-2007 作为灾前阶段、2007-2010 作为灾中阶段、2010-2012 作为灾后阶段， 对这三个阶段的各产业经济增长速度进行 比较。总体来看，第一产业在灾中阶段保持 快速增长，但灾后并未保持高速增长的态势。 第二产业在这次地震灾害中受损严重，与灾 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 14 前相比，大部分产业灾后增长较慢，而第三 产业则保持比灾前更高的速度增长。具体请 见图 1 所示。 图 1 四川省 2002-2007 年、2007-2010 年、2010-2012 年各产业增长速度对比 注：2002-2007 投入产出表中公共管理和社会组织业数值为 0，故未对该产业增速进行分析。 3.2. 2002-2012 年间各产业的偏离份额分析 从灾前阶段来看，各产业的地区增长份 额均大于 0，表明各产业的增速高于全国平 均水平。其中，第一产业具有明显的竞争优 势，第二产业对经济的贡献大于第三产业。 具体而言，农林牧渔业等产业的结构偏离份 额小于 0，表明这些产业对四川省经济增长 的促进作用不明显。化学工业等产业的结构 偏离份额大于 0，表明这些产业对四川省经 济增长有明显的促进作用。其中，非金属矿 物制品业和金属冶炼及压延加工业对经济 增长的贡献较大。此外，农林牧渔业等产业 的区位特征偏离份额大于 0，相对全国平均 水平，这些产业具有较好的竞争力。但是非 金属矿采选业等产业的区位特征偏离份额 小于 0，表明相对于全国而言，这些产业没 有明显的竞争优势。具体请见图 2。 图 2 四川省 2002-2007 年各产业产出增长水平的偏离份额 注：（1）2002-2007 投入产出表中公共管理和社会组织业数值为 0，故未对该产业产出增长水平的偏 离份额进行分析。（2）r(xi)：区位特征偏离份额，p(xi)：产业结构偏离份额，n(xi)：地区增长份额。 -0,2 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 增 长 率 产业编号 2002-2007 2007-2010 2010-2012 -4000000 -2000000 0 2000000 4000000 6000000 8000000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 单 位 ： 万 元 产业编号 r(xi) p(xi) n(xi) 𝑚𝑚𝑖𝑖 /(𝐴𝐴𝐴𝐴 + 𝑓𝑓)，其中mi为第i产业的进口列向 量。则(6)式可变为： (𝐼𝐼 − (𝐼𝐼 − 𝑀𝑀𝑖𝑖)𝐴𝐴)𝐴𝐴 = (𝐼𝐼 − 𝑀𝑀𝑖𝑖)𝑓𝑓 + 𝑒𝑒 (7) 定义𝐷𝐷 = 𝐼𝐼 − 𝑀𝑀𝑖𝑖为区域购买力系数，即 为满足每单位区域产品需求所消耗的区域 内资源，则(7)式可变为： 𝐴𝐴 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1(𝐷𝐷𝑓𝑓 + 𝑒𝑒) (8) 根据(8)式可将总产出列向量分成两部 分：区域内最终需求诱发的产出和出口需求 诱发的产出，分别用𝑋𝑋𝑟𝑟和𝑋𝑋𝑒𝑒表示，如下式所 示： 𝑋𝑋𝑟𝑟 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1𝐷𝐷𝑓𝑓 (9) 𝑋𝑋𝑒𝑒 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1𝑒𝑒 (10) 为进一步分析哪些因素对总产出的增 长变化产生影响，将𝑋𝑋𝑟𝑟和xe进行分解。定义 𝐹𝐹 = 𝐷𝐷𝑓𝑓，上标 1、0 分别表示报告期和基期： 𝑋𝑋𝑟𝑟1 𝑋𝑋𝑟𝑟0 = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 (11) 定义𝐿𝐿𝑟𝑟 = (𝐼𝐼 − 𝐷𝐷𝐴𝐴)−1，运用两极分解法， 综合基期和报告期的分解，(11)式可转换为： 𝑥𝑥𝑟𝑟1 𝑥𝑥𝑟𝑟0 = √(𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴1)−1𝐹𝐹1 × (𝐼𝐼−𝐷𝐷 1𝐴𝐴0 )−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 × (𝐼𝐼−𝐷𝐷 0𝐴𝐴1)−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0)−1𝐹𝐹1 × (𝐼𝐼−𝐷𝐷 1𝐴𝐴1)−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷1𝐴𝐴0)−1𝐹𝐹0 × 𝐿𝐿𝑟𝑟 0𝐹𝐹1 𝐿𝐿𝑟𝑟0𝐹𝐹0 × 𝐿𝐿𝑟𝑟 1𝐹𝐹1 𝐿𝐿𝑟𝑟1𝐹𝐹0 (12) 定义𝐷𝐷1∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1 )−1 𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴1 )−1 𝐹𝐹1 ，𝐷𝐷0∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴0)−1 𝐹𝐹0 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹0 ， 𝐴𝐴1∗ = (𝐼𝐼−𝐷𝐷 0𝐴𝐴1 )−1𝐹𝐹1 (𝐼𝐼−𝐷𝐷0𝐴𝐴0 )−1𝐹𝐹1 ，𝐴𝐴0∗ = (𝐼𝐼−𝐷𝐷 1𝐴𝐴1 )−1𝐹𝐹0 (𝐼𝐼−𝐷𝐷1𝐴𝐴0 )−1𝐹𝐹0 ，𝐹𝐹1∗ = 𝐿𝐿𝑟𝑟1 𝐹𝐹1 𝐿𝐿𝑟𝑟1 𝐹𝐹0 ，𝐹𝐹0∗ = 𝐿𝐿𝑟𝑟 0 𝐹𝐹1 𝐿𝐿𝑟𝑟0 𝐹𝐹0 ，则(6)式可转化为： 𝑋𝑋𝑟𝑟1 𝑋𝑋𝑟𝑟0 √𝐷𝐷1∗ ∗ 𝐷𝐷0∗ ∗ 𝐴𝐴1∗ ∗ 𝐴𝐴0∗ ∗ 𝐹𝐹1∗ ∗ 𝐹𝐹0∗ (13) 利用两极分解法，如下所示： 𝑑𝑑𝐷𝐷 = 1 2 ∗ (∆𝐷𝐷𝑖𝑖𝑟𝑟 − ∆𝐷𝐷𝑖𝑖𝑁𝑁)(∆𝐴𝐴𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 + ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (14) 𝑑𝑑𝐹𝐹 = 1 2 ∗ (∆𝐹𝐹𝑖𝑖𝑟𝑟 − ∆𝐹𝐹𝑖𝑖𝑁𝑁)(∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 + ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑟𝑟 ) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (15) 𝑑𝑑𝐴𝐴 = 1 2 ∗ (∆𝐴𝐴𝑖𝑖𝑟𝑟 − ∆𝐴𝐴𝑖𝑖𝑁𝑁)(∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁 + ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 ) ∗ 𝐴𝐴𝑟𝑟𝑖𝑖 0 (16) 则(5)式可转换为： 𝑅𝑅𝑖𝑖𝑟𝑟 = 𝑋𝑋𝑟𝑟𝑖𝑖 0 (∆𝐷𝐷𝑖𝑖𝑟𝑟 ∗ ∆𝐴𝐴𝑖𝑖𝑟𝑟 ∗ ∆𝐹𝐹𝑖𝑖𝑟𝑟 − ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) (17) 根据同样的转换方法也可以对 𝑋𝑋𝑒𝑒1 𝑋𝑋𝑒𝑒0 进行 拆分，得到下式： 𝑅𝑅𝑖𝑖𝑒𝑒 = 𝑋𝑋𝑒𝑒𝑖𝑖 0 (∆𝐷𝐷𝑖𝑖𝑒𝑒 ∗ ∆𝐴𝐴𝑖𝑖𝑒𝑒 ∗ ∆𝑒𝑒𝑖𝑖𝑒𝑒 − ∆𝐷𝐷𝑖𝑖𝑁𝑁 ∗ ∆𝐴𝐴𝑖𝑖𝑁𝑁 ∗ ∆𝐹𝐹𝑖𝑖𝑁𝑁) (18) 式中，∆d、∆a、∆f、∆e 分别代表区域 购买力系数变化、技术系数变化、区域最终 需求系数变化以及出口系数变化。 2.2 数据说明 本文主要用到与投入产出表相关的数 据。由于 2002 年前后的中国和各省的投入 产出表的分类差异较大，故采用了 2002、 2007、2010 和 2012 年的投入产出表。2012 年投入产出表的产业分类是按照 2011 年的 《国民经济行业分类》（gb/t4754-2011）进 行的，这与 2002、2007 和 2010 年采用的分 类标准略有不同。因此本文对一些产业进行 了归类和合并，如将通用、专用设备制造业、 仪器仪表及文化办公用机械制造业合并为 一个产业，将造纸印刷和文教体育用品、其 他制造产品和废品废料合并为一个产业，科 学研究和技术服务合并为一个产业等。 3. 实证分析 3.1 2002-2012 年间不同阶段各产业增长速 度对比 将各产业总产出作为衡量经济增长的 指标，2002-2007 作为灾前阶段、2007-2010 作为灾中阶段、2010-2012 作为灾后阶段， 对这三个阶段的各产业经济增长速度进行 比较。总体来看，第一产业在灾中阶段保持 快速增长，但灾后并未保持高速增长的态势。 第二产业在这次地震灾害中受损严重，与灾 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 14 15 求诱发的产出的影响较为明显，表明区域购 买力和直接消耗对各产业最终需求诱发的 产出影响较大。其中，区域最终需求系数对 第二产业影响最大，在最终需求诱发产出中 占据了很大比重，这反映出区域购买力是拉 动第二产业产出增加的重要因素，而第三产 业却相反。但第三产业中技术系数贡献明显， 影响程度远大于区域最终需求系数，说明技 术系数对于促进第三产业的发展起到了积 极作用。区域购买力系数对第一产业、第二 产业影响较大，对第三产业影响较小。 具体而言，与2007-2010灾中阶段相比， 食品制造及烟草加工业等产业的区域最终 需求系数由正变负，变化幅度较大，而金属 冶炼及压延加工业等产业的区域最终需求 系数由负变正。农林牧渔业等产业的区域购 买力系数由正变负，表明区域内产品消耗系 数减小，抑制了这些产业的区域内最终需求 诱发的产出。农林牧渔业等产业的技术系数 变为负值，但煤炭开采和洗选业等产业的技 术系数变为正值，通用、专用设备制造业、 仪器仪表及文化办公用机械制造业的技术 系数明显增大，交通运输、仓储和邮政的技 术系数明显减小。此次地震事件中，区域购 买力系数、技术系数、区域最终需求系数的 变化对最终需求诱发的产出产生了十分明 显的影响。 再从出口需求诱发产出的影响因素来 看，在灾中阶段，出口系数拉动了第一产业 出口需求诱发产出的增长。对纺织业等产业 起到了积极的作用，但食品制造及烟草加工 业等产业却相反。区域购买力系数、技术系 数对出口需求诱发产出的影响较小，技术系 数促进了食品制造及烟草加工业等产业的 出口需求诱发产出的增长，煤炭开采和洗选 业等产业的出口系数变为负值，但交通运输 设备制造业等产业的出口系数变为正值。煤 炭开采和洗选业等产业的技术系数值显著 减小，各产业的区域购买力系数较为稳定。 可见出口系数对各产业的出口需求诱发产 出的影响十分明显。 从偏离份额分析和结构分解的结果来 看，汶川地震对四川各产业的影响十分明显， 对总体的产业结构也产生了重要的影响。如 上所述，在2007-2010灾中阶段和2010-2012 灾后阶段，各产业的产出增速呈现出一直下 降、一直增长、先增后降、先降后增的不同 发展趋势，各产业对经济的贡献度以及竞争 力也发生了不同的变化。 4. 结论与建议 本文基于对投入产出表的分析，运用偏 离份额分析法、结构分解法对汶川地震后四 川省各产业的变化情况做了详细的研究。 结果表明，第一，各产业的增长速度变 化不一，各产业产出的增长速度高于全国平 均水平，但灾后第一产业的竞争力减弱，第 三产业的竞争力增强。第二，区域内最终需 求诱发的产出占总产出的比重较大，出口需 求诱发的产出占总产出的比重较小。技术系 数、区域最终需求系数对总产出中区域内最 终需求诱发的产出的影响较大，出口系数对 各出口需求诱发的产出影响十分明显。第三， 灾前阶段，区域购买力系数促进了第一产业 及第二产业中大部分产业的区域内最终需 求诱发产出的增加，区域最终需求系数拉动 第二产业尤其是制造业的增加，技术系数对 第三产业的区域内最终需求诱发的产出贡 献巨大。灾后阶段，区域购买力系数、技术 系数抑制了第一产业的增长，技术系数拉动 了第二产业区域内最终需求诱发产出的增 长，区域最终需求系数促进了第三产业区域 内最终需求的诱发产出的发展，技术系数对 第三产业的区域内最终需求诱发的产出起 反作用。第四，第一产业的感应度系数和影 响力系数降幅明显，第三产业的关联性相对 较弱，第二产业则表现出较高的相关性，但 第二产业内部产业的关联性差异化明显。 从灾中阶段来看，各产业的地区增长份 额大于 0。第一产业仍然具有竞争优势，但 此次地震灾害对第二产业的负面影响较大， 第三产业对经济增长的拉动作用增强。相比 灾前阶段而言，石油和天然气开采业等产业 的产业结构偏离份额由正变负，表明地震使 得这些产业的比重下降，对经济增长的贡献 减弱。建筑业等产业的产业结构偏离份额由 负变正，即这些产业的比重增加。观察各产 业的区位特征偏离份额可以发现，非金属矿 采选业等产业的区位特征偏离份额由负变 正，表明这些产业在地震中受损严重，竞争 力减弱，对经济产出的拉动作用减弱。但燃 气生产和供应业等产业的区位特征偏离份 额由正变负，表明这些产业的竞争力下降。 在灾后阶段，各产业的地区增长份额大 于 0，高于全国的平均水平增长。第一产业 的竞争力大幅削弱；第二产业的恢复较慢， 在第二产业中，不同产业对经济增长所起作 用的差异较大；第三产业的恢复较快。具体 而言，灾后阶段，煤炭开采和洗选业等产业 的产业结构偏离份额由正变负，表明灾后这 些产业在总产业中的相对份额降低。而农林 牧渔业等产业的产业结构偏离份额由负变 正，表明这些产业相对份额变大。各产业的 区位特征偏离份额相对稳定。农林牧渔业和 金融保险业的区位特征偏离份额由正变负， 表明这两个产业的竞争优势减弱；燃气生产 和供应业等产业的区位特征偏离份额由负 变正，表明这些产业的竞争力增强。 总体而言，第一产业在灾中阶段保持高 速增长，但灾后增速放缓，相对竞争优势减 弱。第二产业整体受损严重，相比灾前，第 二产业对经济增长的促进作用减弱，但第二 产业中的产业在不同阶段的结构偏离份额、 区位特征偏离份额差异明显，这表明这些产 业在不同阶段对四川的经济发展起到了不 同作用。第三产业的增速保持快速增长，大 部分产业的竞争力增强，促进了经济的发展。 3.3. 最终需求和出口需求诱发的产出变化 在灾中阶段，第二产业的区域内最终需 求诱发的产出所占比重较大，且高于第三产 业。具体而言，农林牧渔业等产业的区域内 最终需求和出口需求的诱发产出大于 0，表 明这些产业对各产业发展起促进作用。但电 气、机械及器材制造业等产业的区域内最终 需求和出口需求的诱发产出小于 0，表明这 些产业的促进作用不明显。此外，食品制造 及烟草加工业等产业的区域内最终需求诱 发的产出大于 0、出口需求诱发的产出小于 0，表明区域内最终需求诱发的产出促进了 总产出的增加，但出口需求却相反。而纺织 业等产业的区域内最终需求诱发的产出小 于 0、出口需求诱发的产出大于 0。 在灾后阶段，区域内最终需求诱发的产 出在总产出中仍占据很大比重，出口需求诱 发的产出所占比重较小。可见灾后区域内最 终需求促进了大部分产业产出的增加。其中， 农林牧渔业等产业的出口需求诱发的产出 由正变负，表明这些产业的产出水平受出口 因素的影响减弱。食品制造及烟草加工业等 产业的出口需求诱发的产出由负变正，表明 出口促进了这些产业的发展。此外，农林牧 渔业等产业的区域内最终需求诱发的产出 由正变负，表明内需对产出的促进作用减弱。 纺织业等产业的区域内最终需求诱发的产 出的值由负变正，表明区域内需求对这些产 业产出的增长起到了促进作用。 3.4. 最终需求和出口需求诱发的结构分解 此外，本文还分析了各产业的区域购买 力系数、技术系数、区域最终需求、出口系 数对区域内最终需求诱发产出和出口需求 诱发产出的影响。在灾中阶段和灾后阶段， 区域最终需求和技术系数的变化对最终需 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 16 求诱发的产出的影响较为明显，表明区域购 买力和直接消耗对各产业最终需求诱发的 产出影响较大。其中，区域最终需求系数对 第二产业影响最大，在最终需求诱发产出中 占据了很大比重，这反映出区域购买力是拉 动第二产业产出增加的重要因素，而第三产 业却相反。但第三产业中技术系数贡献明显， 影响程度远大于区域最终需求系数，说明技 术系数对于促进第三产业的发展起到了积 极作用。区域购买力系数对第一产业、第二 产业影响较大，对第三产业影响较小。 具体而言，与2007-2010灾中阶段相比， 食品制造及烟草加工业等产业的区域最终 需求系数由正变负，变化幅度较大，而金属 冶炼及压延加工业等产业的区域最终需求 系数由负变正。农林牧渔业等产业的区域购 买力系数由正变负，表明区域内产品消耗系 数减小，抑制了这些产业的区域内最终需求 诱发的产出。农林牧渔业等产业的技术系数 变为负值，但煤炭开采和洗选业等产业的技 术系数变为正值，通用、专用设备制造业、 仪器仪表及文化办公用机械制造业的技术 系数明显增大，交通运输、仓储和邮政的技 术系数明显减小。此次地震事件中，区域购 买力系数、技术系数、区域最终需求系数的 变化对最终需求诱发的产出产生了十分明 显的影响。 再从出口需求诱发产出的影响因素来 看，在灾中阶段，出口系数拉动了第一产业 出口需求诱发产出的增长。对纺织业等产业 起到了积极的作用，但食品制造及烟草加工 业等产业却相反。区域购买力系数、技术系 数对出口需求诱发产出的影响较小，技术系 数促进了食品制造及烟草加工业等产业的 出口需求诱发产出的增长，煤炭开采和洗选 业等产业的出口系数变为负值，但交通运输 设备制造业等产业的出口系数变为正值。煤 炭开采和洗选业等产业的技术系数值显著 减小，各产业的区域购买力系数较为稳定。 可见出口系数对各产业的出口需求诱发产 出的影响十分明显。 从偏离份额分析和结构分解的结果来 看，汶川地震对四川各产业的影响十分明显， 对总体的产业结构也产生了重要的影响。如 上所述，在2007-2010灾中阶段和2010-2012 灾后阶段，各产业的产出增速呈现出一直下 降、一直增长、先增后降、先降后增的不同 发展趋势，各产业对经济的贡献度以及竞争 力也发生了不同的变化。 4. 结论与建议 本文基于对投入产出表的分析，运用偏 离份额分析法、结构分解法对汶川地震后四 川省各产业的变化情况做了详细的研究。 结果表明，第一，各产业的增长速度变 化不一，各产业产出的增长速度高于全国平 均水平，但灾后第一产业的竞争力减弱，第 三产业的竞争力增强。第二，区域内最终需 求诱发的产出占总产出的比重较大，出口需 求诱发的产出占总产出的比重较小。技术系 数、区域最终需求系数对总产出中区域内最 终需求诱发的产出的影响较大，出口系数对 各出口需求诱发的产出影响十分明显。第三， 灾前阶段，区域购买力系数促进了第一产业 及第二产业中大部分产业的区域内最终需 求诱发产出的增加，区域最终需求系数拉动 第二产业尤其是制造业的增加，技术系数对 第三产业的区域内最终需求诱发的产出贡 献巨大。灾后阶段，区域购买力系数、技术 系数抑制了第一产业的增长，技术系数拉动 了第二产业区域内最终需求诱发产出的增 长，区域最终需求系数促进了第三产业区域 内最终需求的诱发产出的发展，技术系数对 第三产业的区域内最终需求诱发的产出起 反作用。第四，第一产业的感应度系数和影 响力系数降幅明显，第三产业的关联性相对 较弱，第二产业则表现出较高的相关性，但 第二产业内部产业的关联性差异化明显。 从灾中阶段来看，各产业的地区增长份 额大于 0。第一产业仍然具有竞争优势，但 此次地震灾害对第二产业的负面影响较大， 第三产业对经济增长的拉动作用增强。相比 灾前阶段而言，石油和天然气开采业等产业 的产业结构偏离份额由正变负，表明地震使 得这些产业的比重下降，对经济增长的贡献 减弱。建筑业等产业的产业结构偏离份额由 负变正，即这些产业的比重增加。观察各产 业的区位特征偏离份额可以发现，非金属矿 采选业等产业的区位特征偏离份额由负变 正，表明这些产业在地震中受损严重，竞争 力减弱，对经济产出的拉动作用减弱。但燃 气生产和供应业等产业的区位特征偏离份 额由正变负，表明这些产业的竞争力下降。 在灾后阶段，各产业的地区增长份额大 于 0，高于全国的平均水平增长。第一产业 的竞争力大幅削弱；第二产业的恢复较慢， 在第二产业中，不同产业对经济增长所起作 用的差异较大；第三产业的恢复较快。具体 而言，灾后阶段，煤炭开采和洗选业等产业 的产业结构偏离份额由正变负，表明灾后这 些产业在总产业中的相对份额降低。而农林 牧渔业等产业的产业结构偏离份额由负变 正，表明这些产业相对份额变大。各产业的 区位特征偏离份额相对稳定。农林牧渔业和 金融保险业的区位特征偏离份额由正变负， 表明这两个产业的竞争优势减弱；燃气生产 和供应业等产业的区位特征偏离份额由负 变正，表明这些产业的竞争力增强。 总体而言，第一产业在灾中阶段保持高 速增长，但灾后增速放缓，相对竞争优势减 弱。第二产业整体受损严重，相比灾前，第 二产业对经济增长的促进作用减弱，但第二 产业中的产业在不同阶段的结构偏离份额、 区位特征偏离份额差异明显，这表明这些产 业在不同阶段对四川的经济发展起到了不 同作用。第三产业的增速保持快速增长，大 部分产业的竞争力增强，促进了经济的发展。 3.3. 最终需求和出口需求诱发的产出变化 在灾中阶段，第二产业的区域内最终需 求诱发的产出所占比重较大，且高于第三产 业。具体而言，农林牧渔业等产业的区域内 最终需求和出口需求的诱发产出大于 0，表 明这些产业对各产业发展起促进作用。但电 气、机械及器材制造业等产业的区域内最终 需求和出口需求的诱发产出小于 0，表明这 些产业的促进作用不明显。此外，食品制造 及烟草加工业等产业的区域内最终需求诱 发的产出大于 0、出口需求诱发的产出小于 0，表明区域内最终需求诱发的产出促进了 总产出的增加，但出口需求却相反。而纺织 业等产业的区域内最终需求诱发的产出小 于 0、出口需求诱发的产出大于 0。 在灾后阶段，区域内最终需求诱发的产 出在总产出中仍占据很大比重，出口需求诱 发的产出所占比重较小。可见灾后区域内最 终需求促进了大部分产业产出的增加。其中， 农林牧渔业等产业的出口需求诱发的产出 由正变负，表明这些产业的产出水平受出口 因素的影响减弱。食品制造及烟草加工业等 产业的出口需求诱发的产出由负变正，表明 出口促进了这些产业的发展。此外，农林牧 渔业等产业的区域内最终需求诱发的产出 由正变负，表明内需对产出的促进作用减弱。 纺织业等产业的区域内最终需求诱发的产 出的值由负变正，表明区域内需求对这些产 业产出的增长起到了促进作用。 3.4. 最终需求和出口需求诱发的结构分解 此外，本文还分析了各产业的区域购买 力系数、技术系数、区域最终需求、出口系 数对区域内最终需求诱发产出和出口需求 诱发产出的影响。在灾中阶段和灾后阶段， 区域最终需求和技术系数的变化对最终需 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 16 17 附录 附表 1 各产业说明 序号 部门 1 农林牧渔业 2 煤炭开采和洗选业 3 石油和天然气开采业 4 金属矿采选业 5 非金属矿及其他矿采选业 6 食品制造及烟草加工业 7 纺织业 8 纺织服装鞋帽皮革羽绒及其制品业 9 木材加工及家具制造业 10 石油加工、炼焦及核燃料加工业 11 化学工业 12 非金属矿物制品业 13 金属冶炼及压延加工业 14 金属制品业 15 通用、专用设备制造业、仪器仪表及文化办公用机械制造业 16 交通运输设备制造业 17 电气、机械及器材制造业 18 通信设备、计算机及其他电子设备制造业 19 造纸印刷和文教体育用品、其他制造产品、废品废料 20 电力、热力的生产和供应业 21 燃气生产和供应业 22 水的生产和供应业 23 建筑业 24 批发和零售贸易业 25 交通运输、仓储和邮政 26 住宿和餐饮业 27 信息传输、计算机服务和软件业 28 金融保险业 29 房地产业 30 租赁和商务服务业 31 科学研究和技术服务 32 水利、环境和公共设施管理业 33 居民服务和其他服务业 34 教育 35 卫生、社会保障和社会福利业 36 文化、体育和娱乐业 37 公共管理和社会组织 基于以上研究，可提出以下建议： 首先，灾后应继续保持对第一产业的援 助，确保第一产业的竞争优势，更好地为其 他产业的发展提供基础性支持。对第二产业 实行有差别的援助政策，精确扶助相关产业， 利用第二产业的高影响力，充分发挥其对经 济增长的拉动作用。同时，应当注重利用第 三产业推动社会经济的全面进步。兼顾各方 利益、统筹各行业的协调发展。 其次，区域内最终需求和出口需求对产 业竞争力的影响较大，灾后部分产业内需疲 软，区域内消耗的减弱对第一产业的影响较 大。因此应当注重扩大内需，增强产业竞争 力，促进该部分产业对经济的拉动作用。同 时，应发掘各产业的出口潜力，通过增加出 口促进各产业的发展。还应注重技术系数的 变化对第三产业的负面影响，提高第一、二 产业的技术创新能力，促进各产业均衡协调 发展。 参考文献 1. m. skidmore, h. toya. do natural disasters promote long-run growth? j. economic inquiry, 40(4) (2010) 664-687. 2. j. c. cuaresma, j. hlouskova, m. bersteiner. natural disasters as creative destruction? evidence from developing countries, j. economic inquiry, 46(2) (2010) 214-226. 3. e. cavallo, s. galiani, i. noy, et al.. catastrophic natural disasters and economic growth, j . research department publications, 95(5) (2013) 1549-1561. 4. y. okuyama. economics of natural disasters: a critical review, j. research paper, (2003). 5. h. fujiki, h. cheng. disentangling the effects of multiple treatments—measuring the net economic impact of the 1995 great hanshin-awaji earthquake, j. journal of econometrics, 186(1) (2015) 66-73. 6. s. e. chang. urban disaster recovery: a measurement framework and its application to the 1995 kobe earthquake, j. disasters, 34(2) (2010) 03-327. 7. w. dupont, i. noy. what happened to kobe? a reassessment of the impact of the 1995 earthquake in japan, j. economic development & cultural change, 63(4) (2015) 777-812. 8. w. li, j. wen, x. li. progress of research on economic loss assessment of disasters in industrial networks, j. progress in geography, 37(3) (2018) 330-341. 李卫江, 温家洪, 李仙德. 产业网络灾害经济 损失评估研究进展[j]. 地理科学进展, 2018, 37(3):330-341.. 9. j. klomp, b. hoogezand. natural disasters and agricultural protection: a panel data analysis, j. world development, 104 (2018) 404-417. 10. y. okuyama, j. r. santos. disaster impact and input–output analysis, j. economic systems research, 26(1) (2014) 1-12. 11. r. akhtar, j. r. santos. risk-based input–output analysis of hurricane impacts on interdependent regional workforce systems, j. natural hazards, 65(1) (2013) 391-405. 12. l. zhou, x. wu, g. gao. analysis of indirect economic loss of meteorological disasters among the belt and road typical countries based on the mrio model——taking the china’s typhoon rammasun in 2014 as an example, j. journal of natural disasters, 10 (2018) 1-10. 周蕾，吴先华，高歌. 基于 mrio 模型的“一带 一路”国家间的气象灾害间接损失分析——以 2014 年中国“威尔逊”台风灾害为例[j]，自然灾 害学报，2018(10):1-10. 13. x. wu, l. tan, j. guo, l. zhou. how much damage does resilience reduce: an empirical study based on improved cge model, j. journal of management sciences in china, 7 (2018) 66-76. 吴先华, 谭玲, 郭际,等. 恢复力减少了灾害的 多少损失——基于改进 cge 模型的实证研究 [j]. 管理科学学报, 2018(7): 66-76. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 18 附录 附表 1 各产业说明 序号 部门 1 农林牧渔业 2 煤炭开采和洗选业 3 石油和天然气开采业 4 金属矿采选业 5 非金属矿及其他矿采选业 6 食品制造及烟草加工业 7 纺织业 8 纺织服装鞋帽皮革羽绒及其制品业 9 木材加工及家具制造业 10 石油加工、炼焦及核燃料加工业 11 化学工业 12 非金属矿物制品业 13 金属冶炼及压延加工业 14 金属制品业 15 通用、专用设备制造业、仪器仪表及文化办公用机械制造业 16 交通运输设备制造业 17 电气、机械及器材制造业 18 通信设备、计算机及其他电子设备制造业 19 造纸印刷和文教体育用品、其他制造产品、废品废料 20 电力、热力的生产和供应业 21 燃气生产和供应业 22 水的生产和供应业 23 建筑业 24 批发和零售贸易业 25 交通运输、仓储和邮政 26 住宿和餐饮业 27 信息传输、计算机服务和软件业 28 金融保险业 29 房地产业 30 租赁和商务服务业 31 科学研究和技术服务 32 水利、环境和公共设施管理业 33 居民服务和其他服务业 34 教育 35 卫生、社会保障和社会福利业 36 文化、体育和娱乐业 37 公共管理和社会组织 基于以上研究，可提出以下建议： 首先，灾后应继续保持对第一产业的援 助，确保第一产业的竞争优势，更好地为其 他产业的发展提供基础性支持。对第二产业 实行有差别的援助政策，精确扶助相关产业， 利用第二产业的高影响力，充分发挥其对经 济增长的拉动作用。同时，应当注重利用第 三产业推动社会经济的全面进步。兼顾各方 利益、统筹各行业的协调发展。 其次，区域内最终需求和出口需求对产 业竞争力的影响较大，灾后部分产业内需疲 软，区域内消耗的减弱对第一产业的影响较 大。因此应当注重扩大内需，增强产业竞争 力，促进该部分产业对经济的拉动作用。同 时，应发掘各产业的出口潜力，通过增加出 口促进各产业的发展。还应注重技术系数的 变化对第三产业的负面影响，提高第一、二 产业的技术创新能力，促进各产业均衡协调 发展。 参考文献 1. m. skidmore, h. toya. do natural disasters promote long-run growth? j. economic inquiry, 40(4) (2010) 664-687. 2. j. c. cuaresma, j. hlouskova, m. bersteiner. natural disasters as creative destruction? evidence from developing countries, j. economic inquiry, 46(2) (2010) 214-226. 3. e. cavallo, s. galiani, i. noy, et al.. catastrophic natural disasters and economic growth, j . research department publications, 95(5) (2013) 1549-1561. 4. y. okuyama. economics of natural disasters: a critical review, j. research paper, (2003). 5. h. fujiki, h. cheng. disentangling the effects of multiple treatments—measuring the net economic impact of the 1995 great hanshin-awaji earthquake, j. journal of econometrics, 186(1) (2015) 66-73. 6. s. e. chang. urban disaster recovery: a measurement framework and its application to the 1995 kobe earthquake, j. disasters, 34(2) (2010) 03-327. 7. w. dupont, i. noy. what happened to kobe? a reassessment of the impact of the 1995 earthquake in japan, j. economic development & cultural change, 63(4) (2015) 777-812. 8. w. li, j. wen, x. li. progress of research on economic loss assessment of disasters in industrial networks, j. progress in geography, 37(3) (2018) 330-341. 李卫江, 温家洪, 李仙德. 产业网络灾害经济 损失评估研究进展[j]. 地理科学进展, 2018, 37(3):330-341.. 9. j. klomp, b. hoogezand. natural disasters and agricultural protection: a panel data analysis, j. world development, 104 (2018) 404-417. 10. y. okuyama, j. r. santos. disaster impact and input–output analysis, j. economic systems research, 26(1) (2014) 1-12. 11. r. akhtar, j. r. santos. risk-based input–output analysis of hurricane impacts on interdependent regional workforce systems, j. natural hazards, 65(1) (2013) 391-405. 12. l. zhou, x. wu, g. gao. analysis of indirect economic loss of meteorological disasters among the belt and road typical countries based on the mrio model——taking the china’s typhoon rammasun in 2014 as an example, j. journal of natural disasters, 10 (2018) 1-10. 周蕾，吴先华，高歌. 基于 mrio 模型的“一带 一路”国家间的气象灾害间接损失分析——以 2014 年中国“威尔逊”台风灾害为例[j]，自然灾 害学报，2018(10):1-10. 13. x. wu, l. tan, j. guo, l. zhou. how much damage does resilience reduce: an empirical study based on improved cge model, j. journal of management sciences in china, 7 (2018) 66-76. 吴先华, 谭玲, 郭际,等. 恢复力减少了灾害的 多少损失——基于改进 cge 模型的实证研究 [j]. 管理科学学报, 2018(7): 66-76. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 11–19 18 19 received 13 february 2016 accepted 1 march 2016 the effects of the social context on pre-decisional processes of protective action in beijing communities tiezhong liu, hubo zhang, xiaowei li, huiru li school of management and economics, beijing institute of technology, beijing 100081, china abstract this paper examined the effects of the social context (media communication, face-to-face communication, and friend and kinship network) on pre decisional processes of individual protective action. a two-stage survey was carried out aimed at residents in beijing communities in half a year, and then the data were disposed from three aspects: common method variance, non-parametric test, and exploratory factor analysis. based on the pretest data (n=344) and the posttest data (n=351), analyses were implemented from aspects of competing model analysis, hypothesis path test, and multiple group analysis with the method of structural equation modeling. results: first, media communication activities can influence individual reception of risk information, which influence is the strongest among social context variables; moreover education attainment and age can influence it significantly. then media channel should be diverse with respect to the different groups in order to increase the degree of information reception. second, face-to-face communication activities can influence individual attention behavior; at the same time, it can also influence media communication; in addition, education attainment and income level are can influence it significantly. then physical activities should be specific with respect to different socioeconomic strata in order to raise the degree of information attention. third, friendly and kinship network does not influence any pre decisional processes variable, but it can influence face-to-face communication instead. then it can be seen that the role of social conformity should be emphasized in order to enhance the degree of individual participation. fourth, no social context variable can exert significant influence on comprehension behavior. then it can be seen that the literal problem should be emphasized in order to improve the degree of understanding of risk communication. future research should cover the particular social context as regards demographic characteristics, the facilitating and barriers inside risk information, and the relationship between social conformity and risk communication. keywords: risk communication; health behavior; social influence; community involvement; demographic characteristics 1. introduction people usually acquire environmental risk information from direct observation or indirect perception. when the physical environment is complex or rapidly changing and supplies only subtle cues for predicting threats to people and property, physical reality testing tends to give way to social reality testing (lindell and perry, 2004, p.27). for example, although the public might not really understand complex technology, a series of environmental problems have become the hot topics of china, such as px factories and inland nuclear power plants as a result of widespread media coverage (zhu, 2014). in addition, disaster cues, such as earthquake cannot even predict by scientists, let alone lay people. local people and organizations are always the main actors in disaster risk reduction activities (maskrey, 1989). therefore, since the public usually lacks direct ability to perceive hazard from their physical environment, it is important to discuss the role of social context on individual behavior. in theory, such topics belong to the field of risk communication (health & o’hair, 2009) and health behavior theory (hayden, 2009). based on previous theories (such as shannon, 1948; lasswell, 1948), lindell and perry (2004, 2012) have developed a journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 21-30 published by atlantis press copyright: the authors 21 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities protective action decision model (padm) to describe the information source and individual protection action. in which they state that: risk information initiates a series of pre decisional processes, elicits core perceptions, and then motivates protective actions (p. 48). it can be seen that pre decisional processes acts as the mediator between the social context and individual protection action. although the primary focus of the padm is on decision-making processes that take place in conscious awareness (lindell and perry, p.48), but the relationship between the social context and pre decisional processes has not been explained in detail, as well as later studies about the application of the padm, such as terpstra and lindel (2012) and wang (2014). thus, this paper contributes to the discussion about the effects of social context on pre decisional processes. the focus will be put on three aspects: to begin with, we will discuss the interactions between different social context variables; furthermore, we will discuss the specific relationships between social cues (such as friends, kinships, media) and individual psychological processes before decision-making; finally, we will discuss the variance of significant paths across demographic groups in order to clarify the circumstance of given connections. the discussion will be useful for perfecting the theory of padm; at the same time, the clarification of social context will be helpful for advancing the coverage and effectiveness of risk communication. 2. methodology 2.1variables 2.1.1 independent variables social context variables were designated as independent variables. the padm offers two kinds of social context in long-term hazard adjustment: friend and kinship networks and community networks (lindell & perry, 2004, p.139). since the social context plays the sources of risk information, community networks can be represented by individual preference of the communication channel in the community. in view of risk communication activities in chinese community, media style (such as tv, radio, and newspaper) and physical activities (such as class, poster and emergency exercise), three social context variables were designed as follows. first, media communication (mc).media is the common channel for residents to acquire risk information, which are controlled by central and municipal government. receiving information from media is self-driven behavior, so the influence of media communication can be seen as intrinsic motivation (lazarus &folkman, 1984). this path is the same with the central route of elaboration likelihood model (elm), in which individual attention behavior is influenced more by the cognitive contents of the message (petty & cacioppo, 1986). therefore mc was designed to be measured by residents’ attitude toward different media channel. second, face-to-face communication (ffc). there are many activities related to risk communication in chinese community. generally, the participation of these activities depends on compulsory or incentive factors, which reflects individual compliance with the authority or power (lazarus& folkman, 1984). this path is the same with the peripheral route of elm, in which individual attention behavior is influenced more by superficial cues available in the persuasive context (petty & cacioppo, 1986). therefore ffc was designed to be measured by residents’ attitude toward different community activities. third, friend and kinship network (fkn). according to wirth (1938)’s theory, complex social function creates unique personal interaction in mega city, which depending on specific identity status (occupation or role) instead of personality (emotion).previous studies also confirmed this phenomenon in beijing (zhang, 2006). therefore, fkn was designed to be measured with formal circles, such as community circles, working circles, and political circles. 2.1.2 dependent variables predecisional processes variables were taken as dependent variables. predecisional processes comprise three factors, which are reception of, attention to, and comprehension of socially transmitted information (fiske & taylor, 1991). then following variables were designed. first, reception (rp).the reception means people’s ability to receive information from external environment. it was measured with the individual access to the disaster information (such as rainfall, earthquake, and fire disaster)because information reception would depend on the invasion of normal activities (lindell & perry 2004, p.76). second, attention (at).the attention means people’s ability to accept the cues, which depend on the expectations of threat, competing demands, and intrusiveness of hazard information (lindell, 2004, p.49). it was measured with individual interest in different disaster information. published by atlantis press copyright: the authors 22 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities third, comprehension (cp).generally, the comprehension refers to one’s ability to understand the meaning, nature, or importance of something correctly. it was measured with a series of true or false questions related to protective knowledge of specific disaster types. 2.1.3 moderator variables it is well-documented that demographic characteristics can influence protective action, such as enarson, fothergill and peek (2007)’s discussion about gender and disaster, gaillard (2012)’s argument about caste, ethnicity, religious affiliation and disaster. in view of the characteristics of residents in beijing, six demographic variables were selected as moderator variables, including marital status, gender, age, ethnicity, education attainment, and income level. in sum, we acquired twelve variables, including three independent variables (social context), three dependent variables (predecisional processes), and six moderator variables (demography). their definition and measuring items are represented in table 1. table 1. definition of twelve variables type variables name symbol measuring items demographic variables (moderator) marital status mr/um married or unmarried gender me/fm male or female age md/yu middle-aged (45-59) or youth (18-44) ethnicity mj/mn majority group (han) or minority group education attainment he/le higher level (with bachelor degree) or lower level (without bachelor degree) income level hi/li higher level ($5,000.00-$13,000.00) or lower level (0-$5,000.00) social context variables (independent) media communication mc attitude toward media activities face-to-face communication ffc attitude toward training activities friend and kinship network fkn personal social circles predecisional processes variables (dependent) reception rp concern of the disaster information attention at interest of the disaster information comprehension cp true or false questions of disaster knowledge 2.2 hypothesis and flow chart 2.2.1 hypothesis based on the padm and previous analyses, the relationship between the social context and predecisional processes was focused in this paper. then the hypothesis was stated as “if the social context was taken as information source, it would influence predecisional processes in different ways”. it can be divided into three parts: first, suppose mc would influence rp (h1a), at(h1b), and cp(h1c); second, suppose ffc would influence rp (h2a), at (h2b), and cp (h2c); third, suppose fkn would influence rp (h3a), at (h3b), and cp (h3c). 2.2.2 flow chart in order to describe the relationships between different variables, a flow chart was designed with respect to above hypotheses and previous theories, as is shown in figure 1. first part presents the interactions between social context variables (fkn, fft, mc). the padm can’t include these links, so they should be redesigned. because friend and kinship network is relatively stable, we let fkn influence both ffc and mc; then we let ffc influence mcor let mc influence fft. after a series of combination of “fkn, fft, and mc”, sixteen competing models can be obtained, which are show in table 2. in which, the best fit model should represent the best combination of social context variables(byrne, 2001). published by atlantis press copyright: the authors 23 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities second part presents the relationships between three predecisional processes (rp, at, cp). in learning theories, constructivism emphasizes the importance of the background knowledge, which believes that when new information is approached, the learner faces a loss of equilibrium with their previous understanding, and then an improved cognitive schema can be formed (bodner, klobuchar, &geelan, 2001). then it can be seen that the order of three predecisional processes should be “from reception (rp) to attention (at)”, and “from attention (at) to comprehension (cp)” in the disaster schema. third part presents the specific effects between two types of variables. according to above hypothesis (section 2.2.1), we supposed every social context variables (fkn, fft,mc) can influence every predecisional processes variables (rp, at, cp). then we acquired nine hypothesis paths, which should be tested later, as is shown in figure 1. fourth part presents the role of specific demographic variable. for example, aimed at gender variable, we separated the sample into male group and female group, and then check the variance of particular relationship across groups. 2.3 data collection 2.3.1 sampling the subjects are designed as residents in the haidian district in beijing. the population in beijing is representative because 62% of them are registered residents, and 38% of them are non-registered residents, which come from all over the china. furthermore, haidian district is a typical district because it locates at the edge of beijing, and its territory includes both urban communities and rural communities(bmbs, 2014). then sampling was implemented with stratified sampling technique (cochran, 1977). first, community stratum: concerning the community category (cssb, 2015), four types of communities were selected: new community (building before 1999), old community (building after 2000), rural community, and employer-governance community. second, resident stratum: minimum sample size was calculated by the statistics equation of “(z-score)² * stddev*(1-stddev) / (margin of error)²” (smith, 2013), and then a threshold of 384 was found. meanwhile, taking into account the problem of method bias, a two-stage survey was carried out with half a year’s interval: the pretest survey was implemented in september 2013, and the posttest survey was implemented in january 2014. finally, each respondent was given $5 gift as an incentive of completing the questionnaire. 2.3.2 questionnaire data were collected with method of the questionnaire survey. the questionnaire was designed concerning the definition of twelve variables. the draft was modified through the primary test and with the method of brainstorming. then a questionnaire with five parts and 63 items was obtained: part i is basic information of residents, which include personal information and social circle information, therefore it was used to measure demographic variables and variables of “fkn”; part ii was designed to measure variables of “rp” with a series of questions of “did you notice the following disaster information”; part iii was designed to measure variables of “at” with a series of questions of “did you interest in the following disaster information”; part iv was designed as to measure variables of “cp” with a series of true or false questions about disaster preparedness knowledge; part v was designed to measure variables of “ffc” and “mc” with questions of “how about your attitude of the following risk communication activities”. 2.4 data analysis 2.4.1 primary analysis and missing data data were analyzed to check the reliability and validity with spss 20.0, and then following values were found: cronbach’s α is 0.867 for the pretest, 0.886 for the posttest; the parameter of kmo is 0.861 (sig.=0.000) for the pretest, 0.894 (sig.=0.000) for the posttest, which show that the data are both reliable and valid. the missing data were approached with two methods (bouza-herrera, 2010): to begin with, the samples with rp cp at fkn mc ffc demographic variables figure 1. flow chart about relationships between different variables published by atlantis press copyright: the authors 24 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities overmuch missing data (exceeding 30% information) were deleted directly; furthermore, the method of hot-deck imputation was used to fill up the blank cells with the median value considering that those in the risk area are likely to engender protective actions by observing the behavior of others (lindell& perry, p.55). then the final data were composed of 344 acceptable pretest samples, and 351 acceptable posttest samples. demographic distribution are as follows: (a) marital status, mr/um are 282/62 (pretest) and 278/73 (posttest); (b) gender, “me/fm are 142/199 (pretest) and 229/116 (posttest); (c) age, md/yu are 100/81 (pretest) and 98/110 (posttest); (d) ethnicity, mj/mn are 316/28(pretest) and 325/26 (posttest); (e) educational attainment, he/le are 113/153 (pretest) and 151/113 (posttest); (f) income level, hi/liare 175/109 (pretest) and 165/121 (posttest). 2.4.2 common method variance in order to overcome the problem of method bias, common method variance (cmv) was controlled with procedural remedy method and statistical method(podsakoff, mackenzie, lee, & podsakoff, 2003).first, the procedural remedy was implemented through three procedures: (a) dividing the questionnaire survey into two stages; (b) collecting data from four types of communities; (c) using anonymous questionnaire. second, the statistical control was implemented with harman’s single-factor test. the basic assumption of this method is that “if cmv was substantial, one general factor would account for the majority of the whole information (>50%)” (podsakoff, mackenzie, & podsakoff, 2012). factor analysis was implemented with spss 20.0 software to check the parameter of “% of variance”, and the results show that: its value is 22.226 for the pretest data, 25.405 for the posttest data, which means that the cmv of survey was acceptable. 2.4.3 non-parametric test in view of the problem of sampling consistency, non-parametric test was implemented(hollander, wolfe, & chicken, 2013). the parameter of kolmogorov-smirnov was used to compare the characteristics of demographic groups regarding the two-stage survey. the results are as follows: gender (sig.=0.220, r), age (sig.=0.439, r), ethnicity (sig.=1.000, r), education (sig.=0.131, r), income (sig.=0.997, r). symbol “r” represents “retains”, which means there is no significant variance between two-stage surveys. 2.4.4 exploratory factor analysis creating good measures involves both writing good questions and organizing them to form the questionnaire (ussr,n.d). because of this, more items were designed to measure the variables, which should be refined when the data have been collected. explore factor analysis (efa) was implemented based on spss 20.0, and then the items in the questionnaire were processed according to two rules (fabrigar, 1999): first, item loading should approach 0.5 (hair, black, babin, anderson, 2009) unless there are not enough items; second, no more than four items should be selected for each variable taking into account the balance between variables. at last, a total of 29 measuring items were left, in which 11 items was used to measure three social context variables, 12 items was used to measure three predecisional variables, and 6 items was used to measure six demographic variables. 3. results 3.1 interactions between social context variables in order to discuss the interactions between social context variables, sixteen competing models were designed with the combinations of “fkn, ffc, and mc”. structural equation modeling (sem) was used to rank the competing models with four criteria: first, the model should be convergent; second, lower ecvi is better fit (cudeck, 1989); third, one with the smallest aic would be chosen (kieseppä, 2003); fourth, the model is better if it can comprise the characteristics of rival models (mizon,1984). sem model was built and run based on amos 20.0 software, and then the results were obtained. table 2 represents the combinations and corresponding parameters. published by atlantis press copyright: the authors 25 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities table 2.combinations of social context variables and corresponding parameters model combinations cvg. evci aic pre post pre post pre post model 1 mcfkn, ffcfkn n n 1.96 2.01 673.77 703.51 model 2 mcfkn, ffcfk, ffcmc y n 1.93 2.03 660.66 710.08 model 3 mcfkn, ffcfkn, mcffc y n 1.93 - 660.66 - model 4 mcfkn, ffcfk, ffcmc, mcffc n n -- -- model 5 mcfkn n n 1.96 2.01 671.86 712.97 model 6 mcfkn, ffcmc y n 1.93 2.01 661.07 711.91 model 7 mcfkn, mcffc y n 1.93 - 660.66 - model 8 mcfkn, ffcmc, mcffc y n 1.93 2.03 660.66 710.08 model 9 ffcfkn n y 1.97 2.00 675.97 701.22 model 10 ffcfkn, ffcmc y n 1.93 - 660.66 - model 11 ffcfkn, mcffc y y 1.93 2.01 660.91 704.16 model 12 ffcfkn, ffcmc, mcffc y n 1.93 2.00 660.66 700.82 model 13 null y n 1.93 - 660.66 - model 14 ffcmc n n 1.93 2.01 661.07 702.79 model 15 mcffc y y 1.93 2.01 660.91 704.16 model 16 ffcmc, mcffc n n -- -- note: “” means the influence from the latter variable to the former variable; “--” represents “no influence”; y represents “convergent”; n represents “no convergent”; cvg. represents “convergence status”. it can be seen from the table 2 that: first, only model 11 and model 15 were valid regarding the convergence rule; second, the values of evci, aic, and bic are the same between model 11 and model 15; third, model 11 is better than model 15 taking into account the encompass principle. depending on the structure of model 11, the interactions between social context variables can be found, which was presented as “mcffcfkn”. it can be seen that immersing in friend and kinship network is useful to face-to-face communication; participating in face-to-face communication is beneficial to media communication. 3.2 effects of social context on predecisional processes in order to analyze hypothesis paths, sem was built according to model 11 and above hypothesis paths (section 2.2.1) with amos 20.0 software. the results were checked with five criteria: c.r. should be greater than 1.96 (hanneman, kposowa, & riddle, 2013); p-value should be less than 0.05 (nuzzo, 2014); s.e. should be positive; the bigger the value of unstandardized and standardized regression coefficients, the stronger the hypothesis (hargens, 1976). the results of nine hypothesis paths are represented in table 3, in which the result cannot be accepted unless the pretest result and posttest result are significant simultaneously. published by atlantis press copyright: the authors 26 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities table 3. the testing parameters of hypothesis paths hypothesis unstd. s.e. c.r. p std. prepostprepostprepostprepostprepost h1a: rpmc 0.59 0.83 0.12 0.15 5.07 5.62 *** *** 0.57 0.51 h1b: atmc -0.03 0.15 0.08 0.08 -0.46 1.87 0.65 0.06 -0.05 0.14 h1c: cpmc -0.01 -0.14 0.11 0.09 -0.10 -1.62 0.92 0.11 -0.01 -0.36 h2a: rpffc -0.14 -0.02 0.10 0.11 -1.48 -0.15 0.14 0.89 -0.16 -0.01 h2b: atffc 0.24 0.38 0.07 0.07 3.58 5.64 *** *** 0.40 0.42 h2c: cpffc -0.05 0.09 0.10 0.07 -0.49 1.40 0.63 0.16 -0.07 0.28 h3a: rpfkn 0.05 -0.65 0.08 0.38 0.56 -1.70 0.58 0.09 0.04 -0.17 h3b: atfkn 0.00 0.06 0.05 0.10 0.09 0.57 0.93 0.57 0.01 0.02 h3c: cpfkn 0.37 -0.07 0.14 0.09 2.69 -0.84 0.01 0.40 0.42 -0.08 note: “***” means the value which is less than 0.001; “” have the same meaning with table 2.;unstd.represents“unstandardized regression weights”; std. represents “standardized regression weights”. the results show that: first, in light of consistency of two-stage data, two hypothesis paths of “rpmc (c.r.: 5.07/5.62) and atffc” (c.r.: 3.58/5.64) are significant. second, in light of the value of unstandardized regression weights and standardized regression weights, the influence of mc (unsd.: 0.59/0.83; sd.: 0.57/0.51) is stronger than ffc (unsd.: 0.24/0.38; sd.: 0.40/0.42). third, fkn is negative because there is no significant hypothesis path related to it. 3.3 variance across demographic groups in order to discuss the variance of significant paths across demographic groups, multiple group analysis(mga) was implemented. sem models of path “rpmc” and path “atffc” were built separately. we run sem model based on amos 20.0 software, and then check the following parameters: first, model fit parameter was checked to make sure that the model selected achieves the minimum value. second, “ecvi” and “aic was checked to rank the model, lower ecvior aicis better fit (cudeck, 1989). third, significant level was checked with pairwise parameter of specific path (>1.64). if pairwise parameter was not reported, unstandardized and standardized regression coefficients (>0.4) would be checked as an alternative (richards, 1982). the final results were represented in table 4, which represent the variance of two paths across demographic groups. the results are summarized from two aspects: first aspect is about the path of “rpmc”. to begin with, we checked the pretest data with the parameter of p.p (>2.20), and the posttest data with parameters of unstd. and std. (>0.4), and then found that there are variances across education attainment groups. furthermore, we checked the pretest data and the posttest data with parameters of unstd. and std. (>0.4), and found that there are variances across gender groups. second aspect is about the path of “atffc”. we checked the pretest data and the posttest data with parameters of unstd. and std. (>0.4), and found that there are variances across education attainment groups, as well as variances across income level groups. published by atlantis press copyright: the authors 27 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities table 4.mga parameters of two paths across demographic groups demographic variables rpmc atffc p.p. unstd. std. p.p. unstd. std. age preed -0.07 0.38 0.51 0.96 0.78 0.51 yu -0.07 0.38 0.49 0.96 1.02 0.58 posted -0.85 0.69 0.40 3.53 0.36 0.67 yu -0.85 0.49 0.46 3.53 0.83 0.76 education prehe 2.20 0.29 0.42 -0.83 0.53 le 2.20 0.62 0.44 -0.83 0.53 posthe -0.55 0.49 -0.55 0.71 le -0.55 0.52 -0.55 0.69 gender preme -0.49 0.37 -0.77 0.41 fm -0.49 0.49 -0.77 0.51 postme -0.76 0.51 0.27 0.58 0.63 fm -0.76 0.51 0.27 0.61 0.70 income prehi -2.83 0.72 0.50 -0.69 0.37 li -2.83 0.31 0.30 -0.69 0.48 posthi -0.39 0.76 0.46 -0.55 0.71 li -0.39 0.68 0.43 -0.55 0.69 note: p.p.represents “pairwise parameter”; unstd. and std. have the same meaning with table 3.; “--” represents “no report”. 4. discussion the findings from this research suggest that: the social context can influence predecisional processes in two ways, which are “rpmc” and “atffc”; furthermore, the order of influence strength is “mc>ffc>fkn”; at last, there are interactions between social context variables, which is “fknffcmc”. the detailed explanations are as follows. on the one hand, the media communication does influence individual reception of risk information. each medium has its own technological capabilities that influence the delivery of text, sound, and visual images (croteau, hoynes,& milan, 2011, pp.285-286), so it is natural that the access to different media depends on individual habits and lifestyle. furthermore, the findings show education attainment and age can influence media communication significantly, which might be explained by the theory that education can develop habits, skills, and abilities (mirowsky, 1998), as well as the theory that age can engender different circadian rhythmicity (minors, atkinson, bent, rabbitt, & waterhouse, 1998).in addition, the impact of media communication is the strongest among three social context variables, which shows that media channel should be emphasized because of its capacity to cover extensive population. thus, it can be deduced that media channel should be diverse with respect to the different groups, such as elder/youth groups and blue-collar/white-collar groups, in order to increase individual reception of risk information, which should be helpful for expanding the coverage of risk communication. on the other hand, the face-to-face communication does influence individual attention to risk information. humans are driven to protect against external threats (lawrence, nohria, 2002), so they are easy to be impressed by the vivid on-site activities. furthermore, the findings show that education attainment and income level can influence face-to-face communication significantly. it is well-known that socioeconomic status is related to participation in physical activity. therefore, the social status might become “perceived barriers” (cerin& leslie, 2007) to impede face-to-face activity. in addition, the findings also show that face-to-face communication can influence media communication. thus, it can be deduced that physical activities should be specific with respect to the different socioeconomic published by atlantis press copyright: the authors 28 tiezhong liu et al. / pre-decisional processes of protective action in beijing communities groups, such as the middle class and the lower class, in order to enhance individual attention to risk information, which should be helpful for improving the effectiveness of risk communication. surprisingly, friend and kinship network does not influence any predecisional processes variables, but it influences face-to-face communication activity. the former finding might be explained by lindell and perry (2004, p.139)’s theory that available response time impede individual decision behavior in the long-term hazard adjustment. the latter finding has ever been interpreted by many researchers (such as voorhees, et.al., 2005; nauert, 2012), whose theories showed that social networks can influence physical activity in a positive or negative way. thus, it can be seen that social influence should be emphasized, such as the power of kids and elder, in order to motivate public participation towards communication activities. quite surprisingly, no social context variable does exert significant influence on individual comprehension of risk information. the terminology of risk information is usually hard to be understood by the local communities (pearson, 2012). therefore, there might be some mismatch between the information people receive and what they understand(hhs, 2010). the findings demonstrate that there are big challenges in the comprehension of risk information concerning the residents in beijing. thus, it can be seen that the problem of translating scientific jargon into plain language should be emphasized in order to improve public comprehension of risk information. overall, this research suggests that the social context can influence predecisional processes through media communication and face-to-face communication, but the methods should be diverse and specific. in addition, the topic of social conformity, as well as the problem of translating terminology into plain language should be emphasized. 5. conclusions it can be found that media communication, face-to-face communication and friend and kinship network play different role in the link between social context and predecisional processes. they should be designed diversely and specifically concerning the characteristics of different demographic groups. it should be emphasized that there are big problem related to public comprehension of social transmitted information. in the future, the research should cover the particular social context regarding demographic characteristics, the facilitating and barriers in the social context, and the role of social conformity. acknowledgements with the sponsor of china scholarship council (201306035027) this paper was completed at the disaster research center of university of delaware. i would like to thank dr. james kendra for his recommendation of corresponding literature, as well as dr. sarah deyoung for her help in the field of social psychology. meanwhile, i would like to express my gratitude to mrs. xao han of beijing municipal office of civil air defense and mr. zhen xu of the haidian district government for their help in the survey. reference bmbs, beijing municipal bureau of statistics. (2014, june 18). data interpretation of permanent residents of beijing in 2013. retrieved from http://www.bjstats.gov.cn/sjfb/bssj/ndsj/ndsjfpfb/2013n/20 1406/t20140618_274951.htm (in chinese) bodner,g., klobuchar,m., &geelan,d. (2001). the many forms of constructivism. journal of chemical education, 78, 1107-1134. bouza-herrera,c.n. (2010). handling missing data in ranked set sampling. new york, ny: springer, 2010. doi: 10.1007/978-3-642-39899-5 byrne,b.m. (2001). structural equation modeling with amos: basic concepts, applications, and programming. mahwah, nj: lawrence erlbaum associates. cerin,e., & leslie, e. (2007, february). socioeconomic status and perceived barriers to physical activity.presentation at the 2007 active living research annual conference. retrieved from http://activelivingresearch.org/socio-economic-status-and-p erceived-barriers-physical-activity cochran,w.g.(1977). sampling techniques (3rd edition). wilmington, wiley. croteau,d., hoynes,w., milan,s. (2011). media society: industries, images, and audiences. los angels, sage. cscb, community service center of beijing. (2015, may 16). community construction. retrieved from http://hd.bjcs.gov.cn/wps/portal/template1/bjssq cudeck,m.r. (1989). single sample cross-validation indices for covariance structures. multivariate behavioral research, 24(4): 445-455. enarson,e., fothergill,a., &peek,l. (2007). gender and disaster: foundations and directions. (pp. 130-146). in rodríguez,h., quarantelli,e.l., dynes,r.r. (eds.). handbook of disaster research, springer. fabrigar,l.r., wegener,d.t., maccallum,r.c., &strahan,e.j. (1999). evaluating the use of exploratory factor analysis in psychological research. psychological methods, 4(3), 272-299. fiske,s.t., &taylor,s.e.(1991). social cognition (2nd edition). new york: mcgraw-hill. published by atlantis press copyright: the authors 29 http://www.bjstats.gov.cn/sjfb/bssj/ndsj/ndsjfpfb/2013n/201406/t20140618_274951.htm http://www.bjstats.gov.cn/sjfb/bssj/ndsj/ndsjfpfb/2013n/201406/t20140618_274951.htm tiezhong liu et al. / pre-decisional processes of protective action in beijing communities gaillard, jc.(2012). case, ethnicity, religious affiliation and disaster. (pp. 459-469). in b.wisner, j.c gaillard, &i.kelman.the routledge handbook of harzards and disaster risk reduction. london, routledge. hair,j.f., black,w.c., babin,b.j., &anderson,r.e. (2009). multivariate data analysis (7th edition). upper saddle river, nj: prentice hall, 2009. hanneman,r.a., kposowa,a.j., &riddle,m.d. (2013). basic statistics for social research. wilmington, de: jossey-bass. hargens,l.l. (1976). a note on standardized coefficients as structural parameters. sociological methods and research, 5, 247-256. hayden,j (ed.). (2009). introduction to health behavior theory.boston, jones and bartlet publishers. health,.l., &o’hair,h.d.(eds.).(2009). handbook of risk and crisis communication.new york, ny, routledge. hhs, u.s. department of health and human services, office of disease prevention and health promotion. (2010). plain language: a promising strategy for clearly communicating health information and improving health literacy. healthy people 2010 health communication focus area. retrieved from http://health.gov/communication/literacy/plainlanguage/pla inlanguage.htm hollander,m., wolfe,d.a., &chicken,e. (2013). nonparametric statistical methods. hoboken, nj: wiley. kieseppä,i.a. (2003). aic and large samples. philosophy of science, 70(5): 1265-1276. lasswell,h.d. (1948). the structure and function of communication in society.in l. bryson (ed.). the communication of ideas: a series of addresses (p. 117). new york, ny: institute for religious and social studies. lawrence,p.r., nohria,n. (2002). driven: how human nature shapes our choices.san francisco, ca, jossey-bass. lazarus,r.s., &folkman,s. (1984). stress, appraisal, and coping. new york: springer. lindell,m.k., perry, r.w. (2012). the protective action decision model: theoretical modifications and additional evidence. risk analysis, 32(4), 616-632. lindell,m.k., &perry,r.w. (2004). communicating environmental risk in multiethnic communities. los angeles: sage publications. maskrey a. (1989). disaster mitigation: a community-based approach. oxford, oxfam. minors d., atkinson g., bent n., rabbitt p., & waterhouse j., (1998). the effects of age upon some aspects of lifestyle and implications for studies on circadian rhythmicity. age ageing, 27(1):67-72. mirowsky,j. (1998). education, personal control, lifestyle and health: a human capital hypothesis. research on aging, 20(4), 415-449. doi: 10.1177/0164027598204003 mizon,g.e. (1984). the encompassing approach in econometrics. canberra: australian national university, faculty of economics and research school of social sciences. nauert,r. (2012, september 11). does social networking limit physical activity? retrieved from http://psychcentral.com/news/2012/09/11/does-social-netw orking-limit-physical-activity/44412.html nuzzo,r. (2014). scientific method: statistical errors. nature, 506: 150-152. pearson,l. (2012, november, 11). early warning of disasters: facts and figures. scidev net. retrieved from http://www.scidev.net/global/communication/feature/earlywarning-of-disasters-facts-and-figures-1.html petty,r., &cacioppo,j. (1986). the elaboration likelihood model of persuasion. in l. berkowitz (ed.), advances in experimental social psychology (pp.123-205).new york: academic press. podsakoff,p.m., mackenzie,s.b., &podsakoff,n.p. (2012). sources of method bias in social science research and recommendations on how to control it. annual review of psychology, 63, 539-569. doi: 10.1146/annurev-psych-120710-100452 podsakoff,p.m., mackenzie,s.b., lee,j., & podsakoff, n. p. (2003). common method biases in behavioral research: a critical review of the literature and recommended remedies. journal of applied psychology, 88(5), 879-903. richards,j.m. (1982). standardized versus unstandardized regression weights.applied psychological measurement, 6, 201-212. shannon,c.e. (1948). a mathematical theory of communication. the bell system technical journal, 27, 379-423. smith,s.m. (2013, april 8). determining sample size: how to ensure you get the correct sample size. retrieved from https://www.qualtrics.com/blog/determining-sample-size/ terpstra,t., lindell,m.k. (2012). citizens’ perceptions of flood hazard adjustments: an application of the protective action decision model. environment and behavior, 45(8), 993-1018. ussr, u.s. survey research. (n.d.).questionnaire design. retrieved from http://www.pewresearch.org/methodology/u-s-survey-resea rch/questionnaire-design/ voorhees.c.c., murray,d., welk,g., birnbaum,a., ribisl,k.m., johnson,c.c., pfeiffer,k.a., saksvig,b., &jobe,j.b. (2005). the role of peer social network factors and physical activity in adolescent girls. american journal of health behavior, 29(2), 183-190. wang,f. (2014). modeling public s protective behavior response from the perspective of risk perception .dissertation, university of science and technology of china. (inchinese) wirth l.(1938). urbanism as a way of life. american journal of sociology, 189-197. zhang,y.x. (2006). community, urbansim and social networks: research about social ties in urban. social sciences in guangdong, 5, 183-188. doi: 10.3969/j.issn.1000-114x.2006.05.029 (in chinese) zhu y. (2014, september 23). china's nuclear expansion threatened by public unease. china dialogue. retrieved from https://www.chinadialogue.net/article/show/single/en/7336 -chinese-protesters-threaten-nuclear-expansion (in chinese) published by atlantis press copyright: the authors 30 https://www.chinadialogue.net/article/show/single/en/7336-chinese-protesters-threaten-nuclear-expansion https://www.chinadialogue.net/article/show/single/en/7336-chinese-protesters-threaten-nuclear-expansion 1. introduction 2. methodology 3. results 4. discussion 5. conclusions acknowledgements reference << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /usedeviceindependentcolorforimages /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <feff06270633062a062e062f0645002006470630064700200627064406250639062f0627062f0627062a002006440625064606340627062100200648062b062706260642002000410064006f00620065002000500044004600200645062a064806270641064206290020064406440637062806270639062900200641064a00200627064406450637062706280639002006300627062a0020062f0631062c0627062a002006270644062c0648062f0629002006270644063906270644064a0629061b0020064a06450643064600200641062a062d00200648062b0627062606420020005000440046002006270644064506460634062306290020062806270633062a062e062f062706450020004100630072006f0062006100740020064800410064006f006200650020005200650061006400650072002006250635062f0627063100200035002e0030002006480627064406250635062f062706310627062a0020062706440623062d062f062b002e0635062f0627063100200035002e0030002006480627064406250635062f062706310627062a0020062706440623062d062f062b002e> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <feff004b0069007600e1006c00f30020006d0069006e0151007300e9006701710020006e0079006f006d00640061006900200065006c0151006b00e90073007a00ed007401510020006e0079006f006d00740061007400e100730068006f007a0020006c006500670069006e006b00e1006200620020006d0065006700660065006c0065006c0151002000410064006f00620065002000500044004600200064006f006b0075006d0065006e00740075006d006f006b0061007400200065007a0065006b006b0065006c0020006100200062006500e1006c006c00ed007400e10073006f006b006b0061006c0020006b00e90073007a00ed0074006800650074002e0020002000410020006c00e90074007200650068006f007a006f00740074002000500044004600200064006f006b0075006d0065006e00740075006d006f006b00200061007a0020004100630072006f006200610074002000e9007300200061007a002000410064006f00620065002000520065006100640065007200200035002e0030002c0020007600610067007900200061007a002000610074007400f3006c0020006b00e9007301510062006200690020007600650072007a006900f3006b006b0061006c0020006e00790069007400680061007400f3006b0020006d00650067002e> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <feff004b00e40079007400e40020006e00e40069007400e4002000610073006500740075006b007300690061002c0020006b0075006e0020006c0075006f00740020006c00e400680069006e006e00e4002000760061006100740069007600610061006e0020007000610069006e006100740075006b00730065006e002000760061006c006d0069007300740065006c00750074007900f6006800f6006e00200073006f00700069007600690061002000410064006f0062006500200050004400460020002d0064006f006b0075006d0065006e007400740065006a0061002e0020004c0075006f0064007500740020005000440046002d0064006f006b0075006d0065006e00740069007400200076006f0069006400610061006e0020006100760061007400610020004100630072006f0062006100740069006c006c00610020006a0061002000410064006f00620065002000520065006100640065007200200035002e0030003a006c006c00610020006a006100200075007500640065006d006d0069006c006c0061002e> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 85-92 doi: https://doi.org/10.2991/jracr.k.190703.004; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). applicability of regional evaluation for rapid assessment models of earthquake disaster life loss – a case study of gansu province wen li, wenkai chen, hanran zhang, zijing su institute of lanzhou earthquake research, china earthquake administration, gansu lanzhou 730000, china received december 30, 2018 accepted june 24, 2019 abstract in this paper, typical earthquake disaster life loss assessment models are applied to verify and calculate the historical earthquake cases in the gansu province since 1966. the assessment accuracy and applicability of various models in the gansu province are studied by means of actual earthquake cases. results show that: (1) for ms<5.5 and 5.5≤ms≤6.0, a life loss assessment model is based on population density, another model based on epicentral intensity proposed, and the national earthquake disaster emergency assessment model are relatively accurate and basically included in the reasonable range; for ms>6.0, characteristics of the assessed regions, such as physical geography and social economy, should be analyzed to correct the model calculation results and to receive comprehensive assessment results, so as to support emergency decisions. (2) earthquake disaster life loss assessment models are regionalized and the assessment results are obviously regional. assessment results of the earthquakes occurred in hedong region are not accurate for hexi region of gansu province. keywords: gansu province; earthquake; life loss model; rapid assessment; applicability evaluation 1. introduction china is one of the countries that suffer the severest earthquake disasters all over the world (fu, 1994), and as gansu province is located on the north segment of the south-north seismic belt, it is one of the multi-earthquake provinces in china. it is recorded that gansu province has ever had the earthquakes of ms≥8.0 for four times, resulting in mass casualties, wherein haiyuan had an earthquake of ms=8.5 in 1920, resulting in death toll of 273,465±9,700 (liu, 2003); gulang had an earthquake of ms=8.0 in 1927, resulting in death toll of 41,471. in recent years, gansu province and its surrounding areas have increased seismic activities. for example,  *corresponding author. tel.: +86 13993146183; fax: +86-0931-8268536, e-mail: cwk2000@yeah.net post address: no.450, donggang west road, lanzhou city, gansu province, china. wenchuan of sichuan province had an earthquake of ms=8.0, jiuzhaigou of sichuan province had an earthquake of ms=7.0, minxian county and zhangxian county of gansu province had an earthquake of ms=6.6, and menyuan of qinghai had an earthquake of ms=6.4, etc. since earthquakes are characterized in instantaneity and suddenness, violent earthquakes often lead to serious loss of life. according to the chinese rules for natural disaster relief, “people first” is the primary principle (the state council of the p.r.c., 2010). besides, in actual earthquake emergency response, response level is in direct proportion to relief materials. if emergency response level is too high, it will lead to mass waste of relief materials; if emergency response level is too low, it goes against disaster relief. as a key factor, “casualties” directly determines the earthquake emergency response start level of the government (national earthquake emergency plan, 2012). rapid assessment of earthquake journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 86 disaster life loss is a scientific problem to be solved in actual earthquakes. studies on rapid assessment models of life loss can provide technical support for instant emergency command and decision-making of the governments, thus have very important scientific and practical significance. in this paper, typical earthquake disaster life loss assessment models are applied to verify and calculate the historical earthquake cases of gansu province since 1966. assessment accuracy, applicability and existing problems of different models by means of actual earthquake cases are discussed. 2. research progress at present, the life loss risk assessment oriented to earthquake emergency is mainly focused on probabilistic risk assessment and deterministic risk assessment. there are mainly two assessment directions: one is the earthquake-caused human mortality rate model obtained through building vulnerability analysis. this generally contains seismic damage matrix method based on historical data and experts’ experience and vulnerability analysis method based on performance. for example, yin (1991) proposed seismic vulnerability classification method for buildings and facilities, vulnerability analysis method for various building structures, and earthquake-caused loss prediction method; he (2005) obtained the house building vulnerability matrix in different parts of sichuan province through the empirical statistics of ten earthquake cases in sichuan province; fema commissioned the national institute of building sciences (nibs) for theoretical research and demonstration and launched a set of standardized methods hazus99 for earthquake-caused loss prediction in 1999 (fema, 1999). the former does not require much on the detail level of the data about the disaster-affected carriers, but gives a relative rough classification for building types, therefor the accuracy and reliability of the seismic damage matrix may be affected by the limitation of the local seismic history conditions; the latter fully considers the influence of structure types and post-earthquake damage state on mortality and systematically conducts calculations for earthquake-caused mortality. however, it requires detailed classification and careful investigation of the buildings in the area, establishment of a detailed database about buildings, and a large number of numerical calculations and statistical analyses (li, 2014). the other is the empirical formulae of human mortality rate and number based on seismic parameters (i.e., magnitude, intensity, etc.) obtained by regression analysis of historical seismic damage data without consideration of the damage to buildings. the first category is based on population density. for example, samardjieva et al. (2002) conducted research on earthquake cases which occurred since 1990. they, based on population density (respectively below 25 people/km2, 25 people/km2 to 50 people/km2, 50 people/km2 to 100 people/km2, 100 people/km2 to 200 people/km2，and above 200 people / km2), obtained the empirical formulae about earthquake-caused mortality number and earthquake magnitude for the earthquakes occurring in 1900 to 1950 and 1950 to 1999 on a global scale; on this basis, badal et al. (2005), by means of magnitude, people density and other related indicators, constructed a human mortality assessment model and used it to assess the human mortality number in eight spanish cities hypothetically hit by 6.0-magnitude and 6.5-magnitude earthquakes. the second category is based on time. for example, chen (2005) conducted a research on the earthquake cases which occurred in china from 1980 to 2000. he classified the earthquake occurrence time by day and night, and obtained the empirical formulae related to human mortality number and earthquake intensity based on time. the third category is based on per capita gdp. for example, chen et al. (1999) researched 207 earthquake cases occurred in china from 1989 to 2004. he set cn ¥2,700 yuan – the per capita gdp represented by constant price in 2000 – as the classification threshold, and obtained the relationship between life loss rate and earthquake intensity; liu jifu et al (liu, 2009), based on the previous research on macroscopic vulnerability, allotted population and gdp according to actual earthquake intensity, and established an earthquake-related life vulnerability model that is more similar to the real earthquake situations. the empirical formulae based on seismic parameters under statistical laws are dependent highly on historical data, so the selection of different historical data sets imposes great influence on the establishment of calculation methods. the selection of major influencing factors is the main journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 86 87 cause of the difference in the assessment results from various models. simple parameter obtaining is more applicable for rapid post-earthquake human mortality assessment. 3. data source considering the reliability and integrity of earthquake disaster data, the earthquake disaster loss assessment reports of gansu province having officially have field investigation of earthquake damage and loss assessment since 1966. they include earthquake-caused loss data in gansu and were collected from the compilation of earthquake disaster loss assessment in mainland china for 1966-1989, 1990–1995, 1996–2000, 2001–2005, 2006–2010, and 2011–2015 (china earthquake administration, 1996; 2000; 2010; 2015); besides, the data of towns, townships and sub-districts according to the population census of gansu province in 2010 as well as statistical yearbooks of gansu province over the years are also collected. upon data analysis, 16 earthquake cases of ms≥6.0 are selected to verify the earthquake disaster life loss assessment models and take corresponding earthquake parameters, including the earthquake date, earthquake time, latitudinal and longitudinal coordinates, magnitude (because the china seismological network uses the surface wave magnitude ms, this magnitude is used in the rest of the paper), and mortality number. the intensity maps of the 16 earthquakes were scanned one by one. thereafter, the arcgis software developed by esri was used to register and digitize the intensity maps to obtain the area and distribution range of each intensity zone of an earthquake. in some previous studies, population data are distributed in macroscopic epicentral intensity (the highest intensity), which is quite different from the actual condition of earthquake; besides, actual life loss distribution is closely related to the distribution of population in intensity map. in this paper, the actual intensity area is used as the allocation unit. according to the township population involved in each intensity area during the earthquake, the number of people in the area of each intensity area is converted into the population of the intensity zones. 4. method when collecting and collating the earthquake-caused life loss assessment models published since 2000, the author finds that different models stem from different sources, employ different expressions, and have different applicable spatial ranges and time periods. in the case of earthquake emergency response, the human mortality assessment method based on building vulnerability requires more accurate classification of buildings and collapse rate of earthquake-affected buildings. in addition, the construction of earthquake-caused life loss risk assessment model requires the following principles: the model is simple and easy to popularize, the parameters are easy to obtain, the human judgment-based factors are less, and the assessment results are relatively reliable. in view of this, the author finally determines four earthquake risk models (which are named after the researchers’ names) of two types, respectively deterministic risk assessment models and probabilistic risk assessment models. 1) nonlinear regression model based on population intensity the author selects the morality assessment model based on magnitude, population density and other related indicators, which was constructed by badal et al. (2005). the formula is: log 𝑁𝑁𝑘𝑘 (𝐷𝐷) = 𝑎𝑎(𝐷𝐷) + 𝑏𝑏(𝐷𝐷)𝑀𝑀 (1) where m is the magnitude, d is the population density, and 𝑁𝑁𝑘𝑘is the predicted mortality number. the population is divided by population density: <25 people/km2, 25–50 people/km2, 50–100 people/km2, 100–200 people/km2, and >200 people/km2. the values of a and b differ for different population density levels. 2) probabilistic model based on macro economic indicator liu et al. (2009), based on the previous research on macroscopic vulnerability, set cn ¥2,700 yuan – the per capita gdp represented by constant price in 2000 – as the classification threshold, and divided all the statistical data into two groups for regression statistics: 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀 = 𝐶𝐶𝐶𝐶𝐼𝐼𝐵𝐵 (2) where 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀 represents life loss rate; i is seismic intensity; a and b are coefficients; c is correction coefficient. the model is used to disaster life loss is a scientific problem to be solved in actual earthquakes. studies on rapid assessment models of life loss can provide technical support for instant emergency command and decision-making of the governments, thus have very important scientific and practical significance. in this paper, typical earthquake disaster life loss assessment models are applied to verify and calculate the historical earthquake cases of gansu province since 1966. assessment accuracy, applicability and existing problems of different models by means of actual earthquake cases are discussed. 2. research progress at present, the life loss risk assessment oriented to earthquake emergency is mainly focused on probabilistic risk assessment and deterministic risk assessment. there are mainly two assessment directions: one is the earthquake-caused human mortality rate model obtained through building vulnerability analysis. this generally contains seismic damage matrix method based on historical data and experts’ experience and vulnerability analysis method based on performance. for example, yin (1991) proposed seismic vulnerability classification method for buildings and facilities, vulnerability analysis method for various building structures, and earthquake-caused loss prediction method; he (2005) obtained the house building vulnerability matrix in different parts of sichuan province through the empirical statistics of ten earthquake cases in sichuan province; fema commissioned the national institute of building sciences (nibs) for theoretical research and demonstration and launched a set of standardized methods hazus99 for earthquake-caused loss prediction in 1999 (fema, 1999). the former does not require much on the detail level of the data about the disaster-affected carriers, but gives a relative rough classification for building types, therefor the accuracy and reliability of the seismic damage matrix may be affected by the limitation of the local seismic history conditions; the latter fully considers the influence of structure types and post-earthquake damage state on mortality and systematically conducts calculations for earthquake-caused mortality. however, it requires detailed classification and careful investigation of the buildings in the area, establishment of a detailed database about buildings, and a large number of numerical calculations and statistical analyses (li, 2014). the other is the empirical formulae of human mortality rate and number based on seismic parameters (i.e., magnitude, intensity, etc.) obtained by regression analysis of historical seismic damage data without consideration of the damage to buildings. the first category is based on population density. for example, samardjieva et al. (2002) conducted research on earthquake cases which occurred since 1990. they, based on population density (respectively below 25 people/km2, 25 people/km2 to 50 people/km2, 50 people/km2 to 100 people/km2, 100 people/km2 to 200 people/km2，and above 200 people / km2), obtained the empirical formulae about earthquake-caused mortality number and earthquake magnitude for the earthquakes occurring in 1900 to 1950 and 1950 to 1999 on a global scale; on this basis, badal et al. (2005), by means of magnitude, people density and other related indicators, constructed a human mortality assessment model and used it to assess the human mortality number in eight spanish cities hypothetically hit by 6.0-magnitude and 6.5-magnitude earthquakes. the second category is based on time. for example, chen (2005) conducted a research on the earthquake cases which occurred in china from 1980 to 2000. he classified the earthquake occurrence time by day and night, and obtained the empirical formulae related to human mortality number and earthquake intensity based on time. the third category is based on per capita gdp. for example, chen et al. (1999) researched 207 earthquake cases occurred in china from 1989 to 2004. he set cn ¥2,700 yuan – the per capita gdp represented by constant price in 2000 – as the classification threshold, and obtained the relationship between life loss rate and earthquake intensity; liu jifu et al (liu, 2009), based on the previous research on macroscopic vulnerability, allotted population and gdp according to actual earthquake intensity, and established an earthquake-related life vulnerability model that is more similar to the real earthquake situations. the empirical formulae based on seismic parameters under statistical laws are dependent highly on historical data, so the selection of different historical data sets imposes great influence on the establishment of calculation methods. the selection of major influencing factors is the main journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 88 assess life loss rate. in practical application, the earthquake-caused mortality number can finally be obtained only after the affected population is obtained. 3) nonlinear regression model based on epicenter intensity liu et al. (2012), based on the data samples of the destructive earthquakes causing mortality in china from 1990 to 2006 along with several major earthquakes damage data in china, found out the main mortality-affecting factors. through gaussian function fitting and regression analysis; they obtained a mortality prediction model with epicenter intensity as a main parameter and magnitude and population density as auxiliary parameters for correction. the final mortality model is shown below: d = 𝑒𝑒12.2𝛼𝛼𝑑𝑑𝑑𝑑𝑑𝑑𝛼𝛼𝑚𝑚∙𝑒𝑒−(ln(ln𝑡𝑡)−2.445) 2∕0.32 (3) where d represents mortality number; ln𝑡𝑡 represents epicenter intensity; 𝛼𝛼𝑚𝑚 is magnitude correction coefficient; 𝛼𝛼𝑑𝑑𝑒𝑒𝑑𝑑 is population density correction coefficient. 4) earthquake damage assessment method stated in the national standard gb∕t 30352-2013 the author selects the mortality assessment method stated in the normative appendix of the earthquake damage assessment promulgated on december 31, 2013 and implemented in july 1, 2014 (china earthquake administration, 2014). the formula is as follows: 𝑁𝑁𝐷𝐷 = ∑ 𝐴𝐴𝑗𝑗 𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚 𝑗𝑗=6 𝜌𝜌𝑅𝑅𝑗𝑗 (4) where 𝑁𝑁𝐷𝐷 is mortality number; 𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚 is the intensity of meizoseismal zone (it is the highest intensity zone of earthquake.); 𝐴𝐴𝑗𝑗 is the distribution area of the jth intensity value; ρ is population density; 𝑅𝑅𝑗𝑗 is the mortality rate corresponding to the jth intensity value, and the statistical relationship of mortality rate and intensity is hereby given. 5. applicability evaluation 5.1. comparison between predicted value and actual value of death toll based on the four types of assessment models, this paper assesses the death toll of the selected 16 earthquakes with epicentral intensity of ⅵ and above, and compares the results with the actual death toll. for the convenience of statistics, these 16 earthquakes are sampled and numbered. the evaluation results are shown in table 1. table 1. comparison between the assessment results of four models and the actual death tolls id date magnitude name of meizoseismal area death toll/person liu jifu badal gb/t30352-201 3 liu jinlon g 1 19840106 5.5 tianzhu 0 1 2 0 1 2 19870108 5.9 diebu 0 34 5 2 2 3 19871025 5 lixian county 0 14 4 0 1 4 19881122 5.7 su’nan 0 0 2 0 1 5 19901020 6.2 tianzhu 1 26 9 2 9 6 19920112 5.4 su’nan 0 2 2 0 1 7 19950722 5.8 yongdeng 12 117 11 14 12 8 19960601 5.4 tianzhu gulang 0 30 6 1 2 9 19990415 4.7 wenxian county wudu 1 9 2 0 1 10 20000606 5.9 jingtai 0 11 9 2 10 11 20021214 5.9 yumen 2 16 5 1 2 12 20031025 6.1 minle shandan 10 553 26 117 13 13 20031113 5.2 minxian county lintan 1 133 14 31 16 14 20040907 5 minxian county zhuoni county 1 37 6 2 2 15 20060621 5 wudu wenxian county 1 7 3 0 1 16 20130722 6.6 minxian county zhangxian county 95 4132 52 864 15 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 88 89 it can be concluded from the assessment results that relative error values of the four models fluctuate obviously; predicted death tolls fluctuate greatly; results of some cases are consistent with the actual results, but results of some cases basically don’t belong to the same order of magnitude and are obviously different. in order to further verify accuracy of the overall assessment results of the four models, mean relative error and root-mean-square error are applied to measure the deviation between estimated value and truth value, as shown in table 2. in terms of overall accuracy, the earthquake damage life loss assessment model proposed by badal and liu performs well in the verification of historical examples in gansu province. table 2. mean relative error & root-mean-square error of the assessment results of the four models model liu jifu badal gb/t30352-2013 liu jinlong mean relative error 96% 81% 142% 96% root-mean-square error 1019.46 12.58 194.18 20.70 5.2. applicability evaluation after analysis and comparison of the above predicted and actual values of death toll, it is found that: (1) these earthquakes are divided into three magnitude levels, and root-mean-square error is applied to measure the relative difference degree between predicted and actual values of the death toll, as shown in table 3. as to 𝑀𝑀𝑀𝑀 < 5.5 and 5.5 ≤ 𝑀𝑀𝑀𝑀 ≤ 6.0, the life loss assessment model based on population density proposed by badal, the life loss assessment model based on epicentral intensity proposed by liu jinlong, and the national earthquake disaster emergency assessment model are relatively accurate, so that they have reference significance when determining earthquake emergency response level and provide scientific support to emergency decision-making and deployment. however, assessment result gained from the life loss assessment model based on gdp per capita and the actual value basically don’t have the same order of magnitude and the value is relatively large in general. this model might be not applicable for gansu province as per the division standard of gdp per capita, resulting in noticeable error of assessment result. generally speaking, when assessing actual earthquake death toll, macroscopic life loss research method might be not applicable for some regions in gansu province. as to 𝑀𝑀𝑀𝑀 > 6.0 , assessment results are obviously not as accurate as those of 𝑀𝑀𝑀𝑀 ≤ 6.0. the difference between some assessment and actual results even exceeds 1.5 times with big contingency. the reason for such obvious difference might be that with the rise of magnitude, the factors that impact earthquake deaths increase or a certain factor is enhanced, so that special earthquake cases need to be assessed by the models of specific applicable range. as to the assessment of death toll after actual occurrence of earthquake, it can be concluded as per experts’ experience on the basis of model assessment. table 3. assessment result error comparison of the four models under different magnitude levels magnitude level root-mean-square error liu jifu badal gb/t30352-201 3 liu jinlon g ms＜5.5 53.34 6.09 11.49 5.79 5.5≤ms≤6.0 45.44 4.44 1.61 4.39 ms＞6.0 2352.04 26.79 448.10 46.57 (2) earthquake disaster casualties are regionalized in gansu province to a certain extent. as hedong region is mountainous with high mountains and deep valleys, it often suffers post-earthquake disasters (landslip, landslide and debris flow), so that the casualties there are higher than hexi region (pei, 2015). for example, the earthquake of m6.6 occurred in minxian county and zhangxian county in 2013 caused death toll of 95, including 14 deaths due to secondary disasters of earthquake. in this paper, this earthquake case is divided into two regions: hedong region and hexi region. after calculating root-mean-square error of the assessment results of the four models respectively and verifying the accuracy, it is found that accuracy of model assessment results of the earthquakes occurred in hedong region is quite different from hexi region, as shown in figure 1. the reason for such obvious difference is the increase in regional impact factors for the assess life loss rate. in practical application, the earthquake-caused mortality number can finally be obtained only after the affected population is obtained. 3) nonlinear regression model based on epicenter intensity liu et al. (2012), based on the data samples of the destructive earthquakes causing mortality in china from 1990 to 2006 along with several major earthquakes damage data in china, found out the main mortality-affecting factors. through gaussian function fitting and regression analysis; they obtained a mortality prediction model with epicenter intensity as a main parameter and magnitude and population density as auxiliary parameters for correction. the final mortality model is shown below: d = 𝑒𝑒12.2𝛼𝛼𝑑𝑑𝑑𝑑𝑑𝑑𝛼𝛼𝑚𝑚∙𝑒𝑒−(ln(ln𝑡𝑡)−2.445) 2∕0.32 (3) where d represents mortality number; ln𝑡𝑡 represents epicenter intensity; 𝛼𝛼𝑚𝑚 is magnitude correction coefficient; 𝛼𝛼𝑑𝑑𝑒𝑒𝑑𝑑 is population density correction coefficient. 4) earthquake damage assessment method stated in the national standard gb∕t 30352-2013 the author selects the mortality assessment method stated in the normative appendix of the earthquake damage assessment promulgated on december 31, 2013 and implemented in july 1, 2014 (china earthquake administration, 2014). the formula is as follows: 𝑁𝑁𝐷𝐷 = ∑ 𝐴𝐴𝑗𝑗 𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚 𝑗𝑗=6 𝜌𝜌𝑅𝑅𝑗𝑗 (4) where 𝑁𝑁𝐷𝐷 is mortality number; 𝐼𝐼𝑚𝑚𝑚𝑚𝑚𝑚 is the intensity of meizoseismal zone (it is the highest intensity zone of earthquake.); 𝐴𝐴𝑗𝑗 is the distribution area of the jth intensity value; ρ is population density; 𝑅𝑅𝑗𝑗 is the mortality rate corresponding to the jth intensity value, and the statistical relationship of mortality rate and intensity is hereby given. 5. applicability evaluation 5.1. comparison between predicted value and actual value of death toll based on the four types of assessment models, this paper assesses the death toll of the selected 16 earthquakes with epicentral intensity of ⅵ and above, and compares the results with the actual death toll. for the convenience of statistics, these 16 earthquakes are sampled and numbered. the evaluation results are shown in table 1. table 1. comparison between the assessment results of four models and the actual death tolls id date magnitude name of meizoseismal area death toll/person liu jifu badal gb/t30352-201 3 liu jinlon g 1 19840106 5.5 tianzhu 0 1 2 0 1 2 19870108 5.9 diebu 0 34 5 2 2 3 19871025 5 lixian county 0 14 4 0 1 4 19881122 5.7 su’nan 0 0 2 0 1 5 19901020 6.2 tianzhu 1 26 9 2 9 6 19920112 5.4 su’nan 0 2 2 0 1 7 19950722 5.8 yongdeng 12 117 11 14 12 8 19960601 5.4 tianzhu gulang 0 30 6 1 2 9 19990415 4.7 wenxian county wudu 1 9 2 0 1 10 20000606 5.9 jingtai 0 11 9 2 10 11 20021214 5.9 yumen 2 16 5 1 2 12 20031025 6.1 minle shandan 10 553 26 117 13 13 20031113 5.2 minxian county lintan 1 133 14 31 16 14 20040907 5 minxian county zhuoni county 1 37 6 2 2 15 20060621 5 wudu wenxian county 1 7 3 0 1 16 20130722 6.6 minxian county zhangxian county 95 4132 52 864 15 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 90 earthquake with casualties where the earthquake magnitude is generally in direct proportion to casualties. but some earthquakes may be different due to earthquake occurrence site, and main impact factors might include population density, geographical environment, origin time, building structure, etc. figure 1. root-mean-square error comparison of the assessment results of four models of the two regions in gansu province based on the above analysis, it is found that earthquake casualties are of great uncertainty and it is really hard to make the assessment results totally the same. thus, comparison in order of magnitude between the assessment results and the actual death toll is considered at the same time, as shown in table 4. earthquake assessment results of the models at hexi region are obviously better than those of hedong district. assessment results of the model proposed by badal and the model proposed by liu are of the same order of magnitude, so that such results can provide accurate basis for the determination of emergency response level when an earthquake occurs. thus, when making life loss assessment in gansu province, regional characteristics of hedong region and hexi region need to be considered and assessment conclusions should be given combining experts’ experience. in the studies of earthquake disaster life loss assessment models, it is especially important to build typical assessment models targeted on regional earthquake life loss characteristics. table 4. consistency comparison in order of magnitude between predicted values and actual values of death toll of the four models model verified earthquake cases same order of magnitude different orders of magnitude hedong hexi hedong hexi liu jifu’s model 16 20% 50% 80% 50% badal’s model 16 80% 100% 20% 0% gb/t30352-2013 16 80% 83% 20% 17% liu jinlong’s model 16 70% 100% 30% 0% 6. conclusion and discussion 6.1. conclusion after destructive earthquakes, the primary task of post-earthquake rescue and relief is to make timely, efficient and orderly rescue and to reduce casualties. as the emergency relief strength input is closely related to casualties, it is of important research significance and practical value to finish rapid assessment of the death toll after earthquake and clarify credibility of results. in journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 90 91 this paper, with applicability comparison of the earthquake disaster life loss assessment models as research objects, conclusions are drawn below: (1) from the theoretical point of view, the empirical formula based on earthquake parameters follows statistical laws and relies highly on historical data. selection of different historical data sets will have great impacts on the establishment of computing methods. but simple parameters acquisition is applicable for the demand of rapid assessment for post-earthquake casualties. thus, empirical formula for casualty assessment is a relatively ideal model for rapid assessment of post-earthquake casualties. (2) from the perspective of practical application, casualties of 16 earthquakes at gansu province are assessed with the four earthquake disaster life loss assessment models and the death tolls concluded from assessment are compared with the actual values. results show that: ① as to 𝑀𝑀𝑀𝑀 < 5.5 and 5.5 ≤ 𝑀𝑀𝑀𝑀 ≤ 6.0,, the life loss assessment model based on population density proposed by badal, the life loss assessment model based on epicentral intensity proposed by liu jinlong and the national earthquake disaster emergency assessment model are relatively accurate, so that they have reference significance when determining earthquake emergency response level and have important support to emergency decision-making and deployment. as to 𝑀𝑀𝑀𝑀 > 6.0, due to less data of such earthquake cases, the number of statistical fitting samples is small, resulting in discrete fitting and error. thus, obvious characteristics in physical geography, social economy, specific customs and other relevant aspects are analyzed to receive comprehensive assessment results, so as to support emergency decisions. ② even though the formulas and methods given by these models basically meet the requirements of regression fit and research results can be concluded from statistical analysis, most models are limited by certain applicable range or input condition. assessment results are obviously regional: accuracy of the model assessment results for the earthquakes occurred in hedong region is not as good as that of hexi region in gansu province. in the studies of earthquake disaster life loss assessment models, it is especially important to build typical assessment models targeted on regional earthquake life loss characteristics. 6.2. discussion from the perspective of practical application, earthquake life loss rapid assessment models are important parts of earthquake emergency relief decision-making and deployment. at present, many types of assessment models have been studied, but few of them are actually applied in relevant earthquake departments, such as earthquake emergency center. the reason is that such assessment models involve too many parameters and limits, so that it is really hard to apply these models in actual conditions. when any destructive earthquake occurs, china earthquake networks center and the provincial earthquake bureau of the earthquake location will release three elements (time, place and magnitude) of the earthquake to the public in about ten minutes and report the possible earthquake disaster affection to the government. the government will start the earthquake emergency relief of rapidly according to earthquake disaster level. it is an urgent scientific problem to input less estimated parameters, reduce subjective factors and improve credibility of earthquake life loss rapid assessment results at present. however, assessment models are often built on the basis of accurately knowing the hardest-hit area, hit area, epicentral intensity and hit population, which leads to higher requirements for deployment of the entire earthquake emergency command technology system. in order to improve the accuracy of earthquake disaster life loss assessment models to better serve earthquake disaster emergency command decision-making of the governments at all levels, it should be studied from the following three aspects: ①essential data should be collected about the disaster situation in an affected zone. a comprehensive collection and classification of management standards should be established. standards for the collection of essential data about disaster situation and effective information resource sharing mechanisms should be established. ② post-earthquake life vulnerability models involve many life loss influencing factors. it is impossible to consider all influencing factors and collect the relations between casualties and all factors so as to obtain a universal empirical earthquake with casualties where the earthquake magnitude is generally in direct proportion to casualties. but some earthquakes may be different due to earthquake occurrence site, and main impact factors might include population density, geographical environment, origin time, building structure, etc. figure 1. root-mean-square error comparison of the assessment results of four models of the two regions in gansu province based on the above analysis, it is found that earthquake casualties are of great uncertainty and it is really hard to make the assessment results totally the same. thus, comparison in order of magnitude between the assessment results and the actual death toll is considered at the same time, as shown in table 4. earthquake assessment results of the models at hexi region are obviously better than those of hedong district. assessment results of the model proposed by badal and the model proposed by liu are of the same order of magnitude, so that such results can provide accurate basis for the determination of emergency response level when an earthquake occurs. thus, when making life loss assessment in gansu province, regional characteristics of hedong region and hexi region need to be considered and assessment conclusions should be given combining experts’ experience. in the studies of earthquake disaster life loss assessment models, it is especially important to build typical assessment models targeted on regional earthquake life loss characteristics. table 4. consistency comparison in order of magnitude between predicted values and actual values of death toll of the four models model verified earthquake cases same order of magnitude different orders of magnitude hedong hexi hedong hexi liu jifu’s model 16 20% 50% 80% 50% badal’s model 16 80% 100% 20% 0% gb/t30352-2013 16 80% 83% 20% 17% liu jinlong’s model 16 70% 100% 30% 0% 6. conclusion and discussion 6.1. conclusion after destructive earthquakes, the primary task of post-earthquake rescue and relief is to make timely, efficient and orderly rescue and to reduce casualties. as the emergency relief strength input is closely related to casualties, it is of important research significance and practical value to finish rapid assessment of the death toll after earthquake and clarify credibility of results. in journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 85-92 92 statistical formula. according to actual demands, based on the space-time characteristics of earthquake disaster life losses in details, regional assessment models are developed to improve or solve the problems related to the accuracy of assessment models, such as building regional models or building assessment models in different magnitude levels; ③ existing models of high accuracy have complex parameters and require much for assessed data, so that it is really hard to collect and update relevant data in practical application, resulting in poor practicability. the life vulnerability models suitable for practical application should be characterized by few parameter input, easy collection and update of basic data with scientific and reasonable model assessment results. it needs further study to build simple and practical life vulnerability models. acknowledgement the article is supported by national key r&d program of china (no. 2017yfc1500906) and science and technology support program of gansu province（no. 1504fkca065） references badal j, vázquez-prada m, álvaro gonzález, “preliminary quantitative assessment of earthquake casualties and damages”, natural hazards, vol.34, no.3, (2005) pp.353-374. chen q f, hongliang m i, huang j, “a simplified approach to earthquake risk in mainland china”, pure & applied geophysics, vol.162, no.6-7, (2005) pp.1255-1269. chen yong, liu jie, chen qifu et al. 1999. analysis of earthquake hazard and forecast of earthquake disaster. beijing: earthquake press, china earthquake administration, china statistics bureau.2015. zhong guo dalu dizhen zaihai sunshi pinggu huibian(1990-1995). beijing, china: seismological press. china earthquake administration. 2014. gb/t30352 2013 assessment of earthquake disaster situation in emergency period. beijing: china standard press. emergency rescue department of china earthquake administration. 2000. zhong guo dalu dizhen zaihai sunshi pinggu huibian (1996-2000) beijing, china: seismological press. emergency rescue department of china earthquake administration. 2010. zhong guo dalu dizhen zaihai sunshi pinggu huibian (2001-2005). beijing, china: seismological press. emergency rescue department of china earthquake administration. 2015. zhong guo dalu dizhen zaihai sunshi pinggu huibian (2006-2010). beijing, china: seismological press. emergency rescue department of china earthquake administration.2015. zhong guo dalu dizhen zaihai sunshi pinggu huibian(1966-1989). beijing, china: seismological press. federal emergency management agency (fema). hazus99 earthquake loss estimation methodology, technical manual, washington. dc, united states: federal emergency management agency. 1999. fu zhengxiang, jiang lixin, li geping, “analysis of temporal-spatial and intensity distribution of life losses caused by earthquake hazards”, earthquake, no.2, (1994) pp.1-10. he yuling, fan lingchun, li dahu et al, “ design of earthquake disaster prediction &evaluation system in sichuan province and its function”, north china earthquake sciences, vol.23, no.2, (2005) pp.32-36. li yuanyuan, su guofeng, wengwenguo et al, “areview of researches on seismic casualty estimation”, journal of catastrophology, vol.29, no.2, (2014) pp.223-227. liu baichi, zhang junling, wu jianhua, guo hua. reevaluating on casualty in haiyuan ms8.5 earthquake on december 16.1920. earthquake research in china, vo19, no.4, (2003) pp.386-399. liu jifu, chen yu, shi peijun et al, “a study on seismic risk assessment in mainland china (ii): life vulnerability model”, journal of beijing normal university (natural science), vol.45, no.4, (2009) pp.404-407. liu jinlong, lin junzhi, “study on assessment method for earthquake casualty based on epicentral intensity”, journal of natural disasters, vol.21, no.5, (2012) pp.113-119. ministry of civil affairs of the people’s republic of china. interpretation of the regulations on natural disaster relief. natural disaster reduction in china. no.19, (2010) pp.4-7. national earthquake contingency plan, gazette of the state council of the people's republic of china, no.28, 2012. pei huijuan, zhou zhonghong, sun yanping, chen wenkai. spatiotemporal distribution feature of earthquake disaster in gansu province. journal of natural disasters. vol. 24, no.3, (2015) pp.67-75. samardjieva e, “estimation of the expected number of casualties caused by strong earthquakes”, bulletin of the seismological society of america, vol.92, no.6, (2002) pp.2310-2322. yin zhiqian. “a study for predicting earthquake disaster loss”, earthquake engineering and engineering vibration, vol.11, no.4, (1991) pp.87-96. microsoft word introduction for volume 3, issue 1 editor’s introduction modern society is a competitive society and to reduce existing costs, which is the ultimate goal of risk analysis and crisis response, is an important means to enhance the competitiveness. in the present issue of journal of risk analysis and crisis response (jracr), volume 3, issue 1 (2013), such efforts have been embodied in these papers. this issue contains 8 papers. the first 5 contributions are written in english and others in chinese with english abstracts. the papers can be divided into five categories: risk assessment, risk management, credit risk, accident risk and risk analysis. the risk assessment category includes two papers. the first paper "risk assessment of radio frequencies and public information" by marc poumadere and anne perrin, explores the lesserknown notion of socio-cognitive exposure, characterized by chronic exposure of populations to potentially worrying information when various health consequences are evoked in the literature and by the media. this raises the possibility of a link between risk information and health. several specific explanations are explored here, notably: nocebo, stress, and the symmetry rule. the second paper "a risk assessment model based on attribute theory" by feng and sheng, uses qualitative theory method to investigate the deep relationship of the five parties and how they impact others and builds a mathematical model to guide us how to make a proper interest rate to deal with various economical issues. there are two papers in the risk management. the first paper "risk scenes of managerial decisionmaking with incomplete information" by dusan marcek, explains the relation of a degree of risk expressed by the classical information measure, bit, by the concept of confidence intervals, or possibly by the standard deviation. forecasting models are applied which are based on a statistical theory and a neural approach. the aim is also to examine whether potentially highly non-linear neural network models outperform the advanced statistical methods and better reduce risk in managerial decision-making, or they yield competitive results. a method for finding the forecasting horizon within which the risk is minimal is also presented. the second paper "an operational drought risk management framework based on stream-flow intelligent internet control" by li et al. in which an operational drought risk management framework based on the stream-flow intelligent internet control is proposed. in the proposed framework drought can be predicted, evaluated and mitigated by using a dynamic stream-flow control under the sensors detection. the credit risk includes two papers. the first paper "the research on applicability of amended kmv model with different industries" by yang et al. amends the parameters of default point. further the paper also analyzes the applicability of amended kmv model in different industries based on the data of chinese listed companies. the second paper "types of credit risks and strategies to improve risk identification by internet of intelligences" by huang, shows four categories’ intensions, respectively, as well as overviews the main methods to analyze them. the direct identification strategy in internet of intelligences can improve the recognition of credit risk to the firm which closely relates to public life, and there is a lot of information available to public. the indirect feeling strategy in internet of intelligences can improve the recognition of credit risk that would be seriously affected by changes in the political, economic and social environment. journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 1-2 published by atlantis press copyright: the authors 1 there is one paper on accident risk, "analytical research on the accident risk in three industries of china and usa in the period of 2006-2010" by wu, zeng and zhang. it analyzes the accident risk in three industries of china and usa in the period of 2006-2010, to explore effective channels attributable to improving safety situation. in conclusion, effective channels attributable to improving safety situation are to change industry structure, develop the tertiary industry and plan the tertiary industry development in a reasonable way. the last category includes one paper, "the spatial symmetry axis of earthquake hazard in china" by yan, li, bai and liu. it takes the spatial symmetry axis of earthquake hazard in china as the key of assessing natural disaster risk to emphatically analyze the rationality of the existence of the spatial symmetry axis of the regions of western china, yunnan province and taiwan and find out the dynamics basis of spatial symmetry axis combing with those presentational features, so as to provide some theoretical thinking and methods for reference to identify the time, spatial trends of the earthquake disaster occurrence. the results show that the earthquakes have the quality of spatial symmetry in certain regions and the symmetry has the basis of dynamics, which can work as the basis of the regional earthquake trend. and further improvements are needed to judge the earthquake trend to combine the spatial imagery, dynamics basis and physical mechanism. we sincerely thank the referees, who reviewed submissions to this issue, for their strong support and kind help. and also we thank the authors very much for all their contributions. editor-in-chief chongfu huang professor, beijing normal university e-mail: hchongfu@gmail.com editorial assistant ye xue associate professor, taiyuan university of technology e-mail: xueye0412@126.com published by atlantis press copyright: the authors 2 microsoft word 170628-a study on construction method of consensus measure space of macro-seismic anomalies.doc a study on construction method of consensus measure space of macro-seismic anomalies weidan wang1,2, chongfu huang1,2,3,* 1 state key laboratory of earth surface processes and resources ecology, beijing normal university, beijing 100875, china 2 academy of disaster reduction and emergency management, faculty of geographical science, beijing normal university, beijing 100875, china 3 beijing cazl technology service co., ltd., beijing 100088, china abstract macro-anomalies are so complex that it is difficult to carry out the systematic quantitative analysis macro-anomalies because it is lack of measure space. it’s more difficult to find the relationship between micro-anomalies and earthquakes. in the paper, the theory system of macro-anomalies group’s measure space which is expressed by the fuzzy relationship matrix is set up. the concept of macro-anomalies group is put forward. scene, elements and attributes are used to describe the relationship among macro-anomalies group, macro-anomalies and basic information of macro-anomalies. based on the expression of basic information and the evaluation information, according to properties of the elements of a scene, basic indicators that describe macro-anomalies group can be calculated. the generation of results based on consensus of the first-line earthquake workers’ evaluation of macro-anomalies group’s comprehensive abnormal degree is expounded. the specific process and related algorithm is raised that use information diffusion technique calculate the fuzzy relationship matrix to express macro-anomalies group’s measure space, and it takes basic indicators of macro-anomalies group as input, the evaluation information of first-line earthquake workers as output. keywords: earthquake, macro-anomaly, measure space, information diffusion, internet of intelligences. 共识性的地震宏观异常群测度空间及其构建方法研究 王蔚丹 1,2 ，黄崇福 1,2,3,∗ 1.北京师范大学地表过程与资源生态国家重点实验室，北京 100875 2.北京师范大学地理科学学部减灾与应急管理研究院，北京 100875，中国 3.北京崇安智联科技服务有限责任公司，北京 100088，中国 摘 要:针对人们缺少综合度量地震宏观异常的测度空间，难以开展系统性的量化分析，更难以寻找宏观异 常与地震之间可能蕴含的关系等问题，本文建立了用模糊关系矩阵表达地震宏观异常群测度空间的理论体系：提 出了宏观异常群的概念，利用情景、要素、属性来描述宏观异常群、宏观异常和宏观异常基本信息之间的关系； 进行了宏观异常群基本信息及评判信息的表达，根据构成情景的要素的基本属性—宏观异常基本信息，可以得到 描述情景—宏观异常群的基本指标；阐述了一线地震工作者对宏观异常群情景的综合异常程度进行评判得到共识 ∗ corresponding author: e-mail: hchongfu@bnu.edu.cn. post address: academy of disaster reduction and emergency management, beijing normal university, no.19 xinjiekouwai street, beijing 100875, china this project was supported by the national natural science foundation of china (41671502), and partly supported by the beijing cazl technology service co., ltd.. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 166 received 28 june 2017 accepted 10 july 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). javascript:void(0); 性结果的过程；将宏观异常群的基本指标作为输入，一线地震工作者对宏观异常群情景整体异常程度的共识性结 果作为输出，提出了用信息扩散技术计算模糊关系矩阵表达地震宏观异常群测度空间的具体过程和相关算法。 关键词:地震，宏观异常，测度空间，信息扩散，智联网 1. 引言 现阶段对宏观异常的统计分析缺乏大量可靠、完 备的数据，导致宏观异常的研究裹足不前。现阶段利 用统计方法对宏观异常现象进行分析，很难再有突破 性进展，只能不断积累数据。宏观异常种类繁多，表 现形式多样，仅已知的动物异常就有几十种之多，各 个异常又有具体的表现形态。在缺乏大量可靠、完备 宏观异常数据，进行统计分析难以得到行之有效规律 的情况下，拿到宏观异常信息时，该如何分析处理这 些信息成为一个问题。 目前的宏观异常研究缺乏系统化的量化工具。宏 观异常的认识和运用主要由地震专家来完成，无论是 宏观异常的现场核查，还是对宏观异常进行宏观把握， 辅助地震预报，都离不开专家的参与。我们把专家看 作一个个高级智能体，基于智联网平台 [1] ，利用一线 地震工作者的经验和知识，建立一个共识性的测度空 间综合度量宏观异常，以便开展系统性的科学研究。 2. 宏观异常群相关概念 宏观异常：指人类用感官可直接觉察到的自然界 中的不正常现象 [2] 。 大震前一般是有宏观异常的。我国近四千年的文 化历史资料中有丰富的关于宏观异常资料的记载。将 历史资料中出现频率较高的宏观异常，如井水浑浊， 翻花冒泡，断水自流；蛇冬眠时出洞，老鼠惊慌乱跑， 狗成群乱咬乱叫，大猫叼着小猫搬家等，称为典型异 常。宏观异常一般不会孤立出现，宏观异常的“宏” 不仅体现在人类用感官就可以察觉到，还表现为它一 般成群地出现在一定的空间范围内。 宏观异常表现形式多样，种类多达几百种，异常 现象的表现更是数以千计。宏观异常如此复杂，其是 否出现，又有很大的不确定性。相对单个的宏观异常 而言，大规模的群体性的宏观异常，对辅助判断地震 是否有可能发生，更有指示意义。由此我们提出宏观 异常群的概念。 宏观异常群：指在一定时空范围发生的多个宏观 异常组成的群体。 不同地震宏观异常现象存在地域上的相关性和时 间上的同步性特征，历史资料及文献中对于比较显著 的群发性宏观异常，一般用种类多、数量多、范围广、 规模大、随时间剧烈变化等来描述。由于缺少综合量 化的理论和方法，加之人们还不能在震前及时汇集和 处理宏观异常群信息，宏观异常对地震预报的辅助决 策尚待改进。 我们把一个个的宏观异常看成一个个的异常要 素，由若干个宏观异常构成了宏观异常群，也就在某 个时间点上形成了关于一个地域的异常情景。例如，7 个宏观异常组成了一个宏观异常情景，投影在平面上 用地图显示就形成了图 1 中所示。为了综合度量某一 时空范围出现的宏观异常，提出宏观异常群严重度的 概念。 宏观异常群严重度：指宏观异常群的综合异常程 度。 宏观异常群严重度类似于地震灾害中建筑物的破 坏严重度程度，是一个综合的量。一个宏观异常群的 异常程度，由所出现的宏观异常的种类、数量、范围、 规模以及单个宏观异常的异常幅度等指标所决定。用 区间[0,10]中一个适当的数值来表示严重程度。0 为无 异常；10 为非常严重。 本文的主要工作，就是探讨如何根据宏观异常群 中的异常种类、数量、范围、规模以及单个宏观异常 的异常幅度等，计算宏观异常群严重度，从而综合度 量宏观异常，为辅助地震预测提供信息。 3. 宏观异常群信息的表达 3.1 宏观异常群信息的表达 我们把一个宏观异常群看成一个情景，它由若干 个要素——宏观异常组成。每个宏观异常都有它自身 的属性，如宏观异常所属类别、发生时间、持续时间、 发生地点（经纬度）、异常描述及异常所处环境等。图 2 展示了宏观异常群层次结构示意图。 一个宏观异常群记为 g，它由多个宏观异常 m1, journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 167 m2, …构成。一个宏观异常由多个属性 f1, f2, …进行描 述。一个宏观异常所属类别、发生时间、持续时间、 发生地点（经纬度）、异常描述及异常所处环境等，均 是其属性。式（1）表达了由 n 个宏观异常构成的一个 宏观异常群 g。式（2）表达了由 l 个属性描述的一个 宏观异常 m。 g=（m1, m2, …, mn） （1） m=（f1, f2, …, fl） （2） 图 1. 宏观异常群情景 图 2. 宏观异常群层次结构示意图 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 168 假设宏观异常群 g 由 n 个基本单元——宏观异常 来组成，如式（1）所示，mj 表示构成宏观异常群的第 j 个宏观异常，j=1,2,…,n。根据宏观异常历史资料，一 条宏观异常信息通常包括异常的类别、种类、出现时 间、结束时间、出现地点以及宏观异常现象的详细描 述等。若宏观异常 mj 可以通过 l 个属性特征来表示， 则 mj 的属性特征集 fj 可具体地表示为式（3）： {( , , , , , , ) 1, 2, , }j j j j j j j jf id v t x y d a j n= = l （3） id 是异常的唯一标识，vj 表示宏观异常群 g 的第 j 个 宏观异常 mj 所属的异常类别，t 是异常发生的时间， 异常发生的地点这里用 x,y 表示，d 是对异常现象的具 体表述，a 表示异常幅度，即该异常区别于正常的显 著程度。 为了尽可能地将所有的异常都包括进来，参照已 有的宏观异常历史资料，我们采用《地震宏观异常预 报方法》一书的分类体系，许敦煌对汶川地震前宏观 异常进行现场调查 [3] ，张小涛等对汶川地震前后出现 的宏观异常现象收集整理和统计分析 [4] 就采用了该分 类方法，上面所述的气象异常有中期异常（旱涝异常） 和短期异常（气温、地温、湿度、气压、降雨、风等 气象变化异常），本文将第五类宏观异常表述为气候气 象异常，具体分类情况见文献 [5] 。 v 是属性异常类别对应的属性特征值的集合，前 文我们把宏观异常分为五大类，可用式（4）来表示： v={v1,v2,。。。,v5} ={生物异常，地下流体异常，地球物理场异常， 地质现象异常，气候气象异常} （4） 以往研究地震和宏观异常的关系时，我们常常统 计震前出现的宏观异常种类、数量，时空分布规律等， 较少将宏观异常本身反应的强烈、明显程度加以考虑。 我们把宏观异常反应的强烈、明显程度称为异常幅度。 异常幅度也是描述宏观异常的一个重要指标。 地震宏观异常涉及到异常种类繁多，异常表现形 式多样，很难有一个统一的标准。异常幅度a带有很大 的主观性，可以根据宏观异常具体表现形态加以判断， 或者从异常描述判断其异常幅度。为了研究方便，可 将异常幅度按照级别加以划分，如在描述地下水和地 下油气时可能出现水位大幅度上升；水位大幅度下降； 枯井开始冒水；水位不断升高，同时伴有响声；泉水 流量剧增；泉水流量减少或断流；井水（泉水、水库、 龙洞等）发浑；水变色、翻花冒泡、变味或有异味； 飘油、打旋、有响声；田里冒水或喷沙；水温升高或 降低等词句。李坤、李秀英在研究地下水和强震之间 的关系时，将异常分为一般异常和重要明显异常两种。 一般异常，主要表现为水位上升、下降；冒泡、水变 浑等。重要明显异常，主要以如井水突然自流、地面 翻砂冒水、水位忽升忽降和水变色等形式出现，他们 提出后者对发震时间和震中预报有极其重要的作用 [6] 。根据他们的区分方式，为了我们的研究方便，这 里将异常幅度分为两类，一般异常和重要明显异常。 异常幅度a对应的属性特征值的集合可具体地表示为 式（5）： 1 2={ , }a a a ={一般异常，重要明显异常} （5） 具体地，对地下流体异常来说，将变色、自流、 井喷、翻砂冒水、水位忽升忽降作为重要明显异常， 水位上升、下降、流量改变、翻花冒泡、水变浑等作 为一般异常。对生物异常来说，区域内群体性异常、 或者反应特别剧烈等可作为重要明显异常，如 2000 年 9 月 12 日青海兴海 6.6 级地震前，在“震中区温泉一 带 95%以上狗有异常反映，狗异常狂犬，且声音凄凉， 拖声很长，似狼嗥声”将其看作重要明显异常；1997 年 5 月 31 日福建永安地震前“胶南县百灵鸟笼之中乱 飞乱跳致死”也可作重要明显异常。其它的可看作一 般异常。地声、地磁、地光及有感地动等地球物理场 异常以及地质现象和气象气候异常难以判别其异常幅 度，暂时将其看作一般异常。 3.2 宏观异常群的描述指标 以往对于比较显著的群发性宏观异常，一般用种 类多、数量多、范围广、规模大、随时间剧烈变化等 来描述。这些都是对出现在一定时空范围内根据式 （3），对某个宏观异常群中的单个宏观异常的异常类 别、发生时间、发生地点等属性信息进行统计，可得 到该群出现的宏观异常个数，出现的异常种类数，时 间、空间分布情况等。除此之外，群体的综合异常程 度还和个体的异常幅度有关。由此，我们用种类数量、 分布情况和个体强度这些参量来描述宏观异常群，具 体地，种类数量指宏观异常出现种类数和异常总数， 即出现了哪几类，共多少起异常；分布情况主要用空 间密集度和时间频度来表示；个体强度主要是指一起 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 169 宏观异常表现的强烈程度。其中，空间密集度指宏观 异常总量在空间范围的分布密度，时间频度指宏观异 常在单位时间内出现的次数。 不同异常现象在空间上的相关性和时间上的同步 性是在宏观异常研究中尤其要关注的内容，在本文中， 为了简化研究，遵循从简单到复杂的原则，采用种类、 总量、空间密集度、异常幅度来建立相应的测度空间。 假定宏观异常群分布在一定的时间范围内，时空演化 的整体研究可以在后续研究中来实现。 根据前面的分析，把宏观异常划分为 5 大类，n 表示宏观异常群中各类异常出现的次数集合，用式（6） 表示。 1 2 5 ( , , , )n n n n= l （6） n1 表示宏观异常群中第一类异常出现的个数,nj 表示 宏 观 异常 群中 第 j 类 异常 出 现的 个数 ， 其中 ， 。这样，该宏观异常群出现的异常类别数 及异常总起数就能计算出来。 1, 2, , 5j = … 空间密集度即宏观异常群在空间上密集程度。可 以用单位面积内的宏观异常总量来表示，即总起数/面 积。例如，在一定时间范围内出现三起异常，分别分 布在西北、东南和东北，可知其分布及其稀疏，密集 程度低。相对来说，如果在云南省内东北部出现了三 起宏观异常，其在空间上的分布密集程度较高。(xj, yj) 表示宏观异常群中第 j 起异常的位置，据此可计算出 宏观异常分布范围，用其覆盖面积来表示，利用宏观 异常群出现的异常总量和面积可得到空间密集度指标 。如此，只要将宏观异常群的面积表现出来即可。 接下来看一下宏观异常群情景的形式化表达。 c 3.3 宏观异常群情景及其评价信息的表达 宏观异常群情景的表达。一个宏观异常群可以看 作一个宏观异常情景，不失一般性，设有 个宏观异 常情景，则 κ （7） 1 2{ , , , }s s s sκ= l 其中，si 表示第 i 个宏观异常情景，其中 i=1,2,…, κ 。 假设第 i 个情景 si 由 n 个基本单元组成，则 si 可 表示为式（8）： （8） 1 2, , ,is f f f= l（ n） 1 2ij ij ij ijl 其中，fij 表示构成第 i 个情景的第 j 个宏观异常要素， 其中 i=1,2,…, ；j=1,2,…,n。若宏观异常要素 fij 可以 通过 l 个属性特征来表示，则 fij 的属性特征集 fij 可表 示为式（9）： κ { , , }f f f f= l （9） 其中，fijb 表示构成第 i 场情景的第 j 个要素所具有的第 b 个属性特征。 为了后续计算方便，我们对宏观异常群构成的情 景信息进行综合表达。第 i 场情景可用式（10）来表 示。 1 2 5{( , , , , , , , ) 1, 2, , }i i i i i i i is id nt nv n n n iγ κ= =l l， （10） 其中，si 表示第 i 个情景，id 是情景的唯一标识， nt 表示情景 s 包含的要素总数，即宏观异常个数， nv 是情景中异常类别数， 1in 表示第一类异常 在情景 si 中出现的个数， 以此类推。 1v ,2 5, ,in nl ，i iγ 表示情景 i 中宏观异常覆盖的面积。 宏观异常情景评价信息的表达。假设呈现出一个 宏观异常群情景，一线地震工作者对该情景的综合异 常程度进行评判，这个评判信息怎么表达呢？目前地 震的相关研究大多还处于初级阶段，特别是宏观异常 在实际工作中发挥作用，更多的是依靠经验。我们要 建立相关的智联网服务平台，一线地震工作者们可根 据自身的经验和智慧对平台推送的宏观异常群情景进 行整体把控，得出一个判断结论。 可以直接把自然语言映射成某个数值或数值区 间。例如把严重程度从不严重到非常严重映射到[0,10] 区间上，“比较小，不太严重”转换为数值 3。有时为 了方便，我们常常分区间来表达，例如可将[0,10]划分 为 5 个级别，[0-2]表示不严重，[2-4]表示不太严重， [4-6]表示严重，[6-8]表示比较严重，[8-10]表示非常严 重。直接选择和判断最接近的区间，这个过程其实是 进行逻辑匹配的过程。不论是直接映射成数值还是直 接选择区间，都不可避免地丢失了很多信息。把这些 描述转换成某个论域上的模糊子集可以更准确地表达 模糊评价信息。我们要做的是将这种模糊判断通过画 图画出来。 如图 2 所示，对出现的宏观异常组成的群体强度 做一个整体评判。横轴表示针对某一宏观异常群中出 现的全部宏观异常的种类、数量、范围、规模以及单 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 170 个宏观异常的异常幅度等做出的严重程度判断，数值 越大表示越严重，用区间[0,10]中一个适当的数值来表 示，这个数值就是前面提到的宏观异常群严重度。这 里不严重为 0，非常严重为 10。纵轴表示对相对应严 重程度的可能性判断，可以理解为信度或者隶属度， 是对相对宏观异常群严重度的可能性判断。 一线地震工作者对群体异常严重程度的评判是模 糊的，前面提到了宏观异常群严重度和隶属度的概念， 我们以 来表示宏观异常严重度。出现一个宏观异常 群，一线地震工作者通过画曲线的形式模糊输入进行 评判，这个判断是连续的，存储的时候就需要根据需 z 要进行采样，得到离散化的一组数据。这样每次评判 都可以得到一组 ( , )pυ 数据，其中υ 可看做宏观异常群 严重度 z 的监控点空间，p 是对应的隶属度。 假设一共进行了 k 次评价，记 {1, 2, , }θ β= l ，取 宏观异常群严重度的离散论域 1 2={ }βυ υ υ υl， ， ， ， [0,10]jυ ∈ ， 第 i 次评价结果记为 ji，则 1 1 2 2{( , ),( , ), ,( , )}i i i i i i ij p p pβ βυ υ υ= l （11） pij 表示 ijυ 对应的隶属度，其中 1, 2, , ; 1, 2, ,i k j β= … = … 。 那么 1 2i i i i{ , , , }p p p p β= l （12） 式（12）表示组成离散论域的这些标准点属于第 i 次评价信息这个模糊子集的隶属度集合。 用 表示第 i 次评价中最大隶属度，则,i maxip maxi θ∃ ∈ ，使得 , =max ( )i maxi ip p （13） 其中， 1, 2, ,i k= l 。 ,i m pυ 表示最大隶属度对应的宏观异常群严重度， 称其为最可能宏观异常群严重度。 , ,i mp i maxiυ υ= （14） 其中， 1, 2, ,i k= l 。同样 mp θ∈ 。 例如，假设图 3 所示是第 3 次评价所绘制的模糊 输入曲线，令 1 2={ , }={0,1,2, ,10}mυ υ υ υl l， ， 根据所绘制的曲线，评价结果可用一组数值表示 3 3,1 3,1 3, 2 3,2 3,11 3,11{( , ), ( , ), , ( , )} ={(0, 0), (1, 0), (2, 0), (3, 0), (4, 0), (5, 0), (6, 0.328), (7, 0.552), (8, 0.635 0.469), (9, ), (10, 0.141)} j p p pυ υ υ= l 其对应的隶属度集合如下： 3 {0,0,0,0,0,0,0.328,0.552,0.635,0.469,0.141}p = 其中最大隶属度为 p3,maxi=max(p3)=0.635，它所对应的 宏观异常群严重度 3, m p =8υ 。 4. 宏观异常测度空间建立 宏观异常现象复杂，宏观异常研究相关数据缺乏， 在这样情况下，要想对某一时空范围出现的宏观异常 群体进行综合度量是不容易的。而对宏观异常的认识 和运用目前主要由一线地震工作者来完成，由此提出 图 3. 一线地震工作者将评判结果进行模糊输入 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 171 一种测度宏观异常群的思路，即一线地震工作者根据 自身知识和经验将对宏观异常群的综合异常程度的评 判结果进行模糊输入，对多个评判结果归一化后进行 综合处理，得到一个共识性的结果。其产生过程如图 4 所示。 4.1 宏观异常群测度空间的具体建立过程 我们构建测度空间的过程主要是计算描述宏观异 常群的指标和一线地震工作者评判宏观异常群情景的 综合异常程度结果之间的模糊关系，具体过程如图 5 所示。假设生成了 λ 个宏观异常群的模拟情景，一线 地震工作者对随机出现的宏观异常情景进行评判，得 到 k 次评价结果，对宏观异常群的描述指标做降维处 理，得到描述空间异常群的两个指标：空间密集度 c 和非空间综合指标 s。空间密集度可由情景中出现的宏 观异常位置坐标计算出；非空间综合指标是对宏观异 常群中出现的宏观异常类别、各类别出现次数及单个 宏观异常的异常幅度进行综合得到的，综合的过程用 到了各异常类别的权重，它的计算来源于 k 次评判结 果。根据得到的样本数据自身构造各自的监控空间， 和评判信息的归一化结果结合进行三维信息扩散，得 到宏观异常群的测度空间。 图 4. 一个模拟情景共识性结果的产生过程. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 172 4.2 宏观异常群的综合指标计算 为了更好地建立宏观异常的测度空间，综合度量 宏观异常群，同时由于样本数量较少，需要对描述宏 观异常群的变量降维处理，即把异常种类，异常起数 和异常幅度综合成一个非空间综合变量作为一个输 入，空间聚集度作为第二个参量，把一线地震工作者 的评判结果作为输出，通过建立三者之间的输入—输 出关键构建宏观异常群的测度空间。首先我们可以从 单个宏观异常信息的形式化表达式里计算出整个宏观 异常群的空间密集度。 4.3 宏观异常群空间综合指标的建立 用空间密集度表征宏观异常群的空间属性，指一 定时空范围内出现的宏观异常其在空间上的聚集程 度。本文建立宏观异常群的测度空间时，其空间综合 指标暂时用单位面积内的宏观异常总量来表示，即总 起数/面积。(xij,yij)表示第 i 个宏观异常群情景中第 j 起 异常的位置。据此可计算出空间密集度指标 ci。对一 个一线地震工作者来说，他进行某一次评价，异常情 景是随机出现的。假设进行第 e 次评价时出现了第 i 个情景。对一个异常情景来说，它可能出现不止一次。 假设有λ 个虚拟情景被随机推送，对第 i 个宏观异 常群情景 si 来说，设其有 n 起异常，则所有异常横坐 标和纵坐标集合分别为： 1 2 1 2 ={ , , } ={ , , } i i i in i i i in x x x x y y y y ⎧ ⎨ ⎩ l l （15） 其中， 1, 2, , 1, 2, ,i j nλ= =l l； 。 一定时空范围内宏观异常出现的次数统计即可， 但对于面积的计算有多种方法，最简单的方法是根据 宏观异常群中异常的位置坐标(xij,yij)的最大值最小值 确定的矩形来计算。 本文采用生成最小凸多边形的方法来计算宏观异 常群的面积，其具体的计算过程如下： 进行坐标转换。对宏观异常在 unity3d 中的世界 图 5. 宏观异常群测度空间的构建过程 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 173 坐 标 进 行 坐 标 转 换 ， 得 到 转 换 后 的 坐 标 集 合 为 1 1 2 2{ ( , ), ( , ), , ( , )}i i i i in inx y x y x y% % % % % %l 。 生成异常点对应的圆。一个宏观异常对应空间上 一个点，考虑一个宏观异常辐射周边区域，对周边区 域有大致的影响范围，这里假设这个影响范围是以宏 观异常点为圆心，以一定范围为半径的圆形区域。将 该圆形以离散点的集合 1 1 2 2{( , ), ( , ), , ( , )}i i i i im imx y x y x y ) ) ) ) ) ) l 来表示。 生 成 最 小 凸 多 边 形 。 根 据 离 散 点 的 集 合 1 1 2 2{( , ), ( , ), , ( , )}i i i i im imx y x y x y ) ) ) ) ) ) l 生成最小凸多边形，它必 须包含所有的点，同时必须尽可能获得最小的面积。 具体采用的是分而治之算法，时间复杂度接近于 o(n*logn)，算法流程如下： （1） 找到对角线。找出点集中左边最上的点，找到 右边最下的点，可以确定从左上到右下的一个向量 a r 。 （2） 获取对角线左边的顶点序列。将向量 a r 左边的 距向量最远的点插入序列中，形成两个新的向量 1b r 和 。获取第一个向量 的左边顶点序列，形成新的两 个向量 和 ，如此这般，直到没有新的顶点序列产 生。接着获取第二个向量 的左边顶点序列。直到遍 历所有向量，不再有左边顶点序列产生。这是一个递 归的过程。 2b r 1b r 3b r 4b r 2b r （3） 翻转对角线，获得一个右下到左上的向量 b r 。 （4） 获取对角向量 b r 左边的顶点序列。如同步骤（2） 一样进行递推，直到不再产生左边顶点。最后得到的 顶点序列中所有的点就可以构成一个凸多边形，它就 是点集 1 1 2 2{( , ), ( , ), , ( , )}i i i i im imx y x y x y ) ) ) ) ) ) l 的最小凸多边 形。 计算最小凸多边形的面积 r。通过前面的处理，利 用百度地图中开源库的 api 函数计算我们生成顶点序 列的面积，函数为 bmaplib.geoutils.getpolygonarea 得到的就是最小凸多边形的面积。该函数返回 number 数据类型，能够计算出多边形面积，或者由点数组构 建的图形面积，值得注意的是：所计算的 polygon 坐标 类型只能是经纬度，且不适合计算自相交多边形的面 积 [7] 。 用最小凸多边形方法确定矩形法计算出的面积如 图 6 所示，红色气泡表示宏观异常点，红色虚线表示 最小凸多边形方法的面积范围。 如此就可以计算出宏观异常群的空间密集度，如 式子(16)所示。 nt c γ = (16) 宏观异常群非空间指标的建立。除了空间因素之 外，我们把一个宏观异常群情景中出现的种类、数量、 异常幅度等指标进行处理，得到一个非空间的综合指 图 6. 最小凸多边形计算宏观异常群面积结果示意 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 174 标 s，其中宏观异常一共分为 5 类。建立该指标的过程 中除了需要情景中出现的宏观异常相关的数据外，还 需要借助一线地震工作者对相应情景的评判信息。设 第 e 次评价时，随机出现了第 i 个情景 si。共有λ 个情 景，进行了 k 次评价，则第 e 次评价对应的 si 的非空 间信息及其评价信息表示为式（17）： 1 2 ,5 ,( , , , , , , )e i i i ei e i n n n mpυ= l （17） 其中 。把1, 2, , ; 1, 2, ,e k i= =l l n ei 作为矩阵的一 行，共有 k 行。 ei 表示这是第 e 次评价，e 是评价 id，i 是情景 id， ni1 表示第一类异常 v1 在情景 si 中出现的个数， ,e m pυ 表 示最大隶属度对应的宏观异常群严重度。 宏观异常群非空间指标的建立主要分两部分进 行。首先进行种类重要性排序，计算各类别权重值。 对于式（17）的所有类别来说，若 >0 则令 =1。 把无差异情景行即每类异常都有的异常情景所在的行 去除，即若 >0， ，则把第 i 个情景对应 的所有评价信息行去除。再分别针对某类异常进行处 理计算。如对第一类异常 来说， ，则把相对 应的第 i 个情景对应的所有评价信息行去除，然后把 剩下的所有评价 ijn ijn ijn 1, 2, , 5j = l 1v 1 0in = ,e m pυ 相加并除以剩下的所有评价个 数。最后将所有类别对应的值进行归一化处理。得到 式（18）： 1 2 5{ , , }w w w w= l （18） 第二步把式（18）中 作为权重值计算综合指标 s。对第 i 个情景来说， w 1 , 1, 2, , ij n i v ij j s w a i m = = × =∑ l （19） 其中 表示第 i 个情景宏观异常 j 的异常类别对应的 权重值， 表示情景 i 中第 j 个宏观异常的异常幅度。 ijv w 4.4 建立用模糊关系矩阵表达测度空间的具体算法 经过前面的预处理，我们已经得到了描述宏观异 常群的两个指标：非空间指标 s 和空间分布指标 c。 设有 n 个评价资料，每个资料有三个分量 s、c 和 j，记第 i 个评价资料为 di=(si,ci,ji)。视 di 为一个信息， si,ci,ji 为信息分量。其中 si,ci 对应着某一个宏观异常群 情景。 为了寻求 s，c，j 三个信息分量之间的输入-输出 关系，得到从 s，c 的离散论域到 j 的离散论域的模糊 关系矩阵 r 的数值，采用三维信息扩散公式对评价资 料 d 构造三维信息矩阵。由于信息分量 ji 的特殊性， 我们不需要再对此分量进行扩散，只需要对其进行归 一化处理，就可以将 ji 分配到监控空间中。这样得到 的样本在每个监控点的信息量，可以完整的嵌入到三 维信息扩散公式当中。 （1）分量 j 归一化处理 取宏观异常群严重度离散论域为 1 2={ , , , }βυ υ υ υl (20) 因为我们的信息分量 ji 比较特殊，它本身就是一 个模糊子集，用扎德记号可表示为 1 2 1 2 + ii ii pp p j β β υ υ = + +l υ (21) 可对 ji 进行信息量总和为 1 的归一化处理，可得 ji 在宏观异常群严重度论域上各个分量上的信息分配。 1 1 1 2 2 2 1 2 1 2 ( ) ( ) ( ) i i i i i i i i i i i i i i i p q p p p p q p p p p q p p p β β β β β υ υ υ ⎧ =⎪ + +⎪ ⎪ =⎪⎪ + +⎨ ⎪ ⎪ ⎪ =⎪ + +⎪⎩ l l m l (22) 我 们 可 以 选 取 一 个 例 子 进 行 简 单 计 算 ， 取 1 2 β={ , , , }={0,1,2, ,10}υ υ υ υl l ，第 3 次评价结果为 3 1 3,1 2 3,2 3,11 3,11{( , ), ( , ), , ( , )} ={(0,0),(1,0),(2,0),(3,0),(4,0),(5,0),(6,0.328), (7,0.552),(8,0.635 0.469),(9, ),(10,0.141)} j p p pυ υ υ= l ija 它是一个模糊子集，用扎德记号可表示为 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 175 1 2 3 1 2 3,1131 32 1 2 11 = + + 0 0 0 0 0 0 0.328 0.552 = + + + + + + + + 0 1 2 3 4 5 6 7 0.635 0.469 0.141 + + 8 9 10 ii i pp p j pp p β βυ υ υ υ υ υ + + = + + l l 隶属度为 0 时可以不表示，则 3 0.328 0.552 0.635 0.469 0.141 = + + + + 6 7 8 9 10 j 根据式(2-50)对其进行归一化处理： 3 1 3 2 3 11 0.469 0.469 0 ( ) =0 0.328+0.552+0.635+ +0.141 0 ( ) =0 0.328+0.552+0.635+ +0.141 0.141 ( ) =0.066 0.3 0.46928+0.552+0.635+ +0.141 q q q υ υ υ ⎧ =⎪ ⎪ ⎪ =⎪ ⎨ ⎪ ⎪ ⎪ =⎪⎩ m （2）构造监控空间 对于 si 和 ci 两个指标，可以用常规信息扩散方法 来处理。首先构造关于 s 和 c 的监控点序列。设非空 间指标 s 的监控点序列为 1 2{ , , , }ku u u u= l 设空间指标 c 的监控空间为 1 2={ , , }ro o o ol ， 以宏观异常群的非空间指标 s 为例，来说明如何 构造相应的离散论域作为监控空间。监控空间通常数 据本身息息相关，具体构造过程如式（23）所示。 1 2 1 2 1 1 , , , } { , , , } min( ) 2 (max( ) min( )) / ( ) 1 0 min( { ) , 1, 2, , i j n i i i js s u u u u uu s u u u s s u u s s s s s s i j n ρ ρ + ≠ ⎧ ⎪ =⎪ ⎪ δ= −⎪ ⎪ = + δ⎨ ⎪ = − δ⎪ ⎪ < δ ≤ − = ⎪ ⎪ =⎩ l l l + （23） 式（23）中，s 表示原始样本数据，u 表示监控点 序列， 表示序列 u 中的第 i 个监控点，iu ρ 表示 u 中 需要构造的监控点个数， 表示监控点序列的步长。 同理可构造宏观异常群的空间密集度 c 的监控空间 o。 uδ （3）计算扩散系数和信息矩阵 三维正态信息扩散估计函数如式（24）所示[8]： 2 22 2 2 ( ) ( )( ) exp[ ] exp[ ] exp[ ] 2 2 2k j j i j ik i u o s c o c zu s h h υ μ − −− = − − − 2 zh (24) 其中， ( , , )i i is c z 是样本点， , ,k j lu o υ 是对应的监控点， , ,s c zh h h 是扩散系数。 将式（22）和式（24）结合，进行改造得到适合 本研究的三维信息扩散函数如式（25）所示： 22 2 2 ( )( ) ( , , ) exp[ ]exp[ ] ( ) 2 2k j l j ik i u o i i i i l s c o cu s s c e q h hυ μ υ −− = − − (25) 其中，扩散系数 hs，hc 可以根据式（26）计算得到， 0.8146( ), 5 0.5690( ), 6 0.4560( ), 7 0.3860( ), 8 0.3362( ), 9 0.2986( ), 10 2.6851( ) / ( 1), 11 b a n b a n b a n h b a n b a n b a n b a n n − =⎧ ⎪ − =⎪ ⎪ − = ⎪ = − =⎨ ⎪ − = ⎪ − =⎪ ⎪ − − ≥⎩ （26） 其中， 1 max( )ii n b x ≤ ≤ = ， 1 min( )ii na x≤ ≤= 。 ( , , ) k j lu o g i i i s c jμ 可以简写为 kjlμ ，接下来要将各样 本对 ( , , )k j lu o υ 的信息扩散量进行叠加，以 表示， kjlq 1 = ( , , k j l n kjl u o i i i i q sυμ = ∑ )c j （27） 为了更清晰地表达三维信息矩阵，我们对其做一 下简单变换处理，用 表示 。 ( )j klq kjlq （4）计算模糊关系矩阵 将原始信息矩阵用式（28）和式（29）进行处理， 可得到空间指标 c（空间密集度）和宏观异常群自身的 非空间指标 s 与宏观异常严重度的模糊关系矩阵 ( ){ }jklr r= 。这个模糊关系矩阵就是得到的测度空间。 ( ) 1 1 max{ }jl kj j k k h q ≤ ≤ ≤ ≤ = l l （28） ( ) ( ) / , 1, 2, ; 1, 2,j jkl kl lr q h j k= = =l （29） 5. 结论和讨论 本文主要基于宏观异常研究进展缓慢，缺乏测度 空间的支撑，很难开展系统性科学研究的现状，提出 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 176 了宏观异常群的概念，研究了如何描述和表达宏观异 常群信息，怎样建立描述宏观异常群的空间综合指标 ——空间密集度和非空间综合指标，如何基于此构建 宏观异常群的测度空间，那就是研制智联网平台，邀 请多位一线地震工作者在线对模拟的宏观异常情景进 行评判，将评判结果作为输出，宏观异常群的空间综 合指标和非空间综合指标作为输入，利用三维信息扩 散技术形成共识性的地震宏观异常测度空间，为实现 地震宏观异常的综合量化提供理论基础。同时，将模 糊量化输入和综合处理的理念引入，为智联网的信息 沟通提供了一个新的手段。 宏观异常种类繁多，异常形态复杂多样。异常形 态的复杂性和多样性是异常形成条件和异常机理复 杂、多样性的反应。例如不同井水位的异常幅度差别 很大，大者可达几十厘米，甚至 1 米以上，小者仅几 毫米。从统一层面上对宏观异常的异常幅度分级划分 是很困难的，但是异常幅度本身也存在一定信息量， 本研究为了将单个异常的异常幅度作为描述宏观异常 的一个属性，从而为描述宏观异常群贡献信息，将其 划分为两个级别。划分方法虽然稍显粗糙，也是我们 度量宏观异常群的综合异常程度的一个必不可少的指 标，在实际的宏观异常核查过程中，专家或者一线地 震工作者可以根据具体异常点过往表现情况，结合自 身经验知识对单个异常的强烈明显程度进行判断。 参考文献 [1]c. f. huang. internet of intelligences in risk analysis for online services, journal of risk analysis and crisis response,1(2):110-117,2011. [2]车用太,鱼金子,刘成龙.地震宏观异常与临震预测成功的 希望[j].国际地震动态,2012(6):138-138. [3]许 敦 煌 . 汶 川 大 地 震 前 宏 观 异 常 的 现 场 调 查 [j]. 地 震,2010,30(2):121-133. [4]张小涛,张永仙,许敦煌.汶川 8.0 级地震前后宏观异常现象 分析[j].地震,2009,29(2):104-117. [5]王蔚丹,黄崇福.建立共识性地震宏观异常测度空间的研究. 大数据时代的风险分析和危机反应--中国灾害防御协会 风险分析专业委员会第六届年会论文集[c].中国,长 沙,2016.11.4-6:908-913. [6]李坤,李秀英.地下水宏观异常与强震临震预报[j]. 华北地 震科学,1997,15(4):57-63. [7]http://api.map.baidu.com/library/geoutils/1.2/docs/symbols/ bmaplib.geoutils.html. [8]黄崇福.自然灾害动态风险分析的一个虚拟案例[j]. 灾害 学,2015(4):1-11. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 166–177 ___________________________________________________________________________________________________________ 177 http://api.map.baidu.com/library/geoutils/1.2/docs/symbols/bmaplib.geoutils.html http://api.map.baidu.com/library/geoutils/1.2/docs/symbols/bmaplib.geoutils.html p1.pdf 共识性的地震宏观异常群测度空间及其构建方法研究 p12.pdf 参考文献 research article discussing the need to manage uncertainty relating to users in road tunnel fire risk assessment panagiotis ntzeremes* , konstantinos kirytopoulos, vrassidas leopoulos school of mechanical engineering, national technical university of athens, 9 iroon polytechniou str., athens 15780, greece 1. introduction although studies indicate the significant growth of maritime and the rise of intermodal transportation during the last decade, they also pinpoint that road transportation continues to hold the largest share in the transportation sector worldwide, having a considerable margin from the rest [1]. in that view, modern roads face the challenge to provide an adequate, intelligent as well as safe road network. therefore, road infrastructure must be formed in order to cope with the arisen challenges. although each part of the infrastructure should fulfill these challenges, special emphasis should be primarily given on its critical elements, such as tunnels [2]. nowadays, tunnels have been an essential part for road networks to successfully meet these challenges. in general, the use of tunnels is strongly linked to the particular operational characteristics of the road network. to this respect, tunnels are divided into two major types. in summary, tunnels in urban areas aim to enable transportation flow in these areas by easing traffic congestion. in doing so, they also contribute to the enhancement of environmental quality of densely populated areas. on the contrary, the significance of tunnels belonging to rural areas and motorways is related to the reduction of travel time and cost commonly by creating shortcuts in mountainous areas [3]. recent studies have shown that the number of tunnels is rising over the last two decades [4]. this progress has been supported by the enhancement of the mining technology, (e.g. improvement of tunnel boring machines), which has made tunnels become an attractive cost-effective engineering solution [5]. as a result, the share of both passenger and freight transport passing through tunnels is continously increasing. as far as the safety issue is concerned, it has been named as the major challenge in the operation of road networks [6]. therefore, each element of the road infrastructure should be characterised by a high safety level. although tunnels do provide roads with numerous advantages, they also include significant risks for the operation of road networks. in particular, the type of accidents that can result in fire are the biggest threats [7]. the reasons are the impact of these accidents not only on the tunnel system itself but also on the rest road network. the devastating experience from fire accidents in the past has been didactic of the severity of such accidents [8]. the increase in the number and the length of tunnels implies that fire accidents will remain a substantial threat for each road network in the future. thus, the need to adopt efficient safety strategies should be of high priority. given that an efficient fire safety strategy is related to the loss reduction rate succeeded in case of a fire accident, users are in the centre of attention when designing fire safety strategies for road tunnels. nowadays, risk assessment has the predominant role for managing fire safety in road tunnels [9]. risk assessment enables the safety analyst to estimate adequately potential losses amongst tunnel users in order to obtain an accurate description for the system’s level of safety. but, capturing and examining human behaviour in fires is a difficult task in every risk assessment process due to the significant uncertainty embeds. road tunnels do not constitute an exception. however, existing risk assessment methods exhibit significant lack in dealing with tunnel users’ uncertainty in a rigorous way. ntzeremes and kirytopoulos [10] study has illustrate that the absence on treating users’ uncertainty can raise serious questions a r t i c l e i n f o article history received 24 february 2020 accepted 09 march 2020 keywords fire risk safety tunnel users uncertainty a b s t r a c t although tunnels provide roads with numerous advantages, they also include significant risks. especially, fire accidents are the biggest threats. nowadays, risk assessment has the predominant role for managing tunnel fire safety and users are in the centre of attention. however, existing methods exhibit lack in treating uncertainty stemming from users. this paper aims to discuss current ways of representing tunnel users’ uncertainty as well as to introduce the importance of distinguishing these representations in the wider context of risk assessment. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: ntzery@mail.ntua.gr; pntzeremes@gmail.com journal of risk analysis and crisis response vol. xx(z); month (year), pp. xx–yy doi: https://doi.org/10.2991/jracr.k.200421.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr in press, uncorrected proof journal of risk analysis and crisis response vol. 10(1); april (2020), pp. 12–18 doi: https://doi.org/10.2991/jracr.k.200507.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0003-1407-5014 http://creativecommons.org/licenses/by-nc/4.0/ mailto:ntzery%40mail.ntua.gr?subject= mailto:pntzeremes%40gmail.com?subject= p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 13 on the credibility of risk assessment studies. based on this study, only a small change, about 30s in the parameter of the beginning of the evacuation process can change the acceptability of the safety level of the tunnel. therefore, the purpose of this study is to discuss current ways of representing tunnel users’ uncertainty as well as to introduce the importance of distinguishing these representations into aleatory and epistemic uncertainty in the wider context of risk assessment for road tunnels. 2. key features of fire safety approach tunnels prove to be safer road sections compared to the rest elements of the road network. there are plenty of reasons to justify this outcome. in brief, tunnels are not affected by dangerous weather conditions because of their enclosed environment, i.e. rainfall or snowfall. additionally, they commonly lack junctions, pedestrians, stopped vehicles, and advertisement signages, which distract driver’s attention [11]. all these elements have been highlighted either as root causes of accidents or as aggravating factors. however, the most important factor that makes tunnels be safer is the driving behaviour of users inside tunnels. relevant studies have indicated that users are more cautious when passing through tunnels [12]. despite the lower accident rate in tunnels, the severity level of potential accidents is much higher. as far as fire accidents is concerned, their rates are even lower in tunnels than in open roads. however, the facts do not allow any complacency on this issue. indicatively, in norway, the country with more than a 1000 road tunnels and with the longest ones, the vehicle fires in tunnels average up to 20 per year between 2008 to 2011 [13]. china, which due to the economic growth is enhancing its road network, has more than 15,000 tunnels with a total length above 16,000 kilometers. nevertheless, 161 serious fire accidents have been recorded in chinese tunnels over the last 15 years [14]. the severity of fires in road tunnels depends on their intrinsic particularities, which are [15]: • the absence of natural light that hinders drivers’ visibility when entering or exiting a tunnel, • the difference in pressure between tunnel portals together with the piston effect developed by the vehicles movement. both of them impact on the movement of the air, and in many cases they hinder the successful operation of ventilation, • the difficulties for emergency services to get close to the fire promptly, and • the unique characteristics of fires in enclosed environments. modelling fires in tunnels is a complicated issue because the behaviour of fire depends on both the attributes of burning objects and the environmental conditions, which affected by the enclosed nature of tunnels [16]. geometrical, construction and operational characteristics of tunnels, which have differences to one another, impact too. for instance, tunnels can be unidirectional or bidirectional, which affect the forced movement of the air. additionally, they have different length or width, different structure, and different traffic conditions (e.g. average volume, permission of dangerous goods, etc.). the aforementioned factors impact on both fire development and evolution. in general, the conditions created from tunnel fires shape a highly unsafe environment for trapped users because they are affected by the high radiation released and the propagated smoke. these conditions can be even worse especially during the first minutes of fire evolution because the emergency ventilation system could not manage to impede the fire plume (i.e. hot gases and smoke) spread upstream of the fire location, creating thus the well-known backlayering phenomenon [17]. as a result of the significant losses caused in the trans-alpine tunnel accidents occurred between 1999 and 2001 in france (mont blanc tunnel), in austria (tauern tunnel), and in switzerland (st. gotthard tunnel), tunnel managers and safety analysts raised serious concerns on the safety of these infrastructures [18]. before these tragedies, safety management in road tunnels was carried out based on each country’s “prescriptive” approach. on a nutshell, each tunnel’s infrastructure and facilities should comply with the prescriptive requirements had been imposed. more details about prescriptive requirements exist in piarc [19] study. however, the adverse effects of the aforementioned accidents uncovered with emphasis the deficiencies of the “prescriptive” approach. indicatively, deficiencies arose regarding the surveillance of tunnels since the absence of a single control room led to significant reaction errors or the absence of escape routes associated with safety shelters in order to head trapped users to the external environment, which led to lot of users trap in these and burnt [20]. the aftermath of these disastrous events was that tunnel systems have become complex infrastructures making safety a complicated task. therefore, fire safety management should fulfill the developed requirements [2]. indeed, tunnels are the most significant structural element of the road network [3]. a quick look on their structure, their facilities, and their equipment features this point of view. indicatively, complex surveillance systems are installed for monitoring each tunnel section. their role is to provide sufficient as well as direct information of an incident/accident, namely its location and its evolution, in order to enable tunnel operators to activate the appropriate response protocols. surveillance systems include either closed circuit television for continuously supervising traffic flows and identifying potential incidents/accidents or include sensors for detecting smoke and temperature from potential fire accidents. other advanced technical installations existed in road tunnels are the ventilation and the fixed firefighting systems. their existence is considered vital because they can control drastically the evolution of a fire event, enabling tunnel users trapped inside the tunnel to successfully evacuate themselves and mitigating damage to tunnel’s structure and facilities. in view of that, the systematic approach of risk assessment has been used in conjunction with the prescriptive approach. risk assessment is the proper tool to assess the safety level of the tunnel since it facilitates the examination of specific accidents as well as the observation of potential residual hazards, while assesses their inherent characteristics [21]. 14 p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 fire risk assessment framework, incorporated in the field of road tunnel safety, consists of five stages forming a repetitive circle (figure 1): (i) establish context. this stage consists of the system definition, and the specification of risk assessment’s goals and criteria. (ii) identify hazards. this stage aims to discover and record the hazards that can cause, directly or combined with others, to critical events, like fires. (iii) calculate the risks. (iv) evaluate the risks based on the risk criteria in order to judge their significance. (v) treat the unacceptable risks. in the end, the process is repeated in order to re-assess the renewed safety level of the tunnel system. understanding the operation of tunnel systems is a substantial step in order to conduct risk assessment. to simplify this process, the safety elements circle has been introduced (figure 2). by grouping the various tunnel system parameters based on source characteristics, five discrete elements are developed. these elements are: (i) the users, i.e. walking speed, behavioural intentions, etc. (ii) the facilities, i.e. emergency exits, mechanical ventilation, etc. (iii) the infrastructure, i.e. tunnel length, slope, etc. (iv) the vehicles, i.e. heavy good vehicles (hgv), bus, etc. (v) the traffic, i.e. volume, etc. the use of risk assessment has introduced also the notion of the holistic approach, which every safety plan should follow. therefore, each standard or additional safety measure, apart from the requirements of the specific discrete element, should also take into account all the potential interdependencies with the others. 3. understanding tunnel users’ uncertainty and link to fire risk assessment 3.1. types, definitions and measures provided that road tunnels are complex systems (refer to section 2), safety analysts commonly deal with problems throughout risk assessment process that require decisions under conditions of uncertainty. in general, two types of uncertainty are featured in risk assessment with which safety analysts come up against [22]. on the one hand is the so-called aleatory uncertainty. this type of uncertainty is variability that stems from diversity and stochasticity. on the other hand, there is the epistemic uncertainty. epistemic uncertainty derives from either the partial ignorance or systematic measurement error or lack of knowledge. this type of uncertainty is the most common in risk analysis [23]. human behaviour is treated to be non-deterministically in nature. in addition, a lack of complete knowledge is also presented when modelling human behaviour. thus, human behaviour is characterised by both types of uncertainty. tunnel users do not constitute an exception. surely, the lack of knowledge and ignorance is related to the definition of a problem, the following methodologies, and the nature of the outcomes. however, due to the complexity of problems, (evaluating tunnel users’ self-evacuation is such a problem), safety analysts unavoidably will need to make decisions under conditions of limited knowledge and resources, thus necessitating increased dependence on the proper treatment of uncertainty [23,24]. given that uncertainty is a decisive factor for the analysis and evaluation of risks, safety analyst should have to both recognise and quantify uncertainty sources throughout the risk assessment process. in doing so, the framework of probability is just following in order to represent it [23]. probability theory is regarded as the better tool for coping with aleatory uncertainty. bearing in mind that this kind of uncertainty is variability stemming from diversity or stochasticity, probabilities do provide an adequate mathematical concept for representing random or uncertain quantities. however, there are different views on what is the best approach for describing uncertain quantities associated with partial ignorance or systematic measurement error or lack of knowledge. subjective probabilities are often used for representing this type of uncertainty. although a subjective estimate of this probability is used, with some additional effort, it can be treated as a random variable bounded using either probability intervals or probability bound analysis or bounds based on figure 1 | the risk assessment process. figure 2 | safety elements circle, adapted from ntzeremes and kirytopoulos [4]. p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 15 evidence theory or percentile ranges [23]. the better the enhancement of knowledge and the awareness is, the better the ranges of the probabilities are updated. tunnel users’ losses are reasonably regarded as the representative parameter that provides the preparedness of the system in confronting fire accidents [25]. therefore, each risk assessment method focuses on estimating the possible losses amongst users and subsequently is interested in mitigating them. mitigating the losses in road tunnels means to minimise as low as reasonably practicable the potential losses amongst the users being involved in the accident. the term “reasonably practicable” means that “the risk must be insignificant in relation to the sacrifice (in terms of money, time or trouble) required to avert it” or else “risks must be averted unless there is a “gross disproportion” between the costs and the benefits of doing so” [26; p. 91]. meanwhile, risk assessment should also be considered as the method for describing, the safety analyst’s together with the rest stakeholders of the system, uncertainties [22]. 3.2. evacuation process on the view of uncertainty each of the elements of the safety circle (figure 2) includes parameters characterised by uncertainty of both types (see section 3.1). although a few studies have dealt with uncertainty of certain parameters (for more details refer to ntzeremes and kirytopoulos [4]), little has been done regarding the users. users consist the most vulnerable factor of the system. they are the first who confront with the fire consequences and in most of the cases without being experienced in such critical events. moreover, they do not have appropriate equipment with them and they do not have the education of other groups on how to react in critical situations, like the members of the rescue teams [4]. furthermore, fire in tunnels has much different behaviour than fire in the open road (i.e. developing of backlayering phenomenon, etc., see section 2). as a result, users trapped in the tunnel have to evacuate themselves immediately after the spark of the fire without any delay in the beginning of the self evacuation process. during their self-evacuation users at the same time receive, recognise and interpret cues that affect their decisions before beginning their self-evacuation process. subsequently, once movement gets started on, relevant cues such as the information received by the control room, the mental state or potential influence amongst users along with the exposure to heat, smoke and toxic gases may influence user’s responses and in turn his/her movement time. all these factors characterised by high uncertainty should be taking into account from the analyst [27]. undoubtedly time is the basic engineering measure of trapped users’ evacuation for rescuing themselves making rational decisions. therefore, risk assessment is primarily focused on calculating the time it takes for trapped users to evacuate the tunnel walking away from the fire environment and heading towards a safe place. a basic representation of the fire emergency timeline that depicts the critical human responses or behaviours that impact and contribute to the evacuation process is illustrated in figure 3. the majority of the evacuation models are used in order to calculate the time needed for trapped users to evacuate the tunnel walking away from the fire environment and heading toward a safe place as the emergency exit doors or the tunnel portals (it depends from both the location of the fire and the user) as soon as possible. an important prerequisite in this direction is the establishment of two basic time parameters, the available safety egress time (aset) and the required safety egress time (rset). aset is defined as the time which is actually available for trapped users from fire sparking and the time point at which conditions become inadequate for human life, because of the high rates of pollutant concentration and radiation. on the other hand, rset refers to the time that trapped users actually need for a successful outcome of their self-evacuation. the aim of the safety analysts is to achieve the aset to be greater than rset [28]. to do so, they have to translate the stages of users’ evacuation process (see figure 3) into time required to perform. along with the stages of the evacuation process, substantial information in order to comprehend better users’ behaviour in fires can be found in the relevant theoretical framework. these theories try to interpret the judgements of the actors in a critical situation. fridolf et al. [20] study summarises the most commonly figure 3 | evacuation process, adapted from ntzeremes [27]. 16 p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 used theoretical frameworks. all of them, feature emphatically the inherent variability derived from users during their self-evacuation process. these are: • the behaviour sequence model, which considers that actions of users consist of four distinct phases, namely: receive cues, interpret cues, prepare actions, and act. • the role–rule model, which considers that each user acts following the rules associated with his/her professional status, • the affiliation model, which considers that a user moves towards familiar to him/her places, and • the social influence model, which considers that the behaviour of the crowd influences the evacuation process of an individual user. however, fire risk assessment seems to neglect the presence of uncertainty. most of the methods as well as evacuation models make various assumptions about the behaviour of trapped users during their evacuation. these approaches although simplify the complicated nature of the problem, they also hide the variability existed in human behaviour. additionally, the deterministic approach of each country-specific regulatory requirements can cause similar consequences [21]. thus, the uncertainty included in users can lead to a considerable uncertainty of the whole safety approach. hence, it is important to design models taking into account potential information or clues from both real accidents and studies about human behaviour in fire evacuation and adapting them to the models, accordingly. if so, the uncertainties regarding the overall level of tunnel safety could be diminished. 3.3. critical analysis and suggestions the need for predicting how users would react just in case a fire spark in a tunnel has emerged intensely over the last two decades (refer to section 3.1). post-accident reports were the most valuable sources of data [29], which indicated that many tunnel users did not behave appropriately during their evacuation. for instance, some remained in their cars expecting emergency services to rescue them or some others considered shelters of the tunnel as fireproof safe places and stayed inside these facilities [8]. consequently, a significant number of projects and fullor small-scale experiments has been interested in examining evacuation behavioural intensions of tunnel users. reviewing the relevant to the subject of users’ self-evacuation behaviour literature, one can found a significant number of valuable contributions on it during the last two decades, mainly after the trans-alpine accidents. these studies can be utilised by safety analysts in order to deal with existing uncertainties. ultimately, their contributions could be valuable in the subjective estimation of aleatory probabilities as well as in the minimisation of the epistemic uncertainty. the inherent variability derived from users during their self evacuation process is emphatically featured considering the various theoretical approaches having developed (refer to section 3.2). moreover, various experiments have been highlighted by different approach each time. for example, users’ behaviour in the upgrading of existing tunnels (uptun) project is interpreted through a mixed theoretical approach synthesised by both the behaviour sequence and the role–rule model. detailed information can be found in papaioannou and georgiou [30] study. in addition, the behavioural sequence approach combined with the affiliation model are used in seike et al. [31] study. in this particular case, one of the aims was to measure the movement speed of users in a japanese road tunnel filled with smoke. in the studies of kinateder et al. [32] social influence approach is the instrument in order to examine users’ evacuation in road tunnels experiment with virtual fires. furthermore, in nilsson et al. [33] experiment carried out in sweden, the importance of social influence is examined for the success of the self-evacuation process. ronchi et al. [34] study examined in a virtual reality small scale experiment the factors that impact on the choice of evacuation strategy of users by taking into account both the affiliation approach (distance to familiar safe places) and social influence. as mentioned in section 3.2, time is the basic engineering measure of trapped-users’ evacuation for rescuing themselves. however, fire risk assessment seems to neglect the presence of uncertainty when this measure is calculated. piarc’s [19] reports provide some clues about evacuation behaviour and walking speeds. in addition to these reports, some methods provide specific walking speeds of the trapped-users deterministically, and in relation to the developed tunnel environment [35], although in a qualitative correlation in between. fridolf et al. [36] study is the most recent report on this issue. although provisions enable safety analysts in represent users’ evacuation, their deterministic approach can hide or disregard several aspects of the natural variability of users’ behaviours. besides, the majority of the methods either do not give any information on this issue or provide oversimplified assumptions on the behaviour of users. ntzeremes and kirytopoulos [4] refer on the deficiencies of the european methods in addressing this issue. consequently, the effectiveness of fire risk assessment can be drastically limited. last but not least, studies confirm that the level of education of users can determine significantly their actions in such cases. for instance, the study conducted in the benelux tunnel illustrated that users who did not have any knowledge about the safety facilities existed in the tunnel were far away from reacting according to the expected norms. in particular, some users (i.e. motorists) did not recognise the criticality of the event heading on the fire location [37]. similar conclusion was drawn in gandit et al. [38] study, performed in a tunnel located in the french rhone-alpes region, where evacuation behaviours of participants proved to be far away from the expected ones. the acteurs project examined how awareness of users combined with successful communication channels can impact on the evacuation process decreasing the wrong behaviours [39]. 4. concluding remarks bearing in mind that tunnels are critical elements of the road network, safety is recognised a major challenge for their operation. although the official accident rates showcase that tunnels are safer road sections compared to the rest network and despite their numerous advantages, yet they include significant risks for the road network. in particular, the accidents that can result in fire are the biggest threats. the increase in the number and the length of tunnels implies that fire accidents will remain a substantial threat for p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 17 each road network in the future. thus, the need to design efficient fire safety strategies should be of high priority. the ultimate goal of an efficient safety strategy in the event of a fire accident in a tunnel is to minimise as low as reasonably practicable the potential losses amongst the users being involved in the accident. therefore, by employing an adequate risk assessment method, risk analysts focus on estimating the possible losses amongst users and subsequently are interested in mitigating these losses. given that tunnels are complex systems, safety analysts commonly deal with problems throughout risk assessment process that require decisions under conditions of uncertainty, both aleatory and epistemic. analysing users’ behaviour in a fire accident and especially their self-evacuation process for estimating tunnel’s level of safety constitutes such case. capturing and examining human behaviour is a difficult task in every risk assessment process due to the significant uncertainty embeds. road tunnel users do not constitute an exception. however, fire risk assessment in tunnels seems to neglect the presence of users’ uncertainty. thus, the uncertainty is increased that subsequently leads to a considerable uncertainty of the whole safety approach. most of the methods as well as evacuation models make various assumptions about the behaviour of trapped users during their evacuation. these approaches although simplify the complicated nature of the problem, they also hide the variability existed in users’ behaviour, concealing thus the level of safety of the tunnel. as a result, the decision of the analysts and tunnel managers regarding the need for receiving or not additional safety measures can be misled. similar results can cause the deterministic approach commonly used by each country’s specific regulatory requirements. hence, it is important to use risk assessment methods and design evacuation models by taking into account potential information or clues from both real accidents and studies about human behaviour in fire evacuation. in particular, time is the basic engineering measure for estimating users’ evacuation. therefore, risk assessment is primarily focused on calculating the time it takes for trapped users to evacuate the tunnel walking away from the fire environment and heading toward a safe place. the literature indicates that during their self-evacuation users at the same time receive, recognise and interpret cues that affect their decisions before beginning their self-evacuation process. subsequently, once movement gets started on, relevant cues such as the information received by the control room, the mental state or potential influence amongst users along with the exposure to heat, smoke and toxic gases may influence user’s responses and in turn his/her movement time. all these factors characterised by high uncertainty, which should be taking into account from the analyst. critical analysis of the literature illustrated the various sources of uncertainty embedded in human behaviour during evacuation process. in doing so, it aims to discuss current ways of representing users’ uncertainty as well as to introduce the importance of distinguishing these representations into aleatory and epistemic uncertainty in the wider context of fire risk assessment for road tunnels. thus, safety analysts and others relevant to fire risk assessment on road tunnels can be motivated to modify the current framework. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments we would like to express our gratitude to the editors and the two anonymous reviewers for both their time and the valuable comments provided that helped the improvement of this paper. this research has been co-financed by the european union and greek national funds through the operational program competitiveness, entrepreneurship and innovation, under the call research – create – innovate (project code: t1edk-02374). references [1] rodrigue jp. the geography of transport systems. 4th ed. new york: routledge; 2017. [2] piarc. road tunnels: complex underground road networks. paris: world road association, technical committee 3.3, road tunnel operation; 2016, p. 80. [3] maidl b. thewes m, maidl u. handbook of tunnel engineering, vols. i and ii. london: wiley; 2014, p. 940. [4] ntzeremes p, kirytopoulos k. evaluating the role of risk assessment for road tunnel fire safety: a comparative review within the eu. j traffic transport eng 2019;6:282–96. [5] kaliampakos d, benardos a, mavrikos a. a review on the economics of underground space utilization. tunnell undergr space technol 2016;55:236–44. [6] piarc. strategic plan 2016-2019. paris: world road association; 2016. [7] ingason h. new challenges in tunnel fire safety. fire technol. 2016;52:1445–7. [8] voeltzel a, dix a. a comparative analysis of the mont blanc, tauern and st. gotthard tunnel fires. routes/roads mag piarc virtual libr 2004;324:18–34. [9] european commission. study on the implementation and effects of directive 2004/54/ec on minimum safety requirements for road tunnels in the trans-european road network. brussels: european commission directorate general for mobility and transport; 2015. [10] ntzeremes p, kirytopoulos k. a stochastic-based evacuation model for risk assessment in road tunnel fire accidents and the importance of educating users. in: haugen s, barros a, van gulijk c, kongsvik t, vinnem je, editors. proceedings of the 28th european safety and reliability conference (esrel 2018). trondheim, norway: crc press, european safety and reliability association; 2018, pp. 2185–91. [11] ntzeremes p, kirytopoulos k, leopoulos v. developing a riskbased method for predicting the severity of fire accidents in road tunnels. in: huang c, nivolianitou zs, editors. risk analysis based on data and crisis response beyond knowledge. proceedings of the 7th international conference on risk analysis and crisis response (racr). athens, greece: crc press; 2019, pp. 518–25. [12] zeeri mo, kirytopoulos k, ntzeremes p. exploring road tunnel drivers’ behavior: the case of greece. in: beer m, zio e, editors. proceedings of the 29th european safety and reliability conference (esrel). hannover, germany: research publishing, european safety and reliability association; 2019, pp. 253–60. https://doi.org/10.1016/j.jtte.2018.10.008 https://doi.org/10.1016/j.jtte.2018.10.008 https://doi.org/10.1016/j.jtte.2018.10.008 https://doi.org/10.1016/j.tust.2015.10.022 https://doi.org/10.1016/j.tust.2015.10.022 https://doi.org/10.1016/j.tust.2015.10.022 https://doi.org/10.1007/s10694-016-0607-1 https://doi.org/10.1007/s10694-016-0607-1 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9781351174664-274 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.1201/9780429286346-74 https://doi.org/10.3850/978-981-11-2724-3_0654-cd https://doi.org/10.3850/978-981-11-2724-3_0654-cd https://doi.org/10.3850/978-981-11-2724-3_0654-cd https://doi.org/10.3850/978-981-11-2724-3_0654-cd https://doi.org/10.3850/978-981-11-2724-3_0654-cd 18 p. ntzeremes et al. / journal of risk analysis and crisis response 10(1) 12–18 [13] nævestad to, meyer s. a survey of vehicle fires in norwegian road tunnels 2008–2011. tunnell undergr space technol 2014;41:104–12. [14] ren r, zhou h, hu z, he s, wang x. statistical analysis of fire accidents in chinese highway tunnels 2000–2016. tunnell undergr space technol 2019;83:452–60. [15] ntzeremes p, kirytopoulos k, benekos i. exploring the effect of national policies on the safety level of tunnels that belong to the trans-european road network: a comparative analysis. int j crit infrastr 2018;14:40–58. [16] ingason h, li yz, lonnermark a. tunnel fire dynamics. london: springer; 2014. [17] stec a, hull r. fire toxicity. 1st ed. oxford: woodhead publishing limited; 2010. [18] beard a, carvel r. road tunnel fire safety. 2nd ed. london: thomas telford; 2012. [19] piarc. fire and smoke control in road tunnels. paris: world road association; 1999. [20] fridolf k, nilsson d, frantzich h. fire evacuation in underground transportation systems: a review of accidents and empirical research. fire technol 2013;49:451–75. [21] ntzeremes p, kirytopoulos k. applying a stochastic-based approach for developing a quantitative risk assessment method on the fire safety of underground road tunnels. tunnell undergr space technol 2018;81:619–31. [22] aven t. on the need for restricting the probabilistic analysis in risk assessments to variability. risk anal 2010;30:354–60. [23] ayyub mb, klir jg. uncertainty modeling and analysis in engineering and the sciences. 1st ed. new york: chapman & hall/ crc press; 2006. [24] aven t, renn o. on risk defined as an event where the outcome is uncertain. j risk res 2009;12:1–11. [25] piarc. risk analysis for road tunnels. paris: world road assosiation, technical committee 3.3, road tunnel operation; 2008. [26] ale bjm, hartford, dnd, slater d. alarp and cba all in the same game. saf sci 2015;76:90–100. [27] ntzeremes p. developing a quantitative frire risk assessment method for critical infrastructures: the case of road tunnels. phd thesis. athens; 2019. [28] kinateder mt, kuligowski ed, reneke pa, peacock rd. risk perception in fire evacuation behaviour revisited: definitions, related concepts and empirical evidence. fire sci rev 2015;4:1–26. [29] aadt. task force for technical investigation of the march 1999 fire in the mont blanc vehicular tunnel. paris: ministry of equipment, transportation and housing; 1999. [30] papaioannou p, georgiou g. human behaviour in tunnel accidents and incidents: end users, operators and response teams. report of european project uptun, no. grd1-2001-40739. brussels: eu; 2003. [31] seike m, kawabata n, hasegawa m. experiments of evacuation speed in smoke-filled tunnel. tunnell undergr space technol 2016;53:61–7. [32] kinateder m, ronchi e, gromer d, müller m, jost m, nehfischer m, et al. social influence on route choice in a virtual reality tunnel fire. transport res f 2014;26:116–25. [33] nilsson d, johansson m, frantzich h. evacuation experiment in a road tunnel: a study of human behaviour and technical installations. fire saf j 2009;44:458–68. [34] ronchi e, nilsson d, kojić s, eriksson j, lovreglio r, modig h, et al. a virtual reality experiment on flashing lights at emergency exit portals for road tunnel evacuation. fire technol 2016;52:623–47. [35] piarc. current practice for risk evaluation for road tunnels. paris: world road association, technical committee 3.3, road tunnel operation; 2013. [36] fridolf k, ronchi e, nilsson d, frantzich h. the representation of evacuation movement in smoke-filled underground transportation systems. tunnell undergr space technol 2019;90: 28–41. [37] boer lc. behaviour by motorists on evacuation of a tunnel. report tm-02-c034. soesterberg, the netherlands: tno human factors; 2002. [38] gandit m, kouabenan dr, caroly s. road-tunnel fires: risk perception and management strategies among users. saf sci 2009;47:105–14. [39] noizet as, dedale a, richard f. acteurs: improving understanding of road tunnel users with view to enhancing safety. in: proceedings of the 5th international conference on safety in road and rail tunnels. marseille, france; 2003. https://doi.org/10.1016/j.tust.2013.12.001 https://doi.org/10.1016/j.tust.2013.12.001 https://doi.org/10.1016/j.tust.2013.12.001 https://doi.org/10.1016/j.tust.2018.10.008 https://doi.org/10.1016/j.tust.2018.10.008 https://doi.org/10.1016/j.tust.2018.10.008 https://doi.org/10.1504/ijcis.2018.090650 https://doi.org/10.1504/ijcis.2018.090650 https://doi.org/10.1504/ijcis.2018.090650 https://doi.org/10.1504/ijcis.2018.090650 https://doi.org/10.1007/s10694-011-0217-x https://doi.org/10.1007/s10694-011-0217-x https://doi.org/10.1007/s10694-011-0217-x https://doi.org/10.1016/j.tust.2018.08.020 https://doi.org/10.1016/j.tust.2018.08.020 https://doi.org/10.1016/j.tust.2018.08.020 https://doi.org/10.1016/j.tust.2018.08.020 https://doi.org/10.1111/j.1539-6924.2009.01314.x https://doi.org/10.1111/j.1539-6924.2009.01314.x https://doi.org/10.1080/13669870802488883 https://doi.org/10.1080/13669870802488883 https://doi.org/10.1016/j.ssci.2015.02.012 https://doi.org/10.1016/j.ssci.2015.02.012 https://doi.org/10.1186/s40038-014-0005-z%0d https://doi.org/10.1186/s40038-014-0005-z%0d https://doi.org/10.1186/s40038-014-0005-z%0d https://doi.org/10.1016/j.tust.2016.01.003 https://doi.org/10.1016/j.tust.2016.01.003 https://doi.org/10.1016/j.tust.2016.01.003 https://doi.org/10.1016/j.trf.2014.06.003 https://doi.org/10.1016/j.trf.2014.06.003 https://doi.org/10.1016/j.trf.2014.06.003 https://doi.org/10.1016/j.firesaf.2008.09.009 https://doi.org/10.1016/j.firesaf.2008.09.009 https://doi.org/10.1016/j.firesaf.2008.09.009 https://doi.org/10.1007/s10694-015-0462-5 https://doi.org/10.1007/s10694-015-0462-5 https://doi.org/10.1007/s10694-015-0462-5 https://doi.org/10.1007/s10694-015-0462-5 https://doi.org/10.1016/j.tust.2019.04.016 https://doi.org/10.1016/j.tust.2019.04.016 https://doi.org/10.1016/j.tust.2019.04.016 https://doi.org/10.1016/j.tust.2019.04.016 https://doi.org/10.1016/j.ssci.2008.01.001 https://doi.org/10.1016/j.ssci.2008.01.001 https://doi.org/10.1016/j.ssci.2008.01.001 microsoft word introduction for volume 2, issue 2 editor's introduction in the middle of 2012, most of the countries in the world are undergoing depression or suffering debt crisis. the world we live in is far more interconnected than it was in the past. highly interconnected systems are vulnerable to catastrophic failures. it implies that, using traditional technology, it is very difficult to capture today's fast-changing risks. obviously, human beings need more powerful tools to analyze the risks and manage the causes so that we could prevent the outbreak of the crisis. in the present issue of journal of risk analysis and crisis response (jracr), volume 2, issue 2 (2012), such efforts have been vividly in papers. swain’s paper and tichý’s paper are the most representatives. this issue contains 8 papers. the former 4 contributions are written in english and others in chinese with english abstracts. the papers can be divided into four categories: terrorist attack, natural disaster, methodology and credit rating. in the first category, “explanation of risk and uncertainty in news coverage of an anthrax attack” by swain, analyzed 833 u.s. news stories about the 2001 anthrax attacks, to examine explanations of risk and uncertainty. findings suggest that vague advice, sensational language, anonymous sourcing, premature coverage of hoaxes and false alarms, and speculation may contribute to public misunderstandings about the risks involved. news coverage of a bioterrorism threat can amplify the perception of indeterminate risk. providing more explanatory content can help citizens frame risks rationally and put them into context. the three papers in the natural disaster category offer tools for decision-making. the first paper “urban seismomorphoses seismic vulnerabilities, an embarrassing legacy” by cartier, vallette and mediene, suggests a methodological pattern to identify “urban seismic patrimonial strategies”, crossing urban historical heritage with public investment and patrimonial care permit to situate natural hazards mitigation in urban policy. the second paper “the meteorological disaster risk assessment based on the diffusion mechanism” by guo, proposes the concept of diffusion of meteorological disaster risk. the result from a risk diffusion network is useful for government and relevant departments to make the decisions in dealing with meteorological disaster risk. the paper “the nature event's probabilistic risk assessment of nuclear power plants with improvements” by chen, yu, zhang and chai, introduces the principle of risk assessment of the nature disaster to study risk of nuclear plants with hazard factor, vulnerability analysis, loss risk and risk zone map in nature disaster. there also are three papers in the methodology category. the first paper “entity risk mechanics”, by tichý, is undoubtedly an important explore for innovation of risk theory. concepts of structural mechanics and structural reliability, common in design of constructed facilities, could upgrade the risk theory. the second paper “risk analysis techniques in construction engineering projects” by baloi, reviews and discusses some of the risk analysis techniques available to tackle different types of uncertainty inherent in construction projects. these include probability, certainty factors, evidence theory and fuzzy logic. it argues that the quality of risk analysis depends largely upon the appropriateness of the approach. strengths and weaknesses of each technique are presented and journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 79-80 published by atlantis press copyright: the authors 79 discussed. the paper “a risk-benefit analysis model for project investment based on the normal distribution” by chen, presents the solutions to apply normal distribution to the evaluation of continuous random variable, and to optimize the random variables by adopting optimization principles. the last category includes one paper “a credit rating model for enterprises based on projection pursuit and k-means clustering algorithm” by zhang and zhou, proposes a new credit rating model for enterprises based on projection pursuit and k-means clustering algorithm. in this model, the initial cluster centers are determined by the combination of projection pursuit and kernel density estimation method, which can reduce iterative times, increase the convergence speed and improve the clustering effect of k-means clustering algorithm. taking the high-tech listed companies in china as samples, it is proved that the model proposed by this paper is feasible and effective. i am grateful to the referees for their kind support and help. thanks also go to the authors for all their submissions. editor-in-chief chongfu huang professor, beijing normal university email:hchongfu@gmail.com published by atlantis press copyright: the authors 80 research article investigation of the risk awareness of wearing masks by the public during covid-19 according to the health belief questionnaires of shanxi province residents hou ruyi1, duan tingyu1, sui chunying2, wu yibo2,3,*, wang xiujun1, wang yujie1, sun yu1 1department of humanities, arts and communication, changzhi medical college, shanxi 046000, china 2key research base of philosophy and social sciences in shaanxi province, health culture research center of shaanxi, xi’an 712046, china 3department of medicine, health science center, peking university, beijing 100191, china 1. literature review novel coronavirus (covid-19), which appeared in 2019 and spread around the world, is a newly discovered coronavirus strain. the pneumonia caused by it was confirmed to be an acute respiratory infectious disease [1]. as a newly discovered acute respiratory infectious disease. it poses a great threat to public health. covid-19 prevention is necessary due to its high speed of transmission and the lack of therapeutic means. according to “covid-19 diagnosis and treatment plan (trial version 7)” [2], respiratory droplets and close contact are the main transmission routes of covid-19, and aerosol transmission may exist when exposed to high concentration of covid-19 contained aerosol in a relatively closed environment for a long time. in today’s epidemic, wearing masks can effectively prevent the spread of the virus. based on the current situation, on may 26, 2020, disease prevention and control bureau issued notice on the issuance of guidelines for the public to wear masks scientifically, it divides the population into the general public, people in specific places, key people, and occupationally exposed people, etc., and put forward scientific advice to wear masks under different places. although a great deal of research has been conducted on risk perception, there is still no consensus on the dimensions and measures of risk perception: is risk strictly cognitive or emotional; risk may differ before and after preventive actions (such as wearing a mask) [3]. this study believes that risk is a strict cognitive process, which is measured in two dimensions: perceived susceptibility, or the probability of the outcome, and perceived severity, or the size of the outcome. studies on wearing masks during the epidemic also include the protection of masks against covid-19 susceptibility and severity. in order to make the model more complete, based on the health belief model, this study supplemented the perceived benefit, perceived barriers and self-efficacy. the behavior of wearing masks is a kind of health behavior, which is generally studied through health belief model. the model has been relatively mature. health belief model can change people’s behavior by interfering with people’s attitude and other psychological activities. the structural equation model can deal with the relationship between multiple causes and multiple results at the same time. in this study, it can effectively deal with the relationship between multiple independent variables on the public behavior of wearing masks, and analyze the correlation and quantitative relationship between potential variables. in the context of covid-19, structural equation model plays an important role in a r t i c l e i n f o article history received 17 august 2020 accepted 07 january 2021 keywords risk covid-19 mask wearing behavior health belief model diffusion of innovation theory structural equation model influencing factors a b s t r a c t when some parts of the world are still debating whether the public wearing masks will help prevent and control the covid-19. we have used questionnaires to investigate the wearing of masks and its influencing factors among residents in 11 cities in shanxi province. the questionnaire was designed based on the health belief model. we processed the collected data using the structural equation method. the results we got are as follows: (1) the average score of perceived severity is 3.14, indicating that the respondents had a strong sense of risk; (2) the average score of perceived benefits was 4.00, indicating that respondents thought that wearing masks can prevent covid-19; (3) the average score of perceived barriers was 2.42, indicating that the respondents believed that there were fewer barriers against wearing masks during the epidemic; (4) the average score of perceived susceptibility was 3.23, indicating that respondents believed that they had a strong risk awareness of covid-19 infection; (5) the average score of self-efficacy was 4.00, indicating that respondents had a strong belief in wearing masks correctly. the structural equation model shows that self-efficacy (l = 0.40) and perceived susceptibility (l = −0.15) had direct effects on the behavior of wearing masks. a wide range of mediating effects exists in the model. perceived severity has an effect on the behavior of wearing masks through perceived susceptibility (l = 0.41), perceived benefits through self-efficacy (l = 0.68) and perceived susceptibility (l = −0.35), and perceived barriers through self-efficacy (l = −0.28). people are in the early adopter phase. each dimension has a direct or indirect impact on the wearing of masks during the covid-19 epidemic. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: bjmuwuyibo@outlook.com journal of risk analysis and crisis response vol. 10(4); december (2020), pp. 168–175 doi: https://doi.org/10.2991/jracr.k.210111.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:bjmuwuyibo%40outlook.com https://doi.org/10.2991/jracr.k.210111.001 https://www.atlantis-press.com/journals/jracr h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 169 studying the behavior of the public wearing masks. based on the structural equation model and the health belief model, the results and the conclusion can improve the public’s cognition of wearing masks, so as to promote the change of public behavior. due to the lack of relevant studies on the behavior of wearing masks during covid-19, this study is based on these two models to conduct investigation on the behavior of the public wearing masks, hoping to promote the change of the behavior of the public wearing masks. 2. research objects and methods 2.1. research objects the residents of shanxi province were taken as the research object. the inclusion criteria are as follows: long-term residence in shanxi; have a certain ability to take care of themselves; can complete the network questionnaire or complete the questionnaire with the help of the investigator; understand each item in the questionnaire; volunteer to participate in this study. exclusion criteria are as follows: questionnaires with repeated ip addresses; with inconsistent logic. incomplete questionnaires. 2.2. research methods 2.2.1. taking samples this study adopts the questionnaire through stratified cluster random sampling, draws a certain number of shanxi residents for investigation shanxi province was divided into 11 prefecture-level cities, taiyuan city, datong city, yangquan city, changzhi city, jincheng city, shuozhou city, jinzhong city, yuncheng city, xinzhou city, linfen city and luliang city, and 1–2 communities were selected for random sampling in each prefecture-level city. 2.2.2. questionnaire design the first part of the questionnaire is the general demographic characteristics of the sample population, including gender, age, current city, whether the permanent residence is an urban or rural area, highest educational level, ethnic group, religious belief, marital status, monthly income, occupation, etc. the second part of the questionnaire is the investigation of the behavioral influence factors based on the health belief model, and set questions on the influence factors of the behavior of wearing masks during the epidemic: a total of 21 items, include five dimensions: severity (three items), perceived benefits (five items), perceived barriers (four items), self-efficacy (five items) and perceived susceptibility (four items). each item is scored by likert 5-level scoring method, which were 1–5 points from low to high. the third part of the questionnaire is based on the diffusion of innovation theory, through this part, we divide the behavior of the public to wear masks during the epidemic into five behavioral stages: innovators (the first to adopt the behavior and to spread and persuade others), early adopters (the earlier group to adopt the behavior and to spread one’s subjective view), early public (adopt the behavior after careful consideration), late public (adopt the behavior later and easily changed by external influence), laggards (resistance or late adoption). 2.2.3. survey methods due to the epidemic, the questionnaires were mainly distributed online. questionnaires were distributed one-on-one through wechat, qq and other social platforms to residents of 11 prefecture level cities in shanxi province. in order to ensure the validity of the questionnaire, the investigator answered the questions raised by the respondents, and a total of 1603 valid questionnaires were recovered in the end. 2.2.4. statistical analysis ibm spss21.0 software (armonk, ny, usa) was used for statistical analysis. the mean value and standard deviation were used to study the central trend 95% of the sample statistics were taken as the confidence interval. the structural equation model was established by using ibm spss amos24.0 for latent variable analysis. chi-square/df, goodness-of-fit index (gfi), adjusted goodnessof-fit index (agfi), comparative fit indexes (cfi) and root mean square of approximate error (rmsea) are used in the structural equation model, take a = 0.05 as the standard of statistical test significance. 2.2.5. quality control before the questionnaires were distributed, we conducted a unified training for the investigators to ensure the standardization of the survey process and improve the reliability and validity of the questionnaire survey. all questionnaires are filled in anonymously, which can reduce the concerns of respondents. after the questionnaires were collected, we checked the logical relationship of the questionnaires, and eliminated questionnaires with inconsistent logic before and after the options and incomplete questionnaires. the survey was reviewed and approved by the ethics committee of changzhi medical college rt2020018, and all the respondents understood the project content and signed the informed consent form. 3. results 3.1. basic information and descriptive statistics a total of 1635 questionnaires were collected, and 1603 were finally confirmed to be the valid questionnaires, with an effective rate of 98.04%. among the valid 1603 questionnaires, the basic information of 10 respondents was randomly selected, as shown in table 1. there are 625 innovators (39.0%), 570 early adopters (35.6%), 249 early public (15.5%), 110 late public (6.9%), and 49 laggards (3.1%) of mask wearing. the scores of innovator, early adopter, early public, late public and laggards are 5 points, 4 points, 3 points, 2 points and 1 point respectively. according to the survey, the mean of the overall score is 4.01 points (3.95, 4.06), which is in the transition 170 h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 stage from early adopter to innovator. univariate factors showed that there were significant statistical differences among different age groups, cities, education level, marital status, smoking status, infection of relatives and friends, etc., in the behavior of wearing masks during the epidemic period (see figure 1 for details). 3.2. health belief theory of the public’s behavior of wearing masks during the epidemic the average score of perceived severity was 3.14 (3.10, 3.18); perceived benefits 4.00 (3.96, 4.03); perceived barriers 2.42 (2.37, 2.46); self-efficacy 4.00 (3.96, 4.04); perceived susceptibility 3.23 (2.42, 4.04). the responses of the above 10 respondents to each item are shown in table 2. table 3 shows the correlation between the factors. 3.3. structural equation model of public behavior when wearing masks during the epidemic structural equation model includes measurement model and structural model. perceived severity, perceived benefits, perceived barriers, self-efficacy and perceived susceptibility were included in the model. the factor load of 21 items in the scale was all higher than 0.5, the combined reliability of the five dimensions of the scale was all higher than 0.6, and the average variance extract of the five dimensions was all higher than 0.5. therefore, the scale had good validity. the fitting index of measurement model is as follows: chi-square/df = 4.755 < 5, gfi = 0.947 & gt; 0.9, agfi = 0.933 & gt; 0.9, cfi = 0.966 > 0.9, rmsea = 0.048 < 0.08, all within the acceptable range, indicating that the model and the survey data are compatible as a whole and the model is relatively scientific and effective. in the structural equation model, self-efficacy and perceived susceptibility have a direct impact on behavior, with path coefficients of 0.40 and −0.15 respectively, indicating that self efficacy positively affects behavior, while perceived susceptibility negatively affects behavior. in other words, the stronger self-efficacy is, the more compliance is with the behavior of wearing masks during the epidemic. the lower the susceptibility, the stronger the compliance with wearing masks during the epidemic. perceived severity, perceived benefits and perceived barriers play a mediating role in the public behavior of wearing masks. perceived severity has an effect on the public’s behavior of wearing masks through perceived susceptibility, with a path coefficient of 0.41. perceived benefits has an effect on the public behavior of wearing masks through self-efficacy and perceived susceptibility, and the path coefficient is 0.68 and −0.35 respectively. perceived barriers has an effect on the behavior of the public wearing masks through self-efficacy, with a path coefficient of −0.28 (see table 4 and figure 2 for details). 4. discussion 4.1. analysis of the current situation of the public wearing masks during the epidemic the five stages of innovators, early adopters, early public, late public, and laggards of the public wearing masks during the epidemic were scored 5 points, 4 points, 3 points, 2 points, and 1 point. the overall average score of 4.01 (3.95, 4.06). during the epidemic, the public are early adopters of wearing masks. this is consistent with the research results of sim et al. [4], zai et al. [5] and zhan et al. [6] on the behavior of public wearing masks [6]. there were 625 innovators (39.0%), 570 early adopters (35.6%), 249 early public (15.5%), 110 late public (6.9%) and 49 laggards (3.1%). therefore, during the epidemic, the innovative act of wearing masks by the table 1 | basic information of 10 randomly selected respondents serial number q1 q2 q3 q4 q5 q6 q7 q8 q9 q10 q11 q12 q13 q14 1 2 1 3 2 5 5 1 1 1 4 2 3 3 3 2 2 4 1 1 5 4 2 2 3 5 2 2 3 3 3 2 6 6 1 1 7 1 4 1 5 1 1 1 1 4 1 5 11 1 4 2 1 2 4 6 1 2 2 1 5 2 7 5 2 6 6 1 2 5 4 2 1 3 3 6 1 5 9 1 6 3 2 2 3 3 2 2 3 1 7 1 2 4 2 7 1 2 1 3 8 1 2 3 1 8 1 3 8 2 5 1 2 2 4 6 1 2 3 3 9 2 2 2 1 7 1 2 1 4 8 1 2 3 3 10 1 1 7 1 4 1 2 1 3 7 1 2 3 3 note: q1 your gender: 1 = male, 2 = female; q2 your age: 1 = under 18 years old, 2 =18–25 years old, 3 = 26–30 years old, 4 = 31–40 years old, 5 = 41–50 years old, 6 = 51–60 years old, 7 = over 60 years old; q3 your city: 1: taiyuan, 2: datong, 3: shuozhou, 4: xinzhou, 5: yangquan, 6: lvliang, 7: jinzhong, 8: changzhi, 9: jincheng, 10: linfen, 11: yuncheng; q4 your usual place of residence: 1: urban, 2: rural; q5 your highest education: 1: no formal education, 2: primary school, 3: junior high school, 4: senior high school, 5: technical secondary school, 6: junior college, 7: university undergraduate, 8: master’s degree, 9: doctoral degree; q6 your nationalities: 1: han nationality, 2: hui nationality, 3: zhuang nationality, 4: tibetan nationality, 5: miao nationality, 6: manchu nationality, 7: others. q7: religious belief: 1: have, 2: have nothing; q8 your marital status: 1: unmarried, 2: married, 3: divorced, 4: widowed; q9 per capita monthly income of your family: 1: no income: less than 2.00 yuan, 3: 1000-2999 yuan, 4: 3000-4999 yuan, 5: 5000 and above; q10 what is your current or retired career: 1: office/business unit personnel, 2: professional and technical personnel, 3: business unit personnel, 4: business, service personnel, 5: agriculture, forestry, animal husbandry and fishery, water conservation, production personnel, 6: manufacturing, transportation equipment operators and related personnel, 7: soldiers, 8: student: 9 other employed and unemployed; q11 so far, are there any confirmed cases in your current city?: 1: there are 2: there are no 3: it is not clear; q12 have any of your family or friends infected with novel coronavirus in this epidemic?: 1: there are 2: there are no 3: it is not clear; q13 do you smoke?: 1: smoking 2: have quit smoking 3: never smoking; q14 do you drink alcohol?: 1: drinking 2: abstaining from alcohol 3; never drink. h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 171 figure 1 | comparison of individual characteristics with behaviors of wearing masks during the epidemic. 172 h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 public was carried out in depth, during which the government played an important role with the high frequency and multi-form publicity of personal protection measures. demographic characteristics can influence the behavior of the public to wear masks during the epidemic. through one-way analysis of variance, eight factors were obtained, including age, education level, religion, marital status, occupation, whether there were confirmed cases in the city of the resident, whether any of the family or friends infected with covid-19 and whether they smoked. these factors had a significant impact on the behavior of the public wearing masks during the epidemic period. in terms of age distribution, residents aged 18–25 have an earlier stage of innovation behavior on the whole, and young people can table 2 | the responses of the 10 respondents to each item serial number 1 2 3 4 5 6 7 8 9 10 the covid-19 epidemic will last a long time. 5 5 4 4 4 3 3 3 3 2 covid-19 affects my relationship with my family or friends. 4 4 3 4 4 2 4 1 2 1 if i have covid-19, i will probably lose my life. 3 4 4 4 4 2 3 3 2 1 wearing masks can reduce my chance of suffering from covid-19. 5 5 4 4 5 5 4 4 4 5 wearing masks can protect my family and reduce their risk of infection. 5 4 4 4 5 4 4 1 4 4 wearing masks is an expression of responsibility to my family. 5 5 4 4 4 5 4 1 4 4 wearing masks can make me at ease about my health. 4 4 3 4 4 4 4 4 4 4 i can take advantage of wearing masks. 3 4 3 4 4 5 4 4 3 4 i think the cost of buying masks is high. 3 4 4 3 3 1 2 2 3 2 it is uncertain whether the mask wearing behavior is effective to avoid the infection. 2 3 3 2 1 1 2 3 2 1 i often forget to wear a mask. 3 4 2 2 3 1 2 1 2 1 the process of wearing a mask will make me feel uncomfortable. 4 5 4 2 4 1 2 3 4 1 i can pay attention and frequently check information on how to prevent covid-19. 4 5 2 4 5 5 4 1 4 4 i’m sure i can wear a mask actively. 3 4 3 4 4 5 4 3 4 4 i am sure i can wear a mask every time i go out. 4 5 4 4 5 5 4 4 4 4 if i feel uncomfortable, i will go to the hospital in time. 5 4 4 4 4 5 4 5 4 4 even i am not sick, i still stick to wearing a mask. 4 5 3 4 4 5 3 1 4 4 i have high-risk factors for contracting covid-19 5 4 3 2 5 3 1 3 3 2 maybe one day i will be infected with novel coronavirus. 4 4 3 2 4 2 5 2 3 1 i have a high probability of contracting covid-19. 4 5 3 2 5 3 1 3 3 1 before the outbreak is over, i am at high risk of getting covid-19. 4 4 3 2 4 1 1 2 3 2 note: 1 = strongly disagree, 2 = disagree, 3 = not sure, 4 = agree, 5 = strongly agree. table 3 | correlation among factors correlation perceived severity perceived benefit perceived barriers self-efficacy perceived susceptibility perceived severity 1 perceived benefits −0.087** 1 perceived barriers 0.359** −0.513** 1 self-efficacy −0.178** 0.763** −0.589** 1 perceived susceptibility 0.377** −0.328** 0.541** −0.415** 1 **significantly correlated at 0.01 level (bilateral). h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 173 get information faster and more easily to be exposed to information related to covid-19 prevention, so they adopt healthy behaviors to prevent diseases earlier. the overall stage of innovation behavior of residents over 60 years old is relatively backward, which is related to the elderly’s weak ability to accept new information and stubborn and backward thinking. it is inconsistent with the results of a telephone survey conducted by taylor et al. [7]. this may be due to the fact that the survey was conducted in the early stage of the prevalence of covid-19, and that the elderly were slower to obtain protective measures and lower in the level of healthy behavior than figure 2 | structural equation model. note: f1: perceived severity, f2: perceived benefits, f3: perceived barriers, f4: self-efficacy, f5: perceived susceptibility, q7: behavior stage; the path coefficient in the figure is the standardized coefficient, and the path coefficient in the figure is statistically different (p ≤ 0.05). v1–v21: entry. table 4 | parameter estimation in the structural equation model latent variables independent variable a unstd. s.e. t-value p perceived severity v1 0.77 1.00 v2 0.69 1.07 0.05 23.84 <0.01 v3 0.81 1.13 0.05 24.84 <0.01 perceived benefits v4 0.76 1.00 v5 0.84 1.25 0.04 34.33 <0.01 v6 0.78 1.05 0.03 31.38 <0.01 v7 0.81 1.31 0.04 32.73 <0.01 v8 0.68 1.24 0.05 26.86 <0.01 perceived barriers v9 0.79 1.00 v10 0.81 1.07 0.03 32.69 <0.01 v11 0.84 1.15 0.03 34.45 <0.01 v12 0.61 0.85 0.03 24.83 <0.01 self-efficacy v13 0.78 1.00 v14 0.86 1.30 0.04 37.29 <0.01 v15 0.87 1.36 0.04 38.43 <0.01 v16 0.78 1.10 0.03 33.17 <0.01 v17 0.83 1.34 0.04 35.79 <0.01 perceived susceptibility v18 0.84 1.00 v19 0.84 0.99 0.02 40.69 <0.01 v20 0.89 1.06 0.02 45.15 <0.01 v21 0.83 0.96 0.02 40.58 <0.01 the young. however, once the elderly accept this health behavior, it will last for a long time and will not be easily changed. in terms of educational background, residents with the highest educational background in high school and below have a low level of cognition of disease-related knowledge and a weak ability to acquire the latest knowledge, so they have a weak behavioral ability to take proper preventive measures. people with a bachelor’s degree or above have a higher literacy level. they have a strong ability to prevent and control diseases and can take correct measures to prevent covid-19, which is consistent with the research results of kuo et al.’s [8] risk awareness and prevention behavior of avian influenza and lau et  al.’s [9] prevention behavior in the h5n1 epidemic period. in terms of marital status, the married people are more obedient to wearing masks during the epidemic than unmarried, divorced or widowed residents. due to the sense of family responsibility, married people pay more attention to their own health or the health of family members, which is consistent with the research of tang and wong [10]. people who believe that there are confirmed cases in their city have higher levels of wearing masks than those who do not know whether there are confirmed cases in their city. people who have quit smoking or never smoke are more compliant to wearing masks than those who smoke. therefore, people with good health habits are more likely to take effective measures to protect themselves during the epidemic than those with bad health habits. when publicizing the knowledge and protective measures about the virus during the epidemic period, it is necessary to take into account the residents with low education level and long age, take pictures, videos and other health communication forms that do not require high level of cultural knowledge for health communication, expand the spread and coverage of health information, and make more people take health protective measures to effectively prevent diseases. 174 h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 4.2. analysis of the public’s behavioral health beliefs in wearing masks during the epidemic the scores of self-efficacy, perceived benefits, perceived severity, perceived barriers and perceived susceptibility ranged from 1 to 5 in order from low to high. the average score of self-efficacy was 4.00 (3.96, 4.04), which was higher among five stages. residents’ attention to the prevention of covid-19 information, their awareness of going to the hospital voluntarily due to physical discomfort, their correct choice despite of illness and their insistence on wearing a mask have an impact on their self-efficacy. thus, it can be seen that covid-19 information can be effectively released during the epidemic period, and the public should be encouraged to seek medical treatment in time when they are sick. this will have an important impact on the self-efficacy of the mask wearing behavior during the epidemic. the average score of perceived benefits was 4.00 (3.96, 4.03), which was higher among five dimensions. the satisfaction of benefiting from wearing a mask, the certainty that wearing a mask reduces the risk of being infected by covid-19 for oneself and the family, and the sense of responsibility to one’s family and oneself positively affected the perception of benefits. therefore, the effective prevention of infection by wearing masks by the public during the epidemic, and the publicity and popularization of the impact of wearing masks on their own health and the health of their families play a vital role in improving the public’s awareness of the benefits of wearing masks during the epidemic. the average score of perceived severity was 3.14 (3.10, 3.18), which was medium. the average score of perceived susceptibility was 3.23 (2.42, 4.04), with the lowest score. the impact of illness on one’s relationship with family and friends, the public’s perception of the long duration of the epidemic and the threat of the disease to life, in turn, have a positive impact on the perceived severity. the public’s perception of the possibility of future illness, the perception of whether they have disease influencing factors and the probability of their infection with virus will influence the public’s perceived susceptibility. therefore, under the premise of not affecting the normal life of the public, if the government and the media strengthen publicity on the duration of the epidemic, the general susceptibility of the population, and the harm of covid-19, it will help improve the public’s cognition for the extent of severity and susceptibility of the virus. however, it should be noted that the government and the media’s overemphasizing the severity of covid-19 will definitely increase the psychological burden of the public and patients with covid-19 and, subsequently, cause social panic. meanwhile, the government and the media are suggested to achieve precise health communication and adopt different intervention measures for different groups of people. the average score of perceived barriers was 2.42 (2.37, 2.46), which was low. the discomfort of wearing a mask, the aversion to learning to wear a mask correctly, the uncertainty about the effectiveness of wearing a mask, and forgetting to wear a mask due to negligence all have positive effects on the perception of disability. therefore, when promoting the correct way of wearing masks, it is necessary to adopt a popular way as far as possible, so as to improve the public’s enthusiasm for learning how to wear masks correctly and develop good healthy habits. the repeated emphasis in various forms on the usefulness of wearing masks for virus prevention and the adoption of appropriate life tips will affect the public perception of the barrier of wearing masks during the epidemic. 4.3. analysis of influencing factors of public behaviors of wearing masks during the epidemic the structural equation model showed that self-efficacy and perceived susceptibility had significant direct effects on the public wearing masks behavior during covid-19. self-efficacy is the most important factor influencing the public wearing masks behavior during the epidemic, and it has a positive effect on the behavior of wearing masks, which is consistent with the research results of babatunde [11], li [12] and others. therefore, relevant departments should be coordinated to improve the public’s perceived susceptibility, and at the same time, the public’s self-efficacy should be improved through health education and health promotion, so as to appeal for more people to wear masks. there is a mediating effect in the structural equation model. the perceived severity, perceived benefits and barrier perception have significant mediating effects on the public wearing masks behavior during covid-19 epidemic. self-efficacy and perceived susceptibility are mediating variables in this model. self-efficacy is the most important mediating variable, which is consistent with the research results of mcneill et al. [13] and ling et al. [14]. selfefficacy is the mediating variable of perceived barriers and perceived benefits. the perceived benefits has the greatest influence on self-efficacy. the higher the perceived benefits, the higher the self-efficacy. the greater the risk awareness of covid-19 infection among the public, the greater the self-efficacy of the public, and the more likely it is to promote the implementation of mask wearing. therefore, health education practitioners should guide the public to wear masks scientifically to emphasize their preventive effect on covid-19. perceived susceptibility is a mediating variable of perceived severity and perceived benefits. perceived severity has the greatest impact on perceived susceptibility. the more serious the consequences of covid-19 are considered by the public, the stronger the awareness that they may contract covid-19. therefore, health education practitioners should appropriately increase the publicity of the harm caused by covid-19. 4.4. limitations this study also has some limitations. first, the nature of cross sectional data limited the analysis of causal relationships between variables. secondly, due to the impact of the epidemic, the questionnaires for the study were mainly distributed online. and the respondents’ answers were so uneven that the validity of questionnaires could not be guaranteed. all above may easily lead to data bias and serial errors. third, questionnaires and the collected data used for investigation are presented in the form of self-report, which may cause the difficulty of elimination of the recall bias, thus, affects the scientific nature and reliability of the results to some extent. h. ruyi et al. / journal of risk analysis and crisis response 10(4) 168–175 175 conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution hr and dt are responsible for questionnaire design, data analysis and thesis writing; sc is responsible for translating the thesis; w. yibo is the corresponding author; wx is responsible for thesis guidance; w. yujie and sy are responsible for distributing and collecting the questionnaire. funding to the key research base of social science foundation of shaanxi province (grand: 2020m016) and the qingdao key health discipline development fund (grand: 2020b047) and the regular project of shaanxi sports bureau in 2020 (grant no. 2020148) for their support. acknowledgments thanks to li yunshan of shandong university for her support in data processing. we would like to thank hou qiang, wang wenjin, xie yanfei, zhang huihui, li liting, liu xiaofeng and zhang lu from changzhi medical college for their contributions in the questionnaire. references [1] guo x, lin j, lai c, et al. novel coronavirus and real-time fluorescent rt-pcr nucleic acid detection. shaanxi western med j 2020;49:264–6 (in chinese). [2] general office of the national health commission. diagnosis and treatment plan for covid-19 (trial version 7). 2020. [3] freimuth vs, jamison a, hancock g, musa d, hilyard k, quinn sc. the role of risk perception in flu vaccine behavior among african american and white adults in the us. risk anal 2017;37:2150–63. [4] sim sw, moey ksp, tan nc. the use of facemasks to prevent respiratory infection: a literature review in the context of the health belief model. singapore med j 2014;55:160–7. [5] zai c, zuo h, jiang l, et al. kap survey of community residents on sars and public health. beijing china public health 2004;20:385–6. [6] zhan s, wu t, ren t, et al. cognition and behavior of community residents on atypical pneumonia. beijing j pek univ (med sci ed) 2003;35:95–8. [7] taylor m, raphael b, barr m, agho k, stevens g, jorm l. public health measures during an anticipated influenza pandemic: factors influencing willingness to comply. risk manag healthc policy 2009;2:9–20. [8] kuo pc, huang jh, liu md. avian influenza risk perception and preventive behaviour among traditional market workers and shoppers in taiwan: practical implications for prevention. plos one 2011;6:e24157. [9] lau jtf, kim jh, tsui hy, griffiths s. anticipated and current preventive behaviours in response to an anticipated the humanto-human h5n1 epidemic in the hong kong chinese general population. bmc infect dis 2007;7:18. [10] tang csk, wong cy. factors influencing the wearing of facemasks to prevent the severe acute respiratory syndrome among adult chinese in hong kong. prev med 2004;39: 1187–93. [11] babatunde ot. importance of self-efficacy and knowledge to physical activity behavior in older african americans. j health care poor underserved 2015;26:1223–34. [12] li s. a study on the influencing factors of cervical cancer prevention behavior in community women of childbearing age in hangzhou based on health belief model. zhejiang: hangzhou normal university; 2019. [13] mcneill lh, wyrwich kw, brownson rc, clark em, kreuter mw. individual, social environmental, and physical environmental influences on physical activity among black and white adults: a structural equation analysis. ann behav med 2006;31:36–44. [14] ling j, robbins lb, mccarthy vl, speck bj. psychosocial determinants of physical activity in children potential afterschool programs: a path analysis. nurs res 2015;64:190–9. https://doi.org/10.1111/risa.12790 https://doi.org/10.1111/risa.12790 https://doi.org/10.1111/risa.12790 https://doi.org/10.1111/risa.12790 https://doi.org/10.11622/smedj.2014037 https://doi.org/10.11622/smedj.2014037 https://doi.org/10.11622/smedj.2014037 https://doi.org/10.2147/rmhp.s4810 https://doi.org/10.2147/rmhp.s4810 https://doi.org/10.2147/rmhp.s4810 https://doi.org/10.2147/rmhp.s4810 https://doi.org/10.1371/journal.pone.0024157 https://doi.org/10.1371/journal.pone.0024157 https://doi.org/10.1371/journal.pone.0024157 https://doi.org/10.1371/journal.pone.0024157 https://doi.org/10.1186/1471-2334-7-18 https://doi.org/10.1186/1471-2334-7-18 https://doi.org/10.1186/1471-2334-7-18 https://doi.org/10.1186/1471-2334-7-18 https://doi.org/10.1016/j.ypmed.2004.04.032 https://doi.org/10.1016/j.ypmed.2004.04.032 https://doi.org/10.1016/j.ypmed.2004.04.032 https://doi.org/10.1016/j.ypmed.2004.04.032 https://doi.org/10.1353/hpu.2015.0116 https://doi.org/10.1353/hpu.2015.0116 https://doi.org/10.1353/hpu.2015.0116 https://doi.org/10.1207/s15324796abm3101_7 https://doi.org/10.1207/s15324796abm3101_7 https://doi.org/10.1207/s15324796abm3101_7 https://doi.org/10.1207/s15324796abm3101_7 https://doi.org/10.1097/nnr.0000000000000084 https://doi.org/10.1097/nnr.0000000000000084 https://doi.org/10.1097/nnr.0000000000000084 atlantis press journal style received 3 december 2016 accepted 12 december 2016 path optimization in dynamic adverse weathers mingkong zhang1,2,3, xiaobing hu1,2,3, jianqin liao4 1state key laboratory of earth surface processes and resources ecology (beijing normal university) beijing 100875, china 2key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing 100875, china 3academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education the peoples' republic of china, beijing normal university, beijing 100875, china 4miidshare technology ltd, chengdu, 610000, china abstract as we all know, weather condition is constantly changing. then, when an adverse weather event occurs during travelling period, path planning becomes a dynamic path optimization (dpo) problem. a common practice of dpo is to conduct real-time online path optimization based on the current weather condition. however, the result of online optimization under the current weather condition is rarely optimal for future weather conditions. we are concerned with how to achieve optimal actual travelling trajectory by just a single offline optimization, given the dynamics of weather conditions is pre-known. to this end, the concept of co-evolutionary path optimization (cepo) is introduced, where the weather condition in a single run of offline optimization is not static, but keeps changing during the single run of offline optimization. existing dpo methods can hardly address cepo, because they do not allow the weather condition to change in a single run of online optimization. to address the cepo in dynamical adverse weathers, this paper proposes a ripple-spreading algorithm (rsa), which can achieve optimal actual travelling trajectory by a single offline calculation. the reported cepo and rsa are then tested on a typhoon scenario in hainan province of china, and the advantages against traditional dpo methods are clearly demonstrated. keywords: co-evolutionary path optimization, adverse weathers, ripple-spreading algorithm, typhoon. 动态极端天气情况下的路径优化问题 张明空 1,2,3 ，胡小兵 1,2,3 ，廖建勤 4 1.北京师范大学地表过程与资源生态国家重点实验室，北京 100875 2. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875 3. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875 4. 成都鸣先科技有限公司，成都，610000 摘要：众所周知，天气情况是实时变化的。所以，当旅途中遭遇极端天气时，路径规划问题就成为了动态路径 优化问题（dynamic path optimization，dpo）。动态路径优化的普遍求解思路是：基于当前天气情况实时优化 路径。然而，虽然在线优化的结果对于当前天气情况而言是最优的，但对于未来天气情况而言不见得是最优的。 journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 197-205 published by atlantis press copyright: the authors 197 m.k. zhang et al. / path optimization in dynamic adverse weathers 所以,基于当前天气情况而不断在线实时优化所得到的实际旅行路径一般都不是最优的。本文关心的问题是：假 定已知天气变化的动态规律，如何通过一次性的离线优化就能使实际旅行路径达到理论最优？为此，本文提出 了协同进化路径优化（co-evolutionary path optimization，cepo）概念，它要求：在一次性的离线路径优化计 算中充分考虑天气的动态变化过程。目前已有的 dpo 方法很难解决 cepo 问题，因为在 dpo 的每次在线优化 计算中，未来的天气变化情况并没有被考虑。本文提出了一种涟漪扩散算法（ripple-spreading algorithm， rsa），通过一次性的离线优化就能找到动态变化天气情况下的最优实际旅行路径，从而解决 cepo 问题。为 了验证 cepo 和 rsa，本文以中国海南岛路网为背景，模拟海南岛台风登陆的动态过程，并在这种动态天气情 况下进行路径优化仿真实验。实验结果证明：本文所提出的基于 rsa 的 cepo 方法较传统的 dpo 方法，不论 在计算时效还是在求解效果上，都更具优势。 关键词： 协同进化路径优化；动态路径优化；极端天气；涟漪扩散算法；台风 1. 前言 日常出行中的路径规划通常都需要考虑天气情 况。如果旅途中遇到恶劣天气，那么，不仅我们出 行的舒适性会受到影响（例如，极端天气造成的交 通拥堵）[1]-[2]，我们出行的安全性也会受到威胁 （例如，台风、龙卷风、暴风雨和暴风雪经常会引 起致命的交通事故）[3]-[4]，所以，在极端天气情 况下进行路径优化具有重要的现实意义。优化路径 一方面可以避开受极端天气影响的区域或道路，降 低气象灾害风险[5]，增加旅行的舒适性和安全性， 提高面对气象灾害时，降低应急时承灾体的脆弱性 [6]。另一方面也可以降低相关旅行成本。在一般的 路径优化问题中，旅行成本可以用距离、油耗和路 桥费等指标来衡量，然而，在极端天气情况下，旅 行时间通常会被作为优先考虑的指标，因为在极端 天气情况下，安全问题是重中之重，对于旅行者而 言，如何以最短的时间安全完成旅行任务才是最重 要的。而且，旅行时间最小化在自然灾害疏散、救 援和应急管理中也具有重要地位[7]。许多自然灾害 通常会伴随有极端天气情况发生（例如，灾难性的 地震之后很可能会伴随暴风雨）[8]。最小化旅行时 间就意味着尽可能缩短暴露在极端天气情况下的时 间，从而降低受灾风险。所以，在减灾和风险管理 研究领域中，极端天气情况下的路径优化问题具有 重要的研究价值[9]-[10]。 显然，极端天气情况下的旅行是一个动态路径 优化问题。例如，当台风中心随时间移动时，其经 过的区域就会有暴风雨，暴风雨将导致该区域的通 行条件恶化，只有等暴风雨过境后才能恢复有效通 行，从而在一定时间内对旅行造成影响。其实，动 态路径优化问题并不是一个新问题，国内外的学者 已经提出了很多求解方法。例如，文献[11]-[14]研 究了针对单一路网改变的动态路径优化问题；文献 [14]-[17]则研究了针对一系列路网改变的路径优化 问题。在文献[11]和[17]的方法中，基本上，原始 路网一旦有任何改变，优化算法就需要在线重新计 算以确定新的最优路径。事实上，每次在线路径优 化计算所求解的是一个基于当前路网的静态路径优 化问题（static path optimization，spo），因为单 次在线路径优化计算中所使用的路网都是静止不变 的。由于受在线计算时间的限制，文献[11]-[14]中 的动态路径优化方法的关注重点是：能否不进行完 全的重新计算，而是结合当前路网对以前的优化路 径进行恰当的改进就达到优化的效果？换句话说， 文献[11]-[14]中的动态路径优化方法强调在每次路 网发生改变时求解新 spo 的速度。在线求解 spo 的结果对于当前路网来说可能是最优的，但是在实 际旅行中，不断重复在线求解 spo 所走出的实际 旅行路径，对于给定的路网动态变化规律而言，很 难保证最优性。图 1 展示了 dpo 的求解过程，以 及 dpo 所导致的实际旅行路径与理论最优的实际 旅行路径之间的差异。事实上，dpo 方法不能得 到理论最优的实际旅行路径，其主要原因是忽略了 路网未来变化的可预测性。在 dpo 的单次在线路 径优化计算中，动态变化的路网被简化成当前时刻 的路网（静态路网），因而单次在线路径优化的结 果对于未来的路网而言，一般很难是最优的。 图 1. 动态路径优化（dpo）不能实现最优实际旅行路径 published by atlantis press copyright: the authors 198 m.k. zhang et al. / path optimization in dynamic adverse weathers 事实上，得益于天气预报技术的快速发展，许 多由于天气原因对路网造成的改变或多或少是可以 预测的。例如，对台风中心的位置、移动轨迹、以 及暴风雨的持续时间，目前都可以提前几个小时准 确预报。显然，充分利用天气预报信息进行动态极 端天气情况下的路径优化，对于提升优化效果是非 常重要的。当前，对于灾害发生规律的研究已经有 相应基础，例如连续的 markov 决策模型模拟灾害 随机发生过程 [18]和预测灾害的有效发生时间[19]； 对于社会灾害，城市火灾发生时蔓延的规律及模型 的研究也有不少相关报道[20]-[21]。这些研究都有 助于提高灾害预测技术的发展，帮助增强 cepo 单 次离线路径最优结果的解算精度。 本 文 通 过 改 进 文 献 [22] 中 的 涟 漪 扩 散 算 法 （ripple-spreading algorithm，rsa），提出了一 种解决动态极端天气情况下路径优化问题的新方法， 通过利用天气预报提供的未来天气变化信息，执行 一次性的离线优化，从而找到在所预报的天气动态 变化规律下的理论最优的实际旅行路径。为了实现 这个目标，就必须在一次性的离线路径优化过程中 把天气变化对路网的动态影响考虑进去。换句话说， 天气动态变化情况和离线优化过程必需协同进行。 由于天气情况和路径优化求解是一起协同进化的， 因此我们将此新方法称为协同进化路径优化方法 （co-evolutionary path optimization，cepo）。传 统的 dpo 方法很难处理 cepo，因为天气情况的 变化没有（事实上是没法）在 dpo 的单次在线优 化计算中得到考虑。文献[22]中的 rsa 区别于传统 路径优化算法的最大特点就是：rsa 实际上是一种 确定性的去中心化的多智体仿真模型，而不是传统 类型的集中式搜索算法。因为集中式搜索算法需要 进行全局比较，所以在算法的单次运行中，路网不 允许变化（否则全局比较就会失去意义和作用）， 因而在单次运行中无法考虑天气情况的变化。去中 心化的多智体仿真模型则不一样，对于仿真模型中 的单个智体而言，其行为只取决于当前该智体所处 位置的路网部分的情况，路网中其它部分的变化并 不会影响该智体的行为。因此天气情况的动态变化 可以自然而有机地整合到多智体仿真过程当中。正 是基于去中心化的多智体仿真模型的特点，本文通 过对 rsa 进行改进，从而可以有效解决 cepo 这 一传统 dpo 方法所不可能完成的任务：即，通过 一次性的离线优化就能找到动态变化天气情况下的 最优实际旅行路径（而传统 dpo 方法需要不停地 重复在线实时优化，可是 dpo 方法下实际走出的 旅行路径却通常并不是最优的）。 2. cepo 问题描述 用 g(v,e)来表示路径网络，v 表示结点集合， e 表示结点间的链接集合。v 有 nn 个不同的结点， 包括起点和终点，e 有 ne 条链接。路径网络可以 表示成 nn *nn 的邻接矩阵 m。矩阵 m(i, j)=1, i=1,…,nn，j=1,…,nn,表示结点 i 和 j 之间存在一条 链接，否则 m(i, j)=0，即结点 i 和 j 之间没有链接。 本文不考虑结点自我连接的情况，因而有 m(i,i)=0。 每条链接上的旅行速度 sl(i, j)与链接 m(i,j)有关。 c(i, j)为链接上的成本，也与 m(i, j)有关。不同等 级的链接分别有不同的旅行速度，根据链接 m(i, j) 的旅行速度 sl(i, j)，就可以计算从结点 i 到结点 j 的所用时间。 假设候选路径通过整数型向量 r 来记录，其元 素 r(i)=j 表示结点 j 是路径 r 中的第 i 个结点， i=1,…,np, j=1,…,nn,np 表示路径 r 中有多少个结点， 包括起点和终点。显然 r(1)是起点，r(np)是终点。 本文研究中，不允许任何一结点在同一条路径 r 中 出现次数多于一次，即整数型向量 r 满足如下条件： r(i)≠r(j), 如果 i≠j, i=1,…,np, j=1,…,np. (1) 为了在 t=0 时刻计算出最优路径，需要基于已 知的天气预报信息，预测在未来各个时刻有多少结 点和链接会受到极端天气的影响，从而邻接矩阵 m 就会随时间发生改变。mk|0 表示在当前时刻 t=0 所 预测的未来时刻 t=k 时的邻接矩阵。 基于上述数学符号定义，传统 dpo 在当前时 刻 t=0 的单次在线优化计算问题可以描述如下： 约束条件为：公式（1）和 ( )min ,dpo p r f r n ∈ω (2) m0(r(i), r(i+1))=1, i=1,…,np-1, (3) ω 是所有可能连结起点与终点的路径集合, ( ),1 0dpof r = (4) ( ) ( ) ( )( ), , 1 1 , ( )dpo dpof r i f r i c r i r i= − + − 1 < i ≤ np. (5) 如公式（3）所示，dpo 在当前时刻 t=0 的单 次在线优化计算中，路径网络固定不变的。一旦路 径网络在未来时刻 t=k>0 时刻发生改变，dpo 就需 要基于 mk 重新在线优化计算一次（注意：时刻 t=k 的起点与时刻 t=0 的起点一般是不一样的）。 与 dpo 彻底不同的是，本文所提出的 cepo 在其一次性的离线优化计算中，路径网络不是固定 不变的，而是随预测时间 k|0 而不断变化的。cep o 问题的数学描述如下： published by atlantis press copyright: the authors 199 m.k. zhang et al. / path optimization in dynamic adverse weathers ))(,(min rlrfcepor ω∈ , (6) 约束条件： 1))1(),((0| =+irirm k , 1,..., += ii kkk (7) 0)1,( =rf cepo , (8) ))1(),(()0|),,(max()1,( ++=+ irirckirfirf icepocepo , 1 ≤i< np -1, (9) fcepo(r,i)是用来计算沿路径 r 到达结点 r(i)的 时间，ki|0 是当沿路径 r 到达结点 r(i)后，结点 r(i) 变为可以通行的预测时刻。换句话说，当沿路径 r 于时刻 fcepo(r,i)到达结点 r(i)时，结点 r(i)并不一 定可以通行，因此需要在结点 r(i)处等待，预测需 要等到时刻 ki|0 结点 r(i)才变为可以通行，此时旅 行才得以继续。所以沿路径 r 通过结点 r(i)的预计 时间是 max(fcepo(r,i),ki|0)。mk|0(r(i), r(i+1))=1 表 示的意思是：根据在初始时刻 t=0 的预测，结点 r (i)和 r(i+1)之间在预测时刻 t=k|0 存在可通行的链 接。 公式(7)和公式(9)清楚的表示：在正在通过的 r 的一个子路径中，其连通性和预测旅行时间是相 互依赖的，这一点 dpo 并没有考虑。根据预测时 间 k|0 改变 mk|0 是由动态变化的极端天气情况决定 的，公式如下： )( 0|0|1 kfawdk mfm =+ , k≥0, (10) 约束条件是 m0|0=m0，ffawd 是由天气预报的极端天 气动态变化规律所决定的函数，m0 是 t=0 时的邻接 矩阵。 从公式(6)和(7)对 cepo 的数学描述可以看出， 如果 ffawd 已经确定，即，极端天气情况的动态变 化过程能够预测，并且给定 t=0 时的 m0，那么我 们就能进行一次性离线优化来找到动态变化天气情 况下的最优实际旅行路线，而完全不需要像 dpo 那样不断在线重新计算 t=k>0 时的最优路径（仅仅 对于时刻 t=k 的路网 mk 而言是最优的）。 比较 dpo 中的公式(3)和 cepo 中的公式(7)， 我们能发现，在一次优化过程中，dpo 只关心两 个结点间是否有链接，而 cepo 需要知道什么时候 哪些链接可以通过，即，需要预测某条链接是否在 恰当的时间变为可以通行。换句话说，在 dpo 中 的基本计算操作是分析链接，例如，一个链接需要 多长的旅行时间；而在 depo 中的基本计算操作是 分析各个预测时刻的仿真时间单位长度内的路网变 化和旅行行为，即，在模拟一个仿真时间单位长度 内，路径网络如何改变，然后基于这种改变，根据 公式(9)沿着某条链接的旅行时间可能就会受到什么 影响，例如， )0|),,(max( icepo kirf 就意味着旅行 者在结点 r(i) 处可能需要等到时刻 0|ik （假设 fcepo(r,i)<ki|0）才能够继续进行旅行。 为了解决 cepo 问题，必须有新方法能够进行 基于仿真时间单位长度的基本计算分析操作。不幸 的是，现存的 dpo 方法仅仅能够进行基于链接的 基本计算分析操作。将基于链接的方法转换成可以 进行基于仿真时间单位长度的仿真分析是非常困难 的，因为一条链接可能跨越的仿真时间段及仿真时 间段长度是不确定的（因计划的旅行路径而异）， 且各条链接所跨越的仿真时间段及仿真时间段长度 通常也各不相同，所以基于链接的分析与基于仿真 时间单位长度的分析是很难相互兼容通用的。因此， 必需研发全新的能够基于仿真单位时间长度进行分 析操作的方法，以解决 cepo 问题。 图 2. 涟漪扩散算法求解 cepo 的过程的图解 3. 涟漪扩散算法（rsa） 3.1 涟漪扩散算法的基本理念 正如文献[22]中揭示的，自然界的涟漪扩散现 象本身反映了一种优化原则，即，涟漪在水面上以 相同的速度向四周扩散，因而总会率先到达离涟漪 中心最近的空间点。涟漪到达空间中某点的时间是 由该点到涟漪中心的距离决定的。直觉上,涟漪扩 散现象可以用于寻找到最近的兴趣结点,例如，找 到附近最近的加油站；更进一步的扩展应用就是找 到最短路径。 published by atlantis press copyright: the authors 200 m.k. zhang et al. / path optimization in dynamic adverse weathers 基于自然界的涟漪扩散现象反映的优化原则， 文献[22]中开创性地提出了涟漪扩散算法（ripplespreading algorithm，rsa）用以求解多种最短路径 问题。简单说，这就像是在路网结点间进行一次涟 漪接力赛跑，初始激励涟漪从起点开始，当一个涟 漪到达一个未被激活过的结点后，新的结点就会被 激活产生涟漪。当与某个涟漪的波源结点有链接的 所有结点都已被激活过（不一定是被同一涟漪激 活），那么该涟漪就停止扩散而消亡。当目的地结 点第一次被某个涟漪到达时，那么第一最短路径就 已经找到，涟漪接力赛也因此终止。在整个过程中， 所有涟漪都以相同的预设固定速度扩散。 正如文献[22]中强调的，rsa 并不是集中式的、 自上而下的逻辑算法，而是分散式的、基于多智体 的、自下而上的仿真模型。在 rsa 中，不需要计算 或排序从起点到中间结点的距离，而是只需模拟结 点的涟漪激活和扩散行为，涟漪扩散的最终结果自 然而然就是最短路径。这种基于自然的多智体分散 式仿真模型使得 rsa 具有更高的修改灵活性，例如， 我们只需要根据给定问题的特点来设置结点的涟漪 激活和扩散行为，不需要考虑任何全局搜索逻辑和 准则，从而方便我们解决各种路径优化问题。例如， rsa 可以用来解决前 k 条最短路径问题,完全不同于 现有的解决方法，这一点文献[23]中有详细的介绍。 现有方法需要基于前（j-1）条最短路径不断重建网 络并计算比较，以便确定第 j 条最短路径。rsa 只 需在最初的网络中进行一次涟漪扩散接力赛，就可 以确定所有的前 k 条最短路径，而完全不需要任何 网络重建和迭代计算过程。rsa 也可以处理一些更 复杂、且现有方法无法解决的问题。例如，目前还 没有切实有效的其它方法可以解决前 k 个最短工程 时间问题，并且保证结果的最优性；而通过在 rsa 中引入结点上的等待行为，就为解决这一问题带来 了希望[24]。接下来，我们将阐释如何改进[22]中的 rsa，以便解决动态极端天气情况下的 cepo 问题。 3.2 应用 rsa 解决动态极端天气情况下的 cepo 正如第二节强调的，cepo 和传统的 dpo 根 本不同之处是：在单次优化过程中，dpo 把天气 情况当作静止处理，而 cepo 是在天气变化的同时 进行路径搜索。所以，要想解决 cepo，就应该有 能力应对和处理天气变化在单次优化过程中对路网 造成的影响。由于天气情况的变化会影响路径网络， 从而改变路网，所以解决 cepo 的方法的单次优化 过程就必须结合网络变化同时进行。 正如上一节分析的，天气情况的变化是基于仿 真时间单位的模拟过程，例如，在每个仿真时刻的 单位时间长度内，需要计算天气情况的改变和以及 所引起的路网改变。幸运的是，rsa 中的涟漪激活 和扩散行为也是基于仿真时间单位的模拟操作，例 如，在每个仿真时刻的单位时间长度内，涟漪如何 扩散，结点对激励涟漪如何反应。所以，在 rsa 中引入天气和路网变化是非常自然和容易的。换句 话说，在每个仿真时刻的单位时间长度内，一方面， 我们要根据预报的天气动态变化规律更新的天气和 路网情况，另一方面，我们要同时模拟该单位仿真 时间长度内涟漪的激活和扩散行为。 当然，文献[22]中的 rsa 并不能直接应用于 cepo，而需要做必要的改进。首先，我们需要将 路网更新操作（也就相当于天气变化）整合到文献 [22]的涟漪接力赛中。我们也可以根据需要将链接 进行分级处理，不同等级的链接有不同的涟漪扩散 速度。最后，需要为涟漪扩散引入在结点处的等待 行为，因为，一旦某个结点由于极端天气原因关闭 而暂时无法通行，此时可以选择绕道走或在该结点 处等待，有时等待结点重新开放可能比绕道走更能 节约旅行时间。针对 cepo 改进的 rsa 伪代码给 出如下。假设网络总共有 nn 结点，为方便描述， 我们假设起点和终点分别是结点 1 和结点 nn。sl(n, m)是结点 n 和 m 链接上的旅行速度。在本文研究 中，假设旅行者总是按链接的旅行速度旅行。在涟 漪接力赛中，每个结点最多能被激活一次，结点 i 被激活时产生的涟漪被称为激励涟漪 i。fr(i) = j > 0 表示涟漪 i 是被涟漪 j 激活的。rr(i,:)是一个向量， 表示当前涟漪 i 的半径。sr(i)记录涟漪 i 的状态， sr(i)=0,1,2,3 表示涟漪 i 所处的状态，分别表示不活 跃、等待、活跃和消亡。 第 1 步，初始化，fr(i) = 0, sr(i)=0, rr(i,:) = 0， i = 1, …, nn。 第 2 步，k=0，当前时间 t=k|0。设 fr(1) = 1, sr(1)=2。 第 3 步，如果 fr(nn) = 0， 第 3.1 步，k=k+1,更新时间 t = k|0。 第 3.2 步，更新 mk|0，即，根据预报的天气动 态变化规律，更新路网。 第 3.3 步，对于任一等待涟漪 i，即，sr(i)=1， 如果根据更新的 mk|0，结点 i 变为可以通行，则涟 漪 i 的状态更新为活跃，即，设置 sr(i)=2。 第 3.4 步，对于任一活跃涟漪 i，即，sr(i)=2， 更新涟漪 i 沿着与结点 i 相连的链接扩散的半径， 例如，设结点 i 的第 l 条链结与结点 m 相连，那么， rr(i, l) = rr(i, l) + sl(i, m)。 第 3.5 步，根据 mk|0 对于任意可通行的结点 n, published by atlantis press copyright: the authors 201 m.k. zhang et al. / path optimization in dynamic adverse weathers 如果 sr(n)=0，且它刚被至少一个活跃涟漪到达， 假设是激励涟漪 i，沿着与结点 i 相连的第 l 条链接 扩散第一个到达结点 n，那么根据涟漪 i 到达结点 n 的时间 tr 更新 rr(n,:)，例如：假设结点 n 的第 l 条链接与结点 m 相连，则 rr(n,l) = (k-tr)×sl(n, m)。 设置 fr(n) = i。设置涟漪 n 作为活跃涟漪，即， sr(n)=2。 第 3.6 步，根据 mk|0，对于任意不可通行结点 n，如果 sr(n)=0，且它刚被至少一个活跃涟漪到达， 那么，假设是涟漪 i 第一个到达结点 n，则设置 rr(n)=0, fr(n) = i,同时设置涟漪 n 为等待涟漪，即， sr(n)=1。 第 3.7 步，对于任何一个涟漪 i，无论活跃还 是不活跃，如果对于每个具有 mk|0(i, n) = 1 的结点 n，fr(n) > 0，那么设置 sr(i)=3，即，涟漪 i 消亡。 第 4 步，从 fr(nn)回溯，就可确定最优路径。 步骤 3.2 和步骤 3.4 很好地描述了路网和涟 漪的协同进化关系。根据步骤 3.5，一旦一个可 以通行的结点第一次被任何一个涟漪到达后，那 么在该结点就会立刻产生新的活跃涟漪。步骤 3.4 和步骤 3.5 允许不同的链接可以有不同的涟 漪扩散速度。步骤 3.6 定义了涟漪的等待行为， 一个不可通行结点第一次被任何一个涟漪到达后， 该结点的涟漪虽然会被激活，但会保持等待而不 扩散。根据步骤 3.7，如果与一个涟漪的波源结 点相连的所有结点都已被涟漪到达过（不管是被 哪个涟漪到达），则该涟漪就会消亡。 有了上述伪代码中所引入的协同进化的关系、 以及多涟漪扩散速度和等待行为，rsa 就可以有效 解决 cepo 问题，即，通过一次性的离线优化就能 找到动态变化天气情况下的最优实际旅行路径。图 2 展示了涟漪扩散算法 rsa 是如何解决 cepo 问题 的。简单起见，图 2 中所有链接具有相同的旅行速 度。从图 2 可以看到，由于天气的动态变化，随着 预测时间 k|0，k=1,…,5，路网变化与涟漪接力赛是 协同进行的。如果 dpo 方法运用于该情境中，它 的第一步是先走向结点 9，同时结点 4 在时间 1|0 成为不可通行结点；然而当到达结点 9 时，结点 9 已变成不可通行结点；于是在 4|0 时刻，dpo 方法 会返回结点 1。这就会使得实际旅行时间变长。而 根据本文提出的 cepo, rsa 总是可以找到能够在 恰当的时间避开不可通行点的路径，因此成功地找 到图 2 所给的极端天气动态变化情况下的理论最优 实际旅行路径：1→2→4→5→7→10。 4. 实验结果 为了验证本文所提出的方法，在此，我们以中 国海南省路网为例来研究 cepo。海南省是中国的 第二大岛屿，拥有相对现代化的路网，路网主要包 括 4 大类，环岛的高速网络、遍布全岛的国道、省 级道路和县/村级道路。省级道路连接着各个城市 和县，而县/村级道路连接着各个镇和村。图 3（a） 中给出了海南岛的路网图。由于这里并没有其他外 界路径可以利用，相对封闭，因此就构成了非常适 合路径优化仿真实验的案例路网。众所周知，海南 岛每年都会有台风登陆，所以，在台风季节里，通 过有效的路径优化方法可以合理规划出行路线，提 高旅行安全性、舒适性和旅行效率。 (a) 海南省路网 (b) 从万宁市和三亚市登陆的台风运动示意图 图 3. 海南省路网图和台风运动示意图 台风：三 亚 西北： 37° 台风：万 宁市 西北 45° 时速： published by atlantis press copyright: the authors 202 m.k. zhang et al. / path optimization in dynamic adverse weathers 在实验中，我们首先模拟了从万宁市登陆的台 风，台风中心以每小时 30 公里的速度向西北方向 移动穿过海南岛（图 3(b)模拟了从万宁市登陆的台 风的运动情况），假设所有的距离台风中心 50 公 里的道路都将受台风影响而关闭。假设旅行者在高 速公路上的旅行平均速度是 80km/h，国道的平均 速度是 60km/h，省道的平均速度是 40km/h，乡村 道路平均速度是 30km/h。实验中，从路网里随机 选择旅行起点和终点对，旅行者从所选的起点城市 出发到终点城市处理紧急事情，因而需要以最短的 时间到达目的地。仿真结果见表 1。 表 1：台风从万宁市登陆向西北方向运动情景下的两种 方法仿真结果（pl 单位：km; ct 单位：s； tt 单位：h） 起 点 和 终 点 916 和 274 664 和 436 190 和 576 299 和 464 dpo pl 320.00 344.90 312.30 282.30 tt 9.10 9.85 8.92 8.06 ct 3.81 6.58 5.07 2.70 cepo pl 257.80 189.60 241.80 220.30 tt 7.36 5.41 7.30 6.43 ct 5.68 4.29 5.58 5.04 注：起点 916：莺歌海市 终点 274：琼海市 起点 664： 儋州市 终点 436：陵水市 起点 190：文昌市 终点 576： 乐东市 起点 299: 安定市 根据参考文献[25],海南岛的三亚市也是台风经 常登陆的地方（如图 3(b)，从三亚登陆的台风的运 动方向为西偏北 37°，时速 30km/h），因此本研究 又模拟了台风从三亚市和万宁市同时登陆的情况， 然后随机选取起点终点对进行路径优化，实验结果 如表 2。 表 2：台风从三亚市和万宁市同时登陆情景下的两种方 法仿真结果（pl 单位：km; ct 单位：s； tt 单位：h） 起点和终点 664 和 436 190 和 576 800 和 464 408 和 51 dpo pl 无解 无解 无解 122.90 tt 3.51 ct 4.91 cepo pl 189.50 241.90 148.00 122.90 tt 5.41 7.30 4.22 3.51 ct 7.13 9.45 5.55 1.39 注：190：文昌市 结点 576：乐东市 结点 800：昌河市 结点 464：保亭市 结点 408：琼中市 结点 51：临高市； 为了更好评估本文提出的 cepo 和 rsa 方法， 我们将其与传统的 dpo 方法做对比。在 dpo 方法 中，不断重复的在线优化需要在每个仿真时刻根据 当 前 旅 行 者 的 位 置 和 台 风 中 心 位 置 ， 应 用 dijkstra’s 算法计算当前的最优路径。如本文一直所 强调的，基于天气预报的台风中心移动路径，本文 的 cepo 和 rsa 方法，仅仅需要在起点进行一次 性的离线优化就能找到最优的实际旅行路径。模拟 台风从万宁市登录，随机选取四个起点终点对的仿 真实验结果如表 1。图 4 是台风从万宁市登陆，旅 行者从起点 664 儋州市和到终点 436 陵水市的仿真 路径结果示意图。 表 2 是台风从三亚市和万宁市同时登录时随机 选取四个起点终点对的仿真实验结果。表 1、2 中 的 pl、tt 和 ct 分别代表路径长度、旅行时间和 计算时间。从表 1、2 和图 4 中，有以下观测结果：  cepo 的 pl 和 tt 都比 dpo 的方法短。尤其， cepo 的真实路径长度相比 dpo 缩短 20%45%，cepo 的实际旅行时间也比 dpo 减少 18%-45%，所以，在 tt 和 pl 上，cepo 都比 dpo 更有明显优势。  图 4 展示了 cepo 方法为何比 dpo 方法有更 短的 pl 和 tt。在图 4 的实验中，起初 dpo 运行结果是从东南方向行走，由于海南岛的东 南部遭遇台风，dpo 又准备绕西北部避开走， 一段时间过后，最初绕西北部走的计划又被移 动到西北部的台风打乱，所以，dpo 又返回 绕东南部走。不同的是，cepo 的离线优化可 以直接找到最优路线，并且在合适的时间避开 台风中心，不仅节约了旅行成本，又满足了最 小化时间的要求。  表 2 展示的是台风从三亚市和万宁市同时登录 的仿真结果，表二中的第四组结果显示，dpo 和 cepo 的仿真结果是一致的，这说明：如果 动态天气变化过程不影响 dpo 的最优性时， cepo 的求解结果与 dpo 的一致，从而间接 证明了 cepo 的最优性。  在两个台风同时登陆海南省的时候，表 2 中除 第四组数据，剩余的其它三组仿真结果说明： dpo 在面对更复杂的动态障碍区时（例如， 多个台风同时登录情景）常常无法找到可行解 （更不要说最优解），只能停止运动等待台风 过境，但是这种坐以待毙的方法并不是旅行者 所希望的。相反，cepo 结合台风预报信息， 离线优化，引入结点等待行为使得 cepo 在处 理多个台风同时登录情景时，优化路径能力更 有突出优势。 published by atlantis press copyright: the authors 203 m.k. zhang et al. / path optimization in dynamic adverse weathers 图 4. dpo 和 cepo 结果仿真结果图  表 1 和表 2 就 ct 而言，cepo 拥有和 dpo 相 似的计算效率。值得注意的是，cepo 的 ct 是某个起点终点对下采用 cepo 方法离线优化 所用的时间，而 dpo 的 ct 是旅行者到达终 点时所有在线优化所消耗的时间总和。在本次 实验中，所有的 dpo 和 cepo 都是在同一计 算机上运行。事实上，cepo 的离线优化可以 很容易的采用更多更强大的硬件系统，然而 dpo 却受各种在线优化环境条件的限制而很 难采用大型复杂的硬件系统。所以，在现实生 活中，cepo 有更好的潜力去实现更短的 ct。 5. 结论 本文主要提出了协同进化路径优化（cepo） 和涟漪扩散算法（rsa）在动态极端天气情况下路 径优化中的应用。与现有的动态路径优化（dpo） 方法不同，新方法不需要在线实时优化。由于天气 变化的同时也会影响路网中各个结点和链接的状态， cepo 在一次性的离线优化过程中的每一个单位时 间内都会同时考虑该时刻天气对路网的影响，所以， 离线优化的结果就能保证实际旅行路径的最优性。 rsa 是实现这种离线优化的关键，涟漪扩散优化原 理保证了新方法的最优性。以中国海南省台风为背 景，我们进行了仿真实验，实验结果清晰的表明了 新方法在旅行时间和计算时间上相对 dpo 的优势。 未来的研究将会考虑更复杂的路径网络，同时考虑 更多真实的天气情景以便更好的了解和挖掘新方法 的优势和应用潜力。 参考文献 [1] wolfgang w., weisser, volkl w., hassell m., the importance of adverse weather conditions for behaviour and population ecology of an aphid parasitoid. journal of animal ecology, 1997, 66:386400. [2] william h., lam k., hu s. and sumalee, a., modeling impacts of adverse weather conditions on a road network with uncertainties in demand and supply. transportation research part b : methodological , 2008, 42:890-910. [3] keay k. and simmonds i., the association of rainfall and other eather variables with road traffic volume in melbourne, australia. ccident analysis &prevention, 2005, 37:109–12433. [4] keay k. and simmonds i., road accidents and rainfall in a large australia city. accident analysis &prevention, 2006, 38:445–459. [5] guo s.j. the meteorological disaster risk assessment based on the diffusion mechanism. journal of risk analysis and crisis response, 2012（2）:124-130. [6] jiang t.h. assessment method of emergency preparedness system vulnerability based on the complex network theory. journal of risk analysis and crisis response, 2012，2（3）:195-200. （a）dpo 的结果 pl: 344.90km； tt：9.85h； ct：6.58s （b）cepo 的结果 pl：189.60km ；tt： 5.41h； ct: 4.29s published by atlantis press copyright: the authors 204 m.k. zhang et al. / path optimization in dynamic adverse weathers [7] lakshay, agarwal a., bolia n. route guidance map for emergency evacuation. journal of risk analysis and crisis response，2016, 6(3): 135-144. [8] maarten k, aalst v., the impacts of climate change on the risk of natural disasters. disaster, 2005, 30:5-18. [9] quan c., trani a., and srinivas, s., modeling the economic impact of adverse weather into route flights. journal of the transportation research board, vol. 1788, doi: 10.3141/1788-10. [10] krozel j., lee c.and mitchell j., estimating time of arrival in heavy weather conditions. guidance, navigation, and control conference and exhibit, guidance, navigation, and control and co-located conferences, http://dx.doi.org/10.2514/6.1999-4232. [11] ramalingam g. and reps t., on the computational complexity of dynamic graph problems. theoretical computer science, 1996,158:233–277. [12] frigioni d., marchetti-spaccamela a. and nanni u., fully dynamic algorithms for maintaining shortest paths trees. journal of algorithms,2000, 34:251–281. [13] frigioni d., marchetti-spaccamela a. and nanni u., semidynamic algorithms for maintaining single-source shortest path trees. algorithmmica, 1998, 22:250–274. [14] narvaez p., siu k. and tzeng h., new dynamic algorithms for shortest path tree computation. ieee/acm transactions on networking, 2000, 8:734– 746. [15] ramalingam g. and reps t., an incremental algorithm for a generalization of the shortest-path problem. j. algorithms, 1996, 21:267–305. [16] bauer r. and wagner d., batch dynamic single-source shortest-path algorithms: an experimental study. experimental algorithms, 2009, 51–62. [17] taoka s., takafuji d., iguchi t. and watanabe t., performance comparison of algorithms for the dynamic shortest path problem. ieice transactions on fundamentals of electronics, communication and computer sciences, 2007,90. [18] 叶柳儿，黄香香.连续时间 markov 决策过程的均值方 差 优 化 问 题 . 中 国 科 学 ： 数 学 , 2014 ， 44 （8）:883-898. [19] 李俊平，张利娜.带有灾难和依赖状态控制的成批到 达成批服务排队模型. 中国科学：数学, 2015，45 （5）:527-538. [20] 黄维章，张锁春，雷光耀.城市火灾蔓延的数学模型 和计算机模拟.计算物理, 1993，10（1） [21] 马鑫，黄全义，刘全义.基于物联网的建筑火灾动态 监测方法. 清华大学学报（自然科学版）, 2010，52 （11）. [22] hu, x.b., wang, m., m.s. leeson, e.l. hines, and e. di paolo, a deterministic agent-based path optimization method by mimicking the spreading of ripples. evolutionary computation, 2016, 24:319-346. [23] hu, x.b., wang, m., d. hu, m.s. leeson, e.l. hines, and e. di paolo, a ripple-spreading algorithm for the k shortest paths problem. 2012 the 3rd global congress on intelligent systems (gcis 2012), wuhan, china,2012:202-208. [24] hu, x.b., sun, q., wang, m.s. leeson, and paolo e., a ripple spreading algorithm to calculate the k best solutions to the project time management problem. 2013 ieee symposium series on computational intelligence (ieee ssci 2013), 16-19 april 2013, singapore. [25] 张凯荣, 宋长远, 陈钰祥.近 50 年来海南岛台风记录 及 其 灾 害 性 评 价 . 安 徽 农 业 科 学 , 2010 ， 38 （23）:12280-12882. published by atlantis press copyright: the authors 205 1. 前言 2. cepo问题描述 3. 涟漪扩散算法（rsa） 4. 实验结果 5. 结论 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <feff005500740069006c0069007a007a006100720065002000710075006500730074006500200069006d0070006f007300740061007a0069006f006e00690020007000650072002000630072006500610072006500200064006f00630075006d0065006e00740069002000410064006f00620065002000500044004600200070006900f900200061006400610074007400690020006100200075006e00610020007000720065007300740061006d0070006100200064006900200061006c007400610020007100750061006c0069007400e0002e0020004900200064006f00630075006d0065006e007400690020005000440046002000630072006500610074006900200070006f00730073006f006e006f0020006500730073006500720065002000610070006500720074006900200063006f006e0020004100630072006f00620061007400200065002000410064006f00620065002000520065006100640065007200200035002e003000200065002000760065007200730069006f006e006900200073007500630063006500730073006900760065002e> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <feff04180441043f043e043b044c04370443043904420435002004340430043d043d044b04350020043d0430044104420440043e0439043a043800200434043b044f00200441043e043704340430043d0438044f00200434043e043a0443043c0435043d0442043e0432002000410064006f006200650020005000440046002c0020043c0430043a04410438043c0430043b044c043d043e0020043f043e04340445043e0434044f04490438044500200434043b044f00200432044b0441043e043a043e043a0430044704350441044204320435043d043d043e0433043e00200434043e043f0435044704300442043d043e0433043e00200432044b0432043e04340430002e002000200421043e043704340430043d043d044b04350020005000440046002d0434043e043a0443043c0435043d0442044b0020043c043e0436043d043e0020043e0442043a0440044b043204300442044c002004410020043f043e043c043e0449044c044e0020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200431043e043b043504350020043f043e04370434043d043804450020043204350440044104380439002e> /sky <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> /slv <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> /suo <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> /sve <feff0041006e007600e4006e00640020006400650020006800e4007200200069006e0073007400e4006c006c006e0069006e006700610072006e00610020006f006d002000640075002000760069006c006c00200073006b006100700061002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e007400200073006f006d002000e400720020006c00e4006d0070006c0069006700610020006600f60072002000700072006500700072006500730073002d007500740073006b00720069006600740020006d006500640020006800f600670020006b00760061006c0069007400650074002e002000200053006b006100700061006400650020005000440046002d0064006f006b0075006d0065006e00740020006b0061006e002000f600700070006e00610073002000690020004100630072006f0062006100740020006f00630068002000410064006f00620065002000520065006100640065007200200035002e00300020006f00630068002000730065006e006100720065002e> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style a two-step water-management approach for nuclear power plants in inland china xiaowen ding1,2,3, * wei wang3, ** guohe huang1,2, *** 1. key laboratory of regional energy and environmental systems optimization, ministry of education north china electric power university, no.2, beinong road, beijing 102206, china 2. institute for energy, environment and sustainable communities, university of regina 120, 2 research drive, regina, saskatchewan s4s 7h9, canada 3.research center for beijing energy development, north china electric power university, no.2, beinong road, beijing 102206, china **e-mail: wwfancy@163.com ***e-mail: huangg@uregina.ca qingwei chen water resources management centre, ministry of water resources, no.10, nanxiange street, beijing, 100053, china e-mail: chenqw@mwr.gov.cn guoliang wei nuclear and radiation safety centre, ministry of environmental protection, beijing 100082, china e-mail: weiguoliang78@163.com abstract nowadays, effective management of water withdraw, water consumption and wastewater discharge is desired for nuclear power plants in inland china. in this paper, the inland nuclear power industry and its policies in china were reviewed, a two-step water-management (tswm) approach for the plants was proposed. the framework includes the flow process, main tasks, and tools of tswm management for any nuclear power plant in inland china. finally, suggestions on future development of the management were also put forward. keywords: two-step water-management; water resources assessment; nuclear power plant; inland china * corresponding author: binger2000dxw@hotmail.com. 1. introduction due to economic development and population growth, energy demands are increasing greatly in many countries across the world. at the same time, various energy activities based on fossil fuels lead to the emission of a tremendous amount of greenhouse gases (ghg), which is considered as the primary contributor to climate change (barrie, 2011; wang and sun, 2012). to deal with reliable energy supply and ghg emission journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 184-202 published by atlantis press copyright: the authors 184 mailto:chenqw@mwr.gov.cn willieb typewritten text received 9 may 2014 willieb typewritten text accepted 27 august 2014 willieb typewritten text mitigation, governments around the world have implemented a series of energy-related strategic policies of which new energy (particularly nuclear power) exploitation is probably the most direct and effective one (sovacool, 2008; jasmina et al., 2012). as for nuclear power utilization, inland nuclear power plants (npps) have been developed in 14 countries including the usa (bezdek, 2009), france (kopytko and perkins, 2011), canada (lane et al., 2013) due to limited ideal coastal sites. moreover, npps are also planned to be constructed in several developing countries including china (ma et al., 2010). however, fresh water consumption (khamis and kavvadias, 2012) and radioactive wastewater discharge (nuclear energy institute, 2009) of inland npps are major issues which have attracted high attention, particularly under accident conditions (international atomic energy agency (iaea), 2011). for example, huge water consumption is becoming the most important factor to decide on the survival of inland npps in china (zai, 2012). more importantly, surface water is the main source of drinking water in many countries (zhang et al., 2011). so management of water withdraw, water consumption and wastewater discharge for inland npps based on its quantitative assessment are crucial to not only water resources utilization and conservation but also sustainable development of inland nuclear power industry. researches on water resources management for inland npps have been carried out by several countries such as the usa, france, canada and china. as for water withdraw, it was proposed that the amount of an inland npp adopting ap1000 (advanced passive pressurized water reactor 1000) technology was between 0.96 and 1.27 m3/s·gw (united states nuclear regulatory commission (usnrc), 2009; changjiang water resources commission (cjwrc), 2009; wuhan university, 2009a; wuhan university, 2009b). in the usa, usnrc (1984) required that an assessment of the impact on the flow of the river, instream and riparian ecological communities must be carried out if an inland npp withdrew make-up water from a river whose annual flow rate was less than 9×1010m3/y. with respect to water consumption, khamis and kavvadias (2012) presented that inland npp generally consumed 20 to 83% more of water comparing with that based on coal and at the same capacity. the amount for an inland plant was regarded as approximately 35 to 65 million litres per day depending on the types of reactors (khamis and kavvadias, 2012). moreover, some reference standards rather than compulsory requirements also have been developed. for example, parliament of australia (2006) suggested the maximum amount of cooling water consumption for an inland npp as 0.76 m3/s·gw. besides water withdraw and consumption, much more efforts were made to control wastewater discharge. in the usa, usnrc (1998) required that wastewater discharge of npps should comply with the requirements of clean water a ct as well as the regulations issued by united states environmental protection agency, usnrc and local governments. usnrc (2011a) developed a liquid effluent release standard which included three levels of dose limits to individual members (i. e. 3, 25 and 100 mrem per year). such requirements were also developed in france (nuclear safety authority of france, 2006), canada (canadian nuclear safety commission, 1997), china (ministry of environmental protection of the people's republic of china (mepprc), general administration of quality supervision, inspection, and quarantine of the prc (gaqsiqprc), 2011a; 2011b). in the mean time, wastewater discharge control in successive dry seasons and water conservation in the emergency condition were also improved by countries (nuclear safety authority of france, 2006) and international agencies (iaea, 1986a; 1986b; 1994). however, such water resources assessment of npps were mainly based on a one-step approach such as environmental impact assessment (eia) (department of justice canada, 2012) or radioactive effluent release assessment (usnrc, 2011b) which emphasized on wastewater discharge control but ignored water withdraw and consumption. in fact, for countries with huge population and various water users, more effective approaches are desired considering their serious conflicts between water supply and demand. such approaches should be able to tackle these water conflicts through decomposing various impact factors into detailed issues that are of practical implications. however, there have no published reports on it. therefore, in this research, a two-step watermanagement (tswm) approach will be proposed to address the issues of water conflicts, where various impact factors will be decomposed into phases of eia and water resources assessment (wra). the eia will be useful for assessing the possible published by atlantis press copyright: the authors 185 impacts that inland nuclear power plants may have on the environmental (soil, land use, water quality, air quality, noise environment, etc) biological, social and economic aspects. moreover, the wra will help to obtain reasonable and feasible water-related schemes for inland npps. the eia will be the basic step in which water resources is considered through assessing the impacts of wastewater discharging on local water quality. subsequently, the wra will be advanced as a secondary assessment procedure which focuses on water resources exclusive, and more important, wra includes whole water-related processes (i.e. water withdraw, water consumption and wastewater discharge) and concerns both water quality and water quantity. therefore, as an extension of the previous efforts on water resources management, a management framework of the tswm approach for npps in inland china is to be proposed. rationalities of water withdraw, water consumption and wastewater discharge of npps, as well as the associated impacts are to be evaluated. the major procedure, tasks and approaches are to be discussed, as well as the completing requirements of tswm reports for china’s inland npps. as a result, a series of suggestions for advancing water resources management for nuclear power plant in inland china will be presented. this research is also expected to provide a reference for countries such as india, pakistan who are operating or planning to construct inland npps and with problems of population explosion and water shortage. 2. the development of inland nuclear power industry in china according to the medium and l ong-term d evelopment plan of n uclear p ower ( 2005 t o 2020) approved by china’s state council, the capacity for nuclear power generation will have reached 40 gw by 2020 and will have approximately accounted for 4% (i.e., 260 to 280 billion kwh) of the total generated electricity in china (national development and reform commission of the prc (ndrcprc), 2007). due to limited land resources in coastal areas and the increasing electricity demands, siting of npps has extended to inland china. in february 2008, ndrcprc identified hunan taohuajiang npp, hubei dafan npp and jiangxi pengze npp (yang and huang, 2010) as the three pioneering npps to develop nuclear power industry in inland china. the central government has thus identified sites, construction units, and the associated preliminary works for each npp. in march 2008, ndrcprc and national energy administration of the prc (neaprc) proposed that the medium and longterm d evelopment p lan of n uclear p ower (2005 to 2020) should be modified and the proportion of nuclear power to the installed total capacity would have been over 5% by 2020 (wang, 2009). in december 2009, the second national energy conference was held in which nuclear power industries in inland china were proposed to be booming in the next 5 to 20 years. nuclear energy development in inland areas of the country was shifted from active to aggressive states (zhou and zhang, 2010). up to june 2010, preliminary feasibility reports of the 31 npps in inland china had been approved (dai and yi, 2010; table 1 and figure 1). in january 2011, china energy conference was held in which the three pioneering npps were suggested to be constructed during the period of “the 12th five-year” (i.e. 2011 to 2015). at the same time, many other inland provinces have already carried out preliminary works of nuclear industry development, such as anhui, jilin, and henan provinces (yan et al., 2011). on march 11th, 2011, due to the m 9.0 earthquake, one of the most serious nuclear accidents in history occurred in the fukushima dai-ichi npp, japan (nuclear accident independent investigation commission, 2012). on march 16th, 2011, china's state council made a decision that no npps should be approved until the 12 th five-year plan and 2020 vision of nuclear safety and radioactive pollution prevention and c ontrol ( i.e., nuclear s ecurity plan) have been confirmed (china’s state council, 2011), indicating that china attaches great importance to security and stability of its nuclear industry. especially for any npps in inland china, considering that the ap1000 is a new technology without operating practice, security is considered as the most important factor. by the end of 2011, earthworks of the three pioneering npps in inland china had been completed. site analysis of npps had been carried out in more than 10 provinces in china (zhang et al., 2011). in 2012 r eport on t he government’s w ork (wen, 2012) and president jintao hu's speech at 2012 seoul nuclear security summit (hu, 2012), china’s nuclear power industry was presented to be developed in a security and sustainable way in the near future. on october 16th, 2012, nuclear security p lan was formally proposed by the mepprc, ndrcprc, the ministry of finance of the prc, published by atlantis press copyright: the authors 186 neaprc, and administration of the defence science and technology industry (mepprc, 2012). on october 24th, 2012, china issued nuclear p ower saf ety p lan (2011 t o 20 20) as well as the medium an d l ong-term table 1. nuclear power plant projects in inland china (up to june, 2010). no. nuclear power plant projects province main construction unit reactor design units planned total capacity (mw) site watershed the progress of wdua 1 dafan hubei cgnpc ap1000 4 5000 yangtze river wdua reports are complete 2 taohuajiang hunan cnnc ap1000 4 5000 3 pengze jiangxi cpic ap1000 4 5000 4 wuhu anhui cgnpc cpr-1000 4 5000 yangtze river not be carried out 5 anqing anhui chg htgr 6 1200 6 jiyang anhui cnnc ap1000 4 5000 7 fengdu chongqing cdc ap1000 4 5000 8 fuling chongqing cpic ap1000 4 5000 9 nanyang henan cnnc ap1000 6 7500 10 songzi hubei cgnpc ap1000 4~6 5000~7500 11 xishui hubei cgnpc ap1000 4 5000 12 zhongxiang hubei cgnpc&cdc ap1000 4 5000 13 changde hunan cgnpc ap1000 4 5000 14 hengyang hunan cgc ap1000 4 5000 15 xiaomoshan hunan cpic ap1000 4 5000 16 zhuzhou hunan cdc ap1000 4 5000 17 xiajiang jiangxi cdc&cgnpc ap1000 4 5000 18 yanjiashan jiangxi cnnc ap1000 4 5000 19 yingtan jiangxi chg ap1000 4 5000 20 sanba sichuan cgnpc ap1000 4 5000 21 shaoguan guangdong cgnpc ap1000 4 5000 pearl river 22 yunfu guangdong cpic ap1000 4 5000 23 zhaoqing guangdong cgnpc ap1000 6 7500 24 guidong guangxi cpic ap1000 4 5000 25 jingyu jilin cpic ap1000 4 5000 song-liao river 26 songjiang jilin cgnpc ap1000 4 5000 27 huanren liaoning cpic ap1000 4 5000 28 longyan fujian chd ap1000 4 5000 rivers in china’s southeast coast 29 sanming fujian cnnc bn800 4 3200 30 longyou zhejiang cnnc ap1000 4 5000 31 xinyang henan cgnpc ap1000 4 5000 huaihe river fig. 1. distribution of nuclear power plants in inland china that have passed preliminary feasibility studies. published by atlantis press copyright: the authors 187 development p lan of n uclear p ower (2010 t o 2020) , and decided to resume the construction of npps (zhong, 2012). 3. complexities of water resources assessment related to inland nuclear plants inland npps, a whole new kind of power plant in china, the construction of which hasn’t been approved up to the present considering the complexities of their waterrelated issues. the most important concerns are huge water consumption and low-level radioactive wastewater discharge under normal working conditions. other major considerations are large quantities of water needed for cooling and middle even high-level radioactive wastewater discharge under accident conditions. besides the characteristics of water withdraw, consumption and discharge of inland npps, national conditions and water resources conditions of china exacerbate the complexities of water-related issues of inland npps. firstly, china, with the largest population in the world, has a considerable conflict between water supply and demand. especially for the site areas of inland npps, there are intensive municipal, industrial, agricultural intakes in site watershed, which makes the assessment of water withdraw and consumption critical. some of the watersheds almost have reached the limits of water resources development and utilization (zhou, 2009). however, the operation of a npp really consumes large amount water for cooling. therefore, huge water consumption becomes a crucial factor to decide on the survival of inland npps in china (zai, 2012). secondly, for more convenient water withdraw and wastewater discharge, sites of inland npps are likely to be close to inland water bodies. different levels of radioactive wastewater under normal working conditions and accident ones would be discharged into the nearby rivers, lakes and reservoirs. more important, different from other countries who have developed inland nuclear power industry (e.g. france), china, with high degree correlation of river networks, takes surface water as the predominant drinking water source, especially for inland npp site areas where surface water is abundant and closely related to inland npps. a balanced solution is called for to resolve the conflict between radioactive wastewater discharge and the conservation of drinking water source considering that there are many drinking water sources in site watersheds and their downstream. therefore, not only water quality but also water quantity of discharged wastewater should be assessed and a series of drainage schemes under different conditions should be proposed subsequently. thirdly, even though the probability of a nuclear accident is as low as 105/unit·year (usnrc, 1990), the supply of huge accident handling water and the impacts of radioactive wastewater discharge on receiving water bodies must be careful deliberated. the amounts of accident handling water and the methods of water supply for possible nuclear accidents should be fully considered. furthermore, the effects of middle even high-level radioactive wastewater discharge under accident conditions on site watersheds, especially for local and downstream drinking water sources, should be assessed quantitatively. nowadays, china’s each construction project is required to carry out eia which merely focuses on wastewater discharge control as far as water resources is concerned. however, it is insufficient for inland npps considering the complexities of water-related issues for nuclear power plant in inland china. in addition, ap1000 adopted by npps in inland china is a totally new technology without any operating practice in the world. therefore, a two-step approach is demanded to address these complexities. in other words, wra needs to be conducted following eia to assess the whole water water-related processes of inland npps and to propose the feasible schemes. according to current nuclear power policies, the three pioneering npps may be constructed during the period of the 13th five-year (i.e., the years from 2016 to 2020). considering that wra approval is a legal process before the construction of any npps, wra management system for it should be constructed initially before the beginning of the 13th five-year. therefore, wra of inland npps is an innovative but urgent work in china. in the following, the article tentatively put forward a management framework of tswm for npps in inland china in order to promote the establishment of the system. 4. development of two-step water-management approach for nuclear power plants a management framework of the tswm approach for npps in inland china is to be proposed. the first step is eia, which has been developed for npps in china. the further step is wra, which is required by management approaches of water-draw and utilization assessment published by atlantis press copyright: the authors 188 of construction projects and guidelines o f water-draw and u tilization a ssessment for c onstruction p rojects (trial) but is still in the stage of management framework construction. therefore, the major procedure, tasks and approaches are to be presented and discussed. furthermore, the completing requirements of assessment reports for china’s inland npps are also to be proposed. 4.1. flow process of two-step water management the flow process of tswm for npps in inland china should include: (a) initially, implementing eia management for a inland npp, (b) secondly, carrying out wra with the beginning of the construction unit entrusting a qualified agency, (c) the qualified agency drafting an outline of the wra report completion, (d) the related watershed authority approving the outline and providing necessary comments, (e) the qualified agency carrying out the assessment and completing report based on the outline and comments, (f) the construction unit submitting a application to the watershed authority with the required documents, (g) the watershed authority approving the report preliminarily, providing basic comments, and submitting the documents to ministry of water resources of the prc (mwrprc) for administrative approval, (h) mwrprc reviewing the application documents and making a decision for acceptance or rejection, (i) mwrprc organizing site survey and holding a hearing in the site if the application is accepted, (j) review committee reviewing the report and providing suggestions and comments, (k) mwrprc making the decision according to review comments of the committee, records of site survey and hearing, and suggestions of local water resources management departments, (l) mwrprc issuing an approval letter or the reasons for rejecting the water-related schemes proposed in the wra report to the construction unit who may legally apply for re-approval, and (m) the water resources management departments supervising design, construction and operation of the inland npps according to the wra report and the approval documents if the issued file is an approval one. figure 2 shows the flow process chart of tswm for npps in inland china. 4.2. main tasks of two-step water management the main tasks of tsym for npps in inland china should mainly consist of the following five tasks: i）carrying out eia management for a npp in inland china one of the main tasks of tswm is carrying out eia management in the stage of feasibility study. according to the e nvironmental i mpact a ssessment law of the people's republic of china (the president of the people's republic of china, 2003), the construction unit of an inland npp shall entrust a qualified agency with the task of eia. then the qualified agency should develop the assessment and compile the eia report. as for water resources, the eia report of an inland npp mainly addresses the radioactive concentration of receiving water body and its mitigation measures. furthermore, public participation is to be conducted by the methods of holding hearings, holding colloquia and so on to ask for the public opinions. finally, the mepprc shall make an administrative decision based on the eia report and public opinions. the first step of tswm (i.e. eia management) for npps in inland china is conducted in accordance with the environmental i mpact assessment l aw of t he people's republic of china and has been performed in china. however, the second step of tswm (i.e. wra management) for npps in inland china is still in the initial and exploration period. therefore, main tasks of it are proposed specifically as following. ii) stating the requirements for completing the wra report according to management a pproaches of w ater resources assessment of c onstruction p rojects and management a pproaches of w ater r esources assessment qualification of c onstruction projects (mwrprc, ndrcprc, 2002b), the qualified agency shall carry out the wra report completion according to the related requirements and regulations. considering its particular water-related characteristics, proposing special requirements and regulations should be one of the major tasks of the management system of wra for npps in inland china. water resources management departments should carry out investigations and researches on the issues including current situation of inland npps and the water resources management in other countries, control indicators of water use of inland npps, low-level radioactive wastewater discharge published by atlantis press copyright: the authors 189 control of inland npps, dilution and diffusion ability of receiving water, the impact of accidents on water resources and the countermeasures, and water resources constraints for inland npp siting and so on. in the near future, technical standards and guidelines of wra report completion for inland npps needs to be put forward. iii) organizing review panels of wra reports according to administrative p ermission l aw, management a pproaches of w ater r esources the construction unit entrusting a qualified agency with wdua report completion the qualified agency drafting an outline of the wdua report completion the related watershed authority approving the outline and provides necessary comments the qualified agency carrying out the assessment and report completion the watershed authority approving the report preliminarily, providing basic comments, and submitting the documents to mwr the approval file is issued the construction unit submitting a application to the watershed authority mwr reviewing the documents and making a decision for acceptance or rejection mwr organizing site survey and holding a hearing conference in the site review committee reviewing the report and providing suggestions and comments mwr making the administrative approval decision mwr issuing a approval or the reasons for rejecting to the construction unit the water resources management departments supervising design, construction and operation of the nuclear power plant according to the wdua report and the approval files whether the construction unit applying for re -approval? yes no the reasons for rejecting is issued the construction unit modifying the water-related designs or abandons the project yes no whether the submitted files are desirable? eia management for a inland npp fig. 2. the flow chart of tswm for nuclear power plants in inland china. published by atlantis press copyright: the authors 190 assessment of construction projects, as well as management r egulations of r eviewing of water resources assessment reports for construction projects (trial), the water resources management departments shall organize the outline and report review of a npp. mwrprc or the watershed authority shall identify more than seven experts from the expert database established by mwrprc to form a review committee. background, education, experiences and achievements should be mainly considered in expert selection. as far as current expert database established by mwrprc is concerned, most of the experts are major in waterrelated areas such as hydrology, hydrogeology, water resources management, and water pollution control. in future, experts should understand the water-related processes of construction and operation of npps. moreover, nuclear experts need to be involved in the expert database. among the experts of a review committee, one of them should be appointed as the director to preside over the review meeting. major duties of the director should include: (a) hosting the review meeting, (b) carrying out principal technical review of a wra outline or a report, (c) declaring the review comment achieved by the committee, (d) signing the review comment, and (e) taking technical responsibility for the review comment. in the review meeting, committee members should express their opinions on water-related schemes proposed in an outline or a report, and give suggestions on water withdraw, water consumption, wastewater discharge, as well as water conservative measures for improving the report. the opinions and suggestions should be significant references for the following administrative approval. iv) carrying out administrative approval of wra reports according to the laws and regulations mentioned above, administrative approval of a wra report for a npp in inland china is an official decision-making which give a government’s opinion on the water-related schemes proposed in the report. the decision should be based on the requirements of state laws and regulations, the characteristics of a inland npp, the impact on the water body (including the intake and receiving ones) and other water users, rationality and feasibility of water resources conservative and mitigation measures, the comment of the review committee, the suggestions of local water resources management departments, the opinions of other water users of local watersheds, and other related conditions. in china, watershed authorities should take the responsibilities of outline approving as well as preliminary report approving of npps, and then mwrprc should approve reports and make final decisions. the emphases of administrative approval should be the rationalities, feasibilities and reliabilities of the schemes of water withdraw, water consumption, wastewater discharge, and water resources protection. mwrprc and watershed authorities shall give approval documents within 20 days if the submitted files are well prepared. administrative approval will ensure the legality and validity of wra management, and promote construction units to fulfill their obligations of water resources protection actively. v) supervising and monitoring the enforcement of wra reports and approval documents according to management a pproaches of w ater resources as sessment of c onstruction p rojects, management a pproaches of w ater r esources assessment qualification of construction projects, and management r egulations o f r eviewing o f water resources assessment reports for construction projects (trial), after the wra report of a npp is approved by mwrprc, the construction unit can apply for water withdraw permission and intake water from the specified water body. design, construction and operation of the power plant should be consistent with the wra report and approval document, which should be also supervised by water resources management departments. methods of supervising may include remote monitoring, site inspection, post-assessment and so on. 4.3. management tools of two-step water management i) making laws, regulations and standards legislation should be a legal method to facilitate the tswm for npps in inland china. besides the environmental i mpact assessment law of t he people's republic of china, tswm for npps in inland china also involves regulations relating to wastewater discharge, such as integrated w astewater discharge standard (gb 8978-1996) (mepprc, gaqsiqprc, 1996), environmental quality st andard f or surface water (gb 3838-2002) (mepprc, gaqsiqprc, 2002) and so on. particularly, in 2005, mwrprc issued published by atlantis press copyright: the authors 191 requirements o f w ater r esources a ssessment r eports for construction p rojects (sl/z322-2005) (i.e., the annex of management a pproaches of w ater r esources assessment o f construction p rojects). it gives specific requirements on technical levels, scopes, contents, and approaches for the two steps of tswm. but none of them is specifically for npps in inland china. different from other construction projects, npps in inland china involves not only huge fresh water consumption, but also the discharge of radioactive wastewater and cooling water. so legislation on wra for npps in inland china is urgently needed. in the near future, water resources management departments should issue a special technical guideline in which the scope and depth of assessment, the content and focus of analysis, the indicators of rationality analysis of water withdraw and consumption, as well as the methods of impact estimation of water withdraw and wastewater discharge are specified. ii) field trip and site survey field trip and site survey are helpful for decision makers and review experts to investigate the planned sites of npps and identify potential risks and water-related problems. by this ways, they can obtain direct and useful information about current situations of water resources in local area. at the same time, they can capture the opinions of local water resources management departments through acquiring. so field trip and site survey should be the tools to support management for npps in inland china. field trip and site survey should focus on issues related to water withdraw and radioactive discharge, such as water withdraw sites, wastewater treatment facilities, locations of discharge outlets, current situation of water resources of intake and receiving water bodies, and the public’s perception. meanwhile, records, photographs, videos about the processes should be preserved. figure 3 shows the process of field trip and site survey. iii) public participation beside for industry, water is also an essential resource for residents’ living. water withdraw and consumption of a npp in inland china would greatly impact on other water users. at the same time, discharge of wastewater threaten and may adversely affect public health. particularly, under accident conditions, middle or highlevel radioactive wastewater may be released to vicinal water bodies and cause disastrous effects on sources of drinking water. thus, public participation should be an important process and significant tool for tswm management. public’s opinions need to be taken in the procedures of decision making and final approval (or rejection). in the site, holding a hearing, a legal way of public participation, should be necessary. it provides an opportunity for interested people and organizations to express their opinions, participate in wra management, protect their interests, and provide decision support for approval authorities. iv) decision-making decision-making, involving report review and administrative approval, should be a crucial process of tswm management. for the first step of tswm, it should follow the e nvironmental i mpact a ssessment law of t he p eople's r epublic of c hina. as for the second step of tswm, it should include three phases. firstly, a related watershed authority should approve the outline of a wra report completion and the wra report primarily. secondly, the review committee should make a decision whether the report is qualified from the technical point of view. the review committee should work as a third-party organization to the construction unit and approval authorities, which makes the decisionmaking process scientific and democratic. thirdly, the decision-making process of administrative approval should be organized and carried out by mwrprc. it should follow the principle of “government-oriented” considering that it affects public’s interests. 4.4. reporting requirements of two-step water management for tswm approach, eia and wra reports are important considering that they reflect the impacts of the construction and operation of npps on water resources as well as their mitigation schemes. eia report compiling should comply with the environmental impact a ssessment l aw of t he p eople's republic of china and technical g uidelines f or e nvironmental impact a ssessment (mepprc, 2011). moreover, detailed requirements for completing a wra report for an inland npp project are presented as follows (table 2). i) general description in this part, general information related to the report needs to be described such as background information, published by atlantis press copyright: the authors 192 main purposes, major tasks, assessment scope, and target years. ii) executive summary in this part, basic information of the npp needs to be briefly presented, such as: (a) name, investment source, and scale, (b) site, floor space, land use, and implemental suggestions, (c) necessary approval documents by other related authorities, (d) water withdraw schemes including possible sources of water resources, water levels of intake water bodies, quantity of intake water, quality of intake water, temperature of intake water, water withdraw location, water withdraw inlet setting, and water consumption techniques and processes, and (e) wastewater discharge schemes including the radioactive concentration of wastewater, wastewater quantity, discharge methods, and discharge location and outlet setting. iii) current situation of water resources in local area this part should mainly include the following six aspects: (a) basic information of physical geography and social economy of local area, (b) quantity, dynamics, and temporal-spatial distribution of water resources, (c) water quality, the relevant database, and the assessment methods, (d) current situation of water consumption, (e) major problems of water consumption, and f) attached drawings and graphs. iv) water sources assessment this part should mainly include water quantity and quality assessment of water sources. reasonable structure and scale of economic development should be also proposed based on current situation of water resources in local area. water quality and the related requirements should be discussed. also, reasonable locations of water withdraw need to be justified and determined. surface water sources assessment should focus on reliability of water sources based on hydrological serial materials. in china, water sources of construction projects include surface water, groundwater and unconventional water. but for npps, groundwater and unconventional one shouldn’t be involved (even as auxiliary ones) considering of huge fresh water consumption. if surface water in site area is decision-makers and a review committee read tswm reports the decision-makers define the content and procedure of field trip and site survey the decision-makers and the review committee have a spot investigation in the site the decision-makers and the review committee collect site informations by observing, investigating, listening, questioning and discussing (the whole process is recorded) whether the describe of the report is reliable, the assessment is accurate, and the measures are feasible? yes end no the qualified agency modifies and improves the report fig. 3. the process of field trip and site survey. published by atlantis press copyright: the authors 193 inadequate, a construction unit should relocate the site rather than use groundwater or unconventional one as water resources. v) rationality analysis of water withdraw and consumption in this part, impacts of water withdraw and consumption on other water users, regional water resources allocation, and even regional water consumption structure need to be assessed. the following seven aspects should be contained, including: (a) water withdraw amount and its rationality analysis, (b) water balance analysis, (c) establishment of water consumption indicators based on assessment i ndicators and calculation methods f or industrial w ater (cj 42-1999), evaluating g uide f or water s aving e nterprises (gb/t 7119-2006) (gaqsiqprc, 2006) and other relevant standards, (d) rationality analysis of water consumption, (e) analysis of water-saving measures, and (f) analysis of the impact of surface water withdraw. according to the standards mentioned above, construction st andard of n pps (exposure draft) (china power engineering consulting group corporation, 2011), as well as design documents of inland ap1000 npps in china (cjwrc, 2009; table 2. requirements for completing tswm reports. step no. part requirements ⅰ 1 eia reports comply with the environmental impact a ssessment l aw of t he p eople's r epublic of china and technical guidelines for environmental impact assessment ⅱ 1 general description describing background information, main purposes, major tasks, assessment scope, and target years of the report 2 executive summary introducing basic information about the project 3 current situation of water resources in local areas analyzing basic information of water resources in local areas, current situation of water resources utilization 4 water sources assessment assessing the quantity and quality of the water sources 5 rationality analysis of water-draw and utilization evaluating impacts of water-draw and utilization on regional water resources allocation, other water users, and even regional water utilization structure (as for an inland npp with 4 ap1000, water-draw amount should be no more than 1.2 m3/s·gw, total water use amount should be no more than 55 m3/s·gw, water consumption amount should be no more than 1.0 m3/s·gw, and reuse rate of water resources should be no less than 97.5 %.) 6 impact analysis of wastewater discharge describing the wastewater discharge system and its components, assessing the rationality of the outlet setting, discussing the pollutants in discharged wastewater, analyzing impact of wastewater discharge on the receiving water bodies, and simulating the temporal variation of pollutant discharge and the spatial-temporal variations of pollutants (as for an inland npp with 4 ap1000, wastewater discharge amount should be no more than 0.35 m3/s·gw and the flow of the receiving river should be no less than 57.1m3/s.) 7 impact analysis of the project on the regional water resources and other water users analyzing the effects of water-draw and utilization on regional water resources allocation and water intake of other users, discussing the adverse impact of radioactive wastewater discharge on drinking water quality and other water users, and proposing necessary compensation measures preliminarily 8 water conservative measures proposing water conservative measures, and analyzing the applicability and feasibility of the engineering measures for emergency 9 mitigation schemes for the adverse impact on other water users presenting compensation measures in detail, estimating the mitigation effects and the final impacts on other water users quantitatively 10 conclusions giving a comprehensive evaluation on the schemes of water-draw, water utilization and wastewater discharge of the project values for tswm published by atlantis press copyright: the authors 194 wuhan university, 2009a; wuhan university, 2009b) and the usa (usnrc, 2009), water consumption indicators of inland npps with ap1000 reactors and the recommended values for wra are shown in table 3. as far as an inland npp with 4 ap1000 reactors (as the three pioneering ones in china) is concerned, water withdraw and consumption amounts are recommended as no more than 1.2 m3/s·gw for water withdraw, no more than 55 m3/s·gw for total water use, no more than 1.0 m3/s·gw for water consumption, and no less than 97.5 % for reuse rate of water resources. vi) impact analysis of wastewater discharge in this part, the wastewater discharge system and its components shall to be described. also, the rationality of the outlet setting needs to be assessed. the concentrations, total amounts, discharging standards, temporal and spatial variations, and compliance rates of pollutants need to be depicted. whether wastewater discharge meets the total allowable quota and the requirements of receiving water bodies also needs to be assessed. the quantitative and cumulative impacts of wastewater discharge on water resources under normal working conditions should be analyzed. the rationalities of low-level radioactive wastewater discharge scheme (under normal working conditions) as well as medium or high-level radioactive wastewater discharge scheme (under accident conditions) shall also be demonstrated. reasonable wastewater discharge schemes and suggestions for mitigation shall be proposed. considering that wastewater of a npp includes radioactive one and non-radioactive one, the former shall attract special attention and the latter may be controlled similar to that of a thermal power plant. particularly, impacts of medium and high-level radioactive wastewater discharge on water resources security under nuclear leakage accident conditions need to be evaluated. take an ap1000 reactor as an example, wastewater discharge amount should be no more than 0.35 m3/s·gw (see table 3). the design source terms of tritium and other liquids are 3.74×1013 bq/y and 9.46×1010 bq/y respectively. in this part, a safety factor of 1.2 can be adopted, so those for an npp with 4 ap1000 units are listed in table 4. china’s standards for radioactive pollutant discharge of inland npps also are also presented in table 4. according to the discharge amounts and standard values, the minimum flow of the receiving water is deduced and the recommended one is 57.1m3/s (see table 4). in addition, for riverside npps, outlets should be near main-flow areas of rivers, chutedischarge mode should be adopted, and discharging effluents into floodplains should be prohibited. besides annual discharge control, quarterly and monthly ones also need to be controlled. that’s to say, the discharge amounts of per quarter and per month should be no table 3. water utilization indicators of an inland npp with ap1000 reactors and recommended. no. indicator design values of npps values proposed in construction standard of npps (draft) recommended value for tswm taohuajiang npp dafan npp pengze npp vogtle npp 1 total water use amount (m3/s·gw) 54.35 60.33 54.25 54.07 — ≤55 2 water-draw amount (m3/s·gw) 1.24 1.27 1.16 0.96 1.2 ≤1.2 3 water consumption amount (m3/s·gw) 0.88 0.92 0.81 0.72 1.0 ≤1.0 4 reuse rate of water resources (%) 97.72 97.59 97.90 98.23 — ≥ 97.5 5 wastewater discharge amount (m3/s·gw) 0.36 0.35 0.36 0.24 — ≤0.35 table 4. wastewater discharge indicators for a npp with 4 ap1000 units in inland china. indicators radioactive pollutant value designed discharge amount of per unit (bq/y) (ontario power generation, 2009) liquid h-3 3.74×1013 sum of other liquids 9.46×109 discharge amount of a npp with a safety factor of 1.2 (bq/y) liquid h-3 1.80×10 14 sum of other liquids 4.54×1010 concentration of the receiving water downstream 1 km from the origin of sewage discharge (bq/l) (mep, general administration of quality supervision, inspection, and quarantine of the prc, 2011a & 2011b) liquid h-3 ≤100 sum of other liquids ≤1 required minimum flow of receiving water to meet the concentration standard of the pollutant (m3/s) liquid h-3 57.1 sum of other liquids 1.44 recommended value of minimum flow of receiving water (m3/s) 57.1 published by atlantis press copyright: the authors 195 more than half and one fifth of the approved annual one (mepprc, gaqsiqprc, 2011a). vii) impact analysis of the npp on the regional water resources and other water users in this part, the effects of water withdraw and consumption on regional water resources allocation as well as water intake of other users need to be analyzed. the adverse impact of radioactive wastewater discharge on drinking water quality and other water users shall be evaluated in detail. considering the large amount of water consumption of an inland npp, necessary compensation measures shall be proposed preliminarily for people and organizations those are affected. viii) water conservative measures in this part, water conservative measures need to be proposed and assessed. particularly, applicability and feasibility of the engineering measures for any emergencies shall to be presented. the construction unit and qualified agency shall follow principles of strengthening water reuse, reducing water consumption, and ensuring water security. ix) mitigation schemes for the adverse impact on other water users even if necessary measures are adopted, water withdraw and wastewater discharge would still affect the interests of other water users. in the part, compensation measures need to be proposed in detail. the mitigation effects of compensation measures and the final impacts on other water users need to be quantitatively evaluated. x) conclusions in this part, the entire assessment work needs to be summarized and a comprehensive evaluation on the schemes of water withdraw, water consumption and wastewater discharge of the power plant needs to be presented. this would provide a scientific support for wra administrative approval and water withdraw permission. in this part, the following key points need to be covered: (a) the attitude to construction and operation of the npp based on current situation of water resources in local area, (b) the water allocation plans and strategies of the corresponding watershed or region, (c) the guarantee rate of water withdraw sources, (d) the total water demand of the power plant as well as the rationalities of water consumption and saving potentials, (e) the amount, pollutant concentration and discharge mode of wastewater, (f) the impact of water withdraw and wastewater discharge on other water users and its conservative measures, and (g) suggestions for wra approval and supervision. take the three pioneering npps in inland china as examples. according to the current design, they will be located in the vicinity of inland water bodies in the yangtze river watershed. yangtze river is the largest river in china and significant for water supply to municipal, agricultural and industrial sectors. so in the wra reports, water sources, quantities of intake water, water use efficiencies, wastewater discharge, hydrodynamic conditions of receiving water bodies, adverse impacts on local water bodies and other water users, mitigation schemes should be the emphases. 5. enhancement of water management for nuclear power plants through tswm as for water management for nuclear power plants in inland china, it can be enhanced through tswm approach by the following ways. i) advancing involvement time with regard to current water management for other construction projects, eia and wra will be carried out when the site of the project is confirmed. but it is not suitable for npps in inland china. because once the site of a npp is confirmed, a huge amount of fund and manpower will be invested and lots of water-related issues will be identified. meanwhile, the total allowable quota of receiving water bodies, local water demands, and watershed management strategies is also confirmed. on one hand, it is difficult for assessors and water resources management departments to propose negative opinions on a confirmed site location for huge investments and manpower in preliminary phases. on the other hand, current situation of water resources may not be suitable and potential risks to water bodies should be also further identified. that is an important reason why the three pioneering plants are stagnant nowadays. take hunan taohuajiang npp for example, the total expense of the preliminary actions after siting was around 600 million dollars (li, 2012). nowadays, the huge money has been spent, but the wra management has not involved. so the inland npp has to be suspended considering its serious influence on water published by atlantis press copyright: the authors 196 resources and its probable nuclear leakage accidents (even with very extremely low probability). in future, water management for npps should be involved at the preliminary phases (planning and siting ones) of an inland npp. necessary regulations and policies need to be put forward and implemented to guarantee it. thus, the water resources management departments can have a full understand of the npp and give contemporaneous suggestions on the designs related to water resources, such as the siting of a plant, the distance between the plant and water bodies, the allowable pollution carrying capacity of the receiving river and so on. ii) strengthening the effectiveness of administrative approval the first step of tswm is required by the environmental i mpact assessment law of the p eople's republic of c hina and administrative approval of it is legal and effective. however, administrative approval of the second step needs to be strengthened. nowadays, local governments (including provincial, municipal and county ones) support inland npps strongly due to the huge stimulation of local economy. therefore, they and construction units are not quite willing to carry out wra considering that it may delay or reject npps and associated investments. meanwhile, the authority of water resources management departments, particularly watershed authorities, is inadequate in china. so in the future, the effectiveness of administrative approval of tswm needs to be further strengthened. specific laws should be legislated for wra and administrative approval of wra should be enforced at the preliminary phases of a npp project. in this way, if water resources condition in the local area is not appropriate for supporting the operation of an inland npp, water resources management departments could require the construction unit to adjust the original site. iii) improving the technical level of assessment reports till now, there are several technical issues need to be tackled. for example, (a) how to accurately determine the minimum safe distance between an inland npp and the nearby water bodies, (b) what are the typical transport and decay processes of radioactive substances in receiving water bodies, and (c) and what would be the cumulative effects of radioactive substances on aquatic organisms? such questions are complicated and need to be answered on an interdisciplinary basis (e.g., hydrology, radiology, and biology). at the same time, there is a lack of special technical guidelines and standards, which poses challenges to assessors. especially for the second step of tswm, it concerns several unprecedented and uncertain technical issues including those mentioned above. therefore, mwrprc and the related watershed authorities have supported a series of research efforts in studying the necessary standards, guidelines, and regulations. for example, research on analyzing the impacts of radioactive wastewater of npps on aquatic organisms has been carried out. based on it, appropriate models and monitoring methods are expected to be proposed. in the near future, a number of technical training programs should be provided for assessors to enhance their capabilities in conducting wra for inland npp projects. moreover, a series of management measures also need to be adopted. for example, water resources management departments should take reward and penalty measures, by which the superior qualified agencies will be encouraged while the inferior ones will be discarded. at the same time, technical exchange among qualified agencies needs to be carried out, which will strengthen the responsibility of assessors and improve their abilities. iv) enhancing the ability of review committees because inland npps is a kind of new projects in china, reviewers included in the existing expert databases may not be well familiar with all aspects of the construction and operation processes of inland npps. as for tswm, china’s environmental protection and water resources management departments should make efforts in enhancing the ability of reviewers. especially for the second step of tswm, it is necessary to invite nuclear experts to take part in the technical review due to the lack of experts majored in nuclear science and engineering. the cooperation between water resource professionals and nuclear ones (such as experts of national nuclear safety administration of the prc) needs be also strengthened. moreover, further communication between review committee and public (especially water users of receive water bodies) shall be developed considering that review committees have the responsibilities to reflect the public interest effectively and incorporated it into decision-making processes. published by atlantis press copyright: the authors 197 http://www.iciba.com/inadequate v) improving public participation as for tswm, mepprc (2008) has released implementing measures of public participation o f environmental i mpact a ssessment f or nuc lear p ower plants (exposure d raft), which gives general requirements, implementing regulations and organizational forms of public participation in different phases of npps for the first step of tswm. however, that of the second step of tswm should be improved considering of its close relationship to water resources and great public concerns. for designed receiving water bodies of radioactive wastewater in inland china, almost all of the lower reaches of them are with drinking water sources. radioactive wastewater discharge also has various adverse impacts on human health, water quality and aquatic ecosystems. in future, model prediction needs to be used to reflect the sensitivity of public concerns. public participation is an effective way to resolve water disputes, improve the effectiveness of decision-making, and enhance the democracy of wra management. considering that, water resources management departments shall increase the degree of public participation in wra through various ways. in the process of the second step of tswm, the information of a npp and its impacts on water resources and other water uses needs to be released in time. in the wra reports, the forms and approaches of public participation need to be specified. with the enhancement of the transparency of wra management, the public’s misunderstanding on any npps will be reduced. in the foreseeable future, posting notice in local areas, releasing notification in web sites, as well as holding hearings and colloquia shall be developed. through the above measures, the public can get necessary information conveniently. public opinions shall be taken into account when qualified agencies complete the chapters of impact analysis of a npp on the regional water resources and other water users. 6. conclusions nuclear power is an important energy source to deal with climate change and energy shortage. nowadays, a number of inland nuclear power plants (npps) are operated or planned to be constructed across the world. take china for example, construction of three pioneering inland npps (hunan taohuajiang npp, hubei dafan npp, and jiangxi pengze npp) may be started in a few more days. it poses great challenges to china’s water resources management due to large quantity of fresh water consumption and radioactive wastewater discharge. thus, it is desired that an effective water-management approach for such projects shall be conducted in china. a two-step watermanagement approach (tswm), proposed in this article, presents a comprehensive and feasible management approach of water withdraw, water consumption, wastewater discharge of such kind of construction projects. in this research, the flow process of tswm for npps in inland china was proposed. it should include eia management, entrusting the work of wra report completion, drafting an outline, a watershed authority approving the outline, report completion, the watershed authority approving the report preliminarily, site survey and hearing, review committee reviewing the report, final administrative approval, and regulating. in addition, it was put forward that main tasks of the management system should consist of carrying out eia management for a npp in inland china, proposing the requirements for report completion, organizing report review, carrying out administrative approval, and supervising the enforcement of approval documents. moreover, management tools of tswm for npps were also presented. it may contain constituting laws and standards, field trip and site survey, public participation, as well as decision-making. as far as the decisionmaking is concerned, “three-step process” was suggested to be taken besides eia management. the process may contain watershed authority making a decision whether the work outline is appropriate, a review committee making a decision whether the wra report is qualified, and mwrprc making a decision whether the water-related schemes proposed in the report is reasonable and feasible. furthermore, detailed requirements for completing tswm reports for an inland npp were also proposed. eia report compiling should comply with the relevant laws and technical guidelines. a wra report for an inland npp is proposed to cover general description, executive summary and conclusions. it also shall depict current situation of water resources and water sources, analyze the rationality and feasibility of water withdraw, assess water consumption and wastewater discharge of the project, evaluate the impact of water withdraw and wastewater discharge on water environment and other water users, and propose water conservative measures and mitigation schemes for adverse impacts. as far as published by atlantis press copyright: the authors 198 an inland npp with 4 ap1000 reactors (as the three pioneering ones in china) is concerned, water withdraw and consumption amounts are recommended as no more than 1.2 m3/s·gw for water withdraw, no more than 55 m3/s·gw for total water use, no more than 1.0 m3/s·gw for water consumption, no less than 97.5 % for reuse rate of water resources, and no more than 0.35 m3/s·gw for wastewater discharge. the flow of the receiving river of an inland npp with 4 ap1000 reactors should be no less than 57.1m3/s. suggestions for establishing the effective and feasible management for the new kind of projects were provided. tswm should be involved at the planning and siting phases of an inland npp. if water resources condition in the local area is not appropriate for supporting a plant operation, water resources management departments should have the right to negative the site. in the future, several researches on safe distance between a npp and water bodies, transport and decay processes of radioactive substances in receiving water bodies, and cumulative effects of radioactive substances on aquatic organisms should be carried out. special standards, guidelines, and regulations for the management system should be constituted. measures such as cooperation and communication between water resources governors and nuclear ones, technical training for assessors, invite nuclear experts to take part in the technical review, as well as reward and penalty measures should be also taken. in addition, public participation needs to be strengthened by the ways of specifying its forms and approaches, releasing information in time, posting notice, releasing notification, as well as holding hearings and colloquia. it is expected that tswm will become a powerful and effective management method to support sustainable consumption of water resources and security development of nuclear power industry. acknowledgements this research work was funded by ministry of education, special items fund of beijing municipal commission of education, national key scientific and technological projects of prc (2011zx06004-008) and the fundamental research funds for the central universities of prc (jb2014129). the authors gratefully acknowledge the financial support of the programs and agencies. references barrie p. co-benefits of large-scale renewables in remote australia: energy futures and climate change. rangeland journal, 33(4 si)（2011）315325. bezdek r. h. nuclear power prospects in the usa: the continuing problem of the waste issue. energy & environment, 20(3) （2009）375-385. china power engineering consulting group corporation. construction standard of nuclear power plants (exposure draft), in chinese , 2011. china's state council. state council’s decision on setting administrative permission for the projects with necessary administrative approval, in chinese, 2004. china's state council. management regulations of water withdraw permission and water resources fee collection, in chinese, 2006. china's state council. four decisions of state council concerning the development of nuclear power industry, in chinese, 2011. dai j. s., yi y.g. condition of inland nuclear power plant construction is basically ready. state grid news，2010-07-15, in chinese,2010. department of justice canada. canadian environmental assessment act, 2012. general administration of quality supervision, inspection, and quarantine of the prc, standardization administration of the prc. evaluating guide for water saving enterprises (gb/t 7119-2006), in chinese , 2006. hu j. t. president hu's speech at 2012 seoul nuclear security summit, in chinese, 2012. cjwrc. hunan taohuajiang nuclear power plant phase i project 2×1000mw units water resources assessment report, in chinese, 2009. iaea. convention on assistance in case of a nuclear accident or radiological emergency. , 1896. iaea. convention on early notification of a nuclear accident, 1986. iaea. convention on nuclear safety, 1994. iaea. iaea international fact finding expert mission of the fukushima dai-ichi npp accident following the great east japan earthquake and tsunami, 2011. jasmina v., dragoljub p. a., zorka v. environmental impact and cost analysis of coal versus nuclear power: the u.s. case. energy, 45(1) （2012）31-42. khamis i., kavvadias k.c. trends and challenges published by atlantis press copyright: the authors 199 toward efficient water management in nuclear power plants. nuclear engineering and design, 248 （2012）48-54. kopytko n., perkins j. climate change, nuclear power, and the adaptation-mitigation dilemma. energy policy, 39(1)（2011）318-33. lane r., dagher e., burtt j., thompson p. a. radiation exposure and cancer incidence (1990 to 2008) around nuclear power plants in ontario, canada. journal of environmental protection, 4(9) （2013） 888-913. li y. f. hunan taohuajiang nuclear power plant is suspended and the cost of the preliminary work is about 6 hundred million dollars. dongfang morning paper, 2012-02-20, in chinese, 2012. ma b., li j., han b., han x. discussions on experiences of environmental investigation around inland nuclear power plant. radiation protection bulletin, (5)(2010) 32-37. ministry of environmental protection of the prc. “the 12th five-year plan and 2020 vision of nuclear safety and radioactive pollution prevention and control” is approved by state council. available at: < http://haq.mep.gov.cn/gzdt/201210/t20121016_2384 21.htm>, in chinese, 2012. ministry of environmental protection of the prc, general administration of quality supervision, inspection, and quarantine of the prc. integrated wastewater discharge standard (gb 8978-1996), in chinese, 1996. ministry of environmental protection of the prc, general administration of quality supervision, inspection, and quarantine of the prc. technical requirements for discharge of radioactive liquid effluents from nuclear power plant (gb145872011) , 2011a. ministry of environmental protection of the prc, general administration of quality supervision, inspection, and quarantine of the prc. regulations for environmental radiation protection of nuclear power plant (gb6249-2011), 2011b. ministry of environmental protection of the prc, general administration of quality supervision, inspection, and quarantine of the prc. environmental quality standard for surface water (gb 3838-2002), in chinese, 2002. ministry of construction of the prc. assessment indicators and calculation methods for industrial water (cj 42-1999), in chinese,1999. ministry of water resources of the prc. claim on enhancing water resources assessment work of constructing projects, in chinese, 2002. ministry of water resources of the prc. management regulations of reviewing of water resources assessment reports for construction projects (trial), in chinese, 2003. ministry of water resources of the prc. guidelines of water resources assessment for construction projects (trial) (sl/z322-2005). beijing: china water power press, in chinese, 2005. ministry of water resources of the prc, national development and reform commission of the prc. management approaches of water resources assessment of construction projects, in chinese, 2002a. ministry of water resources of the prc, national development and reform commission of the prc. management approaches of water resources assessment qualification of construction projects, in chinese, 2002b. national people's congress of the prc. water law of the prc, in chinese, 2002. national people's congress of the prc. administrative permission law of the prc, in chinese, 2003. national development and reform commission of the prc. medium and long-term development plan of nuclear power (2005-2020), in chinese, 2007. nuclear energy institute. water use, electric power, and nuclear energy-a holistic approach to environmental stewardship, 2009. nuclear safety authority of france. act no. 2006-686 transparency and security in the nuclear field, 2006. ontario power generation. plant parameter envelope: concept and application for opg new nuclear at darlington, 2009. parliament of australia. water requirements of nuclear power stations , 2006. sovacool b. k. valuing the greenhouse gas emissions from nuclear power: a critical survey. energy policy, 36(8)（2008） 2950-2963. the diet's fukushima nuclear accident independent investigation commission. the final report of the fukushima nuclear accident independent investigation commission, 2012. united states nuclear regulatory commission. 10 cfr published by atlantis press copyright: the authors 200 part 51-environmental protection regulations for domestic licensing and related regulatory functions, 1984. united states nuclear regulatory commission. regulatory guide 4.7: general site suitability criteria for nuclear power stations, 1998. united states nuclear regulatory commission. vogtle electric generating plant license renewal application, 2009. united states nuclear regulatory commission. backgrounder on tritium, radiation protection limits, and drinking water standards, 2011a. united states nuclear regulatory commission. susquehanna steam electric station radioactive effluent release report and offsite dose calculation manual, 2011b. wang q. a comparison of the nuclear options for greenhouse gas mitigation in china and in the united states. energy policy, 37(7)（2009） 24872491. wang y. x., sun t. y. life cycle assessment of co2 emissions from wind power plants: methodology and case studies. renewable energy, 43（2012） 30-36. wen j. b. 2012 report on the government’s work, delivered at the fifth session of the eleventh, national people’s congress of the prc on march 5, 2012, in chinese, 2012. wuhan university of the prc. hubei xianning nuclear power plant water resources assessment report, in chinese, 2009a. wuhan university of the prc. jiangxi pengze nuclear power plant phase i project 2×1000mw units water resources assessment report, in chinese ,2009b. yan q., wang a. j., wang g. s. nuclear power development in china and uranium demand forecast: based on analysis of global current situation. progress in nuclear energy, 53(6) （2011）742-747. yang c. j. a comparison of the nuclear options for greenhouse gas mitigation in china and in the united states. energy policy, 39(6)（2011）30253028. yang g., huang w. j. the status quo of china's nuclear power and the uranium gap solution. energy policy, 38(2)（2010）966-975. zai r. m. three key factors that hinder the development of nuclear power plants in inland china. neteasenews, 2012-10-25, in chinese, 2012. zhang s. l., ding x. w., chen q. w. management of water resources argumentation for inland nuclear power plants of china. beijing: china water power press, 2011. zhong j. j. china restarts nuclear power construction and inland projects won’t be arranged in the next three years. beijing news, available at: < http://news.sina.com.cn/c/2012-1025/023925432152.shtml>, in chinese, 2012. zhou s., zhang x. l. nuclear energy development in china: a study of opportunities and challenges. energy, 35(11 si), 4282-4288, 2010. zhou y., rengifo c., chen p. p. is china ready for its nuclear expansion? energy policy, 39(2) （2011） 771-781. published by atlantis press copyright: the authors 201 abbreviation list abbreviation full name ap1000 advanced passive pressurized water reactor 1000 cdc china datang corporation cgc china guodian corporation cgnpc china guangdong nuclear power holding co. ltd chd china huadian corporation chg china huaneng group cjwrc changjiang water resources commission cnnc china national nuclear corporation cpic china power investment corporation cpr commercial power reactor eia environmental impact assessment ghg greenhouse gas htgr high temperature gas-cooled reactor iaea international atomic energy agency mepprc ministry of environmental protection of the prc mwrprc ministry of water resources of the prc ndrcprc national development and reform commission of the prc neaprc national energy administration of the prc npcprc national people's congress of the prc npp nuclear power plant usnrc united states nuclear regulatory commission prc people's republic of china tswm two-step water-management wra water resources assessment published by atlantis press copyright: the authors 202 1. introduction 2. the development of inland nuclear power industry in china 3. complexities of water resources assessment related to inland nuclear plants 4. development of two-step water-management approach for nuclear power plants 4.1. flow process of two-step water management 4.2. main tasks of two-step water management 4.3. management tools of two-step water management 4.4. reporting requirements of two-step water management 5. enhancement of water management for nuclear power plants through tswm 6. conclusions acknowledgements references 60 journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 60-61 doi: https://doi.org/10.2991/jracr.k.190703.001; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). introduction for volume 9, issue 2 this issue contains 6 papers. there are five contributions in english and one contribution in chinese with english abstracts. the papers can be divided into five topics: disaster response, earthquake risk assessment, ecological environment risk management, urban vulnerability assessment and agricultural weather insurance. concerning to disaster response, the paper“competency of the infantry troops of the nepalese army in disaster response”by shrijan bahadur malla, et al., analyses the nepalese army’s disaster response’s soft skill knowledge, technical skill knowledge and preparedness activities along with the perception of senior army officers. there are two papers earthquake risk assessment. the paper “seismic hazard model harmonization in tienshan area” by changlong li, mengtan gao, provides a comparison between central asia, the middle east hazard models and the chinese hazard result in tienshan area, based on reshaped seismogenic sources, recalculated related seismicity parameters, and calibrated ground motion models. this paper concluded that in the most areas of tienshan, the peak ground acceleration (pga) with 10% probability of exceedance in 50 years is around 0.2g~0.3g, and northern part of pakistan has the highest seismic hazard in tienshan area. the second paper “applicability of regional evaluation for rapid assessment models of earthquake disaster life loss -a case study of gansu province” by wen li, et al., uses the typical earthquake disaster life loss assessment models to verify and calculate the historical earthquake cases in the gansu province since 1966. the assessment accuracy and applicability of various models in the gansu province are studied by means of actual earthquake cases. results show that: (1) for ms<5.5 and 5.5≤ms≤6.0, the life loss assessment model based on population density proposed by badal, the life loss assessment model based on epicentral intensity proposed by liu jinlong, and the national earthquake disaster emergency assessment model are relatively accurate and basically included in the reasonable range; for ms>6.0, characteristics of the assessed regions, such as physical geography and social economy, should be analyzed to correct the model calculation results and to receive compre hensive assessment results, so as to support emergency decisions. (2) earthquake disaster life loss assessment models are regionalized and the assessment results are obviously regional. assessment results of the earthquakes occurred in hedong region are not accurate for hexi region of gansu province. in the studies on earthquake disaster life loss assessment models, development of the typical assessment models targeted on life loss characteristics of regional earthquakes has practical significance to actual earthquake emergency response and it is convenient to apply scientific research results in practical work. there is one paper in ecological environment risk management. the paper“evaluation of guizhou ecological civilization construction level based on intuitionistic fuzzy analytic hierarchy process” by fangping wang, et al., calculates the nine province each secondary index to the weight of a layer and target layer by using ifahp, the establishment of intuitionistic fuzzy judgment matrix and consistency test. comprehensive evaluation value is calculated and sorted. the empirical research shows that the ecological civilization construction level of guizhou province is relatively high, but the level distribution of all prefectures and cities is very uneven, and the difference between the highest score southeast guizhou and the lowest score liupanshui is 4.87. therefore, it is necessary to accelerate the construction of ecological civilization in various prefectures and cities, so as to improve the construction level of ecological civilization in the whole province and promote regional economic development. there is also one paper in urban vulnerability assessment. the paper “spatial and temporal patterns of urban vulnerability in guangzhou”by bo tang, et al., develops an urban vulnerability index system from four aspects of population vulnerability, economic vulnerability, social vulnerability, and ecological vulnerability. the spatial and temporal patterns of urban vulnerability in guangzhou were evaluated by a comprehensive index model and gis analysis from three time sections （2005、2010、2014）. the results indicated that: (1) economic density, population density, and per capita gdp are the main factors affecting the vulnerability of guangzhou. (2) the population journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 60–61 60 61 introduction for volume 9, issue 2 this issue contains 6 papers. there are five contributions in english and one contribution in chinese with english abstracts. the papers can be divided into five topics: disaster response, earthquake risk assessment, ecological environment risk management, urban vulnerability assessment and agricultural weather insurance. concerning to disaster response, the paper“competency of the infantry troops of the nepalese army in disaster response”by shrijan bahadur malla, et al., analyses the nepalese army’s disaster response’s soft skill knowledge, technical skill knowledge and preparedness activities along with the perception of senior army officers. there are two papers earthquake risk assessment. the paper “seismic hazard model harmonization in tienshan area” by changlong li, mengtan gao, provides a comparison between central asia, the middle east hazard models and the chinese hazard result in tienshan area, based on reshaped seismogenic sources, recalculated related seismicity parameters, and calibrated ground motion models. this paper concluded that in the most areas of tienshan, the peak ground acceleration (pga) with 10% probability of exceedance in 50 years is around 0.2g~0.3g, and northern part of pakistan has the highest seismic hazard in tienshan area. the second paper “applicability of regional evaluation for rapid assessment models of earthquake disaster life loss -a case study of gansu province” by wen li, et al., uses the typical earthquake disaster life loss assessment models to verify and calculate the historical earthquake cases in the gansu province since 1966. the assessment accuracy and applicability of various models in the gansu province are studied by means of actual earthquake cases. results show that: (1) for ms<5.5 and 5.5≤ms≤6.0, the life loss assessment model based on population density proposed by badal, the life loss assessment model based on epicentral intensity proposed by liu jinlong, and the national earthquake disaster emergency assessment model are relatively accurate and basically included in the reasonable range; for ms>6.0, characteristics of the assessed regions, such as physical geography and social economy, should be analyzed to correct the model calculation results and to receive compre hensive assessment results, so as to support emergency decisions. (2) earthquake disaster life loss assessment models are regionalized and the assessment results are obviously regional. assessment results of the earthquakes occurred in hedong region are not accurate for hexi region of gansu province. in the studies on earthquake disaster life loss assessment models, development of the typical assessment models targeted on life loss characteristics of regional earthquakes has practical significance to actual earthquake emergency response and it is convenient to apply scientific research results in practical work. there is one paper in ecological environment risk management. the paper“evaluation of guizhou ecological civilization construction level based on intuitionistic fuzzy analytic hierarchy process” by fangping wang, et al., calculates the nine province each secondary index to the weight of a layer and target layer by using ifahp, the establishment of intuitionistic fuzzy judgment matrix and consistency test. comprehensive evaluation value is calculated and sorted. the empirical research shows that the ecological civilization construction level of guizhou province is relatively high, but the level distribution of all prefectures and cities is very uneven, and the difference between the highest score southeast guizhou and the lowest score liupanshui is 4.87. therefore, it is necessary to accelerate the construction of ecological civilization in various prefectures and cities, so as to improve the construction level of ecological civilization in the whole province and promote regional economic development. there is also one paper in urban vulnerability assessment. the paper “spatial and temporal patterns of urban vulnerability in guangzhou”by bo tang, et al., develops an urban vulnerability index system from four aspects of population vulnerability, economic vulnerability, social vulnerability, and ecological vulnerability. the spatial and temporal patterns of urban vulnerability in guangzhou were evaluated by a comprehensive index model and gis analysis from three time sections （2005、2010、2014）. the results indicated that: (1) economic density, population density, and per capita gdp are the main factors affecting the vulnerability of guangzhou. (2) the population vulnerability is considerably higher for the central group cities than for the peripheral group cities. the economic vulnerability reflects a crisscross pattern, consisting of a central group and a peripheral group. the social and ecological vulnerabilities are higher for the peripheral group than for the central group. (3) the comprehensive vulnerability has a layered structure, with a high vulnerability in the inner ring layer, low vulnerability in the middle layer, and medium vulnerability in the outer layer. there is one paper in agricultural weather insurance. the paper“summary of researches on basis risk in weather index insurance” by wang yueqin, et al., summarizes researches on the theories and practices of weather index insurance at home and abroad, and systematically comments on the basis risk’s definition, formation mechanism, quantitative methods and management measures. in order to provide useful ideas and references for the development of china's weather index insurance and the future research on basis risk. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca atlantis press journal style multicriteria decision aid applications to support risk decisions in the marine environment: locating suitable transshipment areas dimitrios i. stavrou, nikolaos p. ventikos national technical university of athens, school of naval architecture and marine engineering, laboratory for maritime transport (maritime risk group), 9 iroonpolytechniou str. zografou 15773, greece, e-mail: dstaurou@mail.ntua.gr, niven@deslab.ntua.gr abstract the aim of this paper is to develop an mcda model to support decision makers in the marine environment. the utastar method is employed for the selection of the best area for ship-to-ship transfer of cargo. the method relies on the hypothesis that both the actions and the corresponding criteria interact with each other over time, constructing and formulizing the decision aid model. the use of real data for the tuning of the model parameters may lead to the optimal compatibility between model and decision-maker cognition. keywords: multicriteria decision aid models, ship-to-ship transfer of cargo, aggregation–disaggregation methods, global criterion model. 1. introduction shipping is a risky business as well as a risky operation due to different issues involving both operational (ballis and stathopoulos, 2003) and trading factors (roumboutsos et al. 2004); the operational factors are related to the complex and frequently hostile sea environment, whereas the trading factors refer to the primary target of the shipping companies, which is to maximize their profits in a highly competitive and demanding market (sambracos and ramfou, 2001). an interesting and very promising way to deal with decision support problems in the marine environment comes from the employment of different multicriteria decision aid (mcda) methodologies. in most cases, mcda methods are preference modelling approaches that are based on binary relations of the alternatives under the outranking relation law or ordinal regression approaches, which employ methodologies based on additive value models (lopez et al. 2008). the modelling of the decision makers’ (dms’) preferences in outranking relations is achieved by the direct interrogation of the dm and the analyst, in which the dm must determine the weights of the evaluation criteria, as well as the preference, indifference, and veto thresholds (zopounidis, 2001). there are several representative works on this approach (roy, 1985; roy and bouyssou, 1993). a plethora of related examples can be found in the work of siskos and tsotsolas (2015). on the other hand, the additive value models refer to the order of preference of a set of actions or alternatives followed by a consistent family of criteria (greco et al., 2008, 2011; hurson and siskos, 2014). there exist many different approaches when referring to the additive value functions. for example, multiattribute utility theory (maut) is based on the construction of a global utility function that corresponds to the dm preferences (keeney and raiffa, 1976; farquhar, 1984; fishburn, 1967; figueira et al. 2005). in journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 3 received 13 february 2017 accepted 26 february 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). maut theory, a direct interrogation process is employed to elicit information from the dm concerning the trade-offs among the conflicting criteria, attributes, or points of view with the objective of constructing the global preference model in the form of a utility function that the dm implicitly uses to make decisions (zopounidis, 2001). thus, maut offers a consistent and analytical evaluation of the alternative actions with regard to the corresponding criteria. alternatively, additive value functions can be built by employing the global criterion method (siskos, 2008). this method is based on the aggregation– disaggregation or, in other words, analytical approach that refers to the analysis (disaggregation) of the global preferences (judgement policy) of the dm to identify the criteria aggregation model that underlies the preference result (doumpos and zopounidis, 2001) for the development of one or more additive value functions. thus, the analytical approach theory relies on the assumption that the knowledge of the mechanism that dms have in mind can be extracted and used to rank different alternatives that are governed by the same principles or rules. the process is applied to a predefined set of actions or alternative solutions 𝐴𝑅. at first, an initial order of preference integrating both real and imaginary alternatives is built. the ultimate purpose is to construct a model that consists of the marginal functions of the selected criteria by applying special lp techniques. the ranking of the actions obtained through these functions on 𝐴𝑅 should as consistent as possible with the initial ranking of actions (lakiotaki et al. 2009). finally, the extracted model is extrapolated to the rest of the alternative set of actions (spyridakos and yannacopoulos, 2015). the main representative of the analytical approach comes from the uta (utility additives) family, which was initially proposed by jacquet-lagreze and siskos (1982, 2001). the primary goal of the uta methods is the inductive inference of one or more additive value functions using a preference order of reference actions. the uta methods belong to the wide family of analytic global criterion methods with monotonic regression. this paper focuses on the implementation of the utastar method, which was initially proposed by siskos and yannacopoulos (1985) as an improved version of the original uta method. in the original uta method, each alternative action comes with a unitary defined error (grigoroudis et al. 2004), which corresponds to the correction of the value of each criterion in order, the additive value function, to verify the order of preference of the reference actions. the uta method finally calculates the marginal value function of each criterion as a result of the solving a linear problem (lp) with the objective function, the error function, and the constraints that arise from the nature of the problem. the utastar method aims to remedy the weakness of the original uta method by minimizing, through the error function, the total dispersion of the points that corresponds to the value of each criterion with regard to the reference actions. more specifically, the utastar introduces the meaning of the double error function (𝜎+(𝑎𝑘) and 𝜎−(𝑎𝑘)), taking into account situations of both underestimation and overestimation of the marginal value function (matsatsinis, 2010). to examine the effectiveness of the utastar method for decision models from the marine sector, the method is applied to the problem of choosing a suitable area for ship-to-ship (sts) transfer of cargo. thus, the aim of this paper is to develop an mcda model based on the analytical global criterion method with monotonic regression for the location of the most suitable sts transfer area. to achieve this goal, the rest of the paper is organized as follows: first, the current context of the mcda additive value functions is presented. next, in section 2, the link between the analytical approaches and the utastar method is properly described, and in section 3, the method is applied for the selection of the most appropriate sts transfer location. in section 4, the results are presented and discussed. finally, section 5 concludes the paper. 2. from the analytical approach to the utastar methodology when dealing with multi-attribute decision problems by applying roy’s (1985) general methodological framework, the majority of the developed decision models are based on the axiom that the final decision is the result of the evaluation of the different actions or alternatives followed by a consistent family of criteria. hence, there is a type of inductive inference process in which the general conditions of the problem determine the final decision. this “cause and effect” approach was reconsidered by jacquet-lagreze and siskos (1982), who stated that the final decision can be the result of knowledge of the cognitive mechanism by which the dms decide instead of an inductive inference process. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 4 thus, the analyst can model the dms’ way of thinking and apply this knowledge to make decisions regarding other alternatives from the same set of actions. this philosophy is the fundamental principle of aggregation– disaggregation or simply analytical methods where the decision and criteria interact with each other over time and co-construct the decision model (siskos, 2008). according to the theory of analytical methods, the first step is to collect the data relevant to the decision problem and then to determine the actions and the corresponding criteria. next, a model is constructed based on the dm’s order of preference for the predefined actions. finally, the constructed model is compared with the order of preference of actions and, for absolute relevance, the constructed model is extrapolated to the rest of the actions; otherwise the model is properly corrected to fit the initial order of preference. the most representative family of analytical methods is the uta family. a flowchart of the analytical approach is shown in figure 1. fig. 1. the concept of the analytical approach through the uta methodologies includes a given order of preference of the selected actions which is used as a mean to determine the corresponding value functions during the disaggregation process. next comes the aggregation process where the extracted model is extrapolated to the rest of the alternative set of actions. in uta methods, the analyst takes for granted the preferences of the dm regarding a given set of reference actions 𝛢𝑅. the preference actions could be either real, taken from the original set of actions, or imaginary, which means that they are built on the selected criteria. given the order of preference of the selected actions, the uta method applies special techniques of linear programming to determine the corresponding value functions, through which the criteria are aggregated into a global criterion per action. that is why this kind of modelling is also known as the “global criterion model”. the value function 𝑢 of a given set of actions 𝛢 = {𝑎1, 𝑎2, … , 𝑎𝑚} , followed by a consistent family of criteria 𝑔1, 𝑔2, … , 𝑔𝑛, is expressed by the relationship: 𝑢(𝒈) = 𝑢(𝑔1, 𝑔2, … , 𝑔𝑛) (1) 𝒈(𝑎) → 𝑢[𝒈(𝑎)] (2) where 𝑔𝑖∗ 𝑎𝑎𝑎 𝑔𝑖 ∗ , correspond to the worst and best values of the criterion 𝑔𝑖, and 𝑢[𝒈(𝑎)] is a real number that is known as the value of action α. thus, the global value of each action can be expressed by an additive value function of the following form: 𝑢(𝒈) = �𝑢𝑖(𝑔𝑖) 𝑛 𝑖=1 (3) under the normalization constraints, �𝑢𝑖(𝑔𝑖 ∗) = 1 𝑛 𝑖=1 (4) 𝑢𝑖(𝑔𝑖∗) = 0, ∀𝑖 = 1,2, … (5) where 𝑢𝑖, 𝑖 = 1,2, … 𝑎 represent monotonic non-descent functions 𝑔𝑖 known as marginal value functions. in order to implement a uta model, two conditions should be in force: preference independence among the selected criteria and the property of consistency or monotony, which means that: when 𝑢[𝒈(𝑎)] > 𝑢[𝒈(𝑏)], action α is preferred to b (6) when 𝑢[𝒈(𝑎)] = 𝑢[𝒈(𝑏)], dm is indifferent between action α and b (7) the utastar is an improved version of the original uta method (jacquet-lagreze and siskos, 1982). it refers to the evaluation of a set of actions 𝐴 = {𝑎1, 𝑎2, … , 𝑎𝑛} according to a consistent family of criteria 𝑔𝑖 = {𝑔1, 𝑔2, … , 𝑔𝑚} based on an order of preference of actions given by the dm. during the model implementation, an additive value function is constructed from the selected criteria with the objective of being consistent with the initial order of preference. the improvement of the utastar method comes from the introduction of the meaning of double error to deal with underestimations (which is also dealt by the original uta method) as well as overestimations of the journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 5 marginal functions. another innovative characteristic of the utastar is the transformation of the partial values of each criterion scale of the reference actions through equidistant spaces 𝑤𝑖𝑖 to facilitate the resolution of the lp (siskos, 2008). a detailed description of the utastar method can be found in the work of siskos (2008) or matsatsinis (2010). in brief, the utastar consists of the following four successive steps: step 1: the actions and the corresponding criteria are determined. discretized criterion scales are made and the global value of the actions 𝑢[𝒈(𝑎𝑘)], 𝑘 = 1,2, … , 𝑚 is initially expressed through the use of the marginal value functions 𝑢𝑖(𝑔𝑖). next, the transformation 𝑤𝑖𝑖 of each value function is conducted according to the relationships: 𝑤𝑖𝑖 = 𝑢𝑖�𝑔𝑖 𝑖+1� − 𝑢𝑖�𝑔𝑖 𝑖� ≥ 0, (8) ∀𝑖 = 1,2 … 𝑎 𝑎𝑎𝑎 𝜉 = 1,2, … 𝛼𝜄 − 1 𝑢𝑖(𝑔𝑖 1) = 0, ∀𝑖 = 1,2 … 𝑎 (9) 𝑢𝑖�𝑔𝑖 𝑖� = �𝑤𝑖𝑖, 𝑖−1 𝑖=1 (10) ∀𝑖 = 1,2, … 𝑎 𝑎𝑎𝑎 𝜉 = 2,3, … 𝛼𝑖 − 1 step 2: the error functions 𝜎+ and 𝜎− of the underestimation and overestimation of the preference actions accordingly are integrated to each value function and the differences between the pairs (first, second), (second, third), and so on of the order of preference are written accordingly: 𝛥(𝛼𝑘, 𝑎𝑘+1) = 𝑢[𝑔(𝑎𝑘)] − 𝜎+(𝑎𝑘) + 𝜎−(𝑎𝑘) − [𝑢[𝑔(𝑎𝑘+1)] − 𝜎+(𝑎𝑘+1) + 𝜎−(𝑎𝑘+1)] (11) step 3: the following lp is solved: [min]𝑧 = �[𝑢[𝜎+(𝑎𝑘) + 𝜎−(𝑎𝑘)] 𝑚 𝑘=1 (12) under the constraints: 𝛥(𝛼𝑘, 𝑎𝑘+1) ≥ 𝛿, 𝑤ℎ𝑒𝑎 𝛼𝑘 ≻ 𝑎𝑘+1 (13) 𝛥(𝛼𝑘, 𝑎𝑘+1) = 0, 𝑤ℎ𝑒𝑎 𝛼𝑘 ∼ 𝑎𝑘+1 (14) � � 𝑤𝑖𝑖 𝑎𝑖−1 𝑖=1 𝑛 𝑖=1 = 1 (15) 𝑤𝑖𝑖 ≥ 0, 𝜎+(𝑎𝑘) ≥ 0 (16) 𝜎−(𝑎𝑘) ≥ 0, ∀𝑖, 𝑗 𝑎𝑎𝑎 𝑘 (17) step 4: the existence of multi-optimal or semi-optimal solutions of the relative lp is checked through the calculation of the weighted average of the additive value functions that maximize the following objective functions. when a lack of robustness occurs, an additional constraint is applied to the existing ones of the previous hyper-polyhedron, which is: � [ 𝑎∈𝐴𝑅 𝜎+(𝑎) + 𝜎−(𝑎)] ≤ 𝑧∗ + 𝜀 (18) where 𝑧∗ is the optimal error value of step 3, and 𝜀 corresponds to a very small positive number (or zero). the new linear program should give n new solutions that are able to optimize, in the new hyperpolyhedron, the linear functions: 𝑢𝑖(𝑔𝑖 ∗) = � 𝑤𝑖𝑖 𝑎𝑖−1 𝑖=1 , ∀𝑖 = 1,2 … 𝑎 (19) if the stability of the model is adequate, the final decision corresponds to the mean utility function of n meta-optimal solutions. 3. use of utastar to choose sts transfer area according to the problem, a large oil cargo is to be delivered from the persian gulf to a final destination somewhere in north europe (point b of fig. 2). for both operational and financial reasons, the stakeholders have decided that the oil will be carried in parcels by suezmax tankers from the persian gulf (point a of fig. 2) through the suez canal with the objective of delivering each parcel to a very large crude carrier (vlcc) tanker by conducting an sts transfer operation. as the stakeholders now have to decide on the most suitable sts transfer area, there are four alternative locations under consideration: cyprus (cr), crete (k), malta (m), and gibraltar (g). information regarding the special characteristics of a sea area journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 6 suitable for sts transfer operations can be found in imo (2010) and ocmf (2013). after careful examination of the alternative locations, which is beyond the scope of this paper, four different categories of criteria have been chosen to evaluate each sts transfer area: operational, economic, environmental, and safety/security criteria. operational criterion: this criterion refers to the operational potentialities that each location can offer to fig. 2. problem definition of the sts transfer area. the transfer operation. for example, it includes the different ways (stationary or under way) in which the transfer operation can be conducted, the familiarization of the local authorities with the facilitation of the sts transfer process, the relative experience of the sts transfer provider who is responsible for providing the person in overall advisory control (poac), the relevant sts transfer equipment, and finally the proximity of the facilities to provide critical equipment in case of emergency, such as oil pollution. the operational criterion takes values from zero, which refers to the worst operational conditions, where no sub-criterion is satisfied, to four, which refers to the best option, where all sub-criteria are satisfied. finally, this is a positive qualitative criterion, which means that the larger the value of the criterion, the higher the value of the action for the stakeholder. economic criterion: this consists of the subcriteria that are relevant to the expenses that the stakeholder should cover to conduct the sts transfer operation in each suggested area. for example, it refers to the cost of the fees the stakeholder must pay to the authorities, the total cost of fuels and personnel for the transition to the sts transfer area, the payment for the sts provider services, and other additional operational costs. in order to evaluate the economic criteria, it is assumed that each economic sub-criterion makes a weighted contribution to the total cost of the sts transfer. the economic criterion is a quantitative criterion and is negative, which means that the lower the value, the more suitable is the location for the transfer and vice versa. environmental criterion: this is related to the prevailing weather or sea conditions that may affect the success of the operation, including the existence of sufficient room in which to conduct the transfer (especially transfers carried out under way), the traffic density of the proposed transfer area (too high a density may lead to early transfer suspension), the proximity of sensitive or protected areas, and finally the presence of a sheltered environment able to protect vessels from the forces of nature. this criterion is a qualitative criterion and is given by an empirical equation explained in detail in the work of stavrou et al. (2016). finally, the criterion has a negative meaning, which means that the lower the value, the more suitable is the location for the transfer. to evaluate the environmental criterion, it is assumed that each environmental sub-criterion makes a weighted contribution to the total cost of the sts transfer. safety and security criterion: this criterion refers to the geo-political status of the proposed transfer area, including the potential for unpredictable situations due to unstable geo-political relations between nearby nations. this criterion also accounts for the possibility of terrorist activity that may endanger crew, vessel, and cargo safety. a further factor is historical data regarding incidents or accidents within the area under consideration. it is a qualitative criterion and has a positive meaning, which means that the higher the value, the more suitable is the location for the transfer. the summarized data for each location according to the selected criteria are shown in table 1. table 1. multi-criteria evaluation of the proposed sts locations criterion operational environmental economic safety/security scale (0–4) (0–100) ($ × 103) (0–2) sense positive negative negative positive cyprus 2 44 31.7 1 crete 0 48 25.3 2 malta 4 59 15.9 2 gibraltar 3 38 19.2 2 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 7 after a short discussion with the stakeholder, the analyst proceeds with the construction of four virtual actions or reference profiles. for example, regarding the question of choosing by instinct one of the four different alternatives, the stakeholder chooses gibraltar due to its better environmental profile. the second choice of the stakeholder is malta due to its low cost, and so on. the order of preference according to the stakeholder’s cognition is shown in table 2, while table 3 depicts the corresponding discretized criterion scales. table 3: discretized criterion scales criterion scale low boundary value operational [𝒈𝟏∗, 𝒈𝟏 ∗ ] = [0, 2, 4] u1(0) = 0 environmental [𝒈𝟐∗, 𝒈𝟐 ∗ ] = [59, 53, 47, 41, 35] u2(59) = 0 economic [𝒈𝟑∗, 𝒈𝟑 ∗ ] = [32, 26, 20, 14] u3(32) = 0 safety/security [𝒈𝟒∗, 𝒈𝟒 ∗ ] = [0, 1, 2] u4(0) = 0 from the implementation of the first step of the algorithm, the value function of each action is calculated, applying the linear regression technique whenever needed. to do so, the following boundary conditions are also taken into account: 𝑢1(0) = 𝑢2(59) = 𝑢3(32) = 𝑢4(0) = 0. thus, the value function of each action is given by the following equations: 𝑢[𝑔(𝐴)] = 0.5𝑢1(4) + 0.5𝑢1(2) + 0.5𝑢2(41) + 0.5𝑢2(35) + 0.83𝑢3(20) + 0.17𝑢3(14)+𝑢4(2) 𝑢[𝑔(𝐵)] = 𝑢1(4) + 0.33𝑢3(20) + 0.67𝑢3(14) + 𝑢4(2) 𝑢[𝑔(𝛤)] = 𝑢1(2) + 0.5𝑢2(47) + 0.5𝑢2(41) + 0.83𝑢3(26) + 0.17𝑢3(20) + 𝑢4(1) 𝑢[𝑔(𝛥)] = 𝑢1(0) + 𝑢2(48) + 𝑢3(32) + 𝑢4(2) next, the transformation of equation (1) is applied: 𝑢[𝑔(𝐴)] = 𝑤11 + 0.5𝑤12 + 𝑤21 + 𝑤22+ 𝑤23 + 0.5𝑤24 + 𝑤31 + 𝑤32 + 0.17𝑤33 + 𝑤41 + 𝑤42 𝑢[𝑔(𝐵)] = 𝑤11 + 𝑤12 + 𝑤31 + 𝑤32 + 0.67𝑤33 + 𝑤41 + 𝑤42 𝑢[𝑔(𝛤)] = 𝑤11 + 𝑤21 + 𝑤22+0.5 𝑤23 + 𝑤31 + 0.17𝑤32 + 𝑤41 𝑢[𝑔(𝛥)] = 𝑤21 + 0.83𝑤22 + 𝑤41 + 𝑤42 according to the second step, the differences between the corresponding pairs are calculated according to the order of preference of the actions: 𝛥(𝛢, 𝛣) = 0.5𝑤12 − 0.67𝑤21−0.67𝑤22 − 𝑤23 − 0.5𝑤24 − 0.33𝑤31 − 0.33𝑤32 + 0.5𝑤33 − 𝜎𝛢+ + 𝜎𝛢+ + 𝜎𝛣+ − 𝜎𝛣+ 𝛥(𝛣, 𝛤) = 𝑤11 + 0.5𝑤12 + 0.17𝑤22 + 𝑤23 + 0.5𝑤24 + 𝑤31 + 𝑤32 + 0.17𝑤33 − 𝜎𝛣+ + 𝜎𝛣+ + 𝜎𝛤+ − 𝜎𝛤+ 𝛥(𝛤, 𝛥) = −𝑤11 − 0.17𝑤22 − 0.5𝑤23 − 𝑤31 − 𝑤32 + 𝑤42 − 𝜎𝛤+ + 𝜎𝛤+ + 𝜎𝛥+ − 𝜎𝛥+ in the third step, the lp is solved, taking into account a threshold value of δ = 0.01. the last row of the corresponding table shows the optimal value of the relative weights. the solution of the lp gave a feasible solution but not a unique one. for that reason, it is necessary to carry out a robustness analysis in order to determine the corresponding solutions that verify the problem. to achieve this goal, the objective function is transformed per equation (17). according to the results, z = 0, and thus the robustness analysis refers to those solutions that maximize the weight of each criterion (ε = 0). as all errors have zero values, they can be omitted and thus a new lp arises where the four objective functions are shown. table 2: adapted table for the utastar implementation. criterion operational environmental economic safety/ security order of preference scale (0–4) (+) (0–100) (–) ($ × 103) (–) (0–2) (+) a 3 38 19 2 1 b 4 59 16 2 2 c 2 44 25 1 3 d 0 48 32 2 4 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 8 4. results and discussion the final solution of the lp corresponds to the marginal functions of table 7, while the functions are depicted in figure 3. the grey lines correspond to the initial solution of the lp, while the black line shows the final solution after the meta-optimization process. according to the results, the order of preference of the shareholder is verified: 𝑢[𝒈(𝐴)] = 0.875, 𝑢[𝒈(𝐵)] = 0.745, 𝑢[𝒈(𝛤)] 𝑢[𝒈(𝛥)] as the model verifies the initial preference order, it can be further extrapolated to the evaluation of the rest of the selected actions. for example, in the case of malta, the global value function will be: 𝑢[𝒈(𝑀𝑎𝑀𝑀𝑎)] = 𝑢1(4) + 𝑢2(59) + 𝑢3(15.9) + 𝑢4(2) = 0.75 thus, the final ranking is as shown in table 8. in this paper, the order of preference of the reference actions emerged as a result of the interaction of the dm with the analyst. from the dm–analyst interaction, four fictitious actions emerged and were used to determine the corresponding marginal value functions. the results verified the initial order of preference and the value functions were used for the ranking of the rest of the actions. the interaction between dm and analyst was an table 4: table of the lp. 𝒘𝟏𝟏 𝒘𝟏𝟐 𝒘𝟐𝟏 𝒘𝟐𝟐 𝒘𝟐𝟑 𝒘𝟐𝟒 𝒘𝟑𝟏 𝒘𝟑𝟐 𝒘𝟑𝟑 𝒘𝟒𝟏 𝒘𝟒𝟐 errors 𝝈+ 𝒂𝒂𝒂 𝝈− 0 -0.5 0 0 0.5 0.5 0 0 0.5 0 0 -1 1 1 -1 0 0 0 0 ≥ 0.01 0 0.5 0 0 -0.5 -0.67 0 0.83 0.17 0 0 0 0 -1 1 1 -1 0 0 ≥ 0.01 1 0 1 1 1 0.67 1 0.17 0 0 1 0 0 0 0 -1 1 1 -1 ≥ 0.01 0 -0.5 0 0 0.5 0.5 0 0 0.5 0 0 0 0 0 0 0 0 0 0 = 1 0.09 0.353 0.109 0.08 0.01 0.11 0 0 0.12 0.132 0 0 0 0 0 0 0 0 0 z table 5. lp of meta-optimization for the problem of the sts transfer area location. 𝒘𝟏𝟏 𝒘𝟏𝟐 𝒘𝟐𝟏 𝒘𝟐𝟐 𝒘𝟐𝟑 𝒘𝟐𝟒 𝒘𝟑𝟏 𝒘𝟑𝟐 𝒘𝟑𝟑 𝒘𝟒𝟏 𝒘𝟒𝟐 0 -0.5 0 0 0.5 0.5 0 0 0.5 0 0 ≥ 0.01 0 0.5 0 0 -0.5 -0.67 0 0.83 0.17 0 0 ≥ 0.01 1 0 1 1 1 0.67 1 0.17 0 0 1 ≥ 0.01 1 1 1 1 1 1 1 1 1 1 1 = 1 1 1 0 0 0 0 0 0 0 0 0 [max] 𝑢1(𝑔1∗) 0 0 1 1 1 1 0 0 0 0 0 [max] 𝑢2(𝑔2∗) 0 0 0 0 0 0 1 1 1 0 0 [max] 𝑢3(𝑔3∗) 0 0 0 0 0 0 0 0 0 1 1 [max] 𝑢4(𝑔4∗) after solving the meta-optimization lp, the following solutions are extracted: table 6: robustness analysis and the final solution of the meta-optimization problem. 𝒘𝟏𝟏 𝒘𝟏𝟐 𝒘𝟐𝟏 𝒘𝟐𝟐 𝒘𝟐𝟑 𝒘𝟐𝟒 𝒘𝟑𝟏 𝒘𝟑𝟐 𝒘𝟑𝟑 𝒘𝟒𝟏 𝒘𝟒𝟐 [max] 𝒖𝟏(𝒈𝟏∗ ) 0.96 0.02 0 0 0.02 0 0 0 0 0 0 [max] 𝒖𝟐(𝒈𝟐∗ ) 0 0.02 0 0 0.02 0.96 0 0 0 0 0 [max] 𝒖𝟑(𝒈𝟑∗ ) 0 0 0 0 0.01 0 0.31 0.68167 0 0 0 [max] 𝒖𝟒(𝒈𝟒∗ ) 0 0 0 0 0.01 0 0.02 0 0 0.96 0.02 meta-optimal solution 0.24 0.01 0 0 0.015 0.24 0.08 0.17042 0 0.24 0 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 9 essential part that determined the nature of the obtained results. table 8: final ranking of the corresponding alternatives for the selection of the sts transfer area. the implementation of the utastar method demonstrates that the method has the capability to deal adequately with criteria of qualitative origins. this capability is very important especially when dealing with problems from the marine sector, where the need to apply qualitative scales in combination with qualitative ones is often unavoidable. another important observation comes from the step of the meta-optimization process. during the initial determination of the marginal value functions, the lp gave a feasible solution. this solution satisfies the problem constraints but is not the only one that does so. thus, it was necessary to perform a robustness analysis in order to check the stability of the model through the determination of other solutions that satisfy the problem. thus, the robustness analysis through the metaoptimization process gives the ability to control the stability of the model, and the solutions of the lp constraints are the most representative ones. moreover, the use of the weighted average for the selection of the optimal solution during the meta-optimization process confines the modelling of the dm’s behavior within narrow limits. the weighted average solution corresponds to a risk-neutral dm and excludes cases of dms who have lowor high-risk attitude. last but not least, it is essential to mention that an important advantage of the utastar method is the ability to overcome problems of incomparability that can be met during the implementation of methods that are governed by the outranking relation law when a binary comparison between the actions is conducted. this is due to the fact that the final result is based on the aggregation of the marginal value functions of the selected criteria for each action rather than a binary comparison of the alternative actions. the incomparability remains a significant problem especially in the case of the employment of the electre methods. ranking candidate area value 1 gibraltar 0.8800 2 malta 0.7500 3 cyprus 0.5015 4 crete 0.3498 table 7: marginal value functions for the problem of the sts transfer area location criteria feasible solution meta-optimized solution operational 𝒖𝟏(𝟎) 0 0 0 𝒖𝟏(𝟐) 𝒘𝟏𝟏 0.09 0.24 𝒖𝟏(𝟒) 𝒘𝟏𝟏 + 𝒘𝟏𝟐 0.443 0.25 environmental 𝒖𝟐(𝟓𝟓) 0 0 0 𝒖𝟐(𝟓𝟑) 𝒘𝟐𝟏 0.1094 0 𝒖𝟐(𝟒𝟒) 𝒘𝟐𝟏 + 𝒘𝟐𝟐 0.1894 0 𝒖𝟐(𝟒𝟏) 𝒘𝟐𝟏 + 𝒘𝟐𝟐 + 𝒘𝟐𝟑 0.2594 0.015 𝒖𝟐(𝟑𝟓) 𝒘𝟐𝟏 + 𝒘𝟐𝟐 + 𝒘𝟐𝟑 + 𝒘𝟐𝟒 0.3724 0.255 economic 𝒖𝟑(𝟑𝟐) 0 0 0 𝒖𝟑(𝟐𝟐) 𝒘𝟑𝟏 0 0.08 𝒖𝟑(𝟐𝟎) 𝒘𝟑𝟏 + 𝒘𝟑𝟐 0 0.25 𝒖𝟑(𝟏𝟒) 𝒘𝟑𝟏 + 𝒘𝟑𝟐 + 𝒘𝟑𝟑 0.12 0.25 safety and security 𝒖𝟒(𝟎) 0 0 0 𝒖𝟒(𝟏) 𝒘𝟒𝟏 0.1317 0.24 𝒖𝟒(𝟐) 𝒘𝟒𝟏 + 𝒘𝟒𝟐 0.1317 0.24 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 10 5. conclusions the implementation of multi-attribute methodologies for decision problems in the marine environment is a very promising way to overcome problems of conflicting criteria in order to find a reliable and effective solution after the evaluation of the alternative actions. often the comparison of two or more actions becomes very difficult due to the complex and frequently hostile sea environment. the unavoidable possibility of human error in combination with the unpredictable behaviour of the natural elements may compromise the success of a marine operation, causing adverse effects on humans and the environment as well as property loss. thus, there is a pressing need to make the right decisions that will support operators when dealing with the various risks. on the other hand, the primary goal of the shipping companies always remains the maximization of profit together with the minimization of the use of necessary resources. the above two pillars can be adequately combined in the mcda models, giving the shipping industry a new perspective as a result of the optimization of the evaluated alternatives. fig. 3. normalized additive value functions for the problem of the sts transfer area location. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 11 references 1. ballis a, stathopoulos a (2003) criteria for selecting innovative technologies for maritime transhipment facilities, oper res int j vol.3 no.3 (2003): 213-227 2. doumpos m, zopounidis c (2001) business failure prediction: a comparison of classification methods, oper res 2:303. 3. farquhar p h, (1984) utility assessment methods. management science 30(11) 1283-1300 4. figueira j, greco s, ehrgott m (2005) multiple criteria decision analysis. state of the art surveys, springer. 5. fishburn p c (1967) methods for estimating additive utilities. management science 13(7): 435-453 6. greco s, mousseau v, słowiński r (2008) ordinal regression revisited: multiple criteria ranking using a set of additive value functions. eur j oper res 191(2): 416-436 7. greco s, kadziński m, mousseau v, słowiński r (2011) electregkms: robust ordinal regression for outranking methods. eur j oper res 214(10): 118135 8. hurson c, siskos y, (2014) a synergy of multicriteria techniques to assess additive value models. eur j oper res 238: 540-551 9. imo (2010) manual on oil pollution, section i, prevention of pollution. imo publication 10. jacquet-lagreze e, siskos j (1982) assessing a set of additive utility functions for multicriteria decisionmaking: the uta method. eur j oper res 10: 151-164 11. jacquet-lagreze e, siskos j (2001) preference disaggregation: 20 years of mcda experience. eur j oper res 130: 233-245 12. juan carlos leyva lo´pez, lizbeth dautt sa´nchez, miguel angel aguilera contreras (2008) a multicriteria decision support system with an evolutionary algorithm for deriving final ranking from a fuzzy outranking relation, oper res int j (2008) 8:47–62, doi 10.1007/s12351-008-0006-6 13. keeney r l, raiffa h (1976) decisions with multiple objectives: preferences and value trade-offs. john wiley and sons 14. lakiotaki k, delias p, sakkalis v, matsatsinis n f (2009) user profiling based on multi-criteria analysis: the role of utility functions, oper res vol 9 (1): 3-16 15. ocmf (2013) ship to ship transfer guide for petroleum, chemical and liquefied cases, first edition, scotland, uk 16. roy b (1985) methodologie multicritere d’aide a la decision, economica paris 17. roy b, bouyssou d (1993) aide multicritere a la decision: methodes et cas. paris: economics 18. roumboutsos a, litinas n, tobaloglou e, (2004) management by projects in maritime organizations, oper res int j. vol.4, no.3 (2004): 399-409 19. sambracos, ramfou i, (2001) shipment tracking technologies and customer satisfaction, oper res int j vol 1 no i (2001): 43-54 20. siskos e, tsotsolas n (2015) elicitation of criteria importance weights through the simos method: a robustness concern. eur j oper res (in press) 21. spyridakos a, yannacopoulos d (2015) incorporating collective functions to multicriteria disaggregationaggregation approaches for small group decision making, annals of operations research, april 2015 vol 227(1):119-136, doi 10.1007/s10479-014-1609-3 22. stavrou d, ventikos n, siskos y, (2016) locating ship-to-ship (sts) transfer operations via multicriteria decision analysis (mcda): a case study, real case studies of multiple criteria decision making c. zopounidis (ed) springer (in press) 23. zopounidis c, (2001) preference disaggregation in financial modeling: basic features and some examples, oper res, 1:263:263-283 24. grigoroudis ε, doumbos m zopounidis n, matsatsinis n (ed) 2004 multicriteria decision analysis: methodological approaches and applications, new technologies, athens 25. matsatsinis n (2010) decision support systems, new technologies athens 2004 26. siskos i (2008) decision models 1st edition athens isbn: 978-960-6759-10-9 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 3–12 ___________________________________________________________________________________________________________ 12 1. introduction 2. from the analytical approach to the utastar methodology 3. use of utastar to choose sts transfer area 4. results and discussion 5. conclusions references research article analysis of death risk of covid-19 under incomplete information1 chongfu huang* key laboratory of environmental change and natural disaster, ministry of education, beijing normal university, beijing 100875, china state key laboratory of earth surface processes and resource ecology, beijing normal university, beijing 100875, china faculty of geographical science, academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china 1. introduction the covid-19 outbreak was first reported in the city of wuhan, central china’s hubei province, in december 2019. experts have attributed the outbreak to a novel coronavirus that has since spread across china and abroad with confirmed cases exceeding 234,000 globally, death toll of coronavirus tops 11,000 [1], on march 21, 2020. the disease has been named “coronavirus disease 2019” (abbreviated “covid-19”). when the disease started to spread in china, authorities reacted with historically unprecedented quarantines of cities. on february 27, 2020, the world health organization (who) raised the risk assessment of covid-19 from “high” to “very high” at global level. however, the concept of the risk of covid-19 is still unclear today, and risk communication is not smooth. before the outbreak, information about the disease was incomplete and traditional risk analysis tools were unable to provide any support. holding the hope that the virus does not pass from human to human, meanwhile some political considerations, people lost the opportunity to control the source of infection early, causing serious losses. if a more reliable risk analysis was carried out before and at the beginning of the outbreak, and if a strengthen crisis response was took immediately, the situation will certainly be much better. it is the same for almost all disasters, the story after the events is easy to tell, but the risk analysis before the events is not easy. one reason is that the information available for risk analysis before and at the beginning of the outbreak is incomplete. it is notable that the concept of risk in the glossary of who’s international health regulations is different from the definition of risk in iso 31000 recommended by international organization for standardization (iso). who defines public health risk as the likelihood of an event that may adversely affect the health of human populations [2], and iso defines risk as effect of uncertainty on objectives [3]. there are at least three terms related to who’s definition: the risk of infection [4], epidemic risk and social risk [5]. it is difficult to establish a quantitative relationship between these risks and the public health risk. when who declares covid-19’s risk is high, the meaning is not clear. if it refers to the risk of infection, the warning is not enough to attract people’s attention, because the risks infected by influenza viruses are also high, but the mortality rates of the influenzas are very low. international organization for standardization’s definition cannot be used to describe any risk of covid-19. a risk in iso’s formwork must be related to one or more objectives which are managed by individuals or organizers. for example, “annual profit” is an objective of the risk management in many companies. in other words, the level of risk is relative to managed objectives. at the beginning of january 2020, covid-19 already shown its initial shape in wuhan, what is the risk faced by the residents who will the baibuting banquet [6]? which was held for the 20th time in 2020. article info article history received 19 april 2020 accepted 19 june 2020 keywords coronavirus death risk evidence experience internet of intelligences set-valued statistics abstract it is easy to write a story about the coronavirus disease 2019 (covid-19) when everything about covid-19 is known. it is difficult to analyze the death risks of covid-19 with a few evidences collected before and at the beginning of the outbreak. in this paper, we suggest a hybrid model to analyze the death risk under incomplete information. the hybrid model would be supported by the internet of intelligences, being a platform interacting with infectious disease specialists and local doctors who fuse the evidences with the experience of the known infectious diseases and provide a series of judgments related to the death risk of a human population in a given period to covid-19. the hybrid model consists of two models of set-valued statistics and a formula. the set-valued statistics integrate the judgments for constructing (1) a probability distribution of the percent of patients, as the exposure of the population, and (2) a mortality curve with respect to the percent, as the vulnerability of the population. the suggested formula calculates the expected value of death toll. we give a virtual case to show how to use the hybrid model. © 2020 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *email: hchongfu@bnu.edu.cn 1 this project was supported by the national key research and development plan (2017yfc1502902) and the national natural science foundation of china (no. 41671502). author is the president of society for risk analysis china (sra-china), https://www.sra.org/china-sra. journal of risk analysis and crisis response vol. 10(2); july (2020), pp. 43–53 doi: https://doi.org/10.2991/jracr.k.200709.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto: hchongfu@bnu.edu.cn https://doi.org/10.2991/jracr.k.200709.002 https://www.atlantis-press.com/journals/jracr%20 44 c. huang / journal of risk analysis and crisis response 10(2) 43–53 if the “objective” is to set no one to be infected with the new crown pneumonia virus, the risk level will be high to reach the sky. no banquet! if the “objective” is to set 50% to be infected, the risk is low. do banquet! the question is who is authorized to set a “objective”. without any risk assessment, the banquet is held on january 18, 2020, more than 40,000 families in the baibuting community presented their own 13,986 dishes. in this paper, we will suggest a hybrid model to analyze the death risk caused by covid-19 under incomplete information. the paper is organized as follows: section 2 defines the death risk caused by covid-19 as the expected value of death toll; section 3 reviews three approaches to assess health risk; section 4 suggests a theoretical model to assess the death risk with complete information, where a formula is given to calculate the expected value of death toll; section 5 suggests two models to integrate the judgments given by infectious disease specialists and local doctors for constructing the probability distribution and the mortality curve, serving for the formula in section 4. in section 6, we give a virtual case, based on the information collected from the internet during december 8, 2019january 10, 2020, to show how to use the hybrid model. we conclude this paper with section 7. 2. death risk of covid-19 our world is fraught with uncertainties and hazards, called the risk society [7], as featuring a less-controllable world and increasing challenges to define risks. the less-controllable world results from the diversification of human social activities, science and technology are changing with each passing day, climate change, and environmental deterioration, in which the unpredictable consequences of modernization cause social effects that are beyond the ability of experts and institutions to narrate them. however, people do not just sitting by and watching tragedies. ancient chinese people invented a word “风险” (fēng xiǎn) to express an uncertainty tragedy. “风” means “wind”, and “险” means “danger”. chinese “风险” means english “risk”. it may be argued that, risk is a synthetic phenomenon. particularly, a risk must connect with the future and adverse incidents. according to the meaning of risk in the chinese word, the following definition [8] is much clearer to answer the question: what is a risk? definition 1. risk is a scene in the future associated with some adverse incident. scene means something seen by a viewer, or felt by individuals or various societal groups. it is a view or prospect. adverse is contrary to one’s interests or welfare. it is harmful or unfavorable. a scene must be described with a system consisting of time, a site and objects. the association would be measured with a metric space, such as probability. and, the adverse incident would be scaled with a magnitude, such as injure or death. any risk must related to some time or period, a site and some objects that would suffer adverse incidents. obviously, no past or existing scene can be called risk. for example, an outbreak cannot be called a risk. the risk analysis is for the future. the measurement of a risk depends on our knowledge for the risk. for example, the risk of aviation accidents is a statistical risk for an insurance company, but is a perceived risk for a passenger. from the perspective of epistemology, risks should be classified into four categories [9]: • a pseudo risk is that we are able to accurately predict it by using system models and currently available data. • a probability risk is that we are able to statistically predict it by using probability models and a lot of data. • a fuzzy risk is that we are able to approximately infer it by using fuzzy logic and incomplete information. • an uncertain risk is that we cannot predict or infer it by using any existing approach and available data. definition 1 is the basic definition of the risk rooted in the chinese interpretation. according to a specific type of adverse events, more specific risk definitions can be derived. for example, the definition “natural disaster risk is a scene in the future associated with the adverse incident caused by natural events or forces [10]” is derived from the basic definition. similarly, we give the following definition: definition 2. death risk of covid-19 is a scene in the future associated with death caused by covid-19. we denote the death risk as d. for a risk analyst, the categories of death risks of covid-19 are different for different regions, different times, and different populations. many years later, the death risk of covid-19 might become a pseudo risk. after an outbreak, the risk could be considered as a probability risk. for wuhan in early january 2020, the death risk of covid-19 is a fuzzy risk. the death risk of covid-19 is not an uncertain risk. 3. three approaches to assess health risks currently, there are at least three approaches to assess health risks: pathogenic modules, epidemiological models and media templates. 3.1. pathogenic modules for assessing infection risks the term pathogenicity denotes the ability of micro-organisms to cause infectious diseases within the body. the pathogenicity of a micro-organism is markedly coincident with inoculated dosage. today, some study of pathogenic modules have reached the human genome. a risk map indicating the vulnerability of different organs to covid-19 infection was constructed based on that lung, heart, esophagus, kidney, bladder, and ileum, and located specific cell types are vulnerable [11], where the authors analyzed the single-cell rna sequencing datasets derived from major human physiological systems, including the respiratory, cardiovascular, digestive, and urinary systems. they explored the data from the digestive system and known that extremely high angiotensin converting enzyme ii (ace2) expression was found in ileal epithelial cells; thus, ileum could be at high risk. more than 1% ace2 positive esophagus epithelial cells were found, and esophagus can thus be regarded as high risk. by contrast, the cells from stomach and liver showed lower ace2 expression levels. c. huang / journal of risk analysis and crisis response 10(2) 43–53 45 a study has demonstrated that the aa genotype of rs2268690 (59¢-flanking region), which leads to a higher serum concentration of a-2-heremans-schmid glycoprotein, was significantly associated with protection against severe acute respiratory syndrome (sars) development [4]. the statistical analysis of the rs2248690 genotype data among the patients and healthy controls in the cohort of health care workers, who were all similarly exposed to the sars virus, also supported the findings. more specifically, individuals with the aa genotype have a 41% lower risk of developing sars than those with the tt/at genotype. in the genetic risk association study, the odds ratio and a 95% confidence interval were used to measure the strength of association. the pathogenic modules for assessing infection risks are limited to human organs or individuals. they are not suitable for assessing the health risks faced by among people in villages, cities, areas, territories, or countries. 3.2. epidemiological models for predicting the spread of human diseases epidemiology is the study of human health in populations, rather than in individuals. it studies the causes of illness, how a particular disease is spread, and disease control. epidemiology uses a wide array of scientific tools, from medicine and statistics to sociology and anthropology. epidemiology is concerned not only with the spread of diseases, but also with their cause-if an outbreak of cholera can be traced back to a particular contaminated water source. for the covid-19 pandemic, the imperial college covid-19 response team modified an individual-based simulation model developed to support pandemic influenza planning [12,13] to explore scenarios for covid-19 in great britain (gb). the basic structure of the model remains as previously published. in brief, individuals reside in areas defined by high-resolution population density data. contacts with other individuals in the population are made within the household, at school, in the workplace and in the wider community. census data were used to define the age and household distribution size. data on average class sizes and staff-student ratios were used to generate a synthetic population of schools distributed proportional to local population density. data on the distribution of workplace size was used to generate workplaces with commuting distance data used to locate workplaces appropriately across the population. individuals are assigned to each of these locations at the start of the simulation. the team’s report [14] predicted, in the absence of any control measures or spontaneous changes in individual behavior, approximately 510,000 deaths in gb and 2.2 million in the us, not accounting for the potential negative effects of health systems being overwhelmed on mortality. the most important statistics used in the model are that, (a) incubation period are 5.1 days 9 and 10; (b) infectiousness is 6.5-day; (c) reproduction number r0 = 2.4, based on fits to the early growth-rate of the epidemic in wuhan [15,16]; (d) 50% of those in critical care will die. the report does not clearly give the mortality rate of covid-19 used in the model. it seems that the report of the team finally got the uk and us to realize that this is going to require a massive social-distancing and lockdown to avoid killing millions. it is interesting to note that, if the report was released at the beginning of january 2020, few would believe its results. the outbreak data over the past 2 months have supported the report. the results of epidemiological models depend on the statistics of diseases. changing any of the important parameters in the model will change the results. the parameters produced without a large amount of data are unreliable. they are not suitable for predicting the spread of the human diseases caused by newly discovered pathogens, such as the novel coronavirus appeared in 2019. 3.3. media templates for showing potential health risks the media templates are the paradigmatic examples, would be established and maintained by source strategies, social power relations and journalistic/audience reception processes, a journalist also examined how templates operate in relation to existing theories around key events, framing and news icons. the templates are acting to provide context for new events, and helping to shape the ways in which we make sense of the world. template events help to shape news narratives and guide thinking not only about the past, but also of the present and the future [17]. the model of media template might be illustrative of the making of social facts against the backdrop of risk society. for example, 1079 press items from the apple daily, ming pao and wen hui pao were selected for the content analysis to constructed a three-step model to examine how highly unknown social risks are constructed as social facts in the process of news-making, demonstrating how hong kong’s news discourse about the swine flu pandemic (sf) in 2009 drew upon the exemplar of the hong kong’s sars epidemic in 2003. the data coding was conducted in april 2011. before the coding, one tenth of the samples were selected randomly to conduct an inter-coder reliability test [5]. in self-media era, a lot of important and timely information may not come from the selected press agencies, but from the internet. to cover different ideologies of the local press, in the media template for sf, the apple daily (widely regarded as the most sceptical and critical newspaper), ming pao (a commercial paper widely perceived as politically neutral and elitist) and wen hui pao (a leftist, partisan paper that maintains the chinese state discourse and the official discourse of the government) were selected. it is very difficult to select the self-media platforms to serve for a media template. on the self-media platforms, such as wechat and twitter, there is not only an amount of message about truth but also lies. none of the above approaches alone can provide effective services for early risk assessment. under incomplete information, we need a hybrid model to assess death risk, in which all information from observations, analogies and public participations can be used. 4. theoretical models for assessing death risks with complete information any theoretical model is based on some assumption. when we assume that the information about risk source, risk bearing body, exposure and vulnerability are complete, there are at least three 46 c. huang / journal of risk analysis and crisis response 10(2) 43–53 theoretical models could be used to calculate or assess death risks caused by a deadly virus, with respect to specific individuals, non-specific individuals, or groups. 4.1. risk source, risk bearing body, exposure and vulnerability definition 3. a source that causes risk scenes is called the risk source. for example, the novel coronavirus is the risk source of the covid-19’s death risk; flood, earthquake and tropical cyclone are the risk sources of natural disasters; the employee errors are the risk sources of the operational risks faced by financial institutions [18]. we suggest term “scale” to measure the size of a risk source. for example, the number of covid-19 cases in a city is the scale of the risk source “covid-19”. definition 4. an object that will bears adverse incidents is called the risk bearing body. for example, a healthy person in a deadly virus-infected area is a risk bearing body. a building in a seismic activity area, a village in the flood area, and a city in a region where multiple hazards frequently occur are risk bearing bodies. a risk bearing body can be an object, individual, group, community, area, country or other entity to incur the possible damages from a risk source [19]. definition 5. the degree to which a risk bearing body is affected by a risk source is called the risk exposure. for example, the risk exposure of a healthy person to covid-19 is the number of covid-19 patients who would directly contact the person. the risk exposure of a city to covid-19 is the percent of patients who would be infected in the future. in natural management, exposure usually refers to that which is affected by natural disasters, such as people and property. “risk exposure” has more definitions than “risk”. holton defined risk as exposure to a proposition of which one is uncertain, where exposure means possible consequences and a proposition is the content of an assertion [20]. the concentration and effect levels of a chemical is called the human exposure to the chemical in a scenario-based risk assessment method [21]. the transaction volume of a bank is regarded as a risk exposure of the bank in a model to assess credit risk of banks to non-financial firms [22]. we suggest term “intensity” to measure the size of a risk exposure. for example, the percent of patients to covid-19 is the intensity of the risk exposure of a city to covid-19. the seismic intensity is the risk exposure of a building to earthquake. definition 6. the degree of adverse incident of that a risk bearing body would bear is called the vulnerability. for example, the vulnerability of a healthy person to covid-19 is degree of injury after illness caused the coronavirus. the vulnerability of a community to covid-19 could be measured in mortality rate. the vulnerability of a building to earthquake is the frangibility of the building. the means of the vulnerability is similar to the weakness in a risk bearing body. a scale of the vulnerability is from 0 (no adverse incident) to 1 (total adverse incident), or in percent. the vulnerability of a risk bearing body could be measured by a relationship between the intensities of a risk exposure and the possible adverse incidents on the body. in the risk-based decision making for terrorism applications, the definition of the vulnerability of a facility is the probability that an attack against that facility will succeed, given the attack was initiated [23]. we suggest term “curve” to describe the vulnerability of a risk bearing body to a risk source. for example, the mortality curve with the percent of patients as independent variable is the vulnerability of a city to covid-19. 4.2. pseudo risk model for calculating death risks to specific individuals let w be a set of the individuals with n persons w1, w2,..., wn, i.e., w w w wn= { }1 2, ,..., , (1) and each element w have two attribute values: exposure e and vulnerability v, which are assigned by equations (2) and (3). e w ww = 1 0 , , will be infected, will not be infected.    (2) v w ww = 1 0 , , will die after infection, does not die affter infection.    (3) that the information is complete for analysis of death risks of w as a set of specific individuals means that, all attribute values have been known, i.e., we known who will be infected by the given virus, and also know he must die or cure if he is infected by the virus. in this case, the death risk of the individuals is a pseudo risk. then, the death risk of a person w in the set can be calculated by using equation (4), d e vw w w= × . (4) for example, suppose john and mary are two persons of the specific individuals. the collected data provide complete information to assign their exposures and vulnerabilities to the virus: ejohn = 1, emary = 1, vjohn = 1, and vmary = 0. then, as two risk bearing bodies, their death risks to the given virus, as a risk source, are d e v d e vjohn john john mary mary mary= × = × = = × = × =1 1 1 1 0 0, . obviously, it is very difficult to assign exposure and vulnerability of a risk bearing body even there are a lot of data of death cases and cure cases. it must be noted that, no risk bearing body is included in the death cases or cure cases which have been recorded. any case is an occurred event, but a risk is a scene in the future. a risk bearing body would bears adverse incidents which would occur in the future. 4.3. probability risk model for assessing death risks to non-specific individuals that the information is complete for a deadly infectious disease means that, at least the basic reproduction number r0 and the c. huang / journal of risk analysis and crisis response 10(2) 43–53 47 mortality rate r have been known, which could be estimated by using a large number of epidemiological cases. let x be a set of the individuals in the epidemiological cases with m persons x1, x2,..., xm, i.e., x x x xm= { }1 2, ,..., , (5) and each element x have two attribute values, reproduction number r and vulnerability v, i.e., x = (r, v). the variable r is a nonnegative integer. v is 1 if the person is in a death case, 0 otherwise. the expectations r0 and r are the average values of r1, r2,..., rm, and v1, v2,..., vm, respectively. any risk assessed by using r0 and r is a probability risk, because r and v are random variables. the set x in equation (5) is different from the set w in equation (1), although both of their elements are persons. the elements of x are the epidemiological cases, but w’s elements are risk bearing bodies, who haven’t been infected but may be infected in the future. that the information is complete for analysis of death risks of w as a set of non-specific individuals means that, although we do not know who will be infected by the given virus, and we do not know whether they will die after they are infected, but we know r0 and r of the individuals, which are the same as ones calculated by using x. furthermore, the completeness implies that, all probability distributions related to the disease are known. it means that, for a person w, there is enough information to calculate the probability pw of his infection, according to his environment and the r0 of the disease. then, the death risk d of a person w in the set w can be calculated by using equation (6), d pw w= × r. (6) for example, let pjohn = 0.326 be the probability of john’s infection in covid-19 and rcovid-19 = 0.0387, his death risk is d pjohn john covid-19= × = × =r 0 326 0 0387 0 01262. . . . when w’s size is 10,000 and the probability of infection is the same regardless of the individuals, d = 0.01262 means that the people in w is at risk of 126.2 death, i.e., the mortality rate rw = 0.01262, which is for people who may be infected in the future, but rcovid-19 is for cases. in the real world, even the information is complete, it is not easy to estimate the probability of the infection, which is mainly determined by personal behavior and environment factors. there are too much uncertainty in routes of infection. no environmental model can estimate the probability that a person in a set of non-specific individuals will be infected. 4.4. probability risk model for assessing death risks to groups when the information about risk source, risk bearing body, exposure is complete, it is possible to estimate the probability distribution of the percents of a group facing an infectious disease. when the information about vulnerability is complete, it is possible to find a function expressing the relationship between the percent and the mortality rate, determined by the vulnerability of the group and the medical resources. the more patients, the more scarce the medical resources are, and the higher the mortality rate of patients. let t be a period in the future, and g be a group in which some persons might be infected during t by virus carriers as risk sources. that the information is complete for a deadly infectious disease means that, there are enough observations that various groups of people are infected. with the observations, we can estimate the probability distribution p(g) of the percent g of patients in g during t. with the observations of death cases and cure cases under various medical resources, we can find a function f(g) expressing the relationship between the percent g and the mortality rate r, i.e., r = f(g). the expected value of mortality rate r is the death risk d of group g in period t, which is calculated by using equation (7), d p f dg = ∫ ( ) ( ) .g g g0 1 (7) for example, let t = 30 days, and g be a group which is composed of all residents in a community. we suppose that the probability distribution p(g) of the percent g of patients in g during t is, p( ) . exp (ln . ) . , ,g g p g g= − + ×     < < 1 0 4 2 1 7 2 0 4 0 1 2 2 (8) and the relationship between the percent g and the mortality rate is, f e ( ) . , .g gg= + ≤ <− 0 1 1 0 18 (9) p(g) reflects the exposure of g to the infectious disease, and f(g) reflects the vulnerability of g under corresponding medical system to the disease. figure 1 shows the two supposed functions. with them, the death risk of group g to the infectious disease during t is assessed by equation (10), i.e., the risk is that 8.12% of the individuals in the community will die. d p f dg = =∫ ( ) ( )g g g0 1 0.0812. (10) it is interesting to note that, when the information about risk source, risk bearing body, exposure is complete, the epidemic might be over. any analysis is storytelling. the analysis results can no longer be used for risk management of the epidemic. however, the results should be useful for risk analysts to assess the risks of another similar epidemics, such as that the results of sars in 2003 should be useful for us to assess the risks of covid-19. before the outbreak and at the beginning of the outbreak of a human disease caused by a newly discovered pathogen, the information supporting the analysis risk of the disease is seriously incomplete. in the situation, it is impossible to directly estimate the probability distribution of the percents, or find a function expressing the relationship between the percent and the mortality rate. it is necessary to have model to assess the risk under incomplete information. in the case, the exposure p(g ) and the vulnerability f(g ) resulted from the similar infectious diseases, that occurred in the past, can be used for reference, but cannot be used directly. 48 c. huang / journal of risk analysis and crisis response 10(2) 43–53 figure 1 | two functions serving for assessing the death risk of a group to the infectious disease. the probability distribution reflects the exposure of the group and the relationship reflects the vulnerability of the group under corresponding medical system to the disease. (a) probability distribution of the percent g that will be infected. (b) relationship between the percent g and the mortality rate. 5. two set-valued statistics of fuzzy judgments given by stakeholders when the information of a new infectious disease is seriously incomplete, it is necessary to fuse the information with the experience about historical infectious diseases. it is possible to develop a platform of the internet of intelligences (ioi) [24] to do that, where experts, citizens, officer, as elements of a set of agents, do not only provide information and the experience but also fuse them. for a given population and time period, on ioi, each agent proposes a rough judgment of the percent of patients and the relationship between the percent and the mortality rate. then, the model serving for ioi will deal with the judgments to produce the exposure p(g ) and the vulnerability f(g ), which could be used to assess the death risk of the population to the new infectious disease, more early. 5.1. framework supporting by the internet of intelligences on january 23, 2020, chinese authorities placed a lockdown on wuhan where covid-19 outbreak. until this day, the information people have about the infectious disease is still incomplete. both of the exposure and the vulnerability of the city were unknown. the government and experts know neither risk sources (virus carriers) nor the mortality rate. however, they still remember the terrible sars. it is the shock from sars that help china controlled covid-19 in march. if decisive measures had been taken earlier, such as before january 10, the disease might not be outbreak globally. if there is a more reliable risk warning, it is possible to avoid this mistake. a believable risk warning is based on sufficient evidences. when the information is incomplete, it is necessary to have a new model that can make full use of a few observations, expert knowledge and public wisdom for the risk analysis. to do this, first we review what data is available on january 10, 2020. integrating messages from official reports and public media, at least we known the following five facts: (1) on december 8, 2019, the first case of unexplained pneumonia was officially reported in wuhan. (2) on december 31, 27 cases have been found in wuhan, of which seven are in serious condition, and two cases are expected to be discharged in the near future. (3) on january 5, 2020, 59 cases of the unknown strain of viral pneumonia have been reported in wuhan, and officials say there has been no “clear evidence” of human-to-human transmission [25]. (4) on january 10, 2020, 41 cases of pneumonia infected by the new coronavirus, including seven severe cases and one death was officially reported in wuhan [26]. (5) on january 10, 2020, experts discover that most of the infectious disease contains 80% of the same genetic code as sars occurred in 2003 [27]. obviously, the information provided by these facts is seriously incomplete for understanding a new infectious disease. however, the epidemic’s 34-day record and the high similarity with sars are sufficient to preliminarily analyze the death risk of a group to the new infectious disease. certainly, the risk analysis only based on the incomplete information is unreliable. fusing the information from the facts with the experience about historical infectious diseases, we should reasonably assess the risk. considering theoretical model in equation (7), we suggest a framework figure 2 to fuse the incomplete information with the experience, and we restrict attention to construct a probability distribution p(g ) of the percent g of patients in g during t and a function f(g ) expressing the relationship between the percent and the mortality rate. the framework is supported by ioi. a network connecting agents by the internet and integrating scattered wisdoms to be a great wisdom by embedded models is called an ioi, where an individual who can provide message, experience, knowledge, and judgment to a customer to solve problems is called an intelligent agent or simply an agent. a system consisting of a group of two or more computers linked together is called a network. the internet is the most convenient network. any mathematical expression describing the relationships among variables, any mathematical method processing data, and any human brain paradigm analyzing questions can be called a model. for an ioi, when we talk about the model, unless stated otherwise, it is assumed as a mathematical model or a human a b c. huang / journal of risk analysis and crisis response 10(2) 43–53 49 figure 2 | the framework to fuse the collected evidences of a new infectious disease with the experience of the known infectious diseases. the elements a1, a2, a3, ... are agents, such as infectious disease specialists and local doctors. s is a network server. m is a set of models, including two models to integrate the judgments given by agents for constructing the probability distribution p(g ) and the mortality curve f(g ), where g is the percent of people who will be infected. the similarities connect the new infectious disease with known ones in terms of viruses, patient symptoms, mode of transmission, and so on. brain paradigm for processing information. the formal definition of ioi is as follows [28]: let a be a set of agents, n be a network used by a, and m be a model to process information provided by a; a triple <a, n, m> is called the internet of intelligences, denoted as f. in next two subsections, we suggest two models to construct the probability distribution p(g ) and the mortality curve f(g ) used in equation (7), respectively, for assessing the death risk of a population to a new infectious disease, with respect to a time period in the future. 5.2. constructing a probability distribution by using possibility distributions for a new infectious disease, it is easy to collect the following four types of information: the times when the cases were recorded, places where the cases occurred, symptoms of the cases, and similarities to known infectious diseases. we can use the first three types of information to construct three evidence templates. the fourth type of information could play a role of bridge to connect the experience about historical infectious diseases. according to the connection, anyone can give their own judgments about the severity of the new infectious disease. the evidence templates should be post on ioi. the experience of historical infectious diseases should be stored in the mutual database of ioi, and automatically organize the templates corresponding to the evidence templates. the similarities between the evidence templates and experience templates are a series of basis to fuse the evidences with the experience. there are a lot of fuzzy uncertainties in evidence templates, experience templates, and similarities, due to linguistic terms, unstructured information, and vague similarities, and so on. only the human brain, as a super model, can handle the fuzzy information to fuse them efficiently and reliably. no mathematics model can do that. ibm’s watson cannot do it, and google’s alphago cannot. any artificial intelligence (ai) on the basis of existing mathematics is intelligence in the sense of automation, which should be called “smart”, not real intelligence that can create knowledge. the work of fusing information, for assessing death risk of covid-19 under incomplete information, can only be carried out by the human brain with the help of a series of analytical tools. there is no loss in generality when we supposed that, for a new infectious disease, q stakeholders (in risk issues, socially organized groups that are or will be affected by the outcome of the event or the activity from which the risk originates and/or by the risk management options taken to counter the risks) were ordered to give their judgments about the percent g of patients in g during t according to the collected evidences, the experience of the known infectious diseases, and the similarities between the new infectious disease with known ones in terms of viruses, patient symptoms, mode of transmission, and so on. the set of the q stakeholders is written as a a a aq= { }1 2, , ..., . (11) for a given group g and a time period t, nobody can accurately judge how percent of patients will be. each stakeholder, as an agent on the platform of ioi serving for analysis of the death risk of g in t to the new infectious disease, is requested to give a fuzzy judgment of the percent, distributing on the values in equation (12), that called the universe. u u u ud= { }1 2, , ..., . (12) for example, u1 = 0, u2 = 0.1 (i.e., 10%),..., u11 = 1 (i.e., 100%). a fuzzy judgment is a possibility distribution to distinguish the 50 c. huang / journal of risk analysis and crisis response 10(2) 43–53 possible degree of different percents. a possibility distribution is written as p = { }, , ..., .p p p1 2 d (13) an element of the possibility distribution p is also written as p(u), u ∈ u, mathematically. for example, for the universe with respect to group g and period t, u u u ug t, , , ..., , . , . , . , . , . , . , . = = { }1 2 11 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 ,, . , . , , 0 8 0 9 1{ } the possibility distribution given by a stakeholder a, pg t a u u u, ( ) { ( ) ( ) ( )}, , ..., , , . , , , , , = = p p p1 2 11 0 1 0 4 0 0 0 0 0 ,, , , , 0 0 0{ } means that the stakeholder thinks that the possibility of that 10% of patients in people of g in the future t will be infected is the largest. it is also that 20% people may be infected with a small possibility. no other percentage is possible. the main benefit of that possibility distributions are used to express the judgments is that the flexible retains useful transition information. this is very helpful to improve the accuracy of the risk assessment under incomplete information. the traditional statistics generally restricts itself to deal with the samples consisting of observations on one or more variables. the value of a variable is a real number. such observation is called the singleton observation. a fuzzy judgment with a possibility distribution is a set-valued observation, where the element of the universe is one or more variables, but more than one value would be taken. we also can use a set of set-valued observations to estimate a probability distribution with respect to a random variable. suppose now that q stakeholders have given q possibility distributions expressing their judgments of the percent of patients with respect to group g and period t, pg t i ia u i q u u, ( ) { ( )}, , ,..., ; .= = ∈p 1 2 (14) to make each judgment have the same importance, we use equation (15) to normalize them, ¢ = = ò p p p i i iu u u u du i q( ) ( ) ( ) , , , , .1 2 … (15) let p p( ) ( ), ,u u u ui i q = ∈ ≤ ≤ ∑ ′ 1 (16) which is a possibility distribution integrating the q judgments. according to the consistency principle of possibility/probability shown in equation (17) [29], prob poss( ) ( ),u u≤ (17) let g = u, p(u) could be used to infer a probability distribution in equation (18), p u du u u ( ) ( ) ( ) , ,g p g p g= ∈ ∫ (18) to calculate the death risk of group g in period t by equation (7), after we obtain mortality curve f(g). 5.3. constructing a mortality curve by rough judgments for covid-19, it took 34 days from the first case to the first deceased. although it is not a short period, it also is impossible to estimate a mortality curve of the new infectious disease by using the evidence. however, considering the similarities with the known infectious diseases, any stakeholder can give a rough judgment on the death rate. suppose that, based on the evidence templates, experience templates and the similarities which are shared on an ioi, q stakeholders have given q mortality curves expressing their judgments of the relationship between the percent g and the mortality rate r with respect to group g and period t, f a f u i q u ug t i i, ( ) { ( )}, , ,..., ; .= = î1 2 (19) let g = u, the average of these curves fi(u) in equation (20) could be used as the mortality curve in equation (7). f f q u i i q( ) ( ) , .g g g= ∈≤ ≤ ∑ 1 (20) 6. a virtual case to assess the death risk of covid-2019 since the author of this paper is neither an infectious disease specialist nor a doctor, but only a risk analyst, he can neither obtain the scientific data of the covid-19 cases nor have the experience of any known infectious disease. therefore, he can only use the suggested hybrid model for virtual data, based on the information collected from the internet during december 8, 2019january 10, 2020, to analyze the death risk of covid-19 under incomplete information. in the case, the group g studied is the population of a city, where 10 million people are living. the period t is 30 days in the future. the starting point of time starts from the day when 18 infectious disease specialists and local doctors, as stakeholders, who are virtually invited to give their judgments. suppose that, before the t, there is a small number of covid-19 cases and early outbreaks. in the virtual case, the information available for risk analysis is seriously incomplete. as a virtual case, it is reasonable to assume that there is an ioi where the invited stakeholders can work on, and they are required to give their judgments under following two constraints: (1) the current medical resources (local hospital facilities and doctors); (2) the current administration (no control over patients and communities). let the universe of percent of patients, for analyzing the death risk of the city to covid-19 in future 30 days, be c. huang / journal of risk analysis and crisis response 10(2) 43–53 51 table 1 | the fuzzy judgments of that the percent of patients who might be infected by the covid-19 virus in the studied city in the next 30 days u1 u2 u3 u4 u5 u6 u7 u8 u9 u10 u11 ids1 0.1 1 0 0 0 0 0 0 0 0 0 ids2 0 0.8 1 0.1 0 0 0 0 0 0 0 ids3 0.5 1 0.6 0 0 0 0 0 0 0 0 ids4 0.2 1 0.1 0 0 0 0 0 0 0 0 ids5 1 0.5 0 0 0 0 0 0 0 0 0 ids6 0.3 0.2 0.1 0 0 0 0 0 0 0 0 ids7 0 0.4 0.8 1 0.7 0.6 0.2 0.1 0 0 0 ids8 0 0.9 1 0.7 0.5 0 0 0 0 0 0 ids9 0.4 0.5 0.7 0.8 0.9 1 0.9 0.8 0.7 0.1 0 ids10 0 0.8 0.9 1 0.7 0.5 0.1 0 0 0 0 ld1 0 0.5 0.6 0.8 1 0.7 0.6 0.2 0.1 0 0 ld2 0 0.8 1 0.8 0 0 0 0 0 0 0 ld3 0 0.4 0.8 1 0.8 0.4 0 0 0 0 0 ld4 0 0 0.3 0.6 0.8 1 0.7 0.2 0.1 0 0 ld5 0 0.1 0.4 0.8 1 0.9 0.8 0.4 0.3 0.1 0 ld6 0 0 0 0.6 1 0.6 0.3 0 0 0 0 ld7 0.2 0.8 1 0.9 0.7 0.6 0.5 0.2 0.1 0 0 ld8 0.6 1 0.6 0 0 0 0 0 0 0 0 ids, infectious disease specialist; ld, local doctor, the value in the table is possibility p. u u u u= = { }1 2 11 0 0 1 0 2 0 3 0 4 0 5 0 6 0 7 0 , ,..., , . , . , . , . , . , . , . , . 88 0 9 1, . , . { } (21) suppose that, there are 10 infectious disease specialists and eight local doctors among the 18 stakeholders, who given fuzzy judgments of that the percent of patients to covid-19 in equation (21) are the possibility distributions shown in table 1. taking the seventh local doctor’s possibility distribution as an example, we use equation (15) to normalize the judgment. let h u du u u j j ld7 ld7 ld7 0.2 + 0.8 + 1 + 0.9 + 0.7 = = = ∫ ∑ ≤ ≤ p p( ) ( ) 1 11 + 0.6 + 0.5 + 0.2 + 0.1 = 5, then, p′g,t ld7 ld7 ld7 ld7ld7 ( ) { ( ), ( ), ( ), , ( )}= ′ ′ ′ ′p p p pu u u u1 2 3 11� ld7 ld7 ld7 ld7 ld7 ld7 = p p p( ) , ( ) , ( ) , u h u h u h 1 2 3 ��, ( ) . , . , , . pld7 ld7 u h 11 0 2 5 0 8 5 1 5 0       = 99 5 0 7 5 0 6 5 0 5 5 0 2 5 0 1 5 0 5 0 5 , . , . , . , . , . , ,       0.04, 0.16, 0.2, 0.18, 0.14, 0.12, 0.1, 0.04, 0.02, 0,= 0{ } . summing the 18 normalized judgments by equation (16), we obtain an integrated possibility distribution, pg,t { ( ), ( ), ( ), , ( )} ( ), ( ) p p p p p p u u u u u ui i i i 1 2 3 11 1 1 18 2 1 � = ′ ′ ≤ ≤ ≤ ∑ ≤≤ ≤ ≤ ≤ ≤ ∑ ∑ ∑′ ′       = 18 3 1 18 11 1 18 , ( ), , ( ) { p pi i i i u u� 2.021, 5.083,, 3.307, 2.438, 2.075, 1.521, 0.919, 0.366, 0.235, 0.036, 0} . for example, p p( ) ( )u ui i 2 2 1 18 = ′ = ≤ ≤ ∑ 0.909 + 0.421 + 0.476 + 0.769 + 0.333 + 0.333 + 0.105 + 0.29 + 0.074 + 0.2 + 0.111 + 0.308 + 0.1118 + 0 + 0.021 + 0 + 0.16 + 0.455 5.083.= p p( ) ( )u ui i 2 2 1 18 = ′ = ≤ ≤ ∑ 0.909 + 0.421 + 0.476 + 0.769 + 0.333 + 0.333 + 0.105 + 0.29 + 0.074 + 0.2 + 0.111 + 0.308 + 0.1118 + 0 + 0.021 + 0 + 0.16 + 0.455 5.083.= let h u du u = = + + + + + + ∫ p( ) 2.021 5.083 3.307 2.438 2.075 1.521 0.919 ++ + + = ≈ 0.366 0.235 0.036 18 001 18. , using equation (18), we infer a discrete probability distribution of the percent of patients who might be infected by the covid-19 virus in the studied city in the next 30 days, p u p u p u p u p u u h u h u h ( { ( ), ( ), ( ), , ( )} ( ) , ( ) , ( ) , , ) = = 1 2 3 11 1 2 3 � � p p p p(( )u h 11     = 2.021 18 , 5.083 18 , 3.307 18 , 2.438 18 , 2.075 18 , 1.521 18 , 0.919 18 , 0.366 18 , 0.235 18 , 0.036 18 , 0 18       = {00.112, 0.282, 0.184, 0.135, 0.115, 0.084, 0.051, 0.020, 0..013, 0.002, 0}. for example, p(u2) = 0.282 means that, according to the judgments given by the 18 stakeholders, the probability of infection of 10% of people in the city in the 30 days is 0.282. similarly, suppose that, the 18 stakeholders given 18 mortality curves expressing their judgments of the relationship between the percent and the mortality rate, shown in table 2, including the average of the mortality rate with respect to each percent shown in equation (21). then, we obtain a discrete mortality curve of the percent of patients who might be infected by the covid-19 virus in the studied city in the next 30 days, 52 c. huang / journal of risk analysis and crisis response 10(2) 43–53 table 2 | the judgments of the mortality curve f(u) with respect to percent u of patients who might be infected by the covid-19 virus in the studied city in the next 30 days u1 u2 u3 u4 u5 u6 u7 u8 u9 u10 u11 ids1 0 0.10 0.50 0.6 0.8 1 1 1 1 1 1 ids2 0 0.8 1 1.1 1.5 1.5 1.5 1.5 1.5 1.5 1.5 ids3 0 0.15 0.55 0.9 1 1.2 1.2 1.2 1.2 1.2 1.2 ids4 0 0.08 0.3 0.5 0.8 0.9 0.9 0.9 0.9 0.9 0.9 ids5 0 0.03 0.2 0.4 0.6 0.6 0.6 0.6 0.6 0.6 0.6 ids6 0 0.02 0.3 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 ids7 0 1.5 2.1 3.5 4.02 4.02 4.02 4.02 4.02 4.02 4.02 ids8 0 1.9 2.7 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 ids9 0 2.5 6.04 6.04 6.04 6.04 6.04 6.04 6.04 6.04 6.04 ids10 0 2 2.5 3.5 4 4 4 4 4 4 4 ld1 0 2 3 4 5 5 5 5 5 5 5 ld2 0 0.7 0.8 1.3 1.3 1.3 1.3 1.3 1.3 1.3 1.3 ld3 0 1.4 1.8 3 3.5 3.5 3.5 3.5 3.5 3.5 3.5 ld4 0 3 5 5.8 5.8 5.8 5.8 5.8 5.8 5.8 5.8 ld5 0 2.6 4 5.1 5.1 5.1 5.1 5.1 5.1 5.1 5.1 ld6 0 2.5 3.8 4.9 4.9 4.9 4.9 4.9 4.9 4.9 4.9 ld7 0 1.2 1.3 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4 ld8 0 0.13 0.45 0.8 0.9 1 1 1 1 1 1 average 0 1.256 2.019 2.619 2.831 2.864 2.864 2.864 2.864 2.864 2.864 ids, infectious disease specialist; ld, local doctor, the value in the table is the mortality rate r. f u f u f u f u f u( { ( ), ( ), ( ), , ( )} { , . , . , . , . ) = = 1 2 3 11 0 1 256 2 019 2 619 2 8 � 331 2 864 2 864 2 864 2 864 2 864 2 864 , . , . , . , . , . , . }. according to equation (7), we infer the death risk of covid-19, d p f d p u f ui i i = = = ´ + ´ + ò å £ £ ( ) ( ) ( ) ( ) . . g g g 0 1 1 11 0 112 0 1 2560.282 0.1844 0.135 0.115 0.084 0.051 0 ´ + ´ + ´ + ´ + ´ + 2 019 2 619 2 831 2 864 2 864 . . . . . ..020 0.013 0.002 1.8917. ´ + ´ + ´ + ´ = 2 864 2 864 2 864 0 2 864 . . . . considering that there are ten million people in the studied city, d = 1.8917 means that, in the next 30 days, the death toll to covid-19 might be 189170 (0.018917 × 10000000 = 189170), if both of the medical resources and the administration are unchanged. this is a terrible figure of deaths. this figure is enough to attract the attention of the authorities, and will immediately take a strong crisis response measures. 7. conclusion under the condition of incomplete information, there still is a possibility to reasonably analyze the death risk of covid-19. however, the first of all, the death risk must be defined clearly. there is no abstract risk, only concrete risk. secondly, there must be some infectious disease specialists and local doctors, who can fuse the evidences of the new infectious disease with the experience of the known infectious diseases, to give fuzzy judgments: (1) possibility distributions of the percent of patients in the population during a period, and (2) mortality curves with respect to the percent. third, with the set-valued statistic, we can change the fuzzy judgments into the probability distribution of the percent of patients and the mortality curve. finally, as the expected value of death toll, the death risk can be calculated by using the probability distribution and the mortality curve. simply speaking, the suggested hybrid model consists of three statistics models, one is to calculate an expected value, others are set-valued statistics to integrate fuzzy judgments. the hybrid model would be performed on an ioi where the infectious disease specialists and local doctors share information and experience and give their judgments. it is easy to write a story when everything has been known. but, any risk analysis is not about writing stories. the death risk is a scene in the future, not a story. it is difficult to analyze the death risks of covid-19 with a few evidences collected before and at the beginning of the outbreak. the death risk of covid-19, which is calculated by using the suggested hybrid model, should not be the same as the future story, but if the assessment can promote reducing catastrophes, the value of the model will be revealed. the meaning of “the risk of covid-19” is ambiguous, which may be infection risk, death risk, disability risk, or economic systemic risk. particularly, any health risk must be related to individuals or a population, as the risk bearing body, and the risk is constrained by a time period. the risk within 5 days is quite different from one within two months. neither iso’s risk definition: “effect of uncertainty on objectives” or who’s definition of public health risk: “likelihood of an event that may adversely affect the health of human populations” can lead out the definition of death risk. without an managed “objective”, no risk can be measured by iso’s definition. the who’s risk warning cannot provide the connotation of a risk. according to the meaning of risk in the chinese word, “risk” should be defined as: “a scene in the future associated with some c. huang / journal of risk analysis and crisis response 10(2) 43–53 53 adverse incident”, which leads out the definition of death risk of covid-19: “a scene in the future associated with death caused by covid-19”. as the expected value of death toll, the death risk of covid-19 can be assessed by the exposure and vulnerability of a human population to covid-19. in the suggested model, the probability distribution of the percent of patients in the population during a period represents the exposure, and the mortality curve with respect to the percent represents the vulnerability. as evidence increases, medical resources improve, and crisis response strengthen, the death risk must change. the ioi with updated risk function is a risk radar. today, the most of commercial risk radars only have the function to show risks, as same as a set of risk matrixes. the risk radar driven by ioi monitors dynamic risks for emergency management in community. conflicts of interest the author declares no conflicts of interest. references [1] cgtn. global death toll of coronavirus tops 11,000: jhu. available from: https://www.cgtn.com/special/battling-the-novelcoronavirus-what-we-know-so-far-.html (accessed march 21, 2020). [2] world health organization. joint external evaluation tool: international health regulations (2005). geneva, switzerland: world health organization; 2016. available from: https://apps. who.int/iris/handle/10665/204368. [3] iso. 31000: 2009 risk management principles and guidelines. geneva, switzerland: international organization for standardization; 2009. [4] zhu x, wang y, zhang h, liu x, chen t, yang r, et al. genetic variation of the human α-2-heremans-schmid glycoprotein (ahsg) gene associated with the risk of sars-cov infection. plos one 2011;6:e23730. [5] chan ck. defining health risk by media template: hong kong’s news discourse of the swine flu pandemic. journalism 2016;17:1018–36. [6] wu yl, shi yq. wuhan community hosts the wanjia banquet. beijing: china news.com; 2020. available from: https://baijiahao. baidu.com/s?id=1656118483387696299&wfr=spider&for=pc (in chinese). [7] beck u. risk society: towards a new modernity. london: sage publications; 1992. [8] huang c, ruan d. fuzzy risks and an updating algorithm with new observations. risk anal 2008;28:681–94. [9] huang c, huang y. an information diffusion technique to assess integrated hazard risks. environ res 2018;161:104–13. [10] huang cf, liu al, wang y. a discussion on basic definition of disaster risk. j nat dis 2010;19:8–16 (in chinese). [11] zou x, chen k, zou j, han p, hao j, han z. single-cell rna-seq data analysis on the receptor ace2 expression reveals the potential risk of different human organs vulnerable to 2019-ncov infection. front med 2020;14:185–92. [12] ferguson nm, cummings da, fraser c, cajka jc, cooley pc, burke ds. strategies for mitigating an influenza pandemic. nature 2006;442:448–52. [13] halloran me, ferguson nm, eubank s, longini jr im, cummings dat, lewis b, et al. modeling targeted layered containment of an influenza pandemic in the united states. proc natl acad sci u s a 2008;105:4639–44. [14] ferguson nm, laydon d, nedjati-gilani g, imai n, ainslie k, baguelin m, et al. impact of non-pharmaceutical interventions (npis) to reduce covid19 mortality and healthcare demand. london: imperial college london: 2020, pp. 1–20. available from: https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/imperial-college-covid19npi-modelling-16-03-2020.pdf. [15] li q, guan x, wu p, wang x, zhou l, tong y, et al. early transmission dynamics in wuhan, china, of novel coronavirus– infected pneumonia. n engl j med 2020;382:1199–207. [16] riou j, althaus cl. pattern of early human-to-human transmission of wuhan 2019 novel coronavirus (2019-ncov), december 2019 to january 2020. euro surveill 2020;25:2000058. [17] kitzinger j. media templates: patterns of association and the (re)construction of meaning over time. media culture soc 2000;22:61–84. [18] huang y. a naive uncertainty model for measuring operational risks faced by financial institutions. stochastic environ res risk assess 2009;23:507–16. [19] coppola dp. introduction to international disaster management. mutat res 2015;8:409–11. [20] holton ga. defining risk. finan anal j 2004;60:19–25. [21] scheringer m, vögl t, von grote j, capaul b, schubert r, hungerbühler k. scenario-based risk assessment of multi-use chemicals: application to solvents. risk anal 2001;21:481–97. [22] accornero m, cascarino g, felici r, parlapiano f, sorrentino am. credit risk in banks’ exposures to non-financial firms. eur finan manage 2018;24:775–91. [23] dillon rl, liebe rm, bestafka t. risk-based decision making for terrorism applications. risk anal 2009;29:321–35. [24] huang c. multiple internet of intelligences for risk analysis. j risk anal crisis response 2014;4:61–71. [25] wuhan municipal health commission. wuhan municipal health and health committee’s report on unexplained viral pneumonia. available from: http://wjw.hubei.gov.cn/fbjd/dtyw/202001/ t20200105_1903304.shtml (accessed january 21, 2020). [26] liao j, li cz. 41 cases of new coronavirus pneumonia infected in wuhan. beijing: xinhuanet; 2020. available from: http://www. xinhuanet.com/local/2020-01/11/c_1125448269.htm (accessed january 21, 2020). [27] yan qr. race against time, second-generation technology shortens viral gene sequencing time. beijing: china business journal. available from: https://xw.qq.com/cmsid/20200123a0hwut00 (accessed january 24, 2020). [28] huang c. internet of intelligences can be a platform for risk analysis and management. human ecol risk assess int j 2015;21:1395–409. [29] klir gj, harmanec d. on some bridges to possibility theory. in: de cooman g, ruan d, kerre ee, editors. foundations and applications of possibility theory. singapore: world scientific; 1995, pp. 3–19. https://apps.who.int/iris/handle/10665/204368 https://apps.who.int/iris/handle/10665/204368 https://doi.org/10.1371/journal.pone.0023730 https://doi.org/10.1371/journal.pone.0023730 https://doi.org/10.1371/journal.pone.0023730 https://doi.org/10.1371/journal.pone.0023730 https://doi.org/10.1177/1464884915595473 https://doi.org/10.1177/1464884915595473 https://doi.org/10.1177/1464884915595473 https://doi.org/10.1111/j.1539-6924.2008.01057.x https://doi.org/10.1111/j.1539-6924.2008.01057.x https://doi.org/10.1016/j.envres.2017.10.037 https://doi.org/10.1016/j.envres.2017.10.037 https://doi.org/10.1007/s11684-020-0754-0 https://doi.org/10.1007/s11684-020-0754-0 https://doi.org/10.1007/s11684-020-0754-0 https://doi.org/10.1007/s11684-020-0754-0 https://doi.org/10.1038/nature04795 https://doi.org/10.1038/nature04795 https://doi.org/10.1038/nature04795 https://doi.org/10.1073/pnas.0706849105 https://doi.org/10.1073/pnas.0706849105 https://doi.org/10.1073/pnas.0706849105 https://doi.org/10.1073/pnas.0706849105 https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/imperial-college-covid19-npi-modelling-16-03-2020.pdf https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/imperial-college-covid19-npi-modelling-16-03-2020.pdf https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/imperial-college-covid19-npi-modelling-16-03-2020.pdf https://www.nejm.org/doi/10.1056/nejmoa2001316 https://www.nejm.org/doi/10.1056/nejmoa2001316 https://www.nejm.org/doi/10.1056/nejmoa2001316 https://doi.org/10.2807/1560-7917.es.2020.25.4.2000058 https://doi.org/10.2807/1560-7917.es.2020.25.4.2000058 https://doi.org/10.2807/1560-7917.es.2020.25.4.2000058 https://doi.org/10.1177/016344300022001004 https://doi.org/10.1177/016344300022001004 https://doi.org/10.1177/016344300022001004 https://doi.org/10.1007/s00477-008-0240-8 https://doi.org/10.1007/s00477-008-0240-8 https://doi.org/10.1007/s00477-008-0240-8 https://doi.org/10.1111/0272-4332.213127 https://doi.org/10.1111/0272-4332.213127 https://doi.org/10.1111/0272-4332.213127 http://dx.doi.org/10.1111/eufm.12138 http://dx.doi.org/10.1111/eufm.12138 http://dx.doi.org/10.1111/eufm.12138 https://doi.org/10.1111/j.1539-6924.2008.01196.x https://doi.org/10.1111/j.1539-6924.2008.01196.x https://doi.org/10.2991/jrarc.2014.4.2.2 https://doi.org/10.2991/jrarc.2014.4.2.2 https://doi.org/10.2991/jrarc.2014.4.2.2 http://wjw.hubei.gov.cn/fbjd/dtyw/202001/t20200105_1903304.shtml http://wjw.hubei.gov.cn/fbjd/dtyw/202001/t20200105_1903304.shtml http://www.xinhuanet.com/local/2020-01/11/c_1125448269.htm http://www.xinhuanet.com/local/2020-01/11/c_1125448269.htm https://xw.qq.com/cmsid/20200123a0hwut00 https://doi.org/10.1080/10807039.2014.955396 https://doi.org/10.1080/10807039.2014.955396 https://doi.org/10.1080/10807039.2014.955396 https://doi.org/10.2807/1560-7917.es.2020.25.4.2000058 基于行走时间最短的机场停机位分配问题的一种混合算法：遗传算法加涟漪扩散模型 principle of internet of intelligences and development of its core technology chongfu huang1,2,3 1. key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing 100875, china 2. academy of disaster reduction and emergency management, faculty of geographical science, beijing normal university, beijing 100875, china 3. beijing cazl technology service co., ltd., beijing 100088, china e-mail: hchongfu@bnu.edu.cn abstract the network platform connecting multiple agents, which is embedded in the models to integrate individual wisdoms into a great wisdom, is called internet of intelligences (ioi). the risk radar can be driven by ioi with mathematical models written by php language. a measure space of the seismic macro-anomalies group was set up in ioi based the consensus of the staffs participating earthquake monitoring and prediction. the core technology in ioi is information diffusion technology, and its intelligent development depends on the appearance of intelligent mathematics on earth. by means of the development of the factor space theory to construct ioi expressing knowledge and being cognizant of thinking, we should promote the research of "internet + risk analysis" and promote the development of web mathematics, and lay the foundation for intelligent mathematics. the ioi supported by web mathematics is expected to decompose and dissolve wisdom ingredients in some factor spaces so that individual wisdoms are integrated to be a great wisdom. keywords: internet of intelligences; risk; risk radar; information diffusion; factor space; web mathematics; intelligent mathematics 智联网原理及其核心技术的发展 黄崇福 1,2,3 1. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875，中国 2. 北京师范大学地理科学学部减灾与应急管理研究院，北京 100875，中国 3. 北京崇安智联科技服务有限责任公司，北京 100088，中国 摘 要：由互联网联结多个智能体，并通过嵌入的模型实现集个体小智慧为群体大智慧的网络平台，称为智 联网。用 php 语言编写数学模型所支持的智联网平台，已经实现了驱动风险雷达和建立共识性地震宏观异常 度量空间。智联网的核心技术是信息扩散技术，其智能化的发展，依赖于地球上出现智能数学。通过发展因 素空间理论来构建表达知识和认知思维的智联网，既能推动“互联网+风险分析”的研究，还能促进网络数 学的形成，为智能数学打下基础。网络数学支持的智联网，有望用因素空间来分解、溶合智慧成分，将小智 慧提升为大智慧。 关键词：智联网；风险；风险雷达；信息扩散；因素空间；网络数学；智能数学 ∗本项目受国家自然科学基金项目(no.41671502)的资助，部分受北京崇安智联科技服务有限责任公司的资助。 received 5 september 2017 accepted 28 september 2017 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 146 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 1.引言 用有限的知识，认识无限的世界，是人类永恒 的追求。用不完备信息和有限的知识，研究不确定 性极大的与不利事件有关的未来情景，这就是风险 分析的工作。风险是一种扑塑迷离的现象，风险分 析不是将数据放入模型中算一算的机械性工作，而 是一项智能工作。随着互联网的广泛应用，智联网 技术应运而生 [1] ，为集众人之所长，生超级之智能， 解风险之万千，提供了一个重要的渠道。同时，相 关的研究，将为互联网时代人工智能的研究提供一 个检验平台。 由互联网联结多个智能体，并通过嵌入的模型 集个体小智慧为群体大智慧的网络平台，称为智联 网。目前，人们已经研制出 9 个智联网平台。 高考志愿填报智联网服务平台 [2] 是首个研制成 功的智联网演示系统。考生以购买服务的方式，通 过智联网获得信息和咨询，并由数学模型处理后提 供决策参考，有效规避填报失误的风险。台风灾害 风险分析智联网 [3] 中出现了柔性知识捕获器，以在温 州地区开展台风灾害水产养殖保险的可行性需求调 查为例，对智联网平台的理念和模型进行了成功的 验证。智联网支撑的“优秀论文”评选系统 [4] 能评出 共识性最高的优秀论文, 多次用于中国风险分析与 管理精英杯优秀论文的评选。内涝风险分析智联网 服务平台 [5] 使用属性拼图技术对原始的经验信息进 行优化整合，再通过“雨强-水深”模糊关系模型， 在线实现基于降雨强度，估计易涝点积水深度，进 而给出积水风险值。智联网驱动的风险雷达 [6] 通过社 区居民的参与，进行风险信息搜集、动态风险评估， 实现了风险事件的动态追踪。该项发明已由北京崇 安智联科技服务有限责任公司申请了名为“一种社 区安全风险雷达智联网服务系统”的发明专利。海 洋环境风险管理智联网平台 [7] 通过去中心化信息收 集的技术，完成了围填海造地项目对天津自然灾害 抵御能力的影响评价。地震宏观异常的智联网服务 平台 [8] 将众多一线地震工作者对地震宏观异常群强 度的目视判断，以模糊隶属函数的曲线方式输入， 在智联网中建立以模糊关系矩阵表达的共识性宏观 异常群测度空间，有望为震前宏观异常辅助地震预 报提供帮助。风险时效性评价的智联网服务平台 [9] 是一个因素藤智联网，被用于对北京延庆区果树冰 雹灾害概率风险的时效性进行评价。风险沟通智联 网服务平台 [10] 提供了洪水灾害风险沟通的便捷渠 道，并尝试用于宁波市洪水灾害的风险管理。 上述这些智联网，无论是其遵循的原理，还是 集值统计意义上的信息扩散核心技术，均是传统数 学模型；加之缺少合适的商业模式，目前还没有被 大规模地应用。如果能跳出传统数学模型的约束， 实现较高水平的智能化，哪么，既使不靠商业模式， 仅仅靠技术优势，也会有更多的人依托智联网开展 风险分析的工作，使之成为“互联网+风险分析”的 主流。 本文将在第 2 节介绍智联网原理，给出相关定 义和编程方式；第 3 节介绍智联网中的核心技术“信 息扩散”；第 4 节提出了对网络数学的设想；第 5 节 研究因素空间理论作为网络数学出发点的问题；第 6 节探讨如何网络数学支持智联网，第 7 节给出结论 和一些后续问题的讨论。 2．智联网原理 人类社会经历了从农业阶段到工业阶段，再到 信息化阶段三次巨大浪潮。目前，人类正处于第三 次浪潮之中，其特点是形形色色的互联网无处不在。 移动通信、物联网、战术数据链等大型的信息系统， 无一例外皆有互联网的烙印。然而，今天的互联网 系统，远非智能系统，更多的是信息发布和交换系 统。 今天，以深度学习为特征，以大数据为背景的 人工智能被炒得沸沸扬扬。alphago 横扫 60 位围棋 大师；watson 制定癌症治疗方案，都是标志性的事 件。一夜之间，似乎人工智能已从虚无缥缈的幻想 成为了现实。如此这般，人工智能完成风险分析的 工作，应该更不在话下。然而，事实并非如此，这 些奇妙的人工智能，本质上与工业革命时代的自动 化同出一辙，并非智能化。正如日本发明的模糊技 术控制的洗衣机被冠以智能洗衣机之名，但其实并 无智能，只是自动化程度高一点而已。人类智能的 标志是能创造知识。只有能创造知识的机器智能， 才是真正的人工智能。计算机系统用某种技巧拼凑 出乐曲，用某些模板几十秒内写出重大自然灾害事 件的新闻稿子，已经被商业包装为“人工智能”产 品，但其与真正的人工智能，毫无关系。自动化意 义下的智能，应该被称为“灵巧”，英文为 smart； 能创造知识的智能，实为“智慧”，才称得上英文的 intelligence。 显然，依靠人类智能进行的风险分析工作，不 可能由 smart 式的人工智能替代。依靠网络的支持， journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 147 做大做强人类智能，或许是人工智能热潮中，提高 风险分析水平的一条重要渠道。建设这条渠道的工 具之一，是智联网。 “智联网”(internet of intelligences) 是 在互联网的服务器上，使用一系列数学模型，利用 人机交互技术，实时处理来自众多利益相关者的信 息、经验和判断，对客观现象做出综合分析的网络 平台。 我们可以将智联网与物联网做一个比较，说明 智联网的本质所在。物联网无需人的干预，其实质 是利用射频自动识别技术，通过计算机互联网实现 物品的自动识别和信息的互联与共享。智联网须有 人参与，其实质是用数学模型整合信息和经验，将 个人判断的小智慧提升为群体大智慧，生成智能产 品。物联网是一个信息管理系统，智联网是一个智 慧加工和生产系统。 智联网原理由两大部分构成，一是形式化定义； 二是编程原理。 2.1 智联网的形式化定义 定义 1.具有观察、演绎、推理和解决问题能力的个 体称为智能体。 例如，正常的个人、决策支持系统、恒温调节 器、模糊洗衣机等，均是智能体。 定义 2.设 a 是一个智能体集合，n 是 a 使用的一个 网络，m 是处理 a 所提供信息的模型，三元体<a, n, m>称为一个智联网，记为φ。 换言之，一个由智能体、计算机网络和信息处 理器组成的系统，称为一个智联网。 最简单的智联网可以用图 1 示之，该系统由三 个智能体 a1,a2,a3，三台终端机 c1,c2,c3，一台服务 器 s 和一个信息处理模型 m 组成。它的拓扑结构见 图 2。智联网的工作原理见图 3。 互联网是一个动态混沌系统，没有人知道每一 个节点何时会发生什么变化，也不知这种变化对其 它节点会产生什么影响。智联网则是一个智力提升 系统，只有当其数学模型 m 能加工生产出智慧产品 时，系统才有存在的意义。智联网具有动态自组织 性，目标是形成一个高级智能体。互联网从混沌系 统走向自组织系统，必将是一次革命。 事实上，互联网应用模式的发展和人类大脑的 结构机理，有惊人的重合。例如，从电子邮件、电 子公告牌到博客、社会化网络，互联网的每一个应 用创新都能映射到人脑的功能结构中 [11] 。人们预测， 互联网的应用模式将形成一个与人类大脑功能结构 高度相似的网络虚拟结构，互联网正朝着使人脑充 分互联的方向发展和进化，全球脑呼之欲出 [12] 。 a1 s m a2 a3 图 2. 一个简单智联网的拓扑结构[13]，由一个网络服务器 s， 一个数学模型 m 和的三个智能体 a1,a2,a3 组成。 多个智能体 客户的问题 数据库 在线信息 文件 网络服务器 数学模型 解答 图 3. 智联网工作原理[13]。智联网用数学模型处理来自智能 体的在线信息，在一些背景数据库的支持下，智联网为客 户解答问题。 c1 c2 c3 a1 a2 a3 s(íø âç·þ îñ æ÷) m(ðåï¢ ´¦ àíä£ðí) 图 1. 最简单的智联网[1]。图中的三台计算机形成网络，三 个工作人员是三个智能体。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 148 2.2 智联网的编程原理 智联网是一个计算机网络系统平台，用网络编 程语言来实现。编程原理是：用 html 语言实现终 端界面，用 mysql 建设服务器上的数据库，用 php 语言连接界面和数据库并编写数学模型。这一原理 如图 4 所示。 2.2.1 html 语言 智联网的终端界面，由网页编写技术实现。 html（hypertext mark-up language）即超文本标 记语言，是目前网络上应用最为广泛的网页编写语 言，它提供了终端界面的编写技术。 html 文本是由 html 命令组成的描述性文 本，html 命令可以说明文字、图形、动画、声音、 表格、链接等。html 的结构包括头部（head）、主 体（body）两大部分，其中头部描述浏览器所需的 信息，而主体则包含所要说明的具体内容。 用记事本编写 html 语言的文件，并以后缀名 “html”存之，用浏览器就可以打开所设计的终端界 面。为了方便开发终端界面，人们研制了诸如 editplus、dreamweaver 和 zend studio 等大量的开发 工具，使 html 文件的编写，更加方便。 2.2.2 mysql 数据库 互联网本身就是一个大型的分布式超媒体信息 数据库，但服务于智联网的数据库，是一个能用互 联网访问的专用数据库。mysql、oracle 和 sql server 数据库都可以使用，但 oracle 过于庞大，sql server 的安全性较差，在智联网中我们使用 mysql。 mysql 是一种开放源代码的关系型数据库管 理系统，任何人都可以下载并根据个性化的需要对 其进行修改。通过数据库管理系统，管理者可以轻 松地实现各种数据库对象的访问（增、删、改、查 等操作），并可以轻松地完成数据库的维护工作（备 份、恢复、修复等操作）。 mysql 数据库的管理者，通过调用操作系统的 进程管理、内存管理、设备管理以及文件管理等服 务，为数据库用户提供管理、控制数据库容器中各 种数据库对象、数据库文件的接口。 不掌握数据库管理系统的数据库用户，则无法 经由网络获取数据库文件中被管理者屏蔽了的具体 内容。 mysql 的功能并不很强大，但因为它的开源、 广泛传播，是智联网首选的数据库。 2.2.3 php 语言 连接界面和数据库并能运行数学模型的网络程 序，用 php 语言编写。php（hypertext preprocessor） 即超文本预处理器，是一种通用开源脚本语言，其 语法吸收了 c 语言、java 和 perl 的特点，程序嵌入 到 html 文档中去执行。 php 脚本在服务器上执行，然后向浏览器发送 回 html 结果。php 脚本可放置于 html 文档中 的任何位置，通常以“ <?php”开头，以 “?>” 结尾。 在 php 中，所有用户定义的函数、类和关键词（例 如 if、else、echo 等等）都对大小写不敏感，而对 所有变量的大小写敏感。 参照最流行和使用最为广泛的能编排漂亮数学 公式的 latex 排版系统，php 对数学变量使用“$” 进行标注。php 程序中的变量以 $ 符号开头，其后 是变量的名称。一个用一维信息分配方法进行概率 分布估计的 php 程序见图 5。 html html html mysql php php php 图 4. 智联网编程原理示意图。智联网平台由三个部分构 成：终端界面、服务器上的数据库和连接界面和数据库并 能运行数学模型的网络程序。它们分别由 html 语言、 mysql 数据库和 php 语言来实现。 function distrprob($rr,$uu) { $rs=$uu[1]-$uu[0]; for($j=0;$j<count($uu);$j++) { for($i=0;$i<count($rr);$i++) { if(abs($rr[$i]-$uu[$j])<$rs||abs($rr[$i]-$uu[ $j])==$rs) $q[$j]+=(1-(abs($rr[$i]-$uu[$j]))/$rs); } 图 5. 一维信息分配的概率分布估计 php 代码。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 149 http://baike.baidu.com/item/%e8%b6%85%e6%96%87%e6%9c%ac http://baike.baidu.com/item/%e8%b6%85%e6%96%87%e6%9c%ac http://baike.baidu.com/item/%e5%bc%80%e6%ba%90 http://baike.baidu.com/item/%e5%bc%80%e6%ba%90 http://baike.baidu.com/item/%e8%af%ad%e6%b3%95 http://baike.baidu.com/item/c%e8%af%ad%e8%a8%80 http://baike.baidu.com/item/java http://baike.baidu.com/item/perl http://baike.baidu.com/item/%e7%a8%8b%e5%ba%8f http://baike.baidu.com/item/html http://baike.baidu.com/item/html 3.智联网的核心技术：信息扩散技术 针对具体问题，能拥有信息和经验参与智联网 解答问题的人员并不多。大多数情况下，智联网是 在小样本条件下进行工作。这与欧盟第七框架计划 大型项目 integ-risk 推出的欧洲新兴风险雷达---风 险滴嗒器 [14]---大不相同。 风险滴嗒器也是一个计算机网络系统平台，但 它是通过网络爬虫实时抓取互联网中的灾害报道并 计算出风险指数，它还能通过推特(twitter)中某地对 一种传染病的讨论人数多少来评估此种传染病在该 地的风险。讨论的人数越多，被认为风险水平越高。 风险滴嗒器追踪的是宏观风险，使用的是互联网中 大量的信息。依赖网络大数据的风险滴嗒器，监测 不了微观的风险。 例如，对于一个社区的火灾风险，只有发现了 火灾隐患的少数居民和相关管理人员，才能提供有 价值的信息和有针对性的经验，智联网中的数学模 型才能计算出该社区的风险。而社区外议论火灾风 险的人再多，网络数据再丰富，对该社区的实时风 险监测，提供不了任何帮助。智联网驱动的风险雷 达，是为某一区域专门研制的网络平台，配备有该 区域的基础数据库，只关注该区域的风险，追踪的 是微观风险。由于大多数风险源显露时，区域中往 往只有几个人发现，为实时而快捷地追踪风险，须 在只有少量信息的情况下做出风险评估。少量信息 进入评估模型，我们就遇到了小样本问题。 智联网对问题解答的质量由微观、实时和快捷 来决定。这就要求智联网必须具备处理小样本的能 力。网络中的大数据，只具有补充和改善基础数据 库的作用。 信息扩散原理保证了信息扩散技术能优化处理 小样本，显著提高分析精度。信息扩散的基本思想 是把一个传统的数据样本点变成一个模糊集合。由 于信息扩散的目的是挖掘出尽可能多的有用信息， 以此提高系统识别的精度，这种技术，也被称为模 糊信息优化处理技术。最简单的扩散技术是线性信 息分配方法，应用最广泛的扩散技术，是正态扩散。 令 }...,,,{ 21 nxxxd = 是一个给定的样本点集 合， }{uu = 是它的论域（定义域）。将 nxxx ...,,, 21 视为物理化学意义上的分子，将 u 视为分子扩散的 空间，模仿分子扩散，可以给出信息扩散方程，得 到一个扩散函数： .,], 2 )( exp[ 2 1 ),( 2 2 uudx h ux h ux ∈∈ − −= p µ (1) 由于 ( , )x uµ 表达的，给定样本点 x 沿论域点 u 进行 扩散的量值分布，正好与概率论中的正态分布相同， 所以 ( , )x uµ 被称为正态扩散函数。其归一化的模糊 集由式（2）表达。 .,], 2 )( exp[)( 2 2 uudx h ux ux ∈∈ − −=µ (2) 用式（1）和式（2）对样本点 x 在 u 上进行的 信息扩散处理，都称为正态信息扩散。由于对模糊 集样本点进行统计分析时，须进行归一化处理，让 各个模糊集样本点具有相同的统计地位，所以，式 （1）中的第一个因子不影响统计结果。通常提及的 正态信息扩散，是指用式（2）进行的扩散。 对正态扩散的计算机仿真实验表明，当每一个 信息接收点，从邻近两个样本点扩散得到信息量， 等于或多于从其它样本点得到的信息量总和时，扩 散估计的结果最好。这一现象被称为两点择近。根 据两点择近原则和样本点平均距离假设，可以推导 出一个计算扩散系数的简单公式：            ≥−− =− =− =− =− =− =− = 11. 1),)/(2.6851( 10; ),( 0.2986 9; ),( 0.3362 8; ),( 0.3860 7; ),( 0.4560 6; ),( 0.5690 5; ),( 0.8146 nnab nab nab nab nab nab nab h （3） 这里 }.{min},{max 11 i ni i ni xaxb ≤≤≤≤ == 使用信息扩散方法，不需要关于样本点之间内 在联系（诸如概率分布类型等）的任何人为假设， 从而推断出来的结论不会因人而异。 目前，用于智联网的信息扩散技术，是一种统 计样本的集值化技术，旨在提高统计精度。这种技 术的作用域是概率空间，而非物理空间。使用这种 技术，不需要任何空间性质的支持。 人们正在研究的“空间不完备信息的扩散技 术”，试图通过扩散函数与地理空间中自然属性和社 会经济属性的共振，建立弥补地理单元数据缺失的 信息扩散技术，使之比假设空间数据连续而插值的 数据更符合实际，比诸如反距离移动平均法、反距 离移动表面法和普通克里格插值法等得到的数据都 更具可解释性。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 150 4.智联网与网络数学 智联网的核心，不是商业模式，不是互联网技 术，更不是数学模型的复杂度，而是能否集小智慧 为大智慧。 举例来说，10 个刚出道的风险投资人，对一个 项目的风险评估，各人都会有明显的考虑不到之处。 而一名成功的风险投资家（例如，“股神”巴菲特）， 会考虑得更加周到。是不是综合 10 个新手的评估结 果，就能有所改进了呢？当然不是，众人看走眼的 事，很多。在众人均看错的情况下，无论用什么好 的传统数学模型去综合众人的观点，结果也还是错。 如果有一个平台，能帮助这 10 个新手协同工作，评 估出与成功风险投资家一样的或差不多的结果，这 个平台就具有了集小智慧为大智慧的功能。 作者认为，目前的数学，不是用来表述人类思 维活动的工具，不是智能数学，不能提供集小智慧 为大智慧的数学支持。使用传统数学模型的智联网， 只具有在线整合部分信息和经验，快速做出较合理 判断的能力。社区安全风险雷达，较好地发挥了现 有智联网的这一特长。 在思维科学领域，人们最早关注的是计算思维 (computational thinking)。主流学派是把人脑设想 成一套复杂的开关网络，努力发展数学模型来模拟 大脑思维。人工神经元网络，是较为成功的模型之 一。受此启发，上世纪末，人们发展了大量的生物 计算模型，自然计算（natural computation）概念应 运而生。以模糊逻辑为主的软计算，有力地推动了 计算智能(computational intelligence)的发展。这些用 现有数学原理开发的数学模型，既使模拟了人的某 种思维过程，在认识问题和解决问题方面有所发展， 甚至在计算机上得以实现，但仍是 smart 意义下的 智能，而非 intelligence 意义下的智能，因为这些 数学模型并不具备集小智慧为大智慧的潜能。 人们停留在 smart 水平的根本的原因，不是开 发数学模型的努力不够，也不是计算机还不够发达。 根本的原因是现有的数学原理不支持集小智慧为大 智慧的数学模型的出现。 事实上，人类从第一代数学进化到第二代数学 后，就没有质的进步。本文将研究静态问题的数学 称为第一代数学；研究动态问题的数学称为第二代 数学。只有出现第三代数学，人类才具有研究智慧 问题的数学工具。本文将第三代数学称为智能数学。 当然，并非文[15]认为的，具有格结构的模糊数学， 就是智能数学。模糊数学是第二代数学中一个很小 的部分。 人们从简单的计数到美轮美奂的代数系统，并 坚韧不拔地在证明诸如哥德巴赫猜想等大量关于 “数”的定律，历经几千年，构建了第一代数学； 从牛顿研究天体运动规律的需要，到柯西简洁而严 格地证明了的牛顿-莱布尼茨微积分学，再到充满争 议的主观-非主观概率论，历经三百多年，人们构建 了第二代数学。这两代数学，组成了体系宏大的传 统数学，严密而富有活力，为现代科学和技术的发 展，提供了重要的工具。但是，到目前为止，没有 一门数学学科是因为研究智慧的生成、发展和增强 而发展起来。人们已经满足于用传统数学的良性拓 展来研究天地万物，也包括研究智能问题。 超越 smart 的智能数学，目前只能是一个泛泛 而谈的第三代数学，实质性的探索的工作之一，或 许可以从建立网络数学开始。 目前，无论是 imb 的“watson”，还是在线或非 在线的网页语义分析，抑或是棋技超群的 google 的 alphago，大凡用到数学，无一不是以传统数学为基 础。没有一门数学学科是因为研究互联网中的各种 现象而发展起来的。 然而，正如将传统数学的十进制数用于计算机 设计，电路很难实现一样，以传统数学为基础来研 究互联网中的各种现象，注定效率低下。虽然十进 制表述的任何数学问题均可转化为用二进制表述， 成为计算机能读懂的问题，但如果没有二进制，没 有布尔代数，就没有今天的计算机科学。同理，没 有网络数学的出现，基于传统数学的模型，不可能 支持高效的智联网。 新一代数学出现，必然出现一套全新的数学符 号。研究静态问题的第一代数学，其代表性符号是 阿拉伯数字、四则运算符、几何图形、英文字母和 矩阵等。研究动态问题的第二代数学，其代表性符 号是极限、无穷大、无穷小、微分、积分、矢量和 形形色色的空间符号等。 我们不知道研究智慧问题的第三代数学，其代 表性符号将是什么。本文建议，在网络数学中，可 使用互联网中常用的符号，例如“@”和图 6 所示的 “二维码”等。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 151 根据网络数学研究的对象和数学符号变化规 律，我们给出网络数学的定义： 定义 3. 研究互联网世界中各种信息的数量关系和 动态混沌空间，并用互联网中主要符号加以表达的 数学，称为网络数学。 正如第二代数学大量使用第一代数学提供的工 具但与第一代数学有本质区别一样，第三代数学也 必须能够使用第一代数学和第二代数学的工具但与 它们有本质区别。网络数学作为智能数学的一部分， 必须能够使用传统数学的工具，但却不能是传统数 学简单拓展，须有自己的内核。 之所以网络数学有可能成为智能数学的一部 分，是因为互联网中以信息为载体，流动着大量的 人类智慧。这些智慧在互联网中的累积、交溶、嬗 变和服务，形态万千，绝非个体人脑中的智慧形态。 网络数学哪怕触碰到互联网世界中人类智慧的一小 部分，其成就也会远在前述提及的人工智能之上。 一旦人们有了网络数学的帮助，就能设计出挖 掘人类智慧的智联网界面，就能设计出集小智慧为 大智慧的数学模型来，智联网就能被大规模地应用， 智联网也必将成为“互联网+风险分析”的主流。 就目前的各种数学理论而言，因素空间理论有 可能为网络数学的形成提供帮助。 5.基于因素空间理论的网络数学框架 因素空间的最初思想，是汪培庄在解释随机性 的根源及概率规律的数学实质时产生的。1982 年在 文献[16]中给出了一个传统数学意义上的定义。 “因素”是一个泛泛的概念，既指构成事物本 质的成分，也指决定事物成败的原因或条件，又称 因子。生物学中的“基因”，最初就被称为“因素”。 在科学试验中，影响试验指标的要素或原因，称为 因素。 在因素空间理论中，因素的要意被数学符号替 代，因素被数学地定义为一种映射。它把一个对象 f 映射成为一个属性值 xf。例如，身高是一个映射， 它把一个对象“张三”映射成类似“很高”这样的 属性。这里，“身高”就是全面描述人的“因素”之 一。满足一定数学结构的，由诸因素组成的一个集 合族，就称为一个因素空间。形式化定义如下： 定义 4. 令 )}({ lfx f ∈ 是一个因素集合族。如果 l 是一个布尔代数， )1,0,,,,( cll ∧∨= ，且满足下列条 件 ， 则 称 此 集 合 族 为 一 个 因 素 空 间 ， 记 为 ),( lfx f ∈ 。 （1） }{0 φ=x ； （2）若 l 的子集 t 独立（即， t∈∀ βα , ，当 βα ≠ ，必 0=∧ βα ），则 ∏ ∈ = t xx α αγ 这里 }|{ t∈∨= ααγ ，0 和 1 分别是 l 中的最小和最 大元素，π表示笛卡尔乘积。 虽然在过去的 30 多年中，因素空间的研究，一 直在传统数学的框架内，并且以代数系统为主，但 正如模糊集、粗糙集在传统数学的基础上拓展了人 们的视野一样，因素空间理论提供了从“因素”看 问题的视角。有了这种视角，在一个以模糊概念构 成的状态空间中，由信息扩散技术处理观测样本生 成的原始信息矩阵，可以转化为因果型模糊关系 [17] 。 尽管因素空间理论处理实际问题的能力仍停留在传 统数学可以触及的问题上，但这并不意味着因素空 间理论不具有超越传统数学的发展空间。关键是因 素空间理论和方法的研究人员能否挣脱传统思维惯 性的束缚，放弃诸如“因素库”，“知识挖掘算法”， “定性相空间”等的研究，重点放在传统数学难以 发挥作用的问题上，展开全新的研究。建设网络数 学框架，或许是因素空间理论全新发展的第一步。 中国传统数学的基本框架，由方田、粟米、差 分、少广、商功、均输、盈不足、方程、旁要等构 成，包括四则运算法则等。它们在《九章算术》中 有详细介绍。西式的现代数学，其基本框架由代数、 几何和微积分等构成。概率论，也是由这个框架中 的元素构成。集合论，是这个框架的“地基”。动一 动地基，框架就伸出去一大块。模糊数学就是这么 图 6. 表示文字数值信息的“二维码”。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 152 来是。网络数学，不可能由这个框架中的元素构成， 因为集合论的“映射”决定了传统数学只能处理结 构化信息(structured information)，而在网络中流动 的代表人类智慧的很多信息，是非结构化的，它们 不能按预先定义的方式进入网络。现有用传统数学 研制的语义分析工具，对非结构化的柔性信息无能 为力，但人脑却能处理这类信息。例如，“洪水齐腰 深”，没有明确定义，不能形式化，用模糊集表达， 也难尽人意，因为个头高矮的人，齐腰深的深度并 不一样。 基于因素空间理论的网络数学框架，首要目标 是能处理柔性信息，这就注定了不能在传统数学的 框架中打圈圈，首先要从数学符号上有所突破，更 是要设法离开代数系统。为此，我们先定义网络数 学的符号如下： 定义 5. 设@是一个在互联网中通用的符号，称它是 一个网络数学的符号。 例如，微信中使用的图 7 中一些有代表性的符 号，可选作网络数学的符号。图 6 中的“二维码”， 也可作网络数学符号的一员。 我们将网络数学中使用的所有符号组成的集合 记为ω，其内的元素记为@。网络数学中的结构定义 如下： 定义 6. 设ω是ω的一个非空子集，一个能在互联网 中描述ω的某种群体行为的方式，称为一个结构，记 为§(ω)。 例如，以网络数学符号表征因素的一个因素藤， 或许能成为网络数学中的一个结构（图 8）。 因素空间与信息科学的相结合，正在为揭示人 类对客观世界认知的原理，为表达知识，开展网络 背景下的人工智能研究，提供新的数学工具。因素 空间架构体系在智联网上的实现 [9] ，或许意示着因素 空间理论已经开始走向网络数学。 6.网络数学支持的智联网 正如现在的计算机仿真用面向对象技术进行模 块化搭建，不需在底层编程一样，网络化数学将为 人们提供面向智慧的技术，构造智联网不再需要懂 得网络语言，而是用一些模块搭建。有了这些工具， 构造智联网就会变得简单，使用智联网就会便捷。 当因素空间成为网络数学的一部分时，其功能 之一，是用多种形态的因素空间来承担分解和溶合 多个智能体所提供信息中蕴藏的智慧成分，并将小 智慧提升为大智慧，努力达到“三个臭皮匠顶个诸 葛亮”的作用。 不失一般性，我们用图 9 来代表三种基本的因 素空间。图 9（a）中的“藤式因素空间”用于表达 错综复杂的因素关系；图 9（b）的“动作因素空间” 用于表达复杂而有规律（含“随机规律”）的变化事 物；图 9（c）的“趋势因素空间”用于表达因素发 表动态。 正如传统数学的符号举不胜举，但大多是一些 基本符号的变体一样，描述主客观世界的因素空间 符号，应该很多，但也会有一些是基本的。 用图 9 的因素空间符号 ℜℑℵ ,, 加上“@”和“二 维码”，可构造一个简单的，由网络数学支持的智联 网模型。 针对一个具体的问题，我们用@1，@2，...， @n 代表 n 个智能体提供的信息、经验和判断。每一 个@进入智联网，就类似于一封电子邮件发给一位 综合分析大师一样，内含丰富的信息和智慧。 @1，@2，...，@n 经由网络数学中的因素空间 进行分解、溶合，集成智慧，形成产品，其输出用 二维码表示。对图 2 进行拓展，得到一个网络数学 支持的简单智联网，如图 10 所示。 图 8. 因素藤结构。 图 7. 微信中使用的符号。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 153 图 10. 用因素空间分解、溶合并提升智慧的智联网 7. 结论和讨论 智能化风险分析的工作，并非用传统意义上的 人工智能进行，更不是天罗地网的监测系统的自动 化分析，而是依托互联网平台，以集小智慧为大智 慧为准则的智能化工作。因为只有人类才具有“用 有限的知识，认识无限的世界”的动机，才有对“与 不利事件有关的未来情景--风险”展开分析的能力。 今天，商业化的人工智能，是自动化意义下的 “灵巧”（smart）系统，并非能创造知识的智能， 不具有“智慧”（intelligence）。 发端于在线风险分析服务的智联网，追求的是 依托互联网，发挥利益相关者的群体智慧，对具体 的风险进行跟踪、识别、判断和管理，使“互联网+ 风险分析”成为可能。 目前研制出来智联网，受传统数学的限制，只 具有共享信息、实时在线分析的能力，其核心技术 是信息扩散技术。为了提高智联网的能力，须发展 智能数学。网络数学是实质性探索智能数学的开始。 因素空间理论有可能为网络数学的形成提供帮助。 因素空间在智联网中或许能承担分解和溶合多 个智能体所提供信息中蕴藏的智慧成分，并将小智 慧提升为大智慧的任务。 一旦在智能数学的支持下，智联网具有了“三 个臭皮匠顶个诸葛亮”的能力，“互联网+风险分析” 就会成为人们进行风险分析的首选。 参考文献 [1] huang c f. internet of intelligences in risk analysis for online services. journal of risk analysis and crisis response, 2011, 1(2): 110-117. [2] 北京崇安智联科技服务有限责任公司. 高考志愿填报 智联网服务平台—联接智慧,网获知识 (available at: http://www.cazl.cn/ioiserver/index.php?r=site/main), 2013. [3] 艾福利. 自然灾害风险分析智联网服务平台构建与应 用研究. 北京师范大学博士学位论文,2014. [4] 北京崇安智联科技服务有限责任公司. 风险学会“优 秀论文奖”评选系统(available at: http://www.cazl.cn/ risktech/yxlw/),2014. [5] 曾凡雷. 属性拼图智联网风险分析方法及其服务平台 的构建与应用研究. 北京师范大学博士学位论文， 2016. [6] huang c f, wu t, renn o. a risk radar driven by internet of intelligences serving for emergency management in community. environmental research, 2016, 148: 550-559. [7] 苏妩. 信息收集去中心化在风险评估中的应用---海洋 环境风险管理智联网平台研究. 北京师范大学硕士学 位论文,2016. （a）藤式因素空间 （b）动作因素空间 （c）趋势因素空间 图 9. 网络数学中三种基本的因素空间。 internet @2 @3 @1 =ℵ )(ω =ℑ )(ω =ℜ )(ω journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 154 [8] 王蔚丹. 地震宏观异常的综合度量及其在智联网服务 平 台 上 的 应 用 研 究 . 北 京师 范 大 学 博 士 学 位 论 文,2017. [9] 郭君. 自然灾害概率风险时效性评价的因素空间架构 体系及其在智联网上的实现. 北京师范大学博士学位 论文,2017. [10] 梁屹. 网络互动模式下洪水灾害风险沟通共识获得的 方法—基于智联网途径. 北京师范大学博士学位论 文,2017. [11] 刘锋,彭赓,刘颖. 从人脑的结构机理看互联网的进化. 人类工效学,2009,15(1): 11-14. [12]heylighen f. conceptions of a global brain: an historical review, in evolution: cosmic, biological, and social, eds. grinin l e, carneiro r l, korotayev a v and spier f. (uchitel publishing, volgograd, russia, 2011) pp.274-289. [13] huang c f. multiple internet of intelligences for risk analysis. journal of risk analysis and crisis response, 2014, 4(2): 61-71. [14] steinbeis advanced risk technologies. integ-risk riskticker(available at: http://www.integrisk.eu-vri.eu/ #bl2099),2013. [15] 刘云丰,陈洪,刘忠,史光亚. 智能数学:内容、方法及意 义. 自然杂志,1987,(4):255-262. [16] 汪培庄,sugeno m. 因素场与模糊集的背景结构. 模 糊数学, 1982,(2):45-54. [17] 黄崇福,王家鼎. 模糊信息优化处理技术及其应用. 北京:北京航空航天大学出版社,1995. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 146–155 ___________________________________________________________________________________________________________ 155 1.引言 2．智联网原理 3.智联网的核心技术：信息扩散技术 信息扩散原理保证了信息扩散技术能优化处理小样本，显著提高分析精度。信息扩散的基本思想是把一个传统的数据样本点变成一个模糊集合。由于信息扩散的目的是挖掘出尽可能多的有用信息，以此提高系统识别的精度，这种技术，也被称为模糊信息优化处理技术。最简单的扩散技术是线性信息分配方法，应用最广泛的扩散技术，是正态扩散。 令是一个给定的样本点集合，是它的论域（定义域）。将视为物理化学意义上的分子，将u视为分子扩散的空间，模仿分子扩散，可以给出信息扩散方程，得到一个扩散函数： (1) 由于表达的，给定样本点x沿论域点u进行扩散的量值分布，正好与概率论中的正态分布相同，所以被称为正态扩散函数。其归一化的模糊集由式（2）表达。 (2) 对正态扩散的计算机仿真实验表明，当每一个信息接收点，从邻近两个样本点扩散得到信息量，等于或多于从其它样本点得到的信息量总和时，扩散估计的结果最好。这一现象被称为两点择近。根据两点择近原则和样本点平均距离假设，可以推导出一个计算扩散系数的简单公式： （3） 使用信息扩散方法，不需要关于样本点之间内在联系（诸如概率分布类型等）的任何人为假设，从而推断出来的结论不会因人而异。 目前，用于智联网的信息扩散技术，是一种统计样本的集值化技术，旨在提高统计精度。这种技术的作用域是概率空间，而非物理空间。使用这种技术，不需要任何空间性质的支持。 人们正在研究的“空间不完备信息的扩散技术”，试图通过扩散函数与地理空间中自然属性和社会经济属性的共振，建立弥补地理单元数据缺失的信息扩散技术，使之比假设空间数据连续而插值的数据更符合实际，比诸如反距离移动平均法、反距离移动表面法和普通克里格插值法等得到的数据都更具可解释性。 4.智联网与网络数学 5.基于因素空间理论的网络数学框架 6.网络数学支持的智联网 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <feff005500740069006c006900730065007a00200063006500730020006f007000740069006f006e00730020006100660069006e00200064006500200063007200e900650072002000640065007300200064006f00630075006d0065006e00740073002000410064006f00620065002000500044004600200070006f0075007200200075006e00650020007100750061006c0069007400e90020006400270069006d007000720065007300730069006f006e00200070007200e9007000720065007300730065002e0020004c0065007300200064006f00630075006d0065006e00740073002000500044004600200063007200e900e90073002000700065007500760065006e0074002000ea0074007200650020006f007500760065007200740073002000640061006e00730020004100630072006f006200610074002c002000610069006e00730069002000710075002700410064006f00620065002000520065006100640065007200200035002e0030002000650074002000760065007200730069006f006e007300200075006c007400e90072006900650075007200650073002e> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word revised rovira-jracr-castellolí-2011.doc environmental pollution and human health risks near a hazardous waste landfill. temporal trends joaquim rovira, montse mari, marta schuhmacher departament d’enginyeria quimica, universitat rovira i virgili, av. països catalans 26 43007 tarragona, catalonia, spain martí nadal, josé l. domingo laboratory of toxicology and environmental health, iispv, universitat rovira i virgili, sant llorenç 21 43201 reus, catalonia, spain e-mail: marti.nadal@urv.cat abstract the human health risk assessment associated to the human exposure to polychlorinated dibenzo-p-dioxins and dibenzofurans (pcdd/fs) and metals for the population living near a hazardous waste landfill (hwl) in catalonia (spain) is here presented. in summer of 2009 and 2010, two monitoring campaigns were performed by analyzing the levels of pcdd/fs and some elements (as, cd, cr, hg, ni, and pb) in air and soil samples around the hwl. airborne pollutant levels were very low and without tendency found depending the distance to the facility, while higher levels were found in soils inside and close to the facility. anyhow, current health risks associated with those contaminants were within acceptable ranges. notwithstanding, it is recommended to continue with the monitoring program to detect any potential future change in the health risks. keywords: hazardous waste landfill; pcdd/fs; heavy metals; risk assessment; castellolí (catalonia, spain). 1. introduction the european directive 2008/98/ec establishes that the priorities for waste management are prevention, re-use, and recycling. if none of these options are feasible, the following step in the prioritization ranking is incineration with energy recovery (waste-to-energy). however, incineration is not the final step, as some byproducts (i.e., bottom ash, fly ash, and residues from air pollution control devices) may be produced. among those, fly ashes can be especially dangerous as they may concentrate important amounts of metals and organic pollutants such as polychlorinated dibenzo-p-dioxins and dibenzofurans (pcdd/fs).1-3 dioxins and furans are well-known persistent, toxic, and bioaccumulative chemicals, included in the original list of the stockholm convention treaty, which aims at eliminating and/or restricting the environmental release of persistent organic pollutants (pops). because of their accumulation in soils, pcdd/fs may be a problem for environment and human health.4 moreover, ashes may also contain important quantities of elements such as arsenic, cadmium, or mercury, which have been associated with a wide range of toxic effects. incineration ashes, together with other hazardous wastes, may pose substantial or potential threats to public health or the environment.5 therefore, these residues must be properly disposed in controlled sites such as hazardous waste landfills (hwls) in order to assure their harmlessness. in the last decade, the number of studies focused on assessing the potential health effects for people living nearby municipal solid waste incinerators (mswis) and other waste management plants has increased.6 some other investigations have examined the human health effects in relation to residence near landfill sites,7 and findings seem not to journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 13-20 published by atlantis press copyright: the authors 13 administrateur texte tapé à la machine administrateur texte tapé à la machine received 10 october 2011; accepted 12 march 2012 j. rovira, m. mari, m. schuhmacher, m. nadal and j.l. domingo 2 demonstrate an increased risk for the population.8 nevertheless, a great concern still remains on this topic, as there are many uncertainties involved in the assessment of human exposure and health risks. in 1998, a hwl located in the village of castellolí (barcelona, catalonia, spain), which in the past had been used as a chromate-derivatives deposit, started to receive other hazardous materials. fly ashes from municipal solid waste incinerators (mswis) operating in catalonia began to be stored there, together with other solid inert wastes and asbestos. the hwl of castellolí is currently working according to the technical standards established in both the catalan and european legislations regarding sealing criteria, leachate management, and water control. despite the rigorous restraint measurements, the population and local authorities have shown some worries about that facility, including the potential environmental impact and adverse health effects derived from the continued operations in the facility. to address these concerns, in july and december 2007 a surveillance program was initiated. the main goal was to determine the environmental concentrations and human health risks derived from the exposure to metals and pcdd/fs by the population living in the vicinity of the hwl of castellolí.9 air and soil samples were collected at sites located at different distances from the hwl. the levels of metals and pcdd/fs were analyzed. that initial study concluded that the additional risks for the population living nearby the hwl were not significant. nevertheless, a regular monitoring program was recommended in order to assure the proper operation of the facility. when evaluating the environmental impact of industrial facilities, the importance of performing temporal investigations, rather than specific campaigns, has been highlighted.10 moreover, the use of various environmental monitors to provide complementary information is recommended to avoid a possible misinterpretation of the data.11 as result, two new monitoring campaigns were again performed in 2009 and 2010. in this paper, we present the results of the temporal trends in the levels of heavy metals and pcdd/fs in soil and air samples collected around the hwl of castellolí, as well as the trends regarding human health risks. 2. materials and methods 2.1. sampling in july 2009 and 2010, air and soil samples were collected at the same 4 sampling points around the hwl of castellolí, where samples had been collected in july and december 2007.9 during both campaigns, mean temperature was around 25ºc, no rainfall was recorded, and the wind origin was from south, southeast, and west directions. samples were taken in the hwl (on site), in the nearest villages of castellolí (519 inhabitants) and òdena (3,442 inhabitants), and in the village of jorba (823 inhabitants), considered as control. as in 2007, an additional soil sampling point was located in the village of copons (320 inhabitants), considered also as control. geologically, the terrain is dominated by the presence of clay rocks. this last village is unaffected by heavy traffic emissions of a main highway which crosses the area under study. collection sites are shown in figure 1. sampling methods were recently described by rovira et al.12 briefly, air samples were collected by using highvolume active samplers (te-1000 puf for pcdd/fs and te-6070dv for metals adsorbed to pm10 (particulate matter), tisch environmental, cleves, oh, usa). volumes sampled ranged between 600-700 m3 and 2,000-2,300 m3 for pcdd/fs and pm10, respectively. in turn, each soil sample consisted on four subsamples collected within an area of 25 m2, taken from the upper 5 cm of ground and stored in polyethylene bags. the bulked samples were dried at room temperature and sieved through a 2 mm mesh screen until analysis. fig. 1. sampling area. published by atlantis press copyright: the authors 14 human health risks near a hazardous waste landfill 3 2.2. analytical methods the concentrations of arsenic (as), cadmium (cd), chromium (cr), mercury (hg), nickel (ni), and lead (pb) were determined in air and soil samples. metal levels were measured by inductively coupled plasma spectrometry (icp-ms, perkin elmer elan 6000) after sample digestion. for soils, 0.5 g of sample were digested with 5 ml of hno3 (65% suprapur, e. merck, darmstadt, germany) in a milestone start d microwave digestion system for 10 min until reaching 165°c, and kept at this temperature for 20 min. for air, around 6.3 cm2 of each filter was digested with 2 ml hno3 (65% suprapur, e. merck) and 3 ml hf (37.5%, panreac sa, barcelona, spain) in hermetic teflon bombs. in both cases, blank and control samples, as well as reference materials (soil, loamy clay, crm-052, resource technology corporation, laramie, wy, usa), were used to check the accuracy of the instrumental methods. detailed additional information concerning the analytical procedures can be found elsewhere.13 the recovery percentages were 86-124% and 82-101% for soils and air, respectively. on the other hand, pcdd/f concentrations were determined by high-resolution gas chromatography/ high-resolution mass spectrometry (hrgc/hrms), following the german vdi 3499 method. prior to extraction, 13c12-pcdd/f congeners were spiked as recovery internal standards to detect any potential loss during the process. samples were extracted with toluene by accelerated solvent extraction (ase). the extract was subsequently subjected to an acid/base clean-up procedure followed on micro columns of silica gel and alumina. the final extract was again spiked with isotopic labelled internal standards and then analyzed by hrgc-hrms (agilent 6890-waters autospec ultima). the recovery percentage ranges were 58-107% and 59121% for soils and air, respectively. in this study, total pcdd/f concentrations were calculated by using the most recent data of world health organization toxic equivalency factors (who-tefs),14 consequently, levels are given as who toxic equivalents (whoteq). in terms of comparison, values from the legislation and old studies are given in international toxic equivalents (i-teq), as these were calculated according to the old i-tefs. 2.3. human health risks the concentrations of pcdd/fs and metals in soils and air were used to estimate the human exposure and health risks at the different sampling points in/around the hwl in both periods of time (2009 and 2010). three different routes of exposure were considered for the pollutants: soil ingestion, dermal contact, and air inhalation. the equations to calculate exposure and risks were recently reported.15 the non-carcinogenic risks were calculated using the hazard quotient (hq), which is defined as the quotient between the predicted exposure and the respective reference dose. cancer risks were calculated by multiplying the predicted exposure by the respective slope factor. dermal reference doses were calculated multiplying the respective oral reference dose by the gastrointestinal absorption factor, table 1. levels of metals and pcdd/fs in air and soil samples collected around the hwl in july 2009 and 2010. july 2009 july 2010 hwl castellolí òdena jorba copons hwl castellolí òdena jorba copons air as 0.32 0.30 0.28 0.28 na 0.32 0.19 0.22 <0.02 na cd 0.08 <0.01 <0.01 0.05 na 0.06 <0.01 0.06 <0.01 na cr <0.08 <0.08 <0.08 <0.08 na <0.08 <0.08 0.30 2.50 na hg <0.03 <0.03 <0.03 <0.03 na <0.03 <0.03 <0.03 <0.03 na ni 2.04 14.6 0.76 2.08 na 3.48 1.23 3.58 1.07 na pb 4.81 2.70 4.65 3.06 na 3.68 1.28 3.16 1.24 na pcdd/fs 6.70 3.23 3.56 2.45 na 6.93 5.41 4.39 9.86 na soil as 10.2 9.47 1.83 1.75 1.02 8.83 7.64 1.95 3.95 1.18 cd 0.39 0.14 0.07 0.07 0.05 0.60 0.14 0.08 0.07 0.06 cr 27.0 12.7 8.83 11.9 11.3 52.2 16.1 9.18 13.7 9.48 hg <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 <0.10 ni 16.2 12.6 4.18 5.49 3.66 19.1 15.3 4.69 8.90 3.87 pb 50.0 10.5 7.74 9.75 8.80 50.1 10.5 8.44 11.3 8.29 pcdd/fs na na na na na 1.07 0.13 0.20 0.16 0.13 na: not analyzed. units. air: pcdd/fs: fg who-teq·m-3; metals: ng·m-3; soil: pcdd/fs: ng who-teq·kg-1; metals: mg·kg-1 published by atlantis press copyright: the authors 15 j. rovira, m. mari, m. schuhmacher, m. nadal and j.l. domingo 4 while dermal slope factors were calculated by dividing the respective oral slope factor by the gastrointestinal absorption factor.16 reference doses, slope factors and gastrointestinal factors were taken from the risk assessment information system database.17 pcdd/f exposure was estimated by calculating separately the exposure to each one of the 17 pcdd/f congeners and summing the results, which were the basis for the health risk assessment. 3. results and discussion 3.1. air concentrations of metals and pcdd/fs table 1 summarizes the levels of metals and pcdd/fs in air samples collected around the hwl in 2009 and 2010. mercury concentration in all air samples was below its detection limit (<0.03 ng·m-3) in both sampling periods. cadmium was detected only in the hwl (0.08 ng·m-3) and jorba (0.05 ng·m-3) sampling sites in 2009, as well as in the hwl and òdena (0.06 ng·m-3 in both sampling sites) in 2010. chromium levels were below detection limit (<0.08 ng·m-3) in most samples collected in 2009 and 2010, with the only exception of òdena (0.30 ng·m-3) and jorba (blank site) (2.50 ng·m-3) in 2010. in 2009, as levels in air were similar in the four sampling sites, ranging from 0.28 to 0.32 ng·m-3, while in 2010 very similar levels were found everywhere, with the exception of jorba, for which as was undetected. the highest metal concentrations corresponded to ni and pb (14.6 and 4.81 ng·m-3, respectively). anyhow, these levels are still below the guidelines established by the eu legislation for pb, as, cd, and ni, which were set at 500, 6, 5, and 20 ng·m-3, respectively. the variation percentages with respect to data from the baseline survey (july 2007) are shown in table 2. for these calculations, castellolí and òdena were grouped as near sites, while jorba and copons as control sites. heterogeneous fluctuations were observed, and a common increase in the levels of the near-site was not noted. those variations would be more related to point peaks of unknown source associated with local emission sources, rather than due to the influence of the facility. the levels of pcdd/fs in air collected in the 2009 survey ranged from 2.45 to 6.70 fg who-teq·m-3, with a mean concentration of 3.99 fg who-teq·m-3 (table 1). in 2010, pcdd/f values were between 4.39 and 9.86 fg who-teq·m-3, being the arithmetic mean 6.65 fg who-teq·m-3. the concentration of pcdd/fs increased at the hwl sampling site in the periods 20072009 and 2007-2010 (table 2), while different variations were observed in the remaining areas. anyhow, the concentrations within and around the hwl were very low in comparison with recent data from various national and international studies. pcdd/f ranges of 5-7 fg who-teq·m-3, 5-45 fg i-teq·m-3, 3172 fg who teq·m-3, and 22-125 fg i-teq·m-3 were found in ambient air of rural areas of south catalonia, spain, northeast china, and italy, respectively.18-20 table 2. temporal trends of the concentrations of metals and pcdd/fs. 3.2. soil concentrations of metals and pcdd/fs the concentrations of metals and pcdd/fs in soil samples collected in july 2009 and july 2010 in/around the hwl are shown in table 1. in general terms, the hwl sampling site presented the highest metal concentrations. however, all metals were below the most restrictive limits established by the catalan agency of waste to protect human health.21 when compared with the european geochemical atlas developed by eurogeosurveys,22 only the concentrations of cd, cr, and pb inside the hwl were higher than baseline topsoil levels. in the 2009 and 2010 surveys, hg concentrations in all soil sampling points were below its detection limit (<0.10 mg·kg-1). in contrast, pb and cr presented the highest levels. lead concentrations ranged between 7.74 and 50.1 mg·kg-1 in both sampling campaigns, while cr levels ranged from 8.83 to 52.2 mg·kg-1, with slightly higher values in % var july 2007-2009 % var july 2007-2010 hwl near site control site hwl near site control site air as -24 287 367 -24 173 -83 cd 60 900 20 550 cr -99 -99 325 6150 hg ni -42 186 -5 -1 -10 -51 pb 3908 3873 61100 2967 2300 6150 pcdd/fs 72 -9 -37 78 31 153 soil as 357 147 164 296 110 389 cd 22 -16 50 88 -12 63 cr 273 18 41 621 39 41 hg ni 174 28 59 223 52 121 pb 169 -52 7 170 -50 13 pcdd/fs na na na -29 -85 -15 near site: castellolí and òdena; control site: soil: jorba + copons, air: jorba. published by atlantis press copyright: the authors 16 human health risks near a hazardous waste landfill 5 2010. notwithstanding, the concentrations of metals in soils found at the different sites in 2009 and 2010 were similar, being generally higher at the hwl sampling site, followed by castellolí. òdena and the two reference locations (jorba and copons) presented the lowest levels. lead was the only element which did not follow this pattern. lead level in castellolí soils was not different from that found in other sites. when assessing the temporal trends in the soil concentrations, all metals presented important increases in the hwl and the reference sampling sites, being a first indication that these raises could be more associated to meteorological conditions and/or other sources, rather than to the presence of the hwl. finally, moderate decreases were noted for pb and cd at the near sites during the period 2007-2010. because soils are a cumulative matrix, pcdd/f levels were only measured in july 2010. the highest concentrations of pcdd/fs in soils were noted at the hwl sampling site (1.07 ng who-teq·kg-1), while those found in the remaining sites were very similar (range: 0.13-0.20 ng who-teq·kg-1). anyway, the levels of pcdd/fs in soils at the hwl were generally lower than the diverse international screening levels available for pcdd/fs. threshold values of 10 i-teq ng·kg-1 in austria and finland, and 4 i-teq ng·kg-1 in canada,23 have been established to regulate the levels of pcdd/fs in agricultural soils. the pcdd/f congener profile (as contribution percentage) in soil samples collected in the vicinity of the hwl is depicted in figure 2. the hwl sample presented the lowest ocdd contribution (near 30%), in comparison with the other sites where this was the predominant congener (>50%). in contrast, ocdf showed an opposite profile, being proportionally higher inside the landfill. when evaluating the temporal variation between the baseline (2007) and the last (2010) surveys (table 2), a decrease in the pcdd/f levels was noted in all sampling points, which was being especially remarkable at the closest sites. 3.3. principal component analysis (pca) in recent years, the use of statistical and mathematical tools to process amounts of data has become a standard practice in environmental studies. self-organizing maps and principal component analysis (pca) are good examples.24, 25 in the present investigation, a pca was performed in order to compare the profiles of metals and pcdd/fs found in soils of the different sites, and consequently, to investigate the potential influence of the hwl as a pollutant source in the surrounding environment. soil samples were chosen to perform the pca as it is a cumulative matrix, contrasting with air, which is a more variable environmental compartment and highly dependent on the meteorological conditions during sampling. those pcdd/f congeners and metals showing concentrations below their respective detection limits in more than 50% of the samples, were not taken into account. the whole set of soil samples, corresponding to july and december of 2007 and july of 2010, was included. pca provided a threedimensional model with pc1, pc2, and pc3 explaining 60%, 19% and 14% of the total variance, respectively. pc1 was positively correlated with penta-, hexa-, hepta and octa-furans, as well as with hexacdd. pc2 was positively correlated with as, cr, and ni, while pc3 was positively correlated with 1,2,3,4,6,7,8-hpcdd and ocdd. the score plot (figure 3) showed a main cluster containing most of the sampling points, independently of their distance to the hwl. however, those samples collected inside the hwl showed a different pattern, indicating that this particular zone could be somehow impacted by some pollutants. therefore, further studies should corroborate the lack of increase in the environmental burden of pcdd/fs, as this issue could have important implications for the hwl workers potentially exposed to these contaminants. 0 10 20 30 40 50 60 70 80 2 ,3 ,7 ,8 -t c d d 1 ,2 ,3 ,7 ,8 -p e c d d 1 ,2 ,3 ,4 ,7 ,8 -h x c d d 1 ,2 ,3 ,6 ,7 ,8 -h x c d d 1 ,2 ,3 ,7 ,8 ,9 -h x c d d 1 ,2 ,3 ,4 ,6 ,7 ,8 -h p c d d o c d d 2 ,3 ,7 ,8 -t c d f 1 ,2 ,3 ,7 ,8 -p e c d f 2 ,3 ,4 ,7 ,8 -p e c d f 1 ,2 ,3 ,4 ,7 ,8 -h x c d f 1 ,2 ,3 ,6 ,7 ,8 -h x c d f 1 ,2 ,3 ,7 ,8 ,9 -h x c d f 2 ,3 ,4 ,6 ,7 ,8 -h x c d f 1 ,2 ,3 ,4 ,6 ,7 ,8 -h p c d f 1 ,2 ,3 ,4 ,7 ,8 ,9 -h p c d f o c d f % hwl castellolí odena jorba copons fig. 2. pcdd/f congener profile in soil samples collected in 2010 around the hwl. published by atlantis press copyright: the authors 17 j. rovira, m. mari, m. schuhmacher, m. nadal and j.l. domingo 6 3.4. human health risks table 3 shows the exposure to pcdd/fs and metals through 3 routes at the different sampling points. the main pathways of metal exposure were soil and dust ingestion, with percentages of contribution between 67% and 95%, except for as, whose predominant route was dermal absorption (53%). regarding pcdd/fs, only inhalation was evaluated in 2009 since soil concentrations were not determined that year. in 2010, the major route of exposure to pcdd/fs was soil ingestion in the hwl, while air inhalation was the leading pathway in residential sites, with individual contributions from 44% to 76%. inhalation and dermal contact are 2 routes of exposure to pops in general, and table 3. predicted daily exposure for metals (mg·kg-1·day-1) and pcdd/fs (ng who-teq·kg-1·day-1) for the population living around the hwl of castellolí, catalonia, spain. july 2009 july 2010 hwl castellolí òdena jorba copons hwl castellolí òdena jorba copons ingestion as 1.59e-05 1.48e-05 2.86e-06 2.73e-06 1.59e-06 1.38e-05 1.19e-05 3.05e-06 6.17e-06 1.84e-06 cd 6.09e-07 2.19e-07 1.09e-07 1.09e-07 7.81e-08 9.37e-07 2.19e-07 1.25e-07 1.09e-07 9.37e-08 cr 4.22e-05 1.98e-05 1.38e-05 1.86e-05 1.76e-05 8.15e-05 2.51e-05 1.43e-05 2.14e-05 1.48e-05 hg nc nc nc nc nc nc nc nc nc nc ni 2.53e-05 1.97e-05 6.53e-06 8.57e-06 5.72e-06 2.98e-05 2.39e-05 7.32e-06 1.39e-05 6.04e-06 pb 7.81e-05 1.64e-05 1.21e-05 1.52e-05 1.37e-05 7.82e-05 1.64e-05 1.32e-05 1.76e-05 1.29e-05 pcdd/fs na na na na na 8.26e-11 8.09e-12 2.86e-11 3.13e-11 7.86e-12 dermal as 1.81e-05 1.68e-05 3.25e-06 3.11e-06 1.81e-06 1.57e-05 1.36e-05 3.46e-06 7.01e-06 2.09e-06 cd 2.16e-07 7.77e-08 3.88e-08 3.88e-08 2.77e-08 3.33e-07 7.77e-08 4.44e-08 3.88e-08 3.33e-08 cr 1.50e-06 7.05e-07 4.90e-07 6.60e-07 6.27e-07 2.90e-06 8.93e-07 5.09e-07 7.60e-07 5.26e-07 hg nc nc nc nc nc nc nc nc nc nc ni 8.99e-07 6.99e-07 2.32e-07 3.05e-07 2.03e-07 1.06e-06 8.49e-07 2.60e-07 4.94e-07 2.15e-07 pb 2.77e-06 5.83e-07 4.29e-07 5.41e-07 4.88e-07 2.78e-06 5.83e-07 4.68e-07 6.27e-07 4.60e-07 pcdd/fs na na na na na 2.93e-11 2.87e-12 1.02e-11 1.11e-11 2.79e-12 inhalation as 8.77e-08 8.22e-08 7.67e-08 7.67e-08 na 8.77e-08 5.21e-08 6.03e-08 nc na cd 2.19e-08 nc nc 5.48e-09 na 1.64e-08 nc 1.64e-08 nc na cr nc nc nc nc na nc nc 8.22e-08 6.85e-07 na hg nc nc nc nc na nc nc nc nc na ni 5.59e-07 4.00e-06 5.70e-07 2.08e-07 na 9.53e-07 3.37e-07 9.81e-07 2.93e-07 na pb 1.32e-06 7.40e-07 1.27e-06 8.38e-07 na 1.01e-06 3.51e-07 8.66e-07 3.40e-07 na pcdd/fs 2.18e-11 1.93e-11 2.37e-11 1.48e-11 na 3.54e-11 2.31e-11 3.00e-11 4.44e-11 na nc: not calculated, as its concentration was <lod ; na: not analyzed table 4. non-carcinogenic and carcinogenic risks (adimensional) of metals and pcdd/fs. july 2009 july 2010 hwl castellolí òdena jorba coponsb hwl castellolí òdena jorba coponsb hqtotal as 1.34e-01 1.25e-01 3.83e-02 3.74e-02 1.13e-02 1.42e-01 1.17e-01 4.29e-02 5.27e-02 1.58e-02 cd 5.70e-03 1.77e-03 1.37e-03 8.86e-04 6.33e-04 9.80e-03 2.13e-03 1.91e-03 1.06e-03 9.11e-04 cr 3.40e-02 1.60e-02 1.11e-02 1.50e-02 1.42e-02 3.46e-02 1.36e-02 1.12e-02 4.04e-02 8.03e-03 hg nc nc nc nc nc nc nc nc nc nc ni 9.46e-02 5.36e-02 8.98e-02 5.95e-02 5.40e-04 8.22e-02 3.00e-02 8.29e-02 2.56e-02 4.36e-04 pb 2.31e-02 4.85e-03 3.58e-03 4.50e-03 4.07e-03 2.75e-02 5.76e-03 4.63e-03 6.20e-03 4.55e-03 pcdd/fs 1.93e-04a 9.31e-05a 1.02e-04a 7.05e-05a na 3.88e-03 5.72e-04 1.11e-03 1.14e-03 4.34e-04 cancer risk ingestion as 1.02e-05 9.51e-06 1.84e-06 1.76e-06 1.02e-06 8.86e-06 7.67e-06 1.96e-06 3.97e-06 1.18e-06 pcdd/fs na na na na na 9.27e-08 1.13e-08 1.70e-08 1.41e-08 1.12e-08 dermal as 2.72e-05 2.52e-05 4.87e-06 4.66e-06 2.72e-06 1.01e-05 8.72e-06 2.23e-06 4.51e-06 1.35e-06 pcdd/fs na na na na na 3.29e-08 4.01e-09 6.03e-09 5.02e-09 3.97e-09 inhalation as 5.65e-07 5.30e-07 4.95e-07 4.95e-07 na 5.65e-07 3.36e-07 3.89e-07 nc na cd 5.92e-08 1.48e-08 3.70e-08 1.48e-08 na 4.44e-08 nc 4.4e-08 nc na cr (vi) nc nc nc nc na nc nc 1.73e-06 1.44e-05 na ni 2.01e-07 1.44e-06 2.05e-07 1.25e-08 na 3.43e-07 1.21e-07 3.53e-07 1.06e-07 na pcdd/fs 1.05e-07 5.05e-08 5.56e-08 3.83e-08 na 1.08e-07 6.85e-08 8.44e-08 1.54e-07 na nc: not calculated, pollutant below detection limit; na: not analyzed b published by atlantis press copyright: the authors 18 human health risks near a hazardous waste landfill 7 chlorinated compounds in particular.26 notwithstanding, it must be highlighted that dietary intake is the major route of exposure to pops, especially through consumption of fish and seafood from contaminated waters.26 considering recent data on the human intake through food for the catalan population,10 the environmental exposure to pcdd/fs and metals here calculated would only mean less than 0.01% of the total exposure. the carcinogenic and non-carcinogenic risks of exposure to heavy metals and pcdd/fs are summarized in table 4. the non-carcinogenic risks for metals and pcdd/fs, expressed as hazard quotient (hq), did not exceed the safety value of 1 in any case. for metals, the highest hq corresponded to as in the hwl, with values of 0.134 and 0.142 in 2009 and 2010, respectively. as for pcdd/fs, the highest hq also corresponded to the hwl, ranging from 1.93·10-4 to 3.88·10-3 in 2009 and 2010, respectively. the carcinogenic risks derived from exposure to metals and pcdd/fs were below the threshold recommended by the spanish legislation, which is 10-5. the only exceptions were cr inhalation in jorba (control site) in 2010, as well as the as dermal absorption and soil ingestion in the hwl in both sampling periods. anyhow, they were within the range 10-6-10-4, which is considered as assumable taking into account the differences between individuals.27 in addition, two important aspects should be highlighted: 1) although it is generally known that inorganic as is the only carcinogenic species, risk calculations were performed by using total as concentration; 2) background as levels in soils are extremely affected by the regional geology.28 with respect to cr, total levels were also evaluated, but a slope factor is only defined for cr(vi), the carcinogenic form. in this case, the risk was estimated by considering that the level of hexavalent chromium was 1/6 of the total cr.29 the above results show that there were not differences in air samples between metal (as, cd, cr, hg, ni, and pb) and pcdd/f levels measured at the hwl and the near/blank sampling sites. in turn, metal and pcdd/f soil levels found inside the hwl were slightly higher than those observed in the other sampling points. anyway, pollutant levels in samples collected inside and outside the landfill presented different profiles. it would indicate that the hwl is not an important release source of these pollutants in the area under its direct influence. moreover, the (carcinogenic and non-carcinogenic) health risks for the local residents were considered as acceptable according to national and international regulations. in spite of that, a continuous monitoring is recommended to detect any possible increase in the environmental levels of metals and pcdd/fs, which might be of concern for both the hwl workers and the population living in the vicinity of the plant. acknowledgments this study was financially supported by atlas gestión medioambiental, s.a., spain. references 1. m. cobo, a. gálvez, j. a. conesa and c. montes de correa, characterization of fly ash from a hazardous waste incinerator in medellin, colombia, j. hazard. mater. 168 (2009) 1223-1232. fig. 3. pca a) pc1 vs. pc2, and b) pc1 vs. pc3 of metals and pcdd/f congeners in soil. 2007d: samples collected in december 2007; 2007j: samples taken in july 2007. published by atlantis press copyright: the authors 19 j. rovira, m. mari, m. schuhmacher, m. nadal and j.l. domingo 8 2. t. chung, c. liao and g. chang-chien, distribution of polycyclic aromatic hydrocarbons and polychlorinated dibenzo-p-dioxins/dibenzofurans in ash from different units in a municipal solid waste incinerator, waste manage. res. 28 (2010) 789-799. 3. j. yao, w. li, q. kong, y. wu, r. he and d. shen, content, mobility and transfer behavior of heavy metals in mswi bottom ash in zhejiang province, china, fuel 89 (2010) 616-622. 4. m. schuhmacher and j. l. domingo, long-term study of environmental levels of dioxins and furans in the vicinity of a municipal solid waste incinerator. environ. int. 32 (2006) 397-404. 5. s. uğurlu and c. kahraman, fuzzy multicriteria risk assessment for hazardous waste management: the case of istanbul, j. risk anal. crisis response 1 (2011) 29-41. 6. l. giusti, a review of waste management practices and their impact on human health, waste manage. 29 (2009) 22272239. 7. m. vrijheid, health effects of residence near hazardous waste landfill sites: a review of epidemiologic literature, environ.health perspect. 108 (2000) 101-112. 8. e. davoli, e. fattore, v. paiano, a. colombo, m. palmiotto, a. n. rossi, m. il grande and r. fanelli, waste management health risk assessment: a case study of a solid waste landfill in south italy, waste manage. 30 (2010) 16081613. 9. m. mari, m. nadal, m. schuhmacher and j. l. domingo, exposure to heavy metals and pcdd/fs by the population living in the vicinity of a hazardous waste landfill in catalonia, spain: health risk assessment, environ. int. 35 (2009) 1034-1039. 10. m. nadal, m. schuhmacher and j. l. domingo, longterm environmental monitoring of persistent organic pollutants and metals in a chemical/petrochemical area: human health risks, environ. pollut. 159 (2011) 1769-1777. 11. m. nadal, m. mari, m. schuhmacher and j. l. domingo, multi-compartmental environmental surveillance of a petrochemical area: levels of micropollutants, environ. int. 35 (2009) 227-235. 12. j. rovira, m. mari, m. nadal, m. schuhmacher and j. l. domingo, environmental monitoring of metals, pcdd/fs and pcbs as a complementary tool of biological surveillance to assess human health risks, chemosphere 80 (2010) 11831189. 13. j. rovira, v. linares, m. bellés, m. nadal and j. l. domingo airborne levels of uranium in the surroundings of various industrial facilities: human health risks, j. risk anal. crisis response 1 (2011) 42-47. 14. m. van den berg, l.s. birnbaum, m. denison, m. de vito, w. farland, m. feeley, h. fiedler, h. hakansson, a. hanberg, l. haws, m. rose, s. safe, d. schrenk, c. tohyama, a. tritscher, j. tuomisto, m. tysklind, n. walker, r.e. peterson, the 2005 world health organization reevaluation of human and mammalian toxic equivalency factors for dioxins and dioxin-like compounds. toxicol. sci. 93 (2006) 223-241. 15. j. rovira, m. mari, m. schuhmacher, m. nadal and j. l. domingo, monitoring environmental pollutants in the vicinity of a cement plant: a temporal study, arch. environ. contam. toxicol. 60 (2011) 372-384. 16. us epa, risk assessment guidance for superfund volume i: human health evaluation manual, united states environmental protection agency, epa/540/1-89/002, 1989. 17. rais, the risk assessment information system, available at: http://rais.ornl.gov/. accessed: 7 october 2011. 18. j. rovira, m. mari, m. nadal, m. schuhmacher and j. l. domingo, partial replacement of fossil fuel in a cement plant: risk assessment for the population living in the neighborhood, sci. total environ. 408 (2010) 5372-5380. 19. e. abad, k. martínez, l. gustems, r. gómez, x. guinart, i. hernández and j. rivera, ten years measuring pcdds/pcdfs in ambient air in catalonia (spain), chemosphere 67 (2007) 1709-1714. 20. m. onofrio, r. spataro and s. botta, the role of a steel plant in north-west italy to the local air concentrations of pcdd/fs, chemosphere 82 (2011) 708-717. 21. acr, soil reference levels. agència catalana de residus. generalitat de catalunya (in catalan). available at: http://www20.gencat.cat/portal/site/arc/, accessed: 2 june 2011. 22. eurogeosurveys, geochemical atlas of europe, available at: www.gsf.fi/publ/foregsatlas/index.php. accessed: 7 october 2011. 23. us epa, review of international soil levels for dioxin. united states environmental protection agency, 9200.3-54. oswer, 2009. 24. r. carafa, l. faggiano, m. real, a. munné, a. ginebreda, h. guasch, m. flo, l. tirapu and p. carsten von der ohe, water toxicity assessment and spatial pollution patterns identification in a mediterranean river basin district. tools for water management and risk analysis, sci. total environ. 409 (2011) 4269-4279. 25. m. mari, m. nadal, m. schuhmacher and j. l. domingo, application of self-organizing maps for pcdd/f pattern recognition of environmental and biological samples to evaluate the impact of a hazardous waste incinerator, environ. sci. technol. 44 (2010) 3162-3168. 26. d.o. carpenter, polychlorinated biphenyls (pcbs): routes of exposure and effects on human health, rev. environ. health 21 (2006) 1-23. 27. us epa, soil screening guidance: user’s guide. united states environmental protection agency, epa/540/r-96/018, 1996. 28. x. martinez-lladó, m. vilà, v. martí, m. rovira, j. a. domènech and j. d. pablo, trace element distribution in topsoils in catalonia: background and reference values and relationship with regional geology, environ. eng. sci. 25 (2008) 863-878. 29. m. nadal, m. schuhmacher and j. l. domingo, metal pollution of soils and vegetation in an area with petrochemical industry, sci. total environ. 321 (2004) 59-69. published by atlantis press copyright: the authors 20 atlantis press journal style earthquake insurance, catastrophe insurance and earthquake risk & loss: a comparative analysis based on citespace cong lang 1 , xinyan wu 1, *, haoyu wu 2 1 institute of geophysics, china earthquake administration, beijing 100081, china 2 shanxi earthquake agency，taiyuan 030021, china received october 1, 2018 accepted october 25, 2018 abstract china suffers severe earthquake disasters. earthquake insurance can effective transfer and disperse earthquake risk. this paper compared the keyword cluster and institution network of earthquake insurance, other catastrophe insurance and earthquake risk & loss. the keyword cluster showed that earthquake insurance and other catastrophe study relied on non-engineering approach, such as institutions, policy and risk management. while study about earthquake risk & loss mainly relied on engineering approach. besides, the institution network of earthquake insurance and earthquake risk & loss indicated that there was a lack of institutional cooperation and gaps between natural science and social science. as earthquake insurance is a multidiscipline reform of risk management, earthquake insurance study should need multidiscipline studies. keywords: earthquake insurance, catastrophe insurance, earthquake risk & loss, citespace, keyword cluster. 地震保险、巨灾保险和地震风险损失——基于 citespace 的比较分析 * 郎 从 1，吴新燕 1,*，吴昊昱 2 1 中国地震局地球物理研究所, 北京 100081, 中国 2 山西省地震局, 太原 030021, 中国 摘 要 中国地震灾害频繁而强烈，地震保险是常用风险转移工具之一，本文使用 citespace 分析“地震保 险”、“其他巨灾保险”、“地震风险和损失”三大研究主题的研究文献。关键词共现分析表明，地震巨 灾保险和其它巨灾保险研究较注重制度等非工程方面的研究，相关研究较少结合工程技术。而地震风险和 损失研究则多注重工程技术，较少结合制度、风险管理等非工程防御手段。研究机构共现网络分析则表 明，地震巨灾保险和地震风险和损失研究存在机构交流合作少、学科割裂的现象。因此，应结合工程和非 工程的研究，进行跨学的巨灾保险和地震风险损失研究。 关键词：地震保险，巨灾保险，地震风险和损失，citespace，关键词聚类 1. 引言  中国地震灾害频繁而强烈 [1]-[2] 。作为世界第一 人口大国，第二经济大国，中国面临极高的地震风 险，根据 gar2009 报告，中国具有全球最高的地震 * corresponding author: wuxy1977@sina.com 人口死亡风险指数 [3] 。地震给我国造成巨大的损 失，历史上死亡人口最多的地震就发生在中国， 1556 陕西地震造成约 83 万人死亡 [4]-[6] 。改革开放 后，地震也给我国带来巨大的人口伤亡和经济损 失，2008 年汶川地震造成 8 万余人死亡和失踪， 6920 亿元直接经济损失 [7] 。 209 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 保险是进行巨灾风险转移的手段之一。相比财 政拨款、慈善捐款、国际捐款和灾民自救，保险可 以克服国家财政拨款有限，捐赠不可预知的缺点， 在国际上广为应用。 党的十八届三中全会通过的《中共中央关于全 面深化改革若干重大问题的决定》，明确提出“完 善保险经济补偿机制，建立巨灾保险制度”。《国 务院关于加快发展现代保险服务业的若干意见》提 出“建立巨灾保险制度”。2016 年 5 月，中国保监 会、财政部印发了《建立城乡居民住宅地震巨灾保 险制度实施方案》，指出要“结合我国国情，以地 震巨灾保险为突破口，探索建立专项巨灾保险制 度”。我国也在 2015 年 8 月，在云南省大理白族自 治州启动全国首个农房地震保险试点。预计在 3 年 的试点期限内，为大理州所辖 12 县（市）82.43 万 户农村房屋及 356.92 万居民提供风险保障。 目前，国际上开展较多的巨灾保险包括洪水保 险、台风保险和地震保险 [8] 。地震保险是基于地震 风险和地震损失为基础的研究，有必要将之与已有 地震风险和地震损失研究进行比较，了解地震保险 对于基础研究的应用和发展情况，寻找未来可能的 突破点。同时，也有必要与其他巨灾保险对比，了 解巨灾保险的总体发展。 因此本文使用 citespace [9] 对 1980-2018 年的 “地震保险”、“其他巨灾保险”、“地震风险和 损失”三大研究主题的研究文献进行分析，展示研 究领域的热点趋势和主题演进，对比研究差异，寻 找未来研究突破点。 2. 数据来源与方法 citespace 软件 [9] 是文献热点趋势和演进研究软 件，目前国内外学者已应用此方法开展多研究领域 的分析，本文使用此软件对“地震保险”、“其他 巨灾保险”和“地震风险和损失”三大研究主题的 研究文献开展关键词共性分析和时间演进分析，研 究目标的热点趋势和主题演进。 本文指定检索范围为 cnki 期刊数据库中的 “sci 来源期刊”、“ei 来源期刊”、“核心期 刊”和“cssci”。检索时间段为“1990 年-2018 年”数据检索时间为 2018 年 9 月 28 日。 地震保险相关内容的检索条件为标题或关键词 为“地震”和“保险”。得到检索文献共 93 条。 其他巨灾保险检索条件为标题或关键词为“洪 水”和“保险”，标题或关键词为“台风”和“保 险”，标题或关键词为“巨灾”和“保险”且不含 “地震”，得到检索文献共 358 条。 地震风险和地震损失方面的文献检索条件为标 题或关键词为“地震”和“风险”，“地震”和 “损失”，得到检索文献共 367 条。 表 1 三大研究主题检索条件和检索结果 研究主题 检索方式 文献 数目 地震保险 关键词：地震 并含 保险 或篇名：地震 并含 保险 93 篇 其他巨灾保 险 关键词：台风 并含 保险 或篇名：台风 并含 保险 且关键词和篇名不含 地震 358 篇 关键词：洪水 并含 保险 或篇名：洪水 并含 保险 且关键词和篇名不含 地震 关键词：巨灾 并含 保险 或篇名：巨灾 并含 保险 且关键词和篇名不含 地震 地震风险和 地震损失 关键词：地震 并含 风险 或篇名：地震 并含 风险 367 篇 关键词：地震 并含 损失 或篇名：地震 并含 损失 3. 结果和分析 3.1. 关键词共现分析 关键词是对文章主题的高度凝练，本文首先对 所有文献的关键词频率进行分析，统计三大研究主 题文献的关键词频率。表 2 给出三大研究主题文献 的主要关键词对比（为便于研究，去掉了与主题相 同、前后重复和无意义的关键词）。表格中可以看 出，地震保险研究主要围绕地震保险制度、地震风 险度量、使用金融方法进行巨灾风险转移、再保险 对巨灾风险的处理、国外已有经验、费率厘定等内 容。其它巨灾保险研究主要围绕再保险风险分散方 式、制度、建立巨灾保险基金、农业巨灾、债券方 式转移风险、美国已有经验。而地震风险和损失研 究主要围绕风险和损失的评估、各种评估方法和地 震的易损性、危险性、不确定性。 表 2 三大主题文献的关键词前十名统计 排 名 地震保险 a 其它巨灾保险 b 地震风险 和损失 c 1 地震保险制度 巨灾风险 风险评估 2 巨灾风险 再保险 损失评估 3 地震保险基金 巨灾保险制度 地震保险 4 地震风险 巨灾保险基金 地震易损性 5 再保险 农业巨灾保险 巨灾风险 6 日本 巨灾风险管理 增量动力分析 7 风险管理 巨灾债券 评估方法 8 保险费率 分保 层次分析法 9 保险基金 美国 地震危险性 10 巨灾保险制度 风险分散 不确定性 210 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 a 去掉了地震保险、灾害保险、地震灾害、地震、巨灾保 险、中国人民保险公司、人保公司、保险业。 b 去掉了巨灾保险、洪水保险、农业保险、巨灾、防洪保 险、保险、保险制度、灾害保险、保险业、洪水风险、赔 款、巨灾损失、洪水灾害、再保险人。 c 去掉了地震、地震灾害、汶川地震、评估、风险、地震 风险、风险分析、直接经济损失、经济损失、地震损失、 风险评价、损失、地震灾害损失、易损性、滑坡、中国大 陆、间接经济损失、风险感知。 使用 citespace 对关键词的进行共现和聚类分 析，可以确定相关研究的关键词联系，找到核心研 究内容。图 1、图 2、图 3 分别给出地震保险、其它 巨灾保险、地震风险和损失三大主题的关键词聚类 分析。 从图中可以看出，地震保险研究主要围绕地震 保险制度、巨灾基金、免赔额和保费、包头市的案 例、风险转移研究。 其它巨灾保险主要围绕农业巨灾、洪水保险和 洪水保险基金、巨灾保险制度、巨灾债券开展研 究。而地震风险和损失研究主要围绕风险评价或评 估、地震易损性、地震损失和汶川地震的风险和损 失研究几个方面。 3.2. 关键词时间线分析 时间线可以展示关键词的时间演进。图 4、图 5、图 6 分别是地震保险主题文献的关键词共现时间 线图、其它巨灾保险主题文献的关键词共现时间线 图和地震风险损失主题文献的关键词共现时间线 图。 早在 1992 年国务院就通过国办发文的形式明确 地震保险的地位，指出“开展地震保险是实现社会 互助、减轻国家财政负担、提高抗震救灾能力的有 效途径”，而地震保险的研究也在上世纪九十年代 初就已经存在（如图 4 所示）。1996 年丽江 7.0 级 地震造成保险公司偿付能力危机，于是 1996 年中国 人民银行将地震作为一般财产保险的除外责任，另 一方面中国人民银行特别召开全国地震保险座谈 会，并资助《中国地震保险保险费率区划图编制暨 地震保险条款的制定》、《中国地震保险基金的建 立和渠道》、《中国地震再保险运作机制的研究》 三个地震保险课题，对地震保险开展专门研究 [10] 。 自此地震保险的相关研究和讨论逐渐展开。但 19972007 年，地震保险相关研究保持平稳，期间仅有少 数新增关键词出现。2008 年，中国经历改革开放以 来最大损失的地震——汶川地震，它给我国人民的 财产和生命带来了极大的损失。以此事件为标志， 举国上下开始重视地震灾害及其影响，地震保险也 随之迎来又一个研究高峰。之后地震保险研究进入 持续热烈讨论期，研究内容和主题也不断增加和扩 图 1 地震保险主题关键词共现聚类图 图 2 其它巨灾保险主题关键词共现聚类图 图 3 地震风险和损失研究主题关键词共现聚类图 211 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 充，形式从个人财产保险到对公指数保险，风险分 散方式也扩充到了国际金融领域，模拟方法也从定 性或概率性的方式进化到如蒙特卡罗为代表的随机 模拟阶段。但从整体而言，地震保险相关研究更多 注重于公共管理制度的研究和经济金融方面风险转 移和分担的研究，对于工程性的技术类的研究几乎 没有。 图 5 给出其它巨灾保险文献研究的时间线。农 业巨灾保险是其它巨灾保险一直最为热烈的研究 点，从 1994 年美国农作物保险改革一直持续到现 在。洪水保险则在 1998 年长江中下游洪水灾害的背 景下，在 1998 年成为研究爆发点。巨灾保险制度研 究则在 1997 年开始，而巨灾债券 1995 年美国芝加 哥交易所最先推出，1998 年被介绍到中国学术界。 但总体而言，其它巨灾保险相关研究同样多注重于 制度研究和风险转移和分散的研究，对于工程性的 技术类的结合研究几乎没有。 图 6 给出地震风险和损失主题文献的关键词时 间线。地震风险和损失的研究热点主要集中在风险 评价、风险评估、易损性等方面。研究技术手段多 样，从模糊方法到云模型再到增量动力分析，从概 率性风险分析到解析函数法。研究内容不仅包括地 震本身，也包括地震引起的次生灾害——如滑坡、 地震相关的堤防工程以及水土流失。研究内涵上， 不仅包括工程方面，也包括社会人文方面的风险感 知。但总体而言，对于风险和损失的研究多集中在 技术手段，而风险管理相关研究较少，风险分散研 究几乎没有，社会人文方面涉及很少，只有汶川地 震相关的风险感知研究。 因此，本文认为地震巨灾保险和其它巨灾保险 研究具有注重制度、技术等非工程方面的研究的相 似性；无论是地震巨灾保险还是其它巨灾保险都较 少结合工程技术。另一方面，地震风险和损失研究 则多注重工程技术，较少结合制度、风险管理等非 图 4 地震保险文献关键词时间线趋势图. 图 5 其它巨灾保险文献关键词时间线趋势图. 212 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 工程防御手段。巨灾保险研究和地震风险损失研究 之间跨学科研究较少，缺少融合式发展。 3.3. 机构共现分析 本文进一步对地震保险和地震风险损失两大主 题的研究机构之间的共现关系进行分析。图 7、图 8 分别是地震保险主题文献的研究机构共现图和地震 风险损失主题文献的研究机构共现图。 由图 7 所示，参与地震巨灾主题的研究机构 中，各单位联结较少，只出现了 7 个组团的关系网 络，并且这仅有的 7 个组团网络的联结也都比较分 散，多为两点或三点型，只有国科大-四川大学-同济 大学-上海财经大学关系网络是多合作，且合作的关 系网中，也只有此关系网络是经管类学科和工程类 学科合作，其它组团的关系网络均为单一学科合 作。 如图 8 所示，地震风险和损失的研究也存在类 似现象。首先各个单位的交流合作不多，只有 7 个 关系网络形成，其中只有一个关系网格是 7 点，一 个关系网格是 4 点，且这两个多点网格还未形成闭 合网络，各单位之间缺少交流合作。其次，同样存 在学科割裂的现象。参与地震风险和损失研究的机 构和单位绝大多数都是理工科院校或院系，其中只 有对外经贸大学保险学院，武汉大学经管学院和防 灾科技学院经管系三个非理工科院系。并且这个三 个非理工科院系也和其它单位缺少有效交流，没有 图 6 地震风险和损失文献关键词时间线趋势图. 图 7 地震保险主题文献的研究机构共现图 213 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 形成联系。 因此，本文认为地震巨灾保险和地震风险和损 失研究都存在交流合作少、学科割裂的现象。 4. 结论与展望 本文使用 citespace 软件对“地震保险”、“其 他巨灾保险”、“地震风险和损失”三大研究主题 的研究文献开展关键词共性分析和时间演进分析。 分析表明，地震巨灾保险和其它巨灾保险研究 具有注重制度等非工程方面的研究的相似性；无论 是地震巨灾保险还是其它巨灾保险都较少结合工程 技术。而地震风险和损失研究则多注重工程技术， 较少结合制度、风险管理等非工程防御手段。巨灾 保险研究和地震风险损失研究之间的跨学科研究较 少，缺少融合式发展。 进一步，地震保险和地震风险损失文献的研究 机构共现网络分析则表明，地震巨灾保险和地震风 险和损失研究存在机构交流合作少、学科割裂的现 象。 地震是我国发生频繁且严重威胁国民安全制约 经济社会发展的自然灾害之一。为减少地震风险， 应同时设法减少暴露性和脆弱性，增加社会韧性， 做到真正的融合式发展，使工程措施和非工程措施 相辅相成，降低全社会的地震风险。 5. 致谢 本项目由中国地震局地球物理研究所基本科研 业务经费资助（项目编号 dqjb17c10）。 this project is sponsored by: special fund of the institute of geophysics, china earthquake administration (grant number dqjb17c10). references [1] 胡聿贤.地震工程学(第二版).北京:地震出版社,2006. y. hu, earthquake enginering. 2nd edition. (earthquake publish company, beijing, 2006) [2] 李善邦.中国地震.北京:地震出版社,1981. s. li, earthquake in china. (earthquake publish company, beijing, 1981) [3] unisdr, global assessment report on disaster risk reduction: risk and poverty in a changing climate. (unisdr, geneva, 2009) [4] usgs, historic world earthquakes. (2004) http://earthquake.usgs.gov/earthquakes/world/historihis. php. 图 8 地震风险和损失主题文献的研究机构共现图. 214 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 [5] t. utsu, a list of deadly earthquakes in the world: 1500–2000, in international handbook of earthquake and engineering seismology, (2002) 691-717. [6] noaa, the national oceanic and atmospheric administration (noaa) online catalog. (2010) http://www.ngdc.noaa.gov/hazard/earthqk.shtml. [7] 郑通彦,李洋,侯建盛等,2008 年中国大陆地震灾害损 失述评,灾害学,25(2)(2010)112-118. t. zheng, y. li, j. hou, h. mi, review on earthquake disaster loss in chiness mainland in 2008. journal of catastrophology. 25(2)(2010)112-118. [8] p. grossi, h. kunreuther, catastrophe modeling: a new approach to managing risk. (springer, new york, 2005) [9] c. chen, citespace. http://cluster.cis.drexel.edu/~cchen/citespace/. [10] 邹其嘉,顾建华,我国地震保险的研究与展望,城市与 减灾,(2),(2001),8-11. q. zou, j. gu, research and prospect of earthquake insurance in china. cities and disaster reduction, (2),(2001),8-11. 215 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 209 215 © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 101-110 doi: https://doi.org/10.2991/jracr.k.190703.006; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr spatial and temporal patterns of urban vulnerability in guangzhou bo tang1, , jinan qiu2, jiaying huang1, yuanyuan zhang1 and feipeng qiu1 1. department of resources and the urban planning, xinhua college of sun yat-sen university, guangzhou 510520, china 2. school of geography and planning, sun yat-sen university, guangzhou 510275, china received march 27, 2019 accepted april 29, 2019 abstract the evaluation of urban vulnerability is of great significance to improve the quality of urbanization. an urban vulnerability index system was developed from four aspects of population vulnerability, economic vulnerability, social vulnerability and ecological vulnerability. the spatial and temporal patterns of urban vulnerability in guangzhou were evaluated by a comprehensive index model and gis analysis from three time sections (2005、2010、2014). the results indicated that: (1) economic density, population density, and per capita gdp are the main factors affecting the vulnerability of guangzhou. (2) the population vulnerability is considerably higher for the central group cities than for the peripheral group cities. the economic vulnerability reflects a crisscross pattern, consisting of a central group and a peripheral group. the social and ecological vulnerabilities are higher for the peripheral group than for the central group. (3) the comprehensive vulnerability has a layered structure, with a high vulnerability in the inner ring layer, low vulnerability in the middle layer, and medium vulnerability in the outer layer. keywords: urban vulnerability, guangzhou, comprehensive index model, gis  corresponding author: tballen196@163.com 1. introduction urban vulnerability is an important part of urban sustainable development research. it mainly refers to the city's ability to resist the interference of natural and human factors such as resources, ecological environment, economic and social development in the process of development (fang, wang, & fang, 2016). in addition, new-type urbanization requires the reduction of urban vulnerability and improvement in urbanization quality. the concept of ―vulnerability‖ originated from natural disasters (white & haas, 1975). vulnerability research outside of china has mainly concentrated on two aspects, including the concept’s connotation and the research paradigm. with regard to the concept’s connotation, the research has moved away from the limits of ―internal risk‖, extending to external factors such as nature, food security, society, economy, poverty and development, global environmental change, as well as ecology, resulting in an evolution from single-factor to multi-factor evaluation and from assuming a unitary structure to multiple structures (chambers, 1989; janssena et al., 2006; janssen, 2007; briguglio et al., 2009; gaillard, 2010; li & zhang, 2011; wang et al., 2013). in particular, vulnerability has become a collective concept, which includes the components of risk, sensitivity, adaptation, and response (newell et al., 2005). with regard to the research paradigm, there are five models: the risk hazard model (r-h model; burton, kates, & white, 1978), the pressure and release model (par model; wisner et al., 2004), the hazard of place model (hop model; cutter, 1996), the ―diamond model‖ (cutter, 2003), and the coupling system model (turner et al., 2003). for instance, the coupling system, proposed by turner et al. (2003) considers the degrees of exposure and sensitivity and restoring force to be at the core of vulnerability. in addition, it emphasizes the vulnerability from multiplicity and multi-scale disturbance, highlighting the formation mechanism of system vulnerability (turner et al., 2003). in china, urban vulnerability research has journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 102 concentrated catastrophology and ecology. a number of research studies on urban vulnerability of disaster have investigated areas in the circum-bohai-sea region, the yangtze river delta, and the pearl river delta (shi, 2010; tang, liu, & li, 2013; gang, han, & chu, 2015). in addition, several research studies on the ecological vulnerability of certain regions have involved the area of the yangtze river, dianchi lake in yunnan province, and the tarim river basin in xinjiang (gong et al., 2007; yu, li, & zhang, 2015; shao et al., 2016). in terms of research reviews, the vulnerability of resource cities, social vulnerability about the concepts, characteristics, methods, and preventive measures have been summarized (li, zhang, & cheng, 2008; guo, 2010; wang, fang, & zhang, 2013). in background of the new urbanization process in china, the coordinated and sustainable development of cities is becoming increasingly important (ma, cheng & qi, 2018). 2. study area and data sources 2.1. study area guangzhou is the national central city, the capital of guangdong province, and the core city of the pearl river delta urban agglomeration (figure 1). located at 112° 57'–114° 3' e, 22° 26'–23° 56' n, it covers an area of 7434.4 km2. guangzhou is located in a hilly area, with higher terrain in the northeast and lower terrain in the southwest, in the downstream area of the pearl river. the area has a typical subtropical marine monsoon climate. as a typical megacity, guangzhou is susceptible to two threats: disturbances within the urban system and the potential of an external disaster (e.g., floods and typhoons). because guangzhou carried out a larger reform of administrative divisions in 2015, so this paper adopted the administrative divisions of 2014. 2.2. data sources we used data on population, socio-economics, infrastructure, and other statistics, obtained from the yearbooks of various districts in guangzhou city, the china city yearbook, and the guangdong province construction yearbook. due to the lack of data in some areas, missing data was replaced with the subsequent year. 3. methodology 3.1. index system of urban vulnerability exposure refers to the degree of contact that a person, group or system has with a particular stress or impact factor (sanjeev, william, & dana, 2003). the degree of exposure should not only be related to the characteristics of the hazard factors themselves, such as the frequency, intensity, and duration of disasters, but also have a direct relationship with the population size, age structure, population quality, occupational composition, and the level of economic development. sensitivity refers to the degree of influence of a region affected by stress or disturbance. in the turner framework, it is related to the interaction of figure 1. location of the study sites in southeast china development of guangzhou. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 102 103 concentrated catastrophology and ecology. a number of research studies on urban vulnerability of disaster have investigated areas in the circum-bohai-sea region, the yangtze river delta, and the pearl river delta (shi, 2010; tang, liu, & li, 2013; gang, han, & chu, 2015). in addition, several research studies on the ecological vulnerability of certain regions have involved the area of the yangtze river, dianchi lake in yunnan province, and the tarim river basin in xinjiang (gong et al., 2007; yu, li, & zhang, 2015; shao et al., 2016). in terms of research reviews, the vulnerability of resource cities, social vulnerability about the concepts, characteristics, methods, and preventive measures have been summarized (li, zhang, & cheng, 2008; guo, 2010; wang, fang, & zhang, 2013). in background of the new urbanization process in china, the coordinated and sustainable development of cities is becoming increasingly important (ma, cheng & qi, 2018). 2. study area and data sources 2.1. study area guangzhou is the national central city, the capital of guangdong province, and the core city of the pearl river delta urban agglomeration (figure 1). located at 112° 57'–114° 3' e, 22° 26'–23° 56' n, it covers an area of 7434.4 km2. guangzhou is located in a hilly area, with higher terrain in the northeast and lower terrain in the southwest, in the downstream area of the pearl river. the area has a typical subtropical marine monsoon climate. as a typical megacity, guangzhou is susceptible to two threats: disturbances within the urban system and the potential of an external disaster (e.g., floods and typhoons). because guangzhou carried out a larger reform of administrative divisions in 2015, so this paper adopted the administrative divisions of 2014. 2.2. data sources we used data on population, socio-economics, infrastructure, and other statistics, obtained from the yearbooks of various districts in guangzhou city, the china city yearbook, and the guangdong province construction yearbook. due to the lack of data in some areas, missing data was replaced with the subsequent year. 3. methodology 3.1. index system of urban vulnerability exposure refers to the degree of contact that a person, group or system has with a particular stress or impact factor (sanjeev, william, & dana, 2003). the degree of exposure should not only be related to the characteristics of the hazard factors themselves, such as the frequency, intensity, and duration of disasters, but also have a direct relationship with the population size, age structure, population quality, occupational composition, and the level of economic development. sensitivity refers to the degree of influence of a region affected by stress or disturbance. in the turner framework, it is related to the interaction of figure 1. location of the study sites in southeast china development of guangzhou. the social/human capital (population size, social institutions, and economic structures) and the environmental conditions (turner et al., 2003). with increasing urbanization, there is a clear agglomeration of urbanization, the city’s population, economy, and space. the occurrence of external interference and internal disasters will affect the whole system: secondary disasters caused by the main disaster, indirect losses caused by the direct loss, and a single disaster developing into a variety of disasters. restoring force refers to the ability of the region to resist external risks and internal disturbances without damaging its function and structure (adger & hughes, 2005). resilience becomes the adaptive capacity of the regional response to a disturbance. it is a positive feedback mechanism (liu, shi, & xu, 2006). research on the restoring force includes two parts: engineering resilience and ecological restoring force. engineering resilience emphasizes the maintenance of functional effectiveness. its primary objective is security engineering and has a relatively clear measure. meanwhile, the ecological restoring force emphasizes on retaining the system function. research shows that a scientific and rational green space ecosystem plays an important role in disaster prevention and mitigation in urban areas (holling, 1996; huang et al., 2009; ge et al., 2010). according to the characteristics of the urban area and the mechanisms of vulnerability formation, an urban vulnerability assessment framework is established （tang, liu, & shang, 2012） (figure 2). in this paper, we developed an urban vulnerability index system from population vulnerability, economic vulnerability, social vulnerability and ecological vulnerability (table 1). 3.2. measurement model for urban vulnerability 3.2.1. normalizing of raw data to facilitate the comparison of the indicators in the evaluation index system, it is necessary to standardize each evaluation index one by one, so that the value of each evaluation index is in the range of 0–1. the evaluation indicators can be either a positive or a negative index. positive parameters: a larger value indicates a higher vulnerability. the function is: )/(( minxmaxx)minxixpositivex  (1) negative parameters: a larger value indicates a lower vulnerability. the function is: ))/(( minxmaxxixmaxxnegativex  (2) xi is the nondimensionalized value of the 18 parameters, while xmax and xmin are the maximum and minimum values of the same parameter, respectively. 3.2.2 .determining the weight of the indicator the weight of each parameter is calculated using the topsis method. the main advantages of this method are the ability to describe objectively the importance of the index in the entire evaluation system, weaken the influence of individual anomalies, and avoid the interference of subjective factors of weight. figure 2. framework of urban vulnerability journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 104 table 1. evaluation index system of urban vulnerability target hierarchy parameter hierarchy calculation formula unit population vulnerability population density resident population / area population age structure proportion of population over 60 years of age (%) population education level junior high school or above / total population (%) natural population growth rate population born in that year / total population of the year (‰) population immigration rate number of incoming population / total population (‰) economic vulnerability per capita gdp gdp/total population _ foreign exports foreign exports billion dollars proportion of the increased value of tertiary industry in gdp output value of tertiary industry/gdp (%) consumption level index total retail sales of social consumer goods/gdp (%) economic density gdp/area _ social vulnerability engel coefficient household food expenditure / total expenditure (%) employment rate of urban unemployed unemployed persons employed / registered unemployed (%) thousands of people have hospital beds number of hospital beds / resident population _ investment in fixed assets per capita fixed assets input / resident population _ ecological vulnerability overall energy consumption per gdp overall energy consumption per gdp tons/104 yuan urban centralized wastewater treatment ratio urban centralized wastewater treatment ratio (%) ratio of days with good air quality ratio of days with good air quality (%) green coverage ratio in built-up areas green coverage ratio in built-up areas (%) 3.2.3. comprehensive measurement model based on the data standardization and comprehensive weight, we can determine the vulnerability of target hierarchy (population vulnerability, economic vulnerability, social vulnerability, and ecological vulnerability): 𝑊𝑊𝑉𝑉𝐷𝐷′ 𝑊𝑊𝑖𝑖𝑛𝑛𝑖𝑖 𝑈𝑈𝑖𝑖 (3) where wvd' is the vulnerability of target hierarchy, wi the comprehensive weight value; ui represents the normalized results for the indicators; i is the number of indicators. subsequently, the urban comprehensive vulnerability (wvd) was calculated according to the equation below: 𝑊𝑊𝑉𝑉𝐷𝐷 𝑊𝑊𝑉𝑉𝐷𝐷′𝑝𝑝𝑖𝑖 (4) where wvd' is the vulnerability of target hierarchy, p is the number of target hierarchy (p = 4). 4. results 4.1. analysis of the contribution rate of urban vulnerability index as shown in figure 3, the contribution rate of economic density and population density is the highest at approximately 0.15, with a gradual upward trend between 2005 and 2014, indicating that the urban population and the degree of economic agglomeration are the most important perturbation factors for urban risk, per capita gdp, and per capita investment in fixed assets. the proportion of tertiary industry and the number of hospital beds per capita are important negative indicators, with values in the range of 0.05–1, journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 104 105 table 1. evaluation index system of urban vulnerability target hierarchy parameter hierarchy calculation formula unit population vulnerability population density resident population / area population age structure proportion of population over 60 years of age (%) population education level junior high school or above / total population (%) natural population growth rate population born in that year / total population of the year (‰) population immigration rate number of incoming population / total population (‰) economic vulnerability per capita gdp gdp/total population _ foreign exports foreign exports billion dollars proportion of the increased value of tertiary industry in gdp output value of tertiary industry/gdp (%) consumption level index total retail sales of social consumer goods/gdp (%) economic density gdp/area _ social vulnerability engel coefficient household food expenditure / total expenditure (%) employment rate of urban unemployed unemployed persons employed / registered unemployed (%) thousands of people have hospital beds number of hospital beds / resident population _ investment in fixed assets per capita fixed assets input / resident population _ ecological vulnerability overall energy consumption per gdp overall energy consumption per gdp tons/104 yuan urban centralized wastewater treatment ratio urban centralized wastewater treatment ratio (%) ratio of days with good air quality ratio of days with good air quality (%) green coverage ratio in built-up areas green coverage ratio in built-up areas (%) 3.2.3. comprehensive measurement model based on the data standardization and comprehensive weight, we can determine the vulnerability of target hierarchy (population vulnerability, economic vulnerability, social vulnerability, and ecological vulnerability): 𝑊𝑊𝑉𝑉𝐷𝐷′ 𝑊𝑊𝑖𝑖𝑛𝑛𝑖𝑖 𝑈𝑈𝑖𝑖 (3) where wvd' is the vulnerability of target hierarchy, wi the comprehensive weight value; ui represents the normalized results for the indicators; i is the number of indicators. subsequently, the urban comprehensive vulnerability (wvd) was calculated according to the equation below: 𝑊𝑊𝑉𝑉𝐷𝐷 𝑊𝑊𝑉𝑉𝐷𝐷′𝑝𝑝𝑖𝑖 (4) where wvd' is the vulnerability of target hierarchy, p is the number of target hierarchy (p = 4). 4. results 4.1. analysis of the contribution rate of urban vulnerability index as shown in figure 3, the contribution rate of economic density and population density is the highest at approximately 0.15, with a gradual upward trend between 2005 and 2014, indicating that the urban population and the degree of economic agglomeration are the most important perturbation factors for urban risk, per capita gdp, and per capita investment in fixed assets. the proportion of tertiary industry and the number of hospital beds per capita are important negative indicators, with values in the range of 0.05–1, indicating that the regional per capita income level, infrastructure investment and construction, optimization of economic industrial structure, and the improvement of medical facilities play an important role in the reduction of urban vulnerability. although the other factors caused minimal change (less than 0.05), but they cannot be ignored. 4.2. population vulnerability from 2005 to 2014, the total population and population density of guangzhou increased steadily, while the overall education level of the population also increased. with the continuous immigration of people from outside the city and the aging of the population, spatial and temporal differences in the population vulnerability of guangzhou have become obvious. using the natural breakpoint method in gis, this is a statistical classification method that is based on the law of numerical statistical distribution. in this method, the population vulnerability is automatically divided into three levels, as shown in figure 4. in general, population vulnerability is higher in the central cities than in the peripheral cities, especially for the three districts (yuexiu, haizhu, and liwan), which exhibit a high population vulnerability for the three years because of their high population density, constant aging of the population, and high immigration rates. according to the analysis of the demographic data for guangzhou in 2015, the average aging rate of the household population in figure 3. contribution of urban vulnerability indicators figure 4. spatial distribution of population vulnerability in guangzhou in 2005, 2010, and 2014 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 106 the three old districts reached 20%, with the liwan district showing a rate of 21.37%. in addition, the population migration rate indicates that the urban household population of the central group remains between 15% and 25%. as a result, to a certain extent, the increase in the urban vulnerable groups and the management of the floating population intensify the difficulty of the central group to cope with urban risk. despite the relatively low urban vulnerability of the peripheral groups, the general population growth rate is increasing, with the vulnerability of these cities in the later period representing a larger, hidden danger. 4.3. economic vulnerability economic vulnerability mainly focuses on urban economic quality and economic benefits, such as industrial structures, the degree of openness to foreign economies, and the level of consumption. during 2005 to 2014, guangzhou experienced its fastest economic development, with the degree of economic agglomeration enhancing openness to foreign economies, optimizing industrial institutions, and greatly stimulating the per capita gdp and consumption index of guangzhou. however, because of the geographical location of guangzhou district and the economic basis of the differences, the economic vulnerability of guangzhou city exhibits large spatial differences. as shown in figure 5, the medium to high vulnerability is mainly distributed in the central group and the western part. the northern and the central groups have a high vulnerability with upward tendencies, such as in yuexiu and tianhe. the main reason for the high vulnerability values in these areas is the high economic density. the eastern and northern peripheral groups, such as zengcheng and conghua, also show a high vulnerability, which is mainly related to per capita gdp and relatively low openness to foreign economies. low economic vulnerabilities occur mainly in the center of luogang and the outer group of the eastern and southern areas, such as nansha and panyu, due to geographical and policy advantages (technology development zone, nansha new area, ―the belt and road‖). the top three exports from guangzhou in 2014 originate in these three regions. with a low economic density, the economic vulnerability is relatively low. figure 5. spatial distribution of economic vulnerability in guangzhou in 2005, 2010, and 2014 4.4. social vulnerability social vulnerability is an important part of urban vulnerability and reflects the regional social problems from the aspects of income distribution, social security, and social welfare. it also reflects the happiness of a city’s population and the quality of urbanization. figure 6 show that the social vulnerability of guangzhou exhibits a slightly upward trend from 2005 to 2014. while the spatial pattern of the peripheral group is consistently higher than that of the central group. peripheral groups such as conghua, zengcheng, and huadu have a greater deficiency with regard to medical care and medical facilities, with the distribution of income being more unequal. in addition, the engel coefficient is higher, while the social investment in fixed assets is lower, attributing social vulnerability to a high value area. being located in the center of relatively high urbanization, central groups such as tianhe, yuexiu, and haizhu have a relatively greater journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 106 107 the three old districts reached 20%, with the liwan district showing a rate of 21.37%. in addition, the population migration rate indicates that the urban household population of the central group remains between 15% and 25%. as a result, to a certain extent, the increase in the urban vulnerable groups and the management of the floating population intensify the difficulty of the central group to cope with urban risk. despite the relatively low urban vulnerability of the peripheral groups, the general population growth rate is increasing, with the vulnerability of these cities in the later period representing a larger, hidden danger. 4.3. economic vulnerability economic vulnerability mainly focuses on urban economic quality and economic benefits, such as industrial structures, the degree of openness to foreign economies, and the level of consumption. during 2005 to 2014, guangzhou experienced its fastest economic development, with the degree of economic agglomeration enhancing openness to foreign economies, optimizing industrial institutions, and greatly stimulating the per capita gdp and consumption index of guangzhou. however, because of the geographical location of guangzhou district and the economic basis of the differences, the economic vulnerability of guangzhou city exhibits large spatial differences. as shown in figure 5, the medium to high vulnerability is mainly distributed in the central group and the western part. the northern and the central groups have a high vulnerability with upward tendencies, such as in yuexiu and tianhe. the main reason for the high vulnerability values in these areas is the high economic density. the eastern and northern peripheral groups, such as zengcheng and conghua, also show a high vulnerability, which is mainly related to per capita gdp and relatively low openness to foreign economies. low economic vulnerabilities occur mainly in the center of luogang and the outer group of the eastern and southern areas, such as nansha and panyu, due to geographical and policy advantages (technology development zone, nansha new area, ―the belt and road‖). the top three exports from guangzhou in 2014 originate in these three regions. with a low economic density, the economic vulnerability is relatively low. figure 5. spatial distribution of economic vulnerability in guangzhou in 2005, 2010, and 2014 4.4. social vulnerability social vulnerability is an important part of urban vulnerability and reflects the regional social problems from the aspects of income distribution, social security, and social welfare. it also reflects the happiness of a city’s population and the quality of urbanization. figure 6 show that the social vulnerability of guangzhou exhibits a slightly upward trend from 2005 to 2014. while the spatial pattern of the peripheral group is consistently higher than that of the central group. peripheral groups such as conghua, zengcheng, and huadu have a greater deficiency with regard to medical care and medical facilities, with the distribution of income being more unequal. in addition, the engel coefficient is higher, while the social investment in fixed assets is lower, attributing social vulnerability to a high value area. being located in the center of relatively high urbanization, central groups such as tianhe, yuexiu, and haizhu have a relatively greater advantage with regard to social infrastructures, such as fixed assets and per capita hospital beds. it is worth mentioning that the re-employment rate of urban unemployed people in guangzhou remained at approximately 72% in 2014, demonstrating that solving the unemployment problem to improve social security is an effective measure to reduce urban vulnerability. figure 6. spatial distribution of social vulnerability in guangzhou in 2005, 2010, and 2014 4.5. ecological vulnerability the ecological vulnerability of guangzhou city was evaluated based on four aspects: energy consumption, sewage treatment rate, air quality rating, and urban green space rating. in figure 7, we can see that from 2005 to 2014, the vulnerability of the ecological environment in guangzhou decreased slowly, while the area of high vulnerability decreased from four in 2005 to two in 2014. in guangzhou, the variation in values for the management of air, energy, and ecological environment is evident, especially the effects of the centralized sewage treatment. in 2005, the centralized sewage treatment in guangzhou city covered only 71.34%, which increased to 93.22% in 2015. the spatial pattern indicates that the peripheral group has higher values than the central group, which is attributed to the sewage treatment in the central city and the significantly higher energy consumption in the central group. the report ―the overall planning of guangzhou city sewage treatment‖ indicates that the peripheral groups such as panyu, nansha, and huadu are key areas targeted for planning and management. although the overall urban environmental vulnerability of guangzhou has declined, the air quality rate was only 85.5% in 2015, which was clearly lower than the 92% and 97.5% in 2010 and 2005, respectively. in addition, relevant research has shown that urban ecological environment can be maintained when the proportion of green space is greater than 50% (ye, 2001). however, in the built-up areas of guangzhou, this proportion is only 40%, indicating that there is potential for improvement with regard to air quality and urban green space. 4.6. comprehensive vulnerability there are some spatial differences in the comprehensive vulnerability of guangzhou city during 2005–2014 and that the average values are higher for the central group than for the outer group. the urban infrastructure has been improved in recent years, the community service and community security are closer to the people's livelihoods, the environment and the economic structure have been strengthened, and the residents' risk awareness has also improved. however, the urbanization process has resulted in a rapid increase in the urban population of guangzhou’s downtown area, and the population’s mobility and the continuous improvement of economic density have rendered these areas more vulnerable to the city’s internal perturbations and external risks, especially in some districts such as yuexiu, tianhe, and liwan. peripheral groups (such as panyu and nansha in the south, luogang, zengcheng, and conghua in the north, and huadu in the west) have a lower economic development and less urbanization than groups in the central urban district. therefore, if the aggregation is not high, the district will fall into the category of low vulnerability. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 108 figure 7. spatial distribution of ecological vulnerability in guangzhou in 2005, 2010, and 2014 finally, the comprehensive vulnerability of guangzhou city was divided into three classes using a gis breakpoint method. as shown in figure 8, the comprehensive vulnerability of guangzhou city exhibits a circular structure, where the inner circle layer has a high vulnerability, the middle layer has a low vulnerability, and the outer layer has a medium vulnerability. figure 8. spatial distribution of comprehensive vulnerability in guangzhou in 2005, 2010, and 2014 5. conclusions and discussions the comprehensive vulnerability evaluation index is established based on the aspects of population vulnerability, economic vulnerability, social vulnerability and ecological vulnerability. through the calculation of the index weights, it was determined that the main impact factors of urban vulnerability in guangzhou are economic density, population density, gdp per capita, the proportion of tertiary industries, and the investment in fixed assets. these indicators provide an important direction for the management and regulation of guangzhou’s urban vulnerability. during 2005, 2010, 2014, the urban vulnerability of guangzhou underwent some spatial and temporal evolution, which is especially evident in the difference between the inner and outer groups. the population vulnerability is clearly higher for the central group than for the peripheral group, especially in the three older districts. the economic vulnerability follows the same pattern as population vulnerability for the central and peripheral groups, while the social vulnerability is higher for the peripheral group than for the central group. the ecological vulnerability exhibits a slow decline. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 108 109 figure 7. spatial distribution of ecological vulnerability in guangzhou in 2005, 2010, and 2014 finally, the comprehensive vulnerability of guangzhou city was divided into three classes using a gis breakpoint method. as shown in figure 8, the comprehensive vulnerability of guangzhou city exhibits a circular structure, where the inner circle layer has a high vulnerability, the middle layer has a low vulnerability, and the outer layer has a medium vulnerability. figure 8. spatial distribution of comprehensive vulnerability in guangzhou in 2005, 2010, and 2014 5. conclusions and discussions the comprehensive vulnerability evaluation index is established based on the aspects of population vulnerability, economic vulnerability, social vulnerability and ecological vulnerability. through the calculation of the index weights, it was determined that the main impact factors of urban vulnerability in guangzhou are economic density, population density, gdp per capita, the proportion of tertiary industries, and the investment in fixed assets. these indicators provide an important direction for the management and regulation of guangzhou’s urban vulnerability. during 2005, 2010, 2014, the urban vulnerability of guangzhou underwent some spatial and temporal evolution, which is especially evident in the difference between the inner and outer groups. the population vulnerability is clearly higher for the central group than for the peripheral group, especially in the three older districts. the economic vulnerability follows the same pattern as population vulnerability for the central and peripheral groups, while the social vulnerability is higher for the peripheral group than for the central group. the ecological vulnerability exhibits a slow decline. however, it is slightly higher for the peripheral group than for the central group. finally, the comprehensive urban vulnerability of guangzhou exhibits a circular structure, where the inner layer has high vulnerability, the middle layer has low vulnerability, and the outer layer has medium vulnerability. these variations in vulnerability also reflect the characteristics and problems of urban development in guangzhou. based on the results of study, we recommend some actions or measures: a balanced development of regional economy should be achieved through industrial transfer and support for the economic development of outlying urban areas (conghua, zengcheng) to reduce economic density in the future, as mentioned in the draft ―guangzhou city master plan (2017–2035)‖. with regard to population density, the local government should gradually relocate the population of the inner group to the periphery of the city to prevent excessive concentration of population.in view of the unbalanced gdp per capita in guangzhou, the government should transform the economic development model by strengthening institutional and technological innovation, while improving production efficiency. in order to increase the proportion of tertiary industries, the government should reduce industrial dependence on labor and natural resources, while strengthening technological innovation and structural optimization. in terms of investment in fixed assets, it should increase transport, medical care, and infrastructure development, while expanding the basic public service coverage of various areas. these measures will improve guangzhou’s urban resilience and promote sustainable development so that the city can respond better to the challenges arising from the ―one belt and one road initiative‖ and construction of the guangdong-hong kong-macao greater bay area. urban vulnerability is a comprehensive problem, and the regulation path of urban vulnerability is one of the directions to be explored in the future. every city will face various problems at different stages. however, excessive population concentration, irrational economic and urban structure, imperfect infrastructure, and resource-environment disruption are problems that cannot be avoided. acknowledgments we are grateful for the financial support for our initial and ongoing research from guangdong province philosophy and social science planning project (no. gd17ygl10), the youth innovation talent project of department of education guangdong province (no. 2015kqncx226), funding project for scientific and technological innovation and cultivation of college students in 2018 (climbing plan)( no. pdjh0645). references fang, c.l.; wang, y. comprehensive assessment of urban vulnerability and its spatial differentiation in china. journal of geographical sciences, 70(2) (2015) 234-247. white g f, haas j e. assessment of research on natural hazards. ( cambridge: mit press,1975). chambers r, (1989). editorial introduction: vulnerability, coping and policy. institute of development studies bulletin, 2,1–7. janssena m a, schoon m l, ke w et al. scholarly networks on resilience, vulnerability and adaptation within the human dimensions of global environmental change. global environmental change, 3 (2006) 240–252. janssen m a.an update on the scholarly networks on resilience, vulnerability, and adaptation within the human dimensions of global environmental change. ecology and society, 2 (2007) 9. briguglio, l., cordina, g., farrugia, n., & vella, s. economic vulnerability and resilience: concepts and measurements. oxford development studies, 37 (2009) 229–247. gaillard,j.c. vulnerability, capacity and resilience: perspectives for climate and development policy .journal of international development ,22 (2010) 218-232. li h, zhang p.y. research progress and prospective applications of vulnerability approach under global change. progress in geography, 7 (2011) 920–929. wang,y.,fang,c.l.,zhang,q. progress and prospect of urban vulnerability. progress in geography,32 (2013) 755-768. burton, i.; kates, r.w.; white, g. f. the environment as hazard. (oxford university press: usa. 1978). wisner, b.; blaikie, p.; cannon, t.; davis, i. at risk: natural hazards, people’s vulnerability and disaters. (psychology press: london ,2004). cutter, s.l. vulnerability to environmental hazards. progress in human geography, 20 (1996) 529-539. cutter, s.l. social vulnerability to environment al hazard social science quarterly. social science quarterly, 84 (2003) 242-261. turner, b.l.; kasperson, r.e.; matson, p.a.; mccarthy, j.j.; corell, r.w.; christensen, l.; eckley, n.; kasperson, j.x.; luers, a.; martello, m.l.; polsky, c.; pulsipher, a.; schiller, a. a framework for vulnerability analysis in sustainability science. proceedings of the national academy of sciences, 100 (2003) 8074-8079. gang, x.d.; han, z.l.; chu, w.z. assessment of economic system vulnerability in circum-bohai sea region urban agglomeration. journal of liaoning normal university (natural science edition). 38 (2015) 137-144. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 101-110 110 shi, c.h. urban vulnerability assessment research under disaster scenario-in shanghai. east china normal university (doctoral thesis). (2010). tang, b.; liu, x.l.; li, y. hazard vulnerability analysis of the spatial temporal pattern differences in pearl river delta urban agglomerations. economic geography, 33 (2013) 72-78, 85. shao, q.f.; peng, p.h.; huang, j.; liu, z.; sun, x.f.; shao, h.y..remote sensing monitoring of ecological environment vulnerability in the yangtze river upstream anning valley. remote sensing for land and resources, 28(2016)175-181. yu, z.y.; li, b.; zhang, x.s.. space-time evolution and regulation research of ecosystem vulnerability in the society located at lake basin—dian lake, as an example. human geography, 30 (2015)110-116. gong, x.m.; ma, y.; gui, w.d.; yuan, z.y.; lu, g.h. ecological vulnerability assessment research of natural grassland in a typical river basin in arid areas. journal of chengdu university of technology (natural science edition), 34 (2007) 216-220. li, h.; zhang, p.y.; cheng, y.q..the concept of vulnerability and its evaluation method. progress in geography, 27(2008) 18 -25. guo, y. the study of social vulnerability of natural disasters and its influencing factors. journal of catastrophology, 25 (2010) 84-88. wang, y.; fang, c.l.; zhang, q. comment and prospective of urban vulnerability research. progress in geography, 32 (2013) 755-768. ma,l.b.; cheng,w.j.; qi,j.h. coordinated evaluation and development model of oasis urbanization from the perspective of new urbanization: a case study in shandan county of hexi corridor, china, sustainable cities and society,39 (2018) 78-92. sanjeev, k.; william, c.; dana, f.r. from the environment and human security to sustainable security and development. journal of human development, 4 (2003) 289-313. adger, w.n.; hughes, t.p. social-ecological resilience to coastal disaster. science, 309 (2005) 1036-1040. liu, j.; shi, p.j.; xu, w. new progress and comment of restoring force research. advances in earth science, 21 (2006) 211-218. holling, c. s. engineering within ecological constraints －engineering resilience versus ecological resilience. (washington,d.c.: national academies press,31－43. 1996). ge, y.; shi, p.j.; xu,w.; liu.j.; qian,y.; chen, l. the new progress in the study of restoring force and review. journal of catastrophology, 25 (2010) 119-124. huang, p.y.; zhang, j.x.; lu,x.l. urban planning, urban disasters and its prevention. urban planning international, 24, (2009)51-55. godschalk, d.; x, c. urban hazard mitigation: creating resilient cities. urban planning international, 30 (2015) 22-29. cai, h., lam, n. s.-n., zou, l., qiang, y., & li, k. assessing community resilience to coastal hazards in the lower mississippi river basin. water, 8(2) (2016) 46. cutter, s. l. the landscape of disaster resilience indicators in the usa. natural hazards, 80(2)(2016b) 741–758. tang, b.; liu, x.l.; shang, z.h. urban disaster vulnerability and its evaluation index. journal of catastrophology, 27 (2012) 6-11. atlantis press journal style editor’s introduction in risk society, we need more knowledge and innovations to understand complicated questions. in the present issue of journal of risk analysis and crisis response (jracr), volume 4, issue 2 (2014), such efforts have been embodied in these papers. lotfi a. zadeh, creator of fuzzy logic, brings us a new concept in his paper “the crisis of undercoordination”. he views the debt crisis of 2008 as an eruption of the crisis of undercoordination. the paper analyses the crisis of undercoordination in democratic societies due to that there is an imbalance between interdependence and coordination/regulation. we believe that the note will become a new masterpiece by zadeh as his another note “what is soft computing”, published in 1997. in the paper “multiple internet of intelligences for risk analysis”, chongfu huang designs a multiple internet of intelligences to process homological information that is employed to answer a question of an insurance company about the typhoon dynamic risk in wenzhou, china. on the website, the fuzzy mathematics method of normal diffusion is suggested to transform the summaries into fuzzy sets so that a satisfactory answer to the question is given. the third paper “construction of dynamic risk maps for large metropolitan areas” by e.s. guryev, l.v. poluyan and s.a. timashev, describes a methodology used for constructing dynamic risk map for a virtual large “russian gothamrg” city. the methodology for constructing individual risk maps is an important and useful tool for providing insight into every decision made by the decision making persons that govern the large metropolitan areas. in the paper “volatility forecasting in financial risk management with statistical models and arch-rbf neural networks”, dusan marcek and lukas falat, investigate the volatility dynamics of eur/gbp currency. this paper suggests the arch-rbf model that combines information from arch with rbf neural network for volatility forecasting. it also uses a large number of statistical models as well as different optimization techniques for rbf network such as genetic algorithms or clustering. both in-sample and out-of-sample forecasts are evaluated using appropriate evaluation measures. milík tichý discusses risk concepts in paper “analogies in entity risk mechanics”. he analyses the basic analogies between load-subjected structures and hazard-exposed companies. important analogies can be identified in various specific concepts, e.g., in size effect, instability, fatigue, and in many others. the last paper“parts of speech of anquan” by ming xu, zongzhi wu and yun luo is written in chinese with an english abstract. the authors propose a recommendation on revising the parts of speech of anquan in the chinese dictionary, namely, increasing the interpretation of anquan in case of that it is a noun for the demand of application and research in domestic and facilitating international convenient communication and learning. many thanks to referees for their careful reading and commenting, which led our journal to have quality manuscripts. and very much thanks to the authors for all their submissions. editor-in-chief chongfu huang professor, beijing normal university email: hchongfu@bnu.edu.cn director of editorial department junxiang zhang associate professor, huangshan university email: peonyzjx@126.com journal of risk analysis and crisis response, vol. 4, no. 2 (june 2014), 58 published by atlantis press copyright: the authors 58 mailto:hchongfu@bnu.edu.cn title of article (use “title of paper” style) wenzhou real estate bubble and rupture risk warning anping pan1, *,chongfu huang2,dehua jiang3 1college of civil engineering, wenzhou university, wenzhou 325035, china 2academy of disaster reduction and emergency management, faculty of geographical science, beijing normal university, beijing 100875, china 3wenzhou economic broadcasting media group, wenzhou 325000, china abstract it is known to us all that the excessive bubble in the real estate market can seriously affect the normal development of the national economy, which could cause the hidden dangers of financial crisis. in 2011, a wave of defaults destabilized markets across wenzhou, which punctured the property price bubble and caused a painful lesson for the regional economy. therefore, this paper reviews the trend of housing price in wenzhou from 2006 to 2015, and analyzes the causes of house bubble. in addition, the adjustment process of wenzhou real estate is of typical significance, which could provide references for current market trend. keywords: housing bubble, wenzhou, housing price, risk 温州房地产泡沫产生及破裂的风险警示 潘安平 1,* 黄崇福 2 姜德华 3 1.温州大学建筑工程学院，浙江温州 325035，中国 2.北京师范大学地理科学学部减灾与应急管理研究院，北京 100875，中国 3.温州广播电视传媒集团经济广播频率，浙江温州 325000，中国 摘要: 众所周知，房地产市场的过度泡沫化会严重影响国民经济的正常发展，急剧膨胀的房地产泡沫为金融危 机的爆发埋下了众多隐患。2011 年，温州爆发民间借贷危机，刺破了温州房地产价格泡沫，造成区域经济受严 重波及的惨痛教训。本文通过回顾的 2006 年到 2015 年温州房价的变迁走势，分析了温州房地产泡沫产生及破 灭的原因。温州房地产市场的调整过程具有典型意义，可为当前楼市走向提供借鉴和参考。 关键词: 房地产泡沫；温州；房价；风险 * corresponding author: e-mail: pap3888@163.com. 1. 引言 2016 年底以来，中央提出“房子是用来住的， 不是用来炒的”，说明中国政府对房地产泡沫本身 一直保持着警觉的。相对应的是，市场对于我国房 地产行业的房价是否存在泡沫，以及该如何应对潜 在的风险也保持高度关注。随着房地产价格的上 涨，对我国宏观经济运行将产生重要影响。 2017 年以来，全国各地的房价问题热度升温， 一些地区房价上涨过快，越来越多的人加入到了炒 房大军中，危机或许正在逼近。对房地产的泡沫会 否因为过度膨胀而破灭的担心和讨论也越来越多。 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 179 received 14 september 2017 accepted 3 november 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 2017 年 7 月 17 日，人民日报在头版刊发评论员文章 《有效防范金融风险》，文中提到：防范化解金融 风险，需要增强忧患意识。……既防“黑天鹅”， 也防“灰犀牛”，对各类风险苗头既不能掉以轻 心，也不能置若罔闻。文中将房地产泡沫就列为存 在的“灰犀牛”风险隐患。而所谓的灰犀牛，首先 要体量足够大，其发生之后才能造成“影响巨大的 潜在危机”。“灰犀牛事件”则是太过于常见以至 于人们习以为常的风险。如果是从房地产对整个经 济运行的影响程度来看，影响巨大是够得上条件 的。房价会不会崩盘？什么时候会崩盘？这些疑问 如同幽灵般在中国房地产市场上空一直徘徊不去。 大家不禁要问，若房价泡沫是中国目前最大的“灰 犀牛”，那么人人都知道房价泡沫巨大，为什么还 要 抢 着 买 房 呢 ？ 其 实 在 心 态 方 面 ， 面 对 “ 灰 犀 牛”，人们往往会经历“警惕——尝试改变——失 败——危机之前赌一把”的心态变化。房地产可能 带来的危险并不都来源于突如其来的灾难、或者太 过微小的问题，更多只是因为我们长久地视而不 见。从 2016 年 7 月份以来，房地产政策演变来看， 中国房地产泡沫问题被多次提及，可见我国房地产 市场确实存在着一定程度的泡沫，亟需采取相应措 施抑制风险。 面对房地产领域可能存在的“灰犀牛”，我们 只有发现“灰犀牛”、盯紧“灰犀牛”，才谈得上 有效应对“灰犀牛”。纵观中国的房地产发展历 史，从局部情况看，海南和温州等地区均曾经出现 过严重房地产泡沫破灭的情况，造成区域经济受严 重波及的惨痛教训。在温州是真实存在过房地产抽 干实体经济的现象，历史教训值得深思。本文通过 对温州历史上发生的房地产泡沫问题的实证研究， 对当前如何防范泡沫破裂以及该如何正确引导房地 产市场发挥对经济增长的拉动作用具有现实意义。 2. 房地产泡沫的文献综述 房地产泡沫一直以来倍受政府决策人士、学术 界和社会公众的广泛关注,在已有的文献中,学术界 在房地产泡沫的定义、成因等重要问题上仍然存在 争论。campbell 等(1988)给出了一个理性泡沫的明 确定义。allen 等(1991)形成了非完全理性泡沫理 论，该理论以人的行为特征、心理倾向以及主观非 理性预期作为研究对象, 视角更加符合现实。贾生 华等（2014）将行为经济和金融学中非理性泡沫的 研究引入房地产市场, 基于 35 个大中城市的实证研 究后认为，噪声交易者的市场预期的确在很大程度 上决定了房地产市场的泡沫化程度。实证结果显示, 噪声交易者的非理性预期确实是影响房地产价格的 重要因素。当噪声交易者对房地产未来价格有较为 乐观的预期时, 他们会在当期大量地购买房地产, 从而拉动房地产需求的迅速增长, 使得房地产价格 偏离基础价值, 产生了非理性泡沫。在房地产泡沫 的测度方面，洪开荣等(2006)利用物业空置率、gdp 增长率、房地产业增长率以及个人购房比例四个指 标构造出衡量房地产泡沫大小的一个综合指标。王 浩等（2015）认为只有以房价收入比为指标测度我 国 城 市 住 宅 市 场 泡 沫 水 平 具 有 适 用 性 。 李 平 等 （2015）研究表明：研究期内,中国房地产泡沫拉动 了宏观经济增长；基于重要相关产业的分析表明,房 地产泡沫发展至一定程度对建筑业和工业的拉动效 应减弱。王永钦等 (2011)则认为高房价可能导致房 地产部门吸引过多本应用于实体工业部门的投资,造 成实体部门投资不足。就泡沫形成机制方面，李杰 等（2016）认为过度宽松的货币政策在恶性房地产 泡沫产生的过程中扮演了重要的角色。从政策意义 来说,考虑到恶性房地产泡沫可能造成的巨大系统性 风险,我国应当避免以货币宽松的方式来推动房地产 市场的发展。田利辉等（2015）认为银行信贷、gdp 比重和市场化进程是房地产价格泡沫的构成因素。 从金融功能说来看,房地产的金融属性对房地产价格 泡沫具有推动作用；而从金融产品说来看,房地产的 金融属性对房地产泡沫具有抑制作用。此外,房地产 的政治属性和市场化水平对房地产泡沫有明显的推 升效应。张炜（2017）认为引导消费者预期与合理 控制房地产市场信贷量增长是有效抑制房地产泡沫 过快膨胀的有效方法。 3. 温州房地产泡沫的形成 3.1. 2004-2011 年间温州房价快速上涨 在中国经济变迁的版图中，温州曾经扮演过一 个十分特殊的角色，“温州模式”曾经一度风靡大 江南北。在房地产市场上，温州也曾同样创造了很 多传奇，温州炒房团一度横扫包括上海、北京等在 内的多地楼市。与此同时，温州本地的房价也伴随 着开始水涨船高。当市场暗藏巨大的利益时，人们 往往逐利而往，将风险压缩至最小或者忽略。 根据笔者的统计，从 2004 年开始，温州楼市炒 作的现象已表现的十分明显。很多新开的楼盘，开 盘之后，马上遭到爆炒。2004 年温州市区住宅销售 均价在每平米 6000 元左右。随着房价的持续上涨， 越来越多的温州人将房子作为投资品，也吸引众多 追求暴利的炒作资本进入楼市。一些楼盘外加价甚 至高达 20 多万元（由于是期房，买卖双方根本不具 备办理房产证的条件）。2006 年，温州市区房价加 速上升，房价在万元以上的楼盘比比皆是。尽管国 家连续出台宏观调控政策，但还是不能改变当年房 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 180 价的上升趋势。2008 年，温州市区商品住房销售均 价每平米 15477 元。接下来温州楼市的表现只能用 “疯狂”两个字形容。到 2011 年，市区新建商品房 全年均价为每平方米 34674 元。当年 11 月份均价最 高，为每平方米 39623 元。房价最高时的房价收入 比约为杭州市和宁波市的 2 倍。峰值与 2004 年均价 相比，涨幅近 7 倍。 3.2. 2006-2011 年温州房价上涨的原因 造成温州楼市上涨的原因相当复杂。有宏观经 济面的影响，也有温州自身的特殊情况。既有制度 性的，又有发展理念和模式偏差层面的。既有政 府、银行的助推，更有房地产商和投机者“操纵市 场、恶意炒作”方面的。一旦房地产市场脱离经济 发展水平出现“疯涨”，势必激发一些人的赌博心 理。他们（也就是所谓的噪声交易者）的非理性预 期，对房地产未来价格有较为乐观的预期时, 就会 在当期大量地购买房地产, 从而拉动房地产需求的 迅速增长, 使得房地产价格偏离基础价值, 产生了 非理性泡沫。 （1）当时土地出让供应量少。作为浙江人口最 多、城市化率不足的地级市，2005 年至 2009 年温州 市区平均每年供地只有 500 亩左右。以市区房价上 涨较为明显的 2009 年为例，温州市区土地出让总面 积 437.91 亩，住宅建设用地仅 256.74 亩，可供应住 宅建筑面积约 30 万平方米，总体供应量严重匮乏。 土地因此显得稀缺，市场看涨的预期非常强烈。土 地供应稀缺，市区挂牌出让的地块自然也成了抢手 货。温州市区土地出让市场高潮迭起，大部分地块 都要经过十几轮，甚至几十轮竞标才能取得开发 权，溢价率超过 100%、甚至 200%以上的地块也不 在少数。在土地出让市场表现高调的背后，是成规 模的产业资本进入地产开发市场。当时温州民营经 济十分发达，温州作为中国民营经济发展的先发地 区与改革开放的前沿阵地，在改革开放初期，“温 州模式”享誉全国。所谓的“温州模式”就是浙江 省东南部的温州地区以家庭工业和专业化市场的方 式发展非农产业，从而形成小商品、大市场的发展 格局。在温州皮鞋、眼镜、打火机等轻工特色产业 闻名遐迩，电气、鞋业、服装、汽摩配和泵阀等构 成了温州五大传统支柱产业。面对当时红红火火的 房地产市场，许多搞实体产业的企业主纷纷坐不住 了。于是，有一些产业资金抱团组成所谓的“财 团”进军房地产业。比如，名噪一时的“中瑞财 团”就是温州的神力集团、奥康集团、法派集团、 泰力实业等九家制造业企业组成的，据称“可运作 资金超过 100 亿元”。该财团 2004 年成立以后，当 年 11 月，就以 15.3 亿元的夺标价拿下温州江滨路原 东方造船厂地块，进军房地产业。财团演变成了温 州炒房资金一个隐蔽的新出口。 （2）人们投资渠道比较狭窄，楼市是人们闲余 资金比较好的去处。温州本身民营经济极其发达， 民众手里头都有一定的余钱，在通胀压力下面临缩 水风险，因此老百姓的保值增值压力突出。过去民 间就有买金买银，买地买房之说，在温州人的投资 观念中对房地产投资似乎情有独钟，一旦条件成熟 便会显身手。当时，“温州炒房团”在国内闻名遐 迩，甚至成了温州的代名词。炒房是一种市场趋利 行为，当房价上涨时，不仅诱使居民涌入楼市，也 诱导企业涌入楼市。从而引导更多资本“脱实入 虚”，对实体经济形成“挤出效应”。与此同时， 由于成本上升，产业空心化，温州出现了制造业危 机，传统的眼镜，服装，制鞋，打火机等行业风光 不再，企业利润率下降。当时温州流行的一个笑话 就是：丈夫辛辛苦苦办个厂一年赚到的钱，还不如 天天在家打麻将的妻子买几套房赚的钱多。导致很 多人误以为光靠炒房就能一夜暴富了。 （3）房价经历一个从价值再发现到形成泡沫的 过程，房价处于上升周期。2008、2009、2010 年连 续三年，温州市区商品住房销售均价实现了上涨三 连跳，连上三个台阶。分别为 15477 元/平方米、 22042 元/平方米、25032 元/平方米。特别是在 20092010 年间，全国房价经历了一轮上涨，温州上涨势 头更猛，其中有多个月份是领涨全国。比如，2010 年 1 月，70 个大中城市住宅价格指数同比上涨 9.5%，温州同比上涨高达 15.7%，位居第一。值得一 提的是，2009 年 8 月，温州商品住房每平方米销售 均价一跃达到 19860 元，超过北京、上海、深圳， 位列全国各城市第一。根本原因就是住宅供应量 少，需求旺盛。据笔者的统计，2009 年温州市区销 售商品住房面积 106.87 万平方米，而当年可供应的 住宅建筑面积却仅为约 30 万平方米。如此匮乏的商 品住房供应，直接促就了温州建设用地被疯抢并被 抬高的原因所在，无疑也推高了温州市区的商品住 房价格。开发商也很疯狂，当年的 11 月 29 日，位于 温州市区的原温师院地块以总价 37.02 亿的价格被温 州本地的房地产开发企业置信房产竞得，楼面价高 达 3.7 万元/平方米，成为当时全国住宅用地单价地 王。与此同时，受城市区域发展限制等因素影响， 2010 年前后，温州当时的房地产开发主要集中在老 城区、瓯江路沿线、温州新城区域等市中心位置， 为追求利润化，众多开发商纷纷将项目定位成高端 产品。如位于学院路附近的金色尚品，瓯江路沿线 的中瑞曼哈顿、外滩国际公馆、绿城鹿城广场等项 目，套型面积多在 160 平方米以上，有些单套面积 甚至高达五六百平方米。 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 181 （4）民间借贷加杠杆现象严重，充裕的流动性 是房价上涨的主要“推手”。2008 年国家应对国际 金融危机而推出 4 万亿经济刺激政策，导致银根放 松。许多温州企业得到了巨额贷款后，不是进行原 有产业的升级换代，而是流进了楼市，温州实体企 业涉及房地产的投资比重高达 66%。大批企业逐渐 放弃实业，转向投资房地产以及虚拟经济，从而埋 下了危机的伏笔。市场暂时表现出的虚假繁荣又会 引起炒房人士预期房地产未来价格的进一步上升, 这 就形成了一个不断自我强化的正反馈循环, 并最终导 致房地产价格的长时间持续上扬以及房地产泡沫的 形成并持续。当时不少温州人炒房加杠杆的路线图 如下：先是向银行按揭买房，然后再拿着的房子到 银行做消费抵押贷款，接下来再拿贷款的钱买房， 形成了一个“购房—抵押贷款—再购房”的循环。 从事后观察分析，这些行为其实都是在搞击鼓传花 式的博傻炒作，谁接最后一棒，那就成了高山上的 守望者了。一直到 2017 年底，有些楼盘的接盘者还 没有解套。前面提到的原温师院地块，楼面价 3.7 万 /平方米，后来房子建成后，二手房市场楼盘售价也 不到 4 万/平方米。套牢了一大批人。这些过度杠杆 化的房地产市场投机炒作直接导致了 2011 年的温州 金融市场危机。 4. 温州房地产泡沫的破灭 一旦房价上涨就像黑洞一样，吸金无数，再多 的钱投入进去都看不到一丝动静。带来的结果就是 实体企业不但没钱，而且还要把钱倒回房地产，于 是，楼市对土地、资金等主要社会资源的虹吸效应 已经成为实体经济持续低迷的重要原因之一。带来 的结果是房地产绑架地方实体经济就不可避免，造 成全社会对于房价带来的各种收益依赖程度日益加 深。特别是当时温州房地产的各项税收曾经占到了 地方财政总规模的 45%，政府离开房地产，直接就 会面临断粮的风险。 温州房地产泡沫破裂后，房价下跌过程经历了 多个阶段，持续时间很长。从 2011 年 9 月算起，温 州房价经历了连续 40 多个月的下跌，一直到 2015 年 中才触底。温州许多小区的房价与顶峰相比缩水近 50%（如图 1 所示）。 4.1. 温州房价僵持阶段（2010 年初至 2011 年 8 月） 2010 年 4 月 17 日，国务院发布“新国十条”， 当时房价一直居高不下的温州在同年 10 月 12 日出台 了相关意见：暂定本市户籍居民家庭及能够提供 1 年以上本市纳税证明或社会保险缴纳证明的非本市 户籍居民家庭（包括夫妻双方及未成年子女）只能 在本市新购买 1 套商品住房（含二手存量房）。也 就是说，某个居民家庭当时无论名下有几套房，只 能再买一套。从温州一手房市场成交情况看，成交 量走势随政策变化明显。自 4 月国十条限购政策出 台后，市场成交情况逐渐进入颓势。尤其是 4 月份 图 1：2010 年 6 月-2016 年 9 月温州房价走势示意图. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 182 以后，新开盘的商品住房一度出现阶段性“零”供 应的局面，一方面是由于前几年温州的土地供应量 不大；另一方面温州实施的“限购令”政策，也影 响了一批存在“双拼”户型（当时住建部有个规定 就是楼盘开发有个 90/70 政策，凡新审批、新开工的 商品住房建设，套型建筑面积 90 平方米以下住房面 积所占比重，必须达到开发建设总面积的 70%以 上。若开发商开发的是高档楼盘，大多数是面积 90 平方米以上，于是一套房给登记 2 个甚至 2 个以上房 产证。这些有两个以上房产证的户型就是“双拼” 户型）项目无法正常开盘；另外为遏制房价过快上 涨的势头，临时干预了部分高价楼盘定价（比如不 给核发商品房预售许可证），从而影响了这些高价 楼盘的上市进度。 从图 2 中可以明显发现温州房价造顶的过程。 临界点发生在 2010 年 10 月，此时，国家四万亿刺激 计划进入尾声。央行的货币政策转向明显，原来极 度宽松的货币政策转向紧缩。央行先后连续 5 次加 息、12 次提高存款准备金率，结果是银行资金流动 性紧张起来。于是，银行开始抽资压贷，要企业归 还贷款。到了 2011 年，大多数中小企业已经无法从 银行获得贷款，由此爆发了温州民间借贷危机。由 于温州民间借贷盛行“联保联贷”，互相担保，一 家企业倒闭，上下游的一系列贷款者、担保者都将 跟着遭殃。民间借贷就像是一个套环，一环套一 环。无数企业和个人牵扯其中。只要其中有一个环 节断裂，就会形成整个多米诺骨牌连锁反应，进而 造成普遍踩踏事件。当时曾经发生过有些温州的老 板甚至因为钱还不上就去跳楼。毫无疑问，资金危 机也就成为房价下跌的导火索。 2010 年，温州房地产市场表现转冷，虽然商品 房销售面积比上年下降 27.2%，但房价没有转跌，而 是由以前的快速上涨转为小幅波动，房价表现为僵 持状况，该阶段时间跨度 2010 年初至 2011 年 8 月。 4.2. 温州房价快速下跌阶段（2011 年 9 月至 2012 年 11 月） 2011 年，温州房地产市场状况没有好转，反而 进一步趋冷，新建商品房、二手房交易，均出现了 同 比 大 幅 下 降 。 比 如 ， 全 市 商 品 房 销 售 面 积 为 141.71 万平方米又比 2010 年（187.88 万平方米）下 降了 24.5%（如图 3 所示）。 2011 年 1 月 26 日， 国务院常务会议研究部署进 一步做好房地产市场调控工作，再度推出八条房地 产市场调控措施（俗称“新国八条”），其中“限 购”是国八条奏效的关键。文件要求，各直辖市、 计划单列市、省会城市和房价过高、上涨过快的城 市，在一定时期内，要从严制定和执行住房限购措 施。温州作为“房价过高、上涨过快的城市”相应 图 2：2009 年 1 月-2013 年 1 月温州住宅历史单价市场行情. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 183 的出台了《关于落实住房限购政策有关问题的通 知》，当时民间称之为“温州版”限购令。具体规 定如下：2011 年 3 月 14 日之后，本地户籍家庭及纳 税或社保满一年的非本地户籍家庭只能新购买１套 住房，已拥有 2 套及以上住房的本市户籍居民家庭 等暂停购房。该政策限制本地户籍家庭购买第 2 套 住房，这与新“国八条”要求、大多数城市实行的 “限购第 3 套”政策相比，更加严苛。 房地产调控措施特别是限购措施已经在温州发 挥明显作用。在“史上最严调控”的发力下，温州 楼市终于降声四起，进入了“以价换量”的大促销 阶段，前期过热的房价和地价得到回归。长时间的 持续大幅下滑使得房地产企业资金链崩裂，房价的 僵持阶段终于被打破，进入快速下跌的被动调整阶 段。这一阶段的房价下跌幅度为 18.5%，持续时间为 15 个月。 面 对 低 迷 的 市 场 ， “ 购 房 送 宝 马 ” 、 “ 6.8 折”、“保值回购”等等营销玩意破茧而出，完全 颠覆了温州惯用的高调、饥饿式营销方式，楼市已 经进入“竞相降价”的流潮。 在楼市持续调控的背景之下，伴随着房价的下 跌，开发商的信心也降到冰点。表现在温州的土地 出让市场上，2012 年，温州土地成交面积 190.28 万 方（2854.1 亩），成交金额 21.18 亿，同比大幅下降 84%（如图 4 所示）。温州市区出现过连续多月没有 住宅用地成功出让的现象。2011 年 1 月份，计划出 让的 7 宗南湖地块和 3 宗仙岩地块全部遭遇流标，后 来没有办法，政府甚至想出了“勾地”办法，即在 挂牌出让前，先找好买家，来解决土地出让问题。 否则结果只有一个——流拍。 图 3：2010 年-2015 年温州全市商品房供销走势图. 图 4：2010 年-2015 年温州全市土地成交情况. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 184 由于大量房产通过银行抵押贷款购买的，在房 价下跌过程中，断供房不断涌现，密集涌现的时点 是 2011 年民间借贷危机爆发期间。信息首先递延到 银行，银行为了处置这些不良资产，只能起诉到法 院，寻求司法途径处置。自 2012 年 6 月 26 日司法拍 卖在淘宝网上线以来，浙江各级法院纷纷试水，将 司法拍卖搬上了淘宝网，做起“掌柜”，专拍卖房 产和名车。司法拍卖的房源中来自温州的最多。与 传统拍卖相比，网络司法拍卖的零佣金、透明度 高、快捷高效吸引了不少关注者。司法拍卖作为获 得房产的一种“非主流”方式，正逐步被广大市民 所接受。大量的二手房拍卖，对二手房市场造成的 冲击是最大。每当某小区有司法拍卖房挂到网上， 该小区的房价就意味着又要下一个台阶了。 在房价下跌过程中，还引发了“房闹”现象。 温州多个楼盘价格跳水，引发部分已购房业主的不 满。“降价退房”的屡屡上演，当业主遇到损失 后，知道一个人“维权”势单力薄，没有效果。于 是业主们便纠结在一起，采取共同“维权”举措。 好几处楼盘的售楼处发生前期业主和开发企业的纠 纷和冲突，甚至维权行为演变成现场打闹。 4.3. 温州房价下跌过渡阶段（2012 年 12 月至 2013 年 6 月） 古往今来对“安居乐业”的向往，日常生活中 对“衣食住行”的重视，居住条件都是不可或缺的 一环。作为社会中的一员，总得有个遮风蔽雨的地 方。居民购买住房可以满足居住的需要。 因此，当房价下跌时，会有效刺激老百姓的正 常住房消费需求（也就是通常讲的刚需或者改善性 住房需求），再加上全国房地产市场迅速升温，温 州房地产市场也明显好转，商品房销售面积呈现恢 复性增长，2012 年比上年增长 71.8%，房价在 2012 年底开始也由下跌转为稳定，但走势明显弱于全 国。在全国各大城市房价普遍上涨的情况下，温州 的房价仅为稳定。 虽然 2012 年房价处于下跌过程，但是从成交量 来看，相较于 2011 年楼市的全线低迷，2012 年市民 显然对买房卖房的热情增长不少，而且与往年不同 的是，特别青睐“一手房”。温州市房产登记中心 发布的《2012 年温州市区房产交易分析白皮书》显 示，温州市区新建商品房 2012 年成交量为 67.638 万 平方米、5651 套，较 2011 年同期的 11.715 万平方 米、956 套，分别增加了 477.36％、491.1％。成交 金额达 174.882 亿元（如图 5 所示）。 4.4. 温州房价缓慢下跌阶段（2013 年 7 月至 2014 年 6 月） 受资金流动紧张和全国房地产市场大环境的影 响，温州房价在 2013 年 7 月再次转跌，到 2014 年 5 月份末下跌 5.1%，下跌速比前一波慢一些。 面对房价的持续走低以及库存的居高不下的局 面，温州市政府开始出手了。在政策面上，2013 年 8 月，温州对原先严苛的限购政策微调，回归 2011 年新“国八条”框架，规定温州本地户籍家庭可购 买第 2 套住房，意味着已购买１套住宅的本地户籍 家庭，可再购买１套住宅，但名下所有房产只能是 2 套，之前名下无房产的可再购买 2 套。 2013 年温州限购政策微调政策影响，这一阶段 的销售状况继续好转，温州市区当年新开楼盘以较 为合理的价格、日益完善的配套设施和户型设计等 图 5：2010 年-2015 年温州市区住宅供销及价格走势图. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 185 因素，越来越得到市场的认可，多个楼盘热销甚至 重现“日光盘”现象。温州市区 2013 年的商品房销 售面积达到 116.91 万方，创出历史新高，比上年增 长 72.7%（如图 4 所示）。住宅销售均价则由 2012 年的 25886 元/m2 降到了 22541 元/m2。经过长时间调 整，温州房价正在向合理水平回归，但离合理水平 仍有相当差距。2013 年，温州市城镇居民人均住房 建筑面积 41.67 平方米，但即使按 3 口之家 80 平米 计算，房价收入比依然超过 10 倍，远高于合理水 平。商品房去库存的压力仍很大，温州商品房网上 销售管理系统显示，截止 2014 年 6 月 30 日，公开发 售房源 48527 套、6082779 平方米，按 2013 年的销 售面积计算，需要 21 个月才能消化完。 不过，商品房销售形势的旺盛还是带动了土地 出让市场，在土地出让方面终于一改 2012 年的颓 势，全市挂牌出让土地 689.04 万方（10335.5 亩）， 实际成交 578.69 万方（8680.3 亩），可建的建筑面 积达到 1457.29 万平方米（如图 6 所示），为近年来 之最。其中温州市区共出让商品房建设用地 65 宗， 出让土地面积 230.75 万平方米（3401.2 亩），可建 面积 545.3 万平方米；其中出让商品住房建设用地 28 宗，面积 101.94 万平方米（1529.05 亩），可建的 建筑面积 280.23 万平方米，可供应的住宅建筑面积 达到了 229.08 万平方米。 4.5. 温州房价基本企稳见底（2014 年 7 月至 2015 年 6 月） 在 2014 年 3 月召开的全国两会上，李克强总理 在政府工作报告中指出：要针对不同城市的具体情 况实施“因城施策、分类调控”。当年 7 月 29 日， 温州宣布放开限购。具体措施包括在房屋交易登记 时不再核查现有住房情况，在外温州人和新温州人 在温购房享受与温州市本地户籍居民同等待遇。 2015 年，房地产政策利好的频频出台落地。在 全国层面的政策利好有：3.30 房贷新政、二套房首 付降至 4 成、营业税免征期限 5 年改 2 年、央行连续 多次降息、住房公积金新政等。而温州市于 4 月 2 日 出台了《关于促进房地产市场持续平稳健康发展的 若干意见》（俗称为“温八条”房产新政）。其中 明确提出，对首次购买新建普通商品住房的购房者 给予补助，打出政策组合拳，直指“促量稳价”的 目标。在多重政策利好下，温州 2015 年 4 月份价格 指数环比出现了回升，新建商品住宅价格指数环比 上升 0.7%，二手住宅价格指数环比上升 0.4%，房价 基本企稳见底。老百姓入市意愿强烈，2015 年度温 州全市供应商品房面积 429.58 万方，同比下降 32.9%，供应套数 38095 套；全市商品房成交面积 527.35 万方, 同比上涨 16.8%，成交套数 42791 套； 全年商品房供销比为 0.81（如图 3 所示）。 5. 温州房地产泡沫的破灭对当前楼市发展的启示 温州作为中国楼市“风向标”，在历史上曾经 出现过的房价下跌、资本撤离、“弃房”显现等楼 市泡沫破裂迹象，为全国其他城市狂飙发展的楼市 提出警示。 5.1. 房价过度投机炒作是引发下跌的总根源 当前有些城市与当年的温州一样，房地产市场 图 6：2010 年-2015 年温州全市成交土地建筑面积和容积率情况图. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 186 的繁荣主要靠投资需求拉动，而实际需求明显不 足。大量的投机性购房，会放大虚拟需求传递给人 一种虚假的需求信息。若投机需求占比过大，一旦 房价失去上涨动力，房地产市场的大幅波动就无法 避免，房价的剧烈调整也在所难免。因为要想维持 住严重过高的房价，方法只有一个，那就是房价继 续上涨，并且上涨的速度要足够快。房市作为一个 完全依赖于资金推动的市场，这需要源源不断的资 金流入楼市。越来越多购房者现金流面临可持续性 和成长性问题时，楼市的上涨也变得岌岌可危。一 旦资金投入乏力，就犹如人之失血，出现调整也是 理所当然。也就是说，脱离实业的楼市“炒作”只 能是浮萍，一旦投资需求被限制，结局是楼市崩 盘、实业受损。 5.2. 警惕房地产业“绑架”实体经济 经常有人开玩笑说，中国经济就是房地产经 济。为什么老百姓这么疯狂地买房？是因为他们相 信房价会一直涨。对房地产投机的巨额利益，也会 促使企业大举进军这个产业。有的企业通过买楼来 炒房，有的企业干脆抱团组成财团入股投资房地 产。这样的话，房价事实上成为错综复杂民间借 贷、企业互保链条、银行信贷行为相互交织的风暴 眼。一旦地产市场缩水后，就会影响企业抵押物的 市值，必然会导致银行缩小这些企业的信贷规模， “弃房”现象出现。结果必然会导致银行不良贷款 大幅上升，民间借贷信用破产。2011 年温州房地产 泡沫破裂后，温州银行业长达数年陷入困境，期间 不良率的高点是 2014 年的 4.68%。信贷资产变差， 致使银行不愿放贷，企业融资更加困难。经济因此 受到了致命打击。以 2013 年为例，公开数据显示， 在 16 个主要经济指标中，温州大多数处于浙江省倒 数行列，人均 gdp、gdp 增幅、财政总收入等 9 项 指标倒数第一，另有 4 项倒数第二。 房地产发展的根本还在于“人、财、物”的支 撑。人是人口的需求，财是资本的供给，物是土地 的供给。相对而言，“财”才是最关键的，只有财 来了，才能吸引人的需求，吸引物的投资和供应。 温州的经验表明：任何一个地方实体经济是基础， 实体经济的疲软使“财”不足，“人”迁移，温州 经济不旺，导致温州的常住人口近年来的常住人口 均值保持在 910 万人左右，没有多少新增人口。 房价是附属在实体经济之上的一个产物，脱离 了实体经济支撑的房价必然走向泡沫。从长远的角 度来看，决定性的资本供给，还是由产业决定。对 于那些没有经济基本面和人口流入支撑的城市而 言，房价暴涨只不过是一场击鼓传花的游戏摆了。 因此，只有发展好地方的实体经济，引导实体经济 实现转型升级，从而吸引更多的人口与资金回归， 才是防止当地楼市泡沫的根本途径。 5.3. 严格控制银行信贷资金及各类“热钱”进入 房地产业 温州的经验表明，一旦房地产市场急剧调整， 极易引爆民间借贷危机。具体表现在：当房价下跌 的时候，原先“购房-抵押贷款-再购房”的杠杆操作 模式不灵了，很多人开始变卖房产还贷，那些企业 主的资产也开始恶化，还不上的干脆跑路。而企业 主加快套现房产还债的行为，又加快房价回落，从 而形成螺旋式传导链条，使得风险继续扩散发酵， 导致银行不良贷款急剧增加。与此同时，面对不断 上升的风险，银行只能收紧信贷甚至停贷，从而进 一步加剧资金紧张。在市场大环境发生质变的情况 下，若不及时主动地对房价进行调整，那么后果就 只能是被动调整，而被动调整往往会相当惨烈。 5.4. 房价回归合理水平其实是有利于房地产业长 远发展 房子不是一般的消费品，而是事关百姓民生、 长远来看也关乎中国经济总体格局的特殊商品。住 房市场的核心问题是社会问题，不是经济问题，甚 至是一个要建设什么样的社会的问题。发展住房市 场的根本目的是解决广大居民的居住问题，而不应 该推行把住房作为投资品为主的发展模式。当下中 央提出“房子是用来住的 不是用来炒的”的定位顺 应了住房领域需求的变化，是完全正确的。 房价下跌会导致房地产投资性需求大幅减少， 未来房地产的稳定与增长，只能依靠消费性需求。 在“衣食住行”中，居民对住房的消费意愿最为强 烈、潜力最大，而实现难度也是最大的。从理论上 讲，凡是从消费角度出发的需求，都对消费成本特 别敏感，特别是首套住房购买者或咬牙买房者，温 州情况正是如此。房价下跌对住房消费需求的刺激 作用非常大，住房消费需求的价格弹性相当高。房 价下跌后的 2012 年和 2013 年，商品房销售面积均大 幅增长，增速分别为 50.1%和 71.2%。 参考文献 [1] 本报评论员.有效防范金融风险——二论做好当前金融 工作[n].人民日报 2017-07-17. [2] 余根钱. 温州房价下跌的启示 https://www.qianzhan.com/analyst/detail/329/1407046e7dd407.html. [3] 温州市房产登记中心. 2012 年温州市区房产交易分析白 皮书,2013.04. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 187 [4] 周知客. 脱实向虚:温州地产泡沫启示录. http://www.p5w.net/weyt/201611/ t20161118_1640872.htm. [5] 徐明 胡丹.温州楼市正回暖 曾经房价连续 40 多个月同 比下跌. http://biz.zjol.com.cn/system/2015/06/29/020715476.shtm l. [6] 海通证券. 温州房地产泡沫对当前楼市的经验与教训. http://www.guandian.cn/article/20161018/ 179283.html1. [7] campbell j y, kyle a s.smart money, noise trading and stock price behavior[j].berd technical working paper, 1988 (236) :124-131. [8] allen f, gorton gb.rational finite bubbles[j].nber working papers, 1991, 9 (23) :121-132. [9] 贾生华 李航.噪声交易者预期与房地产泡沫——基于 35 个 大 中 城 市 的 实 证 研 究 [j]. 审 计 与 经 济 研 究 2014,03(29),85-92. [10] 洪开荣,刘亦.近期我国房地产泡沫研究的文献综述[j]. 湖南财经高等专科学校学报,2006(103).31-33. [11] 王浩 穆良平.当前我国房地产泡沫两种主要测度方法: 研究思路、误区及适用性分析 [j]. 宏观经济研究 2015,02,78-85. [12] 李平 张玉 杨立娜 江强.中国房地产泡沫对宏观经济影 响的演变路径[j]. 经济问题探索 2015,04,37-42. [13] 王永钦,包特.异质交易者、房地产泡沫与房地产政策 [j].世界经济,2011(11):84-102. [14] 李杰 沈良延 赵玉丹.恶性房地产泡沫产生原因的经验 分析[j]. 中央财经大学学报 2016,09,101-111. [15] 田利辉 马君壮.我国房地产泡沫成因分析[j]. 广义虚拟 经济研究 2015,01(6),48-56. [16] 李杰 沈良延 赵玉丹.恶性房地产泡沫产生原因的经验 分析[j]. 中央财经大学学报 2016,09,101-111. [17] 张炜.预期、货币政策与房地产泡沫——来自省际房地 产市场的经验验证[j]. 中央财经大学学报 2017,08,7790. [18] 李杰 沈良延 赵玉丹.恶性房地产泡沫产生原因的经验 分析[j]. 中央财经大学学报 2016,09,101-111. [19] 笑银.温州房地产市场多坎坷 城市框架未拉开是主要 原因[n]. 温州日报 2015-4-17. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 179–188 ___________________________________________________________________________________________________________ 188 3wenzhou economic broadcasting media group, wenzhou 325000, china 摘要: 众所周知，房地产市场的过度泡沫化会严重影响国民经济的正常发展，急剧膨胀的房地产泡沫为金融危机的爆发埋下了众多隐患。2011年，温州爆发民间借贷危机，刺破了温州房地产价格泡沫，造成区域经济受严重波及的惨痛教训。本文通过回顾的2006年到2015年温州房价的变迁走势，分析了温州房地产泡沫产生及破灭的原因。温州房地产市场的调整过程具有典型意义，可为当前楼市走向提供借鉴和参考。 关键词: 房地产泡沫；温州；房价；风险 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <feff005500740069006c0069006300650020006500730074006100200063006f006e0066006900670075007200610063006900f3006e0020007000610072006100200063007200650061007200200064006f00630075006d0065006e0074006f00730020005000440046002000640065002000410064006f0062006500200061006400650063007500610064006f00730020007000610072006100200069006d0070007200650073006900f3006e0020007000720065002d0065006400690074006f007200690061006c00200064006500200061006c00740061002000630061006c0069006400610064002e002000530065002000700075006500640065006e00200061006200720069007200200064006f00630075006d0065006e0074006f00730020005000440046002000630072006500610064006f007300200063006f006e0020004100630072006f006200610074002c002000410064006f00620065002000520065006100640065007200200035002e003000200079002000760065007200730069006f006e0065007300200070006f00730074006500720069006f007200650073002e> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <feff0054006900650074006f0020006e006100730074006100760065006e0069006100200070006f0075017e0069007400650020006e00610020007600790074007600e100720061006e0069006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b0074006f007200e90020007300610020006e0061006a006c0065007001610069006500200068006f0064006900610020006e00610020006b00760061006c00690074006e00fa00200074006c0061010d00200061002000700072006500700072006500730073002e00200056007900740076006f00720065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f00740076006f00720069016500200076002000700072006f006700720061006d006f006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076016100ed00630068002e> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style editor’s introduction for volume 5, issue 1 this issue contains six papers which can be divided into five topics: accident calamity, financial risk, risk analysis, social security, and natural hazard assessment. the last contribution is written in chinese with english abstract. in paper “bp's reputation repair strategies during the gulf oil spill”, swain and jordan believe that analysis of 1,161 bp tweets during the crisis response reflected unexpected reputation repair strategies and responsibility attribution. situational crisis communication theory suggests that after an accident, pr messages typically reflect low responsibility attribution. although the official investigation initially did not suggest a preventable crisis, 90% of bp’s tweets reflected high responsibility. there are two papers on financial risk. the first paper “banking balance sheet channel of systemic risk” by gan explores the role of the banking balance sheet as the source and transmitter of systemic risk. this paper discusses the key balance sheet channels of systemic risk. the second paper "non-diversifiable risk in investment portfolios an aid to investment decision making", by anyika, puts forward modeling non diversifiable risk in investment portfolios together with redefinition of estimators of diversifiable risk and portfolio expected returns to reflect normal market conditions. he uses garch models to make forecasts of given time series, from which future predictions of nondiversifiable risk, diversifiable risk and portfolio expected returns are made. in paper “quantum-behaved particle swarm optimization with nelder-mead simplex search method”, yao proposes a novel hybrid algorithm for risk analysis, which based on quantum-behaved particle swarm optimization algorithm and nelder-mead simplex search method for continuous optimization problems. it seems that the proposed algorithm is more effective and efficient at locating optimal solutions for continues optimization. the paper “research on society risk evolution mechanism and countermeasures in severe emergency infectious disease— in the case of h7n9 avian influenza”, by xu, cai and wang, analyzes and study the evolution and control of society risks caused by severe emergency infectious disease. firstly, the evolution chain of society risks caused by severe emergency infectious disease is constructed to analyze the evolution rule of society risks and identify the essential factors in countermeasures. then, the system dynamics model is established and employed to simulate the effects of society risk control with various countermeasures. finally, based on the simulation results, a conclusion is drawn: improving medical treatment capacity, strengthening quarantine level under epidemic situation and enhancing the effectiveness of dealing with public opinions are effective in controlling society risks. the paper in natural hazard assessment is written in chinese, with title “research of city rainstorm waterlogging scene simulation -in daoli district of harbin city as an example”, submitted by chen, zhang et al, where authors use a onedimensional and two-dimensional unsteady flow as the basic equation, with irregular grid as the basic frame, builds urban rainstorm waterlogging deposition numerical simulation model, combining with the information diffusion theory calculation rainstorm probabilities in the study area. on this basis, this paper sets the study area rainstorm waterlogging situation, realizes the different waterlogging scene simulation and visualization. many thanks to the referees for their strong support and kind help. and also very much thank to the authors for all their submissions. editor-in-chief: prof. chongfu huang email: hchongfu@gmail.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 1 published by atlantis press copyright: the authors 1 http://star.co.ke/node/41796 http://star.co.ke/node/41796 mailto:hchongfu@gmail.com mailto:rim_007@163.com mailto:jracr.srachina@126.com received 11 august 2015 accepted 15 november 2015 sweden’s capacity to prepare and respond to a terrorist attack on rail-bound traffic – promising practices and obstacles to inter-organizational collaboration veronica strandh department of political science andcentre for research and development in disaster medicine, department of surgery, umeå university, se-90187, umeå, sweden e-mail: veronica.strandh@umu.se abstract this article takes an interest in evolving collaborative practices in crises caused by terrorism targeting rail-bound traffic. sweden provides the empirical focus, by examining current preparedness processes, this article offers an important perspective on inter-organizational collaboration; that is, the perspective of the involved actors themselves. this study relies on 20 interviews with key actors in the response system and rail bound traffic actors.this article combines insights from the literature on governance with the research on disasters and crises. promising practices and obstacles to inter-organizational collaboration are identified and analyzed. keywords: crisis response, terrorism, rail-bound traffic, networks, crisis management actors 1. introduction terrorism continues to plague societies, and major mass-casualty terrorist events in the past have demonstrated how complicated and organizationally demanding the crisis response to such events is. crisis management, including both preemptive and preparatory efforts as well as response and recovery actions, requires the knowledge, resources, and interactions between many different types of organizations (dynes 1970; mileti 1999). the importance of collaborative actions in crisis management has been stressed by both practitioners and researchers. in fact, there is a vast body of literature suggesting inter-organizational collaboration as a success factor in effective crisis management (boin and ´t hart 2010; moynihan 2009). however, in practice, it turns out that collaborative actions are often difficult and imbued with challenges (boin and bynander 2014). therefore, how to achieve inter-organizational collaboration in crisis management continues to be a topic for scholarly attention. not least, it is an ongoing discussion within the disaster and crisis management field, in which scholars seek to gain a better understanding of the ideas, resources, and interactions taking place in collaborative response networks (comfort, boin, and demchak 2010; crunin et al. 2015; hu, knox, and kapucu 2014; kapucu 2008; kapucu 2012). this article sets out to contribute to that discussion by studying collaborative efforts in the context of terrorist-induced crises. more specifically, this article takes an interest in evolving collaborative practices in crises caused by terrorism targeting rail-bound traffic. thus, in terrorists’ attempts to maximize the number of casualties from their terror events, rail-bound traffic has become an attractive target (jenkins 2001, 2012; strandberg 2013; waugh 2004). sweden provides the empirical focus, and by examining the ongoing processes to enhance the journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 215-225 published by atlantis press copyright: the authors 215 capacity in sweden to prepare for and respond to a major terrorist attack on rail-bound traffic, this article offers an important perspective on inter-organizational collaboration; that is, the perspective of the involved actors themselves. this study relies on 20 interviews with key actors in the response system and major rail operators and railway authorities in sweden. the following questions are posed: how do the different actors perceive the threat of terrorism, and how do they experience their respective roles in enhancing the preparedness capacity for responding to a terrorist attack on rail-bound traffic? in terms of interorganizational collaboration, what do the different actors perceive as promising practices and obstacles to their current crisis management capacity? the outline is as follows: the next section begins with a short background, followed by a section presenting the perspective applied to analyze the research questions. in order to understand the complexity associated with inter-organizational collaboration, this article combines literature on governance with the crisis and disaster literature that focuses on collaborative response networks. the notes on methodology then follow. a fictional scenario of a multi-site terrorist attack targeting rail-bound traffic in sweden’s capital, stockholm, was developed and used during the 20 interviews. the findings from the interviews are structured to reflect practices in different phases of crisis management, with a special focus on preparedness and response practices. the concluding section discusses the main findings and answers the research questions posed. 2. background sweden has experienced major crises, including the estonia ferry disaster in 1994 and the assassinations of the prime minister in 1986 and the minister of foreign affairs in 2003, as well as suffered greatly from the indian ocean tsunami in 2004. in the 1970s, the country suffered from acts of terrorism, including the bulltofta skyjacking event in 1972 and the seizure of the west german embassy in 1975 (hansén 2007). however, sweden has no experience of managing terroristinduced crises that have resulted in major mass-casualty situations. thus, it is reasonable to believe that actors in sweden suffer from a so-called paucity of experience problem (levitt and march 1988). it is assumed that a management and knowledge gap exists in terms of the practical experience of dealing with a mass-casualty event caused by terrorism. nevertheless, sweden now faces a situation when the palette of risks, hazards, and vulnerabilities is changing. terrorism is increasingly often identified as a concern in many public organizations’ risk and vulnerability analyses, and we can therefore assume that preparedness practices are also taking place for terrorist-induced crises. a terrorist attack on rail-bound traffic in sweden will require response actions from fire fighters, emergency medical personnel, police, intelligence agencies, and crisis actors at different levels of government. in addition, transportation actors are assumed to participate. in the case of sweden, it happens that rail-bound traffic is a deeply fragmented environment; no less than 17 companies are running passenger traffic and 21 companies are running freight traffic (registered members at astoc 2015). hence, an attack on rail-bound traffic would affect and require actions from a multitude of different actors. the many different actors are is expected to operate in complex networks under the guiding norm of interorganizational collaboration. characteristic for the swedish response system is its very strong emphasis on shared and collaborative actions in crisis management (wimelius and engberg 2014). the importance of collaboration isfirmly embedded in policy documents central for crisis management such as government bill cooperation in emergency – for a more secure society (2005/06:133) and government communiqué society’s crisis preparedness: strong collaboration for increased safety (2009/10:124). moreover, one can observe how the centrality of inter-organizational collaboration is reflected in the actual daily language used by practitioners in the field; there are collaboration courses, collaboration conferences, collaboration exercises, forums for cooperation and collaboration, etc. danielsson, johansson, and kvarnlöf (2013:17) make a telling observation about inter-organizational collaboration in sweden: “the concept of interorganizational coordination has sunk in to the point where it is imbued with ideological rather than practical meaning; coordination has become a goal in and of itself, as opposed to being a means to an end.” it illustrates how the particular concept of interorganizational collaboration has reached a rather exceptional position in the swedish crisis management context. hence, empirically speaking, sweden constitutes an interesting case to further explore, the support for collaborative actions appears to be resounding. however, in the particular context of collaborative preparedness for a terrorist-induced crisis targeting rail-bound traffic in sweden, there is no previous systematic research. in addition, and analytically speaking, sweden is also an informative case when aiming to gain a better understanding of how organizations operate in collaborative, networked environments. hence, during the last decade, public services in sweden have become increasingly characterized by collaborations and networked forms of organization. published by atlantis press copyright: the authors 216 3. collaborative actions and network theory in crisis management in the endeavor to understand collaborative actions in crisis management, this article combines insights from the literature on governance with the research on disasters and crises. in line with researchers such as drabek (2007) and moynihan (2009), interorganizational collaboration is used as the key concept. the terminology appears problematic because it varies extensively. inter-organizational coordination, collaboration and cooperation, multi-agency cooperation, and cross-sectorial cooperation are often used interchangeably in both crisis research and governance literature. however, the author interprets the core of the concept inter-organizational collaboration as referring to the fact that organizations need knowledge and resources from other organizations; therefore, they need to interact in networks to achieve a common goal. a great amount of researchers have drawn attention to how overarching changes are taking place in crisis management. the american response system is frequently studied, not least since the september 11 attacks. hu et al. (2015) drew attention to how the american emergency management system has gradually shifted from a centralized command and control system to a more collaborative approach. such a development can also be observed in other countries. it is common to contrast a traditional hierarchal response system to a horizontal, networked system. an overwhelming bulk of the crisis and disaster research stresses how hierarchical networks perform poorly under crisis because they are assumed to isolate large networks from each other. instead, a more horizontal form of collaboration is increasingly put forward as the preferred system (comfort 2007; dynes and quarantelli 1977; kapucu 2006; moynihan 2008). a fundamental assumption is that society’s many “complex threats” require the involvement of a multitude of actors, including their combined knowledge and resources (handmer and dovers 2013). the observed shift in how crisis management is organized also relates to a broader discussion on political steering and public administration, which according scholars have using a common formulation, “experienced a shift from government to governance” (rhodes 1996; peters and pierre 1998). this shift is described as a movement away from principles such as bureaucracy and command and control in favor of collaboration and flexibility. klijn and koppenjan (2000:130) described the characteristic of the network approach as: “policy is made in complex interaction processes between a large numbers of actors which takes place within networks of interdependent actors.” this shift from government to governance not only implies a change in structure but also opens up for a greater influx of different ideas and views among the many actors. one can here refer to “frame reflection” (schön and rein 1994), which makes us aware of how different actors make strategies based on their own frame. therefore, we must assume that there might be several frames present in collaborative networks, thus indicating different views on solutions as well as problems in networks. as stated, a more horizontal form of collaboration in crisis management has support in research. the benefits from collaborative networks can vary depending on context; however, in the context of complex crises, some aspects are frequently highlighted in the literature. networks are seen as flexible and are thereby regarded as effective in crisis situations. according to comfort (2007), a key characteristic of collaborative actions is that the participating actors align their activities voluntarily. therefore, the importance of personal relations and trust is often mentioned in this regard. building confidence among actors is seen as a key to reduce the costs of collaboration (coleman 1990; edelenbos and klijn 2007). networked forms of collaboration are also assumed to enable rapid dissemination of information among their members, which in turn is assumed to enable collective sensemaking (weick 1988). comfort (2007) argued how cognition, understood as the capacity to recognize the degree of emerging risks and act on such information, is central to crisis management practices. moreover, networked systems are assumed to foster creativity, an aspect which is frequently regarded as important in complex crisis management processes (kapucu 2008). in light of examining inter-organizational collaboration in the context of rapid evolving crises, which are associated with stress, uncertainty, and time pressure (boin et al. 2005), the actual management of response networks becomes crucial. klijn and koppenjan (2012) underlined how networks require guidance and management. when the management of networks is discussed in the literature, the state is often referred to as being an enabler of collaboration and coordination, and its steering capacities should seek to unite the various perceptions of actors, thus including negotiating skills (klijn and koppenjan 2000). the bottom line of the governance literature, and the crisis research that focuses on networked forms of crisis management, can be interpreted somewhat simply as complex problems should be met with complex network solutions; therefore, a question arises: is it so easy? one should be aware that the network approach has encountered criticism; for instance, it is criticized for placing too much emphasis on collaboration and thereby downplaying the risk of conflict and power differences. moreover, critics argue that the approach considers published by atlantis press copyright: the authors 217 government organizations to be seen as any other organization, neglecting their role as “guardian[s]” of the public (brans 1997; klijn and koppenjan 2012). one can also refer to moynihan (2008), who emphasized the complexity of the so-called “crisis management paradox,” which means that crises require an inter-organizational response, while at the same time requiring rapid and decisive coordination, which is seen as an inherent challenge to networks. thus, there are diverging views of collaborative networks. in our endeavor to understand collaborative actions in the context of terrorist-induced crises directed against rail-bound traffic in sweden, questions were raised during the interviews that, in different ways, captured the dynamics of in-organizational collaboration with regard to both the presumed merits and limits. 4. methodology empirically, this article builds on 20 semi-structured interviews with actors who have a central role in preparing and responding to terrorism targeting railbound traffic. the main data collection was carried out between february and november 2014, and the interviews were conducted in person. a broad range of organizations were included: 1) first responders at an operational and at a strategic level (the police, the security service, the fire service, and the medical services); 2) actors from rail-bound traffic (key actors in the scenario include sj ab, a-train ab, mtr stockholm, jernhusen, the swedish transport agency, and the swedish transport administration); and 3) representatives from central actors in the response system such as the city of stockholm, the county administrative board in stockholm, and the civil contingencies agency (msb). a scenario-based methodology has guided the research. a terrorist attack scenario was developed in which stockholm was struck by four bomb explosions. the merit of using scenariobased interviews is that practitioners in the field of crisis management are used to working with scenarios in various methods as a way to plan for potential future crises (stern and sundelius 2002). a scenario was also useful since it indicated the scale of the attack and thereby reduced the risk that the term preparedness or response capacity had very different meanings for the interviewee and researcher. moreover, a semi-structured interview guide, capturing different aspects of interorganizational collaboration, was used. the data were later transcribed by the author and analyzed on the basis of the pre-defined themes: threat perception, preparedness activities, immediate response capacity, inter-organizational collaboration, recovery, and learning processes. 5. analysis 5.1. crisis management in sweden the swedish crisis management system is often described according to its three guiding principles. the principle of responsibility establishes that the authority who has responsibility over an activity under normal circumstances will remain the responsible authority during a crisis. the principle of parity states that an activity during a crisis should, as far as possible, be carried out in the same manner as it would under normal situations. the principle of proximity implies that a crisis should be managed at the lowest possible level (krisinformation.se 2012). municipalities have geographic responsibilities at the local level. the 21 county administrative boards have a geographic responsibility at the regional level. they have an important role in coordinating society’s preparedness through, for example, exercises and risk and vulnerability analyses. at the time of a major crisis, the county administrative board has an important function to coordinate different measures with relevant actors (government bill 2007/98:92). moreover, at the regional level, county councils play a central role when it comes to providing health care and ensuring that there is disaster medicine preparedness in place (the national board of health and welfare 2013). at the national level, the political responsibility for civil crisis preparedness belongs to the ministry of justice, which acts through the swedish civil contingencies agency. there is also a crisis management coordination secretariat at the government offices in order to handle cross-sector crises in a coordinated way (ministry of justice 2014). 5.2 preparedness practices a starting point for examining ongoing processes to enhance the capacity in sweden to prepare for and respond to a major terrorist attack on rail-bound traffic is with the joint efforts carried out early in the process. given the many different involved actors, one can assume that the perceptions of the nature and extent of the actual hazard, in this case terrorism, might vary. interviewees were asked to describe how they perceived the threat of terrorism against sweden and how they viewed the risk of an attack targeting rail-bound traffic. the three quotations below illustrate the most recurring reflections. “absolutely, it is not unlikely that it might happen. it is likely that it is going to happen; the only question is when?” (interview 5). published by atlantis press copyright: the authors 218 “it is going to happen. so, it is good to be as prepared as can be” (interview 12). “i don’t know how today’s terrorists think, but historically—madrid, london—there’s a tendency for them to appreciate rail-bound traffic as a target. why wouldn’t they be thinking the same thing in sweden, if the threat scenario is there?” (interview 15). sweden is seen as a potential target for terrorism, and based on the logic that terrorists appear to target soft targets in order to achieve extensive damage, the interviewees found rail-bound traffic to be a likely target for attacks. hence, the empirical material reveals a uniformed picture when it comes to the risk of an attack and its potential consequences. however, on the question of how each organization relates to the threat level and how knowledge and intelligence information are communicated in the current preparedness practices, a more ambiguous picture emerges. the swedish security service (säpo) is perceived as the most central actor in the context of counterterrorism. the agency is also seen as a node or a source of information when other actors prepare for acts of antagonistic violence. however, the empirical material shows different views in how collaboration actually works between intelligence actors and other actors in the crisis response system. one line of reasoning that emerged during the talks was that the abilities to acknowledge risks and threats as well as take preventive actions belong to the domain of säpo. respondents expressed trust for their work, and they commented that they can only act upon their assessments. another line of reasoning reveals how respondents perceive law enforcement, particularly säpo, as working in a closed or protective manner. hence, they experience a lack of information. the two different views are exemplified below: “i trust that säpo is doing their job and that they are conducting the right threat assessments; i can only act upon their information” (interview 20). “sometimes it feels like a one-way communication. i have to call säpo and ask, “is this affecting us?”.... even though they make all the decisions when it comes to threat levels and so forth, we at least need to be informed” (interview 12). despite an awareness of the complexity of intelligence, several organizations stress the need for improved flows of information. however, at the same time and in a selfcritical manner, there seem to be uncertainties about the different organizations’ capabilities to receive increased information and then adapt their behavior accordingly. it is a precarious task to establish a link between a certain-threat scenario as well as prepare to manage the situation if the threat is realized. training and exercises are common tools for enhancing preparedness. since there are many players in a response to a large-scale crisis such as a terrorist attack, there are also different exercise routines. a training-intensive organization such as the fire brigade, which trains on a weekly (or daily) basis for a crisis situation, differs widely from a rail operator who conducts, in general, one to two trainings a year. this difference is expected due to their respective roles and responsibilities. still, the essence of a more collaborative approach to crises implies that both public and private organizations are assumed to interact during crises; therefore, increased training opportunities, spanning over organizational boundaries that involve different cultures of working, are therefore regarded as important. from an inter-organizational perspective, it is an interesting finding that the many different actors appear to prepare for terrorist-induced crises in profoundly different ways, and some of them not at all. as stated, it is a shared perception that rail-bound traffic is at risk for terrorism, and most of the respondents find it likely that sweden will suffer from an attack in the future. at the same time, however, one of the most obvious findings in the material is the disparate answers on whether there is a need to carry out training and preparedness exercises with a specific focus on terrorism. “we do not practice how to respond to terrorism. i’m not sure, but i do not think that a crisis caused by terrorism differs from a crisis caused by something else” (interview 10). “sometimes we practice a scenario including a bomb or so, but we never really practice all the extra dimensions one can assume come with a terrorist attack…you can assume for example, a second explosion, and when something like that occurs, i think we are not as prepared as one could wish” (interview 3). the two opinions expressed above come from two different people, each with a central operational function in the case of a major crisis, yet they present two diverging views. hence, there seems to be a tension between those who practice or believe that it is necessary to practice terrorism-specific scenarios since they encompass particular security challenges and bring certain challenges, and there are others who believe that it is better to concentrate the available resources on practicing general and more recurring response challenges. published by atlantis press copyright: the authors 219 a point of reference for collaborative preparedness practices is 2007, when msb, which organizes the yearly multi-sector exercise “samö,” used a scenario simulating multiple attacks on rail-bound traffic in stockholm. the evaluation report revealed a rather extensive critique of the response; interestingly, several aspects referred to inter-organizational collaboration. a lack of routines for establishing a shared perception of the threat level and shortcomings in coordinating communication to the media and the public were identified as critical aspects (msb 2007). this particular exercise was brought to our attention in several of the 20 interviews, and it appears to be a shared opinion that the preconditions for carrying out effective inter-organizational collaboration during the exercise were poor. importantly, many respondents expressed that they have experienced progress since 2007; not least in terms of an increased commitment to work together. yet, there has been no follow-up exercise on this particular scenario of a terrorist-induced crisis on rail-bound traffic since. “i think that it is still rather controversial to talk about terrorism against rail-bound traffic… if an imminent threat against the railway system would emerge, how are we supposed to act? who will work with this? who speaks the language of law enforcement? do we have security clearance?...it is one thing to collaborate during normal circumstances, but it is a completely different thing to collaborate when the actual crisis occurs.” (interview 4) hence, there are respondents who indicated that despite the obvious challenges identified in 2007, there is still much that remains to be done when it comes to preparedness for this particular terrorist-induced crisis. 5.3 response practices as stated, in order to discuss collaborative response practices during the most acute phase of a crisis, a fictional scenario was developed. this scenario illustrates how central stockholm has become the target for an unknown terrorist group.1 hidden in backpacks, the first bomb explodes at an sj ab regional passenger train while still standing at the central station. two subsequent explosions take place at a subway train between the two central stations: hötorget and rådamansgatan; and an additional bomb explodes at an arlanda express train heading for the international 1 the scenario builds on an empirical overview of around 4,000 attacks carried out against rail-bound traffic between 1970–2010 (strandberg 2013). based on the 20 most large-scale attacks, a scenario has been developed. airport. in all, the four bombs explode during a friday morning rush hour in the short time span of 08.12– 08.16. it causes the deaths of 80 people, and 300 people are in need of immediate medical care. in addition, there is a large number of so-called walking wounded. the scenario also entails a full closure of the transportation system. each interviewee was asked to describe his or her immediate actions following the four bomb explosions, as depicted in the scenario. the reported tasks cover a wide range of activities. core emergency management actions include saving lives, fighting fires, securing buildings, evacuating the train and subway system, conducting law enforcement procedures, collecting evidence, and responding to new threats. several actors described their immediate tasks as being about coordination, such as coordinating distribution of information to the media, establishing a common progress report, sharing information, holding coordination meetings, and planning for future actions. other actors interpreted their immediate actions to be about taking care of their own companies, such as activating catastrophe plans. additional examples concerned activities such as electricity restoration, volunteer coordination, provision of food supplies, and supporting commuters. this description serves the purpose of underlining that one of the most obvious characteristics of the responding network in the case of a major attack on rail-bound traffic is its sheer size and the many tasks that are to be performed in the first critical hours. without a doubt, in terms of management skills and resources, the different actors anticipate that their collaborative response will face some critical challenges. difficulties adhering to the initial sensemaking process appear as a main challenge. in particular, it was underlined by several respondents that it is assumed to be more difficult to make sense of the situation when it is a terrorist-induced crisis compared to a more recurring or familiar crisis situation: “you need a common ground for how you approach incident sites after a terrorist attack….in today’s situation, it would be like ants moving in on an anthill, from all directions, and each ant going to work on its own little bit” (interview 3). first responders express concern that a slow sensemaking process can have serious consequences in terms of security. the quotation above demonstrates how a respondent, who isfamiliar with working at incident scenes, anticipates how front liners will rush into the situation to save lives without carefully considering potential security risks. further explosions and “dirty bombs” are other potential security dangers. most first published by atlantis press copyright: the authors 220 responders seem to agree that their usual quick response routines can become a vulnerability in these particular contexts since, most likely, they will not be reached by the information about the antagonistic nature of the attack before they have entered the site. a second point refers to the increased confusion that three different incident sites apply to the sense-making process. a third critical point refers to the dilemma of whether the police need to secure the site before ambulance and firefighters can gain access. if so, how long can such a decision take, considering the first “critical hour” and the need for people to receive immediate medical treatment? it is highly important to remember that the immediate response to a multi-site attack takes place in a very time-critical environment. as the fictional scenario entails, the rail-bound traffic system becomes the target of bomb explosions. persons injured in bomb explosions (and in shootings) tend to bleed to death, and meaningful life-saving efforts must be initiated extremely rapidly. hence, much of the most imitated collaborative actions, taking place at the different incident sites, revolve around saving lives. a common formulation among the first responders interviewed for this study was “we will manage the situation, one way or another.” however, the scale of the attack seems to add an extra dimension to the response and to collaboration in particular. the role of disaster medicine emerges here as central; that is, the capacity to handle demanding situations in which the available resources are insufficient in relation to the immediate need of medical care (lennqvist 2005:300). a concern was expressed that the current resources invested in specific disaster medicine competence fall short. one respondent explained how it is reasonable to believe that, in the case of a major terrorist attack of the same scale as the scenario, the hospitals in stockholm will be able to deploy a significant amount of medical personnel. however, the respondent questions to what extent the personnel hold a specific competence in disaster medicine. “there is a belief that the local level—that is, hospitals and our many local health facilities—should also have a competence when it comes to disaster medicine. but in the world we are working in today, with so many demands on efficiency and demands on making profits, it’s not going to happen. they many different actors will not invest in specific trainings or disaster medicine competence unless they are paid for, or unless it is a requirement.” (interview 15) it is important to note that the scenario is taking place in the capital; thus, the preconditions for responding to an attack in more sparsely populated places in sweden are different. it is emphasized in the interviews that the stockholm regionhas plenty of resources. for instance, the two major fire brigade organizations have around 30 fire stations at their disposal, and there are seven emergency hospitals in stockholm. yet, as the quotation below illustrates, one should not underestimate the enormous resources that a terrorist attack on rail-bound traffic would require: “although we might have plenty of resources in stockholm, we cannot handle everything….three different incident sites is an extremely demanding task to manage, and it would most likely require three separate command and control functions” (interview 15). at first, resources will be scarce and will not be available at the moment when the first bomb explodes. the medical “surge capacity” in stockholm was described as consisting of seven emergency hospitals, 68 ambulances, one helicopter, and more than 200 health centers. importantly, many of these resources are already in use on a daily basis, and access to intensive care beds is particularly challenging because they are already being used by critically ill patients who cannot be transferred to other places without difficulty. thus, rapid action is necessary to make resources available and request assistance from other regions. once again, referring to the scenario and its bomb explosions, we must expect that the injury pattern following the attack will include severe burn injuries. there are no special units for burn injuries in stockholm; hence, rapid collaboration becomes crucial since the severely injured must be transported to the cities of uppsala or linköping. so far, it has been illustrated how the immediate response to four bomb explosions tends to revolve around a specific “subnetwork” consisting of the ambulance, fire, and police services. respondents report trust and confidence in the well-practiced interactions taking place among the front liners. above all, first responders often have similar command structures and are used to working together, and they have a confidence in their respective roles. if we expand the horizon, collaborative networks in crisis management are not confined to specialized first responders and another subnetwork emerges. that is a network consisting of transportation actors; they are not specialized in crisis management, but they are nevertheless assumed to participate in interorganizational collaboration. in the fictional scenario used in this article, the railway operators, infrastructure owners, and transportation authorities, such as sj ab, arlanda express, mtr stockholm, jernhusen, and the swedish transport administration, are immediately affected and put under pressure. in contrast to first published by atlantis press copyright: the authors 221 responders, rail-bound traffic actors’ actions in the immediate response take place in more peripheral parts of the response networks; for example, assisting with buses at evacuations, providing information to passengers, or dealing with infrastructure damage. furthermore, respondents emphasized how a response to a terrorist attack in central parts of the capital will cause an initial chaos. thus, in addition to established response procedures, many respondents stressed the need to “think outside the box.” one cannot fully predict how a crisis will develop and which ad-hoc networks and ad-hoc solutions will take place. respondents are open to the development that ad-hoc solutions can serve as a complement to (or even replace according to some) pre-planned procedures, such as turning sport venues into meeting points for relatives or stranded commuters. although respondents expressed diverging views as to what extent and when volunteers can be utilized in the response, volunteers were mentioned as an additional subnetwork of potential resources. inter-organizational collaboration cannot only be understood in terms of the influx of resources and knowledge. in fact, a conclusion from the 22 july commission in norway (2012:19:1) is illustrative in this regard by concluding “resources count, but their use decides.” hence, all these potential resources put forth by the respondents also require management and coordination, which in turn requires clear roles and responsibilities among the many actors. in light of that, some reflections from the respondents deserve our attention: “we have to concretize our efforts to enhance interorganizational collaboration. we cannot only emphasize it when we are participating in a meeting or workshop and then go home and do nothing” (interview 4). “from the perspective of organizational-collaboration, i think we are pretty well prepared. we should probably be able to take prompt collaborative actions. however, if your scenario, including hundreds of casualties, would actually occur, then i am not completely sure if we could ‘manage the situation’ or not” (interview 15). it appears as if the different organizations relevant in a response network, in the case of a major terrorist attack on rail-bound traffic, increasingly pay attention to interorganizational collaborations and actions to enhance knowledge about different subnetworks and how different ways of working and responsibilities take place. however, the empirical material also echoes a concern that increased knowledge about one another will not always turn into actual action. therefore, several respondents referred in a positive manner to the programme for cooperation – stockholm region as a concrete example of inter-organizational coordination. the initiative that was taken after the coordination deficiencies during the “samö 2007” was revealed. the county administrative board has a central function at this regional level, and additional participants include sos alarm, municipalities in stockholm, the stockholm county council, representatives for public transport, fire, and rescue services, the ports of stockholm, and the police authority. the guiding idea outlines that efficient collaboration in everyday life will generate good preparedness and the ability to deal with a major incident in a time of crisis (programme for cooperation – stockholm region 2014). the county administrative board plays a key role in carrying out recurrent collaboration conferences during a crisis, and interviewees expressed expectations that these meetings will “move the management forward” through establishing a common progress report and making decisions about how to proceed. the respondents were asked whether there are any concrete mechanisms to resolve possible differences in opinion among players in a collaboration conference. this is how one respondent answered: “i actually don’t know. it’s about leadership, and in this case the leadership belongs to the county administrative board…you have to negotiate and be convincing that your idea is the right way forward, and if the others think differently, well, then you have to come up with an ad-hoc solution.” (interview 13) the citation above shows that potential diverging views are expected to be bridged through discussions. that is interesting, at least when one recalls “the crisis management paradox” (moynihan 2008), which is identified as the tension between the need for many actors and interactions in a crisis and for them to perform rapid and decisive coordination at the same time.” a terrorist attack on rail-bound traffic would cause vast societal consequences and thereby result in national ramifications. some interviewees stressed that the mere location of the attack, in the center of stockholm and close to the political power of the country, would trigger an enormous security apparatus, and information and coordination activities would take place, as one interviewee expressed, “from all over.” no single government agency or jurisdiction has the required resources and expertise for coordinated crisis management; thus, once again, interactions over organizational boundaries are needed. msb is frequently referred to in the interviews due to its mandate to “support the coordination of measures taken published by atlantis press copyright: the authors 222 by local, regional, and national authorities during a crisis or disaster” (government bill 2007/98:92). interviewees expressed expectations of msb playing a role in allocating additional resources, activating the swedish national air medivac if necessary, and requesting international support as well as playing a role in information sharing. however, the exact meaning of common formulations such as support to coordination and enabling coordination comes across as unclear to some respondents. thus, a common contradiction arises. on one hand, it seems to be a fundamental assumption that inter-organizational collaboration and horizontal cooperation are desirable in such situations because expertise and resources from many actors are necessary. on the other hand, concepts associated with a hierarchical steering model, such as control, command, and decisive actions, were frequently mentioned as useful during a complex large-scale crisis with rapid developments. 6. ideas, resources and interactions in interorganizational collaboration in this article, the concept of inter-organizational collaboration has been understood as referring to how organizations need knowledge and resources from other organization and therefore need to interact effectively in crisis management networks in order to achieve a common goal. in our endeavor to gain an understanding of the interplay between resources and structures and ideas about how to use them, it was suggested that we benefit from turning to the involved actors themselves. hence, drawing attention to what both crisismanagement actors and rail-bound traffic actors regard as opportunities for and challenges to collaborative actions, we can better capture and understand interorganizational collaboration in the context of terroristinduced crises targeting rail-bound traffic. in terms of opportunities, the respondents in this material have discussed how the idea of interorganizational collaboration has earned increased support over the years. the “samö 2007 exercise” was mentioned as a point of reference for progress. since then, the actual hazard, in this case terrorism, has drawn more attention, and through that process, the need for collaborative action has emerged. workshops related to terrorism prevention and various knowledge-sharing activities are seen as steps toward filling the knowledge gap that has resulted from sweden’s limited experience of dealing with actual mass-casualty attacks. moreover, different players also report that closer inter-agency relations in some cases have resulted in successful collaborations. most notable among these was the “programme for cooperation – stockholm region.” in terms of challenges; the ideas, resources, and collaborative networks encompass many twists and turns. we should note that today’s complex crises, including terrorism, are assumed to require collaborative efforts. however, at the same time, it appears that the trend towards more horizontal, collaborative networks also give rise to new challenges. replacing the often contrasted system, that is a more hierarchical response system, with a collaborative network system, is by no means an automatic solution to the demanding organizational challenges associated with complicated crises. hence, this concluding discussion will address three challenging points; (i) actors appear to work in subnetworks, (ii)a perceived difficulty of concretizing inter-organizational collaboration and (iii) network concepts contribute to our understanding of obstacles to inter-organizational collaboration, however, it contributes with less regarding promising practices the first point relates to how actors appear to work in subnetworks, or alternatively expressed, in “silos.” the swedish response system builds on the idea of collaborative actions in which many actors’ combined resources and knowledge are a precondition for successful crisis management. however, a picture has been revealed in which many actors tend to continue working primarily in already-known constellations. the different subnetworks can be described as having their own frames and behaviors. for example, one can discern major differences between how a subnetwork of first responders operate compared to a network of transportation actors; and in the context of terrorism, intelligence actors emerge as an additional, very central subnetwork. hence, even though the swedish response systems is talked about as a major network spanning different sectors and levels, different subnetworks are easily identified. an interesting finding is therefore that the respondents appear to relate to and interpret this reality very differently. according to many respondents, working in silos is perceived as the biggest obstacle to effective inter-organizational collaboration. they underline how networks are working side-by-side, but do not necessarily interact with each other on a deeper level. interestingly, not all respondents regard the feature of working in silos as a challenge or an obstacle to inter-organizational collaboration. instead, they argue that one must acknowledge that, to some extent, the many assumed actors in a response are very different, with different functions and different mandates, and it is not time-efficient for all actors to interact during a response. however, this does not mean that increased cross-sectoral preparedness is not desirable. one can conclude that there appears to be many interpretations of the idea of collaborative crisis-management networks. this indicates that today we have a situation published by atlantis press copyright: the authors 223 in which many actors hold very different understandings of what knowledge, resources and interactions are needed in order to enhance preparedness and the capacity to respond to terrorism targeting rail-bound traffic. a second point relates to a perceived difficulty of concretizing inter-organizational collaboration among actors. the respondents have revealed a frustration with the difficulty of translating the concept of interorganizational coordination into practice during an extraordinary event. how should all actors, interests, perceptions, and goals be organized in practice? interorganizational collaboration has been put to the test in connection with major preplanned events and sudden crises, but never in an extraordinary terrorism context. the findings in the material underline the difficulties of this particular extraordinary event. the following concerns have been raised: initial sense-making is seen as problematic, and an insufficient mental preparedness for the complexity of a terrorist attack can hinder a rapid sense-making process and rapid decision-making. confusion remains over procedures to secure incident sites before entering them. coordination of several incident sites is another challenge, and there are uncertainties about different limiting factors for capacity in terms of resources. it is important to remember that anticipated difficulties must be dealt with under severe time pressure with many players involved. it is therefore noteworthy that no clear mechanisms for resolving potential conflicts seem to exist. what happens if the many actors disagree or take different directions concerning, for example, progress reports, information provided to the public, and decisions on what to do next? once again, there is a resounding support for collaborative actions in principle, but as these problems illustrate, there is a delicate challenge to turn the idea of inter-organizational collaboration into prompt, effective action in times of an extraordinary crisis. theoretically, and third,results point to the inherent complexity of inter-organizational collaboration by networking. the interplay between ideas, resources and their flows within and between inter-organizational networks seems to preclude any unified, systemic understanding. whereas network concepts contribute to our understanding of inter-organizational collaboration by sharpening our analysis of potential obstacles, it contributes with less regarding promising practices. to conclude, two main obstacles to interorganizational coordination seem to exist: despite promising practices, extensive work is still carried out in different subnetworks, and players experience difficulties in translating inter-organizational coordination into practical means. a paradox arises: if work continues to be performed in isolated silos, the ability to concretize truly inter-organizational collaboration will be hindered. references 22 july commission in norway nou 2012:14, rapport fra 22. julikommisjonen, office of the prime minister, oslo. astoc association of swedish train operating companies, medlemmar. avialabe from: http://www.tagoperatorerna.se/medlemmar_3. (accessed 2015-10-20). a. boin, and p. hart, organizing for effective emergency management: lessons from research, australian journal of public administration. 69(4) (2010) 357-371. a. boin, and f.bynander, explaining success and failure in crisis coordination, geografiskaannaeler: series a, physical geography, 97(1) (2014) 123-135. m. brans, challenges to the practice and theory of public administration in europe,journal of theoretical politics. 9 (3) (1997) 389-415. j. coleman,foundations of social theory. (harvard university press, cambridge, 1990). l. comfort, a. boin, and c. demchak, c. (eds.) designing resilience –preparing for extreme events. (pittsburg: university of pittsburg press, 2010) l. comfort, crisis management in hindsight: cognition, communication, coordination, and control,public administration review.67 (1) (2007) 189-197. s. curnin, c. owen, d. paton, c. trist and d. parsons, role clarity, swift trust and multi-agency coordination, journal of contingencies and crisis management, 23(1) (2015) 29-35. e. danielsson, r. johansson and l kvarnlöf, samverkan under räddningsinsatser – en slutrapport från ett forskningsprojekt, rcr working paper series, mid sweden university, 2013. t. drabek, community processes: coordination in handbook of disaster research(springer: new york), pp 217-233. r. dynes and e. quarantelli. organizational communications and decision making in crisis, the ohio state department of sociology, disaster research center, 1977. r. dynes, organized behavior in disaster (lexington: lexington books, 1970). j. edelenbos and e-h klijn, trust in complex decisionmaking networks: a theoretical and empirical exploration, administration and society. 39 (1) (2007) 2550. d. hansén, crisis and perspectives on policy change swedish counter-terrorism policymaking. dissertation. swedish national defence college, 2010. https://www.msb.se/upload/utbildning_och_ovning/ovning/ utv%c3%a4rdering_uppf%c3%b6ljning/utvardering/s am%c3%b6%202007.pdf (accessed: 2010-10-01). b. jenkins, protecting public surface transportation against terrorism and serious crime: an executive overview. report fhwa/ca/or-2001-29. san jose: the mineta transportation institute, 2010. b. jenkins, thetransportation threat to surface transportation: the challenges of securing public places inthe mcgrawhill homeland security handbook: strategic guidance for a coordinated approach to effective security and published by atlantis press copyright: the authors 224 emergency management(mcgraw hill, new york), pp 2131. n. kapucu, collaborative emergency management: better community organizing, better public preparedness and response, disasters. 32(2) (2008) 239-62. n. kapucu, interagency communication networks during emergencies: boundary spanners in multiagency coordination,the american review of public administration. 36 (2) (2006) 207-225. n. kapucu, network governance in response to acts of terrorism: comparative analyses (new york: routledge). e. klijn and j. koppenjan, public management and policy networks: foundations of a network approach to governance, public management: an international journal of research and theory, 2(2) (2000) 135-158. krisinformation.se krishanteringens grunder, available from:http://www.krisinformation.se/web/pages/page____1 1260.aspx (accessed 2014-03-23). s.lennqvist, education and training in disaster medicine,scandinavian journal of surgery. 94(4) (2005) 300-310. b. levitt and j. march, organizational learning, annual review of sociology. 14(1) (1988) 319-338. d, mileti, disasters by design: a reassessment of natural hazards in the united states (washington dc: joseph henry press, 1999). ministry of justice, regeringskansliets krishantering, available from: http://www.regeringen.se/regeringenspolitik/krisberedskap/regeringskansliets-krishantering/ (accessed: 2014-12-10). d. moynihan, combining structural forms in the search for policy tools: incident command systems in u.s. crisis management, governance. 21(2) (2008) 205-229. d. moynihan, the network governance of crisis response: case studies of incident command systems,journal of public administration research and theory. 19(1) (2009) 895-915. msb /krisberedskapsmyndigheten, utvärdering and samverkansövning 2007, available from: national board of health and welfare, socialstyrelsens föreskrifter och allmänna råd om katastrofmedicinsk beredskapavailable from: http://www.socialstyrelsen.se/publikationer2013/2013-546 (accessed: 2014-04-20). g. peters and j. pierre, governance without government? rethinking public administration, the journal of public administration research and theory. 8(2) (1998) 223-243. program för samverkan stockholms regionen. http://www.pfssr.se/ (accessed2014-04-23) h. qian, c. knox and n. kapucu, what have we learned since september 11, 2001? a network study of the boston marathon bombings response, public administration review. 74(6) (2014) 698-712. r.rhodes, the new governance: governing without government, political studies. 44(4) (2006) 652-667. e. stern and b. sundelius, crisis management europe: an integrated regional research and training program,international studies perspectives, 3(1) (2002) 71-88. v. strandberg, rail bound traffic—a prime target for contemporaryterrorist attacks?,journal of transportation security. 6 (3) (2013) 271–286. w. waugh, securing mass transit: a challenge for homeland security, review of policy research. 21 (3) (2004) 307316. k. weick, enacted sensemaking in crisis situations, journal of management studies. 25(4) (1988) 305-317. m. wilmelius and j. engberg, crisis management through network coordination: experiences of swedish defence directors, journal of contingencies and crisis management. 23(3) (2015) 129–137. published by atlantis press copyright: the authors 225 references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <feff03a703c103b703c303b903bc03bf03c003bf03b903ae03c303c403b5002003b103c503c403ad03c2002003c403b903c2002003c103c503b803bc03af03c303b503b903c2002003b303b903b1002003bd03b1002003b403b703bc03b903bf03c503c103b303ae03c303b503c403b5002003ad03b303b303c103b103c603b1002000410064006f006200650020005000440046002003c003bf03c5002003b503af03bd03b103b9002003ba03b103c42019002003b503be03bf03c703ae03bd002003ba03b103c403ac03bb03bb03b703bb03b1002003b303b903b1002003c003c103bf002d03b503ba03c403c503c003c903c403b903ba03ad03c2002003b503c103b303b103c303af03b503c2002003c503c803b703bb03ae03c2002003c003bf03b903cc03c403b703c403b103c2002e0020002003a403b10020005000440046002003ad03b303b303c103b103c603b1002003c003bf03c5002003ad03c703b503c403b5002003b403b703bc03b903bf03c503c103b303ae03c303b503b9002003bc03c003bf03c103bf03cd03bd002003bd03b1002003b103bd03bf03b903c703c403bf03cd03bd002003bc03b5002003c403bf0020004100630072006f006200610074002c002003c403bf002000410064006f00620065002000520065006100640065007200200035002e0030002003ba03b103b9002003bc03b503c403b103b303b503bd03ad03c303c403b503c103b503c2002003b503ba03b403cc03c303b503b903c2002e> /heb <feff05d405e905ea05de05e905d5002005d105d405d205d305e805d505ea002005d005dc05d4002005db05d305d9002005dc05d905e605d505e8002005de05e105de05db05d9002000410064006f006200650020005000440046002005d405de05d505ea05d005de05d905dd002005dc05d405d305e405e105ea002005e705d305dd002d05d305e405d505e1002005d005d905db05d505ea05d905ea002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e05d005de05d905dd002005dc002d005000440046002f0058002d0033002c002005e205d905d905e005d5002005d105de05d305e805d905da002005dc05de05e905ea05de05e9002005e905dc0020004100630072006f006200610074002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word introduction for volume 3_ issue 2 editor’s introduction modern society is a competitive society where reducing existent costs is very important, which also is the ultimate goal of risk analysis and crisis response. in the present issue such efforts have been embodied in these papers. this issue contains 6 papers. the first three contributions are written in english and the last three in chinese with english abstracts. risk assessment category includes the following three papers. the first paper "the assessment of risk caused by fire and explosion in chemical process industry: a domino effect-based study" by farid kadri, eric chatelet and patrick lallement, presents a method for risk assessment of domino effects caused by heat radiation and overpressure on industrial sites. this methodology is based on the probabilistic models and the physical equations. it allows quantifying the effect of the escalation vectors (physical effects) in industrial plants. the second paper “on the renewal risk model with constant interest force" by thampi and jacob, considers a renewal risk model with constant interest force for an insurance portfolio. in the paper, the authors discuss equations for the survival probability and its laplace-stieltjes transforms have been obtained. it provides recursive algorithm for the upper and lower bounds for the ruin/survival probability under interest force. finally, an exponential integral equation for the survival probability is derived. the third paper is “decision-making model in the environment of complex structure data” submitted by fusheng yu and shihu liu, aims to discover a underlying community structure of the data by taking all aspects of original information into account. in this paper, the considered data is diversity, not only in structure but also in representation. what is more, a missing data compensation method is proposed by considering the information losing situation in practical decision making problem. research shows that this model has great maneuverability. especially, the proposed decision model seems more consistent with the actual decision problem, than decision model with single data structure. in order to evaluate the environmental impact of a hazardous waste incinerator, in paper "health risks of environmental exposure to pcdd/fs near a hazardous waste incinerator in catalonia, spain " by montse mari, et al., 30 vegetation and soil samples were alternatively collected. the data were compared with those of previous campaigns, and especially with pcdd/f levels in the same monitors reported in the baseline study. it demonstrates that the health risks associated to environmental exposure of pcdd/fs are currently within acceptable ranges. the paper “meteorological grading indexes of water–saving irrigation for corn" by xiao, huo, et al., builds the water– saving irrigation meteorological grading indexes of different development stages of maize, aiming to optimize irrigation and provide a technical guide on agricultural production. the data are very useful for researchers who study agrometeorological hazard risk. there is one paper in the risk analysis, titled "research on characteristics and formation mechanism of landslide disaster in red soil hilly region of south china" by gong, huang, zhang, analyzes the regularity of landslides regional differentiation based on the field investigations and the available data in the red soil hilly region of south china. this paper reveals the growth characteristics and disaster mechanism of landslide by study the relationship between the characteristics of intensity and frequency of landslide and meteorological factors, hydrologic geology, geologic factors, landform and human factors and so on. based on the above results, a regional landslide prediction model is obtained by quantification approach in this paper. many thanks to all of referees who have given strong support to jracr. and we also thank the authors very much for their submissions and revising. editor-in-chief chongfu huang professor, beijing normal university email:hchongfu@gmail.com editorial assistant junxiang zhang associate professor, huangshan university email: peonyzjx@126.com journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 65 published by atlantis press copyright: the authors 65 atlantis press journal style editor’s introduction for volume 5, issue 4 the changing world always is uncertain so that various risks exist everywhere, endless crises always happen. jracr, launched by sra-china in 2011, is becoming an important journal in risk analysis and management, although it hasn’t been indexed by sci. this issue contains 6 papers. among them, 2 are written in english and 4 in chinese with english abstracts. in the paper “analysis of a risky two unit system under marked process incorporating two repairmen with vacations ” by tiwari and singh, the supplementary variable technique, laplace transformation and copula are used to obtain various transition state probabilities, reliability, availability, cost analysis and the steady state behaviour of the system. to make a comparison between the measures obtained using two families of copulas gumbel-hougaard family and bivariate clayton, the model has been solved with the help of gumbel-hougaard family of copula and bivariate clayton copula both in two different cases. the paper “sweden’s capacity to prepare and respond to a terrorist attack on rail-bound traffic” by veronica strandh, takes an interest in evolving collaborative practices in crises caused by terrorism targeting rail-bound traffic. sweden provides the empirical focus, by examining current preparedness processes, this article offers an important perspective on inter-organizational collaboration; that is, the perspective of the involved actors themselves. this study relies on 20 interviews with key actors in the response system and rail bound traffic actors. there are two papers in risk assessment. in the first paper “research on urban waterlogging disaster risk assessment based on arcgis and mike flood” by zhang, et al., chunyang zhang, an urban drainage system model of shijiazhuang city was built based on the mike21, mike urban and mike 11 modules in mike flood platform in order to provide a storm sewer drainage capacity and waterlogging risk evaluation. the results show that, in shijiazhuang, 64.5% of storm sewer drainage criteria is less than 1-year return period storm, and only 28.9% of storm sewer can meet the drainage criteria of 2-year return period storm. the second paper “regional risk assessment of earthquake-triggered landslides” by tian, et al., presents a review on the methods for risk assessment include the evidence-weight model, information value model and certainty factor method, logistic regression model, artificial neural networks, support vector machine method, newmark displacement model, analytic hierarchy process, and so on, and puts forward some ideas about the research trend of seismic landslide hazard evaluation. the paper “research on construction of major drainage system for urban area of shijiazhuang” by zhang, et al., introduces the relationship between major and minor drainage systems, as well as the significance of major drainage system construction based on the comprehensive plan of drainage and waterlogging prevention of shijiazhuang. according to the current problems of urban drainage systems of shijiazhuang, two-dimensional surface flow modelling tools was applied, with the consideration of urban master planning, major drainage system was established. in “the variable characteristics and response to climatic factors of the runoff in the downstream areas of the yellow river under the background of global change”, hong, et al. calculated and analyzed the runoff of huayuankou and lijin hydrologic stations of the areas from 1951 to 2012, and obtained four important conclusions. many thanks to the referees for their strong support and kind help. and also thank the authors very much for all their submissions. editor-in-chief: prof. chongfu huang email: rim_007@163.com email: hchongfu@126.com publication chair of sra-china: prof. mu zhang director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 199 published by atlantis press copyright: the authors 199 microsoft word introduction for volume 2, issue 3 editor's introduction in the global risk society, particularly in developing counties, facts are more uncertain, data are incomplete, values in dispute, takes high and decisions urgent. to reduce existent costs, there are more awakening of risk consciousness and there more actions in crisis response. in the present issue of journal of risk analysis and crisis response (jracr), volume 2, issue 3 (2012), such efforts have been embodied in these papers. this issue contains 8 papers. the former 4 contributions are written in english and others in chinese with english abstracts. the papers can be divided into four categories: crisis response, risk assessment, safety management and applications of information technology in crisis response. the first category includes three papers. the first paper "crisis transmission: global financial crisis" by alam, analyzed the emp-based crisis proxy of eight countries from q1 2001 – q2 2010, to explore the channels of crisis transmission with regard to the global financial crisis. it is concluded that the competitive devaluation effect did not play significant role in the transmission of the global financial crisis whereas wake-up call and cash-in effects were the major contributing transmission channels. based on the author’s estimations, indonesia, japan and russia showed signs of wake-up call effect, whereas brazil and japan recorded cash-in effect. the second paper "research framework for emergency rescue of disasters and crises " written by chinese police officers, zhang, you, li and du, comes up with research framework of professional rescue: the emergency organization and command of multi-forces on strategy-battle-tactics level, the risk analysis of multi-disasters events and the risk management and control of rescue, the standardization of rescue skills and equipments and professional training, and the information platform combining application system, database and the preliminary analysis of main sources. the paper "assessment method of emergency preparedness system vulnerability based on the complex network theory" by jiang, proposes a novel assessment method of emergency preparedness system vulnerability (epsv) to evaluate emergency system construction. the results from the assessment epsv of 3 regional governments show that the emergency preparedness complex network is rational, and the method is effective and feasible. the three papers in the risk assessment category offer tools for disaster reduction. the first paper “risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time scales” by zhao and zhang, suggests a new method on base of metrological data and crop’s vulnerability models assess risk variability among months on county scale. through the case study in three northeastern provinces of china, a serial of risk maps on county-level and month-level scales were produced to be used to recognize the rules of risk variability in space and time. the second paper “risk analysis of water resources crisis in the lancang mekong river drainage basin under the background of climate change” by zhang, zou, hong, zhou, zeng, carries out the risk assessment of water resources crisis in the lancang mekong river basin under climate change, aiming at the deteriorating its variation trends and potential crisis of water resources. the third paper "disaster management system in nepal – policy issues and solutions" by gangalal, former education minister of nepal, points out that nepal is highly vulnerable to natural and human journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 155-156 published by atlantis press copyright: the authors 155 induced disasters due to various aspects and then proposes that advanced technology and adequate resources are needed to alleviate the disasters. besides, training and other educational programs are suggested to raise awareness and disaster management act and policy for disaster risk reduction activities are also needed in nepal. there is one paper in the third category. “research progress on risk acceptance criteria of longdistance oil and gas transportation pipeline” by wu and zhang, summarizes the concepts and definition methods of acceptable risk, investigates several common determination principles and their application status of risk acceptance criteria, analyze the differences between long-distance oil and gas pipeline and other industrial facilities in several aspects and also discusses the determination methods and standards of acceptable criteria of different kinds of risks. last, suggestions were proposed on the part of investigating risk acceptance criteria and the applicability of the standards abroad. the last category includes one paper “gis-based micro-simulation queue model for vehicle evacuation” by ma, liu, wang, lin and lo, proposes a so called gis-simqueue vehicle evacuation model based on gawron’s queue model. this model comprises a two-stage process by considering both the dynamic capacity and static capacity of the routes and integrates with a geographic information system (gis) component for network analysis and post-processing of the simulation results. results indicate that the model can reproduce the congestion and spillback of peak traffic, which is in agreement with empirical ones. i sincerely thank the area editors and referees for their strong support and kind help. and also thank the authors very much for all their submissions. thanks jracr team members: miss jun guo，dr. junxiang zhang, dr. ye xue, and dr. fuping yan. editor-in-chief chongfu huang professor, beijing normal university email:hchongfu@gmail.com published by atlantis press copyright: the authors 156 atlantis press journal style editor’s introduction for volume 6, issue 2 during editing this issue, many of us have realized that the uncertainty in the world increases. jun 24, 2016, the united kingdom has voted to leave the european union. severe weather, including heavy rain, strong winds and hail, has continued to affect southern china, causing more damage and casualties. risk analysis and crisis response are becoming powerful tools to study and manage the uncertainty related to safety in the real word. such efforts have been embodied in the six papers published in this issue, where 4 contributions are written in english and 2 contributions in chinese with english abstracts. the objectives of the paper “living with flood: a livelihood resilience approach of rural people in bihar, india” by li and et al., are (1) to assess the mediating role of risk perception, and flood preparedness between flood experience and livelihood resilience; and, (2) to assess the mediating role of risk perception, and flood preparedness between flood education and livelihood resilience. the primary data were collected from 472 households by using multi-stage random sampling technique from seven blocks in river basins of ganga and kosi in the district of bhagalpur, bihar. the study recommends that the active involvement of the local people can be made mandatory with due consideration to their indigenous knowledge, flood experience, and flood education in order to make flood measures effective and successful. the paper “overview on the developments and applications of hesitant fuzzy sets: an uncertain decision making tool” by bin zhu, zeshui xu, reviews the main relevant papers published from 2010 to 2016, relating to the interpretations, modellings, and applications of hesitant fuzzy sets. it seems that the suggested hesitant fuzzy sets could help us to deal with real-life decision maker problems related risk issues. “application of risk estimation of noise-induced hearing loss method in evaluations of occupational disease hazards in construction projects in china” by li and et al., shows that the iso 1999:2013(e) risk model is effective to evaluate the occupational disease hazards in construction projects in china. based on the data collected by occupational hygienic investigations and measurements, the consequences and probabilities of noise-induced hearing loss could be predicted by using the model when the appropriate frequency combinations of interest and fences have been chosen according to the need, and then the corresponding management measures should be taken. in paper “study on performance evaluation of government comprehensive supervision for safety production based on balanced score card --a case study in shandong province, china”, author, yu hao, established a performance evaluation indicator system of safety production supervision within city and county two levels for the local government according to the theory of key performance indicators. 4 first-level indicators and 40 secondary indicators have been designed. the indicator system was used to study the safety production of tai’an city, shandong province, china. there are two papers in regional response to climate change. the first paper “a study of the impact on soybean potential under climate change” by ding and et al., extended and improved the parameters of soybean in agricultural ecology zone (aez) based on the 22 soybean observation stations in the major planting area from 1981-2011 to achieve china-aez. and then the authors simulate the impact of climate change on soybean. the results show that: the simulation of china-aez has been improved a lot. the second paper “the analysis on the effecting factor of drought disease in qingyang, gansu” by xiaodong wang, analyzes change of air temperature and precipitation change occurred from last half century to the beginning of this century in the qingyang region, gansu province, china. considering the relation between el-nino, la-nina and drought disease, the author wants to find out the reason of drought disease under the global warming, and to contribute to the theory studying of relieving poverty. we sincerely hope our reader will find this issue’s information on risk analysis and crisis response interface useful. thanks to the referees for their strong support and kind help. and also thank the authors very much for all their outstanding contributions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 47 published by atlantis press copyright: the authors 47 atlantis press journal style introduction for volume 7, issue 1 a modern society is full of competitions. to reduce existent costs, the ultimate goal of risk analysis and crisis response, is an important means to enhance competitiveness. in this issue of the journal of risk analysis and crisis response (jracr), volume 7, issue 1 (2017), such efforts are embodied in the included papers. this issue contains 6 papers. there are 5 contributions written in english and 1 contribution in chinese with english abstracts. the papers can be divided into five topics: risk decisions in the marine environment, financial risk, disaster economics, risk analysis related to chemical engineering and equipment, and rural tourism safety. the paper on risk decisions in marine environment entitled“multicriteria decision aid applications to support risk decisions in the marine environment: locating suitable transshipment areas”by stavrou and ventikos, aims to develop an mcda model to support decision makers in marine environments. the utastar method is employed for the selection of the best area for ship-to-ship transfer of cargo. this method relies on the hypothesis that both the actions and the corresponding criteria interact with each other over time, constructing and formulizing the decision aid model. the use of real data for the tuning of the model parameters may lead to the optimal compatibility between model and decision-maker cognition. there are two papers on financial risk. the first one“research on initial trust model of mobile banking users”by sun, et. al. studies the key factors that affect the trust mechanism of mobile banking, such as structure assurance, ubiquity, information quality, initial trust, perceived ease of use and experience. the results show that structure assurance is an important factor influencing trust tendencies, and ubiquity, information quality, perceived satisfaction and perceived ease of use are the key factors that affect initial trust and experience. the initial trust directly affects users' sense of experience of mobile banking and decides the tendency of trust, which affects the actual use. therefore, mobile phone banks need to pay more attention to the initial sense of trust and experience to promote user acceptance and trust mobile banking services. the second paper “review of research on credit risk management for rural credit cooperatives” by song, li, xiao, is an examination and summary of the research on credit risk management for rural credit cooperatives. the main content of this paper contains four aspects of credit risk: classification and characteristics, problems and factors, model and evaluation, and countermeasures and suggestions for rural credit cooperatives. the paper on disaster economics, entitled“a critical review on the economics of disasters”by yu and tang, presents a research framework of the economics of disasters, including disaster damage assessment, mitigation theory, disaster short-term economic impact, relationship between disasters and long-term economic growth, the value of life assessment, disaster risk management, and disaster recovery and reconstruction theory. there is one paper on risk analysis related to chemical engineering and equipment. the paper “optimization of ventilation and alarm setting during the process of ammonia leak in refrigeration machinery room based on numerical simulation”by liu, et al, discussed the velocity and concentration field distribution of an ammonia leak through simulation of a refrigeration machinery room using fluent software. the ventilation system of the room is optimized in three aspects, which are air distribution, ventilation volume and discharge outlet. the influence of the ammonia alarm system through ventilation is also analyzed. the results show that it is better to set the discharge outlet at the top of the plant than at the side of the wall, and the smaller the distance between the air outlet and the ammonia gathering area, the better the ventilation will be. air flow can be improved and the vortex flow can be reduced if the ventilation volume, the number of air vents and the exhaust velocity are reasonably arranged. not only does this ensure a functioning alarm system, but also the scope of the detection area could be enlarged if the detectors are set on the ceiling of the refrigeration units or the ammonia storage vessel. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 1–2 ___________________________________________________________________________________________________________ 1 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). the last paper focuses on rural tourism safety. “research on mechanism of risk coupling for rural tourism safety” by luo, presents an analysis model of risk coupling for rural tourism safety accidents. first, based on regional environmental risk system theory, the rural tourism safety system is analyzed. second, under the four dimensions of man, machine, environment, and management, the diamond model is established for rural tourism safety risk. then, based on the connotation of coupling, the author provides a definition and description of the types of risk coupling in rural tourism safety accidents. finally, based on the periphery theory and trigger working principle, a risk coupling mechanism model of rural tourism safety was established according to the “ rural tourism risk trigger”. the results provide countermeasures and suggestions for rural tourism safety and risk management. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang and prof. junxiang zhang, the respective publication chair and deputy chair of the society for risk analysis – china, who devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 1–2 ___________________________________________________________________________________________________________ 2 http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cmfd&sfield=kw&skey=+regional+environmental+risks http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cmfd&sfield=kw&skey=+regional+environmental+risks http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cjfq&sfield=kw&skey=the+diamond+model http://210.34.248.29/kcms/detail/%20%20%20%20%20%20%20%20%20%20%20%20search.aspx?dbcode=cjfq&sfield=kw&skey=types+of+risks microsoft word a study of seismic macroeconomic losses based on monte carlo method.doc a study of seismic macroeconomic losses based on monte carlo method—take tangshan city as an example qing wu*, mengtan gao institute of geophysics, china earthquake administration, beijing, 100081, china received october 26, 2018 accepted november 10, 2018 abstract the monte carlo method is used to simulate seismic sequences. for each earthquake in the sequence, the ground motion parameters of each site are calculated by the attenuation relationship, which is introduced into the vulnerability of the macroeconomic with gdp loss as an indicator. the model gives the exceeding probability curve of macroeconomic losses. this method provides a straightforward way to estimate the level of economic loss of a city in future earthquakes. it is of great importance to decision-making reference for formulating the corresponding earthquake prevention and mitigation countermeasures at present stage, and also provides a reference for insurance companies to determine the rates. keywords: monte carlo method, macroeconomic vulnerability model, exceeding probability curve 基于蒙特卡罗方法的地震宏观经济损失研究——以唐山市为例 吴 清*，高孟潭 1.中国地震局地球物理研究所，北京 100081 摘 要：本文利用蒙特卡罗方法模拟地震序列，对于地震序列里的每一个地震通过衰减关系计算得到各场 点地震动参数值，带入到以国内生产总值 gdp 损失为指标的宏观经济易损性模型，给出了地震宏观经济损 失超越概率曲线。此方法为估计一个城市在未来地震中的经济损失水平提供了一种简单直接的思路，对于 在现阶段制订相应的防震减灾对策具有十分重大的决策参考意义，更为保险公司厘定费率提供借鉴。 关键词：蒙特卡罗方法；宏观经济易损性模型；超越概率曲线 1. 引言 灾害定量化是地球科学、工程科学和社会经济 科学的交叉领域。人口城市化是全球的发展趋势。现 代城市在抗御自然灾害方面，比过去任何时候都显得 更为脆弱[1]。城市化在给人类带来富裕的物质生活和 丰富多彩的文化生活的同时，也会因人口、财富在有 限空间的高度集中，造成维系城市正常运转的基础设 *通讯作者：吴清(wuqing908@sina.com) 施、生命线工程、建筑物的日益庞大和复杂。尤其是 gdp值高的城市，经济发达、人口稠密，单位面积的 社会财富远远高于全国的平均水平。一旦这些重要城 市遭受大地震袭击，将会造成重大的人员伤亡和巨额 的经济损失，甚至会对整个国家的经济发展造成巨大 冲击。我国许多大城市和城市群位于易发生大地震的 大型活动断裂带附近，其中包括 12 个省会城市。改 革开放以来，大量的人口、资产和重大设施快速向这 些城市集中。暴露在具备发生 7 级以上大地震的危险 区域的人口、资产以指数方式迅速累积，导致地震风 192 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 险飙升 [2] 。如何分析一个城市在未来地震中的经济损 失水平，对于在现阶段采取相应的防震减灾对策具有 重大的决策参考意义。 传统的地震灾害损失研究通常采用易损性分类 清单方法，要求对研究区内建筑设施建立分类数据 库，并详尽的收集资料用以建立震害矩阵。但面对现 代社会的快速发展，使得传统的方法在收集并及时更 新资料方面都存在着明显的局限性，而这种详细的分 类资料在许多地区甚至是难以获得的。陈颙等[3]在执 行联合国“国际减轻自然灾害十年”计划中论证了可 以将承灾体脆弱性转化为国民生产总值（gdp）脆弱 性，gdp资料容易获得且更新及时，随后陈颙等[4-5]、 陈棋福等 [7-8] 、米宏亮 [9] 建立了适用于不同尺度和地区 的以gdp损失为指标的宏观经济易损性模型。刘吉夫 等 [10-12] 在此思路上开展了地震灾害损失预测研究，完 善了该分析方法的人口和gdp分配方案，以烈度区为 分配单位，建立了地震宏观风险分析模型。王晓青等 [13] 将此方法用于汶川地震损失快速评估中，其结果与 袁一凡等 [14] 进行现场调查得出的经济损失结果进行 了比较，结果较为一致。杨斌等 [15] ，范志伟等 [16] 都通 过实际震例验证了王晓青等 [13] 提出的宏观经济易损 性模型在小尺度空间范围内震后经济损失快速评估 中的可行性。 过去的震害损失预测分析，一般在考虑地震危险 性时是有概率水平的，可是在分析震害及经济损失 时，没有充分考虑这些因素的概率特征。刘本玉等[17] 以全概率的公式为基础，建立了区域震害经济损失预 测的概率分析模型，综合考虑了地震、场地、结构和 当地经济发展水平的影响。但是在公式推导过程中， 由于各因素之间的关系很难得到，而做了许多简化与 平均，在实际应用中比较难以实现与推广。 蒙特卡罗方法则是一种基于合成地震序列的地 震危险性计算方法，该方法根据地震活动性模型可以 生成满足某一区域内地震时空分布规律的地震序列， 对地震序列里的每一个地震通过衰减关系计算得到 各场点地震动参数值，进一步带入易损性模型，可以 对地震风险进行评估[18-23]。本文拟将蒙特卡罗方法与 宏观经济易损性模型结合起来，将蒙特卡罗方法模拟 计算得到的地震事件影响场带入到以国内生产总值 gdp 损失为指标的宏观经济易损性模型，给出地震 宏观经济损失超越概率曲线。 2. 方法介绍 蒙特卡罗（monte carlo）方法是一种应用随机数 来进行计算机模拟的方法。此方法对研究的系统进行 随机观察抽样，通过对样本值的观察统计，求得所研 究系统的某些参数。其基本思想是当实验次数充分多 时，某一事件出现的频率近似于该事件发生的概率， 即 x p x ≈ 错误！未找到引用源。 (当 x 充 分大时) （1） 式中 p 是某一事件发生的概率；x 是实验次数； x 是 在 x 次实验中该事件出现的频次。 中国地震动参数区划图（gb18306-2015）依据 各个地区的地球物理资料，划分了地震带和潜在震源 区，建立了相宜的地震发生概率模型和空间分布模 型，并给出了各个地震带的基本参数。根据中国地震 动参数区划图的基本假定和地震活动性参数，可采用 以下步骤合成地震事件集[22]： （1）基于地震带内地震发生满足泊松分布的假 定，首先确定所要模拟地震序列的时间长度 t 和地震 带内地震的年平均发生率 4ν 。随机产生一个以 t 和 4ν 为参数的泊松分布随机数 r，那么 r 就是所要模 拟时间长度为 t 的该地震带上地震的数量。 （2）基于地震带内地震活动的震级分布满足截 断的 gutenberg-richter 关系（震级-频度关系）的假 定，以及 小震级水平 0 m 和 大震级上限 uz m 来确 定震级。 震级-频度关系表达式为： （2） logn a bm= − 式中，a, b 为系数，n 为震级大于等于 m 的地震个数， 4, 4.1, 4.2, , uz m m= l ，中国地震动参数区划图的起算 震级为 4 级。则地震事件累积数量的值为： （3） (m) ea bmn −= 取 0.1mδ = ，则有 （4） (m) n(m m)n > +δ 随机生成一个符合在 0 和 1 之间均匀分布的随 机数，判断它是否在 (m m) (4) n n + δ 错误！未找到引用源。 与 (m) (4) n n 之间，如果是，则确定地震事件的震级 m。 （3）震中位置的确定。首先，确定地震事件所 在的潜在震源区 h。根据上步所确定的震级 m 判定 地震所在的震级档 d，因为每个震级档的地震落在各 个潜在震源区的概率 是已知的并且是归 1 的， 那么随机产生一个符合在 0 和 1 之间均匀分布的随机 (h)pd 193 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 数，判断它在是否在 错误！未找到引用源。 与 之间，如果是，则确定地震事件所在的潜 在震源区 h。基于震中在潜在震源区内均匀分布的假 定，随机抽取一点作为地震的震中位置。 h 1 (h) h d p = ∑ 1 h 1 (h) h d p − = ∑ （4）根据潜在震源区衰减长轴方位角的概率，抽 样确定地震方位角θ 。 至此，确定了一个地震的基本要素。重复（2）~ （4）步直至所需的该地震带上的地震数目 r，综合 考虑所有可能对场点造成影响的地震带，由此确定出 来一个地震序列，完成一次抽样。通过大规模抽样模 拟，形成地震概率事件集。 时间长度若设置为 1 年，则完成一次抽样模拟得 到的地震序列本文称为 1 年地震序列。时间长度设置 为 10 年，即称为 10 年地震序列。 根据蒙特卡罗方法原理，抽样次数越多结果越精 确，但抽样次数增多会显著增加计算量。因此，为了 统筹考虑计算量和结果精确程度，需进行不同抽样次 数的试验，当计算结果趋于稳定时，即认为不再需要 增加抽样次数。 对于概率事件集里的每一个地震，以地震的经纬 度为原点，衰减长轴方向为正方向，将所有场点转换 到平面坐标下。根据中国地震动参数区划图的椭圆烈 度衰减关系，采用 优椭圆搜索的数值计算方法计算 地震在各个场点处的烈度值 i。 宏观经济易损性模型可表示为： (i, gdp) c a i b f = ⋅ ⋅ （5） 式中，f(i,gdp)代表 gdp 损失率，i 为地震烈度，a、 b 为系数，c 为修正系数。a、b 值直接从回归关系 式获得，修正系数 c 一般取 1.0。 对于一个行政区域（市），将其所辖的区县或乡 镇各自视为一个场点，每个场点附有各自的 值。 对于用蒙特卡罗方法模拟得到的地震序列里的每一 个地震，通过烈度衰减关系可以同时计算得到所有场 点的影响烈度值 i gdp i i ，带入宏观易损性模型，可以计算 得到这个地震对行政区域可能造成的损失： （6） j 1 b i i loss c a i gdp = = ⋅ ⋅ ⋅∑ i 式中 j 是区县或乡镇总数。这样地震序列里每一个地 震对应一个 gdp 损失值，选取序列里的 大值作为 一次模拟的结果。随机模拟 x 次，得到 x 个地震序 列，统计超越给定经济阈值的相应地震序列个数 x， 计算得到超越概率值。 在计算过程中需考虑经济及社会财富的动态发 展，即 gdp 值的增长变化。如果以当前资料搜集年 的 gdp 产值为基础，用简单的年均增长率 q 来考虑 的话，则第 k 年的国内生产总值为 1(1 )kgdp q −× + ， 如此 t 年内的 gdp 总值为 1 1 (1 q) 1 (1 ) tt k k y gdp q gdp q − = + − = × + = ×∑ （7） 则在 t 年内以 gdp 描述的社会平均财富为 /gdp y t= （8） 将之代替损失公式中的 gdp 值。 本文收集整理了自 1989 年以来，开展正式地震 现场震害调查和损失评估的中国大陆地震的灾害损 失资料，拟回归宏观经济易损性模型的系数。但由于 收集到的灾评报告基本以划分的评估区或者行政区 为单位给出经济损失估计值，而评估区和行政区往往 与烈度区并不重合，导致经济损失和人员伤亡有时无 法归算到烈度区。而宏观经济易损性模型的高烈度部 分是根据拟合公式外推的，因此需要更多的大地震震 例来验证。 王晓青等[13]给出了宏观经济易损性模型： （9） 8 9.8082 10 11.585 11 11.377 2 10 2700 (i, gdp) 2 10 2700 10000 4 10 10000 i gdp f i gdp i gdp − − − × ⋅ < = × ⋅ ≤ < × ⋅ ≥ ⎧ ⎪ ⎨ ⎪ ⎩ 人均 人均 人均 该模型用于汶川地震损失快速评估中，其结果与 袁一凡等 14 进行现场调查得出的经济损失结果进行 了比较，结果较为一致。杨斌等 [15] ，范志伟等 [16] 都通 过实际震例验证了王晓青等 [13] 提出的宏观经济易损 性模型在小尺度空间范围内的可行性。因此，本文也 借鉴这一模型进行计算。 3. 计算实例 1976 年 7 月 28 日，唐山发生了举世震惊的 7.8 级强烈地震，使百年工业城市瞬间毁于一旦。地震造 成 24.2 万人死亡，16.4 万人重伤，铁路交通、邮电 通信、供水供电完全中断，道路桥梁、工厂设备和农 田水利设施遭到严重破坏，直接经济损失高达 54 亿 元。灾情之重、损失之巨，举世罕见。经过 10 年的 恢复重建和 30 年的振兴发展，今天的唐山已然成为 一个现代化新型都市，成为国际灾后城建的典范。然 而唐山依然处于发震构造带上，其地震风险随着唐山 194 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 市经济的高速发展和人口密度的集中高度飙升，甚至 远超 40 年前。本文拟通过蒙特卡罗方法给出唐山市 t 年内地震宏观经济损失超越概率曲线，为城市减灾 战略规划需要和地震应急需求提供参考。 图 1(a)为唐山市行政区域分布图，将 18 个区县 各自视为一个场点，每个场点附有各自的 值； 图 1(b)是唐山地区周边地震带分布图；图 1(c)是覆盖 唐山地区的潜在震源区分布图；图 1(d)为唐山地区蒙 特卡罗方法 1 年地震序列模拟 100000 次中的 5.0≤ m<8.0 的地震震中分布示意图。表 1 是计算所用到的 中国地震动参数区划图中地震带的地震活动性参数 列表；表 2 是 2016 年唐山市各区县 gdp 值（当年价） 和年增长率[24]。 i gdp 利用gb18306-2015 中国地震动参数区划图华北 地区椭圆烈度衰减模型，按 优椭圆搜索法逐次计算 地震序列里的每一个地震在 18 个场点处的烈度值i。 华北地区烈度椭圆衰减关系如下[25]： 长轴方向 （10） 5.7123 1.3626 4.2903 lg(r 25)i m= + ⋅ − ⋅ + 短轴方向 （11） 3.6588 1.3626 3.5406 lg(r 13)i m= + ⋅ − ⋅ + 0.5826σ = （12） 式中，i 为烈度，m 为震级，r 为震中距，σ 为 标准差。 由于蒙特卡罗方法抽样次数越多结果越精确，为 了统筹考虑计算量和结果精确程度，分别进行了 1 年 地震序列的 10 万次、100 万次、1000 万次以及 10 年 地震序列的 10 万次和 100 万次抽样计算，统计不同 模拟次数下和不同时间长度的经济损失超越概率，结 果见图 2。由图 2 可以见到，对于 1 年地震序列，1000 万次的模拟足以给出稳定平滑的年超越概率曲线；而 对于 10 年时间长度的地震序列，100 万次的模拟也 呈现稳定平稳的超越概率曲线。 由超越概率曲线可以看到，以 2016 年的经济水 平为基础，唐山市 100 亿元地震经济损失的年超越概 率约为 2%，1000 亿地震经济损失的年超越概率约为 0.38%，10000 亿地震经济损失的年超越概率约为 0.02%，10 年的宏观经济损失超越概率也符合概率计 算规律。 表 1 计算所用到的地震区带地震活动性参数列表 地震区带名称 b 值 4ν uzm 潜在震源区数目 郯庐地震带 0.85 4.0 8.5 107 华北平原地震带 0.86 4.6 8.0 97 汾渭地震带 0.78 2.5 8.5 57 东北地震区 1.00 5.0 7.5 117 表 2 2016 年唐山市各区县生产总值（gdp） 县（市）区 地区生产总值（亿元） 累计增长% 全市 6354.87 6.8 曹妃甸区 367.88 7.2 海港开发区 130.59 8.5 芦台开发区 43.81 12.6 汉沽管理区 32.09 10.0 路南区 123.10 7.5 路北区 152.05 7.0 古冶区 193.26 4.3 开平区 125.55 6.8 丰南区 619.36 6.9 丰润区 641.97 4.6 高新开发区 126.07 8.0 迁安市 920.16 3.7 遵化市 512.24 5.9 滦 县 462.07 6.2 滦南县 336.55 6.1 乐亭县 345.18 7.6 迁西县 425.97 6.5 玉田县 380.91 5.8 195 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 图 1(a) 唐山市行政区域分布图 图 1(b) 唐山地区及周边地震带分布图 图 1(c) 唐山地区潜在震源区分布图 图 1(d)唐山地区 1 年地震序列模拟 100000 次中 5.0≤ m<8.0 的地震震中分布示意图 4. 结论与讨论 基于中国地震动参数区划图（gb18306-2015）， 利用蒙特卡罗方法建立概率地震事件集和地震动影 响场，带入到宏观经济易损性模型，简单直接的统计 归算未来一段时间内城市经济损失水平的超越概率。 通过大规模模拟抽样，以 2016 年经济水平为基础， 得到唐山市未来可能面临的地震经济损失超越概率 曲线，此方法给地震经济损失估计赋予了概率的意 义。 196 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 由于蒙特卡罗方法是逐次计算每个模拟地震可 能对城市造成的经济损失，那么很容易找到给城市造 成 大损失的那些地震，进而构建更为详细的地震情 景，以更精确的估计它们可能的影响。 蒙特卡罗方法原理简单直接，可以很容易被规划 者、施政者和保险公司理解接受。以此方法来估计一 个城市在未来地震中的经济损失水平，对于在现阶段 制订相应的防震减灾对策具有十分重大的决策参考 意义，更为保险公司厘定费率提供借鉴。 致谢 this study was supported by major research plan of institute of geophysics, china earthquake administration, basal research fund (dqjb17c08). 参考文献 [1] c. yong, p. wengtao, x. wenli, new feature of earthquake disaster in 21st century, advance in earth sciences, 19(3) (2004) 359-363. 陈颙,彭文涛,徐文立,21 世纪地震灾害的一些新特点[j], 地球科学进展,19(3),2004:359-363. [2] g. mengtan, a new generation of seismic zonation map of china for national social and economic development, city and disaster reduction, (3) (2016) 1-5. 10 0 10 1 10 2 10 3 10 4 10 5 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 经济损失(亿元) 超 越 概 率 1年地震序列10万次抽样 1年地震序列100万次抽样 1年地震序列1000万次抽样 10年地震序列10万次抽样 10年地震序列100万次抽样 图 2 宏观经济损失超越概率曲线 高孟潭,新一代国家地震区划图与国家社会经济发展, 城市与减灾,(3),2016:1-5. [3] c. yong, z. hongren, research on quantification of earthquake disasters, recent developments in world seismology, (5) (1991) 5-9. 陈颙,朱宏任.地震灾害定量化研究.国际地震动态,(5), 1991:5-9. [4] c. yong, c. qifu, c. ling, vulnerability analysis in earthquake loss estimate, earthquake research in china, 15(2) (1999) 97-105. 陈颙,陈棋福,陈凌,地震损失预测评估中的易损性分析, 中国地震,15(2),1999:97-105. [5] chen yong, chen qifu, chen ling, 2001, vulnerability analysis in earthquake loss estimate, natural hazards, (16):1-16. [6] chen yong , chen qifu, liu jie , et al . , 2002,seismic hazard and risk analysis : a simplified approach . beijing, science press. [7] c. qifu, c. yong, c. ling, forecast and assessment of earthquake disaster losses using gross domestic product and population data, acta seismologica sinica, 19(6) (1997) 640-649. 陈棋福,陈颙,陈凌,利用国内生产总值和人口数据进行 地震灾害损失预测评估,地震学报,19(6),1997:640-649. [8] c. qifu, w. xiaoqing, l. jifu, study on seismic vulnerability of macroeconomic indicators (gdp) in mainland china, prediction of earthquake hazard zone and earthquake disaster loss in mainland china from 2006 to 2020, beijing, seismological press, 2007. 197 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 陈棋福,王晓青,刘吉夫,中国大陆宏观经济指标(gdp)的 地震易损性研究[a].2006—2020 年中国大陆地震危险 区与地震灾害损失预测研究编委会.2006—2020 年中国 大陆地震危险区与地震灾害损失预测研究[c].北京:地 震出版社,2007:252-264. [9] m. hongliang, research on macro-vulnerability of earthquake disasters and network realization of rapid assessment of earthquake disasters, m.s. thesis, beijing: the center for analysis and prediction, china earthquake administration, (2002), pp. 19-48. 米宏亮,地震灾害宏观易损性研究与震害快速评估的网 络 实 现 [d]. 硕 士 论 文 , 中 国 地 震 局 分 析 预 报 中 心,2002,19-48. [10]l. jifu, the study on applicability of macroscopic indicator vulnerability to a small-scale region, phd thesis, beijing: institute of geophysics, china earthquake administration, (2006), pp. 21-38. 刘吉夫,2006,宏观震害预测方法在小尺度空间上的适 用性研究 [d]. 博士论文 , 中国地震局地球物理研究 所,21-38. [11]l. jifu, c. yong, s. peijun, et al, on seismic risk assessment in mainland china, journal of beijing normal university (natural science), 44(5) (2008) 520-523. 刘吉夫,陈颙,史培军等,2008,中国大陆地震风险分析模 型研究,北京师范大学学报(自然科学版),44(5):520-523. [12]l. jifu, c. yong, s. peijun, et al, a study on seismic risk assessment in mainland china(ⅱ): life vulnerability model, journal of beijing normal university (natural science), 44(4) (2009) 404-407. 刘吉夫,陈颙,史培军等,2009,中国大陆地震风险分析模 型研究(ⅱ):生命易损性模型,北京师范大学学报(自然 科学版),45(4):404-407. [13]w. xiaoqing, d. xiang, w. long, et al, a study on fast earthquake loss assessment and its application to 2008 wenchuan m8 earthquake, acta seismologica sinica, 31(2) (2009) 205-211. 王晓青,丁香,王龙等,四川汶川 8 级大地震灾害损失快 速评估研究[j].地震学报,2009,31(2):205-211. [14]y. yifan, loss assessment of wenchuan earthquake, journal of earthquake engineering and engineering vibration, 28(5) (2008) 10-19. 袁一凡,四川汶川 8.0 级地震损失评估[j].地震工程与工 程振动,28(5),2008:10-19. [15]y. bin, m. zhaohui, the application of the disaster losses macroscopic forecasting model in the rapid assessment after earthquake, seismological and geomagnetic observation and research, 34(3/4)(2013) 213-216. 杨斌,马朝晖,灾害损失宏观预测模型震后快速评估,地 震地磁观测与研究,34(3/4),2013:213-216. [16]f. zhiwei, y. fan, c. xiaoyan, economic loss of rapid assessment after earthquake in zhangjiakou, journal of disaster prevention and reduction, 32(3) (2016) 15-18. 范志伟,杨凡,陈晓燕,基于宏观经济指标对张家口地区 地震灾后经济损失快速评估,防灾减灾学报, 32(3), 2016:15-18. [17]l. benyu, s. jingyu, j. jianjing, et al, total probability forecasting method for economic loss of seismic damage, world earthquake engineering, 19(1) (2003) 15-20. 刘本玉,苏经宇,江见鲸,李刚,震害经济损失的全概率预 测,世界地震工程,19(1),2003:15-20. [18]gao mengtan, pan hua, characteristics of random field for seismic zoning results. acta eismologica sinica, 15(1) (1993) 53-60. 高孟潭 , 潘华 , 地震区划结果随机场特征 [j], 地震学 报,15(1),1993:53-60. [19]musson, r. m. w. 1999, determination of design earthquakes in seismic hazard analysis through monte carlo simulation, j. earthquake. eng., vol.3, 463-474. [20]musson, r.m.w., 2000, the use of monte carlo simulations for seismic hazard assessment in the u.k., annaili di geofisica, 43(1): 1-9. [21]smith, w. d., 2003, earthquake hazard and risk assessment in new zealand by monte carlo methods, seismological research letters, 74(3): 298-304. [22]g. xing, probabilistic seismic hazard assessment based on monte carlo simulation, m. s. thesis, beijing: institute of geophysics, china earthquake administration, (2008), pp. 16-18. 郭星,2008,基于蒙特卡罗模拟的概率地震危险性分析 方法,中国地震局地球物理研究所硕士论文,16-18. [23]c. kaufmann, j. schwarz,2014, earthquake hazard and risk assessment tool using monte-carlo simulation techniques, tenth u.s. national conference on earthquake engineering frontiers of earthquake engineering july 21-25,anchorage, alaska. [24]tangshan statistical bureau, tangshan investigation team of national statistical bureau, tangshan statistical summary 2016, hebei: printing factory of tangshan municipal committee of the communist party of china, 2017, 46. 唐山市统计局,国家统计局唐山调查队,2017,唐山统计 概要 2016,河北:中共唐山市委机关印刷厂,46. [25]x. liang, study on the attenuation relationships of horizontal ground motion parameters near the source of rock site, phd thesis, beijing: institute of geophysics, china earthquake administration, (2011), pp. 104. 肖亮,水平向基岩强地面运动参数衰减关系研究[d],北 京:中国地震局地球物理研究所,2011,104. 198 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 192 -198 1. 引言 2. 方法介绍 3. 计算实例 4. 结论与讨论 致谢 参考文献 atlantis press journal style review of research on credit risk management for rural credit cooperatives* xin song1, li li1 †, lei xiao2 1pan-asia business school, yunnan normal university, 650092 kunming, china; email:1593666405@qq.com, 43063010@qq.com 2school of marxism studies, kunming university, 650214 kunming, china; email:391644664@qq.com abstract with the rapid development of rural micro-credit, whether the "agriculture, rural areas and farmers" problems have been effectively solved, whether the credit risk has been effectively controlled, these have become the focus of our attention to the rural economic environment. the main content of this paper contains four aspects: the classification and characteristics of credit risk, the problems and factors of credit risk, the model and evaluation of credit risk, the countermeasures and suggestions of credit risk. this paper reviews the research of credit risk management for rural credit cooperatives from the above four aspects, and makes a brief summary. keywords: rural credit cooperatives, credit risk, research review *corresponding author: li li, pan-asia business school, yunnan normal university, kunming, china. email:43063010@qq.com. †address：no. 298, 121 street, wuhua district, kunming 650092, china 1. introduction since the popularizing of rural credit cooperatives in china after 2000, the development of rural credit cooperatives in china has achieved a leap. especially in rural areas, the role of financial services for farmers is obvious. rural credit cooperatives have become major financial institutions of credit for farmers. in recent years, rural areas, agriculture and farmers issues have become the focus of rural economic development. because rural credit cooperatives play a most important role in chinese rural financial environment, their credit risk has also become the important question in rural economic research. relating to the personal financial services of rural credit cooperatives in china, credit business mainly includes rural household micro-credit loans and farmers group guaranteed loans. and micro-credit for farmers occupies the dominant position in credit business. rural household micro-credit loans are micro-credit loans that rural credit cooperatives issued to farmers for agricultural production and consumption, within the approved quota and time limit based on the credit reputation, assets and repayment ability of farmers. at present, rural household micro-credit has become one of the main ways to obtain credit loans in rural areas of china, which plays an important role in accelerating rural economic development, improving agricultural structure and promoting the coordinated development of urban and rural areas. on the research of farmers' credit risk, foreign scholars began earlier (70s), domestic scholars began late (90s). this paper reviewed the research based on the domestic and foreign credit risk in recent years, and this paper is divided into four parts: the classification and characteristics of credit risk, the problems and journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 21 received 29 december 2016 accepted 22 january 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). factors of credit risk, the model and evaluation of credit risk, the countermeasures and suggestions of credit risk. we will give a more comprehensive analysis of the current situation of credit risk in rural credit cooperatives, and make comments and conclusions. 2. the classification and characteristics of credit risk 2.1. the types of credit risks in rural credit cooperatives credit risk refers to the fact that the borrower cannot return the credit principal and interest for various reasons on time, which makes the actual income result deviate from the expected profit goal and suffer the loss of assets in the operation and management of the rural credit cooperatives. domestic scholars study earlier on the type of credit risk, can be divided into five categories, including moral hazard, credit risk, operational risk, management risk and natural risk. huang jian (2003) [1] points that the potential risks of micro-credit farmers cannot be ignored, destroys the rural economic environment, risk mainly credit risk, operational risk, management risk and market risk four aspects. wu baoshan (2006) argues that the main forms of rural credit risk are operational risk, management risk, local policy risk, farmer credit risk and loan transfer risk. dai lixin (2008) [2] divides credit risk into natural risk, borrower's risk, debtor's risk and government's risk by using 30 rural credit cooperatives and two banking regulatory bureaus in five districts of bd. zhu xinle (2010) [3] takes jilin province rural credit cooperatives as an example to study the risk types of rural micro-credit, the main risk is credit risk, operational risk and natural risk. based on the survey data of rural microfinance companies in jiangsu province, tan zhongming (2011) [4] finds that there are many risks in rural microfinance companies, including legal risk, credit risk, operational risk, liquidity risk and regulatory risk. in summary, although most scholars put the credit risk into five categories, but the credit level of rural credit cooperatives and farmers, natural risk and market risk are the objective factors, rural financial institutions face is mainly from the farmers' credit risk. when the farmer defaults on the small loan, the unsecured microloan is just relying on farmers' credit guarantee to repay the loan, there is a great deal of uncertainty. 2.2. characteristics of credit risk in rural credit cooperatives generally speaking, credit risk of rural credit cooperatives has three main characteristics. the first is objectivity, as long as there is credit activities, credit risk will not be transferred based on people's will and objective existence. it can be said, as long as the market economy, credit risk is inevitable. the second is spreading, the loss of funds caused by credit risk that not only affect the survival and development of rural credit cooperatives, more is caused by the associated chain reflected. the third is controllable, the rural credit cooperatives in accordance with certain methods and system can advance the credit risk identification, prediction, prevention and post resolve the matter, and the risk can be minimized by controlling. 3. problems and factors for credit risk 3.1. information asymmetry between borrowers and lenders duan wenjie (2009) [5] from the farmers point of view, that farmers in order to obtain loans, will inevitably hide their own negative information, coupled with staff negligence, rural credit rating is not necessarily reliable, leading to loans to non-trustworthy farmers, resulting in moral hazard. zeng zhiming (2010) [6] from the information asymmetry theory, believes that the credit sides often cannot have enough information, cannot make accurate judgments, which affect both the transaction behavior and market efficiency. in this case, farmers become a party with information advantage, resulting in moral hazard and adverse selection. sheng guanghua (2014) [7] argues that rural credit cooperatives cannot fully grasp the actual situation of farmer loan projects because of information asymmetry. therefore, from the farmer's point of view, the expected return of observing contracts are less than expected return of not complying with the contract. so the dominant strategy is farmers choose not to comply with the contract. 3.2. imperfect pre examination mechanism of rural credit cooperatives pre examination of the relevant qualifications of farmers is a prerequisite for the signing of the loan contract, but also an essential program of borrowing behavior. the journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 22 significance of the review is to assess the size of the loan risk, and thus determine the success of the loan transaction. yang xisun (2007) [8] believes that rural credit cooperatives lack of perfect credit system, a number of rural credit cooperatives in the credit rating of the farmers when only to see whether the credit situation of the farmers early, there is no complete economic household files data. zeng zhiming (2010) finds that personal information storage is only farmers basic information and loan repayment information through the investigation of hunan rural credit cooperatives. such as product quality, household and personal gains and losses default status are not complete collected, lack of query service. tan zhongming (2011) believes that the internal restraint mechanism of rural credit cooperatives is not perfect, the operational process becomes a mere formality, business inspection is mainly checked by credit workers, and risk awareness is very weak. 3.3. lower risk management level of rural credit cooperatives compared with the commercial financial institutions, huang boyong (2008) [9] argues that both hardware and staff quality are at a low level. in the identification, prevention of financial risks and innovative financial products, are obviously inadequate. zhu xinle (2010) believes that rural credit cooperatives, savings banks and other rural financial institutions in the large range of outlets, the good and bad personnel qualities are intermingled, the work hard avoids negligence and mistake. 3.4. other factors chavan p (2002) [10] reviews the empirical evidence of ngo-led micro-credit schemes in several developing countries and compared it with the indian rural credit relief program, which concludes that the practice of grameen bank creditors in bangladesh has led to the credit debt cycle. cipollini (2012) [11] studies the factors that affect the credit risk of european commercial banks, including liquidity and credit risk, asset size, income diversification and market power. jansson kh (2013) [12] presents and compares institutional frameworks for agricultural credit markets in selected european countries, conducts risk assessments of agricultural credit markets based on questionnaires of selected agricultural financial experts, finds that the most common reasons for refusing the loan applications are related to economic performance and the situation of farmers .li zhengbo (2006) [13] uses the logit model to analyze the sample survey data of shandong, shanxi and shaanxi provinces, and finds that age and educational level have a positive effect on the farmers' default. but the expenditure structure and income structure of peasant households will affect the farmers' default probability. in addition, the loan interest rate and loan term have an obvious impact on the probability of default. the higher the loan interest rate is, the shorter the loan term is, and the more likely the farmer defaults. wei lan (2013) [14] through the survey data of 300 rural households in liaoning province, to establish logistic regression model to identify the default farmers, believes that the number of family labor, farm income, total household assets, the use of the loans with peasant household credit risk is negatively related; family population, cultivated land area, household expenditures, the amount of loans with credit risk is positively related. 4. model and evaluation of credit risk for the evaluation of credit risk, foreign scholars mostly evaluate the credit rating or credit score of the company or institution, or calculate the default probability of the large commercial banks to manage the risk. for the evaluation of farmers and individuals, foreign scholars from developed countries have studied less. cipollini and fiordelisi (2012) takes the panel probit regression model to establish the credit risk score model by using the stable covariance matrix. malik (2012) [15] applies the data of a uk bank credit card portfolio to construct a markov chain behavior score model, incorporating economic variables and loan terms, to establish an investment portfolio in credit risk consumer loans for corporate credit risk research. dadson awunyo-vitor (2014) [16] uses descriptive statistics, variance analysis (anova) and heckman's two-stage regression model to determine the factors affecting agricultural credit rationing on data from 595 households in the brong ahafo district of ghana. on the contrary, the study of domestic scholars is more in-depth. in the aspect of system analysis and grading, analytic hierarchy process (ahp), delphi method and fuzzy mathematics method are more primitive, but these methods have too much subjective factor to determine the weight of the index, so credit journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 23 evaluation of farmers is difficult to be objective and impartial. hu yu (2007) [17] uses uncertain type of analytic hierarchy process (utahp) to determine the weight of indicators, which reduces the subjective degree of each index weight, and he divides the index system of farmer credit rating into four categories according to different attributes: family structural characteristics, solvency, business conditions, the credibility of the situation. nie yong (2009) [18] applies the multi-objective decision analysis method to the performance evaluation of farmers’ micro-credit, and can scientifically evaluate the comprehensive benefits of farmers’ micro-credit. wang jing (2011) [19] attempts to use the mutation evaluation method to classify three types of mutation system, including cusp mutation, dovetail mutation and butterfly mutation. this method not only reduces the subjectivity of multiobjective analysis when determining factors of index weight but also absorb the advantages of the method of fuzzy mathematics and analytic hierarchy process (ahp).finally, the sample data of rural credit cooperatives in yangling of shaanxi province are used to carry out an empirical study, which verifies the objective and rationality of the method. in addition, yao shuqiong (2012) [20] by empirical analysis of survey data of agriculture, rural areas and farmers credit cooperatives in yangling district of shaanxi province, the index variables were normality test, variance test and multicollinearity test, and establishes credit evaluation model of bp neural network. the results show that the overall accuracy of the model is 100% for the training samples, and 90% for the default samples of the test.hu haiqing (2012) [21] uses 151 small and medium enterprise sample data for empirical analysis, and constructs credit evaluation model based on svm and bp neural network. the results show that classification accuracy of svm model is significantly higher than bp neural network model. chi guotai (2015) [22] uses a complex comprehensive discriminate analysis to empirically study 2044households of large-scale commercial banks in china, the results show that education and engel coefficient is a key indicator to distinguish farmers’ default. and he demonstrates that the farmer loans of the ccc level can achieve the target profit of the bank and the farmer loans of cc level above can achieve the breakeven of the bank. 5. countermeasures and suggestions of credit risk 5.1. establish a guarantee system the credit business of rural credit cooperatives, especially micro-credit, is uncovered and unsecured, which is the root cause of credit risk. the basel accord also points out that the guarantee is the necessary tool to mitigate the credit risk. the establishment of group cooperation guarantee system, on the one hand, farmers can reduce the asymmetry of information with the lending institutions; on the other hand, group cooperation can obtain a larger amount of loans to alleviate the difficulty of the loans to farmers and the problem of low amount. xiong xueping (2005) [23] analyzes the gb model in bangladesh, and concludes that the system of farmer household guarantee and group center meeting system is a strong system guarantee for credit risk, which can be energetically implemented in densely populated areas. however, based on sample survey of 113 farmers’ cooperative economic organizations in guangxi province,yang xisun (2007) [8] finds that chinese implementation of group guarantee system is difficult. there are three reasons: the first is the trust between members of the group is difficult to estimate; the second is the group of farmers between the demand for loans and loan demand time cannot reach agreement; the third is implementation of the responsibility for breach of contract is not clear, the principal repayment of default is ambiguous. starting from the dual structure of the rural financial supply, han xiping (2014) [24] fully understands the connotation and the boundary of the form of security, to build a "dual structure, four levels, n form" credit guarantee system. 5.2. improve the internal management mechanism of rural credit cooperatives from the perspective of self-construction of rural credit cooperatives, ding yezhen (2006) [25] believes that credit institutions in rural credit cooperatives should strengthen staff training, improve the quality of staff, and establish employee incentive mechanism to reduce the credit risk. based on the credit control system, yang hongli (2009) [26] increases the macroeconomic information and agricultural information to establish early warning system and the buffer system of credit risk, which improves internal risk prevention and journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 24 control organizations of rural credit cooperatives and forms effective self-discipline mechanism. ma xiaonan (2013) [27] believes that the difficulty of credit risk control of rural credit cooperatives is the inherent risk of internal management mechanism. rural credit cooperatives should strengthen the reform of corporate governance structure to improve the property rights system of rural credit cooperatives, to achieve the interests to match the responsibilities of borrowers and lenders. tong yuanbao (2014) [28] studies the farmer credit risk in hainan province and finds that the rural credit cooperatives in hainan province innovatively uses the "four-deliver" mechanism and "integrity award" system to encourage farmers to accumulate good credit records. 5.3. other factors agriculture is a basic industry, relating to the livelihood of the farmers, agriculture with the characteristics of natural risk makes it more important to strengthen the construction of agricultural insurance system. the implementation of the agricultural insurance system, exploration of rural credit cooperatives and agricultural insurance cooperation mechanism, are conducive to improve the tolerance of agricultural economy risk. corresponding to a certain extent, it can also avoid the occurrence of credit risk of farmers. throughout the country, the number of rural credit cooperatives in china is very large, dealing with credit risk is bound to need hardware and software and financial support. from the government point of view, national and regional governments should come up with supporting policies to provide big subsidy efforts for rural financial institutions and help to enhance the stability of rural credit and financial environment. basu p (2005) [29] conducts a survey using 6000 rural households in india and conducts an empirical analysis using the shg bank linkage model. based on international experience, learning from micro-credit potential in india, he analyzes the government departments to pay attention to policy of this field, and proposes to expand the scale, and puts forward the way to provide funds for farmers. based on empirical research on micro-credit in india, s jha and ks bawa (2007) [30] argues that the establishment of institutions, the diversification of knowledge and skills, the input of traditional and scientific knowledge and the sustainable protection of enterprises in rural areas can improve economic and environmental benefits. from the questionnaire and survey data, based on model analysis, some domestic scholars propose credit risk settlement recommendations. according to the results of the model, find an obvious correlation between interest rate and default possibility of farmers, the significance level of the loans term is also very high. he believes that rural credit cooperatives should be appropriate to reduce the interest rate in providing loans to farmers and should extend the loan period. liu liyan (2013) [32] conducts an empirical study on rural credit cooperatives in heilongjiang province and 143 rural households. it is believed that they should focus on their per capita net income and asset value when providing loans to rural households, and takes these asset endowments as an important indicator when considering repayment possibility of farmers. at the same time, it is necessary to consider the age and the cultural level of the farmers who apply for loans, and rural credit cooperatives should control the amount of loans for older peasants whose cultural level is too low. 6. conclusions and discussions based on the above research, the domestic and foreign scholars mainly analyze the risk factors and risk evaluation, including the qualitative analysis and the quantitative analysis. qualitative analysis is putting forward the credit risk coping strategies from the macromanagement and institutional measures, quantitative analysis is studying credit assessment and impact factors of the credit risk from the construction of the model. at present, domestic scholars on credit risk research has taken shape. in the early stage, the theoretical analysis of credit risk of peasant households is more, the representative figures are du xiaoshan and so on. in the latter stage, the research on credit risk of rural households is mainly transferred to quantitative analysis. there are many types of measurement models, but there are few effective models for rural financial environment. and it is not enough in the index design, because the determination of index weight mainly comes from the subjective evaluation, such as qualitative score or expert judgment. the index design on the risk factor is still the difficulty of the current research. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 25 acknowledgements this research is supported by 2016 year yunnan philosophy and social science planning project (yb2016016), scientific research foundation of yunnan provincial education department (2016zzx080) and doctor research foundation of yunnan normal university. references 1. huang j. on the policy effect and risk prevention of farm household microcredit loan. china rural finance, 2003 (12): 24-27. 2. dai l.x., li z.h. and yang f.w.research on the risks and evaluation of farm household microcredit loan. science and technology, 2008,8 (3): 20-24. 3. zhu x.l.study on risk management of rural microcredit loan. economic survey, 2010 (12): 80-83. 4. tan z.m. and liang j.operational risks of rural microcredit companies and their precautions based on the survey and analysis of jiangsu province. rural economy, 2011 (10): 51-53. 5. duan w.j. an empirical analysis on the risk and benefit of farm household microcredit loan. wuhan finance, 2009 (1): 48-49. 6. zeng z.m. and yue y.d.mechanism innovation of microcredit credit risk management based on game theory. business economics and management, 2010 (8): 60-66. 7. shi g.h., pang y. and zhang z.y. a stochastic regulatory game analysis of moral risks of farmers' microcredit loans. china rural survey, 2014 (6): 49-58. 8. yang x.x., xue r.x. and ye h. problems and countermeasures of the sustainable development of rural credit cooperatives microfinance. rural economy, 2007 (5): 57-61. 9. huang b.y. on the sustainable development of microfinance in rural credit cooperatives. economic restructuring, 2008, 10 (2): 101-104. 10. chavan p,ramakumar r. micro-credit andrural poverty: an analysis of empirical evidence. economic and political weekly, 2002: 955-965. 11. cipollini a,fiordelisi f. economic value, competition and financial distress in the european bankingsystem. journal of banking & finance, 2012, 36 12. jansson k h, huisman c j, lagerkvist c j, et al. agricultural credit market institutions: a comparison of selected european countries. 2013. 13. li z.b., gao j.and cui w.j. study on credit risk evaluation of rural credit cooperatives. journal of beijing electronics science and technology institute, 2006, 14 (1): 69-74. 14. wei l. study on themicro credit risk evaluation system of peasant household. research on finance and economics, 2013 (8): 125-128. 15. malik m, thomas l c.transition matrix modelsof consumer credit ratings. international journal offorecasting, 2012, 28（1）: 261-272. 16. awunyo-vitor d, mahama al-hassan r, bruce sarpong d, et al. agricultural credit rationing in ghana: what do formal lenders look for?.agricultural finance review, 2014, 74(3): 364-378. 17. hu y., xu h.l.and wang x.study on the credit rating of micro-credit loans for farmers.financial theory and practice, 2007, 28 (1): 30-33. 18. nie y. study on the performance evaluation model of rural microfinance based on multi objective decision making. journal of huazhong agricultural university (social science edition), 2009 (1): 6-9. 19. wang j.and zhu m.h. study on credit risk evaluation of microfinance for farmers based on catastrophe theory.financial theory and practice, 2011 (7): 20-23. 20. yao s.q.and qiang j.h. research on credit risk assessment of farmers' microfinance based on bp neural network. journal of northwest sci-tech university of agriculture and forestry (social science edition), 2012, 12 (2): 78-83. 21. hu h.q., zhang l.and zhang d.h. credit risk assessment model of smes from the perspective of supply chain finance a comparative study based on svm and bp neural network. management review, 2012,24 (11): 70-80 22. peng m.d.and cheng y.q. finance credit evaluation model of peasant household micro credit based on comprehensive judgment ability. review of management, 2015, 27 (6): 42-57. 23. xiong x.p.and yi f.h.farmer credit risk of rural households: performance, causes and countermeasures. east china economic management, 2005, 19 (11): 49-52. 24. han x.p. andjin y. construction of rural credit guarantee system in china. journal of agricultural economics, 2014, 35 (3): 37-43. 25. ding y.h. rural microfinance risk and its prevention. rural economy, 2006 (12): 67-69. 26. yang h.l.and xin r.foundation of micro-credit risk early-warning system in peasant households. business research, 2009 (6): 164-165. 27. ma x.n.research on the micro-credit risk control strategy of rural credit cooperatives. agricultural economy, 2013 (7): 105-106. 28. tong y.b.and qi w.n. innovative development and sustainability of microfinance in rural credit cooperatives: a case study of hainan. agricultural economics, 2014, 35 (5): 75-80. 29. basu p, srivastava p. exploring possibilities: microfinance and rural credit access for the poor in india. economic and political weekly, 2005: 1747-1756. 30. s jha, k s bawa. the economic and environmental outcomes of microfinance projects: an indian case study.environment, development and sustainability,2007,(9): 229-239. 31. li z.b.and gao j. discriminant analysis of credit risk of rural credit cooperatives. journal of shandong institute of business and technology, 2007, 21 (1): 76-82. 32. liu l.y.and li x.g. study on factors influencing rural household credit risk of rural credit cooperatives: aaasestudy of heilongjiang province. anhui agricultural sciences, 2013, 41(13): 6023-6025. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 21–26 ___________________________________________________________________________________________________________ 26 1. introduction 2. the classification and characteristics of credit risk 2.1. the types of credit risks in rural credit cooperatives 2.2. characteristics of credit risk in rural credit cooperatives 3. problems and factors for credit risk 3.1. information asymmetry between borrowers and lenders 3.2. imperfect pre examination mechanism of rural credit cooperatives 3.3. lower risk management level of rural credit cooperatives 3.4. other factors 4. model and evaluation of credit risk 5. countermeasures and suggestions of credit risk 5.1. establish a guarantee system 5.2. improve the internal management mechanism of rural credit cooperatives 5.3. other factors 6. conclusions and discussions acknowledgements references atlantis press journal style introduction for volume 7, issue 2 in the present issue of journal of risk analysis and crisis response(jracr) , volume 7, issue 2 (2017), there are 4 contributions written in english and 1 contribution in chinese with english abstracts. the papers can be divided into four topics: political risk, hazard assessment, risk analysis and management, and eco-environmental quality evaluation. there is one paper in political risk,“the risk of power imbalance in project delivery: a study of large victorian public infratrucure projects”by zarei et al.. in this paper, seven large infrastructure projects in the state of victoria, australia are examined through a workshop involving key stakeholders who had played active roles in these projects. the findings revealed that power asymmetry between central and delivery agencies exist and would lead to optimism bias, which in turn creates uncertainty and risk of over-promising in the business case. power asymmetry exist in large infrastructure projects because the central agencies usually only have the responsibility but not the skill set needed to measure the robustness of the business case. these types of political risks are difficult to quantify and even detect. this paper recommends a few managerial strategies that have referential values and/or can be used to mitigate and circumvent this risk. there is one paper in hazard assessment, “an algal specific growth rate relation model to assess dynamical brown tide disasters” by wen et al., in which the authors consider the coupling effect of nutrient and temperature on algal growth. they propose an algal specific growth rate relation model to assess dynamical brown tide disasters caused by aureococcus anophagefferens. to verify the model, we processed data from coastal waters along qinhuangdao, china, during january 2014 to december 2014, where gis spatial analysis tools were used to process satellite data on nutrient, chlorophyll a, and temperature, and then calculated the hazard degree of aureococcus anophagefferens brown tide. the results show that (1) the suggested model is in accordance with shelford’s law of tolerance, which states that each species thrives best at a particular value of an environmental variable (the optimum) and cannot survive when the value is either too low or too high; (2) the coastal water along qinhuangdao was more likely prone to brown tide blooms from may to july than the other months of the year in 2014; (3) the hazard degree of aureococcus anophagefferens in areas near the shore was higher than that in areas far from the shore in the same months. these findings would provide a technical basis to effective response to brown tide disasters in qinhuangdao coastal waters. there are two papers in risk analysis and management. in the first paper “risk combinations in the risk analysis and management” by milík tichý, the author thinks of that the analogies between the risks carried by entities and the physical loads affecting building and engineering structures can be applied in the studies of risks forming groups arranged according to specified definitions. four classes of relations among risks belonging to a particular hazard are to be distinguished: existential, sequential, physical, and statistical. the type and number of possible groupings of combined risks, called "risk combinations", depends upon the type of existential and sequential relations among the risks affecting the respective entity. the results of the analysis from this study can be used in decision makings on which risks are to be considered in the decisions. the combined-risk problem relates to various fields of risk engineering and management as, such as seismic engineering, medicine, agriculture, economic engineering. the second paper “a study on marine vessels’ path optimization under typhoon scenarios” proposes a hybrid algorithm integrating genetic algorithm (ga) with receding horizon control (rhc), which considers real-time typhoon data, marine environment data including ocean currents and reefs etc., as well as safety operation requirements for marine voyage. then, simulation experiments are conducted to test the proposed method. the experimental results indicate that the new algorithm is effective and efficient to optimize voyage paths for marine vessels under typhoon scenarios, i.e., when compared with global path optimization, the new algorithm can improve the cost-efficiency of voyage path with safety guarantee. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 51–52 ___________________________________________________________________________________________________________ 51 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). there is one paper in eco-environmental quality evaluation,“evaluation on eco-environmental quality of dongsheng district ” by wu et al., in which authors selectes biological abundance index, vegetation coverage index, land desertification index, water density index and pollution load index to evaluate the quality of ecological environment. the evaluation result shows that the ecological environment quality of dongsheng district is "normal", the degree of vegetation coverage is moderate, so as to the general level of biological diversity, and it is suitable for human to live, but sometimes exist restrictive factors limit human survival. the ei value increased from 52.06 of 2005 to 53.11 in 2014,and land ecological environment quality of dongsheng district has rising trend from the overall view, but the amplitude variations in the ecological environment condition has no obvious change, land ecological environment is good, the land ecosystem structure is complete,and its function is basically perfect. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang and prof. junxiang zhang, the respective publication chair and deputy chair of the society for risk analysis – china, for that they devoted their time to overseeing the reviews editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 51–52 ___________________________________________________________________________________________________________ 52 atlantis press journal style assessment of earthquake prevention and disaster reduction capability of county-level administrative units in gansu province* jin chen†, wen li, wenkai chen‡, suping zhang lanzhou earthquake research institute of china earthquake administration, lanzhou 730000, china abstract firstly, this paper establishes the assessment indicator system for the earthquake prevention and disaster reduction capability of county-level administrative units in gansu, and formulates three criteria, ten indicators and fifteen variables centered on one target. then, the analytic hierarchy process is applied to determine the indicator weights, and the environmental supporting capability, infrastructure supporting capability and resource supporting capability of different counties in gansu in the handling of earthquake disasters are calculated to obtain the assessment figure for the earthquake prevention and disaster reduction capability of county-level administrative units in gansu. the research results indicate that anning district of lanzhou has the strongest earthquake prevention and disaster reduction capability while zhouqu county possesses the weakest earthquake prevention and disaster reduction capability; the overall earthquake prevention and disaster reduction capability of northwestern and northeastern regions in gansu is higher, in which the earthquake prevention and disaster reduction capability of aksay-subei district, jinchuan district, xifeng district, linxia city district and lanzhou city district is particularly prominent; the earthquake prevention and disaster reduction capability of sunan-tianzhu district and gannan-longnan district is relatively weaker; the geographical environment has a great impact on the earthquake prevention and disaster reduction capability, which is demonstrated as that the earthquake prevention and disaster reduction capability of regions with violent tectonic activities and dramatic topographic inequality tend to be weak; the whole province should improve the construction of infrastructure and fundamental resources in order to compensate for the insufficiency in environmental supporting capability. in addition, the expert scoring method is adopted in the determination of indicator weights, which is subject to high subjectivity; considering that some data are missing and the influential indicators are not considered comprehensively, the assessment indicator system requires further improvement. keywords: gansu, county-level administrative units, earthquake prevention and disaster reduction capability, analytic hierarchy process * fund project: co-funded by special project of basic scientific research of earthquake forecasting research institute of china earthquake administration (2015ieslz06), gansu provincial support program of science and technology (1504fkca065) and national key r&d program of china (no. 2017yfb0504104). †author: jin chen (1993-), male (han), born in shanxi, master’s degree, mainly research in application research of remote sensing and gis technology. email: 37329554@qq.com. ‡corresponding author: wenkai chen (1983-), male (han), associate researcher. e-mail: cwk2000@yeah.net . journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 214 received 10 september 2017 accepted 3 november 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). mailto:37329554@qq.com 1. introduction as one of the few chinese provinces with frequency earthquake activities, gansu tops the ranking of regional earthquake activity level. on its history, many destructive earthquakes have occurred, which led to severe personal casualties and economic loss in gansu (pei huijuan, 2015). inside gansu, there are qilian mountains earthquake belt, tianshui earthquake belt and alkin earthquake belt, where the multiple destructive earthquakes are concentrated, including ms8.5 haiyuan earthquake in 1920, ms8.0 gulang earthquake in 1927 and ms7.25 shandan earthquake in 1954, etc. in recent years, the occurrence frequency of earthquakes in gansu and its surrounding regions has increased obviously (including ms8.0 wenchuan earthquake in 2008, ms6.6 minxian-zhangxian earthquake in 2013, ms6.4 menyuan earthquake in 2015 and ms7.0 jiuzhaigou earthquake in 2017). in addition, with the quantum leaps in regional economy, the impacts of earthquake disasters on gansu society and economy will be further aggravated (nie gaozhong, 2002). earthquake prevention and disaster reduction capability is the ability of a region to reduce earthquake damage and loss in the face of an earthquake disaster. the higher the capability of earthquake prevention and disaster reduction is, the less damage the earthquake will cause to the local social and economic development. in the 1990s, all countries in the world started to carry out scientific research on earthquake prevention and disaster reduction one after another, and established various indicator systems for earthquake prevention and disaster reduction capability assessment (liu xiaojing, 2012; guo yan, 2013; deng yan, 2013; liu li, 2008; li zhi, 2011; davidson rachel, 1997; kunihiro a, 2000). the most representative overseas result is the earthquake disaster risk indicator edri (davidson rachel, 1997) in while the most representative domestic result is the system put forward by academician xie lili to assess the capacity of urban earthquake prevention and disaster reduction based on the triple criteria (liu li, 2008). targeted at gansu province, this paper establishes an assessment indicator system for earthquake prevention and disaster reduction capability based on the county-level administrative unit and to assess the earthquake prevention and disaster reduction capability of all districts and counties in gansu province so as to correctly understand the distribution characteristics of earthquake prevention and disaster reduction capability in all districts and counties of gansu province and to provide scientific basis for the realization of the scientific configuration of earthquake prevention and disaster reduction resources. 2. data and method 2.1. data source and processing the accuracy of the assessment indicators directly affects the reliability of the assessment result. therefore, this paper uses the statistical data released by the state authority to conduct the calculation and analysis. the data of population, economy, health care, financial savings, traffic and communications are from the statistical yearbook of gansu province in 2016 (the deadline for statistics is december 2015) and the statistical yearbook of china county (the volume of cities and counties) 2016; the topographic data are from japan asem satellite dem data (spatial resolution of 30m); administrative divisions and active fault data are from gansu province earthquake emergency basic database. the spatialization of each indicator in this paper is done in the arcgis software platform. the coordinate system adopts wgs-1984. 2.2. analysis method please make sure that no page numbers appear in your paper. sections, sub-sections and sub-subsections are numbered in arabic. use double spacing before all section headings and single spacing after section headings. flush left all paragraphs that follow after section headings. the analytic method adopted in this paper is ahp (analytic hierarchy process), which is a multi-criteria decision-making method proposed by saaty (saaty tl, 1980) et al. in the early 1970s. in the decision-making process, the factors are classified to form a hierarchical analysis model. after that, the factors are mutually compared on different hierarchies, and then the discriminant matrix is established, and then the weight on the importance of each factor is finally determined to provide a comparative quantitative basis for the development analysis and prediction (satty tl, 1994). the method used in the processing of statistical data in journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 215 this paper is linear dimensionless method to eliminate the impacts of the indicator dimension on the result. the first step of ahp is to set up the hierarchical model. based on profound analysis of practical problems, the relevant factors are decomposed into several hierarchies from top to bottom according to different attributes. the factors on the same hierarchy are subordinated to or have an impact on the factors on the above hierarchy, and dominate or are under the influence of the factors on the lower hierarchy. the top hierarchy is the target hierarchy, usually with only one factor, and the bottom hierarchy is usually a factor hierarchy. in the middle, there can be one or several hierarchies, usually the criteria hierarchy and indicator hierarchy. the second step is to construct the comparison matrix. starting from the second hierarchy of the hierarchical model, pairwise comparison method and 19 comparison scale are used to construct the comparison matrix for the factors on the same hierarchy subordinate to factors on the above hierarchy until the bottom. for n factors, we can get pair comparison judgment matrix  ij n nc c  , whose general form is expressed as: bk c1 c2 … cn c1 c11 c12 … c1n c2 c21 c22 … c2n … … … … … cn cn1 cn2 … cnn matrix c possesses following properties: (1) 0 ij c  ; (2) 1 / ( ) ij ji c c i j  ; (3) 1( 1,2,..., ) ii c i n  . 1-9 comparison scale method is illustrated in following table 1: table 1 definition of comparison scale no. importance level value of cij 1 i and j are equally important 1 2 i is slightly more important than j 3 3 i is obviously more important than j 5 4 i is highly more important than j 7 5 i is extremely more important than j 9 6 i is slightly less important than j 1/3 7 i is obviously less important than j 1/5 8 i is highly less important than j 1/7 9 i is extremely less important than j 1/9 other relative importance levels can be assigned as 2, 4, 6, 8, 10, 1/2, 1/4, 1/6, 1/8, 1/10, and the important level can be determined in comparison with the 1-9 scale. the third step is to calculate the weight vector and does the consistency test. for each paired comparison matrix, the largest eigenvalue and the corresponding eigenvector are calculated, and the consistency test is performed by using consistency indicator, random consistency and consistency ratio. the calculation process of the largest matrix eigenvalue max is as follows: (1) calculate the product of each row element of the judgment matrix c. 1 , 1,2,..., n i ij i m c i n    (2) calculate the nth power root of i m n i i w m (3) if the normalized 1 i i n j i w w w    then wi is the sought eigenvector. (4) calculate the largest eigenvalue max 1 n i i i cw nw     cwi denotes the negative average value of the other eigenvalues except for the largest eigenvalue of the judgment matrix after the ith component of the vector cw has been calculated to serve as a measure on the deviation of the judgment matrix from the consistency indicator, which means: max 1 n ci n     the larger the ci is, the greater the deviation of the judgment matrix from the complete consistency will be. a smaller ci value indicate a better consistency of the judgment matrix; when the matrix has satisfactory consistency, max is slightly larger than n and the other eigenvalues are close to 0. hereunder, a measure of satisfactory consistency is proposed. when the order of the matrix is greater than 2, the ratio of consistency indicator ci of the judgment matrix to the same-order mean coincidence index ri is called the random consistency ratio and is denoted as cr. when journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 216 0.10 ci cr ri   it is assumed that the judgment matrix has satisfactory consistency. otherwise, the judgment matrix needs to be adjusted to a satisfactory consistency. the value of ri changes with the change of matrix dimension n, as shown in table 2: table 2 values of random consistency indicator n 1 2 3 4 5 6 7 8 9 ri 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 3. assessment and earthquake prevention and disaster reduction capability 3.1. assessment process in this paper, the earthquake prevention and disaster reduction capability assessment is conducted on countylevel administrative unit in gansu, and the overall assessment process is shown in figure 1. the target hierarchy is the calculated earthquake prevention and disaster reduction capability of the counties (districts). 3.2. assessment indicator system for earthquake prevention and disaster reduction capability this paper selects the assessment indicators and factors from three aspects of environmental supporting capability, infrastructure supporting capability and resource supporting capability, and establishes the assessment indicator system for earthquake prevention and disaster reduction capability of county-level administrative units in gansu province (table 3). environmental supporting capability includes the distribution of topographic relief and active faults. it is more difficult to carry out disaster relief and rescue work in places where the topography fluctuates greatly. there is also a relatively higher risk of earthquakes in areas with more active faults. infrastructure supporting capability includes building quality, traffic conditions, medical conditions, communication conditions and shelter premises. the finally-determined factor hierarchy includes topographic relief, active fault distribution, building seismic capacity, average resident gdp, road density, medical beds, fixed telephone density, satellite telephone density, average personal refuge, number of young males, emergency rescue forces, fiscal revenue density, average resident savings material reserves, population density above high school and others, making 15 factors in total. among them, the building quality depends first and foremost on the overall structure of the local buildings. the data of the building structures in all districts and counties in gansu province are used. then, the seismic damage indicator of buildings in all districts and counties is calculated . figure 1 assessment process journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 217 based on the earthquake damage indicator of each structural building to represent the seismic capacity of buildings (wang xianghao, 2008). in addition, the building quality is closely linked with the local economy and people's living standard. the higher the local economy and people's living standard are, the higher the overall quality of living buildings is. resource supporting capability includes labor resources, property resources and cognitive level. after the earthquake, more human and financial resources that can be mobilized for rescue provides better disaster relief effect. besides, the knowledge level of local residents also exerts some influence on the implementation of earthquake prevention and disaster reduction policies. the weights of assessment indicators are determined by ahp. according to the relative importance of each indicator, the factors on different hierarchies are mutually compared to establish the judgment matrices, and the consistency test of each judgment matrix is carried out. combined with the scoring of experts in the field of earthquake, there are 8 final judgment matrices, 1 on the criteria hierarchy, 3 on the indicator hierarchy and 4 on the factor hierarchy. the judgement matrix of environmental supporting capability, infrastructure supporting capability and resource supporting capability is determined as: 1 1 2 4 1 / 2 1 4 1 / 4 1 / 4 1 a            through calculation, the weight coefficient of a1 is w1=（0.547，0.345，0.108） t . the judgment matrix of topographic relief and active fault is determined as: 2 1 1 / 2 2 1 a        through calculation, the weight coefficient of a2 is w2=（0.333，0.667） t . the judgment matrix of building quality, traffic conditions, medical conditions, communication conditions and shelter premises is determined as: table 3 the assessment indicator system for earthquake prevention and disaster reduction capability of county-level administrative units in gansu province. target hierarchy criteria hierarchy indicator hierarchy factory hierarchy earthquake prevention and disaster reduction capability environmental supporting capability (0.547) topographic relief (0.182) topographic relief (0.182) active fault (0.365) active fault distribution (0.365) infrastructure supporting capability (0.345) building quality (0.168) building seismic capacity (0.112) average resident gdp (0.056) traffic conditions (0.078) road density (/km2) (0.078) medical conditions (0.049) medical beds (/10000 persons) (0.049) communication conditions (0.03) fixed telephone density (/km2) (0.01) satellite telephone density (/km2) (0.02) shelter premises (0.02) average personal refuge (/10000 persons) (0.02) resource supporting capability (0.108) labor resources (0.035) number of young males (/km2) (0.007) emergency rescue forces (/km2) (0.028) property resources (0.045) fiscal revenue density (/km2) (0.013) average resident savings (0.008) material reserves (/10000 persons) (0.024) cognitive level (0.028) population density above high school (/km2) (0.028) journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 218 3 1 3 4 5 6 1 / 3 1 2 3 4 1 / 4 1 / 2 1 2 3 1 / 5 1 / 3 1 / 2 1 2 1 / 6 1 / 4 1 / 3 1 / 2 1 a                through calculation, the weight coefficient of a3 is w3=（0.487，0.228，0.142，0.087，0.056） t . the judgment matrix of labor resources, property resources and cognitive level is determined as: 4 1 1 1 1 1 2 1 1 / 2 1 a            through calculation, the weight coefficient of a4 is w4=（0.327，0.413，0.26） t . the judgment matrix of building seismic capacity and average resident gdp is determined as: 5 1 2 1 / 2 1 a        through calculation, the weight coefficient of a5 is w5=（0.667，0.333） t . the judgment matrix of fixed telephone density and satellite telephone density is determined as: 6 1 1 / 2 2 1 a        through calculation, the weight coefficient of a6 is w6=（0.333，0.667） t . the judgment matrix of number of young males and emergency rescue forces is determined as: 7 1 1 / 4 4 1 a        through calculation, the weight coefficient of a7 is w7=（0.2，0.8） t . the judgment matrix of fiscal revenue density, average resident savings and material reserves is determined as: 8 1 2 1 / 2 1 / 2 1 1 / 3 2 3 1 a            through calculation, the weight coefficient of a8 is w8=（0.297，0.163，0.540） t . it is tested that the random consistency ratios of all the judgment matrices are lower than 0.1, which means that the judgment matrices possess satisfactory consistency. as shown in table (from up to bottom), the calculated criteria weights on the criteria hierarchy are respectively (0.547, 0.345, 0.108), and the calculated indicator weights on the indicator hierarchy are respectively (0.182, 0.365, 0.112, 0.056, 0.078, 0.049, 0.01, 0.02, 0.02, 0.007, 0.028, 0.013, 0.008, 0.024, 0.028). the calculation formula of the final criteria hierarchy is given as: 1 n i i i c i    where 1 m j j j i f    the calculation formula of the target hierarchy is given as: 1 h k k k t c    t represents the target hierarchy, c represents the criteria hierarchy, i represents the indicator hierarchy, f represents the factor hierarchy, δ represents the weight of each indicator hierarchy, σ represents the weight of each factor hierarchy, μ represents the weight of each criteria hierarchy, n represents the number of factors on the indicator hierarchy which is contained in the corresponding criteria hierarchy c, m is the number of factors on the factor hierarchy which is contained in the corresponding indicator hierarchy, and h is the number of factors on the criteria hierarchy which is contained in the corresponding target hierarchy. the final value the above-calculated criteria hierarchy c and target hierarchy t ranges from 0 to 1, which is because the dimensionless method is adopted for all the factors in the calculation process to eliminate the influence of different dimensions on the result. the final calculation results have also been normalized to facilitate the follow-up classification analysis. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 219 3.3. result and analysis (1) environmental supporting capability in combination of the topographic relief and active fault, the distribution figure of environmental supporting capability of counties and districts in gansu province is obtained. the capability is divided into five levels, respectively strong (0-0.09), relatively strong (0.09-0.21), medium (0.21-0.33), relatively weak (0.330.53) and weak (0.53-1). hereunder, the classification is done in the arcgis software platform, and the method used is natural breaks (jenks) (natural breakpoint classification, the same for below). the principle of classification is to put the similar individuals together and divide them into several classes. statistics can be measured by the variance, which refers to calculating the variance of each class, and then calculating the sum of these variances and the size of the variance to compare the classification quality. the classification with the minimum variance is the optimal method. natural breaks classification can well classify things alike together with obvious differences among classes but small differences within one class. it can be seen from figure 2 that the districts with strong environmental supporting capability are mainly located in the four regions of dunhuang-guazhou district, jinchang-minqin district, dingxi-pingliangqingyang district and lanzhou city district. the topographic relief in these districts is relatively small, with dunhuang-guazhou district, jinchang-minqin district and lanzhou city district relatively flat. the district with weak environmental supporting capability are mainly gannan-longnan district and sunantianzhu district. sunan-tianzhu district is located on the edge of the qilian mountains with a great topographic relief, while gannan-longnan district is located in the main active fault in the province, kunlun fault zone with strong tectonic activity. figure 2 the distribution of environmental supporting capability of various districts and counties in gansu province. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 220 (2) infrastructure supporting capability in combination of the building quality, traffic conditions, medical conditions and communication conditions, the distribution figure of infrastructure supporting capability of districts and counties in gansu province is obtained (figure 3). the capability is divided into five levels, respectively strong (0.51-1), relatively strong (0.32-0.51), medium (0.23-0.32), relatively weak (0.13-0.23) and weak (0-0.13). as shown in fig.3, the districts with strong infrastructure supporting capability mainly includes aksay-subei district, jiayuguan city, jinchang city district, baiyin city district and lanzhou city district. among them, the aksay-subei district is sparsely populated with rich mineral resources. its average gdp is in the forefront among the province, and it maintains a good overall building quality. population is densely distributed throughout central and eastern gansu province, which pulls down the average gdp. the infrastructure supporting capability cannot meet the demand of excessive population. the gannan-linxia-dingxi district has a weak overall infrastructure supporting capability. (3) resource supporting capability through the labor resources, property resources and cognitive level, the distribution figure of resource supporting capability of districts and counties in gansu province is obtained (figure 4). the capability is divided into five levels, respectively strong (0.66-1), relatively strong (0.25-0.66) medium (0.11-0.25), relatively weak (0.05-0.11) and weak (0-0.05). fig.4 shows that the overall resource supporting capability of gansu province is weak, with only relatively strong capability of anning district of lanzhou, linxia city district, subei county and qinzhou district. the appearance of this phenomenon is due to the backward economic development of gansu province, and the government revenue and the average resident savings are less. in addition, the provincial education is underdeveloped, resulting in a low cognitive level of the people. figure 3 the distribution of infrastructure supporting capability of various districts and counties in gansu province. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 221 (4) earthquake prevention and disaster reduction capability in combination of the environmental supporting capability, infrastructure supporting capability and resource supporting capability, the distribution figure of earthquake prevention and disaster reduction capability of various districts and counties in gansu province is finally calculated. the capability is divided into five levels, respectively strong (0.76-1), relatively strong (0.63-0.76), medium (0.50-0.63), relatively weak (0.260.50) and weak (0-0.26). as shown in fig.5, the earthquake prevention and disaster reduction capability of west section and mid-east section of hexi, longdong district and lanzhou in gansu province is relatively strong, where the earthquake prevention and disaster reduction capability of anning district of lanzhou is the strongest, followed by aksay district, jinchuan district, xifeng district and linxia city district. then, the earthquake prevention and disaster reduction capability of sunan-tianzhu district and gannanlongnan district is relatively weak, and zhouqu county has the weakest earthquake prevention and disaster reduction capability in all the counties and districts in gansu province. 4. conclusion in this paper, the assessment indicator system of earthquake prevention and disaster reduction capability of county-level administrative units in gansu province is established. the analytic hierarchy process is used to determine the weight of each indicator factor. the arcgis software platform is used to calculate the environmental supporting capability, infrastructure supporting capability and resource supporting capability of all districts and counties in gansu province. on this basis, the assessment figure of earthquake prevention and disaster reduction capability is obtained. the research results indicate that anning district of lanzhou has the strongest earthquake prevention and figure 4 the distribution of resource supporting capability of various districts and counties in gansu province. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 222 disaster reduction capability while zhouqu county possesses the weakest earthquake prevention and disaster reduction capability; the overall earthquake prevention and disaster reduction capability of northwestern and northeastern regions in gansu is higher, in which the earthquake prevention and disaster reduction capability of aksay-subei district, jinchuan district, xifeng district, linxia city district and lanzhou city district is particularly prominent; the earthquake prevention and disaster reduction capability of sunan-tianzhu district and gannan-longnan district is relatively weaker; the geographical environment has a great impact on the earthquake prevention and disaster reduction capability, which is demonstrated as that the earthquake prevention and disaster reduction capability of regions with violent tectonic activities and dramatic topographic inequality tend to be weak; the whole province should improve the construction of infrastructure and fundamental resources in order to compensate for the insufficiency in environmental supporting capability. in addition, the assessment method used in this paper provides a hierarchical thinking framework. by combining qualitative judgment with quantitative inference, this paper assesses the earthquake prevention and disaster reduction capability of county-level administrative units in gansu province and obtains their spatial distribution characteristics, providing certain theoretical reference. however, this assessment method is subject to obvious shortcomings in adopting expert scoring method during the determination of the judgment matrices and the factor weights, which involves subjective human factors, easily leading to controversy. furthermore, due to the lack of individual data and collection difficulties, the assessment indicator system established in this paper is not perfect enough. therefore, it is more convincing to further explore the independent factors that have an impact on earthquake figure 5 the distribution of earthquake mitigation capacity of various districts and counties in gansu province. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 223 prevention and disaster reduction capability and include them in the assessment indicator system. references pei huijuan, zhou zhonghong, sun yanping, et al, spatiotemporal distribution feature of earthquake disaster in gansu province, journal of natural disasters. 24(3) (2015) 67-75. nie gaozhong, gao jianguo, ma zongjin, et al, on the risk of earthquake disaster in china in the coming 10-15 years, journal of natural disasters. 11(1) (2002) 68-73. liu xiaojing, bo tao, guo yan, index system in evaluating comprehensive earthquake disaster mitigation capacity in china: taking tangshan city as an example, journal of natural disasters. 6 (2012) 043-49. guo yan, bo tao, liu xiaojing, assessment methodology of earthquake disaster prevention and reduction capacity and its application to wenchuan earthquake, journal of natural disasters. 22(5) (2013) 036-43. deng yan, su guiwu, wu huanjie, comprehensive assessment of the earthquake prevention and disaster reduction ability of china, seismology and geology. 35(3) (2013) 584-592. liu li, xie lili, application of analytic hierarchy process in assessment of urban earthquake prevention and disaster reduction ability, journal of natural disasters. 17(2) (2008) 48-52. li zhi, zhao xiaohui, qu le, a case study in casualty on the assessment for the ability of earthquake disaster reduction in regions, journal of disaster prevention and reduction. 27(3) (2011) 1-6. davidson rachel, haresh sban, an urban earthquake disaster risk index. report 121(the john a. blume earthquake engineering center, stanford, california: blume center, 1997). kunihiro a.,takahisa e.and toshio m, intercity comparison using evaluation techniques on vulnerability of earthquake disaster--a case study of ordinance-designated city, journal of japan society for natural disaster science. 18(4) (2000) 489-500. liu li, xie lili, study on factors influencing on ability of earthquake disaster reduction in a city, world earthquake engineering. 24(1) (2008) 88-92. saaty tl, the analytic hierarchy process. (new york mcgraw-hill. 1980). saaty tl, how to make a decision: the analytic hierarchy process, european journal of operational research. 48(1) (1994) 9-26. wang xianghao, characteristics of chinese village earthquake disasters and types of building structures, architecture and planning design. 2(2) (2008) 39-41. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 214–224 ___________________________________________________________________________________________________________ 224 1. introduction 2. data and method 2.1. data source and processing 2.2. analysis method 3. assessment and earthquake prevention and disaster reduction capability 3.1. assessment process 3.2. assessment indicator system for earthquake prevention and disaster reduction capability 3.3. result and analysis 4. conclusion references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <feff004b0069007600e1006c00f30020006d0069006e0151007300e9006701710020006e0079006f006d00640061006900200065006c0151006b00e90073007a00ed007401510020006e0079006f006d00740061007400e100730068006f007a0020006c006500670069006e006b00e1006200620020006d0065006700660065006c0065006c0151002000410064006f00620065002000500044004600200064006f006b0075006d0065006e00740075006d006f006b0061007400200065007a0065006b006b0065006c0020006100200062006500e1006c006c00ed007400e10073006f006b006b0061006c0020006b00e90073007a00ed0074006800650074002e0020002000410020006c00e90074007200650068006f007a006f00740074002000500044004600200064006f006b0075006d0065006e00740075006d006f006b00200061007a0020004100630072006f006200610074002000e9007300200061007a002000410064006f00620065002000520065006100640065007200200035002e0030002c0020007600610067007900200061007a002000610074007400f3006c0020006b00e9007301510062006200690020007600650072007a006900f3006b006b0061006c0020006e00790069007400680061007400f3006b0020006d00650067002e> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <feff004200720075006b00200064006900730073006500200069006e006e007300740069006c006c0069006e00670065006e0065002000740069006c002000e50020006f0070007000720065007400740065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e00740065007200200073006f006d00200065007200200062006500730074002000650067006e0065007400200066006f00720020006600f80072007400720079006b006b0073007500740073006b00720069006600740020006100760020006800f800790020006b00760061006c0069007400650074002e0020005000440046002d0064006f006b0075006d0065006e00740065006e00650020006b0061006e002000e50070006e00650073002000690020004100630072006f00620061007400200065006c006c00650072002000410064006f00620065002000520065006100640065007200200035002e003000200065006c006c00650072002000730065006e006500720065002e> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <feff04120438043a043e0440043804410442043e043204430439044204350020044604560020043f043004400430043c043504420440043800200434043b044f0020044104420432043e04400435043d043d044f00200434043e043a0443043c0435043d044204560432002000410064006f006200650020005000440046002c0020044f043a04560020043d04300439043a04400430044904350020043f045604340445043e0434044f0442044c00200434043b044f0020043204380441043e043a043e044f043a04560441043d043e0433043e0020043f0435044004350434043404400443043a043e0432043e0433043e0020043404400443043a0443002e00200020042104420432043e04400435043d045600200434043e043a0443043c0435043d0442043800200050004400460020043c043e0436043d04300020043204560434043a0440043804420438002004430020004100630072006f006200610074002004420430002000410064006f00620065002000520065006100640065007200200035002e0030002004300431043e0020043f04560437043d04560448043e04570020043204350440044104560457002e> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice bp’s reputation repair strategies during the gulf oil spill received 9 september 2014 accepted 15 january 2015 bp’s reputation repair strategies during the gulf oil spill kristen alley swain meek school of journalism and new media, university of mississippi, box 1848, university, ms 38677, usa lindsay a. jordan profiles incorporated, 3000 chestnut avenue, suite 201, baltimore, md 21211, usa e-mail: lajordan13@gmail.com abstract on april 20, 2010, british petroleum’s deepwater horizon drilling rig in the gulf of mexico exploded, creating the largest oil spill in u.s. history. analysis of 1,161 bp tweets during the crisis response reflected unexpected reputation repair strategies and responsibility attribution. situational crisis communication theory suggests that after an accident, pr messages typically reflect low responsibility attribution. although the official investigation initially did not suggest a preventable crisis, 90% of bp’s tweets reflected high responsibility. keywords: reputation repair, responsibility attribution, oil spill, situational crisis communication theory 1. introduction on april 20, 2010, british petroleum’s deepwater horizon drilling rig located in the gulf of mexico ignited and exploded, leaving 11 members of its 126member crew missing and spewing thousands of gallons of oil into the gulf. transocean, the world’s largest offshore drilling contractor, had been drilling an exploration well using the deepwater horizon as a contractor to bp. transocean’s emergency and family response team, bp, and the u.s. coast guard immediately began searching for the missing people, but suspended the search after covering 5,000 square miles. the 11 workers were pronounced dead, and their bodies were never recovered. on april 22, bp ceo and british native tony hayward said, “we are determined to do everything in our power to contain the oil spill and resolve the situation as rapidly, safely, and effectively as possible.” the macondo well, from which the deepwater horizon was drilling for oil, would not be declared “dead” for nearly five months. when the deepwater horizon rig sank, bp executed a major oil spill response. initial response efforts included the implementation of a small fleet of response vessels, relief well planning, skimming of oily surface water, a protective boom to prevent oil from reaching the shoreline, and chemical dispersants to break up the oil, in an attempt to keep it from destroying marshes, mangroves, and beaches. efforts to seal the well began as early as april 25, but many were unsuccessful. in early may, the drilling of a relief well and a back-up relief well began. a riser insertion tube tool (ritt), which attempts to bring oil flow to the surface by inserting a tube into the broken end of the deepwater horizon’s riser, and a lower marine riser package (lmrp) containment cap were also used to collect oil and pump it to the surface. a sealing cap was put in place on july 5, and oil no longer flowed freely from the well. by journal of risk analysis and crisis response, vol. 5, no. 1 (april 2015), 2-15 published by atlantis press copyright: the authors 2 aug. 4, the u.s. government reported that three-fourths of the 4.9 million barrels of spilled oil had been evaporated, burned, skimmed, recovered, or dispersed. on sept. 16, the relief well drilled by development driller iii successfully intercepted the annulus of the macondo well. the void in the leaking valve on the deepwater horizon was sealed. finally, on sept. 19, the u.s. federal government declared the well “effectively dead.” monetary reimbursement efforts began shortly after the explosion with the initiation of a claims process, the opening of several claims offices, and the activation of a toll-free call center. bp’s claims process followed the “responsible party” guidelines of the oil pollution act of 1990, as the u.s. coast guard designated bp as a responsible party. from the time of the explosion to that of the interception by the relief well, bp paid out billions of dollars for statewide oil spill contingency plans, protection plans, research of the oil spill’s effect on the environment and public health, tourism promotion for affected states, restoration of wildlife habitats, unemployed rig workers, the construction of barrier islands, personal claims, etc. bp claims to have spent more than $13 billion on the cleanup, $500 million on scientific studies, and $280 million on wildlife rescue and rehabilitation. 2. background while bp was waging war against the oil in the gulf, a war waged against its reputation. not only did bp have to fix the largest oil spill in the history of the u.s., it also had to fix its damaged reputation. on the image front, bp had much to defend. bp spent heavily to position itself as an environmentally friendly company, redesigning its logo into a green-and-yellow sunburst and advertising its $4-billion-alternative-energy push to move beyond petroleum (m. debord, 2010). in bp’s post-crisis phase of crisis communications, many experts debated the effectiveness of bp’s pr campaign that accompanied the gulf oil spill. after one year, and a pr bill of more than $90 million, bp’s reputation was still in question. bp’s biggest pr mistake was lying to the public and regulators about its level of unpreparedness in combating a spill of such catastrophic proportions (j. vidal, 2011). some argued that bp’s pr campaign will serve as a “how not to” case study for future crises (r. blake, 2010). many of bp’s so-called oil spill related pr blunders can be traced back to tony hayward, the company’s then chief executive officer (associated press, 2010). although traditional crisis public relations campaigns suggest appointing one person, usually a company’s ceo, as the general spokesperson, british-born hayward may not have been bp’s best choice for an american tragedy. some experts have suggested that bp’s biggest mistake was its decision to have hayward serve as the most prominent spokesperson on the oil spill (e. reguly, 2010), as his lack of experience with u.s. culture was highlighted. although hayward got points for making his way to the gulf shortly after the explosion, people wanted to see him making an effort to fix the problem. daniel keeney, president of a dallas-based pr firm said, “you want to get him right in the thick of things, even if he looks somewhat uncomfortable doing it”(e. mcclam & h. weber, 2010). on june 1, 2010, national incident commander admiral thad allen was announced as a spokesperson separate from bp. the public was excited to learn of allen’s new position. “allen was the reassuring spokesperson, the steady hand that the public needed to see. he consistently refused to put numbers on the flow rate or predict the final plugging of the macondo” (d. lenci & j. mullane, 2010). however, having allen serve as a spokesperson further confused bp’s public relations tactics since allen was represented an entity separate from bp. one basic rule of pr is to avoid finger pointing, as it can diminish a company’s level of trustworthiness. unfortunately, that is exactly what hayward did during the immediate days following the explosion. he declared that the incident was “not our accident” (a. brownsell, 2010); “what the hell did we do to deserve this?”; and “the gulf of mexico is a very big ocean. the amount of volume of oil and dispersant we are putting into it is tiny in relation to the total water volume” (b. snyder, 2010). fraser seitel, a communication consultant who writes for jack o’dwyer’s pr newsletter, claims that bp should not have made early predictions about the amount of oil involved in the spill, as it weakened the company’s credibility (a. caruba, 2010). u.s. coast guard rear admiral and deputy on-scene commander mary landry initially said no oil was leaking from the macondo well. landry’s estimates published by atlantis press copyright: the authors 3 on leaking oil during the initial days of the spill rapidly increased from seepage, to 1,000 barrels per day, to 5,000 barrels per day, to 20,000 barrels per day, and finally to 100,000 barrels per day (d. lenci, denise & j. mullane, 2010). 2.1 use of social media as a crisis communication tool along with executing elements of a traditional pr campaign (i.e. press releases, press conferences, etc.), bp also launched an extensive social media campaign. however, the success of this social media campaign continues to be disputed by social media experts (t. bergin, 2010). prior to the oil spill in the gulf, bp did have a facebook page, a twitter feed, and a linkedin account, but none were updated regularly. in the wake of the oil spill, a full week passed before bp initiated any sort of social media response. for example, @bp_america, bp’s twitter account, did not tweet about the oil spill until april 27, a full seven days after the spill. in comparing the communications tactics pertaining to the deepwater horizon spill with that of the 1989 exxon valdez spill, harlan loeb, director of crisis and issues management at edelman, a leading global pr firm, said that nowadays companies operate in a “24/7 risk environment” (a. symington, 2010). according to loeb, two key factors behind this change were the emergence of social media and nongovernmental organizations’ increase in influence. loeb also argues that social media offers companies “an extraordinary opportunity to break through the static of information flow to tell your own story” (a. symington, 2010). with the help of oglivy public relations worldwide and purple strategies, bp launched an aggressive social media campaign after the first few weeks following the explosion. the company added a “gulf of mexico response” page to its corporate website providing information on all aspects of the spill; began updating its facebook page and twitter feed on a daily and then hourly basis; and broadcasted videos on its youtube channel and uploaded pictures to its flickr account of response and recovery effort (b. morrissey, 2010). bp also disabled comments from users who posted negative comments (a. seitz-wald, 2010). 2.2 twitter bp had a twitter account prior to the deepwater horizon oil spill but failed to update the feed regularly. following the spill, bp repurposed its account, @bp_america, to serve as a communication hub for updates and breaking news pertaining to the spill. however, it took bp several days to make this happen. in fact, bp did not tweet about the spill until april 27, seven days after the explosion. this tweet read, “bp pledges full support for deepwater horizon probes.” in 2010, bp’s twitter profile read: “updates of bp’s ongoing response efforts are provided by our own social media team, as well as onthe-ground personnel working in affected gulf regions. from time to time, mike utsler, coo of bp’s gulf coast restoration, will be giving first-hand updates via this channel.” from its first oil spill tweet on april 27, bp’s tweets gradually increased to multiple tweets per day. as of september 2014, @bp_america had 95,000 followers. more than a year after the incident, @bp_america currently reads, “official account of bp america. stay current on our commitment to the gulf and our work toward secure, affordable energy while addressing climate change.” while bp was busy repurposing its twitter account, a spoof account, @bpglobalpr, was quickly gaining followers. as of september 2014, it had 132,000 followers, 28% more than the official bp account. bp public relation’s (@bpglobalpr) profile reads, “this page exists to get bp’s message and mission statement out into the twitterverse.” this twitter feed was quite detrimental to bp’s twitter efforts, publishing satirical tweets, such as, “yes, we disabled the alarms on the deepwater horizon. oh, like you’ve never hit the snooze button?” (b. hatton, 2010). miles nadal, the chief executive of advertising conglomerate mdc partners inc., said, “the brand detractor is more influential and more vehement than the than the brand evangelist” (h. shaw, 2010). this fact was quite unfortunate for bp’s social media campaign. 2.3 crisis communication theory a crisis is defined as “a major occurrence with a potentially negative outcome affecting the organization, company, or industry, as well as its publics, products, services or good name” (k. fearn-banks, 2002). a crisis can present itself in many forms, and it will published by atlantis press copyright: the authors 4 almost always disturb normal business activities. depending on the severity of the crisis, it may even threaten the entire existence of the company or organization. common types of crises include chemical spills or leaks, contamination, earthquakes, fires, hurricanes, product failures, and terrorism. in order to successfully manage and overcome a crisis, companies must have a specific crisis management plan. crisis management is the process of strategic planning for a crisis or negative turning point. it seeks to removes risk and uncertainty from the negative occurrence, so the organization can be in greater control of its own identity. crisis management is the dialogue between the organization and its public prior to, during, and after the negative occurrence. these communications tactics are specifically designed to minimize damage to the image of the organization (k. fearn-banks, 2002). there are three phases of crisis management: precrisis, crisis response, and post crisis. the pre-crisis phase focuses on crisis prevention and preparation. the crisis response phase focuses on successfully responding to an occurring crisis. the post-crisis phase focuses on preparing for future crises while fulfilling promises made during the actual crisis. pre-crisis best practices include having a continually updated crisis management plan, having a designated and properly trained crisis management team, implementing mock crisis exercises to test the reliability of the plan, and pre-drafting messages to combat specific crises. post-crisis best practices include delivering promised information to stakeholders, continually updating stakeholders on the progress of the crisis recovery efforts, and closely reviewing the crisis management actions to see if they should be integrated into future crisis management plans. when a crisis does occur, the crisis response consists of what management does and says after the crisis hits in terms of its initial crisis response and its reputation repair and behavioral intentions. initial crisis response best practices include giving a response to the crisis within the first hour, being accurate with facts, keeping the spokesperson(s) informed of key messages; focusing on public safety, successfully implementing the use of all available communication mediums, providing a human element be expressing concern and/or sympathy (t. coombs, 2007). during this phase, an organization’s management is required to communicate important information to affected and/or interested audiences, and the public relations practitioner is responsible for shaping these messages. most practitioners advise management to abide by three rules during the initial crisis response phase: be quick, be accurate, and be consistent. of these three rules, a quick response is the most vital. in today’s world of 24/7 digital media, people expect an immediate response, and this is exactly what an organization should give them. even if the organization does not have an answer, an immediate response could simply state that the organization is in the process of gathering important information. “an early response may not have much ‘new’ information, but the organization positions itself as a source and begins to present its side of the story.” (t. coombs, 2007, p.2). the following list breaks down the best practices of the initial crisis response: (i) release an initial response as quickly as possible, preferably within the first hour following the crisis. (ii) be as accurate as possible by checking all facts and quickly correcting misinformation. (iii) make sure that the spokesperson(s) is consistent by having the public relations practitioner keep him or her updated with all information pertaining to crisis events and key messages. (iv)designate public safety as the crisis communications plan’s top priority. (v)get key messages out by using all available communication channels. this include, but are not limited to, newspapers, radio, television, the organization’s website, and social media outlets (t. coombs, 2007). 2.4 evolution of crisis communications in the early hours of friday, march 24, 1989, one of the most infamous crises in american history began to unfold. the 987-foot exxon valdez oil tanker was en route off the coast of alaska to sound to long beach, california, when it struck the bligh reef and began spewing 11 million gallons of crude oil into the waters of prince william sound. the largest oil spill in u.s. history, it killed an estimated two million sea animals. people gradually learned of the massive spill over the weekend through broadcasts on cbs, nbc, abc, and local news networks. published by atlantis press copyright: the authors 5 compared with the 2010 deepwater horizon oil spill, which instantly made news around the world due to the thousands of websites and blogs, exxon had considerably more time to draft key response messages. bp had an hour, tops (k. fearn-banks, 2002). due to the lack of 24/7 digital media during the time of the valdez spill, exxon was not held as accountable for certain actions as bp would have been. frank iarossi, then-president of exxon shipping and exxon’s main representative of the valdez site, was left to initiate the crisis response, as exxon ceo lawrence rawl, an employee at exxon for 37 years, was missing in action. in contrast, bp ceo tony hayward was crucified by the public, even though he immediately made himself present at the spill site and available to the media and public. this highlights the extent of the evolution of crisis communications attributable to the internet, new media, and especially social media (k. fearn-banks, 2002; r. parekh & e. lee, 2010). 2.5 situational crisis communication theory since current study pertains to bp’s use of crisis communications during the crisis response phase, the situational crisis communication theory serves as a guiding theoretical framework. scct is an attribution theory-based approach to crisis (r. heath & d. miller, 2004). during the crisis response phase, a company or organization will attempt to protect its reputation by including specific response strategies within their communications tactics. the main objective of scct is to develop a working system of strategies that can effectually protect an organization’s reputation – or, at the very least, diminish the threat of having the organization’s image tarnished (r. heath & d. miller, 2004). scct was chosen as the foundation for the current study because it links attribution of responsibility with crisis type and because its main components include reputation repair strategies, crisis types, and attribution of responsibility. coombs outlined three phases of crisis management: pre-crisis, crisis response, and post crisis. the pre-crisis phase focuses on crisis prevention and preparation. the crisis response phase focuses on successfully responding to an occurring crisis. the post-crisis phase focuses on preparing for future crises while fulfilling promises made during the actual crisis. pre-crisis best practices include having a continually updated crisis management plan, having a designated and properly trained crisis management team; implementing mock crisis exercises to test the reliability of the plan, and pre-drafting messages to combat specific crises. postcrisis best practices include delivering all promised information to stakeholders, continually updating stakeholders on the progress of the crisis recovery efforts, and closely reviewing the crisis management actions to see if they should be integrated into future crisis management plans (coombs, 2007). 2.6 the role of pr in crisis communication grunig’s excellence theory asserts that all communications must be strategically coordinated and executed by the public relations department. granting management powers to this department empowers the company to successfully respond to a crisis, if and when one does occur (l. hagan, 2007). the federal emergency management agency defines emergency management as “the managerial function charged with creating the framework within which communities reduce vulnerability to hazards and cope with disasters” (fema, 2007). mitigation, when successfully executed, can drastically lessen the negative impact disasters may wreak on people and/or property. examples of mitigation include purchasing flood insurance, enforcing building codes, or using fireretardant materials during construction. preparedness includes plans or procedures designed to save lives and minimize damage when an emergency occurs. vital aspects of the preparedness phase include planning, training, and exercising disaster drills. preparedness is a critical phase of emergency management because it ensures that crisis managers know what specific actions to take when a certain type of disaster occurs. response involves the actions taken to save lives and prevent further damage in a disaster or emergency situation. examples of responses include damage assessment, search and rescue, fire fighting, and sheltering victims. recovery is action taken to return the community to normal following a disaster. examples of recovery include repairing, replacing, or rebuilding property. 2.7 reputation repair strategies just as important as the initial crisis response in the crisis response phase is that of reputation repair and behavioral intentions. much research has been completed in order to compile a database of successful published by atlantis press copyright: the authors 6 reputation repair strategies. coombs’ list of reputation repair strategies is often used by companies or organizations when a crisis occurs. this list includes nine strategies of reputation repair: attack the accuser, denial, scapegoat, excuse (provocation, defeasibility, accidental, or good intentions), justification, reminder, ingratiation, compensation, and apology (coombs, 2007). attack the accuser exists when a crisis manager confronts the individual who is making a claim against the organization. denial exists when a crisis manager denies the existence of a crisis. scapegoat exists when a crisis manager places blame for the crisis on an individual outside of the organization. a provocation excuse exists when a crisis manager asserts that the crisis resulted out of a response to an outside individual’s actions. a defeasibility excuse exists when a crisis manager claims to have had a deficient amount of information prior to the crisis. an accidental excuse exists when a crisis manager had no control over events prior to the crisis. a good intentions excuse exists when a crisis manager insists that the organization meant to do good. justification exists when a crisis manager attempts to minimize the resulting damage of the crisis. reminder exists when a crisis manager reminds the organization’s stakeholders of positive works that were completed by the organization in the past. ingratiation exists when a crisis manager applauds the organization’s stakeholders for specific actions. compensation exists when a crisis manager offers a form of compensation, usually money, to victims of the crisis. apology exists when a crisis manager apologizes for the crisis and takes full responsibility on behalf of the organization (coombs, 2007). also present during the crisis response phase are the elements of crisis type and attribution of crisis responsibility, which typically go hand in hand. these elements are extremely important in determining the reputational threat of a crisis. there are three types of crises: victim crisis; accident crisis; and preventable crisis. a victim crisis may consist of a natural disaster, such as a hurricane or earthquake, a rumor, workplace violence, or product tampering. an accident crisis may consist of a challenge, technical error, or product harm. a preventable crisis may consist of a human-error accident, human-error product harm, or organizational misdeed. usually, a victim crisis charges the organization with minimal crisis responsibility, an accident crisis charges an organization with low crisis responsibility, and a preventable crisis charges an organization with strong crisis responsibility. 2.8 research questions the purpose of this study is to better understand bp’s use of twitter as a crisis communication tool during the gulf oil spill during the crisis response phase. by analyzing the content of bp’s tweets during this time period, one may understand its public relations, crisis communications, and social media strategies. specific topics that will be analyzed include: emergency management phases, reputation repair strategies, crisis types, attribution of responsibility, and risk smart items. the following research questions about @bp_america’s tweets guided this study: rq1: which phase(s) of emergency management was the most prevalent? rq2: what reputation repair strategies were most prevalent? rq3: which crisis types were reflected most often? rq4: which type(s) of attribution of responsibility were reflected most often? all corporate companies must have an online strategy to accompany a crisis communication plan, if they wish to be successful in the face of a crisis. this study examines how a lack of an online crisis communication plan, prior to a crisis, can prove detrimental to a company’s reputation. it also highlights how social media networks such as twitter can be used successfully as a crisis communication tool. although a social media network may be used for many reasons during a crisis, this study examines how such a platform may be used for purposes pertaining to emergency management, reputation repair strategies, different crises, and different levels of responsibility attribution. 3. methods this study used a systematic, quantitative content analysis of all 1,161 tweets posted online by bp_america during its crisis response phase, april 20july 15, 2010. this period extended from the drilling rig explosion through the date the wellhead of the macondo prospect was capped. tweets from april 20 published by atlantis press copyright: the authors 7 to june 4, 2010 and june 7-9, 2010 were obtained from http://topsy.com, a website that allows for real-time searches of the social web. in addition, tweets from june 5-6, 2010, and all remaining tweets (june 10-july 15, 2010), were provided upon request by bp in the form of a spreadsheet titled “twitter content log.” coder a coded all tweets, while coder b coded 20% of the tweets (i.e. 232) by coding every fifth tweet of the sample. the 44-item code sheet identified concepts within each tweet. items included the date in which the tweet was broadcast, actual text of the tweet, whether the tweet was an original post, retweet (rt), or @reply, whether or not the tweet pertained to the deepwater horizon oil spill crisis, number of hashtags included in the tweet, terms used in hashtags, number of links included in the tweet, and where the link redirected the user. the code sheet examined tweet content during different phase(s) of emergency management (mitigation, preparedness, response, and recovery), and strategies of reputation repair reflected in tweets (attack the accuser, denial, scapegoat, provocation excuse, defeasibility excuse, accidental excuse, good intentions excuse, justification, reminder, ingratiation, compensation, and/or apology). the study also identified crisis type(s) reflected in tweets (victim crisis, accident crisis, and preventable crisis), and type(s) of attribution of responsibility reflected in tweets (none, low/minimal, or a strong/high level of crisis responsibility). the code sheet included items for month and date, and asked coders to record links that redirected the user to different websites. coders identified whether each tweet was an original post, a retweet (rt), or a reply. an original post was defined as a tweet organically composed by bp_america. a retweet was defined as a tweet originally tweeted by another twitter user that bp_america chose to retweet, using its twitter account. a retweet was identified by the “rt” abbreviation plus the other twitter user’s twitter name (ex: rt @johndoe13: see spot run...). a reply was defined as a tweet composed specifically in reply to another twitter user or directed at a specific twitter user. the reply was identified by the @ symbol at the beginning of a tweet, followed by the other twitter user’s twitter name. each tweet was coded for whether it pertained to the gulf of mexico crisis, how many hashtags were included in the tweet, what term(s) were included in the hashtag, how many links were present within the tweet, and where the link(s) redirected the user. coders then identified any of the four phases of emergency management (r.r. dynes, e.l. quarantelli & g.a. kreps, 1981) that were reflected in the tweet. the four phases of emergency management are mitigation, preparedness, response, and recovery. more than one phase could be present within a tweet. a tweet in which mitigation was reflected would include language that attempted to reduce the impact of the deepwater horizon explosion and/or oil spill. mitigation, the cornerstone of emergency management, is the continuing effort to lessen the impact that disasters have on people and property. mitigation is sustained action that reduces or eliminates long-term risk to people and property from natural hazards and their effects. if the word “mitigation,” or a variation of that word, was included in the tweet, the coder chose this crisis phase. a tweet in which preparedness was reflected would include language that revealed the existence and/or development of risk management plan(s) for the current and/or future oil spill(s), such as training, drills, plans or procedures designed to save lives or to minimize damage when an emergency occurs. a tweet in which response was reflected would include mentions of oil spill emergency services, such as actions taken to save lives or prevent further damage in a disaster or emergency situation. response is putting preparedness plans into action (r.r. dynes, e.l. quarantelli & g.a. kreps, 1981). this phase also includes protecting wildlife and/or shorelines by way of booms, dispersants, and skimming. a tweet in which recovery was reflected would include mentions of oil spill recovery efforts to rebuild and repair damage caused by the explosion or oil spill. recovery efforts may include mentions of repair, replacement, rebuilding, cleanup, drilling of relief wells, capping of the well, etc. recovery includes monetary response efforts, such as donations from bp, fines assigned to bp, personal claims, etc., or other actions taken to return the community to normal. reputation repair strategies were coded in terms of whether the tweet: (i) confronted a person or group claiming that something is wrong with bp published by atlantis press copyright: the authors 8 (ii) asserted that no crisis, or specific aspect of the crisis, is present (iii)placed blame of the crisis on an outside group or person (iv)attempted to make an excuse by claiming the crisis was a result of someone else’s actions (v)attempted to make an excuse by claiming a lack of information about events leading up to the explosion or oil spill (vi)attempted to make an excuse by claiming a lack of control over events leading up to the explosion or oil spill (vii)attempted to make an excuse by claiming that the organization meant to do well (viii)included a justification that minimizes the perceived damage caused by the explosion or oil spill (ix)reminded people of past good works completed by bp (x)reminded people of a specific amount of oil collected, money paid in claims, boom deployed in response to the oil, or wildlife rescued (xi)praised bp stakeholders for their actions pertaining to the oil spill, including the thanking or praising of volunteers (xii)included offers of money or other gifts from bp to direct and indirect victims of the explosion or oil spill such as claims given to individuals, research funds, or wildlife funds (xiii)indicated that bp accepted full responsibility for the explosion or oil spill or asked stakeholders for forgiveness. coders recorded whether the tweet reflected a victim, accident, or preventable crisis. only one crisis type could be chosen per tweet. a tweet in which victim crisis was reflected would claim that bp was the victim of the explosion or oil spill, such as the victim of the crisis was a result of a natural disaster, rumor, or product tampering/malevolence (ex: the oil spill being compared to hurricane katrina). a tweet in which an accident crisis was reflected would claim that the explosion or oil spill was an accident. an accident crisis could include a challenge, such as a stakeholder claiming the organization operated in an inappropriate manner, a technical error accident such as the explosion and/or oil spill being caused by equipment or technology failure, or technical error product harm such as an equipment or technology failure caused a product to be defective or harmful. a tweet in which a preventable crisis was reflected would claim that the explosion or oil spill could have been prevented. for example, the crisis may have been caused by a human-error accident (i.e. the explosion and/or oil spill was caused by human error), human-error product harm (i.e. a product was defective or harmful because of a human error), or organizational misdeed (i.e. bp management actions put stakeholders at risk and/or violated the law). coders evaluated three types of attribution of responsibility reflected in the tweets: no crisis responsibility, low or minimal crisis responsibility, and strong or high crisis responsibility. only one type of attribution of responsibility was coded for each tweet. a tweet in which no crisis responsibility was reflected would indicate that bp was not responsible for the explosion or oil spill or that the explosion or oil spill was a victim crisis, such as hurricane katrina. a tweet in which low/minimal crisis responsibility was reflected would indicate that bp was minimally responsible for the explosion or oil spill. this type of attribution of responsibility indicated that the explosion or oil spill was an accident crisis. for low/minimal crisis responsibility, no specific actions taken by bp were noted. however, if bp tweeted about an entity separate from itself taking an action against the oil spill, this would reflect low/minimal crisis responsibility. a tweet in which strong/high crisis responsibility was reflected indicated that bp took full responsibility for the explosion and/or oil spill. for strong/high crisis responsibility tweets, a specific action was noted when one was mentioned. during the initial coding process, recurring themes present in the hashtag term(s) and link redirection site(s) were noted, and a secondary coding process was implemented to quantify recurring hashtag terms and urls. recurrent and/or unique hashtag terms and urls were then given a label and were included in the spreadsheet so they could be accounted for as well. because of the approach used, the hashtag term and url coding was not exhaustive. the following hashtags were included in the spreadsheet: #bp, #oilspill, #oil_spill_2010, #gulf, #air, #water, #birds, #topkill, #bpcares, #uscg, and #usfws. published by atlantis press copyright: the authors 9 4. findings the greatest proportion of bp’s tweets (40%) appeared in the last two weeks of the crisis response phase, while only 0.3 percent of the tweets appeared in the first two weeks. about 23% appeared in may, and 37% appeared in june. about 85% of all bp tweets were original posts, while only eight percent were replies and six percent were retweets. out of all tweets, 99.5% were relevant to the oil spill. a total of 572 hashtags appeared within the tweets; 129 tweets included a single hashtag; 200 tweets (18 percent) included two hashtags; 13 tweets included three hashtags; and one tweet included four hashtags. a total of 815 links were included within the tweets; 763 tweets included one link, 20 tweets included two tweets, and four tweets included three links. urls directed the user from bp’s tweets to bp’s website (general); bp’s press page on website; bp’s gulf of mexico restoration page on website; facebook; youtube; flickr; restorethegulf.gov; deepwater horizon memorial; abc; cnn; cspan; cbs; npr; ap; pbs; msnbc; the wall street journal; the new york times; the los angeles times; the huffington post; the boston globe; usa today; u.s. coast guard; geoplatform.gov/gulfresponse; u.s. environmental protection agency; tri-state bird rescue; u.s. fish and wildlife service; international bird rescue; national fish and wildlife foundation; national oceanic and atmospheric administration. rq1 examined which phases of emergency management were most prevalent in bp’s tweets. the four phases were reflected in the 1,161 tweets on 1,090 separate occasions. of the four phases of emergency management (mitigation, preparedness, response, recovery), the recovery phase was most prevalent in tweets from bp, appearing in 54% of the tweets. the response phase appeared in 36% of the tweets, while the preparedness phase appeared in three percent of the tweets, and the mitigation phase appeared in only 0.4% of the tweets. over the 13-week time period, the response phase surfaced most often in tweets during week 12, when bp’s final attempts to sealed the leak were going well and it saw a rise in shares for the first time since the explosion. the fewest response tweets appeared in the second week of the crisis, when the government investigation began, state emergencies were declared, and the rate of oil flow was declared to be far worse than originally thought. the recovery phase surfaced most in tweets during the last week of the crisis, when bp stopped the oil leak for first time since the explosion (figure one). meanwhile, weeks three and seven had the fewest tweets that referenced recovery. during week three, bp blamed transocean for the oil spill, and during week seven, bp faced political flack over its decision to pay out more than $10 billion in dividends to its shareholders rather than to victims. fig. 1. response vs. recovery in post-spill tweets published by atlantis press copyright: the authors 10 rq2 examined which reputation repair strategies were most prevalent in tweets (table 1). these reputation repair strategies included attack the accuser, denial, scapegoat, excuse through provocation, excuse of defeasibility, excuse about an accidental situation, excuse from good intentions, justification, reminder, ingratiation, compensation, and apology. of these 12 reputation repair strategies, eight were reflected within the tweets: compensation (14%), reminder (12%), denial (0.9%), ingratiation (0.5%), apology (0.5%), justification (0.3%), attack the accuser (0.1%), and accidental excuse (0.1%). none of the tweets reflected the scapegoat, excuse (provocation), excuse (defeasibility), and excuse (good intentions) strategies. table 1. reputation repair strategies in bp tweets strategies frequency percent compensation 163 14.0 % reminder 139 12.0 % denial 11 0.9 % ingratiation 6 0.5 % apology 6 0.5 % justification 4 0.3 % attack the accuser 1 0.1 % excuse (accidental) 1 0.1 % scapegoat 0 0.0 % excuse (provocation) 0 0.0 % excuse (defeasibility) 0 0.0 % excuse (good intentions) 0 0.0 % reminder and compensation strategies of reputation repair were most prevalent (figure two). reminder tweets appeared most frequently during week 10, when bp shares hit a 14-year low, and least often during weeks three and four, when bp blamed transocean for the crisis. compensation tweets appeared most often in week nine, when the white house demanded that bp pay $20 billion in compensation to victims and no dividends to shareholders. compensation was mentioned least often during week 6, when the u.s. launched a criminal investigation into the oil spill. fig. 2. reminder vs. compensation strategies in bp tweets published by atlantis press copyright: the authors 11 rq3 identified which of the three crisis types (victim, accident, preventable) were reflected most often in tweets. accident crisis was reflected most often, appearing in 97% of tweets, while victim crisis was reflected in only 0.2% and preventable crisis in 0.1%. preventable crisis was reflected in only one tweet (0.1%). rq4 examined which responsibility attribution types were reflected most frequently. the three types were no crisis responsibility, low/minimal crisis responsibility, and strong/high crisis responsibility. the strong/high crisis responsibility type was reflected most often, appearing in 90% of all tweets. low/minimal crisis responsibility was reflected in only seven percent, while no crisis responsibility was reflected in 0.2% of the tweets. among tweets reflecting that the oil spill “had strong political attributes tied to it,” most were tweeted during week nine, when congress accused bp’s ceo of "stonewalling" after failing to answer questions about the spill's causes. the fewest were tweeted in the following week, week nine, when the bp ceo failed to appear at a gathering of the oil industry and greenpeace protesters stormed the stage at the convention. most tweets reflecting that the oil spill “caused serious financial harm” appeared in week nine, when bp's credit rating was downgraded by moody's after expressing concern at the escalating cost of the cleanup and potential cost of litigation claims. the fewest were tweeted during week 6, when bp began an advertising campaign in the u.s. aimed at boosting opinion. among tweets reflecting that the oil spill “caused death/injury through action or inaction,” most appeared during week ten, when hundreds of activists staged protests against bp in london. the fewest of these tweets appeared during weeks two, when catastrophic contamination and wildlife destruction was reported, and in week six, when the u.s. launched a criminal investigation into the oil spill. all tweets were coded for the following hashtags: #bp; #oilspill; #oil_spill_2010; #gulf; #air; #water; birds; #topkill; #bpcares; #uscg; #usfws. the most prevalent hashtag was #oilspill, which appeared in 26%, followed by #bp, which appeared in 19%. #oil_spill_2010 appeared in only 0.5%, #topkill and #usfws each appeared in 0.3%. #gulf and #uscg appeared in 0.3%, and #air, #water, #birds, and #bpcares appeared in 0.1%. bp seldom used redirects in its tweets (table two). urls redirected users to restorethegulf.gov in ten percent of the tweets, followed by the gulf of mexico restoration page on bp’s website (nine percent), youtube (eight percent), and bp’s general website (eight percent). urls redirected users to the press page on bp’s website on 77 occasions (seven percent). urls redirected users to facebook on 57 occasions (five percent). urls redirected users to cnn on 46 occasions (four percent). urls redirected users to flickr and cspan on 19 occasions, respectively (two percent). urls redirected users to the u.s. environmental protection agency on 14 occasions (1%). urls redirected users to msnbc on 12 occasions (one percent). urls redirected users to geoplatform.gov/gulfresponse on 11 occasions (one percent). urls redirected users to abc on nine occasions (one percent). urls redirected users to the national oceanic and atmospheric administration on eight occasions (one percent). urls redirected users to the ap and tri-state bird rescue on five occasions, respectively (0.4%). urls redirected users to cbs and the u.s. coast guard on four occasions, respectively (0.3%). bp seldom redirected users to news media outlets. urls redirected users to the wall street journal, new york times, los angeles times, npr, pbs, the boston globe, but only about two mentions for each of these media outlets (0.4% across all tweets). even less frequently, urls redirected users to huffington post or usa today (0.1%). 5. discussion this study evaluated bp’s use of twitter as a crisis communication tool during the 2010 deepwater horizon crisis response phase. the research questions guided an analysis of the emergency management phases, reputation repair strategies, crisis types, and types of attribution of responsibility within bp’s tweets posted april 15-july 15, 2010. the recovery phase of emergency management was most prevalent, reflected in 54% of the tweets. bp used most of its tweets during the recovery phase to inform published by atlantis press copyright: the authors 12 its audience of its efforts to rebuild or repair damages in the gulf and on the coast, particularly the repair or rebuilding of the damaged macondo well. these efforts included the lower marine riser package, the riser insertion tube tool, and drilling relief wells. table 2. website redirects within bp tweets websites frequency percent webpage on bp corporate site 273 23.4 % social media 169 14.5 % government oil spill response 124 10.6 % national broadcast 101 8.5 % government agency 27 2.3 % elite newspaper 13 1.0 % environmental agency 8 0.7 % other 1 0.1 % recovery tweets also included information about coastal cleanup efforts and wildlife rehabilitation and release. the response phase was the second-most prevalent emergency management phase seen during this period, appearing within 36.3% of the tweets. this means that bp also used a large number of tweets to provide information about saving wildlife and preventing further damages. response tweets included information about rescuing wildlife and protecting shorelines from oil using booms, dispersants, and skimming. response tweets were most prevalent during the early weeks of the crisis response, and they gradually decreased over time. the opposite is true of recovery tweets, which increased over time. this shows that bp acted in way that was directly in line with emergency management protocol. that cycle begins with mitigation, moves into preparedness, then response, then recover, and back to mitigation. among the reputation repair strategies reflected in tweets, the compensation and reminder strategies appeared at a much higher rate than the other ten strategies. the compensation strategy appeared in 14% of the tweets, revealing that bp only occasionally provided information about claims given to individuals who were financially harmed as a direct result of the oil spill or about donations made to research and wildlife funds. the reminder strategy appeared in 12% of the tweets, indicating the extent that bp provided reminders of good works pertaining to the oil spill that had been completed. these good works included individual claims checks cut, money donated to research or wildlife funds, amount of oil collected, and amount of boom deployed. during the ninth week, the most compensation tweets appeared, in the same week that bp announced its agreement to pay $20 billion over 3.5 years to meet the obligations arising from the oil spill. week ten contained the most reminder tweets; the same week that bp announced it would donate the net revenue from the sale of oil recovered from the macondo well to the national fish and wildlife foundation. they immediately donated $5 million. the accident crisis type was reflected most often, appearing in 97% of tweets. an overwhelming majority of tweets during the analyzed 13-week crisis response period suggested that the explosion and/or oil spill was an accident. most tweets during this period suggested an accident crisis, since the investigation into the cause of the explosion and oil spill had not found bp responsible. ongoing investigations never proved that bp was guilty of a human-error accident, humanerror product harm, or an organizational misdeed. a strong/high level of crisis responsibility was most often reflected within tweets, appearing in 90% of all tweets. this means that an overwhelming majority of tweets during the analyzed 13-week crisis period indicated that bp took full responsibility for the explosion, oil spill, and all resulting issues. since admiral thad allen, a retired u.s. coast guard admiral, was appointed as national incident commander of the unified command for the deepwater horizon oil spill, his name was mentioned in a large amount of tweets, implying strong political attributes. bp included hashtags in many tweets in an attempt to be included in trending conversations in the twitterverse. although a variety of hashtag terms were used during the crisis response phase, #oilspilll published by atlantis press copyright: the authors 13 and #bp were by far used the most. this is probably because these terms were more popular trending terms; therefore, if bp used these specific hashtag terms, they were able to secure a place in trending conversations. hashtags that did not trend well they immediately retired. for example, the hashtag #bpcares appeared in only one tweet during the analyzed 13-week period. bp strategically included certain links in its tweets, as well. most links either redirected users to restorethegulf.gov, a government website created solely for the gulf oil spill, bp’s website, or to one of bp’s other social media outlets. bp probably redirected users most to the government website to gain or retain the trust of its followers, while it probably redirected users to its other websites to remain the main source of information relating to the oil spill. after the explosion of the deepwater horizon drilling rig and the resulting gulf oil spill, bp implemented an extensive online crisis communication and public relations campaign. one major leg of this campaign consisted of repurposing its twitter account, @bp_america, to serve as a hub for updates and information pertaining to the oil spill response. @bp_america tweeted its first oil spill-related tweet on april 27, seven days after the explosion, and continued to tweet thousands of oil spill-related tweets as the response and recovery efforts continued. bp used its twitter account for a variety of reasons, and quite strategically. the tweets reflected various emergency management phases, reputation repair strategies, crisis types, types of attribution of responsibility, and public perceptions. bp implemented specific reputation repair strategies in a manner in response to current events. for example, bp employed a compensation strategy after making a large monetary donation or a reminder strategy after collecting a large amount of oil. one area of bp’s crisis communication plan was not in line with past crisis communication research. coombs’ situational crisis communication theory suggests that accident crisis types usually indicate a low or minimal attribution of responsibility. however, most of bp’s tweets revealed an accident crisis with a high or strong attribution of responsibility. although the investigation into the explosion and oil spill did not point to a preventable crisis, bp still took full responsibility for the incident. also, preventable crisis was only reflected in one tweet, as ongoing investigations during the crisis response period never proved that bp was responsible for a human-error accident, human-error product harm, or organizational misdeed. although bp provided a twitter content log for most tweets analyzed in this content analysis, this log only dated back to june 5, 2010, so missing tweets from june 6-9 were obtained through topsy.com, a credible search engine with an extensive twitter library. however, there is always the possibility that some tweets from @bp_america might be missing from its library. bp extensive online crisis communication campaign included the use of online sites other than twitter, including its corporate website, facebook, flickr, and youtube. future research might study how bp used all social media outlets across all crisis phases of the oil spill, in light of reputation repair strategies and public perceptions. one could then compare the use of these social media outlets to twitter and uncover important similarities and differences. finally, future research might compare bp’s twitter response to tweets used in a different corporate crisis, such as the 2010 toyota recall for brake problems, to examine how they responded differently in admitting mistakes to the public. references m. debord, crisis pr must not confront its own problems, the washington post (july 25, 2010). j. vidal, bp’s pr campaign fails to clean up reputation after gulf oil spill, guardian, 2011. http://www.guardian.co.uk/environment/blog/2011/apr/14 /bp-pr-campaign-gulf-oil-spill r. blake, “bp shares sink amid oil spill bankruptcy chatter.” abc news (june 10, 2010). http://abcnews.go.com/business/bp-survive-companyresult-oil-spill-gulf-mexico/story?id=10874702 associated press, “bp chief at yacht race while oil spews into gulf,” new zealand herald (june 20, 2010). e. reguly, bp’s pr woes start at the top, the globe and mail (june 16, 2010). e. mcclam and h. weber, bp’s failures made worse by pr mistakes, msnbc.com, 2010. d. lenci and j. mullane, communicating with the public: how bp told the macondo story, oil & gas journal (december 6, 2010). a. brownsell, bp, marketing (july 2010). b. snyder, tony hayward’s greatest hits, cnn money, june 10, 2010. http://money.cnn.com/2010/06/10/news/companies/tony_ hayward_quotes.fortune/index.htm published by atlantis press copyright: the authors 14 a. caruba, bp, pr, and the oil spill, canada free press (june 7, 2010). d. lenci, denise and j. mullane, communicating with the public: how bp told the macondo story, oil & gas journal (december 6, 2010). t. bergin, “analysis: bp pr blunders carry high political cost.” reuters (june 29, 2010). a. symington, legacy of the bp spill: what’s a reputation worth? the bp spill in the gulf of mexico caused a public outcry and savaged bp’s share price. image repair won’t be easy, christian science monitor (october 18, 2010). b. morrissey, bp gets aggressive, adweek (june 21, 2010). a. seitz-wald, “bp launches ‘aggressive’ social media campaign but disables comments from users who don’t ‘like’ it,” truth-out.org (june 11, 2010). http://www.truth-out.org/bp-launches-aggressive-socialmedia-campaign-disables-comments-users-who-dontlike-it60633 b. hatton, bp’s pr crisis shows why social media matters, daily post (august 2, 2010). h. shaw, the truth shall set you free of pr hell, national post (june 25, 2010). k. fearn-banks, crisis communications: a casebook approach (mahwah, nj: lawrence erlbaum, 2002), p. 8. t. coombs, crisis management and communications, gainesville, fl: institute for public relations, 2007: http://www.instituteforpr.org/topics/crisis-managementand-communications/ r. parekh and e. lee. “how to succeed when it’s time to make your social-media mea culpa.” advertising age 81 (19): 2, 2010. r. l. heath and d. p. miller, responding to a crisis: a rhetorical approach to crisis communication (mahwah, nj: lawrence erlbaum, 2004), p. 97. l. m. hagan, “for reputation’s sake: managing crisis communication,” in the future of excellence in public relations and communications management, ed. e. l. toth (mahwah, nj: lawrence erlbaum, 2007), p. 413. fema, principles of emergency management supplement (sept. 11, 2007), p. 4. r. r. dynes, e. l. quarantelli, and g. a. kreps, report series #11, a perspective on disaster planning, 3rd ed., newark, de: university of delaware disaster research center, 1981. published by atlantis press copyright: the authors 15 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjdffile false /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /description << /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /dan <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> /deu <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> /esp <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> /fra <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /ptb <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> /suo <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> /sve <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style editor’s introduction two invited papers from two presidents open the present issue, volume 4, issue 3 (2014). their themes are emerging risk and disaster resilience, which are two focal concerns in risk analysis and crisis response. the other four articles following them are also wonderful. ortwin renn, president of the society for risk analysis (sra), in paper “emerging risks: methodology, classification and policy implications”, gives a first application of the irgc protocol for classifying emerging risk clusters. the article specifies six risk patterns that match the three categories for emerging risks. each of these patterns constitutes a special threat to the global community. the irgc remerging risk protocol might help risk managers to acquire the necessary knowledge, to articulate the insightful scenarios and to develop the most appropriate risk management strategies for finding the right balance between opportunities and risks. k. harald drager, president of the international emergency management society (tiems), with thomas v. robertson, in paper “tiems drevs initiative: disaster resilience establishment for vulnerable societies”, introduces an initiative toward to (1) establishing a global pool of em expertise, with shared frames of reference and shared best practices; and (2) creating a foundation for better global collaboration by developing and testing models of local participatory governance in international collaboration for disaster preparedness. the paper “identification of global indicators for regulatory risk management in production platforms”, by alex almeida and figueiredo, selects and applies methods to normalize and compose global indicator by using obtained information related to risk management system and incidents from 67 oil and gas production platforms in brazil. this study aims to analyze different indicator’s programs from regulators and other organizations. there are two papers in the crisis analysis. the first paper “a gamer’s nightmare: an analysis of the sony playstation hacking crisis” by olaniran and et al, focuses on assessment of the sony playstation hacking using the anticipatory model of crisis management. it was found that sony corporation could have handled the crisis better. the second paper “shipwreck: a crisis with challenging solutions" by nikolaos p. ventikos and et al., attempts to highlight the main parameters of a wreck removal operation and thus define a solid framework for the preliminary assessment of different wreck removal methods. the paper aims at presenting the challenges faced and the risks assumed by salvors during wreck removal operations, which is done by examining an illustrative case study and implementing a scenario based analysis. there is one paper in insurance, “an empirical research on the relationship between property insurance premiums and macroeconomic variables based on ardl model”, where guiyun you and his colleagues established an autoregressive distributed lag model (ardl) which considers adding lags of the dependent variable and/or lags of some independent variables. based on china’s insurance premiums, gross domestic product, consumer price index and fixedasset investment data from 1980 to 2012, we analyze the long-term and short-term relationship between them using the ardl bounds testing approach. the results indicate that gdp is the major factor driving the growth of property insurance premiums in china; fixed-asset investment has significant impact on china’s property insurance premiums and they are adversely related; consumer price index has almost no effect on the premium income. the society for risk analysis china many thanks to the referees for their strong support. and also thank the authors very much for all their submissions. chongfu huang and junxiang zhang september 1, 2014 journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 113 published by atlantis press copyright: the authors 113 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). the design of a drought weather index insurance system for summer maize in anhui province, china * ying xu 1 , chao gao 2,† , xuewen li 3 , taiming yang 1 , xibo sun 4 , congcong wang 2 , de li 5 1 agricultural meteorological center of anhui province, hefei 230031, china 2 department of geography & spatial information techniques, ningbo university, ningbo 315211, china 3 college of territorial resources and tourism, anhui normal university, wuhu 241000, china 4 guoyuan agricultural insurance company, hefei 230031, china 5 meteorological of anhui province，suzhou 234000，china received november 23, 2017 accepted january 8, 2018 abstract we analyzed precipitation anomalies of summer maize in order to establish a drought weather index model that is based on tassel appearance and maturity stage as well as drought yield reduction rate. we utilized daily precipitation data encompassing the period between 1971 and 2010 from 15 agricultural meteorological observation stations within a summer maize planting area in anhui province, china. a compensation standard for insurance claims resulting from summer maize drought is proposed in this paper, and revisions are presented to the drought weather index by incorporating relative humidity from the initial insurance period. the results of this study enabled the generation of a summer maize drought weather index insurance product which was then tested within the city of huaibei in suixi county, anhui province. results also demonstrate that application of a weather index insurance product can rapidly and objectively enable the provision of economic compensation in the aftermath of agricultural disasters. keywords: summer maize; drought; weather index; insurance * foundation item: under the auspices of ―research on threshold of meteorological factors of drought and waterlogging disasters in different spatial scale—a case study in the upper reaches of huaihe river basin (41571018)‖. † corresponding author: e-mail: gaoqinchao1@163.com post address: department of geography & spatial information techniques, ningbo university. 1. introduction agriculture is a risky industry. indeed, meteorological disasters are the main kind of natural catastrophes that endanger agricultural production and are closely linked to economic benefits (zhang et al., 2011; antón et al., 2013). agricultural insurance is therefore an important mechanism that can help to reduce the cultivation risks faced by farmers, and can also mitigate perceived threats to agricultural production. thus, if losses are suffered to crops covered by insurance, compensation will be paid to farmers (coble et al., 2003; thomas et al., 2007; panda et al., 2013). the concept of meteorological index insurance has been widely studied and applied in various countries over recent years. this kind of product is a new form of agricultural protection which determines both the insurance rate and loss compensation using objective meteorological data (barnett et al., 2010; bobojonov et al., 2014; finger et al., 2014; mornet et al., 2015). a number of steps have contributed to the development of meteorological index insurance, including an initial series of research studies that explored different crop 14 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 types, including rice, maize, apples, and citrus (leblois et al., 2013; chen et al., 2017; zhang et al., 2017). on this basis, a meteorological disaster claim index model was then designed to generate insurance contracts based on the relationship between reductions in crop yield and meteorological hazard factors (jin et al., 2016). judgements are therefore made about the level of disaster loss in order to carry out a division of risk and an assessment of meteorological index insurance that combines risk similarities and the variability of characteristics within a disaster environment (liu et al., 2010; cao et al., 2013). in terms of research methods, it is then important to select relevant meteorological indicators that are based on regional characteristics, including selecting approaches that incorporate the concepts of moving averages and polynomial fitting in order to accurately separate meteorological yields (blanc et al., 2012). probability functions, such as beta and gamma distributions, can then be applied to determine the risk probability of a given yield reduction rate at all levels and thus to determine the premium rate (lu et al., 2008; ozaki et al., 2009). in an international context, precipitation index insurance is commonly utilized in the usa, south africa, and mexico, for example, to mitigate the risks of drought and waterlogging disasters (raphael et al., 2006; barnett et al., 2010), while a meteorological index of citrus frost damage has been proposed for use in china as the compensation standard for this kind of issue. this index is based on risk analysis of annual minimum temperatures within the main production areas for this citrus fruit in zhejiang province (mao et al., 2007). in another study, liu et al. (2010) proposed the use of the apple florescence freezing injury insurance index developed on the basis of different levels of low temperature freezing damage as measured using this effect in shaanxi apples, while yang et al. (2013) proposed the winter wheat planting weather index. this latter measure was designed in anhui province with reference to the five main kinds of meteorological disasters that influence agricultural wheat growth and has been successfully applied within the city of suzhou (yang et al., 2013). designing weather index-based insurance products is difficult if they are expected to encompass a large area; indeed, few such products are currently available for use at the provincial level. the products that are available have mostly been based on research areas selected at municipal or county levels; a few studies have even utilized villages and towns as their research areas (lou et al., 2011). summer maize is the third most important grain crop grown in anhui province, as well as a key feed and industrial raw material (sheng et al., 2014). the main agricultural areas for this crop are concentrated in the arid huaibei region within the huaihe basin, a zone that experiences marked precipitation variability throughout the growing season. the tassel appearance maturity stage of summer maize is the most critical period for a water shortage, as during this phase, precipitation only encompasses 64% of water demand. this phase is also a critical production time as other precipitation periods can generally meet the water demand requirements of maize (yang et al., 2016). an analysis of precipitation anomalies and the drought yield reduction rate of summer maize in anhui province is therefore presented in this study, alongside a drought insurance claim compensation standard for this crop. a revision of the drought weather index is also presented in this study that takes into account relative humidity during the initial phase of the insurance period. finally, an insurance product based on the drought weather index for summer maize is proposed. 2. materials and methods 2.1. materials the meteorological data utilized in this study consists of precipitation, temperature, and hours of sunshine records collected at 15 stations within the anhui province summer maize planting area between 1971 and 2015 (fig. 1). geographic data was also collected for analysis in this study, including the latitude, longitude, and elevation of each county as well as summer maize yields and observational records collected across the growth period for nearly 40 years. these data were supplied by ―the meteorological information center of anhui province‖. the output data of summer maize in anhui province come from statistics bureau of anhui province. 15 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 observational data for the summer maize growth period, yield data treatments, and the weather index model were compared via conventional statistical analyses using the software packages spss and excel. 2.2. summer maize drought weather index applying the design principle (yang et al., 2013) that underlies the anhui province winter wheat weather index, a drought index for summer maize should incorporate factors that are less influenced by human activities, should be simple to calculate, and easy to understand and popularize. among the currently available domestic drought monitoring indices (table 1), the precipitation anomaly percentage index does possess these characteristics and utilizes data that is easy to obtain. this measure has also been widely applied to assess meteorological and agricultural droughts and utilized in other research areas. this study therefore fig. 1. map to show the distribution of summer maize agro-meteorological observation stations throughout anhui province. table 1. comparison of the advantages and disadvantages of different drought indices. drought index advantages disadvantages drought index based on precipitation rainfall index quick, simple, and practical takes just a single precipitation factor into account and does not consider the effects of crops, the underlying surface, and other related factors percentage precipitation anomaly (pa) clear meaning and simple method does not consider the effects of soil moisture and relies heavily on the average values standardized precipitation index calculation is stable and spatiotemporal distribution differences in precipitation are eliminated calculation is tedious, difficult to utilize and promote drought index based on soil moisture soil relative humidity intuitive, data is easy to access determination difficulty and index is poorly representative palmer drought index crop water supply can respond sensitively calculation is tedious, difficult to utilize and promote crop drought index leaf water potential, stomatal conductance, and cell sap concentration can directly and sensitively reflect crop water supply deficit a large number of field tests are required to obtain the necessary parameters, data representation is limited, and accuracy needs to be improved comprehensive index crop water stress index reflects the combined effects of soil, plant factors, and meteorological conditions calculation is tedious, difficult to utilize and promote 16 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 builds on the anhui province drought weather index for summer maize and analyzes the relationship between negative precipitation anomalies at different growth stages and the crop yields of summer maize across this region using meteorological and agricultural meteorological station data from 15 cities province-wide. the negative precipitation anomaly of summer maize during the tassel appearance maturity stage (i.e. between the middle of august and early september) was chosen in this study as the summer maize drought weather index for anhui province. this negative precipitation anomaly has a negative percentage value. precipitation anomaly percentage was calculated as follows: 100% a p p p p    (1). in this expression, pa denotes the percentage of precipitation anomalies, while p denotes precipitation over a given period, and p refers to annual average precipitation over the same period. 2.3. revising the summer maize drought weather index for use in the huaibei region the huaibei region of anhui province is characterized by the presence of sandy black soils which have poor water retention capabilities and are therefore sensitive to both droughts and waterlogging. the distribution of summer maize roots is also relatively shallow, mainly within the top 20 cm of the soil layer; thus, if the soil surface dries out, crops cannot grow normally. thus, in the absence of supplementary water from precipitation or irrigation, soil moisture will gradually decrease (i.e. water is consumed) over time; the soil moisture loss rate varies significantly depending on weather, soil and crop type, and other factors. in previous research, ma et al. (2001) noted that mild droughts can be significantly alleviated, or relieved, in the summer (between july and august) by between 30 mm and 40 mm of precipitation, while a 50 mm level can relieve a moderate drought, and more than this can relieve a severe drought (table 2) (ma et al., 2001). as the initial state of the insurance period is not considered arid when constructing the summer maize drought weather index because soil relative humidity at the start of the study period (august 11th) is suitable (i.e. wet), we revised equation (1) on the basis of the data in table 2. the modified expression used in this study is as follows: 65% 0 ( ) 100% 65% 55% 30 55% 50 a sw p p p p p sw p p sw p                (2). in this expression, pa denotes the precipitation anomaly percentage, p refers to precipitation (mm) over a given time period, δp denotes the precipitation (mm) required to relieve a drought, p denotes average precipitation (mm) over the 40 years of the study period, and sw denotes the relative humidity (%) of the top 20 cm soil layer on august 11th. 3. results 3.1. the relationship between drought weather index and summer maize yield reduction there are many reasons for the decline in summer maize. not all of them are affected by drought. therefore, this study select the year of drought-induced reduction in summer maize as the data sample to calculate the relationship between summer drought index and yield reduction. selection of a typical year of drought: calculate the cumulative wetting index of each station by ten –days periods from mid-june to mid-september (ma et al. 2008). cumulative wetting index by ten – days periods ≤-0.40 indicates the drought. if the rate of cut-off is selected, the year of dryness corresponding to the period table 2. the level of precipitation required to relieve drought in the huaibei region of anhui province (mm). season mild drought (65% ≥ sw > 55%) moderate drought (55% ≥ sw > 45%) severe drought (sw ≤ 45%) summer 30-40 50 ≥ 50 abbreviation: sw, relative humidity of the top 20 cm soil layer 17 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 of development will be eliminated only for a few days. if for some drought years, compare that with its historical facts, and if it is prone to drought, it is reserved for this ten days. building on data that describes the growth and development of summer maize in anhui province, precipitation records for this crop during the tassel appearance maturity stage were collected from 15 meteorological stations within the huaibei region encompassing the period between 1971 and 2010. these data were then used to calculate values of the summer maize drought weather index. a trend equation was then deduced based on summer maize yields by applying a linear sliding average method to simulate this change (xu et al., 2009). the correlation between summer maize drought weather index and yield was then expressed as a coefficient. the expression used to calculate the drought weather index (i.e. negative precipitation) of summer maize during the tassel appearance maturity stage can be deduced from equation (1), as follows: 100%c c c p p h p     (3). in this expression, h denotes the drought index of summer maize during the tasseling milk stage, pc denotes actual precipitation (i.e. during the period between the middle of august and september 10th) throughout this stage, and cp refers to the average precipitation seen during this stage over the last 40 years. the resultant regression equation for the summer maize drought index and yield reduction rate within anhui province is therefore as follows: 0.278 11.8x h  (4). in this expression, x denotes the yield reduction rate of summer maize, and h denotes to the drought weather index of summer maize. the data collected for this study were used to verify this equation and resulted in p values of 0.019 and 0.007, both below the 0.05 significance level. similarly, the coefficient of determination in the regression equation for summer maize drought index and yield reduction rate is 0.336, also below the 0.05 significance level; this result demonstrates that the regression equation fits well to the data and can adequately reflect summer maize drought disasters in anhui province and their relationship with production. thus, based on equation 4, data that describes the relationship between drought weather index and summer maize yield rate is presented in table 3. the data presented in table 3 show that when the drought weather index of summer maize was 50%, the yield reduction rate of local maize was 26%. similarly, when the drought weather index was 70%, the yield reduction rate was 31%, whereas when it was 100% (i.e. no precipitation throughout this period), the yield reduction rate of maize was 40%. the relationship between calculated drought weather index and yield reduction rate based on summer maize drought grade across anhui province was also calculated (table 4). these data show that the grade remained consistent, which means that equation (4) can be used to adequately express the actual local situation. 3.2. developing a compensation standard based on the drought weather index of summer maize records from local farmers show that the maize yield in the huaibei region of anhui province can reach 7,500 kg/ha if there are no meteorological disasters. thus, as table 3. the relationship between drought weather index (h) and yield reduction rate for summer maize. h (%) yield reduction rate (%) h (%) yield reduction rate (%) h (%) yield reduction rate (%) 10 15 45 24 80 34 15 16 50 26 85 35 20 17 55 27 90 37 25 19 60 29 95 38 30 20 65 30 100 40 35 22 70 31 40 23 75 33 18 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 the current market price for maize is 2.2 rmb/kg, the harvest output value can reach 16,500 rmb/ha. the data presented in table 3 also show that a drought during the growing season can lead to a 40% reduction in summer maize yield. this means that if the output value per hectare of harvest is 16,500 rmb/ha, a drought can cause a maximum summer maize loss of 6,600 rmb/ha. a guarantee level of 70% was therefore selected, taking the local situation into account; this translates to an insurance value of rmb 4,620 (i.e. 0.7 × 6,600). the data presented in fig. 2 show that the amount of compensation per hectare is equal to the amount of insurance per hectare multiplied by the compensation ratio. the premium rate can be calculated from the insurance company's claim data for many years. the local weather index and crop yield reduction are important factors affecting the rate. the data collected in this study reveal an average compensation probability of 10.8% at each site within the huaibei area over the period between 1971 and 2010. this equates to an average compensation payout amount per year of 2,194.5 rmb/ha. thus, as the insurance rate for the drought index of summer maize in the huaibei region is equal to the compensation probability multiplied by the average payout amount over a compensation year, 10.8% × 2,194.5 equates to 237 rmb/ha. data therefore show that the insurance rate for summer maize droughts in the huaibei region should be 237 rmb/ha. the equation used to calculate an insurance premium is as follows: insurance premium = insurance rate × insurance quantity (ha) (5). thus, the insurance amount for summer maize in this region is rmb 4,620/ha; compensation standards for situations when the summer maize drought index is greater than 40% are shown in fig. 2. fig. 2. payment standards based on summer maize drought weather index insurance. table 4. summer maize drought grade divisions in the huaibei region of anhui province. drought grade yield reduction rate (%) negative precipitation anomaly of summer maize during the tassel-milk stage (%) light drought < 10 < 10 moderate drought 10-23 10-40 heavy drought 23-31 40-70 severe drought > 31 > 70 19 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 http://dict.cnki.net/dict_result.aspx?searchword=%e4%ba%a9%e4%ba%a7%e5%80%bc&tjtype=sentence&style=&t=output+value+per+mu 3.3. revision of the drought weather index of summer maize as discussed above, because soil relative humidity is arid at the start of the insurance period (august 11th), a premium rate based on the summer maize drought weather index was revised on the basis of equation (3), as follows: 65% 0 ( ) 100% 65% 55% 30 55% 50 c c c sw p p p p h sw p p sw p                 (6). in this expression, h refers to the dry weather index of summer maize during the tasseling milk stage (i.e. between the middle of august and early september), while pc denotes actual precipitation (mm) during this stage, δp denotes the precipitation (mm) required to relieve drought (mm), and sw refers to the relative humidity (%) of the top 20 cm of soil on august 11th. 3.4. testing the drought weather index of summer maize the drought weather index of summer maize is expressed as a percentage of negative precipitation anomalies. thus, both the role of soil moisture and data that could be used for modeling the correspondence with initial soil moisture non-arid status are not usually taken into account as components of this index. according to the ―china meteorological disaster ceremony (anhui volume)‖, the period between august and early september 1985 in this province was hot and experienced just a small amount of precipitation because of the presence of a strong and stable subtropical higher. in addition to these conditions in the northern huaibei region, other areas of anhui province were also hot at this time and experienced little rain; this year was also characterized by a significant autumn drought controlled by the subtropical high. indeed, throughout the period between august and early september 1986, all regions of this province experienced continuously high temperature, little rain, and both summer and autumn droughts; for these reasons, the drought weather index of summer maize was calculated using precipitation data for the period between august 11th, 1985, and september 10th, 1986. comparing the drought indices calculated before, and after, this revision (table 5) shows that most of the sites in huaibei in 1985 and all sites during 1986 experienced droughts. this revised drought weather index is therefore more realistic than those previously employed. table 5. standard payments for summer maize drought weather index insurance site 1985 1986 (before revision) (after revision) (before revision) (after revision) dangshan -111.41 -65.25 -24.26 21.90 xiaoxian -40.46 8.65 1.97 51.08 bozhou -66.65 -18.20 -24.02 24.43 linquan 57.57 102.61 8.39 53.43 jieshou 57.84 102.13 -23.12 21.17 taihe -31.52 17.48 -12.70 36.30 suixi -216.45 -165.01 -41.25 10.19 guoyang -145.96 -97.49 -44.07 4.41 lixin 46.27 98.74 -27.92 24.55 mengcheng 8.26 63.39 -29.24 25.90 suzhou -157.10 -107.83 -38.85 10.42 lingbi 32.32 77.44 -30.76 14.36 sixian 31.77 79.02 21.65 68.91 guzhen 41.61 92.21 0.32 50.92 3.5. application in august 2015, the anhui meteorological bureau commissioned the ―guoyuan agricultural insurance company‖ to carry out a pilot study on the maize drought index within the city of huaibei in suixi county, anhui province, china. the bureau therefore signed a maize drought index insurance contract with six wealthy and influential families who cultivate around 188 hectares of summer maize with an insurance value of 869,176 rmb. compensation was set to start for these farmers when the drought index was greater than 40% over the set time period defined above (between august 11th and september 10th). continuous precipitation was recorded within the huaibei region throughout august 2015, and meteorological observation data from suixi county station revealed a drought index of 38.7% over this time period. as this index value did not meet the predefined insurance payment conditions, claims were not produced. indeed, subsequent to this insurance pilot experiment, investigations showed that maize plots remained in 20 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 good condition and no drought disaster was recorded; this experiment confirmed that maize drought index insurance can realistically describe actual risk. 4. discussion and conclusion 4.1. discussion a summer maize drought weather index is proposed in this study and practical applications were used to derive a compensation rate for each county within anhui province on the basis of objective meteorological data, independent of actual crop yields. in addition to the advantages of easy handling and simple claims, this index is also conducive to promoting the use of more risky areas to farmers in order to enhance crop management and adjust planting structures. thus, compared with traditional agricultural (makki et al., 2001) and regional production index-based insurance (ozaki et al., 2009), the summer maize drought weather index proposed in this study is based on a model of the actual relationship between meteorological factors and crop yield reduction rates. this means that meteorological data provided by a third party independent of the insurance companies and farmers is used as the basis for both protection and subsequent compensation payments. this approach therefore effectively avoids disputes that might occur between parties during the processes of insurance and payment. building on the previous work of mao et al. (2007), the drought index of summer maize is also revised in this study, utilizing the status of soil moisture during the initial insurance period. the critical summer maize production period (i.e. the tassel appearance maturity stage) was selected for application in this study; as this time period is even shorter in our study, calculated insurance rates are even more accurate. however, whether or not a drought weather indexbased insurance product is practical depends on a number of factors, including the distance between a reference station and adjacent crop site; it is important to ensure that climatic conditions at a planting site and associated reference weather station are comparable. in addition, as yield loss is not entirely related to weather index (skees, 2008; turvey et al., 2010), undesirable effects could also occur; crop losses without compensation, and payments without losses. finally, the demand for weather-based insurance products is generally low amongst target customers, mainly because low income farmers who are most likely to benefit from these schemes tend to be suspicious of their credibility (binswanger-mkhize, 2012). nevertheless, the drought risk insurance product outlined in this study was piloted within the city of huaibei in suixi county, anhui province, over the course of august 2015; the results of this initial test showed that a drought index-based insurance product does accurately reflect actual risk. a number of issues should be addressed in future research, including the fact that this study did not take into account the influence of heavy rainfall or long-term droughts on summer maize yields before the critical period of summer growth. continuous improvements are therefore required to develop the drought weather index model proposed in this study. indeed, the data used in the design of this index, and whether or not environmental differences between collection sites and insured land will lead to compensation errors should also be addressed in future research. 4.2. conclusion building on the previous domestic and international research results, this study addressed the growth of summer maize in anhui province, china, in order to elucidate the relationship between yield reduction rate and drought index. a drought index model for summer maize claims within the province was therefore constructed, a series of disaster insurance rates were determined, and an insurance product was designed based on a statistical analysis of meteorological, crop yield, and growth period data. a regression approach was used to evaluate the relationship between the summer maize drought weather index proposed in this study and yield reductions in anhui province. a significance test was then applied to show that the resultant regression equation model fitted well with available data and can be used to accurately predict the drought grade of summer maize in anhui province, indicative of the actual local situation. a series of insurance coverage standards are proposed based on the summer maize drought weather index, and a huaibei region summer maize premium rate of 237 rmb/ha is suggested. the results of this study also show that the volume of insurance units up to a level of 4,620 rmb/ha can be maintained when a higher than 40% summer drought index is seen. relevant insurance probabilities are also revised in this 21 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 study, based on the soil relative humidity during the initial insurance period of drought. the results of a pilot study on summer maize drought index insurance carried out in august 2008 within the city of huaibei in suixi county, anhui province, confirm that this product does accurately reflect actual risk. acknowledgements this research was supported by the national natural science foundation of china (grant no. 41571018). references 1. antón, j., cattaneo, a., & kimura, s. et al. (2013). agricultural risk management policies under climate uncertainty. global environmental change, 23(6): 17261736. doi.org/10.1016/j.gloenvcha.2013.08.007. 2. binswanger-mkhize, h. p. (2012). is there too much hype about index-based agricultural insurance? journal of development studies, 48(2): 187-200. doi: 10.1080/00220388.2011.625411. 3. blanc, e. (2012). the impact of climate change on crop yields in sub-saharan africa. american journal of climate change, 1(1): 1-13. doi: 10.4236/ajcc.2012.11001. 4. bobojonov, i., awhassan, a., & sommer r. (2014). index-based insurance for climate risk management and rural development in syria. climate and development, 6(2): 166-178. doi.org/10.1080/17565529.2013.844676. 5. cao, j., li, s., & li, y. (2013). design of products for winter wheat agro-meteorological index insurance in jiangsu province. international journal of hybrid information technology, 6(5): 37-48. doi: 10.14257/ijhit.2013.6.5.04. 6. chen, w., hohl, r., & tiong, l. k. (2017). rainfall index insurance for corn farmers in shandong based on highresolution weather and yield data. agricultural finance review, 77(2) 35-41. doi.org/10.1108/afr-10-20150042. 7. coble, k. h., hanson, t., & miller, j. c. et al. (2003). agricultural insurance as an environmental policy tool. journal of agricultural & applied economics, 35(2): 391-405. doi: 10.1017/s1074070800021350. 8. finger, r., pelka, n., & musshoff, o. (2014). hedging effectiveness of weather index-based insurance in china. china agricultural economic review, 6(2): 212-228. doi: 10.1108/caer-11-2012-0124. 9. jin, j., wang, w., & wang, x. (2016). farmers’ risk preferences and agricultural weather index insurance uptake in rural china. international journal of disaster risk science, (4): 1-8. doi: 10.1007/s13753-016-0108-3. 10. turvey, c. g., & kong, r. (2010). weather risk and the viability of weather insurance in china's gansu, shaanxi, and henan provinces. china agricultural economic review, 2(1): 5–24. 11. leblois, a., & quirion, p. (2013). agricultural insurances based on meteorological indices: realizations, methods and research challenges. meteorological applications, 20(1): 1–9. 12. liu, y. n., he, w. l., & li, y. l. et al. (2010). a study on the risk index design of agricultural insurance on apple florescence freezing injury in the shaanxi fruit zone. chinese journal of agrometeorology, 31(1): 125-124. (in chinese) doi:10.3969/jissn.1000-6362. 13. lou, w. p., ji, z. w., & qiu, x. f. et al. (2011). design of weather index insurance contact for tea frost. journal of natural resources, 26(12): 2050-2060. (in chinese) doi: 10.11849/zrzyxb.2011.12.005. 14. lu, y., ramirez, o. a., rejesus, r. m., & knight, t. o. et al. (2008). empirically evaluating the flexibility of the johnson family of distributions: a crop insurance application. agricultural & resource economics review, 37(1): 79-91. doi: 10.1017/s1068280500002161. 15. makki, s. s., & somwaru, a. (2001). evidence of adverse selection in crop insurance markets. journal of risk and insurance, 68(4):685-708. doi: 10.2307/2691544. 16. ma, x. q., sheng, s. x., & zhang, a. m. (2001). influence of precipitation on drought developing in north of the yangtze river in anhui. agricultural meteorology, 22(3): 15-19. (in chinese) 17. ma, x. q., wu, w. y., & zhang, h. (2008). analyzing spatial and temporal variations of agricultural drought and waterlog in jianghuai area through accumulated humidity index. resources science, 2008, 30(3): 371-377. (in chinese) 18. mao, y. d., wu, l. h., & miao, c. m. et al. (2007). a reference design for citrus freeze damage insurance by using meteorological index in zhejiang province. chinese journal of agrometeorology, 28(2): 226-230. (in chinese) 19. mornet, a., opitz, t., & luzi, m. et al. (2015). index for predicting insurance claims from wind storms with an application in france. risk analysis, 35(11): 2029-2056. doi: 10.1111/risa.12395. 20. ozaki, v. a., & silva, r. s. (2009). bayesian ratemaking procedure of crop insurance contracts with skewed distribution. journal of applied statistics, 36(4): 443-452. doi: 10.1080/02664760802474256. 21. panda, a., sharma, u., & ninan, k. n. et al. (2013). adaptive capacity contributing to improved agricultural productivity at the household level: empirical findings highlighting the importance of crop insurance. global environmental change, 23(4): 782-790. doi: 10.1016/j.gloenvcha.2013.03.002. 22 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 22. sheng, s. x., zhang, j. j., & wang, x. d. (2014). spatialtemporal distribution of climatic suitability of summer maize in anhui province. geographical research, 35(1): 1467-1476. (in chinese) doi: 10.11821/dlyj201408007. 23. skees, j. r., 2008. challenges for use of index-based weather insurance in lower income countries. agricultural finance review, 68(1): 197-217. doi: 10.1108/00214660880001226. 24. thomas, d. s. g., twyman, c., & osbahr, h. et al. (2007). adaptation to climate change and variability: farmer responses to intra-seasonal precipitation trends in south africa. climatic change, 83(3): 301-322. doi: 10.1007/s10584-006-9205-4. 25. xu, y., ma, x.q., & yue, w. (2009). study on quantitative assessment of water logging disaster losses of single-season rice in anhui province. journal of anhui agricultural sciences, 37(2): 722-726. (in chinese) doi: 10.13989/j.cnki.0517-6611. 26. yang, t. m., liu b. c., & sun, x. b. et al. (2013). design and application of the weather indices of winter wheat planting insurance in anhui province. chinese journal of agrometeorology, 34(2): 229-235. (in chinese) doi: 10.3969 /j.issn.1000-6362. 27. yang, t. m., xu, y., & sun x. b. et al. (2016). design and application of the drought weather index insurance of summer maize in anhui province. meteorological monthly, 42(4): 450-455. (in chinese) 28. zhang, j., zhang, z., & tao, f. (2017). performance of temperature-related weather index for agricultural insurance of three main crops in china. international journal of disaster risk science, 8(1): 78-90. doi: 10.1007/s13753-017-0115-z. 29. zhang, z. (2014). spatial pattern and decadal change of agro-meteorological disasters in the main wheat production area of china during 1991-2009. journal of geographical sciences, 24(3): 387-396. doi: 10.1007/s11442-014-1095-1. 23 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 14-23 asc 2014 paper-gomal zam dam.docx the risk of power imbalance in project delivery: a study of large victorian public infrastructure projects hamzeh zarei, felix kin peng hui, colin duffield department of infrastructure engineering, university of melbourne parkville, victoria, 3010, australia e-mail: hamzeh.zarei@unimelb.edu.au, kin.hui@unimelb.edu.au, colinfd@unimelb.edu.au george wang department of construction management, east carolina university greenville, north carolina, 27858, usa e-mail: wangg@ecu.edu abstract in large public infrastructure projects, political risks due to the power imbalance between central and delivery agencies are often overlooked or underestimated. the primary motive of the delivery agency in distorting information for political gains should be deemed a risk that creates uncertainty for large projects planning the outcome. in this study, seven large infrastructure projects in the state of victoria, australia are examined through a workshop involving key stakeholders who had played active roles in these projects. the findings revealed that power asymmetry between central and delivery agencies exist and would lead to optimism bias, which in turn creates uncertainty and risk of overpromising in the business case. power asymmetry exist in large infrastructure projects because the central agencies usually only have the responsibility but not the skill set needed to measure the robustness of the business case. these types of political risks are difficult to quantify and even detect. this paper recommends a few managerial strategies that have referential values and/or can be used to mitigate and circumvent this risk. keywords: political risks, power asymmetry, infrastructure projects 1. introduction risk assessment is an essential part of an organization’s strategic planning process. in aiming to achieve its strategic objectives, risk evaluation in the strategic planning process normally calls for some risk reporting taking into account normal and worstcase scenarios. managerial decisions are then formed, and a risk management framework developed to retain a “no surprise” project operation status. risk evaluation is also a vital part of project portfolio selection when a project is appraised for the risk and value in comparison to other initiatives. due to limited resources, only the best value for money should permit to proceed, and hence robust risk assessment is a critical step in short listing and choosing the projects for investment. risk can come in the form of external and internal risks. external risks are the risks caused by elements external to the organization such as economic downturns and natural disasters. traditionally a special category of external risk known as political risks specifically, have been largely ignored and only considered informally by project managers. examples of political risk include external disruptive behavior that can deeply affect project cost, time and performance. however, it was difficult to quantify political risks due to their ever-changing nature. a common practice for accounting this type of risks would be to use an activity-based costing technique that increases the variance while estimating the performance of the activities in the project. however, this method is not robust or sufficient; probabilistic risk analyses may be better options to deal with political risks. the specific aim of this research is to explore and investigate political risks in large infrastructure received 9 may 2017 accepted 26 may 2017 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 53 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). projects using cases of large victorian public infrastructure projects as our basis for examination. 2. literature review 2.1 accounting for political risks past studies previous research in this area includes the neelum jhelum hydroelectric project (njhp) (kiani, 2014; mukhtar, 2013). in this study, the authors tried to evaluate whether project managers working on large infrastructure projects in a volatile environment can have sufficient objective information in order to be able to utilize any form of risk modeling techniques. the political risk is defined, in the study as any changes resulted from the hierarchy of control that alters the expected outcome and value of a given economic action by changing the probability of achieving business objectives. this massive infrastructure project is carefully chosen using flyvberg’s (2016) argument that a carefully chosen experiment, case or experience can lead to the critical situation, where one case may be able to provide sufficient evidence to justify an argument. neelum-jhelum hydro power plant (njhp) is a run of the river hydroelectric generation project that is located in the vicinity of muzaffarabad in the territory of azad jammu and kashmir (aj&k) in pakistan. this megaproject diverts the neelum river water through a tunnel, into a powerhouse and out into the jhelum river. the project experienced significant delays, funding issues and political disputes with regards to the indus river treaty signed between india and pakistan. based on the detailed project study and the specific project accounts, twenty political risk factors were found for the njhp – some of which had caused an effect on the cost, time and performance. objective project information was then obtained to model these risks using a probabilistic model. the probabilistic risk assessment utilized two core principles: the principle of decomposition and bayes probabilistic principle. it was outside of the scope of this research to determine the presuppositions and conditional probability data for a full set of scenarios and the posterior probability of the identified risks. by implementing the probabilistic model and costs derived for each of the identified risks, a total amount of approximately $596 million usd was determined to be the impact of the identified risks on the $5 billion usd project. so were the managers able to obtain all the relevant objective data to run the probabilistic models? a closer look revealed that the majority of the information was derived from experts, who were on-site and had been on the project since its beginning. an analysis of the impact of subjectivity in the data provided by experts showed that the probabilistic model loses its robustness and ability to assist project managers in real life – shown in the study by implementing a small change to conditional probabilities during the different design phases, resulting in a change of over $250 million usd of cost variance. the answer to the research problem is that, although conditional and a priori, the probabilities were being determined with experts are at best a guess – simply due to the nature of political risks. the authors then went on to conclude that the vast majority of objective data, such as costs, delays and impacts can be obtained objectively for the use of a probabilistic model to predict the impact to a project from political risks, the requirement of conditional and prior probabilities from expert opinions cannot be considered objective data, resulting in a model that may or may not be entirely realistic. 2.2 agency theory revisited agency theory (mitnick, 1975) attempts to explain the relationship between principals (defined as the entity with control over resources and legitimacy to make a decision including approval of a proposed solution) and the agent or the contractors (defined as a delegate of the principal to perform a task). the agent is expected to have knowledge and skills to explore the need, identify the solution and implement the work with the highest value for the principal and is also supposed to offer a solution for principal’s approval. eisenhardt (1989) provided an overview of the agency theory by stating that principal-agent relationships should reflect efficient organization through information sharing and risk-bearing costs. key assumptions are made about human nature being self-interest, bounded rationality, and risk aversion. the gap of information and incentive between principal and agent results in problems such as risk sharing, relationships with differing goals and risk preference. the focus of the theory is determining the most efficient contract governing the principal-agent relationship given assumptions about people (e.g., self-interest, bounded rationality, risk aversion), organizations (e.g., goal conflict among members), and information (e.g., information is a commodity which can be purchased). agency theory assumes the pursuit of self-interest at the individual level and goal conflict at the organizational level (march, 1962). in both perspectives, information asymmetry creates a power in the lower order participants (pfeffer& pettigrew, 1976). agency theory is a kind of information processing approaches that assumes journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 54 individuals are rational and that information is distributed asymmetrically throughout organizations contingency theory (chandler, 1962; galbraith, 1973; lawrence &lorsch, 1967). in political models, goal conflicts are resolved through bargaining, negotiation, and coalitions however in agency theory they are settled by the co-alignment of incentives using the price mechanism (eisenhardt, 1989) or reward power in the contractual framework. agency theory is most relevant in situations in which contracting problems are difficult. these include situations in which there is (a) substantial goal conflict between principals and agents, such that agent opportunism is likely (e.g., owners and managers, managers and professionals, suppliers and buyers); (b) sufficient outcome uncertainty to trigger the risk implications of the theory (e.g., new product innovation, young and small firms, recently deregulated industries); and (c) unprogrammed or team-oriented jobs in which evaluation of behaviors is difficult (eisenhardt, 1989). in some cases, it is difficult or expensive for the principal to verify what the agent is doing or whether the agent has behaved appropriately. there is also a problem of risk sharing which arises when the principal and agent have different attitudes toward risk, and this results in different actions (eisenhardt, 1989). agency theory broadened this risk-sharing literature to include the agency problem that occurs when cooperating parties have different goals and division of labor (jensen & meckling, 1976; ross, 1973). researchers are divided on the usefulness and application of the agency theory. jensen (1983) argued that agency theory provides an important foundation while others have argued that the theory address no clear problem, is narrow, lacks testable implications, and is dangerous (perrow, reiss, & wilensky, 1986). agency theory has been critiqued to having only a view of the world that ignores the complexity of organizations. another view states that agency theory provides a unique, realistic, and empirically testable perspective on problems of cooperative effort (eisenhardt, 1989). 2.3 power being identified as a universal driver (nietzsche, 1968), power is a fundamental concept in political science (falkemark, 1982; isaac, 1987). a common definition of power comes from the causal relation between two entities one desires to influence, and one is affected (isaac, 1987). “power is the ability of those who possess it to bring about the outcome(s) they desire” (salancik & pfeffer, 1974). such power could extend over a broad span of means and leverage embedded in an actor's potential activities, called power field (kurt lewin, 1935). according to raven (1993), lewin (1952) defines power as the ability to induce forces' of a certain magnitude on another and is a fundamental phenomenon to explicate actors’ behavior during interactions. power is defined as the capacity to influence the conduct of others. it comes with exclusive access to resources less available to other e.g. access to information, knowledge, skills and expertise. the classic theory of power identifies expert power, reward power, referent power, legitimate power, and coercive power as the basis of power (french & raven, 1959). the model was then completed by informational power as the sixth bases of power (raven, 2008). power aims to explain the intention of the powerful in changing the behavior of others. it originates from information, knowledge, skills, expertise, and ability to reward/punishment, rightful position, reputation and force. means of exertion of power are diverse and changes according to the base of power. although the theories of power have been cultivated in the context of political science, it explains the means of compliance and influence in social systems (dahl, 1957; falkemark, 1982; isaac, 1987; lukes, 1974). in behavioral compliance or ‘decisionist’ view “a has power over b to the extent that a can get b do something that b would not otherwise do” (dahl, 1957). when power is viewed as a ‘non-decision’ influence or an invisible influence (szalai, 1997), there will be “mobilization of bias”. bias emerges when something favors a few, but suppress others (schattschneither, 1960). “[mobilization of bias is] is a set of predominant values, beliefs, rituals, and institutional procedures that operate systematically and consistently to the benefit of particular groups and persons at the expense of others” (szalai, 1997). pfeffer (2005) also distinguishes power from authority. in his view, power suggests hard power that its exertion diminish it, while authority is a legitimate position that might even reinforce when applied (pfeffer, 2005). it could be said that authority is a power sourced from legitimacy and strengthened by reward and punishment. nevertheless, the classic theory of power identifies and clusters the origin of power that explains the influence of actors more cohesively that acknowledges the variety of bases of power. we take the classic theory of power as the cornerstone of our research. measurement of power is controversial and problematic as the precise measurement is not possible (pfeffer, 2005). when used between a pair of entities, power is a relative term. it does not exist in solitude. that is to say, the power of a powerful journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 55 resonates with the perceived picture of others. in a relationship between a principal and an agent, the perimeters of their behavior will set the scope of power that each party might have over the other. not only the conduct of actors may severely change under observation, but a proper study also requires two sets of data for actor behavior when power exerted and when power not exerted that is hard to structure and collect. 2.3.1 power in project delivery in a project, stakeholders have a different level of influence to impact the project (cleland & ireland, 2007). stakeholder salience theory considers power an indicator of the stakeholder influence (mitchell, agle, & wood, 1997). kernaghan (1993) defines stakeholder relation to a project “a relationship involving sharing of power, work, support and information with others for the achievement of joint goals and mutual benefits”. the participants’ ability to influence the project decision-making process is laid in their power comprises proficiency as well as influence. the term ‘powerful’ in projects also implies competency, skillfulness, expertise, experience, qualification and adequacy. a knowledgeable consultant has an impact on a client, and a skilled contractor has the power to demand a higher price. in projects, power is not only perceived as the ability to change other’s decision but the competency that convinces others to comply. that is to say; power is a combination of authority and competency in project systems. a right mix of project people competency and motivation moreover to well-empowered project parties have a significant effect on success (ayas, 1996; nagadevara, 2012). that is to say, participants’ power has a great role in project outcome through the effect it has on their communications and collaborations within the project that is reflected in their decisions. the classic theory of power recognizes knowledge, skills, reputation, reward, legitimacy and coercion as origins of power. we try to map the six sources of power classic theory to authority and competency as the major pillars of power in projects. authority resonate legitimacy, reward and fuels coercion power. on the other hand, competency relates to knowledge and skill power. it is rather puzzling how to associate reputation, as a source of power, with authority or competency. reputation is not an actual characteristic but a perceived effect on others in a long-term collaboration. reputation or referent power as means of influencing decision should be investigated in the context project delivery to explore cause and effects of any possible risk it may impose on the project actors’ behavior. 2.3.2 the concept of power asymmetry there is a link between the utility of decisions and the cognition reflected in the level of knowledge and required resources to make an informed choice. besides, there is no guarantee that the degree of authority and cognition in participants are in the balance when the opposite is suggested by literature when relates the irrationality to power (flyvbjerg, 1998). power has been blamed for being the cause of optimism bias (inesi, 2010; tost, gino, & larrick, 2012), over-confidence (fast, sivanathan, mayer, & galinsky, 2012), constraint ignorance (lammers, gordijn, & otten, 2008), and lack of advice taking (see, morrison, rothman, & soll, 2011). actors’ power in a project should be investigated as it originates the behavior and hence the quality of decisions. participants’ power is an undividable part of any project, and its role should not be overlooked in project success. while power is the ability to influence other’s decision, it is the capacity of one to fulfill the allocated task. power asymmetry or imbalance exits when one party has more power than its working partners. in projects, a lack of in-house knowledge in public sector was a serious problem that contributes to time and cost overrun (patel & robinson, 2010). competence is a collection of knowledge, attitude, skills, and experience needed to successfully perform a function (international project management association, 2006). participants have a different level of competency in a project. inappropriate level of competency hinders effective communication, cooperation and flexibility to change; new ideas are ignored, and collaboration is diminished. a possible extreme scenario is where project clients have supreme authority but inadequate resources to make an informed decision. the gap between participants’ authority and competency prevent an optimum utilization of resources or maximized decision. the outcome would be a ‘limited’ decision (sanderson, 2012) that occurs where either all options are not identified, or the best possible option is not preferred amongst them. 2.3.3 power in central government and delivery agencies central agencies in government organizations typically only have executive powers to implement and enforce governmental decisions. central agencies may or may not have the expertise to decide and the control of resources at the grassroots level or the delivery level. it is the essence of delegation process journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 56 that a principal asks agencies to do a task that requires ad-hoc skills and knowledge. the central agency finds it inefficient and burdensome to keep all the necessary resources in-house but instead, to breakdown the work and allocate it to delivery agencies. nonetheless, in a delegation, an implicit form of influence exists within approval process. informal power is thought to be held by delivery agencies due to the knowledge, expertise and referent power the delivery agency has over the central agencies that will influence the approval of the proposed business case. there is, therefore, an element of exposure to the risk arising from the informal power that resides in these delivery agencies. a closer look at the arrangement reveals that the power balance may influence information flow, decision making and dynamics between policy makers and project delivery team. while the central agencies use the business case as a means of evaluating the value of an investment proposal over other candidates, the robustness of the business case becomes important as it may influence the selection process. that is to say; delivery agencies may affect the content of business cases which is the primary form of information flow used for decision making. it is a critical risk if the business case may be made to look more optimistic to push the project through the initial approval gates. power comes from the delivery agencies close interaction and engagement with two first groups of agents: (a) users i.e. these are the people who finally use the infrastructure; (b) market i.e. these are the individuals or groups who design, who constructs and who supplies parts and components to bring the project to fruition. the superior knowledge of a delivery agency of user’s requirements and market condition provide higher information power to delivery agencies in its communication with central institutions. being close to the users and market brings a particular type of advantage in that the delivery can choose to withhold information from the agency or embellish the information or even to distort the information to their organization advantage or to enhance their positions. although they are expected to act on behalf of the central agency, organizational need for survival and their ignorance of the government priorities of resource allocation may make them a bias toward their proposed business case. the approval process inevitably faces a risk of optimism in a proposed business case where the central agencies rely on a delegation process of delivery agencies prepare and propose a business case in line with government strategies. in effect, acting on behalf of a higher authority brings about the risk of optimism bias and over promising in proposed business case that might lead to opportunity loss or lesser value for money. 3. research method in the first part of this research, publicly available parliamentary interviews with experts from government and private sector were indexed and analyzed to find the current challenges in the delivery of public infrastructure. the analysis allowed the root causes of the current shortcomings to be uncovered. grounded research method and case study are used to assess the risk of over-promising in victorian infrastructure delivery. we incorporated focus group discussions to capture the process of decision making in the different stages of a project. a focus group discussion was then organized. the participants for the workshop were drawn from the people from the public and the private sector in victoria. the aim of the seminar was too organized to fine-tune and validate the previous findings. during the workshop, the identified challenges, causes, and recommendations are presented and discussed. the workshop was held in march 2015 at a convenient location at a university premise. forty-four experts were contacted via telephone, 26 were formally invited, 19 expressed their interests, and finally, 17 attended the workshop. table 1.a shows the affiliated sector of the seminar delegates and table 1.b depicts their depth of knowledge of the case study. during the workshop, the experts’ feedback was collected in every step. at the same time, seven case projects were presented, and learning lessons are solicited. the workshop was conducted by a professional moderator. specialized software was used to capture the expert’s statements (recording was evaded to encourage candid disclosure of participants). 4. findings the workshop reviewed seven victorian case projects. project approval process is considered in a broader framework of central and delivery agencies. in an extension to the literature of organizational theories e.g. agency theory, the research splits up central and delivery agencies. it provides a better understanding of the current state of public procurement and highlights the existing organizational challenges such as asymmetry of power and conflict of incentives. the findings verify that underestimating the overpromising behavior of project delivery agencies during the planning stage of infrastructure is a critical risk that would impact the perception of success. it is postulated that the risk of journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 57 table 1.the demography of the workshop delegates a. affiliated sector; b. knowledge of the studied project cases a. government 33% private sector 42% academic 17% other 8% b. extensive 54% some 31% peripheral 15% optimism bias could be attributed to the power balance of the participants. power imbalance or asymmetry can be linked to the present shortcomings in public infrastructure by uncovering the people’s behavior within project processes. whereas the common understanding of the literature implies that success is an outcome of project performance according to the plan, the majority of participants, 58%, believe that success is mainly measured in the shadow of the initial expectations. more remarkably, 64% of experts find inflated expectations the leading cause of disappointment in projects. optimism bias in planning the value and overpromising behavior of sponsors are the key factors in deterring a project from success. the findings also endorsed the risk of overambitious decisions particularly those made of a business case. the quality of the business case is severely affected by client’s fallacy of initial estimations especially if exaggerated benefits or optimistic cost are prime considerations. many other factors were also found to be a possible cause of the over-promising behavior. the insufficient understanding of requirements, errant assumptions, inadequate technical knowledge, strategic misrepresentation and insufficient supervision are identified as leading factors that might cause a project bite off more than they can chew. the workshop outcomes confirm the initial hypothesis that the distribution of power among participants explains the risk of bias in preparing a robust business case. suboptimal decisions are partially attributed to the participants’ conflicting interests when concurred with asymmetrical power. there is a risk of overpromising business case when resources are constrained and governed by central agencies, but the field knowledge of user requirements and market data rest with the delivery agencies. if the central agency is less informed of the decentralized project requirements, the risk becomes more severe. 4.1 informal authority in this research, the notion of informal authority is conceptualized to explain the agency’s referent power to manipulate and influence principal’s decision. informal authority contrasts the formal authority, a known legitimacy power of an entity that possesses it lawfully as a form of authority to make a decision. in public delivery system, the central agencies have the formal authority because of their given position in the approval process such as gateway review. central agencies allocate resources and control the progress. conversely, informal authority is a tacit power possessed by an entity, usually a subordinate or a reporting organization. in other words, informal authority influences a formally authorized entity’s decision. in projects, the informal authority may originate from information, expertise or reputation power of an agency to manipulate other organizations’ behavior. every time a business case is found trustworthy the reputation of the delivery agency increases. the more the reputation of the delivery agency, the more the informal authority due to the increase of reputation power. as a conclusion, the concept of informal authority is relevant to the organizational challenge of victorian public delivery, and so it is embedded in the theoretical framework of our proposed theory. 4.2 power and optimism bias the intrinsic worth of power asymmetry theory is cross-checked against seven recent case projects in victoria. these case studies are selected across the extreme range of performance. through a series of questions, the participants evaluated the level of success whereas informal authority. failure is measured through questioning the experts of the gap between expectations and journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 58 outcome. it shows a mix of perceived success and failure among the case studies. the popular projects in the eyes of experts are convention centre, children hospital, and regional rail link. on the opposite, myki, healthsmart and market relocation are deemed disappointing; finally, victorian desalination is in limbo. it was found that informal authority exists among the central and delivery agencies. it is partly due to the governance structure frequently adopted in australia where delivery agencies are delegated to communicate the field data. the role of central agencies is merely direct and audit. the information gap and distribution of skills among central and delivery agencies creates an informal authority for the delivery agencies that might create a risk to the project. it appears that if a delivery agency has an accurate understanding of the requirements, risks may be minimized; see children hospital. the contract structure of children hospital project is ppp (public private partnerships) that allocate the risk of cost overrun and time delay to the private sector. the procurement strategy for this project plays an active role to provide a robust project estimation. a solid understanding of the project and user requirements provides unswerving information to the central agency to make a decision that preserves value for money. an adequate understanding of the requirements, make a client an informed buyer that profoundly contributes to project success. in the view of stakeholders and experts, the risk of informal authority and failure are connected. the correlation factor of 0.71 shows that the higher informal authority, the greater risk of a failure; see figure 1. the data shows that informal authority is not unavoidable nor does it always lead to a failure. however, it might be suggested that there is a threshold that beyond which the risk of manipulation is high and the disaster is probable e.g. myki and healthsmart. within the threshold level, the central agencies' understanding of the requirements is a key to defusing the threat of informal authority. informal authority if not managed through another mechanism of informing the decision maker e.g. proper procurement strategy, will cause risk of failure. among the case studies, those show high informal authority are ended with a perception of failure. 5. discussion 5.1 sense of power contribute to optimism bias in our study of victorian large infrastructure projects, we recognized the peculiar role of central agencies figure 1. informal authority vs. failure in seven victorian infrastructures; correlation factor = 0.71 that hold the formal authority and approved the funding for proposed project i.e. the cabinet and the treasury, or retrospectively audit project delivery process i.e. the parliament and auditor-general that provide specialized audit and reports to the central agencies. central agencies make the final decision and tend to have an over-reliance on the delivery agencies for developing a business case. the delivery agencies, on the other hand, are expected to have field knowledge while being in relation with end users and market entities. they initiate or develop a business case for the central agencies that compete for the limited funding and approvals. the race for limited resources provokes informal authority that raises the power of the delegate in swaying the final decision. the relationships between the agencies are depicted in figure 2. the issue of informal authority introduces risk into the system as central agencies do not have the first-hand information and expertise. emerging power is masqueraded in this informal authority as the delivery agencies now have influence over final approval decision through manipulating the quality of the business case. they have the power to make the business case optimistic and more appealing to the central agencies either intentionally in a strategic misrepresentation of swaying the approval process or unintentionally as a result of a psychological process of discounting worst case scenarios or absence of realistic assumptions. 1 2 3 4 5 6 7 8 9 10 c o n ve n ti o n c e n tr e h e a lt h s m a r t m a rk e t r e lo ca ti o n c h il d re n 's h o sp it a l m y k i v ic to ri a n d e sa li n a ti o n r e g io n a l r a il l in k n um be r o f r es po ns es perception of failure risk of informal authority journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 59 5.2 uncertainties and risks arising uncertainty risk stems from one party holding more power and influences others. these results in the uncertainty (risk) those results in 1) an overpromising behavior and 2) under-delivering when constraints cannot be overcome. due to the lack of knowledge and expertise, central agencies may not have a sound check and balance system in place as they cannot control due to not physically present all the time. secondly, they are continually relying on the delivery agency for information and advice. when the business case is made too optimistic due to a motivation of getting initial project approval and funding, the chance of approval increases. it leads to a kind of overpromising behavior which introduces uncertainty and risk where resources are limited, and investment opportunities substitute each other. 6. managerial implications and significance of the research projects with a known boundaries and accurate estimation models, such as buildings, are less risky than those of complex interfaces with latent conditions such as it, earthworks, or unpredicted market demand. furthermore, a delivery agency with an on-going stream of projects accumulates experiences and gets a true understanding of the user requirements. that might explain why a department that routinely delivers hospitals has a higher chance of being successful, even if it has a medium level of informal authority. nonetheless, informal authority remains a threat for early stage decisions particularly if the delivery agency is not an informed buyer. even a reliable and positive record of a delivery agency in fulfilling expected value for money is not enough to assure the central agencies as it raises the reputation of the delivery agency that sources informal figure 2. power fields of public agencies in infrastructures delivery systems. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 60 authority. a change in the approval process is required to deal with this uncertainty. to mitigate this risk, different strategies to manage participants’ risk of power asymmetry are proposed. three strategies are found more effective to control the risk of planning fallacy by balancing the participant power. these strategies are detailed in subsections 6.1 to 6.3 below. 6.1 state-wide centre of expertise lack of adequate expertise in the public sector is a danger in infrastructure delivery. not only there are examples of the public sector being incompetent in addressing the proper requirements of a project, but there is also a growing concern for managing and preserving the existing corporate memory. therefore, external expertise is preferred and called upon by the delivery agencies find outsourcing an efficient approach to access occasional skills. consequently, lack of in-house expertise is more likely in the major one-off projects that require unique know-how normally non-existent in agencies’ skill set. although it might not be efficient for every deliverer to keep a separate skill set, having a statewide center of excellence would be an answer that has been applied in other jurisdictions too e.g. ontario, canada as it engages in every large project on a frequent basis. a center of expertise is proposed. the aim of this center of expertise is to create knowledge, implement large projects, maintain public competency and accumulate the experience. such a center will benefit delivery agencies as well as central agencies by providing a reliable benchmark in assessing the optimism in a business case. 6.2 stable delivery pipeline although central public agencies have the authority, usually by people’s election, to direct public investments, they are political entities with political gain as the motive and incentive. the appetite for political gain is a pitfall that might be abused if a business case targets to stimulate it by overpromising that intrigues central agencies for a higher political weight. public infrastructures are typically large, and their delivery cycle is likely to be lengthier than a political cycle. that is to say; a government might not last long enough to witness the end product of an incepted infrastructure. the longer life cycle of infrastructure delivery than political elections may create a risk of a government becoming bias toward early benefits of a project that could be projected as a political gain. investment priorities and project pipeline are strategies to emphasize the long-term value of a project for the community. a steady delivery pipeline has the advantage of retaining public skills in-house or helps outsource it more efficiently. a delivery pipeline provides ample opportunity for stakeholders to engage and share their requirements. a qualified public agency with sufficient understanding of user requirement has a higher chance of success. this is also true in the private sector where a reliable schedule of infrastructure facilitate skills management is beneficial. private sector being already informed of the project queue, organize the necessary resources. it collectively increases competition and eventually benefits the community. even though it is accepted that a delivery pipeline should be kept away from short-term political influence, the legitimacy of a long-term isolated pipeline is under question too. an elected government set forward to fulfill the people’s direct interests. it is easily conceivable that a plan of a voted government might be different to those of a long-term infrastructure pipeline. it remains to be investigated the optimum mechanisms of a government to influence the infrastructure pipeline. 6.3 early engagement of stakeholder through financial commitments for some stakeholders, such as end users of an infrastructure, the first engagement might occur the first day of the project operation. that is to say; these secondary stakeholders feel no legitimate link, no urgency and no power to influence the project in the early days. nonetheless, the impact they receive from the project is substantial. that might explain why social infrastructure is prone to create a controversy just close to the operation day. early engagement of dormant stakeholders may stimulate streams of communication that assist transferring the stakeholder requirements when the cost of change is reasonably low. users may have less incentive to engage since they underestimate the value of early decisions, and also they are not contractually bonded with no authority to raise a voice. infrastructure users should be given an opportunity to express their expectations and advise their requirements. this objective is likely if the stakeholder has the right and need of engaging with project process. while fees traditionally start after the operation, infrastructure users might be summoned to discuss the payment regime long before the asset is delivered. a form of financial commitment ahead of product delivery would intrigue the stakeholder and attract journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 61 their attention. this financial structure could be optimized to raise the voice of silent stakeholders. scope and requirements might change in early stage, or the investment might fail the test of feasibility, the early financial engagement of the stakeholders should carefully avoid to inspire any sense of guarantee to the project team and to hold ready a mechanism of snap back. 7. conclusion governing a construction project includes many constituencies working together. public projects are defined and proposed in the context of political interaction of government agencies and the private sector. breaking down the government agencies into delivery agencies and central agencies may unravel the complexity of decision processes. the behavior of agencies in decision and approval process determines the quality of planning and ultimately the project outcome. during the planning stage, a business case defines the project scope and conceptual design that affects the project perception of success in long term. while the quality of business case is the cornerstone of project success, optimistic plan often ends up with a perceived failure. the risk of influence over the approval process may hinder the government achieving the expected value for money. agency theory assumes an information and incentive gap between principal and delegates. the theory tries to explain shortcomings in the quality of early decisions when the agencies may not follow the interests of the principal as it would follow their own. in public infrastructure, early decisions are made by a business case that a delivery agency proposes to central agencies for review and approval. in the presence of misaligned objectives of agencies and unbalanced distribution of information, the delivery agency may manipulate the process in favor of its organizational preferences in preparing the business case such as scenario making and analysis, estimation of time and cost, identifying and analysis of project options, and selection of procurement strategy. power plays in any multiplayer settings. power comes from an efficient access or exclusive control over relevant resources and is always a perceived concept rather than an absolute term. principal’s power is the ability to align the agent’s action with the principal value vis-à-vis the agent power is the ability to influence the principal decision. the classic theory of power identifies knowledge, skills, reputation, legitimacy, and force as sources of power. authority and competence are the two axes of power in projects. bringing the classic theory of power to the context of projects shows that reputation (or referent) is an informal power that emerges and develops in a long-term collaboration. we have proposed a concept of informal authority to explicate the risk of overpromising as a behavior that delivery agency exhibits during the approval process. in a visual framework that graphically illustrates the power field of project agencies, the dissimilar perspectives of central and delivery agencies are studied and the risks of making suboptimal decision investigated. it was observed that decision makers’ power balance is a significance contributor to the quality of early decisions. in a dedicated workshop attended by project associates, we collected stakeholders’ perspective on seven infrastructure projects in the state of victoria, australia. the perception of failure and the informal authority are measured, and the correlation is indicated. expectedly, informal authority exists across the case studies but remains benign when low or when the buyer is informed through other mechanisms such as procurement strategy. in the case of a significant informal authority, the risk of failure is high. references 1. ayas, k. (1996). professional project management: a shift towards learning and a knowledge creating structure. international journal of project management, 14(3), 131–136. https://doi.org/10.1016/ 0263-863(95)00080-1. 2. chandler, a d. (1962). strategy and structure: chapters in the history of american enterprise. massachusetts institute of technology cambridge, 349–407. 3. cleland, d., & ireland, l. (2007). project management strategic design and implementation. 4. dahl, r. (1957). the concept of power. behavioral science, 2(3), 201. https://doi.org/10.1002/ bs.3830020303. 5. eisenhardt, m. (1989). agency theory: and assessment review. the academy of management review, 14(1), 57–74. 6. falkemark, g. (1982). power, theory and value. cwk gleerup. 7. fast, n. j., sivanathan, n., mayer, n. d., & galinsky, a. d. (2012). power and overconfident decisionmaking. organizational behavior and human decision processes, 117(2), 249–260. https://doi.org/10.1016/ j.obhdp.2011.11.009. 8. flyvbjerg, b. (1998). rationality and power: democracy in practice. university of chicago press. 9. flyvbjerg, b. (2016). five misunderstandings about case-study research. qualitative inquiry, 12(2), 219– 245. https://doi.org/10.1177/1077800405284363. 10. french, j. r. p., & raven, b. (1959). the bases of social power. in d. cartwright & a. arbor (eds.), studies in social power (pp. 150–167). university of journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 62 michigan press. 11. galbraith, j. r. (1973). designing complex organizations. organization development. https:// doi.org/10.2307/248839. 12. inesi, m. e. (2010). power and loss aversion. organizational behavior and human decision processes, 112(1), 58–69. https://doi.org/10.1016/ j.obhdp.2010.01.001. 13. international project management association. (2006). ipma competence baseline. (g. caupin, h. knoepfel, g. koch, k. pannenbäcker, f. pérez-polo, & c. seabury, eds.) (3rd ed.). international project management association. 14. isaac, j. (1987). power and marxist theory. cornell university press. 15. jensen, c., & meckling, h. (1976). theory of the firm: managerial behavior, agency costs and ownership structure. journal of financial economics, 3, 305 –360. https://doi.org/10.1016/0304-405x(76)90026-x. 16. jensen, m. c. (1983). organization theory and methodology. the accounting review, 58(2), 319– 339. https://doi.org/10.2307/24683kernaghan, k. (1993). partnership and public administration: conceptual and practical considerations. canadian public administration, 36(1), 57–76. https://doi.org/10.1111/j.1754-7121.1993.tb02166.x. 17. kiani, k. (2014). neelum-jhelum project facing financial problems. https://www.dawn.com/news/ 1129742. 18. lammers, j., gordijn, e. h., & otten, s. (2008). looking through the eyes of the powerful. journal of experimental social psychology, 44(5), 1229–1238. https://doi.org/10.1016/j.jesp.2008.03.015. 19. lawrence, p. r., & lorsch, j. w. (1967). organization and environment. organization and environment. 20. lewin, k. (1935). a dynamic theory of personality. journal of heredity. https://doi.org/10.1109/ iembs.2008.4649489. 21. lewin, k. (1952). constructs in field theory [1944]. in field theory in social science: selected theoretical papers by kurt lewin (vol. 191, pp. 30–42). https://doi.org/10.1037/10269-014. 22. lukes, s. (1974). power: a radical view. london: macmillan. 23. march, j. (1962). the business firm as a political coalition. the journal of politics, 24(4), 662–678. 24. mitchell, r. k., agle, b. r., & wood, d. j. (1997). toward a theory of stakeholder identification and salience: defining the principle of who and what really counts. the academy of management review, 22(4), 853. https://doi.org/10.2307/259247. 25. mitnick, b. m. (1975). the theory of agency the policing “paradox” and regulatory behavior. public choice, 24(1), 27–42. https://doi.org/10.1007/ bf01718413. 26. mukhtar, i. (2013). nab, pm secretariat reluctant to take action. http://nation.com.pk/national/03-jan2013/nab-pm-secretariat-reluctant-to-take-action. 27. nagadevara, v. (2012). project success factors and inter-sectorial differences. review of business research, 12(1), 115–121. 28. nietzsche, f. (1968). the will to power. new york (vol.21). https://doi.org/10.1038/ scientificamericanmind0310-24. 29. patel, m., & robinson, h. (2010). impact of governance on project delivery of complex nhs pfi/ppp schemes. journal of financial management of property and construction, 15(3), 216–234. https://doi.org/10.1108/13664381011087489. 30. perrow, c., reiss, a., & wilensky, h. (1986). complex organizations: a critical essay. random house (vol. 3). https://doi.org/10.2307/2062232. 31. pfeffer, j. (2005). understanding the role of power in decision making. in classics of organization theory (pp. 289–303). 32. pfeffer, j., & pettigrew, a. m. (1976). the politics of organizational decision-making. contemporary sociology, 5(4), 478. https://doi.org/10.2307/2063824. 33. raven, b. h. (1993). the bases of power: origins and recent developments. journal of social issues, 49(4), 227–251. https://doi.org/10.1111/j.1540-4560.1993. tb01191.x. 34. raven, b. h. (2008). the bases of power and the power/interaction model of interpersonal influence. analyses of social issues and public policy, 8(1), 1– 22. https://doi.org/10.1111/j.1530-2415.2008.00159.x. 35. ross, s. a. (1973). the economic theory of agency: the principal’s problem. the american economic review, 63(2), 134–139. https://doi.org/10.2307/ 1817064. 36. salancik, g. r., & pfeffer, j. (1974). the bases and use of power in organizational decision making: the case of a university gerald r . salancik and jeffrey pfeffer. administrative science quarterly, 19(4), 453– 473. https://doi.org/10.2307/2391803. 37. sanderson, j. (2012). risk, uncertainty and governance in megaprojects: a critical discussion of alternative explanations. international journal of project management, 30(4), 432–443. https://doi.org/10.1016/ j.ijproman.2011.11.002. 38. schattschneither, e. e. (1960). the semi-sovereign people. dryden press. 39. see, k. e., morrison, e. w., rothman, n. b., & soll, j. b. (2011). the detrimental effects of power on confidence, advice taking, and accuracy. organizational behavior and human decision processes, 116(2), 272–285. https://doi.org/10.1016/ j.obhdp.2011.07.006. 40. szalai, j. (1997). power and poverty. social research. new york: oxford university press. 41. tost, l. p., gino, f., & larrick, r. p. (2012). power, competitiveness, and advice taking: why the powerful don’t listen. organizational behavior and human decision processes, 117(1), 53–65. https://doi.org/ 10.1016/j.obhdp.2011.10.001. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 53–63 ___________________________________________________________________________________________________________ 63 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjdffile false /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /description << /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /dan <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> /deu <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> /esp <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> /fra <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /ptb <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> /suo <feff004b00e40079007400e40020006e00e40069007400e4002000610073006500740075006b007300690061002c0020006b0075006e0020006c0075006f00740020006c00e400680069006e006e00e4002000760061006100740069007600610061006e0020007000610069006e006100740075006b00730065006e002000760061006c006d0069007300740065006c00750074007900f6006800f6006e00200073006f00700069007600690061002000410064006f0062006500200050004400460020002d0064006f006b0075006d0065006e007400740065006a0061002e0020004c0075006f0064007500740020005000440046002d0064006f006b0075006d0065006e00740069007400200076006f0069006400610061006e0020006100760061007400610020004100630072006f0062006100740069006c006c00610020006a0061002000410064006f00620065002000520065006100640065007200200035002e0030003a006c006c00610020006a006100200075007500640065006d006d0069006c006c0061002e> /sve <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice living with flood: a livelihood resilience approach of rural people in bihar, india received 2 december 2015 accepted 12 april 2016 living with flood: a livelihood resilience approach of rural people in bihar, india madhuri department of humanities and social sciences, iit kharagpur kharagpur, west bengal 721302, india email: madhuriptp09@gmail.com, madhuri@iitkgp.ac.in h. r. tewari department of humanities and social sciences, iit kharagpur kharagpur, west bengal 721302, india email: hrt@hss.iitkgp.ernet.in p. k. bhowmick rural development centre, iit kharagpur kharagpur, west bengal 721302, india email: pradipb@hijli.iitkgp.ernet.in abstract flood is a perennial problem in the state of bihar, india with devastating impact on the livelihood of people. in spite of the government’s measures of flood mitigation, households continue to live with sufferings on account of severe damage to their material and non-material assets. in this background, the objectives of the study are (1) to assess the mediating role of risk perception, and flood preparedness between flood experience and livelihood resilience; and, (2) to assess the mediating role of risk perception, and flood preparedness between flood education and livelihood resilience. the primary data were collected from 472 households by using multi-stage random sampling technique from seven blocks in river basins of ganga and kosi in the district of bhagalpur, bihar. to analyze the data descriptive statistics and structural equation modelling were used. however, risk perception is not found to mediate between flood experiences, flood education, and livelihood resilience. households adapt the strategy of ‘wait-watch-act’. households do not perceive flood as a threat but they have learnt to ‘live with flood’ as a ‘way of life’. the study recommends that the active involvement of the local people can be made mandatory with due consideration to their indigenous knowledge, flood experience, and flood education in order to make flood measures effective and successful. keywords: flood, vulnerability, resilience, livelihood, household 1. introduction the inundation of a vast area of land and the resultant loss of property, human lives, and livelihood of households by recurring river flood is the most challenging phenomenon in the agrarian state of bihar, india. the livelihood structure created after years of hard work by households is lost in no time, and its restoration takes longer than the expected time depending on pace and expediency of the relief assistance received from the government as well as non-governmental organizations. however, it is primarily the resilience measures of the local community (saavedra and budd, 2009) that matter the most and hence, ought be placed centrally and enhanced further (srivastava and laurian, 2006) in strategic interventions to cope with and recover from the shocks caused by flood (bosher et al., 2009). flood, all its devastating effects, severely jeopardizes the livelihood of people who live in active zone of flooding. the livelihood structure created after years of hard work is suddenly washed away in no time. households adapt strategies to journal of risk analysis and crisis response, vol. 6, no. 2 (july 2016), 48-66 published by atlantis press copyright: the authors 48 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india reconstruct their livelihood structure according to their capability, adaptability, in view of their experience and knowledge and with the support they receive from their own community, the government, and other external agencies including ngos. the government’s approach to livelihood includes both structural as well as non-structural measures. included in the structural measures are construction of dams, floodwall, and levee, while in non-structural measures, regulations, zoning, and protecting floodplains and wetlands are included (heidari, 2009). nevertheless, these measures neither succeed in checking flood damages, nor address the related issues of water logging and drainage congestion (vari and ferencz, 2006). failure of these measures brought in focus on reducing vulnerability, and thereafter resilience building of the affected households was accorded prominence in flood mitigation measures (hyogo framework for action 2005-2015: building the resilience of nations and communities to disasters). timmerman (1981) in his paper ‘vulnerability, resilience, and the collapse of societies’ introduced the term ‘resilience’, i.e., capacity to absorb, resist and recover from disaster (klein et al., 2003). tierney and bruneau (2007) stated that ‘resilience building’ is key to livelihood because (1) it nurtures and enhances the ability of social and physical systems to absorb, resist and recover from disaster; (2) the pre and post measures to mitigate disaster help in reducing harms it might cause (maguire and hagan, 2007; tierney and bruneau, 2007); (3)resilience building is an important trait for both the social and physical systems which support sustainability (tierney and bruneau, 2007); (4)it increases the capability in dealing with uncertainties and unexpected changes (berkes, 2007); (5)the measures of resilience building are proactive and emphasize on the collective efforts of the whole community; and (6) it focuses on the areas through which the capability of the community can be enhanced. 1.1. flood experience households carry experience of negative feelings of flood, according to ‘inoculation hypothesis’ (norris and murrell, 1988; slovic, 1987). the fear of impending risk makes them careful in confronting and overcoming flood (chongfu, 2014). attentiveness is more in the person with risk experience than the one without it (brilly and polic, 2005). however, there is difference in risk perception even in cases with similar experience of flood (ruin et al., 2007, kaiser et al., 2004, siegrist et al., 2008). weinstein (1989) described three ways by which experience can affect risk perception (1) societal attention, at the time of flood occurrence; (2) victim-directed influence like education and social norms; and, (3) intraindividual response (cited from howe, 2009). the approach and outlook which develop after the assessment of risk (lazo et al., 2010) lead to adaptation (tompkins and adger, 2004), of protective actions (weinstein, 1989) to resist future flooding (raaijmakers et al., 2008). the impetus for adaptation of measures to cope flood depends on the level of perceived risk of households (reid et al., 2007). if risk is not a fear, there will be no effort for adaptation (smit and wandel, 2006). it is flood experience and risk perception, which enforce households to make livelihood sustainable (harvatt et al., 2011) in perilous situation (berkes, 2007). thus, the cognitive factors play significant role in livelihood resilience (grothmann and reusswig, 2006). the experience and consciousness of impending risk is vital in adaptation of livelihood resilience (grothmann and patt, 2005) by adapting different activities to sustain and maintain livelihood (ellis, 1998; niehof, 2004). the diversification and adaptation of livelihood strategies to refrain flood is influenced by flood experience and risk perception (slovic et al., 2004). according to protection motivation theory (pmt), which takes into account the threat appraisal process and coping appraisal process. flood experience and response help households residing in flood prone areas to safeguard themselves through adaptation (bubeck, botzen and aerts, 2013). the threat appraisal process is based on the apprehension of risk and its repercussions, while coping appraisal process is based on response efficacy (individual’s expectancy that carrying out recommendation can remove the threat), and self-efficacy (belief in one’s ability to execute the recommend courses of action successfully) to withstand, confront, and resist flooding (rogers,1983). the protection motivation theory (pmt) has divided responses into two categories, i.e., protective responses and non-protective responses. protective responses are applied when threat appraisal and coping appraisal are high to prevent monetary or physical damage. on the other hand, in non-protective responses, the threat appraisal is high and coping appraisal is low as was found by milne et al. (2000) in the metaanalysis of pmt research, considering 27 studies published by atlantis press copyright: the authors 49 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india involving 7694 participants. high-risk perception makes households to adapt coping responses, which may be either protective or non-protective response (abraham et al., 1994). the model of private proactive adaptation to climate change (mppacc) renamed threat appraisal to risk appraisal and coping appraisal to adaptation appraisal. however, adaptation appraisal, or perceived adaptive capacity has not been included in studies of risk perception after flood. the study (grothmann and reusswig, 2006) which reveals the relation between risk perception and adaptation processes is based on a case study. there is a lack of empirical study showing the relation between the role of risk perception and adaptation of households for livelihood resilience. the acceptance and avoidance of risk is voluntary to households (raaijmakers et al., 2008). the fear of being vulnerable and consciousness of risk boost households’ resilience and actions to cope with flood (grothmann and reusswig, 2006; ludy and kondolf, 2012). however, studies differ about the nature of relationship between flood experience, risk perception, and mitigation method. in the study of hurricane and storm risk by peacock et al. (2005), the experience and damage of flood led to enhancement of risk perception resulting in improved precautionary actions (lindell and perry, 2000) to mitigate flood risk (mileti, 1999). the threat of flood enhances adaptation of livelihood strategies of households (botha et al., 2011). flood experiences amplify risk perception and consequently the behavior of recovery and adaptation. these are important components of livelihood resilience building and risk reduction (bubeck, botzen and aerts, 2012) after flood. the households’ flood experience and perception of risk influence their responses to control (harvatt et al., 2011) and initiate precautionary measures to overcome flood (harvatt et al., 2011) and make themselves resilient (fatti and patel, 2013). the way households perceive flood risk and resulting damages determine households’ responses to and management of flood risk (harvatt et al., 2011). studies reveal that the personal experience of destruction of livelihood increases the fear of flood risk perception (plapp and werner, 2006; siegrist and gutscher, 2006) and motivate households to accelerate efforts for livelihood resilience. on the other hand, there are studies which do not find any influence of risk perception on private mitigation strategies (bubeck, botzen and aerts, 2012; nyakundi, mogere, mwanzo and yitambe, 2010) or decreases risk perception after flood experience (brilly and polic, 2005; botzen, aerts and van den bergh, 2009). therefore, the differences in approaches of households with regard to acceptance or ignorance of risk are reflected in their livelihood resilience. therefore, the present study looks into risk perception as a mediator between flood experience and livelihood resilience. the experience of flood is a key component in flood risk management. it determines households’ responses to flood warnings and their efforts to improve preparedness (botzen, aerts and van den bergh, 2009b; aboagye, dari and koomson, 2013). it always alerts households to remain prepared to meet flood challenges (few et al., 2005), and enables them to adjust with and respond to the flooding situation to minimize flood damages (mccarthy et al., 2001). the self-protective measures based on personal experience keep households ready for future ones (weinstein, 1989). jackson (1981) found that experience of earthquake influence households’ preventive measures. households use their experience in adapting and coordinating flood mitigation measures like laying sand bags in flood prone areas (anderson, 1965), and how to participate in rescue activities (perry and lindell, 1978). they themselves attempt to save their lives and property (keogh et al., 2011). knowledge and experience provide strategic input to the affected households in their efforts to mitigate flood effects (the theory of bounded rationality), (smith, 2001). it comes in handy to households in designing suitable measures to mitigate flood impacts (tapsell, 2001) and also broadens their perception and responses to an event (cutter,1993). households develop ways to minimize damages and devise livelihood strategies in view of the nature and extent of damages, on the one hand, and make optimal use of available resources, on the other. they remain prepared to handle the dreadful effects of flooding (cash and moser, 2000). in course of time, it becomes their strength to withstand the unstable situation (robyn, 2012), according to the social constructionist theory (loseke, 1999). households’ knowledge and experience of flood find to be better equipped in dealing with flooding (brilly and polic, 2005; wilson, 2012) and always keep them alert and motivated with emergency preparedness to meet flood challenges (mileti,1999). the social constructionists explain how households through published by atlantis press copyright: the authors 50 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india personal experience and interaction respond to flood (stabllings, 1995). h1: risk perception would mediate between flood experience and livelihood resilience. households’ experience of flood and worries to it are reflected in their preparedness level (raaijmakers et al., 2008). raaijmakers et al., (2008) observed that experience generates awareness, understanding along with worry, which decreased after long time. the community disaster resilience framework emphasized that more the households have resilience building capacity, less would be their worries about repercussions of flood (raaijmakers et al., 2008) (fig.1). fig. 1: relationship between flood characteristics (cited from raaijmakers, 2008) households’ repeated experience to flood enriches their understanding of the nature of flood, which is ultimately helping them in their livelihood resilience (grothmannw and reusswig, 2006). furthermore, flood experience helps household in overcoming flood impact (pagneux et al., 2011) by responding through preparedness measures which facilitate their efforts in livelihood resilience (paton et al., 2006). livelihood resilience measures do not only mean to protect from flood but also taking proactive measures to control flood (longstaff, 2005). it enhances households’ efforts and abilities (folke et al., 2003) which assist them to adapt in hazardous and uncertain conditions (lengnick-hall and beck, 2005). the anticipation of flood risk and the destruction it causes, trouble the households, who accordingly may remain prepared for future event, so that the impacts can be reduced, and their property and life can be saved (howe, 2011). there are a number of studies of impact of disaster preparedness (lindell and perry, 2000) on socio-demographic characteristic of the households. on the contradiction, there are studies, which show that there is no correlation between flood experiences and flood preparedness measures (lin et al., 2008) despite being frequently affected by flood. in a nationwide survey of earthquake risk in taiwan, lin et al. (2008) did not find any relation between experience, risk perception, and households’ preparation. in survey of flood preparedness in the maribyrnong (victoria) (victoria state emergency service (vicses), 2008), and in 2005 survey of flood prone properties in maitland, nsw (new south wales) (huntercentral river catchment management authority, 2005);shows that there is minimum or no preparedness plan to combat flood (gissing, keys and opper, 2012). in another survey of landslide victims in taiwan, support shows that disaster experience does not have any bearing on preparedness (ho et al., 2008). while on the other hand, in the study by slovic (2000) of hurricane reveals that flood experience stimulates flood preparedness. however, there is paucity of study, to establish that flood preparedness stimulates households’ livelihood resilience after flood. h2: flood experience would influence livelihood resilience through flood preparedness. therefore, the first objective of the study is to assess the mediating role of risk perception, flood preparedness on flood experience, and livelihood resilience. 1.2. flood education education, in general, broadens households’ understanding of the social and physical world around them; flood education creates awareness with regard to pros and cons of flooding in particular. flood education is defined as ‘any learning process or activity that builds households resilience to flooding’ (dufty, 2008). in a critical situation of flood and in recovery of livelihood, it comes in handy to households’ rescue, improves their capabilities, and knowledge to marshalling political and economic advantages (srinivas, 1996). in flood management, awareness creation and issuance of warning before hand play crucial role (elliott et al., 2003). webber and dufty (2008) identified ‘preparedness conversion’, ‘mitigation behavior’; ‘adaptive capability’, ‘community competencies’ and, ‘post-flood learning’ as an important function of disaster education. though in the guidelines for disaster management (mariasabo and gavrila, 2011; unisdr, 2009) it is suggested that disaster education plays a vital role in minimizing flood impact, but it is not included in the government’s mitigation measures. moreover, the government’s awareness programs do not properly emphasize on provisioning, protection, published by atlantis press copyright: the authors 51 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india and promotion of livelihood resilience. in livelihood provisioning, food and health relief are provided to households who are chronically vulnerable, whereas livelihood protection intervention includes income transfer, infrastructure repair, rehabilitation, and improvements, besides food or cash for work or other means and compensation for assets such as tools, boats and seeds. livelihood promotion is a set of development initiatives of households to diversify their livelihood strategies; create alternative income-generating activities; provide financial services, such as loans and insurance and strengthen markets (cited from the household livelihood security concept, retrieved from ftp://ftp.fao.org/docrep/fao/x0051t /x0051t05.pdf). in most of the studies, flood education had not been looked as an important factor in constructing livelihood resilience (paton et al., 2006). paton et al. (2006) stated that resilience is a measure of how well households and societies can adapt to a changed reality and capitalize on new possibilities offered. flood education by making forecasts, issuing early warnings, outlines recovery procedures (heinz, 2000) which help households to resist and overcome flood impact. it further generates awareness among them with regard to adaptive measures (botzen et al., 2009) for livelihood restoration (berkes, 2007). however, the attitude and sensitivity of households towards risk is important in adaption of strategies to diversify income sources, and self-organize before, during and after flood. the efforts and responses of households to resist flood depend on their consciousness (fielding et al., 2005). however, palm (1981) found that disaster education does not guarantee ‘risk avoidance behavior’, but is about ‘preventive measures’ to protect households (howe, 2009). therefore, there is a need to create consciousness, and competencies among households for their livelihood resilience (dufty, 2008). studies show how consciousness is significant in flood mitigation measures of households (pagneux et al., 2011; burningham et al., 2008; miceli et al., 2008; weinstein, 1989) which may be enhanced through training to improve their adjustment and adaption of opportunities of income and potential precautionary measures to enable them in restoring their livelihood bases. therefore, risk perception plays a prominent role in creating awareness and enhancing understanding about future occurrences of flood (raaijmakers et al., 2008). thus, the study tries to explore the mediation of risk perception between flood education and livelihood resilience. h3: risk perception would mediate between livelihood resilience and flood education. not only risk perceptions of households motivate them towards resilience measures, but flood preparedness also facilitates them in their livelihood restoration. furthermore, dufty (2008) emphasized that ‘flood education can facilitate the community to build its capability (networks, leadership, and competencies) for preparedness, response, and recovery and involve the community in the planning, implementation and evaluation phases’. households with knowledge and being aware of potential impact of hazards remain prepared with contingency plans to meet challenges arising out of flooding conditions (bauman, 1983). the protective measures give proper direction and stimulate the process of resilience so that livelihood can be regenerated without any heavy loss (dufty, 2008). the attentiveness of potential hazard keep household prepared with emergency plans to meet the challenges arising from flood (bauman, 1983). study show how disaster education increases preparedness for hurricane (faupel et al., 1992). on the contrary, studies also show that in spite of learning through flood education, it does not stimulate households for livelihood resilience (boura, 1998; paton et al., 2003). along with flood control and mitigation, the pre-flood preparedness practices help in restructuring livelihood activities is an important concern, which needs attention. the improvement in capabilities and diversification of livelihood help households to bounce back to normal life (walker et al., 2004) without experiencing much destruction in their day-to-day life. the awareness programs mainly focuses on mitigation and coping measures but there is dearth of study which aid in adaptation and restoration of livelihood. the needs and involvement of households in flood prone area may be given proper consideration instead of fixing the flood education process only to disseminate knowledge so that it cannot be one-way approach (o’neill, 2004). flood education is a most essential factor in mitigation and management (berkes and folke, 1998) of flood control (berke, 1998; burby et al., 2000) but flood preparedness enhance livelihood resilience is not elucidated (burby, 1998; burby et al., 2000). hence, the study articulates flood preparedness as an intervening published by atlantis press copyright: the authors 52 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india aspect between flood education and livelihood resilience. h4: flood experience would influence livelihood resilience through flood preparedness. hence, the second objective of the study is to find out the mediating role of risk perception, and flood preparedness between flood education and livelihood resilience. 2. study area and sampling strategy the study was conducted in the district of bhagalpur, bihar. it has an area of 2570 sq km and the rivers ganga and kosi traverse throughout the district. the southern part of the district falls in the badua-koa sub-basin of the river ganga, and the area in north of ganga falls under the baghmatikosi sub-basin. these two sub-basins constitute the mid-ganga basin that causes severe damage to life and property. the district is principally drained by the river ganga, which enters the district at sultanganj. the northern boundary of the district is marked by the river kosi (ghugri) heavily laden with silt and sand. geomorphologically, the district forms a part of the mid-ganga foreland basin (ministry of water resources, 2009). fig.2 shows the flood inundated area of bhagalpur district. the study is based on the primary and secondary data, and focus group discussion. the primary data were collected by using multi-stage random sampling technique. in the first stage, the purposive sampling method was used to identify the blocks, which have remained inundated in the last 6 years during successive floods with the help of data provided by the bihar disaster management department, and in consultation with block development officers (bdos). based on the data obtained, out of 13 blocks, which are often affected by flood, 7 blocks (bihpur, ismailpur, gopalpur, rangra chowk, kharik, narayanpur, and naugachhia) are selected for study. table 1 shows the number of village and size of population of the blocks. the secondary data were collected from the records of the district planning board, various depts. of the governments of bihar dealing and several other documents and reports published from time to time with regards to the issue of flood in the state. 2.1. focus group discussion focus group discussions were arranged in each block of the area under investigation to ascertain the views of the village people with regard to the issue of livelihood resilience of the flood affected households and other flood related issues. in the fig. 2: administrative map of bhagalpur district, bihar (central ground water board, ministry of water resources, 2009). published by atlantis press copyright: the authors 53 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india table 1: number of village and size of population of the blocks sl. no blocks villages population in the block 1. ismailpur 6 6,277 2. gopalpur 11 76,420 3. rangra chowk 9 72,780 4. kharik 19 102,825 5. bihpur 22 97,033 6. narayanpur 15 81,971 7. naugachia 12 122,809 source: http://www.allaboutbihar.com focus groups discussions, the issues of livelihood resilience measures, households apply to cope with flood were discussed. groups were constituted in all the blocks consisting of randomly selected ten members in each group. in groups only those members were included whose names did not figure in the list of respondents for quantitative data collection. before the focus group discussions were held in actual, the researcher met the village persons and explained about the purpose of discussion. they were requested to express their opinion frankly without any fear or prejudice. initially, the village people were very reluctant and were not agreeing for discussion due to their bitter experience of the government’s approach to their plight during flooding. it was a tough time for the researcher to convince. it was then that village people very hesitatingly agreed to take part in discussion. after that, the place, time, and dates were decided as suggested by village people. before, the actual discussions, the members were requested to follow an order and express their opinion without hesitation one by one. tea with biscuits was arranged for adult members and toffee for young children. all the members were then requested to ensure everyone’s participation. members were than frankly requested to share their experience and opinion on the issue raised. the questions raised for discussion were the following:  what do you do to cope with flood within the household and in the farm?  how do you manage to reconstruct your livelihood after flood, or how do you survive in the flood prone area? with regard to the first question, members stated, “since flood a recurring phenomenon in the area, they have become habituated to live with it.” they closely watch over flood progress and remain ready to evacuate any time. they also identify places where they can keep their belongings safely when floodwater will cover the danger level. they move to safer places with their belongings, which they could carry with them. the heavy items of the household are kept on the rooftop of the dwellings, or tied with the roof. temporary shelters are made of plastic on the roadside, railway track, national highway, and other high land areas. food is rarely cooked and water and other essentials of life are arranged with great difficulty. the government though provides food packets and other items but it is not being made available to everyone. it is often too short of households’ requirements. in order to meet food requirement in emergency conditions, they keep which are traditional fast food, items ready to eat during flood. vegetables are grown in house premises. as floodwater stays for three to four months, households go for traditional farming. one of the members mentioned that it is because of her (ganga river) displeasure that she brought silt and sand to our land. ganga river is worshipped as holy mother. one of the respondents said, “we are poor people and have no savings but we work as daily wage laborer (roj kamne and khane wale log). the damage and destruction caused by flood has made our life hell. flooding is a curse to them in their opinion. it destroys all their resources support base resultantly many households migrate to other places in search of livelihood. the government officials in connivance with the political leaders takes away for themselves all the relief assistance government sends for distribution among flood victims. it is an open loot of the government resources, and no action is taken against erring officials and persons in spite of several complaints. 2.2. measures the following section details the measurement scales and the items (table 2) used in measuring the constructs. in-depth interviews were conducted with key informants at three study sites. information from the qualitative research was used for designing the structured interview schedule for household survey. the final measurement scales and design of the interview schedule on a five point likert scale. amos and spss software was used to analyze the data. 3. results and analysis 3.1. preliminary data analysis t he preliminar y data analysis, reliabilit y estimation, confirmatory factor analysis through published by atlantis press copyright: the authors 54 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india structural equation modeling was used to analyze and interpret the data. the normality of the variables was tested by skewness and kurtosis. 3.2. reliability of the measurement scales in accordance with the chronbach alpha test, all the constructs except flood preparedness obtained an acceptable of a coefficient alpha above .70 (table 3) indicating that the measurement scales were reliable and appropriate for further data analysis. 3.3. correlations between variables during the preliminary analysis of data, the mean and standard deviation were obtained for each variable. the correlation gave preliminary support for the hypotheses of the study (table 4). flood experience was negatively correlated and insignificant to livelihood resilience (r = -.075, p > .01). flood education was also negatively correlated to livelihood resilience but significant (r = .272, p < .01). however, risk perception was positively correlated to livelihood resilience and significant (r = .172, p < .01). the correlation between flood preparedness and livelihood resilience was significant but negative (r = -.413, p < .01). the study also explored the relation between flood experience and risk perception (r = .585, p > .01) but the correlation between flood table 2: measurement and scales of items. measured variable no. of items measurement scale source flood experience 11 1=not exposed to flood 5= very severely exposed to flood tyler and hoyt (2000); norris and murrell (1998); susan et al. (1996) flood education 6 1= strongly disagree 5= strongly agree dufty (2008); mishra and suar (2005) livelihood resilience 15 1=not at all 5=much more than usual household questionnaire: survey of living conditions, uttar pradesh and bihar adapted from world bank (1997); hahn et al. (2009); dhs (2006); who/rbm (2003); fernando (2003); patnaik and narayanan (2010); scoones (1998); little et al. (2001) risk perception 6 1=not at all 5=much more than usual slovic et al. (1980) flood preparedness 11 1=not at all 5= much more than usual mulilis et al. (1990); brun et al. (1997); van der veen and logtmeijer (2005) table 3: summary of measurement reliability measured variables no. of items items remained cronbach alpha 1. flood experience 11 11 .87 2. flood education 4 4 .71 3. livelihood resilience 15 15 .78 4. risk perception 6 6 .80 5. flood preparedness 11 11 .69 table 4: correlation between the variables m sd flex fled lvrs rspr flpp flex 45.84 11.50 1 fled 6.77 4.38 .306** 1 lvrs 50.21 .550 -.075 -.413** 1 rspr 17.90 -623 .585** .207** -.011 1 flpp 17.76 .147 .084 .526** -.397** 0.17 1 published by atlantis press copyright: the authors 55 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india experience and flood preparedness (r = .084, p < .05) was also significant. the correlation linking flood education and risk perception (r = .207, p < .01) and flood preparedness (r = .526, p < .01) was also significant. 4. factor analysis 4.1. exploratory factor analysis factor analysis is an interdependent technique, which aims to determine the underlying structure among variables (hair et al., 1998). as livelihood resilience scale was prepared by the researcher through focus group discussion and interview with informants, therefore, in order to ascertain how and up to what extent the items of the scale were linked to the construct in a different context, exploratory factor analysis (efa) method was applied on the sample (n=472). the sample was subjected to principal components analysis (pca). prior to performing pca the suitability of data for factor analysis was assessed. inspection of the correlation matrix revealed the presence of correlation coefficients of .3 and above. the kaiser-meyeroklin value of the variable (.628) was exceeding the recommended value of .6 (kaiser, 1970) and the barlett’s test of sphericity (bartlett,1954) reached statistical significance (p<.01), supporting the factorability of the correlation matrix (hair et al., 1998). in efa, first the unrotated factor matrix was computed, containing the factor loading for variable on factor. factor loading is the correlation of variables and the factor. the higher loading makes the variable representative of the factor. the factor-loading matrix was examined and significant loading was identified. in the sample size of 472, factor loading of .35 and higher was considered significant for interpretive purpose. variables having communalities of greater than .50 were retained in the analysis. the unrotated factor solution did not provide adequate information of variables under examination. in the unrotated factor loading variables, having cross loading and those with low significance was identified. catell (1966) scree test was also used to identify the optimum number of factors that can be extracted. this was further confirmed by the results of parallel analysis, which showed components with eigen values exceeding the corresponding criterion values for a randomly generated data matrix of the same size. to achieve simpler and theoretically more meaningful factor solution, orthogonal approach with varimax factor rotation method was used, which attempted to minimize the number of variables that had a high loading on factor. the rotated solution revealed the presence of simple structure (thurstone, 1947) with factors showing a number of strong loadings. the loadings were improved for almost every item. the factor loading of the items ranges between .543 to .657, the eigen value was 3.976 and variance was 25.6. for further analysis of the constructs, latent variable structural equation modeling method was used. 4.2. latent variable structural equation modeling (lvsem) latent variable structural equation modeling (lvsem) tests the sequential relationship between a series of independent and dependent variables. it assists in specifying measurement model as well as structural models there are two components of sem, measurement model and structural model. the measurement model was evaluated by using confirmatory factor analysis (cfa). the measurement model specifies the posited relationships of the observed indicators to the latent construct. therefore, before testing the overall measurement model, each construct in the model was evaluated and analyzed separately by respecification of the model. the model respecification procedure was used to identify the source of misfit and to generate a model that achieves better fit to the data. the measurement model was modified by examining the standardized residuals, modification indices, and the standardized loading estimates (hair et al., 2006). each of the measures was examined together with the model fit indices to ascertain if respecification was needed. the model fit was examined using multiple indices such as χ² and the χ²/df, gfi, cfi, tli, rmsea. after the modification of the fit indices, the final cfa model was improved. the revised measurement model fit the data well. further, when each construct had shown an acceptable fit to the model, then all constructs were evaluated together (table 5). hypothesis (h1): risk perception would mediate between flood experience and livelihood resilience. the present study also examined (h1) the relationship between flood experience and livelihood resilience mediated by risk perception. at first the direct relationship was determined (table 6), i.e., (a) regression was run to predict published by atlantis press copyright: the authors 56 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india livelihood resilience from flood experience (β = .37, p <.01, c.r.= 3.2) was significant and supported; (b) the regression coefficient for the prediction of risk perception from flood experience was negative and statistically insignificant (β1 = -.90, p > .05, c.r = -1.79); (c) next, regression was performed to predict the effect of risk perception on livelihood resilience which was significant and positive (β3 = .174, p <. 01, c.r. = 4.0); and, (d) with risk perception in equation, the regression coefficient was not smaller than the β1 (β4 = .082, p > .05, c.r = 1.73), further the β1 was not significant; risk perception had no mediation in the relationship between flood experience and livelihood resilience. table 7 reveals the indirect effect was not supporting the mediation. therefore, h1 was refuted. hypothesis (h2): flood experience would influence livelihood resilience through flood preparedness. next, the hypothesis (h2) of the study is when mediator flood preparedness was taken into account the relationship between flood experience and livelihood resilience would be more powerful. before considering the position of flood preparedness as mediator in the analysis, at first the bivariate association between flood experience and livelihood resilience was established (table 8) (a) flood experience significantly account for variations in livelihood resilience (β = .37, p < .01 ,c.r. = 3.2); (b) the relationship between flood experience and flood preparedness was significant (β2 = .158, p < .01, c.r. = 4.4);(c)flood preparedness significantly account for variation in livelihood resilience (β3 = .164,p <.01, c.r = 2.7); and,(d) with flood preparedness in equation, the regression coefficient of flood experience to livelihood resilience was smaller than the β1 (β4 = .037, p < .05, c.r. = 2.3), further the β1 was still significant, flood preparedness had a partial mediating effect on the relationship between flood experience and livelihood resilience. table 9 shows the direct, indirect, and total effect of mediation. therefore, h2 was supported. figure 3 shows the hypothesis h1 and h2. table 5: summary of cfa result and model fit indices (n=472) construct cfa items reliability 4. flood experience 6 .82 χ² = 42.3; χ²/df = 3.25; p = .000; rmsea = .06; gfi = .96; tli = .96; cfi =.98 5. flood education 3 .69 χ² = 48.8 ; χ²/df = 2.9; p = .000; rmsea = .00; gfi = .99; tli = 1.00; cfi = 1.00 6. livelihood resilience 6 .70 χ² = 18.6; χ²/df = 2.3; p = .000; rmsea = .05; gfi = .98; tli = .97; cfi = .98 7. risk perception 5 .78 χ² = 17.14; χ²/df = 3.4; p = .004; rmsea = .07; gfi = .98; tli = .98; cfi = .99 8. flood preparedness 5 .70 χ² = 21.77; χ²/df = 2.7; p = .000; rmsea = .06; gfi = .98; tli = .90; cfi = .94 table 6: hypothesis testing of risk perception as mediator between flood experience and livelihood resilience hypothesized path direction beta estimate c.r/ t value se decision lvrs ← flex flpp ← flex lvrs ← flpp + + + .037* .158* .164* 3.28 4.43 2.74 .037 .058 .060 supported supported supported note: β = regression weight; *p<.05, **p<.01 (2-tailed) table 7: direct, indirect, and total effects of risk perception as mediator between flood experience and livelihood resilience variable variable rspr lvrs direct indirect total direct indirect total flex -.90** .00 -.90** .37** .82 .119** rspr .00 .00 .00 .17 .00 -.09 note: path estimates were reported; *p<.05, **p<.01 (two-tailed) published by atlantis press copyright: the authors 57 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india hypothesis (h3): risk perception would mediate between livelihood resilience and flood education. the study also looked that (h3) risk perception would mediate between livelihood resilience and flood education. table 10 shows (a) flood education was significantly accounted for variations in livelihood resilience (β1 = .224, p < .01, c.r. = 2.3); (b) the relationship between flood education and risk perception was significant (β2 = .122, p < .05, c.r. = 3.4); (c) risk perception was significantly accounted for variation in livelihood resilience (β3 = .174, p < .01, c.r. = 4.0); and,(d) with risk perception in equation, the regression coefficient was significant but negative β2 (β3 = .296, p < .05, c.r. = -3.1). hence, there was no mediation of risk perception between flood education and livelihood resilience. table 11 shows that the direct effect with mediator was negative. therefore, h3 was refuted. hypothesis (h4): flood education would influence livelihood resilience through flood preparedness. the study also explored the relationship (h4) table 8: hypothesis testing of flood preparedness as mediator between flood experience and livelihood resilience hypothesized path direction beta estimate c.r/ t value se decision lvrs ← flex flpp ← flex lvrs ← flpp + + + .037* .158* .164* 3.28 4.43 2.74 .037 .058 .060 supported supported supported table 9: direct, indirect, and total effects flood preparedness as mediator between flood experience and livelihood resilience variable variable flpp lvrs direct indirect total direct indirect total flex .37** .00 .37** .37** .37** .74** flpp .00 .00 .00 .16** .00 .16** note: path estimates were reported; *p<.05, **p<.01 (two-tailed) note. dotted lines show indirect effects **p < .01, *p<.05 fig 3: risk perception and flood preparedness as a mediator between flood experience and livelihood resilience table 10: hypothesis testing of risk perception as mediator between flood education and livelihood resilience hypothesized path direction beta estimate c.r/ t value se decision lvrs ← fled rspr ← fled lvrs ← rspr + + + .224 ** .122* .174 ** 2.35 3.45 4.08 .095 .042 .043 supported refuted supported note: β = regression weight; *p<.05, **p<.01 (2-tailed) published by atlantis press copyright: the authors 58 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india between flood education and livelihood resilience. table 12 reveals (a) flood education significantly accounted for variations in livelihood resilience (β1 = .224, p< .01, c.r. = 2.3); (b) the relationship between flood education and flood preparedness was significant (β2 = .472, p < .05, c.r. = 4.02); (c) flood preparedness was significantly accounted for variation in livelihood resilience (β3 = .178, p < .05, c.r. = 2.8) and, (d) with risk perception in equation, the regression coefficient between flood education and livelihood resilience was smaller than the β1 (β4 = .164, p < .01, c.r. = 2.7); but the β4 was still significant, flood preparedness has a partial mediating effect on the relationship between flood education and livelihood resilience. table 13 shows the direct, indirect and total effect of mediation analysis, which was significant and supported. therefore, h4 was supported. the fit measures of the direct and indirect path (including mediators) model indicate that chi square of all the models were highly significant (p < .001). as chi-square was sensitive to sample size, so chi-square for degree of freedom (χ²/df) was estimated. the result shows that the relative chi-square was not below the required limit of 3 (kline, 1998). gfi, cfi, nfi was close to .90 which reveals the good fit of the model. the parsimonious (pgfi, pcfi, pnfi) measures were acceptable in both the models. however, it was slightly low in the direct model because the mediation was absent in that model (table 14). rmsea show the approximation of the observed model to the true model, which was acceptable in this model. however, the indirect model was better than the direct model because it included all possible paths. therefore, the result of the hypothesized structural model reveals that the initial model did not fit the data well, and so, it was not acceptable (χ² = 380.6, χ²/df = 2.45, p = .01, gfi = .90; tli = .91, cfi = .92, rmsea = .06). the results indicate that risk perception does not mediate between livelihood resilience and flood experience and flood education. the two hypothesized paths do not show any significant relationship between exogenous and endogenous variable. therefore, these hypotheses paths were removed because the construct does not have any significant affect on livelihood resilience (fig. 3) revised model with standardized and unstandardized path table 12: hypothesis testing of flood education would influence livelihood resilience through flood preparedness. hypothesized path direction beta estimate c.r/ t value se decision lvrs ← fled flpp ← fled lvrs ← flpp + + + .224 ** .472* .178* 2.35 4.02 2.88 .095 .117 .178 supported supported supported note: β = regression weight; *p<.05, **p<.01 (2-tailed) table13: direct, indirect, and total effects flood preparedness as mediator between flood education and livelihood resilience variable variable flpp lvrs direct indirect total direct indirect total fled .22* .00 .15 .22** .16** .48** flpp .00 .00 .00 .17* .00 .17* note: path estimates were reported; *p<.05, **p<.01 (two-tailed) table 14: fit measures of two models dealing with mediation χ2/df p gfi tli cfi pgfi pcfi pnfi rmsea direct model 2.10 .001 .91 .91 .92 .60 .68 .66 .05 indirect model 2.45 .001 .90 .91 .92 .62 .69 .67 .06 table 11: direct, indirect, and total effects risk perception as mediator between flood education and livelihood resilience variable variable rspr lvrs direct indirect total direct indirect total fled .22** .00 .22** .22** .29** .51** rspr .00 .00 .00 .17** .00 .17** note: path estimates were reported; *p<.05, **p<.01 (two-tailed) published by atlantis press copyright: the authors 59 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india coefficient). consequently, a more parsimonious model (table 15) was made after deleting the nonsignificant path from the initial model (χ² = 326.8, χ²/df = 2.04, p = .00, gfi = .94; tli=.96, cfi=.98, rmsea= .06). figure 4 shows the hypothesis h3 and h4. 5. discussions objective1: to assess the mediating role of risk perception, flood preparedness on flood experience, and livelihood resilience risk perception is not found to mediate between flood experience and livelihood resilience as the findings of the study reveal. although, flood experience plays significant role in both flood protective behavior and flood risk perception of households, their perceived flood probabilities and perceived flood consequences do not intervene in livelihood resilience. this is consistent with the heuristic theory, which emphasizes that households do not unreasonably worry about flood impact because of their flood experience (grothmann and reusswig, 2006). rather, learning from repeated exposure to flood comes in handy to households in their efforts to skillfully arrange livelihood, as it provides them added advantage when it is most needed. the learning from flood experience enables households to resist, recover, and bounce back to normal livelihood after flood. they adjust and learn to live in hazardous condition on the basis of their previous flood experience which saves them from much of their likely losses. flood experience is further found to provide stability and enhances households’ capability to skillfully and effectively handle flood challenges (fig 5 and fig. 6). fig. 5: households shifting their belonging to a safer place fig. 6: households tie their belongings to the rooftop of the house table 15: goodness of fit results for structural model measurement model goodness-of-fit indices χ² χ²/df p gfi tli cfi rmsea initial model 380.6 2.45 .001 .90 .91 .92 .06 final model 326.8 2.04 .000 .94 .96 .98 .07 note. dotted lines show indirect effects **p < .01, *p<.05 fig 4: risk perception and flood preparedness as a mediator between flood education and livelihood resilience published by atlantis press copyright: the authors 60 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india the study is consistent with the previous literature (bubeck, botzen and aerts, 2012), which found weak relationship between flood risk perceptions and mitigation measures. households’ views and perceptions about flood changes in course of time because of their repeated flood experience. as a result, they no more remain fearful of flood impacts. their flood experience and capabilities instill further confidence in them with regard to livelihood resilience, diversification of livelihood, and income generating activities. the protection motivation theory (threat coping and threat appraisal behavior) does not hold relevance and is not imperative in livelihood resilience. the study is contrary to the study by wachinger and renn (2010) which found that risk perceptions influence vulnerability of the households. the study also differs with the study of bradford et al. (2012) which proposed to develop a model of social resilience based on risk perception. however, the study confirms the assumption that risk perception largely depends on recency (hertwig et al., 2004), frequency, and intensity of households’ flood experience, but does not play role in adaptation of resilience measures. flood preparedness partially mediates between flood experience and livelihood resilience. the study is in contrast to the findings of the study which had found high level of emergency preparedness (lindell and prater, 2000), but is consistent with the study which found low and minimal preparedness (jackson, 1981; turner et al., 1986; miceli et al., 2008). the study is further consistent with the study of residents in a floodprone area of italy, which showed low level of individual-level domestic protection (de marchi et al., 2007). hung-chih (2009) found a similar pattern among taiwanese participants. flood experience is also found to stimulate households in flood preparedness based on their understanding and familiarity with flood devastation as was found in previous studies (basolo et al., 2009; harvatt, petts and chilvers, 2010; kirschenbaum, 2002). the result is consistent with the study of nyakundi, mogere, mwanzo, and yitambe (2010) which concluded that households’ indigenous knowledge play crucial role in livelihood resilience. the resilience measures of flood control are a byproduct, or an outcome of households’ direct experience of flooding and reactive mechanism to reinstate livelihood to its normal condition. households do not evacuate or shift their belongings and livestock to safer places unless all options have not been ruled out, and households do not find themselves placed in much deep water. absence of boat, dinghy and other means of transportation complicate the process of households’ evacuation. as a result, household follow their own ways and shift their belongings to the rooftop of the house. dwellings of households in active flood zones are generally built on silt or elevated ground. objective 2: to assess the mediating role of risk perception, and flood preparedness between flood education and livelihood resilience. the results of the study do not find mediation by risk perception between flood education and livelihood resilience as because households, in course of time, adjust with flood, live with uncertainty, and accordingly adapt resilience measures. though there is no formal flood education programs in the study areas, households always learn from the experience of the senior persons of the community about the pros and cons of flood, and what measures may be more effective in different flooding situations. this facilitates households in making better use of the available resources to recreate livelihood. this, possibly, may be for the reason that households are neither warned, nor informed beforehand about the probable occurrences of flood and its impact on their livelihood. besides, the government’s awareness programs are often superficial and perfunctory, do not lay focus on households’ attitudinal change. it does neither focus on which adaptive measures should they follow, nor which new opportunities are available to them to earn their livelihood. it only focuses on preventive measures, which though provide temporary relief to households, but does not offer any effective or lasting solution to livelihood resilience of households to reduce vulnerability. the government sponsored flood mitigation programs do not focus further on livelihood promotion. the result of the study shows that flood preparedness mediates between flood education and livelihood resilience. though, there is no formal flood education in the blocks of the study area, households informally learn from the experience of elders and senior members of the community about what steps should be taken in what kind of flood situation, how to mobilize resources, what measures should be adapted to protect life and property of households effectively, published by atlantis press copyright: the authors 61 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india what preparatory measures should be taken to meet flood challenges, and how to approach political leadership and other external agencies to support livelihood resilience of the affected households etc. all potential problems are discussed in informal village gatherings and village chitchat sessions at length along with their possible solutions to safeguard the interests of the households in the flooding situations. in fact, households realize the significant role of elders and senior members of the community in creating awareness, dissemination of information, and flood preparatory mechanism more than what may be achieved through formal system of education (fig. 7). fig. 7: households take their livestocks to safer place(s) however, households do acknowledge the importance of flood education with regard to various aspects of flooding. i.e., knowledge about potential impact of flooding, training to effectively handle flood situation, viability of various strategies to meet flood challenges and how to make preparation, and how to make optimal use of available resources in the given situation of flooding etc. in the opinion of majority of the households, though education about flood does exist informally, formal flood education may further add, supplement, provide or train ‘flood folks’ 1 about how to behave them in a flooding situation, and how to act effectively to meet flood challenges. in the community, threadbare discussions take place with regard to pros and cons of various strategies of livelihood resilience and various strategic measures the community should initiate to safeguard their interest and contain flood fury. households’ live with the flood ‘as a way of life’. resources are pulled, specific steps of action are chalked out, round the clock monitoring of floodwater is arranged, responsibilities are 1 it refers to a group of households who are affected by flood. distributed and all households are asked to standby if situations so warrant. the community is found to remain alert to collectively act to meet any flood eventuality. throughout the entire flooding period, the community remains together and jointly initiates actions to support each other at the time of flooding. loud speakers are used to update households, and petromax and other lighting arrangements are made to keep close watch over flood development. households flood preparedness includes storage of relief materials, storage of seeds, monitoring of floodwater, and exchange of information about safe places of shelter. flood preparedness may further be improved and flood impacts be lessened by making arrangements for formal flood education, by organizing awareness programs, and by issuing early warning to households to remain prepared to meet any flood eventualities. the study envisages the necessity for accurate flood education program to create awareness in households in respect to flood and its accompanying problems as it may install confidence and create hope in households in the situation of despair (berke, 1998; burby et al., 2000; botzen, 2009). in the opinion of households, since the government relief assistance always comes late it may be taken care of by arranging a camp office at the site of flooding. the camp office may be interested with the responsibility to maintain liaison with the flood headquarters of the government to continuously update the government about the flood progress. a roster of households may be maintained closely monitor and watch the flood movement for onward transmission to the flood headquarters. 6. conclusions flood experience of households and their knowledge work as strategic inputs in their livelihood resilience. flood experience is acquired from their frequent exposure is far more the most effective and powerful weapon, which households use advantageously to fight against flood vagaries. it is used as a strategic resource input by them in their livelihood resilience. it is a process of continuous learning, which creates confidence in households and enables them to adapt measures to meet any flood eventuality. as a result, they do not perceive flood as a threat because of their experience and knowledge and their confidence in finding ways to overcome it. households adapt the strategy of ‘watch and act’ and accordingly remain published by atlantis press copyright: the authors 62 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india prepared with contingency plans to overcome flood impacts. although, flood education is not found to exist formally, younger persons informally learn from the experience of senior persons about the preventive measures to be adapted and likely to be effective in coping with flood. wide discussion takes place among the households and in the community with regard to proactive measures when flood is visualized and considered inevitable. livelihood resilience depends on the ingenuity of the households, the support they receive from the community and the government, and how optimally do they use their flood experience and knowledge. in the flooding situation, the community acts together as a cohesive unit, for all being risked to the same problem. the long experience of flood enables affected households’ learning and adaptation of suitable ways and means to livelihood resilience, which may further help policy makers in devising effective strategy. therefore, there is a need to take benefit of the strategic value of people’s indigenous knowledge and flood experience in livelihood resilience, as it does not depend only on distribution of relief materials, but more so depends on people’s access to resources and their attitude and aspirations for livelihood resilience. the government focuses on preparedness activities but does not promote livelihood diversification strategy, which may help households in generating income and different livelihood activities. moreover, lack of formal flood education programs does not pose any problem in households’ awareness as it takes place in different informal meetings of the households. in order to meet flood challenges, involvement of all the stakeholders need to be duly considered at all levels of flood operations. people’s experience and knowledge to flood which come in handy and guide them at each step in their efforts to reestablish livelihood may be advantageously utilized in strategic intervention by the government. it helps households in their strategies, according to situational requirement. 7. limitations the study though shows risk perception as mediator, but risk perception changes over time and cognitive emotions may enhance or lessen after certain time and thus, may influence livelihood resilience. this may be taken care-off by making pre and post studies, but this could not be done in this present study. an effective model by involving different stakeholders and by incorporating local issues including households’ strategies may be prepared for different phases of livelihood resilience. acknowledgement we are very grateful to iit kharagpur for providing fund to carry out the research. references s. c. s. abraham, p. sheeran, d. abrams and r. spears, exploring teenagers’ adaptive and mal adaptive thinking in relation to the threat of hiv infection, psychology and health 9 (1994) 253-272. d. aboagye, t. dari and j. koomson, risk perception and disaster management in the savannah region of ghana, international journal of humanities and social sciences 3(3) (2013) 157-172. w. a. anderson, some observations on a disaster subculture: organizational response of cincinnati ohio to the 1964 flood. (note bc. columbus, ohio: the disaster research center, the ohio state university) 1965. d. bauman, determination of the cost effectiveness of flood hazard information, papers and proceedings of the applied geography conference 6 (1983) 292. m. s. bartlett, a note on the multiplying factors for various chi square approximations, journal of the royal statistical society 16 (1954) 296-298. v. basolo, l. j. steinberg, r. j. burby, j. levine, a. m. cruz and c. huang, the effects of confidence in government and information on perceived and actual preparedness for disasters, environment and behavior 41 (2009) 338-364. http://dx.doi.org/ 10.1177/ 001 391650 8317222. f. berkes, understanding uncertainty and reducing vulnerability: lessons from resilience thinking, natural hazards 41(2) (2007) 283-95. f. berkes and c. folke, linking social and ecological systems: management practices and social mechanisms for building resilience. (cambridge university press, cambridge, uk) 1998. d. botha, d. van niekerk, g. wentink, t. tshona, y. maartens, k. forbes, e. annandale, c. coetzee and e. raju, disaster risk management status assessment at municipalities in south africa, south african local government association (salga), pretoria, 2011. l. s. bosher, a. r. j. dainty, p. m. carrillo, j. glass and a. d. f. price, attaining improved resilience to floods: a proactive multi-stakeholder approach, disaster prevention and management18 (1) (2009) 9-22. w. j .w. botzen, j. c. j. h. aerts and j .c. j. m. van den bergh, willingness of homeowners to mitigate climate risk through insurance, ecological economics 68 (2009) 2265-2277. published by atlantis press copyright: the authors 63 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india m. brilly and m. polic, public perception of flood risks, flood forecasting, and mitigation, nat hazards earth system science 5 (3) (2005) 345-355. r. a. bradford et al, risk perception-issues for flood management in europe, natural hazards and earth system science 12 (7) (2012). p. bubeck, w. j.w. botzen, h. kreibich and j. c. j. h. aerts, detailed insights into the influence of floodcoping appraisals on mitigation behavior, global environmental change (2012),doi:10.1016/j.gloen vcha.2013.05.009. r. j. burby, cooperating with nature: confronting natural hazards with land use planning for sustainable communities, united states: joseph henry press, 1998a. r. j. burby, natural hazards and land use in an introduction in cooperating with nature: confronting natural hazards with land use planning for sustainable communities, ed. r.j. burby (united states: joseph henry press, 1998b). pp. 1-28. d. w. cash and s. moser, linking global and local scales: designing dynamic assessment and management processes, global environmental change (10) (2000) 109-120. r. b. cattell, the scree test for the number of factors, multivariate behavioral research 1 (1966) 245-276. s. l. cutter, living with risk: the geography of technological hazards, new york: edward arnold, 1993. h. chongfu, multiple internet of intelligences for risk analysis, journal of risk analysis and crisis response 4 (2) (june 2014) 61-71. b. de. marchi, a. scolobig, g. delli zotti del and m. zotto, risk construction and social vulnerability in an italian alpine region, floodsite report, 2007. n. dufty, a new approach to community flood education, the australian journal of emergency management 23 (2) (2008). h. a. elliott et al., improving flood warning -which way forward?, the australian disaster conference canberra, 2003. f. ellis, survey article: household strategies and rural livelihood diversification, journal of development studies 35 (1) (1998)1-38. c. e. faupel, s. p. kelley and t. petee, the impact of disaster education on household preparedness for hurricane hugo, international journal of mass emergencies and disaster 10 (1) (1992) 5-24. j. k. fielding gray, k. burningham and d. thrush, flood warning for vulnerable groups: secondary analysis of flood data, environment agency r and d report w5c-018/2.environment agency, bristol, uk, (2005). c. folke, freshwater and resilience: a shift in perspective. philosophical transaction of royal society, london series b 358 (2003) 2027-36. a. gissing, m. morgan and c. ronan, planning for the inevitableemergency planning for floods in new south wales, proceedings of the 47th floodplain management conference, gunnedah, 2007. t. grothmann and f. reusswig, household at risk of flooding: why some residents take precautionary action while others do not, natural hazards 38 (2006) 101-120. t. grothmann and a. patt, adaptive capacity and human cognition: the process of individual adaptation to climate change, global environmental change part a15 (3) (2005) 199-213. j. harvatt, j. petts and j. chilvers, understanding householder responses to natural hazards: flooding and sea level rise comparisons, journal of risk research 14 (1) (2011) 63-83 doi:10.1080/1366 987 7.2010.503935. a. heidari, structural master plan of flood mitigation measures, natural hazards and earth system sciences 9 (2009) 61-75. j. h. heinz, iii. the hidden costs of coastal hazards: implications for risk assessment and mitigation. center for science, economics, and the environment (washington, dc, island press, 2000), pp. 252. r. hertwig, g. barron, e. u, weber and i. erev, decisions from experience and the effect of rare events, psychological science 15 (2004) 534-539. m. ho, d. s. shaw, lin and y. chiu, how do disaster characteristics influence risk perception? risk analysis 28 (3) (2008) 635-643. h. hung-chih, the attitude towards flood insurance purchase when respondents’ preferences are uncertain: a fuzzy approach, journal of risk research 12 (2009) 239-258, doi:10.1080/13669 870802497702. e. l. jackson, response to earthquake hazard: the west coast of north america, environment behavior 13(4) (1981) 387416. g. kaiser, s. reese, h. j. markau and h. sterr, public perception of coastal flood defense and participation in coastal flood defense planning unpublished final report, 2004. d. keogh, a. apan, s. mushtaq, d. king and m. thomas, resilience, vulnerability, and adaptive capacity of an inland rural town prone to flooding: a climate change adaptation case 3 studies of charleville, queensland, australia. natural haz ards 59 (2011) 699-723. a. kirschenbaum, disaster preparedness: a conceptual and empirical reevaluation, international journal of mass emergencies and disasters 20 (2002) 5-28. r. j. t. klein, r. j. nicholls and f. thomalla, the resilience of coastal megacities to weather-related hazards: a review in building safer cities: the future of disaster risk, eds. a. kreimer., m. arnold, a. carlin, (disaster risk management series no. 3, world bank, washington, dc, usa, 2003). pp. 101-120. j. k. lazo, d. m. waldman and b. h. morrow, household evacuation decision making and the benefits of improved hurricane forecasting: developing a framework for assessment, american meteorological society 25 (2010) 207-219 doi: 10.1175/ 2009 waf 2222310.1. c. lengnick-hall and t. beck, adaptive fit versus robust transformation: how organizations respond to environmental change, journal of management 31 (5) 738-757 (2005). published by atlantis press copyright: the authors 64 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india m. k. lindell and r. w. perry, households adjustment to earthquake hazard, environment behavior 32 (2000) 590-630. c. w. lin, s. t. lu, t. s. shih, w. h. lin, y. c liu and p. t. chen, active faults of central taiwan, centre for geological survey 21(2008) 148. d. loseke, thinking about social problems: an introduction to constructionist perspectives. new york: aldine de gruyter, 1999. p. longstaff, security, resilience, and communication in unpredictable environments such as terrorism, natural disasters, and complex technology. syracuse, ny: author, 2005. j. ludy and g. kondolf, flood risk perception in lands “protected” by 100-year levees, natural hazards 61 (2012) 829-842. b. maguire and p. hagan, disasters and communities: understanding social resilience, the australian journal of emergency management 22 (2007)16-20. a. j. milne, h. h. wollford, j. masonand and e. hatzidimitriadou, early diagnosis of dementia by gps: an exploratory study of attitudes, aging and mental health 4 (2000) 292-300. r. miceli, i. sotgiu and m. settanni, disaster preparedness and perception of flood risk: a study in an alpine valley in italy, journal of environment psychology 28 (2008) 164-173. d. mileti, disasters by design: a reassessment of natural hazards in the united states. washington, dc: joseph henry press, 1999. a. niehof, the significance of diversification for rural livelihood systems, food policy 29(2004) 321-338. f. h. norris and s. a. murrell. prior experience as a moderator of disaster impact on anxiety symptoms in older adults, american journal of community psychology 16 (1998) 665-683. h. nyakundi, s. mogere, i. mwanzo and a. yitambe, community perceptions and response to flood risks in nyando district, western kenya, jàmbá. journal of disaster risk studies 3(1) (2010) 346-365. doi: 10.4102 /jamba.v3i1.35. d. paton, l. smith, d. johnston, m. johnston and k. ronan, developing a model to predict the adoption of natural hazard risk reduction and preparatory adjustments (research project no. 01-479) eqc research report (2003). r. i. palm, public response to earthquake hazard information, annals of the association of american geographers 71(3) (1981)389-399. e. pagneux, g. gısladottir and s. jonsdottir, public perception of flood hazard and flood risk in iceland: a case study in a watershed prone to ice-jam floods, nat. hazards 58 (2011) 269-287. d. paton, j. mcclure and p. t. burgelt, natural hazard resilience: the role of individual and household preparedness in disaster resilience: an integrated approach, eds. d. paton and d. johnston, (springfield, illinois: charles c. thomas 2006). pp.105-124. r. w. perry and m. k. lindell, the psychological consequences of natural disaster: a review of research on american communities, mass emergencies 3 (1978) 105-115. t. plapp and u. werner, understanding risk perception from natural hazards: examples from germany in risk 21, coping with risks due to natural hazards in the 21st century, eds. w. j. amman., s. dannenmann and l. vulliet, (proceeding of risk 21 workshop 2004, taylor and francis, london, 2006) pp. 101108. r. raaijmakers, j. r. krywkow and a. van der veen, flood risk perceptions and spatial multi-criteria analysis: an exploratory research for hazard mitigation, nat. hazards 46 (2008) 307-322. b. reid, w. smit, b. o. caldwell and s. belliveau, vulnerability and adaptation to climate risks in ontario agriculture, mitigation adaptation strategy global change 12 (2007) 609-637.doi10.1007/s 11027-006-9051-8. r. w. rogers, cognitive and physiological processes in fear appeals and attitude change: a revised theory of protection motivation in social psychophysiology, eds. j. cacioppo and r. petty (new york: guilford press, 1983). i. ruin, j. c. gaillard and c. lutoff, how to get there? assessing motorists’ flash flood risk perception on daily itineraries, environment hazards 7(3) (2007) 235-244, doi:10.1016/j. envhaz.2007.07.005. c. saavedra and w. budd, climate change and environmental planning: working to build community resilience and adaptive capacity in washington state, usa, habitat international 33(3) (2009) 246-252. m. siegrist and h. gutscher, natural hazards and motivation for mitigation behavior: people cannot predict the affect evoked by a severe flood, risk analysis 28 (3) (2008) 771-778,doi: 10.1111/j. 15396924.2008.01049.x. p. slovic, perception of risk reflections on the psychometric paradigm in social theories of risk, eds. s. krimsky., and d. golding praeger, (westport 1992)pp.117-152. p. slovic, m. l. finucane, e. peters and d.g. macgregor, risk as analysis and risk as feelings: some thoughts about affect, reason, risk, and rationality, risk analysis 24 (2) (2004) 311-322. b. smit and j. wandel, adaptation, adaptive capacity and vulnerability, global environmental change 16 (2006) 282-292. d. i. smith, j. w. handmer, j. m. mckay, m. a. d. switzer and b. j. williams, non-destructive measures in australian urban floodplain management: barriers and recommendations, the australian journal of emergency management 11(2) (1996) 51-56. r. srivastava and l. laurian, natural hazard mitigation in local comprehensive plans: the case of flood, wildfire and drought planning, disaster prevention and management: an international journal 15 (3) (2006) 461-483. m. n. srinivas, caste, its twentieth century avatar, new york, penguin books (1996). r. a. stablling. promoting risk: constructing the earthquake threat. new york: aldine de gruyter, 1995. published by atlantis press copyright: the authors 65 madhuri et al. / living with flood: a livelihood resilience approach of rural people in bihar, india s. tapsell, the health effects from fluvial flooding. report to the environment agency, flood hazard research centre, enfield, 2001. l. l. thurstone, multiple factor analysis. chicago: university of chicago press, 1947. k. tierney and m. bruneau, conceptualizing and measuring resilience: a key to disaster loss reduction.tr news, may-june 2007, 14-1. p. timmerman, vulnerability, resilience and the collapse of society: a review of models and possible climatic applications. institute for environmental studies, university of toronto, canada, 1981. e. l. tompkins and w. n. adger, responding to climate change: implications for development, id21 insights 53 (2004) 4-4. r. h. turner, j. m. nigg and d. h. paz, waiting for disasters: earthquake watch in california. university of california press, berkeley, california, usa, 1986. unisdr, global assessment report on disaster risk reduction, united nations international strategy for disaster reduction secretariat: geneva, 2009. a. vari and z. ferencz, flood research from the social perspective: the case of the tisza river hungary in flood research: le point de la recherche sur les crues, eds. i. tchiguirinskaia., ni. ni. thein and k. p. hubert (frontiers in iahs press, 2006) pp.155172. b. walker et al., resilience, adaptability, and transformability, ecology and society 9 (2) (2004) 5,http://www. ecology andsociety.org/vol9/iss2/art5. g. wachinger and o. renn, risk perception and natural hazards, caphaz-net wp3 report, dialogik non-profit institute for communication and cooperative research, stuttgart,2010klk, (2010), http://caphaz-net.org /outcomes results/caphaznet_ wp3_ riskpercep tion.pdf. n. d. weinstein, effects of personal experience on selfprotective behavior, psychological bulletin 105 (1989) 31-50. published by atlantis press copyright: the authors 66 1. introduction 1.1. flood experience 1.2. flood education 2. study area and sampling strategy 2.1. focus group discussion 2.2. measures 3. results and analysis 3.1. preliminary data analysis 3.2. reliability of the measurement scales 3.3. correlations between variables 4. factor analysis 4.1. exploratory factor analysis 4.2. latent variable structural equation modeling (lvsem) 5. discussions 6. conclusions 7. limitations << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <feff05d405e905ea05de05e905d5002005d105d405d205d305e805d505ea002005d005dc05d4002005db05d305d9002005dc05d905e605d505e8002005de05e105de05db05d9002000410064006f006200650020005000440046002005d405de05d505ea05d005de05d905dd002005dc05d405d305e405e105ea002005e705d305dd002d05d305e405d505e1002005d005d905db05d505ea05d905ea002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e05d005de05d905dd002005dc002d005000440046002f0058002d0033002c002005e205d905d905e005d5002005d105de05d305e805d905da002005dc05de05e905ea05de05e9002005e905dc0020004100630072006f006200610074002e002005de05e105de05db05d90020005000440046002005e905e005d505e605e805d5002005e005d905ea05e005d905dd002005dc05e405ea05d905d705d4002005d105d005de05e605e205d505ea0020004100630072006f006200610074002005d5002d00410064006f00620065002000520065006100640065007200200035002e0030002005d505d205e805e105d005d505ea002005de05ea05e705d305de05d505ea002005d905d505ea05e8002e> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <feff0055007300740061007700690065006e0069006100200064006f002000740077006f0072007a0065006e0069006100200064006f006b0075006d0065006e007400f300770020005000440046002000700072007a0065007a006e00610063007a006f006e00790063006800200064006f002000770079006400720075006b00f30077002000770020007700790073006f006b00690065006a0020006a0061006b006f015b00630069002e002000200044006f006b0075006d0065006e0074007900200050004400460020006d006f017c006e00610020006f007400770069006500720061010700200077002000700072006f006700720061006d006900650020004100630072006f00620061007400200069002000410064006f00620065002000520065006100640065007200200035002e0030002000690020006e006f00770073007a0079006d002e> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word introduction for volume 3, issue 3 editor’s introduction during a slow economic recovery in developed countries appears and the average growth in emerging countries keep down, we publish the present issue, which only contains 4 papers. the first two contributions are written in english and the last two in chinese with english abstracts. the first paper, “risk assessment for a chemical spill into a river” by mccready and williams, advises that a chemical spill can be a significant threat to human health and the aquatic environment. it describes a mathematical model to predict chemical concentrations and it presents an approach to characterize the risk as acceptable or unacceptable. the second paper, “the volatility of viet nam listed banking, insurance and financial services company groups after the financial crisis 2009-2011” by huy, shows us the banking and financial market in viet nam after the crisis during the period 2009-2011. there are decreases in equity and asset beta var values in the insurance industry, less risk dispersion, but there are increases in asset beta mean value in the investment and finance industry (from 0.41 to 0.52) and in equity beta mean value (from 0.81 to 0.96) in this industry. however, these numbers are reduced for the whole period 2007-2011. it is noted that these data in the banking industry do not vary much. in the stock investment industry, equity and asset beta mean values are slightly increasing (from 0.51 and 0.48 to 0.56 and 0.53). the market risk has been affected by the crisis and the impact is extended during the period 2009-2011 and for the whole period 2007-2011, the market risk decreases in the investment and finance industry. the third paper, “mcsim-based occupational health risk assessment on benzene” by li, huang and liu, employs the monte carlo simulating software (mcsim) to build a model for estimating the concentrations in target tissues of human in distribution forms. furthermore, the authors use fluent software to simulate the external dose, so that the internal dose estimated by mcsim can be used to calculate the lifetime cancer risk range of exposed workers if the benzene occupational exposure scenarios are complicated. the final paper, “on the coastal erosion risk assessment indexes” by xu, wen, zhao and xu, tells us that erosion has taken place widely along coastal area in china since 1970s, which results in the width of beach being narrowed, bathing beach destruction, coast-protection facilities and roads collapse, and wetland deterioration. the authors suggest the coastal erosion risk assessment index system which can be divided into hazard index system and vulnerability index system separately. the index system is expected to provide a practical theoretical basis for coastal erosion risk assessment. many thanks to all of referees who given strong support to jracr. and also thank the authors very much for their submissions and revising. editor-in-chief chongfu huang professor, beijing normal university email: hchongfu@gmail.com journal of risk analysis and crisis response, vol. 3, no. 3 (november 2013), 115 published by atlantis press copyright: the authors 115 atlantis press journal style editor’s introduction for volume 5, issue 2 this issue contains six papers. there are three contributions written in english and three contributions in chinese with english abstracts. the papers can be divided into four topics: risk-based mitigation planning, supply chain management, literature statistics and vulnerability assessment. there are three papers on risk-based mitigation planning. the first paper “modeling of municipal drainage and urban channel flooding in coastal city in the south of china” by xiyan ren, demiao yu, yongjun ruan, et al., puts forward a numerical model which is formatted including ocean, rain drainage, urban channels, and pumps by the software mike flood. the second paper “pluvial flooding risk analyzing and the solutions to risk mitigation for dangyang city in china” by yuhong li, lian shi, jiazhuo wang, et al., introduces urban storm water drainage and localized pluvial flooding mitigation planning of dangyang city of hubei province in china. with the help of arcgis, mike flood and other software, this paper found that the two main courses of pluvial flooding in dangyang city are the flash floods around the city and the insufficient drainage river. this paper also gives several solutions for mitigation pluvial flooding risk based on analyzing the cause of pluvial flooding, identifying risk area and assessing the risk of pluvial flooding in dangyang city. the third paper “utilization of flood simulation technique in urban flood warning — a case study on fuzhou” by shuaijie li, yingxia xie, xiaotao chengl, studies an emergency plan for flood control. the flood simulation technique was applied in urban flood warning and emergency response to improve the practicability and operability of the emergency plan. several calculation schemes, involving flood scenarios with high possibility or the most dangerous situations, were designed for the case study. there is one paper in supply chain management, titled “a study of the role of customs in global supply chain management and trade security based on the authorized economic operator system” by liwen chen, yongfei ma, that conducts an in-depth study of customs management thinking regarding the aeo system and actively explores its implementation and integration with common practices of international customs standards and fulfilling the requirements of any international obligations. the paper “disaster risk research literature on statistics analysis in china journal net”, by shuzhen li and alateng tuya, features the literature in china related to disaster risks which are collected from china national knowledge infrastructure (cnki). statistical analysis shows most studies in disaster risks are beginning from 2007, when the number of documents increased. core journal literature citations are high, indicating that the field has a central effect, and receives attention in academia. also there is one paper on vulnerability assessment. the paper “gis-based assessment of vulnerability to landslide hazards in lushan earthquake-stricken areas” by mingtao ding and cheng miao, establishes a gis-based system of vulnerability assessment to landslide hazard for lushan earthquake-stricken areas. they extract six vulnerability assessment indexes including population, economic, and road densities, building and forestland coverage, and landslide hazard affected areas. the results show that the distribution of landslide hazard affected areas is closely correlated to the intensity of human engineering activities. we thank the referees for their strong support and kind help. also, very much thanks to the authors for all their submissions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 5, no. 2 (july 2015), 73 published by atlantis press copyright: the authors 73 mailto:jracr.srachina@126.com atlantis press journal style received 17 june 2016 accepted 18 july 2016 route guidance map for emergency evacuation lakshaya*, amit agarwalb, nomesh b. boliaa adepartment of mechanical engineering, indian institute of technology delhi, india, hauz khas, new delhi, india-110016 e-mail: lakshaytaneja01@gmail.com btransport systems planning and transport telematics, technische universität berlin, germany 10587 abstract an efficient process of emergency evacuation must be guided. in the event of an evacuation instruction, a significant amount of time is spent by evacuees looking for a place of relative safety or an exit. due to the ensuing stress and confusion evacuees try to follow others, consequently, all the exits are not used effectively. therefore, it is important to develop a route guidance map for the emergency. the focus of the map is to help both, the evacuees and the authorities to perform evacuation efficiently. this paper presents a route guidance map for pedestrians that aims an efficient evacuation in case of an emergency. an agent-based simulation framework is used for the simulation of various scenarios to prepare the guiding map. a real world case study of sarojini nagar, delhi is presented to test the presented methodology. eventually, several strategic recommendations are provided for improving safety of existing infrastructure. keywords: disaster preparedness, public safety, simulation, emergency, evacuation plan, strategic planning. 1. introduction in recent years, public safety and disaster preparedness have become a prime focus for national authorities, urban planners and civic agencies due to losses of human lives. in year 2014, at least 32 people were killed and 26 injured in a stampede shortly after the celebration of festival dussehra in patna, india (express news service, 2014). there are many similar examples across the world (see table 1 for similar examples), where due to lack of efficient evacuation planning, people have suffered. the recurring stampedes occur mainly at places of mass gatherings for example religious places, railway stations, sports/political/social events etc. there are many causers * corresponding author and triggers for the crowd disaster including structural design, fire, rumors, and sudden mass evacuation (ndma, 2014). evacuation is a process in which endangered people are moved from a dangerous place to a safe place in order to reduce the vulnerability during these dangerous circumstances. in order to mitigate impacts of disasters, proper evacuation planning is required. in many of the past events, lack of evacuation planning has resulted in loss of human lives, particularly in india (see table 1). improper selection of exit or failure to avoid the obstacles may lead to either serious injury or death. therefore, a proper route guidance map in terms of an “evacuation plan” is required that can help evacuees to find the suitable exits and the route to be followed to evacuate the journal of risk analysis and crisis response, vol. 6, no. 3 (october 2016), 135-144 published by atlantis press copyright: the authors 135 lakshay et al. / route guidance map for emergency evacuation endangered area in minimum time and with minimum loss of life. this guiding map can also serve as a reference for security staff to guide the evacuees on the route to take for evacuation. this evacuation plan may also suggest the structural improvements, which can be helpful for further reduction of the evacuation time. limited research on developing evacuation plan has been done and reported in literature. the ability to evacuate people depends mainly on two factors viz.: structural design and behavior. inefficient design and panic behavior may lead to overcrowding, which in turn may lead to crushing, suffocation and trampling. besides planning the infrastructure efficiently, it is also essential to understand the movement and flow behavior, which may help planners and civic agencies to reduce the severity. evacuation, where there may be a transition from normal behavior to irrational panic behavior, is governed by factor of “nervousness” which leads to slow down the crowd and tendency to follow others (helbing et al., 2002). a great share of literature focus is on simulating a single room evacuation pattern (casadesús et al., 2009; takahashi et al., 1989; taylor, 1996) where in true sense little evacuation planning take place. on the other hand, to evacuate a larger area, egress route have to be defined first, which requires optimization techniques. in a similar research direction, this study aims to investigate sudden mass evacuation from a crowded place. this paper presents a real-world case study for evacuation preparedness due to disastrous events in large-scale pedestrian areas. a majority of crowd disasters have occurred at shopping malls, music concerts, and stadium in developed countries (ndma, 2014). with increasing population, developing countries are also susceptible to crowd disaster at such venues (ndma, 2014). therefore, main focus of this study is to evacuate persons from congested areas such as market places or mass gathering venues. the objective is to make recommendations to improve the evacuation time of all people in the identified area. the key outcome is an “evacuation plan” for designated sites. the event of potential bombing is used as an example of the disaster where evacuation is required. the methodology presented in this study is applied to a market place, sarojini nagar, new delhi, however, it can be applied to any scenario wherever evacuation is required. also, in order to check the applicability and robustness of the approach, the same methodology is applied to two other areas namely lajpat nagar and laxmi nagar, new delhi. table 2 shows several models that have been used in the past to reduce the response time for an evacuation. in a study by flötteröd & lämmel (2010), the authors suggested to adopt dynamic traffic assignment model to develop an evacuation plan for open spaces. in general, table 1. past mishappenings due to improper evacuation modeling. year place reason casualties 1903 iroquois theatre fire, chicago (u.s) no exit signs; no emergency lighting; exit routes were confusing (disaster, 2015) 602 1913 italian hall disaster, michigan (u.s) escape from a falsely shouted of fire at a party (halldisaster, 2016) 73 1995 dabwali, haryana (india) synthetic tent caught fire, blocking main entrance (ndma, 2014) 446 1997 uphaar cinema, delhi (india) smoky cinema hall (ndma, 2014) 59 2000 night club lisbon (portugal) head for main exit and ignore alternative exit (helbing et al., 2002) 7 2006 jamrat bridge (saudi arabia) overcrowding and poor crowd management (still, 2016) 363 2008 chamunda devi temple, jodhpur, rajasthan (india) stampede due to false rumors of bomb (ndma, 2014) 249 2010 love parade (duisburg) trying to escape the overcrowded tunnel (still, 2016) 21 2011 amri hospital, calcutta (india) basement fire, suffocation causing deaths (ndma, 2014) 89 2014 patna stampede (india) mass exit from a single gate and rumors also that live electric had fallen on ground. (express news service, 2014) 33 2015 mina stampede blockage of route to jamrat bridge (still, 2016) 2110 published by atlantis press copyright: the authors 136 lakshay et al. / route guidance map for emergency evacuation dynamic traffic assignment relies on microscopic models. zheng et al. (2009) compares several modeling approaches at different scopes, for e.g. a) cellular automata models based on lattice gas or social force models b) agent-based models based on cellular automata or social force models etc. most of these models are detailed models which are resource hungry and need higher computational time. however, the aim of the present study is to develop and test a route guidance map using an approach that is computationally efficient for large-scale scenarios. therefore, the present study uses an evacuation planning approach in a multi-agent simulation based framework (lämmel et al., 2010). the multi-agent systems are preferred for crowd simulation modeling (almeida et al., 2013).this approach has also been applied to a real-world evacuation scenario of patna city, india (agarwal and lämmel, 2016) under mixed traffic conditions. this simulation framework is suitable for large scale scenarios due to its queuing model (see agarwal et al. (2015); balmer et al., (2009) and also section 2.1). every person in the area under consideration may not be familiar with the prevailing traffic conditions and alternative exit routes during evacuation situation, therefore, this study proposes an evacuation plan and subsequently, investigates the response time when this evacuation plan is used under different situations. the rest of the paper is organized as follows. first, the detail of simulation framework for the present study is explained in section 2. section 3 exhibits the methodology and the case study of sarojini nagar market is illustrated in section 4. section 5 shows the impact of “evacuation plan” and its usefulness. finally, the last section concludes the overall work and provides some outlook for future work. 2. evacuation modelling three different modelling approaches can be applied to an evacuation process (schadschneider et al., 2009): a) risk assessment, b) optimization and c) simulation. simulation of pedestrians is generally used for two purposes: to gain insight on a particular situation and to prove/disprove a hypothesis (still, 2007). the output of a simulation mainly includes: distribution of evacuation time, evacuation curves (number of people evacuated with respect to time), sequence of evacuation (snapshot at a specific time), and identification of congestion (schadschneider et al., 2009). in this article, multi-agent simulation framework (matsim) is used to identify the evacuation time, congested links and sequence of evacuation. in this, all evacuees are modeled as individual agents. a geographical information system (gis) based risk assessments information, planning system toolkit (grips) is used along with matsim (taubenböck et al., 2009). matsim has an evolutionary algorithm which consists of mainly three steps as shown in figure 1 table 2. method used in literature to reduce time. past study model description taylor (1996) macroscopic find minimum time to evacuate building and optimal plan in terms of exit usage. casadesús pursals & garriga garzón (2009) macroscopic find the distribution of exit usage for minimum evacuation time. takahashi, tanaka, & satoshi (1989) macroscopic optimal exit from a room is chosen for evacuating. han, yuan, chin, & hwang (2006) macroscopic routing for reducing total evacuation time. klüpfel (2003) macroscopic shows connection between choice of exit and individual egress time. stepanov & smith (2008) microscopic potential egress route described by kth shortest path using distance, travel time and level of congestion as objective function. abdelghany, abdelghany, & mahmassani (2014) microscopic show that evacuation time reduces significantly by optimizing the temporal distribution of evacuation and exit gate selection. kneidl, thiemann, hartmann, & borrmann (2011) microscopic find the probability of choosing a route to reduce the evacuation time. published by atlantis press copyright: the authors 137 lakshay et al. / route guidance map for emergency evacuation (balmer et al., 2009; horni, nagel, & axhausen, 2015). in this iterative cycle, an agent learns and adapts to the system. the minimal inputs are network and daily plans of the individual agents. • execution (mobsim) in this step, all the plans are simultaneously executed using predefined mobility simulation (mobsim) on the network. the network loading algorithm is a queuing model (cetin, burri, & nagel, 2003; gawron, 1998). 1 the queue model tracks every agent only at entry and exit and never in between which makes it computationally efficient. hence, a large-scale scenario can be simulated in reasonable computational time (agarwal, lämmel, & nagel, 2016a). • scoring various plans of an individual are compared using a utility function. the utility function consists of utility of performing an activity, (dis)utility of traveling etc. all executed plans are evaluated using the default scoring function (charypar & nagel, 2005). • re-planning for some of the agents, a new plan is generated by modifying an existing plan depending on the so-called innovative strategies (choice modules). several choice dimensions are available for e.g. reroute, time mutation, mode choice etc. the new plan is then executed in the next iteration. the innovation is used until a fixed number of iterations (for e.g. for 80% of the iterations). therefore, rest of the agents until innovation and all the agents after innovation select a plan from their choice set according to a probability distribution which converges to the multi-nomial logit model. 1 refer to (agarwal et al., 2016b, 2015)for details about the queue model and its extensions in the matsim. for simplicity, the present study uses first-in-first-out (fifo) approach of the queue model. 3. survey methodology typical crowd density at various sections of the road is estimated as illustrated further. a travel count survey data is conducted as follows to identify the initial person density on each link. 1) on every link of the road network, three surveyors are placed to count a) the number of persons present initially at time t, b) number of persons entering and c) number of persons leaving the link in 5 min time bin. 2) thus, number of persons on a road at any time t is given by equation (1) ( ) ( ) ( ) ( )n t t i t o tλ= + − (1) where, ( )n t is number of persons on a road at time t ; i(t) is number of persons entering the link in time bin, o(t) is number of person leaving in that time bin and λ(t) is number of persons initially present on the link at time t. 3) thus, total number of people to evacuate from a link is given by equation (2), which includes the persons on the road and also persons inside the shops. let s(t) be the number of people present inside shops on the link in time bin t (counted by fourth surveyor). then the total number of persons to evacuate on the concerned road tp(t) is: ( ) ( ) ( )tp t n t s t= + (2) 4) thus tp(t) is computed from the survey data. the pre-evacuation coordinates of all agents are assigned randomly on corresponding link. 5) in an evacuation problem, destination and route choice are interrelated. for the simplicity, in the present study, only one destination is used which reduces the whole problem to one dimension only. fig. 1. iterative appraoch of matsim (horni et al. 2015). published by atlantis press copyright: the authors 138 lakshay et al. / route guidance map for emergency evacuation the post-evacuation location coordinate (destination) of each agent is modeled as a virtual point formed far away from the center of the evacuation area. all the exits are connected to this artificial node termed as “super sink” (see figure 2). fig. 1. exit connected to a super sink (a virtual destination). 4. scenario set up sarojini nagar is located in the south west district of delhi. it is one of the most popular market in delhi. this was one of the site which was bombed in delhi on 29 october 2005 and resulted in many deaths and major injuries (nctc, 2006). thus, because of its past history, it is chosen as a site for potential disaster location. 4.1. case study: sarojini nagar market the surveys were conducted between 4 september and 7 september, 2014 as illustrated in section 3. generally, evacuation planning is composed of the following steps (lahmar, assavapokee, & ardekani, 2006): 1) impact zone: in this step, the evacuation zone is identified. it is generally dependent on the type of emergency. in some cases only small area needs to be evacuated while in some cases complete area needs to be evacuated, in the present study, for the case of potential bombing scenario, complete market is considered as evacuation area. figure 3 shows the complete market of sarojini nagar market. fig. 3. market area (in red) under consideration for evacuation. 2) assignment of evacuees to shelter: after defining area to be evacuated, next is to decide where to evacuate people. in the case study all the exits are assumed as a potential shelter. once the agent is out of the particular exit, he/she is assumed to be safe. 3) traffic routing (determining driving direction at each road): in this step, the best route to reach the shelter is determined. different strategies have been considered in the case study presented in the section 4.3. 4) self-evacuation: agents starts evacuating as soon as warning is announced. they follow the evacuation plan considered in various scenarios. 4.2. simulation inputs the sarojini nagar market area remains crowded most of the time of the day. motorized and non-motorized vehicles are rarely used inside this market area and therefore all vehicles are ignored in the present study. only the walk mode is considered in the simulation framework. • network the desired evacuation area of sarojini nagar market is taken from open street map (openstreetmap, 2015). all exits of this area are connected to a safe virtual destination, which is far away from the centre of selected area. all exits have published by atlantis press copyright: the authors 139 lakshay et al. / route guidance map for emergency evacuation same exit capacities. it is assumed that travel time from exit points to the safe virtual destination is zero. the network is converted to desired format of the simulation framework. the width of the streets are measured during survey and eventually, due to heavy encroachment, the effective width of link is estimated as 4 m which results in a capacity of about 1000 pcu/h per direction. the roads in the sarojini nagar market are partially tiled and partially concrete and are in good condition. a poor condition of the surface may increase the evacuation time. • plans in the evacuation situation, only two types of activities i.e. pre-evacuation and post-evacuation are considered. the activity locations of these activities are the locations of agents before and after the evacuation respectively. all agents use walk mode to travel between these activities. it is assumed that all agents start moving out of the market area as soon as warning is announced. initial positions and the density of the agents on each link is calculated from the survey (see section 3). in the simulation, the speed of the agent is assumed as 6 km/h. passenger car unit (pcu) of the agent is taken as 0.08 (tiwari, fazio, & gaurav, 2007). overall, about 8430 agents are evacuated from the market area. • choice dimension in this study, 20% of agents are allowed to change their route until 80% of the iterations. simulation is run for 100 iterations. rest of the agents until 80% iterations and all agents after that select a plan from their choice set only which stabilizes the demand. 4.3. scenarios the first step is to identify the bottleneck links in order to identify the cause and then propose necessary strategic decisions to rectify and improve the overall evacuation time. no single hypothetical scenario is expected to perfectly emulate a real event that will occur in future. thus, different situations are considered for evacuation of pedestrian in market place. these scenarios help in generating the evacuation plan for an open space environment. the following scenarios are considered. • scenario 1: (no evacuation plan): in absence of any evacuation plan, all agents are left to themselves. this would replicate the existing situation of the market area. • scenario 2: (shortest path): in this scenario, it is assumed that all agents will evacuate by running to the nearest exit, taking the shortest path between their current location and exit. this is recorded as the “shortest path evacuation time”. • scenario 3: (benchmark evacuation with encroachment): in this scenario, evacuation time is identified based on nash equilibrium (lämmel, rieser, & nagel, 2008). evacuation time calculated using this approach is termed as the “benchmark evacuation time”. in reality, it is not possible to achieve benchmark evacuation time (since this is a result of several iterations of matsim with learning of each outcome) but it is useful to generate a feasible evacuation plan and compare it with benchmark time. streets of the market area are heavily encroached therefore, in this scenario, the evacuation time is estimated with the existing situation. • scenario 4: (planned evacuation): as discussed before, in this scenario, an evacuation plan is proposed aiming to achieve the evacuation time same as benchmark evacuation time and in turn expecting to be better than shortest path scenario or no evacuation plan scenario. the resulting time is called “planned evacuation time”. this scenario will result in the development of a route guiding map. in case of an emergency, these routes can be followed from the current locations of all agents. • further, after analyzing the scenario based on the link flow in peak hours, more recommendations such as widening of bottleneck links by removing encroachments and adding new emergency exits will also help in further reduction in the evacuation time. 5. results it is clear that in the absence of any planning and signage (scenario 1), all agents may produce herding behaviour which results in early degradation of network supply. thus, evacuation time will be higher than all other scenarios due to sheer chaos. in scenario 2 (shortest path evacuation plan) everybody moves to their geographically nearest exit point. this kind of plan is most easy to implement, because of its unique solution. it is only required to put sign at crossing of street network. the big disadvantage of such strategy is that, it does not take congestion in consideration. congestion avoidance is important in case published by atlantis press copyright: the authors 140 lakshay et al. / route guidance map for emergency evacuation of evacuation. according to schadschneider et al. (2009), to reduce the congestion, two corrective actions can be taken: change of geometry (wider escape paths) and proper guidance through signage which helps in improving orientation capability. our methodology intervenes at two levels: it develops signage for the existing geometry and also makes recommendation on specific geometry change for faster evacuation. in the scenario 3 (benchmark scenario), the fastest route is computed using iterative algorithm of matsim. this kind of plan (benchmark) can only be implemented with proper training and repetitive mock drills, which is not feasible in practice. therefore, a practically feasible evacuation plan is proposed in scenario 4, in which consistent direction signs are placed at all relevant locations. a snapshot of such a plan is shown in figure 4. heuristic (colour scheme) and evacuation time of agents help in making the evacuation plan. routes that are closer to exit but with high congestion are bypassed, agents diverted to a route where there is lesser congestion. the main advantages of this plan over shortest path plan are that it considers the congestion effects into consideration and it is easy to implement. this route guidance map will also serve as a reference for concerned authorities to provide evacuation route related instructions to evacuees. different scenarios (from section 4.3) are compared based on total evacuation time and average evacuation time per person. the former is the total time to evacuate all the agents out of the evacuation area. statistically total evacuation time is not a good measure for finding effectiveness of a given evacuation strategy. thus, average evacuation time is required, which not only minimizes the response time but at the same time also maximizes the flow at given time (hamacher & tjandra, 2001). table 3 shows the results obtained from these scenarios. clearly, as expected, benchmark scenario has the least total evacuation time and average evacuation time. this corresponds to a first-best condition in which everyone knows the prevailing congestion conditions and the best route to exit. further, for the case of planned scenario, the total and average evacuation time is significantly shorter than shortest path scenario and marginally higher than benchmark scenario. the comparison of evacuation progress is shown in figure 5. it can be observed that evacuation time is the same for all three scenarios until 50% of the agents are evacuated. afterwards, evacuation progress for the shortest path scenario becomes slowest and evacuation progress of the benchmark scenario become the fastest. the links towards exit a (see figure 4) become bottlenecks in the shortest path scenario (observation from simulation output). thus to make an effective use of all the exits, some agents are diverted to another exit. the procedure is repeated for all other exits. in this way, planned scenario routing strategy is developed making use of benchmark routing strategy. the evacuation progress of planned scenario is marginally slower than the benchmark scenario. effectiveness of the approach: in order to see the effectiveness of the route guidance map, the same methodology has been applied to two other markets of delhi (india) namely, lajpat nagar and laxmi nagar. evacuation plan for these sites are developed. it can be fig. 4. guiding map: a feasible solution. table 3. average evacuation time per person and total evacuation time scenario evacuation time (min) average total shortest path 10.05 23.05 benchmark 8.94 16.46 planned 9.45 17.32 published by atlantis press copyright: the authors 141 lakshay et al. / route guidance map for emergency evacuation observed from table 4 that planned scenario response time is better than the shortest path evacuation strategy. thus, clearly, the methodology is transferable to any scenarios where such kind of short-notice evacuation is required. 6. discussion the present study shows the necessity of an evacuation plan for improving safety and response efforts. the study provides strategic and tactical recommendations to improve the response time in case of an emergency evacuation. strategic recommendations include increasing network supply side by making new routes or by widening the existing roads. these strategic recommendations help planners to decide the increase in the capacity of roads, or where an emergency exit should be made to further improve evacuation response. statistical and tactical recommendations deal with effective utilization of existing capacity. this can be achieved by properly routing the evacuee through a street network in order to minimize danger and ensure safety. the simulation returns the route assignment policy to reduce congestion and improve response time, which eventually will reduce the collateral damage. a policy imperative from this study is that even a static plan would help in reducing the evacuation time. further this work and methodology can also be used to determine the maximum allowed safe occupancy for an event in open area, a parameter that can be imposed by decision makers by way of policy. for this, a safe level of evacuation time must be determined through consultation with relevant experts. the evacuation time consists of two time components: reaction time and egress time (kuligowski, 2013). in the present study only latter is considered and estimated whereas it is assumed that the agents react instantly after the warning. this lays a future research direction to incorporate the different reaction times for different group of persons depending on the factors such as age and sex similar to the work by (agarwal, lämmel, & nagel, 2016b) in which the authors incorporated a uniform reaction time for all drives in the queue model. another important observation of the study is that the egress time is highly affected by heavy encroachments. clearly, removing these encroachments will ease some capacities and would reduce evacuation time significantly. in the literature, it has been argued that people tend to misjudge the likelihood of a disaster event and range of severity of its impact. this would in turn result in a different outcome; such behaviors are out of the scope of the present study. 7. conclusion evacuation time is a critical factor for developing evacuation strategies. in this work, a methodology to prepare a guidance map was developed using an agentbased simulation framework. initial inputs were calculated from different surveys. an event of potential bombing was considered as an example of the disaster where immediate evacuation is required due to a disaster on the same location in the past. a real-world case study of sarojini nagar market, delhi was considered. different scenarios were considered and their total and average evacuation time were compared. it was shown that with the help of proper signages in planned scenario; the total and average evacuation time would be significantly lower than shortest path or no evacuation plan scenarios and marginally higher than benchmark scenario. the planned scenario routing map would help fig. 5. plot of evacuation time for different scenario. table 4. total evacuation time for lajpat nagar and laxmi nagar markets. scenario total evacuation time (min.) lajpat nagar market laxmi nagar market shortest path 13.25 11.51 benchmark 10.01 7.34 planned 10.46 9.53 published by atlantis press copyright: the authors 142 lakshay et al. / route guidance map for emergency evacuation the authorities (the security staff) to guide or push evacuees. the effectiveness of the proposed methodology was shown using the same approach for two more markets of delhi. future work includes an analysis of the robustness of the suggested evacuation plan, particularly with respect to distribution of people in the market. development of a dynamic evacuation plan, that is, one with message signs that change dynamically with congestion distribution is another possible extension. with this, accounting for behavioral characteristics of agents, after developing appropriate models for the same, can induce even more realism in the recommendations. acknowledgments this project is funded by the human settlement management institute, housing and urban development corporation limited (hudco). the authors also would like to thank dr. gregor lämmel for his helpful comments. lastly the authors would like to acknowledge the contribution of prem chand and rama shankar for carrying out several supporting tasks throughout the duration of the project. references abdelghany, a., abdelghany, k. f., & mahmassani, h. s. (2014). modeling framework for optimal evacuation of large-scale crowded pedestrian facilities. european journal of operational research, 237(3), 1105–1118. http://doi.org/10.1016/j.ejor.2014.02.054 agarwal, a., & lämmel, g. (2016). modeling seepage behavior of smaller vehicles in mixed traffic conditions using an agent based simulation. transportation in developing economies, 2(2), 1–12. http://doi.org/10.1007/s40890-016-0014-9 agarwal, a., lämmel, g., & nagel, k. (2016a). incorporating within link dynamics in an agent-based computationally faster and scalable queue model. agarwal, a., lämmel, g., & nagel, k. (2016b). modelling of backward travelling holes in mixed traffic conditions. in v. l. knoop & w. daamen (eds.), traffic and granular flow ’15 (1st ed.). delft, nl: springer international publishing. http://doi.org/10.1007/978-3-319-33482-0_53 agarwal, a., zilske, m., rao, k. r., & nagel, k. (2015). an elegant and computationally efficient approach for heterogeneous traffic modelling using agent based simulation. procedia computer science, 52(c), 962–967. http://doi.org/10.1016/j.procs.2015.05.173 almeida, j. e., rosseti, r. j. f., & coelho, a. l. (2013). crowd simulation modeling applied to emergency and evacuation simulations using multi-agent systems. arxiv preprint arxiv:1303.4692. balmer, m., rieser, m., meister, k., charypar, d., lefebvre, n., & nagel, k. (2009). matsim-t: architecture and simulation times. multi-agent systems for traffic and transportation engineering. http://doi.org/10.1140/epjb/e2008-00153-6 casadesús pursals, s., & garriga garzón, f. (2009). basic principle for the solution of the building evacuation problem. journal of industrial engineering and management, 2(3), 499–516. http://doi.org/10.3926/jiem.2009.v2n3.p499-516 cetin, n., burri, a., & nagel, k. (2003). a large-scale agentbased traffic microsimulation based on queue model. charypar, d., & nagel, k. (2005). generating complete all-day activity plans with genetic algorithms. transportation, 32(4), 369–397. http://doi.org/10.1007/s11116-004-8287-y disaster. (2015). fire breaks out in chicago theater. retrieved december 1, 2015, from http://www.history.com/thisday-in-history/fire-breaks-out-in-chicago-theater express news service. (2014, october 4). patna stampede: 33 killed, 29 injured during dussehra celebration at gandhi maidan. patna. retrieved from http://indianexpress.com/article/india/india-others/32people-killed-15-injured-in-stampede-in-patna/ flötteröd, g., & lämmel, g. (2010). evacuation simulation with limited capacity sinks an evolutionary approach to solve the shelter allocation and capacity assignment problem in a multi-agent evacuation simulation. in proceedings of the international conference on evolutionary computation. gawron, c. (1998). an iterative algorithm to determine the dynamic user equilibrium in a traffic simulation model. international journal of modern physics c, 9(3), 393–407. halldisaster. (2016). the italian hall disaster, calumet, michigan. retrieved january 1, 2016, from http://www.genealogia.fi/emi/emi3d31e.htm hamacher, h. w., & tjandra, s. a. (2001). mathematical modelling of evacuation problems: a state of the art. (vol. 24). fraunhofer-institut für techno-und wirtschaftsmathematik, fraunhofer (itwm). http://doi.org/citeulike-article-id:6650160 han, l. d., yuan, f., chin, s. m., & hwang, h. (2006). global optimization of emergency evacuation assignments. interfaces, 36(6), 502–513. http://doi.org/10.1287/inte.1060.0251 helbing, d., farkas, i. j., molnár, p., & vicsek, t. (2002). simulation of pedestrian crowds in normal and evacuation situations. pedestrian and evacuation dynamics, 21, 21–58. retrieved from http://www.researchgate.net/publication/224010870_sim ulation_of_pedestrian_crowds_in_normal_and_evacuatio n_situations/file/d912f50eb0d9bb6224.pdf horni, a., nagel, k., & axhausen, k. w. (2015). the multiagent transport simulation matsim. retrieved from published by atlantis press copyright: the authors 143 lakshay et al. / route guidance map for emergency evacuation http://ci.matsim.org:8080/view/all/job/matsimbook/ws/main.pdf klüpfel, h. l. (2003). a cellular automaton model for crowd movement and egress simulation. tese de doutorado, duisburg-essem university. kneidl, a., thiemann, m., hartmann, d., & borrmann, a. (2011). combining pedestrian simulation with a network flow optimization to support security staff in handling an evacuation of a soccer stadium. in proceedings of the european conference forum 2011, cork (pp. 72–80). retrieved from 2011_kneidl_et_al_fbi.pdf kuligowski, e. (2013). predicting human behavior during fires. fire technology, 49(1), 101–120. http://doi.org/10.1007/s10694-011-0245-6 lahmar, m., assavapokee, t., & ardekani., s. (2006). a dynamic transportation planning support system for hurricane evacuation. in 2006 ieee intelligent transportation systems conference (pp. 612–617). http://doi.org/10.1109/itsc.2006.1706809 lämmel, g., grether, d., & nagel, k. (2010). the representation and implementation of time-dependent inundation in large-scale microscopic evacuation simulations. transportation research part c: emerging technologies, 18(1), 84–98. http://doi.org/10.1016/j.trc.2009.04.020 lämmel, g., rieser, m., & nagel, k. (2008). large scale microscopic evacuation simulation. pedestrian and evacuation dynamics, 2(2), 1–6. http://doi.org/10.1007/978-3-642-04504-2 nctc. (2006). nctc report on incidents of terrorism 2005. ndma. (2014). a guide for state government, local authorities, administrators and organizers. openstreetmap, f. (2015). openstreetmap. retrieved july 4, 2015, from www.openstreetmap.org schadschneider, a., klingsch, w., klüpfel, h., kretz, t., rogsch, c., & seyfried, a. (2009). evacuation dynamics: empirical results, modeling and applications. in encyclopedia of complexity and systems science (pp. 3142–3176). springer. http://doi.org/10.1016/j.ssci.2011.12.024 stepanov, a., & smith, j. m. (2008). multi-objective evacuation routing in transportation networks. european journal of operational research, 198(2), 435–446. http://doi.org/10.1016/j.ejor.2008.08.025 still, g. k. (2007). review of pedestrian and evacuation simulations. international journal of critical infrastructures, 3(3-4), 376–388. http://doi.org/10.1504/ijcis.2007.014116 still, g. k. (2016). crowd disasters. retrieved may 1, 2016, from http://www.gkstill.com/expertwitness/crowddisasters.ht ml takahashi, k., tanaka, t., & satoshi, k. (1989). an evacuation model for use in fire safety design of buildings. fire safety science, 2, 551–560. http://doi.org/10.3801/iafss.fss.2-551 taubenböck, h., goseberg, n., setiadi, n., lämmel, g., moder, f., oczipka, m., … klein, r. (2009). “last-mile” preparation for a potential disaster – interdisciplinary approach towards tsunami early warning and an evacuation information system for the coastal city of padang, indonesia. natural hazards and earth system science, 9(4), 1509–1528. http://doi.org/10.5194/nhess-9-1509-2009 taylor, i. r. (1996). a revised interface for evacnet+. in proceedings of 7th international fire science and engineering conference, interflam. tiwari, g., fazio, j., & gaurav, s. (2007). traffic planning for non-homogeneous traffic. in sadhana academy proceedings in engineering sciences (vol. 32, pp. 309– 328). http://doi.org/10.1007/s12046-007-0027-5 zheng, x., zhong, t., & liu, m. (2009). modeling crowd evacuation of a building based on seven methodological approaches. building and environment, 44(3), 437–445. http://doi.org/10.1016/j.buildenv.2008.04.002 published by atlantis press copyright: the authors 144 1. introduction0f( 2. evacuation modelling 3. survey methodology 4. scenario set up 4.1. case study: sarojini nagar market 4.2. simulation inputs 4.3. scenarios 5. results 6. discussion 7. conclusion acknowledgments references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /cmyk /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <feff005400610074006f0020006e006100730074006100760065006e00ed00200070006f0075017e0069006a007400650020006b0020007600790074007600e101590065006e00ed00200064006f006b0075006d0065006e0074016f002000410064006f006200650020005000440046002c0020006b00740065007200e90020007300650020006e0065006a006c00e90070006500200068006f006400ed002000700072006f0020006b00760061006c00690074006e00ed0020007400690073006b00200061002000700072006500700072006500730073002e002000200056007900740076006f01590065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f007400650076015900ed007400200076002000700072006f006700720061006d0065006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076011b006a016100ed00630068002e> /dan <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> /deu <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> /esp <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> /eti <feff004b00610073007500740061006700650020006e0065006900640020007300e4007400740065006900640020006b00760061006c006900740065006500740073006500200074007200fc006b006900650065006c007300650020007000720069006e00740069006d0069007300650020006a0061006f006b007300200073006f00620069006c0069006b0065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e00740069006400650020006c006f006f006d006900730065006b0073002e00200020004c006f006f0064007500640020005000440046002d0064006f006b0075006d0065006e00740065002000730061006100740065002000610076006100640061002000700072006f006700720061006d006d006900640065006700610020004100630072006f0062006100740020006e0069006e0067002000410064006f00620065002000520065006100640065007200200035002e00300020006a00610020007500750065006d006100740065002000760065007200730069006f006f006e00690064006500670061002e000d000a> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word volume 13, issue 2-2 journal of risk analysis and crisis response, 2023, 13(2), 117-135 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i2.360 117 article role of social networks in crowdfunding performance during the covid-19 pandemic in africa lenny phulong mamaro 1 and athenia bongani sibindi 1,* 1 school of economic and financial sciences, university of south africa, pretoria (0003), gauteng, south africa * correspondence: sibinab@unisa.ac.za received: march 7, 2023; accepted: may 5, 2023; published: june 30, 2023 abstract: the covid-19 pandemic made it more difficult for entrepreneurs to obtain capital finance from banks. as a result of disruptions caused by the covid-19 pandemic, entrepreneurs found alternative means of obtaining capital finance in the form of crowdfunding. crowdfunding has the potential to help entrepreneurs get access to finance. the prevalence of crowdfunding campaigns in africa is relatively low compared to most developed economies around the world. the objective of the study was to determine the role of social networks on crowdfunding performance during the covid-19 pandemic in africa. a total of 859 crowdfunding projects in africa were analysed in the study. the ordinary least squares (ols) and probit models were estimated in the analysis. the duration, comments, and target amount variables were found to be negatively and significantly associated with crowdfunding success in africa. conversely, the presence of videos, updates and backer’s variables were positively and significantly associated with the success of crowdfunding campaigns. the study contributes to new knowledge by revealing the influence of the covid-19 pandemic and the role of social networks on crowdfunding performance in africa. furthermore, the study makes a policy recommendation to african governments on the need to craft policy instruments that will foster the development of crowdfunding sources which will ultimately mitigate the problem of credit rationing by banks which curtail the growth of entrepreneurs. keywords: crowdfunding; covid-19 pandemic; financing; social network; africa 1. introduction restricted access to finance was a growing concern for new small to medium-sized enterprises (smes) after the global financial crisis. the covid-19 pandemic made it even more difficult for entrepreneurs to get access to finance to ensure their survival. crowdfunding has increased in popularity as part of a trend towards the disintermediation of the finance market and gives smes greater access to finance. crowdfunding is an alternative source of finance that could provide a sustainable solution to entrepreneurs’ lack of access to finance [1]. the covid-19 pandemic influenced the use of crowdfunding as a means of accessing finance. crowdfunding requires the involvement of an internet-based platform and many crowds to raise funds for capital openly and transparently. the role of entrepreneurs and smes in fostering economic growth and job creation in many economies is well documented in extant studies [2]. another strand of studies has advanced limited access to finance from traditional sources of finance as the major impediment to the lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 118 development of economies. given the rapid advancements in technology, investors allocate more risk capital to digital equity crowdfunding sites [3]. the involvement of social networks for entrepreneurs mitigates the challenges posed by the covid-19 pandemic [4]. however, the crowdfunding industry is characterised by a lower success rate in respect of crowdfunding campaigns [5, 6]. several studies conducted on the role of social network on the crowdfunding success were limited to mainly developed nations such as uk, malaysia, china, and usa [7, 8]. empirical research on the role of social networks in the success of crowdfunding is limited to developed nations. to the best of the researchers’ knowledge, the present study is the first study with a large sample to be conducted on the african continent. the wider context of the study is economic growth, which is predicated on the role of entrepreneurs (smes) in economic activity. the specific context of the study is that of the financial inclusion of entrepreneurs (smes). thus, the real-world problem that this study seeks to address is that of limited access to finance for entrepreneurs or smes, or the lack of financial inclusion of entrepreneurs or smes, on the african continent. the purpose of the study is to investigate the influence of social networks on crowdfunding performance in africa and, specifically, to determine the factors that contribute to the success of crowdfunding campaigns in order to alleviate the high incidence of unsuccessful campaigns on the continent. this study is based on the research objectives set out below.  to determine the influence of social media on crowdfunding performance during the covid-19 pandemic.  to determine the influence of backers on crowdfunding performance in africa.  to determine the success rate of crowdfunding during the covid-19 pandemic in africa. the paper is organised as follows; section 2 presents the literature review and develops the hypotheses for the study; section 3 sets out the research methodology applied in the study; section 4 presents the results of the study. section 5 discusses the findings of the study and draws conclusions thereof by outlining the implications of the study the limitations of the study and suggestions for future research. 2. literature review and hypotheses development limited access to finance for smes and entrepreneurs around the world, particularly on the african continent, is a major impediment to economic development [9]. unsurprisingly, the greatest challenge confronting african smes and entrepreneurs is limited access to finance [9, 10, 11]. an estimated 84% of smes and entrepreneurs on the african continent do not have access to finance, and there is a funding gap of about $70 to $170 billion [12]. many studies in this area originate from developed economies, hence the current study focuses on the african continent. 2.1. information asymmetry information asymmetry is inherent in the relationship between the project creator and the backers of the campaign. the risk and untrustworthiness of the crowdfunding campaign project signal unsuccessful project because it discourages the crowd from supporting the campaign [13, 14]. the effectiveness of these signals depends on investor characteristics [15]. the problem of information asymmetry exists between the project creator who does not disclose information to backers concerning the crowdfunding campaign features. it is therefore important to overcome the lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 119 problem of information asymmetry. the theory of information asymmetry originates from akerlof [16], rothschild and stiglitz [17]. information asymmetry exists when the other parties between the project creator (entrepreneur) and the backers (crowd) have less information about the quality of the product, specifically, the crowdfunding campaign [18]. information asymmetry has a negative influence on the success of a crowdfunding campaign [19]. signalling theory was adapted to the crowdfunding market to mitigate the risk associated with information asymmetry [18, 20]. the consistent presence of information asymmetry decreases backers’ participation in crowdfunding campaigns and consequently lead to failed projects. the ability of smes to send convincing signals could encourage both existing and potential funders to participate in new funding campaigns. therefore, fundraisers must convey signals that reduce information asymmetry and vagueness in the crowdfunding market [21]. 2.2. signalling within crowdfunding the reward-based crowdfunding signals the crowdfunding success performance due return promised to backers [22]. on a reward-based crowdfunding platform, the signals of the backers of projects can be identified from the comments on the projects. signals from the creators can be identified based on the number of updates to the projects. comments can only be considered as comments from backers and not general users of the crowdfunding platform, since only supporters of a project can comment on it. to overcome the negative effects of asymmetric information, the more informed party of a project can create an action to signal the less informed party about the quality of the product. this signalling process regarding labour market signal was first identified in the seminal work of ross [18] and spence [20]. these authors highlight the asymmetric information in the labour market and point out that without signals of some kind, employers are unable to distinguish highskilled workers from low-skilled ones. information asymmetry puts distant funders at a disadvantage compared to nearby funders. quality signals are valuable to informationally disadvantaged backers. 2.3. social capital theory the social network is an essential driver which attracts potential investors to support the crowdfunding project hence increases the crowdfunding success [23]. it is commonly accepted that social networks play an important role in the way in which backers and entrepreneurs communicate publicly on a digital platform [24]. the success of any entrepreneur’s survival is influenced by the accessibility of financial resources [25]. the social network creates business opportunities, help to overcome challenges concerning resources and threats and are adaptable in the crowdfunding market, thereby creating opportunities to fund smes [26]. the success of social networking sites lies in the ability of users to create content, to consume content created by other users and to identify and interact with users who have similar or opposing views [27]. social media platforms include webbased and mobile tools and apps that allow users to create content or to consume content created by creators and that ultimately facilitate connections [27]. the social media or networks are applied by role-players as follows to: (i) reveal themselves, (ii) know if others are available, (iii) relate to one another, (iv) know the social standing of others, (v) form communities, (vi) communicate with one another and (vii) exchange, distribute and receive content [28]. the crowdfunding platform connect participation between entrepreneurs and backers digitally in exchange of funds contributed to the lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 120 crowdfunding campaign project. social media platforms like facebook, linkedin, twitter, and instagram are based on network interactions, which include data analytics used for digital marketing and better consumer targeting [29]. figure 1 depicts the conceptual model of the study which shows how the social capital and signalling theories mitigate information asymmetry to help ensure the success of a crowdfunding campaign. figure 1. conceptual framework (authors’ compilation). the crowdfunding factors do not contribute equal chances to the project success, hence there are factors which contribute to failed crowdfunding campaign projects [30]. generally, the duration, the target amount and the covid-19 pandemic have a negative effect on crowdfunding performance [31, 32, 19]. in some studies, two parameters, namely, the funding rate and the number of donors were adopted to measure the performance of crowdfunding [33, 34]. crowdfunding campaigns are more likely to succeed when they manage to attract both many investors and a large amount of funds [34]. in other studies, the success of campaigns (that is, the projects raised at least the funding goal) is regarded as a proxy for crowdfunding performance [35, 36]. the factors that influence the performance of crowdfunding campaigns are the subject of several in-depth studies [37, 38, 39]. 2.4. hypotheses development the target amount of a crowdfunding project is the minimum amount of money that is needed for the project to get funded [40]. a project is considered successful when it has reached the minimum amount of money raised through backers [41]. therefore, the target amount signals a negative association with crowdfunding success, as supported by signalling theory. agrawal, catalini and goldfarb [42] found no relationship between the target amount and the backers of a project, whereas mollick [35] found a positive relationship between the backers and the target amount. there is a negative signal between the target amount and the probability of crowdfunding success, thus leading to the following hypothesis: 𝐇𝟏: the target amount of money requested is negatively related to crowdfunding performance. the large number of backers supporting crowdfunding campaign increases the probability of success compared to lower number of backers [43]. the presence of many backers influences investors’ confidence and crowdfunding success. therefore, the presence of many backers signals crowdfunding success. social networks play an important role in crowdfunding success since they help fundraisers to attract backers. the more backers, the higher the probability of crowdfunding success and the lower the degree of information asymmetry. therefore, the following research hypothesis is put forward: social capital and signalling theories mitigate problems of information asymmetry improvement crowdfunding campaign success lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 121 𝐇𝟐: the higher backers positively increase the probability of crowdfunding campaign success. duration is the period for which a project can receive financial support from backers. the prolonged duration provides potential backers with more time to comprehend the information on the crowdfunding platform [54]. however, the findings were not conclusive on whether prolonged duration signal success or not. authors such as frydrych et al. [44] and nyberg and aberg [2] found that a longer duration led to either an increase or a decrease in crowdfunding success [45], whereas mollick [35] and frydrych et al. [44] found a negative relationship between duration and crowdfunding success. thus, the following research hypothesis is put forward: 𝐇𝟑: the longer the duration of a crowdfunding campaign, the greater the expected success. an update refers to information that is posted on a crowdfunding platform during and after the fundraising period of a project [35]. updates represent the efforts of the founder of a crowdfunding project to reach out to current and potential funders. the founder also informs interested backers about the development of the project through videos, comments and information associated with crowdfunding success [46]. updates have a positive impact on crowdfunding success [47]. the posting of frequent updates about a campaign on a crowdfunding platform attracts investors or backers to the campaign [48]. signal theory holds that founder responses, formal websites and frequent project updates are clear signals from project founders of their intentional efforts to build social bonds with sponsors. these measures can thus strengthen the trust of sponsors and promote follow-up investment behaviour [49]. thus, the following hypothesis is put forward: 𝐇𝟒: frequent updates have a positive influence on crowdfunding performance. according to kim and hollingshead [50] describes as the change in crowd’s thoughts, feelings, behaviour of one or more people is influenced by the engagement and communication made via social network. the social network users post messages to interact with others and to construct their social image, leading to greater social acceptance [51]. the messages posted by the users of a social network convey information about them through which the audience can get to know more about them. network involvement is a means by which potential backers are informed about a crowdfunding project. the social media platform links backers to the campaign creator [35]. the social interaction between the stakeholders of a crowdfunding campaign signals the probability of its success [35]. the literature shows that network involvement has a positive influence on crowdfunding success. successful crowdfunding projects have been shown to rely on the information provided to the crowd via social networks. in line with the findings of previous studies, the following hypothesis is proposed in this study: 𝐇𝟓: the network involvement in a crowdfunding project is positively associated with crowdfunding performance. the comments made on a crowdfunding platform by the creator (entrepreneur) and the crowd (backers) signal the effectiveness of transparent communication and a reduction of information asymmetry [53]. the comments feature on a crowdfunding platform allows backers and the project creator to post comments concerning the progress of the crowdfunding campaign. previous studies have reported positive association between comments and crowdfunding success [54]. bi, liu and usman [55] observe that comments may convince potential backers to contribute to a crowdfunding campaign, known as electronic word of mouth, hence overcomes the problems of information lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 122 asymmetry. additionally, comments are commonly used as a marketing strategy to attracts backers for crowdfunding campaigns. bi, liu and usman [55] found that comments could persuade potential backers to participate in a crowdfunding campaign. consequently, comments about crowdfunding campaigns could affect entrepreneurs’ access to funding through crowdfunding platforms. therefore, the research hypothesis put forward: 𝐇𝟔: the more comment on the crowdfunding platform increases the probability of crowdfunding campaign success. backers easily understand the visuals that are used on a crowdfunding platform, which ultimately increases the success of campaigns and overcomes information asymmetry. therefore, videos that are displayed on the page of a crowdfunding campaign must provide valuable information to backers, hence, increases the crowdfunding success performance [56]. videos serve as means of communication between entrepreneurs (creators) and backers concerning the purpose of crowdfunding campaigns and the amount of money needed. the use of videos may encourage the crowd’s participation or contribution to a crowdfunding campaign. the presence of a video signals trustworthiness, preparedness, and quality. bi et al. [55] and mollick [35] note that crowdfunding projects without videos do not achieve the target amount. thus, the following hypothesis is put forward to determine whether videos increase the probability of crowdfunding campaign success or not: 𝐇𝟕: the presence of a video on a crowdfunding page has a positive influence on crowdfunding performance. displaying pictures on a crowdfunding page enables potential backers to understand the campaign better [57]. hence, pictures increase the probability of crowdfunding success. the presence of images on a crowdfunding page encourages potential backers to continue browsing the crowdfunding platform before they decide whether to support the campaign or not. a crowdfunding page with images has a higher probability of success than a page without images [58]. the presence of images provides a positive signal of crowdfunding success and overcomes information asymmetry. zhu and zhou [59] found that the availability of images on a crowdfunding platform increases the probability of success. however, the overuse of images may have an unpleasant appeal for potential backers, which decreases the probability of success. therefore, the following research hypothesis is put forward: 𝐇𝟖: the presence of images on the crowdfunding page increases the probability of crowdfunding campaign success. the covid-19 pandemic has shifted the focus from traditional forms of financing to innovative forms. many crowdfunding projects around the world were affected by the disruption caused by the pandemic [60]. the presence of covid-19 pandemic has negatively influenced the crowdfunding performance [61]. the popularity and the usage of social network has grown during the presence of covid-19 pandemic [31]. therefore, covid-19 pandemic has potential to change investors and backers’ behaviour in the financial markets and crowdfunding sector [31]. in order to verify if the social network influence the crowdfunding performance before covid-19 is the same today thus, the hypothesis is put forward: 𝐇𝟗: the covid-19 pandemic influenced the main drivers of crowdfunding performance. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 123 3. research methodology a quantitative research method, which is based on a deductive and postpositivist philosophy, was adopted to test the research hypotheses formulated for this study. secondary data covering the period 2019 to 2020 were collected from a crowd datacentre. convenience sampling was used for purposes of the study. the below figure 2 represent the research methodology flow chart of the study below: figure 2. research methodology flow chart (authors’ compilation). the sample for the study consisted of 859 crowdfunding projects. projects with missing information were excluded from the sample. the sample is depicted in table 1. table 1. sample description. crowdfunding campaign period number of projects success rate amount raised (us dollar) before covid-19 – 2019 749 7% 1 368 829 during covid-19 – 2020 110 12% 403 389 total 859 19% 1 772 218 source: authors’ compilation. figure 3 depicts the sampling process and the data coding to reach a final sample. data was collected from 1032 crowdfunding projects from 2019 to the year 2020. hence, the invalid and incomplete crowdfunding projects from african continent (specifically reward based) were removed to ensure consistency. the variables that were employed in the study are depicted in figure 4. these factors constitute the independent, dependent and control variables. the dependent variables include backers, success rate and success, whereas the independent variables include the covid-19 pandemic, social media, images and videos. lastly, the control variables consist of backers, comments, duration and updates. campaign success dummy variable that equals 1, if the project was successful and 0 if it is a failed project. the variables are defined more formally in table 2. quantitative research strategy (post positivism) deductive research approach (existing theory) non-probability convenience sampling lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 124 figure 3. the sampling and data coding process (source: adapted from (mamaro and sibindi 2023)). independent variables control variables figure 4. research model (source: authors’ compilation). the econometric analyses that were conducted in the study were ordinary least squares (ols) and probit regression. the following models were specified: 𝐶𝑆𝑅 = 𝛽 + 𝛽 𝑆𝑁 + 𝛽 𝐶𝑂𝑉𝐼𝐷 + 𝛽 𝑉𝐷 + 𝛽 𝐷𝑅𝑁 + 𝛽 𝐼𝑀 + 𝛽 𝐵𝐶𝐾 + 𝛽 𝑈𝑃𝐷 + 𝛽 𝐶𝑀𝑀 + 𝛽 𝑇𝐴 (1) 𝐶𝑆 = 𝛽 + 𝛽 𝑆𝑁 + 𝛽 𝐶𝑂𝑉𝐼𝐷 + 𝛽 𝑉𝐷 + 𝛽 𝐷𝑅𝑁 + 𝛽 𝐼𝑀 + 𝛽 𝐵𝐶𝐾 + 𝛽 𝑈𝑃𝐷 + 𝛽 𝐶𝑀𝑀 + 𝛽 𝑇𝐴 (2) 𝐵𝐶𝐾 = 𝛽 + 𝛽 𝑆𝑁 + 𝛽 𝐶𝑂𝑉𝐼𝐷 − 19 + 𝛽 𝑉𝐷 + 𝛽 𝐷𝑅𝑁 + 𝛽 𝐼𝑀 + 𝛽 𝑇𝐴 + 𝛽 𝑈𝑃𝐷 (3) dependent variables campaign success rate; campaign success; backers images videos covid-19 social media backers comments updates duration step 1 all projects from crowd datacenter step 2 removal of ongoing, invalid and duplicated projects (173) step 3 final sample of the study 1 032 859 859 projects lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 125 table 2. variable definition. names of variables measurement of variables references dependent variables -- campaign success rate (csr) a continuous variable is the ratio that is calculated by dividing each actual amount raised by each fundraising goal, which describes what extent can be achieved. [35, 62] campaign success (cs) a dummy variable of 1 if a campaign was successful and 0 if is failed. [63] backers (bck) the number of supporters who contributed to a project (transformed as log). [32, 64, 35] independent variables -- social network (sn) a binary variable of 1 if any of the following mediums are used: facebook, twitter, blogs, linkedin and myspace, and 0 otherwise. [47] video (vd) a dummy variable of 1 if a video is available on a campaign website and 0 otherwise. [56, 65] image or visuals (im) a dummy variable of 1 if an image is available on a campaign website and 0 otherwise. [65] covid-19 pandemic a dummy variable of 1 if a campaign during the covid-19 pandemic and 0 if was launched before the presence of covid-19 pandemic. [31, 79] control variables -- backers (bck) the number of supporters who contributed to a project (transformed as log). [32] comments (cmm) the number of comments about a project by the entrepreneur and backers (transformed as log). [67] duration (dr) the number of days for a campaign to raise funds (transformed as log). [68, 69] number of updates (upd) the number of updates on a crowdfunding platform (transformed as log). [70] 4. empirical findings the descriptive and correlation analysis results are presented in table 3. the correlation coefficients range between -0.02 and 0.60. the correlation coefficients fall within the threshold of 0.80 suggested by wooldridge [71]. it can be concluded correlation among the variables are not far less than 0.80. consequently, there is no problem of multicollinearity between variables. both descriptive and correlation of all variables reported in table 3. additionally, the variance inflation factor (vif) was calculated to validate whether the model is free of multicollinearity. the vif was found to be between 1,04 (minimum) and 2,00 (maximum). the vif falls within the threshold suggested by mcdonald and moffit [72]. therefore, the model is free of problems associated with multicollinearity. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 126 table 3. descriptive statistics and correlations. probability mean sd vif cs 0.083 0.28 dp sn 0.866 0.34 2.00 vd 0.47 0.50 1.41 im 0.68 0.47 1.80 drn 44.54 17.26 1.13 bck 19.64 118.47 1.73 cmm 2.08 28.61 1.66 covid-19 0.13 0.34 1.04 ta 4.07 0.80 1.12 upd 0.92 3.16 1.33 table 3. descriptive statistics and correlations (continued table). probability cs sn vd im drn bck cmm covid19 ta upd cs 1.000 sn 0.094*** 1.000 vd 0.156*** 0.373*** 1.000 im 0.123*** 0.578*** -0.049 1.000 drn -0.162*** 0.0152 -0.045 0.0427 1.000 bck 0.444*** 0.059* 0.141*** 0.0272 -0.113*** 1.000 cmm 0.223*** 0.028613 0.071** 0.0231 -0.059* 0.609*** 1.000 covid-19 0.047 0.070** 0.080** 0.128*** -0.059 0.089*** 0.098*** 1.000 ta -0.123*** 0.033 0.082 0.0507 0.293*** 0.058* 0.0145 -0.013 1.000 upd 0.493*** 0.114*** 0.177 0.141*** -0.103*** 0.4161 0.393*** 0.0610* -0.020 1.000 notes: number of observations 851; * p < 0.1, ** p < 0.05, *** p < 0.01. source: authors’ compilation. the results of estimating model 1 in table 4 shows that social networks and covid-19 are negatively and insignificantly associated with crowdfunding campaign success, whereas duration, comments and target amount (ta) are negatively and significantly associated with crowdfunding success (β = -0.01, -0.002 and -0.045 respectively; p < 0.1 and p < 0.01 for comments and target amount). the presence of updates, pictures and videos on a crowdfunding page increases the probability of success (β = 0.032, 0.067 and 0.047 respectively; p < 0.01). lastly, the presence of many backers positively influences crowdfunding performance. in the second model, campaign success was measured by a dummy variable of 1 if the projects have reached the target amount and 0 if not. the logistic regression results (model 2) suggest that sn, vd and drn are negatively but insignificantly associated with crowdfunding success (β = -0.046, 0.033 and -0.01 respectively). additionally, the results suggest that ta is negatively and significantly associated with crowdfunding success (β = -1.323; p < 0.01). conversely, covid-19, im, upd and cmm are positively but insignificantly associated with crowdfunding success (β = 0.020, 0.559, 0,033 and 0.001 respectively). lastly, bck is positively and significantly associated with crowdfunding performance (β = 0.039; p < 0.01). lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 127 table 4. regression analysis results. variables model 1 (ols) model 2 (probit) model 3 csr cs bck sn -0.046 (0.031) -0.046 (0.552) 3.648 (12.90) covid-19 -0.013 (0.023) 0.020 (0.292) 6.813 (9.398) vd 0.047*** (0.018) -0.033 (0.259) 12.694* (7.370) drn -0.001* (0.0005) -0.001 (0.006) -0.546*** (0.019) im 0.067*** (0.022) 0.559 (0.384) -4.796 (8.960) upd 0.032*** (0.003) 0.033 (0.033) 7.347*** (1.1038) cmm -0.002*** (0.0003) 0.001 (0.021) 2.156*** (0.119) ta -0.045*** (0.010) -1.323*** (0.247) -18.561* (18.130) bck 0.001*** (0,000) 0.039*** (0.005) -- c 0.235*** (0.045) 2.086 (0.822) -- pseudo 𝑹𝟐 0.3589 -0.663105 -0.421222 - number of observations 851 -851 -851 - notes: * p < 0.1, ** p < 0.05, *** p < 0.01. source: authors’ compilation. in model 3, backers were used as the dependent variable. the findings reveal that sn, covid19 and vd are positively but insignificantly associated with crowdfunding success (β = 3.648 and 6.813 respectively). the presence of vd is positively and significantly associated with crowdfunding success (β = 12.69; p < 0.1). further, upd and cmm are positively and significantly associated with crowdfunding success (β = 7.347 and 2.156 respectively; p < 0.01). by contrast, im is negatively and insignificantly associated with crowdfunding performance (β = 4.796). additionally, vd is negatively and significantly associated with crowdfunding success (β = 7.347 and 2.156 respectively; p < 0.01). the regression results of estimating in model 1 and model 2 suggest that social networks have a negative influence on crowdfunding success in africa. the findings are in line with the findings of tan and reddy [73] and bukhari, usman and hussain [74], that is, that the presence of social networks does not persuade potential backers to contribute to crowdfunding campaigns. hence it does not overcome the problems associated with information asymmetry and signal the success of crowdfunding performance. by contrast, [60, 67, 75] found that entrepreneurs can use social networks are more likely to be successful than entrepreneurs without social network usage. the ability of campaigns to meet fundraising targets on a finite timeline is typically regarded as crowdfunding success [76]. however, the results of estimating model 3 show a positive relationship between social networks and backers which were in line with the findings of renko, moss and lloyd [76]. several studies suggest that social networks persuade potential backers to invest in crowdfunding campaigns [8, 64, 78]. the covid-19 pandemic had a negative influence on crowdfunding success in africa, as reflected in the results of estimating model 1. this finding is consistent with zribi’s finding [31] that the covid-19 pandemic disrupted crowdfunding performance. during the covid-19 pandemic there was a drastic decline in the number of crowdfunding projects in africa, hence there is a negative association between covid-19 and crowdfunding success. by contrast, the results of estimating models 2 and 3 reflect a positive signal between the covid-19 pandemic and crowdfunding success lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 128 [66, 79, 80, 81]. a reliable communication strategy and social networks provide effective information to potential investors and reduce the distrust between backers and project creators [82]. despite the popularity and the usage of crowdfunding globally, remains limited on the african continent owing to limited information and communications technology (ict) infrastructure. the use of videos increases crowdfunding success, as deduced by from the regression results in models 1 and 3. this is in line with the findings of petitjean [68] and aleksina, akulenka and lublóy [83]. the presence of videos on crowdfunding platforms signals the success of crowdfunding while reducing information asymmetry. some backers may not be able to read textual information concerning a crowdfunding campaign. therefore, the presence of videos reduces demand uncertainty [84]. conversely, schraven, van burg, van gelderen and masurel [85] found a negative relationship between the use of videos and crowdfunding success, which is in line with the findings presented in model 2. however, the results of the study point to a mixed association between the use of videos and crowdfunding success. the use of images increases crowdfunding success, as reflected in models 1 and 2. therefore, the use of images is a persuasive signal to potential backers to support a campaign, which ultimately leads to success [86, 87]. aleksina et al. [83] and petitjean [68] found a negative signal between the use of images and crowdfunding success, which is supported by model 3 of the study. the duration of crowdfunding campaigns has a negative effect on their success, as reflected in models 1, 2 and 3. this result is consistent with the findings of zhang, liu, wang, zhao and zhang [88] and dikaputra, sulung and kot [53]. many backers and consistent updates on crowdfunding platforms have a positive influence on crowdfunding success, as reflected in all models presented in table 4 [52, 53]. many backers and consistent updates on crowdfunding platforms signal the success of crowdfunding and enhance communication transparency. in contrast, [31, 68, 89] found a negative association between backers and updates on crowdfunding success, additionally, the target amount decreases crowdfunding success, as reflected in table 4 [73, 90, 91]. therefore, the findings are inconclusive. the comments feature on crowdfunding platforms has a positive influence on crowdfunding success, as reflected in models 2 and 3. this result is supported by the findings of pinkow and emmerich [92], zhang et al. [6], petitjean [68] and aleksina et al. [83]. comments reduce information asymmetry by ensuring transparent and effective communication between backers and project owners (smes). however, model 1 reflects a negative relationship between backers and comments. the results suggest that continuous comments on crowdfunding platforms decrease crowdfunding success. this finding is not supported by the signalling and information asymmetry theories. one of the possible reasons for this finding could be the fact that crowdfunding projects that are based on the african continent differ from projects in developed economies owing to differences in ict infrastructure. 5. discussion and conclusion the purpose of the study was to investigate the influence of social networks on crowdfunding performance in africa and, specifically, to determine the factors that contribute to the success of crowdfunding campaigns in order to alleviate the high incidence of unsuccessful campaigns on the continent. the study contributes to the streams of research on the role of social networks in crowdfunding success. the findings of the study provide greater insight into the relationship lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 129 between social networks and crowdfunding success. the study provides several theoretical implications in respect of the role of social networks in crowdfunding performance. among the few studies that have examined the role of social networks in successful crowdfunding campaigns, primary attention is given to equity-based and reward-based crowdfunding campaigns. most of studies conducted on crowdfunding originate from developed countries [78, 8, 93]. hence, the current study fills a research gap by investigating the role of social network in crowdfunding success performance on the african continent. the current study has several theoretical and practical implications. firstly, the study applies contributions in the crowdfunding literature in examining the role of social networks in crowdfunding success from the perspective of signalling theory and social capital theory. most former studies that investigated signalling dynamics were limited to a single economy or country [5, 67, 89]. to the best of the researchers’ knowledge, the current study is one of the first to investigate the importance of signalling in crowdfunding success and its role in mitigating the problem of information asymmetry in the crowdfunding market from the perspective of african countries. a limited number of studies have examined crowdfunding success and market development in the context of a cross-country analysis [14, 69, 88]. this study represents an attempt at exploring signalling dynamics and advancing theoretical insight into the association between project comments, project updates, backers, images, videos, duration, target amount, covid-19 pandemic and social networks, on the one hand, and success in a donationand reward-based crowdfunding market, on the other. moreover, the study contributes to an understanding of the influence of social networks and backers on project success. the findings in respect of the variables are not consistent with the findings of previous studies. the study validates previous studies concerning the role of social networks in crowdfunding performance. furthermore, it highlights the complicated influence of the predictors of success, which was not previously explored. entrepreneurs or creators of crowdfunding projects could use the findings of the study to design more successful campaigns. lastly, the study indicates that project updates and comments signal quality and trustworthiness; these two aspects strongly influence backers’ participation and ultimately increase the success of crowdfunding campaigns. both the level of involvement of different campaign elements and their association with successful outcomes can differ, depending on the level of social trust that prevails in the context in which campaigns are launched. there is no study without limitations. the sample of this study was limited to the african continent crowdfunding market. future studies could investigate similar platforms established in other markets. the findings of the study cannot be generalised to all developing countries owing to differences in factors that play a role in the development and growth of crowdfunding, such as ict infrastructure, culture, and internet participation. moreover, all social ties were treated as one-way in this study, that is, a distinction was not made between the different types of social ties on social networks (for example: friends, families, or strangers). further work could differentiate between relationship types in measuring perceived social influence. the success of most crowdfunding campaigns depends on the rewards offered to backers [94]. consequently, the success of crowdfunding, especially donation-based crowdfunding, was influenced by the covid-19 pandemic. the secondary data collected in the study were limited to non-investment and donation-based crowdfunding models. therefore, it is recommended that future lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 130 studies focus on investment crowdfunding models, namely, lending and equity-based models. additionally, the current study emphasises that the presence of comments, updates and social networks diminishes the trust deficit of backers and information asymmetry [95]. it is recommended that future studies examine other variables, such as spelling mistakes, creators’ experience, frequently asked questions and textual descriptions, in exploring crowdfunding success. lastly, the study did not investigate the influence of personal and demographic characteristics, such as the gender, the age, the educational level, the income, or the interest category, of potential supporters. all these characteristics could influence potential backers’ decision-making patterns. therefore, future researchers are advised to consider the above aspects when studying project success in the crowdfunding market. contributions: conceptualisation, l.p.m. and a.b.s.; methodology, l.p.m.; software, l.p.m.; validation, l.p.m and a.b.s.; formal analysis, l.p.m.; investigation, l.p.m.; resources, l.p.m.; data curation, l.p.m.; writing – original draft preparation, l.p.m.; writing – review and editing, l.p.m. and a.b.s; visualisation, l.p.m.; supervision, a.b..s; project administration, l.p.m.; funding acquisition, l.p. m. all authors have read and agreed to the published version of the manuscript. acknowledgments: the authors would like to thank the anonymous editor at the university of south africa language services department for the excellent technical support provided in editing this article. funding: the study was funded by the national research framework (nrf) in south africa. conflicts of interest: the authors declare no conflict of interest. the funders did not play any role in the design of the study, data collection, data analysis, the discussion of the results, the writing of the manuscript or the decision to publish the article. references [1] mollick, e.r.; venkat k. after the campaign: outcomes of crowdfunding. unc kenan-flagler research paper, 2014, 2376997. available at ssrn: https://ssrn.com/abstract=2376997. doi: http://dx.doi.org/10.2139/ssrn.2376997. [2] nyberg, a.; åberg, m. crowdfunding social entrepreneurship: the influential factors in crowdfunding success for social entrepreneurs, 2017. available at http://hdl.handle.net/2077/53751. [3] brown, r.; mawson, s.; rowe, a.; mason, c. working the crowd: improvisational entrepreneurship and equity crowdfunding in nascent entrepreneurial ventures. international small business journal, 2018, 36(2):169-193. doi: https://doi.org/10.1177/0266242617729743. [4] bacq, s.; lumpkin, g.t. social entrepreneurship and covid-19. journal of management studies, 2021, 58(1):283-286. [5] courtney, c.; dutta, s.; li, y. resolving information asymmetry: signalling, endorsement, and crowdfunding success. entrepreneurship theory and practice, 2017, 41(2):265-290. doi: https://doi.org/10.1111/etap.122. [6] zhang, y.; tan, c.d.; sun, j.; yang, z. why do people patronize donation-based crowdfunding platforms? an activity perspective of critical success factors. computers in human behaviour, 2020, 112:1-9. doi: https://doi.org/10.1016/j.chb.2020.106470. [7] wahjono, s.i.; marina, a.; sezeli, s.s.; mahardika, b.w. the role of social awareness over the success of crowdfunding platform. humanities & social sciences reviews, 2019, 7(3):534-540. doi: https://doi.org/10.18510/hssr.2019.7379. [8] liu, y.; chen, y.; fan, z.p. do social network crowds help fundraise campaigns? effects of social influence on crowdfunding performance. journal of business research, 2021, 122:97-108. doi: https://doi.org/10.1016/j.jbusres.2020.08.052. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 131 [9] beck, t.; cull, r. sme finance in africa. journal of african economies, 2014, 23(5):583-613. doi: https://doi.org/10.1093/jae/eju016. [10] adebayo, n.a.; nassar, m.l. impact of micro and small business entrepreneurship on poverty reduction in ibadan metropolis, south western nigeria. international review of management and business research, 2014, 3:1603-1626. [11] ayyagari, m.; demirgüç-kunt, a.; maksimovic, v. financing of firms in developing countries: lessons from research. world bank policy research working paper, 2012, (6036). [12] hiller, a. an empirical analysis of crowdfunding in sub-saharan africa. 2017, hhl leipzig graduate school of management, leipzig, dissertation available at https://nbn-resolving.org/urn:nbn:de:bsz:14qucosa2-163562. [13] cerpentier, m.; vanacker, t.; paeleman, i.; bringmann, k. equity crowdfunding, market timing, and firm capital structure. the journal of technology transfer, 2022, 47(6):1766-1793. doi: https://doi.org/10.1007/s10961-021-09893-y. [14] johan, s.; zhang, y. investors’ industry preference in equity crowdfunding. the journal of technology transfer, 2022, 47(6):1737-1765. doi: https://doi.org/10.1007/s10961-021-09897-8. [15] kleinert, s.; mochkabadi, k. gender stereotypes in equity crowdfunding: the effect of gender bias on the interpretation of quality signals. the journal of technology transfer, 2021:1-22. doi: https://doi.org/10.1007/s10961-021-09892-z. [16] akerlof, g.a. the market for “lemons”: quality uncertainty and the market mechanism. in uncertainty in economics (pp. 235-251). academic press. 1978:237-251. doi: https://doi.org/10.1016/b978-0-12-2148507.50022-x. [17] rothschild, m.; stiglitz, j. equilibrium in competitive insurance markets: an essay on the economics of imperfect information. in uncertainty in economics (pp. 257-280). academic press. 1978:259-280. doi: https://doi.org/10.1016/b978-0-12-214850-7.50024-3. [18] ross, s.a. the determination of financial structure: the incentive-signalling approach. the bell journal of economics, 1977:23-40. doi: https://doi.org/10.2307/3003485. [19] de larrea, g.l.; altin, m.; singh, d. determinants of success of restaurant crowdfunding. international journal of hospitality management, 2019, 78:150-158. doi: https://doi.org/10.1016/j.ijhm.2018.10.003. [20] spence, m. job market signalling. in uncertainty in economics (pp. 281-306). academic press.1978:283-306. doi: https://doi.org/10.1016/b978-0-12-214850-7.50025-5. [21] di pietro, f.; grilli, l.; masciarelli, f. talking about a revolution? costly and costless signals and the role of innovativeness in equity crowdfunding. journal of small business management, 2020:1-32. doi: https://doi.org/10.1080/00472778.2020.1816435. [22] kromidha, e.; robson, p. social identity, and signalling success factors in online crowdfunding. entrepreneurship & regional development, 2016, 28(9-10):605-629. doi: https://doi.org/10.1080/08985626.2016.1198425. [23] nahapiet, j.; ghoshal, s. social capital, intellectual capital, and the organizational advantage. academy of management review, 1998, 23(2):242-266. doi: https://doi.org/10.5465/amr.1998.533225. [24] lynn, t.; rosati, p.; nair, b.; mac an bhaird, c. an exploratory data analysis of the# crowdfunding network on twitter. journal of open innovation: technology, market, and complexity, 2020, 6(3):1-22. doi: https://doi.org/10.3390/joitmc6030080. [25] baum, j.a.; oliver, c. toward an institutional ecology of organizational founding. academy of management journal, 1996, 39(5):1378-1427. doi: https://doi.org/10.5465/257003. [26] johannisson, b. business formation—a network approach. scandinavian journal of management, 1988, 4(3-4): 83-99. doi: https://doi.org/10.1016/0956-5221(88)90002-4. [27] hoffman, d.l.; novak, t.p. toward a deeper understanding of social media. journal of interactive marketing, 2012, 26(2):69-70. doi: https://doi.org/10.1016/j.intmar.2012.03.001. [28] kietzmann, j.h.; hermkens, k.; mccarthy, i.p.; silvestre, b.s. social media? get serious! understanding the functional building blocks of social media. business horizons, 2011, 54(3):241-251. doi: https://doi.org/10.1016/j.bushor.2011.01.005. [29] ratten, v. digital platform usage amongst female sport technology entrepreneurs. journal of small business & entrepreneurship, 2022:1-24. doi: https://doi.org/10.1080/08276331.2022.2116678. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 132 [30] beier, m.; wagner, k. individual and corporate decisions to adopt social media: perception of risks and benefits in small firms. 2014 (january 24, 2014). available at ssrn: https://ssrn.com/abstract=2384900. doi: http://dx.doi.org/10.2139/ssrn.2384900. [31] zribi, s. effects of social influence on crowdfunding performance: implications of the covid-19 pandemic. humanities and social sciences communications, 2022, 9(1):1-8. doi: https://doi.org/10.1057/s41599-022-01207-3. [32] fourkan, m. crowdfunding: antecedents of number of backers and success of a project. fourkan, m. (2021). crowdfunding: antecedents of number of backers and success of a project. international journal of small business and entrepreneurship research, 2021, 9(2):1-13. doi: retrieved from https://www.eajournals.org/wp-content/uploads/crowdfunding.pdf. available at ssrn: https://ssrn.com/abstract=3865312. [33] vulkan, n.; åstebro, t.; sierra, m. f. equity crowdfunding: a new phenomena. journal of business venturing insights, 2016, 5:37-49. doi: https://doi.org/10.1016/j.jbvi.2016.02.001. [34] lukkarinen, a.; teich, j. e.; wallenius, h.; wallenius, j. success drivers of online equity crowdfunding campaigns. decision support systems, 2016, 87:26-38. doi: https://doi.org/10.1016/j.dss.2016.04.006. [35] mollick, e. the dynamics of crowdfunding: an exploratory study. journal of business venturing, 2014, 29(1): 1-16. doi: https://doi.org/10.1016/j.jbusvent.2013.06.005. [36] colombo, m.g.; franzoni, c.; rossi–lamastra, c. internal social capital and the attraction of early contributions in crowdfunding. entrepreneurship theory and practice, 2016, 39(1):75-100. doi: https://doi.org/10.1111/etap.12118. [37] pyayt, p.o.; thomas, h.a.; arvin, s.; sakdipon, j. user entrepreneurs’ multiple identities and crowdfunding performance: effects through product innovativeness, perceived passion, and need similarity. j bus venture, 2019, 34(5):105895.1–105895.16. doi: https://doi.org/10.1016/j.jbusvent.2018.08.005. [38] rose, s.; wentzel, d.; hopp, c.; kaminski, j. launching for success: the effects of psychological distance and mental simulation on funding decisions and crowdfunding performance. journal of business venturing, 2021, 36(6):106021. doi: https://doi.org/10.1016/j.jbusvent.2020.106021. [39] zhang, y.; tan, c. d.; sun, j.; yang, z. why do people patronize donation-based crowdfunding platforms? an activity perspective of critical success factors. computers in human behaviour, 2020, 112:106470. doi: https://doi.org/10.1016/j.chb.2020.106470. [40] neugebauer, p.; medziausyte, j. financing success through equity crowdfunding: the case of start-ups and smes funded on a european crowdfunding platform. 2017. [41] wachira, v.k. crowdfunding in kenya: factors for successful campaign. 2021, 3:413428. available at http://ir.kabarak.ac.ke/handle/123456789/901. [42] agrawal, a. k.; catalini, c.; goldfarb, a. the geography of crowdfunding (no. w16820). national bureau of economic research. 2011, february 2011: doi 10.3386/w16820. [43] ralcheva, a.; roosenboom, p. on the road to success in equity crowdfunding (november 1, 2016). available at ssrn: https://ssrn.com/abstract=2727742. doi: http://dx.doi.org/10.2139/ssrn.2727742. [44] frydrych, d.; bock, a.j.; kinder, t.; koeck, b. exploring entrepreneurial legitimacy in reward-based crowdfunding. venture capital, 2014, 16(3):247-269. doi: https://doi.org/10.1080/13691066.2014.916512. [45] burtch, g.; ghose, a.; wattal, s. secret admirers: an empirical examination of information hiding and contribution dynamics in online crowdfunding. information systems research, 2016, 27(3):478-496. doi: https://doi.org/10.1287/isre.2016.0642. [46] drabløs, c. what influences crowdfunding campaign success (master's thesis, universitetet i agder; university of agder). 2015. [47] koch, j.a.; siering, m. crowdfunding success factors: the characteristics of successfully funded projects on crowdfunding platforms (april 4, 2015). proceedings of the 23rd european conference on information systems (ecis 2015); muenster, germany 2015. available at ssrn: https://ssrn.com/abstract=2808424. [48] kuppuswamy, v.; bayus, b.l. does my contribution to your crowdfunding project matter? journal of business venturing, 2017, 32(1):72-89. doi: https://doi.org/10.1016/j.jbusvent.2016.10.004. [49] xie, k.; liu, z.; chen, l.; zhang, w.; liu, s.; chaudhry, s.s. success factors and complex dynamics of crowdfunding: an empirical research on taobao platform in china. electronic markets, 2019, 29(2):187-199. doi: https://doi.org/10.1007/s12525-018-0305-6. [50] kim, y.j.; hollingshead, a.b. online social influence: past, present, and future. in communication yearbook, 2015, 39 (pp. 185-214). routledge. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 133 [51] toubia, o.; stephen, a.t. intrinsic versus image-related motivations in social media: why do people contribute content to twitter. 2012, ssrn electron j. doi: https://doi.org/10.2139/ssrn1949833. [52] ho, h.c.; chiu, c.l.; mansumitrchai, s.; yuan, z.; zhao, n.; zou, j. the influence of signals on donation crowdfunding campaign success during covid-19 crisis. international journal of environmental research and public health, 2021, 18(14):7715. doi: https://doi.org/10.3390/ijerph18147715. [53] dikaputra, r.; sulung, l.a.k.; kot, s. analysis of success factors of reward-based crowdfunding campaigns using multi-theory approach in asean-5 countries. social sciences, 2019, 8(10):293. doi: https://doi.org/10.3390/socsci8100293. [54] lagazio, c.; querci, f. exploring the multi-sided nature of crowdfunding campaign success. journal of business research, 2018, 90:318-324. doi: https://doi.org/10.1016/j.jbusres.2018.05.031. [55] bi, s.; liu, z.; usman, k. the influence of online information on investing decisions of reward-based crowdfunding. journal of business research, 2017, 71:10-18. doi: https://doi.org/10.1016/j.jbusres.2016.10.001. [56] kunz, m.m.; bretschneider, u.; erler, m.; leimeister, j.m. an empirical investigation of signalling in reward-based crowdfunding. electronic commerce research, 2017, 17(3):425-461. [57] prasobpiboon, s.; ratanabanchuen, r.; chandrachai, a.; triukose, s. success factors in project fundraising under reward-based crowdfunding platform. academy of entrepreneurship journal, 2021, 27:1-20. [58] kuppuswamy, v.; bayus, b.l. does my contribution to your crowdfunding project matter? journal of business venturing, 2017, 32(1):72-89. doi: https://doi.org/10.1016/j.jbusvent.2016.10.004. [59] zhu, h.; zhou, z.z. analysis and outlook of applications of blockchain technology to equity crowdfunding in china. financial innovation, 2016, 2(1):1-11. doi: https://doi.org/10.1186/s40854-016-0044-7. [60] mamaro, l.p.; sibindi, a.b. entrepreneurial financing in africa during the covid-19 pandemic. journal of risk and financial management, 2022, 15(11):511. doi: https://doi.org/10.3390/jrfm15110511. [61] sansa, n.a. the impact of the covid-19 on the financial markets: evidence from china and usa. electronic research journal of social sciences and humanities, 2020, 2(2). available at ssrn: https://ssrn.com/abstract=3567901. doi: http://dx.doi.org/10.2139/ssrn.3567901. [62] kim, p.h.; buffart, m.; croidieu, g. tmi: signalling credible claims in crowdfunding campaign narratives. group & organization management, 2016, 41(6):717-750. doi: https://doi.org/10.1177/105960111665118. [63] anglin, a.h.; short, j.c.; drover, w.; stevenson, r.m.; mckenny, a.f.; allison, t.h. the power of positivity? the influence of positive psychological capital language on crowdfunding performance. journal of business venturing, 2018, 33(4):470-492. doi: https://doi.org/10.1016/j.jbusvent.2018.03.003. [64] greenberg, j.; mollick, e. activist choice homophily and the crowdfunding of female founders. administrative science quarterly, 2017, 62(2):341-374. doi: https://doi.org/10.1177/0001839216678847. [65] xu, t. learning from the crowd: the feedback value of crowdfunding (september 30, 2018). available at ssrn: https://ssrn.com/abstract=2637699. doi: http://dx.doi.org/10.2139/ssrn.2637699. [66] cummings, d.; reardon, r.s. covid-19 and entrepreneurial processes in us equity crowdfunding. journal of small business management, 2022:1-24. doi: https://doi.org/10.1080/00472778.2022.2051178. [67] wang, n.; li, q.; liang, h.; ye, t.; ge, s. understanding the importance of interaction between creators and backers in crowdfunding success. electronic commerce research and applications, 2018, 27:106-117. doi: https://doi.org/10.1016/j.elerap.2017.12.004. [68] petitjean, m. what explains the success of reward-based crowdfunding campaigns as they unfold? evidence from the french crowdfunding platform kisskissbankbank. finance research letters, 2018, 26:914. doi: https://doi.org/10.1016/j.frl.2017.11.005. [69] song, y.; berger, r.; yosipof, a.; barnes, b.r. mining and investigating the factors influencing crowdfunding success. technological forecasting and social change, 2019, 148:119723. doi: https://doi.org/10.1016/j.techfore.2019.119723. [70] giudici, g.; guerini, m.; rossi-lamastra, c. reward-based crowdfunding of entrepreneurial projects: the effect of local altruism and localized social capital on proponents’ success. small business economics, 2018, 50(2):307-324. doi: https://doi.org/10.1007/s11187-016-9830-x. [71] wooldridge, j.m. introductory econometrics: a modern approach. cengage learning. 2018. [72] mcdonald, j.f.; moffit, r.a. the uses of tobit analysis. the review of econometrics and statistics. kristjanson, p.; okike, i.; tarawali s.; singh, b.b. 1980, 1000:195-210. doi: https://doi.org/10.2307/1924766. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 134 [73] tan, y.h.; reddy, s.k. crowdfunding digital platforms: backer networks and their impact on project outcomes. social networks, 2021, 64:158-172. doi: https://doi.org/10.1016/j.socnet.2020.09.005. [74] bukhari, f.a.s.; usman, s.m.; usman, m.; hussain, k. the effects of creator credibility and backer endorsement in donation crowdfunding campaigns success. baltic journal of management, 2019, 15(2):215235. doi: https://doi.org/10.1108/bjm-02-2019-0077. [75] vismara, s. equity retention and social network theory in equity crowdfunding. small business economics, 2016, 46:579-590. doi: https://doi.org/10.1007/s11187-016-9710-4. [76] allison, t.h.; davis, b.c.; webb, j.w.; short, j.c. persuasion in crowdfunding: an elaboration likelihood model of crowdfunding performance. journal of business venturing, 2017, 32(6):707-725. doi: https://doi.org/10.1016/j.jbusvent.2017.09.002. [77] renko, m.; moss, t.w.; lloyd, a. crowdfunding by non-profit and social ventures. handbook of research on crowdfunding, 2019:249-268. doi: https://doi.org/10.4337/9781788117210.00017. [78] farhoud, m.; shah, s.; stenholm, p.; kibler, e.; renko, m.; terjesen, s. social enterprise crowdfunding in an acute crisis. journal of business venturing insights, 2021, 15:e00211. doi: https://doi.org/10.1016/j.jbvi.2020.e00211. [79] igra, m.; kenworthy, n.; luchsinger, c.; jung, j.k. crowdfunding as a response to covid-19: increasing inequities at a time of crisis. social science & medicine, 2021, 282:114105. doi: https://doi.org/10.1016/j.socscimed.2021.114105. [80] cummings, d.; reardon, r.s. covid-19 and entrepreneurial processes in us equity crowdfunding. journal of small business management, 2022:1-24. doi: https://doi.org/10.1080/00472778.2022.2051178. [81] liu, z.j.; panfilova, e.; mikhaylov, a.; kurilova, a. assessing stability in the relationship between parties in crowdfunding and crowdsourcing projects during the covid-19 crisis. journal of global information management (jgim), 2021, 30(4):1-18. doi: https://doi.org/10.4018/jgim.297905. [82] wu, w.; huang, x.; wu, c.h.; tsai, s.b. pricing strategy and performance investment decisions in competitive crowdfunding markets. journal of business research, 2022, 140:491-497. doi: https://doi.org/10.1080/17517575.2020.1734239. [83] aleksina, a.; akulenka, s.; lublóy, á. success factors of crowdfunding campaigns in medical research: perceptions and reality. drug discovery today, 2019, 24(7):1413-1420. doi: https://doi.org/10.1016/j.drudis.2019.05.012. [84] miglo, a. crowdfunding in a competitive environment. journal of risk and financial management, 2020, 13(3): 39. doi: https://doi.org/10.3390/jrfm13030039. [85] schraven, e.; van burg, e.; van gelderen, m.; masurel, e. predictions of crowdfunding campaign success: the influence of first impressions on accuracy and positivity. journal of risk and financial management, 2020, 13(12):331. doi: https://doi.org/10.3390/jrfm13120331. [86] zhou, m.; lu, b.; fan, w.; wang, g.a. project description and crowdfunding success: an exploratory study. information systems frontiers, 2018, 20:259-274. doi: https://doi.org/10.1007/s10796-016-9723-1. [87] shneor, r.; mrzygłód, u.; adamska-mieruszewska, j.; fornalska-skurczyńska, a. the role of social trust in reward crowdfunding campaigns’ design and success. electronic markets, 2021:1-16. doi: https://doi.org/10.1007/s12525-021-00456-5. [88] zhang, x.; liu, x.; wang, x.; zhao, h.; zhang, w. exploring the effects of social capital on crowdfunding performance: a holistic analysis from the empirical and predictive views. computers in human behavior, 2022, 126:107011. doi: https://doi.org/10.1016/j.chb.2021.107011. [89] usman, s.m.; bukhari, f.a.s.; you, h.; badulescu, d.; gavrilut, d. the effect and impact of signals on investing decisions in reward-based crowdfunding: a comparative study of china and the united kingdom. journal of risk and financial management, 2020, 13(12):325. doi: https://doi.org/10.3390/jrfm13120325. [90] hapsari, n.s.; sulung, l.a.k. the role of social capital and reward factor in the success of crowdfunding project fundraising: case study of emerging market countries. in asia-pacific research in social sciences and humanities universitas indonesia conference (aprish 2019), 2019 may (pp. 591-599). atlantis press. doi: https://doi.org/10.2991/assehr.k.210531.074. [91] tian, z.; guan, l.; shi, m. the key factors of successful internet crowdfunding projects-an empirical study based on different platforms. in 2018 15th international conference on service systems and service management (icsssm) (pp. 1-6). ieee. doi: https://doi.org/10.1109/icsssm.2018.8465009. lenny phulong mamaro, et al. / journal of risk analysis and crisis response, 2023, 13(2), 117-135 doi: https://doi.org/10.54560/jracr.v13i2.360 135 [92] pinkow, f.; emmerich, p. re-examining crowdfunding success: how the crowdfunding goal moderates the relationship of success factors and crowdfunding performance. central european business review, 2021, 10(2):91-114. doi: https://doi.org/10.18267/j.cebr.263. [93] kubo, t.; veríssimo, d.; uryu, s.; mieno, t.; macmillan, d. what determines the success and failure of environmental crowdfunding? ambio, 2021, 50:1659-1669. doi: https://doi.org/10.1007/s13280-021-01522-0. [94] bargain, o.; cardebat, j.m.; vignolles, a. crowdfunding in the wine industry. journal of wine economics, 2018, 13(1):57-82. doi: https://doi.org/10.1017/jwe.2018.3. [95] liang, x.; hu, x.; jiang, j. research on the effects of information description on crowdfunding success within a sustainable economy—the perspective of information communication. sustainability, 2020, 12(2): 650. doi: https://doi.org/10.3390/su12020650. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 licence (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style editor’s introduction for volume 6, issue 1 it is the sixth year since we founded this international journal as an official publication of society for risk analysis – china (sra-china, http://www.sra.org/china-sra). the journal is designed to promote a higher level of methodology and practice, and provide an international platform for sharing theory and applications of risk analysis (ra) and crisis response (rc). particularly, it is considered as a journal in science and technology rather than policy and law related to risk issues. this issue contains 6 papers. there are 3 contributions written in english and 3 contributions in chinese with english abstracts. the papers can be divided into four categories: experimental riskology, risk perception, financial risk, and public security. based on the thinking of experimental riskology, the paper “an approach to ranking integrated models for risk assessment by using the internet of intelligences”, by wang and et al., built an internet of intelligences to play the gobang game for ranking the analytic hierarchy process, the median method corresponding to time and comprehensive evaluation method, according their scores. the results show that for gobang, the effect of simple comprehensive evaluation method is better than analytic hierarchy process. there are two papers in risk perception. the first paper “eutrophication hazard evaluation using copula-cloud”, by liu, wang and wang, proposes a copula-cloud to model the randomness, fuzziness and dependence of each risk criterion in assessment. the suggested technique has been applied on cases of representative lakes or reservoirs in china. results in comparison with relevant proposed methods verify the completeness and effectiveness of the technique. the second paper “high-rise building group regional fire risk assessment model based on ahp”, by zhu and you, puts forwards a regional fire risk assessment model for high-rise building group considering the fire danger degree, basic characters of the high-rise building group and fire control capability based on analytic hierarchy process (ahp). the impact factors, values of typical risk parameter and the weight of each index are analyzed. an assessment program is further developed using matlab, to perform the risk assessment procedures. the model is then applied to the regional fire risk evaluation of a high-rise building group. there are two papers in financial risk. the first paper “research on the performance evaluation of government venture capital fund based on factor analysis and dea model”, by zhang, chen and wang, analyzes and determines the evaluation index system relevant in the performance evaluation of government venture capital fund. the suggested system could help us to effectively evaluate the performance of the guidance fund. the majority of the fund operation used in this paper is comprehensive and effective. the second paper “economic fluctuations and insurance industry crisis: the history, status quo and the future”, by guo, analyzes the reasons and mechanisms of insurance industry crisis in japan and taiwan, found the insurance industry crisis closely related to the economic fluctuation, and provided some warnings and references to the insurance industry of china. there is a paper in public security, “the effects of the social context on pre-decisional processes of protective action in beijing communities”, by liu, zhang, li and li, which examined the effects of the social context on pre-decisional processes of individual protective action. a two-stage survey was carried out aimed at residents in beijing communities in half a year. main results are : (1) media communication media communication activities can influence individual reception of risk information; (2) face-to-face communication activities can influence individual attention behavior; (3) friendly and kinship network does not influence any pre-decisional processes variable, but it can influence face-to-face communication instead. we sincerely hope our reader will find this issue’s information on ra and cr interface useful. thanks to the referees for their strong support and kind help. and also thank the authors very much for all their outstanding contributions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 6, no. 1 (april 2016), 1 published by atlantis press copyright: the authors 1 atlantis press journal style research on promoting effect of tourism industry development on tourism poverty alleviation in guizhou siqi li, mu zhang, xinpu wang school of finance, guizhou university of finance and economics, guiyang 550025, china e-mail:546932209@qq.com, rim_007@163.com ,1207112376@qq.com received december14, 2017 accepted march 3, 2018 abstract the traffic in the counties of guizhou is inconvenient and the economic location is remote. how to transform the characteristics of natural resources and national cultural resources into economic advantages? this is the key of this paper. in this paper, the tourism industry of 50 poverty-stricken counties in guizhou province is selected as the representative, and the anti-poverty effect of guizhou tourism industry is measured by using panel data model. the empirical results show that the development of tourism industry in 50 poverty-stricken counties in guizhou has promoted the development of the national economy, thus reducing the poverty rate of the counties. it is concluded that guizhou tourism industry has a certain promoting effect on the precision poverty alleviation in guizhou. keywords: tourism industry, guizhou county, targeted poverty alleviation, panel data model. 1. preface the development of guizhou's socioeconomic development and the gap between the developed regions of the eastern coastal areas requires the development of guizhou. however, how to realize the crossing has always been a problem for guizhou's development. despite the advantages of learning, imitation, development and opportunities, at the same time, inequality, the existence and development foundation development assignments, endogenous dynamic development mechanism is difficult to establish, and many other factors. at present, the academic circles mainly analyze how to develop the characteristic industries of this underdeveloped region from the perspectives of culture, technology and management, but make quantitative analysis from the perspective of economics is less. this article explores the promotion function of tourism industry development to the accurate poverty alleviation in guizhou counties, and provides some theoretical and empirical research for the economic development of guizhou province to make up for the deficiency of this research. 2. writing background at present, our country is in an extraordinary period of accelerating the transformation of the mode of economic growth. general secretary jinping xi stressed at the 2014 national economic work conference that the relationship between sustained economic development and gdp growth should be correctly dealt with in future work. development is not equal to the increase of gdp, accelerating the adjustment of industrial structure, maintaining the rational and stable growth of the national economy, and realizing the sustainable development of the economy in the future, that is, the economic growth can not be measured by digital gdp alone. it is necessary to coordinate the development of industry, ensure the process of industrialization and urbanization, and develop modern agriculture to ensure food security, stabilize the price level, and at the same time, ensure stable growth. employment should also protect people's livelihood. preliminary accounting by the national bureau of statistics: in 2013, china's gdp was 56.8845 trillion 110 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 s. li, m. zhang yuan, an increase of 7.7 percent, while guizhou's gdp8006 grew by 12.5 percent, although guizhou had a relatively fast growth. however, the economic level is still below the national average. compared with other provinces and cities in the west, the economic level is still not high. if we want to achieve the overall well-off level in the whole province early, we must strive to solve the poverty problem, especially in the rural areas. at the same time, the gap between the rich and the poor is widening, especially the gap between urban and rural income is more prominent, urban and rural income the widening of the level gap does not only affect the reform of the rural production mode, the development of the agricultural market, and the farmers' enthusiasm for production. it will also have a significant impact on the overall level of economic development of the whole society. in the "outline for poverty alleviation and development in china's rural areas 2011-2020", published by the public office of the state council for poverty alleviation in 2011, it is pointed out that the overall goal is to promote comprehensive and coordinated economic and social development in poverty-stricken areas. by 2020, the whole society will be able to build a well-off society in an all-round way and achieve common prosperity. at the same time, it is proposed that by 2015, the per capita net income of the poor will have doubled to more than 3000 yuan, and the number of poor people will have decreased by 15 percent. by 2020, the per capita net income of the poor will quadruple, with a net income of 6000 yuan, and the poor will achieve collective eradication of poverty and basically eliminate the phenomenon of absolute poverty. it is hoped that the overall goal of eradicating poverty and achieving a well-to-do life by 2020 will be achieved in advance of 2017. the main poverty-stricken areas in the country are concentrated in remote mountainous areas. at present, poverty relief work is mainly carried out in the poverty-stricken special zones of guizhou. of the 88 counties in guizhou, 56 are focused on helping the poor and 48 are at the national level. it is urgent to eliminate poverty and promote economic development in guizhou. there are many international measure of poverty indicators, including poverty headcount, also called poverty gap, sen poverty index, fgt poverty index, s index (sst) index system, we study the problem of poverty in guizhou province, combined with the mathematical statistics principle, to quantitative research, and taking into account the availability and regularity data show the data, select the poverty rate as the research object. in the study of poverty in guizhou province, not only does the poverty problem of guizhou province be considered as a whole, but also goes deep into 88 counties (cities) of guizhou province. therefore, it is more reasonable to consider the establishment of panel data model. cross-sectional data series pooled crosssectional data and panel data are three main economic data structure forms. time series data and cross section data are all data sets with one dimension. where time series data are individual variables in the observed values vary in sequence over a certain period of time; section data are spatial data of multiple individuals at a certain point in time. panel data are formed by mixing time series and cross section data in time and cross section. the characteristics of individuals in time and cross-section can be reflected from different angles. therefore, the study of panel data can take into account both the effect of time and the effect of influencing factors on a single variable. it is more comprehensive, more systematic and more stable than simple onedimensional data. in addition, at present, most of the research on poverty tends to the emergence of poverty, the evolution of poverty and other qualitative analysis. there is no indepth study on the huge data released by the guizhou bureau of statistics. based on this situation, in order to make full use of the abundant data resources and make better quantitative research on the poverty problem, this paper intends to carry out a further study on the poverty problem in guizhou province. 3. literature review 3.1. domestic research status the articles and achievements of "targeted poverty alleviation" in domestic research are relatively limited, but some outstanding research ideas are also obtained. feng chen (2014) mentioned the need to improve the object recognition mechanism of poverty alleviation, and advocated the development of specific support plans and measures in accordance with the characteristics of different households in different villages [1]. lijie zhang (2016) pointed out that in the current our country economy under the macro background of the new normal, precise poverty 111 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 alleviation work is facing new development opportunities, we must seriously implement the xi general secretary precision poverty alleviation idea, promote the poverty alleviation work theory innovation, system innovation and mechanism innovation [2]. qianqian tang (2016) mentioned when it comes to "precise poverty alleviation" as the 18th national congress emphasis on the implementation of poverty alleviation strategy after the new policy, which is china's poverty alleviation work mechanism to construct important train of thought in the new period, but due to the vast and complex conditions, and the obvious regional difference, the accuracy of the national poverty alleviation in different regions would present different results [3]; haiying liu (2011) advocated providing a fair share of economic and social development to poor areas through poverty reduction initiatives [4]. domestic scholars for accurate precision knowledge poverty alleviation policy for poverty alleviation mechanism study is a reference, but for how to effectively carry out precise poverty alleviation work is still exist some missing, which provides a reference for the discussion and development of this thesis. 3.2. foreign research status researches on poverty, foreign scholars have abundant theoretical achievements, especially the foreign research on the concept of poverty reduction. it has an important meaning to the construction and practice of our country's precision anti-poverty theory. amartya (2016) attributed the causes of poverty to unequal enjoyment of rights, and the contradiction between rights and distribution resulted in the reality of poverty [5]. so, amartya argues that the solution to the famine is to tackle inequality first. not only that, but also to give the poor people the basic rights and equal opportunities in the production, exchange and circulation. under the guidance of the amartya theory, the asian development bank in 2007 promoted the theory of poverty reduction with "equal opportunity" and "equitable sharing". this theory argues that the economic growth in earnings and the fair distribution to promote poverty reduction effect, to some extent, sharing type growth poverty reduction concept is "all equal opportunity or all share achievements" effective poverty reduction paradigm [6]. all in all, experts and scholars have expressed their concerns and opinions. it is not easy to think that the role of government in the targeted poverty alleviation is very important and the poverty alleviation itself is very complicated and tedious work. 4. establishment of panel data model 4.1. overview of panel data panel data refers to a collection of two dimensional data that is a combination of time series data and cross sectional data by intercepting multiple surfaces or multiple time points at a time sequence point. the panel data is used to describe the data sets of individual data and time series formed by a specific individual through repeated observation, which is the three-dimensional information of time, section and individual: from the point of time series, panel data can describe changes over time of different changes in law; from the cross-sectional data, the differences between individual individuals at a particular point in the panel data. its manifestations are as follows: ,y 11it itkitktittit xx μββα ++…++= ttni ,,2,1;,,2,1 …=…= among them, itα represents the constant term of the model, ),,,( 21 kititit xxxx …= represents k dependent variables in the model, and ),,,( 21 itttt ββββ …= is the coefficient matrix for k variables, itμ represents the random error terms and the assumptions that are satisfied: itμ are independent of each other, and the mean is 0 and the variance is 。 2 μσ 4.2. panel data model the regression model based on panel data is called panel data model. usually, linear panel data regression model is called panel data model. for convenience, linear panel data regression model is called panel data model. the general form of panel data model is: t,…1,2,tn;,…,2,1,y 1 ==++= ∑ = ∗ iux ithit k h hititit βα where is the observed value of the explained variable for the individual i at , is the explanatory variable for the observation of individual at ; ity t hitx thh i t hitβ to be estimated for the explanatory variable; is the random error term; the panel model is usually divided into three categories. thh itu 112 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 s. li, m. zhang that is, the mixed model, the fixed effect model and the random effect model, in which the fixed effect model is a fixed parameter, and the random error term denotes the influence of the neglected variation with time or individual. and given the non-observational effect and the explanatory variable, the expectation of the random error term is equal to zero. has the same square difference, and different individuals and different points correspond to each other. is independent of each other. its model takes the form of: ∗ itα itu itu itu itu if there are only individual effects in the model, the model form is as follows: iα as a random variable, this paper describes the difference between regression functions established by different individuals, which is called individual fixed effect model because it is not observable and is associated with the change of observable explanatory variable . hitx if there is only a time effect in the model, the form of the model is: tα as a random variable, this paper describes the difference between regression functions established by different individuals. because it is not observable and is associated with the change of observable explanatory variable t, it is called the time-point fixed effect i hitx model. α is a random variable, indicating that there are n different intercept terms for n individuals, and their variation is related to itx , and tγ is a random variable, indicating that there ar differe t intercept terms for t cross section (time), and that the variation is related to itx . this model is called individual time point double d effect model. e t n 4.3. selection of models in empirical analysis, we usually use the hausman test to decide whether to use the fixed effect model or the random effect model. the idea of the hausman test is that, under the original assumption 0)( = itit xve , (generalized least squares estimator) and glsβ̂ withinβ (the least squares estimator in the group) are unbiased and consistent, but in the case of the alternative hypothesis 0)( ≠itit xve , is not only partial but non-uniform, glsβ̂ withinβ is unbiased and consistent through the elimination of individual effects in groups. ithit k h hit ux ++= ∑ =1 ty βα if the original hypothesis is established, the probability limit of the two estimators is similar, whereas the probability limit is different. so hausman tests the probability of using the difference between and glsβ̂ withinβ to determine whether or not to reject the original hypothesis. ithit k h hiit ux ++= ∑ =1 y βα 5. empirical analysis 5.1. data source and sample selection in this paper, the annual data of all regions comes from the guizhou statistical yearbook from 2014 to 2016 and the government work report from 2014 to 2016 issued by the official website of each district county. part of the data is simply processed with excel tables to get schedule 1. ithit k h hit ux ++= ∑ =1 ty βα (1) poverty level index (pov): indicators of poverty level include poverty incidence, income gap ratio and the fgt index. this paper starts from the convenience of obtaining data and uses the poverty rate to measure the extent of poverty. (2) tourism development index (tod): a regional tourism revenues accounted for the size of the gross national product (gnp) is undoubtedly reflect the most compelling of the local tourism development index, this paper adopts the counties tourism income / gdp to reflect the status of each county tourism development. ithit k h hit ux +++= ∑ =1 tty βγα (3) the tourists attraction index (toa): the development of tourism is bound to be accompanied by an increase in the number of tourists, this paper adopts the counties with the number of visitors a year to illustrate the attraction of regional tourism to tourists. fixe (4) ecological-economic development index (daf): ecological tourism has become the new direction of tourism development. therefore, we also listed the ecological-economic development as an indicator to be 113 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 examined. the added value of ecological-economic are used to indicate here. (5) economic growth index (rgp): in the empirical study, gdp per capita tends to be better than that of gdp, so this paper measures economic growth by gdp per capita. (6) income distribution index (ig): the main body of this article is key county of guizhou province 50 poverty alleviation and development. considering the comparability and data item, we use urban per capita disposable income / rural per capita net income of is adopted to measure the equality of income distribution. based on the hypothesis that tourism development has a positive promoting effect on reducing poverty, this article selects development level of tourism, tourist attraction and ecological-economic development level as explanatory variables of the empirical model, considering the national consumption level in guizhou county precision for poverty alleviation, the influence of long economic rights and income distribution is chosen as the control variables of the model. the variables are shown in table 1: table 1. model variable scale variable types variable name variable definitions interpreted variable pov poverty degree tod tourism development level toa tourist attraction explanatory variables daf development level of agriculture, forestry and animal husbandry rgp economic growth control variables ig income distribution in addition, there are many factors that can reduce the incidence of poverty, such as the investment of government departments and the degree of education of residents, which we use itμ to represent the random error terms. get the following basic model: itititititit igrgpdaftoatodpov μβββββα ++++++= 54321 among them, represents 50 counties of poverty alleviation and development in guizhou province; means different years. 50,,3,2,1 …=i 2015,2014,2013=t 5.2. impact of tourism on targeted poverty alleviation 5.2.1 hausman test we believe that the poverty rate will be affected by tourism development, tourist attraction, agriculture and forestry, fishery development, economic growth and income distribution, also will be affected by other factors associated with particular observation object, or period. according to the original data of tourism industry development of 50 poverty-stricken counties in guizhou province from 2013 to 2015, the hausman test was carried out using the stata11.0 software, and the results were as table 2. table 2. hausman test results (b) (b) (b-b) sqrt(diag (v_b-v_b)) fe re difference s.e. tod -0.6618238 -0.4157728 -0.246051 0.0495937 doa -0.3819703 -0.790043 0.4080727 0.1682498 daf 0.6975831 0.8270739 -0.1294908 rgp -10.34866 -8.609656 -1.739001 0.3655479 ig 1.96336 1.96336 -1.213396 chi2(5)=(b-b)'[(v_b-v_b)^(-1)](b-b)=8.61 prob>chi2=0.1255 can be seen from table 2, prob > chi2 = 0.1255, hausman test is not significant under the confidence level of 95%, so accept the null hypothesis, that estimate relative to the use of fixed effects model, estimate the sample data by using random effects model is more suitable. 5.2.2 stochastic effect model regression according to the original data of 50 national povertystricken counties in guizhou province from 2013 to 2015, the data of the stata11.0 software was used to estimate the parameters, and the results of the fixed effect model were obtained as shown in table 3. table 3. regression results of random effects model pov coef. std. err. t p>|t| [95% conf. interval] tod -0.4157728 0.1061118 -3.92 0 -0.6237482 -0.2077975 doa -0.790043 0.3230725 -2.45 0.014 -1.423253 0.1568325 daf 0.0270739 0.2466582 3.35 0.268 0.3436327 1.310515 rgp -8.609656 0.8618021 -9.99 0 -10.29876 -6.920555 ig 3.176756 1.476361 2.15 0.031 0.2831411 6.070371 _cons 30.64046 5.939359 5.16 0 18.99953 42.28139 sigma_u 4.1794156 sigma_e 1.3203599 rho 0.90925186 corr(u_i,x)= 0(assumed) prob>chi2=0.0000 from the point of significance level, tourism development, tourist attraction, economic rights and income distribution coefficient of indicators 114 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 s. li, m. zhang significantly long is not zero; p value is less than 0.05. it is indicates that tourism development index and tourist attraction index coefficient is zero the possibility of less than 5%; and the p value of ecologicaleconomic development index is 0.268, shows an ecological-economic development indicators under 26.8% probability coefficient may be zero, 73.2% of the time coefficient is not zero, so the index can't pass the test of significance and should be deleted from the model. therefore, tourism development, tourist attraction, long economic rights and income distribution have significant influence to the poverty rate of 50 counties in guizhou, its elasticity were 0.4158, 0.79, 8.6097, 0.4158, while there is no significant impact on ecological-economic development indicators. 5.2.3 hausman test after deleting the no significant in the original data, the ecological-economic development index (daf) was removed and the hausman test was carried out with the stata11.0 software. the results were shown in table 4. table 4. hausman test results (b) (b) (b-b) sqrt(diag (v_b-v_b)) fe re difference s.e. tod -0.6296584 -0.4067843 -0.2228471 0.0547451 doa -0.2666472 -0.7460522 0.4794049 0.1825376 rgp -11.43676 -9.641491 -1.795274 0.3757357 ig 2.748999 4.232755 -1.483756 chi2(4)=(b-b)'[(v_b-v_b)^(-1)](b-b)=3.84 prob>chi2=0.4279 can be seen from table 4, prob > chi2 = 0.4279, hausman test is not significant under the confidence level of 95%, so accept the null hypothesis, that estimate relative to the use of fixed effects model, estimate the sample data by using random effects model is more suitable. 5.2.4 regression of random effect model according to the 2013 to 2015, 50 state-level povertystricken counties in guizhou raw data, delete the ecological-economic development indicators, again use stata11.0 software for parameter estimation, concluded as shown in table 5 the regression results of random effects model. table 5. regression results of random effects model pov coef. std. err. t p>|t| [95% conf. interval] tod -0.4067843 0.1095295 -3.71 0 -0.6214581 -0.1921105 doa -0.7460522 0.3336913 -2.24 0.025 -1.400075 0.0920293 rgp -9.641491 0.8373706 -11.51 0 -11.28271 -8.000274 ig 4.232755 1.477959 2.68 0.004 1.336009 7.129501 _cons 34.80641 5.991294 5.81 0 23.06369 46.54914 sigma_u 4.4968712 sigma_e 1.3852328 rho 0.913333 corr(u_i,x)= 0(assumed) prob>chi2=0.0000 from table 5, we can see from the estimated coefficients of variables, tourism development indicators, tourist attraction indicators and economic growth indicators are negatively correlated with poverty rates in guizhou poverty-stricken counties. it is suggested that the increase of tourism development, tourist attraction and economic rise can slow the poverty status of guizhou national poverty-stricken counties. income distribution and poverty rate were positively related, namely the wider income distribution gap , the higher incidence of poverty-stricken counties in guizhou, which conform to the real economic significance. from the point of the estimated coefficients of variables, the p values of tourism development, tourist attraction, long economic rights and income distribution indicators were less than 0.05, which is under the 95% confidence level were significantly. therefore, the tourism development index and tourist attraction index of explanatory variables have a positive effect on reducing the poverty rate in guizhou poverty-stricken counties. therefore, the regression equation can be obtained: ititititit igrgptoatodpov μ++= 2328.46415.9-7461.0-4068.0-8064.34 6. conclusion by using the stochastic effect model for empirical research, we find that the development of tourism, the attraction of tourists and the increase of economic growth can slow the poverty situation in guizhou's national poverty-stricken counties, the wider income distribution gap, the higher incidence of povertystricken counties in guizhou. therefore, in addition to increasing input in order to promote the development of the tourism industry, increasing the visibility of tourist attractions or developing tourism economy is also an important way to reduce poverty. from the significance level, it is the development of tourism industry and 115 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 economic rise of the economic rise, which has a significant influence on the poverty rate in guizhou. the tourism population is slightly lower, which may be caused by the uneven tourism consumption level in guizhou province. through the establishment of panel data model research, it is shown that increasing the investment of tourism industry in all regions has a certain positive effect on reducing the incidence of poverty. investment is more than just the poverty alleviation office government to invest more, but it can also encourage entrepreneurs to help the poor and provide funds or technology for win-win results. at the same time, the government should increase financial input, pay more attention to the construction of public infrastructure and alleviate the road traffic problems in remote areas. in recent years, scholars have done a lot of research on poverty, but there are few studies on the incidence of poverty. based on the tourism industry, this paper collects panel data of 50 counties in guizhou. after simple processing of the data, a panel data model is established, we compared the advantages and disadvantages to determine the final proper model, and then draw the conclusion. but because my ability is limited, in the depth of the research and analysis, variable selection method and depth is relatively shallow, so this paper in terms of data extension and influencing factors of development still has a certain research space. 7. propose based on the above results, the data of poverty incidence in 88 counties (cities and districts) of guizhou province are analyzed. it is found that the change of poverty incidence is influenced by a series of influencing factors. some of the factors we consider can be adjusted through national and local policies... in the process of poverty reduction and development, while ensuring stable and rapid economic growth in all districts and counties of guizhou province, we can also adopt local pro-poor and pro-agricultural policies and social security. other factors affecting poverty can be changed in terms of multi-channel financing. this not only guarantees stable economic growth, but also improves people's living standards. to reduce the scope of poverty and reduce the degree of poverty, to achieve stable economic growth income steadily improve poverty reduction effect of good win-win situation. this paper studies the incidence of poverty in all counties and cities of the province as the starting point. according to the situation of guizhou province, the countermeasures and suggestions for poverty reduction in guizhou province are put forward. (1) in terms of economic development, it can be roughly seen from the distribution map of poverty in various regions that key poverty counties in guizhou province are mainly located in remote areas with low economic levels, so we must first improve the overall poverty situation in the whole province. first of all, it is necessary to grasp the lifeblood of the overall economy, vigorously develop the market economy, encourage technological innovation and research and development, move closer to technology-intensive industries, rationally develop the economy and implement differentiated economic development according to the geographical regions and resource advantages of the 88 counties. in addition, the development of economy in the whole province should grasp the rational and sustainable development mode, protect the construction of ecological civilization, rationally utilize the exploitation of resources, and realize the sustained and steady growth of the economy. (2) the province has actively promoted the development of industrialization, increased employment posts, reduced the urban unemployment rate, and provided job opportunities for migrant workers in cities. however, at present, the income gap between urban and rural residents is widening, leading to a polarization between the rich and the poor. seriously restricting the development of rural economy, and because the various social security systems for migrant workers are not perfect, it is easy to cause the old labor force to lose and return to poverty again. therefore, in the course of promoting the urbanization and industrialization construction actively in the province, actively ensure the employment of rural personnel and low-income urban residents, feed modern agricultural production with industry, and promote the upgrading and transformation of rural industries by the development of urban economy. it is not a jump process that needs to increase the overall development of urban and rural areas, adjust the traditional urban-rural dual economic structure, enhance the interaction between urban and rural areas, and actively guide and cooperate with each other, so as to put into practice the system of ensuring the overall development of urban and rural areas 116 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 s. li, m. zhang formulated by the state. do not allow national preferential policies from the central to local gradually devalued. (3) research on panel data model shows that increasing fixed asset investment in various regions has a certain positive effect on reducing the incidence of poverty. since the new round of poverty alleviation work has been carried out by the public office of the state council for poverty alleviation, efforts to alleviate poverty and development have been intensified. through such effective means as continuous development, village promotion and industrial poverty alleviation, we should increase investment in povertystricken areas. investment is not just an increase in government investment in poverty alleviation offices and an increase in rural financing channels. at the same time, entrepreneurs can be encouraged to help industries in poor areas, providing capital or technology to achieve win-win results. the government should increase financial expenditure, increase the construction of rural public infrastructure, and effectively alleviate the problem of water consumption for agricultural production in high mountain areas in summer. (4) rural workers go to cities to work and raise family income effectively, but the problem of the disadvantaged situation of children left behind and the elderly living alone in rural areas arising gradually from going out of the young labor force should be guaranteed both to increase their income and to ensure a smooth solution to such problems. we must vigorously support the development of township and township industries in counties, cities and towns, actively build local characteristic industries in combination with local advantages, adopt rural cooperative management methods, and integrate resources for long-term development; in addition, farmers who are willing to engage in secondary and tertiary industries, rely on immigrants to expand towns and new rural construction, through industrial skills training, promote employment and other forms of industrialization poverty alleviation, adhere to rural areas to adhere to rural poverty alleviation labor force nearby employment principle, avoid rural labor force outflow, raise local economic level strength. acknowledgements this paper is one of the periodic results of the study on the study on the efficiency and mechanism of financial support for the poverty-alleviation of guizhou specialty industry. references [1] chen f. statistical analysis of poverty rate in guizhou province. guizhou university of finance and economics, 2014. [2] zhang l j. study on the implementation of precision poverty alleviation policy of county-level government. zhengzhou university, 2016. [3] tang q q. research on the internal mechanism and efficiency of financial support for strategic emerging industries. guizhou university of finance and economics, 2016. [4] liu h y. great poverty alleviation: practice and suggestion of public welfare organizations. social sciences literature publishing house, 2011. [5] he x. poverty alleviation in zang people's regions of yunnan province from the perspective of accurate anti-poverty. international conference on society science. 2017. [6] yang y, zhang a. precise poverty alleviation problems and countermeasures in xinjiang uygur autonomous region. international education, economics, social science, arts, sports and management engineering conference. 2016. appendix a. related index data of 50 state-level poor counties in guizhou province in the past 3 years area year pov fs fe fsr rgp ig 2013 0.1997 1.5620 1.7100 0.0000 2.0371 3.2719 2014 0.1698 1.3608 2.0656 0.0000 2.5648 3.0126 liuzhi special zone 2015 0.1390 1.3238 2.3333 0.0000 2.9673 2.9963 2013 0.2886 4.7495 0.5234 0.0000 1.8619 3.3206 2014 0.2481 5.2849 0.5956 0.0000 2.4496 3.0630 shuich eng county 2015 0.2100 5.2321 0.6428 0.0000 2.7828 3.0200 2013 0.2349 0.9845 1.9424 0.0000 3.5049 3.1507 2014 0.1879 0.9092 2.3680 0.0000 4.0896 2.9045 pan county 2015 0.1430 1.0151 2.2980 0.0000 4.5397 2.8861 2013 0.2162 2.3025 0.7822 0.0000 1.1360 3.4571 2014 0.1638 2.0470 0.8642 0.0000 1.5592 2.9436 zheng ’an count 2015 0.1510 2.3394 0.8765 0.0000 1.8475 2.9222 2013 0.1961 2.4699 0.7183 0.0000 1.3184 3.6118 2014 0.1636 2.3703 0.7921 0.0000 1.6275 2.9399 daozh en county 2015 0.1160 2.3788 0.8025 0.0000 1.9703 2.9214 117 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 2013 0.2353 2.5731 0.7207 0.0000 1.0852 3.6755 2014 0.1874 2.4613 0.8310 0.0000 1.3473 2.9448 wuchu an county 2015 0.1370 2.7818 0.7780 0.0000 1.6098 2.9316 2013 0.2284 1.5973 1.4299 0.0000 1.7580 3.5500 2014 0.1778 0.9054 3.1268 0.0000 2.1734 2.9301 xishui county 2015 0.1320 1.6925 1.6957 0.0000 2.4662 2.9142 2013 0.2547 1.5579 1.2553 0.0000 1.5692 3.3232 2014 0.2077 1.4798 1.7643 0.0000 1.8760 3.2807 puding county 2015 0.1510 1.6890 1.4374 0.0000 2.1876 3.2072 2013 0.2574 2.1540 0.7718 1.0000 1.8314 3.3715 2014 0.2176 1.8775 1.1052 1.0000 2.2544 3.3173 zhenni ng county 2015 0.1870 1.9718 1.0534 1.0000 2.6310 3.2871 2013 0.2865 1.5957 0.5869 0.0000 1.6678 3.3401 2014 0.2416 1.3686 0.9200 0.0000 2.0134 3.2758 guanli ng county 2015 0.1800 1.5269 0.8919 0.0000 2.4092 3.2343 2013 0.2868 1.8025 0.6788 0.0000 1.3071 3.3083 2014 0.2427 1.6404 0.9083 0.0000 1.6589 3.2834 ziyun county 2015 0.1930 1.8093 0.9055 0.0000 1.8985 3.1803 2013 0.2776 1.0846 1.5632 0.0000 1.6673 3.3181 2014 0.2351 0.9710 1.8221 0.0000 2.0372 3.2741 dafang county 2015 0.1910 1.0792 1.6908 0.0000 2.2897 3.2005 2013 0.3195 1.5369 1.8684 0.0000 1.3712 3.4599 2014 0.2732 1.4326 2.4169 0.0000 1.7361 3.3147 zhijin county 2015 0.2250 1.7864 1.7137 0.0000 1.9573 3.2461 2013 0.2685 0.9434 2.3442 0.0000 1.8370 3.5131 2014 0.2239 0.9022 2.7703 0.0000 2.2095 3.4727 nayon g county 2015 0.1810 0.9971 2.3715 0.0000 2.5354 3.3979 2013 0.2405 0.9114 2.2688 0.0000 0.9767 3.3157 2014 0.1863 0.8969 2.2487 0.0000 1.1992 3.2773 weinin g county 2015 0.1450 0.9656 1.8456 0.0000 1.4882 3.1898 2013 0.2605 1.2804 1.4626 0.0000 1.0138 3.5010 2014 0.2177 1.0915 1.6651 0.0000 1.3937 3.4606 hezha ng county 2015 0.1790 1.1539 1.3654 0.0000 1.7112 3.3678 2013 0.2350 2.6010 1.1867 0.0000 1.6017 3.3829 2014 0.1878 2.4635 1.5219 0.0000 1.9289 3.2618 jiangk ou county 2015 0.1380 2.3778 1.4809 0.0000 2.3858 3.2857 2013 0.2586 2.0468 0.7560 0.0000 1.2592 3.3528 2014 0.2120 1.9425 0.8501 0.0000 1.5657 3.1903 shiqia n county 2015 0.1650 2.1168 0.7530 0.0000 1.9547 3.1842 2013 0.2543 2.0290 1.3253 0.0000 1.4778 3.5352 2014 0.2132 1.9354 1.4767 0.0000 1.7667 3.2799 sinan county 2015 0.1640 2.0426 1.3880 0.0000 2.0282 3.2694 yinjian g county 2013 0.2364 2.0946 1.0596 0.0000 1.7001 3.4483 2014 0.1913 1.9265 1.2793 0.0000 2.0722 3.2351 2015 0.1430 1.9747 1.1913 0.0000 2.6070 3.2720 2013 0.2880 1.4384 1.1893 0.0000 1.6341 3.6942 2014 0.2579 1.4075 1.3866 0.0000 1.9852 3.3293 dejian g county 2015 0.2020 1.5073 1.3448 0.0000 2.2807 3.3124 2013 0.2713 1.6461 1.4640 0.0000 1.3106 3.5447 2014 0.2295 1.5026 1.6962 0.0000 1.6125 3.3216 yanhe county 2015 0.1700 1.6187 1.5581 0.0000 1.8766 3.2763 2013 0.2333 1.7564 1.1832 0.0000 1.4971 3.5246 2014 0.1889 1.5295 1.4697 0.0000 1.8223 3.2933 songta o county 2015 0.1420 1.6677 1.3339 0.0000 2.1132 3.2999 2013 0.2070 1.4665 1.5932 0.0000 1.8413 3.3526 2014 0.1612 1.4060 1.9500 0.0000 2.1951 3.2807 xingre n county 2015 0.1060 1.4848 0.8705 0.0000 2.7007 3.2391 2013 0.1963 1.5688 1.1677 0.0000 1.6741 3.6717 2014 0.1517 1.4999 1.4472 0.0000 1.9705 3.4866 pu’an county 2015 0.1050 1.5515 1.4830 0.0000 2.3385 3.4370 2013 0.3741 1.4027 1.4547 0.0000 1.4956 3.8001 2014 0.3277 1.3350 1.7043 0.0000 1.7919 3.6365 qinglo ng county 2015 0.2570 1.3734 2.1179 0.0000 2.2231 3.5218 2013 0.2562 1.4204 1.0073 0.0000 2.0521 3.4162 2014 0.2065 1.3977 1.1601 0.0000 2.5059 3.2789 zhenfe ng county 2015 0.1610 1.3313 0.6826 0.0000 2.9813 3.2789 2013 0.3302 2.1944 0.7123 0.0000 1.0232 4.0494 2014 0.2782 1.9558 0.8495 0.0000 1.4544 3.7188 wang mo county 2015 0.2120 1.8425 0.6996 0.0000 1.9023 3.6595 2013 0.3273 2.3924 0.6503 0.0000 1.2101 3.8645 2014 0.2917 2.2186 0.7470 0.0000 1.5128 3.6773 cehen g county 2015 0.2090 1.9262 2.0920 0.0000 1.9221 3.6284 2013 0.1877 1.3995 0.8662 0.0000 1.7737 3.4391 2014 0.1440 1.3832 1.0388 0.0000 2.0314 3.2890 anlong county 2015 0.0940 1.7626 0.4974 0.0000 2.4505 3.2979 2013 0.3004 1.9860 0.7621 0.0000 1.1815 3.7572 2014 0.2790 1.8816 0.9165 0.0000 1.4245 3.5437 huang ping county 2015 0.2300 2.0605 0.6161 0.0000 1.6621 3.4960 2013 0.3150 1.6230 1.1558 0.0000 1.7763 3.3146 2014 0.2681 2.0655 0.7784 0.0000 2.0910 3.1398 shibin g county 2015 0.2190 2.0764 1.1981 0.0000 2.3598 3.1285 2013 0.3379 1.7702 1.1428 0.0000 1.6071 3.5407 2014 0.2909 1.6284 1.6463 0.0000 1.9833 3.3523 sansui county 2015 0.2380 1.8556 1.4249 0.0000 2.2779 3.3075 2013 0.2637 1.9118 1.0224 0.0000 1.6613 3.5417cengo ng 2014 0.2539 1.7591 1.2600 0.0000 1.9863 3.3480 118 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 s. li, m. zhang county 2015 0.1990 2.2469 1.0210 0.0000 2.2779 3.2971 2013 0.3176 1.7351 0.9835 0.0000 1.8871 3.3986 2014 0.2765 1.6470 1.1170 0.0000 2.2678 3.2196 tianzh u county 2015 0.2250 1.6809 1.1953 0.0000 2.5857 3.1822 2013 0.3275 2.0483 0.5913 0.0000 1.6432 3.8270 2014 0.2854 1.9548 0.6780 0.0000 1.9746 3.5833 jinping county 2015 0.2370 2.1287 0.6660 0.0000 2.2632 3.5319 2013 0.3248 1.7745 0.9054 0.0000 1.4525 3.7346 2014 0.2829 1.7187 1.0360 0.0000 1.7250 3.5258 jianhe county 2015 0.2370 1.8588 0.9957 0.0000 1.9953 3.4880 2013 0.3201 2.6150 1.2902 0.0000 1.6713 3.9039 2014 0.2878 2.3378 1.6208 0.0000 1.9982 3.6416 taijian g county 2015 0.1990 3.0975 1.1386 0.0000 2.2931 3.6023 2013 0.2738 2.1670 0.8683 0.0000 1.2132 3.6510 2014 0.2448 1.9845 1.0485 0.0000 1.4719 3.4456 liping county 2015 0.2250 2.0841 1.0524 0.0000 1.7300 3.4425 2013 0.3622 2.1586 0.9597 0.0000 1.2366 3.7655 2014 0.3373 1.9742 1.1567 0.0000 1.4460 3.5307 rongji ang county 2015 0.2760 2.1531 1.0975 0.0000 1.6868 3.4802 2013 0.3168 1.4098 1.5943 0.0000 1.1850 3.5328 2014 0.2901 1.3438 1.8033 0.0000 1.4128 3.3457 congji ang county 2015 0.2390 1.4400 1.6609 0.0000 1.6614 3.3071 2013 0.2798 2.0373 0.8388 0.0000 1.4951 3.5638 2014 0.2648 2.1268 0.8765 0.0000 1.7947 3.3613 leisha n county 2015 0.2080 2.3375 0.8282 0.0000 2.0324 3.3134 2013 0.3768 1.9331 1.0642 0.0000 1.5451 3.7132 2014 0.3276 2.3109 0.6275 0.0000 1.8280 3.5102 majian g county 2015 0.2060 2.8289 0.5394 0.0000 2.1728 3.4473 2013 0.3373 2.3746 1.3205 0.0000 1.4748 3.7290 2014 0.2924 2.2849 1.5348 0.0000 1.7435 3.5065 danzh ai county 2015 0.2381 2.4103 1.5601 0.0000 1.9967 3.4564 2013 0.3092 1.8763 1.0938 0.0000 2.4050 3.2383 2014 0.2556 1.7939 1.2754 0.0000 3.2619 3.0447 libo county 2015 0.2109 1.9702 1.2838 0.0000 3.5620 3.0310 2013 0.2730 1.9787 0.7490 0.0000 1.6214 3.2014 2014 0.2273 1.7662 0.9972 0.0000 2.0422 2.9973 dusha n county 2015 0.1759 2.0870 0.9221 0.0000 2.3035 2.9623 2013 0.3151 1.6425 1.2050 0.0000 1.3571 3.2600 2014 0.2598 1.5143 1.3828 0.0000 1.7760 3.0491 pingta ng county 2015 0.1955 2.0246 1.1493 0.0000 1.9885 3.0271 2013 0.3551 1.6731 1.1905 1.0000 1.5948 3.0516 2014 0.3027 1.5089 1.4432 1.0000 1.9817 2.8634 luodia n county 2015 0.2416 1.6755 1.4443 1.0000 2.2175 2.8479 2013 0.3183 1.4607 1.3845 0.0000 1.7166 3.1800 2014 0.2656 1.3031 1.8096 0.0000 2.2211 2.9792 changs hun county 2015 0.2052 1.7363 1.3846 0.0000 2.4685 2.9309 2013 0.3562 1.8810 1.0051 0.0000 1.1991 3.2038 2014 0.3230 1.6408 1.1906 0.0000 1.6361 2.9842 sandu county 2015 0.2526 1.9236 1.0769 0.0000 1.8281 2.9600 119 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 110-119 1. preface 2. writing background 3. literature review 3.1. domestic research status 3.2. foreign research status 4. establishment of panel data model 4.1. overview of panel data 4.2. panel data model 4.3. selection of models 5. empirical analysis 5.1. data source and sample selection 5.2. impact of tourism on targeted poverty alleviation 5.2.1 hausman test 5.2.2 stochastic effect model regression 5.2.3 hausman test after deleting the no significant 5.2.4 regression of random effect model 6. conclusion 7. propose acknowledgements references atlantis press journal style the human health risk assessment based on process simulation and uncertainty analysis yue pan, xiankui zeng*, xinyu gao, jichun wu, dong wang school of earth sciences and engineering, nanjinguniversity, nanjing 210023, china received november 1, 2017 accepted november 25, 2018 abstract water resource is an important part of human living and production. with the modernization of cities, the problem of groundwater contamination is becoming more and more serious. it is outstanding to effectively assess the risk of groundwater contamination to human health. this paper proposes a human health risk assessment method based on the simulation of contaminant transport process and parameter uncertainty analysis. the method combines health risk assessment with groundwater contaminant transport processes and applies the tough2 program to simulate the temporal and spatial distribution of contaminant in groundwater. in order to improve the accuracy of the assessment results, the influence of the uncertainty of the transport model on human health risk assessment is taken into account. and the key parameter (e.g., the permeability) of the transport model is inversed by markov chain monte carlo simulation method. based on a sandbox experiment, the human health risk exposed to a typical dense non-aqueous phase liquids contaminant tetrachloroethylene (pce) is evaluated by considering the parameter uncertainty of the pce transport model, which confirmed that the method can provide decision-making information for the management and prevention of contaminated sites. keywords: human health risk, dense non-aqueous phase liquids, model parameter, uncertainty 基于过程模拟与参数不确定性的人体健康风险评价 潘玥，曾献奎*，高鑫宇，吴吉春，王栋 南京大学地球科学与工程学院，南京 210023, 中国 摘要：水资源是人类生存与生产的重要组成部分，随着城市的现代化发展，地下水污染问题日益严峻，有 效评估地下水污染对人体健康的风险显得尤为重要。本文提出了一种基于污染物运移过程模拟及参数不确 定性分析的人体健康风险评价方法。该方法将健康风险评价与地下水污染物运移过程相结合，利用 tough2 程序模拟有机污染物在地下水中的时空分布。为了提高评价结果的可靠性，考虑污染物运移模拟 模型的不确定性对人体健康风险评价的影响，通过马尔科夫链蒙特卡洛模拟方法对模拟模型的关键参数 （如介质渗透率）进行反演识别。基于一个室内砂箱实验，选取典型重非水相污染物四氯乙烯（pce）作 为污染物对象，并通过考虑 pce 运移模拟模型的参数不确定性来评价人体健康风险，证实了该方法能够为 污染场地的管理与防治提供决策信息。 关键词：人体健康风险，重非水相流体（dnapl），模型参数，不确定性 1. 引言 * 水是人类赖以生存的重要资源。随着社会经济 的快速发展，地下水污染已成为威胁生态系统与人 体健康的重要环境问题。1 除了常规污染物重金属离 * corresponding author: xiankuizeng@nju.edu.cn 子、氮元素、氯元素外，各种理化性质较为复杂的 典型有机污染物逐渐引起社会的重视，这一类有机 污染物密度比水大且不溶于水，又被称为重非水相 流体（dnapl）。2 由于这类污染物的理化性质比 可溶污染物更复杂，增加了人们对地下水污染监测 及防治工作的难度，为了控制地下水中 dnapl 对 人类健康的危害，针对典型有机污染物 dnapl 的 199 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 199 -202 y. pan, x. k. zeng, x. y. gao, et al 健康风险评价已成为地下水环境管理的重要组成部 分。 人体健康风险评价是指评价人体在潜在污染环 境中受到的可能危害健康的影响概率。3 评价内容包 括识别潜在污染源，预估人体所接触的污染物浓度， 以及量化暴露于污染的健康风险。常用的评价模型 是 美 国 环 保 署 提 出 的 rags （ risk assessment guidance for superfund）模型，评价过程可以分为四 步：（1）危险识别，（2）剂量反应，（3）暴露评 估，（4）风险表征。人体健康风险评价的重点在于 准确计算污染物的浓度。4 常规的污染物浓度获取办 法是实地取样与实验室分析。5-6 但事实上，污染物 对人体的影响是持续的，随着时间的推移，污染物 浓度的时空分布会发生显著变化，因此，借助数值 模拟技术利用实时动态变化的浓度值进行人体健康 风险评价是十分有必要的。 由于地下水系统的复杂性与多变性，污染物的 运移过程受到诸多因素的影响，如流速、含水层性 质、污染物降解作用等。因此，利用模拟模型描述 污染物运移迁移过程时，模拟结果（如污染物浓度） 通常会受到这些因素的影响而具有不确定性，导致 依据模拟结果进行人体健康风险评价时，评价结果 具有不确定性。本次研究通过室内砂箱 dnapl 运 移实验进行人体健康的风险评价，运用 tough2 程 序模拟污染物的运移过程，采用马尔科夫链蒙特卡 洛（mcmc）方法分析了 dnapl 运移模型关键参 数（介质渗透率）的不确定性，利用 dnapl 饱和 度观测数据进行参数反演获得其后验分布，在考虑 模型参数不确定性的基础上进行人体健康风险评价。 2. 研究方法 2.1. 人体健康风险评价 人体健康风险评价基于 rags 模型，采用常规 评价指标：人体一生患癌症风险（elcr）进行定量 评价，本次选取的风险阈值为 10 -6 。elcr 是一种随 时间变化的综合指标，仅依赖于研究区内的环境敏 感位置。本次的环境敏感位置由控制平面 x 表示， 与地下水流向相垂直。dnapl 污染物为四氯乙烯 （pce），elcr 的表达式可以写作： elcr cdi cpf  （1） c ir ef ed cdi bw at      （2） 其中，cdi 表示慢性每日摄入量（mg/kg·d）； cpf 表示潜在致癌因子代谢率；ir 表示每日饮水量 （l/d）；bw 表示体重（kg）；ed 表示暴露时长 （y）；ef 表示暴露频率（d/y）；at 表示人的寿 命（d）；c 表示地下水中的污染物浓度（mg/l）。 2.2. dnapl 运移模型 dnapl 是一种与水不混溶的液体，因此其运移 问题可以理解为多相流动问题，各相（如水相、气 相、dnapl 相）在地下水系统中的运移过程以达西 定律为基础得到平衡控制方程：     n n n n v r n n v s x dv k k p d q dv                                   g n （3） 其中 t 表示时间（s）；φ 表示孔隙度；sβ 表示 β 相 的饱和度；ρβ 表示 β 相的密度（kg/m 3 ）；xβ κ 表示 β 相中 κ 质所占的质量分数；vn 表示流域；γn 表示流 域表面积（m 2 ）；k 表示绝对渗透率（m 2 ）；krβ 表 示 β 相的相对渗透率（m 2 ）；μβ 表示 β 相的动态粘 滞系数（cp）；pβ 表示 β 相的流体压力（pa）；n 表示法向量；g 表示重力加速度（m/s 2 ）；q κ 代表单 位体积内生成或消耗热量的速率。本次研究利用多 相流模拟程序 tough2 模拟 dnapl 的运移过程。 2.3. 马尔科夫链蒙特卡洛模拟 针对一个非线性方程（如 dnapl 运移方程） m=f(θ)，其观测数据可以表示为： ( )f d   （4） ( | ) ( ) ( | ) ( | ) ( ) p p p p p d   d d d       （5） 其中 p(θ)代表参数 θ 的先验分布密度；p(d|θ)代表参 数的似然函数，即模型参数为 θ 时，模拟结果与观 测数据 d 之间的拟合程度。 马尔科夫链蒙特卡洛模拟的基本思路可以总结 为：（1）根据专家知识或经验，设置模型参数 θ[θ1, θ2…, θd]的先验分布，如分布函数、分布范围等信息； （2）定义似然函数来描述模型模拟值 f(θ)与观测数 据 d 之间的拟合程度，通常采用高斯似然函数； （3）选择一种抽样算法在模型参数 θ 的先验分布范 围内随机抽样，通过构建平稳分布的马尔科夫链， 实现对目标函数 p(θ)概率分布空间的充分搜索，获 得模型参数的后验概率分布 p(θ|d)；（4）根据获得 的模型参数的后验概率分布，输入模拟模型 f，获得 200 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 199 -202 模型输出的后验概率分布。本次研究步骤（1）中， 需要反演识别的模型参数 θ 为介质渗透率，步骤（3） 中，使用高效稳定的 dream 算法作为抽样算法。 3. 案例分析 使用的二维砂箱尺寸为长（x）0.60m，宽（y） 0.45m，厚（z）0.016m，在砂箱内均匀填充粒径范 围在 0.5mm~0.35mm 的石英砂，在 x=0 与 x=0.6m 处 设定水头边界，用蠕动泵控制砂箱内水流流速，使 其 稳 定 在 1m/d ， 污 染 物 注 入 点 位 于 x=0.3m ， y=0.4m，z=0.008m，如图 1 所示。 在模拟开始时刻，恰好 pce 停止注入，砂箱中 pce 分布如图 2，此时 pce 已经运移至砂箱底部并 形成堆积相，运移路径上存在 pce 残余相。环境敏 感面取靠近左边界的一侧的剖面。在 dnapl 数值 模拟过程中，介质渗透率、孔隙度、dnapl 进液压 力倒数等参数均对 dnapl 分布具有显著影响，其 中介质渗透率的敏感性较高。8 常规的模拟方法是采 用石英砂介质的经验渗透率进行模拟，但得到的结 果具有一定的不确定性。本案例假设具有较高敏感 性的石英砂介质渗透率是未知变量（k），并以此为 前提建立 pce 运移数值模型。利用光透法监测污染 物运移期间在砂箱中的饱和度分布情况，得到数字 图像，并采用 matlab r2014(a)软件编译程序处理数 字图像，得到砂箱内 x=0.008m 剖面处 pce 饱和度 数值。7 在整个砂箱内均匀选取 608 个观测点，其中 x 方向上等距布设 32 列，y 方向上等距布设 32 列， y 方向上等距布设 19 行（32×19），观测时间选取 污染物停止注入后的第 0 分钟、第 20 分钟、第 40 分钟三个时刻。 4. 结果分析与讨论 为对比参数不确定性对风险评价的影响，对介质 渗透率进行参数识别，利用识别后的渗透率进行人 体健康风险评价，随后将渗透率设为固定的参考值 再次进行风险评价，并对比考虑与不考虑渗透率不 确定性条件下的人体健康风险评价结果。 （1）参数不确定性分析。利用 dream 算法对 模型参数的概率空间进行搜索时，设置三条平行的 马尔可夫链，每条链的长度设定为 25,000，准备期 （burn-in period）的长度设定为 10,000，利用每条 链余下长度（15,000）生成模型参数的后验分布， 最终获得 45,000 个参数后验样本，其概率密度分布 如图 3 所示。 从图中可以看出，参数的分布得到明显收敛， 图 1 砂箱示意图 fig.1. schematic diagram of the sandbox 图 2 dnapl 饱和度分布图 fig.2. the saturation distribution of dnapl 图 3 渗透率参数 k 的后验概率分布图 fig.3. the probability density function of permeability 201 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 199 -202 y. pan, x. k. zeng, x. y. gao, et al 表明渗透率对于观测值（饱和度）具有较高的敏感 性，k 的先验分布范围为[2.0×10 -10 ,6.0×10 -10 ]，识别 得到的后验分布范围缩小至[3.51×10 -10 ,3.87×10 -10 ]。 渗透率的后验分布显示了双峰分布的特征，表明了 参数概率空间的复杂性。 （2）参数不确定性对人体健康风险评价的影响。 在后验参数样本的基础上，利用 rags 模型得到 elcr 的频数分布。如图 4 所示，受渗透率后验分 布的影响，elcr 的分布同样表现出明显的多峰分 布特征，结合模型参数的不确定性，这表明 elcr 的概率空间非常复杂。为了对比参数不确定性对风 险 评 价 结 果 的 影 响 ， 将 参 数 k 固 定 在 经 验 值 （ k=1.33 × 10 -10 ）， 7 模拟得到 elcr 的经验值 （elcr=1.357×10 -5 ），在图中用黑色虚线表示， 该值在 elcr 分布中出现的概率较低。因此，将污 染物运移模型的参数设定为确定值，忽略参数不确 定性来进行人体健康风险评价是不准确的。 5. 结论 人体健康风险评价是污染场地修复的重要基础 工作，为了使得评价结果更加真实可信，通常需要 建立污染物的运移模型。本次研究评价了暴露于 dnapl 的人体健康风险，并使用患癌风险指标 elcr 量化评价结果。基于室内砂箱实验，通过贝 叶斯参数不确定性分析方法，对比研究了 pce 运移 模型的参数不确定性对健康风险评价结果的影响。 结果表明，用确定的模型参数值得到的人体健康风 险评价结果是不可靠的，而基于贝叶斯不确定性分 析得到的风险评价结果可以为决策者提供更多的信 息，具有更高的可信度。 致谢 感谢评审专家对论文提出宝贵修改意见。本研究受 到中央高校基本科研业务费资助（0206-14380040）。 references j. chen, h. qian, h. wu, et al, assessment of arsenic and fluoride pollution in groundwater in dawukou area, northwest china, and the associated health risk for inhabitants. environ. earth sci., 76(8)(2017):314-328. m. l. brewster, a. p. annan, j. p. greenhouse, et al, observed migration of a controlled dnapl release by geophysical methods. groundwater, 33(6)(1995):977– 987. u. s. epa, risk assessment guidance for superfund volume i: human health evaluation manual (part f, supplemental guidance for inhalation risk assessment). in: u.s. epa (editor), washington dc, 2009. u. s. epa, risk assessment guidance for superfund (rags) part a. in: u.s. epa (editor), rep. epa, washingtondc, 1989. h. mishra, v. yadav, s, karmakar,et al, quantification of human health risk to msw landfill leachate contamination for major indian cites, international conference on solid waste technology and management. (2017). r. a. fallahzadeh, m. miri, m. taghavi, a. gholizadeh, et al, spatial variation and probabilistic risk assessment of exposure to fluoride in drinking water. food chem. toxicol., 113(2018):314-321. m. wu, z. cheng, j. f. wu, quantifying representative elementary volume of connectivity for translucent granular materials by light transmission microtomography. j. hydrol., 545(2017):12-27. 施小清,吴吉春,刘德朋等.饱和介质中重非水相液体运移的 数 值 模 拟 及 敏 感 性 分 析 , 南 京 大 学 学 报 ( 自 然 科 学)47(3)(2011)299-307. x. q. shi, j. c. wu, d. p. liu, et al, numerical simulation of transportation of dense nonaqueous phase liquids in the subsurface environment. j. nanjing univ., 47(3)(2011): 299-307. 图 4 elcr 概率分布图（黑色虚线表示 elcr 经验值） fig.4. the probability density function of elcr (the black dash represented the reference value of elcr) 202 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 199 -202 atlantis press journal style identification of global indicators for regulatory risk management in production platforms case study of brazilian industry alex garcia de almeida technical advisor at safety and environmental department, national agency of petroleum, natural gas and biofuels av rio branco 65, 18 andar, centro, rio de janeiro-rj, brazil e-mail: agalmeida@anp.gov.br marco antonio gaya de figueiredo operation a industrial project department, state university of rio de janeiro são francisco xavier street , 524, maracanã rio de janeiro cep 20550-900, brazil e-mail: mgaya@uerj.br abstract this study aims to analyze different indicator’s programs from regulators and other organizations, select and apply methods to normalize and compose global indicator. using obtained information related to risk management system and incidents from 67 oil and gas production platforms in brazil, the results show a new method to follow offshore performance and its companies, in lagging and leading focuses, showing a tool to improve regulator’s overview. keywords: offshore platforms, process safety, leading indicator, lagging indicator, global indicator 1. introduction the selection and monitoring of performance indicators, both in a leading focus to monitor the performance of safety barriers, as in lagging focus to monitor precursors events of major accidents, are the best tools to provide information about how a risk management process is conduced [4][5][6]. it can indicate if corrective actions are needed to achieve management goals at various levels, and, in the end, monitor the risk of major accidents. [7]. the experience of industry shows that without an effective regulatory oversight at oil and gas industrial activities, the risk of accidents will not be reduced, nor will industry prepare effectively to respond to major emergencies [8]. to be sure that all actions are going in a right direction, regulator’s can use safety indicators to follow safety performance and show deviations and the need to add corrective actions. as examples of this approach, the project trends in risk level (rnnp) [9] and the united kingdom health and safety executive (hse) [10] are nowadays in use. in this context, it becomes reasonable that in the regulatory level that exist tools for monitoring the platform’s performance and oil companies’ risk management to have a correct approach to set rules to regulated activities. all of it in sense to achieve sustainable industry’s growth ensuring a good level of safety in regulated activities and applying regulators efforts in a reasonable. journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 141-150 published by atlantis press copyright: the authors 141 willieb typewritten text received 20 april 2014 willieb typewritten text accepted 1 august 2014 the brazilian regulator offshore authority1 released in 2007 a non prescriptive regulation based on a safety management system and has been implementing actions to increase industry’s risk awareness and lead a safety continuous improvement environment of all offshore activities, as regular regulator’s audits, accident investigations, reports, meetings and workshops. this regulatory risk control regulation based on a process safety management system (psms) allows to achieve an indicators monitoring program for a new step on this development, giving information to improve the results evaluation of regulatory activities. 2. process safety management systems the psms have a variety of meanings and purposes. the aiche defines psms as a management system focused on prevention, preparedness, mitigation, response and recovery from catastrophic events associated with plant processes [3]. also according to the aiche, the psms is the discipline for managing the integrity of the operation of systems and processes that handle hazardous substances by applying good engineering practices, operation and maintenance [7]. it can be translated into the prevention and control of risks that have the potential for release of hazardous substances and energy. the organization of procedures, behaviors and actions to be taken and checked for systematic implementation of an effective process risk management is usually done by adoption of a management practices in a psms format 2. 2.1. performance indicators used in risk management the majority of indicators definitions tells that they are mainly used to information aquisition for management decision making related to a relevant issue [11]. for the word “performance”, it is indicated that it contains the idea of something that has already been done, executed or performed [12]. therefore, performance indicators have the function to transmit information about processes that occured in the past for adoption of actions to follow an objective. frequently, a retrospective analysis shows that early signs arising from safety indicators could avoid major 1 agência nacional de petróleo, gás natural e biocombustíveis (anp) 2 more information in sepeda, adrian l.; understanding process safety management, cep magazine, aug. 2010, aiche, 2010. accidents if detected and managed timely [13]. thus, safety performance indicators or just “safety indicators” are measures that seek to give information if a system is operating within chosen safety bounds to ensure that major accident risks are under control. anyway, they intend to indicate that the preventive and mitigative safeguards and assumptions considered for risk acceptance process are applicable, reliable and active. in the other hand can denounce irrelevance, unreliability or inactivity for an early detection of a non tolerable risk to demand pro-active actions to its reduction. 2.1.1. leading indicators according to hopkins [14], leading indicators are able to measure the ongoing risk management process and make predictions at sufficiently early stages to enable interruption of evolution course, reverse the process and avoid the fact. another definition describes leading indicators as a type of accident precursors and reports that this kind of indicators are conditions, events and actions that precede an undesirable events and which have value in predict the proximity of an accident, near miss incident or undesirable safety condition. in this sense, leading indicators are related to proactive activities that identify and evaluate hazards, eliminate, control, or mitigate the risk [15]. hw heinrich in 1931 introduced the widespread accident pyramid based on his experience with the insurance industry (fig. 1). it has two basic concepts: one is that accidents can be placed on a scale representing the degree of impact and the other is that many precursors incidents occurs with minor fig. 1. process safety indicator pyramid [15]. published by atlantis press copyright: the authors 142 consequences compared to accidents with large consequences. this representation establishes a predictive relationship between minor and major incidents [16]. large explosions and other disasters are rare events and many companies did not record any event or only few were recorded in the last 50 years at the same industry (top of the pyramid of accidents). thus, the number of occurrence of these events is not a good indicator of the probability of occurrence of major accidents [6]. 2.1.2. lagging indicators the lagging indicators generally reflect past experience and are associated with undesirable events and adoption of corrective actions to avoid the recurrence of such events. these indicators are often not sufficient for a preventive program, because the accidental events that generate this type of measure indicators are rare and do not allow a statistical prediction [17]. lagging indicators are generally viewed as direct measures of damages, however, may also be included incidents of unexpected losses of containment and failure on demand of safety critical systems. the fact is that not all damages and failures are recorded as lagging data, which complicates the design of a boundary between different types of indicators [18]. 2.2. global indicators the search for global indicators has been currently quite intensified as a way to manage different activities using the lowest possible number of information, or an indicator that integrates different information, allowing a responsible decision making to manage applied resources. in the chemical and petrochemical industries, the implementation of global indicators is already an established fact [16]. a global indicator represents a way to specify the evaluation of organizational performance, based on indicators combined and grouped for a determined focus. global indicators allow us to conclude about implementation effectiveness measurement of an aspect and can be a tool to correlate complex concepts and translate this in simple numerical information [19]. 3. methodology 3.1. identification of requirements to be considered in the formation of global safety indicators the platforms that are in operation in brazil are required to comply with a safety regulation established by anp regulation 43/2007 [20] (also called sgso). it demands the implementation of a set of good practices developed as a psms focused operational risk control. it relies on the adoption of seventeen management practices, showed in table 1, that are interrelated and are incorporated into all phases of a project from its conception until its decommissioning [20]. table 1. management practices (mp) required by sgso [19]. mp description mp 1 safety culture, commitment and responsibility management mp 2 personnel involvement mp 3 personal qualification, training and performance mp 4 working environment and human factors mp 5 selection, management and control of contractors mp 6 monitoring and continuous improvement performance mp 7 audits mp 8 management of information and documentation mp 9 incident investigation mp 10 project, construction, installation and deactivation mp 11 operational safety critical elements mp 12 risk identification and analysis mp 13 asset integrity mp 14 planning and management major emergencies mp 15 operational procedures mp 16 management of change mp17 work safety practices and control procedures for special activities 3.1.1. identified leading indicators checking guidelines and reports of safety indicators key programs already disclosed by regulatory authorities [21], it was observed that information taken from leading indicators contribute to act in a preventive way, to identify and correct non acceptable behaviors aiming to follow risk acceptance criteria in a regulatory point of published by atlantis press copyright: the authors 143 view prior to the occurrence of major accidents [6]. as sgso is a regulatory instrument of risk control and has a primarily preventive focus, leading indicators was settled based on information acquisition through specific parameters and sub groups as shown in table 2. table 2. subgroups and parameters to be monitored under leading focus. indicators related to leading focus subgroup parameters leadership, management and personnel involvement of personnel qualifications, training and personal performance monitoring and continuous performance improvement audits accident investigation systems and technology identification and risk analysis mechanical integrity and safety critical systems planning and management of major emergencies operational practices change management safe work practices beaconed in leading focus established subgroups and considering identified parameters based on sgso, indicators were identified for each subgroup. 3.1.2. identified lagging indicators accidental events classified as catastrophic, involving the death of many people, severe environmental and other damage to businesses rarely occur, not allowing to perform a statistical analysis. however, prior to these disasters, less damage incidents (precursor events) occur and are good indicators for monitor and prevent major accidents. thus, in this text, lagging indicators were grouped according to their kind and characteristics in order to support its proposition, from less to more severe events [22] [23]. the groups defined are shown in table 3 and are divided into operational deviations; loss of primary containment; fires and collisions. table 3. parameters identified for indicators related to the acquisition of lagging focus. subgroup related to lagging focus subgroup parameters operational deviations emergency shutdowns loss of primary contention oil spills without fire oil spills with fire major spills significant spills major natural gas leaks significant natural gas leaks h2s release fires major fire significant fire collisions major collision significant collision 3.2. the global indicator indicators related to leading and lagging focus the first step to create global indicators was the validation of chosen indicators by experts in the field of risk management in offshore activities. each subgroup was evaluated by 9 skilled professionals, with plenty of knowledge in sgso and risk management of offshore activities which evaluated the significance of the proposed indicators. this validation was made together to the prioritization evaluation, using criteria showed in tables 4 and 5 [24][25]. table 4. grades used in validation of the indicators. criteria note justifications not relevant 1,0 2,5 the proposed indicator has no relevance in the composition of a global indicator. low relevance 2,6 – 6 the proposed indicator has low relevance in the composition of a global indicator even though related to risk area average relevance 5,1 – 7,5 the proposed indicator is important in composing a global indicator related risk area high relevance 7,6 10 the proposed indicator is very relevant in the composition of a global indicator related to risk area table 5. weights used for prioritization of indicators. criteria wheight justifications hardly relevant 1,0 the indicator derives from subjective assessments or have an history of low frequency of occurrence relevant 3,0 the indicator stems from objective assessments with lower severity very relevant 5.0 the indicator runs from objective assessments with high severity for this purpose, it was established a prioritization matrix composed of columns and rows which was prepared for each subgroup related to both preventive and reactive focus. in columns were listed for each subgroup indicators, and for each one were assigned published by atlantis press copyright: the authors 144 grades and weights. this methodology adaptation was already tested in the environmental area for the global suitability indicators and proved to be satisfactory [25]. this method allows separate evaluations on criteria that point the relevance of each indicator through grades and use this value associated to given weights (weighted by multiplying), we have direct relationship with its importance in a general scenario, i.e., the higher these factors, the greater relative importance of such a specific indicator [26]. 3.2.1. normalization the normalization of the indicators used in each subgroup was conducted according to procedures established by the united nations (un) in the composition of the human development index (hdi) for the data comparison between different countries. it consists in adjust data within interval [0,1] and make it comparable to each other so that they can compose a global indicator, as showed in equation 1 [27] [28] [29]. , , , , , min( ) max( ) min( ) i j i j i j w w w i j i j w w x x x x x − = − (1) the meaning of variables is: i : platform j : company w : index of specific indicator: ji wx , specific indicator w ji wx , : normalized indicator derived from specific indicator xw )(min , jiwx : minimum value of the specific indicator w across all platforms )max( , jiwx : maximum value of the specific indicator w across all platforms 3.2.2 the global indicators the leading and lagging variables were incorporated, respectively, as global leading indicator (igprv) and global lagging indicator (igrtv) through a sum according to equation 2. , , 1 *w n i jx wi j n ig v x = = ∑ (2) the meaning of variables is: ig : global indicator for the company's j and platform i wxv :coefficient contribution percentage assigned to specific indicator within the prioritization performed for the respective segment; ji wx , : normalized indicator derived from specific indicator xw; n : number of specific indicators of the respective segment. 4. results and discussion the table 6 and 7 shows a summary of prioritization results obtained in leading and lagging indicators, showing the relative importance of specific indicators using the methodology already presented. the tables 6 and 7 show the relative importance of information raised by a specific indicator compared to others. each one has a weight that put it in a position that will impact on the final result of the respective global indicator and its position was obtained based on opinion of specialists that are the most used method for this kind of hierarchy process [29]. to analyze the results, real data from 2011 was used in 67 offshore production platforms and after normalization of specific indicators using equation 1, equation 2 was applied. the previous obtained values ( ji wx , ) have been multiplied by respective weighting values ( wxv ) to achieve both global indicators that add the contribution of each specific indicator to achieve one final number. the final results were divided into quartiles as shows figure 2. for igprv case, the greater value of the indicator, better is the platform safety practices compared to others. in a opposite way, as higher is the igrtv, worse is the results of the risk management for the specifc platform. as presented, the igprv and igrtv are showed respectivily at upper and lower part of figure 2 and allow the behavior identification of implementation effort to follow sgso requirements (igprv) and also give us information about the precursor events of major accidents (igrtv), based on the lowest and highest values of specific good safety practices indicators. as igrtv exhibits low values for a considerable number of platforms, it can deliver a false output that might indicates that no further actions are needed. however, if this kind of data is used alone in a safety analysis, it may lead actions that are not enough to avoid major accidents. this results may indicate that the cut lines established in the lagging approach may need a published by atlantis press copyright: the authors 145 further stratification to incorporate a greater number of related events or a more extended period of time [31]. in the other hand, it is not possible aim a good safety performance analysis looking just at leading indicators and forget the results showed by lagging indicators [6]. however, important management information can be raised from global indicators like showed in fig. 2. for instance, if a non desirable performance is identified, the weakness in risk management can be analyzed as showed in fig. 3. table 6. prioritization results of indicators in the leading focus. indicator results normalized value ( w xv ) ( %) compliance with recommendations of risk assessments 367,0 6,35 risk assessment in change management 324,0 5,61 esdvs tests 323,2 5,60 psvs tests 323,2 5,60 bdvs tests 322,0 5,58 fire detections system test 317,3 5,49 gas detections system test 317,0 5,49 risk assessment in the work permits 284,5 4,93 personal qualification 283,1 4,90 start up testing of fire pumps 280,6 4,86 deluge system test 273,3 4,73 implementation of audit corrective actions 263,1 4,56 implementation of incident investigations corrective actions 258,5 4,48 awareness and participation 248,8 4,31 risk analysis recommendations acceptance 367,0 3,90 temporary management of changes 219,0 3,79 accidental scenarios drills 215,0 3,72 level of investigation 214,0 3,71 goal attainment 212,7 3,68 deadline attendance for corrective actions of audits 180,6 3,13 evaluation level 172,5 2,99 work permits 150,0 2,60 total / normalized 5775,4 100 table 7. prioritization results of indicators in the lagging focus. indicator results normalized value ( w xv ) (%) h2s release 420,5 10,32 major natural gas leaks 419,0 10,29 major fire 419,0 10,29 oil spills with fire 406,0 9,97 major spills 405,0 9,94 significant fire 360,0 8,84 major collision 339,1 8,32 significant natural gas releases 282,5 6,94 significant spills 278,0 6,82 oil spills without fire 264,8 6,50 significant collision 245,5 6,03 plant shutdowns in emergency 234,0 5,74 total / normalized 4073,4 100 fig. 2. leading global indicator (igprv) and lagging global indicator (igrtv) to 67 production platforms. published by atlantis press copyright: the authors 146 fig. 3 shows that once a bad result is identified for a specif platform, the global information can be analysed in a more deep perpective, looking for the causes of this output. on the left its shows the subgroups of igprv results compared to a general average. after find the lower performance subgroup, the specific indicators can be analysed as showed on the right side of figure 3 and specific actions can be adressed to improve these results. furthermore, this kind of analyses can be expanded to an oil companies by adding the diferent results of their own platforms for a more amplied scope. 5. conclusions this study aimed to generally discuss the topic of process safety in offshore oil and gas production platforms in order to propose a model of global fig. 3. analyses of one specific result for igprv for a chosen platform. published by atlantis press copyright: the authors 147 indicators to turn easier to regulators to identify weaknesses in a process safety management based regulation. it can be useful to point possible improvement opportunities, optimize regulators resources finding inspection opportunities and enforce an operational safety continuous improvement environment. thus, from a leading point of view, it was found that knowing the real position of companies in meeting the need of implementing a regulatory psms, identifying key practices for the adoption of such a system and, therefore, measuring the implementation ongoing efforts has direct impact in avoiding major accidents. in other hand, measuring the occurrence of minor damage precursors events (near misses) and accidents has to be the focus to a lagging monitoring. the adequacy of the proposed indicators as instruments of information and evaluation came from numerous studies that incorporated different perceptions and different methods of analysis and safety monitoring, enabling the proposal of a mechanism for the safety assessment process [30]. in this context and within analyzed data, it was found that the proposed global indicators allow weaknesses identification at individual platforms level, which can be expanded to oil companies and platform operators, allowing regulatory and other management levels to act aiming to avoid undesirable behaviors and, in the end, avoid main situations that can lead a major accident. in this aspect, the indicators used and normalization procedure allow different levels of comparisons analysis and lead to compare platforms inside an oil company, platforms from different companies and even companies. it turns easier the establishment of performance goals and proactive regulatory oversight of the industry. it was found in the data analysis, confirming texts that performance monitoring programs based on purely leading or lagging data can lead weak or insufficient outputs, which do not allow a proper conclusion of the current status of safety of ongoing processes [6] [31]. on this way and to improve results analysis was found a need to improve lagging data analysis, maybe changing the data horizon, expanding the analysis period. other study opportunity is try to find the best fit to correlate both leading and lagging global indicators. as result of this research, considering the methodology procedure, it has allowed to measure the performance of global indicators in a given period of time, observing their variations in a large number of production platforms operating in brazil. in this same perspective, it could be seen that the use of global indicators proposed can address the proposal needs. however, a demand of this kind of method is the adoption of continuous improvement safety culture for companies which data are required and for regulator’s analysis mechanisms, in order to maintain the characteristics, sensitivity, robustness and simplicity in the conditions of measurement and analysis to reach desirable results [6] [29]. in this sense, it is expected that the evaluation of data from igprv and igrtv may be added with additional information from other processes such as information audits conducted by the regulator and causes of incidents correlated to psms, among others, could significantly improve the regulatory management process. it can lead a designing of companies risk management profiles, or other results outside the scope of this text, but preliminarily discussed in the data analysis. the presented indicators are designed to simplify analysis by quantifying subjective and complex concepts for establishing goals and to allow assess performance in regulated industry. as showed, it can be used to articulate concepts of process safety and regulatory action to allow identification and dissemination of best practices in risk management [5] [9] [10] [32]. this perception aims to improve resources destination in regulatory bodies improving actions to achieve the main focus of avoid major accidents. references [1] harsem, ø.; eide, a.; heen, k.; factors influencing future oil and gas prospects in artic, energy policy v.39, p. 8037-8045, 2011. [2] us energy information administration (eia), annual energy outlook – reference case, 2011. [3] american institute of chemical engineers (aiche), center for chemical process safety (ccps), guidelines for risk based process safety, wiley, 2007. [4] skogdalen, j.e.; utne, i.b.; vinnem, j.e. developing safety indicators for preventing offshore oil and gas deepwater drilling blowouts, safety science, n. 49, p. 1187-1199, 2011. published by atlantis press copyright: the authors 148 [5] united kingdom health and safety executive (hse), developing process safety indicators: a step-by-step guide for chemical and major hazard industries, 2006. [6] vinnem, j. e. risk indicators for major hazards on offshore installations. safety science. 2010. [7] american institute of chemical engineers (aiche), center for chemical process safety (ccps), guidelines for process safety metrics, wiley, 2010. [8] deepwater horizon incident joint investigation (dhjit). marine board of investigation into marine casualty, explosion, fire, pollution and sinking of mobile offshore drilling unit deepwater horizon, with the loss of life in the gulf of mexico, april 20-27, 2010. the us coast guard/bureau of ocean energy management, regulation and enforcement (boemre) joint investigation team, 2010. [9] petroleum safety authority norway (psa), summary report 2010 norwegian continental shelf trends in risk level in petroleum activity, 2011. [10] united kingdom health and safety executive (hse), key programme 3 report, asset integrity programme, hse's hazardous installations directorate. 2007. [11] bullen, p. performance indicators new management jargon, political marketing, or one small element in developing quality services? caring, the association of children’s welfare agencies newsletter (editorial comment), september edition, 1991. [12] kiyan, f.m.; proposal for development of performance indicators to support strategic, universidade de são paulo, 2001. [13] øien, k.; utne, i.b.; herrera, i.a.; building safety indicators: part 1 – theoretical foundation, safety science, n. 49, p. 148-161, 2010. [14] hopkins, a., the limits of lost injury frequency rates positive performance indicators for ohs part 1. proceedings. canberra: worksafe australia, 1994. [15] lethinen, e.; wahlstorom, b., safety performance measurement in process industries, third international conference on performance measurement and management, 2002. [16] international association of oil and gas producers (ogp), process safety – recommended practice on key performance indicator, report n. 456, 2011. [17] vinnem, j. e., aven, t., husebø et al., major hazard indicators for monitoring of trends in the norwegian offshore petroleum sector, reliability engineering & system safety,n. 91, p. 778-791, 2006. [18] seath, i.; lead vs. lag indicators, ian seath’s blog, http://ianjseath.wordpress.com/2009/08/26/lead-vs-lagindicators/, 2009, (accessed jan. 6, 2013). [19] levy, j.de q. beaumont, j. indicador global de ruído http://www.ecoservicos.pt/page.jsp?section =cientificos&page=pubs (accessed dez. 02, 2009) [20] brazilian national agency of oil, natural gas and biofuels (anp), anp resolution 43/2007, 2007. [21] almeida, a. c. c.; bardy, m. b.; risk assessment to the business due to accidents in process plants, det norske veritas (dnv), 2006. [22] giannetti, e.; almeida, c. industrial ecology – concepts, tools and applications. são paulo: ed. edgard blücher, 2006 apud sâo josé, andreia silva; identification of global indicators for social responsability and sustainabilityl, 2010. 125 f.,universidade do estado do rio de janeiro, 2010. [23] kraemer, m. e. p. tinoco, j. e. p. contabilidade e gestão ambiental. são paulo: ed. atlas, 2004. [24] freeman, h.m., hazardous waste minimization, mcgraw-hill publishing company. new york, usa 1990 [25] sâo josé, andreia silva; identification of global indicators for social responsability and sustainabilityl, 2010. 125 f.,universidade do estado do rio de janeiro, 2010. [26] matarazzo, dante c. análise financeira de balanços: abordagem básica e gerencial. 5ª ed. são paulo: atlas, 1995 apud sâo josé, andreia silva; identification of global indicators for social responsability and sustainabilityl, 2010. 125 f.,universidade do estado do rio de janeiro, 2010. [27] united nations (un), human devellopment report 2011, 2011. [28] organization for economic co-operation and development (oecd), handbook on constructing composite indicators – methodology and user guide, 2008. [29] organization for economic co-operation and development (oecd), handbook on constructing composite indicators – methodology and user guide, 2008 apud carvalho, j.a.b.; uma proposta de agupamento de indicadores para a avaliação da efetividade da segurança de usinas nucleares, dissertação de mestrado, ufrj – coppe, 2009. [30] almeida, a.g.; henrique , n. i.; lemos, t.s.m.; analysis from released data by offshore safety regulators: typical profiles and parameters for the establishment of a brazilian methodology, rio oil and published by atlantis press copyright: the authors 149 http://www.ecoservicos.pt/page.jsp?section=cientificos&page=pubs http://www.ecoservicos.pt/page.jsp?section=cientificos&page=pubs gas conference and expo, 2012. [31] international association of oil and gas producers (ogp), 2011.process safety recommended practice on key performance indicator, 2011. [32] united kingdom health and safety executive (hse), a guide to health & safety performance measuring, 2001. published by atlantis press copyright: the authors 150 1. introduction 2. process safety management systems 2.1. performance indicators used in risk management 2.1.1. leading indicators 2.1.2. lagging indicators 2.2. global indicators 3. methodology 3.1. identification of requirements to be considered in the formation of global safety indicators 3.1.1. identified leading indicators 3.1.2. identified lagging indicators 3.2. the global indicator indicators related to leading and lagging focus 3.2.1. normalization 3.2.2 the global indicators 4. results and discussion 5. conclusions references microsoft word introduction for special issue vol_2 no_4 introduction for volume 2, issue 4 the special issue on risk management in construction construction is one of the high-risk industries. constructed facilities are becoming more sophisticated and complex and risk management in construction is imperative. i would like to thank the journal of risk analysis and crisis response (jracr) for providing the platform for construction professionals and educators to interact and exchange research experience and results on risk management in construction. the special issue contains seven papers. the topics cover the development of new models, methods, and framework of risk management in construction; dispute analysis based on numerous data, and the impact of climate on property values. the first paper titled ‘analysis of disputes in transportation projects’ by m. emre bayraktar, cagri cinkilic and farrukh arif presents an analysis of the disputes for transportation projects and the related lessons learned. it was found that majority of disputes in the 262 cases studied were due to unforeseen conditions. the analysis is based on the data stored in the dispute resolution board (drb) database of the florida department of transportation (fdot). the second paper by payam bakhshi and ali touran proposes a new approach for contingency determination in a portfolio of construction projects. a bayesian approach is employed to modify a model on regular intervals that helps an agency find the level of confidence needed for individual projects to ensure that the portfolio budget will meet the minimum level of confidence based on available funding and the agency’s policy goals. the third paper contributed by dean kashiwagi and jacob kashiwagi presents a new risk management model designed by using deductive logic. the new model identifies risk, mitigates risk by use of transparency and dominant information which has been used by the expert contractor to minimize the risk they do not control. the fourth paper contributed by lingguang song, sang-hoon lee and fitria h. rachmat deals with a new method for scheduling repetitive construction operations with stochastic simulation to incorporate uncertainty in activity performance during field operations. the proposed stochastic linear scheduling method (slsm) has been implemented and demonstrated in a pipeline construction project. the fifth paper titled ‘risk management framework for the construction industry according to the iso 31000:2009 standard’ and contributed by vitor sousa, nuno marques de almeida and luís alves dias discusses the risk management framework and proposes a risk management framework for the construction industry. the sixth paper by babak a. samani and farzad shahbodaghlou emphasizes the need for a fuzzy systematic structural approach to the risk assessment of construction projects and introduces the processes required to form a hierarchical systematic structure based on fuzzy logic using the fuzzy decision making trial and evaluation laboratory. the last paper, contributed by huili hao, patrick long and scott curtis examines property owners’ attitudes regarding the impacts of climate on property ownership and future property values in coastal communities. special thanks go to dr. xianjun guan of the school of economics and management at tongji university, who had been of great assistance during calling for and reviewing manuscripts. again, i sincerely thank professor chongfu huang, the editor-in-chief, and editors of jracr for their strong support and assistance. thanks also go to all authors for their contributions, hard work and patience during the process of making the special issue become a reality. guest editor george c. wang, ph.d., p.eng. department of construction management east carolina university, greenville, north carolina wangg@ecu.edu journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 214 published by atlantis press copyright: the authors 214 atlantis press journal style editor’s introduction for volume 6, issue 3 while the real estate bubble in china is becoming more and more serious and the risk to runaway, this issue will be published. due to the lack of reliable data and the risk is highly sensitive, it is very difficult to analyze the risk of real estate bubble in china, and difficult to study how to response the crisis caused by the risk. until now, we have not received any contributions related to the topic. very regrettable! this issue contains 5 papers related to crisis threat assessment, financial risk, emergency management, fire risk analysis and marine disaster warning, respectively. among them, 4 contributions are written in english and 1 contribution in chinese with english abstract. the paper “identifying crisis threats: a partial synthesis of the literature on crisis threat assessment with relevance to public administrations” by christian kalbassi identifies crucial analytical approaches to crisis threat assessment with relevance to public administrations using existing literature regarding crisis definitions, crisis classifications, and crisis threat assessment tools. a critical evaluation of the specific research reveals gaps in the theoretical knowledge base and identifies three conceptual requirements for promoting a profound theorization of the field: the provision of relevant crisis threat variables, the specification of these variables, and the consolidation of the existing crisis threat assessment literature. the contribution of he et al., “connection parameters of heavy-tailed operational risk measurement model and management model”, puts forward a model to identify the crucial supervising parameters of operational risk after the heavy-tailed operational var’s sensitivity is theoretically researched by the elasticity analysis method. further, the analysis of model application is illustrated with a numerical example. the crucial supervising parameters connect the operational risk measurement model and management model, which make the operational risk management frameworks to be a complete system. and a dynamical supervising system of operational risk is established. this research in theory improves the application of loss distribution approach to the operational risk measurement and management. there is one paper in emergency management. the paper “route guidance map for emergency evacuation” by et al., re presents a route guidance map for pedestrians that aims an efficient evacuation in case of an emergency. an agent-based simulation framework is used for the simulation of various scenarios to prepare the guiding map. a real world case study of sarojini nagar, delhi is presented to test the presented methodology. eventually, several strategic recommendations are provided for improving safety of existing infrastructure. the paper “spatial inequality analysis of fire risk in china” by li et al. aims to introduce gini coefficient and lorenz curve, coefficient of variation, and spatial equilibrium index to explore the spatial inequality and spatial-temporal evolution of fires in china based on the fire statistics from 2003 to 2012. the spatial inequality of fire risk is influenced by population, provinces, regions, and fire causes. the results indicate that the spatial inequality of the fire deaths tends to become more intense over time, and shows spatial concentration. however, the number of fires becomes more decentralized distribution. the provinces with a high level of fire risk for different factors are depicted in maps by using arcgis, which can explicitly illustrate the spatial distribution characteristics of fire risk. “a case study of the severe convective in bohai sea and the establishment of early warning index on the sightseeing boat” by zhou et al. analyzes a severe convective weather process on august 31, 2015 in bohai sea area using the synchronous data such as weather radar, oceanic wrf model and et al.. in this paper, the authors also suggested an early warning model about the severe convective weather for the sightseeing boat in bohai sea. we sincerely hope our reader will find this issue’s information on risk analysis and crisis response interface useful. thanks to the referees for their strong support and kind help. and also thank the authors very much for all their outstanding contributions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 6, no. 3 (october 2016), 109 published by atlantis press copyright: the authors 109 microsoft word a_risk-benefit_analysis_model_for_project_investment_based_on_the_normal_distribution_1 a risk-benefit analysis model for project investment based on the normal distribution liwen chen school of economics and management,hebei university of technology, tianjin 300401, china email:lwchen@hebut.edu.cn abstract normal distribution formula is very important to analyze the reciprocity between the random event probability and the events. there exists certain kind of reciprocity between random variable that conforms to normal distribution and its parameter. solutions to apply normal distribution to the evaluation of continuous random variable, optimize the random variables by adopting optimization principles and methods were proposed in this paper. the purpose is to establish random variable standard deviation expected value function that conforms to normal distribution. the reciprocity among random variable, expected value and standard deviation, that is, the “mutual restraint principle of the three elements” were explored in this paper, in order to improve the accuracy and scientificity of decisions of random variable that conforms to normal distribution, avoid decision-making errors, and enhance the policy-making reliability. keywords: normal distribution, random variable, parameter, reciprocity, project investment 一个基于正态分布的项目投资风险收益分析模式 1 陈立文 河北工业大学经济管理学院，天津 300401 摘要：正态分布公式是分析随机事件概率与事件之间关系的一种重要公式，而符合正态分布的随机变 量与其参数存在着某种关系，本文提出如何应用正态分布于连续型随机变量的评价中，采用最优化的原理 与方法，进行随机变量的优化研究。目的是建立符合正态分布的随机变量标准差期望值函数，研究了随机 变量、期望值、标准差三要素相互关系，即“三要素相互制约准则”。进而为了提高符合正态分布的随机 变量决策的正确性和科学性，避免决策失误，提高决策可靠性。 关键词：正态分布，随机变量，参数，相互关系，项目投资 1. 引言 数学的发展来源于社会实践、生活实际，它是 人类实践经验的总结，它属于客观世界，并服务于 社会。 在概率及数理统计中，一般说来若影响随机变 量的因素很多，那么，每个因素单独引起的作用很 小，不能起到压倒一切的作用，甚至可以忽略不计， 各随机因素相互独立，它们的作用可以叠加，那么 这个随机变量服从正态分布。在自然现象和社会现 1 基金项目：国家自然科学基金(70872029)、河北省百名优秀创新人才支 持计划(sprc016)和河北省人才工程培养经费资助 作者简介：陈立文，男(1964-)，河北工业大学经济管理学院副院长、博士、 教授、博士生导师，研究方向：项目管理与风险控制、技术经 济与投资决策 象中，大量的随机变量都服从或近似地服从正态分 布。例如：人的身高、农作物收获量、测量某零件 长度的误差、海洋波浪的高度、电子管或半导体器 件中的热噪声电流或电压、空气分子运动速度等都 服从正态分布，因此，正态分布在概率统计中的基 本理论及其应用中占有特殊重要地位。本文研究就 是为了拓宽正态分布研究的领域 [1-3] 。 随着项目投资规模越来越大，无论从时间方面 还是从空间方面考虑，工程投资项目具有实施周期 长、不确定因素多、经济风险和技术风险大，对生 态环境的潜在影响严重，在国民经济和社会发展中 journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 146-154 published by atlantis press copyright: the authors 146 administrateur texte tapé à la machine received 17 december 2011 administrateur texte tapé à la machine accepted 6 march 2012 administrateur texte tapé à la machine liwen chen 占有重要的战略地位等特征。现代工程项目投资规 模大，投资大，影响深远，因而所面临的风险种类 繁多，各种风险之间的相互关系错综复杂，投资项 目从立项到完成后运行的整个周期中都必须重视风 险管理。 由于科技的飞速发展及其在社会生产各方面的 广泛应用，从而使各种风险因素及风险发生的可能 性大大增加，并且扩大了风险事件造成的损失，这 使各项目组不得不加强自身的风险管理水平，从而 使风险管理的各种手段应用于工程投资项目中。 项目投资经济效果评价指标的风险分析，指的 是如何依据各影响因素的概率分布来推求经济效果 评价指标的概率分布。有了这个概率分布也就有了 项目投资经济效果评价指标的大小及实现其大小的 可能性。一般来讲，风险表示一种可能性，衡量其 大小通常用可能出现风险的概率表示，由于概率的 大小反映了风险的大小，所以产生风险的关键性变 量可以由期望值和方差两个特征参数所反映的概率 分布表达式中确定出来。如果我们给出项目投资经 济效果评价指标与累积概率的关系曲线，也就有了 风险发生的全部信息，这样决策者也就可以很方便 地进行决策了，会做到胸中有数，措施得当，因此， 它为合理决策提供了依据 [4] 。 经济效果评价指标是多种多样的，它们从不同 角度反映了项目投资的经济性，这些指标主要可以 分成两大类：一类是以货币单位计量的价值型指标， 如净现值、净年值、费用现值、费用年值等；另一 类是反映资金利用效率的效率型指标，如投资收益 率、内部收益率、净现值指数等。 一个理智决策人对项目投资决策的意愿是选取 收益相对高，风险相对小的项目。也就是要求项目 投资的经济效果风险收益最好，那么，如何在给定 的条件下，即收益、风险一定的条件下，确定风险 收益条件下的最优经济效果，以及在最优经济效果 的条件下如何确定收益和风险。 本文提出如何应用正态分布于连续型随机变量 的评价中，采用最优化的原理与方法，进行项目投 资经济效果的随机变量优化研究。目的是建立符合 正态分布的项目投资经济效果随机变量标准差期望 值函数，研究了随机变量、期望值、标准差三要素 相互关系，即“三要素相互制约准则”。进而为了提 高符合正态分布的随机变量决策的正确性和科学 性，避免决策失误，提高决策可靠性。 2. 符合正态分布的随机变量标准差期望值函数 的建立 在项目不确定风险分析中，当某项目投资经济 效果随机变量期望值相对较低，如净现值(npv)，需 进一步了解项目经济效益发生在某一区间的可能性 有多大，则应计算这个区间内所有可能取值的概率 之和，即累积概率，用p(npv≥0)表示。 设符合正态分布的某项目投资经济效果随机变 量为 z ，实现其大小的累积概率为 )( zp ，其密度函 数为 )(zf ，分布函数为 )(zf ，则： 22 2)( 2 1 )(      z ezf (1)   z ttfzf d)()( (2) )(1 d)(1 d)()( zf ttf ttfzp z z        (3) 则实现该随机变量z 的期望值为： 2 2 2 2 22 2 2 )( 2 )( 22 2 1 2 2 d 2 d 2 d 2 d 2 1 d)( d)(d)( d)()( 22 2)(                                                                                 z z yzyz yz z z z z e z e z ye y ye ye y tet tttf tttftttf tttfzg t (4) 称 )(1 zg 为该随机变量z 的标准差期望值函 数。 1 ( )g z 有唯一最大值(证明见附录1)。 published by atlantis press copyright: the authors 147 a risk-benefit analysis model for project investment based on the normal distribution 则 最 低 实 现 该 随 机 变 量 z 的 期 望 值 为 ：  )(1 )( d)( d)()(2 zfz zzp ttfz ttzfzg z z         (5) 称 )(2 zg 为该随机变量 z 的最低标准差期望值 函数。 )(2 zg 也有唯一最大值(证明见附录2)。 由(4)式知：    22 )(1 g (6) (6)式说明随机变量 z 的标准差期望值函数 )(1 zg ，当z = μ 时， 1 ( )g  既与期望值 μ 有关系， 同时与标准差σ 也有关系。 由(5)式知： 2 d)( d)()(2            ttf ttzfg z (7) (7)式说明随机变量 z 的最低标准差期望值函 数 2 ( )g z ，当z = μ 时， 2 ( )g  只与期望值 μ 有关系。 比较(6)式与(7)式可得：    2 )(-)( 21 gg (8) 3. 随机变量、期望值、标准差三要素相互关系 研究 要满足最低标准差期望值函数条件下的随机变 量 z 最优时，需使： 0)(2  zg (9) 即：   z ttfzzf d)(-1)( (10) 将(1)式代入(10)式得： tee z t z z d 2 1 1 2 2 2 2 2 2 )( 2 )(            (11) 由(11)式可知，在最低标准差期望值函数条件 下，随机变量z 、期望值 μ 、标准差σ 三要素之间 存在着相互制约的关系，(11)式没有解析解，但可 采用计算机可以求出(11)的数值解，三者之间的关 系如图1所示。 图1 随机变量z 、期望值 μ 、标准差σ 三要素之间相互制约的关系图 在三者之间，只有两个要素保持独立或不固定， 最后一个要素必然是由已决定的两个要素中推出的 值，即它们之间存在着两个自由度。即三要素之间 存在着相互制约的关系，称上述关系为“三要素相 互制约准则”。由此可知，对于符合正态分布的随机 变量 z ： ①给定一组参数 μ 、σ 后，在满足随机变量最 低标准差期望值函数的条件下，必存在唯一的随机 变量z 的值。 published by atlantis press copyright: the authors 148 liwen chen ②要达到一定的随机变量 z ，在满足最低标准 差期望值函数的条件下，给定随机变量 z 的期望值 μ ，可唯一地确定标准差σ 的值，二者关系如图2 所示。 图2 随机变量z 与标准差σ 二要素之间相互制约的关系图 ③要达到一定的随机变量 z ，在满足最低标准 差期望值函数的条件下，给定随机变量 z 的标准差 σ ，可唯一地确定期望值 μ 的值，二者关系如图3 所示。 图3 随机变量z 与期望值 μ 二要素之间相互制约的关系图 当z =  时，(11)式就可简化成(12)式： 5.0 2    (12) 则离散系数为： 7979.0 2    (13) (13)式说明，当随机变量 z 取期望值 μ 标准差 时，单位期望值的标准差为一常数，即离散系数约 为0.7979。离散系数的大小说明了分布离散程度， 离散系数越大说明分布越离散，否则分布越集中。 这个离散系数对符合正态分布的随机变量标准差期 望值函数的决策提供了一个参考依据。这个系数对 published by atlantis press copyright: the authors 149 liwen chen 项目投资风险收益等问题研究具有非常重要的意 义。 4. 计算实例 已知某项目投资收益率 z 服从正态分布，其期 望值  =24.8%，标准差 =7.9%，现作下列优化及 测算，估计出该项目投资的风险性。计算可得如下 结果： ①绘制投资收益率z～累计概率 )(zp 的关系曲 线 投资收益率z ～累计概率 )(zp 关系为：                      z te ttfzp z z z 1 d 2 1 d)()( 2 2 2 )( 通过查标准正态分布表，可求得投资收益率z 、 累计概率 )(zp 、风险收益函数 )(1 zg ，最低风险收 益函数 )(2 zg 的关系列入表1中： 表1 )(zp 、 )(1 zg 及 )(2 zg 计算表 z (100%) 0 0.05 0.10 0.15 0.16 0.17 0.18 0.19 )(zp (100%) 0.999 0.993 0.970 0.892 0.866 0.839 0.805 0.767 )(1 zg 0.2490 0.2480 0.2460 0.2360 0.2320 0.2270 0.2210 0.2150 )(2 zg 0 0.0397 0.0970 0.1338 0.1386 0.1326 0.1339 0.1359 z (100%) 0.20 0.21 0.22 0.23 0.24 0.248 0.25 0.26 )(zp (100%) 0.728 0.685 0.638 0.590 0.530 0.500 0.390 0.330 )(1 zg 0.2070 0.1980 0.1880 0.1770 0.1650 0.1560 0.1530 0.1400 )(2 zg 0.1356 0.1339 0.1303 0.1357 0.1296 0.1230 0.1225 0.1133 z (100%) 0.27 0.28 0.29 0.30 0.31 0.35 0.50 1 )(zp (100%) 0.371 0.332 0.298 0.255 0.216 0.099 0.001 0 )(1 zg 0.1270 0.1140 0.1010 0.0890 0.0770 0.0380 0.0004 0 )(2 zg 0.1002 0.0958 0.0863 0.0765 0.0670 0.0337 0.0004 0 绘制z ～ )(zp 关系曲线，如图4所示。 图4 z ～ )(zp 的关系曲线图 published by atlantis press copyright: the authors 150 liwen chen 这条关系曲线不仅反映了项目投资经济效果的 大小，同时也反映了实现其大小的可能性，决策者 依据这条曲线做决策会胸中有数，措施得当，因此， 它为合理决策提供了依据。 ②建立风险期望收益函数 )(1 zg dtet dtttfzg t z z 2 2 079.02 )248.0( 1 079.02 1 )()(           ③建立风险最低期望收益函数 )(2 zg                                         tez tez ttfz zfzzg t z t z z d 079.02 1 1 d 2 1 1 d)(1 )(1)( 2 2 2 2 079.02 )248.0( 2 )( 2     ④求最优值 )(max),(max 21 zgzg 及最佳经济 效果 * 2 * 1 , zz 2485.0 079.0 248.0 248.0 2 079.0 248.0 2 )(max 2 2 2 2 079.02 248.0 2 1                              e ezg 0z1 1359.0)(max 2 19.2   zg 0z ⑤绘制风险期望收益函数 )(1 zg 和风险最低期 望收益函数 )(2 zg 的曲线 )(1 zg 的关系曲线类同示意图5， )(2 zg 的关系 曲线类同示意图6。 ⑥求投资收益率在 ],[ ba 区间(设[10%,15%])的 概率 由概率论理论得：                        ab ttfbzap b a d)()( 图5 )(1 zg 示意图 图6 )(2 zg 示意图 则 0768.0 )87.1()24.1( 079.0 248.010.0 079.0 248.015.0 d)(%)15%10( %15 %10                      ttfzp 这说明项目投资收益率在区间[10%,15%]的概 率为7.68%。 ⑦求投资收益率小于零的概率 0009.0 )14.3( 0 079.0 248.0 0 )0(                                zp published by atlantis press copyright: the authors 151 a risk-benefit analysis model for project investment based on the normal distribution 这说明该项目投资收益率小于零的概率约为 0.09%，即：该项目亏损的可能性只有0.09%，这是 一种小概率事件，因此，正常情况下可以认为该项 目是不会发生亏损的。 ⑧求投资收益率不小于某个值(设 20%)的概率 7291.0 079.0 248.020.0 1 20.0 )%20(                               zp 这说明该项目投资收益率达到或超过20%的概率 为72.91%。 ⑨如果项目有亏损，求平均的投资收益(亏损) 率 由条件期望得： 004.0 0009.0 079.0 2 1 248.0 2 1 d)()( d)( d)( )0( 2 2 2 2 079.02 248.0 2 0 0                                e e zp ttft ttf tttf zze              这说明，如果该项目可能出现各种亏损，那么 该项目投资平均亏损率为 0.4%。 ⑩如果项目有收益，求平均的投资收益率 由条件期望得： )0(1 d)( d)( )0( 0 0       zze ttf tttf zze 24823.0 0009.0 079.0 2 1 248.0 1 2 1 2 2 2 2 079.02 248.0 2               e e         这说明该项目若有收益，则该项目平均收益率 为24.823%。 5. 结论 本文采用最优化的原理与方法，对符合正态分 布的项目投资经济效果随机变量标准差期望值函数 进行了深入研究，其目的是为了提高符合正态分布 的项目投资经济效果随机变量决策的正确性和科学 性，避免决策失误，提高决策可靠性。本文的主要 工作有：(1)从理论上证明了所建的符合正态分布的 项目投资经济效果随机变量标准差期望值函数有唯 一最大值，从而为决策者提供了可靠的依据。(2) 提出了符合正态分布的项目投资经济效果随机变量 z 、期望值  、标准差 三要素之间相互制约的关 系，即决策三要素相互制约的准则。(3)项目投资经 济效果随机变量 z 的标准差期望值函数 )(1 zg 与最 低 标 准 差 期 望 值 函 数 )(2 zg ， 当 z 时 ，    2 )(-)( 21 gg 。(4)提出了符合正态分布的 项目投资经济效果随机变量取期望值时，其离散系 数为一定值，即约为0.7979。(5)通过某项目投资收 益率 z 服从正态分布，在已知其期望值和标准差情 况下估计出了该项目投资的风险性。 参考文献 [1] 浙江大学数学系高等数学教研组编．概率论与数理统 计[m]．人民教育出版社，1982.8 [2] 周誓达编著．概率论与数理统计[m]．中国人民大学出 版社，2005.6 [3] 陈梅荪．论正态分布曲线与太极图的联系性及物理意 义[j]．井冈山师范学院学报，2001，22(5)：4-5+7 [4] 陈立文著．项目投资风险分析理论与方法[m]．北京： 机械工业出版社，2004.9 published by atlantis press copyright: the authors 152 liwen chen 附录1： 定理1 满足(4)式函数 )(1 zg 有唯一最大值。 证明 为了证明(4)式函数 )(1 zg 有唯一最大 值，现在先来研究函数 )(1 zg 的极值，由(4)式知： 2 2 2 2 22 2 2 )( 2 )( 22 2 1 2 2 d 2 d 2 d 2 d 2 1 d)( d)(d)( d)()( 22 2)(                                                                                    z z yzyz yz z z z z e z e z ye y ye ye y tet tttf tttftttf tttfzg t (14) 为了判断函数 )(1 zg 的特性，先来求 )(1 zg 、 g z1 ( ) ： )()(1 zzfzg  (15) 由于 0)( zf ，所以说 )(1 zg 的正负性是由z 的 正负性决定的，即：          01 00 01 ))(sgn( 1 z z z zg (16) 此处 sgn( )x 为符号函数，即：          01 00 01 )sgn( x x x x (17) 这 说 明 函 数 g z1 ( ) 在 ( , ) 0 上 单 调 增 加 ， 在 ( , )0  上单调减少，在z  0处取得唯一最大值， 且为：   2 2 2 2 2 22 22 2 2 2 2 2 2 2 2 222 2 0 2 )( 0 11 2 2 2 d 22 d 2 d 2 1 d 2 1 d 2 1 d)( )0()(max 2                                                                                                                       e e e ye y e yedyey yey tte tttf gzg y yy yy y t z (18) 证毕。 published by atlantis press copyright: the authors 153 a risk-benefit analysis model for project investment based on the normal distribution 附录 2： 定理2 满足(5)式函数 )(2 zg 有唯一最大值。 证明 为了证明(5)式函数 g z2 ( ) 有唯一最大 值，现在先来研究函数 g z2 ( ) 的极值，为此先来求 )(2 zg 、 )(2 zg  ： )()(1 )()(1)(2 zzfdttf zzfzfzg z     (19)  222 )( 3 2 2 2 1 )()(2)( 2 2         zze zfzzfzg z (20) 令 22 2 2)(   zzz (21) 由于 0 2 1 2 2 2 )( 3       z e ，所以 )(2 zg  的正负性 是由2 ( )z 的正负性决定的，即：          0)(1 0)(0 0)(1 ))(sgn( 2 2 2 2 z z z zg    (22) 由于 0)(2 z 的两个根 2 8 22 2,1   z ， 又由于 )(2 zg 连续可微，则(22)式又可以写成下式：                       2 8 2 8 1 2 8 0 2 8 , 2 8 1 ))(sgn( 2222 22 2,1 2222 2    z z zz zg (23) 这说明：          1))(sgn( 0))(sgn( 1))(sgn( )( 2 2 2 2 zgfunctionconcave zgvaluepointinflection zgfunctionconvex zg (24) 又由于 0z 时， 0)(1  zf 0)( zf ， 0)( zzf ， 0)(  zzf ，则 0)(2  zg ，所以当 0z 时， )(2 zg 为 单 调 增 函 数 ， 也 就 是 说 )(2 zg 在           2 8 , 22  区间内有唯一最大值。 当 2 8 22   z 时， 0)(2  zg ，所以 g z2 ( ) 是凸函数，又由于：   0 1 2 lim 2 lim )(lim 1 d)(1 lim d)(1lim )(1lim)(lim 2 2 2 2 2 )( 2 )( 2 2 2                                       zz zz z z z z z zz e z e z zfz z ttf ttfz zfzzg (25) 因此，函数 g z2 ( ) 在           , 2 8 22  区间内单 调 递 减 ， 又 由 于 g z2 ( ) 连 续 可 微 ， 则 g z2 ( ) 在 ( , )  内 有 唯 一 最 大 值 ， 且 在          2 8 ,0 22  区间内取得。 证毕。 published by atlantis press copyright: the authors 154 rainstorm warning information in beijing: exploring the local perceptions and views fangping wang1,2, hanping zhao1,2*, weihua cao3, xiaoxue zhang1,2 1 key laboratory of environmental change and natural disaster, ministry of education, faculty of geographical science, beijing normal university, beijing 100875, china 2 academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education, faculty of geographical science, beijing 100875, china 3 the institute of urban meteorology, cma, beijing 100089, china received september 23, 2018 accepted november 11, 2018 abstract based on the public's cognition, evaluation and expectation of rainstorm warning information in beijing, the descriptive statistics and non-parametric test methods were used for data analysis. the results show that more than 80% of the public can recognize the importance of rainstorm warning, but only half of the public pay attention to the rainstorm warning information. as for the method of issuing early warning, the village broadcasting and electronic display, and network communication channels need to be strengthened. although the rainstorm warning information was released in time, the update process was not effectively communicated, and the information update period was long, it could not meet the public's demand for real-time attention to heavy rain. as far as the content of the warning information is concerned, false alarm effects of the rainstorm warning exist. according to the results, it is concluded that lead time can be extended by adding the update of warning information, and the probability of rainfall could be introduced to rainstorm warning information. these guide the design of rainstorm warning and promote understanding how the public responds to the warning. keywords: warning information, rainstorm warning, cry-wolf effect, beijing 北京市暴雨预警信息:探索地方认知和观点 王方萍 1,2，赵晗萍 1,2 ﹡ ，曹伟华 3，张晓雪 1,2 1. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875，中国 2. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875，中国 3. 中国气象局北京城市气象研究所，北京 100089，中国 摘要：基于北京市公众对暴雨预警信息发布方式和内容的认知、评价与期望，采用描述统计和非 参数检验方法进行数据分析。研究表明，80%以上公众能认识到暴雨预警的重要性，但只有一半 的公众每天关注暴雨预警信息。就预警发布方式而言，村镇广播和电子显示屏及网络传播渠道有 待加强；暴雨预警信息虽然发布及时但过程更新未能有效传达，且信息更新周期较长，不能满足 公众对暴雨实时关注的需求。就预警信息内容而言，存在暴雨预警虚警效应。根据结果得出可通 过预警信息更新将预警发布提前期延长。发布暴雨预警等级信号的同时，增加降雨概率的信息。 本研究为指导预警信息设计，促进理解公众对预警的响应提供思路。 ﹡corresponding author: 201531480035@mail.bnu.edu.cn 通讯作者简介：赵晗萍（1977-），女，汉，副教授，博士 研究方向：灾害风险评价、离散系统仿真、应急响应技术 资助项目：国家自然科学基金(41471424)；中央高校基本科研业务费专项资金资助 43 journal of risk analysis and crisis response vol. 9(1), march (2019), pp. 43–51 doi: https://doi.org/10.2991/jracr.b.190328.005; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 正确的风险认知和对预警的信任是预警 系统成功的基础[16]。mileti 和 o'brien（1992 年， 第 42 页）指出公众收到的预警信息如来源、 渠道、准确性、频率、一致性和清晰度是决定 其进行风险认知的关键因素[17]。由此可知，预 警信息发布渠道、更新频率、预警信息质量、 公众风险认知等在预警响应中发挥着重要作 用。通过了解公众如何回应预警对于确定应向 人群传播哪些信息，进而改善虚警效应和把握 合适的提前期至关重要，以增加公众对当局的 信任以及灾害发生时采取行动的能力[18]。本文 尝试通过实证研究，获取公众对暴雨预警信息 的评价及期望，进而提取改进暴雨预警信息传 播方式、信息内容和发布形式的方法，以期设 计个性化的预警信息，更贴合公众对预警的实 际需求。 2. 方法与材料 2.1 问卷设计 利用陈述偏好（sp）调查[19]方法分析北京 市突发地质灾害隐患点居民对暴雨预警的认 知、评价及期望。sp 调查方法也称“意向调 查”，指调查者根据假设情境，设计调查问卷， 获取被调查者自述信息中反映的个人偏好。sp 调查方法是获取公众风险认知和主观意愿的 有用工具，以便理解决策者对预警的响应行为 [19]。它也提供了从同一受访者收集多个回复的 机会。matyas 等[20]使用 sp 调查方法分析佛罗 里达游客在飓风威胁下风险认知的和疏散决 策。 为了评估暴雨预警信息发布效用，并获取 公众对暴雨预警信息的需求，使用家庭调查收 集数据。调查问卷的设计和修订基于预警响应 过程研究的文献综述和一系列小组讨论和预 调研。mileti 和 o'brien 指出风险认知是预警 信息和预警接收者个人特征的函数[17]。回顾预 警响应过程研究，以保证合理和科学地设计预 警认知因子。小组讨论在相关政府机构之间进 行。小组讨论的目的是评估当地的暴雨预警信 息内容涉及范围。随后进行了预调研，以进一 步辨别和纠正调查问卷的潜在问题。 问卷包括四个主要部分。第一部分讨论了 公众对灾害的风险认知及其灾害经历。受访者 被问及近年北京市最严重的灾害或环境问题 （1. 空气污染，2. 城市火灾，3. 城市内涝， 4. 暴雨，5. 山林火灾，6. 高温热害，7. 扬沙， 8. 干热风，9. 沙尘暴，10. 山崩，11. 雷电， 12. 雹灾，13. 低温冻害，14. 泥石流，15. 滑 坡，16. 冻雨，17. 龙卷风，18. 干旱，19. 积 雪，20. 霜冻）。暴雨会对自己生活的哪些方 面造成影响？（1. 交通出行，2. 房屋，3. 果 园，4. 生意，5. 庄稼，6. 人身安全，7. 无影 响）。还有是否遭受过暴雨损失（1. 是，2. 否）。 为了提升暴雨预警信息发布有效性，在第 二部分，要求受访者对暴雨预警信息的多方面 内容进行评价。mileti 和 o’brien 总结了公众 对传达的风险信息进行回应的预警信息内容 包括其来源、渠道、准确性、频率、清晰度和 一致性[17]。这里设计预警信息接收渠道、准确 性、信息更新、发布及时性和重要性五个预警 信息指标获取公众对暴雨预警的综合评价。首 先，受访者被问及平时得知暴雨预警信息的渠 道（1. 从未收到，2. 短信，3. 电视天气预报， 4. 网络、软件推送，5. 报纸，6.朋友、邻里转 告，7.电子显示屏，8.其他渠道）；其次受访者 需对以往暴雨预警的可信度进行打分（0-100 分）；出现一次暴雨预警不准，对下次的暴雨 预警的可信度打分（0-100 分）；出现两次暴 雨预警都不准，对下次的暴雨预警的可信度打 分（0-100 分）；出现一次暴雨预警准确，对 下次的暴雨预警的可信度打分（0-100 分）； 出现两次暴雨预警都准确，对下次的暴雨预警 的可信度打分（0-100 分）。然后受访者被问 及在一次暴雨过程中是否收到过暴雨预警信 息的更新（1.每次都能收到，2.大多数能收到， 3.收到一两次，4.没收到过，5.没注意）。收到 的预警信息是否及时（1.非常及时，2.及时，3. 不一定，4.不及时，5.非常不及时），暴雨预警 作为提前提醒进行暴雨防范有多重要（1.非常 重要，2.比较重要，3.一般，4.不重要）。 为进一步加强公众对预警的信任度以及 灾害应对能力，第三部分旨在检查公众对目前 暴雨预警发布的期望。对应公众对暴雨预警信 息的评价，这里询问公众预期在一次暴雨过程 中预警信息的更新频率（1. 半小时/次，2. 一 小时/次 3. 两小时/次，4. 三小时/次，5. 四小 时/次，6. 五小时/次，7. 六小时/次，8. 七小 时/次，9. 八小时/次，11.九小时/次，11. 十小 时/次，12. 十一小时/次，13. 十二小时/次，14. 十二小时以上/次，15. 无所谓）；暴雨预警发 布提前期（1. 半小时以内，2. 一小时以内，3. 两小时以内，4. 三小时以内，5. 四小时以内， 6. 五小时以内，7. 六小时以内，8. 六到十二 小时，9. 十二小时以上）；与降低虚警水平相 比，天气灾害预测中增加预测概率可以提高响 应率和决策质量[11]。这里设计公众期望接收到 的暴雨预警信息形式（1.降雨量的预报，2.降 雨概率的预报，3.降雨量和降雨概率预报都希 望有） 第四部分检查公众对四级暴雨预警的响 应情况。暴雨预警分为蓝色预警、黄色预警、 橙色预警和红色预警四个等级。受访者被问及 发布什么等级的暴雨预警会开始响应（1. 蓝 色预警，2. 黄色预警，3. 橙色预警，4. 红色 关键词：预警信息；暴雨预警；虚警效应；北京市 1. 引言 尽管极端天气事件预测有所改善，但公众 不响应天气预警仍然是一个问题。虽然可能有 很多原因导致公众不响应天气预警，但一个重 要原因可能是预警发布方式。如北京市 2012 年“七•二一”特大暴雨事件，气象部门于 7 月 21 日 09 时 30 分发布暴雨蓝色预警，14 时 升级为暴雨黄色预警，18 时 30 分升级为暴雨 橙色预警，22 时发布继续保持暴雨橙色预警， 并 22 日 01 时降级为暴雨蓝色预警，3 时 50 分 解除暴雨蓝色预警。预警信息平均每四小时更 新一次，尽管这样，预警并未被公众和相关防 灾部门接受并重视，没有及时采取有效的应对 措施，2012 年 7 月 21 日至 22 日两日内造成 经济损失 116.4 亿元，至少 79 人死亡。预警发 布渠道、预警信息更新频率或确定性预警发布 形式等都可能是导致此次事件的原因。有研究 [1]表明，对夜间低温的概率预测会产生比确定 性预测更好的决策，因为它可以让人们更好地 区分需要采取预防措施的情况。且当单值预测 误差增加时，这种概率预测的优势增加，表明 承认预测中的不确定性可以抵消预测误差的 负面影响。 对基于预测的天气预警，另外可能影响公 众对天气预警的响应的因素是预警发布提前 期和预警信息质量。预警系统减少危险后果的 前提是可以在合适的提前期内采取降低风险 的措施。先前的研究已经反复证明，预警系统 过于敏感，会产生过多的误报，错误警报会降 低公众对后续预警的信任度从而降低对预警 的响应[2-7]，这种现象被称为“cry wolf”效应 [8]。如 2013 年 7 月 15 日，北京市气象台 00 时 40 分发布暴雨黄色预警。有了“七•二一”特 大暴雨的经历，各部门对此投入了大量的物力 人力进行应对，但这次降雨并未形成暴雨过程。 从新浪微博的公众评论可知（如图 1），公众却 因此次错误预警对预警信息的可靠性产生了 怀疑。相反，如果它不够敏感，它可能无法在 事件即将发生时发出信号，或者它可能会在非 常短的提前期内发出警告，以至于不能采取保 护措施来减少损失[9-10]。 由于预测不确定性最初得到承认，因此在 预测中尽管存在错误预警，只要包含不确定性 估计就会保持预警的可信度。因此，不确定性 信息也可以保持存在错误预警时的可靠性[11]。 此外，不确定性信息可以使用户更有效的理解 所面临风险[12-14]。事实上，日常用户可能会设 想确定性预测和高的预警错误率的范围值，当 出现极端预测时他们会认为预测结果是夸张 的[12,15]。由于这些直觉，如果没有提供有效的 不确定性估计，人们实际上可能会低估某些情 况下的风险。而且也可能会将排除不确定性估 计的预警视为不完整和不可靠。这些因素组合 可能会影响公众对天气预警的响应。 图 1 2013 年 7 月 15 日北京市暴雨预警公众评论 figure 1 public comment on rainstorm warning in beijing on july 15, 2013 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 44 正确的风险认知和对预警的信任是预警 系统成功的基础[16]。mileti 和 o'brien（1992 年， 第 42 页）指出公众收到的预警信息如来源、 渠道、准确性、频率、一致性和清晰度是决定 其进行风险认知的关键因素[17]。由此可知，预 警信息发布渠道、更新频率、预警信息质量、 公众风险认知等在预警响应中发挥着重要作 用。通过了解公众如何回应预警对于确定应向 人群传播哪些信息，进而改善虚警效应和把握 合适的提前期至关重要，以增加公众对当局的 信任以及灾害发生时采取行动的能力[18]。本文 尝试通过实证研究，获取公众对暴雨预警信息 的评价及期望，进而提取改进暴雨预警信息传 播方式、信息内容和发布形式的方法，以期设 计个性化的预警信息，更贴合公众对预警的实 际需求。 2. 方法与材料 2.1 问卷设计 利用陈述偏好（sp）调查[19]方法分析北京 市突发地质灾害隐患点居民对暴雨预警的认 知、评价及期望。sp 调查方法也称“意向调 查”，指调查者根据假设情境，设计调查问卷， 获取被调查者自述信息中反映的个人偏好。sp 调查方法是获取公众风险认知和主观意愿的 有用工具，以便理解决策者对预警的响应行为 [19]。它也提供了从同一受访者收集多个回复的 机会。matyas 等[20]使用 sp 调查方法分析佛罗 里达游客在飓风威胁下风险认知的和疏散决 策。 为了评估暴雨预警信息发布效用，并获取 公众对暴雨预警信息的需求，使用家庭调查收 集数据。调查问卷的设计和修订基于预警响应 过程研究的文献综述和一系列小组讨论和预 调研。mileti 和 o'brien 指出风险认知是预警 信息和预警接收者个人特征的函数[17]。回顾预 警响应过程研究，以保证合理和科学地设计预 警认知因子。小组讨论在相关政府机构之间进 行。小组讨论的目的是评估当地的暴雨预警信 息内容涉及范围。随后进行了预调研，以进一 步辨别和纠正调查问卷的潜在问题。 问卷包括四个主要部分。第一部分讨论了 公众对灾害的风险认知及其灾害经历。受访者 被问及近年北京市最严重的灾害或环境问题 （1. 空气污染，2. 城市火灾，3. 城市内涝， 4. 暴雨，5. 山林火灾，6. 高温热害，7. 扬沙， 8. 干热风，9. 沙尘暴，10. 山崩，11. 雷电， 12. 雹灾，13. 低温冻害，14. 泥石流，15. 滑 坡，16. 冻雨，17. 龙卷风，18. 干旱，19. 积 雪，20. 霜冻）。暴雨会对自己生活的哪些方 面造成影响？（1. 交通出行，2. 房屋，3. 果 园，4. 生意，5. 庄稼，6. 人身安全，7. 无影 响）。还有是否遭受过暴雨损失（1. 是，2. 否）。 为了提升暴雨预警信息发布有效性，在第 二部分，要求受访者对暴雨预警信息的多方面 内容进行评价。mileti 和 o’brien 总结了公众 对传达的风险信息进行回应的预警信息内容 包括其来源、渠道、准确性、频率、清晰度和 一致性[17]。这里设计预警信息接收渠道、准确 性、信息更新、发布及时性和重要性五个预警 信息指标获取公众对暴雨预警的综合评价。首 先，受访者被问及平时得知暴雨预警信息的渠 道（1. 从未收到，2. 短信，3. 电视天气预报， 4. 网络、软件推送，5. 报纸，6.朋友、邻里转 告，7.电子显示屏，8.其他渠道）；其次受访者 需对以往暴雨预警的可信度进行打分（0-100 分）；出现一次暴雨预警不准，对下次的暴雨 预警的可信度打分（0-100 分）；出现两次暴 雨预警都不准，对下次的暴雨预警的可信度打 分（0-100 分）；出现一次暴雨预警准确，对 下次的暴雨预警的可信度打分（0-100 分）； 出现两次暴雨预警都准确，对下次的暴雨预警 的可信度打分（0-100 分）。然后受访者被问 及在一次暴雨过程中是否收到过暴雨预警信 息的更新（1.每次都能收到，2.大多数能收到， 3.收到一两次，4.没收到过，5.没注意）。收到 的预警信息是否及时（1.非常及时，2.及时，3. 不一定，4.不及时，5.非常不及时），暴雨预警 作为提前提醒进行暴雨防范有多重要（1.非常 重要，2.比较重要，3.一般，4.不重要）。 为进一步加强公众对预警的信任度以及 灾害应对能力，第三部分旨在检查公众对目前 暴雨预警发布的期望。对应公众对暴雨预警信 息的评价，这里询问公众预期在一次暴雨过程 中预警信息的更新频率（1. 半小时/次，2. 一 小时/次 3. 两小时/次，4. 三小时/次，5. 四小 时/次，6. 五小时/次，7. 六小时/次，8. 七小 时/次，9. 八小时/次，11.九小时/次，11. 十小 时/次，12. 十一小时/次，13. 十二小时/次，14. 十二小时以上/次，15. 无所谓）；暴雨预警发 布提前期（1. 半小时以内，2. 一小时以内，3. 两小时以内，4. 三小时以内，5. 四小时以内， 6. 五小时以内，7. 六小时以内，8. 六到十二 小时，9. 十二小时以上）；与降低虚警水平相 比，天气灾害预测中增加预测概率可以提高响 应率和决策质量[11]。这里设计公众期望接收到 的暴雨预警信息形式（1.降雨量的预报，2.降 雨概率的预报，3.降雨量和降雨概率预报都希 望有） 第四部分检查公众对四级暴雨预警的响 应情况。暴雨预警分为蓝色预警、黄色预警、 橙色预警和红色预警四个等级。受访者被问及 发布什么等级的暴雨预警会开始响应（1. 蓝 色预警，2. 黄色预警，3. 橙色预警，4. 红色 关键词：预警信息；暴雨预警；虚警效应；北京市 1. 引言 尽管极端天气事件预测有所改善，但公众 不响应天气预警仍然是一个问题。虽然可能有 很多原因导致公众不响应天气预警，但一个重 要原因可能是预警发布方式。如北京市 2012 年“七•二一”特大暴雨事件，气象部门于 7 月 21 日 09 时 30 分发布暴雨蓝色预警，14 时 升级为暴雨黄色预警，18 时 30 分升级为暴雨 橙色预警，22 时发布继续保持暴雨橙色预警， 并 22 日 01 时降级为暴雨蓝色预警，3 时 50 分 解除暴雨蓝色预警。预警信息平均每四小时更 新一次，尽管这样，预警并未被公众和相关防 灾部门接受并重视，没有及时采取有效的应对 措施，2012 年 7 月 21 日至 22 日两日内造成 经济损失 116.4 亿元，至少 79 人死亡。预警发 布渠道、预警信息更新频率或确定性预警发布 形式等都可能是导致此次事件的原因。有研究 [1]表明，对夜间低温的概率预测会产生比确定 性预测更好的决策，因为它可以让人们更好地 区分需要采取预防措施的情况。且当单值预测 误差增加时，这种概率预测的优势增加，表明 承认预测中的不确定性可以抵消预测误差的 负面影响。 对基于预测的天气预警，另外可能影响公 众对天气预警的响应的因素是预警发布提前 期和预警信息质量。预警系统减少危险后果的 前提是可以在合适的提前期内采取降低风险 的措施。先前的研究已经反复证明，预警系统 过于敏感，会产生过多的误报，错误警报会降 低公众对后续预警的信任度从而降低对预警 的响应[2-7]，这种现象被称为“cry wolf”效应 [8]。如 2013 年 7 月 15 日，北京市气象台 00 时 40 分发布暴雨黄色预警。有了“七•二一”特 大暴雨的经历，各部门对此投入了大量的物力 人力进行应对，但这次降雨并未形成暴雨过程。 从新浪微博的公众评论可知（如图 1），公众却 因此次错误预警对预警信息的可靠性产生了 怀疑。相反，如果它不够敏感，它可能无法在 事件即将发生时发出信号，或者它可能会在非 常短的提前期内发出警告，以至于不能采取保 护措施来减少损失[9-10]。 由于预测不确定性最初得到承认，因此在 预测中尽管存在错误预警，只要包含不确定性 估计就会保持预警的可信度。因此，不确定性 信息也可以保持存在错误预警时的可靠性[11]。 此外，不确定性信息可以使用户更有效的理解 所面临风险[12-14]。事实上，日常用户可能会设 想确定性预测和高的预警错误率的范围值，当 出现极端预测时他们会认为预测结果是夸张 的[12,15]。由于这些直觉，如果没有提供有效的 不确定性估计，人们实际上可能会低估某些情 况下的风险。而且也可能会将排除不确定性估 计的预警视为不完整和不可靠。这些因素组合 可能会影响公众对天气预警的响应。 图 1 2013 年 7 月 15 日北京市暴雨预警公众评论 figure 1 public comment on rainstorm warning in beijing on july 15, 2013 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 44 45 是电视天气预报。短信方式接收预警信息是当 局层面的发布渠道，而只有不到一半的人是通 过短信获取暴雨预警信息，表明手机短信覆盖 率欠缺。村委会是通过广播和显示屏传达预警 信息的直接官方渠道，但村里广播和电子显示 屏的使用率较低，所以应加强村委会对暴雨预 警信息传播的重视。在当今发达的网络时代， 预警信息应重视网络渠道的利用，一方面从普 及网络的使用方面促进村民了解预警信息，一 方面通过网络信息传播，便于村民了解实时的 降雨情况。 表 1 预警接收渠道 table 1 the major source of information of rainstorm warnings 预警信息接收渠道 比例（%） 电视天气预报 67.6 短信 43.3 自己感觉 35.2 村里广播 34.7 网络、软件推送 22.4 朋友、邻里转告 14.6 报纸 9.7 电子显示屏 2.7 从未收到 1.8 暴雨预警信息发布及时与否，一次暴雨过程中 是否有实时的预警信息更新，对公众认知风险， 乃至采取防护措施至关重要。由图 4 可知，仍 有 29%的人认为暴雨预警发布的不及时。图 5 表明有一部分人并未能收到暴雨预警信息的 更新。反映了一次暴雨过程中，预警信息更新 的传播与内容可能存在不足。 图 4 暴雨预警及时性 figure 4 timeliness of rainstorm warnings 图 5 暴雨预警信息更新接收情况 figure 5 reception of rainstorm warning information update （2）暴雨预警信息内容 这里基于预警信息质量（准与不确）和两 个时间维度（上次和前两次）定义虚警效应。 图 6 和表 2 展示了五种虚警效应定义下公众 对预警信任度的统计量和分布情况。结果显示， 23% 48% 19% 8% 2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 14% 22% 14% 40% 10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 图 3 各种社会问题严重性排序分析 figure 3 an sorting analysis of the severity of various social problems 空 气 污 染 暴 雨 城 市 内 涝 高 温 热 害 沙 尘 暴 扬 尘 雷 电 城 市 火 灾 雹 灾 干 旱 干 热 风 积 雪 滑 坡 泥 石 流 霜 冻 持 续 低 温 山 林 火 灾 龙 卷 风 山 崩 0 10 20 30 40 50 60 70 80 90 100 预警）。 2.2 抽样方案 为了研究公众对暴雨预警信息的认知和 反馈，本研究小组于 2017 年在北京市郊区进 行实地调研，暴雨预警自主响应研究区见图 2。 为了获得较为全面的资料，本次调研联系了当 地的气象部门，并获得他们的积极配合。根据 北京市突发地质灾害隐患点资料，以家庭为单 位，每个区随机选取村镇进行入户一对一面对 面调研。在本次调查之前已在北京居住超过一 年的居民被选中参加，这些人满足暴雨经历的 条件。共有 2000 名受访者被选中。北京师范 大学地理科学学部 20 名研究生和本科生接受 了有效进行面对面访谈的培训。所有这些人都 可以随时回答数据收集过程中出现的任何问 题。所有受访者都被告知该研究的目标和资金 组织，并且在调查之前从受访者那里获得了知 情同意。参与者获得匿名保证。总共回收 1539 份，回收率为 77%。 2.3 统计方法 描述统计方法用于分析数据的统计特征。 卡方检验用于名义数据，以检验四个等级暴雨 预警下不同预警信任组中公众预警响应的差 异。显著性水平设定为 p<0.1。所有统计分析 在 ibm spss statistics 23.0 软件中实现。 3. 结果分析 3.1 公众对暴雨的风险认知和受损经历 列举北京市主要灾害：空气污染、暴雨、 城市内涝、高温、沙尘暴、扬尘、雷电、城市 火灾、雹灾、干旱、干热风、暴雪、滑坡、泥 石流、霜冻、持续低温、山林火灾、山崩、龙 卷风，选择影响大且希望发布预警的灾害，结 果显示公众对空气污染预警的期望最高，其次 是暴雨预警（如图 3）。 对空气污染预警期望最高，与北京市近年 来严重的空气污染状况有关。对暴雨预警的期 望，与 2012 年“7.21”北京暴雨造成的严重 损失有关。此结果与北京市突发事件预警信息 发布中心统计结果基本一致。北京市郊区有 41% 的人遭受过暴雨的损失。其中受暴雨影响最频 繁最严重的是交通出行、房屋、庄稼、果园。 3.2 暴雨预警信息评价 （1）暴雨预警信息发布方式 评价预警的前提是关注预警及预警对公众的 重要性。结果表明有 54%的人每天都关注暴雨 预警信息，34%的人有时关注，12%的人从不 关注暴雨预警。88%的人肯定暴雨预警信息的 必要性。暴雨预警信息主要的接收渠道如下表 1。可见郊区暴雨预警信息接收最普遍的渠道 图 2 北京市暴雨预警公众响应调研地分布图 figure 2 location of survey for rainstorm warning response journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 46 是电视天气预报。短信方式接收预警信息是当 局层面的发布渠道，而只有不到一半的人是通 过短信获取暴雨预警信息，表明手机短信覆盖 率欠缺。村委会是通过广播和显示屏传达预警 信息的直接官方渠道，但村里广播和电子显示 屏的使用率较低，所以应加强村委会对暴雨预 警信息传播的重视。在当今发达的网络时代， 预警信息应重视网络渠道的利用，一方面从普 及网络的使用方面促进村民了解预警信息，一 方面通过网络信息传播，便于村民了解实时的 降雨情况。 表 1 预警接收渠道 table 1 the major source of information of rainstorm warnings 预警信息接收渠道 比例（%） 电视天气预报 67.6 短信 43.3 自己感觉 35.2 村里广播 34.7 网络、软件推送 22.4 朋友、邻里转告 14.6 报纸 9.7 电子显示屏 2.7 从未收到 1.8 暴雨预警信息发布及时与否，一次暴雨过程中 是否有实时的预警信息更新，对公众认知风险， 乃至采取防护措施至关重要。由图 4 可知，仍 有 29%的人认为暴雨预警发布的不及时。图 5 表明有一部分人并未能收到暴雨预警信息的 更新。反映了一次暴雨过程中，预警信息更新 的传播与内容可能存在不足。 图 4 暴雨预警及时性 figure 4 timeliness of rainstorm warnings 图 5 暴雨预警信息更新接收情况 figure 5 reception of rainstorm warning information update （2）暴雨预警信息内容 这里基于预警信息质量（准与不确）和两 个时间维度（上次和前两次）定义虚警效应。 图 6 和表 2 展示了五种虚警效应定义下公众 对预警信任度的统计量和分布情况。结果显示， 23% 48% 19% 8% 2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 14% 22% 14% 40% 10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 图 3 各种社会问题严重性排序分析 figure 3 an sorting analysis of the severity of various social problems 空 气 污 染 暴 雨 城 市 内 涝 高 温 热 害 沙 尘 暴 扬 尘 雷 电 城 市 火 灾 雹 灾 干 旱 干 热 风 积 雪 滑 坡 泥 石 流 霜 冻 持 续 低 温 山 林 火 灾 龙 卷 风 山 崩 0 10 20 30 40 50 60 70 80 90 100 预警）。 2.2 抽样方案 为了研究公众对暴雨预警信息的认知和 反馈，本研究小组于 2017 年在北京市郊区进 行实地调研，暴雨预警自主响应研究区见图 2。 为了获得较为全面的资料，本次调研联系了当 地的气象部门，并获得他们的积极配合。根据 北京市突发地质灾害隐患点资料，以家庭为单 位，每个区随机选取村镇进行入户一对一面对 面调研。在本次调查之前已在北京居住超过一 年的居民被选中参加，这些人满足暴雨经历的 条件。共有 2000 名受访者被选中。北京师范 大学地理科学学部 20 名研究生和本科生接受 了有效进行面对面访谈的培训。所有这些人都 可以随时回答数据收集过程中出现的任何问 题。所有受访者都被告知该研究的目标和资金 组织，并且在调查之前从受访者那里获得了知 情同意。参与者获得匿名保证。总共回收 1539 份，回收率为 77%。 2.3 统计方法 描述统计方法用于分析数据的统计特征。 卡方检验用于名义数据，以检验四个等级暴雨 预警下不同预警信任组中公众预警响应的差 异。显著性水平设定为 p<0.1。所有统计分析 在 ibm spss statistics 23.0 软件中实现。 3. 结果分析 3.1 公众对暴雨的风险认知和受损经历 列举北京市主要灾害：空气污染、暴雨、 城市内涝、高温、沙尘暴、扬尘、雷电、城市 火灾、雹灾、干旱、干热风、暴雪、滑坡、泥 石流、霜冻、持续低温、山林火灾、山崩、龙 卷风，选择影响大且希望发布预警的灾害，结 果显示公众对空气污染预警的期望最高，其次 是暴雨预警（如图 3）。 对空气污染预警期望最高，与北京市近年 来严重的空气污染状况有关。对暴雨预警的期 望，与 2012 年“7.21”北京暴雨造成的严重 损失有关。此结果与北京市突发事件预警信息 发布中心统计结果基本一致。北京市郊区有 41% 的人遭受过暴雨的损失。其中受暴雨影响最频 繁最严重的是交通出行、房屋、庄稼、果园。 3.2 暴雨预警信息评价 （1）暴雨预警信息发布方式 评价预警的前提是关注预警及预警对公众的 重要性。结果表明有 54%的人每天都关注暴雨 预警信息，34%的人有时关注，12%的人从不 关注暴雨预警。88%的人肯定暴雨预警信息的 必要性。暴雨预警信息主要的接收渠道如下表 1。可见郊区暴雨预警信息接收最普遍的渠道 图 2 北京市暴雨预警公众响应调研地分布图 figure 2 location of survey for rainstorm warning response journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 46 47 表 4 一次预警不准下虚警效应检验.表中数字表示响应比例（%） table 4 cry wolf effect for one wrong warning test. the numbers are the percentage of people response to warnings in each level. 预警可信度分组 蓝色 预警 卡方 （p 值） 黄色 预警 卡方 （p 值） 橙色 预警 卡方 （p 值） 红色 预警 卡方 （p 值） 预警信任降低组 30.3 12.903 （0.002） 54.5 4.555 （0.103） 72.7 0.523 （0.77） 86.8 2.164 （0.339 ） 预警信任不变组 46.5 62 73.9 89.7 预警信任加强组 35.8 54.8 75.5 90.9 表 5 两次预警不准下虚警效应检验.表中数字表示响应比例（%） table 5 cry wolf effect for two wrong warnings test. the numbers are the percentage of people response to warnings in each level. 预警可信度分组 蓝色 预警 卡方 （p 值） 黄色 预警 卡方 （p 值） 橙色 预警 卡方 （p 值） 红色 预警 卡方 （p 值） 预警信任降低组 26.5 15.325 （0.00） 54.8 5.296 （0.071 ） 72.9 0.611 （0.737 ） 85.5 5.946 （0.051 ） 预警信任不变组 48.2 63.1 75.4 89.7 预警信任加强组 35.8 55.1 75.5 95.9 虽然各组响应比例无显著差异（一次预警不准 检 验 ： 2=4.555, p=0.103; 2=0.523, p=0.77; 2=2.164, p=0.339），（两次预警不准检验： 2=5.296，p=0.071；2=0.611，p=0.737；2=5.946， p=0.051）。结果表明，在四级预警等级下，相 对于预警不变组和预警信任加强组，因错误预 警造成的预警信任度降低组的人群对预警的 响应最低，说明虚警效应在引起公众对预警的 信任度降低时，也进一步影响公众对预警的响 应。 3.3 公众对暴雨预警信息的期望 针对暴雨不及时的评价，其中 50%以上的 公众希望提前至少六小时发布暴雨预警（图 7）。 而预警发布提前期越长，预警信息质量越差[7]， 在预期提前至少六小时发布暴雨预警的同时， 公众期望可以在半小时、一小时或者三小时内 更新预警信息（图 8）。 图 7 预期暴雨预警信息发布提前期 figure 7 expected lead time of rainstorm warning information 图 8 预期暴雨预警信息更新频率 figure 8 expected update frequency of rainstorm warning information 7% 19% 16% 13% 41% 4% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 26% 20% 23% 13% 17% 2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 出现一次错误预警，公众对预警的信任度低于 公众对暴雨预警的整体信任度。出现两次错误 预警，同样出现公众对预警的信任度低于公众 对暴雨预警的整体信任度。由此表明，出现错 误预警会使得公众对暴雨预警的信任度下降。 进一步检验公众对暴雨预警的响应是否存在 虚警效应。将公众对暴雨预警的信任度进行分 组，分组依据如表 3 所示。总共分为三组：出 现预警不准可信度降低的划为预警信任降低 组，出现预警不准可信度不变的划为预警信任 不变组，出现预警不准可信度上升的划为预警 信任加强组。由表 4 和表 5 可知，仅在蓝色预 警下，出现一次或两次错误预警，不同预警信 任度分组下的预警响应有显著差异（2=12.903, p=0.002; 2=15.325, p=0.000），即错误预警的 出现降低了公众对预警的信任度，从而降低公 众对预警的响应比例。在黄色预警、橙色预警 和红色预警等级下，出现一次或两次错误预警， 不同预警信任分组同样呈现虚警减少响应比 例的趋势。 图 6 暴雨预警虚警效应 figure 6 cry-wolf effects 表 2 不同虚警效应定义下的暴雨预警可信度统计量 table 2 summary statistics for warning credibility under cry-wolf effects are defined based on the accuracy of warnings in the last time or the first two times 均值 标 准 偏 差 中 位 数 众 数 下 四 分 位数 上四分位数 偏度 峰度 以往预警的可信度 78.74 16.817 80 80 70 90 -1.604 4.27 上次不准可信度 71.24 18.66 74 80 60 81 -1.171 2.478 上两次不准可信度 60.59 25.588 60 50 50 80 -0.7 -0.035 上次准可信度 83.37 14.848 85 80 80 90 -1.873 6.192 上两次准可信度 87.73 14.417 90 100 80 100 -2.436 8.972 表 3 预警可信度分组 table 3 the credibility of warning group 条件 人群分组 以往预警可信度>如上次不准现在的可信度 1=预警信任降低组 以往预警可信度=如上次不准现在的可信度 2=预警信任不变组 以往预警可信度<如上次不准现在的可信度 3=预警信任加强组 以往预警可信度>如前两次不准现在的可信度 1=预警信任降低组 以往预警可信度=如前两次不准现在的可信度 2=预警信任不变组 以往预警可信度<如前两次不准现在的可信度 3=预警信任加强组 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 48 表 4 一次预警不准下虚警效应检验.表中数字表示响应比例（%） table 4 cry wolf effect for one wrong warning test. the numbers are the percentage of people response to warnings in each level. 预警可信度分组 蓝色 预警 卡方 （p 值） 黄色 预警 卡方 （p 值） 橙色 预警 卡方 （p 值） 红色 预警 卡方 （p 值） 预警信任降低组 30.3 12.903 （0.002） 54.5 4.555 （0.103） 72.7 0.523 （0.77） 86.8 2.164 （0.339 ） 预警信任不变组 46.5 62 73.9 89.7 预警信任加强组 35.8 54.8 75.5 90.9 表 5 两次预警不准下虚警效应检验.表中数字表示响应比例（%） table 5 cry wolf effect for two wrong warnings test. the numbers are the percentage of people response to warnings in each level. 预警可信度分组 蓝色 预警 卡方 （p 值） 黄色 预警 卡方 （p 值） 橙色 预警 卡方 （p 值） 红色 预警 卡方 （p 值） 预警信任降低组 26.5 15.325 （0.00） 54.8 5.296 （0.071 ） 72.9 0.611 （0.737 ） 85.5 5.946 （0.051 ） 预警信任不变组 48.2 63.1 75.4 89.7 预警信任加强组 35.8 55.1 75.5 95.9 虽然各组响应比例无显著差异（一次预警不准 检 验 ： 2=4.555, p=0.103; 2=0.523, p=0.77; 2=2.164, p=0.339），（两次预警不准检验： 2=5.296，p=0.071；2=0.611，p=0.737；2=5.946， p=0.051）。结果表明，在四级预警等级下，相 对于预警不变组和预警信任加强组，因错误预 警造成的预警信任度降低组的人群对预警的 响应最低，说明虚警效应在引起公众对预警的 信任度降低时，也进一步影响公众对预警的响 应。 3.3 公众对暴雨预警信息的期望 针对暴雨不及时的评价，其中 50%以上的 公众希望提前至少六小时发布暴雨预警（图 7）。 而预警发布提前期越长，预警信息质量越差[7]， 在预期提前至少六小时发布暴雨预警的同时， 公众期望可以在半小时、一小时或者三小时内 更新预警信息（图 8）。 图 7 预期暴雨预警信息发布提前期 figure 7 expected lead time of rainstorm warning information 图 8 预期暴雨预警信息更新频率 figure 8 expected update frequency of rainstorm warning information 7% 19% 16% 13% 41% 4% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 26% 20% 23% 13% 17% 2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 出现一次错误预警，公众对预警的信任度低于 公众对暴雨预警的整体信任度。出现两次错误 预警，同样出现公众对预警的信任度低于公众 对暴雨预警的整体信任度。由此表明，出现错 误预警会使得公众对暴雨预警的信任度下降。 进一步检验公众对暴雨预警的响应是否存在 虚警效应。将公众对暴雨预警的信任度进行分 组，分组依据如表 3 所示。总共分为三组：出 现预警不准可信度降低的划为预警信任降低 组，出现预警不准可信度不变的划为预警信任 不变组，出现预警不准可信度上升的划为预警 信任加强组。由表 4 和表 5 可知，仅在蓝色预 警下，出现一次或两次错误预警，不同预警信 任度分组下的预警响应有显著差异（2=12.903, p=0.002; 2=15.325, p=0.000），即错误预警的 出现降低了公众对预警的信任度，从而降低公 众对预警的响应比例。在黄色预警、橙色预警 和红色预警等级下，出现一次或两次错误预警， 不同预警信任分组同样呈现虚警减少响应比 例的趋势。 图 6 暴雨预警虚警效应 figure 6 cry-wolf effects 表 2 不同虚警效应定义下的暴雨预警可信度统计量 table 2 summary statistics for warning credibility under cry-wolf effects are defined based on the accuracy of warnings in the last time or the first two times 均值 标 准 偏 差 中 位 数 众 数 下 四 分 位数 上四分位数 偏度 峰度 以往预警的可信度 78.74 16.817 80 80 70 90 -1.604 4.27 上次不准可信度 71.24 18.66 74 80 60 81 -1.171 2.478 上两次不准可信度 60.59 25.588 60 50 50 80 -0.7 -0.035 上次准可信度 83.37 14.848 85 80 80 90 -1.873 6.192 上两次准可信度 87.73 14.417 90 100 80 100 -2.436 8.972 表 3 预警可信度分组 table 3 the credibility of warning group 条件 人群分组 以往预警可信度>如上次不准现在的可信度 1=预警信任降低组 以往预警可信度=如上次不准现在的可信度 2=预警信任不变组 以往预警可信度<如上次不准现在的可信度 3=预警信任加强组 以往预警可信度>如前两次不准现在的可信度 1=预警信任降低组 以往预警可信度=如前两次不准现在的可信度 2=预警信任不变组 以往预警可信度<如前两次不准现在的可信度 3=预警信任加强组 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 48 49 [10] pate-cornell m e. probabilistic assessment of warning system. report no. 13, center for economic policy research, stanford university, stanford, california, 1984. [11] leclerc j, joslyn s s. the cry wolf effect andweather-related decision making. risk analysis, 2015, 35 (3): 385-395. doi: 10.1111/risa.12336. [12] joslyn s l, savelli s. communicating forecast uncertainty: public perception of weather forecast uncertainty. meteorological applications, 2010, 17(2): 180–195. [13] morss r e, demuth j l, lazo j k. communicating uncertainty in weather forecasts: a survey of the us public. weather and forecasting, 2008, 23(5): 974–991. [14] lazo j k, morss r e, demuth j l. 300 billion served: sources, perceptions, uses, and values of weather forecasts. bulletin of the american meteorological society, 2009, 90(6): 785–798. [15] savelli s, joslyn s l. boater safety: communicating weather forecast information to high-stakes end users. weather, climate, and society, 2012, 4(1):7–19. [16] dash n, gladwin h. evacuation decision making and behavioural responses: individual and household. nat hazards rev, 2007, 8: 69–77. [17] mileti d, o’brien p. warning during disaster: normalizing communicated risk. soc probl,1992, 39(1): 40–55 [18] bradford r a, o’sullivan j j, van der craats i m, krywkow j, rotko p, aaltonen j, bonaiuto m, de dominicis s, waylen k, schelfaut k. risk perception – issues for flood management in europe, nat. hazards earth syst. sci., 12, 2299– 2309, doi:10.5194/nhess-12-2299-2012, 2012. [19] louviere j j, hensher d a, swait j d. stated choice methods analysis and applications. cambridge university press, uk, 2000. [20] matyas c, srinivasan s, cahyanto i, thapa b, pennington-gray l, villegas j. risk perception and evacuation decisions of florida tourists under hurricane threats: a stated preference analysis. nat hazards, 2011, 59: 871–890. 图 9 预期暴雨预警发布形式 figure 9 expected issuance forms of rainstorm warnings 由图 9 可知，更多公众期望在暴雨预警信 息中包括将于概率的预报。有研究表明不确定 性信息可以在存在错误预警时保持公众对其 的信任，改善决策，并提高响应率[11]。 4. 结论与讨论 虽然有 80%以上的人群能认识到暴雨预 警存在的重要性，但有一半的公众对暴雨预警 信息缺乏关注。一方面可能因为村民的灾害风 险意识淡薄，另一方面预警信息传播渠道闭塞。 结果表明，除了较多使用的固定端如电视天气 预报，移动端如手机短信、网络和能保证覆盖 率的村镇广播及电子显示屏传播渠道接收率 都很低。因此首先需对村民加强普及灾害风险 知识，增加其灾害风险意识，可通过村委会组 织灾害风险学习教育。其次应重视移动端预警 信息传播渠道的使用，保证任何时刻公众能接 收到预警信息。 虽然大多数人认为暴雨预警信息发布及 时，但仍然有 30%的人认为预警发布不及时， 且更多的公众希望至少提前 12 小时收到预警 信息，而有研究表明随着预警提前期的增加， 预测精度会下降[7]。根据本文研究结果，公众 期望以小时为单位更新预警信息，所以可通过 在提前期内增加更新预警信息环节，解决提前 期长带来的虚警效应，同时也可避免提前期不 够导致来不及采取防护措施。 经检验，北京市暴雨预警确实存在虚警效 应，错误预警不但会降低公众对预警的信任度， 还会降低对预警信号的响应率。因此在发布暴 雨预警等级信号的同时，可引入降雨量发生概 率的预报，帮助公众更好的作出响应决策。 本研究也存在一定的局限性，首先，只考 虑北京郊区公众整体的预警信息认知；其次， 因为人们对预警响应决策过程是复杂的，涉及 许多因素，其中只有一些在本文中进行了探讨。 一些影响公众风险认知的预警信息仍需要进 一步讨论，如预警信息的清晰度、预警信息的 一致性等。此外，需设计更详细的问卷，提取 一些现象的原因。总之，本研究对促进预警沟 通很有意义。继续研究影响天气突发事件决策 的认知因素，可以在提高公众对预警的响应和 减少与天气有关的伤害和死亡方面发挥关键 作用。 参考文献 [1] joslyn s l, leclerc j e. uncertainty forecasts improve weather-related decisions and attenuate the effects of forecast error. journal of experimental psychology: applied, 2012, 18(1): 126–140. [2] mileti d s, drabek t e, haas j e. human systems in extreme environments: a sociological perspective. boulder: inst. behav. sci., univ. colo. 1975. [3] bliss j p, gilson r d, deaton j e. human probability matching behavior in response to alarms of varying reliability. ergonomics. 1995, 38: 2300–2312. [4] breznitz s. cry wolf: the psychology of false alarms, erlbaum, hillsdale, nj, 1989. [5] dejoy d m, cameron k a, lindsay j d. postexposure evaluation of warning effectiveness: a review of field studies and population-based research, in: handbook of warnings (human factors & ergonomics), edited by s.wogalter, m., pp. 35–48, lawrence erlbaum associates, 2006. [6] wolshon b, urbina e, wilmot c, et al. review of policies and practices for hurricane evacuation. i: transportation planning, preparedness, and response. natural hazards review. 2005, 6(3): 129–142. [7] pate-cornell m e. warning systems in risk management. risk analysis. 1986, 6: 223–234. [8] breznitz s. cry wolf: the psychology of false alarms, 1st ed. hillsdale, nj: lawrence erlbaum associates, 1984. [9] pate-cornell m e. warning system and risk reduction. proceedings of the workshop of the society for risk analysis, new york, 1983. 23% 37% 40% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 50 [10] pate-cornell m e. probabilistic assessment of warning system. report no. 13, center for economic policy research, stanford university, stanford, california, 1984. [11] leclerc j, joslyn s s. the cry wolf effect andweather-related decision making. risk analysis, 2015, 35 (3): 385-395. doi: 10.1111/risa.12336. [12] joslyn s l, savelli s. communicating forecast uncertainty: public perception of weather forecast uncertainty. meteorological applications, 2010, 17(2): 180–195. [13] morss r e, demuth j l, lazo j k. communicating uncertainty in weather forecasts: a survey of the us public. weather and forecasting, 2008, 23(5): 974–991. [14] lazo j k, morss r e, demuth j l. 300 billion served: sources, perceptions, uses, and values of weather forecasts. bulletin of the american meteorological society, 2009, 90(6): 785–798. [15] savelli s, joslyn s l. boater safety: communicating weather forecast information to high-stakes end users. weather, climate, and society, 2012, 4(1):7–19. [16] dash n, gladwin h. evacuation decision making and behavioural responses: individual and household. nat hazards rev, 2007, 8: 69–77. [17] mileti d, o’brien p. warning during disaster: normalizing communicated risk. soc probl,1992, 39(1): 40–55 [18] bradford r a, o’sullivan j j, van der craats i m, krywkow j, rotko p, aaltonen j, bonaiuto m, de dominicis s, waylen k, schelfaut k. risk perception – issues for flood management in europe, nat. hazards earth syst. sci., 12, 2299– 2309, doi:10.5194/nhess-12-2299-2012, 2012. [19] louviere j j, hensher d a, swait j d. stated choice methods analysis and applications. cambridge university press, uk, 2000. [20] matyas c, srinivasan s, cahyanto i, thapa b, pennington-gray l, villegas j. risk perception and evacuation decisions of florida tourists under hurricane threats: a stated preference analysis. nat hazards, 2011, 59: 871–890. 图 9 预期暴雨预警发布形式 figure 9 expected issuance forms of rainstorm warnings 由图 9 可知，更多公众期望在暴雨预警信 息中包括将于概率的预报。有研究表明不确定 性信息可以在存在错误预警时保持公众对其 的信任，改善决策，并提高响应率[11]。 4. 结论与讨论 虽然有 80%以上的人群能认识到暴雨预 警存在的重要性，但有一半的公众对暴雨预警 信息缺乏关注。一方面可能因为村民的灾害风 险意识淡薄，另一方面预警信息传播渠道闭塞。 结果表明，除了较多使用的固定端如电视天气 预报，移动端如手机短信、网络和能保证覆盖 率的村镇广播及电子显示屏传播渠道接收率 都很低。因此首先需对村民加强普及灾害风险 知识，增加其灾害风险意识，可通过村委会组 织灾害风险学习教育。其次应重视移动端预警 信息传播渠道的使用，保证任何时刻公众能接 收到预警信息。 虽然大多数人认为暴雨预警信息发布及 时，但仍然有 30%的人认为预警发布不及时， 且更多的公众希望至少提前 12 小时收到预警 信息，而有研究表明随着预警提前期的增加， 预测精度会下降[7]。根据本文研究结果，公众 期望以小时为单位更新预警信息，所以可通过 在提前期内增加更新预警信息环节，解决提前 期长带来的虚警效应，同时也可避免提前期不 够导致来不及采取防护措施。 经检验，北京市暴雨预警确实存在虚警效 应，错误预警不但会降低公众对预警的信任度， 还会降低对预警信号的响应率。因此在发布暴 雨预警等级信号的同时，可引入降雨量发生概 率的预报，帮助公众更好的作出响应决策。 本研究也存在一定的局限性，首先，只考 虑北京郊区公众整体的预警信息认知；其次， 因为人们对预警响应决策过程是复杂的，涉及 许多因素，其中只有一些在本文中进行了探讨。 一些影响公众风险认知的预警信息仍需要进 一步讨论，如预警信息的清晰度、预警信息的 一致性等。此外，需设计更详细的问卷，提取 一些现象的原因。总之，本研究对促进预警沟 通很有意义。继续研究影响天气突发事件决策 的认知因素，可以在提高公众对预警的响应和 减少与天气有关的伤害和死亡方面发挥关键 作用。 参考文献 [1] joslyn s l, leclerc j e. uncertainty forecasts improve weather-related decisions and attenuate the effects of forecast error. journal of experimental psychology: applied, 2012, 18(1): 126–140. [2] mileti d s, drabek t e, haas j e. human systems in extreme environments: a sociological perspective. boulder: inst. behav. sci., univ. colo. 1975. [3] bliss j p, gilson r d, deaton j e. human probability matching behavior in response to alarms of varying reliability. ergonomics. 1995, 38: 2300–2312. [4] breznitz s. cry wolf: the psychology of false alarms, erlbaum, hillsdale, nj, 1989. [5] dejoy d m, cameron k a, lindsay j d. postexposure evaluation of warning effectiveness: a review of field studies and population-based research, in: handbook of warnings (human factors & ergonomics), edited by s.wogalter, m., pp. 35–48, lawrence erlbaum associates, 2006. [6] wolshon b, urbina e, wilmot c, et al. review of policies and practices for hurricane evacuation. i: transportation planning, preparedness, and response. natural hazards review. 2005, 6(3): 129–142. [7] pate-cornell m e. warning systems in risk management. risk analysis. 1986, 6: 223–234. [8] breznitz s. cry wolf: the psychology of false alarms, 1st ed. hillsdale, nj: lawrence erlbaum associates, 1984. [9] pate-cornell m e. warning system and risk reduction. proceedings of the workshop of the society for risk analysis, new york, 1983. 23% 37% 40% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 43–51 50 51 atlantis press journal style introduction for volume 7, issue 3 this issue contains 7 papers. there are 3 contributions written in english and 4 contributions in chinese with english abstracts. the papers can be divided into three topics: hazard assessment, risk assessment and internet of intelligences for monitoring risks there are three papers in hazard assessment. in the first paper “a note on non-parametric estimation of the conditional hazard quantile function” by el hadj hamel, nadia kadiri and abbes rabhi, the authors study an kernel estimator of the conditional hazard quantile function of a scalar response variable y given a random variable x taking values in a semi-metric space and using the proposed estimator based of the kernel smoothing method. the almost complete consistency and the asymptotic normality of this estimate are obtained when the sample is an independence sequence. the second paper “maximum entropy-based model of high-threat landslide disaster distribution in zhaoqing, china” by shao-xiong yuan, guang-qing huang, hai-xian xiong, et al., puts forward a maximum entropy-based model of high-threat landslide disaster distribution in zhaoqing, china. landslide disaster that threatened over 100 people in zhaoqing, china, were taken as samples. sixteen environmental factors were selected. the suggested model was employed for simulation analysis of landslides. in the third paper “major element geochemistry of longshan loess profile in the central shandong mountainous regions, northern china” by min ding，shuzhen peng, longjiang mao, et al., the major elements of the longshan loess profile on the northern piedmont zones and intermountain valleys of mountainous regions in central shandong province in northern china, have been systematically tested and been compared with the yhc loess in the loess plateau to reveal the geochemical characteristics and material sources of ls loess. it is found that the average chemical composition of shandong ls profile is similar to that of typical loess at yhc profile. there is one paper in risk assessment. the paper “time limit of the probabilistic risk for natural disaster”, by jun guo & chongfu huang, studies the period of validity of probabilistic risk for natural disaster. they considerred that risk is for the future and meanwhile dynamic, the result of probabilistic risk analysis in natural disaster could just represent risk for a limited time, which is the timeliness of probabilistic risk. there are three papers related to the internet of intelligences. in the first paper “principle of internet of intelligences and development of its core technology ” by chongfu huang, the author describe the principle of internet of intelligences (ioi) and discuss the core technology in ioi, information diffusion technology. then, the concept of intelligent mathematics is suggested to develop ioi, where the factor space theory could play role to express knowledge and being cognizant of thinking. web mathematics would lay the foundation for intelligent mathematics. the second paper “a research on community risk radar with presetting forms to structure information in internet of intelligence” by yifang leng & chongfu huang, put forward some data forms of the collecting page in the community risk radar driven by ioi to reduce the proportion of unstructured information. increase the image upload and download function can help residents building the risk scenarios. in the risk event assessment stage, coding the php web page can achieve the online processing of risk data. third paper is“a study on construction method of consensus measure space of macro-seismic anomalies”, by weidan wang & chongfu huang. with the ioi help, the authors set up the theory system of macro-anomalies group’s measure space which is expressed by the fuzzy relationship matrix, and puts forward the concept of macro-anomalies group. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 100 ___________________________________________________________________________________________________________ 100 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). introduction for volume 8, issue 1 this issue contains 6 papers which can be divided into five topics: eco-environment risk, crop meteorological disaster risk, tourism disaster risk analysis，emergency management and financial risk. there are two papers in eco-environment risk. the first paper “a fuzzy method for assessing eco-environmental disaster risk caused by coalbed methane in china” by ye xue, et al., constructs an eco-environmental disaster risk assessment model during coalbed methane industrialization development in china by using intuitionistic fuzzy sets to describe the uncertain risk information, and the transformed interval value of mamdani intuitionistic fuzzy neural networks model. its validity is verified by simulation tests. furthermore, the assessment results are compared with those obtained by fuzzy neural networks model. the study supplies a decision support for routine supervision in ecoenvironmental risk assessment of coalbed methane industrialization developmen. in the second paper“geochemical and clay-size minerals evidence for the provenance of lqc loess deposits in the central shandong, northern china” by min ding, et al., the clay-size(< 5μm) minerals and major elements composition of bulk samples and of two grain-size fractions (< 20μm and 20~63μm) were analyzed in order to determine the provenance of the longqiaocun(lqc) loess in the central shandong based on comparisons with loess from the chinese loess plateau and with yellow river sediments. statistical analysis of the sedimentary clay-size minerals reveals that lqc loess deposits in the central shandong adjacent to the floodplain of the yellow river were not blown directly from the northern deserts as is the case of loess deposits in the chinese loess plateau. there is one papers in crop meteorological disaster risk, “the design of a drought weather index insurance system for summer maize in anhui province, china”by ying xu, et al., where the authors analyzed precipitation anomalies of summer maize in order to establish a drought weather index model that is based on tassel appearance and maturity stage as well as drought yield reduction rate. they utilized daily precipitation data encompassing the period between 1971 and 2010 from 15 agricultural meteorological observation stations within a summer maize planting area in anhui province, china. paper “a study of spatial evolution patterns of tourist destinations disaster risks”, by yingyue sun, et al., focuses on tourism disaster risk analysis, where the authors built a comprehensive disaster risk evaluation model using selected 17 economic and social indicators from tourist destinations in jilin province based on the four-factor theory of natural disaster risk formation. the model was used to assess tourist destination disaster risk in the province for the period of 2009 2014. using gis technology, they developed a tourist destination disaster risk zoning map of 2009–2014 for jilin province, classified the spatial evolution patterns of tourist destination disaster risk. there is one paper in emergency management, which is “wargame mapping and implementation for emergency evacuation of residents in urban waterlogging disaster” by peng chen, et al., where the authors introduced the wargame excise to the residents’ emergency evacuation in urban waterlogging disaster. as an essential component of wargame, the chessboard (map) is an important tool for the study of deduction and command training. in this paper, the research focuses on the methods of generating wargame map based upon vector map data, from the viewpoints of mathematic foundation as well as quantitative treatments of point, line and plane for the map composed by regular hexagonal grids. the last article in this issue, titled“research on the development of county finance in guizhou province in the promotion of precise poverty alleviation” by yu ding and mu zhang, is related to financial risk, where the authors selects the 50 state-level poverty-stricken counties in guizhou province as the research object, and uses financial scale, financial efficiency and financial structure to represent the level of financial development in each county, using 1 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 1–2 economic growth and income distribution as controlled variable. the poverty of every county is expressed by poor slow index. applying the panel data model, the promotion of local financial development to the targeted poverty alleviation is studied. the empirical evidence shows that the financial scale, financial efficiency, financial structure, economic growth and poverty reduction of 50 national poverty-stricken counties in guizhou province are positively correlated. however the financial scale is more significant to reduce the incidence of poverty than that of financial efficiency. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca 2 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 1–2 基于行走时间最短的机场停机位分配问题的一种混合算法：遗传算法加涟漪扩散模型 a study on marine vessels’ path optimization under typhoon scenarios li xu1,2, xiaobing hu1,2 1 state key laboratory of earth surface processes and resource ecology, beijing normal university, beijing 100875, china 2academy of disaster reduction and emergence management, beijing normal university, beijing 100875, china abstract typhoon may cause a huge impact on the safety and costs of marine vessels on the voyage. in order to effectively solve the path optimization problem for marine vessels under typhoon scenarios, this paper proposes a hybrid algorithm integrating genetic algorithm (ga) with receding horizon control (rhc), which considers real-time typhoon data, marine environment data including ocean currents and reefs etc., as well as safety operation requirements for marine voyage. then, simulation experiments are conducted to test the proposed method. the experimental results indicate that the new algorithm is effective and efficient to optimize voyage paths for marine vessels under typhoon scenarios, i.e., when compared with global path optimization, the new algorithm can improve the cost-efficiency of voyage path with safety guarantee. keywords: marine vessel; real-time path optimization; typhoon; genetic algorithm; receding horizon control 船舶海上躲避台风航线优化研究 徐里 1,2 ，胡小兵 1,2 1.北京师范大学 地表过程与资源生态国家重点实验室，北京 100875 2.北京师范大学 减灾与应急管理研究院，北京 100875 摘要：台风对海上航行船舶的安全和效益具有巨大的影响。为了更好地解决船舶海上航行避台路径优化问题， 以达到兼顾船舶安全性及经济性运营，本文提出了一种遗传算法（ga）加滑动地平线策略（rhc）的混合 算法；并结合台风实时数据、海洋洋流及岛礁等环境数据、船舶避台安全运营数据，开展了仿真实验。结果 表明，该算法在船舶海上航行躲避台风问题上有效可行，并相对于传统的全局路径规划算法，在保证船舶安 全航行的前提下，能更好的改善船舶运行的成本效益。 关键词：船舶路径优化；台风；应急管理；混合算法；滑动地平线策略 1 引言 自然灾害系统是由致灾因子、孕灾环境和承灾 体共同组成的复杂系统 [1] 。灾害风险往往是天、地、 人综合作用的结果 [2] 。对于台风这种自然灾害，其 致灾因子有大风、强降雨、巨浪和风暴潮：台风的 大风具有极强的摧毁力，可拔树、倒屋、翻船、甚 至摧毁城镇；台风过境时所形成的强降雨，在短时 间内形成大量的雨水汇集，可导致山洪，城市内涝， 农田淹没等灾害，还会导致山体滑坡，泥石流等次 生灾害的发生；巨浪和风暴潮与天文大潮共同作用 会在沿海地区形成多达几米的增水，造成沿海城市 海水倒灌、港口基础设施损毁、沿海养殖业毁灭性 损失等严重后果 [3] 。就航运领域而言，台风对海上 航行船舶的安全和效益具有巨大的影响。 在海上，强大台风中心附近的最大风速可达到 100～120 米/秒，形成的海浪可达到十几米。在风 速和巨浪的影响下，船舶承受着巨浪的撞击。当船 长小于涌浪波长时，船舶在巨浪的涌动过程中，船 舶螺旋桨会露出水面，使螺旋桨空转，极易造成螺 received 7 april 2017 accepted 7 may 2017 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 72 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 旋桨掉落，尾轴断裂等事故；当船长接近于涌浪波 长时，船舶可能会同时跨一个或两个波峰，导致船 舶发生中拱或中垂，船体结构变形，甚至船体断裂 等事故；另外船舶顶风航行时，船舶的机电设备长 时间高负荷运转，极易导致机械故障，而一旦发生 机械故障，船舶的动力受到影响，船舶将随风飘荡 而发生触礁、搁浅甚至倾覆的事故，尤其是老旧船 舶，在遭遇台风时，由于船体结构和机电设备老化， 抗台能力大大下降，更容易发生损失 [4] 。由于船舶 在抗台过程中面临各种危险，因此，各船舶管理公 司在船舶应对台风问题上均极为重视，但主要采取 被动措施，例如锚泊避风、长距离绕行，或较长时 间的滞航，其结果是虽然保证了船舶的安全，却造 成船舶燃油消耗的增加，并导致航期延误，无论是 对货主还是船务公司或租船人而言，都产生了不必 要的经济损失。由于台风对于海上航行船舶造成的 巨大影响，因此，船舶在海上航行应对台风时，寻 找一条兼顾安全性和经济性的航行路径是一个值 得研究的课题。本文提出一种将滑动地平线控制 （rhc）策略和遗传算法（ga）相结合的混合算 法，应用于船舶海上航行躲避台风航线优化问题， 以期达到在保证船舶安全航行的前提下，减少因绕 行和延期所带来的经济损失，为船舶企业运营带来 更好的成本效益。本文的余下部分安排如下。第 2 节给出了船舶在海上自由海域的航线实时优化问 题的数学描述。第 3 节给出了遗传算法（ga）加 滑动地平线控制（rhc）策略的混合算法的设计细 节和主要技术参数设定。第 4 节通过仿真实验证实 此算法的有效性。第 5 节给出了本研究的主要结 论。 2 船舶在海上自由海域航行时的路径实时优化问 题 路径优化和障碍规避是交通研究邻域里的两 个重要问题 [5-10] 。传统的路径优化问题通常都是基 于特定的路径网络进行求解 [5,6] ，而路径网络一般 是基于有限数目的结点和联结结点的链接来定义 的。有了结点和结点间的链接，路径优化的目的就 是找出一组结点和链接的子集，从而连通给定的起 点和终点，并使得给定的路径指标最优化。自由区 域路径优化问题则不同，因为在自由区域路径优化 问题中不存在路径网络，而是可以在除障碍区之外 的整个区域内自由规划路径（因而可以视为有无穷 多的结点和链接）。障碍区可以是固定于某位置的 障碍物，也可以是动态变化的特殊区域，例如台风 区域、其它运动物体等等。因此，自由区域路径优 化问题通常要求针对动态障碍区进行实时路径优 化。带动态障碍区的自由区域路径优化问题具有十 分广泛的应用背景。船舶在公海自由航行时，需要 根据暗礁区域、台风区域、及其它船舶的位置等障 碍区信息进行实时的航路规划。 目前，国内外学者已开展了许多关于船舶在海 上躲避台风航线优化这类自由区域路径优化问题 的研究，提出了扇形避台法 [11] ，目标圆法 [12,13] ，这 些方法对于船舶躲避台风优化路径选择的研究，多 凭借航海经验进行避台决策制定，可以肯定的是上 述研究对于船舶避台的安全性均有显著的效果，但 难以实现船舶运营的最佳的成本效益。文献 [14] 提出 了一种改进的遗传算法用以求解带有动态障碍区 的自由飞空域中的飞行路径优化问题。然而，由于 该改进算法采用了全局优化策略，所以很难保证在 动态环境中应用的时效性。 为了更好地解决船舶在海上躲避台风航线优 化问题，本文提出一种结合遗传算法（ga）加滑动 地平线控制（rhc）策略的混合算法。rhc 策略又 称为模型预测控制（mpc），专门用于解决在动态 环境下的实时优化问题。简单地讲，rhc 策略是提 前 n 步进行的在线优化策略。在每个时间间隔起 点，基于当前的可用信息，rhc 策略优化接下来最 近的 n 段时间间隔的路径，同时只执行当前的一 个时间间隔的优化路径。到下一个时间间隔，rhc 将基于更新的信息，重复计算接下来的 n 段时间 间隔的优化路径，以此类推。rhc 策略现在已经被 控制工程领域广泛接受，与其他控制方法相比， rhc 策略具有许多优点 [15] 。最近，rhc 策略在运筹 学领域的应用研究也受到了很多重视。例如，在 文献 [16] 讨论了关于将 rhc 策略应用于各类离散事 件系统的理论研究工作。文献 [17] 总结了许多 rhc 策略在管理学领域的应用研究。文献 [18] 尝试了应 用 rhc 策略进行有效的动态实时飞机进近排序研 究。 2.1 可选自由航线 在船舶可自由航行的海域进行路径优化，有无 数可选的子航线。因此，有必要合理并适当地简化 初始问题。使用传统的结构化的路径网络就是简化 此问题的一个简单方式。为了能充分利用好船舶可 自由航行的海域，我们可定义一个有限的可选方向 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 73 的集合，并引入“时间段”概念，以此将船舶可自 由航行的海域转换成一个动态路径网络。这样就可 以实现在简化问题和使用的恰当算法找到的最优 方案之间来进行合理的折中。 图 1 船舶离散化自由海域航线 fig. 1. discretization of free navigation area for marine vessel’s path planning 使用一个有限的数值的集合来代表船舶可选 航线方向，以代替原来的无限的船舶航线方向的集 合，是离散化船舶可自由航海域的关键技术之一。 本文中，有限的船舶可选航线方向集合假定如下： ω = [   50,,20,10,0 , direθ ] (1) 其中 dire θ 表示从当前船舶位置点到终点的直接方 向。所以，每次需要确定船舶可选航线方向时，仅 有 37 个值可供选用。 关于“时间段”的概念，我们假设有一个信息 发布系统将周期性地发布船舶航行海域的环境数 据，即台风七级风圈影响海域数据，这个周期叫“时 间段”。船舶使用最新更新的信息以便在下一个时 间段开始来优化剩余的航线。一个可选的航线包括 一系列和“时间段”相关的子航线。用于当前时间 段的子航线是由之前的优化来决定的。图 2 示例了 船舶离散化自由海域航线选择的过程。如果时间段 太长，而在 ω 中的可选航线方向太少，则离散化 后的航线网络可能和传统的结构化的航线网络类 似，从而几乎无法包括全局最优的航线。另一方面， 如果时间段太短，而可选航线方向太多，将会有许 多可选的航线，从而导致找到最优航线的计算时间 大大增加，实时执行可能性就大大降低了。 2.2 航线优化的性能指标 如 2.1 节所述，一个可选的船舶航线包括一 系列和时间段相关的子航线。每个子航线（可选航 线的最后一个子航线除外）的航行时间都可以被看 做一个时间段的长度。然后，可选航线的总航行 时间由其所包括的子航线的数值来决定。因此， 虽然指标是总航行时间，但现在优化的基本变量 是子航线起始点和终点的坐标。这些基本变量和 一些重要参数如示意图 2 所示，（x,y）是一个点的 坐标，sab 是 a 和 b 两点的距离，（ , vϕ ）是一个 点的洋流运动方向和速度，θ 表示船舶航向（本文 中所有方向都是相对于正北方向而定义的）。严格 讲，计算一个子线段的终点坐标，即（xb,yb），是 不可能的，因为（xb,yb）和（ bb v,ϕ ）互为前提条 件。但是，如果时间段定义得足够短，通常可以 假设子线段终点处的洋流运动参数与起始点处一 致，即（ bb v,ϕ ）=（ aa v,ϕ ）。根据此假设，计 算（xb,yb）就不再需要（ bb v,ϕ ）。 (a) 目标点 (b) 变量 图 2 计算船舶航行的一些重要参数 fig. 2. key parameters for calculating the movement of vessel θr 点 a 点 b abs baθ 坐标：（ aa yx , ），（ bb yx , ） 点 a 处的洋流参数: （ aa v,ϕ ） 点 b 处的洋流参数: （ bb v,ϕ ） 正北方向 船舶绝对速度： vabs 洋流运动速度：ve 船 舶 相 对 速 度：vr θabs θe 正北方向 船 舶 当 前 位 置 目 的 港 口 船舶自由 航行海域 台风七级风圈 或暗礁、其它船舶 可能的子航线 优化后船舶航线 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 74 一条子航线的终点坐标（xb,yb），可通过如 下计算。 cosb a ab bax x s θ= + (2) sinb a ab bay y s θ= + () 其中 tsabsab tvs = , absba θθ = (4) )cos(222 rerereabs vvvvv θθ −++= (5) )/)sin((sin 1 absreerabs vv θθθθ −+= − (6) ae ϕθ = , ae vv = w av v= (7) 其中 tts 是一个时间段的长度。坐标（xb,yb）随后 将被用于下个子航线的起始点。然后，基于气象 服务部门发布的洋流运动参数，新子航线的终点 坐标也可通过相同的方式进行计算。子航线的计 算可以一直进行到到达目的地港口为止。 对于一个可选航线最后的子航线，其终点就 是目的港，因此，(xb,yb) 是已知的，不再需要按 公式（1）和（2）计算。但是，最后子航线的航行 时间一般不再等于一个时间段的长度，而是需要 进行计算。假定图  中的 b 点就是目的港，则最后 子航线的航行时间可以通过如下计算： absablast vst /= (8) 其中 22 )()(),( bababaab yyxxppdiss −+−== (9) ))(),(((2tan90 ababbaabs xxyya −−−== θθ (10) )/)sin((sin 1 reabseabsr vv θθθθ −−= − (11) )cos()cos( eabserabsrabs vvv θθθθ −+−= (12) 而“ )(2tan ⋅a ”是计算四象限反正切的函数。 假设，排除最后的子航线，在一条可选航线 里有 n 条子航线。那相应的航行时间成本为： lastts tntj +=1 (1) 3 遗传算法（ga）加滑动地平线控制（rhc）策略 的混合算法 应用于船舶在海上自由海域航线实时优化问 题的现存方法有一个共同点：即在每一个时间段 ，优化的对象都是从当前子航线的结束点到目的 港的剩余航程。因此，这些方法总会面临两个常 见问题。一个问题是，在受台风风圈影响的自由 海域中，要在一个很短的时间段内完成对于长途 航线优化，通常很难实现。另一个问题是，在受 台风风圈影响的自由海域环境中，由于未来不确 定信息的原因，从目的港反推到当前子航线结束 点的优化方法所计算出的优化航线，其性能和可 靠性都会大大降低。 3.1 滑动地平线策略（rhc） 本文提出使用滑动地平线策略（rhc）来克服 现存方法中的上述问题。在每个阶段，即在每个 时间段，rhc 策略将只优化近期 n 个时间段内的航 线。这样，无论到目的港距离有多远，每次优化 所需的计算时间都有一个上限，这个计算时间的 上限主要取决于 n，即滑动地平线的长度。而且， 适当地选择滑动地平线的长度能像过滤器一样滤 除未来的不可靠信息。图 3 直观地显示了 rhc 策略 以及其相对于传统全局路径动态优化策略的潜在 优势。 图  rhc 策略与传统全局路径优化策略在船舶避台航线 优化问题的比较 fig. . comparison between rhc and traditional global dynamic optimization strategy in the route optimization problem for marine vessels under typhoon scenarios 公式（1）中所定义的 j1 是根据传统全局路 径动态优化策略而设计的优化性能指标。对于 rhc 策略，在每个时间段，仅需优化滑动地平线 内的航线，这只与 n 的大小有关，而与到目的港 的距离无关。因此，最小航行时间看起来对 rhc 策略没有什么意义。但事实是，在船舶可自由航 船 舶 1 和 船 舶 2 目 的 港 目 的 港 船舶 1（rhc 策略） 船舶 2（全路 径优化策略） （rhc 策略） （全路径优化策略） journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 75 行海域内，与 n 个时间段对应的多数可选航线都 是 z 字型（因为如果每个时间段都可以随机选取方 向的话，正如后文中遗传算法的做法，那么所得 到的 n 个时间段的总航线很少可能会是直线），这 意味着捷径通常存在于 z 字型路径的子航线之间 ，如图 4 所示。显然，即使原来计划的可选 z 字型 航线是基于 n 个时间段的固定长度的滑动地平线 ，但走捷径后，除了实际航线长度缩短了，沿捷 径从 z 字型航线起点到其终点的航行时间也会缩 短，不过具体缩短了多少并不确定。因此，通过 找捷径降低基于固定长度的滑动地平线的航行时 间仍是有意义的。事实上为了找到最短航线，传 统的全局路径动态优化策略也需要找捷径。 假定，在第 k 个时间段，在走捷径后，原来的 可选 z 字型航线变为了 m（k）个时间段，其中 0 ≤ m（k） ≤ n 是一个实数，而 m（k）的小数部 分等于通过对应捷径上最后一个子航线的航行时 间除以 tts，即有： ( ) ( ( )) /last tsm k floor m k t t− = (14) 其中“floor”使 m（k）四舍五入向负无穷大就近 取整。 图 4 n 个时间长度的 z 字型航线和捷径 fig. 4. zigzag route of n-time-slice-long and shortcut 这样一来，rhc 策略采用的性能指标如下： j2(k)=m(k)tts+wterm(k) (15) 其中 wterm（k）是一个与最后一条子航线和目的港 相关的终端加权函数。稍后我们会提供更多关于 wterm（k）细节的论述。然后，在一个船舶可自由 航行海域环境下优化航线的 rhc 策略可以按如下 描述： 步骤 1：当船舶从起点出发，让 k =0。 步骤 2：接收气象部门发布的更新台风数据 ，以当前运动物体所在点 p（k）作为开始航线优 化的起始点，然后解决如下的最小化问题 )(min 2)|(),...,|2(),|1( kjknkpkkpkkp +++ (16) 其中 p（k+i|k）， 1, ,i n=  是与第 k 时刻所对应 的滑动地平线的一个可选 z 字型航线中的第 i 子航 线的终点，这些点不能位于台风七级风圈等船舶 不 可 航 行 海 域 之 内 。 假 设 最 优 解 为 )]|(),...,|1([ ** knkpkkp ++ ，而相应的 捷 径 为 )]|))(((),...,|1([ ** kkmceilkpkkp ff ++ ，其中“ceil”将 m（k）四舍五入向正无穷大就近 取整。 步骤 ：当船舶到达 p（k），设定 )|1()1( * kkpkp f +=+ (17) 则船舶在当前时间段按照[p（k）, p（k+1）]所确 定的子航线航行。 步骤 4：如果 p（k+1）不是目的港，让 k=k+1 ，然后去步骤 2。否则，计算结束。 3.2 基于遗传算法（ga）的优化程序 第 2 节里的数学模型提供了离散化的船舶可自 由航行海域的航线，求解最小化问题（16）首先就 需要在离散化的船舶可自由航行海域的航线中快 速高效地找到可行的船舶航线。众所周知，遗传算 法（ga）是一个大规模平行随机搜索的优化算法， 它能很好地解决最小化问题（16）。本文则提出基 于 rhc 策略的 ga 算法，用以在船舶可自由航行 海域的实时航线优化问题。在 ga 中的染色体是根 据原始的 z 字型航线或捷径航线中的子航线端点 来构造的。由于 m（k）是一个不确定的有限实数， 不同的染色体可以有不同的长度。换句话说，本文 中的遗传算法采用的是可变长度的染色体设计。因 此，一个染色体的构造可以是：第一个基因记录 m （k）的值，“ceil（m（k））”是相应船舶航线中子 航线端点的数目，而接下来的基因按顺序记录下这 些端点的坐标，如图 5 所示。 基于一个染色体上记录的信息，相应的船舶航 线的 j2（k）可以根据公式（2）到公式（15）计算 出来。在第 k 时刻，假定 ga 中每代群体有 n 个染 色体，第 i 个染色体的 j2（k）值是 qi（k），而 qmax （k）和 qmin（k）表示 j2（k）在此代中的最大和最 小值。那么，第 i 个染色体的适应性由式（18）定 船舶 与有 n 个时间段的滑动地 平线对应的 z 字型航线 与 z 字型航线对应的捷 径 z 字型航线中的子航线 的终点 捷径中的子航线的终点 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 76 义。 图 5 ga 中的染色体的结构和含义 fig. 5. the structure and meaning of chromosomes in ga ( ) ( ) ( ( ) ( )) / , ( ) ( ) max max min max min( ) ( ) ( ) ( ( ) ( )) / ( ), ( ) ( ) max max min max max min q k q k q k q k n q k q k if k i q k q k q k q k n q k q k q k i − + − ≠ = − + − + =    (18) 在文献 [15] 中，提出了一些改进 ga 性能的有 效技术，如自适应交叉和变异概率，以及启发法 规则等。本文中的 ga 将采用这些技术，但限于篇 幅原因，不多赘述。 3.3 滑动地平线的长度和终端加权 滑动地平线的长度，即 n 的选择，是至关重 要的。rhc 策略每轮在线计算时间有一个上限， 主要取决于 n 的大小，通常可以通过模拟仿真进 行估计。因此，只要 rhc 策略中的时间段大于这 个计算时间的上限，则无论全程距离有多远，我 们总能保证算法的实时有效性。另外，适当地选 择滑动地平线的长度，可以像过滤器一样滤除长 远未来的不可靠信息。如果 n 太大，rhc 策略将 面临和现存的全局路径优化方法一样的关于实时 计算和台风等环境信息动态变化的问题。反之， 如果 n 太小，rhc 策略将变得短视，而且性能将 极大的降低。因此，需要合理选择 n 的大小，以 使 rhc 策略在在线计算时间和算法的稳定性能之 间得到最好的平衡。 但是，滑动地平线的性质让 rhc 策略在某种 意义上不可避免地变得短视，特别是和静态环境 中的传统全局路径优化方法相比较时。因此，我 们需要精心设计 j2（k）中的终端加权 wterm（k） ，以求能有效降低 rhc 策略的短视性。 在控制工程领域，无终端加权的 rhc 策略曾 被广泛应用，然而，这种 rhc 策略经常导致系统 变得不稳定。为了解决这个问题，人们专门引进 了终端加权的技术 [15] 。现在，终端加权已成为保 证 rhc 策略的稳定性的一个关键技术。本文针对 于船舶在自由海域航行遭遇台风的情况 假设 wterm（k）从 j2（k）中移除，即 0term k =w ( ) (19) 那么，如果目的地不在第 k 个时间段的范围内的话， j2（k）将没有目的港的信息。这种情况下，航线优 化的结果将导致一个随机的船舶航线，这很可能永 远不会指向目的港，如图 6 中的虚线所示。这是一 个由于在 j2（k）中没有终端加权所造成的不稳定 情况。 将目的地港信息加到 j2（k）中的一个简单方 法是使用如下的终端加权计算： absadlastterm vpkpdiskw /)),(()( ..= (20) 其中 plast（k）是可选航线中的最后一条子航线的终 点，pd.a.是目的港，而地上速率 vabs 和函数“dis” 在 公式（9）和公式（12）中分别给出了。在公式 （20）中的 wterm（k）能有效避免由于公式（19） 而产生的随机船舶航线，并可以在许多条件下让船 舶抵达目的港。但是，公式（20）中没有台风等不 可航海域的信息，有时会产生一个新的问题，即船 舶被困在一个海域内，rhc 策略几乎无法使船舶 驶离该海域，如图 6 上的点虚线所示。 图 6 各种终端加权的效果 fig. 6. effects under different terminal weighting terms 为了避免被困的现象，应在终端加权中加入一 些必要的障碍区域的信息。基本上，在 pd.a.（目的 地）和 plast（k）（可选航线中的最后一条子航线的 终点）之间存在的七级台风风圈、附近航行船舶、 无目的港的 wterm 可能被困的 wterm 最 优 化 的 可能绕远的 wterm 船舶 起始港 目的港 ga 中的一条染色体： (x1, y1) … m（k）的 数值 表达航线的第一 条子航线的终点 表达航线的最后一 条子航线的终点 m(k) (xceil(m(k)),yceil(m(k))) (x2, y2) (x1, y1) 4. (x2, y2) (x, y) (x4, y4) journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 77 岛礁等不可航行海域是导致船舶被困问题的主要 原因。为了方便考虑问题，下文中将这些在 pd.a. 和 plast（k）之间的海域叫做 iw 不可航行海域，其 它海域叫 ow 不可航行海域。如果没有 iw 不可航 行海域，那么 wterm（k）就如公式（20）。否则， 到 plast（k）最近的 iw 不可航行海域（可能包括相 互重叠的一些台风风圈、岛礁等不可航海域）可以 用图 7 进行分析： 图 7 静态环境下第一种终端加权说明 fig. 7. illustration of the first terminal weighting term in static environment 其中 1θ 和 2θ 是显示在图 7 上的角度。wterm（k）能 防止运动物体被困在某一区域，因为在潜在的受困 区域， ),max(/),min( 2121 θθθθ 接近于 1，将导致 很大的终端加权函数值。 然而，按照图 7 中的 wterm（k）可能会导致较长 的航行时间。如图 6 中的双点虚线所示，为了避免 被困，运动物体可能会偏离直接航向 direθ 很远。为 了使 rhc 算法能更有效地找到最优航线，而不只 是可行的航线，需要对终端加权做更多修改。假设 pprev（k）是遵照可选航线的点 plast（k）而延伸出 来的点。如果 iw 不可航行海域的数目不为零，那 么一个更有效的终端加权可通过图 8 进行分析。图 中 θ 和 4θ 是图 8 所示的角度。  0θ > 表示可选路 径中的最后一条子线段在向外偏离目的地。相反，  0θ < 表示在向内偏离。无论哪种情况，都会通过 终端加权进行调整，保证的 wterm（k）是一个很有 效的定义，其效果如图 6 中的实线所示。 然而，船舶在海上航行遭遇台风的情况，iw 不可航行海域是动态变化的，面积的大小、形状、 方向都随时发生改变甚至消失。为了方便处理问 题， 图 8 静态环境下第二种终端加权说明 fig. 8. illustration of the second terminal weighting term in static environment 不可航行海域的动态特征也可以简单地包括在 wterm（k）。某种程度上，利用不可航行海域的动态 特征的一个简单方式就是考虑到 plast（k）最近的 iw 不可航行海域的移动方向，即： absadlastterm vpkpdiskw /)),(()1/||)1(()( ..4 ++= θθbr (21) 6 5 6iwsign signr g θ θ θ θ= − −( ) ( ) (22) 其中 θ 和 4θ 如图 8 所示， 0α > 是一个系数，而 0b > 是一个系数， 5θ ， 6θ 和 iwθ 是相对于正北 方向的顺时针方向旋转角，如图 9 所示， iwθ 是到 plast（k）最近的 iw 不可航行海域的移动方向， 0g > 是一个转向参数，而“sign”是一个符号函 数。 目前，本文的终端加权计算仅使用了到 plast（k） 最近的 iw 不航行海域的信息。而关于如何使用其 它的 ow 不可航行海域信息可以进行进一步的研 究。例如，结合具体某一次航行任务中的不可航行 海域的空间分布特征和动态变化特性开展研究，但 已就不在本文的讨论范围内了。 图 9 动态环境下终端加权说明 fig. 9. illustration of terminal weighting term in dynamic environment θ ( 此 图 中  0θ < ) 4θ 起始港 pprev(k) pd.a. 船舶 已走航行过的航线 可能的计划航线 ow 不可航行海域 iw 不可航行海域 plast(k) 5θ θ pprev(k) pd.a. 船舶 可能的计 划航线 ow 不可 航行海域 iw 不可 航行海域 plast(k) 6θ 1θ 2θ 起始港 plast(k) pd.a. 船舶 已航行过的航线 可能的计划航线 ow 不可航行海域 iw 不可航行海域 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 78 4 仿真结果 为了评估本文所提出的遗传算法（ga）加滑 动地平线控制（rhc）策略的混合算法对解决船舶 海上躲避台风的航线优化问题的有效性，本文建立 了的一套仿真系统来模拟船舶在海上自由海域航 行的环境，并以 2016 年 14 号“莫兰蒂”台风为例， 开展了一个仿真案例。图 10 是“莫兰蒂”台风的 实际路径，为了更准确的验证算法的效果，仿真案 例中的台风相关数据设置均以实际莫兰蒂台风的 实际数据为基础进行设定。 在航线设计方面，假定一艘集装箱船舶正由香 港港口以 90 度角的方向台湾航行，航速的设定以 参照大型集装箱船舶的运营航速，从船舶航行的安 全性及实验的数据误差两方面考虑，本文设定船舶 受台风影响的不可航行海域是以台风的七级风圈 影响范围作为数据基础。 台风所在位置的设计，假定仿真实验开始时， “莫兰蒂”台风正由东南向西北方向迅速移动，台 风中心在台湾高雄以南附近海域。仿真案例的具体 参数设定如下： 参数 船舶速 度 （km） 船舶航 向 （度） 台风移 动速度 （km） 台风移 动方向 （度） 台风七 级风圈 半径 （km） 数值 7 90 18 25 200 （备注：本文中所有方向都是相对于正北方向而定义的） 图 10 “莫兰蒂“台风实时路径 fig. 10. the“meranti”typhoon real-time path 为了对比算法的效果，本文特地引用了基于 ga 的传统全路径动态优化算法。为了便于区别， 此后使用了 rhc 策略的混合算法简称为 rhc，而 传统全路径优化的遗传算法为 cdo。由于它们都 使用了相同的 ga 作为航线优化算子，因此将 rhc 和 cdo 进行对比就很公平。关于 ga 优化算子的 更多细节本文不加以赘述。在仿真试验中，除非特 别指定，仿真试验中一个时间段是 10 分钟长，滑 动地平线的长度为 n=12（表 5 对应的试验除外， 因为表 5 专门研究不同 n 的影响），或 2 小时长， rhc 使用公式（21）中定义的终端加权计算 wterm （k）（表 6 对应的试验除外，因为表 6 专门研究不 同终端加权的影响）。 在表 1 中定义了 4 个仿真案例，其自由航行海 域环境的复杂程度都各不相同，dd 表示从始发港 到目的地港的直接距离，ur 表示不可航行区域。 在案例 1-2 中，ur 是静态的，而在案例 -4 中， 其随着时间而产生变化，换言之，他们可以移动， 面积可以发生改变，消失或可能随机出现新的 ur。 此仿真主要用于比较 rhc 和 cdo 的在线计算时间 （oct）及性能，即从始发港到目地港的实际航行 时间（ast）由于篇幅限制，我们仅采用并显示了 整个航行过程中 8 个时间段相关的结果。数值仿真 结果见表 2-表 6，其中对于每个静态案例，rhc 或 cdo 进行了 10 次仿真模拟，而对于每个动态方案 则进行了 200 次仿真模拟。 表 1 仿真试验中的 4 个仿真案例 table 1. four simluation cases 静态环境 动态环境 case 1 case 2 case  case 4 dd (km) 500 1000 500 1000 ur 的数目 1 6 1 6 虽然 rhc 主要应用于带台风等不可航海域的 动态方案，但在静态方案中它也需要能正常运转。 表 2 给出了根据不同算法案例 1-2 的仿真模拟结果， 从中可以看出，在两个案例中 cdo 实现了最好的优 化性能，即取得了最小的 ast。这是因为从理论上 说，在静态案例中像 cdo 这样的传统动态全路径优 化策略本就应该达到最好的性能。表 2 也显示了 rhc 的性能非常接近 cdo，这表示 rhc 在静态案 例中的运行效果不错。关于 oct，rhc 很显然比 cdo 更有效率。由于一个时间段是 10 分钟长，我 们可以看出 rhc 在实时运行方面应该没有问题，而 在一些案例中 cdo 却很难实现在线运行。 我们主要的关注点是带台风等不可航海域的动 态案例，相关的仿真模拟结果见表 。在性能方面， journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 79 对于像案例  和 4 这样相对简单的案例，cdo 和 rhc 的 oct 虽然相差不大，但可以看出 rhc 的性 能要优于 cdo。原因已在第 2 节和第  节中详细阐 述过了。因此，rhc 能提供比 cdo 更加可靠的实 时运算性能。 表 2 静态案例的仿真试验结果 table 2. simuluation results in static cases （单位: 分） cdo rhc 0.1590 1.0489 0.218 0.6079 平均 oct 871.84 1628.28 871.97 161.40 平均 ast 0.5900 4.0975 0.68 0.8099 最大 oct 872.1 164.5 872.52 1644.5 最大 ast 0.1590 1.0489 0.218 0.6079 如表 1 所示，从案例  到案例 4，dd 和 ur 的数值都增加了。那么是 dd 还是 ur 的数值更能 影响 rhc 的 oct 呢？表 4 回答了这个问题，其中 dd 在[500,1000]之间变更，ur 数值在[1,6]之间变 更，而所有案例都是动态的。从表 4 中可以看出， rhc 的 oct 主要取决于 ur（因为 ur 的数值极 大地影响了寻找可选飞行路径和计算终端加权的 ga 优化算子的计算负荷），和 dd 关系不大（因为， 对于 rhc，不是由 dd 而是由 n 来决定一条可选 飞行路径的最长飞行时间）。 表 3 动态态案例的仿真试验结果 table 3. simuluation results in dynamic cases （单位： 分） cdo rhc case  case 4 case  case 4 平均 oct 0.1206 1.1844 0.125 0.4816 平均 ast 928.21 164.0 928.12 168.61 最大 oct 0.581 4.260 0.6449 0.6908 最大 ast 928.6 1866.8 928.45 1757.65 表 4 dd 和 ur’s 对 rhc 性能的影响 table 4. influence of dd and ur on the oct of the rhc oct (分) dd=500 (km) dd=1000 (km) ave. max. ave. max. 1 个 ur 0.125 0.6449 0.77 0.7072 6 个 ur 0.495 0.7455 0.4816 0.6908 表 4 很清楚地表明，应合理地选择 n，即滑动 地平线的长度。如果 n 太小，rhc 的优化性能就 差，如表 5 中 n=1 和 n= 的案例。而如果 n 太大， oct 增加了，但性能并没有进一步提高。相反，在 动态案例中，n 即太大时 rhc 的优化性能还可能 降低，如表 5 中 n=9 时所示。 表 5 滑动地平线的长度 n 对 rhc 性能的影响 table 5. influence of n on the rhc （单位：分） 静态环境 动态环境 case 1 case 2 case  case 4 n=1 oct 0.1045 0.1174 0.0920 0.1061 ast 880.8 541.76 928.89 175.47 n= oct 0.160 0.2445 0.1618 0.182 ast 871.71 1717.02 929.20 1644.8 n=6 oct 0.218 0.6079 0.125 0.4816 ast 871.97 161.40 928.12 168.61 n=9 oct 0.5799 1.290 0.515 1.0750 ast 871.91 1628.5 928.19 1640.9 表 6 显示了 rhc 中终端加权对优化性能的影 响。因为公式（19）中定义的 wterm（k）使算法变 得不稳定，所以表 6 中没有给出相关的结果。基本 上，在 oct 保持在相同水平时，可以看出在分别 使用公式（20），公式（21）中定义的 wterm（k） 后，rhc 的性能一步步地得到了改善。原因已在 . 节中详细阐述了。 表 6 终端加权对 rhc 性能的影响 table 6. influence of terminal weighting on the rhc wterm(k) 静态环境 动态环境 case 1 case 2 case  case 4 公式 (20) oct 2.8102 4.9121 2.8994 .845 ast 904.67 164.44 978.54 1647.80 公式 (21) oct 2.5675 4.8498 2.490 .8419 ast 871.97 161.40 928.12 168.61 下面的图 11 截取了带有单个台风动态不可航行 区域的仿真案例中，船舶避台航线规划过程及结果， 共有 8 张图片展示船舶避台航行的航线优化的过程。 5 结论 船舶海上躲避台风航线优化问题与传统的路径 优化问题不同，船舶在海上躲避台风时的不可航海 域是实时变化的，不可航海域可以是固定的岛礁、 浅滩，也可以是动态变化的台风风圈和其它船舶在 航海域等等，因此，船舶海上躲避台风航线优化问 题要求针对动态变化的不可航海域进行实时航线优 化。本文提出了一种基于遗传算法（ga）和滑动地 平线控制（rhc）策略的混和算法来解决此问题。 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 80 为此目的，本文首先建立了带动态变化不可航海域 备注：图中的横纵坐标代表仿真实验航行海域范围，单 位：km） 图 11 rhc 仿真案例 fig. 11. an example of rhc simulation 的自由海域路径优化问题的数学模型后，然后详细 介绍了混和算法的设计思路，全面探讨了此算法在 如何选择 rhc 的滑动地平线的长度，以及如何使 用终端加权等主要技术问题。仿真试验结果显示， 在缺少台风等动态变化环境因素时，混和算法可以 和现存算法达到一样好的性能，而在带有台风等动 态变化的环境中，混和算法的性能（不论求解质量， 还是求解速度）则优于现存算法，混合算法在解决 船舶海上躲避台风航线优化问题能够做到在保证 船舶安全航行的前提下，更好的实现船舶运营的成 本效益。 参考文献： [1] 史培军. 论灾害研究的理论与实践. 南京大学学报 (自然科学版). 自然灾害研究专辑, 1991(11): 7-42. [2] 张兰生，史培军，方修琦. 我国农业自然灾害灾情分 析 . 北京师范大学学报 ( 自然科学版 ). 1990(0): 94-100. [] 梁必骐，梁经萍，温之平. 中国台风灾害及其影响的 研究. 自然灾害学报. 1995(01): 84-91. [4] 夏施敏. 台风对沿海船舶的影响及防抗措施研究. 中 国水运. 2008(09): 50-51. [5] cormen t h, leiserson c e, rivest r l, et al. introduction to algorithms, second edition. cambridge: mit press and mcgraw-hill, 2001. [6] 韩雪, 刘英舜, 郭唐仪. 城市轨道交通网络中断下的 有效路径搜索模型. 公路交通科技, 2015, 2(10): 97-101. han xue, liu ying-shun, guo tang-yi. a valid path searching model for interrupted urban rail transit network. journal of highway and transportation research and development, 2015, 2(10): 97-101. [7] wickens c d, mavor as, parasuraman r,et al. airspace system integration-the concept of free flight. washington: national academy press, 1998. [8] kahne s. research issues in the transition to free flight. annual reviews in control, 2000, 24(1): 21-29. [9] mclean d. operational requirements for fully automatic flight. aircraft engineering and aerospace technology, 200, 75(6): 570-574. [10] 胡晓明, 李万莉, 朱为国, 等. 半挂液罐车的避障运 动仿真分析. 公路交通科技 201, 0(7): 151-158. hu xiao-ming, li wan-li, zhu wei-guo, et al. simulation of obstacle avoidance motion of liquid tank semi-trailer. journal of highway and transportation research and development, 201, 0(7): 151-158. [11] 刘大刚, 吴兆麟, 李国平. 船舶绕避热带气旋技术方 法述评. 航海技术, 2004(5): 9-11. [12] 刘涛，刘大刚. 目标圆方法在船舶导航中的应用. 气 象科技. 2007, 5(6): 867-87. [1] 林明智, 庄丽, 张楠, 等. 船舶避台技术方法新探. 气 象科技. 1999(4): 49-5. [14] hu x b, wu s f, jiang j. on-line free-flight path optimization based on improved genetic algorithms. engineering applications of artificial intelligence, 2004, 17(8): 897-907. [15] clarke d w. advances in model-based predictive control. new york: oxford university press, 1994. [16] bart d s, ton v d b. model predictive control for max-plus-linear discrete event systems. automatica, 2001, 7(7): 1049-1056. [17] chand s, hsu v n, sethi s. forecast, solution, and rolling horizons in operations management problems: a classified bibliography. manufacturing & service operations management, 2002, 4(4): 25-4. [18] hu x b, chen w h. receding horizon control for aircraft arrival sequencing and scheduling. ieee transaction on intelligent transportation system, 2005, 6(2): 189-197. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 72–81 ___________________________________________________________________________________________________________ 81 http://en.wikipedia.org/wiki/introduction_to_algorithms a study on marine vessels’ path optimization under typhoon scenarios 1 引言 2 船舶在海上自由海域航行时的路径实时优化问题 2.1 可选自由航线 2.2 航线优化的性能指标 3.2 基于遗传算法（ga）的优化程序 3.3 滑动地平线的长度和终端加权 （单位：分） wterm(k) [12] 刘涛，刘大刚. 目标圆方法在船舶导航中的应用. 气象科技. 2007, 35(6): 867-873. atlantis press journal style time limit of the probabilistic risk for natural disaster* jun guo1,2, chongfu huang1,2,† 1 key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing 100875, china 2 academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education, the peoples' republic of china, beijing 100875, china abstract since natural disaster risk is a feature of the future and is also dynamic, the result of probabilistic risk analysis for natural disaster has its own time limit. considering that risk is for the future and meanwhile dynamic, the result of probabilistic risk analysis in natural disaster could just represent risk for a limited time, which is the timeliness of probabilistic risk. besides, the length of the timeliness is just called the period of validity of probabilistic risk for natural disaster. because the system of natural disaster risk is complex and comprehensive, the dynamics of natural disaster risk is caused by numerous reasons related to both natural and social environment and is with great uncertainty. therefore, qualitative analysis could be a feasible method for the timeliness evaluation of probabilistic risk for natural disaster. keywords: natural disaster, probabilistic risk, time limit, period of validity 自然灾害概率风险及其时效性 郭 君 1,2，黄崇福 1,2 1. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875，中国 2. 民政部/教育部减灾与应急管理研究院，北京 100875，中国 摘 要：由自然和社会组成的自然灾害风险系统是一个变化的系统，因此不存在一劳永逸的风险分析结果。风 险分析结果的有效性只能维持一定的时间长度，即风险分析结果具有时效性。概率分析方法作为一种常用的自 然灾害风险定量分析方法，其结果，必然具有时效性。考虑到风险是面向未来的和动态变化的，在具体的研究 目标设定下，概率风险仅能表征未来一定时间长度的灾害风险，该特性即为自然灾害概率风险的时效性，该特 定的时间长度称为其有效期。由于自然灾害风险系统的复杂性和综合性，造成自然灾害风险动态变化的原因众 多、存在着极大的不确定性，通过定性分析出相关因素发生较大变化所需的时间长度并进一步选定其中最快因 素的时间长度，作为自然灾害概率风险有效期的最大值，可为自然灾害概率风险的时效性评价提供可行办法。 关键词：自然灾害，概率风险，时效性，有效期 *this project was supported by the national natural science foundation of china (41471426,41671502). †corresponding author: e-mail: hchongfu@bnu.edu.cn. post address: academy of disaster reduction and emergency management ,beijing normal university, no.19 xinjiekouwai street,beijing 100875, china journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 137 received 14 june 2017 accepted 21 july 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 1. 引言 自然灾害是由自然事件或力量为主因造成的生 命伤亡和人类社会财产损失的事件。由于未来的自 然灾害难以精确预测，对其进行的不确定意义下的 量化分析，称为自然灾害风险分析。由自然和社会 组成的自然灾害风险系统是一个变化的系统，因此 不存在一劳永逸的风险分析结果，风险分析结果的 有效只能维持一定的时间长度，即风险分析结果是 有时效性的。例如，假定存在某种较为可靠的方法， 在 2000 年能分析出四川省汶川县映秀镇未来十年面 对的地震风险，即 2001 年到 2010 年内的地震风险。 那么，2008 年汶川大震后，在 2000 年分析出的地震 风险，将不再适用于 2009 年和 2010 年。也就是说， 当地的地震灾害系统发生了巨大变化，人们曾经分 析出的风险结论必然失效。 如今，人们耗巨资进行了从全球尺度到村落尺 度的大量自然灾害风险分析工作，绘制了相应的自 然灾害风险图，然而很少有制作者能向用户提供其 风险图保持有效的时间长度也即其有效期。如同失 效的食品会影响食用者的身体健康，失效的自然灾 害风险图也会给使用者带来额外的风险。 在实际中，为淘汰陈旧的自然灾害风险图，一 些研究机构或人员会对其进行更新。例如，bündnis entwicklung hilft 公司与联合国大学环境与人类安全 研究所从 2011 年开始每年发布一个世界灾害风险指 数（world risk index）报告，内容包括世界上 170 多个国家由暴露度、敏感性、应对能力和适应性的 多个指标构成的灾害风险指数分布图 [1] ，每年根据 获得的最新数据情况对相应的指标进行更新，获得 新的世界风险指数分布图；美国地质勘探局（usgs） 根据最新的研究成果、观察资料等，从 1976 年美国 第一张地震危险性概率区划图发布以来，分别于 1990 年、1996 年、2002 年、2008 年、2014 年和 2016 年对其进行了更新[2]。在我国，从新中国成立 以来到现在，随着地震台网观测数据的积累、理论 方法研究的深入及重大地震灾害的催生，地震区划 图于 1957 年、1977 年、1992 年、2001 年、2016 年 分别进行了五次更新，第五代区划图于 2016 年 6 月 1 日起在全国正式实施；而为了适应 2008 年汶川特 大地震后的恢复重建工作的需要，基于对地震灾区 的新资料和新认识，特就汶川地震灾区的地震动参 数区划图进行了局部更新 [3] 。同时，也有学者研究 了不同时段的风险的确存在差异。例如，庞西磊依 据不同的五年计划来划分历史数据，分析显示“七 五”到“十二五”期间汕头市城市人口面临的暴雨 洪涝灾害风险存在差异 [4, 5] ；蒋卫国等分析的马来西 亚吉兰丹州洪水灾害风险指数分布在 1990 年和 2000 年是不同的[6]，路明浩指出安徽省沿江地区 1990、2000 和 2007 年的风险指数分布存在差异[7]， 苏飞等研究表明 1979、1990、2000 和 2009 年在不 同洪水强度下黄浦江流域的洪灾风险等级显著不同 [8] 。 上述实践工作和理论研究表明了风险分析结果 存在时效性。同时，也有部分研究者关注了度量灾 害风险时效的问题。例如，苏桂武和高庆华依据风 险载体的“价值”存在时间期限，定性给出了不同 载体的灾害风险有效时间尺度 [9] ；赵思健概述了综 合考虑致灾因子、孕灾环境、承灾体的有效时间长 度而确定自然灾害风险评估的有效时间广度，并定 性分析了不同灾种的风险有效时间长度 [10, 11] 。 然而，已有研究并没有给出时效性的具体研究。 由于自然灾害风险分析的方法众多，从普通预测方 法到复杂系统分析法，从概率分析法到层次分析法， 不同方法依据的原理不同，得到的风险分析结果所 具有的时效性的具体内涵必然是不同的，而现有的 研究只是模糊地、一概而论地表达为自然灾害风险 的有效时间长度等概念。在众多方法中，概率分析 法是自然灾害风险分析中最常用的定量分析方法， 主要原因是自然灾害事件大量出现，可将自然灾害 看做随机事件继而采用概率统计方法而得到灾害事 件统计特征值来表征风险 [12] ，尤其是用期望损失值 作为风险值，具有明确的物理意义。 鉴于此，本文就概率风险的时效性展开研究， 以期为风险更新等动态风险管理提供理论参考。首 先，本文从风险的系统性出发，介绍了自然灾害概 率风险的形式化模型和应用模型；继而，界定了自 然灾害概率风险的时效性内涵；最后，给出了自然 灾害概率风险时效性评价的形式化表达。 2. 自然灾害概率风险 根据其数学形式的具体化程度，风险分析模型 可分为“概念化模型”、“形式化模型”和“应用 模型” [12] ，如下从上述三个层次对自然灾害概率风 险进行分别介绍。 2.1. 自然灾害风险的概念化模型 从系统论出发，自然灾害风险是风险源和风险 承受体综合作用的结果 [12-14] 。 风险源是指引发损害（包括人员伤亡、经济损 失、物体破坏等）的原因，构成了风险的诱因，在 自然灾害领域通常称为致灾因子。为了突出风险的 研究主体，本文统一称作风险源。自然灾害风险源 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 138 是可能造成损害的自然事件、力量或过程 [15] ，是自 然灾害风险是否存在的先决条件。例如，《国家综 合减灾“十一五”规划》中列出了我国常发生的 13 种 自然灾害风险源：洪涝、干旱、台风、风雹、雷电、 高温热浪、沙尘暴、地震、地质灾害、风暴潮、赤 潮、森林草原火灾和植物森林病虫害 [16] 。 风险承受体是风险源的作用对象和受体，存在 风险源但没有风险承受体时不能构成风险，在自然 灾害领域通常称为承灾体。同样地，本文统一称作 风险承受体。概括地，自然灾害风险承受体包括人 类自身及其构筑物，例如人口、财产、建筑物、农 作物、基础设施以及地域综合体等 [13] 。 风险源和风险承受体在灾害风险中的各自作用 和特性，决定了两者成为自然灾害风险分析的两个 基本组成成分，自然灾害风险的基本模式 [12] 可以概 括为 [17-19] ： 𝑅 = 𝐻 ∘ 𝐷 ( 1 ) 其中 h 代表风险源、d 代表风险承受体、“∘”表示 合成方法。这一基本模式，是一个概念化模型。 采用不同的方法确定 h、d 和“∘”，可以得到不同 的自然灾害风险分析的形式化模型。例如，指标体 系法，通过选取 h 和 d 的多个表征指标，使用加权 综合等方式作为合成方法“∘”；而概率分析方法， 利用风险源的发生规律表示 h、风险承受体的脆弱 性表示 d，采用概率统计方法获取 h 和（或）d， 使用微分求和等方式作为合成方法“∘”。 需要说明的是，通常表述中的“自然灾害风险” 指的是一定区域的自然灾害风险。自然灾害的发生 与发展须依附于一定的地理单元 [20] ，也就是说，提 到“干旱、台风、冰雹的发生规律”等，必然是指 某一地区的干旱、台风、冰雹的发生规律。同时， 自然灾害风险承受体的分布和抗灾能力也有着明显 的空间差异，同样地，当提到“小麦的干旱风险、 建筑物的地震风险”等时，必然是针对某一地区的 上述承受体的风险。也就是说，采用某种方法评估 得到的自然灾害风险，一定是针对某一地理单元的， 因此自然灾害概率风险也必然如此。由于人们对自 然灾害风险的地理属性共识性较高，因此在表述中 往往省略区域的限定，直接表达为某自然灾害风险， 本文也是采用默认的方式表述自然灾害概率风险。 也就是说，本文中的自然灾害风险、自然灾害概率 风险等均指某地区的风险。 2.2. 概率风险的形式化模型 由于自然灾害事件时而发生、时而不发生的外 在表现，自然灾害事件可以被看作随机事件 [21] ，用 于描述随机现象的概率论等数理统计方法被用于自 然灾害风险的评估中。自然灾害概率风险评估方法 的出发点是将自然灾害事件看作随机事件，采用概 率统计等数理方法研究灾害事件的相关规律，最后 使用某些统计指标作为风险的度量，如超越概率、 期望损失值、最大可能损失值等 [22-25] 。 基于基本模式( 1 )，利用风险源的发生规律 p(x)表示 h、风险承受体的脆弱性 f(x)表征 d，一般地选用损 失程度的期望值为内涵，则积分或累加求和等方式 作为合成方法“∘”，如此可得到概率风险的形式化 模型（见图 1）： ∫= 2 1 )()( u u dxxpxfr （2）式中，风险 r 的内涵为灾害损失的期望值；x 为表征风险源强度 的因子；p(x)为表征风险源发生强度取值的概率分 布函数，通过使用历史灾害记录集构成的灾害强度 样本集估计得到；f(x)为表征风险承受体的脆弱性 函数，可以理解为灾害强度与损失的输入-输出函数 关系，使用灾情记录集的灾害强度-损失数据对构成 的样本集进行识别；u1、u2 为 x 的积分上、下限。 概率 1 风险源强度 0 风险源发生规律 风险承受体脆弱性 灾损 图 1 自然灾害概率风险的形式化模型示意图[12] 3. 自然灾害概率风险的时效性界定 自然灾害风险是动态变化的，尤其是风险承受 体的变化不可忽视。因此，基于历史记录得到的概 率风险只能代表未来一定时间长度内的风险，即自 然灾害概率风险具有时效性。由于风险的动态变化 是自然灾害概率风险存在时效性的原因，这里先对 风险的动态变化进行介绍，继而再说明概率风险的 时效性概念。 3.1. 自然灾害风险的动态变化性 自然灾害风险的动态变化性是由其组成成分即 风险源和风险承受体相关特性的动态变化决定，其 中与人类主观能动性直接相关的风险承受体是变化 速率最快的成分 [26-30] 。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 139 3.1.1. 风险源发生规律的动态变化性 尽管自然灾害风险源由自然环境的演变规律决 定，然而人类的足迹早已布满世界的各个角落，现 在的自然已远远偏离原始的自然，是人化的自然， 人类活动对自然环境的影响不容置疑。由包括人类 活动在内的生物过程引起的全球变化问题，在 1982 年首次提出之时便得到广泛响应，至今仍然是研究 的焦点之一。尤其是在人类活动剧烈的几十年到几 百年时间尺度上的全球变化问题，由于与人类社会 可持续发展密切相关，更是研究的重点 [31] 。 全球变化问题包括全球气候变化，如温室气体 的增加、臭氧层空洞、酸雨增多等问题；也包括全 球地表覆盖变化，如森林面积的减少、湿地的减少、 人为修筑物覆盖面积的增加，等等。全球气候变化 会导致极端天气灾害事件的发生频率出现变化，而 全球地表覆盖变化可能会影响滑坡、泥石流等地质 灾害的发生规律等。 3.1.2. 风险承受体脆弱性的动态变化性 人类活动的诸多方面都会影响风险承受体的脆 弱性，如社会经济发展 [32] 、土地利用布局的改变 [33] 、 执行建筑规范的变革、公民责任感的增强、保险制 度的实施、所用材料和生产技术的改进等 [34] ，尤其 是有效防灾减灾措施 [27, 35] 更是人类为降低脆弱性而 主动采取的举措，因此风险承受体的脆弱性是动态 变化的。 防灾减灾措施可以分为两类，即工程性措施和 非工程性措施，兵库会议报告中也称为结构性措施 和非结构性措施 [36] 。结构性措施指为减少或避免危 害可能带来的影响而进行有形实物的修筑或技术的 实施行为，包括工程项目以及抵御和防御危害的结 构和基础设施等。例如，人工加固边坡措施，如设 置坡脚护墙、覆盖护网等阻止边坡滑动；建筑物的 抗震、抗风加固结构的修筑等；人工影响天气技术， 如人工增雨雪、防雹、消雨、消雾、防霜等措施。 非结构措施指从政策、认识、知识开发、公众承诺、 方法、机制等方面开展的防灾减灾行动，例如面对 居民的灾害风险知识的宣传教育和灾害应急能力培 训和灾害预警机制的完善等。 3.1.3. 自然灾害风险动态变化的特点 由于风险的未来性和不确定性，无法得到客观 的灾害风险。基于历史记录使用概率分析方法等得 到的概率风险结果是对客观灾害风险的一种近似估 计，其潜在前提假设便是该结果能够表征未来的灾 害风险（准确地表达应该为“能够表征风险”，因 为风险本身就是未来的，但为了突出其未来性，本 文采用“未来的风险”等术语，特此说明。）。 自然灾害风险的动态变化性是必然的，从其对 概率风险表征客观灾害风险的有效性的影响效果来 看，自然灾害风险的动态变化具有以下两个特点： （1）风险动态变化的凸显需要一定的时间长度 理论上，自然灾害风险系统的变化可能时时刻 刻存在。然而，鉴于客观灾害风险的不可准确获得 性和概率风险计算结果的近似性，自然灾害风险的 动态变化需要一定时间长度才能在概率风险分析结 果上凸显，这也是概率风险分析方法有意义的前提 基础。 如前所述，灾害事件具有低发生率性，为了获 得研究灾害事件统计规律的样本数据，需使用过去 一段时间长度内的灾害事件发生记录来求解其统计 特征值，以此来表征灾害风险。由于风险是未来的 又是动态变化的，概率风险有意义的前提假设便是 历史记录所处的过去、当前以及要表征的未来三个 时间段内灾害风险系统认为是相同或相似的。也就 是，使用历史记录获得的灾害事件统计规律可以延 展应用到未来一定时间尺度。由于概率风险采用灾 害事件的统计特征值来表征风险，该潜在假设的成 立便意味着灾害风险的动态变化需要一定时间长度 才能在概率风险分析结果上凸显。事实上，灾害风 险动态变化的上述特点是成立的。 一方面，由于自然环境演变过程缓慢，自然灾 害风险源发生规律的变化往往需要较长的时间尺度。 另一方面，相对来说，由人类自身决定或由人类主 导的自然灾害风险承受体的变化速率非常快，而且 人类活动的干预往往是显著性的，因此风险承受体 的脆弱性会在相对较短的时间内发生显著的变化。 例如，当一些措施以集中性、全面实施时，脆弱性 的变化可能会以突变的形式出现，例如灾害预警机 制可以在较短的时间得到全面实施，进而引起区域 脆弱性函数出现突变 [37] ；而当一些措施的实施需要 一定的时间时，脆弱性的变化则也需要一定的时间 长度，例如新的建筑物材料在一个区域的全面普及 需要一定的时间长度，那么经过该时间长度的推广 后，区域建筑物的地震脆弱性则会发生显著变化， 等等。然而，无论是突变还是渐变的形式，风险承 受体脆弱性的变化必然需要一定的时间长度。也就 是说，不管是缓慢的风险源发生规律的变化还是快 速的风险承受体脆弱性的变化，都需要一段时间才 能凸显，那么在该段时间内灾害风险系统是可以看 做不变的。也即一个灾害系统的状态会持续一定的 时间长度，因此在近似不变的时间长度内，基于历 史记录得到的概率风险可以表征未来一定时间尺度 的风险。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 140 （2）自然灾害风险系统动态变化的不可逆性 在人类可持续发展关注的时间尺度即几十年到 几百年内，可以认为灾害风险系统是无后效性，即 阶段 i 的状态只与阶段 i-1 的状态有关，而与其他状 态无关。 自然灾害概率风险的研究目的是为人类社会发 展的相关决策、规划等提供科学依据，例如研究区 域灾害损失的均值表征的概率风险可为政府的救灾 金储备提供依据，往往人类社会发展关注的时间尺 度短则为几年、长可至几百年，则该时间长度便是 自然灾害概率风险研究关注的时间长度。在几年到 几百年的时间长度内，自然灾害风险的动态变化可 认为不可逆的，一则在人类认知范围内灾害发生和 发展的自然孕育环境早已远远偏离了原始状态，二 则风险承受体所处的人类社会环境的变化更是翻天 覆地。 上述自然灾害风险动态变化的不可逆性，一方 面决定了可用于评估概率风险的历史资料不可能过 于久远，因为太久远的历史资料用作当前概率风险 评估样本的有效性值得斟酌，例如我国改革开放前 的灾害经济损失记录用于当前灾害风险分析的合理 性值得怀疑，该问题便是用于概率风险分析的历史 资料有效性问题 [37] ；另一方面决定了评估得到的自 然灾害概率风险结果只能表征距离当前最近的一定 时间长度未来的风险，不再考虑更久远的未来，也 就是说更久远的未来必然与当前的自然灾害风险系 统不同，该问题便是本研究关注的自然灾害概率风 险的时效性问题。 3.2. 概率风险的时效性 在新华汉语词典中，时效的解释为“在一定时 间内能起的作用”，并列出常用的表达为“时效长、 失去时效” [38] 。根据列出举出的常用表达以及在实 际中的用法，“时效”有两层具体含义：一是“在 一定时间能起的作用”，常用在“保证时效、失去 时效、具有时效”等表达中；二是“能起作用的时 间”，用于“时效长或短”等表达中。事实上，时 效的两层含义是其分别侧重于时间和作用的两种表 现，当然两者均同时包括随时间动态变化的特性和 围绕某一目标的作用两个方面。 关于时效性的含义，词典并没有明确给出解释， 然而可以确定的是时效性表达是一个特性。这里， 根据“时效”的两层意思，可以确定出时效性的两 个具体含义：第一，表示“在一定时间内能起作用” 的性质，用于“快递、物流服务要保证时效性、不 能失去时效性”等表达中，该含义侧重于对象应能 起的作用；第二，表示“作用的发挥具有一定的时 间限度”的性质，用于“新闻具有时效性、政策具 有时效性、时效性长”等表达中，该含义侧重对象 起作用的时间有限。 如前所述，由于风险的未来性和不确定性，自 然灾害概率风险的潜在假设是基于历史记录得到的 概率风险结果能够表征未来的灾害风险。与此同时， 自然灾害风险又是动态变化的，而且自然灾害风险 的动态变化需经过一段时间才会在概率风险上凸显 出来，因此可以认为自然灾害概率风险能够表征未 来有限时间长度的灾害风险，超过该时间长度则认 为概率风险失效。上述自然灾害概率风险与时间和 对灾害风险表征的效用两个方面相关的特性，正符 合时效性的第二层含义，因此称自然灾害概率风险 具有时效性。 本质上，造成事物或现象具有时效性的根本原 因可分为两类。一类是事物本身的生化、物理特性 或者内部组分随着时间发生了变化，继而导致事物 的效用逐渐降低直至消失。例如，药品或食品存在 时效性是因为其所包含成分具有不稳定性，导致在 放置的条件下其成分会随时间发生变化，进而导致 药品的效力或食品的食用功能逐渐丧失。另一类是 事物或现象所表征的对象随着时间发生了变化，继 而导致事物或现象设计时被用来承担的作用变得减 弱甚至丧失。例如，污染源监测报告具有时效性是 因为污染源状况是随时间变化的、而监测报告是静 态的，所以随着时间的变化已有的监测报告逐渐不 能继续用来有效表征污染源状况，进而失去其应有 的功能。 自然灾害概率风险存在时效性的原因属于第二 类。由于概率风险是根据一定时间的历史记录所计 算出的静态结果，其所刻画的自然灾害风险是动态 变化的，因此随着时间的变化而导致概率风险不能 继续有效表征灾害风险，也即失去应有的效用。 本文界定自然灾害概率风险时效性的内涵为： 自然灾害风险是动态变化的，针对具体的设定目标， 基于历史记录的自然灾害概率风险具有仅能有效表 征未来一定时间内的灾害风险的特性。此外，根据 “有效”的词典解释即“能达到预期目的或有效果” [39] ，有效期是指事物或现象改变到某一给定值而导 致其原有的效果丧失所需的时间长度，特将自然灾 害概率风险时效性中的时间长度称为自然灾害概率 风险的有效期。在本研究中，自然灾害概率风险时 效性评价是指对其有效期的估计，即估计“能起作 用的时间”。根据“时效”的第二层含义，也可称 为“时效评价”。然而为了不引起误解，本研究使 用“估计有效期”来表示自然灾害概率风险时效性 评价的任务。 需要指出的是，本文研究的自然灾害概率风险 时效性针对的是能够通过合适的历史记录、合理的 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 141 分析方法而得到的满足应用目标的概率风险结果， 例如用于保险业务的期望损失情况等表征的概率风 险分析结果。也就是说，本文假定能够得到这样一 个满足应用需求的概率风险，并不研究通过哪些资 料和方法可以得到满足要求的概率风险分析结果， 而是假定得到的概率风险能够满足要求。此外，本 文针对的概率风险也不是具体的分析结果，因为理 论上存在着满足应用目标的概率风险分析结果在风 险源发生规律和风险承受体脆弱性动态变化前后的 是一样的情况，然而根据自然灾害风险动态变化的 不可逆性，本文并不指具体的概率风险分析结果值， 而是指其中任一组分的动态变化引起灾害风险的动 态变化而造成概率风险的时效性问题。 4. 自然灾害概率风险时效性评价 本质上，药物的有效期是指药物的成分变化到 一定阶段而失去药用价值时所需的时间长度，污染 源监测报告的有效期是指污染源状况变化到与报告 内容相差较大而报告不能用于指示当前状况时所需 的时间长度。与之相似，自然灾害概率风险的有效 期便是指自然灾害风险发生变化到一定程度而导致 分析结果失去对灾害风险表征的有效性时所需的时 间长度。分析灾害风险动态变化的时间演变情况可 得到自然灾害概率风险的有效期，进而完成自然灾 害概率风险的时效性评价。 4.1. 时效性评价的形式化表达 如前所述，自然灾害概率风险的时效性评价便 是确定概率风险的有效期，自然灾害风险动态变化 到一定程度所需的时间长度便是自然灾害概率风险 的有效期。因此，自然灾害概率风险有效期的长短 由自然灾害风险的变化速率决定。当自然灾害风险 系统所涉及的综合环境或（和）内在属性变化速率 快时，基于历史记录评估得到的概率风险与客观风 险的偏离速度快，则概率风险的有效期短；反之， 概率风险的有效期长。如此，概率风险有效期的形 式化公式为： 0 er rt v − = ( 2 ) 其中：t 表示概率风险的有效期； r0 概率风险评估时灾害风险的状态； re 表示灾害风险出现较大变化时的状态； v 表示灾害风险动态变化的速率。 从具有有效期的典型物品即医用药物来看，单 一成分的医用药物具有有效期的原因是生产出来的 药品在一定储存条件下，随着时间的流失内部成分 发生变化，导致能起药效的成分减小甚至完全消失 而失去效用。因此，单一成分药物的有效期便对应 着起药效的成分变化到某一临界状态所需的时间长 度。进一步地，对于多成分的医用药物，其效用是 由多成分共同决定，药物失效的原因可能是一种成 分变化到一定程度或者若干个成分同时变化到各自 的某个程度。此时，导致药物具有时效性的情况更 加复杂，其有效期的确定也变得困难。类比来看， 由于自然灾害风险是风险源和风险承受体的综合结 果，自然灾害概率风险具有时效性的原因与多成分 的医用药物类似，其涉及多个组分的动态变化情况， 即式( 2 )可具体为： 0 0( , )s se o oe s o i i i i t c v v − − = ( 3 ) 其中：t 表示概率风险的有效期； is0 概率风险评估时风险源的状态；ise 表示风 险源出现较大变化时的状态；vs 表示风险源 动态变化的速率； io0 概率风险评估时风险承受体的状态；ioe 表示风险承受体出现较大变化时的状态；vo 表示风险承受体动态变化的速率； c 表示导致灾害风险出现较大变化的风险源 和风险承受体动态变化的各个可能情况的组 合。 再进一步分析，引起风险源动态变化的原因涉 及全球气候或地表覆盖变化等自然方面，而造成风 险承受体脆弱性变化的原因则涉及生产材料（如建 筑物材料、农作物品种等）、生产技术（房屋结构、 农作物种植技术等）和多方面的防灾减灾措施（灾 前预警能力、灾中应对、灾后救援等）等社会方面。 因此，式( 3 )可分解为更具体组分的动态变化情况， 即： 𝑇 = 𝐶( 𝐼10 − 𝐼1𝑒 𝜈1 , … , 𝐼𝑖0 − 𝐼𝑖𝑒 𝜈𝑖 , … , 𝐼𝑛0 − 𝐼𝑛𝑒 𝜈𝑛 ) ( 4 ) 其中：t 表示概率风险的有效期； ii0 概率风险评估时某组分状态；iie 表示该组 分出现较大变化时的状态；vi 表示该组分变 化的速率； c 表示导致灾害风险出现较大变化的各个组 分动态变化的不同情况组合。 依据式( 4 )确定概率风险的有效期，需确定其中 的各个参数，即：第一，确定造成风险动态变化的 诸组分即因素，也即确定 n；第二，确定每个因素 的变化情况，包括变化的速率 vi、初始状态 ii0 和出 现较大变化时的状态 iie；第三，确定导致灾害风险 出现较大变化的各组分变化情况的组合 c。 如前所述，自然灾害风险系统涉及复杂的自然社会系统，造成灾害风险动态变化的因素较多，因 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 142 此 n 的大小难以精确获得。此外，关乎自然（或） 和社会系统的因素的动态变化难以准确确定，尤其 是属于社会系统的因素，如前所述某些因素的变化 形式可能以突变的形式呈现，此时要建立因素的动 态变化状态方程，需考虑使用法国数学家勒内·托 姆的突变理论 [40, 41] 进行构建，其复杂程度可想而知， 也即各因素的变化速率 vi 等难以确定。而因素的多 样性导致的复杂性和诸因素自身动态变化情况的难 确定性等，决定了组合形式 c 的难获知性。 由此可知，无法得到自然灾害概率风险时效性 评价的具体数学模型，也无法采用试验法确定概率 风险的时效性，因此无法准确评价自然灾害概率风 险的时效性，事实上也无需给出一个准确的评价结 果 [42] 。那么，在上述情形下，如何合理地开展自然 灾害概率风险的时效性评价，为概率风险更新等动 态风险管理工作提供理论参考呢？该问题便是本研 究的关注点。 4.2. 时效性评价的一个应用模型 考虑到自然灾害概率风险时效性评价的高复杂 性以及大致准确性，通过借助人的智慧分析、综合、 判断认知思维，对造成自然灾害风险动态变化的诸 因素的变化情况进行判断，定性分析各因素出现较 大变化的时间长度，找到最关键和最快因素的变化 时间长度，作为自然灾害概率风险有效期不大于的 时间长度而得到概率风险的区间限制，可作为自然 灾害概率风险时效性评价的可行方法，为概率风险 更新等工作的具体开展提供时机选择的理论参考。 在条件允许、风险更新的成本不高的情况下，在概 率风险有效期的区间范围内，更新越及时越好；在 更新成本高时，概率风险的有效期的区间范围可作 为更新时间间隔的时间范围限定参考，更新的时间 长度不可长于区间的最大值。 之所以选择最快的因素是因为在一定程度上可 以认为造成灾害风险动态变化的因素是独立的或者 弱相关的，或者说各因素有着自身动态变化的主要 动力。具体来看： （1）作为风险源的灾害事件的自然孕育规律的 变化，在灾害风险管理关注的时间尺度内是由自然 环境的演变主导，人类活动的影响占次要地位。例 如，农作物品种的变化可认为不会改变暴雨灾害的 自然孕育规律。当然，某些人类活动会影响风险源 发生规律，例如人类对地表植被覆盖的改造活动是 通过改变大气中二氧化碳的含量而间接引起极端天 气事件的自然孕育规律的变化。 （2）与社会系统相关的风险承受体脆弱性的变 化，可认为主要由社会发展决定，当然也可能受风 险源发生规律变化的影响。例如，极端降雨事件变 得频繁会促使人类更加天气学问题而可能加快应对 洪涝能力的提高速率，然而人类对洪涝灾害的应对 能力更大程度上受社会经济发展水平和科学认识水 平的制约。 （3）不同因素之间也可能存在一定程度上的弱 相关，然而各个因素的变化有着自身的主要动力。 例如人工影响天气技术的增强可能造成人类应对极 端天气事件的意愿降低，然而人工影响天气技术的 改变主要由相关技术的研究进展决定，而人类应对 灾害事件的意愿由灾害认知水平、求生意愿等自身 条件决定。 因此，在因素之间相互独立或者弱相关的基础 上，由于其中某一或者若干因素的动态变化便会引 起的自然灾害概率风险的失效，因此自然灾害概率 风险的有效期必然不大于其中变化最快的因素发生 较大变化所需的时间长度，如此可得到式( 4 )的一个 应用模型： 𝑇 ≤ min (𝑇𝐼1, 𝑇𝐼2, … , 𝑇𝐼𝑖, … , 𝑇𝐼𝑛) (5) 其中：t 表示概率风险的有效期； tii 表示组分 ii 发生较大变化所需的时间长度， 可借助相关人员的认识定性分析而得到。 如此，式( 5 )的应用模型可作为自然灾害概率风 险时效性评价的一个可行方法，获得自然灾害概率 风险有效期的时间范围限制，为风险更新等动态风 险管理工作提供理论值参考。 5. 结论 自然灾害概率风险分析方法的出发点是将灾害 事件的发生与否看作随机事件，将历史灾害事件记 录作为样本，采用概率统计等数理方法获得灾害事 件的统计特征值来表征风险。从自然灾害风险的系 统性出发，自然灾害概率风险是风险源发生规律和 风险承受体脆弱性的综合作用结果。 由于其组成成分的动态变化性，自然灾害风险 是动态变化的，继而导致基于历史记录的自然灾害 概率风险具有只能在一段时间内能够有效表征风险 的特性，超过该段时间则可认为概率风险失效，该 特性便为概率风险的时效性，该段时间长度称为其 有效期。 自然灾害概率风险的时效性评价便是对其有效 期进行估计，也即估计灾害风险动态变化情况所需 的时间长度。由于引起自然灾害风险动态变化的因 素众多而且存在很大的不确定性，定性分析出相关 因素发生较大变化所需的时间长度并进一步选定其 中最关键和最快因素的时间长度，作为自然灾害概 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 143 率风险有效期不大于的时间长度而得到有效期的区 间范围，可为自然灾害概率风险的时效性评价提供 可行办法。 参考文献 [1] unu-ehs (united nations universityinstitute for environment and human security) world risk report. http://weltrisikobericht.de/english/. [2] 美 国 地 质 勘 探 局 (united states geological survey). http://earthquake.usgs.gov/hazards/products/conterminou s/. [3]高孟潭, 陈国星, 谢富仁, 等. 《四川、甘肃、陕西部分地 区地震动参数区划图》编制. 国际地震动态, 2008, (6): 9-12. [4]庞西磊. 自然灾害动态风险分析基本模式的探讨及其应 用研究. 北京: 北京师范大学, 2012. [5]庞西磊, 黄崇福, 张英菊. 自然灾害动态风险评估的一种 基本模式. 灾害学, 2016, 31(1): 1-6. [6]蒋卫国, 盛绍学, 朱晓华, 等. 区域洪水灾害风险格局演变 分析——以马来西亚吉兰丹州为例. 地理研究, 2008, 27(3): 502-508+727. [7]路明浩. 安徽沿江地区洪涝灾害风险时空演变研究[d]. 芜湖: 安徽师范大学, 2011. [8]苏飞, 殷杰, 尹占娥, 等. 黄浦江流域洪灾动态风险演化趋 势研究. 地理科学, 2014, 34(5): 621-626. [9]苏桂武, 高庆华. 自然灾害风险的行为主体特性与时间尺 度问题. 自然灾害学报, 2003, 12(1): 9-16. [10]赵思健. 自然灾害风险分析的时空尺度初探. 灾害学, 2012, 27(2): 1-6+18. [11]赵思健. 基于情景的自然灾害风险时空差异多维表达框 架. 自然灾害学报, 2013, 22(1): 10-18. [12]黄崇福. 自然灾害风险分析与管理. 北京: 科学出版社, 2012: 32, 42, 63-64, 73-74, 130, 208, 215. [13]coppola d p. introduction to international disaster management. 3rd ed. oxford: butterworthheinemann elsevier, 2015: 150. [14]温家洪, 黄蕙, 陈珂, 等. 基于社区的台风灾害概率风险 评估——以上海市杨浦区富禄里居委地区为例. 地理 科学, 2012, 32(3): 348-355. [15]undp (united nations development programme). reducing disaster risk: a challenge for development. new york: bureau for crisis prevention and recovery, 2004. [16] 国 务 院 办 公 厅 . 国 家 综 合 减 灾 “ 十 一 五 ” 规 划 . http://jnjd.mca.gov.cn/article/zyjd/zhxxy/201004/201004 00069153.shtml. [17]kron w. flood risk = hazard • values • vulnerability. water international, 2005, 30(1): 58-68. [18]马保成. 自然灾害风险定义及其表征方法. 灾害学, 2015, 30(2): 16-20. [19]heneka p, hofherr t, ruck b, et al. winter storm risk of residential structures-model development and application to the german state of baden-wurttemberg. natural hazards and earth system sciences, 2006, 6(5): 721-733. [20]郭君, 黄崇福. 自然灾害风险研究进展的一个调查. 风险 分析和危机反应的创新理论和方法---中国灾害防御协 会风险分析专业委员会第五届年会论文集(南京， 2012 年 10 月 27-28 日)，黄崇福，翟国方编，巴黎， atlantis 出版社， 2012: 168-176. [21]chai b, tang z h, zhang a, et al. an uncertainty method for probabilistic analysis of buildings impacted by rockfall in a limestone quarry in fengshan, southwestern china. rock mechanics and rock engineering, 2015, 48(5): 1981-1996. [22]cardona o d, ordaz m g, reinoso e, et al.capra comprehensive approach to probabilistic risk assessment: international initiative for risk management effectiveness. 15th world conference on earthquake engineering", lisbon, portugal. 1-10. [23]joyette a r, nurse l a, pulwarty r s. disaster risk insurance and catastrophe models in risk-prone small caribbean islands. disasters, 2015, 39(3): 467-492. [24]赵思健, 张峭, 王克. 农业生产风险评估方法评述与比较. 灾害学, 2015, 30(3): 131-139. [25]郭君, 黄崇福, 艾福利. 与月份及预警有关的广东省台风 动态风险研究. 系统工程理论与实践, 2015, 35(6): 1604-1616. [26]dewan a. floods in a megacity: geospatial techniques in assessing hazards, risk and vulnerability. dordrecht, heidelberg: springer geography, 2013: 35-74. [27]jongman b, winsemius h c, aerts j c, et al. declining vulnerability to river floods and the global benefits of adaptation. proceedings of the national academy of sciences of the united states of america, 2015, 112(18): 2271-2280. [28]mitchell j k. megacities and natural disasters a comparative analysis. geojournal, 1999, 49(2): 137-142. [29]郭跃. 自然灾害的风险特征及风险管理模型的探讨. 水 土保持研究, 2006, 13(4): 15-18. [30]chang s e, gregorian m, pathman k, et al. urban growth and long-term changes in natural hazard risk. environment and planning a, 2012, 44(4): 989-1008. [31]孙鼎国, 王杰. 西方思想 3000 年. 下册. 北京: 九洲图书 出版社, 1998: 1516. [32]kellenberg d k, mobarak a m. does rising income increase or decrease damage risk from natural disasters?. journal of urban economics, 2008, 63(3): 788-802. [33]zhai g f, li s s, chen j. reducing urban disaster risk by improving resilience in china from a planning perspective. human and ecological risk assessment, 2015, 21(5): 1206-1217. [34]gfdrr (global facility for disaster reduction and recovery). understanding risk in an evolving world, emerging best practices in natural disaster risk journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 144 assessment. washington dc: the world bank, 2014. [35]pappenberger f, cloke h l, parker d j, et al. the monetary benefit of early flood warnings in europe. environmental science & policy, 2015, 51: 278-291. [36]联合国减灾署(unisdr). 加强国家和社区的抗灾能力: 2005--2015 年兵库行动纲领. http://www.un.org/zh/humanitarian/disaster/hfa.shtml. [37]guo j, zhao s, huang c. valid historical data for probabilistic risk analysis in natural disaster. human and ecological risk assessment: an international journal, 2017, 23(3): 474-493. [38]任超奇. 新华汉语词典. 武汉: 崇文书局, 2006: 805, 822, 1030. [39]董大年. 有效. 现代汉语分类大词典. 上海: 上海辞书出 版社, 2007: 1003. [40]凌复华. 突变理论——历史、现状和展望. 力学进展, 1984, 14(4): 9-24. [41]何文平. 动力学结构突变检测方法的研究及其应用. 兰 州: 兰州大学, 2008. [42]朱吉祥, 张礼中, 周小元, 等. 区域地质灾害评价的有效 性周期分析. 安全与环境学报, 2012, 12(3): 253-258. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 137–145 ___________________________________________________________________________________________________________ 145 自然灾害概率风险及其时效性 1. 引言 2. 自然灾害概率风险 2.1. 自然灾害风险的概念化模型 2.2. 概率风险的形式化模型 3. 自然灾害概率风险的时效性界定 3.1. 自然灾害风险的动态变化性 3.1.1. 风险源发生规律的动态变化性 3.1.2. 风险承受体脆弱性的动态变化性 3.1.3. 自然灾害风险动态变化的特点 3.2. 概率风险的时效性 4. 自然灾害概率风险时效性评价 4.1. 时效性评价的形式化表达 4.2. 时效性评价的一个应用模型 5. 结论 参考文献 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <feff03a703c103b703c303b903bc03bf03c003bf03b903ae03c303c403b5002003b103c503c403ad03c2002003c403b903c2002003c103c503b803bc03af03c303b503b903c2002003b303b903b1002003bd03b1002003b403b703bc03b903bf03c503c103b303ae03c303b503c403b5002003ad03b303b303c103b103c603b1002000410064006f006200650020005000440046002003c003bf03c5002003b503af03bd03b103b9002003ba03b103c42019002003b503be03bf03c703ae03bd002003ba03b103c403ac03bb03bb03b703bb03b1002003b303b903b1002003c003c103bf002d03b503ba03c403c503c003c903c403b903ba03ad03c2002003b503c103b303b103c303af03b503c2002003c503c803b703bb03ae03c2002003c003bf03b903cc03c403b703c403b103c2002e0020002003a403b10020005000440046002003ad03b303b303c103b103c603b1002003c003bf03c5002003ad03c703b503c403b5002003b403b703bc03b903bf03c503c103b303ae03c303b503b9002003bc03c003bf03c103bf03cd03bd002003bd03b1002003b103bd03bf03b903c703c403bf03cd03bd002003bc03b5002003c403bf0020004100630072006f006200610074002c002003c403bf002000410064006f00620065002000520065006100640065007200200035002e0030002003ba03b103b9002003bc03b503c403b103b303b503bd03ad03c303c403b503c103b503c2002003b503ba03b403cc03c303b503b903c2002e> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word volcanic risk perception, locus of control, stress and coping responses of people living volcanic risk perception, locus of control, stress and coping responses of people living near the popocatépetl volcano in mexico esperanza lópez-vázquez psychology faculty, universidad autónoma del estado de morelos.   cuernavaca, morelos, 62209, méxico e-mail:esperanzal@uaem.mx www.uaem.mx  maria luisa marván institute of psychological research, universidad veracruzana. méxico xalapa, veracruz city, 9110, méxico. e-mail: mlmarvan@gmail.com www.uv.mx  abstract the influence of locus of control and stress on coping strategies in relation to volcanic risk were studied in 156 people living near the popocatépetl volcano in mexico. the participants with a limited educational background were more likely than those with higher education to use passive coping strategies. linear regression showed that a higher level of psychological stress predicts more active strategies. these results confirm some of our previous studies. probable explanations for our findings are discussed. keywords: locus of control, stress, coping strategies, volcanic risk 1. introduction there are numerous volcanoes in the world that are active, and around 50 or 60 volcanoes erupt annually1. even though volcanic threats have always existed and have brought disaster to afflicted populations, research on the social impact of such risks did not develop until well into the 20 century. since the 1990s, a larger number of works have shown the link between risk perception and people’s behavior in a volcanic risk environment 2, 3, 4, 5. natural hazards have been frequently analyzed as being part of the social relationship of production 6, 7. some topics developed in these studies are: the perception of risk and peoples’ behaviors in a volcanic risk scenario and/or the influence of religious beliefs 8, 9, 10, 11; peoples’ responses to volcanic risks 12, 13, 14; the impact and efficiency of risk communication for people exposed to risk 15, 16, 17; risk perception and socioeconomic variables 18, 15, 10; risk perception, knowledge and preparedness 19, 13, 20, 21; risk perception and risk management 22; and culture and volcanic risk mitigation 23, 24. volcanic risk* and volcanic risk perception are a main topics of interest in méxico due to the many volcanoes currently active in the country. there are 16 active volcanoes25. the majority of the volcanoes are located in the center of the country where there are at least 36 million people exposed to the risk. the popocatépetl volcano is one of the most active, with a high risk level a volcanic risk can be understood as the likelihood of a person or a property to be injured/killed or damaged by a volcanic hazard. it depends on the timescale, the location of the person/property and the current state of the volcano (http://www.volcanodiscovery.com/volcanic_risk.html). journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 3-12 published by atlantis press copyright: the authors 3 administrateur texte tapé à la machine received 18 february 2012 administrateur texte tapé à la machine accepted 12 march 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine esperanza lópez-vázquez & ma. luisa marván and intermittent periods of activity. the current period of volcanic activity began in december 1994, presenting intense seismic activity, followed by explosions. ash falls have been the principal problem arising from the recent activity. the hazards surrounding popocatépetl have been divided into three risk zones according to proximity to the source. these zones are classified on a volcanic hazard map for pyroclastic flows (composed of blends of solid or liquefied particles and gases at very high temperature that may act as a liquid with high mobility and destructive power), tephra falls (different volcanic materials such as pumice stone with very little pyroclastic elements), landslides (movements of the land provoked by the activity of a volcano), and lahars (flows containing volcanic stone fragments that may incorporate water and mud when descending the volcano’s slopes) 26. the divisions are determined by the distance from the crater and by the risk intensity of each zone 27. it is important to clarify that “risk” is a concept than can be defined from two perspectives: 1) the objective evaluation that define risk as a function of the interaction between the harmful effects of a hazard and the vulnerability of individuals to these effects 21, 28; 2) and the subjective evaluation that involves the cognitive analysis that each individual make of a hazard that is constructed along its own life 28. thus, risk perception, defined from the psychosocial point of view, is understood as a process that is influenced by different factors that are personal, social, political, historical and cultural 30. the perception of a risk also depend on specific factors such as familiarity with the hazard, the voluntary acceptance of the activity, the degree of the exposure one is submitted to, the knowledge one has of said hazard, the degree of control one has to avoid or mitigate the effects that said hazard has, the novel or chronic nature of the hazard, the fear that is generated by the constant exposition to the risk situation, and the actual real severity of the threat 31, 32; the social benefit that the hazard offers, the acceptability people express regarding said hazard, and the trust people have in the regulatory and official bodies in charge33, 34, 35, 36, 37, and other factors that make exposure to volcanic risk a complex process to study. the perception of risk, when added to a threat that may disrupt people’s stability, generates tension in a person’s life. this tension has been called “stress”, and is provoked by exterior stimuli that threaten the organism and generate the organic response we call stress38. in his studies on the subject, lazarus concludes that a certain stimulus is capable of generating stress only if it is evaluated (cognitive process) by the subject as being stressful39, 40, 41. considering that a hazard, which represents a potential threat, can generate the stress response; this process is strongly related to other variables such as personality, experience with the hazard, the knowledge one has of the hazard42 and the educational, socio-economic levels of individuals,39 and subjective evaluations. behaviors that are generated by confronting a stressful stimulus are called coping strategies, and are designed to defend and protect the individual from the stressor that is evaluated as threatening43, 44. these strategies can be classified in different ways, but in this paper they will be separated into two large groups: active and passive coping strategies. the first group corresponds to all those active behaviors that directly confront a threat (such as gathering information, direct evasive actions, or attempts to control or mitigate the situation); the second group refers to behaviors characterized by rejection or denial of the event, withdrawal, or passive acceptance45. taking into account lazarus’s model of stress, duval and mulilis have proposed the person relative event (pre) model. this model says that that increasing levels of threat, when resources of an individual are appraised as sufficient relative to the magnitude of the threat, will increase problem-focused coping, (in other words, the active coping). on the contrary, increasing levels of threat when resources are appraised as unsatisfactory relative to threat scale will decrease problem-focused coping (pfc)46, 47, 48, 49. according to the authors and studies using this pre model, theory is related to locus of control, which refers to generalized beliefs about the causes of people’s outcomes50, 51. the concept of locus of control refers to the individual’s perception of his or her own capability to control a given situation, and which in turn depends on certain reinforcing events that allow the person to test the level of control that he or she feels towards the given object, person or event. through such relationships, individuals may recognize the existence of certain situations they themselves can control or mitigate by their own intervention (internal control), or that may be controlled or mitigated by external forces or other persons published by atlantis press copyright: the authors 4 volcanic risk, locus of control, stress and coping (external control) 51, 52, 53, 54. in the case of natural hazards, it has been found that people with an internal locus of control believe that life events may be at least partially controlled by their own actions, they are well prepared for tornadoes55, they also buy more flood insurance56, and see earthquake possible damages as more preventable than do people with an external locus57. locus of control also predicts mitigation actions more than survival actions58. overview many studies have been carried out examining the perception of volcanic risk in different contexts and different populations around the world, but little has been done to evaluate the level of stress produced by such exposure to risk or the types of coping strategies that people use. a previous study on people living in different volcanic risk zones around the volcano popocatépetl showed that one factor of the perception of risk is the feeling of insecurity linked to the volcanic risk, and that this factor influences the level of stress in people who live around the volcano. in that study, when interaction effects between risk zones and the other variables were analyzed, it was found that active coping strategies were moderated according to risk zone. results also demonstrated that risk zone operates as a moderating variable, acting with coping strategies to determine stress levels14. the goals of this study were: a) to study the level of stress, the type of locus of control (internal-external), and the type of coping strategies (active-passive) used by people living near a volcanic risk area, according to their education level; b) to study the influence of locus of control and stress on coping strategies in relation to volcanic risk. we expect that people with a higher education use more active and less passive coping strategies, and show a more internal locus of control than people with lower education level, and more stress. furthermore, we expect that both locus of control and stress have influence on coping strategies used by participants. 2. method 2.1 participants the sample was composed of 156 adults between the ages of 18 and 60. all of them were residents of the cities of san pedro cholula or cuautla, in mexico. both cities are located at 20 km from the popocatépetl volcano crater. to make sure that participants had lived a considerable time under the volcanic risk, an inclusion criterion was that participants had to have lived in the risk zone for at least five years. the participants were divided into two groups according to their educational level: a) people with a limited educational background (37 women and 36 men), whose schooling varied from six to nine years. they were blue-collar workers and had jobs that are usually poorly paid. women who did not work were housewives married to blue-collar workers. in accordance with the limited educational level of these participants as well as the nature of their employment, we inferred they held a lower or lower-middle socioeconomic status. b) people with higher education, whose schooling was at least 12 years (51 women and 32 men). these participants were white-collar workers and held positions that are usually well paid. women who did not work were housewives whose husbands were whitecollar workers. in mexico, people with higher education earn, on average, 300% more than those who did not complete their basic education53. thus, we inferred that these participants had an upper-middle socioeconomic status. characteristics of the visited cities in 2005, the population of the city of san pedro cholula was well over 110,000 inhabitants. most frequent occupations include agriculture (corn, beans, alfalfa, and certain species of flowers); stockbreeding (goats, cows, pigs and sheep); industry (mainly fabrication of bricks and tiles for construction and decorative use); as well as commerce, government officials, teachers, technicians and blue-collar labor59. the municipality is well known both nationally and internationally due to its historical importance during the pre-hispanic period of mexican history. it was a very important ceremonial center. the city of cuautla had more than 150,000 inhabitants in 2005 and notable occupations included agriculture (beans, onions, rice, cucumber, green tomatoes, zucchini and sugar cane), stockbreeding, poultry farming, beekeeping, aquaculture, agro-industry, commerce and tourism60. published by atlantis press copyright: the authors 5 esperanza lópez-vázquez & ma. luisa marván it is important to point out that both cities are located in risk zones which, in the event of a volcanic eruption, would be exposed to pyroclastic flows, lahars, and ash fall25. 2.2 instruments in the first part of the survey participants completed a basic information sheet that asked general questions in order to know how many years had they lived in that city, and other socio-economic data. after that, participants answered three different questionnaires: a) coping strategies scale (css). this scale was adapted from the “echelle toulousaine de coping”45 and validated in mexico by lópez-vázquez and marván61. this instrument is a 5-point likert scale from 1“i never react that way” to 5“i always react that way”. the questionnaire has 26 items that measure two factors: 1) active coping strategies, which includes items that refer to direct active behaviors relating to the problem, searching for information, strategies that anticipate the problem, as well as self-control and control of circumstances; and 2) passive coping strategies which includes items that refer to behaviors such as rejection or denial of the event, withdrawal or passive acceptance. participants were asked to respond to this scale according to their experience of the popocatépetl volcanic risk. the questionnaire’s reliability was .86 using chronbach’s alpha coefficient. b) stress scale (ss). this scale was adapted and validated in mexico by jiménez62 from the “echelle toulousaine de stress”45. it is composed of 27 items that examine typical reactions that most people present when they are exposed to a stressful event. the first factor in this scale is psychological stress, which refers to different feelings such as threatened, helpless, lonely, lack of control or lack of understanding about the stressful situation. the second factor concerns the psychophysical exhaustion stress, which refers to the presence of physical manifestations such as having a knot in one’s throat or being exhausted most of the time. the third factor is psycho-physiological stress, which refers to feeling tired, being unable to sleep or becoming agitated. finally, the fourth factor is physiological stress, which refers to physiological reactions caused by stress as having stifling heat or accelerated heart beat. this instrument is a 5-point likert scale from 1“i never react that way” to 5“i always react that way”. the participants were asked to answer according to the stress they felt related to the volcano popocatépetl (chronbach’s alpha = .94). the coping and stress scales were chosen because they were written in a general sense that can be adapted to any stressing event. the instructions allow to the researcher to specify clearly the situation we are studying, and for our study, the items were adapted to volcanic risk. c) locus of control scale (lcs). this scale was developed in mexico by la rosa53. the original scale has 61 items, but we used 48 items that loaded onto five subscales: the first subscale, "instrumental internality," refers to situations that individuals are able to control in accordance with their personal capabilities. the second subscale is the "affective scale" and describes objectives reached by affective relationships between people. the third factor, "fatalism/luck," consists in reliance on fortune or faith to face different situations (external factor). the fourth factor, called "microcosm powerful," refers to people that are close to an individual (family or friends) having some power over that person’s life (external factor). the fifth factor, called "macrocosm powerful," describes people that are more distant from an individual (frequently they have never talk to them as religious guides or politicians) having an influence on them (external factor). this scale was scored in a likert 5-points scale from 1"completely disagree", to 5"completely agree"(chrombach’s alpha = .78). 2.3 procedure the survey method was employed in this research. this method allows recruiting participants in their natural environment. an interviewer visited both cities over a period of about two months. during these visits, adults were approached in their workplaces, on the streets or in stores and were asked if they would be willing to participate in a study regarding the popocatépetl volcano. after ascertaining that individuals fulfilled the criteria mentioned, interviewer set up a time and a place to meet with the participant in order to conduct the survey. people were polled individually. the interviewer gave the questionnaires to the participants and made sure that they understood the meaning of the published by atlantis press copyright: the authors 6 volcanic risk, locus of control, stress and coping items. if the person understood, the interviewer let him/her complete the questionnaire independently; if not, the interviewer read each of the items in the surveys to make sure that the participants properly understood each question and recorded their answers. each individual survey took from 2030 minutes to complete. 2.4 data analyses: data analyses were performed using the statistical package spss version 16.0. independent sample t-tests were conducted to compare the locus of control, the stress level and the coping strategies of participants depending on their educational level. linear regressions were performed with the coping strategies scores as dependent variables, and the independent variables were: years of schooling, the five factors comprised in the locus of control scale, and the four factors comprised in the stress scale. 3. results the results of both groups of participants according of their educational level are shown in table 1. the participants with limited educational backgrounds were more likely than those with higher education to use passive coping strategies. however, there are no significant differences when comparing active strategies. table 1. means (m) and standard deviation (sd) for coping strategies, locus of control, and stress for participants with limited educational backgrounds (n = 73) and with higher levels of education (n = 83). limited education higher education m (sd) m (sd) student t test coping active strategies 3.15 (0.77) 2.90 (0.84) 1.90 coping passive strategies 2.99 (0.76) 2.71 (0.65) 2.46 ** stress exhausting 1.89 (0.73) 1.93 (0.73) 0.31 stress physical 1.70 (0.66) 1.79 (0.83) 0.70 stress psychophysical 1.94 (0.71) 1.81 (0.64) 1.24 stress psychological 2.23 (0.78) 2.12 (0.76) 0.87 l.c. affective scale 3.23 (0.98) 3.20 (0.95) 0.19 l.c. instrumental internality 1.67 (0.51) 1.86 (0.61) 2.14 * l.c. macrocosm powerful 3.04 (0.83) 3.01 (0.77) 0.21 l.c. microcosm powerful 2.56 (0.90) 2.53 (0.72) 0.24 l.c. fatalism/luck 2.73 (1.12) 2.32 (0.91) 2.54** l.c. = locus of control *p value < .05, ** p value <.01 concerning locus of control, there were significant differences in two subscales of the lcs: participants with higher education scored higher than those with limited educational background on the “instrumental internality” subscale and they scored lower on the "fatalism/luck" subscale. there were no significant differences when any of the stress subscales were compared between both groups of participants. in order to identify some of the possible variables that influence coping strategies related to volcanic risk, two linear regression analyses were conducted. in the first one the score of the passive coping strategy was entered as dependent variable, and in the second the score of the active coping strategy was the dependent variable. both models were statistically significant. as can be seen in table 2, we found that passive coping strategies were predicted by: (a) higher scores on the "fatalism/luck" subscale of the locus of control scale; (b) higher scores on the “psycho-physical exhaustion” subscale of the stress scale; and (c) lower schooling. additionally, we found that active coping strategies were predicted by published by atlantis press copyright: the authors 7 esperanza lópez-vázquez & ma. luisa marván both higher scores on the “psychological stress” subscale of the stress scale and lower scores on the “"macrocosm powerful" subscale of the locus of control scale. table 2. hierarchical regression analysis for variables predicting coping strategies. coping strategies passive strategies active strategies variables b se b beta b se b beta schooling -.08 .04 -.13* -.08 .06 -.10 l.c. affective scale -.14 .07 -.19* .02 .10 .03 l.c. instrumental internality -.03 .08 -.02 -.15 .12 -.11 l.c. macrocosm powerful -.07 .07 -.08 -.27 .10 -.27** l.c. microcosm powerful -.08 .08 -.09 .05 .12 .05 l.c. fatalism/luck .29 .06 .42** .03 .08 .04 stress exhausting .21 .10 .22* -.17 .15 -.15 stress physical .02 .10 .02 -.04 .14 -.04 stress psychophysical -.05 .11 -.05 .15 .16 .12 stress psychological .14 .10 .15 .38 .15 .36** note:passive strategies: adjusted r2 = .425 (p < .001); active strategies: adjusted r2 = .090 (p < .01); l.c. = locus of control * p < .05; ** p <.01 according to the variance inflation factor (vif) test, there were no co-linearity problems in our data since all vif scores were lower than three. 4. discussion the results obtained in the present study show that some subscales of the variables studied have a clear influence on the coping strategies used by participants. we observed that the control locus “fatalism/luck” factor tends to predict the use of passive strategies. this makes sense if we consider that, according to the authors of the locus of control scale, people who tend to control externally feel less able to handle environmental problems than those who do so internally55, and such individuals also have a tendency to believe that control of the situation lies beyond their ability. according to the pre model, under conditions in which resources are appraised by individuals as insufficient relative to threat, increasing absolute levels of appraised menace will decrease the problem focused coping (pfc), independently of the level of appraised resources58. in that case, more passive coping strategies are able be used by people. moreover, according to mulilis and duval 48, 49, this predicted effect is more evident under conditions of high personal responsibility for preparation, as opposed to low levels of the same variable, and so, it has less active implications as it might happened in our present study. these results could be also explained taking into account the socio-cultural aspects of the mexican tendency to favor passive behaviors and to promote disengagement from social problems. díaz-guerrero, in his studies on the “psychology of the mexican”, describes how the most effective strategy that the mexican has to face stress in life is to do so passively, due to the social assumption that passivity is a sign of “virtuosity”. this is promoted by the influence of the catholic religion, which favors the tendency to accept one’s fate without qualms63. other variables that predicted the use of passive strategies were low educational level and psychological exhaustion. in this regard, a study analyzing risk perception of precollege students for 20 environmental hazards, and its relationship to gender, community socioeconomic setting (cses), age, and locus of published by atlantis press copyright: the authors 8 volcanic risk, locus of control, stress and coping control, found that lower cses students had less internal control locus than higher cses students, and the authors confirmed that these relationships are generally consistent with previous research64. another study conducted in sonora, mexico, on risk perception and behavior towards protecting the environment, showed that people from lower economic levels, and those of a more advanced age, perceive greater social and personal risk from environmental hazards, which, according to the authors, might reflect a greater state of defenselessness due to their lower economic resources65. as mentioned earlier, lazarus39 has shown how personal resources and variables such as demands, limitations or opportunities, as well as other variables that include intelligence, social abilities, recent friendships, physical health, and of course education and money, may influence an individual’s stress response. this suggests that both the level of stress a person develops, and the type of response he or she makes to a stressing situation will be in accordance with his/her available resources. if the individual lacks available resources that he or she considers necessary, no adaptation or strategy will permit him or her to satisfy the external demand. this is when the stress response is stifled by the impossibility to act and may in the long run negatively affect the individual’s health66, 39, 44. another of our findings demonstrated that higher levels of psychological stress predict more active coping strategies. thus, the stress generated by exposure to volcanic risk tends to stimulate actions by the individual rather than reducing or annulling them. in a previous study carried out in a population near popocatépetl, the researchers looked for a relationship between feelings of insecurity and stress and coping strategies. this study showed that people using more active coping strategies reported experiencing more stress. in that study, the authors concluded that the stress and concomitant risk caused by volcanic activity probably had the role of a positive stress14, or maybe in the sense of lazarus, a challenge40. we consider that our finding is directly related with the individual’s evaluation of the stressful situation and his/her determination of whether it can be controlled, depending on perceived resources and other factors that tend to reduce anxiety. some of these factors related to the perception of risk may include: (a) familiarity with the hazard from daily exposure to it, (b) specific knowledge people have concerning the risk, (c) peoples’ toleration of, or willingness to be exposed to, volcanic risks, (d) the control people may feel they have because volcanic risk may not be a daily source of stress in their lives, and (e) the extent of knowledge that people may have concerning the dangers and possible preventive measures that may be available to them 31,67. considering the pre model, the resources of our participants are certainly assessed as sufficient to face the threat, and we can interpret that these people are using pfc46, 47, 48, 49. this kind of coping allows people to be centered in preparedness actions to face natural hazards. it is important to state that after the 1994 eruptions, the government informed people by tv, radio, newspaper, internet, and printed announcements in bus stops and highways stands, which stated the actions the population should carry out in case of a volcanic eruption. some of these instructions included having important papers at hand, storing reserves of water and food, protecting windows and entries from ash falls, taking out ash from weak building structures. other actions were carried out by the university center for the prevention of disasters (centro universitario de prevensión de desastres – cupreder) and by the government through a program called the popocatépetl plan68. public awareness campaigns were carried out to inform populations dwelling near the popocatépetl, concerning the principal preventive measures to be taken. such activities may foster more positive feelings of control over the situation in case the threat of volcanic activity becomes imminent. another finding of the current research is that low scores in external locus of control (those that are powerful in the macrocosm) also predict active coping strategies. this corresponds to the locus of control theory which holds that people who use less external control will use more internal control by applying their direct coping capabilities to a greater extent 51, 52, 48. we also noticed that the internality factor of locus of control is relatively low in both groups. considering the pre model, we can think that our participants evaluate their resources as insufficient, and this could be a reason of their low internality. we cannot say more about this result because our measure was not concentrated in responding about natural hazards, it was a general scale that shapes a personality feature. published by atlantis press copyright: the authors 9 esperanza lópez-vázquez & ma. luisa marván finally, this study could be replicated in mexico or in other countries wherever populations are exposed to volcanic hazards in order to compare results and validate our findings. we believe it is important to continue exploring these variables and others such as cognitive bias, knowledge of preventive measures, attitudes, etc. with populations exposed to natural hazards, for the purpose of increasing our knowledge of cognitive mechanisms involved in the perception of risk and peoples’ responses when faced with such hazards. such studies can be of great use in improving our methods of communicating information about the potential risks and preventive measures to people that are exposed to these hazards. acknowledgments: the authors wish to thank mónica encinas herrera who assisted us in the collection of data. references 1. simkin t. terrestrial volcanic eruptions in space and time. proceedings of kagoshima international conference on volcanoes. kagoshima prefectural government. (kagoshima, japan, 1988). pp. 412-415. 2. lachman, r. & bonk, w. j. behavior and beliefs during the recent volcanic eruption in kapolo, hawaii. science, 131 (3407) (1960). 1095-1096. 3. greene, m.r., perry, r.w., lindell, m.k. the march 1980 eruptions of mt. st. helens: citizen perceptions of volcano hazard. disasters, 5 (1) (1981). 49-66. 4. perry, r.w., lindell, m.k. & greene, m.r. threat perception and public response to volcano hazard. the journal of social psychology, 116, (1982). 199-204. 5. gaillard, j. c. volcanic risk perception and beyond. journal of volcanology and geothermal research, 172. (2008). 163-169. 6. odum, e. p. fundamentals of ecology, 3rd ed. (philadelphia, saunders college publishing 1971). 7. worster, d., the ecology of order and chaos. environmental history review, 14 (1/2) (1990). 1-18. 8. dove, m. r. perception of volcanic eruption as agent of change on merapi volcano, central java. journal of volcanology and geothermal research. 172. (2008). 329337. 9. chester, d. k., duncan a. m., dibben & c. j.l. the importance of religion in shaping volcanic risk perception in italy, with special reference to versuvius and etna. journal of volcanology and geothermal research. 172. (2008). 216-228. 10. lavigne, f., de coster b., juvin, n., flohic, f., gaillard j. c., texier, p., morin, j. & sartohadi, j. people’s behaviour in the face of volcanic hazards: perspective from japanese communities, indonesia. journal of volcanology and geothermal research. 172. (2008). 273287. 11. perry, r.w., lindell, m.k. volcanic risk percepcion and adjustment in a multi-hazard environment. journal of volcanology and geothermal research, 172. (2008). 170178. 12. gaillard, j. c. alternative paradigms of volcanic risk perception: the case of mt. pinatubo in the philippines. journal of volcanology and geothermal research. 172. (2008). 315-328. 13. gregg, c.e., houghton, b.f., paton, d., swanson, d.a. & johnston, d. m. community preparedness for lava flows from mauna loa and hualalai volcanoes, kona, hawai’i. bulletin of volcanology. 66. (2004). 531-540. 14. lópez vázquez, e., marván m. l., flores f. & peyrefitte a. volcanic risk exposure, feelings of insecurity, stress and coping strategies in mexico. journal of applied social psychology. 38(12). (2008) pp. 2885–2902. 15. baberi, f., davis, m.s., isaia r., nave, r. & ricci, t. volcanic risk perception in the vesuvis population. journal of volcanology and geothermal research. 172. (2008). 244-258. 16. haynes, k., barclay, j. & pidgeon, n. woshe reality counts? factors affecting the perception of volcanic risk. journal of volcanology and geothermal research. 172. (2008). 259-272. 17. solana, m.c., kilbum, c.r.j., & rolandi, g. communicating eruption and hazard forecasts on versuvius, southern italy. journal of volcanology and geothermal research. 172. (2008). 308-314. 18. dibben, c. j. l. leaving the city for the suburbs – the dominance ordinary decision making over volcanic risk perception in the production of volcanic risk on mt. etna, sicily. journal of volcanology and geothermal research. 172. (2008). 288-299. 19. carlino, s., somma r. & mayberry g.c. volcanic risk perception in the vesuvius population. journal of volcanology and geothermal research. 172. (2008). 229243. 20. lópez vázquez, e. risk perception and coping strategies for risk from popocatépetl volcano, mexico. geofísica internacional. 48 (1). (2009). 301-315. 21. paton, d., smith, l. & johnston, d. volcanic hazards: risk perception and preparedness. new zealand journal of psychology. 29 (2). (2000). 86-91. 22. gavilanes-ruíz, j. c., cuevas-muñiz, a., varley, n., gwynne, g., stevenson, j., saucedo-girón, r., pérezpérez, a., aboukhalil, m. & cortés-cortés, a. exploring the factors that influence the perception of risk: the case of vocán de colima, mexico. journal of volcanology and geothermal research. 186. (2009). 238-252. 23. gregg c.e., houghton, b.f., paton, d., swanson, d.a., lachman r. & bonk, w.j. hawaiian cultural influences published by atlantis press copyright: the authors 10 volcanic risk, locus of control, stress and coping on support for lava flow hazard mitigation measures durin the january 1960 eruption of kilauea volcano, kapoho, hawaii. journal of volcanology and geothermal research, 172, (2008). 300-307 24. paton, d., smith, l., daly, m. & johnston, d. risk perception and volcanic hazard mitigation: individual and social perspectives. journal of volcanology and geothermal research. 172. (2008). 179-188. 25. macías, j. l. & caprapedol, l. los volcanes y sus amenazas. (méxico, fondo de cultura economica. 2005). 26. de la cruz, r. s. & ramos, j. e. definición y clasificación de calamidades de origen volcánico y sus efectos. (méxico, cenapred. 1995). 27. macías, j.l., carrasco, g., delgado, h., martin del pozzo, a.l., siebe, c., hoblitt, r., sheridan, m.f. & tilling, r.i. mapa de peligros volcánicos del popocatépetl, scale 1:25,000 (méxico: universidad nacional autónoma de méxico, instituto de geofísica. (1995). 28. weyman, a. k. y clarke, d.d. the effect of familiarity on workplace risk assessment contextual insight or cognitive availability? contemporary ergonomics, publ. taylor and francis, (2003). 463-470. 29. assally, j. p. les jeunes et le risque: une approche psychologique de l’accident. (paris, vigot. 1992). 30. slovic, p. perception of risk. in p. slovic (ed.), the perception of risk. (london, earthscan publications ltd. 2000). 31. slovic, p. (1987). perception of risk. science ,236, 280– 85. 32. fischhoff, b., slovic, p., lichtenstein, s., read, s. & combs, b. how safe is safe enough? a psychometric study of attitudes toward technological risks and benefits. in p. slovic (1 eds.), the perception of risk, (80-103). (london, earthscan publications ltd. 2000). 33. bronfman, n. perception and acceptability of technological and environmental hazards in chile, (phd thesis pontifica universidad católica de chile, 2005). 34. bronfman, n. c., lópez-vázquez, e., gutiérrez, v. v. &cifuentes, l. a. trust, acceptance and knowledge of technological and environmental hazards in chile. journal of risk research 11(6), (2008). 755-773. 35. bronfman, n. c., lópez-vázquez, e. & dorantes, g. an empirical study for the direct and indirect links between trust in regulatory institutions and acceptability of hazards. safety science. 47. (2009). 686-692. 36. siegrist, m. the influence of trust and perceptions of risk and benefits on the acceptance of gene techonology. risk analysis 20(2). (2000). 195-203. 37. siegrist, m., & cvetkovich, g. t. perception of hazards: the role of social trust and knowledge. risk analysis. 20(5). (2000). 713-719. 38. selye, h. la tensión de la vida (el estress). (argentina, compañía general fabril editora. 1960). 39. lazarus, r. s. estrés y emoción. manejo e implicaciones en nuestra salud. (españa: desclée de brower. 2000). 40. lazarus, r. s., & folkman, s. stress, appraisal, and coping. (new york, springer pub. 1984). 41. appley, m. h. & trumbull, r. psychological stress: issues in research. (new york: appleton-century-crofts. 1967). 42. lazarus, r. s. & launier, r. stress related transactions between person and environment. perspectives in international. (new york: psychology new york presses. 1978). 43. lazarus, r. s., & folkman estrés y procesos cognitivos. (méxico: ediciones martínez roca. 1991). 44. esparbès, s., sordes-ader, f. & tap, p. l’échelle toulousaine de coping: un instrument pour l’analyse des liens entre personnalisation et stratégies de coping. data report 1993. st. cricq, (france, journées du labo pcs, université de toulouse le-mirail. 1993). 45. duval, t. s., duval. v. h., & mulilis, j. p. self-focus, degree of discrepancy between self and standard, outcome expectancy favorability and behavior. journal of personality and social psychology, 62, (1992). 340-348. 46. duval, t.s., mulilis, j.p. a person-relative-to-event (pre) approach to negative threat appeals and earthquake preparedness: a field study. journal of applied social psychology 29, (1999). 495-516. 47. mulilis, j.p., & duval, t. s. negative threat appeals and earthquake preparedness: a person-relative-to-event (pre) model of coping with threat. journal of applied social psychology, 25, (1995). 13 191339. 48. mulilis, j-p., & duval, t. s. the pre model of coping and tornado preparedness: moderating effects of responsibility. journal of applied social fësychology, 27, (1997). 1750-1766. 49. carver, c. s. the internal-external scale confounds internal locus of control with expectancies of positive outcomes. personality and social psychology bulletin,23, (1997). 580-585. 50. rotter, j. b. generalized expectancies for internal versus external control of reinforcement. psychological monographs, 80, 1-609. (1966). 51. dubois, n. la psychologie du contrôle. les croyances internes et externs. (france, presses universitaires de grenoble. 1987). 52. la rosa, j. escalas de locus de control y autoconcepto: construcción y validación. ph. d. thesis in social psychology. (méxico, universidad nacional autónoma de méxico. 1986). 53. anuies asociación nacional de universidades e instituciones de educación superior. (2003). méxico. retrieved november 17, 2010, from: http://www.anuies.mx/e_proyectos/html/parte%201/cua dros%20diagnostico.htm 54. lefcourt, h. m. the construction and development of the multidimensional-multiattributional causality scales. published by atlantis press copyright: the authors 11 esperanza lópez-vázquez & ma. luisa marván research with the locus of control construct. 1. (1981). 245-277. 55. sims, j., & bauman, d. the tornado threat: coping styles of the north and south. science. 176. (1972). 1386-1392. 56. baumann, d. d., & sims, j. h. flood insurance: some determinants of adoption. economic geography, 54, (1978). 189-196. 57. mcclure,j.,walkey,f.,& allen,m. when earthquake damage is seen as preventable: attributions, locus of control and attitudes to risk. applied psychology:an international review, 48, (1999). 239-256. 58. spittal, m. j., mcclure, j., siegert, r. j., walkey, f. h. predictors of two types of earthquakes: survival activities and mitigation activities. environmental and behavior, 40, 6, (2008). 798-817. 59. instituto nacional para el federalismo y el desarrollo municipal gobierno del estado de puebla enciclopedia de los municipios del estado de puebla, 2005. (2009). available from: http://www.elocal.gob.mx/work/templates/enciclo/puebla/mpios/21140 a.htm [accesed 17 november 2010]. 60. instituto nacional para el federalismo y el desarrollo municipal gobierno del estado de morelos (2005). enciclopedia de los municipios del estado de morelos. available from: http://www.elocal.gob.mx/work/templates/enciclo/morelos /municipios/17006a.htm [accesed 17 november 2010]. 61. lópez-vázquez, e., & marván, m. l. validación de una escala de afrontamiento frente a riesgos extremos. salud pública de méxico, 46, (2004). 216-221. 62. jiménez, a.. percepción de riesgo, estrés y afrontamiento en zonas de alto riesgo de méxico: validación de instrumentos de medición. unpublished master’s thesis, universidad de las américas-puebla (2002). 63. diáz-guerrero, r. psicología del mexicano: descubrimiento de la etnopsicología. 5th edition. (méxico, trillas. 2002). 64. riechard, d. & peterson, s.j. perception of environmental risk related to gender, community socioeconomic setting, age, and locus of control. the journal of environmental education. 30 (1). (1998). 65. corral-verdugo, v., frías, m. & gonzález, d. percepción de riesgos, conducta proambiental y variables demográficas en una comunidad de sonora, méxico. región y sociedad. 15. (2003). 49-72. 66. bruchon-schweitzer, m. (2002). psychologie de la santé. modèles, concepts et méthodes. paris: dunod. 67. fischhoff, b., p. slovic, s. lichtenstein, s. read, & b. combs. (1978). how safe is safe enough? a psychometric study of attitudes towards technological risks and benefits. policy sciences, 9, 127–52. 68. patiño, l.m. (2005). comunicación de riesgo y su impacto en las poblaciones afectadas por el volcán popocatépetl. bd thesis. universidad de las américas-puebla. published by atlantis press copyright: the authors 12 atlantis press journal style editor’s introduction in order for the international journal of risk analysis and crisis response (jracr) to be indexed in science citation index (sci), the society for risk analysis china (sra-china) is strengthening the relevant work to jracr. srachina appointed prof. mu zhang publication chairman, who works in guizhou university of finance and economics. he will be responsible for jracr’s editorial board. sra-china appointed dr. junxiang zhang director of jracr editorial department, who works in huangshan university. in the last issue of this year, the six papers can be divided into three categories: risk management, risk assessment and financial risk. there is one paper on risk management, “a two-step water-management approach for nuclear power plants in inland china” by xiaowen ding，guohe huang,wang wei, et al., which reviews the inland nuclear power industry and its policies in china. a two-step water-management (tswm) approach for the plants is proposed in this paper. the framework includes the flow process, main tasks, and tools of tswm management for any nuclear power plant in inland china. finally, suggestions on future development of the management are also put forward. there are three papers on risk assessment. the first paper "ship to ship transfer of cargo operations: risk assessment applying a fuzzy inference system”, by dimitrios i. stavrou and nikolaos p. ventikos, applies fuzzy inference system as a novel approach for risk assessment for ship to ship transfer operations. the advantages of the fuzzy approach are highlighted through the application of the mamdani’s method for hypothetical accident scenarios. the second paper, "study on risk evaluation based on occupational exposure evaluation and carcinogenic risk simulation” by qian zhang, deyin huang and mao liu provides the quantitative risk evaluation for 1,3-butadieneexposure by using physio-logically based pharmacokinetic model and dose-response model. a large chemical enterprise in tianjin, china, is selected as the case study and carcinogenic risk evaluation for 1,3-butadiene-exposed workers is calculated. the results show that the cancer risk of 1,3-butadiene-exposed workers obviously exceeds 1×10-4 the maximum acceptable risk level. the third paper “flood disaster risk assessment and spatial distribution characteristics along the yangtze river in anhui province” by xianfu cheng, honghu sun, zhang yuan and guanglai xu uses analytic hierarchy process and entropy method to determine the weight of index. a fuzzy comprehensive evaluation model of risk assessment was constructed in this paper. flood risk index and risk rating of area along the yangtze river in anhui province were drawn by using gis spatial analysis technology. from disaster causing factors, inducing environment and bearing body, average maximum 3-day rainfall, rainstorm number, vegetation coverage, river network density, standard deviation of the elevation, population density, proportion of rural population, density of gdp, per capita net income of farmers, cultivated land area ratio were selected as evaluation indicators. the results show that flood risk rating and risk index of area along the yangtze river in anhui province has a spatial distribution characteristics from south to north gradually reduced. there are two papers on financial risk. the first paper "individual, community and societal effects of the global financial crisis: a bioecological model” by francesca eleuteri and lisa saskia arduino aims to understand the individual, community and societal effects of the global financial crisis by examining several studies using the bioecological model. the analysis suggests that coping methods for the global financial crisis are highly associated with: social policies and labour market programmes, social support, employment status and individual vulnerability. the second paper “bill business risk and prevention of small and medium commercial bank”, by yanling hu, reviews main risks in bill business of small and medium commercial banks, analyzes the reasons of the risks, and suggests five preventive measures to them. many thanks to the referees for their strong support and kind help. and also thank the authors very much for all their submissions. editor-in-chief chongfu huang professor, beijing normal university email: hchongfu@gmail.com journal of risk analysis and crisis response, vol. 4, no. 4 (december 2014), 183 published by atlantis press copyright: the authors 183 atlantis press journal style optimization of ventilation and alarm setting during the process of ammonia leak in refrigeration machinery room based on numerical simulation dongliang liu, liang wang, tong wang, yan liu, ya chen beijing municipal institute of labor protection, beijing 100054, china e-mail: ldlbmilp@163.com abstract in order to optimize the ventilation effect of ammonia leakage in the refrigeration machinery room, a food processing enterprise is selected as the subject investigated. the velocity and concentration field distribution during the process of ammonia leakage are discussed through simulation of refrigeration machinery room using cfd software. the ventilation system of the room is optimized in three aspects which are named air distribution, ventilation volume and discharge outlet. the influence of the ammonia alarm system through ventilation is also analyzed. the results show that it will be better to set the discharge outlet at the top of the plant than at the side of the wall, and the smaller of the distance between the air outlet and the ammonia gathering area, the better of the effect of ventilation will be. the air flow can be improved and the vortex flow can be reduced if the ventilation volume, the number of air vents and the exhaust velocity are reasonably arranged. not only the function of the alarm could be ensured, but also the scope of the detection area could be enlarged if the detectors are set on the ceiling of the refrigeration units or the ammonia storage vessel. keywords: numerical simulation; ammonia leakage; refrigeration machinery room; ventilation optimization 1. introduction in china, according to the statistics, over 50% of the cold storages are built 30 years ago and more than 80% of the refrigeration machinery rooms choose ammonia as refrigerants 1. the count from state administration of work safety shows that, from 2005 till now, there have been 24 major accidents of ammonia leakage or explosion happened in china, causing 1171 people injured and 187 people died. ammonia leakage or explosion is becoming a sensitive issue because its influence to people’s life is far beyond its own damage24. many researchers studied the law of gaseous diffusion by using cfd methods. james s. bennett5 proved that balancing the air inlet and outlet can enhance the control of exposure and pollutant diffusion after simulating the painting process in navy aircraft paint warehouse using cfd software. galani6 studied the concentration of co in london using cfd software, finding that the calculated result has a tendency of overestimate because of the uncertainty of urban wind field. mazzoldi7 studied the law of co2 leakage by cfd simulation method, and the detailed data for the calculation of public safety distance was provided by his result. chao c8 proved that the cfd method can be used to the standard design by simulating the process of continuous diffusion and the ammonia absorption effect of water spray system. z.h guan9 simulated how the diffusion influencing factors such as facility layout and the obstacles position affect the diffusion process. overall ventilation system is the most important occupational disease prevention facilities in the ammonia refrigeration plant10. numerical simulation method is able to simulate the ventilation effect quantitatively, and it can provide effective guidance to the design and optimization of the ventilation facilities, journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 37 received 22 november 2016 accepted 16 january 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). so as to offer data support to the prevention, control and response of ammonia leakage accident in the workplace. in this research, by using cfd methods, the diffusion process is simulated and its influence factors are studied. consequences and evacuation after the accidents, ventilation and spraying and other emergency rescue measures are also discussed11. 2. selection of numerical model 2.1 solving equations during the process of ammonia diffusion, there is no chemical reaction and it belongs to single phase, multicomponent diffusion. therefore, continuity equation, momentum equation, energy equation, and component equation are established12. 2.1.1 continuity equation because the process of ammonia leakage is isothermal process, and the concentration of ammonia gas is different from the air density in the environment, so the continuity equation form selected is: ( ) 0 1i ixt uρ ρ∂ ∂ + = ∂∂ （） where ρ is the density of the mixed gas, kg/m3； iu is the time averaged velocity in different directions, m/s. 2.1.2 momentum equation in the inertial coordinate system, the general equation of momentum in direction is: ( ) ( ) ( ) 2iji j i i i i j iu t p u u g f x x c ρ ρ t ρ ∂ ∂ ∂ + = − + + + ∂ ∂ ∂ ∂ ∂ （ ） where u is the time averaged velocity in different directions, m/s; p is the absolute pressure, and the pressure in the momentum equation is ( )p gzρ+ , pa; µ is the viscosity of the fluid turbulence; g is gravitational acceleration, m/s2; igρ and if are the gravity volume force and other volume forces; i ju uρ is turbulence stress; ijt is stress tensor, defined as: 2 3 3 ji ij ij j i uu k x x t µ ρ δ   ∂∂ = + −   ∂ ∂    （ ） 2.1.3 energy equation because the component transfer model is used, the energy equation must be applied. and its form is: ( ) ( ) 1 4 i i pv pa t i t p i i p c i i u t x t t c c c t k c x x c x t x ρ ρ µ σ ∂ + = ∂ ∂ ∂  ∂  ∂    +    ∂ ∂ ∂ ∂    ∂  ∂ （ ） where t is time averaged temperature of fluid, k; tk is thermal conductivity of fluid, w/(m·k) ； pc is specific heat capacity of the mixed gas at constant pressure, j/(kg·℃)； pac is specific heat capacity of air at constant pressure in the environment, j/(kg·℃); pvc is specific heat capacity at constant pressure, j/(kg·℃); ic is mass concentration of mixed gases, kg/m3. and there are: r 5 p tt t k k µ σ µ   =     （ ） r 6p pc k µ = （ ） where rp is the prandtl number; tσ is constant, with interval from 0.9 to 1.0. put formula (5), (6) into formula (4), there is: ( ) ( ) 7 i i t pc pa t i i t i p c i i t u t x t c c c t x x x t c x ρ µ µ σ σ ρ ∂ + = ∂ ∂ ∂ −  ∂  ∂    +    ∂ ∂ ∂ ∂  ∂   ∂   （ ） 2.1.4 component equation multicomponent transport equation without chemical reaction is derived based on mass conservation of components. and its form is: journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 38 ( ) ( ) 8 jt j t i j d x x x ω ρµ ω ρρω ∂ ∂ ∂  + =  ∂ ∂ ∂  ∂ ∂ （ ） where ω is component mass fraction; td is turbulent diffusion coefficient. and there is: 9 c c t t d sd µ σ µ   =     （ ） where cs is the turbulent schmidt number; cσ is constant, with general value 1.0. with ( )cs dµ ρ= and formula (9), formula (8) is changed as: ( ) ( ) 10tj j i c jt x x x µ ω ρρ µ ωω σ ∂ ∂ ∂  + =  ∂ ∂ ∂  ∂ ∂ （ ） 2.2 project overview a refrigeration machinery room in a large food enterprise in beijing is selected as the research object. it is an old refrigeration machinery room and it covers an area of 550m2. its building structure is a layer of frame structure. in the east, west, and north of the building there are solid walls without windows and doors. on the south side of the wall, there are some windows, entrance and exits doors. the clearance height of the building is 7.7m. there are 2 high pressure ammonia storage devices, 2 horizontal evaporators, 2 low pressure circulating barrels and 2 chain axial flow fans. in this area, the average temperature is 29.9℃ in july, and the summer dominant wind is from the south, with the average speed of 1.9m/s. 2.3 geometric model parameters and assumptions of the refrigeration machinery room geometric model are as follows:  the space size is 17 meters long, 11 meters wide and 7 meters high; the doors size on both west and south side walls is 2.0m×1.5m; the size of the 6 windows on south side wall is 0.7m×0.7m; the diameter of the 2 fans is 0.6m.  the model is established according to the actual size and location of the storage tanks. the ammonia pipelines (except the pipelines with leakage), valves, pump body, operating platforms, the cofferdam structure of the liquid ammonia tanks are neglected. the bearing beams on the top of the workshop and the sidewalls except the parts within windows are all considered as the plane walls.  supposing that the valve sealing material of ammonia transmission line on the high pressure ammonia storage device has been aging, leading to the micro ammonia leakage. and it is completely volatilized after the leakage. the influence of ambient temperature and humidity is neglected.  the influence of the fans and the outside wind is considered. supposing that the wind go into and out of the room vertically, and the wind speed does not change with the height and time.  as is shown in figure 1, the three-dimensional model is built using gambit2.4 fig.1. 3d geometric model of the refrigeration machinery room. 2.4 boundary conditions and solution parameters the reference pressure, gravity and operating temperature are set as 101325pa, 9.8m/s2 and 301.2k. ambient air density defaults 1.225kg/m3. cfd simulation parameters and boundary conditions are set as following table 1 and table 213-15. table 1. calculation parameters setting computational model solving parameters turbulence model standard k ε pressure velocity coupling method simple solver pressure based solver discretizatio n scheme standard / first order upwind component transport model start-up convergence criterion 10 -3 energy equation start-up notes: the ammonia leakage rate is derived from the concentration data of the ammonia leakage accident in the enterprise in the past (according to the original data collected from the accident, the ammonia leakage rate is journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 39 about 13l/s, which can be exchanged as 0.01kg/s) and in this leakage rate, the monitoring data are in good agreement with the numerical simulation data. air exhaust rate is obtained according to the ventilation rate of 12times/h. table 2. boundary condition setting 3. analyses of numerical simulation results the effect of overall ventilation mainly depends on two factors: ventilation volume and the air distribution in the plant. this section studies the velocity and concentration distribution in the plant by changing the air distribution in the plant, the ventilation volume and the number of air vents. 3.1 the effect of air flow on the overall ventilation the main factors that affect the air flow organization are the position of the air supply and the air outlet and the way of the air distribution. in the most refrigeration machinery rooms, the air comes into them from the doors and windows. and the air trajectory is from the bottom to the top of the building. therefore, this section mainly considers the effect of the air outlet on the ventilation result. on the basis of the leakage rate and air exhaust volume setting in the former section, the air outlet is arranged on the side of the wall and the top of the building. figure 2 is about two vector maps of space velocity field. the setting of the air outlet is the main factor that affects the airflow distribution in the plant, if the ventilation volume and the fan position are definite in different cases. the more uneven distribution of air flow is, the more serious of vortex, the greater the diffusion range of ammonia will be, which indicates that the ventilation effect of exhaust fans is not significant. the air exhaust port is set on the ceiling of the work shop, above the leakage point, which can effectively reduce the diffusion range of ammonia gas. and it is beneficial to the discharge of ammonia gas. figure 3 shows two kinds of installation modes of the air vent. concentration distribution of the ammonia in the air after 10mins ventilation is shown in the picture. the position of air vent has a significant effect on the concentration distribution of ammonia gas in the whole space. the top exhaust can obviously reduce the ammonia accumulation, the ammonia concentration in the plant, and the diffusion range of ammonia. therefore, the effect of the top exhaust ventilation is better than that of the side exhaust. fig.2. flow field distribution of velocity vector with different exhaust position. fig.3. distribution of ammonia concentration with different exhaust position. position boundary type measureme nt turbulent coefficient hydraulic diameter ammonia inlet mass flow inlet 0.01kg/s 4.7% 0.05 exhaust fan exhaust fan 2.94kg/s 10% 0.6 door/ window pressure inlet 0pa 10% 1.7/0.7 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 40 3.2 the influence of ventilation volume on the overall ventilation effect according to code for design of cold storage (gb50072-2010)16, air flow rate of accident in ammonia refrigerating plant room should be 183m3/（m2·h）and the minimum exhaust air volume should not be less than 34000m3/h. according to code for heating ventilation and air conditioning design of civil buildings （ gb50736-2012）17, the air flow of the accident should be determined by calculation according to the process design requirements. and the ventilation should not be less than 12 times /h. on the basis of the top exhaust setting, the changes of the ammonia concentration field and the velocity field along with the changes of the ventilation rate are observed. the influence of air distribution on the overall ventilation effect is analyzed. two ventilation plans are as follows, 1) ventilation 12 times /h; 2) 34000m3/h. figure 2 (b) and figure 4 are vector maps of space velocity field under the two ventilation plans. if the ventilation volume increases, the airflow rate will increase at the air inlet, the air velocity in the whole plant will be greater, and the air distribution will be more uneven, which will be more likely to form a local vortex at the corner of the wall. in the same condition of the installation of air inlet and air outlet, the ventilation volume mainly affects the air flow and the air velocity of the plant. increasing the ventilation volume, which is conducive to the discharge of ammonia, will make the gas flow inside the plant more complex, however. fig.4. flow field distribution of velocity vector when ventilation volume is 34000 m3/h. figure 3 (b) and figure 5 are spatial distribution maps of the ammonia diffusion (400s after the leakage) under the two ventilation plans. increasing the amount of ventilation volume can significantly accelerate the discharge and reduce the accumulation of ammonia. but in a certain amount of air vents, the greater the air flow, the more complex the airflow inside the plant, the easier it is to form the local eddy current, which will increase the diffusion range of ammonia gas. fig.5. distribution of nh3 concentration when ventilation volume is 34000m3/h. 3.3 the influence of the number of air vents on the ventilation effect on the basis of the setting of top exhaust and 12 times ventilation mentioned above, ammonia concentration and velocity field distribution are studied by the simulation calculation under the settings of different number (1, 2 and 3) of air vents. influence of the number of air vents on the ventilation effect is also analyzed by numerical simulation. according to the cfd analysis report, the exhaust rate are 16.9m/s, 8.5m/s and 4.3m/s corresponding to the three kinds of settings of the air vents. figure 6 is the vector map of velocity field of different number of air vents (1, 2 and 3). if the ventilation volume is certain, the more the number of air vents is, the smaller the air exhaust rate will be. when the air vent number is 1 and 3, the eddy current is formed in different degrees. when the air vent number is 2, the intake airflow distribution is more uniform and the eddy current is the least, which proves that setting the number of air vents and exhaust velocity reasonably can improve the uniformity of the intake air flow and reduce the eddy current, so as to improve the ventilation effect. figure 7 (a), figure 3 (b) and figure 7 (b) are the overall spatial concentration distribution of ammonia diffusion in corresponding with number of air vent 1, 2 and 3. the case comparison shows that when there are 2 air vents, the ammonia accumulation and the diffusion area are the smallest, and the ventilation effect is the best. the ammonia gathers at above the leak point. the faster the exhaust air velocity, the closer the distance from air outlet to the gathering area, the more easily the ammonia gas is discharged. it is proved that if the ventilation volume is certain and setting the number of air vents and exhaust velocity reasonably can reduce the eddy current so as to improve the uniformity of the air inlet and the ventilation effect. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 41 fig.6. flow field distribution of velocity vector in the surface of z=5.5m fig.7. distribution of nh3 concentration with different number of outlet 4. ammonia detection and alarm setting according to the code for design of combustible gas and toxic gas detection and alarm in petrochemical industry（gb50493-2009）18, ammonia is poisonous gas. therefore, ammonia refrigerating plant room should be set with fixed ammonia gas detection alarm and control system according to the specification. so that the accident can be found and the ventilation system can be paused-on in time to reduce the loss. 4.1 setting of the alarm value according to the data access, as is shown in table 319, there is no uniform provision on the alarm setting value of all countries and regions. the ammonia alarm value in table 3 is set based on the code for design of combustible gas and toxic gas detection and alarm in petrochemical industry in china. and the second stage alarm value is much higher than that in other developed countries and regions. code for design of cold storage is a national code for the development of refrigeration machinery room, which requires that the alarm signal should be automatically sent out, and the accident exhaust fan should be automatically opened when the ammonia concentration reaches 100ppm or 150ppm. in the engineering design, the ammonia alarm value of the first stage in the refrigeration machinery room is set as 40ppm based on code for design of combustible gas and toxic gas detection and alarm in petrochemical industry. and the ammonia alarm value of the second stage is 150ppm. this kind of setting greatly reduces the alarm value of the second stage. and the safety of the personnel is ensured to a great extent. table 3. alarm point of nh3 in different countries or regions country/region value of the first stage/ppm value of the second stage/ppm china 40 100 or 150 america 100 or 150 hongkong 25 250 japan 50 200 4.2 setting position of the detector code for design of combustible gas and toxic gas detection and alarm in petrochemical industry 4.2 requires that the detectors should be set above the release sauce with the distance less than 1m and toxic gas detectors should also be set at the ceiling of the plant. 6.1.2 requires that installation height of a detector that detects toxic gases lighter than the air should be higher than the release source, with the distance of 0.5~2m. code for heating ventilation and air conditioning design of civil buildings 7.2.1 requires that ammonia detectors should be installed on the ceiling of the refrigeration machinery room, which is above the ammonia refrigerating unit and ammonia storage container. the standards are not uniform in china. so in the engineering design, there are many ways to set the ammonia detectors in the refrigeration machinery room. in the most of the situations, the detectors are set at the ceiling or above the release source with the distance of 1~2m. in europe, japan and hong kong, the ammonia detectors are set at the ceiling in the refrigeration machinery room19. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 42 in the refrigeration machinery room researched in this paper, two detectors are set at above the high pressure ammonia storage device, horizontal evaporator and low pressure buffer barrel pipeline valve (with the distance of 1.8~2.0m) respectively. when the detection points s1 and s2 at above the high pressure ammonia storage device move to the detection points s1-1 and s21at the ceiling of the plant, the change curve of the ammonia concentration along with time is obtained by the optimization model. fig.8. the change of nh3 concentration along with time at the top of the high pressure ammonia storage device from the time beginning to get the presence of ammonia to the time in theory to cause the alarm, the sequence of time consuming of the four detection points are as follows: s2<s2-1<s1-1<s1. the time it takes in theory at the detection points s2 and s2-1 are 3.2s and 11.5s with the time difference of 8.3s. while the response time of the ammonia detector equipment is about 30s ， indicating that the time for ammonia diffusion is much less than that for detector response. therefore, whether the ammonia detector is set at the ceiling of the plant or at above the equipment, the impact it brings on the initiation response time of the alarm and exhaust system is little. as long as the exhaust system can operate normally, the ventilation effect can be guaranteed. setting the detector at the top of the plant has the advantage of increasing the detecting area. 5. conclusions  changing the air flow organization can obviously improve the ventilation effect, the more uniform airflow distribution in the workshop, the less the vortex, the more conducive to the discharge of ammonia, the more effectively the ammonia diffusion region can be reduced. the ventilation effect will be better if setting the air outlet at the top of the plant than that at the side walls of the plant. reducing the distance between the air outlet and the ammonia gathering area can improve the ammonia removal efficiency.  increasing the ventilation rate can obviously accelerate the discharge of ammonia gas, but it will increase the flow of air in the plant to form a vortex, and increase the diffusion region of ammonia gas. setting the number of air vents and the exhaust velocity reasonably can reduce the vortex and improve the airflow uniformity and the ventilation effect.  in the ammonia refrigeration machinery room, the first and the second level of ammonia concentration alarm value can be set as 40ppm and 150ppmrespectively.when the first level alarm is activated, the ventilation system will automatically start and when the second level alarm is activated, all the equipment and facilities of the refrigeration unit will automatically shut down to ensure the safety of the workers.  setting the ammonia detection device at the ceiling above the ammonia refrigerating machine and the ammonia storage container in the refrigeration machinery room, not only can meet the alarm requirement, but also can increase the detection region. references 1. zhen bing, wang lei, guo jing, et al. study on diffusion law of ammonia leakage in cold storage. heilongjiang science, 2015,10:4-7 2. wu xintian. application of numerical simulation on diffusion of liquid ammonia leak. tianjin: tianjin university of technology, 2014. 3. zhao qingyue. analyzation on the case of particular serious fire and explosion on 3rd jun. baoyuanfeng poultry industry ltd. in changchun of jilin province. jilin labor protection, 2013, 7:32-35. 4. approval on regarding the agreements of “investigation report of major ammonia leak accident on 31th aug. in weng refrigeration industrial co ltd. shanghai” by shanghai municipal government, no. 90 [2013] hu. 5. james s b, david a m, duane r. h, et al. experimental and numerical research on the performance of exposure control measures for aircraft painting operations, part i. ephb report 2011, no.329-12a. 6. galani a, neofytou p, venetsanos a, et al. prediction and research of pollutant dispersion in a street canyon in london using computational fluid dynamics techniques. global nest journal, 2009,11(4):434-439. 7. mazzoldi a, picard d, sriram p g, et al. simulation based estimates of safety distances for pipeline transportation of carbon dioxide. greenhouse gases: science and technology, 2013,3(1):66-83. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 43 8. zhao huijun, zhang qingsong, zhou shidong, et al. study on the numerical simulation of gas leakage in gas tank area based on phoenics. china safety science journal, 2007,02:39-43+3. 9. chao c, wei t, liu ly, numerical simulation of water curtain application for ammonia release dispersion. journal of loss prevention in the process industries. 2014,30:105-112. 10. chen jiang. study on the evaluation index of protection effect for ventilation facilities .journal of safety science and technology, 2013.05:63-66.guan 11. zhong hui. liquid ammonia station ventilation safety and influence on leakage and diffusion. harbin: harbin university of science and technology, 2013. 12. sorensen dn, nielsen pv. quality control of computational fluid dynamics in indoor environments .indoor air, 2003, 13: 2~17. 13. wang zhipeng. numerical simulation and emergency response measure for ammonia leakage. beijing: china university of geosciences, 2013. 14. yan yunfei. numerical simulation of the influence between spray system and the leaking ammonia accident in the refrigeration station. harbin: harbin university of science and technology, 2013. 15. yu jiashou. liquid ammonia leakage diffusion law and the study of evacuation in open space. beijing: china university of geosciences, 2014. 16. gb50072-2010, code for design of cold storage. beijing: standards press of china, 2010. 17. gb50736-2012, code for design heating ventilation and air conditioning in civil buildings. beijing: standards press of china, 2012. 18. gb50493-2009, code for the design of combustible gas and toxic gas detection and alarm for petrochemical industry. 19. shi jichun. experience learned on the design of explosion-proof and gas-defense in the ammonia refrigeration machinery room. china refrigeration association refrigeration freezing specialized committee. in 2011 the refrigeration industry and shandong refrigeration and air conditioning industry annual conference and green low-carbon new technology seminar paper. china refrigeration association refrigeration freezing specialized committee: 2011:6. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 37–44 ___________________________________________________________________________________________________________ 44 1. introduction 2. selection of numerical model 2.1 solving equations where is component mass fraction; is turbulent diffusion coefficient. 2.2 project overview 2.3 geometric model 2.4 boundary conditions and solution parameters 3. analyses of numerical simulation results 3.1 the effect of air flow on the overall ventilation 3.2 the influence of ventilation volume on the overall ventilation effect 3.3 the influence of the number of air vents on the ventilation effect 4. ammonia detection and alarm setting 4.1 setting of the alarm value 4.2 setting position of the detector 5. conclusions atlantis press journal style editor’s introduction for volume 5, issue 3 this issue contains six papers. there are four contributions written in english and two contributions in chinese with english abstracts. the papers can be divided into three topics: emergency management, risk perception and risk assessment. there are two papers in emergency management. the first paper “dealing with emergencies: the case of a heavy disruption of the mexico city metro system” by diego padilla-pérez, jaime santos-reyes and samuel olmos-peña, presents the results of a forecasting model associated with the affluence of users of the metro line-b of mexico city's metro system. it also presents in a way a retrospective analysis of the metro incident that occurred on september, 2011, in the same metro line; the incident affected seven metro stations and about 17 thousand commuters. in the second paper “agent-based simulation of fish boats evacuation”, authors, hanping zhao, huiyan ding and han wang, suggest a fisher boats evacuation simulation model with considering individual desire of back harbor. in the agent based model, typhoon and harbor are inactive agents, government and fisher boats are active agents. fisher boats make decisions whether back to harbor and which harbor to be choose according of typhoon forecast information and government instruction. there are two papers in risk perception. the first paper “spatial effect on public risk perception of natural disaster: a comparative study in east asia” by zhongyu he and guofang zhai, explores the effect of geographical location on public risk perception of natural disasters. by conducting an identical questionnaire survey across china, japan and south korea, the authors discovered that different country has its unique structure of risk perception. generally, the risk perception of sample residents in japan weakens as the distance from the risk source increases, which indicates japanese people’s risk perception reflects the actual risk probability. on the other hand, korean partially and chinese hardly perceive the probability of existing risks. the second paper “crisis communication about nuclear accidents with psychological approaches” by yanran yang, lina jin, jinbin li and chao fang, analyzed risk and crisis communication in three nuclear accidents: chernobyl, three mile island and fukushima. this paper also studies the public risk and crisis communication after these accidents with psychological methods. it shows that new media such as wechat and microblog play very important role in this process. there are two papers in risk assessment. the first paper “evaluating the three methods of goodness of fit test for frequency analysis” by xiankui zeng, dong wang and jichun wu, applies three methods of goodness of fit test that include chi-square (c-s), kolmogorov-smirnov (k-s), and anderson-darling (a-d) tests. the results of power test indicate that the most powerful tests for normal, uniform, p3, and weibull distribution are k-s, c-s, and a-d tests, respectively. the test method with the best comprehensive power is c-s test, followed by k-s and a-d test. the second paper “a study on spatial-temporal rainstorm risk at civil airports in china” by xiaomei guo, xiaobing hu, hang li and zhen xu, carries out a study on risk analysis of rainstorm at chinese civil airports, based on daily precipitation data of 174 meteorological stations near civil airports in china, flight data over the same period (1994-2013) and other relevant airport data. this paper uses the platform of arcgis, analyzes the four essential factors of disasters, combines the traditional model of meteorological disaster risk assessment and characteristics of chinese civil aviation, and then develops monthly rainstorm disaster risk zoning maps for civil airports in china. we sincerely thank the referees for their strong support and kind help. also, very much thanks to the authors for all their submissions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com journal of risk analysis and crisis response, vol. 5, no. 3 (october 2015), 141 published by atlantis press copyright: the authors 141 atlantis press journal style a critical review on the economics of disasters* xi yu1,2, yandong tang1 1. institute of disaster prevention, sanhe hebei 065201, china 2. institute of engineering mechanics, china earthquake administration, haerbin heilongjiang150090, china e-mail: puyuxi@126.com abstract the idea of economics of disasters can be traced back to the mid-19th century. the field of economics of disaster continues to be expand after 1970s and mainly concentrating in the disaster loss assessment methods, disaster economic impact model, the value of life assessments and disaster insurance and disaster risk securitization. this paper presents the research framework of economics of disasters, including disaster damage assessment, mitigation theory, disaster short-term economic impact, relationship between disasters and long-term economic growth, the value of life assessment, disaster risk management and disaster recovery and reconstruction theory. keywords: economics of disaster, loss assessment, value of statistical life, economic model *supported by science and technology planning project of hebei province, china（15275412） the fundamental research funds for the central universities（zy20160103） 1. introduction tsunamis, droughts, earthquakes, hurricanes and floods are a constant threat to society. these hazards affect human beings in many ways, for example, they can destroy houses, factories, buildings and infrastructures, and they can cause casualties, bring psychological damage, even destroy the ecological environment. some disaster losses can be quantified in currency easily, such as the loss of housings, buildings. but some losses are difficult to evaluate in money, such as environmental disruption, mental trauma and inconvenience of life, etc. although these losses is difficult to calculate statistically, there is a growing concern for the environment, health, safety, eco-systems and other related problems in post-disaster situations. economics of disaster is a subdiscipline of economics that applied basic principles and methods of modern economics to study the economic relations between human society and disaster. it studies disasters from an economics’ perspective and deals with problem that how to configure scarce resources under disaster conditions. it studies the economic relations between disasters and corporate, family or individual, relations between disasters and sectors of the economy, relations between disasters and macroeconomics. 2. the origin of economics of disasters earlier research on economics of disasters is unsystematic. the idea of economics of disasters can be traced back to the late of 18th century. british economist thomas robert malthus predicted that the rise in population would lead to widespread famine and catastrophe in his famous an essay on the principle of population. john stuart mill argued that all traces of the mischiefs done by earthquakes, floods, hurricanes and war will disappear in a short period of time in his principles of political economy with some of their journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 27 received 30 november 2016 accepted 24 january 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). applications to social philosophy published in1848. mill though the perpetual consumption and reproduction of capital affords the explanation of what has so often excited wonder, the great rapidity with which countries recover from a state of devastation [1].almost a century later, john kenneth galbraith corroborated mill's observations. he investigated the impact of allied bombing raids of hamburg and concluded that the raids had had little impact [2]. with continued population growth and rapid economic development since the 1950s, industrial waste and pesticide pollution has become increasingly serious, which raised human awareness of environmental protection. arthur cecil pigou developed alfred marshall’s concept of externalities and used it to analyze environmental disasters, such as air and water pollution. pigou argued that the existence of negative externalities is sufficient justification for government intervention. he advocated a tax, now called pigovian taxes, on such activities to discourage them. ronald harry coase suggested that well-defined property right could overcome the problem of externalities. coase theorem states that if trade in an externality is possible and there are no transaction costs, bargaining will lead to an efficient outcome regardless of the initial allocation of property rights. jack hirshleifer was the first economist that studied disasters deeply, he analyzed the choices people make in times of disaster and conflict to explain the possibilities and limits of human cooperation under severe environmental pressure [3]. howard kunreuther, a professor of wharton school of the university of pennsylvania, published an article entitled the alaskan earthquake: a case study in the economics of disaster in 1966. the term economics of disasters was used for the first time in this paper [4]. in 1967, kunreuther, conducted an empirical analysis to study the price changes of emergency supplies in the peculiar economics of disaster. he concluded that most supply and demand problems facing an area hit by a natural disaster are generally short-run in nature because of the aid forthcoming from outside regions. even when the threat of shortages does exist, the concern of residents in the community for the plight of others helps to minimize serious problems during the emergency period. these short-term response patterns indicate that sociological and psychological factors change to utility functions of residents during the short-run period and thus explain why economic behavior following a disaster appears at first glance to be somewhat peculiar [5]. dacy and kunreuther published the book the economics of natural disasters: implications for federal policy on the basis of previous studies in 1969, following the national flood insurance act of the unitedstates in 1968 and devastating losses from the alaska earthquake in 1964 [6]. in this book, the authors claimed that the main objective of the book is “to formulate a clear-cut case for the development of a comprehensive system of disaster insurance as an alternative to the current paternalistic federalpolicy” [6]. some others, for example, sorkin( 1982) and albalabertrand (1993), aim to offer the generalized framework of disaster analysis, but they are yet oriented to investigate empirical cases and/or to provide the empirical modeling frameworks for the analysis, and lack the theoretical development and/or analysis of disasters and their impacts to economy [7,8]. in china, the well-known economist yu guangyuan first pointed out the importance of studying disaster from an economic point of view in the 1980s. subsequently, zheng gongchenghe aiping and tang yandong published their works to discuss the basic principles of economics of disasters, and initially establish a theoretical framework [9-11]. in addition, many scholars have also published papers to discuss the issue related to economics of disasters [12-15]. 3. recent advances in economics of disasters the economics of natural disasters has gotten initial development during the 1970s. the research is heavily focused on the direct economic losses assessment or business interruption losses evaluation caused by hazards. in the 1990s, the research of this field have developed and deepened. the content of economics of disasters research is enriching and extending from direct loss to indirect loss, economic effect models, value of statistical life, recovery theory and disaster risk management. the research method has also reached an unprecedented level of innovation. these researches mostly concentrated in the disaster loss assessment, risk analysis, value of life assessments, hazard insurance and disaster risk securitization. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 28 3.1. disaster loss assessment timely and accurate assessment of losses caused by disasters can help us to determine disaster spatial extent, severity and distribution of loss, affected object and proportion. it’s very importance for emergency rescue, recovery and risk management. in addition, loss prediction is the foundation to determine risk severity and type in risk analysis. research focuses on basic concepts of disaster loss and loss assessment framework proposed by international organizations, countries, institutions or scholars. national research council of us completed a report the impacts of natural disasters: a framework for loss estimation in 1999 [16].in order to standardize the methods and process of loss assessment, economic commission for latin america and the caribbean published report: manual for estimating the socioeconomic effects of natural disasters [17] and the organization revised the report in 2003. the revised one entitled handbook for estimating the socio-economic and environmental effects of disasters [18]. in these two handbooks, the basic concepts, methods of disaster losses and sectors damage assessment method are described. 3.1.1. direct and indirect loss distinction of direct and indirect losses seem to be simple, in fact, it is a controversial problem. some scholars distinguish direct and indirect losses based on the relationship between losses and hazards or on the chronological sequence [1]. such as, cochrane (2004) argued that direct loss is a loss linked directly to disaster, indirect loss is any loss other than direct loss. some scholars believe that direct losses are losses that occur on the spot of disasters. these are not valid criterion for distinguishing them. for example, earthquake destroyed the gas pipelines, gas exploded and burned buildings. the damage of buildings linked earthquake indirectly, but the loss should be direct loss. economics, business, accounting, and related fields often distinguish between quantities that are stocks and those that are flows. a stock variable is measured at one specific time, and represents a quantity existing at that point in time. a flow variable is measured over an interval of time. many economists believe that the distinction between direct and indirect losses should be based on the principle of the stocks and flows. direct losses relate to the damage to the stock of capital assets, including building, infrastructure, industrial plants, machinery equipment, and inventories of finished, intermediate, and raw materials destroyed by hazards. property damage represents a decline in stock value and usually leads to a decrease in service flows. indirect losses refer to damage to the flow of goods and services including lower output of business interruption, and the increased costs associated with the use of more expensive inputs because of the destruction of cheaper usual sources of supply. adam rose disagrees the view of stocks and flows. he suggests that the value of an asset is the discounted flow of net future returns from its operation. so the measures both stocks and flows represent the same things, including both would involve double-counting. rose argues that flow measures are superior to stock measures and proposes the use of the term “higher-order effects”. this point of view is suitable for overall equity that can produce cash flow. we cannot calculate the flow of a simple equipment, such as computer, transport vehicle, and weld equipment, because they cannot produce output alone. a secondary disaster caused by a specific hazard can also damage property that pertains to direct losses, for example, earthquake destroyed the gas pipelines, gas exploded and burned buildings. the damage of buildings is a part of direct losses. so direct losses consist of two more refined types of losses. primary direct losses are those resulting from the primary hazard, such as shake damage from an earthquake or water and wind damage from a hurricane. secondary direct losses are those additional impacts resulting from secondary disasters, such as the building burned by fire following an earthquake (fig. 1). for example, more than half of the 700 thousands buildings destroyed in great kanto earthquake in 1923 were damaged by fire. indirect losses are the losses of flow. primary indirect losses are induced by a company's own property damage by primary or secondary disasters, for example, lost production stemming from direct losses of factories, equipment, and instruments. primary indirect losses relate to business interruption, which means a flow. the extent of primary indirect losses does not stop here, but sets off a chain reaction. lower production of a factory may affect the output of other factories through forward and backward linkages. we can use the concept of linkage indirect losses to distinguish them from primary indirect losses (fig. 2). journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 29 fig. 1. primary direct losses and secondary direct losses fig. 2. compensating variation (cv) and equivalent variation (ev) 3.1.2. economic assessment indices hazards can destroy the infrastructure, such as transportation, water supply, power supply, communication system, hamper the normal operation of society, and deteriorate human living environment. hazards affect people's welfare by influencing the level of utility. therefore, the measure the disaster loss may be changed into welfare changes. on the one hand, the disasters may change the supply or demand of the commodities, then the price change can affect the welfare of the people. on the other hand, people's income levels can decrease caused by hazards. in addition, the disasters have an effect on the quantity or quality of some market/ non-market goods, such as quality of water, traffic, air quality. so welfare economics provides the theoretical basis of the disaster damage assessment. indicators to measure the welfare changes include consumer surplus (cs) and producer surplus (ps), compensating variation (cv) and equivalent variation (ev), compensation surplus (cs) and equivalent surplus (es) [11]. at the background of economics of disasters, compensating variation refers to the amount of additional money a consumer would need to reach its initial utility after a change in prices after disasters. o u0 a e b u1 x1 d · · · e′ x2 ev cv · c b b′ damaged property primary direct losses fire （secondary disaster） indirect losses earthquake （primary disaster） indirect losses damaged property secondary direct losses journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 30 equivalent variation is a measure of how much more money a consumer would pay before a price increase to avert the price increase (fig. 2).where eb and eb′are the old and new budget lines before and after the change in price. u0 and u1 are the old and new utility levels respectively. cv and ev can be defined by the solution of indirect utility function as follows. ' 1 2 1 2 0, , , ,v p p y v p p y cv u= − =（ ） （ ） ' 1 2 1 2 1, , , ,v p p y ev v p p y u+ = =（ ） （ ） where 1 2, ,v p p y（ ） is indirect utility function, y is income, p1 and p2 are the prices of the two goods, p1′ is affected price of goods one by disasters. for the price change caused by disasters, compensating variation is simply the area under the hicksian demand curve evaluated at the initial utility level and the two prices. similarly, equivalent variation is simply the area under the hicksian demand curve evaluated at the new utility level and the two prices. fig. 3 depicts these two measures for the price change. 1 1 ' 1 1 1 2 0 1 1 1 2 0 1' 1 ( , , ) ( , , ) p p h p p e p p u cv dp x p p u dp p ∂ = = ∂∫ ∫ 1 1 ' 1 1 1 2 1 1 1 1 2 1 1' 1 ( , , ) ( , , ) p p h p p e p p u ev dp x p p u dp p ∂ = = ∂∫ ∫ where 1 2 0( , , )e p p u is expenditure function at utility level u0, and x1 h is hicksian demand function. the relation of three indices cv, cs and ev is as follows. cv a cs a b ev a b c = = + = + + when consumers can choose the purchase of the two commodities freely, they will choose the quantities of two goods determined by the point of tangency of the budget line and indifference curve to reach maximum utility. however, in some cases, consumer cannot choose goods freely, for example, under the condition of shortage of foodstuffs after disasters, and the utility maximization conditions will change. the quantity of goods determined by the point of intersection of budget line and indifference curve will be chosen. compensating surplus and equivalent surplus can be used in these cases. these two measurements are closely related to the compensating variation and equivalent variation respectively. the only difference is the restriction on adjusting the purchase of goods. fig. 3. the relation of cv, cs and ev 3.1.3. disaster economic impact model since the pioneering work of dacy and kunreuther and cochrane, significant progress has been made in recent years for economic analysis of natural disasters, especially in the field of modeling economic impacts of disasters in a regional context. the recent advancements focus on empirical analysis and modeling extensions and modifications to fit them to disaster situations. many problems remain unsolved, such as the treatment of time, space, and counteractions. input-output (i-o) model, social accounting matrix (sam), computable general equilibrium (cge) model and econometric models are widely adopted to estimate indirect loss caused by various hazards. i-o model may be the most widely used model framework to estimate natural and manmade disasters (cochrane, 1974, 1997; wilson, 1982; kawashima et al., 1991; boisvert, 1992; gordon and richardson, 1996; rose et al., 1997; rose and benavides, 1998; and okuyama et al., 1999). as early as 1974, harold cochrane used the model to assess the earthquake damage. assessment of the man-made disasters can be traced back to the strategic bombing in world war ii and extend to the world trade center terrorist attacks. the u.s. federal emergency management agency and the u.s. national academy of building research jointly developed hazus loss assessment system also use input-output method to assess the indirect losses. the o p ' 1p x1 x1（p,y） xh1（p,u1） a c xh1（p, u0） b p1 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 31 http://en.wikipedia.org/wiki/measurement main reason that input-output model is popular is that it can reflect the regional economic interdependence within the relationship. furthermore, the model is relatively simple and easily modified and integrated with other models. the simplicity of the io model creates a set of weaknesses, including its linear structure, its rigid coefficients, no supply capacity constraint, no response to price change, a lack of explicit resource constraints. because of this, it is generally believed that the model will overestimate the impact caused by disasters. although these limitations of io model, a number of scholars have utilized and improved it to assess disaster indirect loss. in recent years, social account matrix (sam), lagging the expenditure model, inter-industry time-series model (sim) and the regional econometric input-output model (reim) have been developed based on basic input-output model [24]. computable general equilibrium model (cge) was developed because of the defects of input-output model. cge model overcomes some of the shortcomings of the input-output method, it is non-linear, can resolve the interaction between quantity and price and it take into account resource constraints, the utility maximization problem. cge model can simulate external shocks on the economy [25]. 3.1.4. nonmarket effects evaluation disasters can cause a variety of effects. some of them have market price and can be traded in the market, such as property loss, income reduction and yield decline. some impacts, called ‘nonmarket effect’, are particularly challenging to value in monetary terms because their very nature is difficult to measure and quantify. examples of nonmarket effect include death and injury, environmental damage, damage to cultural artifacts, and losses of memorabilia (photographs, books, toys and personal original work). non-market impact caused by disasters does not have market value, its value cannot or difficult to be measured in the market. there are, however, a variety of non-market valuation methods that can be used to assess the value people attribute to nonmarket natural disaster impacts. care is needed when using these methods to estimate the value of disaster impacts, as they can be complicated and time-consuming. if the methods are beyond the time and resource capacities available, simply list the important nonmarket effect of a natural disaster in as much detail as possible. we will start simply with the most obvious, most easily valued impacts. this may mean looking for impacts resulting in changes in productivity that can be valued using market prices. river pollution may disrupt a traditional downstream fishery or some agricultural activity. the net change in fish or crop production can be identified and valued. the change in the quality of water flowing downstream and its effect on the coastal mangroves or on offshore coral reefs is a secondary effect. secondary effects may be very important, both ecologically and economically, but the analyst would do best to start with the fishery or the agricultural activity. the other two categories of non-market valuation methods are known as ‘revealed preference methods’ and ‘stated preference methods’. revealed preference methods include defensive expenditure approach, hedonic price method and travel cost approach. these methods estimate the environmental and other nonmarket products by observing people's market behavior, to "expressed preferences". the principal difference between revealed preference and stated preference methods is that the latter draw their data from people’s responses to hypothetical questions rather than from observations of real-world choices. stated preference methods gained popularity after the two major non-use values, namely, option and existence values, have been recognized as important components of the total economic values in environmental economics literature, especially during the 1960s. while the conventional revealed preference method cannot assess these non-use values (smith, 1993), the only method that is identified for estimating these values is the stated preference method. this method elicits willingness to pay (wtp) information that cannot be inferred from markets. the contingent valuation method (cvm) is a widely used nonmarket valuation method especially in the areas of environmental cost–benefit analysis and environmental impact assessment. hence, a considerable amount of studies on cvm have emerged in the economic valuation literature, including a large number of studies criticizing the cvm method. the criticism revolves mainly around two aspects, namely, the validity and the reliability of the results, and the effects of various biases and errors. table 1 shows the three nonmarket impact valuation methods and disaster impact can be assessed. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 32 table 1. nonmarket impact valuation methods methods disaster impact market method changes in productivity productivity replacement/restoration costs capital assets natural resource cost of illness health(mortality) human capital health(mortality) revealed preference method preventive expenditures health, productivity, capital assets, natural resource hedonic price method hedonic property pricing method environmental quality hedonic wage method health travel cost method natural resources stated preference methods contingent valuation method health natural resources choice modeling method 3.2. value of statistical life as early as the late 1970s and early 1980s, there is literature to assess the value of statistical life in the united states. but policy makers used human capital approach to calculate the income loss to assess the value of human life and thought it is immoral to assess the value of life. this continued until the reagan administration. human capital approach makes economic sense when how much compensation should pay for dependents in wrongful death settlements. but it is incorrect in principle because it ignores the individual's own welfare, preference, and wtp and it defines vsl in a narrow way. viscusi used wtp to assess the risk reduction benefit in cost-benefit analysis of an expensive regulation firstly. from then on, wtp approach has been accepted to estimate the value of a statistical life by scholars rapidly [19-22].individuals make decisions every day that reflect how they value health and mortality risks, such as driving an automobile, smoking a cigarette, crossing the street. many of these choices involve market decisions, such as the purchase of a hazardous product or working on a risky job. using evidence on market choices that involve implicit tradeoffs between risk and money, economists have developed estimates of the value of a statistical life (vsl). vsl is the marginal willingness to pay (mwtp) for a small reduction in the probability of death during a given period or how much compensation that individual would require to accept a small increase in that probability. in fact, the mwtp is also the marginal rate of substitution (mrs) between goods and risk of death, as follows: , 0 lim (1)w p dp vsl mwtp mrs dπ π π π∆ → ∆ = = = = ∆ where p is payment,π is the probability of death. if a person is willing to spend $100 to reduce the ten-thousandth of the probability of death suffering from hepatitis a, then the value of his life will be one million dollar. assume that an individual obtains utility from wealth, the utility of death is zero, the state-dependent expected utility is ( ) (1 ) ( )e u u wπ= − ⋅ where w is wealth or income, ( )u w is utility to consume wealth w. we can obtain an expression for the individual's vsl, as follows ( ) (2) (1 ) ( ) u w vsl mu wπ = − where mu (w) is marginal utility. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 33 equation 2 shows that individuals' marginal willingness to pay depends on their wealth and survival situation. all other things being equal, the rich have higher total utility and lower marginal utility and then they have higher vsl. another results is that the higher the risk of death is, the higher will be the marginal willingness to pay to reduce the mortality risk. some scholars argue that assessing the value of life violates ethical because life is priceless. but if human life is priceless, all the mitigation measures regardless how expensive are worth taking. such an assertion does not make sense economically. under realistic conditions, we have to make trade-offs between changes in the probability of death and other goods that have monetary values in decision making. vsls are often used in cost-benefit analysis; but they are not a measure of what is "lost" when a person dies. what people "buy" and "sell" is small changes in the probability of dying and is not life or death. the economic approach of vsl avoids the issue of valuing life. vsl focuses on risk. the concept of vsl is very easy for us to associate it with compensation for deaths. some scholars believe that the estimates of vsl can become the basis of compensation for wrongful deaths. in fact, this is a misconception because the concept cannot applied under conditions of certainty. fig. 4. possible long-run impact of a disaster on gdp per capita 3.3. disaster and economic growth an important research question is relations between disasters and long-term economic growth. this question has been investigated empirically and theoretically by, e.g., albala-bertrand (1993), benson (2003), and skidmore and toya (2002), hallegatte and dumasa (2009), but results are not conclusive. albala-bertrand (1993) found, in a statistical analysis of 28 disasters from 1960 to 1976 in 26 countries, that the long-run growth rate is unaffected by natural disasters. benson (2003) carried out an analysis on 115 countries and found that the growth rate was lower in the countries that experienced more disasters. the analysis of skidmore and toya (2002) shows that climatic disasters are positively correlated with economic growth, whereas geologic disasters are negatively correlated with growth. the report of the world bank show that the growth rate can be like these (fig. 4). time gdp/capita disaster time disaster time time disaster (a) （b） （c） （d） o o o o disaster gdp/capita gdp/capita journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 34 3.4. disaster insurance and catastrophe risk securitization in the past few decades, our understanding of the extreme events or natural disasters has been greatly improved. natural disaster management system including hazard map, building codes, the government emergency response to natural disaster management system, commercial insurance, has played an increasingly important role in disaster prevention and reduction of losses. risk transfer financing methods include insurance financing and non-insured financing. insurance is an economic institution that allows the transfer of financial risk from an individual to a pooled group of risks by means of a two-party contract. aware of insurers’ need for more capacity to finance catastrophe risk, entrepreneurs have been devising solutions that would spread catastrophe risk to investors by “securitizing” it. that is, entrepreneurs have been developing means of packaging insurers’ catastrophe risk as securities that could be sold to investors. such solutions, however, will only be successful if they simultaneously meet insurers’ need to spread risk efficiently while offering investors opportunities to improve the performance of their portfolios. the securitization of catastrophe risk has taken several forms, each with advantages and disadvantages. primary insurers can use all of the forms of securitizing insurance risk to supplement traditional reinsurance, and reinsurers can use them to supplement traditional retrocessions. large self-insureds may also be able to use securitization to share their catastrophe risk with investors. to date, the principal forms of securitization include contingent surplus notes, catastrophe or “act of god” bonds, and exchange traded catastrophe options. 4. conclusion the theoretical discussion and analysis of disaster is important to advance our knowledge for the impacts and consequences of a disaster. the new and special discipline of economics of disasters is in establishment and progress has been made in it’s basic law, research methods and contents. the main areas of research are as follows: (1) disaster loss assessment methods (2) theory of value of statistical life (3) disaster economic impact model (4) disaster insurance and catastrophe risk securitization in addition, many researchers have studied the longterm effects of disasters on economic growth, but there is no consistent point of view. some scholars believe that disaster will lead to long-term negative economic effects, such as the slow economic growth, debt raising, area income gap increasing [26]. some scholars support that disasters will not have a long-term effect on the economy. recovery and reconstruction measures postdisaster can even increasing productivity, thus beneficial to the long-term development of the economy [27-30]. the current economic theory does not give a clear conclusion. references 1. mill j s. principles of political economy with some of their applications to social philosophy. london: j.w. parker, 1848 2. william b. the economics of natural and unnatural disasters. michigan: we upjohn institute, 2010. 3. hirshleifer j. economic behaviour in adversity. chicago: the university of chicago press, 1987. 4. kunreuther h, fiore e s. the alaskan earthquake: a case study in the economics of disaster[r]. 1966. 5. kunreuther h. the peculiar economics of disaster. public choice. 1967, 3 (1): 67-83. 6. dacy dc, kunreuther h. the economics of natural disasters: implications for federal policy. new york: the free press, 1969. 7. sorkin a l. economic aspects of natural hazards. lexington, ma: lexington books, 1982. 8. albala-bertrand j m. political economy of large natural disasters: with special reference to developing countries. oxford: oxford university press, 1993. 9. zheng gongchengthe economics of disasters. changsha: hunan people's publishing house, 1998. 10. he aiping. the economics of disasters. xi'an: northwest university press, 2001. 11. tang yandong. the economics of disasters. beijing: tsinghua university press, 2011. 12. wang yan-yan, liu shu-kun. disaster economic research.joural of catasrtophology, 2005, 20 (1): 104109. 13. xu juan. mitigation investment and the cost of disaster economics. disaster of 2006, 21 (2): 103-105. 14. he aiping developing countries, disasters, economic characteristics, causes and countermeasures. disaster science 2000, 15 (2): 91-96. 15. zhang xiandong, mei guangqing review on economics of disaster. joural of catasrtophology 1998, 13 (4): 81-87. 16. national research council (nrc). the impacts of natural disasters a framework for loss estimation. washington, dc: national academy of sciences press, 1999. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 35 17. united nations. economic commission for latin america and the caribbean, world bank. manual for estimating the socio-economic effects of natural disasters, 1991. 18. united nations. economic commission for latin america and the caribbean, world bank. handbook for estimating the socio-economic and environmental effects of disasters. new york: united nations publications, 2003. 19. viscusi wk, aldy j e. the value of a statistical life: a critical review of market estimates throughout the world. journal of risk and uncertainty. 2003, 27 (1): 5-76. 20. mei, lu yumei. the value of human life assessment methods.china safety science. 2007, 17 (3): 56-61. 21. tang yandong, liu chun-ping, assessment of the value of life and death compensation, china safety science. 2010, 20 (4): 14-21. 22. yu xi tang yandong, liu chun-ping, and other disasters of life valuation theoretical studies.china safety science. 2009, 19 (12): 17-22. 23. okuyama y, chang s. modeling spatial and economic impacts of disasters. berlin: springer verlag, 2004. 24. okuyama y, chang s. modeling spatial and economic impacts of disasters. berlin: springer verlag, 2004: 1. 25. rose a, liao s y. modeling regional economic resilience to disasters: a computable general equilibrium analysis of water service disruptions. journal of regional science. 2005, 45 (1): 75-112. 26. benson c, clay e. understanding the economic and financial lmpacts of natural disasters. washington dc: world bank publications, 2004. 27. albala-bertrand j m. political economy of large natural disasters: with special reference to developing countries. oxford: oxford university press, 1993. 28. albala-bertrand j m. the unlikeliness of an economic catastrophe. 2006. 29. toya h, skidmore m. economic development and the impacts of natural disasters. economics letters. 2007, 94(1): 20-25. 30. hallegatte s, dumas p. can natural disasters have positive consequences? investigating the role of embodied technical change. ecological economics. 2009, 68(3): 777-786. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 27–36 ___________________________________________________________________________________________________________ 36 1. introduction 2. the origin of economics of disasters 3. recent advances in economics of disasters 3.1. disaster loss assessment 3.1.1. direct and indirect loss 3.1.2. economic assessment indices 3.1.3. disaster economic impact model 3.1.4. nonmarket effects evaluation 3.2. value of statistical life 3.3. disaster and economic growth 3.4. disaster insurance and catastrophe risk securitization 4. conclusion references microsoft word editor's introduction editor's introduction the present issue of journal of risk analysis and crisis response (jracr), volume 2, issue 1 (2012) contains 8 papers. the former 5 contributions are written in english and others in chinese with english abstracts. the papers can be divided into five categories: natural disaster, environmental risk, crisis management, industry safety and investment risk. there are three papers in the first category, the paper “volcanic risk perception, locus of control, stress and coping responses of people living near the popocatépetl volcano in mexico” by esperanza lópez-váquez and maria luisa marván, presents the difference between coping strategies of people living near the popocatépetl volcano in mexico. the residents with a limited educational background were more likely than those with higher education to use passive coping strategies. the paper “a study on residents’ risk perception in abrupt geological hazard” by anping pan, carried out a questionnaire survey in qingyuan county of china to understand residents disaster perception and actual evacuation behaviors. the results show that most of the population considers that their possessions and services would be affected. 54.9% residents in geological hazard risk believe their lives will be threatened. the paper “research on risk assessment and regionalization of forest and grassland fires” by li, wu, lv and zhang, suggests a risk regionalization of grassland fire in inner mongolia. the authors consider rule of fires, distribution of high temperature points, characteristic of distribution of vegetation types, and the extent of road coverage. the second category, the “environmental pollution and human health risks near a hazardous waste landfill. temporal trends” by rovira, mari, schuhmacher, nadal and domingo, presents the human health risk assessment associated to the human exposure to polychlorinated dibenzo-p-dioxins, dibenzofurans and metals for the population living near a hazardous waste landfill in catalonia, spain. the current health risks in the area were within acceptable ranges, but it is recommended to continue the monitoring. the paper in the crisis management category is “how crises model the modern world” by lagadec and topper. the authors believe that “black swans” are increasingly becoming the norm and proposed a new theoretical vision of crisis and crisis management. their recommendations include leadership, rapid reflection force, facing the real world and the center for transatlantic relations advanced seminars. there are two papers in the industry safety category. the first paper “perception of risks in nanotechnology: determining key aspects in chile” by pozo, schmessane and et al., provides preliminary base-line data on public perceptions about the risks associated with nanotechnology and analyzes the current development of nanotechnology in chile and the risks associated with the incorporation of this new technology. in the country, the main risks associated with nanotechnology relate to the lack of legislation and/or journal of risk analysis and crisis response, vol. 2, no. 1 (march 2012), 1-2 published by atlantis press copyright: the authors 1 regulations in chile in the area of science and technology. the second paper “a method for assessing risk rating of natural gas pipeline based on accident statistics” by xu, liu, guo and ji, analyzed 199 natural gas pipeline leakage accidents occurred in china and proposed a method to assess the natural gas pipeline risk rating. the last category includes one paper “evaluating of loan guarantees between parent and subsidiary based on vulnerable option” by li, zhou and lei, gives the valuation model under two different security methods, theoretically analyzes the influence of such critical factors as companies asset value and equity ratio to loan guarantees’ value and provides theoretical basis for the decision-making of loan guarantees. in 2012, more people are wondering any threats to the earth. yes, the mayan calendar ends on december, 2012, but very few people believe that we are in the risk of the end of the world. meanwhile “black swans” are increasingly becoming the norm, human risk analysis technology and crisis response capacity are growing rapidly. i am grateful to the referees for their kind support and help. thanks also go to the authors for all their submissions. editor-in-chief chongfu huang published by atlantis press copyright: the authors 2 atlantis press journal style acknowledgements to referees and staffs for jracr in 2016 by chongfu huang the quality of journal of risk analysis and crisis response (jracr) depends on the qualified and regular collaboration of renowned scientists, who devoted their time to constructively review the submitted articles. i wish to thank the following individuals who acted as referees for the journal in 2016. it is a great pleasure to thank the staffs serving for jracr. i appreciate these volunteers who have donated their time without compensation to promote the journal. referees of jracr: christopher c. allan agnes allansdottir arjen boin an chen gianpaolo cimellaro lei deng min ding christoph doktor edna einsiedel alexander fekete zhaoyong guan zhenhua guo jinqiu hu carel tj hulshof chengjun ji tae-woong kim emérite andré laurent supeng leng husheng li jingyi li qiang li shuangshuang li xu li kai liu tiezhong liu xilin liu yun luo jianming mo bin pan lei pang a. pearson andy pearson salvador casadesús pursals huanguang qiu manuel d. rossetti yanrui shang geetam tiwari ruipeng tong donald a. wilhite chaowu xie xinliang xu xuanhua xu fei you zhuliang yu hongmei zhang junye zhao staffs of jracr: editorial board office mu zhang hongmei zhang wen han ziyan wang xiaobo lu editorial department junxiang zhang fei wen hongbing zhu jingjing cheng reviewer center xilei pang ye xue fuping yang chongfu huang, professor, ph.d. academy of disaster reduction and emergency management beijing normal university no.19 xinjiekouwai street beijing 100875, china journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 221 published by atlantis press copyright: the authors 221 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style major element geochemistry of longshan loess profile in the central shandong mountainous regions, northern china min ding 1,2*,shuzhen peng 1*, longjiang mao 3, wei zhang 1, qiuyue zhao 1 1 the key laboratory of tourism and environment in universities of shandong, taishan university, tai’an 271000, china 2 the college of tourism and environment, shaanxi normal university, xi’an 710062, china 3 college of marine science, nanjing university of information science and technology, nanjing 210044, china *corresponding author, email:dingmintsjy@163.com, shuzhenpeng@sohu.com abstract in this paper, the major elements of the longshan loess profile on the northern piedmont zones and intermountain valleys of mountainous regions in central shandong province in northern china, have been systematically tested and been compared with the yhc loess in the loess plateau to reveal the geochemical characteristics and material sources of ls loess. it is found that the average chemical composition of shandong ls profile is similar to that of typical loess at yhc profile. the cia, na2o/k2o and al2o3-cao+na2o-k2o triangles show that the ls loess is in moderate weathering stage and the chemical weathering degree is higher than that of the yhc profile in the loess plateau, but it is still in the na, ca removal of stage. element activity sequence at the ls profile is ca> na> mg> si> al> k> fe, and element activity capacity at ls profile is higher than that of at yhc profile. the ratio of tio2/ al2o3 and k2o /al2o3 of the material source index clearly distinguishes ls loess and yhc loess, and it is concluded that the ls loess has different source from the loess plateau loess. it is not the result of dust storm direct from the northwest of china; otherwise, the yellow river sediments and the north china plain material nearby may be its main sources when it was colder and drier during the glacial period. cia and k2o/na2o, and the migration rate of fe, al, k and na is restored to the depositional environment of research. the winter monsoon in the late glaciation is strong and the chemical weathering is weak. in the early holocene winter monsoon is weakened and the chemical weathering is enhanced. in the middle of holocene, the summer monsoon dominated and the chemical weathering is strongest; in late holocene, the climate deteriorated again, the chemical weathering is weakened. key words: loess in shandong; elemental geochemistry; weathering intensity; environmental significance; material source 1. introduction element geochemistry plays an important role in studying quaternary sediments (guo, et al. 2009; huang, et al. 2009; hao, et al. 2010; taylor. 1985). in the process of eolian transport, eolian accumulation and soil formation, it is accompanied by the migration, transformation and enrichment of elements (guo, et al. 2009; huang, et al. 2009). therefore, the composition and evolution of elements have unique advantages in regional environmental reconstruction and provenance studies (guo, et al. 2009; huang, et al. 2009), many studies on major elements have been carried out about loess (peng, et al. 2001; xiong, et al. 2008; chen, et al. 2001; ding, et al. 2011) in the loess plateau and its surroundings (huang, et al. 2009; hao, et al. 2010; li, journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 127 received 16 july 2017 accepted 13 august 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). （b） （a） et al. 2013; li, et al.2016; li, et al. 2007; zhang, et al. 2013; cao, et al. 1987). the loess of shandong is located in the link between northern and southern loess, and plays a key role in regional environmental evolution and global monsoon reconstruction. in the past, traditional stratigraphy, sedimentology and other means have been used to study the causes of loess (cao, et al. 1987; zhang.1995; zhao, et al. 1996), material sources (peng, et al. 2007; peng, et al. 2011; xu, et al. 2014) and climate change (li. 1987; ding, et al. 2011), element geochemical research has achieved certain results (diao. 1994; peng, et al. 2016; xu, et al. 2016; ni.2015), but the high-resolution research is still relatively rare. enriching and perfecting the data of loess elements in shandong will undoubtedly help to reveal the internal relationship between the loess plateau and the surrounding loess, and help to reveal the process of regional environmental change. this paper systematically analyzes the major geochemical characteristics of the longshan loess profile in the central shandong mountainous regions in northern china, and compares it with typical loess sections such as yhc profile in the loess plateau to reveal its chemical weathering characteristics, migration laws and material sources. material source and transport path and dust accumulation law will contribute to the prediction of modern dust storm risk. 2. materials and methods 2.1 study area and profile characteristics the study area is located at the central shandong mountains of north china plain (fig.1)，it is covered with the alluvial flood plain only along the yellow river, the rest mostly consists of the hills and low mountains, elevating between 200 m and 1545 m. the region has a warm temperate and semi-humid monsoon climate with distinct season, prevailing northwesterly wind in winter and southeasterly wind in summer, respectively. annual temperature and precipitation are about average 12.6-14.5 °c and 615.3-793.9 mm (peng et al., 2016). thick loess deposits in central shandong mountains mainly spread in the east-west direction along the northern piedmont regions of the central shandong mountains (fig.1a). (a) the locations for the research areas and the sampling points (b) the photos of the sections figure.1. research areas, ampling point location and the profile of field picture the section of longshan is located in shandong (hereinafter referred to as ls) (117°21′46.2″e, 36°43′59.4″n), the exposed depth is about 4m (bottom not seen), the elevation is 56m (fig. 1b). 120 samples are abtained continuously from upside to downside with an interval of 2.5cm. the depth of sampling is 3m. after the field profile morphological characteristics observation and 14c dating (data published elsewhere), it is determined that the section has formed since the last glacial, we can divide the ls profile into the following layers: l1 (190cm below) sediments formed in the last glacial period, it is yellow, clay silt. lt (190~160cm deep) is the transition layer, which formed in the early holocene, it is grey yellow; s0 (160~50cm deep) formed in mid-holocene, it is brown ancient soil, texture is hard, the structure is prismatic; l0 (50~0cm deep), is the modern dust accumulation layer which formed in the journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 128 late holocene, it is orange and sandy soil. the loess plateau is the result of wind transport and accumulation from the northwest desert and gobi. the yhc section of the loess plateau is chosen as the representative of the northwest desert and gobi sediments the profile of yaohe village (hereinafter referred to as yhc) in baishui, shanxi is located on the plateau (109°29′4.7″e , 35°15′57.3″n, the elevation is 960m), and is in the west of the ls section in shandong, the two section latitudes match each other. the layer for the yhc section is clear, it is divided from top to bottom as: the surface layer (ms), the modern loess layer (l0), the paleosol layer (s0), the malan loess layer (l1), baishui belongs to the warm temperature semi-arid climate, the annual average temperature and the annual precipitation in baishui are 11.4 ℃ and 577.8 mm respectively . 2.2 the experiment method the determination for the chemical element is applied with the pw2403 x-ray fluorescence spectrometer produced by the company of panalytical from the netherlands. we grind the natural dried samples until the diameter of the particle is smaller than 200 eyes, and then weigh for 4g of the samples with the use of the method for chrome acid to press the tablets on the pressure protot of yy-60; put the samples after the press of tablets into the sample cup by sequence, conduct the determination for the content of the chemical elements by the pw2403 x-ray fluorescence spectrometer. in order to control the stability of the measurement and the error, the national standard samples of gss-1 and gsd-12 are added in the process of measurement for the control, the error is controlled within 5%. 3. the result of the experiment as it is indicated in fig.2 and fig.3, the major elements in the ls section of shandong have the following characteristics: • the composition of major elements in ls section as follows, the content of sio2 is the highest which is 52.07%~65.53%, the content of al2o3 is 12.36%~15.85%, the content of fe2o3 is 3.90%~6.63%, the content of k2o is 1.96%~2.66%, the average content of the four main chemical components in the ls section which are sio2, al2o3, fe2o3, k2o, the average content of the four reaches 82.22% (69.78%~99.30%), the composition sequence for the average content of the elements is sio2 (59.30%) >>al2o3 (14.01%) > fe2o3 (5.14%) > cao (4.32%) > k2o (2.40%) > mgo (2.04%) > na2o(1.10%) > tio2 (0.68%). it has the characteristics of uniform composition of wind loess such as yhc section on chinese loess plateau. • the vertical variation trend of the major elements such as al2o3, fe2o3, k2o and tio2 in the ls profile is basically the same, and they are relatively enriched in the paleosol layer (s0), and relatively deficient in the loess layer (l0, l1, lt). the change trend of na2o is opposite to that of al2o3, fe2o3, k2o and tio2. the content of na2o in paleosol (s0) is lowest, and the content is higher in loess layer (l0, l1 and lt). • sio2 and cao appear with highly significant negative correlation (r=-0.901) which is the same as the relationship with the two disclosed in the yhc section in the central shaaxi plain (m. ding, et al.2011) and the zltc section in the southern shanxi (li, et al. 2013). cao is mainly present in the form of silicate and caco3, and the cao content is low and the sio2 content is high in s0 and l0, and they are opposite in lt and l1. • the distribution of the constant elements of the ls profile is related to the geochemical characteristics of the elements and the soil environment. in addition to cao (coefficient of variation of 0.72), the vertical coefficient of variation (cv) of other major elements is less than 0.25, and the coefficient of variation is ti (0.02) and al (0.05). the results show that the material composition of ls loess-paleosol sequence in shandong has the characteristics of uniform composition of wind loess. • it can be known from image 3, compared with ucc, the ls section of longshan has the characteristics of poor na and rich ti, the content of other elements is similar. compared with the different layers of ls section, the dropping na, k and the rich ti is almost at the same degree which means the characteristics of the source area; the difference between the layers of ca is obvious. l1, lt are in rich of ca in an apparent way, while s0, l0 drop ca in an apparent way, it is the reason that the element ca obviously migrated after the journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 129 accumulation of wind dust. compared with the yhc section, in the paleosol layer of s0 of the ls section, the elements of ca, mg and na are relatively lost, especially ca of which the loss is the most obvious one; fe, al, k and si are relatively concentrated, the concentration of fe and al is relatively obvious. in the loess layer of l1, the element of mg, na are relatively lost, ca, fe and al are relatively concentrated. fig. 2. change curves of major elements content and caco3 at ls loess-soil profile in shandong province fig. 3. curves of major elements standardized by ucc at ls profile in brief, the compositions of major elements in both ls section and yhc section are quite similar; compared with ucc the difference is mainly caused by weathering. it is indicated that the ls profile is characterized by eolian deposits and is preliminarily judged to be mainly a eolian. this is consistent with previous findings (peng, et al. 2016; xu, et al. 2016). 4. the discussion for the problems 4.1 the characteristics of weathering (cia, na/k and a-cn-k) the a-cn-k (al2o3-cao*+na2o-k2o) triangle reflects the trend of chemical weathering, the changes in principal composition and mineralogy during chemical weathering, and the palaeo climate environment during sediment deposition (gu. 1999; wang, et al. 1995; gallet, et al. 1982; nesbitt, et al. 1980 and 1982). in figure 4a, the distribution of data points of ls profile are in the above the plagioclase-feldspar baseline, its weathering trend line substantially parallel to the a-cn line, just lying on the line of continental crust ucc and terrigenous shale paas connection, close to the plagioclase and not reach the side of a-k line, which implies that the loess feldspar mineral chemical weathered weakly, and experienced dropping na and ca , but potassium feldspar almost no changes. so, it may still be in the early stages of chemical weathering (j.chen, et al. 2001), weathering products are mainly kaolinite, illite and montmorillonite. the loss rate of na and ca is the lowest in l1 and the highest in s0. in addition, we select the s0 with the strongest soil formation and l1 with the weakest soil formation to project point on the triangle of a-cn-k digram, the data points of ls in shandong province is closer to point a, according to this order (yhc-l1 →yhc-s0 →ls-l1 → ls-s0), the degree of weathering is enhanced, generally the strength of dropping ca and na of loess in ls section is higher than that of the yhc section in loess plateau (fig. 4b). the data points of both ls section and yhc section are concentrated and distributed at ucc→paas chemical weathering trend line (parallel with the connection line of a-cn), this characteristics again explains that ls section is to some extent the same with the yhc section of the loess plateau , which both are originated from the wide upper continental crust with a certain mixing before the sedimentation. l0 s0 lt l1 60 64 68 72 sio2（%） 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0 深 度 （ c m ） 14 16 18 al2o3（%） 4 6 8 fe2o3（%） 0 6 12 18 cao（%） 1.8 2.4 3 k2o（%） 0.6 1.2 1.8 na2o（%） 1.8 2.4 3 mgo（%） 0.72 0.78 0.84 tio2（%） 0 4 8 12 caco3(%) journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 130 a=al2o3,cn=cao+na2o,k=k2o, sm = montmorillonite, il = illite, pl = plagioclase, ks=potash feldspar fig. 4. ls(a) and other dust deposits(b) plotted in a cn – k ternary diagram the molecular ratio of the elements reduces the impact from the elements of disturbance to some extent which can effectively disclose the weathering degree for the sediments, cia and na2o/k2o is the commonly used alteration index for the chemical weathering. the chemical index of alteration (cia) may effectively indicate the degree of weathering from the feldspars into the clay mineral, the larger the value is, and the stronger the degree of the weathering reflected will be. the calculation formula of cia is: cia=[al2o3/(al2o3+cao*+k2o+na2o)]×100, in the formula: all are the mole numbers for the oxide molecules, of which cao* is the mole number in the mineral of silicate. all the calculation methods for the value of mcao in this article are as follows: when the mole number of cao is larger than na2o, mcao﹡=mna2o, when it is smaller than na2o, then mcao ﹡ =mcao (s.m.mclennan.1993). research finds out that the value of cia is between 50%~65% which reflects the degree of chemical weathering of lower level under the condition of cold climate; when the value of cia is between 65%~85%, it reflects the strength of the chemical weathering of medium level under the warm and humid climate; when the value of cia is between 85%~100%, it reflects the strength of the chemical weathering is very strong under the condition of subtropical climate. na2o/k2o is also the effective index for the degree of weathering of feldspar which is negatively related to the soil formation of weathering. the cia of ls section varies between 66.3~74.5, the average value is 69.1. the weathering sequence for the different layers of the section is s0 (71.5) > l0 (68.2) > l1 (67.7) ≈ lt (67.2) >> ucc, it means that the whole section of ls in shandong province is at the medium weathering stage. while the yhc section of the loess plateau which is at the transition stage of lower weathering and medium weathering, the average value of cia is 65.6 (63.7~67.7), the strongest weathering layer is s0 which is close to the medium weathering degree. compared with those on the same layer, cials>ciayhc. it explains that the weathering degree of ls section is obviously higher than the yhc section in the loess plateau., and the lt with the weakest weathering in the ls section (cia is 67.2) is roughly the same as the s0 with the strongest weathering in the yhc section (cia is 66.5) in the weathering degree. na2o/k2o ratio in both the ls and yhc section shows obvious the systematic changes in cycle which is lower in the paleosol layer and higher in loess layer, and has a good negative correlation with the numerical value of cia. the weathering sequence it expressed is similar with that of cia. the average value of na2o /k2o for the ls section is 0.71（0.46~0.94）, the lowest value appears in the paleosol and the highest value in l1, the sequence for the claying degree of ls is s0 (0.58)>l0 (0.73)> lt (0.80) ≈l1 (0.81), the one layer with the strongest claying of yhc is also s0 (0.81), the second is l0 (0.86), the claying degree of l1 is the lowest, na2o/k2o is 0.94. compared with the two sections, the na2o/k2o value of the ls section is obviously smaller than that of the corresponding layer of yhc section in the loess plateau. it is consistent with the extent of weathering revealed by cia. in conclusion, the ls section in shandong has experienced stronger feldspar journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 131 weathering. it is completely consistent with the weathering degree disclosed by cia. in short, the ls section in shandong has experienced a stronger feldspar weathering than yhc section on the loess plateau. the cia of the ls section (69.1) is slightly lower than that of the xiashu loess in the southern china whose cia is 70.45 (the annual average temperature is 15.4℃, the precipitation is 450mm) (li, et al. 2007), higher than that of shanxi luochuan section whose cia is 62.5 (the annual average temperature is 9.2℃, the precipitation is 620mm) (huang, et al. 2009), and shanxi xiangfen section whose cia is 63.6 (the annual average temperature is 11.5℃, the precipitation is 550mm) (li, et al. 2013), and shanxi yhc section whose cia is 65.5 (the annual average temperature is 11.4℃, the precipitation is 577.8mm) (ding, et al. 2011), which are slightly higher than the loess at zhoujiagou, liaonan whose cia is 66.1 (the annual average temperature is 9.6℃, the precipitation is 708mm) (zhang, et al. 2013). the weathering degree of ls section in shandong is slightly lower than that of southern loess, higher than that of western loess on loess plateau and northeastern liaoning loess, which is not so hot as that in southern china, but wetter and warmer than that of the west and warmer than liaonan in the north east. and which is corresponding with large bio-climatic conditions. it is shown that the dust deposits are mixed and the parent material impact is small, and the cia is mainly controlled by the change of climate change, especially the precipitation (j.x.cao, et al. 1987; z.l.zhang. 1995; s.l.zhao, et al. 1996; s.z.peng, et al. 2007). however, the cia reveals the comprehensive weathering information of dust accumulation, including the source, the process of sorting and environmental changes in the sedimentary area, such as the degree of weathering of the ls section (cia is 69.1) is significantly higher than that of shandong pingyin loess (cia 42.9 ~ 63.3) (xu, et al. 2016) and the miaodao islands (59.7, averages of 4 profiles) (ni. 2015), indicating that the apparent differences in cia under similar climatic conditions are related to material source and particle size sorting, significant differences in cia in shandong may be caused by different material sources. 4.2 the activity of the elements in conclusion, different major elements have different geochemical behaviors in the process of loess formation, and are closely related to the environment of the soil formation t. most elements have particle size sorting effect. in addition, they are impacted by the leaching deposition of element ca. absolute content of major elements does not reflect its true geochemical properties (li, et al. 2007; zhang, et al. 2013). elements in ls and yhc profile were not subjected to acid-soluble treatment before the test. in order to eliminate the influence of carbonate leaching, the change rate of other elements in the sample was calculated by using the stability element ti as reference to obtain the transportation and concentration degree of the elements. using the method proposed by gallet et al. (s.gallet, et al. 1982). the calculation formula is: δ(%) ＝[(x(s)/i(s))/(x(l)/i(l))-1]×100; in the section of ls and yhc, we select the ti with the smallest coefficient of variation as the reference standard for the calculation of the migration rate of the elements in the other soil layer of the paleosol relative to l1, of which x(s), i(s) represents the content of element x and the reference element i respectively, x(l), i(l) represents the content of the above stated elements in the malan loess (l1) respectively (because the parent material of the two sections cannot be known before the soil formation, this article will consider l1 with the weak weathering degree as the parent material). when δ<0, it explains that element x is migrated out in this layer compared with the reference element; when δ>0, it represents its relative concentration. for the observation in the wild field, the soil formation of s0 is the strongest; its difference with l1 is the largest for which it can best show the difference of activity for the elements in the two sections. compared with the l1, element migration rate in s0 in the section ls and yhc has the following characteristics (fig. 5): element ca, na and mg δ<0, all of them are leaching relative to l1, while element k, fe, al δ>0, all are relatively concentrated. the activity order of the elements in the two sections is basically the same as that of ca > na > mg > si > al > k > fe. but the activity capability of the elements in the two sections (indicated with the migration rate) is not the same apparently, the activity of the elements in ls section is higher than that in the section of yhc, particularly, in the ls section, element migration rate ca>> na> 20%, which it reaches journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 132 the strong activity stage, element migration rate only ca>20% at the yhc section, all the other elements reach no more than 10%. ca na mg si al k fe -100 -80 -60 -40 -20 0 20 40 ls yhc fig. 5. migration ratios of major elements of the ls and yhc profile relative to the stable element ti nesbitt etc. divided the process of chemical weathering into the early stage of dropping na, ca, the middle stage of dropping k and the late stage of dropping si in accordance with the activity sequence of the elements. according to the geochemical characteristics of element migration for ls and yhc, both the section of ls in shandong and the yhc in the loess plateau have not entered the stage of dropping k and si, they are still at the stage of dropping na and ca. the loss rate of na and ca in the ls section is higher than that of yhc section on the loess plateau, it explains that the ls section has experienced the stronger weathering process than that of yhc section on loess plateau which is the same as the result cia and a cn –k ternary diagram have disclosed. 4.3 material source the ratio of tio2/al2o3 can be used to trace the material source (hao, et al. 2010; peng, et al. 2016). al, ti are relatively stable in the loess which are often used as the standard elements, al often is stable existing in the form of oxide or aluminosilicate in nature. ti is contained in extremely stable rutile and ilmenite, and it is difficult to migrate in the epigenetic environment (hao, et al. 2010; wang, et al. 1995). k elements are mainly present in potassium feldspar and mica, although it is easy to be weathered compared with al and ti, but it is very easy to be absorbed by the clay due to the large radius of k ion, obvious migration can only occur in the soil with strong weathering. the research in the article afore discovered that the weathering degree of both ls section in shandong and the yhc section on the loess plateau have not reached the senior weathering strength. k2o/al2o3 is suitable for the two sections to trace the source. in this paper, tio2/al2o3 and k2o/al2o3 were used to identify the source of ls in shandong and select the l1 layer with the weakest weathering to conduct. the trace index plot of tio2/al2o3 and k2o/al2o3 in the section of ls and yhc are distributed as it is shown in fig. 6, they are distributed in two different areas apparently. the value of tio2/al2o3 for the ls section in shandong province is smaller than that of the yhc section on the loess plateau, most of the value of tio2/al2o3 in ls section is between 0.060-0.065, the data for yhc section is between 0.066~0.069; the distribution of k2o/al2o3 on the ls section in shandong is very scattered (0.172~0.190), while the distribution of k2o/al2o3 for yhc on the loess plateau is comparatively concentrated (0.188~0.193), and the value of k2o/al2o3 on the ls section is lower than that of the yhc section. it explains that the material source of the ls loess in shandong is not consistent with the yhc section material source on the loess plateau, it is not directly blowming from the northwest area in the pattern of dust storm. as the study before, the k2o/al2o3 ratio of yellow river sediments is lower than the ratio of loess from chinese loess plateau is due to the loss of illite, smectilte, mica and other clay minerals comminuting during fluvial transport (peng.2016). k often is hosted in illite, smectilte, mica and other clay minerals, the loss of k leads to decrease of the ratio k2o/al2o3. the value of k2o/al2o3 on the ls section is lower as that of yellow river sediments. it is speculated that ls loess is coming from the sediment of the yellow river and north plain transported by wind transport. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 133 0.05 0.055 0.06 0.065 0.07 0.075 0.08 tio2/al2o3(molar ratio) 0.17 0.175 0.18 0.185 0.19 0.195 k 2o /a l 2o 3( m ol ar ra tio ) fig. 6. the trace index plot of tio2/al2o3 vs. k2o/al2o3 4.4 environment reconstruction the migration rate of elements is to eliminate the impact of the parent material, it is the reliable index that directly discloses the difference of environment for sedimentation, cia and k2o/na2o disclose the weathering strength, it is often used as the substitute index for the recovery of the environment. so, cia and k2o/na2o, and the migration rate of fe, al, k and na are effectively used to restore the depositional environment of ls section region (fig.7). l1 (190cm~300cm deep, bottom not seen,) at ls section , it was formed in the last glacial period, both the cia value and k2o/na2o value are the smallest, the degree of concentration for fe, al and k is low, the leaching degree of na is the lowest. when the climate was dry and cold. it is presumed that the water level of the yellow river in the north of the section is lowered, the flood land of the yellow river is accumulated as the thick beach. in the meantime the plain in the north china are exposed, the coverage of vegetation is low. so, the northwest monsoon dominated the winter monsoon, in the direction of the wind blowing, a lot of sand blowing from the yellow river beach and the plain in the north china to jinan, zhangqiu, zibo and other places, due to the southern mountains block in central shandong, it settled in the northern foot of the mountain, which become the main source of ls section. lt (190~160 cm deep), the transition layer, formed in the early holocene, the cia value and the k2o/na2o value starts to increase, the degree of weathering starts to be strengthened, fe, al and k starts to concentrate, na also has a certain leaching. the weathering is strengthened, the concentration degree of fe, al and k is still low, the monsoon is transformed. in the layer of s0 (160~50cm), both the cia value and the k2o/na2o reach the maximum value of the whole section, the leaching of na is also stronger, the concentration degree of fe, al and k is also the maximum, it explains that the weathering degree reaches the maximum value of the whole section, when we enter the middle holocene, the monsoon of winter is gradually weakened, the soil is the strongest .at this time the winter monsoon is weakest, the summer monsoon takes the advantage, and therefore, the effect of dust storm is reduced in an apparent way. while in the last interglacial period and the holocene epoch, the pollen records show that the west bank of bohai and the laizhou bay are the landscapes of broad leaved forest meadow or the swamp meadow, it does not have the condition for the supply dust (cao, et al. 1987). therefore, it can be initially presumed that the holocene epoch, especially the formation of s0 when the monsoon of winter is the weakest, basically all the materials on the ls section are from the flood land of the yellow river in the north of the research area and the sedimentation of the plain in the north china. l0 (50~0cm), is the accumulation layer of the modern wind dust, it formed in the late holocene, the cia value and the k2o/na2o value are reduced all in a sudden, the weathering is weakened, the concentration degree of fe, al and k is still low, the climate is deteriorating. 66 72 78 cia 300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0 d e p t h （ c m ） 0.8 1.2 1.6 2 na2o/k2o -20 0 20 40 fe migration rate(%) -8 -4 0 4 8 al migration rate( %) -8 -4 0 4 8 k migration rate (%) -40-30-20-10 0 10 na migration rate(%) l0 s0 lt l1 fig. 7. major element ratios and migration ratios of major elements at ls profile in shandong journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 134 5. conclusion this article systematically analyse the composition of the major elements in the ls loess section in the central shandong mountainous regions and compares it with the yhc section of the loess plateau to study the geochemical characteristics and the material source of ls loess. it is discovered as that: （1）the average chemical composition in the ls section in shandong province is very similar to the composition of the yhc section which is the typical aeolian, sio2, al2o3, cao and fe2o3 in the composition takes the absolute advantage, the average content of the four reaches 82.77%. （2）the cia, na2o/k2o and al2o3-cao+na2ok2o triangles consistently discloses that the ls section belongs to the medium weathering, the weathering degree is higher than that of the yhc section on the loess plateau but it is still at the stage of dropping na, ca, it has not entered the stage of dropping k, si. the activity sequence for the elements is ca > na > mg > si > al > k > fe, the activity capability is ls > yhc. （3）the ratio of tio2/al2o3 and k2o/al2o3 for the index of material source is clearly distinguished the ls loess and the yhc loess, it is presumed that the loess in the ls section and the loess plateau have different wind force transportation system, it is not the result of dust storm direct from the northwest of china, otherwise, the yellow river sedments and the north china plain material nearby may be its main sources when it was colder and drier during the glacial period. （4）cia and k2o/na2o, and the migration rate of fe, al, k and na is restored to the depositional environment of research. the winter monsoon in the late glaciation is strong and the chemical weathering is weak. in the early holocene winter monsoon is weakened and the chemical weathering is enhanced. in the middle of holocene, the summer monsoon dominated and the chemical weathering is strongest; in late holocene, the climate deteriorated again, the chemical weathering is weakened. acknowledgement: this study was supported by national natural science foundation of china (41402319,41472313,41771218,41602353), national key laboratory of loess and quaternary geology open fund (skllqg1002), taishan university talent research fund (y-01-2016001). min ding and shuzhen peng both are corresponding authors. references z.t.guo, s.z.peng, q.z.hao, et al. late miocene-pliocene development of asian acidification as recorded in an eolian sequence in northern china, global and planetary change, 41(2004)135-145. c.c.huang, j.l.pang, h.x.su, et al. the ustic isohumisol (chernozem) distributed over the chinese loess plateau: modern soil or palaeosol, geoderma, 150(2009)344-358. q.z.hao, z.t.guo, y.s.qiao, et al. geochemical evidence for the provenance of middle pleistocene loess deposits in southern china, quaternary science reviews, 29(23)(2010)3317-3326. s.r.taylor, s.m.mclennan. the continental crust: its composition and evolution, london: blackwell, (1985)277. s.z.peng, z.t.guo. geochemical indicator of original eolian grain size and implications on winter monsoon evolution, science in china, series d 44(suppl.), (2001)261–266. s.f.xiong, y.j.zhu, r.zhou, et al. chemical weathering intensity and its grain-size dependence for the loess-red clay deposit of the baishui section, chinese loess plateau, the research on the quaternary, 28(5)(2008) 812-821. j.chen, z.s.an, l.w.liu, et al. the change for the chemical composition of the wind dust in the loess plateau and the chemical weathering for the inland of asia since 2.5ma, china science, earth science, 31(2)(2001) 136-145. m. ding, j.l.pang, c.c.huang, et al. geochemical characteristics of major elements in holocene loesspaleosol sequence in the eastern guanzhong basin, shaanxi, china, china desert, 31(4)(2011)862-867 t.y.li, d.w.mo, g.r.zhu, et al. the geochemical characteristics and its paleo-environment significance of the invariable elements in the leoss section of holocene in jinnan, geographical research, 32(8)(2013)1411-1420. n.li, q.z.hao, x.j.zhang, et al. geochemical evidence for the provenance of loess deposits in the eastern qinling mountains, central china , the research on the quaternary, 36(2)(2016)332-346. x.s.li, z.y.han, s.y.yang, et al. chemical weathering intensity and element migration features of the xiashu loess profile in zhenjiang, the journal of geographical sciences, 62(11)(2007)1174-1184. w.zhang, y.w.dong, y.yu, et al. the weathering characteristics and its environmental significance for the loess in liaonan, the marine geology and the quaternary geology, 33(5)(2013)163-171. j.x.cao, p.y.li, n.shi. the loess at the group islands of miao dao in shandong , china science b edition , (10)(1987)123-127. z.l.zhang. research on the loess at the plain near the sea in the south of bohai, science bulletin, 40(1)(1995)57-63. s.l.zhao, h.j.yu. shelf desertization environment in the bohai and yellow seas during the last glaclal maximum, the research on the quaternary, 2(1)(1996)42-44. s.z.peng, z.d.gao, x.p.wu, l.b.zhang, m.y.liang, y.s.qiao. grain—size distribution and genesis of loess in journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 135 the qingzhou area, shandong , journal of geomechanics ,13(4)(2007)315-321. s.z.peng, l.j.zhu, g.q.xiao, et al. magnetostratigraphy and provenance of the qingzhou loess in shandong province, journal of arid land, 3(3)(2011)184–190. s.j.xu, x.c.ding, z.c.ni. the sedimentary characteristics of buxi loess profile in shandong province and their paleoclimatic and palaeoenvironment significance, the journal of geographical sciences, 11(2014)1707-1717. p.y.li. late cenozoic erathem and environment vicissitudes of miaodao archipelago, shandong province, marine geology and the quaternary geology ,7(4)(1987)111-122. m.ding, s.z.peng, j.l.pang, w.zhang, d.d.chen, h.j.yang. holocene environment change and its impacts on cultural development in the central region of shandong province, the notice of soil, 42(06)(2011)1281-1287. g.y.diao. caco3 and ree in the loess of shandong, geology, geochemistry, (02)(1994)98-102. s.z.peng, q.z.hao, l.wang, et al. geochemical and grainsize evidence for the provenance of loess deposits in the central shandong mountains region, northern china,quaternaryresearch(2015),http://dx.doi.org/10.101 6/j.yqres,(2016).01.005 j.s.xu, z.c.ni, x.c.ding. geochemical characteristics of major elements of the pingyin loess in shandong province , bulletin of mineralogy;petrology and geochemistry ,35(2)(2016)353-359. z.c.ni. the osl age for the loess on the group islands of miao dao, shandong province and the discussion for the material source, jinan ： shandong normal university, (2015 ) z.y.gu. weathering histories of chinese dust deposits base on uranium and thoriun series nuclides, 10be and major elements, beijing: institute of geology and geophysics, chinese academy of sciences,(1999) y.wang, f.s.wei. the element chemistry of soil environment, beijing: china environmental science press,(1995)362379, 75-86, 180-192. s.gallet, b.jahn, m.torii. early proterozoic climates and plate motions inferred from major element chemistry of lutites , nature, 229(1982)715-717. h.w.nesbitt, g.m.young. early proterozoic climates and plate motions inferred from major element chemistry of lutites, nature, 299(1982)715-717. h.w.nesbitt, g.markovics, r.c.price. chemical processes affecting alkalis and alkaline earths during continental weathering,geochem cosmochim acta, 44(11)(1980)1659-1666. s.m.mclennan. weathering and global denudation, journal of geology, 101(1993) 295-303. h.j.yang. research on the geochemical characteristics of the soil element and the evolution of the environment for the soil formation for the loess in holocene at the great wall in ningxia , xi’an,：shanxi normal university, (2010) x.y.yang, c.c.huang, j.l.pang, et al. dust source of the holocene loess-soil and pedogenic environmental changes in the upper huaihe river,journal of geographical sciences, 19(1)(2009)107-117. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 127–136 ___________________________________________________________________________________________________________ 136 1. introduction 2. materials and methods 2.1 study area and profile characteristics (b) the photos of the sections figure.1. research areas, ampling point location and the profile of field picture the section of longshan is located in shandong (hereinafter referred to as ls) (117 21′46.2″e, 36 43′59.4″n), the exposed depth is about 4m (bottom not seen), the elevation is 56m (fig. 1b). 120 samples are abtained continuously from upside to downsid... 2.2 the experiment method 3. the result of the experiment 4. the discussion for the problems 4.1 the characteristics of weathering (cia, na/k and a-cn-k) 4.2 the activity of the elements 4.3 material source 4.4 environment reconstruction 5. conclusion references << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <feff005500740069006c0069006300650020006500730074006100200063006f006e0066006900670075007200610063006900f3006e0020007000610072006100200063007200650061007200200064006f00630075006d0065006e0074006f00730020005000440046002000640065002000410064006f0062006500200061006400650063007500610064006f00730020007000610072006100200069006d0070007200650073006900f3006e0020007000720065002d0065006400690074006f007200690061006c00200064006500200061006c00740061002000630061006c0069006400610064002e002000530065002000700075006500640065006e00200061006200720069007200200064006f00630075006d0065006e0074006f00730020005000440046002000630072006500610064006f007300200063006f006e0020004100630072006f006200610074002c002000410064006f00620065002000520065006100640065007200200035002e003000200079002000760065007200730069006f006e0065007300200070006f00730074006500720069006f007200650073002e> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <feff004200720075006b00200064006900730073006500200069006e006e007300740069006c006c0069006e00670065006e0065002000740069006c002000e50020006f0070007000720065007400740065002000410064006f006200650020005000440046002d0064006f006b0075006d0065006e00740065007200200073006f006d00200065007200200062006500730074002000650067006e0065007400200066006f00720020006600f80072007400720079006b006b0073007500740073006b00720069006600740020006100760020006800f800790020006b00760061006c0069007400650074002e0020005000440046002d0064006f006b0075006d0065006e00740065006e00650020006b0061006e002000e50070006e00650073002000690020004100630072006f00620061007400200065006c006c00650072002000410064006f00620065002000520065006100640065007200200035002e003000200065006c006c00650072002000730065006e006500720065002e> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <feff04120438043a043e0440043804410442043e043204430439044204350020044604560020043f043004400430043c043504420440043800200434043b044f0020044104420432043e04400435043d043d044f00200434043e043a0443043c0435043d044204560432002000410064006f006200650020005000440046002c0020044f043a04560020043d04300439043a04400430044904350020043f045604340445043e0434044f0442044c00200434043b044f0020043204380441043e043a043e044f043a04560441043d043e0433043e0020043f0435044004350434043404400443043a043e0432043e0433043e0020043404400443043a0443002e00200020042104420432043e04400435043d045600200434043e043a0443043c0435043d0442043800200050004400460020043c043e0436043d04300020043204560434043a0440043804420438002004430020004100630072006f006200610074002004420430002000410064006f00620065002000520065006100640065007200200035002e0030002004300431043e0020043f04560437043d04560448043e04570020043204350440044104560457002e> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice atlantis press journal style editor’s introduction for volume 6, issue 4 risk, similar as a ghost, is a scene in the future associated with some adverse incident. if we can accurately predict the incident, it is called a pseudo risk. in this case, it is not a wandering ghost, but a familiar thing. for example, if a person falls to the ground from a plane in 500 meters high without parachute, he will die. there is not any suspense, it is a pseudo risk. therefore, the most of risk analysts consider risk as some uncertainty and used to measure risk with probability. however, “probability” does not appear in the papers of this issue. it implies that the denotation of risk is larger than the denotation of probability. this issue contains 6 papers. there are 3 contributions written in english and 3 contributions in chinese with english abstracts. the papers can be divided into five topics: emerging risks related to new technologies and environment, risk issues in industrial systems, emergency management of natural disasters, emerging risks related to climate change and risk identification ＆ management of natural disasters. there are two papers in emerging risks related to new technologies and environment. the first paper “risk considerations in the use of unmanned aerial vehicles in the construction industry” by george wang, donna hollar, suaan sayger, et al is to provide construction professionals with timely and pertinent information on unmanned aerial vehicles use with a focus on risk management based on current industry practice, experience and literature review. the second paper “monster cows and the doing of modern biotechnology in sweden: an intersectional risk analysis” by anna olofsson ＆ susanna öhman is to analyze how the insemination of images of a cow is entangled with perceptions of gene technology through the ‘doing’ of risk, nature, periphery and gender in sweden. two types of data have been used; the 1996 eurobarometer survey on public opinion and a corpus of articles from the swedish daily press. the conclusion was that terms like gene technology are so abstract that people have problems in understanding the meaning of the concepts and then turn to easily accessible images for instance in the media. there is one paper in risk issues in industrial systems. the paper “cfd analysis of oxygen and carbon dioxide recovery during ventilation in municipal confined space” by yan liu, cong tan＆ dongliang liu, simulates ventilation process in municipal heating confine space by cfd (computational fluid dynamics) numerical simulation method, and studies the distribution of temperature, oxygen and carbon dioxide fraction in municipal heating confined space during ventilation process. the simulation results show that the airflow and temperature are obviously stratified in vertical direction during the ventilation. the temperature is low and the airflow velocity is high at the bottom part of the tunnel. while the temperature is high and the airflow velocity is low at the top of the tunnel. by contrast, recovery of the oxygen and carbon dioxide content is relatively faster at the bottom than it at the top. there is one paper in emergency management of natural disasters. the paper “path optimization in dynamic adverse weathers”, by mingkong zhang, xiaobing hu ＆ jianqin liao, concerned with how to achieve optimal actual travelling trajectory by just a single offline optimization, given the dynamics of weather conditions is pre-known. to this end, the concept of co-evolutionary path optimization (cepo) is introduced, where the weather condition in a single run of offline optimization is not static, but keeps changing during the single run of offline optimization. existing dpo methods can hardly address cepo, because they do not allow the weather condition to change in a single run of online optimization. to address the cepo in dynamical adverse weathers, this paper proposes a ripple-spreading algorithm (rsa), which can achieve optimal actual travelling trajectory by a single offline calculation. the reported cepo and rsa are then tested on a typhoon scenario in hainan province of china, and the advantages against traditional dpo methods are clearly demonstrated. there is one paper in emerging risks related to climate change. the paper “the response characteristics of xilingol grassland to uneven distribution of precipitation at temporal and spatial scale” by qiaofeng zhang, hongbo yu, guixiang liu, and et al, uses pearson correlation analysis to analysis the response characteristics of ndvi to precipitation at spatial and temporal scale based on the precipitation of 15 weather stations and modis mod13a3 ndvi data. results showed that from one month view, vegetation growth was the highest correlation with the precipitation of this month from may to july, but august and september were the highest correlation with the precipitation in july. accumulated month view, the accumulated precipitation in the last three months was very important to the vegetation growth, and snowfall in the winter of previous year played an important role in may of next year. spatially, from one month view, erenhot and sonid youqi（desert steppe region）were the highest correlation with the precipitation in two months before, the others were the highest correlation with the precipitation in one month journal of risk analysis and crisis response, vol. 6, no. 4 (december 2016), 163-164 published by atlantis press copyright: the authors 163 before. accumulated month view, erenhot and sonid youqi were the highest correlation with the accumulated precipitation in the last four months, the others were the highest correlation with the accumulated precipitation in the last three months. there is one paper in risk identification ＆management of natural disasters. the paper “public risk perception and risk communication of typhoon disaste” by zhihai shang ＆ liping li, constructs the theory model of 12 factors affecting risk perception from three dimensions by public individual characteristics, risk situational characteristics, risk communication efficiency, which influence the input, output and transmission of risk information. then the theory was applied in typhoon disaster in zhanjiang city, and it comes out that the theory is reliable. finally, the contents of typhoon disaster risk perception capacity construction were given in this paper as risk communication capacity building, special platform construction and cultural environment of risk communication, which is the only way for effective risk communication. we sincerely thank the referees who provided such insightful comments and helpful. and also thank the authors very much for all their outstanding contributions. editor-in-chief: prof. chongfu huang email: hchongfu@126.com publication chair of sra-china: prof. mu zhang email: rim_007@163.com director of editorial department: prof. junxiang zhang email: jracr_srachina@126.com published by atlantis press copyright: the authors 164 航空应急救援地面保障能力评价 capability evaluation of ground support for emergency rescue helicopters xiuyan zhang1,2 ran liu1 aiwen yu1 1 civil aviation university of china, economics and management college, tianjin 300300, china 2 university of chinese academy of sciences, institutes of science and development, beijing 100190, china received april 16, 2018 accepted april 28, 2018 abstract under the background that the demand for air emergency rescue is increasingly urgent, the ground support for helicopters’ taking off and landing safely in disaster area is studied. by defining the connotation and identifying the influencing factors, the evaluation index system is established, including three level indicators. the weights of the indexes are determined by means of questionnaires. the evaluation model is established by using fuzzy comprehensive evaluation method. finally, the feasibility of the evaluation model is verified taking a helicopter airport in china as an example. the verification results show that the model is beneficial for the government to fully understand the present situation of the ground support for emergency rescue helicopters, and also to provide direction for the related ground support organizations to improve their ground support capability. keywords: emergency rescue; helicopter; ground support; capability evaluation 应急救援直升机地面保障能力评价 张秀艳 1,2 刘然 1 余爱文 1 1.中国民航大学, 经济与管理学院, 天津 300300, 中国 2.中国科学院大学, 科技战略咨询研究院, 北京 100190，中国 摘要：在航空应急救援需求愈发迫切的时代背景下，针对直升机在受灾地区安全起降所需的地面 保障工作进行研究。通过内涵界定与影响因素识别，构建了应急救援直升机地面保障能力评价三 级指标体系，根据问卷调查结果确定了指标权重，运用模糊综合评价法建立了应急救援直升机地 面保障能力评价模型。 后，以国内某直升机机场为例验证了评价模型的可行性，验证结果表明 该模型既有利于政府部门充分了解我国应急救援直升机地面保障能力的现状，也可为相关地面保 障单位提高自身地面保障能力提供方向。 关键词：应急救援；直升机；地面保障；能力评价 1 作者简介：张秀艳，1983.8，讲师（博士研究生），安全管理方向，天津市东丽区津北公路 2898 号中国民 航大学，300300，13802104873，13802104873@163.com 基金项目号：国家重点研发计划 2016yfc0802601 航空应急救援装备体系和标准体系研究 国家级大学生创新创业训练计划项目 201710059014 航空应急救援单位运行模式构建研究 93 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 1 引言 近年来，我国公共安全呈现出大灾多发、 多灾并发的态势，突发事件发生总量大、损失 大、对社会造成的影响越来越引起人们的关 注。航空应急救援作为应急救援中 快速、 灵活的救援方式，在各种救援任务中可以发挥 不可替代的作用。而直升机作为航空应急救援 中使用频率 高、参与救援任务种类 多的救 援装备，却由于起降、天气或油料等保障不力 而无法随叫随到、随救随走。在 2013 年 4 月 发生的雅安地震中，受灾重灾区适合直升机使 用的起降点较少，且起降条件较差，缺少相应 的灯光等助航设施设备，而直升机很难在能见 度不够的情况下进入灾区展开救援工作，直升 机顺利运行所需要的油料也得不到保障 [1] 。种 种地面保障条件不到位，使得直升机不能在第 一时间赶赴灾区，这极大地浪费了我国的应急 救援资源，也导致灾区人民错过了黄金救援时 间。 直升机地面保障以通航固定起降点地面 保障为主，以临时铺设起降点为辅。在通航发 展较为成熟的国家中，通航保障体系中的通航 机场、固定运营基地（fixed base operator, fbo）、飞行服务站（flight service station, fss） 以及维修站（maintenance repair& operations, mro）的数量都较多，且配备了相应的通信、 导航、气象等设施设备，并安排了专业人士来 进行操作 [2,3] 。我国目前仅有的几家fbo均属于 公务型fbo，在航空应急救援中能发挥的作用 有限，而专门为直升机提供维修服务的mro 还尚未设立 [4] 。直升机地面保障严重不足所带 来的后果就是直升机在许多地方无处可落，大 大影响应急救援效率。在这些已有的通航机场 中，还有许多缺少或者没有及时更新相应的配 备，如：导航设施、场道灯光设施，除雪车、 除冰车等。 而要提高应急救援直升机地面保障水平， 增强航空应急救援能力，就必须首先掌握和了 解我国当前应急救援直升机地面保障的基本 状况和综合水平。目前，尚无学者直接针对应 急救援直升机地面保障能力评价问题进行研 究，但该问题本质上就是能力评价问题。国内 外学者关于能力评价问题的研究主要集中在 应急救援能力、公共服务能力、工作能力和保 障能力评价几个方面。应急救援能力评价主要 关注应急响应处置过程，通过识别影响因素、 或者进行场景模拟，进而构建评价模型 [5-13] ； 公共服务能力侧重于从社会一线获取数据，根 据公共服务类型不同分别构建评价指标体系 [14,15] ；工作能力评价集中表现为单个人员工作 能力评价，如维修人员、医护人员、运动员等， 针对不同的岗位需求分别识别能力影响因素 构建评价模型 [16-21] 。 保障能力评价问题集中体现在对装备设 备、维修、运行、物流、物资相关内容等方面 的保障能力进行评价。如装备设备方面，张兴 旺等人针对民航空管设备综合保障能力评价 指标的相对性和模糊性，运用专家函询法和层 次分析法，分析了影响系统保障能力的关键要 素，构建了评价指标体系，运用层次分析法确 定各指标的权重，运用模糊综合评价法建立了 评估模型 [22] ；在维修方面，林波在系统分析装 备维修保障能力评价理论的基础上，确定了装 备维修保障能力的评价内容，建立起具有较强 可操作性的炮兵部队装备维修保障能力评价 指标体系框架 [23] ；在运行方面，江涛分析了机 场运行特征和保障流程，探讨了机场安全运行 保障能力的内涵和构成要素，在此基础上提出 了机场安全运行保障能力的评价指标体系和 模型 [24] ；在物流方面，丁鹏玉运用访谈法和文 献统计方法提取应急物流保障能力要素初始 指标，编制了我国应急物流保障能力评价指标 体系测评量表，通过对量表调查数据的探索性 因子分析，构建出我国应急物流保障能力评价 体系 [25] ；在物资方面，陈璐在根据不同突发事 件对应急资源的需求差异进行情景设定的基 础上，构建了基于情景分析的应急资源保障能 力评价指标体系和评价模型，并用算例验证了 该模型用于评价应急资源保障能力的科学性 和适用性 [26] 。 综上，根据各种能力评价问题的研究方法 与思路，得出应急救援直升机地面保障能力评 价问题的解决思路为：在界定应急救援直升机 地面保障能力内涵的基础上，识别出应急救援 地面保障能力影响因素，根据影响因素构建应 急救援直升机地面保障能力评价指标体系，运 用数学方法构建应急救援直升机地面保障能 力评价模型。 94 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 2 应急救援直升机地面保障能力的内涵及 影响因素 本文研究的重点为直升机在参与应急救 援过程中安全起降所需的地面保障，据此给出 应急救援直升机地面保障的定义为：在灾害事 件发生后，直升机在受灾地区起飞和降落所需 要的所有地面工作的统称。具体可以分为起降 场地保障、目视助航保障、航空气象保障、通 信导航监视保障与后勤保障。与此相对应，应 急救援直升机地面保障能力则可以划分为起 降场地保障能力、目视助航保障能力、航空气 象保障能力、通信导航监视保障能力与后勤保 障能力。 2.1 起降场地保障能力影响因素分析 起降场地保障是保证直升机顺利起飞降 落的首要条件。直升机起飞区、进近区和离地 区是此类保障的关键因素；净空条件以及在遇 到不良天气时对于起降场地的保障能力对于 总体起降保障工作有较为重要的影响 [27] ；此 外，保障人员的经验技能和资格管理，以及起 降场地的维护保养也是影响起降场地保障的 重要因素。 2.2 目视助航保障能力影响因素分析 目视助航保障是整个直升机地面保障工 作当中较为重要的一环。目视助航设施设备的 情况是该类保障的关键，包括指示标、飞行区 地面标志和灯光系统 [28] ，其中，指示标又包括 风向标、着陆方向标等标志，对于机场地面保 障工作来说指示标的大小、颜色以及形状都代 表了不同的含义，因此，机场应保证拥有齐全 的指示标，且指示标的相关特性符合国家相关 要求。飞行区地面标志的相关要求与指示标的 类似，主要涉及标志的颜色、大小和形状。灯 光系统用于指示飞行区的边界轮廓、障碍物、 本场标志、风向指示、下滑坡度等，是航空器 空间或低能见度情况下进行起降的必备系统， 它包括边界灯、障碍灯、风向标志袋、下滑坡 度指示器、标志灯、泛光灯等多种灯具以及控 制器，对于灯光系统的要求则主要集中在灯具 质量、灯光强度、灯光颜色、灯光发光频率、 灯光的闪光顺序及灯光的摆放位置 [29] 。 2.3 航空气象保障能力影响因素分析 灾害发生后出现恶劣气象条件是不可避 免的，因此航空气象保障在整个地面保障工作 中显得尤为重要。气象人员的数量与专业技能 是影响航空气象保障能力的关键因素 [30] 。航空 气象设备包括常规观测设备（如电传式风向风 速仪）、图文传真设备、天气雷达、卫星云图 接受处理系统等 [31] ，相关要求主要包括在气象 设备的探测范围、分辨率、抗电磁干扰能力、 误码率（即接收云图的清晰程度）以及探测所 允许的 大误差 [32] 。 2.4 通信导航监视保障能力影响因素分析 通信导航监视保障工作对于直升机的航 行和定位至关重要。通信设备包括高频通信设 备、甚高频通信设备和手持地空对讲机；导航 设备包括测距仪、全向信标、指点信标等设备； 监视设备则包括一次监视雷达、二次监视雷达 和其他监视设备 [33] 。这些设备需要配套使用， 且需要安装在合理的位置，其探测范围也是衡 量通信导航监视保障工作的依据。此外电磁环 境与地理环境的良好性也是影响上述设施设 备工作的重要因素 [34] 。 2.5 后勤保障能力影响因素分析 后勤保障工作主要包括油料供应保障、供 电保障、特种车辆保障等。由于后勤保障工作 内容较多，因此保障人员的数量、技术水平和 管理水平均应达到要求。油料保障主要是保证 能够提供充足的质量达标的航油，拥有足够的 储油设备也是影响油料保障的因素 [35] 。供电保 障则主要是保证供电的稳定性，防止因为电压 不稳而对相应的航空设施设备造成损坏，此 外，备用电源的可靠性对于后勤保障也非常重 要 [36] 。特种车辆保障也是后勤保障的一部分， 主要涉及特种车辆的数量和质量。 3 应急救援直升机地面保障能力评价指标 体系 在分析了应急救援直升机地面保障能力 影响因素的基础上，构建应急救援直升机地面 保障能力评价三级指标体系，一级指标和二级 指标体现应急救援直升机地面保障能力的基 95 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 本构成要素，方便管理者找到地面保障有缺陷 的方面；三级指标是衡量应急救援直升机地面 保障能力 有效、 直接的基层元素。不同指 标对于地面保障能力的影响不同，通过权重大 小来体现。根据指标体系编制影响力度调查问 卷，通过计算每个选项对应的选择人数占总选 择人数的百分比来确定权重。通过网络向通用 航空工作人员和研究人员发放问卷共计 139 份，回收 139 份。得到的应急救援直升机地面 保障能力指标体系及权重如表 1 所示： 表 1 应急救援直升机地面保障能力评价指标体系及权重 一 级 指标 二级指标 三级指标 起降场地保障人员数量 0.073 起降场地保障人员专业技能 0.543 起降保障人员能力 0.331 起降场地保障人员的工作经验 0.384 区域大小 0.351 区域表面强度 0.441 进近起飞和离地区保障能力 0.298 区域坡度 0.208 起降场地的净空条件 0.573 起降场地环境保障能力 0.144 起降场地应对不良天气的能力 0.427 起降场地保障人员的培训与考核制度 0.505 起 降 场 地 保 障 能力 0.328 起降管理保障能力 0.227 起降场地的维修保养制度 0.495 目视助航保障人员的数量 0.082 目视助航保障人员的专业技能 0.553 目视助航保障人员能力 0.206 目视助航保障人员的工作经验 0.365 指示标的齐全性 0.455 指示标的颜色 0.221 指示标的大小 0.267 指示标保障能力 0.137 指示标的形状 0.057 飞行区地面标志的颜色 0.402 飞行区地面标志的形状 0.445 飞行区地面标志保障能力 0.217 飞行区地面标志的大小 0.153 灯具的质量 0.155 灯光强度 0.262 灯光发光频率 0.188 灯光颜色 0.122 灯光位置 0.203 灯光保障能力 0.220 灯光闪光顺序 0.070 目视助航设备所在地的地理位置 0.418 目视助航环境保障能力 0.090 目视助航保障工作应对不良天气的能力 0.582 目视助航设备管理人员培训考核制度 0.427 目 视 助 航 保 障 能力 0.132 目视助航管理保障能力 0.130 目视助航设备的维护检查制度 0.573 气象保障人员的数量 0.054 气象保障人员的专业技能 0.561 气象人员保障能力 0.322 气象保障人员的工作经验 0.385 气象设备的探测范围 0.205 航 空 气 象 保 障 能力 0.189 气象设备保障能力 0.311 气象设备的分辨率 0.194 96 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 气象设备的探测允许 大误差 0.224 气象设备的抗电磁干扰能力 0.221 气象设备的误码率（接收云图清晰程度）0.156 气象设备所处地区的电磁环境 0.398 气象环境保障能力 0.195 航空气象保障工作应对不良天气的能力 0.602 航空气象人员的培训与考核 0.524 气象管理保障能力 0.172 航空气象设备的维护制度 0.476 通信导航监视保障人员的数量 0.105 通信导航监视保障人员的专业技能 0.562 通信导航监视人员保障能力 0.300 通信导航监视保障人员的沟通能力 0.333 通信导航监视设备的配套率 0.327 通信导航监视设备安装位置合理性 0.530 通信导航监视设备保障能力 0.372 通信导航监视设备的信号覆盖率 0.143 通信导航监视设备所处的电磁环境 0.602 通 信 导 航 监 视 环 境 保 障 能 力 0.186 通信导航设备所处的地理环境 0.398 引导保障人员的培训考核制度 0.476 通 信 导 航 监 视 保 障 能力 0.216 通 信 导 航 监 视 设 备 管 理 能 力 0.142 设备维护保养制度 0.524 后勤保障人员的数量 0.177 后勤保障人员的技术水平 0.464 后勤人员保障能力 0.268 后勤管理人员的管理水平 0.359 储油设备的充足性 0.326 油料质量良好性 0.399 油料保障能力 0.22 油料的充足性 0.275 供电的稳定性 0.689 供电保障能力 0.208 备用电源的可靠性 0.311 特种车辆的数量 0.291 特种车辆保障能力 0.168 特种车辆的质量 0.709 后勤保障人员的培训考核制度 0.524 后 勤 保 障 能力 0.135 后勤管理能力 0.136 后勤设备等资源的库存补充制度 0.476 4 应急救援直升机地面保障能力评价模型 根据模糊综合评价法构建应急救援直升 机地面保障能力评价模型，其评价步骤 [37] 如图 1 所示。 4.1 建立因素集 评价因素集是指以被评价系统中各个影 响因素为元素所组成的集合，通常用 u 表示。 建立第一层次因素集：u={u1,u2,u3,u4,u5}， 其中 u1 代表起降场地保障能力，u2 代表目视 助航保障能力，u3 代表航空气象保障能力，u4 代表通信导航监视保障能力，u5 代表后勤保障 能力。 建立第二层次因素集：ui=｛ui1,ui2,...uin｝， 以 起 降 场 地 保 障 能 力 u1 为 例 ， 即 u1= ｛u11,u12,u13,u14｝，其中 u11 代表起降场地保障 人员能力，u12 代表进近起飞与离地区保障能 力，u13 代表起降场地环境保障能力，u14 代表 起降场地管理能力。 建立第三层次因素集，uij=｛uij1,uij2,...uijr｝， 以 起 降 场 地 保 障 人 员 能 力 为 例 ， 即 u11= ｛u111,u112,u113｝，其中，u111 代表起降保障人 员数量，u112 代表起降保障人员专业技能，u113 代表起降保障人员工作经验。 4.2 建立权重集 根 据 上 节 内 容 建 立 权 重 集 ： a= ｛a1,a2,...,am｝，ai(i=1,2,....,m)代表相应的因素 97 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 图 1 应急救援直升机地面保障能力评价步骤 对应的权重，以第一层次为例，a=｛a1,a2, a3, a4, a5｝，其中，a1=0.328，a2=0.132，a3=0.189， a4=0.216，a5=0.135。 4.3 建立评判集 评判集是评判者对评判对象可能做出的 各种评判结果所组成的集合，通常表示为 v= ｛v1, v2,..., vn｝。将评判结果分为很强、较强、 一般、较弱、很弱五种，则评判集 v=｛很强， 较强，一般，较弱，很弱｝。 4.4 单因素模糊评判 设对因素集 u 中第 i 个元素 ui 进行评判， 对于评判集 v 中第 j 个元素 vj 的隶属度为 rij, 则按照第 i 个元素 ui 的评判结果，可得出模糊 集合： ri=(ri1, ri2,..., rin) 同理，可得到相应于每个元素的单因素评 判集： r1=(r11, r12,..., r1n) r2=(r21, r22,..., r2n) … … … … rm=(rm1, rm2,..., rmn) 4.5 模糊综合评判 各单因素评判集的隶属度组成矩阵，又称 为评判矩阵。 r= ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎣ ⎡ mnm n rr rr ... ... 1 111 lll 根据评判矩阵 r 和权重集 a，则模型综合 决策模型为： b=a·r 5 实例分析 以国内某直升机机场为例，该机场是经军 方批准修建的飞行训练机场。机场包括 1600 米跑道和 68×800 米停机坪，建设 8 个 35x30 米机库和 2200 余平方米的航站航管综合楼， 航行气象、通信导航、供电供油、机务保障、 消防保安等各类设备和车辆一应俱全。能提供 飞行培训、公务飞行、低空旅游、飞行器试航、 应急救援、空中巡查等全面的通航飞行服务， 可满足大多数中小型喷气式公务机的起降。运 用上述评价模型对其应急救援直升机地面保 障能力进行评价，评价结果如表 2 所示. 根据表 2 后一行模糊综合评判结果可 以看出，第三项 0.295 明显大于其他四项，而 根据评判集 v=｛很强，较强，一般，较弱， 很弱｝，第三项 0.295 对应的是“一般”，因 此可以说明该直升机机场的航空应急救援地 面保障能力一般。 此外，还可看出具体某类保障工作的保障 能力，如起降场地保障能力一般，其中起降场 地人员保障能力较强，进近起飞和离地区的保 障能力一般，起降场地环境保障能力较强，起 降管理保障能力一般。其他类型保障工作的保 障能力也可同理得出。 6 结语 本文针对应急救援直升机在受灾地区安 全顺利起降所需的地面保障工作展开研究，得 出以下结论： （1）通过对能力评价问题进行研究，提 炼出了能力评价问题的共性的解决思路，即； 界定问题内涵、识别问题影响因素、构建评价 指标体系、建立评价模型； 98 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 表 2 某直升机机场地面保障能力评价结果 评价结果 目 标 层 一级 指标 二级指标 很强 较强 一般 较弱 很弱 起降场地人员保障能力 0.1543 0.347 0.206 0.2384 0.0543 进近起飞和离地区保障能力 0.2143 0.2441 0.4208 0.1 0.0208 起降场地环境保障能力 0.2 0.3247 0.3 0.1 0.0573 起降 场地 保障 能力 起降场地管理保障能力 0.1495 0.2495 0.3 0.1505 0.1505 起降场地保障能力 0.178 0.291 0.305 0.158 0.068 目视助航人员保障能力 0.1082 0.2529 0.3928 0.1553 0.0918 指示标保障能力 0.2057 0.3488 0.2676 0.0512 0.0267 航空地面标志保障能力 0.2402 0.2359 0.3641 0.1445 0.0153 灯光保障能力 0.2709 0.3048 0.2553 0.1483 0.0225 目视助航环境保障能力 0.2164 0.2582 0.3254 0.1 0.1 目视 助航 保障 能力 目视助航管理保障能力 0.3 0.1427 0.2427 0.1573 0.1573 目视助航保障能力 0.221 0.260 0.313 0.146 0.060 航空气象人员保障能力 0.1615 0.3284 0.3331 0.1385 0.0385 航空气象设备保障能力 0.2927 0.2639 0.3221 0.0989 0.0224 航空气象环境保障能力 0.3602 0.3 0.1398 0.1602 0.0398 航空 气象 保障 能力 航空气象管理保障能力 0.2476 0.2 0.3524 0.1524 0.0476 航空气象保障能力 0.256 0.281 0.295 0.133 0.035 通信导航监视人员保障能力 0.2105 0.4229 0.2333 0.1 0.0333 通信导航监视设备保障能力 0.2327 0.2143 0.3184 0.1673 0.0673 通信导航监视环境保障能力 0.2602 0.3 0.2398 0.1398 0.0602 通信 导航 监视 保障 能力 通信导航监视管理保障能力 0.1476 0.2 0.4 0.2 0.0524 通信导航监视保障能力 0.219 0.291 0.290 0.147 0.053 后勤人员保障能力 0.2928 0.3105 0.2249 0.1359 0.0359 油料保障能力 0.2725 0.3399 0.2 0.1275 0.0601 供电保障能力 0.2689 0.3689 0.2689 0.0311 0.0622 特种车辆保障能力 0.3 0.2291 0.2709 0.1291 0.0709 后勤 保障 能力 后勤管理能力 0.2524 0.2 0.4 0.1 0.0476 航 空 应 急 救 援 保 障 能 力 后勤保障能力 0.279 0.301 0.260 0.106 0.054 航空应急救援地面保障能力 0.221 0.286 0.295 0.142 0.056 （2）通过对直升机在受灾地区安全起降 所需的地面保障工作进行分析，得出应急救援 直升机地面保障能力可以分解为起降场地保 障能力、目视助航保障能力、航空气象保障能 力、通信导航监视保障能力和后勤保障能力， 为系统地建立健全直升机地面保障工作提供 切入点； （3）根据每种类别保障能力影响因素的 识别结果，从人、机、环、管四种系统要素构 成的角度提出了 23 个二级指标和 64 个三级指 标，为受灾地区直升机地面保障能力改进与提 高工作的具体实施提供指导； （4）通过实例验证得出，本文构建的应 急救援直升机地面保障能力评价模型能够为 地面保障单位评估自身保障能力和水平提供 有效参考。 基于此，未来研究工作将围绕以下两方面 展开： （1）对应急救援直升机地面保障能力评 价指标进行优化，使其评价结果客观化与定量 99 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 化； （2）以应急救援直升机地面保障能力评 价工作为基础，进行应急救援直升机地面装备 配备和人员资质研究及标准研究。 参考文献 [1] 韩榕,安锦.通航应急救援体系有待建立.中国航 空报,2013,4(2). [2] national air transportation association. general aviation in the united states american: nata , 2009. [3] rod beck a short of general aviation fbo’s and the aviation client [eb/ol]. http://www.get aviation.com/blog/airplanes-business-use-produc tivity/a-short-history-of-general-aviation-fbo%e2 %80%99s-and-the-aviation-client. 2011-10-13. [4] 于清媛.我国通航运行保障模式发展建议.中国 民用航空,2014,28(3):99-101. [5] shaohua shang; chunghwei su. the evaluation of fire rescue capability from post-earthquake fires in taipei, taiwan.journal of earthquake and tsunami. 2014,8(5):145-147. [6] 吴新燕.城市灾害救援能力评价指标体系的研 究:(硕士学位论文).河北:河北理工学院,2003. [7] 迟娜娜.城市灾害应急能力评价指标体系研 究 :( 硕士学位论文 ). 北京 : 首都经济贸易大 学,2006. [8] 田依林,杨青.基于 ahp_delphi 法的城市灾害 应急能力评价指标体系模型设计.武汉理工大 学学报,2008,32(1):168-171. [9] 徐吉辉,楚维.基于 ahp_fce 的航空应急救援能 力评价研究.航空维修与工程,2012,24(3):60-62. [10] 张丽敏,张琪.基于 iahp 法的民用机场应急救 援 能 力 评 价 . 消 防 技 术 与 产 品 信 息,2013,18(12):21-24. [11] 王晓东,吴群红.突发公共卫生事件应急能力评 价指标体系构建研究.中国卫生经济, 2013, 32(6): 47-50. [12] olivier salvi, samuel delsinne, biogas a european perspective on safety and regulation, journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). [13] shuzhen li, alateng tuya, disaster risk research literature on statistics analysis in china journal net, journal of risk analysis and crisis response, vol. 5, no. 2 (july 2015), 129 140. [14]邵旭.成都市安全生产公共服务能力评价指标 体系构建研究:(硕士学位论文)成都:电子科技 大学,2015. [15] 杨丽.贫困地区农村卫生室公共卫生服务能力 评价指标体系初步研究:(硕士学位论文)武汉: 华中科技大学,2014. [16] nai ruo xing,xiao bin hui. the post ability evaluation of maintenance personnel based on the g1-topsis algorithm. applied mechanics and materials,2013,347(12):3051-3055. [17] 周厚秀.护理本科生临床能力评价指标体系的 研究:(硕士学位论文)重庆:第三军医大学,2006. [18] 熊磊.篮球运动员竞技能力评价指标体系的构 建:(硕士学位论文)武汉:武汉体育大学,2014. [19] 王彩霞.博士研究生科研能力评价指标体系及 评价方法研究:(博士学位论文)成都:西南交通 大学,2006. [20]k. harald drager, tiems drevs initiative: disaster resilience establishment for vulnerable societies, journal of risk analysis and crisis response, vol. 4, no. 3 (september 2014), 133 140. [21] tianhan jiang, assessment method of emergency preparedness system vulnerability based on the complex network theory, journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 195-200. [22] 张兴旺,陶煜.基于模糊综合评价的民航空管设 备保障能力评估.价值工程.2012,14(6):25-26. [23] 林波.基于灰色聚类炮兵装备维修保障能力评 价及应用研究:(硕士学位论文)长沙:国防科学 技术大学,2009. [24] 江涛.机场安全运行保障能力模型研究:(硕士 学位论文)天津:中国民航大学,2014. [25] 丁鹏玉.我国应急物流保障能力评价指标体系. 物流技术,2015,34(9):140-142. [26] 陈璐.基于情景分析的应急资源保障能力评价 模式研究:(硕士学位论文)郑州:河南理工大 学,2014. [27] 刘志浩.机场净空安全保障水平提升方法研 究:(硕士学位论文)天津:中国民航大学,2016. [28] gb/t17836-1999,通用航空机场设备设施.北京: 中国标准出版社,1999. [29] 徐承东.初探连云港机场灯光保障及战备应急 抢修.江苏航空,1999,28(3):178-179. [30] matt ebbatson, don harris. crosswind landings in general aviation: a modified method of reporting wind information to the pilot. the international journal of aviation psychology, 2007, 17(4): 353-370. [31] mh/t4016.4-2004,民用航空气象第四部分:设 备配备.北京:中国民航出版社,2004. [32] mh/t4016.5-2004,民用航空气象第五部分:设 备技术要求.北京:中国民航出版社,2004. [33] mh/t5026-2012,通用机场建设规范.北京:中国 民航出版社,2012. [34] gb6364-86,航空无线电导航台站电磁环境要求. 北京:中国标准出版社,1986. [35] 张凤.我国通用机场航油保障模式探析.西安航 空学院学报,2016,34(2):9-11. [36] 王志勇,陈峰.机场供电及助航灯光保障体系设 计与分析.民用航空,2001,24(3):9-10. [37] 张景林,崔国璋.安全系统工程.北京:煤炭工业 出版社,2002. 100 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 93-100 1 引言 2 应急救援直升机地面保障能力的内涵及影响因素 3 应急救援直升机地面保障能力评价指标体系 4 应急救援直升机地面保障能力评价模型 5 实例分析 6 结语 © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 111-122 doi: https://doi.org/10.2991/jracr.k.190703.007; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr summary of researches on basis risk in weather index insurance yueqin wang1,2, sijian zhao1,2* 1. agricultural information institute of chinese academy of agricultural sciences, key laboratory of digital agricultural early-warning technology, moa, beijing 100081, china 2. china institute of actuarial science, central university of finance and economics, beijing 100081, china received january 1,2019 accepted june 25, 2019 abstract agricultural weather index insurance overcomes the technology and management problems existing in traditional agricultural insurance, and can effectively transfer agricultural meteorological disasters risk, providing strong guarantee for farmers with small production scale and scattered land. although weather index insurance has many advantages, governments around the world have made a lot of efforts to expand the coverage, but farmers' demand is still not high. one of the reasons is basis risk. how to reduce it, and improve the operational efficiency of weather index insurance is an important issue that needs to be solved urgently. this paper has summarized researches on the theories and practices of weather index insurance at home and abroad, and systematically comments on the basis risk’s definition, formation mechanism, quantitative methods and management measures. in order to provide useful ideas and references for the development of china's weather index insurance and the future research on basis risk. keywords: agricultural weather index insurance, basis risk, formation mechanism, quantitative methods, risk management 农业天气指数保险中基差风险的研究进展 王月琴 1,2 , 赵思健 1,2* 1. 中国农业科学院农业信息研究所/农业部智能化农业预警技术重点开放实验室，北京 100081 2. 中央财经大学中国精算研究院，北京 100081 摘要：农业天气指数保险克服了传统农业保险中的技术管理难题，能够有效转移分散农业气象灾害风险， 为生产规模小、地块分散的发展中国家提供有力保障。虽然天气指数保险有诸多优势，各国政府也为扩大 天气指数保险的承保范围做出很多努力，但农户对天气指数保险的需求仍然不高，造成这种现象的主要原 因之一是基差风险的存在。如何降低基差风险、提高天气指数保险的实施效果是当前亟待解决的重要问题。 本文在总结国内外天气指数保险理论与实践的基础上，针对天气指数保险基差风险的界定、影响机理、量 化方法、管理手段等方面进行了系统的梳理及评述，以期为我国天气指数保险的发展和今后基差风险的研 究提供有益的思路和参考。 关键词：农业天气指数保险，基差风险，形成机理，量化方法，风险管理 * 通讯作者：赵思健，中国农业科学院农业信息研究所副研究员，邮箱：zhaosijian@caas.cn。 第一作者：王月琴，中国农业科学院农业信息研究所博士研究生，邮箱：1105118850@qq.com。 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 112 1. 引言 全球气候变化增加了气象灾害风险的发生，频 率高、强度大的气象灾害给农业生产带来巨大的损 失。我国农业生产大都是小规模经营，基于“个体” 的传统农业保险在推行过程中出现了勘查定损难、 交易成本高等问题，大大削弱了农业保险作为风险 管理手段的作用。因此，创新农业保险机制、降低 定损理赔成本、拓宽保障范围，已成为关乎我国农 业保险未来发展的重要命题 1 。农业天气指数保险 （以下天气指数保险均指农业领域）是将一个或几 个气象要素对农作物的损害程度指数化，当天气指 数对产量的影响达到一定水平，农户就可获得相应 标准的赔偿，是一种对气象导致的农作物产量下跌 风险进行保障的保险产品。天气指数保险不基于农 户的实际损失，而是基于区域气象部门的数据进行 赔付，无需勘察定损，理赔迅速，可有效降低交易 成本，有助于应对农业生产中的系统性气象风险 2 ， 因此在实践中得到了广泛的应用。 天气指数保险适合“农民数量多、生产规模小” 的发展中国家，被越来越多的学者和保险公司视为 发展中国家应对农业气象灾害风险的有效替代方式 3 。 2002 年，墨西哥首先将天气指数保险运用于农业领 域。在世界银行的协助和推广下，天气指数保险陆 续在印度、马拉维、孟加拉、埃塞俄比亚等发展中 国家开展起来。其中，印度是天气指数保险商品化 程度最高的国家 4 。在我国，天气指数保险引进时间 较短。2007 年上海安信农业保险公司推出全国首个 西瓜天气指数保险产品。随后各种天气指数保险产 品陆续在全国各地开展起来，涉及的农产品多样， 主要有水稻、小麦、玉米等粮食作物，柑橘、杨梅 等水果，茶叶、棉花等特色经济作物，虾、蟹等水 产品，牛、羊等牲畜，其中特色经济作物居多。随 着我国农村保险需求的不断提高，天气指数保险已 成为农业保险多元化创新发展的重要驱动力。政府 对天气指数保险的重视程度越来越高，在农业保险 的相关政策中也越来越多被提到（表 1）。可以预 见，在未来，天气指数保险将成为我国转移气象灾 害风险的重要方式。 尽管与传统农业保险相比，天气指数保险优势 明显，但作为指数保险的一种，天气指数保险不可 避免的存在基差风险（basis risk），即根据指数得 到的理赔金额与投保人实际受损可能并不吻合。例 如，sarah 等指出基差风险是指数保险等区域保险产 品所固有的，当保险赔付依靠指数而不取决于投保 人的实际损失时就会发生，使保险保障偏离投保人 的预期 5 。几乎所有涉及天气指数保险的文献都达成 表 1. 与天气指数保险相关的国家政策 年份 部门 文件 内容 2014 年 国务院 《关于加快发 展现代保险服 务业的若干意 见》 明确提出要鼓励探索天 气指数保险等新兴产品 和服务，丰富农业保险 风险管理工具。 2015 年 保监会 《关于做好农 业气象灾害理 赔和防灾减损 工作的通知》 要求各财产保险公司 “加快推进天气指数保 险”。 2016 年 中共中 央、国 务院 “中央一号” 文件 提出“积极开发适应新 型农业经营主体需求的 保险品种。探索开展重 要农产品目标价格保险 以及收入保险、天气指 数保险试点”。 2016 年 农业部 《关于开展 2016 年度金融 支农服务创新 试点的通知》 重点提到通过创新天气 指数保险等方式探索运 用保险以及保险与其他 金融工具融合，促进我 国农业现代化发展的有 效模式。 2017 年 国务院 《关于加快构 建政策体系培 育新型农业经 营主体的意 见》 提出要加快开展天气指 数保险的试点。 2018 年 国务院 《乡村振兴战 略规划（2018 －2022 年）》 提出要完善农业保险政 策体系，设计多层次、 可选择、不同保障水平 的保险产品，鼓励开展 天气指数保险等试点， 完善农业风险管理和预 警体系。 资料来源：作者整理。 普遍共识：基差风险是天气指数保险的最大缺陷 6-8 ； 基差风险并不能完全消除，但可通过保险产品的合 理设计来降低 9 ；天气指数保险的最大挑战是如何降 低基差风险 10 。目前国内外关于天气指数保险基差 风险的研究并没有很深入，只是认识到基差风险是 天气指数保险的劣势，影响了农民参与保险的意愿， 但基差风险究竟在多大程度上影响天气指数保险的 推广、如何量化基差风险以及具体的降低基差风险 的方法，目前的研究很少涉及 11-13 。基于上述背景， 本文系统梳理和评述了国内外天气指数保险基差风 险的研究进展，以期为进一步研究基差风险提供理 论和方法指引。主要从以下几个方面展开：第 1 部 分引言部分交代研究的背景和意义；第 2 部分是关 于基差风险的概念界定研究，第 3 部分主要阐述了 基差风险如何降低天气指数保险的需求；第 4 部分 归纳总结基差风险的形成机理，第 5 部分和第 6 部 分分别梳理量化和管理基差风险的方法，最后一部 分是本文的结论和研究展望。 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 112 113 1. 引言 全球气候变化增加了气象灾害风险的发生，频 率高、强度大的气象灾害给农业生产带来巨大的损 失。我国农业生产大都是小规模经营，基于“个体” 的传统农业保险在推行过程中出现了勘查定损难、 交易成本高等问题，大大削弱了农业保险作为风险 管理手段的作用。因此，创新农业保险机制、降低 定损理赔成本、拓宽保障范围，已成为关乎我国农 业保险未来发展的重要命题 1 。农业天气指数保险 （以下天气指数保险均指农业领域）是将一个或几 个气象要素对农作物的损害程度指数化，当天气指 数对产量的影响达到一定水平，农户就可获得相应 标准的赔偿，是一种对气象导致的农作物产量下跌 风险进行保障的保险产品。天气指数保险不基于农 户的实际损失，而是基于区域气象部门的数据进行 赔付，无需勘察定损，理赔迅速，可有效降低交易 成本，有助于应对农业生产中的系统性气象风险 2 ， 因此在实践中得到了广泛的应用。 天气指数保险适合“农民数量多、生产规模小” 的发展中国家，被越来越多的学者和保险公司视为 发展中国家应对农业气象灾害风险的有效替代方式 3 。 2002 年，墨西哥首先将天气指数保险运用于农业领 域。在世界银行的协助和推广下，天气指数保险陆 续在印度、马拉维、孟加拉、埃塞俄比亚等发展中 国家开展起来。其中，印度是天气指数保险商品化 程度最高的国家 4 。在我国，天气指数保险引进时间 较短。2007 年上海安信农业保险公司推出全国首个 西瓜天气指数保险产品。随后各种天气指数保险产 品陆续在全国各地开展起来，涉及的农产品多样， 主要有水稻、小麦、玉米等粮食作物，柑橘、杨梅 等水果，茶叶、棉花等特色经济作物，虾、蟹等水 产品，牛、羊等牲畜，其中特色经济作物居多。随 着我国农村保险需求的不断提高，天气指数保险已 成为农业保险多元化创新发展的重要驱动力。政府 对天气指数保险的重视程度越来越高，在农业保险 的相关政策中也越来越多被提到（表 1）。可以预 见，在未来，天气指数保险将成为我国转移气象灾 害风险的重要方式。 尽管与传统农业保险相比，天气指数保险优势 明显，但作为指数保险的一种，天气指数保险不可 避免的存在基差风险（basis risk），即根据指数得 到的理赔金额与投保人实际受损可能并不吻合。例 如，sarah 等指出基差风险是指数保险等区域保险产 品所固有的，当保险赔付依靠指数而不取决于投保 人的实际损失时就会发生，使保险保障偏离投保人 的预期 5 。几乎所有涉及天气指数保险的文献都达成 表 1. 与天气指数保险相关的国家政策 年份 部门 文件 内容 2014 年 国务院 《关于加快发 展现代保险服 务业的若干意 见》 明确提出要鼓励探索天 气指数保险等新兴产品 和服务，丰富农业保险 风险管理工具。 2015 年 保监会 《关于做好农 业气象灾害理 赔和防灾减损 工作的通知》 要求各财产保险公司 “加快推进天气指数保 险”。 2016 年 中共中 央、国 务院 “中央一号” 文件 提出“积极开发适应新 型农业经营主体需求的 保险品种。探索开展重 要农产品目标价格保险 以及收入保险、天气指 数保险试点”。 2016 年 农业部 《关于开展 2016 年度金融 支农服务创新 试点的通知》 重点提到通过创新天气 指数保险等方式探索运 用保险以及保险与其他 金融工具融合，促进我 国农业现代化发展的有 效模式。 2017 年 国务院 《关于加快构 建政策体系培 育新型农业经 营主体的意 见》 提出要加快开展天气指 数保险的试点。 2018 年 国务院 《乡村振兴战 略规划（2018 －2022 年）》 提出要完善农业保险政 策体系，设计多层次、 可选择、不同保障水平 的保险产品，鼓励开展 天气指数保险等试点， 完善农业风险管理和预 警体系。 资料来源：作者整理。 普遍共识：基差风险是天气指数保险的最大缺陷 6-8 ； 基差风险并不能完全消除，但可通过保险产品的合 理设计来降低 9 ；天气指数保险的最大挑战是如何降 低基差风险 10 。目前国内外关于天气指数保险基差 风险的研究并没有很深入，只是认识到基差风险是 天气指数保险的劣势，影响了农民参与保险的意愿， 但基差风险究竟在多大程度上影响天气指数保险的 推广、如何量化基差风险以及具体的降低基差风险 的方法，目前的研究很少涉及 11-13 。基于上述背景， 本文系统梳理和评述了国内外天气指数保险基差风 险的研究进展，以期为进一步研究基差风险提供理 论和方法指引。主要从以下几个方面展开：第 1 部 分引言部分交代研究的背景和意义；第 2 部分是关 于基差风险的概念界定研究，第 3 部分主要阐述了 基差风险如何降低天气指数保险的需求；第 4 部分 归纳总结基差风险的形成机理，第 5 部分和第 6 部 分分别梳理量化和管理基差风险的方法，最后一部 分是本文的结论和研究展望。 2. 基差风险的界定研究 风险是指在未来一定的时间和空间范围内，由 于决策者无法确定与控制的外在因素，导致决策者 在特定的活动中获得的实际收益低于预期收益的可 能性 14 。基差风险也是风险的一种，是农户预期损 失的不确定性，是农户承担的未被保险的剩余风险 （residual risk），属于保险赔付的公平性问题 15-17。 从已有文献可以发现，国内外对基差风险的界定主 要是从根据保险赔付与实际损失关系这一角度，或 体现在二者不完全相关，或体现在二者值的差异。 学者们认为按照区域性指数而未按实际损失的赔付 方法，虽然数据客观、理赔方便迅速，但由于指数 和实际损失之间存在不完全相关性 9 ，保险赔付无法 完全匹配每个农户的实际损失 6, 7, 10, 13, 15, 18-20 ：有些 农户虽未达到区域指数触发值却得到赔偿；有些农 户虽发生严重损失却因未触发指数而没有得到赔偿 或者得到的赔偿不足以弥补损失。即使同一个种植 区域的农户受到同种气象灾害的侵袭，灾后的实际 受损情况也可能不同，故保险保障的效果也会因人 而异。 3. 基差风险对保险需求的影响研究 目前天气指数保险的试点区域有限且区域内有 效需求十分不足。在已有关于天气指数保险需求的 研究方面，大多人认为农业收入占比、农户受教育 程度、种植规模、农作物受灾程度等是影响天气指 数保险需求的因素 21 。基差风险的存在到底是否会 影响天气指数保险的推广，在多大程度上会降低农 户对天气指数保险的需求，是未来值得深入探讨的 问题。国内外有部分学者研究了基差风险对保险需 求的影响，主要从以下两个方面展开： 一是理论分析层面。carter 等将风险/模糊厌恶 ambiguity aversion）理论引入基差风险影响天气指 数保险需求的研究中，指出农户（风险者）不知道 天气指数保险赔付的概率分布，保险的赔付出现不 确定性，而且从历史上看很可能出现有损失但得不 到赔付的情况，所以农户不愿意冒着基差风险而去 购买指数保险 22 。elabed 和 carter 利用样本中马里 棉农偏好特征分布来预测指数保险需求如何随着基 差风险而变化，分别考虑基差风险对个人期望效用 最大化需求和考虑风险厌恶需求的影响，结果显示 二者都随基差风险的增加而降低，但考虑了风险厌 恶的需求与个人期望效用最大化的需求相比，保险 需求降低了一半，得出需要降低基差风险来提高指 数保险需求的结论 23 。 二是定量分析层面。mobarak 等基于基差风险 和指数保险需求有相关关系的假定，将到气象站的 距离作为基差风险的替代性指标，使用随机实验法 研究距自动降雨站不同距离处的农户的保险需求。 研究结果显示距离每增加 1 公里，指数保险的需求 就下降 6.4％24。hill 等同样采用这种替代性指标计 算方法证明了基差风险的存在会显著降低指数保险 的需求 25 。jensen 等使用家庭尺度数据来研究影响 牲畜指数保险需求的因素，结果表明基差风险和空 间逆向选择在牲畜指数保险需求方面起着重要作用 26 。除此之外，还有用问卷调查的方法研究基差风险 对保险需求的影响。例如，王振军等对陇东黄土高 原区 524 户农户进行问卷调查，研究显示相同保障 水平下由于基差风险存在，天气指数保险的购买意 愿要低于传统农业保险 27 。 4. 基差风险的形成机理研究 在确定基差风险会降低天气指数保险的需求后， 了解基差风险是如何形成以及受到哪些因素的影响 具有十分重要的意义。纵观国内外对基差风险形成 机理的研究，可以发现影响基差风险的因素大抵可 以归为三类（表 2）——作物生长期变化带来的时 间基差风险、空间异质性导致的空间基差风险、保 险设计产生的产品基差风险 28 。 4.1. 时间基差风险 总结国内外关于时间基差风险的研究，可以发 现时间基差风险主要是由于历史年际间作物的生长 期变化引起的，若用固定生长期设计保险期间会产 生时间基差风险。diaz nietoa 等 29，collier 等 6 认为 基差风险的来源之一是存在时间风险，即天气事件 对农作物每个生长阶段产量的影响在不同年份可能 有所不同。牛浩，陈盛伟也认为目前的天气指数保 险产品存在作物生长阶段难以随年际间气象条件的 变化作灵活调整的问题。农作物生长期不是固定不 变的，年际间种植时间的偏差和气象条件的不同都 会造成作物生长期时间范围发生变化。因此，在不 同年份采用固定的生长期时间会存在时间基差风险 30 。 4.2. 空间基差风险 从已有对基差风险形成机理的研究中可以发现， 大多数人认为空间差异是基差风险的一个重要的来 源。空间基差风险是由于空间异质性造成的，这种 异质性最终导致个体农户的实际损失与区域损失不 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 114 完全匹配。目前对空间基差风险的认识主要有两个 方面： 一是区域内系统性（covariate risk）天气风险 导致的空间基差风险。diaz nietoa 等 29，collier 等 6 认为由于空间异质性的存在导致同一天气事件对不 同农户作物产量的影响可能不同。barnett 和 mahul 认为基差风险来自于天气变量的空间差异性，即天 气变量在空间上并非高度共变（ highly spatially covariate），目标区域可能存在区域性微气候，导 致个体农户之间的实际受灾程度并不完全一致 7 。 giné 和 townsend31，rao32 认为小范围天气的差异性 使个体农户实际受损不一致进而导致基差风险的存 在。尹东也认为天气指数保险中的气象要素存在分 布不统一的空间风险 33 。牛浩，陈盛伟认为天气指 数设计忽略了不同地域之间的差异，区域内小气候 的存在造成了天气指数保险空间误差的发生 30 。 二是除天气外的非系统性风险（idiosyncratic risk ） 导 致 的 空 间 基 差 风 险 。 例如， barnett 和 mahul 指出引起农作物减产的原因除了天气风险之 外，还可能与病虫害、个体农户的生产技能、土壤 质量、农作物品种等其他因素有关，这些因素导致 了基差风险的发生 7 。giné 和 townsend31，rao32 也 认同受灾情况不光与天气有关，还有病虫害等其他 非天气因素有关。conradt 指出天气指数保险只承保 天气等系统性风险，没有将病虫害等非系统性风险 包含其中，而且承保地域越大，投保人面临的风险 异质性越大，指数保险越难预测产量 5 。陈晓峰认为 区域内的气候、土壤、地形、海拔等自然禀赋的同 质化程度以及田间管理的标准化程度都会直接影响 基差风险的大小 34 。丁少群，罗婷指出基差风险产 生的原因之一是保险产品中没有考虑个体之间的差 异（地形，防范措施等），这会增大基差风险 13 。 4.3. 产品基差风险 产品基差风险是人为设计误差造成的，没有将 影响作物损失的所有气象灾害考虑在内、模拟拟合 不准确等都可能造成基差风险。例如，rao 指出产 品设计不合理会产生基差风险 32 。conradt 指出若气 象站与个别农场距离太远，天气指数的设计则无法 准确反映天气条件的变化；若产品未能考虑天气条 件对作物不同生长阶段的敏感性影响时，会导致对 实际损失的估计较差，从而出现设计错误 5 。牛浩， 陈盛伟认为所有气象因素未能被全部考虑在内、历 史数据缺乏、数据质量低、风险自救措施，这些因 素的存在会导致只根据气象统计数据来构建保险产 品产生设计误差。另外，模型方法准确性也值得商 榷 30 。丁少群，罗婷就指出若保险定价模型出现差 错则特定气象因子与实际损失率之间的关系就不会 被准确刻画出来，会产生产品设计基差风险 13 。 表 2. 基差风险的类型 类型 影响因素 说明 时间基差 风险 作物生长期变 化 年际间种植时间的偏差和气象条 件的不同会造成作物生长期时间 范围发生变化。 空间基差 风险 地理空间异质 性 （1）系统性天气风险：气象站 与个体之间的距离不同，气象变 量在空间上并非高度共变，存在 区域微气候； （2）非系统性风险：除天气之 外，作物的产量还受病虫害、管 理水平、土壤质量、作物品种等 其他非系统性风险影响。 产品基差 风险 产品设计准确 性 产品设计的气象要素和产量并不 完全相关，模型拟合效果不好， 天气指数选取不准确等。 5. 基差风险的量化研究 如何衡量基差风险的大小是天气指数保险基差 风险研究至关重要的一步。目前学者们普遍认为指 数设定与承保区域产量越相关，基差风险就越小， 也有学者认为这种相关性体现在天气指数和农户所 在地的气候条件上。量化基差风险的方法一般分为 两种：一种是衡量基差风险的程度；另一种是衡量 基差风险的数量。 （1）衡量基差风险的程度。目前大多数学者是 利用作物区域气象/产量与个体农户的相关性来衡量 基差风险，这种方法只能间接地定性描述基差风险 的风险程度高低，无法定量得出基差风险到底有多 大。例如，世界银行 9 ，fuchs 和 wolff35 指出指数和 农场一级的产量相关性的大小决定了基差风险的大 小。张峭 17 ，张惠茹 10 也认为如果天气指数和个体 农户的气候条件高度相关，基差风险就会相对较小。 徐磊，张峭进一步指出基差风险的大小与地区总水 平和个体农户相关度成反比，在其他条件一定的情 况下，相关度越小，基差风险越大 36 。除了理论上 的分析，还有部分学者利用实证验证了基差风险的 存在。例如，clarke 和 mahul 等使用研究区域作物 平均产量占其历史（1999 2007 年）平均产量的比 重来表示增产减产程度，用 2006 年印度降雨指数保 险的赔付率来表示保险赔付情况，二者进行相关性 分析发现作物产量与保险赔付之间存在弱的相关性， 用这种弱相关性表示该保险产品存在的基差风险较 大 37 。杨太明，刘布春等在设计安徽省宿州市冬小 麦天气指数保险时，将历史天气指数赔付率与历史 产量损失率的对比定义为基差比函数（公式 1）， journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 114 115 完全匹配。目前对空间基差风险的认识主要有两个 方面： 一是区域内系统性（covariate risk）天气风险 导致的空间基差风险。diaz nietoa 等 29，collier 等 6 认为由于空间异质性的存在导致同一天气事件对不 同农户作物产量的影响可能不同。barnett 和 mahul 认为基差风险来自于天气变量的空间差异性，即天 气变量在空间上并非高度共变（ highly spatially covariate），目标区域可能存在区域性微气候，导 致个体农户之间的实际受灾程度并不完全一致 7 。 giné 和 townsend31，rao32 认为小范围天气的差异性 使个体农户实际受损不一致进而导致基差风险的存 在。尹东也认为天气指数保险中的气象要素存在分 布不统一的空间风险 33 。牛浩，陈盛伟认为天气指 数设计忽略了不同地域之间的差异，区域内小气候 的存在造成了天气指数保险空间误差的发生 30 。 二是除天气外的非系统性风险（idiosyncratic risk ） 导 致 的 空 间 基 差 风 险 。 例如， barnett 和 mahul 指出引起农作物减产的原因除了天气风险之 外，还可能与病虫害、个体农户的生产技能、土壤 质量、农作物品种等其他因素有关，这些因素导致 了基差风险的发生 7 。giné 和 townsend31，rao32 也 认同受灾情况不光与天气有关，还有病虫害等其他 非天气因素有关。conradt 指出天气指数保险只承保 天气等系统性风险，没有将病虫害等非系统性风险 包含其中，而且承保地域越大，投保人面临的风险 异质性越大，指数保险越难预测产量 5 。陈晓峰认为 区域内的气候、土壤、地形、海拔等自然禀赋的同 质化程度以及田间管理的标准化程度都会直接影响 基差风险的大小 34 。丁少群，罗婷指出基差风险产 生的原因之一是保险产品中没有考虑个体之间的差 异（地形，防范措施等），这会增大基差风险 13 。 4.3. 产品基差风险 产品基差风险是人为设计误差造成的，没有将 影响作物损失的所有气象灾害考虑在内、模拟拟合 不准确等都可能造成基差风险。例如，rao 指出产 品设计不合理会产生基差风险 32 。conradt 指出若气 象站与个别农场距离太远，天气指数的设计则无法 准确反映天气条件的变化；若产品未能考虑天气条 件对作物不同生长阶段的敏感性影响时，会导致对 实际损失的估计较差，从而出现设计错误 5 。牛浩， 陈盛伟认为所有气象因素未能被全部考虑在内、历 史数据缺乏、数据质量低、风险自救措施，这些因 素的存在会导致只根据气象统计数据来构建保险产 品产生设计误差。另外，模型方法准确性也值得商 榷 30 。丁少群，罗婷就指出若保险定价模型出现差 错则特定气象因子与实际损失率之间的关系就不会 被准确刻画出来，会产生产品设计基差风险 13 。 表 2. 基差风险的类型 类型 影响因素 说明 时间基差 风险 作物生长期变 化 年际间种植时间的偏差和气象条 件的不同会造成作物生长期时间 范围发生变化。 空间基差 风险 地理空间异质 性 （1）系统性天气风险：气象站 与个体之间的距离不同，气象变 量在空间上并非高度共变，存在 区域微气候； （2）非系统性风险：除天气之 外，作物的产量还受病虫害、管 理水平、土壤质量、作物品种等 其他非系统性风险影响。 产品基差 风险 产品设计准确 性 产品设计的气象要素和产量并不 完全相关，模型拟合效果不好， 天气指数选取不准确等。 5. 基差风险的量化研究 如何衡量基差风险的大小是天气指数保险基差 风险研究至关重要的一步。目前学者们普遍认为指 数设定与承保区域产量越相关，基差风险就越小， 也有学者认为这种相关性体现在天气指数和农户所 在地的气候条件上。量化基差风险的方法一般分为 两种：一种是衡量基差风险的程度；另一种是衡量 基差风险的数量。 （1）衡量基差风险的程度。目前大多数学者是 利用作物区域气象/产量与个体农户的相关性来衡量 基差风险，这种方法只能间接地定性描述基差风险 的风险程度高低，无法定量得出基差风险到底有多 大。例如，世界银行 9 ，fuchs 和 wolff35 指出指数和 农场一级的产量相关性的大小决定了基差风险的大 小。张峭 17 ，张惠茹 10 也认为如果天气指数和个体 农户的气候条件高度相关，基差风险就会相对较小。 徐磊，张峭进一步指出基差风险的大小与地区总水 平和个体农户相关度成反比，在其他条件一定的情 况下，相关度越小，基差风险越大 36 。除了理论上 的分析，还有部分学者利用实证验证了基差风险的 存在。例如，clarke 和 mahul 等使用研究区域作物 平均产量占其历史（1999 2007 年）平均产量的比 重来表示增产减产程度，用 2006 年印度降雨指数保 险的赔付率来表示保险赔付情况，二者进行相关性 分析发现作物产量与保险赔付之间存在弱的相关性， 用这种弱相关性表示该保险产品存在的基差风险较 大 37 。杨太明，刘布春等在设计安徽省宿州市冬小 麦天气指数保险时，将历史天气指数赔付率与历史 产量损失率的对比定义为基差比函数（公式 1）， 基差比越接近于 0，表明基差风险越小。例如，历 年平均赔付率为 5. 84% ，历年平均产量损失率为 -5. 33% ，基差比则为 -8.73% 38 。 （2）衡量基差风险的数量。有部分国外学者利 用替代性指标——样本数量到气象站的距离来直接 量化基差风险的大小。例如 woodard 和 garcia39 、 wang 和 zhang40、 odening 和 musshoff 等 41 认为基 差风险来自于天气变量之间的空间相关性，并将其 描述为气象站间距离的函数。mobarak 等 24 ，hill 等 25 基于基差风险和指数保险需求有相关关系的假 定，将样本数量到气象站的距离作为基差风险的替 代性指标，用距离的大小来刻画基差风险的大小。 朱俊生也赞同天气指数保险中气象站与保险标的之 间的距离远近与空间基差风险的大小有关 42 。另外， 还有利用保费空间差异性来量化基差风险的方法。 例如，norton 和 turvey 等认为应该根据相邻位置保 险保费支出的不同，考量空间地理特征（观测站之 间的高度、经纬度）的差异进而量化天气指数保险 的基差风险 43 。 6. 基差风险的管理研究 虽然天气指数保险能弥补传统农业保险的很多 缺陷，但并不是所有地区都有适宜发展天气指数保 险，若在气象对产量影响不大的地区实施天气指数 保险则会产生严重的基差风险问题。因此，保险公 司在开展天气指数保险时，首先要判断保险实施的 可行性。例如，barnett 和 mahul 7 ，yang44，张跃华 和张琦 12 ，张峭等 45 指出若目标区域空间差异巨大 （气候多变或地势复杂多变），天气指数则不能够 准确衡量农户的作物损失，该地区不适宜引入天气 指数保险。woodard 和 garcia 还认为指数应该反映 的是密切影响特定地区作物产量的天气状况 39 。总 结实施天气指数保险应该至少具备三个特点：一是 气象条件与产量之间有相关性，即天气状况对作物 的产量产生决定性影响；二是天气指数可观察、易 于监测、客观可靠 46 ；三是空间差异性较小，即不 存在极大的区域微气候和复杂多变的地形。天气指 数保险中的基差风险无法完全消除，但可以通过合 理的手段把基差风险控制在可接受的范围 28, 36 。本 文在梳理相关研究的基础上，得出管理基差风险的 技术流程（图 1）： 基差比 − − 历史平均产量损失率 历史天气指数保险平均赔付率 （1） 图 1 基差风险风险管理的技术流程 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 116 6.1. 优化保险产品设计 为降低基差风险，优化天气指数保险设计是一 种事前预防措施 47 ，本文主要从以下几个方面梳理 国内外研究进展。 （1）筛选天气指数 在设计天气指数保险合同时，应首先从各地区历 史气象数据出发，考察分析气象要素和产量损失的 相关性，筛选出与当地农作物减产相关的关键气象 要素。综合已有研究，发现目前国内对天气指数的 确定主要采取如下方法： ①根据调研情况，结合农户的种植经验确定当地 的主要气象灾害。 ②减产量与可能影响农作物减产的因子之间做回 归分析 48-50 。 对于受到多个气象灾害影响的农作物，牛浩，陈 盛伟认为可用隶属度函数（公式 2）来确定主因子 天气指数，函数值越大表明该天气指数风险权重越 大。 minmax min xx xx xf   )( (2) 式中， )(xf 表示隶属度，x 表示天气指数，xmin 表示 天气指数最小值，xmam 表示天气指数最大值。也可 通过数学降维的处理方法将具有一定相关性的多个 指标重新组合成一个新的综合指标 51。③在实验室 中模拟农作物气象受损情况，将处理后的减产量与 天气指数构建相关关系 52。 多数农作物并非只遭受一种气象灾害的影响， 农作物的损失是多种气象灾害共同作用的结果 53, 54 。 若保险产品设计只涵盖一种气象灾害会增大基差风 险。smith 和 watts 论证了仅仅依靠一个指数难以开 发出有效的天气指数保险产品这一观点。他们使用 来自多个领域的文献研究发现降水量与测量站点附 近作物产量的相关关系为 46%-53%，可见在监测点 缺乏的情况下降水量与作物产量的相关性会更低。 同时，他们在不同产量损失程度下进行了降水指数 保险基差风险的模拟分析，结果发现降水量与作物 产量的相关程度不高，存在很高的基差风险 55 。丁 少群，罗婷也认为应该将多个气象因子囊括在产品 设计中，承保不同的天气风险引起的损失。但他们 也担忧多指数保险产品设计难度大，难以精确度量 致灾气象因子与产量损失的关系 13 。产品设计中倘 若覆盖所有天气风险，基差风险会相对较小，但是 也有人认为这种做法有一定的弊端：保险公司的风 险会很大，保险产品的保费也会相对较高，对精算 技术要求会较高，指数的设计会变得更加复杂和难 以理解 56 。 （2）调整作物生长期 年际间气象条件会随全球气候的变化而变化， 因此农户种植作物的时间并不是完全一致，作物的 生长期日期也不是一成不变。天气指数保险的赔付 往往是根据生长期来赔付，在不同年份采用固定的 生长日期设计保险产品会存在时间基差风险。因此， 在优化产品设计时应该根据保险实施的时间调整生 长期。国外针对精细区分作物生长阶段有专门的研 究：conradt 等利用 1980-2009 年哈萨克斯坦北部农 场尺度上的小麦产量数据来研究灵活性天气指数保 险的实施效果。他们使用 gdd（growing degree days）来确定每个保险期间变量开始和结束的日期， 研究证明这种灵活确定天气指数的方法比固定日期 法更能准确描述作物物候生长期的进程，能降低指 数保险中的基差风险 5 。牛浩，陈盛伟认同应该随年 际间气象条件的变化灵活调整天气指数保险中的生 长期时间段 30 。 （3）优化精算模型 天气指数保险产品的核心在于利用历年气象与 作物产量数据准确分析出指数与损失之间的关系， 从而根据该关系制定赔付标准和费率。天气指数和 作物产量之间的相关程度的高低直接影响保险产品 基差风险的大小 20 。目前已有研究主要是从作物单 产时序数据出发，遵循“作物单产－趋势剔除－指 数选取—关系确定—费率厘定”的范式 57, 58。但在各 环节具体模型技术的选择上还未达成一致。 ①产量去趋势模型。单产数据中包含多种影响 作物生长的因素，如技术进步、作物品种、管理技 能和气象条件等。除气象条件外，其他影响作物生 长的因素视为趋势产量，将趋势产量去除是气象产 量分离过程的核心。目前模拟方法主要有 5a 滑动平 均 48 、灰色系统模型逐步滑动平均 59、多元多项式 回归方程 38 ，hp 滤波法 51 、arima 模型 60，直线 滑动平均法 54, 61 。利用滑动平均法模拟作物趋势过 于简单，受异常值影响较大，步长的选择也具有较 大的主观性；arima 模型方法可能存在对作物产量 趋势的拟合过于准确，即拟合优度过高，从而低估 单产波动水平 57 。但目前的文献很少提到优化比较 趋势模型的方法，而仅仅选择一类模型进行产量的 拟合，容易导致较大的基差风险。因此在产量去趋 势的环节，需要进行多种模型方法的比较，进行合 理性检验，选择最符合实际受灾的最优模型。 ②拟合产量与天气变量的关系。产量与气象之 间的关系是确定赔付标准和最终费率的关键。不同 的模型方法的拟合效果有很大的差异，进而影响天 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 116 117 6.1. 优化保险产品设计 为降低基差风险，优化天气指数保险设计是一 种事前预防措施 47 ，本文主要从以下几个方面梳理 国内外研究进展。 （1）筛选天气指数 在设计天气指数保险合同时，应首先从各地区历 史气象数据出发，考察分析气象要素和产量损失的 相关性，筛选出与当地农作物减产相关的关键气象 要素。综合已有研究，发现目前国内对天气指数的 确定主要采取如下方法： ①根据调研情况，结合农户的种植经验确定当地 的主要气象灾害。 ②减产量与可能影响农作物减产的因子之间做回 归分析 48-50 。 对于受到多个气象灾害影响的农作物，牛浩，陈 盛伟认为可用隶属度函数（公式 2）来确定主因子 天气指数，函数值越大表明该天气指数风险权重越 大。 minmax min xx xx xf   )( (2) 式中， )(xf 表示隶属度，x 表示天气指数，xmin 表示 天气指数最小值，xmam 表示天气指数最大值。也可 通过数学降维的处理方法将具有一定相关性的多个 指标重新组合成一个新的综合指标 51。③在实验室 中模拟农作物气象受损情况，将处理后的减产量与 天气指数构建相关关系 52。 多数农作物并非只遭受一种气象灾害的影响， 农作物的损失是多种气象灾害共同作用的结果 53, 54 。 若保险产品设计只涵盖一种气象灾害会增大基差风 险。smith 和 watts 论证了仅仅依靠一个指数难以开 发出有效的天气指数保险产品这一观点。他们使用 来自多个领域的文献研究发现降水量与测量站点附 近作物产量的相关关系为 46%-53%，可见在监测点 缺乏的情况下降水量与作物产量的相关性会更低。 同时，他们在不同产量损失程度下进行了降水指数 保险基差风险的模拟分析，结果发现降水量与作物 产量的相关程度不高，存在很高的基差风险 55 。丁 少群，罗婷也认为应该将多个气象因子囊括在产品 设计中，承保不同的天气风险引起的损失。但他们 也担忧多指数保险产品设计难度大，难以精确度量 致灾气象因子与产量损失的关系 13 。产品设计中倘 若覆盖所有天气风险，基差风险会相对较小，但是 也有人认为这种做法有一定的弊端：保险公司的风 险会很大，保险产品的保费也会相对较高，对精算 技术要求会较高，指数的设计会变得更加复杂和难 以理解 56 。 （2）调整作物生长期 年际间气象条件会随全球气候的变化而变化， 因此农户种植作物的时间并不是完全一致，作物的 生长期日期也不是一成不变。天气指数保险的赔付 往往是根据生长期来赔付，在不同年份采用固定的 生长日期设计保险产品会存在时间基差风险。因此， 在优化产品设计时应该根据保险实施的时间调整生 长期。国外针对精细区分作物生长阶段有专门的研 究：conradt 等利用 1980-2009 年哈萨克斯坦北部农 场尺度上的小麦产量数据来研究灵活性天气指数保 险的实施效果。他们使用 gdd（growing degree days）来确定每个保险期间变量开始和结束的日期， 研究证明这种灵活确定天气指数的方法比固定日期 法更能准确描述作物物候生长期的进程，能降低指 数保险中的基差风险 5 。牛浩，陈盛伟认同应该随年 际间气象条件的变化灵活调整天气指数保险中的生 长期时间段 30 。 （3）优化精算模型 天气指数保险产品的核心在于利用历年气象与 作物产量数据准确分析出指数与损失之间的关系， 从而根据该关系制定赔付标准和费率。天气指数和 作物产量之间的相关程度的高低直接影响保险产品 基差风险的大小 20 。目前已有研究主要是从作物单 产时序数据出发，遵循“作物单产－趋势剔除－指 数选取—关系确定—费率厘定”的范式 57, 58。但在各 环节具体模型技术的选择上还未达成一致。 ①产量去趋势模型。单产数据中包含多种影响 作物生长的因素，如技术进步、作物品种、管理技 能和气象条件等。除气象条件外，其他影响作物生 长的因素视为趋势产量，将趋势产量去除是气象产 量分离过程的核心。目前模拟方法主要有 5a 滑动平 均 48 、灰色系统模型逐步滑动平均 59、多元多项式 回归方程 38 ，hp 滤波法 51 、arima 模型 60，直线 滑动平均法 54, 61 。利用滑动平均法模拟作物趋势过 于简单，受异常值影响较大，步长的选择也具有较 大的主观性；arima 模型方法可能存在对作物产量 趋势的拟合过于准确，即拟合优度过高，从而低估 单产波动水平 57 。但目前的文献很少提到优化比较 趋势模型的方法，而仅仅选择一类模型进行产量的 拟合，容易导致较大的基差风险。因此在产量去趋 势的环节，需要进行多种模型方法的比较，进行合 理性检验，选择最符合实际受灾的最优模型。 ②拟合产量与天气变量的关系。产量与气象之 间的关系是确定赔付标准和最终费率的关键。不同 的模型方法的拟合效果有很大的差异，进而影响天 气指数产品设计中基差风险的大小。评价天气变量 和产量之间的关系，经常性使用的方法是通过一个 回归模型评价产量对天气指数的敏感性，其中隐含 假定是产量和天气指数之间是一种线性依存关系， 每种气象灾害对作物的影响是累计相加影响。采用 线性模型主要是因为拟合方法较为简单，这一假设 过于严格。正如 bokusheva 所说的那样，气象与产 量的相关关系并不一定是线性的，可能存在其他函 数关系 62 ，各天气指数对作物产量损失的影响是否 存在交叉影响（如经过持续干旱后，突降暴雨缓解 旱情，干旱对农作物产量的影响会降低）还有待继 续研究。储小俊，曹杰提出了新的非线性模型方法， 他们通过支持向量机模型拟合天气变量和作物产量 之间的关系，相对于线性回归，支持向量机预测模 型的应用可以提高预测精度，有助于降低基差风险 63 。 ③风险拟合分布的准确性。目前国内外学者对 拟合分布的研究归纳起来大体分为参数法、非参数 法和半参数方法三种。其中，参数方法包括正态分 布 ， 偏 态 分 布 （ 如 beta 分 布 、 gamma 分 布 、 weibull 分布等）；非参数法包括 kernel 核密度模 型和贝叶斯模型，半参数模型实际上是参数模型和 非参数模型的综合 64 。不同的模型对风险拟合的效 果可能不同，有学者针对风险拟合效果做了研究证 实了以上观点。如 thomas 等研究了珠江流域的降水 指数保险，用四种分布 gamma3、gev、广义帕累 托和 wakeby 来拟合降雨量的概率分布函数，对于 192 个气象台站的 1961-2007 年的时间序列数据的分 布函 数 应 用了 三 个 拟合 优度 检 验 （ kolmogorovsmirnov、anderson-darling 和 chi squared），结果 表明最大降水量和 5 天最大降水量最好用 wakeby 分布来描述，但在流域尺度上，gev 是估算珠江流 域降水指数保险最可靠和最稳健的分布 65 。 6.2. 加强空间相关性 空间异质性是基差风险的重要来源。因此，在 设计天气指数保险时需要更多的考虑天气与产量的 潜在时空变化。纵观已有研究，为降低基差风险， 加强保险产品的空间相关性，主要有以下两种方法： 一是缩小保险覆盖范围。如果天气指数保险承 保区域内使用统一费率，则承保范围越小，指数与 保险标的的相关程度就越高，基差风险就越低 19 。 保险产品覆盖的范围过大会导致部分农户的高风险 被区域水平平均化，进而低估农户风险，且空间尺 度越大风险低估程度也越大 57 。因此，在成本可控 的情况下缩小承保区域尺度有利于降低基差风险。 朱俊生就提出与以县为单位相比，以乡镇为单位测 产理赔的方式可以有效降低县域范围内的基差风险 42 。claasse 等也指出相比于农场水平的数据，在县 级水平上说明时空异质性较为困难 66 。clarke 和 mahul 等 37，丁少群等 13，张峭等 45 都认为天气指数 保险的设计应针对更小区域范围内的作物 37 。与传 统农业保险相比，天气指数保险对相关数据要求更 为苛刻，普遍认为需要有 30 年以上的数据才能很好 的模拟指数的概率分布 30, 47 。根据国际要求一个标 准的气象观测站的观测范围为 20 平方公里，大多数 发展中国家因气象站基础设施落后很难达到这个标 准。在我国乡镇一级气象站点的分布密度小，目前 在乡镇一级开展天气指数保险的可能性低。 二是提高数据获取能力。专家学者们一致认为 完整可靠的历史气象和产量数据是降低基差风险的 基础，是成功研发天气指数保险产品的保证 30, 67 。 相对于气象数据，产量数据人为干预因素更大，其 获取及准确性更加难以估计。因此需要采用高科技 技术、数据处理方法等来弥补和精确保险设计所需 的数据。在科技上，通过卫星遥感技术、无人机技 术、计算机仿真等现代高科技技术可以填补天气指 数保险设计数据缺失的缺陷 20, 68 。例如，殷剑敏等 采用高分辨率的 gis 小网格推算技术对南丰市蜜桔 冻害风险作出考量 69 。chantarat 等利用 ndvi 指数 在高空间分辨率下反映植被的数量和活性，将卫星 数据与地面农户层面数据相结合来降低设计基差风 险 70 。由于我国组织化程度低，地块分散，所以在 应用遥感技术时未免会存在地理信息与实际有差距、 精度提高难度大、成本较高的问题。在数据处理方 法上，目前大多采用空间插值法来弥补缺失的气象 数据。鲁振宇等通过反距离加权法、张力样条函数 法和普通克里格法对黄河源区的降水量进行空间插 值 71 。对于整体数据的缺失，可选取最小临近点的 已有数据代替。 6.3. 开展基差风险区划 个体农户与区域产量之间存在着空间差异性， 空间差异性越大，基差风险也就越大。因此，在天 气指数保险实施过程中要合理划分保险区域，针对 损失各不相同的农户制定“异质性”费率，使其缴 纳的保险费与其生产风险水平尽可能匹配，进而降 低基差风险。目前关于风险区划的方法以聚类分析 为主，但是对于风险区划的指标体系看法不一。现 阶段主要有三大风险区划指标体系： （1）综合指标体系。按照区内相似性与区间差 异性的原则使区域内气象条件、作物品种、土壤质 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 118 量、地理特征、生产技能等综合影响因素尽可能的 一致。国外发达国家按照该指标体系对农业保险进 行精细的风险区划，例如，加拿大曼尼托巴省根据 土质、气候、地理和农作物生产历史情况将全省划 分为 16 个风险大区，在每个风险大区中又按照作物 种类、土壤类型进一步细分亚区，亚区内再依据农 户特征作进一步调整，以使农户缴纳的保费与其风 险水平相匹配 16 。庹国柱等 72 提出了作物产量水平、 气候综合评判值等 9 个综合指标，并提供了指标图 重迭法和模糊聚类分析法两种分类方法。王克以县 域内所有农户的平均风险水平厘定县域农业保险平 均费率，然后对所有农户按照土壤环境、生产条件、 产量波动等因素进行聚类分析划分不同的类组，根 据不同类组风险和县域平均风险的对比给予不同群 组风险折算系数 57 。 （2）主导指标体系。主导指标体系主要是定义 一个或几个主要的计量指标来对风险区域进行区划。 邢鹂等将定性和定量分析相结合，选择粮食单产变 异系数、农作物成灾概率、专业化指数和效率指数 4 个主导指标，采用聚类分析法对我国粮食生产地 进行了风险等级划分 73 。陈晓峰将广西糖料蔗区按 照灾害风险分布、生产能力划分为不同的保险区域 单元，设定不同的产量基准来控制基差风险 34 。陈 雅子等设计江苏夏季水稻高温热害天气指数时，根 据风险指标将全省划分为 2 个风险区域，在此基础 上计算出 2 个区域的保险费率。对比发现保险费率 的分布与风险指标的分布相似，得出该天气指数保 险的设计能有效地减小基差风险的结论 74 。梁来存 认为气象、水利、土壤等各种复杂因素对农作物的 影响最终表现在产量的波动上，他从产量的角度建 立包括单产和种植面积两个子系统在内的风险评价 指标体系，利用聚类分析法、判别分析法进行风险 区划并用 probit 模型对结果进行验证 75。 6.4. 实施组合保险 传统农业保险的承保理赔模式可用于分散非系 统性风险，而指数类保险则用于分散系统风险 76 。 实际操作中可以利用组合保险的方式取长补短，即 首先利用天气指数保险分散区域所有农户的系统天 气风险，然后对个体农户的自身异质性风险再通过 传统保险的方式进行承保。梳理已有研究，发现组 合保险主要有附加说明、备用条款、个例补偿三种 形式。 （1）附加说明。美国团体风险计划（group risk plan，grp）的经验或许对天气指数保险合同 的设计有一定的参考意义。skees 等指出基差风险对 于农户来说是一个全新的概念，农户可能无法准确 地理解这些条款。美国 grp 的做法是在保险合同中 对基差风险进行专门的明示，农户需要专门签署一 份文件以表示自己已经了解并承认了区域保险产品 与传统农业保险产品的差异 77 。这种在保险合同中 添加附加说明的做法，只能提升农户对基差风险的 认知，无法真正的降低产品中的基差风险，在实践 中需要将这种做法与实际降低基差风险的做法相结 合。 （2）备用条款。国外有部分研究通过设置备用 条款来降低基差风险，主要包含设置备用指数和双 层触发指数合同两种。carter m 等的研究指出指数 保险设定的触发指数是一个区域的平均产量，一经 赔付就要赔付整个区域损失，保险公司的风险较大， 所以设定的触发水平往往比较低。如果承保的地区 面临严重的基差风险问题，在保险合同中需要设置 备用指数（secondary/backup/audit index）。备用指 数是指设定一个范围，如果超过这一范围或者个体 农户的损失占整个区域的比例超过一定范围，就启 动备用指数。即如果农民遭受了损失，但天气指数 并没有被触发，则实施备用指数。备用指数的保费 可能较高，多用在基差风险高、风险异质性高的区 域。埃塞俄比亚的 ifpri / i4 项目采用了这种方法， 实施效果还有待观察 22 。elabed 等提出了一种新的 思路，他们以平均产量指数为基础设计了一个双层 触发指数保险旨在减少马里棉花种植者的基差风险。 第一份合同是传统的单一触发指数合同，根据被保 人所在地区的平均产量来赔付。第二份是双层触发 （two-scale）指数合同。如果区域产量足够低，村 里的产量也很低，就会赔偿。模拟结果表明双层触 发指数合同相对于传统的单一指数保险合约降低了 基差风险，农户对第二种合同的需求比对传统合同 的需求高出约 40％78。 （3）个例补偿。在国内，朱俊生等在对北京区 域产量指数保险研究时，创造性地提出采取区域产 量保险与传统农业保险相配合的方式 76。王克也认 可这种方式既可以降低基差风险，又可以利用指数 保险的优势降低农业保险经营管理成本，保障大部 分参保农民的损失 57 。同样，牛浩，陈盛伟提出针 对一些小气候频发地区，可以采用“统一赔付 + 个 例补偿”的设计方式进行调整。当按照区域指数统 一赔付后，在部分风险空间差异性较大的地区，农 户遭受损失较大或几乎绝产，保险公司可对其进行 补偿 30 。 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 118 119 量、地理特征、生产技能等综合影响因素尽可能的 一致。国外发达国家按照该指标体系对农业保险进 行精细的风险区划，例如，加拿大曼尼托巴省根据 土质、气候、地理和农作物生产历史情况将全省划 分为 16 个风险大区，在每个风险大区中又按照作物 种类、土壤类型进一步细分亚区，亚区内再依据农 户特征作进一步调整，以使农户缴纳的保费与其风 险水平相匹配 16 。庹国柱等 72 提出了作物产量水平、 气候综合评判值等 9 个综合指标，并提供了指标图 重迭法和模糊聚类分析法两种分类方法。王克以县 域内所有农户的平均风险水平厘定县域农业保险平 均费率，然后对所有农户按照土壤环境、生产条件、 产量波动等因素进行聚类分析划分不同的类组，根 据不同类组风险和县域平均风险的对比给予不同群 组风险折算系数 57 。 （2）主导指标体系。主导指标体系主要是定义 一个或几个主要的计量指标来对风险区域进行区划。 邢鹂等将定性和定量分析相结合，选择粮食单产变 异系数、农作物成灾概率、专业化指数和效率指数 4 个主导指标，采用聚类分析法对我国粮食生产地 进行了风险等级划分 73 。陈晓峰将广西糖料蔗区按 照灾害风险分布、生产能力划分为不同的保险区域 单元，设定不同的产量基准来控制基差风险 34 。陈 雅子等设计江苏夏季水稻高温热害天气指数时，根 据风险指标将全省划分为 2 个风险区域，在此基础 上计算出 2 个区域的保险费率。对比发现保险费率 的分布与风险指标的分布相似，得出该天气指数保 险的设计能有效地减小基差风险的结论 74 。梁来存 认为气象、水利、土壤等各种复杂因素对农作物的 影响最终表现在产量的波动上，他从产量的角度建 立包括单产和种植面积两个子系统在内的风险评价 指标体系，利用聚类分析法、判别分析法进行风险 区划并用 probit 模型对结果进行验证 75。 6.4. 实施组合保险 传统农业保险的承保理赔模式可用于分散非系 统性风险，而指数类保险则用于分散系统风险 76 。 实际操作中可以利用组合保险的方式取长补短，即 首先利用天气指数保险分散区域所有农户的系统天 气风险，然后对个体农户的自身异质性风险再通过 传统保险的方式进行承保。梳理已有研究，发现组 合保险主要有附加说明、备用条款、个例补偿三种 形式。 （1）附加说明。美国团体风险计划（group risk plan，grp）的经验或许对天气指数保险合同 的设计有一定的参考意义。skees 等指出基差风险对 于农户来说是一个全新的概念，农户可能无法准确 地理解这些条款。美国 grp 的做法是在保险合同中 对基差风险进行专门的明示，农户需要专门签署一 份文件以表示自己已经了解并承认了区域保险产品 与传统农业保险产品的差异 77 。这种在保险合同中 添加附加说明的做法，只能提升农户对基差风险的 认知，无法真正的降低产品中的基差风险，在实践 中需要将这种做法与实际降低基差风险的做法相结 合。 （2）备用条款。国外有部分研究通过设置备用 条款来降低基差风险，主要包含设置备用指数和双 层触发指数合同两种。carter m 等的研究指出指数 保险设定的触发指数是一个区域的平均产量，一经 赔付就要赔付整个区域损失，保险公司的风险较大， 所以设定的触发水平往往比较低。如果承保的地区 面临严重的基差风险问题，在保险合同中需要设置 备用指数（secondary/backup/audit index）。备用指 数是指设定一个范围，如果超过这一范围或者个体 农户的损失占整个区域的比例超过一定范围，就启 动备用指数。即如果农民遭受了损失，但天气指数 并没有被触发，则实施备用指数。备用指数的保费 可能较高，多用在基差风险高、风险异质性高的区 域。埃塞俄比亚的 ifpri / i4 项目采用了这种方法， 实施效果还有待观察 22 。elabed 等提出了一种新的 思路，他们以平均产量指数为基础设计了一个双层 触发指数保险旨在减少马里棉花种植者的基差风险。 第一份合同是传统的单一触发指数合同，根据被保 人所在地区的平均产量来赔付。第二份是双层触发 （two-scale）指数合同。如果区域产量足够低，村 里的产量也很低，就会赔偿。模拟结果表明双层触 发指数合同相对于传统的单一指数保险合约降低了 基差风险，农户对第二种合同的需求比对传统合同 的需求高出约 40％78。 （3）个例补偿。在国内，朱俊生等在对北京区 域产量指数保险研究时，创造性地提出采取区域产 量保险与传统农业保险相配合的方式 76。王克也认 可这种方式既可以降低基差风险，又可以利用指数 保险的优势降低农业保险经营管理成本，保障大部 分参保农民的损失 57 。同样，牛浩，陈盛伟提出针 对一些小气候频发地区，可以采用“统一赔付 + 个 例补偿”的设计方式进行调整。当按照区域指数统 一赔付后，在部分风险空间差异性较大的地区，农 户遭受损失较大或几乎绝产，保险公司可对其进行 补偿 30 。 7. 结论 关于基差风险的讨论普遍存在于所有涉及天气 指数保险的文献中，主要围绕基差风险的界定、对 保险需求的影响、形成机理和量化等方面，然后提 出管理基差风险的方法。在基差风险的界定研究方 面，普遍认为基差风险是一种由农户承担的未被保 险的剩余风险，指数和实际损失之间无法完全匹配。 在研究基差风险对保险需求的影响时，一致认为基 差风险会降低天气指数保险的需求，但文献中大多 未使用实际农户水平数据来验证基差风险对保险需 求会产生多大的影响。从基差风险形成机理的角度， 可以将基差风险分为时间、空间和产品基差风险。 在量化基差风险方面，主要是衡量基差风险的程度 和使用替代性指标量化基差风险的大小。在管理基 差风险方面，主要是通过优化保险产品的设计，加 强空间相关性，开展基差风险区划，实施组合保险 这些方法来降低基差风险对天气指数保险的影响， 各种研究中都可以发现这些方法并不是独立的，可 以结合起来共同使用。 总的来说，国内外学者只是在设计天气指数保 险产品时零星提到基差风险，专门系统研究基差风 险的文献较少，对于如何降低基差风险的研究还未 真正展开。但研究天气指数保险的学者们都十分关 注基差风险的问题，一致认为天气指数保险需要落 脚到降低基差风险上，这是天气指数保险未来的研 究方向。因此，本文认为以后的研究至少可以从以 下几个方面展开：（1）基差风险的概念和内在机理 理论研究；（2）通过实证研究基差风险究竟在何种 程度上降低了天气指数保险的需求；（3）采用何种 方法量化基差风险；（4）采用何种方法降低天气指 数保险中的基差风险。 致谢 国家自然科学基金面上项目（41471426），教育部 人文社会科学重点研究基地重大项目《基于多源数 据融合的农业生产风险评估研究》（项目批准号： 17jjd910002）。 参考文献 1.research group of coverage level of china agricultural insurance, research report on the coverage level of china agricultural insurance, (china financial publishing house, beijing, 2017) 中国农业保险保障水平研究课题组,中国农业保险保障 水平研究报告,(中国金融出版社,北京,2017). 2.cao qianjin, solutions to the problems with agriculture insurance: innovation, finance & economics. (03)(2005)155-160. 曹前进,农业保险创新是解决农业保险问题的出路,财经 科学. (03)(2005)155-160. 3.zheng jun, cao cong, review on the researches of crop regional index insurance, journal of shandong institute of business and technology. (01)(2015)106-112. 郑军,曹翀,农作物区域指数保险研究述评,山东工商学院 学报. (01)(2015)106-112. 4.chen xiaofeng , huang lu, experience of drought index insurance in malawi and its inspiration for development of sugarcane insurance in guangxi, journal of regional financial research. (10)(2010)53-56. 陈晓峰,黄路,马拉维干旱指数保险试点经验及其对广西 甘蔗保险发展的启示,区域金融研究. (10)(2010)53-56. 5.sarah conradt and robert finger, et al, flexible weather index-based insurance design, climate risk management. 10(2015)106-117. 6.benjamin collier and jerry skees, et al, weather index insurance and climate change: opportunities and challenges in lower income countries, geneva papers on risk & insurance issues & practice. 34(3)(2009)401-424. 7.barry j. barnett and olivier mahul, weather index insurance for agriculture and rural areas in lowerincome countries, american journal of agricultural economics. 89(5)(2007)1241-1247. 8.p. hazell and j. anderson, et al, the potential for scale and sustainability in weather index insurance for agriculture and rural livelihoods, ifad. (2010). 9.world bank, china: innovations in agricultural insurance (promoting access to agricultural insurance for small farmers), sustainable development, east asia & pacific region finance and private sector development, 2007. 10.zhang huiru, index based insurance contract一innovation in agricujturaj insurance, journal of central university of finance & economics. (11)(2008)49-53. 张惠茹,指数保险合约——农业保险创新探析,中央财经 大学学报. (11)(2008)49-53. 11.chu xiaojun, caojie, review of weather index insurance research, inquiry into economic issues. (12)(2012)135140. 储小俊,曹杰,天气指数保险研究述评,经济问题探索. (12)(2012)135-140. 12.zhang yuehua, zhang qi, literature review of weather index insurance development in africa: farmers' production behavior and risk dispersion mechanism, african studies. 1(12)(2018)134-150. 张跃华,张琦,非洲气象指数保险发展综述——农户生产 行为与风险分散机制,非洲研究. 1(12)(2018)134-150. 13.ding shaoqun, luoting, evaluation and analysis of weather index insurance pilots situation in china, shanghai insurance. (05)(2017)56-61. 丁少群,罗婷,我国天气指数保险试点情况评析,上海保险. (05)(2017)56-61. 14.zhao qiao, wang ke, et al, china agricultural risk management. (china agricultural science and technology press, beijing, 2015) 张峭,王克等,中国农业风险综合管理,中国农业科学技术 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 120 出版社,北京, 2015. 15.syroka joanna and bryla erin, developing index-based insurance for agriculture in developing countries, department of economic and social affairs, united nations, 2007. 16.zhang qiao, wang ke, assessment and regional planning of chinese agricultural natural disaster risks, chinese journal of agricultural resources and regional planning. 32(03)(2011)32-36. 张峭,王克,我国农业自然灾害风险评估与区划,中国农业 资源与区划. 32(03)(2011)32-36. 17.tuo guozhu, zhang qiao, weather index insurance enters "test field" in anhui province, china financial and economic news, 2008. 庹国柱,张峭,气象指数保险进入安徽"试验田”,中国财经 报, 2008. 18.michael r. carter and francisco galarza, et al, underwriting area-based yield insurance to crowd-in credit supply and demand, savings & development. 31(3)(2007)335-362. 19.zhang yuhuan, an exploration of weather index-based insurance for agriculture in the world, chinese rural economy. (12)(2017)81-92. 张玉环,国外农业天气指数保险探索,中国农村经济. (12)(2017)81-92. 20.chen kecun, chen shengwei. research on the development of agricultural index insurance in china, contemporary economics. (32)(2015)40-43. 陈克存,陈盛伟,我国农业保险指数化发展研究,当代经济. (32)(2015)40-43. 21.niu hao, chen shengwei, research summary of the agricultural meteorological index insurance, journal of shandong agricultural university (social science edition). 18(03)(2016)69-74+132. 牛浩,陈盛伟,农业气象指数保险研究综述,山东农业大学 学报(社会科学版). 18(03)(2016)69-74+132. 22.carter m and de janvry a, et al, index-based weather insurance for developing countries: a review of evidence and a set of propositions for up-scaling, ferdi, afd, 2014. 23.ghada elabed and michael carter, compound-risk aversion and the demand for microinsurance: evidence from a wtp experiment in mali, university of california, 2014. 24.mobarak and ahmed mushfiq, et al, selling formal insurance to the informally insured, social science electronic publishing. (2012). 25.ruth vargas hill and john hoddinott, et al, adoption of weather-index insurance: learning from willingness to pay among a panel of households in rural ethiopia, agricultural economics. 44(4-5)(2013)385-398. 26.nathaniel d. jensen and andrew g. mude, et al, how basis risk and spatiotemporal adverse selection influence demand for index insurance: evidence from northern kenya, food policy. 74(2018)172-198. 27.wang zhenjun, analysis on the scenarios of farmers' willingness to purchase agricultural insurance—a questionnaire survey on 524 farmers in the regions of the loess plateau in eastern gansu province, journal of lanzhou university (social sciences). 42(02)(2014)132138. 王振军,不同保险方式下农户购买农业保险的意愿分析 ——陇东黄土高原区524户农户问卷调查,兰州大学学报 (社会科学版). 42(02)(2014)132-138. 28.futoshi okada, sustainable growth in crop natural disaster insurance: experiences of japan, in fftc-rda international seminar on implementing and improving crop natural disaster insurance program, 2016. 29.j. diaz nieto, a system of drought insurance for poverty alleviation in rural areas, epilepsia journal of the international league against epilepsy. 39(2)(2006)213– 231. 30.niu hao, chen shengwei, developing situation, current difficulty and solutions &suggestions of china's agricultural meteorological index insurance products, forum on science and technology in chian. (07)(2015)130-135. 牛浩,陈盛伟,中国农业气象指数保险产品的发展现状、 面临难题及解决建议,中国科技论坛. (07)(2015)130-135. 31.xavier giné and robert townsend, et al, statistical analysis of rainfall insurance payouts in southern india, american journal of agricultural economics. 89(5)(2007)1248-1254. 32.kolli n. rao, index based crop insurance, agriculture & agricultural science procedia. 1(2010)193-203. 33.yindong, study on weather index agricultural insurance and its technical issues, xiandai nongye keji.(06)(2014)330-332+335. 尹东,气象指数农业保险及其技术问题探讨,现代农业科 技. (06)(2014)330-332+335. 34.chen xiaofeng, crop regional production insurance: international practice and applicability analysis, journal of financial development research. (02)(2014)9-16. 陈晓峰,农作物区域产量保险:国际实践及适用性分析, 金融发展研究. (02)(2014)9-16. 35.alan fuchs and hendrik wolff, concept and unintended consequences of weather index insurance: the case of mexico, american journal of agricultural economics. 93(93)(2011)505-511. 36.xu lei, zhang qiao, assessment approach for agricultural catastrophic risk in china, scientia agricultura sinica. 44(09)(2011)1945-1952. 徐磊,张峭,中国农业巨灾风险评估方法研究,中国农业科 学. 44(09)(2011)1945-1952. 37.daniel j. clarke and olivier mahul, et al, weather based crop insurance in india, policy research working paper. (2012)1-31. 38. yang taiming, liu buchun, et.al, design and application of the weather indices of winter wheat planting insurance in anhui province, chinese journal of agrometeorology. 34(02)(2013)229-235. 杨太明,刘布春,等,安徽省冬小麦种植保险天气指数设计 与应用,中国农业气象. 34(02)(2013)229-235. 39.joshua d. woodard and philip garcia, basis risk and journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 120 121 出版社,北京, 2015. 15.syroka joanna and bryla erin, developing index-based insurance for agriculture in developing countries, department of economic and social affairs, united nations, 2007. 16.zhang qiao, wang ke, assessment and regional planning of chinese agricultural natural disaster risks, chinese journal of agricultural resources and regional planning. 32(03)(2011)32-36. 张峭,王克,我国农业自然灾害风险评估与区划,中国农业 资源与区划. 32(03)(2011)32-36. 17.tuo guozhu, zhang qiao, weather index insurance enters "test field" in anhui province, china financial and economic news, 2008. 庹国柱,张峭,气象指数保险进入安徽"试验田”,中国财经 报, 2008. 18.michael r. carter and francisco galarza, et al, underwriting area-based yield insurance to crowd-in credit supply and demand, savings & development. 31(3)(2007)335-362. 19.zhang yuhuan, an exploration of weather index-based insurance for agriculture in the world, chinese rural economy. (12)(2017)81-92. 张玉环,国外农业天气指数保险探索,中国农村经济. (12)(2017)81-92. 20.chen kecun, chen shengwei. research on the development of agricultural index insurance in china, contemporary economics. (32)(2015)40-43. 陈克存,陈盛伟,我国农业保险指数化发展研究,当代经济. (32)(2015)40-43. 21.niu hao, chen shengwei, research summary of the agricultural meteorological index insurance, journal of shandong agricultural university (social science edition). 18(03)(2016)69-74+132. 牛浩,陈盛伟,农业气象指数保险研究综述,山东农业大学 学报(社会科学版). 18(03)(2016)69-74+132. 22.carter m and de janvry a, et al, index-based weather insurance for developing countries: a review of evidence and a set of propositions for up-scaling, ferdi, afd, 2014. 23.ghada elabed and michael carter, compound-risk aversion and the demand for microinsurance: evidence from a wtp experiment in mali, university of california, 2014. 24.mobarak and ahmed mushfiq, et al, selling formal insurance to the informally insured, social science electronic publishing. (2012). 25.ruth vargas hill and john hoddinott, et al, adoption of weather-index insurance: learning from willingness to pay among a panel of households in rural ethiopia, agricultural economics. 44(4-5)(2013)385-398. 26.nathaniel d. jensen and andrew g. mude, et al, how basis risk and spatiotemporal adverse selection influence demand for index insurance: evidence from northern kenya, food policy. 74(2018)172-198. 27.wang zhenjun, analysis on the scenarios of farmers' willingness to purchase agricultural insurance—a questionnaire survey on 524 farmers in the regions of the loess plateau in eastern gansu province, journal of lanzhou university (social sciences). 42(02)(2014)132138. 王振军,不同保险方式下农户购买农业保险的意愿分析 ——陇东黄土高原区524户农户问卷调查,兰州大学学报 (社会科学版). 42(02)(2014)132-138. 28.futoshi okada, sustainable growth in crop natural disaster insurance: experiences of japan, in fftc-rda international seminar on implementing and improving crop natural disaster insurance program, 2016. 29.j. diaz nieto, a system of drought insurance for poverty alleviation in rural areas, epilepsia journal of the international league against epilepsy. 39(2)(2006)213– 231. 30.niu hao, chen shengwei, developing situation, current difficulty and solutions &suggestions of china's agricultural meteorological index insurance products, forum on science and technology in chian. (07)(2015)130-135. 牛浩,陈盛伟,中国农业气象指数保险产品的发展现状、 面临难题及解决建议,中国科技论坛. (07)(2015)130-135. 31.xavier giné and robert townsend, et al, statistical analysis of rainfall insurance payouts in southern india, american journal of agricultural economics. 89(5)(2007)1248-1254. 32.kolli n. rao, index based crop insurance, agriculture & agricultural science procedia. 1(2010)193-203. 33.yindong, study on weather index agricultural insurance and its technical issues, xiandai nongye keji.(06)(2014)330-332+335. 尹东,气象指数农业保险及其技术问题探讨,现代农业科 技. (06)(2014)330-332+335. 34.chen xiaofeng, crop regional production insurance: international practice and applicability analysis, journal of financial development research. (02)(2014)9-16. 陈晓峰,农作物区域产量保险:国际实践及适用性分析, 金融发展研究. (02)(2014)9-16. 35.alan fuchs and hendrik wolff, concept and unintended consequences of weather index insurance: the case of mexico, american journal of agricultural economics. 93(93)(2011)505-511. 36.xu lei, zhang qiao, assessment approach for agricultural catastrophic risk in china, scientia agricultura sinica. 44(09)(2011)1945-1952. 徐磊,张峭,中国农业巨灾风险评估方法研究,中国农业科 学. 44(09)(2011)1945-1952. 37.daniel j. clarke and olivier mahul, et al, weather based crop insurance in india, policy research working paper. (2012)1-31. 38. yang taiming, liu buchun, et.al, design and application of the weather indices of winter wheat planting insurance in anhui province, chinese journal of agrometeorology. 34(02)(2013)229-235. 杨太明,刘布春,等,安徽省冬小麦种植保险天气指数设计 与应用,中国农业气象. 34(02)(2013)229-235. 39.joshua d. woodard and philip garcia, basis risk and weather hedging effectiveness, agricultural finance review. 68(1)(2007)99-117. 40.holly wang and hao zhang, on the possibility of a private crop insurance market: a spatial statistics approach, journal of risk & insurance. 70(1)(2003)111124. 41.martin odening and oliver musshoff, et al, analysis of rainfall derivatives using daily precipitation models: opportunities and pitfalls, agricultural finance review. 67(1)(2007)135-156. 42.zhu junsheng, tuo guozhu, innovation in agricultural insurance business requires policy support, china's insurance quote,2016. 朱俊生,庹国柱,农业保险经营模式创新需要相应政策支 持,中国保险报, 2016. 43.michael t. norton and calum turvey, et al, quantifying spatial basis risk for weather index insurance, the journal of risk finance. 14(1)(2012)20-34. 44.yong yang, weather index derivatives in risk transfer for agricultural natural hazards, agriculture & agricultural science procedia. 1(2010)100-105. 45.zhang qiao, li yue, et al, development, application and suggestion of agricultural index insurance, rural finance ressarch. (06)(2018)14-20. 张峭,李越,等,农业指数保险的发展、应用与建议,农村 金融研究. (06)(2018)14-20. 46.huang yingjun, pu yuecheng, review on the challenges facing the weather index insurance and their countermeasures, journal of regional financial research. (08)(2016)4-10. 黄英君,蒲玥成,天气指数保险面临的挑战及其应对策略 研究述评—— 基于印度等地的经验 ,区域金融研究. (08)(2016)4-10. 47.k. warner and k. yuzva, et al, innovative insurance solutions for climate change: how to integrate climate risk insurance into a comprehensive climate risk management approach, united nations university institute for environment and human security (unuehs), 2013. 48.lou weiping, yao yiping, et al, design of products for rice agro-meteorological index insurance: a case in zhejiang province, scientia agricultura sinica. 43(23)(2010)4942-4950. 吴利红,娄伟平,等,水稻农业气象指数保险产品设计—— 以浙江省为例,中国农业科学. 43(23)(2010)4942-4950. 49.lu ping, design of weather index crop insurance contract for northeast china, master student thesis, beijing: tsinghua university,(2010) 路平,东北地区粮食作物气象指数农业保险合同设计,清 华大学学位论文,清华大学, 2010. 50.wang ren, deng chao, et al, weather index insurance design for rice based on panel data of 14 cities in hunan province, seeker. (01)(2015)69-73. 王韧,邓超,等,基于湖南省14地市面板数据的水稻气象指 数保险设计,求索. (01)(2015)69-73. 51.niu hao, chen shengwei, design and research of corn storm and rainfall index insurance products——taking ningyang county of shandong province as an example, journal of agrotechnical. (12)(2015)99-108. 牛浩,陈盛伟,玉米风雨倒伏指数保险产品设计研究—— 以山东省宁阳县为例,农业技术经济. (12)(2015)99-108. 52.sun qing, yang zaiqiang, et al, estimation of premium rates of high temperature disaster for early rice in jiangxi, chinese journal of agrometeorology. 35(05)(2014)561-566. 孙擎,杨再强,等,江西早稻高温逼熟气象灾害指数保险费 率的厘定,中国农业气象. 35(05)(2014)561-566. 53.guo shujun. the meteorological disaster risk assessment based on the diffusion mechanism[a]. journal of risk analysis and crisis response[c]. atlantis press, 2012. 54.zhao sijian, zhang qiao, et al, weather related diaster risk recognizing and assessment of crop, journal of catastrophology. (02)(2018)51-57. 赵思健,张峭,等,农作物气象灾害风险识别与评估研究, 灾害学. (02)(2018)51-57. 55.vincent h. smith and myles watts, index based agricultural insurance in developing countries: feasibility, scalability and sustainability, gates foundation. (2009). 56.wu shan, study on the weather index insurance and its pricing apporoach, master thesis, sichuan: southwestern university of finance and economics,(2014). 吴姗,天气指数保险及其定价研究,西南财经大学学位论 文, 西南财经大学, 2014. 57.wang ke, effect evaluation of chinese crop insurance and its improvement, phd thesis, beijing: chinese academy of agricultural sciences, (2014) 王克,中国农作物保险效果评估及相关政策改善研究,中 国农业科学院学位论文,中国农业科学院, 2014. 58.zhang qiao, china's crop production risk assessment and zoning theory and practice, (china agricultural science and technology press, beijing,2013). 张峭,中国农作物生产风险评估及区划理论与实践,(中 国农业科学技术出版社,北京, 2013). 59.ren yifang, zhao yanxia, et al, winter wheat drought disaster insurance risk assessment and regionalization in henan province, journal of applied meteorological science. 22(05)(2011)537-548. 任义方,赵艳霞,等,河南省冬小麦干旱保险风险评估与区 划,应用气象学报. 22(05)(2011)537-548. 60.li wenfang, rating area yield crop insurance by nonparameter kernel density model, ecological economy. (04)(2012)61-64. 李文芳,基于非参数核密度法的农作物区域产量保险费 率厘定研究,生态经济. (04)(2012)61-64. 61.wang ke, zhang qiao, a new approach to assess crop yield risk based on mixed source of data, scientia agricultura sinica. 46(05)(2013)1054-1060. 王克,张峭,基于数据融合的农作物生产风险评估新方法, 中国农业科学. 46(05)(2013)1054-1060. 62.raushan bokusheva, measuring dependence in joint distributions of yield and weather variables, agricultural finance review. 71(1)(2011)120-141. 63.chu xiaojun, caojie, research on agricultural weather journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 111-122 122 risk management and products pricing, statistics & information forum. (05)(2015)99-104. 储小俊,曹杰,农业天气风险管理及产品定价研究,统计与 信息论坛. (05)(2015)99-104. 64.wang ke, a study on the influence of flexible crop yield distributions on crop insurance premium rate, master thesis, beijing: chinese academy of agricultural sciences, (2008) 王克,农作物单产分布对农业保险费率厘定的影响,中国 农业科学院学位论文,中国农业科学院, 2008. 65.fischer thomas and su buda.et al, probability distribution of precipitation extremes for weather index-based insurance in the zhujiang river basin, south china, journal of hydrometeorology. 13(3)(2012)1023-1037. 66.roger claassen and richard e. just, heterogeneity and distributional form of farm-level yields, american journal of agricultural economics. 93(1)(2011)144-160. 67.yao qinghai, meteorological disaster and weather index insurance research, shanghai insurance. (01)(2015)711+20. 姚庆海 , 气象灾害与天气指数保险研究 , 上海保险 . (01)(2015)7-11+20. 68.hou maozhang, wumin, the review of weather-index based insurance, journal of central south university of forestry & technology (social sciences). 9(06)(2015)3742. 侯茂章,吴敏,天气指数保险研究进展,中南林业科技大学 学报(社会科学版). 9(06)(2015)37-42. 69.yin jianmin, miao qilong, et al, climate index and risk assessment of frozen injury of nanfeng tangerine(citrus reticulalta), chinese journal of agmneteorology. (04)(2008)507-510. 殷剑敏,缪启龙,等,南丰蜜桔冻害的气候指标及风险评估, 中国农业气象. (04)(2008)507-510. 70.sommarat chantarat and andrew g. mude, et al, designing index based livestock insurance for managing asset risk in northern kenya, journal of risk and insurance. 80(1)(2013)205–237. 71. application of the spatial interpolation of rainfall—a case study of the headstream region of the yellow river, journal of lanzhou university (natural sciences). (04)(2006)11-14. 鲁振宇,杨太保,等,降水空间插值方法应用研究——以黄 河源区为例, 兰州大学学报. (04)(2006)11-14. 72.tuo guozhu, dingshaoqun, discussion on risk division and rate division of crop insurance, chinese rural economy. (08)(1994)43-47+61. 庹国柱,丁少群,农作物保险风险分区和费率分区问题的 探讨,中国农村经济. (08)(1994)43-47+61. 73.xing peng, zhong funing, research on grain production and risk zoning, journal of agrotechnical. (01)(2006)1923. 邢鹂,钟甫宁,粮食生产与风险区划研究,农业技术经济. (01)(2006)19-23. 74.chen yazi, shen shuanghe, high temperature and heat damage weather index insurance for rice in jiangsu province, jiangsu agricultural sciences. (10)(2016)461464. 陈雅子,申双和,江苏省水稻高温热害保险的天气指数研 制,江苏农业科学. (10)(2016)461-464. 75.liang laicun, china's grain crop insurance risk zoning and test of its probit ordered choice model, economic survey. (06)(2009)68-72. 梁来存,我国粮食作物的保险风险区划及其probit排序选 择模型验证,经济经纬. (06)(2009)68-72. 76.zhu junsheng, zhaole, et al, area-based crop yield insurance in beijing, insurance studies. (02)(2013)76-86. 朱俊生,赵乐,等,北京市农业区域产量保险研究,保险研 究. (02)(2013)76-86. 77.jerry r. skees and j. roy black, et al, designing and rating an area yield crop insurance contract, american journal of agricultural economics. 79(2)(1997)430-438. 78.ghada elabed and marc f. bellemare, et al, managing basis risk with multiscale index insurance, agricultural economics. 44(4-5)(2013)419-431. international journal of computational intelligence systems vol. xx(z); month (year), pp. xx–yy doi: 10.1080/xxxxxxxxxxxxxx; issn xxxx–xxxx online https://www.atlantis-press.com/journals/ijndc journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 128–133 doi: https://doi.org/10.2991/jracr.k.191024.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article a critique of pandemic catastrophe modeling daniel j. rozell* department of technology and society, 1432 computer science, stony brook university, stony brook, ny 11794-3760, usa 1. introduction over the past three decades, catastrophe modeling has become an essential tool of the insurance industry for estimating potential financial losses from disasters such as earthquakes and hurricanes [1]. the basic premise of catastrophe modeling is to combine stochastic hazard models with asset models to assess financial risk exposure from large events. while the original intent was to assess systemic risk, the influence of catastrophe models eventually expanded into related practical tasks—setting rates, allocating capital reserves, selecting markets, etc. considering the substantial variation in the nature of hazards, one would also expect considerable variation in the utility of catastrophe models depending on the availability of data and science. that is, more common events should generally yield more reliable projections. this requires a skeptical eye when evaluating new models, yet catastrophe models have become widely adopted and viewed as established (and therefore implicitly reliable) technology. it is normal to want to replicate success in one area to new situations. but the temptation to apply existing catastrophe modeling techniques to more uncertain events has gotten ahead of methodological development. the result is inevitable disappointment. a prime example is pandemic catastrophe modeling where the hazard is a pandemic-inducing pathogen that results in substantial life and health insurance losses. while the insurance industry should be commended for tackling such an important topic, the uncertainties are daunting. the difficulty of quantitative pandemic risk assessment is summarized by harvey fineberg, the former president of the u.s. national academy of medicine, “major flu pandemics arise on average only about three times every century, which means scientists can make relatively few direct observations in each lifetime and have a long time to think about each observation. that is a circumstance that is ripe for over-interpretation” [2]. major limitations of hurricane catastrophe modeling—primarily in the form of hidden subjectivity—have already been well described [3]. many of these limitations are generalizable to similar catastrophe models: flooding, earthquakes, etc., yet there is a new class of models with exceptional levels of uncertainty that require their own discussion. this critique explores the additional limitations unique to pandemic catastrophe modeling that further limit the value of these types of models for estimating financial exposure, disaster response, or social policymaking. 2. lessons from ebola while pandemic catastrophe modeling has appropriately focused on avian influenza—the most likely cause of a pandemic—the current state of pandemic risk assessment can be evaluated using another recent deadly outbreak—the ebola virus disease epidemic that emerged in west africa in late 2013 [4]. much like avian influenza, ebola virus disease has zoonotic origins, but in this case the primary natural reservoir is bats rather than birds. in addition to the much higher mortality rate, ebola is much less transmissible than influenza. these combined factors tend to limit outbreaks and the pandemic potential of ebola is much lower than influenza [5]. despite its more modest growth characteristics, ebola outbreaks provide some insight into the current state of pandemic forecast modeling. during the ebola outbreak, multiple epidemiological models (table 1) that made short-range predictions in 2014 considerably overestimated its severity— some by an order of magnitude. unless the world health organization (who) underreported cases (which would be a separate serious problem in itself ), the a r t i c l e i n f o article history received 12 august 2019 accepted 30 august 2019 keywords catastrophe modeling pandemics uncertainty insurance risk assessment a b s t r a c t catastrophe modeling is a popular risk assessment tool for the insurance industry and has been applied to a variety of natural disaster events. more recently, catastrophe modeling techniques have been extended to events, such as pandemics, where the range of possible scenarios is less understood due to the complexity of the hazard and the dependence of event magnitude to human response. some general limitations of catastrophe modeling are discussed in the context of pandemics—such as the failure to distinguish natural variability from incertitude and the difficulty of ensuring a representative model—along with recommendations for minimizing surprises. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: daniel.rozell@stonybrook.edu https://doi.org/10.2991/jracr.k.191024.002 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ daniel.rozell%40stonybrook.edu d.j. rozell / journal of risk analysis and crisis response 9(3) 128–133 129 general inaccuracy of these predictions suggests that the progression of an epidemic is still difficult to model. pandemics move quickly and the early stages are sensitive to assumptions regarding public health interventions. likewise, failure to understand all the transmission routes (e.g., sexual transmission of ebola by recovered patients) can make the end stages of a pandemic equally difficult to predict [10,11]. and it is not that the ebola epidemic modeling was obviously too alarmist in overestimating the severity of the outbreak. for example, in july 2014, a liberian man infected with ebola flew to lagos, nigeria. if that case had not been quickly detected and contained, the epidemic could have entered a metropolitan area of 21 million people. had that happened, the who mortality projections most likely would have been grossly underestimated. the difficulty is that these model estimates are strongly dependent upon many smaller events within the larger epidemic. one potential solution is to move toward agent-based modeling—computer simulations that estimate the aggregated consequences of the independent behavior of many individuals [12]. however, this approach is still very resource intensive, complex to build, and challenging to calibrate [13]. in addition, the translation of ebola transmission risk studies performed in less-industrialized nations to more-industrialized nations presents a considerable challenge [14]. the response to the ebola outbreak among the most industrialized nations was not uniform and there was even a noticeable difference in how scientists and public officials estimated risk. because the transmissibility of ebola is relatively low, medical experts recommended against the quarantine of u.s. medical workers returning from west africa. however, the cost and effort of tracking and monitoring potential exposures led public officials to lean toward precautionary quarantines that varied from state-to-state. estimating the human response is challenging and averaged data tell a very incomplete story. any useful pandemic risk model would need to account for local and regional conditions that affect an outbreak’s progression. the financial effects of a pandemic may be even more difficult to model. for example, the who’s estimate of the cost of controlling the west african ebola outbreak jumped from less than 5 million usd$ in april 2014 to almost 1 billion usd$ in september 2014 [4,15]—an increase by a factor of 200 within 6 months. this demonstrates how difficult it can be to estimate the cost of a potential outbreak when it is not known at what stage the outbreak will be discovered or contained. the secondary economic impacts of a pandemic would also be expected to change rapidly and would need to include a range of extraneous factors that would only increase the uncertainty. one might assume that after all the analysis of the west africa ebola outbreak [16], at least ebola would now be easier to model. yet the ongoing ebola outbreak in the democratic republic of congo (drc) has been extremely difficult to predict. initial opinion on the outbreak was cautious [17], but optimistic because the previous outbreak had made the world realize that ebola was not just a self-limiting african disease, but had pandemic potential. thus, the drc’s prompt call for assistance and the faster who response—along with a new highly effective vaccine—all suggested rapid containment of the outbreak [18]. however, the outbreak has continued longer than expected due to multiple factors: persistent regional conflict [19], rampant disinformation [20], widespread use of inadequate health clinics [21], and even the limited availability of diagnostic tests [22]. the important lesson from this discussion of ebola is that there is an extra layer of complexity inherent to pandemic modeling that is missing from most other natural disasters. the severity of damage from an earthquake is dependent on infrastructure built before the event, but the magnitude of the event itself is essentially independent of human activity. however, the magnitude of a pandemic is also dependent on the human response after the beginning of the disaster. political inaction or inadequate public health systems will not make an earthquake larger, but can allow a small local outbreak to spread into a pandemic. likewise, changing individual behaviors, such as increased social distancing or just more frequent hand-washing can substantially influence the progression of an infectious disease outbreak [23]. because social conditions vary widely by region and can change unpredictably, a useful pandemic catastrophe model is much more difficult to construct than other natural hazard models. a realistic accounting of uncertainty in a pandemic risk model should show that our current incertitude is so large as to preclude standard probabilistic quantitative decision-making. 3. underlying problems with catastrophe models so, if we are asking too much of pandemic catastrophe modeling, what is wrong with the models and how do we fix them? the simple answer is that overconfidence in models is the underlying problem, but this overconfidence has some specific causes. the following outlines two of these problems and proposes mitigating actions. 3.1. the mistreatment of uncertainty it is difficult to specifically critique commercially available pandemic catastrophe models because they are proprietary. however, the documentation for several hurricane catastrophe models is available for public inspection as a requirement of the florida commission on hurricane loss projection methodology. table 1 | accuracy of ebola virus disease predictions made in late 2014 model timeframe (weeks) prediction actual (who data) total cases in liberia 10 9400–47,000 [6] <7000 daily cases in liberia 15 249–545 [7] <50 reported cases in liberia 12 20,471–94,143 [8] 7819 total confirmed and probable cases 6 5740 in guinea 1820 in guinea 9890 in liberia 4240 in liberia 5000 in sierra leone [9] 4602 in sierra leone 130 d.j. rozell / journal of risk analysis and crisis response 9(3) 128–133 given the basic similarities of catastrophe models, some general remarks can be made regarding their attempts to characterize uncertainty by aggregating scenarios through a large number of stochastic simulations. in these cases, the numbers of simulations or aggregation methods are not the issue, but the method of representing uncertainty. there is no current consensus on the best way to represent uncertainty or when to use one form over another [24]. furthermore, there is no consensus on the number of forms of uncertainty. uncertainty has been categorized across a broad spectrum that ranges from certainty to nescience (unresolvable ignorance) [25]. bayesians [26,27] generally argue that there is only one type of uncertainty—a measure of belief irrespective of its source—and that probability is the best way to express it [28]. at the other extreme, complex uncertainty typologies have been created that distinguish it by location, nature, and level with multiple subtypes [29,30]. the utility of this level of detail is debatable when there is no method for treating each distinct form of uncertainty [31,32]. while uncertainty representation is an unsettled matter, distinguishing between incertitude (i.e., lack of knowledge) and natural variability does have known advantages over a single treatment methodology. for example, when an analyst has no knowledge of the value of a parameter within a bounded range, it is common to represent that interval using a uniform distribution (a frequently used “uninformed prior” in bayesian inference). however, treating interval data as uniform distributions requires an equiprobability assumption—an idea that traces back to bernoulli’s and laplace’s ‘principle of insufficient reason’ and more recently critiqued by keynes under the name ‘principle of indifference’ [33]. under this assumption, each possible value is considered to be equally likely and thus the interval can be represented by the interval’s midpoint—the mean and median of a uniform distribution. while the uniform distribution approach is computationally expedient and easy to understand, it can also disguise uncertainty. one result of the central limit theory is that any distribution with finite variance, such as a uniform distribution, will converge to a normal distribution approximation when repeatedly convolved [34]. as shown in figure 1, adding two uniform distributions results in a triangle distribution that would appear to have more certainty than the original uniform distributions. as the number of distributions increases, the resulting normal approximation does spread, but the assumed central tendency is maintained. by comparison, the addition of two intervals of [0, 1] by interval arithmetic yields the even wider interval [0, 2]. repeating the process yields ever wider and more uncertain intervals without an artifact of precision. the practical implication of this phenomenon is that standard bayesian and monte carlo techniques have an important limitation. these approaches will yield valid results when the model inputs represent normally distributed natural variability. however, when the uncertainty due to lack of knowledge is large compared to the natural variability, the uncertainty in the model output will be under-represented. unfortunately, this situation frequently occurs in rare natural disasters, such as pandemics. in practice, uncertainty representation may be based on extraneous factors such as familiarity, academic tradition, or lack of knowledge of alternatives. catastrophe modelers have been warned in the past to avoid becoming enamored with extreme precision that may not have a basis in reality [35], but they need further guidance. fortunately, there has been considerable work in the area of intervalized probabilities [36,37], second-order monte carlo techniques, and robust bayesian or bayesian sensitivity analysis [38,39] which allows for the recognition of incertitude within bayesian inference [36]. models that more explicitly account for incertitude can sometimes yield results that seem disappointingly vague, but an honest portrayal of uncertainty is always informative. however, catastrophe model users typically come from the world of financial modeling where data are abundant and historical prices are precisely known. it is not surprising that they prefer data analysis methods that are not optimal for addressing scenarios with sparse data and large incertitude. adopting new methods may be a difficult transition despite their utility. 3.2. failures of imagination in a workshop on the west africa ebola epidemic, virologist daniel bausch noted how prior relatively small outbreaks influenced expectations, “i think if you’d asked [ebola virus experts] … a year and a half ago ‘are we going to have 25,000 cases of ebola in west africa?,’ most of us would have not said that was a likelihood” [4]. it is common for modelers to claim that their models have been figure 1 | convolution of n uniform distributions with bounds [0, 1]. d.j. rozell / journal of risk analysis and crisis response 9(3) 128–133 131 “validated” because they can replicate past data. however, this is history matching, not validation [40,41]. acknowledging this distinction creates less surprise if we encounter extreme events not previously experienced or we discover that a model contains non-stationary processes that invalidate a model for future use. in terms of pandemic catastrophe models, diseases can behave unpredictably due to evolving pathogens or changing demographic conditions. for example, the 2017 outbreak of plague in madagascar—normally a relatively controllable bacterial disease— was exacerbated by its spread to growing urban centers. the user’s trust in any catastrophe model depends on assurances of validation (i.e., history matching) which may be acceptable in the case of relatively stable processes, such as earthquakes, but more tentative for pandemic catastrophe models that depend on non stationary processes [42]. after catastrophic events, modelers have an opportunity to check their model against actual losses, so it’s not surprising that catastrophe model outputs will change after these recalibrating updates. however, it can be rather difficult to detect changes in the probabilities of extreme events caused by non-stationary processes [43]. one practical output of a catastrophe model is an exceedance probability curve, which summarizes the annual probability of exceeding particular losses—most importantly, a loss that would lead to the insolvency of an insurer. to generate these curves, catastrophe models will typically simulate many potential scenarios that include both historical data as well as scenarios that attempt to address limitations in the representativeness of the historical data set. this is especially necessary for models that include known non-stationary processes such as sea level rise, changing building technology, increased coastal development, etc. however, the ebola epidemic serves as an example of a previously unforeseen risk that would not be captured merely by modestly expanding the range of a catastrophe model’s simulations. another lesson from the ebola epidemic is that the various outbreaks within the larger epidemic had very different transmission dynamics depending on the level of cooperation and trust between the international public health workers, government officials, and affected communities. the impact of these social factors is critical to any risk assessment and difficult to characterize. furthermore, new factors and previously unknown relationships in epidemic evolution are emerging as large epidemiological data sets are analyzed [44]. it would seem that a truly representative catastrophe model of a real-world epidemic would tend towards substantial complexity. one thoughtfully constructed pandemic catastrophe model [45–47] considers a myriad of factors including: wild and domestic viral reservoirs; the virulence and transmissibility of potential pandemic pathogens; the age and density distribution of populations; seasonal impacts; social factors such as air travel, work commutes and travel restrictions; and the availability of quality health care, antiviral medications, and vaccines. whether these are enough factors is debatable. estimating the insurance losses from a hurricane is fairly well-known and straightforward by comparison, yet building construction and risk mitigation details frequently ignored in standard catastrophe models can have a substantial impact on risk exposure estimates [48]. furthermore, the impact, timing, and magnitude of each factor must be estimated along with any correlation with other factors. in the end, we are left with the question of whether to pursue a more detailed pandemic catastrophe model that might be unusably complex or a simpler model that may be missing critical factors. there is a common, but often untested, assumption that complex problems call for complex solutions and that it is always better to use complex methods that make use of all available information [49]. however, there is a history of studies showing that very simple predictive models can often have better performance than complex models [50,51]. the question of best modeling approach has been explored in recent years in the epidemiological community in a series of infectious disease forecasting challenges [52]. for example, in 2015, the research and policy for infectious disease dynamics (rapidd) ebola forecasting challenge compared the predictive abilities of eight ebola epidemiological models against a synthetic set of data over four scenarios at multiple points in an outbreak [53]. models varied in complexity and ranged from 2 to 27 input parameters. the results did not point toward an “ideal” level of complexity because one of the best individual performers was a two-parameter stochastic model, while the worst performer was a two-parameter logistic growth model. the overall best performer was a bayesian average of all the models which suggests that using multiple independent models is the current best approach. however, the last challenge scenario of an uncontrolled ebola outbreak with noisy data was poorly predicted by all models, so even multi-model averaging has its limitations. in a similar forecasting challenge for influenza in the u.s. over seven flu seasons, almost three-quarters of the 22 models evaluated made better seasonal predictions than the historical base-line model and the best performer was again a weighted combination of other models [54]. however, none of the models replicated the observed data more than 50% of the time and the challenge did not include an influenza pandemic—the models were optimized to predict seasonal flu. current epidemiological modeling for emerging infectious diseases tends toward using relatively simple models that only require the estimation of a few parameters primarily due to the scarcity of data [55,56]. more complex tools for assessing the pandemic risk of influenza [57,58] have not been widely adopted—one reason being their substantial data requirements. ultimately, choosing the level of model detail is a professional judgement in the art of modeling until the state of the art is improved. the current best solution to the problem of potentially non-representative data, processes or models is more attitudinal than technical. in the face of ignorance, the best approach is to use multiple competing models and constantly question their assumptions and results. in the case of epidemic modeling, real-time prediction remains a difficult challenge [59]. furthermore, longer-term forecasting and predicting the emergence of new pathogens with pandemic potential will likely remain beyond our capabilities for the near future [60]. given the dynamic nature of factors underlying epidemic progression, there may be limits on long-term predictability of infectious disease outbreaks and new analytical methods may be needed [61]. 4. conclusion catastrophe models can err in disaster forecasting for a variety of reasons. here i have outlined one reason that is essentially an error of mathematical method and another based on an important psychological bias related to natural human optimism. however, there are others. for example, a pernicious cause of overconfidence in modeling is the human tendency toward excessive admiration of our own creations [62,63]. 132 d.j. rozell / journal of risk analysis and crisis response 9(3) 128–133 ultimately, savvy users may know what level of confidence to put in their models, but shortcomings in current practices in uncertainty quantification require that model interpretation remain an art as much as a science until more progress is made. although the focus of this discussion was pandemic modeling, catastrophe models are tackling other complex hazards with strong sociological components, such as cybersecurity and terrorism risks. furthermore, standard catastrophe models are expanding into secondary areas such as business interruption losses [64]. given the potential for overreach, catastrophe model users will need to practice shrewd professional judgment for some time to come. despite the critiques presented here, it is important to clarify that planning for disasters using data and quantitative methods remains an essential approach that can provide significant insight. while there is danger in putting too much confidence in a model’s results, failing to consider modeling at all presents the much larger danger of missing critical trends and vulnerabilities hidden within an overwhelming mountain of data. the only way forward is to keep improving the models while continuing to use them with a healthy skepticism. conflicts of interest the author declare they have no conflicts of interest. acknowledgment i would like to thank sheldon reaven for insightful questions and comments regarding this work. references [1] grossi p, kunreuther h, windeler d. an introduction to castrophe models and insurance. in: grossi p, kunreuther h, editors. catastrophe modeling: a new approach to managing risk. boston, ma: springer; 2005, pp. 23–42. [2] fineberg hv. swine flu of 1976: lessons from the past. an interview with dr harvey v fineberg. bull world health organ 2009;87:414–5. [3] weinkle j, pielke r. the truthiness about hurricane catastrophe models. sci technol human values 2017;42:547–76. [4] mack a, snair mr, mundaca-shah c. the ebola epidemic in west africa: proceedings of a workshop. washington, dc: national academies press; 2016, p. 120. [5] adalja aa. ebola in west africa: a familiar pattern? biosecur bioterror 2014;12:161–2. [6] gomes mf, pastore y piontti a, rossi l, chao d, longini i, halloran me, et al. assessing the international spreading risk associated with the 2014 west african ebola outbreak. plos currents 2014;6. [7] pandey a, atkins ke, medlock j, wenzel n, townsend jp, childs je, et al. strategies for containing ebola in west africa. science 2014;346:991–5. [8] lewnard ja, ndeffo mbah ml, alfaro-murillo ja, altice fl, bawo l, nyenswah tg, et al. dynamics and control of ebola virus transmission in montserrado, liberia: a mathematical modelling analysis. lancet infect dis 2014;14:1189–95. [9] who ebola response team, aylward b, barboza p, bawo l, bertherat e, bilivogui p, et al. ebola virus disease in west africa—the first 9 months of the epidemic and forward projections. n engl j med 2014;371:1481–95. [10] enserink m. infectious diseases. in guinea, a long, difficult road to zero ebola cases. science 2015;348:485–6. [11] hayden ec. ebola virus and its legacy linger on. nature 2016;537:291–2. [12] macal c, north m. introductory tutorial: agent-based modeling and simulation. in: proceedings of the winter simulation conference (wsc) 2014. savanah, ga, usa: ieee press; 2014, pp. 6–20. [13] venkatramanan s, lewis b, chen j, higdon d, vullikanti a, marathe m. using data-driven agent-based models for forecasting emerging infectious diseases. epidemics 2018;22:43–9. [14] haas cn. the role of risk analysis in understanding ebola. risk anal 2015;35:183–5. [15] butler d, morello l. ebola by the numbers: the size, spread and cost of an outbreak. nature 2014;514:284–5. [16] sabeti p, salahi l. outbreak culture: the ebola crisis and the next epidemic, 1st ed., cambridge, ma: harvard university press; 2018, p. 288. [17] frieden tr. still not ready for ebola. science 2018;360:1049. [18] cohen j. congo rapidly curtails ebola. science 2018;361:211–12. [19] maxmen a. violence propels ebola outbreak towards 1,000 cases. nature 2019;567:153–4. [20] spinney l. in congo, fighting a virus and a groundswell of fake news. science 2019;363:213–4. [21] maxmen a. ebola detectives race to identify hidden sources of infection as outbreak spreads. nature 2018;564:174–5. [22] cnops l, de smet b, mbala-kingebeni p, van griensven j, ahuka-mundeke s, ariën kk. where are the ebola diagnostics from last time? nature 2019;565:419–21. [23] eksin c, paarporn k, weitz js. systematic biases in disease forecasting the role of behavior change. epidemics 2019;27:96–105. [24] flage r, aven t, zio e, baraldi p. concerns, challenges, and directions of development for the issue of representing uncertainty in risk assessment. risk anal 2014;34:1196–207. [25] tannert c, elvers h-d, jandrig b. the ethics of uncertainty. in the light of possible dangers, research becomes a moral duty. embo rep 2007;8:892–6. [26] apostolakis g. the concept of probability in safety assessments of technological systems. science 1990;250:1359–64. [27] winkler rl. uncertainty in probabilistic risk assessment. reliabil eng syst safety 1996;54:127–32. [28] o’hagan a, oakley je. probability is perfect, but we can’t elicit it perfectly. reliabil eng syst safety 2004;85:239–48. [29] van asselt mba, rotmans j. uncertainty in integrated assessment modelling. climatic change 2002;54:75–105. [30] walker we, harremoës p, rotmans j, van der sluijs jp, van asselt mba, janssen p, et al. defining uncertainty: a conceptual basis for uncertainty management in model-based decision support. integr assess 2003;4:5–17. [31] norton jp, brown jd, mysiak j. to what extent, and how, might uncertainty be defined? comments engendered by “defining uncertainty: a conceptual basis for uncertainty management in model-based decision support” (walker et al., 2003). integr assess 2006;6:83–8. [32] krayer von krauss mp, walker we, van der sluijs jp, janssen p, van asselt mba, rotmans j. response to “to what extent, and https://doi.org/10.2471/blt.09.040609 https://doi.org/10.2471/blt.09.040609 https://doi.org/10.2471/blt.09.040609 https://doi.org/10.1177/0162243916671201 https://doi.org/10.1177/0162243916671201 https://doi.org/10.1089/bsp.2014.1563 https://doi.org/10.1089/bsp.2014.1563 https://doi.org/10.1371/currents.outbreaks.cd818f63d40e24aef769dda7df9e0da5 https://doi.org/10.1371/currents.outbreaks.cd818f63d40e24aef769dda7df9e0da5 https://doi.org/10.1371/currents.outbreaks.cd818f63d40e24aef769dda7df9e0da5 https://doi.org/10.1371/currents.outbreaks.cd818f63d40e24aef769dda7df9e0da5 https://doi.org/10.1126/science.1260612 https://doi.org/10.1126/science.1260612 https://doi.org/10.1126/science.1260612 https://doi.org/10.1016/s1473-3099%2814%2970995-8 https://doi.org/10.1016/s1473-3099%2814%2970995-8 https://doi.org/10.1016/s1473-3099%2814%2970995-8 https://doi.org/10.1016/s1473-3099%2814%2970995-8 https://doi.org/10.1056/nejmoa1411100 https://doi.org/10.1056/nejmoa1411100 https://doi.org/10.1056/nejmoa1411100 https://doi.org/10.1056/nejmoa1411100 https://doi.org/10.1126/science.348.6234.485 https://doi.org/10.1126/science.348.6234.485 https://doi.org/10.1109/wsc.2014.7019874 https://doi.org/10.1109/wsc.2014.7019874 https://doi.org/10.1109/wsc.2014.7019874 https://doi.org/10.1109/wsc.2014.7019874 https://doi.org/10.1016/j.epidem.2017.02.010 https://doi.org/10.1016/j.epidem.2017.02.010 https://doi.org/10.1016/j.epidem.2017.02.010 https://doi.org/10.1111/risa.12361 https://doi.org/10.1111/risa.12361 https://doi.org/10.1038/514284a https://doi.org/10.1038/514284a https://doi.org/10.1126/science.aau3345 https://doi.org/10.1126/science.361.6399.211 https://doi.org/10.1038/d41586-019-00805-7 https://doi.org/10.1038/d41586-019-00805-7 https://doi.org/10.1126/science.363.6424.213 https://doi.org/10.1126/science.363.6424.213 https://doi.org/10.1038/d41586-018-07618-0 https://doi.org/10.1038/d41586-018-07618-0 https://doi.org/10.1038/d41586-019-00212-y https://doi.org/10.1038/d41586-019-00212-y https://doi.org/10.1038/d41586-019-00212-y https://doi.org/10.1016/j.epidem.2019.02.004 https://doi.org/10.1016/j.epidem.2019.02.004 https://doi.org/10.1111/risa.12247 https://doi.org/10.1111/risa.12247 https://doi.org/10.1111/risa.12247 https://doi.org/10.1038/sj.embor.7401072 https://doi.org/10.1038/sj.embor.7401072 https://doi.org/10.1038/sj.embor.7401072 https://doi.org/10.1126/science.2255906 https://doi.org/10.1126/science.2255906 https://doi.org/10.1016/s0951-8320%2896%2900070-1 https://doi.org/10.1016/s0951-8320%2896%2900070-1 https://doi.org/10.1016/j.ress.2004.03.014 https://doi.org/10.1016/j.ress.2004.03.014 https://doi.org/10.1023/a:1015783803445 https://doi.org/10.1023/a:1015783803445 https://doi.org/10.1076/iaij.4.1.5.16466 https://doi.org/10.1076/iaij.4.1.5.16466 https://doi.org/10.1076/iaij.4.1.5.16466 https://doi.org/10.1076/iaij.4.1.5.16466 d.j. rozell / journal of risk analysis and crisis response 9(3) 128–133 133 how, might uncertainty be defined” by norton, brown, and mysiak. integr assess 2006;6:89–94. [33] keynes jm. a treatise on probability. london, uk: macmillian and co.; 1921, p. 539. [34] grinstead cm, snell jl. introduction to probability. rhode island: american mathematical society; 1997, p. 510. available from: http://www.dartmouth.edu/~chance/teaching_aids/books_ articles/probability_book/amsbook.mac.pdf. [35] jablonowski m. do catastrophe models mislead? risk manage 2005;52:32–8. [36] walley p. statistical reasoning with imprecise probabilities. london: chapman and hall; 1991, p. 706. [37] ferson s, ginzburg lr. different methods are needed to propagate ignorance and variability. reliabil eng syst safety 1996;54:133–44. [38] berger jo. statistical decision theory and bayesian analysis, 2nd ed., new york: springer science & business media; 1985, p. xvi, 618. [39] berger jo, moreno e, pericchi lr, jesús bayarri m, bernardo jm, cano ja, et al. an overview of robust bayesian analysis. test 1994;3:5–124. [40] bredehoeft jd. from models to performance assessment: the conceptualization problem. ground water 2003;41:571–7. [41] oreskes n, shrader-frechette k, belitz k. verification, validation, and confirmation of numerical models in the earth sciences. science 1994;263:641–6. [42] manheim d. questioning estimates of natural pandemic risk. health secur 2018;16:381–90. [43] ceres rl, forest ce, keller k. understanding the detectability of potential changes to the 100-year peak storm surge. climatic change 2017;145:221–35. [44] dalziel bd, kissler s, gog jr, viboud c, bjørnstad on, metcalf cje, et al. urbanization and humidity shape the intensity of influenza epidemics in u.s. cities. science 2018;362: 75–9. [45] air worldwide. the air pandemic flu model; 2014. available from: https://www.air-worldwide.com/publications/air-currents/2014/ the-air-pandemic-flu-model/. [46] air worldwide. air worldwide expands its global pandemic model; 2016. available from: https://www.globenewswire.com/ news-release/2016/07/25/858573/10164236/en/air-worldwideexpands-its-global-pandemic-model.html. [47] air worldwide. the air pandemic model; 2018. available from: https://www.air-worldwide.com/models/life-and-health/. [48] karl jb, medders la, maroney pf. the effects of revealed information on catastrophe loss projection models’ characterization of risk: damage vulnerability evidence from florida. risk anal 2016;36:1224–50. [49] goldstein dg, gigerenzer g. fast and frugal forecasting. int j forecast 2009;25:760–72. [50] makridakis s, hibon m, moser c. accuracy of forecasting: an empirical investigation. j r stat soc ser a 1979;142:97–145. [51] makridakis s, taleb n. decision making and planning under low levels of predictability. int j forecast 2009;25:716–33. [52] viboud c, vespignani a. the future of influenza forecasts. proc natl acad sci u s a 2019;116:2802–4. [53] viboud c, sun k, gaffey r, ajelli m, fumanelli l, merler s, et al. the rapidd ebola forecasting challenge: synthesis and lessons learnt. epidemics 2018;22:13–21. [54] reich ng, brooks lc, fox sj, kandula s, mcgowan cj, moore e, et al. a collaborative multiyear, multimodel assessment of seasonal influenza forecasting in the united states. proc natl acad sci u s a 2019;116:3146–54. [55] metcalf cje, lessler j. opportunities and challenges in modeling emerging infectious diseases. science 2017;357:149–52. [56] tuite ar, fisman dn. the idea model: a single equation approach to the ebola forecasting challenge. epidemics 2018;22:71–7. [57] trock sc, burke sa, cox nj. development of an influenza virologic risk assessment tool. avian dis 2012;56 4 suppl:1058–61. [58] cox nj, trock sc, burke sa. pandemic preparedness and the influenza risk assessment tool (irat). curr top microbiol immunol 2014;385:119–36. [59] kraemer mug, cummings dat, funk s, reiner rc, faria nr, pybus og, et al. reconstruction and prediction of viral disease epidemics. epidemiol infect 2018;147:1–7. [60] holmes ec, rambaut a, andersen kg. pandemics: spend on surveillance, not prediction. nature 2018;558:180–2. [61] scarpino sv, petri g. on the predictability of infectious disease outbreaks. nat commun 2019;10:898. [62] festinger l. a theory of cognitive dissonance. redwood city, ca: stanford university press; 1957, p. 291. [63] norton mi, mochon d, ariely d. the ikea effect: when labor leads to love. j consumer psychol 2012;22:453–60. [64] rose a, huyck ck. improving catastrophe modeling for business interruption insurance needs. risk anal 2016;36:1896–915. https://doi.org/10.1016/s0951-8320%2896%2900071-3 https://doi.org/10.1016/s0951-8320%2896%2900071-3 https://doi.org/10.1016/s0951-8320%2896%2900071-3 https://doi.org/10.1007/978-1-4757-4286-2 https://doi.org/10.1007/978-1-4757-4286-2 https://doi.org/10.1007/bf02562676 https://doi.org/10.1007/bf02562676 https://doi.org/10.1007/bf02562676 https://doi.org/10.1126/science.263.5147.641 https://doi.org/10.1126/science.263.5147.641 https://doi.org/10.1126/science.263.5147.641 https://doi.org/10.1089/hs.2018.0039 https://doi.org/10.1089/hs.2018.0039 https://doi.org/10.1007/s10584-017-2075-0 https://doi.org/10.1007/s10584-017-2075-0 https://doi.org/10.1007/s10584-017-2075-0 https://doi.org/10.1126/science.aat6030 https://doi.org/10.1126/science.aat6030 https://doi.org/10.1126/science.aat6030 https://doi.org/10.1111/risa.12524 https://doi.org/10.1111/risa.12524 https://doi.org/10.1111/risa.12524 https://doi.org/10.1111/risa.12524 https://doi.org/10.1016/j.ijforecast.2009.05.010 https://doi.org/10.1016/j.ijforecast.2009.05.010 https://doi.org/10.1016/j.ijforecast.2009.05.013 https://doi.org/10.1016/j.ijforecast.2009.05.013 https://doi.org/10.1073/pnas.1822167116 https://doi.org/10.1073/pnas.1822167116 https://doi.org/10.1016/j.epidem.2017.08.002 https://doi.org/10.1016/j.epidem.2017.08.002 https://doi.org/10.1016/j.epidem.2017.08.002 https://doi.org/10.1073/pnas.1812594116 https://doi.org/10.1073/pnas.1812594116 https://doi.org/10.1073/pnas.1812594116 https://doi.org/10.1073/pnas.1812594116 https://doi.org/10.1126/science.aam8335 https://doi.org/10.1126/science.aam8335 https://doi.org/10.1016/j.epidem.2016.09.001 https://doi.org/10.1016/j.epidem.2016.09.001 https://doi.org/10.1016/j.epidem.2016.09.001 https://doi.org/10.1637/10204-041412-resnote.1 https://doi.org/10.1637/10204-041412-resnote.1 https://doi.org/10.1007/82_2014_419 https://doi.org/10.1007/82_2014_419 https://doi.org/10.1007/82_2014_419 https://doi.org/10.1017/s0950268818002881 https://doi.org/10.1017/s0950268818002881 https://doi.org/10.1017/s0950268818002881 https://doi.org/10.1038/d41586-018-05373-w https://doi.org/10.1038/d41586-018-05373-w https://doi.org/10.1038/s41467-019-08616-0 https://doi.org/10.1038/s41467-019-08616-0 https://doi.org/10.1016/j.jcps.2011.08.002 https://doi.org/10.1016/j.jcps.2011.08.002 https://doi.org/10.1111/risa.12550 https://doi.org/10.1111/risa.12550 journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 149–154 doi: https://doi.org/10.2991/jracr.k.191024.006; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article exploration of risk oriented fund performance audit—take rural endowment insurance as an example ronggang zhang*, lanzhihao liu business school, northwest university of political science and law, xi’an 710122, china 1. introduction the performance of rural endowment insurance fund is the core goal of the establishment of rural endowment insurance, and also the key to its sustainable development. with the increasingly serious aging of china’s population, there are more and more participants in rural endowment insurance. audit is an important channel to ensure the performance of rural endowment insurance fund. starting in the 1980s, our government is exploratory created the system of rural endowment insurance, the notice a full audit, audit in 2015 but our country rural endowment insurance fund audit supervision time is very short, covering wide social group, its policies and complicated content, means and methods, combined with the current audit result in relative to the general audit of rural endowment insurance fund risk is much larger. at present, the traditional audit method can not meet the requirement of rural pension fund performance audit. dubnick [1] mentioned in his paper that public fiduciary responsibility should be the sum of various methods adopted to manage the expectations of the principal and achieve the organizational goals. the application effect and function of these methods will constantly change due to the position of the principal. behn robert [2] proposed that the theoretical framework gradually converges, and it should emphasize the comprehensive evaluation of public expenditure at various stages before, during and after the event. li [3] proposed to carry out comprehensive audit and innovate audit methods and means. qiu [4] proposed that the minimum pension benefits may not be able to maintain the minimum living standards of the masses, and the adjustment of benefits performance audit needs to be paid attention to. lv [5] proposed to pay attention to the value added and maintenance of pension fund and improve the fund performance audit system. tan [6] proposed that china’s pension fund audit should carry out performance audit, and converge with the world. wang [7] mentioned the existence of government performance audit risk, and risk-oriented government performance audit has become an irresistible trend. chinese scholars have begun to actively explore the performance audit of chinese rural pension fund. in addition, some scholars have proposed to apply wind direction to policy performance audit, but few scholars have studied new audit methods and ideas for rural pension fund performance audit. actively carrying out rural pension fund performance audit is a necessary measure to ensure reasonable, compliant and efficient use of the fund, an important way to explore and innovate the national audit function, and the only way to ensure the healthy operation of social economy. we should actively explore new ideas of performance audit optimization. nowadays risk-oriented audit has been widely used in social audit, and its application in national policy should arouse our attention and thinking. introducing risk orientation into performance audit will be a new idea and breakthrough of fund performance audit in china. 2. status quo of performance audit of rural endowment insurance fund 2.1. the social demand is too low in the guidance opinion on the audit operations of local audit institutions in 2015 issued by the audit office, it is mentioned that the national unified organization will do a good job in the audit of pension funds from august to october 2015 to find out the fund raising, management, use, value-added and preservation. a r t i c l e i n f o article history received 31 may 2019 accepted 10 july 2019 keywords risk oriented audit fund performance audit rural endowment insurance a b s t r a c t the performance audit risk of rural pension fund is worth studying deeply. based on the cost-benefit principle and the public fiduciary economic responsibility theory, this paper analyzes the current situation and causes of the performance audit of rural pension fund and finds that the audit demand degree is not enough. this paper proposes to integrate risk orientation into performance audit and adopt different audit methods and evaluation systems to improve audit efficiency and reduce costs by dividing risk levels and optimizing performance audit. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 463214273@qq.com https://doi.org/10.2991/jracr.k.191024.006 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:463214273%40qq.com?subject= 150 r. zhang and l. liu / journal of risk analysis and crisis response 9(3) 149–154 the public generally do not understand the performance audit of the rural endowment insurance fund. the insufficient publicity of the performance of the endowment insurance fund results in insufficient social demand for the audit of the endowment insurance and low public attention. the overall social performance of rural endowment insurance fund performance ignored. public fiduciary economic responsibility refers to the obligation to manage fiduciary economic resources and report their operating conditions in accordance with specific requirements or principles. according to the theory of public trust economic responsibility, the neglect of the performance of pension fund by the public will result in the neglect of the performance audit of pension fund by the state to a great extent, which will lead to the stagnation of the performance audit of pension fund in china. 2.2. the focus of auditing deviates from the target the author thinks that the audit of rural pension fund should pay attention to the performance audit of pension fund. endowment insurance fund performance audit is refers to the special audit institutions and auditors accept authorized or entrusted, in accordance with the relevant laws and regulations, using special audit method, for the auditees configuration, management and utilization of resources of economy, efficiency and effect audit, supervision, evaluation and verification activities, its purpose is to improve the rural old-age insurance fund management, increasing the service efficiency of funds. however, in the current pension fund management, the value added function of rural pension fund is weakened. as a result, the main focus of pension fund audit is to review the empty account of rural pension fund. at present, the clear direction of fund performance audit is only to audit the appreciation and preservation of the remaining funds, and the results of fund management audit are mainly reflected by the rate of return on fund investment. the audit direction is too single; the audit result is also relatively onesided. the performance audit of rural pension fund requires the economy, efficiency and effectiveness of the fund. it is not enough to audit the fund performance of rural endowment insurance only by auditing the appreciation and preservation of the surplus fund of rural endowment insurance. 2.3. audit cost is too high at present, our rural old-age insurance audit focuses mainly on the issue of empty accounts of individual accounts. according to statistics, the empty accounts of rural old-age insurance in china in 2016 reached 360 million yuan. the so-called “empty account” refers to the fact that in the actual implementation of the system, because no one bears the transformation cost from the pay-as-you-go system to the part accumulation system, the individual account payment has to be used to protect the current pension payment, thus forming a “empty account” affecting the sustainable development of the pension, reducing the rural pension insurance credibility. because of the wide coverage of rural endowment insurance, the problem of personal empty account makes the rural endowment insurance fund which should be used to realize value-added is used to issue rural endowment insurance. in order to carry out a comprehensive audit of the legitimacy, rationality, compliance, economy, efficiency and effectiveness of the endowment insurance fund, the large-scale existence of individual empty accounts raises the audit risk of the performance audit of the rural endowment insurance fund on the one hand, and on the other hand increases the audit cost of the performance audit of the rural endowment insurance fund. 3. analysis of causes for performance audit of rural endowment insurance fund because the performance audit of rural endowment insurance fund in china is still in the process of continuous exploration, there are many problems such as: there is no special method of performance audit of endowment insurance fund; it is difficult to obtain evidence; the focus of performance audit is not comprehensive enough; the problems of internal control of endowment insurance fund; the performance audit of endowment insurance fund. the evaluation system is not perfect enough. but we can not only wait for the maturity of the policy and the institutionalization of the management to promote the development of the rural endowment insurance, nor can we expect the comprehensive audit of the rural endowment insurance with increasing investment. we must start from the performance audit of the endowment insurance fund and look for the audit of the rural endowment insurance itself to be able to do so. in order to promote the development of the performance audit of china’s rural endowment insurance fund, we should change the passive measures to take the initiative. 3.1. audit work less attention 3.1.1. the government department lack of understanding the rural endowment insurance fund is a fund specially set up by the government to improve the living conditions of the elderly through the rural endowment insurance and provide them with a stable and reliable source of livelihood. although the provisional regulations on the collection and payment of social insurance premiums issued by the state council in 1999 stipulates in article 22, “auditing departments shall supervise the income and expenditure of social insurance funds according to law.” however, china’s pension fund performance audit started very late, is still in the exploratory stage. at present, most of the rural pension fund audit in china is still at the traditional financial audit level. departments of the government ignored the performance audit of endowment insurance fund and endowment insurance fund performance audit of endowment insurance fund profitability important influence, at present our country’s endowment insurance fund performance audit has no legal basis, performance audit of endowment insurance fund is the basis of the theory research and practice is weak, cause the endowment insurance fund performance audit in conducting many blame mistakes and not willing to cooperate with related department. it is difficult to carry out the rural pension fund performance audit. r. zhang and l. liu / journal of risk analysis and crisis response 9(3) 149–154 151 3.1.2. public neglect according to the data survey, in rural areas, the respondents are over 35 years old, mainly between 45 and 64 years old. nearly half of the respondents have an annual income of less than 1000 yuan, and most of the rest are between 1000 and 2000 yuan. through the investigation, it is found that 99% of the insured people do not understand the pension fund audit, do not know audit is an important means to supervise the security of pension insurance. however, the public generally do not understand the performance audit of rural pension fund, and the insufficient publicity and popularization of the performance audit of pension fund leads to the insufficient social demand for the audit of pension insurance and the low public attention. the whole society has ignored the performance audit of rural pension fund. the public neglect of pension fund performance can not produce the demand to promote the development of pension fund performance audit in china. for the public, the days when managers of government-owned pension funds merely maintained their value are over. the performance of people’s livelihood fund has become the focus of all guarantee objects in the 21st century. if the government can pay attention to and consider the fund performance, we can reasonably adopt different audit methods and develop effective system and evaluation system for the pension fund performance. and the comprehensive coverage of audit has put forward new requirements for fund performance audit. if the fund performance management cannot be fundamentally valued, the comprehensive audit of people’s livelihood security and other aspects cannot fundamentally achieve its audit objectives. 3.2. the elements of performance audit are not mature 3.2.1. auditing methods are relatively backward because each process design link of rural endowment insurance fund is very complex, the transmission and summary of information needs to be done through the digital equipment, to a great extent, restricts the traditional auditing technique method, method of obtaining audit evidence is still in the original manual audit method, raises the fund audit evidence collection difficulty, at the same time in the process of performance audit, many problems can only qualitative analysis, a variety of reasons, led to the current our country performance audit can not meet the requirements of deeper, these problems will increase the audit of the audit risk of rural endowment insurance, weaken the depth and effectiveness of performance audit, it cannot reach the expected audit target of rural endowment insurance. 3.2.2. poor personnel ability the neglect of the rural pension fund performance audit by the audit department largely limits the professional ability of the auditors. at present, most of the auditors in our country are accounting personnel and the awareness of audit professional ability is not enough. china’s current pension fund performance audit is very immature. in the absence of policy enforcement, the state lacks awareness cultivation and ability improvement of fund performance audit in the training of auditors. when carrying out the performance audit of rural pension fund, auditors are reluctant but unable to achieve the purpose of the performance audit of pension fund. 3.2.3. the fund performance audit evaluation system is not mature our country does not have the present rural endowment insurance fund performance audit evaluation system the fund performance audit development lacks the guidance document. the scientific and standard pension fund performance audit evaluation system is an important basis for the orderly development of pension fund performance audit in china. the lack of audit evaluation system seriously weakens the authority and credibility of china’s fund performance audit, leading to a large discount of fund performance audit results. so far, china’s official only issued three sets of the performance evaluation index system, in 1995 “the ministry of finance issued the economic benefit evaluation index system for enterprises”, 1999 joint state economic and trade committee issued “enterprise performance evaluation system”, issued by the state council in 2004 the local government performance evaluation index system of the first two of endowment insurance fund performance audit basic does not have the reference value; the index system of local government performance evaluation mainly evaluates the work performance of local governments and has very limited guiding significance for the performance audit of rural pension fund. therefore, it is urgent to establish an evaluation system of rural pension fund performance audit in china. analyzing the three aspects from another aspect, we can all attribute them to the neglect of risk. if risks are not taken seriously and considered, we can reasonably adopt different audit methods and have a more appropriate evaluation index system for the changes of risks. risk is a problem that pension fund performance audit must pay attention to and think about. 4. based on fund performance audit of public entrusted economic responsibility appraisal public entrusted economic responsibility mainly aimed at the public resources, and the government accepts the commission of the people, to exercise various functions and powers, in entrusted with the responsibility of public relations, as the government has the absolute advantage of the trustee, and as the principal of the social public in a disadvantage position, this kind of information asymmetry can lead to moral hazard and adverse selection. auditing, as a professional and independent third-party organization, can externally supervise the performance of the government’s public fiduciary responsibility, minimize agency risks and guarantee the fund performance of citizens. based on the analysis of public entrusted economic responsibility, rural endowment insurance fund is the public special fund that the insurance management department raises from individuals and 152 r. zhang and l. liu / journal of risk analysis and crisis response 9(3) 149–154 their units according to law. its owner is all the insured personnel, and its management is the government’s relevant social insurance agency. in the management of rural endowment insurance, the ownership and management rights of rural endowment insurance have been separated, resulting in agency risk. at the same time, in the management and investment process of fund, rural endowment insurance has generated secondary agency risk, and its own risk is greater. according to the audit risk = material misstatement risk * inspection risk the audit risk of the pension fund will continue to expand with the increase of the risk of the pension fund itself. with the increasing development of rural endowment insurance, the public’s attention on the performance of rural endowment insurance is gradually increasing. once problems arise in the performance audit of pension fund, serious social effects will be caused, affecting the government’s credibility and the sustainable development of pension policy. from the perspective of public fiduciary responsibility, we can conclude that the problems in pension fund audit can be attributed to the neglect of pension fund performance. at the 19th national congress of the communist party of china (cpc), it was proposed to “establish a comprehensive, standardized and transparent budget system with scientific standards and strong constraints, and fully implement performance management”. for the public, the days when managers of government-owned pension funds merely maintained their value are over. the performance of people’s livelihood fund has become the focus of all guarantee objects in the 21st century. if the government can pay attention to and consider the fund performance, we can reasonably adopt different audit methods and develop effective system and evaluation system for the pension fund performance. and the comprehensive coverage of audit has put forward new requirements for fund performance audit. if the fund performance management cannot be fundamentally valued, the comprehensive audit of people’s livelihood security and other aspects cannot fundamentally achieve its audit objectives. 5. performance audit of rural endowment insurance fund based on risk orientation 5.1. risk oriented audit approach risk oriented auditing refers to the audit conducted by certified practising accountant (cpa) based on the audit risk model. we should guide the risk assessment of major misstatements and the whole audit process with the concept of strategy and system. it is a kind of audit mode based on the risk assessment of the auditees, comprehensively analyzing the factors that affect the economic activities of the auditees, and determining the scope and focus of implementation according to the quantified risk level, so as to conduct substantive review (professor chunyuan hu, “risk-oriented auditing”). the author believes that risk-oriented audit will control risks and inherent risks after considering the major misstatement risk, through the analysis of the environment, strategy, management and other factors, the formulation of the audit plan is more suitable for different circumstances of the audit business. compared with other audit models, wind-guided audit only needs to perform limited substantive testing according to audit plan to reduce audit risk. at the same time, it is emphasized that no matter the size of audit risk, we must conduct substantive tests on important trading activities. when it is found that there is a misreporting risk in a transaction, it can also timely adjust the audit direction, improve the audit efficiency and save the audit cost; risk-oriented audit has a strong applicability, it can be matched with other audit modes, as the risk-oriented rural pension fund performance audit proposed in this paper, on it is the combination of risk-oriented audit and performance audit to audit and supervise the rural endowment insurance fund. 5.2. risk oriented fund performance audit in the performance audit of rural endowment insurance fund, the audit risk mainly comes from the major misrepresentation risk in the use of rural endowment insurance fund. the core idea of risk oriented audit is to reduce the risk of rural endowment insurance fund in economic, effective and effective audit [8]. through risk oriented auditing, we can control the risk of major misstatement in the performance audit of pension fund. we should make a detailed investigation and careful examination of some economic activities that have been assessed to be more risky, so as to reduce the risks that may arise in the performance audit of rural endowment insurance funds. 5.2.1. risk assessment public entrusted economic responsibility mainly aims at public resources. the government accepts the people’s entrustment and exercises various functions and powers. in the relationship of public entrusted responsibility, the government as the trustee has absolute advantages, while the public as the trustee is in a disadvantaged position. this information asymmetry may lead to moral hazard and reverse choice. as a professional and independent third party organization, auditing can supervise information from outside and reduce agency risk as much as possible. rural endowment insurance fund based on public trusted economic responsibility is a special public fund raised by insurance management departments from individuals and their units according to law. the owner of the fund is all the insured persons, and the managers are government-related social insurance agencies. in the management of rural endowment insurance, the ownership and management rights of rural endowment insurance have been separated, which has resulted in agency risk. at the same time, in the process of fund management and investment, rural endowment insurance has produced secondary agency risk, and its own risk is even greater. according to audit risk = major misstatement risk * risk checking. the audit risk of endowment insurance fund will continue to expand with the increasing risk of pension fund itself. with the development of rural endowment insurance, the public’s attention to the performance of rural endowment insurance has gradually increased. once the performance audit of endowment insurance fund has problems, it will cause serious social effects and affect the credibility of the government and the sustainable development of endowment policy. the author believes that risk-oriented audit should be introduced, and the performance audit should be carried out after the risk r. zhang and l. liu / journal of risk analysis and crisis response 9(3) 149–154 153 assessment of the whole by the relevant auditors before the performance audit of rural endowment insurance. risk levels are classified according to the degree of risk: level 1 risk (slightly risky), level 2 risk (general risk), and level 3 risk (serious risk). 5.2.2. update audit method cost-benefit theory refers to that under the condition of unbalanced supply and demand, a project is feasible only if the benefit is greater than the cost, and through comparative analysis can find the best way to obtain the maximum benefit with the minimum cost. it is not difficult to find that the performance audit of rural endowment insurance funds in china is blindness after the comprehensive audit, because it involves a wide range of economic activities in the audit process, generally high audit costs, not in line with the principle of cost-effectiveness. risk-oriented audit is introduced into performance audit, and different audit methods are adopted for different risks through risk assessment, so as to reduce the cost of performance audit of pension funds, rationally allocate audit resources, and optimize audit methods to maximize audit benefits. level 1 risks: mainly using sampling method, rural pension fund audited more units, and the data is more complex, auditors can not be checked one by one, should be less risk fund activities reasonable use of sampling audit. its main purpose is to audit the efficiency of funds. level 2 risks: mainly financial audit, supplemented by performance audit. in order to ensure the authenticity, legality and compliance of the fund sources, the economic, efficient and effective use of the fund can be guaranteed by the forward or backward investigation method. level 3 risks: financial audit, performance audit at the same time. two-way audit, on the one hand, from the collection of funds, accounts of the fund details to the specific use of the fund for positive inspection, review of the current account details, should adopt the method of field investigation, and actively look for external evidence. on the other hand, tracking method can be actively used to ensure the authenticity and reliability of the audit from the audit process of the inversion of fund accounts. in the process of audit, we actively adopt advanced audit methods such as computer audit, cloud audit and other new ideas. at the same time, we use factor analysis method to conduct a comprehensive and detailed audit of the three-level risk assessment results of the project. to ensure the rationality, legitimacy, economy, efficiency and effectiveness of the rural endowment insurance fund audit. 5.2.3. improved evaluation system the new public management theory advocates that the government should adopt the advanced management methods and competition mechanism which absorb the private sector. the government departments should establish the “effect-oriented government” and pay close attention to the efficiency of public management activities. at the same time, the establishment of “measurable performance indicators” public management theory to a large extent supports the performance audit of rural endowment insurance funds, but also puts forward requirements for the performance audit of rural endowment insurance, its audit system to be assessable, which is the current performance audit of rural endowment insurance funds has not yet achieved the goal [9]. the introduction of risk-oriented thinking can improve the performance audit evaluation system of pension funds to a certain extent. in the measurement system, it is more effective to use different evaluation index system to evaluate the audit risk according to the classification of risk and different audit risk [10]. the evaluation system of the performance audit of the pension fund should be considered comprehensively by combining qualitative analysis with quantitative analysis. economic evaluation indicators: (a) fund management costs: fund management costs, operating supervision costs can also be measured by their ratio to the total social insurance funds; (b) the total weight of pension expenditure in gdp = current pension insurance fiscal expenditure/current gdp 3 pension insurance expenditure in the proportion of fiscal expenditure = current pension insurance risk expenditure/current fiscal expenditure [11]. efficiency evaluation indicators: (a) pension fund investment rate of return = current pension fund investment income/current pension fund total; (b) pension fund investment loss rate = current pension fund investment loss/current pension fund total; (c) pension insurance expenditure benefit = pension insurance. effective evaluation indicators: (a) actual levy rate = actual levy rate of endowment insurance fund/plan levy of endowment insurance fund * 100%, if the levy rate is greater than 100%, it shows that the fund over levy; (b) actual payment rate of endowment insurance expense = actual payment of endowment insurance expense/plan payment of endowment insurance expense; (c) rural endowment insurance base. the rate of gold collection = the actual receipts of the current rural endowment insurance fund/the current age pension fund receivables [11]. level 1 risks is mainly the effectiveness audit. the final results of the rural endowment insurance fund performance audit, the use of quantitative analysis of the fund in the process of investment financial efficiency indicators are calculated and combined with qualitative analysis of the social and economic effects of its evaluation, in order to determine the rural endowment insurance fund management investment objectives to achieve the pros and cons. level 2 risks is on the basis of the authenticity and legality of the endowment insurance fund itself, the economic and efficient audit of the fund activities with level 2 risk is emphasized. this paper investigates the rationality and economy of the anticipated investment of the second-level venture project fund and the efficiency audit of each link in the fund turnover process. using quantitative analysis to compare the expected investment and expected income, actual investment and actual income of pension fund, the paper focuses on the value-added and preservation of the fund. combined with qualitative analysis, this paper analyzes whether the relevant management and internal control system of the audit of rural endowment insurance fund is effective in operation. through the above analysis to determine whether the pre-management and operation of the rural endowment insurance fund meets the requirements of the national rural endowment insurance fund management, in order to promote the management efficiency of the endowment insurance fund. 154 r. zhang and l. liu / journal of risk analysis and crisis response 9(3) 149–154 level 3 risks is for the major risk fund events resulting from the risk assessment results, financial audit and performance audit should be combined with the economic nature of the fund, efficiency and effectiveness. qualitative analysis and quantitative analysis are flexibly used to conduct comprehensive audit of the pension fund from the collection of the pension fund to the management of the fund and then to the value-added and preservation of the fund. 6. summary china’s rural endowment insurance fund is the ginseng and guarantee the material conditions of self-interest, rural endowment insurance fund itself span length, the amount of funds is differ, but the sum is very large, the range is wide, whether the rural endowment insurance fund management, affects the social stability and economic development of our country, to our country’s modernization construction, comprehensive sustainable development has important significance. rural pension insurance audit is the most important goal. we must pay attention to the pension fund performance audit and actively explore the new vision and new methods of pension fund performance audit. on the one hand, introducing riskoriented audit into fund performance audit will improve the efficiency and audit quality of current fund performance audit; on the other hand, audit cost can be reduced through more reasonable allocation of audit resources relying on risk-oriented audit. meanwhile, the most important role of risk-oriented audit is to reduce the risk of rural pension fund performance audit. this will become a new idea of fund performance audit optimization in china. since the risk-oriented audit focuses on the audit of major risk projects, it pays little attention to the rural endowment insurance fund projects with little risk and ignores the problems and potential risks it may cause. pension insurance fund performance audit may not meet the expected requirements; meanwhile, china’s pension fund performance audit is in the exploration stage, and risk oriented rural pension fund performance audit is a new idea about performance audit; its specific application needs to be further verified and practiced by other scholars. conflicts of interest the authors declare they have no conflicts of interest. references [1] romzek bs, dubnick mj. accountability in the public sector: lessons from the challenger tragedy. public adm rev 1987;47:227– 238. [2] rober b. public organization theory, translated by zhu ganwei, beijing: renmin university of china press, 2003. [3] li a. discussion on audit work of new rural social endowment insurance. finan econ acad ed 2011;7:213. [4] qiu y. performance audit analysis of adjustment of basic old-age insurance benefits in enterprises based on big data—a case study of x province. audit research 2014;3:106–12. [5] lv z. analysis on performance audit of endowment insurance fund. finan econ acad ed 2015;11:290–1. [6] tan m. a brief analysis of the problems in the performance audit of china’s endowment insurance fund. modern econ inform 2016;4:172. [7] wang a. research on risk oriented government performance audit beijing. capital univ econ business 2016. [8] ningbo audit association research group. special funds performance audit research. audit res 2014;3–8. [9] yu t. design of performance audit evaluation index system of social insurance fund based on balanced scorecard. account chinese township enterpr 2011;12. [10] audit commission. guidance of the audit commission on the audit work carried out by local audit institutions in 2015. [11] qiu y, lv t, li j. study on audit index system of basic endowment insurance policy implementation. audit res 2013;1:20–7. atlantis press journal style introduction for volume 8, issue 3 this issue contains 6 papers. there are 4 contributions written in english and 2 contributions in chinese with english abstracts. the papers can be divided into four topics: risk assessment, financial risk mitigation strategy, risk perception and risk analysis. there are two papers in risk assessment. the paper “assessing the influence of environmental factors on spatial soil erosion risk based on the certainty coefficient method”by jun wang, et al., assessed the spatial relationships between soil erosion risk and the environmental factors affecting soil erosion. the research is significant for monitoring future land use/cover changes, including agricultural expansion and deterioration of forest resources. the results indicate that soil erosion is highly correlated to specific slope categories, elevation zones, distance to rivers, land use/cover type, stratum lithology, soil types and annual 24h maximum rainfall. the second paper “study and application on risk assessment method of coal worker pneumoconiosis based on logistic regression model ” by qian zhang, et al., research the risk assessment method of coal-worker's pneumoconiosis caused by coal dust exposure, predict the incidence of coal-worker's pneumoconiosis, and provide technical basis for occupational risk management . in this study, a practical quantitative approach was proposed to assess the risk of coal-worker's pneumoconiosis caused by coal dust exposure. this method could be applied in evaluations of occupational disease hazard in construction projects and help to control and manage the risk of coal-worker's pneumoconiosis. there is two papers in financial risk mitigation strategy. the first paper“study on the relationship between financial development and poverty alleviation in guizhou province based on kuznets curve” by jianchun yang, et al., used granger causality test and regression analysis to analyze the relationship between financial development and poverty alleviation in guizhou province. the results present that the relationship between financial development and poverty alleviation follows the kuznets inverted u-curve, which means that financial development may have positive effects on poverty alleviation in the early stage and then have negative effects on poverty alleviation in later stage. based on this, in-depth analysis was conducted and development suggestions were put forward. the second paper “research on the sme’s collateral credit rationing under loan risk compensation mechanism” by changbing yang, et al., discusses the function of loan risk compensation in the collateral credit rationing model.according to the derivation of my model, we find when the bank have some requirements on interest and collateral, at the same time, we add the loan risk subsidy variable into the above model, loan risk subsidy can lower the bank's collateral requirements for the enterprise,therefore, we can conclude that loan risk compensation can effectively alleviate the credit rationing faced by enterprises. there is one paper in risk perception. the paper “ a description of earthquake risk perception in intelligent mathematics”by chongfu huang, proposes an framework of intelligent mathematics, which consists of factor space, information diffusion, and internet of intelligences. this paper demonstrates how to use intelligent mathematics to describe earthquake risk perception. the concept of earthquake risk is described in the factor space, and the knowledge of earthquake risk is formed in association learning based on information diffusion. different knowledge systems precipitate different acquired consciousnesses. the security instinct affects the perception of risk. for a region, the risk consciousnesses of the stakeholders could be integrated with the internet of intelligences to form a consensus of the earthquake risk perception. there is one paper in risk analysis. the paper “analysis of ultraviolet radiation characteristics and related factors in huangshan scenic area” by jianyong xu, et al., analyzes the relationship between the daily, monthly and seasonal variation characteristics of ultraviolet radiation intensity and its related meteorological elements by using the ultraviolet observation data of huangshan scenic area from march 2017 to february 2018. the results show that the diurnal variation of ultraviolet radiation intensity in huangshan scenic area is normal distribution at noon; the monthly average ultraviolet radiation intensity reaches the maximum in april and may; the ultraviolet radiation intensity in all seasons is above 4, and the largest in spring. autumn and winter 129 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 129-130 are second, and summer is the smallest. cloud amount, relative humidity and air pressure are all significantly negatively correlated with uv radiation intensity, and cloud amount, especially low cloud amount, is the most critical factor affecting uv radiation intensity. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca 130 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 3 (october 2018) 129-130 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). a fuzzy method for assessing eco-environmental disaster risk caused by coalbed methane in china ye xue1,2,*, xiaoxiao li1, wen sun1, baozhang chen2 1college of economics and management, taiyuan university of technology, taiyuan 030024, china 2institute of geographic sciences and natural resources research, chinese academy of sciences, beijing 100101, china received september 29, 2017 accepted january 8, 2018 abstract based on an improved risk assessment index system, this paper constructs an eco-environmental disaster risk assessment model during coalbed methane industrialization development in china by using intuitionistic fuzzy sets to describe the uncertain risk information, and the transformed interval value of mamdani intuitionistic fuzzy neural networks model. then, the validity of the model is verified by simulation tests. furthermore, the assessment results are compared with those obtained by fuzzy neural networks model. results show that suggested model has multidimensional nonlinearity and global approximation characteristics. by the procedure of “fuzzification fuzzy-rules defuzzifier”, the output conversion from uncertainty quantitative indicators to accurate risk assessment values can be effectively realized. compared with the fuzzy neural networks model, the suggested model has better accuracy and stability. the risk assessment value calculated by the suggested model fairly matches the expected one. the study supplies a decision support for routine supervision and risk precaution and management on one hand, and enriches the theoretical research of the eco-environmental risk assessment of coalbed methane industrialization development on the other hand. keywords: coalbed methane; eco-environment; risk assessment; intuitionistic fuzzy set; fuzzy neural system 1. introduction coalbed methane (cbm) is one of the most reliable supplements to oil and gas. however, with the rapid increase of cbm production, regional ecological risk problems become increasingly prominent. let’s take the eco-environmental deterioration issue in qinshui basin, shanxi province of china as illustration. wang (2014) conducted an indoor experiment of soil column simulated irrigation for a long time and finally drawn the conclusions: the produced water of south shizhuang cbm test area was with high-salinity, high-chloride and alkalinity characteristics, and the soil salinity increased, water retention capacity declined after a long-time irrigation, which resulted in massive crop failures. liu et al. (2014) selected water-formation samples, reservoir hydrops samples, bottom sediment samples of fanzhuang area for test, and confirmed that the heavy metal content of arsenic, cadmium, stannum, etc. in produced water is high. meanwhile, the topsoil around cbm wells were confronted with the contamination risk of heavy mentals when the accumulation, migration and enrichment of such kinds of chemical elements happened. hu et al. (2009) insisted that during the process of cbm production, a large amount of draining water from coal seams would lead to the contamination of surface water, the decline of groundwater level, the disruption of aquatic ecological environment, and the denudation of soil. li et al. (2014) investigated the region of zhengzhuang and suggested that the severely supracrustal and hypogeal disturbance during the construction and production period of cbm industry made great changes on regional underlying topography, which may result in the disruption of vegetation, the newly arose soil erosion, and the damage of climatope. besides, since the dominant component of cbm is methane (ch4), the greenhouse effect is 22 times over carbon dioxide (co2) (song et al., 2012), and the leak of cbm may bring the conditions of safety production and ecological civilization into the grave threat. therefore, we should strength the risk management of ecological environment on cbm industrialization development, whose premise is to reasonably and reliably assess such risks by quantitative approaches. america and canada early carried out the study the issues of the eco-environmental risk assessment on cbm exploitation. in the late twentieth century, beckstrom and ___________________________________________________________________________________________________________ 3 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 boyer (1993) analyzed the aquifer-protection works on cbm producing wells and deep cathodic-protection groundbeds. chafin et al. (1996) studied the effect of hydrofracture and the degradation of groundwater level on eco-environmental risk. u.s. environmental protection agency (epa) evaluated the impact of the hydrofracture of cbm reservoirs on the drinking water resource by stages, and respectively from three aspects: actual observation records, experimental data, and theoretical analysis. cheung et al. (2009, 2010) assessed the impact of the geochemistry of trace metals and rare earth elements in produced fluids or shallow groundwater in cbm producing regions of alberta, canada. dahm et al. (2013) utilized 3-d fluorescence spectroscopy to distinguish the naturally occurring organic from anthropogenic chemical compounds in the cbm produced water in order to recognize the risk of aquifer contamination. aguiler et al. (2014) analyzed the effect of “resource endowments, economics and the environment” in conventional, tight gas, shale and coalbed methane reservoirs. the studies in china mainly took qualitative methods, and the quantitative approaches are relatively few. for instance, pei et al. (2015) applied the evaluation model of pressure-state-response (psr) to construct the framework and index system of cbm ecological system. sun et al. (2015) set up an index system of ecological risk assessment on cbm industrialization development, based on the logic framework “hazard-induced environment – caused-hazards – hazard-affected bodies – disaster losses”. nonetheless, there is a lack of comprehensive and effective quantitative evaluation models. the reason is twofold. on one hand, it is hard to acquire the actual data, and there is much uncertainty in the obtained data information. as a result, the specialized database has not been set up that can efficaciously link among the environmental data from different regions. on the other hand, it is due to the lack of the unified evaluation index system and valid quantitative criteria. therefore, this paper plans to screen and optimize the index system proposed by sun et al. (2015), and then improve the quality of data processing with uncertain risk information by means of the intuitionistic fuzzy numbers transformed by interval values. ultimately, based on the mandani intuitionistic fuzzy neural network, an eco-environmental risk assessment model for the cbm industrialization development is constructed. 2. optimization of eco-environmental risk assessment index system sun and xue (2015) constructed an eco-environmental risk assessment index system of cbm industrialization development based on the theory of disaster system, and defined the eco-environmental risk of cbm, rcbm, as the set {hazard-induced environment, caused-hazards, hazard-affected bodies, disaster losses}, where the stability of hazard-induced environment ( e ), the risk of caused-hazards ( c ), and the vulnerability of hazard-affected bodies ( b ) are defined to represent the probability of eco-environmental risk while the damage of disaster losses ( l ) are designed to reflect the probability of risk losses. the combination of them constitutes the eco-environmental risk of cbm industrialization development. based on this model, the paper makes the following modifications: (1) transform some indexes into relative ones. for instance, use “proportion of economic loss” instead of “economic loss”. in doing so, the indexes become comparable and the calculation is simplified. (2) optimize the exposure level, adaptive capacity and anti-risk capability of hazard-affected bodies from three aspects: personnel, property and materials. for instance, in considering that with the increase of cbm resource exploration, the more hazard-affected bodies ( b ) are exposed in ecological risk and thus the greater probability of the risk could happen. hence, the index “resource exposure degree of cbm” is added. (3) specialize the artificial caused-hazard in caused-hazards. for instance, to assess the groundwater pollution and soil pollution, we can specialize in the impact of comprehensive hazard behaviors by means of “standard evaluation index of major groundwater pollution factors” and “nemero pollution index of soil”. based on the theory of rough set and rosetta software, the paper finally determines an index system by screening and optimizing evaluation indexes, which is shown in table 1. 3. quantification of risk assessment indexes 3.1. fuzzification method as shown in table 1, the ecological risk information itse lf has uncertainty in general. for example, in the practical ___________________________________________________________________________________________________________ 4 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 table 1. eco-environmental risk assessment index system of cbm industrialization development first level index second level indexes third level indexes 1e complexity of the hydrogeological condition 2e mechanization degree the stability of hazard-induced environment e 3e reliability of mechanical equipment 1c impact of bad weather 2c ratio of maximum ground-level concentration to standard concentration of major air pollutants 3c standard evaluation index of groundwater major pollution factors 4c nemero pollution index of soil the risk of caused-hazards c 5c percentage of forestry and grass coverage 1b resource exposure degree of cbm 2b ability of personal risk prevention the vulnerability of hazard-affected bodies b 3b investment proportion of prevention construction funds 1l proportion of economic loss 2l incidence of occupational diseases eco-environmental risk of cbm industrialization development cbm r the damage of disaster losses l 3l casualty rate risk assessment process, we may frequently encounter the situation that experts are hesitant in their judgments over risk information, or that experimenters are hard to obtain the data precisely due to the systematic error. with the emergence of intuitionistic fuzzy sets theory (atanassov, 1986), the nonmembership functions and hesitancy degree functions are added to general fuzzy numbers, which improves the quality of data processing and realizes the precise expression of fuzzy uncertainty information. the definition of an intuitionistic fuzzy set is as follows: definition 1 let x be a given universe of discourse, and an intuitionistic fuzzy set is defined by       , ,a aa x x x x x x      (1) where ( )a x  and ( )a x  are respectively the membership function and non-membership function of set a ~ , which respectively means the degree to which x belongs to a ~ and the degree to which x does not belong to a ~ such that 0 ( ) 1a x  , 0 ( ) 1a x  and 0 ( )a x  ( ) 1a x  . in addition, the hesitancy degree function is ( ) 1 ( ) ( ) a a a x x x       , and the smaller value corresponds to the more certain risk information. an intuitionistic fuzzy set is equivalent to an interval-valued fuzzy set (yuan 2013). considering the inequality ( ) 1 ( )a ax x    , the intuitionistic fuzzy set a~ can be transformed into an interval-valued fuzzy set using the following formula:      , ,1a a a aa x a x            (2) the membership function ( ) a x  can be of any type including gaussian, sigmoid or triangular function etc. we select the gaussian membership function in this paper in considering that the function has high smoothness. the gaussian membership function is defined by and lower bounds of the interval-valued fuzzy numbers respectively as shown in fig. 1. 2,( ) ( , ) exp ( )a xx f x               (3) where  and  represent respectively the curve center and curve width of the gaussian membership function. substituting (3) into (2), and combining with ___________________________________________________________________________________________________________ 5 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 the features of gaussian function, intuitionistic fuzzy numbers can be transformed into interval valued fuzzy numbers with the same curve center ij and different curve widths ij , as shown through the following (4) to (6):       iijiiji xxxa   ,~ (4) 2( ) exp[ ( ) ] , 1, 2 ; 1, 2i ijij i ij x x i n j m            (5) 2( ) exp[ ( ) ] , 1, 2 ; 1, 2i ijij i ij x x i n j m            (6) where ( )ij ix  and ( )ij ix  denote the upper in this way, fuzzy uncertainty information is theoretically expressed as gaussian intuitionistic fuzzy number so that the actual expression of fuzzy information is realized. 3.2. fuzzification of risk evaluation indexes following the method of fuzzification method in section 3.1, the indexes from table 1 can be fuzzified. (1) complexity 1 ~e of the hydrogeological condition is a key index of natural hazard-induced environment, which is determined by three factors: the stability of coal seam, the vulnerability of aquifer and the sensibility of groundwater environment. for its fuzzification, we refer to “if-then” risk decision rules (xue 2014). furthermore, we can obtain five output parameter values: very simple, simple, general, complex, very complex after the experts input the fuzzy linguistic variables of the above three factors as shown in table 2 and fig. 2. by section 3.1, 1 ~e is an intuitionistic fuzzy number with the gaussian membership function. (2) mechanization degree 2 ~e is determined by the utilization rate of mechanical equipments, the mine integrated unit yield of cbm and the informationization degree of management (xie et al, 2012). 2 ~e can be fuzzified by means of “if-then” rules presented in table 2. and fig. 2. fig. 1. transformation of a gaussian intuitionistic fuzzy number into an interval-valued fuzzy number. table 2. “if-then” risk decision rules input parameters output parameters coal seam stability aquifer vulnerability groundwater sensitivity hydrogeological condition complexity instable vulnerable sensitive very complex instable vulnerable sensitive very complex     stable invulnerable insensitive very simple ___________________________________________________________________________________________________________ 6 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 (3) reliability 3 ~e of mechanical equipmen -ts is another key index of anthropogenic hazard induced environment. we focus on the security attribute, i.e. the ability with which mechanical equipments can prevent from the environmental disruption, casualties and property loss. 3 ~e could be readily calculated by the following:  3 exp s me t    (7) where s ~ is the fuzzy failure rate of the equipments or the system which may cause the contamination accident, and mt ~ is the fuzzy time of fulfilling cbm productive tasks. (4) impact degree 1 ~c of bad weather is determined by three factors: the damage degree {strong, general, weak}, the time of duration {long, medium, short} and the affected areas {large, medium, small}. 1 ~c can be estimated by experts, and fuzzified in the same way as 1 ~e . (5) ratio 2 ~c of maximum ground-level concentration to standard concentration of major air pollutants is used for evaluating the impact of atmospheric environment. usually, 1 to 3 major pollutants are chosen and the ratio of each pollutant is calculated, and 2 ~c is the fuzzification of the maximum ratio:   3 1 2 01 max 100%m mmc c c     (8) where mc ~ denotes the fuzzy value of maximum ground-level concentration of the -m th pollutant, and 0 m c is the standard concentration value of ambient air quality (ministry of environmental protection of the prc, 2008). (6) the standard evaluation index 3 ~c of major pollution factors of underground water is used to estimate the impact on groundwater due to cbm project. the major pollution factors include each pollutant (water quality factor) whose ratio of accumulative pollution load is over 70%. 3 ~c could be computed as follows:   1-3 ~max~ skkk qqc  (9) where kq ~ denotes the fuzzy mass concentration of the -k th major pollution factor and skq ~ denotes the corresponding standard mass concentration(ministry of environmental protection of the prc, 2008). 1~3 c indicates that the monitoring value is over the stipulated one, and the larger value means the severer pollution. (7) the nemero pollution index 4 ~c of soil is used to estimate the comprehensive environmen tal quality which is computed using the following equation:     222 max 4 2 ~~ ~            avezzzz scsc c (10) where zc ~ stands for the fuzzy value of heavy metal content in soil, zs denotes the standard value of soil environmental quality,  max~ zz sc and  avezz sc~ indicates the maximum and average fuzzy values of a certain heavy mental pollutant respectively. (8) percentage 5 ~c of forestry and grass fig. 2. the complexity 1 ~e of hydrogeological conditions. ___________________________________________________________________________________________________________ 7 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 coverage is designed to estimate the situation of soil and water conservation and ecological functions, which is computed by the formula:   15 ~~  tfg sse (11) where fgs ~ denotes the fuzzy total area of vegetation, and t s denotes the affected area due to cbm project. (9) exposure degree 1 ~ b of cbm resource is computed by   1 1 ~~~   am qqb (12) where mq ~ denotes the fuzzy value of exploitation quantity cbm, and aq ~ denotes the fuzzy value of cbm recoverable reserves. (10) risk prevention ability 2 ~ b of personnel is to reflect the anti-risk and adaptive capabilities of hazard-affective bodies (personnel), including the employees’ professional quality, the situation of risk prevention skills training and the standardization of emergency plan, which are described by means of language values: good, medium, bad. then 2 ~ b is obtained in the same way as 1 ~e . (11) investment proportion 3 ~ b of prevention construction funds reflects the anti-risk and adaptive capabilities of hazard-affected bodies (property), which is defined by   13 ~~~   tcrp gfb (13) where rpf ~ denotes the ratio of risk prevention funds and tcg ~ is the total construction funds. (12) proportion 1 ~ l of economic loss reflects the economic loss caused by eco-environmental risk, which is calculated by    11 ~~~~   ovieldel gffl (14) where delf ~ is the direct economic loss, ielf ~ is the indirect economic loss and ovg ~ is the gross output value. (13) incidence 2 ~ l of occupational diseases indicates the injury caused by nonvital but persistent disaster losses, which is calculated by 1 2 ( )od wp pl    (15) where odp ~ is the number of patients with occupational disease and wp ~ is the total numbers of people in cbm production activities. (14) casualty rate 3 ~ l reflects the damage and loss caused by abrupt momentous injury accidents which is calculated by   13 ~~~  wcf ppl (16) where cf ~ p is the number of fatalities due to the pollution or destructive accident and wp ~ is the same as in formula (15) . 4. construction of comprehensive assessment model based on the fuzzy quantization in section 3, by inputting fuzzy rules and training samples obtained from expert system into mamdani intuitionistic fuzzy neural networks (mifnn) using comprehensive generated method (lei et al, 2014) and by adjusting the learning algorithm of the model, we can get the actual outputs of eco-environmental risk assessment in cbm industrialization development. 4.1. model structure the mifnn model is a fuzzy multilayer feed forward neural network system with multi-input and single-output structure. by referring to the hierarchical structure of self-organizing intuitionistic fuzzy neural network model proposed by xu et al (2010) and lei et al (2014) and combining with fuzzy control theory, the mifnn model can be described as a six-layer networks model as shown in fig.3. the first layer, the layer for input variables, inputs variable values into the fuzzy neural system. owing to the fuzzifier in mamdani model, the input values should be the exact values 1 3( , , ) te l . the second layer, the layer for fuzzy linguistic variables, is designed for inputting attribute functions (membership functions and non-membership functions). by formula (4) through (6), we denote the attribute functions by interval-valued gaussian membership functions, and the output values gained by each node are interval-valued fuzzy sets [ , ]ij ij    . then the vector of the fuzzy sets  tle 31 ~,,~  is obtained. ___________________________________________________________________________________________________________ 8 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 the third layer, the layer for inference rules, is designed for marching the antecedent of interval-valued fuzzy rules. each node in this layer represents an antecedent of the fuzzy rules, and the output value of the thj node is ,i i i     . meanwhile, in order to reduce the information loss, we replace the and  in the model respectively by the addition and multiplication operations in real numbers. 14 1 j ij i       , 14 1 j ij i       (17) the forth layer, the layer for output linguistic variables, outputs the consequent of fuzzy rules of each node in the upper layer. the output linguistic variable of the thj node is the interval-valued fuzzy set:      31 ~~,,~ leijijjjj    (18) the fifth layer, the layer for output processing, is defuzzification for which the center average defuzzifier is utilized. the output values are   rrr ,~ and r and r are defined by 1 1 u j j j u j j r            , 1 1 u j j j u j j r            (19) where u is the number of fuzzy rules. the sixth layer, the layer for output results, shows the risk assessment value cbm r of cbm industrialization development, based on the intuitionistic fuzzy sets theory of interval values and the defuzzification of center average defuzzifier. the exact value cbm r , is the satisfactory stack result of the input information from the risk assessment indexes in the network system. 2cbm r r r    (20) after the processing of six layers of network structure in the mifnn model, we obtain cbmr as 14 1414 14 1 1 1 11 1 14 14 1 11 1 ( ) ( ) 2 2 u u ij ij i ij ij i j i j ii i cbm u u ij ij j ji i x x r                               (21) meanwhile, according to the universal approximation theory, there must be a gaussian fuzzy logic system, in which any given function can be approached with any precision. the global approximation of mifnn model can be readily verified. 4.2. self-adaptive learning algorithm of the model with the mifnn model for eco-environmental risk assessment of cbm industrilization development, the correspondence between risk assessment indexes and the real output value cbmr is obtained after some repeated learning training. now assume that the desired output value is r . then the squared error function is 2 2 ( ) cbm p r r e   . substitute (5) and (6) into fig. 3. mamdani intuitionistic fuzzy neural network model. ___________________________________________________________________________________________________________ 9 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 (21) to obtain the transformed fuzzy system which possesses the curve center ij , the curve width ij (calculated by the average value of ij  and ij  ) of gaussian membership function and ij ( calculated by the average value of ij  and ij  ). by the algorithm of first-order gradient search, the adjusted variables ij , ij and ij can be expressed by ( 1) ( ) pij ij ij e n n          (22) ( 1) ( ) ( 1) )( p ij ij ij p ij ij ij e n n e n n                           (23) ( 1) ( ) ( 1) ( ) p ij p ij ij ij ij ij e n n e n n                           (24) then, we can obtain the calibrated output surface. 5. case study in this section, we utilize the sample data obtained by laboratory simulation to analyze the reasonability feasibility and superiority of the mifnn model in section 4. 5.1. data sources and preprocessing our data come from the corresponding scenario simulations and experts’ experiences. the fuzzy rules and training samples are obtained from the comprehensive generated method. the input-output relation surfaces are obtained after repeated learning trainings by matlab software, which are shown in fig. 4. these 3-dimensional relation surfaces are relatively smooth, which verifies the reasonability of fuzzy rules. 5.2. results analysis now we analyze the simulation results of the model through 50 groups of cases which are （a）ecological risk-occupational disease-casualties （b）ecological risk-groundwater-soil contamination （c）ecological risk-personnel prevention-air contamination （d）ecological riskprecaution funds-personnel prevention fig. 4. relation surfaces selected randomly and partial input and output values are shown in table 3. take the no.3 group of data in table 3 as an example. the attribute characteristics of eco-environmental risk in cbm industrialization development in this group are: the hydrogeological condition is complex (the stability of coal seam is medium, ___________________________________________________________________________________________________________ 10 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 the vulnerability of aquifer is medium, and the sensitivity degree of the groundwater environment is sensitive); the degree of mechanization is high (the utilization rate of mechanical equipments is high, the mine comprehensive yield of cbm is high, and the informationization degree of management is low); the average failure rate of mechanical equipments is 4 16.25 10 h  , and they can perform production tasks normally within 7 24h ; the impact of bad weather is low (the damage degree is weak, the time of duration is medium, and the affected area is small); the ratio of the maximum ground-level concentration to the standard concentration of major air pollutants is 73%; the standard index evaluation of groundwater major pollution factors is 1.15; the nemero pollution index evaluation of soil is 1.45; the percentage of forestry and grass coverage is 18.3%; the resource exposure degree of cbm is 23.3%; the ability of personal risk prevention is good (the employees’ professional quality is medium, the situation of risk prevention skills training is good, and the standardization of emergency plan is medium); the investment proportion of prevention construction funds is 1.78%; the proportion of economic loss is 2.65%; the incidence of occupational diseases is 6.1%; and the casualty rate is 0.96%. calculated with the mifnn model, the risk assessment value is 0.4574, which is the medium risk level as anticipated, which indicates that the eco-environment is mildly contaminated, and the situation is manageable and can be controlled. the administrators should strength the supervision and precaution. table. 3. the input and output values of attribute parameters in cbm eco-environmental risk num. 1e 2e 3e 1c 2c 3c 4c 5c 1b 2b 3b 1l 2l 3l outputs 1 0.83 0.67 0.12e 0.33 39 0.86 0.67 21.5 13.3 0.50 2.52 1.76 4.33 0.64 0.2639 2 0.83 0.67 0.18e 0.33 80 1.27 1.52 16.4 36.7 0.67 1.01 2.96 9.96 1.63 0.5215 3 0.67 0.67 0.11e 0.17 73 1.15 1.45 18.3 23.3 0.67 1.78 2.65 6.10 0.96 0.4574 4 0.83 0.50 0.32e 0.50 86 1.48 1.83 7.7 33.3 0.33 0.94 3.83 8.29 1.41 0.6605 5 0.50 0.83 0.03e 0.17 57 0.76 0.95 23.9 16.7 0.83 2.62 1.85 5.53 0.92 0.1517 … … … … … … … … … … … … … … … … 46 0.83 0.67 0.06e 0.33 55 1.12 1.72 13.9 26.7 0.67 2.03 2.61 5.31 0.90 0.4050 47 0.67 0.67 0.19e 0.33 87 2.35 3.01 6.6 50.0 0.33 0.82 6.17 13.22 2.19 0.8982 48 0.83 0.67 0.21e 0.50 63 0.93 0.97 15.3 23.3 0.50 2.41 2.35 6.39 1.06 0.3046 49 0.50 0.83 0.17e 0.33 45 0.82 0.86 16.5 16.7 0.50 2.37 2.23 5.28 0.89 0.2797 50 0.83 0.50 0.32e 0.67 92 1.89 2.23 9.8 40.0 0.17 0.56 4.85 10.35 1.74 0.7992 5.3. comparative analysis with the fnn model in order to illustrate the superiority of the mifnn model, we compare it with the fuzzy neural networks (fnn) model. to increase the comparability, we employ the simulation cases in table 3 and apply the fnn model. the comparative analysis is based on the upper and lower bounds of the interval values in formula (4)-(5). fig.5 shows that both the mifnn and fnn models could reflect the level of eco-environmental risk assessment in general. however, some values obtained by the fnn model are out of the interval values of intuitionistic fuzzy sets, e.g. the data in the no.13 and no.41 groups. in view of this, the further analysis for the evaluation quality of the both models is needed. calculate respectively the degree of dispersion （ mifnn ， fnn ）and the coefficient of variation ( mifnncv ， fnncv ) which are shown in table 4. ___________________________________________________________________________________________________________ 11 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 table. 4. the evaluation comparison of mifnn and fnn mifnn fnn numeric comparison result analysis  0.0507 0.0646 mifnn fnn   the stability of mifnn model is superior to fnn model cv 0.4940 0.5712 mifnn fnn cv cv the accuracy of mifnn model is superior to fnn model it can be seen from fig.5 and table 4 that the stability and accuracy of the mifnn model are superior. therefore, it is more suitable for the eco-environmental risk assessment in cbm industrialization development. 6. conclusions this paper provides a method of eco-environ -mental risk assessment during cbm industrialization development based on the mifnn model with both nonlinearity and global-approximation characteristics, by optimizing the index system and utilizing atanassov intuitionistic fuzzy sets to improve data quality. the rationality and validity of the model are verified by simulation cases. it can be concluded that the expected risk level of eco-environment is in conformity with the assessment results. the comparative analysis indicates that both the mifnn and fnn models are effective in assessing the risk level of eco-environment; however, the former model is superior to the latter as far as stability and accuracy are concerned. therefore the application of the mifnn model on empirical study will provide a decision basis for ecological supervision and risk precaution and management. nevertheless, the number of attribute indexes in the model is relatively large, which results in some defects on precision, though we effectively adjust parameters in fuzzy rules and training samples. therefore, we will further improve the risk assessment model by trimming and optimizing fuzzy inference rules and taking multidisciplinary complementary advantages in future. meanwhile, we will develop evaluation software to implement the computer management of eco-environmental risk assessment on cbm industrialization development. acknowledgements this work was supported by the national science foundation for young scholars of china (grant no. 41101507); the national science foundation of china (grant no. 71373170); the fig. 5. the comparison of inference results between mifnn and fnn. ___________________________________________________________________________________________________________ 12 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 program for the philosophy and social sciences research of higher learning institutions of shanxi (pssr), china (grant no. 2017314); the program for the top young academic leaders of higher learning institutions of shanxi (tyal), china; and by the soft science foundation of shanxi province, china (grant no. 2017041025 -2). references j. wang, irrigation effects on soil and plant by using coal bed mathane produced water in qinshui basin, china university of mining & technology, (2014) h. h. liu, s. x. sang, l. w. cao, et al. heavy metal pollution feature analysis and potential ecological risk assessment of the coalbed methane production on the topsoil quality of the mining area, journal of safety and environment, (3) (2014) 288-293. l. w. hu, h. x. chen, w. g. yang, study on environmental impact and management of coal bed methane development, environmental protection science, (6) (2009) 51-54. y. y. li, j. w. kang, analysis of potential ecological influence during grounding mining of coal bed methane, china coal, (3) (2014) 24-127. y. song, x. m. zhang, s. b. liu, et al. basic theory of coalbed methane geology and exploitation in china, science press, (2012). j. g. li, study on simulation and risk assessment of leakage and diffusion of coalbed methane field gathering and transportation, china university of mining & technology, (2014). j. a. beckstrom, d. g. boyer, aquifer-protection considerations of coalbed methane development in the san-juan basin, society of petroleum engineers formation evaluation, 8(1) (1993) 71-79. d. chafin, d. swanson, d. gery, methane-concentration and methane-isotope data for groundwater and soil gas in the animas river valley. colorado and new mexico, (1996). us environmental protection agency. study design for evaluating of impacts to underground sources of drinking water by hydraulic fracturing of coalbed methane reservoirs. washington dc, 2001. us environmental protection agency. evaluation of impacts to underground sources of drinking water by hydraulic fracturing of coalbed methane reservoirs. washington dc, 2004. k. cheung, h. sanei, p. klassen, et al. produced fluids and shallow groundwater in coalbed methane (cbm) producing regions of alberta, canada: trace element and rare earth element geochemistry, international journal of coal geology, 77 (2009) 338-349. k. cheung, p. klassen, b. mayer, et al. major ion and isotope geochemistry of fluids and gases from coalbed methane and shallow groundwater wells in alberta, canada, applied geochemistry, 25 (2010) 1307-1329. k. j. dahm, c. m. van straaten, j. munakata-marr, et al. identifying well contamination through the use of 3 d fluorescence spectroscopy to classify coalbed methane produced water, environmental science and technology, 47 (2013) 649-656. r. f. aguilera, r. d. ripple, r. aguilera, link between endowments, economics and environment in conventional and unconventional gas reservoirs, fuel, 126 (2014) 224-238. y. t. pei, l. k. zhang, j. k. he, et al. ecological effect evaluation of cbm development based on psr framework model, environmental protection of oil & gas fields, (1) (2015) 25-27. w. sun, y. xue, x. j. ren, the index system of ecological risk assessment based on the industrialization development of coal-bed methane. in: huang, c.f. emerging economies, risk and development, and intelligent technology: proceedings of the 5th international conference on risk analysis and crisis response. morocco: crc press, (2015) 495-502. p. j. shi, theory and practice on disaster system research in a fifth time, journal of natural disasters, (5) (2009) 1-9. y. xue, the study of soft hierarchical model on integrated risk assessment for natural disaster. meteorological press, (2014). k. atanassov, intuitionistic fuzzy sets, fuzzy set sys, 20(1) (1986) 87-96. x. h. yuan, h. x. li, j. j. song, interval-valued level cut sets, decomposition theorems and representation theorems on interval-valued fuzzy sets. fuzzy systems and mathematics, (5) (2013) 1-10. h. p. xie, j. h. wang, b. h. shen, et al. new idea of coal mining: scientific mining and sustainable mining capacity, journal of china coal society, (7) (2012) 1069-1079. j. feng, q. sun, p. c. luo, et al. equipment reliability and comprehensive security, national university of defense technology press, changsha, hunan, china, (2008). china's ministry of environmental protection. guidelines for environmental impact assessment: atmospheric environment. environmental science press, beijing, china, 2008. y. j. lei, j. zhao, y. l. lu, et al. intuitionistic fuzzy set theory and application. science press, beijing, china, (2014). x. l. xu, y. j. lei, w. b. xie, self-organising intuitionistic fuzzy neural networks based on ukf. acta electronica sinica, (3) (2010) 638-645. ___________________________________________________________________________________________________________ 13 journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 3–13 microsoft word editors’ introduction for volume 3, issue 4 editors’ introduction modern society is a competitive society where reducing existent costs is very important, which also is the ultimate goal of risk analysis and crisis response. in the present issue such efforts have been embodied in these papers. this issue contains six papers. the first contribution is written in english and the last five in chinese with english abstracts. the papers can be divided into four categories: crisis management, risk assessment, time series prediction and cloud model assessment technique. the crisis management category includes the paper “gezi park crisis” by ali ekşi. gezi park is a public park located in the center of istanbul, turkey. the objective of this article is to evaluate the gezi park incident which started on 28 may 2013 and continued until the end of june in terms of crisis management. the reasons why a civil opposition act that started against the decision taken by the government regarding the environment transformed into a serious crisis due to an unmanageable process by the government will be questioned. the reasons for the incident and why it could not be prevented will be evaluated from the perspective of crisis management. problem areas focusing especially on crisis communication during the crisis management process, the effect of messages given by the administrators along with the use of disproportionate force in deepening the crisis along with the lessons learned from the crisis will be evaluated. the most important factors that transformed the events starting with representation problem into a crisis have been communication problems and the disproportionate force used on the protesters. turkey should develop a proactive approach in crisis management and should develop permanent solutions to solve the communication problem. the study is important since this was the first time that an environmentalist action transformed into a social movement causing a serious crisis in a country like turkey. the risk assessment category includes two papers. the first paper is “dynamic vulnerability analysis of population for flood disaster in urban area” by xilei pang. in this paper, based on information diffusion theory and incomplete information, the changes of the affected population in urban flood is analyzed and the function relationship between the hazard (rainstorm) and disaster-bearing body (population), which is called vulnerability curve of population in flood disaster, is identified. this curve not only can present some changing extent of urban flood risk, but also could provide an important basis for integrated risk assessment of urban flood. the second paper “internal exposure simulation based on exposure related dose estimating model” by qian zhang, deyin huang, mao liu and minyan li. in this paper, occupational chemical poisons’ physiologically based pharmacokinetic (pbpk) models are established to simulate metabolic processes in vivo based on exposure simulation software-exposure related dose estimating model (erdem). pbpk model establishing progress is described based on the development and utilization of erdem. the time series prediction category includes one contribution of “wd-rbf model and its application of hydrologic time series prediction” by dengfeng liu, dong wang, yuankun wang, lachun wang and xinqing zou. a method of radial basis function network based on wavelet de-nosing (wd-rbf) was proposed according to the nonlinear problem and noise in hydrologic time series. wavelet coefficients of each scale were calculated through wavelet transform; soft-threshold was used to eliminate error in series. reconstructed series were predicted by rbf network. the simulation and journal of risk analysis and crisis response, vol. 3, no. 4 (december 201 ), 156-157 published by atlantis press copyright: the authors 156 prediction of wd-rbf model were compared with arima and rbf network to show that wavelet de-nosing can identify and eliminate random errors in series effectively; rbf network can mine the nonlinear relationship in hydrologic time series. examples show that wd-rbf model has superiority in accuracy compared with arima and rbf network. the cloud model assessment technique category includes two papers. the first paper is “cloud model assessment research of bijie drought risk” by ying he, mu zhang and wei li. bijie area locates in the northwest of guizhou, china. in this paper, eight cities and counties of bijie area are chosen as the research target. with the research foundation of natural disaster risk theory and drought risk formation principle, we start from dangerousness, exposure , vulnerability and the ability to prevent disaster to filter out 23 indexes range from social status, economy, weather and geology to construct the assessment system of bijie drought risk. to solve the uncertainty of quantitative description for drought risk assessment index and the judge of assessment result, cloud model and entropy weight method are implemented to decide how much percentage the city or county belong on each index. in the end, we use the weight of 23 indexes to get the drought risk level each city or county belongs to. the last contribution “water security risk assessment of the yellow river basin based on conserained-random weight and cloud mode” by chenye song, ren zhang, aixia zhou, et al. in this paper, to overcome the difficulty of the uncertainty hardly fusing into risk assessment outputs, a new risk assessment technique of combining constrained-random weight and cloud model is introduced to perform the risk assessment of water security in the yellow river basin. the effect of knowledge deficiency on risk assessment can be illustrated on the instability of risk series by the constrained-random weight method; the result of assessment can incorporate the randomness and fuzziness of risks and corresponding assessments, and transition of quantified risk series into qualitative risk level can be possible as the presence of the cloud model. experiment about the water security risk assessment of the yellow river baisin proved the validity and applicability of this method. kindly thanks to all of referees and reviewers who have given strong help and support to jracr. and we also thanks for the authors very much for their contributions and revising. guest editors dong wang professor, nanjing university email: wangdong@nju.edu.cn ren zhang professor, pla university of science and technology email: zren63@126.com published by atlantis press copyright: the authors 157 atlantis press journal style introduction for volume 8, issue 2 this issue contains 7 papers. there are 3 contributions written in english and 4 contributions in chinese with english abstracts. the papers can be divided into five topics: emergency management, security risk, geological hazard analysis， real estate risk management and financial risk analysis. there are two papers in emergency management. the first paper“review and prospect of emergency logistics under uncertainty conditions” by hanping zhao, et al., reviews the mainly research of emergency logistics under uncertainty conditions in recent years. the characteristics of uncertainty appearing in emergency logistics process were analyzed, and the expression of uncertain factors, demand forecasts and objective function definition in decision-making optimization model were summarized; moreover, this article analyzed the key points of research in aspects of emergency logistics operations, uncertainties origination and emergency risks, and provided useful reference and directions for future research. in the second paper “capability evaluation of ground support for emergency rescue helicopters” by xiuyan zhang, et al., the evaluation index system is established, including three level indicators by defining the connotation and identifying the influencing factors. the weights of the indexes are determined by means of questionnaires. the evaluation model is established by using fuzzy comprehensive evaluation method. finally, the feasibility of the evaluation model is verified taking a helicopter airport in china as an example. the verification results show that the model is beneficial for the government to fully understand the present situation of the ground support for emergency rescue helicopters, and also to provide direction for the related ground support organizations to improve their ground support capability. there are two papers in security risk. the paper “using trust game for cross-strait security risk analysis” by cheng-kuang wu and chongfu huang, simulated a mixed strategy nash equilibrium. experimental results show that the china’s expected payoffs of n.e. fix and taiwan’s expected payoffs of n.e. are increased if the payoffs of chinataiwan bilateral trust are increased. the chinese mainland released new measures on exchanges and cooperation with taiwan. taiwan accepts these measures results in favor of taiwan’s expected payoffs and decreases the level of risk. in the second paper “research on promoting effect of tourism industry development on tourism poverty alleviation in guizhou” by siqi li, et al., the tourism industry of 50 poverty-stricken counties in guizhou province is selected as the representative, and the anti-poverty effect of guizhou tourism industry is measured by using panel data model. the empirical results show that the development of tourism industry in 50 poverty-stricken counties in guizhou has promoted the development of the national economy, thus reducing the poverty rate of the counties. it is concluded that guizhou tourism industry has a certain promoting effect on the precision poverty alleviation in guizhou. there is one paper in geological hazard analysis. the paper “geomorphology characteristic and tectonic response of the meijiang watershed in the southern china represented by hypsometric integral”by qinghua gong, et al., take meijiang river basin as study area because it is a typical representative in the mountains of south china. firstly, the study area was classified into several catchments in arcgis platform based on dem data. secondly, this paper withdraws the measuring indicators of the geomorphology and hypsometric integral from the basins. they studied the information of geological structure and development of disaster situation. finally, revealing the relationship between hypsometric integral and tectonic activity, lithological association and disaster intensity. it provides an important theory and method for studying geomorphic spatial pattern and characteristics of development as a whole. the result indicates that hypsometric integral has a spatial scale effect, geomorphology, lithology and disaster development can be simulated by small watersheds. for one thing, the studying area is in the mature phase, which is a high-incidence area with many kinds of mountain disasters, for another thing, the erosion resistance of biotite monzonitic granite is the most strongly. on the contrary, the sandstone, conglomerate, glutenite, siltstone, mudstone and limestone of it are weakly. at last, regional tectonic activity can be reflected by the hypsometric integral of big watershed units. 61 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 61-62 there is one paper in real estate risk management. the paper“project portfolio management strategies in housing estate development organizations in nigeria’s built environment” by sadiq gumi abubakar, et al., aims to assess the ppm strategies used by housing estate development organizations in nigeria’s built environment with a view of identifying, examining, and highlighting the impact of these strategies on housing estate development project portfolios in nigeria. literature reviewed led to identification and assessment of five ppm strategies. krejcie and morgan table was used to determine the sample size, while cronbach’s alpha, mean item scores, relative importance index, t-test statistics were used for data analyses. the results show that any chosen ppm strategy(ies) by housing estate development organizations will not be easy to apply, use, implement nor will it be excellent in the strive to achieve the organizational objectives in any given portfolio. there is one paper in financial risk analysis. the paper “simulation study on evolutionary game model between technological small and medium enterprises and banks under verification system” by huafeng chen & mu zhang, adds risk compensation funds and establishes an evolutionary game model for technological small and medium enterprises (smes) and bank loans under the verification mechanism based on the evolutionary game theory. and the stability analysis and simulation study of the game results of the four evolutionary strategies of the model are carried out. finally, we have put forward some suggestions on the strategy selection and the development of the loan system. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca 62 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 61-62 atlantis press journal style introduction for volume 7, issue 4 this issue contains 5 papers. there are 3 contributions written in english and 2 contributions in chinese with english abstracts. the papers can be divided into three topics: financial risk, disaster reduction & emergency management, and animal infectious disease, there are two papers in financial risk. the first paper“research on evaluation of equity financing efficiency of listed companies in strategic emerging industries” by yaxi huang & mu zhang, chooses 198 listed companies in strategic emerging industries, using dea model to study the efficiency of equity financing, and carries on efficiency analysis, investment redundancy and output shortage analysis and industry comparative analysis. the results show that the efficiency of equity financing of listed companies in strategic emerging industries is inefficient. the comprehensive efficiency, pure technical efficiency and scale efficiency are 0.370, 0.603 and 0.563. from the scale pay, the economic scale of chuanrungufen should be increased, zhongguobaoan and other 179 decision-making units should be reduced; dongxulantian and other 169 decision-making units have different levels of input redundancy and lack of output; equity financing efficiency is unevenly developed between different industries. the second paper “wenzhou real estate bubble and rupture risk warning” by anping pan, chongfu huang & dehua jiang, reviews the trend of housing price in wenzhou from 2006 to 2015, and analyzes the causes of house bubble. in addition, the adjustment process of wenzhou real estate is of typical significance, which could provide references for current market trend. there are two papers in disaster reduction and emergency management. the first paper “assessment of earthquake prevention and disaster reduction capability of county-level administrative units in gansu province” by jin chen, wen li, wenkai chen, et al, establishes the assessment indicator system for the earthquake prevention and disaster reduction capability of county-level administrative units. an analytic hierarchy process is applied to determine the indicator weights, and the environmental supporting capability, infrastructure supporting capability and resource supporting capability of different counties in gansu in the handling of earthquake disasters are calculated to obtain the assessment figure for the earthquake prevention and disaster reduction capability of county-level administrative units in gansu. the research results indicate that anning district of lanzhou has the strongest earthquake prevention and disaster reduction capability while zhouqu county possesses the weakest earthquake prevention and disaster reduction capability. the second paper “research on the emergency shelter accessibility in urban communitie” by bo tang， danni wang, yun song, et al, analyzes the accessibility of emergency shelter in yuexiu district under microcosmic perspective by using the improved two-step floating catchment area method and gis. the conclusions are as follows: firstly, the total amount and type of emergency shelter are abundant in yuexiu district, per capita emergency shelter is high, but spatial distribution is unreasonable；secondly, the overall accessibility of emergency shelter is poor, north region better than south region, and supply and demand ratio is an important factor about the difference of accessibility；thirdly, when d0 to 1500m, emergency shelter accessibility reaches the maximum and proportion of service area is the best. the related conclusions can provide some reference for the optimization and planning of emergency shelter in yuexiu district. there is one paper in disaster risk assessment, which is “risk assessment of canine distemper in the distribution area of giant panda in sichuan, shaanxi and gansu provinces, china” by weigeng shao, feng jiang,liya huang,et al, where the authors used maxent model and combined with arcgis analysis to predict the potential risk of canine distemper to giant panda habitat in sichuan, gansu and shaanxi. the results showed that 35.05% and 19.47% of the giant pandas were in the high risk and medium risk of canine distemper, respectively. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 178 ___________________________________________________________________________________________________________ 178 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <feff04180441043f043e043b044c04370443043904420435002004340430043d043d044b04350020043d0430044104420440043e0439043a043800200434043b044f00200441043e043704340430043d0438044f00200434043e043a0443043c0435043d0442043e0432002000410064006f006200650020005000440046002c0020043c0430043a04410438043c0430043b044c043d043e0020043f043e04340445043e0434044f04490438044500200434043b044f00200432044b0441043e043a043e043a0430044704350441044204320435043d043d043e0433043e00200434043e043f0435044704300442043d043e0433043e00200432044b0432043e04340430002e002000200421043e043704340430043d043d044b04350020005000440046002d0434043e043a0443043c0435043d0442044b0020043c043e0436043d043e0020043e0442043a0440044b043204300442044c002004410020043f043e043c043e0449044c044e0020004100630072006f00620061007400200438002000410064006f00620065002000520065006100640065007200200035002e00300020043800200431043e043b043504350020043f043e04370434043d043804450020043204350440044104380439002e> /sky <feff0054006900650074006f0020006e006100730074006100760065006e0069006100200070006f0075017e0069007400650020006e00610020007600790074007600e100720061006e0069006500200064006f006b0075006d0065006e0074006f0076002000410064006f006200650020005000440046002c0020006b0074006f007200e90020007300610020006e0061006a006c0065007001610069006500200068006f0064006900610020006e00610020006b00760061006c00690074006e00fa00200074006c0061010d00200061002000700072006500700072006500730073002e00200056007900740076006f00720065006e00e900200064006f006b0075006d0065006e007400790020005000440046002000620075006400650020006d006f017e006e00e90020006f00740076006f00720069016500200076002000700072006f006700720061006d006f006300680020004100630072006f00620061007400200061002000410064006f00620065002000520065006100640065007200200035002e0030002000610020006e006f0076016100ed00630068002e> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word assessment of covid-19 epidemic control efficiency based on sirs model journal of risk analysis and crisis response, 2021, 11(3), 129-138 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i3.302 129 article assessment of covid-19 epidemic control efficiency based on sirs model kai xu 1,*, xiaofeng xie 2, xing hang 1, fengying zhang 2, qian qian 3, zongfang zhou4, dongyang li 1 and yu jiang 1 1 business school, chengdu university, chengdu (610106), sichuan, china 2 west china school of nursing /west china hospital, sichuan university, chengdu (610041), sichuan, china 3 school of business, sichuan normal university, chengdu (610101), sichuan, china 4 school of management and economics, university of electronic science and technology of china, chengdu (610054) sichuan, china * correspondence: xukai@cdu.edu.cn received: june 24, 2021; accepted: october 27, 2021; published: october 30, 2021 abstract: based on the sirs epidemic model embedded in complex network theory and the covid19 spreading characteristics, the influence of prevention & control and treatment on the contagion of covid-19 and the stability of social network is analyzed separately in this paper. the results show that the contagion of covid-19 leads to the risk stability of social network. the number of infected persons is decreased by prevention & control and treatment which drives social network to risk-free stability. the treatment is more effective than prevention & control against covid-19. compared with prevention & control, treatment can make social network more risk-free and stable faster. keywords: control efficiency; covid-19 epidemic; sirs model; social network stability 基于 sirs 模型的新冠肺炎疫情管控效率评估 徐 凯 1,*，谢小凤 2，杭 行 1，张凤英 2，钱 茜 3，周宗放 4，李东阳 1，江 宇 1 1 成都大学 商学院，四川 成都 610106 2 四川大学 华西护理学院/华西医院，四川 成都 610041 3 四川师范大学 商学院，四川 成都 610101 4 电子科技大学 经济与管理学院，四川 成都 610054 * 通讯作者: xukai@cdu.edu.cn 摘 要：文章基于嵌入复杂网络理论的 sirs 传染病模型，结合新冠肺炎疫情蔓延所呈现的特征，分 别分析了防控和救治措施对新冠肺炎病毒传染和社会网络稳定性的影响。研究结果表明：新冠肺炎病 毒的传染导致社会网络最终趋于有风险稳定，通过防控救治可以降低感染人群数量，驱使社会网络达 到无风险稳定；相较于防控，救治对新冠肺炎病毒传染的遏制具有更高的效率；且相较于防控，救治 能更快的使社会网络趋于无风险稳定。 关键词：管控效率；新冠肺炎疫情；sirs 模型；社会网络稳定性 1. 引言 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 130 据世界卫生组织（who）数据显示，截至 2021 年 6 月 17 日，全球 168 个国家和地区报告确诊新 型冠状病毒肺炎（以下简称新冠肺炎）病例达 17669.3988 万，累计死亡超过 383 万人[1]。近半年来， 中国大陆新冠肺炎患者多为境外输入或与其密切接触者，充分表明中国新冠肺炎疫情已进入相对稳定 状态，故全国范围内已在 2020 年夏秋之际全面复学复工复产；但随着我国大部分地区寒冬的来临，特 别是近期的河北、辽宁和黑龙江等地，新冠肺炎病毒卷土重来，疫情防控形势依然严峻。早在 2020 年 初，为了阻断病毒随人口迁徙而迅猛扩散，从中央到地方，中国各级政府迅速采取空前的限行和隔离 防控措施（以下简称防控）。各地相继启动重大突发公共卫生事件 i 级响应，投入巨大人力和物力，对 来自或路经湖北人员跟进排查，对疑似或确诊患者进行隔离观察或者治疗，疫情最严重的湖北省实施 了严格的出行限制，全国城镇陆续关闭了公共交通，实施了严格的封闭管理，全社会自觉在家隔离， 戴口罩出门，人员聚集场所关闭，企业延期复工，学校延期开学，春节假期延长。上述措施有助于隔断 病毒传播，降低了疾病传染率，同时经过专门定点医院药物和器械治疗（以下简称救治），疫情得到有 效缓解，并于 2020 年 3 月份始逐步趋于稳定。 目前，国内外学者已对新冠肺炎防控、患者救治以及对经济的影响进行了卓有成效的研究。jia 等 [2]基于人口移动数据，构建人口流动-风险源模型，为疫情早期的应急管理和相关决策提供依据。wu 等[3]运用青海鼠疫模型，研究青海新冠肺炎传染及患者治疗情况，发现青海采取的预防和控制措施很 有效。shen 等[4]介绍了中国加强人员管理、个人防护、环境清洁消毒和健康教育等提高公共交通安全 的预防和控制新冠肺炎的经验。rutayisire 等[5]针对非洲医疗资源缺乏的困难，提出合作预防和控制 是消除新冠肺炎的希望路径。liu 等[6]的研究为新冠肺炎大流行期间，其他国家的外科医生和卫生保 健提供了决策参考。tartari 等[7]研究表明有效隔离是防控的关键，并强调国际统一个人防护的重要性。 wang 等[8]认为新冠肺炎病毒在特定环境中传染性更强，并分别为疗养院、孤儿院和监狱的健康管理、 个人防护和预防措施提供了实际指导。白宁等[9]基于非自制动力学方程，对湖北省疫情进行预测，并 评估了相应控制策略的有效性，揭示了对密切接触者的追踪隔离是最有效的疫情控制措施。潘理虎等 [10]运用多智能体构建病毒防控模型，分析病毒传播趋势，为城市疫情防控提高决策支持。yang 等[11]、 wang 等[12]、陈彬等[13]、朱翌民等[14]、孙皓宸等[15]和尹楠[16]基于 sir 模型，分别从不同侧面对 疫情走势进行研究，并对校园、工作场区和市中区等重点区域的疫情传播风险进行了预测与防控措施 评估。喻孜等[17]和 li 等[18]对中国政府行政干预和防控疫情的效果进行了初步评估。朱启荣等[19]实 证研究了新冠疫情对我国宏观经济和产业的影响。nishi 等[20]从社交网络的视角，研究控制病毒传播 和维持经济发展之间的平衡问题，为政策制定者提供参考，同时强调了戴口罩和消毒等行为习惯可有 效降低传染规模。上述有关新冠肺炎疫情的研究，为我们提供了思路，激发了笔者对疫情防控救治效 率的思考。而现有新冠肺炎疫情的研究中，鲜见针对限制出行、隔离防控和专门救治效果的综合性评 估文献。因此，有必要重点针对政府限行隔离防控和专门救治新冠肺炎的效率进行科学评估，研究疫 情发展趋势，分析防控和救治措施对疫情稳定状态的影响。 2. 研究设计 2.1. 基本假设 随着通讯技术的深入发展，社会越来越呈现网络化的趋势，某个区域如学校、医院、公司或村落 等成员聚集，形成众多区域社会网络，区域社会网络彼此相链接，形成网络社会。现有研究表明，网络 社会中，新冠肺炎病毒传播的主要途径是呼吸道飞沫和密切接触，相对密闭或污染环境中也存在气溶 胶或接触传播的可能[21]。为讨论方便，本文把病毒传播途径归类为飞沫和接触传播，给出如下几点假 设： 假设 1：网络社会中，以人与人之间飞沫或接触关系为边，以每个人为节点构成社会网络。在社会 网络中，节点总数量保持不变，记为 n，人群可以被划分为三类：易感者 s，感染者 i，康复者 r。 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 131 （1）易感者 s：社会上各年龄段人群对新冠肺炎病毒没有免疫力，普遍容易感染，故均为易感者， 即普通社会网络环境中，新冠肺炎病毒经飞沫和密切接触等传染的人群。 （2）感染者 i：有新冠肺炎疫情爆发区域旅居史者或与爆发区域旅居者有密切接触者或与潜在新 冠肺炎感染者有密切接触者，且出现发热、乏力、干咳，逐渐显现呼吸困难等症状者，称为感染者，包 括官方披露的确诊者和疑似者。 （3）康复者 r：感染者接受专业门诊医院治疗一段时间之后，免疫力得到增强，体温、呼吸道和 肺部炎症等符合出院标准，称为康复者，康复者源于易感者和感染者。 假设 2：感染者通过飞沫传播或密切接触易感者，使易感者以概率 变为感染者。称 为新冠肺 炎病毒的传染概率，表征了新冠肺炎病毒在社会网络中的传播扩散速度。 随着新冠肺炎病毒防控技术的提高，以及防控的进一步落实，易感者中部分人员抵抗病毒的免疫 能力得到提升。并且，由于医疗条件的限制，少部分感染者未能如期入院治疗，居家隔离用药而逐渐 康复。一般来讲，相较于感染者，易感者更易转变为康复者。为不失一般性，假设如下： 假设 3：易感者以概率 转化为康复者，感染者以概率  转化为康复者，其中   。 对于社会网络中的感染者，为降低其病毒传播的危害及对社会网络的冲击，定点医院收受治疗。 大多新冠肺炎病毒患者经过医院治疗能够康复。不失一般性，假设如下： 假设 4：如果通过专业门诊医院救治，比例为 的感染者转化为康复者，称 为新冠肺炎病毒传播 的救治率， 代表救治对新冠肺炎病毒传播的管控效率。 对于社会网络中的易感者，可以通过切断或远离与感染者或潜在感染者的接触等防控，避免飞沫 传播或密切接触，从而提高自身的免疫能力，直接转化为康复者。不失一般性，假设如下： 假设 5：如果通过限制出行和隔离防控，比例为 的易感者转变为康复者，称 为新冠肺炎病毒传 播的防控率， 表征了防控对新冠肺炎病毒传播的管控效率。 由于企业复工、学生复学、防控疲劳和境外输入等外部环境的变化，以及康复者自身保护意识的 淡化，康复者对新冠肺炎病毒的免疫能力可能会降低，重新转变为易感者，故假设如下： 假设 6：康复者以概率 ，重新转变为易感者。 2.2 建立模型 由 n 个人构成的社会网络中，网络的节点为每个人，网络的边为各节点间的飞沫或接触关系。节 点i 的度 ik 定义为与该节点相连的其他节点数目，网络中所有节点i 的度 ik 的平均值称为网络平均度， 记为 k  ，  ip k 表示随机选定一个节点的度为 ik 的概率，可得   1 n i i i k k p k    。由假设 2 至假设 5，在社会网络中，易感者可能直接转化为康复者，也可能首先转化为感染者，感染者居家用药或住院 治疗，转化为康复者，且康复者具有暂时免疫能力，内外环境的变化，使得康复者可能逆变为易感者。 根据上述转换过程，构建嵌入防控和救治两项管控措施的 sirs 传染演化模型（1）。                                   k k k k k k k k k k k k k k k ds t k t s t s t s t r t dt di t k t s t i t i t dt dr t i t i t s t s t r t dt                                  （1） kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 132 其中，    ,k ks t i t 和  kr t 分别表示度为 k 的三类人群在 t 时刻网络中所占比例，分别称之为易感 者密度、感染者密度和康复者密度，故       1k k ks t i t r t   。  t 表示t 时刻随机取一条边与 感染者飞沫或接触的概率。即   ( ) ( ) 1 ( ) ( ) ( ) k k k k s kp k i t t kp k i t sp s k     （2） 2.3 社会网络稳定性划分 模型（1）刻画了易感者、感染者和康复者之间的转化关系。经过一段时间的传染演化之后，若新 冠肺炎病毒在社会网络中的传染处于不确定或无序状态，则认为社会网络处于不稳定状态；若新冠肺 炎病毒在社会网络中的传染趋于稳定，即社会网络中易感者、感染者和康复者三类人群各自密度趋于 定值，则认为社会网络处于稳定状态，这正是文章所关注的状态。因此，本文主要讨论模型（1）的稳 定解。社会网络的稳定性可以划分为有风险稳定和无风险稳定。 定义 1. 有风险稳定：社会网络中的有风险稳定是指，经过一段时间的传染演化后，新冠肺炎病毒 的传染趋于稳定，此时，社会网络中仍然存在感染者，但感染者在社会网络中的密度最终趋于某一常 量 i*，称易感者、感染者和康复者各自在社会网络中的密度向量 , ,s i r   为有风险稳定点 e 。 根据定义 1，当社会网络处于有风险稳定时，感染者始终存在于网络中，但其数量不再变化，此时， 新冠肺炎病毒仍具传染性，但社会网络处于相对有序和确定的状态。当外部环境发生变化时，有风险 稳定状态可能被打破，社会网络重新回到无序和不确定状态。 定义 2. 无风险稳定：社会网络中的无风险稳定是指，在经过一段时间的传染演化后，新冠肺炎 病毒不再具有传染性，此时，社会网络中不再存在感染者，即 * 0i  ，称易感者、感染者和康复者各 自在社会网络中的密度向量 0 0, 0,s r  为无风险稳定点 0e 。 根据定义 2，当社会网络处于无风险稳定时，网络中不存在感染者，新冠肺炎病毒不再具有传染 性。外部条件未发生变化时，社会网络处于可持续发展的健康状态。 2.4 新冠肺炎病毒的传染阈值分析 直接求解模型（1）比较困难，而在稳定状态条件下求解则比较容易。令 ,s i 和 r 分别为社会网 络处于有风险稳定时，整个网络中易感者、感染者和康复者密度。 定义 3：当新冠肺炎病毒传染概率 小于某一概率值 c 时，新冠肺炎病毒在社会网络中不再具有 传染性，称 c 为新冠肺炎病毒在社会网络中的传染阈值。 命题 1：社会网络中，考虑防控救治的新冠病毒存在一个传染阈值    2c k k            ， 即当 c  时，新冠病毒在社会网络中存在传染性，社会网络最终趋于有风险稳定；而当 c  时， kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 133 新冠病毒在社会网络中不再具有传染性，社会网络最终趋于无风险稳定。 证明：由模型（1）可得：      * k k k k i s k k i r k                            （3） 结合（2）与归一化条件       1k k ks t i t r t   ，可得：   k k i k k                        （4） 将（4）代入（2），可得：        1 = = k k kp k f k k k                           （5） 可见（5）是关于  的自相容方程， =0 显然是（5）的一个平凡解。 由 于   0 df d      ， 故  f  关 于  严 格 单 调 递 增 ， 因 此 当      2 0 1 df k d k                   时，方程（5）存在 0 1   的非平凡解，从而得出 新冠肺炎病毒传染阈值    2c k k            。将（5）的非平凡解  代入（4），可得 ki  ， 代入（3）可得 ks  和 kr  。又因   k k s p k s   ， * *( ) k k i p k i  ， * *( ) k k r p k r  。故存在社会 网络的新冠病毒稳定点  , ,e s i r    。即当   2 k k            时，新冠病毒在社会 网络中存在传染性，社会网络最终趋于有风险稳定点 e 。 当社会网络处于无风险稳定时，即 0i  ，可得新冠病毒在社会网络中的无风险稳定点  0 1 1, 0,e s r  ，即当    2 k k            时，新冠病毒在社会网络中消失，社会网络 最终趋于无风险稳定点 0e 。 文献研究表明社会网络具有较小的平均路径长度和较大的集聚系数，即社会网络具有小世界特性 [22]，故有 k k ，可得新冠病毒在社会网络中的传染阈值：   1 c k            。 2.5 传染阈值对防控救治的敏感性分析 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 134 上述阈值表达式 c 和 1 c 表明，阈值均随着和 的增加而增大。因此，无论新冠病毒传染概率 大小，随着防控救治力度加大和经验积累，社会网络趋于无风险稳定的概率增加。 命题 2：在社会网络结构等其他条件不变的情景下，当采取相同强度的防控和救治，即 =  时， 救治比防控更能提高新冠病毒传染阈值，即救治比防控的效率更高。 证明：由命题 1 中 c 的表达式，可得：   2 0c kd d k        ，   2 0c kd d k          因此，新冠病毒传染阈值随着防控和救治力度的增大而增大。由命题 2 的 =  、假设 3 的   和 0  ，可以得到：    + +      ， 显而易见，传染阈值对救治率敏感性大于传染阈值对防控率敏感性，即相较于防控，救治更能提高新 冠病毒传染阈值，故相较于防控率，救治率对传染阈值的贡献更大。因此，救治比防控的效率更高。 3. 数值仿真分析 实际上，由新冠病毒传染阈值的表达式可见，阈值均随着防控率和救治率的增大而增大，因此无 论网络是否均匀，阈值与防控救治效果之间的变化规律不受影响。基于此，为简化研究，本文选用均 匀网络进行数值仿真。小世界网络被视为均匀网络的代表，具有较大的聚集系数和较小的平均路径长 度，被广泛应用于流行病和经济管理领域的研究[23]。 为了更好的刻画防控救治对新冠病毒传染和社会网络稳定性的影响，结合目前疫情形势，本文仅 对有风险稳定性进行数据仿真分析。当社会网络处于有风险稳定时，分别就采取防控和救治时，考察 风险管控措施对新冠病毒传染和社会网络稳定状态的影响。为此，假设社会网络中人群总数为 100000， 初始时刻网络中有 30 个感染者，以下仿真图中横轴代表时间，纵轴代表人数。借鉴文献[4]、[9]、[24] 和[25]的参数设置和研究成果，设易感者转变为感染者的传染概率 =0.3 ；康复者转变为易感者的概 率 =0.3 ；网络平均度 3k  。 3.1 社会网络中三类人群总量变化趋势 在不考虑防控救治的情景下，即当 =0, =0  时，令易感者转化为康复者的概率 =0.2 ，感染者 转化为康复者的概率 =0.2 ；根据上述参数，可得   1 0.11 0.3c k              ， 即 1 c  。 根据模型（1）和上述参数的设定，进行数值仿真，如图 1 所示。 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 135 图 1 未采取防控救治的三类人群随时间变化趋势 由图 1 可见，当传染概率大于传染阈值时，社会网络最终趋于有风险稳定，且前期感染者数量增 长较快，这与命题 1 的结论一致，并与现实中新冠肺炎传染蔓延现象一致。现实中，新冠肺炎病毒的 传染性较强，易感人群被感染的概率较大，因而社会网络中感染者数量增多，在不采取限制出行、隔 离等防控和专门救治的情景下，网络中感染者数量增长速度较快。另外，感染者以概率 =0.6 转变为 康复者，故感染人群不会无限制的增加，最终整个社会网络中感染人群、易感人群和康复人群趋于某 三个常数。 同时考虑防控救治的情景下，当 =0.1, =0.2  时，令易感者转化为康复者的概率 =0.2 ，感染者 转化为康复者的概率 =0.2 ；根据上述参数，可得   1 0.27 0.3c k              ，即 1c  。 根据模型（1）和上述参数的设定，进行数值仿真，如图 2（a）所示。 图 2（a） 采取防控救治的三类人群随时间变化趋势 由图 2（a）可以发现，当传染概率大于传染阈值时，社会网络最终趋于有风险稳定，这与命题 1 的结论一致，并与现实中新冠肺炎传染蔓延现象一致。对比图 1 和图 2（a），不难发现，由于采取了防 控救治措施，前期感染者数量增长较慢，且感染人群数量比未采取措施时少很多，进一步说明防控救 治措施有效。图 2（a）表明最终整个社会网络中感染人群、易感人群和康复人群趋于某三个常数。 同时考虑防控救治的情景下，当 =0.4, =0.5  时，令易感者转化为康复者的概率 =0.2 ，感染者 转化为康复者的概率 =0.2 ；根据上述参数，可得 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 136   1 0.9 0.3c k              ， 即 1 c  。 根据模型（1）和上述参数的设定，进行数值仿真，如图 2（b）所示。 图 2（b） 采取防控救治的三类人群随时间变化趋势 图 2（b）表征了当传染概率小于传染阈值时，虽初始阶段有不少感染者，但社会网络最终趋于无 风险稳定，这与命题 1 的结论一致。现实中，也可以观察到此现象，新冠病毒在社会网络中传染，由 于防控措施得力，易感者被传染的可能性大大降低，同时感染者会得到专门救治，当救治率和防控率 大于传染概率时，网络中的感染者数量将逐步减少，最终导致社会网络中不再存在新冠病毒的传染。 3.2 防控对社会网络稳定状态的影响 在上述参数设定的基础上，仅考察限制出行和隔离防控及防控率 对新冠病毒传染和社会网络稳 定状态的影响。当 =0 时，即顺其自然，不采取任何防控措施时，显而易见，社会网络趋于有风险稳 定，且感染者数量会比较大。由图 2 可见，当 =0.9 时，社会网络中感染者数量趋于 0，即趋于无风 险稳定；当 =0.6 时，社会网络中最终仍然存在一定数量的感染者，且网络趋于有风险稳定。当 0.6  时，随着防控率的增加，网络中感染者数量减少，并且存在一个防控率阈值 c ，当防控率大于此阈值 时，社会网络从有风险稳定转变为无风险稳定。 图 3 感染者数量对防控率的敏感性 3.3 救治对社会网络稳定状态的影响 在上述参数设定的基础上，仅考察专门救治及救治率 对新冠病毒传染和社会网络稳定状态的影 响。当 =0 时，即顺其自然，不采取任何专门救治措施时，显而易见，社会网络趋于有风险稳定，且 kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 137 感染者比例会比较大。由图 3 可见，当 =0.6 时，社会网络中感染者数量趋于 0，即趋于无风险稳定； 当 =0.4 时，社会网络中最终仍然存在一定数量的感染者，且网络趋于有风险稳定。当 0.4  时， 随着救治率的增加，网络中感染者数量减少，并且存在一个救治率阈值 c ，当救治率大于此阈值时， 社会网络从有风险稳定转变为无风险稳定。 图 4 感染者数量对救治率的敏感性 由图 3 和图 4 可以看出，采取防控救治措施能够使社会网络从有风险稳定转化为无风险稳定。对 比防控和救治两类措施发现，当社会网络处于有风险稳定时，同时将防控率 和救治率 提高到 0.6 时， 采取救治措施已使社会网络处于无风险稳定状态，而防控措施情景下社会网络仍然处于有风险稳定， 表明相较于防控，救治能更快的使社会网络转变为无风险稳定，即救治比防控效率更高，这与命题 2 一 致。 4. 结论 文章结合 2019 年 12 月爆发的新冠肺炎疫情实时发展状况，运用 sirs 模型，对 2020 年 1 月 25 日 以来政府限行、隔离防控和专门救治效果进行评估。首先，推导出社会网络中新冠肺炎病毒传染阈值 表达式，发现防控率和救治率均与传染阈值正相关，因此可以采取适当的防控和救治措施抑制新冠肺 炎病毒的传染。其次，数学推导表明，传染阈值对救治率的敏感性大于其对防控率的敏感性，即相较 于防控率，救治率对新冠肺炎病毒传染阈值的贡献更高，也就是说以相同幅度提高防控率与救治率， 救治比防控更有效。最后，数值仿真表明，防控和救治均对社会网络的稳定性有显著影响，均会降低 新冠肺炎病毒传播中感染人群数量；当社会网络处于有风险稳定时，通过防控和救治都可以使其趋于 无风险稳定，且相较于防控，救治能更快地使社会网络趋于无风险稳定。因此，为快速抑制新冠肺炎 疫情蔓延，中国政府及时强力推出的限行、隔离、筹建或指定定点医院等防控和救治措施效果显著， 虽然境外输入引发的新冠疫情反复在全国各地时有爆发，但实际数据表明疫情早已得到控制，人们的 生产生活早已正常化。与此同时，诸如美国、印度、巴西、英国、意大利、法国、伊朗等国家新冠肺炎 疫情肆虐，我国管控新冠肺炎疫情的有效措施，为境外疫情管控提供了重要的决策参考。 funding: this research was funded by the national natural science foundation of china, grant number 71871147; this research was funded by the sichuang provincial social science planning funding project, grant number sc19b004. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. kai xu, xiaofeng xie, xing hang, et al. / journal of risk analysis and crisis response, 2021, 11(3), 129-138 doi: https://doi.org/10.54560/jracr.v11i3.302 138 references [1] who. who coronavirus disease (covid-19) dashboard. geneva, switzerland: who; 2020[eb/ol]. available from: https://covid19.who.int. [2] jia j, lu x, yuan y, et al. population flow drives spatio-temporal distribution of covid-19 in china[j]. nature, 2020: 1-11. [3] wu h, zhang q, wu h, et al. tackling covid-19: insights from the qinghai province plague prevention and control (ppc) model[j]. biosafety and health, 2020, 2(4): 187-192. [4] shen j, duan h, zhang b, et al. prevention and control of covid-19 in public transportation: experience from china[j]. environmental pollution, 2020, 266(pt 2): 115291. [5] rutayisire e, nkundimana g, mitonga h k, et al. what works and what does not work in response to covid-19 prevention and control in africa[j]. international journal of infectious diseases, 2020, 97: 267269. [6] liu f, wang w, yu h, et al. prevention and control strategies of general surgeons under covid-19 pandemic[j]. surgery in practice and science, 2020: 100008. [7] tartari e, hopman j, allegranzi b, et al. perceived challenges of covid-19 infection prevention and control preparedness: a multinational survey[j]. journal of global antimicrobial resistance, 2020, 22(9): 779-781. [8] wang j, yang w, pan l, et al. prevention and control of covid-19 in nursing homes, orphanages, and prisons[j]. environmental pollution, 2020, 266: 115161. [9] 白宁, 宋晨玮, 徐瑞. 基于动力学模型的 covid-19 疫情预测与控制策略研究[j]. 应用数学学报, 2020, 43(03): 483-493. [10] 潘理虎, 秦世鹏, 李晓文, 芦飞平, 杨芬玉. covid-19 病毒防控多智能体仿真模型[j]. 系统仿真学报, 2020, 32(11): 2244-2257. [11] yang z f, zeng z q, wang k, et al. modified seir and ai prediction of the epidemics trend of covid-19 in china under public health interventions[j]. journal of thoracic disease, 2020, 12(3): 165-174. [12] wang x, tang s y, chen y, et al. when will be the resumption of work in wuhan and its surrounding areas during covid-19 epidemic? a data-driven network modeling analysis[j]. scientia sinica mathematica, 2020, 50: 1-10. [13] 陈彬, 杨妹, 艾川, 马亮, 朱正秋, 陈海亮, 朱蒙娜, 许伟. 基于人工社会的疫情传播风险预测和防控措施评估 [j]. 系统仿真学报, 2020, 32(12): 2507-2514. [14] 朱翌民, 黄勃, 王忠震, 巨家骥, 朱良奇. 隔离措施对 covid-19 疫情控制的模型分析[j]. 武汉大学学报(理学 版), 2020, 66(05): 442-450. [15] 孙皓宸, 刘肖凡, 许小可, 吴晔. 基于连续感染模型的新冠肺炎校园传播与防控策略分析[j]. 物理学报, 2020, 69(24): 74-83. [16] 尹楠. 基于 sir 模型的有限区域内新冠肺炎疫情传播仿真模拟[j]. 统计与决策, 2020, 36(05): 15-20. [17] 喻孜, 张贵清, 刘庆珍, 吕忠全. 基于时变参数-sir 模型的 covid-19 疫情评估和预测[j]. 电子科技大学学报, 2020, 49(03): 357-361. [18] li q, med m, guan x h, et al. early transmission dynamics in wuhan, china, of novel coronavirusinfected pneumonia. 2020, 382(13): 1199-1207. [19] 朱启荣, 孙明松, 杨伟东. 新冠肺炎疫情对我国经济影响的评估：基于 gtap 模型的实证[j]. 统计与决策, 2020, 36(21): 91-96. [20] nishi a, dewey g, endo a, et al. network interventions for managing the covid-19 pandemic and sustaining economy[j]. the national academy of sciences, 2020, 117(48): 30285-30294. [21] 李兰娟, 朱雪灵. 新型冠状病毒肺炎疫情防控相关进展[j].浙江医学,2021,43(01):1-8. [22] milgram s. the small world problem[j]. psychology today, 1967, 2(1): 60-67. [23] watts d j, strogatz s h. collective dynamics of ‘small-world’ networks[j]. nature, 1998, 393(6684): 440. [24] 耿辉, 徐安定, 王晓艳, 张勇, 尹小妹, 马茂, 吕军. 基于 seir 模型分析相关干预措施在新型冠状病毒肺炎疫 情中的作用[j]. 暨南大学学报(自然科学与医学版), 2020, 41(02): 175-180. [25] 钱茜, 周宗放, 李永奎. 管控策略对关联企业网络稳定状态的影响—基于关联信用风险传染的视角[j]. 中国 管理科学, 2019, 27(06): 21-29. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word a credit rating model for enterprises based on projection pursuit and k-means clustering a credit rating model for enterprises based on projection pursuit and k-means clustering algorithm mu zhang school of finance, guizhou university of finance and economics guiyang 550004, guizhou, china, e-mail: rim_007@163.com zongfang zhou school of management and economics, university of electronic science and technology of china chengdu 610054, sichuan, china abstract this paper proposes a new credit rating model for enterprises based on projection pursuit and k-means clustering algorithm. firstly, using projection pursuit, the comprehensive credit score of each sample is obtained, so as to reflect the structure or characteristics of original multi-dimensional data. secondly, the distribution density of the comprehensive credit score series is estimated by the kernel density estimation method, and then the initial cluster centers in original high dimension space are determined according to the local maximum points of density function. finally, starting from the initial cluster centers above, using k-means clustering algorithm, the final cluster centers are obtained, and then the credit grades are partitioned. thus, the credit rating for enterprises is realized. taking the high-tech listed companies in china as samples, it is proved that the model proposed by this paper is feasible and effective. keywords: enterprise credit rating; projection pursuit; kernel density estimation; initial cluster centers; k-means clustering algorithm 基于投影寻踪和 k-均值聚类的企业信用评级模型 张目 1 周宗放 2 1. 贵州财经大学/金融学院，贵阳 550004 2. 电子科技大学/经济与管理学院，成都 610054 摘要：提出一种基于投影寻踪和 k-均值聚类的企业信用评级模型。首先，运用投影寻踪对样本企业进行信 用综合评分，以反映原高维数据的结构或特征；然后，利用核密度估计法对信用综合得分序列进行分布密 度估计，并根据密度函数的局部极大值点确定原高维空间中的初始聚类中心； 后，从给出的初始聚类中 心出发，运用 k-均值算法获得 终聚类中心，并划分企业信用等级，从而实现对样本企业的信用评级。以 我国高技术产业上市公司为例，应用实例证明了该模型的可行性和有效性。 关键词：企业信用评级，投影寻踪，核密度估计，初始聚类中心，k-均值聚类算法 1. 引言 企业信用评级是运用科学的指标体系、定量分 析和定性分析相结合的方法，通过对企业信用记 录、经营水平、外部环境、财务状况、发展前景以 及可能出现的各种风险等进行客观、科学、公正的 分析研究之后，就其信用能力所做出的综合评价， 并用特定的等级符号标定其信用等级[1]。信用评级 有助于企业防范商业风险，为现代企业制度的建设 提供良好条件；信用评级有利于资本市场的公平、 公正和诚信；同时，信用评级也是商业银行确定贷 款风险程度的依据和信贷资产风险管理的基础。 目前，信用评级 常用的方法是基于分类的方 法。在 altman (1968)[2]做出开创性工作之后，多元 判别分析(mda)[2-3]、logistic 回归模型[4]、probit 回 归模型[5]等统计方法在信用评级中获得了广泛应 journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 131-138 published by atlantis press copyright: the authors 131 administrateur texte tapé à la machine received 7 february 2012 administrateur texte tapé à la machine accepted 14 march 2012 administrateur texte tapé à la machine zhang and zhou 用。然而，这类统计方法存在着诸多局限，如： mda 要求样本数据服从正态分布和等协方差，而现 实中大量数据并不服从这些假定[6]；logistic 回归模 型不仅对中间区域的差别敏感性较强，而且当样本 点完全分离时，模型参数的 大似然估计可能不存 在[7]。20 世纪 90 年代以来，以聚类分析[8]和 k-近邻 法[9]为代表的非参数统计方法被引入到信用风险分 析中，其中，聚类分析具有不要求样本数据服从具 体分布，并且，具有可对变量采用名义尺度和次序 尺度等优点，适于信用风险分析中按照定量指标和 定性指标对并不服从一定分布特性的数据信息分类 的要求 [10] 。在众多的聚类算法中， k均值（ kmeans）算法[11]是一种基于划分的聚类算法，因其理 论上可靠、算法简单、收敛速度快、能有效处理大 数据集而得到 为广泛的使用[12]。参考文献[13]和 [14]对 k-均值算法在企业信用评级中的应用进行了 有益的尝试，其基本思路是：首先采用 z 评分法、 因子分析法等对样本企业进行信用评分，然后，在 系统自动指定初始聚类中心下，运用 k-均值算法对 信用得分序列进行聚类。上述研究存在以下两个方 面的问题：（1）将高维数据“降维”后进行聚类分 析，易丢失数据信息；（2）由系统自动指定初始聚 类中心，导致聚类结果缺乏可靠性。 众所周知，k-均值算法对初始聚类中心较为敏 感，对于给定的聚类数目 k，从不同的初始聚类中 心出发，可能得到不同的聚类结果[15-16]。现有文献 提出的优选初始聚类中心的方法主要有：密度评估 法、距离优化法、基于遗传算法的方法和基于取样 的方法等[17-22]。这些方法在一定程度上优化了初始 聚类中心，减少了聚类的迭代次数。然而，上述方 法均是在高维空间中进行计算，其算法复杂度较 高，且某些方法存在输入参数难以确定的不足。 投影寻踪（projection pursuit，pp）[23-24]是一种 直接由样本数据驱动的探索性数据分析方法，特别 适用于分析和处理非线性、非正态的高维数据，其 基本思想是把高维数据投影到低维子空间上，寻找 出能反映原高维数据的结构或特征的投影，以达到 研究分析高维数据的目的。有鉴于此，本文受参考 文献[25]的启发，将投影寻踪与核密度估计结合运用 于优选初始聚类中心，从而提出一种基于投影寻踪 和 k-均值聚类的企业信用评级模型。本文的研究逻 辑是：首先，运用投影寻踪对样本企业进行信用综 合评分，以反映原高维数据的结构或特征；然后， 利用核密度估计法对信用综合得分序列进行分布密 度估计，并根据密度函数的局部极大值点确定原高 维空间中的初始聚类中心； 后，从给出的初始聚 类中心出发，运用 k-均值算法获得 终聚类中心， 并划分企业信用等级，从而实现对样本企业的信用 评级。 2. k-均值算法原理 k-均值算法的基本思想是通过迭代把数据对象 划分到不同的簇中，以求目标函数 小化，从而使 生成的簇尽可能的紧凑和独立。给定样本集和正整 数 k，k-均值算法将样本集分割成 k 个簇，每个聚 类中心是簇中样本的均值；将其余对象根据其与各 个簇的中心的距离分配到 近的簇；然后，求出新 形成的簇的中心。这个迭代重新定位过程不断重 复，使得每个簇中所有样本与其中心的距离总和 小，直到目标函数 小化为止[11-12]。 k-均值聚类过程是通过反复移动簇中心以 小 化簇集内的总度量（如：距离、相似度等）来完成 的。设样本为 ix （ 1, 2, ,i n  ），给定一组初始 聚类中心点 kc （ 1, 2, ,k k  ），初始聚类中心可 以从样本集中随机选择，也可以根据实际需要来指 定。k-均值聚类算法交替执行以下两步[11-12]： （1）对每个样本 ix ，找出距离其 近的中心 点（簇） argk   1,2, ,min ( , )k ik k d c x  ， 1, 2, ,k k  （1） （2）计算每个簇中样本的均值，该均值向量即 成为该簇新的中心 ( ) 1 1 kn k k j jk c x n    ， 1, 2, ,k k  （2） 其中， kn 为第 k 簇中的样本数。 重复以上两步，直到没有样本或很少的样本被 分配到不同的簇中。 3. 企业信用评级模型的构建 对于多分类的企业信用评级问题，设有 m 个企 业组成训练样本集  1, ,ia a i m   ，企业信用评 级指标集  1, ,jc c j n   ， ijx 为训练样本 ia 在 published by atlantis press copyright: the authors 132 a credit rating model for enterprises 指标 jc 下的指标值。基于投影寻踪和 k-均值聚类的 企业信用评级模型构建步骤如下： 步骤 1：指标值的归一化处理。为消除各指标 的量纲、统一各指标的变化范围和方向，须对指标 值进行极值归一化处理。 对于成本型指标，令 max max min j ij ij j j x x y x x    ， 1, 2, , ; 1, 2, ,i m j n   （3） 对于效益型指标，令 min max min ij j ij j j x x y x x    ， 1, 2, , ; 1, 2, ,i m j n   （4） 式（3）-（4）中， maxjx 、 min jx 分别为第 j 个指标的 大值和 小值。 对于固定型指标，即指标值越接近某一固定值 越好的指标，有 1 max ij j ij ij j i x x y x x       ， 1, 2, , ; 1, 2, ,i m j n   （5） 式（5）中， jx  为第 j 个指标的 佳稳定值。 步骤 2：构造信用评分函数及投影指标函数。 pp 方法就是把 n 维数据  1, ,ijy j n  综合成以 1 2( , , , )na a a a  为投影方向的一维投影值 iz ： 1 n i j ij j z a y    ， 1, 2, ,i m  （6） 上式中， a 为单位长度向量。 iz 近似刻画了样本企 业的信用状况[7]，投影值越低，信用风险越高，则 称式（6）为样本企业的信用评分函数， iz 为样本 企业的信用综合得分。 pp 方法在综合 iz 时，要求 iz 的散布特征应 为：局部投影点尽可能密集， 好凝聚成若干个点 团，而在整体上投影点团之间尽可能散开。由此， 投影指标函数可构造为[23] ( ) z zq a s d （7） 式中， zs 为 iz 的标准差， zd 为 iz 的局部密度，即 2 1 1 ( ) 1 m z i i s z z m      （8） 1 1 ( ) ( ) m m z ij ij i j d r r i r r      （9） 其中， 1 1 m i i z z m    为 iz 的均值； r 为求局部密度 的窗口半径，它的选取既要使包含在窗口内的投影 点的平均个数不太少，避免滑动平均偏差太大，又 不能使它随着 m 的增大而增加太快， r 一般可取值 为 0.1 zs [26-27]；点间距离 | |ij i jr z z  ； ( )i t 为单位 阶跃函数，当 0t 时其函数值为 0，当 0t 时其 函数值为 1。 步骤 3：优化投影指标函数。当样本集给定 时，投影指标函数只随投影方向的变化而变化。不 同的投影方向反映不同的数据结构特征， 佳投影 方向就是 大可能暴露高维数据某类特征结构的投 影方向 [26-27]。通过求解投影指标函数 大化问题可 估计出 佳投影方向，即 2 1 max ( ) . . 1 z z n j j q a s d s t a    （10） 式（10）所设定的问题是一个以 1, ,ja j n  为优化变量的复杂非线性优化问题，常规优化方法 较难处理。模拟生物优胜劣汰规则与群体内部染色 体信息交换机制的实码加速遗传算法(real coded accelerating genetic algorithm， raga)是一种通用 的全局优化方法，用它来求解该问题则十分简便而 有效。raga 的具体算法参见参考文献[26]和[27]。 published by atlantis press copyright: the authors 133 zhang and zhou 步骤 4：计算训练样本的信用综合得分，对信 用综合得分序列进行分布密度估计。将步骤 3 估计 出的 佳投影方向 a 代入式（6）后可得训练样本 的信用综合得分 iz  。初始聚类中心要求是一组能尽 量反映数据分布特征的数据对象[28]；由投影寻踪原 理可知， iz  的散布特征反映了原高维数据的某种结 构或特征；因此，可以通过分析 iz  的散布特征来优 选初始聚类中心。为此，本文利用核密度估计法对 信用综合得分序列 iz  进行分布密度估计。其中，核 密度估计定义如下 [29-30]： 定义 1：设 ( )k 为 1r 上一个给定的概率密度函 数， 0mh  是一个与 m 有关的常数，满足 m   ， 0mh  ，则称 1 1 ( ) ( ) m i m im m z z f z k mh h        （11） 为 ( )f z 的一个核密度估计，其中 ( )k 为一已知核 函数，满足 sup ( ) u k u     ， ( ) ( )k u k u  （12） ( )k u du     （13） lim ( ) 0 u uk u   （14） mh 称为窗宽或光滑参数。 步骤 5：确定初始聚类中心，运用 k-均值算法 划分信用等级。由步骤 4 得出密度函数 ( )mf z  及相 应的核密度估计曲线。在已知数据分布的条件下， 一个优良的初始聚类中心应满足[31]：（1）选择的初 始聚类中心点各属于不同的类，即任意两个初始聚 类中心点不能属于同一类；（2）选择的初始聚类中 心点应能够作为该类代表，即应该尽量靠近类中 心。据此，可直观搜索出密度函数 ( )mf z  的局部极 大值点，并选取与局部极大值点 临近的样本投影 点在原高维空间中所对应的点为初始聚类中心点。 在运用 k-均值算法对企业进行信用评级时，首 先根据信用评级的实际需要设定 k 个信用等级，则 应有 k 个聚类数目与之对应，从而需选取 k 个样本 点组成初始聚类中心。假设密度函数 ( )mf z  有 n 个 局部极大值点，当 k=n 时，初始聚类中心随即确 定；当 k＜n 时，从我国商业银行“区别对待，择 优扶持”的信贷原则出发，在 n 个局部极大值点中 选取数值较大的前 k 个点来确定初始聚类中心；当 k＞n 时，则需通过增加训练样本数量来使得 k≤ n。 在确定初始聚类中心后，运用 k-均值算法对训 练样本进行聚类分析，从而得到 k 个 终聚类中心 点。由式（6）计算 k 个 终聚类中心点的信用综合 得分，然后，根据信用综合得分的大小，建立聚类 类别与信用等级的一一对应关系，从而划分出 k 个 信用等级，并实现对训练样本的信用评级。 步骤 6：对新样本进行信用评级。对于一个新 的测试样本，首先，运用式（3）-（5）对测试样本 的信用评级指标值进行标准化处理，特别地，当测 试样本的第 j 个指标值在训练样本指标值区间 min max,j jx x   （ 1, 2, ,j n  ）内时，即为归一化处 理。然后，分别计算测试样本与步骤 5 得出的 k 个 终聚类中心点的欧式距离，找出距离其 近的中 心点，该中心点对应的信用等级即为测试样本所属 的信用等级。特别地，当测试样本与 2 个或 2 个以 上 终聚类中心点的欧式距离相等时，则可通过计 算联系向量距离[32]来加以区分。 4. 应用实例 4.1. 指标体系与样本数据 本文参照国家财政部统计评价司的企业绩效评 价指标体系和中国工商银行企业资信评估指标体 系，遵循指标选取的系统性、科学性、客观性、可 比性及可操作性等原则，从偿债能力、营运能力和 盈利能力等三个方面构建企业信用评级指标体系。 该指标体系包括以下 12 个指标：流动比率、速动比 率、资产负债率、利息保障倍数、存货周转率、应 收账款周转率、总资产周转率、固定资产周转率、 总资产报酬率、净资产报酬率、销售净利率、股本 报酬率等。 选取沪、深股市中的高技术产业上市公司作为 实验样本，样本区间选定为 2005-2007 年，数据来 源于国泰安数据库。剔除异常数据样本后， 终获 得 112 家样本企业，其中，有 74 家为“非 st ”企 业，这类企业称之为“正常企业”；其余 38 家为 published by atlantis press copyright: the authors 134 a credit rating model for enterprises “ st 或 st ”企业，这类企业称之为“违约企 业”。将实验样本集划分为训练样本集和测试样本 集。随机抽取 50 家“正常企业”和 25 家“违约企 业”作为训练样本，剩余的 24 家“正常企业”和 13 家“违约企业”作为测试样本。使用 matlab7.1 工具包、eviews6.0 和 spss16.0 软件进行实验分析 。 4.2. 信用评分及分布密度估计 按照第 3 节步骤 1，对训练样本指标值进行归 一化处理。运用 raga 求解式（10）所设定的 优 化 问 题 ， 得 出 大 投 影 指 标 函 数 值 ： max ( )q a =0.9539， 佳投影方向： a  =（0.0171， 0.0140，0.3387，0.1517，0.4217，0.3066，0.3451， 0.2032，0.2904，0.2252，0.3880，0.3793）。将 a 代入式（6），计算出训练样本的信用综合得分 iz  。 根据定义 1 对信用综合得分序列 iz  进行分布密 度估计。首先，采用 silverman（1986）提出的经验 法则[33]计算初始光滑参数，即：假定 ( )f z 为正态 密度函数 2(0, )n  ，选取正态核函数，则根据经验 法可得 佳渐进光滑参数为： 1 5ˆ ˆ1.06amiseh m   （15） 其中̂ 为信用综合得分序列 iz  的标准差估计值。 将 75m  ， ˆ 0.1166  代 入 式 （ 15 ） 得 出 ˆ 0.0521amiseh  。 其次，选取正态（gaussian）核函数： 2 1( ) ( 2 ) exp( ) 2 g u k u    ，  ,u    （16） 设置格点数为 200，利用 eviews6.0 软件实现核密度 估计，由于拟合曲线不光滑，本文还采用尝试法[30] 对光滑参数进行适当调整，当光滑参数为 0.0180 时 得到较为满意的结果。信用综合得分序列 iz  的核密 度估计曲线见图 1。 4.3. 初始聚类中心的确定 从图 1 可以看出，密度函数 ( )mf z  共有 7 个局 部极大值点，通过对 eviews6.0 软件输出的数据矩阵 的直观搜索，得出这 7 个局部极大值点分别为： 0.7459，0.8152，1.0193，1.1156，1.2350，1.3082， 1.4700。本文根据我国商业银行贷款五级分类的实 际需要，设定 5 个信用等级，则应有 5 个聚类数目 与之对应。由第 3 节步骤 5，在上述 7 个局部极大值 点中选取数值较大的前 5 个点来确定原高维空间中 的初始聚类中心。 4.4. k-均值聚类分析与信用等级的划分 导入 4.3 节得出的初始聚类中心进行 k-均值聚 类分析。设置聚类数目为 5， 大迭代次数为 20， 收敛准则为 0.01，经过 5 次迭代后，达到聚类结果 的要求，聚类分析结束，得出 终聚类中心。由式 （6）计算出 5 个 终聚类中心点的信用综合得分 为：1 类-1.1080、2 类-1.2293、3 类-1.2486、4 类1.2700、5 类-1.3501。根据信用综合得分的大小， 建立聚类类别与信用等级的一一对应关系，即有：5 类-ⅰ、4 类-ⅱ、3 类-ⅲ、2 类-ⅳ、1 类-ⅴ，其 中，数字序号ⅰ-ⅴ分别代表 5 个信用风险从低到高 的 信 用 等 级 。 0 1 2 3 4 5 6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 d e n si ty z fig. 1. kernel density estimation curve 图 1 核密度估计曲线 published by atlantis press copyright: the authors 135 zhang and zhou 4.5. 信用评级结果与对比分析 根据聚类类别与信用等级的对应关系及聚类成 员，即可实现对训练样本的信用评级。对于新的测 试样本，则按照第 3 节步骤 6 评定其信用等级。训 练样本和测试样本的信用评级结果见表 1。 由表 1 可知，训练样本的信用评级结果表现 为：从第ⅰ级到第ⅴ级，随着信用等级的降低，违 约企业比例（可近似看成违约率）呈单调递增趋 势。即，企业信用等级越低，其违约概率越大，信 用风险越高，这与信用风险管理理论相吻合。测试 样本的信用评级结果呈现出与训练样本相似的特 征，表明本文模型具有良好的泛化能力，能够满足 实际应用的需要。 为便于比较，本文还采用由系统自动指定初始 聚类中心的 k-均值算法（以下简称为传统模型）对 样本企业进行信用评级。设置聚类数目为 5， 大 迭代次数为 20，收敛准则为 0.01，经过 8 次迭代 后，达到聚类结果的要求，聚类分析结束，得出 终聚类中心。由式（6）计算出 5 个 终聚类中心点 的信用综合得分为：1 类-0.8896、2 类-1.1551、3 类 -1.3143、4 类-1.2287、5 类-1.3153。训练样本和测 试样本的信用评级结果见表 2。 由表 2 可知，虽然训练样本的信用评级结果与 信用风险管理理论相符，但测试样本的信用评级结 果未呈现出与训练样本相似的特征，说明传统模型 的泛化能力较差，不能满足实际应用的需要。另 外，传统模型的聚类分析迭代次数为 8 次，高于本 文模型的 5 次，表明本文模型优选的初始聚类中心 减少了 k-均值算法的迭代次数，加快了算法的收敛 速度，提高了算法的运算效率。 此外，为进一步考察本文模型的聚类效果，本 文还比较了上述两个模型的 小目标函数值： 5 1 k k j j    ，其中， kj 表示第 k 类中聚类成员与其中 table 1. the result of enterprise credit rating based on the new model 表 1 企业信用评级结果（本文模型） 聚类类别 5 4 3 2 1 信用等级 ⅰ ⅱ ⅲ ⅳ ⅴ 训练样本（75） 企业总数 5 21 25 12 12 违约企业比例 0.0000 0.0476 0.1600 0.7500 0.9167 测试样本（37） 企业总数 2 11 13 6 5 违约企业比例 0.0000 0.0909 0.1538 0.8333 1.0000 注：表中括号内为样本个数。 table 2. the result of enterprise credit rating based on the traditional model 表 2 企业信用评级结果（传统模型） 聚类类别 5 3 4 2 1 信用等级 ⅰ ⅱ ⅲ ⅳ ⅴ 训练样本（75） 企业总数 8 27 18 20 2 违约企业比例 0.0000 0.0370 0.2778 0.8500 1.0000 测试样本（37） 企业总数 5 12 9 11 0 违约企业比例 0.2000 0.0833 0.3333 0.7273 —— 注：表中括号内为样本个数。 published by atlantis press copyright: the authors 136 a credit rating model for enterprises 心的距离总和。计算结果显示，本文模型的 j 值为 18.57，小于传统模型的 19.08，表明本文模型的聚 类效果优于传统模型。 5. 结束语 本文将投影寻踪与核密度估计结合运用于优选 初始聚类中心，从而提出一种基于投影寻踪和 k-均 值聚类的企业信用评级模型。该模型具有以下特 点：（1）运用投影寻踪对样本企业进行信用综合评 分，以反映原高维数据的结构或特征；利用核密度 估计法对信用综合得分序列进行分布密度估计，并 根据密度函数的局部极大值点来确定原高维空间中 的初始聚类中心，具有合理性和可操作性；（2）把 高维数据投影到低维子空间上，在低维子空间进行 初始聚类中心的优选，计算相对简单，且不需要任 何输入参数，具有直观性和便捷性；（3）从给出的 初始聚类中心出发，在原高维空间中运用 k-均值算 法进行聚类分析， 大限度的保留了原始数据的信 息，并提高了聚类结果的可靠性。本文的研究为拓 展 k-均值算法在企业信用评级中的应用提供了新的 方法和思路。k-均值算法是基于梯度下降的算法， 不可避免地常常陷入局部极优[16]，因此，将基于遗 传算法、免疫规划或粒子群优化的 k-均值算法引入 到企业信用评级中有待于进一步研究。 致谢 本 文 获 得 国 家 自 然 科 学 基 金 面 上 项 目 （70971015）、教育部人文社会科学研究规划基金 项目（11yja630196）和贵州财经学院金融学院科 研项目（2009-04）的资助，在此表示衷心的感谢。 参考文献 [1] li shimei, the theoretical thinking of credit capacity evaluation of the industrial enterprises in china, j. jilin university journal social sciences edition. 48(4) (2008) 107–112. 李士梅.我国工业企业信用能力评价的理论思考 [j].吉林大学社会科学学报, 2008,48(4):107–112. [2] altman e i, financial ratios, discriminant analysis and the prediction of corporate bankruptcy, j. journal of finance. 23(4) (1968) 589–609. [3] altman e i, haldeman r g and narayanan p, zeta analysis: a new model to identify bankruptcy risk of corporations, j. journal of banking and finance. 1(1) (1977) 29–54. [4] ohlson j, financial rations and the probabilistic prediction of bankruptcy, j. journal of accounting research. 18(1) (1980) 109–130. [5] gentry j a, whitford d t and newbold p, predicting industrial bond ratings with a probit model and funds flow components, j. the financial review. 23(3) (1988) 269–286. [6] eisenbeis robert a, pitfalls in the application of discriminant analysis in business, finance, and economics, j. journal of finance. 32(3) (1977) 875–900. [7] wang chunfeng and li wenhua, credit risk assessment in commercial banks: projection pursuit discriminant model, j. journal of industrial engineering and engineering management. 14(2) (2000) 43–46. 王春峰,李汶华.商业银行信用风险评估:投影寻踪 判别分析模型[j].管理工程学报, 2000,14(2):43– 46. [8] lundy m (eds.), cluster analysis in credit scoring. credit scoring and credit control, 1nd edn. (oxford university press, new york, 1993, 78–90). [9] henley w e and hand d j, a k-nearest-neighbor classifier for assessing consumer credit risk, j. statistician. 45(1) (1996) 77–95. [10] zhang wei and li yushuang, credit risk analysis in commercial bank: an overview, j. journal of management sciences in china. 1(3) (1998) 20–27. 张维,李玉霜.商业银行信用风险分析综述[j].管理 科学学报,1998,1(3):20–27. [11] macqueen j, some methods for classification and analysis of multivariate observations, in proceedings of the fifth berkeley symposium on mathematical statistics and probability, (university of california press, berkeley, calif., 1967, 1), pp. 281–297. [12] liang xun (eds.), data mining algorithms and applications, 1nd edn. (peking university press, peking, 2006, 16–22, 193). 梁循,数据挖掘算法与应用[m].北京:北京大学出 版社,2006.16–22,193. [13] zuo ziye and zhu yangyong, credit scoring and rating based on clustering technology of data mining, j. computer applications and software. 21(4) (2004) 1–3. 左子叶,朱扬勇.基于数据挖掘聚类技术的信用评 分评级[j].计算机应用与软件,2004,21(4):1–3. [14] zhang guiqing and liu shulin, empirical study of credit risk evaluation in china’s commercial banks, j. journal of hebei university of economics and trade. 26(4) (2005) 41–45. 张贵清,刘树林.我国商业银行信用风险评级实证 分析[j].河北经贸大学学报,2005,26(4):41–45. [15] pena j m, lozano j a and larranaga p, an empirical comparison of four initialization methods for the k-means algorithm, j. pattern recognition letters. 20(10) (1999) 1027–1040. [16] liu jingming, han lichuan and hou liwen, cluster analysis based on particle swarm optimization algorithm, j. systems engineering-theory and practice. 25(6) (2005) 54–58. published by atlantis press copyright: the authors 137 zhang and zhou 刘靖明,韩丽川,侯立文.基于粒子群的 k 均值聚类 算法[j].系统工程理论与实践,2005,25(6):54–58. [17] kaufman l and rousseeuw p j (eds.), finding groups in data: an introduction to cluster analysis, 1nd edn. (john wiley and sons, new york, 1990, 64–75). [18] li chunsheng and wang yaonan, new initialization method for cluster center, j. control theory and applications. 27(10) (2010) 1435–1440. 李春生,王耀南.聚类中心初始化的新方法[j].控制 理论与应用,2010,27(10):1435–1440. [19] katsavounidis i, jay kuo c.-c. and zhang zhen, a new initialization technique for generalized lloyd iteration, j. ieee signal processing letters. 1(10) (1994) 144–146. [20] xiong zhongyang, chen ruotian and zhang yufang, effective method for cluster centers' initialization in k-means clustering, j. application research of computers. 28(11) (2011) 4188–4190. 熊忠阳,陈若田,张玉芳.一种有效的 k-means 聚类 中 心 初 始 化 方 法 [j]. 计 算 机 应 用 研 究,2011,28(11):4188–4190. [21] phanendra babu g and narasimha murty m, a nearoptimal initial seed value selection in k-means algorithm using a genetic algorithm, j. pattern recognition letters. 14(10) (1993) 763–769. [22] bradley p s and fayyad usama m, rifining initial pionts for k-means clustering, in proceedings of the fifteenth international conference on machine learning, (morgan kaufmann publishers, san francisco, ca., 1998), pp. 91–99. [23] friedman j h and turkey j w, a projection pursuit algorithm for exploratory data analysis, j. ieee transactions on computer. 23(9) (1974) 881–890. [24] huber p j, projection pursuit (with discussions), j. the annals of statistics. 13(2) (1985) 435–475. [25] gan wenyan and li deyi, hierarchical clustering based on kernel density estimation, j. journal of system simulation. 16(2) (2004) 302–305, 309. 淦文燕,李德毅.基于核密度估计的层次聚类算法 [j].系统仿真学报,2004,16(2):302–305,309. [26] fu qiang and zhao xiaoyong (eds.), the principle and application of projection pursuit model, 1nd edn. (science press, peking, 2006, 1–119). 付强,赵小勇.投影寻踪模型原理及其应用[m].北 京:科学出版社,2006.1–119. [27] jin juliang and ding jing (eds.), water resources systems engineering, 1nd edn. (sichuan science and technology press, chengdu, 2002, 37–179). 金菊良,丁晶.水资源系统工程[m].成都:四川科学 技术出版社,2002.37–179. [28] lai yuxia and liu jianping, optimization study on initial center of k-means algorithm, j. computer engineering and applications. 44(10) (2008) 147– 149. 赖玉霞,刘建平.k-means 算法的初始聚类中心的 优化[j].计算机工程与应用,2008,44(10):147–149. [29] parzen e, on estimation of a probability density function and the mode, j. the annals of mathematical statistics. 33(3) (1962) 1065–1076. [30] li zhuyu, lu wanbo and gong jinguo (eds.), the non-parametric estimation techniques in economic, financial econometrics, 1nd edn. (science press, peking, 2007, 7–58). 李竹渝,鲁万波,龚金国.经济、金融计量学中的非 参数估计技术[m].北京:科学出版社,2007.7–58. [31] liu liping and meng zhiqing, an initial centrepoints selection method for k-means clustering, j. computer engineering and applications. 40(8) (2004) 179–180. 刘立平,孟志青.一种选取初始聚类中心的方法[j]. 计算机工程与应用,2004,40(8):179–180. [32] zhang mu and zhou zongfang, an improved topsis method based on connection degree, j. systems engineering. 26(8) (2008) 102–107. 张目,周宗放.一种基于联系度的改进 topsis 法 [j].系统工程,2008,26(8):102–107. [33] silverman b w (eds.), density estimation for statistics and data analysis, 1nd edn. (chapman and hall, london, 1986, 43–60). published by atlantis press copyright: the authors 138 journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 123–127 doi: https://doi.org/10.2991/jracr.k.191024.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article re-examining political risk assessments in volatile regions ghaidaa hetou* department of political science, rutgers university, istrategic llc, galloway, nj 08205, usa department of political science, rutgers university, new brunswick, nj 08854, usa 1. introduction political risk assessments are increasingly becoming a business imperatives for corporations interested in expanding their business operations into developing and emerging markets, which usually exhibit political uncertainty. political risk associated with market entry and subsequent business operations is a slowly evolving subcategory of risk analysis, and one that has not lent itself to strict mathematical modeling and probability assessments. to counter the subjective nature of non-scientific assessments, conventional political risk analysis and country risk indexes aimed at standardizing the process within formulas to account for political and social factors, mimicking thereby a standardized approach, applicable to all countries. the persisting shortcoming in this process is that political and social factors are necessarily context specific, i.e. interconnected and part of a complex adaptive system, thereby cannot be correctly evaluated as independent static variables. the resulting indexes hence remained at an artificial interval level spanning from low to high risk, and in most cases devout of valuable and relevant content for the private sector or government agencies. yet political risk, including cultural, social and economic factors are crucial component of evaluating a proposed investment environment, not to mention their utility for security and governance, all of which underscore the demand for developing reliable political risk analysis and processes. there are three justifications to re-evaluate political risk assessments: • providers of political risk insurance such as overseas private investment corporation (opic) and zurich insure against a specific set of political risks: expropriation, political violence including wars and terrorism, and currency inconvertibility. business entities can incur costs and losses at earlier stages, from different sources, before these extreme events manifest. • political and socio-economic variables have open continuous payoffs, because their statistical distribution does not have an upper bound. hence, risk assessment and risk management cannot rely on probability distributions.1 • corporations need to be proactive in volatile regions and fend against increasing losses related to political and geopolitical risks that are not governed by market and finance logic. on the path of conceptually developing this crucial subcategory of risk analysis, this paper answers the following foundational question: how can we first conceptualize and integrate the nature of political risk in order to derive a better approach to risk characterization and improve assessment methodology? this paper takes on this question and suggests a path forward through the example of the middle east and north africa region. 2. understanding the nature of political risk the intersection of politics and international business continues to be a focal point of interest to the business community and government agencies alike. while [1] contend that no universally accepted definition of what constitutes political risk exists [1], political risk is generally defined as the possibility of a business incurring loss or harm due to political changes in a country whether internal or a r t i c l e i n f o article history received 19 may 2019 accepted 20 august 2019 keywords risk characterization complex adaptive systems middle east and north africa risk assessment risk analysis a b s t r a c t conventional country risk and political risk indexes, to formalize the process, have attempted to standardize and generalize assessment models for factors that are highly context specific. hence, the value derived from traditional political risk indexes lack precision and are therefore less reliable. this paper re-examines political risk analysis and explains how understanding the topology and nature of political risk in emerging and developing markets is a crucial advancement in developing political risk analysis for the private sector and government agencies. particular focus is given to develop political risk characterization as a risk analysis category. to bridge the conceptual gap between risk assessment and risk management, this paper proposes the concept of complex adaptive systems as the backdrop for emerging political risk scenarios. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: ghaidaa@i-strategic.com 1continues payoffs exist within an interval, not a finite set such as x: q → [a, ∞) v (−∞, b] v (−∞, ∞) see nassim nicholas taleb in fooled by randomness and on the statistical differences between binary forecasts and real world payoffs. https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:ghaidaa%40i-strategic.com?subject= 124 g. hetou / journal of risk analysis and crisis response 9(3) 123–127 external. while modes of entrance in international markets can range from exports, licensing, franchising, joint venture or wholly owned subsidiaries, political risk is a present factor in each entrance mode at different degrees. the literature identifies sources of political risk as either emanating from environmental factors, or from constrains on operations such as discriminatory taxes and foreign direct investment regulations [2]. scholars have attempted to frame political risk through two distinct approaches and analysis types. first type focuses on the country’s attributes, processed as independent variables, such as evaluating political stability, corruption, and social cohesion and then estimating the probability of occurrence and impact ([3], p. 7). the second type evaluates and projects the behavior of government and economic functions through future oriented risk scenarios ([3], p. 9). both approaches however attempt to frame political risk through simplifying and abbreviating a complex set of circumstances. scholars have also realized that political risk reaches beyond the conventional risks of confiscation, asset recovery, expropriation and nationalization [4–7]. the literature is increasingly pointing to the complexity and interrelated nature of political risk factors as these factors exhibit a dynamic relation with the operational context. this realization however prompts us to question the utility of generalizable risk assessment models. does corruption or ethnic tension in country a have a similar effect on business operations as corruption and ethnic tension in country b? if the answer is no, due to the contextual and interrelated nature of each risk factor, then generalized risk assessment models that codify the probability and impact of each so called “independent variables”, across diverse set of countries, are of minimum value to corporate and government decision makers. in order to arrive at a better understanding and appreciation of political risk, let us examine its nature first. to understand the nature of something is to understand its defining characteristics, which prompts us to localize our understanding of political risk, and proceed in the opposite direction of generalization. within the context of the middle east and north africa (mena) region, political economic and social risk factors are products of their respective environments and are in constant dynamism, both shaping and being shaped by a complex adaptive system. the so called arab spring in 2011 for example had far reaching consequences within mena countries and across the region. in this case, evaluating economic disparity, corruption and regime type would not be adequate to construct a comprehensive evaluation of the types of risks a company may face in country a with regards to social and political instability. this distinct interrelated and dynamic nature can be summarized through the main characteristics of political risk: • interrelated factors, in a nonlinear relation, that loose representation value when studied as independent variables. • risk factors exist in a complex system (region/country) in which the sum of individual components is not a reflection of the whole system. • the context within which risk factors exist is an adaptive system that adjusts to disruption or intervention, which in turn changes the effects and impact of these risks. • emergent system properties and cascading effects cannot be predicted from precedent and historical data. • second and third order effects are a core characteristic. to bring this understanding of the nature of political risk from the abstract to practical considerations, this paper will proceed with a discussion of regions as complex adaptive systems, and integrate assessment methods, in order to develop an integrated understanding of political risk characterization. 3. regions as complex adaptive systems regions, such as latin america, western europe or the mena, have distinctive characteristics stemming from their history, geography, geopolitics, markets, governing structures, and deterministic historical development path of their respective polities. observed similarities, and shared culture and norms within each region is a crucial representation of generational survival and evolution of norms, heuristics and historical memory. the background within which regulations, laws, trade, governing regimes, corruption, market competition, capital flows and other factors exist is part of the complex context of the region, state and market sector. the concept of complex adaptive systems (cas) is useful as it captures a fundamental characteristic that the behavior, or the politics of the system, cannot be accurately identified by the observed trends of its components without nesting the observations in context. in this sense, understanding the limitations of mathematical modeling or quantitative simulations of uncertainties and risks in complex systems is key. a number of scholars have eluded to these limitations, and contended that mathematical models are less than useful in social sciences, this finding was mostly effectively proven by gödel’s incompleteness phenomenon2 [8,9]. researchers have also pointed out that normal probability distribution, most prominent in linear regression analysis among others, are not representative of the real world because these distributions underestimate the occurrence of risk categories of “known unknowns” and “unknown unknowns.”3 fat tail probability distributions on the other hand account for such realities, yet researchers contend that forecasting is not possible in complex systems [10]. in order to identify risk factors and mitigate their possible impact, investing in acquiring region and country specific data and indigenous and local knowledge is a core component of an effective analysis. in addition to giving priority to context in mena when conducting political risk analysis, local perceptions of what constitutes ethnic strife, economic equity, foreign intervention, and elite monopoly for example are part of acquiring a local understanding of the operational environment. the point is to give value to indigenous knowledge that has, over generations, not only survived but internalized and became the transmitters of constant feedback loops within cas, where emergent properties allow the culture and polity to adapt and shape future states. middle east and north africa region is also a strategic competitive space for global powers; the united states, china and russia, as well as between regional powers such as turkey, iran, israel and saudi arabia. the dynamics of this competition, which represents opportunities and constrains for each mena country, affects and shapes the political risk environment in this complex system. 2gödel’s incompleteness theorem states that “all consistent axiomatic formulations of number theory include undecidable propositions.” 3known unknowns and unknown unknowns were two terms revived by defense secretary donald rumsfeld in february 2002 in a briefing to the us department of defense. g. hetou / journal of risk analysis and crisis response 9(3) 123–127 125 4. conventional risk assessment methods kaplan and garrick [11–13] established the set definition of risk, as r: r s l xi i i= < >{ }, , where si is the ith risk scenario, l is the likelihood of that scenario, and x is the resulting consequences. researchers have traditionally leaned toward objective approaches to risk assessment. advanced quantitative methods such as monte carlo simulations have focused on repeated random sampling, and capturing the nonlinear evolution of an equation by drawing from a sequence of probability distributions [14]. monte carlo simulations have been used to anticipate financial risk, with regards to political risk however it carries little value for practitioners without understanding and connecting results to the context. other less sophisticated quantitative methods are not representative of complex systems as they still assume a normal probability distribution, expected value (average/mean) and linear associations, despite the ability of incorporating trends and seasonality. a combination of quantitative and qualitative methods is also used in political risk assessment such as the cross-impact matrix method [15] as an example, in addition to scenarios and decision trees. in the cross-impact matrix method, after an initial set of possible events is compiled, and experts estimate the initial probability of each event in isolation, a conditional probability is estimated satisfying the consistency requirement as follows: p p p p c p c ( ) ( ) ( )1 2 1 2 2 1 2 = ×     + ×     where: p(1) = probability that event 1 will occur; p(2) = probability that event 2 will occur; p(1/2) = probability of event 1 given the occurrence of event 2; p(2c) = probability that event 2 will not occur; and p(1/2c) = probability of event 1 given the nonoccurrence of event 2. other more straightforward methods rely on weighing the impact of possible risk factors across countries, and then aggregate and rank the risk score along the spectrum of high, medium to low risk. these simplistic approaches and reliance on current and historical data however limits the predictive capability of quantitative and qualitative models unless we are dealing with a stable environment, where the future state of the system is expected to resemble its past. this particular caveat is a crucial determinant in the urgency of re-evaluating political risk assessment in volatile regions such as mena. in addition and as mentioned in the introduction, political and socio-economic variables have open continuous payoffs, because their statistical distribution does not have an upper bound. hence, risk assessment and risk management cannot rely on probability distribution and historical data solely. monitoring and understanding historical data on certain factors however remains crucial and is part of the quantitative assessment, here is an example of such data points in table 1. unlike relatively stable environments, mena is prone to chronic instability that manifests in its security, economic, social and political environments, which have immediate and long term repercussion on international business, and strategic foreign government policy making. by examining the fragility index for 2019, the majority of mena states are on the 80–120 scale on the fragility spectrum (0–120), where a score of 120 denotes a failed state [16]. mena’s inherent regional characteristics are structural, hence, they shape and influence outcomes, yet these structural characteristics are less susceptible to change. for example the evolution of state-system and central governments in mena after wwi, the underdevelopment of national identities, the oil resource dependent economies, and ethnic and religious heterogeneity are some of the structural constrains within which political, social and economic policy is exercised by regional governments. given the aforementioned, how can we integrate the main defining characteristics of political risk in mena so as to reflect the complex and dynamic reality of the operational environment in this region? the following section proposes a revised definition and positioning of risk characterization on the risk analysis continuum. 5. advancing the concept of political risk characterization risk analysis literature describes risk characterization as the bridge between risk assessment and risk management. risk characterization is defined as integrating information from the risk assessment stage and synthesizing an overall conclusion of the main findings, including the context of those findings [17]. most prominent in health, environmental, cyber, and infrastructure risk analysis, risk characterization is the decisive step prior to setting strategies to manage and table 1 | sample of data points for political risk assessment in mena political economic social trends of civil unrest real gdp growth (%) religious and ethnic tensions historical military in politics presidential elections unemployment/employment trends by category poverty and marginalized communities by district legislative and bureaucratic reliability foreign debt as % of gdp and debt service as a % of xgs and types of loans youth mobilization and civil society regional conflict spillovers, refugee populations inflation powers brokers within society religious and civil monitoring elite monopolies by sector foreign reserves and government budget deficits intelligence apparatus-community relations counter-terrorism policy oil and non-sectors media, social networks and press control regional alliances regulations and implementation trends by sector islam-christian and minority relations © istrategic llc. 126 g. hetou / journal of risk analysis and crisis response 9(3) 123–127 mitigate these risks. re-evaluating and redefining risk characterization to better fit the nature of political risk is a crucial step forward, as risk characterization precedes risk management and risk mitigation strategies, which in turn are the main objectives of international corporations and foreign governments interested in mena. on the risk analysis continuum, political risk characterization has to exist within a vertical and along a horizontal dynamic. first, the vertical section comes after identifying non-market capabilities and risk tolerance of an organization, and then identifying its main objectives in the host country. then, a vertical dynamic integrates two subcategories in constant adaptive mode, incorporating the analogy of feedback loops to denote how the context (complex system’s evaluation) shapes and adjusts risk assessments (data dominant), and how preliminary risk estimations have to go through the litmus test of a system’s evaluation. only when risk assessments have made a few system’s evaluation rounds are they ripe and realistic enough to inform a set of risk scenarios. even then, a horizontal dynamic positions risk scenarios for a possibility of a re-evaluation through the vertical risk characterization dynamic. the following illustrates how we may redefine risk characterization to reflect and integrate the complex nature of political risk within political risk analysis. unlike conventional approaches to political risk analysis, this illustration combines both analytic and quantitative methods while accounting for the context (complex system’s evaluation) as a vital source of feedback to adjust quantitative risk assessments. only after a satisfactory completion of this vertical dynamic process can we proceed with risk characterization to inform risk scenario formations. the following step after risk characterization is: risk scenarios, which ought to be perceived as an integral stage of political risk analysis (and not a substitute or alternative risk assessment method). through risk scenario planning can the full potential of previous assessed risks be displayed through plausible narratives, accounting for second and third order effects. the following section gives an overview of how risk scenarios are framed. 6. modeling political risk scenarios a number of scholars have explored how to integrate the reality of complex environments into scenario planning. scenario analysis (sa) has been defined as “..a description of a possible set of events that might reasonably take place. the main purpose of developing scenarios is to stimulate thinking about possible occurrences, assumptions and risks, and courses of action” [18]. another definition describes sa as “..conjunctures about what might happen in the future” [19]. sa’s objectives may range from exploratory to advisory and primary support for decision-makers [20]. the stage of scenario planning is crucial as it generates an ordering mechanism for possible alternative future environments for the decision-maker. chermack [21] argued that the scenario planning stage is best modeled and understood through systems theory, in his study however he argues that the organization itself is conceptualized as a system and is assumed to self-organize through feedback networks. while this is a valuable contribution to sa literature, the operational environment in mena requires at outward oriented systems’ approach as well. scenario planning may be prompted by a diverse set of questions, and the most raised question in sa is “what can go wrong?.” the scenario organizing mechanism of hierarchical holographic modeling (hhm) is useful to demonstrate one available framing tool to answer this question. hhm is regarded as “..a holistic philosophy/methodology aimed at capturing and representing the essence of the inherent diverse characteristics and attributes of a system” [22,23]. the term holographic refers to the multiview image of the system that include views of economic risk, social risk, political risk, …etc. the term hierarchical points to the importance of understanding risk factors at all levels of the system from the macro to the micro level, which in turn organizes scenarios into sets and subsets. an added development to sa was the russian method of anticipatory failure determination where the question, instead of: what can go wrong?, becomes: what should happen for things to go wrong? then both types of scenario planning questions were added in what we now call the theory of scenario structuring [24,25]. when a corporation is interested in understanding and investigating the political landscape in a potential host country in mena, the importance of a comprehensive risk evaluation becomes crucial. referring to the risk analysis model presented in this paper, in volatile regions, data analysis is not sufficient by itself, hence the vertical risk assessment dynamic has to be integrated with a context/system’s evaluation in order to arrive at comprehensive political risk characterization. the analogy of feedback loops within this vertical dynamic illustrates how quantitative risk assessments have to be nested and checked against the context where these risk factors have the potential to cause harm or create opportunities. only then can a scenario analysis, that answers both questions of: what can go wrong? and what should happen for things to go wrong? become a sufficient representation of possible alternative future scenarios for the decision-maker, ranked in a hierarchy of likely occurrence (see figure 1). figure 1 | political risk analysis model. g. hetou / journal of risk analysis and crisis response 9(3) 123–127 127 7. conclusion in volatile regions like mena, political risk analysis becomes an imperative for corporations, ngos and foreign governments. yet simple, standardized, one-dimensional approaches to political risk assessment are not only insufficient but downright harmful to the objective of sustaining and expanding business operations. this paper aimed at contributing to the development of political risk analysis by re-evaluating the concept of risk characterization as the gateway to integrate predominantly quantitative risk assessment methods with a complex system’s evaluation. understanding regions and states as complex systems necessitates grounding data in indigenous knowledge and intelligence. the main objective of improving political risk analysis is to equip the corporate leadership and government policy makers with political risk management and risk mitigation strategies. understanding and anticipating disruption by proactively monitoring warning signs is an essential operations mode in volatile regions. developing professional corporate adaptive and proactive capabilities tailored for the political environment in a host mena country is a competitive advantage in a region ripe with opportunities. conflicts of interest the author declare they have no conflicts of interest. references [1] hood j, shahid nawaz m. political risk exposure and management in multi-national companies: is there a role for the corporate risk manager? risk manage 2004;6:7–18. [2] kobrin sj. political risk: a review and reconsideration. j int business stud 1979;10:67–80. [3] howell ld. the handbook of country and political risk analysis. nyc: public risk services group; 2001. [4] schmidt a. analyzing political risk. business horizon 1986;29: 43–50. [5] howell ld, chaddick b. models of political risk for foreign investment and trade: an assessment of three approaches. columbia j world business 1994;29:70–91. [6] sottilotta ce. political risk: concepts, definitions, challenges. working paper series, luiss guido carli school of government; 2013 (sog-wp6/2013). [7] sottilotta ce. political risk assessment and the arab spring: what can we learn?. thunderbird inter busi rev 2015;57:1–12. [8] hofstadter dr. gödel, escher, bach: an eternal golden braid. new york: vintage books; 1989, p. 17. [9] doria fa. the limits of mathematical modeling in the social sciences: the significance of gödel’s incompleteness phenomenon. singapore: world scientific publishing company; 2017. [10] taleb nn. the black swan: the impact of the highly improbable. ny: random house; 2010. [11] kaplan s, garrick bj. on the quantitative definition of risk. risk analysis 1981;1:11–27. [12] kaplan s. the general theory of quantitative risk assessment its role in the regulation of agricultural pests. in: proceedings of the aphisa/nappo international workshop on the identification, assessment, and management of risks due to exotic agriculture pests. 1993, pp. 123–46. [13] kaplan s. an introduction to triz. the russian theory of incentive problem solving. southfield, mi: ideation international inc.; 1996. [14] kolokoltsov vn. nonlinear markov processes and kinetic equations. uk: cambridge university press; 2010. [15] gordon tj. cross-impact method. ny: ac/unu millennium project; 1996. [16] fund for peace. fragile states index annual report 2019. available from: http://fundforpeace.org/wp-content/uploads/2019/04/ 9511904-fragilestatesindex.pdf (last accessed may 1, 2019). [17] usepa. guidance for risk characterization. washington, d.c.: united states environmental protection agency, science and policy council; 1995. [18] jarke m, tung bx, carroll jm. scenario management: an interdisciplinary approach. j req eng 1998;3:155–73. [19] cornish e. futuring: the exploration of the future. bethesda, md: world future society; 2004, p. 313. [20] tourki y, keisler j, linkov i. scenario analysis: a review of methods and applications for engineering and environmental systems. environ syst decisions 2013;33:3–20. [21] chermack t. the role of system theory in scenario planning. j future stud 2004;8:15–30. [22] haimes yy, tarvainen k, shima t, thadathil j. hierarchical multi objective analysis of large-scale systems. new york: hemisphere publishing co.; 1990. [23] haimes yy, sage ap. risk modeling, assessment and management. hoboken, nj: john wiley & sons; 2015. [24] kaplan s, zlotin b, zusman a, vishnipolski s. new tools for failure and risk analysis, anticipatory failure determination and the theory of scenario structuring. southfield, michigan us: ideation international inc.; 1999. [25] kaplan s, haimes yy, garrick bj. fitting hierarchical holographic modeling into the theory of scenario structuring and a resulting refinement to the quantitative definition of risk. risk anal 2001;21:807. https://doi.org/10.1057/palgrave.rm.8240169 https://doi.org/10.1057/palgrave.rm.8240169 https://doi.org/10.1057/palgrave.rm.8240169 https://doi.org/10.1057/palgrave.jibs.8490631 https://doi.org/10.1057/palgrave.jibs.8490631 https://doi.org/10.1016/0022-5428%2894%2990048-5 https://doi.org/10.1016/0022-5428%2894%2990048-5 https://doi.org/10.1016/0022-5428%2894%2990048-5 https://doi.org/10.1007/s10669-013-9437-6 https://doi.org/10.1007/s10669-013-9437-6 https://doi.org/10.1007/s10669-013-9437-6 https://doi.org/10.1111/0272-4332.215153%0d https://doi.org/10.1111/0272-4332.215153%0d https://doi.org/10.1111/0272-4332.215153%0d https://doi.org/10.1111/0272-4332.215153%0d microsoft word editor's introduction-tbc!! editor's introduction the present issue of journal of risk analysis and crisis response (jracr), volume 1, issue 2 (2011) contains 8 papers. the former 2 contributions are written in english and others in chinese with english abstracts. the papers can be divided into six categories: industry safety, internet of intelligences，investment risk, environmental risk, disaster risk and food safety risk. in the first category, the paper “efficient applications of risk analysis in the chemical industry and emergency response” by roberto bubbico, presents some of the most common applications of the risk analysis technique in the industrial sector and in the civil protection and emergency planning area. the paper “risk management in process industry practical approach in poland” by kozak, presents the weak points of risk management procedures in the process industry during hazard and operability study. a technical assessment of the industrial environment and installations is proposed as a background to rbi procedures. the second category, the “internet of intelligences in risk analysis for online services” by huang, proposed a new concept “internet of intelligences” (ioi). an ioi collects and processes the information provided by the agents with experiences and judgment in a risk. it serves for the customer who hasn’t any experience in a specific risk. the online services of ioi can help us, such as choosing health foods, correctly filling in the college candidate voluntary and guaranteeing love affairs more romantic. the paper in the environmental risk category is “environmental risk discrimination and assessment for municipal solid waste secure landfill site” by cao. it summarizes the risks in waste secure landfill site, including water environment risks, air environment risks and waste pile sliding risks, which are results caused by anthropogenic, equipment and natural factors. there are two papers in the investment risk category, the paper “the errors estimate of the multistage combined investment risk assessment” by zhou and yu, discusses the estimation of the errors from the simple average method and the general mcira model. meanwhile, the mathematics analysis techniques are applied to determine the existent maximum of the errors square sum of the general mcira models. the paper “research on model for evaluating risks of venture capital projects” by li and zhou, establishes an index system for evaluating the risks of venture capital projects and presents a model for evaluating the risk degree of single venture capital project by applying a hierarchy process and the ewaa operator. journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 90-91 published by atlantis press copyright: the authors 90 in the disaster risk category. the paper “the research of a hail risk evaluation under imperfect information” by wang, zhou and du, sets up a hail risk evaluation method and utilizes a nonlinear finite element analysis software to simulate the case that hails dash on the steel body and glass material to evaluate hail risks. the paper “dynamics of urban fire correlations with detrended fluctuation analysis” by wang, sun and lo, uses the detrended fluctuation analysis to study the long-range correlations of urban fires in both small and large spatial scales with a case of chinese city. it is found that the time series of direct economic loss caused by urban fires presents stable anti-power law correlations, while the time interval series of urban fire occurrence presents persistent long-range power-law correlations with two scaling exponents. the last category includes one paper “study on food safety risk analysis under the condition of traditional agriculture” by yin, gao and pei, discusses the problems of food safety and suggests a framework to analyze food safety risk under the condition of traditional agriculture. as a new international journal, jracr will to attract various contributions in risk analysis (ra) and crisis response (cr) from authors around the world in sharing their research findings and experience. let us work hard to construct a safety world! editor-in-chief chongfu huang published by atlantis press copyright: the authors 91 a note on nonparametric estimation of conditional hazard quantile function el hadj hamel, nadia kadiri, abbes rabhi laboratory of mathematics, university djillali liabes of sidi bel abbés, po. box. 89, sidi bel abbés 22000, algeria e-mail: rbe0222@yahoo.com, nad.kad06@yahoo.com, rabhi abbes@yahoo.fr abstract in this paper, we study an kernel estimator of the conditional hazard quantile function (chqf) of a scalar response variable y given a random variable (rv) x taking values in a semi-metric space and using the proposed estimator based of the kernel smoothing method. the almost complete consistency and the asymptotic normality of this estimate are obtained when the sample is an independante sequence. keywords: asymptotic normality, conditional hazard quantile function, functional data, kernel smoothing, nonparametric estimator. 1. introduction the goal of this paper is to study a nonparametric estimator of the chqf when the explanatory variable is functional. this is motivated by the increasing number of situations in which the collected data are curves (consecutive discrete recordings are aggregated and viewed as sampled values of a random curve) where it used to be numbers and vectors. functional data analysis (see ferraty and vieu (2006)) can help to analyze such data sets in a nonparametric framework. recently, many authors are interested in the estimation of conditional quantiles for a scalar response and functional covariate. ferraty et al. (2005) introduced a nonparametric estimator of conditional quantile defined as the inverse of the conditional cumulative distribution function when the sample is considered as an α -mixing sequence. they stated its rate of almost complete consistency and used it to forecast the well-known el niño time series and to build confidence prediction bands. ezzahrioui et al. (2008) established the asymptotic normality of the kernel conditional quantile estimator under α -mixing assumption. recently, and within the same framework, dabo-niang and laksaci (2012) provided the consistency in lp norm of the conditional quantile estimator for functional dependent data, bouchentouf et al. (2015) provided the consistency and asymptotic normality of the smoothing conditional quantile density function. in an earlier contribution of the estimator of the chqf (see sankaran and unnikrishnan (2009)) we established the consistency and asymptotic normality of the kernel smoothing estimator for the independent sequence and reel case. the present work gives a generalization to the functional data, we investigate the asymptotic properties and the asympjournal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 101 received 14 march 2017 accepted 18 april 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). totic normality of the chqf of a scalar response and functional covariate. the interest comes mainly from the fact that application fields for functional methods need to analyze continuous-time stochastic processes. in what follows, the rest of the paper is organized as follows. section 2 we present our estimation procedure and recall the definition of the the functional kernel estimates property. section 3 formulates main results of strong consistency (with rate) and asymptotic normality of the estimator with gives proofs of the main results. section 4 is devoted provides a brief conclusion of the study. 2. the functional kernel estimates we consider a random pair (x,y ) where y is valued in r and x is valued in some infinite dimensional semi-metric vector space (f,d(·,·)). let (xi,yi),i = 1,...,n be the statistical sample of pairs which are identically distributed like (x,y ),but not necessarily independent. from now on, x is called functional random variable f.r.v. let x be fixed in f and let fy |x (y,x) be the conditional cumulative distribution function (cond-cdf) of y given x = x, is defined by: ∀y ∈ r,fy |x (x,y) = p(y = y|x = x). let qy |x (γ) be the γ -order quantile of the distribution of y given x = x. from the cond-cdf fy |x (·,x), it is easy to give the general definition of the γ -order quantile: q(γ|x = x) ≡ qy |x (γ) = in f {t : fy |x (t,x) = γ},0 6 γ 6 1. then, the definition of conditional quantile implies that fy |x (qy |x (γ)) = γ. on differentiating partially w.r.t. γ we get fy |x (qy |x (γ)) = 1 ∂ ∂ γ (qy |x (γ)) . thus, the condition quantile density function can be written as follows qy |x (γ) = 1 fy |x (qy |x (γ)) . let us now, define the kernel estimator f̂y |x (·,x) of fy |x (·,x) f̂y |x (x,y) = n ∑ i=1 k(h−1k d(x,xi))h(h −1 h (y −yi)) n ∑ i=1 k(h−1k d(x,xi)) . (1) where k is a kernel function, h a cumulative distribution function and hk = hk,n(resp.hh = hh,n) a sequence of positive real numbers. roussas (1969) introduced some related estimate but in the special case when x is real, while samanta (1980) produced previous asymptotic study. as a by-using of nair and sankaran (2009) and xiang (1995), it is easy to derive an estimator q̂y |x of qy |x : qy |x (γ) = inf{t : f̂y |x (t,x) = γ} = f−1y |x (qy |x (γ)). let now defined the conditional density function is the derivative of conditional distribution function. f̂y |x (x,y) = h−1h n ∑ i=1 k(h−1k d(x,xi))h(h −1 h (y −yi)) n ∑ i=1 k(h−1k d(x,xi)) . (2) parzen (1979) and jones (1992) defined the quantile density function as the derivative of q(γ), that is, q(γ) = q′(γ). note that the sum of two quantile density functions is again a quantile density function. nair and sankaran (2009) have defined the hazard quantile function as follows: r(γ) = r(q(γ)) = f (q(γ)) 1 − f(q(γ)) = ((1 − γ)q(γ))−1. (3) thus hazard rate of two populations would be equal if and only if their corresponding quantile density functions are equal. this has been used to construct tests for testing equality of failure rates of two independent samples. now, from this definition, let journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 102 us introduce the γ -order conditional quantile of the conditional hazard function r(γ) = ry |x (qy |x (γ)) = fy |x (qy |x (γ)) 1 − fy |x (qy |x (γ)) (4) =((1 − γ)qy |x (γ))−1. (5) consequently, the conditional quantiles of conditional hazard function operator is defined in a natural way and can be estimated by using kernel smoothing methods by rn(γ) = r̂y /x (q̂y /x (γ)) = f̂y /x (q̂y /x (γ)) 1 − f̂y /x (q̂y /x (γ)) . (6) now we proposed the other estimator of rn(γ) using the kernel smoothing method, define by: rn(γ) = 1 hh ∫ 1 0 1 [1 − f̂y |x (q̂y |x (t))].q̂y |x (t) h ( t − γ hh ) dt. (7) in the next section derive the asymptotic properties of our conditional quantile hazard function 3. assumptions and main results 3.1. general assumptions our results are stated under some assumptions we gather hereafter for easy reference. (h1) for all h > 0, p(x ∈ b(x,h)) =: ϕx(h) > 0. moreover, ϕx(h) > 0 −→ 0 as h −→ 0. (h2) for all i ̸= j ,0 < supi ̸= j p[(xi,x j) ∈ b(x,h)× b(x,h)] = p(wi 6 h,wj 6 h) 6 ψx(h), where ψx(h) −→ 0 as h −→ 0. furthermore, we assume that ψx(h) = o(ϕ x2 (h)). (h3) h is such that, for all (y1,y2) ∈ r2, |h(y1)− h(y2)| = c|y1 − y2| and its first derivative h(1) verifies∫ |t|b2 h(1)(t)dt < ∞. (h4) k is a nonnegative bounded kernel of class c1 over its support [0,1] such that k(1) > 0. the derivative k′ exists on [0,1] and satisfy the condition k′(t) < 0 , for all t ∈ [0,1] and∫ 1 0 (k) j(t)dt < ∞ for j = 1,2. (h5) lim n−→∞ hk = 0 with lim n−→∞ logn nϕx(hk) = 0. remark 1. hypothesis (h1) is the classical concentration assumption.(h3) allows to get the convergence rate in the independent case. assumption h3 is classical in nonparametric estimation and is satisfied by usual kernels such as epanechnikov, biweight, whereas the gaussian density k is also possible, it suffices to replace the compact support assumption by: ∫ rd |t| b2 h(t)dt < ∞ . assumption h3 ensures the existence and uniqueness of the quantile estimate qγ (x), see ferraty et al. (2005). a mild regularity hypothesis (h4) is assumed for the distribution function. hypothesis (h3) is technical and is imposed only for the brevity of proofs. finally the choice of bandwidth is given by (h5). 3.2. asymptotic properties in this section, we prove strong consistency and asymptotic normality of the estimator 7. theorem 1. let fy |x be continuous.assume that k(·) satisfies the conditions (h1)-(h5) in secestimator rn(γ) is uniformly strong consistent. proof. we can write equation (7) as journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 103 rn(γ) = 1 hh ∫ 1 0 h ( t − γ hh ) dt [1 − f̂y |x (q̂y |x (t))].q̂y |x (t) − 1 hh ∫ 1 0 h ( t − γ hh ) dt [1 − fy |x (qy |x (t))].q̂y |x (t) + 1 hh ∫ 1 0 h ( t − γ hh ) dt [1 − fy |x (qy |x (t))].q̂y |x (t) = 1 hh ∫ 1 0 h ( t − γ hh ) 1 q̂y |x (t) [ 1 [1 − f̂y |x (q̂y |x (t))] − 1 [1 − fy |x (qy |x (t))] ] dt + 1 hh ∫ 1 0 h ( t − γ hh ) dt [1 − fy |x (qy |x (t))].q̂y |x (t) = 1 hh ∫ 1 0 h ( t − γ hh ) f̂y |x (q̂y |x (t)− fy |x (qy |x (t))dt q̂y |x (t)[1 − fy |x (qy |x (t))][1 − f̂y |x (q̂y |x (t))] + 1 hh ∫ 1 0 h ( t − γ hh ) dt [1 − fy |x (qy |x (t))]q̂y |x (t) (8) since sup t |f̂y |x (t)− fy |x (t)| −→ 0 almost surely, equation (8) is asymptotically equal to rn(γ) = 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 − fy |x (qy |x (t))q̂y |x (t) . thus, rn(γ)− r(γ) = 1 hh ∫ 1 0 h ( t − γ hh ) 1 (1 − fy |x (qy |x (t)))[ 1 q̂y |x (t) − 1 qy |x (t) ] dt+ 1 hh ∫ 1 0 h ( t − γ hh ) 1 qy |x (t) dt (1 − fy |x (qy |x (t))) − 1 (1 − γ)qy |x (γ) = 1 hh ∫ 1 0 1 (1 −t) h ( t − γ hh ) [ qy |x (t)− q̂y |x (t) ] dt q̂y |x (t)qy |x (t) dt + 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 − fy |x (qy |x (t)))qy |x (t) − 1 (1 − γ)qy |x (γ) (9) denoting k∗(t,γ) = (h((t − γ)/hh))/(1 − t)q̂y |x (t)qy |x (t), on using integration by parts, equation (9) reduces to rn(γ)− r(γ) = 1 hh ∫ 1 0 (qy |x (t)− q̂y |x (t))dk∗(t,γ)+ 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 −t)qy |x (t) − 1 (1 − γ)qy |x (γ) . since supt |q̂y |x (t) − qy |x (t)| −→ 0 almost surely, equation (10) is asymptotically equal to journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 104 rn(γ)− r(γ) = 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 − t)qy |x (t) − 1 (1 − γ)qy |x (γ) . (10) setting (t − γ)/hh = v, in equation (10), rn(γ)− r(γ) = 1 hh ∫ (1−γ)/hh −γ/hh h(v) fy |x qy |x (γ + vhh) 1 −(γ + vhh) dv − 1 (1 − γ)qy |x (γ) = 1 hh ∫ (1−γ)/hh −γ/hh h(v) 1 1 − γ [ 1 − vhh 1 − γ ]−1 fy |x qy |x (γ + vhh)dv − 1 (1 − γ)qy |x (γ) . (11) by taylor’s series expansion of qy |x (γ + vhh) around γ , equation (11) becomes rn(γ)− r(γ) = 1 hh(1 − γ) ∫ (1−γ)/hh −u/hh ( h(v) [ 1 + vhh 1 − γ +······ ] × fy |x [ qy |x (γ)+ vhh dqy |x (γ)+ ······ ]) dv − 1 (1 − γ)qy |x (γ) . (12) as n −→ ∞,we have hn −→ 0 and ∫ +∞ −∞ h(v)dv = 1, so that equation (12) reduces to |rn(γ)− r(γ)| = ∣∣∣∣ fy |x (qy |x )(γ)1 − γ − 1(1 − γ)qy |x (γ) ∣∣∣∣, which tends to zero as n −→ ∞. this completes the proof. 3.3. asymptotic normality in this section we give the asymptotic normality of rn(γ). theorem 2. under assumptions (h1)(h5) and suppose that fy |x is continuous, for 0 < γ < 1,√ n(rn(γ) − r(γ)) is asymptotically normal with mean zero and variance σ 2(γ) as given in equation (13). σ 2(γ) = n 1 h2h e [∫ 1 0 qy |x (t)dm ′(t,γ) + ∫ 1 0 f̂y |x (q̂y |x )(t) m(t,γ) (1 − t) dqy |x (t) ]2 . (13) m(t,γ) = h((t − γ)/hh)/qy |x (t) and m′(t,γ) is the derivative of m(t,γ) with respect to t. proof. √ n(rn(γ)− r(γ)) = √ n 1 hh ∫ 1 0 h ( t − γ hh ) [ 1 1 − f̂y |x (q̂y |x )(t) . 1 q̂y |x (t) − 1 (1 − f̂y |x (q̂y |x )(t))qy |x (t) ] dt + √ n 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 − f̂y |x (q̂y |x )(t))qy |x (t) − √ n (1 − γ)qy |x (t) , journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 105 which can be written as √ n(rn(γ)− r(γ)) = √ n hh ∫ 1 0 h ( t − γ hh ) 1 (1 − f̂y |x (q̂y |x )(t))[ qy |x (t)− q̂y |x (t) qy |x (t)q̂y |x (t) ] dt + √ n hh ∫ 1 0 h ( t − γ hh ) 1 qy |x (t)  ( f̂y |x (q̂y |x )(t)− fy |x (qy |x )(t) ) dt (1 − f̂y |x (q̂y |x )(t))(1 − fy |x (qy |x )(t))   + √ n hh ∫ 1 0 h ( t − γ hh ) dt (1 − fy |x (qy |x )(t))qy |x (t) − √ n (1 − γ)qy |x (γ) . (14) since supt |f̂y |x (q̂y |x )(t)− fy |x (qy |x )(t)| −→ 0 and fy |x (qy |x )(t) = t, equation (14) is asymptotically equal to = √ n 1 hh ∫ 1 0 h ( t − γ hh ) (qy −x (t)− q̂y −x (t))dt (1 − t)(qy |x (t))2 + √ n hh ∫ 1 0 h ( t − γ hh ) 1 qy |x (t) 1 (1 − t)2[ f̂y |x (q̂y |x )(t)− fy |x (qy |x )(t) ] dt + √ n 1 hh ∫ 1 0 h ( t − γ hh ) dt (1 − fy |x (qy |x )(t))qy |x (t) − √ n (1 − γ)qy |x (γ) . (15) setting m(t,γ) = h((t − γ)/hh)/qy |x (t) in equation (15) and applying integration by parts, we obtain √ n(rn(γ)− r(γ)) = √ n hh ∫ 1 0 m(t,γ)qy |x (t) (1 − t)[ f̂y |x (q̂y |x )(t)− fy |x (qy |x )(t) ] dt + √ n hh ∫ 1 0 h ( t − γ hh ) dt (1 − fy |x (qy |x )(t))qy |x (t) − √ n (1 − γ)qy |x (γ) + √ n hh ∫ 1 0 (q̂y |x (t)− qy |x (t))dm′(t,γ), (16) where m′(t,γ) is the derivative of m(t,γ) with respect to t. from equation (12), we can obtain equation (16) as √ n(rn(γ)− r(γ)) = √ n hh ∫ 1 0 m(t,γ)qy |x (t) (1 − t)[ f̂y |x (q̂y |x )(t)− fy |x (qy |x )(t) ] + √ n hh ∫ 1 0 (q̂y |x (t)− qy |x (t))dm′(t,γ). (17) note that from ezzahrioui and ould-saı̈d (2008), laksaci et al. (2011) and chaouch and khardani (2015) for 0 6 γ 6 1, √ n(q̂y |x (γ) − qy |x (γ)) is asymptotically normal with mean zero and variance σ 2(γ) = 1√ ϕx(h) β2 β 21 γ(1 − γ) ( fy |x (qy |x (γ)))2 , β j = k j(1)− ∫ 1 0 (k j)′(s)τ0(s)ds, and τ0 is a nondecreasing bounded function such that, uniformly in s ∈ [0,1], ϕ(hs) ϕ(h) = τ0(s)+ o(1) as h ↓ 0 and for j > 1,∫ 1 0 ((k) j(t))′τ0(t)dt < ∞. thus √ n(f̂y |x (q̂y |x )(γ)−fy |x (qy |x )(γ)) is also asymptotically normal with mean zero and variance σ 21 (γ), since d/dγ fy |x (qy |x )(γ) = 1. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 106 now from equation (17), we can show that√ n(rn(γ) − r(γ)) is asymptotically normal with mean zero. the expression of variance can be obtained from equation (17), which is given by σ 21 (γ) = n h2h e [∫ 1 0 q̂y |x (t)dm ′(t,γ) + ∫ 1 0 f̂y |x (q̂y |x )(t) m(t,γ) (1 − t) dqy |x (t) ] .(18) this completes the proof. remark 2. • the function τ0(·) defined in by there exists a function τ0(·) s.t. for all s ∈ [0,1], lim r−→0 ϕx(sr)/ϕx(r) = ϕx(s), permits to get the variance term explicitly. this condition is classical and related to a non vanishing conditional density. the second one means that a small amount a concentration is needed in order to ensure asymptotic normality. • the present study provided a nonparametric estimator, for the conditional hazard quantile function, its based and using the kernel smoothing method, and the asymptotic properties of the kernel estimator were studied. the kernel based estimator seems to perform satisfactorily except for large values of γ . references 1. d. bosq, linear processes in function spaces, (lecture notes in statistics 149, springer-verlag, berlin, 2000). 2. a.a. bouchentouf, y. hammou, k. nedjadi and a. rabhi, nonparametric estimation of a conditional quantile density function for time series data, j. stat. appl. prob. 4(1) (2015), pp. 13–21. 3. m. chaouch and s. khardani, randomly censored quantile regression estimation usinf functional stationary ergodic data, journal of nonparametric statistics, 27 (1) (2015), pp. 65–87. 4. m. ezzahrioui, and e. ould-saı̈d, asymptotic results of a nonparametric conditional quantile estimator for functional time series, comm. statist. theory methods. 37(16-17) (2008), pp. 2735-2759. 5. s. dabo-niang, and a. laksaci, nonparametric quantile regression estimation for functional dependent data, comm. statist. theory methods, 41(7) (2012), pp. 1254–1268. 6. f. ferraty, a. laksaci, a. tadj, and p. vieu, rate of uniform consistency for nonparametric estimates with functional variables, j. stat. plann. and inf., 140 (2010), pp. 335–352. 7. f. ferraty, a. rabhi, and p. vieu, conditional quantiles for functionally dependent data with application to the climatic el niño phenomeno, sankhya: the indian journal of statistics, special issue on quantile regression and related methods, 67(2) (2005), pp. 378–398. 8. f. ferraty, a. rabhi, and p. vieu, estimation nonparamétrique de la fonction de hasard avec variables explicatives fonctionnelles, rev: rom. j. pure & applied math, 52 (2008), pp. 1–18. 9. f. ferraty, and p. vieu, nonparametric functional data analysis: theory and practice, (springer series in statistics, springer, new york, 2006). 10. m.c. jones, estimating densities, quantiles, quantile densities and density quantiles, ann. inst. statist. math., 44(4) (1992), pp. 721–727. 11. a. laksaci, m. lemdani, and e. ould-saı̈d, asymptotic results for an l1 -norm kernel estimator of the conditional quantile for functional dependent data with application to climatology, sankhya: the indian journal of statistics, 73 (a, part 1) (2011), pp. 125– 141. 12. n.u. nair, and p.g. sankaran, quantile based reliability analysis, communications in statistics theory and methods, 38 (2009), pp. 222–232. 13. g. roussas, nonparametric estimation of the transition distribution function of a markov process, ann. math. statist. 40 (1969), pp. 1386–1400. 14. m. samanta, nonparametric estimation of conditional quantiles, j. of statist. proba. lett. 7 (1989), pp. 407– 412. 15. p.g. sankaran, and n. unnikrishnan, nonparametric estimation of hazard quantile function, journal of nonparametric statistics, 21(6) (2009), pp. 757–767. 16. r.j. serfling, approximation theorem of mathematical statistics, (john wiley, new york 1980). 17. e. parzen, nonparametric statistical data modelling, journal of american statistical association 74 (1979), pp. 105–122. 18. j.o. ramsay, and c.j. dalzell, some tools for functional data analysis (with discussion), j. roy. statist. soc., ser. b (1991), pp. 539-572. 19. j.o. ramsay, and b.w. silverman, applied functional data analysis: methods and case studies, (springer series in statistics, springer-verlag, new york, 2002). 20. j.o. ramsay, and b.w. silverman, functional data analysis, (2nd ed.), (springer series in statistics, springer-verlag, new-york, 2006). 21. x. xiang, estimation of conditional quantile density function, journal of nonparametric statistics, 4 (1995), pp. 309–316. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 101–107 ___________________________________________________________________________________________________________ 107 journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 93-100 doi: https://doi.org/10.2991/jracr.k.190703.005; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). evaluation on ecological civilization construction level in guizhou based on intuitionistic fuzzy analytic hierarchy process xinpu wang1,*, mu zang2, † 1. college of big data application and economics, guizhou university of finance and economics huaxi guiyang guizhou, china 2. college of big data application and economics, guizhou university of finance and economics huaxi guiyang guizhou, china received january 7, 2019 accepted march 4, 2019 abstract in order to better build the ecological environment of guizhou province, accelerate the marketization and capitalization of guizhou province's ecological resources, and evaluate and study the construction level of guizhou's ecological civilization. this paper, by using ifahp, the establishment of intuitionistic fuzzy judgment matrix, consistency test, calculate the nine-province each secondary index to the weight of a layer and target layer, the comprehensive evaluation value is calculated and sorted. the empirical research shows that the ecological civilization construction level of guizhou province is relatively high, but the level distribution of all prefectures and cities is very uneven, and the difference between the highest score southeast guizhou and the lowest score liupanshui is 4.87. therefore, it is necessary to accelerate the construction of ecological civilization in various prefectures and cities, so as to improve the construction level of ecological civilization in the whole province and promote regional economic development. keywords: construction of ecological civilization; the level of ecological civilization construction; ifahp; comprehensive evaluation *guizhou university of finance and economics, huaxi guiyang guizhou 550025, china. †corresponding author: rim_007@163.com 1. introduction at the fifth plenary session of the eighteenth central committee of the communist party of china, the development concept of "innovation, coordination, green, open and sharing" was put forward, which put forward higher requirements for the construction of ecological civilization. the construction of ecological civilization is of great significance to solve the resource and environment problems facing our country. "green water and green hills are jinshan and yinshan". green development has become the theme of china's economic and social development in the 13th five-year plan period and even in the longer term. the key to building a resource-saving and environment-friendly society lies in the improvement of ecological environment, which plays an important role in promoting economic growth and industrial upgrading, promoting energy conservation and emission reduction, protecting the environment and improving ecological efficiency. guizhou, as a national experimental area of ecological civilization, guian new area is the first batch in china and the only national experimental area in southwest china to be allowed to carry out green financial reform and innovation. firstly, the improvement of the level of ecological civilization construction plays a positive role in promoting the marketization and capitalization of ecological resources in guizhou. focusing on the three strategies of poverty journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 94 alleviation, big data and big ecology, guizhou will focus on developing the "four-type" industries, which will bring huge investment opportunities and financial needs, provide support for guizhou's ecological construction, and promote the realization of the "three orientations" in the new area. finally, it will be conducive to fostering new momentum of economic development and providing loans for regional economic development. this paper uses intuitionistic fuzzy analytic hierarchy process (ifahp) to comprehensively evaluate the level of ecological civilization construction in nine cities and municipalities of guizhou and to provide development suggestions for improving the level of ecological civilization construction in guizhou province. 2. an overview of domestic and foreign research with the improvement of environmental awareness, great progress has been made in the evaluation of the level of ecological civilization construction. early scholar zhao fang (2010) [1] decomposed the evaluation index of ecological civilization construction into four core areas: production mode, lifestyle, ecological protection and construction, social development, and set up several specific indicators, and constructed a threelevel, 27 single evaluation index structure from "target level-criterion level-index level". the evaluation index system of ecological civilization construction was established, and the progress rate analysis method was used to evaluate the construction of ecological civilization. in recent years, zhang huan et al. (2015) [2] has established an evaluation index system of ecological civilization in mega-cities, which includes four aspects: the degree of ecological environment health, the intensity of resource and environment consumption, the efficiency of non-point source pollution control and the livability of residents, with a total of 20 indicators, and put forward measures to strengthen the construction of ecological civilization. zheng yanling et al. (2016) [3] taking hebei province as an example, this paper constructs a comprehensive evaluation index system of ecological civilization construction and regional economic development level, adopts the measurement model of ecological civilization construction and regional economic coordinated development, and puts forward that with the advancement of ecological civilization construction, the relationship between ecological civilization and economic development will enter into mutual complementarity. good circulation. cheng jinhua et al. (2018) [4] taking 35 provincial capitals and sub-provincial cities in china and 7 small and medium-sized cities in hubei province as objects, the index system is constructed from four dimensions: land and space optimization, resource and energy utilization, ecological environment protection and governance, and ecological system construction. the index weight is given by the method of entropy weight. empirical research finds that big cities are the main cities. cities with a high index of ecological civilization often have the characteristics of balanced development of the economy and environment. there are few studies on the evaluation of guizhou's ecological civilization construction, and more qualitative analysis is used to analyze the current situation of guizhou's ecological civilization construction (jiangchuan et al. [5], 2014; shipeixin [6], 2010, etc.). van laarhoven and edrycz (1983) [7] proposed for the first time to fuzzify the analytic hierarchy process (ahp). buckley (1985) [8] developed the fuzzy analytic hierarchy process (fahp) on this basis. it is more complete and accurate to evaluate the target level indicators, which can reflect the uncertainty of the decision-makers in the evaluation of objective things. the subjectivity of experts is fuzzified, which is more objective and accurate than the analytic hierarchy process, but it can not accurately express the situation of abstention or hesitation, and the quantitative and qualitative indicators need to be aggregated by different methods. in order to solve this problem, an intuitionistic fuzzy analytic hierarchy process (ifahp) based on the concept of fuzzy sets is proposed, which unifies the aggregation of qualitative and quantitative indicators. intuitionistic fuzzy analytic hierarchy process (ifahp) has also been widely used in the field of a comprehensive evaluation in china, and the effectiveness of this method has been proved [9]. for example, gu jing et al. (2015) [10] used the intuitionistic fuzzy analytic hierarchy process to evaluate the performance of venture capital guiding fund. compared with the traditional fuzzy analytic hierarchy process, it was found that this method has better performance in the evaluation accuracy. zhang shang-wai (2016) [11] put forward that the intuitionistic fuzzy analytic hierarchy process (ifahp) should be used to evaluate the operation status of power grid scientifically and reasonably and to rank the comprehensive evaluation journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 94 95 alleviation, big data and big ecology, guizhou will focus on developing the "four-type" industries, which will bring huge investment opportunities and financial needs, provide support for guizhou's ecological construction, and promote the realization of the "three orientations" in the new area. finally, it will be conducive to fostering new momentum of economic development and providing loans for regional economic development. this paper uses intuitionistic fuzzy analytic hierarchy process (ifahp) to comprehensively evaluate the level of ecological civilization construction in nine cities and municipalities of guizhou and to provide development suggestions for improving the level of ecological civilization construction in guizhou province. 2. an overview of domestic and foreign research with the improvement of environmental awareness, great progress has been made in the evaluation of the level of ecological civilization construction. early scholar zhao fang (2010) [1] decomposed the evaluation index of ecological civilization construction into four core areas: production mode, lifestyle, ecological protection and construction, social development, and set up several specific indicators, and constructed a threelevel, 27 single evaluation index structure from "target level-criterion level-index level". the evaluation index system of ecological civilization construction was established, and the progress rate analysis method was used to evaluate the construction of ecological civilization. in recent years, zhang huan et al. (2015) [2] has established an evaluation index system of ecological civilization in mega-cities, which includes four aspects: the degree of ecological environment health, the intensity of resource and environment consumption, the efficiency of non-point source pollution control and the livability of residents, with a total of 20 indicators, and put forward measures to strengthen the construction of ecological civilization. zheng yanling et al. (2016) [3] taking hebei province as an example, this paper constructs a comprehensive evaluation index system of ecological civilization construction and regional economic development level, adopts the measurement model of ecological civilization construction and regional economic coordinated development, and puts forward that with the advancement of ecological civilization construction, the relationship between ecological civilization and economic development will enter into mutual complementarity. good circulation. cheng jinhua et al. (2018) [4] taking 35 provincial capitals and sub-provincial cities in china and 7 small and medium-sized cities in hubei province as objects, the index system is constructed from four dimensions: land and space optimization, resource and energy utilization, ecological environment protection and governance, and ecological system construction. the index weight is given by the method of entropy weight. empirical research finds that big cities are the main cities. cities with a high index of ecological civilization often have the characteristics of balanced development of the economy and environment. there are few studies on the evaluation of guizhou's ecological civilization construction, and more qualitative analysis is used to analyze the current situation of guizhou's ecological civilization construction (jiangchuan et al. [5], 2014; shipeixin [6], 2010, etc.). van laarhoven and edrycz (1983) [7] proposed for the first time to fuzzify the analytic hierarchy process (ahp). buckley (1985) [8] developed the fuzzy analytic hierarchy process (fahp) on this basis. it is more complete and accurate to evaluate the target level indicators, which can reflect the uncertainty of the decision-makers in the evaluation of objective things. the subjectivity of experts is fuzzified, which is more objective and accurate than the analytic hierarchy process, but it can not accurately express the situation of abstention or hesitation, and the quantitative and qualitative indicators need to be aggregated by different methods. in order to solve this problem, an intuitionistic fuzzy analytic hierarchy process (ifahp) based on the concept of fuzzy sets is proposed, which unifies the aggregation of qualitative and quantitative indicators. intuitionistic fuzzy analytic hierarchy process (ifahp) has also been widely used in the field of a comprehensive evaluation in china, and the effectiveness of this method has been proved [9]. for example, gu jing et al. (2015) [10] used the intuitionistic fuzzy analytic hierarchy process to evaluate the performance of venture capital guiding fund. compared with the traditional fuzzy analytic hierarchy process, it was found that this method has better performance in the evaluation accuracy. zhang shang-wai (2016) [11] put forward that the intuitionistic fuzzy analytic hierarchy process (ifahp) should be used to evaluate the operation status of power grid scientifically and reasonably and to rank the comprehensive evaluation values, which is of great significance for the prevention of major blackouts. in view of this, on the basis of previous studies, this paper establishes an index system including four primary indicators: ecological economy, ecological environment, ecological livability and ecological culture, and then evaluates the level of ecological civilization construction in nine cities and municipalities of guizhou province by using intuitionistic fuzzy analytic hierarchy process. 3. establishment of an index system of ecological civilization construction level the selection of ecological civilization indicators plays an important role in accurately evaluating the level of ecological civilization construction. this paper follows the principles of comprehensiveness and representativeness, pertinence and comparability, orientation and foresight when constructing the index system. it emphasizes the sustainable development of nature, environment, human and society, and takes into account the carrying capacity of ecological environment, to promote the harmonious development of human and nature on the basis of improving social-ecological civilization. reference [12], four first-level indicators and 23 second-level indicators were developed, such as table 1. t he d ata so urces o f p aper : china regio nal economic statistics yearbook, guizhou statistical yearbook, nine prefecture, and city statis tical yearbooks, prefecture and city water resources bulletin, environment bulletin and national economic development bulletin. among them, the per capita housing area of anshun, southeastern guizhou and southwestern guizhou is estimated by the data of past years; the proportion of added value of cultural industry in gdp in anshun, zunyi, and bijie, and the per capita green area of parks in southwestern guizhou are estimated by the government's press announcement, and the sulfur dioxide emissions and industrial solid waste synthesis in southern guizhou are estimated. utilization ratio, the proportion of added value of cultural industry to gdp, average temperature and annual precipitation all come from the official website of the state government; the comprehensive utilization ratio of industrial solid waste in tongren city, the per capita green space area of parks in southeastern guizhou and southern guizhou and the proportion of added value of table 1. evaluation index system of ecological civilization construction level in guizhou. first level index two level index eco economy a1 per capita gross domestic product(rmb) b1 the growth rate of the added value of the tertiary industry as a proportion of gdp (%) b2 per capita growth rate of general public budget income (%) b3 energy consumption per unit gdp(ton of standard coal/10000 yuan) b4 power consumption per unit gdp (kwh/10000 yuan) b5 general public budget expenditure energy conservation and environmental protection (rmb 10,000) b6 ecological environment a2 forest cover (%) b7 excellent rate of urban ambient air quality (%) b8 water quality of centralized drinking water sources above county level (%) b9 increase rate of fertilizer application for agriculture (%) b10 comprehensive utilization rate of industrial solid waste (%) b11 total sulphur dioxide emissions (10,000 tons) b12 ecological livability a3 per capita park green space area (square metre/person) b13 harmless disposal rate of municipal solid waste (%) b14 average temperature (。c) b15 annual precipitation (mm) b16 per capita highway mileage (km) b17 per capita housing area in urban areas (square meters) b18 ecological culture a4 general public budget expenditure education services (rmb 100 million) b19 general public budget expenditure culture, sports and media (rmb 100 million) b20 the proportion of the added value of the cultural industry to gdp (%) b21 students in higher education (persons) b22 students in ordinary secondary schools (persons) b23 cultural industry in gdp in southeastern guizhou and southwestern guizhou are estimated by the government's news bulletins. 4. models and methods 4.1 standardization of data according to the attributes of the indicators, the indicators in this paper can be divided into benefit type, journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 96 cost type, and medium type. benefit-oriented indicators refer to the indicators with bigger values, while costoriented indicators refer to the indicators with smaller values. for benefit, cost and medium-sized indicators, the standardization methods [13] are: among them, i = 1, 2,..., m, m is the number of evaluation objects; j= 1, 2,..., n, n is the number of attributes, *, ,max min-j ij j ij jx x x x x   is the best stable value of j index. after data standardization, the values measured by various attributes change from 0 to 1. 4.2 comprehensive assessment steps atanassov further extends the theory of fuzzy sets proposed by zadeh and defines intuitionistic fuzzy sets. recently, intuitionistic fuzzy set theory has been widely used in decision-making field. the comprehensive assessment steps of ifahp are as follows: the first step: determine the index system. this paper constructs the index system of ecological civilization construction from four perspectives of ecological economy, ecological environment, ecological livability, and ecological culture, as shown in table 1. the second step: establish an intuitionistic fuzzy judgment matrix. this paper compares the importance of each factor of the same level index to a certain index of the upper level in two ways. through expert scoring, an intuitionistic fuzzy judgment matrix  ij n nr r  , ji, is established to represent the rows and columns of the intuitionistic fuzzy judgment matrix, in which ),(ij ijij vur  and iju represent the membership degree, i.e., the degree of superiority of index i to index j , ijv denotes non-membership, i.e. the degree of inferiority of index i to index j , and ij denotes hesitation, ijijij vu 1 . the correspondence between qualitative evaluations and intuitionistic fuzzy numbers is shown in table 2. step 3: consistency test. the consistency test of the intuitive judgment matrix is carried out. if consistency is satisfied, jump to step 5; otherwise, move to step 4. using the test formula table 2. correspondence tables between qualitative evaluations and intuitionistic fuzzy numbers. qualitative evaluation language intuitionistic fuzzy number qualitative evaluation language intuitionistic fuzzy number extremely important (0.90,0.10,0.00) less important (0.4,0.45,0.15) very important (0.80,0.15,0.05) unimportant (0.30,0.60,0.10) important (0.70,0.20,0.10) not important (0.20,0.75,0.05) more important (0.60,0.25,0.15) extremely unimportant (0.10,0.90,0.00) equally important (0.50,0.30,0.20) _ _       n i n j ijijijijij vvuu nn rrd 1 1 ij )( )2)(1(2 1 ),( ππ r is an intuitionistic fuzzy judgment matrix and an intuitionistic fuzzy consistency judgment matrix,  ij n nr r  . (1)when j>i+1, ( ) ( , )ij ij ijr r u v  ; (2)when j=i + 1, order ij ijr r ; (3)when j<i + 1, order ( , )ij ij ijr v u . the intuitionistic fuzzy consistency judgment matrix ( )ij n nr r  is substituted into the test formula for consistency test. if 1.0, ）（ rrd it passes the consistency test, otherwise it fails. step 4: revise the intuitive judgment matrix which does not satisfy the consistency test. if it fails to pass the consistency test, it is necessary to set the parameter σ for iteration, where  1,0 , until it passes the consistency test, the following formula is used: 1 ~ 1 1 1~ 1 1 ; (1 ) (1 ) , , 1, 2,..., (1 ) (1 ) ij ij ij ij ij ij ij ij ij ij ij ij ij ij u u u u u u u v v v i j n v v v v                        （ ） （ ） （ ）（ ） （ ） （ ） （ ） （ ） （ ） （ ） （ ） （ ） step 5: calculate the weight. according to the 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 , 1 (1 )(1 ) , 1 (1 )(1 ) j-1 j-i-1 it tj t i ij j j j i j iit tj it tj t i t i j j i it tj t i ij j j j i j iit tj it tj t i t i u u u j i u u u u v v v j i v v v v                                              * * ; ; 1 , 1, 2,..., max j ij ij j ij ij j j j j ij j ij ij j x x x x z z x x x x x x z i n x x                 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 96 97 cost type, and medium type. benefit-oriented indicators refer to the indicators with bigger values, while costoriented indicators refer to the indicators with smaller values. for benefit, cost and medium-sized indicators, the standardization methods [13] are: among them, i = 1, 2,..., m, m is the number of evaluation objects; j= 1, 2,..., n, n is the number of attributes, *, ,max min-j ij j ij jx x x x x   is the best stable value of j index. after data standardization, the values measured by various attributes change from 0 to 1. 4.2 comprehensive assessment steps atanassov further extends the theory of fuzzy sets proposed by zadeh and defines intuitionistic fuzzy sets. recently, intuitionistic fuzzy set theory has been widely used in decision-making field. the comprehensive assessment steps of ifahp are as follows: the first step: determine the index system. this paper constructs the index system of ecological civilization construction from four perspectives of ecological economy, ecological environment, ecological livability, and ecological culture, as shown in table 1. the second step: establish an intuitionistic fuzzy judgment matrix. this paper compares the importance of each factor of the same level index to a certain index of the upper level in two ways. through expert scoring, an intuitionistic fuzzy judgment matrix  ij n nr r  , ji, is established to represent the rows and columns of the intuitionistic fuzzy judgment matrix, in which ),(ij ijij vur  and iju represent the membership degree, i.e., the degree of superiority of index i to index j , ijv denotes non-membership, i.e. the degree of inferiority of index i to index j , and ij denotes hesitation, ijijij vu 1 . the correspondence between qualitative evaluations and intuitionistic fuzzy numbers is shown in table 2. step 3: consistency test. the consistency test of the intuitive judgment matrix is carried out. if consistency is satisfied, jump to step 5; otherwise, move to step 4. using the test formula table 2. correspondence tables between qualitative evaluations and intuitionistic fuzzy numbers. qualitative evaluation language intuitionistic fuzzy number qualitative evaluation language intuitionistic fuzzy number extremely important (0.90,0.10,0.00) less important (0.4,0.45,0.15) very important (0.80,0.15,0.05) unimportant (0.30,0.60,0.10) important (0.70,0.20,0.10) not important (0.20,0.75,0.05) more important (0.60,0.25,0.15) extremely unimportant (0.10,0.90,0.00) equally important (0.50,0.30,0.20) _ _       n i n j ijijijijij vvuu nn rrd 1 1 ij )( )2)(1(2 1 ),( ππ r is an intuitionistic fuzzy judgment matrix and an intuitionistic fuzzy consistency judgment matrix,  ij n nr r  . (1)when j>i+1, ( ) ( , )ij ij ijr r u v  ; (2)when j=i + 1, order ij ijr r ; (3)when j<i + 1, order ( , )ij ij ijr v u . the intuitionistic fuzzy consistency judgment matrix ( )ij n nr r  is substituted into the test formula for consistency test. if 1.0, ）（ rrd it passes the consistency test, otherwise it fails. step 4: revise the intuitive judgment matrix which does not satisfy the consistency test. if it fails to pass the consistency test, it is necessary to set the parameter σ for iteration, where  1,0 , until it passes the consistency test, the following formula is used: 1 ~ 1 1 1~ 1 1 ; (1 ) (1 ) , , 1, 2,..., (1 ) (1 ) ij ij ij ij ij ij ij ij ij ij ij ij ij ij u u u u u u u v v v i j n v v v v                        （ ） （ ） （ ）（ ） （ ） （ ） （ ） （ ） （ ） （ ） （ ） （ ） step 5: calculate the weight. according to the 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 , 1 (1 )(1 ) , 1 (1 )(1 ) j-1 j-i-1 it tj t i ij j j j i j iit tj it tj t i t i j j i it tj t i ij j j j i j iit tj it tj t i t i u u u j i u u u u v v v j i v v v v                                              * * ; ; 1 , 1, 2,..., max j ij ij j ij ij j j j j ij j ij ij j x x x x z z x x x x x x z i n x x                 intuitionistic fuzzy consistency judgment matrix, the weight is calculated. n,...,,i), n i n j vij n j )uij( , n i n j )vij( n j uij (i 21 1 1 1 1 1 1 1 1 1                 step 6: using intuitionistic fuzzy number operator to calculate the combined weights of each level of indicators on ecological civilization. wa is the weight of the first level index; wb is the weight of the second level index. )wajwbij( n j wj )vv,uuuu( )vvvv,uu(      1 1 21212121 21212121   step 7: substitute standardized index values into ) 1 i ( jij n j wzf   . the comprehensive evaluation value is obtained, and then the ranking of the construction level of ecological civilization in nine cities and municipalities is obtained. the bigger the ranking, the higher the construction level of ecological civilization. 5. empirical analysis according to step 2, the intuitionistic fuzzy judgment matrix is established. according to the experts' opinions, in order to evaluate the level of ecological civilization construction, the importance of the first-level indicators from big to small is the ecological environment, ecological livability, ecological culture, and ecological economy. taking the ecological environment as an example, this paper introduces the process of empirical analysis in detail. under this first-level index, according to the importance degree of the second-level index, the good rate of urban environmental air quality, forest coverage, total so2 emission, comprehensive utilization rate of industrial solid waste, water quality standard rate of centralized drinking water sources above county level and the growth rate of agricultural chemical fertilizer application rate are respectively. according to the expert opinions, the former is more important than the forest coverage, and the corresponding intuitionistic fuzzy number is (0.6, 0.25); the former is very important, and the intuitionistic fuzzy number is (0.8, 0.15); the former is important, and the intuitionistic fuzzy number is (0.8, 0.15); the comprehensive utilization ratio of forest coverage and industrial solid waste is important, the former is important, and the intuitionistic fuzzy number is (0.7, 0.2). by analogy, experts' opinions corresponded to intuitionistic fuzzy numbers, and the intuitionistic fuzzy judgment matrix is established as shown in table 3. the calculation method and process of the other matrices are the same. set up the iteration parameters, make σ = 1/4, and use step 4 to calculate ), ~~ vu（ , as shown in table 4. the consistency test was carried out again, d = 0.048 < 0.1. through the consistency test, the intuitionistic fuzzy consistency judgment matrix was established as shown in the table. the calculation method and process of the other matrices are the same. according to step 6, the combination weight is calculated. the combined weights of six secondary indicators under ecological environment are (0.058, 0.895), (0.070, 0.883), (0.025, 0.946), (0.038, 0.930), (0.032, 0.935) and (0.046, 0.913). it is extended to the other four matrices. according to the first step to the fourth step of calculation, after establishing the intuitionistic fuzzy judgment matrix in the form of expert scoring, the consistency test is carried out. it is found that the four intuitionistic fuzzy judgment matrices do not satisfy the consistency test. the four matrices ),( rrd are 0.1333, 0.186, 0.153 and 0.11 respectively, which are greater than 0.1 and do not satisfy the consistency test. therefore, setting parameters to σ=1/4, adjusting intuitionistic fuzzy judgment matrix, and then conducting consistency test, ),( rrd are 0.035, 0.049, 0.038 and 0.046 respectively, which are less than 0.1, satisfying the consistency test. the weight of the index is calculated from the fifth and sixth steps as shown in table 5. according to the three cases of i and j in step 3, and are calculated respectively, and the intuitionistic fuzzy consistency judgment matrix is established, as shown in table 6. the calculation method and process of the other matrices are the same. the consistency test was carried out by using the test formula, d = 0.0208 > 00.1, which did not pass the consistency test. therefore, it can be said that the above table is not an intuitionistic fuzzy consistency judgment matrix. the calculation method and process of the other matrices are the same. the step 7 is used to calculate the comprehensive evaluation values of nine prefectures and cities, and the journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 98 table 3. intuitionistic fuzzy judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.8,0.15) (0.8,0.15) (0.7,0.2) (0.6,0.25) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.9,0.1) (0.8,0.15) (0.7,0.2) b9 (0.15,0.8) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.25,0.6) (0.2,0.7) table 4. intuitionistic fuzzy consistency judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.57,0.2) (0.67,0.17) (0.72,0.13) (0.49,0.28) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.57,0.2) (0.8,0.08) (0.62,0.23) b9 (0.2,0.57) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.44,0.27) (0.1,0.73) b10 (0.17,0.67) (0.2,0.57) (0.6,0.25) (0.5,0.5) (0.7,0.2) (0.61,0.17) b11 (0.13,0.72) (0.08,0.8) (0.27,0.44) (0.2,0.7) (0.5,0.5) (0.4,0.45) b12 (0.28,0.49) (0.23,0.62) (0.73,0.1) (0.17,0.61) (0.45,0.4) (0.5,0.5) table 5. intuitionistic fuzzy consistency judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.75,0.16) (0.77,0.15) (0.71,0.,18) (0.57,0.26) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.85,0.12) (0.80,0.13) (0.68,0.21) b9 (0.16,0.75) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.29,0.51) (0.17,0.71) b10 (0.15,0.77) (0.12,0.85) (0.6,0.25) (0.5,0.5) (0.7,0.2) (0.23,0.66) b11 (0.18,0.71) (0.13,0.80) (0.51,0.29) (0.2,0.7) (0.5,0.5) (0.4,0.45) b12 (0.26,0.57) (0.21,0.68) (0.71,0.17) (0.66,0.23) (0.45,0.4) (0.5,0.5) table 6. total weight tables of indicators. first level index weight two level index weight total weight first level index weight two level index weight total weight (0.137,0.785) (0.092,0.832) (0.013,0.964) (0.237,0.676) (0.137,0.795) (0.033,0.934) (0.144,0.772) (0.020,0.951) (0.137,0.805) (0.032,0.937) (0.081,0.868) (0.011,0.972) (0.229,0.712) (0.054,0.907) (0.187,0.729) (0.026,0.942) (0.186,0.754) (0.044,0.920) (0.105,0.818) (0.014,0.961) (0.040,0.891) (0.010,0.965) (0.185,0.722) (0.025,0.940) (0.088,0.867) (0.021,0.957) (0.324,0.587) (0.180,0.746) (0.058,0.895) (0.149,0.772) (0.142,0.754) (0.021,0.944) (0.215,0.716) (0.070,0.883) (0.159,0.740) (0.024,0.941) (0.077,0.869) (0.025,0.946) (0.222,0.662) (0.033,0.923) (0.117,0.830) (0.038,0.930) (0.141,0.747) (0.021,0.942) (0.098,0.844) (0.032,0.935) (0.105,0.797) (0.016,0.954) (0.142,0.788) (0.046,0.913) journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 98 99 table 3. intuitionistic fuzzy judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.8,0.15) (0.8,0.15) (0.7,0.2) (0.6,0.25) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.9,0.1) (0.8,0.15) (0.7,0.2) b9 (0.15,0.8) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.25,0.6) (0.2,0.7) table 4. intuitionistic fuzzy consistency judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.57,0.2) (0.67,0.17) (0.72,0.13) (0.49,0.28) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.57,0.2) (0.8,0.08) (0.62,0.23) b9 (0.2,0.57) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.44,0.27) (0.1,0.73) b10 (0.17,0.67) (0.2,0.57) (0.6,0.25) (0.5,0.5) (0.7,0.2) (0.61,0.17) b11 (0.13,0.72) (0.08,0.8) (0.27,0.44) (0.2,0.7) (0.5,0.5) (0.4,0.45) b12 (0.28,0.49) (0.23,0.62) (0.73,0.1) (0.17,0.61) (0.45,0.4) (0.5,0.5) table 5. intuitionistic fuzzy consistency judgment matrix. b7 b8 b9 b10 b11 b12 b7 (0.5,0.5) (0.25,0.6) (0.75,0.16) (0.77,0.15) (0.71,0.,18) (0.57,0.26) b8 (0.6,0.25) (0.5,0.5) (0.8,0.15) (0.85,0.12) (0.80,0.13) (0.68,0.21) b9 (0.16,0.75) (0.15,0.8) (0.5,0.5) (0.25,0.6) (0.29,0.51) (0.17,0.71) b10 (0.15,0.77) (0.12,0.85) (0.6,0.25) (0.5,0.5) (0.7,0.2) (0.23,0.66) b11 (0.18,0.71) (0.13,0.80) (0.51,0.29) (0.2,0.7) (0.5,0.5) (0.4,0.45) b12 (0.26,0.57) (0.21,0.68) (0.71,0.17) (0.66,0.23) (0.45,0.4) (0.5,0.5) table 6. total weight tables of indicators. first level index weight two level index weight total weight first level index weight two level index weight total weight (0.137,0.785) (0.092,0.832) (0.013,0.964) (0.237,0.676) (0.137,0.795) (0.033,0.934) (0.144,0.772) (0.020,0.951) (0.137,0.805) (0.032,0.937) (0.081,0.868) (0.011,0.972) (0.229,0.712) (0.054,0.907) (0.187,0.729) (0.026,0.942) (0.186,0.754) (0.044,0.920) (0.105,0.818) (0.014,0.961) (0.040,0.891) (0.010,0.965) (0.185,0.722) (0.025,0.940) (0.088,0.867) (0.021,0.957) (0.324,0.587) (0.180,0.746) (0.058,0.895) (0.149,0.772) (0.142,0.754) (0.021,0.944) (0.215,0.716) (0.070,0.883) (0.159,0.740) (0.024,0.941) (0.077,0.869) (0.025,0.946) (0.222,0.662) (0.033,0.923) (0.117,0.830) (0.038,0.930) (0.141,0.747) (0.021,0.942) (0.098,0.844) (0.032,0.935) (0.105,0.797) (0.016,0.954) (0.142,0.788) (0.046,0.913) ranking results are shown in table 7, where the comprehensive evaluation value of southeastern guizhou is 12.72, with the highest score, followed by guiyang, southern guizhou, southwestern guizhou, tongren, and zunyi, with 11.94, 11.88, 11.42, 10.97 and 10.72 respectively. bijie, anshun and liupanshui scored lower, 9.72, 9.55 and 7.81 respectively. the construction level of ecological civilization in nine cities and municipalities is relatively high, but the development level is unbalanced. table 7. ranking table of comprehensive evaluation results. region comprehensive evaluation value sort qiandongnan 12.72 1 guiyang 11.94 2 qiannan 11.88 3 qianxinan 11.42 4 tongren 10.97 5 zunyi 10.72 6 bijie 9.72 7 anshun 9.55 8 liupanshui 7.81 9 6. conclusion this paper applies the intuitionistic fuzzy analytic hierarchy process (ifahp) to evaluate the level of ecological civilization construction in guizhou province and establishes an index system including four firstlevel indicators of the ecological economy, ecological environment, ecological livability and ecological culture, and 23 second-level indicators. the index weights are expressed by the intuitionistic fuzzy numbers, and then the scores are calculated and compared. however, limited by the statistical data of each prefecture and city, the representative of individual indicators is not strong. the empirical study of this paper shows that the level of ecological civilization construction in guizhou province is relatively high, but the level of each prefecture is quite different. therefore, it is necessary for the governments, environmental protection bureaus and water conservancy bureaus of various prefectures and municipalities in guizhou province to take measures to further improve the ecological quality. for example, guizhou province should make full use of ecological resources, promote the marketization and capitalization of ecological resources in guizhou province, and respond to the national strategy to achieve green development, improve the efficiency of clean energy use, improve citizens'awareness of environmental protection, and increase the penalties for excessive pollutants discharged by industrial enterprises. acknowledgments fund projects: regional project of national natural science foundation of china (71861003); the second batch projects of basic research program (soft science category) in guizhou province in 2017 (foundation of guizhou-science cooperation [2017] 1516-1). references [1] zhao fang. establishment and empirical study of evaluation index system for ecological civilization construction [d]. chinese academy of forestry sciences, 2010. [2] zhang huan, cheng jinhua, feng yin, chen dan, ni lin, sun han. evaluation index system and its application for the construction of ecological civilization in mega-cities: a case study of wuhan [j]. journal of ecology, 2015, 35 (02): 547-556. [3] zheng yanling, gao jianshan, han fubin. study on the performance evaluation of the harmonious development of ecological civilization and regional economy: a case study of hebei province [j]. eco-economy, 2016, 32(12): 198-203. [4] cheng jinhua, peng xinjie, feng yin. evaluation of chinese urban ecological civilization [j]. journal of china university of geosciences (social science edition), 2018, 18 (02): 102-113. [5] jiangchuan, wang zhiming, huang wen hu. the present situation and countermeasures of ecological civilization construction in guizhou province [j]. china environmental monitoring, 2014, 30 (03): 13-17. [6] shi ba xin. some thoughts on the construction of ecological civilization in guizhou [j]. journal of guizhou teachers college, 2010, 26 (02): 8-11. [7] van laarhoven plm pedrycz w.a fuzzy extension of saaty’s priority theory [j].fuzzy sets and systems, 1983, 11(1): 229-241. [8] buckley j j. fuzzy hierarchical analysis [j].fuzzy sets and systems, 1985, 3(17): 233－247. [9] gao hongyun, wang chao, haminghu. intuitionistic fuzzy analytic hierarchy process [j]. journal of hebei university of engineering (natural science edition), 2011, 28 (04): 101-105. [10] zhang shang, wang tao, gu xueping. a comprehensive evaluation of power network operation status based on intuitionistic fuzzy analytic hierarchy process [j]. power system automation, 2016, 40 (04): 4149. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 93-100 100 [11] gu jing, ren bijia and xu zeshui. research on the performance evaluation method of venture capital guidance fund based on intuitionistic fuzzy analytic hierarchy process [j]. china management science, 2015, 23 (09): 124-131. [12] lin zhen, shuangzhimin. a comparative study on the evaluation index system of eco-civilization construction in provincial capitals: taking guiyang, hangzhou, and nanjing as examples [j]. journal of beijing university of aeronautics and astronautics (social sciences edition), 2014, 27 (05): 22-28. [13] mu zhang, zongfang zhou.a credit rating model for enterprises based on projection pursuit and k-means clustering algorithm[j].jracr, 2012(2): 131-138. 62 journal of risk analysis and crisis response vol. 9(2), june (2019), pp. 62-73 doi: https://doi.org/10.2991/jracr.k.190703.002; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). competency of the infantry troops of the nepalese army in disaster response shrijan bahadur malla1, shuichi hasegawa2, ranjan kumar dahal3 1. faculty of engineering, graduate school of engineering, kagawa university, 2217-20 hayashi-cho, takamatsu, kagawa 761-0396, japan. 2. department of safety systems construction engineering, kagawa university, takamatsu city, japan. 3. central department of geology, tribhuvan university, kritipur, nepal. received april 7, 2019 accepted april 27, 2019 abstract although the nepalese army’s regular infantry troops are the tools of disaster response their competency at times becomes questionable. their disaster response’s soft skill knowledge, technical skill knowledge and preparedness activities along with the perception of senior army officers have been analyzed. as per the result the disaster response competency of such troops needs improvements. the organization should focus on competency building activities. besides, the perception of senior army officers is positive and convincing towards disaster response engagement. keywords: nepalese army, infantry troops, disaster response, competency, knowledge, preparedness 1. introduction the risk of natural disaster arises when hazards interact with the physical, social, economic and environmental vulnerabilities and exposure to population (selmi et al., 2015). today the increased frequency of natural disaster has accumulated immense challenge to economic, social, health, cultural and environmental issues mostly in developing countries2. data show that in the world every year around 70,000 people loses their lives due to various natural disaster (unisdr, 2015: dahal et al., 2013). and most of them are from low and middle-income countries like nepal. undp (2004) once reported that 11% of the people exposed to natural disaster hazards live in poor countries and those countries account more than 53% of the total disaster-related deaths. in nepal alone every year more than 500 people die due to various natural disasters since it is the 20th most disaster prone country in the world (moha, 2018). it ranks 4th vulnerable in terms of climate change and 11th vulnerable in terms of earthquake (nepal disaster report, 2017). earthquake, flood, landslide, debris flow, avalanche, fire, and draught are amongst the disasters nepal is always prone to (moha, 2015). disaster management is a complex cyclic process that has four phases i.e. mitigation, preparedness, response and recovery (smet et al, 2015). coming to this age although the disaster scientists have chiefly emphasized on mitigation process to reduce societal vulnerabilities to prevent disasters from occurrence the mitigation efforts only cannot prevent all disasters from occurrence hence requires preparedness activities to be ready for effective response(lindell & perry, 1992). indeed, despite good mitigation and preparedness effort disaster always occurs that renders a response to be critical to reduce casualty and economic losses and to prevent the situation from further escalation (wildavsky, 1988; coppola, 2011; wex et al, 2014). for poor countries like nepal where investment in disaster mitigation is inadequate the need for an effective response mechanism is vital. today, in the worldly context, the militaries are somehow effective tools of disaster response mechanism. higher magnitude and frequency of disasters overwhelm the capability of civilian government thus gives an opportunity to militaries to join hand in relief and rescue operations (sahashi, 2015). their specific assets (human and material), capacity to respond quickly on a larger scale and national political imperative makes them reliable tools of the governments to engage in the venture (elizabeth, 2012; hoff, 1999). especially in developing countries, the reliance on the military for disaster response is high since these countries lack effective and comprehensive civilian structures for disaster response (malesic, 2015). in such situation, their regular infantry troops, along with specialized military units, become the part of their response mechanism (raj, 2008; hall & anita, 2010; keen et al., 2011; frank, 2005). but the effectiveness of such regular infantry’s engagement at times becomes questionable. frederick c. cunny (1991) argues that regular infantries are not trained for humanitarian roles. their directives and training are inappropriate. many scholars argue that military support is sought during disaster response because only of their logistic capability and human resource. hofmann and hudson (2009) write that the regular infantries’ involvement in disaster response is inefficient, inappropriate, inadequate and expensive. similarly, weeks (2007) also suggests that in many occasions the military troops (regular infantry) were found ill-prepared to deal with the natural disaster since they hardly have any experience in the past and much military personnel find themselves experiencing myriad problems associated with a disaster situation. philippine mudslide of 2006, china sichuan earthquake of 2008, pakistan flood 2010, and japan nuclear disaster of 2011are some examples where the engagement of such regular infantry troops bore criticism (hall & anita, 2010; manish, 2018; ajay & kundrat, 2011; banyan, 2014; martinez et al., 2015; raja & arshad, 2015). jake hooker (2008) writes that the ‘during the sichuan earthquake the pla soldiers (regular infantries) were mobilized quickly but couldn’t do anything for the first 72 hours’. later on, it was investigated that the disaster specialized pla soldier was 5.6 times more effective than a regular infantry soldier (li et al., 2009). in nepal as well the regular infantry troops of the nepalese army are the tools of natural disaster response (manandhar et al., 2017; thapa, 2016; marshall & adkin, 2016; the constitution of nepal, 2015; disaster risk reduction and management act, 2017). for security and disaster management their presence has been ensured in every district of the country. today an infantry battalion has been deployed in every district of the country. 2017 terai flood, 2015 gorkha earthquake, 2014 jure landslide, and 2008 koshi flood are some examples where these troops have played significant role. but in contrary to this urgency, disaster response preparedness of those regular infantry troops has remained in shadow. it is because the organization is moreover focused on strengthening its war-waging capability. the secondary mandate like disaster management has been facing chronic resource shortage (poudel, 2016). in such a scenario full-fledged preparedness of such troops in disaster response is hardly attainable. and this reality was clearly evident during the response operations of gorkha earthquake 2015. various national and international after-action reports of gorkha earthquake have emphasized that there is a need to ‘enhance search and rescue capability of nepali army’(moha, 2015; moha, 2016). they have also stated that the troops were less trained and less equipped for disaster response (manandhar et al., 2017; grunewald & burlet, 2016). indeed, the effectiveness of response operation chiefly relies on the competency of the responders. various disaster scientists have indicated that ‘knowledge’ and ‘preparedness activities’ are the fundamental pillars of disaster response competency (barsky et al., 2007; gallardo et al., 2015). harris et al. (2018) defines that the knowledge gives awareness of the given role(s) and task(s), that helps responder to work within the outlined framework, whereas preparedness renders him/her a platform to carry out effective, realistic and coordinated planning, reduces duplication of effort and increases overall effectiveness. in the void of these pillars, challenges persist in coordination, integration, and logistics, along with health and safety problems of the responder that may jeopardize the entire holistic effort of disaster response (twigg & mosel, 2017). furthermore, disaster response knowledge consists of two categories i.e. ‘soft skill knowledge (ssk)’ and ‘technical skill knowledge (tsk)’ (barelli et al., 2014). the skk is non-technical, interpersonal and cognitive like principles, leadership, teamwork, and communication42 whereas the tsk is procedural and systematic to function technical work like search and rescue, debris management, dead body management, casualty evacuation etc (poul, 2018). and both the knowledge is prerequisite for the normal infantry troops since they have to carry out multiple functions during a disaster scenario. the gorkha earthquake is an example how these same troops carried out series of response operations like search and rescue, casualty evacuation, debris management, dead body management simultaneously (barsky et al., 2007). similarly, training, exercise, and rehearsal are the cornerstone activities for building disaster response journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 62 63 reliable tools of the governments to engage in the venture (elizabeth, 2012; hoff, 1999). especially in developing countries, the reliance on the military for disaster response is high since these countries lack effective and comprehensive civilian structures for disaster response (malesic, 2015). in such situation, their regular infantry troops, along with specialized military units, become the part of their response mechanism (raj, 2008; hall & anita, 2010; keen et al., 2011; frank, 2005). but the effectiveness of such regular infantry’s engagement at times becomes questionable. frederick c. cunny (1991) argues that regular infantries are not trained for humanitarian roles. their directives and training are inappropriate. many scholars argue that military support is sought during disaster response because only of their logistic capability and human resource. hofmann and hudson (2009) write that the regular infantries’ involvement in disaster response is inefficient, inappropriate, inadequate and expensive. similarly, weeks (2007) also suggests that in many occasions the military troops (regular infantry) were found ill-prepared to deal with the natural disaster since they hardly have any experience in the past and much military personnel find themselves experiencing myriad problems associated with a disaster situation. philippine mudslide of 2006, china sichuan earthquake of 2008, pakistan flood 2010, and japan nuclear disaster of 2011are some examples where the engagement of such regular infantry troops bore criticism (hall & anita, 2010; manish, 2018; ajay & kundrat, 2011; banyan, 2014; martinez et al., 2015; raja & arshad, 2015). jake hooker (2008) writes that the ‘during the sichuan earthquake the pla soldiers (regular infantries) were mobilized quickly but couldn’t do anything for the first 72 hours’. later on, it was investigated that the disaster specialized pla soldier was 5.6 times more effective than a regular infantry soldier (li et al., 2009). in nepal as well the regular infantry troops of the nepalese army are the tools of natural disaster response (manandhar et al., 2017; thapa, 2016; marshall & adkin, 2016; the constitution of nepal, 2015; disaster risk reduction and management act, 2017). for security and disaster management their presence has been ensured in every district of the country. today an infantry battalion has been deployed in every district of the country. 2017 terai flood, 2015 gorkha earthquake, 2014 jure landslide, and 2008 koshi flood are some examples where these troops have played significant role. but in contrary to this urgency, disaster response preparedness of those regular infantry troops has remained in shadow. it is because the organization is moreover focused on strengthening its war-waging capability. the secondary mandate like disaster management has been facing chronic resource shortage (poudel, 2016). in such a scenario full-fledged preparedness of such troops in disaster response is hardly attainable. and this reality was clearly evident during the response operations of gorkha earthquake 2015. various national and international after-action reports of gorkha earthquake have emphasized that there is a need to ‘enhance search and rescue capability of nepali army’(moha, 2015; moha, 2016). they have also stated that the troops were less trained and less equipped for disaster response (manandhar et al., 2017; grunewald & burlet, 2016). indeed, the effectiveness of response operation chiefly relies on the competency of the responders. various disaster scientists have indicated that ‘knowledge’ and ‘preparedness activities’ are the fundamental pillars of disaster response competency (barsky et al., 2007; gallardo et al., 2015). harris et al. (2018) defines that the knowledge gives awareness of the given role(s) and task(s), that helps responder to work within the outlined framework, whereas preparedness renders him/her a platform to carry out effective, realistic and coordinated planning, reduces duplication of effort and increases overall effectiveness. in the void of these pillars, challenges persist in coordination, integration, and logistics, along with health and safety problems of the responder that may jeopardize the entire holistic effort of disaster response (twigg & mosel, 2017). furthermore, disaster response knowledge consists of two categories i.e. ‘soft skill knowledge (ssk)’ and ‘technical skill knowledge (tsk)’ (barelli et al., 2014). the skk is non-technical, interpersonal and cognitive like principles, leadership, teamwork, and communication42 whereas the tsk is procedural and systematic to function technical work like search and rescue, debris management, dead body management, casualty evacuation etc (poul, 2018). and both the knowledge is prerequisite for the normal infantry troops since they have to carry out multiple functions during a disaster scenario. the gorkha earthquake is an example how these same troops carried out series of response operations like search and rescue, casualty evacuation, debris management, dead body management simultaneously (barsky et al., 2007). similarly, training, exercise, and rehearsal are the cornerstone activities for building disaster response journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 64 competency. descy and tessaring (2002) define that vocational training provides people with competence that is necessary and sufficient to perform a job. ingrassia pl et.al (2014) explains that all the professionals involved in disaster response should receive specific disaster-related training, exercise, and rehearsal regardless of their profession. as per hsu et al. (2013) ‘the quality, consistency, and frequency of disaster response training and exercise are acknowledged to notably impact selfperceived disaster preparedness of potential responders’. jay levinson (2008) writes in his article military involvement in disaster response that ‘providing soldiers with disaster awareness training can only give better understanding when catastrophe arises’. he also adds that only technical units in military have skills with disaster response application, to field other units require training, drills and periodic refresher courses. when it comes to competency building, the national strategy for disaster risk management 2009, the national framework for disaster management 2013 and the disaster risk reduction and mitigation act 2017 have also guided the nepalese army to carry out preparedness activities of its troops. similarly, the nepali army disaster management doctrine has also defined that training, exercise, and rehearsal are the regular tasks for the competency building. in this backdrop, it is important that the knowledge and preparedness of normal infantry troops should be studied. but hardly this issue has garnered interest in the academic research field. julia hornyacsek (2018) admits that the ‘disaster response competency of such regular infantry forces has hardly been studied’. in this regard, this research aims to study the disaster response competency of regular infantry troops of the nepalese army setting following research questions (1) whether the skk and tsk of the regular infantry troops of the nepalese army adequate enough to carry out effective response operations? (2) did the preparedness activities i.e. training, exercise and rehearsal adequate of the general infantry troops of the nepalese army adequate enough to carry out effective disaster response operations? (3) is there any gap in the skk and tsk, and preparedness of the regular infantry troops of the nepalese army? (4) what is the perception of the senior army officers working at policy level about the nepalese army’s engagement in disaster management? this last question is equally important because the perception of headquarters’ officers ultimately helps to drive the army’s disaster response engagement since they play a crucial role to design plan, policies, and directives including the operational activities of the army. ultimately this research paper will also be an aid to scholars interested to learn the competency of regular infantry troops of the nepalese army in disaster response. 2. methodology 2.1. overview the research emphasized both qualitative and quantitative approach in the data collection and analysis. during survey qualitative interviews with various government officials, subject matter experts, retired and active the nepalese army officers have been carried out. analysis of their experience, attitude, belief and thoughts, the study of national and international legal documents, binding and non-binding guidelines, plans and policies of the nepalese army have also been carried out. the quantitative analysis was conducted through two sets of questionnaire surveys. one was for the infantry troops and other was for the senior army officers working at army headquarters level. the first survey was primarily focused on the study of the ssk, tsk and disaster response preparedness of the troops. whereas, the second survey was focused studying the perception of high ranking officers about the army’s engagement in disaster management. the questions were designed relatively cognitive to answer presuming that the nepalese army personnel are somehow acquainted with the disaster response venture through various pieces of training, educational programs, and operational engagements. 2.2. sample population and its analysis for quantitative survey altogether 270 active nepalese army personnel were selected as study participants. amongst them, for first survey 245 personnel were randomly selected from 35 infantry battalions deployed in 35 districts. mid-level officers below the rank of major, junior commissioned officers (jcos) (also known as warrant officers) and other ranks (from the rank of soldier till the rank of staff sergeant) were the participants for the first survey. their percentage was 19.7%, 20.6%, and 59.7% respectively. similarly, 25 officers above the rank of lieutenant colonel, working at various branches of army headquarters, were randomly selected for the second survey. the service year of all the study participants spanned from 2 to 36 years, with a mean value of 14.52 years and the standard deviation (sd) of 5.97. during the first survey, the junior officers working in the respective units, who have basic knowledge in disaster management, were selected as enumerators. they were asked to select the study participants who could understand and answer the questions. the survey was conducted in the presence of those enumerators themselves in an interview style for the clarity of the question. in average, it took 20-30 minutes for the participants to answer all the questions of the surveys. 2.3. research design in both surveys, closed-ended self administered questions (saq) was asked to the responders as per the established methodology well described by brancato et al. (2004), ronan et al. (2010), kuroiwa (1993), and arya (1993). during the first survey total of 22 questions configured into three categories were asked to the responders. first eight questions were asked about the ssk of disaster management. fundamental issues like command control, civil-military relations, response priority, lead agency etc were the part of the questions. these questions were asked to the officers and jcos since they are the field level commanders to lead disaster response operations. subsequently, in the second category, ten questions about disaster response’s tsk were asked. study of their perception of disaster response operations like search and rescue operations, medical assistance, debris management, dead body management etc was conducted. study of the underlying technical issues was not part of the survey because it requires a separate scientific technical survey. similarly, in the end, four questions related to disaster response preparedness activities i.e. training, exercise, rehearsal, and operational engagement were asked to the participants. likewise, the second survey was conducted amongst policy level officers. altogether 7 questions were asked to them. how do they perceive about the engagement of the nepal army in disaster response operations was the scope of this survey. fundamental issues of present engagement and its effect on civil-military relations was the range of the study. ‘how the policymakers want to drive this venture in future’ was also the part survey. three types of closed-ended saqs were designed for the surveys. first was of likert scale of five optional indicators of intensity rating scale i.e. (1) strongly disagree (2) disagree (3) uncertain (4) agree (5) strongly agree. the second was of ‘check all that apply’ type questions and third was dichotomous questions (callegaro, 2014). during analysis, all the replies were re-coded into intensity rating scale. 2.4. method of analysis fundamentally, the descriptive analysis method and inferential analysis method were adopted during research that helped to examine the relationship between the nepal army’s disaster response competency building initiative and its effect on personnel competency. for the purpose, three key independent variables i.e. ssk, tsk, and disaster preparedness adaptation were considered as study factors. these variables were again analyzed on the basis of two dependent variables i.e. ‘rank’ (officer, jco and other ranks) and ‘training’ (trained and untrained troops). whether these variables make any significant difference in the study participants’ knowledge and preparedness or not was the aim of the analysis process. similarly during the second survey prevailing perception of senior army officers on the engagement of the nepalese army in disaster response was analyzed. analytical tools such as bar graphs and comparison table, and tools of linear regression such as pearson correlation matrix, model summary matrix, anova, coefficients matrix were used during the analysis process. through cross tabulation frequency distribution of the responses was studied. and the results were illustrated through bar graphs. for analysis the results of indicators of intensity rating scale were rephrased. for example, in an issue, if the respondent participant replied ‘strongly agree’ then it was considered that he/she was well informed in that issue. again through linear regression, the relationship between independent and dependent variables were examined. for example, the ‘ssk’ was examined amongst the officers and jcos through cross tabulation and whether that relation is statistically significant or not was verified through linear regression. results of bar graphs, model summary, anova, and coefficients were illustrated in figures. detail analysis was conducted using available commercial software and tools. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 64 65 various branches of army headquarters, were randomly selected for the second survey. the service year of all the study participants spanned from 2 to 36 years, with a mean value of 14.52 years and the standard deviation (sd) of 5.97. during the first survey, the junior officers working in the respective units, who have basic knowledge in disaster management, were selected as enumerators. they were asked to select the study participants who could understand and answer the questions. the survey was conducted in the presence of those enumerators themselves in an interview style for the clarity of the question. in average, it took 20-30 minutes for the participants to answer all the questions of the surveys. 2.3. research design in both surveys, closed-ended self administered questions (saq) was asked to the responders as per the established methodology well described by brancato et al. (2004), ronan et al. (2010), kuroiwa (1993), and arya (1993). during the first survey total of 22 questions configured into three categories were asked to the responders. first eight questions were asked about the ssk of disaster management. fundamental issues like command control, civil-military relations, response priority, lead agency etc were the part of the questions. these questions were asked to the officers and jcos since they are the field level commanders to lead disaster response operations. subsequently, in the second category, ten questions about disaster response’s tsk were asked. study of their perception of disaster response operations like search and rescue operations, medical assistance, debris management, dead body management etc was conducted. study of the underlying technical issues was not part of the survey because it requires a separate scientific technical survey. similarly, in the end, four questions related to disaster response preparedness activities i.e. training, exercise, rehearsal, and operational engagement were asked to the participants. likewise, the second survey was conducted amongst policy level officers. altogether 7 questions were asked to them. how do they perceive about the engagement of the nepal army in disaster response operations was the scope of this survey. fundamental issues of present engagement and its effect on civil-military relations was the range of the study. ‘how the policymakers want to drive this venture in future’ was also the part survey. three types of closed-ended saqs were designed for the surveys. first was of likert scale of five optional indicators of intensity rating scale i.e. (1) strongly disagree (2) disagree (3) uncertain (4) agree (5) strongly agree. the second was of ‘check all that apply’ type questions and third was dichotomous questions (callegaro, 2014). during analysis, all the replies were re-coded into intensity rating scale. 2.4. method of analysis fundamentally, the descriptive analysis method and inferential analysis method were adopted during research that helped to examine the relationship between the nepal army’s disaster response competency building initiative and its effect on personnel competency. for the purpose, three key independent variables i.e. ssk, tsk, and disaster preparedness adaptation were considered as study factors. these variables were again analyzed on the basis of two dependent variables i.e. ‘rank’ (officer, jco and other ranks) and ‘training’ (trained and untrained troops). whether these variables make any significant difference in the study participants’ knowledge and preparedness or not was the aim of the analysis process. similarly during the second survey prevailing perception of senior army officers on the engagement of the nepalese army in disaster response was analyzed. analytical tools such as bar graphs and comparison table, and tools of linear regression such as pearson correlation matrix, model summary matrix, anova, coefficients matrix were used during the analysis process. through cross tabulation frequency distribution of the responses was studied. and the results were illustrated through bar graphs. for analysis the results of indicators of intensity rating scale were rephrased. for example, in an issue, if the respondent participant replied ‘strongly agree’ then it was considered that he/she was well informed in that issue. again through linear regression, the relationship between independent and dependent variables were examined. for example, the ‘ssk’ was examined amongst the officers and jcos through cross tabulation and whether that relation is statistically significant or not was verified through linear regression. results of bar graphs, model summary, anova, and coefficients were illustrated in figures. detail analysis was conducted using available commercial software and tools. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 66 3. result and analysis 3.1. soft skill knowledge (ssk) eight fundamental questions were asked to the participant jcos and officers to study their ssk. the result of the survey indicated that more than 50% of the participants were not familiar with ssk issues (see figure 1). during the survey, only 35% of the participants strongly accepted that the army is the secondary responder. rest 65% didn’t recognize that the civilian mechanism is primarily responsible to save lives and property of the victims. they lacked to understand that the mobilization of the army is to substantiate the civilian effort. similarly, more than 65% of the responders seemed confused about the issue that the army is the last resort for disaster response. only 6% of the responders accepted that the army is the last resort for disaster response. this result illustrates that the participant officers and jcos have a tendency of overestimating the role of the army in disaster response. likewise, only 40% of the participants were found to be familiar with the fact that the engagement of the army in disaster management will help to enhance cmr. only about 30% of participants gave their consent that the role of disaster management is suitable for the army. in one hand overemphasizing the role of the army and in the other hand underestimating the significance of such engagement illustrates the confused state of knowledge of study participants. but surprisingly in command control aspects, the responders were found quite knowledgeable. 70% responder strongly agreed that for an effective response there should be joint command system comprised of civilian and military officials. similarly, 63% strongly agreed that the local authority should take the lead of response operation and 51% strongly believed that the moha is the leading agency for disaster response. but only 42% of participants were found aware that local government is the most responsible government for disaster management. still, 58% of participants were unaware of this knowledge. however, the overall responses of the participants showed that more than 50% of the participant jocs and officers are still confused about ssk. indeed such deficiency in junior leadership may create a dilemma in mobilization, command control and timely response. 3.2. technical skill knowledge (tsk) during the survey, ten questions were asked to the study participants to study their perception about fundamental technical tasks of disaster response. although all the tasks are equally important for effective response only less than 30% of the participants recognized this fact. around 70% of participants were found less aware of tsk (see figure 2). during the survey, only 50% of participants strongly agreed that the task of information collection and dissemination during disaster’ and ‘providing medical     assistance to victims’ are important for the army. but it is really surprising that more than 70% of the participants were not convinced that search and rescue, debris management; dead body management are important for the army during disaster response. only less than 30% of participants choose them as the most significant. indeed such confusion is counterproductive for an effective response because it creates negligence and lethargy amongst the responders resulting ignorance for preparedness. this result has indeed raised a question in the understanding level of study participants in tsk. similarly, 73% of respondents have failed to identify engineering assistance during post-disaster scenario as an important task for the army. only 27% of participants strongly agreed that the army should construct temporary shelters for victims in the aftermath of the disaster and 23% of the participants considered assisting local authorities in restructuring damaged structures as most significant. likewise, only 30% of the participants strongly believed that the task of providing life-saving services such as food and water to disaster victim should be conducted by the troops. this is again another surprising result derived from the survey. similarly, 40% of responders identified that educating local people for disaster preparedness and assisting local authority to formulate disaster response plan is equally important for the army. an overall impression of the result suggested that nearly 70% of the respondents are still confused about prioritizing their operational tasks for disaster response. the level of their tsk is quite low. 3.3. disaster response preparedness activities four questions were asked to the participants to study their involvement in disaster response preparedness activities. the survey showed that amongst total participants 57% have taken part in some sort of disaster management training in their career whereas 52 % have participated in a disaster response exercise (see figure 3). indeed this is an encouraging finding that half of the participants have been acquainted with some kind of disaster preparedness activities. but in the part of the rehearsal, the response was low. only 30% of participants replied that they use to rehearse on a regular basis at their unit. during survey weekly and monthly rehearsal was considered a regular rehearsal. rest 70% of the respondents didn’t have consistency in their rehearsal activities. similarly when asked about their engagement in disaster response operations 73 % replied that they have been involved in some kind of rescue operations in the past (see figure 3). if we look at the data still 16% of participants involved in disaster response operations were not trained in response operations. this result illustrates the nepalese army’s trend of using untrained troops in disaster response operations. similarly, in comparison to operational engagement, the trend of regular rehearsal is also low. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 66 67   assistance to victims’ are important for the army. but it is really surprising that more than 70% of the participants were not convinced that search and rescue, debris management; dead body management are important for the army during disaster response. only less than 30% of participants choose them as the most significant. indeed such confusion is counterproductive for an effective response because it creates negligence and lethargy amongst the responders resulting ignorance for preparedness. this result has indeed raised a question in the understanding level of study participants in tsk. similarly, 73% of respondents have failed to identify engineering assistance during post-disaster scenario as an important task for the army. only 27% of participants strongly agreed that the army should construct temporary shelters for victims in the aftermath of the disaster and 23% of the participants considered assisting local authorities in restructuring damaged structures as most significant. likewise, only 30% of the participants strongly believed that the task of providing life-saving services such as food and water to disaster victim should be conducted by the troops. this is again another surprising result derived from the survey. similarly, 40% of responders identified that educating local people for disaster preparedness and assisting local authority to formulate disaster response plan is equally important for the army. an overall impression of the result suggested that nearly 70% of the respondents are still confused about prioritizing their operational tasks for disaster response. the level of their tsk is quite low. 3.3. disaster response preparedness activities four questions were asked to the participants to study their involvement in disaster response preparedness activities. the survey showed that amongst total participants 57% have taken part in some sort of disaster management training in their career whereas 52 % have participated in a disaster response exercise (see figure 3). indeed this is an encouraging finding that half of the participants have been acquainted with some kind of disaster preparedness activities. but in the part of the rehearsal, the response was low. only 30% of participants replied that they use to rehearse on a regular basis at their unit. during survey weekly and monthly rehearsal was considered a regular rehearsal. rest 70% of the respondents didn’t have consistency in their rehearsal activities. similarly when asked about their engagement in disaster response operations 73 % replied that they have been involved in some kind of rescue operations in the past (see figure 3). if we look at the data still 16% of participants involved in disaster response operations were not trained in response operations. this result illustrates the nepalese army’s trend of using untrained troops in disaster response operations. similarly, in comparison to operational engagement, the trend of regular rehearsal is also low. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 68   3.4. effect of rank and training in the competency the competency of the study participants was also evaluated using correlation matrix pearson correlation result suggests that the participants who were knowledgeable in tsk were also knowledgeable in ssk and adopted in preparedness, whereas participants who were knowledgeable in ssk were also knowledgeable in tsk but not necessarily adopted in preparedness(see table 1). and those who were adopted in preparedness were not necessarily bear ssk. table 1. correlation matrix of competency competency ssk tsk preparedness activities pearson correlation 1 .205* -.051skk sig. (2-tailed) .041 .566 pearson correlation .205* 1 -.193*tsk sig. (2-tailed) .041 .041 pearson correlation -.051 -.193* 1preparedness activities sig. (2-tailed) .566 .041 * correlation is significant at the 0.05 level (2-tailed). to study whether the participants’ training in disaster response makes any significant difference in their competency linear regression test was conducted. as per the equation derived (f (3, 82) =11.384, p <0.000) the disaster response training of participants is statistically significant to their response competency as a whole (see table 2). but when compared individually, coefficients test result illustrated that only the training has statistically significant relation with preparedness activities where the p-value is less than 0.05 (p<0.000) and t value is 5.539 (see table 3). that means the trained troops were more prepared than untrained troops, whereas with skk and tsk training factor do not have any significant relation since p values are higher than 0.05. similarly the rank factor also didn’t make any statistically significant difference in the competency since the equation derived is (f (3, 83) =2.222, p >0.092) (see table 4). and coefficients result also illustrated that there was no statistical significance of rank with ssk, tsk, and preparedness individually since all p values are greater than 0.05 (see table 5). during analysis, it was found that in some issues of ssk only small percentage of officers have better table 2. anova model sum of squares df mean square f sig. regression 6.045 3 2.015 .000a11.384 residual 14.514 82 .177 1 total 20.558 85 a. predictors: (constant), preparedness activities, ssk, tsk b. dependent variable: disaster management training table 3. coefficients knowledge than jcos. otherwise there is no such difference in tsk of officer, jcos and other ranks (see tables 6 & 7). similarly in preparedness as well, officers are marginally less trained, exercised and rehearsed than jocs and other ranks (see figure 4). but the involvement in response operations is almost equal. when it comes to training, besides preparedness there is no evidential difference in the percentage of trained and untrained troops in ssk and tsk (see tables 6 & 7 and figure 5). 3.5. prevailing engagement perception at the policy level ten questions were asked to the participant officers from policy level to analyze their perception for the engagement of the nepalese army in disaster response. 84% of participants agreed that the army should be engaged more in disaster response in the future. but there also prevailed a huge gap (75%) in the understanding that the government should invest more in establishing an effective civil defense mechanism for disaster response (see figure 6). even 32% denied the fact that the nepalese army is filling the gap of lacked civil defense mechanism in the country. similarly, 84% of the participant agreed that in the federal system there should be dedicated disaster response units at each provincial level. another 88% of respondents agreed that there should be specialized teams for search and rescue operations at each battalion level. although this provision has been ingrained in the nepalese army’s policy its robust implementation has not been achieved yet. many scholars have pointed out that over-reliance on the military for disaster response might garner threat to civil-military relations (cmr) (madiwale & virk, 2011). about 88% of the respondents agreed that there should be joint command control set up of civilian and unstandardized coefficients standardized coefficients model b std. error beta t sig. (constant) -.417 .655 -.636 .526 ssk .145 .146 .095 .995 .323 tsk .003 .076 .004 .045 .964 1 prep1 .798 .144 .527 5.539 .000 a. dependent variable: disaster management training table 4. anova model sum of squares df mean square f sig. regression 4.827 3 1.609 .092a2.222 residual 60.092 83 .724 1 total 64.920 86 a. predictors: (constant), preparedness activities, ssk, tsk b. dependent variable: rank table 5. coefficients unstandardized coefficients standardized coefficients model b std. error beta t sig. (constant) -1.626 1.324 -1.228 .223 ssk .557 .295 .206 1.888 .062 tsk .169 .153 .122 1.101 .274 1 prep1 .238 .291 .089 .819 .415 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 68 69 table 3. coefficients knowledge than jcos. otherwise there is no such difference in tsk of officer, jcos and other ranks (see tables 6 & 7). similarly in preparedness as well, officers are marginally less trained, exercised and rehearsed than jocs and other ranks (see figure 4). but the involvement in response operations is almost equal. when it comes to training, besides preparedness there is no evidential difference in the percentage of trained and untrained troops in ssk and tsk (see tables 6 & 7 and figure 5). 3.5. prevailing engagement perception at the policy level ten questions were asked to the participant officers from policy level to analyze their perception for the engagement of the nepalese army in disaster response. 84% of participants agreed that the army should be engaged more in disaster response in the future. but there also prevailed a huge gap (75%) in the understanding that the government should invest more in establishing an effective civil defense mechanism for disaster response (see figure 6). even 32% denied the fact that the nepalese army is filling the gap of lacked civil defense mechanism in the country. similarly, 84% of the participant agreed that in the federal system there should be dedicated disaster response units at each provincial level. another 88% of respondents agreed that there should be specialized teams for search and rescue operations at each battalion level. although this provision has been ingrained in the nepalese army’s policy its robust implementation has not been achieved yet. many scholars have pointed out that over-reliance on the military for disaster response might garner threat to civil-military relations (cmr) (madiwale & virk, 2011). about 88% of the respondents agreed that there should be joint command control set up of civilian and unstandardized coefficients standardized coefficients model b std. error beta t sig. (constant) -.417 .655 -.636 .526 ssk .145 .146 .095 .995 .323 tsk .003 .076 .004 .045 .964 1 prep1 .798 .144 .527 5.539 .000 a. dependent variable: disaster management training table 4. anova model sum of squares df mean square f sig. regression 4.827 3 1.609 .092a2.222 residual 60.092 83 .724 1 total 64.920 86 a. predictors: (constant), preparedness activities, ssk, tsk b. dependent variable: rank table 5. coefficients unstandardized coefficients standardized coefficients model b std. error beta t sig. (constant) -1.626 1.324 -1.228 .223 ssk .557 .295 .206 1.888 .062 tsk .169 .153 .122 1.101 .274 1 prep1 .238 .291 .089 .819 .415 journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 70 table 6. ssk as per rank and training rank training description jco officers trained untrained ministry of home affairs (moha) is the lead agency for disaster management 77.10% 54.30% 61.40% 72.30% local authority should take lead in disaster response 67.40% 72.70% 68.20% 70.50% joint command concept is ideal for effective disaster response 68.20% 74.40% 73.20% 65.10% nepalese army is the secondary responder for disaster response 25% 47.80% 31.80% 40.40% nepal army is the last resort for disaster response 4.30% 2.20% 0.90% 6.40% engagement in disaster management will enhance civil military relation (cmr) 34.10% 69.80% 54.80% 50% local government is the most responsible government for disaster management 44.40% 48.70% 42.50% 52.40% nepalese army should play active role in disaster management 29.90% 36.40% 25% 30.20% table 7. tsk as per rank and training rank training description or jco officer trained untrained educate and train local communities about disaster preparedness 40% 35.70% 30.25% 32.80% 42% provide life saving services like food and water to victim 35.90% 37.50% 25% 33.60% 35.10% build temporary shelter to disaster victim 29.40% 28.60% 20% 26.90% 26.80% debris management 14.90% 22.50% 22.20% 18.70% 18.40% collect and disseminate information during disaster 50.40% 48.80% 62.80% 50% 56.50% search and rescue operation during disaster 18.50% 19.50% 17.90% 22% 13.30% assist local authority to formulate disaster response plan 27.70% 19.50% 35% 27.50% 27% provide medical assistance to disaster victim 53.30% 35.90% 34.10% 44.30% 47.40% dead body management 20.60% 12% 7% 15.70% 15% assist local authority to restructure damaged infrastructure 28.80% 13.50% 19% 27.40% 17.30% military authorities for effective response. and this is a promising gesture that nepalese army has accepted civilian control over response operations. to make the venture more effective 80% of participants believed that there should be an autonomous civilian entity for disaster response instead of giving full authority to moha. this is true since moha itself is an over-loaded agency with various responsibilities. similarly, only 28% of officers agreed that disaster response training should be made compulsory for all ranks in the army. the overall result of this survey gives an impression that around 70% of senior officer participants have position perception on the engagement of the nepalese army in disaster response.       4. discussion this research has ultimately helped to understand the competency of the general infantry troops of the nepalese army in disaster response operations. indeed, in the army, junior officers and jcos are the key personalities. they are the one to lead disaster response operations in the field. but the result showed that the junior officers and jcos have limited skk. the issues of civil-military relation, a priority of their engagement during a disaster, and the responsibilities of local government in disaster management are much important at their respective level. it is because at the local level they are the one to play interface between civilian authorities, civil societies and responding troops. they have to establish effective communication, coordination, and cooperation with those civilian bodies and civil communities at the time of preparedness and response. in the void of such knowledge, they can’t drive the venture in an effective way. similarly, tsk was also found less promising amongst the participant infantry troops. this indicates that the general infantry of the nepalese army has confusions on their precise disaster response tasks. only 30% of the participants strongly believed that search and rescue, debris management, dead body management, constructing temporary shelters for victims are important for the disaster response (moha, 2016). such lacking will garner negligence and lethargy during preparedness that will ultimately hamper the response operations during a disaster. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 70 71     4. discussion this research has ultimately helped to understand the competency of the general infantry troops of the nepalese army in disaster response operations. indeed, in the army, junior officers and jcos are the key personalities. they are the one to lead disaster response operations in the field. but the result showed that the junior officers and jcos have limited skk. the issues of civil-military relation, a priority of their engagement during a disaster, and the responsibilities of local government in disaster management are much important at their respective level. it is because at the local level they are the one to play interface between civilian authorities, civil societies and responding troops. they have to establish effective communication, coordination, and cooperation with those civilian bodies and civil communities at the time of preparedness and response. in the void of such knowledge, they can’t drive the venture in an effective way. similarly, tsk was also found less promising amongst the participant infantry troops. this indicates that the general infantry of the nepalese army has confusions on their precise disaster response tasks. only 30% of the participants strongly believed that search and rescue, debris management, dead body management, constructing temporary shelters for victims are important for the disaster response (moha, 2016). such lacking will garner negligence and lethargy during preparedness that will ultimately hamper the response operations during a disaster. journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 72 in the aspect of training, more than 50% of the participants were found participated in some kind of disaster-related training and exercise in their career. it is true that following the essence of national policies and guidelines (moha, 2017; nepalese army, 2017) the organization has credibly invested to train its infantry troops in disaster response training like collapsed structure search and rescue training, medical first responders training, dead body management training, deep water rescue training, and sphere training (nepalese army, 2017). but the result of skk and tsk doesn’t warrant qualitatively conduct of such training. the understandings of trained participants are low in many fundamental issues. and there is not much difference amongst trained troops’ and untrained troops’ responses. surprising only 22% of the trained participants replied that search and rescue is the most suitable tasks for them for disaster response. this sort of gap clearly indicates the need for better orientation of the infantry troops in disaster response venture. further research is suggested to investigate in-depth issues of this aspect. similarly, the result also revealed that the regular disaster response rehearsal is lacking in the regular infantry troops of the nepalese army. margaret o’leary (2004) states that 'the rehearsal is amongst the five core processes of disaster response preparedness that improves responders’ future performance’. it doesn’t only help to test the ability but also helps to enhance muscle memory. regular counseling of the troops, practicing drills, tabletop exercise, and mock exercise are some examples of rehearsal activities. but their regularity is not consistency in the army. only 30% of the responders replied that their units carry out such activities regularly. indeed the revealed result of the level of knowledge and preparedness has raised a question on the disaster response effectiveness of the regular infantry troops of the army. levinson (2008) writes that the militaries only have logistics and human resource for disaster response. hardly the ordinary troops are competent. and the tendency of engagement also reveals that disaster untrained troops are also the parts of disaster response operations. during the survey, around 20% of participants replied that they were untrained though they were engaged in some sort of response operations. this tendency also raises a question in the effectiveness of response operations. during correlation analysis also it was found that who were knowledgeable were better prepared since they have a significant correlation. in that sense also quality training and regular rehearsal are indispensable for effective preparedness of the troops. but during research hesitation was found amongst senior officers to make disaster response training compulsory for regular infantries despite their claim that the disaster response competency of such troops should be enhanced. such tendency will not help the organization to achieve its aim. similarly, their disagreement in the need of an effective civilian response mechanism for disaster also generates a question that ‘whether they want to militarize the disaster response venture in nepal’. indeed such feelings contradict the international principle of disaster management, which warrants civilian supremacy for the venture (miller et al., 2008). such feeling may cause friction amongst civilian and military leadership in the long run. however, the overall result of this research has illustrated that the knowledge base competency and preparedness of the general infantry troops of the nepalese army has limitations. nevertheless, less number of samples populations and exclusion of detail study of disaster response training curriculum of the nepalese army are some limitation of this study. despite such limitation whatever result has been derived from the study is definitely useful for other scholars who intend to learn about the competency of general infantry troops of the nepalese army in disaster response. 5. conclusion there is no doubt that the nepalese army is amongst the key responding tools of the nepalese government to respond to natural disaster. but the knowledge-based disaster response competency and preparedness of the general infantry troops of the nepalese army has remained deficit. although disaster response training facility is satisfactory in the army the prevalent unawareness amongst troops in disaster management fundamental issues raises the question in the quality of those training programs. but in other hands at the policy level, the environment is aspiring. a better result can be expected in the future. for preparedness, regular rehearsal of the troops is the missing part of the organization. such a trend makes troops lethargic during a response operation. also, the contradiction amongst senior officers in making disaster response training compulsory for all troops is also a hurdle to some extent for enhancing disaster response capability of the nepalese army. definitely due to legal mandates the nepalese army has larger obligation to respond effectively during a disaster in the country and for that the maintenance of the competency of its troops is mandatory. and towards achieving the aim this kind of research plays an eyeopening role for the concerned authority. acknowledgements this research has been conducted with the extensive support from the officials and staffs of himalayan conservation group, nepal and the nepalese army. the support from the active and retired senior army officers is very much acknowledged. similarly, the officials from government authorities who rendered their guidance on policy matters are also note worthy of recognition for the success of this research paper. references a. a. selmi, s. egawa, h. sasaki, et al., the sendai framework for disaster risk reduction: renewing the global commitment to people’s resilience, health, and well being, in int j. disaster risk sci. 6 (2015), pp.164-176. unisdr, sendai framework for disaster risk reduction 20152030 (2015). r. k . dahal, g. tuladhar, r. yatabe, et al., knowledge of disaster risk reduction among school students in nepal. geomatic, in natural hazard and risk (2013). moha, national position paper on disaster risk reduction and management nepal, in amcdrr 2018 (ulaanbaatar, 2018) nepal disaster report, the road to sendai (final draft) (2017). moha, nepal disaster report 2015 (2015). h. d. smet, b. schreurs, j. leysen, the response of the disaster management life cycle revisited within the context of “ disasters out of box”, in homeland security & emergency management, 12(2) (2015), pp. 319-350. m. k. lindell, r.w. perry, behavioral foundations of community emergency planning (hemisphere publishing corporation, washington, dc, 1992). a. b. wildavsky, searching for safety (transaction publishers, new brunswick, 1998). d. p. coppala, introduction to international disaster management (elsevier inc. burlington, 2011). f. wex, g. schryen, s. feuerriegel,et al., emergency response in natural disaster management: allocation and scheduling of rescue units, in european journal of operational research 235 (2014), pp. 697-708. r. sahashi, disaster relief and risk reduction in east asia: the role of japan-asean cooperation, in navigation change: asean-japan strategic partnership in east asia and in global governance (2015), p.132. http://jcie.org/researchpdfs/aseanjapan/navchange/8.pdf f. elizabeth, future directions in civil-military responses to natural disasters, conflict prevention in practice: from rhetoric to reality, in acmc paper (2012). a. p. hoff, an analysis of disaster relief and humanitarian supply chains (master’s dissertation) (cranfield university, shrivenham, england, 1999). m. malesic, the impact of military engagement in disaster management on civil-military relations, in current sociology (2015). a. raj, armed forces in disaster response: role reappraisal, in claws journal (2008), pp. 163-176. r. a. hall, c. anita, civil military relations in disaster rescue and relief activities: response to the mudslide in southern leyte, philippines, in south african journal of military studies, 38 (2010). p. k. keen, f. p. v. neto, c.w. nolan, et al., relationships matter: humanitarian assistance and disaster relief in haiti. jiim challenges in the geographic combatant commands, in newsletter (2011,11-23), pp.129-139. j. frank, a neutral’s perspective: the role of austrian armed forces in homeland security, in the quarterly journal (2005), pp.97-120. f. c. cuny, dilemmas of military involvement in humanitarian relief, in soldiers, peacekeepers, and disasters (st. martin’s press, new york, 1991), pp. 52-81. c. a. hofmann, l. hudson, military responses to natural disaster: last resort or inevitable trend, in british red cross (2009). m. r. weeks, organizing for disasters: lessons from the military, in business horizons 50(6) (2007), pp. 479-489. m. thapa, out of barracks: civil military relations in disaster management, a case study of nepalese army’s humanitarian response during 2015 earthquake in nepal (university for peace, san jose, 2016). m. ajay, v. kundrat, civil-military relations in natural disaster: a case study of the 2010 pakistan floods, in international review of the red cross 93(884) (2011). banyan, the 2011 earthquake, tsunami, and nuclear accident in japan: coordinating the u.s. government response, in a banyan analytics case study (2014). g. martinez, j. flint, k. sutton, civil-military-police coordination in disaster management, perspectives from south east asian countries (humanitarian advisory group, 2015).   s. raja, a. arshad, armed forces and disaster management, in american journal of social science 1(3) (american institute of science, 2015),pp.152-157. j. hooker, quake revealed deficiencies of china’s military, in the new york times (2 july, 2008). y. li, p. huang, x. sun, an analysis of the legal basis of pla's participation in earthquake rescue operations, in journal of the institute of disaster prevention science and technology 1(2) (2009), pp. 117–120. m. d. manandhar, g. varughese, a. m. howitt, et al., disaster preparedness and response during political transition in nepal: assessing civil and military roles in the aftermath of the 2015 earthquakes, in the asia foundation (2017), pp. 5-6. a. r. c. marshall, r. adkin, nepalese army gets image boost from quake relief work, in reuters (2015). the constitution of nepal, part 28, article 267 (2015), p.197. disaster risk reduction and management act, part 8, article 18 (2017), pp. 24-26. k. poudel, nepal army by people's side, in new spotlight news magazine 09(16) (march, 2016). journal of risk analysis and crisis response vol. 9, no. 2 (june 2019) 62-73 72 73 in the aspect of training, more than 50% of the participants were found participated in some kind of disaster-related training and exercise in their career. it is true that following the essence of national policies and guidelines (moha, 2017; nepalese army, 2017) the organization has credibly invested to train its infantry troops in disaster response training like collapsed structure search and rescue training, medical first responders training, dead body management training, deep water rescue training, and sphere training (nepalese army, 2017). but the result of skk and tsk doesn’t warrant qualitatively conduct of such training. the understandings of trained participants are low in many fundamental issues. and there is not much difference amongst trained troops’ and untrained troops’ responses. surprising only 22% of the trained participants replied that search and rescue is the most suitable tasks for them for disaster response. this sort of gap clearly indicates the need for better orientation of the infantry troops in disaster response venture. further research is suggested to investigate in-depth issues of this aspect. similarly, the result also revealed that the regular disaster response rehearsal is lacking in the regular infantry troops of the nepalese army. margaret o’leary (2004) states that 'the rehearsal is amongst the five core processes of disaster response preparedness that improves responders’ future performance’. it doesn’t only help to test the ability but also helps to enhance muscle memory. regular counseling of the troops, practicing drills, tabletop exercise, and mock exercise are some examples of rehearsal activities. but their regularity is not consistency in the army. only 30% of the responders replied that their units carry out such activities regularly. indeed the revealed result of the level of knowledge and preparedness has raised a question on the disaster response effectiveness of the regular infantry troops of the army. levinson (2008) writes that the militaries only have logistics and human resource for disaster response. hardly the ordinary troops are competent. and the tendency of engagement also reveals that disaster untrained troops are also the parts of disaster response operations. during the survey, around 20% of participants replied that they were untrained though they were engaged in some sort of response operations. this tendency also raises a question in the effectiveness of response operations. during correlation analysis also it was found that who were knowledgeable were better prepared since they have a significant correlation. in that sense also quality training and regular rehearsal are indispensable for effective preparedness of the troops. but during research hesitation was found amongst senior officers to make disaster response training compulsory for regular infantries despite their claim that the disaster response competency of such troops should be enhanced. such tendency will not help the organization to achieve its aim. similarly, their disagreement in the need of an effective civilian response mechanism for disaster also generates a question that ‘whether they want to militarize the disaster response venture in nepal’. indeed such feelings contradict the international principle of disaster management, which warrants civilian supremacy for the venture (miller et al., 2008). such feeling may cause friction amongst civilian and military leadership in the long run. however, the overall result of this research has illustrated that the knowledge base competency and preparedness of the general infantry troops of the nepalese army has limitations. nevertheless, less number of samples populations and exclusion of detail study of disaster response training curriculum of the nepalese army are some limitation of this study. despite such limitation whatever result has been derived from the study is definitely useful for other scholars who intend to learn about the competency of general infantry troops of the nepalese army in disaster response. 5. conclusion there is no doubt that the nepalese army is amongst the key responding tools of the nepalese government to respond to natural disaster. but the knowledge-based disaster response competency and preparedness of the general infantry troops of the nepalese army has remained deficit. although disaster response training facility is satisfactory in the army the prevalent unawareness amongst troops in disaster management fundamental issues raises the question in the quality of those training programs. but in other hands at the policy level, the environment is aspiring. a better result can be expected in the future. for preparedness, regular rehearsal of the troops is the missing part of the organization. such a trend makes troops lethargic during a response operation. also, the contradiction amongst senior officers in making disaster response training compulsory for all troops is also a hurdle to some extent for enhancing disaster response capability of the nepalese army. definitely due to legal mandates the nepalese army has larger obligation to respond effectively during a disaster in the country and for that the maintenance of the competency of its troops is mandatory. and towards achieving the aim this kind of research plays an eyeopening role for the concerned authority. acknowledgements this research has been conducted with the extensive support from the officials and staffs of himalayan conservation group, nepal and the nepalese army. the support from the active and retired senior army officers is very much acknowledged. similarly, the officials from government authorities who rendered their guidance on policy matters are also note worthy of recognition for the success of this research paper. references a. a. selmi, s. egawa, h. sasaki, et al., the sendai framework for disaster risk reduction: renewing the global commitment to people’s resilience, health, and well being, in int j. disaster risk sci. 6 (2015), pp.164-176. unisdr, sendai framework for disaster risk reduction 20152030 (2015). r. k . dahal, g. tuladhar, r. yatabe, et al., knowledge of disaster risk reduction among school students in nepal. geomatic, in natural hazard and risk (2013). moha, national position paper on disaster risk reduction and management nepal, in amcdrr 2018 (ulaanbaatar, 2018) nepal disaster report, the road to sendai (final draft) (2017). moha, nepal disaster report 2015 (2015). h. d. smet, b. schreurs, j. leysen, the response of the disaster management life cycle revisited within the context of “ disasters out of box”, in homeland security & emergency management, 12(2) (2015), pp. 319-350. m. k. lindell, r.w. perry, behavioral foundations of community emergency planning (hemisphere publishing corporation, washington, dc, 1992). a. b. wildavsky, searching for safety (transaction publishers, new brunswick, 1998). d. p. coppala, introduction to international disaster management (elsevier inc. burlington, 2011). f. wex, g. schryen, s. feuerriegel,et al., emergency response in natural disaster management: allocation and scheduling of rescue units, in european journal of operational research 235 (2014), pp. 697-708. r. sahashi, disaster relief and risk reduction in east asia: the role of japan-asean cooperation, in navigation change: asean-japan strategic partnership in east asia and in global governance (2015), p.132. http://jcie.org/researchpdfs/aseanjapan/navchange/8.pdf f. elizabeth, future directions in civil-military responses to natural disasters, conflict prevention in practice: from rhetoric to reality, in acmc paper (2012). a. p. hoff, an analysis of disaster relief and humanitarian supply chains (master’s dissertation) (cranfield university, shrivenham, england, 1999). m. malesic, the impact of military engagement in disaster management on civil-military relations, in current sociology (2015). a. raj, armed forces in disaster response: role reappraisal, in claws journal (2008), pp. 163-176. r. a. hall, c. anita, civil military relations in disaster rescue and relief activities: response to the mudslide in southern leyte, philippines, in south african journal of military studies, 38 (2010). p. k. keen, f. p. v. neto, c.w. nolan, et al., relationships matter: humanitarian assistance and disaster relief in haiti. jiim challenges in the geographic combatant commands, in newsletter (2011,11-23), pp.129-139. j. frank, a neutral’s perspective: the role of austrian armed forces in homeland security, in the quarterly journal (2005), pp.97-120. f. c. cuny, dilemmas of military involvement in humanitarian relief, in soldiers, peacekeepers, and disasters (st. martin’s press, new york, 1991), pp. 52-81. c. a. hofmann, l. hudson, military responses to natural disaster: last resort or inevitable trend, in british red cross (2009). m. r. weeks, organizing for disasters: lessons from the military, in business horizons 50(6) (2007), pp. 479-489. m. thapa, out of barracks: civil military relations in disaster management, a case study of nepalese army’s humanitarian response during 2015 earthquake in nepal (university for peace, san jose, 2016). m. ajay, v. kundrat, civil-military relations in natural disaster: a case study of the 2010 pakistan floods, in international review of the red cross 93(884) (2011). banyan, the 2011 earthquake, tsunami, and nuclear accident in japan: coordinating the u.s. government response, in a banyan analytics case study (2014). g. martinez, j. flint, k. sutton, civil-military-police coordination in disaster management, perspectives from south east asian countries (humanitarian advisory group, 2015).   s. raja, a. arshad, armed forces and disaster management, in american journal of social science 1(3) (american institute of science, 2015),pp.152-157. j. hooker, quake revealed deficiencies of china’s military, in the new york times (2 july, 2008). y. li, p. huang, x. sun, an analysis of the legal basis of pla's participation in earthquake rescue operations, in journal of the institute of disaster prevention science and technology 1(2) (2009), pp. 117–120. m. d. manandhar, g. varughese, a. m. howitt, et al., disaster preparedness and response during political transition in nepal: assessing civil and military roles in the aftermath of the 2015 earthquakes, in the asia foundation (2017), pp. 5-6. a. r. c. marshall, r. adkin, nepalese army gets image boost from quake relief work, in reuters (2015). the constitution of nepal, part 28, article 267 (2015), p.197. disaster risk reduction and management act, part 8, article 18 (2017), pp. 24-26. k. poudel, nepal army by people's side, in new spotlight news magazine 09(16) (march, 2016). atlantis press journal style collective risk generalization to creditrisk+ reza habibi iran banking institute, central bank of iran, tehran, iran received 19-june 19, 2018 accepted july 25, 2018 abstract using the collective risk models of actuarial science, the creditrisk+ is extended to the case of random number obligors. first, mathematical methods to compute the distribution of total loss are studied. then, the mathematical results are applied and verified numerically. the insufficiency data in risk management is a big problem. thus, the case of data scarce is studied using a bayesian approach. finally, a concluding remarks section is also given. keywords: bayesian inference; collective risk; creditrisk+; modelrisk; data scarce problem; moment generating function; monte carlo simulation 1. introduction in the last two decades, the credit risk modeling has been received considerable attentions in financial literatures (avesani et al., 2006). indeed, evaluation of the default probability of any borrower is the main concern of bankers when they lend to their clients (liao et al. 2009). almost for all debtors, the quantitative modeling of the credit risk is too important. to this end, some risk measures such as the credit value at risk are computed. there are two approaches for modeling credit value at risk: creditmetrics and creditrisk+ (avesani et al., 2006). these approaches are applicable by regulators and risk managers which make decisions about the capital adequacy ratio. the creditmetrics is used for rating (lee, 2011). by this approach, the credit risk is defined as the risk that the security keepers don't materialize the security expected value because the borrower’s credit quality is deteriorated (jarrow, 2011). using creditrisk+ the default models are constructed. that is, in this approach, credit risk is considered as risk that borrower of security failures on his/her promised obligations. therefore, default of borrowers may make losses in the portfolio (huang and yu, 2010). another difference is that the creditrisk+ applies actuarial methodologies to derive the loss distribution of a financial portfolio. in this approach, just the default risk is modeled, and downgrade risk is not considered (xiaohong et al., 2010) as it is stated, the creditrisk+ provides an actuarial based framework for quantitative credit risk management. this software computes the portfolio and other debt instruments loss distributions which lead to determination of the required economic capital. avesani et al. (2006) reviewed basic creditrisk+ models exist in the literatures and proposed some generalizations including latent factors and random probabilities. creditrisk+ basic model is very similar to the individual risk model of actuarial risk theory (see, kaas et al., 2008) at which, following notation avesani et al. (2006), the total normalized loss λ of 𝑛 obligor's normalized losses 𝜆𝑖 = 𝐷𝑖 𝑣𝑖 , 𝑖 = 1, … , 𝑛 is given by eq. (1), as follows 𝜆 = 𝜆𝑖 𝑛 𝑖=1 = 𝐷𝑖 𝑣𝑖 𝑛 𝑖=1 , (1) where the i -th default di of eq. (1) occurs (is one) with probability 𝑝𝑖 and is zero with probability 1 − 𝑝𝑖 and 𝑣𝑖 is the 𝑖 -th normalized 185 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 exposure. although, avesani et al. (2006) assumed that 𝑣𝑖 's are fixed, however, it is possible to assume that they are random variables with gamma or log-normal (generally positive skewed) distributions. a natural extension (and different of latent factors considered by avesani et al.) to the basic model of creditrisk+ is the collective risk model where the number of obligors 𝑛 who contribute exposure, itself, is also a random variable denoted by 𝑁. it is assumed that 𝑁 is independent of, 𝐷𝑖 's and 𝑣𝑖 's. however, usually 𝐷𝑖 's and 𝑣𝑖 's are correlated. in literatures, usually, selected distributions for 𝑁 are negative binomial and poisson laws. thus, like eq. (1), the collective (credit) risk of a portfolio is presented as a random sum as λ = 𝐷𝑖 𝑣𝑖 . 𝑁 𝑖=1 when 𝑁 = 0, then λ=0. this fact implies that λ has a mixture distribution. thus, has a compound distribution. although, avesani et al. (2006) proposed an excel addin mcm cr+, however, for computational purposes of some generalizations of creditrisk+, even generalizations proposed by avesani et al. (2006), the modelrisk excel add-in of vose (2015) seems to be very useful. this paper considers the credit risk management. an interesting fact is that similar pattern of the above mentioned random sum occurs in banking operational risk management using the ama method (see, shevchenko, 2011). the rest of paper is organized as follows. in the next section, mathematical results are presented and creditrisk+ is generalized to the case of collective creditrisk+. the numerical examples are developed in the section 3. the big problem of risk management is the scarcity of data available in hand. hence, the data scarce case is studied using bayesian inferential tools. finally, a concluding remarks section is also proposed. 2. mathematical expression of creditrisk to compute risk measure like value at risk (var), the distribution of λ is needed. as follows, some stylized facts about actuarial risk theory are updated and modified for creditrisk+ designed for credit risk management. for each fact, necessary conditions are stated. (a) the first fact is about the moment generating function of collective impact of credit risk. to this end, assuming all 𝜆𝑖 's have the same distributions, one can see that the moment generating function of λ is given by eq. (2), as follows 𝑀λ 𝑥 = 𝑀𝑁 (log 𝑀𝜆𝑖 𝑥 , (2) (kaas et al., 2008, see page 43, eq. (3.5)). for example, when 𝑁 has geometric distribution with parameter 0 < 𝑝 < 1 and λ𝑖 's are exponentially distributed with parameter 1 and independent, then eq. (2) reduces to 𝑀𝜆 𝑥 as follows 𝑀𝜆 𝑥 = 𝑝 + (1 − 𝑝) 𝑝 𝑝 − 𝑥 which is the moment generating function of a mixture distribution. although, there is a closed form for 𝑀𝜆 𝑥 , however, this is not true, generally and monte carlo simulation should be used. (b) a natural method in actuarial science is the approximating the distribution of total claim. one can see that in collective risk models, the central limit theorem (clt) approximation doesn't work well and two more accurate approximations for distribution of λ are translated gamma (tg) and normal power (np) approximations. traditionally, there exist two justifications for np approximations. first, based on approximating the distribution of 𝑍 + 𝛾𝜆 6 (𝑍2 − 1) where 𝑍 has standard normal distribution. the second justification is derived using edgeworth expansion. to apply both tg and np methods, the first three moments are needed. let the cumulant generating function of λ be 186 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 𝐾𝜆 𝑥 = log 𝑀λ 𝑥 . indeed, in the case of identically distributed 𝜆𝑖 's, then the cumulant function is given by eq. (3), as follows 𝐾𝜆 𝑥 = 𝐾𝑁 𝐾𝜆𝑖 𝑥 . (3) notice that the cumulant function of eq. (3) is approximated by eq. (4), as given in below 𝐾𝜆 𝑥 = 𝐸 λ 𝑥 + 𝑣𝑎𝑟 λ 𝑥 2 2 + 𝐸 λ − e λ 3 𝑥 3 6 + 𝑂 𝑥4 , (4) (kaas et al., 2008, see page 29). (c) here, the mean and variance of 𝜆 are derived. they are given in eq. (5) 𝐸 λ = 𝐸( 𝐸(𝜆𝑖 )), 𝑁 𝑖=1 𝑣𝑎𝑟 λ = e( 𝑣𝑎𝑟(𝜆𝑖 )) + 𝑣𝑎𝑟( 𝐸(𝜆𝑖 )). 𝑁 𝑖=1 𝑁 𝑖=1 (5) again, when 𝜆𝑖 's are identically distributed, then these moments are reduced to the eq. (6) 𝐸 𝜆 = 𝐸 𝑁 𝐸 𝜆𝑖 , 𝑣𝑎𝑟 λ = 𝐸 𝑁 𝑣𝑎𝑟 𝜆𝑖 + 𝐸 2 𝜆𝑖 𝑣𝑎𝑟 𝑁 , (6) (kaas et al., 2008, see page 29, eqs. (3.3, 3.4)). (d) as follows, the skew and kurtosis indices of are derived. these values are necessary to find the distribution of . to this end, notice that the following eq. (7) is correct that 𝐸 λ − e λ 3 = 𝐸( 𝐸(𝜆𝑖 − 𝐸(𝜆𝑖 )) 3𝑁 𝑖=1 . (7) when 𝜆𝑖 's are identically distributed, then equation (7) is changed to the eq. (8), as follows 𝐸 λ − e λ 3 = 𝐸 𝑁 𝐸(𝜆𝑖 − 𝐸(𝜆𝑖 )) 3, (8) (lehmann and casella 1998, see page 29, eq. (5.26)). thus, to find the moments of , the marginal distributions of 𝜆𝑖 's are needed. to this end, notice that, each 𝜆𝑖 has a mixture distribution. to see this, notice that following eq. (9) as follows, we have 𝑀𝜆𝑖 𝑥 = 𝐸 𝐸 𝑒 𝐷𝑖𝑣𝑖𝑥 𝐷𝑖 = 𝐸 𝑀𝑣𝑖 𝐷𝑖 𝑥 = 𝑝𝑖 𝑀𝑣𝑖 𝑥 + 1 − 𝑝𝑖 . (9) this shows that the marginal distribution of 𝜆𝑖 is mixture of two components laws of 𝑣𝑖 and degenerate distribution on zero. to find moments of 𝜆𝑖 , it is seen that the following eq. (10) is correct that 𝐸 𝜆𝑖 = 𝜕 𝑀𝜆 𝑖 (𝑥 ) 𝜕𝑥 |𝑥 =0 = 𝑝𝑖 𝐸 𝑣𝑖 , 𝐸 𝜆𝑖 𝑘 = 𝑝𝑖 𝐸 𝑣𝑖 𝑘 , 𝑘 ≥ 2. (10) thus, the variance of skewness of 𝜆𝑖 are given by eq. (11), as follows 𝑣𝑎𝑟 𝜆𝑖 = 𝑝𝑖 𝐸 𝑣𝑖 2 − 𝑝𝑖 2𝐸2 𝑣𝑖 , 𝐸((λ𝑖 − 𝐸(λ𝑖 )) 3 = 𝑝𝑖 𝐸 𝑣𝑖 3 + 2𝑝𝑖 3𝐸3 𝑣𝑖 − 3𝑝𝑖 𝐸 𝑣𝑖 𝐸 𝑣𝑖 2 . (11) (e) here, the density function of 𝜆 is derived. to this end, let 𝑓𝜆1→𝑖 ∗𝑖 be the convolution of densities of 𝜆𝑗 , 𝑗 = 1, … , 𝑖 for 𝑖 ≥ 1 . then, the density function 𝑓𝜆 𝑦 , is given by eq. (12) as below 𝑓𝜆 𝑦 = 𝑃 𝑁 = 0 + 𝑓𝜆1→𝑖 ∗𝑖 𝑦 𝑃 𝑁 = 𝑖 .∞𝑖=1 (12) assuming 𝜆𝑗 has normal distribution with parameter 𝜇 and variance 𝜍2, then 𝑓𝜆1→𝑖 ∗𝑖 𝑦 is the density of normal distribution with mean 𝑖𝜇 and 𝑖𝜍2 . the same result is correct for 𝜆𝑗 has cauchy distributions, i.e. if 𝜆𝑗 is 𝐶(𝜋, 𝛿), then 𝑓𝜆1→𝑖 ∗𝑖 𝑦 is the density of 𝐶(𝑖𝜋, 𝑖𝛿), (kaas et al., 2008, see page 44, eq. (3.10)). to see why the formula 𝑓𝜆 𝑦 is correct, it is enough notice that the moment generating function of 𝜆 by eq.(3) is given by eq. (13), as follows 187 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 𝑀λ 𝑥 = 𝐸 𝑒 𝑥𝜆 = 𝐸 𝐸 𝑒 𝑥𝜆 𝑁 = 𝑃 𝑁 = 0 + 𝐸 𝑒𝑥𝜆 𝑁 = 𝑖 𝑃 𝑁 = 𝑖 . ∞𝑖=1 (13) (f) again, the density function of 𝜆 is derived in alternative method. let 𝑓𝜆1→𝑖 ∗𝑖 𝑦 be the result of the fast fourier transform (fft) procedure, when 𝑁 = 𝑖 is kept fixed and denote it by 𝐹𝐹𝑇𝑖 (𝑦), hence, it is given by eq. (14) 𝑓𝜆 𝑦 = 𝑃 𝑁 = 0 + 𝐹𝐹𝑇𝑖 𝑦 𝑃 𝑁 = 𝑖 . ∞ 𝑖=1 (14) indeed, for each 𝑖, excel add-in mcm cr+ gives the 𝐹𝐹𝑇𝑖 (𝑦) and therefore, 𝑓𝜆 𝑦 is obtained. since, 𝑃 𝑁 = 𝑖 decreases as 𝑖 gets bigger, thus, it is enough to compute 𝐹𝐹𝑇𝑖 (𝑦) for suitable 𝑖 's. the maximum 𝑖 's is chosen such that 𝑃(𝑁 ≥ 𝑖) is negligible. the aggregate panjer option of modelrisk software computes the density 𝑓𝜆 𝑦 and thus the var is obtained. the following proposition summarizes the above discussion. proposition. in a collective creditrisk+ format, assuming λ𝑖 's are identically distributed, then (i) the moment generating function, mean and variance of are given by eq. (15), as follows 𝑀𝜆 𝑥 = 𝑀𝑁 (log 𝑀λ𝑖 𝑥 , 𝐸 λ = e n e 𝜆𝑖 , 𝑣𝑎𝑟 λ = e n var 𝜆𝑖 + 𝐸 2 𝜆𝑖 𝑣𝑎𝑟 𝑁 . (15) (ii) generally, 𝐸 𝜆𝑖 𝑘 = 𝑝𝑖 𝐸 𝑣𝑖 𝑘 , 𝑘 ≥ 1. (iii) np approximation. for 𝜆∗ ≥ 1, then np approximation is given by eq. (16) 𝑃 λ−𝐸 λ 𝑣𝑎𝑟 λ ≤ 𝜆∗ + 𝛾λ 6 𝜆∗2 − 1 ≈ 𝑃 𝑍 ≤ 𝜆∗ . (16) here, 𝛾λ is the skewness of 𝜆 and 𝑍 has standard normal distribution with zero mean and variance one. (iv) tg approximation. let 𝐹λ be the distribution function of and 𝐺(. , 𝛼, 𝛽) be the distribution function of gamma distribution with parameters 𝛼, 𝛽 . then, 𝐹λ 𝜆 ∗ is approximated as eq. (17) 𝐹λ 𝜆 ∗ ≈ 𝐺 𝜆∗ − 𝜆0 , 𝛼, 𝛽 . (17) here, 𝜆0 , 𝛼, 𝛽 are chosen such that 𝐸 𝜆 = 𝜆0 + 𝛼𝛽, 𝑣𝑎𝑟 𝜆 = 𝛼𝛽2 and 𝛾λ = 2 𝛼 . (v) the density function 𝑓𝜆 𝑦 , is given by eq. (18) 𝑓𝜆 𝑦 = 𝑃 𝑁 = 0 + 𝐹𝐹𝑇𝑖 𝑦 𝑃 𝑁 = 𝑖 , ∞ 𝑖=1 (18) for each 𝑖, excel add-in mcm cr+ gives the 𝐹𝐹𝑇𝑖 𝑦 . remark 1. avesani et al. (2006) showed that 𝐷𝑖 has poisson distribution as 𝑝𝑖 = 𝑝 → 0. let 𝑣𝑖 = 𝑣. then, 𝑀𝜆 𝑥 = 𝐸 𝑒 𝑥𝑣 𝐷𝑖 𝑁 𝑖=1 = 𝐸 𝐸 𝑒 𝑥𝑣 𝐷𝑖 𝑁 𝑖=1 𝑁 = 𝐸 𝑒𝑁𝑝 𝑒 𝑥𝑣 −1 = 𝑒𝜃 (𝑒 𝑢 −1), where 𝑢 = 𝑝 𝑒 𝑥𝑣 − 1 . also, let 𝜆 = 𝜆𝑖 𝑁 𝑖=1 and 𝑁 has poisson distribution with parameter 𝜃 and 𝜃 is a gamma variable with parameters 𝛼, 𝛽, then, it is easy to see that the moment generating function is given by eq. (19) 𝑀𝜆 𝑥 = 1 (1−𝛽 𝑒 𝑔 𝑥 −1 )𝛼 , (19) where 𝑔 𝑥 = 𝑙𝑜𝑔𝑀𝜆𝑖 (𝑥). suppose that 𝜆𝑖 's have power series distributions with density given by 𝑓𝜑 𝜆𝑖 = 𝑒 𝜑 𝜆 𝑖 𝑕 (𝜆𝑖 ) 𝑐 (𝜑 ) . the moment generating function is 𝑀𝜆𝑖 𝑥 = 𝑐(𝜑 +𝑥 ) 𝑐 (𝜑 ) . it is easy to see that the moment generating function given by eq. (20) 𝑀𝜆 𝑥 = 𝑒 𝜃 ( 𝑐 𝜑 +𝑥 𝑐 𝜑 −1) . (20) remark 2. an alternative method to approximate distribution of is to use simultaneously both monte carlo or bootstrap (to simulate moments of 𝜆 ) and edgeworth or cornish-fisher expansions (to simulate 𝐹λ , 188 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 i.e., the distribution function of 𝜆). to this end, the urnto-urn method should be applied in running the monte carlo method to simulate mixture distributions. 3. examples here, using some examples, the above results are applied and verified numerically. example 1. let 𝑁 = 1 is observed and 𝑣 takes two values 𝑎 and 𝑏. here, 𝜆 takes three values 0, 𝑎, 𝑏. let 𝑃 𝑣 = 𝑎, 𝐷 = 1 = 𝑝𝑎 and 𝑃 𝑣 = 𝑏, 𝐷 = 1 = 𝑝𝑏 where 𝑝𝑎 + 𝑝𝑏 = 𝑃 𝐷 = 1 = 𝑝. then, 𝑃 𝜆 = 𝑎 = 𝑝𝑎 and 𝑃 𝜆 = 𝑏 = 𝑝𝑏 . thus, 𝑃 𝑣 = 𝑎|𝐷 = 1 = 𝑝𝑎 𝑝𝑎 +𝑝𝑏 = 1 − 𝑃 𝑣 = 𝑏|𝐷 = 1 . next, suppose that 𝑁 has poisson distribution with intensity parameter 10 and 𝑃 𝑣 = 𝑎|𝐷=1=0.1 and 𝑝=0.1. it is seen that 𝑝𝑎=0.01 and 𝑝𝑏 = 0.09 . indeed, 𝑃 𝜆 = 0 = 0.9 , 𝑃 𝜆 = 𝑎 = 0.01 and 𝑃 𝜆 = 𝑏 = 0.09. for 𝑎 = 10 and 𝑏 = 20 thousand dollars, 𝐸 𝜆𝑖 = 1.8, 𝑣𝑎𝑟 𝜆𝑖 = 33.1 and 𝐸(𝜆𝑖 − 𝐸(𝜆𝑖))3=553.605. thus, 𝐸λ=18, 𝑣𝑎𝑟λ=363.4 and 𝛾λ = 29.07. to apply the tg approximation, it is seen that 𝛼 = 0.00473, 𝛽 = 277.18 and 𝜆0 = 16.69 . however, it doesn't work well, here. using the np approximation, the 0.99 var, in this case, is 317.8 thousand dollars. the exact var is 240 thousand dollars. example 2. suppose that obligors are categorized to three categories. the probability of each category is 𝑝𝑖 . the amount of loss severity is 𝑣𝑖 = 0, 𝑎, 𝑏 for each category. let 𝑁1 , 𝑁2 , 𝑁3 be the number of obligors in the first, second and third category with intensities parameters 𝜃1 , 𝜃2 , 𝜃3 , respectively. thus, 𝐸 𝜆 = 𝑎𝜃2𝑝2 + 𝑏𝜃3𝑝3 𝑣𝑎𝑟 𝜆 = 𝑎2𝑣𝑎𝑟 𝑁2𝐷2 + 𝑏 2𝑣𝑎𝑟 𝑁3𝐷3 , where 𝑣𝑎𝑟 𝑁2𝐷2 = 𝜃2𝑝2 + 𝜃2 2𝑝2 (1 − 𝑝2 ) and 𝑣𝑎𝑟 𝑁3𝐷3 = 𝜃3𝑝3 + 𝜃3 2𝑝3 (1 − 𝑝3 ) . assuming 𝑝2 = 0.02, 𝑝3 = 0.03, 𝑎 = 1, 𝑏 = 2, 𝜃2 = 10 and 𝜃3 = 20, then the 99 percent var is . example 3. assume that a portfolio contains two types of loans. the probability of default in each type is 𝑝𝑖 . if there is a default, then the severity of default has distribution 𝑞𝑖 (𝑣𝑖 ). assuming each type contain 𝑁𝑖 , 𝑖 = 1,2 numbers of obligors having poisson distributions with parameters 𝜃𝑖 , 𝑖 = 1,2.. then, 𝐸 𝜆 = 𝐸(𝑣1 )𝜃1𝑝1 + 𝐸(𝑣2 )𝜃2𝑝2 𝑣𝑎𝑟 𝜆 = 𝑣𝑎𝑟 𝑁1𝐷1𝑣1 + 𝑣𝑎𝑟 𝑁2𝐷2𝑣2 . let 𝜃1 = 10, 𝜃2 = 20, 𝑝1 = 0.01, 𝑝2 = 0.02 and losses 𝑣1 and 𝑣2 are 5 or 10 dollars with probability of 0.5 and 0.5. then, the 99 percent var is . example 4. let 𝑣𝑖 = 0,2 with probability of 0.25 and 0.75, respectively. then the 99 percent var is . suppose that 𝑃 𝑣𝑖 = 𝑣 ∗ 𝐷𝑖 = 1 = 𝑞𝑖 and 𝑃 𝑣𝑖 ∈ 𝑣,𝑣+𝑑𝑣=1−𝑞𝑖𝑣∗) for 0<𝑣<𝑣∗. again, let 𝜆𝑖=𝑣𝑖𝐷𝑖 and 𝑁 has poisson distribution with intensity parameter 𝜃. here, when 𝑞𝑖 = 0.3 , 𝑣 ∗ = 5, 𝜃 = 20 , then the 99 percent var is . example 5. let 𝜆𝑖 = 𝑣𝑖 𝐷𝑖 where 𝑣𝑖 has exponential distribution with scale parameter 𝛽 . therefore, 𝜆𝑖 = 0 with probability of 𝑝𝑖 and 𝜆𝑖 has density of 1 𝛽 𝑒 − 𝜆 𝑖 𝛽 with probability of 1 − 𝑝𝑖 . for 𝑝𝑖 = 0.05, 𝛽 = 1, and 𝑁 is poisson with parameter , the 99 percent var is . 4 data scare case. data scarce is a small sample problem. both shevchenko (2011) and svensson (2015) studied the operational risk management when the data is scarce. here, their approach in operational risk managements is extended to the credit risk management. the main idea of bayesian method to compute 𝑓𝜆 , the density of 𝜆 = 𝜆𝑖 𝑁 𝑖=1 , assuming the priors for 𝑁 and 𝜆𝑖 's are given, is to derive the posterior of them and then to simulate 𝜆, using monte carlo or panjer recursion methods. the aggregate panjer option of modelrisk software is good tool, in this way. also, the excel addin mcm cr+ is a useful instrument to compute the var, directly. here, throughout two examples, these methods are studied. 189 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 remark 3. when we take bayesian approach to deal with it, a priori knowledge is required. how to solve this problem is worth considering. generally, there are two approaches. first, because of computational simplicity, the conjugate priors are selected with hyper-parameters that are estimated using historical data based on empirical bayes method. to this end, some computational methods like em algorithms may be used (lehmann and casella (1998)). these approaches are chosen in the current paper. the second approach searches for suitable prior distribution for parameters to derive the posterior distributions. some computational methods such as monte carlo markov chain (mcmc) are applied, in this case. however, in this paper, since the aim is development of creditrisk+ and mostly data sets are simulated one, therefore, the problem of prior selection and its subjective is not the main body of study. example 6. here, first, the bayesian inference for frequency component of collective creditrisk+ is given. following shevchenko and wuthrich (2006), let 𝑁1 , … , 𝑁𝑘 be a sequence of independent and identically distributed poisson random variables with intensity parameter 𝜃, and suppose that 𝜃 has gamma distribution with parameters 𝛼, 𝛽. shevchenko and wuthrich (2006) showed that 𝜃 has the posterior density of gamma with parameters 𝛼 𝑘 and 𝛽 𝑘 given by 𝛼 𝑘 = 𝛼 + 𝑁𝑖 𝑘 𝑖=1 = 𝛼 𝑘−1 + 𝑁𝑘 and 𝛽 𝑘 = 𝛽 𝑘𝛽 +1 = 𝛽 𝑘−1 1+𝛽 𝑘−1 and 𝐸 𝑁𝑘 +1 𝑁1 , … , 𝑁𝑘 = 𝑧𝑁 + 1 − 𝑧 𝛼𝛽, where 𝑧 (based on credibility theory) is the credible factor given by 𝑧 = 𝑘𝛽 𝑘𝛽 +1 . again, the bayesian modeling of severities are proposed. here, in spite of shevchenko and wuthrich (2006) which assumes a log-normal law for severity distribution, it is assumed that 𝜆𝑗 = 𝐷𝑗 𝑣𝑗 , 𝑗 = 1,2, … , 𝑀 where 𝐷𝑗 has bernoulli distribution with parameter 𝑝 and 𝑝 has beta distribution with parameters 𝜔, 𝜑 . thus, 𝑝 has beta distribution with parameters 𝜔 + 𝐷𝑗 𝑀 𝑗 =1 and 𝜑 + 𝑘 − 𝐷𝑗 𝑀 𝑗 =1 . also, 𝑣𝑖 has log-normal distribution with parameters 𝜇 and 𝜍2 . here, for simplicity arguments, it is assumed that 𝜍2 is known and 𝜇 has prior normal with mean к and standard deviation 𝜁 . again, shevchenko and wuthrich (2006) showed that 𝜇 has normal distribution with mean m 𝜍 2 +𝜁 2 𝑣𝑗 𝑀 𝑗 =1 m 𝜍 2 +𝜁 2 and variance 𝜍 2𝜁 2 m 𝜍 2 +𝜁 2 . indeed, using hyperparameters and real data, posteriors estimates of parameters are computed. then, 𝐷𝑗 , 𝑣𝑗 , then 𝜆𝑗 as well as n are simulated. finally, using the panjer recursive "eq. the density of λ are computed and the bayesian risk measure var is computed example 7. following shevchenko (2006), let 𝜆 = 𝜆𝑥 + 𝜆𝑦 . here, 𝜆𝑥 = 𝜆𝑗 𝑥𝑁1 𝑗 =1 and 𝜆 𝑦 = 𝜆𝑗 𝑦𝑁2 𝑗 =1 with 𝜆𝑗 𝑥 = 𝐷𝑗 𝑥 𝑣𝑗 𝑥 and 𝜆𝑗 𝑦 = 𝐷𝑗 𝑦 𝑣𝑗 𝑦 where 𝑣𝑗 𝑥 and 𝑣𝑗 𝑦 are correlated with correlation coefficient 𝜌 . indeed, gaussian, t, clayton, frank and gumbel copulas are used to model dependency between 𝑣𝑗 𝑥 and 𝑣𝑗 𝑦 . 4. conclusions this paper considers the collective creditrisk+ extensions to the regular creditrisk+ version. indeed, here, it is assumed that the number of default obligors are random obeys a specified distribution say poisson or negative binomial. also, the severity of exposures are assumed to be random. in the new format, some extended approaches are given to compute the density of total loss variable which is needed to calculate the var risk measure. the bayesian solution is developed to the problem of scarcity of data as well as correlations between exposures. extending the collective creditrisk+ models to latent factors seems to be straightforward which is omitted. acknowledgements author thanks the referee for several improving suggestions. references r. g. avesani, k. liu, a. mirestean and j. salvati, review and implementation of credit risk models in the financial sector assessment program, imf working paper 06/134 (washington: international monetary fund). (2006). 190 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 available via http:// www .imf .org /external /pubs /cat /longres.aspx ?sk=19111. s. j. huang, and j. yu, bayesian analysis of structural credit risk models with microstructure noises. journal of economic dynamics & control 34, (2010) 2259-2272. r. a. jarrow, credit market equilibrium theory and evidence: revisiting the structural versus reduced form credit risk model debate. finance research letters 8 (2011), 2–7. r. kaas, m. goovaerts, j. dhaene, and m. denuit, modern actuarial risk theory using r (springer, 2006, usa). w. c. lee, redefinition of the kmv model’s optimal default point based on genetic-algorithms–evidence from taiwan. expert systems with applications 38 (2011), 1010-1022. h. h. liao, t. k. chen and c. w. lu, bank credit risk and structural credit models: agency and information asymmetry perspectives. journal of banking & finance 33 (2009), 1520-1530. p. v. shevchenko, modeling operational risk. technical report. csiro division of mathematical and information sciences, sydney (2006). australia. p. v. shevchenko, modeling operational risk using bayesian inference (springer, 2011, usa). p. v. shevchenko and m. v. wüthrich, the structural modeling of operational risk via bayesian inference: combining loss data with expert opinions. the journal of operational risk 1 (2006), 3-26. k. p. svensson. a bayesian approach to modeling operational risk when data is scarce. master's theses in mathematical sciences. lund university. sweden. (2015) d. vose, risk analysis. vailable:http://www.vosesoftware.com (2015). c. xiaohong, w. xiaoding, and w.d desheng, credit risk measurement and early warning of smes: an empirical study of listed smes in china. decision support systems 49 (2010), 301–310. 191 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 185 -191 microsoft word volume 11, issue 4-3 journal of risk analysis and crisis response, 2021, 11(4), 176-188 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i4.311 176 article research on enterprise credit risk prediction based on text information haonan zhang 1,2, hongmei zhang 1,2,* and mu zhang 1 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhm1035@qq.com; tel.: +86-0851-88510575 received: september 3, 2021; accepted: december 26, 2021; published: january 25, 2022 abstract: this paper uses the text data mining method to separate the intonation in the annual reports of credit risk enterprises and non-credit risk enterprises, quantify it, and study the impact of annual report intonation on the effectiveness of credit risk prediction. in the empirical research, this paper uses the factor analysis method for some traditional financial variables, and uses the extracted components and intonation variables to predict the credit risk through the logistic model. the results show that the tone of enterprises with credit risk is more negative, and the degree of pessimism is significantly positively correlated with the probability of credit risk. by comparing the roc curves of the prediction results before and after the addition of intonation variables, adding intonation variables to the credit risk prediction based on financial variables can improve the effectiveness of the prediction. keywords: credit risk; text data mining; factor analysis; logistic model; text intonation 1. introduction credit is the foundation of financial development, and credit risk is also an uneasy factor enough to destroy the whole financial system. preventing and resolving credit risk is a necessary means to maintain social stability and ensure the healthy development of economy. nowadays, with the rapid development of finance and the increasingly frequent financial exchanges among social subjects, it also brings complex interest relations. once a credit risk occurs in a certain interest link, the associated losses will be immeasurable. therefore, scholars at home and abroad regard the prevention of credit risk as an important research object. credit risk usually refers to the default caused by the reluctance or inability of the borrower, securities issuer, or transaction party to perform the contract [1]. yang lian and shi baofeng [2] introduced the focal loss modified cross entropy loss function into the credit risk evaluation model to predict the risk of several individual samples. the empirical results show that this prediction method can improve the identification ability of difficult samples. wang chongren and han dongmei [3] proposed a bayesian parameter optimization method and xgboost algorithm for personal credit risk assessment of internet credit industry. the empirical results show that this method is superior to traditional prediction models such as support vector machine. luo fangke and chen xiaohong [4] brought the internet financial personal microfinance data of commercial banks haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 177 into the logistic model to screen out the factors that have a significant impact on credit risk. as companies have greater influence than individual borrowers and investors, and the harm caused by credit risk is more destructive, improving the prediction accuracy of corporate credit risk is also a hot issue in the field of risk management. zhang tong and chi guotai [5] empirically analyzed the data of 2169 chinese a-share listed companies from the perspective of credit characteristics, and concluded that the model of feature division has higher discrimination accuracy. compared with credit characteristics, more scholars' research on credit risk is based on the perspective of the optimal combination of credit risk indicators. zhou ying and su xiaoting [6] found that different financial indicators have different effects on the prediction of long-term and short-term default status. li zhe and chi guotai [7] screened 31 indicators with strong ability to distinguish default status from 610 indicators by using the data of listed companies. li meng and wang jin [8] investigated the impact of enterprise internal control level on its debt default risk through traditional financial indicators. the results show that enterprises with high internal control quality tend to have lower debt default risk. previous studies mostly focused on the analysis of financial data. with the progress of computer technology and the rapid development of the internet, more and more unstructured data are applied to the research of financial problems [9]. structured data is field variable data. for example, wu fei et al. [10] collected the keywords related to "digital transformation" in the enterprise annual report through crawler technology to describe the intensity of enterprise digital transformation. li bin et al. [11] identified 29 important risk points in the insurance industry by mining 1682 financial report texts of listed insurance companies in the united states, and analyzed the change trend of important risks in the insurance industry. liang kun and he jun [12] believes that text information effectively alleviates information mismatch and significantly improves the predictability of credit evaluation model. therefore, text big data can also be applied to the field of credit risk. cecchini m, et al. [13] extracts the effective information of the management analysis and discussion module in the annual report, and integrates other financial data to improve the prediction default accuracy of the traditional prediction model. liu yishuang and chen yiyun [14] studied the relationship between text emotion and financial distress through the management tone in the company's annual report. wang xiaoyan et al. [15] constructed a priori word frequency of credit risk indicators by mining the text information in journal papers. the empirical results show that the classification effect of credit risk model is significantly improved after using such a priori word frequency. wang z, et al. [16] and others believe that in addition to the traditional hard information, soft information can also enter the loan decision-making process, and the effect of credit risk assessment is significantly improved after adding semantic indicators. zhang yiwei and gao weihe [17] took the borrower's sms data as the text mining object, analyzed the relationship between the expression of "我" and "我们" and default, and found that the cultural level adjusted the role of these two words in credit risk prediction. wang shuxia et al. [18] identified the characteristics of the lender from the text description and used these characteristics to evaluate the credit risk of the loan. the empirical results show that the text data can effectively replace the traditional financial data, and the combination of structured data and unstructured data can improve the performance of the credit risk evaluation system. it can be seen from the existing literature that the research on the use of text information for credit risk prediction at home and abroad mostly focuses on individual investors, while the research on corporate credit risk mainly focuses on traditional structured data, but there is also a lot of information in many public information such as the company's annual report. obtaining this kind of haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 178 text information will help us reduce the impact caused by information asymmetry, so as to improve the effectiveness of credit risk prediction. in view of this, this paper will combine the real default data, select 25 listed companies with credit risk and 53 listed companies without credit risk from 2018 to 2020 as the total sample, peel the tone from the company's annual report, conduct quantitative analysis, and predict the company's probability of credit risk combined with traditional structured data. the research value of this paper is to expand the traditional credit risk identification indicators, prove that it is meaningful to add intonation to the risk identification model, provide new ideas for predicting credit risk, increase china's identification means of credit risk, and enhance the monitoring of systemic financial risk. 2. research methods and index selection of enterprise credit risk identification 2.1. logistic model and research ideas logistic regression is a common machine learning method, which is mainly used to classify samples and belongs to "generalized linear regression". this model is often used in credit risk research, such as zhang jie and zhang yuansheng [19], bian yuning et al. [20], liang weisen and wen simei [21], etc. the reason is that the logistic model has the excellent characteristics that the value of dependent variable is between 0 ~ 1 and does not need to obey normal distribution [22]. the expression of logistic model is: ln(p/1-p) = β0+∑βi*xi (1) in this paper, the enterprise with credit risk is marked as 1. in formula (1), p represents the probability of credit risk, β0 is a constant term, xi is a dependent variable affecting the predicted credit risk, βi is the influence degree of each dependent variable on credit risk. the research idea of this paper is as follows: firstly, the dimensionality of multiple financial indicators is reduced, and three main components are extracted by factor analysis method. secondly, the logistic model is used to predict credit risk in two steps. in the first step, only three principal components are input to predict credit risk, and in the second step, three principal components and intonation variables are used as input data to predict credit risk. finally, the roc curve is used to compare the credit risk prediction effect of the model before and after adding intonation variables, and the bp neural network model is used to test the robustness of the empirical results. 2.2. data selection since the financial status and annual report of the enterprise in the year of credit risk will not be known to investors in that year, the annual report and financial data of the year before the occurrence of credit risk are the main basis for investors to predict whether the enterprise has credit risk. when selecting the data of defaulting enterprises, the company's annual report and financial data of the year before the occurrence of credit risk are selected as risk identification indicators. when selecting the data of non-defaulting enterprises, the 2019 annual report and financial data are uniformly selected as risk identification indicators. the annual report data are from the public disclosure of listed companies on shanghai stock exchange and shenzhen stock exchange, and the financial data are from the resset financial research database. 2.3. data processing haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 179 financial data processing: referring to the selection method of financial indicators in the construction of credit risk identification system by wang qianhong and zhang min [22] and liu xiangdong and wang weiqing [23], this paper divides the traditional financial indicators into 5 primary indicators and 12 secondary indicators, as shown in table 1. due to the vacancy value in some original financial data, this paper fills it with the average value of this index. table 1. credit risk identification index. primary index secondary index symbol debt service level quick ratio x1 asset liability ratio x2 profit level operating profit margin x3 net profit margin on sales x4 return on assets x5 operational capability turnover rate of fixed assets x6 total asset turnover x7 turnover rate of noncurrent assets x8 cash flow indicators cash content of operating income x9 growth index growth rate of net assets x10 growth rate of total assets x11 growth rate of main business income x12 unstructured indicators annual report intonation tone quantitative processing of text intonation: this paper uses the "dictionary model" to construct the text intonation of the annual report, and refers to hownet dictionary [13] and actual financial terms as the emotion dictionary. the dictionary is divided into positive emotion dictionary and negative emotion dictionary. when quantifying the text intonation, first convert the format of the company's annual report downloaded by shanghai stock exchange and shenzhen stock exchange, convert the pdf file format into txt file format, and then use the jieba word segmentation package in python to segment the annual report [24]. then remove the stop words such as "的" and "了", and make word frequency statistics according to the emotional dictionary. the statistical method is as follows: if there are words in the negative emotion dictionary in the annual report, such as "怀疑", " 难", "疑惑", etc., sum the occurrence times of such words, and use neg to represent the total occurrence times of negative words in an annual report. if the words in the dictionary of positive emotion appear in the annual report, such as "奖励", "引领", "支持", etc., sum the occurrence times of such words, and use pos to represent the total occurrence times of positive words in an annual report. since negative intonation often has a greater impact on decision makers [14], this paper quantifies the text intonation with formula (2), in which the meanings of neg and pos have been introduced above. tone indicates the text intonation, that is, the larger tone, the stronger the negative emotion revealed in the text, otherwise it indicates that the text intonation is more positive. in order to facilitate readers to intuitively understand the positive and negative words in the annual report, this paper generates word clouds from the high-frequency words in the two types of words, as shown in figure 1 and figure 2. tone = neg/(pos+neg) (2) haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 180 figure 1. positive word cloud. figure 2. negative word cloud. 3. empirical analysis 3.1. descriptive statistics and inter group difference test this paper uses spss 21 software to carry out descriptive statistics and mean "independent sample t-test" on the data of credit risk group and non-credit risk group, and observe the characteristics of the two groups of data and whether there is significant difference [24]. the enterprise with credit risk is marked as 1 and the enterprise without credit risk is marked as 0. the descriptive statistics of main variables and the results of "independent sample t-test" are shown in table 2. it can be seen from table 2 that the p value of four variables x6 (turnover rate of fixed assets), x8 (turnover rate of noncurrent assets), x9 (cash content of operating income) and x10 (growth rate of net assets) is greater than 0.05, that is, the difference of these four indicators is not significant and cannot better reflect the difference between different types of samples. the other 9 variables including tone passed the "independent sample t-test", which proved that the remaining 9 variables could significantly reflect the differences between groups. in addition, through the analysis of descriptive statistics, the average, maximum and minimum values of tone of enterprises with credit risk are significantly higher than those without credit risk, which indicates that negative emotions are widespread in the annual report of enterprises one year before credit risk. haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 181 table 2. descriptive statistics of main variables and t-test results of independent samples. explanatory variable enterprise defaults minimum maximum mean standard deviation p value tone 0 0.02263 0.04906 0.03484 0.00662 0.031** 1 0.02733 0.06407 0.03873 0.00857 x1 0 0.27020 4.82480 1.38805 1.17206 0.007*** 1 0.08130 3.30150 0.68852 0.64901 x2 0 12.68890 88.53180 47.17486 21.25192 0.000*** 1 30.96460 175.83540 74.88301 26.74978 x3 0 -18.53640 36.24070 8.94938 10.60676 0.001*** 1 -440.15130 52.32540 -35.86637 92.32086 x4 0 -15.05380 32.30110 7.52735 8.59006 0.001*** 1 -502.99920 40.38250 -43.95637 108.17100 x5 0 -3.67370 16.37470 5.30242 4.43122 0.000*** 1 -109.20290 12.94880 -8.63132 26.38773 x6 0 1.16790 102.52400 8.30737 14.71136 0.762 1 0.34160 94.96940 9.50070 19.00566 x7 0 0.06230 1.22330 0.60662 0.30374 0.006*** 1 0.05700 1.30740 0.39472 0.32921 x8 0 0.09130 4.86020 1.66720 1.09202 0.375 1 0.11960 7.39710 1.37499 1.78710 x9 0 30.15580 132.85340 99.72378 18.74031 0.110 1 65.39690 186.61100 108.65078 29.70843 x10 0 -11.74240 26.17940 5.85995 8.03406 0.093* 1 -136.35060 257.63410 -10.39964 68.99186 x11 0 -15.97900 34.99220 6.18474 8.52548 0.009*** 1 -75.05330 46.43540 -3.97918 24.77621 x12 0 -59.68250 156.73350 9.72063 26.17668 0.022** 1 -68.08130 63.29080 -6.46469 33.30012 *** indicates p<0.01; **p<0.05; *p<0.1. 3.2. factor analysis based on the theoretical and practical impact analysis, the correlation test is conducted for the variables that pass the "independent sample t-test". it can be obtained from table 3 that tone has no significant correlation with the financial variables, but the correlation between the financial variables is relatively significant, which indicates that there may be some same information between the variables and can be explained to each other. if all variables are input into logistic model to predict credit risk, it may lead to wrong conclusions. therefore, this paper makes factor analysis on the other 8 variables except. this paper uses spss 21 software to conduct factor analysis on 8 financial variables. firstly, the data are processed by z-score standard method to eliminate the influence of sample data dimension [25]. it can be seen from table 4 that the kmo test value is 0.549, which is greater than the standard haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 182 value of 0.5 and the p value is 0. again, the eight financial variables contain more similar information and are suitable for factor analysis. table 3. correlation coefficient of each variable. tone x1 x2 x3 x4 x5 x7 x11 x12 tone 1.000 x1 -0.070 1.000 x2 0.066 -0.633*** 1.000 x3 -0.107 0.154* -0.406*** 1.000 x4 -0.047 0.079 -0.376*** 0.971*** 1.000 x5 -0.062 0.115 -0.637*** 0.480*** 0.537*** 1.000 x7 0.140 0.103 -0.154* 0.173* 0.184* -0.023 1.000 x11 -0.070 0.056 -0.385*** 0.336*** 0.320** 0.677*** -0.098 1.000 x12 -0.113 0.030 -0.166*** 0.412*** 0.377*** 0.282** 0.166* 0.366*** 1.000 *** indicates p<0.01; **p<0.05; *p<0.1. table 4. kmo and bartlett test. kaiser-meyer-olkin 0.549 bartlett's sphericity test χ2 436.809 df. 28 p value 0.000 table 5. explains the total variance. ingredients initial eigenvalue extract sum of squares load total variance % accumulate% total variance % accumulate% 1 3.404 42.550 42.550 3.404 42.550 42.550 2 1.366 17.075 59.626 1.366 17.075 59.626 3 1.242 15.527 75.153 1.242 15.527 75.153 4 0.833 10.413 85.566 5 0.670 8.381 93.947 6 0.326 4.080 98.027 7 0.140 1.750 99.778 8 0.018 0.222 100.000 table 6. composition matrix. variable ingredients 1 2 3 x1 0.340 -0.834 0.149 x2 -0.725 0.599 0.056 x3 0.834 0.267 0.282 x4 0.825 0.316 0.267 x5 0.803 -0.013 -0.394 x7 0.187 -0.072 0.773 x11 0.650 0.110 -0.548 x12 0.533 0.352 0.118 it can be seen from table 5 that three principal components with eigenvalues greater than 1 are extracted from factor analysis, and the three components contain 75.153% of the total variables. the original financial variable data can be reduced by nearly two-thirds through factor analysis, indicating that the result of factor analysis is good. let the extracted three components be f1, f2 and f3 respectively (see table 6). the scores of each variable in f1, f2 and f3 are shown in table 7. the following expressions are listed according to the scores. haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 183 f1 = 0.1x1-0.213x2+0.245x3+0.242x4+0.236x5+0.055x7+0.191x11+0.157x12 (3) f2 = -0.610x1+0.439x2+0.196x3+0.231x4-0.01x5-0.052x7+0.081x11+0.257x12 (4) f3 = 0.12x1+0.045x2+0.227x3+0.215x4-0.317x5+0.622x7-0.441x11+0.095x12 (5) table 7. component score coefficient matrix. variable ingredients 1 2 3 x1 0.100 -0.610 0.120 x2 -0.213 0.439 0.045 x3 0.245 0.196 0.227 x4 0.242 0.231 0.215 x5 0.236 -0.010 -0.317 x7 0.055 -0.052 0.622 x11 0.191 0.081 -0.441 x12 0.157 0.257 0.095 3.3. logistic regression when using logistic regression for the first time, only f1, f2 and f3 are used as input variables. the results are shown in table 8. according to the prediction results of the model, equation (6) can be obtained. logistic (p) = -2.46f1+0.638f2-0.709f3-0.995 (6) table 8. logistic regression without tone variable. explanatory variable coefficient standard error wald df f1 -2.460*** 0.683 12.963 1 f2 0.638 0.533 1.434 1 f3 -0.709 0.508 1.947 1 constant -0.995*** 0.373 7.123 1 *** indicates p<0.01; **p<0.05; *p<0.1. table 9. logistic regression with tone variables. explanatory variable coefficient standard error wald df tone 0.774** 0.343 5.090 1 f1 -2.628*** 0.705 13.909 1 f2 0.693 0.584 1.412 1 f3 -1.015* 0.578 3.080 1 constant -1.108*** 0.396 7.832 1 *** indicates p<0.01; **p<0.05; *p<0.1 according to the logistic regression results including principal components f1, f2 and f3, only f1 is significant at the level of 1%, and the pre f1 coefficient is -2.460, which is negatively correlated haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 184 with the probability of default. according to formula (3), the variables with higher scores in f1 are x2 (asset liability ratio), x3 (operating profit margin), x4 (net profit margin on sales) and x5 (return on assets), of which x3, x4 and x5 are variables reflecting profitability, and the coefficient of such variables in f1 is positive. therefore, it can be inferred that the profitability of a company is the main factor affecting its default. when the profitability of the company is good, credit risk is not easy to occur. the worse the profitability, the higher the probability of credit risk. when using logistic regression again, input the three components f1, f2 and f3 together with tone variables into the logistic model to predict credit risk. according to table 9, the logistic expression (7) with tone can be obtained, where p represents the probability of occurrence of credit risk. logistic (p) = 0.774tone-2.628f1+0.693f2-1.015f3-1.108 (7) according to the logistic regression results with tone variable, the pre tone coefficient is 0.774, which is significant at the level of 5%, indicating that tone is significantly positively correlated with the probability of credit risk. according to equation (2), the larger the value of tone, the more negative the tone in the annual report, that is, the more pessimistic the tone in the annual report of the enterprise in the previous year, the more likely the company is to have credit risk, on the contrary, it is less prone to credit risk. it is worth mentioning that after the tone variable is added, the f3 component has changed from having no significant impact on the credit risk prediction in the first regression to being significant at the 10% level. the variables with higher scores in the f3 component are x7 (total asset turnover rate) and x11 (total asset growth rate). because this is not the focus of this paper, this phenomenon has not been analyzed in detail. 3.4. roc curve comparison when most scholars use logistic model to predict credit risk, they usually use out of sample resampling method, take the occurrence probability of credit risk of 0.5 as the critical value of risk occurrence, and judge the accuracy of the model to predict credit risk [22, 23, 25], but few articles discuss the scientificity of the critical value. therefore, this paper uses roc curve to study the effectiveness of credit risk prediction before and after adding tone to logistic model. the ordinate of roc curve represents sensitivity, and the higher the index, the higher the diagnostic accuracy; the abscissa represents 1-specificity. the lower the index, the lower the misjudgment rate. therefore, in general, the closer the point to the upper left corner of the coordinate, the better the diagnostic effect, that is, the larger the area at the lower right side of the roc curve, the better the credit risk prediction effect. table 10. area under roc curve test result variable tone prediction probability non-tone prediction probability area under curve 0.892 0.855 standard error 0.036 0.049 significance 0.000 0.000 asymptotic 95% confidence interval lower limit 0.821 0.760 upper limit 0.963 0.950 haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 185 figure 3 shows the comparison of roc curves of logistic model for predicting the occurrence probability of credit risk before and after the addition of tone variables. the blue curve represents the prediction probability curve of credit risk with tone, and the green curve represents the prediction probability curve of credit risk without tone. it can be clearly seen that the blue roc curve is closer to the upper left of the coordinate than the green roc curve. according to table 10, the area under the roc curve with tone variable prediction results is 0.892, which is greater than the area under the roc curve without tone prediction results by 0.855. both results show that adding intonation to predict enterprise credit risk can improve the effectiveness of credit risk identification. figure 3. roc curve. 4. robustness test this paper uses bp neural network model to test the robustness of the empirical results. firstly, it forecasts the credit risk of traditional financial variables, sets the maximum number of iterations for 10000 times, and trains the total samples with 30% test set and 70% training set. through multiple training comparisons, the model is optimal when there are 12 neuron nodes, and the results are shown in table 11. table 11. identification results of 12 nodes without tone samples in bp model. predicted correct number 48 prediction errors number 30 correct rate 61.5% error rate 38.5% table 12. identification results of tone samples at 13 nodes in bp model. predicted correct number 63 prediction errors number 15 correct rate 80.1% error rate 19.9% then, the traditional financial variables are combined with tone to predict the credit risk. the maximum number of iterations, the total number of samples, and the distribution ratio of training set haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 186 and test set remain unchanged. through comparison, the model is optimal when there are 13 neurons, and the results are shown in table 12. from the comparison of the two output results, the number of correct predictions increased significantly after adding intonation variables, and the prediction accuracy increased from 61.5% to 80.1%, an increase of 18.6%, indicating that adding intonation variables can improve the accuracy of model prediction, which is consistent with the empirical results. 5. conclusions taking 25 listed companies with credit risk and 53 listed companies without credit risk from 2018 to 2020 as the research object, this paper uses text data mining method to capture and quantify the intonation, and uses factor analysis method to extract three principal components from the traditional financial data. finally, we compare the impact of logistic model on the accuracy of credit risk prediction before and after adding intonation variables, and draw the following conclusions. first, when using traditional financial data to predict credit risk, profitability has a great impact on credit risk prediction. the stronger the profitability, the less prone to credit risk. second, the tone of the annual report of enterprises with credit risk in the previous year is more pessimistic than that of enterprises without credit risk. investors can observe the tone of the company's annual report to reduce the impact of information asymmetry. third, according to the empirical results, in the logistic regression, the probability of credit risk is significantly positively correlated with the pessimistic degree of the quantified text tone at the level of 5%, that is, the more negative the tone of the enterprise in the previous year, the greater the probability of credit risk in that year. this result also shows that the tone of the company's annual report contains information related to credit risk, which can solve the problem of information asymmetry between investors and company subjects to a certain extent. fourth, the roc curve is used to test the prediction results of the logistic model twice. the results show that compared with the logistic model which only uses the traditional financial data as the input, the effectiveness of the model prediction is improved after adding the text intonation index. it also shows that although enterprises can beautify financial data and increase investor confidence, the negative emotions revealed in the annual report are widespread. by mining the text information of the annual report, we can expand the credit risk identification indicators and improve the effectiveness of credit risk identification. this paper separates the intonation from the company's annual report, quantifies it, and supplements the traditional credit risk identification system based on structured data. according to the research conclusion of this paper, the most of investors, commercial banks and other financial institutions should strengthen the acquisition of text information when predicting enterprise credit risk, build a risk prediction system from multiple dimensions, improve the efficiency of credit risk identification and reduce the loss caused by information asymmetry. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003 and guizhou university of finance and economics is in urgent need of the support of the special subject (2020zjxk20). conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 187 references [1] chen yanli, jiang qi. business environment, real earnings management and credit risk identification [j]. journal of shanxi university of finance and economics, 2021, 43(09): 98-110. [2] yang lian, shi baofeng. credit risk evaluation model and empirical evidence based on focal loss modified cross entropy loss function [j/ol]. chinese management science: 1-12 [2021-10-10]. https://doi.org/10.16381/j.cnki.issn1003-207x.2020.2188. [3] wang chongren, han dongmei. internet credit personal credit evaluation based on hyper-parameter optimization and integrated learning [j]. statistics and decision-making, 2019, 35(01): 87-91. [4] lou fangke, chen xiaohong. credit risk assessment and application of individual microfinance based on logistic regression model [j]. financial theory and practice, 2017, 38(01): 30-35. [5] zhang tong, chi guotai. default discrimination model based on optimal credit feature combination a case study of chinese a-share listed companies [j]. system engineering theory and practice, 2020, 40(10): 2546-2562. [6] zhou ying, su xiaoting. enterprise credit risk prediction based on optimal index combination [j]. journal of system management, 2021, 30(05): 817-838. [7] li zhe, chi guotai. research on credit risk of listed companies based on maximum index discrimination and optimal relative membership [j]. china management science, 2021, 29(04): 1-15. [8] li meng, wang jin. internal control quality and corporate debt default risk [j]. international finance research, 2020(08): 77-86. [9] shen yan, chen yun, huang zhuo. the application of text big data analysis in economics and finance: a literature review [j]. economics (quarterly), 2019, 18(04): 1153-1186. [10] wu fei, hu huizhi, lin huiyan, ren xiaoyi. digital transformation of enterprises and capital market performance empirical evidence from stock liquidity [j]. managing the world, 2021, 37(07): 130-144+10. [11] li bin, wang yinghui, zhu xiaoqian, li jianping. identification and evolution analysis of important risk points in insurance industry based on text risk information disclosed in financial report [j/ol]. system engineering theory and practice: 1-15 [2021-10-10]. http://kns.cnki.net/kcms/detail/11.2267.n.20210528.0838.002.html. [12] liang kun, he jun. analyzing credit risk among chinese p2p-lending businesses by integrating textrelated soft information [j]. electronic commerce research and applications, 40. https://doi.org/10.1016/j.elerap.2020.100947. [13] cecchini m, aytug h, koehler g j, et al. making words work: using financial text as a predictor of financial events [j]. decision support systems, 2010, 50(1): 164-175. [14] liu yishuang, chen yiyun. management tone and credit risk early warning of listed companies based on content analysis of annual report [j]. financial economics research, 2018, 33(04): 46-54. [15] wang xiaoyan, zhang zhongyan, ma shuangge. credit risk assessment model based on text prior information [j]. chinese management science, 2021, 29(05): 34-44. [16] wang z, jiang c, zhao h, et al. mining semantic soft factors for credit risk evaluation in peer-to-peer lending [j]. journal of management information systems, 2020, 37(1): 282-308. [17] zhang yiwei, gao weihe. self-construction, cultural differences and credit risk empirical evidence from internet finance [j]. financial studies, 2020, 46(01): 34-48. [18] wang shuxia; qi yuwei; fu bin; liu hongzhi. credit risk evaluation based on text analysis [j]. international journal of cognitive informatics and natural intelligence (ijcini), 2016, 10(1): 1-11. [19] zhang jie, zhang yuansheng. research on the measurement of liquidity risk of p2p lending platform based on logistic taking shangjinfu as an example [j]. friends of accounting, 2019(21): 124-127. [20] bian yuning, lu likun, li yeli, zeng qingtao, sun yanxiong. implementation of financial venture capital scoring card model based on logical regression [j]. computer science, 2020, 47(s2): 116-118. [21] liang weisen, wen simei. study on loan default risk assessment of small and medium-sized agricultural enterprises based on data of agriculture, forestry, animal husbandry and fishery enterprises in 'new third board' [j]. rural economy, 2019(11): 93-100. [22] wang qianhong, zhang min. empirical study on credit default risk identification of smes in china [j]. shanghai economy, 2017(01): 91-100. [23] liu xiangdong, wang weiqing. multi-model comparative study on credit risk identification of commercial banks in china [j]. economic latitude and longitude, 2015,32(06): 132-137. haonan zhang, hongmei zhang and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 176-188 doi: https://doi.org/10.54560/jracr.v11i4.311 188 [24] zhang shuhui, zhou meiqiong, wu xueqin. annual report text risk information disclosure and stock price synchronicity [j]. modern finance and economics (journal of tianjin university of finance and economics), 2021, 41(02): 62-78. [25] zhang jingui, hou yu. empirical analysis on credit risk of smes based on logit model [j]. friends of accounting, 2014(30): 40-45. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response vol. 9(3); october (2019), p. 155 doi: https://doi.org/10.2991/jracr.k.191024.007; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr correspondence editors-in-chief: prof. chongfu huang academy of disaster reduction and emergency management, beijing normal university, no. 19 xinjiekouwai street, beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina, regina, sk s4s 0a2, canada email: gordon.huang@uregina.ca © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). introduction this issue contains six papers. there are six contributions in english. the papers can be divided into six topics: risk assessment related to ecological environment, risk assessment related to new technology, political risk assessment, financial risk analysis, emergency response and methods on risk analysis. there is one paper in political risk assessment. the paper “re-examining political risk assessments in volatile regions” by ghaidaa hetou, re-examines political risk analysis and explains how understanding the topology and nature of political risk in emerging and developing markets is a crucial advancement in developing political risk analysis for the private sector and government agencies. particular focus is given to developing political risk characterization as a risk analysis category. to bridge the conceptual gap between risk assessment and risk management, this paper proposes the concept of complex adaptive systems as the backdrop for emerging political risk scenarios. there is one paper in methods on risk analysis. the paper “a critique of pandemic catastrophe modeling” by daniel j. rozell, discusses some general limitations of catastrophe modeling in the context of pandemics—such as the failure to distinguish natural variability from incertitude and the difficulty of ensuring a representative model—along with recommendations for minimizing surprises. there is one paper in emergency response. the paper “a game theory approach for multi-agent system resources allocation against outside threats” by cheng-kuang wu et al., proposes an integrated model for the deployment of multiagent resources for resisting outside threats. the proposed two-stage model applies the divide-and-conquer strategy to solve the resources allocation problem. first, the interactive actions between an external attack and a response agent are modeled as a non-cooperative game, after which the external threat value is derived from the nash equilibrium. second, the threat values of all response agents are utilized to compute each agent’s shapley value. then, an acceptable resource allocation of agents based on their expected marginal contribution creates a minimum set of resource deployment costs. the experimental results show that our approach is feasible as a means to mobilize search and rescue resources from a non-affected district and to improve relief efforts against earthquake damage. the shapley value allocation approach proposed in this study; the percentage of resources allocation of districts is closer to death rate of each district than the proportional division of resources. there is one paper in risk assessment related to new technology. the paper “evaluation on technology innovation efficiency of big data enterprises based on dea” by xinpu wang et al., uses dea-bcc model to evaluate the technological innovation efficiency of 21 big data enterprises. it is found that the technological innovation efficiency of big data enterprises is better. the average values of technological efficiency, pure technological efficiency and scale efficiency are 0.587, 0.772, and 0.750, respectively. the technological efficiency of big data enterprises is lower, and the individual differences are larger. at the same time, the research shows that there are 15 big data enterprises need to further expand the scale, and then improve production efficiency. there is one paper in risk assessment related to ecological environment. the paper “dynamic comprehensive evaluation of ecological environment of 12 provinces and cities in western china” by hongmei zhang et al., selects data from 12 provinces and cities in the western part of china for 6 years, adopts multiple evaluation methods to conduct a dynamic comprehensive assessment of the western ecological environment, and provides some corresponding countermeasures for the coordinated development of the environment and economy in the western ecologically fragile region. there is one paper in risk assessment related to ecological environment. the paper “exploration of risk oriented fund performance audit—take rural endowment insurance as an example” by ronggang zhang et al., analyzes the current situation and causes of the performance audit of rural pension fund based on the cost-benefit principle and the public fiduciary economic responsibility theory and finds that the audit demand degree is not enough. this paper proposes to integrate risk orientation into performance audit and adopt different audit methods and evaluation systems to improve audit efficiency and reduce costs by dividing risk levels and optimizing performance audit. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis—china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. https://doi.org/10.2991/jracr.k.191024.007 https://www.atlantis-press.com/journals/jracr mailto:hchongfu%40126.com?subject= mailto:gordon.huang%40uregina.ca?subject= http://creativecommons.org/licenses/by-nc/4.0/ microsoft word volume 12, issue 1-4 journal of risk analysis and crisis response, 2022, 12(1), 45-54 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i1.321 45 review a systematic literature review on the influence mechanism of digital finance on high quality economic development mu zhang 1,* and qi luo 1,2 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: september 9, 2021; accepted: march 28, 2022; published: april 15, 2022 abstract: to further promote the theoretical research that digital finance effectively supports the high-quality development of the real economy, this paper reviews the current situation of domestic and international research on the connotation of digital finance, the connotation of high-quality development of the real economy, and the influence mechanism of digital finance on high-quality economic development. at present, the research on the connotation of digital finance and the connotation of high-quality development of the real economy at home and abroad has been relatively mature. domestic and foreign research on the influence mechanism of digital finance on high-quality economic development mainly includes direct influence mechanism featuring capital allocation effect and inclusive effect, and indirect influence mechanism featuring innovative development, coordinated development, green development, open development, shared development, and industrial structure transformation and upgrading as channels. future research directions include: deepening the research on the indirect influence mechanism of digital finance on the high-quality economic development; deepening the research on the influence mechanism of digital finance on the high-quality development of the real economy. keywords: digital finance; high quality economic development; real economy; influence mechanism; literature review 1. introduction digital finance is the product of combining financial services with digital technologies including the internet (mobile internet and internet of things), big data, distributed technologies (cloud computing and blockchain), artificial intelligence, and information security (biometrics and encryption), and so on. the ability of digital finance to promote the achievement of high-quality economic development has begun to attract the attention of some scholars [1]. xue ying and hu jian [2] discussed the theoretical logic, practical basis and realistic path of fintech to promote high-quality development, and believed that fintech has resource (including but not limited to capital) allocation effect and innovation effect, which helps to boost the financial industry's ability to de-emphasize the virtual to the real and enhance the ability of financial services to the real economy, which provides objective and realistic conditions to promote high-quality and sustainable economic development. fintech can directly provide diversified financing channels for enterprises and improve capital mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 46 utilization efficiency through the savings-investment conversion mechanism, help realize investment capital accumulation, play the role of "shadow banking" through non-bank credit intermediaries, help relevant enterprises integrate into the global value chain, and expand the economic opening pattern [3]. as a combination of digital technology and finance, digital finance uses digital technology to expand the scope of financial services and improve the accessibility of financial services for many small and disadvantaged economic agents, while its digital technology also makes financial services more targeted and can provide more efficient financing services for various economic agents. this "inclusive" and "precise" characteristic makes financial services for the real economy both efficient and fair. therefore, it can help achieve high-quality development [4]. whether high-quality development can be achieved depends on whether the economic growth model can be truly shifted from factor-driven to innovation-driven, and the transformation of economic driving forces will inevitably require the transformation of the financial services model, which cannot be achieved without the technical transformation of traditional finance by digital technology [5]. this paper will review the current situation of domestic and international research on the connotation of digital finance, the connotation of high-quality development of the real economy, and the influence mechanism of digital finance on high-quality economic development, with a view to further promoting the theoretical research on the effective support of high-quality development of the real economy by digital finance. the remainder of this paper is structured as follows: part 2 introduces the connotation of digital finance; part 3 introduces the connotation of high-quality development of the real economy; part 4 introduces the influence mechanism of digital finance on high-quality economic development; and part 5 is a brief review. 2. connotation of digital finance with the continuous application of modern cutting-edge science and technology in the financial field, the understanding of the connotation of digital finance is deepened in both academia and industry. for example, some scholars believe that digital finance is a new type of financial industry based on modern information technologies such as the internet and big data, and it is a new thing arising from the process of infiltration of modern information enterprises into finance [6]. while some scholars believe that digital finance, or fintech, is a product of the combination of finance and technology [7]. a more comprehensive understanding is that digital finance generally refers to the use of digital technology by traditional financial institutions and internet companies to realize financing, payment, investment, and other new financial business models [8]. this concept is consistent with the definition of "internet finance" (a new financial business model in which traditional financial institutions and internet companies use internet technology as well as information and communication technology to realize capital financing, payment, investment, and information intermediary services) by the people's bank of china with the remaining nine ministries and commissions, as well as the definition of "fintech" (promoting financial innovation through technological means, and forming business models, technology applications, business processes and innovative products that have a significant impact on financial markets, institutions and financial services) by the financial stability board (fsb). however, intuitively, internet finance is seen more as internet companies engaging in financial business, while fintech is more technical in nature. in comparison, the concept of digital finance is more neutral and covers a somewhat broader spectrum. in addition, some scholars have also put forward a similar view that digital finance is the product of mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 47 combining financial services with digital technologies including the internet (mobile internet and internet of things), big data, distributed technologies (cloud computing and blockchain), artificial intelligence, information security (biometrics and encryption), and so on. it is a series of financial innovation spectrum concepts with internet finance and fintech, and providing inclusive and accurate financial services is its core attribute [9, 10]. 3. connotation of high-quality development of the real economy since the 19th nation congress of the communist party of china proposed promoting highquality development in 2017, many scholars have discussed this issue in depth. changes in the main contradictions of chinese society are the stage characteristic of china's economic development, which is also the logical starting point and logical basis for proposing high-quality development. therefore, the goal of high-quality development is to meet the people's growing demand for a better life. to achieve this goal, it is necessary to uphold the development concepts of innovation, coordination, green, openness, and sharing [11]. he lifeng [12] believes that high-quality development is necessarily development that implements the five development concepts, and achieving high-quality development requires unified promotion from six aspects, includes an industrial system led by innovation and coordinated development, a unified and open market system which is competitive and orderly, an income distribution system which reflects efficiency and promotes fairness, an urbanrural regional development system which highlights advantages and coordinates linkages, a green development system which is resource-saving and green-friendly, and an open system which is diversified and balanced. the connotation of high-quality development has multidimensional characteristics, and the choice of paths is also plentiful in nature, with a new dynamic mechanism of true value rationality, which is manifested in the supply side as innovation-led and the demand side as people's aspiration [13]. the research group of the economic research institute of the national development and reform commission [14] focused on the supply system to understand the connotation of high-quality development, arguing that high-level quality, efficiency and stability of the supply system are the inevitable requirement for high-quality development. despite the discussion perspectives on the connotation of high-quality development are difficult, these studies generally have a relatively consistent economic meaning, that is, high-quality development covers a broader dimension and richer content than pure economic growth, and its multidimensional characteristics are unanimously agreed by the researchers. the real economy is one of the organic components of the whole economic system, as opposed to the virtual economy. according to huang chongying [14], the real economy in the broadest sense refers to the primary industry, the secondary industry and the tertiary industry in the national economy minus the financial industry and the real estate industry. high-quality development is the theme of china's economic work in the new era. high-quality development of china's real economy is relative to the high-speed growth in the past, and it is an inevitable requirement for adapting to changes in the main social contradictions, building a moderately prosperous society in all respects and building a great modern socialist country in every dimension. it is also a necessary stage in line with the laws of economic development, and its scientific connotation can be understood from the following four aspects [15, 16]: (1) the theme of high-quality development of the real economy is quality reformation. quality reformation refers to improving the supply quality of the real economy as the main direction, not only to actively align with international advanced standards, but also to impact the mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 48 dominance of international standard formulation, to carry out quality improvement actions, as well as to provide more high-end products and quality services; not merely to carefully cultivate a number of high-quality brand enterprises and products with strong international competitiveness, but also to constantly improve the green low-carbon cycle development of the economic system, and strive to make green development universal form, and the metamorphosis from "high speed" to "high quality" is realized, and made in china and chinese services become synonymous with high quality. (2) the core of high-quality development of the real economy is efficiency reformation. the level of input-output efficiency is crucial to the success or failure of entity enterprises in the market competition. under the background of high costs of entity enterprises and the carrying capacity of resources and environment is near the limit, deepening the reform of financial, energy, transportation and other administrative monopoly areas, enhancing the capacity of financial services to the real economy, reducing the costs of entity enterprises, optimizing the business environment, improving the allocation and utilization efficiency of production factors, effectively stimulating the production potential of factor resources and improving input-output efficiency are the keys to enhancing the attractiveness and competitiveness of the real economy and promoting the highquality development of the real economy. (3) the basis of high-quality development of the real economy is the power reformation. promoting the dynamic reformation of the real economy growth, and realizing the transformation from factor-driven to innovation-driven is the meaning of building the modern economic system with high-quality development. we should take innovation as the first driving force to lead the development of the real economy, increase the contribution rate of scientific and technological innovation to economic growth, as well as cultivate and strengthen new kinetic energy. we should take talent as the first resource to support the development of the real economy, increase the protection of intellectual property rights, stimulate and protect the entrepreneurial spirit, give full play to the role of entrepreneurs, at the same time accelerate the modernization of education, build a knowledge-based, skilled and innovative workforce, focus on mobilizing and protecting the enthusiasm of workers at all levels, and continuously inject new and strong impetus for the real economy to improve quality and increase efficiency. (4) the purpose of high-quality development of the real economy is to serve the people. the high-quality development of the real economy is an inevitable trend in the national economic development, and it is also a strategic move to adapt to the upgrading trend of resident’s consumption. the main contradiction in chinese society in the new era is the contradiction between the people's growing needs for a better life while unbalanced and inadequate development. the people expect clean water, fresh air, healthy food and high-quality services. china's high-quality development of the real economy is the point of constantly meeting the growing needs of the people for a better life, so that the people have a greater sense of gain, happiness and security. 4. the influence mechanism of digital finance on high-quality economic development 4.1. digital finance helps to achieve high-quality economic development it has been generally recognized that high-quality development has multi-dimensional characteristics, and digital finance is also a multi-dimensional concept, including multiple dimensions such as the breadth of financial services coverage, the depth of use and the degree of mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 49 digitization, with the core attribute is the unification of inclusive services and precise services. this core attribute of digital finance can consider the issues of efficiency and equity in economic and social development, which not only contributes to the healthy operation of the economic system, but also helps the whole social enterprise to move forward and develop with high quality. on the one hand, digital finance increases the value of relevant information through data collection and mining, reduces information asymmetry in the process of financial services, and reduces moral hazard and adverse selection in the process of investment and financing. at the same time, the lowering of the market threshold has widened the channels of capital sources, making the cost of capital lower and the transaction more efficient. the matching of risks and benefits based on massive data information mining and artificial intelligence analysis makes risk management in the capital allocation process more efficient. this capital allocation effect provides more efficient financial support for economic development, thus promoting the economy to achieve high-quality development [1]. on the other hand, digital finance has broken the long-standing "28 laws" of financial services, that is, the financial long-tail customers who have been discriminated against by capital for a long time can also obtain the required financial support through digital finance development, thus enabling financial services to benefit disadvantaged groups, including small and micro enterprises, innovative and entrepreneurial subjects, and residents in remote areas. by providing these long-tail groups with financial services such as entrepreneurship, employment and consumption, the problem of unbalanced and insufficient development in china can be solved to a large extent, enabling backward regions to obtain financial support to achieve leapfrog development, enabling areas that are not sufficiently developed to obtain the financial support needed for further development, and forming a more rational industrial system, market system, regional development system, opening-up system, and optimizing the distribution system. to ultimately realize the aspiration of all residents for a better life. the logical starting point of high-quality development is the change of the main contradiction in society, and this inclusive effect of digital finance obviously helps the whole society to achieve high-quality development in all fields and solve the main contradiction in chinese society [10, 17]. 4.2. digital finance can help achieve high-quality economic development by promoting innovation development innovation as the core driving force to lead high-quality development has been widely recognized, but for market entities, innovative r&d activities themselves have the characteristics of high investment, high risk, long cycle and irreversibility which highlights the financing constraints faced by enterprises to a certain extent, digital finance provides convenient financing channels for enterprise innovation, research, and development, which is conducive to the development of their innovation activities. on the one hand, digital finance uses digital technology to alleviate the information asymmetry between the supply and demand sides in the investment and financing process, reducing information search costs and transaction costs, and making risk and benefits more compatible [18]. more information from logistics, information flow, and capital flow is incorporated into the risk and pricing assessment system, which reduces financing frictions and increases the probability of successful financing. on the other hand, digital finance also efficiently attracts many small-scale investors to the capital supply system through scene reshaping and model reformation, increasing the level of capital supply, reducing the degree of credit distortion, and lowering the cost mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 50 of corporate financing, all of which provide the possibility of easing corporate financing constraints and thus promoting r&d innovation. in addition, digital finance not only provides financing support for innovation and entrepreneurship activities, but also leads to the continuous emergence of various new business models and service industries due to the innovative ways of providing capital, which also greatly contributes to the development of regional innovation [10]. therefore, digital finance can help achieve high-quality economic development by promoting innovative development. 4.3. digital finance can help achieve high-quality economic development by promoting coordinated development industrial synergy and urban-rural regional linkage are important connotations of high-quality development, and are also an inevitable requirement for cultivating a healthy industrial system and manifesting institutional advantages. china's long-standing uncoordinated industrial development, obvious urban-rural dual structure and huge differences in regional development levels are certainly due to historical and resource endowments, but the systemic defects of the financial system itself are also important reasons. the inefficiency and distortion of the financial system are manifested in the influx of capital from the disadvantaged agriculture-related industries to the secondary and tertiary industries, from the rural areas to the cities and towns, the long-term circulation from the middle and lower ends of the value chain, and even the internal idling of the financial system, which have exacerbated the incongruity in china's development. digital finance has largely improved the financial system's attribute of being "poor but rich", and its inclusive and inclusive attribute can bring capital to disadvantaged industries and disadvantaged regions, and use digital technology to precisely match capital support to the financiers of these industries and regions. at the same time, digital finance has expanded the scope of information search and changed the way of data processing, enabling more effective identification and prevention of risks [19], thus also providing a guarantee for the healthy development of these fields and the financial system itself. the concept of "unbounded finance" brought about by digital technology has further blurred the differences in original financial endowments between urban and rural areas and different regions, putting different regions under the same financial services coverage and promoting the coordinated development of each region [10, 1, 17]. therefore, digital finance can contribute to achieving high-quality economic development by promoting coordinated development. 4.4. digital finance can help achieve high-quality economic development by promoting green development high quality development cannot be achieved without the harmonious development of human beings and nature; therefore, green development is also an important part of high-quality development. although green finance is an economic activity that influences environmental improvement, climate change and resource conservation through the financial system. but its development has been hampered by information asymmetry and difficulty in data sharing over the years. digital finance has green attributes in itself, which can link the way of ordinary people behave with carbon emissions and green transactions, thus promoting green finance and environmental friendliness in a bottom-up manner. in a narrow sense, digital financial activities represented by digital payment, online credit, etc. are themselves environmentally-friendly financial services, while in a broader sense, the core of green finance is to internalize the externalities of environmental problems and to balance environmental behaviors that are not paid or compensated through policy mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 51 guidelines or pricing mechanisms, while digital finance is also a public good with obvious externalities, and its compensation mechanism should and can be aligned with green finance through appropriate arrangements. through the digital financial platform, green financial resources can be guided to the public, environment-friendly enterprises as well as the green production and life style of the industrial economy, can establish an ecosystem of green financial resource allocation and management based on the digital financial platform, and can transmit the informationized environment-related data to the government environmental governance department. finally, a closed loop of environmental governance is created in which financial resources deal with the ecological environment by "rewarding the good and punishing the bad" [10]. therefore, digital finance can contribute to the realization of high-quality economic development by promoting green development. 4.5. digital finance can help achieve high-quality economic development by promoting open development the level of openness to the outside world demonstrates the vitality of economic development and confidence in future development, so open development is the sure way to high-quality development. the technical structure of a country's trade largely influences the quality of its economic growth, and this structure depends not only on comparative advantages but also on possible potential costs [20]. according to this theory, digital finance, on the one hand, through the inclusive and precise financial services it provides, provides broader and lower-cost financing support to the majority of enterprises engaged in export-oriented production, thus helping them to expand production and capture external markets. at the same time, digital finance helps these enterprises to invest more capital in r&d and innovation, thus integrating into the middle and high end of the global value chain. on the other hand, digital finance has contributed to the optimization of china's financial structure and the upgrading of the factor endowment structure, which enhances china's comparative advantage in foreign trade. in addition, the financial service processes, such as payment clearing, transformed by digital technology have brought convenience to cross-border capital in terms of extremely fast arrival, end-to-end payment tracking, and etc., while distributed ledger technology and digital encryption technology enable institutions to process payment in parallel with greater security and safety, which provides great convenience for china to attract foreign investment and domestic capital out of the country [10]. therefore, digital finance can help achieve high-quality economic development by promoting open development. 4.6. digital finance can help achieve high-quality economic development by promoting shared development the goal of high-quality development is to meet the growing needs of people for a better life, and whether the achievements of development can be shared by all people is an important criterion to test the quality of development. the core attribute of digital finance is financial inclusion, which is to achieve inclusive growth shared by all strata through the provision of inclusive financial services, this is coupled with the ultimate goal of high-quality development. as a great innovation to promote the development of inclusive finance in the past decade, digital finance has innovated third-party payment, biometric password identification and other service methods in payment and clearing, brought about network lending, crowdfunding, big data credit and other service methods in lending and financing, has developed internet funds, securities, robo-advisor, quantitative investment and mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 52 other service models in investment and financial management, and has formed big data intelligent risk control, blockchain encryption and other service models in risk management. while digital financial services such as cloud computing platform, digital currency and functional collection have also been spawned. all of these have successfully improved the economic welfare of vast low-income people, young people, women, rural residents, and other people who were originally excluded from traditional finance, and have provided a convenient channel for these people to identify and take advantage of entrepreneurial opportunities, increase personal income, borrow small amounts of money, and achieve credit consumption [10]. therefore, digital finance can help achieve high-quality economic development by promoting shared development. 4.7. digital finance can promote high-quality economic development by promoting the transformation and upgrading of industrial structure digital finance has changed the production and consumption patterns and has promoted highquality economic development from the supply side and the demand side, and there is a transmission mechanism of "digital finance industrial structure transformation and upgrading high-quality economic development" [17]. the optimization and upgrading of industrial structure is an inevitable requirement for china to adjust its development strategy, balance the supply and demand structure, and promote high-quality economic growth [21]. digital finance, as an innovative tool with both technical and financial in nature, can promote the development of high-tech industries by relying on digital kinetic energy, and can also realize the transformation of traditional industries by bringing into play technological benefits and competitive effects, to create an excellent financial environment and jointly promote the transformation and upgrading of the industrial structure. first, from the supply side, digital finance optimizes the industrial investment environment [22] and effectively promotes the development of financial infrastructure. this will prompt the entire financial industry to use internet thinking to build and operate, rely on digital technology to improve the productivity within the industry, and continuously strengthen the "boost effect" of digital finance on the transformation and upgrading of the industrial structure [23]. at the same time, the development of financial infrastructure will also drive the financial industry to gather as well as promote the optimization and upgrading of the industrial structure by relying on the network economy effect and spatial spillover effect [24]. second, from the demand side, the technological benefits brought by digital finance significantly stimulate the release of financial consumption potential. digital finance provides financial products that are safer, more efficient, and more convenient than traditional finance, as well as increases residents' investment choices and wealth accumulation channels, and promotes structural optimization at the demand level. at the same time, the upgrading of financial consumption also greatly alleviates the bottleneck of production factors and overcapacity in traditional industries, and stimulates innovative development momentum in other demand sectors based on competitive pressure, and promotes the improvement and innovation of industrial structure [25]. 5. brief comment at present, the research on the connotation of digital finance and the connotation of high-quality development of real economy at home and abroad has been relatively mature. it is foreseeable that with the continuous deepening of digital finance practice and the practice of high-quality mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 53 development of real economy, the understanding of the connotation of digital finance and the connotation of high-quality development of real economy will also be deepened continuously. at present, domestic and foreign research on the influence mechanism of digital finance on high-quality economic development mainly includes direct influence mechanism featuring capital allocation effect and inclusive effect, and indirect influence mechanism featuring innovation development, coordinated development, green development, open development, shared development and industrial structure transformation and upgrading as channels. however, scholars at home and abroad have conducted rich research on the direct influence mechanism of digital finance on highquality economic development, while the research on the indirect influence mechanism is relatively weak, and the research on the indirect influence mechanism needs to be further deepened. in addition, although domestic and foreign scholars have studied the influence mechanism of digital finance on high-quality economic development, most of them have not clarify whether the "high-quality economic development" what they study is the "high-quality development of the real economy". therefore, the research on the influence mechanism of digital finance on the high-quality development of real economy needs to be further deepened. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] yu, c.y.; wang, x.b.; sun, g.l. digital finance and the quality of economic growth in china: internal mechanism and empirical evidence. inquiry into economic issues, 2020, (07), 1-14. [2] xue, y.; hu, j. fintech promotes high-quality economic development: theoretical logic, practical basis and path choices. reform, 2020, (03), 53-62. [3] li, y.; cheng, b.q. driving force of fintech development for economic growth in china: measurement and mechanism. social sciences in guangdong, 2018, (03), 44-52. doi: https://doi.org/10.3969/j.issn.1000114x.2018.03.006. [4] he, h.q. digital finance: an important driver of china’s high-quality economic development. journal of xi'an university of finance and economics, 2019, 32, (02), 45-51. doi: https://doi.org/10.19331/j.cnki.jxufe.2019.02.007. [5] li, d. support high-quality development with financial innovation (zheshang bank releases "2019 · jingshan report" in hangzhou). china financialyst, 2019, (10), 129-130. doi: https://doi.org/10.19294/j.cnki.cn11-4799/f.2019.10.040. [6] zhang, k. focusing on security supervision of digital finance. the chinese banker, 2013, (11), 114-117. doi: https://doi.org/10.3969/j.issn.1671-1238.2013.11.031. [7] lee, s.h.; lee, d.w. fintech-conversions of finance industry based on ict. journal of the korea convergence society, 2015, 06, (03), 97-102. doi: https://doi.org/10.15207/jkcs.2015.6.3.097. [8] huang, y.p.; huang, z. the development of digital finance in china: present and future. china economic quarterly, 2018, 17, (04), 1489-1502. doi: https://doi.org/10.13821/j.cnki.ceq.2018.03.09. [9] ma, d.g.; teng, l. digital finance, entrepreneurship and inclusive growth. collected essays on finance and economics, 2020, (09), 54-63. doi: https://doi.org/10.13762/j.cnki.cjlc.2020.09.006. [10] teng, l.; ma, d. g. can digital finance help to promote high-quality development? statistical research, 2020, 37, (11), 80-92. doi: https://doi.org/10.19343/j.cnki.11-1302/c.2020.11.007. [11] gao, p.y. comprehending, grasping, and promoting high-quality economic development. economic perspectives, 2019, (08), 3-9. mu zhang and qi luo / journal of risk analysis and crisis response, 2022, 12(1), 45-54 doi: https://doi.org/10.54560/jracr.v12i1.321 54 [12] he, l.f. implementing the new concept of development to lead china's economy to high-quality development. macroeconomic management, 2018, (04), 4-5+14. doi: https://doi.org/10.19709/j.cnki.113199/f.2018.04.003. [13] jin, b. study on the “high-quality development” economics. china industrial economics, 2018, (04), 5-18. [14] research group of institute of economic research. a study on promoting high-quality development of china's economy. macroeconomics, 2019, (02), 5-17+91. doi: https://doi.org/10.16304/j.cnki.113952/f.2019.02.002. [15] huang, c.y. research on the direction and path choice of the high-quality development of china's real economy. journal of fujian normal university (philosophy and social science edition), 2019, (03), 51-61+168. doi: https://doi.org/10.12046/j.issn.1000-5285.2019.03.007. [16] guo, w.; yang, j.; zeng, x.x. scientific guidelines for promoting high-quality entity economic development: studying president xi jin-ping’s important expositions on the real economy. studies on mao zedong and deng xiaoping theories, 2020, (04), 16-25+107. [17] wang, m.; li, z.w.; lv, h. digital finance and economic high-quality growth: analysis of mechanism, effect and heterogeneity. journal of harbin university of commerce, 2021, (03), 18-34. doi: https://doi.org/10.3969/j.issn.1671-7112.2021.03.002. [18] wang, j.y.; zhou, q.; xiao, y. digital finance, financial constraint and enterprise innovation. economic review, 2020, (01), 71-83. doi: https://doi.org/10.19361/j.er.2020.01.05. [19] song, x.l. empirical analysis of digital inclusive finance bridging the urban-rural residents' income gap. finance & economics, 2017, (06), 14-25. [20] hausmann, r.; klinger, b. structural transformation and patterns of comparative advantage in the product space. ssrn electronic journal, 2006, (03), 65-77. doi: http://dx.doi.org/10.2139/ssrn.939646. [21] zhou, s.l.; wang, w.g. scientific-technical innovation, and optimalization upgrade of industrial structure. management world, 2001, (05), 70-78+89-216. doi: https://doi.org/10.19744/j.cnki.111235/f.2001.05.009. [22] ren, b.p.; dou, y.b. the path and policy of the new economy to promote the upgrading of china's industrial structure during the 14th five-year plan. review of economy and management, 2021, 37, (01),1022. doi: https://doi.org/10.13962/j.cnki.37-1486/f.2021.01.002. [23] watanabe, c; naveed, k; tou, y; et al. measuring gdp in the digital economy: increasing dependence on uncaptured gdp. technological forecasting & social change, 2018, 137, 226-240. doi: https://doi.org/10.1016/j.techfore.2018.07.053. [24] yu, b.b. can financial agglomeration promote industrial structure upgrading from perspective of spillovers effect: based on the analysis of dynamic spatial panel model and city data in china. studies of international finance, 2017, (02), 12-23. doi: https://doi.org/10.16475/j.cnki.1006-1029.2017.02.002. [25] cai, h.y.; xu, y.z. does trade openness affect the upgrading of china's industrial structure? the journal of quantitative & technical economics, 2017, 34, (10), 3-22. doi: https://doi.org/10.13653/j.cnki.jqte.2017.10.001. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style introduction for volume 8, issue 4 this issue contains 6 papers. there are 2 contributions written in english and 4 contributions in chinese with english abstracts. the papers can be divided into four topics: health risk, earthquake risk, credit risk and ecological risk. there are two papers in health risk. the paper “research on the occupational noise exposure health risk managem nt standard”by minyan li,et al., proposes the procedures and methods of risk assessment and management of occupational noise hazard in the workplaces as technical guidelines in the new standard based on the current status and technical requirements of occupational noise hazard in china. the second paper “the human health risk assessment based on process simulation and uncertainty analysis” by yue pan, et al., proposes a human health risk assessment method based on the simulation of contaminant transport process and parameter uncertainty analysis. the method combines health risk assessment with groundwater contaminant transport processes and applies the tough2 program to simulate the temporal and spatial distribution of contaminant in groundwater. the key parameter of the transport model is inversed by markov chain monte carlo simulation method. there is two papers in earthquake risk. the first paper“earthquake insurance, catastrophe insurance and earthquake risk & loss: a comparative analysis based on citespace” by cong lang, et al., compared the keyword cluster and institution network of earthquake insurance, other catastrophe insurance and earthquake risk & loss. the keyword cluster showed that earthquake insurance and other catastrophe study relied on non-engineering approach, such as institutions, policy and risk management. the second paper “a study of seismic macroeconomic losses based on monte carlo method— —take tangshan city as an example” by qing wu, et al., uses the monte carlo method to simulate seismic sequences. the model gives the exceeding probability curve of macroeconomic losses. there is one paper in credit risk. the paper “collective risk generalization to credit risk”by reza habibi, uses the collective risk models of actuarial science, to extend the creditrisk+ to the case of random number obligors. first, mathematical methods to compute the distribution of total loss are studied. then, the mathematical results are applied and verified numerically. the insufficiency data in risk management is a big problem. thus, the case of data scarce is studied using a bayesian approach. finally, a concluding remarks section is also given. there is one paper in ecological risk. the paper “characteristic analysis of ecological water use in xilin river basin based on remote sensing” by hongbo yu, et al., takes the xilin river basin as the study area and use remote sensing approach to retrieve ecological water use by using modis images from from apr. to sept. of 2000、2007、20102014, and using auxiliary environmental data from the same time periods. the results showed the spatial distribution of ecological water use was consistent with the land surface condition. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university no.19 xinjiekouwai street beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca 184 ___________________________________________________________________________________________________________ copyright © 2019, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 4 (december 2018) 184 microsoft word volume 12, issue 1-2 journal of risk analysis and crisis response, 2022, 12(1), 25-35 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i1.319 25 article financial distress prediction for digital economy firms: based on pca-logistic dongyang li 1, kai xu 1,*, yun li 1, yu jiang 1, ming tang 1, yangdan lu 1, chun cheng 1, chunxiao wang 1 and guanbing mo 1 1 business school, chengdu university, chengdu (610106), sichuan, china * correspondence: xukai@cdu.edu.cn received: february 16, 2022; accepted: march 30, 2022; published: april 15, 2022 abstract: financial distress prediction is important for risk prevention and control of digital economy firms, as well as going concern guarantee. this paper takes 100 chinese a-share listed digital economy firms from 2017 and 2021 as samples, obtains financial indicators by combining the characteristics of digital economy firms, the first three periods of financial distress are systematically modeled employs logistic regression, while we use the principal component analysis method to deal with the problem of multicollinearity. the results show that the profitability factor has the greatest contribution to the predictive role; the closer to the year in which the financial distress occurred, the higher the prediction accuracy rate. finally, this model achieves 86% prediction accuracy. the successful modelling provides a basis for information users to determine the financial distress of firms accurately and prospectively in the digital economy. keywords: financial distress; digital economy; principal component analysis; logistic regression 1. introduction in the economic globalization and the new economy era, the financial situation and financial distress of digital economy firms are under attention of the society. compared with other industries, the digital economy industry has typical characteristics of high growth, high profitability, focus on innovation, and highly competitive intensity in the field, which also makes the financial situation of digital economy firms dynamic and high-risk. when financial risks are not controlled and accumulate to a certain level to produce qualitative changes, it will cause financial distress. financial distress can be considered as a business failure, bankruptcy crisis [1], debt crisis or credit risk [2]. during a firm's decline, the indicators reflecting the firm's operations, assets, and liabilities often show abnormal changes from those of a normally operating firm, followed by difficulties and eventual bankruptcy. timely predicting of corporate financial distress purposes to reduce investors' losses and help management make strategic adjustments to avoid facing operating crisis. the so-called financial distress prediction, that is based on risk management, according to the combination of theory and methodology, analysis of the macro environment, micro governance, business conditions and financial management faced by firms [3]. pca (principal component analysis) constructs a few representative principal factor indicators from many indicators of the firms, which provide a method to avoid the multicollinearity caused by high correlation among indicators and achieve the purpose of reducing and simplifying variables. the advantages of pcalogistic are, firstly, using the symmetric feature of pca covariance matrix to weaken the sensitivity dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 26 of logistic model to covariance. secondly, replacing most indicators with few indicators to realize the initial data dimensionality reduction of logistic model. thirdly, reducing the workload and errors of manual screening indicators and improving the overall prediction accuracy of the model. in view of this, we aim to contribute to research into the accounting-based distress model, and specifically in two ways. firstly, by adding r&d investment, a core characteristic indicator of digital economy firms. secondly, pca-logistic method is used to make the model more fit for digital economy firms while having high accuracy. therefore, this paper utilizes a combination of pca and logistic to build a financial distress prediction model with a sample of chinese a-share listed digital economy firms from 2017-2021 to improve the accuracy of prediction. the innovations of this paper are, firstly, the combination of pca method and logistic model to build a financial distress prediction model compared to the single prediction model used in the past. secondly, digital economy firms are emerging industries and growing rapidly in china, previous studies lacked a financial predicting methodology for the digital economy industry, and this paper complements this. the remainder of this paper is organized as follows. section 2 gives overview of previous literature in the field of distress prediction. section 3 describes the research design for building the distress prediction model. section 4 presents the empirical analysis, which contains pca and logistic models. section 5 concludes. 2. literature review the study of corporate financial distress has been carried out since the early 20th century, and the first research scholar fitzpatrick (1932) defined financial distress as the inability of a firm to pay its debts as they fall due [4]. foster (1985) considered the existence of financial distress when a firm needs to face a reorganization situation [5]. laitinen (1991) in his study classifies firms with different levels of distress and thus identifies the firms that are in financial distress [6]. following behind, chinese researchers have also conducted in-depth studies on financial distress. wu and huang (1987) argued that a firm is found to be in financial distress when it cannot pay its debts as they fall due and loses its legal personality [7]. wu (2011) defines financial distress as a situation in which an enterprise's ability to pay is insufficient after weighing the force of financial distress occurrence against its resisted force in the context of embedded stakeholder behavior [8]. based on previous researchers' definitions, this paper defines a financially distressed firm as a firm that has been given special treatment and is at risk of delisting due to consecutive years of losses. the construction of financial distress prediction models is the key to ensure the efficiency of forecasting, which has been fruitfully studied in the academic. fitzpatrick (1932) pioneered the univariate analysis of financial indicators of listed companies; subsequently, altman (1968) used multivariate linear discriminant methods to build multivariate financial prediction model [9]. since multivariate discriminant models require strict distribution and covariance of independent variables, which limit the selection of samples, in response maitin (1977) explored and used logistic model to construct a bank warning system [10]. based on this, ohlson (1980) introduced four non-financial variables and used logistic regression for predictive analysis [11], and since then more researchers have used logistic model for financial distress prediction. chinese researcher jiang (2001) proved that logistic model has high predictive accuracy through model testing [12]. li et al. (2011) concluded that logistic model is more suitable for analyzing the actual situation of financial distress compared to linear models [13]. li (2018) predicted financial risk with high accuracy through logistic regression dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 27 [14]. based on the comparison of previous studies, it is found that the use of logistic model is more representative, and given that machine learning is still developing, this paper uses the widely used logistic regression to construct a financial distress prediction model. 3. data and methodology 3.1. sample selection according to the research of altman et al (2017) and sun et al (2014), firms with two consecutive years of losses, insolvency [15], and audit negativity are defined as firms in financial distress [16]. in accordance with the rules of china shanghai and shenzhen a-share markets, this paper uses firms that have been st or *st for the above reasons as a sample of firms in financial distress, and firms that have not been st or *st as a sample of healthy firms. figure 1. industry distribution of digital economy firms. previous researchers and institutions have not been consistent in defining the digital economy, but their definitions have some common points, which emphasize digital technologies, networks, industrial convergence and their impact on the economy [17]. hence, based on the characteristics of digital economy, this paper defines digital economy firms as relying on digital platforms and applications, etc., providing products or services with digital technology innovation as the core, at the same time promoting the industrial integration of digital technology and real economy. according to the "statistical classification of digital economy and its core industries" released by china statistics bureau in 2021 to determine the classification of digital economy firms, this paper obtains a-share listed firms in china comes from between 2017 and 2021 as the sample, the year in which the financial distress occurred is taken as period t. referring to the method of wang et al. (2017), data from the previous three years, i.e. period t-1, period t-2 and period t-3 are selected as forecast data according to the ratio of 1:1 between financially distress firms and financially healthy firms [18]. a total of 50 firms in financial distress and 50 healthy firms are used as controls. according to the industry classification of china securities regulatory commission (csrc), the sample can be divided into four industries, namely, telecommunications, radio and television and satellite transmission services (i63); internet and related services (i64); software and information technology services (i65); computer, communication and other electronic equipment manufacturing (c39). and the industry dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 28 distribution of the sample is shown in figure 1. the data are obtained from a database in which the indicators are calculated according to the annual reports by listed firms. 3.2. variable definition financial indicators are an intuitive indicator to reflect the financial and operational status of firms. in this paper, variables are selected according to five dimensions: solvency, profitability, operational capacity and development capacity as well as cash flow status of each firm. in addition, a total of 20 independent variables are selected considering the characteristics of digital economy firms focusing on innovation and r&d. the definition of each variable is described in table 1. table 1. variable definition. variables abbrev formulas financial distress fail financial distress using 1, financial health using 0 current ratio x1 current assets / current liabilities quick ratio x2 (current assets inventories) / current liabilities cash ratio x3 ending balance of cash and cash equivalents / current liabilities debt-to-asset ratio x4 total liabilities / total assets accounts receivable turnover ratio x5 operating income / accounts receivable inventory turnover x6 operating costs / inventory current asset turnover x7 operating income / current assets total assets turnover x8 revenue from main business / total assets return on assets x9 (total profit + finance costs) / total assets total net asset margin x10 net profit / total assets return on net assets x11 net income / shareholders' equity balance gross operating margin x12 (operating revenues operating costs) / operating revenues total assets growth rate x13 (closing assets opening assets) / opening assets operating income growth rate x14 (ending operating revenue opening operating revenue) / opening operating revenue sustainable growth rate x15 return on net assets * earnings retention rate / (1 return on net assets * earnings retention rate) net assets per share growth rate x16 (closing net assets per share opening net assets per share) / opening net assets per share net profit growth rate x17 (closing net income opening net income) / opening net income net cash content of operating income x18 net cash flow from operating activities / total operating income operating index x19 net cash flows from operating activities / cash generated from operations r&d investment ratio x20 r&d investment / total assets 3.3. model design due to the limited number of digital economy firms in a-shares, but the variables for discriminating financial distress are more and multidimensional, using a large number of financial dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 29 variables will have multicollinearity and overfitting problems. meanwhile, logistic regression is sensitive to the covariance of variables, and manual elimination of variables is easy to ignore multidimensional discrimination. to deal with this issue, the advantage of pca lies in the dimensionality reduction of variables by factor coefficients to avoid over-fitting problems caused by too many independent variables. the concept of pca is based on the maximum variance theory, which maps the original features to feature vectors, and the feature value corresponding to each feature vector is the variance after the projection of the original features to the projection surface of the feature vector. in order to ensure that the information is not lost as much as possible, pca will select the projection plane with larger variance as the projection plane of the original features to get the feature value with the maximum information, so that the loss of information after dimensionality reduction is minimized. in this study, we first discriminate the significance of the difference of variables between the two types of samples, use pca method to reduce the dimensionality of variables when there are more variables, and use logistic regression for the obtained principal component factors, finally discriminate the accuracy of the model for financial distress prediction of digital economy firms. the expression about the logistic model is show in equation (1): , 0 ,1 ( ) ln( ) 1 ji i t j i jj i p logistic fail β β x p       (1) among other things, i=1, …, n, j=1, …, j, xi,j is the jth variable of the ith firm, variable j has a total of 20 variables, and t is the prediction period. logistic regression, a linear model commonly used to deal with dichotomous problems, is widely applicable in the field of predicting financial distress, besides alifiah and norfian (2014) showed a high prediction accuracy [19]. in this study, the explanatory variables are dichotomous {0, 1} variables to measure whether a firm is in financial distress. x is the financial index of the firm, and the coefficient β is obtained by the method of great likelihood estimation, which ultimately leads to the estimated fail=1 probability ,ˆ i tp for the ith firms in period t. the expression is as in equation (2): 0, , ,1 , , , 0, , ,1 ˆ ˆexp( ) ˆ ( 1| ) ˆ ˆ1 exp( ) j t j t i tj i t i t i t j t j t i tj β β x p fail x β β x          (2) if ,ˆ i tp is greater than 0.5, there is a financial distress signal. conversely, if ,ˆ i tp is less than 0.5, there is no financial distress signal [20]. 4. empirical results 4.1. k-s and mann-whiteoy u test we first apply kolmogorov-smirnov(k-s) to test whether the variables of the sample overall are normally distributed, and the results show that period t-3, period t-2 x4 and period t-1 x8, x13 are in line with normal distribution using independent samples t-test, and the remaining indicators are subjected to mann-whiteoy u test for significance of differences in variables between the financial distress sample and the financial health sample. the indicators find to be significantly different in dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 30 period t-3 are x1, x3, x5, x8, x20. the indicators with significant differences in period t-2 are x3, x5, x7, x8, x9, x10, x11, x13, x14, x15, x16, x17, x19, x20. in period t-1, the indicators with significant differences are x1, x2, x3, x4, x7, x8, x9, x10, x11, x12, x13, x14, x15, x16, x17, x18, x19. since there are only five significantly different indicators in period t-3, which is not suitable for principal component factor construction, this paper uses logistic regression after testing the covariance to construct the model for this period. for the remaining two periods, the indicators with significant differences are first downscaled using principal component analysis, and then a logistic model will build. 4.2. principal component analysis the kaiser-meyer-olkin test (kmo) and the bartlett's test for sample indicators is first used to determine the suitability for principal component analysis. the results in table 2 show that the kmo is greater than 0.5 in both periods and the bartlett's test of sphericity is significant at the 1% level of the mean, indicating that the data in both periods are suitable for principal component analysis. table 2. kmo and bartlett's test results. t-2 t-1 kmo measure of sampling adequacy quantity 0.554 0.563 bartlett's test of sphericity sig 0.000 0.000 table 3. component matrix. t-2 t-1 f1 f2 f3 f4 f5 f6 f1 f2 f3 f4 f5 f6 x1 -0.006 0.944 0.132 0.186 0.060 0.008 x2 -0.029 0.943 0.131 0.189 0.071 0.004 x3 -0.019 0.003 0.235 0.618 0.660 0.198 0.135 0.871 0.038 0.151 -0.047 -0.038 x4 -0.569 -0.444 -0.209 0.038 0.379 -0.084 x5 0.099 0.797 0.094 0.081 0.295 0.004 x6 x7 0.038 0.916 0.075 -0.134 -0.117 -0.060 0.226 -0.433 0.647 0.427 -0.001 0.012 x8 0.049 0.919 0.094 -0.044 -0.059 -0.105 0.244 -0.383 0.668 0.431 0.013 -0.005 x9 0.851 0.067 -0.341 0.083 -0.130 -0.021 0.921 -0.134 -0.168 -0.038 -0.114 -0.011 x10 0.894 0.082 -0.300 0.098 -0.100 -0.023 0.929 -0.114 -0.168 -0.022 -0.128 -0.012 x11 0.721 0.072 -0.257 0.164 0.095 -0.008 0.226 0.064 0.027 0.026 -0.141 0.826 x12 0.382 0.189 -0.144 -0.134 0.718 0.038 x13 0.470 -0.141 0.639 -0.317 -0.138 -0.064 0.782 -0.062 -0.312 -0.156 -0.178 0.045 x14 0.065 -0.174 -0.329 -0.053 0.206 -0.572 0.524 -0.007 -0.124 0.031 -0.019 -0.401 x15 0.787 -0.145 0.114 0.036 0.135 0.082 0.397 0.139 0.540 -0.659 -0.124 -0.010 x16 0.736 -0.188 0.526 -0.082 0.025 -0.009 0.101 -0.065 -0.492 0.748 -0.209 0.083 x17 0.213 -0.054 -0.139 -0.542 0.403 0.214 0.499 0.082 0.104 0.222 0.402 -0.004 x18 0.546 -0.107 0.028 0.135 0.469 -0.016 x19 0.028 0.084 -0.215 -0.054 -0.209 0.762 0.102 0.170 0.076 0.144 -0.343 -0.412 x20 0.063 -0.031 0.169 0.582 -0.471 -0.046 dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 31 in extracting the common factors, the paper takes the eigenvalues with a cumulative contribution greater than 70% as the main components. a total of 6 principal component factors are extracted in period t-2, with a cumulative contribution of 75.01%, and a total of 6 principal component factors are extracted in period t-1, with a cumulative contribution of 75.34%. according to table 3 of the component matrix, the economic meaning of each principal factor can be determined. in period t-2, the significance of principal factor f1 is profitability (x9, x10, x11), development capacity (x15, x16); the significance of principal factor f2 is operating capacity (x5, x7, x8); the significance of principal factor f3 is development capacity (x13, x16); the significance of principal factor f4 is solvency (x3), innovation capacity (x20); the significance of the main factor f5 is solvency (x3), development capacity (x17), innovation capacity (x20); the significance of the main factor f6 is development capacity (x14), cash flow (x19). in period t-1, the significance of principal factor f1 is profitability (x9, x10), development capacity (x13, x14); the significance of principal factor f2 is solvency (x1, x2, x3, x4); the significance of principal factor f3 is operating capacity (x7, x8); the significance of principal factor f4 is development capacity (x15, x16); the significance of principal factor f5 is profitability (x12), development capacity (x17), cash flow (x18, x19); the significance of the main factor f6 is profitability (x11), cash flow (x19). the calculated two-period principal component factor score matrices, which are omitted from the table to save space, and the two-period score matrices as coefficients of the six principal component factors, are expressed as follows. each principal factor score in period t-2 equations: 1 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.006 0.028 0.011 0.014 0.244 0.256 0.207 0.135 0.018 0.255 0.211 0.061 0.008 0.018 f x x x x x x x x x x x x x x                (3) 2 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.001 0.325 0.373 0.375 0.027 0.033 0.029 0.057 0.071 0.059 0.077 0.022 0.034 0.012 f x x x x x x x x x x x x x x               (4) 3 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.188 0.075 0.060 0.075 0.273 0.240 0.205 0.511 0.263 0.091 0.421 0.112 0.172 0.135 f x x x x x x x x x x x x x x               (5) 4 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.515 0.068 0.112 0.037 0.069 0.082 0.137 0.265 0.044 0.030 0.069 0.452 0.045 0.485 f x x x x x x x x x x x x x x               (6) 5 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.609 0.272 0.108 0.055 0.120 0.093 0.087 0.127 0.190 0.124 0.023 0.371 0.193 0.434 f x x x x x x x x x x x x x x               (7) 6 3 5 7 8 9 10 11 13 14 15 16 17 19 20 0.194 0.004 0.059 0.103 0.021 0.023 0.008 0.062 0.560 0.081 0.008 0.209 0.745 0.045 f x x x x x x x x x x x x x x               (8) each principal factor score in period t-1 equations: 1 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.001 0.007 0.034 0.143 0.057 0.061 0.232 0.234 0.057 0.096 0.197 0.132 0.100 0.025 0.126 0.138 0.026 f x x x x x x x x x x x x x x x x x                   (9) 2 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.294 0.293 0.271 0.138 0.135 0.119 0.042 0.035 0.020 0.059 0.019 0.002 0.043 0.020 0.026 0.033 0.053 f x x x x x x x x x x x x x x x x x                  (10) dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 32 3 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.078 0.077 0.023 0.124 0.384 0.396 0.099 0.100 0.016 0.085 0.185 0.074 0.320 0.292 0.062 0.017 0.045 f x x x x x x x x x x x x x x x x x                  (11) 4 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.117 0.119 0.095 0.024 0.268 0.271 0.024 0.014 0.016 0.084 0.098 0.019 0.414 0.470 0.139 0.085 0.090 f x x x x x x x x x x x x x x x x x                  (12) 5 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.046 0.054 0.036 0.289 0.001 0.010 0.087 0.098 0.107 0.548 0.136 0.014 0.095 0.159 0.307 0.358 0.262 f x x x x x x x x x x x x x x x x x                  (13) 6 1 2 3 4 7 8 9 10 11 12 13 14 15 16 17 18 19 0.008 0.004 0.037 0.081 0.012 0.005 0.011 0.012 0.800 0.037 0.043 0.389 0.009 0.081 0.004 0.016 0.399 f x x x x x x x x x x x x x x x x x                  (14) 4.3. logistic regression in this paper, the t-3 period variables through the presence of differences and the t-2 and t-1 period principal factors derived through principal component analysis are used as independent variables in the logistic regression. the significant difference indicators for t-3 periods are x1, x3, x5, x8, x20. after the cointegration test the variance inflation factor (vif) of x1 and x3 is greater than 5, which indicates that there may be cointegration. considering that the current ratio (x1) includes cashlike assets, the cash ratio (x3) is excluded. the logistic regression coefficients for the three periods are shown in table 4. from the regression results, it can be seen that the r&d investment ratio (x20) in period t-3 is a significant predictor of financial distress, and the lower the r&d investment ratio is, the higher the possibility of financial distress. additionally, the principal factor f1 in both period t-2 and period t-1 is negatively and significantly correlated with financial risk at the 1% level, and f1 in both periods includes corporate profitability, indicating that the profitability factor has the largest contribution to the financial distress prediction of digital economy firms. table 4. logistic model estimated coefficients. period/factors t-3 t-2 t-1 x1 -0.100 x3 -0.016 x8 0.294 x20 -25.827** f1 -1.234*** -3.158*** f2 -0.079 -0.099 f3 0.078 -1.625 f4 -0.578** 1.699 f5 -0.005 0.624 f6 -0.331 0.033 constant 0.740 0.130 0.901** observations 100 100 100 -2 log likelihood 130.227 117.612 85.595 nagelkerke r2 0.107 0.253 0.549 note: *, **, *** indicate statistically significant at 10%, 5%, 1% level. dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 33 accordingly, the equation of the logistic model for each period can be derived , 3 1 5 8 20( ) 0.740 0.100 0.016 0.294 25.827i tlogistic fail x x x x      (15) , 2 1 2 3 4 5 6 ( ) 0.130 1.234 0.079 0.078 0.578 0.005 0.331 i tlogistic fail f f f f f f         (16) , 1 1 2 3 4 5 6 ( ) 0.901 3.158 0.099 1.625 1.699 0.624 0.033 i tlogistic fail f f f f f f         (17) 4.4. model accuracy table 5. distress prediction accuracy. period firms financial health financial distress prediction accuracy t-3 financial health 25 25 50% financial distress 14 36 72% total 61% t-2 financial health 43 7 86% financial distress 16 34 68% total 77% t-1 financial health 45 5 90% financial distress 9 41 82% total 86% figure 2. financial distress prediction roc curve. the model is used to further calculate the accuracy of financial distress prediction for each period, and the results in table 5 show that the prediction accuracy for firms in financial distress in period t-3 reached 72%, and the total prediction accuracy is 61%. the total prediction accuracy in period t-2 is 77%, and the total prediction accuracy in period t-1 is 86%. this result shows that the dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 34 closer the year of financial distress, the higher the accuracy of the prediction. in conclusion, the model has a high accuracy rate and a strong prediction function for the financial distress situation of digital economy firms. in addition, according to the roc curve in figure 2, the auc (area under the roc curve) is 0.818, which imply that the model has predictive value for the financial distress of digital economy firms. 5. conclusion and discussion in this paper, we screen indicators reflecting five dimensions of firms solvency, profitability, operating capacity and development capacity as well as cash flow status by k-s test, and has increased the proportion of r&d investment considering the innovation characteristics of digital economy firms, then we find that there are significant differences in indicators between financial distress firms and healthy firms in each dimension. based on the analysis above, pca-logistic regression is applied to predict the financial distress situation of digital economy firms, and the indicators with differences are downscaled by pca to obtain the principal component factors and then logistic regression, finally the prediction model and prediction accuracy of digital economy firms three years before the occurrence of financial distress are obtained. the results show that the model achieve high accuracy and the closer to the year of financial distress the higher the prediction accuracy, indicating that the three distress prediction models are good enough to predict the financial distress of digital economy firms, and the model is constructed successfully and can provide a basis for decision making for management and information users such as investors and government. it is not difficult to find that the significant contribution of profitability and innovation capacity of digital economy firms to the prediction model, which reflects that the characteristics of digital economy firms are significantly related to the prediction of financial distress and the deterioration of related indicators is the direct cause of the distress of firms. the financial data of digital economy firms listed in china a-shares have become easily accessible, and firms in the digital economy should make full use of these data to dig deeper into the risk patterns of various factors in the industry in order to obtain risk evolution patterns to prevent and guide their own financial risk management. in this context, the regulator is responsible for ensuring that firms operating have efficiency and sustainability incentives and can avert financial distress. at the same time, it is also able to promote the healthy development of the industry and improve risk identification and resilience. funding: this research was funded by the chengdu research base of philosophy and social sciences--research center project of chengdu chongqing twin city economic circle (no. cysc21b002), the youth foundation project of social science and humanity, china ministry of education (no. 17yjc790083), and the sichuan county economic development research center (no. xy2018016). conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] barboza f, kimura h, altman e. machine learning models and bankruptcy prediction[j]. expert systems with applications, 2017,83:405-417. doi: https://doi.org/10.1016/j.eswa.2017.04.006. [2] uthayakumar j, vengattaraman t, dhavachelvan p. swarm intelligence-based classification rule induction (cri) framework for qualitative and quantitative approach: an application of bankruptcy prediction and dongyang li, kai xu, yun li, et al. / journal of risk analysis and crisis response, 2022, 12(1), 25-35 doi: https://doi.org/10.54560/jracr.v12i1.319 35 credit risk analysis[j]. journal of king saud university computer and information sciences, 2020, 32(6):647-657. doi: https://doi.org/10.1016/j.jksuci.2017.10.007. [3] ashraf s, félix g s, elisabete, serrasqueiro, zélia. do traditional financial distress prediction models predict the early warning signs of financial distress?[j]. journal of risk and financial management, 2019, 12(2):138-143. doi: https://doi.org/10.3390/jrfm12020055. [4] fitzpatrick p j. a comparison of the ratios of successful industrial enterprises with those of failed companies[j]. certified public accountant, 1932,2:589-605. [5] foster, george. financial statement analysis[m]. prentice hall, 1985,2:625. [6] laitinen e k. financial ratios and different failure processes[j]. journal of business finance & accounting, 1991,18(5):649-673. doi: https://doi.org/10.1111/j.1468-5957.1991.tb00231.x. [7] wu s, huang s. analytical indicators and prediction models of enterprise bankruptcy[j]. china economic issues, 1987(06):8-15. doi: https://doi.org/10.19365/j.issn1000-4181.1987.06.002. [8] wu x. financial crisis early warning research: problems and framework reconstruction[j]. accounting research, 2011(02):59-65+97. doi: https://doi.org/10.3969/j.issn.1003-2886.2011.02.009. [9] altman e i. financial rations, discriminant analysis and the prediction of corporate bankruptcy[j]. journal of finance, 1968,23(4):589-609. doi: https://doi.org/10.1111/j.1540-6261.1968.tb00843.x. [10] martin d. early warning of bank failure: a logit regression approach[j]. journal of banking & finance, 1977,1(3):249-276. doi: https://doi.org/10.1016/0378-4266(77)90022-x. [11] ohlson j. a. financial ratios and the probabilistic prediction of bankruptcy[j]. journal of accounting research, 1980,18(1):109130. doi: https://doi.org/10.2307/2490395. [12] jiang x, sun z. weak governance and financial crisis: a predictive model[j]. nankai management review, 2001(05):19-25. doi: https://doi.org/10.3969/j.issn.1008-3448.2001.05.005. [13] li h, chen y, zhao g. research on cash flow-based financial early warning: a comparison of linear probability model and logistic model applications[j]. exploration of economic issues, 2011(06):102-105+111. doi: https://doi.org/10.3969/j.issn.1006-2912.2011.06.022. [14] li c. construction of prediction model of enterprise financial risk based on logistic regression method[j]. statistics and decision, 2018,34(06):185-188. doi: https://doi.org/10.13546/j.cnki.tjyjc.2018.06.045. [15] altman e i, małgorzata iwanicz-drozdowska, laitinen e k, et al. financial distress prediction in an international context: a review and empirical analysis of altman's z-score model[j]. journal of international financial management & accounting, 2017,28(2):131–171. doi: https://doi.org/10.1111/jifm.12053. [16] sun j, li h, huang q h, et al. predicting financial distress and corporate failure: a review from the stateof-the-art definitions, modeling, sampling, and featuring approaches[j]. knowledge-based systems, 2014, 57(feb.):41-56. doi: https://doi.org/10.1016/j.knosys.2013.12.006. [17] zeng y. research on the trend and social effects of digital economy[m]. china social science press, 2021:11-12. [18] wang x, zhang l, he x. a comparative study on the effectiveness of financial distress prediction based on consolidated statements and parent company statements[j]. accounting research, 2017(06):38-44+96. doi: https://doi.org/10.3969/j.issn.1003-2886.2017.06.007. [19] alifiah m n, norfian m. prediction of financial distress firms in the trading and services sector in malaysia using macroeconomic variables[j]. procedia social and behavioral sciences, 2014,129(6):90-98. doi: https://doi.org/10.1016/j.sbspro.2014.03.652. [20] he p, lan w, ding y. is china's stock market predictable? -a perspective based on a combined lasso logistic approach[j]. statistical research, 2021(05):82-96. doi: https://doi.org/10.19343/j.cnki.111302/c.2021.05.007. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). evaluation on eco-environmental quality of dongsheng district evaluation on eco-environmental quality of ordos plateau ruiyan wu, ming zhao, ruiping zhou geograghic science college, inner mongolia normal university, hohhot, 010022, china e-mail: 1362607775@qq.com abstract according to “technical criterion for ecosystem status evaluation” issued by state environmental protection administration, on the basis of arcgis platform, using the comprehensive evaluation index method, the paper takes dongsheng district as a case, selected biological richness index, vegetation coverage index, land stress index, water network denseness index and pollution load index to evaluate the quality of ecological environment. the evaluation result shows that the ecological environment quality of dongsheng district is "normal", the degree of vegetation coverage is moderate, the biological diversity is general, and it is suitable for human to live, but sometimes exist restrictive factors limit human existence. the ecological index value increased from 52.06 of 2005 to 53.11 in 2014, and land ecological environment quality of dongsheng district has rising trend from the overall view, but the amplitude variations in the ecological environment condition has no obvious change, land ecological environment is good, the land ecosystem structure is complete, and its function is basically perfect. keywords: eco-environmental assessments, ecological index, eco-environmental quality 1. introduction eco-environment is the basis of humanity survival and development is a social economic and natural ecological system which can provide people natural resources and living environment. but ecological environment are always polluted even destroyed because of development of social and economy rapid population growth. once polluted or destroyed, it will influence even limit the development in all aspects. we are supposed to launch scientific and comprehensive evaluation on ecological environment quality, as well as understand the causes and principle of dynamic change as ecological environment construction has become tough and urgent task[1]. eco-environment quality assessment means to evaluate the advantages and disadvantages of the regional ecological environment according to selected index system[2]. according to the difference of research process and the object of the evaluation of the ecological environment, the study can be divided into 2 categories: one is evaluation on the status of the ecological environment (ecological environment quality), two is to evaluate the service function of the ecological environment (the value of the ecological environment)[3]. ecological environmental assessment is to propose strategies and measures to reduce influence or improve ecological environmental condition after analysis, prediction and assessment of the impact human economic and social construction activities may cause. it is an important basis for prevention of ecological environment problems, rational exploitation and utilization of resources, environmental protection and sustainable economic and social development planning. therefore, to strengthen the monitoring and evaluation of ecological environment has important practical significance to the protection of the ecological environment, management and regional sustainable development strategy[4]. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 82 received 19 december 2016 accepted 12 april 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 2. research area and data sources 2.1 research area dongsheng district is located in the southwest of the inner mongolia autonomous region, central part of the ordos plateau. the geographical coordinates of east longitude 109 degrees 08′04″—110°23′11″, north latitude 39°30′42″— 39°58′44″, it is also one of the city's core area with its city's largest built area, the most perfect functional facilities and the highest degree of economic activity. the study area possess 2512.3 km2, 60.12 ten thousands population. 2.2 data source dongsheng district’s statistical bulletin of the national economic and social development (2005-2014), dongsheng statistical yearbook (2005-2014), final database of land change survey (2005-2014), some data are provided by bureau of statistics and meteorological administration. 3. framework of evaluation index system in order to carry out “environmental protection law of people’s republic of china”, strengthen ecological environment protection, evaluate the status and trends of ecological environment, ministry of environmental protection of the people’s republic of china issued and carried “technical criterion for ecosystem status evaluation” at 13 march, 2015.the technical criterion set annual evaluate index system, calculation method and classification standard to town, provincial and ecological environment status and dynamic trend. 3.1 evaluation index system of ecological environment condition ecological environment’s status evaluation is to reflect the overall state of the regional ecological environment using a composite index (ecological index). the index system includes biological richness index, vegetation coverage index, water network denseness index, land stress index, pollution load index, partial indexes are reflect biological abundance, high or low vegetation coverage, water abundance, stress intensity and pollutant loading pressure of evaluated area relatively. 3.2 weight and calculation method of evaluation index 3.2.1 weight and calculation method of biological richness index bioi refers to the degree of abundance of biodiversity in the area being evaluated. (1)weight table 1. weight of biological richness index class forest land grass land water area farm land construction land unused land weight 0.35 0.21 0.28 0.11 0.04 0.01 (2)calculation method r ucawgfbio bio s ssssssa i )01.004.011.028.021.035.0( +++++ = (1) where bioa is normalization coefficients of biological richness, value is 511.2642131067. fs is forest land area, gs is grassland area, ws is water area, as is farmland area, cs construction land area, us unused land area, rs is whole area of evaluated region. 3.2.2 weight and calculation method of vegetation coverage index vegetation coverage index uses normalized difference vegetation index (ndvi) of regional unit area to refer to the extent of vegetation coverage. (1)weight table 2. weight of vegetation coverage index class forest land grass land farm land construction land unused land weight 0.38 0.34 0.19 0.07 0.02 (2)calculation method r ucagfveg veg s sssssa i )02.007.019.034.038.0( ++++ = (2) where vega is normalization coefficient of vegetation coverage index, value is 121.165124. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 83 3.2.3 calculation method of water network denseness index ) resourceswater areawater lengthriver ( 3 1 r res r lak r riv wat s a s a s a i × + × + × = (3) where riva is normalization coefficient of river length, value is 84.3704083981; laka is normalization coefficient of water area, value is 591.7908642005; resa is normalization coefficient of water resources quantity, value is 86.3869548281, and rs is whole area of evaluated region. 3.2.4 weight and calculation method of land stress index lani means stress intensity of evaluated region, soil erosion, land desertification are involved to express land stress index. (1)weight table 3. weight of land stress index degree of erosion serious moderate construction land slight weight 0.4 0.2 0.2 0.2 (2)calculation method area)erosion slight 0.2area landn constrctio0.2 areaerosion moderate0.2areaerosion severe4.0( ×+× +×+×= r ero lan s a i (4) where eroa is normalization coefficients of land stress index, value is 236.0435677948. 3.2.5 weight and calculation method of pollution load index (1)weight table 4. weight of pollution load index pollution index sulfur dioxide(so2) cod solid waste other weight 0.4 0.2 0.2 0.2 (2)calculation method p va s vava i codcod r yfcyfcsoso pd ×× + ××+×× = 33.04.0 22 (5) where 2soa is normalization coefficients of so2, dust, cod, values are 0.0648660287, 4.0904459321, 4.3937397289. 2sov is sulfur dioxide emission, yfca is smoke (powder) dust emission, codv is cod emission, p is regional average annual rainfall . 3.3 evaluation on ecological environmental quality ecological index (ei) is used to describe, and its value range 0-100. 3.3.1 weight table 5. weight of each index index biological richness index vegetation coverage index water network denseness index land stress index pollution load index weight 0.35 0.25 0.15 0.15 0.10 3.3.2 calculation method of ecological quality index ei=0.35×biological richness index+ 0.25×vegetation coverage index+ 0.15×water network denseness index + (6) 0.15× (100-land stress index)+ 0.10×(100-pollution load index) table 6 shows degree of ecological environmental quality, and range of ecological index change is always divided into 4 grades, that is not sensitive, slight change, obvious change, significant change (table 7). 4. evaluation on ecological environmental quality in dongsheng district 4.1 analysis on land use structure land cover information was extracted according to “code for classification of urban land use and planning standards of development land (gb50137-2011)” issued by ministry of housing and urban-rural development of the people’s republic of china and general administration of quality supervision, inspection and quarantine of the people's republic of china, and then class the land types. standard specifies classification should according to the main character of land use; land classification has 3 classification systems, that are large class, middle class and small class, including urban and rural land classification, urban construction land classification two parts, which can be divided into 10 large categories, 44 middle categories, 59 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 84 small categories.10 large categories contain construction land, water land, residential land, public management and public service land, land for commercial use, industrial land, warehouse land, transportation use, public facilities, green space. land types used in this paper are grassland, forest land, farmland, water area, urban land, construction land, unused land. then draw these objects on the basis of arcgis platform, the results are shown in figure1, figure 2. 4.2 evaluation on ecological environment on the basis of “technical criterion for ecosystem status evaluation”, combine with table1-7, put the area of difference land use types into the formula, finally can get to ecological quality index during 2005-2014 (table 8). the evaluation results show that biological richness of dongsheng at the high level in 2005, the value is 83.95; biological richness index has great changes during 2005-2014, down from 83.95 in 2005 to 74.64 in 2014, the main reason is the reduce of water wetland area, the cultivated area (figure 2), the weight of water area reached 0.28 to the biological richness index , and most important is dongsheng district is a water shortage region, large changes in water area influence biological richness index significantly. dongsheng district is the economy, politics and culture center of erdos city, the population is relatively concentrated, so ecological types like farmland, woodland and grassland possess high proportion in the total area which was 10.99%, 20.89%, and 23.05%. in addition, the reduction of unused land area is consistent with the increasing trend of the construction land area, which indicates that the economic construction intensity of dongsheng district is larger in the past ten years, the human activities are frequent, and the ecological table 6. degree of ecological index degree excellent good normal poor bad range ei≥75 55≤ei<75 35≤ei<55 20≤ei<35 ei<20 conditi on high vegetation coverage, rich biodiversity, ecosystem is stability, the most suitable degree for human life vegetation coverage and biodiversity is well, it is suitable for people to live. the vegetation coverage is moderate, the biological diversity is general, and suitable for human life, but sometimes there is some limitation for the survival of human beings. poor vegetation cover, severe drought and little rain, fewer species, there are restricting factors the condition is poor, most of land are gobi, desert, saline alkali land, human’s survival environment is poor table 7. range of ecological index change level not sensitive slight change obvious change significant change change value |δei|≤2 2<|δei|≤5 5<|δei|≦10 |δei|>10 description there was no obvious change in ecological environment if 2<δei≤5，the ecological environment is slightly going better; if -2>δei≥-5, going poor. if 5<δei≤10, the ecological environment is obviously going better; if -5>δei≥-10, obviously going poor. if δei>10, the ecological environment status is significantly better; if δ ei<-10, then environment status is significantly going bad. table 8. ecological environment status and change in dongsheng district during 2005-2014. time biological richness index vegetation coverage index water network denseness index land stress index pollution load index ei 2005 83.95 20.40 31.39 39.80 61.59 52.06 2008 75.89 22.46 26.21 39.01 27.75 52.48 2011 75.23 22.28 27.40 39.31 55.35 49.58 2014 74.64 22.15 26.77 31.04 29.12 53.11 journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 85 environment pressure is heavy. the maximum value of vegetation coverage is the highest in 2014 with22.15.vegetation cover index increased from 20.40 to 22.15 between 2005 and 2014, an increase of 1.75. water density index decreased from 31.39 in 2005 to 26.77 in 2014, reduced by 4.62.this is mainly because the water area land area is consistent with the increasing trend of the construction land area, which indicates that the economic construction intensity of dongsheng district is larger in the ten years, the human activities are frequent, and the ecological environment pressure is heavy. the maximum value of vegetation coverage is the highest in 2014 was 22.15. vegetation cover index increased from 20.40 to 22.15 between 2005 and 2014, an increase of 1.75. water density index decreased from 31.39 in 2005 to 26.77 in 2014, reduced by 4.62. this is mainly because the water fig. 1. remote sensing interpretation of land use / cover of dongsheng district in 2014. fig. 2. area of various land use types journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 86 area decreased significantly from2014 to 2005, from 79.53km2 (2005) to 370.59km2 (2014).water density index can reflect the regional ecological environment distribution and climatic characteristics, the more water density index value is large, the more suitable for human to live. when calculating land degradation index, soil erosion degree was mainly observed. the types of soil erosion were divided into mild, moderate and severe erosion. mild soil erosion area in dongsheng region decreased from 767.59km2 in 2005 to 585.33km2, decreased 182.26km2 and moderate erosion area were 664.02km2, 528.71km2, 456.98km2, 398.67km2, severe erosion area reduced from 342.11km2 (2005) to 313.33km2 (2014).the erosion area continued to decrease, land degradation index has an upward trend, and its value is 39.80 in 2005 decreased to 31.04 in 2014.this is mainly attributed to the implementation of the policy of returning farmland to forest and grassland after the second survey of land resources in 2008. the pollution load index is mainly composed of so2, cod and rainfall to calculated, so2 emissions in 2005, 2014 were 42765 tons, 20217 tons, and average annual rainfall was 220.3 mm, 386.9 mm, 252.3 mm, and 450.8 mm. 5. conclusion the evaluation of land ecological environmental quality in dongsheng district results (table 8) shows that the ecological environment quality of dongsheng area belongs to the "normal", the vegetation coverage is moderate, the biological diversity is general, and suitable for human life, but sometimes there is some limitation for the survival of human beings. from 2005 to 2014, ecological index is 52.06, 52.48, 49.58, 53.11, from the overall trend, the land ecological environment quality of dongsheng area improved year by year, the main reason is due to the increase of vegetation index and land degradation, pollution load index decreased year by year, and the average contribution rate of these indexes to ecological index was 16.86%, 28.8%, 33.77%. according to table7, the value of ecological index change is 0.42 during 2005-2008 with no significant changes. in 2008-2011, δei is -2.9, in interval of-2>δei≥-5, the ecological environment is slightly worse. during the period of 2011-2014, δei is 3.53, belonging to 2<δei≤5, ecological environment is slightly better. but from 2005 to 2014, δei is 1.05, indicating that the quality of land ecological environment has not changed significantly, the land ecological environment is better, the structure of land ecosystem is more complete, and the land ecosystem service function is basically perfect. references [1] song j, wang h, wang f. research progress of ecological environment quality assessment and methods review. environmental science & technology, 2013, 36: 448-453. [2] zhou h. study on ecological environmental quality as assessment index system of xinjiang. china environmental science, 2000, 20(2):150-153. [3] zheng x, wang a. study on the integrated assessment of regional ecological environmental quality based on rs & gis for shandong province. acta scientiae circu mstantiae, 2000, 20(4): 489-493. [4] liu h. ecological environment evaluation research in beijing based on rs, gis. capital normal university, 2008. [5] liu j, li x, gao l. application of rs technologies in monitoring and synthetically evaluation on the eco-environment in xinjiang. arid land geography, 2004, 27(4):508-511. [6] xiong y, wang k, huang d. eco-environmental synthetic evaluation based on gis tecnology in hunan province. journal of soil and water conservation, 2004, 18(5): 174-178. [7] cui x, liu g. researches on urban ecosystem environmental assessment in huhhot city. areal research and development, 2011, 30(6): 79-83. [8] ma r. evaluation on eco-environmental quality of hainan island. rural eco-environment, 2000, 16(4):11-14, 32. [9] huang b, wang s, yang h. city eco-environment quality assessment based on gis and rs. journal of tongji university (natural science), 2009, 37(6):805-809. [10] ministry of environmental protection of the people's republic of china. technical criterion for ecosystem status evaluation (hj192-2015). beijing, china environmental science press, 2015. journal of risk analysis and crisis response, vol. 7, no. 2 (july 2017) 82–87 ___________________________________________________________________________________________________________ 87 1. introduction 2. research area and data sources 2.1 research area 2.2 data source 3. framework of evaluation index system 3.1 evaluation index system of ecological environment condition 3.2 weight and calculation method of evaluation index 3.2.1 weight and calculation method of biological richness index 3.2.2 weight and calculation method of vegetation coverage index 3.2.3 calculation method of water network denseness index 3.2.4 weight and calculation method of land stress index 3.3.1 weight 3.3.2 calculation method of ecological quality index 4. evaluation on ecological environmental quality in dongsheng district 5. conclusion references microsoft word evaluating of loan guarantees between parent and subsidiary based on vulnerable option evaluating of loan guarantees between parent and subsidiary based on vulnerable option li li school of management and economics, university of electronic science and technology of china chengdu, 610054, china school of mathematics, yunnan normal university kunming, 650092, china zongfang zhou school of management and economics, university of electronic science and technology of china chengdu, 610054, china xiao lei ideological and political theory teaching and research department, kunming university kunming, 650214, china e-mail: leisurelily@yahoo.cn abstract loan guarantees between parent and subsidiary is the main related transaction in business group, its correct estimation is the foundation of controlling group’s credit risk. applying option theory, this paper researches the vulnerable option characters of loan guarantees, gives the valuation model under two different security methods, theoretically analyzes the influence of such critical factors as companies asset value and equity ratio to loan guarantees’ value and provides theoretical basis for the decision-making of loan guarantees. keywords: business group, vulnerable option, loan guarantees, credit risk. 基于脆弱期权的母子公司贷款担保价值研究 李丽 1, 2 周宗放 1 肖磊 3 1. 电子科技大学/经济与管理学院，成都 610054 2. 云南师范大学/数学学院，昆明 650092 3. 昆明学院/思政部，成都 650214 摘要：母子公司贷款担保是企业集团主要的一种关联交易行为，对担保价值的正确估算是控制集团整体信 用风险的基础。本文基于期权理论分析了贷款担保的脆弱期权特性，给出了两种不同担保方式下的担保价 值模型，从理论上分析了母子公司资产价值和股权比例等关键因素对担保价值的影响，为企业集团母子公 司贷款担保决策提供了理论依据。 关键词：企业集团，脆弱期权，贷款担保，信用风险 1. 引言 近年来，随着农凯、德隆、铁本等企业集团相 继出现财务危机，由此而引发的信用危机给企业经 营发展带来的不利影响已经引起集团经营管理者的 注意。一些大型企业集团纷纷成立风险管理部门来 评估各类交易活动带来的风险，以期将风险控制在 集团整体的风险容忍范围内，从而增强企业自身的 抗风险能力。企业集团在进行外部融资时，既可以 以母公司作保证人，由子公司向银行申请贷款，也 可以以子公司作保证人，由母公司向银行申请贷 款，这两种母子公司之间的关联担保行为不仅在企 业集团内母子公司关联交易记录中占有较大比例， 同时其担保的贷款金额占据了商业银行大客户对外 journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 56-60 published by atlantis press copyright: the authors 56 administrateur texte tapé à la machine administrateur texte tapé à la machine received 26 november 2011 administrateur texte tapé à la machine accepted 12 march 2012 li li, zhou zongfang, xiao lei 担保的大部分资金。无论哪一方做保证人，当借款 人和保证人都不能履行还款责任时，为维护集团整 体信誉和获得银行长久的信贷支持，企业集团通常 需要动用内部资本市场上的闲散资金来弥补资金缺 口。如果集团没有足够的风险储备金，担保带来的 风险超出集团整体的风险容忍水平，关联担保的传 染性、聚集性将会导致企业集团整体发生信用危 机，给集团的经济发展带来严重后果。然而另一方 面，过多的风险储备金会降低资源使用效率，导致 企业集团内部资本市场的投资效率不足，限制集团 的健康发展。因此，对母子公司关联担保行为价值 的科学估算和正确认识是衡量贷款担保风险与未来 补偿能力的基础，有助于加强企业集团信用风险的 管理。 早期关于贷款担保价值的研究文献通常假定保 证人一定具有偿付能力，即担保本身不存在违约风 险。这是因为在早期的西方发达国家，保证人一般 是政府或政府机构，可以看作是无违约风险的担保 人。merton(1977)首次揭示了无违约风险贷款担保的 看跌期权特性 1，并指出保证人的担保行为类似于承 约看跌期权。利用担保责任与看跌期权之间的同构 关系，merton 推导出由政府或政府机构提供的贷款 担保价值的解析解。jones and mason(1980)拓展了他 的研究工作，运用连续时间期权定价理论，对无违 约风险保证人担保的可赎回、附息票债券的担保进 行定价 2。 随着人们对期权面临的风险的不断认识，特别 是场外市场交易的期权，其承约方（writer）也有可 能违约，这类期权被 johnson 和 stulz（1987）3 称之 为脆弱期权（vulnerable option）。类似于看跌期权 承约方的保证人同样也有可能违约。lai(1992)4 首次 关注了由私人或私人机构提供担保的贷款担保估值 问题，他将这种担保称为有违约风险的担保。lai 采 用离散时间框架下的均衡分析，将担保价值视为有 担保的贷款价值与无担保的贷款价值之差。孙艳、 郭菊娥等(2008)5 沿用 lai 的研究思路，得到担保物 权未按比例分配下的贷款担保价值。chang (2006)6 研究了多个担保对象和多个担保企业的担保定价问 题。孙艳，郭菊娥等人（2009）7 则在 chang 的工作 基础上，建立了担保人与借款人一对多情形下，存 在优先求偿权的贷款担保定价模型，侧重剖析了求 偿权的优先次序对担保价值的影响。上述研究都是 以离散时间框架下的均衡分析为基础，对所得各类 担保价值的比较静态分析采用的是蒙特卡洛模拟方 法。本文将在连续时间框架下，利用脆弱期权的性 质，通过严谨的证明得到贷款担保价值的比较静态 分析结论。 另外，前人的研究都是针对没有股权联系的借 款人与保证人进行的，所得结论如果直接应用于母 子公司间的贷款担保定价中，必然会扭曲担保的真 实价值，不利于控制集团整体信用风险。因此，有 必要对具有股权联系的母子公司间的担保行为价值 进行科学评估。 本文第二部分首先介绍模型基本假设，然后分 别建立两种关联担保方式下的贷款担保价值模型， 并利用脆弱期权的性质严格地推导出担保价值的比 较静态分析结论，重点考察母子公司资产价值、股 权比例、资产相关系数以及贷款金额等重要参数对 担保价值的影响。最后是算例分析。第三部分是本 文的不足和进一步的研究展望。 2. 贷款担保价值模型 本节将分别建立以母公司作保证人和以子公司 作保证人，这两种担保方式下的母子公司贷款担保 价值模型，并作比较静态分析。 2.1. 模型假设和符号说明 本文主要分析企业集团内母公司 p 和一个子公 司 s 之间关联担保的情形。假定贷款期限为[0,t], 到 期还款金额为 d ，为简化问题起见，在贷款期内不 考虑两公司各自原有负债。沿用文献 8 中的假设， 在任意的 t∈[0,t]时刻，母公司 p 和子公司 s 的资产 价值分别记为 p(t)和 s(t)， 且 p(t)和 s(t)服从二维几 何布朗运动，即：                   )( )( ))(ln( ))(ln( 2 1 tdw tdw dt tvd tsd p s   （1） 其中， s 、 p 是常数，表示资产价值漂移 率， w1(t)和w2(t)是独立的布朗过程，  是常数二 阶矩阵，满足        2 2 ' ppsps pspss   （2） p ， s 分别表示母子公司资产价值波动率， ps 是 它们之间的相关系数。 根据文献8的分析，母子公司资产价值关系可以 表示为 )()()( tstytp  。这里，y(t)示母公司p的 总价值减去所含子公司s的那部分价值后剩下的价 值。 表示母公司p持有子公司s的股权比例。 published by atlantis press copyright: the authors 57 evaluating of loan guarantees between parent and subsidiary based on vulnerable option 令 2 ( ( ))y var y t  ， ( ( ), ( )) ys y s cov y t s t     表示母公司 总价值中不包含子公司的那部分价值y(t)与其子公司 资产价值s(t)变化的相关系数，则有 syps ys p p         （3） 上式表明，当固定  、 p 、 s 与 y 时， ps 与 ys 成正比，记作： ps ys  。这表示在其它条 件不变的情况下，y(t)部分资产与子公司资产价值相 关性越强，母子公司资产价值相关性越强；类似 地，当固定 p 、 s 、 y 与 ys 时， ps 与  成正 比，记作： ps  。表示当其它条件不变时，母公 司对子公司的持股比例越大，母子公司资产价值相 关性越强。 2.2. 贷款担保价值 2.2.1 母公司 p 为子公司 s 贷款提供担保情形 假设子公司 s 向银行申请到期还款金额为 d 的 贷款，母公司 p 为这笔贷款提供担保。在贷款期末 t 时刻，考虑到保证人也有可能不能足额偿付剩余 贷款，因此，根据母子公司资产价值情况，可以得 到母公司作为保证人所需提供的支付具有以下三种 情形： (1) 子公司的资产 s(t)足以支付贷款 d 时，母公 司无需承担连带担保责任，支付为 0。 (2) 子公司的资产 s(t)不足以支付贷款 d，母公 司的剩余资产 y(t)足以偿付贷款余额 d-s(t)，此时 母公司因为担保，所需提供的支付为 d-s(t)。 (3) 子公司的资产 s(t)不足以支付贷款 d，母公 司的剩余资产 y(t)也不足以偿付贷款余额 d-s(t)， 此时母公司因为连带担保责任，将其所有剩余资产 支付给银行，提供的支付为 y(t)。 综上所述，母公司因为提供担保，在t时刻所需 提供的支付结构为: 0, ( ) ( ) ( ), ( ) ( ) ( ) ( ), ( ) ( ) ( ) min[ ( ), max( ( ), 0)] min[ ( ) ( ), max( ( ), 0)] ps if s t d l t d s t if s t d and y t d s t y t if s t d and y t d s t y t d s t p t s t d s t                 （4） 由以上支付结构可知，母公司为子公司的贷款 d 提供担保相当于以母公司做承约人，出售一份以 子公司资产为标的、执行价格为贷款金额的欧式脆 弱看跌期权。因此，贷款的担保价值为欧式脆弱看 跌期权价值。 根据johson3, 可知贷款担保价值lps满足边界条 件为（4）式的下列微分方程 2 2 2 2 2 2 2 2 2 1 2 1 2 s p ps p s x x x rx rp rs s p s s x x x p p s tp                         （5） 在风险中性假设下，贷款担保的价值为 [ ( )] {min[ ( ) ( ), max( ( ), 0)]} ps q rt ps rt q l e e l t e e p t s t d s t       （6） 其中 eq[.]表示在风险中性概率空间下求期望， r 是无风险利率。 2.2.2 子公司 s 为母公司 p 贷款提供担保情形 同上节分析方法类似，当子公司 s 为母公司 p 的贷款提供担保责任时，考虑到子公司也有可能不 能足额偿付剩余贷款，因此根据期末母子公司资产 价值情况，可以得到子公司作为保证人所需提供的 支付结构为 0, ( ) ( ) ( ), ( ) (1 ) ( ) ( ) (1 ) ( ), ( ) (1 ) ( ) ( ) min[(1 ) ( ), max( ( ), 0)] sp if p t d l t d p t if p t d and s t d p t s t if p t d and s t d p t s t d p t                      （7） 同理根据 johson3, 可知贷款担保价值 lsp 满足边 界条件为（7）式的如（5）式的微分方程。 在风险中性假设下，贷款担保的价值为 [ ( )] {min[(1 ) ( ), max( ( ), 0)]} sp q rt sp rt q l e e l t e e s t d p t       （8） 2.3. 比较静态分析 引理：记 p(s(t),v(t),x,t)为欧式脆弱看跌期权在 t 时刻的价值，其中 s(t)表示标的资产价值，v(t)表 示期权承约方资产价值，x 为执行价格，t 为到期 日，则有以下三式成立： 0   s p ， 0   v p ， 0   x p （9） 证明：易知 p(s(t),v(t),x,t)在到期时的支付为 min[v(t), max(x-s(t),0)]。构造一人造资产 )(ta ， published by atlantis press copyright: the authors 58 li li, zhou zongfang, xiao lei 满足条件 ))(),(max()( tvxtsta  ，则有下式成 立 max( ( ), 0) min( ( ), max( ( ), 0))x a t v t x s t   即 以 ( )a t 为 标 的 资 产 的 普 通 欧 式 看 跌 期 权 ),),(( txtap a 与脆弱看跌期权p(s(t),v(t),x,t)具有 相同的价值。令e(x,y,t)表示在t时刻以价值为y的资 产 交 换 价 值 为 x 的 资 产 的 欧 式 期 权 的 现 值 。 由 margrabe9，有 ( ) ( ) [ exp( ( )) ( ), ( ), ]a t s t e x r t t v t s t t     ( ) is t e  ( ) exp( ( )) ( ) ( ) [ ( ), exp( ( )) ( ), ] s t x r t t v t s t e s t x r t t v t t           exp( ( )) ( ) [ ( ), exp( ( )) ( ), ] exp( ( )) ( ) ii x r t t v t e s t x r t t v t t x r t t v t e              上 式 表 明 1 0 iia e s     ， 1 0 ia e v      ， ( ) 1 0 r t t ia e e x      。其中，下标1表示对第一个分量 取偏导数。进一步，由 merton10中普通欧式看跌期 权的性质有： 1 0 ap ap s s       ， 1 0 ap ap v v       ， 1 0 ap ap x x       。 证毕 引理表明欧式脆弱看跌期权的价值是关于标的 资产价值递减，关于承约方资产价值和执行价格递 增的函数。 结论 1：当母公司为子公司贷款提供担保时， 贷款担保价值关于子公司资产价值和母公司对子公 司的持股比例单调递减，关于母公司资产价值和贷 款金额单调递增。 证明：令 h=max(d-s(t),0)，由引理可知， 1 2 1 2( ) 0 ps ps ps ps psl y h hl l l l s s s s                （10） 1 0 ps psl l p     （11） 2 0 ps psl hl d d       （12） 1( ) 0 ps psl s l       （13） 下标 1、2 分别表示对第 1、2 个分量取偏导数。 这表明，当母公司为子公司贷款提供担保时， 在其他条件不变的情况下，子公司资产价值越大， 贷款担保价值越低；母公司资产价值越大，贷款担 保价值越高；贷款金额越大，贷款担保价值越高； 母公司对子公司的持股比例越小，贷款担保价值越 高。 类似地可以证明，当子公司为母公司贷款提供 担保时，有下列各式成立 0 spl s    ， 0 spl p    ， 0 spl d    ， 0 spl     于是有结论 2：当子公司为母公司贷款提供担 保时，贷款担保价值关于母公司资产价值和母公司 对子公司的持股比例单调递减，关于子公司资产价 值和贷款金额单调递增。 这表明当子公司为母公司贷款提供担保时，在 其他条件不变的情况下，母公司资产价值越大，贷 款担保价值越低；子公司资产价值越大，贷款担保 价值越高；贷款金额越大，贷款担保价值越高；母 公司对子公司的持股比例越小，贷款担保价值越 高。 综合以上分析发现，无论是母公司做保证人还 是子公司做保证人，借款人资产价值越低，贷款担 保价值越高；保证人资产价值越高，贷款担保价值 越高；贷款金额越大，担保越有价值。这符合人们 的经济直觉。另外，母公司对子公司持股比例越 小，保证人的支付越大，贷款担保价值就越高。 注意到 ps  ，因此母子公司资产价值相关性 越弱，贷款担保价值越高。这是因为，相关性越弱 说明母子公司之间分散风险的能力越强，担保越有 价值。 另一方面，如果母公司与子公司分别处于行业 互补型企业，且剥离子公司资产后的母公司剩余资 产 y 与子公司资产 s 负相关。由 ps ys  知，这种 关系下的贷款担保价值要高于同行业内母子公司之 间的贷款担保价值。换句话说，行业互补型母子公 司关联担保给集团控制整体信用风险带来的压力要 小于同行业内母子公司关联担保，原因在于行业互 补有助于风险的分散。 2.4. 算例分析 假设某集团母公司资产初值为p(0)=4000（单 位：万元），旗下一子公司资产初值为s(0)=2400 （单位：万元），母公司对子公司的持股比例为 0.7  ，子公司于2008年初通过由母公司作担保从 银行获得一笔贷款，承诺在2011年初归还贷款的面 值为d=2000（单位：万元）。其余参数取值同参考 published by atlantis press copyright: the authors 59 evaluating of loan guarantees between parent and subsidiary based on vulnerable option 文 献 8 ， 具 体 如 下 ： 0.05p  ， 0.4p  ， 0.05s  ， 0.35s  ， 0.2y  ， 0.8ps  ， r=5%。将参数代入（6）式，模拟50000次，计算得 到lps=163.53（万元），即这笔贷款的担保价值为 163.53万。如果集团内闲散资金多于163.53万，那么 这种担保行为不会增加集团整体的信用风险。 假设这笔贷款是由母公司向银行申请，而子公 司做保证人，其余参数同上，代入（8）式，模拟计 算可得 lsp=44.51（万元）。如果风险管理部门认 为，集团的风险承受能力或者能够承受的最大风险 损失为 100 万，那么根据上述计算结果，建议企业 集团进行外部融资时，由子公司作保证人，以母公 司名义向银行申请贷款，则可以降低集团整体的信 用风险。 3. 小结 本文应用期权理论分析了母子公司贷款担保的 脆弱期权特性，给出了以母公司作保证人和以子公 司作保证人，这两种不同担保方式下的担保价值模 型，从理论上分析了母子公司资产价值、股权比 例、资产价值相关性等关键因素对担保价值的影 响。结果表明，无论是母公司作保证人，还是子公 司作保证人，保证人资产价值以及贷款金额的增大 都会增加担保价值，借款人资产价值的增大会降低 担保价值；在其他因素不变的情况下，母公司对子 公司持股比例的减少会增加保证人的支付，进而增 大担保价值；同时，母子公司资产价值相关性越 弱，贷款的担保价值就越高。特别是从控制风险角 度来看，行业互补型母子公司关联担保给集团控制 整体信用风险带来的压力要小于同行业内母子公司 关联担保。算例分析发现，即使向银行申请同样金 额的贷款，在不同的担保方式下，贷款的担保价值 不同。因此，企业集团可以选择合适的担保方式， 以降低集团整体的信用风险。 集团风险管理者在评估担保行为价值时，还应 注意资产波动对担保价值的影响，但是标的资产和 承约方资产波动对脆弱期权价值的影响较为复杂， 本文未能从理论上得出母子公司资产波动对担保价 值的影响，这将是我们要进一步深入研究的问题。 致谢 本文感谢下列基金的资助：国家自然科学基金 项目（70971015,11061041）；云南省教育厅科学研 究基金项目（2011c109）。 参考文献 1. r. c. merton, an analytic derivation of the cost of loan guarantees and deposit insurance: an application of modern option pricing theory, journal of banking and finance. (6) (1977) 3-11. 2. e. p. jones and s.p. mason, the valuation of loan guarantees, journal of banking and finance. 4(1980) 89-107. 3. h. johnson and r. stulz, the pricing of options with default risk, journal of finance. (42) (1987) 267-280. 4. v. s. lai, an analysis of private loan guarantee, journal of financial services research. 6(1992) 223248. 5. y. sun etc, the research on the value of the loan guarantee under security interest in no proportion, forecasting. 27(2) (2008) 73-76. 孙艳,郭菊娥,王乐,曹华, 担保物权未按比例分配 的贷款担保价值研究, 预测. 27(2) (2008)73-76. 6. c. c. chuang, s. l. chung and m. t. yu, loan guarantee portfolios and joint loan guarantees with stochastic interest rates, the quarterly review of economics and finance. (46) (2006) 16-35. 7. y. sun etc, research on the loan guarantee value with the claim priority, management review. 21(4) (2009) 34-41. 孙艳,郭菊娥,王乐,曹华, 优先求偿权条件下贷款 担保价值研究, 管理评论, 21(4) (2009) 34-41. 8. l. chen and z. f. zhou, a research based shareholding ratio on default contagion between parent and subsidiary company in an enterprise group, journal of industrial engineering and engineering management. 3(2009) 80-84. 陈林,周宗放, 基于股权比重的企业集团内母子公 司之间信用风险传递研究, 管理工程学报. 3(2009) 80-84. 9. w. margrabe, the value of an option to exchange one asset for another, journal of finance. 33(1978) 177186. 10. r. c. merton, the theory of rational option pricing, journal of economics and management science. 4(1973) 141-183. published by atlantis press copyright: the authors 60 microsoft word introduction for volume 4, issue 1 editor’s introduction modern society is a competitive society. reducing existent costs is the ultimate goal of risk analysis and crisis response, which is an important means to enhance the competitiveness. in the present issue of journal of risk analysis and crisis response (jracr), volume 4, issue 1 (2014), such efforts have been embodied in these papers. this issue contains 6 papers, which can be divided into four categories: risk management, human health risk, and risk analysis and risk assessment. the risk management includes three papers. the first paper “risk mitigation in managing a mega project a case study of gomal zam dam construction” by george wang, wei hu and colin duffield, provides the background information of the dam, the specifics of the project and challenges encountered. it also presents the countermeasures and management techniques used under the multitude of uncertainties in this mega infrastructure project, the management lessons learned during the six years construction. the second paper “average life expectancy as a criterion for regional risk management” by prof. timashev, describes principles and methodology of average life expectancy as a criterion for regional risk management. the third paper “on risk and disability – investigating the influence of disability and social capital on the perception and digital communication of risk” by jörgen sparf and susanna öhman, aims to examine how disabled people’s risk perception and preference for digital risk communication channels are influenced by disability in itself and by social capital. in this paper, a quantitative survey (n=6 500) was carried out in sweden showing that while disability is not influential, social capital is. this differs from the results of previous studies on other population groups, which show that group specific factors do influence risk perception. this study suggests that due to the importance of social capital, institutions communicating risk information should build strong relationships with (local) disability associations and networks in order to communicate more effectively. there is one paper on human health risk, "new england compounding center meningitis outbreak: a compounding public health crisis" by bolanle a. olaniran and juliann c. scholl, that focuses on assessment of the necc meningitis outbreak using the anticipatory model of crisis management (amcm). using the amcm principles, it was found that necc did not engage in an adequate vigilant decision making process. there is one paper on the risk analysis. the paper “a cooperative game theoretical approach to risk analysis, using network structure” by jun-ichi takeshita and hiroaki mohri, introduces the framework for a cooperative game theoretical approach to risk analysis using network structure. the risk assessment includes one paper “review of the strengths and weaknesses of risk matrices” by mustafa elmontsri. risk matrices are commonly encountered devises used for rating hazards in different areas of risk management and almost in all industries. the apparent simplicity and transparency of risk matrices have made it very popular but recent research has identified a number of mathematical defects with risk matrices. this review article looks at the development of risk matrices and provides some of the advantages and disadvantages about the three types of risk matrices. risk matrices are helpful tools for risk assessment as they use quantitative measures to ensure consistent method of determining risk but organizations should adjust the design and size of risk matrices to suit their needs. we sincerely thank to the referees for their strong support and kind help. and also thank the authors very much for all their contributions. editor-in-chief chongfu huang professor, beijing normal university email:hchongfu@gmail.com director of editorial department junxiang zhang associate professor, huangshan university email: peonyzjx@126.com journal of risk analysis and crisis response, vol. 4, no. 1 (march 2014), 1 published by atlantis press copyright: the authors 1 microsoft word volume 12, issue 4-6 journal of risk analysis and crisis response, 2022, 12(4), 235-238 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i4.346 235 book review book review of guide for travelling in an uncertain world: what is risk? chongfu huang 1,2,* 1 key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing (100875), china 2 academy of disaster reduction and emergency management, faculty of geographical science, beijing normal university, beijing (100875), china * correspondence: hchongfu@bnu.edu.cn received: november 18, 2022; accepted: december 28, 2022; published: december 31, 2022 abstract: this paper reviews the main contents of the following picture book: guide for travelling in an uncertain world: what is risk? zhang, j., and huang, h.w. 2022. shanghai: tongji university press (isbn: 978-7-5765-0351-7) (in chinese). this picture book can help children and their parents become aware of risk, learn the knowledge about risk analysis, and cultivate interests related to the science, technology, engineering, and mathematics behind risk analysis. it is a welcome and successful contribution to the education of risk analysis to the child and the public. keywords: book review; guide for travelling; uncertain world; risk the popular science picture book guide for travelling in an uncertain world: what is risk? (fig. 1) by professor jie zhang and professor hongwei huang from tongji university, china details the story of a child called qiqi from shanghai, china and his robot friend kuqi when they experience a lot of adventures in an uncertain world. there are four chapters in this picture book. fig. 1. cover of the picture book “guide for travelling in an uncertain world: what is risk?”. in chapter 1, qiqi’s family is introduced (fig. 2). his father is a scientist working on nature hazards, and his mother is a financial analyst. qiqi is a curious boy with a lot of questions such as whether aliens have visited earth. he likes to watch outdoor adventure movies. qiqi’s parents are very busy, and kuqi is a robot created by his father to be a friend for him in case he is lonely. during the dinner, the family discusses qiqi’s examination which will happen tomorrow. qiqi’s is indeed chongfu huang / journal of risk analysis and crisis response, 2022, 12(4), 235-238 doi: https://doi.org/10.54560/jracr.v12i4.346 236 anxious about the examination, but he tells his parents that he is fine. he asks kuqi to come to his room after the dinner. fig. 2. qiqi’s family. in chapter 2, qiqi asks kuqi a lot of questions regarding the coming exam in qiqi’s room, which then initiates the discussions about uncertainties. kuqi explains the natures of the two types of uncertainties to qiqi, i.e., natural variability and knowledge-based uncertainty, and tells qiqi that it is the uncertainties that make him stressful. many examples like the height of children with the same age and the cause about the distinction of dinosaurs are used to illustrate such uncertainties (fig. 3). using examples like weather forecast in the ancient times and nowadays, the role of the development of science and technology in uncertainty reduction is emphasized. fig. 3. examples about the inherent variability. in chapter 3, kuqi explains to qiqi the relationship between hazard and risk. they then look at typical risks caused by traffic accidents, earthquakes, and financial crisis. kuqi explain that only when likelihood and adverse consequences are involved, the risk is real. such an idea is illustrated using examples such as the presence of a rattlesnake in the remote desert or in downtown shanghai. how to reduce risk through decreasing the chances and the consequences are also explained, using examples like building dams to reducing the likelihood of floods, and issuing early warning messages when earthquakes are arriving. in chapter 4, kuqi teaches qiqi how to assess the risk like a detective through asking a series of interesting questions (fig. 5), including: (1) what can go wrong? (2) how it can happen? (3) how likely is it? and (4) what are the consequences? the importance of collecting evidence to support the answers to these questions is emphasized. the potential of recent developments in wireless sensors chongfu huang / journal of risk analysis and crisis response, 2022, 12(4), 235-238 doi: https://doi.org/10.54560/jracr.v12i4.346 237 network, data analytics, and artificial intelligence on answering these questions is also explained. qiqi and kuqi together analyze the risk on the campus of a primary school supposing the typhoon is coming. after these talks, qiqi understands why he worries and becomes less stressful about the coming exam. fig. 4. examples about the impact of chance and consequence on the risk. fig. 5. how to analyze the risk like a detective through asking a series of questions. both authors of this book are well-known experts in the field of risk analysis and management. professor jie zhang obtained his phd from the hong kong university of science & technology in 2009. his research mainly focuses on risk assessment and management of geohazards. he is one of the founding managing editors of the journal of underground space, and is currently the vice-chair of the engineering practice of risk assessment and management committee (tc304) of the international society for soil mechanics and geotechnical engineering (issmge). he is the recipient of the early achievement award from the international association for structural safety and reliability (iassar). professor hongwei huang obtained his phd from tongji university in 1993. his research mainly focuses on risk assessment and management in underground engineering. professor huang is the founding director of the international joint research center for resilient infrastructure of tongji university, and is currently the chair of the risk and insurance research branch of china civil engineering society. he is the recipient of many academic awards, including the second-class scientific and technological progress award from the state council of china, and the first-class natural science award from the government of shanghai, china. as noticed by peter bernstein, the notion of bringing risk under control is one of the central ideas that distinguishes modern times from the more distant past in the thousands of years of history of chongfu huang / journal of risk analysis and crisis response, 2022, 12(4), 235-238 doi: https://doi.org/10.54560/jracr.v12i4.346 238 humankind. in recent years, risk analysis has evolved into a science-based powerful tool to assess, communicate, and manage risk across different disciplines, and has infiltrated almost every field of our society. while it has been widely used in our daily life, the science behind risk analysis is still not well known to the public. this picture book can help children and their parents become aware of risk, learn the knowledge about risk analysis, and cultivate interests related to the science, technology, engineering, and mathematics behind risk analysis. it is a welcome and successful contribution to the education of risk analysis to the child and the public. prof. chongfu huang editor-in-chief of journal of risk analysis and crisis response copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response vol. 9(4); january (2020), pp. 185–193 doi: https://doi.org/10.2991/jracr.k.200117.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article prospects of guizhou province’s ecological agriculture benefit evaluation index system jian he1,2, hongmei zhang1,2,* 1school of big data application and economics, guizhou university of finance and economics, guiyang, huaxi, china 2guizhou institution for technology innovation and entrepreneurship investment, guizhou university of finance and economics, guiyang, huaxi, china 1. introduction 1.1. ecological agriculture ecological agriculture is developed using modern scientific and technological achievements and modern management methods, drawing on the effective experience of traditional agriculture, using ecological principles and economic principles. ecological agriculture is a modernization that can obtain higher economic, ecological and social benefits of agriculture. eco-agriculture requires that chemical fertilizers and chemical pesticides are not used, and that the use of farm manure increases soil fertility, uses biological methods to control crop diseases and insect pests, and protects the living environment. ecological agriculture is a type of agriculture that is dominated by ecological theories, and uses a systematic engineering method, based on the premise of rational use of agricultural natural resources and protection of a good ecological environment, and plans, organizes, and conducts agricultural production according to local conditions. utilization rate, conversion rate of bioenergy, recycling utilization rate of waste, etc., to promote the recycling and multiple reuse of materials within the agricultural ecosystem, with as little input as possible, and get as much output as possible, and obtain the unified comprehensive effects of production development, energy reuse, ecological environmental protection, economic benefits, etc., and put agricultural production in a virtuous circle. ecological agriculture has high ecological, economic and social benefits, and is one of the important components of agricultural modernization. it has a great impact on the sustainable development of agriculture. in the 1920s, ecological agriculture first emerged in europe, and in this concept was first proposed by american soil scientist will albert wedge in 1970, and it has been developing slowly until the early 1990s. western governments officially introduced ecological agriculture to the field of agricultural development [1], and systematically explained various models of eco-agricultural development in foreign countries [2]. with the development of agricultural modernization, ecoagriculture has received widespread attention from chinese government departments and scholars in the 21st century. compared with the traditional agricultural development model, eco-agriculture pays more attention to sustainability and coordination of agriculture, resources and environment [3]. ecological agriculture mainly solves how to promote the coordinated development of agricultural economy and resources, the environment, and optimize the economic, ecological and social benefits of agriculture. agricultural development is of great significance. the development of ecological agriculture is the focus of implementing sustainable development strategies. china is a developing country. the prerequisite for sustainable development is development. poverty cannot reach the goal of sustainable development. in order to improve social productivity, enhance comprehensive national strength and continuously improve people’s living standards we must give top priority to economic development. while maintaining sustained, rapid, and healthy economic development, we must pay attention to the rational use of resources and the continuous improvement of the environment. it should be noted that the implementation of sustainable development strategies is a longterm process. international, regional, national, or local sustainable development strategies should be based on currently formulated reasonable economic, social, ecological and environmental policies and plans, and coordinate with each other. ecological a r t i c l e i n f o article history received 03 august 2019 accepted 22 october 2019 keywords ecological agriculture guizhou province benefit evaluation index system a b s t r a c t ecological agriculture is a major reform in the history of world agriculture. the study of ecological agriculture benefits is of great significance to the development of ecological agriculture. in this paper, by referring to the evaluation index system of ecological agriculture at home and abroad, and according to the actual situation of the development of ecological agriculture in guizhou province, it is introduced into the evaluation of ecological agriculture benefits in guizhou province, and established the evaluation index system of ecological agriculture, which has certain practical and theoretical significance for the evaluation of the development of ecological agriculture in guizhou province. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: zhm1035@qq.com https://doi.org/10.2991/jracr.k.200117.002 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:zhm1035%40qq.com?subject= 186 j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 agriculture focuses on the pursuit of ecological benefits, emphasizes the self-sustainment of biological processes in agricultural production, and seldom considers economic benefits. china is a large agricultural developing country. at present, in the process of rapid economic growth, it faces the existence of many developed countries in the world. problems such as insufficient resources and environmental pollution, as well as a large population and relative resources, the problems of lack of economy, underdeveloped economy, extensive operation, and low level of productivity, so we cannot copy the practices of developed countries. instead, we should explore a new way of investing in provinces with good returns and pursue sustainable development in accordance with china’s national conditions and economic affordability. specifically, china’s ecological agriculture should be based on a careful summary of the rich production practices accumulated in china’s rural areas for thousands of years, learn from foreign experience in ecological agriculture, and take development as the primary goal. ecological agriculture is based on the premise of increasing agricultural comprehensive production capacity and increasing farmers’ income. it goes without saying that the focus of china’s agricultural implementation of sustainable development strategy is to develop ecological agriculture, which is an inevitable requirement of china’s national conditions. the development of ecological agriculture is the only way for guizhou province to achieve sustainable agricultural development and agricultural modernization. as a “underdeveloped” western province, guizhou is one of the typical economic and agricultural poor provinces. guizhou’s agricultural development faces many problems and many people. the lack of land, ecological fragility, weak foundation, and low comprehensive agricultural production capacity have always restricted guizhou’s economic development. at the same time, guizhou’s ecological environment belongs to the extremely fragile karst landform mountain area. it is a mountainous agricultural province without plain support. in ecological agriculture construction there are various problems and difficulties. poverty in rural guizhou is the root of the problem, and the lack of food and self-sufficiency is the core of the problem. the agricultural development of guizhou’s karst areas is facing a growing agricultural population, a fragile agricultural environment, worsening. there are a series of very serious problems such as the lack of agricultural resources such as water and land resources, backward economic development in rural areas, low average annual income of farmers, and slow economic growth. therefore, the construction and development of ecological agriculture will affect the whole of guizhou. increased economic growth rate. ecological agriculture is a comprehensive agriculture, because the comprehensive development of agriculture, forestry, animal husbandry, fisheries and the joint operation of agriculture, industry and commerce in accordance with local conditions and conditions will not only help solve the problem of the transfer of rural surplus labor to non-agricultural industries, but also gradually develop regional products to drive rural people out of poverty and become rich. continuously increasing farmers’ income has played an important role in building a new socialist countryside. guiyang is the provincial capital of guizhou. in april 2012, the guiyang municipal people’s government issued guiyang’s implementation opinions on adhering to the path of ecological agriculture development and accelerating the adjustment of the agricultural industrial structure. in order to seriously implement the central and provincial party committee rural work conference, the city the spirit of the ninth party congress, and the implementation of the spirit of the “opinions of the people’s government of guiyang municipality of the communist party of china on vigorously developing ecological agriculture,” the local governments adopted the concept of ecological cluster economic development, highlighting regional advantages, breaking administrative boundaries, and implementing regional vertical and horizontal integration and vigorous development of ecological agriculture. therefore, vigorously developing high-efficiency ecological agriculture with guizhou characteristics and in line with guizhou’s actual conditions will be an effective way to comprehensively solve the agricultural development in guiyang. 1.2. foreign research status foreign studies on ecological agriculture started earlier than domestic ones. the agricultural development of western developed countries generally first experienced the modern agricultural stage. after the high-input and -output production process, the ecological environment was caused great damage, and the government began to reflect on it. and gradually embarked on the path of ecological agriculture. we will summarize it as “destroy first, then protect”. at present, western countries have made world-renowned achievements in the development of ecological agriculture, and foreign countries’ evaluation and research on ecological agriculture systems are gradually being carried out. from the perspective of evaluation content, foreign scholars usually focus on the evaluation of construction effectiveness, and combine economic, ecological and social benefits for comprehensive evaluation. barnet et al. [4] summarized various indicators that can be used to measure sustainable agricultural development, including the total factor productivity index p), and the input–output ratio related to soil health index, and concluded that the tornqvist-theil index is the most suitable for estimating the sustainable agriculture. the tornqvist index used to measure the growth of total factor productivity (tfp) is an exact index, and the tornqvist index is a measure of the divix index. obtained by discrete approximation. herdt and steiner [5] pointed out that the sustainable development capability of the system should be measured from three aspects: biophysical, economic, and social benefits. nowadays, the more widely used evaluation systems include tfp, actuator sensor interface (asi) and other systems, but due to tfp there are some disadvantages that cannot accurately measure the role of sustainable development in the system, and this is exactly the basis for the sustainable and stable development of agriculture. there are also systems that focus on the evaluation of ecological benefits and cannot compare different ecological agricultural systems. foreign studies on ecological agriculture mostly focus on evaluation methods, evaluation criteria, and evaluation index systems. among them, the design of evaluation systems is the focus of ecological agriculture evaluation. in recent years, most of the foreign countries have adopted frameworks for the creation of ecological agriculture evaluation systems. systems, such as goal oriented-concept-system-stress-state-response-system, etc., the development trend of evaluation methods has gradually changed from qualitative analysis to quantitative evaluation analysis, of which the research on index method is more and more mature, and it also appears newer methods, such as gray system theory, fuzzy mathematical statistics, etc., and more evaluation content focus on con j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 187 struction effectiveness. barnett also believes that the tornqvist-theil index is the best factor for assessing agricultural sustainability, and he has constructed various indicators to measure the sustainable development of agriculture. in terms of economic benefits, agricultural costs, agricultural production, agricultural profits and other indicators are usually selected. in terms of ecological benefits, it is mostly evaluated from the aspects of energy conversion, ecological environment status, and nutrient conversion. how to develop ecological agriculture in poor and backward areas does not provide practical experience for our country, so we must study the development path of ecological agriculture suitable for our country on the basis of learning from foreign experiences, but we must not go abroad to “damage first, then protect” “old road”. 1.3. status of domestic research although domestic research on ecological agriculture started relatively late, in recent years china has made great breakthroughs in the development of ecological agriculture. li and zhu used a combination of qualitative analysis and quantitative judgment to build a three-level, 18 an index system for evaluating the comprehensive benefits of ecological agriculture based on specific indicators, using the analytic hierarchy process (ahp) method to calculate the weight of each indicator, using the method of fuzzy membership function to perform dimensionless processing, and linearly weighting the function of each indicator’s weight and dimensionless value. finally, the comprehensive benefit index of ecological agriculture is calculated, and the statistical analysis method of comprehensive benefits is provided [6]; zhou [7] established the comprehensive benefit index system of ecological agriculture, specifically analyzed the development status of ecological agriculture in counties in jiangsu province, and proposed the realization the path of ecological agriculture; jun and jianmin used the ahp method and followed a series of paradigms to build an indicator system for the development of china’s ecological agriculture and reached the final conclusion: the most important factors affecting the development of ecological agriculture is the main body of agricultural production and the benefits of agricultural production; according to the principles of systems engineering, ye et al. [9] use the principle of combining ahp and mathematical statistics to study and establish an ecological agricultural comprehensive benefit evaluation system with three sub-benefits of ecological and environmental benefits, economic benefits and social benefits, and divide the three sub-benefits into several specific indicators. ahp calculates the index weight coefficient of the contribution of the specific response index. after obtaining the weighted index value of each index, it is summed to obtain the sub-benefit and comprehensive benefit values. zhu and ma [10] established a comprehensive benefit indicator system for ecological agriculture, analyzed the development of ecological agriculture in counties in jiangsu province, and proposed a path to achieve ecological agriculture. zhao and fu [11] proposed an effective projection pursuit (pp) model using the projection direction information to evaluate the development level of ecological agriculture and provide a basis for decision-making for the rational construction of ecological agriculture. yang et al. [12] analyzed the system coupling relationship between ecological agriculture and tourism from the perspective of industry coupling, and concluded that the development of eco-agriculture not only promotes ecological benefits, it can also make a great breakthrough in economic benefits, especially in the ecological tourism industry. wang [13] and others comprehensively used the analytic hierarchy process and delphi method to discuss the comprehensive evaluation index and the weight of each index. deng and xiao [14] used a multi-level fuzzy evaluation method to evaluate ecological agriculture. wang used factor analysis to analyze the impact. the factors of comprehensive benefits of ecological agriculture are divided into comprehensive capacity factors, balance capacity factors and sustainable capacity factors. based on the data indicators of 18 provinces and municipalities in henan province, a neural network model is constructed for simulation testing. the comprehensive agricultural benefits are ranked, and it is found that the comprehensive capacity factor has the greatest impact on the benefits of ecological agriculture. therefore, we should start from the aspects of increasing the economic output ratio, accelerating the transfer of rural surplus labor, increasing the per capita net income of farmers, and the comprehensive production capacity of ecological agriculture. in the development of ecological agriculture, li and zhang [16] explored the efficiency, regional differences, and influencing factors of agricultural waste conversion and utilization from the perspective of industrial linkage, and pointed out that the current scale efficiency of the development of the ecological edible fungus industry in china is low, which is the development of the ecological edible fungus industry seeking a path breakthrough. domestic scholars have systematically researched and discussed the development of the system of eco-agriculture and the sustainable development of eco-agriculture. according to the above research, the comprehensive benefits of eco-agriculture have become a hot topic that society and scholars pay close attention to, so based on the previous literature, this article explores how to construct an index system for evaluating the benefits of eco-agriculture through field investigations in parts of guizhou province, with a view to providing a reference for the development of eco-agriculture in guizhou province. 2. determination of guizhou province’s ecological agriculture benefit evaluation index system 2.1. principles for setting up an ecological agriculture indicator system the term indicator is derived from the latin indicator. it has the meaning of revealing, indicating, or making the public aware. it is a “sign” that reflects the characteristic information of things. according to the different forms and functions of indicators, it can be divided into quantitative indicators and non-quantitative indicators are qualitative indicators. both types of indicators can reflect the “quality” of things. indicators can make concepts and theories easier to understand and operate, the goals are clear and clear, and people’s actions and efforts are more effective. during monitoring, the indicator system is a structural system that can comprehensively reflect all aspects of “characteristics” of the characteristic information of things. 188 j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 for the evaluation of the benefits of ecological agriculture, we must examine the ecological agriculture indicator system of guizhou province from the perspective of systems science. at the macro level, we must fully consider the characteristics of the integrity, dynamics, hierarchy, and diversity of the system. we must also consider that the eco-agricultural indicator system is a large system with a strong hierarchy. in this large system, there are multiple subsystems, and each subsystem contains many single indicators according to the system’s coherent action principle and diversity characteristics. when we establish the ecological construction indicator system of guizhou province, we must try to be comprehensive and not repeated. the indicator system is used to describe and evaluate the goals, conditions, levels, and trends of ecological agriculture construction in guizhou province. the basic purpose is to seek a set of characteristic indicators that have typical representative meanings and can comprehensively reflect all aspects of ecological construction requirements. these indicators and their combinations can appropriately expressing people’s quantitative judgments on the construction goals of ecological agriculture. this requires us to start from the system structure and elements of ecological agriculture when researching the indicator system, and set the indicators based on some basic principles. it is generally believed that the following principles should be followed: (1) purpose: the evaluation index system must be able to reflect the final effect of the comprehensive benefits of ecological agriculture. (2) comprehensiveness: the evaluation system can systematically reflect the organic unity of ecological, economic and social benefits. (3) scientific: the selection of evaluation indicators must strictly abide by certain principles of ecological economics, and the evaluation methods selected should be able to be weighed and expressed quantitatively using modern means as far as possible. (4) importance: ecological agriculture is a complex system. for the current production status and development trend, it is necessary to choose indicators that can reflect the final effect. (5) comparability: the selected index can be quantified without dimensions, and it is easy to collect, and it is feasible and comparable within the range. 2.2. setting of guizhou province’s ecological agriculture index system based on the principle of the selection of the above index system, this article combines the actual investigation of guizhou province, selects the ecological agriculture benefit evaluation index from the theoretical and practical aspects, and builds an evaluation index system based on this. the constructed evaluation index system can be comprehensively reflected. in the development of ecological agriculture in guizhou province, the following factors are considered when selecting evaluation indicators: 1. eco-agriculture in our country is an agricultural ecoeconomic composite system with synergistic growth of economic, ecological and social benefits established in accordance with the principles of ecology and economics, and the structure and function of which are constantly optimized within different ecological conditions and business scopes. the selection of methods and indicators should conform to the following principles: (1) whether the system structure of ecological agriculture is optimized depends on whether it can increase the biomass production and conversion efficiency, and the focus should be on the evaluation of the system’s function. (2) the economic benefit is the primary purpose of building ecological agriculture. whether the system structure is reasonable depends on whether the resource advantage can be converted into the product advantage and the commodity advantage to the greatest extent. the development speed of commodity production is the key to measuring economic benefits. (3) social benefits are related to the material and cultural living standards of the rural population. the selection of evaluation indicators should reflect the relationship between people and the system, and the indicators that are most closely related to the function of the system should be found. (4) environmental benefits are related to urban and rural areas, people’s production and life, and the ecological and social benefits produced by a beautiful, comfortable, and pollution-free environment are difficult to calculate by value, but they can be comprehensively reflected using fuzzy evaluation methods. (5) the evaluation index system must finally reflect the overall effect of the system, and it must consider dynamic changes, compare history with the current situation, and predict the development trend. (6) the source of the variables for the evaluation index cannot be separated from the existing statistical system, but also convenient for the necessary field surveys, measurements, and supplements. therefore, based on the above principles, this article summarizes the comprehensive benefits of guizhou’s ecological agriculture into three aspects, namely: ecological benefits, economic benefits and social benefits. these three indicators are highly unified, and these three indicators should be fully considered when evaluating. level and status. 2. eco-agriculture is a regional and composite agricultural production system. it is a composite agricultural ecosystem that is a combination of natural and economic systems under artificial control. therefore, it is necessary to consider the agricultural ecosystem and macroeconomics in a comprehensive manner. restrictive and coordinated relationships between systems and social systems. 2.3. development status of ecological agriculture in guizhou province on the premise of theoretical factor analysis, combined with the actual situation in some areas of guizhou province, through field surveys and data collection of various factors affecting eco agriculture, the current status of the development of ecoagriculture in guizhou is deeply explored: 1. jiuzhou town, anshun city, guizhou province is one of the top 10 historical and cultural towns in guizhou province and one of the 20 national villages and towns that are important for protection and construction. it is one of the five economically strong j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 189 towns in southeast guizhou. acres, of which 65,000 acres of land are occupied; 86,000 acres of forest and grassland, 104,000 acres of pasture grassland; 15,000 acres of economic fruit forest. the old state has a milder environment and abundant rainfall, which belongs to the subtropical warm and humid climate, with an average annual rainfall of 1121 mm. the temperature is 15.7°c, which is a relatively high-yield commodity grain and livestock base in the eastern part of guizhou. in addition, the local environment is pleasant and it is very suitable for agricultural development. at the same time, its cultural tourism is developing. the unique “tunbao” culture is a cultural treasure accumulated by the accumulation of local history. this shows that the old state has been a town suitable for survival since ancient times. the agricultural products are rich, and the development of the original ecological agriculture has well protected the local environment and climate. the development of ecological agriculture in jiuzhou town has a high degree of correlation with the development of human history, population, and fertilizer use, such as per capita farmland, factors such as the per capita net income of farmers have an effect on the development of eco-agriculture by affecting the production consciousness of farmers in the cultivated land. projects such as the “mountain river south” tourism complex attract agricultural transfers to the towns and beautiful villages, while concentrating small towns the combination of construction and relocation of poverty alleviation in different places will relocate the poor households living in areas with extremely poor public security conditions, fragile ecological environments, and frequent natural disasters. they will be concentrated near the townships and help them find employment. xinwan modern agricultural industry ecological core demonstration park in xinwan village, jiuzhou town surrounds the fange industry and serves as a key project of characteristic agricultural upgrading projects. the demonstration park promoted the adjustment of agricultural structure, boosted farmers’ income, and made a solid step forward in the construction of ecological villages. 2. xifeng county is a county under the jurisdiction of guiyang city, guizhou province. it is located in the qianzhong economic zone in guizhou province and is an important area for ecological protection in guiyang city. its total area is 1036.5 km2 and the total population in 2013 was 260,000. the area is located in the northern subtropical and southern temperate monsoon climate zone. the climate is more suitable for living and pleasant. the average annual rainfall is 1111 mm and the temperature is 14.5°c. it has the three dimensional characteristics of “one mountain with four seasons and ten miles of different days” in climate. the growth of a variety of agricultural and forestry crops in the warm temperate zone. in 2013, the per capita disposable income of rural residents in xifeng county was 8988 yuan; the total agricultural output value was 1.432 billion yuan; the total grain output was 67,400 tons. figure 1 shows the interest rate in october 2015 county survey. through field investigations in xifeng county, we found that the local place attaches great importance to the protection of the ecological environment and the rich growth of agricultural and forestry vegetation. this not only protects the ecological environment, but also provides important support for the development of ecological agriculture. at the same time, the sales of agricultural products is high, which indicates that local agriculture is more developed, and in addition, the quality of agricultural products is high, and original agricultural products are favored by the consumer market, which indicates that the development of ecological agriculture is positively related to factors such as the degree of forest coverage and the degree of commercialization of agricultural products. forest coverage is an important indicator for evaluating the benefits of ecological agriculture in guizhou, and the degree of commercialization of agricultural products is also an important indicator for evaluating the economic benefits of ecological agriculture in guizhou. at the same time, the climate of many villages in xibei county is due to the special micro-climate formed by the wujiang reservoir area, the agricultural infrastructure is weak, and farmers’ incomes are increasing slowly. in recent years, with the support of higher authorities, they have been actively seeking funds and projects due to the situation. qianben village, liuchang township, xi’an county has helped farmers to grow fruit trees in seedling subsidies, technical training, marketing, and infrastructure construction. the potential rich function of agriculture, seized the favorable time for the xisha jinsha figure 1 | survey of xijing county, guiyang city, guizhou province, october 2015. 190 j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 figure 2 | survey of liupanshui city, guizhou province, november 2015. wujiang bridge to be completed and opened to traffic, and hired experts from guizhou province to carry out scientific planning for the planting of many rural fruit forests, focusing on the creation of qianben pear, xiaonanjiang orange, yingzhong the “four” fruit bases of purple grapes and tea garden jujube and pears are used to develop ecological agriculture and promote farmers to become rich. through this survey, we have provided some ideas for the study of the ecological agricultural indicator system in guizhou province. we can study from economic benefits and other aspects the impact of agricultural product commodity rate and economic output-to-investment ratio on ecological agriculture. on august 11, 2015, xiyi county introduced the linyiju modern leisure and tourism ecological agriculture industrial park project, which adopts the company + cooperative + farmer model and relies on the modern tourism ecological agriculture project built in the tourism resources of the wujiang reservoir basin. mainly, the development of a core area of 0.2 acres in yanglongsi town has formed a demonstration driving effect, driving seven towns including yongjing town, xiaozhaiba town, shijie town, qingshan miao township, jiuzhuang town, and luwo township. the towns and villages form a radiation zone. at the same time, the ecological agriculture demonstration zone in xifeng county has been officially listed, which shows that the vigorous development of ecological agriculture in xifeng county has produced certain benefits. 3. liupanshui city, guizhou province is called “china’s liangdu”. it is located in the west of guizhou. at the end of 2013, liupanshui covers an area of 9965 km2. it has a humid climate zone in the north subtropical monsoon climate 14°c. according to the 2010 national census data survey, the number of permanent residents was 2.85 million, which is seventh in guizhou province. compared with 2000, the birth rate decreased by 8.9%, the natural growth rate decreased by 7.7%, and liupanshui most of the local population has gone out to work in recent years, and the situation of left-behind children and women is more obvious. figure 2 is part of the results of a survey in liupanshui city, guizhou province in november 2015. through the survey, we found that the local natural environment is superior and there are extensive agricultural production areas and natural conditions. however, as the labor force continues to work outside, with the continuing shortage of labor for future generations, and the prospect of left-behind children receiving education is not optimistic, which has caused the problem of insufficient motivation for the development of eco-agriculture. this article believes that the consideration of the comprehensive benefits of eco-agriculture should consider the number of rural surplus labor and the education level of the labor. indicators such as population growth rate and population growth rate are included in the system construction. the number of rural surplus labor force, labor education level, and rural population growth rate are the social benefits in the study of ecological agriculture benefits in guizhou province, which is conducive to the development of rural human resources in guizhou province. to open up employment opportunities for the remaining millions of labor in rural areas. ecological agriculture is a comprehensive agriculture that comprehensively develops agriculture, forestry, animal husbandry, vice, fisheries, and joint operations of agriculture, industry, and commerce in accordance with local conditions, which is beneficial to guizhou provincial poverty-stricken areas to solve the problem of the transfer of rural surplus labor to non-agricultural industries, but also the gradual development of the region competitive products, promote rural masses to become rich, have an important role in building a new socialist countryside. the liupanshui research will certainly help to build social indicators of ecological agriculture. at the same time, national agricultural science and technology office [2015] document no. 76 approved “guizhou liupanshui national agricultural science and technology park” as the seventh batch of national agricultural science and technology parks. guizhou liupanshui national agricultural science and technology park is divided into three areas: core area, demonstration area, and radiation area. the overall layout is divided into “four parks, one belt and one center”, including panxian pear industrial park, liangdu kiwi industrial park, shuicheng taxus industrial park, liuzhi konjac industrial park, and dalaxianguniangniang. the mountain plateau ecological wetland health tourism belt, high-tech technology incubation platform and industrial service center, with the core of increasing the income of the poor and poverty alleviation, industrial development and capacity building as the starting point, focusing on the goal of building a “modern and highly efficient agriculture”, focusing on j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 191 table 1 | evaluation index system of comprehensive benefits of ecological agriculture in guizhou province the target layer first grade indexes the secondary indicators the evaluation index system of comprehensive benefit of ecological agriculture in guizhou ecological benefits forest cover success rate water and soil loss area ratio system vulnerability organic matter content of soil the system energy ratio farmland irrigation area accounted for saving land cropping index economic benefits the per capita net income of rural areas farmers per capita occupancy of grain the vcr agricultural land productivity agricultural labor productivity agricultural and sideline products processing capacity agricultural machinery use intensity the first total industrial production of gross domestic product (gdp) social benefits rate of the rural population health rural labor force employment proportion of the population engel coefficient agricultural science and technology contribution ratio the urbanization level forage-livestock system of agricultural and sideline products fixed number of year of the rural labor force per capita by education creating advantageous features. the agricultural industry focuses on scientific and technological innovation transformation capabilities, scientific and technological entrepreneurship service capabilities, driving industry development capabilities, and increasing farmers’ income to become rich. it promotes scientific and technological research and development, technical training, information networks, and infrastructure construction; a new journey of poverty. 2.4. construction of guizhou ecological agriculture benefit index system on the basis of the above research, the selected indicators were collected according to the actual situation of guizhou province, and the highly relevant and difficult to identify indicators were further eliminated, and finally a three-level ecological agricultural benefit evaluation with a total target, three sub-targets, and 22 specific indicators was formed. index system: based on the collection of relevant opinions, through analysis and screening, construct the index system shown in table 1. based on field investigations and investigations, through analysis of the relevant theoretical content of ecological agriculture development, the following policy implications for ecological agriculture development are proposed: 1. intensify the government’s support and protection for ecological agriculture. the government should formulate relevant policies for the reasonable transfer of land. the high efficiency of ecological agriculture development must form economies of scale, and the government must formulate corresponding policies to allow reasonable transfer of land, improve agricultural efficiency to form economies of scale, and promote the implementation of ecological agriculture. vigorously strengthen government support establish a reliable and stable input system. at present, the agricultural development of developed and newly industrialized countries is inseparable from the country’s vigorous support. aimed at the shortage of agricultural funds and insufficient investment in guizhou province, the establishment of a sustainable agricultural development system attention should be paid to the adjustment and application of input policies. on the basic policy of stabilizing the rural joint production contract responsibility system, it is necessary to strengthen the leadership of ecological agriculture, especially paying attention to establishing typical models and organizing promotion in time. through typical demonstrations, mobilize the enthusiasm of the majority of farmers to lead the masses to modernize ecological agriculture. due to the characteristics of agriculture, while actively exploring collective, individual, foreign investment and other input channels, the state input should still be the main channel, and a stable input ratio should be gradually defined. in the national income distribution pattern, the establishment of fixed funding channels for investment is inherent special funds for water conservancy construction, cultivated land reclamation, and agricultural wasteland should be opened as soon as possible. due to the large price gap between industrial and agricultural products, the prices of agricultural production materials (fertilizers, pesticides, mulch, diesel, etc.) have risen sharply, which is difficult for farmers alone continue to increase agricultural inputs, so increasing agricultural inputs is the key to increasing agricultural stamina for sustainable and rapid development. of course, it is also necessary to classify inputs scientifically, highlight priorities, clarify responsibilities, and prevent interception and misappropriation. to ensure timely arrival and produce maximum benefits. 192 j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 2. improve the rural financial system and strengthen regulations. at present, the scale of rural areas is small and the ability to resist risks is low. large-scale funding is needed to implement ecological agriculture, so guizhou should increase the bank’s credit for agriculture. the state should allocate special funds to support the pilot work of ecological agriculture in guizhou province and improve the financial system. increase farmers’ income and encourage farmers to invest in the construction of ecological agriculture. ecological agriculture is a revolution in agriculture, which has transformed agriculture from a traditional production method to a new one. to regulate people’s behavior. according to the actual situation of guizhou, we should formulate laws and regulations that are suitable for guizhou, and we can learn from the advanced ideas and successful experiences of agricultural development in western developed countries to ensure the healthy development of regional ecological agriculture. 3. establish a technical system suitable for agricultural development in guizhou province. in the construction of ecological agriculture, the establishment of a technological system suitable for agricultural development in guizhou is a guarantee for the sustainable development of agriculture in guizhou. advanced and practical agricultural technology is the main means and basic work for current agricultural growth, and it is also one of the main indicators for measuring the contribution rate of science and technology. to strengthen and improve agricultural science and technology innovation capacity, reserve capacity, and transformation capacity. implement a closer integration of agriculture, science and education, grasp major scientific research projects that affect grain development in guizhou province, and effectively strengthen supporting facilities focusing on the selection and improvement of major grain and oil crops. research on technology and original technology should promote high-yield, stable-yield, high-quality crop varieties according to local conditions. according to the basis of past work, the following technologies should be promoted: construction of hybrid seed production bases and construction of seed projects to ensure self-sufficiency of hybrid seeds; may adopt the results of the new technological revolution, advanced practical technologies and various supporting measures to improve cultivation techniques, implement intensive cultivation and increase the output rate of agricultural products; strengthen infrastructure construction, promote efficient water-saving irrigation technologies, improve agricultural production conditions, and improve agricultural disaster resistance disaster prevention capabilities to lay a solid foundation; it was found in the agricultural process that the use of pesticides and chemical fertilizers is very large, resulting in pollution of water resources and soil, so reducing the use of pesticides and using biological fertilizers is the key to the construction of ecological agriculture. at the same time, we must increase fertilizer inputs and expand green fertilizers. optimize fertilizer use technology; improve soil and land, increase slope and ladder improvement, expand the use of biotechnology, increase cultivated land, and improve the quality of cultivated land; use advanced technology to prevent various disasters and reduce losses; organize large-scale high-quality and high-yield demonstration projects efforts should be made to increase the scientific and technological content of agricultural products and achieve a comprehensive and balanced increase of agricultural products. 4. improve agricultural ecological conditions. at present, due to the low efficiency of grain production in guizhou, land, capital, technology, labor and other factors of production flow from grain production to other industries, resulting in insufficient agricultural infrastructure, reduced disaster resistance, and worsening agricultural ecological conditions. in order to further improve agricultural ecological conditions, construction a good agricultural ecosystem must first strengthen the construction of farmland water conservancy and other infrastructure in various regions of guizhou province, improve agricultural ecological conditions, and develop ecological agriculture. governments at all levels must closely integrate the construction of ecological agriculture with the strategic goals of the western development. make ecological agriculture construction an important part of the region’s economic and social development goals. second, we must make full use of market mechanisms to promote the continuous improvement and development of the joint production contract responsibility system, strengthen the protection of cultivated land, and accelerate the construction of basic farmland protection areas. investigate and punish violations of national interests such as illegal land occupation, excessive land grants, and the use of powers to substitute laws. take back barren land, expand cultivated land resources, further improve the land contract system, and improve collective unified management. combine adjustments to the agricultural structure and improve the agricultural ecological environment. natural ecological characteristics of guizhou province, karst vegetation should be protected to prevent soil erosion, gradually establish a new industrial economic structure that focuses on the development of forestry and animal husbandry products and combines agriculture, forestry, and animal husbandry, rational use of natural resources, and gradually establish a virtuous cycle of agricultural ecology. finally, we must strengthen the environmental management of township and village enterprises and protect the agricultural ecology. environment: according to the state’s regulations for township and village enterprises, develop industries that are pollution-free and less polluting according to local conditions, and carefully investigate industries that are more polluted, carry out technical transformation and environmental treatment within a time limit, and meet emission standards in accordance with national regulations. the development of township and village enterprises in the protection of agroecological environment, and the protection of agro-ecological environment in the development of township and village enterprises. 5. strengthening the popularization and education of the sustainable development strategy of ecological agriculture in guizhou province. the construction and development of agro-ecological environment is not only a problem of agricultural production development, but also related to the implementation of the entire economic development strategy of guizhou province. it can only be done by the joint efforts of the cadres and the masses. one of the main reasons for the serious pollution and destruction of the agricultural environment in guizhou province it is because people lack understanding of the serious consequences j. he and h. zhang / journal of risk analysis and crisis response 9(4) 185–193 193 of environmental pollution and ecological damage, and do not act in accordance with ecological laws and sustainable development strategies. therefore, strengthening the popularization and education of ecological agriculture and sustainable development strategies has become a protection and construction of agricultural ecology an important issue for the environment. in addition, the environmental protection legal system should be one of the components of china’s socialist legal system construction. popular legal knowledge should include the content of environmental law, vigorously publicize the principles and policies of environmental protection, and environmental protection regulations. the vast number of cadres and the masses pay attention to the protection of the agricultural environment and consciously abide by the law. 3. summary of indicator system construction 1. through the study of the above indicators, a more objective index system of ecological agricultural benefit evaluation was constructed, which provided a certain theoretical basis for the evaluation of the development performance of ecological agriculture in guizhou province. 2. if comprehensive evaluation is made on ecological agriculture in different regions of guizhou province, the selected evaluation indicators and weight distribution can be adjusted appropriately by experts according to the specific development of different regions and different ecological agriculture. 3. according to the principles of system engineering, a comprehensive evaluation system of ecological agriculture benefits was established, and the comprehensive benefits of ecological agriculture were divided into three sub-benefits of ecological and environmental benefits, economic benefits, and social benefits. the idea of sustainable development provides a theoretical basis and reference for the construction and benefit evaluation of ecological agriculture in guizhou province. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments this research was financially supported by the regional project of national natural science foundation of china (71861003) and the second batch projects of basic research program (soft science category) in guizhou province in 2017 (foundation of guizhouscience cooperation [2017] 1516-1). references [1] li x. theoretical basis and research trends of ecological agriculture in china. res agric modern 2000;21:341–5. [2] liu x, yang z. development paths and models of japanese rice ecological agriculture. econ geogr 2011;31:1890–6. [3] zhai y, yang s, han q. evaluation theory and empirical research on ecological agriculture. j northw a & f univ nat sci ed 2006;34:54–60. [4] johnston ae, landau s, payne rw, welham sj, rayner ai. sustainability the rothamsted experience. in: barnett v, steiner r, payne r, editors. agricultural sustainability: economic, environmental and statistical considerations. chichester, west sussex: john wiley & sons; 1995, pp. 171–206. [5] herdt rw, steiner ra. agricultural sustainability: concepts and conundrum. in: barnett v, payne r, steiner r, editors. agricultural sustainability: economic, environmental and statistical considerations. chichester, uk: john wiley & sons; 1995, pp. 3–13. [6] li h, zhu k. evaluation index system and evaluation method of comprehensive benefits of ecological agriculture. china forestry econ 2007;9:19–22. [7] zhou r. evaluation and realization of comprehensive benefits of ecological agriculture in county areas. agric econ issues 2002;9:36–9. [8] zheng j, shi j. evaluation of eco-agriculture competitiveness index system based on ahp method. chinese j eco-agric 2010;18:1087–92. [9] ye r, zeng c, hu x. analysis of comprehensive evaluation methods of ecological agriculture benefits. chinese j eco-agric 2006;7:184–7. [10] zhu k, ma c. research on the comprehensive evaluation method of ecological agriculture. j ecol 1991;10:67–70. [11] zhao x, fu q. improvement of projection pursuit model and its application in comprehensive evaluation of ecological agriculture construction. trans chinese soc agric eng 2006;22:222–5. [12] yang h, dong y, yin x. a new path for industrial structure adjustment in underdeveloped areas: eco-agricultural eco-tourism coupling industry development model. j yunnan univ finance econ 2013;16:149–52. [13] wang f. comprehensive evaluation of ecological agricultural management benefits. agric syst sci comp res 1994;10: 69–71. [14] deng z, xiao w. research on evaluation of water-saving agriculture benefits based on multi-level fuzzy evaluation method. water conserv sci econ 2012;18:94–7. [15] wang x. eco-agricultural benefit evaluation of henan province based on factor analysis-bp neural network. econ survey 2015;32:37–42. [16] li p, zhang j. study on regional differences of recycling performance of resource agricultural wastes. econ geogr 2013;24: 150–5. microsoft word volume 11, issue 4-1 journal of risk analysis and crisis response, 2021, 11(4), 146-162 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i4.309 146 article financial technology, big data enterprise financing constraints and big data industry development: empirical analysis based on mediating effect and threshold effect mu zhang 1,*, cheng cao 1,2 and zhiyuan lv 1,2 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: june 6, 2021; accepted: november 26, 2021; published: january 25, 2022 abstract: based on theoretical analysis, we select the relevant data of 30 provinces (autonomous regions and municipalities) in china from 2013 to 2019, and empirically test the impact of financial technology on the development of big data industry and its mechanism using dynamic panel data model, mediating effect test method and threshold effect model. the benchmark regression results show that the regression coefficient of financial technology to big data industry is significantly positive at the significance level of 10%, indicating that the financial technology can directly promote the development of big data industry. the regression coefficient of the dynamic lag term of big data industry is negative, but not significant, indicating that the dynamic lag effect of big data industry is not obvious. the mediating effect test results show that the financial technology can indirectly promote the development of big data industry by alleviating the big data enterprise financing constraints. the big data enterprise financing constraints have a partial mediating effect, and the mediating effect account for 27.63% of the total effect. in addition, the threshold effect test results show that the direct effect of financial technology on big data industry is significantly enhanced when the development level of financial technology is higher than 5.8790, that is, there is a positive threshold effect of financial technology directly promoting the development of big data industry. however, the indirect effect of financial technology on big data industry is relatively weak when the development level of financial technology is higher than 5.4328, that is, financial technology indirectly promotes the development of big data industry by alleviating the big data enterprise financing constraints, which has a negative threshold effect. keywords: financial technology; big data enterprise; financing constraints; big data industry; mediating effect; threshold effect 1. introduction big data industry refers to related economic activities focusing on data production, collection, storage, processing, analysis, and services, including data resource construction, big data software and hardware product development, sales and leasing activities, and related information technology services. seizing the opportunity to promote the development of the big data industry is of great significance to improving government governance capabilities, optimizing people's livelihood public mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 147 services, promoting economic transformation and innovative development ("big data industry development plan (2016-2020)" (ministry of industry and information technology (2016)) no. 412)). the state council put forward the policy opinion of “encouraging financial institutions to strengthen and improve financial services and increase support for big data companies” in the "action program for promoting the development of big data" (guo fa [2015] no. 50). in the context of big data, encouraging and guiding financial institutions to apply big data technology to improve financial service level and risk prevention and control capabilities has become an important way to accelerate the integrated development of finance and big data industries. the definition that financial technology is a technology-driven financial innovation, as proposed by the financial stability board (fsb) in 2016, becomes a global consensus. financial technology aims to use modern science and technology to transform or innovate traditional financial products, financial business processes, and financial industry business models, to fully empower financial development to improve quality and increase efficiency [1-2]. according to the research of fsb, the development of modern emerging cutting-edge technologies such as big data, artificial intelligence, cloud computing, and blockchain has brought a significant impact on the traditional financial market and traditional financial service industry, forcing financial institutions to accelerate the development of financial technology [3]. according to the axiom of action and reaction, the big data industry acts on financial technology, it is countered by financial technology at the same time. on the one hand, the development of financial technology has stimulated the demand for big data technology products and technical services in the financial industry, thereby directly promoting the development of the big data industry; on the other hand, the development of financial technology can also broaden financing channels, reduce financing costs and increase financing efficiency, which can alleviate the financing constraints faced by big data companies [4], thereby indirectly promoting the development of the big data industry. therefore, in-depth exploration of the impact of financial technology on the development of the big data industry and its mechanism is of great significance for accelerating the integrated development of the financial and big data industries and cultivating new momentum for high-quality economic and social development. at present, researches resulting on financial technology promoting the development of the big data industry are relatively rare, and the relevant researches mainly focus on the discussion of the relationship between financial technology and industrial structure upgrading [5-6]. on that account, we select relevant data from 30 provinces (autonomous regions and municipalities) in china from 2013 to 2019, and use dynamic panel data models, intermediary effect test methods, and threshold effect models to empirically test the impact of financial technology on the development of the big data industry and its mechanism. the remainder of this article is organized as follows: in section 2, we provide theoretical analysis and research hypotheses. section 3 refers to empirical model setting, variable selection, research samples and data sources. in section 4, we take benchmark regression, intermediate effect test, threshold effect test and robustness inspection. finally, conclusion and policy recommendations are given in section 5. 2. theoretical analysis and research hypotheses financial technology is the result of the in-depth application of modern emerging frontier technologies in the financial industry, and is an inevitable product of a new round of information technology progress [7]. chishti and barberis [8] believe that financial technology refers to start-up mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 148 or small and medium-sized technology companies that continuously provide innovative applications and financial product development in the financial industry according to the needs of the financial industry. arner et al. [9] think that financial technology is a new combination of financial services and information technology, and a financial solution supported by modern information technology. based on the definition of financial technology in current researches, it can be considered that financial technology is the financial innovation that modern information technology is applied to improve financial products, improve the quality of financial services, strengthen the governance of financial institutions, thus improving the efficiency of financial markets [10]. it can be seen from the connotation of financial technology that the development of financial technology can stimulate the demand for big data technology products and technical services in the financial industry, thereby directly promoting the development of the big data industry. as a result, hypothesis h1 is proposed. hypothesis h1: financial technology can directly promote the development of the big data industry. from a macro perspective, financial technology can promote economic growth. financial technology can also bring about changes in traditional financing methods and payment methods from a micro perspective [11]. on the one hand, compared with traditional sources of capital, financial technology, a new and alternative financing method, can provide companies with lowercost and more convenient financing channels, thereby improving the availability of company financing to a certain extent. on the other hand, the widespread application of financial technology in digital payment systems helps to establish corporate credit records, thereby improving the availability of corporate formal financing through data driven [12]. with the help of digital payment systems and the relevance of various financing channels, financial technology can ease the credit constraints of enterprises to a certain extent [13]. relevant empirical studies have proved the effect and mechanism of financial technology in alleviating corporate financing constraints [4,12]. as a result, hypothesis h2 is proposed. hypothesis h2: financial technology can effectively alleviate the financing constraints of big data companies. it is generally believed that corporate financing constraints will restrict the growth of corporate performance to a certain extent, thereby inhibiting the development of related industries. rajan and zingales [14] selected sample data from 41 countries and empirically examined the relationship between financing constraints and industrial growth. they found that financing constraints have a significant negative impact on industrial growth. based on the research of rajan and zingales, xie and zhang [15] further empirically analyzed the relationship between financing constraints, foreign direct investment and industrial growth, and the results showed that financing constraints had a significant negative impact on industrial growth. in addition, yang et al. [16] showed that corporate financing constraints are an important reason for the low-end evolution of strategic emerging industries. as a result, combined with hypothesis h2, hypothesis h3 is proposed. hypothesis h3: financial technology can indirectly promote the development of the big data industry by alleviating the financing constraints of big data companies. financial technology may have different impacts on the big data industry in different development stages. in initial stage, due to insufficient infrastructure, laws and regulations, low coordination between departments, untimely financial supervision, and increased financial risks, the cost of financial technology development is relatively high while the process of it is far from mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 149 satisfactory [17]. under this circumstance, the demand of financial technology for big data technology products and technical services in the financial industry is relatively limited. the effect that financial technology eases the financing constraints of big data companies by broadening financing channels, reducing financing costs, and improving financing efficiency is not strong enough, and the indirect effects of financial technology on the big data industry are relatively limited. in the middle and advanced stage, with sound infrastructure, laws and regulations, financial risk is under control, and the linkage between financial technology and the big data industry is enhanced continuously [17], the direct and indirect effects of financial technology on the big data industry become increasingly prominent. as a result, hypothesis h4a and hypothesis h4b are proposed. hypothesis h4a: financial technology directly promotes the development of the big data industry with a positive threshold effect. hypothesis h4b: financial technology indirectly promotes the development of the big data industry by alleviating the financing constraints of big data companies, and there is a positive threshold effect. 3. research design 3.1. empirical model setting economic behavior has dynamic characteristics for the fact that it has continuity and inertia, and is affected by factors such as preference. the dynamic panel data model introduces the dynamic lag term of the explanatory variable into the static panel data model to reflect the dynamic lag effect. due to the correlation between the dynamic lag term of the explained variable and the individual effect, the endogeneity of the coefficient estimate is formed [18]. we select the dynamic panel data model to test the direct effect of financial technology on the big data industry, and the dual logarithmic model is set as follows: 1 1 1 1 1 1 1ln ln ln ln k k it it it it i itdldbi dldbi fintech x u          (1) where i indicates region, t indicates year, itdldbi indicates big data industry development level, 1itdldbi  indicates the dynamic lag term of big data industry development level, itfintech indicates financial technology development level, kitx indicates the kth control variable, iu is the individual effect, and it is random disturbance term. the following three double logarithmic models are set up to test the indirect effect of financial technology on the big data industry through the financing constraints of big data companies, according to the causal stepwise regression test method of intermediary effects [17]: 2 2 1 2 2 2 2ln ln ln ln k k it it it it i itsa sa fintech x u          (2) 3 3 1 3 3 3 3ln ln ln ln k k it it it it i itdldbi dldbi sa x u          (3) 4 4 1 4 5 4 4 4ln ln ln + ln ln k k it it it it it i itdldbi dldbi sa fintech x u           (4) where sait is the financing constraint of big data companies, and sait-1 is the dynamic lag term of the financing constraints of big data companies. the mediating effect is tested as follows: the first step is to test formula (1). we can say financial technology directly promote the development of the big data industry if the regression coefficient 1 mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 150 is significant. if not, the mediating effect test is over. the second step is to test formula (2). we can say financial technology significantly affect the financing constraints of big data companies if the regression coefficient 2 is significant. if not, the mediating effect test is over. the third step is to test formula (3). we can say big data corporate financing constraints significantly affect the big data industry, and the mediation effect test is passed if the regression coefficient 3 is significant. if not, the mediating effect test is over. finally, we test formula (4). big data corporate financing constraints have complete mediation effect if the regression coefficient 5 is not significant, while the regression coefficient 4 is significant. big data corporate financing constraints have a partial mediation effect if both the regression coefficients 4 and 5 are significant. the following two threshold effect models are set up to examine the different effects of financial technology and big data enterprise financing constraints on the development of the big data industry under different financial technology development level. firstly, the double logarithmic model is set as follows to examine the different effects of financial technology on the development of the big data industry under different financial technology development level: 5 5 1 6 1ln ln ln (ln )it it it itdldbi dldbi fintech fintech       7 1 5 5 5ln (ln ) ln k k it it it i itfintech fintech x u        (5) secondly, the double logarithmic model is set as follows to examine the different effects of big data enterprise financing constraints on the development of the big data industry under different financial technology development level: 6 6 1 8 2ln ln ln (ln )it it it itdldbi dldbi sa fintech       9 2 6 6 6ln (ln ) ln k k it it it i itsa fintech x u        (6) where 1 and 2 are the threshold for the development level of financial technology. 3.2. variable selection (1) explained variable: big data industry development level (dldbi) due to the lack of big data industry statistics and considering that the electronic information industry is the basic industry among big data industry, this article refers to the literature [19] and uses the industrial scale and product type of the electronic information industry to approximate those of the big data industry. according to the connotation of the development of the big data industry, relevant research results [20-23], and the compilation principle of the "china electronic information industry comprehensive development index" issued by the operation monitoring and coordination bureau of the ministry of industry and information technology, following the selection principles of scientific, objective, systematic, performance, functionality, dynamics, relative independence, feasibility (or operability), and comparability, the evaluation index system is constructed from three dimensions of industry scale, product type, and infrastructure, including three primary indicators and 17 secondary indicators, which is shown in table 1. the entropy weight method [24] is used to determine the index weight (see table 1), and the topsis method [25] is used to evaluate big data industry development level in 30 provinces (autonomous regions and municipalities) in china from 2013 to 2019 (evaluation results are available on request). mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 151 table 1. big data industry development level evaluation index system. destination layer criterion layer index layer (unit) index weight big data industry development level industry scale number of enterprises above designated size (enterprise) 0.043 main business income (billion yuan) 0.053 total profit (billion yuan) 0.064 main business cost (billion yuan) 0.050 total asset (billion yuan) 0.060 total liability (billion yuan) 0.064 product type (export) communication equipment (thousand dollars) 0.089 computer (thousand dollars) 0.081 household appliances (thousand dollars) 0.115 electronic components (thousand dollars) 0.094 electron device (thousand dollars) 0.073 electronic material (thousand dollars) 0.042 electronic equipment (thousand dollars) 0.107 infrastructure mobile phone base station (base station) 0.015 mobile phone exchange capacity (thousand) 0.014 length of long-distance optical cable line (kilometer) 0.017 internet broadband access port (thousand port) 0.018 (2) core explaining variable: financial technology development level (fintech) this article refers to the design ideas of huang et al. [12], tian and zhang [17], and use the provincial digital inclusive finance index issued by the digital finance research center of peking university(https://idf.pku.edu.cn/yjcg/zsbg/485016.htm) as a proxy variable for the level of financial technology development in china’s 30 provinces (autonomous regions and municipalities). (3) mediating variable: big data enterprise financing constraints (sa) enterprise financing constraints refer to the difference between internal financing cost and external financing cost caused by market incompleteness (asymmetric information, agency cost, etc.), which are generally measured by enterprise behavior characteristics such as investment-cash flow sensitivity [26]. the financing constraint index calculated based on enterprise-level data [27] is not applicable for regional data, as a result, this article draws on the method of harrison and mcmillan [28] and uses a single variable to measure the level of financing constraints of regional big data companies. considering that current structure of corporate financing is still dominated by indirect financing, namely, bank loans, which often require fixed assets or intellectual property as collateral [29]. there is a certain negative correlation between collateral and bank loan cost, so as enterprise financing constraints. therefore, we use the reciprocal of the sum of fixed assets and intangible assets of the regional electronic information industry to approximate the level of financing constraints of regional big data enterprises. the larger the reciprocal is, the greater the level of financing constraints of regional big data companies. (4) control variable we select the basic support capacity of big data industry development, regional technological innovation capacity, the degree of opening, and the degree of government connection as the control variables. among them, the regional per capita gdp (pgdp) and the proportion of regional tertiary mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 152 industry (pti) are used to measure the basic support capacity for the development of the big data industry. the regional technology innovation ability is measured by the intensity of regional r&d expenditure (rds) and the number of regional patents granted (npg). the degree of openness by the ratio of regional export value to total industrial output value (dou). the degree of government connection by the proportion of government funds in the regional r&d expenditure (gc) [30-32]. 3.3. research samples and data sources we take 30 provinces (autonomous regions and municipalities) in china as research object (note: the original data in tibet is seriously missing, so it is not listed as the research object). the selected time range is 2013-2019. among them, the data related to the big data industry comes from the "statistical yearbook of china's electronic information industry" and the website of the national bureau of statistics. the original financial technology data comes from the peking university digital finance research center. the original data on big data enterprise financing constraints comes from the "china electronic information industry" statistical yearbook. the original data of each control variable comes from "china statistical yearbook" and "china science and technology statistical yearbook". since the provincial digital inclusive finance index released by the digital finance research center of peking university is only updated to 2018, the provincial digital inclusive finance index in 2019 is estimated by the exponential smoothing method. the remaining missing data were estimated by means of imputation and manual imputation. the descriptive statistics of all variables are shown in table 2. table 2. descriptive statistics of variables. variable (unit) number of samples average standard deviation minimum maximum dldbi 210 0.088 0.148 0.000 0.879 fintech 210 237.398 57.782 118.010 377.730 sa (1/million) 210 1.644 20.450 0.001 296.472 pgdp (thousand yuan) 210 57.700 26.130 23.150 140.210 pti (%) 210 0.482 0.089 0.342 0.810 rds (%) 210 0.017 0.011 0.005 0.062 npg (grant) 210 56526.173 78508.547 502 478082 dou (%) 210 0.326 0.345 0.025 1.595 gc (%) 210 0.241 0.136 0.069 0.573 4. empirical analysis 4.1. benchmark regression we use the system generalized moment (sys-gmm) method in stata16.1 to estimate the coefficient of formula (1) (benchmark regression). in order to choose between a fixed-effects model and a random-effects model, a hausman test is first performed before the benchmark regression. the hausman test results show that the p value is 0.0006<0.01, that is, the null hypothesis of random effects is rejected at the 1% significance level. therefore, the fixed effects model is selected for benchmark regression. the results of the benchmark regression are shown in regression i in table 3. mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 153 it can be seen from regression i in table 3 that the regression coefficient of financial technology to the big data industry is positive and significant at 10% significance level. the results of the arellano-bond autocorrelation test accept the null hypothesis that there is no first-order autocorrelation in the residual series. the hansen over-identification test results accept the null hypothesis of the validity of the instrumental variables, indicating that the setting of formula (1) is reasonable and the instrumental variables are efficient. therefore, accept the null hypothesis h1: financial technology can directly promote the development of the big data industry. the regression coefficient of the dynamic lag of the big data industry is negative, but not significant, indicating that the dynamic lag effect of the big data industry is not yet obvious, which may be related to the short development history of the big data industry. among the control variables, the regression coefficients of the proportion of regional tertiary industry, the intensity of regional r&d expenditures, the ratio of regional export value to total industrial output value, and the proportion of government funds in regional r&d expenditures are significantly positive under the significance level of 5%, 5%, 5%, and 10% respectively, indicating that these control variables can significantly promote big data industrial development. in addition, the regression coefficients of regional gdp per capita and the number of regional patent grants are both positive, but not significant, indicating that the impact of regional per capita gdp and regional patent grants on the development of the big data industry is not yet obvious. 4.2. intermediate effect test under the framework of the dynamic panel data model, the causal stepwise regression test method [17] is used to test the intermediate effect of big data enterprise financing constraints. step 1 is to use equation (1) to test the impact of financial technology on the big data industry, and get regression i in table 3. the results show that the regression coefficient of financial technology on the big data industry is significantly positive at significance level of 10%. step 2 is to use formula (2) to test the impact of financial technology on big data enterprise financing constraints, and get regression ii in table 3. the results show that the regression coefficient of the constraint is significantly negative at significance level of 10%, indicating that financial technology is effective in alleviating big data enterprise financing constraints, that is, accepting the null hypothesis h2: financial technology can effectively alleviate the financing constraints of big data companies. step 3 is to use equation (3) to test the impact of big data enterprise financing constraints on the big data industry, which is shown in the regression iii in table 3. the results show that the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative at a significance level of 10%, indicating that alleviating big data enterprise financing constraints will promote the development of the big data industry. step 4 is to use formula (4) to test the impact of financial technology and big data enterprise financing constraints on the big data industry, and obtain the regression iv in table 3. the results show that the regression coefficient of financial technology to the big data industry is significantly positive at significance level of 5%, however, the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative significance level of 5%, indicating that big data enterprise financing constraints have intermediate effect partly, the ratio of the intermediate effect to the total effect is: (-4.0716)×(-0.1194)/1.7595=27.63%, that is, accepting the null hypothesis h3: financial technology can indirectly promote the development of the big data industry by alleviating big data enterprise financing constraints. mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 154 table 3. regression results of benchmark regression and intermediate effect test. variable regression ⅰ regression ⅱ regression ⅲ regression ⅳ dlbdi sa dlbdi dlbdi lag1 -0.1198 0.0157 0.0207 -0.0842 (0.3480) (0.9080) (0.9060) (0.3640) fintech 1.7595* -4.0716* 1.8812** (0.0590) (0.0710) (0.0120) sa -0.1078* -0.1194** (0.0500) (0.0360) pgdp 0.8583 -1.4893 0.2828 1.7066* (0.3150) (0.4110) (0.5290) (0.0600) pti 2.7963** 6.3937** 0.9414 3.2920* (0.0210) (0.0480) (0.2610) (0.0750) rds 1.6306** -2.9760** -0.1744 0.1178 (0.0240) (0.0280) (0.5520) (0.9070) npg 0.1293 -1.1776** 0.5989** 0.1942 (0.6960) (0.0300) (0.0120) (0.5670) dou 0.6120** 0.5350 0.6324** 0.6632** (0.0170) (0.2970) (0.0140) (0.0120) gc 0.4335* 4.8790 0.2336 0.4160 (0.0620) (0.1280) (0.6300) (0.3180) _cons -7.9334** -1.7637 -9.6530*** -18.1407** (0.0130) (0.8240) (0.0070) (0.0190) ar(1) -1.93 -3.13 -1.75 -1.72 (0.0540) (0.0020) （0.0810） (0.0850) ar(2) -1.63 0.92 -1.18 -1.41 (0.1030) (0.3580) (0.2390) (0.1600) hansen test 26.94 25.22 24.76 19.08 (0.9660) (0.9860) (0.9840) (1.0000) ①***p<0.01, **p<0.05, *p<0.1, and the value in parentheses is the p value. ②lag1 corresponds to the lagging period of the explained variable. 4.3. threshold effect test under the framework of the dynamic panel data model, formulas (5) and (6) are used to test the different effects of financial technology and big data enterprise financing constraints on the development of the big data industry under different financial technology levels. first, we use equation (5) to test the different effects of financial technology on the development of the big data industry under different financial technology levels (recorded as regression v). second, we use equation (6) to test the different effects of big data enterprise financing constraints on the development of the big data industry under different financial technology levels (denoted as regression ⅵ). we test the existence of the threshold effect before the threshold effect regression. for regression v, the p value of the single-threshold test is 0.067, so the null hypothesis is rejected, mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 155 indicating that there is a threshold. the p value of the double-threshold test is 0.244, so the null hypothesis is accepted, indicating that there is no dual threshold. therefore, there is a single threshold in regression ⅴ. for regression vi, the p value of single-threshold test is 0.050, so the null hypothesis is rejected, indicating that there is a threshold. the p value of double-threshold test is 0.122, so the null hypothesis is accepted, indicating that there is no double threshold. therefore, regression vi also exists single threshold. the single-threshold likelihood ratio function diagrams of regression v and regression vi are shown in figure 1 and figure 2. figure 1. single-threshold likelihood ratio function diagram of regression v. figure 2. single-threshold likelihood ratio function diagram of regression vi. we use formula (5) to perform threshold effect regression, and get regression v in table 4, the results show that when the development level of financial technology is lower than 5.8790, the regression coefficient of financial technology to the big data industry (2.0884) is significantly positive at significance level of 1%, indicating that financial technology can significantly promote the development of the big data industry. when the development level of financial technology is higher than 5.8790, the regression coefficient of financial technology to the big data industry (2.1684) is 0 2 4 6 8 1 0 l r s ta ti s ti c s mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 156 significantly positive at significance level of 1%, indicating that financial technology can more significantly promote the development of the big data industry. it shows that there is a bottleneck in financial technology's direct promotion of the development of the big data industry. when the development level of financial technology breaks through this bottleneck, the direct effect of financial technology on the big data industry is significantly enhanced. therefore, we accept the null hypothesis h4a: financial technology directly promotes the development of the big data industry with a positive threshold effect. table 4. regression results of threshold effect. variable regression ⅴ regression ⅵ coefficient t value p value coefficient t value p value lag1 -0.0413 -0.80 0.423 -0.0237 -0.45 0.654 fintech (fintech≤γ1) 2.0884*** 3.40 0.001 fintech (fintech＞γ1) 2.1684*** 3.55 0.001 sa (fintech≤γ2) -0.1761*** -3.01 0.003 sa (fintech＞γ2) -0.0362 -0.58 0.560 pgdp 1.2243** 1.98 0.049 0.6450 1.30 0.196 pti -2.7037 -1.41 0.162 -1.2723 -1.53 0.128 rds 0.2532 0.67 0.502 0.2317 0.59 0.558 npg 0.2176 1.05 0.294 0.4759** 2.35 0.020 dou 0.2764** 2.08 0.039 0.5457*** 3.61 0.000 gc -0.4650 -1.53 0.128 -0.3860 -1.41 0.162 _cons -16.1841*** -3.98 0.000 -8.9509*** -2.66 0.009 threshold 5.8790 [5.8665, 5.8978] 5.4328 [5.3962, 5.4353] ①***p<0.01, **p<0.05, *p<0.1. ②lag1 corresponds to the lagging period of the explained variable. ③the confidence interval is shown in square brackets. we use equation (6) to perform threshold effect regression, and get regression ⅵ in table 4. the results show that when the development level of financial technology is lower than 5.4328, the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative at significant level of 1%, indicating that the alleviation of big data enterprise financing constraints can significantly promote the development of the big data industry. when the development level of financial technology is higher than 5.4328, the regression coefficient of big data enterprise financing constraints to the big data industry is negative, but not significant, indicating that the relief of big data enterprise financing constraints no longer significantly promotes the development of the big data industry. it shows that there is also a bottleneck in financial technology’s indirectly promotion the development of the big data industry by alleviating big data enterprise financing constraints, when the development level of financial technology breaks through this bottleneck, although the indirect effect of financial technology on the big data industry is enhanced, there are many internal and external causes of big data enterprise financing constraints [33], leading to the result that the increase in the indirect effect of financial technology on the big data industry may be smaller than the increase in the direct effect of financial technology on the big data industry . there are many internal and external causes of big data enterprise financing constraints [33]. the mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 157 increase in the indirect effect of financial technology on the big data industry may be less than the increase in the direct effect of financial technology on the big data industry, that is, compared with the direct effect, the indirect effect of financial technology on the big data industry is relatively weakened. therefore, we reject the null hypothesis h4b: financial technology indirectly promotes the development of the big data industry by alleviating the financing constraints of big data companies, and there is a positive threshold effect. 4.4. robustness inspection in order to test the robustness of the regression results of the dynamic panel data model, this article starts from the econometric model and uses the static panel data model to re-estimate the coefficients of formulas (1)-(6). regression ⅰ shows that the regression coefficient of financial technology to the big data industry is significantly positive at significance level of 5%. regression ⅱ shows that the regression coefficient of financial technology to big data enterprise financing constraints is significantly negative at significance level of 5%. regression iii shows that the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative at significance level of 10%. regression iv shows that the regression coefficient of financial technology to the big data industry is significantly positive at significance level of 5%, while the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative at significance level of 10%. regression v shows that when the development level of financial technology is lower than 5.7323, the regression coefficient of financial technology to the big data industry (2.0363) is significantly positive at significance level of 5%. when the development level of financial technology is higher than 5.7323, the regression coefficient of financial technology to the big data industry (2.1143) is significantly positive at significance level of 5%. the regression vi shows that when the development level of financial technology is lower than 5.2972, the regression coefficient of big data enterprise financing constraints to the big data industry is significantly negative at significance level of 5%. when the development level of financial technology is higher than 5.2972, the regression coefficient of big data enterprise financing constraints to the big data industry is negative, but not significant. comparing the coefficient estimation results of the dynamic panel data model and the static panel data model, it can be found that the regression coefficient signs of important variables are the same, and the significance is slightly different, indicating that the regression results of the dynamic panel data model are relatively robust. 5. conclusions, discussions and policy recommendations 5.1. conclusions (1) results of regression i (benchmark regression) show that the regression coefficient of the development level of financial technology (fintech) to the development level of the big data industry (dlbdi) is 1.7595, and the p value is 0.0590<0.10, indicating that under significance level of 10%, if fintech increase by 1%, dlbdi will increase by 1.7595%. the regression coefficient of the dynamic lag item (lag1) to the development level of the big data industry (dlbdi) is -0.1198, and the p value is 0.3480, indicating that lag1 has no significant effect on dlbdi. financial technology can directly promote the development of the big data industry, but the dynamic lag effect of the big data industry is not yet obvious. mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 158 among the control variables, the regression coefficient of the proportion of the regional tertiary industry (pti) to the development level of the big data industry (dlbdi) is 2.7963, and the p value is 0.0210<0.05, indicating that under significance level of 5%, if pti increase by 1%, dlbdi will increase by 2.7963%. the regression coefficient of regional r&d investment intensity (rds) to dlbdi is 1.6306, and the p value is 0.0240<0.05, indicating that under significance level of 5%, if rds increase by 1 %, dlbdi will increase by 1.6306%. the regression coefficient of the ratio of regional export value to total industrial output value (dou) to dlbdi is 0.6120, and the p value is 0.0170<0.05, indicating that under significance level of 5%, if dou increase by 1%, dlbdi will increase by 0.6120%. the regression coefficient of the proportion of government funds in the regional r&d expenditure (gc) to dlbdi is 0.4335, and the p value is 0.0620<0.10, indicating that under significance level of 10%, if gc increase by 1%, dlbdi will increase by 0.4335%. the regression coefficient of the regional per capita gdp (pgdp) on dlbdi is 0.8583, and the p value is 0.3150, indicating that pgdp has no significant impact on dlbdi. the regression coefficient of the number of regional patent grants (npg) to dlbdi is 0.1293, and the p value is 0.6960, indicating that npg has no significant impact on dlbdi. the basic support capacity for the development of the big data industry, the regional technological innovation capacity, the degree of openness, and the degree of government connection can significantly promote the development of the big data industry to varying degrees. (2) the regression ⅱ shows that the regression coefficient of the development level of financial technology (fintech) on big data enterprise financing constraints (sa) is -4.0716, and the p value is 0.0710<0.10, indicating that if fintech increase by 1%, sa will fell by 4.0716%. financial technology can effectively alleviate big data enterprise financing constraints, which is like the research results of literature [4] and [12]. regression iii shows that the regression coefficient of sa to the development level of the big data industry (dlbdi) is -0.1078, and the p value is 0.0500=0.05, indicating that under significance level of 5%, if sa drop by 1%, dlbdi will increase by 0.1078%. the alleviation of big data enterprise financing constraints will promote the development of the big data industry, which is like the research results of literature [14], [15] and [16]. the results of regression ⅰ, regression ⅱ and regression ⅲ show that big data enterprise financing constraints have an intermediary effect, and financial technology can indirectly promote the development of the big data industry by alleviating big data enterprise financing constraints. regression iv shows that the regression coefficient of the development level of financial technology (fintech) to the development level of the big data industry (dlbdi) is 1.8812, and the p value is 0.0120<0.05, indicating that under significance level of 5%, if fintech increase by 1%, dlbdi will increase by 1.8812%. at the same time, the regression coefficient of big data enterprise financing constraints (sa) to dlbdi is -0.1194, and the p value is 0.0360<0.05, indicating that under significance level of 5%, if sa decrease by 1%, dlbdi will increase by 0.1194%. big data enterprise financing constraints have a partial intermediary effect, and the intermediary effect accounts for 27.63% of the total effect. (3) regression v shows that when the development level of financial technology (fintech) is lower than 5.8790, the regression coefficient of fintech to the development level of the big data industry (dlbdi) is 2.0884, and the p value is 0.001<0.01, indicating that under significance level of 1%, if fintech increase by 1%, dlbdi will increase by 2.0884%. when fintech is higher than 5.8790, the regression coefficient of fintech to dlbdi is 2.1684, and the p value is 0.001<0.01, indicating that under significance level of 1%, if fintech increase by 1%, dlbdi will increase by 2.1684%. there is a positive mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 159 threshold effect in the direct promotion of the development of the big data industry by financial technology, which is like the research results of the literature [17]. the regression vi results show that when the financial technology development level (fintech) is lower than 5.4328, the regression coefficient of the big data enterprise financing constraint (sa) on the big data industry development level (dlbdi) is -0.1761, and the p value is 0.003<0.01, indicating that under significance level of 1%, if sa drop by 1%, dlbdi will increase by 0.1761%. when fintech is higher than 5.4328, the regression coefficient of sa to dlbdi is -0.0362, and the p value is 0.560, indicating that the impact of sa on dlbdi is not significant. financial technology indirectly promotes the development of the big data industry by alleviating big data enterprise financing constraints, and there is a reverse threshold effect. 5.2. discussions we select relevant data from 30 provinces (autonomous regions and municipalities) in china from 2013 to 2019, and use dynamic panel data models, intermediary effect testing methods, and threshold effect models to empirically test the direct promotion of financial technology to the development of the big data industry, the indirect promotion of financial technology to the development of the big data industry by alleviating big data enterprise financing constraints, and the threshold effect of financial technology in promoting the development of the big data industry. compared with literature [4] and [12], this article extends the research on financial technology to ease enterprise financing constraints by studying how it promote industrial development. compared with literature [14], [15] and [16], this article expands the research on alleviating enterprise financing constraints and promoting industrial development to alleviating enterprise financing constraints and promoting industrial development through financial technology. compared with the literature [17], this article expands the intermediary effect and the threshold effect of financial technology on economic growth by improving the efficiency of financial resource allocation to the intermediary effect and threshold effect of financial technology that affect the development of the industry by alleviating enterprise financing constraints. this paper analyzes the impact of financial technology on the development of the big data industry and its mechanism in depth, which is of great significance for clarifying the interaction mechanism between financial technology and the big data industry and accelerating the integrated development of the finance and big data industry. due to the difficulty of data acquisition and the limitation of cognition, many shortcomings still exist, which need to be improved in the future. future research directions are as follows: first, establishing a comprehensive evaluation index system for the development level of regional financial technology, and using a comprehensive evaluation method based on fuzzy sets to evaluate the development level of regional financial technology. second, establishing a complete evaluation of the development level of the regional big data industry, which uses a comprehensive evaluation method based on fuzzy sets to evaluate the development level of regional big data. third, exploring the construction of a more scientific and reasonable method for measuring the financing constraint level of regional big data companies, and improving the convincing power of the empirical analysis results. fourth, testing the normality of the panel data. if the panel data is skewed normal, use the skewed normal panel data model for empirical analysis. fifth, selecting more control variables that have a stronger impact on the development of the big data industry to improve the convincing power of the empirical analysis results. sixth, considering the spatial spillover effects of financial technology, and mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 160 establishing a spatial measurement model to measure the direct, indirect, and total effects of financial technology on the big data industry. 5.3. policy recommendations based on the research conclusions above, combined with the spirit of relevant departmental documents, the following policy recommendations are put forward: (1) in terms of direct effects, first, encourage financial institutions to actively use big data, artificial intelligence, and other technologies to deeply analyze customer financial needs and create smart financial products and services. second, promote the transformation of traditional financial entity outlets to marketing and experience smart financial outlets, and improve the operating efficiency of financial outlets. third, speed up the improvement of the credit process and credit evaluation models of enterprises in key areas. fourth, optimize the mobile payment technology architecture system and increase technology-enabled payment services. fifth, improve the financial business risk prevention and control system. sixth, actively explore the innovation of financial big data application and further expand the demand for big data technology products and technical services in the financial industry. (2) in terms of indirect effects, first, accelerate the application of mobile internet, big data, cloud computing, internet of things, artificial intelligence, and other emerging cutting-edge technologies in the financing of big data companies, and adhere to the balanced development of online and offline, and provide multi-level and all-round financing services for big data companies. second, vigorously promote the construction of credit information sharing platforms to realize cross-level, crossdepartment, and cross-regional interconnection, improve the acquisition, analysis, and application capabilities of big data enterprise financing-related data, and reduce information asymmetry, reduce financing costs, improve financing efficiency, and effectively alleviate the financing constraints faced by big data companies. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] sun xuran, wang kangshi, wang jinfeng. fintech, competition and bank credit structure: based on the perspective of sme financing[j]. journal of shanxi university of finance and economics, 2020, 42(6): 5972. [2] jin hongfei, li hongji, liu yinlu. financial technology, banking risks and market crowding effect[j]. financial research, 2020, 46(5): 1-14. [3] liao min. the current status and future trends of global financial technology regulation[j]. new finance, 2016(10): 12-16. [4] wei chenglong, guo feinan. research on fintech innovation and mitigation of corporate financing constraints——based on the calculation and empirical analysis of fintech index[j]. price theory and practice, 2020(1): 163-166. [5] zhu junjie, wang yanxi, zhang zeyi. the influence of the development of financial technology on the upgrading of my country's industrial structure[j]. science and technology management research, 2017, 37(19): 31-37. mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 161 [6] chen deyu, tang yonggang, zhang shaohe. empirical analysis of industrial structure transformation and upgrading, financial technology innovation and regional economic development[j]. science and technology management research, 2018, 38(15): 105-110. [7] zheng nanlei. fintech: the commanding heights of the future development of the financial industry[j]. securities market herald, 2017(1): 1-6. [8] chishti s, barberis j. the fin-tech book: the financial technology handbook for investors, entrepreneurs and visionaries[m]. chichester, uk: john wiley & sons, ltd, 2016. [9] arner d, et al. the development of financial technology: a new model after the financial crisis[j]. new finance, 2018(5): 9-15. [10] qiao haishu, huang jianxuan. research on financial technology development momentum index[j]. financial forum, 2019, 24(3): 64-80. [11] yosepha s y. the role of fintech encourages the export of small medium enterprises in indonesia[j]. journal of social and development sciences, 2018, 9(3): 66-77. [12] huang rui, lai xiaobing, tang song. how does financial technology affect corporate financing constraints? — — dynamic effects, heterogeneity characteristics and macro-micro mechanism inspection[j]. international finance research, 2020(6): 25-33. [13] yin z, gong x, guo p, et al. what drives entrepreneurship in digital economy? evidence from china[j]. economic modelling, 2019, 82: 66-73. [14] rajan r g, zingales l. financial dependence, and growth[j]. american economic review, 1998, 88(3): 559586. [15] xie qiaoxin, zhang yu. an empirical analysis of the impact of financing constraints and fdi inflows on industrial growth: taking zhejiang province as an example[j]. technical economy, 2014, 33(12): 54-58+79. [16] yang yuanyuan, yu jinping, yang dongxu. are financing constraints hindering the high-end of strategic emerging industries? [j]. economic review, 2018(5): 60-74. [17] tian xinmin, zhang zhiqiang. financial technology, resource allocation efficiency and economic growth——analysis based on the threshold effect of china's financial technology[j]. statistics and information forum, 2020, 35(7): 25-34. [18] bai zhonglin. econometric analysis of panel data[m]. tianjin: nankai university press, 2008. [19] pang lujing, zhang mu. research on the coordinated development of china's provincial big data industry and financial agglomeration coupling [j]. science and technology management research, 2020, 40(15): 209-216. [20] tang zhongfu, gu peiliang. comprehensive evaluation of the development level of high-tech industry[j]. economic theory and economic management, 2003(10): 23-28. [21] yang jianhua, lu bo. research on the evaluation index system for the development of regional electronic information industry[j]. journal of irbin university of technology (social science edition), 2005(4): 56-60. [22] ren weihong, huang lucheng. the evaluation model and application of the development level of r&d industry based on entropy[j]. technological progress and countermeasures, 2009, 26(3): 109-111. [23] wang hailong, liu jia. evaluation and model analysis of the development level of high-tech industries in my country's provinces and regions[j]. technological progress and countermeasures, 2011, 28(22): 113-117. [24] qiu wanhua. management decision and applied entropy[m]. beijing: machinery industry press, 2002. [25] yue chaoyuan. decision theory and method[m]. beijing: science press, 2003. [26] kaplan s, zingales l. do investment-cash flow sensitivities provide useful measures of financing constraints? [j]. quarterly journal of economics, 1997, 112(1): 169-215. [27] hadlock c j, pierce j r. new evidence on measuring financial constraints: moving beyond the kz index[j]. review of financial studies, 2010, 23(5): 1909-1940. [28] harrison a e, mcmillan m s. does direct foreign investment affect domestic credit constraints? [j]. journal of international economics, 2003, 61(1): 73-100. [29] wang feibo, xie qiaoxin. fdi inflow, financing constraints and export spillover effects[j]. science and technology management research, 2014, 34(22): 97-101. [30] shen junxin, li shuang, zhang jingyang. research on the factors influencing the development capability of big data industry——based on the fsqca method[j]. science and technology management research, 2019, 39(7): 140-147. [31] fu yongping. an empirical study on the key factors influencing the development of china's strategic emerging industries[j]. journal of donghua university (natural science edition), 2019, 45(4): 605-611. mu zhang, cheng cao and zhiyuan lv / journal of risk analysis and crisis response, 2021, 11(4), 146-162 doi: https://doi.org/10.54560/jracr.v11i4.309 162 [32] fan xiaoli, huang lingxiang, lu jing, wang liyan. agglomeration development of strategic emerging industries and analysis of influencing factors[j]. statistics and decision, 2017(14): 139-143. [33] ma shuqin, wang jianghang. a review of frontier research on financing constraints and heterogeneous enterprise exports[j]. international trade issues, 2014(11): 164-176. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). study on meteorological service policy for agricultural insurance in hebei province under the background of climate change kaicheng xing 1,2 shujun guo 3,* 1. key laboratory of meteorological and ecological environment of hebei province, shijiazhuang 05002, china 2. hebei climate center, shijiazhuang 050021, china 3. hebei meteorological bureau, shijiazhuang 050021,china received october 23, 2018 accepted december 22, 2018 abstract hebei province is a region sensitive to global climate change. under the background of climate change and frequent extreme climate events, existing agricultural production structure is facing enormous risk of climate disasters. in order to reduce the risk level of agricultural meteorological disasters, it is imperative to implement effective agro-insurance meteorological services. this paper analyses the facts of climate change, the trend and influence of extreme weather and climate events in recent years. based on the simulation results of climate models under moderate and high emission scenarios, the distribution characteristics of mean annual and seasonal air temperature in hebei province in the first 50 years of the 21st century are given, and the main problems in the practice of agricultural insurance in hebei province are pointed out. this paper also puts forward some countermeasures and suggestions for optimizing the agricultural insurance policy and improving the meteorological service effect of agricultural insurance. key words: climate change, insurance, meteorology, policy 气候变化背景下河北省农业保险气象服务对策研究 邢开成 1,2 郭树军 3,* 1.河北省气象与生态环境重点实验室，石家庄 050021，中国 2.河北省气候中心，石家庄 050021，中国 3.河北省气象局，石家庄 050021，中国 摘要：河北省是对气候变化非常敏感的区域，在气候变化和极端气候事件频发的背景下，现有农 业生产结构面临着巨大的气象灾害风险，为降低农业气象灾害风险水平，实施有效的农业保险气 象服务已经势在必行。本文在分析近年来河北省气候变化事实、极端天气气候事件趋势及影响的 基础上，基于中等和高排放情景下气候模式模拟结果，给出了河北省 21 世纪前 50 年的年、季平 均气温均变化趋势分布特征，并针对河北省农业保险实践中存在的主要问题，提出了优化农业保 险政策，提高农业保险气象服务效果的对策建议。 关键词：气候变化；保险；气象；对策 *corresponding author: gsjun888@sina.com. post address: hebei meteorological bureau, no.178 tiyu south street , shijiazhuang, china this study is supported by the national key research and development program of china (no. 2018yfa0606302). 1.引言 河北作为农业大省，2012-2016 年河北粮 食产量年均增速为 1.65%，高于全国平均水平 0.52 个百分点。2016 年河北粮食产量在全国 排第七位，较 2012 年前移一位，仅低于黑龙 江、河南、山东、吉林、四川、江苏 6 省，高 出第八名安徽 42.7 万吨，比第六名江苏少 5.8 万吨。2016 年，河北人均粮食占有量达到 463.22 千克，比全国平均水平高 17.54 千克， 比世界平均水平高约 65 千克（数据来源于国 家统计局河北调查总队）。在全球变暖气候背 景下，极端天气气候事件多发、频发，现有农 业生产结构所面临的气象灾害风险越来越大 [1]。河北省所在的环渤海地区属于对气候变化 响应非常敏感的区域[2]，2018 年 8 月，受台风 “温比亚”影响，山东部分地区和河北东部沿 海县市因强降水导致内涝严重，许多农户大田 作物绝收、蔬菜大棚损毁，农业保险赔付和灾 后重建等问题受到社会各界的广泛关注，也被 媒体和广大网友热议。河北省农业保险气象服 务方式的针对性、服务内容精细化、服务水平 的专业化与社会需求还有很大的差距[3]。提高 农业大灾保险气象服务保障水平，事关现代农 业健康可持续发展和农民增收、社会和谐稳定， 不仅是技术问题，也是经济问题，更是社会问 题。本文开展河北省气候变化事实和高温干旱 等极端气候事件的分析，并对农业气象灾害、 农业大灾保险气象服务有关政策、措施、技术 等进行研究论述，并提出农业保险优化工作对 策建议。 2.河北省气候变化事实及农业气象灾害极端 性影响 河北省地处温带大陆性季风气候区，地貌 多样，四季分明，寒暑悬殊，雨热集中，农业 气象资源丰富。但年降水量时空分布极不均匀， 降水变率大，多雨年和少雨年降水量相差可达 4~ 5 倍甚至更多，致使境内经常出现旱涝灾 害[2]。河北省干旱、洪涝、风雹、小麦干热风、 冻害、寡照等农业气象灾害类型复杂、发生频 繁、影响范围广、灾害损失重，成为威胁粮食 安全、影响现代农业可持续发展的重要因素。 统计分析显示，在全球气候变暖背景下， 近 50 年河北省各地平均气温上升显著，全省 平均每 10 年升高 0.24℃，是全球地表平均温 度升高速率（0.12℃/10a）的两倍，其中冬季 （12 月-来年 2 月）增温最为显著，高达 0.45℃ /10a（图 1） 1961-2015 年，河北省大部分地区年平均 最高气温为上升趋势，河北省中部、廊坊北部 及以北地区升温速率在 0.20℃/10a 以上，北部 部分地区超过 0.30℃/10a（图 2）。20 世纪 90 年代以来，全省各年代平均高温日数均超过常 年平均值，90 年代末至 2010 年日最高气温突 破历史极值的范围较前期明显增多，其中有 6 年出现 14-43 个县突破历史极值（图 3、图 4）。 图 1 全球、中国及河北历年平均气温及变化趋势图（单位：℃） 36 journal of risk analysis and crisis response vol. 9(1), march (2019), pp. 36–42 doi: https://doi.org/10.2991/jracr.b.190328.004; eissn: 2210-8505, issn: 2210-8491 https://www.atlantis-press.com/journals/jracr © 2019, the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). study on meteorological service policy for agricultural insurance in hebei province under the background of climate change kaicheng xing 1,2 shujun guo 3,* 1. key laboratory of meteorological and ecological environment of hebei province, shijiazhuang 05002, china 2. hebei climate center, shijiazhuang 050021, china 3. hebei meteorological bureau, shijiazhuang 050021,china received october 23, 2018 accepted december 22, 2018 abstract hebei province is a region sensitive to global climate change. under the background of climate change and frequent extreme climate events, existing agricultural production structure is facing enormous risk of climate disasters. in order to reduce the risk level of agricultural meteorological disasters, it is imperative to implement effective agro-insurance meteorological services. this paper analyses the facts of climate change, the trend and influence of extreme weather and climate events in recent years. based on the simulation results of climate models under moderate and high emission scenarios, the distribution characteristics of mean annual and seasonal air temperature in hebei province in the first 50 years of the 21st century are given, and the main problems in the practice of agricultural insurance in hebei province are pointed out. this paper also puts forward some countermeasures and suggestions for optimizing the agricultural insurance policy and improving the meteorological service effect of agricultural insurance. key words: climate change, insurance, meteorology, policy 气候变化背景下河北省农业保险气象服务对策研究 邢开成 1,2 郭树军 3,* 1.河北省气象与生态环境重点实验室，石家庄 050021，中国 2.河北省气候中心，石家庄 050021，中国 3.河北省气象局，石家庄 050021，中国 摘要：河北省是对气候变化非常敏感的区域，在气候变化和极端气候事件频发的背景下，现有农 业生产结构面临着巨大的气象灾害风险，为降低农业气象灾害风险水平，实施有效的农业保险气 象服务已经势在必行。本文在分析近年来河北省气候变化事实、极端天气气候事件趋势及影响的 基础上，基于中等和高排放情景下气候模式模拟结果，给出了河北省 21 世纪前 50 年的年、季平 均气温均变化趋势分布特征，并针对河北省农业保险实践中存在的主要问题，提出了优化农业保 险政策，提高农业保险气象服务效果的对策建议。 关键词：气候变化；保险；气象；对策 *corresponding author: gsjun888@sina.com. post address: hebei meteorological bureau, no.178 tiyu south street , shijiazhuang, china this study is supported by the national key research and development program of china (no. 2018yfa0606302). 1.引言 河北作为农业大省，2012-2016 年河北粮 食产量年均增速为 1.65%，高于全国平均水平 0.52 个百分点。2016 年河北粮食产量在全国 排第七位，较 2012 年前移一位，仅低于黑龙 江、河南、山东、吉林、四川、江苏 6 省，高 出第八名安徽 42.7 万吨，比第六名江苏少 5.8 万吨。2016 年，河北人均粮食占有量达到 463.22 千克，比全国平均水平高 17.54 千克， 比世界平均水平高约 65 千克（数据来源于国 家统计局河北调查总队）。在全球变暖气候背 景下，极端天气气候事件多发、频发，现有农 业生产结构所面临的气象灾害风险越来越大 [1]。河北省所在的环渤海地区属于对气候变化 响应非常敏感的区域[2]，2018 年 8 月，受台风 “温比亚”影响，山东部分地区和河北东部沿 海县市因强降水导致内涝严重，许多农户大田 作物绝收、蔬菜大棚损毁，农业保险赔付和灾 后重建等问题受到社会各界的广泛关注，也被 媒体和广大网友热议。河北省农业保险气象服 务方式的针对性、服务内容精细化、服务水平 的专业化与社会需求还有很大的差距[3]。提高 农业大灾保险气象服务保障水平，事关现代农 业健康可持续发展和农民增收、社会和谐稳定， 不仅是技术问题，也是经济问题，更是社会问 题。本文开展河北省气候变化事实和高温干旱 等极端气候事件的分析，并对农业气象灾害、 农业大灾保险气象服务有关政策、措施、技术 等进行研究论述，并提出农业保险优化工作对 策建议。 2.河北省气候变化事实及农业气象灾害极端 性影响 河北省地处温带大陆性季风气候区，地貌 多样，四季分明，寒暑悬殊，雨热集中，农业 气象资源丰富。但年降水量时空分布极不均匀， 降水变率大，多雨年和少雨年降水量相差可达 4~ 5 倍甚至更多，致使境内经常出现旱涝灾 害[2]。河北省干旱、洪涝、风雹、小麦干热风、 冻害、寡照等农业气象灾害类型复杂、发生频 繁、影响范围广、灾害损失重，成为威胁粮食 安全、影响现代农业可持续发展的重要因素。 统计分析显示，在全球气候变暖背景下， 近 50 年河北省各地平均气温上升显著，全省 平均每 10 年升高 0.24℃，是全球地表平均温 度升高速率（0.12℃/10a）的两倍，其中冬季 （12 月-来年 2 月）增温最为显著，高达 0.45℃ /10a（图 1） 1961-2015 年，河北省大部分地区年平均 最高气温为上升趋势，河北省中部、廊坊北部 及以北地区升温速率在 0.20℃/10a 以上，北部 部分地区超过 0.30℃/10a（图 2）。20 世纪 90 年代以来，全省各年代平均高温日数均超过常 年平均值，90 年代末至 2010 年日最高气温突 破历史极值的范围较前期明显增多，其中有 6 年出现 14-43 个县突破历史极值（图 3、图 4）。 图 1 全球、中国及河北历年平均气温及变化趋势图（单位：℃） journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 36 37 20 世纪 80 年代至 90 年代，重旱发生频率明 显增加，干旱化趋势较显著（图 5）。近 55 年 来，部分地区暴雨日数也呈增加趋势（0.010.21d/10a），主要分布在张家口西北部、承德中 部、河北省南部的山前平原等地区（图 6）。 从 20 世纪 80 年代至 2013 年，河北省暴 雨洪涝灾害的人口脆弱性（受灾人口/总人口） 增加，由 80 年代的 1~2%增加至 2000 年以来 的 2~5%；经济脆弱性（直接经济损失/gdp） 表现为先增加后减少的趋势，20 世纪 80 年代 图 2 河北省 1961-2015 年平均最高气温变化趋势空间分布 图 3 河北省历年高温日数变化图（红线为常年（1981-2010 年）高温日数） 图 4 河北省历年日最高气温超过历史极值的站数 经济脆弱性为 0.25~0.5%，90 年代增加至 0.5~1.0%，2000 年以来又降至 0.25%以内；干 旱灾害的人口脆弱性明显持续增加，由 80 年 代脆弱性不足 2%，持续增加至 2000 年以来的 10~15%；干旱灾害的经济脆弱性先增加后减 少（主要得益于良种优化、耕作和灌溉技术水 平的提高），20 世纪 80 年代经济脆弱性仅为 0.05~0.1%，90 年代增至 0.2~0.5%，2000 年以 来又降至 0.10~0.20%；低温冷害的人口脆弱性 和经济脆弱性均表现为持续增加趋势，低温冷 害人口脆弱性在 20 世纪 80 年代的不足 1%， 2000 年以来其脆弱性增加至 2~4%，经济脆弱 性由 80 年代的 0.016~0.05%增加至 2000 年以 来的 0.05~0.07%[1]。 20 世纪 80 年代以来，河北省气温显著上 升、降水减少、水资源匮乏、气候干化突出， 致使气象灾害频繁发生，气候灾害每年给河北 带来巨大经济损失，90 年代后平均每年达上 百亿元，且呈逐年增加趋势。2009 年 6 月 20 日～ 7 月 5 日，石家庄市连续 15 天最高气 温超过 35℃，日平均气温达 38.3℃，比常年偏 高 6.2℃，且其中 3 天超过 40℃，其高温持续 时间之长创石家庄气象站 1955 年建站以来的 最长记录。2015 年 6 月至 8 月，河北省 11 个 地市 133 个县（市）不同程度遭受旱灾，农作 物受灾面积 771000 公顷，直接经济损失达 45.75 亿元。2016 年河北省干旱、风雹、寡照 等农业气象灾害发生总面积 2062.9 万亩，种 植业直接经济损失约 72.6 亿元。其中，2016 年 7 月 19 日特大洪涝灾害受灾面积 1335.5 万亩， 图 5 河北省历年平均每站发生的重旱日数及 1980-1999 年重旱日数变化趋势 图 6 河北省 1961-2015 年暴雨日数变化趋势 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! 暴雨日数变化趋势(天/10年) ! -0.32 -0.20 ! -0.19 0 ! 0.01 0.10 ! 0.11 0.21 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 38 20 世纪 80 年代至 90 年代，重旱发生频率明 显增加，干旱化趋势较显著（图 5）。近 55 年 来，部分地区暴雨日数也呈增加趋势（0.010.21d/10a），主要分布在张家口西北部、承德中 部、河北省南部的山前平原等地区（图 6）。 从 20 世纪 80 年代至 2013 年，河北省暴 雨洪涝灾害的人口脆弱性（受灾人口/总人口） 增加，由 80 年代的 1~2%增加至 2000 年以来 的 2~5%；经济脆弱性（直接经济损失/gdp） 表现为先增加后减少的趋势，20 世纪 80 年代 图 2 河北省 1961-2015 年平均最高气温变化趋势空间分布 图 3 河北省历年高温日数变化图（红线为常年（1981-2010 年）高温日数） 图 4 河北省历年日最高气温超过历史极值的站数 经济脆弱性为 0.25~0.5%，90 年代增加至 0.5~1.0%，2000 年以来又降至 0.25%以内；干 旱灾害的人口脆弱性明显持续增加，由 80 年 代脆弱性不足 2%，持续增加至 2000 年以来的 10~15%；干旱灾害的经济脆弱性先增加后减 少（主要得益于良种优化、耕作和灌溉技术水 平的提高），20 世纪 80 年代经济脆弱性仅为 0.05~0.1%，90 年代增至 0.2~0.5%，2000 年以 来又降至 0.10~0.20%；低温冷害的人口脆弱性 和经济脆弱性均表现为持续增加趋势，低温冷 害人口脆弱性在 20 世纪 80 年代的不足 1%， 2000 年以来其脆弱性增加至 2~4%，经济脆弱 性由 80 年代的 0.016~0.05%增加至 2000 年以 来的 0.05~0.07%[1]。 20 世纪 80 年代以来，河北省气温显著上 升、降水减少、水资源匮乏、气候干化突出， 致使气象灾害频繁发生，气候灾害每年给河北 带来巨大经济损失，90 年代后平均每年达上 百亿元，且呈逐年增加趋势。2009 年 6 月 20 日～ 7 月 5 日，石家庄市连续 15 天最高气 温超过 35℃，日平均气温达 38.3℃，比常年偏 高 6.2℃，且其中 3 天超过 40℃，其高温持续 时间之长创石家庄气象站 1955 年建站以来的 最长记录。2015 年 6 月至 8 月，河北省 11 个 地市 133 个县（市）不同程度遭受旱灾，农作 物受灾面积 771000 公顷，直接经济损失达 45.75 亿元。2016 年河北省干旱、风雹、寡照 等农业气象灾害发生总面积 2062.9 万亩，种 植业直接经济损失约 72.6 亿元。其中，2016 年 7 月 19 日特大洪涝灾害受灾面积 1335.5 万亩， 图 5 河北省历年平均每站发生的重旱日数及 1980-1999 年重旱日数变化趋势 图 6 河北省 1961-2015 年暴雨日数变化趋势 ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !! ! ! ! ! ! ! 暴雨日数变化趋势(天/10年) ! -0.32 -0.20 ! -0.19 0 ! 0.01 0.10 ! 0.11 0.21 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 38 39 直接经济损失 42.2 亿元。 气候模式模拟显示，未来 40 年河北省年 平均气温将继续上升，到 21 世纪末温度将会 大幅增暖，降水可能趋于增加。在中等（rcp4.5： 预计 2100 年辐射强迫稳定在 4.5 w/m2）和高 等（rcp8.5：2100 年辐射强迫上升至 8.5 w/m2） 排放情景下，预估 21 世纪前 50 年河北省年、 季平均气温均呈上升趋势，年平均气温上升速 率为 0.25~0.41℃/10a，与 1986—2005 年平均 相比，21 世纪 20 年代河北省年平均气温可能 变化范围为-0.15~0.65℃，30 年代可能变化范 围为 0.24~1.03℃，40 年代可能变化范围为0.12~0.63℃，21 世纪前 50 年河北省季平均气 温冬季升温速率最大，春季最小（图 7）；降水 量年际波动较大，中等和高等排放情景下，总 体呈增加趋势，增速分别为 1.0%~2.2%/10a， 0.6%~1.7%/10a（图 8）。 综上所述，在全球气候变暖背景下，河北 省农业气象灾害的极端性有显著增强的趋势， 高附加值现代农业的脆弱性和暴露度不同程 度提高[3]，灾害传导链越来越长、越来越复杂， 同等强度农业气象灾害所造成的损失越来越 大[4-8]。在干化加重的情况下，引发大面积的干 旱灾害发生的可能性加大的同时，自然降水更 可能以短历时强降水的方式出现，不但容易造 成局地山洪地质灾害和城市渍涝成灾，而且雨 水的有效利用率低，降水停止后，由于气温高、 土壤蓄积水分少、蒸发快，会很快转变为干旱 灾害。 3.农业保险在抗灾救灾中发挥的作用 图 7 在 rcp4.5（中等）和 rcp8.5（高等）排放情景下，河北省历年地表气温距平（相对于 19862005 年）预估数据图，图中蓝线和红线是区域模式下预估的地表气温；绿线和紫线是全球气候模式预估 的地表气温。 图 8 在 rcp4.5（中等）和 rcp8.5（高等）情景下，河北省历年降水量距平百分比（相对于 19862005 年）预估数据图。 河北省做为主要农作物和养殖业政策性 农业保险重点省份，农业保险规模和覆盖范围 不断扩大，市场主体不断增多，保险密度和深 度不断提高，经济保障功能越来越明显。各级 政府大力推进，相关保险机构及时对受灾农户 基于保险金补偿，充分保障农民利益，对农民 抗灾减灾、生产自救、恢复重建和社会和谐稳 定起到了重要作用。 2012-2016 年，河北省政策性农业保险共 为 5877.60 万户次参保农户提供了 2304.82 亿 元风险保障，实现签单保费收入 91.78 亿元； 已决赔款金额达 49.93 亿元，简单赔付率达 54.40%，受益农户达 866.14 万户次，户均赔款 576.43 元。其中，2015 年，河北全省种植业投 保 7090.68 万亩，其中小麦 2506.78 万亩、玉 米 4223.00 万亩、棉花 166.69 万亩、水稻 51.51 万亩、大豆 3.18 万亩、花生 14.75 万亩、油菜 1.82 万亩、马铃薯 118.18 万亩、甜菜 4.77 万 亩。养殖业投保 1138.38 万头，其中能繁母猪 90.39 万头、奶牛 65.72 万头、育肥猪 982.27 万头。全省森林投保 3404.91 万亩，其中公益 林 2249.49 万亩、商品林 1155.42 万亩。全省 设施农业投保 1.3 万亩，其中蔬菜日光温室 0.18 万亩、蔬菜塑料大棚 0.24 万亩、蔬菜塑料 中小棚 0.88 万亩。综合投保率为 62.88%。农 险经保机构共处理玉米保险理赔案件 3.31 万 件，赔款金额 10.97 亿元，受益农户达 174.19 万户次（数据来源于河北省保监局）。 2016 年全省干旱、风雹、寡照等农业气象 灾害发生总面积 2062.9 万亩，种植业直接经 济损失约 72.6 亿元，尤其是“7·19” 特大洪 涝灾害受灾面积 1335.5 万亩，直接经济损失 42.2 亿元。截至 2016 年底，农险经办机构共 赔付暴雨灾害损失 3.89 亿元，其中种植险赔 付 3.49 亿元，农房保险赔付 0.30 亿元（数据 来源于河北省农业厅），在此巨灾下保险补偿 金也一定程度上解决了部分农户的燃眉之急。 4.农业保险存在的主要问题 河北省地貌独特，气候类型多样，特色农 业种类丰富、分布广泛。另外，环京津的区位 优势，使得设施农业、错季蔬菜等高附加值的 现代农业成为河北省农业产业结构特征重点 发展领域。极端气候事件和频繁的气象灾害加 剧了农业的脆弱和高风险[9-11]。目前河北省农 业保险“高成本、高风险、高赔付和低收费、 低保障、低保额”的特点，使得农业保险的指 数化趋势对特色农业、设施农业大灾农业保险 的现实作用越来越凸显。农业气象指数保险有 利于减少或消除传统农业保险的道德风险和 信息不对称引起的逆选择等问题，可以降低管 理成本，便于转移保险市场风险，特别适合地 域性强、 对气象灾害高度敏感、物化成本和产 业附加值高、对农民增收关系重大、农户投保 热情高的特色农业和设施农业，有助于实现政 府化解系统性风险、激励农户加强生产管理、 推进农业保险的资本化运作和现代化进程、推 动农业保险市场的创新与健康发展等多重政 策目标[12-15]。 但是，目前在河北省农业气象保险产品推 广过程中存在着农户保险意识不强，特色农产 品气象指数保险产品单一，保险范围、保险额 度吸引力不大，市县级地方政府对政策性农业 保险补贴的财政压力大等等问题。现行的政策 性农业保险“低保障、广覆盖”原则，主要针 对国家粮食安全保障，一旦出现重大气象灾害， 目前各保险公司推广的以物化成本为主要标 的物的政策性指数保险和普通灾害险，赔付保 障对于大灾、巨灾后生产自救和恢复重建来讲 却是“杯水车薪”，更不能解决农民致富、特色 农业升级等问题。 5 农业保险优化工作对策建议 5.1 健全大灾农业保险工作体制机制 各级地方政府做好制度性安排，积极参与 保险业务体系建设，进一步加大对气象指数农 业保险的补贴投入力度，在税费等方面对承保 企业给予更多的优惠政策；省金融办组织建立 由多家保险公司共同经营的大灾农业保险 “共保体”制度模式，引进退出机制，健全保 险联席会议和沟通协调长效工作机制，有利于 充分发挥政府和监管部门的职责，落实有关特 色农业保险各项财政补贴和惠民政策；充分发 挥政府、气象部门、保险公司、新型农业经营 主体的积极性，推进更加市场化，更加透明、 多元的农业保险环境，增强农业气象灾害防御 和抗灾自救互救保障能力，促进现代农业保险 服务的健康发展。 5.2 做好大灾农业保险气象服务 气象部门利用双重管理体制的行业特点， 省市县三级上下联动，尤其是农业大市、农业 大县，对地方支柱型特色农业产业在大灾农业 保险气象保障服务工作中，主动与地方政府和 农保经办企业对接，投入更多人才、利用更多 先进技术，结合需求设计更为科学、适用的农 业保险产品，在特色农业气象指数保险产品试 点、推广、运行、评估、认证等环节提供技术 支持以及精细化、针对性强的全程技术支撑。 配合做好基于影响的气象灾害风险区划、灾害 评估、灾害风险预警、气象灾情调查鉴定、气 象防灾减灾科普宣传等全程专业化气象保障 服务。 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 40 直接经济损失 42.2 亿元。 气候模式模拟显示，未来 40 年河北省年 平均气温将继续上升，到 21 世纪末温度将会 大幅增暖，降水可能趋于增加。在中等（rcp4.5： 预计 2100 年辐射强迫稳定在 4.5 w/m2）和高 等（rcp8.5：2100 年辐射强迫上升至 8.5 w/m2） 排放情景下，预估 21 世纪前 50 年河北省年、 季平均气温均呈上升趋势，年平均气温上升速 率为 0.25~0.41℃/10a，与 1986—2005 年平均 相比，21 世纪 20 年代河北省年平均气温可能 变化范围为-0.15~0.65℃，30 年代可能变化范 围为 0.24~1.03℃，40 年代可能变化范围为0.12~0.63℃，21 世纪前 50 年河北省季平均气 温冬季升温速率最大，春季最小（图 7）；降水 量年际波动较大，中等和高等排放情景下，总 体呈增加趋势，增速分别为 1.0%~2.2%/10a， 0.6%~1.7%/10a（图 8）。 综上所述，在全球气候变暖背景下，河北 省农业气象灾害的极端性有显著增强的趋势， 高附加值现代农业的脆弱性和暴露度不同程 度提高[3]，灾害传导链越来越长、越来越复杂， 同等强度农业气象灾害所造成的损失越来越 大[4-8]。在干化加重的情况下，引发大面积的干 旱灾害发生的可能性加大的同时，自然降水更 可能以短历时强降水的方式出现，不但容易造 成局地山洪地质灾害和城市渍涝成灾，而且雨 水的有效利用率低，降水停止后，由于气温高、 土壤蓄积水分少、蒸发快，会很快转变为干旱 灾害。 3.农业保险在抗灾救灾中发挥的作用 图 7 在 rcp4.5（中等）和 rcp8.5（高等）排放情景下，河北省历年地表气温距平（相对于 19862005 年）预估数据图，图中蓝线和红线是区域模式下预估的地表气温；绿线和紫线是全球气候模式预估 的地表气温。 图 8 在 rcp4.5（中等）和 rcp8.5（高等）情景下，河北省历年降水量距平百分比（相对于 19862005 年）预估数据图。 河北省做为主要农作物和养殖业政策性 农业保险重点省份，农业保险规模和覆盖范围 不断扩大，市场主体不断增多，保险密度和深 度不断提高，经济保障功能越来越明显。各级 政府大力推进，相关保险机构及时对受灾农户 基于保险金补偿，充分保障农民利益，对农民 抗灾减灾、生产自救、恢复重建和社会和谐稳 定起到了重要作用。 2012-2016 年，河北省政策性农业保险共 为 5877.60 万户次参保农户提供了 2304.82 亿 元风险保障，实现签单保费收入 91.78 亿元； 已决赔款金额达 49.93 亿元，简单赔付率达 54.40%，受益农户达 866.14 万户次，户均赔款 576.43 元。其中，2015 年，河北全省种植业投 保 7090.68 万亩，其中小麦 2506.78 万亩、玉 米 4223.00 万亩、棉花 166.69 万亩、水稻 51.51 万亩、大豆 3.18 万亩、花生 14.75 万亩、油菜 1.82 万亩、马铃薯 118.18 万亩、甜菜 4.77 万 亩。养殖业投保 1138.38 万头，其中能繁母猪 90.39 万头、奶牛 65.72 万头、育肥猪 982.27 万头。全省森林投保 3404.91 万亩，其中公益 林 2249.49 万亩、商品林 1155.42 万亩。全省 设施农业投保 1.3 万亩，其中蔬菜日光温室 0.18 万亩、蔬菜塑料大棚 0.24 万亩、蔬菜塑料 中小棚 0.88 万亩。综合投保率为 62.88%。农 险经保机构共处理玉米保险理赔案件 3.31 万 件，赔款金额 10.97 亿元，受益农户达 174.19 万户次（数据来源于河北省保监局）。 2016 年全省干旱、风雹、寡照等农业气象 灾害发生总面积 2062.9 万亩，种植业直接经 济损失约 72.6 亿元，尤其是“7·19” 特大洪 涝灾害受灾面积 1335.5 万亩，直接经济损失 42.2 亿元。截至 2016 年底，农险经办机构共 赔付暴雨灾害损失 3.89 亿元，其中种植险赔 付 3.49 亿元，农房保险赔付 0.30 亿元（数据 来源于河北省农业厅），在此巨灾下保险补偿 金也一定程度上解决了部分农户的燃眉之急。 4.农业保险存在的主要问题 河北省地貌独特，气候类型多样，特色农 业种类丰富、分布广泛。另外，环京津的区位 优势，使得设施农业、错季蔬菜等高附加值的 现代农业成为河北省农业产业结构特征重点 发展领域。极端气候事件和频繁的气象灾害加 剧了农业的脆弱和高风险[9-11]。目前河北省农 业保险“高成本、高风险、高赔付和低收费、 低保障、低保额”的特点，使得农业保险的指 数化趋势对特色农业、设施农业大灾农业保险 的现实作用越来越凸显。农业气象指数保险有 利于减少或消除传统农业保险的道德风险和 信息不对称引起的逆选择等问题，可以降低管 理成本，便于转移保险市场风险，特别适合地 域性强、 对气象灾害高度敏感、物化成本和产 业附加值高、对农民增收关系重大、农户投保 热情高的特色农业和设施农业，有助于实现政 府化解系统性风险、激励农户加强生产管理、 推进农业保险的资本化运作和现代化进程、推 动农业保险市场的创新与健康发展等多重政 策目标[12-15]。 但是，目前在河北省农业气象保险产品推 广过程中存在着农户保险意识不强，特色农产 品气象指数保险产品单一，保险范围、保险额 度吸引力不大，市县级地方政府对政策性农业 保险补贴的财政压力大等等问题。现行的政策 性农业保险“低保障、广覆盖”原则，主要针 对国家粮食安全保障，一旦出现重大气象灾害， 目前各保险公司推广的以物化成本为主要标 的物的政策性指数保险和普通灾害险，赔付保 障对于大灾、巨灾后生产自救和恢复重建来讲 却是“杯水车薪”，更不能解决农民致富、特色 农业升级等问题。 5 农业保险优化工作对策建议 5.1 健全大灾农业保险工作体制机制 各级地方政府做好制度性安排，积极参与 保险业务体系建设，进一步加大对气象指数农 业保险的补贴投入力度，在税费等方面对承保 企业给予更多的优惠政策；省金融办组织建立 由多家保险公司共同经营的大灾农业保险 “共保体”制度模式，引进退出机制，健全保 险联席会议和沟通协调长效工作机制，有利于 充分发挥政府和监管部门的职责，落实有关特 色农业保险各项财政补贴和惠民政策；充分发 挥政府、气象部门、保险公司、新型农业经营 主体的积极性，推进更加市场化，更加透明、 多元的农业保险环境，增强农业气象灾害防御 和抗灾自救互救保障能力，促进现代农业保险 服务的健康发展。 5.2 做好大灾农业保险气象服务 气象部门利用双重管理体制的行业特点， 省市县三级上下联动，尤其是农业大市、农业 大县，对地方支柱型特色农业产业在大灾农业 保险气象保障服务工作中，主动与地方政府和 农保经办企业对接，投入更多人才、利用更多 先进技术，结合需求设计更为科学、适用的农 业保险产品，在特色农业气象指数保险产品试 点、推广、运行、评估、认证等环节提供技术 支持以及精细化、针对性强的全程技术支撑。 配合做好基于影响的气象灾害风险区划、灾害 评估、灾害风险预警、气象灾情调查鉴定、气 象防灾减灾科普宣传等全程专业化气象保障 服务。 journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 40 41 5.3 加强特色农业气象指数保险市场化运作 在政策性农业保险中引入市场机制，通过 财税政策、激励措施鼓励商业保险公司推出不 同档位的保险业务产品，针对各地实际情况因 地制宜推进特色农业气象指数保险商业化和 市场化，实现农业保险的分层布局和多元运作。 积极探索推进农产品价格指数保险、产量保险、 收入保险等农业保险产品，推动气象灾害风险 管理和防灾减灾上新水平[10]，为现代农业保险 提供重大气象灾害保险转移市场支撑，做到承 保机构的保险产品“广覆盖、深介入、高保障”， 相关技术服务单位“全流程、多层次、强支撑”， 农户“买得起、用得上、靠得住”。 “适销对路”的农业保险产品是农民抵御 气象灾害的一道坚固防线，是美丽乡村建设的 重要保障，农保市场广阔、意义重大。做好大 灾农业保险是现代保险服务业的发展方向，不 仅是经济问题，更是事关乡村振兴、精准扶贫、 农业生产安全和农村社会稳定的政治问题。需 要各相关部门和单位共同努力，协同配合，为 提高农业保险服务保障水平，增强农业气象灾 害防御能力作出应有贡献。 参考文献： 1. 秦大河. 中国极端天气气候事件和灾害风险管 理与适应国家评估报告. 2015. 2. 张可慧. 全球气候变暖对京津冀地区极端天气 气候事件的影响及防灾减灾对策.干旱区资源与 环境,2011, 25(10) :122-125. 3.《河北省 2016 年气候变化监测公报》,河北省气候 中心,2017. 4. guo j, zhao s, huang c. valid historical data for probabilistic risk analysis in natural disaster. human and ecological risk assessment: an international journal, 2017, 23(3): 474-493. 5. 黄崇福,郭君,艾福利,等. 洪涝灾害风险分析的基 本范式及其应用. 自然灾害学报,2013,(4):13-25. 6. 黄崇福. 综合风险管理的地位、框架设计和多态 灾害链风险分析研究. 应用基础与工程科学学 报，2006, 增刊，29-37. 7. huang c, shi x. a discrete model of the expected loss for catastrophe insurance in natural disasters. journal of risk analysis and crisis response, 2011,1(1) :48-58 8. guo j, huang c. time limit of the probabilistic risk for natural disaster. journal of risk analysis and crisis response, 2017,7(3): 137–145. 9. 魏瑞江, 姚树然, 王云秀. 河北省主要农作物农 业气象灾害灾损评估方法,中国农业气象,2000, 21(1): 27-31. 10.王炜, 权循刚, 魏华. 从气象灾害防御到气象灾 害风险管理的管理方法转变 ,气象与环境学 报,2011,27(1):7-13. 11.郭丽丽. 农业气象灾害风险评估研究进展与展 望,农业科技与信息,2016, 11:52-53. 12.王春乙, 张继权, 霍治国,等. 农业气象灾害风险 评估研究进展与展望,气象学报,2015, 1:1-19. 13.张强, 韩兰英, 张立阳,等. 论气候变暖背景下干 旱和干旱灾害风险特征与管理策略,地球科学 进,2014,29(1) :80-91. 14.祁添垚, 张强, 孙鹏,等. 气候暖化对中国洪旱极 端事件演变趋势影响研究,自然灾害学报,2015,3: 143-152. 15.陈琳, 陈鑫, 陆鹏程,等. 农业气象保险在防御农 业气象灾害中的作用 ,现代农业科技 ,2016,7: 231-232. journal of risk analysis and crisis response vol. 9, no. 1 (march 2019) 36–42 42 microsoft word volume 12, issue 3-4 journal of risk analysis and crisis response, 2022, 12(3), 143-151 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i3.335 143 review poly vinyl chloride additives and applications a review abdallah s. elgharbawy 1,2,* 1 materials science department, institute of graduate studies and research (igsr), alexandria university, alexandria (21526), alexandria province, egypt 2 the egyptian ethylene and derivatives company (ethydco), alexandria (21526), alexandria province, egypt * correspondence: ab_pet_88@hotmail.com received: february 23, 2022; accepted: april 16, 2022; published: september 30, 2022 abstract: a pure polyvinyl chloride (pvc) is a white, brittle material and it is the third-largest polymers produced after polyethylene and polypropylene as 40 million tons of pvc are produced yearly. the basic structure of pvc is (c2h3cl)n and it is produced by polymerization of the vinyl chloride monomer (vcm) with a polymerization degree ranges from 300 to 1500. the chlorine content in pvc is about 57% by weight, which makes it less dependent on hydrocarbon content. in this paper, we are going to reveal the pvc additives and applications. keywords: poly vinyl chloride; pvc application; pvc additives 1. introduction there are two types of pvc which are rigid and flexible. the rigid pvc is a type of pvc that do not contain plasticizers and used in the construction activities. whereas, the flexible type is made by adding plasticizers to the rigid form and can replace rubber. flexible pvc can be used in plumbing, electrical cable insulation, imitation leather, and flooring [1, 2]. rigid pvc is very important type of pvc as it is used in the building construction, piping, signs, roofing sheet material and other products [3]. the types of pvc that are more flexible serve as an alternative to rubber and are widely used in the manufacturing of clothing, footwear/boot grades, upholstery, electrical cable compounds, and many other everyday products [4]. pvc is thermoplastic with a low thermal stability, so the use of pvc polymer is limited in the industry that requires high temperature [5, 6]. neat pvc polymer is a brittle, not flexible material with limited commercial possibilities. in addition, the pvc processing in its raw form by using high temperature and pressure leads to severe degradation of the pvc structure. therefore, before using pvc, it must be treated with suitable additives. the most crucial additives are heat stabilizers and lubricants, fillers, processing aids, impact modifiers, pigments, and plasticizers [7, 8]. the role of additives is to improve the mechanical, electrical, thermal, light, color, and clarity properties of pvc. the additives are blended with the pvc via a compounding process [5, 9]. the resin of pvc is categorized by the k-value, which refers to the polymerization degree and the molecular weight [10]. pvc that has a k-value of 57 is a low molecular weight type and it is used abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 144 in packaging films, blow sheets, bottles and injection shapes. whereas, pvc that has a k-value of 7075 is a high molecular weight type has a high mechanical property but it is processed difficultly [11]. this type of pvc is used for flooring, rigid parts, high performance cables, pipes, and profiles. emulsion pvc is used in the form of plastisol or latex and it is used in special applications such as coatings, multilayer films, and battery separators [12, 13]. not only does pvc offer the flexibility necessary for medical uses such as blood bags and intravenous therapy (iv) containers, but it can also be relied upon for its strength and durability, even under different temperatures and conditions. pvc can also be easily processed to make iv tubing, thermoformed to make ‘blister’ packaging, or blow molding to make hollow rigid containers. this versatility is a major reason why pvc is the material of choice for medical product and packaging designers. in this paper, we will investigate the pvc additives, which are required to improve the pvc properties to be able to perform its functions in the dedicated applications. we will also mention the pvc applications and the mixing techniques. 2. materials and methods 2.1. dry mixing the pvc mixing is either dry mixing that uses a very high-speed mixer that blends all the component to produce powder [14, 15]. 2.2. extruder mixing extruder mixes the components using a low or high-speed extruder mixer. after that, the mixed components are transferred through heaters to melt the mixture. eventually, the mixed components are cooled and cut into granules for processing [16, 17]. 3. pvc additives 3.1. plasticizers at normal temperature, pvc is naturally rigid. this is due to the short distances between the molecules since there are strong intermolecular forces between them. when plasticizers are added to pvc, the plasticizer molecules make their way between the pvc molecules, preventing the pvc polymer molecules from getting closer to each other. therefore, the polymer molecules are kept away even at normal temperature and softness is achieved [13, 18]. 3.1.1. primary plasticizers primary plasticizers have a good dispersion inside the pvc resin and used in a large content such as 140-150 parts per hundred (phr) and it is mainly used for super soft products. all plasticizers are almost liquids and must be absorbed in a suspension resin in heated mixers. plasticizers are mixed with pvc resin in dry blend in high or low shear mixer [14, 15]. 3.1.2. secondary plasticizer abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 145 secondary plasticizers have a limited dispersion inside the pvc resin. the main use of secondary plasticizer is to decrease the pvc price [19]. chlorinated paraffin wax (cpw) is the most common secondary plasticizers that consist of high viscous wax with a 50 % of chlorine content. the other common secondary plasticizers are chlorinated paraffin oils (cpo) since its viscosity and plasticizing efficiency is much better than cpw [14]. table 1 summarizes the plasticizers types’ performance and specialty [8, 19]. 3.1.3. plasticizing performance the plasticizing performance of di, (2-ethyl hexyl) phthalate (dop) is 1. the plasticizing performance of any other plasticizer type is calculated as a ratio with dop for producing a molding of same softness [15, 20]. table 2 shows the plasticizing efficiency for many plasticizers. for example, if 100 phr of dop yields a hardness of 50, and 110 phr of another plasticizer gives the same hardness, the plasticizing efficiency of the plasticizer equals: = 100*(100/110) = 91 %. 3.1.4. plasticizers for medical applications dioctyl phthalate (dop) is the most common plasticizer used in medical industry due to its high efficiency, availability at high purity and low cost. other plasticizers such as tri, (2-ethyl hexyl) mellitate (tehm, also called totm), n-butyrul, tri n-hexyl citrate (bthc) and di (n-decyl), phthalate (dndp) is used for other medical applications such as the platelets storage [21, 22]. 3.2. stabilizers the pvc molecule is not stable at high temperature and ultra violet light. heating pvc leads to the polymer chains breaking and free the toxic hydrochloric acid gas. the released hcl causes the pvc degradation. the pvc exposure to the uv radiations dissociate the polymer chains but its impact is lesser than the heat degradation [22, 23]. 3.2.1. heat stabilizers pvc must be thermally stabilized during its processing at high temperature. most of heat stabilizers contain metal elements that interact with hcl and prevent pvc degradation. the most common heat stabilizers are metal salts, soaps. table 3 shows heat stabilizers types [24, 25]. 3.2.2. light stabilizers most of mixed metal stabilizers save pvc from the uv radiation attack. dbl phosphate is a light stabilizer that has some uv resistance properties. light stabilizers are complex chemicals, very expensive, and are efficient at low amount (0.1-0.3 phr) [24]. 3.3. fillers fillers are inorganic, inactive materials and its main role is to decrease the pvc processing cost and enhance the mechanical properties of pvc such as the fracture strength and impact performance of pvc. it can be used as a pigment and improve the pvc chemical resistance. the most common type of filler is calcium carbonate, titanium dioxide, talc, glass, and calcined clay [26, 27]. abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 146 3.3.1. nano fillers nano fillers are mixed with polymer from 1 to 10 wt. %. the most common nano fillers are nano fibers such as glass and carbon fibers. the other types of nano fillers are nano-clays, nano-oxides, carbon nanotubes. nano-clays fillers is a silicate of transition metals and it is natural or synthetic clays. nano metal oxide such as rutile tio2 is used as a white pigment [27, 28]. 3.4. impact modifiers impact modifiers (im) should have rubber properties such as acrylic rubber, chlorinated polyethylene, or butadiene styrene. im are mixed with pvc with an amount of 5-20 wt.% to give the pvc a toughening property [29, 30]. the aim of im is to enhance the pvc impact resistance and the role of impact modifiers molecules is to absorb the impact energy and eliminate the fracture of the pvc product. the particle size of im is in the micro range [31]. 3.5. pigments pvc pigments is a colored material that is insoluble in water generally is categorized into [32]:  inorganic.  organic. pigments for pvc must be thermally and light stable, have good dispersibility, and be compatible within the formulation. table 4 shows the different types of pigment [33, 39]. 4. pvc applications 4.1. piping applications pvc has been improved for many piping applications such as exterior body trim and molding, chemical resistant trim, gear shifter knobs, hoses, tubing, and interior console covers [14, 34]. 4.2. construction activities pvc has been used extensively in a wide range of construction products for more than 50 years. the pvc strength, lightweight, and durability make pvc perfect for many applications as outlined below [35, 36]:  window seals.  weather-strip and screen spline.  concrete water stops.  weatherable capstock for exterior trim.  dry wall accessories.  flooring.  roofing.  electrical insulators  sunroof 4.3. medical applications abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 147 pvc is used in the medical applications for more than 50 years due to its safety, chemical stabilization, low cost, durability, its resistance to chemical stress cracking, and biocompatibility [37, 38]. 4.4. figures, tables and schemes figure 1. extruder components. table 1. plasticizers types. family general purpose performance of plasticizers specialty plasticizers strong solvent low temperature low volatility low diffusion stability flame resistance phthalates p s s s s s trimellitates s p s aliphatic esters p polyesters p p epoxides s s p phosphates s s p extenders p miscellaneous p p p p: primary performance function. s: secondary performance function. abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 148 table 2. plasticizing efficiency for many plasticizers. plasticizer plasticizing efficiency di butyl phthalate 1.05 di octyl phthalate 1.00 di iso octyl phthalate 1.00 di nonyl phthalate 0.98 di iso decyl phthalate 0.95 table 3. heat stabilizers types. type heat stability main applications not recommended for leads very good cables, pipes, fittings, sleeves, profiles. clear application. can be toxic. tins best transparent tubing and sheeting, high quality pipes, medical. staining with leads cd-zn moderate electrical cables for high heat history ba-cd good leather cloth, calendared products, footwear. nontoxic applications metallic stearates low co-stabilizer with lubricating action sole stabilizer table 4. pigment types. inorganic pigments organic pigments type color type color ultramarine blue and violets phthalocyanines blue, green chromes yellows, orange chromophthals red, orange cadmiums reds, orange, yellow azos wide range iron oxide brown, black toners wide range abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 149 5. conclusions pvc is categorized into two types: rigid and flexible. neat pvc uses is negligible and need additives to be more appropriate for many applications such as bags, bottles, toys, construction, petroleum pipes, and medical applications. pvc is mixed with additives through two techniques: dry mixing or extruder mixing. pvc additives are various such as plasticizers, stabilizer, fillers, impact modifier, and pigments. many research is required to investigate the pvc industry and widen the uses of pvc. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] elgharbawy, a.s., a review on high density poly ethylene as engineering polymer. quaestus, 2021(18): p. 455-459. [2] allam, e.m., et al., rare earth group separation after extraction using sodium diethyldithiocarbamate/polyvinyl chloride from lamprophyre dykes leachate. materials, 2022. 15(3): p. 1211. doi: https://doi.org/10.3390/ma15031211. [3] patrick, s., practical guide to polyvinyl chloride. 2005: ismithers rapra publishing. [4] giacomucci, l., et al., polyvinyl chloride biodegradation by pseudomonas citronellolis and bacillus flexus. new biotechnology, 2019. 52: p. 35-41. [5] elgharbawy, a.s.a.a., review on corrosion in solar panels. international journal of smart gridijsmartgrid, 2018. 2(4): p. 218-220. [6] elgharbawy, a.s., et al., maximizing biodiesel production from high free fatty acids feedstocks through glycerolysis treatment. biomass and bioenergy, 2021. 146: p. 105997. doi: https://doi.org/10.1016/j.biombioe.2021.105997. [7] ding, l., et al., the photodegradation processes and mechanisms of polyvinyl chloride and polyethylene terephthalate microplastic in aquatic environments: important role of clay minerals. water research, 2022. 208: p. 117879. doi: https://doi.org/10.1016/j.watres.2021.117879. [8] choi, d.-s., et al., emotion-interactive empathetic transparent skin cushion with tailored frequencydependent hydrogel–plasticized nonionic polyvinyl chloride interconnections. chemical engineering journal, 2022: p. 136142. doi: https://doi.org/10.1016/j.cej.2022.136142. [9] elgharbawy, a. and a. sayed, production of biodiesel from used cooking using linear regression analysis. journal of petroleum and mining engineering, 2020. 22(2): p. 92-99. doi: https://doi.org/10.21608/jpme.2020.39252.1044. [10] mahmoud, k., et al., investigation of the gamma ray shielding properties for polyvinyl chloride reinforced with chalcocite and hematite minerals. heliyon, 2020. 6(3): p. e03560. [11] elgharbawy, a., a review on vinasse a by-product from sugarcane industry. trends petro eng, 2021. 1(2): p. 1-3. doi: https://doi.org/10.53902/tpe.2021.01.000506. [12] elgharbawy, a. and a. sayed, a review on natural gas previous, current and forecasting prices and demand. journal of petroleum and mining engineering, 2020. 22(1): p. 61-64. [13] xia, x., et al., polyvinyl chloride microplastics induce growth inhibition and oxidative stress in cyprinus carpio var. larvae. science of the total environment, 2020. 716: p. 136479. doi: https://doi.org/10.1016/j.scitotenv.2019.136479. [14] boyle, d., et al., polyvinyl chloride (pvc) plastic fragments release pb additives that are bioavailable in zebrafish. environmental pollution, 2020. 263: p. 114422. [15] farjami, m., v. vatanpour, and a. moghadassi, fabrication of a new emulsion polyvinyl chloride (epvc) nanocomposite ultrafiltration membrane modified by para-hydroxybenzoate alumoxane (phba) additive to improve permeability and antifouling performance. chemical engineering research and design, 2020. 153: p. 8-20. doi: https://doi.org/10.1016/j.cherd.2019.10.022. abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 150 [16] sun, y., et al., thermal behavior of the flexible polyvinyl chloride including montmorillonite modified with iron oxide as flame retardant. journal of thermal analysis and calorimetry, 2018. 131(1): p. 65-70. doi: https://doi.org/10.1007/s10973-017-6117-7. [17] bagherinia, m.a., m. sheydaei, and m. giahi, graphene oxide as a compatibilizer for polyvinyl chloride/rice straw composites. journal of polymer engineering, 2017. 37(7): p. 661-670. [18] gao, m., m. wan, and x. zhou, thermal degradation and flame retardancy of flexible polyvinyl chloride containing solid superacid. journal of thermal analysis and calorimetry, 2019. 138(1): p. 387-396. [19] zhang, y., et al., flotation separation of hazardous polyvinyl chloride towards source control of microplastics based on selective hydrophilization of plasticizer-doping surfaces. journal of hazardous materials, 2022. 423: p. 127095. [20] fang, l.-f., et al., improved antifouling properties of polyvinyl chloride blend membranes by novel phosphate based-zwitterionic polymer additive. journal of membrane science, 2017. 528: p. 326-335. doi: https://doi.org/10.1016/j.memsci.2017.01.044. [21] li, d., et al., effects of molecular design parameters on plasticizer performance in poly (vinyl chloride): a comprehensive molecular simulation study. chemical engineering science, 2022. 249: p. 117334. [22] zhang, h. and j. zhang, rheological behaviors of plasticized polyvinyl chloride thermally conductive composites with oriented flaky fillers: a case study on graphite and mica. journal of applied polymer science, 2022: p. 52186. [23] elmahdy, m.m., et al., tetrabutylammonium tetrafluoroborate electrolyte as poly (vinyl chloride-co-vinyl acetate-co-2-hydroxypropyl acrylate) plasticizer: thermal degradation and optical characteristics. 2022. [24] jubsilp, c., et al., effects of organic based heat stabilizer on properties of polyvinyl chloride for pipe applications: a comparative study with pb and cazn systems. polymers, 2021. 14(1): p. 133. doi: https://doi.org/10.3390/polym14010133. [25] yu, j., et al., thermal degradation of pvc: a review. waste management, 2016. 48: p. 300-314. [26] dutta, n., s. hazarika, and t.k. maji, study on the role of tannic acid–calcium oxide adduct as a green heat stabilizer as well as reinforcing filler in the bio-based hybrid polyvinyl chloride–thermoplastic starch polymer composite. polymer engineering & science, 2021. 61(9): p. 2339-2348. [27] deng, x., et al., crush behaviors of polyvinyl chloride cellular structures with liquid filler. composite structures, 2018. 189: p. 428-434. [28] mansour, s.a., r. elsad, and m. izzularab, dielectric spectroscopic analysis of polyvinyl chloride nanocomposites loaded with fe2o3 nanocrystals. polymers for advanced technologies, 2018. 29(9): p. 2477-2485. doi: https://doi.org/10.1002/pat.4359. [29] khosya, s., s. kushwaha, and a.a.d. dung, polyvinyl chloride (pvc) acting as occupational hazard in a factory worker presented with acute toxic encephalopathy: a case report. j clin toxicol, 2018. 8(400): p. 2161-0495.1000400. doi: https://doi.org/10.4172/2161-0495.1000400. [30] hammiche, d., et al., effects of types of pvc-g-ma on wettability and dynamical behavior of polyvinyl chloride/alfa composites. materials today: proceedings, 2021. 36: p. 10-15. [31] fakhri, m., e. shahryari, and t. ahmadi, investigate the use of recycled polyvinyl chloride (pvc) particles in improving the mechanical properties of stone mastic asphalt (sma). construction and building materials, 2022. 326: p. 126780. [32] kumar, s., et al., orange light spectra filtered through transparent colored polyvinyl chloride sheet enhanced pigment content and growth of arthrospira cells. bioresource technology, 2021. 319: p. 124179. [33] gangwar, a.k., k. nagpal, and b.k. gupta, triluminescent functional composite pigment for nonreplicable security codes to combat counterfeiting. chemistryselect, 2018. 3(33): p. 9627-9633. [34] wu, p., et al., adsorption mechanisms of five bisphenol analogues on pvc microplastics. science of the total environment, 2019. 650: p. 671-678. [35] mukhamediev, m. and d.z. bekchanov, new anion exchanger based on polyvinyl chloride and its application in industrial water treatment. russian journal of applied chemistry, 2019. 92(11): p. 14991505. [36] petrović, e.k. and l.k. hamer, improving the healthiness of sustainable construction: example of polyvinyl chloride (pvc). buildings, 2018. 8(2): p. 28. [37] feit, c.g., m.k. chug, and e.j. brisbois, development of s-nitroso-n-acetylpenicillamine impregnated medical grade polyvinyl chloride for antimicrobial medical device interfaces. acs applied bio materials, 2019. 2(10): p. 4335-4345. doi: https://doi.org/10.1021/acsabm.9b00593. abdallah s. elgharbawy / journal of risk analysis and crisis response, 2022, 12(3), 143-151 doi: https://doi.org/10.54560/jracr.v12i3.335 151 [38] beveridge, j.m., et al., covalent functionalization of flexible polyvinyl chloride tubing. langmuir, 2018. 34(35): p. 10407-10412. doi: https://doi.org/10.1021/acs.langmuir.7b03115. [39] elgharbawy, a.s., & ali, r.m. (2022). techno-economic assessment of the biodiesel production using natural minerals rocks as a heterogeneous catalyst via conventional and ultrasonic techniques. renewable energy. doi: https://doi.org/10.1016/j.renene.2022.04.020. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style research on initial trust model of mobile banking users baolin sun1, chaohao sun2, chang liu1, chao gui1 1school of information engineering, hubei university of economics, wuhan 430205, china, wuhan 430205, china e-mail: blsun@163.com, liu.chang@vip.126.com, gui_chao@126.com 2china huarong financial leasing co., ltd, hangzhou 310007, china e-mail: skypedestrian@163.com abstract the trust mechanism and trust model of mobile banking is especially urgent because of its high cognitive risk and poor experience. this article studies several key factors that affect the trust mechanism of mobile banking, such as structure assurance, ubiquity, information quality, initial trust, perceived ease of use and experience. the results show that structure assurance is a most important factor influencing trust tendencies, and ubiquity, information quality, perceived satisfaction and perceived ease of use are some other key factors that affect initial trust and experience. the initial trust directly affects the users' sense of experience of mobile banking and decides the tendency of trust, which affects the actual use. therefore, the mobile phone banks need to pay more attention to the initial sense of trust and experience to promote user acceptance and trust mobile banking services. keywords: mobile banking, structure assurance, initial trust, sense of experience, trust tendencies. 1. introduction mobile banking, also known as mobile bank, is a typical application in the mobile financial field. it refers to the integration of electronic money and mobile communication services through multi-industry and multi-platform cooperation between mobile telecom carriers and banking institutions. with the help of the fourth generation mobile communication technology (4g) for mobile phone users, it provided a new mode of financial payment services. mobile banking can provide bank account information inquiries, account transfers, bank and securities transfer, securities trading, payment, financial information, and many other financial services. the main advantage of mobile banking is that it’s “everywhere” as compared with internet banking (usually refers to the pc side), because the mobile banking service is mainly based on the binding of bank card and credit card on the basis of mobile phone number to complete the financial payment service. in other words, with the help of mobile communication platform, mobile banking payment system and smart phone, smartphone users can accomplish the financial payment service at any time and any place, thus promoting the use of mobile banking for smartphone users. however, smart phones also have limitations, such as the small screen, inconvenient input and slow processing speed, as well as the instability of the mobile operating system, mobile communication security loopholes, distrust between smart phone users and banks. in this distributed, dynamic and mobile financial application system is composed of multiple platforms, multiple service organizations and multiple technical services, the lack of a restrictive mechanism for financial institutions, mobile operators and smartphone users, and the lack of confidence among entities, the journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 13 received 18 february 2017 accepted 28 february 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). large number of fraudulent acts and untrusted services events, result in financial transactions among financial institutions, mobile operators and mobile phone users which those entities behaviors can be presented as dynamic, security and uncertainty, restricting the smart phone users on mobile banking acceptance and use. in the mobile communication environment, mobile phone user’s trust, banking services, 4g communication technology play an important role in the success of mobile banking. therefore, the establishment of an effective trust mechanism is very helpful to assess the conduct of the credibility of the mobile banking financial security and credibility of great significance among financial institutions, mobile operators and smart phone users. according to “china’s e-banking survey in 2015”, it showed that in 2015, the proportion of individual online banking users was 35.6%, increased by 10% compared to 2014; personal mobile banking users ratio was 32% which increased by 14.5%; the proportion of mobile phone users was 88%, increasing by 5%; the proportion of mobile banking users was 58%, decreased by 18%. according to the theory of innovative diffusion, mobile banking has crossed the 'take-off point'". the report also pointed out that the ratio of use of mobile phone app orders, wechat, mobile payment alipay and other third-party payment is 31%, qr code scanning payment is 24%, smartphone wallet payment is 10 %, and mobile phone acoustic payment is 6%. the trend of mobile payment business is very obvious, once mobile phone users start using mobile banking, they will quickly transfer their bank account information inquiries, micropayment services, mobile securities, counter the business to the mobile phone bank, and gradually fall into a habit of using mobile banking. 2. mobile banking initial trust mechanism and model of the status quo in recent years, many foreign and domestic scholars have used different theories and methods to carry out effective research on trust-related issues, and put forward many trust evaluation models, social norms and other important trust structures. as for the extension of mobile banking, in previous research on the consumer demographic characteristics, we identified the role of mobile banking risk in ways of using mobile banking with or without trust mechanism. tian junfeng (2011) studied key mechanisms of the trust model, methods and existing problems such as a review of the trust model for further study to provide a good reference and guidance. the results of shaikh (2012) research showed that the technical structure of mobile banking, mobile communication security and ease of use, perceived usefulness, and quality of service on the use of mobile banking have an important role. fang zeyin (2012) analyzed the security factors of existing mobile banking security and its shortcomings, and put forward secure solutions for mobile banking security encryption card, thus establishing the theoretical foundation of financial security, business development, and large-scale application for mobile banking. david carf (2012), an american scholar, studied the operational risks of mobile banking, and put forward major operational risks: mobile phone operation risk, information asymmetry risk, credit risk, and mobility risk. through empirical research, afshan (2016) analyzed the security structure of mobile banking, the quality of mobile communications technology, making a significant contribution on the use of mobile banking. meanwhile, structural assurance is an important guarantee for mobile phone users to the initial trust of mobile banking, and the convenience of operation system will be more conducive to users of mobile banking. jiabao lin (2011) carried out research on mobile trust for mobile securities that impacts mobile phone investors to adopt securities trading activities, and pointed out that the initial trust of mobile securities is a trading determinant of mobile security. lin (2011) studied the trust and payment process of mobile banking with the theory of diffusion of innovation and trust theory. the results show that the relative advantage and perceived ability have a strong influence on mobile phone users' attitude towards using mobile banking. in addition, shuihong yao (2013) studied the relationship between the service quality of mobile banking and the continuous trust of customers, and established trust model between mobile banking service quality and customer continuous trust. the empirical test shows that the service quality of mobile banking to mobile banking customer's ongoing trust has an important role. the study of zhou (2012) points out that the impact of trust on mobile banking users' behavior is significant. in our previous study (2016), we analyzed several main factors that affect the quality of communication, selfefficacy, perceived risk, ubiquity, perceived satisfaction, structural assurance, initial trust, experience, and trust journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 14 tendencies of mobile banking, this paper presents a multi-dimensional trust mobile banking risk evaluation index system based on mobile banking customer direct trust, mobile banking customer trust decay, mobile banking financial business trust, and mobile banking customer final trust evaluation. 3. mobile banking multi-dimensional trust relationship structure and risk evaluation model 3.1. mobile banking experience in payment mobile banking payment experience is defined as a sense of the overall, which is the overall feeling when they use the mobile phone operators to pay for mobile banking. the mobile payment process is characterized by: (1) operations on the smart phone according to information related on the screen; (2) inner feelings and self-awareness of the operation; (3) operational awareness of the subjective initiative in payment; (4) enhanced awareness of the operation of the mobile payment system. the payment process also reflects the degree of responsiveness and the associated balance between mobile phone users and the financial system. 3.2. the relationship between mobile banking multi-dimensional trust structures fig. 1 shows the structure of mobile banking multidimensional trust relationship. the main influencing factors are structure assurance, ubiquity, information quality, perceived satisfaction and perceived ease of use, and the influence of initial trust and experience. smartphone users' experience perception can determine the tendency of trust in a better way. fig. 1. mobile banking multidimensional trust relationship structure. as an important and fundamental trust mechanism, the structure assurance can effectively establish the trust of the smart phone users for the financial payment and reduce the perceived risk of their online transactions. structure assurance means sufficient technical and legal structure to ensure that the mobile banking payment architecture and payment security structure in order to establish initial trust with the mobile phone users. structural security can also enable mobile phone users have better experience with mobile banking. ubiquity means that in the mobile terminal and the internet, smartphone users can access financial system at anytime, anywhere, so that the financial system can be presented to smart phone users services everywhere. information quality refers to the quality of the technology, skills and transmission data required by the mobile operators and the banking service providers to fulfill their payment processes. quality of information reflects the technical assurance provided by service providers, promise, and no deceit to users. the quality of information affects the initial trust and experience of smartphone users. in the mobile banking environment, user perceived satisfaction refers to the behavior or willingness of mobile phone users to participate in the mobile banking process and share the mobile banking. perceived satisfaction can improve the initial trust and experience. perceived ease of use mainly reflects the difficulty of smart phone users to use mobile banking to pay in the process. because of some limitations of smart phones, such as mobile phone small screen, multi-screen inconvenient operation and input, smartphone users cannot operate so easy as when they does with computer. perceived ease of use is conducive to improving the trust of mobile phone users and the use of mobile banking experience. trust reflects the relationship between mobile banking service providers and smartphone users, and common interests. initial trust also affects smartphone users' experience. trust provides a guarantee that mobile banking users buy mobile banking services provided, they also feel that the use of mobile banking is guaranteed and will have a better sense of experience. initial trust and experience will affect the using willingness of mobile banking users. in addition, initial trust mitigates some uncertainties and risks in the process of using the mobile banking payment and promotes the payment behavior of mobile banking structural assurances ubiquity information quality perceived satisfaction perceived ease of use initial trust experience perception experience perception journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 15 users, making the good sense of experience reach its culmination, and benefiting mobile phone banking user's in trust tendencies. 3.3. financial risk assessment model of multidimensional trust mobile banking 1. mobile banking customer direct trust model: in order to improve the accuracy of the trust evaluation index and dynamic adaptability, so a period of time the business is divided into several business, set t1, t2, ..., tn. in the nth business time, we assume that the number of times between the mobile banking service i and the mobile banking service j is m, then the direct trust degree dn(i, j) can be expressed as 1 ( , ) ( , ) , 0; 0, 0. m k n f i j d i j m m m =   = ≠  = ∑ (1) 2. mobile banking financial trust model: we use pt(i, j) as the trust that mobile banking services i to mobile banking financial services j, and consider the nth financial payment business factors, then pt(i, j) can be calculated as ptn(i, j) = αdn(i, j)+(1-α)rn(i, j), α∈[0, 1] (2) among them, α (0<α<1) is the trust adjustment factor, which is related to the degree of importance of direct trust and indirect trust. 4. empirical research and data analysis of mobile banking 4.1. collection of samples in this paper, 10 senior college students are invited to collect and distribute questionnaires to citizens who own smartphones. a total of 500 questionnaires were distributed, then 452 questionnaires were retrieved, and 28 filled-out cases were removed from the 15 offices of five banks and the six operating rooms of the three telecom-operators in wuhan optics valley. except for invalid and incomplete questionnaires, a total of 424 valid ones were collected. of these valid questionnaires, 45 were bank employees, 134 were employees, 127 were employees of institutions (including schools), and 86 were of other occupations. the distribution of the sample characteristics is shown in table 1. table 1. sample feature distribution table personnel structure option structure quantity proportion percentage (%) gender male 221 52.1 female 203 47.9 age <20 36 8.5 20-30 78 18.4 30-40 103 24.3 40-50 115 27.1 >50 92 21.7 education high school 42 9.9 junior college 102 24.1 undergraduate 219 51.6 graduate and above 61 14.4 occupation bank clerk 45 10.6 enterprise employee 134 31.7 civil servant 32 7.5 institution employee 127 29.9 others 86 20.3 monthly income <2000 35 8.3 2000-4000 53 12.5 4000-6000 121 28.5 6000-8000 128 30.2 >8000 87 20.5 4.2. the design of questionnaire the questionnaire consists of two parts: the first part is the demographic characteristics; the second part of the survey is the measured value of structure variables in the studying model. the questionnaire was distributed to experts in two mobile banking banks and two professors of financial research at universities to review and revise according to their opinions in china. table 2 is the main content of the questionnaire. 4.3. data analysis table 2 lists the distribution of the sample the following year, the main test of people's sex, age, education, occupation, and income and as such having impacts on trust, and the test of factors of table 1 in the mobile communication technology platform, mobile financial trading platform, and smart phones and as such having impacts on the trust. from the 424 questionnaires collected, we can see that men have better sense of experiences than women in the use of mobile banking for mobile financial transactions, and can feel that mobile banking brings fast, convenience, and safe in a better way. in terms of journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 16 age, because young people accept new technology very quickly and can experience mobile banking mobile banking financial transactions in a better way than older people do, the proportion of its use of mobile banking decreases as age increases. then speaking to the level of education, people with higher degrees of education have higher awareness of new technologies and the stronger the sense of experience than those with lower ones. in addition, from the perspectives of occupations, the results of use of mobile banking comparison are: corporate employee is of the highest, and then followed by others, institution employee, bank clerk, and civil servant. last but not least, in views of wage income, impacts of wage income on use of mobile banking experience are not as obvious as what have discussed above. table 2. part of the questionnaire. measurement factors measured values main content measurement factors measured values main content structure assurance 1 mobile banking financial trading system provides adequate technical assurance measures, allowing people to feel very comfortable in the process of financial transactions perceived satisfaction 1 the bank's reputation in mobile banking is very good 2 the bank of the mobile banking provides a good service 2 i think that telecom operators, mobile banking take measures to fully guarantee the use of mobile banking without technical problems 3 4g communication quality is very good, and mobile banking also provides good service 3 i believe that 4g communication technology, mobile banking authentication, data security and other technologies can guarantee security and reliability of mobile banking financial trading system perceived ease of use 1 mobile banking can provide a convenient mobile financial operation interface 2 mobile banking can provide convenient mobile phone safety certifications for users 4 i believe that 4g communication technology, mobile banking certification, data security and mobile banking financial transactions have legal guarantee mechanism initial trust 1 mobile banking has always been able to provide accurate and convenient financial transaction services 2 mobile banking can provide secure and reliable financial transaction services ubiquity 1 use of mobile banking can carry out financial transactions at any time, without time constraints 3 mobile banking has been able to provide safe, reliable, and fast financial transaction services 2 use of mobile banking can carry out financial transactions anywhere, regardless of time or geographical restrictions experience sense 1 i plan to open mobile banking services information quality 1 mobile banking can provide a secure, stable financial transaction information platform trust trend 1 i have opened mobile banking financial services 2 mobile banking can provide timely financial transaction information 2 i am ready to launch mobile banking services 3 mobile banking can provide reliable users’ authentications, data security, and timely financial transaction information journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 17 in order to test the influence of various factors on the tendency of trust and the moderating effect, we use structural equation model to analyze the adjustment of each factor and trust tendency. besides having direct effects on the initial trust of the mobile phone banks, mobile banking users' trust tendencies also play an important regulatory role on other factors such as structure assurance, ubiquity, information quality, initial trust, perceived ease of use, and so on. fig. 2 shows the relationship between trust tendency and structural assurance. as can be seen from the figure, structural assurance measures can be more perfect with better trust tendency. fig. 2. the relationship between trust orientation and structural assurance. fig. 3 shows the relationship between trust tendency and ubiquitous regulation, which can be seen from the figure that trust tendency are better with wider coverage area of 4g mobile communication under no restrictions of time for mobile payment. fig. 3. regulatory relationship between trust tendency and ubiquity. fig. 4 shows the relationship between trust tendency and information quality. it can be seen from the figure that the data transmission of 4g mobile communication technology fast, reliability, and data security plays an important role in trust tendency. fig. 4. the relationship between trust tendency and information quality. fig. 5 shows that the initial trust of mobile banking has a good perceptual effect on trust tendency, showing that it is very important for mobile banking to provide safe, reliable, and fast financial payment service. fig. 5. relationship between trust tendency and initial trust. fig. 6 shows that perceived satisfaction has a trust relationship with the trust tendency, which shows that service quality of mobile banking and of mobile telecom operators has a significant effect on user satisfaction. fig. 7 is the relationship between trust and the number of mobile banking payments, which can be seen from the figure, the more the number of trust payments, the higher the trust, indicating that their mobile financial payment security and reliability are better. 0 1 2 3 4 0 1 2 3 4 5 t rust t endency structure assurance 0 1 2 3 4 t rust t endency ubiquity 0 1 2 3 4 5 0 1 2 3 4 0 1 2 3 4 5 t rust t endency information quality t rust t endency initial trust 0 1 2 3 4 0 1 2 3 4 5 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 18 fig. 6. the relationship between trust inclination and perceived satisfaction. fig. 7. the relationship between trust and the number of payment. 5. conclusion and discussion this article studies several important factors that affect initial trust in mobile banking. according to the results of questionnaire survey and previous data analysis, there are some important conclusions which are: (1) structure of quality assurance and information are the most important factors that affect the emergence of initial trust in mobile banking, showing that mobile banking users will take a fancy to the architecture of mobile banking, trading platform, the service quality of mobile communication operators, and the security of financial transactions; (2) the ubiquity embodies the advantages of mobile financial transactions, and directly influences the mobile banking users of mobile operators, mobile banking confidence, and experience. ubiquity is able to pay for mobile banking users to provide timely financial activities, for example, when business trip or travel, mobile banking users can pay by mobile banking activities and query account information; (3) the structure of guarantee is the most significant factor that influences the mobile banking initial trust. when the mobile banking and mobile banking users are not in interaction, users will pay special attention to the security of mobile banking payment environment, the reliability of the data. therefore, mobile banking and mobile communication operators should provide adequate security and protection measures to reduce the risk of mobile banking users attention; (4) mobile banking users trust tendency not only has important and direct influence on the initial trust, but also adjust the influence of other factors, including the structure quality, perceived ease of use, and information on the influence of initial trust; (5) the effects of sense of satisfaction on initial trust of mobile banking users are not pronounced. a reasonable explanation may be that domestic mobile communication operators and the support of mobile phone bank scale are relatively large, then their reputation are better in return; (6) the quantity of users’ payments has a direct and important influence on mobile banking trust, meaning that trust of mobile banking increases as the quantity of payment goes up, so plays an important role on use and promotion of mobile banking. acknowledgements this article is supported by the national social science foundation of china (no. 14bjy171), the national natural science foundation of china (no. 61572012), the key natural science foundation of hubei province of china (no. 2014cfa055). references 1. china financial certification center. 2015 china ebanking survey. http://www.financialnews.com.cn/yw/jryw/201512/t2015 1207_88383.html, 2015-12-07. 2. jun-feng tian, and hong-yun cai. relationship between trust model and its development. journal of hebei university (natural science edition), 2011, 31(5): 555–560. 3. a. a. shaikh, h. karjaluoto. mobile banking adoption: a literature review. telematics and informatics, 2015, 32(1): 129–142 4. ze-nan fang, wei-gang liu, and guang-zhuo yu. study on the security of mobile banking. microcomputer and applications, 2013, 32 (24): 4-6. 5. david carfi and francesco musolino. a coopetitive approach to financial markets stabilization and risk management. communications in computer and information science, 2012, 300(4): 578–592. 20 40 60 80 100 0 20 40 60 80 100 t rust the number of payment direct trust business trust t rust t endency perceived satisfaction 0 1 2 3 4 0 1 2 3 4 5 journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 19 6. s. afshan, a. sharif. acceptance of mobile banking framework in pakistan. telematics and informatics, 2016, 33(2): 370–387. 7. jiabao lin and yaobin lu. study on initial trust model of mobile securities users. management review, 2011, 23 (11): 59-68. 8. hsiu-fen lin. an empirical investigation of mobile banking adoption: the effect of innovation attributes and knowledge-based trust. international journal of information management, 2011, 31(3): 252–260. 9. shuihong yao and zhenzhen chen. the impact of mobile banking service quality on continuous trust. economy and management, 2013, 27 (11): 78-82. 10. tao zhou. examining mobile banking user adoption from the perspectives of trust and flow experience. information technology and management, 2012, 13(1): 13:27–37. 11. baolin sun, chao gui, chang liu, xuan-yu chen, and chaohao sun. multi-dimensional trust evaluation of mobile banking risk index system. journal of risk analysis and crisis response, vol. 7, no. 1 (april 2017) 13–20 ___________________________________________________________________________________________________________ 20 1. introduction 2. mobile banking initial trust mechanism and model of the status quo 3. mobile banking multi-dimensional trust relationship structure and risk evaluation model 3.1. mobile banking experience in payment 3.2. the relationship between mobile banking multi-dimensional trust structures 3.3. financial risk assessment model of multidimensional trust mobile banking 4. empirical research and data analysis of mobile banking 4.1. collection of samples 4.2. the design of questionnaire 4.3. data analysis 5. conclusion and discussion acknowledgements references journal of risk analysis and crisis response vol. 10(1); april (2020), p. i doi: https://doi.org/10.2991/jracr.k.200415.003; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr introduction for volume 10, issue 1 on 22 january 2020, a statement from the world health organization (who) showed that there is a high risk of an outbreak of the novel coronavirus in china. to respond the emergency, on the same day, the journal of risk analysis and crisis response (jracr) decided to edit and publish a special issue on risk analysis and crisis management for epidemic diseases, meanwhile chongfu huang (professor of beijing normal university) and fengying zhang (deputy dean of the west china school of nursing at sichuan university) were assigned to be the guest editors. in order to share the papers for the special issue as soon as possible, they will be published in regular issues with a distinction of covid-19 topic. in this issue, two papers are in covid-19 topic. the first paper “the diamond princess cruise: an accidentally experimental model of virus pneumonia” by liyuan liu, introduces infection process and mechanism of covid-19 on the diamond princess cruise. the second paper “risk analysis and crisis management of viral pneumonia in wuhan” by weixi xu and xuanhua xu, attempts to use the wisdom of the super large groups in the social network, and will be divided into two parts to elaborate, exploring the group decision-making scheme under the big data and the risk analysis under different schemes. both of the contributions are written in chinese with english abstracts. other four papers, two contributions in english and two contributions in chinese with english abstracts, can be divided into four topics: fire risk assessment, risk decision making, earthquake risk analysis, and credit risk analysis. “discussing the need to manage uncertainty relating to users in road tunnel fire risk assessment” by panagiotis ntzeremes et al. aims to discuss current ways of representing users’ uncertainty as well as to introduce the importance of distinguishing these representations in the wider context of risk assessment in road tunnel fire. finite investment decision making using real market risk (non-diversifiable risk) was undertaken in paper “kernel density estimation of white noise for non-diversifiable risk in decision making” by emma anyika et al. to derive probability estimates of the non-diversifiable risks of the various stocks based on a finite data sample. the paper “review the parameters of historical strong earthquakes in tianshui and its surrounding areas” by qing wu, proposed an elliptical intensity distribution model based on modern events with both instrument records and macro investigation records that suitable for the western region of china, then estimated the historical strong earthquake parameters in tianshui and its surrounding areas. there is one paper in credit risk analysis. the paper “research on the contagion mechanism of associated credit risk in the supply chain” by xiaofeng xie et al. reveals the contagion mechanism of the associated credit risk in the supply chain based on multiple perspectives such as the origin, contagion path and basic characteristics of the associated credit risk in the supply chain, which is helpful for the supervision authorities to strengthen the management of the enterprise credit risk in the supply chain. we sincerely thank the referees for their strong support and kind help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, being technology editor, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time in overseeing the reviews. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). editors-in-chief: prof. chongfu huang beijing normal university, no. 19 xinjiekouwai street, beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina, regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca https://doi.org/10.2991/jracr.k.200415.003 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:hchongfu@126.com mailto:gordon.huang@uregina.ca microsoft word a systematic literature review on the credit risk management of big tech lending journal of risk analysis and crisis response, 2021, 11(3), 139-145 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i3.303 139 review a systematic literature review on the credit risk management of big tech lending mu zhang 1,* and cheng cao 1,2 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: october 24, 2021; accepted: october 27, 2021; published: october 30, 2021 abstract: this article reviews the relevant research of big tech lending credit risk management in order to promote the research on the theory and method of credit risk management of big tech lending. at present, relevant research results at home and abroad mainly concentrated on internet finance, credit risk analysis and control of digital inclusive finance and financial technology, and large-tech credit risk management framework research. under the background of financial technology, the research on the theory and method of credit risk management of big tech lending is in its infancy at home and abroad. the explanatory nature of the risk control methods of largescale technology credit is relatively low, resulting in the fact that government supervision department cannot well identify the stability of such loans, which is a key issue that needs to be solved urgently in the credit risk management of large-scale technology credit. the research directions of big tech lending credit risk management in the future include: big tech lending credit risk perception and its influencing factors, big tech lending credit risk control mechanism research and big tech lending credit risk early warning model research. keywords: smes; big tech lending; credit risk management; big tech ecosystem; big data risk control model; literature review 1. introduction small and medium-sized enterprises (smes) refer to enterprises established in accordance with the law within the territory of the people's republic of china with relatively small staff and operating scales, including medium-sized enterprises, small enterprises and micro-enterprises. private enterprises with small and medium-sized enterprises as the main body have occupied a very important position in the national economy, and their contributions to gdp growth, intellectual property rights and urban employment have exceeded 60%, 70% and 80% respectively. therefore, whether the "financing difficulties" of small and medium-sized enterprises can be effectively alleviated is not only related to the survival of micro-enterprises, but also affects the development of the macro-economy. there are many reasons for the "financing difficulties" of smes (demirgüç-kunt and klapper, 2012), which can be attributed to the difficulties of "acquiring customers" and "risk control" mostly. most smes are small in scale, large in number, and geographically dispersed, which increases the cost of acquiring customers for banks. at the same time, banks usually use three methods to manage corporate credit risk: financial data, mortgage assets, and "relational credit". mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 140 however, it is difficult for these methods to be applied to small and medium-sized enterprises on a large scale. in recent years, the rapid development of internet finance, digital inclusive finance, financial technology, and especially large-scale technology company credit (big tech lending) has brought dawn to the real solution to the problem of "financing difficulties" for smes. the so-called big tech lending mainly refers to the use of digital technology to provide loans by some large technology companies (frost et al., 2019). from the perspective of technological process, service scale and asset quality, big tech lending is likely to be a revolutionary financial innovation. the financial business model of big tech lending originated in china and has been implemented in many countries such as the united states, argentina, india, and south korea. according to data from the bank for international settlements (bis), the total credit of global financial technology and large technology companies reached nearly 800 billion us dollars in 2019. china, united states and united kingdom are the largest markets for financial technology credit (chen hua and li qingchuan, 2021). big tech lending has creatively formed an effective credit risk management framework. the two pillars of this framework are the big technology ecosystem based on the big technology platform and the big data risk control model (huang yiping and qiu han, 2021). the outstanding advantage of the big technology ecosystem is that it can cover most of the activities in the entire process from customer acquisition, loan application, risk assessment, monitoring to repayment management, thereby providing an effective way to reduce information asymmetry and control adverse selection and moral hazard. the massive digital footprint left in the big technology ecosystem is gathered to form big data. on one hand, these big data can be used for real-time monitoring to help the platform quickly adjust business strategies, on the other hand, it can support big data risk control models. this new type of risk control model based on big data and machine learning can not only predict the default rate more accurately and robustly (huang et al., 2020), but also provide credit services for many smes that lack bank credit history. it fully embodies the inclusive characteristics of big tech lending. the new credit risk management framework of big tech lending makes it possible for large-scale, low-cost, high-efficiency and sustainable smes loans. this article will review the current research status at home and abroad of internet finance, digital inclusive finance, financial technology credit risk analysis and control, and big tech lending risk management framework, with a view to promoting the research of big tech lending risk management theories and methods. the research in this article has important theoretical and practical significance for promoting the healthy development of big tech lending business and alleviating the "financing difficulties" of smes. 2. related research of internet finance, digital inclusive finance and financial technology compared with traditional financial credit risk, internet financial credit risk has three characteristics: credit risk is highly concealed, credit risk of a single platform spreads widely, and credit risk supervision is more difficult (li guoyi, 2017). short loan terms in the internet lending market can easily cause small and micro enterprises to face serious mismatches in investment and financing terms, resulting in higher credit risks (hu jinyan and shui bingbing, 2020). the main borrowers in the internet financial market are low-income groups with insufficient credit records, which will bring greater credit risk (graves, 2003; dobbie and skiba, 2013; li kemu, 2016), which brings huge risks to financial technology companies. operational risks may even affect financial mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 141 security and social stability. therefore, performing well in risk control and accurately assessing the credit of borrowers is of great significance for promoting the healthy and orderly development of the internet financial market (zhang yiwei and gao weihe, 2020). digital inclusive finance provides financial services through the internet, however, its impact on credit expansion cannot be underestimated due to wide coverage and large amount. in particular, current financial regulatory agencies are unable to accurately monitor the liquidity absorbed by digital finance and incorporate the risks of digital finance into the prudential regulatory framework for systemic risks in the entire financial market, leading to excessive credit expansion (wei xiaofeng, 2019). financing internet platforms have the dual business attributes of "information servicesfinancial services", and there are price transmission mechanisms and systemic risk channels between "money market-information market". therefore, digital financial supervision should take moral hazard and credit risk as focus. the extensive use of financial technology has not only improved the convenience and availability of financial services, but also lowered the barriers to entry for customers. it has introduced a large number of high-risk customers with varying qualifications, besides, credit information system, credit reporting system and credit reporting supervision is not perfect in china, financial technology is more likely to induce credit risk (liu mengfei, 2020). although financial technology adopts different technologies, causing credit risk to have many manifestations, its essence is still the risk caused by lack of information disclosure or default of market participants. when the economic situation declines, if there exists a large scale of failure to repay on time or even failure to repay, banks and other financial institutions will be affected, which will lead to large-scale credit problems (chen hong and guo liang, 2020). internet banking credit risk presents highly different risk characteristics from traditional banks in that: loan default rate is small, but potential risks are worthy of attention. the success of big data risk control requires cross-cycle and long-term calibration verification. post-loan management is difficult (research group of chengdu branch of the people's bank of china, 2020). although fintech 2.0 has financial attributes, it is dominated by technical attributes. therefore, the nature of fintech companies is a mixed business of technology companies and financial companies. for financial regulators, it is difficult to define the target of supervision. it is necessary to have a clear boundary between technology and finance, otherwise it will cause regulatory loopholes, duplicate supervision or regulatory mismatch, making it difficult to discover potential financial risks in time (meng xiantong, 2020). 3. related research of big tech lending the business model of the american fintech company nav can be summarized as: acting as a platform and channel to match service needs of small and micro enterprises with the products and services of credit bureaus and financial institutions to achieve one-stop service (liu xinhai, 2019). born in the internet boom, fintech companies in china are in a leading position of the world whether in terms of revenue scale, investment and financing scale, or the strength of individual companies. however, the development of the regulatory field is relatively lagging. the construction of an effective regulatory framework is still in the process of exploration (sun fei, 2019). from the involvement of technology companies in the financial field to the online transformation and development of traditional financial institutions, online credit has gradually evolved into two modes of weak-scene finance represented by p2p and strong-scene finance represented by e-commerce and mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 142 supply chain financing. a more inclusive, prudent, and appropriate regulatory environment will ensure a deeper integration of finance and technology (icbc online financing center, 2020). frost et al. (2019) proposed the concept of big tech lending, which mainly refers to the use of digital technology to provide loans by some large technology companies. huang yiping and qiu han (2021) further proposed the concept that big technology companies use the big technology ecosystem and the big data risk control model to provide credit services, innovate the credit risk management framework, and analyze the working mechanism of big tech lending on the credit risk management framework in the field of sme loans. on one hand, the big technology ecosystem based on the big technology platform achieves customer acquisition by connecting hundreds of millions of users, accumulates a large number of digital footprints, supports real-time monitoring and credit risk assessment, and designs corresponding incentive mechanisms to improve repayment management. gambacorta et al. (2019) found that the higher the importance of the network, the higher the loan amount that users can spend. huang et al. (2020) also found that the importance of borrowers in the ecological network can help predict their default. on the other hand, big data risk control models based on big data and machine learning have outstanding information advantages and model advantages (frost et al., 2019; gambacorta et al., 2019). they can be used to predict defaults more accurately and robustly (huang et al., 2020), and there will be a heterogeneous impact on different groups (fuster et al., 2018; huang et al., 2020). the combination of the two pillars of the big technology ecosystem and the big data risk control model constitutes a new type of credit risk management framework for big tech lending, making large-scale, low-cost, high-efficiency and sustainable sme loans possible. big tech lending promotes the inclusiveness of finance, but at the same time brings some problems and challenges, which require further research and improvement. for example, the risk control method of big tech lending is low in explanatory, similar to a "black box". the government supervision department cannot well identify the stability of this type of loan. the effectiveness of big tech lending in terms of risk control has not yet been tested by the complete financial cycle. the digital footprint may bring fraud risks and problems such as data privacy and "algorithm discrimination". big tech lending may also increase the "digital divide" of some special groups (such as middle-aged and elderly groups) (huang yiping and qiu han, 2021). 4. brief comment at present, relevant research results at home and abroad mainly concentrated on internet finance, digital inclusive finance and financial technology credit risk analysis and control, and big tech lending risk management framework research. in summary, in the context of financial technology, research on theories and methods of credit risk management for big tech lending just started. although big tech lending has achieved initial success, the maturity and popularization of its business model still requires the joint efforts of big tech lending institutions and government regulatory agencies. at present, the explanatory nature of the risk control method of big tech lending is low, and the government supervision department cannot well identify the stability of this type of loan (huang yiping and qiu han, 2021), which is a key issue that needs to be solved urgently in the credit risk management of big tech lending. the future research direction of credit risk management of big tech lending includes the following three aspects： mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 143 （1）research on big tech lending risk perception and influencing factors based on system simulation perceived credit risk is defined as: the possibility of loan loss for smes and its probability distribution that big tech lending institutions can perceive in credit transactions, consisting of basic credit risk and dynamic credit risk. based on this, the system composed of big tech lending risk perception and its influencing factors is defined as a big-tech credit risk perception system, and the system dynamics characteristics such as complexity, nonlinearity and time-varying characteristics are analyzed. according to the causal relationship among the various influencing factors in the system, a system dynamics model for the perception of credit risk of big tech lending is established. using the number of credit transactions as the simulation unit, through the vensim software, we can simulate the change trend of the perceived credit risk rate in the process of n credit transactions, and research the influence of changes in factors such as industry, technology, market, supervision, culture, transaction, big tech lending institutions, and big technology ecological environment on perceived credit risk. （2）research on credit risk control mechanism of big tech lending based on differential game big tech lending features "small amount, short term, and repayment on demand". therefore, the differential game can accurately describe the dynamic changes of the decision-making process of big tech lending institutions and smes in the process of big tech lending, and describe the differences of interests between the two parties in each stage of the game. a simplified big tech lending system is formed by big tech lending institutions (c) and smes (e) based on the big technology ecosystem. smes may default in the process of big tech lending transactions. therefore, smes may default institutions need to establish supervision and punishment mechanisms to supervise the transaction process of smes in order to reduce the occurrence of defaults and control the number of transaction defaults. smes choose the degree of effort to control breach of contract according to existing conditions to maximize their own interests. first, introduce characteristic variables such as the political benefits brought about by smes may default institutions’ regulatory defaults and the reputation benefits of smes in the large technology ecosystem brought about by the performance of smes. under the assumptions of model, the game model is constructed based on the differential game theory of the default supervision behavior of big tech lending institutions and the default control behavior of smes in the process of big tech lending transaction. then, solve different game equilibrium strategies under the situation of independent decision-making between big tech lending institutions and smes, the stackelberg game situation under the incentives of big tech lending institutions, and in the case of consistent decision-making between big tech lending institutions and smes, and analyze the degree of default supervision effort of big tech lending institutions and the degree of default control effort of smes in the three game situations. finally, make numerically simulation analysis of the theoretical model. （3）research on credit risk warning model of big tech lending based on cnn-lstm neural network using the good feature extraction ability of convolutional neural network (cnn) and the special memory prediction function of long and short-term memory (lstm) neural network, a big tech lending risk warning model based on cnn-lstm neural network is constructed. the first step is to collect sme credit data. the second step is to use nuclear principal component analysis (kpca) to mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 144 extract the principal components of numerical data, and use word2vec technology to convert text data into word vectors, thereby converting high-dimensional sparsity data into low-dimensional dense data. the third step is to input the extracted principal components and word vectors into cnn, and use cnn's good feature extraction function to find out the potential relationships in the historical data, and then integrate them into effective feature vectors. the fourth step is to input the integrated feature vector into the lstm neural network, memorize and filter the features, and perform fitting predictions, so as to output the credit risk prediction results through the fully connected layer. the fifth step is to set the early warning threshold to achieve credit risk early warning. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. acknowledgments: this research obtained some technical support from master si-si li in the field of financial technology. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] demirgüç-kunt a, klapper l. measuring financial inclusion: the global findex database[r]. the world bank, 2012. [2] dobbie w, skiba p m. information asymmetries in consumer credit markets: evidence from payday lending[j]. american economic journal: applied economics, 2013,5(4):256-282. [3] frost j, gambacorta l, huang y, shin h s, zbinden p. bigtech and the changing structure of financial intermediation[j]. economic policy, 2019,34:761-779. [4] fuster a, goldsmith-pinkham p, ramadorai t, walther a. predictably unequal? the effects of machine learning on credit markets[z]. cepr working paper, 2018, dp12448. [5] gambacorta l, huang y, qiu h, wang j. how do machine learning and non-traditional data affect credit scoring? new evidence from a chinese fintech firm[z]. bis working papers, 2019, no.834. [6] graves s m. landscapes of predation, landscapes of neglect: a location analysis of payday lenders and banks[j]. the professional geographer, 2003, 55(3): 303-317. [7] huang y, zhang l, li z, qiu h, sun t, wang x. fintech credit risk assessment for smes: the case in china[z]. imf working paper, 2020, 20/193. [8] chen h,guo l. the causes, negative effects and prevention system construction of financial technology risks [j]. reformation, 2020(3): 63-73. [9] chen h,li qc. six relationships that fintech companies should handle well for high-quality development [j]. shanghai enterprise, 2021(1): 79-84. [10] hu jy, shui bb. research on the relationship between internet financial lending term, credit risk and pricing——based on the perspective of borrower’s mismatch of investment and financing term [j]. modern finance (journal of tianjin university of finance and economics), 2020, 40(9): 3-18. [11] huang yp, qiu h. big technology credit: a new credit risk management framework [j]. management world, 2021, 37(2): 12-21+50+2+16. [12] jiang qp. "information-financial" quantity price transmission mechanism and digital financial risk supervision under the new financial order [j/ol]. price theory and practice: 1-5[2021-03-04]. https://doi.org/10.19851/j.cnki.cn11-1010/f.2021.01.05. [13] li gy. research on the formation mechanism of credit risk in internet finance [j]. journal of harbin university of commerce (social science edition), 2017(3): 38-45. [14] li km. innovation and risks of internet finance [j]. management world, 2016(2): 1-2. [15] liu mf. potential risks of financial technology and regulatory response[j]. southern finance, 2020(6): 4555. mu zhang and cheng cao / journal of risk analysis and crisis response, 2021, 11(3), 139-145 doi: https://doi.org/10.54560/jracr.v11i3.303 145 [16] liu xh. the new model of internet service for small and micro enterprises' credit investigation——taking the us silicon valley financial technology company nav as an example [j]. tsinghua financial review, 2019(6): 99-100. [17] meng xt. the development significance of fintech 2.0 and risk prevention strategies[j]. management modernization, 2020, 40(5): 12-14. [18] sun f. policy suggestions for improving the regulatory framework of financial technology companies in china [j]. research of development, 2019(11): 66-70. [19] wei xf. risks and supervision of digital inclusive finance[j]. financial theory and practice, 2019(6): 49-54. [20] zhang yw, gao wh. self-construction, cultural differences and credit risk: empirical evidence from internet finance[j]. financial research, 2020, 46(1): 34-48. [21] research group of the internet financing center of industrial and commercial bank of china, zheng x. international comparison and prospects of online credit evolution models——research on the evolution of the role of fintech companies[j]. modern finance guide, 2020(9): 66-70. [22] research group of chengdu branch of the people's bank of china, gong zq, jiang jq. internet banking risk characteristics and development constraints from the perspective of fintech [j]. southwest finance, 2020(8): 3-12. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 12, issue 2-4 journal of risk analysis and crisis response, 2022, 12(2), 95-105 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i2.328 95 review a systematic literature review on the influence mechanism of digital finance on economic growth mu zhang 1,* and yu-tong luo 1,2 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: march 22, 2022; accepted: may 31, 2022; published: july 10, 2022 abstract: to further promote the theoretical research on digital finance's effective support for highquality development of the real economy, this paper reviews the current status of domestic and foreign research on the impact mechanism of digital finance on the real economy and the impact mechanism of digital finance on economic growth. at present, domestic and foreign research on the impact mechanism of digital finance on the real economy mainly includes the basic transmission mechanism featuring the improvement of financing availability and guiding the flow of funds in the real economy, and the indirect transmission mechanism using innovation and r&d as a channel. domestic and foreign research on the impact mechanism of digital finance on economic growth mainly includes: the direct impact mechanism featuring the improvement of financial services universality and efficiency, and the indirect impact mechanism with resident consumption, innovation and entrepreneurship, industrial structure transformation, and business environment as the channels. future research directions include: deepening the research on the indirect transmission mechanism of digital finance to the real economy; expanding the research on the indirect impact mechanism of digital finance on economic growth. keywords: digital finance; economic growth; real economy; influence mechanism; literature review 1. introduction digital finance is a product of the combination of financial services with digital technologies including the internet (mobile internet and internet of things), big data, distributed technologies (cloud computing and blockchain), artificial intelligence, and information security (biometrics and encryption), which, together with internet finance and fintech, are a series of financial innovation spectrum concepts, and the provision of inclusive and accurate financial services is its core attribute (ma, degong, and teng, lei 2020; teng, lei, and ma, degong, 2020) [1,2]. along with the wide proliferation and deep penetration of digital finance, the existing pattern of difficult and expensive financing for the real economy has undergone a remarkable transformation. with features such as sustainability, wide-coverage, and near-zero marginal cost, digital finance can effectively support the recovery of the real economy. because of this, some scholars have systematically explored the impact effect and transmission mechanism of digital finance on the real economy (peking university digital finance research center group, 2018; wang yannan et al., 2020) mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 96 [3,4]. at the same time, the impact of digital technology on the traditional financial sector and its farreaching impact on the economy has received increasing attention from policymakers and scholars (wang yongcang and wen tao, 2020) [5]. digital financial development may contribute to economic growth in several ways (huang yiping and tao kunyu, 2019; qian haizhang et al., 2020) [6,7]. in this paper, we will review the current status of domestic and international research on the impact mechanism of digital finance on the real economy and the impact mechanism of digital finance on economic growth, to further promote the theoretical study of digital finance to effectively support the high-quality development of the real economy. the remainder of this paper is structured as follows: part 2 introduces the impact mechanism of digital finance on the real economy; part 3 introduces the impact mechanism of digital finance on economic growth, and part 4 is a brief review. 2. study on the influence mechanism of digital finance on the real economy 2.1. basic transmission mechanism: digital finance can boost the real economy, and there are structural and regional differences in this effect on the one hand, digital finance improves the availability of finance to the real sector, eases liquidity constraints, and contributes to the real economy ratio. appropriate relaxation of credit constraints is an important way to promote economic development (banerjee, 2003) [8]. guo, feng et al. (2019) [9] argue that traditional financial institutions are increasingly focusing on the concept of digital finance, and they have continuously expanded the service scope and reach of digital finance and reduced financial discrimination with the help of fintech such as big data and cloud computing. new internet financial platforms, such as alipay, have lowered the threshold of financial services (payments, insurance, money funds, etc.) and provided financial services within reach of the financially disadvantaged (yi xingjian and zhou li, 2018) [10]. in addition, digital finance provides insurance services for the development of the real economy, which to some extent hedges the operational risks of the real economy and re-motivates the real investment. financial development can improve productivity, reduce economic volatility, and promote economic development through better risk-sharing (insurance) (ramey and ramey, 1995) [11]. the credit function and insurance function of digital finance effectively provide entrepreneurs with capital and reduce entrepreneurial risk, which in turn increases their entrepreneurial motivation (xie, xuanli et al., 2018) [12]. on the other hand, digital inclusive finance is policy-oriented and targeted, which can, to a certain extent, curb the profit-seeking nature of capital and optimize the direction of capital allocation, guiding capital out of the overly prosperous financial and real estate industries and flowing to the real economy. the government attaches great importance to financial reform, and the central bank has repeatedly issued targeted downgrading policies for small and micro enterprises and the "three rural areas", mobilizing financial institutions to provide inclusive loans to disadvantaged economic groups. micro and small enterprises have poor or no business records (stiglitz and weiss, 1981) [13], which makes it difficult for financial institutions to assess the integrity of enterprises (agarwal and hauswald, 2008) [14], and the lack of collateralized assets for micro and small enterprises makes financial institutions assume greater risks. digital finance collects and integrates a large amount of behavioral data deposited by mses on the internet, and with the help of financial innovation technologies such as big data, cloud computing, and artificial intelligence, and then constructs credit mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 97 assessment models for mses. this shows that digital finance is a feasible solution to improve the information asymmetry dilemma (duarte et al., 2012; jizun li, 2015) [15,16]. it is worth mentioning that there may be structural and regional differences in the stimulation effect of digital finance on the real economy. firstly, the structural differences are clarified. digital finance includes three dimensions: breadth of coverage, depth of use, and degree of digitization (guo, f. et al., 2019) [9]. the breadth of coverage and depth of use belong to traditional finance, and they have quite rich practical experience in supporting the real economy. china's finance is in the transition period to digital finance, digital financial facilities are not yet sound, and in general, the digitalization of china's financial development still needs to be improved, so the boosting effect of digital finance on the real economy still needs time to be tested. secondly, we analyze the regionalization differences. the distribution of china's real economy shows the obvious characteristics of being strong from the east and weak from the middle and west (wang qian and dong yanling, 2018) [17]. at the same time, financial resources show a clustering trend (zhang and hui, 2016) [18], with the pearl river delta, yangtze river delta and bohai sea rim regions having the richest financial resources. on the contrary, the central and western regions have the most severe bank exclusion and insurance exclusion in the central rural areas, and the most severe internet financial exclusion in the western rural areas (su, f. and fang, l., 2016) [19], and these regional differences will certainly limit the boosting effect of digital finance on the real economy. thus, digital finance can boost the real economy, and there are structural and regional differences in this effect. 2.2. indirect transmission mechanism: digital finance can boost the real economy through innovative mechanisms over the past 40 years of reform and opening-up, china's real economy once relied on a factor and investment-driven crude growth model, which gradually revealed a series of problems such as overcapacity in low-end industries, environmental pollution, and resource shortage, foreshadowing the urgent need for technological innovation changes in china's real economy to continue to provide a strong impetus for economic growth. waldorf (1989) [20] empirically found that in the united states about two-thirds of the contribution of technological progress to economic growth can be explained by talent innovation. zhang lin (2016) [21] found that technological innovation has significantly contributed to the growth of the real economy in both the short and long term. the development of the real economy is the foundation of economic growth, and the fundamental path to promoting the real economy cannot be separated from science and technology innovation (long, 2018) [22]. it can be inferred from this that innovation r&d is the fundamental driving force to solidify the sustainable growth of the real economy. according to practical experience, the intrinsic mechanism of the driving role played by innovation r&d can be summarized as follows. first, innovative r&d can nurture the birth and growth of new industries and promote the transformation of china's real economy from low value-added to high value-added; second, innovative r&d can provide traditional manufacturing industries with automated production lines, optimize the combination of factor inputs, reduce energy consumption, pollution and production costs, and thus improve the production efficiency, product quality and profitability of traditional manufacturing industries; third, china's high-tech third, china's high-tech development zones and technology towns strongly support the industrial clustering effect of innovative r&d, which enables each region to form unique industrial advantages and avoid the tidal wave phenomenon of mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 98 homogeneous production and cheap products, thus optimizing the industrial layout of each city cluster according to local conditions. technological progress is the source of economic growth, while innovative r&d is an important guarantee to promote technological progress. numerous previous studies have confirmed that financial development can significantly promote technological innovation (maskus et al., 2011; miaomiao li et al., 2015) [23,24]. specifically, chowdhurya and maung (2012) [25] proposed that financial development enhances investment in innovation r&d by reducing the degree of information asymmetry. zhang (2012) [26] suggested that both the scale and efficiency of financial development can stimulate innovative r&d. as research continues to advance, the impact of digital finance on innovation r&d, in addition to traditional finance, cannot be ignored. liang, pang and zhang (2019) [27] found that digital finance can promote innovation in smes by easing access to financing. this shows that digital finance can boost the real economy through innovative mechanisms. 3. study on the influence mechanism of digital finance on economic growth 3.1. digital finance enables financial services to reflect better inclusiveness and higher efficiency digital finance lays the foundation for low-income and vulnerable groups to access low-cost financial services and is, therefore, an important source for achieving low-cost, broader coverage and more sustainable financial inclusion (xie, 2016) [28]. while providing effective and sustainable financial services to smes and low-income households is extremely challenging for any country around the world, the rise of the digital finance industry in china proves that with the help of digital technology, digital finance, especially mobile payments and online lending, overcomes geographical barriers and the blind spots of traditional risk assessment to achieve lower-cost access to smes and all types of people everywhere, especially in addition, with the help of "big data", digital finance, especially mobile payment and online lending, has overcome geographical barriers and traditional blind spots in risk assessment to provide more convenient financial services at a lower cost to smes and all kinds of people everywhere, especially those in less developed areas and low-income society. in addition, credit scoring models that leverage "big data" information can help mitigate financial frictions such as information asymmetries caused by the fragmentation of china's credit market, and it also pushes the "frontier" of credit accessibility to companies with lower credit scores by significantly improving risk control methods. it also pushes the "frontier" of credit availability to companies with low credit scores by significantly improving risk control methods. internet lenders have an information advantage over traditional financial institutions in credit assessment (frost et al., 2019) [29], and digital financial credit providers such as ant financial and wezuo bank have a competitive advantage over traditional commercial banks due to stronger credit rating systems, cheaper distribution channels, and more pronounced information advantages (hau et al., 2019) [30]. for micro-entities, digital finance development reduces the financial transaction costs for residents and small, medium, and weak entities. information technology-based digital finance is an important driving force of business model change (jiang xiao juan and luo libin, 2019) [31], which has changed the face-to-face transaction model in traditional business models, greatly reducing the transaction costs of traditional financial services and improving transaction efficiency (zeng and reinartz, 2003) [32]. compared with the traditional financial business model, digital finance pays more attention to the scale effect and tail effect, the accumulation of existing data reduces the mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 99 marginal cost of developing related business, and the internalization of the business model also makes the coverage of digital finance unrestricted by time and space (lu, 2018) [33], which precisely solves the problem of diseconomies of scale of traditional finance. in addition, the development of digital finance facilitates the service upgrade of traditional financial institutions and effectively alleviates the problem of financial exclusion. due to risk considerations, traditional financial institutions are often reluctant to provide financial services to small, medium, and weak subjects, who face a difficult capital shortage problem (li and han, 2019) [34]. digital financial development, through information technology, is convenient and fast, but also has the advantage of low cost, which enables underdeveloped areas to enjoy the same financial services, cracking the long-standing problem of insufficient financial services and making up for the lack of traditional financial services. taking commercial banks as an example, the use of internet technology by commercial banks has created rich application scenarios and enhanced user experience, which is conducive to the improvement of commercial banks' efficiency and undoubtedly helps alleviate the problem of financial exclusion. 3.2. the development of digital finance has boosted consumer spending by facilitating the customer experience one of the reasons for the decline in china's economic growth is the low consumption, which has also become an important driving force of china's economic growth under the new economic normal. mobile payment in digital finance facilitates customers' payment experience, and risk assessment based on real transaction behavior weakens the requirement of collateral and facilitates customers to carry out cross-consumption. from this perspective, the development of digital finance has changed customers' attitudes towards consumption and facilitated the transformation and upgrading of consumption. in the case of service consumption, for example, through digital diffusion, residents are more willing to consume food and beverage, healthcare, entertainment, and education and culture, and digital financial development has played an active role in promoting consumption upgrading and encouraging consumer demand (yi, xingjian, and zhou, li, 2018) [10]. digital finance has an impact on residents' consumption and consumption upgrading in three main ways (yang, w. et al., 2021) [35]. first, digital finance reaches users mainly through an online model, which can expand the coverage of financial services and make credit support more accessible to the general population. during 2019 double 11, more than 8 million products in tmall and taobao were available in interest-free chanting instalments, and the transaction volume of jingdong white strip broke 100 million in 10 seconds. zhao baoguo and gai nian (2020) [36] found that compared to basic subsistence consumption expenditure, the credit allocation strength of internet finance has a higher contribution rate to residents' development of enjoyment consumption expenditure. second, digital finance contains a diversity of financial services such as funds, insurance, and investment. compared with the low return of traditional savings, digital finance can improve the level of residents' return on investment and financial management (yang, weiming et al., 2020) [37]. according to xing and zhang (2019) [38], the wealth effect brought by internet finance can effectively increase the consumption level of residents. in addition, data show that as of the end of 2019, the overall wealth management scale of yu’e bao was about 1.09 trillion yuan, of which 99.8% was occupied by individual investors, bringing 21.3 billion yuan of income to customers throughout the year, and the higher level of income has laid a certain foundation for promoting consumption. third, mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 100 the convenience and low transaction costs of digital finance have an important impact on residents' consumption. expanding leisure tourism spending is one of the important elements of consumption upgrading, and guo yingzhi and li xiaomin (2018) [39] found that future travelers will use mobile payments more often to purchase tourism products. 3.3. for society as a whole, digital financial development promotes technological innovation and regional entrepreneurship as a representative of the new financial model, digital finance development provides new opportunities to solve the financing of small and medium-sized enterprises and promote the development of innovation and entrepreneurship (yin et al., 2019; xie et al., 2018; liang and zhang, 2019) [40,12,27]. along with china's economy entering a new normal, the chinese government has implemented an innovation-driven development strategy to achieve high-quality economic development. by alleviating financial exclusion and increasing the availability of financial services, digital finance provides appropriate funding for the promotion of innovation and entrepreneurship, which is conducive to the formation of "mass entrepreneurship and innovation" and thus promotes china's economic growth. for example, beck et al. (2018) [41] construct a general equilibrium model using mobile payments in digital finance as an example and confirm that mobile payments in kenya can ultimately promote economic development by enhancing entrepreneurship. yanqin lv and bin zhao (2020) [42] argue that digital inclusive finance better serves rural msmes, alleviates financing difficulties, promotes entrepreneurship and employment, increases the flow of capital, and increases aggregate social demand, thus boosting economic growth. ma degong and teng lei (2020) [1] argue for the transmission mechanism of the role of digital finance development in alleviating the financial constraints of entrepreneurship and thus achieving inclusive growth. some studies suggest that digital finance can create economic value by promoting technological innovation and supporting entrepreneurship. for example, credit histories based on internet transaction information can effectively facilitate e-commerce sales and transactions, and this promotion is more significant for young firms and young entrepreneurs (hau et al., 2019) [30]. online loans offered to e-commerce entrepreneurs can improve the quality of service for such consumers and help them mitigate financial and operational shocks in a more effective manner (huang et al., 2018) [43]. chinese digital finance companies that have leveraged the online credit market can offer a richer product range than firms not involved in internet lending (frost et al., 2019) [29]. in addition, it was also found that by combining the digital inclusive finance index with household survey data, the use of mobile payments was found to significantly increase the likelihood that farmers would become selfemployed in the future and would also significantly increase their income (wang, 2019) [44]. 3.4. digital finance promotes regional economic growth by facilitating the structural transformation of industries in addition to directly promoting economic growth, digital finance can also promote economic growth by facilitating industrial structural transformation (linhan li and weimin tian, 2021) [45]. the development of digital finance promotes industrial structure transformation mainly in the following ways: 1) the development of digital finance requires the completion of digital infrastructure. the development of digital infrastructure will prompt enterprises to use internet thinking and rely on digital technology to optimize enterprise management mode, improve mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 101 enterprise productivity and informatization level, and promote the transformation of enterprises to intelligent networking, that is, digital finance can promote the transformation and upgrading of industrial structure through the development of digital infrastructure (xi enchong et al., 2013) [46]. 2) the development of digital finance is mainly reflected in the information service industry development level. the so-called information service industry is an industry that uses the advanced technology of the internet to collect, process and handle network information, and feeds it back to users with information-based products. with the progress of digital technology such as the internet, the information service industry has become more and more closely connected with the manufacturing industry, and within the manufacturing industry, the degree of automation and production efficiency has also been improved with the help of the improvement of information technology level. the integration of the information technology industry and manufacturing industry has become the current trend, and the two promote each other, which can realize the transformation of the manufacturing industry, that is, digital finance can promote the transformation and upgrading of industrial structure through the deepening of digital industry (tao changqi and zhou xuan, 2015) [47]. 3) the development of digital finance must not leave the promotion of digital technology, and the upgrading of industrial structure is also the process of continuous improvement of technological innovation. at the enterprise level, the emergence of emerging technologies such as the internet can help enterprises reduce information acquisition costs and transaction costs and improve operational efficiency, while in terms of digital technology it is necessary to integrate existing technologies faster and better, promote the emergence of new technologies, achieve the optimization of the internet of things, artificial intelligence and big data, and help enterprises achieve data integration and intelligent manufacturing, i.e., digital finance can continuously enhance through technological innovation the boosting effect on industrial structure transformation (chihiro et al., 2018) [48]. 3.5. digital finance can promote economic growth by improving the business environment digital finance can optimize the governmental, market and legal environments by reducing transaction costs and easing financing constraints, as well as enhancing the human environment by releasing the vitality of innovation and entrepreneurship, thus playing a role in improving the business environment. first, digital finance relies on internet technology to shorten the time for traditional financial institutions to review and issue loans from months to seconds, and the speeding up of the loan approval process reduces transaction costs and improves transparency in the process of corporate financing, thus effectively optimizing the governmental environment. on the one hand, digital finance can rely on soft information such as credit records and transaction data tracked by big data technology as the basis for credit assessment, breaking through the financing dilemma suffered by small and micro enterprises due to the disadvantage of hard information such as financial statements, and on the other hand, it can expand the coverage of financial services through information technology and lower the access threshold of the credit market, both of which will alleviate the financing constraints and credit discrimination of small and micro enterprises and enhance the market environment of fair competition; thirdly, digital finance can reduce the nonproductive expenses of enterprises. traditional financial institutions use credit rents of enterprises as an important identification tool for whether to borrow or not, which can breed rent-seeking behaviors among approval and credit-granting personnel in charge of credit and increase additional costs of enterprise financing, while the emergence of digital finance reduces manual intervention, eliminates mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 102 the space for rent-seeking in the approval process, and reduces the human and relationship fees of enterprises in the financing process, thus playing a role in improving the legal environment. fourth, the emergence of digital finance also promotes the rapid development of e-commerce, releasing more business opportunities and greater entrepreneurial vitality, which breeds fertile soil for "mass entrepreneurship and innovation" and thus enhances the human environment (zhang and yu, 2021) [49]. the positive effect of the business environment on economic growth can be realized through at least three channels. first, where the business environment is more perfect, the laws for property rights protection are also more strictly enforced, and the protection of good property rights is conducive to absorbing more fdi, which also helps to increase domestic private investment, thus promoting economic growth (dong et al., 2012) [50]; second, a good business environment helps to reduce rent-seeking behavior and lower transaction costs, releasing greater market dynamics, breeding more independent innovation and promote economic growth (xia houxue et al., 2019) [51]; finally, the optimization of the business environment can also break administrative monopolies, reduce inequality of opportunity, and narrow the income gap (shi xinjie et al., 2018) [52]. 4. brief review at present, domestic and foreign research on the impact mechanism of digital finance on the real economy mainly includes: the basic transmission mechanism featuring the improvement of financing availability and guiding the flow of funds in the real economy, and the indirect transmission mechanism with innovation and r&d as the channel. domestic and foreign studies on the impact mechanism of digital finance on economic growth mainly include the direct impact mechanism characterized by improving the universality and efficiency of financial services, and the indirect impact mechanism with resident consumption, innovation and entrepreneurship, industrial structure transformation, and business environment as the channels. however, scholars at home and abroad have studied the basic transmission mechanism of digital finance to the real economy more abundantly, while the research on the indirect transmission mechanism is relatively weak; therefore, the research on the indirect transmission mechanism of digital finance to the real economy needs to be further deepened. in addition, although domestic and foreign scholars have conducted rich research on the indirect influence mechanism of digital finance on economic growth, the mediating variables involved are still not comprehensive enough, therefore, the research on the indirect influence mechanism of digital finance on economic growth needs to be further expanded. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] ma degong, teng lei. digital finance, entrepreneurship and inclusive growth [j]. collected essays on finance and economics, 2020(09): 54-63. doi: https://doi.org/10.13762/j.cnki.cjlc.2020.09.006. [2] teng lei, ma degong. can digital finance help to promote high-quality development? [j]. statistical research, 2020, 37(11): 80-92. doi: https://doi.org/10.19343/j.cnki.11-1302/c.2020.11.007. mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 103 [3] research group of digital finance research center of peking university. the power of digital finance: empowering the real economy [m]. huang zhuo (editor-in-chief), beijing: renmin university of china press, 2018. [4] wang yanan, ye xin, xu lin. can digital finance boost the real economy [j]. finance & economics, 2020(03): 1-13. [5] wang yongcang, wen tao. research on economic growth effect and heterogeneity of digital finance [j]. modern economic research, 2020(11):56-69. doi: https://doi.org/10.13891/j.cnki.mer.2020.11.009. [6] huang yiping, tao kunyu. revolution of digital finance in china: experience, impacts and implications for regulation [j]. international economic review, 2019(06): 24-35+5. [7] qian haizhang, tao yunqing, cao songwei, cao yuyang. theoretical and empirical analysis on the development of digital finance and economic growth in china [j]. the journal of quantitative & technical economics, 2020, 37(06): 26-46. doi: https://doi.org/10.13653/j.cnki.jqte.2020.06.002. [8] banerjee a v. advances in economics and econometrics: theory and applications, eighth world congress [m]. cambridge university press, 2003:1-33. [9] guo feng, wang jingyi, wang fang. measuring china’s digital financial inclusion: index compilation and spatial characteristics [r]. working paper of peking university digital finance research center, 2019. doi: https://doi.org/10.13821/j.cnki.ceq.2020.03.12. [10] yi xingjian, zhou li. does digital financial inclusion significantly influence household consumption? evidence from household survey data in china [j]. journal of financial research, 2018(11): 47-67. [11] ramey g, ramey v a. cross-country evidence on the link between volatility and growth [j]. american economic review, 1995,85(05):1138-1151. doi: https://doi.org/10.3386/w4959. [12] xie xuanli, shen yan, zhang haoxing, guo feng. can digital finance promote entrepreneurship? —— evidence from china [j]. china economic quarterly, 2018,17(04):1557-1580. doi: https://doi.org/10.13821/j.cnki.ceq.2018.03.12. [13] stiglitz j e, weiss a. credit rationing in markets with imperfect information [j]. american economic review, 1981,71(03):393-410. doi: http://www.jstor.org/stable/1802787. [14] agarwal s, hauswald r b h. the choice between arm’s-length and relationship debt: evidence from eloans [r]. working paper, 2008. [15] duarte j, siegel s, young l. trust and credit: the role of appearance in peer-to-peer lending [j]. review of financial studies, 2012,25(08):2455-2484. doi: https://doi.org/10.1093/rfs/hhs071. [16] li jizun. reflections on internet finance [j]. management world, 2015(07):1-7,16. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2015.07.002. [17] wang qian, dong yanling. research on the regional difference and distributional dynamic evolution of real economy development [j]. the journal of quantitative & technical economics, 2018(05):77-94. doi: https://doi.org/10.13653/j.cnki.jqte.20180503.003. [18] zhang hui, liu peng, yu tao. financial spatial distribution, heterogeneity, and industrial allocation [j]. china industrial economics, 2016(12):40-57. doi: https://doi.org/10.19581/j.cnki.ciejournal.2016.12.004. [19] su fang, fang lei. regional differences in financial exclusion in china's rural areas: insufficient supply or insufficient demand? a comparative analysis of banking, insurance and internet finance [j]. management world, 2016(09):70-83. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2016.09.007. [20] waldorf w h. trends in american economic growth:1929-1982 [j]. business economics, 1989,24(03):65. https://link.gale.com/apps/doc/a7729745/aone?u=anon~ec7fee02&sid=googlescholar&xid=cc03dede. [21] zhang lin. financial development, technological innovation and real economic growth——an empirical study based on spatial econometrics [j]. journal of finance and economics, 2016(01):14-25. [22] long xiaoning. technological innovation and growth of the real economy [j]. china economic studies, 2018(06): 21-30. doi: https://doi.org/10.19365/j.issn1000-4181.2018.06.03. [23] maskus k e, neumann r, seidel t. how national and international financial development affect industrial r&d [j]. european economic review, 2011,56(01):72-83. doi: https://doi.org/10.1016/j.euroecorev.2011.06.002. [24] li miaomiao, xiao hongjun, zhao shuang. study on the relationship among the financial development, technological innovation and economic growth based on china's provincial panel data [j]. chinese journal of management science, 2015(02): 162-169. doi: https://doi.org/10.16381/j.cnki.issn1003207x.2015.02.020. mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 104 [25] chowdhury r h, maung m. financial market development and the effectiveness of r&d investment: evidence from developed and emerging countries [j]. research in international business and finance, 2012,26(02):258-272. doi: https://doi.org/10.1016/j.ribaf.2011.12.003. [26] zhang zhiqiang. financial development, r&d innovation and regional technology deepen [j]. economic review, 2012(03): 82-92. doi: https://doi.org/10.19361/j.er.2012.03.010. [27] liang bang, zhang jianhua. can the development of digital inclusive finance stimulate innovation? —— evidence from chinese cities and smes [j]. modern economic science, 2019(05):74-86. [28] xie p, zou c, liu h. the fundamentals of internet finance and its policy implications in china[j]. china economic journal, 2016,9(03):240-252. doi: https://doi.org/10.1080/17538963.2016.1210366. [29] frost j, gambacorta l, huang y, shin h s, zbinden p. bigtech and the changing structure of financial intermediation [r]. bis working paper, no.779, 2019. doi: https://doi.org/10.1093/epolic/eiaa003. [30] hau h, huang y, shan h, sheng z. how fintech enters china’s credit market [j]. aea papers and proceedings, 2019,109(05):60-64. doi: https://doi.org/10.1257/pandp.20191012. [31] jiang xiaojuan, luo libin. globalization of services in the internet age—a new engine, acceleration and the competitiveness of major powers [j]. social sciences in china, 2019(02):68-91+205-206. [32] zeng m, reinartz w. beyond online search: the road to profitability [j]. california management review, 2003,45(02):107-130. doi: https://doi.org/10.2307/41166168. [33] lu l. promoting sme finance in the context of the fintech revolution: a case study of the uk’s practice and regulation [j]. banking and finance law review, 2018,33(03):317-343. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3144767. [34] li jianjun, han xun. the effect of financial inclusion on income distribution and poverty alleviation: policy framework selection for efficiency and equity [j]. journal of financial research, 2019(03):129-148. [35] yang weiming, su lin, sun ruili, yuan weipeng. has digital finance promoted consumption upgrading? —— evidence based on panel data [j]. studies of international finance, 2021(04):13-22. doi: https://doi.org/10.16475/j.cnki.1006-1029.2021.04.002. [36] zhao baoguo, gai nian. the impact of internet consumer finance on the consumption structure of domestic residents: an empirical study based on var model [j]. journal of central university of finance & economics, 2020(03):33-43. doi: https://doi.org/10.19681/j.cnki.jcufe.2020.03.004. [37] yang weiming, su lin, wang mingwei. digital financial inclusion and income of urban and rural residents: based on the intermediary effect of economic growth and entrepreneurial behavior [j]. journal of shanghai university of finance and economics, 2020, 22(04): 83-94. doi: https://doi.org/10.16538/j.cnki.jsufe.2020.04.006. [38] xing tiancai, zhang xi. the impact of internet consumer finance on urban household consumption upgrade and changes in their consumption tendency [j]. contemporary economic research, 2019(05):8997+113. [39] guo yingzhi, li xiaomin. an empirical study on consumers' intention of buying tourism products with mobile payments——an integration model of tam and tpb [j]. journal of sichuan university (philosophy and social science edition), 2018(06): 159-170. [40] yin z, gong x, guo p, wu t. what drives entrepreneurship in digital economy? evidence from china [j]. economic modelling, 2019,82(03):66-73. doi: https://doi.org/10.1016/j.econmod.2019.09.026. [41] beck t, pamuk h, ramrattan r, uras b r. payment instruments, finance and development[j]. journal of development economics, 2018,133:162-186. doi: https://doi.org/10.1016/j.jdeveco.2018.01.005. [42] lv yanqin, zhao bin. research on digital finance and rural economic development [j]. wuhan finance, 2020(03):79-84. [43] huang y, lin c, sheng z, wei l. fintech credit and service quality [r]. mimeo, 2018. [44] wang x. can mobile payment facilitate entrepreneurial activity: evidence from household panel data [r]. china & world economy (forthcoming), 2019. [45] li linhan, tian weimin. digital finance, industrial structure transformation and regional economic growth: an empirical research based on spatial dubin model [j]. financial theory and practice, 2021(02):8-16. [46] xi enchong, xu zhipeng, zhang dan. research on the total factor productivity effect of china's infrastructure investment [j]. statistics and decision, 2013(23):137-140. doi: https://doi.org/10.13546/j.cnki.tjyjc.2013.23.028. mu zhang and yu-tong luo / journal of risk analysis and crisis response, 2022, 12(2), 95-105 doi: https://doi.org/10.54560/jracr.v12i2.328 105 [47] tao changqi, zhou xuan. effect analysis of industrial structure optimization and upgrading—— empirical research on coupling of information industry and manufacturing [j]. industrial economics research, 2015(03): 21-31+110. doi: https://doi.org/10.13269/j.cnki.ier.2015.03.003. [48] chihiro w, et al. measuring gdp in the digital economy: increasing dependence on uncaptured gdp [j]. technological forecasting & social change, 2018(137):226-240. doi: https://doi.org/10.1016/j.techfore.2018.07.053. [49] zhang rui, yu jintao. digital finance, business environment and economic growth [j]. modern economic research, 2021(07): 1-9. doi: https://doi.org/10.13891/j.cnki.mer.2021.07.002. [50] dong zhiqiang, wei xiahai, tang canqing. institutional soft environment and economic development: an empirical study based on the business environment of 30 big cities [j]. management world, 2012(04):9-20. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2012.04.002. [51] xia houxue, tan qingmei, bai junhong. business environment, enterprise rent-seeking and market innovation: evidence from the china enterprise survey [j]. economic research journal, 2019, 54(04): 84 98. [52] shi xinjie, wei longbao, fang shile, gao xuwen. inequality of opportunity in china's income distribution [j]. management world, 2018, 34(03): 27-37. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2018.03.003. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time scales sijian zhao1,2 1key laboratory of digital agricultural early-warning technology, moa, agricultural information institute of caas, no. 12, zhongguancun south street, haidian district, beijing, 100081, china 2academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education, beijing normal university, no.19 xinjiekouwai street, haidian district, beijing, 100875, china e-mail: scanzhao@hotmail.com qiao zhang key laboratory of digital agricultural early-warning technology, moa, agricultural information institute of caas, no. 12, zhongguancun south street, haidian district, beijing, 100081, china abstract [purpose] crop’s risk assessment induced by natural disasters needs to develop in the fine scale. it is required that risk assessment can be performed on county-level space scale and month-based time scale. but with th e thinning of space and time scales, disaster samples would decl ine greatly and severely limit the us age of risk assessment method based on the statistic of samples. therefore, in this paper, a new method on base of metrological data and crop’s vulnerability models was proposed to assess risk variability among months on county scale. [methods] since the weather stations cannot cover all the counties, the idw method was used to interpolate the daily rainfall for each coun ty by its nearby 3 stations. moreover, ba sed on the flood recor d data, the vulnerability function between crop’s flood-effected area and av erage daily rainfall in a storm wa s regressed. using the informatio n diffusion model of nonp arametric kernel density, the probability density function of average daily rainfall in a storm for each month of each county was fitted. finally, the expected disaster-effected area of crops as risk measure for each month of each county was calculated on the combination of probability function and vulnerability function. [results] through the case study in three northeastern provinces of china, a serial of risk maps on county-level and month-level scales were produced to be used to recognize the rules of risk variability in space and time. keywords: crop; flood; risk; county-level scale; month-level scale 小时空尺度下的东北三省农作物洪涝风险评估 赵思健 1,2 张峭 2 1. 中国农业科学院农业信息研究所，农业部智能化农业预警技术重点开放实验室，北京，100081 2. 北京师范大学，民政部/教育部应急与减灾管理研究院，北京，100875 摘要：（目的）农作物灾害风险评估逐步向尺度精细化的方向发展，评估结果要求在空间上达到县级尺 度，在时间上达到月级尺度。然而，随时空尺度的细化，灾害样本数量急剧减少，严重限制了基于样本统 计的风险评估方法的运用。为此，本文提出一种气象数据与农作物洪涝脆弱性相结合的方法，以东北三省 为研究区开展以县和月为单位的小时空尺度下农作物洪涝风险评估。（方法）由于气象站点无法覆盖所有 县级行政区，利用与县级行政区最临近的 3 个气象站点日降雨数据，采用反距离权重法（idw）插值出各 县的日降雨数据；收集农作物洪涝灾情数据，回归出东北三省农作物洪涝受灾率与过程平均降雨量之间的 脆弱性函数；分县分月提取过程平均降雨量，构建非参数核密度的信息扩散模型拟合降雨量的概率分布； 最后，综合概率分布与脆弱性函数，计算出分县分月的农作物洪涝期望受灾率。（结果）通过实例应用， 制作出东北三省县级尺度下 4 至 9 月的农作物洪涝风险图，并对风险时空差异特征进行剖析。 关键词：农作物，洪涝，风险，县级尺度，月尺度 journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 201-208 published by atlantis press copyright: the authors 201 administrateur texte tapé à la machine received 13 august 2012; accepted 26 september 2012 administrateur texte tapé à la machine sijian zhao, qiao zhang 1. 引言 我国是一个农业大国，农业是一个安天下、稳 民心的基础产业，在我国受到了越来越高的重视。 但同时，我国又是一个自然灾害多发、频发的国 家。与世界平均水平相比，我国自然灾害发生频率 高出了 18 个百分点。据联合国开发计划署的标准， 我国农业是世界上遭遇灾害最频发、受灾面积最广 和灾害损失最严重的国家之一。 作为灾害防范的一种重要手段，农业生产灾害 风险评估是预测与量化未来自然灾害事件对农业生 产可能造成的损失程度，是指导农业生产灾害风险 管理的一项重要工作，逐渐受到了人们的重视。在 我国，农业生产灾害风险评估工作开展得较晚，在 方法上分为基于指标体系的评估法与基于损失统计 的评估法。其中，基于损失统计的评估法运用得较 为广泛，它是在损失数据样本的基础上经统计建模 后计算出代表风险的某一定量指标，属于纯定量的 方法。根据损失数据源的不同，又可分为基于单产 损失 1,2 与基于灾情损失 3,4 的评估法。在时空尺度 上，绝大多数的风险评估研究集中在中大尺度上， 空间上以省级、市级行政区为单元，时间上以年为 单位。以县级行政区和以月为单位的小时空尺度风 险评估常常因为损失样本的严重不足甚至完全缺 失，无法实施统计建模而被放弃。然而，小时空尺 度的风险评估能更加有效地识别出高风险区与高风 险期，为实现更有针对性的风险管理提供强有力的 支撑，值得深入探索。 为此，本研究尝试在有限的灾情数据上结合气 象数据，实现以县和以月为单位的小时空尺度下农 作物灾害风险评估，并探讨小时空尺度下的评估结 果所包含的风险时空差异特征。 2. 研究区概况 黑龙江、吉林和辽宁俗称东北三省，位于 115º 32’e 至 135º10’e，38º43’n 至 53º25’n 之间（如图 1 所示），土地总面积占全国土地总面积的 8.33%， 耕地面积占全国耕地总面积的 16.68%5，2011 年粮 食产量达到 2155 亿斤，占全国粮食产量的 18.9%6， 主要粮食作物包括玉米、大豆、春小麦、水稻等， 是我国重要的粮食生产基地。同时，东北三省又是 我国受全球气候变化影响最显著的地区之一 7，气候 变化加剧了暴雨洪涝灾害事件的发生，使东北三省 成为洪涝的多发区，严重阻碍了区域农业的发展。 作为重要的产粮基地，防范洪涝灾害风险，保障东 北三省的农业发展及粮食生产安全对于保障我国的 整体粮食安全有着极其重要的战略意义。 基于上述背景，本研究选择东北三省为研究 区，以洪涝灾害作为影响农作物生产的代表性灾 种，开展以县和以月为单位的小时空尺度下农作物 洪涝风险评估。 图 1 东北三省县级行政区划及气象站点分布 3. 数据收集与处理 3.1. 数据收集 本研究收集了三类分析数据：（1）农作物洪涝 灾情数据，（2）农作物种植面积数据和（3）东北 三省气象站点日降雨量数据。 农作物洪涝灾情数据来源于多种渠道，包括中 国气象灾害大典、地方气象局统计资料和地方农业 局统计资料等。灾情数据总量达 1871 条（如表 1 所 示），在时间序列上从 1984 年至 2009 年，空间尺 度上精确到县级，时间尺度上精确到年月日，灾情 统计的数据指标包括了农作物受灾面积（公顷）、 成灾面积（公顷）和绝收面积（公顷）。其中，成 灾面积与绝收面积数据存在严重的缺失，因此在后 续的分析中主要运用受灾面积作为农作物灾情损失 指标。农作物洪涝灾情数据具体格式如表 2 所示。 published by atlantis press copyright: the authors 202 risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time scales 农作物种植面积数据来源“中国农业部县级农 村经济统计数据库”，时间序列上从 1981 年至 2011 年，空间尺度上精确到县级，种植面积单位为 公顷。 气象站点日降雨数据来源“中国气象科学数据 共享服务网”（http://cdc.cma.gov.cn/）。整个东北 三省共设 86 个气象站点（站点分布如图 1 所示）， 降雨量数据在时间序列上是从 1959 年至 2011 年， 在时间尺度上精确到日，单位为 mm。 表 1. 东北三省农作物洪涝灾情数据统计（1984-2009） 省份 洪涝记录（条） 黑龙江 894 吉 林 567 辽 宁 410 合 计 1871 表 2. 东北三省农作物洪涝灾情记录（1984-2009） 区县 年 起始 终止 受灾面积 （公顷） 月 日 月 日 嘉荫县 1984 6 15 6 18 28000 铁力市 1985 7 27 7 28 1781.3 五营区 1995 7 23 7 24 340 … … … … … … … 3.2. 数据处理 3.2.1. 农作物洪涝受灾率计算 假设某县发生的某一次洪涝灾害造成农作物洪 灾面积 a ，对应县相应年份的农作物种植面积 s ， 则此次农作物洪涝受灾率 y 计算如下： say /= (1) 选取农作物洪涝受灾率作为农作物灾情指数。 3.2.2. 县级行政区日降雨量插值计算 图 1 可以看出，东北三省全境范围内 86 个气象 站点无法完全覆盖 182 个县级行政区。利用气象站 点日降雨量数据直接评价县级行政区日降雨量显得 比较粗糙。为了体现出县级行政区日降雨量的差 异，可先查找与县级行政区中心点最近的 3 个气象 站点，再利用这 3 个站点的日降雨数据，采用反距 离权重法（idw）8 插值获得县级行政区的日降雨 量。 图 2 示意了县级行政区日降雨量插值过程。假 设待计算的县级行政区中心点为 c ，查找到与该中 心点最近的 3 个气象站点（ 3,2,1, =kmk ），并计算 出中心点与 3 个气象站点的距离（ 3,2,1, =kd k ） 后，县级行政区日降雨量 ci （mm）的 idw 插值计 算如下： ∑ ∑ = = ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⋅ = 3 1 2 3 1 2 1 1 k k k k k c d d i i (2) 其中， ki （ 3,2,1=k ）表示 3 个气象站点的日降雨量 （mm）。 1d 2d 3d c )( 11 im )( 22 im )( 33 im 图 2 县级行政区日降雨量插值示意 4. 农作物洪涝风险评估 4.1. 农作物洪涝脆弱性分析 承灾体是致灾因子作用的对象，脆弱性用于衡 量承灾体遭受损害的程度，是致灾因子与灾情联系 的桥梁 9。当承灾体的脆弱性侧重于因灾造成的灾情 水平方面时，通常可用致灾因子（ h ）与成害 （ d ） 之 间 的 关 系 曲 线 或 方 程 式 表 示 ， 即 ),( dhfv = ，又叫做脆弱性曲线或灾损（率）曲 线，用来衡量不同的致灾因子强度与相应损失 （率）之间的关系。在本研究中，洪涝灾害的致灾 published by atlantis press copyright: the authors 203 sijian zhao, qiao zhang 因子是强降雨，承灾体是农作物，农作物洪涝脆弱 性就要建立降雨与农作物受灾率之间的曲线（函 数）关系。 洪涝通常是发生在一次强降雨过程中。所谓一 次强降雨过程是指一日或连续多日降雨量超过某一 阈值 0i 的阶段性降雨（如图 3 所示）。根据经验， 阈值 0i 可取 10mm。描述一次强降雨过程的数据指 标有过程降雨量极值 pi （mm）、过程降雨总量 ti （mm）、过程平均降雨量 mi （mm）和降雨持续时 间 d （日）。其中，过程平均降雨量 mi 与过程降雨 总量 ti 的之间关系表示如下： dii tm /= (3) 0 50 100 150 200 250 300 350 400 1 6 11 16 21 26 31 日 降 雨 量 （ m m ） 日 图 3 一次强降雨过程图示 利用收集的洪涝灾情数据，根据每条灾情记录 中洪涝发生时间提取对应降雨过程的数据指标 （ pi ， ti 和 mi ），并进行农作物洪涝受灾率与各数 据指标之间的回归分析，分析发现过程平均降雨量 mi 与受灾率 y 之间的相关性最大。因此，本研究选 择过程平均降雨量作为致灾因子指标，借助回归分 析（其中线性回归拟合效果最佳）建立东北三省农 作物洪涝脆弱性函数，即： 0283.00016.0 −⋅= miy (4)   图 4 过程平均降雨量与农作物洪涝受灾率的线性回归 图 4 是东北三省过程平均降雨量与农作物洪涝 受灾率线性回归分析结果。结果可以看出，相关系 数 2r （0.2318）比较低。究其原因，一方面是洪涝 灾害本身存在较强的随机性，受随机性影响相关系 数不会太高；另一方面，一个地区的农作物洪涝受 灾率不仅与强降雨有密切关系，还与该地区的地 形、地貌（河流和植被）等环境因子有关系，例如 地形低洼、河流密集和植被稀疏的地区易发生洪 涝，农作物受灾率会较高，因此仅用降雨量进行一 元回归导致相关系数不高。但从图 4 可以看出，回 归曲线还是能够客观、有效地反映过程平均降雨量 与受灾率之间的发展趋势，因此被用于后续的风险 评估中。 4.2. 分县分月过程平均降雨量概率分布拟合 4.2.1. 洪涝灾害月发生频次分析 在不考虑洪涝灾害强度及造成损失的前提下， 先利用收集的洪涝灾情数据对洪涝灾害不同月份上 的发生频次进行统计，结果如图 5 所示。结果显 示，东北三省历史上有 98%以上的洪涝灾害发生在 4 至 9 月间，且 6，7 和 8 月是洪涝的频发期，其他 月份的洪涝可忽略不计。因此，下面仅对 4 至 9 月 （6 个月）的过程平均降雨量概率分布进行分月拟 合。 图 5 洪涝灾害月发生频次统计 4.2.2. 过程平均降雨量概率分布的非参数拟合 分布拟合的方法通常有参数估计法与非参数估 计法。参数估计法虽运用较广，但要求有先验的分 布函数、样本容量足够大且已知关于总体分布的有 关信息，而实际应用中往往很难获得总体分布信 息，选择先验分布函数的过程也比较繁琐。相比之 下，非参数估计法不限制样本母体的分布形式，估 计结果相对稳定。为此，本文采用非参数核密度的 信息扩散模型 10 进行降雨概率分布的拟合。 首先，利用县级行政区的日降雨数据，分县分 月（仅 4 至 9 月）提取过程平均降雨量样本，即： ctlmkmmmctm iiiii },..,,...,,{ ,,2,1,, = (5) published by atlantis press copyright: the authors 204 risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time scales 其中， c 表示县级行政区单元， t 表示月份， ctmi , 表示县级行政区月份上过程平均降雨量样本集 合， kmi , 表示集合中的过程平均降雨量样本， l 表示 集合中的样本个数。 假设过程平均降雨量的样本空间u 为[0,300],并 将样本空间分成 s 等分，获得u 的离散论域： },....,,{ 21 suuuu = (6) 对样本集合 mi 中的任一个样本 kmi , ，利用正态 分布规律扩散到u 中的所有点 bu ，即： ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − −= 2 2 , 2 )( exp 2 1 )( h ui h uf bkmbk π (7) 其中， h 称为扩散系数，即组间宽度参数，这个参 数决定了密度函数的平滑程度， h 越大密度越平 滑，选择一个合适的组间宽度参数是非参数核密度 估计的重要步骤。在实践中，选择组间宽度的方法 有交叉验证法、silverman 的“经验法则”和插入法 11 。本文选择 silverman 的“经验法则”，根据 barry g k 和 alan p k 的工作 12，优化的参数 h 为： 5/1 ˆ06.1 l h σ = (8) 其中， ⎭ ⎬ ⎫ ⎩ ⎨ ⎧ = 34.1 ,minˆ q sσ (9) σ̂ 为样本标准差， q 为四分位数间距，即第 75 百分 位点样本值减去第 25 百分位点样本值。在实际工作 中，由于数据常常偏离正态分布，式（8）中的范围 因子 1.06 降到 0.9 效果较好 13。 令： ∑ = = s b bkk ufc 1 )( (10) 则： k bk bi c uf u km )( )( , =μ (11) 令： ∑ = = l k bib uuq km 1 )()( , μ (12) 再令： ∑ = = s b buqq 1 )( (13) 易知： q uq up bb )( )( = (14) 就是样本空间u 中样本点 bu 的频率值，可作为概率 的估计值。 利用分县分月的过程平均降雨量样本，采用 silverman 的“经验法则”计算出最优组间宽度 h ， 取 100=s 运用非参数核密度的信息扩散模型，逐一 拟合各县各月的过程平均降雨量核密度。图 6 是地 处东北三省中心的三个市（铁力市、九台市和灯塔 市）的过程平均降雨量核密度拟合图。图中可以看 出，三个省 7、8 月的过程平均降雨量大于其他月 份。从 6 个月过程平均降雨量分布的总体趋势上 看，辽宁省大于吉林省和黑龙江省，这与辽宁省临 近渤海受海洋影响降雨量丰富的事实相符。   (a) 黑龙江省铁力市   (b) 吉林省九台市   (c) 辽宁省灯塔市 图 6 分月过程平均降雨量概率拟合图 published by atlantis press copyright: the authors 205 sijian zhao, qiao zhang 4.3. 分县分月农作物洪涝风险评估 4.3.1. 风险评估 依据风险的定义，农作物洪涝风险可以表达为 洪涝造成农作物的期望受灾率。根据该定义，分县 分月的农作物洪涝风险 ctr 可表达为： ( ) ct s b bbctct uyupyer ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ ⋅== ∑ =1 )()( (15) 其中， c 表示县级行政区单元， t 表示月份， e 表示 受灾率 y 的期望， p 表示过程平均降雨量 bu 出现的 概率， y 表示过程平均降雨量 bu 造成的受灾率。利 用分县分月的过程平均降雨量核密度统计结果 （14），结合农作物洪涝脆弱性函数（4），采用公 式（15）便可计算出东北三省分县分月的农作物洪 涝风险，并以风险图方式进行展示，如图 7 所示。 4.3.2. 风险时空差异分析 从图 7 可以看出，东北三省农作物洪涝风险在 时空维上存在显著的差异。 4 月份，东北三省的农作物洪涝风险总体水平 较低，风险均值在 0.09 至 0.14 之间（黑龙江省 0.091，吉林省 0.105，辽宁省 0.140）。在黑龙江 省，除 16 个县级行政区风险值大于 0.10 外，剩余的 70 个县级行政区的风险值处在 0.05 至 0.10 之间，风 险值排在前 2 位是齐齐哈尔市市辖区（0.1227）和 泰来县（0.1190）。在吉林省，有超过半数的县级 行政区（28 个，共 48 个）的风险值大于 0.10，且集 中分布在东部地区，风险值排在前 2 位是集安市 （0.1489）和通化县（0.1441）。相比较，辽宁省风 险水平较高，所有 58 个县的风险值都超过 0.10，其 图 7 东北三省农作物洪涝分月风险图 published by atlantis press copyright: the authors 206 risk assessment of crops induced by flood in the three northeastern provinces of china on small space-and-time scales 中有 17 个县级行政区的风险值超过了 0.15 且集中分 布在辽宁省的东南部地区，风险值排名前 2 位的是 鞍 山 市 市 辖 区 （ 0.1780 ） 和 丹 东 市 市 辖 区 （0.1772）。4 月份，风险重点防范的区域应在辽宁 省的东南部。 5 月份，东北三省的农作物洪涝风险水平较 4 月 份有所提升，风险均值升到 0.10 至 0.17 之间（黑龙 江 0.114，吉林省 0.136，辽宁省 0.166）。黑龙江省 的绝大部分县级行政区（66 个）风险值在 0.10 至 0.15 之间，其中海伦市和克山县的风险值最高，分 别为 0.1443 和 0.1355。吉林省全境内所有县级行政 区的风险值都超过了 0.10，另有 12 个县级行政区的 风险值超过 0.15，集中分布在东南部，其中集安市 和通化县的风险值最高，分别为 0.1832 和 0.1745。 辽宁省内有 5 个县级行政区风险值超过 0.20，分布 在东部，36 个县级行政区风险值在 0.15 至 0.10，分 布在中部，17 个县级行政区风险值在 0.10 至 0.15， 分布在西部，其中鞍山市市辖区和丹东市市辖区风 险值依旧最高，分别是 0.2131 和 0.2094。5 月份风 险重点防范的区域是吉林省东南部和辽宁省东部。 6 月份，东北三省的农作物洪涝风险水平仍在 上升，风险均值升到 0.14 至 0.20 之间（黑龙江 0.149，吉林省 0.172，辽宁省 0.195）。黑龙江省中 部的 32 个县级行政区风险值在 0.15 至 0.20 之间， 其余县级行政区风险值在 0.10 至 0.15 之间，其中位 于中部地区的海伦市和庆安县的风险值最高，分别 是 0.1996 和 0.1912。吉林省西北部的 5 个县级行政 区风险值在 0.10 至 0.15 之间，其余 43 个县级行政 区风险值大于 0.15，另有 3 个县级行政区风险值大 于 0.20，其中位于中东部地区的靖宇县和四平市市 辖区的风险值最大，分别为 0.2074 和 0.2030。相比 较 4、5 月份，6 月份黑龙江省和吉林省的最高风险 值之间差距缩小。辽宁省有 22 个县级行政区的风险 值超过 0.20，分布在东部、南部和中西部，另有 36 个县级行政区的风险值在 0.15 至 0.20 之间，其中， 普兰店市和抚顺市市辖区的风险值最高，分别为 0.2343 和 0.2260。6 月份，风险重点防范的区域是黑 龙江省的中部，吉林省的东部，辽宁省的西部、东 部和南部。 7 月份，东北三省的农作物洪涝风险达到了最 高水平，风险均值达到了 0.21 至 0.26 之间（黑龙江 0.213，吉林省 0.233，辽宁省 0.259）。黑龙江省有 4 个县级行政区的风险值超过 0.25 且分布在中北 部，48 个县级行政区的风险值在 0.20 至 0.25 间且集 中分布在中西部，剩下的 24 个县级行政区的风险值 在 0.15 至 0.20 之间，其中绥化市市辖区和明水县的 风险值最高，分别为 0.2773 和 0.2619。吉林省有 13 个县级行政区风险值超过 0.25 且分布在中部，剩下 的 35 个县级行政区风险值在 0.20 至 0.25 间，其中 梅河口市和集安市的风险值最大，分别为 0.2765 和 0.2706。7 月份，黑龙江省与吉林省之间的高风险值 相当。辽宁省全境内的风险值均超过了 0.20，有 35 个县级行政区的风险值在 0.25 值 0.30 之间且分布在 东部地区，另有 1 个县级行政区的风险值甚至超过 了 0.30，即宽甸满族自治县的风险值达到 0.3233， 其次是凤城市的风险值达到 0.2962。7 月份，风险 重点防范的区域是黑龙江省的中北部，吉林省的中 部和辽宁省的东部。 8 月份，东北三省的农作物洪涝风险开始衰 减，但总体水平高于 6 月份，风险均值在 0.20 至 0.25 间（黑龙江 0.201 ，吉林省 0.222 ，辽宁省 0.245）。黑龙江省有 38 个县级行政区的风险值在 0.20 至 0.25 之间且分布在中部，其余的县级行政区 的风险值在 0.15 至 0.20 之间，其中依兰县和牡丹江 市市辖区的风险值最高，分别为 0.2464 和 0.2317。 吉林省有 6 个县级行政区的风险值高于 0.25 且分布 在东南部与辽宁省交界的地区，35 个县级行政区的 风险值在 0.20 至 0.25 之间，剩余 7 个县级行政区的 风险值在 0.15 至 0.20 之间，其中通化县和通化市市 辖区的风险值最大，分别是 0.2735 和 0.2582。辽宁 省全境范围内的风险值在 0.20 以上，有 26 个县级行 政区风险值在 0.25 至 0.30 之间且分布在中东部，其 中清原满族自治县和鞍山市市辖区的风险值最高， 分别为 0.2958 和 0.2818。8 月份，风险重点防范的 区域是黑龙江省中部，吉林省东南部和辽宁省的中 东部。 9 月份，东北三省的农作物洪涝风险进入全面 的衰退期，总体水平甚至低于 5 月份，风险均值降 至 0.14 至 0.20 间（黑龙江 0.143，吉林省 0.154，辽 宁省 0.200）。黑龙江省全境内风险值高于 0.10，有 23 个县级行政区的风险值在 0.15 至 0.20 之间且呈现 零散分布，其中庆安县和富裕县的风险值最大，分 别为 0.1773 和 0.1724。吉林省的风险值介于 0.10 至 0.20 之间，其中 28 个县级行政区的风险值大于 0.25 且分布在中南部，通化县和四平市市辖区的风险值 最大，分别为 0.1949 和 0.1907。辽宁省有 30 个县级 行政区风险值在 0.15 至 0.20 之间，27 个县级行政区 风险值在 0.20 至 0.25 之间，1 个县级行政区风险值 大于 0.25，高风险值依旧集中在中东部，其中营口 市市辖区风险值为 0.263，宽甸满族自治县风险值为 0.2412，排在前两位。8 月份，风险重点防范的区域 应在辽宁省的中东部和东南部。 5. 总结 本文以东北三省为研究区，在洪涝灾情数据、 气象数据和种植面积数据的基础上，通过构建气象 published by atlantis press copyright: the authors 207 sijian zhao, qiao zhang 插值模型、农作物洪涝脆弱性模型和非参数核密度 的信息扩散模型，开展了以县、月为单元的小时空 尺度下农作物洪涝风险评估研究，并在评估结果的 基础上探讨了风险的时空差异特征。本研究是一个 全新的尝试，它为灾害样本严重不足情况下的小时 空尺度风险评估提供了一个解决方案。 另一方面，在研究中发现，农作物洪涝脆弱性 模型是整个风险评估的关键，本研究仅建立了单变 量（降雨量）与受灾率之间的回归关系，关系合理 但并不理想。在接下来的研究中，作者将会深入探 讨如何建立多因子农作物洪涝脆弱性关系模型来进 一步完善风险评估。 致谢 本研究得到了国家自然科学基金项目(40901274)的资 助。 参考文献 1. zhang qiao and wang ke, met hods and models in risk analysis of crops’ production, agricultural outlook, 8 (2007) 7-10 张峭，王克. 农作物生产风险评估的方法和模型[j]. 农 业展望, 2007, 8:7-10 2. huo zhiguo, li shikui, wang suyan, et al. stud y on the risk evaluation technologies of main agrometeorological disasters and their application, journal of natural resources, 18(6) 2003 692-702 霍治国, 李世奎, 王素艳等. 主要农业气象灾害风险评 估技术及其应用研究[j]. 自然资源学报, 2003, l8(6): 692-702 3. li yunhui, he yimei, yang zisheng. analy sis on reduced grain yield from agricultural natural disaster in jinsha river basin of yunnan province. journal of mountain science. 20(sup.) 2002 43-48 李云辉，贺一梅，扬子生. 云南金沙江流域因灾减产 粮食量分析[j]. 山地学报. 2002.20 (增刊): 43-48 4. zhang xing, zhang chungui, wu juxin , et al. risk assessment of yield losses from a gro-meteorological disasters in fujian province, journal of natural disasters, 18(1) 2009 90-94 张星，张春桂，吴菊薪等. 福建农业气象灾害的产量 灾损风险评估[j]. 自然灾害学报，2009，18(1): 90-94 5. http://www.cma.gov.cn/qxxw/t20071115_220733.phtml 6. http://www.tech-food.com/news/2012-314/n0732086.htm 7. zhao junfang, yan xiaodong, jia gensuo. simulating the responses of forest net primary productivity and carbon budget to climate ch ange in northeast china. acta ecologica sinca, 28(1) 2008 92-102 赵俊芳，延晓冬，贾根锁. 东北森林净第一性生产力 与碳收支对气候变化的响应[j]. 生态学报，2008， 28(1): 92-102 8. chen shupeng, lu xuejun, zhou chenghu. introduction to geographical information systems. science press, beijing, 2009, 119 陈述彭，鲁学军，周成虎. 地理信息系统导论[m]. 北 京: 科学出版社，2002: 119 9. zhou yao, wang jingai. a review on development of vulnerability curve of natural disaster. advances in earth science, 27(4) 2012 435-442 周瑶，王静爱. 自然灾害脆弱性曲线研究进展[j]. 地球 科学进展, 2012, 27(4): 435-442 10. wang lihong, yang ruihua, tian zhihong, et al. maize grp rate of pr emium deciding by nonprarametric kernel density: a case study on ang uo city,hebei province. journal of china agricultural university, 12(1) 2007 9094 王丽红，杨华，田志宏, 等. 非参数核密度法厘定玉米 区域产量保险费率研究—以河北安国市为例[j]. 中国 农业大学学报，2007，12(1): 90-94 11. guo yingchun, yan yiling, wang wei, et al. the methods of agricultu ral natural risk ev aluation and regional insurance premium calculation. quarterly journal of applied meteorology. 5 (1998) 232-237 郭迎春，闫宜玲, 王卫, 等. 农业自然风险评估及区域 农 业 保 险 费 率 的 确 定 方 法 [j]. 应 用 气 象 学 报 ， 1998(5):232-237 12. barry k g, alan p k. nonpa rametric estimation of cro p yield distributions: implications for rating group-risk crop insurance contracts. amer j agr econ, 80 (1998) 139-153 13. alan p k, b arry k g. nonpar ametric estimation of crop insurance rates revisited. amer j agr econ, 83 (2000) 463-478 published by atlantis press copyright: the authors 208 zk-entity risk mechanics final approved 120629 entity risk mechanics milík tichý faculty of civil engineering, czech technical university solinova 7, 16629 praha 6, czech republic milik.tichy@volny.cz www.fsv.cvut.cz abstract the actual theory of risk is based on phenomenological approach. however, it seems feasible to start thinking about entities exposed to hazards in a similar way as about structural systems exposed to loads. concepts of structural mechanics and structural reliability, common in design of constructed facilities, can be implemented into the risk theory, which, in this way, might be upgraded. as a result, new tools could be given to the project risk management enabling better risk predictions than available today. the aspects of the problem are demonstrated by comparing two specific entities – a building structure exposed to loads, and a company exposed to hazards. important conceptual analogies between the two entities are identified. keywords: risk theory; structural design; entity; structure; company; risk mechanics 1. concepts and the reference bases the principles of the actual risk theory, as formulated by kaplan and garrick [1], are based l on identifications of hazards, hz, and hazard scenarios, sc, and l on de moivre's concept of risk (1711, in de mensura sortis, see bernstein, [2]), formulated, in today words, as the product of the hazard scenario probability, pr(sc), and the respective damage, dm(sc). the identification of hazards and hazard scenarios is based on experience and judgment while the probabilities and damage are established from a priori observations of past events or from a posteriori forecasting of the future. experience is obviously the governing factor in risk analysis. thus, it can be stated that the actual risk theory is essentially phenomenological. a phenomenological approach is typical for the starting periods of whatever science branch. sooner or later, it will have been followed by elaborate theories based on generalizing mathematical models. doubtlessly, the science of risk can also be expected to pursue a similar track. a wider, more extensive and more comprehensive theoretical framework for solving systems "hazards– risks–risk exposed entity" than that being now dealt with can be foreseen. the concept of entity used in the author's following propositions can be defined as: any concrete or abstract thing of interest, including associations among things; for example, a person, object, event, or process that is of interest in the context under consideration, and about which data may be stored in a database.1 thus, "entity" covers universally all objects and processes of any kind – material, non-material, or mixed.—a few examples of entities: a manufacturing company, a vacation travel, academic staff of a university, a political party, a philosophical system. the risk analysis terms, in particular hazard, hz, hazard scenario, sc, and risk, rs, will be applied as outlined by kaplan and garrick [1]—other concepts 1 see definitions through google.com journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 107-113 published by atlantis press copyright: the authors 107 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine received 1 march 2011; accepted 7 may 2012 milík tichý and terms have been recently discussed – see merkelsen [3]. time and space shall be specified in any solutions, and so all variables, conditions, and concepts in general, mentioned in the following, shall be always referred to time and space, or possibly also to other reference bases. two parameters of time are of interest in analyses of whatever systems: a reference period, tref, and its timeline position, t0. similarly as time, the space considered shall be specified in terms of a reference space, ωref, situated at ω0. space shall be conceived not only geometrically. in the risk analyses space may have various meanings: geometrical space, management space, marketing space, operation space, and others. both reference bases can be continuous or discrete: tref can consist of several time periods (e.g., winter periods, flight time between two airports), ω ref can be considered as a set of partial spaces (e.g., classrooms of a school building, departments of a company). any of the reference parameters, tref and t0, ωref and ω0, though usually assumed by fixed deterministic values, can be time-dependent, and space-dependent as well. moreover, they can be random. however, the features of discontinuity, hyperdependences, and randomness of the reference parameters will not be discussed here. 2. structure and company for sake of simplicity, only two model entities – of substantially different and not related kind – will be compared to show the author's conception. constructed facilities, c.f., i.e. buildings, bridges, dams, and many other engineering works, are exposed to environmental effects that can be of various natures: physical, chemical, economic, political, and many others. c.f.'s are expected to deal with such effects adequately. this means, in technical words, that the respective c.f. will remain fixed in its space and time, and will be sufficiently firm to meet the requirements imposed by its users2. these properties are provided by 2 other engineering facilities exist whose main role is to move – the mechanical facilities (cars, ships, dvd drive mechanics, etc.) –, to transfer energy and information, etc. the carrying structures – foundations, columns, beams, slabs, roof trusses, etc. – joined in one or more systems. structure will be taken as one of the two model entities in this study. basic features of structures exposed to loads will be compared here with basic features of another model entity: company, exposed to hazards. a company is described in predominantly management terms, though some amount of physical description cannot be escaped. companies and structures are systems that substantially differ by their inherent natures; yet, both have surprisingly much in common. although the two model entities are unrelated, the following comparisons will show striking analogies between both. the left-hand column of the following broken table is reserved for engineering concepts concerning structures, while on the right-hand side risk-related concepts referred to companies are outlined. 3. loads and hazards observe various parallel aspects of loads and hazards: passive role aspect structures are exposed to loads – fig. 1 companies are exposed to hazards – fig. 2. imposed loads self weight loads structure firm ground surface soft ground and fill reactions fig. 1. a simple structure exposed to imposed and self-weight loads. beams, columns, and footings transmit the load-effects into the firm ground, and the structure is loaded by upside reactions. all loads acting on the structure must be in equilibrium. published by atlantis press copyright: the authors 108 entity risk mechanics hazards: generatedexternal internal fig. 2. the organization chart of a company exposed to external and internal hazards and generating hazards to its environment. active role aspect structures generate loads affecting other structures and ground. companies generate hazards to other companies or to other entities (e.g., to the environment). time aspects loads on structures are usually classified into two principal groups: ● permanent loads, ● transient loads. hazards threatening companies can be classified in the same way: ● permanent hazards, ● transient hazards. utility aspects ● self-weight loads (no terrestrial structure can exist without self-weight), ● imposed loads (live load, dead load, transient load, wind load, snow load, etc.) ● existence-based hazards (company cannot avoid them, whether they were active or not), ● activity-based hazards (affecting the company or created by the company as a result of its activity). further aspects, divisions and subdivisions of loads and of hazards, might be easily demonstrated, and the respective classification would become more detailed. 4. load combinations and hazard combinations in current structures loads never occur in isolated form. as soon as two or more loads occur simultaneously, we talk about a load combination. as a rule, several load combinations have to be investigated in structural analysis and design the effect of a single load can be either amplified in combination with other loads, or, on the contrary, diminished. a company is exposed to a set of hazards, many of them can occur simultaneously, and hazard combinations should be faced. again, amplification or lessening of the effects of a particular hazard is possible in a hazard combination. the hazard combination feature is typical in the insurance theory and practice where combined hazards and the ensuing risks are dealt with under different schemes. 5. load-effects and hazard scenarios structures react to loads in their appropriate way. the resulting load-effects depend upon geometric and physical properties of the carrying system. under various arrangements and levels of load the relative distributions of loadeffects are, in general, not identical since they depend upon the response of the structure. one or more hazard scenarios can be identified for each single hazard or for a hazard combination. a set of hazard scenarios must be considered when analyzing a specific hazard affecting a company. the more complex the company, the more scenarios are possible for any hazard, some of them dominating the others. hazard scenarios are the possible results of hazards imposed or proper, and of the company's hazard response (or risk response). published by atlantis press copyright: the authors 109 milík tichý 6. structure and company properties structures are made of specific materials, which possess sets of properties (strength, ductility, durability, etc.). the result of these material properties combined with the arrangement of a given structure in a system consisting of carrying members (beams, columns, slabs, arches, etc.) provides for structure's resistance with respect to loads and stiffness under their action. companies are built up of units according to various patterns, each unit (e.g., accounting department, shipping and delivering, ceo's office) having a defined specific role in the respective company's system. every unit thus plays a role in company's resistance and stiffness with respect to hazards. of course, structures can be conceived as riskologic entities exposed to hazards, too. some of structureaffecting hazards stem straight from physical loads, yet there are many hazards that have nothing common with the load category (e.g., hazards due to bad maintenance). on the other hand, a company can be subjected to "loads" that cannot be considered hazards – e.g., debt loads, contractual loads. 7. kinematic compatibility equations and constitutive laws 7.1. structure how and why structures remain fixed and firm? this crucial double-question had not been asked in the beginnings of the mankind's construction activities; everybody had been happy that buildings, bridges, and other facilities did not systematically break down or fall apart. if, nevertheless, something did happen, the immediate observations were transferred into current practice as experience without going into the background of the event. no theories did exist to generalize the phenomena. yet, questions started to be raised as soon as extrapolations of experience became dangerous. answers had been found in due course in a couple of centuries, and, as a result, structural mechanics based on physical laws has developed into a complex system of knowledge. stresses in a fixed and firm structural system must be balanced, and all strains and displacements must be mutually compatible. this condition can be symbolically expressed by a set of kinematic compatibility requirements: ∀t∈tref , ∀ω∈ ωref : f(σ, ε) = 0 (1) where σ represents the stress state, and ε the strain and displacement state. note the time and space quantifiers. whenever the equilibrium of forces is lost due to loss of support and compatibility, the structure or its parts start moving until a new equilibrium has been reached. when the ultimate equilibrium state is exceeded, the system collapses and, as a result, the structure ceases to exist as such. structural mechanics does not present the kinematic compatibility requirement in the above symbolic form. as a rule, a system of equations has to be formulated where physical dependences between strains on one side and stresses on the other must be known or assumed. these dependences are usually denoted as constitutive laws of materials (also called constitutive equations). there always is a relationship between stress and strain in a material. it can be defined mathematically in terms of a function g(σ, ε) based on experimental observations, the graph of which is often called the "stress-strain diagram", ssd. this function can have various forms in dependence upon the properties of the material applied, loading rate, temperature, and many other factors. for illustration, two simple ssd's, typical for many materials, are shown in fig. 3.3 7.2. company an analogous set of concepts can be followed considering the company model. however, a specific concept shall be introduced, viz. the concept of concern, cn. risk generates concern in individuals and groups at various time and space positions (managers, family heads, farmers, sailors, parliaments, nations, etc.). 3 for certain materials under specific conditions ssd can be assumed linear, and is well known from elementary physics as hooke's law. published by atlantis press copyright: the authors 110 entity risk mechanics concern can be measured by methods of psychology, and dependence between risk and concern can be empirically found for each case. such dependence can be taken as the constitutive law of the company. it can be graphically presented as a risk-concern diagram, rcd, fig. 4. s 0 0 18% e a ulte s 4% e s ulte s (a) (b) fig. 3. stress-strain diagram, ssd, of a structural material: (a) – material softens with growing strain, (b) – material hardens with growing strain. rs rsult cncnult 0 rs cnult 0 rsult cn (a) (b) fig. 4. risk-concern diagram, rcd, of a company: (a) – risk-averse company, (b) – risk-reverse company. it can be approximately said that concern, cn, covers the qualitative aspect of a hazard situation while risk, rs, expresses the quantitative aspect. again, rcd can be mathematically described by a function, g(rs, cn), which seems to be a challenging task for future investigations. now, a general risk compatibility requirement can be formulated: ∀t∈tref , ∀ω∈ ωref : f(rs, cn) = 0 (2) all company risks, i.e., downside risks and upside risks as well, should enter the requirement. in fact, an equilibrium and compatibility of risks is a condition sine qua non for the survival of any entity. observe that company a with rcd according to fig. 4a is more risk-sensitive than the company b with rcd of the type in fig. 4b. while the company a gets increasingly concerned with small increments of risk, the company b does not mind, and its concern growth slows down. it is evident that the respective constitutive law depends on the company and its environment in general concept. note that similar considerations can be made for the relationship between hazard severity, sv, and concern, cn. many situations exist where the hazard severity, not the risk, does govern decisions. this particularly happens when important hazards with very small probability of getting materialized are dealt with, and the precautionary principle thinking becomes a necessity. 8. limit states when the intensity of loads increases, or the structural resistance and stiffness diminish 4 , an ultimate equilibrium is reached beyond which any change in load parameters results in collapse. the ultimate equilibrium state is one of several limit states that are a basic concept in the design of structures. since mayer [4], 1926, the theory of limit states had successively developed into a now commonly stabilized system, defined, e.g., in structural design codes. again, analogies between the two model entities examined can be observed: two basic families of structural limit states are distinguished in the design of structures: ● serviceability limit states, referred to the reliability of structures under "average conditions"; a company's behavior under current conditions is definitely different from its behavior in an exceptional situation. some functions get intentionally or automatically suppressed, some must be started to 4 the two processes can be mutually dependent. published by atlantis press copyright: the authors 111 milík tichý this group mainly concerns various forms of displacements and deformations of structural systems, ● ultimate limit states, dealing with structures under extreme loads. keep the company running. thus, a set of company limit states, analogous to structural limit states, can be distinguished. 9. structural reliability and risk acceptability the loads and load-effects, and the material properties as well, vary in certain ranges. in general, these variations are random in time and space. therefore, a simple deterministic request for a fixed and firm structure does not suffice in the design and assessment of c.f. what we are interested in, is how far from the total or partial failure the given structure under the expected load system is. in other words, we want to know whether the structure will be and will remain reliable, i.e. safe, durable, etc., during its assumed use in the time and space envisaged. in practice, we want to design the structure in such a way that its system remains fixed and firm with a certain margin of reliability, z(t, ω), during the reference period, tref, and in the reference space, ω ref, specified. to meet this, a probability-based structural reliability requirement, psrs, shall be fulfilled (see, e.g., tichý [5]): ∀t∈tref , ∀ω∈ ωref : pr(z(t, ω) ≤ 0) ≤ pfd (tref , ωref ) (3) where pfd is the design value of the failure probability, which is a decision-based parameter, often depending on the reference period and reference space. the magnitude of pfd is established on a set of miscellaneous considerations – economic, psychologic, etc. – of deterministic and statistical character.—for simplicity, the time and space arguments, (t, ω), are not indicated in the equations that follow. the margin of safety, z, can be expressed in various ways, in dependence upon the situation considered. in general, z is formed by a set of elementary random variables describing loads, material properties, geometry of the structure, and also the respective environmental properties the c.f. is exposed to. thus z is a time-and-space-dependent random variable.—in simplification, z can be understood as the highest load the respective structure is able to carry. if now the analogy track is followed, a probabilitybased risk acceptability requirement, prar, can be formulated for a company: ∀t∈tref , ∀ω∈ ωref : pr(rsacc ≤ 0) ≤ padm(tref , ωref ) (4) where rsacc is the acceptable risk, and padm is an admissible probability, the latter being dependent on tref and ωref. the acceptable risk is defined by (cf. fig. 5) rsacc = rscap – rsexp (5) where rscap is the risk capacity, i.e., the risk the company is able to carry at ultimate conditions, and rsexp is the expected risk the company will supposedly be exposed to from external and internal hazards. the expected risk shall be established by analyzing the company's risk portfolio, and the risk capacity depends upon company's properties and its environment in broad sense. rs rscap rsexp rsacc 0 1t 2t 3t a b t fig. 5. risk capacity, rscap, and expected risk, rsexp, are mutually independent random processes. both can be continuous in time with occasional random or non-random discontinuities at t1, t2. at t2 the acceptable risk, rsacc, becomes negative for a certain period, a. in case the risk capacity process is not governable by the company's management, a temporary solution, b, must be found in reducing the expected risk, rsexp .— similar situations can be demonstrated for a structure when it gets temporarily overloaded at some period of its lifetime. note that, for simplicity, prar is presented here in scalar form while it definitely is a multidimensional requirement. both the expected risk and the risk capacity are timeand-space dependent and random, and so is the acceptable risk. for illustration, fig. 5 shows the time dependence of the expected risk, rsexp, and of the risk capacity, rscap. observe that the curves representing the two quantities may not be continuous. at t1 there is a jump rise of the expected risk, and at t2 the risk capacity suddenly drops; the jumps can be caused, e.g., by a change in regulations that govern the company's activity and the company's environment. measures must be published by atlantis press copyright: the authors 112 entity risk mechanics taken to cope with such situations and the expected risk must be treated to pass periods of a possible company failure. the admissible probability, padm, refers to the limit state when the company becomes a hazard to itself and to its environment. while the concept of pfd in psrs is more or less well understood in the structural engineering community, not too much can be said about padm at this moment. the risk resistance of companies in many countries is often regulated by governments or by international bodies, which are, of course, risk-exposed entities themselves. e.g., in the banking area, basel iii is a well known regulatory document, similarly as solvency ii is for the insurance industry (see [6], [7], and [8]). if the structure-company analogies would be developed, padm may become a basic regulatory parameter. 10. conclusions (1) a principal analogy between structure and company can be identified: loads on structures and hazards affecting companies can be considered parallel concepts. (2) the ensuing analogies between concepts of structural mechanics and concepts of risk analysis can be concised into the following list: structural mechanics risk analysis load hazard load effect hazard scenario stress risk strain concern reliability requirement risk acceptability requirement stress-strain diagram risk-concern diagram (3) the analogy approach can be applied in comparing structures with other entities of very different kind. (4) principles of structural mechanics and structural reliability theory can be applied in risk analyses of financial institutions of different kind, political bodies, biological systems, and scores of many other material and non-material entities of surprisingly differing kind. (5) many research topics are offered by the analogy approach. theoretical as well as experimental research will be needed to define mathematical models of entity risk behavior. (6) there is a good chance that an entity risk mechanics be built, analogous to the existing structural mechanics. (7) the actual risk theory can be raised from its phenomenological platform to a higher level. note that in the foregoing discussion the term "company" can be simply substituted by "entity". acknowledgements i would like to thank professor jiří šejnoha for his insightful suggestions concerning some aspects of the analogies presented. however, any errors are my responsibility. references [1] s. kaplan, b. j. garrick, on the quantitative definition of risk, risk analysis, 1(1) (1981), 11-27 [2] p. l. bernstein, against the gods (john wiley, new york, 1996) [3] h. merkelsen, the constitutive element of probabilistic agency in risk: a semantic analysis of risk, danger, chance, and hazard, journal of risk research, 14(7) (2011), 881-897 [4] m. mayer, sicherheit der bauwerke und ihre berechnung nach grenzkräften anstatt nach zulässigen spannungen (julius springer, berlin, 1926) [5] m. tichý, applied methods of structural reliability (kluwer academic publishers, dordrecht, 1994) [6] a global regulatory framework for more resilient banks and banking systems – basel iii, (basel, bank for international settlements, 2010, rev. 2011) [7] international framework for liquidity risk measurement, standards and monitoring – basel iii (basel, bank for international settlements, 2010) [8] directive 2009/138/ec of the european parliament and of the council of 25 november 2009 on the taking-up and pursuit of the business of insurance and reinsurance – solvency ii published by atlantis press copyright: the authors 113 microsoft word the meteorological disaster risk assessment based on the diffusion mechanism the meteorological disaster risk assessment based on the diffusion mechanism guo shujun hebei province meteorological bureau, shijiazhuang 050021, china e-mail:gsjun888 @sina.com abstract at present, the study on meteorological disasters risk is mostly for single hazard-affected body. with the economic and social development in recent years, meteorological disaster risk is appearing more "system" characteristics. meteorological disasters not only impact on direct hazard-affected bodies, but also on other bodies through the direct victims. in general, the influence of the direct victims transmits to other bodies in some diffusion ways, which is rather different from the disaster chain, which we well known. in this paper, we propose the concept of diffusion of meteorological disaster risk, study the diffusion mechanism of the meteorological disaster risk, make a simulation system of risk diffusion analysis, and provide the analysis method and the calculation model for more reasonably and more comprehensively understanding the diffusion of meteorological disaster risk. taking blizzard of north china in early winter 2009 as an example, we construct a risk diffusion network of the low temperature and blizzard disaster, systemically analyze what would be affected by the disaster, and give the scene expression of risk with prompt function, which is useful for government and relevant departments to make the decisions in dealing with meteorological disaster risk. keywords: meteorological disaster, risk diffusion, diffusion network, simulation 基于风险扩散机制的气象灾害风险评估 郭树军 河北省气象局，石家庄 050021 摘要：目前有关气象灾害风险的研究，大多是针对单承灾体的风险分析。随着经济社会的发展，近年来气 象灾害风险呈现出“系统型”的特点。气象灾害不仅对直接承灾体产生影响，还通过直接承灾体对其它相 关的承灾体产生影响，通常以扩散方式传导，与人们熟知的灾害链并不一样，相关研究较少。本文提出了 气象灾害风险扩散的概念，揭示了气象灾害风险的扩散机理，建立了风险扩散分析的仿真模拟系统，为较 正确和更全面地认识气象灾害风险扩散现象提供了分析方法和计算模型。以 2009 年华北初冬暴雪为例进行 了风险扩散的过程分析，建立了低温暴雪灾害的风险扩散图，对低温暴雪灾害可能影响的主要领域进行了 系统分析，给出了具有风险预警提示的情景表示，为政府和有关部门应对气象灾害风险提供了决策依据。 关键词：气象灾害，风险扩散，扩散网络，仿真模拟 1. 引言 按照世界气象组织的统计，气象灾害约占自然 灾害的 70%左右。随着经济社会的发展，气象灾害 的连锁效应显著，连锁效应造成的影响有时会比灾 害本身造成的损失还要大，气象灾害风险存在扩散 效应 [1-3] 。目前的气象灾害风险研究虽然取得了不少 的进展，但是，国内外大部分气象灾害的风险评价 工作都是基于“风险度=危险度×易损度”这一传统 的风险评价体系 [4-7] ，该评价体系主要针对单一承灾 体，并不能完全解释气象灾害风险在经济社会领域 中的扩散效应。 近年来，众多科学研究机构启动了综合风险的 相关研究计划,国际科学联盟（icsu）2008 年启动 了灾害风险综合研究的计划（irdr），提出在风险 和减轻灾害风险研究中，要对各种灾害或灾害链进 行多学科、多尺度的综合性深入研究 [8] 。2009 年 ihdp 启动了新一轮综合风险防范（ihdp-irg）的科 学研究计划,重点强调了灾害链风险评价模型的改进 和灾害风险情景模拟工具的完善。有学者利用基于 journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 124-130 published by atlantis press copyright: the authors 124 administrateur texte tapé à la machine received 9 september 2011 administrateur texte tapé à la machine accepted 27 april 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine guo shujun 宏观经济学的投入产出关系研究灾害的宏观效应 [910] 。但是，针对风险扩散机制的研究及应用较为鲜 见。 本文拟从风险产生的关键环节着手，进行气象 灾害风险扩散机理的研究，揭示气象灾害风险产生 和发展、扩散的全过程，为政府和风险管理相关部 门全面认识与有效管理风险提供理论依据和技术支 撑。 2. 气象灾害风险扩散机理 2.1. 定义 为了更好地描述气象灾害风险的扩散现象，给 出以下定义。 定义 1 气象灾害风险是由气象因素导致人类社 会造成损失的一种未来情景。 该定义是基于黄崇福的自然灾害风险的情景定 义提出的气象灾害情景定义，优点是为描述风险提 供了一个新的更加符合现实的表达方式，为气象灾 害风险扩散的情景表示提供了依据。 定义 2 气象灾害风险扩散是指气象致灾力作用 于承灾体后，由于经济社会的系统性和承灾体的关 联性，对其他承灾体产生的波及效应。 定义 3 风险节点是指受到气象致灾力直接作用 或间接作用的承灾体。 定义 4 气象灾害风险由一个风险节点向另一个 风险节点传播扩散的路径称为气象灾害风险扩散通 道。 定义 5 由风险扩散的起始点对直接承灾体产生 的扩散称为一级扩散，由直接承灾体对与其它相关 的承灾体产生的扩散称为二级扩散，以此类推，构 成一个分层级的扩散结构体，该结构体称为风险扩 散网络。 定义 6 由下级风险节点向上级风险节点进行的 扩散称为逆向风险扩散。 2.2. 气象灾害风险扩散的拓扑结构分析 根据气象灾害对相关领域的影响建立气象灾害 风险扩散物理模型，不失一般性，我们假设通过风 险专家建立了如下的气象灾害风险扩散物理模型， 如图 1 所示。 图中 v1 是风险的扩散源点，v1,v2…v7 表示风险 的各个节点，箭头表示风险的传递方向。例如：v1 到 v2、v5 是风险的传递，v5、v4 对 v1 有风险的逆向扩 散。下面给出风险扩散网络拓扑结构的层级分析算 法，识别扩散网络的扩散和逆向扩散关系。 v1 v2 v3 v4 v7 v6 v5 图 1. v1 作用于其它节点的气象灾害风险扩散与逆向 扩散物理模型 1）节点连通路径信息矩阵 根据风险节点间的关系，建立一个 n 维信息矩 阵，n 表示风险扩散图的风险节点个数。表示为：   nnij la   （1） 其中         连接无通路到若从风险节点 连接有通路到若从风险节点 jiji jiji ij vvvv vvvv l ,,0 ,,1 以图 1 为例，我们得到节点连通路径信息矩阵如式 （2）所示。                        0100000 0000100 0101001 1010101 1100010 0000000 0010010 7 6 5 4 3 2 1 7654321 v v v v v v v vvvvvvv a (2) a 可以看成由风险节点 i 到风险节点 j 是否有 直接扩散通路的描述。矩阵的第 i 行表示节点 i 到 所有节点有无扩散，矩阵的 j 列表示所有节点有没 有到节点 j 的汇聚。 2)风险扩散数据结构 风险扩散网络结构可以用节点集合和节点间关 系集合来表示， },,{ rdps  其中， ps 表示风险扩 散网络，d 表示节点集合，r 表示节点间关系集合。 为了将节点进行分级，进一步将风险扩散网络 细分为分层级的节点集合和带方向的扩散关系集 published by atlantis press copyright: the authors 125 the meteorological disaster risk assessment 合， },,,{ nrsrlps  其中，l 表示节点的层级集 合，sr 表示风险扩散关系集合，nr 表示风险逆向扩 散关系集合。 3）确定风险扩散网络的层级结构 遍历节点连通路径信息矩阵的第一行所有节 点，选取受致灾力节点 v1 直接作用的节点，将它们 建为一级扩散节点集合 l1，并将关系集合中的对应 关系移致 sr 中。 以一级风险扩散节点集合 l1 的每个节点扫描信 息矩阵，选取该节点直接作用的节点，将它们加入 到二级扩散节点集合 l2 中，如果节点已在上层级或 本层级中出现，则剔除，同时将关系集合中的对应 关系移致 sr 中，如果直接作用的节点出现在本级， 说明存在同级节点间的扩散，扩散关系要移致 sr 中。依次类推，重复以上操作，直至完成所有节点 的扫描，最后，在关系集合中保留的就是逆向扩散 关系集合 nr。 由此得到分层级的节点集合和分方向关系集 合。 l0={v1} l1={v2,v5} l2={v4,v6} l3={v7,v3} sr={ r12,r15,r54,r56,r47,r43,r63,r37} nr={r51,r41,r45,r76,r36,r32} 基于以上的分级，建立的风险扩散网络拓扑结 构如图 2 所示。 图 2. v1 作用于其它节点的气象灾害风险扩散模型 利用上述方法可以将风险扩散的物理网络转变 为由风险扩散和风险逆向扩散组成的逻辑分析结 构，为风险扩散计算奠定了基础。同理，按照逆向 风险扩散关系集合，建立的逆向风险扩散网络拓扑 结构如图 3 所示。 图 3. 逆向扩散模型 2.3. 风险扩散的计算 根据实际情况，给出两类风险扩散的计算方 法： 1)概率风险扩散的计算 上游节点风险乘以上下游节点间的风险扩散系 数得到上游节点对下游节点的风险扩散结果，假设 节点 i 是节点 j 的上游节点，那么由节点 i 扩散给节 点 j 的风险为： jiij spp  （3） 其中， ip 为上游节点 i 的风险值， jis  为节点 i 对节点 j 的风险扩散系数， jp 为节点 i 对节点 j 的风险扩散结果。 2)模糊风险扩散的计算 假设节点 i 是节点 j 的上游节点，如果节点间风 险扩散系数是模糊集合，则风险扩散的计算采用下 面公式： )},(),({sup),( jirpippjp i ii j   （4） 其中， ),( pipi 为上游节点 i 的模糊风险， ),( jir 为节点间的模糊系数， ),( pjpj 为节点 i 对节点 j 扩散的模糊风险。 2.4. 节点的风险融合 图 4 表示风险节点 r 有多个来源的风险,那么， 节点 r 的综合风险值由各个来源的风险值，按照给 定的权重相加求得。即： i n i i pwp    1 （5） v1 v2 v3 v4 v7 v6 v5 v1 v2 v3 v7 v6 v5 v4 published by atlantis press copyright: the authors 126 guo shujun 式中，wi 表示权重，pi 表示各风险源的风险值。 如果各个来源的风险等级不一致，则不能直接 融合。 图 4. 多来源风险的融合 3. 气象灾害风险扩散分析仿真系统 根据风险扩散的基本原理，利用面向对象的计 算机软件技术，建立了气象灾害风险扩散分析仿真 系统（如图 5 所示），能够进行定量与定性相结合 的风险扩散分析。气象灾害风险扩散分析仿真系统 主要包括风险扩散网络构建、风险分析计算以及风 险的情景表示等功能。 图 5.风险分析仿真系统界面图 4. 实例分析—以 2009 年华北暴雪灾害为例 4.1. 研究范围 研究区域为石家庄地区，研究的灾害为持续暴 雪，研究的领域为区域内的农业、蔬菜损失风险及 其对相关领域的影响。 4.2. 风险扩散的识别 根据暴雪的灾害影响按照“先面后点、先上后 下、横向交叉”的分析原则，进行灾害风险扩散的 识别，建立暴雪灾害的风险扩散网络。 暴雪首先对大田作物冬小麦、设施蔬菜等农 业，以及交通造成直接影响，此为一级风险的扩 散。冬小麦将受到霜冻减产的威胁，蔬菜大棚垮塌 对设施蔬菜的生产造成影响，降雪使交通受阻直接 造成运输不畅，此为暴雪的二级风险所在。蔬菜减 产和运输不畅会造成蔬菜涨价，运输不畅会还造成 日常生活用品的供给不足，引发涨价，粮食是基础 性战略物资，粮食的减产会给相关的产业造成影 响，此为三级风险扩散。蔬菜、日用品涨价，会给 中低收入的居民生活带来影响，此为四级风险扩 散。建立的扩散网络如图 6 所示。 4.3. 确立节点间的风险扩散关系 节点间风险扩散关系的确立是进行风险扩散计 算的基础，下面对风险扩散网络中的主要关系进行 说明。 图 6. 2009 年华北雪灾风险扩散的识别 1)冬小麦霜冻与粮食减产的关系 表 1 为从气象和农业部门收集的近年来冬小麦 霜冻与产量的资料。 i … n r 持续低温 暴雪 生活品涨价 食品制 农业 纺织品 城市交通 受阻 冬小麦霜 粮食减产 下游产业 影响 大棚垮塌 蔬菜减产 蔬菜涨价 菜农减收 运输不畅 交通受阻 中低收入居 民生活困难 published by atlantis press copyright: the authors 127 the meteorological disaster risk assessment 表 1.冬小麦霜冻年份的产量与减产率 年份 温度 (摄氏度) 单产 (公斤/亩) 平均单产 (公斤/亩) 减产率 （%） 2000 -5.1 362.3 375.4 -3.49 2001 -4.2 359.1 -4.34 2003 -5.5 371.8 -0.96 2004 -4.8 376.2 0.22 2009 -7.2 363.6 -3.14 鉴于资料样本少，采用信息扩散技术建立冬小 麦霜冻与产量的关系，得到冬小麦霜冻与粮食减产 的关系如表 2 所示。 表 2.冬小麦霜冻与减产的模糊关系 减产 5% 减产 4% 减产 3% 减产 2% 减产 1% 减产 0% -10 度 0.233 0.232 0.23 0.226 0.222 0.215 -9 度 0.404 0.405 0.403 0.399 0.393 0.386 -8 度 0.617 0.619 0.618 0.615 0.61 0.603 -7 度 0.828 0.831 0.831 0.83 0.827 0.822 -6 度 0.973 0.976 0.977 0.977 0.976 0.974 -5 度 1 1 1 1 1 1 -4 度 0.894 0.891 0.888 0.887 0.886 0.886 -3 度 0.692 0.687 0.682 0.679 0.677 0.676 -2 度 0.463 0.456 0.451 0.447 0.444 0.442 -1 度 0.266 0.261 0.256 0.253 0.25 0.248 0 度 0.131 0.128 0.125 0.122 0.12 0.119 2)粮食减产对其他相关行业的关系 依据石家庄统计局的投入产出分析表中各行业 对农业的直接消耗系数 [11] ，可知当地的农业对食品 制造及烟草加工业、农业、纺织业、造纸印刷及文 教用品制造业等行业的贡献较大（表 3）。 由于篇幅所限，其他风险扩散的关系建立过程 见文献 [12] ，这里不再赘述。 4.4. 基于风险扩散的系统风险计算 本案例中气象致灾力有三个输入条件，即：{温 度,雪量,致灾因子发生的概率}，下面以输入条件{6 度，30mm,0.9 }为例，说明主要节点风险扩散计 算的过程。 表 3. 2002 年石家庄市投入产出表中 8 个 主要行业对农业的直接消耗系数 序 号 行业 直接消耗 系数 01 食品制造及烟草加工业 0.243 02 农业 0.151 03 纺织业 0.089 04 造纸印刷及文教用品制造业 0.057 05 木材加工及家具制造业 0.055 06 住宿及餐饮业 0.038 07 交通运输及仓储业 0.033 08 服装皮革羽绒及其制品业 0.029 1)霜冻对粮食减产的风险 根据霜冻与粮食减产的模糊关系，根据公式 （4）进行模糊推理，当出现-6 度时，粮食减产率 约为 2(%)，对应的风险值为 0.977。那么，由严重 霜冻风险（-6 度）为 0.9 导致的粮食减产 2%的风险 是 0.9×0.977 =0.879。 2) 计算暴雪对蔬菜大棚垮塌的风险 采用公式（4）进行模糊推理，当降雪量 30mm 时，由暴雪对蔬菜大棚垮塌率的关系可求得大棚的 垮塌率为 23(%)，从模糊关系中取接近的 24%，即暴 雪 30mm 造成大棚垮塌率 24%的可能性最大，对应的 风险值为 0.8，则大棚垮塌 24%的风险为：0.9× 0.8=0.72。 同理，可以计算其他节点的风险值，得到低 温、暴雪的风险。 风险的情景表示能够给出风险的直观表达，表 4 给出了一般情况下的风险等级说明。 表 4. 风险等级说明 风险 级别 颜色 标志 风险值区间 用语 红色 0.8＜p≤1 将发生 橙色 0.7＜p≤0.8 极有可能发生 黄色 0.6＜p≤0.7 很可能发生 蓝色 0.5＜p≤0.6 可能发生 灰色 p≤0.5 不提示 风险扩散结果的情景表示如图 7 所示。 published by atlantis press copyright: the authors 128 guo shujun 图 7. 2009 年暴雪的风险情景表示 由图 7 可知，如果降雪量为 30mm、最低气温 达到-6 度，则将会对以下方面带来影响： （1）暴雪将使冬小麦发生霜冻，并导致夏粮减 产 2%；降雪将极有可能导致 24%的温室大棚垮塌， 造成设施蔬菜减产 24%，进而将使蔬菜价格大幅上 涨； （2）暴雪将使公路交通受阻率达到 70%，由于 运输不畅将导致生活日用品价格大幅上涨，蔬菜和 日用品价格的上涨将对中低收入的居民造成生活支 出增加； （3）暴雪还可能造成城市交通瘫痪，综合生活 日用品和蔬菜价格的上涨因素，将对部分居民生活 造成短时性的困难。 5. 结论 本文对气象灾害风险扩散的机理进行了研究， 得到以下结论： 1）提出了气象灾害风险扩散的概念模型。气象 灾害风险扩散是气象致灾力作用于承灾体后，由于 经济社会的系统性以及承灾体间的关联性，致使对 其他承灾体产生的波及效应。 2）提出了气象灾害风险扩散的分析计算模型。 基于气象灾害风险扩散的概念模型，提出了风险扩 散网络拓扑结构的层次分析数学算法，给出了风险 扩散关系、风险扩散计算和节点风险融合的一般计 算处理方法。 3）开发了气象灾害风险扩散的仿真模拟系统。 根据气象灾害风险扩散的分析计算模型，建立了一 个可视化的风险分析仿真工具，以人机交互的方式 实现了风险扩散网络结构的构建、风险扩散的计 算。以 2009 年初冬的华北暴雪为例进行了实际案例 风险扩散的分析，给出了具有风险预警提示的风险 情景表示图，进而验证了风险扩散分析理论与技术 方法的合理性。 参考文献 [1] 国家气候中心,2008 年中国气候公报.北京：中 国气象局，2009. national climate center, 2008 china climate report, cma, 2009. [2] 国家气候中心.2009 年中国气候公报.北京：中 国气象局，2010. national climate center, 2009 china climate report, cma, 2010. [3] 国家气候中心.2010 年中国气候公报.北京：中 国气象局，2011. national climate center, 2010 china climate report, cma, 2011. [4] roadmap for adapting to coastal risk. http://www.csc.noaa.gov/rvat/rvat.html [5] ken granger. quantify storm tide risk in chains. natural hazards, 2003,30(1): 165-185. [6] 刘希林.区域泥石流灾害风险评价研究.自然灾害 学报,2000,9(1):54-61. liu xilin. regional risk assessment on debris flow. journal of natural disasters, 2000,9(1):54-61. [7] zhang jiquan. risk assessment of drought disaster in the mainze-growing region of songliao plain, china. agriculture ecosystems & environment, 2004, 102(2):133-153. [8] icsc a science plan for integrated research on disaster risk. paris,2008. [9] 路踪，魏一鸣，范英等.灾害对国民经济影响的 定量分析模型及其应用.自然灾害学报，2002, vol.11 no.3,15-20. lu zong, wei yiming, fan ying. quantitatively analytic model for the impact of natural disaster on national economy. journal of natural disasters, 2002, vol.11 no.3,15-20. [10] 林均岐，钟江荣.地震灾害产业关联损失评估.世 界地震工程，2007 vol.23 no.2，37-41. lin junqi, zhong jiangrong. assessment of 冬小麦霜冻 0.9 大棚垮塌率 24% 0.72 粮食减产 2% 0.879 蔬菜减产率 24% 0.72 蔬菜大幅涨价 0.802 生活品大幅涨价 0.884 中低收入居民影响 0.843 下游产业受 到 2%的影响 菜民减收 24% 0.72 食品制 造 0.214 农业 0.133 纺织品 0.079 城市交通瘫痪 0.54 低温暴雪 0.9 交通受阻率 70% 0.884 运输不畅率 70% 0.884 published by atlantis press copyright: the authors 129 the meteorological disaster risk assessment earthquake induced production sections-related loss. world earthquake engineering, 2007 vol.23 no.2，37-41. [11] 石家庄市统计局.2002 年度石家庄市投入产出表. 石家庄：石家庄市统计局，2005. shi jia zhuang statistical bureau .2002 shi jia zhuang input-output table. shi jia zhuang: shi jia zhuang statistical bureau, 2005. [12] 郭树军.气象灾害风险扩散机理研究及其在气象 灾害风险评估中的应用，北京师范大学博士学 位论文，2011. guo shujun, the meteorological disaster risk diffusion study and application on meteorological disaster risk assessment, phd thesis, beijing normal university, 2011. published by atlantis press copyright: the authors 130 microsoft word research on risk assessment and regionalization of forest and grassland fires research on risk assessment and regionalization of forest and grassland fires xinghua li the centre of inner mongolia ecology and agricultural meteorology, huhhot 010051, china wenjie wu the centre of the command attemper about preventing fire in inner mongolia forest grassland, huhhot 010019, china dibo lv，cunhou zhang the centre of inner mongolia ecology and agricultural meteorology, huhhot 010051, china abstract the hazard risk and disaster degree of damage was confirmed about the forest and grassland fire in inner mongolia by the occurrence rule of fires, distribution of high temperature points, characteristic of distribution of vegetation types, and the extent of road coverage. production of risk assessment and regionalization of forest and grassland fires was made by cis, risk of fires in forest and grassland is classified as safe zones, warning zones, general risk zones, moderate risk zones and high risk zones. on the basis of conclusions of risk regionalization, forest and grassland fires will be subdivided into non-defense zones, the general defense zones, the strengthening defense zones, the key zones and the special defense zones. keywords: forest and grassland fires, risk assessment, risk regionalization, gis, fire-fighting 森林草原火灾风险评估及区划研究  李兴华 1 武文杰 2 吕迪波 1 张存厚 1 1.内蒙古生态与农业气象中心,呼和浩特 010051 2.内蒙古森林草原防火协调调度中心，呼和浩特 010020 摘要：通过对内蒙古森林草原火灾发生规律，高温点分布状况，植被类型分布特征，道路交通覆盖程度等 综合评估，确定内蒙古森林草原火灾致灾过程的危险性和受灾损害程度；利用 gis 制作森林草原火灾风险 评估和区划，将森林草原火灾风险划分为安全区、警戒区、一般危险区、中度危险区和高度危险区；根据 风险区划结论，将森林草原火灾区划分为非防御区、一般防御区、加强防御区、重点防御区和特别防御区。 关键词：森林草原火灾，风险评估，风险区划，gis，火灾防御 基金项目：内蒙古自治区科技攻关项目“内蒙古森林草原火灾预警和评估研究”（20070501）；内蒙古气象局科技创新项目“精细化森林草原火险等级 气象预报技术”（nmgqxjcx201004）资助 作者简介：李兴华（1971），男，高级工程师，主要从事气候变化、生态学、森林草原火灾、气象灾害评估方法、遥感应用等研究。 e-mail:lxh101900@163.com journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 69-77 published by atlantis press copyright: the authors 69 administrateur texte tapé à la machine administrateur texte tapé à la machine received 27 july 2011 administrateur texte tapé à la machine accepted 12 march 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine xinghua li, wenjie wu, dibao-lv，et.al 1.引言 森林草原火灾风险是指森林或草原主体发生火 灾的程度和可能性，表现为在特定时空环境中，森林 或草原受火灾威胁的可能性 [1] 。一般火灾强度越大， 频次越高，火灾所造成的破坏损失越严重，火灾的风 险也越大 [2] 。一个区域面临何种火灾风险水平，要采 取何种应对措施，是森林草原火灾风险决策需要考虑 的核心问题 [3] 。全球每年发生森林、草原火灾几十万 次；尤其近年来森林火灾发生的频率和规模更加反 常，已远远超过自然界的自我调节范畴，对人类、环 境和经济所造成的危害也超过了历年。我国每年发生 森林、草原火灾上万次，燃烧面积几十万公顷。内蒙 古历来是森林、草原火灾的高发区之一，尤其每年的 春秋两季，火灾发生更加频繁 [4] 。据统计1949～2000 年，内蒙古年均发生森林火灾92.1次，年均森林受害 面积23.6万hm 2 ；年均发生草原火灾138.4次，年均草 原受害面积1566万hm 2 [5] 。内蒙古森林草原火灾2000 年以前呈下降趋势 [6] ；2000年后森林草原火灾逐年增 加，从1999年62次增加到2004年311次，2005～2007 年每年都在160次以上。因此，加强内蒙古森林草原 火灾风险区划，确定防火区域，才能更好的做好森林 草原火灾的防御。 森林草原火灾风险评估是对致灾风险、脆弱性和 承灾体的综合评估。森林草原火灾风险的环境因子包 括森林草原主体所处空间的气候、立地条件、植被类 型以及人类活动等，森林草原火灾风险在时间上表现 为不同时间的风险不同，在空间上表现为不同水平和 垂直梯度上风险不同 [7-8] 。过去对森林草原火灾研究大 多以火行为和火险等级预报研究为主，少数对森林草 原火灾风险的研究也主要是以可燃物状态等单因素 进行风险评价 [9] 和利用历史资料进行起火概率与损失 变化计算，由于森林草原火灾受多种因素的制约，诸 如可燃物、天气条件、地形、人类活动等，因此这种 用单一方法进行的火灾风险区划，很难真实反应森林 草原火灾风险状况。 因此，本文以内蒙古火灾历史资料为基础，利用 地理信息系统，结合内蒙古森林草原区气候特点，森 林草原火灾发生的规律、严重程度、植被分布特征， 卫星监测的高温点分布状况，人类活动等开展森林草 原火灾风险评估和区划研究。对防火部门加强春秋防 火管理，有效控制火源，开辟防火道，制定防火和灭 火预案，提高防火能力具有重要意义。 2. 研究区域和数据来源及方法 2.1. 研究区域概况 2.1.1. 内蒙古地理概况 内蒙古自治区地处祖国北部边疆，东与黑龙江、 吉林、辽宁三省接壤，南与河北、山西、陕西、宁夏 相连，西与甘肃为邻，北与蒙古、俄罗斯两国交界， 东西长约 2500km，南北宽约 1700km，总土地面积为 118.3 万 km 2 ，呈狭长弧形由东北向西南方向延伸。境 内从东北向西南以大兴安岭、燕山、阴山、贺兰山、 龙首山、北山横贯全区，山地以北主要是内蒙古草原 牧业区，山地以南主要以丘陵、山地和平原为主，境 内镶嵌分布着 68.2×10 4 km 2 的天然草场，23.7×10 4 km 2 的森林，7.1×10 4 km 2 耕地，五大沙漠和五大沙地。 2.1.2. 内蒙古森林草原资源分布概况 受地貌、气候、土壤等自然因素影响，内蒙古森 林草原的水平分布自东北向西南地带性明显，分别为 森林、草甸草原、典型草原、荒漠草原、沙漠化草原 和荒漠。内蒙古森林总面积约 2366.4×10 4 hm 2 ，森林 覆盖率达 20.0%，高于全国 18.0%的水平 [10] 。森林总 蓄积量 13.6×10 8 m 3 。树木种类繁多，全区乔灌树种 达 350多种,既有寿命长、材质坚硬的优良用材林树 种，又有耐旱耐风沙运作防护林的树种，还有经济树 种和列入国家保护的珍贵树种。内蒙古自治区拥有呼 伦贝尔市东北部原始林区、锡林郭勒盟东北部和兴安 盟阿尔山原始林区、大青山原始林区、贺兰山原始林 区等六大原始林区，原始林区面积达到18.7×10 4 km 2 ， published by atlantis press copyright: the authors 70 study on risk assessment and regionalization of forest and grassland fires 居全国第二位，尤其呼伦贝尔市北部和阿尔山的樟子 松等珍贵树种保持着原始森林的特征，是区域环境变 化的敏感地带。另外呼伦贝尔市北部和阿尔山又具有 不同的气候变化特征，阿尔山林区是草典草原包围下 的林区，呼伦贝尔市北部是中国北方原始林区覆盖 区，而贺兰山原始林区被沙漠包围。人工林是内蒙古 森林中不可缺少的组成部分，不仅在条件较好的平原 区开展造林，而且深入河区、水土流失区及牧区进行 造林。防护林、用材林、经济林、薪炭林等都发展较 快。 草原是我国国土的主体和陆地生态系统的主体， 是畜牧业发展的重要物质基础和农牧民赖以生存的 基本生产资料。我国有天然草原面积3.93×10 8 hm 2 ， 约占国土总面积的41.7％，仅次于澳大利亚，居世界 第二位。内蒙古天然草场面积辽阔，是国内重要的畜 牧业生产基地，草原总面积达8666.7×10 4 hm 2 ，其中 可利用草场面积达6818.0×10 4 hm 2[10] 。内蒙古草原类 型分布多样，气候干燥、枯草期长，草原火灾频繁发 生，给畜牧业生产、人民生活及草地生态系统带来巨 大损失，严重地制约和影响了牧区经济可持续发展。 2.1.3. 内蒙古气候概况 内蒙古自治区地处中纬度地带，基座占据中温 带，东西狭长，地域辽阔，气候资源丰富多变，属于 温带大陆性气候。气候特点是气温自东北向西南逐渐 增加，在-4～8℃之间变化，气温日较差平均 12～ 16℃；全区无霜期 100～140 天之间，呼伦贝尔市东 北部只有 60～80 天。日照时间长，大部地区年日照 时数在 2700 小时以上。降水量集中于夏季，夏季降 水量占全年降水量的 60%～75%，雨热同期，比较适合 植物生长；春季干旱少雨，多大风和沙尘暴；秋季温 度急剧下降，昼夜温差大，时常有霜冻发生；冬季漫 长、严寒干燥 [11] 。 2.2. 数据资料来源 森林草原火灾资料选 1981～2007 年，由内蒙古 森林草原防火信息中心提供 [5] ；高温点资料采用 2000～2009 年极轨气象卫星（noaa 系列和 fy 系列） 遥感监测的高温点记录资料；地理信息数据采用国家 测绘局编制的 1：25 万地形图，内容包括 2005 年土 地利用和植被覆盖信息 [12] ；公路和铁路交通里程利用 2009 年最新统计数据 [10] 。 2.3. 研究方法 通过对内蒙古森林草原火灾、植被类型分布特征 和气象条件的统计分析，研究森林草原火灾发生的特 点和规律；以相关分析和聚类分析为基础，分析影响 森林草原火灾发生的主要因子，确定森林草原火灾致 灾过程的危险性和受灾损害程度，根据森林草原火灾 发生不同时间的严重程度和火灾成因，制作森林草原 火灾风险时段评估和区划。采用因子加权叠置法 [13] ， 依据森林草原火灾空间分布特征，高温点的分布状 况，植被类型分布特征，铁路和公路交通覆盖程度确 定危险等级，根据危险等级划分防治区域。 因子加权叠置法主要是充分利用 gis（地理信息系 统）能够对地理数据进行采集、管理、查询、计算、 分析与可视化表现的功能，在统一地理坐标基础上对 各因子指数绘图，通过指数计算和图层叠加 [11] ，绘制 内蒙古森林草原火灾风险区划图。具体方法见技术流 程图 1。 published by atlantis press copyright: the authors 71 xinghua li, wenjie wu, dibao-lv，et.al 3. 森林草原火灾风险评估和区划内容 3.1. 森林草原火灾风险时段评估和区划 内蒙古森林草原火灾的时间格局与季、月和日的 气象条件息息相关。 研究表明，降水量与森林草原火灾发生次数的相 关系数为-0.19，当日降水量大于 10mm 时，24 小时内 无火灾发生，48 小时内发生火灾数占样本数的 2%； 日降水量 5～10mm 时，当日无火灾发生，24 小时发生 火灾数占样本数的 2%，48 小时发生火灾数占样本数 的 6.6%；日降水量小于 5mm 时，火灾发生次数逐渐 增多，其中 24 小时内发生火灾数占样本数的 23%。另 外，冬季积雪也与火灾密切相关，当积雪深度大于 3cm 并可以座住时，基本无火灾发生 [14] 。 月平均气温在一定区间内与森林草原火灾呈正 相关，相关系数为 0.27；当最高气温低于-11℃时， 由于积雪和人类的活动减少，一般不会发生火灾；当 日最高气温大于 20℃时，由于草木生长旺盛，降水季 节性增多，也不易发生火灾；日平均气温在-9.5～ 17.7℃之间时，由于此时为春种秋收的季节，草木含 水量小，植被干枯，是火灾发生的高峰期 [14] 。 相对湿度与森林草原火灾的相关系数为-0.88 ** ， 即湿度越大火灾发生次数越少；湿度越小火灾发生的 次数越多。当空气中的相对湿度大于 62%时，基本无 火灾发生；相对湿度在 52～62%之间时，火灾发生的 次数也较少；相对湿度在 35～52%之间时火灾发生的 次数最多 [14] 。 风速与森林草原火灾的相关系数为 0.86 ** 。风可 以加速地被物的水分蒸发，加快干化；风对火灾具有 助燃作用，能把燃着物及余烬带到新的可燃物上，使 火灾蔓延扩散。统计表明，在火灾发生当日，最大风 速大于 7m/s 的机率为 78.3%，风大使火蔓延加速，使 火场的氧气得到补充，加剧燃烧 [14] 。 内蒙古近 27 年总计发生森林草原火灾 5598 次， 平均每年 207.3 次。内蒙古森林草原火灾次数冬季 （12～2 月）最少，占全年火灾总数的 0.98%；春季 （3～5 月）火灾次数最多，占全年火灾总数的 66.1%； 夏季（6～8 月）火灾次数较少；秋季（9～11 月）火 灾次数次多，占全年火灾总数的 19.1%。依据火灾发 生次数将每年 4 月和 5 月划分为极度危险期，6 月和 10 月划分为高度危险期，3 月划分为中度危险期，9 月和 11 月划分为一般危险期，7、8 和 2 月划分为林 火警戒期，1 月和 12 月为安全期。从气候条件来看， 春季降水少，气温回升快，大风日数多，地被物干燥， 极容易发生火灾；秋季降水减少，地被物干枯，气温 下降快，昼夜温差大，人们早晚野外用火增加，容易 引发火灾；冬季地被物被积雪覆盖，可燃物覆盖度和 连续度降低，人为火源减少，不易发生火灾 [2] ；夏季 随着雨季的来临，植被返青，空气湿度增大，可燃物 燃烧性能降低，不易发生火灾，但 2000 年以后夏旱 频繁，雷击火灾次数明显增加。 图 2 内蒙古森林草原火灾的日变化图 由图 2 可知，每日 8～19 时是火灾的危险期，其 中 10～15 时是火灾的极度危险期；8～9 时和 16～17 时火灾的高度危险期；18～19 时火灾的中度危险期； 6～7 时和 20～21 时是火灾的警戒期；22 时～次日 5 时是火灾的相对安全期区。火灾的日变化与人们的生 产生活用火和气候条件息息相关，12 时前后是午饭时 间，也是每日温度最高、相对湿度最小、风速较大、 蒸发力较强的时期，地表可燃物很快变干，在加上局 地对流增强，着火容易蔓延 [15] ；22 时～次日 5 时是人 0 200 400 600 800 1 2 3 4 5 6 7 8 9101112131415161718192021222324 小时 火 灾 次 数 published by atlantis press copyright: the authors 72 study on risk assessment and regionalization of forest and grassland fires 类活动最少的时期，也是气候最湿润的时期，因此火 灾次数最少。 3.2. 森林草原火灾空间风险评估和区划 森林草原火灾的发生必须具备可燃物、火源和可 燃烧气象条件才能发生火灾 [16] ，因此森林草原火灾空 间风险评估和区划必需充分考虑这些因子。火灾风险 评估是对致灾风险、脆弱性和承灾体的综合评估 [17] 。 致灾风险评价主要是分析火灾发生的起因，根据孕灾 环境建立自然、人为和境外入侵的评价因素，结合历 年火灾数据统计起火频率，即火灾致灾风险。脆弱性 评价对承灾体以及周围孕灾环境进行综合评价，利用 历年火灾数据分布规律， 结合孕灾环境评价火灾发 生区域火灾蔓延的可能性。承灾体分为自然承灾体和 社会承灾体 [1] ，自然承灾体评价主要是对可能发生火 灾区域的植被生物量进行评估，社会承灾体主要是对 人类社会和生物界造成的影响评估。 3.2.1. 森林草原火灾可燃物类型分布 内蒙古森林草原区地面植被的分布决定了火灾 次数空间分布呈东多西少走势 [2] 。由内蒙古植被分布 状况可知，呼伦贝尔市北部原始林区，人为火源少， 主要以雷击火灾为主；呼伦贝尔市东部为原始林区、 高覆盖度草甸草原区和农林交错区，植被连续度好， 人口较多，起火原因复杂，是森林草原火灾高发区； 呼伦贝尔市南部和兴安盟的植被类型为农林交错区， 人员活动频繁，火源增加，发生火灾次数最多；呼伦 贝尔市西部和锡盟东北部的植被类型为草甸草原，由 于本区生产以畜牧为主，人员活动相对较少，火灾发 生的次数略少；通辽市北部、赤峰市北部、锡盟中部 和鄂尔多斯市东部为典型草原分布区，由于牧草高 度、盖度和连续度受降水的影响较大，火灾次数明显 下降；锡盟西部、乌兰察布市北部、包头市、鄂尔多 斯市中西部的植被类型为荒漠化草原，植被长势较 差，不易发生火灾；阿拉善盟大部为荒漠，植被稀疏， 人烟稀少，基本无火灾发生。 3.2.2. 森林草原火灾空间分布 内蒙古近 27 年总计发生森林草原火灾 5598 次， 平均每年 207.3 次。由图 3 可知，内蒙古森林草原火 灾空间分布呈东多西少。呼伦贝尔市是火灾的高发 区，其中鄂伦春旗年平均火灾次数最多为 33.7 次， 其次牙克石市和莫力达瓦旗年平均火灾次数为 13～ 14 次，都属高度危险区；呼伦贝尔市西部和南部、兴 安盟北部和锡林郭勒盟东北部年平均火灾次数为 5～ 9 次，属中度危险区；满洲里市、兴安盟大部、通辽 市北部、赤峰市北部和锡林郭勒盟中部年平均火灾次 数为 2～5 次，属一般危险区；通辽市南部、赤峰市 南部、锡林郭勒盟西南部、鄂尔多斯市东部年平均火 灾次数为 1～2 次，属警戒区；其他地区的年平均火 灾次数在 1 次以下，植被稀疏，属安全区。 图 3 内蒙古森林草原火灾年平均次数分布图 3.2.3. 森林草原高温点空间分布 通过对内蒙古 2000～2009 年卫星遥感监测的高 温火点资料统计，内蒙古近 10 年高温火点平均每年 595.3 次。由图 4 可知，呼伦贝尔市是高温火点密集 区，年平均 15 次以上，属高度危险区，其中鄂伦春 旗年平均高温火点最多为 157.4 次，其次额尔古纳市、 莫力达瓦旗、牙克石、阿荣旗、扎兰屯、陈旗年高温 火点次数为 30～71.3 次；兴安盟北部和锡林郭勒盟 东北部的高温火点次数为 10～15 次，也属高度危险 区；兴安盟南部、锡林郭勒盟中部年高温火点次数为 published by atlantis press copyright: the authors 73 xinghua li, wenjie wu, dibao-lv，et.al 5～10 次，属中度危险区；通辽市北部、锡林郭勒盟 西南部年高温火点次数为 2～5 次，属一般危险区； 通辽市大部、赤峰市北部、锡林郭勒盟西部年高温火 点次数为 1～2 次，属警戒区；其他地区年高温火点 次数都在 1 次以下，植被稀疏，属安全区。另外，通 过叠加道路交通图发现，高温点在主要的交通干线周 围也是密集区。 图 4 内蒙古森林草原高温点年平均次数分布图 3.2.4.森林草原火灾成因分析 由图 5 可知，发生火灾后不能查明原因的占火灾 总数的 33%。人为引发火灾次数占火灾总数的 61%， 其中吸烟引起火灾占火灾总数的 19%；机动车失火引 起火灾占火灾总数的 12%；上坟烧纸引起火灾占火灾 总数的 7%；烧茬地引起火灾占火灾总数的 6%；人为 放火引起火灾占火灾总数的 4%；其他原因引起火灾 占火灾总数的 13%。雷击引起火灾占火灾总数的 6%。 吸烟和机车失火是人为引发森林草原火灾的最 主要原因，尤其每年的 4、5 和 10 月份是本区的春耕 和秋收的主要时期，野外人员活动和车辆使用频繁， 极易引发火灾。另外上坟烧纸虽然只占火灾总数的 7%，但是其引发的火灾集中在每年清明前后，这类火 灾经过加强管理，最近几年已经基本杜绝。雷击火是 由于天气干燥，干雷暴引发的火灾；雷击火灾由于受 天气系统的影响，起火时上空有云覆盖，卫星遥感发 现时，火灾已经扩散开来，因此雷击火灾具有随机性 大，人为控制难的特点；预防雷击火的基本原则为早 发现早扑救，严防火灾蔓延 [18] ；雷击火发生的主要区 域为呼伦贝尔市东北部和锡盟东北部，主要发生时期 为每年的 5 月下旬～6 月底、9 月下旬～10 月上旬也 有少量的雷击火发生；其中 2000～2005 年共发生雷 击火 187 次，年均 31 次，比 1981～2000 年均值(9 次) 增多了 23 次，其中 2004 年最多达到 55 次。 图 5 内蒙古森林草原火灾成因分布图 3.2.5. 森林草原火灾空间风险评估和区划 3.2.5.1 森林草原火灾空间风险评估模型 根据森林草原火灾发生的严重程度，高温点的分 布状况，植被类型分布特征等区划结论通过综合评 估，制作内蒙古森林草原火灾风险评估和区划。另外， 由于机车失火、拖拉机喷火和吸烟都是火灾发生的重 要原因，而这些火灾的发生又与道路交通密切相关， 因此将道路交通覆盖程度也作为一个评判指标，并按 道路交通覆盖指数（道路交通覆盖指数=公路（铁路） 长度/总面积）确定危险等级。而气候干燥度因子在 风险区划中只做月季危险度指标，在空间上湿度越大 的地区植被覆盖好，火灾发生的次数多，干燥度越大 的地区，植被稀疏，不易发生火灾，因此气候因子不 作为空间区划的指标。通过表 1 中多指标因子的加权 综合评价，确定森林草原火灾风险评价指标，并根据 指标大小划分危险等级，绘制内蒙古森林草原火灾风 险区划图。 吸烟 19% 上坟烧纸 7% 烧茬地 6%人为放火 4% 雷击火 6% 不明原因 33% 其他原因 13% 机车失火 12% 吸烟 机车失火 上坟烧纸 烧茬地 人为放火 雷击火 不明原因 其他原因 published by atlantis press copyright: the authors 74 study on risk assessment and regionalization of forest and grassland fires 火灾风险评估指数=ｆ（火灾孕灾环境，致灾因 子，承灾体） [1] 其中火灾孕灾环境包括森林草原火灾的历史次 数，卫星遥感监测高温火点次数，道路分布；致灾因 子包括可燃物种类和火源（人为火源、雷击火灾、自 然火灾、机车失火、境外烧入等） [19] ；火灾承灾体主 要是森林和草地。由于火灾指数、高温点指数、公路 交通指数和铁路交通指数与植被之间是交叉关系，植 被的好坏决定了火灾发生和蔓延的程度，因此将植被 类型指数与其他指数之和做乘法处理，具体火灾风险 评估区划模型如下： hzfx=（a+b+d+e）×c 其中 a 是历史火灾次数指数；b 是高温点次数指 数；c 是植被类型指数；d 是公路交通指数；e 是铁路 交通指数。根据各指标的重要程度，确定火灾次数指 数权重为 35%；高温点次数的权重为 30%；植被类型 指数权重为 18%；公路交通指数权重为 9%；铁路交通 指数权重为 8%。火灾风险评价指标见表 1。 表 1 火灾风险评价指数和指标一览表 风险区划因子及权重 指数 指 标 火灾次数 p >15 11～15 5～10 2～5 1～2 ≤1 a 指数 12 10 8 6 4 2 高温点次数 q >15 11～15 5～10 2～5 1～2 ≤1 b 指数 10 8 6 4 2 1 植被类型 z 森林 草甸草原 典型草原 农田 沙地 荒漠化草原 荒漠 c 指数 6 5 4 3 2 1 0 公路交通指数 m >0.5 0.41～0.5 0.31～0.4 0.21～0.3 0.11～0.2 ≤0.1 d 指数 3 2.5 2 1.5 1 0.5 铁路交通指数 n >0.2 0.16～0.2 0.12～0.16 0.08～0.12 0.04～0.08 ≤0.04 e 指数 3 2.5 2 1.5 1 0.5 3.2.5.2 森林草原火灾空间风险评估和区划结论 根据森林草原火灾发生的特点和严重程度，将森 林草原火灾风险划分为高度危险区、中度危险区、一 般危险区、安全区和警戒区 [20] ，各划分指标见表 2。 高度危险区主要分布在呼伦贝尔市中部和东部；中度 危险区主要分布在呼伦贝尔市中部偏西和南部、兴安 盟北部和锡林郭勒盟东北部；一般危险区主要分布在 呼伦贝尔市西部、兴安盟大部、通辽市北部、赤峰市 北部和锡林郭勒盟中东部；警戒区主要分布在通辽 市、赤峰市、锡林郭勒盟西部、乌兰察布市南部、呼 和浩特市、包头市南部、鄂尔多斯市北部和巴彦淖尔 市南部；安全区主要分布在中西部植被稀疏的荒漠化 草原区或荒漠区。根据风险区划结论，将森林草原火 灾区划分为特别防御区、重点防御区、加强防御区、 一般防御区和非防御区，具体见图 6。 表 2 内蒙古森林草原火灾风险指数划分及预防措施表 hzfx 指数 危险程度 预防要求 ≤10 安全区 无需预防 11～30 警戒区 一般预防 31～65 一般危险区 加强预防 66～100 中度危险 重点预防 100< 高度危险 特别防御 published by atlantis press copyright: the authors 75 xinghua li, wenjie wu, dibao-lv，et.al 图 6 内蒙古森林草原火灾风险和防御区划图 3.2.6. 森林草原火灾风险区划应用 开展森林草原火灾时空风险区划的目的是在火 险等级预报、火灾预防和火灾扑救中应用。根据森林 草原火灾月季区划可以将春秋季做内蒙古森林草原 火灾预防的重点时段，并将每年春季 4 月、5 月和秋 季 10 月作为防火关键期；根据火灾发生的日变化规 律和风险评估结论，得出中午前后由于温度高、风速 大、地被物干燥，有利于火灾蔓延，不利于扑救，因 此有经验的扑火指挥员，常常利用早晚和午夜对打火 有利的气象条件，进行包围战，把火场围起来，一举 歼灭 [15] 。 火灾空间风险区划结论，是做好火险等级预报的 基础，由于内蒙古东西跨度大，气候和植被分布的差 异性强，在火险等级预报中单一的预报模型很难满足 预报需求。因此根据火灾空间风险区划将预报区域划 分为五个区，在不同预报区域，根据各自的植被和气 候特点，建立不同的森林草原可燃物可燃烧等级模型 和火险气象等级预报模型，并利用 gis 软件绘制火险 等级预报图。 4 结论与讨论 内蒙古森林草原火灾次数在20世纪末曾经每年 降到100次以下；进入21世纪，森林草原火灾次数呈 逐年递增趋势，已经从2000年的150次上升到2005年 的346次，2006年也达到200次以上。随着国家天然林 保护工程和生态建设工程的实施，内蒙古大部地区的 植被逐年得到恢复，在丰水年，植被长势更好，而许 多地区又实施围栏封育和禁牧等政策，过高的植被得 不到及时合理的利用，必然使这些地区的火灾隐患和 火灾次数增多；另外随着气候变暖 [21-22] ，内蒙古东北 部林区的降水呈逐年减少趋势，春夏秋三季的阶段性 干旱时有发生，夏季干雷暴引发的森林火灾次数也呈 增加趋势。内蒙古森林草原火灾的分布格局也在发生 重大变化，东部森林火灾增加，草地火灾减少；而中 西部地区随着植被恢复，火灾次数呈逐年增加趋势， 历史上没有火灾发生的区域，近几年也有火灾发生。 这使火灾的防御范围不断扩大，防御时间不断延长， 由历史上东部区防火，变成全区性防火，由春秋季防 火，变成全年性预防。因此，本风险评估和区划适应 了当前防火工作的需要，对防火部门有效控制火源， 科学规划开辟防火道，制定防火和灭火预案，提高防 扑火能力具有重要的参考价值。 参考文献 1. fan w.c., sun j.h., lu s.x. and et al., fire risk assessment methodology, beijing: science press, 2004. 范维登,孙金华,陆守香等,火灾风险评估方法学[m],北 京：科学出版社,2004. 2. zhang j.q., liu x.p. and tong z.j., the study of grassland fire disaster risk assessment and regionalization: a case study in the western jilin province, geographical research, 2007, 26(4), 755-762. 张继权,刘兴朋,佟志军,草原火灾风险评价与分区—以 吉林省西部草原为例[j],地理研究,2007,26(4),755-762. 3. chen y., study on the establishment of district fire risk criteria, fire science and technology, 2006, 25(5), 646-648. 陈也, 区域火灾风险准则的确立研究[j],防火科学与 设计,2006,25(5),646-648. 4. li x.h., guo x.x. and li s.s., the occurrence rule and character of forest and grassland fire in inner mongolia, meteorological monthly, meteorology journal of inner mongolia, 2003,28-30. 李兴华,郭西峡,李树森,内蒙古自治区森林草原火灾发 生的规律及特征,内蒙古气象,2003,28-30. 5. the centre of the command attempter about preventing fire in inner mongolia forest grassland (1981-2007). 内蒙古森林草原防火协调调度中心(1981-2007 年). 6. li x.h., guo x.x., li y.p and et al., the evolution of forest and grassland fire and the relationship of it to meteorological condition in inner mongolia, meteorological monthly, 2003, 6, 15-19. 李兴华,郭西峡,李云鹏等,内蒙古森林草原火险发生的 演变规律及与气象条件关系[j],气象,2003,6,15-19. published by atlantis press copyright: the authors 76 study on risk assessment and regionalization of forest and grassland fires 7. huang c.f, basic principles of risk analysis of natural disasters, journal of natural disasters , 1999, 8 (2) , 21-30. 黄崇福,自然灾害风险分析的基本原理[j],自然灾害学 报,1999,8(2),21-30. 8. huang h.f, zhang j.x., chen z.f. and et al., toward a new kind of natural disaster risk zoning map, journal of natural disasters, 2004,12(3) , 9-15. 黄崇福,张俊香,陈志芬等,自然灾害风险区划图的一个 潜在发展方向[j],自然灾害学报,2004,12(3),9-15. 9. liu x.p., zhang j.q., zhou d.w. and et al., study on grassland fire risk dynamic distribution characteristic and management policy,chinese journal of grassland, 2006, 18(6),77-82. 刘兴朋,张继权,周道伟等,中国草原火灾风险动态分布 特征及管理对策研究[j],中国草地学报,2006,18(6)： 77-82. 10. national bureau of inner mongolia survey team, inner mongolia economic and social survey yearbook-2010, beijing:china statistics press,2010. 国家统计局内蒙古调查队,内蒙古经济社会调查年鉴 ——2010[m],北京：中国统计出版社,2010. 11. wu h.b. and et al., analysis on main meteorological disasters in the inner mongolia autonomous region, bei jing: china meteorological press, 1990. 吴鸿宾等,内蒙古自治区主要气象灾害分析[m],北京： 气象出版社,1990. 12. tong z.j., zhang j.q.,.liao x.y. and et al., a gis—based research on grassland fire disaster risk management decision support system, journal of northeast normal university(natural science edition), 2008, 40(2),115-120. 佟志军,张继权,廖晓玉,刘兴朋,基于 gis 的草原火灾风 险管理辅助决策支持系统的研究.东北师大学报(自然 科学版),2008,40(2),115-120. 13. yin h.w, kong f.h. and li x.z., gis-based forest fire risk zone mapping in daxing’an mountains, chinese journal of applied ecology, 2005, 16(5), 833-837. 尹海伟,孔繁花,李秀珍,基于 gis 的大兴安岭森林火险 区划,应用生态学报,2005,16(5),833-837. 14. li x.h., li. y.p. and yang l.p., inner mongolia forest grassland fire danger rating short term prediction method and system study. meteorological monthly, 2003, 6, 79-82. 李兴华,李云鹏,杨丽萍,内蒙古森林草原火险等级短期 预报方法及系统开发[j],气象,2003,6,79-82. 15. zhou g.s., lu q. and et al., weather and forest grassland fire. bei jing: china meteorological press. 2009. 周广胜,卢琦等,气象与森林草原火灾[m],北京：气象出 版社,2009. 16. zhao f.j., wang y.m., shu l.f. and et al., progress in studies on influences of climate change on forest fire regime, advances in climate change research, 2009, 5(1),50-55. 赵凤君,王玉明,舒立福等,气候变化对林火动态的影响 研究进展[j],气候变化研究进展,2009,5(1),50-55. 17. li s. k., huo z.g., wang s.y. and et al., risk evaluation system and models of agro meteorological disasters, journal of natural disasters, 2004, 13(1), 77-87. 李世奎,霍治国,王素艳等,农业气象灾害风险评估体系 及模型研究[j],自然灾害学报,2004,13(1),77-87. 18. li x.h., yang l.p. and lv d.b, the occurrence reason of forest and grassland fire and fire risk forecast in summer of inner mongolia, meteorology journal of inner mongolia, 2004, 13(1), 77-87. 李兴华,杨丽萍,吕迪波,内蒙古夏季森林火灾发生原因 及火险等级预报[j],内蒙古气象,2004,2,27-29. 19. zhang s.y., zhu c.h. and chen z.h., research on forest fire meteorological environmental elements and large forest fires, journal of natural disasters, 2000, 9(2),111-117. 张尚引,祝昌汉,陈正洪,森林火灾气象环境要素和重大 林火研究[j],自然灾害学报,2000,9(2),111-117. 20. liu x.p., zhang j.q. and fan j.b., historical data-based risk assessment of fire in grassland of northern china, journal of natural disasters, 2007, 16(1),61-65. 刘兴朋,张继权,范久波,基于历史资料的中国北方草原 火灾风险评价,自然灾害学报,2007,16(1),61-65. 21. china’s national assessment report on climate change of editorial board, china’s national assessment report on climate change,beijing: science press, 2007. 气候变化国家评估报告编写委员会,气候变化国家评 估报告[m],北京：科学出版社,2007,7-17。 22. ding y.h., ren g.y., shi g.y. and et al., china’s national assessment report on climate change(i): climate change in china and the future trend, advances in climate change research, 2006, 2(1), 3-8. 丁一汇,任国玉,石广玉,等,气候变化国家评估报告(i): 中国气候变化的历史和未来趋势[j],气候变化研究进 展,2006,2(1),-8。 published by atlantis press copyright: the authors 77 microsoft word how crises model the modern world how crises model the modern world patrick lagadeci senior research scientist at the ecole polytechnique palaiseau, france benjamin topperii theoretical physicist (upmc/ens ulm, france), junior science and technology advisor paris, france abstract crises are our new reality. “black swans” are increasingly becoming the norm; our systems, environments, contexts are structurally prone to crises. doing more of the same will not be the appropriate way to deal with modern crises: a paradigm shift is needed, based on a more accurate understanding of the dynamics of complex systems. this paper is an invitation to change the theoretical vision of crisis and crisis management, and the education and training of all actors involved. keywords: crises, complex systems, black swan, fractal, rapid reflection force 1. global crises and the rise of a new world natural disasters. mega-storms like katrina, “assembly lines” of hurricanes striking vital assets and networks of our nations and continents; earthquakes threatening major dams’ stability; tsunamis triggering large scale technological disasters and threatening millions, as the world population shifts towards the seashores; forest fires leading to cross-continental power failures… the very concept of “natural” disaster has to be revised in a time when nature can trigger technological disasters and technology can spark off natural upheavals. simple categorizations are not valid anymore. industrial disasters. on march 11, a magnitude 9.0 earthquake, one of the most powerful earthquakes the world has ever known since modern recordkeeping began in 1900, combined with a destructive tsunami (with waves over 20 meters), hit japan and triggered a series of nuclear disasters (level 7 on the ines scale). a devastated region, a vast area under nuclear radiation threat, the third greatest economic power in the world was left with shortages in energy and food supplies, and triggered unknown ripple effects for the months to come on a worldwide scale. the japanese proved to be very well prepared for the well-known threat – an earthquake – but unprepared for such an outside-the-box challenge. in the nuclear area in particular, as confessed by the international atomic energy agency chief: “the current international emergency response framework needs to be reassessed. it was designed largely in the wake of the chernobyl disaster in 1986, before the information revolution. it reflects the realities of the 1980s, not of the 21st century” (ref. 1). the global "fall out" appears to be borderless: the very perception of nuclear energy, our faith in technological advancement, and public trust in official governance, are at stake. financial crises. the lehman brothers collapse in september 2008 came as a shock, but on its heels came something more: systemic financial-economic upheaval, with the banking system calling for desperate rescue operations. and now, we are facing the intricacies of financial, economic, and social turbulences behind the incredible and unreadable perspectives of state failures. greece, italy, portugal, spain, … the challenge is defying the best visions, techniques, usual habits – even the best experts on the 1929 crisis, which is not the ideal reference. social unrest. the stage was set for decades to come: nothing new would emerge in the arab countries, under strict control. yet what seemed journal of risk analysis and crisis response, vol. 2, no. 1 (may 2012), 21-33 published by atlantis press copyright: the authors 21 administrateur texte tapé à la machine received 4 august 2011 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine accepted 12 march 2012 administrateur texte tapé à la machine p. lagadec and b. topper impossible the previous day did happen, and the map became globally unpredictable. it all started with just one man, al bouazizi, a poor unemployed 26-year old tunisian, who set himself on fire in front of a government building on december 17 of last year. a month later, ben ali fled to saudi arabia. “the march of folly” (ref. 2) could spread across the board. but many different maps had to be drawn: egypt was not tunisia; libya was not egypt; syria was not libya. the crucial challenge is not the lack of answers, but the difficulty of asking and framing the proper questions. the problem is no longer the “event with ripple effects” but the multifaceted challenging situation mixing hyper-local and hyper-global dynamics. emergency response and recovery. the conventional vision was to rush in and help countries in difficult situations after some disaster occurred. this vision and situation is increasingly shattered. crises appear to be blossoming everywhere: at any one time, the world has to face about forty crises of different natures, with, for some, a risk of regional spillover able to destabilize international security. crises change their scale due to demographic trends, and new types of humanitarian intervention could affect billions by 2030. post-event intervention has now to deal within a proliferation of “failing states”, with the development of extremely violent situations and the development of mafias at the local, national and international levels (ref. 3). whatever the frontline, the same kind of difficulty comes to the fore: conventional visions of crises and crisis management are thrown into an uncharted wilderness. we are confronted with a global volatile world, with novel textures and foundations – where the tiniest change can result in a global and unmapped dynamic. the challenge is to prepare the conditions to deal with this new landscape. our conventional vision of risks and crises has to evolve accordingly. this contribution modestly aims at proposing some preliminary inputs to facilitate such a huge program. a new vision and practice of crisis management is required. 2. why do we continually appear to be a disaster behind? the most recent crises are illustrations of a global dynamic. the words written by the house of representatives after hurricane katrina, come to mind: “why do we continually appear to be a disaster behind?” (ref. 4). 2.1. crisis management in crisis hegel’s remark in lectures on the philosophy of world history that “history is not the soil in which happiness grows” teaches us that crises are a constant of human societies. the names and dates of great crises are remembered and used to recall entire epochs, and they have influenced changes of societies, cultures and civilizations. however, in the past decades, crises have been multiplying exponentially, and man’s power over them is looking increasingly fragile. urlich beck has warned that modern man is presently living in a “risk society,” obsessively preoccupied with his security and very sensitive to the slightest sign of its imperilment. modern crises management practice (refs 5-8) and theory (refs. 9-10), even though it has been more and more precise in the characterization of complex systems with an ever larger set of parameters to act upon, has never taken the decisive step needed to deal with the world as it really is (refs. 11-13). there exists an extensive literature on the topic of crisis resulting in many definitions that overlap and sometimes diverge. many have underlined and criticized such a hazy notion. however, they often miss the point. by essence, crisis is a very complex and elusive phenomenon. the very notion of crisis is structurally resistant to clear-cut definition and capture. hence, crisis management theory cannot be so easily ensnared in usual frameworks. and crisis management cannot be a series of fixed, easy to teach and apply “best practices”. these last decades, crisis management has involved scholars from various disciplines (sociology, public administration, political science, international relations, organizational psychology, epidemiologists …). each discipline has tried to produce definitions, frameworks, and sometimes practical tools. this sophistication has not been a synonym for efficiency – especially when used to understand and deal with recent crises since the turn of the century. any actions performed within the current crisis management paradigm forged in the published by atlantis press copyright: the authors 22 how crises model the modern world 1980s have had limited efficiency, and sometimes counter-productive consequences (refs. 14-15). undoubtedly, the last few decades have witnessed an extraordinary development in the sciences and techniques of risk control and crisis management. but, cases are piling up to show that something is wrong with the current approach (table 1). the core of the challenge appears to be in the bedrock of crisis theory and the very culture of crisis management. so far, the motto has been: let’s define families of risks, let’s plan in advance to prepare the best answers, to detect weak signals, and apply those best answers. today hyper-complex crises demand something else, and probably exactly at the opposite: the capacity to ask the tough questions, the preparation to navigate unmapped situations. 2.2. when black swans become the norm big departures from the mean, usual, or common state, or “black swans” (ref. 20), used to be so rare that their effect could be considered negligible. all practices, regulations, policies, were set to address the regular occurrences only, as they were the one people were facing on a daily basis. until now, we have been reasoning by induction, dealing only with “white swans”, the common accidents. however, in a world where complexity and crises have become the new reality, such rare events have become so dramatic and so frequent that they cannot be excluded as simply marginal. “black swans” are increasingly becoming the norm, forcing us to search for new perspectives and a novel collective attitude towards risks and major crises. the intrinsic quality of the dynamics in question is increasingly eluding our crisis management rationale and governance. observing an event once does not imply that it will occur again in the future. in other words, there is no way to know that somewhere out there a black swan is not hiding, disproving the rule and nullifying our "knowledge" of swans. it points out the problem of inductive reasoning: we cannot really learn from our past experiences, and cannot always apply what has been learned in past crises to new crises. this statement remains true regardless of the number of observations one adds to the pile. david hume, quoted by n.n. taleb, said that the observation of even a million white swans does not justify the statement "all swans are white." planning is becoming much more difficult. although it can help, it cannot be the driver. the importance of knowledge, specifically knowledge of previous crises, is a question that can be raised, as new crises become more and more difficult to apprehend. humans have a tendency to behave -almost without exceptionas though they believe that experience teaches them lessons, in a very cause-consequence, past-present, “linear” line of thought. this can be easily understood, as it has been very effective until recent times. one must understand the limits of induction, which shapes a world where past experience is a more fundamental knowledge than a deep, theoretical understanding of the framework we are working in. we still seem, at every turn, to lack this critical self-awareness, this realization that more of the same is not the solution. this basic challenge becomes strategic when the table 1. recalling some recent events. storm of 1999 over europe. 9/11 katrina, 2005 lehman brothers, 2009 “nobody had anticipated, after the first alarming signals, the runaway scenario that happened.” (ref. 16) “on the morning of 9/11, the existing protocol was unsuited in every respect for what was about to happen” 9/11 commission (ref. 17). “it was not a hurricane, but nobody understood that. it was a weapon of mass destruction without criminal dimension.” admiral thad allen (ref. 18) “no one knows what to do. we are in a new territory here. this is a new game. they don’t know what to do.” senator majority leader harry reid (ref. 19). published by atlantis press copyright: the authors 23 p. lagadec and b. topper issue is not only to be able to detect and deal with one black swan, but when black swans appear to be the “norm” in a very volatile world. then, we are in a very difficult position, urging a more profound understanding of what is happening and more creativity. 2.3. “business as usual” and the collapse of the roman empire: why we need to act our current situation is very similar to the apogee of the roman empire. the situation for someone living today is better than it has ever been in history. we have built connections between parts of the world that had never interacted before, creating a global interdependent economy. like someone living at the apogee of the roman empire, very few clear signs indicate that it has reached the top of the hill. there is a refusal to see them, but the fall is near, unless we raise awareness about this fact on a global scale. during the 3rd century a.d., the roman empire had to face a serious military crisis: invasions of foreign peoples and internal civil wars. the crisis was solved by diocletian by doubling the size of the army, increasing taxes and enlarging bureaucracy; overall it was a considerable increase in complexity. the roman empire could not afford such a large army and, eventually, it destroyed itself in the attempt to maintain it. our society is reacting to the surge of crises much in the same way (ref. 21). despite all the talk of “saving” or “conserving” resources, or “regulating” some new practices, or “controlling” new technologies, it is clear that our society is not doing anything like that. we strive, certainly, towards more efficiency, regulation, control or security, but what is done in some areas or in some branch of the economy is not done in some other areas, in some other branches, due to ever more complex and interconnected societies. and that feedback loop is the loophole of complex societies. complexity is all the economic, social, bureaucratic, and military structures that societies can create and manage, but only up to a certain point. crises are piling up, and we are less and less able to face them. we argue that, as the apogee of the roman empire, we are reaching that point of maximum disorder. and unless we do something about it, the world will slowly enter an era of terra incognita, facing successive global crises which could mean the decline of the industrial society as we know it. however, modern societies have developed a level of knowledge, a science of nature, which can be used to adapt to this growing volatility. in the following part, we will make a careful detour through hard science to show how the appropriate paradigm shift can allow us to change the way we understand the emerging properties of dynamic interactions in a society and why current techniques are failing. 3. deep into science: a call for action in physics, dealing with complex systems usually starts with trying to linearize the equations involved i.e. getting rid of the complexity, exactly in the same way as black swans are diminished when events are modeled with a gaussian curve. this approach leads to quantitative results, which allow physicists to grasp an idea of the processes that are taking place near equilibrium. we argue that the way we are currently dealing with a crisis is similar to that over-simplified approach, but that recent major crises (table 1) indicate that it has become ineffective (ref. 22). 3.1. a failing paradigm crisis management is a comprehensive and cohesive set of processes which consist of four separate but related basic components: anticipation and preparation, rapid response, follow-through, and post-event evaluation. in an ideal “white swan” world, seamless anticipation and preparation would eliminate all major crises. in reality, we are seeing time after time, event after event (table 1), that "the best laid plans" become almost instantaneously irrelevant, if not counter-productive, in modern crises. we will first start by trying to characterize the analogy we are making between linear systems and current crisis theory and discuss the reasons why it is failing. humanity started with small tribes, which grew stronger and larger; struggles over territory ensued. at the same time, agriculture was developed, which published by atlantis press copyright: the authors 24 how crises model the modern world opened the possibility of trade and commercial ties. social, cultural, and educational revolutions made their way, further strengthening the ties among individuals, but also increasing our dependence on one another. the understanding that we could draw greater good from combining efforts and joining forces lead to the industrial era, and the information revolution. yet they also have raised connections among humans up to a whole new and sophisticated level (figure 1). thus, after tens of thousands of years, we are now realizing that, in the last decade or so, the world has become a global village, in which we are all connected socially, politically, and economically. the attributes of the current paradigm of crisis management however still rely on the following assumptions: (i) events happen with a long characteristic time with respect to “regulation” time, i.e. once the risk has been identified, one has enough time to act before something irreversible happens ; (ii) well defined and spatially limited interactions between the various actors i.e. once the risk has been identified, one has enough control over the actors to act before something irreversible happens. these two properties have, until the last decade or so, allowed effective control and limited crisis events. however, it is clear that the validity domain for crisis management has become too small for the world as it is now (figure 1). in finance, decisions have to be taken fast, as the regulation time (~seconds) is much longer than the time of the actors (~milliseconds) (ref. 23). with modern transportation, diseases can spread in a matter of hours: sars, h1n1... modern means of communication lead to an information deluge (which played a key role in the arab world revolutions), and unverified rumors to spread, in matter of instants (twitter, etc.). companies are global, they rely on just-in-time business, but controls and regulations are only effective on a national scale, and in specific sectors. more important, interconnectivity and interdependence has become increasingly important and can lead to major disruption such as the north american power failures of 2003. and at this very point in time, the fundamental change that began in the 21st century has been exposed: the power that has propelled individuals thus far has been inverted from personal ego to global ego, and has locked us all into a vicious circle. warning signals are numerous and the time has come to acknowledge that “more of the same” is not enough. as sunzi would phrase it: “when the paradigm is false, you are bound to fail in every battle”. 3.2. when science leads the way: a theoretical framework for modern crises the world is experiencing one of the worst financial crises in its entire history, therefore an analysis of what is happening can help understand how science can contribute to a better understanding of crisis and help in the stabilization of the global system. 3.2.1. science and finance: a good indication of what must (and must not) be done benoit mandelbrot, father of fractal theory, had understood a long time ago that the basic techniques figure 1. evolution of the number of internet users in the world (2000-2010) published by atlantis press copyright: the authors 25 p. lagadec and b. topper of analysis developed in mathematics were very effective in some contexts but could be proven wrong when applied to complex phenomenon like price variation, which is calculated through the interaction of a large number of actors, using different strategies. he had shown (ref. 24) that the basic market assumption that price variations followed a simple gaussian distribution (bell shaped), in which extreme changes (“black swans”) were quasi inexistent, was deeply incorrect. as anyone who follows price movements knows all too well, statistically significant changes can be seen every day in the financial pages. and most of them are independent on the news (ref. 25) i.e. they are large internal fluctuations, a phenomenon that any scientist knows to happen generically in a complex system, but that models based on gaussian distributions assume to be non-existent. this proves that the efficient market hypothesis and the blackscholes model, commonly used by traders, are flawed. over a year, the most important changes can occur in just a handful of days, and large price changes can happen without any external trigger. one could instead use a pareto distribution (figure 2.) which has a long and thick “tail” and that can therefore account for black swan events. of course, the math behind it is quite different, but the agreement with real data, and especially with extreme events, is much better (ref. 25). however, even though mandelbrot’s ideas were, and are still, considered groundbreaking and remarkable, they imply such a paradigm shift that they are largely ignored in the financial community that keeps working with “white swan only” models . the context is the same on a more global scale. our current scheme does not take into account complexity as being the inherent nature of the world we live in. crises are still seen as an unusual event that can be treated separately, on its own, although they have become the defining process of our modern societies. 3.2.2. the foundations of a new paradigm we do feel it important to recall that, when confronted with a large number of contradictory observations (table 1), a paradigm shift, based on a more precise understanding of the dynamics of complex systems, can offer new insights and effective solutions to seemingly unpredictable behavior (ref. 26). in physics, the lyapunov theorem on stability of systems states that "in the vicinity of its equilibrium points, the solutions of a non-linear system are similar to the ones of the equivalent linear system". this means that as long as your system is near its equilibrium point, you can use the techniques usually used for linear systems to get answers on the behavior of the nonlinear system. it is a non trivial theorem that can explain why techniques in risk and crisis management could still be used quite effectively during the previous decade or so, even though complexity had already become increasingly apparent. however, the need, in the current framework of crisis management, for adding more and more parameters to describe the behavior of the system should have been a clear indication that something had changed. one cannot hope to describe a nonlinear dynamic system with a patchwork of simple parameters. a non-linear system implies, in most cases, unpredictable behavior, such as the inversion of the earth's magnetic field in physics. we must now prepare for the unexpected, and not predict the predictable. we must be more creative, learn to be surprised, and to act rationally and creatively during the phase of ignorance, information surge and shock. figure 2. schematic comparison of the tails of a gaussian and a pareto distribution published by atlantis press copyright: the authors 26 how crises model the modern world (i) be prepared to be surprised. piling up written plans, rules, “coordination” committees and “communication” best practices remains the norm. people are prepared to manage according to solid rules, anchored on solid ground, and selected for their “inside the box” excellence. as soon as people discover that the situation does not correspond to the nominal framework, they tend to suffer instant paralysis and breakdown. "nothing is more difficult and therefore more precious than the ability to decide," said napoleon bonaparte. while those words refer to battlefield commanders' ability to take action when under fire, they also resonate with anyone in charge who has ever had to handle a crisis situation. in such a situation, the challenge is to make a decision quickly and accurately. in the current state of mind, a key component is to balance "gut feeling"--acting without thinking,--with "analysis paralysis"--thinking instead of acting. acting impulsively can lead to a worsening of the situation. likewise, overthinking a situation, weighing options endlessly can end up in leaving the entire operation in a state of stasis instead of recovering from the initial setback. this is an invitation to radically change the vision of crisis and crisis handling, the education and training of all actors involved: “when training, federal officials should not shy away from exercising worst-case scenarios that “break” our homeland security system” (ref. 27). it is equally trying for people wrestling at the theoretical level, as todd laporte stressed years ago: “what are our special intellectual obligations when we are burdened with the weight of theories of simple systems persisting in a world of vanishing simplicity?” (ref. 28). (ii) be prepared to be creative. the study of analogous physical systems can therefore teach us some new decisive visions and approaches. in the management of crises, it would mean:  getting rid of automatisms when necessary and being ready not to act according to the plans  taking a step back with the intervention of actors that are not submerged by the immediate operational constraints  shaping a strategic reorientation with the realization that actors all have very different characteristic time-frames; the major challenge is less in terms of “doing” and more in terms of “knowing what to do, and who can help”  considering innovation as a principle of action by constructing new convergences between the actors with regard to the principles, objectives and operating modes  understanding crisis as an opportunity to change visions, resolutely opening new horizons for action. this detour has taught us many lessons on how our views can be changed because of new, emerging properties stemming from the dynamic interactions in our society and how our techniques can be improved and rethought. 4. strategic recommendations: innovative methods put to the test “people in government are overwhelmed by crises […]. they do not have much time to step back and consider the big picture” (ref. 29). the same is true in the private sector. the practical challenge is to implement such a vision within the institutional design. modern crises must be seen as an opportunity for more innovation, for more action. 4.1. leadership as stated earlier, the major challenge today is to prepare leaders so that the creative approach will prevail at the right moment. unfortunately, the entire organizational, administrative and institutional culture normally tends to fall into procedural thinking (ref. 30). and this fundamental logic is not going to be turned around by devoting a few hours a year to formal “crisis management" seminars – essentially oriented towards the teaching of the “good answers” and “best and certified talk-points”. published by atlantis press copyright: the authors 27 p. lagadec and b. topper the leaders themselves should be mentally prepared to take an approach to intelligence and action that is more creative than procedural. the problem is that our habits at times of emergency and crisis are usually just the opposite. with very little information available and even less of it verified, the leader must have the conviction and the vision to lead the community out of its initial disorientation, and to avoid the two pitfalls that are always present in extreme crises: bureaucratic inertia (where each organization waits until the crisis fits its codes and rules), and the general loss of nerve (not only within the public, but throughout the entire chain of command). it is only by inspiring confidence that we can get through the ordeal, renew our energy, and come up with innovative plans and concrete paths to success. and confidence can be gained by the ability to call upon an experimented and ready-to-question crisis team. this innovative method is the “rapid reflection force”, which was theorized a few years ago and has already been experimented in various cases. but, as every new thinking process, the feedback will only be significant once a large scale implementation has been implemented. 4.2. the rapid reflection force: a discovery process the rrf is a group whose task is to help the leader to grasp and confront issues raised by unconventional situations. it does so by developing unconventional responses when usual toolkits and references turn out to be irrelevant, or indeed dangerous. along with the more "tactical" crisis teams, focused entirely on immediate operational responses, plans and logistics, such rrf teams promptly undertake four broad lines of interlinked questioning processes.  what is the essence of the problem? it is crucial to understand the nature of the crisis, and to anticipate the possible mutations of the challenges.  what are the major pitfalls? when the pressure of events becomes extreme, a very normal tendency is to become mired in the most counterproductive ruts. it is crucial, immediately, to think about the major errors to avoid. and the first is a wrong framing of the issue.  what is the map of actors; what networks are needed? extreme crises strike at the system in ways that are hard to anticipate, and involve complex emerging networks of actors. mapping those networks is crucial.  what constructive initiatives can the rrf suggest? the most important thing is not to pore over statistical lists or to compile all the information possible, but rather to try to discern one or a few critical initiatives that could introduce a new set of rules. the ambition is not, or cannot be, to put the finger on ‘the’ magic formula, but to create conditions and avenues for improvement. the point is not only to be successful, but to act wisely. a key success factor for such a support team, directly reporting to the highest level, is diversity of background and an ability to think and deliver creative options under pressure and in “the middle of nowhere”, with no information or worse, only deceptive information. the rapid reflection force must also resist, through its constant questioning process, making the assumption that it has found the “right” course of action. rrf core responsibility is to open options to the highest level, not to substitute leadership. in its essence, the rrf is a genuine “discovery” process applied to a sudden departure into terra incognita. it is a new managerial approach for an increasingly complex world. it clearly translates the shift required: a new vision to grasp the challenges, a genuine effort to escape from the mere application of conventional “best practices” – which is generally a terrible pitfall. 4.3. facing the real world this innovative approach to crisis management has already been implemented on a series of occasions, and is now gaining in maturity. generic lessons have emerged from all of this. published by atlantis press copyright: the authors 28 how crises model the modern world électricité de france was the first to implement such a process within its crisis platform; the rrf was used in multiple circumstances (refs. 31-32). real life incidents and exercises have shown that the rrf can genuinely become a pillar of strength around which an organization can group up:  over a seven-month period, (in august and november 2006, and then again in february 2007), one of edf’s nuclear plants (chinon) was hit by a cluster of tragic events, as three of its employees committed suicide.  in september 2007, a very ambitious simulation exercise was held by edf, based on the scenario of a breakdown in information systems. the ‘fog of war’ was such that it was unclear whether the event was due to a national terrorist attack or merely to a localized disruption. this raised a serious challenge for the ce level, as the appropriate posture would differ dramatically depending on how the situation was interpreted. the rrf proved invaluable in helping the upper echelon make sense of the resulting ‘phony war’, and was the first to understand that the situation was not a case of global terror, but was due to insufficient protection at a single site – a conclusion which called for a specific communication strategy  communication policies have also changed radically. for years, the norm in crisis communication had been to prepare the initial communiqué, followed by a media briefing, and high-profile tv interviews, especially in nationally-televised newscasts. and again, the rrf was crucial numerous times as it helped the ce representative and the communication team to build a strategic response that reflected the new challenges. the same kind of process has been used in 2006 by aéroports de paris. efforts are being devoted just now to train people to use and implement such an rrf in various countries such as belgium in the railway system, or in switzerland. the rrf can benefit all. on a global scale, it can help an entire organization develop strength, coherence, stability, and strategic intelligence, and thereby address the most difficult – and increasingly frequent – challenges of our turbulent times. the rrf is also a steady driver for benchmarking, partnerships, and shared initiatives at the chancellery level. extensive feedback shows that these rapid reflection forces are crucial for leaders, from blowing the whistle – when the challenge is not the frequently mentioned “weak signal”, but the “strange” incomprehensible signal – to re-checking the organizational response, and above all to open entirely new pathways during difficult episodes. it’s not only data harvesting, it’s also data analysis. on the one hand, it is now clear that the rrf can play a crucial role in the decision making process for leaders. but on the other hand it cannot, and should not, replace other functions: neither operations nor communication, nor above all, the organization’s strategic team. this initial feedback suggests where the goalposts are to be set: everyone within the crisis platform should be trained to take full advantage of the rrf, but they must also retain their own crucial mission. but one cannot sufficiently stress that organizations and leaders only trained to consider known, independent, and common risks, only prepared to consider mapped crises, show strong reluctance to accepting any rrf process. they stick to preplanned courses of action, and they fail. the real challenge is a cultural change. 4.4. current developments: the center for transatlantic relations advanced seminars. in 2006, the center for transatlantic relations (sais, johns hopkins university) launched the project “unconventional crises, unconventional responses: reforming leadership in the age of catastrophic crises and ‘hyper complexity’” (ref. 33) it sets up an international platform of leaders and experts that share questions, practices, and innovations. it brings together top officials from public sector (i.e. dhs, fema), critical networks (energy, banks, transport, information, etc.), ngos and the red cross, and academics worldwide with a challenging rule: the only published by atlantis press copyright: the authors 29 p. lagadec and b. topper questions explored are those that are not known, for which there are no ready-to-use solutions or plans. a fresh impulse is now on its way with new and resolute signs of interest from the highest level of key organizations. the cornerstone is the understanding that crisis management is not so much a crucial need for additional “tool-boxes, practices, frameworks”, but rather "fresh vision" approaches and capacities to deal with hypercomplexity, systemic risks, and global volatility. 5. discussion this new frontier will confront us with totally unknown theatres of operations, with overlapping and rippling crises. shaping the future will require new visions and practices if we want to rise to the challenge. 5.1. new frontiers in a volatile world in our search for unconventional territories, we have chosen to pick four domains in which we think crisis reflects the interconnectivity and the complexity our societies have reached, and therefore the sectors that are among the most at risk. this short discussion should not be seen as a prediction game, but as an addendum to our analysis of the global situation and the major nodes of our societies, which are therefore critical components for its stability. 5.1.1. energy the most critical domain for coming crises is probably the energy sector, and specifically the disappearance of cheap and easily extractible oil reserves [34]. without doubt this sector is most likely to trigger a global crisis that could jeopardize global economic growth and bring social upheavals in the coming decade. cheap energy is a key source of economic growth, and not the other way round, because energy is the physical means of transforming the world. the scarcity of this primary energy source, representing approximately 35% of the world energy production, is an event that humanity has never had to face, and that it is about to have to overcome. 5.1.2. cyber attacks world’s public and private institutions are more and more subject to threat coming from the cyber world. the information technology infrastructure is now vital for communication, commerce, and control of physical infrastructures, and is highly vulnerable to terrorist and criminal attacks. the sophistication of our world is in itself a source of vulnerability: information technology drives critical industries such as aviation, electricity or water supply, banking and finance. our time is one of global leaks, with the use of websites such as wikileaks, and cyber attacks, when dispatched groups of individuals, like anonymous and lulzsec decide to steal data or block a website. but also to cyber wars, as when iran had to face a computer virus, stuxnet, affecting its nuclear program. the world institutions are effectively in a state of cyber-siege and cyber-chaos, enduring attacks from people who want to steal or reveal confidential information, to bring websites down, or to destroy as much information as possible. 5.1.3. from unity to separation: towards geopolitical complexity volatility and disconnection appear to transform institutional foundations on the international, national and regional levels. this global shift – from cohesion to dysfunction, from unity to separation – is particularly clear in the emergence of separatist movements. for instance in europe, the severe disagreements in belgium between flemish and walloons. the country was without effective government for over a year. french-speaking wallonia in the south and flemishspeaking flanders in the north disagree on pretty much everything, from the handling of the economy to the war in libya. this situation actually reflects a much broader problem. the eu is getting more and more concerned that the divorce of belgium could spark a domino effect across europe: the basque country leaving spain, scotland leaving the uk (the scottish national published by atlantis press copyright: the authors 30 how crises model the modern world party won a majority of seats in scotland's parliament during last election)... in italy, the separatist attitudes are strong in the industrially advanced northern regions towards the poorer south of the country. but separatist movements are also gaining momentum all over the world, in developed, emerging and developing countries alike. therefore we could witness a chain reaction of small states gaining independence. such a plunge into the unknown would imply weaker governments threatened by inner rebellions, and a rise in tension between states with a surge in border disagreements. this would imply a completely new framework for crisis management, whatever the nature of the crisis. 5.1.4. debt issues and social upheavals after the 2007 financial crisis, most sovereign states had to help the banking system which constitutes the foundation of the global economy, and decided to implement stimulus packages to boost growth, which were all financed by a growing debt. the debt issue is now a huge problem we are facing: investors are losing confidence in the ability of sovereign states to pay, and the subsequent budget cuts needed to restore confidence generates social upheavals. first we heard about greece. then, we learnt about the “piigs” – a very inelegant acronym for portugal, ireland, italy, greece and spain – who are having sovereign debt issues. these countries represent a growing percentage of the world gdp. globalization means that the debt of each of these countries is owned by investors around the world. for a long time, dissemination of risks meant greater safety, as risks were diluted among separate actors. now that the world is global and interconnected, risk still seems to be distributed around the world, but because of global tight coupling nobody knows who is exposed, and to what level. and now, at the time of writing, we are witnessing the emergence of a surprising global movement "occupy wall street". it may be ephemeral. but it may be much more. here again, nobody seems to be controlling or leading. volatility could engender many different and surprising developments. this last point is a perfect analogy for what global crisis has now become. but again, such explorations must not lead back to the common segmented approach to risks and crises. crises today will most probably combine, in a strange and unanticipated way, many dimensions of such challenges and many others emerging brutally from nowhere. the imperative is not to categorize and apply, but to question the present and to shape the future. 5.2. research opportunities: what needs to be done it may be quite unusual for a paper on crisis management not to advocate for new tools or for new "best-practices". our position it that crisis management has to reinvent its foundations if it is to become truly operational. this can only be done through a paradigm shift. of course, this may seem far removed from the immediate realities, but when tactics has proven powerless as it is now the case, the first priority is to restore the strategic depth, global vision, revisiting and strengthening our theoretical grounds instead of rushing down the path of micro-management. calling for additional research efforts is a rule in any academic contribution. in most cases, authors are pleading for more research on specific subjects – such as “children in disasters”, “women in disasters”, “emerging groups in crises”, etc.; or on specific dimensions – such as “communication in crises”, “recovery after catastrophic disasters”, and so on. the new frontiers, the new global contexts that we are facing today call for a more radical shift in the research agenda. in thomas kuhn’s words we need to go beyond “normal science” – which is fully appropriate as long as the usual paradigm works. as our basic assumptions – linearity, independence, etc. – have become ineffective, the time has come to translate that reality into science and research. we need to fully accept the need for a paradigm shift, even if such a leap is always extremely tiring – “creative scientists must occasionally be able to live in a world out of joint” (ref. 26). published by atlantis press copyright: the authors 31 p. lagadec and b. topper some very specific questions have to be dealt with. in fact, every single question, reference, know-how, practice has to be questioned and re-formatted. for instance, on matters of “crisis communication”, research has to address the new environment we live in: how to “communicate” when anyone can instantly reach billions of people. but we certainly need something else. we have to prepare researchers in such a way that they can picture the world differently. as a starting point here, it would be useful to think of the decisive move made by eminent scientists in the usa at the turn of the 19th century in the field of public health: “shortly before the great war began, the men who wanted to transform american medicine succeeded. they created a system that could produce people capable of thinking in a new way, capable of challenging the natural order. they, together with the first generation of scientists they had trained formed a cadre who stood on alert, hoping against but expecting and preparing for the eruption of an epidemic. when it came, they placed their lives in the path of the disease and applied all their knowledge and powers to defeat it. as it overwhelmed them, they concentrated on constructing the body of knowledge necessary to eventually triumph.” (ref. 35) we certainly have to address that kind of vital challenge today. obviously, a huge amount of work still needs to be done, especially in top education programs to train people to new global challenges. new answers must be searched for, and will surely be found, but we have to realize that “they will not simply fall as a gentle rain from the sky.” (w. shakespeare, the merchant of venice). references 1. yukiya amano, international atomic energy agency (iaea) director general, vienna, 21 march 2011. http://www.unmultimedia.org/tv/unifeed/d/172 51.html 2. barbara w. tuchman, the march of folly – from troy to vietnam, ballantine books, new york, 1984. 3. xavier guilhou, “devoir de protéger”: pourquoi le repenser?” la revue de géopolitique, 28 juillet 2011, diploweb.com, http://www.diploweb.com/devoir-de-protegerpourquoi-le.html 4. u.s. house of representatives: a failure of initiative, us house of representative, final report of the select bipartisan committee to investigate the preparation for and response to hurricane katrina, us government printing office, 15 february, 2006. 5. steven fink, crisis management. planning for the inevitable. amacom, american management association, 1986. 6. robert heath, crisis management for managers and executives, financial times, pitman, london, 1998. 7. patrick lagadec, preventing chaos in a crisis strategies for prevention, control and damage limitation, mcgraw hill, 1993. 8. mark haynes daniell, world of risk – next generation strategy for a volatile era, john wiley & sons, new york, 2000. 9. uriel rosenthal, michael t. charles, paul 't hart (ed.): coping with crises. the management of disasters, riots and terrorism, charles c. thomas publisher, springfield, illinois, 1989. 10. uriel rosenthal, r. arjen boin, louise k. comfort (ed.), managing crises – threat, dilemmas, opportunities, charles c. thomas publisher, springfield, illinois, 2001. 11. arjen boin, “from crisis to disaster: towards an integrative perspective”, in ronald w. perry & e.l. quarantelli, editors: what is a disaster? – new answers to old questions, international research committee on disasters, xlibris corporation, lexington, ky, 2005, p. 153-172. 12. patrick lagadec, “crisis management in the twenty-first century –“unthinkable” events in “unthinkable” contexts”, in havidan rodriguez, enrico l. quarantelli, and russel dynes: handbook of disaster research, springer, (chapter 30, pp. 489-507), 2006. 13. patrick lagadec, "over the edge of the world", crisis response journal, volume 3, issue 4, p. 48-49, 2007. 14. karl e. weick and kathleen m. sutcliffe, managing the unexpected, resilient performance in an age of uncertainty, john wiley & sons, inc. 15. joshua cooper ramo, the age of the unthinkable – why the new world disorder constantly surprises us and what we can do published by atlantis press copyright: the authors 32 how crises model the modern world about it, little, brown and company, new york, 2009. 16. gilles sanson (président de la mission interministérielle), « evaluation des dispositifs de secours et d’intervention mis en œuvre à l’occasion des tempêtes des 26 et 28 décembre 1999 », rapport d'étape de la mission interministérielle, premier ministre, juillet 2000. 17. the 9/11 commission report, final report of the national commission on terrorist attacks upon the united states, authorized edition, w.w. norton & company, new york, 2004. 18. admiral thad allen, "hurricane katrina: framing the issue – a weapon of mass destruction without criminal dimension", videotaped communication to hans de smet, department of economics, management & leadership, royal military academy, brussels. 19. senate majority leader on economy http://abcnews.go.com/video/playerindex?id=5 824988 20. nassim nicholas taleb, the black swan – the impact of the highly improbable, allen lane, penguin books, london, 2007. 21. joseph a. tainter, the collapse of complex societies, cambridge university press 22. rj zeckhauser and wk viscusi, risk within reason, science 4 may 1990: 559-564. [doi:10.1126/science.2333509]. 23. joint sec/cftc report, findings regarding the market events of may 6, 2010. 24. benoit mandelbrot, the variation of certain speculative prices. the journal of business of the university of chicago, 36, 394-419. 1963. 25. contributions of sciences and technologies to the evolution of financial markets, public hearing of 14 october 2010, mr claude birraux, deputy, president of opecst 26. tomas kuhn, the structure of scientific revolution, university of chicago press, 1962. 27. the white house: the federal response to hurricane katrina – lessons learned, 2006. 28. todd. r. ed. laporte, organizational social complexity: challenge to politics and policy, princeton university press, 1975. 29. lee hamilton, “prologue”, in thomas h. kean and lee h. hamilton, without precedent, the inside story of the 9/11 commission, alfred a. knoff, new york, 2006. 30. young, james (2007), “leadership in the age of unconventional crises and chaotic environment”, personal video-taped communication. 31. pierre béroux, xavier guilhou, patrick lagadec, "implementing rapid reflection forces", crisis response journal, vol. 3, issue 2, march 2007, pp. 36-37. 32. pierre béroux, xavier guilhou, patrick lagadec, "rapid reflection forces put to the reality test", crisis response journal, vol 4, issue 2, march 2008, pp. 38-40. 33. erwan lagadec, leadership in unconventional crises, a transatlantic and cross-sector assessment, center for transatlantic relations, the paul nitze school of advanced international studies, the johns hopkins university, washington dc., 2009. 34. international energy agence, world energy outlook 2010 35. john m. barry, the great influenza – the epic story of the deadliest plague in history, penguin books, new york, 2004, i senior research scientist at the ecole polytechnique (france); elected member of the french academy of engineering; scientific advisor to various organizations confronted with unconventional risks and crises. mail : plagadec@club-internet.fr ii theoretical physicist (upmc /ens ulm), junior science and technology advisor. mail : benjamin.topper@gmail.com published by atlantis press copyright: the authors 33 journal of risk analysis and crisis response vol. 9(4); january (2020), p. 203 doi: https://doi.org/10.2991/jracr.k.200120.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr introduction for volume 9, issue 4 this issue contains five papers. there are three contributions in english and four contributions in chinese with english abstracts. the papers can be divided into four topics: climate disaster risk, emergency management, ecological risk management, and big data. there are two papers in climate disaster risk. the first paper “interannual variation and hazard analysis of meteorological disasters in east china” by jun shi, linli cui and zhongping shen, analyzes the overall characteristics, interannual variations and the hazards of meteorological disasters in different provinces of east china based on the historical data of meteorological disasters in east china during 2004–2015. the results indicated that flood disaster (including landslide and mud-rock flow) induced by rainstorms had caused the largest affected area and total failure area of crops, and also caused the largest number of affected people and collapsed houses. strong convection weather (including gale, hail, thunder and lightning) disaster resulted in the largest number of deaths and typhoon disaster caused the greatest direct economic losses. there were significant decreasing trends in the affected area and the total failure area of crops, and the number of affected people and deaths during 2004–2015, while the direct economic loss caused by meteorological disasters showed no significant trend in east china. the hazards of meteorological disasters had obvious regional differences. in the northern part of east china, the hazards of drought and strong convection weather disasters were higher, but in the southern part, there were higher hazards of flood disaster. in the eastern coastal areas of east china, the hazards of typhoon disaster were higher. the second paper “climate risk assessment of rooftop solar resource development in inner mongolia” by yanan hu and xinghua li, calculate the solar energy resources abundance and stability index, combined with the annual surface meteorological data of 119 meteorological stations in inner mongolia from 1988 to 2018 by using the meteorological industry standard “solar energy resource assessment method”. the distribution characteristics of solar energy resources and meteorological elements in inner mongolia is analyzed and the suitability and meteorological risks of rooftop solar resources development and utilization are evaluated. there is one paper in emergency management. the paper “summary of intelligent guidance system for fire emergency evacuation in large buildings” by xiangzhi meng et al., reviews researches of intelligent guidance system for fire emergency evacuation in large buildings at home and abroad. three key problems of the fire detection, the evacuation path planning and the evacuation guidance design are presented. the development trend of the intelligent guidance system for fire emergency evacuation is discussed from two aspects, namely, the evacuation path planning methods and the guidance system hardware research. there is one paper in ecological risk management. the paper “prospects of guizhou province’s ecological agriculture benefit evaluation index system” by jian he and hongmei zhang, establishes the evaluation index system of ecological agriculture by reference to the evaluation index system of ecological agriculture at home and abroad. this study has a certain practical significance and provides theoretical guidance for the development of ecological agriculture in guizhou. there is one paper in big data. the paper “the harmonious development of big data industry and financial agglomeration in guizhou by junmeng lu and mu zhang, used 2015 cross-section data, the intuitionistic fuzzy analytic hierarchy process, the intuitionistic fuzzy number score function, the coupling model and the coupling coordination model to empirically research the coupling and coordination level between guizhou big data industry and financial agglomeration. the empirical research shows that there is an obvious imbalance in the coordinated development and obvious spatial heterogeneity of big data industry and financial agglomeration in guizhou. only guiyang and zunyi can achieve the coupling and coordinated development of big data industry and financial agglomeration. we sincerely thank the referees for their strong support and kindly help. thanks to all the authors for their submissions. particularly, thanks to prof. mu zhang, publication chair of the society for risk analysis – china, and thanks to prof. junxiang zhang, manager of journal of risk analysis and crisis response, they devoted their time to overseeing the reviews. editors-in-chief prof. chongfu huang beijing normal university, no. 19 xinjiekouwai street, beijing 100875, china email: hchongfu@126.com prof. gordon huang faculty of engineering and applied science, university of regina, regina, sask s4s 0a2, canada email: gordon.huang@uregina.ca © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). https://doi.org/10.2991/jracr.k.200120.001 https://www.atlantis-press.com/journals/jracr mailto:hchongfu%40126.com?subject= mailto:gordon.huang%40uregina.ca?subject= http://creativecommons.org/licenses/by-nc/4.0/ microsoft word internet of intelligences in risk analysis for online services.doc internet of intelligences in risk analysis for online services chongfu huang1,2,3 1state key laboratory of earth surface processes and resources ecology (beijing normal university) beijing 100875, china, hchongfu@bnu.edu.cn 2key laboratory of environmental change and natural disaster, ministry of education of china, beijing normal university, beijing 100875, china 3academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education the peoples' republic of china, beijing normal university, beijing 100875, china abstract there are always some people who experienced or studied the risks of daily life. based on that, in this paper, we proposed a new concept “internet of intelligences” (ioi), which consists of intelligent agents, the internet and a mathematics model. an ioi collects and processes the information provided by the agents with experiences and judgment in a risk. it serves for the customer who hasn’t any experience in a specific risk. the key of constructing a positive ioi is to filter malicious interference, process inconsistent information, and carry out an optimal treatment on incomplete information. its core technology is the data processing method with information diffusion. the online services of ioi can help us, such as choosing health foods, correctly filling in the college candidate voluntary and guaranteeing love affairs more romantic. keywords: risk, intelligent agent, internet, information diffusion, online service. 风险分析在线服务的智联网 黄崇福 1,2,3 1.北京师范大学地表过程与资源生态国家重点实验室，北京 100875 2. 北京师范大学环境演变与自然灾害教育部重点实验室，北京 100875 3. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875 摘要：基于总有一些人已经历或研究过日常生活中的风险问题，本文首次提出“智联网”的概念，它由 智能体、互联网和一个数学模型构成，以汇集和处理相关信息，为面对风险而又没有经验的人们提供服务。过 滤恶意假信息，处理不协调信息，优化处理不完备信息等，是构建正向智联网的关键，其核心技术是信息扩散 的数据处理方法。智联网的在线服务，有望帮助人们有效解决远离有害食品、正确填报高考志愿和使恋爱更符 合个性等问题。 关键词：风险，智能体，互联网，信息扩散，在线服务 人类社会经历了从农业阶段到工业阶段，再到 信息化阶段三次巨大浪潮。目前，人类正处于第三 次浪潮之中，其特点是形形色色的互联网无处不 在。移动通信、物联网、战术数据链等大型的信息 系统，无一例外皆有互联网的烙印。然而，今天的 互联网系统，远非智能系统，更多的是信息发布和 交换系统。以此支持的在线服务，并没有超出传统 服务的内涵，只不过速度更快，花样更多，成本更 低。换言之，如果不计成本和时间，互联网能提供 的服务，传统手段均能提供。互联网真正的威力尚 在发掘之中，智联网或许将成为其突破点之一。 1. 智联网概念 “智联网”(internet of intelligences)，是由各种 智能体，通过互联网形成的一个巨大网络。其目的 是集小智慧为大智慧，群策群力，帮助人们更好地 认识世界，获得更好的生活质量。 本文首次提出的“智联网”，将是继人工智能 和计算智能之后的智能科学大革命，同时也将推动 世界信息产业再上一个新台阶。智联网的基础是互 联网，智联网的核心技术是信息扩散的数据处理， 智联网的发展动力是有偿智能服务。 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 110-117 published by atlantis press copyright: the authors 110 与传统咨询服务业 大的不同是，提供智联网 服务的不再限于专家。对某些实际问题，民众的经 验可能强于专家的知识。对于巨大系统的复杂问 题，专家群体的意见往往强于个别权威的判断。通 过“智联网”获得尽可能正确的明智选择，将成为 人们日常生活中的重要助手。 设 a={a1,a2,…,an}是含有的 n 个智能体 a1,a2,…,an（例如某大学的 n 个教授）的一个集合。 设 n={s,c1,c2,…,cn}是含有服务器 s 和 n 台计算机网 络终端 c1,c2,…,cn 的，一个可以独立工作的子系统， 例如一个局域网。设 m 是处理 n 个智能体所提供的 信息的模型。三元体（a, n, m）称为一个智联网。 假定存在某一准则 r，可以判定 ai，i=1,2,…n, 的智力水平，记为 qi=r(ai)。假定 r 还能判定三元体 （a, n, m）的智力水平，记为 q=r(a, n, m)。当 q>max{q1,q2,…,qn}时，我们称其为一个正向智联 网。 我们可以举例来说明什么是智联网，什么是正 向智联网。三人团队 a={张三，李四，王五}参加过 传统智力竞赛，即，现场抢答，依标准答案现场判 分。接下来，张三，李四，王五改为各使用一台由 服务器 s 管理的网络 n 参赛，其他参赛队仍沿袭现 场抢答方式。假定我们设计出了某种可以实时处理 a 队所提供信息的模型 m，并依据处理结果回答问 题。由参赛队 a，计算机网络 n 和信息处理模型 m 组成的系统是一个智联网。假定在以往参赛中 高 智力者为张三，称为 a 队的传统智力分，记为 q。 再假定 a 队用网络参赛而得的网络智力分为 q。当 q>q 时，（a, n, m）就是一个正向智联网，此时， a 队的得分一定高于传统方式参赛的得分。 简单的智联网可以用图 1 示之，该系统由三 个智能体 a1,a2, a3，三台终端机 c1,c2,c3，一台服务器 s 和一个信息处理模型 m 组成。 事实上，并非智能体联网就能形成智联网。例 如，医学影像远程网络会诊并不是智联网，因为系 统中并不存在本质上能对专家意见进行处理的任何 模型。并非智联网均有正向性，例如，用一个简单 的统计模型支撑的智联网，理论上来讲，智力水平 为网中智能体的平均值。显然，智联网的核心技术 在于网络信息处理模型 m。下面我们给出智联网及 其有关概念的形式化定义。 定义 1. 具有观察、演绎、推理和解决问题能力的个 体称为智能体。 例如，正常的个人、决策支持系统、恒温调节 器、模糊洗衣机等，均是智能体。 c1 c2 c3 a1 a2 a3 s(网 络 服 务 器 ) m(信 息 处 理 模 型 ) 图 1. 简单的智联网。图中的三台计算机形成网 络，三个工作人员是三个智能体。 定义 2. 智能体解决问题的能力大小称为智力水平。 例如，中学生参加智力竞赛的得分高低通常能 反映其智力水平。由于各种智力竞赛的难易程度和 测试方式不同，由得分反映的智力水平具有相对 性。 定义 3. 设 a 是一个智能体集合，n 是 a 使用的一个 网络，m 是处理 a 所提供信息的模型，三元体（a, n, m）称为一个智联网。 换言之，一个由智能体、计算机网络和信息处 理器组成的系统，称为一个智联网。当其智力水平 超越每个个体智力水平时，称为一个正向智联网。 即， 定义 4. 设（a, n, m）是一个智联网，智力水平为 q。再设 a 中个体的 大智力水平为 q。当 q 大于 q 时，称（a, n, m）为一个正向智联网。 互联网是一个动态混沌系统，没有人知道每一 个节点何时会发生什么变化，也不知这种变化对其 它节点会产生什么影响。智联网则是一个智力提升 系统，只有正向智联网才有存在的意义。智联网具 有动态自组织性，目标是形成一个高级智能体。互 联网从混沌系统走向自组织系统，必将是一次革 命。 经过对比互联网应用模式的发展和人类大脑的 结构机理，人们发现，二者有惊人的重合。例如， 从电子邮件、电子公告牌到博客、社会化网络，互 联网的每一个应用创新都能映射到人脑的功能结构 published by atlantis press copyright: the authors 111 中[1]。人们预测，互联网的应用模式将形成一个与 人类大脑功能结构高度相似的网络虚拟结构，互联 网正朝着使人脑充分互联的方向发展和进化。然 而，现有的互联网技术，尚不足以自动形成超越个 体智能体的智能，ibm 为其百年大庆于 2011 年推出 的，能与冠军一比高下的沃森[2]，不过是以往经验 的学习而已。只有智联网，才有可能使网络成长为 超级智能体，也就是个体智能的组合，超越简单智 能的叠加。 2. 风险分析与智联网 解决如何识别、规避、权衡风险的问题，是人 们得以正常生活的基本保障。如果过马路时不能识 别风险，就可能发生交通事故；如果地震区的建筑 物不设防，就无法规避地震风险；如果感染后过量 吃抗生素，就会让病菌产生耐药性。解决风险问 题，基础是风险分析。 使用有关理论和方法，在相关知识和数据资料 的基础上，对未来不利事件出现的可能性、规模、 影响等进行的分析，称为风险分析。风险分析是为 了认识风险，为风险管理提供科学依据，使未来情 景向好的方向转变。 复杂问题的风险分析，需要相当的智慧。提高 监测手段、综合使用各种技术和信息、大量访问专 家等，是目前解决复杂风险问题的主要途径。然 而，无论监测手段多么完善，大量的风险分析工作 还是遇到信息不完备的问题；无论综合多少技术和 信息，分析中的许多缺陷仍然暴露无遗；无论访问 多少专家，结论也不会比简单统计更有说服力。 由于正向智联网能超越个体智力水平，用智联 网改善现有的风险分析技术，极具发展前途。相应 地，我们可以发展出监测型智联网、综合型智联网 和评价型智联网。 2.1 监测型智联网 设 a 是含有的 n 个智能监测器（例如一台复杂 机器人配置的 n 个智能传感器）的一个集合，n 是 连接这 n 个智能监测器的一个计算机网络，m 是处 理 n 个智能监测器所提供的信息的模型，三元体（a, n, m）称为一个监测型智联网。 例如，当一台机器人识别其环境风险时，监测 型智联网能提高其识别能力。当我们监测自然和社 会的各个机构能形成监测型智联网时，我们对自然 灾害风险的监测能力就会大大提高。 监测型智联网正向化的基础是 m 能使监测器联 动起来，并能生成比单个监测器更具使用价值的信 息。例如，一个能快速捕捉变化目标的监测器能通 过 n 和 m 使高精度观察器及时联动到目标上来。多 个监测器的信息经 m 处理能对被监测对象有超越直 观监测的认识。 监测型智联网有可能推进物联网的智能化。地 震监测智联网有可能为提高地震预报精度创造条 件。 2.2 综合型智联网 设 a 由 n 个智能体组成，他们分别使用不同的 技术（例如，线性回归技术、人工神经元网络技术 等）或信息源(例如，遥感信息、气象信息等）对某 一风险问题进行研究，n 是连接这 n 个智能体的一 个计算机网络，m 是综合处理它们所提供信息的模 型，三元体（a, n, m）称为一个综合型智联网。 例如，某灾害管理机构建有“天―地―现场” 一体化业务平台，分析人员分别使用无人飞机、卫 星遥感和地面观测数据等对气象灾害风险进行分析 并通过计算网络传送给综合模型 m 实时处理，这就 形成一个综合型智联网。 从分布式处理、并行处理和网格计算发展而来 的云计算，将在综合型智联网方面发挥重要作用， 因为系统的综合度越高，实时处理的任务越重。 2.3 评价型智联网 设 a 由 n 个评价人员组成，n 是连接他们的一 个计算机网络，m 是处理 a 所提供评价结果的模 型，三元体（a, n, m）称为一个评价型智联网。 例如，根据欧盟预防和控制工业风险的塞维索 指令 ii，八位专家分别对某化工厂进行风险评价， 并通过计算网络传送给模型 m 进行处理，这就形成 一个评价型智联网。 与传统的综合处理专家评价结果 大的不同 是，评价型智联网不仅使专家间的匿名提示、协 调、互动更加方便、有效，也为评价结果超越参与 专家的水平创造了条件。 3. 由智联网支撑的风险评价在线服务 风险是与某种不利事件有关的一种未来情景 [3]。由个人或社会观察或感觉到的某种事件称为情 景，它必须由包含时间、地点、对象等要素的系统 来加以描述。“有关”之程度应该用某个度量空间来 测度，例如，概率。“不利事件”的强度应该被量 化，例如，损失的量。显然, 任何已发生的灾难都不 是风险；未来的不同时段，风险也不同。 published by atlantis press copyright: the authors 112 显然，风险因人而异。2007 年中石油在国内上 市，随后其股票大跌。对于持有其较多股票的人而 言，风险很大；对于不持有其股票的人，中石油股 票是否大跌，与他们无关，没有风险。 风险问题，举不胜举，既有因人而异的风险， 更有大量人们尚无法认识的风险。今天，在似乎人 人都懂风险，人人都会进行风险评估的时代，五花 八门的预测模型、拟合模型、插值模型等被称为是 风险分析模型，而且，“专家打分”进行风险评 估，似乎成了灵丹妙药。然而，现实情况是，人们 对风险的研究尚处于初级阶段，这不仅反映在人类 甚至于没有一个共同认知的风险概念，而且对地震 风险、经济风险等重大风险问题也缺少令人信服的 分析理论和方法，每每在灾难来临前侥幸多于掌 控。 不过，在大量风险问题中，简单风险在一定的 条件下可以被正确认识并加以规避。简单风险是指 人们有能力认识的风险。条件是曾经有人经历过， 并想出了规避的方法。有经验积累的简单风险，对 经历过的人而言，能造成伤害的可能性大大降低， 而对没有经历过的人，风险就比较大。 例如，摸黑走过一条左边有大坑的路，曾经掉 进坑里的人知道走左边有危险，应该绕着走，但是 对从来没有走过的人，他摸黑走该路就有较大风 险，他或许掉进坑里去，或许顺利通过。这里，危 险是指确定性的伤害，风险是指与伤害有关的一种 未来情景。由于前进路线的不确定性，伤害可能出 现，也可能不出现。 简单风险，也可能涉及复杂的系统，以个人的 经验，尚难全面认识。特别地，当风险系统随着时 间发生明显变化时，个人经验可能失效。此时，用 评价型智联网集成风险经历者的经验，并尽量对集 成经验加以提升，就能提供风险评价的在线服务。 有三类简单风险较适合于在线服务：信心类简单风 险、决策类简单风险和浪漫类简单风险。 3.1 信心类简单风险在线服务 信心的 精辟定义出自《圣经• 希伯来书》： “信心是所望之事的实底，是未见之事的确据。” 换言之，信心就是人所希望实现的事情的根据，是 还没有发生、没有看见的事情必然要发生和被看见 的证据。 信心的来源和本义是宗教领域根本性的问题。 德国威丁堡大学的马丁• 路德教授在 1517 年的宗教 改革运动中提出的“因信称义”的学说认为，人单 单凭着信仰，无需教会这个媒介，就能够直接与上 帝相沟通。只要有信心，而无需其他的条件和资 格，就能获得生命信仰上的解脱，获得上天对你的 拯救。于是，信心成为衡量一个人信仰的标准，惟 有信心能成就一个人的信仰，而其他所有外在的行 为、条件、资格、外力都不能成为信仰的依据。信 心也成为人在世间生活、工作是否有意义的根据， 成为能够战胜艰难险阻、开拓出美好喜乐的生活的 动力。信心在人类史上开始具有了神圣的意味，成 为人们生活和工作当中的强大精神力量。今天的清 教徒，是 为虔诚的新教徒，他们完全秉持“因信 称义”的学说，他们在信仰方面非常强调信心的作 用。正是凭着这种信心，清教徒们不远万里到北美 拓荒。他们坚信，这里肯定会是一个繁荣兴旺、日 新月异的新国度；正是这种信心，使他们获得了独 立战争的胜利，创建了美利坚合众国。可以说，是 信心成就了清教徒，是信心成就了美利坚。 然而，世俗生活中的信心并不等同于宗教上的 信心。毫无根据的盲信，足以带来重大伤害。在日 常生活中，我们对一些事有信心，因为我们有较充 足的证据信任它；对一些事则没有信心，感觉到有 些把握不住，接触它可能会造成伤害。例如，我们 对中国乒乓球队信心十足，因他们在世界乒乓球锦 标赛上连年夺冠。但是，我们对中国的食品安全整 体没有信心，因为转型期的中国对食品安全的监管 严重缺失。2008 年爆发的三聚氰胺污染事件极大地 打击了人们对国产奶粉的信心，全国有近 30 万名婴 幼儿因食用问题奶粉患泌尿系统结石[4]。2011 年曝 光的“全国每年返回餐桌的地沟油 200 万到 300 万 吨[5]”似乎说明中国的食品安全几近失控。 在宗教上，人的信心像芥菜一样，开始看起来 比较小，但未来成就非常巨大，超过开始时几百甚 至几万倍，人就应该拥有这种必将成就万千成果的 信心。信心的大小不是按目前的想法比较的，而是 按照将来要成就的成果来比较大小的。人虽然比山 渺小，但他如果有移山的信心，他的信心就比巍峨 的大山还要大。 在世俗生活中，人的信心像凭据一样，信心的 大小是由目前的处境决定的。信心由三个因素决 定：行动实现难度的认知、个人或团体的情绪和外 在意识。认知是指人们对行为必定成功的认识过 程，由于这种认识过程只能是对行为未来发展状况 的预期，所以这种认识过程实际上又是一种对行为 过程的想象和推断；情绪是指有机体在受到生活环 境中的刺激时，生物需要是否获得满足而产生的暂 时的较剧烈的评价和体验，它包括喜、怒、忧、 思、悲、恐、惊七种；外在意识是指人们在行为中 大脑对外界事物觉察的清醒程度和反应灵敏程度， published by atlantis press copyright: the authors 113 人们在睡眠时意识水平 低，在注意力高度凝聚时 意识水平 高。 与风险有关的信心问题，其实就信心的逆问 题。以人们对消费品的信心为例，如果有大量的真 凭实据表明某一消费品存在明显的安全问题，消费 者使用它就面对较大的被伤害风险。这时人们对此 消费品的信心就低；反之，如果风险很小，信心就 高。 信心类简单风险在线服务，就是构建一个计算 信心指数的信息处理模型 m，由其支撑一个智联网 （a, n, m）。 信心指数原本是债券市场上的一个指标，定义 为 10 个顶级公司债券的平均收益率与 10 个中等级 别公司债券的平均收益率的比值。这个比值总是小 于 100%，因为高级别债券所承诺提供的到期收益率 较小。当债券交易者对经济表示乐观时，他们对低 级别债券所要求违约溢酬率就较低，这两类债券的 收益率差就变小，信心指数接近 100%。因此，较高 的信心指数通常是牛市的信号。这里的模型 m 可表 为 平均收益率10个中等公司债券的 平均收益率10个顶级公司债券的 信心 = 如果从风险分析的角度来看待消费品的信心指 数，我们可将其简单地定义为消费满意度。当所有 消费者都 100%地满意时，信心指数为 1；当所有消 费者都绝对不满意时，信心指数为 0。 现实中的问题是，任何调查机构都只能采取抽 样方式进行调查，并且大多数机构的委托人是商品 生产者，调查机构为了迎合雇主的需要，无法确保 调查结果的客观性。另一个更为棘手的问题是，调 查数据中常常夹杂有许多不实信息，如何排除，并 非易事。我们可以用智联网技术很好地解决这些问 题。 设 o 是待评估对象，a 是由 n 个消费者组成的 智能体，他们经由互联网分别根据自己的消费经历 对 o 进行定性和定量描述及评价。由于智联网的开 放性和成本低，调查的面可以很广。当我们使用基 于信息扩散[6]技术对调查进行分析后，就可以得知 消费群体想表达的意思，从而评估出他们对 o 的信 心。设 o1, o2, …, ot 是由智联网评估的 t 个同类消费 品，将它们依信心指数的大小从上到下排序，就能 提供风险在线服务。 经由智联网提供这类服务的重点是个性化。人 们从信心指数排名上只能看到综合而笼统的结果， 并不提供性价比参数，更没消费品适合对象的信 息。两个信心指数相近的消费品，可能价格相差较 大。信息回溯功能不仅可以让潜在消费者根据自己 选定的指标进行评价，还可以看到一些事故案例。 当自己能承受时，信心指数随即提高。 作者建议首先建立奶粉信心智联网和饮料信心 智联网，并可以尝试建立城市级餐馆信心智联网。 支撑信心类简单风险在线服务的智联网，是一 种评价型智联网，其中的智能体是普通的消费者， 使用的是互联网，评价模型 m 和信息回溯功能决定 服务质量。 3.2 决策类简单风险在线服务 决策，是指管理主体为了实现某种目标而对未 来一定时期内有关活动的原则、方法、技术、途径 等拟定备选方案，并从各种备选方案中作出选择的 活动。备选方案是指供选择用的行动措施（原则、 方法、技术、途径等）的汇集。原则上，决策可通 过下述四步来实现。 第一步：目标拟定。根据管理主体需要解决的 问题，提出管理目标。例如，防灾减灾部门依据国 家的法律、法规，并根据可用资源、基本组织架构 等，确定工作方针、减灾目标、长远减灾规划等， 从而提出监测、预警、应急的目标等等。灾后 24 小 时内救灾物资到达灾民手中，就是民政部门应急管 理的一个具体目标。 第二步：资料分析。根据管理目标，通过各种 途径和渠道，收集管理系统内部的和外部的数据资 料和相关信息。并进行必要的整理和分析。显然， 收集到的资料和信息越多、越准确，通过分析对自 然状态的认识和未来情景的预测也就越接近客观实 际，所作出的管理决策也就越合理。所以，在管理 决策过程中，资料和信息收集是十分重要的。但需 要特别注意的是，由于时间紧迫和成本等问题，资 料和信息收集难以做到又多又准，应适可而止。当 然，资料和信息的收集，一方面要有目的、有针对 性地进行收集、整理、分析；另一方面，也要依靠 平时的积累和存储。 第三步：寻求备选。所谓制定备选方案，就是 以所要解决的问题为目标，对收集到的情报和信息 资料认真整理、分析和科学计算，并以此为依据制 定出几个实现目标的方案，提交管理决策者选定。 制定备选方案，也是一项比较复杂，要求较高的重 要工作，有时还需采用试验的方法，有的要采用数 学的方法，进行可靠性和可行性分析，提出每个方 案的利与弊，然后才能提供备选。例如，自然灾害 风险评价模型的选用，就是一个典型的寻求备选问 题。面对大千世界，风险评价模型举不胜举。所要 解决的问题不同，所能使用的数据资料不同，可供 published by atlantis press copyright: the authors 114 选择的风险模型种类也不同。如果是为保险公司计 算标的费率所用，概率类模型应为备选。 第四步： 优选项。这里所说的选定 优方 案，是在若干个备选方案中，选定一个 佳方案。 这是管理决策的 后阶段，也是关键的一环。管理 决策的成功与否，直接关系着管理涉及的企业或社 会系统的发展，关系到职工或民众的切身利益，甚 至往往决定着企业或社会的命运。 传统上，上述的决策工作都是管理主体组织专 家来进行。由于管理系统的复杂性，人们越来越难 以选出 优方案，更多的时候是选出满意方案。人 们在这方面做了很多有益的探索。其中，遗传算法 就能为寻找满意方案提供许许多多帮助。 这种途径只适于重大项目的决策， 大问题是 高成本。然而，人们在现实生活中常常遇到一些个 人的决策问题，不当的决策潜伏着巨大的风险。另 一方面，大量的人或许已经历类似的决策，或对其 有了相当的研究，可以用智联网来形成重要的经 验。 设 p 是某人待决策的问题，a 由 n 个经历和思 考过此问题的人组成，他们经由互联网 n 分别对 p 进行阐述，并根据自己的经历或思考提出对 p 的决 策建议。由于智联网可以提供足够大的信息兼容空 间，参与者可以用智联网平台提供的标准构件，规 范但较自由地表述对 p 的决策建议。智联网上的标 准构件和处理它们的算法，构成了模型 m。当（a, n, m）为一个正向智联网时，它能给需要决策的人 提供重要帮助。 “志愿无忧网”(http://www.51bzy.com/)，以专 家咨询的方式为千千万万的高考学子提供填报志愿 的咨询，很受考生家长们的欢迎。但是，这类网上 咨询服务，与传统咨询服务并无本质区别，只是当 事人见面的方式有所不同。这种网上咨询，无法提 高产品质量，但有利于扩大市场。 用智联网辅助决策，能集思广义，可望大幅提 高产品质量，极可能将专家咨询提升到一个高度。 当我们使用基于信息扩散的核心技术对参与者提供 的信息进行分析后，就可以得知不同备选方案对面 的风险，从而给使用者提供不同条件下采取何种决 策方案的建议。更进一步地，可以根据模糊风险分 析原理，列出其它参考方案。 作者建议首先建立高考志愿智联网和专业取向 智联网，并可以尝试为年青学者建立选择科研方向 的智联网。 支撑决策类简单风险在线服务的智联网，是一 种综合型智联网，其中的智能体是有一定背景的智 者，使用的是互联网。如何过滤误导信息，融合同 类有效信息，是模型 m 必须考虑的重大问题。 3.3 浪漫类简单风险在线服务 风险渗透于生活的方方面面。谈情说爱、旅游 观光等浪漫的事，时常也不尽如人意，一些严重的 后果甚至会使人痛苦一生。有效规避相关的风险， 锦上添花才能实现。 设 l 是某人拟全程经历的浪漫之旅，a 由 n 个 经历过浪漫之旅或有相当研究的人组成，他们经由 互联网 n，比对 l 与自己的经历或重要案例，从而 判断参与者是否适合此浪漫之旅，并提出建议。由 于智联网能溶百家之精华，显旅程之弯曲，无疑可 以帮助浪漫中人提前看清旅程。如果前面是大坑， 停止前进；如果略有阻力，则设法解决。由于浪漫 之旅的动态性，这类风险问题的在线服务，可能要 有一定的持续性。智联网（a, n, m）中的信息处理 模型 m，将更具挑战性。 我们以高校学生恋爱为例来说明模型 m 的基本 原理。首先是 l 的构建。它由两大部分组成：双方 自然属性和社会发展属性。前者由家庭背景（居住 地、社会地位和经济条件等）和个人特质（学习成 绩、性格和身体条件等）构成；后者由社会价值取 向的变化和双方今后可能的就业构成。接下来是经 历过浪漫之旅或有相当研究的人能剖析 l 和进行对 比并给出建议的空间 x。设 x1, x2, …, xn 是由 n 个智 能体在 t 时给出的信息，f(t, l, x1, x2, …, xn)是对其处 理的结果。由于两人相互了解决的加深，在下一个 时刻，l 会有所变化，一些信息会得到补充，一些 信息会得到进行更正，形成新的 l’。此时，当持 l’ 再次上智联网时，假定另有 n’个智能体在 t’时给出 信息 nxxx ′′′′ ,,, 21 ，则我们有 ),,,,,( 21 nxxxltf ′′′′′′ ， 模型 m 就是要处理一系列这样的结果，给当事人提 供非常负责的建议。 作者建议首先建立高校学生恋爱智联网和“新 加坡、马来西亚、泰国”旅游智联网，并可以尝试 建立国际旅游智联网。 支撑浪漫类简单风险在线服务的智联网，是一 种监测型智联网，其中的智能体是有过浪漫经历或 相当研究的人，使用的是互联网。根据形势变化提 供较为稳定的服务，是模型 m 必须考虑的问题。 4. 智联网中的关键技术 古人云：三个臭皮匠，顶一个诸葛亮。古人又 云：一个和尚挑水吃，两个和尚抬水吃，三个和尚 published by atlantis press copyright: the authors 115 没水吃。前者是说，集思广义能出大智慧；后者是 说，如果没有好的管理，人多了会坏事。 显然，使智联网呈正向，即，智联网（a, n, m）的智力水平大于其中任何一个智能体 ai 的智力 水平，是智联网技术能投入实用的关键。今天，智 能体遍布大江南北，长城内外，世界各地；另一方 面，让智能体 ai 上网已经没有任何技术问题。也就 是说，智联网（a, n, m）中的 a 和 n 已经没有什么 问题，关键在于 m。 传统的信息处理模型，大多假定其处理的数据 信息是基本可靠的，协调性较好的，关注的是信息 中的模式，例如各种函数关系。 智联网中的 m 则完全不同，面对的可能是恶意 假信息，还有客观存在的不协调性，其次才是有效 信息的处理，并且信息永远不完备。因此，智联网 中的关键技术，其实就是解决上述问题的技术。 恶意假信息识别技术 设 xi 是来自于智联网（a, n, m）中智能体 ai 的 信息，如何判定它是恶意假信息？ 简单的技术是 离差技术。设样本 x={x1, x2, …, xn}的中心点是 x0， 如果 xi 与 x0 的距离超过其它点与 x0 平均距离一定倍 数，则判定它是恶意假信息。 关键的问题还在于样本点的全面量化以及缺失项 的处理，涉及不完备信息，需用到信息扩散技术。 不协调信息识别技术 如果 x={x1, x2, …, xn}中有大量的不协调信息， 则是智能体间不同看法的正常反映。用模型自动找 出相对协调的子集，并形成新的样本供更高一个层 次的模型学习，是解决此问题的较好策略。扩散型 人工神经元网络[7]已经提供了这方面的技术。 不完备信息处理技术 当 x={x1, x2, …, xn}是可以直接使用的信息时， 由于智联网成本的控制，n 是一个较有限的整数， 相应的信息 x 源是不完备的。无论是用其估计概率 分布还是识别因素间的关系，均是小样本问题，须 用信息扩散技术处理。 以信息扩散原理为基础，人们已经建立了五个 实用的分析模型： （1）用信息分配方法计算软直方图，比传统直方图 的估计效率提高 28%； （2）自学习离散回归模型，可以直接从给定的样本 生成表达输入输出关系的模糊关系矩阵； （3）基于正态扩散函数的一种混合型人工神经元网 络，有效地解决矛盾样本的学习问题； （4）以历史灾情资料为依据的农业自然灾害风险评 估模型，计算以县市为区域的自然灾害风险水平； （5）计算模糊风险的内集—外集模型，不仅可以替 代专家依据给出的样本进行模糊概率估计，而且实 践证明，用这一模型计算出来的自然灾害模糊风险 能对减灾方案进行合理的筛选。 上述模型经过一定的改造就可用于智联网，只 不过是将自然灾害的数据改为来自智联网的数据。 尤其是自学习离散回归模型，非常适合于无人操作 的系统大量处理数据。 5. 人类将进入智联网时代 1712 年托马斯·纽科门制造了早期的蒸汽机。 从 1765 年到 1790 年，詹姆斯·瓦特运用科学理 论，使蒸汽机的效率提高到原来纽科门机的 3 倍 多， 终发明出了现代意义上的蒸汽机。1807 年罗 伯特·富尔顿第一个成功地用蒸汽机来驱动轮船。 由于蒸气机的出现，人类脱离了繁重的体力劳动， 进入了机械化时代，也称为蒸汽机时代或第一次工 业革命。 从 19 世纪 60-70 年代开始，出现了一系列电气 发明。1866 年德国工程师西门子制成发电机，1870 年比利时工程师格拉姆发明电动机，电力开始用于 带动机器，成为补充和取代蒸汽动力的新能源。电 力工业和电器制造业迅速发展起来，人类跨入了电 气化时代。电力的广泛应用、内燃机的发明和新交 通工具的发明应用，称为第二次工业革命。 1946 年由冯•诺依曼设计的世界上第一台电子管 电子计算机在美国诞生，其快速运算的能力解决了 过去人力所不能完成的大量计算工作。从此，人类 进入计算机时代。晶体管电子计算机的出现，使计 算机开始在工业、农业、商业、医学、军事、科研 等领域发挥巨大。微处理芯片的出现，使计算机深 入到人们的家庭和日常生活。 1974 年美国国防部国防前沿研究项目署 （arpa）的罗伯特·卡恩和斯坦福大学的温顿·瑟 夫开发了 tcp/ip 协议，定义了在电脑网络之间传送 信息的方法，arpa 网在内部投入使用。1990 年欧 洲核子研究组织的蒂姆·伯纳斯-李推出世界上第一 个网页浏览器和第一个网页服务器，推动了万维网 的产生，人类进入互联网时代。尽管 1995 年至 2000 年间大量投机的出现产生了网络经济泡沫， 终导致 nasdaq 和所有网络公司崩溃，但互联网的 发展速度并没有减慢。今天，便捷的互联网不仅催 生了电子商务、网络文化等诸多产业，而且产生了 许多像 myspace、facebook、youtube、维基百科这 published by atlantis press copyright: the authors 116 样的虚拟社区。随着互联网用户的增加，互联网在 现代经济生活中正发挥着日益重要的作用。截至 2010 年 6 月 30 日，中国的网民达 4.2 亿人，互联网 普及率达到 31.8％，继续超过世界平均水平[8]。 今天的互联网，仍以承载信息流为主，尚不具 备本质上的智能。然而，一旦终端上的智能体之智 慧能借助互联网联接起来并被放大，智联网时代就 将来临。如果说 tcp/ip 协议使互联网成为可能，则 解决终端智能协议远不能成就智联网，更关键的是 处理相关信息的模型 m。不同类型的智能问题，相 应的模型应该不同。不远的将来，云计算和物联网 中的模型也有智能化的可能，将它们溶入智联网， 将加速智联网时代的到来。届时，人类个体有限的 智慧有可能被溶合后产生高级智慧，甚至大量出现 不亚于爱因斯坦的超级智慧。果真如此的话，人类 的生活会更加美好，人类的未来会更加光明。 6. 结论与讨论 将智能体用计算机联接起来的网络称为智联 网。智联网将是继人工智能和计算智能之后的智能 科学大革命。 智联网的发展动力是有偿智能服务。与传统咨 询服务业 大的不同是，提供智联网服务的不再限 于专家，而主要是大量有经验的民众。 智联网（a, n, m）正向否，取决于处理 n 个智 能体所提供信息的模型 m。对于提供风险评价在线 服务的智联网，信息扩散理论和方法成为 m 的核心 技术，以解决信息不完备和自学习的问题。 本文首次提出了智联网的概念，并针对风险分 析在线服务展开了研究，但仍处于理论探索阶段， 尚未建成一个运行的智联网。作者建议以 简单、 明了的情景展开实验，形成简单的 m 范式。 一个 简单的实验如下：实验主持人邀请 a, b, c, d 四人参加实验，选定一条正在施工的道路作为 实验对象。 主持人让 a 白天对施工道路的路况进行详细考 查，并写出报告备用。主持人让 b, c, d 三人夜晚通 过施工的道路， 先通过者给予奖励。随后，主持 人让这三人描述道路路况，描述得 详细的给予奖 励。并不知道路况的主持人综合 b, c, d 三人提供的 信息，形成报告。 后，主持人研究 a 提供的报告 和自己综合的报告，从而解决以下两大问题：1. 让 b, c, d 写什么形成的报告才更于自己综合出较完整 的报告；2. 仅仅根据三人提供的信息，如何让自己 的综合报告与 a 提供的报告差异 小。 显然，第一个问题的解决，有利于设计智联网 终端的交互界面；第二个问题的解决，则有利于形 成 m 的内核。 人工智能经历了从形式化到专家系统再到计算 智能的漫长发展过程，效果并不显著，目前仍是步 履维艰。无论知识工程也好，人工神经元网络也 罢，不仅无法生产出超过普通人的实用智能系统， 大多数系统甚至连少儿的智能都不如。智联网的出 现，有望推动人工智能获得长足发展。 风险分析是展示人类智慧 具挑战性的问题之 一，智联网集思广义的能力，有望使人们在日常生 活中从未知的风险世界进入已经的风险世界，从而 在别人的帮助下可以自由地选择是去担当风险还是 规避风险。担当风险的背后可能有较大收益，规避 风险则降低不必要的伤害。 参考文献 [1] 刘锋，彭赓，刘颖. 从人脑的结构机理看互联网的进化. 人类工效学, 2009，15(1): 11-14. [2] 苏锋.“沃森”的胜利. 微电脑世界, 2011,(3): 2. [3] huang c. f. and ruan d. fuzzy risks and an updating algorithm with new observations. risk analysis, 2008, 28(3): 681-694 [4] 赵新培. 问题奶粉致婴儿泌尿系统结石，全国发现 29.4 万名. 北京青年报, 2008 年 12 月 2 日. [5] 汪挺. 食用油市场利润遭受严重挤压，正规军染指地沟 油生产源于暴利. 中国商报, 2011 年 9 月 20 日. [6] huang c. f. and shi y. towards efficient fuzzy information processing --using the principle of information diffusion. heidelberg: physica-verlag (springer), 2002. [7] huang, c.f. and leung, y. estimating the relationship between isoseismal area and earthquake magnitude by hybrid fuzzy-neural-network method. fuzzy sets and systems, 1999, 107(2): 131-146. [8] 祝华新，单学刚，胡江春. 2010 年中国互联网舆情分析 报告. 人民网, [2011-01-23] http://www.people.com.cn/ gb/209043/210110/13740882.html published by atlantis press copyright: the authors 117 基于行走时间最短的机场停机位分配问题的一种混合算法：遗传算法加涟漪扩散模型 a research on community risk radar with presetting forms to structure information in internet of intelligence yifang leng1,2, chongfu huang1,2,3,* 1state key laboratory of earth surface processes and resources ecology (beijing normal university) beijing 100875, china 2academy of disaster reduction and emergency management, ministry of civil affairs & ministry of education the peoples' republic of china, beijing normal university, beijing 100875, china 3. beijing cazl technology service co., ltd., beijing 100088, china abstract in order to improve the community risk radar in the information gathering stage, the problem of unstructured information processing may be able to result by taking advantages of the internet of intelligence. according to the characteristics of the community risk events, setting the data form of the collecting page in advance can help reduce the proportion of unstructured information. adding the image upload and download function can help residents building the risk scenarios. in the risk event assessment stage, coding the php web page can achieve the online processing of risk data. in the stage of risk event display, the risk is displayed by the risk radar and risk map. finally, a community in beijing is taken as an example for specific description. keywords: community, risk radar, internet of intelligences, structuring information 智联网驱动的社区风险雷达中预设表单式信息结构化研究 冷一芳 1,2 ，黄崇福 1,2,3,* 1. 北京师范大学地表过程与资源生态国家重点实验室，北京 100875 2. 北京师范大学，民政部/教育部减灾与应急管理研究院，北京 100875 3. 北京崇安智联科技服务有限责任公司，北京 100088，中国 摘 要：针对现有的风险雷达中非结构化信息的处理问题，本文提出了基于智联网特点的信息搜集方式， 根据社区风险事件的特点，通过预设搜集页面表单的方式减少非结构化信息所占比例。在风险事件评估阶段， 通过 php 网页编码实现风险数据的在线处理；在风险事件展示阶段，以风险雷达图辅以风险地图的形式进行 事件展示。此外还增加了风险事件总结阶段，由管理员对事件的处理状态和处理措施进行更新，从而使得风 险雷达的数据库成为社区的经验库，并提供了查询功能，完成了风险事件排查、匹配以及处理三个环节的统 一。最后，本文以北京海淀区的某社区的试验案例进行了具体说明。 关键词：社区，风险雷达，智联网，信息结构化 * corresponding author: e-mail: hchongfu@bnu.edu.cn. post address: academy of disaster reduction and emergency management, beijing normal university, no.19 xinjiekouwai street, beijing 100875, china this project was supported by the national natural science foundation of china (41671502), and partly supported by the beijing cazl technology service co., ltd.. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 156 received 22 june 2017 accepted 3 august 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 1. 引言 现代社会的基本单位是社区，从社区心理学的 角度来说，社区的成员互相之间有许多的关联和共 同认识 [1] 。社区构成了城市，也是城市管理的基本单 位。二十世纪以来，社区减灾一直都是国内外采用 的重要减灾战略。1999年，在“国际防灾战略” （isdr）中，联合国已经提出为建设21 世纪更安 全的世界，需“从强调政府的作用到重视推进建设 灾害应对能力强的社区” [2] ，国内“十一五”规划中 有关减灾的重要内容是创建综合减灾示范社区作 [3] 。 无论在哪，社区灾害风险管理都是减灾和保障可持 续发展的有效手段 [4] 。 在社区，社区居民从多年的生活中获得了丰富 的应对灾害风险的经验，此外，拥有智能可以进行 深入思考的天赋也使得居民能够根据经验感知未来 的风险，居民通过公共参与的形式加入到社区减灾， 将大大提高社区减灾效率，有利于公众在防灾减灾 中发挥自己独特的作用 [5] 。 风险雷达是能够发现、识别、监测和管理动态 风险的系统，将风险雷达的理念和社区减灾结合在 一起，就形成了社区风险雷达 [6] 。社区风险雷达收集 和监测社区内的潜在风险，它以智联网为驱动，借 助雷达图的形式动态地展示社区风险事件的变化， 从而为社区管理提供风险监测与预警的功能。在“人 人都是传感器”的互联网时代，人机结合的智联网， 正在推动风险分析进入互联网的大数据时代。但是， 目前的社区风险雷达在交叉验证和综合提取，实现 向结构化信息的转化环节的数据处理仍然依靠人工 操作，尚无法应对大面积使用带来的大量数据，不 能达到连续运行的目标。 2. 相关研究现状 2. 1 智联网简介 智联网（internet of intelligences）是由各种智能 体，通过互联网形成的，这一概念最早由黄崇福教 授提出，目的通过网络集小智慧为大智慧，群策群 力 [7] 。在数学模型的帮助下，人们更好地认识世界， 获得更好的生活质量。智联网有三大关键词，智能 体、网络和模型，实际上，智联网是由智能体、网 络、模型组合而成的，如图1所示。当智联网的智力 水平大于智能体中最高智力水平时，这是我们期待 的具有集智功能的智联网。 c1 c2 c3 a1 a2 a3 s(network server) m(information processing model) 图 1：智联网示意图 2.2 智联网应用现状 目前已有若干智联网平台成功开发并投入使 用，“高考志愿填报智联网服务平台”是首个研制 成功的智联网演示系统，可为考生提供有关报考志 愿的参考经验和志愿填报风险评估等功能 [8] ，“温州 台风灾害风险分析智联网”是首个投入实用的智联 网 [9] ，此外，基于智联网的灾害风险评估还在城市内 涝 [10] 、地震群测群防 [11] 、社区风险 [6] 等领域开展 [12] 。 2.3 风险雷达研究现状 十九世纪，科学家在研究蝙蝠夜间飞行原因的 过程中发明了雷达。二十一世纪，人们开始关注风 险，能否用雷达来监测风险？普通的雷达当然不行， 但是在雷达原理的启发下，人们提出了风险雷达的 概念，能够发现、识别、监测和管理动态风险的系 统，被称为风险雷达。 在国内，风险雷达的概念主要运用在企业风 险监测、化工装备监测、舆情监测和社区应急监测 中。在商业银行的风险监测中，“银行风险雷达图” 主要是一种形象化的表达工具，是将银行的财务状 况、经营成果等信息用各类比率的形式在雷达图上 显示出来，以便监管人员了解银行运作的风险状况 [13] ；在化工装备监测中，通过筛选装备运行的实时 状态数据识别故障，获得故障发生的原因和相应补 救措施 [14] ；舆情监测是互联网对境内外的中文舆情 载体（包括新闻网页、论坛、博客、贴吧等）的网 络资源进行采集，挖掘分析，把握网民的舆论动态、 了解隐患、分析趋势，并实现话题监控和各种统计 功能 [15] 。 我国是第一个尝试使用智联网驱动技术将风险 雷达应用于社区，用于提高社区应急管理能力的国 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 157 家。与传统商用风险雷达不同的是，智联网驱动的 风险雷达是一个在线系统，而不再是传统风险矩阵 的一种雷达化显示 [16] ；与jovanovic式欧洲风险雷达 不同的是，智联网驱动的风险雷达是一个人机交互 系统；新兴的风险雷达舆情监测系统不同的是，智 联网驱动的风险雷达不仅仅是对网页海量信息进行 归纳总结的系统。智联网驱动的风险雷达，具有更 强的针对性，提供的风险信息更精细 [17] 。 然而，目前的风险雷达，在信息采集页面的设 计过于简单，导致复式智联网处理社区居民提供的 信息时，在交叉验证和综合提取，实现非结构化信 息向结构化信息的转化环节中，存在问题。由于智 联网获得的风险信息大多是柔性信息，也是非结构 化信息，是人类感知风险的自然语言表述，不能简 单地规范化和数量化，如自然灾害中的“洪水齐腰 深”、“瓢泼大雨”等，这类信息在经验信息中占 有较大比例，而大多数分析模型能够处理的数据是 以规范的和量化的形式表示的 [8] ，因而社区风险雷达 的运行还离不开人工的处理环节。 2.4 社区风险信息结构化研究 随着计算机网络技术的快速发展，信息以爆炸 式的速度飞速增长，数据形式变得更加多样化，信 息的种类也在不断的扩展，其中包括了大量的非结 构化和半结构化数据 [18] 。结构化数据是可以存储在 数据库里，用二维表的逻辑结构来表达的数据。数 据结构不固定，无法使用关系数据库存储，只能够 以各种类型的文件形式存放的文档、文本文件、图 片、图像、音频和视频等称为非结构化数据 [19] 。介 于两者之间的，具有结构但结构变化较大，同时数 据的结构和内容混在一起无法简单区分的数据称为 半结构化数据，代表性数据为xml文档[20]。 非结构化数据存在存储与管理方面的问题，而 解决这一问题目前通常的思路是通过“非结构化数 据—半结构化数据—结构化数据”的方式实现数据 的转换，从而进行统一有效的管理 [18] 。 但目前的研究大多只是针对于某一类型的文件 展开的，而目前研制的社区风险雷达 [6] 在社区运行过 程中搜集的数据类型包括数字、图片、描述性的大 段文字等，当中的非结构化信息必须在人工进行处 理之后才能进入模型进行计算。当风险雷达在更多 社区中运用时，接收到的大量风险信息显然不能再 依靠人工处理，因此，如何将非结构化数据转为结 构化数据，也是风险雷达面临的几大挑战之一。 3. 社区风险信息结构化及社区风险事件 3.1 基于智联网的社区风险信息结构化 智联网获取的信息多为柔性信息，来源广泛， 包括个体的亲身经历、听闻等。形式上包括零散的 文字、数字，或者是具有逻辑的简短语句，事件的 现场照片等，是结构化数据和非结构化数据组成的 信息集。其中，为全面描述风险事件，事件描述信 息中包含大段文字的详情描述，属于非结构化信息， 其中的信息提取、分类、真伪判断等工作普通模型 并不能胜任，虽然也可利用复式智联网中人脑的智 能作用参与对这类非结构化经验信息进行处理与分 析 [21] ，但效率较慢，也不方便连接到模型进行运算。 智联网中的智能体具有柔性感知功能 [8] ，可以分 辨有关风险事件的干扰信息和片面信息，分析提取 出其中的有效信息，做出自己的判断，其原理图如 图2所示。例如，在面对各类纠纷时，当事双方往往 各执一词，他们的陈述信息干扰信息和片面信息较 多，而经验老道者往往可以通过二者的描述还原事 件的真实过程，这一结论的得出即是柔性感知过程。 利用智联网的柔性感知功能，我们也可以提取出风 险事件的有效信息。 图 2：柔性感知示意图 为解决风险雷达中非结构化信息转化问题，在 对社区风险事件研究的基础上，通过预设信息搜集 页面的表单，在信息搜集初期，尽量将更多信息从 非结构化的“事件描述”字段中剥离，通过智能体 的柔性感知过程转化为可以直接利用的信息。换言 之，复杂的详情描述文字信息部分转化为简单的选 择填空形式，以此搜集各类风险问题所需的数据， 剩余部分则依旧依靠复式智能网的人工智能，原理 图见图3。在对社区风险事件深入了解的基础上，如 何通过问题设置来引导和激发智能体的智慧，就成 为智联网表单设计的一大目标。 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 158 3.2 社区风险事件 风险是与某种不利事件有关的一种未来情景 [22] 。社区风险事件，是发生在社区内部的事件，它 可能会导致社区在未来发生某些不利事件。 社区人口密集，基础设施不足和设备失修老化 等问题往往带来较大后果，具有突发性、威胁性和 不确定性，透明性和社会性等特点 [23] 。同时，社区 内发生的风险事件也种类繁多，根据文献阅读和前 期在若干社区的调研情况，依据社区风险事件的不 同属性进行分类，将有利于风险分析阶段模型对信 息的处理。因此，根据事件的类型，事件的发展速 度，事件的处理方式，社区风险事件将分为若干类， 按事件类型分类的结果见表1。 表 1：社区风险事件分类表 编 号 事件 类型 具体 分类 1 自然 灾害 暴雨 积水 大风 雷电 冰雹 雪灾 高温 热害 其 他 2 火情 隐患 危化 品等 消防 设备 燃气 泄漏 触电 危险 其 他 3 治安 交通 交通 堵塞 可疑 人员 盗窃 诈骗 暴力 事件 其 他 4 设备 隐患 路面 损坏 电力 设备 通讯 设备 其他 设备 其 他 5 公共 卫生 水质 异常 垃圾 处理 食品 卫生 疾病 传播 其 他 6 居民 活动 老年 救助 居民 纠纷 装修 干扰 谣言 传播 其 他 4. 社区风险雷达设计与比较 4.1 社区风险雷达设计 根据社区风险雷达的工作原理，以及社区风险 事件的特点，风险雷达由以下几个层面组成：信息 提交层，包括事件提交页面和事件总结页面；风险 分析层，包括事件评价页面和风险评估页面；反馈 展示层主要包括事件展示页面，下面将具体介绍各 页面的设计。 4.1.1 事件提交页面的表单 表 2：事件提交页面表单属性设置 名称 数据类型 备注 pid int(10) 事件编号 name char(20) 事件类型 description varchar(200) 事件描述 lnglat char(20) 位置坐标 location_dis char(20) 位置描述 start_time char(20) 发生时间 emerg char(20) 紧急程度 price char(20) 危险度 sehnfen char(20) 身份 image varchar(40) 照片上传 image_dis text(200) 照片详情 feature2 text(20) 事件状态 feature3 text(20) 消息来源 advise text(20) 建议措施 supply2 create_at text(20) timestamp(10) 补充 提交时间 在事件提交页面，用户需要通过文字、图片等 信息来构建风险情景，尽可能多地提供风险事件相 关信息，社区其他用户才能在此基础上对风险事件 进行风险评价。为了尽可能多地搜集到风险事件的 相关信息，同时尽量将更多信息从非结构化的“事 件描述”字段中剥离，成为可以直接利用的信息， 部分信息以选择或填空形式展现，以此搜集各类风 险问题所需的数据，提交页面表单设置见表2。 通过设置不同的属性信息，使得用户头脑中关 于风险事件的抽象记忆能最大程度的表述出来，也 使得用户提交的信息能够便于理解和处理。其中， 图 3： 预设信息搜集页面表单原理图 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 159 事件类型以选择形式出现，是基于社区风险事件分 类表，由于总结性较强，往往需要辅以简单的事件 描述。位置坐标通过导入百度地图api来获取经纬度 坐标，有利于对事件进行精确定位，以及风险地图 模块的表达，但脱离地理信息系统后的经纬度坐标 如在一般人的头脑中并不能映射到某一地点，因此， 还需添加位置描述用以补充。事件所在地的照片往 往包含许多柔性信息，难以用语言描述全面，因此， 采用照片上传和图片描述简单介绍结合的方式使得 信息既全面又概括。 危害度和紧急程度两项指标用来表征普通居民 眼中的风险事件严重程度。此外，发生时间、事件 状态、消息来源都是风险事件某一属性的具体挖掘： 事件状态包括解决中和已解决，针对的是社区内的 突发性事件和渐发性事件，突发性事件很有可能已 经得到相应处理，此外，在事件得到具体解决后， 事件状态也将在总结页面被管理员修改为已解决， 从而进入社区经验库；消息来源包括了眼见和听闻 两种，有利于风险评价阶段的真伪判定。最后，补 充部分是为了弥补人为分类导致的柔性信息丢失， 对于表单中的未提事宜，用户可提交到此属性栏中。 通过信息采集和表单设定，尽可能多的结构化 信息与非结构化信息一起组成了用以描述风险事件 的信息集，后期的结构化信息的风险量化部分工作 量大大减少，也使得其他社区居民可以更轻松地构 建风险事件的风险情景，并结合自身经验对其进行 评价。 4.1.2 事件评价页面的表单 往往，离风险事件的发生地越远，人们对事件 的了解程度越低，可靠性也不足。因此，事件评价 页面的用户是社区内的人员，需要提交的内容包括 可能性、危险度、紧急程度、建议解决方向、类似 事件描述、其他补充和身份信息，页面表单的设置 见表3。此阶段第一次运用了复式智联网的交叉提取 原理，用户不仅仅提供信息，还参与了信息的处理。 具体来说，其他用户在作为智能体提供的经验信息 进行评价的同时，也在利用人脑知识库及人脑的智 能作用对事件的初始信息进行处理与分析，这是成 为了风险雷达风险分析模型的组成之一。 在风险事件的评价中，参考风险矩阵图的内涵， 用户将根据自身经验，对事件的可能性、紧急度和 危险度三项指标进行打分，采用相对值0-10的表示 方法，用以表征风险事件的风险值。为了尽可能的 获取用户有关风险事件的个人经验，设置了建议解 决方向和类似事件描述两项属性。建议解决方向的 分类基于社区事件的处理部门，引导个体提供相关 经验，加速事件的解决进程。类似事件描述是针对 社区的某些共通问题设立的，如社区内有人举报有 可疑人员在某居民楼前停留并做标记，用户通过智 联网回答该问题，并扩大和补充可疑人员在社区的 活动范围，使得社区类似问题可以尽早暴露，提高 事件的处理效率。在事件评价的过程中，参与其中 的用户不仅能看见事件提交的详细信息，还能参考 其他用户对本事件做出的评价，分辨其中的干扰信 息和片面信息，分析提取出其中的有效信息，并与 自己的个体经验相融合，从而完成评价过程，这一 过程是柔性感知的过程 [8] 。 在复式智联网中，智能体不仅仅提供自身感知 的风险信息，还对其它智能体提供的风险信息进行 柔性感知与捕获，设参与事件提交的用户集为 {a,b,c,d}，参与事件评价的用户集分布为{a1，a2， a3……}，{b1，b2，b3……}，{c1，c2，c3……}， {d1，d2，d3……}，将两类用户集的用户之间进行 比对，将会发现用户a参与了其他事件的评论，这种 情况即为交叉提取。 4.1.3 风险评估页面的原理 在事件评价阶段，复式智联网的参与用户已经 利用人脑的智能作用参与对事件信息进行初步处理 与分析，提交的位置信息、图片信息、消息来源等 为事件评价阶段用户的二次评估和交叉提取提供了 风险事件的具体信息，在此基础上，获取的三项指 标值在经过虚假信息排查后可进入数学模型进行运 算，通过对事件可能性进行阈值设定，在排除掉某 些虚假信息后，采用包括信息扩散在内的数学模型 进行风险评估，其中，风险信息扩散技术是用来弥 补小样本的不完备缺憾，将样本转化为模糊样本， 表 3：事件评价页面表单属性设置 名称 数据类型及宽度 备注 id pid int(10) int(10) 评价编号 事件编号 view_trust float(10) 可能性 view_risk float(10) 危险度 view_emerg float(10) 紧急程度 viewer_shenfen char(20) 身份 viewer_desc text(200) 其他补充 viewer_advise text(200) 建议解决方向 score char(20) 类似事件描述 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 160 通过php编程语言实现了信息扩散技术的模糊样本 优化算法，扩散方法定义如下： 令x为区间[0，1]的随机样本，使用正态扩散函 数将xi变为模糊集[7]，利用公式（1-4）完成信息扩 散计算。 (1) (2) (3) (4) 在对三项指标进行信息扩散计算后，风险值也 可以计算出来。考虑到风险是在特点时间和空间范 围内的潜在损失大小及发生的可能性，将以风险源 的发生概率和承灾体的暴露性和脆弱性表征风险 [24], 可能性指标代表了风险源的发生概率，考虑到社区 风险事件的突发性和威胁性，将危险度和紧急程度 两项指标代表承灾体的暴露性和脆弱性。用p代表可 能性，e代表暴露性和脆弱性，风险评估值r的模型 可形式化表示为式（5）： r=p○e (5) 其中，“○”代表了耦合规则，可以是包括乘 法运算等在内的数学公式 [25] 。考虑到风险与特定时 间相关，具有动态性，随着时间的推移，p、e的变 化将导致风险表征值r也发生变化，因此，评估过程 是不断进行的。 4.1.4 事件展示页面的设计 风险时时刻刻都在发生变化，随着时间的推移， 不断有智能体加入到风险评估中，提供自己的经验 信息，此时发生变化的不仅仅是风险评估的结果， 还包括可能风险事件本身。当事件得到及时解决， 事件状态从待处理变为已解决时，应将该事件从待 评价列表中撤下，使得居民的注意力转移到新的风 险事件上。因此，为了更好的反映事件的变化，评 估结果是实时更新并展示出来的。展示页面包括首 页的风险雷达图和风险地图两个页面。 风险雷达图即图4，社区风险雷达以智联网为驱 动，获得不断更新的风险评估结果，并在，风险雷 达图中表现出来。然而社区风险事件多发，过多的 风险事件展示在风险雷达图上，不仅可能导致社区 内发生恐慌，还可能充塞雷达图，分散居民的注意 力，导致居民对风险事件关注度下降，忽视某些风 险事件。因此，首页风险雷达图中风险事件的显示 仍然离不开系统管理员的审核。 图 4 ：风险雷达图 风险雷达图是基于事件分类进行展示，除此之 外，还有基于地理信息的风险地图展示页面。在风 险地图中，各风险事件依据经纬度坐标展示在百度 地图插件中，以圆形表示，圆形的颜色代表风险事 件的风险值大小，见图10。在长期使用后，数据经 过整理，风险地图可以升级为社区灾害风险地图。 4.1.5 事件展示页面的设计 此阶段第二次运用了复式智联网的原理，主要 用户为社区风险雷达管理者，可以是居委会工作人 员，业主委员会成员，或是经社区认证的热心居民。 在参考事件评价阶段获取的若干个体经验后，事件 得到相应处理，为了促进社区风险雷达数据库建设 成为社区的经验库，用户需要在事件总结页面进行 总结，页面表单的设置见表4。 表4：事件总结页面表单属性设置 名称 数据类型及宽度 备注 pid feature2 s1 s1 s3 int(10) text(20) text(20) text(200) text(200) 事件编号 事件状态 实际解决措施 措施详情 如何规避风险 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 161 总结页面主要包括四项内容：事件状态、解决 措施分类、措施详情、如何规避风险。事件状态需 从待处理改为已解决或解决中，从而使得事件从待 评价列表以及风险雷达图上撤下。公布事件的解决 措施分类以及措施详情将为社区提供更为丰富的应 对经验，减少未来遇到同样问题的应对时间。最后， 利用人脑的智能作用对事件的相关信息进行柔性感 知的处理与分析，参考其他智能体的经验，总结出 规避风险的办法，避免风险事件在未来产生不利后 果。 4.2 与传统社区风险雷达的比较 图 5 为社区风险雷达改进前后的工作流程 对比，与传统社区风险雷达 [6] 相比，改进后的风险雷 达在工作流程上主要区别包括：风险事件提交、评 价页面的改变，由于预设表单页面的设计，新经验 信息集中获取的非结构化信息较少，大部分数据可 以进入模型中直接处理；模型处理和评估阶段，通 过 php 网页编码实现风险数据的在线处理，可以得 到实时的评估结果；反馈展示阶段增加了风险地图 的表达；事件总结页面的设置，增加了社区风险事 件的处理终点，社区经验库的开放使得个体经验成 为社区的共同经验。 5. 案例背景与分析 5.1 案例背景 北京的住宅小区中绝大多数为老旧小区，由多 层住宅楼和平房胡同组成，社区内部问题较多，社 区管理难度也较大。选择老旧小区为实验社区，可 以更快地提高社区风险雷达的实际应用能力。在社 （a）原社区风险雷达工作流程图 (b) 改进的社区风险雷达工作流程图 图 5：社区风险雷达改进前后的工作流程对比. journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 162 区居委会的宣传和帮助下，社区居民在社区风险雷 达平台提交和评论社区风险事件，通过交叉提取和 柔性感知进行风险评估，并实时发布至平台，管理 员结合居民的评估意见和实际情况处理风险事件， 并对风险事件进行总结，完善社区经验库。 实验社区 d 位于北京市海淀区东南部，处于北 下关街道辖区。北下关街道面积约 60.4 万平方米， 辖社区 31 个。实验社区占地 6.6 万平方米，共有楼 房 19 栋，平房院落 5 个，是一个楼院、平房共存的 小区。常住人口户数 1901 户 [26]。社区地处城乡结 合部，区内有大型央企，冷冻食品厂，学校，家乐 福超市，综合性医院，酒店宾馆等，居民生活区域 和生产区域并未分离。由于社区建立较早，是老旧 社区，社区硬件比较差。除治安问题外，社区内平 房区安全用火、用气、用电安全是社区管理工作的 重心。社区所在街道已建立有微信公众号平台，“随 手拍文明”等活动已开展近半年 [27] ，社区居民对于 使用网络平台来反映社区公务事务有较高积极性。 在居委会的宣传和帮助下，2017 年 3 月 28 日 至 4 月 1 日期间社区风险雷达在 d 社区进行了试用。 社区居民可以通过电脑或手机登陆社区风险雷达平 台，提交他们在社区内发现的风险事件。期间共收 集到风险事件 10 起，收集到事件评论 30 多份。其 中，关注度最高的是设备隐患中的电线树枝纠缠事 件，下面将以此社区风险事件进行具体说明。 5.2 案例分析 电线树枝纠缠事件发生在 d 社区丙四号院外， 事件类型属于设备隐患。院内大树的枝干伸出院门， 与院外道路一侧的电力线纠缠在一起，树枝与电线 之间的摩擦将加速电线外壳的磨损速度，在强风、 暴雨、冰雹等灾害性天气出现时，树枝的存在对电 力线的安全威胁更加严重，容易给过往行人和车辆 带来灾难性后果。 本起风险事件中提交者提交信息见图 6，并上传 了图片加以佐证，点击照片上传右侧链接，即可下 载图片进行查看，风险事件图片见图 7。 共有 8 位居民参与了事件评价，通过在社区风 险雷达评价管理页面输入事件编号，可查询本次风 险事件的全部评价，见图 8。 风险事件经过评价阶段的柔性感知和交叉提 取，在风险评估阶段不断更新。随着参与人数的增 加，数据样本不断扩大，风险值也越来越接近真实 值，具体地，根据定义的风险计算公式得到的风险 表征值变化如表 5。 本次社区风险事件经管理员审核后，出现在首 页风险雷达图上，随着评论人数的增加，雷达图中 图标的位置也发生相应变化，表征着风险事件的风 险值变化，见图 9。 风险事件发生前后，风险地图的变化如图所示， 图 10 为发生后，发生地用圆圈圈出。 图 6 ：风险事件提交信息 图 7 ：风险事件现场图 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 163 图 8：风险事件评价查询结果 表 5 ：风险表征值变化 参与人数 危险度 可能性 紧急程度 风险值 1 8 7 3 6.25 2 8 7.5 5 7 3 6.67 6 4.33 5.75 4 7 6.75 4.5 6.25 5 6.70 6.24 44.91 6.02 6 6.97 6.54 5.38 6.36 7 6.92 6.78 5.88 6.59 8 7.07 6.99 5.88 6.73 图 9：风险雷达图变化 图 10 ：某风险事件发生后 由于树枝修剪工作专业性较强，且可能存在触 电风险，社区居民无法自行处理，社区内也无物业 管理公司负责此项事务。在核实情况后，管理员将 事件上报至本区园林部门，请绿化队工作人员携带 专业工具完成修剪工作，并在总结页面提交事件处 理方案，并将事件状态改为“处理中”，该事件处理 进度可在列表中查询到，见图 11，输入事件编号“59” 即可。修剪工作完成后，管理员可将事件状态修改 为“已解决”。 图 11：事件总结信息查询 本次风险事件过程中搜集到的经验信息也已进 入数据库，可以在“查看评价”、“查看总结”模块 查询到具体信息，即历史风险事件匹配过程，在“事 件描述”下方的文本框内输入“电线”，也可以查询 到本起事件。通过事件的积累丰富社区的风险事件 应对经验库，将避免重复工作，大大提高工作效率。 6. 结论 本文以智联网驱动的社区风险雷达为研究对 象，通过尝试深入了解社区风险事件以及社区居民 的风险意识和表述习惯，研制一系列柔性感知过程 journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 164 中使用的表单，实现数据的在线处理功能，形成自 动运行的复式智联网，同时利用智联网数据库实现 了社区经验库的功能，持续驱动社区风险雷达，完 成社区风险雷达设计中隐患排查、历史事件匹配以 及事件处理的目标。针对现有的风险雷达中非结构 化信息的处理问题，提出基于智联网特点的信息结 构化搜集方式，针对社区风险事件的特点，通过预 设搜集页面表单的方式减少非结构化信息所占比 例，并具体介绍了社区风险雷达中各个环节的设计。 通过实验社区的案例，展示了社区风险雷达的试用 情况、评估过程及评估结果等，验证了平台的实用 性。实验过程中，社区居民通过社区风险雷达反映 问题的愿望较强烈，社区工作人员则希望能够通过 社区风险雷达改善社区环境，减轻工作量。 参考文献 [1] 刘视湘. 社区心理学. 开明出版社,2013:60-60. [2] 陈徐东. 社区防灾减灾保障机制的创新. 城市与减 灾,2010(6):5-7. [3] 卫敏丽. 解读《国家综合减灾“十一五”规划》 . disaster reduction in china, 2007,(9):19-22. [4] 陈容,崔鹏. 社区灾害风险管理现状与展望. 灾害 学,2013,28(1):133-137. [5] 冷一芳,黄崇福. 社区应急管理风险雷达中预设表单 式信息结构化研究初探. 大数据时代的风险分析和 危机反应----中国灾害防御协会风险分析专业委员会 第七届年会. 2016,11:890-895. [6] 吴彤. 智联网驱动的在线风险雷达在在社区应急管 理中的应用研究. 北京:北京师范大学,2015. [7] 黄崇福. 自然灾害风险分析与管理. 科学出版社, 2012. [8] 艾福利,黄崇福,王蔚丹. 高考志愿填报风险评估智联 网服务平台. 风险分析和危机反应的创新理论和方 法——中国灾害防御协会风险分析专业委员会第五 届年会论文集,2012. [9] 黄崇福. 自然灾害风险评价:理论与实践. 北京:科学 出版社,2005. [10] 郭君,艾福利,蒋卫国,等. 自然灾害风险分析智联网 中的柔性地理信息及其应用. 风险分析和危机反应 中的信息技术--中国灾害防御协会风险分析专业委 员会年会.2014. [11] 王蔚丹,黄崇福,曾凡雷. 地震群测群防智联网中双层 虚假问题的研究. 风险分析和危机反应中的信息技 术--中国灾害防御协会风险分析专业委员会年会. 2014. [12] 苏妩. 信息收集去中心化在风险评估中的应用 ---海 洋环境风险管理智联网平台研究. 北京:北京师范大 学,2016. [13] 周晓阳. 改进的雷达图分析法在我国商业银行风险 监 测 中 的 应 用 . 今 日 财 富 : 金 融 发 展 与 监 管,2010(6):59-64. [14] 中国专利发布公告. [15] http://epub.sipo.gov.cn/pam.action [16] 苏博. 太原供电公司网络舆情监测系统的设计和实 现. 华北电力大学,2014. [17] 黄崇福. 自然灾害动态风险分析基本原理的探讨,灾 害学,vol.30, no.2, (2015), 1-7, [18] 吴彤,艾福利,郑建军,黄崇福. 智联网驱动的风险雷 达实现探讨[a]. 风险分析和危机反应中的信息技术 --中国灾害防御协会风险分析专业委员会第六届年 会论文集.中国灾害防御协会风险分析专业委员 会,2014,6:468-473. [19] 万里鹏. 非结构化到结构化数据转换的研究与实现. 西南交通大学,2013. [20] 徐宗本,张讲社. 基于认知的非结构化信息处理:现状 与趋势. 中国基础科学,2007,(6):53-55. [21] 许学标,顾宁. 半结构化数据模型及查询语言. 计算 机研究与发展, 1998,(10):896-901. [22] 艾福利. 自然灾害风险分析智联网服务平台构建与 应用研究. 北京:北京师范大学,2013. [23] 黄崇福. 自然灾害风险评价理论与实践. 科学出版 社, 2006. [24] 乔苹. 我国城市公共危机治理中的公民参与机制研 究. 首都经济贸易大学, 2013. [25] 马保成. 自然灾害风险定义及其表征方法. 灾害 学,2015,(3):16-20. [26] 黄崇福,郭君,艾福利,等. 洪涝灾害风险分析的基本 范式及其应用. 自然灾害学报,2013,(4):13-25. [27] 社区简介. http://mp.weixin.qq.com/s/oya07yvr_8oco6y0o_fi 3w [28] “你来拍,我来奖”活动网址. https://sanwen8.cn/p/556cbwf.html journal of risk analysis and crisis response, vol. 7, no. 3 (october 2017) 156–165 ___________________________________________________________________________________________________________ 165 http://mp.weixin.qq.com/s/oya07yvr_8oco6y0o_fi3w http://mp.weixin.qq.com/s/oya07yvr_8oco6y0o_fi3w https://sanwen8.cn/p/556cbwf.html 1. 引言 2. 相关研究现状 3. 社区风险信息结构化及社区风险事件 4. 社区风险雷达设计与比较 5. 案例背景与分析 6. 结论 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <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> /pol <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> /ptb <feff005500740069006c0069007a006500200065007300730061007300200063006f006e00660069006700750072006100e700f50065007300200064006500200066006f0072006d00610020006100200063007200690061007200200064006f00630075006d0065006e0074006f0073002000410064006f0062006500200050004400460020006d00610069007300200061006400650071007500610064006f00730020007000610072006100200070007200e9002d0069006d0070007200650073007300f50065007300200064006500200061006c007400610020007100750061006c00690064006100640065002e0020004f007300200064006f00630075006d0065006e0074006f00730020005000440046002000630072006900610064006f007300200070006f00640065006d0020007300650072002000610062006500720074006f007300200063006f006d0020006f0020004100630072006f006200610074002000650020006f002000410064006f00620065002000520065006100640065007200200035002e0030002000650020007600650072007300f50065007300200070006f00730074006500720069006f007200650073002e> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word volume 12, issue 3-1 journal of risk analysis and crisis response, 2022, 12(3), 110-123 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i3.332 110 article cus-rf-based credit card fraud detection with imbalanced data wei li 1, cheng-shu wu 1 and su-mei ruan 1,* 1 school of finance, anhui university of finance and economics, bengbu (233030), anhui, china * correspondence: ruansumei0116@163.com received: july 2, 2022; accepted: september 4, 2022; published: september 30, 2022 abstract: with the continuous expansion of the banks' credit card businesses, credit card fraud has become a serious threat to banking financial institutions. so, the automatic and real-time credit card fraud detection is the meaningful research work. because machine learning has the characteristics of non-linearity, automation, and intelligence, so that credit card fraud detection can improve the detection efficiency and accuracy. in view of this, this paper proposes a credit card fraud detection model based on heterogeneous ensemble, namely cus-rf (cluster-based under-sampling boosting and random forest), based on clustering under-sampling and random forest algorithm. cus-rfbased credit card fraud detection model has the following advantages. firstly, the cus-rf model can better overcome the issue of data imbalance. secondly, based on the idea of heterogeneous ensemble learning, the clustering under-sampling method and random forest model are fused to achieve a better performance for credit card fraud detection. finally, through the verification of real credit card fraud dataset, the cus-rf model proposed in this paper has achieved better performance in credit card fraud detection compared with the benchmark model. keywords: credit card fraud detection; random forest; imbalanced data; heterogeneous ensemble; fintech 1. introduction credit card fraud has caused immense financial loss to both card-issuing banks and financial institutions. according to the statistical data from china banking association, by the end of 2018, credit card-based transaction in china has attained 38,200 billion yuan rmb at a growth rate of 24.9%; 73.2% of credit cards remain activated, and unpaid balance is 6,850 billion yuan rmb (on a year-onyear growth of 23.2%). it is especially noteworthy that credit card loss rate is 1.27%, slightly higher than 1.17% in previous year. the global credit card fraud-related loss climbs from 7.6 billion dollars in 2010 to 21.81 billion dollars in 2015, with a growth of 300% within 5 years. it is expected to reach 31.67 billion dollars by 2020. a conventional credit card fraud detection model is usually constructed using rules provided by experts. nevertheless, the aforesaid fraud detection models often demand manual parameter tuning and supervision from experts, which makes it impossible for financial institutions to discover fraudulent behaviors in time. moreover, it is a heavy task to check all the transactions one by one. to overcome the shortcomings in such work, financial institutions have employed machine learning algorithm and data mining methods in setting up an artificial intelligence (ai)-based credit card fraud detection model that is different from the traditional detection models. the machine learning algorithm can help financial institutions in constructing automated detection models to significantly wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 111 improve the fraud detection efficiency and speed. driven by machine learning, the novel credit card fraud detection model is trained and parameter-tuned in order to gain the expected effect. however, the ability of fraud detection will be greatly impaired should severe data imbalance exist amid credit card fraud data, so that the model may fail to exert its due performance. data and their features are the most critical factors affecting the operation of a fraud detection model. that said the category of imbalance perplexing data centralization is associated and demands analysis. furthermore, in the real world where financial institutions apply strict censorship, fraudulent credit card swiping is far less frequent than the normal operation. despite this, once such fraudulent transaction occurs, it will be hard for the financial institution involved to get back the money lost. therefore, when studying credit card fraud, it is urgent to considered the data imbalance issue. since automated credit card fraud detection system is investigated on the basis of balanced datasets when being established with machine learning, the model is hard to give a full play to its own strengths. the problem of credit card fraud detection for imbalanced data has been studied by scholars in different countries with different ideas. after a systematic study of category imbalance processing strategies, singh et al. (2021) compared the effectiveness and efficiency of different category imbalance processing methods and state-of-the-art classification methods, evaluating metrics in terms of precision, recall, k-fold cross-validation, auc-roc curve and execution time, and found that oversampling and under-sampling methods performed better for integrated classification models such as adaboost, xgboost and random forest performed better [1]. el-naby et al. (2022) addressed the fraud data imbalanced problem by using mixed sampling and oversampling preprocessing techniques, specifically, in oversampling, smote, bounded smote and adasyn were selected. in mixed sampling, smoteen and smotetomek to eliminate the data set imbalance problem, thereby improving credit card fraud detection accuracy [2]. as for the imbalanced classification problem for credit card fraud detection, makki et al. (2019) found lr, c5.0 decision tree algorithm, svm and ann to be the best methods after comparing imbalanced classification methods and based on three performance metrics: accuracy, sensitivity and average precision (auprc). although the above methods improve the performance of the classifier, when the data is extremely unbalanced, these methods may create a problem of false positives and major credit fraudulent cases may remain undetected [3]. as can be seen, the data imbalance problem interferes with credit card fraud detection and even false positives, and scholars in various countries are working to improve credit card fraud prediction performance and accuracy with imbalanced data. therefore, in creating automated credit card fraud detection model, attention should be paid to the data imbalance issue to evade possible impaired predicting accuracy of the machine learningbased prediction model. in addition, most of sorting algorithms would be undermined in performance in such case. in view of the data imbalance issue, smote (synthetic minority oversampling technique) technology has been extensively applied to financial distress prediction, bankruptcy forecasting, and credit card fraud fields in recent years [4–8]. it exhibits better sorting performance when compared with traditional rus (random under-sampling) and ros (random over-sampling) technologies. in this paper, a brand-new sorting technique based on clustering undersampling that targets imbalanced data is introduced into credit card fraud field. related references indicate this sorting technique has demonstrated outstanding performance in dealing with financial distress prediction and so on [9, 10]. as revealed by the findings of our experiment here, the technique excels smote in sorting imbalanced credit card data containing thousands of samples. it has proven application prospect in the financial field. wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 112 this paper establishes a heterogeneous ensemble model by innovatively ensemble cus (clusterbased under-sampling) with rf (random forest) [11]. to further verify the excellency of the proposed model, cus is blended with five classifiers respectively to form five heterogeneous ensemble models. since each classifier has distinct theoretical backgrounds, the paper combines same imbalanced data processing technique with different base learner methods in the hope to find out which theoretical design stands out in sorting the imbalanced data concerning credit card fraud. besides, in order to explore the performances of cus and smtote, the paper also combines smote with the foresaid base learner and adopt the widely applied evaluation index system in financial field to find out which imbalanced data sorting technique is more suitable for automatic classification of credit card data. the goal of doing so is to propose the best solution for imbalanced data processing and base learner choice. apart from that, another innovative aspect of this paper lies in its embedding of imbalance processing into base learner to form automatic sorting system. the resulting system could process imbalanced data and automatically sort base learners, which greatly cut down the time cost of learning task when compared with existing research achievements. this paper is arranged as follows. section 2 offers a literature review on data imbalance issue and credit card fraud detection model. section 3 describes the cus background model and then proposes an improved model that could deal with imbalanced data. section 4, the proposed model is experimentally testing, and the experimental findings are analyzed. and conclusions are drawn in section 5. 2. literature review as fintech develops by leaps and bounds in recent years, a great number of fintech banks emerges [12, 13]. consumer finance is influencing people’s life in a novel way. this also gives rise to fraud issue. detecting credit card fraud has become one of the much-concerned topics in financial industry. however, public models available for use remain rather limited. one of the major underlying reasons is that credit card transaction data are exclusively kept by the card-issuing agencies. as data owner, card-issue agencies must protect the data security and avoid leaking users’ privacy, so they are not going to disclose the datasets and related models they are using [14]. there are two common types of credit card fraud: application fraud and behavioral fraud. the latter type consists of card stealing, card forgery and non-existing card [15, 16]. fraudulent swiping is the most common type of credit card fraud [17]. in usual cases, after stealing the credit card or obtaining a temporary card, fraudster would use the card for consumption as much as possible. when committing fraud with a stolen card, the criminal usually transacts with the card at high frequency. should the fraudster forge a fake card with the information he has collected, forgery fraud will occur. while the victim still holds his own card for legal transaction, the fraudster transacts also with the fake card. the fake card will be used only for a few times before being abandoned and realized by the victim. the third type of behavioral fraud is card not present fraud which occurs in case of remote transaction. in such case, the transaction is made on the basis of card information only such as cad number, holder name, and valid term [18, 19]. the distinction between fake card fraud and card not present fraud lies in use of solid card in the former case and use of card information only in the latter [20]. there are two primary means of data mining used in credit card fraud detection system creation, namely supervised learning and unsupervised learning [21]. supervised learning aims to training dichotomic model, depending on the detection model trained with datasets marked as “normal” and “fraudulent” to tell fraudulent samples from normal wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 113 ones [22]. this is a most common way of fraud detection. recently, supervise learning algorithm has been applied to the establishment of some fraud detection systems. for example, soemers et al. proposed a dynamic model combining decision-making tree and context-based multi-arm gambler which demonstrated proven effect on identifying credit card fraud [23]. in the work by zliobaite, an adaptive algorithm was put forward, which was able to update fraud detection model with timedependent data flow in order to better adapt to the shifts in fraudulent transaction patterns [24]. blending recursive feature elimination, hyper-parameter optimization and smote technology, naoufal rtayli developed a mixed credit card fraud detection model [5]. other supervised learning methods, including bayes, artificial neural network [25] and support vector machine (svm), are also frequently used in fraud detection [16, 26, 27]. compared with semi-supervised and unsupervised fraud detection systems, supervised systems stand out with sufficient data training time that supports establishment of well-performing models [28]. the output from detection system trained with supervised learning technique has explicit meaning and can be directly applied to mode distinguishment. in unsupervised learning, the dataset samples for constructing fraud detection model carry no tag. instead, the unsupervised machine learning sets out to analyze data from different dimensions and resist fraud by finding out the association or difference between data [21]. for example, the gan model could learn normal data distribution and determine whether unknown test data are normal or fraudulent samples with the proposed abnormality scoring plan. in case of insufficient label data and severely imbalanced data, unsupervised learning model will be a better choice. in addition, unsupervised learning could update model with online unlabeled data from banks or financial institutions, thus rendering it possible to detect use of fraudulent credit card. for instance, an unsupervised learning model, called self organizing map (smo), is came up with for forming an unsupervised credit card fraud detection model [29, 30]. as smo model requires no priori information, the automated system proposed may use newly added transaction data to keep updating the model. besides, k-means clustering algorithm sorts of transaction data according to the similarity concerning credit card fraud features and thus gets used to fraud detection model creation [31]. we have analyzed the application of ensemble to credit card fraud and related fields in recent years. a few comprehensive algorithms have become popular base learners in ensemble algorithms, among which the most noteworthy ones are logistic regression (lr), support vector machine (svm), random forest (rf), k-nearest neighbor (k-nn) and gradient boosting decision tree (gbdt). therefore, we have built some models based on those base learners for comparison purpose. besides, the ensemble models proposed in previous studies are limited to smote and random oversampling when imbalanced data processing is involved. they rarely employ new techniques. this paper has applied cus to the processing of credit card-related imbalanced data for the first time. 3. research design 3.1. data the dataset used in credit card fraud detection is provided by machine learning group (http://mlg.ulb.ac.be), which could be downloaded from kaggle (https://www.kaggle.com/mlgulb/creditcardfraud). in the dataset there are data concerning credit card transactions completed by europeans in september 2013. according to it, among 284,807 transactions within two days, 492 are fraudulent. the dataset appears quite imbalanced, as positive examples (fraudulent swiping) accounts for 0.172% of total transactions. due to privacy protection, we cannot acquire the original wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 114 functions and more background information about related data. features v1, v2, ..., v28 are the major constituents acquired by pca (principal component analysis), while the features undergoing no pca conversion are “class” and “amount”. the feature “class” is a response variable, which is “1” when card fraud occurs and “0” otherwise. the goal of the task is to sort out normal transaction data from abnormal ones in the dataset and predict about the test data. table 1. credit card fraud detection dataset. instance number fraud samples normal samples feature number 284807 492 284315 30 as shown in table 1, there are altogether 284,807 transaction samples in the dataset. among them, only 492 are fraud samples, accounting for 0.17% of total dataset. the proportion of normal samples to fraud ones is as high as 578:1. in other words, this dataset features extremely imbalanced positive and negative samples. if no pre-treatment is made to improve the data imbalance here and data at primitive proportion are directly put into the classifier for training, the classifier is more like to view normal samples as white noise. this would impair the performance of whole combined fraud detection system. our data reveals data imbalance stays as a primary issue in fraud detection process. as a matter of fact, in an imbalanced dataset, we could find the training examples for one class variable are far less than that for the other one. accordingly, the first one is called minority set while the second as majority set. when sorting imbalanced fraud detection transaction dataset, most models perform well in identifying the majority set but much less accurate in the minority one, suggesting they are not good at detecting the minority samples. in order to effectively cope with the class imbalance issue in credit card fraud data, this paper introduces cus and combines it with rf to form machine learning-based heterogeneous ensemble. it succeeds in effectively sorting the credit card fraud data. in addition to that, we use also smote technology for comparison and compare data sorting performance by controlling the base learner. smote is a way of oversampling that generates random examples instead of achieving oversampling by repetition or replacement alone. furthermore, the technology can also progressively increase the learning process of fraud detection algorithm [32]. 3.2. cusboost cusboost is a combination of cus and adaboost algorithm. like rusboost and smote-boost, it contains key difference in sampling technique. smote-boost employs smote to sample minority examples, whereas rusboost chooses random under-sampling over the majority ones. based on comparison, the cusboost proposed by us selects the sampling from majority class based on clustering. cusboost separates first the majority and minority examples from the dataset first, and then applies k-means clustering algorithm to cluster majority examples to k clusters. here, parameter k is determined through hyper-parameter optimization. then, 50% of examples are randomized (or tuned as per field issue or dataset) with the rest being eliminated. random under-sampling is executed to each cluster. since clustering is applied in prior to sampling, theoretically speaking, the algorithm is expected to perform best when dataset is highly clustered. next, those representative samples are combined with the minority ones to form a well-balanced dataset. the strength of our algorithm is displayed on the inclusion of all the subspace examples in considering the majority class, as k-means clustering contains each example in certain cluster. other similar methods usually fail to proper represent the majority class. in fig. 1, the cus proposed is used to choose the majority wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 115 examples, in which red spots indicate the examples sorted out from the majority class while the black and red spots represent all the majority examples. figure 1. cluster-based under-sampling (cus) approach. 3.3. random forest model random forest (rf) is an expansion variant of bagging, which further introduces random attribution selection into decision-making tree training on the basis of base learner [33]. more specifically, traditional decision-making tree is to choose an optimal attribute from the attribute set (suppose there are d attributes) of current node when deciding to divide the attributes. in contrast, rf randomly selects one subset containing k attributes from the attribute set of each node on the base decision-making tree, and then selects one optimal attribute from the resulting subset for division purpose. here the parameter k decides how much randomness should be introduced. if 𝑘 = 𝑑, the base decision-making tree constructed will be nothing different from the traditional one; and if 𝑘 = 1, one attribute will be randomly selected for division. in general cases, 𝑘 = log 𝑑 is recommended. being easily and readily realizable at low computation cost, rf has exhibited strong performance in many practical tasks. though it is adapted from bagging by executing minor changes, its base learner is different from that in bagging, as the diversity of its base learner comes from both sample perturbation and attribute perturbation while that in bagging comes from sample perturbation (of initial training set sampling) alone. therefore, it is made possible to improve the post-ensemble generalization performance by further differentiating individual learners. rf has similar convergence as bagging. its initial performance could be far from being satisfactory, especially when there is only one base learner in the ensemble. however, as the number of individual learners keeps growing, rf usually will converge to lower generalization error. notably, the training efficiency of rf is more than usual superior to that of bagging because bagging employs “certain” tree when constructing individual decision-making tree and has to examine all the attributes of the node in attribute classification, whereas rf adopts “random” decision-making tree wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 116 and need examine one attribute subset only [31]. in view of the strengths displayed by rf model, it is widely applied to a variety of fields including credit card fraud prediction [35–38]. 3.4. model ensemble in consideration of the high imbalance perplexing the dataset used here, this paper attempts to construct a model with good predicting performance on a highly imbalanced dataset. based on cusboost, we replace the target of improvement for cus-adaboost with rf, and thus propose a novel cus employing boosting (rf) and name it as cus-rf. different from adaboost, rf keeps fitting new models during the learning to generate more accurate estimation about the response variable. when constructing the decision-making tree, rf algorithm trains the tree as per the residual error of previous tree in each iteration. finally, the output is the accumulation of all the tree classifications. based on boosting thought, rf serially builds several decision-making trees to predict the data. in other words, it performs gradient boosting in the space where loss function is. in details, it views to-be-resolved decision-making tree model as parameter and fits the negative gradient of loss function in current model upon each iteration to renew the parameter to minimize the loss function. rf could be considered as an extension of adaboost. the latter identifies problem on the basis of mis-classified data point and improves the model by adjusting the weight of such mis-classified data point, whereas rf finds out problem by negative gradient and improves the model through negative gradient computation. in fact, examples containing higher absolute value of negative gradient is going to gain high attention in subsequent training, because the resulting loss is likely to account for a large portion in the final loss function. therefore, it is more depended on to diminish the loss. this is something shred by rf and adaboost. compared with adaboost, rf could invoke more types of loss function and render more problems resolved [39–41]. serving as the basis of cus-rf, cusboost algorithm is based on adaboost algorithm itself. by introducing cus, cusboost does some improvement towards the adaboost to better balance the class distribution, and adaboost improves the classifier performance in virtue of those balanced data. cus-rf does achieve the same goal, but what is improved here is rf instead of adaboost. results indicate this algorithm features quick model training and satisfactory performance. embedding cus algorithm into rf algorithm is both implementable and effective. by following the thought of cusboost in algorithm improvement, we bring cus process into rf. 3.5. evaluation metrics unfortunately, some common classifier evaluation indexes turn out to be inapplicable to imbalanced dataset regardless of their performance in dealing with balanced dataset. this happens with accuracy rate also in credit card fraud detection, a most frequently used index. it does not consider sample distribution which stays as the key issue of imbalanced dataset. besides, accuracy rate may lead to misleading conclusion. for instance, in an imbalanced dataset, 99% of observed values are normal (meaning there is no fraud) while 1% are negative (meaning there is fraud). if taking the positive class (or majority class) prediction as the standard, this model will be deemed as having 99% of accuracy rate, because this method opts to choose the majority class and thus yield the better outcome. the ratio indicates the classifier is accurate, however it ignores the prediction of minority class (or negative class) which should be attached with highest importance in case of dataset imbalance. thus, we need to adjust the model performance evaluation and rely on the evaluation indexes insensitive to sample distribution. we choose four evaluation indexes that are widely used wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 117 in imbalanced datasets, namely area under the receiver operating characteristic curve (auc), tpr, fpr, specificity and maximum ks value. here we are going to sequence the samples by classifierderived prediction outcome and predict all the samples one by one as positive examples. each computation brings in values of two important measures which are used as horizontal and vertical coordinates for mapping. finally, a receiver operating characteristic curve (roc curve) is generated, which has its vertical shaft as true positive rate (tpr) and horizontal one as “false positive rate (fpr)”. definitions of tpr and fpr are offered below: 𝑇𝑃𝑅 = (1) 𝐹𝑃𝑅 = (2) when classifiers are compared in term of performance, if one classifier’s roc curve gets completely covered by that of another, it can be assumed that the latter outperforms the former. if two classifiers have their roc curves overlap each other, it is hard to tell which one is better. in such case, we employ auc (area under roc curve) to determine the inter-model difference in performance. auc is the sum of all the areas under roc. suppose roc is formed with points with coordinates {(𝑥 ,𝑦 ),(𝑥 ,𝑦 ),∙∙∙,(𝑥 ,𝑦 )} that are connected in order, then 𝐴𝑈𝐶 = ∑ (𝑥 − 𝑥 ) ∙ (𝑦 +𝑦 ) (3) specificity is the ratio by which no fraud is predicted as non-fraud. it can be figured out according to following formula: 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐𝑖𝑡𝑦 = 𝑇𝑁 (𝑇𝑁 + 𝐹𝑃)⁄ (4) at last, after all the models gain the optimal hyper-parameters after parameter tuning, we retrain the models with such hyper-parameters on training and development sets before testing again the retrained models on test set. afterwards, five indexes are invoked to evaluate the model performance respectively, as this is an effective index for imbalanced dataset. 4. experimental results and discussion 4.1. feature selection after finishing data preprocessing, we need to input meaningful features into the machine learning algorithm and model for training purpose. in general sense, features should be weighed from two aspects, whether the feature diverges and whether feature remains related to the goal. if a feature does not diverge, such as having a variance nearing 0, it means the samples don’t differ from each other in this feature, and the feature is of little use to sample distinguishment. as for the second aspect which is more apparent, the feature highly concerned with the goal should be prioritized. except for variance method, all other methods introduced in this paper set out from relevance. the feature selection methods could be divided into three types by feature selection form, namely filter, wrapper and embedded. filter method scores every feature as per its divergence or relevance and sets threshold or number of threshold candidates to screen the features. wrapper method selects a few features or eliminate some upon each time by referring to the objective function (or prediction effect score in usual cases). for embedded method, features are trained first with certain machine learning algorithms and models to figure out their weight coefficients, and then features are selected based on the resulting coefficients in a descending order. though having something common with filter, the embedded determines whether a feature is good or not through training. we select features based on xgboost-based approach in the feature selection library of wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 118 sklearn and then apply filter method to generate a database that contains less samples but is more related to the sample type. 4.2. parameter tuning to train models with better performance, we perform further optimization with parameters from such models as cus-gbdt, smote-gbdt, cus-rf and smote-rf in order to exert their best predicting performance. the post-optimization parameters are listed in table 2. table 2. optimized parameters of cus-gbdt and xgboost. classifier parameter description cus-gbdt smote-gbdt max_depth maximum depth of the individual regression estimators. n_estimators the number of boosting stages to perform. subsample the fraction of samples to be used for fitting the individual base learners. loss loss function to be optimized. cus-rf smote-rf n_estimators the number of boosting stages to perform. max_depth maximum depth of the individual regression estimators. scale_pos_weight the weight of positive samples. 4.3. benchmark models table 3. a list of the proposed method and benchmark methods. no. method description 1 cus-gbdt based on cus-gbdt, with cus for preprocessing class imbalance data and gbdt for sample classification. 2 cus-knn based on cus-knn, with cus for preprocessing class imbalance data and knn for sample classification. 3 cus-lr based on cus-lr, with cus for preprocessing class imbalance data and lr for sample classification. 4 cus-rf based on cus-rf, with cus for preprocessing class imbalance data and rf for sample classification. 5 cus-svm based on cus-gbdt, with cus for preprocessing class imbalance data and gbdt for sample classification. 6 smote-gbdt based on smote-gbdt, with smote for preprocessing class imbalance data and gbdt for sample classification. 7 smote-knn based on smote-knn, with smote for preprocessing class imbalance data and knn for sample classification. 8 smote-lr based on smote-lr, with smote for preprocessing class imbalance data and lr for sample classification. 9 smote-rf based on smote-rf, with smote for preprocessing class imbalance data and rf for sample classification. 10 smote-svm based on smote-svm, with smote for preprocessing class imbalance data and svm for sample classification. wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 119 table 3 briefly lists the proposed method and benchmark methods. in the experiment, we adopt gbdt, knn, lr, rf and svm as base learners to ensemble with cus and smote respectively to build heterogeneous ensemble models to determine which imbalanced data processing method or classifier performs well in dealing with imbalanced dataset containing credit card fraud data. 4.4. results and discussion the data screened with feature selection method are brought into the heterogeneous ensemble models composed of imbalanced data processing and classifier. the models are appraised in terms of five indexes, namely auc, tpr, fpr, specificity and precision. the experimental findings are illustrated in fig. 2 and 3 and table 4. in this paper, some common classifiers in credit card fraud field are selected as component classifiers for the heterogeneous ensemble models and then combined with two ways of imbalanced data processing cus and smote in the hope to determine which processing method outperforms and which classifier is better for classifying the credit card fraud. table 4. performance between fraud detection models and benchmark models. models auc tpr fpr specificity precision cus-gbdt 0.964477 0.938776 0.107574 0.892426 0.892426 cus-knn 0.9999 1 0.001565 0.998435 0.998435 cus-lr 0.860443 1 1 0 0 cus-rf 0.999912 1 0.001565 0.998435 0.998435 cus-svm 0.943549 0.785714 0.001811 0.998189 0.998189 smote-gbdt 0.960775 0.816327 0.030758 0.969242 0.969242 smote-knn 0.999804 1 0.002 0.998048 0.998048 smote-lr 0.85719 1 1 0 0 smote-rf 0.999881 1 0.001864 0.998136 0.998136 smote-svm 0.580702 0.555556 0.389474 0.610526 0.610526 figure 2. the column graph of the values of performance measures (cus). in the evaluation index system established by this paper, auc is used to measure model’s comprehensive performance in identifying credit card fraud samples, tpr could figure out the accuracy rate of model in determining sample company’s financial risk, fpr acts to assess the probability for a listed company sample free from financial risk to be wrongfully identified by the model, specificity measures the accuracy rate of a model in determining whether the sample company 0 0.2 0.4 0.6 0.8 1 1.2 cus-gbdt cus-knn cus-lr cus-rf cus-svm auc tpr fpr precision specificity wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 120 contains financial risk, and precision indicates the ratio of correctly classified positive samples in the positive samples identified by classifier. among those five indexes, only fpr score is negatively related to the model’s performance in identifying list company samples free from financial risk, while the scores of rests six methods are all positively related to the model’s performance in the aspect involved. judged from the experimental findings, among five cus-based heterogeneous ensemble models, four models (incl. cus-knn, cus-lr, cus-rf and cus-svm) are either better than or equal to the smote-based heterogeneous ensemble models in terms of five evaluation indexes, suggesting cus enjoys great strengths in dealing with imbalanced credit card fraud data when compared with smote. in the meanwhile, the findings reveal the cus-rf model proposed in this paper harvests the best scores in five indexes, indicating the improved model has great potential for application to credit card fraud data classification. figure 3. the column graph of the values of performance measures (smote). 5. conclusions it is still quite tough and challenging to identify the credit card fraud samples. through the work here, we aim to introduce novel imbalanced data processing techniques to improving the machine learning model’s predicting performance during credit card fraud detection. based on the heterogeneous ensemble principle, we have introduced cus and rf to prediction model creation. to verify the experimental findings, we have also embedded the frequently used classifiers in this field into cus to generate comparative models. in addition, by keeping classifier unchanged, we also create comparative models with smote to prove the superiority of cus over smote in classifying imbalanced credit card fraud data. in the future, it is planned to further testify the reliability of the model with more complicated data from the real world and develop a self-adaptive credit card fraud detection system. acknowledgments: we would like to thank xu-dong du, a doctoral candidate at the school of management, hefei university of technology, for providing experimental support for this paper. funding: this research was funded by the philosophy and social science planning project of anhui province (national social science fund incubation project), grant number ahskf2021d07. 0 0.2 0.4 0.6 0.8 1 1.2 smote-gbdt smote-knn smote-lr smote-rf smote-svm auc tpr fpr precision specificity wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 121 conflicts of interest: the authors declare that they have no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] singh a, ranjan rk, tiwari a. credit card fraud detection under extreme imbalanced data: a comparative study of data-level algorithms[j]. journal of experimental and theoretical artificial intelligence, 2021,34:571-598. doi: https://doi.org/10.1080/0952813x.2021.1907795. [2] el-naby aa, hemdan eed, el-sayed a. an efficient fraud detection framework with credit card imbalanced data in financial services[j]. multimedia tools and applications, 2022. doi: https://doi.org/10.1007/s11042-022-13434-6. [3] makki s, assaghir z, taher y, haque r, hacid m, zeineddine h. an experimental study with imbalanced classification approaches for credit card fraud detection[j]. ieee access, 2019,7:93010-93022. doi: https://doi.org/10.1109/access.2019.2927266. [4] sun j, lang j, fujita h, li h. imbalanced enterprise credit evaluation with dte-sbd: decision tree ensemble based on smote and bagging with differentiated sampling rates[j]. information sciences, 2018,425:76–91. doi: https://doi.org/10.1016/j.ins.2017.10.017. [5] rtayli n, enneya n. enhanced credit card fraud detection based on svm-recursive feature elimination and hyper-parameters optimization[j]. journal of information security and applications, 2020,55:102596. doi: https://doi.org/10.1016/j.jisa.2020.102596. [6] sun j, fujita h, zheng y, ai w. multi-class financial distress prediction based on support vector machines integrated with the decomposition and fusion methods[j]. information sciences, 2021,559:153–170. doi: https://doi.org/10.1016/j.ins.2021.01.059. [7] shen f, liu y, wang r, zhou w. a dynamic financial distress forecast model with multiple forecast results under unbalanced data environment[j]. knowledge-based systems, 2020,192:105365. doi: https://doi.org/10.1016/j.knosys.2019.105365. [8] sun j, li h, fujita h, et al. class-imbalanced dynamic financial distress prediction based on adaboost-svm ensemble combined with smote and time weighting[j]. information fusion, 2020,54:128–144. doi: https://doi.org/10.1016/j.inffus.2019.07.006. [9] du x, li w, ruan s, li l. cus-heterogeneous ensemble-based financial distress prediction for imbalanced dataset with ensemble feature selection[j]. applied soft computing journal, 2020,97. doi: https://doi.org/10.1016/j.asoc.2020.106758. [10] li w, ding s, chen y, et al. transfer learning-based default prediction model for consumer credit in china[j]. journal of supercomputing, 2019,75:862–884. doi: https://doi.org/10.1007/s11227-018-2619-8. [11] khan a, rehman hu, habib u, ijaz u. detecting n6-methyladenosine sites from rna transcriptomes using random forest[j]. journal of computational science, 2020,47:101238. doi: https://doi.org/10.1016/j.jocs.2020.101238. [12] laidroo l, koroleva e, kliber a, et al. business models of fintechs – difference in similarity[j]. electronic commerce research and applications, 2021,46:101034. doi: https://doi.org/10.1016/j.elerap.2021.101034. [13] bollaert h, lopez-de-silanes f, schwienbacher a. fintech and access to finance[j]. journal of corporate finance, 2021,68:101941. doi: https://doi.org/10.1016/j.jcorpfin.2021.101941. [14] west j, bhattacharya m. intelligent financial fraud detection: a comprehensive review[j]. computers and security, 2016,57:47–66. doi: https://doi.org/10.1016/j.cose.2015.09.005. [15] azevedo c da s, gonçalves rf, gava vl, spinola m de m. a benford’s law based methodology for fraud detection in social welfare programs: bolsa familia analysis[j]. physica a: statistical mechanics and its applications, 2021,567:125626. doi: https://doi.org/10.1016/j.physa.2020.125626. [16] forough j, momtazi s. ensemble of deep sequential models for credit card fraud detection[j]. applied soft computing, 2021,99:106883. doi: https://doi.org/10.1016/j.asoc.2020.106883. [17] wang d, chen b, chen j. credit card fraud detection strategies with consumer incentives[j]. omega, 2019,88:179–195. doi: https://doi.org/10.1016/j.omega.2018.07.001. [18] soltani halvaiee n, akbari mk. a novel model for credit card fraud detection using artificial immune systems[j]. applied soft computing journal, 2014,24:40–49. doi: https://doi.org/10.1016/j.asoc.2014.06.042. wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 122 [19] correa bahnsen a, aouada d, stojanovic a, ottersten b. feature engineering strategies for credit card fraud detection[j]. expert systems with applications, 2016,51:134–142. doi: https://doi.org/10.1016/j.eswa.2015.12.030. [20] zhang x, han y, xu w, wang q. hoba: a novel feature engineering methodology for credit card fraud detection with a deep learning architecture[j]. information sciences, 2021,557:302–316. doi: https://doi.org/10.1016/j.ins.2019.05.023. [21] carcillo f, le borgne ya, caelen o, et al. combining unsupervised and supervised learning in credit card fraud detection[j]. information sciences, 2021,557:317–331. doi: https://doi.org/10.1016/j.ins.2019.05.042. [22] błaszczyński j, de almeida filho at, matuszyk a, et al. auto loan fraud detection using dominance-based rough set approach versus machine learning methods[j]. expert systems with applications, 2021,163. doi: https://doi.org/10.1016/j.eswa.2020.113740. [23] soemers djnj, brys t, driessens k, et al. adapting to concept drift in credit card transaction data streams using contextual bandits and decision trees[c]. proceedings of the 30th innovative applications of artificial intelligence conference, new orleans louisiana, usa, february 2 – 7,2018, iaai 2018 7831–7836. [24] wang z, jiang c, zhao h, ding y. mining semantic soft factors for credit risk evaluation in peer-to-peer lending[j]. journal of management information systems, 2020,37:282–308. doi: https://doi.org/10.1080/07421222.2019.1705513. [25] pang x, zhou y, wang p, et al. an innovative neural network approach for stock market prediction[j]. journal of supercomputing, 2020,76:2098–2118. doi: https://doi.org/10.1007/s11227-017-2228-y. [26] craja p, kim a, lessmann s. deep learning for detecting financial statement fraud[j]. decision support systems, 2020,139:113421. doi: https://doi.org/10.1016/j.dss.2020.113421. [27] chen yj, wu ch, chen ym, et al. enhancement of fraud detection for narratives in annual reports[j]. international journal of accounting information systems, 2017,26:32–45. doi: https://doi.org/10.1016/j.accinf.2017.06.004. [28] baesens b, höppner s, verdonck t. data engineering for fraud detection[j]. decision support systems, 2021,150. doi: https://doi.org/10.1016/j.dss.2021.113492. [29] olszewski d. fraud detection using self-organizing map visualizing the user profiles[j]. knowledge-based systems, 2014,70:324–334. doi: https://doi.org/10.1016/j.knosys.2014.07.008. [30] zaslavsky v, strizhak a. credit card fraud detection using self-organizing maps[j]. information & security: an international journal, 2006,18:48–63. doi: https://doi.org/10.11610/isij.1803. [31] srivastava a, kundu a, sural s, majumdar ak. credit card fraud detection using hidden markov model[j]. ieee transactions on dependable and secure computing, 2008,5:37–48. doi: https://doi.org/10.1109/tdsc.2007.70228. [32] gianini g, ghemmogne fossi l, mio c, et al. managing a pool of rules for credit card fraud detection by a game theory based approach[j]. future generation computer systems, 2020,102:549–561. doi: https://doi.org/10.1016/j.future.2019.08.028. [33] akila s, srinivasulu reddy u. cost-sensitive risk induced bayesian inference bagging (ribib) for credit card fraud detection[j]. journal of computational science, 2018,27:247–254. doi: https://doi.org/10.1016/j.jocs.2018.06.009. [34] zhou j, li w, wang j, et al. default prediction in p2p lending from high-dimensional data based on machine learning[j]. physica a: statistical mechanics and its applications, 2019,534:122370. doi: https://doi.org/10.1016/j.physa.2019.122370. [35] jurgovsky j, granitzer m, ziegler k, et al. sequence classification for credit-card fraud detection[j]. expert systems with applications, 2018,100:234–245. doi: https://doi.org/10.1016/j.eswa.2018.01.037. [36] huang yp, yen mf. a new perspective of performance comparison among machine learning algorithms for financial distress prediction[j]. applied soft computing journal, 2019,83:105663. doi: https://doi.org/10.1016/j.asoc.2019.105663. [37] ashraf s, félix egs, serrasqueiro z. development and testing of an augmented distress prediction model: a comparative study on a developed and an emerging market[j]. journal of multinational financial management, 2020,57–58,100659. doi: https://doi.org/10.1016/j.mulfin.2020.100659. [38] petropoulos a, siakoulis v, stavroulakis e, vlachogiannakis ne. predicting bank insolvencies using machine learning techniques[j]. international journal of forecasting, 2020,36:1092–1113. doi: https://doi.org/10.1016/j.ijforecast.2019.11.005. wei li, cheng-shu wu and su-mei ruan / journal of risk analysis and crisis response, 2022, 12(3), 110-123 doi: https://doi.org/10.54560/jracr.v12i3.332 123 [39] pradeepkumar d, ravi v. forecasting financial time series volatility using particle swarm optimization trained quantile regression neural network[j]. applied soft computing journal, 2017,58:35–52. doi: https://doi.org/10.1016/j.asoc.2017.04.014. [40] jones s, johnstone d, wilson r. an empirical evaluation of the performance of binary classifiers in the prediction of credit ratings changes[j]. journal of banking and finance, 2015,56:72–85. doi: https://doi.org/10.1016/j.jbankfin.2015.02.006. [41] he y, zhang w. probability density forecasting of wind power based on multi-core parallel quantile regression neural network[j]. knowledge-based systems, 2020,209:106431. doi: https://doi.org/10.1016/j.knosys.2020.10643. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 12, issue 2-1 journal of risk analysis and crisis response, 2022, 12(2), 62-76 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i2.325 62 article-infus 2021 special issue seismic risk analysis of hospital buildings: a novel interval-valued spherical fuzzy aras akın menekşe 1 and hatice camgöz akdağ 2,* 1 graduate school, istanbul technical university, istanbul (34398), turkey 2 management engineering department, management faculty, istanbul technical university, istanbul (34398), turkey * correspondence: camgozakdag@itu.edu.tr; tel.: +90-212-293-1300 received: november 15, 2021; accepted: march 9, 2022; published: july 10, 2022 abstract: hospitals are vital infrastructures, but they can be particularly susceptible if located in an earthquake-prone area. estimating the performance and projected damage of these structures in the event of future earthquakes is critical to minimize physical damage and disruption of surgery, rehabilitation, laboratory services, and other medical care operations. a quick and low-cost preassessment of the sensitivity of these structures will help hospital managements prepare for suitable retrofitting and reconstruction. on the other hand, a seismic risk assessment requires constructing a model that can offer expert opinions in a quantitative, methodical, and quantifiable way while still reflecting the uncertain and imprecise character of the process. this paper proposes a new decision support model by extending additive ratio assessment (aras) with interval-valued spherical fuzzy sets. the study includes sensitivity and comparison analyses, practical implications, limitations, and future research avenues. the applicability of our methodology demonstrated a numerical example for assessing the seismic risk levels of hospital buildings. keywords: mcdm; interval-valued spherical fuzzy sets; aras; hospital building risk assessment 1. introduction hospitals are essential components of a healthcare system. their position is especially more vital after a catastrophe, such as an earthquake, since they must give medical and surgical treatment as well as nursing care to ill or wounded individuals [1]. in the case of an earthquake, hospitals must be fully functioning to safeguard patients and healthcare staff, as well as to give emergency care and medical treatment to patients who are driven to healthcare facilities. according to assessments undertaken after big earthquakes, hospital performance and functioning are connected not just to structural damage but also to damage to nonstructural materials and medical equipment [2]. hospitals, unlike many other service buildings, have complex and diverse structures. the analysis of the earthquake behavior of these structures, which contain offices, dormitories, conference rooms, and social areas, as well as operating rooms, chemical and biological laboratories, and medical warehouses, is a time-consuming and costly process. a practical methodology of assessing the seismic sensitivity of hospital buildings would enable the efficient use of financial resources for future planning of potential retrofitting and rebuilding plans. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 63 in this context, seismic risk level assessment of hospital buildings can be considered as a decision-making problem with inherent uncertainties such as vague expert evaluations, complicated qualitative and quantitative data, and ambiguous data relationships [3]. on the other hand, multicriteria decision-making (mcdm) models are widely used to rank alternatives from a pool of candidates based on a set of criteria. uncertainty, on the other hand, is a part of the actual world, and it is inherent in most mcdm problems. the criteria are usually qualitative, and decision-makers may be unclear about their assessments and can only evaluate them through linguistic terms, and fuzzy systems give appropriate ways to convert linguistic assessments of decision-makers to a numerical form and model the uncertainty in these problems. spherical fuzzy sets [4] are one of the recent extensions of ordinary fuzzy sets and provide users the opportunity to express their membership, non-membership, and hesitancy degrees independently, and in a spherical volume. interval-valued spherical fuzzy sets [5] on the other hand, provide more space than the first generation for modeling the fuzziness in the problems. 1.1. related work additive ratio assessment (aras) [6] is an mcdm method that evaluates the performance of selected alternatives based on the comparison of each one with the ideal best alternative through a utility function. many academics favor fuzzy aras approaches to handle problem-solving situations. karagoz et al. [7] suggested an interval type-2 fuzzy aras for selecting a location for a recycling facility. jaukovic et al. [8] combined pairwise relative criteria importance assessment (piprecia) and interval-valued valued triangular fuzzy aras and evaluated online learning courses. moreover, heidary et al. [9] integrated data envelopment analysis with aras in a triangular fuzzy environment and ranked human resource practices, and buyukozkan and guler [10] proposed a simple additive weighting (saw) aras model for smartwatch assessment by utilizing hesitant fuzzy sets to model the uncertainty. on the other hand, yildirim and mercangoz [11] worked for a hesitant fuzzy analytical hierarchy process (ahp) aras model to evaluate logistics performances of countries, and gul [12] studied fermatean fuzzy sets to provide a saw aras viekriterijumsko kompromisno rangiranje (vikor) model with an application for selecting a pandemic testing laboratory. interval-valued spherical fuzzy sets have been increasingly used in various decision-making problems. gul and ak [13] proposed an interval-valued spherical fuzzy order preference by similarity to ideal solution (topsis) and evaluated a marble manufacturing facility, and aydin and gundogdu [14] developed a multi-objective optimization by a ratio analysis plus the full multiplicative form (multimoora) model and ranked performances of companies. duleba et al. [15] worked for an ahp methodology to assess public transportation, and jin et al. [16] presented an organization, rangement et synthese de donnes relationnelles (oreste) method for ranking quality characteristics. moreover, lathamaheswari et al. [17] introduced some new operators for interval-valued spherical fuzzy sets and selected a station for air quality analysis. a complete and systematic literature review of recent and state-of-the-art studies on spherical fuzzy studies can be found in the study of ozceylan et al. [18]. in the context of earthquake-related problems, several mcdm models have been proposed. adafer and bensaibi [19] developed a seismic vulnerability assessment method for roads and quantified the criteria that affect the seismic behavior of roads with an ahp model. jena et al. [20] evaluated earthquake vulnerability assessment by using an ahp vikor model, and kumlu and akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 64 tudes [21] used ahp topsis for determining earthquake-risky areas. cremen and galasso [22] proposed a general mcdm model and studied a decision-making problem for an earthquake early warning system of a school. alizadeh et al. [23] assessed the seismic vulnerability of residential houses by utilizing the ahp technique. tyagi et al. [24] used ahp to evaluate the landslide hazard index to be used in a landslide hazard zonation map. in addition, zhang et al. [25] investigated the effects of fault rupture on seismic responses of highway bridges band utilized a topsis weighted sum model (wsm) method for optimizing the fault crossing angle. the above-mentioned authors did not consider uncertainty and presented their works in a crisp environment. the following gaps have been identified as a result of the completed review of the aforementioned research:  the aras method has not benefited from interval-valued spherical fuzzy sets for addressing the vagueness.  the nature of decision-making problems is unclear, which may have a substantial impact on the outcomes. however, the majority of previous earthquake-related mcdm approaches either ignored uncertainty and used crisp integers or only used twodimensional fuzzy sets.  the examination of seismic risk assessment of hospital buildings has received little attention. no previous research provides a model for prioritizing hospital buildings in terms of their seismic risk levels. around earthquake-related mcdm investigations, little has been done in terms of establishing sensitivity and comparative analyses. 1.2. motivation and contribution the major goals of this research are to fill the gaps mentioned above, offer a novel decision support model, and handle a seismic risk evaluation problem. the fuzziness is modeled with interval-valued spherical fuzzy sets, and the alternatives are prioritized by using the aras technique. this work offers a novel approach by providing an interval-valued spherical fuzzy aras and evaluating the seismic risk assessment of hospital buildings in an mcdm structure while modeling the vagueness of experts by using fuzzy logic theory. the rest of the paper is organized as follows: section 2 presents the methodology as follows: first, the structure of traditional aras methodology which forms the backbone of the model is given, then mathematical operations of interval-valued spherical fuzzy sets, which are the building elements of the methodology, are summarized. finally, the proposed interval-valued spherical fuzzy aras method is presented in a step-by-step form. in section 3, the application is presented as follows: first, the criteria are introduced, then the numerical solution is provided, and finally, sensitivity, comparative and rank reversal analyses are given for further discussion. the paper is finalized in section 4 with practical implications, limitations, and future research avenues. 2. methodology aras method evaluates the performance of selected alternatives and compares the scores of those selected alternatives with the ideal best alternative [6]. in this section, the interval-valued spherical fuzzy aras model is developed based on the definition, basic operators, and the linguistic scale presented for interval-valued spherical fuzzy sets. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 65 2.1. classical aras a decision matrix x is defined with n criteria and m alternatives as shown in equation 1. 𝑋 = 𝑥ij mxn = 𝑥 𝑥 . . . 𝑥 𝑥 𝑥 . . . 𝑥 ⋮ ⋮ ⋮ ⋮ 𝑥 𝑥 . . . 𝑥 (1) where, xij is the performance value of the ith alternative with respect to jth criterion. normalization of decision matrix is obtained as equation 2 for beneficial criterion. 𝑁 = (2) where nij is the normalized value of ith alternative. normalization of decision matrix for cost criterion is obtained as follows: 𝑥 ∗ = (3) 𝑁 = ∗ ∗ (4) weighted normalized matrix w is obtained as in equation 5. 𝑊 = 𝑑ij mxn = 𝑑 𝑑 . . . 𝑑 𝑑 𝑑 . . . 𝑑 ⋮ ⋮ ⋮ ⋮ 𝑑 𝑑 . . . 𝑑 (5) 𝑑 = 𝑁 × 𝑤 (6) where nij is the normalized; dij is the weighted normalized value of the ith alternative with respect to jth criterion, and wj is weight of jth criterion. optimality value oi for the ith alternative is obtained as in equation 7. 𝑂 = 𝑑 (7) the degree of utility ui for the ith alternative is obtained as in equation 8. 𝑈 = (8) where, oi is the optimality value for the ith alternative and oo is the optimality value of the optimal alternative. the alternatives are ranked according to utility degrees as the highest is the best. 2.2. interval-valued spherical fuzzy sets a spherical fuzzy set ( sa ~ ) must satisfy the following condition: uuuuvu sss aaa                    ,1)()()(0 2 ~ 2 ~ 2 ~  (9) where  , v and  are the membership, non-membership, and hesitancy degrees, respectively. interval-valued fuzzy sets were later developed and provide an increased fuzziness modeling capacity with an interval-type structure rather than a single point. definition and basic operations [5] of interval-valued spherical fuzzy sets are given below: definition 1. an interval-valued spherical fuzzy set sa ~ of the universe of discourse u is defined as akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 66                          uuuuuvuvuuua ssssss u a l a u a l a u a l as |)(),(,)(),(,)(),(, ~ ~~~~~~  (10) where 0 ( ) ( ) 1; 0 ( ) 1 and 0 ( ) ( ) 1 s s s s s s l u l u l u a a a a a a u u v v u u u                 and 1))(())(())((0 2~2~2~  uuvu sss u a u a u a  . for each )(),(, ~~ uvuuu ss u a u a  and )(~ u s u a  are the upper degrees of membership, non-membership, and hesitancy degrees of u to sa ~ . for each, ,uu  if )()(),(, ~~~~ uvuvu ssss u a l a u a l a   and )()( ~~ uu ss u a l a   then an interval-valued spherical fuzzy set reduces to a single-valued spherical fuzzy set. for an interval-valued spherical fuzzy set, the pair             )(),(,)(),(,)(),( ~~~~~~ uuuvuvuu ssssss u a l a u a l a u a l a  is called an interval-valued spherical fuzzy number. for convenience, the pair             )(),(,)(),(,)(),( ~~~~~~ uuuvuvuu ssssss u a l a u a l a u a l a  is denoted by      fedcba ,,,,,~  where             1;1,0,,1,0,,1,0, 222  fdbfedcba . note that,            1,1 , 0, 0 , 0, 0 ; 0, 0 , 1,1 , 0, 0     and      1,1,0,0,0,0~ *  are the largest, smallest and moderate interval-valued spherical fuzzy numbers, respectively. definition 2. basic operations of interval-valued spherical fuzzy sets are given below: addition;                                                                2/12 2 2 1 2 2 2 1 2 1 2 2 2/12 2 2 1 2 2 2 1 2 1 2 2 2121 2/12 2 2 1 2 2 2 1 2/12 2 2 1 2 2 2 1 21 ))(()()(1(()()(1 ,))(()()(1(()()(1 , ,)()()()(,)()()()( ~~ fffbfb eeeaea ddcc bbbbaaaa  (11) multiplication;                                                        2/12 2 2 1 2 2 2 1 2 1 2 2 2/1 2 2 2 1 2 2 2 1 2 1 2 2 2/12 2 2 1 2 2 2 1 2/12 2 2 1 2 2 2 12121 21 ))(()()(1(()()(1 ,))(()()(1(()()(1 ,)()()()(,)()()()(,, ~~ fffdfd eeecec ddddccccbbaa  (12) multiplication by a scalar  ; 0                                                                          2/1 222 2/1 222 2/1 2 2/1 2 11,11 ,,,11,11 ~.    fbbeaa dcba (13)  th power ~ ; 0 akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 67                                                                          2/1 222 2/1 222 2/1 2 2/1 2 11,11 ,11,11,, ~    fddecc dcba (14) definition 3. interval-valued spherical fuzzy weighted geometric mean ivsfwgm operator is defined as follows:                                                                                                         2/1 1 22 1 2 2/1 1 22 1 2 2/1 1 2 2/1 1 2 11 2121 111,111 ,11,11, ~...~~)~,...,~,~( 21 jjjj jj jj n wn i jj wn i j wn i jj wn i j wn i j wn i j n j w j n j w j w n ww nw fddecc dcba ivsfwgm  (15) definition 4. score and accuracy functions are defined as follows: 𝑆𝑐𝑜𝑟𝑒(𝛼) = ( ⁄ ) ( ⁄ ) (16) 𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦(𝛼) = (17) 2.3. proposed methodology: interval-valued spherical fuzzy aras the flowchart (figure 1) and steps of the proposed model are as follows: figure 1. the flowchart of the proposed model: interval-valued spherical fuzzy aras. define an mcdm problem with a set of alternatives and the most related criteria. let  mi aaaa ,...,, 21 be the alternatives and  nj aaac ,...,, 21 be the selected criteria set. step 1. the selected alternatives are evaluated by decision-makers with respect to criteria by utilizing the linguistic terms that is given in table 1. in table 1, µl, νl and πl are the lower, and µu, νu and πu are the upper degrees of membership, non-membership, and hesitancy degrees. step 2. linguistic evaluations of each decision-maker are transformed to interval-valued spherical fuzzy numbers in a matrix form. the linguistic scale that is given in table 1 is used for the akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 68 conversion operation. then, all matrices are aggregated to obtain one unique collective matrix, which is called the interval-valued spherical fuzzy alternative evaluation matrix. table 1. linguistic terms and corresponding interval-valued spherical fuzzy numbers. linguistic term μl, μu vl, vu πl, πu absolutely most important (ami) [0.85, 0.95] [0.10, 0.15] [0.05, 0.15] very high importance (vhi) [0.75, 0.85] [0.15, 0.20] [0.15, 0.20] high importance (hi) [0.65, 0.75] [0.20, 0.25] [0.20, 0.25] slightly more importance (smi) [0.55, 0.65] [0.25, 0.30] [0.25, 0.30] equal importance (ei) [0.50, 0.55] [0.45, 0.55] [0.30, 0.40] slightly low importance (sli) [0.25, 0.30] [0.55, 0.65] [0.25, 0.30] low importance (li) [0.20, 0.25] [0.65, 0.75] [0.20, 0.25] very low importance (vli) [0.15, 0.20] [0.75, 0.85] [0.15, 0.20] absolutely least important (ali) [0.10, 0.15] [0.85, 0.95] [0.05, 0.15] step 3. decision-makers evaluate the importance level of each criterion by utilizing the same linguistic terms that is given in table 1. step 4. linguistic criterion evaluations are converted to interval-valued spherical fuzzy numbers and aggregated to obtain one unique matrix which is called interval-valued spherical fuzzy criterion weight matrix. step 5. interval-valued spherical fuzzy decision matrix d is obtained by multiplying the alternative evaluation matrix (step 2) by criterion weights (step 3). equation 3 is used for the multiplication operation. the structure of the decision matrix is given in equation 18. 𝐷 = 𝐶 (𝐴 )mxn = �̃� �̃� . . . �̃� �̃� �̃� . . . �̃� ⋮ ⋮ ⋮ ⋮ �̃� �̃� . . . �̃� (18) where �̃� = µi j, νi j, πi j is the interval-valued spherical fuzzy element of the evaluation matrix; the evaluation of alternative ai (i = 1, 2, ..., m) with respect to criterion cj (j = 1, 2, ..., n) is donated by 𝐷 = cj (ai)mxn is an interval-valued spherical fuzzy decision matrix, and µi j, νi j and πi j are the membership, non-membership and hesitancy degrees for the ith alternative and jth criterion. step 6. determine the interval-valued spherical fuzzy ideal solution based on score and accuracy values of attributes in decision matrix 𝐷. equations 7 and 8 are used to calculate score and accuracy values. step 7. determine the interval-valued spherical fuzzy optimality function values oi for all alternatives, as well as for the interval-valued spherical fuzzy ideal best solution oo as given in equations 19 and 20. 𝑂 = [𝑎 , b ][c , d ], [e , f ] + [𝑎 , b ][c , d ][e , f ] + ⋯ + [𝑎 , b ][c , d ][e , f ] (19) where 𝑂 gives the optimality function value for the ith alternative and n is the number of criteria. thus, [𝑎 , b ], [c , d ], [e , f ] are the lower and upper degrees of membership, non-membership and hesitancy for the first criterion of the ith alternative and so on. 𝑂 = [𝑎 , b ][c , d ], [e , f ] + [𝑎 , b ][c , d ][e , f ] + ⋯ + [𝑎 , b ][c , d ][e , f ] (20) akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 69 where 𝑂 gives the optimality function value for the ideal best solution and n is the number of criteria. addition operator that is given in equation 2 is used for calculating 𝑂 and 𝑂 . step 8. defuzzify and obtain crisp form of 𝑂 and 𝑂 . then calculate the utility degree ki of an alternative ai by comparing optimality criterion value of that alternative oi with the optimality criterion value of the ideal best solution oo as given in equation 21. 𝐾 = (21) step 9. rank the alternatives according to utility degrees ki of the alternatives. the higher the utility value, the better the alternative. 3. application in this section, the applicability of the interval-valued spherical fuzzy aras model is shown to evaluate seismic risk levels of hospital buildings. in this regard, four hospital buildings (a1, a2, a3, and a4) are evaluated by three civil engineer decision-makers with respect to eight selected criteria. the structure of the mcdm problem (figure 2) and a brief summary of the selected criteria are given below. figure 2. the structure of the mcdm problem. 3.1. criteria set previous earthquake events have shown that there is a need for the harmony of structural and non-structural elements for the seismic performance of buildings [26]. on the other hand, in seismic risk analysis studies, assessments of critical public institutions such as hospitals in terms of nonstructural factors are very critical [27]. in this study, based on comprehensive literature analysis, five structural (c1, c2, c3, c4, and c8); and three non-structural criteria (c5, c6, and c7) were determined for the seismic risk analysis study as follows: c1 architectural design. architectural design of buildings plays an important role in structural performance and may lead to complex behaviors under earthquake loadings [28]. c2 underlying soil condition. underlying soil condition significantly affects structural performances of buildings and should be considered in seismic vulnerability assessments [29]. c3 foundation system. foundation system is also an important part of a structure which transmits vertical loadings to the soil and may be responsible for failure under seismicity [30]. c4 construction quality. construction material quality is a critical criterion when evaluating seismic vulnerability and it may lead to severe damages after an earthquake if its quality is not high enough [31]. c5 building occupancy. building occupancy is another criterion in determining the seismic vulnerability level of the buildings and can be determined by qualitative assessments of experts [32]. c6 emergency exit system. emergency exit systems in these buildings are critical for exit and access after an earthquake [33]. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 70 c7 falling hazards. falling hazards such as external falling hazards, i.e., unreinforced masonry parapets, chimneys and exterior surface cladding elements; and internal falling hazards, i.e., chemicals in laboratories can fall or block exit ways under earthquake [34]. c8 shear wall system. shear wall system also has an important role in the design of buildings against seismic hazards since these walls make structures more resistant against lateral loads by providing efficient bracing system [35]. 3.2. numerical solution: interval-valued spherical fuzzy aras set step 1. four hospital buildings are evaluated with respect to eight criteria by three civil engineer experts having same experience levels. thus, each decision-maker has an equal weight as 1/3. linguistic evaluations of alternatives are given in table 2. table 2. linguistic evaluations of alternatives. c1 c2 c3 c4 c5 c6 c7 c8 dm1 a1 ei smi smi smi vhi ei ei ei a2 smi vli vhi hi vhi hi li vhi a3 vli vhi smi hi vhi hi vli sli a4 sli smi sli smi ei ei ali sli dm2 a1 smi vhi smi vhi vhi ei ei sli a2 smi sli vhi hi hi vhi sli vhi a3 sli hi hi vhi hi smi ali ei a4 sli sli smi smi smi ei ali ei dm3 a1 ei smi hi smi ami li sli sli a2 vhi li vhi hi hi sli li hi a3 ei smi smi ami sli li li sli a4 vli smi sli ei sli li vli ei step 2. interval-valued spherical fuzzy alternative evaluation matrix is constructed as in table 3. table 3. interval-valued spherical fuzzy alternative evaluation matrix. c1 c2 … c8 a1 ([0.52, 0.58], [0.40, 0.49], [0.29, 0.38]) ([0.55, 0.65], [0.25, 0.30], [0.25, 0.30]) … ([0.31, 0.37], [0.52, 0.62], [0.27, 0.33]) a2 ([0.61, 0.71], [0.22, 0.27], [0.22, 0.27]) ([0.20, 0.25], [0.66, 0.77], [0.20, 0.25]) … ([0.72, 0.82], [0.17, 0.22], [0.17, 0.22]) a3 ([0.27, 0.32], [0.61, 0.72], [0.23, 0.29]) ([0.64, 0.75], [0.20, 0.25], [0.21, 0.25]) … ([0.31, 0.37], [0.52, 0.62], [0.27, 0.33]) a4 ([0.21, 0.26], [0.63, 0.74], [0.21, 0.26]) ([0.42, 0.50], [0.39, 0.47], [0.25, 0.30]) … ([0.40, 0.45], [0.49, 0.59], [0.28, 0.37]) steps 3 and 4. linguistic evaluations of eight criteria by three decision-makers and corresponding interval-valued spherical fuzzy criterion weight matrix are given in table 4. step 5. interval-valued spherical fuzzy decision matrix is obtained as in table 5. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 71 table 4. linguistic evaluations of criteria and corresponding criterion weights. c1 c2 … c8 dm1 hi ami … hi dm2 hi vhi … vhi dm3 smi ami … smi criterion weights ([0.61, 0.72], [0.22, 0.27], [0.22, 0.27]) ([0.82, 0.92], [0.12, 0.17], [0.10, 0.17]) … ([0.64, 0.75], [0.20, 0.25], [0.21, 0.25]) table 5. decision matrix. c1 c2 … c8 a1 ([0.32, 0.42], [0.45, 0.54], [0.34, 0.42]) ([0.45, 0.60], [0.28, 0.34], [0.26, 0.33]) … ([0.20, 0.27], [0.55, 0.65], [0.31, 0.37]) a2 ([0.37, 0.51], [0.31, 0.37], [0.30, 0.36]) ([0.16, 0.23], [0.67, 0.77], [0.21, 0.26]) … ([0.46, 0.61], [0.26, 0.33], [0.26, 0.32]) a3 ([0.16, 0.23], [0.64, 0.74], [0.28, 0.33]) ([0.53, 0.68], [0.24, 0.30], [0.22, 0.30]) … ([0.20, 0.27], [0.55, 0.65], [0.31, 0.37]) a4 ([0.13, 0.19], [0.66, 0.76], [0.26, 0.30]) ([0.34, 0.46], [0.40, 0.49], [0.26, 0.33]) … ([0.26, 0.34], [0.52, 0.62], [0.32, 0.40]) steps 6-9. optimality criterion values for the alternatives and for the ideal best solution, and appraisal scores of alternatives are obtained as in table 6. table 6. optimality criterion values, appraisal scores and final rankings of alternatives. optimality criterion value appraisal score a1 ([0.77, 0.91], [0.00, 0.01], [0.46, 0.36]) 0.772 a2 ([0.84, 0.95], [0.00, 0.00], [0.36, 0.25]) 0.906 a3 ([0.81, 0.93], [0.00, 0.01], [0.39, 0.29]) 0.841 a4 ([0.62, 0.77], [0.01, 0.04], [0.54, 0.51]) 0.487 ideal best ([0.90, 0.98], [0.00, 0.00], [0.30, 0.17]) final ranking a2 > a3 > a1 > a4 a2 has the highest degree of utility which means seismic risk level of hospital building a2 is the most critical. 3.3. sensitivity, comparative and rank reversal analyses sensitivity analysis. to validate our model, sensitivity, comparative, and rank reversal analyses were conducted. for sensitivity analysis of criteria, one-at-a-time analysis was conducted with respect to each criterion, and for this purpose, reference criterion weight values, namely absolutely most important (ami), equal importance (ei) and absolutely least important (ali) were defined to see the effect of the change of criterion weights on the final rankings. by assigning each reference value to each criterion, the proposed model was executed and the scores and the final rankings were calculated. 24 scenarios obtained in this way are given in figure 3. the ranks of a4 and a1 are the same in all scenarios, whereas the rankings of a3 and a2 vary in four of the 24 scenarios. in general, however, the suggested technique produced rather consistent results for a variety of criteria weight situations. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 72 figure 3. sensitivity analysis results for criterion weights. the second analysis was conducted to check the effect of decision-maker weights on the final rankings. in this context, 10 different scenarios were generated, provided that the sum of the weights of three decision-makers is one. the proposed model was executed for these 10 scenarios by assigning extreme values to the decision-maker weights as much as possible. the ranking of alternatives obtained in this way is shown in figure 4. figure 4. sensitivity analysis results for decision-maker weights. in all scenarios, the alternative ranks are the same. it can be said that, the proposed methodology has a robust behavior under different extreme decision-maker weight distribution cases. comparative analysis. for comparing the validity of our method, we compared the appraisal scores of the proposed method with single-valued spherical fuzzy aras [36]. both techniques provided the same outcomes in terms of ranking. it can be said that the proposed interval-valued spherical fuzzy aras gives highly correlated results with single-valued approach. rank reversal analysis. when one of the alternatives is eliminated or a new alternative is introduced to an mcdm model, the ranking result of the model may change, which is known as a rank reversal phenomenon. rank reversal may occur in commonly used mcdm techniques, which can degrade the dependability of the applied method by causing the model to provide inaccurate findings. in this work, the rank reversal capacity of the suggested interval-valued spherical fuzzy aras was also investigated. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 73 table 9. comparative analysis results. proposed interval-valued spherical fuzzy aras single-valued spherical fuzzy aras a1 0.772 0.680 a2 0.906 0.876 a3 0.841 0.786 a4 0.487 0.453 ranking a2 > a3 > a1 > a4 a2 > a3 > a1 > a4 each alternative was successively eliminated from the full model, and the appraisal scores were computed by executing the algorithm. by deleting the alternatives, a total of eight scenarios were obtained. in the first scenario, alternative a1 was deleted from the whole model, and the appraisal scores and final rankings of other alternatives were obtained. alternatives a2, a3, and a4 were deleted one by one in the second, third, and fourth scenarios. in the same way, only alternatives a1 and a2 were evaluated in the fifth scenario; alternatives a3 and a4 were evaluated in the sixth scenario; alternatives a1 and a4 were evaluated in the seventh scenario, and alternatives a2 and a3 were evaluated in the eighth scenario. table 10 shows the appraisal scores and final rankings that were obtained in this manner. table 10. rank reversal analysis results. scenario a1 a2 a3 a4 ranking full model 0.680 0.876 0.786 0.453 a2 > a3 > a1 > a4 1 0.877 0.787 0.453 a2 > a3 > a4 2 0.831 0.962 0.554 a3 > a1 > a4 3 0.725 0.935 0.483 a2 > a1 > a4 4 0.680 0.876 0.786 a2 > a3 > a1 5 0.725 0.935 a2 > a1 6 0.995 0.573 a3 > a4 7 0.992 0.661 a1 > a4 8 0.877 0.787 a2 > a3 4. conclusion and future remarks in this study, a new decision support model is established by integrating aras methodology with interval-valued spherical fuzzy sets. the applicability of the model is illustrated by assessing the seismic risk levels of hospital structures. for this aim, three civil engineer decision-maker experts evaluated four hospital structures based on eight criteria derived from the literature. mathematical operations of interval-valued spherical fuzzy sets were used in the model development. comparative, sensitivity, and rank reversal analyses are given to demonstrate the stability and validity of the proposed model. the main theoretical contributions of this study are as follows: the aras method is extended with interval-valued spherical fuzzy sets which provide users an increased fuzziness modeling capacity. the interval-valued spherical fuzzy addition operation was used to construct optimality akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 74 criterion values. sensitivity analysis for criterion and decision-maker weights distribution scenarios are provided. this study has also some practical implications: a fuzzy decision-making framework is provided for site-specific seismic risk assessment of hospital buildings. the suggested model can be utilized by hospital administrators to analyze the seismic risk levels of hospital structures quickly and cost-effectively for use in retrofitting and rebuilding planning. as a limitation of the study, the following can be said: in the model development, degree of alternative utility was calculated by comparing optimality criterion value of an alternative with the optimality criterion value of the ideal best solution. however, in this final step, the model was converted to a crisp environment due to the absence of spherical fuzzy division operator. the addition operator was used to calculate the spherical fuzzy optimality function in this work. in the future, geometric and arithmetic mean operators might be utilized to do this. for seismic risk assessment, a machine learning reinforced mcdm model may also be created. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] cimellaro, g. p., reinhorn, a. m., & bruneau, m. (2010). seismic resilience of a hospital system. structure and infrastructure engineering, 6(1-2), 127-144. doi: https://doi.org/10.1080/15732470802663847. [2] perrone, d., aiello, m. a., pecce, m., & rossi, f. (2015, august). rapid visual screening for seismic evaluation of rc hospital buildings. in structures (vol. 3, pp. 57-70). elsevier. doi: https://doi.org/10.1016/j.istruc.2015.03.002. [3] delavar, m. r., & sadrykia, m. (2020). assessment of enhanced dempster-shafer theory for uncertainty modeling in a gis-based seismic vulnerability assessment model, case study tabriz city. isprs international journal of geo-information, 9(4), 195. doi: https://doi.org/10.3390/ijgi9040195. [4] kutlu gündoğdu, f., & kahraman, c. (2019a). spherical fuzzy sets and spherical fuzzy topsis method. journal of intelligent & fuzzy systems, 36(1), 337-352. doi: https://doi.org/10.3233/jifs-181401. [5] kutlu gündoğdu, f., & kahraman, c. (2019b). a novel fuzzy topsis method using emerging intervalvalued spherical fuzzy sets. engineering applications of artificial intelligence, 85, 307-323. doi: https://doi.org/10.1016/j.engappai.2019.06.003. [6] zavadskas, e. k., & turskis, z. (2010). a new additive ratio assessment (aras) method in multicriteria decision-making. technological and economic development of economy, 16(2), 159-172. doi: https://doi.org/10.3846/tede.2010.10. [7] karagöz, s., deveci, m., simic, v., & aydin, n. (2021). interval type-2 fuzzy aras method for recycling facility location problems. applied soft computing, 102, 107107. doi: https://doi.org/10.1016/j.asoc.2021.107107. [8] jaukovic jocic, k.; jocic, g.; karabasevic, d.; popovic, g.; stanujkic, d.; zavadskas, e.k.; thanh nguyen, p. a novel integrated piprecia–interval-valued triangular fuzzy aras model: e-learning course selection. symmetry 2020, 12, 928. doi: https://doi.org/10.3390/sym12060928. [9] heidary dahooie, j., estiri, m., zavadskas, e. k., & xu, z. (2021). a novel hybrid fuzzy dea-fuzzy aras method for prioritizing high-performance innovation-oriented human resource practices in high tech sme’s. international journal of fuzzy systems, 1-26. doi: https://doi.org/10.1007/s40815-021-01162-2. [10] büyüközkan, g., & güler, m. (2020). smart watch evaluation with integrated hesitant fuzzy linguistic sawaras technique. measurement, 153, 107353. doi: https://doi.org/10.1016/j.measurement.2019.107353. [11] yildirim, b. f., & mercangoz, b. a. (2020). evaluating the logistics performance of oecd countries by using fuzzy ahp and aras-g. eurasian economic review, 10(1), 27-45. doi: https://doi.org/10.1007/s40822-01900131-3. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 75 [12] gul, s. (2021). fermatean fuzzy set extensions of saw, aras, and vikor with applications in covid-19 testing laboratory selection problem. expert systems, e12769. doi: https://doi.org/10.1111/exsy.12769. [13] gul, m., & ak, m. f. (2021). a modified failure modes and effects analysis using interval-valued spherical fuzzy extension of topsis method: case study in a marble manufacturing facility. soft computing, 25(8), 6157-6178. doi: https://doi.org/10.1007/s00500-021-05605-8. [14] aydin & gundogdu f. k. (2021) interval-valued spherical fuzzy multimoora method and its application to industry 4.0. in decision making with spherical fuzzy sets (pp. 295-322). springer, cham. doi: https://doi.org/10.1007/978-3-030-45461-6 13. [15] duleba, s., kutlu gündoğdu, f., & moslem, s. (2021). interval-valued spherical fuzzy analytic hierarchy process method to evaluate public transportation development. informatica, 1-26. doi: https://doi.org/10.15388/21-infor451. [16] jin, c., ran, y., wang, z., & zhang, g. (2021). prioritization of key quality characteristics with the threedimensional hoq model-based interval-valued spherical fuzzy-oreste method. engineering applications of artificial intelligence, 104, 104271. doi: https://doi.org/10.1016/j.engappai.2021.104271. [17] lathamaheswari, m., nagarajan, d., garg, h., & kavikumar, j. (2021). interval valued spherical fuzzy aggregation operators and their application in decision making problem. in decision making with spherical fuzzy sets (pp. 27-51). springer, cham. doi: https://doi.org/10.1007/978-3-030-45461-6_2. [18] ozceylan, e., ozkan, b., kabak, m., & dagdeviren, m. (2020, july). a survey on spherical fuzzy sets and clustering the literature. in international conference on intelligent and fuzzy systems (pp. 87-97). springer, cham. doi: https://doi.org/10.1007/978-3-030-51156-2_12. [19] adafer, s., & bensaibi, m. (2017). seismic vulnerability classification of roads. energy procedia, 139, 624630. doi: https://doi.org/10.1016/j.egypro.2017.11.263. [20] jena, r., pradhan, b., & beydoun, g. (2020). earthquake vulnerability assessment in northern sumatra province by using a multi-criteria decision-making model. international journal of disaster risk reduction, 46, 101518. doi: https://doi.org/10.1016/j.ijdrr.2020.101518. [21] kumlu, k. b. y., & tüdeş, ş. (2019). determination of earthquake-risky areas in yalova city center (marmara region, turkey) using gis-based multicriteria decision-making techniques (analytical hierarchy process and technique for order preference by similarity to ideal solution). natural hazards, 96(3), 9991018. doi: https://doi.org/10.1007/s11069-019-03583-7. [22] cremen, g., & galasso, c. (2021). a decision-making methodology for risk-informed earthquake early warning. computer-aided civil and infrastructure engineering, 36(6), 747-761. doi: https://doi.org/10.1111/mice.12670. [23] alizadeh, m., hashim, m., alizadeh, e., shahabi, h., karami, m. r., beiranvand pour, a. & zabihi, h. (2018). multi-criteria decision making (mcdm) model for seismic vulnerability assessment (sva) of urban residential buildings. isprs international journal of geo-information, 7(11), 444. doi: https://doi.org/10.3390/ijgi7110444. [24] tyagi, a., tiwari, r. k., & james, n. (2021). gis-based landslide hazard zonation and risk studies using mcdm. in local site effects and ground failures (pp. 251-266). springer, singapore. doi: https://doi.org/10.1007/978-981-15-9984-2_22. [25] zhang, f., li, s., wang, j., & zhang, j. (2020). effects of fault rupture on seismic responses of fault-crossing simply-supported highway bridges. engineering structures, 206, 110104. doi: https://doi.org/10.1016/j.engstruct.2019.110104. [26] gabbianelli, g., perrone, d., brunesi, e., & monteiro, r. (2020). seismic acceleration and displacement demand profiles of non-structural elements in hospital buildings. buildings, 10(12), 243. doi: https://doi.org/10.3390/buildings10120243. [27] dixit, a. m., yatabe, r., guragain, r., dahal, r. k., & bhandary, n. p. (2014). non-structural earthquake vulnerability assessment of major hospital buildings in nepal. georisk: assessment and management of risk for engineered systems and geohazards, 8(1), 1-13. doi: https://doi.org/10.1080/17499518.2013.805629. [28] alecci, v., & de stefano, m. (2019). building irregularity issues and architectural design in seismic areas. frattura ed integrità strutturale, 13(47), 161-168. doi: https://doi.org/10.1016/j.asoc.2020.106322. [29] behnamfar, f., & banizadeh, m. (2016). effects of soil–structure interaction on distribution of seismic vulnerability in rc structures. soil dynamics and earthquake engineering, 80, 73-86. doi: https://doi.org/10.1016/j.soildyn.2015.10.007. akın menekşe and hatice camgöz akdağ / journal of risk analysis and crisis response, 2022, 12(2), 62-76 doi: https://doi.org/10.54560/jracr.v12i2.325 76 [30] adewuyi o, i., & philips o, f. (2019). integrated geophysical and geotechnical methods for prefoundation investigations. journal of geology & geophysics, 08(01). doi: https://doi.org/10.4172/23818719.1000453. [31] kaplan, h., bilgin, h., yilmaz, s., binici, h., & öztas, a. (2010). structural damages of l'aquila (italy) earthquake. natural hazards and earth system sciences, 10(3), 499-507. doi: https://doi.org/10.5194/nhess10-499-2010. [32] ketsap, a., hansapinyo, c., kronprasert, n., & limkatanyu, s. (2019). uncertainty and fuzzy decisions in earthquake risk evaluation of buildings. engineering journal, 23(5), 89-105. doi: https://doi.org/10.4186/ej.2019.23.5.89. [33] li, b., & mosalam, k. m. (2013). seismic performance of reinforced-concrete stairways during the 2008 wenchuan earthquake. journal of performance of constructed facilities, 27(6), 721-730. doi: https://doi.org/10.1061/(asce)cf.1943-5509.0000382. [34] de angelis, a., & pecce, m. (2015). seismic nonstructural vulnerability assessment in school buildings. natural hazards, 79(2), 1333-1358. [35] chandurkar, p. p., & pajgade, d. p. (2013). seismic analysis of rcc building with and without shear wall. international journal of modern engineering research, 3(3), 1805-1810. [36] menekşe, a., & camgöz akdağ, h. (2021, august). seismic vulnerability assessment using spherical fuzzy aras. in international conference on intelligent and fuzzy systems (pp. 733-740). springer, cham. doi: https://doi.org/10.1007/978-3-030-85577-2_86. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 13, issue 1-3 journal of risk analysis and crisis response, 2023, 13(1), 35-48 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i1.354 35 article digital economy, science and technology innovation and carbon emissions a dynamic analysis of pvar based on provincial panel data chen ye 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: 1443795359@qq.com; tel.: +86-18005159270 received: october 26, 2022; accepted: january 10, 2023; published: march 31, 2023 abstract: this paper investigates the dynamic relationship between the digital economy, science and technology innovation and carbon emissions using a panel vector autoregressive model with data from 30 chinese provinces from 2011 to 2019. the results show that: initial development of the digital economy will lead to an increase of carbon emissions, but the effect will be gradually weakened in later stages and become a reverse inhibitory effect; the growth of digital economy development level and the increase of carbon emissions intensity are mutually granger causative; carbon emissions will inhibit the development of the digital economy in the short term, and will have a significant self-promoting effect; the improvement of science and technology innovation level on the growth of digital economy development level is the driving effect of the improvement of sti level on the growth of digital economy development level is not yet stable. keywords: digital economy; science and technology innovation; carbon emission; pvar model 1. introduction the global climate problem is a huge challenge for all mankind, among which carbon emissions, mainly carbon dioxide, have brought a huge negative impact on human production and life. in the face of climate change challenges, china has put forward the goal of achieving peak carbon and carbon neutrality by 2030 and 2060. in october 2021, the central committee of the communist party of china (cpc) and the state council issued the opinions on completely and accurately implementing the new development concept and doing a good job of peak carbon and carbon neutrality, which specifies the goals of green low-carbon development and green transformation, aiming to implement the new development concept and achieve peak carbon and carbon neutrality. at present, china is in the stage of rapid development of its digital economy, with the continuous promotion of digital process, which relies on the wide application of digital technology in production and life. the digital economy is gradually becoming a strong support to achieve the goal of reduced carbon emissions, and the development of the digital economy is inseparable from the key factor of scientific and technological innovation, how to drive the development of digital economy through scientific and technological innovation to reduce carbon emissions is a key issue to be tested empirically. chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 36 therefore, it is of great significance to use the digital economy as the main tool with science and technological innovation as the driving engine to reduce carbon emissions in order to achieve the strategic goal of carbon peaking and carbon neutrality. this paper uses the pvar model to analyze the data of 30 provinces in china except tibet, hong kong, macao and taiwan to investigate the interaction between the development level of the digital economy, the level of science and technology innovation and carbon emissions, which has important theoretical significance and practical value to improve the development of the digital economy, promote science and technology innovation and achieve peak carbon and carbon neutrality. 2. literature review and theoretical analysis 2.1. digital economy and carbon emission in the context of the digital economy, many scholars have studied the relationship between the digital economy and carbon emissions (miao et al., 2022) [1]. xie et al. (2022) [2] have shown that the digital economy significantly suppresses carbon emissions through an empirical analysis of the mediating effect model, and the industrial structure upgrading plays an obvious mediating role in the transmission mechanism of the digital economy on carbon emissions. li and wang (2022) [3] conduct an empirical study on the development of digital economy and spatial carbon emission reduction through a dynamic spatial durbin model, which shows that the digital economy has an inverted u-shaped local emission reduction effect of "promoting and then suppressing" and a ushaped spatial spillover emission reduction effect of "suppressing and then promoting". meanwhile, xie (2022) [4] and xu et al. (2022) [5] found that there is obvious spatial heterogeneity in the effect of digital economy on regional carbon emissions, and the emission reduction effect of digital economy is more significant in eastern regions. 2.2. digital economy and science and technology innovation science and technology innovation provides an important driving force for the development of digital economy, and at the same time, the output of science and technology innovation is inseparable from the support of digital technology. pu (2022) [6] and lai et al. (2022) [7] found that the coupling and coordination degree of digital economy and science and technology innovation in china shows an overall development trend from low coupling to abrasion to high coupling, while there are obvious differences in the coupling degree between regions, overall, the eastern region is superior to other regions. zhao and li (2022) [8] conduct an empirical study on the relationship between science and technology innovation output and digital economy in the yellow river basin, and the research results shows that the development of the digital economy has a significant impact on technological innovation, and its spatial spillover effect is very significant. wang et al. (2022) [9] investigate the driving effect of regional science and technology innovation level on digital economy development through panel quantile regression, and found that the influence of regional science and technology innovation level on digital economy development has a u-shaped pattern and a certain regional heterogeneity. 2.3. science and technology innovation and carbon emission chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 37 in the process of achieving the strategic goals of carbon peaking and carbon neutral, it is always inseparable from the supporting role of science and technology innovation, which is an important tool to optimize the energy structure, improve the energy utilization rate and promote the upgrading and adjustment of industrial structure (wang and hu, 2022) [10]. sun and xue (2022) [11] used the improved stirpat model to analyze the impact of science and technology innovation on carbon emission efficiency, and found that science and technology innovation can significantly improve carbon emission efficiency by improving energy utilization efficiency. cheng et al. (2019) [12] used the gini coefficient, spatial autocorrelation and panel data to study the spatial and temporal variation characteristics of science and technology innovation and carbon emission efficiency in countries along the belt and road, and in general, science and technology innovation can effectively improve carbon emission efficiency in countries along the belt and road. wang and cheng (2020) [13] analyzed the impact of global science and technology innovation on carbon productivity in a global sample of 118 countries and showed that science and technology innovation has an important role in promoting carbon productivity mainly by promoting structural optimization, reducing energy consumption and improving energy efficiency. 2.4. theoretical analysis of digital economy to promote carbon reduction 2.4.1. promote technological innovation and improve energy utilization efficiency at present, china's industrial structure is still dominated by energy-intensive industries, with a high degree of dependence on energy. first, china's energy consumption is mainly coal, which ranks first in the world. secondly, china's energy utilization efficiency is not high and lacks energy technology innovation. finally, fossil energy accounts for a large part of china's energy structure, and the degree of development of new and renewable energy is still insufficient. on the one hand, the digitalization of industry can help existing enterprises use digital technology to complete the transformation of production factors, reduce the dependence on traditional resources, improve the allocation efficiency of energy, and thus promote the reduction of carbon emissions. on the other hand, with the continuous promotion of digital industrialization, digital technologies such as big data and artificial intelligence can be used to realize the development of new energy and upgrade the existing energy structure, build a perfect and reasonable energy structure, and promote the sustainable development of new energy (guo et al., 2022 [14]; xu and ma, 2022 [15]). 2.4.2. improve the level of regulation of carbon emissions and reduce the negative externalities of carbon emissions the accounting and property rights of carbon emissions are complicated, so the government cannot effectively regulate the carbon emissions of enterprises, so enterprises are not willing to pay for the negative externalities caused by carbon emissions, which makes carbon emissions remain high. the negative externalities of carbon emissions can be reduced by clarifying property rights (hu and jin, 2022) [16]. in addition, digital technology can help to establish a carbon emission standard accounting system, through which government departments can calculate the carbon emission index of different enterprises, avoiding the validity of the accounting due to subjective judgment, and at the same time implementing corresponding rewards and penalties for enterprises according to the accounting situation, thus making the administrative control of carbon emission more professional chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 38 and flexible. through deep mining, big data and other digital technologies, government departments can effectively grasp the carbon emission information of enterprises and alleviate the information asymmetry between the government and enterprises, to regulate carbon emission at the macro level (hu and jin, 2022) [16]. 3. model construction and data sources 3.1. model setting the panel vector autoregressive model (pvar model) was first proposed by holtz-eakin et al. in 1988, it inherits the advantages of the vector autoregressive (var model) model, treats each variable as an endogenous variable, and analyzes the interaction effects of each variable and it lags variables on other variables in the model. the pvar model can effectively solve the problem of individual heterogeneity with the use of panel data. based on this, the pvar model is selected to study the dynamic relationship between digital economy, science and technology innovation and carbon emissions, and the pvar model expression is as follows. 𝑌 =∝ + 𝜌 𝑌 , + 𝜀 , (1) among them, 𝑌 = [𝐷𝐸𝐷𝐼,𝑇𝐼,𝐶𝐸𝐼] is a column vector, including the level of digital economy development (dedi), scientific and technological innovation (ti) and carbon emissions (cei). to ensure the stability of each variable, the first order difference method is used to deal with dedi, ti and cei; 𝜀 , represents a random disturbance term. 3.2. data sources and descriptions the three variables of digital economy development level, science and technology innovation level, and carbon emission are denoted as dedi, ti and cei, respectively. descriptive statistics of each variable are shown in table 1. to reduce the effect of heteroskedasticity, the three variables of digital economy development level, science and technology innovation level, and carbon emission are logarithmized. table 1. descriptive statistics of each variable. variable name variable description sample size average value standard deviation minimum value maximum value lndedi development level of digital economy 270 0.2873 0.0067 0.0745 0.6395 lnti scientific and technological innovation level 270 0.1657 0.0085 0.0091 0.6116 lncei carbon emissions 270 0.9912 0.0267 0.3014 2.6851 3.2.1. development level of digital economy this paper measures the development level of digital economy from two dimensions: "digital industrialization" and "digitalization of industry", "digital industrialization" refers to the use of digital chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 39 technology to bring about products and services, such as cloud photo albums, cloud disks, taxi software, digital tv, etc. "digitalization of industry" refers to the use of digital technology to bring about an increase in the quantity and efficiency of production in pre-existing industries, such as enterprises using industrial big data to help optimize their industrial structure, in which the digital financial inclusion index compiled by guo feng et al. (https://www.idf.pku.edu.cn/xz/272857.htm) is chosen to represent the digital financial inclusion index. this paper uses the entropy weight method to calculate the indicators of digital economy development, to measure the comprehensive indicators of digital economy development, and the corresponding indicators are shown in table 2. table 2. indicator system of digital economy development level. level i indicators secondary indicators third level indicators index weights development level of digital economy digital industrialization internet penetration rate (positive) 0.079 number of internet related employees (positive) 0.315 internet related output (positive) 0.423 number of mobile internet users (positive) 0.091 industrial digitalization digital inclusive financial index (positive) 0.094 3.2.2. science and technology innovation level table 3. index evaluation system of science and technology innovation level of each province in china. level i indicators secondary indicators third level indicators units indicator weights science and technology innovation innovation input r&d personnel full time equivalent people 0.0790 r&d expenditure intensity % 0.0522 expenditure on technology acquisition and technological transformation of industrial enterprises above the scale million yuan 0.0619 number of r&d project topics item 0.0561 innovation output number of domestic three kinds of patent applications authorized piece 0.1087 main business income of high-tech industry billion yuan 0.1287 the number of new product research and development of high-tech enterprises item 0.1198 technology market turnover million yuan 0.1553 innovation environment number of r&d institutions piece 0.0287 ratio of local financial expenditure on science and technology to total financial expenditure % 0.0503 average number of students enrolled in higher education institutions per 100,000 people people 0.0191 number of enterprises in high technology industry piece 0.1085 gdp per capita yuan 0.0317 chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 40 based on the study of liu et al. (2022) [17] and chen and cai (2018) [18], this paper uses the entropy method to measure the science and technology innovation level of each province (city and autonomous region) in three dimensions: input, output and environment. the main indicators of innovation input are the elements of r&d personnel, r&d funding and the number of r&d projects; the main indicators of innovation output are the number of patent applications, new product development and sales and technology market turnover; the external conditions of science and technology innovation are the important indicators of innovation environment (see table 3). 3.2.3. carbon emissions in view of the lack of carbon emission data for each province in china in various statistical yearbooks, this paper, with reference to existing research results, adopts the current internationally applicable measurement method to project co2 emissions according to the ipcc 2006 guidelines for national greenhouse gas inventories. the calculation method is as follows. 𝐶 = 𝐶 = 𝐸 × 𝜇 × ℎ (2) where 𝐸 denotes the total consumption of the 𝑖th type of energy source, and seven major energy sources, namely coal, coke, gasoline, kerosene, diesel, fuel oil and natural gas, are selected for measurement in this paper; 𝜇 denotes the discounted standard coal coefficient; and ℎ denotes the carbon emission coefficient. the specific values of 𝜇 and ℎ are shown in table 4. table 4. conversion factor of standard coal and carbon emission factor of major energy species. energy varieties coal coke gasoline kerosene diesel fuel oil natural gas discount factor for standard coal 𝜇 0.7143 0.9714 1.4714 1.4714 1.4571 1.4286 1.2150 carbon emission factor ℎ 2.492 2.977 1.988 2.051 2.167 2.219 2.162 4. analysis of the empirical results 4.1. unit root test for panel data table 5. unit root test results for lndedi, lnti and lncei. test method statistic/p-value lndedi lnti lncei llc t -3.3118 -12.6116 -11.3495 p-value 0.0005 0.0000 0.0000 ips z 0.8637 -5.2908 -1.1664 p-value 0.8061 0.0000 0.1217 adf z 5.3419 -7.8935 1.7030 p-value 1.0000 0.0000 0.9557 pp z 5.9173 -7.7164 1.4179 p-value 1.0000 0.0000 0.9219 test results non-smooth smooth non-smooth chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 41 table 6. unit root test results for d_lndedi, d_lnti and d_lncei. test method statistic/p-value d_lndedi d_lnti d_lncei llc t -16.9488 -10.0748 -15.3267 p-value 0.0000 0.0000 0.0000 ips z -5.3205 -6.3469 -3.1305 p-value 0.0000 0.0000 0.0009 adf z -3.7580 -7.4270 -4.9317 p-value 0.0001 0.0000 0.0000 pp z -7.6136 -12.7327 -6.6126 p-value 0.0000 0.0000 0.0000 test results smooth smooth smooth before using the pvar model for estimation, it is necessary to ensure that all variables have smoothness, otherwise it will lead to the pseudo-regression phenomenon, so this paper uses four tests to check the smoothness of variables, and the test results are shown in table 5 and table 6. from table 5, the variable lnti is smooth, but the variables lndedi and lncei both fail the ips, adf-fisher and pp-fisher tests. therefore, in this paper, the three variables are treated as first-order differences, and as can be seen from table 6, the variables d_lndedi, d_lnti and d_lncei all pass the four tests after first-order differences, so the three variables of digital economy development level, science and technology innovation level and carbon emission are all smooth after first-order differences. 4.2. co-integration test from the perspective of test rigor, this paper adopts three methods to conduct cointegration tests. the results are shown in table 7. the results show that the null hypothesis is rejected in all three tests at the 1% significance level, so it is concluded that there is a long-term cointegration relationship between the level of digital economy development, the level of science and technology innovation and carbon emissions. table 7. results of co-integration relationship test. variables kao-adf pedroni-pp westerlund-adf lnti、lncei、lndedi 4.7620*** (0.0000) -1.9235*** (0.0272) 9.0429*** (0.0000) note: ***, **, * indicate significant at 1%, 5%, 10% confidence level, respectively. table 8. pvar optimal lag order selection. lag aic bic hqic 1 -6.59448* -4.83836* -5.88245* 2 -4.39204 -2.22438 -3.51139 3 -5.96296 -3.24515 -4.85925 4 -2.914 0.585737 -1.5027 5 -6.26354 -1.55127 -4.42031 chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 42 4.3. determining the optimal lag order determining the optimal lag order of the model is a prerequisite for estimating the pvar model. in this paper, the optimal lag order is determined by using the empirical results of the pvar2 package (lian yujun), and the optimal lag order chosen according to the aic, bic, and hqic information criteria can be found in table 8. 4.4. model stability test in order to ensure the rationality of the subsequent impulse response function and anova, the robustness of the model needs to be tested. if all the characteristic roots lie within the unit circle, it indicates that the robustness test is passed, so it can be seen from figure 1 that the pvar model constructed in this paper is robust. figure 1. unit root test. table 9. granger causality test results. equation excluded chi2 df prob>chi2 h_d_lncei h_d_lnti 0.45382 2 0.501 h_d_lncei h_d_lndedi 4.49 1 0.034 h_d_lncei all 5.8953 2 0.052 h_d_lnti h_d_lncei 0.07555 1 0.783 h_d_lnti h_d_lndedi 3.5528 1 0.059 h_d_lnti all 4.505 2 0.105 h_d_lndedi h_d_lncei 8.9563 1 0.003 h_d_lndedi h_d_lnti 0.18943 1 0.663 h_d_lndedi all 9.552 2 0.008 chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 43 4.5. granger causality test the granger causality test was used to determine the causal relationship of each variable and whether the explanatory variables can predict the explained variables. the test results are shown in table 9. from table 9, the d_lncei is the granger cause of d_lndedi, d_lncei is not the granger cause of d_lnti; d_lnti is not the granger cause of d_lncei, d_lnti is not the granger cause of d_lndedi; d_lndedi is the granger cause of d_lncei granger cause of d_lndedi, d_lndedi is not granger cause of d_lnti. 4.6. gmm estimation of pvar model after completing the unit root test and determining the optimal lag order, the helmert transformation was used to eliminate the time effect and fixed effect of the data, and gmm estimation was performed for the three variables, and the results are shown in table 10. table 10. gmm estimation results of pvar model. variables gmm h_d_lndedi h_d_lncei h_d_lnti l.h_d_lndedi b_gmm -0.7368 -0.8945 -6.4703 se_gmm 0.4114 0.4221 3.4327 t_gmm -1.7908 -2.119 -1.8849 l.h_d_lncei b_gmm 0.3016*** 0.3878*** 0.1875 se_gmm 0.1008 0.1403 0.6822 t_gmm 2.9927 2.7642 0.2749 l.h_d_lnti b_gmm 0.0051 -0.0081 -0.4086 se_gmm 0.0118 0.0121 0.0957 t_gmm 0.4352 -0.6737 -4.2678 from the estimation results, it can be seen that when digital economic development growth (d_lndedi) is used as the explanatory variable, the current year's digital economic development growth is negatively affected by the lagged period's digital economic development growth, but this effect is only significant in the national region; the lagged period's carbon emission intensity growth has a significant positive effect on the current period's digital economic development growth, but the effect is not significant in the central region. the positive impact of the increase in the level of science and technology innovation in the lagging period on the growth of the digital economy in the current period is insignificant both nationally and in the eastern, central and western regions. when carbon emissions intensity growth (d_lncei) is used as the explanatory variable, its oneperiod lagged value has a significant positive effect on the current period, but this effect is not significant in the western region; the one-period lagged growth in the level of digital economic development has a negative effect on the current period carbon emissions intensity growth, but this effect is not significant; the one-period lagged increase in the level of science and technology innovation has a non-significant negative effect on the current period carbon emissions intensity growth, but there is a certain degree of negative effect in the central region. chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 44 when using the increase in the level of science and technology innovation (d_lnti) as the explanatory variable, its one-period lagged value has a significant negative effect on the current period, and this effect is significant in the whole country and in the eastern, central and western regions; the one-period lagged increase in the level of digital economic development has a negative effect on the increase in the level of science and technology innovation in the current period, but this effect is not significant; the negative effect of the one-period lagged increase in the intensity of carbon emissions on the increase in the level of science and technology innovation in the current period is significant in the whole country and in the western region. according to the model gmm estimation results, the explanatory power of the growth of digital economy development level and carbon emission growth on the increase of science and technology innovation level is not strong, so this paper focuses on the dynamic influence of carbon emission on the development of digital economy and itself. the paper then goes on to use impulse response and variance decomposition methods to provide an in-depth analysis of the above effects. 4.7. impulse response and variance decomposition the above gmm estimation is a static analysis of the model, and in order to further study the dynamic interaction between the variables, the impulse response plot with a lag of 6 periods is obtained by 200 simulations with the help of monte carlo experiments in this paper, the first to third columns of figure 2 reflect the changes when the growth in the level of development of the digital economy, the increase in the intensity of carbon emissions and the increase in the level of science and technology innovation are hit respectively, the first to third rows show the impact of the increased level of development of the digital economy, the increased intensity of carbon emissions and the increased level of science, technology and innovation, respectively, as shown in figure 2. when the number of lags gradually increases, the impulse response functions of each variable are close to 0, indicating that it is meaningful to study the model. figure 2. graph of impulse response function. chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 45 the following conclusions can be drawn from the impulse response plots in figure 2. firstly, when the growth of digital economic development level, the increase of science and technology innovation level and the increase of carbon emission intensity by its own shock effect will reach a positive maximum in the current period, and then drop sharply to a negative minimum in periods 1 to 2, with the negative effect gradually decreasing until it disappears, indicating that the increase of these three variables has a significant shock effect on itself especially the science and technology innovation level and carbon emission, indicating that these the increase in the intensity of the current period of these variables is likely to lead to a higher intensity level in the next period. secondly, when the growth of the development level of digital economy faces the impact of the increase of carbon emission intensity, the impact is 0 in the current period, and then there is a negative impact, which indicates that there is a certain lag effect of the growth of the development level of digital economy on the increase of carbon emission intensity, and this impact gradually decreases in the later period until it falls to 0. it indicates that in the short term, the increase of carbon emission intensity will crowd out the development of digital economy to a certain extent, but as the digital economy grows, the impact of the increase of carbon emission intensity will be more significant. however, with the continuous improvement and maturity of the digital economy, this inhibiting effect gradually disappears. on the contrary, when the carbon emission intensity increases in the face of the impact of the growth of the digital economy, the impact remains zero in the current period, then the positive impact starts to appear in the first period, but becomes negative in the second period, and then gradually tends to zero until the fluctuation disappears, indicating that at the early stage of the development of the digital economy, it is necessary to purchase industries by increasing the exploitation of resources and energy consumption, thus promoting the increase of carbon emission intensity. however, with the development of digital economy, the optimization and adjustment of the industrial structure of digital economy, digital technology can effectively improve the utilization rate of resources and alleviate the problem of information asymmetry, which has a restraining effect on the increase of carbon emission intensity. thirdly, when the growth of digital economy development level faces the impact of the improvement of science and technology innovation level, the impact is 0 in the current period, then there is a negative impact in the first period, after which the positive and negative impacts occur alternately, and the degree of impact is very small, and finally tends to 0. this indicates that the impact of the improvement of science and technology innovation level on the development level of digital economy is non-linear, and the degree of impact is not significant, indicating that the current science and technology innovation fails to effectively promote the development of digital economy. in order to better reflect the degree of influence of the endogenous variables on the random disturbance term in the model and analyze the relative contribution degree of the structural variables to the changes of other variables in the model, this paper performs variance decomposition, and the results of variance decomposition are shown in table 11. the variance decomposition results are shown in table 11. according to table 11, its own contribution to the growth in the level of development of the digital economy is the highest at 92.0%, followed by the contribution of science, technology and innovation at 67.9% and finally carbon emissions at 39.4%; the contribution to the increase in the intensity of carbon emissions comes mainly from itself, at 60.3%, with the level of development of the digital economy contributing 7.6% and science, technology and innovation contributing 3.9%; for the increase in the level of science and technology innovation, its own chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 46 contribution is 28.2%, the level of development of the digital economy contributes 4.0% and carbon emissions contribute 3.0%. table 11. results of variance decomposition of each variable. variables s d_lndedi d_lncei d_lnti d_lndedi 10 0.920 0.076 0.004 d_lncei 10 0.394 0.603 0.003 d_lnti 10 0.678 0.039 0.282 d_lndedi 20 0.920 0.076 0.004 d_lncei 20 0.394 0.603 0.003 d_lnti 20 0.679 0.039 0.282 d_lndedi 30 0.920 0.076 0.004 d_lncei 30 0.394 0.603 0.003 d_lnti 30 0.679 0.039 0.282 5. conclusion and policy implications this paper constructs a pvar model using provincial panel data from 2011-2019 to empirically study the interaction between digital economy, science and technology innovation and carbon emissions, and the findings show that: 1) the digital economy leads to an increase in carbon emissions at the early stage of development, but changes to a suppressive effect at a later stage. 2) the increase in the level of development of digital economy and the increase in the intensity of carbon emissions are granger causal. 3) carbon emissions will inhibit the development of digital economy in the short term, and will have a significant self-promoting effect. based on the above research, this paper draws the following insights: 1) strengthen the foundation of digital economy development and enhance digital technology research and development. according to the empirical study in this paper, the digital economy cannot suppress the increase of carbon emissions at the early stage of development, so it is necessary to improve the speed and quality of the development of the digital economy and give full play to the key role of digital technology in energy saving and emission reduction. promote a new round of digital infrastructure construction, accelerate the construction of new infrastructure with 5g base stations, big data centers and artificial intelligence platforms as the core, and improve the level of digital infrastructure. accelerate the construction of digital technologies such as big data, cloud computing, internet of things, block chain and artificial intelligence, and use digital technologies to help enterprises achieve green transformation. 2) strengthen digital science and technology innovation and improve the support of science and technology innovation. on the one hand, it is necessary to increase the intensity of investment in scientific and technological research and development, overcome the difficulties at the digital technology level, use digital technology to help enterprises achieve scientific planning and efficient chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 47 allocation of resources, and alleviate the mismatch between innovation input and output. on the other hand, the government level needs to play a good regulatory and control function to create a fair competitive link from multiple directions and angles to motivate enterprises to carry out digital technology innovation. 3) improve the speed and quality of digital economy development, and promote the digital economy to empower carbon emission reduction. avoid the digital economy falling into the traditional way of rough development, focus on low-carbon control of digital economy infrastructure, and promote the transformation of smart city development with developed cities that have a solid digital economy foundation as the first. give full play to the role model of digital economy enterprises in the field of energy saving and emission reduction, promote the continuous improvement of carbon emission information disclosure system, and at the same time implement effective carbon emission incentives and penalties for enterprises to fill the loopholes for enterprises to make use of information asymmetry and gain. funding: this research received no external funding. conflicts of interest: the author declares no conflict of interest. references [1] miao l j, chen j, fan t z, lv y q. impact of digital economy development on carbon emissions--analysis of panel data based on 278 prefecture-level cities[j]. nanfang finance, 2022(02):45-57. [2] xie w q, gao k, yu j f. digital economy, industrial structure upgrading and carbon emission[j]. statistics and decision making, 2022(17):114-118. doi: https://doi.org/10.13546/j.cnki.tjyjc.2022.17.022. [3] li z g, wang j. how does the development of digital economy in the context of economic agglomeration affect spatial carbon emissions?[j]. journal of xi'an jiaotong university (social science edition), 2022,42(05):8797. doi: https://doi.org/10.15896/j.xjtuskxb.202205009. [4] xie y f. the effect of digital economy on regional carbon emission intensity and its mechanism of action[j]. contemporary economic management, 2022,44(02):68-78. doi: https://doi.org/10.13253/j.cnki.ddjjgl.2022.02.008. [5] xu w x, zhou j p, liu c j. spatial effects of digital economy development on urban carbon emissions[j]. geography research, 2022,41(01):111-129. [6] pu g l. measurement of coupled and coordinated development of digital economy and science and technology innovation in china[j]. technology economics and management research, 2022(04):25-29. [7] lai y f, ye l t, xie p j, ma x r. research on the coupling and coordination of regional science and technology innovation and digital economy[j]. science and technology progress and countermeasures, 2022,39(12):31-41. [8] zhao j f, li y. study on the impact of digital economy on science and technology innovation output and spatial spillover effect in the yellow river basin[j]. science and technology management research, 2022,42(09):56-63. [9] wang s p, teng t w, xia q f, bao h. spatial and temporal characteristics of china's digital economy development level and its innovation-driven mechanism[j]. economic geography, 2022,42(07):33-43. doi: https://doi.org/10.15957/j.cnki.jjdl.2022.07.004. [10] wang w j, hu c j. digital finance, technology innovation and carbon emissions-an empirical analysis based on provincial panel data[j]. journal of tianjin university of commerce, 2022,42(03):40-48. doi: https://doi.org/10.15963/j.cnki.cn12-1401/f.2022.03.002. [11] sun j b, xue q. has science and technology innovation improved the efficiency of carbon emissions?[j]. ecological economics, 2022,38(06):22-28+34. [12] cheng y, sun y x, wang x j, yin j z. the impact of global science and technology innovation on carbon productivity and countermeasures[j]. china population-resources and environment, 2019,29(09):30-40. chen ye / journal of risk analysis and crisis response, 2023, 13(1), 35-48 doi: https://doi.org/10.54560/jracr.v13i1.354 48 [13] wang x j, cheng y. study on the mechanism of urbanization's impact on carbon emission efficiency--an empirical analysis based on global panel data of 118 countries[j]. world geography research, 2020,29(03):503511. [14] guo f, yang s g, ren y. digital economy, green technology innovation and carbon emissions--empirical evidence from the city level in china[j]. journal of shaanxi normal university (philosophy and social science edition), 2022,51(03):45-60. doi: https://doi.org/10.15983/j.cnki.sxss.2022.0507. [15] xu h, ma l j. digital economy, resource dependence and green economic development[j]. finance and economics, 2022(01):45-54. doi: https://doi.org/10.19622/j.cnki.cn36-1005/f.2022.01.005. [16] hu y, jin s c. digital technology helps to achieve the goal of "double carbon": theoretical mechanism and practical path[j]. finance and accounting monthly, 2022, no.922(06):111-118. doi: https://doi.org/10.19641/j.cnki.42-1290/f.2022.06.014. [17] liu z h, xu j w, zhang c h. science and technology innovation, industrial structure upgrading and carbon emission efficiency--pvar analysis based on inter-provincial panel data[j]. journal of natural resources, 2022,37(02):508-520. [18] chen q, cai j. country size and technological innovation a comparative study based on countries with different levels of economic development[j]. systems engineering, 2018,36(04):69-77. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word °û?ñ⁄ journal of risk analysis and crisis response, 2022, 12(3), 152-154 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i3.336 152 communication the 10th annual meeting of risk analysis council of china association for disaster prevention was successfully held in beijing tie-fei li 1,* and wang-jing xu 2,3 1 institute of geophysics, china earthquake administration, beijing (100081), china 2 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 3 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: 2fen222@163.com received: september 25, 2022; accepted: september 28, 2022; published: september 30, 2022 the 10th annual meeting of risk analysis council of china association for disaster prevention was successfully held in beijing from september 24 to 25, 2022. the theme of this annual conference was "risk analysis and management of uncoordinated crises". the conference was organized by risk analysis council of china association for disaster prevention, jointly organized by institute of geophysics, china earthquake administration and beijing normal university, and co-organized by journal of catastrophology and city and disaster reduction. the conference was conducted in an online and offline manner and attracted more than 160 experts, scholars and postgraduate students from all over the country. figure 1. opening ceremony of the 10th annual meeting of risk analysis council of china association for disaster prevention. the opening ceremony of the annual conference was held in the grand conference hall of beijing tianhu conference centre on the morning of september 24. the opening ceremony was attended and addressed by mr. chenglei wang, director of the department of risk monitoring and comprehensive disaster reduction of ministry of emergency management, mr. jingbo guan, tie-fei li and wang-jing xu / journal of risk analysis and crisis response, 2022, 12(3), 152-154 doi: https://doi.org/10.54560/jracr.v12i3.336 153 deputy director of the department of seismic damage prevention of china earthquake administration, researcher mr. biao ouyang, secretary of the party committee of institute of geophysics, china earthquake administration, mr. jianliang li, deputy secretary general of china association for disaster prevention, and professor chongfu huang, chairman of risk analysis council of china association for disaster prevention, and other leading guests. the opening ceremony was chaired by researcher sen qiao, executive vice chairman of the specialized committee. at the opening ceremony, the leaders attending the meeting all gave full recognition to the work of risk analysis council over the years. it is hoped that risk analysis council will make efforts to carry out basic research, continuously undertake theoretical innovation, strengthen the transformation of scientific and technological achievements in risk analysis, promote the development of the scientific and technological cause of risk analysis in china, and actively advise and contribute to disaster prevention, mitigation and relief. risk analysis council should carry out the requirements of ministry of emergency management and association for disaster prevention on risk analysis and branch construction, continue to lead the academic exchange of risk analysis theory in china, and strive to promote the development of risk management technology under a high technology platform. figure 2. group photo of some participants at beijing national earth observatory. a total of 11 experts and scholars gave keynote presentations at the conference, focusing on such research hotspots as "the new mission of risk analysis in a changing landscape", "characteristics of natural disaster risk and management strategies in the new era", "public health risk response from the perspective of urban and rural planning", "research on comprehensive risk dynamic evaluation and early warning technology of multi-hazard major agricultural meteorological disasters", "disaster characteristics and scientific and technological support of several recent major earthquakes", "human resource risk factors and response strategies in the context of innovation and development", and "research on risk prevention of buried pipelines in lifeline projects" and so on. at the same time, this annual conference also organized four sessions on "seismic hazard and seismic risk analysis", "agriculture, environment and meteorological disaster risk analysis", "financial risk analysis" and "comprehensive disaster risk analysis". a total of 70 presentations were given in four sessions, which closely followed the characteristics and trends of the times, and focused on theoretical and empirical analysis of mainstream views, reflecting industry dynamics and tie-fei li and wang-jing xu / journal of risk analysis and crisis response, 2022, 12(3), 152-154 doi: https://doi.org/10.54560/jracr.v12i3.336 154 academic frontier directions, with frequent online and offline interactions, which were well received by the participants. the conference is of far-reaching significance to enhance the level of theoretical research on risk analysis, promote the optimization of the training mode of risk analysis talents, strengthen the transformation of scientific and technological achievements of risk analysis, and better serve national needs. in the afternoon of september 25, some of the participants visited the beijing national earth observatory to learn about the exploration plan of china earthquake science array, visit the key laboratories, and watch the rock fracturing and soil dynamics experiments. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). the research of a hail risk evaluation under imperfectly information the research of a hail risk evaluation under imperfect information wang wei 1,2 , zhou chao 2 , du xin 2 1 tianjin meteorology science institute, tianjin 300074, china 2 college of environmental science and engineering , nankai university，tianjin 300071，china abstract based on the situation of the global warming, hail disasters will maybe cause the higher risk of loss in the future. the author established a risk evaluation model of hail disasters in terms of the combination of probabi listic and physical methods. moreover, a nonlinear finite element analysis software (ansys/ls-dyna ) is utilized to simulate the case that hails dash on the steel body and glass material, whose result was one of the parameters of the evaluation model. finally, the article carried out a simulation using the hail risk evaluation model and the monte carlo random simulation method in terms of the historical statistical data in tianjin. the result shows that this model can effectively assess hail disaster risk under the situation of insufficient data and information on hail disasters. keywords: hail disasters, risk evaluation, model, numerical simulation 冰雹灾害的风险评估方法的建立与风险模拟研究 * 王 炜 1,2 周超 2 杜鑫 2 1 天津市气象科学研究所，天津，300074 2 南开大学环境与工程学院，天津，300071 摘要：在全球气候变暖日益加剧的情况下，未来的冰雹灾害造成的各种经济损失风险可能 会更加严重。针对社会经济发展中可能遇到的冰雹灾害的损失风险问题，作者尝试利用概率方法 和物理方法相结合的手段，建立了相应的冰雹灾害损失的评估模型。并且，利用非线性有限元分 析软件 ansys/ls-dyna 对冰雹撞击钢板和玻璃材料进行了数值模拟，使用模拟结果确定冰雹 风险评估模型的损失函数参数。最后，利用天津市的历史统计资料，将蒙特卡罗随机模拟手段和 冰雹风险评估模型相结合，模拟分析了天津地区冰雹灾害的损失风险。分析结果表明，在资料信 息不足的情况下，建立的冰雹灾害风险分析模型可以有效地评估冰雹灾害造成的损失。 关键词：冰雹灾害，风险评估，模型，数值模拟 * 本文得到中国气象局 2010 年软科学重点项目资助（项目编号 2010007） 作者简介：王炜（1965-）, 男（汉族），内蒙古巴彦淖尔盟人，博士，副教授，主要从事风险分析和大气环境数值模拟。 e-mail：wwei356@nankai.edu.cn. 1．引言 冰雹是重要的自然灾害之一。冰雹灾害每 年都给农业、建筑、通讯、电力、交通以及人 民生命财产带来巨大损失。因此，冰雹灾害的 成因和损失情况受到了许多学者的关注 [1-8] 。 据有关资料统计，中国每年因冰雹所造成的经 济损失达几亿元甚至几十亿元。例如，内蒙古 历史年均雹灾造成牲畜伤亡 2818 头，年均造 成人员伤亡 28 人。最严重的一次是在 1966 年 7 月 23 日，鄂尔多斯市鄂托克旗，冰雹砸死 2204 头牲畜，砸伤 25945 头牲畜 [9] 。2006 年 4 月 13-14 日，在美国中西部地区发生的一场冰 雹，造成了大约 18.22 亿美元的损失 [10] 。 在汽车迅速发展的时代，冰雹对汽车的损 害风险也越来越严重。如 2008 年 6 月 22 日， 一场冰雹袭击了德国北部的一个汽车厂。降落 的冰雹导致大约 30,000 新汽车被损害，经济 损失达到 1 亿欧元 [11] 。2010 年 5 月 2 日下午， 中国浙江宁波突降冰雹，降雹时间持续十几分 钟。冰雹过去，部分汽车的引擎盖上被砸出了 一个个小坑，还有不少车辆的挡风玻璃都被砸 坏，更有甚者天窗都被冰雹打碎。 在当前气候变暖的形势下，极端气候条件 下的冰雹灾害风险会更加严重。但是，以前的 科学工作者对冰雹灾害的研究主要集中在冰 雹灾害的成灾机理和冰雹天气预测等方面。由 于对于冰雹灾害的潜在危害认识不足，以前对 于冰雹灾害的风险评估方法研究的研究工作 开展较少。因此，针对冰雹灾害的暴露性和承 灾体的脆弱性等对象，开展定量评估冰雹灾害 的风险的研究工作，对有效地减少冰雹造成的 经济损失具有实际意义。 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 133-141 published by atlantis press copyright: the authors 133 2．冰雹灾害的风险识别 2.1 冰雹灾害的风险成因和风险因素识别 2.1.1 冰雹灾害的风险成因和致灾因子 冰雹是强对流性天气引起的一种剧烈的 气象灾害。一般情况下，冰雹灾害发生的范围 较小、时间较短，但是它由于具有发生突然、 损失强度大的特性，成为严重气象灾害之一。 冰雹灾害的主要风险源是雷暴云中降落 的冰雹颗粒。所以冰雹灾害的风险大小同冰雹 的粒径大小有关。然而，冰雹的粒径大小又同 冰雹的成雹机制密切相关。因为，冰雹形成于 一种同时具有上升气流与下沉气流的有组织 的积雨云中。在这种积雨云中，强烈的上升气 流不仅给雹云输送了充分的水汽，而且支撑着 冰雹粒子在云中的滞留。当冰雹颗粒生长到相 当大程度，上升气流无法支撑冰雹重量时，冰 雹颗粒就会从空中坠落，最后在地面形成了降 雹。因此，冰雹粒径的大小决定于积雨云中上 升气流大小和温度层结等气象条件。 在气象学中，冰雹的严重程度，有时也用 物体的大小来形容。例如，小如绿豆、黄豆， 大似栗子、鸡蛋等形象化词语表示冰雹的形 态。除使用形态分类方法描述冰雹的严重度 外，也可以用定量方法划分降雹的严重度。大 致上，气象学中将降雹分成软雹和冰雹。软雹 的尺寸是 2~5mm，冰雹尺寸大于 5mm。也有 根据冰雹的直径分为三级：（1）轻雹：多数冰 雹直径不超过 0.5cm；（2）中雹：多数冰雹直 径 0.5～2.0cm；（3）重雹：多数冰雹直径 2.0cm 以上。 总之，冰雹危害程度取决于冰雹尺度大 小、冰雹持续时间、冰雹发生频率、冰雹覆盖 区域面积等因素。 2.1.2 冰雹灾害的承灾体 冰雹的发生区域一般在地形复杂的北方 山区及丘陵地区。冰雹对承灾体的伤害与当地 的经济社会情况有密切的关系。在我国经济不 发达时期，冰雹主要是对农业危害很大。猛烈 的冰雹常常毁坏庄稼、损坏房屋、人被砸伤、 打死牲畜等。现在，随着我国经济的发展，城 市公共设施和家庭汽车等成为了新的受灾体。 综合地分析，我国冰雹灾害的主要危害表现在 农作物损失、人员伤亡、牲畜伤亡、汽车损失 和建筑物损失等方面。 2.2 冰雹灾害的风险计算方法 2.2.1 单次冰雹过程的风险计算 冰雹灾害的损失的形成，主要取决于冰雹 的物理特征和风险单元的暴露性和脆弱性。依 靠这些主要的冰雹灾害的风险指标，可以建立 下面的冰雹风险评估的数学模型：    n i ii cdefar 1 1 )*(*)(* （1） 上式中，a 是冰雹的覆盖面积，di 是第 i 种承 载体的密度，ci 是第 i 种承灾体的单位价值， f(e)是依赖于冰雹降落动能的承灾体损失系 数。 2.2.2 年均冰雹风险估算 在某地区，一年中可能发生多起冰雹灾 害，但是冰雹灾害的发生次数可能是一个随机 的过程。 那么，一年的冰雹损失风险可以用 下式估算： 1max *)(* rnnpnr a  （2） 式中,ra 是地区年均冰雹损失，nmax 是某一地 区的最大降雹日数，p（n<n）是小于 n 天的 降雹累积概率，r1 是单次冰雹过程的总损失， 可以利用公式（1）计算。式中 nmax 和 p（n<n） 相乘的结果表示了统计意义上的小于 n 天的 降雹日数的平均值。 3．冰雹灾害承灾体因子的量化方案 暴露量的评估就是定量地估算孕灾环境 中的承灾体暴露程度。在不同的条件下，冰雹 致灾因子对承灾体的危害是有差异的。如果用 概率风险评估承灾体的暴露量，则暴露量的概 率是一个条件概率，即承灾体完全或部分暴露 在致灾因子损害范围中的发生概率。 3.1 冰雹灾害发生日数的计算 估算冰雹灾害发生日数的目的是在给定 的时间 t 内估计冰雹事件的发生概率。一般情 况下，人为或自然现象产生的灾害事件的发生 次数被认为服从泊松概率。因此，冰雹日数的 出现概率可以利用 poisson 分布模型来模拟。 在任意给定时间间隔 t 发生的事件数 n（u） 的泊松分布为： tun etu n tununp )( )),(( ! 1 ]),(|)([     （3） 式中事件发生次数为 n，(u)是指阈值超过 u 的泊松参数。对于单位时间（如一年）中的泊 松分布可以简化为下式： )( ! )( )](|)([ u n e n u ununp     （4） 在一定周期中，多次发生的冰雹灾害事件 可以使用累积分布进行估算。一年中，发生次 数小于 n 次的冰雹发生概率总和可以使用下 面的累积分布计算： published by atlantis press copyright: the authors 134     n i u i e i u nunp 1 )( ! )( ])([  （5） 3.2 冰雹灾害的暴露量的易损价值估算函数 评估冰雹灾害的风险，首先要估算冰雹灾 害的承灾体的经济价值。尤其是冰雹影响区域 的承灾体的易损价值评估。 某一个区域中的承载体的易损价值取决 于承灾体的数量和承灾体价值的密度。由于区 域中承灾体的数量和承灾体的价值分布是一 个随机性很大的事件，同时这些承灾的信息在 现实世界中有具有不完备信息的特征。这中不 完备信息特征为灾害损失分析带来了极大的 困难。但是，这些信息通常又服从于某种统计 分布函数，而且这种分布函数的均值和方差是 较好获取的信息。这样就可以利用承灾体的概 率分布函数可以解决它的信息不完备性的缺 陷。 因此，利用承灾体的价值分布函数可以较 好地估算承灾体暴露量的易损价值。如果某一 地区的承灾体的价值分布符合正态正态分布 规律，则这个区域的承灾体暴露量的易损价值 就可以表示为： dxe mx c x x x     0 2 )( 2 1 2    （6） 在上式中，x 是承担风险的承灾体的数量变 量，估算方法为 x=a*di，a 是冰雹致灾因子 的覆盖面积，di 是某区域中第 i 种承载体的数 量密度。是承灾体数量的方差，是承灾体的 数量均值，m 是单个承载体的价值均值。利用 这种方法，可以将有限的承灾体易损值信息扩 展为较详细的承灾体易损值信息。 3.3 冰雹造成的经济损失率的计算函数 冰雹袭击承灾体后，并非承灾体的全部价 值丧失，有时只是部分价值丧失。这样，为了 有效地评估冰雹灾害的经济损失，就要建立冰 雹灾害的经济损失函数。 冰雹降落对承灾体的损害程度取决于冰 雹的降落动能。因为，冰雹较小时，降落动能 小，对承灾体的损害的可能性小；冰雹较大时， 降落动能大，对承灾体的损害可能性大。这样， 可以利用冰雹动能的特性，在冰雹灾害承灾体 的损失率和冰雹动能之间建立一种函数关系。 冰雹承灾体的经济损失率可以采用 s 型函数 表示：                ae e e ae ef aeb aeb ... 0.1 0.1 ......................0 )( （7） 公式中，e 是冰雹的降落动能。a、b 是 s 型的 调节参数，a 代表了冰雹对承灾体损害的最小 动能，b 是与降雹直径相关的调节参数，它可 以根据实际的统计资料将损失率函数调节到 符合地区的冰雹灾害的损失率。 单个冰雹的降落动能可以用动能公式 2 2 1 mue  计算。在冰雹的动能计算公式中， 一般情况将冰雹颗粒假设为球形。这样，冰雹 颗粒的质量为： 3 12 dm i   （8） 式中，d 是冰雹的直径，i 是冰雹的密度（取 846kg/m 3 ）。 在冰雹动能计算中的另一个变量是冰雹 的速度。冰雹由于受到空气的阻力作用，在快 要降落到地面的一段时间内，冰雹颗粒是处于 近似匀速运动。这时的冰雹运动速度就是冰雹 的降落末速度。冰雹降落末速度公式计算如 下： 2 1 3 4          da i c gd v   （9） 式中，a 是空气密度（取 0.9kg/m 3 ），cd 是阻 力系数（取 0.5）。 综合冰雹颗粒的质量公式和降落末速度 公式，单个冰雹的降落动能可以写成下式： 4 2 9 d c g e da i    （10） 但是在一次冰雹过程中，降落的冰雹颗粒 数量很大，而且冰雹的直径大小是不相等的。 在冰雹的损失评估中，可以用平均动能计算损 失效率。假设冰雹的直径服从正态分布，则冰 雹的平均动能为：           2 1 2 2 142 29 d d d da i dde c dg e     （11） 式中 d1 和 d2 是某次冰雹天气过程中，可以成 灾的冰雹颗粒的最小直径和最大直径。 4．冰雹灾害的承灾体的损失函数的参数确定 冰雹对承灾体的损害程度取决于冰雹的 降落动能和承灾体的材质。一般情况下，承灾 体的材质有钢材、玻璃和农作物等。例如，冰 雹灾害对汽车的直接破坏主要是车身钢板壳 的凹陷或者挡风玻璃的破碎。因此，对于汽车 承灾体造成的直接经济损失也就是车身钢板 外壳和玻璃的置换费用。 published by atlantis press copyright: the authors 135 在实际的冰雹灾害的损失评估工作中，冰 雹造成的各种经济损失的实际数据不易收集， 以及冰雹灾害数据的明显随机特性使得小样 本数据的可靠性低。为了能在评估工作中弥补 这种信息中的不足，本节利用 ansys 软件模 拟各种粒径的冰雹以不同的末速度撞击钢板 和玻璃，通过模拟得到钢板和玻璃的应力场分 布、塑性应变场分布以及变形等多种结果，从 而确定冰雹对钢板和玻璃的损伤。 通过数值模拟不同条件下冰雹对承灾体 的损伤程度，进一步确定冰雹灾害对承灾体造 成的经济损失率函数中的 a 和 b 参数。 4.1 ansys 的应力分析原理 4.1.1 冰雹撞击玻璃材料的非线性分析原理 在 ansys 程序中，分析冰雹对玻璃材料 的伤害效应时，采用了非线性材料的结构分析 方法，即牛顿—拉普森迭代法。牛顿—拉普森 迭代法通常有三种方法：纯牛顿—拉普森迭代 法、修正牛顿—拉普森迭代法和准牛顿—拉普 森迭代法。前两者也称为切线刚度法，而准牛 顿—拉普森迭代法也称为割线刚度法。在 ansys 程序中，割线刚度法被称为弧长法。 在模拟计算中，冰雹对玻璃撞击所造成的 损害主要取决于撞击的冲量。这个冲量一般包 括两部分：冰雹降落产生的竖直冲量和冰雹相 对承灾体的水平冲量。 根据冰雹降落到地面时的末速度计算公 式，将冰雹的各个物理常量代入公式，可以等 到简化后的计算公式： d c gd v da i 7.156 3 4 2 1             （12） 式中，  v 是冰雹末速度（m/s），d 是冰雹的 直径（m）。 同样，冰雹的质量（千克）也可以简化 为： 33 5.221 12 ddm i   （13） 这样，冰雹降落产生的竖直冲量就可以简 化为只与冰雹直径有关的函数关系： 2 7 2 1 3 34714 3 4 12 d c gd dmvi da ii             （14） 根据防雹试验的结果，水平风速对于直径 小的冰雹影响较大，然而对于直径大于 3cm 冰 雹的影响可以忽略。在分析冰雹的竖直冲量 时，直径小于 3cm 的冰雹的实际降落末速度 可以使用静止大气中冰雹下落的末速度乘以 修订系数 1.22。 对于冰雹的水平冲量，可以使用下式计 算： hh i hh vdvdmvi 33 5.221 12   （15） 式中的 vh 是冰雹的水平运动分量。 4.1.2 冰雹撞击玻璃的模拟结果分析 在实际情况中，如果承灾体为汽车时，并 假设汽车挡风玻璃的倾角为 α。则冰雹撞击挡 风玻璃的合冲量为：  sincos h iii   （16） 汽车设计中的挡风玻璃倾角 一般为 30°~45°，而且现在轿车挡风玻璃倾角设计得 越来越小，有些已经达到 20°。为了讨论问题 的方便，本文后面计算中的倾角 α 取 30°，则 冰雹撞击汽车挡风玻璃的合冲量： hh iiiii 5.0866.0sincos    （17） 冰雹对玻璃的集中载荷 f 可以近似通过 冲 量 与 冲 力的 关 系 得到 ， 可以 表 示 为： t if   。根据以往的试验结果，⊿t 一般取 0.001s。 当冰雹粒径达到一定直径大小时，冰雹对 玻璃的集中载荷 f 使得玻璃的形变会变得很 大。当应力达到一定阈值时，会因为应力超过 了玻璃的抗压强度，造成玻璃破碎。 使 用 ansys 的 模 拟 软 件 ， 对 于 0.5~10.0cm 的冰雹撞击普通玻璃和钢化玻璃 进行模拟（见图 1）。模拟结果表明（表 1）， 当冰雹直径达到 4cm 时，钢化玻璃被撞击后 产生的形变约为 1mm，可认为此时玻璃产生 裂纹，因此，冰雹对钢化玻璃造成破坏的最小 粒径为 4cm。 对于普通玻璃，冰雹直径达到 2cm 时， 玻璃的形变已经大于 1mm，超出了钢化玻璃 的形变量。可以认为 2cm 的冰雹会造成普通 玻璃破碎。 表 1 不同粒径冰雹撞击不同速度车辆造成挡风玻璃的形变量 冰雹粒径（cm） 0.5 1.0 2.0 3.0 4.0 6.0 10.0 普通玻璃（mm） 0.011 0.111 1.151 4.564 12.228 81.059 — 钢化玻璃（mm） 0 0.009 0.09 0.365 0.967 5.068 21.953 published by atlantis press copyright: the authors 136 图 1 直径 4cm 冰雹撞击挡风玻璃（钢化玻璃）模拟结果 其中：dmx 为最大形变（单位：mm）；smx 为最大应力（单位：mpa） 4.2 冰雹撞击钢板的模拟原理和分析 4.2.1 冰雹撞击钢板的显式非线性分析 在 ansys 程序中，对于钢板的模拟不同 于 玻 璃 的 模拟 方 法 。 钢 板 的模 拟 采 用了 ls-dyna 的方法。ls-dyna 是一个以显式 为主，兼顾隐式的非线性动力有限元分析程 序。它主要是在 t+δt 时计算位移和平均加速 度：      a tttt fku     1 （18） 对于线性问题，由于 k 是线性的，解是 无条件稳定，计算过程可以采用大的时间步。 对于非线性问题，通过一系列线性逼近（牛顿 -拉普森迭代法）来获取解，要求转置非线性 刚度矩阵 k，收敛需要小的时间步。 ls-dyna 的显式时间积分采用中心差分 法在时间 t 求加速度：         int1 t ext tt ffma   （19） 式中 ft ext ——施加外力和体力矢量； ft int ——内力矢量，它由下式构成： contacthg n t t ffdbf               int （20） 式中三项依次为当前时刻单元应力场等效节 点力、沙漏阻力（为克服单点高斯积分引起的 沙漏问题而引入的粘性阻力）和接触力矢量。 4.2.2 冰雹撞击钢板的模拟试验 冰雹对钢板材料的承灾体撞击时，会对钢 板材料的承灾体造成损坏。然而，这种损坏的 程度与冰雹的特性冰（雹的直径、质量、撞击 的速度）和被撞击物的特性（几何特性和材料 的特性）密切相关。由于真实的损失数据很难 获取，同时在实际操作中又存在着各种各样的 困难，因此冰雹撞击的数值模拟就非常必要。 通过数值模拟实验可以重现撞击过程，分析撞 击结果。 本节对钢板材料的实验模拟，材料参数以 冰雹的承灾体汽车为对象。实验目的是利用 ansys/ls-dyna 软件，通过计算不同直径冰 雹的降落动能，确定不同冰雹动能撞击汽车钢 板的损伤程度。同样，利用公式（16），可以 确定冰雹撞击汽车外壳钢板末速度的计算公 式 h vvv 5.0886.0   。 一 般 来 说 ， 汽 车 外 壳 钢 板 的 厚 度 为 0.6~0.8mm，因此，本文模拟中的钢板厚度取 0.8mm。根据模拟结果，随着冰雹粒径的增大， 冰雹颗粒对钢板的撞击形变也增大（见图 2）。 当冰雹颗粒直径为 3.0cm 时，钢板形变为 3mm （见表 2）。这种变形已对车辆表面造成伤害， 可以认为冰雹对钢板造成破坏的最小粒径为 3.0cm。 (a) （b） （c） 图 2 直径 2.0cm（a）、3.0cm（b）和 6.0cm（c）的冰雹撞击钢板的模拟结果 published by atlantis press copyright: the authors 137 表 2 不同粒径冰雹撞击钢板的形变量 冰雹粒径（cm） 0.5 1.0 2.0 3.0 4.0 6.0 10.0 钢板型变量（cm） 0 0.036 0.13 0.30 0.51 1.05 28 4.3 汽车的总损失系数的确定 公式（7）的损失率函数是估算冰雹灾害 损失的关键性因素。如果确定了冰雹伤害承灾 体的最小直径，就可以确定公式（7）中的冰 雹最小动能参数 a，然而参数 b 可以利用已有 的数据来确定，使得损失函数接近于真实的损 失情况。 现代化城市中的大量汽车的出现，增加了 城市的冰雹灾害承灾体的类型。冰雹对汽车造 成的损害主要包括车身钢板外壳的变形和玻 璃破裂。因此，对于汽车损失的评估分成了钢 板表面破坏和玻璃的破碎两部分。因此，损失 率函数也分成了车身钢板表面损失和玻璃的 损失。 为了讨论汽车损失率问题的方便，分别用 下标 s 和 g 表示钢板和玻璃。对于挡风玻璃来 说，参数 ag 取直径为 4cm 的冰雹的降落动能， 则小于 4cm 的冰雹对挡风玻璃造成的损失为 零，即 f(e)g=0。而对于车身钢板来说，参数 as 取直径为 3.0cm 的冰雹的降落动能，则小于 3.0cm 的冰雹对汽车挡风玻璃造成的损失为 零，即 f(e)s=0。 由于参数 b 的确定与地区的经济发展水 平有密切关系。这里，作者根据天津市的历史 统计数据，确定公式（7）中适合天津市冰雹 灾害对车辆挡风玻璃和车身钢板损失率函数 的系数 bg 和 bs，这两个值分别值 0.01dg 和 0.01ds，其中，dg 和 ds 分别是挡风玻璃和汽 车钢板的形变量，直径单位为毫米。这样，冰 雹 对 汽 车 造 成 的 损 害 的 总 损 失 系 数 f(e)=f(e)g+f(e)s。 4.4 冰雹造成农业损失系数的确定 冰雹对农业生产地区造成的危害主要是 农业作物，以及生产农作物的农业设施（如蔬 菜大棚的塑料薄膜和玻璃）。 冰雹出现时常与 短时大风和强降雨同时出现, 对生长中的农 作物破坏极大。 农作物的枝叶、茎秆、果实受到冰雹颗粒 的打击, 会因叶子损失、茎秆折断、果粒脱落 等伤害而形成减产。苗期的农作物遭受冰雹袭 击后, 可使幼苗受伤而不能正常生长, 若幼苗 被砸伤过重, 则需重新播种而延误农事季节。 农作物在灌浆成熟期受到冰雹袭击, 会直接 影响并阻碍正常灌浆成熟而造成严重减产和 品质变劣。果树在开花座果时遭受冰雹灾害, 会形成严重的落花落果现象而导致大幅度减 产。虽然有的幼果被冰雹打伤后也可以发育成 熟, 但是带有雹伤的果实的商品价值会大幅 度降低。然而，果实在成熟期遭受雹灾后, 果 实容易腐烂, 也不耐贮存, 常带来无法弥补的 经济损失。 冰雹的危害程度主要取决于冰雹的强度、 密度、持续时间和降雹的季节。冰雹直径越大, 密度越大, 危害也越严重, 成灾概率越大。 根据以往研究，当冰雹直径大于 0.5cm 时，就会对农业造成损失。因此，对于农作物 的雹灾风险，公式（7）中的调节参数 a 就可 以取 0.5cm 冰雹的动能。根据天津市冰雹灾害 对农业造成损失的历史统计数据，公式（7） 中的系数 b 可以取值 0.01。 5．冰雹灾害的损失风险评估分析 由于冰雹灾害具有明显的地区性特点，本 节将以天津地区为冰雹灾害的孕灾环境，使用 冰雹灾害评估模型对天津市冰雹灾害的损失 风险进行评估分析。根据天津市的实际情况， 冰雹灾害的承灾体主要为汽车和农作物。 5.1 风险评估中的数据说明 冰雹的发生天数的数据来自天津市气象 部门 1958 年至 2003 年的观测统计数据（见表 3）。冰雹直径数据也是来自气象部门的观测数 据。 轿车统计数量和农业产值来自天津市 2007 年的统计年鉴数据（见表 4 和表 5）。因 为统计年鉴中只将轿车分成微型汽车、小型汽 车和中型汽车三类。为了讨论汽车价值的需 要，分别将这三类车作为低价、中价和高价车。 天津市各个区县的农作物的价值，依据统 计年鉴中的农业总产值计算。 表 3 1958 年至 2003 年天津市及郊区的总降雹频率 年次数 0 1 2 3 4 5 合计 总统计数 79 53 32 18 1 1 184 百分比 42.92 28.81 17.39 9.78 0.55 0.55 100.00 published by atlantis press copyright: the authors 138 表 4 轿车数量统计表（辆） 车类型 微型 小型 中型 数量 102784 504198 26002 表 5 各区县 2006 年农业总产值（万元） 东丽区 西青区 津南区 北辰区 武清区 宝坻区 宁河县 静海县 蓟 县 43939 64145 22252 52169 274835 167013 126997 126366 178131 5.2 冰雹评估模型的模拟分析 5.2.1 冰雹天气的情景设计 由于某一地区的冰雹发生天数，以及冰 雹的粒径都是一个随机变化的数据。因此，在 风险评估中要考虑这些随机特性。 一般情况下，冰雹的发生天数服从泊松 分布。这样，依照泊松分布规律，利用天津市 的冰雹发生日数的历史统计数据，可以确定天 津地区的冰雹发生日的泊松分布函数的参数 为 0.978。而且，表 3 中的历史观测数据表明 天津地区的最大年将雹日数是 5 天。 如果在冰雹灾害风险评估中，假设冰雹 的直径服从正态分布规律。这样就可以建立了 天津地区的冰雹直径的正态分布模型。通过分 析天津的历史降雹的粒径数据，冰雹灾害的平 均冰雹颗粒的直径约 2.5cm，方差约 3.0， 以冰雹发生天数的泊松分布函数和冰雹 直径的正态分布模型，结合模特卡罗随机模拟 方法，生成天津地区冰雹天气的 10 个情景（表 6）。这些以天津市历史统计资料为基础生成的 冰雹天气情景，可以作为分析天津是冰雹灾害 风险的致灾因子样本。 表 6 蒙特卡罗模拟的冰雹天气情景 序号 冰雹日数 p（n<n） 雹径（cm） 1 2 0.548 4.33 2 1 0.368 3.18 3 1 0.368 5.50 4 1 0.368 6.15 5 2 0.548 0.87 6 2 0.548 5.24 7 1 0.368 1.98 8 2 0.548 1.50 9 2 0.548 4.12 10 1 0.368 5.30 平均 1.5 / 4.12 5.2.2 冰雹天气的汽车损失情况评估 根据天津市统计年鉴资料，估算天津市 轿车的价值均值约 7.58 万元，方差 4.74。使 用天津市的 10 个冰雹灾害的天气情景，并采 用蒙特卡罗模拟的方法产生受损的汽车数量， 通过轿车价值的正态分布模型估算出受损汽 车的总价值（表 7）。 随机模拟结果显示，受灾区域的汽车损 失情况与车辆的平均价值关系密切。在受灾车 辆数变化不大的情况下，车辆平均价值越大， 冰雹灾害的损失越大。 同时，模拟结果也显示出冰雹直径的变 化对损失率的影响。当冰雹冰雹直径大于对承 灾体伤害直径时，相应的损失率函数就为大于 0 的正数。如果冰雹的直径小于对承灾体伤害 直径时，相应的损失率函数就为 0。而且冰雹 直径越大，车辆本身的损失率值也越大。 可以看出，冰雹灾害的损失评估模型反映了承 载体的价值和致灾因子的强度等综合因素的 作用。 5.2.3 农作物损失情况评估 对于农业损失情况评估，使用了天津市 的 9 个具有农业生产的区县的统计数据（表 5）。并且假设这 9 个区县中的农作物的产值的 面积密度分布为均匀分布。冰雹灾害的影响区 域，以 6km 宽度为基准，长度是冰雹灾害经 过区县的路径长度。 冰雹天气数据仍然使用表 6 的 10 个冰雹 天气情景数据。估算的天津地区各个农业生产 区的一年内的冰雹灾害造成的损失情况如表 8 所示。 依据表 8 的评估结果和表 6 的冰雹天气情 景数据可以看出，冰雹灾害评估模式评估的农 作物的经济损失是随着冰雹的粒径增加，冰雹 published by atlantis press copyright: the authors 139 灾害的损失也在增加。而且冰雹天数多的情景 造成的农业损失明显高于发生天数小的情景。 表 7 冰雹天气造成的汽车损失（万元） 序号 f(e)g f(e)s f(e) 受损车辆数 受灾车辆均价 受灾车辆总损失 1 2.858e-3 4.416e-3 7.274e-3 317 7.58 9.5780 2 0 1.910e-4 1.910e-4 454 6.07 0.1936 3 6.017e-2 1.996e-2 8.013e-2 748 12.77 281.6772 4 1.594e-1 4.036e-2 1.997e-1 366 16.46 442.7544 5 0 0 0 0 0 0 6 3.629e-2 1.574e-2 5.203e-2 436 11.83 147.0581 7 0 0 0 0 0 0 8 0 0 0 0 0 0 9 9.350e-4 2.930e-3 3.865e-3 552 7.77 9.0851 10 4.087e-2 1.676e-2 5.763e-2 309 10.79 70.7043 平均 359.3 8.431 96.1 表 8 冰雹天气造成的农业损失（万元） 区县 1 2 3 4 5 6 7 8 9 10 东丽区 4590.4 449.43 3943.7 5992.8 6.6865 9739.8 2680.3 65.503 3766.2 3418.2 西青区 6701.3 656.11 5757.3 8748.7 9.7614 14219 3912.8 95.626 5498.1 4990.1 津南区 2324.7 227.6 1997.2 3034.9 3.3863 4932.5 1357.4 33.173 1907.3 1731.1 北辰区 5450.2 533.61 4682.4 7115.3 7.939 11564 3182.3 77.772 4471.6 4058.5 武清区 28712 2811.1 24668 37485 41.824 60922 16765 409.72 23557 21381 宝坻区 17448 1708.3 14990 22779 25.416 37021 10188 248.98 14315 12993 宁河县 13268 1299 11399 17321 19.326 28151 7746.8 189.32 10885 9879.7 静海县 13202 1292.5 11342 17235 19.23 28011 7708.3 188.38 10831 9830.6 蓟 县 18610 1822 15988 24295 27.108 39486 10866 265.55 15268 13858 5.2.4 讨论 模拟实验的结果说明通过概率方法和物 理方法结合构建的冰雹灾害的评估模型，能够 综合反映致灾因子强度和承灾体价值情况。而 且，评估模型可以应用于汽车、建筑玻璃和农 作物等损失评估。 但是，评估模型中的承载体的价值损失率 函数的准确评估是一个复杂的问题，同承灾体 的地理分布关系密切，需要在实际工作中使用 实际数据不断地完善。 6．结论 （1）针对冰雹灾害对社会经济造成的损 失风险问题，作者利用物理方法和概率方法建 立了相应的冰雹灾害的风险评估的数学模型。 （2）利用 ansys 软件和 ls-dyna 软 件分别模拟了不同粒径的冰雹撞击汽车挡风 玻璃（钢化玻璃）和车身钢板以及建筑物玻璃 （普通玻璃）造成的损害情况。模拟结果表明， 而当冰雹直径达到 2.0cm 时，冰雹会对建筑物 的普通玻璃产生破坏。当冰雹颗粒直径为 3.0cm 时，冰雹可以对汽车外壳钢板造成了损 失；而当冰雹直径达到 4.0cm 时，冰雹会对汽 车的挡风玻璃产生破坏。 （3）根据天津市的历史统计资料，利用 建立的冰雹灾害风险评估数学模型，对天津市 的冰雹灾害损失做了模拟分析。分析结果表 明，在资料信息不足的情况下，本数学模型是 有效的冰雹灾害损失的综合评估模型。 本研究只是对冰雹灾害风险评估建模的 理论研究，如果将冰雹风险模型与 gis 系统相 结合，可以做到冰雹灾害风险的定点风险分 析，这会更加有利于冰雹灾害的风险管理。 参考文献： 1. s. a. changnon, the scales of hail. journal of applied meteorology , 16(1977) 626–648. 2. i. kuhnel, the use of a multifactor southern oscillation index for the estimation of annualhailstorm frequencies in the sydney area, int. j. climatol,18(1998) 841–858. 3. w. wang and h. z.jia , forecasting hail with radar vertically integrated liquid data, meteorological monthly, ,28(1) (2002) 47~48. 4. y. l. jin and q. zhang, a case study of synoptic process and inner structure of hails in beijing, meteorological monthly,28(1) (2002)18~20. 5. a. bengtsson and c. nilsson, extreme value modelling of storm damage in swedish forests, nat. hazards earth syst. sci., 7(2007) 515~521. 6. r. leigh and i. kuhnel, hailstormloss modelling and risk assessment in the sydney region, australia. natural hazard , published by atlantis press copyright: the authors 140 24(2001) 171~185. 7. p. damir, comparison of hail characteristics in nw croatia for two periods, natural hazards, 29(2003) 543~552. 8. s. hergarten , aspects of risk assessment in power-law distributed natural hazards, nat. hazards earth syst. sci., 4 (2004) 309~313. 9. j. w. han, h. m. wang and l. wu, the analysis and assessment on thunderstorm and hail disasters and the counter mea ures in innermongolia, journal of arid land resources and environment, 23（7） (2009)31~38. 10. a. c. stanley, increasing major hail losses in the u.s., climatic change, 96(2009) 161–166. 11. m. julian, damaging hail hits vw, weather,63 ( 2008),p218. 12. european commission ( eds.), first report on the harmonisation of risk assessment procedures ,(dg health & consumer protection, brussels,2000). 13. r. edmund, f. peter and r. david, estimating injury and loss of life in floods: a deterministic framework, natural hazard, 36(2005) 43-64. 14. y. zhang and z. lei, static characteristic and impact fracture behavior of automobile glass. journal of mechanical engineering, 45(2)(2009) 64~75. published by atlantis press copyright: the authors 141 microsoft word publication statement-jracr publication statement october 31, 2021 dear authors and readers, the journal was founded by the society for risk analysis – china (sra-china) in july 2011. as a sra-china official publication, jracr aims to promote the development of science and technology in the field of risk analysis and crisis response, to provide an international forum for sharing theory and applications the field, and for exchanging the research findings and case studies. the journal is considered as an international journal publishing papers in science and technology rather than policies and regulations. jracr was published by atlantis press from july 2011 to july 2021. however, springer nature officially completed the acquisition of atlantis press on march 10, 2021. springer nature officially announced the termination of the publishing cooperation agreement with jracr from july 30, 2021. in view of this, jracr signed a pkp publishing services: hosting support agreement with the simon fraser university, pkp publishing services (“pkp services”) on august 10, 2021. according to the above agreement, jracr will be hosted and published in public knowledge project – open journal systems since volume 11, issue 3. thank you very much for your support to our journal. we look forward to receiving you and your friends’ more contributions and attention. with kind regards editorial board journal of risk analysis and crisis response hosted and published in public knowledge project – open journal systems https://jracr.com/ email: jracr_srachina@126.com linkedin: junxiang zhang signature: chongfu huang (editor-in-chief of jracr) journal of risk analysis and crisis response issn print: 2210-8491 issn online: 2210-8505 atlantis press journal style copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). research on the development of county finance in guizhou province in the promotion of precise poverty alleviation yu ding, mu zhang school of finance, guizhou university of finance and economics, guiyang guizhou e-mail: 13399855052@163. com, rim_007@163. com received december 16, 2017 accepted january 28, 2018 abstract this paper selects the 50 state-level poverty-stricken counties in guizhou province as the research object, and uses financial scale, financial efficiency and financial structure to represent the level of financial development in each county, using economic growth and income distribution as controlled variable. the poverty of every county is expressed by poor slow index. applying the panel data model, the promotion of local financial development to the targeted poverty alleviation is studied. the empirical evidence shows that the financial scale, financial efficiency, financial structure, economic growth and poverty reduction of 50 national poverty-stricken counties in guizhou province are positively correlated. however the financial scale is more significant to reduce the incidence of poverty than that of financial efficiency. keywords: financial development, precision for poverty alleviation, panel data model. 1. introduction as research object, guizhou is the largest and most impoverished province in china with the largest and the deepest poverty. for years, it follows policy steps. according to the latest statistics from guizhou statistics bureau, there are still as many as 50 poverty-stricken counties in guizhou province. so there is still a long way to go. in 2014, president xi jinping put forward the concept of "precision for poverty alleviation". this concept has become the basic strategy of the national poverty alleviation work. finance is a resource which can configuring other resources. its total allocation and allocation efficiency directly determine the level of economic development. under the current situation of poverty in guizhou and the policy of "precision for poverty alleviation", how to combine financial development to promote poverty alleviation is worth discussing. 2. literature review set up fixed effects vector decomposition model with 23 provinces 2011-2008 data. it has shown that the rural financial development can significantly alleviate poverty, directly and indirectly. instability in the process of the rural financial development has no significant impact on poverty (tan, 2011) [1] . select the financial development scale and the financial development efficiency as financial development indicators. through empirical test for chinese provincial panel data: financial development can through economic growth, income distribution channels to raise the income level of the poor. but the financial wobbles will offset the effect of financial development of poverty reduction (cui and sun, 2012) [2] . 2001 to 2010, china's rural financial development is negatively related to the incidence of poverty (wu, 2012) [3] . the plight of the rural financial poverty alleviation include: poor credit environment, poor farmers lack of mortgage and financial institutions development is difficult to meet the demand of expanding capital , and regulatory standards hinder financial institutions (guo, 2013) [4] . using the credit risk model analyze the problems of rural credit. credit risk uncertainty even more difficult for the rural registered permanent residence for credit funds, to improve the credit rationing, intensify 52 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 financial poverty alleviation and put forward reasonable suggestions(song, li and xiao, 2017) [5] . analyze sichuan bazhong 3 counties 1 district nearly four years of panel data. the increase of agricultural loans accounted does not improve the farmers' income. in addition, from the perspective of the financial poverty alleviation efficiency index of building, the rural enterprise loan poverty alleviation efficiency during the sample period is on the rise (deng, 2015) [6] . financial development on poverty alleviation role of literature research accurately, but as a serious poverty situation of guizhou province, the current study is seldom the county financial development in guizhou of precise role in promoting research for poverty alleviation. for this reason, this paper uses panel data model for 2013 to 2015, 50 state-level poverty-stricken counties in guizhou financial development study of precise role for poverty alleviation. 3. linear regression model of panel data 3.1. panel data panel data refers to the fixed a group of subjects in the interval of time such as continuous observation data, with cross section and time are the two characteristics of the data. under the panel data in double scalar said, for example , 1, 2, , ; 1, 2, , it y i n t t  where i corresponding to the panel data in different individuals, n said panel data of individual number. t for panel data in different time, t says the maximum length of time sequence. if the fixed t unchanged, yi. , i=1, 2, … , n, is in the cross section of n random variables; if i fixed constant, y. t, t = 1, 2, . . . , t, is the profile of a time series. panel data is divided into two kinds of characteristics. is a section on individual number is little, and each individual time span is long. secondly, the cross section on individual number, and each individual short time span. 3.2. panel data model based on panel data regression model is called the panel data model, usually distinguish between linear and nonlinear etc. for convenience, in this paper, the linear panel data regression model is written as: where yit is interpreted variable for individual values that i at time t, xhit for the hth explained variable for individual values that i at time t; β hit for the hth explained variable be estimated parameters; uit is random error term. panel model is usually divided into three categories, namely the hybrid model, the fixed effect model and random effect model. panel model is usually divided into three categories, namely, hybrid model, fixed effect model and random effect model. in one of the fixed effects model parameters  it  are fixed, the random error term said ignored or changes in the individual, at any time and in a given the observation effect and explanatory variables under the condition of the expectation of a random error term is equal to zero, uit homoscedasticity, different individuals and different point corresponding uit are independent of each other. the model forms as follows: there is only individual effect in the simulation, the model in the form of: i  as a stochastic variable to describe the differences between different individuals establish regression function. because i is invisible, and with the change of the interpretation of the observed variables xhit associated, so called individual fixed effects model. if time effect exists only in the model, the model in the form of: i  is a random variable, said for n individual item has n different intercept, and its change with xit, t is a random variable, said to t section (time) of t different intercept, and its change associated with xit, says this model for individual point double fixed effects model. 4. empirical analysis 4.1. model, index and data 4. 1. 1. model setting the purpose of this paper is to study the financial development on precision of poverty alleviation, choose ithit k h hit ux   1 tt y  ithit k h hiit ux   1 y  ithit k h hit ux   1 t y  t,…1,2,tn;,…,2,1,y 1     iux ithit k h hititit  53 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 some representative index to indicate the level of financial development in guizhou, and poverty slow index is used to represent each county poverty. based on the assumptions about financial development has a positive promoting effect on economic growth, mainly through economic growth and income distribution influence the slow down of poverty. therefore, this article selects three factors as the empirical equation of the independent variable, the level of financial development, economic rights respectively, the income distribution. among them, we will measure the financial development level from the financial scale, financial efficiency and financial structure. in addition, there are many factors to reduce poverty, such as government investment, residents' level of education, etc., these factors we unified with the random error term. get the following basic model: pov said poverty. fs said financial scale. fe said financial efficiency and fsr said financial structure. rgp said the rise of the economy. ig said income distribution. μsaid the other factors that affect poverty to slow, i = 1, 2, 3. . . 50, said 50 key poverty alleviation and development counties in guizhou; t = 1, 2, 3, according to different years. 4. 1. 2 selection of indicators slow poverty indicators (pov): in this paper, the poverty rate is used to measure the degree of poverty. the greater value of the index, the more serious the poverty situation. scale of financial indicators (fs): the total amount of lending and deposit/gdp as the index of financial scale. the greater the value of the index, it shows that the bigger financial system development and the higher the level of financial development. financial structure index (fsr): the existence of economies industry capital structure, is one of the important indicators to measure the financial structure, it is equal to direct financing than indirect financing. according to the situation of county of guizhou province directly, equity financing is the main way of the enterprise financing, bond and other financing way development is not perfect, once again ignored. and since the province a total of 26 listed companies in guizhou, of which only two located in the state-level counties list, so this article uses the 0-1 programming to deal with variable, with the county value of listed companies to "1", the rest of the values for "zero". financial efficiency index (fe): use of the whole society fixed assets investment and the banking financial institutions balance to calculate. the greater value of the index, the higher the efficiency of financial institutions to invest in, and the higher the financial efficiency. economic rights long index (rgp): there are many indicators to measure economic growth. based on the gdp per capita to measure economic growth, if the indicator into a rising trend, it means economic grow and development well. income distribution index (ig): considering the comparability and data item, the urban per capita disposable income and rural per capita net income of is adopted to measure the equality of income distribution. the index's rise in value, the income gap of urban and rural residents is bigger and bigger, the income distribution is unfair. the modeling of variable table is shown in table 1. table 1. the modeling of variable table types of variables variable name variable definition explained variable pov slow poverty indicators explanatory variable fs scale of financial indicators fsr financial structure index fe financial efficiency index control variable rgp economic rights long index ig income distribution index 4. 1. 3. data source and description according to the latest statistics yearbook, there are 50 state-level poverty-stricken counties in guizhou. this paper collected data from 2013 to 2015, 50 statelevel counties. it has the same statistical caliber. and it selected data with high reliability. data mainly comes from <guizhou statistical yearbook> from 2013 to 2013. it used excel deal with some simple data. the original data are shown in table 2. ititititititiit igrgpfsrfefscpov   54 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 table 2. 50 state-level poverty-stricken counties in guizhou in almost three years of relevant indicators data tables. area year pov fs fe fsr rgp ig liuzhi 2013 0. 199712454 1. 561988906 1. 261935037 0 2. 037062147 3. 271872998 2014 0. 1698 1. 360806148 1. 300309598 0 2. 5648 3. 012648345 2015 0. 139 1. 323821759 1. 50255144 0 2. 9673 2. 996302616 shuicheng 2013 0. 28859804 4. 74945391 0. 514804652 0 1. 861926518 3. 320603539 2014 0. 2481 5. 284909723 2. 036130484 0 2. 4496 3. 06296875 2015 0. 21 5. 232148917 2. 249188423 0 2. 7828 3. 019953808 pan 2013 0. 2349 0. 984489981 1. 184473418 0 3. 504879738 3. 150674731 2014 0. 1879 0. 909242318 1. 267221972 0 4. 0896 2. 90451464 2015 0. 143 1. 015130525 1. 377510969 0 4. 5397 2. 886136861 zhengan 2013 0. 21617053 2. 302466037 0. 568922751 0 1. 135984523 3. 457081445 2014 0. 1638 2. 04698211 0. 673866091 0 1. 5592 2. 943560271 2015 0. 151 2. 339418396 1. 056572026 0 1. 847450404 2. 922222222 daozhen 2013 0. 196069861 2. 469862757 0. 537605587 0 1. 318392271 3. 611835045 2014 0. 1636 2. 370314465 0. 569141755 0 1. 6275 2. 939867354 2015 0. 116 2. 37884185 0. 630391482 0 1. 970250348 2. 921421025 wuchuan 2013 0. 235257552 2. 573050159 0. 675316875 0 1. 085231143 3. 675471298 2014 0. 1874 2. 461270063 0. 777029961 0 1. 3473 2. 944797987 2015 0. 137 2. 78176 1. 033144446 0 1. 609810851 2. 931643217 xishui 2013 0. 228370636 1. 597287111 0. 7804341 0 1. 758006561 3. 549972459 2014 0. 1778 0. 905375389 0. 650641026 0 2. 1734 2. 93010449 2015 0. 132 1. 692529881 1. 205118422 0 2. 466186335 2. 914234739 puding 2013 0. 254689753 1. 55787234 1. 815910393 0 1. 5692 3. 323174482 2014 0. 2077 1. 479775281 1. 932032301 0 1. 876 3. 280739045 2015 0. 151 1. 68896147 2. 166130624 0 2. 1876 3. 207245787 zhenning 2013 0. 25737951 2. 154028555 2. 12146189 1 1. 8314 3. 371488108 2014 0. 2176 1. 877455566 2. 205821206 1 2. 2544 3. 317306081 2015 0. 187 1. 971832866 2. 514094518 1 2. 631 3. 287148753 guanling 2013 0. 286480071 1. 595727092 1. 00096707 0 1. 6678 3. 340072553 2014 0. 2416 1. 36863711 1. 073092882 0 2. 0134 3. 275810224 2015 0. 18 1. 52686863 1. 337890201 0 2. 4092 3. 234344272 ziyun 2013 0. 286838134 1. 802484402 1. 116674748 0 1. 3071 3. 308348413 2014 0. 2427 1. 640382052 1. 24827883 0 1. 6589 3. 2834015 2015 0. 193 1. 809332557 1. 449481813 0 1. 8985 3. 180335085 dafang 2013 0. 277637848 1. 084636195 0. 800785621 0 1. 667343057 3. 318108889 2014 0. 2351 0. 971021797 0. 801148796 0 2. 0372 3. 274121406 2015 0. 191 1. 079245957 0. 949919431 0 2. 289736958 3. 20050014 zhijin 2013 0. 319547534 1. 536880811 1. 072905871 0 1. 371192555 3. 459868635 2014 0. 2732 1. 432636597 1. 310437236 0 1. 7361 3. 314683053 2015 0. 225 1. 786448195 1. 679557145 0 1. 957270049 3. 24609209 nayong 2013 0. 268517252 0. 943353851 0. 991139212 0 1. 837045083 3. 513125884 2014 0. 2239 0. 902212657 1. 141195031 0 2. 2095 3. 472671548 2015 0. 181 0. 997070185 1. 412794795 0 2. 535364984 3. 397934332 weining 2013 0. 240462341 0. 911448326 0. 939779449 0 0. 976738675 3. 315709469 2014 0. 1863 0. 896862461 1. 021591949 0 1. 1992 3. 277275662 2015 0. 145 0. 965618926 1. 062023815 0 1. 488182555 3. 18981333 55 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 continued table 2. area year pov fs fe fsr rgp ig hezhang 2013 0. 260545803 1. 28043682 0. 599641977 0 1. 01379099 3. 501040012 2014 0. 2177 1. 091468101 0. 737974962 0 1. 3937 3. 460645382 2015 0. 179 1. 153913071 0. 85954823 0 1. 711211481 3. 367844364 jiangkou 2013 0. 235026029 2. 601040087 1. 148558961 0 1. 6017 3. 382943551 2014 0. 1878 2. 463494913 1. 429294756 0 1. 9289 3. 26176566 2015 0. 138 2. 377842143 1. 579488811 0 2. 385833573 3. 285654703 shiqian 2013 0. 258555943 2. 046813905 0. 888954707 0 1. 2592 3. 352767933 2014 0. 212 1. 942521502 1. 046206353 0 1. 5657 3. 190305791 2015 0. 165 2. 116778523 1. 303385894 0 1. 954739259 3. 184238685 sinan 2013 0. 254316479 2. 028960087 1. 158156172 0 1. 4778 3. 535193549 2014 0. 2132 1. 93537415 1. 168472614 0 1. 7667 3. 279853235 2015 0. 164 2. 042595442 1. 307985486 0 2. 028211285 3. 269407744 yingjiang 2013 0. 236392835 2. 09458126 1. 074512282 0 1. 7001 3. 448269353 2014 0. 1913 1. 926498048 1. 223401746 0 2. 0722 3. 235050522 2015 0. 143 1. 9746699 1. 521337161 0 2. 606954689 3. 272047349 dejiang 2013 0. 28802056 1. 438414918 1. 078973324 0 1. 6341 3. 694211828 2014 0. 2579 1. 407503411 1. 255995961 0 1. 9852 3. 329270357 2015 0. 202 1. 507307556 1. 422050628 0 2. 280744337 3. 312446219 yanhe 2013 0. 271253484 1. 646139756 0. 786537841 0 1. 3106 3. 544692896 2014 0. 2295 1. 50261708 0. 806675939 0 1. 6125 3. 321644295 2015 0. 17 1. 61867534 0. 888931775 0 1. 876595272 3. 276332715 songtao 2013 0. 233279424 1. 756375498 0. 948684227 0 1. 4971 3. 524577178 2014 0. 1889 1. 529517598 0. 958264326 0 1. 8223 3. 293322063 2015 0. 142 1. 667713871 1. 150993333 0 2. 113165152 3. 299930293 xingren 2013 0. 207 1. 46647455 1. 57416672 0 1. 8413 3. 352642451 2014 0. 1612 1. 405961057 1. 807336957 0 2. 1951 3. 280693459 2015 0. 106 1. 484808227 2. 164888052 0 2. 700672571 3. 239052745 puan 2013 0. 196259181 1. 568819818 1. 17560454 0 1. 6741 3. 671681958 2014 0. 1517 1. 499900892 1. 381349206 0 1. 9705 3. 486646884 2015 0. 105 1. 551480865 1. 519460678 0 2. 338521401 3. 4370149 qinglong 2013 0. 374083742 1. 402714016 0. 889113754 0 1. 4956 3. 80012454 2014 0. 3277 1. 334983127 0. 948440208 0 1. 7919 3. 636463482 2015 0. 257 1. 373372342 1. 225945898 0 2. 223100806 3. 521828906 zhenfeng 2013 0. 25616614 1. 420402749 0. 83911166 0 2. 0521 3. 416154374 2014 0. 2065 1. 397700549 0. 990508832 0 2. 5059 3. 278937072 2015 0. 161 1. 331254098 1. 184476412 0 2. 981321802 3. 278937072 wangmo 2013 0. 3302 2. 19442433 1. 196548483 0 1. 0232 4. 049446524 2014 0. 2782 1. 955752212 1. 293567894 0 1. 4544 3. 718786016 2015 0. 212 1. 842509468 1. 4710172 0 1. 90233347 3. 659548717 ceheng 2013 0. 327299562 2. 392394123 1. 43018208 0 1. 2101 3. 864543206 2014 0. 2917 2. 218577348 1. 276337562 0 1. 5128 3. 677299011 2015 0. 209 1. 92618653 1. 38313203 0 1. 922140992 3. 628398759 anlong 2013 0. 18765398 1. 399533825 0. 671555203 0 1. 7737 3. 439123678 2014 0. 144 1. 383162218 0. 720708447 0 2. 0314 3. 289023989 2015 0. 094 1. 762609547 1. 121522008 0 2. 450489025 3. 297945393 56 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 continued table 2. area year pov fs fe fsr rgp ig huangping 2013 0. 300400356 1. 986044487 0. 551266679 0 1. 18148924 3. 757235792 2014 0. 279 1. 881645739 0. 686358447 0 1. 4245 3. 543672627 2015 0. 23 2. 060513209 0. 80948955 0 1. 6621 3. 496015936 shibing 2013 0. 315 1. 623027167 0. 901999773 0 1. 776330908 3. 314560106 2014 0. 2681 2. 065495792 0. 571290634 0 2. 091 3. 139756811 2015 0. 219 2. 076375092 0. 755611408 0 2. 3598 3. 12852186 sansui 2013 0. 337912808 1. 770186807 0. 721300959 0 1. 607073955 3. 540659015 2014 0. 2909 1. 628367413 0. 851790175 0 1. 9833 3. 352302205 2015 0. 238 1. 855583911 0. 958632101 0 2. 2779 3. 307514784 chenggong 2013 0. 263671106 1. 911758775 0. 901263781 0 1. 661290323 3. 541658895 2014 0. 2539 1. 759073842 0. 946095918 0 1. 9863 3. 347981771 2015 0. 199 2. 246878817 1. 348102786 0 2. 2779 3. 297069916 tianzhu 2013 0. 317560422 1. 73507733 0. 54547008 0 1. 887096774 3. 398569493 2014 0. 2765 1. 647038917 0. 588824354 0 2. 2678 3. 219562136 2015 0. 225 1. 680926742 0. 697734382 0 2. 5857 3. 18216399 jinping 2013 0. 327482746 2. 048268621 0. 538821123 0 1. 643229167 3. 826978339 2014 0. 2854 1. 954845089 0. 579051383 0 1. 9746 3. 58334794 2015 0. 237 2. 128669998 0. 670673218 0 2. 2632 3. 531924883 jianhe 2013 0. 324845698 1. 77445503 0. 69124272 0 1. 452532521 3. 734608888 2014 0. 2829 1. 718729955 0. 728682171 0 1. 725 3. 52583691 2015 0. 237 1. 858789305 0. 845340886 0 1. 9953 3. 488044145 taijiang 2013 0. 320123134 2. 614989224 1. 688534946 0 1. 671319226 3. 90386029 2014 0. 2878 2. 337837838 1. 781118461 0 1. 9982 3. 641577061 2015 0. 199 3. 097484388 2. 298387097 0 2. 2931 3. 602333495 liping 2013 0. 273840105 2. 167047478 0. 798506466 0 1. 213165338 3. 651010754 2014 0. 2448 1. 984454148 0. 896921017 0 1. 4719 3. 445584525 2015 0. 225 2. 084132192 1. 054982636 0 1. 73 3. 442538333 rongjiang 2013 0. 362219501 2. 158622636 1. 073681581 0 1. 236649215 3. 765490043 2014 0. 3373 1. 974154589 1. 063465319 0 1. 446 3. 53069719 2015 0. 276 2. 153062949 1. 326029216 0 1. 6868 3. 48019802 congjiang 2013 0. 316824119 1. 409779646 0. 963978359 0 1. 185019756 3. 532828416 2014 0. 2901 1. 343803263 0. 922477441 0 1. 4128 3. 345739471 2015 0. 239 1. 439978029 0. 993257164 0 1. 6614 3. 307121014 leishan 2013 0. 279830763 2. 037258157 0. 940628209 0 1. 495079161 3. 563792528 2014 0. 2648 2. 126787417 1. 006111111 0 1. 7947 3. 361312665 2015 0. 208 2. 337487647 1. 121662182 0 2. 0324 3. 313362702 majiang 2013 0. 376823379 1. 933068536 0. 921480557 0 1. 545126354 3. 713156697 2014 0. 3276 2. 3109319 0. 931235955 0 1. 828 3. 510221465 2015 0. 206 2. 828893229 1. 343370672 0 2. 1728 3. 447252245 danzhai 2013 0. 337322041 2. 374553522 0. 75112077 0 1. 474846626 3. 728992743 2014 0. 2924 2. 284910965 0. 835390947 0 1. 7435 3. 506482429 2015 0. 2381 2. 410324531 1. 003325346 0 1. 9967 3. 456406368 libo 2013 0. 309227981 1. 876319487 1. 005916103 0 2. 404984314 3. 238336279 2014 0. 2556 1. 793926247 1. 204545455 0 3. 2619 3. 044679005 2015 0. 2109 1. 970218543 1. 403795148 0 3. 562 3. 030951868 57 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 the basic statistical characteristics of the sample are shown in table 3. table 3. the basic statistical characteristics of the sample variable mean value standard deviation minimum value maximum value pov 0. 2364972 0. 0614049 0. 094 0. 3768234 fs 1. 840429 0. 6395251 0. 8968624 5. 28491 fe 1. 120168 0. 3989469 0. 5148047 2. 514095 fsr 0. 04 0. 1966157 0 1 rgp 1. 915396 0. 545185 0. 9767387 4. 5397 ig 3. 338166 0. 2438435 2. 847902 4. 049447 4.2. empirical test process 4. 2. 1. fixed effect model test poverty rate as the dependent variable is influenced by independent variables in the model: financial scale, financial efficiency and financial structure, the influence of the economic growth and income distribution, also will be affected by other factors associated with particular observation object or period. fixed effects model, the comprehensive effects of other factors as fixed, to a certain extent, improved the fitting degree of practical data. key county of guizhou province 50 poverty alleviation and development, is the main part of the relatively independent and unified individuals. in order to reflect the differences of the individuals in the whole, this paper first uses the fixed effect variable intercept model, and regression results are shown in table 4. table 4. fixed effect model regression results pov estimate of parameter standard error p>t fs -0. 05774 0. 0110099 0 fe -0. 01377 0. 0074121 0. 06 fsr omitted rgp -0. 11729 0. 0063662 0 ig 0. 02139 0. 0139213 0. 12 c 0. 51337 0. 0616472 0 the result is visible in table 4 that financial structure is omitted. it means the variables are multicollinearity. therefore, remove the variable from the model. the main reason for the financial structure appeared multicollinearity is due to the financial structure in calculating the index of listed companies, and listed companies in 50 state-level poverty-stricken counties in guizhou was small. so data is not ideal, and finally this paper had no way but to remove the index of financial structure. continued table 2. area year pov fs fe fsr rgp ig dushan 2013 0. 273012006 1. 97870742 0. 790070241 0 1. 621393159 3. 201440771 2014 0. 2273 1. 766224323 0. 826587038 0 2. 0422 2. 997296913 2015 0. 1759 2. 087012568 1. 127719347 0 2. 3035 2. 962288054 pingtang 2013 0. 315138198 1. 642500785 0. 996803011 0 1. 357098566 3. 260047272 2014 0. 2598 1. 514265503 0. 988072336 0 1. 776 3. 049054905 2015 0. 1955 2. 024569926 1. 483892488 0 1. 9885 3. 02705992 luodian 2013 0. 3551 1. 673145121 1. 121704211 1 1. 594785095 3. 051587356 2014 0. 3027 1. 508854782 1. 214974913 1 1. 9817 2. 863407821 2015 0. 2416 1. 675535871 1. 4811593 1 2. 2175 2. 847902003 changshun 2013 0. 318250439 1. 460728353 0. 976326357 0 1. 716578884 3. 180008698 2014 0. 2656 1. 303118202 1. 270614278 0 2. 2211 2. 979166667 2015 0. 2052 1. 736274723 1. 513155201 0 2. 4685 2. 930855856 sandu 2013 0. 356204934 1. 881020408 1. 053421689 0 1. 199058579 3. 203770577 2014 0. 323 1. 64083219 1. 074776386 0 1. 6361 2. 984166913 2015 0. 2526 1. 923600327 1. 260590897 0 1. 8281 2. 959970965 58 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 4. 2. 2. delete the common linear variables (1) hausman test remove the financial structure variables from original data. use stata11. 0 software to do hausman test . the results are shown in table 5. table 5. hausman test result hausman test chi(0)=0. 00 prob>chi2=0. 00 due to the p value is 0. 00, so strongly rejected the null hypothesis, it should use a fixed effect model to estimate more appropriate. (2) fixed effect model regression results deleted the financial structure from the original data , parameter estimation using stata11. 0 software, the results as shown in table 6. table 6. the regression results of the fixed effect model after the total linear variables were deleted pov estimate of parameter standard error p>t fs -0. 05830 0. 01101 0 fe -0. 01388 0. 00742 0. 064 fsr -0. 11753 0. 00637 0 rgp 0. 02141 0. 01394 0. 128 c 0. 51502 0. 06168 0 from table 6 shows that financial scale, financial efficiency and economic growth indicators of povertystricken counties in guizhou's poverty rate is negatively related. it illustrate the three indicators can slow the poor development of state-level poverty-stricken counties in guizhou. it accords with the real economic significance. the greater the income distribution gap, the higher the poverty rate in poverty-stricken counties in guizhou. both of them were positively correlated. financial scale, financial efficiency and economic growth indicators of p values are less than 0. 05, which is under the 95% confidence level were significantly. p value of income distribution was 0.128, and it shown that in 12. 8% of cases of income distribution have no effect on state-level poverty-stricken counties in guizhou's slow. it failed to pass the test of significance. so, not significant items will be deleted from the model. income distribution is too poor to slow. the reason mainly is income distribution index using urban per capita disposable income/income. per capita net income of rural income distribution should not only consider between urban income and rural income distribution, industrial management should also be considered. 4. 2. 3. delete distinctive item (1) hausman test remove the income distribution (ig) from the original data. using stata11. 0 software to do hausman test , the results are shown in table 7. table 7. the hausman test results were deleted without significant items hausman test chi(0)=0. 00 prob>chi2=0. 00 (2) fixed effect model regression results according to 2013-2015, 50 state-level poverty-stricken counties in guizhou raw data, to delete the income distribution, parameter estimation using stata11. 0 software, as shown in table 8 results are obtained. table 8. the regression results of the fixed effect model after the total linear variables were deleted pov estimate of parameter standard error p>t fs -0. 05769 0. 011084 0 fe -0. 01582 0. 007365 0. 03 rgp -0. 12475 0. 004328 0 c 0. 60159 0. 025240 0 from table 8, the regression results are significant, and the scale of financial development most important state-level poverty-stricken counties in guizhou's slow, and the regression equation is obtained: 4.3. the empirical result analysis from the point of empirical results, state-level poverty-stricken counties in guizhou about financial scale, financial efficiency and economic growth slow positively to poverty. expand the scale of financial and improve financial efficiency, promote economic growth will reduce the incidence of poverty, and it also can alleviate poverty effectively. ititititit rgpfefspov  12475.001582.005769.060159.0 59 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 from the regression equation of financial scale and financial efficiency of regression coefficient : financial scale relative to financial efficiency, it plays a more important role in slowing poverty. improving financial efficiency can give residents more store credit support services. and the expansion of financial development means that the expanding of the total number of financial institutions and financial, thus it is effective to relief poverty. state-level poverty-stricken counties in guizhou financial development on poverty relief play a positive and effective role. financial development in addition to the direct action slow poverty, and also alleviate poverty through economic growth. 5. conclusions although many literature research in the current financial relationship with poverty alleviation, but study of guizhou county to originality in this paper. in this paper, the results show that 50 state-level povertystricken counties in guizhou about financial scale, financial efficiency and economic growth slow positively to poverty. financial scale relative to financial efficiency, it plays a more important role in slowing poverty. . in addition, economic development is an important role in poverty reduction. the local government should develop the local economy, and poverty reduction effect will increase greatly. in this paper, the original selection of the income distribution is urban per capita disposable income/income and rural per capita net income. eventually the result was not significant. if this paper can use gini coefficient to show income distribution index of each county may be better. but with data is not available, so this paper cannot calculate the gini coefficient. it is a big regret. hope that in later research we can continue to improve and perfect the defects. references tan l l. rural financial and poor. jilin university, 2011. cui y j, sun g. the reason why financial development is slow poverty?evidence from china. journal of financial research, 2012, (11) : 116-127. wu y. poverty reduction effect of china's rural financial development research based on the analysis of national and subregional. journal of southwest university for nationalities (humanities and social science edition), 2012 (7) : 109-113. guo w. the experience of the rural financial poverty relief, the difficulties and countermeasures, in fuchuan county as an example of guangxi. journal of theoretical exploration, 2013, (5) : 98-102. song x, li l, xiao l. review of research on credit risk management for rural credit cooperatives. journal of risk analysis & crisis response, 2017, 7(1):21. deng k. financial poverty alleviation and efficiency evaluation in qinba mountains bazhong, for example. journal of rural economy, 2015, (5) : 8691. cui y j. china's financial development impact on poverty slow: theory and empirical. the northeast university of finance and economics, 2012. 60 ___________________________________________________________________________________________________________ journal of risk analysis and crisis response, vol. 8, no. 1 (march 2018) 52-60 microsoft word analytical research on the accident risk in three industries of china and usa in the period of 2006-2010 analytical research on the accident risk in three industries of china and usa in the period of 2006-2010 zongzhi wu china academy of safety science and technology beijing 100012, china shengzhu zhang, mingrong zeng china academy of safety science and technology beijing 100012, china e-mail:zhangshengzh5168@163.com abstract the related data analysis of china and usa shows that the work-related injuries are closely related to the proportion of each industry. high-incidence period is often in the period when the proportion of the secondary industry rises. increasing proportion of the tertiary industry does good to decrease the number of employees in highly risky industries and lower accident hazard. effective channels attributable to improving safety situation are to change industry structure, develop the tertiary industry and plan three industries development. key words: industrial structure; risk analysis; accident; safe development 2006-2010 年中国和美国三次产业事故风险分析研究 吴宗之 张圣柱* 曾明荣 中国安全生产科学研究院，北京 100012 摘要：中美两国相关数据分析表明，工伤事故情况与各产业的比重密切相关，事故高发期多发生在 第二产业比重上升期。第三产业比重的增加有利于减少高危行业从业人员数量、降低事故风险。转 变产业结构、大力发展第三产业、合理统筹三次产业发展是促进安全发展的有效途径。 关键词：产业结构，风险分析，事故，安全发展 1. 引言 安全生产水平是衡量一个国家现代文明程度 的重要指标。近年来，我国安全生产形势总体趋于 稳定，表现在事故总量下降，工矿商贸行业安全生 产状况趋于好转，建筑、交通、烟花爆竹和民爆器 材等相关行业和领域的安全工作取得成效；但另一 方面安全生产形势依然十分严峻，与欧盟、美国和 日本等发达国家相比，我国生产事故总量较大，重 特大事故频繁发生，职业危害严重。由对各产业事 故风险的分析可知，以采矿业、建筑业和制造业为 代表的第二产业其事故风险较高，第一产业和第三 产业的事故风险较低。大力发展低风险产业，尤其 是第三产业，不仅有利于促进经济结构的优化升 级、提高经济发展的现代化水平，同时由于大多数 服务业的事故风险较低、就业容量大，能在扩大就 业的同时改善安全生产状况。1-4 本文基于三次产业 的发展过程，对比现阶段我国与美国在安全生产方 面存在的差距，分析各产业事故风险特点，探讨转 变产业结构的重要性。 2. 产业事故风险分析 2.1. 我国三次产业事故风险分析 2006-2010 年我国工矿商贸事故按国民经济行 业划分十万人死亡率如表 1 所示。5, 6 journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 52-58 published by atlantis press copyright: the authors 52 z. wu et al 表 1 2006-2010 年各行业十万人死亡率 年份 类别 2006 2007 2008 2009 2010 5 年平均 农、林、牧、渔业 3.28 2.31 2.37 2.81 2.58 2.67 采矿业 135.55 114.05 100.51 75.33 65.91 98.27 制造业 9.98 10.10 9.69 8.58 8.15 9.30 电力、燃气及水的生产和供应业 10.12 10.48 11.42 9.33 6.76 9.62 建筑业 27.98 28.29 27.82 23.44 21.85 25.88 交通运输、仓储和邮政业 3.65 3.29 3.69 3.34 2.98 3.39 信息传输、计算机服务和软件业 3.12 2.92 3.33 2.76 1.99 2.82 批发和零售业 2.49 2.27 2.53 2.57 1.94 2.36 住宿和餐饮业 2.83 2.33 2.86 3.27 2.10 2.68 金融业 0.13 0.06 0.09 0.02 0.04 0.07 房地产业 2.28 2.18 2.42 2.62 1.84 2.27 租赁和商务服务业 7.30 9.01 6.64 5.40 4.48 6.57 科学研究、技术服务和地质勘查业 2.32 2.94 2.58 1.25 1.30 2.08 水利、环境和公共设施管理业 3.13 6.42 4.63 4.42 5.07 4.73 居民服务和其他服务业 50.40 75.20 58.96 51.53 40.37 55.29 教育 0.12 0.12 0.15 0.10 0.07 0.11 卫生、社会保障和社会福利业 0.20 0.25 0.24 0.13 0.05 0.17 文化、体育和娱乐业 1.63 1.60 3.86 3.24 2.89 2.64 公共管理和社会组织 0.02 0.11 0.09 0.11 0.06 0.08 全部行业平均 12.91 12.15 11.17 9.18 8.13 10.71 注：1.十万人死亡率由各行业死亡人数和就业人数求得，各行业死亡人数数据来源于国家安监总局 2006-2011 年事故统计 年报，各行业就业人数数据来源于 2006-2011 年中国劳动统计年鉴； 2.部分行业因未充分将农民工计入就业人数导致十万人死亡率偏高，统计数据不包括交通事故； 3.行业分类参照 gb/t 4754-2011《国民经济行业分类》。 由表 1 可知，我国不同行业事故风险差异巨大。 各行业 2006-2010 年 5 年平均十万人死亡率中，最 高的是采矿业，达 98.27，是最低的金融业(十万人 死亡率为 0.07)的 1400 余倍，其次分别是居民服务 和其他服务业、建筑业、电力、燃气及水的生产和 供应业以及制造业，除居民服务和其他服务业属第 三产业外，其他均属于第二产业。第一产业的十万 人死亡率较低。 第二产业中，采矿业的十万人死亡率是全行业 中最高的，建筑业的十万人死亡率为 25.88，电力、 燃气及水的生产和供应业与制造业的十万人死亡 率在 9～10 之间，第二产业事故风险最高。 第三产业中，居民服务和其他服务业的十万人 死亡率最高，达 55.29，远超过建筑业，其他行业 的十万人死亡率均较低，其中租赁和商务服务业的 十万人死亡率为 6.57，水利、环境和公共设施管理 业的十万 表 2 2006-2010 年我国部分行业一次死亡 10 人以上事故情况 时间 数据 类别 2006 2007 2008 2009 2010 事 故 起数 所 占 比例 (%) 事 故 起数 所 占 比例 (%) 事 故 起数 所 占 比例 (%) 事 故 起数 所 占 比例 (%) 事 故 起数 所 占 比例 (%) 煤矿和非煤矿山 41 42.7 30 35.3 41 42.3 24 35.8 26 30.6 建筑业 1 1.0 7 8.2 5 5.2 4 6.0 6 7.1 工矿商贸合计 48 50.0 44 51.8 50 51.5 33 49.3 37 43.5 注：事故数据来源于国家安监总局事故统计年报。 published by atlantis press copyright: the authors 53 analytical research on the accident risk 人死亡率为 4.73，交通运输、仓储和邮政业的十万 人死亡率为 3.39，信息传输、计算机服务和软件业、 批发和零售业、住宿和餐饮业、房地产业、科学研 究、技术服务和地质勘查业和文化、体计算机服务 和软件业、批发和零售业、住宿和餐饮业、房地产 业、科学研究、技术服务和地质勘查业和文化、体 育和娱乐业的十万人死亡率在 2～3 之间，卫生、 社会保障和社会福利业与教育的十万人死亡率更 低，在 0.1～0.2 之间，金融业与公共管理和社会组 织的十万人死亡率最低，均小于 0.1。第三产业中 的大多数行业事故风险远低于第二产业。 2006-2010 年我国部分行业一次死亡 10 人以上 事故情况如表 2 所示。 由表 2 可知，工矿商贸行业所发生一次死亡 10 人以上事故占全国所发生重特大事故（2006-2010 年分别为 96 起、85 起、97 起、67 起、85 起）的 比例一直在 50%左右，而煤矿和非煤矿山发生的一 次死亡 10 人以上事故又占工矿商贸行业重特大事 故的较大部分，表明煤矿和非煤矿山开采容易导致 重特大事故，给安全生产工作带来巨大负面影响， 建筑业和其他行业的重特大事故相对较少。 2.2. 美国三次产业事故风险分析 2006-2010 年美国各行业十万人死亡率如表 3 所示。 表 3 2006-2010 年美国各行业十万人死亡率 时间 数据 类别 2006 2007 2008 2009 2010 5 年平均 死亡 人数 (人) 死亡 率 死亡 人数 (人) 死亡 率 死亡 人数 (人) 死亡 率 死亡 人数 (人) 死亡 率 死亡 人数 (人) 死亡 率 死亡 人数 (人) 死亡 率 1.商品生产 农林渔牧业 655 30.0 585 27.9 651 29.4 551 26.0 621 27.9 613 28.24 采矿业 192 28.1 183 25.1 175 18.0 101 12.7 172 19.8 165 20.74 建筑业 1239 10.9 1204 10.5 969 9.6 816 9.7 774 9.8 1000 10.10 制造业 456 2.8 400 2.5 404 2.5 304 2.2 329 2.3 379 2.46 2.服务业 批发贸易 222 4.9 207 4.7 175 4.2 186 4.9 191 4.9 196 4.72 零售贸易 359 2.2 348 2.1 290 2.0 291 2.1 311 2.2 320 2.12 运输仓储 860 16.8 890 16.9 762 14.2 579 12.1 661 13.7 750 14.74 信息与传媒 66 2.0 79 2.3 45 1.4 32 1.1 43 1.5 53 1.66 金融、保险、地产、 租赁等 126 1.2 119 1.2 103 1.0 101 1.1 113 1.3 112 1.16 科技、管理及行政 服务 459 3.2 476 3.1 389 2.7 394 2.9 364 2.6 416 2.90 教育、健康及社会 服务 178 0.9 149 0.7 137 0.7 136 0.7 171 0.9 154 0.78 艺术、娱乐及住宿 餐饮 265 2.3 260 2.2 233 2.2 216 2.1 238 2.3 242 2.22 其他 183 2.6 175 2.5 172 2.5 166 2.7 192 3.0 178 2.66 3.政府公共管理 520 2.4 545 2.5 522 2.3 450 1.9 484 2.2 504 2.26 注：1.数据来源于美国劳工局 2006-2011 年的统计数据； 2.数据不包含军事领域，从业人员年龄不小于 16 周岁。 由表 3 可知，美国第一产业的十万人死亡率最 高，达 28.24，就第二产业和第三产业与我国相比 较，尽管美国总体事故死亡人数远少于我国，但其 各行业事故风险特点与我国相似。采矿业事故导致 的死亡人数虽较少，但由于从业人数少，其十万人 死亡率非常高，平均达 20.74，建筑业事故造成的 死亡人数最多，达 1000 人/年，十万人死亡率也较 高，平均达 10.10，制造业事故风险较低，十万人 死亡率为 2.46。第三产业中，除运输仓储业的十万 人死亡率(达 14.74)较高外，其他行业的十万人死亡 率均较低，其中批发贸易的十万人死亡率为 4.72， 科技、管理及行政服务、政府公共管理、艺术、娱 乐及住宿餐饮和零售贸易的十万人死亡率在 2～3 之间，信息与传媒和金融、保险、地产、租赁等行 业的十万人死亡率在 1～2 之间，教育、健康及社 会服务业的十万人死亡率小于 1。总体上讲，第二 产业事故风险远高于第三产业。 published by atlantis press copyright: the authors 54 z. wu et al 2.3. 产业事故风险差异性 不同产业在工作环境、作业条件、工作场所作 业人员密度、生产过程中所涉及的危险物质的及数 量等方面存在的差别，决定其具有不同的事故风 险。第二产业中的采矿业、建筑业和制造业的行业 特点决定其具有高事故风险。例如煤矿井下存在瓦 斯、水害等灾害因素，工作环境复杂，作业条件特 殊，且大量工人处于同一采区进行采掘，易导致重 特大事故；建筑业因需要进行高空作业、施工过程 需要大量起重机械设备以及复杂的地质条件而容 易引发高处坠落事故、机械伤害事故和坍塌事故； 复杂的制造业因生产工艺常常涉及高温、高压，生 产或使用危险材料而容易引发火灾、爆炸和毒物泄 漏扩散等事故。而第三产业的大多数行业特点决定 其具有低事故风险，由于人员在低风险环境中工 作，所接触的物质危险性低，引发人身伤害事故的 概率和严重度均较小。 3. 第二产业部分行业风险分析  2006-2010 年，煤矿、非煤矿山、建筑业和制 造业四个事故风险高的行业每年事故死亡人数（如 表 4 所示）占工矿商贸领域事故死亡人数的总量分 别为 88.9%、87.1%、87.1%、86.1%和 88.9%。 表 4 2006-2010 年我国工矿商贸部分行业事故死亡人数 年份 行业 2006 2007 2008 2009 2010 煤矿 4746 3786 3215 2631 2433 非煤矿山 2277 2188 2068 1542 1271 建筑业 2546 2722 2702 2760 2769 制造业 3244 3393 3225 2993 2966 工矿商贸合计 14412 13886 12865 11532 10616 2006-2010 年，我国原煤产量由 23.25 亿吨增长 到 32.5 亿吨，年均增长 8.7%，煤矿事故死亡人数 由 4746 人下降到 2433 人，年均下降 15.4%，原煤 生产在满足国民经济快速发展对能源大量需求的 同时，安全生产形势逐步好转，但事故死亡人数与 重特大事故起数仍较多（2006-2010 年煤矿重特大 事故起数分别为 39、28、38、20 和 24），尤其是 特别重大事故给煤矿安全生产带来较大负面影响。 2006-2010 年我国原煤产量与煤矿事故死亡人数如 图 1 所示。 0 5 10 15 20 25 30 35 2006 2007 2008 2009 2010 0 1000 2000 3000 4000 5000 原煤产量(亿吨) 死亡人数(人) 图 1 2006-2010 年我国原煤产量与煤矿事故死亡人数 2006-2010 年，我国建筑业实现增加值由 12409 亿元增长到 26451 亿元，年均增长 20.8%，建筑业 在实现增加值翻番的同时,建筑事故死亡人数由 2546 人增长到 2769 人，事故死亡人数居高不下、 波动较小，事故死亡人数呈上升趋势，安全形势严 峻，安全生产问题十分突出。2006-2010 年我国建 筑业增加值与建筑事故死亡人数如图 2 所示。 published by atlantis press copyright: the authors 55 analytical research on the accident risk 0 5000 10000 15000 20000 25000 30000 2006 2007 2008 2009 2010 2400 2500 2600 2700 2800 建筑业增加值 (亿元) 死亡人数(人) 图 2 2006-2010 年我国建筑业增加值与建筑事故死亡人数 由以上数据分析可知，近 5 年来，我国经济在 实现较快增长的同时，安全生产状况总体稳定，但 部分行业安全生产形势不容乐观，表现在建筑业安 全生产形势十分严峻，制造业事故死亡人数较多， 采矿业安全生产状况虽实现逐步好转，但事故死亡 人数和重特大事故起数依然较多，这 3 个行业仍然 是事故多发的高风险行业。 4. 三次产业的发展过程与事故变化趋势 4.1. 我国的产业发展过程与安全生产状况 建国初期，第一产业比重最大，农业实现的增 加值最高，为了发展经济，我国开始制订并实施 5 年规划，优先发展重工业，使第二产业的比重快速 上升，第一产业的比重逐步下降，1950-1977 年间， 第一产业和第二产业的比重波动较大，并呈此消彼 改革开放后的 1978-1984 年间，我国对产业结构进 行了调整，1984 年第一、第二和第三产业的比重分 别为 32.1%、43.1%和 24.8%，总体特点是第一产业 比重提高，第二产业比重趋于下降；1985-1992 年 间，第三产业的比重迅速上升，1992 年第一、第二 和第三产业的比重分别为 21.8%、43.4%和 34.8%； 1993 年至今，我国产业结构变化的总体特点是第一 产业的比重不断下降，第二、第三产业的比重不断 上升。近十年来，我国经济处在高速发展时期，国 内生产总值由 1999 年的 82054 亿元增长到 2010 年 的 397983 亿元，年均增长 15.4%，经济总量是 1999 年的近 5 倍，其中第二产业的增加值一直占国内生 产总值的 50%左右。7-10 1949-2010 年我国三次产业 所占比重与工矿商贸事故死亡人数的变化如图 3 所 示。 长的态势，第三产业的比重变化较小。 年份 三 次 产 业 所 占 g d p比 重 / 工 伤 事 故 死 亡 人 数 / 人 工伤事故死亡人数 第三产业产值比重 第二产业产值比重 第一产业产值比重 24000 22000 20000 16000 18000 10000 8000 12000 14000 6000 4000 0 2000 50 70 60 30 40 20 10 1981 1985 19931989 2005 2009200119971965 1969 197719731957 196119531949 图 3 1949-2010 年我国三次产业所占比重与工矿商贸事故死亡人数 published by atlantis press copyright: the authors 56 z. wu et al 由图 3 可知，建国以来我国工矿商贸事故死亡 人数的波动与第二产业的变化趋势基本相同，5 次 事故高峰（1960 年、1973 年、1979 年、1995 年、 2003 年）均出现在第二产业比重上升时期，第二产 业比重下降时，安全形势又逐步实现好转。 近年来，随着加快转变经济发展方式和调整经 济结构、大力发展现代农业、改造提升制造业、加 快发展服务业等产业结构调整目标的提出 11, 12 ，第 二产业所占比重不断减少，第三产业所占比重持续 增加。与此同时，第二产业不断转型升级，针对煤 矿，提出了建设大型煤炭基地、推进煤炭企业联合 重组等导向政策，对于制造业，提出了淘汰落后产 能、优化结构、发展先进装备制造业等转型升级方 针，对于危险化学品，提出了按照基地化、大型化、 一体化方向调整化学工业布局，对于建筑业，严格 了行业准入条件。第三产业的持续快速发展和第二 产业内部的转型升级使得安全生产状况持续稳定 好转。 4.2. 主要发达国家的产业发展过程与安全生况 目前，第三产业在美国全部经济活动中的比重 已超过 80%，在欧洲和日本经济中所占的比重已超 过 70%，并呈上升趋势。美国三次产业发展过程中， 在第二产业所占比重较大时期，建筑业、制造业等 产业高速发展，出现过事故频发、工伤人数居高不 下的情况，随着第三产业所占比重的增加，安全生 产形势逐步改善，现在工伤事故已经下降到较低的 水平。日本也经历了同样的过程，20 世纪 50 年代 以后，随着制造业的兴起，其安全生产问题突显， 工伤事故死亡人数剧增，l961 年工伤事故死亡人数 达到了历史最高的 6712 人，1956-1973 年完成了重 化工业的调整，1974-1980 年完成了知识密集化和 高附加价值化的调整，随着产业的不断发展、第三 产业所占比重的增加，以及所采取的相关技术、监 管等措施，工伤事故逐步得到控制，从 1998 年起 其工伤死亡人数已控制在 2000 人以下。13-16 4.3. 三次产业发展与事故变化趋势的关系 由国内外三次产业的发展过程与事故变化趋 势的分析可知，在第二产业迅速发展、所占比重较 大时期，工伤事故伤亡人数多，安全生产形势严峻， 当进行产业结构调整并使第三产业的比重明显增 加后，安全生产形势明显改善，大力发展第三产业 有利于减少高风险行业从业人数，规避高风险行 业。 5. 结束语 不同行业的特点决定了其固有的伤亡事故风 险，采矿业、建筑业和制造业是现阶段我国事故多 发的高风险行业，3 个行业的事故死亡人数占工矿 商贸领域事故死亡人数的 86%以上。 安全生产水平与产业发展状况密切相关，事故 高发期多出现在第二产业比重的增加期，而第三产 业比重的增加有助于改善安全生产状况。目前我国 正处于产业结构调整的关键时期，大力发展第三产 业，提高第三产业的比重和水平，减少高风险行业 从业人数，是改善我国安全生产状况的根本途径。 现阶段第二产业仍是我国国民经济的重要支 柱，比重仍较大，在短时间内难以根本改变，应继 续推动第二产业转型升级，依靠科技进步和技术创 新，淘汰落后工艺、技术与装备，优化原材料工业， 发展先进装备制造业，提升消费品工业，对于第二 产业内的高危行业，实行严格的安全准入制度和从 业人员安全资格制度，构建安全、稳定的第二产业 体系。 大力发展安全产业，依靠安全科技和安全装 备，提高三次产业的安全保障能力，形成安全检测 监控、应急救援装备、个体防护用品、本质安全化 装备与材料、安全培训、评价与咨询服务等产业， 实现安全发展。 致谢 本论文的研究工作得到了国家安全生产监督 管理总局课题“安全发展本质内涵及政策措施研 究”的资助。 参考文献 1. general administration of quality supervision, inspection and quarantine of the people’s republic of china, gb/t 4754-2011 sector classification of national economy. (china standards publishing house, beijing, 2011). 国家质量监督检验检疫总局. gb/t 4754-2011 国民经 济行业分类. 北京: 中国标准出版社, 2011. 2. z. z. wu, discussion on the understanding and countermeasures of promoting work safety development, journal of safety science and technology, s(4) (2008): 118-122. 吴宗之, 促进安全发展的若干认识与对策探讨, 中国 安全生产科学技术, s(4) (2008): 118-122. 3. z. z. wu, study on some strategy problems of china’s work safety by 2020, journal of safety science and technology, 3(1) (2007): 3-7. 吴宗之, 面向 2020 年我国安全生产的若干战略问题 的思考, 中国安全生产科学技术, 3(1) (2007): 3-7. 4. x. z. wang , report on occupational safety vs economic and social development. (china coal industry publishing house, beijing, 2006). 王显政, 安全生产与经济社会发展报告, 北京: 煤炭 工业出版社, 2006. 5. national bureau of statistics of china, statistical communiqué of national economic and social development of the people's republic of china from 1999 to 2011. 中华人民 共和国国家 统计局 , 中华 人 民共和国 1999-2011 年国民经济和社会发展统计公报. 6. state administration of work safety, accident statistical communiqué of state administration of work safety from 2006 to 2011. published by atlantis press copyright: the authors 57 analytical research on the accident risk 国家安全生产监督管理总局, 2006-2011 年国家安全 生产监督管理总局事故统计年报. 7. y. n. li, the relationship between economic reforms, economic growth and industrial structure adjustment, the journal of quantitative & technical economics, (12) (1988): 3-9. 厉以宁, 经济改革、经济增长与产业结构调整之间的 关系, 数量经济技术经济研究, (12) (1988): 3-9. 8. s. l. zhou, a few questions of china’s industrial structure adjustment and upgrading, china industrial economics (7) (1998): 22-29. 周叔莲, 我国产业结构调整和升级的几个问题, 中国 工业经济, (7) (1998): 22-29. 9. y. y. song, w. r. pan, discussion of the characteristics and evolvement direction of current industrial structure of china, hlj foreign economic relations & trade (1) (2010): 59-60. 宋迎迎,潘文荣, 论我国当前产业结构的特点及演化 方向, 黑龙江对外贸易, (1) (2010): 59-60. 10. l.song, dynamic econometric analysis on relativity between accident disasters and economic growth & its application, doctor thesis, beijing: china university of mining and technology(beijing), (2010): 61-72. 宋利, 事故灾害与经济增长关联性动态计量分析技 术及应用, 中国矿业大学（北京）博士学位论文[d], (2010): 61-72. 11. 2011 年两会温家宝总理政府工作报告全文[eb/ol]. 2011-03-07.http://news.hexun.com/2011-03-07/127754 083.html 12. research group of institute of industrial economics cass, a study on the adjustments and upgrades to china’s industrial structure during the 12th five-year plan period, china industrial economics (1) (2010): 5-23. 中国社会科学院工业经济研究所课题组, “十二五”时 期工业结构调整和优化升级研究, 中国工业经济, (1) (2010): 5-23. 13. c. x. jin, m. wang, the new trend of industrial structure adjustment in major advanced countries in the world, management. (1-2) (2008): 99-100. 金成晓, 王猛, 世界主要发达国家产业结构调整的新 趋势,管理, (1-2) (2008): 99-100. 14. q. yuan, c. y. liu, adjustment of industrial structure and development of service industry in the u. s. world economy study .(2) (2007): 57-63. 袁奇, 刘崇仪, 美国产业结构变动与服务业的发展, 世界经济研究, (2) (2007): 57-63. 15. w. l. zhao, c. b. wu, change of industrial structure and the function of foreign capital in the asian-pacific region. (xiamen university press, xiamen, 1992) 赵文骝, 吴崇伯, 亚太地区产业结构变化与外资的作 用, 厦门: 厦门大学出版社, 1992. 16. d. y. song, t. z. zhang, optimize industrial development through environmental protection taking japan’s industrial restructuring for example techno economics & management research. (3) (2012): 115-119. 宋丹瑛, 张天柱, 论资源环境优化产业升级—以战后 日本产业结构调整为例, 技术经济与管理研究, (3) (2012): 115-119. published by atlantis press copyright: the authors 58 microsoft word volume 12, issue 1-5 journal of risk analysis and crisis response, 2022, 12(1), 55-61 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i1.322 55 review a literature review on the transmission mechanism of innovation and entrepreneurship for digital finance to effectively support the high-quality development of real-economy wenyan wang 1,2 and hongmei zhang 1,2,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhm1035@qq.com; tel.: +86-0851-88510575 received: september 9, 2021; accepted: march 31, 2022; published: april 15, 2022 abstract: from the perspective of innovation and entrepreneurship, we have collected relevant research results in the past few years (2018–present). based on the collected data, a significant amount of research has been conducted up to the year 2022, indicating that experts and scholars are increasingly focused on the transmission of digital finance to a higher standard of development of the real-economy. existing research is largely devoted to the transmission mechanisms of "digital finance innovation and entrepreneurship" or "innovation and entrepreneurship — the high-quality development of the real-economy". there is little overview of "digital finance—innovation and entrepreneurship—the quality development of the real-economy." in this paper, we review relevant literature in the context of innovation and entrepreneurship, which may provide some useful guidance for improving the digital financial system and enhancing the growth of the real-economy. keywords: digital finance; innovation and entrepreneurship; high-quality development; realeconomy; literature review 1. introduction in march 2021, at the two sessions in beijing, president xi jinping emphasized that the 14th fiveyear plan period marks the beginning of a new journey toward building a modern socialist society. the upcoming journey calls for a clear understanding of the development stage, progress toward the implementation of a novel development concept, acceleration of the development of a novel development pattern, and the facilitation of quality development. according to the outline of the 14th five-year plan and vision 2035 for the national economic and social development of the people's republic of china, it is necessary to increase innovation in cooperation models in the realeconomy. digital finance can eliminate corporate financing constraints, encourage innovation and entrepreneurship, and promote the high-quality development of the real-economy. from the perspective of innovation and entrepreneurship, this paper brings together relevant research results from 2018 to present, which are categorized into two sections: the impact of digital finance on innovation and entrepreneurship, and the impact of innovation and entrepreneurship on wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 56 the high-quality development of the real-economy. the data for this paper is adapted from the cssci chinese social sciences citation index. 2. effects of digital finance on innovation and entrepreneurship 2.1. effect of digital finance on innovation the contribution of digital finance to innovation can be divided into two aspects: macro and micro. the study provides insight into how digital finance can drive regional innovation at the macro level. on the one hand, digital finance eases the pressure on financing directly from the supply side. regional innovation is supported by the extent and depth of digital finance, while the role of digitization is less clear. digital finance plays a significant role in driving innovation in eastern china rather than in central and western china, and its impact is greater in eastern regions or in firstand second-tier cities with greater human capital and economic sophistication. on the other hand, digital finance indirectly contributes to regional innovation capacity by promoting industrial structure upgrading and expanding regional consumption demand [1-4]. the development of digital finance also involves spatial spillover effects and threshold effects. the spatial spillover effect manifests itself in that digital finance stimulates innovation activities in the region, which in turn enhances innovation capacity in neighboring cities [5]. the threshold effect shows that when the threshold value is exceeded, the marginal benefit of the development of digital finance on the innovation impact of cities increases, and its innovation incentive effect shows increasing marginal benefits as the level of regional economic development improves [6-7]. the study also focuses on the role played by digital finance in promoting technological innovation on a micro scale. firstly, digital finance can address the traditional information asymmetry between banks and enterprises, provide insight into enterprises, and eliminate the costs associated with audits of the financial services sector by utilizing big data, the internet of things, and cloud computing. contrary to the superior financing environment for large businesses, smes are more dependent on the financial markets and more sensitive to financial constraints. in addition to improving the availability of funds for "long-tail enterprises," digital means of finance can effectively relieve the pressure of liquidity constraints on enterprises and promote enterprise innovation. by advancing digital finance, shareholders can lessen their financing constraints, address the risk of equity pledges, and enhance corporate governance, thus promoting corporate innovation investments. it is more pronounced among privatized firms and non-big 4 audited firms [8]. it is true that the role of digital finance cannot be separated from the role of effective financial regulation, and in regions with strong financial regulations, its role in promoting corporate innovation is even more significant, not only in terms of "quantity," but also in terms of "quality" of innovation [9]. secondly, digital finance has a positive impact on technological innovation. digital finance can displace the problems of attribute mismatches and domain mismatches seen in traditional finance, allowing for both incremental and breakthrough innovation, and the effect of breakthrough innovation is greater than that of incremental innovation [10-11]. thirdly, digital finance has a significant role to play in all dimensions of the high-quality industrial development, including green transformation, government welfare improvement, and efficiency enhancement [12-13]. the incentive effect of the development of digital finance on corporate green innovation is more pronounced in regions where wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 57 environmental regulations are stronger or economic development is lower in central and western china [14]. 2.2. effect of digital finance on entrepreneurship one can divide the effects of digital finance on entrepreneurship into two categories: those affecting residents' entrepreneurship and business entrepreneurship. recently, research has focused on the impact of digital finance on urban entrepreneurship and farmers' entrepreneurship. on the one hand, farmers who lack credit information and collateral assets face higher financing costs when they begin their businesses. they would rely on cost-effective digital finance to overcome these obstacles [15]. on the other hand, the "pro-poor" nature of digital finance in supporting entrepreneurship is significant. regarding the rural grassroots, the development of digital finance has helped to boost the entrepreneurial behavior of rural residents and equalize entrepreneurial opportunities, thus contributing to inclusive growth in china [16]. entrepreneurship plays a mediating role between digital financial development and shared prosperity, which can also moderate the non-linear spillover effects of digital finance [17]. digital finance is a tool for promoting entrepreneurship by providing financial support to entrepreneurs and easing the financial burdens they face in starting businesses. meanwhile, new enterprises of all types create jobs and power economic development, thus achieving inclusive growth. there is a greater effect of digital finance in promoting entrepreneurial activity among micro and small businesses than it does on medium and large companies [18]. digital finance and entrepreneurial activities also exhibit strong spatial autocorrelations and spillover effects; they form similar spatial agglomeration areas in terms of distribution, and agglomerations of entrepreneurial activities are generally also agglomerations of digital finance. in addition, the development of digital finance in the eastern region has a significant spillover effect on entrepreneurial activities in neighboring provinces, whereas the indirect effects on entrepreneurial activities in the central and western regions are more substantial [19]. 3. effects of innovation and entrepreneurship on the high-quality development of the realeconomy 3.1. impact of innovation on the high-quality development of the real-economy innovation can be categorized as regional innovation and technological innovation; each had a unique impact on the high-quality development of the real-economy. several studies have demonstrated that regional innovation has a positive effect on the high-quality economic development, including spatial spillover and agglomeration effects. in terms of spatial spillovers, innovation activities contribute not only to the high-quality development of the real-economy in the region where they are located, but also to growth in surrounding areas. the agglomeration effect is reflected in the fact that the performance level of innovation-driven the high-quality development varies among regions, with the eastern region significantly outperforming the central and western regions, and the regions with higher overall efficiency assessment values are mostly economically developed regions with strong innovation capabilities [20]. the experience of economic development in developed countries shows that technological innovation has a significant driving capacity for the high-quality development and economic wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 58 development poles [21]. on the one hand, technological innovation can contribute to the upgrading of the industrial structure by participating in the shift from an economy dependent on traditional factors to one based on innovation. in turn, the upgrading of industrial structures will further promote technological innovation, thus promoting the high-quality development of the realeconomy. it means that there is a synergistic effect between technological innovation and industrial structure upgrading, and this synergistic effect helps to contribute to the high-quality economic development [22]. on the other hand, the impact of industrial structure upgrading on the highquality development is different from one region to another. in the eastern region, the upgrading of the industrial structure is the main force behind the quality development of its entity economy; in the central region, on the contrary, it hinders the quality development of its entity economy; in the western region, this does not significantly affect the quality development of its entity economy. while diversification of innovation during periods of high growth can contribute to economic growth, in periods of high-quality development, the role of innovation specialization in stimulating economic growth is even more pronounced [23]. moreover, there is a threshold effect of technological innovation in high-tech industries on the high-quality development of the real-economy, in which the effect of technological innovation efficiency in high-tech industries on the quality of economic development is significantly positive until a threshold is reached, and its effect diminishes afterwards [24]. 3.2. impact of entrepreneurship on the high-quality development of the real-economy mass entrepreneurship is empowered by digital finance, which enhances both the vitality and creativity of the real-economy, and leads to high-quality economic growth. enterprise activities can change traditional economic growth models, enable industrial upgrading, and structural transformation. in addition, digital finance can also promote the equalization of entrepreneurship opportunities, lower the threshold for social entrepreneurship, and promote the inclusive growth of the real-economy. the strengthening of human capital and the stimulation of entrepreneurial dynamism enhance the industrial structure and create favorable conditions for the high-quality development of the real-economy [25]. furthermore, knowledge-intensive entrepreneurship also adds new impetus to the high-quality development of the real-economy of the country [26]. through the combination of digital finance and commodity markets, entrepreneurs can easily access quality innovation resources and thus achieve high levels of entrepreneurship. high-level entrepreneurial activity play a crucial role in improving the quality of economic growth by stimulating market vitality and improving the "quality" of the real-economy [27]. 4. conclusions and recommendations the research selected relevant papers from the cssci database between 2018 and the present, and after an analysis of the literature, the findings are as follows: firstly, innovation is a crucial element in transmitting digital finance to facilitate high-quality development of the real-economy. there are two types of innovation: regional innovation and technological innovation. through spatial spillovers and agglomeration effects, regional innovation drives the high-quality economic progress in the region and surrounding areas; technological innovation drives the digital transformation of non-financial enterprises and makes them more innovative to achieve the high-quality economic growth. wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 59 secondly, entrepreneurship is another effective path for digital finance to promote the highquality growth in the real-economy. entrepreneurship promotion can be categorized into two categories: residential entrepreneurship and corporate entrepreneurship. digital finance is inherently inclusive, and by empowering the masses to start their own businesses, it promotes equality and a higher level of entrepreneurship in the real-economy, which creates new market opportunities and encourages inclusive growth. finally, the sharing, security, and low threshold features of digital finance enable digital financial institutions to provide universal and precise services that traditional financial institutions cannot provide within the real-economy. the model is well suited to the requirements of the highquality development. taking advantage of the resource allocation role of digital finance, it transforms financial services into a popular service capable of serving the real-economy [28-29]. all forms of financial innovation, including digital finance, cannot escape the logical relationship among financial innovation, financial risk, and financial regulation. however, financial regulation often lags. there is the potential for economic volatility due to the varying levels of education of the population receiving financial support through digital finance and the false boom of some start-up businesses. the following recommendations are therefore proposed: firstly, it is essential to integrate financial services and digital technology. building a digital financial infrastructure, identifying exogenous conditions that stimulate strategic digital change in enterprises, and providing references for accelerating the development of a novel pattern of digital transformation, which would facilitate the high-quality economic development [30]. secondly, it is important for governments to formulate policies and institutions for green innovation, optimize the leading mechanisms for green innovation, and encourage the development of a high-quality green innovation industry. in order to play an effective role in transferring the "coordination and guidance" effect of digital finance, it is necessary to enhance the two-way link between the financial market and the environmental protection industry using digital technology platforms. finally, the reform of the digital financial system should be further enhanced to promote the sustainability of the financial industry. the cultivation of high-quality talent in the financial sector has the potential to increase its competitiveness. improve laws and regulations to facilitate a standardized and orderly development of the digital finance industry. funding: this research was funded by the special project of key cultivation disciplines and urgently needed disciplines of guizhou university of finance and economics: "research on credit risk prediction and evaluation of big data enterprises" (2020zjxk20) and the guizhou provincial postgraduate research fund project: "research on digital inclusive finance and multidimensional poverty alleviation in guizhou province: an empirical analysis based on data from 88 counties in guizhou province" (yjskyjj [2021] 127). conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] liu jiaxin, li sha. the development of digital finance and the improvement of regional innovation level in the context of "double cycle"[j]. economic issues, 2021(06):24-32. doi: https://doi.org/10.16011/j.cnki.jjwt.2021.06.004. wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 60 [2] du chuanzhong, zhang yuan. the regional innovation effect of digital finance in the context of "new infrastructure"[j]. finance and economics science, 2020(05):30-42. [3] ren biyun, liu jiaxin. digital inclusive finance development and regional innovation level enhancement-an analysis based on the perspective of internal supply and external demand[j]. journal of southwest university for nationalities (humanities and social sciences edition), 2021,42(02):99-111. doi: https://doi.org/10.3969/j.issn.1004-3926.2021.02.014. [4] zheng wanteng, zhao hongyan. can digital financial development drive regional technological innovation convergence? --empirical evidence from 284 cities in china[j]. contemporary economic science, 2021,43(06):99-111. [5] li linhan, tian weimin. financial deepening, technological innovation and green economy[j]. finance and economics, 2020(03):68-75. doi: https://doi.org/10.19622/j.cnki.cn36-1005/f.2020.03.009. [6] xu zhangxing, zhang bing, liu dan. digital financial development, corporate credit mismatch and labor employment a moderating mediating effect[j]. finance and economics series, 2020(12):40-49. doi: https://doi.org/10.13762/j.cnki.cjlc.2020.12.005. [7] wang wenqian, zhang yu. financial structure, industrial upgrading and economic growth a perspective of technological progress based on different characteristics[j]. the economist, 2022(02):118-128. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2022.02.013. [8] li yukun, ren haiyun, zhu danfeng. digital finance, equity pledges and corporate innovation investment[j]. research management, 2021,42(08):102-110. doi: https://doi.org/10.19571/j.cnki.10002995.2021.08.012. [9] sheng mingquan, zhang yue, wang shun. can the development of digital finance boost the technological innovation of traditional industrial enterprises[j]. statistics and information forum, 2021,36(12):12-22. doi: https://doi.org/10.3969/j.issn.1007-3116.2021.12.002. [10] tang song, wu xuchuan, zhu jia. digital finance and corporate technology innovation structural characteristics, mechanism identification and differences in effects under financial regulation[j]. management world, 2020,36(05):52-66+9. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2020.0069. [11] zheng yuxi, yang rong, mohammad heydari. does digital finance promote breakthrough or incremental innovation?[j]. journal of yunnan university of finance and economics, 2022,38(02):49-69. doi: https://doi.org/10.16537/j.cnki.jynufe.000766. [12] cui gengrui. digital finance and high-quality industrial development[j]. journal of southwest university for nationalities (humanities and social sciences edition), 2022,43(02):136-144. [13] jiang jianxun, tang yuchen, li xiaojing. green innovation of new energy enterprises empowered by digital finance in the context of double carbon: based on the perspective of financing constraints[j/ol]. contemporary economic management:1-19 [2022-04-15]. http://kns.cnki.net/kcms/detail/13.1356.f.20220311.1344.006.html. [14] zhai huayun, liu yisi. research on the relationship between digital finance development, financing constraints and green innovation of enterprises[j]. science and technology progress and countermeasures, 2021,38(17):116-124. doi: https://doi.org/10.6049/kjjbydc.2021030376. [15] zhang biqiong, wu wanting. digital inclusive finance, entrepreneurship and income distribution: an empirical study based on the perspective of urban-rural differences in china[j]. financial review, 2021,13(02):31-44+124. [16] zhang xun, yang tong, wang chen, wan guanghua. digital financial development and residential consumption growth: theory and practice in china[j]. management world, 2020,36(11):48-63. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2020.0168. [17] han liangliang, peng yi, meng qingna. digital inclusive finance, entrepreneurial activism and common wealth: an empirical study based on inter-provincial panel data in china[j/ol]. soft science:1-18 [2022-0415]. http://kns.cnki.net/kcms/detail/51.1268.g3.20220218.1626.004.html. [18] xie huali, shen yan, zhang haoxing, guo feng. can digital finance promote entrepreneurship? --evidence from china[j]. economics (quarterly), 2018,17(04):1557-1580. doi: https://doi.org/10.13821/j.cnki.ceq.2018.03.12. [19] gong xin, tang wenlin. digital finance, spatial spillover and mass entrepreneurship[j]. statistics and information forum, 2021,36(05):71-81. doi: https://doi.org/10.3969/j.issn.1007-3116.2021.05.007. wenyan wang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 55-61 doi: https://doi.org/10.54560/jracr.v12i1.322 61 [20] wang huiyan, li xinyun, xu yinliang. research on the performance evaluation and influencing factors of high-quality development of china's economy driven by science and technology innovation[j]. the economist, 2019(11):64-74. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2019.11.008. [21] shi qi, liang li, zhou ning. identification of high-quality industrial activities and prospects of industrial development in major growth poles[j]. the economist, 2022(02):86-96. doi: https://doi.org/10.16158/j.cnki.51-1312/f.2022.02.010. [22] ren xiaoyan, yang shuihui. technological innovation, industrial structure upgrading and high-quality economic development: a measurement analysis based on independent and synergistic effects[j]. east china economic management, 2020,34(11):72-80. doi: https://doi.org/10.19629/j.cnki.34-1014/f.200317004. [23] zhang jiyang, duan jihong, shi lei. attributes of binary innovation activities in high growth and high quality development stages: diversification or specialization? [j/ol]. contemporary economic management:1-21 [2022-04-15]. http://kns.cnki.net/kcms/detail/13.1356.f.20220308.1348.004.html. [24] ma yu, qiu wanhua, wang xinyu. study on the effect of high-tech industry agglomeration and technological innovation on high-quality economic development based on panel smoothing regression model[j]. industrial technology economics, 2020,39(02):13-20. doi: https://doi.org/10.3969/j.issn.1004910x.2020.02.002. [25] huining, yang xin. digital economy drives and high-quality development of china's manufacturing industry[j]. journal of shaanxi normal university (philosophy and social science edition), 2022,51(01):133147. doi: https://doi.org/10.15983/j.cnki.sxss.2022.0110. [26] qi yudong, ding shulei, liu cuihua. study on the impact of internet use on the wage income of flexibly employed people in the digital economy[j]. journal of social science, 2022(01):125-138+2. [27] he xingbang. quality of entrepreneurship and quality of economic growth in china an empirical analysis based on inter-provincial panel data[j]. statistics and information forum, 2019,34(12):84-93. doi: https://doi.org/10.3969/j.issn.1007-3116.2019.12.011. [28] meng tian, zhang henglong. digital finance and high-quality development of regional economy[j]. journal of social sciences, 2022(01):139-148. [29] wan yongkun, wang chenchen. an empirical test of the digital economy to enable high-quality development[j]. statistics and decision making, 2022,38(04):21-26. doi: https://doi.org/10.13546/j.cnki.tjyjc.2022.04.004. [30] wang hongming, sun pengbo, guo huifang. how can digital finance enable digital transformation of enterprises? --empirical evidence from chinese listed companies[j/ol]. finance and economics series:114 [2022-04-15]. doi: https://doi.org/10.13762/j.cnki.cjlc.20220311.001. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). atlantis press journal style research on evaluation of equity financing efficiency of listed companies in strategic emerging industries yaxi huang*, mu zhang school of finance, guizhou university of finance and economics, guiyang 550025, china abstract this paper chooses 198 listed companies in strategic emerging industries, using dea model to study the efficiency of equity financing, and carries on efficiency analysis, investment redundancy and output shortage analysis and industry comparative analysis. the results show that the efficiency of equity financing of listed companies in strategic emerging industries is inefficient. the comprehensive efficiency, pure technical efficiency and scale efficiency are 0.370, 0.603 and 0.563. from the scale pay, the economic scale of chuanrungufen should be increased, zhongguobaoan and other 179 decision-making units should be reduced; dongxulantian and other 169 decision-making units have different levels of input redundancy and lack of output; equity financing efficiency is unevenly developed between different industries. keywords: strategic emerging industries, financing efficiency, equity financing efficiency evaluation, data envelopment analysis (dea) 1. introduction strategic emerging industries refer to breaking new ground in major cutting-edge technologies, representing the new direction for the development of science and technology and industry, and embodying the trend of development of the knowledge-based economy, circular economy and low-carbon economy in the world today. september 8, 2010 premier wen jiabao chaired a meeting of the standing committee of the state council to consider and adopt the "accelerate the cultivation and development of strategic emerging industries decision" will be energy saving and environmental protection, a new generation of information technology, biomedicine, high-end equipment manufacturing, new energy sources , new materials and new energy vehicles and other seven industries designated as china's key strategic development of new industries. this injected a new * corresponding author: e-mail: 1308113622@qq.com. force into promoting the upgrading of china's industrial structure and effectively meeting social needs and supply. strategic emerging industries are guided by deng xiaoping theory and the important thought of “three represents”. they insist on giving full play to the fundamental role of the market and promoting the government-led role, insist on combining scientific and technological innovation with industrialization, and adhering to the overall promotion the four basic principles of combining the development across key areas with each other and insisting on enhancing the long-term competitiveness of the national economy and supporting the current development are committed to providing strong support for the sustainable economic and social development. the theory of pecking order in corporate financing argues that the best financing order should be endogenous financing first and then exogenous financing. exogenous financing should first be debt financing, and then equity financing [1]. however, the choice of financing methods in chinese companies is journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 189 received 11 september 2017 accepted 3 november 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). the opposite. the financing structure of china's listed companies is mainly based on external financing, and the external financing is up to 80% of the financing structure. the internal financing is generally less than 20% of the financing structure, and a few even depend entirely on external financing. in terms of foreign financing, listed companies in china generally prefer equity financing, and debt financing is not considered. therefore, it is impossible to optimize the company's financing structure by coordinating the proportion of equity financing and debt financing [2]. the sustainable development of strategic emerging industries cannot be separated from the choice of financing methods. and the capital-intensive and technology-intensive strategic emerging industries prefer the of equity financing. [3] equity financing efficiency can effectively measure the degree of perfection of china's capital market and the degree of resource allocation, which is of great significance for national economic growth and sustainable development of strategic emerging industries.[4] therefore, how to accurately and effectively evaluate the efficiency of equity financing of strategic new industry listed companies has become a major issue in the sustainable development of new strategic industries. 2. literature review as a way of exogenous financing, equity financing mainly refers to the behavior of listed companies raising funds by issuing shares, including two ways: public offering and private offering. in order to effectively study the efficiency of equity financing of listed companies in strategic new industries, this paper finds that many domestic and foreign scholars have studied equity financing and equity financing efficiency. the famous american financial expert modigliani and miller (1958) [5] in the "american economic review published entitled" the cost of capital, corporate finance and investment theory "of the thesis, the thesis put forward the famous mm theory, they think, in the condition of perfect capital market, because of arbitrage mechanism, the company issued shares regardless of financing or bond financing will not affect the value of the company, namely, capital structure and company value. sayuri shirai (2004) [6] based on the theory of financing constraints, constructs a regression model of the three ways of equity financing, bank lending and debt financing, which affect the company's investment decisions. the empirical results show that equity financing has not played a significant role, and thus lack of financing efficiency. charnes et al (1989) [7] first introduced the dea method into the evaluation of urban economic growth efficiency, and compared the economic performance level of 28 cities in china in 1983 and 1984. sueyoshi (1992) [8] expanded the application of dea in the area of urban efficiency evaluation, and investigated the resource allocation efficiency of 35 cities in china using dea/ar model. the domestic scholar zhengde xiong, fangjuan yang and jun wan (2014) [9] using two stage relational network dea model, with the cost of debt financing, debt financing risk as input indexes, rate of return on assets, total assets turnover, operating income growth rate as output indexes, the debt financing efficiency of china's new energy automotive industry listed companies and the corresponding sub stage efficiency was calculated. li jingwen, wang yuchun et al (2014) [10] took the 51 strategic emerging industries listed companies in beijing as samples, selected ten quarterly financial data since 2011, and took total assets, assetliability ratio and total operating costs as input indexes, return on net assets, total asset turnover, total revenue growth as output indexes, using dea method to measure and analyze financing efficiency. according to the total assets, assets and liabilities ratio as input indexes, return on net assets and tobin's q as the output indexes, xiaoyan qiao and dongjun mao (2015) [11] used dea method to compare and analyze the efficiency of equity financing for 2010-2013 years in jiangsu province 15 listed companies of the new energy, and based on the above results, the influence factors of efficiency are analyzed by the fixed effect model. qiong wang, chengxuan geng (2016) [12] extended the multi-stage dea model, taking non-flow accountable and capital public reserve as input indexes and net profit and total operating revenue as output indexes to build a six-stage super-sbm model malmquist index model, the static and dynamic evaluation of the financing efficiency of 29 listed companies in strategic emerging industries in jiangsu province for 2009-2014. ruibo liu and xuemei zhang (2009) [13] from the perspective of financing efficiency, journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 190 apply data envelopment analysis (dea) to the analysis and evaluation of the efficiency of equity financing of expressway listed companies. the empirical results show that the efficiency of equity financing of our expressway listed companies is relatively low. we should improve the efficiency of equity financing from reforming input scale of equity financing, equity concentration and other input indexes, main business income growth rate and net assets yield index. lichang liu and genfu feng et al (2004) [14] used the data envelopment analysis method to take the 47 listed companies that initially listed in shanghai stock market for 1998 as the research object, taking asset-liability ratio and ownership concentration as input indexes, net asset return rate, tobin's q value as the output indexes, a comprehensive evaluation of china's equity financing efficiency. haokun yan and honghong zhao (2014) [15] take the total assets, asset liability ratio and financial cost as input indexes, main business revenue growth rate and net assets yield ratio as output indexes. they use dea model to analyze the financing efficiency of 23 listed companies in inner mongolia. the results show that the overall financing efficiency of the listed companies in inner mongolia is low, and only 5 of the 23 companies are satisfied with the dea. on the basis of this, the basic idea of optimizing the financing efficiency of the listed companies is put forward. wen han, kaihong liu (2014) [16] take labor force, operating expenses, paid in capital as input indexes, premium income, payments for the output indexes by dea model showed that dea evaluation sample period the insurance business property insurance company operating performance; then, using two step cluster analysis, clustering analysis, clustering results were obtained clear. the empirical results demonstrate the feasibility and applicability of the dea clustering method in the performance evaluation of the insurance business of property insurance companies. yueping dai and hongmei zhang (2013) [17] choose the data from 2012 and 2011 of china high tech industry statistical yearbook, and use dea model to evaluate the efficiency of input-output of 31 provinces in china. the results show that the independent innovation efficiency of all the provinces in the last 5 years is generally low, and the national efficiency of guizhou province in the last 5 years has been greatly fluctuated and lack of stability. in summary, most foreign scholars focus on the research on the enterprise value and its influencing factors, and research on the financing efficiency of enterprises. domestic scholars have studied the financing efficiency of chinese enterprises, especially the equity financing efficiency from different angles. however, most of them involve strategic emerging industries and their listed companies or their sample size is too small, but they also do not carry out a detailed analysis of their inputs, outputs and industries. therefore, this article chooses 198 representative listed companies in china's strategic emerging industries as the research object, taking equity financing net value, ownership concentration, asset-liability ratio as input indexes, return on net assets, growth rate of main business, tobin's q value as output indexes, using data envelopment analysis (dea) study on the efficiency of equity financing, and analysis of the efficiency, input redundancy and output deficiency of industry analysis and comparative analysis of the financing efficiency of the three aspects of strategic emerging industries listed companies analysis. this is of great theoretical and practical significance to improve the equity financing efficiency of listed companies in strategic emerging industries and to improve the strategic emerging industry market. 3. introduction of dea model 3.1. fundamental data envelopment analysis (dea) was proposed by charnes, coopor and rhodes in 1978. the principle of this method is to maintain the input or input of dmu (decision making units) by means of mathematical programming and statistics identify relatively efficient frontiers of production, project each decision-making unit onto the dea production frontier and evaluate their relative validity by comparing the extent to which decision-making units deviate from the dea frontier. based on the concept of relative efficiency, dea method uses convex analysis and linear programming as a method of evaluation. the mathematical programming model is used to calculate and compare the relative efficiency between the decision-making units and evaluate the evaluation objects. it can give full consideration to the optimal input-output plan for the decision-making unit itself, so it can reflect the information and characteristics of the evaluation object journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 191 more ideally. meanwhile, it has its uniqueness for evaluating multi-input and multi-output of complex systems. it has the following characteristics: first, it is applicable to the comprehensive evaluation of the effectiveness of multi-output and multi-input. it has an absolute advantage in evaluating the effectiveness of multi-output and multi-inputs. second, the dea method does not integrate the data directly. therefore, the optimal efficiency index of decision-making unit has nothing to do with the dimension selection of input index and output index. there is no need to dimensionless the data before establishing the model using the dea method. third, no weight hypothesis is required. the decision-making unit input and output of the actual data to obtain the optimal weight, excluding a lot of subjective factors, with strong objectivity. fourth, the dea method assumes that each input is associated with one or more outputs. and there is indeed some connection between the input and output, but do not have to determine the display of this relationship. 3.2. construction of the model 3.2.1. integrated efficiency model (ccr model) there are n decision units dmu𝑗, where𝑗 = 1,2, … , 𝑛. any dmu has m input vectors (input production factors) and s output vectors (output obtained), then 𝑋𝑗 = （𝑥1𝑗，𝑥2𝑗，𝑥3𝑗 … 𝑥𝑚𝑗） > 0 , 𝑗 = 1,2, … , 𝑛 𝑌𝑗 = （𝑦1𝑗，𝑦2𝑗，𝑦3𝑗 … 𝑦𝑠𝑗） > 0, 𝑗 = 1,2, … , 𝑛 where𝑗 = 1,2, … , 𝑛, x𝑚𝑗denotes that the jth decision unit has m kinds of inputs, and 𝑌𝑠𝑗 denotes the sth input of the jth decision unit. so for the jthdmu𝑗 decision unit based on the minimum, inefficient, convex hypothesis production set: 𝑇 = {(𝑋, 𝑌)| �𝑋𝑗𝜆𝑗 𝑛 𝑗=1 ≤ x, �𝑋𝑗𝜆𝑗 𝑛 𝑗=1 ≥ 𝑌, 𝜆𝑗 ≥ 0, 𝑗 = 1,2, … , 𝑛 the input validity model of the dmu has the following ccrs: ⎩ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎧ 𝑚𝑚𝑛𝑚 �𝑋𝑗𝜆𝑗 + 𝑆− 𝑛 𝑗=1 = 𝑚𝑋0 �𝑌𝑗𝜆𝑗 − 𝑆+ 𝑛 𝑗=1 = 𝑌0 𝜆𝑗 ≥ 0, j = 1,2 … n 𝑆+ ≥ 0, 𝑆− ≥ 0 the ccr model is "comprehensively effective" in terms of the effective and technologically efficient dmu. assuming that the optimal solution is θ, λ, 𝑆+ , 𝑆− , the effective judgments and economic explanations are as follows: (1) ifθ = 1,𝑆+ = 0,𝑆− = 0, then the decision unit dmu𝑗 is said to be valid for the dea under the ccr model, indicating that the decision unit is comprehensive and effective, that is, both the scale efficiency and the technical efficiency best, there is no "excess" investment and "deficit" output; (2) ifθ = 1, 𝑆+ ≠ 0,𝑆− ≠ 0 , the decision unit dmu𝑗 is said to be weakly valid for the dea in the ccr model. although there is no need for isometric compression in terms of input, there are some aspects of "excess" input or "deficit" output; (3) ifθ < 1, then the decision unit dmu𝑗 is said to be valid for non-dea under the ccr model, indicating that the input can be fully compressed by θ [18]. 3.2.2. technical efficiency model (ccgss model) there are n decision unitsdmu𝑗, where 𝑗 = 1,2, … , 𝑛, and 𝑋𝑗 = （𝑥1𝑗，𝑥2𝑗，𝑥3𝑗 … 𝑥𝑚𝑗） > 0 , 𝑌𝑗 = （𝑦1𝑗，𝑦2𝑗，𝑦3𝑗 … 𝑦𝑠𝑗） > 0, the ccgss model is ⎩ ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎪ ⎧ 𝑚𝑚𝑛𝑚 𝑠. 𝑡. �𝑋𝑗𝜆𝑗 + 𝑆− 𝑛 𝑗=1 = 𝑚𝑋0 �𝑌𝑗𝜆𝐽 − 𝑆+ 𝑛 𝑗=1 = 𝑌0 �𝜆𝑗 𝑛 𝑗=1 = 1 𝜆𝑗 ≥ 0, 𝑗 = 1,2, … , 𝑛 𝑆+ ≥ 0, 𝑆− ≥ 0 the optimal value can be obtained by this model. when ∑ 𝜆𝑗𝑛𝑗=1 < 1 is the scale returns increasing, journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 192 ∑ 𝜆𝑗𝑛𝑗=1 = 1 is the scale returns constant; ∑ 𝜆𝑗 𝑛 𝑗=1 > 1is the decreasing returns to scale. suppose the optimal solution is θ, λ, 𝑆+ , 𝑆− , the effective judgments and economic explanations are as follows: (1)if θ < 1 or 𝑆+ ≠ 0 , 𝑆− ≠ 0 , dmu𝑗 is non-dea valid (ccgss); (2)ifθ = 1,𝑆+ = 0, 𝑆− = 0, thendmu𝑗 is dea valid (ccgss). when (2) is satisfied, it indicates that dmu𝑗 is purely technical, otherwise non-technical. when 𝑆− ≠ 0 , it indicates that there is excess investment; when𝑆+ ≠ 0, it indicates that there is a loss output. when 0<θ<1, it indicates that the dmu is improperly input and can be compressed in equal ratio, which is non-technical and effective. to sum up, the ccr model is used to evaluate whether the decision-making unit is both efficient and technically effective. however, the ccgss model is only used to evaluate whether the technical efficiency is the best. combining the two can make the combination of technical efficiency and economies of scale analysis [19]. 3.3. general steps 3.3.1. determine the evaluation objectives evaluation is the most basic function of dea model, which is the basis of our correct application of dea model. only by determining the purpose of evaluation, can we find the right direction, select the appropriate model and collect the appropriate data to substantiate the problems in production and life. this requires that we be able to accurately translate the information in economic activity into the information required by the dea model or to correspond one-on-one with the relevant concepts of dea. 3.3.2. select the decision unit dmu since dea evaluates the relative validity of dmus of the same type, the following two points need to be followed in the selection of dmus: first, dmus must be of the same type, dmus of the same nature or dmus with the same time interval; then, the number of dmu should be selected as the input and output data of the sum of 2 times is appropriate. 3.3.3. establish input and output index system the establishment of the input and output index system needs to pay attention to the following points: first of all, it is necessary to reflect the purpose of evaluation truthfully and comprehensively. secondly, attention should be paid to the relationship between input indicators and output indicators. at the same time, we should try to avoid the multiple linear relationship between input and output indicators; finally, to ensure the diversity and availability of input and output indicators. 3.3.4. select the dea model the choice of dea model to follow the following two requirements: first, pay attention to the actual production and life background; the second is to choose the dea model for evaluation purposes. in addition, different models can be applied for multi-angle analysis in order to arrive at a more comprehensive evaluation. 3.3.5. evaluation and analysis of dea results this is the most critical step in the application of dea model. by collecting data and calculating models, we get the result of dea model. based on this result, we analyze the real economic problems, and provide an accurate direction for policymakers to formulate effective policies and solve practical problems. 3.3.6. adjust the input and output index system when the result of dea evaluation and analysis is unsatisfactory, we should adjust the input and output index system appropriately and reconsider it without violating the purpose of evaluation. by using a variety of dea models to analyze different angles, the different results are compared and the important factors that affect the decision making unit are observed. 3.3.7. draw a comprehensive analysis and evaluate the conclusion by calculating the dea model, we can get the following information: the dea validity of each dmu, the relationship between the relative efficiency of dmu and the input and output indexes, the relative effective production frontier and the projection of dmu on the journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 193 effective production frontier. to make a comprehensive analysis and evaluation of these results, we can formulate a scientific and reasonable policy [20]. 4. empirical analysis 4.1. determination of evaluation index 4.1.1. input indexes (1) equity financing net value since this paper mainly investigates the efficiency of equity financing, after we choose the total amount of equity financing acquired by the company after ipo, deducting the weighted financing cost, we get the net equity financing of the company. in order to ensure the validity of the model, it is transformed into a dimensionless the amount. (2) ownership concentration the degree of ownership concentration affects the company's financing efficiency through its influence on the company's daily operation, so it is used as an input index to investigate the efficiency of financing. the index mainly uses the largest shareholders to share the total number of shares issued. (3) asset liability ratio the target is to reflect the impact of the company's financial structure on equity financing. the calculation method is to divide the total liabilities of the listed company by the total assets. this index mainly investigates the relationship between the tax credit effect and the financing cost of the creditor's rights. 4.1.2 output indexes (1)return on equity this index reflects the profitability of shareholder investment in production and operation, which is the ratio of the company's net profit to the average net asset over a period of time. this paper introduces this index in the empirical evidence, intended to use this index to measure whether the profitability of the company after the equity financing has been enhanced. (2) main business revenue growth rate this index reflects the company's growth ability. it can be calculated from the increase of income from main business over the previous year. (3) tobin q tobin's q reflects the allocation efficiency of equity financing, equal to the total market price divided by the replacement value. since the replacement value of the company is assessed through acquisition, the replacement value is replaced by the net asset value of the company, which is the ratio of total turnover to total turnover over the years since the equity financing. table 4-1 input and output indexes input indexes output indexes equity financing net value return on equity ownership concentration main business revenue growth rate asset liability ratio tobin q 4.2. sample selection and data sources this paper chooses 198 companies with certain representativeness and comparability as the research object after excluding the changes of the financial statements before and after the listing of the company, the negative output indicators and the financial indicators such as the secondary issuance. according to dea's experience, it is only meaningful to analyze the sample size at least twice to three times more than the total number of input variables and output variables. the total number of input and output variables in this paper is 6, which meets the requirements. the data in this paper are all from the database of tai`an (csmar) series research database. the sample table of the company in this paper is shown in table 1 of appendix. strategic new industries are divided into seven industries: energy saving, environmental protection, a new generation of information technology, biomedicine, high-end equipment manufacturing, new energy, new materials and new energy vehicles. among the research objects in this sample, there are 29 energy saving and environmental protection industries, a total of 25 new generation of information technology industry, 34 biomedical industry, 30 high-end equipment manufacturing industry, 27 new energy industry, 25 new material industry, 28 new energy vehicle industry. the proportion of sample companies selected by each industry is shown in fig.4-1. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 194 fig.4-1 distribution of selected samples by industry in this paper, the database of tai`an (csmar) series research database are selected as input indexes and output indexes of each sample company in 2016, and the original data of each sample company are shown in table 2 of appendix. 4.3. evaluation results of equity financing efficiency this article intends to use deap version 2.1 software to research the issue of equity financing efficiency of strategic new listed companies. the original data of input and output indexes in table 2 of appendix are substituted into the model software for calculation, and the shares of the strategic new industry listed companies financing evaluation results and the overall situation (see table 4-2), the evaluation results of equity financing of listed companies in strategic emerging industries are shown in table 3 of appendix. 4.4. analysis of the evaluation results of equity financing efficiency 4.4.1 efficiency analysis (1) comprehensive efficiency the comprehensive efficiency is a comprehensive measure and evaluation of various aspects of the ability of resource allocation and resource utilization of decision-making units. as seen from table 3 of appendix and table 4-2, from the comprehensive efficiency of dea measurement, in 2016, the comprehensive efficiency of 18 decision-making units such as shenzhennengyuan, yichengxinneng, shangqijituan zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxinnengyuan, hebeixuangong, haigetongxin and wandongyiliao and so on reached 1, indicating that the inputs and outputs of the above decision-making units are comprehensive and effective , that is both technically effective and scale effective. the proportion of comprehensive and effective decision-making units is 9.09%. however, the average comprehensive efficiency of the 198 publicly listed companies in strategic emerging industries is 0.370, indicating that the input and output of listed companies in strategic emerging industries are not comprehensively and effectively implemented. among them, zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianchi, yingtejituan, zhongyuanhuanbao and other 180 decision-making units still has some space for improvement and improvement. (2) pure technical efficiency pure technical efficiency is the production efficiency of decision-making unit due to factors such as management and technology. it can be seen from table 3 of appendix and table 4-2,from the purely technical efficiency of dea measurement, in 2016, the pure technical efficiency of 29 decision-making units such as shenzhenneneyuan, yamadun, yinengxincheng, yaxingkeche, hongduhankong, zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxing energy, ankaikeche, hebeixuanonghager and so on reached 1, indicating that at the current technical level, the abovementioned decision-making unit invested in the use of resources is efficient. the proportion of purely technical and effective decision making units is 14.65%. however, the average pure technical efficiency of the table 4-2 the overall efficiency of equity financing efficiency in strategic emerging industries comprehensive efficiency pure technical efficiency scale efficiency dea effective 18（9.09%） 29（14.65%） 18（9.09%） non dea effective 180（90.91%） 169（85.35%） 180（90.91%） mean 0.370 0.603 0.563 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 195 198 publicly listed companies in the strategic emerging industries is 0.603, which is not entirely technically effective. among them, the management and technical level of 169 decision-making units such as zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianchi, yingtejituan, zhongyuanhuanbao, wanxiangqianchao, chinaguangheji, huanruishiji, kaidishentai, xinxianghuaxian and zhongkesanhuan should be improved. (3) scale efficiency scale efficiency is the production efficiency that is affected by the size of the decision unit. it can be seen from table 3 of appendix and table 4-2, from the scale efficiency of dea measurement, in 2016, , the scale efficiency of 18 decision making units such as shenzhennengyuan, yinengxincheng, zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxingnengyuan, hebeiuangong, haigetongxin, daomingguangxue and wandongyiliao reached 1, indicating that these decision-making units are effective in scale. the proportion of effective decision-making units is 9.09%. however, the average of the scale efficiency of the 198 listed companies that represent the strategic emerging industries is 0.563, indicating that the overall performance of strategic emerging industries listed companies is not achieved. among them, the scale efficiency of 180 decisionmaking units such as zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianqi, yingtejituan, zhongyuanhuanbao, wanxiangqianchao, fenghuagaoke, huitianredian and zhongtaiqiche still has some room for improvement and improvement. (4) returns to scale from the returns to scale of view, chuanrungufen shows the increasing returns to scale in production within the boundaries, that should be appropriate to increase the size of its economy, the scale and the input and output matching; zhongguobaoan, dongxulantian, desaidianchi tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianchi, yingtejituan, zhongyuanhuanbao, wanxiangqianchao, fenghuagaoke, zhongguangheji, huanruishiji, kaidishentai, xinxianghuaxian, zhongkesanhuan, zhongtaiqiche, huagongkeji, jingxinyaoye, xinhaiyi, jinzhikeji, leibaogaoje, wohuayiyao, sanweitongxin and other 179 decision-making unit in the production boundary performance of scale returns diminishing, indicating that its economic size should be appropriately reduced to make the scale and investment matching; the remaining 18 decision-making units in the production boundary performance for the same scale returns, then the economy should remain the same size. 4.4.2 analysis of insufficient input redundant output deap version 2.1 software gives the dea evaluation value of equity financing efficiency of listed companies in strategic emerging industries, and also gives the values of slack variables of inputs and outputs of each decision unit, that is, inputting redundant values and outputs insufficient value, the results see table 4 of appendix. table 4 shows that zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianchi, yingtejituan,zhongyuanhuanbao, wanxiangqianchao, fenghuagaoke huitianredian, yingluohua, zhongguangheji, huanruishiji, kaidishentai, xinxianghuaxian, zhongkesanhuan, zhongtaiqiche, huagongkeji, jingxinyaoye and other 169 decisionmaking unit there are varying degrees of input redundancy and output deficiencies. in the case of shenwuhuanbao, there were 0.092 million yuan of net investment in equity financing. there was 1.447% investment redundancy in the ownership concentration and 4.079 in the asset-liability ratio; net assets yield was 91340.018 output deficiency, there are 29.955 output deficits in the main business yield, 0.342 output deficit in tobin q. only after eliminating the above input redundancy and insufficient output can shenwuhuanbao company reach purely technical and effective. the remaining 168 decision-making unit input redundant output analysis, and so on. 4.4.3 industry comparative analysis september 8, 2010 premier wen jiabao chaired a meeting of the standing committee of the state council to consider and adopt the "accelerate the cultivation and development of strategic emerging industries decision" will be energy saving and environmental protection, a new generation of journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 196 information technology, biomedicine, high-end equipment manufacturing, new energy sources , new materials and new energy vehicles and other seven industries designated as china's key strategic development of new industries. among them, the energy-saving and environmental protection industries include 29 companies such as dongxulantian, zhongyuanhuanbao, kaidishentai, huitianxincai, xianhehuanbao, shenwuhuanbao, zhongdianhuanbao, tianhaohianjing, zhongcaijieneng and so on; the new generation of information technology industry include 25 companies such as tefaxinxi, huanruishiji, xinhaiyi, sanweitongxin, beiweikeji, guangxunkeji, shensunda a and so on. the biopharmaceutical industry includes 34 companies such as zhongguobaoan, haiwangshengwu, fengyuanyaoye, jingxinoyey, wohuayiyao and so on. the high-end equipment manufacturing industries include30 companies such as xujidianqi, huagongkeji, hezhongsizhuang, siweituxin, teruide, zhongguoweixin and so on. the new energy industries include 27 companies such as shenzhennengyuan, huitianredian, yingluohua, jinzhikeji, tuorixinneng, yamadun and so on. the new materials industry include25 companies such as desaidianchi, fenghuagaoke, zhongguangheji, xinxianghuaxian, zhongkesanhuan, xinyegufen, zhonggangtianhuan and so on. the new energy automotive industries include 28 companies such as wanxiangqianchao, zhongtaiqiche, yinlungufen, yataigufen and so on. based on the data in table 4-3, we can calculate the dea average of the financing efficiency of the seven major industries such as energy saving and environmental protection, new generation of information technology, biomedicine, high-end equipment manufacturing, new energy, new materials and new energy vehicles. the results, as shown in figures 4-2, 4-3 and 4-4, show that the equity financing efficiency of listed companies in china's strategic emerging industries has an unbalanced development between industries. among them, the average value of comprehensive efficiency from high to low is the highend equipment manufacturing industry, new energy industry, energy saving and environmental protection industry, new energy automotive industry, a new generation of information technology industry, bio pharmaceutical industry, new materials industry; pure technical average value from high to low is the energy saving and environmental protection industry, high-end equipment manufacturing industry, the new energy automotive industry, a new generation of information technology industry, new energy industry, new material industry, bio pharmaceutical industry; the average scale efficiency from high to low is the energy saving and environmental protection industry, high-end equipment manufacturing industry, new energy industry, new energy automotive industry, new material industry, a new generation of information technology industry, bio pharmaceutical industry. fig.4-3 the average of pure technical efficiency of each industry fig.4-4 the average of scale efficiency of each industry fig.4-2 the average of the comprehensive efficiency of each industry 0,8728 0,6434 0,5393 0,7811 0,6273 0,5492 0,7324 0 0,2 0,4 0,6 0,8 1 0,786 0,5245 0,5158 0,6783 0,6273 0,5333 0,6143 0 0,2 0,4 0,6 0,8 1 0,4772 0,3866 0,3137 0,5634 0,4778 0,3127 0,4624 0 0,1 0,2 0,3 0,4 0,5 0,6 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 197 5. conclusions this paper selects a representative 198 listed companies from the strategic emerging industries as the research object, starting from the efficiency of equity financing based on capital input and output efficiency, with the previous definition of the financing efficiency as the theoretical basis established the evaluation index system of equity financing efficiency, and used deabcc model to evaluate the equity financing efficiency of listed companies in strategic emerging industries, thus draw the following conclusions: first, the average comprehensive efficiency of listed companies in strategic emerging industries is 0.370, indicating that the input and output of listed companies in strategic emerging industries are not comprehensively and effectively implemented. among them, shenzhennengyuan, yichengcinneng, zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxingnengyuan, hebeixuanong, haigetongxin, daomingguangxue, metainuo, nandudianyuan, wandongyiliao and other 18 decisionmaking unit to achieve comprehensive and effective; zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianqi, yingtejituan, zhongyuanhuanbao and other 180 decision-making units have not been comprehensively and effectively implemented. second, the average net technical efficiency of listed companies in strategic emerging industries is 0.603, which is not entirely purely technical and effective. among them, 29 decision-making units such as shenzhennengyuan, yamadun, yichengxinneng, yaxingkeche, hongduhankong, zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxingnengyuan, ankaikeche and haigetongxin reached the purely technical and effective level. 169 decision-making units such as zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, yingtejituan, zhongyuanhuanbao wanxiangqianchao, fenghuagaoke, huitianredian, yingluohua, huanruishiji, kaidishentai, xinxianghuaxian and zhongkesanhuan did not reach pure effective technology. thirdly, the average size efficiency of listed companies in strategic emerging industries is 0.563, indicating that the listed companies in strategic emerging industries are not achieving the overall scale effective. among them, shenzhennengyuan, yichengxinneng, zhenhuazhonggong, xugongjixie, liugong, haimaqiche, yinxingnengyuan, hebeixuanong, haigetongxin, daomingguangxue, metainuo, nandudianyuan, wandongyiliao and other 18 decision-making unit to achieve the scale of effective. 180 decision-making units such as zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianqi, yingtejituan, zongyuanhuanbao, wangxiangqianchao, fenghuagaoke, huitianredian, yingluohua, huanruishiji, kaidishentai, xinxianghuaxian, zhongkesanhuan and zhongtaiqiche have not achieved the scale effective. fourth, from the returns to scale of view, chuanrungufen`s economies of scale should be increased; zhongguobaoan, dongxulantian, desaidianchi, haiwangshengwu, fengyuanyaoye, xujidianchi, yingtejituan, zhongyuanhuabao, wanxiangqianchao, fenghuagaoke, huitianredian, , huanruishiji, kadishentai, xinxianghuaxian, zongtaiqiche, huagongkeji, jingxinyaoye, xinhaiyi and other 179 decision making units should be reduced; the remaining 18 decision making units economic scale should remain unchanged. fifth, zhongguobaoan, dongxulantian, desaidianchi, tefaxinxi, haiwangshenwu, fengyuanyaoye, xujidianqi, yingtejituan, zhongyuanhuanbao, wanxiangqianchao, fenghuagaoke, huitianredian, yingluohua, huanruishiji, kaidishentai, xinxianghuaxian, zhongkesanhuan, zhongtaiqiche, jingxinyaoye, xinhaiyi and other 169 decision-making units have varying degrees of input redundancy and output deficiencies. sixthly, the equity financing efficiency of the listed companies in the strategic emerging industries has the unbalanced development among industries. among them, the average value of comprehensive efficiency from high to low is the high-end equipment manufacturing industry, new energy industry, energy saving and environmental protection industry, new energy automotive industry, a new generation of information technology industry, bio pharmaceutical industry, new materials industry; pure technical average value from high to low is the energy saving and environmental protection industry, high-end equipment journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 198 manufacturing industry, the new energy automotive industry, a new generation of information technology industry, new energy industry, new material industry, bio pharmaceutical industry; the average scale efficiency from high to low is the energy saving and environmental protection industry, high-end equipment manufacturing industry, new energy industry, new energy automotive industry, new material industry, a new generation of information technology industry, bio pharmaceutical industry.[21] references [1] s c myers, n s majluf. corporate financing and investment decisions. when firms have information. that investors do not have [a]. the journal of financial economies, 1984. (13):187-221. [2] zhangsheng liu. research on different financing methods and their advantages, [a]. new accounting.2010, (4): 28-32. [3] haibo li. how finance can support the development of strategic emerging industries [n]. people's daily, 201103-31 (007). [4] chen miao, wang liguo. research on equity financing efficiency of strategic emerging industries based on dea model [j]. business, 2015, (16): 202-203. [5] franco modigliani and merton h miller. the cost of capital corporate finance and the theory of investment. the american economic review [j] .1958(6): 261-297. [6] sayuri shirai. testing the three roles of equity markets in developing countries: the case of china. world development [j]. 2004, (9): 1467-1486. [7] charnes a, cooper w, shanling li. using data envelopment analysis to evaluate efficiency in the economic performance of chinese cities [j].socioeconomic planning sciences,1989,23(6): 325-344. [8] toshiyuki sueyoshi. measuring the industrial performance of chinese cities by data envelopment analysis [j]. socio-economic planning sciences, 1992, 26(2): 75-88. [9] zhengde xiong, fangjuan yang, jun wan. debt financing efficiency of listed companies based on two-stage dea model a case study of a strategic emerging industry new energy vehicle [j]. finance and economics theory and practice, 2014, 35 (05): 51-56. [10] jingwen li, yuchun wang, zhengdong yang. study on financing efficiency of listed companies in strategic emerging industries: a case study of beijing [j] .journal of economics and management, 2014, (06): 7482. [11] xiaoyan qiao, dongjun mao. emerging industries equity financing and debt financing support efficiency analysis jiangsu province, the new energy listed companies as an example [j] .academic monthly, 2015, (24): 77-80. [12] qiong wang, xuancheng geng. jiangsu province strategic emerging industry financing efficiency of listed company -based on the panel data of 2009-2014 [j]. east china economic management, 2016, 30 (07): 14-20. [13] ruibo liu, xuemei zhang. analysis of equity financing efficiency of expressway listed companies based on dea [j] .business research, 2009, (12): 102-105. [14] lichang liu, genfu feng, zhang daohong, mao hongxia. evaluation of equity financing efficiency of listed companies based on dea [j]. systems engineering, 2004, (01): 55-59. [15] haokun yan. research on financing efficiency of listed companies in inner mongolia [a].chinese association for disaster prevention risk based on the method of dea analysis of specialized committee. information technology risk analysis and crisis response in the china association for disaster prevention risk analysis of the specialized committee sixth annual meeting of china association for disaster prevention [c]. risk analysis of specialized committee.2014:4. [16] han wen. research on insurance business performance evaluation of property insurance companies based on dea clustering [a] .china disaster defense association risk analysis committee .risk analysis and information technology in crisis response risk analysis of china disaster defense association professional committee of the sixth annual conference proceedings [c]. china disaster defense association risk analysis committee, 2014: 5. [17] yueping dai. dea-based evaluation of china's hightech industry independent innovation efficiency [a].risk analysis council of china association for disaster prevention, 2013: 6. [18] xiaoying peng, qinghua zhang. evaluation of equity financing efficiency of coal listed companies based on dea model [j] .china mines, 2008, 17 (12): 26-29. [19] lichang liu, genfu feng, daohong zhang, hongxia mao. based on dea listed companies equity financing efficiency evaluation [a].systems engineering, 2004, 22 (1): 55-59. [20] chen sheng, juan pang. overview of data envelopment analysis (dea) method [c].science and technology economics, 2016, (20): 5-10. [21] mu zhang, da-zhi huang, hai zhu. evaluation of urban economic efficiency in china considering environmental factors [j] .mathematics and practice, 2016, 46 (07): 71-80. journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 199 appendices table 1: 198 listed companies of strategic emerging industries enterprise name stock code enterprise name stock code zhongguobaoan 000009 shenzhennengyuan 000027 dongxuliantian 000040 desaidianchi 000049 tefaxinxi 000070 haiwangshenwu 000078 fenyuanyaoye 000153 xujidianqi 000400 yingtejituan 000411 zhongyuanhuanbao 000544 wanxiangqianchao 000559 fenhuagongke 000636 huitianredian 000692 yingluohua 000795 zhonguangheji 000881 huanruishiji 000892 kaidishentai 000939 xinxianghuanxian 000949 zhongkesanhuan 000970 zhongtaiqiche 000980 huagongkeji 000988 jingxinyaoye 002020 xinhaiyi 002089 jinzhikeji 002090 laibaogaoke 002106 wohuayiyao 002107 sanweitongxin 002115 yinlungufen 002126 tuobangufen 002139 beiweikeji 002148 laiyinshenwu 002166 tuorixinneng 002218 guanxunkeji 002281 yataigufen 002284 gelinmei 002340 hezhongsizhuang 002383 siweituxin 002405 duofuduo 002407 kanshenggufen 002418 shuanghuanchuangdong 002472 rongjiruanjian 002474 jiangfencicai 002600 yamadun 002623 teruide 300001 yiweilineng 300014 huitianxincai 300041 shuzizhengtong 300075 yichengxinneng 300080 dongfangrishen 300118 xianhehuanbao 300137 shenwuhuanbao 300156 zhengdongzhiyao 300158 zhongdianhuanbao 300172 chulingxixin 300250 tianhaohuanjing 300332 kanxinxincai 600076 shangqijituan 600104 dongkuigufen 600114 zhongguoweixing 600118 mingjiangshuidian 600131 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 200 futianqiche 600166 yaxingkeche 600213 hongduhangkong 600316 zhneghuazhonggong 600320 hangtiandongli 600343 kunyaojituan 600422 hengtongguangdian 600487 guihanggufen 600523 yijingguangdian 600537 guanguyuan 600771 shensangdaa 000032 xugongjixie 000425 liugong 000528 haimaqiche 000572 qidiguhan 000590 shantuigufen 000680 xinyegufen 000751 xinhuazhiyao 000756 zhonghanfeiji 000768 bohaihuosai 600960 qinchuangjichuang 000837 yinxingnengyuan 000862 ankaiqiche 000868 faershen 000890 yunneidongli 000903 shandahuate 000915 hebeixuangong 000923 zhongguozhongqi 000951 fousukeji 000973 jiuzhitang 000989 shirongzhaoye 002016 zhouyankeji 002046 hengdiandongci 002056 zhonggangtianyuan 002057 suzhougude 002079 longjigufen 601012 zhongcaijieneng 603126 woerhecai 002130 yunhaijinshu 002182 zhengtongdianzi 002197 feimaguoji 002210 aotexun 002227 aoweitongxin 002231 dahuagufen 002236 chuanrungufen 002272 zhongdianxinlong 002298 dongfangyuanlin 002310 gellinmei 002340 longjijixie 002363 dongshanjingmi 002384 neimengyiji 600967 shenglutongxin 002446 haigetongxin 002465 fuchunhuanbao 002479 keshida 002518 tianshunfengneng 002531 yataikeji 002540 yishengyaoye 002566 qinxinghuanjing 002573 shenyanggufen 002580 daomingguangxue 002632 maoshuodianyuan 002660 jingweigufen 002662 teyiyaoye 002728 ankeshenwu 300009 jiqiren 300024 meitainuo 300038 hekanxinneng 300048 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 201 yujingangshi 300064 nandudianyuan 300068 danshenkeji 300073 shengyunhuanbao 300090 jiayugufen 300117 zhejiangdingli 603338 dafukeji 300134 weiminghuanbao 603568 yongqinghuanbao 300187 xinwangda 300207 qianshanyaoji 300216 dongfangredian 300217 meichengkeji 300237 dianzhenduan 300244 chanshanyaoye 300255 jinduankeji 300258 baanshuiwu 300262 hejiagufen 300273 yangguangdianyuan 300274 sannuoshenwu 300298 zhongjizhuangbei 300308 bohuichuangxin 300318 jinmokeji 300334 mengcaoshentai 300355 xuelanghuanjing 300385 zhonglaigufen 300393 feilihua 300395 huannengkeji 300425 sitongxincai 300428 shanheyaopu 300452 maikeshanwu 300463 zhongfeigufen 300489 meishangshentai 300495 gaolangufen 300499 wandongyiliao 600055 huarunshuanghe 600062 yutongkeche 600066 jinhuagufen 600080 yongdinggufen 600105 beifangxitu 600111 juhuagufen 600160 jiangsuwuzhong 600200 guangshengyouse 600259 haizhengyaoye 600267 guodiannanzi 600268 hengruiyiyao 600276 taihuagufen 600281 shiyinggufen 603688 hangfakeji 600391 hanlianhuanjing 600323 changjiangtongxin 600345 lianchuangguangdian 600363 ningboyunshen 600366 shangongjintai 600385 wukuangziben 600390 sanyouhuagong 600409 jianghuiqiche 600418 peilingdianli 600452 baotaigufen 600456 guiyanboye 600459 laobaixing 603883 fenhuotongxin 600498 zhongtiankeji 600522 changyuanjituan 600525 feidahuanbao 600526 xiamengwuye 600549 tiandikeji 600582 nanjingxiongmao 600775 shangchaigufen 600841 hangfadongli 600893 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 202 table 2: input,output indexes and related raw datas company abbreviation input indexes output indexes equity financing net value (ten thousand yuan) ownership concentration （%） asset liability ratio return on equity main business revenue growth rate tobin q zhongguobaoan 18000 11.9113 0.623791 0.051567 0.317839 1.029797 shenzhennengyuan 36830 47.8246 0.592068 0.056656 0.016903 0.447504 dongxulantian 12140 30.9812 0.365806 0.015992 1.265375 0.950595 desaidianchi 6436 45.2277 0.708479 0.233108 0.034411 1.677195 tefaxinxi 53460.62 39.1841 0.594942 0.111641 0.882775 1.565169 haiwangshenwu 9667.6 45.9346 0.647212 0.084867 0.223803 1.001336 fenyuanyaoye 45325 11.4827 0.537108 0.039653 0.308429 1.615621 xujidianqi 44700 41.276 0.472021 0.124559 0.307734 1.279725 yingtejituan 4646.4 21.5432 0.757153 0.113082 0.115792 0.656154 zhongyuanhuanbao 15075 56.618 0.152379 0.056745 0.807481 1.82053 wanxiangqianchao 11100 51.5291 0.589862 0.188514 0.053155 2.641354 fenhuagaoke 11002.5 20.0286 0.316424 0.031979 0.430391 1.30635 huitianredian 12719.46 35.1048 0.705711 0.042107 0.166339 0.700617 yingluohua 28325 39.3785 0.20797 0.01569 0.465252 2.841968 zhongguangheji 23625 27.6 0.53008 0.060528 0.439343 0.760075 huanruishiji 24186.25 5.9062 0.178561 0.098234 53.968966 4.174999 kaidishentai 28683 28.4657 0.68962 0.025709 0.430543 0.605557 xinxianghuaxian 56752 30.1696 0.372869 0.032147 0.199627 1.370864 zhongkesanhuan 25215 23.1744 0.133647 0.078054 0.011009 2.544025 zhongtaiqiche 30681.79 19.9883 0.473969 0.04028 0.041599 1.852857 huagongkeji 40680 32.3575 0.417369 0.073051 0.264986 2.496118 jingxinyaoye 16517 23.1578 0.262634 0.089321 0.324762 2.198813 xinhaiyi 13490.32 18.0506 0.62116 0.028052 0.091058 1.814412 jinzhikeji 22973 36.916 0.644611 0.105423 0.509782 1.472001 laibaogaoke 93265.7 20.8423 0.199965 0.059913 0.383746 1.707126 wohuayiyao 17907 50.2671 0.18847 0.107317 0.200293 8.440842 sanweitongxin 16648.17 19.0872 0.621308 0.025444 0.143054 1.679945 yinlungufen 23043.88 11.156 0.466098 0.105559 0.145714 1.457458 tuobangufen 17327.23 19.6085 0.32204 0.083375 0.263597 2.204443 beiweikeji 20715.509 21.2504 0.090645 0.071153 1.037182 3.988281 laiyinshenwu 15016.1 17.5723 0.652109 0.082234 0.110749 2.224834 tuorixinneng 41068.4 32.5529 0.437158 0.047604 0.568097 1.156847 guanxunkeji 61214.91 45.4344 0.395116 0.094016 0.292752 3.416419 yataigufen 42211.337 38.8217 0.4493 0.055966 0.117497 2.110095 gelinmei 70353.976 12.54 0.622371 0.041603 0.531296 0.999681 hezhongsizhuang 104927.53 39.4551 0.270369 0.027372 0.545808 2.467085 siweituxin 136777.3 12.2128 0.230614 0.036656 0.052553 5.006329 duofuduo 99084.837 13.9279 0.452753 0.165225 0.3093 3.259756 kanshengufen 66022.47 15.5804 0.698482 0.099706 0.287662 1.591019 shuanghuanchuangdong 77277.998 8.8614 0.228187 0.061629 0.247183 1.965198 rongjiruanjian 90566.32 20.6293 0.338254 0.017677 0.126668 3.951408 jangfencicai 59961.75 18.4646 0.577755 0.046733 1.475008 0.945552 yamadun 146868 45 0.493956 0.007921 0.27809 1.549758 teruide 77928.3 43.9969 0.748315 0.065754 1.034825 1.473806 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 203 tianlongguangqi 37392.79 42.1827 0.522712 0.150541 0.734491 2.84242 huitianxincai 57117.407 23.2827 0.134061 0.061031 0.160359 2.741069 shuzizhengtong 70255.3 30.4462 0.385291 0.098264 0.485614 3.133993 yichengxinneng 148008 20.0219 0.476753 0.008974 0.405479 0.791892 dongfengrishen 183757.98 32.5652 0.60289 0.183408 0.334125 1.116291 xianhehuanbao 62650.328 13.8599 0.162212 0.078113 0.256952 2.816268 shenwuhuanbao 110894.32 42.6771 0.487197 0.27446 1.572772 5.019875 zhendongzhiyao 130505.44 43.7861 0.233178 0.036636 0.451311 1.480393 zhongdianhuanbao 53860.62 28.9767 0.336518 0.102676 0.06689 2.9211 chulingxinxi 22050.3 42.7341 0.098137 0.082983 0.229803 4.049243 tianhaohuanjing 60990.82 21.2635 0.529938 0.013597 0.770685 0.958072 kanxinxincai 26700 21.3127 0.192468 0.13107 0.268388 2.678145 shangqijituan 208200 74.295 0.601955 0.186995 0.128313 0.437754 dongkuigufen 43199.69 19.4549 0.148709 0.074687 0.072161 2.585874 zhongguoweixin 12370 51.017 0.4703 0.080102 0.163146 3.402302 mingjiangshuidian 17569 23.9211 0.565189 0.153201 0.214414 2.286302 futianqiche 31000 27.0653 0.646346 0.02686 0.368691 0.382292 yaxingkeche 36999.8 51 0.956045 0.319506 0.684865 0.737739 hongduhangkong 87827 43.7703 0.522279 0.002468 0.307081 1.32524 zhenhuanzhonggong 83800 28.8281 0.728936 0.018641 0.046222 0.324244 hangtiandongli 22857.772 28.7781 0.39091 0.013658 0.199757 3.080362 kunyaojituan 39343 29.7873 0.33139 0.113377 0.037617 1.972355 hengtongguangdian 38016.6 19.3351 0.656001 0.224366 0.423001 1.173695 guihanggufen 32755.42 37.0069 0.384845 0.082063 0.050131 1.752411 yijinguangdian 28587.636 30.3608 0.572383 0.118439 0.050501 1.232354 guangyuyuan 6750 23.2198 0.193013 0.089001 1.187008 5.51039 shensangdaa 8400 27.8079 0.18994 0.021402 0.093641 3.226157 xugongjixie 11040 42.5567 0.43458 0.016345 0.010659 0.538602 liugong 20000 34.9758 0.45215 0.012268 0.090789 0.399605 haimaqiche 16000 28.7966 0.1066 0.001935 0.114369 0.484404 qidiguhan 5800 18.6099 0.55682 0.029638 0.022667 6.958307 shantuigufen 37300.6 12.8127 0.53704 0.033263 0.290749 0.740825 xinyegufen 65280 23.591 0.47083 0.061276 0.142051 1.935407 xinhuazhiyao 3275 34.4595 0.56293 0.041142 0.146104 1.200706 zhonghangfeiji 35700 38.1795 0.53861 0.023972 0.093937 1.500922 qinchuangjichuang 19783 4.999 0.48565 0.00598 0.26126 0.751772 yingxingnengyuan 28140 14.6453 0.62288 0.008091 0.078713 0.459694 ankaiqiche 32820 21.1339 0.82949 0.041277 0.22801 0.509507 faershen 36948.6 21.0656 0.75692 0.144655 0.12084 0.519541 yunneidongli 37500 30.9655 0.44611 0.051304 0.381352 0.88804 shandahuate 13499 20.7161 0.12023 0.168022 0.181025 3.059428 hebeixuangong 19567.33 35.5402 0.7105 0.005127 0.02375 3.36523 zhongguozhongqi 32396.54 63.7759 0.73288 0.096596 0.07623 0.444266 fuosukeji 58045 5.8586 0.49031 0.071475 0.036139 1.489305 jiuzhitang 34530 5.8439 0.08758 0.034818 0.114623 3.933827 shirongzhaoye 17737.83 53.5709 0.20849 0.108033 0.432934 1.077128 zhouyankeji 14783.99 39.5579 0.33056 0.034476 0.083552 1.775413 hengdiancidong 61040.62 50.146 0.31203 0.111907 0.128831 1.961213 zhonggangtianyuan 12829.22 25.9393 0.16786 0.019791 0.11988 4.923638 suzhougude 22537.68 34.284 0.12605 0.087356 0.211722 3.80147 longjigufen 151310.5 14.9446 0.38142 0.061057 0.95973 1.670124 zhongcaijieneng 25120 2.2383 0.39708 0.081635 0.088168 2.091867 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 204 wowehecai 20823.57 2.806 0.46708 0.018855 0.119996 1.851254 yunhaijinshu 49632.15 6.4141 0.57318 0.011782 0.310188 2.036891 zhengtongdianzi 23343.26 5.779 0.43747 0.017756 0.700349 1.784756 feimaguoji 25387.15 50.0595 0.84064 0.072063 0.073705 0.911288 aotexun 37539.58 57.5726 0.27216 0.024992 0.068753 5.328808 aoweitongxin 21531.88 18.4978 0.22478 0.011085 0.23733 6.246577 dahuagufen 38955.3 10.3138 0.18514 0.233542 0.302168 2.581213 chuanrungufen 22181.79 3.8122 0.02606 0.003113 0.025215 2.452324 zhongdianxinlong 24013.97 5.2163 0.1709 0.008062 0.074614 1.542406 dongfangyuanlin 80103.85 8.2513 0.60675 0.09416 0.275175 2.016332 gelinmei 70353.98 10.3159 0.53233 0.003037 0.471734 1.109946 longjijixie 50896 45.5516 0.30383 0.03472 0.066615 1.908261 dongshanjimi 95882.38 8.7742 0.76089 0.004494 0.361346 1.097664 neimengyiji 34527 23.6185 0.4673 0.064681 24.638863 2.927052 shenlutongxin 43002.5 4.5006 0.17994 0.030051 0.099744 4.009343 haigetongxin 314314.4 16.9693 0.26216 0.082313 0.041283 2.764868 fuchunhuanbao 133771.2 34.7541 0.20614 0.1019 0.192276 2.220042 keshida 88362.82 60.1006 0.29619 0.132879 0.107002 3.596084 tianshunfenneng 122682.9 29.8115 0.22284 0.058555 0.108446 1.573325 yataikeji 154722.5 10.623 0.06154 0.07064 0.134139 2.717388 yishenyaoye 103922.1 9.771 0.40291 0.028825 0.077862 1.559103 qinxinhuanjiang 159048.6 45.0273 0.58945 0.217156 0.45448 1.887238 shenyanggufen 43837.11 2.8958 0.41865 0.045569 0.119271 2.056385 daomingguangxue 56736.49 42.182 0.17677 0.004946 0.036836 4.526761 maoshuodianyuan 41418.55 7.5847 0.42547 0.015636 0.020701 1.932026 jingweigufen 143213.8 30 0.43974 0.093117 0.199524 1.331809 teyiyaoye 32180.69 2.9 0.44267 0.075389 0.18795 3.176194 ankeshenwu 32109.5 6.7488 0.25408 0.106903 0.156789 7.523596 jiqiren 57590 25.27 0.14804 0.072172 0.204218 5.255439 meitainuo 55130 5.867 0.44249 0.013416 0.001622 2.410953 hekanxinneng 96456.42 21.7151 0.19407 0.039542 0.052915 1.881521 yujingongshi 74502.18 20.4566 0.21699 0.018039 0.464152 1.643184 nandudianyuan 196564.9 4.3561 0.27284 0.022368 0.032202 1.862394 danshengkeji 65537.19 27.0571 0.38001 0.025615 0.41984 4.759231 shwnyunhuanbao 51262.02 6.7223 0.47692 0.008596 0.524326 1.247397 jiayugufen 67769.9 39.6643 0.6294 0.064815 0.0408 1.067788 zhenjiangdingli 43987.43 1.575 0.18195 0.160318 0.415431 5.58102 dafukeji 186988.5 43.3888 0.21451 0.028971 0.082131 2.528019 weiminghuanbao 45147.32 1.6236 0.06583 0.247014 0.09413 5.270439 yongqinghuanbao 61352.57 58.2045 0.39453 0.065618 0.291912 3.375329 xinwangda 82334.43 6.2204 0.71533 0.22284 0.363935 2.80671 qianshanyaoji 46468.2 3.4452 0.65659 0.196326 0.868597 3.403831 dongfangdianre 55924.64 3.9023 0.17367 0.032649 0.003564 2.56007 meichengkeji 33231.09 3.7468 0.24912 0.048871 0.367727 2.276939 dianzhenduan 27027.73 8.3066 0.55446 0.039967 0.299982 3.347799 chanshangyaoye 69726.11 12.5569 0.25662 0.074874 0.28552 2.486892 jingduankeji 58793.76 48.3557 0.23397 0.092779 0.156354 2.923013 baanshuiwu 26742.12 10.4186 0.44756 0.062255 0.408375 2.176984 hejiagufen 61004.08 5.0013 0.2824 0.044699 0.119007 3.763279 yangguangdianyuan 127124.4 7.9717 0.47948 0.098043 0.485882 1.553448 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 205 sannuoshenwu 58083.6 27.6498 0.1304 0.094702 0.234857 4.712694 zhongjizhuangbei 29766.16 46.0071 0.12081 0.01809 0.084099 4.693377 bohuichuangxin 35854.15 19.8865 0.19687 0.016975 0.192963 3.820974 jinmokeji 44544.75 23.1869 0.38736 0.034106 0.210806 2.407568 mengcaoshentai 37361.41 5.9687 0.50505 0.09888 0.600747 1.855393 xuelanghuanjing 25810.54 1.0918 0.48881 0.113478 0.298417 2.804849 zhonglaigufen 35000 7.6141 0.51244 0.221942 0.893895 2.596604 feilihua 26983.01 7.0122 0.21285 0.115276 0.119594 5.003353 huannengkeji 24403 1.2456 0.09319 0.031894 0.308065 4.260348 sitongxincai 26842.62 0.2054 0.13374 0.111696 0.207206 8.550455 shanheyaopu 14868.6 1.8982 0.21451 0.118519 0.085339 9.237657 maikeshenwu 99732 5.9588 0.08749 0.14305 0.137017 5.640249 zhongfeigufen 16203.37 6.7488 0.268 0.063789 0.138595 5.551527 meishangshentai 49591.66 5.1021 0.33097 0.053404 0.193698 4.627519 gaolaigufen 22761.43 1.9472 0.49794 0.120104 0.305244 5.135788 wandongyiliao 12330 46.6818 0.15173 0.058092 0.010364 3.419443 huanrunshuanghe 30503 38.8981 0.11882 0.098426 0.013767 1.78957 yutongkeche 33075 27.114 0.60509 0.303712 0.155955 1.751181 jinhuangufen 17880 15.7225 0.22583 0.026587 0.276149 2.53229 yongdinggufen 23996.03 25.7401 0.21915 0.052393 0.041948 2.727364 beifangxitu 34240 30 0.25873 0.084874 0.05082 2.841912 juhuagufen 44341.68 48.7474 0.09062 0.006217 0.049251 1.927764 jangsuwuzhong 25047.5 17.0103 0.19168 0.022737 0.147961 2.171001 guangshengyouse 29710.11 37.2164 0.37972 0.025531 2.337913 3.147939 haizhengyaoye 45979.83 33.2235 0.44983 0.000534 0.035507 0.700826 guodiannanzi 30771 50.2603 0.74765 0.086113 0.112207 0.488345 hengruiyiyao 46660 24.3076 0.08688 0.22595 0.209272 7.586455 taihuagufen 55755 43.4783 0.69624 0.078889 0.17965 1.62128 shiyinggufen 33084.48 2.828 0.04865 0.07237 0.081654 3.683754 hangfakeji 24523 36.0184 0.61126 0.031833 0.088985 2.336412 hanlanhuanjing 41825 17.9806 0.44463 0.035094 0.143528 0.907461 changjiangtongxin 35460 28.6274 0.07628 0.019998 0.369324 2.519341 lianchuangguangdain 38722 21.7288 0.22522 0.08026 0.168225 2.196961 ningboshenrun 34845.57 31.3721 0.1723 0.137232 0.370272 2.053032 shandongjintai 3977.72 17.3819 8.38 0.025617 0.051392 17.658202 wukuangziben 58120 27.8393 0.10632 0.003036 0.278969 2.069426 sanyouhuangong 57694.5 39.9839 0.4762 0.037353 0.441365 0.803604 jianghuaiqiche 84119.15 7.0963 0.64123 0.073875 0.144266 0.511175 peilingdianli 24348.16 51.6438 0.76181 0.188405 0.332232 2.104195 baotaigufen 31088.4 56.0419 0.47841 0.018563 0.08157 1.150653 guiyanboye 25939.8 39.3433 0.41914 0.038357 0.148653 1.971349 laobaixing 101002.8 2.0548 0.50295 0.034679 0.273318 2.917035 fenghuotongxin 179734.2 46.5749 0.70317 0.054282 0.370349 1.380452 zhongtiankeji 35992 16.5882 0.33742 0.078633 0.268934 1.429016 changyuanjituan 18085.6 3.9149 0.59015 0.038324 0.211132 1.265562 feidahuanbao 27297.22 17.6519 0.64511 0.010392 0.154221 0.918951 xiamengwuye 33354.83 26.0363 0.48819 0.030018 0.563182 1.751385 tiandikeji 30222 36.3103 0.39303 0.022878 0.175307 0.54382 nanjingxiongmao 11199.3 26.4147 0.09628 0.02091 1.0991 2.665993 shangchaigufen 14335 48.0509 0.37862 0.027448 0.175091 1.616131 hangfadongli 29744 29.7823 0.32432 0.035797 0.007234 1.325377 bohaihuosai 30397 18.191 0.26409 0.001953 0.222484 1.449027 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 206 table 3: evaluation results of equity financing efficiency of listed companies in strategic emerging industries company abbreviation comprehensive efficiency pure technical efficiency scale efficiency returns to scale zhongguobaoan 0.291 0.589 0.494 drs shenzhennengyuan 1.000 1.000 1.000 dongxulantian 0.464 0.663 0.700 drs desaidianchi 0.472 0.876 0.539 drs tefaxinxi 0.258 0.611 0.422 drs haiwangshenwu 0.447 0.744 0.600 drs fenyuanyaoye 0.234 0.441 0.532 drs xujidianqi 0.305 0.619 0.493 drs yingtejituan 0.513 0.858 0.598 drs zhongyuanhuanbao 0.355 0.938 0.378 drs wanxiangqianchao 0.267 0.895 0.299 drs fenhuagaoke 0.210 0.398 0.528 drs huitianredian 0.543 0.850 0.639 drs yingluohua 0.240 0.728 0.329 drs zhongguangheji 0.370 0.591 0.625 drs huanruishiji 0.035 0.125 0.284 drs kaidishentai 0.619 0.850 0.728 drs xinxianghuaxian 0.327 0.588 0.555 drs zhongkesanhuan 0.508 0.624 0.813 drs zhongtaiqiche 0.362 0.465 0.777 drs huagongkeji 0.172 0.533 0.323 drs jingxinyaoye 0.121 0.368 0.329 drs xinhaiyi 0.355 0.489 0.725 drs jinzhikeji 0.259 0.587 0.441 drs laibaogaoke 0.263 0.470 0.561 drs wohuayiyao 0.125 0.761 0.165 drs sanweitongxin 0.343 0.521 0.659 drs yinlungufen 0.175 0.377 0.464 drs tuobangufen 0.107 0.322 0.333 drs beiweikeji 0.076 0.341 0.222 drs laiyinshenwu 0.253 0.420 0.603 drs tuorixinneng 0.317 0.607 0.522 drs guanxunkeji 0.184 0.708 0.260 drs yataigufen 0.257 0.656 0.392 drs gelinmei 0.346 0.658 0.525 drs hezhongsizhuang 0.342 0.790 0.433 drs siweituxin 0.481 0.636 0.756 drs duofuduo 0.148 0.402 0.369 drs kanshengufen 0.195 0.566 0.345 drs shuanghuanchuangdong 0.219 0.355 0.616 drs rongjiruanjian 0.329 0.590 0.557 drs jangfencicai 0.296 0.593 0.499 drs yamadun 0.901 1.000 0.901 drs teruide 0.338 0.814 0.415 drs tianlongguangqi 0.159 0.611 0.261 drs huitianxincai 0.176 0.417 0.422 drs shuzizhengtong 0.137 0.496 0.276 drs yichengxinneng 1.000 1.000 1.000 dongfengrishen 0.513 0.841 0.610 drs xianhehuanbao 0.136 0.300 0.454 drs shenwuhuanbao 0.096 0.588 0.163 drs zhendongzhiyao 0.519 0.871 0.596 drs zhongdianhuanbao 0.230 0.498 0.461 drs chulingxinxi 0.152 0.672 0.226 drs tianhaohuanjing 0.551 0.672 0.820 drs kanxinxincai 0.092 0.313 0.293 drs journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 207 shangqijituan 1.000 1.000 1.000 dongkuigufen 0.188 0.355 0.530 drs zhongguoweixin 0.206 0.816 0.253 drs mingjiangshuidian 0.142 0.378 0.376 drs futianqiche 0.765 0.860 0.890 drs yaxingkeche 0.637 1.000 0.637 drs hongduhangkong 0.847 1.000 0.847 drs zhenhuanzhonggong 1.000 1.000 1.000 hangtiandongli 0.181 0.562 0.321 drs kunyaojituan 0.305 0.566 0.539 drs hengtongguangdian 0.249 0.586 0.424 drs guihanggufen 0.303 0.657 0.461 drs yijinguangdian 0.369 0.568 0.651 drs guangyuyuan 0.059 0.365 0.162 drs shensangdaa 0.215 0.490 0.439 drs xugongjixie 1.000 1.000 1.000 liugong 1.000 1.000 1.000 haimaqiche 1.000 1.000 1.000 qidiguhan 0.293 0.433 0.676 drs shantuigufen 0.369 0.584 0.632 drs xinyegufen 0.224 0.483 0.462 drs xinhuazhiyao 0.359 0.680 0.528 drs zhonghangfeiji 0.409 0.736 0.556 drs qinchuangjichuang 0.683 0.726 0.941 drs yingxingnengyuan 1.000 1.000 1.000 ankaiqiche 0.724 1.000 0.724 drs faershen 0.648 0.920 0.704 drs yunneidongli 0.372 0.607 0.613 drs shandahuate 0.077 0.286 0.269 drs hebeixuangong 1.000 1.000 1.000 zhongguozhongqi 1.000 1.000 1.000 fuosukeji 0.497 0.542 0.916 drs jiuzhitang 0.093 0.172 0.540 drs shirongzhaoye 0.459 0.814 0.563 drs zhouyankeji 0.327 0.684 0.478 drs hengdiancidong 0.292 0.755 0.386 drs zhonggangtianyuan 0.171 0.460 0.372 drs suzhougude 0.127 0.536 0.237 drs longjigufen 0.401 0.703 0.571 drs zhongcaijieneng 0.177 0.268 0.660 drs wowehecai 0.279 0.334 0.834 drs yunhaijinshu 0.340 0.468 0.728 drs zhengtongdianzi 0.238 0.324 0.733 drs feimaguoji 0.582 0.922 0.632 drs aotexun 0.495 1.000 0.495 drs aoweitongxin 0.077 0.372 0.208 drs dahuagufen 0.064 0.178 0.360 drs chuanrungufen 0.427 1.000 0.427 irs zhongdianxinlong 0.191 0.255 0.747 drs dongfangyuanlin 0.176 0.498 0.353 drs gelinmei 0.742 0.849 0.875 drs longjijixie 0.436 0.846 0.515 drs dongshanjimi 0.998 1.000 0.998 drs neimengyiji 0.128 0.412 0.311 drs shenlutongxin 0.134 0.223 0.601 drs haigetongxin 1.000 1.000 1.000 fuchunhuanbao 0.309 0.642 0.481 drs keshida 0.310 0.876 0.354 drs tianshunfenneng 0.458 0.689 0.665 drs yataikeji 0.390 0.536 0.728 drs yishenyaoye 0.516 0.646 0.798 drs journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 208 qinxinhuanjiang 0.288 0.724 0.398 drs shenyanggufen 0.193 0.340 0.567 drs daomingguangxue 1.000 1.000 1.000 maoshuodianyuan 0.532 0.573 0.929 irs jingweigufen 0.405 0.707 0.572 drs teyiyaoye 0.118 0.245 0.482 drs ankeshenwu 0.066 0.156 0.423 drs jiqiren 0.127 0.432 0.293 drs meitainuo 1.000 1.000 1.000 hekanxinneng 0.490 0.591 0.830 drs yujingongshi 0.319 0.465 0.687 drs nandudianyuan 1.000 1.000 1.000 danshengkeji 0.154 0.547 0.281 drs shwnyunhuanbao 0.471 0.535 0.880 drs jiayugufen 0.647 0.841 0.769 drs zhenjiangdingli 0.048 0.157 0.307 drs dafukeji 0.764 1.000 0.764 drs weiminghuanbao 0.074 0.147 0.505 drs yongqinghuanbao 0.257 0.946 0.272 drs xinwangda 0.122 0.443 0.276 drs qianshanyaoji 0.086 0.330 0.260 drs dongfangdianre 0.796 0.825 0.965 drs meichengkeji 0.096 0.210 0.459 drs dianzhenduan 0.164 0.281 0.583 drs chanshangyaoye 0.161 0.319 0.503 drs jingduankeji 0.227 0.750 0.303 drs baanshuiwu 0.128 0.294 0.437 drs hejiagufen 0.169 0.278 0.607 drs yangguangdianyuan 0.313 0.631 0.495 drs sannuoshenwu 0.122 0.439 0.279 drs zhongjizhuangbei 0.396 0.830 0.477 drs bohuichuangxin 0.120 0.390 0.307 drs jinmokeji 0.168 0.448 0.374 drs mengcaoshentai 0.126 0.369 0.343 drs xuelanghuanjing 0.101 0.273 0.369 drs zhonglaigufen 0.088 0.306 0.287 drs feilihua 0.074 0.139 0.533 drs huannengkeji 0.059 0.121 0.491 drs sitongxincai 0.036 0.099 0.364 drs shanheyaopu 0.050 0.074 0.678 drs maikeshenwu 0.147 0.320 0.458 drs zhongfeigufen 0.066 0.136 0.487 drs meishangshentai 0.115 0.222 0.521 drs gaolaigufen 0.082 0.180 0.457 drs wandongyiliao 1.000 1.000 1.000 huanrunshuanghe 0.717 0.882 0.813 drs yutongkeche 0.195 0.465 0.419 drs jinhuangufen 0.106 0.293 0.363 drs yongdinggufen 0.221 0.490 0.452 drs beifangxitu 0.211 0.534 0.394 drs juhuagufen 0.837 1.000 0.837 drs jangsuwuzhong 0.140 0.323 0.432 drs guangshengyouse 0.198 0.666 0.297 drs haizhengyaoye 1.000 1.000 1.000 guodiannanzi 0.825 0.967 0.853 drs hengruiyiyao 0.073 0.327 0.224 drs taihuagufen 0.308 0.724 0.425 drs shiyinggufen 0.080 0.118 0.677 drs hangfakeji 0.333 0.652 0.511 drs journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 209 hanlanhuanjing 0.280 0.491 0.570 drs changjiangtongxin 0.199 0.532 0.375 drs lianchuangguangdain 0.133 0.355 0.373 drs ningboshenrun 0.159 0.454 0.351 drs shandongjintai 1.000 1.000 1.000 wukuangziben 0.381 0.643 0.593 drs sanyouhuangong 0.554 0.819 0.677 drs jianghuaiqiche 0.583 0.819 0.712 drs peilingdianli 0.240 0.727 0.330 drs baotaigufen 0.744 1.000 0.744 drs guiyanboye 0.290 0.683 0.424 drs laobaixing 0.306 0.514 0.594 drs fenghuotongxin 0.574 1.000 0.574 drs zhongtiankeji 0.129 0.325 0.395 drs changyuanjituan 0.285 0.509 0.560 drs feidahuanbao 0.645 0.779 0.828 drs xiamengwuye 0.252 0.529 0.476 drs tiandikeji 0.740 0.866 0.855 drs nanjingxiongmao 0.156 0.467 0.334 drs shangchaigufen 0.436 0.841 0.519 drs hangfadongli 1.000 1.000 1.000 bohaihuosai 0.290 0.461 0.629 drs mean 0.370 0.603 0.563 among them, the "irs" indicates that the scale of compensation increases, "-" means that the scale compensation is constant, and "drs" represents diminishing returns of scale. table 4: the input redundancy and output insufficiency of each decision unit company abbreviation input redundancy output insufficiency equity financing net value (ten thousand yuan) ownership concentration （%） asset liability ratio return on equity main business revenue growth rate tobin q zhongguobaoan 0.000 -0.077 0.000 14121.371 10.265 0.436 shenzhennengyuan 0.000 0.000 0.000 0.000 0.000 0.000 dongxulantian 0.000 -1.160 0.000 41564.597 15.716 0.186 desaidianchi -0.166 0.000 -0.839 28397.577 6.402 0.100 tefaxinxi 0.000 -0.790 -0.306 34030.814 24.943 0.379 haiwangshenwu 0.000 0.147 -0.082 21926.638 15.803 0.223 fenyuanyaoye 0.000 -0.212 0.000 57466.912 23.2 0.681 xujidianqi 0.000 -0.215 -0.678 43096.172 25.399 0.290 yingtejituan -0.086 0.000 0.000 58409.159 4.171 0.125 zhongyuanhuanbao 0.000 -0.718 0.000 56101.867 3.727 0.312 wanxiangqianchao -0.107 0.000 -2.065 21917.104 6.070 0.069 fenhuagaoke 0.000 -0.358 0.000 31116.628 30.285 0.478 huitianredian 0.000 -0.108 0.000 50156.266 6.215 0.125 yingluohua 0.000 -0.381 -1.523 10608.972 14.749 0.196 zhongguangheji 0.000 -0.374 0.000 31779.138 19.071 0.366 huanruishiji 0.000 -53.880 -0.446 169317.717 41.347 1.250 kaidishentai 0.000 -0.379 0.000 46273.698 5.013 0.121 xinxianghuaxian 0.000 -0.009 0.000 39750.568 21.132 0.261 zhongkesanhuan -0.028 0.000 0.000 15186.619 0.000 0.099 zhongtaiqiche 0.000 0.000 0.000 35230.028 22.951 0.544 huagongkeji 0.000 -0.173 -1.131 35587.270 28.307 0.365 jingxinyaoye 0.000 -0.244 -1.553 28407.906 39.830 0.452 xinhaiyi 0.000 -0.035 0.000 48759.343 18.834 0.648 jinzhikeji 0.000 -0.428 -0.166 32083.533 25.976 0.454 laibaogaoke 0.000 -0.206 0.000 105313.405 23.535 0.234 wohuayiyao 0.000 -0.101 -5.598 106357.017 15.803 0.251 sanweitongxin 0.000 -0.087 0.000 46424.308 17.569 0.572 yinlungufen -0.049 0.000 0.000 43322.163 18.449 0.771 tuobangufen 0.000 -0.180 -0.766 36438.898 41.236 0.677 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 210 beiweikeji 0.000 -0.951 -0.053 65451.909 41.087 0.175 laiyinshenwu -0.010 0.000 0.000 38233.825 24.307 0.902 tuorixinneng 0.000 -0.436 0.000 26562.311 21.055 0.283 guanxunkeji 0.000 -0.201 -2.009 25203.517 18.706 0.163 yataigufen 0.000 -0.028 -0.794 22175.391 20.395 0.236 gelinmei 0.000 -0.435 0.000 36575.484 23.104 0.324 hezhongsizhuang 0.000 -0.319 -0.339 27886.559 10.486 0.194 siweituxin 0.000 0.000 -2.558 78324.422 6.994 0.132 duofuduo 0.000 -0.049 -0.235 147662.734 24.308 0.675 kanshengufen -0.029 -0.216 0.000 50556.718 26.488 0.535 shuanghuanchuangdong 0.000 0.000 0.000 140576.580 27.02 0.415 rongjiruanjian 0.000 0.000 -1.639 62878.460 14.323 0.235 jangfencicai -0.001 -1.415 0.000 41179.750 17.318 0.397 yamadun 0.000 0.000 0.000 0.000 0.000 0.000 teruide 0.000 -0.841 0.000 17773.739 10.035 0.171 tianlongguangqi 0.000 -0.627 -1.944 102830.115 26.856 0.333 huitianxincai 0.000 0.000 -0.329 79836.760 32.544 0.296 shuzizhengtong 0.000 -0.339 -1.752 71314.742 30.905 0.391 yichengxinneng 0.000 0.000 0.000 0.000 0.000 0.000 dongfengrishen -0.014 -0.219 0.000 34793.913 32.657 0.114 xianhehuanbao 0.000 -0.171 -0.376 146073.198 32.315 0.378 shenwuhuanbao -0.092 -1.447 -4.079 91340.018 29.955 0.342 zhendongzhiyao 0.000 -0.209 0.000 19294.955 6.474 0.27 zhongdianhuanbao 0.000 0.000 -1.745 54193.470 29.156 0.339 chulingxinxi 0.000 -0.140 -0.471 76579.311 20.824 0.292 tianhaohuanjing 0.000 -0.619 0.000 29829.158 10.399 0.259 kanxinxincai 0.000 -0.168 -1.645 90251.771 46.733 0.422 shangqijituan 0.000 0.000 0.000 0.000 0.000 0.000 dongkuigufen 0.000 0.000 -0.654 78442.475 35.326 0.279 zhongguoweixin 0.000 0.086 -1.398 34068.421 11.520 0.106 mingjiangshuidian -0.025 -0.118 0.000 87325.61 39.403 0.931 futianqiche 0.000 -0.315 0.000 48534.629 4.403 0.105 yaxingkeche 0.000 0.000 0.000 0.000 0.000 0.000 hongduhangkong 0.000 0.000 0.000 0.000 0.000 0.000 zhenhuanzhonggong 0.000 0.000 0.000 0.000 0.000 0.000 hangtiandongli 0.000 -0.048 -1.451 25464.296 22.394 0.304 kunyaojituan -0.032 0.000 -1.519 30160.349 22.835 0.261 hengtongguangdian -0.105 -0.162 0.000 26821.650 16.456 0.463 guihanggufen 0.000 0.000 -1.182 17122.758 19.345 0.253 yijinguangdian -0.040 0.000 0.000 21774.204 23.125 0.436 guangyuyuan 0.000 -1.102 -3.212 65717.872 40.481 0.336 shensangdaa 0.000 -0.018 -0.232 25535.176 28.910 0.197 xugongjixie 0.000 0.000 0.000 0.000 0.000 0.000 liugong 0.000 0.000 0.000 0.000 0.000 0.000 haimaqiche 0.000 0.000 0.000 0.000 0.000 0.000 qidiguhan 0.000 0.000 -4.174 14566.975 24.385 0.730 shantuigufen 0.000 -0.168 0.000 26610.877 13.689 0.383 xinyegufen 0.000 0.000 0.000 69752.735 25.207 0.503 xinhuazhiyao 0.000 -0.075 0.000 40120.632 16.200 0.265 zhonghangfeiji 0.000 0.000 0.000 12798.976 13.688 0.193 qinchuangjichuang 0.000 -0.064 0.000 38789.499 8.516 0.184 yingxingnengyuan 0.000 0.000 0.000 0.000 0.000 0.000 ankaiqiche 0.000 0.000 0.000 0.000 0.000 0.000 faershen -0.117 0.000 0.000 25486.441 4.527 0.066 yunneidongli 0.000 -0.280 0.000 24274.507 20.045 0.289 shandahuate 0.000 -0.060 -2.049 174714.085 51.620 0.444 hebeixuangong 0.000 0.000 0.000 0.000 0.000 0.000 zhongguozhongqi 0.000 0.000 0.000 0.000 0.000 0.000 fuosukeji -0.029 0.000 0.000 48994.168 26.473 0.414 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 211 jiuzhitang 0.000 0.000 -1.248 166432.628 28.167 0.422 shirongzhaoye 0.000 -0.328 0.000 107408.687 12.227 0.321 zhouyankeji 0.000 -0.004 0.000 20462.589 18.245 0.152 hengdiancidong 0.000 -0.026 0.000 68113.34 16.251 0.174 zhonggangtianyuan 0.000 -0.041 -2.975 19491.417 30.395 0.271 suzhougude 0.000 -0.120 -0.356 80698.624 29.726 0.273 longjigufen 0.000 -0.688 0.000 64077.563 20.59 0.162 zhongcaijieneng -0.038 -0.029 0.000 68721.151 32.05 1.086 wowehecai 0.000 -0.072 0.000 53865.844 24.482 0.930 yunhaijinshu 0.000 -0.042 0.000 56475.933 12.988 0.652 zhengtongdianzi 0.000 -0.642 0.000 48599.723 20.119 0.911 feimaguoji 0.000 -0.004 0.000 17575.694 4.236 0.071 aotexun 0.000 0.000 0.000 0.000 0.000 0.000 aoweitongxin 0.000 -0.050 0.000 36361.104 31.237 0.380 dahuagufen -0.080 -0.198 -0.494 179403.522 47.499 0.853 chuanrungufen 0.000 0.000 0.000 0.000 0.000 0.000 zhongdianxinlong 0.000 0.000 0.000 70011.371 23.241 0.498 dongfangyuanlin 0.000 -0.072 0.000 80761.497 43.21 0.612 gelinmei 0.000 -0.196 0.000 12517.584 15.02 0.095 longjijixie 0.000 0.000 0.000 9272.367 8.299 0.055 dongshanjimi 0.000 0.000 0.000 0.000 0.000 0.000 neimengyiji 0.000 -24.536 -0.650 49365.468 33.769 0.668 shenlutongxin 0.000 0.000 -1.074 149888.536 15.687 0.627 haigetongxin 0.000 0.000 0.000 0.000 0.000 0.000 fuchunhuanbao 0.000 -0.094 -0.566 74531.812 19.364 0.179 keshida 0.000 0.000 -1.729 56778.582 8.519 0.219 tianshunfenneng 0.000 0.000 0.000 55440.784 13.472 0.216 yataikeji 0.000 -0.064 -0.140 134068.908 9.205 0.229 yishenyaoye 0.000 0.000 0.000 56894.906 8.25 0.221 qinxinhuanjiang -0.054 -0.344 -0.437 60688.857 17.181 0.225 shenyanggufen 0.000 0.000 0.000 85247.268 31.156 0.814 daomingguangxue 0.000 0.000 0.000 0.000 0.000 0.000 maoshuodianyuan 0.000 0.000 0.000 30861.118 17.04 0.317 jingweigufen 0.000 0.000 0.000 59379.072 16.477 0.182 teyiyaoye 0.000 -0.033 0.000 99377.684 41.433 1.367 ankeshenwu 0.000 -0.074 -3.422 174345.975 36.644 1.380 jiqiren 0.000 -0.040 -2.907 75787.729 33.255 0.313 meitainuo 0.000 0.000 0.000 0.000 0.000 0.000 hekanxinneng 0.000 0.000 0.000 66829.264 15.045 0.134 yujingongshi 0.000 -0.214 0.000 85871.670 23.578 0.251 nandudianyuan 0.000 0.000 0.000 0.000 0.000 0.000 danshengkeji 0.000 -0.208 -3.000 54205.159 22.379 0.314 shwnyunhuanbao 0.000 -0.238 0.000 44542.284 7.848 0.414 jiayugufen 0.000 0.000 0.000 12784.866 7.483 0.119 zhenjiangdingli -1.172 -0.373 -0.084 236065.005 15.439 0.976 dafukeji 0.000 0.000 0.000 0.000 0.000 0.000 weiminghuanbao -0.166 -0.053 -2.165 262063.597 15.355 0.382 yongqinghuanbao 0.000 -0.210 0.000 7383.245 3.324 0.023 xinwangda -0.062 -0.249 0.000 103385.651 58.161 0.898 qianshanyaoji -0.081 -0.775 0.000 94193.722 49.249 1.331 dongfangdianre -0.016 0.000 -0.132 11896.317 2.508 0.26 meichengkeji 0.000 -0.054 0.000 124891.553 39.099 0.936 dianzhenduan 0.000 -0.196 0.000 69150.546 25.324 1.419 chanshangyaoye 0.000 -0.094 0.000 148625.157 26.766 0.547 jingduankeji 0.000 -0.061 0.000 50181.755 16.143 0.194 baanshuiwu -0.002 -0.315 0.000 64279.519 25.043 1.076 hejiagufen 0.000 -0.037 -0.209 158250.221 12.974 0.733 yangguangdianyuan 0.000 -0.235 0.000 74286.702 43.072 0.280 sannuoshenwu 0.000 -0.104 -2.207 74177.356 35.311 0.327 zhongjizhuangbei 0.000 0.000 -3.098 6087.037 9.408 0.316 bohuichuangxin 0.000 -0.014 -2.457 55965.323 31.041 0.307 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 212 jinmokeji 0.000 -0.042 -0.410 54844.500 28.548 0.477 mengcaoshentai -0.046 -0.538 0.000 63979.381 30.813 0.865 xuelanghuanjing -0.064 -0.237 0.000 68828.163 34.214 1.303 zhonglaigufen -0.147 -0.820 0.000 79324.005 34.657 1.616 feilihua 0.000 -0.026 -1.388 167210.279 43.454 1.319 huannengkeji 0.000 -0.255 -1.322 177441.023 9.057 0.678 sitongxincai -0.037 -0.165 -3.776 245600.101 6.8191 1.224 shanheyaopu -0.039 -0.028 -2.090 185319.158 23.659 2.674 maikeshenwu -0.058 -0.039 -2.939 211521.844 12.638 0.186 zhongfeigufen 0.000 0.000 -1.265 102966.756 42.886 1.703 meishangshentai 0.000 -0.095 -0.041 174085.933 17.910 1.162 gaolaigufen -0.013 -0.215 0.000 103455.664 47.244 2.263 wandongyiliao 0.000 0.000 0.000 0.000 0.000 0.000 huanrunshuanghe -0.052 0.000 0.000 4082.097 5.206 0.232 yutongkeche -0.173 0.000 0.000 38082.141 31.219 0.697 jinhuangufen 0.000 -0.158 -0.759 43213.663 37.999 0.546 yongdinggufen 0.000 0.000 -0.564 24992.859 26.809 0.257 beifangxitu 0.000 0.000 -2.125 29832.154 26.138 0.347 juhuagufen 0.000 0.000 0.000 0.000 0.000 0.000 jangsuwuzhong 0.000 0.000 -0.694 52472.874 35.635 0.402 guangshengyouse 0.000 -2.244 -1.830 14884.346 18.645 0.190 haizhengyaoye 0.000 0.000 0.000 0.000 0.000 0.000 guodiannanzi 0.000 -0.007 0.000 17710.411 1.739 0.026 hengruiyiyao -0.039 -0.081 -7.149 161540 49.987 0.515 taihuagufen 0.000 -0.082 0.000 21253.228 16.573 0.265 shiyinggufen 0.000 -0.030 -0.745 246135.436 21.039 0.362 hangfakeji 0.000 -0.007 -0.354 13096.106 19.235 0.326 hanlanhuanjing 0.000 -0.052 0.000 43345.526 18.634 0.461 changjiangtongxin 0.000 -0.238 0.000 31172.164 25.166 0.345 lianchuangguangdain 0.000 -0.049 -1.045 70237.035 39.413 0.409 ningboshenrun 0.000 -0.260 -0.113 120932.102 37.786 0.329 shandongjintai 0.000 0.000 0.000 0.000 0.000 0.000 wukuangziben 0.000 0.000 -0.746 32315.936 15.479 0.397 sanyouhuangong 0.000 -0.359 0.000 12707.922 8.807 0.105 jianghuaiqiche -0.028 -0.085 0.000 18642.877 28.835 0.142 peilingdianli -0.011 -0.208 -0.679 172145.014 19.389 0.286 baotaigufen 0.000 0.000 0.000 0.000 0.000 0.000 guiyanboye 0.000 -0.066 -0.783 12054.693 18.284 0.195 laobaixing 0.000 -0.087 0.000 95447.183 13.253 0.475 fenghuotongxin 0.000 0.000 0.000 0.000 0.000 0.000 zhongtiankeji 0.000 -0.110 0.000 74593.280 34.379 0.699 changyuanjituan 0.000 -0.009 0.000 18489.78 17.813 0.570 feidahuanbao 0.000 -0.095 0.000 20890.466 5.011 0.183 xiamengwuye 0.000 -0.446 0.000 29729.828 23.207 0.435 tiandikeji 0.000 -0.147 0.000 4684.167 5.628 0.114 nanjingxiongmao 0.000 -1.022 0.000 22244.929 30.184 0.307 shangchaigufen 0.000 -0.093 0.000 26059.281 9.054 0.081 hangfadongli 0.000 0.000 0.000 0.000 0.000 0.000 bohaihuosai 0.000 -0.023 -0.202 40463.024 21.251 0.309 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 189–213 ___________________________________________________________________________________________________________ 213 1. introduction0f( 2. literature review 3. introduction of dea model 3.1. fundamental 3.2. construction of the model 3.2.1. integrated efficiency model (ccr model) 3.2.2. technical efficiency model (ccgss model) 3.3. general steps 3.3.1. determine the evaluation objectives 3.3.2. select the decision unit dmu 3.3.3. establish input and output index system 3.3.4. select the dea model 3.3.5. evaluation and analysis of dea results 3.3.6. adjust the input and output index system 3.3.7. draw a comprehensive analysis and evaluate the conclusion 4. empirical analysis 4.1. determination of evaluation index 4.1.1. input indexes 4.1.2 output indexes 4.2. sample selection and data sources 4.3. evaluation results of equity financing efficiency 4.4. analysis of the evaluation results of equity financing efficiency 4.4.1 efficiency analysis 4.4.2 analysis of insufficient input redundant output 4.4.3 industry comparative analysis 5. conclusions references appendices << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <feff004e006100750064006f006b0069007400650020016100690075006f007300200070006100720061006d006500740072007500730020006e006f0072011700640061006d00690020006b0075007200740069002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b00750072006900650020006c0061006200690061007500730069006100690020007000720069007400610069006b007900740069002000610075006b01610074006f00730020006b006f006b007900620117007300200070006100720065006e006700740069006e00690061006d00200073007000610075007300640069006e0069006d00750069002e0020002000530075006b0075007200740069002000500044004600200064006f006b0075006d0065006e007400610069002000670061006c006900200062016b007400690020006100740069006400610072006f006d00690020004100630072006f006200610074002000690072002000410064006f00620065002000520065006100640065007200200035002e0030002000610072002000760117006c00650073006e0117006d00690073002000760065007200730069006a006f006d00690073002e> /lvi <feff0049007a006d0061006e0074006f006a00690065007400200161006f00730020006900650073007400610074012b006a0075006d00750073002c0020006c0061006900200076006500690064006f00740075002000410064006f00620065002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006100730020006900720020012b00700061016100690020007000690065006d01130072006f00740069002000610075006700730074006100730020006b00760061006c0069007401010074006500730020007000690072006d007300690065007300700069006501610061006e006100730020006400720075006b00610069002e00200049007a0076006500690064006f006a006900650074002000500044004600200064006f006b0075006d0065006e007400750073002c0020006b006f002000760061007200200061007400760113007200740020006100720020004100630072006f00620061007400200075006e002000410064006f00620065002000520065006100640065007200200035002e0030002c0020006b0101002000610072012b00200074006f0020006a00610075006e0101006b0101006d002000760065007200730069006a0101006d002e> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <feff004b00e40079007400e40020006e00e40069007400e4002000610073006500740075006b007300690061002c0020006b0075006e0020006c0075006f00740020006c00e400680069006e006e00e4002000760061006100740069007600610061006e0020007000610069006e006100740075006b00730065006e002000760061006c006d0069007300740065006c00750074007900f6006800f6006e00200073006f00700069007600690061002000410064006f0062006500200050004400460020002d0064006f006b0075006d0065006e007400740065006a0061002e0020004c0075006f0064007500740020005000440046002d0064006f006b0075006d0065006e00740069007400200076006f0069006400610061006e0020006100760061007400610020004100630072006f0062006100740069006c006c00610020006a0061002000410064006f00620065002000520065006100640065007200200035002e0030003a006c006c00610020006a006100200075007500640065006d006d0069006c006c0061002e> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word explanation of risk and uncertainty in news coverage of an anthrax attack explanation of risk and uncertainty in news coverage of an anthrax attack kristen alley swain meek school of journalism and new media, university of mississippi 135 farley hall, university, ms 38677-1848, usa e-mail: kaswain@olemiss.edu abstract a content analysis of u.s. news coverage of the 2001 anthrax attacks examined explanations of risk and uncertainty. the sample consisted of 833 stories drawn from 272 newspapers, associated press, national public radio, and four television networks (cbs, nbc, cnn, abc). dominant uncertainty factors included outrage rhetoric, speculation, attribution of unnamed sources, and coverage of confusing incidents. overall coverage also promoted comparability, through definitions and explanations about risks and transmission vectors. risk comparisons, specific advice, and process explanations were sparse. keywords: news media, framing, bioterrorism, risk communication, uncertainty, journalism 1. introduction in october 2001, the u.s. news media faced an unprecedented crisis when letters contaminated with deadly anthrax spores began to surface across the nation. spores were spread through the postal system, alarming a public already anxious shortly after the 9/11 attacks in new york city. the first anthrax victim, an employee of a florida supermarket tabloid publisher, died of an illness that doctors could not at first identify. tommy thompson, the secretary of health and human services, suggested the victim might have contracted anthrax by drinking water from a stream, but after several postal workers and journalists tested positive for anthrax exposure, he eventually declared that someone was intentionally trying to kill people. initially, the official response was confused and spread across many different agencies. only four letters containing anthrax, postmarked in trenton, nj, had entered the postal system. these letters resulted in 22 confirmed cases of anthrax infection, including five deaths.1 this small-scale dispersion, which constituted a negligible threat to citizens as compared with other prevalent public health risks, generated confusion and panic and illustrated the challenges of communicating information about risk to an alarmed public. the anthrax outbreaks generated immense media attention2 and dominated the nightly news for two weeks in october. during this period, anthrax knocked the 9/11 attacks and afghanistan bombings out of the top news slot.3 most reporters learned as they went and found themselves in the midst of a story where journalists were both messengers and potential victims.4 maxine isaacs, former press secretary to walter mondale, characterized the anthrax coverage as a “hyperhysterical, meaningless, endless recycling of the same facts over and over again.”5 cbs news anchor dan rather said, "i worry about [excessive media coverage] creating exactly what the people who spread this terrible stuff want, which is spreading fear that they hope will result in panic.”6 the news media serve as the primary source of public information for disaster warnings and predictions7 because of their broad reach and potential to influence knowledge, attitudes, and behaviors.8 however, inaccurate, incomplete, and sensational coverage can contribute to public misunderstanding about the risks involved.9 officials withheld information from journalists because they feared widespread panic, but the lack of information itself alarmed the public10 because the resulting coverage often was conflicting, shallow, and lacked validation by health authorities.11 experts later concluded that a greater public understanding of the anthrax threat would have helped journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 81-95 published by atlantis press copyright: the authors 81 administrateur texte tapé à la machine received 13 february 2012 administrateur texte tapé à la machine accepted 20 june 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine k.a. swain reduce fear and panic.12 shortly after the attacks, most americans believed that government officials were not telling americans everything they needed to know.13 the present study explores news coverage promoting uncertainty in coverage of the anthrax attacks, as well as the use of explanatory coverage to help audiences understand risks and put them in context. uncertainty coverage included outrage rhetoric, speculation, vague advice, conflicting reports, hoaxes, false alarms, and use of off-record interview sources. explanatory content included risk comparisons, relative risk explanations, process explanations, and definitions. the iso 31000 (2010), developed by an international committee of risk experts from more than 30 nations, defines risk as the probability that an activity or inaction will lead to an undesirable outcome. ultimately, risk is the positive or negative impact of uncertainty on objectives. uncertainties include events that may or not happen, as well as events caused by ambiguity. uncertainty involves a lack of information that leads to inadequate understanding of an event, outcome, or likelihood. uncertainty describes a situation where an outcome is unknown, while risk describes the chance of a hazardous incident occurring.14 a threat is a possible danger that could exploit vulnerability, and vulnerability is a weakness that exposes a system to harm.15 mccollum (2006) defines a hazard as a situation or theoretical risk of harm that poses a threat to life, health, property, or the environment. once a hazard becomes active, it can create a crisis incident. determining the likelihood of a hazard occurring, potential seriousness or severity of the incident to various populations if it did occur, and the community’s capacity to mitigate the incident, helps risk managers determine whether and how a hazard should be addressed.16 news audiences may misinterpret risk messages when they have difficulty understanding a lack of scientific certainty.17 media coverage may promote uncertainty when it fails to present scientific knowledge about how a hazard causes adverse health effects, fails to provide precise risk assessment,18 constructs a disagreement among experts or data sources, or uses imprecise language.19 uncertainty also may increase when audiences expect one outcome but something different occurs.20 discussing uncertainties in news coverage of an incident may reinforce anxiety while reducing public confidence.21 the amount of news coverage about a particular hazard can increase risk perception because audiences believe a hazard occurs more frequently when they can easily recall or imagine such instances. when news coverage of a risk increases, this increases the perceived likelihood that it will occur.22 in turn, officials can manipulate news coverage more easily when an event involves greater uncertainty.23 in a crisis, official estimates of risk often are value laden, politically and economically influenced, or based on invalid assumptions.24 1.1. outrage the present study will identify outrage rhetoric in news coverage of the 2001 anthrax attacks. in a threatening situation, uncertainty triggers outrage, the principal determinant of perceived hazard. outrage occurs when individuals are upset about a threat, believe they are in danger,25 and perceive greater risk than actually exists.26 the magnitude of outrage may depend on audience members’ personal knowledge, training, and previous experience with the situation. people tend to focus more on feelings of outrage than the hazard itself27 and may be more outraged by trivial risks that are imposed.28 in anthrax attacks, public outrage is expressed as fear, panic, or anxiety and is provoked by the perception that a hazard is involuntary, unfamiliar, artificial, controlled by others, has no visible benefits, or has delayed effects.29 indeterminate risks breed greater fear,30 and news coverage of a threat can amplify the perception of indeterminate risk. the newer a risk, the more unfamiliar and dreaded it is.31 a risk is dreaded if consequences are potentially catastrophic, uncontrollable, potentially fatal, not equitable in their distribution, pose a high risk to future generations, are not easily reduced, or are involuntarily imposed. a risk is unknown if it is not observable, not evident to those exposed, or if its effects are delayed and not definitively known to science. risks that are both dreaded and unknown are more likely to produce broad social, political, and policy consequences and provoke higherorder concerns such as moral trepidation or perceived threats to future generations.32 published by atlantis press copyright: the authors 82 explanation of risk and uncertainty in news coverage news coverage can promote outrage if it fails to connect specific events to larger issues,33 amplifies or ignores risks, or emphasizes drama over scientific facts.34 media emphasis on a hazard can shape perceived danger, exaggerate social and economic responses, and lead to consequences far more serious than the initial threat.35 misrepresenting the prevalence or causes of death and their risk factors can contribute to distorted perceptions of a health hazard.36 while stating that a risk is insignificant can create public suspicion that officials are suppressing important information that could put citizens at risk,37 exaggerated stories spread quickly when officials overestimate death rates for infected patients.38 when news coverage prompts audiences to worry about getting sick, the more they will consider a hazard to be important.39 during the anthrax attacks, news coverage asserted that cipro was the antidote to anthrax infection, but most americans lacked access to these high-level antibiotics. many who did manage to obtain cipro experienced side effects from misusing it. hospitals became inundated with "worried well," frightened citizens, and these hysterical public reactions hindered the health system’s ability to treat those in need of medical care.40 to reduce outrage, risk messages must reassure, be clear, increase individual knowledge and compliance, provide adequate information, neither undernor overemphasize risk, increase trust,41 and simplify complex information.42 news coverage must thoroughly and precisely present this content from trusted sources, in order to reduce outrage. audiences also must understand the seriousness of a risk and how their practical responses could mitigate possible consequences.43 when individuals perceive a risk as high, they may reject advice presented through public channels, unless the message bolsters enough selfefficacy to adopt the recommended protective behavior.44 top-down, one-way communication, as presented by official sources speaking through daily news coverage, tries to bring public belief in line with expert views.45 when officials speak about a hazard, the main goal is to convey “have faith; we are in charge.”46 official statements, meant to assure the public that the mail, airlines, or water supply is safe, may have the opposite effect. instead of alleviating concern, such statements can increase anxiety and avoidance of an activity previously assumed to be safe. the fact that an investigation is underway can provoke fear and suspicion. warning systems often produce false alarms, leading to confusion, rumors, mistrust in the warning systems, and desensitization to future warnings.47 1.2. other uncertainty factors news coverage of the anthrax attacks promoted uncertainty through the use of speculation, conflicting reports, off-record sourcing, vague advice, and coverage of confusing incidents. speculation can fuel uncertainty and dread. experts often predict future events rather than provide statistical data.48 media speculation occurs in the absence of centralized expertise. during the anthrax attacks, journalists needed instant access to information that was unavailable, experts that were inaccessible, and statements about issues that interview sources felt unprepared to address.49 when the pentagon restricted the information flow, this led to increased speculation in news coverage.50 stories sparked outrage by predicting adverse, uncontrollable outcomes and continually warning of possible dangers.51 journalists frequently asked experts to speculate about possible outcomes. even authoritative sources provided rumors, sweeping claims, and conspiracy theories. repetitive network television news coverage often highlighted speculation about future attacks.52 conflicting statements in news coverage can trigger outrage. in a crisis, communication channels often break down. journalists must filter and interpret multiple, competing sources of information. specialization of expertise and fragmentation of knowledge create the appearance of public disputes among experts. mediaconstructed conflict often portrays a responsible government doing its best to deal with a hazardous situation, pitted against non-experts expressing fear of the unknown. when journalists are unable to sort out whether there is any real threat of harm to citizens, they typically inform the public that a controversy is occurring and identify the players on each side.53 this strategy empowers interview sources to suppress facts, manipulate information, or announce unfounded conclusions. then the public cannot decide which sources to trust and what advice to follow. 54 published by atlantis press copyright: the authors 83 k.a. swain when authoritative sources disagreed during the anthrax attacks, this led to confusing, mixed messages.55 disagreements among experts may emerge when limited authority, data, and resources are available to assess risks, when coverage fails to disclose the uncertainties and limitations of risk assessments, and when news coverage does not examine stakeholder interests and concerns.56 the perceived or actual disagreements erode public trust, leading to the belief that risks are continually underestimated, ignored, or covered up. 57 when journalists cannot access authoritative sources, they turn to off-record or otherwise unnamed sources. in an oct. 25, 2001 npr interview, bioweapons policy consultant matthew meselson remarked, “a political person, or even an outside expert who isn't authorized and fully knowledgeable and fully in contact, may not know exactly what's right. a lot of things have been attributed to unnamed sources, which is certainly the worst thing of all, 'unnamed government sources.'” when official advice is too vague, it can increase outrage, even if it is intended to reassure anxious individuals. although many officials tried to balance uncertainty and reassurance during the anthrax attacks, these messages ultimately sowed chaos and confusion.58 journalists sometimes interpreted experts' hedging language as evidence of stonewalling or incompetence, rather than a portrayal of the uncertain nature of the situation, and then looked for sources who would speak with less caution.59 when stories advise the public to ignore scare-mongering statements, they may imply that those in charge are spreading hysterical lies and deliberate distortions, which can lead to polarization, confusion, and the perception that the hazard is unpredictable and uncontrollable.60 confusing incidents, such as hoaxes and false alarms, may trigger outrage responses to a threat. during the anthrax attacks, the unfamiliar crisis raised daily questions that health experts could not answer quickly or that they did not know how to address. while some interview sources were unwilling to say “i don’t know” when facts were unavailable, others released information before key facts were known. conflicting information heightened journalists’ concern that there was more to the story that the public needed to know.61 within the first month of the attacks, media coverage depicted top officials as bumblers who failed to move aggressively against anthrax-tainted mail while offering shifting explanations of the danger.62 when the fbi and cia were stumped, they second-guessed earlier statements. for example, federal officials initially declared that the anthrax incidents were not acts of terrorism, then linked them to 9/11, and finally concluded that they were probably domestic terrorism unrelated to 9/11.63 1.3. comparability factors as an antidote for outrage, news coverage can help audiences compare or weigh risks.64 to understand a risk message, audiences must consider tradeoffs among different risks or weigh costs and benefits.65 to promote comparability, media accounts must promote rational understanding of an unfamiliar hazard. journalists are faced with the challenge of alarming the public when appropriate, without causing audiences to ignore alarms when danger is still present.66 news coverage that promotes comparability provides understanding of risk comparisons, gradients of risk, or how much it costs to reduce a risk. risk comparisons include estimated deaths or injuries/illnesses across time, time between exposure to a hazard and its effects, links between exposure to a hazard and various health impacts, and ways that citizens can control exposure to a hazard.67 elucidating explanations promote comparability by defining or clarifying a confusing risk concept or listing its essential features.68 risk-reducing statements also can reduce uncertainty by explaining why health effects are unlikely, how contamination is prevented, how anthrax infections can be treated, or how risks are reduced through preventive measures.69 most anthrax coverage failed to report on antidotes, vaccines, as well as the fact that anthrax is not contagious and that the spores are inactivated by ultraviolet light or direct moisture. even when they did report on antidotes, journalists on deadline sometimes had no opportunity to question possible special interest involvement. the media indirectly promoted the widespread use of cipro-brand antibiotic for treating anthrax infection, even though less-expensive generic versions of the same antibiotic were just as effective.70 for example, early in the crisis tom brokaw closed the published by atlantis press copyright: the authors 84 explanation of risk and uncertainty in news coverage nbc nightly news with the remark, “in cipro we trust.” 71 the present study is based on the supposition that news coverage of anthrax provoked outrage and uncertainty but also provided explanations to help citizens frame the risks rationally and put them into context. three research questions were used to guide an examination of the uncertainty and comparability factors used in this coverage, as well as the interrelationships among them: 1） which uncertainty and comparability factors characterized the anthrax coverage? 2） which media channels promoted comparability and uncertainty factors to a greater extent? 3） how did the relationships among these factors highlight strengths and weaknesses of the coverage? 2. method this content analysis examined 833 stories from major u.s. newspapers, as well as the associated press newswire and transcripts from national public radio and four u.s. television news networks (abc, cbs, cnn, nbc). within the sample, 457 (55%) of the stories came from 272 newspapers, 93 (11%) from ap, 168 (20%) from npr, and 114 (14%) from tv networks. the time frame was oct. 1-dec. 31, 2001. the first anthrax report appeared on oct. 4, when a boca raton photo editor lapsed into unconsciousness from exposure to anthrax spores. by mid-october, anthrax-laced letters had been sent to members of the national media and congress, and by nov. 1, five people had died of anthrax infection and 12 others had been infected. the anthrax crisis began to slowly wind down by december. mebane and colleagues identified the anthrax crisis period as oct. 4-dec. 3, 2001.72 the analysis extended through the end of december, so that the overall pattern and eventual decline in coverage could be evaluated. the unit of analysis was an individual story, defined as a news story or opinion article in the lexis-nexis academic universe database mentioning “anthrax” in the headline or lead. lexis-nexis searches rendered 5,389 news stories that fit these inclusion criteria. the final sample represented one in every 7th article in the universe. story corrections, abstracts, letters to the editor, non-u.s. publications, obituaries, reprints, sports stories, and digests/round-up summaries were excluded, as well as stories less than 150 words and material originating from another publication. in order to evaluate search terms and categorization schemes, 20 stories were randomly downloaded and analyzed by three coders. the results of this pilot test were used to further refine the original coding instrument. after categories were tested and coders were trained to reduce intercoder bias, five coders then independently coded the final sample of 833 stories. using cohen’s kappa, intercoder reliability was 0.88. uncertainty factors included outrage rhetoric, speculation, conflicting reports, off-record attribution, vague advice, and confusing incidents. outrage rhetoric included mentions of terrorism/bioterrorism, contagion, fear, scares, panic, or anxiety. types of speculation included food or water contamination, anthrax spraying by crop dusters or aerosol containers, economic consequences, a 9-11 link, or possible perpetrators; coders could select any number of these categories. an article contained conflicting reports if it specifically mentioned conflicting reports; coders did not evaluate whether statements within a story were conflicting. off-record attribution was coded for sources that were not identified by name. vague advice merely recommended that audience members not panic. confusing incidents included suspected but unconfirmed anthrax incidents, deliberate hoaxes, false alarms, scares or negative test results, mysterious pathways of exposure, and media organizations receiving suspicious letters. comparability factors, which may have assisted citizens in assessing their risk of anthrax exposure, included risk explanations, specific advice, antidotes, process explanations, definitions of key terms, and descriptions of transmission vectors. risk explanations included estimates of citizens' general risk of anthrax exposure, estimates of citizens' risk of exposure from handling personal mail, and risk comparisons. risk comparison was selected if a story defined or explained how one risk compared with another or if it discussed tradeoffs. a story contained specific advice if it mentioned a particular tip for avoiding anthrax exposure. stories were coded according to whether they mentioned vaccines as preventives or antibiotics as antidotes to published by atlantis press copyright: the authors 85 k.a. swain anthrax infection. process explanations included dormancy of spores, anthrax testing methods, strategies for identifying a perpetrator strain of anthrax, latency/incubation of spores, and preparedness. definitions of key terms included basic explanations of anthrax, weaponization, and basic infection types (inhalation and cutaneous). transmission vectors were natural sources (streams, dirt, etc.), postal mail, equipment or other items with residue, and air currents. several limitations should be considered when interpreting the findings. the results reflect the subjective views of five raters. stories were drawn from an online database, rather than from a random sample of all coverage of the attacks, which reduces generalizability of the results. it was assumed that stories in the database would not be qualitatively different from stories not in the database. 3. results coverage peaked during the second week of the crisis, but the amount of coverage remained intense for nearly a month after the initial story (figure 1). although a third of newspaper stories appeared on the front page, most page-one stories did not appear until a month after the initial anthrax infection was reported. overall coverage peaked dramatically oct. 15-22, when dan rather’s assistant became infected, various media outlets began receiving powdery letters, sen. tom daschle’s staff member opened a tainted envelope, and spores were discovered in the building where mail is processed for legislators (figure 1). stories containing uncertainty factors, which accounted for 98% of the coverage, included material that could promote irrational risk decisions among audiences through outrage, confusion, panic, or lack of media credibility through sensationalism or off-record attribution. stories containing outrage rhetoric accounted for 77% of the coverage. among all stories, 64% mentioned terrorism or bioterrorism, 42% mentioned fear/panic and 17% mentioned contagion. stories containing outrage rhetoric frequently offered both vague advice and specific advice, and these stories were more likely to speculate about economic impact and mention anthrax definitions and media organizations receiving suspicious letters. npr was more likely than other media to include outrage rhetoric in its coverage, and broadcast outlets were more likely than print media to mention contagion (table 1). stories mentioning terrorism also frequently mentioned contagion, fear, and speculation, including conjecture 0 25 50 75 100 125 150 175 200 225 250 10/1 10/8 10/15 10/22 10/29 11/5 11/12 11/19 11/26 12/3 12/10 12/17 12/24 figure 1: anthrax news coverage, oct. 1-dec. 4, 2001. published by atlantis press copyright: the authors 86 explanation of risk and uncertainty in news coverage table 1: significant differences in coverage, by media type relative amount of media coverage significance (x2, df) uncertainty outrage npr > papers > tv > ap 28.26 (6)** fear/panic/scares npr > tv > papers > ap 28.10 (3)** contagion npr > tv > papers > ap 49.62 (3)** speculation 9-11 link papers > tv > ap > npr 42.70 (3)** crop dusters papers > tv > ap > npr 13.25 (3)** suspects npr > tv > papers > ap 13.47 (3)** conflicting reports npr > tv > papers > ap 21.89 (3)** confusing incidents hoaxes, false alarms papers > tv > ap > npr 14.81 (6)* media receiving letters npr > tv > ap > papers 13.90 (3)** comparability risk explanations npr > tv > ap > papers 53.17 (12)** general risk npr > tv > ap > papers 14.38 (3)** mail-handling risk ap > tv > npr > papers 18.57 (3)** risk comparisons papers > tv > npr > ap 17.58 (3)** antidotes/preventives vaccine npr > ap > papers > tv 18.05 (6)** definitions ap > tv > npr > papers 49.84 (21)** anthrax ap > npr > tv > papers 23.80 (12)* weaponization tv > npr > papers > ap 18.55 (3)** cutaneous infection tv > papers > npr > ap 8.17 (3)* transmission vectors tv > ap > papers > npr 34.56 (15)** postal mail tv > ap > papers > npr 7.84 (3)* equipment residue ap > tv > papers > npr 46.06 (3)** ** = p<.01; * = p<.05; n = 833 about crop dusters, suspects, or a 9-11 link. these stories also frequently mentioned explanations about weaponization, exposure risk, and anthrax testing (table 2). broadcast media were more likely to mention fear than print media. stories that mentioned fear also frequently included mentions of contagion, methods for identifying perpetrator strains, media organizations receiving suspicious letters, and transmission vectors including mail and equipment residue. these stories often included both vague advice and specific advice. coverage of contagion often appeared in concert with speculation about crop dusters and aerosol dispersion, explanations about incubation and anthrax dissemination via natural sources, air currents, and mail, as well as definitions of anthrax and weaponization. these stories were more likely to include vague advice than specific advice, and they were more likely to address inhalation than cutaneous infection. speculation, mentioned in half the coverage, discussed various “what ifs”: suspects (26% of all stories), economic consequences (5%), crop dusters (4%), food/water contamination (3%), and aerosol dispersion of anthrax (2%). speculation frequently accompanied coverage of risk comparisons, as well as explanations about dormancy and incubation of spores, bioterrorism preparedness, vaccines, and equipment residue. however, these speculation stories also were more likely to offer vague advice than specific advice and often speculated about more than one issue. stories that speculated about a 9-11 link were more likely to published by atlantis press copyright: the authors 87 k.a. swain mention crop dusters, food/water contamination, and suspects, while stories mentioning food or water contamination were more likely to mention crop dusters and aerosol dispersion. newspapers were the most likely to speculate about a possible 9-11 link or crop dusters, while broadcast media speculated about suspects more often than print media. vague advice, which merely recommended that people not panic, appeared in nearly 6% of the coverage. stories containing vague advice also frequently mentioned specific advice, contagion, risk comparisons, and transmission vectors. stories that included conflicting reports, 13% of the coverage, often mentioned fear, speculation about a 911 link, food/water contamination and aerosol dispersion, mysterious infections, media organizations receiving suspicious letters, risk comparisons, and transmission vectors including air currents. these stories also were likely to include process explanations, including dormancy and incubation of spores, weaponization, techniques for identifying perpetrator strains, bioterrorism preparedness, anthrax vaccines, and analysis procedures. process explanation stories were more likely to report anthrax infections through inhalation, a more serious infection route than skin contact. broadcast media were more likely than print media to present conflicting reports. stories containing offrecord attribution frequently included speculation, conflicting reports, a 9/11 link, risk comparisons, methods for identifying perpetrator strains, and anthrax dissemination via air currents. however, stories that used traditional attribution were more likely to include uncertainty factors, including outrage rhetoric, conflicting reports, and confusing incidents. stories with conventional attribution were more likely than those citing off-record sources to promote overall table 2: uncertainty and comparability factors in anthrax coverage uncertainty 820 (98.4 %) comparability 833 (100.0 %) outrage: 645 (77.4 %) risk explanations: 427 (51.3 %) fear/panic 349 (41.9 %) general risk 393 (47.2 %) contagion 143 (17.2 %) mail risk 279 (33.5 %) terrorism 530 (63.6 %) risk comparisons 63 (7.6 %) specific advice 137 (16.4 %) vague advice 46 (5.5 %) antidotes/preventives: 428 (51.4 %) antibiotics 418 (49.5 %) speculation: 425 (51.0 %) vaccinations 32 (3.8 %) 9/11 link 265 (31.8 %) process explanations: 115 (13.8 %) food/water contamination 23 (2.8 %) dormancy of spores 18 (2.2 %) economic consequences 38 (4.6 %) anthrax testing 70 (8.4 %) crop dusters 31 (3.7 %) perpetrator strain id 23 (2.8 %) aerosol dispersion 13 (1.6 %) latency / incubation 23 (2.8 %) suspects 213 (25.6 %) preparedness 18 (2.2 %) definitions: 687 (82.5 %) conflicting reports 106 (12.7 %) anthrax 677 (81.3 %) weaponization 118 (14.2 %) off-record attribution 310 (37.2 %) cutaneous infection 238 (28.6 %) inhalation infection 367 (44.1 %) confusing incidents: 510 (61.2 %) transmission vectors: 660 (79.2 %) hoaxes, false alarms 391 (46.9 %) postal mail 548 (65.8 %) media receiving letters 177 (21.2 %) natural anthrax sources 91 (10.9 %) mysterious infections 122 (14.6 %) postal equipment residue 201 (24.1 %) air currents 109 (13.1 %) ** = p<.01; * = p<.05; percentages out of total stories (n = 833) published by atlantis press copyright: the authors 88 explanation of risk and uncertainty in news coverage comparability, mention transmission vectors, and offer specific advice (table 3). most stories, about 61%, mentioned an unconfirmed anthrax incident, hoax, false alarm, negative test result, mysterious pathway of exposure, or a media organization receiving suspicious letters. nearly half of all stories mentioned hoaxes or false alarms, 21% mentioned media organizations receiving suspicious letters, and 13% of stories stated that the source of anthrax was mysterious. stories mentioning a confusing incident often included uncertainty factors: fear, conflicting reports, and speculation including a 9/11 link and aerosol dispersion. comparability factors associated with this coverage included risk comparisons, antidotes/preventives, anthrax testing, methods for identifying perpetrator strains, preparedness, and cutaneous infection. stories covering confusing incidents such as false alarms were likely to mention anthrax transmission vectors, including dissemination via mail and office equipment. broadcast media were more likely than print media to devote attention to journalists receiving suspicious letters. only 7% of stories that mentioned media organizations receiving suspicious letters discussed suspects. stories that mentioned hoaxes or false alarms were more likely to mention fear, speculation about aerosol dispersion of spores, process explanations, transmission vectors, mysterious infections, and anthrax testing. newspapers were more likely than other media to cover hoaxes and false alarms, while npr offered the least amount of this coverage (table 1). comparability factors, which appeared in all coverage, included explanations to help audiences understand a hazard and put it into context. elucidating explanations included descriptions of relative risk, antidotes/preventives, and processes, as well as transmission vectors, specific advice, and definitions of key terms. coverage did not begin to emphasize that anthrax was a threat to citizens until the third week of the crisis; afterward the risk was framed as moderate to serious. nearly half of stories mentioned that an average person is at general risk of anthrax exposure, and half of the general risk coverage appeared in the first three weeks of the crisis. more than a third of coverage mentioned that an average person is at risk of exposure from handling personal mail, and 43% of this coverage appeared in the first three weeks of the crisis. three-fourths of stories stating that citizens were at no risk of exposure appeared in the first three weeks of the scare. among the stories mentioning that citizens were at risk, more than half addressed both general risk and the table 3: attribution patterns on-record attribution significance (x2, df) off-record attribution significance (x2, df) uncertainty 512 (61.5 %) 20.27 (10)* 308 (37.0 %) outrage 401 (48.1 %) 9.92 (2)** 244 (29.3 %) speculation 257 (30.9 %) 168 (20.2 %) 16.32 (6)** conflicting reports 56 (6.7 %) 13.45 (2)** 50 (6.0 %) 5.15 (1) * vague advice 35 (4.2 %) 11 (1.3 %) confusing incidents 286 (34.3 %) 9.71 (4)* 224 (26.9 %) comparability 523 (62.8 %) 39.33 (12)** 310 (37.2 %) risk explanations 277 (33.3 %) 150 (18.0 %) specific advice 83 (10.0 %) 3.61 (1)* 54 (6.5 %) antidotes/preventives 249 (29.9 %) 179 (21.5 %) process explanations 66 (7.9 %) 49 (5.9 %) definitions 414 (49.7 %) 49.08 (14)** 273 (32.8 %) 24.32 (7) ** transmission vectors 390 (46.8 %) 34.57 (5) ** 270 (32.4 %) total sample 523 (62.8 %) 310 (37.2 %) ** = p<.01; * = p<.05; n = 833 published by atlantis press copyright: the authors 89 k.a. swain specific risk of handling postal mail. npr was the most likely channel to provide a variety of risk explanations and to mention that an average person is personally at risk of anthrax exposure from handling mail. television also covered mail risk frequently, but wire stories and newspapers were least likely to mention this risk. however, the general risk of handling mail was mentioned most often in the ap wire stories, followed by tv, npr, and newspapers. only 5% of stories included a risk comparison. however, these stories were twice as likely to mention terrorism and fear as transmission vectors. newspaper stories were more likely than broadcast stories to include risk comparison. stories containing risk comparisons were more likely to speculate about aerosol dispersion, food/water contamination, and economic consequences, mention mysterious infections, media organizations receiving suspicious letters, vaccines, and process explanations about spore dormancy, anthrax testing, perpetrator strains, and preparedness (table 1). stories containing risk comparisons were more likely to mention infection by inhalation than skin contact. process explanations appeared in 14% of the coverage. these stories were more likely to mention speculation about food/water contamination and aerosol dispersion, risk explanations, antidotes/preventives, transmission vectors including air currents, natural sources and definitions including anthrax definitions. about 16% of stories contained specific advice. for example, an oct. 25 npr story reported that “people need to be vigilant. if they receive a package or an envelope that looks suspicious, they should not open it. set it down, wash yourself off, and call law enforcement officials.” stories containing specific advice also often mentioned vague advice, in addition to speculation about suspects and crop dusters, key definitions, and transmission vectors. stories that mentioned anthrax infection antidotes/preventives accounted for 51% of the coverage. these stories often mentioned confusing incidents, and definitions including anthrax definitions. these stories also were more likely to mention incubation and dormancy of spores, and anthrax testing. stories mentioning antibiotics were somewhat more likely to mention infection from skin contact than inhalation. a fourth of all stories specifically mentioned cipro-brand antibiotic. npr was more likely than print media or television to mention vaccines. definitions of key terms appeared in 83% of the coverage. three-fourths of stories defined anthrax. ap defined anthrax more often than broadcast media or newspapers, while the broadcast media defined weaponization more often than print media. tv mentioned cutaneous anthrax more often than any other medium. among newspapers, 44% of stories mentioned inhalation infection, while 28% mentioned skin infection. stories containing key definitions were more likely to contain specific advice. they more frequently included off-record attribution and mentioned speculation about aerosol dispersion and suspects, mysterious infections, antibiotics, perpetrator strains, and media organizations receiving suspicious letters (table 3). stories that mentioned transmission vectors accounted for 79% of the coverage; 66% of these stories mentioned mail as a vector, 24% mentioned equipment residue, 13% mentioned air currents, and 11% mentioned a natural source. these stories often contained confusing incidents, conflicting reports, and speculation about food/water contamination, aerosol dispersion, and crop dusters. these stories were more likely to include definitions and risk explanations including risk comparisons. tv mentioned these vectors more often than print media; tv was also the most likely to mention mail as a vector. ap mentioned equipment residue as a vector more often than tv, and npr was the least likely to mention transmission vectors in its coverage (table 1). 4. conclusions the variables that comprised the uncertainty and comparability factors were statistically clustered, and linkages among the factors revealed complex patterns of reporting; overall anthrax reporting was both scary and beneficial. much of the anthrax coverage was characterized by outrage rhetoric, speculation, and confusing incidents, which ultimately framed the attacks as an involuntary hazard controlled by others – a perception that often leads to public outrage and irrational risk decisions. however, overall coverage also promoted comparability, primarily through definitions and explanations about risk, transmission vectors, and antidotes/preventives. risk comparisons, specific advice, and process explanations were sparse – key weaknesses in the coverage. published by atlantis press copyright: the authors 90 explanation of risk and uncertainty in news coverage much of the outrage rhetoric in the coverage was characterized by references to terrorism and fear. outrage coverage was particularly linked to worries about the economic impact of the attacks. however, many stories that contained coverage that could spark outrage also contained explanatory information. official advice was relatively rare, but stories offering vague advice were more common than those containing practical advice and were more likely to mention fear, contagion, or speculation. stories mentioning terrorism often contained many uncertainty factors, but also were likely to explain risks and anthrax testing. mentions of fear, as a form of outrage rhetoric, were more likely to appear in coverage of other outrage-provoking coverage including mentions of contagion, journalists receiving suspicious letters, hoaxes, and conflicting reports. however, fear content also was more likely to appear alongside practical advice and explanations of transmission vectors. similarly, stories that mentioned contagion were more likely to discuss both known and hypothetical routes of anthrax dissemination. stories that speculated about “what if” scenarios were more likely than not to include risk comparisons and explanations about processes, transmission, and prevention. conflicting reports commonly accompanied scary rhetoric and scenarios, including coverage of hoaxes, but they also were more likely than not to include explanations and risk comparisons. off-record attribution was strongly linked to overall uncertainty (table 3). stories that covered confusing incidents such as hoaxes were more likely than not to promote uncertainty by mentioning fear, speculation, and conflicting reports, but they frequently included comparability factors as well, including risk comparisons and process explanations. stories that reported hoaxes and false alarms promoted high outrage but often did include explanatory content about processes and transmission vectors. the shift in news coverage from an emphasis on outrage rhetoric to more balanced explanations of risk, coincided with increasing public knowledge about anthrax. the chaos of conflicting information about anthrax subsided after the first few weeks, and once audiences learned more about anthrax from different media, the fearful news content gradually subsided. uncertainty factors were mentioned twice as often during the crisis phase of the coverage than during the outbreak phase, and eight times more often during the crisis phase than during the post-crisis phase. similarly, explanatory content was mentioned nine times more often during the crisis phase than during the post-crisis phase and nearly three times as often during the crisis phase than during the outbreak phase. anthrax stories included outrage rhetoric, speculation, conflicting reports, vague advice, or coverage of confusing incidents 33 percent more often than they included explanatory content. future research could examine news coverage of speculation, conflicting reports, and confusing incidents during a disaster, to determine whether there is a causal relationship between this coverage and actual outrage among audience members. a study also might identify risk comparisons that effectively promote rational risk decisions in disasters, particularly when a risk is perceived as more threatening than everyday, comparable risks because of public dread and catastrophic potential. the findings highlight the need for journalists to build greater trust with the official and independent authorities they may need to interview in the midst of a high-threat crisis. when officials suppress information, the balance between overand under-estimating threats suffers. in light of the outrage sparked by media coverage of the anthrax attacks, this study also highlights the necessity for journalists to provide context when discussing uncertainties and speculation, provide elucidating risk explanations, offer practical advice, and clarify contradictions. improved risk coverage of similar crises could strive to connect daily events to larger issues, emphasize facts over drama, and avoid amplifying or ignoring risks. pre-event coverage about bioterrorism preparedness could help audiences anticipate what they might encounter in different scenarios and ways to avoid exposure. these stories also might explain unfamiliar concepts, explore tradeoffs needed to reduce the threat, and address common misconceptions and speculation. acknowledgements the author thanks dr. thomas mason, who obtained funding from the college of public health at the published by atlantis press copyright: the authors 91 k.a. swain university of south florida to support this study. the author also appreciates the efforts of coders patrick lafferty at the university of kansas, keren nishry at the university of arkansas at little rock, and lisa rademakers, linda young, and bryan nichols at the university of south florida st. petersburg. references 1. f. kittler, j. hobbs, l. a. volk, g. l. kreps, and d. w. bates, the internet as a vehicle to communicate health information during a public health emergency: a survey analysis involving the anthrax scare of 2001, journal of medical internet research 6 (january 2004): e8. 2. k. j. dooley and s. r. corman, the dynamics of electronic media coverage, in communication and terrorism: public and media responses to 9/11, ed. b. s. greenberg (cresskill, nj: hampton press, 2002), 121-135; f. mebane, s. temin and c. f. parvanta, communicating anthrax in 2001: a comparison of cdc information and print media accounts, journal of health communication 8 (june 2003), 50-82; b. l. nacos, mass-mediated terrorism: the central role of the media in terrorism and counterterrorism (lanham, md: rowman & littlefield, 2002). 3. k. guckenberger, rising to the occasion, the public perspective 13 (july 2002): 31. 4. d. chandler and i. landrigan, bioterrorism: a journalist's guide to covering bioterrorism (washington, dc: radio and television news directors foundation, 2003). 5. k. guckenberger, rising to the occasion, the public perspective 13 (july 2002): 31. 6. s. ricchiardi, the anthrax enigma, american journalism review, december 2001, 20. 7. a. m. major and l. e. atwood, changes in media credibility when a predicted disaster doesn't happen, journalism and mass communication quarterly 74 (fall 1997): 797-813; l. rainie, pew internet and american life: survey with federal computer week magazine about emergencies and the internet (washington, dc: pew internet and american life project, 2003); w. e. pollard, public perceptions of information sources concerning bioterrorism before and after anthrax attacks: an analysis of national survey data, journal of health communication 8 (june 2003): 93-103; e. m. rogers, diffusion of news of the september 11 terrorist attacks, in crisis communications: lessons from september 11, ed. a. m. noll (lanham, md: rowman & littlefield, 2003), 17-31. 8. c. e. prue, c. lackey, l. swenarski and j. m. gantt, communication monitoring: shaping cdc's emergency risk communication efforts, journal of health communication 8 (june 2003): supplement. 9. d. drache, s. feldman and d. clifton, media coverage of the 2003 toronto sars outbreak: a report on the role of the press in a public crisis (toronto: robarts centre for canadian studies, 2003); k. h. jamieson, k. lammie, c. wardle and s. krutt, questions about hypotheticals and details in reporting on anthrax, journal of health communication 8 (june 2003): 121-131; w. h. reid, bioterrorism: separating fact, fiction, and hysteria, in the psychology of terrorism: a public understanding, ed. c. e. stout (westport, ct: praeger, 2002); j. p. roche and m. a. t. muskavitch, limited precision in print media communication of west nile virus risks, science communication 24 (march 2003): 353-365. 10. d. chandler and i. landrigan, bioterrorism: a journalist's guide to covering bioterrorism (washington, dc: radio and television news directors foundation, 2003). 11. j. hobbs, a. kittler, s. fox, b. middleton and d. w. bates, communicating health information to an alarmed public facing a threat such as a bioterrorist attack, journal of health communication, 9 (january 2004): 67-75. 12. n. cortes, scientists, doctors call coverage of anthrax scare fair, balanced, freedom forum (arlington, va: freedom forum, december 12, 2001): http://www.freedomforum.org/templates/document. asp?documentid=15521. 13. new york times, a nation challenged: poll finds support for war and fear on economy, new york times, 25 september 2001, p. a1. 14. t. s. glickman, d. golding and e. d. silverman, acts of god and acts of man: recent trends in natural disasters and major industrial accidents. (darby, pa: diane publishing, 1993). 15. m. leitch, iso 31000: 2009: the new international standard on risk management, risk analysis 30 (2010), 887-892. published by atlantis press copyright: the authors 92 explanation of risk and uncertainty in news coverage 16. d. v. maccollum, construction safety engineering principles: designing and managing safer job sites. mcgraw-hill professional (december 2006). 17. v. t. covello, d. von winterfeldt and p. slovic, communicating scientific information about health and environmental risks: problems and opportunities from a social and behavioral perspective. in v. t. covello, l. b. lave, a. moghissi and v. r. uppuluri (eds.), uncertainty in risk assessment, risk management and decision making (1987, pp. 112). 18. v. t. covello, p. m. sandman and p. slovic, risk communication, risk statistics and risk comparisons: a manual for plant managers (washington, dc: chemical manufacturers association, 1988), 79. 19. m. g. morgan and m. henrion, uncertainty: a guide to dealing with uncertainty in quantitative risk and policy analysis (new york: cambridge university press, 1992). 20. k. m. thompson, variability and uncertainty meet risk management and risk communication. risk analysis, 22 (2002): 647–654. 21. s. krimsky and a. plough, environmental hazards: communicating risks as a social process (dover, ma: auburn house publishing, 1988), 6. 22. v. covello, the perception of technological risks: a literature review, technological forecasting and social change 23 (1983): 285-297. 23. s. klaidman, health in the headlines: the stories behind the stories (new york: oxford university press, 1991). 24. s. klaidman, health in the headlines: the stories behind the stories (new york: oxford university press, 1991). 25. p. m. sandman, four kinds of risk communication, the synergist (april 2003): 26–27. 26. p. m. sandman, hazard versus outrage in the public perception of risk, in effective risk communication: the role and responsibility of government and nongovernmental organizations, eds. v. t. covello, d. b. mccallum and m. t. pavlova (new york: plenum press, 1989); s. kuznesof and c. ritson, consumer acceptability of genetically modified foods with special reference to farmed salmon, british food journal 98 (may 1996): 39-47; c. vlek and p. j. stallen, rational and personal aspects of risk, acta psychologica 45 (1980): 273300; r. b. schafer, e. schafer, g. l. bultena and e. o. hoiberg, food safety: an application of the health belief model, journal of nutrition education 25 (winter 1993), 17-23. 27. p. m. sandman, environmental risk and the press: an exploratory assessment (piscataway, nj: transaction publishers, 1987). 28. s. klaidman, health in the headlines: the stories behind the stories (new york: oxford university press, 1991). 29. s. g. hadden, regulating product risks through consumer information, journal of social issues 47 (winter 1991): 93-105; s. g. hadden, a citizen's right to know: risk communication and public policy (san francisco: westview press, 1989). 30. l. m. thomas, why we must talk about risk, in risk communication: proceedings of the national conference on risk communication, eds. j. c. davies, v. t. covello & f. w. allen (washington, dc: conservation foundation, 1987), 21. 31. s. c. ratzan, real risks: the need for health leadership and security. journal of health communication 6, 295. 32. p. slovic, perceptions of risk, science 236 (april 7, 1987): 280–285. 33. l. wilkins, shared vulnerability: the media and american perception of the bhopal disaster (westport, ct: greenwood press, 1987). 34. v. t. covello, d. von winterfeldt, and p. slovic, communicating scientific information; s. hornig, reading risk: public response to print media accounts of technological risk, public understanding of science 2 (1993): 95-109; k. lang and g. e. lang, the mass media and voting, in reader in public opinion and communication, eds. b. berelson and m. janowitz (new york: free press, 1966). 35. j. flynn, p. slovic and h. kunreuther, risk, media and stigma: understanding public challenges to modern science and technology (london: earthscan publications, 2001); e. singer and p. m. endreny, reporting on risk (new york: russell sage foundation, 1992). 36. k. frost, e. frank and e. maibach, relative risk in the news media: a quantification of misrepresentation, american journal of public health 87 (may 1997): 842-845. 37. a. alm, t. vacor, et al., responsibilities of risk communicators. risk communication. j. c. davies, published by atlantis press copyright: the authors 93 k.a. swain v. t. covello and f. w. allen (washington, dc, the conservation foundation, 1987): 36. 38. new york times, anthrax missteps offer guide to fight next bioterror battle, new york times, january 6, 2002, p. a1. 39. s. dunwoody and k. neuwirth, coming to terms with the impact of communication on scientific and technological risk judgments, in risky business, eds. l. wilkins and p. patterson (new york: greenwood press, 1991). 40. j. b. tucker, what the anthrax attacks should teach us, hoover digest 3 (winter 2002). 41. v. t. covello, p. m. sandman and p. slovic, risk communication, risk statistics and risk comparisons: a manual for plant managers (washington, dc: chemical manufacturers association, 1988). 42. a. edwards, g. elwyn, j. covey, e. matthews and r. pill, presenting risk information: a review of the effects of framing and other manipulations on patient outcomes, journal of health communication 6 (january 2001): 61-82. 43. c. s. russell, guidebook for risk perception and communication in water resources planning, part i (fort belvoir, va: u.s. army corps of engineers, 1993). 44. j. gordon, risk communication and foodborne illness: message sponsorship and attempts to stimulate perceptions of risk, risk analysis 23 (december 2003): 1287. 45. c. coleman, science, technology and risk coverage of a community conflict, media, culture and society 17 (january 1995): 65-79. 46. s. krimsky and a. plough, environmental hazards. 47. p. slovic, perceptions of risk, science 236 (april 7, 1987): 280–285. 48. r. a. zilinskas, b. hope and d. w. north, a discussion of findings and their possible implications from a workshop on bioterrorism threat assessment and risk management, risk analysis 24 (april 2004): 901. 49. w. lowrey, w. evans, j. a. robinson and k. k. gower, media response to bioterrorism and emerging infectious diseases: pressing problems and plausible solutions, annual meeting of the association for education in journalism and mass communication, toronto, august 2004. 50. k. guckenberger, rising to the occasion, the public perspective 13 (july 2002): 31. 51. f. furedi, culture of fear: risk-taking and the morality of low expectation (new york: continuum international publishing group, 2002). 52. s. c. ratzan, real risks: the need for health leadership and security. journal of health communication 6, 295. 53. s. beder and m. shortland, siting a hazardous waste facility: the tangled web of risk communication, public understanding of science 1 (spring 1992): 139-160. 54. managed care strategies, americans trust physicians, not government, with medical info, managed care strategies 5 (october 1997): p. 119120. 55. f. kittler, j. hobbs, l. a. volk, g. l. kreps, and d. w. bates, the internet as a vehicle to communicate health information during a public health emergency: a survey analysis involving the anthrax scare of 2001, journal of medical internet research 6 (january 2004): e8. 56. v. t. covello, d. von winterfeldt and p. slovic, communicating scientific information about health and environmental risks: problems and opportunities from a social and behavioral perspective. in v. t. covello, l. b. lave, a. moghissi and v. r. uppuluri (eds.), uncertainty in risk assessment, risk management and decision making (1987, pp. 112). 57. f. furedi, culture of fear: risk-taking and the morality of low expectation (new york: continuum international publishing group, 2002). 58. washington post, how do we treat this outbreak of uncertainty? washington post, november 4 2001, p. b1, b3. 59. w. lowrey, w. evans, j. a. robinson and k. k. gower, media response to bioterrorism and emerging infectious diseases: pressing problems and plausible solutions, annual meeting of the association for education in journalism and mass communication, toronto, august 2004. 60. s. beder and m. shortland, siting a hazardous waste facility: the tangled web of risk communication, public understanding of science 1 (spring 1992): 139-160, 139. 61. n. cortes, scientists, doctors call coverage of anthrax scare fair, balanced, freedom forum published by atlantis press copyright: the authors 94 explanation of risk and uncertainty in news coverage (arlington, va: freedom forum, december 12, 2001): http://www.freedomforum.org/templates/document. asp?documentid=15521. 62. washington post, for every cool head, a thousand overheated mouths, washington post, october 29, 2001, p. c1. 63. s. ricchiardi, the anthrax enigma, american journalism review, december 2001, 20. 64. s. klaidman, health in the headlines: the stories behind the stories (new york: oxford university press, 1991). 65. v. t. covello, d. von winterfeldt and p. slovic, communicating scientific information about health and environmental risks: problems and opportunities from a social and behavioral perspective. in v. t. covello, l. b. lave, a. moghissi and v. r. uppuluri (eds.), uncertainty in risk assessment, risk management and decision making (1987, pp. 112). 66. w. lowrey, w. evans, j. a. robinson and k. k. gower, media response to bioterrorism and emerging infectious diseases: pressing problems and plausible solutions, annual meeting of the association for education in journalism and mass communication, toronto, august 2004. 67. b. b. johnson, varying risk comparison elements: effects on public reactions, risk analysis 24 (january 2004): 103. 68. k. e. rowan, a contemporary theory of explanatory writing, written communication 5 (spring 1988): 23-56; k. e. rowan, strategies for explaining complex science news, journalism educator 45 (summer 1990): 25-31; r. d. tennyson and m. j. cocciarella, an empirically based instructional design theory for teaching concepts, review of educational psychology 56 (1986): 40-71. 69. j. gordon, risk communication and foodborne illness: message sponsorship and attempts to stimulate perceptions of risk, risk analysis 23 (december 2003): 1287. 70. s. c. ratzan, real risks: the need for health leadership and security. journal of health communication 6, 295. 71. s. c. ratzan, real risks: the need for health leadership and security. journal of health communication 6, 295. 72. new york times, censored study on bioterror doubts u.s. preparedness, new york times, march 29, 2004, p. a17; mebane, communicating anthrax in 2001, 50. published by atlantis press copyright: the authors 95 atlantis press journal style acknowledgements to referees and staffs for jracr in 2015 by chongfu huang the quality of journal of risk analysis and crisis response (jracr) depends on the qualified and regular collaboration of renowned scientists, who devoted their time to constructively review the submitted articles. i wish to thank the following individuals who acted as referees for the journal in 2015. it is a great pleasure to thank the staffs serving for jracr. i appreciate these volunteers who have donated their time without compensation to promote the journal. referees of jracr: hongbin cao xuan cha xiaoli chen zhifen chen xingwei chen yung-hsiang cheng sing chew tadashi dohi jose l. domingo k. harald drager shanghai du james fogal ilia frenkel duane gill qinghua gong eduardo a. haddad song han vincent ho dana hoag longhua hu juliang jin cengiz kahraman changzhi li jiuyi li shouzhong li kaihua liao dedi liu hao liu tiezhong liu hao liu yaolong liu yuanyuan liu ying long dusan marcek stephen a. nelson jinren ni david pellow longxia qian xiaobo qu guosheng qu mangey ram h.c.ortwin renn naiqi shen congling shi runhe shi douglas smith wenchao sun shujun tian vinay tyagi dong wang hongrui wang dong wang wenguo weng xinliang xu wei xu guiyun you lejiang yu guofang zhai jingyan zhang jiquan zhang mu zhang jun zhao zhou zhao hanping zhao xiaomeng zhou staffs of jracr: editorial board office of jracr mu zhang hongmei zhang wen han ziyan wang xiaobo lu editorial department of jracr junxiang zhang fei wen hongbing zhu jingjing cheng reviewer center of jracr xilei pang ye xue fuping yang chongfu huang, professor, ph.d. academy of disaster reduction and emergency management beijing normal university no.19 xinjiekouwai street beijing 100875, china journal of risk analysis and crisis response, vol. 5, no. 4 (december 2015), 264 published by atlantis press copyright: the authors 264 microsoft word volume 11, issue 4-4 journal of risk analysis and crisis response, 2021, 11(4), 189-197 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i4.312 189 review a literature review on digital finance, consumption upgrading and high-quality economic development gang feng 1,2 and mu zhang 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: december 2, 2021; accepted: december 23, 2021; published: january 25, 2022 abstract: under the background of china's high-quality development and "domestic and international double circulation", it is required to gradually optimize the economic structure and social structure while ensuring economic growth. the role of consumption is increasingly prominent. the development of digital finance can promote residents' consumption and consumption upgrading, which in turn can promote high-quality economic development. in order to promote high-quality economic development, this paper reviews the literature in relevant fields, and finds that the literature that studies the three topics simultaneously is still very rare. the research on digital finance and high-quality economic development from the perspective of consumption upgrading is still very rare. it is suggested that the corresponding research can be conducted from the perspective of consumption upgrading in the future. keywords: digital finance; upgrading of consumption; high-quality economic development; literature review 1. introduction china's economy has turned to a stage of high-quality development. improving the quality of economic development has become the focus of economic work. under the background of highquality development and "domestic and international double circulation", the role of consumption is increasingly prominent. high-quality development is a process of gradually optimizing the economic structure and social structure while maintaining economic growth. the increase of residents' income, the change of consumption demand and the abundance of product types and quality make the contribution of consumption to the economic development increase day by day. the change of economic development mode is stimulated by the consumption upgrading, thus promoting the domestic circulation. the enhancement of the basic effect of consumption in the economic development is an important impetus to realize the high-quality economic development. digital finance refers to the traditional financial institutions and internet companies that use digital technology to realize financing, payment, investment and other new financial business models. its concept is basically similar to "internet finance" defined by ten ministries and commissions such as the people's bank of china and "fintech (financial technology)" defined by the financial stability gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 190 board [1]. practice and many researches show that the development of digital finance can promote the consumption and consumption upgrading of residents, and the consumption upgrading can promote the high-quality development of economy. therefore, this paper will sort out the research status quo of digital finance, consumption upgrading and high-quality economic development, and make a brief review, in order to promote the research of theories and methods among them. 2. research on digital finance and consumption upgrading 2.1. internet finance and consumption upgrading research on internet finance shows that it cannot only promote residents' consumption, but also promote the upgrading of residents' consumption. cui haiyan uses the dynamic time series model, empirical research shows that the development of internet finance can promote the consumption of chinese residents [2]. zhang liyi and tu ben study from the perspective of consumer finance functionality, and conclude that the development of internet finance can significantly affect the upgrading of residents' consumption structure [3]. shang huawei found that the development of internet finance significantly affects the consumption upgrading of urban and rural residents in china, and is more supportive than traditional formal finance and informal finance [4]. the research results of he qizhi and peng mingsheng show that internet finance is stimulating the growth of residents' consumption and at the same time can guide the change of residents' consumption pattern [5]. liu tongtong and wu fuxiang found that internet finance can improve the consumption of rural residents by easing liquidity constraints, while traditional finance has little impact on the consumption of rural residents [6]. qi hongqian and ma qijun research shows that internet finance promotes the upgrading of urban households' consumption structure by improving financial availability, guiding credit demand and increasing urban households' enjoyment-oriented and development-oriented consumption [7]. 2.2. digital inclusive finance and consumption upgrading studies have shown that digital inclusive finance can promote the growth of consumption and the upgrading of its structure in various ways. campbell and mankiw pointed out that the emergence of digital inclusive finance has greatly expanded the scope of financial services and reduced the liquidity constraint, so that residents who were originally excluded from the financial services can also realize the intertemporal smoothing of consumption through the financial services, thus releasing the consumption demand. this indicates that financial constraints will affect residents' consumption [8]. the research of yi hongjian and zhou li shows that digital inclusive finance can significantly promote residents' consumption by easing the mobility constraint and facilitating residents' payment, and mainly promote the consumption expenditure of clothing, residence, daily necessities, transportation and communication and other goods and services [9]. guo hua et al. pointed out that the development of digital inclusive finance can significantly promote the consumption of rural residents, and digital payment services play the most significant role in promoting the consumption of rural residents, while digital credit services play a limited role in promoting consumption [10]. tang shihui proposed that the digital inclusive finance has a significant positive effect on residents' consumption, while the western region is more gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 191 significant than the central and eastern region, and the rural residents are more significant than the urban residents [11]. xiao yuanfei and zhang keyang proposed that digital inclusive finance can improve the consumption level of rural residents, and the coverage and digitalization of digital inclusive finance have significantly improved the consumption level of urban and rural residents, but the depth of use has a certain inhibitory effect on the consumption of urban and rural residents [12]. jiang zhuyuan found that digital inclusive finance can significantly promote the consumption level of residents, the breadth and depth of digital inclusive finance can significantly promote the consumption of residents, while digital inclusive finance can narrow the consumption gap between urban and rural areas [13]. zou xinyue and wang wang found that digital inclusive finance can effectively promote the improvement of residents' consumption level through income, mobile payment, consumer credit and insurance [14]. nan yongqing et al. proposed that digital inclusive finance should rely on the grasp of dynamic consumer behavior information to promote the downward movement of digital financial products and services, so as to promote consumer spending [15]. wang mingyang found that the digital inclusive finance can significantly promote the consumption of residents, and has a significant role in promoting the consumption growth of high-income groups, while the promotion role of middle-income groups is higher than that of high-income groups, and the promotion role of lowincome groups has not been fully released [16]. chen xiaoxia research shows that digital inclusive finance is beneficial to promote the upgrading of residents' consumption, and mainly through the income mechanism to promote the upgrading of consumption [17]. jiang hongli and jiang pengcheng pointed out that digital inclusive finance has improved the residents' consumption level and optimized the consumption structure by narrowing the urban-rural income gap and optimizing the industrial structure [18]. yan jianjun and feng junyi found that digital inclusive finance can significantly promote the consumption upgrading of urban and rural residents. digital inclusive finance promotes the survival and development consumption of rural residents by promoting the development of the tertiary industry, which in turn promotes the consumption upgrading of rural residents [19]. tang yong et al. concluded that digital inclusive finance and its various dimensions can play a role in promoting the upgrading of rural residents' consumption [20]. 2.3. digital finance and consumption upgrading research by many scholars shows that digital finance can increase consumption and promote consumption upgrading. lu caimei and wang haiyan, based on keynes's absolute income hypothesis and using panel data of 280 prefecture-level cities in china, conducted an empirical study on digital finance, income gap and household consumption, and concluded that digital finance significantly promoted household consumption [21]. research by huang kai nan and hao xiangru shows that the development of digital finance has significantly improved the household consumption level of urban and rural residents, but the impact mechanism is different: digital finance promotes the household consumption of urban residents by increasing the income of residents and reducing the risk of household uncertainty, and improves the household consumption of rural residents by facilitating payment and enhancing mobility [22]. li et al. found that online shopping, digital payment, online credit acquisition, purchase of internet wealth management products and commercial insurance are the main intermediary variables of digital finance affecting residents' consumption [23]. gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 192 many scholars' research shows that digital finance can promote consumption upgrading by easing liquidity constraints. jappelli and pagano proposed that the development of finance can ease the liquidity constraints of consumers, spread the risks, smooth the consumption expenditure and improve the consumption [24]. gross and souleles research shows that consumer finance can break the liquidity constraint of consumers and significantly promote consumption [25]. levchenko believes that financial development can be achieved through a reasonable and effective allocation of resources, so that those consumers who are constrained by the liquidity can easily use the financial market to achieve crossperiod smoothing of consumption, thus releasing pent-up consumption demand and thus promoting consumption growth [26]. karlan and zinman found that the increase in credit card limit promoted the consumption of residents, which proved that finance can ease the liquidity constraint [27]. zhao proposed that financial development can promote residents' consumption by easing liquidity constraints and smoothing consumption, reducing the over-sensitivity of consumers' income, diversifying risks and reducing intermediary costs [28]. research by xie jiazhi and wu jingru shows that digital finance, because of its convenience, low transaction cost and universal availability, alleviates household credit constraints, thus stimulating household consumption, and the incentive effect is more obvious for low-income households and rural households [29]. there are also many scholars' research shows that digital finance can make payment more convenient, thus promoting consumption upgrading. grossman and tarazi believe that the development of digital finance can promote household consumption by facilitating payments [30]. he zongyue and song xuguang found that digital finance can significantly promote the consumption of urban residents, especially the basic living consumption in the short term. two possible mechanisms are verified: digital finance facilitates payment and accelerates consumption decisions, thus promoting consumption growth; digital finance reduces household uncertainty, releases consumption demand and promotes consumption growth [31]. key and ma chao research show that: the use of digital finance expands in depth, which can effectively promote household consumption; the promotion effect of digital payment on household consumption is most obvious; digital finance plays the most important role in promoting the basic living expenses of families. the influence of digital finance on the consumption of rural households and low-income households is more obvious [32]. zhang xun et al. pointed out that digital finance has significantly increased residents' consumption, thus contributing to economic growth, and found that digital finance mainly promotes residents' consumption by improving the convenience of payment. the development of digital finance is more to enhance the income of rural residents than consumption [33]. other scholars' research has also proved the promotion effect of digital finance on consumption upgrading from different angles. yang weiming and others found that digital finance significantly improved the consumption level of residents and promoted the consumption upgrading. digital finance not only promoted the consumption level of urban and rural residents, but also promoted the consumption upgrading of urban residents [34]. huang kai nan and hao xiangru concluded that digital finance promoted the upgrading of residents' consumption from two aspects: the improvement of consumption level and the upgrading of consumption structure, and found that digital finance promoted the upgrading of residents' consumption from four aspects: reducing liquidity constraints, optimizing the payment environment, increasing property income and enhancing residents' risk management ability [35]. li jing found that the consumption of rural residents is more affected by digital finance than that of urban residents, and it is mainly manifested in the five categories of food, gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 193 clothing, medical care, transportation and education. the development of digital finance is beneficial to the optimization of the consumption structure in urban and rural areas [36]. the research of yi er et al. shows that the consumption structure of chinese residents shows a trend of accelerating adjustment and upgrading. digital finance plays an important role in this process, and the adjustment of consumption structure in urban areas by digital finance is more obvious [37]. xu zhaofeng found that china's residents' consumption upgrading was significantly driven by digital finance, and the driving effect of digital finance coverage was more obvious, and the effect of digital finance combined with opening to the outside world and technological innovation to drive consumption upgrading was more obvious [38]. 3. research on consumption upgrading and high-quality economic development many scholars have studied consumption upgrading and high-quality development in different ways, which shows that consumption and consumption upgrading can promote high-quality development of economy. wang hongliang and shen yan take the consumption difference between urban and rural residents as the breakthrough point, introduce the relevant theory of moderate consumption, measure the moderate consumption coefficient and its impact on economic growth in china, and obtain the inverse u-shaped correlation between the moderate consumption coefficient and the longterm economic growth trend [39]. ren yong et al. pointed out that consumption is an important kinetic energy to promote high-quality economic development. the green transformation of china's economic dimension developed better, but the social dimension lagged behind [40]. ben lu xin and liu yanqi pointed out that the contribution rate of consumption to gdp is increasing, and the important role of service consumption is also increasingly prominent. however, the service consumption in china has not been fully released at present, so it is necessary to fully stimulate the vitality of service consumption and promote the high-quality development of china's economy [41]. chen yanbin and wang zhaorui proposed that consumption should be expanded to increase the proportion of household consumption in gdp, so as to better promote high-quality economic development [42]. li tao and qiao bin use differential gmm and systematic gmm model to transform the environmental cost of economic development, energy conservation and emission reduction variables into high-quality development. empirical results show that consumption has a positive relationship with high-quality development, and the eastern part is better than the central and western parts [43]. liu jianguo and zhong xianpeng study the impact of healthy consumption on high-quality economic development from the perspective of aging. the results show that the effect of healthy consumption on high-quality economic development is nonlinear threshold. after crossing the second threshold, healthy consumption on aging significantly promotes high-quality economic development [44]. chen chong and wu weicong constructed an economic quality evaluation system based on the characteristics of high-quality development stages, and conducted an empirical study by using provincial panel data. it was concluded that upgrading the consumption structure can significantly promote high-quality development [45]. the empirical study on the provincial panel data constructed by xue junmin and jin mei shows that the upgrading of consumption structure can effectively promote the high-quality economic development, and mainly through the way of promoting the rationalization of industrial structure [46]. luo zhongqing and liu kunxin pointed out that the level gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 194 of china's high-quality economic development has increased year by year, and the optimization of consumption structure has played a significant role in promoting the high-quality development of the regional economy [47]. cui gengrui uses the panel data, uses the system gmm and the intermediary effect model to carry on the empirical research, obtains the consumption promotion can significantly enhance china economy high quality development, its transmission mechanism mainly is the consumption promotion promotes the industrial promotion thus affects the economy high quality development [48]. 4. digital finance, consumption upgrading and high-quality economic development early scholars such as levine and elhorst believe that the development of inclusive finance mainly through the alleviation of corporate financing constraints, and thus reduce the management risk of enterprises, thus promoting economic growth [49, 50]. kapoor studied from the perspective of internal innovation, and found that digital inclusive finance can effectively promote enterprise innovation and further promote economic growth [51]. ozili and others analyzed the impact of digital inclusive finance on finance itself, and believed that digital inclusive finance is the main means of financial services in the future period. with the help of mobile internet, the information medium, it can serve the widest range of customers at the lowest cost, provide convenience for remote areas and low-income groups, help to suppress the occurrence of "financial exclusion" and thus promote economic development [52]. many scholars have conducted research from different perspectives, which shows that digital finance can promote high-quality economic development. he hongqing theoretically analyzed that digital finance is an important driving force for high-quality economic development [53]. the evaluation system constructed by teng lei and ma degong obtains the high-quality development level of 30 provinces in china from 2012 to 2017, and conducts an empirical study in combination with digital inclusive finance of peking university. it is found that digital finance can indeed promote high-quality development [54]. shangguan xuming and ge binhua based on the data of 287 prefecture-level cities and above in china from 2011 to 2018, concluded that digital finance has significant direct promotion effect and positive spatial spillover effect on high-quality economic development [55]. liu wei et al. used the sys-gmm method and the intermediate effect model to conduct empirical research, and found that digital finance drives high-quality economic development by promoting technological progress [56]. 5. a brief review the impact of digital finance on consumption is mainly manifested in the aspects of increasing income, reducing uncertainty, facilitating payment, enhancing liquidity, easing liquidity constraints, easing credit constraints and insurance, etc., and mainly focuses on the aspects of facilitating payment and easing liquidity constraints. the impact of digital finance on consumption upgrading is mainly manifested in raising income, narrowing the income gap between urban and rural areas, optimizing the industrial structure, promoting the development of the tertiary industry, reducing liquidity constraints, optimizing the payment environment, increasing property income and enhancing residents' risk management ability, etc., and mainly focuses on the improvement of consumption level and upgrading of consumption structure. gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 195 consumption and consumption upgrading can effectively promote high-quality economic development. in the existing research, the research is mainly conducted from the perspectives of green consumption, moderate consumption and healthy consumption. the research on its transmission mechanism is still very rare. at present, only consumption upgrading promotes industrial upgrading, thus affecting the high-quality economic development. at present, there are few researches on digital finance, consumption upgrading and high-quality economic development. there are very few literatures that use them as titles. the literature that takes them as the subject is also very rare. few studies have explained the impact of digital finance on highquality economic development from the perspective of consumption upgrading. therefore, in the future, we can take this as the breakthrough point and make corresponding research to supplement it. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. acknowledgments: this research received some support from the risk analysis and management team of guizhou university of finance and economics. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] huang, y.p.; huang, z. china's digital finance development: present and future. economics (quarterly) 2018, 17, 04, 1489-1502. [2] cui, h.y. research on the influence of internet finance on chinese residents' consumption. exploration of economic issues 2016, 01, 162-166. [3] zhang, l.y.; tu, b. the differential impact of internet finance on the consumption of urban and rural residents in china-from the functional perspective of consumer finance. finance and trade research 2017, 28, 08, 70-83. [4] shang, h.wei. the influencing mechanism of financial support for consumption upgrading-based on the comparison of formal finance, informal finance and internet finance. business economics research, 2019, 12, 157-161. [5] he, q.z.; peng, m.s. the influence mechanism and empirical test of internet finance on household consumption. xue hai, 2019, 03, 146-153. [6] liu, t.t.; wu, f.x. internet finance and rural residents' consumption under the strategy of rural revitalization. fujian forum (humanities and social sciences edition) 2020, 03, 115-125. [7] qi, h.q.; ma, k.j. the influence of internet finance on the upgrading of urban household consumption structure. journal of xi 'an jiaotong university (social science edition) 2021, 41, 06, 56-67. [8] campbell, j.y.; mankiw, n. g. the response of consumption to income: a cross-country investigation. european economic review 1991, 35, 04, 723-756. [9] yi, x.j.; zhou, l. does the development of digital inclusive finance significantly affect household consumption-micro evidence from chinese families? financial research 2018, 11, 47-67. [10] guo, h.; zhang, y.; peng, y.l.; he, z.w. study on the regional differences of the influence of the development of digital finance on rural residents' consumption. agricultural technology and economy 2020, 12, 66-80. [11] tang, s.h. measurement and analysis of the influence of digital inclusive finance on household consumption. business economics research 2020, 21, 57-60. [12] xiao, y.f.; zhang, k.y. the influence of digital inclusive finance on the consumption level of urban and rural residents-based on provincial panel data. wuhan finance 2020, 11, 61-68. gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 196 [13] jiang, z.y. the influence of digital inclusive finance on residents' consumption-an empirical study from provincial panel data. business economics research 2020, 10, 56-59. [14] zou, x.y.; wang, w. research on the influence of digital inclusive finance on household consumption-an empirical analysis based on spatial econometric model. financial economics research 2020, 35, 04, 133-145. [15] nan, y.q.; song, m.y; xiao, h.r. digital inclusive finance and the release of consumption potential of urban residents. contemporary economic research 2020, 05, 102-112. [16] wang, m.y. the influence of digital inclusive finance on household consumption from the perspective of income stratification. business economics research 2021, 07, 167-170. [17] chen, x.x. the influence effect of digital inclusive finance to support residents' consumption upgradingan empirical test based on the perspective of income channels. business economics research 2020, 18, 45-48. [18] jiang, h.l.; jiang, p.c. study on the effect of improving the consumption level and structural optimization of digital inclusive finance. modern finance (journal of tianjin university of finance and economics) 2020, 40, 18-32. [19] yan, j.j.; feng, j.y. research on the influence of digital inclusive finance on residents' consumption upgrading. consumer economy 2021, 37, 02, 79-88. [20] tang, y.; lu, t.s.; hou, j.y. digital inclusive finance and rural residents' consumption upgrading. wuhan finance 2021, 07, 18-26. [21] lu, c.m.; wang, h.y. digital finance, income gap and household consumption-an empirical study based on 280 prefecture-level cities in china. finance and economy 2021, 07, 22-30. [22] huang, k.n.; hao, x.r. analysis of the influence of digital finance development on household consumption of urban and rural residents in china-micro evidence from chinese families. journal of social sciences 2021, 04, 110-121+215. [23] li, j.; wu, y.; xiao, j. the impact of digital finance on household consumption：evidence from china. economic modelling 2020, 86, 317~326. [24] jappelli, t.; pagano, m. consumption and capital market imperfections: an international comparison. the american economic review 1989, 79, 05, 1088-1105 [25] gross, d.b.; souleles, n.s. do liquidity constraints and interest rate smatter for consumer behavior? evidence from credit card data. quarterly journal of economics 2002, 117, 01, 149-185 [26] levchenko, a.a. financial liberalization and consumption volatility in developing countries. imf staff papers 2005, 52, 02, 237-259. [27] karlan, d.; zinman, j. expanding credit access: using randomized supply decisions to estimate the impacts， review of financial studies 2010, 23, 01, 433-464. [28] zhao, w. does health insurance promote people’s consumption? new evidence from china, china economic review 2019, 53, 65-86. [29] xie, j.z.; wu, j.r. digital finance, credit constraints and household consumption. journal of central south university (social science edition) 2020, 26, 02, 9-20. [30] grossman, j.; tarazi, m. serving smallholder farmers: recent developments in digital finance. cgap focus note 2014, 94. [31] he, z.y.; song, x.g. how does the development of digital finance affect residents' consumption? finance and trade economics 2020, 41, 08, 65-79. [32] guan, j.; ma. c. digital financial development and heterogeneity of household consumption-empirical evidence from charls. research on financial economics 2020, 35, 06, 127-142. [33] zhang, x.; yang, d.; wang, c.; wan, g.h. digital finance development and resident consumption growth: theory and chinese practice. management world 2020, 36, 11, 48-63. [34] yang, w.m.; su, l.; sun, r.; yuan, w.p. does digital finance promote consumption upgrading? —— evidence based on panel data. international finance research 2021, 04, 13-22. [35] huang, k.n.; hao, x.r. does digital finance promote the upgrading of residents' consumption? shandong social sciences 2021, 01, 117-125. [36] li, j. research on the synergy between digital finance development and urban and rural consumption structure in china. business economics research 2021, 07, 171-175. [37] yi, e.; ma, j.; chen, x. discussion on the relationship among digital financial support, e-commerce embedding and household consumption upgrading. business economics research 2021, 04, 94-97. [38] xu, z.f. digital financial driving effect of china's consumption upgrading under the background of new development pattern. business economics research 2021, 03, 168-171. gang feng and mu zhang / journal of risk analysis and crisis response, 2021, 11(4), 189-197 doi: https://doi.org/10.54560/jracr.v11i4.312 197 [39] wang, h.l.; shen, y. research on the pulling effect of consumption on china's high-quality development stage. price theory and practice 2019, 03, 33-36. [40] ren, y.; luo, m.s.; fan, b.; zhang, j.y.; yu, h.; chen, g. the role of green consumption in promoting highquality development. china environmental management 2020, 12, 01, 24-30. [41] ben, l.x.; liu, y.q. exploring the internal mechanism and path of service consumption and high-quality development of chinese economy. nanjing social sciences 2020, 11, 16-23+48. [42] chen, y.b.; wang, z.r. improving household consumption and promoting high-quality development of china's economy. humanities journal 2020, 07, 97-103. [43] li, t.; qiao, b. gmm analysis of china's high-quality economic development and consumption-based on the perspective of regional energy conservation and emission reduction efficiency. business economics research 2020, 21, 185-188 [44] liu, j.g.; zhong, x.p. the influence of healthy consumption on high-quality economic development from the perspective of aging. statistics and decision 2021, 37, 05, 65-70. [45] chen, c.; wu, w.c. consumption structure upgrading and high-quality economic development: driving mechanism and empirical test. shanghai economic research 2019, 06, 59-71. [46] xue, j.m.; jin, m. residents' consumption upgrading and high-quality economic development-an empirical study based on china's provincial panel data. business economics research 2019, 22, 42-46. [47] luo, z.q.; liu, k.x. analysis of the impact of changes in consumption structure on high-quality economic development-from the perspective of "five development concepts". business economics research 2021, 02, 186-189. [48] cui, g.r. consumption upgrading, industrial upgrading and high-quality economic development. statistics and decision 2021, 37, 15, 114-118. [49] levine, r. finance and growth: theory and evidence. in hand book of economic growth, 2005; volume 1, pp. 865－934． [50] elhorst, p.; gianfranco, p.; arbia, g. growth and convergence in a multiregional model with space time dynamics. geographical analysis 2010, 42, 03, 338-355. [51] kapoor, a. financial inclusion and the future of the indian economy. futures 2014, 56, 01, 35-42． [52] ozili, p.k. impact of digital finance on financial inclusion and stability. borsa istanbul review 2018, 18, 04, 329-340． [53] he, h.q. digital finance: an important driver of high-quality economic development. journal of xi'an university of finance and economics 2019, 32, 02, 45-51. [54] teng, l. ma, d.g. can digital finance promote high-quality development? statistical research 2020, 37, 11, 80-92. [55] shangguan, x.m.; ge, b.h. digital finance, environmental regulation and high-quality economic development. modern finance (journal of tianjin university of finance and economics) 2021, 41, 10, 84-98. [56] liu, w.; dai, b.q.; liu, w.z. can digital finance drive high-quality economic development? —— empirical analysis based on china's provincial panel data from 2011 to 2017. comparison of economic and social systems 2021, 06, 63-75. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 12, issue 2-5 journal of risk analysis and crisis response, 2022, 12(2), 106-109 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i2.329 106 communication 2022 greater china region forum on corporate social responsibility and social business day & academic annual meeting of sichuan society for risk science and emergency management was held online chengyi pu 1,2,3,*, zihan ma 1 and qi luo 4,5 1 school of insurance, central university of finance and economics, beijing (100081), china 2 college of finance and economics, tibet university, lhasa (850014), tibet, china 3 sichuan society for risk science and emergency management, chengdu (610041), sichuan, china 4 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 5 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: pucy2011@126.com received: june 29, 2022; accepted: july 7, 2022; published: july 10, 2022 on june 29, 2022, it was co-sponsored by the sichuan society for risk science and emergency management, the key research base of humanities and social sciences of the ministry of education, china institute for actuarial science, the school of finance and economics of tibet university, the yunus center of zhengzhou university, harvard business review, and the beijing center of the university of hong kong. the "2022 greater china region forum on corporate social responsibility and social business day & academic annual meeting of sichuan society for risk science and emergency management" jointly organized by the college of environment and civil engineering of chengdu university of technology, the faculty of geosciences and environmental engineering of southwest jiaotong university and the school of emergency management of xihua university was successfully held online. figure 1. conference background board. chengyi pu, zihan ma and qi luo / journal of risk analysis and crisis response, 2022, 12(2), 106-109 doi: https://doi.org/10.54560/jracr.v12i2.329 107 the guests attending this academic conference include mr. muhammad yunus, the nobel peace prize winner, as well as more than thirty experts and scholars from the chinese academy of social sciences, central university of finance and economics, the university of hong kong, zhengzhou university, renmin university of china, tibet university, and other universities. the conference was attended by industry elites from caixin media, dream rating group, my bank, and other institutions. professor chengyi pu, president of sichuan risk science and emergency management research association, and professor han huang, grameen china publicity ambassador, co-chaired this academic conference. mr. muhammad yunus delivered a keynote address at the conference. he pointed out that the current covid-19 epidemic and global warming problems have brought great challenges to the world, and the gap between the rich and the poor is widening. he encouraged more enterprises to assume social responsibilities and jointly build a "three-zero world" with zero poverty, zero unemployment and zero net carbon emissions. figure 2. mr. muhammad yunus delivered a keynote address. after the keynote address, the annual meeting carried out a main academic forum and three parallel academic forums, with the theme of consolidating the scientific foundation of risk, paying attention to safety emergency management, building a resilient economic society, improving the quality of development, and achieving common prosperity. figure 3. the opening ceremony of these academic forums. chengyi pu, zihan ma and qi luo / journal of risk analysis and crisis response, 2022, 12(2), 106-109 doi: https://doi.org/10.54560/jracr.v12i2.329 108 first, the experts focused on the topic of corporate social responsibility at the main academic forum. they delivered speeches with the following topics: "social entrepreneurship is the development direction of excellent enterprises ", "successful cases of social enterprises in taiwan", " evolving risk, strengthening resilience, responding to economic and social concerns : the risk management function of insurance ", "exploring the catalyst of social innovation with hong kong experience", "practice of decentralized e-commerce platform of idol group", "the practice and thinking of grameen microfinance in china under the process of comprehensive rural revitalization and common prosperity", "social enterprises, institutional capital and institutional change", and "social capital is the main driving force for human success based on the understanding of grameen's local practice in china". they discussed the necessity and application mode of corporate social responsibility. and they believed that through corporate social responsibility, we can build a more harmonious social relationship among community residents. as a result, we can solve social problems, improve the quality of economic development, and achieve the common prosperity. secondly, scholars discussed specific methods to improve risk management capabilities and economic development quality at the parallel forums a and b. they discussed disaster prediction and prevention issues and delivered speeches with the following topics: "introduction to the spatiotemporal model method of mountain status shift prediction", " introduction to the inflection point deadline structure of coronavirus emergency risk management", "basis and practice of comprehensive disaster reduction work ", "prediction for natural peak years and peaks of carbon emissions in southwest china", "research on disaster risk assessment and prevention of earthquake landslides in jiuzhaigou national geopark", "prevention and emergency handling of underground engineering geological disasters", and "comprehensive risk analysis of flood earthquakes in rural households based on multidimensional information diffusion model". these researches provided some new ideas for disaster prevention and have active application value. figure 4. several experts attended the parallel meeting. what’s more, these scholars also discussed how to deal with economic and social crises at the parallel forums a and b. they delivered speeches with the following topics: "a review and analysis of inflation: will the current high inflation destroy the u.s. and world economies?", "blockchain conjecture, how will the new financial system be constructed?", "financial deepening, new urbanization and high-quality economic development: empirical analysis based on spatial dubin model", "big data industry, regional technological innovation efficiency and fintech development empirical analysis based on spatial dubin model and intermediary effect", "the impact of r&d investment subsidy on the innovation output of agriculture-related enterprises" and "new chengyi pu, zihan ma and qi luo / journal of risk analysis and crisis response, 2022, 12(2), 106-109 doi: https://doi.org/10.54560/jracr.v12i2.329 109 securities law and audit risk control". they expressed their views on frontier problem in the economic field and made some useful suggestions for the development of the economy. at last, 14 graduate students shared their researches at the parallel forums c. professor xianzhi yuan, chief academic expert of the sichuan risk science and emergency management research association, was invited to make some comments. through the preliminary essay collection activities and discussions in the parallel forums, more young students were encouraged to participate in the research of risk science and emergency management. they can work together to promote economic and social development. with the joint efforts of the organizers and participating experts, this academic conference was held online successfully. it provided a platform for experts and scholars at home and abroad to discuss their latest academic achievements. the views they expressed at the conference can help solve the problem of global warming, disaster response, poverty relief, economic development. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). international journal of computational intelligence systems vol. xx(z); month (year), pp. xx–yy doi: 10.1080/xxxxxxxxxxxxxx; issn xxxx–xxxx online https://www.atlantis-press.com/journals/ijndc journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 156–161 doi: https://doi.org/10.2991/jracr.k.191024.005; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article dynamic comprehensive evaluation of ecological environment of 12 provinces and cities in western china hongmei zhang1,2,*, li teng1, chen yazhong1 1college of big data application and economics, guizhou university of finance and economics, guiyang 550025, china 2guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang 550025, china 1. introduction1 in recent years, china’s economy has developed rapidly, and the economic level of the western region has also been greatly improved. the western ecological environment is very fragile, so in the process of development it is necessary to evaluate the development level of existing studies in the west and to find the problems in the development, find a way to develop more efficient and reasonable. at present, both domestic and international scholars have made some achievements in the ecological environment. matthew and luck [1], together with the model of the human ecological footprint model and the ecosystem process model, to get the most important factors that affect the development of the ecosystem, to apply the improved ecological footprint method to the evaluation of the city ecosystem, and to evaluate the ecosystem of 20 major cities in the united states. li yu-ping (2007) models the naturaleconomicsocial concept framework. the index system is constructed from three aspects: land natural ecological security, land social ecological security and land economic ecological security. zhang [2], in the study of ecological environmental assessment of the western region, and the concrete reality of the western provinces, has been building the western eco-environmental evaluation index system on the basis of the environmental support system of the chinese academy for sustainable development. chen yuan-yuan (2010) constructed the evaluation index system from the three aspects of land ecosystem pressure, land ecosystem state and land ecosystem response, fully considering many factors such as nature, economy and society. however, the evaluation accuracy should be affected by the index representativeness and rationality of its weight. dong xiao-xiao (2014) in the western ecological fragile zone land ecological condition evaluation and prediction, the article, with the western ecological fragile areas yuzhong county in gansu province as the research area, using gis to study the change of land use, topography, vegetation, landscape pattern and soil erosion are analyzed, and reflects the land ecological risk condition under the influence of single factor. in this paper, a set of appropriate indicators for the evaluation of ecological environment is established, and various methods are used to conduct dynamic comprehensive evaluation of the ecological fragile areas in the west. according to the research results, 12 provinces and cities in the west are comprehensively ranked and corresponding policy recommendations are put forward. wang et al. [3] divided the ecological environment carrying capacity index system of shandong peninsula into four levels: natural environment, ecological environment, population environment and pollution environment. each level i index selected its typical level ii index and constructed 17 evaluation indexes as its evaluation system. the weight of each evaluation index was determined by the analytic hierarchy process. very et al., with the method of coupling analysis based on the analysis of the new quality of urbanization and ecological environment bearing capacity of main indexes, constructing the new urbanization a r t i c l e i n f o article history received 07 january 2019 accepted 15 may 2019 keywords western ecologically fragile area ecological environment dynamic comprehensive evaluation a b s t r a c t the problem of ecological environment is a global problem. it is a basic material condition for human beings to survive and engage in various production practices, and has a tremendous impact on our economic and social life and development. in recent years, the overall situation of the ecological environment in western china has been severe. on the one hand, it is because of the inherent fragility of the ecological environment in the western region. on the other hand, it is due to human, economic, and political and legal factors. therefore, the western region is faced with the dual task of protecting the environment and developing the economy. in the face of these dual tasks, the western region must choose a development model based on the coordinated development of the environment and economy. therefore, this paper selects data from 12 provinces and cities in the western part of china for 6 years, adopts multiple evaluation methods to conduct a dynamic comprehensive assessment of the western ecological environment, and provides some corresponding countermeasures for the coordinated development of the environment and economy in the western ecologically fragile region. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: zhm1035@qq.com 1introduction: zhang hongmei, professor and tutor of postgraduate program, research direction: finance and regional development and finance in poverty area. email: zhm1035@qq.com. li teng, chen yazhong, male, master’s degree, grade 2014 graduate students. https://doi.org/10.2991/jracr.k.191024.005 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:zhm1035%40qq.com?subject= mailto:zhm1035%40qq.com?subject= h. zhang et al. / journal of risk analysis and crisis response 9(3) 156–161 157 quality and ecology environmental bearing capacity of the coupling coordination model, the calculation and the districts and counties in chongqing from 2000 to 2012 new urbanization quality and ecology environmental bearing capacity of the coupling, evaluating the quality of new urbanization and ecological environment in chongqing coupling stages and levels. zhou et al. [5] constructed a coupling and coordinated evaluation system of regional economy, ecological environment and tourism industry. taking 11 provinces and cities along the yangtze river economic belt as an example, the weighted topsis method was used to evaluate the comprehensive development level of the three major systems of each province and city in this region. secondly, based on the coupling coordination model, this paper analyzes the coupling coordination evolution of the three systems in the yangtze river economic belt from the perspective of space-time. finally, the grey gm model is used to predict the future coupling coordination degree of the three systems. 2. introduction of dynamic model for various methods in view of that dynamic and comprehensive evaluation of the ecological environment in western china, three methods are used to combine namely, the gradation method, ideal point method and entropy value method. we apply these three evaluation methods organically to the evaluation of the ecological environment in the west, which can well reflect that we view and deal with the changing things from the perspective of movement, development and objectivity. 2.1. basic principles of the gradation method the gradation method is a new method to determine the weight coefficient according to the particularity of the comprehensive evaluation of the time-series three-dimensional data table. it is mostly used in economic management, cadre assessment and other multi-index decision-making problems. suppose there are m evaluation objects such as g1, g2, …, gm and there are n evaluation indexes such as x1, x2, …, xn. as time goes by t1, t2, …, tt. the original sequence of {xij(tk)}, we form a sequence of time-sequenced data tables. definition: the comprehensive evaluation problem of the sequential solid data table support, called dynamic comprehensive evaluation problem, can be represented as: y t f w t w t x t x ti k k m k k n k( ) ( ), ..., ( ); ( ), ..., ( )= { }1 1 where, yi(tk) represents the comprehensive evaluation value of the evaluation object of gi at tk, f represents the analytic formula of function, and wi(tk) represents the weight coefficient of the evaluation index at time. the gradation method is to take the comprehensive evaluation function y w x w x w xi i i n in= + + +1 1 2 2 ... the value of wi to be determined is to make the evaluation object such as g1, g2, …, gm, the difference between them is as large as possible, and the difference between evaluation objects can be expressed as s 2 1 2= -å m iy y( ) . to standardize the raw data, we can get function, y y mi= =∑ / 0 , at this point we have s 2 2= = = =å y y y wx xw w hwi t t t( ) , where, y y y y w w w w h x x x x x m m t=           =           = = 1 2 1 2 11 ... , ... , ... 11 1 n m mm k x x t... ... ... ...         when ||w|| = 1 is restricted, that is, when w wn1 2 2 1+ + =� is taken as the eigenvector corresponding to the maximum eigenvalue lmax(h) of w symmetric array h, then s 2 is taken as the maximum value. at this time, max{s 2}lmax(h) applies the grading method to the cross-section data of {xij(tk)}, at tk, respectively, to obtain the weight coefficient such as wi corresponding to the index such as xi, and then substitute in the comprehensive evaluation function, yi = w1xi1 + w2xi2 + ... + wnxin, to obtain the comprehensive score. 2.2. basic principle of entropy value method after getting evaluation object of gi in the moments after the comprehensive evaluation value of yi(tk), because the importance of different times of tk, to get the final evaluation result is also different, so to get reasonable evaluation result, science is the key to determine the weight vector time named as n = (n1, n2, …, nr) t. time weight vector, is to show the degree of attention to different times, according to different criteria, the application of different subjective or objective weighting method to determine. the entropy method [10] is introduced below to determine the time weight vector of n : first define the entropy as i v vr k k= -å 1 ln entropy is a term in thermodynamics. in information theory, it is also called the average amount of information, that is, it is a measure of information. at the same time, the entropy of time weight vector also reflects the degree of information contained in the weight of samples in the aggregation process. redefining “time scale” as l = å 1 1 r k r k r v , especially when n = (1, 0, …, 0), l = 1; when n = (0, 0, …, 1), l = 0; when n = æ è ç ö ø ÷ 1 1 1 r r r , , ..., , l = 0.5 the size of “time degree” l reflects the importance attached to time sequence in the assembly process, as shown in table 1. it can be seen from table 2 that the greater the value of l is, the less the elevator attaches importance to the distance evaluation time tr compared with the recent amount data, and the greater the elevator attaches to the distance evaluation time tr compared with the longterm amount data. when l is close to 0, the relatively long-term amount data of tr at the evaluation time have little effect, which is 158 h. zhang et al. / journal of risk analysis and crisis response 9(3) 156–161 a t t g g y y y y y k m k m mk ij m k= …         = × 1 1 11 1 1 � � � � � � ( ) where yij represents the static comprehensive evaluation value of evaluation object gi at time point of tj. let’s say the growth matrix is b = (bij)m × r, and b i m j r y y y iij ij i j i j 0 1 2 1 2 1 21 1 , , , ..., ; , , ..., , , , .( ) ( ) = = −   =− − ..., ; , , ...,m j r=      1 2 then bij represents the growth and change of the comprehensive evaluation value of the evaluation object gi from time point tj − 1 to tj. in addition, in order to ensure that the increase and decrease of bij are the same as the absolute value added to the denominator here, bi1 = 0 means that the growth and change of the evaluation value of d1 in the base period is 0. the change direction of the static comprehensive evaluation value is consistent. by weighting the static comprehensive evaluation matrix a and the growth change matrix b, the required dynamic comprehensive evaluation matrix c can be obtained. c c c y bij m r ij ij ij= = + + =×( ) , ,a b a b 1 where, a and b represent the relative importance of the static evaluation value and the growth change value. when a = 0 and b = 1, only the growth change degree is considered in the evaluation process. when a = 0 and b = 0, only the static evaluation value situation is considered in the evaluation process. therefore, different values of a and b can more fully consider the content of static evaluation value and dynamic growth and change of evaluation objects. then, we use the ideal point method to construct an ideal time series and a negative ideal sequence for matrix c: c c c c c c c cr r + + + + − − − −= =1 2 1 2, , ..., , , , ..., where, c c j mk ij + = =max{ | , , ..., }1 2 c c j m k rk ij − = = =max{ | , , ..., }, , , ...,1 2 1 2 according to the algorithm of the ideal point, the dynamic comprehensive evaluation value of the evaluation object gk at the time point of tj is cij to c +, and the distance between c− and cij is respectively: d v c c j m d v c c j j kj j r j j kj j r + + − + = −       = = −   ∑ ∑ ( ) , , , ..., ( ) 1 1 2 1 1 2     = 1 2 1 2, , , ...,j m where ni represents the weight of time point of tj , then the relatively close degree between the kth evaluation object and the ideal solution is: s d d d k mk k k k = + = − + −( ) , , , ...,1 2 finally, sk is taken as the final dynamic comprehensive evaluation value of evaluation object gk, where sk î [0, 1]. if the value of sk is larger, the evaluation object gk is closer to the ideal point and away from the negative ideal point. at this time, the value of dynamic table 1 | scale parameters table of “time degree” l assignment instructions 0.1 very little attention has been paid to recent data 0.3 less attention to recent data 0.5 also pay attention to the period data used 0.7 pay more attention to recent data 0.9 take recent data very seriously 0.2, 0.4, 0.6, 0.8 corresponding to the above two adjacent judgments of the middle case table 2 | comprehensive evaluation index system of ecological fragile areas in western china indicator name (unit) resources forest coverage (%) the proportion of nature reserves in the area under jurisdiction (%) soil erosion control area (1000 ha) afforestation area (1000 ha) artificial wetland area (1000 ha) environment total wastewater discharge (10,000 tons) sulfur dioxide emissions (tons) investment in industrial pollution control (10,000 yuan) economic total output value of agriculture, forestry, animal husbandry and fishery (10,000 yuan) per capita disposable income of urban households (yuan) gdp per capita (yuan/person) social total population (10,000) household consumption level (yuan) natural population growth rate (%) per capita water resources (m3/person) mainly used in the dynamic comprehensive evaluation of the completed tense that has occurred; l close to 1 time interval evaluation time tr relatively recent value data is almost useless, this kind of j is mainly used for the future tense with predictive nature of the dynamic comprehensive evaluation of the problem; when l = 0.5, it reflects that the evaluator attaches equal importance to each period and does not favor either side. finally, the determination criterion is introduced: in order to find the time weight vector suitable for sample assembly under the condition of given “time degree”, we use mathematical language to describe the criterion by mining as much as possible the sample information and taking into account the difference information of the evaluated object in time sequence as the standard and solve the following nonlinear programming problem: max ln . . , [ , ] , , ..., −( ) = − − = = ∑ ∑ ∑ 1 1 1 1 1 0 1 1 2 r k k r k r k k s t r k r k n n l n n n ∈ rr        2.3. basic principles of the ideal point method at each point of time, the evaluation value of the static synthetic value of the object and the matrix of the time series of the evaluation is as follows: h. zhang et al. / journal of risk analysis and crisis response 9(3) 156–161 159 comprehensive evaluation is higher and the ranking is higher. on the contrary, the closer the evaluation object gk is to the negative ideal point and away from the ideal point, the lower the value of dynamic comprehensive evaluation will be and the lower the ranking will be. 3. empirical analysis 3.1. establishment of index system based on the references of wang jinye, cheng daopin, hu xintian, li ming, cao lianhai, hao shilong, chen nanxiang [8,9] and others, and combined with the relevant theories and index selection principles of regional ecological evaluation, this paper constructs a set of comprehensive evaluation index system of fragile ecological environment in western china. the system mainly includes four subsystems of resources, environment, economy and society, with a total of 15 indicators, as shown in table 2. the data collected from the 12 cities and cities of the western part of the country for the last 6 years of 2012–2017 are all from the 2012–2017 statistical yearbook. 3.2. comprehensive evaluation of ecological fragile areas in western china 3.2.1. determination of weights of each index at different times based on the method of grading the idea of determining the weight coefficient of each index at different time by using the method of grading is as follows: firstly, the original data of each index in 12 provinces and cities in western china from 2012 to 2017 are processed uniformly and dimensionally. then according to the principle of open class method, get the weight of indexes in different time, in this paper, with the aid of spss20.0 standardization of statistical software for data processing, and then through the computer to calculate the weights of each index, and to get the weight coefficient, the analysis in the 6 years of environmental development, the more influence on the environment of the indicators, to the relevant departments to improve environmental quality to provide some measures and policy recommendations. the original data used in this paper are all from the annual statistical yearbook. setting: t = 6, m = 12, n = 18. firstly, the original data of {xij(tk)} is uniformly and dimensionally indexed, and then the symmetric matrix of hk = (xk) t xk(k = 1, 2, 3, 4, 5, 6) is calculated. according to the previous theory, the maximum eigenvalue hk corresponding to lmax(k) and the corresponding eigenvector lmax(k), namely the weight coefficient vector of each index at different moments, are calculated. therefore, the weight of each indicator at different times is shown in table 3. as can be seen from table 3, the important trends of these 15 indicators can be divided into four categories according to their importance: the first category is the investment completed in industrial governance, sulfur dioxide emissions, the total output value of agriculture, forestry, animal husbandry and fishery, and the total population. the second category is the natural growth rate of population, afforestation area, forest coverage rate, the proportion of nature reserves in the area under jurisdiction, and the importance of soil erosion area. the third category is the total amount of wastewater discharged, the amount of water resources per capita, and the per capita disposable income of urban households. the fourth category is the consumption level of residents, per capita gdp, artificial wetland area. in this way, it can be seen that industrial pollution and its treatment, as well as ecological factors such as afforestation area and population changes have the greatest impact on the comprehensive development of the western region. 3.2.2. comprehensive scores of 12 provinces and cities in western china in order to highlight the role of various indicators at different times, this paper adopts the linear weighting method, namely: y t w t x t w t x t w t x ti k k i k k i k n k in k( ) ( ) ( ) ( ) ( ) ( ) ( )= + + +1 1 2 2 � where yi(tk) represents the comprehensive score value of the ith evaluation object at time slot of tk, wi(tk) represents xi, and the weight of the index at time slot tk. the comprehensive scores of 12 provinces and cities in western china in different years were obtained, as shown in table 4. table 3 | weight coefficients of indicators at different times 指标 2012 2013 2014 2015 2016 2017 x1 0.2578 0.2585 0.2696 0.2407 0.2505 0.2185 x2 0.2494 0.2636 0.2651 0.2516 0.2694 0.2616 x3 0.2466 0.2648 0.2683 0.2770 0.2851 0.2849 x4 0.2703 0.2678 0.2808 0.2641 0.2439 0.2701 x5 0.1014 0.1089 0.1032 0.0868 0.0941 0.1273 x6 0.2363 0.2282 0.2299 0.2176 0.2320 0.2124 x7 0.3268 0.3228 0.3178 0.3072 0.3053 0.3132 x8 0.3467 0.2826 0.2275 0.3072 0.2561 0.2887 x9 0.3012 0.2923 0.2888 0.2693 0.2740 0.2650 x10 0.2210 0.2301 0.2489 0.2725 0.2835 0.2835 x11 0.1787 0.1988 0.1993 0.2108 0.2059 0.2270 x12 0.3100 0.2935 0.2963 0.2704 0.2753 0.2532 x13 0.1853 0.2405 0.2561 0.2676 0.2601 0.2688 x14 0.2966 0.2995 0.3012 0.3061 0.3077 0.2971 x15 0.2333 0.2495 0.2465 0.2428 0.2580 0.2455 table 4 | comprehensive score value y1(tk) of 12 provinces and cities in western china from 2008 to 2013 省市 2012 2013 2014 2015 2016 2017 新疆 0.6399 0.4272 0.5860 0.5002 0.3806 0.2280 甘肃 0.9790 0.9217 1.0434 1.0837 0.7820 1.4441 青海 2.2025 1.9871 1.9800 1.9063 1.9298 1.8497 内蒙古 2.4996 2.5714 2.3273 3.0918 2.6406 3.4214 陕西 0.2462 0.9302 1.1598 1.0565 1.1213 0.7548 宁夏 1.4908 1.5750 1.5948 1.5702 1.5076 1.5060 四川 2.7635 2.1274 2.0897 2.0683 1.7431 1.5521 贵州 0.7284 0.9201 0.9652 0.7891 0.8588 0.5017 云南 0.2890 0.3893 0.6080 0.9529 1.0651 1.0101 重庆 0.1221 0.1641 0.0227 0.1418 0.1203 0.0096 广西 1.7542 1.6045 1.5333 0.7069 0.7875 0.6979 西藏 1.6339 1.6277 1.5715 1.8852 1.7785 1.8972 160 h. zhang et al. / journal of risk analysis and crisis response 9(3) 156–161 3.2.3. entropy value method to determine the time weight vector in determining the time weight by using the entropy value method, we will determine the value of “time” l in advance by the opinion of the relevant experts, where we use the linear programming model: max( ln ) . . , [ , ] , , ..., − = − − = = ∑ ∑ ∑ 1 1 1 1 1 0 1 1 2 r k k r k r k k s t r k r k n n l n n n ∈ 66        using lingo solution above linear programming model, the time weight vector nk = (0.0029, 0.0086, 0.0255, 0.0755, 0.2238, 0.6637) is obtained. 3.2.4. dynamic comprehensive evaluation results of ecological environment in 12 provinces and cities in western china according to the previous theory, we know that the static evaluation matrix a is composed of the comprehensive evaluation value (comprehensive score value), and the growth matrix b is obtained according to the calculation formula of the growth matrix. in this case, we consider that the value of the static evaluation is equal to the value of the increasing variation, so a = b = 0.5, plug in the formula c = (cij)m × r, cij = ayij + b bij, a + b = 1 and you get the matrix c. the matrix c will give you the perfect time sequence and the minus ideal time sequence: c+ = (1.3817, 1.8542, 1.1162, 2.6914, 1.2473, 1.8585) c− = (0.0610, 0.0473, −0.4194, 0.0839, −0.0156, −0.4553) then, the distance formula between ckj and c + and c− of the dynamic comprehensive evaluation value of the evaluation object gk at time point tj, namely: d+ = (1.8275, 0.9637, 0.9287, 0.2746, 1.5061, 1.0772, 1.0585, 1.6669, 1.2928, 1.9994, 1.5161, 0.8729) d– = (0.3367, 1.3273, 1.2488, 2.0464, 0.6724, 1.0839, 1.0948, 0.3778, 0.8504, 0.7167, 0.6784, 1.2865) from the first k degree of evaluation object and relatively close to the ideal solution formula, namely: s d d d k mk k k k = + = − + −( ) , , , ...,1 2 finally, the comprehensive ranking of 12 provinces and cities in western china is obtained, as shown in table 5. according to the analysis in tables 3 and 4, the overall ecological environment of 12 provinces and cities in western china from 2012 to 2017 is the best in inner mongolia, tibet, gansu and qinghai. the ecological environment of sichuan, ningxia, yunnan and guangxi is the second. shaanxi, chongqing, guizhou and xinjiang had the worst comprehensive evaluation. in recent years, with the country’s continuous attention to the ecological development of the west, inner mongolia’s ecological environment has achieved a great reversal. green lock throat, yellow sand oasis, this is only a microcosm of inner mongolia ecological reversal. due to the continuous economic development and urbanization construction in recent years, guizhou, which had a good ecological environment before, paid too much attention to economic development and did not pay attention to pollution control, which made it fall behind in the comprehensive ranking among the 12 western provinces. the overall comprehensive sorting is in good agreement with the actual sorting and has some reference value. 4. conclusion based on the evaluation theory of ecological environment, this paper establishes a dynamic comprehensive evaluation model of ecological environment, and conducts a comprehensive evaluation study on the ecological environment of 12 provinces and cities in western china from 2012 to 2017. the main research conclusions include the following aspects: (1) according to the relevant evaluation theory system of ecological environment and the actual ecological environment of 12 provinces and cities in western china, a set of evaluation index system suitable for the actual ecological environment of 12 provinces and cities in western china is established, which mainly includes four indexes of resources, economy, environment and society; (2) analyze the establishment of a good index system by using the method of grading apart, and obtain the weight of each index at different times. through observation and analysis, the impact of industrial pollution and its treatment, ecological factors such as afforestation area and population changes on the comprehensive development of the western region is the largest; (3) finally, entropy value method, ideal point method and other methods are combined to conduct dynamic comprehensive evaluation of the ecological environment of 12 provinces and cities in western china, and they are comprehensively ranked. according to the comprehensive score, inner mongolia, tibet, gansu and qinghai have the best ecological environment on the whole. the ecological environment of sichuan, ningxia, yunnan and guangxi is the second. shaanxi, chongqing, guizhou and table 5 | comprehensive evaluation ranking of 12 provinces and cities in western china 省 市 新疆 甘肃 青海 内蒙 陕西 宁夏 四川 贵州 云南 重庆 广西 西藏 评 价 值 0.155 6 0.579 3 0.573 5 0.881 7 0.308 6 0.501 5 0.508 4 0.184 8 0.396 8 0.263 9 0.309 2 0.595 8 排 名 12 3 4 1 9 6 5 11 7 10 8 2 h. zhang et al. / journal of risk analysis and crisis response 9(3) 156–161 161 xinjiang had the worst comprehensive evaluation. therefore, the following suggestions are put forward: follow the scientific method and make up the shortage of ecological environment. the comprehensive evaluation of the ecological environment in the 12 western provinces is uneven. we should firmly establish the concept of “environmental protection first, ecological zones”, and enforce environmental protection laws. we will strengthen law enforcement for ecological and environmental protection, regulate enterprises in pollution control, and develop green industries. we will control the size and layout of polluting enterprises, reduce waste discharge, adopt advanced technologies and technologies with low energy consumption, improve energy efficiency, and promote the use of clean energy. conflicts of interest the authors declare they have no conflicts of interest. funding this article is supported by the basic [2016] 1534-4 of guizhou province science program project. project name: study on financial support efficiency of new urbanization in guizhou. unlisted references li yu-ping (2007). chen yuan-yuan (2010). dong xiao-xiao (2014). very et al. references [1] luck ma, darrel jenerette g, wu j, grimm nb. the urban funnel model and the spatially heterogeneous ecological footprint. ecosystems 2001;4:782–96. [2] zhang h. study on comprehensive evaluation of ecological environment in western china. j southwest univ sci technol 2008;25:10–16. [3] wang k, li n, yu x, wang y, liu y. study on the construction and application of evaluation index system of ecological environment carrying capacity of shandong peninsula. chinese geol 2014;41:1018–27. [4] zhang y, yang q, min j. coupling analysis of new urbanization quality and ecological environment carrying capacity in chongqing. j geogr 2016;71:817–28. [5] zhou c, feng x, tang r. analysis and prediction of coordinated development of regional economy, ecological environment and tourism industry—taking provinces and cities along the yangtze river economic belt as an example. econ geogr 2016;36:186–93. [6] zhao g, zhao h. research on evaluation index system of urban competitiveness based on network analytic hierarchy process. sci technol progr countermeasures 2006;126–8. [7] shou h, zhang y. research on early warning of commercial bank system risk indicators based on ahp entropy method—data from 2003–2012. east china econ manage 2013;27:44–9. [8] cao lh, hao sl, chen nx. construction and evaluation of rural ecological environment index system. res soil water conserv 2010;17:238–240+244. [9] wang jy, cheng dp, hu xt, li m. evaluation index system and fuzzy evaluation of ecological environment in guangxi. j northwest forest univ 2006;5–8. [10] he y. assessment research of bijie drought risk based on cloud model. j risk anal crisis response 2013;3:192–200. https://doi.org/10.1007/s10021-001-0046-8 https://doi.org/10.1007/s10021-001-0046-8 https://doi.org/10.1007/s10021-001-0046-8 microsoft word risk management in process industry practical approach in poland.doc risk management in process industry – practical approach in poland andrzej kozak office of technical inspection 02-353 warsaw 34, szczesliwicka street, poland abstract: as one of the risk analysis methods hazop has been successively and commonly applied for many years since it enables the future user of a process installation to check the project in terms of security in a way that minimises the costs of changes. this paper refers to the weak points of risk management procedures in the process industry which may occur when applying analysis techniques during the plant life cycle and typical common errors made during the hazop analysis. a technical assessment of the industrial environment and installations is proposed as a background to rbi procedures. key words: risk-based inspection [rbi], hazard and operability study [hazop], life cycle cost [lcc]. 1. introduction the office of technical inspection (polish: udt) performs about half a million inspections of pressure equipment throughout the country per year. the recorded number of failures of this equipment has been relatively low for many years about 10 cases of technical failures per year. the purpose of this study is not to demonstrate the advantages of the various techniques of risk analysis, but to give guidance regarding the most common weak points of various techniques and the most common difficulties associated with risk management in the installation life cycle. it should also be clear that without proper actions in the earlier phases, performance of rbi type inspections during the operational phase is seriously impeded. 2. safety management in plant life cycle it is possible to take a risk and formulate a premise that approximately 75% of incidents and failures result from the lack or insufficiency of the safety culture in a specific plant in the process industry (hse data [1] relating to the gas industry). at the same time, the most favourable case is that the plant risk management is implemented over the entire life cycle e.g. "from the cradle to the grave" also life cycle cost [2]. a comprehensive approach that the office of technical inspection has successfully applied for several years in poland is presented in fig. 1. the initial phase, or the development application stage, should end with the execution and implementation of the proposals arising from the preliminary risk analysis. during the next stage, the design stage which is a comprehensive, both qualitative and quantitative, risk analysis should be performed. also an explosion protection plan, an action plan in case of failure, and a crisis management plan should be prepared. the next phase, the construction and commissioning stage, is intended to supervise the execution and practical implementation of the conclusions and recommendations from the previous stages. the last phase, the operational stage, is the time for "reaping the fruits" which have previously been worked out, including the risk-based inspection [3] [4] 2.1 development application stage. technique: pha or preliminary hazop and c-hazop. weak points: errors occurring in this phase are most often errors committed by a licensor or a contractor preparing the technical documentation based on the purchased license. thus, they are the errors “purchased” by the future operator of the plant: a. lack of the maturity of technology and/or technical documentation – technology errors on the part of a licensor. it often happens that a technology license is bought, in which the kinetics of the processes are not fully known or verified experimentally. other cases include gaps in the technical documentation such as inadequately known and not thoroughly studied corrosion processes. or, the process apparatus is not well-chosen or incorrectly calculated, neglecting a variety of parallel or side processes. journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 102-105 published by atlantis press copyright: the authors 102 fig. 1 risk management chart within the entire plant life cycle in the process industry b. lack of the implementation of a contracted technical specification – errors of the so-called basic documentation on the part of a licensor or contractor. in the majority of cases, this means omissions or errors in the application of technical standards and best engineering practices. it is a frequent case that appears or is not noticed in due time by the so-called young and dynamic contractor team. 2.2 design stage common errors. the most serious mistakes during this phase can be made by the owner of the plant. mistakes made by other parties involved in the investment project result rather from the owner's tacit acquiescence. technique: hazop weak points: 1. missing plant and/or corporation risk matrix, 2. lack of interest of the end-user's management staff, 3. formal qualitative analysis only, 4. the plant manager is not interested in the analysis or does not have enough knowledge or experience, 5. formal analysis only – very big binder/paper file but without conclusions or important data. the first and very important stage is the design stage. as concerns plant safety, a well executed design stage is like the foundations for a building. errors made at this stage are usually irreparable at the following stages. the plant management staff must clearly declare which losses are acceptable, tolerableacceptable or unacceptable. if it is not the case, the udt inspectors will use udtuniversal risk matrix. 2.2.1 case study from hazop analysis in the course of hazop analysis, it is crucial to determine which of the detected hazards may be contained: a. individually by means of actions undertaken by the operator b. by departmental chemical rescue teams and which of the detected events should be subject to further analysis, e.g. layer of protection analysis (lopa), quantitative risk analysis (qra), safety integrity level (sil) analysis, etc. fig. 2 shows real-life data obtained during hazop analysis of the process installation in the petrochemical industry conducted by the author. the installation was divided into 6 analytical nodes. attention should be focused on node 5 in which there were 8 hazardous events identified and node 1 in which there were 5 hazardous events identified that required development application stage preliminary hazard analysis (pha; preliminary hazop study & c-hazop) hazop study construction safety study eg. ped, atex directives, etc. safety management system, rbi final hazard analysis: qra/lopa sil fire safety & explosion study emergency plan & crisis management validation & independent hazard audits. design stage construction & commissioning stage operational stage p re-approval p ost-approval published by atlantis press copyright: the authors 103 the intervention of the installation operator – fig. 2. hazardous events (that simultaneously occurred in nodes 1 and 5) may transform into a top event that leads to a catastrophic situation. it should be emphasized that, the number of correct human actions undertaken under stress is limited. the detection and prevention of such hazards depends on the experience of the hazop team members. 0 1 2 3 4 5 6 7 8 9 node 1 node 2 node 3 node 4 node 5 node 6 fig. 2 number of interventions undertaken by operations in response to hazardous events detected during hazop analysis for individual nodes. 2.3 common errors construction and commissioning stage. at this stage, the major errors are primarily committed by the plant owner together with its operator. the most frequent errors include: 1. lack of experience with conformity assessment procedures, 2. not certified staff and service for emergency shut down or lock systems, 3. device not reliable enough to be applied in control or emergency shout-down (esd) loops. the design itself is usually made by an experienced and certified team but the construction is carried out by an enterprise preparing a tender under the pressure of cost cutting procedures. cost cutting is usually in conflict with safety! according to [2], the life cycle cost analysis made by third party experts could be counteractive. 2.4 common errors operational stage small improvements without a thorough analysis, particularly of the flow dynamics and the possibility of preventing an ignition reaction. 2.4.1.case study from an organic industry in poland: after the commissioning that followed a minor upgrade, the reactor piping was broken – fig. 3. why? what was the matter? a piston pump was replaced with a rotary pump. the rotary pump mixed liquids in a more turbulent way than the piston one and injected an additional amount of energy into the system which was sufficient to start a chemical reaction in the piping instead of the reactor 1-2m ahead of the reactor resistant to corrosion. fig. 3 a dn250mm tube broken by pressure a few weeks after the minor plant modernization. 2.5 too many different alarms and signals in the control room usually, a well-educated and experienced engineer can effectively identify and analyse a maximum of 5 different signals per minute during 8 hours of work. if there are more signals, the control room engineer can get tired and unfocused. and then the chance of making a mistake increases rapidly. 3. risk-based inspection the rbi method is a modern inspection method primarily applicable in the case of a large difference between the inspection time and the time required to prepare the plant for testing by traditional methods such as the water-pressure test and the plant start-up time after the inspection testing. a flow chart of an rbi process is shown in fig. 4. 3.1 a case study based on a practical inspection time value: customary inspection in a high-tech catalyst reactor plant in the petrochemical industry: 1. inspection time 16 hours published by atlantis press copyright: the authors 104 2. preparation time for inspection 3 days (72 hours) 3. plant start-up time after inspection – 4 days (96 hours) generally, customary inspection staff is not aware of what happened at the previous stages. risk-based inspection time is introduced in this plant without a production break. to perform an inspection using this method, it is necessary to prepare documents at previous stages of the plant life cycle [5]. the application of rbi requires long-time cooperation with different specialists, especially: a. maintenance staff, b. plant corrosion specialist, c. plant engineers, d. others (if needed), and is much better and cheaper when this cooperation and data collection start at the initial stage of the plant life time. additional benefits for the plant operator: the plant is ready to implement the reliability-based inspection (rbi) fig. 4 risk-based inspection flow chart (acc. to [ 4]). 4. conclusions the plant safety system management should be carried out by a multi-disciplinary team of specialists with relevant knowledge and experience of the technology and plant. errors made in the earlier phases have a tendency to accumulate and multiply losses during the following stages. in the course of a complex risk analysis, it is essential to take into consideration the possible mistakes that are likely to happen. in order to avoid them, a group of competent specialists should be formed as a working team, without economic interests to reduce expenditures on safety. the most dangerous mistakes often made by the team are hidden mistakes, especially those that remain undetected at different stages of safety analysis. if not accompanied by a deep insight into the technology of the process, even the best formal approach may lead to shortcomings in the analysis. potential sources of mistakes, especially those referred to as an example in this paper, shall be taken into account during a comprehensive analysis. a simple mistake made at an early stage can cost the end user a lot of money when trying to correct it during the operational phase, possibly after the first accident. 5. references: [1] health and safety executive report crr no. 139 [2] standard en – 60300-3-3 “dependability management – part 3-3. application guide –life cycle costing. [3] standard api rp no. 581 “risk-based inspection. base resource document”. [4] standard api no. 580 “risk-based inspection. recommended practice”. [5] andrzej kozak “reliability analysis and rbi planning for industrial vessels”. international applied reliability symposium. amsterdam 2011. symposium proceedings. bibliographical note: dr andrzej kozak is the manager of the department of functional and process safety at the office of technical inspection (udt) in warsaw, poland and a lecturer at the technical university of łódź. he graduated from the cracow university of technology, where he received a master’s degree in chemical and process engineering. he holds a ph.d. from the polish academy of sciences. he has 35 years experience in process industry and safety management. he has received the title of a certified reliability professional and he is a member of the american chemical society d at a & in fo rm at io n co lle ct io n consequence of failure probability of failure r is k ra nk in g in sp ec tio n pl an mitigation (if any)reassessment published by atlantis press copyright: the authors 105 project portfolio management strategies in housing estate development organizations in nigeria’s built environment sadiq gumi abubakar 1 , jingchun feng 1 , salisu gidado dalibi 1,2,* , xinpeng li 1 , chuanbin zheng 1 ,liang cao 1 1 business school, hohai university, nanjing city, jiangsu province of china. 2 department of quantity surveying, abubakartafawabalewauniversity, bauchi, nigeria. *salgidos@yahoo.com [corresponding author] received march 12,2018 accepted june 19, 2018 abstract nigeria is an african country with abundant natural and human resources with an aspiration to be among the top 20 economies in world by the year 2020. such objective can only be achieved through requisite infrastructural development projects within the nigeria’s built environment such as road, power plants and adequate housing estate development projects etc. however, nigeria’s housing estate developments were insufficient in terms of demands; do not reflect the desired housing needs of the end-users; in some-cases affordable but not qualitative; do not possess green building features. these has led to challenges and questions about project portfolio management (ppm) strategies that align and fit portfolios with the organizational objectives while also delivering estate development projects that are less risky and marketable. the aim of this research paper is to assess the ppm strategies used by housing estate development organizations in nigeria’s built environment with a view of identifying, examining, and highlighting the impact of these strategies on housing estate development project portfolios in nigeria. literature reviewed led to identification and assessment of five ppm strategies. krejcie and morgan table was used to determine the sample size, while cronbach’s alpha, mean item scores, relative importance index, t-test statistics were used for data analyses. the results show that any chosen ppm strategy(ies) by housing estate development organizations will not be easy to apply, use, implement nor will it be excellent in the strive to achieve the organizational objectives in any given portfolio. key words: built environment, factors, housing estates, nigeria, organizations, portfolio, projects, strategies 1. introduction 1.1 background to the study nigeria is often referred to as the "giant of africa", owing to its large population and economy. with its abundant natural and human resources, it also harbors a long term aspiration to be among the top 20 economies in the world by the year 2020. (holmes, 1987; cia, 2014; wikipedia.org/wiki/nigeria; library of congress, 2008; bloomberg.com 2014 and adb, 2013). these lofty objectives can only be realized through the requisite infrastructural development projects such as roads and rail networks, bridges, power generation projects and adequate housing estate development projects etc., within nigeria’s built environment. these will make the built environment economically and investment viable in addition to the aforementioned resources. according to olatunji et al. (2016), the construction industry is vital for the development of any nation. in many ways, the pace of the economic growth of any nation can be measured by the development of physical infrastructures, such as buildings, roads and bridges etc. the pmi (2017), stated that projects enable business value creation. these business value in projects refers to the benefit that the results of a specific project provide to its stakeholders that may be tangible, intangible, or both (pmi, 2017). moreover, projects are known to be the engine and catalyst for developments while adequate financing and funding are simply the fuel that makes the engine (projects) work. the public organizations (government), the private organizations (investors) or a partnership of both known generally as the clients normally initiate such projects. in some cases, many projects will be ongoing simultaneously and each has its own budget and duration while some may be similar, others are entirely different; all are meant to serve a business and or some specific organization’s objectives. a collection of projects is called a program and largely a portfolio (sadiq et al., 2018). while a project is a temporary endeavor undertaken to create a unique product, service, or result; a program is defined as a group of related projects, subsidiary 63 copyright © 2018, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ programs, and program activities managed in a coordinated manner to obtain benefits not available from managing them individually; whereas a portfolio is defined as projects, programs, subsidiary portfolios, and operations managed as a group to achieve strategic objectives. some organizations may employ the use of a project portfolio to effectively manage multiple programs and projects that are underway at any given time. portfolio management (pfm) can be referred to as: combination of projects under the sponsorship of a particular organization sharing scarce resources (archer and ghasemzadeh,1999; jonas et al., 2012); a set of business practices that integrates projects with other business operations (levine, 2005; archer and ghasemzadeh, 2004; dammer and gemunden, 2006); a dynamic decision making process whereby new projects are evaluated, selected, and prioritized; existing projects are accelerated, terminated, or de-prioritized; and resources are allocated and re-allocated to the active projects (cooper et al., 2000); involves projects that are selected and managed in line with strategy and that resources are allocated to projects with the optimization of the entire portfolio in mind (archer and ghasemzadeh,1999a and b; artto and dietrich, 2004; artto et al., 2004). as such, various organizations adopt different approaches and strategies when managing projects and programs within a portfolio. these may be linked to how the organizations’ portfolio of projects are evaluated, selected, and prioritized etc. housing estate development organizations in nigeria are no exceptions. 1.2 the research problem it appears that the attention project portfolio managers give to portfolio activities is inadequate and working with multiple projects overloads the employees (zika-viktorsson et al., 2006; elonen and artto, 2003). the alignment between project portfolios and customer relationship portfolios is a missing link which is implicitly reflected in the objectives of single projects because their results should satisfy a certain hierarchy of their needs and satisfaction (voss, 2012). ppm can be understood as the hub of an intra-company system that connects projects and operations (floricel and ibanescu, 2008). these requires different decision situations and different decision making approaches, which some authors asserted that combining decision-making approaches that were based on different logics might be difficult (floricel and ibanescu, 2008) and it might lead to conflicts within the organization (bessant et al., 2011). in addition, the dilemma in resource sharing is poorly understood and hardly solved in project portfolios and is just one among others. many other deviations from the companies’ ppm frameworks appear in the day-to-day practice (blichfeldt and eskerod, 2008). despite the variety of instructions on how projects should be selected to the portfolio, how resources should be allocated across projects, how to align the entire portfolio with strategy, and how to assess the success of the portfolio, companies still struggle with the resource sharing problem across projects (engwall and jerbrant, 2003) as well as constant changes in their portfolios (elonen and artto, 2003). these may be linked to the organizational choice of ppm strategies. in nigeria’s built environment, housing estate development organizations developed and delivered many housing estates alone or through partnerships with the private sector. however, such housing estate developments were insufficient in terms of demands; do not reflect the desired housing needs of the end-users; in some-cases affordable but not qualitative; do not possess green building features (dalibi et al., 2016). as such, housing estate development organizations are faced with challenges of project portfolio management (ppm) strategies that align and fit with their organizational objectives while also delivering estate development projects that are less risky and marketable. these will require a lot of expertise and effective strategies from the portfolio managers in the housing estate development organizations which is a huge challenge. 1.3 research aim and scope the aim of this research paper is to assess the ppm strategies used by housing estate development organizations in nigeria’s built environment with a view of identifying, examining, and highlighting the impact of these strategies on housing estate development project portfolios in nigeria. the study only focuses on the housing estate development project portfolios. it does not include stocks portfolios in the aforementioned organizations. 1.4 research hypothesis to address the research problem and fully achieve the research aim, the following hypotheses were formulated and tested using the appropriate statistical tool: null hypothesis ho: project portfolio management strategies used by housing estate development organizations within the built environment in nigeria are not significantly effective. alternative hypothesis ha: project portfolio management strategies used by housing estate development organizations within the built environment in nigeria are significantly effective. 2. literature review 2.1 project portfolio management (ppm) and the executing organizations a portfolio is a collection of projects or programs and other work that are grouped together to facilitate effective management of that work to meet strategic business objectives. an organization may have more than one portfolio, each addressing unique business areas or objectives. proposed initiatives become part of the portfolio when they are identified, selected, and/or approved. (pmi, 2008a and b) at any given moment, the portfolio represents a view of its selected components and reflects the strategic 64 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ goals of the organization; however, specific projects or programs within the portfolio are not necessarily interdependent or directly related. by reflecting investments made or planned by an organization, portfolio management includes the processes for identifying the organizational priorities, making investment decisions, and allocating resources. therefore, the portfolio represents the work selected to be done, but not necessarily the work that should be done. if a portfolio’s components are not aligned to its organizational strategy, the organization can reasonably question why the work is being undertaken. therefore, a portfolio is a true measure of an organization’s intent, direction and progress. portfolio management is an opportunity for a governing body to make decisions that control or influence the direction of a group of components (a sub-portfolio, program, projects, or other work) as they work to achieve specific outcomes. an organization uses the tools and techniques described in this standard to identify, select, prioritize, govern, monitor, and report the contributions of the components to, and their relative alignment with, organizational objectives. it is not concerned with managing the components. the goal of portfolio management is to ensure that the organization is ―doing the right work,‖ rather than ―doing work right.‖ (ibid) most projects landscapes are becoming more complex. in addition to effective and efficient single project management, companies require structured and proactive management of the project landscape to stay competitive (elonen and artto, 2003). ppm integrates projects with other business operations and that includes key activities such as decision making on which projects are to be given priority, which projects are to be added to or abandoned /taken out of the portfolio, and how to allocate resources (archer and ghasemzadeh, 2004; dammer and gemünden, 2006). among the key issues has been that projects are selected and managed in line with strategy and that resources are allocated to projects with the optimization of the entire portfolio in mind (archer and ghasemzadeh, 1999a, b; artto and dietrich, 2004; artto et al., 2004; englund and graham, 1999). centering resources for a single project can also benefit the entire portfolio as project execution speed may be maximized and new products can be brought to market rapidly. most of these start with the single projects which is an integral part of a portfolio of an organization (martinsuo and lehtonen, 2009). as the number of projects increases, it is particularly important to guarantee effective and efficient execution of project portfolios. this remains a challenge despite the formalization of single projects, which facilitates faster process implementation and better process quality (ahlemann et al., 2009; garcia, 2005). in parallel, companies are confronted with customers demanding value-adding activities, such as joint product development, financing, or consulting services (homburg et al., 2002). companies introduce concepts and processes such as customer relationship management (crm) or key account management for a closer customer relationship and better service, aiming to create value both for the customer and for the company (ernst et al., 2011; frow and payne, 2009; boulding et al., 2005). furthermore, customer prioritization and the management of customer portfolios have received more attention in research and practice (homburg et al., 2008; homburg et al., 2009; terho, 2009). a customer portfolio should be managed along the customers' valuable contributions to the relationship portfolio, not just the customers' valuable contributions to the firm alone (homburg et al., 2009). however, optimization of individual portfolios does not necessarily optimize the overall business performance; an alignment between the different portfolios is needed (tikkanen et al., 2007). this is where ppm strategies come into play as they provide the requisite direction on how best to align organizational objectives within the portfolios and achieve a favorable outcome for the organization. project portfolio management (ppm) must deal with the coordination and control of multiple projects. as such the project portfolio managers pursue the same strategic goals and compete for the same resources, whereby managers prioritize among projects to achieve strategic benefits (cooper et al., 1997a). companies have adopted project portfolio management frameworks, including the use of project evaluation and decision criteria, project evaluation and control routines, and other means to formalize their project portfolio management (pmi, 2008; benko and mcfarlan, 2003; cooper et al., 2001; martinsuo and poskela, 2011; muller et al., 2008; teller et al., 2012). the rational approach appears to assume that projects are obedient servants that exist primarily to fulfill the strategy of the parent organization.however, innovation projects are frequently used to purposefully question the strategy and are no longer necessarily limited to one company's strategic interests only (artto et al., 2008a, b). also, there is a rational approach which appears to assume that companies are fully aware of all possible factors both internal and external – influencing the projects. many of the previous studies delimit their attention to the projects that are well defined and whose environments are well known, even if also less welldefined projects are being found in portfolios and many portfolio environments are inherently poorly known (blichfeldtand eskerod, 2008; loch, 2000). 2.2 project portfolio management (ppm)strategies ppm execution contexts can somehow be embedded into criteria and routines that align the projects with strategy and, eventually, bring strategic benefits. yet, there is increasing evidence that portfolio managers are not necessarily well informed (blichfeldt and eskerod, 2008; elonen and artto,2003) and the criteria and routines do not solve multi-project problems as expected (engwall and jerbrant, 2003; zika-viktorsson et al., 2006). 65 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ project portfolio managers in housing estate development organizations also faced with dynamic challenges in evaluation and decision criteria, project evaluation and control routines, and other means to formalize their project portfolio management especially the strategy to adopt in the estate development portfolios. the decision to go with quantitative or qualitative housing scheme depends on a wide range of factors such as internal (within the organization) and external (beyond the organizational control) which may include but not limited to housing demand, end-users’ income / buying / purchasing power, the location, stakeholders, sociopolitical stability, availability of funds and resources etc. incorporating some features / elements especially green building elements are basically at the estate developer’s disposal in nigeria’s estate development projects and these may have a significant impact on the total development cost which in turn affects end-users / occupants in terms of rental value, sales value, envisaged savings due to such elements, future asset value of the development etc. thus, those elements if not checked with potential end-users in order to ensure they meet their housing needs, requirements and also their affordability which will also reflect on the developers’ interests in terms of market value and faster sale of the housing units. these all together outlined the need for housing estate development projects with features that can align the interests of the stakeholders in such development projects (dalibi et al., 2016). as such, these must be taken into the ppm strategy of housing estate developments to ensure profitability which also ensures the viability of the portfolio. managing portfolios requires an effective strategy that will ensure success, reduce risk and achievement of the organizational objectives. portfolio management strategies refer to the approaches that are applied for the efficient portfolio management in order to generate the highest possible returns at lowest possible risks. there are two basic approaches for portfolio management including active portfolio management strategy and passive portfolio management strategy others include patient portfolio management strategy, conservative portfolio management strategy and patient portfolio management strategy (sushant, 2018) . 2.2.1 active ppm strategy active portfolio management strategy refers to a portfolio management strategy that involves making precise investments for outperforming an investment benchmark index. the risk reduction is considered a goal of creating an investment return larger than the benchmark by considering and using large number of factors and strategies for constructing the portfolio (sushant, 2018). active portfolio management means allocation of funds based on expectations of future market developments; it is also the implementation of a dynamic investment strategy that over and under weights the predefined investment opportunities over a long-term basis, with the single objective of outperforming the predefined benchmark at a predefined time in order to add value to the portfolio. (hilsted, 2012). the success of an actively-managed portfolio is dependent on combining in-depth research, market forecasting and the experience and expertise of the portfolio manager or management team. the portfolio managers that engaged in active strategy pay close attention to market trends, shifts in the economy, changes to the political landscape and factors that may affect specific companies. it involves all efforts to take advantage of irregularities. active managers claim that these processes will boost the potential for greater returns than those listed on a particular index. (investopedia). active ppm with respect to housing estate development organizations involves: top-down approach: in this approach, managers observe the market as a whole and decide where and what type of estate development projects will perform well in the ongoing economic cycle. bottom-up: in this approach, the housing estate market conditions and its expected trends are ignored and the evaluations of the ppm organizations were based on their respective strengths and capabilities, which if positioned well will allow the organizations perform better irrespective of the prevailing market or economic conditions. the success, performances and effectiveness of an actively managed portfolio is directly dependent on the skills, expertise and research abilities of the ppm team, while also engaging the stakeholders inside and outside the organization. 2.2.2 passive ppm strategy passive portfolio management relies on the fact that markets are efficient and it is not possible to beat the market returns regularly over time and best returns are obtained from the low cost investments kept for the long term (sushant, 2018). this strategy, also involves portfolio rebalancing. this implies that there is much emphasis on asset classes with weights comparable to a benchmark contribute with the largest proportion of the total return of the portfolio (hilsted, 2012). passive ppm focuses on overall sector or asset class is based on the concept of the efficient market, which states that because all investors have access to all the necessary information it's difficult if not impossible to gain an advantage over any other investor. as new information becomes available, market prices adjust in response to reflect a security's true value. the proponents of market efficiency say ―it means that reducing investment costs is the key to improving net returns‖. (raymond jones & associates). as such, housing estate development organizations need them to acquire, disseminate and interpret all the necessary information in this regard and make decisions that fit the organizational objectives within their respective project portfolios. these will also require a lot of expert judgments within the ppm team. 66 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ 2.2.3 patient ppm strategy patient ppm strategy involves making investments in well-known type of projects / programs / operations etc. it involves and represent companies that have classic growth and those expected to generate higher earnings on a regular basis irrespective of financial conditions (sushant, 2018). the knowledge about which projects / programs / operations will perform enables ppm team of various organizations to be in a ―patient mode‖ and take longer periods before making the appropriate decisions on projects / programs / operations within portfolios or the entire portfolio to invest, reinvest or divest. 2.2.4 aggressive ppm strategy aggressive ppm strategy involves making expensive investments that provide good returns and big rewards long with carrying big risks (sushant, 2018). housing estate development organizations employ such strategy for projects that are located in areas that have high prospects for physical, economic and social development bearing in mind that the target population are high earners and big spenders. this will allow the portfolio(s) to generate rapid and faster returns on the projects portfolio investments. such strategy also requires expert judgements and the use of investment appraisal methods or techniques. atrill & mclaney (2011) outlined four main methods of investment appraisal as: accounting rate of return (arr); payback period (pp); net present value (npv) and internal rate of return (irr). the interpretation, application and decision on any of the methods will be based on the skills and expertise of the ppm team. 2.2.5 conservative ppm strategy conservative ppm is an investing strategy that seeks to preserve an investment portfolio's value by investing in lower risk securities and equities (www.investopedia.com). this is done after carefully observing the market returns, earnings growth and consistent dividend history (sushant, 2018). such ppm strategies usually seek to provide both capital appreciation and income to the investing organization. a conservative portfolio is appropriate for an investor with a low risk tolerance and a time horizon from immediate to longer than 3 years. (www.thebalance.com). conservative ppm investors do not have to grapple with the increasing volatility in the markets and is suitable where rates have remained low and demand has been high. (www.forbes.com). table 1 summarizes the ppm strategies with a brief description respectively. irrespective of the ppm strategy used by organizations, expert judgment (from professionals with decades of experience) is vital and is used by the portfolio management team to assess the inputs needed to compare the components and to apply it to any technical and management details during this process. the portfolio management team also applies expert judgment to identify relationships between components, which are under consideration. such relationships may be independent components or components coupled which include: dependencies, redundancies, partial overlap, and mutual exclusivity of components. (pmi, 2008). 3. research methodology the main sources of data were from journals, conference/seminar/workshop papers, textbooks, newspapers, magazines and the internet etc. which were used to review literature in the project portfolio management (ppm), organizations, business, housing estate and its operations related areas which helps to identify and narrow ppm strategies within the nigerian and global context. these identified ppm strategies table 1: portfolio management strategies and their brief description s/n project portfolio management strategy description 1 active portfolio management strategy divided into top-down approach and bottom-up approach; it is based on expectations of future market developments with the single objective of outperforming the predefined benchmark at a predefined time in order to add value to the portfolio. 2 passive portfolio management strategy management and financial investment strategy that requires efficient market information handling in an organization. all investors have access to all the necessary information that makes it difficult if not impossible to gain an advantage over any other investor. 3 patient portfolio management strategy decision-based on the well-knowntype of projects/programs / operations expected to generate higher earnings on a regular basis irrespective of financial conditions 4 aggressive portfolio management strategy decision-based on expensive markets with favorable returns along with carrying big risks. 5 conservative portfolio carefully observing the market returns, earnings growth and consistent dividend history 67 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ form the backbone of the questionnaire, which was distributed to various professionals in ppm housing estate development organizations in nigeria. the research design encompasses the following: krejcie and morgan(1970) table of determining sample size for any given population to determine the research sample size which fixes 384 as the sample size of a maximum number for a given population of 1, 000, 000. as such, up to 740 number of questionnaires were distributed to enable the retrieval of the required sample number. a 5-point likert scale was used in obtaining and analyzing the perceptions of various professionals working in the organizations’ ppm teams. mean item score was calculated using the formula: n xxxxx )1( + )2( + )3(+ )4( + )5( 12345 the values of the mean item scores were used to pass remarks for the strategies analyzed. excellent / extremely difficult for all values between 4.5 to 5; very good / very difficult for all values between 3.5 to 4.4; good / difficult 2.5 to 3.4; satisfactory / not difficult for all values between 1.5 to 2.4 and poor / not at all for all values between 0.5 – 1.4. relative importance indices (rii) were used to rank the perceptions. the rii for a 5-point likert scale were calculated using the formula below. n xxxxx 5 )1( + )2( + )3(+ )4( + )5( 12345 the values for the rii allows for ranking of the ppm strategies in terms of performances, ease of use implementation/applications. cronbach’s alpha was used to measure the reliability of the responses from the 5-point likert scale. in this part, all the questions structured using the 5-point likert scale were subjected to reliability test using the cronbach’s alpha. reliability pertains to the consistency of scores. the less consistency within a given measurement, the less useful the data may be in analysis. cronbach’s alpha provides a useful lower bound on reliability which will generally increase when the correlations between the items increase. as such, the coefficient measures the internal consistency of the test. its maximum value is 1, and the minimum is 0, although it can be negative. tavakol and dennick (2011), concluded that cronbach’s alpha is an important concept in the evaluation of assessments and questionnaires. it is mandatory that assessors and researchers should estimate this quantity to add validity and accuracy to the interpretation of their data. table 2 below shows the corresponding interpretation of the values for cronbach’s alpha. cronbach’s alpha computation was carried out using the formula below: alpha = n / (n 1)  (total variance sum of variance for each question) / total variance. table 2: cronbach’s alpha numerical value remark cronbach’s alpha internal consistency alpha ≥ 0.9 excellent 0.9> alpha ≥0.8 good 0.8>alpha ≥0.7 acceptable 0.7>alpha ≥0.6 questionable 0.6>alpha ≥0.5 poor 0.5>alpha unacceptable where n = k, is the number of items; summation of delta-squared-k is the sum of the k item score variances, and the delta-squared total is the variance of scores on the total measurement. a t-test was used to test the research hypothesis. the formula is shown below. ns xbar ttest 0   where xbaris the mean of a random sample from a given population, µo is the mean specified in the null hypothesis, delta is the standard deviation of a random sample from the population, and n is the size of the sample used to compute and the mean and standard deviation. microsoft excel and statistics resource pack were used for the overall analyses. 4. data presentation and analyses 4.1 research response rate the responses from the research questionnaires that were distributed electronically and manually were shown in table 3 below, which clearly shows that 740 number of questionnaires were distributed, out of which 351 number (47%) were not returned; while 389 number (53%) were returned. among the responsive 389 number of questionnaires, 211 number (54.1%) were responses from the electronically distributed questionnaires (equestionnaires) while 178 number (45.9%) were responses from the manually distributed questionnaires. as such, the response from e-questionnaires was higher in this study. 4.2 research reliability test using cronbach’s alpha a total of ten questions were asked in a table format (shown in the appendix) and analyzed. using the formula for the cronbach’s alpha in table 2 above, the reliability scores and remarks for the 5-point likert scale responses were shown in table 4 below. 68 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ these clearly indicate that there is a good consistency of scores from the likert scale by the various respondents with both the 5-points and 3-points scales having excellent remarks; 4-points and 1-point scales having good remarks while 2-points scale has the least with an acceptable remark which may be some-what good (if approximated). as such the average / overall reliability of the 5-points scale structure for the responses analyzed in this study regarding ppm strategies is 0.89 which is good and may be some-what excellent (if approximated). 4.3 ease of application, use andimplementation of ppm strategies this part shows the perceptions of various ppm professions regarding the ease of implementation/ application for the aforementioned five ppm strategies. these were structured using a five-point likert scale as shown in the table 5 below. from the table above, it can be deduced regarding organizational project portfolio management (ppm) strategies & ease of application within organizations that the active portfolio management strategy and passive portfolio management strategy are the strategies that are very difficult to apply, use and implement by organizations. whereas; patient portfolio management strategy; aggressive portfolio management strategy; conservative portfolio are strategies that are not difficult to apply, use and implement by organizations. active portfolio management strategy ranked 1 st ; this is followed by passive portfolio management strategy which is also very difficult and ranked 2 nd . whereas; patient portfolio management strategy ranked 3 rd ; aggressive portfolio management strategy ranked 4 th ; conservative portfolio ranked 5 th as the least strategy in terms difficulties to apply, use and implement by organizations. table 3: electronically and manually distributed questionnaire responses questionnaires distributed returned non returned percentage electronically distributed 400 211 189 54.1% manually distributed 340 178 162 45.9% total 740 389 351 100% table 4: alpha’s reliability test and remarks questions & their scale components n n-1 variance (v) total variance (tv) tv v cronbach's alpha score remark 5-points 10 9 3740.20 21744.70 18004.50 0.92 excellent 4-points 10 9 5159.89 21744.70 16584.81 0.85 good 3-points 10 9 1015.56 21744.70 20729.14 1.06 excellent 2-points 10 9 6194.64 21744.70 15550.06 0.79 acceptable 1-points 10 9 5634.41 21744.70 16110.29 0.82 good average = 0.89 good 69 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ 4.4 performances of project portfolio management (ppm) strategies for achieving organizational objectives in portfolios this part covers the performance of ppm strategies within organizations. the data and the analyses were shown in table 6 below. table 5: project portfolio management (ppm) strategies & ease of application within organizations s/n project portfolio management strategies & ease of application e x tr e m e ly d if fi c u lt = 5 v e r y d if fi c u lt = 4 d if fi c u lt = 3 n o t d if fi c u lt = 2 n o t a t a ll = 1 t o t a l m e a n it e m s c o r e r e m a r k r e la ti v e im p o r ta n c e in d e x ( r ii ) r a n k 1 active portfolio management strategy 181 149 45 11 3 389 4.27 very difficult 0.85 1st 2 passive portfolio management strategy 171 44 103 61 10 389 3.78 very difficult 0.76 2nd 3 patient portfolio management strategy 32 29 68 187 73 389 2.38 not difficult 0.48 3rd 4 aggressive portfolio management strategy 10 13 109 228 29 389 2.35 not difficult 0.47 4th 5 conservative portfolio management strategy 19 27 56 219 68 389 2.25 not difficult 0.45 5th table 6: performances of project portfolio management strategy in achieving organizational objectives in portfolios s/n performances of project portfolio management strategies in achieving organizational objectives in portfolios e x c e ll e n t = 5 v e r y g o o d = 4 g o o d = 3 s a ti sf a c to r y = 2 p o o r = 1 t o t a l m e a n i te m s c o r e r e m a r k r e la ti v e im p o r ta n c e i n d e x (r ii ) r a n k 1 active portfolio management strategy 57 147 85 88 12 389 3.38 good 0.68 3rd 2 passive portfolio management strategy 71 44 121 133 20 389 3.03 good 0.61 4th 3 patient portfolio management strategy 88 189 68 28 16 389 3.78 very good 0.76 2nd 4 aggressive portfolio management strategy 15 11 19 79 265 389 1.54 satisfactory 0.31 5th 5 conservative portfolio management strategy 136 196 34 12 11 389 4.12 very good 0.82 1st from the table above, it can be deduced regarding performances of project portfolio management (ppm) strategies in achieving organizational objectives in portfolios that the conservative portfolio management strategy and patient portfolio management strategy are the ppm strategies with very good performances in organizations. while passive portfolio management strategy and active portfolio management strategy have good performance. the aggressive portfolio management strategy is the strategy with the satisfactory performanceamong the project portfolio management (ppm) strategies in achieving organizational objectives within portfolios. the conservative portfolio ranked 1 st ; this is followed by patient portfolio management strategy which ranked 2 nd ; passive portfolio management strategy ranked 3 rd ; active portfolio management strategy ranked 4 th and the aggressive portfolio management strategy is the least ranked as 5 th in terms of performancesamong the ppm strategies in achieving organizational objectives in portfolios. 4.5 testing the research hypothesis the values of the mean item scores for all the questions structured using the likert scales (in table 5 and 6) were used to calculate the t-test statistics and the result is shown in table 7 below. 70 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ table 7: result for hypotheses test using t-test statistics ppm strategies m e a n s ta n d a r d d e v ia ti o n s ta n d a r d e r r o r n d f a lp h a ( le v e l o f s ig n if ic a n c e ) p v a lu e t c a l t t a b 0 .0 5 , 9 s ig n if ic a n c e ppm strategies ease of application and performance 3.09 0.92 0.29 10 9 5% 3.73e-02 10.59 1.83 yes with 9 degrees of freedom (df) and 5% level of significance, the t-test calculated (tcal = 10.59) is greater than t-test tabulated (t-tab0.05, 9 = 1.83); the significance level (alpha = 0.05) is greater than the probable value (p-value = 0.037). as such, the null hypothesis was rejected and the alternative hypothesis was accepted. these clearly indicate that the project portfolio management strategies used by housing estate development organizations within the built environment in nigeria are significantly effective. this is evident from the analyses in table 5 and 6 above which shows that four out of five of the strategies perform above the satisfactory level which indicates their effectiveness (as indicated and affirmed by the hypothesis tested above). 5. discussions, conclusions & recommendations the analyses above paved the way for the following discussions and conclusions: any chosen ppm strategy(ies) by housing estate development organizations will not be easy to apply, use and implement. as such, it will require expertise, experience, management and tracking skills from the ppm team to achieve. the performances of any chosen ppm strategy(ies) by housing estate development organization will not be excellent in achieving organizational objectives in portfolios. as such, the performance of any chosen strategy(ies) by the ppm organization can only be very good, good or satisfactory in the strive to achieve the organizational objectives in any given portfolio. these may be related to factors outside the organizational control such as the dynamic market environment. project portfolio management strategies used by housing estate development organizations within the built environment in nigeria aresignificantly effective as attested by the research hypotheses. this may be related to the assessment of the performances of the ppm strategies used in managing project portfolios in the organizations where conservative and patient portfolio management strategy were deemed very good in terms of performances while passive and active portfolio management strategy have good performance. the aggressive portfolio management strategy is the strategy with the satisfactory performanceamong the project portfolio management (ppm) strategy in achieving organizational objectives in portfolios. these clearly indicate that most housing estate development organizations are not willing to adopt and or be active in this regard which may be due to investment risks within nigeria’s built environment and the market or economic related factors. irrespective of the ppm strategy used by organizations, expert judgment (from professionals with decades of experience) is vital and is used by the portfolio management team to ensure that any chosen ppm strategy is workable in the pursuit of organizational objectives in portfolios. assess the inputs needed to compare the components and to apply it to any technical and management details during this process. this study reviewed, identified, assessed and examine five ppm strategies in housing estate development organizations in nigeria’s built environment within a scope and some limitations. as such the following recommendations were proffered for further research: an examination of the factors hindering implementation of any ppm strategy(ies) in nigeria’s built environment. an investigation of the relationship between the dynamic business environment and the performances of ppm strategies. empirical research to examine the factors shaping ppm strategies in nigeria’s built environment. a research to determine the stakeholder’s roles and performances in achieving ppm strategies in nigeria’s built environment. references peter holmes, (1987). nigeria: giant of africa. the oregon press. the cia world fact book (2014). sky horse publishing, inc. 2013. isbn 978-1-62636-073-0. url: https://en.wikipedia.org/wiki/nigeria. library of congress – federal research division (july 2008). "country profile: nigeria". nigeria becomes africa's largest economy. available at: url:https://www.bloomberg.com/news/2014-0406/nigerian-economy-overtakes-south-africa-s-onrebased-gdp.html. accessed december, 2017. african development bank group (2013). federal republic of nigeria country strategy paper, 20132017. samuel olusola, olatunji, ayodeji emmanuel, oke, douglas omoregie, aghimien, seidu, sakiru adeyemi (2016). ―effect of construction project 71 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ performance on economic development of nigeria,‖ journal of economics and sustainable development, vol. 7, no. 12, pp. issn 2222-1700 (paper) issn 2222-2855 (online) , 2016. project management institute – pmi, (2017). a guide to the project management body of knowledge (pmbok® guide) 6th edition. published by pmi, newton square, p.a. sadiq gumi abubakar, salisu gidado dalibi and yuting wang (2017). factors shaping project portfolio management in nigeria’s built environment. international journal of economics, commerce & management, uk. vol vi. issue 3. archer, n., ghasemzadeh, f., (1999a). an integrated framework for project portfolio selection. int. j. proj. manag. 17 (4), 207–216. jonas, d., kock, a., gemünden, h.g., (2012). predicting project portfolio success by measuring management quality—a longitudinal study. ieee transactions on engineering management pp (99), 1–12. http://dx.doi.org/ 10.1109/tem.2012.2200041. levine, h.a., (2005). project portfolio management—a practical guide to selecting projects, managing portfolios, and maximizing benefits. jossey-bass, san francisco, ca. archer, n.p., ghasemzadeh, f., (2004). project portfolio selection and management. in: morris, p.w.g., pinto, j.k. (eds.), the wiley guide to managing projects. john wiley & sons, new york, ny. dammer, h., gemünden, h.g., (2006). multiprojektmanagement—kritische erfolgsfaktoren zum management von projektelandschaften. technische universität berlin, berlin, germany. cooper, r. g., edgett, s. j., & kleinschmidt, e. j. (2000). new problems, new solutions: making portfolio management more effective. research-technology management, 43(2), 18–33. archer, n., ghasemzadeh, f., (1999b). project portfolio selection techniques: a review and a suggested integrated approach. in: dye, l.d., pennypacker, j.s. (eds.), project portfolio management. selecting and prioritizing projects for competitive advantage. center for business practices, usa, pp. 207–238. artto, k., dietrich, p., (2004). strategic business management through multiple projects. in: morris, p.w.g., pinto, j.k. (eds.), thewiley guide to managing projects. john wiley & sons, new york, ny, pp. 1–33. artto, k.a., dietrich, p.h., nurminen, m.i., (2004). strategy implementation by projects. in: slevin, d.p., cleland, d.i., pinto, j.k. (eds.), innovations: project management research. project management institute, newtown square (pa), pp. 103–122. zika-viktorsson, a., sundström, p., engwall, m., (2006). project overload: an exploratory study of work and management in multi-project settings. int. j. proj. manag. 24, 385–394. elonen, s., artto, k., (2003). problems in managing internal development projects in multi-project environments. international journal of project management 21 (6), 395–402. martin voss (2012). impact of customer integration on project portfolio management and its success— developing a conceptual framework. international journal of project management 30 (2012) 567–581. floricel, s., ibanescu, m., (2008). using r&d portfolio management to deal with dynamic risk. r&d management 38 (5), 452–467. bessant, j., b. von stamm, k. m. moeslein and a.o k. neyer (2011). "backing outsiders: selection strategies for discontinuous innovation."r&d management. 40(4): 345o356. stilling blichfeldt, b., eskerod, p., (2008). project portfolio management there's more in it than what management enacts. international journal of project management 26, 357–365. engwall, m., jerbrant, a., (2003). the resource allocation syndrome: the prime challenge of multiproject management? international journal of project management 21 (6), 403–409. salisu gidado dalibi, hadiza balarabe and jamilu bala maiauduga (2016): green buildings: a concept aligning the interests of stakeholders (developers / clients and end-users) in estate development projects in abuja f.c.t (federal capital territory), nigeria. sbe16 hamburg: strategies – stakeholders – success factors. international conference on sustainable built environment in hamburg, germany. 7-11th march, 2016. project management institute –pmi, (2008a). a guide to the project management body of knowledge (pmbok® guide) 4th edition. published by pmi, newton square, p.a. project management institute, 2008b. the standard for portfolio management, second edition. published by pmi, newton square, p.a. englund, r.l., graham, r.j., (1999). from experience: linking projects to strategy.j. prod. innov. manag. 16 (1), 52–64. martinsuo, m., lehtonen, p., (2009). project autonomy in complex service development networks. int. j. manag. proj. bus. 2 (2), 261–281. ahlemann, f., teuteberg, f., vogelsang, k., (2009). project management standards — diffusion and application in germany and switzerland. international journal of project management 27 (3), 292–303. garcia, s., (2005). how standards enable adoption of project management practice. ieee software 22, 22– 29. homburg, c., workman, j.p., jensen, o., (2002). a configurational perspective on key account management. journal of marketing 66 (2), 38–60. ernst, h., hoyer,w., krafft, m., krieger, k., (2011). customer relationship management and company performance—the mediating role of new product performance. journal of the academy of marketing science 39 (2), 290–306. 72 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ frow, p. & payne, a. j bus market manage (2009). customer relationship management: a strategic perspective. journal of business market management february 2009, volume 3, issue 1, pp 7–27. 3: 7. https://doi.org/10.1007/s12087-008-0035-8. boulding,w., staelin, r., ehret,m., johnston,w.j., (2005). a customer relationship roadmap: what is known, potential pitfalls, and where to go. journal of marketing 69 (4), 155–166. homburg, c., droll, m., totzek, d., (2008). customer prioritization: does it pay off, and how should it be implemented? journal of marketing 72 (5), 110–130. homburg, c., steiner, v.v., totzek, d., (2009). managing dynamics in customer portfolio. journal of marketing 73 (5), 70–89. terho, h., (2009). a measure for companies' customer portfolio management. journal of business-tobusiness marketing 16 (4), 374–411. tikkanen, h., kujala, j., artto, k., (2007). the marketing strategy of a project based firm: the four portfolios framework. industrial marketing management 36 (2), 194–205. cooper, r., edgett, s., kleinschmidt, e., (1997a). portfolio management in new product development: lessons from the leaders i. research technology management 40 (5), 16–28. benko, c., mcfarlan, f.w., (2003). connecting the dots. aligning projects with objectives in unpredictable times. harvard business school press, usa. cooper, r.g., edgett, s.j., kleinschmidt, e.j., (2001). portfolio management for new products, second edition. perseus, cambridge (ma). m. martinsuo, j. poskela (2011). use of evaluation criteria and innovation performance in the front end of innovation. j. prod. innov. manag., 28, pp. 896914. müller, r., martinsuo, m., blomquist, t., 2008. project portfolio control and portfolio management performance in different contexts. project management journal 39 (3), 28–42. teller, j., unger, b., kock, a., gemünden, h.g., (2012). formalization of project portfolio management: the moderating role of project portfolio complexity. int. j. proj. manag. 30 (5), 596–607. artto, k., kujala, j., dietrich, p., martinsuo, m., (2008a). what is project strategy? international journal of project management 26 (1), 4–12. artto, k., martinsuo, m., dietrich, p., kujala, j., (2008b). project strategy – strategy types and their contents in innovation projects. international journal of managing projects in business 1 (1), 49–70. loch, c., (2000). tailoring product development to strategy: case of a european technology manufacturer. european management journal 18 (3), 246 –258. sushant n.d. (2018). active portfolio management strategy available at http://www.portfoliomanage ment.in/active-portfolio-management-strategy.html. johan christian hilsted. active portfolio management and portfolio construction implementing an investment strategy master thesis, fall 2012 cand. merc. applied economics and finance copenhagen business school. date of submission: december 14th 2012. url: https://www.investopedia.com/ask/answers/040315/w hat-difference-between-passive-and-active-portfoliomanagement.asp. raymond james and associates (n.d.) active vs. passive portfolio management. available at: https://www.raymondjames.com/weisswealthmanage ment/pdf/active_vs_passive.pdf. atrill, p. & mclaney, e. (2011) accounting and finance for non-specialists. 7th edition. prentice hall. url:https://www.investopedia.com/terms/c/conservative investing.asp. url: https://www.thebalance.com/conservative-mutualfund-portfolio-example-2466527. url:https://www.forbes.com/sites/duncanrolph/2013/10/ 30/5-rules-of-the-road-how-to-invest-a-conservativeportfolio-in-the-current-environment/#6fc9f24244a5. robert v. krejcie and daryle w. morgan (1970). determining sample size for research activities. educational and psychological measurement. 1970, 30, 607-610. mohsen tavakol and reg dennick (2011). making sense of cronbach’s alpha. international journal of medical education. 2011; 2:53-55. issn: 2042-6372. doi: 10.5116/ijme.4dfb.8dfd. cronbach, l. j. (1951). coefficient alpha and the internal structure of tests. psychometrika, 16,297-334. 73 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 ___________________________________________________________________________________________________________ table: project portfolio management (ppm) strategies & ease of application within organizations s/n project portfolio management strategies & ease of application e x tr e m e ly d if fi c u lt = 5 v e r y d if fi c u lt = 4 d if fi c u lt = 3 n o t d if fi c u lt = 2 n o t a t a ll = 1 t o t a l 1 active portfolio management strategy 181 149 45 11 3 389 2 passive portfolio management strategy 171 44 103 61 10 389 3 patient portfolio management strategy 32 29 68 187 73 389 4 aggressive portfolio management strategy 10 13 109 228 29 389 5 conservative portfolio management strategy 19 27 56 219 68 389 table: performances of project portfolio management strategies in achieving organizational objectives in portfolios s/n performances of project portfolio management strategies in achieving organizational objectives in portfolios e x c e ll e n t = 5 v e r y g o o d = 4 g o o d = 3 s a ti sf a c to r y = 2 p o o r = 1 t o t a l 1 active portfolio management strategy 57 147 85 88 12 389 2 passive portfolio management strategy 71 44 121 133 20 389 3 patient portfolio management strategy 88 189 68 28 16 389 4 aggressive portfolio management strategy 15 11 19 79 265 389 5 conservative portfolio management strategy 136 196 34 12 11 389 74 journal of risk analysis and crisis response, vol. 8, no. 2 (june 2018) 63-74 appendix – the research data obtained ___________________________________________________________________________________________________________ microsoft word study on food safety risk analysis under the condition of traditional agriculture.doc a study on food safety risk analysis under the condition of traditional agriculture jingyuan yin1 haiyan gao2 wei pei2 1school of computer engineering and science, shanghai university shanghai 200072, china, jyyin@staff.shu.edu.cn 2school of life sciences, shanghai university, shanghai 200444, china abstract with the development of economy, the problem of food safety has caused extensive concern. agricultural production node is on the front of the food safety management chain. the combination of which and modern economic social development is an important character to reflect the food safety situation in china. in a way of food safety risk analysis, the major problem under the condition of traditional agriculture is the whole product and sale process does not match with the modern information-based management. for one reason or another, it is hard for people to acquire kinds of planting and breeding data, which bring the great difficult to the work of food safety risk evaluation and supervision. the premise to solve the problem is to invest more in agriculture, substantially improve the quality of persons engaged in agricultural production in our country, change some backward product process modes under the condition of traditional agriculture as soon as possible and quicken the pace of informatization construction in rural area. and enacting and improving relevant laws and regulations of food safety by borrowing experience from the developed countries. keywords: traditional agriculture, food safety, risk analysis. 传统农业条件下的食品安全风险分析研究 尹京苑 1 高海燕 2 裴炜 2 1.上海大学计算机工程与科学学院，上海 200072 2. 上海大学生命学院，上海 200444 摘要：随着经济的不断发展，食品安全的问题会越来越引起人们的关注。作为食品安全管理链条的最前 端 － 农业生产环节，其与现代经济社会发展的融合度是体现我国食品安全状况好坏的一个重要特征指标。而从 食品安全风险分析的角度来看，传统农业条件下进行此项工作的最大问题在于，其整个生产销售过程与现代社 会的信息化管理格格不入，由于种种原因，人们很难真实有效地获取各类种植养殖数据，这给食品安全的风险 评估与监管工作带来了巨大的困难。而解决这一困难的前题是加大农业投入，切实提高我国农业生产人员的素 质，尽快改变传统农业条件下的一些落后生产加工模式，加速农村信息化建设步伐。同时借鉴国外发达国家的 先进经验，建立完善与之相关的食品安全法律法规体系。 关键词：传统农业，食品安全，风险分析 1. 引言 随着工业化的进程和科技的进步，国际食品安 全事件不断发生，1999 年比利时的“二恶英事件” 事件爆发之后，食品安全问题愈演愈烈，成为世界 各国关注的焦点。如何保证食品的安全，维护消费 者权利，促进食品贸易健康发展，是摆在各国面前 必须思考解决的问题。过去食品安全的管理是以对 不安全食品的立法、清除市场上的不安全食品和责 任部门的认可等项目的实施为基础的。然而，这些 传统的管理方法由于缺乏预防性手段，因此对食品 安全现行存在及可能出现的危害不能做出及时的应 答。 风险分析是近 20 年发展起来的一种解决食品安 全问题的手段，可为食品安全决策提供系统和规范 的科学数据。随着食品问题的不断复杂和多样，传 统的监管方式表现出了极大的效果和效率方面的无 能。而风险分析这种监管方式由于能够最有效地将 风险控制在萌芽状态，因此，起到了很好地预防食 品安全风险的作用[1]。食品风险分析代表了现代科 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 118-125 published by atlantis press copyright: the authors 118 学技术最新成果在食品安全管理方面实际应用的发 展方向，已经成为世界各国普遍遵循的规则，是制 定食品安全标准和解决国际食品贸易争端的重要依 据。 农业是食品的基础，而中国是一个农业历史悠 久的国家，农业在整个国民经济中占有重要的地 位。食品安全问题的发生与农业生产条件有着很大 的关系，我国目前的农业在很大程度上还是传统 的、原始的小农小本耕作经营，高成本、低效率是 它的显著特征，整个农业生产无序又疏于管理，科 技含量和知识含量很低，农业产业基础非常薄弱， 与发达国家相比，农业的落后程度远远超过其他行 业。我国人口众多，农业资源紧缺，农业发展受到 很大的资源制约，加上长期以来实行高度集中的计 划经济体制，商品经济没有得到充分发展，这些特 殊的国情，决定了我国未来农业发展必须走一条适 合中国国情的、具有中国特色的、又能与国际接轨 的农村改革和发展道路。 2. 食品安全风险分析的基本原理 食品安全风险分析始于20 世纪80 年代末，最先 应用于航天食品。1998 年在罗马召开的fao/who 联合专家咨询会上，形成了《风险情况交流在食品 标准和安全问题上的应用》的报告，标志着食品安 全风险分析的理论框架已经形成。如图1所示[2]，食 品安全风险分析包括风险评估、风险管理和风险交 流这3 个互为前提、相互作用的部分。这3 个部分形 成了一个科学的管理框架，其基本原理如下。 图 l 食品安全风险分析各部分之间的关系 风险分析是指对食品中可能存在的危害进行预 测，并在此基础上采取规避或降低危害影响的措 施，包括风险评估、风险管理和风险交流三个部 分。 风险评估是以科学为基础对食品可能存在的危 害进行界定，特征描述，暴露评估和描述的过程。 具体说就是通过现有的资料包括毒理学数据、污染 物残留数据、统计手段、暴露量及相关参数的评估 等系统的、科学的步骤，对食品中生物、化学或物 理因素对人体健康产生的不良后果进行识别、确认 和定量，决定某种食品有害物质的风险。它为风险 管理提供依据，其实施主体是食品科学、卫生学、 毒理学等方面的科学家。 风险管理是根据风险评估的结果，对公平贸易 的影响程度进行评估，选择和实施适当的管理措 施，尽可能有效地控制食品的风险，从而保障大众 的健康。其目标是选择措施，把食品风险降低到可 接受的水平，风险管理的实施主体是政府，政府主 要通过制定食品中有害物质或微生物的最高限量、 卫生标准、食品标签标准、食品法律法规，以及采 取公众食品安全教育等措施，保障公众健康。一般 包括风险评价、风险管理政策评估、执行管理决定 以及监控和审查几个过程。 风险交流是指在食品安全科学工作者、管理 者、生产者、消费者以及感兴趣的团体之间进行风 险评估结果、管理决策基础意见和见解传递及交换 的过程。这样通过公开、透明的信息交流，使各方 全面了解影响食品安全的各种危害、危害的特征、 危害的严重程度、危害的变化趋势、最高风险人 群、风险人群的特点和规模、风险人群对风险的接 受程度、风险人群的利益等信息。风险交流在于使 社会公众，尤其是广大消费者参与到食品安全管理 中去，促进政府综合考虑各种信息，提高决策的透 明度和科学性，制定更加合理的食品安全政策，将 食源性风险减少到最低限度，实现食品安全水平的 不断提高。风险交流的目标就是确保将所有关于有 效风险管理的信息和意见考虑进决策过程中[3 ，4，5， 6] 。 3. 传统农业条件下的食品安全问题 “传统农业”与“现代农业”的划分，是以农 业生产过程中生产技术手段应用的水平、程度和范 围来确定的。凡是大量应用现代先进的科学技术， 使劳动生产率大幅度提高，并有较发达的社会经济 环境和其他高技术配套产业的密切配合，为现代农 业；如果继续使用常规的生产技术手段，主要还是 人畜耕作、人工收种、靠天吃饭等，则是传统农 业。目前，中国农业虽然取得了许多可喜的成绩， published by atlantis press copyright: the authors 119 但从总体上说，特别是以占总人口百分之八十的农 民在广大农村所从事的生产来说，占主导地位的还 是以手工劳动为主的耕作方式，人力畜力使用还比 较普遍，农业劳动生产率还比较低，所通行的仍然 是以农户为单位的小规模生产，因此仍属于传统农 业方式的范畴[7]。 随着人类活动的发展，食品行业日益国际化， 开放、合作等因素使食品供给变得异常复杂，因此 现代食品安全问题作为一个社会及经济问题的存 在，也随着食品经济体系的变化而变化，并随其发 展而发展。不同的农业生产条件下，食品安全产生 的原因与存在的问题也各有差异，中国作为一个仍 处于传统农业方式下的国家。一般而言，食品供应 链是由农业、食品加工业和物流配送业等相关企业 构成的食品生产与供应的网络系统，主要围绕食品 生产、食品供应、食品物流与食品需求等四个主要 领域来组织实施的。处于建设初期的我国食品供应 链管理领域，一方面承受着国际化竞争的巨大压 力，一方面面临着许多亟待解决的问题[8]。 3. 1生产环节的食品安全问题 主要是农产品源头污染问题。一方面，随着社 会的不断发展，人口的不断增加，人们对粮食的需 求也在不断增加，因此为了满足人们日益增长的对 食物的需求，在现实中，农民往往投入更多的农 药、化肥来达到这种目的。同时，工业的发展让农 村环境受到污染，工业三废、城市废弃物的大量排 放，造成许多有毒、有害物质排放到土壤、水和空 气中。造成了农产品的种植、养殖正在不同程度地 受到农药、化肥、三废的影响与污染，人体在食用 了这些被污染的农产品后不仅将产生直接的健康危 害，而且还可造成食源性疾病的增加，同时这些化 学性、生物性的污染，往往会在食物链中进行蓄 积，给食品安全的风险分析与评估带来挑战。 3.2加工环节的食品安全问题 我国农业生产目前的现状为，在相当一部分地 区，农业还是小农经济的模式，农业设备极其落 后，生产工艺粗糙落后，管理混乱，许多生产者忽 视了质量安全，甚至根本没有质量安全的概念，这 是加工环节食品安全问题产生的主要原因。 这些安全问题主要表现在：一是企业生产规模 过小，管理混乱，未能按照工艺要求操作。以大米 为例，占总销量 50%以上的产品都是由 10 人以下、 设施简陋的家庭作坊式企业生产，这样厂家为追求 利润最大化，往往减少了设备、设施和管理的投 入，导致产品在加工、贮运过程中受到污染而出现 安全问题[8]。 二是超量使用食品添加剂，甚至非法使用非添 加物而引起食品安全问题。为了改善食品品质，延 长保质期，在食品加工中必须使用一定数量的食品 添加剂，但一些生产企业却为了降低生产成本，不 惜滥用非法使用一些添加物，增加了市场监管的难 度。 三是应用新技术、新工艺、新原料带来的问 题。如转基因技术、现代生物技术、益生菌和酶制 剂等技术在食品中的应用，辐照食品的安全性评价 等已经成为人们关注的热点问题。 3.3流通环节的食品安全问题 受经济发展等许多因素的影响，许多农产品或 食品仍然以传统的批发市场和农贸市场流通的形式 进行，这样存在着许多安全隐患。由于检测农药残 留量等会影响农产品的进场流量，因此许多小批发 市场和农贸市场缺乏有效的管理手段和机制，同时 由于销售场所简陋、卫生条件缺乏，许多新鲜果 蔬、水产品等产品极易腐败变质，增大了食品安全 问题发生的几率。 据统计，我国每年有总值 750 亿元人民币的食 品在运送过程中发生腐败。许多食品生产企业为了 追求利益的最大化，对物流设备的保养检修的重视 与投资力度不够，尤其是冷链物流及其相关设施严 重落后且缺乏维修更新，远远达不到食品保藏的温 度要求，致使大部分食品在运输过程中变质或质量 水平降低，结果不但造成了经济损失，而且影响了 企业的信誉和形象。 3.4消费环节的食品安全问题 现阶段，随着现代家庭结构的缩小与人口流动 性的增大，人们在外就餐消费等活动大大增多，使 得消费环节的食品安全问题变得更加严重。一方 面，一般消费者对日常使用的食品都会产生相关的 信息判断，这种判断直接左右了消费者的购买与使 用行为，久而久之甚至养成一种消费爱好或习惯； 另一方面由于受消费者相关知识水平等的限制、生 活习惯的影响、食品安全信息的不对称的影响，造 成这些消费习惯和信息判断上的不准确，往往引领 消费者走向一个对于食品的误区，从而使人们很难 根据自身的力量来保护自己，造成了消费环节二次 污染等许多食品安全问题频繁发生。 3.5食品安全问题产生的原因 published by atlantis press copyright: the authors 120 3.5.1 经济利益的驱动是产生食品安全问题最直接的 原因 在社会主义市场经济条件下，从事农产品种 植、食品生产、加工、销售的个人或企业都追求自 身的经济利益。为了追求经济利益的最大化，大量 以次充好，以假乱真的食品安全事件频繁发生。致 使在生产环节超量使用农药、化肥，滥用激素、抗 生素；在加工环节使用劣质原料或滥用及违法使用 一些添加物；在销售环节销售一些假冒伪劣产品。 3.5.2 法律法规体系的不完善，缺乏系统性和完整 性。 一方面，我国现有的这方面法律法规比较零 散，大多都是针对某个特定领域而制定的，国家的 许多规章制度含有关于食品安全的相关内容，但没 有形成一个有机的整体，这就极易导致食品生产、 加工、销售、消费某个环节缺乏相应的法律法规规 范和调整，而某些方面或环节，不同的法律法规又 有重复规定，这样在流通环节和农产品的再次加工 中给违法分子以可乘之机[9]。 3.5.3 食品安全的技术支撑体系不完善。 由于我国还是发展中国家，限于科技发展水 平，相关食品安全检测技术、检测设备都还比较落 后，同时我国食品的安全标准设定较低，许多的标 准没有建立在科学的风险分析基础上，不利于食品 安全监管的进行。 3.5.4 食品安全的各主体未充分发挥相应的责任。 首先，企业的社会责任不高，自律意识不强， 表现在生产、加工、销售过程中不能够严格地遵循 食品安全法律法规和标准，为追求经济利益极易违 反行业道德和标准甚至相关的法律法规。其次，广 大消费者食品安全意识比较淡薄，相关知识缺乏 [10] 。 3.5.5 缺乏一套行之有效食品安全预警体系 食品安全预警是指通过对食品安全隐患的监 测、追踪、量化分析、信息通报预报等，建立起一 整套针对食品安全问题的功能体系。对潜在的食品 安全问题及时发出警报，从而达到早期预防和控制 食品安全事件、最大限度地降低损失、变事后处理 为事先预警的目的[11]。 食品安全预警系统是食品安全控制体系不可缺 少的内容，是实现食品安全控制管理的有效手段。 食品安全预警通过指标体系的运用来解析各种食品 安全状态、食品风险与突变现象，揭示食品安全的 内在发展机制、成因背景、表现方式和预防控制措 施，从而最大限度地减少灾害效应，维护社会的可 持续发展。预警的关键在于及时发现高于预期的食 品安全风险，通过提供警示信息来帮助人们提前采 取预防的应对策略[12]。 目前我国还缺乏一套行之有效的食品安全预警 体系为消费者提供充足、可靠的安全信息；及时发 布食品安全预警信息、帮助社会公众采取防范措 施；对重大食品安全危机事件进行应急管理，尽量 减少食源性疾病对消费者造成的危害与损失。 4. 风险分析在现代化农业条件下发达国家的应用 近年来，发达国家为了提高食品安全监管效 率、保护消费者健康和促进食品与农产品国际贸 易，在食品安全风险分析制度建设方面取得了很大 进展。食品法典委员会和一些发达国家开展了疯牛 病( bse )、沙门氏菌、李斯特菌、o157: h7 、二恶 英、多氯联苯、丙烯酰氨等的系统研究[13]，已经形 成了化学危害物、微生物、真菌毒素等风险分析指 南和程序。当前风险评估技术已发展到能够对多种 危害物同时形成的复合效应进行评估，并且更加注 重随机暴露量的评估。另外，国际社会对转基因食 品(gmo)的安全性评价问题也形成了评价原则和程 序， 以美国、德国、日本等 3 种农业现代化道路国 家情况为例，可看出风险分析在各发达国家的应用 情况和特点。 4.1风险评估机构独立，与风险管理职能分工明确。 美国的食品安全风险分析由卫生部食品药品管 理局(fda)、农业部(usda)、环境保护署、海关与 边境保护局等按照各自的职能，实行共同监管， usda 负责肉类和禽类等食品的管理，监督食品安 全法规的执行。卫生部食品药品管理局(fda)负责其 他类食品和进口食品的管理，制订相关法规和最高 残留限量(mrls)等食品安全标准。此种多机构联合 监管的模式体现在联邦政府、州和地方等层面，可 开展纵向的风险分析，也可跨机构进行横向合作。 早在 1997 年 5 月，美国就成立了风险评估联盟，旨 在协调风险评估，交流信息，促进食品特别是动物 源食品的风险评估和科学研究。 2002 年 11 月，依据《消费者健康保护和食品 安全重组法案》，德国联邦消费者保护、食品与农 业部设立了联邦风险评估研究所和联邦消费者保护 与食品安全局。其中，联邦风险评估研究所负责风 险评估和风险交流；联邦消费者保护与食品安全局 负责风险管理及与联邦、其他国家及欧盟机构的协 调。联邦风险评估研究所下设 7 个部门，每个部门 下设工作组或实验中心，负责本领域内风险评估的 相关工作，通过各部门的明确分工与相互协作，联 published by atlantis press copyright: the authors 121 邦风险评估研究所可完成对食品中生物与化学的风 险评估，并以风险评估结果为基础提出降低风险的 管理选项，为联邦政府部门和其他风险管理机构提 供建议。 2003 年 7 月 1 日，依据《食品安全基本法》， 日本成立了食品安全委员会，专门从事食品安全风 险评估和风险交流工作。该委员会由 7 名食品安全 专家组成，委员全部为民间专家，经国会批准，由 首相任命，任期 3 年。委员会下设事务局和专门调 查会，事务局负责日常工作，专门调查会负责专项 案件的检查评估。食品安全委员会的主要职责是实 施食品安全风险评估；对风险管理部门进行政策指 导与监督；负责风险信息的沟通与公开。从发达国 家的经验来看，独立的风险评估确保了食品安全立 法与标准制定的科学性，为有效的风险管理提供了 依据[14 ，15] 。 4.2 风险管理较为有效 在管理思想上，各发达国家一改过去基于事后 处置的管理体系，转为进行事前干预的有效管理。 另外从政策效率角度出发，让生产者承担生产安全 食品的主要责任，让政府实行有效的监督，为政府 的管理节约了大量的人力、物力与财力。 在管理体制上，国际食品安全管理体制的核心 是加强食品安全管理各部门之间的协调。为此大多 数国家改变了原先按食品品种或按生产阶段来划分 监管部门的职能的分割式体制，将原有的食品安全 管理部门统一到一个独立的食品安全机构，如加拿 大、德国、丹麦和澳大利亚等国家。由于分割式体 制转变为统一式体制的运作成本较高，因此，也有 许多国家食品安全管理机构依然分布在不同部门， 但通过明确分工来协调实现食品安全，比如美国和 日本。 在管理手段上，发达国家更多的是鼓励企业进 行自我质量管理。食品加工企业进行自我质量管理 的主要手段有良好生产规范（gmp）、良好卫生规 范(ghp)、危害分析与关键控制点(haccp)。其中， gmp 和 ghp 是 haccp 实施的前提条件，而 haccp 则融合了风险评估和风险管理两个阶段，既 是企业控制食品安全的方法，又是政府进行管理的 有效手段，为发达国家广泛采用。美国是最早应用 haccp 系统的国家，目前美国海产品的加工者、肉 禽产品的加工者及果蔬汁加工者都被要求执行 haccp 系统，同时欧共体、日本、澳大利亚、新西 兰也将实施 haccp 体系作为解决本国食品安全问 题的有效措施[16]。 4.3风险交流比较充分 食品安全管理的最终目标是减少食源性疾病的 发生，保护公众健康，其中消费者作为食品安全管 理的最终服务目标，对政策的理解影响着他们对市 场的信心，消费者在整个风险分析中占据着十分重 要的地位，因此也受到了各国政府的重视。 在德国，风险交流作为联邦风险评估研究所的 一项法定任务，应向消费者提供有关食品和产品中 可能存在及已被评估的风险信息。当需要在较大范 围内通知公众时，除媒体外，消费者建议中心、产 品比较团体、消费者保护、食品与农业信息服务部 都会成为风险交流的重要手段。为了实现风险信息 交流的持续和互动，联邦风险评估研究所定期组织 专家听证、科学会议及消费者讨论会，并面向一般 公众、科学家和其他相关团体公开其评估工作与结 果，并通常会在其网站上公布专家意见和评估结 果，还积极寻求以简易的方式与普通公众对评估过 程进行交流，并向消费者提供可见和可用的科学研 究成果。通过全面的风险交流，一方面，尽早发现 潜在的健康风险并及时通知有关部门和消费者；另 一方面，参与交流的各相关方会对风险评估的过程 与结果进行讨论，通过工作的透明度，在风险评估 涉及的各方之间建立起足够的信任。在日本人，食 品安全委员会主要通过召开国际会议与国外政府、 国际组织和相关部门进行风险交流，来交换各方意 见和建议，确定自身食品安全风险评估的方向。同 时通过网站、热线和专人信息采集与公众进行风险 交流，来听取消费者和公众的意见与建议。食品安 全委员会通过每周 1 次公开召开委员会议，并在其 网站公布会议议程来保证实施风险评估的透明性。 通过建立“食品安全热线”，专门用来接收日本民 众对于食品安全的要求和意见。同时，食品安全委 员会从各县选拔任命了 470 名“食品安全监督 员”，监督员通过发放调查问卷来了解食品安全事 件引起人们关注的程度，及时汇报相关信息，并且 协助各地方组织进行信息交流[15]。 5. 风险分析在传统农业条件下的应用情况 中国的食品安全风险分析主要参考美国模式。 我国在传统农业条件下的食品安全管理方面最初应 用风险分析是在 20 世纪 90 年代中后期，在农产品 质量安全方面谈论风险分析最初始于 2001 年。2002 年农业部也专门成立了农业转基因生物安全评价专 家委员会，对农业转基因植物、动物和微生物开展 风险评估和安全评价。2006 年 11 月 1 日施行的 《农产品质量安全法》确立了风险评估的法律地 published by atlantis press copyright: the authors 122 位，要求把风险评估结果作为制定农产品质量安全 标准的重要依据，农业部门要根据风险评估结果采 取相应的管理措施。2009 年施行的《中华人民共和 国食品安全法》也对食品安全风险评估及其结果的 利用做出明确规定，表明我国的风险评估已进入法 制轨道。目前，正在修订的《食品卫生法》将首次 增加食品安全风险评估这一重要内容，政府有关监 管部门可以根据风险评估的结果对高风险的食品重 点监管。新修订的《食品中农药最大残留限量标 准》(gb276320 05) 和《食品中污染物限量》(gb 2762200 5)较大程度地引用了 cac 标准的风险评估 数据，等同性均提高了 85%以上。兽药典也大幅度 采用 cac 标准的风险评估数据，等同性均提高到了 41.4%[17]。《农业转基因安全评价管理办法》规定对 人类、动植物、微生物和生态环境构成的危害进行 风险评估和分析。《兽药注册办法》规定: 农业部对 申请进口注册的兽药进行风险分析，经风险分析存 在安全风险的，不予注册。 近年来，我国商务、卫生、农业和检验检疫部 门针对食品方面的危害分析做了大量工作，检验检 疫部门结合我国进出口贸易中出现的热点问题和国 际热点问题在口岸开展了应用实践，如对酱油中三 氯丙醇；苹果汁中甲胺磷、乙酰甲胺磷残留；禽 肉、水产品中氯霉素残留；冷冻加工水产品中金黄 色葡萄球菌及其肠毒素；油炸马铃薯食品中丙烯酰 胺；水产品中金属异物；牡蛎食用中感染副溶血性 弧菌；进境冻大马哈鱼携带溶藻弧菌可能影响人体 安全和水产动物健康的风险评估等等的风险评估 [18] 。 然而，由于起步较晚，法规不健全，风险分析 缺乏完整、配套的制度保障。食品安全风险分析只 停留在对某一种食品和产业链的某一环节上，风险 分析体系尚未形成。中央一级的相关工作由农业 部、卫生部、国家质量监督检验检疫局等多部门负 责，没有设立专门的机构，风险评估能力弱。信息 传播机制不完善，风险交流不充分。食品行业缺乏 诚信，影响风险分析效果。 5.1 传统农业条件下食品安全风险分析工作相关的法 律法规不完善 由于我国在传统农业条件下应用食品安全风险 分析的起步时间较晚，又多以国外的相关研究为基 础，在我国尚未摸索出一套适应我国国情的，能够 有效指导传统农业条件下进行食品安全风险分析工 作的法律法规。 5.2传统农业条件下的食品安全风险分析框架尚未形 成 我国现在并不缺食品安全风险分析的专业人 员，缺的是一种制度保障。比如，风险评估需要大 量的监测数据，而我国并没有完善的、畅通的监测 网络，无法有效地变事后监管为事前预防。在以往 的食品安全事件的处理上，往往只是停留在对某一 种食品或者一起食品安全事件的评估上，没有专门 的权威机构从整体上对该类食品的整条生产消费链 的风险进行评估。这种方式一味的追求当前利益和 效果，治标不治本无异于“饮鸩止渴”。 5.3传统农业条件下适应食品安全风险分析的外部环 境尚未成熟 在我国农业区域辽阔、各地农业资源差异巨 大、农业技术应用不充分、农业科技应用水平参差 不齐、农村劳动力数量庞大而素质低下和农业投入 不足等客观现实下，虽然有时用了风险评估的方 法，但由于样本量小，检测监测手段有限，食品安 全风险分析体系所需的信息交流渠道闭塞，技术人 员缺乏，使得获得的相关数据品质不高，这些因素 对于评价水平、评价结果有很大的负面影响。 5.4传统农业条件下的食品行业缺乏诚信，影响风险 分析效果 开展食品安全风险分析必须有一个诚实信用的 环境。技术、设备、人才、资金等是食品安全风险 分析的重要硬件，而企业诚信是食品安全分析的首 要软件。如果食品生产者为了追求经济利益，不顾 消费者健康，向食品中添加有毒有害物质，那么再 完美的食品安全风险分析也无法保障食品安全。食 品安全风险分析在发达国家得到很好应用，一个重 要的原因是发达国家食品行业的诚信水平普遍较 高，保证了食品安全法律标准严格的执行[4]。 6. 传统农业条件下进行食品安全风险分析的建议 6.1借鉴发达国家应用风险分析的先进经验，构建适 合我国国情的食品安全风险分析体系. 我国作为 wto 成员，应遵循国际组织相关规 定，积极进行国际交流，及时了解并学习发达国家 的先进经验，逐步建立、完善适合我国国情的食品 安全风险分析体系。 6.2建立和完善食品安全相关的法律法规体系 published by atlantis press copyright: the authors 123 国家立法部门要根据现阶段我国食品安全问题 的实际情况，对现有的食品安全法律、法规、条 例、标准、规范等进行整合，使之系统化；同时需 要制定、出台相应的新的食品安全法律法规，拒绝 法律的空白或薄弱环节和领域，使之完整化。在食 品安全监管过程中，一方面用法律形式明确各级相 关政府部门的职责，针对具体问题要确保法律的可 操作性，做到为实际问题服务；另一方面用法律形 式明确食品领域生产者、消费者等主体的权利义务 关系。从而真正做到食品安全有法可依，有法必 依。 6.3政府、企业、消费者协调统一发挥各自作用 政府应致力于食品安全法规的建立和实施，在 法律的框架范围内改进国家食品安全管理机构的组 织形式和管理模式。同时协调与国际组织，尤其是 国内涉及食品安全的各行政机构的信息收集、交换 和整合工作。收集国内外的有关信息，尽早掌握事 态的发展，协调有关行政机关迅速应对。 食品企业应该在生产过程中充分发挥 haccp 管理的作用。haccp 通过对整个食品链，包括原辅 材料的生产、食品加工、流通、乃至消费的每一环 节中的物理性、化学性和生物性危害进行分析、控 制以及控制效果验证的完整系统。风险分析是对 haccp 管理体系的进一步补充和完善，也是实行 haccp 管理体系的基础。作为主体的企业的自律行 为和实践经验，行业协会的监督约束，均为进行风 险管理提供了宝贵的经验。 消费者对风险分析的认知程度是实施风险分析 的基础。随着社会发展和人们对食品安全的了解程 度及关注程度的不断提高，借助行业协会和团体会 议的交流、网页、公开发行的刊物等媒体的作用， 通过消费者的积极参与和信息交流，最终能为实施 风险分析奠定良好的基础。[19] 6.4改善食品安全监督管理模式，加大食品安全的技 术支持 在相关法律法规的调整下，执法部门要把法律 法规落实到位。一是要整合执法力量，形成统一、 权威、高效的食品安全监管模式，解决多头执法、 重复执法等问题。二是要加快食品安全标准体系建 设。政府有关部门应加快健全食品安全标准体系， 加快检验检疫农药残留限量、食品添加剂等方面的 食品质量安全卫生标准的制定修订步伐。包括与国 际标准、国家标准以及行业标准相配套的地方标准 体系建设，并向农民、企业推荐和介绍国际标准和 国外先进标准，积极引导和鼓励他们达到先进标 准。三是要努力提高食品安全检测、控制技术，完 善检测网络体系。一方面加大投入，提高我国食品 特别是农产品安全生产和检测技术，强化农药残留 检验手段和技术；另一方面加大投入，努力更新完 善检验检测设备，同时管理部门可以把农业教学、 农技推广等部门的检验设施充分利用起来，从而形 成完整的检测网络。 6.5实施风险评估和风险管理职能分离机制 风险评估不仅是风险分析的核心，也是进行风 险管理的科学依据。风险评估这一个科学过程，应 由科学家独立完成，不应受政治、经济、文化的影 响。而风险管理是由政府完成的一个决策过程，必 然会受政治、经济、文化的影响。所以为保障风险 评估的科学性、客观性、透明度和有效性，应将风 险评估和风险管理职能分开，成立专门的风险评估 机构，遵循内外一致的原则进行风险评估[20]。 6.6建立科学、统一的食品安全检测体系 食品安全检验检测是食品安全风险管理的重要 手段之一，它为食品安全风险管理提供重要的技术 支持和管理政策依据。目前，我国食品安全检验检 测体系的基础框架虽然已经初步形成，但是食品安 全检验检测机制、设施、技术力量和手段等还不够 完善。当前存在的突出问题是质监、卫生、农业等 部门往往按照本部门颁布的有关规定进行检测，检 测的结果比较独立，部门与部门之间缺乏良好的共 享和互认机制。我国必须建立科学、统一的食品安 全检测体系，强化技术监管，加强食品安全风险管 理。通过监测收集一批与中国农产品质量安全相关 的分析数据，根据中国农产品质量安全预警体系， 适时公告一批高风险危害因子和对应的产品，提请 公众关注和警觉，政府部门也要采取相应的防范措 施；以正确引导生产消费，增强消费信心[17]。 7. 结论 一般来说，传统农业下的生产模式与食品安全 问题并没有直接的关系，因为即便在生产力十分落 后的封建社会，人们对大部分直接种植养殖出的食 用农产品的“安全”问题也基本不用担心。然而随 着经济的发展，城市化的推进，传统农业生产条件 下引发的食品安全问题却越来越引起人们的关注， 这期间除了农业种植养殖自身发生了一些改变之外 （如越来越多地使用农药、化肥和各类饲料添加 剂；以及现代市场经济对传统农业生产条件下的道 德、诚信规范的冲击），而从食品安全风险评估的 published by atlantis press copyright: the authors 124 角度来看，传统农业生产与现代物流、加工业的信 息化管理严重脱节，是引发食品安全诸多问题的重 要原因之一。这是因为引发食品安全的因素众多， 对其的监管是一个十分复杂的系统工程，问题的产 生即与源头的生态环境保护程度和农产品种植养殖 的过程有关，也与运输、加工方式，以及消费者对 食品安全的认知程度紧密相连。进行食品安全风险 评估首先需要有效地获得食品生产、食品供应、食 品物流与食品需求等四个主要领域产生的检验检 测、问卷调查等相关数据。由于我国广大农村目前 信息化管理水平低下，有些地方在这方面基本就是 空白，因此很难保证风险评估的基本数据的获取； 加上经济飞速发展下的利益诱惑，诚信体制严重缺 失，更给食品安全监管过程中的源头控制工作带来 了巨大的困难。 参考文献 [1] 石兴，自然灾害风险可保性理论及其应用研究，北京师 范大学博士学位论文，2009。 s. xing, insurability s tudy and i ts applic ation o n natura l disasters, phd thesis, beijing normal university, 2009. [2] 张兆本，公共经济学, 北京：人民出版社，2005 z. b. zh ang, public e conomics, beijing: the people' s press, 2005 [3] 曾莉娜，中国食品安全风险分析机制研究，上海师范大 学, 2010 [4] 吴培，许喜林，张毅，食品安全风险分析与应用，现代 食品科技，2006，22(04)，200-203 [5] 魏益民，基于风险分析原理的中国食品安全法规体系建 设，中国食物与营养，2009，8:11-13 [6] 徐成德，我国开展食品安全风险分析的问题与对策，农 产品加工，2009，2:61-66 [7] 袁宗辉，我国应重视食品安全风险分析，华中农业大学 学报，2010，(3):8-12 [8] ariane koing ，harry a. kuiperc ，hans j. p. marvin，et al.the safe food s framework fo r improved risk analysis o f foods. food control 2010，21:1566-1587 [9] http://www.gzu521.com/paper/article/industry/200607/2272 7.htm [10] 张卫斌，顾振宇，基于食品供应链管理的食品安全问题 发生机理分析，食品工业科技，2007，28(1):215-220 [11] 张森富，赵婷，现阶段我国食品安全问题的成因及其对 策研究，长春理工大学学报，2010，23(1)：38-39 [12] r.b.tompkin. interactions be tween gov ernment and industryfood safety activities. food control ， 2001 ， 12:203-207 [13] 叶存杰，基于.net 的食品安全预警系统研究，科学技 术与工程，2007，(2):258-260 [14] 杨君，黄丽，袁利鹏，关于完善我国食品安全风险预警 系统的思考，食品科技，2009 年第 34 卷第 12 期 [15] 刘春霞，浅谈食品安全风险分析，2008 年度标准化学 术研究论文集, 2009 [16] 滕月，发达国家食品安全规制风险分析及对我国的启 示，哈尔滨商业大学学报，2008，(5):55-57 [17] 王芳，陈松，钱永忠，国外食品安全风险分析制度建立 及特点分析，世界农业，2008，(9)；44-47 [18] 周应恒，彭晓佳，风险分析体系在各国食品安全管理中 的应用，世界农业，2005(3): 4–6 [19] 范梅华，张建华，风险分析-我国食品安全管理的必由 之路，中国禽业导刊，2007 ，24（24） [20] 刘志英，风险分析—我国食品安全管理新趋向，内蒙古 科技与经济，2006，4( 5) : 141143 [21] 孙俐，贾伟，食品安全风险分析的发展与应用，食品研 究与开发，2008，29（6） [22] 王成，陈小莉，食品安全风险分析是我国食品安全管理 必经之路，上海食品药品监管情报研究，2008，12(95) published by atlantis press copyright: the authors 125 microsoft word research on characteristics and formation mechanism of landslide disaster in red soil hilly region of south ch research on characteristics and formation mechanism of landslide disaster in red soil hilly region of south china qinghua gong guangzhou institute of geochemistry, chinese academy of sciences, guangzhou 510640, china guangzhou institute of geography, guangzhou 510070, guangdong, china university of chinese academy of sciences, beijing 100049, china guangqing huang guangzhou institute of geochemistry, chinese academy of sciences，guangzhou 510640，china junxiang zhang tourism college, huangshan university, huangshan245021, china abstract the mechanism of landslide hazard in weathering crust of granites in south china was chosen as our research object. landslides, as a common geo-hazard, can result in huge economic losses and enormous casualties in mountainous regions of the world. therefore, the mechanism of landslide hazard has become a topic of major interest for both geoscientists and engineering professionals. take the landslides mechanism as the key problem. firstly, we will analyze the regularity of its regional differentiation in those area based on the field investigations and the available data in the area according to the features of this kind of landslide. secondly, we will reveal the growth characteristics and disaster mechanism of landslide by study the relationship between the characteristics of intensity and frequency of landslide and meteorological factors, hydrologic geology, geologic factors, landform and human factors and so on. finally, based on these result, the regional landslide prediction model was obtained by quantification approach. the final goal is to reduce the landslides risk as low as possible. our subject will provide scientific basis for risk precaution and disaster emergency response. key words：south china; the red soil hilly region; landslide; mechanism 华南红壤丘陵区滑坡灾害特点及致灾因素分析 宫清华 1，2，4 ，黄光庆 1，2 ,张俊香 3 1. 广州地理研究所，广州 510070 2.中科院广州地球化学研究所，广州 510640 3. 黄山学院旅游学院, 黄山 24502 4.中国科学院大学，北京 100049 摘要：本文针对华南沿海风化壳滑坡的特点，以近年华南沿海地区的千余处滑坡灾害点的空间分布和地质 环境特征等资料为基础，研究花岗岩风化壳滑坡的孕灾环境和灾害强度、灾害规模、范围等空间分布特征 地域分异规律；解析华南沿海花岗岩风化壳滑坡灾害的发生规律和形成机理；分析滑坡灾害强度和频度与 地质、地貌、气象、水文、生态人文的关系。本研究对滑坡灾害机理研究有重要的科学意义，对华南沿海 未来在滑坡灾害风险防范的对策和灾害应急预案的制定具有极重要的参考依据。 关键词：华南，红壤丘陵区，滑坡，形成机理 1.引言 华南沿海地区以花岗岩风化壳分布最广，该地区 在雨季，地下水沿不透水风化层面流动使得滑坡灾害 频发，华南地区山地处在强烈的季风作用交互带上， 在全球气候变化和不断加剧的人类活动影响下，山地 系统脆弱性更为明显。特别是华南山地区域目前正处 于快速城市化和城乡统筹发展这一整体社会发展背景 下，华南红壤丘陵区正在发生巨大的变化，也使得山 区地质环境面临着巨大的威胁，而滑坡灾害时威胁山 区人居环境的突出生态问题。据统计，连降雨偏少三 成的 2010 年广东省发生较大规模的突发性地质灾害 也达到 238 宗，其中，滑坡 103 宗，华南山区滑坡灾 害防治任务刻不容缓，但灾害防治领域的一系列科学 问题还有待突破，尤其是滑坡灾害的诱发机理问题一 直是世界上公认的难题[1]，21 世纪初美国地质调查局 的滑坡灾害减灾战略规划，将滑坡过程和诱发机理研 究列为首要的任务，这不仅因为滑坡、泥石流形成机 journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 110-114 published by atlantis press copyright: the authors 110 willieb typewritten text received 28 june 2013 willieb typewritten text accepted 7 july 2013 willieb typewritten text 理和诱发机理研究是至今没有突破的难题，更重要的 是它成为制约地质灾害预测预警和防灾、减灾研究的 瓶颈问题。因此，长期以来，国内外许多地学专家、 学者都将其作为攻克目标，经过多年的发展，对滑坡 发育机理的研究取得了一些探索性的成果。但对于华 南花岗岩风化壳地区的滑坡，在很多方面的问题尚待 解决。本文试图通过解析华南红壤丘陵区滑坡灾害的 形成机理，探索滑坡灾害发育的本质规律，并对华南 山区未来的滑坡灾害发育趋势做出基本的判断。这对 系统地发展有中国地理特色的地球系统科学具有特殊 的科学价值和重要科学意义，并对提高我国在国际地 球系统科学领域的科技影响力和引领作用具有更重大 的意义。 2. 华南红壤丘陵区地质地貌特征 2.1 研究区概况 华南地区的滑坡灾害与地质环境，尤其与风化壳 发育深厚的地质环境有着密切的关系 [2] 。华南地区的 滑坡几乎大部分都分布于各种岩类的风化壳发育地 带，尤其在花岗岩类风化壳区域的案例相对较多。花 岗岩风化壳又是华南沿海分布最广的风化壳类型。广 东省梅州地区是属于典型的花岗岩地区，根据前期工 作的调查表明，该区域花岗岩风化壳厚度较大，约有 16-19.5m，因此，本项研究选择梅州市域作为红壤丘 陵区滑坡灾害发生规律和成灾机理研究的范围。梅州 市位于广东省东北部，地理坐标东经 115°18′116°56′， 北纬 23°23′24°56′。东南与潮州市、揭阳市、汕尾市 接壤，东北与福建省南靖、永定、武平毗邻，西部与 河源市、江西省寻乌相连。面积 15876.05 km2。梅州 市的地貌格局和灾害点分布情况如图 1。 图 1 灾害点与高程分布图 表 1 红壤型风化带物理力学性质表（据何珊儒，2005） 风化带 密度 ( g/ cm 3 ) 干容重 ( g/ cm 3 ) 吸水率 ( %) 孔隙率 ( %) 饱和吸水率 ( %) 极限抗压强度 ( mpa) 极限抗拉强 度( mpa) 强风化带 2.622.64 2.402.50 2.50-2.55 5.306.10 2.512.83 3.509.30 2.403.27 中风化带 2.63-2.65 2 2.382.56 0.903.92 2.809.93 1.234.69 38.40-65.70 1.4454.31 微风化带 2.622.67 2.55-2.63 0.40-0.45 1.56-2.30 0.300.45 65.50138.20 5.8612.12 新鲜基岩 2.64-2.67 2.58-2.64 0.380.42 1.181.50 0.350.43 143.76238.12 5.6313.46 2.2 红壤丘陵区风化壳剖面划分 华南地区地处亚热带，气候热湿，干湿季节明显， 在地势低平和坡度和缓的丘陵低山区，形成了巨厚的 花岗岩红色风化壳 [2] 。华南沿海约有 70%的面积为中生 代花岗岩和火山岩所占，其余为新生代玄武岩、各时 代的沉积岩和少量的变质岩以及松散的第四系。基岩 的表层在高温(年平均气温 21-28℃)和潮湿(年降雨 量 000-2000mm，年相对湿度 75-85%)的气候条件下， 分解和淋溶作用极强，发育为富铝风化壳，相对富集 了铁、铝、钦等活性较小的元素，而以红色为特征， 红壤型风化壳，是华南沿海分布最广的第四系类型。 不同岩性其风化壳的厚度差异很大, 砂页岩的风化壳 一般仅几米厚, 石灰岩风化壳的厚度一般小于 10m, 花岗岩由于节理特别发育(包括水平和垂直节理)，所 以比其他岩石风化作用更为深入，因而形成了较深厚 的风化壳。根据化学风化程度的不同, 把花岗岩风化 壳自上而下分为四层: 全风化层、强风化层、弱风化 层和微风化层 【3】 ，最厚的可达 50-60m。细粒花岗岩为 20-30m(广东梅州), 而粗粒或斑状花岗岩风化壳的厚 度可达 70-80m(广东陆丰)，这也为滑坡的发育提供了 物质基础 [4，5] 。 2.3 风化壳的地质力学特性 随着风化程度的加深,风化层的物理力学性质也 随之变化，风化岩石的比重、干容重等指标多呈降低 的趋势，而其孔隙率和吸水率等指标多呈增加的趋势； 而随着风化程度的加深，岩石强度呈逐渐下降的趋势。 [6，7] 花岗岩风化壳的各物理力学特征如表 1。 斜坡整体的稳定状态主要取决于斜坡内部的构造 力学特性和物质组成，从上表可以看出，红壤型风化 壳中吸水性和孔隙度随着风化程度的增加而减少，到 基岩为不透水层，而风化土是一种随着含水量而表现 出显著软化的特殊性土，在表层风化壳软化，基岩不 透水，表层土体松散软弱，基岩完整坚硬的垂直分布 格局下，若有含水性质或力学荷载发生变化时，斜坡 内部平衡极易收到破坏，进而形成浅层滑坡。 2.4 红壤丘陵区的地貌形态特征 在华南沿海，发育厚层风化壳的花岗岩山丘，形 态多圆浑，一般坡度多在 30 度左右。研究区梅州市位 于粤东低山-丘陵区的东部，以山地丘陵为主，山谷相 间的地貌格局，地势起伏较大。海拔 100m 以下的平原、 阶地、台地占土地面积 16.54％，100-400m 的丘陵占 54.36％，400-800m 的低山占国土面积的 20.13％， 800-1000m 以上中山占 3.21％。三列东北-西南走向的 山系有：罗浮山系、莲花山系、凤凰山系；三列西北published by atlantis press copyright: the authors 111 东南或近乎南北向的山地有：七目嶂山地、铁山嶂山 地、蕉平山地，构成梅州的地形格局（图 1）。中部 梅江河及其支流盆谷区，由梅江河及其支流沿线冲积 盆地和谷地组成；东南部山高坡陡，河谷深切，地形 反差大；西北低山-丘陵区，区内陵谷相间，山势迤逦 起伏，地形高差一般小于 500m。梅州市地形格局总体 上北东向展布，以遭受流水切割、搬运和综合外力风 化、剥蚀地质作用为主，由于雨量充沛,沟谷密度大, 切割深, 故普遍发育天然的切割面,这些切割面常以 风化壳底面为侵蚀下限。另一方面,发育的地表径流不 断冲刷坡脚而塑造不同规模的临空面，为风化壳滑塌 创造了各种边界条件。地表水沿节理和表层残积物渗 入, 极大地降低了风化层的抗剪强度, 导致滑塌频繁 发生。 图 2 灾害点与地质构造的关系 3.华南红壤丘陵区滑坡特性分析 3.1 时间分布特征 根据野外调查以及收集当地资料，本文统计了有 记录的 1058 个滑坡灾害点的发生月份以及当月的雨 量，滑坡灾害的发生具有明显的季节性，这种季节性 与降雨的季节性基本吻合。灾害点分布与雨量分布如 图。在全梅州市已知发生时间的地质灾害点中，有 92.1%处发生在 4～8 月的雨季。发生于 4 月、5 月、6 月、7 月、8 月的滑坡数分别占总数的 10.9%、23.3%7、 19.6%、13.8%、24.5%，各发生滑坡的百分比与当月降 水量基本成正相关。而干旱季节，降雨较少，滑坡次 数也相对较少，仅占滑坡总次数的 8.9%。滑坡一般发 生在暴雨期，大规模的滑坡暴雨强度临界值为 200mm 以上。   图 3 滑坡灾害点与发生时间及降雨量统计图 3.2 空间分布特征 根据区域调查，目前梅州市共有滑坡 1448 处，地 质灾害主要分布在丰顺、五华、兴宁、梅县、大埔各 县。 滑坡发育与地形地貌密切相关。从分布空间看，绝大 多数滑坡灾害点分布在海拔 100～300m 的丘陵地区， 这些区域坡度适宜，有利于形成较厚的风化壳堆积层。 从坡度分布来看，最易发生滑坡和崩塌的斜坡坡度为 2040°，占总地质灾害总数的 76.8%；从坡向分布 来看，阴坡发生地质灾害的概率比阳坡大。     图 4 灾害点与坡度分布图 3.3 强度特征 红壤丘陵区的滑坡灾害一般为小型浅层滑坡，滑 体厚度约 1m 左右，且滑动距离短，牵引式（自坡脚逐 级向上牵引发展），在山丘陡壁，陡坡上则常沿裂隙 面发生整体滑塌。据调查,研究区域内，灾害体规模属 小型，占 96.36%；中型以上占 3.64%。 4.华南红壤丘陵区滑坡灾害发育影响因素分析 红壤丘陵区滑坡灾害的形成与发生是多重致灾因 素交互作用的结果，总体来说，巨厚的红风化壳和底 部不透水基岩是滑坡灾害发生的物质基础，其岩性特 征和力学性质是滑坡形成的内在因素，它在一定程度 上决定着滑坡灾害的发育程度和类型，地形是地质灾 害形成的外在条件，它制约着滑坡等致灾作用的形成， 而降水和人类工程活动是地质灾害形成和发育的激发 published by atlantis press copyright: the authors 112 因素，决定着灾害发生的速度和时间。下面从几个方 面探讨红壤区滑坡的生成机理。 4.1 风化壳的影响 华南红层风化壳是滑坡形成的物质基础和内在条 件。按照滑坡的物质组成，对研究区 1448 处滑坡进行 分类统计，结果表明，发生在风化壳的土质滑坡有 1392 处，占滑坡总数的 96.6%，而岩质滑坡仅有 56 处，占 3.4%。 从基岩性来看，滑坡主要发生在与岩浆作用有关、风 化层较厚的块状较硬-坚硬岩组。岩浆岩经强烈的红土 化作用，脱硅富铁铝、盐基遭受较强烈的淋失，因而 土体结构疏松，孔隙较多，加上风化壳不同层位粒度、 成份、结构等各方面差异，使得风化壳土体抗冲抗蚀 能力差 [8，9] 。 除了风化壳本身力学性质差而容易造成滑坡外， 另一个造成滑坡的重要因素是风化壳内部存在大量的 软弱结构面，它们包括原岩风化后残留下来的构造裂 隙或原生节理，风化过程中产生的风化节理和卸荷节 理。花岗岩岩体形成时常伴生 3 组以上不同方向的原 生节理。通常在区域构造复杂、褶皱强烈、断层众多、 岩体裂隙发育、新构造引动比较活跃的地区，往往岩 体破碎、沟谷深切，较大规模的崩塌、滑坡也比较常 见。梅州市地质构造与滑坡形成和发展的关系表现为 两个方面：滑坡沿断裂破碎带往往成群成带分布。各 种软弱结构面（如断层面、岩层面、节理面、片理面 及不整合面等）控制了滑动面的空间展布及滑坡的范 围如图 2。常见的顺层滑坡的滑动面大部分是由岩层 层面或泥化夹层等软弱结构面构成的 [5，7] 。 4.2 地形的影响 地形地貌为滑坡灾害的发育提供有利的地形坡度 和临空面,主要因素有高程、地形坡度和坡向 [8-10] 。 4.2.1 高程的影响 地质灾害发育与地形地貌密切相关。高程的变化 对滑坡的影响主要表现在提供高于侵蚀基准面的斜 坡，使其有可移动的临空面。从分布空间看，绝大多 数滑坡、崩塌灾害点分布在海拔 100～500m 的丘陵地 区，这与丘陵区坡度适宜，有利于形成较厚的堆积层 有关。结合梅州市 dem 图， 可得不同高程条件对山地 灾害的敏感性( 表 2)。由表 2， 对山地灾害最敏感的 是 0-300m 高程区域； 最不敏感的为>450m 的区域。 表 2 不同高程条件对滑坡灾害的影响 高程（m） <15 0 150-30 0 300-45 0 450-60 0 >60 0 灾害点个数（个） 94 205 68 10 7 所占比例（%） 24 53 18 3 2 4.2.2 坡度的影响 崩塌、滑坡的产生、发展及稳定过程，实质是斜 坡的抗滑力与下滑力失衡过程。对于某一个具体斜坡， 其形态、重度和结构是基本确定的，在这种情况下， 斜坡的稳定就主要取决于岩土体力学性质，如 c、φ 值大小。（c—粘聚力，是土体颗粒间吸附能力的体现； φ—内摩擦角，是土颗粒间错动能力的体现）。根据 对研究区斜坡体的调查发现，斜坡坡度受岩土体的物 理力学性质所控制，特别是岩土体的抗剪强度影响最 显著。在很大程度上，斜坡单元的坡度可以反映岩土 体力学性质。对研究区有坐标信息的灾害点所在的斜 坡单元的平均坡度分级进行统计对比，结果如表 3。 从表 3 中可以看出，最易发生滑坡和崩塌的斜坡坡度 为 2040°，占总地质灾害总数的 76.8%；在小于 15° 的斜坡上发生滑坡和崩塌较少。在斜坡单元内有坡 度 >60°的人工斜坡，多数以崩塌为主。 4.2.3 坡向的影响 坡向反映了山体斜坡所面对的方向。阳坡（或南 坡)与阴坡（或北坡)之间温度、降水和植被有一定差 异。由于光照、温度、雨量、风速、土壤质地等因子 的综合作用，坡向能够对斜坡稳定性产生影响。通过 对灾害点与坡向的相关分析表明(如图 1)：在坡向 0-112.5 o 共有地质灾害点 212 个，占所有滑坡点总数 的 35.15%；在坡向 112.5 o -292.5 o 共有滑坡点 220 个， 占总数的 36.48%，在坡向 292.5 o -360 o 共有地质灾害 点 171 个，占总数的 28.35%。可以看出，在红层风化 壳丘陵区，阴坡发生地质灾害的概率比阳坡大。 表 3 滑坡个数与斜坡坡度关系 斜坡度 （°） <15 15-19 20-24 25-29 30-34 35-39 40-45 >45 灾害点 数 7 49 94 132 155 82 56 28 百分数 （%） 1.2 8.1 15.6 21.9 25.7 13.6 9.3 4.6 4.3 降水的影响 红层风化壳的力学性质受水影响显著[]；由于风 化壳裂隙发育，遇雨后土体大量吸水饱和使负孔隙水 压消散，土体增重，强度显著降低，导致内摩擦角减 小，抗剪强度下降，在重力和地表径流作用下极易发 生土体失稳崩塌、倾覆而形成滑坡。在梅州市已确定 成因的 351 处地质灾害中，由降雨和人类工程活动因 素共同造成的有 247 处，占 75.1%；由纯降雨因素造 成有 46 处，占 14.0%；由纯人类工程活动因素造成的 有 36 处，占 10.9%；与洪水有关的地质灾害有 22 处， 占 6.2%。这里所指的人类工程主要是指开垦、建房、 修建公路等进行削坡而形成的斜坡。可见，与降雨有 关系的地质灾害占 89.1%。 4.4 人为活动的影响 华南红壤丘陵地区的人为活动以农业为主，主要 以开垦种植和修筑道路居民点，山坡地开挖而带来的 地形和地下水循环的改变，进而影响了斜坡稳定性。 本文分析了梅州市土地利用的垦殖指数和建设强度指 数，（垦殖指数指特定区域内已开垦种植的耕地面积 占区域面积的百分比。垦殖指数高，说明该地区的耕 地开发利用程度较高。建设强度指数指特定区域内已 建设用地面积占总面积的百分比。建设强度指数高， 说明该地区的开发建设程度较高）。结果发现梅州市 滑坡多发生在垦殖指数小于 0.3 的区域，占地质灾害 总数的 90%以上。这主要是由于垦殖指数小于 0.3 的 published by atlantis press copyright: the authors 113 区域主要为大于坡度 15°以上的地区，这些地区难以 开垦作为耕地，所以其垦殖指数一般较小，但一旦被 开挖就容易发生崩塌和滑坡等地质灾害。从建设强度 分析看，滑坡和崩塌等地质灾害多发生在建设强度指 数在 0.05-0.5 的坡地。影响地质灾害形成的工程活动 主要有削坡建房、修建公路、开垦等，尤以削坡建房 为甚。在全市工程活动因素造成的 601 处灾害隐患点 中，削坡建房造成的有 518 处，占 86. 12%：修建公 路造成的 64 处，占 10.64%。 5.结论与讨论 （1）红壤丘陵区风化壳滑塌是华南沿海地区常见 的一种地质灾害，由于该区独特的地质环境特点，使 其规模虽然一般较小，但其发生频率较高，常威胁丘 陵山区的农村生产生活。红层风化壳在华南沿海大面 积出露，其组分和原始的结构构造决定了它的风化壳 具有最大的厚度和最完整的剖面，红层风化壳构成了 华南丘陵区滑坡灾害的物质基础。 （2）红壤丘陵区的滑坡灾害以小型浅层滑坡为 主，灾害的发生具有明显的季节性，这种季节性与降 雨的季节性基本吻合，在空间上，主要分布于 300m 以下的坡度适宜、有较厚的风化层的低山丘陵区。 （3）华南红壤丘陵区滑坡灾害的形成与发生是多 种致灾因素交互作用的结果。红层风化壳的地层岩性 和岩土力学性质是灾害形成的内在要素，它在一定程 度上决定着地质灾害的发育程度和类型；地形是灾害 形成的外在条件，它制约着崩、滑、塌等致灾地质作 用的形成；大气降雨与人类工程活动是地质灾害形成 与发生的激发因素，决定着灾害发生的速度和时间。 致谢 本文受国家自然科学青年基金（41201557））、广 东省自然科学基金项目（s2011010004555）、广东省 科学院青年基金项目（qnjj201007）；广东省科技计 划项目（2010a030200020，2011a030200006）资助。 参考文献 [1] 吴树仁,田震远,吴锡浩,石菊松,雷伟志,范树应,胡江,孙 应举. 四川古蔺县蒿枝湾滑坡灾毁土地整理可行性研 究, 水文地质工程地质, (4)2002:21-23. [2] 张虎男.华南沿海风化壳滑塌灾害, 海洋地质与第四纪 地质, 15(3) (1995):149-160. [3] 王清,唐大维,张庆云.中国东部花岗岩残积土物质成分 和结构特征的研究.长春地质学院学报, 21(1) (1991): 73-81. [4] 钟继洪,唐淑英,谭军.南方山区花岗岩风化壳崩岗侵蚀 及其防治对策——以广东省五华县新一村为例.水土 保持通报,11(4)(1991):25-28. [5] 吴志峰,王继增.华南花岗岩风化壳岩土特性与崩岗侵 蚀关系.水土保持学报,14(2)(2000):31-35. [6] 何珊儒,魏国灵,张福林.东莞地区花岗岩风化剖面特征 研究.西部探矿工程, (10)(2005):77-79. [7] 陈明晓.广东沿海地区花岗岩风化剖面及其差异风化 现象的特征分析.广东公路交通, (1)(2003):59-63. [8] 陈海滨,陈志彪,陈志强.红壤侵蚀区花岗岩地质地貌对 水土流失的影响.亚热带水土保持, 22(4) (2010): 7-9+55. [9] 丘世钧.红土坡地崩岗侵蚀过程与机理.水土保持通报, 14(6)(1994):31-40. [10] 阮伏水,周伏建.福建省地貌特征与土壤侵蚀的关系初 探.水土保持通报,14(1)(1994):23-29. published by atlantis press copyright: the authors 114 microsoft word call for papers-infus 2021 special issue infus 2021 special issue of jracr on fuzzy techniques for risk analyses under emerging conditions call for papers october 25, 2021 emerging conditions such as pandemic, wars, natural disasters, and various high technologies force us for significant changes in our business and social life. pandemic caused all of us to live under quarantine for a certain period and serious restrictions in our business and social life. we clearly saw how important digital technologies are and how great the need for them is during this period. digital transformation is the adoption of digital technologies to transform services or businesses, through replacing non-digital or manual processes with digital processes or replacing older digital technology with newer digital technologies. this may enable in addition to efficiency via automation new types of innovation and creativity, rather than simply enhancing and supporting traditional methods. this special issue focuses on revealing the reflection of digital transformation in our business and social life under emerging conditions through intelligent and fuzzy systems. this special issue is on the theory and practice of fuzzy techniques for smart and innovative solutions. topics of interest include, but are not limited to, the followings: theoretical and/or practical developments of the followings for emerging conditions and digital transformation:  type-2 fuzzy sets  hesitant fuzzy sets  intuitionistic fuzzy sets  spherical fuzzy sets  picture fuzzy sets  pythagorean fuzzy sets  q-rung orthopair fuzzy sets  neutrosophic sets  fermatean fuzzy sets and integration of fuzzy sets theory with the followings for smart and innovative solutions:  bayesian networks  chaotic systems journal of risk analysis and crisis response issn print: 2210-8491 issn online: 2210-8505  combinatorial search  complex systems  distributed artificial intelligence  embedded systems  evolutionary systems  genetic algorithms  genetic programming  machine learning  multi-agent systems  neural fuzzy systems  neural genetic systems  neural network  pattern recognition  qualitative reasoning  quantum computing  reinforcement learning  support vector machines  swarm intelligence special issue important dates deadline for submission of papers: november 15, 2021 the end of the first round: january 15, 2022 the end of the second round: march 15, 2022 deadline for the acceptance / rejection decisions: april 15, 2022 guest editor prof. cengiz kahraman istanbul technical university department of industrial engineering 34367 macka, besiktas, istanbul, turkey e-mail: kahramanc@itu.edu.tr +90-212-2931300 (ext. 2035) atlantis press journal style risk management framework for the construction industry according to the iso 31000:2009 standard vitor sousa*, nuno marques de almeida and luís alves dias department of civil engineering, architecture and georesources, technical university of lisbon avenida rovisco pais lisbon, 1049-001, portugal *e-mail: vitor.sousa@ist.utl.pt abstract with stakeholders seeking for reducing the uncertainties in construction projects, the application of risk management has gained increased interest over the past decades, with a particular emphasis on the risk management process. however, very little attention has been given to the framework that supports the implementation of risk management processes within the construction sector. one of the most relevant particularities of the iso 31000:2009 standard1, on risk management principles and guidelines, is the promotion of a risk management framework overseeing the implementation of risk management processes. the present paper proposes a risk management framework for the construction industry, taking into due consideration the specificities of construction-related companies and construction projects seeking to integrate risk management within their overall management system. keywords: construction sector, construction-related companies, construction projects, management framework, risk management. 1. introduction according to carr2, successful implementation of risk management in organizations involves, among other issues, the prevention of:  a risk denial culture;  a lack of organizational support;  a deficient infrastructure supporting risk management. some organizations, when unprepared, tend to follow negative attitudes when facing risks. those who normally point these risks out are often considered barriers and obstacles in the operation and development of the organization. this risk denial attitude inhibits a proactive approach and weakens management. on the contrary, the management of risk is based on the recognition that uncertainty exists in most every-day decisions and on acknowledging that the effects of uncertainty on the organizations’ objectives should be explicitly considered in their decision making processes and activities. this protects (from threats) and enhances (the opportunities) the organizations goals and objectives. this uncertainty can arise either from the natural randomness of phenomena and/or from the limitations of knowledge in understanding and representing reality, both of which affect the capability to predict the future. the implementation of risk management requires an organizational support. such implementation should not be grounded in political, marketing or other motivations of the kind. organizations as a whole must be aware of the advantages and benefits of explicitly taking risks into account while performing their activities and also their decision making. this awareness will make it journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 261-274 published by atlantis press copyright: the authors 261 administrateur texte tapé à la machine received 29 may 2012 administrateur texte tapé à la machine accepted 14 august 2012 administrateur texte tapé à la machine sousa, almeida and alves dias possible to detect and understand threats and opportunities, allow risk-conscientious decision making and accumulate experience along with knowledge regarding uncertainties that can affect organizations. formal support from the organization top management is essential for a successful implementation of risk management, as it will enable the concepts and methods of risk management to be embedded at the various levels of the organization. complementarily, all aspects referred above should be effectively and efficiently framed, in order to prevent the development of independent and non-related risk management processes. framing these aspects ensures that risk management processes articulate the specificities of the organization, and that such processes are continuously improving with company experience and project realization. one of the most relevant differences between the iso 31000:2009 standard1 and the majority of the standards that reference risk management before its publication (with the exception of onr 49000:20043) is the consideration of a framework for risk management. this framework is to be embedded within the organization's overall strategic and operational policies and practices and includes a "set of components that provide the foundations and organizational arrangements for designing, implementing, monitoring, reviewing and continually improving risk management throughout the organization1". the foundations include a policy, objectives, mandate and commitment to manage risk. the organizational arrangements embrace plans, relationships, accountabilities, resources, processes and activities for managing risk1. this paper proposes a framework for the integration of risk management into the construction industry following the principles and guidelines of the iso 31000:2009 standard1. the proposed framework is generally applicable both at the company level and the project level. for convenience, the term organization is used to cover both levels, but distinctions between construction-related companies and construction projects are clearly identified. 2. a framework for managing risk 2.1. general definition a risk management framework links the management of risk with other management activities within the organization. therefore, such frameworks should be embedded in the overall strategy of the organization, as well as in its policies and practices. this is crucial for the success of risk management, as for the "effectiveness of the management framework providing the foundations and arrangements that will embed it throughout the organization at all levels1". in particular, the risk management framework should ensure1:  continual improvement;  full accountability for risks;  application of risk management in all decision making;  clear and continual communications;  full integration in the organization governance structure. a framework for managing risk should emphasize continual improvement of management in general, and the management of risk in particular, by setting performance goals and by measuring, reviewing and adjusting processes, systems, resources, capabilities and competencies. this focus on continual improvement is the major distinction between purely bureaucratic management and effective management. to this end, the assessment of the risk management performance should be an integral part of the overall performance evaluation system of the organization, covering both departments and individuals. selected performance parameters and indicators must represent the objectives of the organization and be compatible with the activities carried out at different levels. accountability for the risks, including the responsibility for their controls and treatment, must be set out in a comprehensive, fully defined and accepted way. to this end, the individuals with responsibilities should meet the required competencies and experience to identify, analyze and evaluate risk and detail risk treatment solutions. these individuals should have the resources needed to monitor risks, to check the controls, to implement treatment measures, and to report and communicate results of risk management. the organization should also delegate authority, grant the necessary time for risk management tasks and activities, as well as provide the required training for individuals to assume their responsibilities with success. all decision-making processes in the organization, regardless of the level of importance or relevance, must explicitly consider risks and the implementation of risk management. records shall be maintained demonstrating risks have been discussed in key decision published by atlantis press copyright: the authors 262 risk management framework for the construction industry: detailing the iso 31000 model making procedures, regardless if the discussion has been carried out in a free and non-structured fashion (e.g. in a meeting), or through the complete application of all steps of the risk management process referred in the iso 31000:2009 standard1. risk management should be a central element in the management of construction-related companies and construction projects. there are three perspectives to establish the structural relation between the organizations management and risk management, namely4:  the traditional option considers risk management as a component of the organization management, with top management delegating the responsibility for it (fig. 1 a);  another option is to consider risk management to be incorporated within the organization top management, so that risk management is not reduced to an administrative task (fig. 1 b);  another option is to consider risk management throughout all aspects of the organization management, including the possibility of delegating risk management task to external entities (fig. 1 c). the first perspective, somehow, isolates risk management as a separated additional task. the second perspective is more applicable for sectors such as insurance. for the construction industry, it is the authors’ believe that risk management should be integrated according to the last perspective. globally, the ultimate goals of risk management can be summarized as follows1:  provide the organization an updated, accurate and comprehensive knowledge regarding the risk in their activities;  ensure that the (level of) risk of the organization is consistent with the limits defined. fig. 2 presents the iso 31000:2009 standard1 risk management framework, including its relation with the risk management process. this model is being adopted by most of the national bodies developing or revising standards for risk management (e.g., bs 31100:20115). it is worth noting that the risk management framework proposed in the iso 31000:2009 standard1 follows the deming's cycle (plan-do-check-act: pdca). the nomenclature may vary depending on the context and the scope in which pdca cycle is applied (e.g., the six sigma model uses definition, measurement, analysis and improvement and control), but the meaning remains substantially identical. the risk management process is the operational component in the management of risk that enables the transposition of company goals into day to day project realization. this component has already been extensively discussed for projects in general8-11 and for projects in particular sectors, including construction projects12-16. also, most standards dealing with risk management (e.g., can/csa-q850-97:19976; jis q 2001:20017; as/nzs 4360:200417) that were developed before the publication of the iso 31000:2009 standard1 were focused mainly on the risk management process. however, little attention has been given to the framework needed to support the implementation of the process regularly and consistently. therefore, the present paper is mostly focused on the risk management framework. rm rm om rm om b)a) c) om fig. 1. accommodation of risk management (rm) into the organization management (om) published by atlantis press copyright: the authors 263 sousa, almeida and alves dias 2.2. mandate and commitment in organizational terms, the mandate and commitment are the guidelines of the institutional strategy that the management is committed to implement, given the various interested parties. the mandate and the commitment reflect the legitimacy and the objectives of the organization’s administration concerning the management of risk, while representing its stakeholders. the management of the organization must initiate risk management and also ensure its efficiency. this requires a strong and sustained commitment, as well as a rigorous design of a strategic plan to obtain involvement at all organization levels. in this regard, the management should1:  define and endorse the risk management policy, ensuring that it is compatible and complementary to other organization policies (e.g., quality policy, environmental policy, health and safety policy, social responsibility policy);  ensure alignment between organization's culture and risk management policy in order to facilitate implementation;  define risk management performance indicators that are aligned with organizational performance indicators, making it easier to interpret results and allow benchmarking studies;  align risk management objectives organizational objectives and strategies, ensuring compliancy with laws and regulations;  assign accountabilities and responsibilities and empower individuals at appropriate levels of the organization, ensuring adequate competences;  ensure that the necessary resources are allocated to risk management, providing the means for implementing the management of risks in operational terms;  communicate the benefits of risk management to all stakeholders, capturing their support and acceptance for risk taking;  ensure that the framework for managing risk continues to remain appropriate, adjusting to the changes in the surrounding environment and the organization dynamics. 2.3. framework design 2.3.1. understanding the organization and its context understanding the organization and its context includes the evaluation and understanding of "both the external and internal context of the organization1". this contextualization is intended to define, with a certain degree of accuracy, the nature of the challenges faced by the organization and the objectives to be attained. it aims at identifying external and internal parameters that influence the pursuit of the organization’s objectives and, therefore, must be taken into consideration while managing risks. the contextualization includes the definition of the external and internal context to which the organization is bounded and wherein risk management will be implemented (fig. 3). evaluating and understanding the external and internal contexts of the organization is the starting point for designing the risk management framework. a correct contextualization ensures that the resources allocated to risk management processes are design of framework for managing risk implementing risk management monitoring and review of the framewok continual improvement of the framework mandate and commitment risk assessment risk identification risk analysis risk evaluation establishing the context risk treatment m o n it o ri n g a n d r e v ie w c o m m u n ic a ti o n a n d c o n su lt a ti o n risk management processrisk management framewok fig. 2. risk management framework1 published by atlantis press copyright: the authors 264 risk management framework for the construction industry: detailing the iso 31000 model directed towards core issues. the implementation of risk management in a company, a department or sector, or in a project, a portion of a project or a project activity, asks for different preponderance and relevance of the internal and external context parameters. the definition of which internal or external parameters are relevant and what is their importance depends on the purpose and objectives set for the risk management. these parameters can be determined using the 6 w's model proposed by chapman and ward9. this model identifies 6 main questions, namely who (parties), why (motives), what (design), which way (activities), wherewithal (resources), and when (timing), to which the answers provide the basis for defining the internal and external contexts. 2.3.2. establishing risk management policy the iso 31000:2009 standard1 defines risk management policy as a “statement of the overall intentions and direction of an organization related to risk management”. the risk management policy should set out the organization's attitude towards risk and its approach to risk management20. also, the objectives and the organization's commitment to risk management must be clearly specified, namely regarding1:  the organization's rationale for managing risk;  links between the organization's objectives and policies and the risk management policy;  accountabilities and responsibilities for managing risk;  the way in which conflicting interests are dealt with;  commitment to make the necessary resources available to assist those accountable and responsible for managing risk;  the way in which risk management performance will be measured and reported;  commitment to review and improve the risk management policy and framework periodically and in response to an event or change in circumstances. two of the points mentioned above are crucial for the risk management framework. one, the connection between the objectives and policies of the organization and the risk management policy in order to avoid incompatibilities and the repetition of tasks, since risk is a concern transversal to the organization activity. this connection will serve as a base model for the integration of risk management in the organization, and this will be detailed next. the other, the measurement and reporting of the risk management performance, must be adequately adjusted to the way that risk management processes are intended to be implemented, in order to standardize the monitoring and reviewing of organization management. 2.3.3. integration into organizational processes for its efficiency, effectiveness and, also, for the sake of simplicity, risk management should be integrated with management systems or practices that already are familiar to organizations within the construction industry. most construction-related companies have already adopted management processes addressing quality, environment, occupational safety and health, social responsibility, and others, along with the current practices of cost, time and scope management. concerning the requirements laid down by standards or guidelines that organizations of the construction industry often use to develop environmental management systems (iso 14001:2004 standard21) and occupational safety and health (ohs) management systems (ilo-osh 200122 or ohsas 18001:2007 standard23), these already have risk management concepts explicitly embedded. despite the fact that the organization strategies policies resources capabilities decisions processes governance structure financial political economicenvironment tecnological social drivers trends stakeholders values stakeholders perceptions legal and regulatory cultural fig. 3. risk framework management contextualization1, 18, 19. published by atlantis press copyright: the authors 265 sousa, almeida and alves dias standard most commonly used by the organizations in the construction industry for implementing quality management systems (iso 9001:2008 standard24) does not make an explicit reference to risk management concepts, several of its requirements are indirectly related to risk management25. moreover, the international quality management standard that envisages the sustained success of organizations (iso 9004:2009 standard26) makes explicit use of harmonized risk terminology (iso/iec guide 73:2009 standard27) and it also explicitly states that "a quality management system is influenced by its organizational environment, changes in that environment, and the risks associated with that environment25". at a construction project level, it is worth mentioning that the iso 10006:2003 standard28 (an international quality management guideline applicable to projects) and the iso/fdis 21500 standard19 (an international project management guideline) incorporate a complete risk management process. the latter follows closely the project management body of knowledge (pmbok) guide model29. in both cases, risk management is always depicted as a process transversal to all areas of interest in the project. for example, the iso 10006 standard28 establishes that "risk identification should consider not only risks in cost, time and product, but also risks in areas such as product quality, security, dependability, professional liability, information technology, safety, health and environment". hence, risk management is applicable to the activities of construction-related companies and constructions projects alike. fig. 4 depicts a general model for integrating risk management into the overall management framework (through the risk management framework) and the management processes (through the risk management process) of construction organizations. this model was conceived to be particularly suitable for the construction industry. however, its general conception is grounded in well established management standards, models and practices and this enables it to be adopted for other industries and economic sectors. it articulates top management, supporting management processes and procedures (including those pertaining to formally established management systems), functional areas or departments in organizations and the inputs and outputs that are generated by the organizations. within this general model, the management framework is grounded in the pdca methodology which is usually adopted by the top management of construction-related organizations (following quality management principles). the integration of different management processes can be operated through communication and consultation, on the one hand, and monitoring and review, on the other. one of the relevant features of the proposed general framework is that the risk management policy should not only influence the internal practices and processes of construction-related companies, but should also become part of the linked construction projects. in other words, the proposed general framework stresses the importance of risk management being embedded into construction project management plans and management manuals of construction-related companies. the present paper is restricted to risk management and focuses in the risk management framework alone. further details about integrated management system for the construction industry can be found in the works of dias30 and almeida31. 2.3.4. accountability and resources at every level, from individuals (for personal or professional reasons) to the institutions (for social or commercial reasons), there are liabilities associated with the choices and decisions taken and/or the attitudes and actions adopted. regarding risk management, "the organization should ensure that there is accountability, authority and appropriate competence for managing risk, including implementing and maintaining the risk management process and ensuring the adequacy, effectiveness and efficiency of any controls1". this accountability for risk management has to be adequately supported by the necessary resources, both human and material, including time to perform the tasks and activities involved. following the guidelines which are laid down in the iso 31000:2009 standard1, the authors consider that it is particularly important for a construction-related company (see fig. 5.a): published by atlantis press copyright: the authors 266 risk management framework for the construction industry: detailing the iso 31000 model  to designate a risk administrator accountable for developing, applying, reviewing and updating the risk management manual as a result of implementing the risk management framework;  to choose risk coordinators accountable for ensuring the uniformity of the operational context and the compatibility of the risk criteria throughout the company and its projects. also, for construction projects, the following key players concerning risk management should be identified (see fig. 5.b):  risk managers accountable for developing, applying, reviewing and updating the risk management plan as a result of implementing the risk management process;  risk owners that have the accountability and authority to manage risks through the implementation of the risk treatment procedures/measures. regarding risk management, the link between the construction-related company and the construction projects is established through the communication between risk managers and risk coordinators. managing risks in the construction industry is intimately related with the allocation of people with the appropriate skills, experience and competence, with the provision of the resources needed for each step of the risk management process, and also with the selection of methods and tools that are compatible with the specificities of this industry. the degree of success in the development, implementation and maintenance of a risk management framework which is applicable to construction projects is directly related with the commitment and empowerment of the participating companies (e.g. designers and contractors), but above all with the commitment of the highest levels of the hierarchy of responsibility within every construction project, namely that of owners and their representatives. r is k m a n a g e m e n t p la n construction project strategic planning programming design construction use revalorization r is k m a n a g e m e n t p la n construction project strategic planning programming design construction use revalorization r is k m a n a g e m e n t m a n u a l construction-related companies owners and their representatives consultants designers contractors and sub-contractor suppliers construction-related companies r is k m a n a g e m e n t m a n u a l owners and their representatives consultants designers contractors and sub-contractor suppliers planning implementating monitoring and review continual improvement policy management framework secretariat technical • studies and designs • proposals • production administrative and financial • administrative services • treasury • accounting • human resources legal support public relations • publicity • marketing information systems research and development assets • operation / maintenance • rehabilitation / substitution commercial relations • aquisitions and sales • contract management functional areas risk assessment identification analysis evaluation establishing the context risk treatment m o n it o r in g r e v ie w c o m m u n ic a t io n c o n s u lt a t io n integrated management quality-related environment-related osh-related others risk management processes cost/time/scope-related r is k m a n a g e m e n t p la n construction project strategic planning programming design construction use revalorization r is k m a n a g e m e n t m a n u a l construction-related companies owners and their representatives consultants designers contractors and sub-contractor suppliers fig. 4. integration of risk management into construction organizations published by atlantis press copyright: the authors 267 sousa, almeida and alves dias within construction projects, a risk manager with the adequate capability is paramount for the implementation of a successful risk management plan. also, a risk owner is the person with the obligation to manage risks (or the state of its management) relating the area of the construction project over which it is accountable and exercises authority. risk owners are allies of the risk manager in the sense that their expertise and knowledge support the application of the risk management process, especially regarding the risk assessment (risk identification, risk analysis and risk evaluation) and risk treatment stages, to all activities of the construction project, within or outside their particular area of influence. fig. 5 suggests a risk management organogram for a construction-related company (fig. 5 a) and for a construction project (fig. 5 b). in the former, there can coexist several risk coordinators depending on the company dimension and internal structure (e.g., different risk coordinator for different construction project types or different project locations). the risk coordinators integrate the risk commission. regarding the later, the level 1 risk owners are usually those directing large processes of the construction project (e.g. financing, design, construction, etc.) and level 2 risk owners are usually hierarchically attached to level 1 risk owners. the risk team is formed by all level 1 risk owners. risk-related accountability may be allocated to different participants. tables 1 and 2 provide examples of how this allocation may be accomplished for different risk management tasks, respectively, in a construction-related company and in a construction project. namely, for a construction project, the authors suggest that:  the construction project manager should be accountable for approving and verifying risk management activities and related documents;  the risk manager should be accountable for the updates of the risk management plan and its associated documents, for validating the riskrelated activities of risk owners, for supporting the risk management team and for reporting the risk and the state of its management;  the risk management team should be accountable for supporting the construction project manager regarding risk-related activities, for assisting risk reporting activities and for issuing risk-related recommendations; administration risk coordinator a (projects) risk manager (project 1) …. risk manager (project n) risk coordinator b (departments) department manager (department 1) …. department manager (department n) risk administrator risk commission a) project manager risk owner 1 (level 1) risk owner 1a (level 2) …. risk owner ... (level 2) ….. risk owner n (level 1) risk owner ... (level 2) …. risk owner ... (level 2) risk manager risk team b) fig. 5. risk management organogram for a) a constructionrelated company and b) a construction project. published by atlantis press copyright: the authors 268 risk management framework for the construction industry: detailing the iso 31000 model  the risk owners should be accountable for the execution of risk management activities and riskrelated documentation (including for the activities and documentation generated by hierarchically dependent risk owners), for reporting threats and opportunities and for validating and implementing risk treatments activities. 2.3.5. communication and consultation in order to be effective, management generally requires a continuous flow of information to keep track of the dynamics affecting the organizations. according to the iso 31000:2009 standard1, the communication and consultation is a "continual and iterative processes that an organization [public, private or community enterprise, association, group or individual] conducts to provide, share or obtain information and to engage in dialogue with stakeholders regarding the management of risk". in the case of risk management, the communication and consultation of subjective aspects can be as relevant as that of objectives issues, in order to correctly characterize the full spectrum of the positions, values and interests of the various stakeholders, both internal and external. this is clearly a core issue for implementing risk management, for stakeholders in the construction industry range from authorities and official bodies, to owners and their representatives, banks and insurance companies, conformity assessment bodies, designers, contractors, subcontractors and suppliers, and, finally but definitely not the least, end-users16. moreover, these are stakeholders that can strongly affect, be affected by, or perceive themselves to be affected by a risk management decision or activity. in fact, risk perception is significantly influenced by risk information and by the manner by which it is conveyed32-36. as such, communication and consultation must ensure that relevant information is conveyed in a manner appropriate to its receptors, so as to avoid irrational reactions that may lead to losses or, which amounts to the same, to the impossibility of attaining expected benefits. for this reasons, faber37 highlights table 1. suggested allocation of risk-related accountability in construction-related companies. risk management components administration risk administrator risk coordinators risk management manual a e p establishing the context external context a e p internal context a e p operational context a v e risk criteria a v e risk treatment strategies/protocols a e p monitoring and review a e p a – approval; v – validation; e – execution; p – participation table 2. suggested allocation of risk-related accountability in construction projects. risk management components project manager risk manager risk owners risk management plan a e p risk identification detection a v e recognition and description a e p risk analysis a v e risk evaluation a e p risk treatment procedures/measures a v e monitoring and review a e p a – approval; v – validation; e – execution; p – participation published by atlantis press copyright: the authors 269 sousa, almeida and alves dias the importance of communication, and refers to it as a special means of risk treatment. cooper et al.10 also stress the importance of communication and consultation for the purposes of justifying the allocation of resource needed to implement risk management solutions or alternatives, ensuring that key stakeholders are satisfied with risk management and recording the performance of risk management. biocca38 has concluded that communication is particularly relevant for managing environmental and health and safety risks. the same author considers that communication is both a technique, a right and a condition. communication techniques must be adapted according to the scenarios in which they are used. successful communication and consultation depends on the capability of effectively and clearly transmitting or grasping an idea. from the organization point of view, an effective communication strategy must take into account the characteristics of the interlocutors, the legitimacy of the content and trustworthiness of the communication process. however, communication is a bi-directional process in which the challenge lies more in the ability to formulate or grasp the message correctly than in the ability to convey it. all interested parties have the right to be informed and should be given the possibility to express opinions that may influence the decision making. communication contributes to create a condition conducive to highly consensual decisions, and to ensure that all interested parties are duly informed about particular aspects and assumptions supporting the decision and know and understand the different points of view. the mechanisms and protocols for communicating and consulting within construction-related companies must support and encourage the responsibility and duties in managing risk. also, those should include tools and rules to obtain and consolidate the relevant information regarding risk management from different sources. communication and consultation within construction projects should take place during all stages of the implementation of a risk management process. it is convenient that a communication and consultation program is available at the earliest stages of the construction project. this program should cover all stages of the construction project (strategic planning, programming, design, construction, use and revalorization – see fig. 2) and should undertake a consultative team approach involving multiple stakeholders, which allows:  to establish the context of risk management appropriately;  the understanding and consideration of the interests of various stakeholders;  taking into account different views when defining risk criteria and in evaluating risks;  the identification of risks;  the merging of different areas of expertise for risk analyzes;  the execution, validation and approval of risk treatment plans;  enhanced change management during the risk management process; the communication and consultation program helps establishing more detailed and context specific communication and consultation plans. these plans should address issues relating to the risk itself, its causes, its consequences (if known), and the measures being taken to treat it1. the goals of the communication and consultation plans may vary, depending on the stage and on the type of stakeholders that are covered by it. communication and consultation should in all cases facilitate truthful, relevant, accurate and understandable exchanges of information, taking into account confidential and personal integrity aspects1. construction projects gather stakeholders with varying perceptions and attitude towards risk. this schema of communication and consultation allows stakeholders' perceptions to be identified, recorded, and it is important for supporting future applications of the risk management processes10. 2.4. implementation the implementation of risk management in the construction industry can be envisaged at two complementary levels: i) construction-related companies' level, and ii) construction projects level. regarding the first level, it is the authors believe that the construction-related companies will have to adhere to risk management principles in order to cope with risk-related demands of interdependent markets such as those of property, finance and insurance – markets that seek some kind of demonstration that the risk of the activities and the deliverables of the construction industry is being properly managed. this adherence does not necessarily imply the development and implementation of a “risk management system” published by atlantis press copyright: the authors 270 risk management framework for the construction industry: detailing the iso 31000 model within the construction-related company. instead, construction-related companies should integrate risk management into their existing overall management system, by adapting components of the iso 31000:2009 standard1 to their specific needs. construction-related companies should, therefore, implement risk management by1:  defining the appropriate timing and strategy for adapting risk management components;  developing and applying a risk management policy and process to existing organizational processes;  complying with legal and regulatory requirements;  ensuring that decision making, including the development and setting of objectives, is aligned with the outcomes of risk management processes;  holding information and training sessions;  communicating and consulting with stakeholders to ensure that its risk management components remain appropriate. regarding the second level, a successful risk management is dependent of the degree of commitment of construction project owners and their representatives. this is because they are in a privileged position to enable risk management to be relevantly, effectively and efficiently embedded into the construction project from its early stages. construction projects owners should, hence, ensure that a risk management plan is designed and implemented throughout the various phases of the construction projects they oversee, in order to establish the approach, the management components and the resources to be applied to manage risk. the authors suggest that risk management plans should include all information needed to support the implementation of risk management at the construction project level. these plans should be customized according with the specific context and particular objectives of each construction project, and should detail the applicable rules for establishing the context of the risk management process and for other core risk management activities such as risk identification, risk assessment, risk treatment, communication and consultation and monitoring and review. fig. 6 situates risk management plans within the hierarchy of risk documentation that is generated throughout the execution of a construction project. the risk register summarizes the construction project risk information regarding risk assessment (risk identification, risk analysis and risk evaluation) and risk treatment. it provides a ranked list of risk and may include prioritized actions or tasks that can then become the basis for risk treatment plans39. the purpose of the risk treatment plans is to stipulate how the chosen treatment options will be implemented for each risk or set of risks1. risk reports facilitate the exchanges of information between stakeholders about the risk of the construction project and the state of its management. these reports should take into account confidential and integrity aspects. these report may describe1:  the objectives and the scope of the risk management process;  a description of relevant parts of the construction project;  a summary of the external and internal context of the construction project;  risk criteria and their justification, limitations, assumptions and justification of hypotheses;  risk assessment methodologies;  risk identification results;  data, assumptions and their sources and validation;  risk analysis results and their evaluation;  sensitivity and uncertainty analysis;  critical assumptions and other factors which need to be monitored;  discussion of results;  conclusions and recommendations;  references. 2.5. monitoring and review alike other elements of overall management, risk management culminates with concrete proposals for amendment of policies, strategies, processes, projects, operations or activities. however, there are no guarantees that these proposals will be correctly implemented, that they remain adequate or that they risk management plan risk register risk treatment plans risk reports fig. 6. risk management documentation. published by atlantis press copyright: the authors 271 sousa, almeida and alves dias contribute to the achievement of the organization objectives. hence, it is crucial to undertake monitoring and review activities. the iso 31000:2009 standard1 defines monitoring as "continual checking, supervising, critically observing or determining the status in order to identify change from the performance level required or expected and review as the activity undertaken to determine the suitability, adequacy and effectiveness of the subject matter to achieve established objectives". monitoring and review can be applied at different levels (e.g. construction-related company level, construction project level) and may envisage different object (e.g., framework risk management manual, process risk management plan). at the construction-related company level, monitoring should cover the status of risk and its management and check if the level of performance complies or deviates from what is established in the risk management manual. on the other hand, review should determine the appropriateness, adequacy and effectiveness of the risk management manual to achieve the objectives laid down in the risk management policy. review should also promote the necessary changes, if necessary, towards continual improvement. at the construction project level, monitoring and review should be covered within the risk management plan and be a planned part of the risk management process. the risk manager should be accountable for both activities. the risk manager should collect and register the information generated by the risk owners, and report in an adequate manner to the project manager and to the risk team, which on their turn should provide feedback to the risk manager. reporting should cover issues such as the effectiveness and efficiency of existing controls, lessons from successes and failures, the need for changes in risk criteria or other aspects of the context, the need to revise risk treatments and priorities and new emerging risk. measurement of the adequacy of the risk management plan regarding the overall performance of the construction project should be encouraged (e.g. using proactive and reactive risk management key-performance indicators) and these results should also be used as an input for continual improvement of risk management activities. monitoring and review is crucial to overcome two major difficulties40:  risk management is, by its nature, orientated to the future (if a risk does not occur, it can be difficult, if not impossible, to show that was the result of the management's actions);  risk management, like many other forms of management, involves delegation of many management functions to the organizations being managed (if the organizations use different models or codes to appreciate risks, the interpretation and comparison of results may become impractical). the former, since it is natural to risk management, is extremely difficult to eliminate or mitigate, and it might only be possible to draw indirect observations in the long range through trend analysis or similar approaches that must be incorporated in monitoring and review procedures. the later can be significantly mitigated through the implementation of an efficient and effective monitoring and review procedure at a construction-related company level. these procedures must be extended to the construction project level. this has to be promoted by the owner and their representatives during project's inception. 2.6. continual improvement the suggested framework includes monitoring and reviewing activities that, alongside with communication and consultation, provide a solid ground for enhancing a risk management culture within the construction industry. the authors suggest that risk managers should be alert to identify and report opportunities for continual improvement of risk management components in order to mature its applicability to the construction industry, both at the construction-related company level and at the construction project level. 3. final remarks the risk management process is a core operational component of risk management, which has already been extensively discussed for projects in general and for construction projects in particular. in fact, there were already many standards on the risk management process before the publication of the iso 31000:2009 standard1. however, very little attention has yet been given to the framework needed to support the implementation of risk management processes within the construction industry. according to hillson41, organizations can be ranked into four increasing levels of risk maturity: i) naive; ii) novice; iii) normalized; and iv) natural. the published by atlantis press copyright: the authors 272 risk management framework for the construction industry: detailing the iso 31000 model implementation of a fully operational risk management framework is part of the requirements towards a natural risk organization, with a risk-aware culture and a proactive approach to the management of risk in all aspects of its activities. the present paper envisages promoting higher risk maturity levels for the construction industry by proposing a risk management framework applicable to the specificities of construction-related companies and projects. the framework proposal is grounded in the iso 31000:2009 standard1internationally accepted principles and guidelines. unlike most sectors, the construction projects involve companies with different core businesses and motivations (e.g., property investors, engineering companies) interacting with varying degrees of involvement throughout successive stages (e.g., design; build; operate). however, it is the authors’ belief that, from a project stand point, the successful implementation of risk management should be started and guided by the owner and their representatives, in order to capture the full spectrum of interested parties involved and focus the product performance on the end user demands. following this, the designers, the contractors and others, should then implement the risk management based on the owner’s guidelines, in order to ensure that the objectives of the end users are met. acknowledgements the authors gratefully acknowledge the support icist research institute, from ist. vitor sousa and nuno almeida were financially supported by fct, the portuguese national science foundation, through their phd grants sfrh/bd/35925/2007 and sfrh/bd/39923/2007, respectively. vitor sousa. thanks the fulbright commission for the award of a fulbright/flad grant supporting is research at ucdavis. nuno almeida acknowledges the calouste gulbenkian foundation grant supporting his research at ryerson university and at florida international university. references 1. iso 31000:2009, risk management: guidelines on principles and implementation of risk management (international organization for standardization iso, geneva, switzerland, 2009). 2. m. j. carr, risk management may not be for everyone, ieee software 14(3) (1997) 21-22. 3. onr 49000:2004, risk management for organizations and systems-terms and principles (austrian standards institute on, vienna, austria, 2004). 4. s. grey, practical risk assessment for project management (wiley, chichester, 1995). 5. bs 31100:2010, code of practice for risk management (british standards institute bsi, london, uk, 2010). 6. can/csa-q850-97:1997, risk management: guideline for decision makers (canadian standards association csa, toronto, canada, 1997). 7. jis q 2001:2001, guidelines for development and implementation of risk management system. (japanese standards association jis, tokyo, japan, 2001). 8. ice, ramp: risk analysis and management for projects. (institution of civil engineers ice, thomas telford, london, 1998). 9. c. chapman and s. ward, project risk management: processes, techniques and insights, 2nd edn. (wiley, chichester, 2003). 10. d. f. cooper, s. grey, g. raymond and p. walker, project risk management guidelines. managing risk in large projects and complex procurements. (wiley, chichester, 2005). 11. pmi, practice standard for project risk management. (project management institute pmi, 2009). 12. j. al-bahar and k. crandall, systematic risk management approach for construction projects, j. construct. eng. manag. 116(3) (1990) 533-546. 13. r. flanagan and g. norman, risk management and construction (blackwell scientific publications, london, 1993). 14. a. del caño and m. de la cruz, integrated methodology for project risk management, j. construct. eng. manag. 128(6) (2002) 473-485. 15. n. almeida, v. sousa, l. alves dias and f. branco, a framework for combining risk management and performance-based building approaches, build. res. inform. 38(2) (2010) 157-174. 16. n. m. almeida, v. sousa, l. alves dias and f. branco, engineering risk management in performance-based building environments: the case study of building structures, (2012, in submission). 17. as/nzs 4360:2004, risk management (standards of australia as, standards of new zealand nzs, sidney, australia, 2004). 18. a. lester, project management, planning and control: managing engineering, construction and manufacturing projects to pmi, apm and bsi standards, 5th edn. (elsevier, oxford, uk, 2007). 19. iso/fdis 21500:2012, guidance on project management (international organization for standardization iso, geneva, switzerland, 2012). 20. irm, airmic and alarm, a risk management standard (the institute of risk management irm, the association of insurance and risk managers airmic, the national forum for risk management in the public sector alarm, uk, 2002). published by atlantis press copyright: the authors 273 sousa, almeida and alves dias 21. iso 14001:2004, environmental management systems: requirements with guidance for use (international organization for standardization iso, geneva, switzerland, 2004). 22. ilo-osh 2001, guidelines on occupational safety and health management systems (international labour office ilo, geneva, switzerland, 2001). 23. ohsas 18001:2007, occupational health and safety management systems: requirements. (2007). 24. iso 9001:2008, quality management systems: requirements (international organization for standardization iso, geneva, switzerland, 2008). 25. e. avanesov, risk management in iso 9000 series standards, in international conference on risk assessment and management, 24-25 november, geneva, switzerland (2009). 26. iso 9004:2009, managing for the sustained success of an organization: a quality management approach (international organization for standardization iso, geneva, switzerland, 2009). 27. iso/iec guide 73:2009, risk management. vocabulary: guidelines for use in standards (international organization for standardization iso, geneva, switzerland, 2009). 28. iso 10006:2003, quality management systems: guidelines for quality management in projects (international organization for standardization iso, geneva, switzerland, 2003). 29. pmi, pmbok guide: a guide to the project management body of knowledge, 4th edn. (project management institute -pmi, pennsylvania, usa, 2008). 30. l. a. dias, integrated management system in construction projects imsincons (internacional social security association issa, construction section, paris: issa construction, 2004). 31. n. m. almeida, framework for risk-managed performance-based buildings, ph.d. thesis (department of civil engineering, architecture and georesources, ist technical university of lisbon, lisbon, portugal, 2011). 32. p. slovic, perception of risk, science 236 (1987) 280285. 33. e. roth, m. g. morgan, b. fishhoff, l. lave and a. bostrom, what do we know about making risk comparisons? risk anal.10(3) (1990) 375-387. 34. b. b. johnson, advancing understanding of knowledge's role in lay risk perception, risk-issues in health and safety 4 (1993) 189-211. 35. j. c. m. van eijndhoven, r. a. p. m. weterings, c. w. worrell, j. de boer, j. van der pligt and p.-j. stallen, risk communication in the netherlands: the monitored introduction of the ec post-seveso directive, risk anal. 14(1) (1994) 87-96. 36. d. okrent, risk perception and risk management: on knowledge, resource allocation and equity, reliab. eng. syst. safety 59 (1998) 17-25. 37. m. h. faber (ed.), risk assessment in engineering principles, system representation & risk criteria (joint committee of structural safety jcss, 2008). 38. m. biocca, risk communication and the precautionary principle, human and ecological risk assessment: an international journal 11(1) (2005) 261-266. 39. iso/iec 31010:2009, risk management: risk assessment techniques. (international organization for standardization iso, geneva, switzerland, 2009). 40. j. black and r. baldwin, really responsive risk-based regulation, law & policy 32(2) (2010) 181-213. 41. d. a. hillson, towards a risk maturity model, the international journal of project & business risk management 1(1) (1997) 35-45. published by atlantis press copyright: the authors 274 international journal of computational intelligence systems vol. xx(z); month (year), pp. xx–yy doi: 10.1080/xxxxxxxxxxxxxx; issn xxxx–xxxx online https://www.atlantis-press.com/journals/ijndc journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 134–144 doi: https://doi.org/10.2991/jracr.k.191024.003; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article a game theory approach for multi-agent system resources allocation against outside threats cheng-kuang wu1,*, xingwei hu2 1school of computer science and software engineering, zhaoqing university, zhaoqing road, zhaoqing city, guangdong province 526021, china 2information technology department, international monetary fund, 700 19th st nw, washington, dc 20431, usa 1. introduction emergency response systems (ers) and homeland security advisory systems (hsas) are both centralized multiagent systems (mass) which delegate multiple interacting agents to resist outside attacks. however, the effectiveness of these mass as a means to defend entire large-scale geographic regions is constrained by the available resources. the system administrator faces a density of agent deployment dilemmas, where the disposition of more agents easily leads to higher costs. these systems also lack specific measures for rational decision-making, and do not apply mathematical models to capture the interactions between attacker and defender. the administrator of a mas should have a tool to measure the strength of the attacks and the resistance capability of the response agents. by considering the utilities of moves available to the attacker and the defender, we can find a way to build a rating system for decision making [1]. game theory tools provide analytical techniques that are already applied in many other research areas, where multiple agents compete and interact with each other within a specific system. in most multiple agent interactions, the overall outcome depends on the choices made by all self-interested agents. the goal is to make choices that optimize the outcome [2]. game theory also provides general mathematical techniques for analyzing goal-conflict and cooperation between two or more individuals [3]. this study surveyed several game theoretic approaches to be applied when external attacks threaten the mas. these mass include ers, hsas, and security force reallocation systems. however, there are common problems related to extension and scalability that are often encountered [4]. when these game theory models are played by more than four players (i.e., agents), it is not easy to compute the nash equilibrium (n.e.) of the payoff matrix in a multi-dimensional game. because mass consist of multiple agents, these computations are complex and therefore pose a challenge for classical game theory. the model uses two steps for solving the growing amount of mas agent work. a framework is formed for deploying multiple agents using game theory, which analyzes the detected threats (or attacks) and allocates the resources of the mas as much as efficiently as possible. the game theory model is appropriate for analyzing the interactions of mas agents and to deal with the resource allocation problem (rap) [5,6]. however, when many agents face external attacks, the whole mas may suffer from limited resources to resist these attacks. the two-stage model applies the divide-andconquer idea to divide the scalability problem into two parts. first, a non-cooperative game is applied to model the conflict of goals between an attacker and a response agent, after which the external threat value (i.e., the strength of the attacker) is derived from the nash equilibrium. second, a cooperative game is used to solve the rap among mas agents, in particular, by increasing the number of agents. the threat values of all response agents are utilized to compute each agent’s shapley value. then an acceptable resource allocation of agents based on the expected marginal contribution creates a minimum set of resource deployment costs. a r t i c l e i n f o article history received 28 july 2019 accepted 25 august 2019 keywords multi-agent system external threat value resources allocation nash equilibrium shapley value a b s t r a c t this study proposes an integrated model for the deployment of multiagent resources for resisting outside threats. the proposed two-stage model applies the divide-and-conquer strategy to solve the resources allocation problem. first, the interactive actions between an external attack and a response agent are modeled as a non-cooperative game, after which the external threat value is derived from the nash equilibrium. second, the threat values of all response agents are utilized to compute each agent’s shapley value. then an acceptable resource allocation of agents based on their expected marginal contribution creates a minimum set of resource deployment costs. the experimental results show that our approach is feasible as a means to mobilize search and rescue resources from a non-affected district and to improve relief efforts against earthquake damage. the shapley value allocation approach proposed in this study; the percentage of resources allocation of districts is closer to death rate of each district than the proportional division of resources. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: shapleyvalue@hotmail.com https://doi.org/10.2991/jracr.k.191024.003 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:shapleyvalue%40hotmail.com?subject= c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 135 the proposed model is applied to reallocate search-and-rescue (sar) resources after a strong earthquake event, to maximize the capacity of the security response teams and minimize the number of emergency responders needed. the experimental results show that the model is feasible as a means of mobilizing sars from a non-affected district and to improve relief efforts against earthquake damage. 2. mas response region in general, the mas agent is regarded as a software agent although they could be robots, humans or human teams, or even a combination of human-agent teams. mass can represent self-organized and complex behaviors even when the individual skills of all their agents are simple [7]. in this study, various mass including intrusion detection system (ids), hsas, and ers are surveyed. the agents possess skills and offer emergency services to protect the mas from external threats. a response agent can be a node in the idss software, a response crisis unit comprised of human teams in hsas, or agencies which assist in dealing with any emergency, such as a detector, sar-group, security force agency, etc. figure 1 shows the interactions of an agent in the mas under external threats (or attacks). as the external attacker chooses its moves to attack the mas, the response agent uses their moves to resist the external attack. this study assumes that the mas architectures (e.g., ers or hsas) will face external threats, rather than internal attacks. each agent has a possibility of an attack that is an external threat to the mas. an attack is an event which happens in the mas. in contrast, a threat is not an actual event. therefore, we define an external threat as “a possibility of outside attack” created by an attacker who will attack the mas, or by an unexpected disaster, such as a strong earthquake. in this situation, each agent possesses detection techniques, raises alerts, and improves emergency responses against the external attacks (e.g., large earthquake attacks) (see figure 1). figure 2 shows the five interaction steps among sar agents faced with external attacks. (i) each agent faces a sudden attack in the external threat environment. (ii) the agent’s goal is to resist the external attack. he/she utilizes specific resources, such as fire fighters, rescue-groups, ambulances, equipment and statistical analysis to report and relieve damage. these specific techniques refer to the sar agent’s capability which controls and handles its resources. (iii) each agent reports to the coordinator (e.g., emergency operation center: eoc) about its own capability and resources as it resists the attack. (iv) the only goal in this environment is the optimization of mas resource allocation. after the coordinator measures the value of a threat to all agents and analyzes its decisions, he/she determines and commands all sar agents to mobilize and redistribute resources. figure 1 | behaviors and interactions between response agents and external attacks. figure 2 | an illustration representing the interactions of an sar agent with its external attacker and eoc. 136 c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 (v) according to the commands of the eoc, each agent begins to employ his/her own resources or make them available for someone else. in other words, some agents’ resources are redistributed to the critical agents so as achieve the goal of resisting serious attacks and discarding nonaffected agents. 3. related works several game theory approaches have been applied in which external attacks threaten a mas which could include ers, hsas, and security force reallocation systems. various approaches have been applied to model the interactions between attackers and defenders for computer network intrusion detection. burke [8] proposed a two-player game framework to model the information warfare between two players, an attacker and a system administrator. in his model, the mixed strategy equilibrium was used to construct a player’s optimal solution. the strategies and scenarios of the model are simple extensions of a few simulations to analyze how it would behave in real situations. other approaches also have been applied to model the interactions between terrorists and defenders for counter-terrorism purposes. paruchuri et al. [9] deployed a software assistant agent that aided police or other security agencies in randomizing their security schedules. they improved the security capability of anti-terrorism forces when dealing with multiple agents posing an intentional threat. the model maps the problem of check-point scheduling as a bayesian stackelberg game. the proposed model is faster than the multiple-lp method because it simply solves the mixed strategy n.e. however, bayesian concepts do not fully apply in real situations, and the model did not consider incomplete information regarding the type of adversary. the solution of the bayesian n.e. is an np computation. their model also encountered the many players problem, which involves the extension and scalability of the proposed model. alpcan and basar [10] proposed the utilization of cooperative and non-cooperative game theory concepts to address some of the basic network security tradeoffs. they constructed a sensor warning system and a security game for making various decisions, analysis and, ids control schemes. the sensor warning system generates a security risk value for the ids agent (i.e., sensor). this value could have various levels and the calculation is based on simple detection output. however, the strategy between the ids and the attacker was incomplete, because it did not consider the propagation attacks in a game, therefore their model could not satisfy the decisionmaking requirements for ids control. their proposed model consists of two independent schemes. our study combines their proposed non-cooperative game with a cooperative game to solve the rap in a mas. 3.1. why game theory? gupta and ranganathan [11] surveyed several optimization approaches such as genetic algorithms (ga), the bayesian search (bs) method, as well as random search (rs) and greedy allocation (gs) methodologies. these algorithms cannot be applied to the multi-crisis management problem because they do not provide individual rationality and some of the members of the population become dominant as the algorithm progresses. however, using game theory, scenarios that optimize multiple competing objectives can be modeled. game theory presents interactions between self-interested agents and analysis as to which strategies can be designed that will maximize the benefits of an agent in a mas. many of the applications of game theory have been to analyze the negotiation and coordination of multiple agents. game theory can be a useful tool for building future generations of mixed game theory and decision theory agents [3]. in a non-cooperative game, each player tries to utilize resources at minimum cost and the coordination is not enforced externally but is self-enforcing. all players optimize their decisions which maximize their payoffs in a non-cooperative game. n.e. is a solution concept for a non-cooperative game which identifies a prediction of the game outcome such that every player in the game is satisfied with respect to every other player [12]. n.e. provides a theoretical prediction of attack in a conflict situation, where individual mas agents suffer from external attacks. the interactions between an external attacker and a response agent can be modeled as a non cooperative game. the shapley value is a solution concept for cooperative games which computes the power index of an individual for cost allocation [13]. the cooperative game provides a suitable model for the design and analysis of response agent deployment, and it has been shown that the famous shapley value rule satisfies many nice fairness properties [14]. the shapely value also identifies a socially fair, good quality allocation for all agents. here, the individual fairness for each player is optimal and the average fairness of the mas is high. the social optimality property ensures that each player in the game receives the best utility for herself/himself and the complete mas. a power index in the form of the shapley value is applied to calculate the marginal contribution among agents and achieve a mutually agreeable division of cost for mas deployment [2]. 4. the proposed model there are two schemes applied in the proposed model. a simplified workflow chart describing the principles for optimal agent deployment is given in figure 3. in the first scheme, the interaction processes between the attacker (or attacks) and the defensive agent are modeled as a two-person noncooperative game (zeroor general-sum game), taking into account the interactions of security measures among agents and attackers. the proposed payoff functions utilize the external threat measures (such as defensive or offensive capability) for two players. the n.e. derived from these functions is calculated to assign a unique external threat value for the agent. then, the second scheme constructs a resource allocation game. the power index of the shapley value is applied to calculate the marginal contribution among mas agents, to find a mutually agreeable division of cost for deployment. this study revises owen’s method [15] to propose a shapley value formula. mas agents are grouped into coalition groups based on the threat levels to provide an acceptable mas deployment. the shapley value can be used to assure the derivation of a unique solution for each agent in the mas c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 137 resource sharing interactions. let z h h h h i h= { }w w w w1 2 3, , , ..., be the minimum set of response agents, "i î n, and n is the number of agents, which is subject to the threat level h. finally, we obtain the appropriate shapley value vector and compute the number of reallocation resources of each agent. these optimal reallocation resources of the agents in the emergency response enable us to mitigate the damage from external attacks. 4.1. the external threat game the first game models the interactions of an attacker (or attack) and a response agent. when an agent confronts an external attacker, both face a competitive situation. the relationships between external attacker and response agents include pure and impure conflicts, and their behaviors are non-cooperative. thus, in the first stage, a non-cooperative game is applied to model decisions and to analyze the best responses. the two-player non-cooperative game is defined as follows: (i) the set of players a: the first model only has two players. the attacker is player 1 and the response agent is player 2. a = {attacker, response agent}. ( ii) strategy space si: a set of all possible strategies available to two players in a game. (iii) payoff pi: the expected utility to a player as a function of the strategies chosen by an attacker or response agent. let g = <a, (si), (pi)> be such a normal form game. sometimes this game cannot determine a pure n.e. strategy, because there is probably no n.e. however, every finite normal form game has a mixed-strategy n.e [16]. thus, this study can also derive another strategic game from g, called the “mixed extension” of g, in which the set of actions ai of each player i is the set of mixed strategies in g. more generally, suppose that player i has k pure strategies: si = {si1, …, s1k}. then, a mixed strategy for player i has a probability distribution (pi1, pi2, …, pik), where pik is the probability that player i will play sik, for k = 1, …, k. since pik is a probability, this study requires 0 ≤ pik ≤ 1 for k = 1, …, k and pi1 + … + pik = 1, where pi is used to denote an arbitrarily mixed strategy from the set of probability distributions over si, just as si is used to denote an arbitrary pure strategy from si. let j denote the number of pure strategies in s1 and k indicate the number in s2. now, s1 = {u1, u2, …, uj} and s2 = {d1, d2, …, dk} are rewritten using uj and dk to denote pure arbitrary strategies from s1 and s2, respectively. a j × k payoff matrix is created for the external threat game based on the two players’ strategies and interactions (see table 1). if player 1 believes that player 2 will play the strategies {d1, d2, …, dk} with the probability q = (q(d1), q(d2), …, q(dk)) then player 1’s expected payoff from playing the pure strategy uj is [equation (1)] q d u dkk k j k( ) ( , )=∑ 1 1p (1) and player 1’s expected payoff from playing the mixed strategy r = (r(u1), r(u2), …, r(uj)) is [equation (2)] m r q r u q d u d r u q jj j k j kk k jk k j j 1 1 1 11 ( , ) ( ) ( ) ( , ) ( ) =     = = = == ∑ ∑ ∑∑ p1 (( ) ( , )d u dk j kp1 (2) where r(uj), q(dk) is the probability that player 1 plays uj and player 2 plays dk. thus, 0 ≤ r(uj), q(dk) ≤ 1 for k = 1, …, k and j = 1, …, j; r(u1) + r(u2) + … + r(uj) = 1 and q(d1) + q(d2) + … + q(dk) = 1. figure 3 | flowchart of the proposed model. 138 c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 player 1’s expected payoff from the mixed strategy r, given in equation (2), is the weighted sum of the expected payoff for each of the pure strategies {u1, u2, …, uj}, given in equation (1), where the weights are the probabilities r(u1), r(u2), …, r(uj). thus, for the mixed strategy r(u1), r(u2), …, r(uj) is the best response for player 1 and 2’s mixed strategy q(dk), and it must be that r(uj) > 0 only if [equation (3)] q d u d q d u dkk k j k kk k j k( ) ( , ) ( ) ( , )= = ′∑ ∑≥1 1p p1 1 (3) for every uj′ in s1. that is, for a mixed strategy to be the best response to q it must put a positive probability on a given pure strategy only if the pure strategy is itself the best response to q. conversely, if player 1 has several pure strategies that are best responses to q, then any mixed strategy that puts all its probability on some or all of these pure-strategy best responses (and zero probability on all other pure strategies) is also the best response for player 2 to q [17]. this study computes player 2’s expected payoff when players 1 and 2 play the mixed strategies r and q, respectively. if player 2 believes that player 1 will play the strategies (u1, u2, …, uj) with the probabilities (r(u1), r(u2), …, r(uj)), then player 2’s expected payoff from playing the strategies (d1, d2, …, dk) with the probabilities (q(d1), q(d2), …, q(dk)) is [equation (4)] m r q r u j q d k u d r u j j j j kk k k k j 2 1 1 11 ( , ) ( , ) ( , ) ( , ) ( , ) =     = = = == ∑ ∑ ∑ p2 jj j kq d k u d∑ ( , ) ( , )p2 (4) given m1(r, q) and m2(r, q) this study restates the n.e. requirement that each player’s mixed strategy is the best response to the other player’s mixed strategy: for the pair of mixed strategies (r*, q*) to be a n.e., r* must satisfy [equation (5)] m r q m r q1 2( , ) ( , ) * * *≥ (5) for every probability distribution r over s1, and q * must satisfy [equation (6)] m r q m r q2 2( , ) ( , ) * * *≥ (6) for every probability distribution q over s2, where r * and q* represent the optimal mixed strategy for the attacker and response agent respectively. a mixed strategy n.e. is similar to a stochastic state that predicts the outcome of a game and it can capture stochastic regularity. the attacker and agent have information about their payoffs from actions which were taken in the past; each player applies these payoffs to form his belief about the future behaviour of the other players, and thereby formulate his optimal mixed strategy. in table 1, player 2’s expected payoff is computed when players 1 and 2 play mixed strategies r and q respectively. if the game g has no pure strategy n.e., a mixed n.e. pair (r*, q*) exists in this game, which is an optimal strategy [18]. the mixed n.e. for the probability vector is r* = {r*(u1), r *(u2), ..., r *(uj)} with actions {u1, u2, ..., uj} for the external attacker and the vector q * = {q*(d1), q *(d2), ..., q*(dk)} with actions {d1, d2, ..., dk} for the response agent. the external attacker’s expected payoff for a n.e. (pure or mixed strategy) is defined as its external threat value. player 1 (i.e., the external attacker) obtains the positive payoff (+), because player 1 gains a profit from player 2’s resource responses. player 2 (i.e., the response agent) obtains a negative payoff (–), which means that player 2 pays for player 1’s attack. when the response agent (i.e., player 2) gets a low expected payoff, the external attacker (i.e., player 1) gets a high expected payoff and the threat of external attack is high. because player 1’s expected payoff always represents a gain (+), this study defines the vi as the ith response agent’s threat value, given by [equation (7)] v p q p u q d u d u di jk k j j k j k j k= = == ∑∑p p1 1 n.e( , ) ( ) ( ) ( , ), ,* * * * * * * *11 ∈ .. (7) therefore, vi is derived from the attacker optimal strategies’ expected payoff which represents the threat value in the first game model. the next proposed model uses the threat value vi to compute the shapley value of each agent within the cooperative game. 4.2. resource reallocation game in this study, it is assumed that a central planner of the security of center (soc) or emergency opera tion center (eoc) can command and control mas resource allocation [19]. each agent reports its degree of threat (or threat value) to the central planner. the central planner computes the shapley value based on the threat values of all agents in order to command agents to dispatch resources. in this second game, the interactions of all response agents in the mas response region are likened to the playing of a cooperative game. an efficient method is needed to decide the number and priority of the deployment of resources to various response events when multiple attacks occur simultaneously. the shapley value is applied to create an optimal allocation of resources for the soc commander during multiple attacks. we utilize the concept of the majority coalition in a party voting game to deal with mas resource reallocation. a majority of voters can pass any bill in the majority game. the power of a voter will depend on how crucial that voter is to the formation of a winning coalition [15]. the shapley value can table 1 | a payoff matrix for the external threat game response agent d1 d2 ... dk r-mix attacker u1 p1(u1, d1), p2(u1, d1) p1(u1, d2), p2(u1, d2) ... p1(u1, dk), p2(u1, dk) r(u1) u2 p1(u2, d1), p2(u2, d1) p1(u2, d2), p2(u2, d2) ... p1(u2, dk), p2(u2, dk) r(u2) ... ... ... ... ... ... uj p1(uj, d1), p2(uj, d1) p1(uj, d2), p2(uj, d2) ... p1(uj, dk), p2(uj, dk) r(uj) q-mix q(d1) q(d2) ... q(dk) c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 139 provide a measure of the power of each voter, which is represented by the power index. when the sum of external threat values of some attacks passes the threshold of the majority level, the formation of a winning coalition is enabled, and the power index of each attack can be computed. we define y: v → r+ as a one-to-one function by assigning a positive real number to each element of v (i.e., external threat value) and y(0) = 0, v = {v1, v2, …, vi}, i ∈ n. the majority of the attack level hh is derived from a majority of all external threat values which represents the corresponding threshold value mh. the allocation of the response agent resources is based on the concept of the majority of external threat values. given the output vector of all external threat values of attacked events, the majority level is hh, if the sum of the external threat values of attacks is greater than or equal to mh [equation (8)] h m mh k hk n hif u u u u ≥ =∑ = + −   1 2 , min max min (8) the external threat values of attacks can be grouped into the majority level hh. this is divided by 2 from the maximum value vmax to minimum value vmin. all response agents can be modelled as an n-person game with x = {1, 2, …, n}, which includes a set of players (i.e., response agents) and each subset v ⊂ x, and where vj ≠ 0, ∀j ∈ v is called a coalition [2]. the coalition of x response agent groups in the mth threshold of the threat level, and each subset of x (coalition), represents the observed threat pattern for different threat levels h. the aggregate value of the coalition is defined as the sum of the threat values of the response agent, y(c) = ∑i∈c vi and is called a coalition function. each response agent coincides with one or another given m thresholds of the threat level. therefore, the different priorities for the response agent deployment are derived from the thresholds. based on the external threat for each mas response agent with respect to others, and the effect of the threshold values on various threat levels, the shapley value represents the relative importance of each response agent. now let y(c) = ∑i∈c vi vi ∈v, c ⊂ x be the value of coalition c with a cardinality of c. the shapley value of the ith element of the response agent vector is defined by [equations (9) and (10)] w( ) ( )!( )! ! [ ( ) ( { })]i c n c n y c y c ic x i c n = − − − −⊂ ∈ ∑ 1 (9) þ = ¢ì î ¢ åw( ) ( )!( )! ! i c n c n c x i c n 1 (10) equation (9) can be simplified to equation (10), because the term y(c) − y(c − {i}) will always have a value of 0 or 1, taking the value 1 whenever c′ is a winning coalition. if it is not a winning coalition, the terms c − {i} and y(c) are 0 [15]. hence, the shapley value is w(i), where c′ denotes the winning coalitions with ∑ i∈c vi ≥ mi. the shapley value of the ith response agent output indicates the relative threat value for the thresholds mh (i.e., threat levels). therefore, a shapley value vector shows the strength of all external attacks. we can use all shapley values to compute the reallocation amount of emergency response resources of the agent. the numbers of resources of type k reallocated to the ith agent are defined by [equation (11)] e i i o i k nk k( ) ( ) ,,= × ∈w all (11) where oall, k is the total number of resource available of type k in a mas (such as ambulances when an earthquake occurs). the reallocated numbers of the ith agent’s resources ek(i) are derived from the shapley value of the ith response agent w (i) multiplied by the total number of resource available of type k oall, k. finally, the commander can reallocate any kind of agents’ resources so as to fairly deploy resources to resist attacks in multiple attack regions. 5. sar response case the september 21, 2009 earthquake claimed 2297 fatalities, according to the national fire agency and ministry of the interior of taiwan. almost 90% of these victims died within 1 day of the event and 56% died within 6 hours (nearly 1284 victims) [20]. as we know, the key point to minimizing the total number of fatalities after an earthquake is the deployment of sar efforts in the first few hours. in an emergency response region, there are usually a number of alarm districts-each equipped with resources that must be able to be deployed quickly and effectively after any kind of emergency. however, the sars (e.g., fire fighters, rescue-groups, ambulances, and equipment) available to some affected districts may be limited. a devastating earthquake brings with its multiple disasters such as fire, building collapse and a combination of fire and collapse. in cases of multiple emergencies, each affected district might possess different response capabilities. some form of the central emergency management system must be present in order to allocate response resources. this study focuses on the problem faced by the emergency manager: which affected districts are the major ones and which ones are minor? which affected districts require first aid and to need more sar resources? this is a deployment efficiency problem. an emergency manager must find an optimal resource allocation for assigning response resources in space and time to the affected districts and improve the relief efforts against earthquake attacks. we apply the proposed model to solve the sar resources allocation problem. two games are constructed, representing the two stages needed for the economical deployment of available resources. one is the earthquake damage game which calculates the value of the earthquake threat in each district. another is the sar resource reallocation game which efficiently mobilizes the available resources of the response agent to increase the survival rate of earthquake victims. 5.1. the model assumptions certain assumptions must be made when treating a multiemergency and response management scenario as a game theory framework. the assumptions are listed below. (i) after the deadliest quakes, emergencies will occur in different districts in a time-overlapped manner, and resource requests will be simultaneously received by the central response agent managers (see table 2). we assume the multi-emergency event is player 1 and the district response agent is player 2. hence, two players can use a static single step. each player plays a non-cooperative and a zero-sum game. a response agent manages one district. three types of earthquake emergencies (i.e., fire, house collapse, and fire and house collapse) could 140 c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 table 2 | emergency priorities, types, and the ratio of four resource requests rescue groups ambulances fire engines machine and equipments priority (1–3) fire 0 1 2 0 1 collapse 3 2 0 0.3 2 fire and collapse 3 3 2 0.5 3 occur simultaneously when evaluating the value of the earthquake threat in a specific time period. (ii) the eoc has developed a training program to improve the emergency response agents’ capabilities. as seen in table 3, the annual emergency training gives us the time needed to perform search and rescue activities from each of the resource sites. these data are generated by the training program through annual earthquake drills or exercises. (iii) the first model is a normal form game for capturing the interactions between two players. this model assumes complete information; that is, every player knows the available strategies of every other player. each district has a different number of resources available. in multiple emergencies, there is also complete information regarding the resource requirements, emergency priorities and the population density of the affected district (see tables 2 and 4). 5.2. earthquake damage game this game applied the first scheme of the proposed model, which is designed to obtain the threat value of the earthquake in each district. we assume that the multi-emergency event is player 1 and the district response agent is player 2. the interactions between multi-emergency event and response agent are modeled by a two-person game. it is assumed that both players are forming a set of non-cooperative players i = {i1, i2}, where i1 generates multiple emergency events; and i2 is the district response agent responding to allocate finite sar resources, obtained by prioritizing the different districts. the parameters for determining the threat measures are defined in the following paragraphs. 5.2.1. multi-emergency events in the earthquake damage game, we assume the multi-emergency event to be player 1; three emergencies could happen simultaneously as a result of the player’s actions. s1 denotes the set of strategies available to player 1: s1 = {u1, u2, u3} = {fire, house collapse, and fire and house collapse}. the greater the seriousness of the emergency events, the more resources are needed. w denotes the set of resource requirements for multiple emergencies in one urban region. w = {w1, w2, ..., wn}. the variable wj,k denotes the number of resources required by uj emergency event from resource dk. moreover, the priority for each emergency is related to the resource requirement. typically, in a priority-based system, a high priority emergency needs more resources, and a lower priority emergency needs fewer resources. each event is assigned a priority p on a scale of 1–3 indicating the severity of the emergency, which is used as a weight in the payoff function to facilitate the calculation of losses. pj denotes the priority of the jth emergency: pj = 1, 2, 3, where 1 is the lowest level and 3 is the highest. an emergency event on a densely populated target would increase the number of fatalities and thus would gain greater benefit for player 1, while the response agent must pay a higher rescue cost. bi denotes the population density of the ith district in the whole region. table 3 | average times needed to perform search and rescue activities from the resource sites in ith district rescue groups ambulances fire engines machine and equipments fire 5 5 5 5 collapse 5 5 5 5 fire and collapse 10 10 10 10 total 20 20 20 20 these average times (or min) are measured by the annual earthquake exercise. table 4 | population density and resource available in 15 disaster districts districts the number of available resources population density rescue workers ambulances fire engines machine and equipments a1 taipei 94 350 250 600 100 a2 taoyuan 100 100 100 200 60 a3 hsinchu 37 50 60 100 50 a4 miaoli 19 25 40 50 45 a5 taichung 73 125 100 250 65 a6 changhua 74 75 50 150 65 a7 nantou* 8 25 50 50 45 a8 yunlin 34 25 50 50 45 a9 chiayi 25 30 60 70 47 a10 tainan 52 80 80 180 60 a11 kaohsiung 57 125 200 250 65 a12 pingtung 19 50 40 100 50 a13 taitung 4 20 20 30 44 a14 hualien 4 20 20 30 44 a15 yilan 13 25 20 40 45 total 613 1125 1140 2150 830 *a strong earthquake occurred. c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 141 5.2.2. response agent each alarm district has different response agents, which in turn have varied resources to prepare for multiple emergencies. in the earthquake damage game, the response agent is player 2, who holds four resources (i.e., four strategies). s2 denotes the sets of strategies player 2 has access to: s2 = {d1, d2, d3, d4} = {rescue worker, ambulance, fire engine, and machine and equipment}. o denotes the set of resources available to response agents in one district: o = {o1, o2, o3, o4}. ok denotes the number of resources available at resource dk, and tj,k denotes the time for resource dk to reach the uj emergency; k ∈1, 2, 3, 4. we assume these times are measured based on the annual emergency drill or exercise. two players will simultaneously make their strategic decisions. a 3 × 4 payoff matrix for the earthquake damage game is created based on two players’ strategies and interactions as seen in table 5. the payoff to player 1 for choosing a particular strategy when player 2 makes his selection can be represented as a gain by player 1(+) or a loss for player 2(–). in this model, a summation of the losses of player 2 is depicted, and player 1 tries to maximize this loss. player 2 tries to minimize the losses. player 1 gains a profit from player 2’s effort of resource responses. player 2 pays as a result of player 1’s multi-emergency events. we assume that each player’s aim in the game is to achieve as high a payoff for player as possible. the payoff for the jth strategy for player 1 when player 2 chooses the kth strategy to the response can be formulated as [equation (12)] p b1 all = −       × ×== ∑∑ w o w t pi k k j k kj j k i j , , , ,1 4 1 3 (12) where oall, k is the total number of sar resources of type k; wi,k indicates the total number sar resources needed of type k by the ith district. the proposed model assumes that the non-cooperative game is a zero-sum game; therefore, the payoff function of player 2 is given by [equation (13)] p b2 all = − −       × ×== ∑∑ w o w t pi k k j k kj j k i j , , , ,1 4 1 3 (13) a mixed nash equilibrium pair (r*, q*) exists in the normal form game if this game has no pure strategy nash equilibrium, which is an optimal strategy. the mixed n.e. for the probability vector is r* = {r*(u1), r *(u2), r *(u3)} with actions {u1, u2, u3} for the multi-emergency event and the vector q* = {q*(d1), q *(d2), q *(d3), q *(d4)} with actions {d1, d2, d3, d4} for the response agent. now, the proposed model lets vi be the ith threat value of a strong earthquake in a district which is the expected payoff of player 1. vi is computed by equation (7). u p p i j kk j k j kj p q p u q d u d u d = = == å 1 11 3 1 3 ( , ) ( ) ( ) ( , ), , * * * * * * * * în.e.åå (14) 5.3. reallocation of sar resources game this game applies the second scheme of the proposed models as explained in section 4.2. we assume that the 15 sar districts of taiwan are modeled as a 15-person cooperative game (see figure 4). the threshold value mh is generated by equation (8). given the mh threshold of the majority value, we can compute the majority of coalitions which are pivotal to each district. the different priorities for response resource deployment are derived from the majority of coalitions. based on the earthquake threat value for each district with respect to the others, and the impact of the threshold values on the majority of emergency level represents the relative importance of each district. equations (9) and (10) are used to compute the shapley value for each agent and the amount of reallocation of each agent’s resources in section 4.2. according to sar resource reallocation, the eoc mobilizes resources from all districts to assure that even rescue of victims in multiple emergency events across the whole region is possible. 5.4. simulation experiments the simulation takes advantage of the 921-earthquake data to verify the proposed model. we hypothesize that a strong earthquake occurs as it did in jiji, nantou county, taiwan. the eoc deploys 15 response agents from the district (or county) in the whole region (from a1 to a15) and provides central management and monitoring for homeland security of taiwan (see figure 4). in the first stage, the information in tables 2–4 is used, regarding the resource types and the availability, population density, emergency requests, emergency priorities, and time taken to reach the crisis area from each of the resource centers. each district’s emergency resource request is derived from the number of emergency calls (i.e., 119) multiplied by the ratio of the resource requests (see tables 2 and 6). we model the non-cooperative game and generated simulation sets of threat measures for the strong earthquake and the response agent in each district. the payoff matrix of the zero-sum earthquake damage game for each district is modeled according to equations (12) table 5 | a payoff matrix for the external threat game response agent d1(rescue groups) d2(ambulances) d3(fire engines) d4(machine and equipments) multi-emergency events u1(fire) w o w t pi , , , , 1 1 1 1 1 1 1 all −       bi w o w t pi i , , , , 2 2 1 2 1 2 1 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 3 3 1 3 1 3 1 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 4 4 1 4 1 4 1 all æ è çç ö ø ÷÷ b u2(collapse) w o w t pi i , , , , 1 1 2 1 2 1 2 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 2 2 2 2 2 2 2 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 3 3 2 3 2 3 2 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 4 4 2 4 2 4 2 all æ è çç ö ø ÷÷ b u3(fire and collapse) w o w t pi i , , , , 1 1 3 1 3 1 3 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 2 2 3 2 3 2 2 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 3 3 3 3 3 3 3 all æ è çç ö ø ÷÷ b w o w t pi i , , , , 4 4 3 4 3 4 3 all æ è çç ö ø ÷÷ b 142 c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 figure 4 | 15 search and rescue (sar) districts of taiwan. and (13) and table 5. then, 15 earthquake threat values are calculated using equations (12–14). table 6 shows 15 earthquake threat values of the response agent derived from nash equilibrium points by enumerating the strategies in the normal form game. in the second stage, all earthquake threat values are utilized to compute the shapley value for each district. given the vector output for all the values, we use equation (8) to compute the majority of the threat values; the threshold value (i.e., mh) is 310. according to this threshold value 15 shapley values of the district are calculated using equation (10). the results obtained are shown in table 6. assume that all sar resources available are unable to satisfy the requirements for all the earthquake-related emergencies occurring in taiwan. table 4 provides the experimental number of total available sar resources. according to these experimental data, the number of reallocated sars for each district can be computed using equation (11). the results obtained are presented in table 7. 5.5. discussion we analyze the application of the proposed model to determine whether the relief effort of sar resources is improved after the 921-earthquake or not. assume specific sar resources for the entire taiwan region. when the 921-earthquake occurred, the results calculated by our model guided the eoc to mobilize resources from all districts to assure even rescue of victims in multiple emergency events across the whole region (see table 7). the number of people killed in all districts during the 921-earthquake is revealed by the ministry of the interior of taiwan statistics (see table 7). central taiwan suffered major damage, particularly nantou county and taichung county. the majority of the casualties were concentrated in these two districts. table 7 is interesting; because the proposed model dispatches most sar resources to the two most severely affected districts. in addition, it shows the average distribution table 6 | fifteen examples for computing earthquake threat values and shapley values districts number of emergency calls requirements earthquake threat values shapley valuesfire collapse fire and collapse rescue groups ambulances fire engines machine and equipments a1 taipei 5 50 40 275 225 180 46 160 0.086353 a2 taoyuan 5 10 10 65 55 50 11 80 0.036489 a3 hsinchu 8 16 20 116 100 92 19.6 53 0.022389 a4 miaoli 8 20 30 158 138 126 26.6 37 0.015335 a5 taichung 30 200 200 1230 1030 860 206 618 0.334121 a6 changhua 5 150 150 605 505 410 101 89 0.040354 a7 nantou* 20 300 350 1990 1690 1430 333 610 0.334121 a8 yunlin 4 100 100 604 504 408 50.4 129 0.062895 a9 chiayi 10 10 20 200 170 160 17 31 0.013536 a10 tainan 8 8 16 212 196 188 19.6 75 0.034316 a11 kaohsiung 4 4 8 40 36 36 3.6 15 0.005728 a12 pingtung 2 2 20 68 66 66 6.6 23 0.011142 a13 taitung 2 2 10 38 36 36 3.6 3 0.001931 a14 hualien 2 2 10 38 36 36 3.6 3 0.001931 a15 yilan 5 5 5 35 30 30 3 4 0.00129 total 118 879 989 2229 1 *epicenter. c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 143 rate (adr) for nantou county (i.e., 33.4%) which almost matches the death rate (i.e., 36.91%). therefore, the proposed model is able to improve relief effort to counteract damage from external attacks. we also evaluated the proposed division of resources by comparing the results obtained with those from the proportional division. the epicenter of the 921-earthquake was in nantou county (i.e., agent a7) where the seismic intensity measured 7 on the richter scale. different magnitudes were measured at other counties in taiwan when the 921-earthquake occurred. we assume that the higher the magnitude in a district, the more sars resources needed. therefore, we can use the proportional division to compute the amount of reallocation of each agent’s resources. the numbers of resources of type k reallocated to the ith agent are defined by [equation (15)] p i m m o i nk k( ) ,,=     ∈1 all × all (15) where oall,k is the total number of sar resources of type k; mi indicates the seismic intensity of the ith agent in the district when a serious quake occurred. the reallocation of the ith agent’s resources in district pk(i) is derived from the proportion of mi/mall multiplied by the total number of resources available of type k oall,k. table 8 shows the experimental results for 15 districts (or agents) using the proportional division of resources. in figure 5, it can be seen that with the shapley value allocation approach proposed in this study, the average distribution of resource rate is closer to the death rate of each district than table 7 | our proposed division of resources for 15 affected districts districts resource allocation after earthquake resource allocation rate after earthquake (%) average distribution rate (%) number of deaths death rate (%) rescue workers ambulances fire engines machine and equipments rescue workers ambulances fire engines machine and equipments a1 taipei 97 98 186 72 8.62 8.60 8.65 8.67 8.64 134 5.40 a2 taoyuan 41 42 78 30 3.64 3.68 3.63 3.61 3.64 1 0.04 a3 hsinchu 25 26 48 19 2.22 2.28 2.23 2.29 2.26 0 0.00 a4 miaoli 17 17 33 13 1.51 1.49 1.53 1.57 1.53 6 0.24 a5 taichung 376 381 718 277 33.42 33.42 33.40 33.37 33.40 1305 52.58 a6 changhua 45 46 87 33 4.00 4.04 4.05 3.98 4.01 30 1.21 a7 nantou* 376 381 718 277 33.42 33.42 33.40 33.37 33.40 916 36.91 a8 yunlin 71 72 135 52 6.31 6.32 6.28 6.27 6.29 82 3.30 a9 chiayi 15 15 29 11 1.33 1.32 1.35 1.33 1.33 7 0.28 a10 tainan 39 39 74 28 3.47 3.42 3.44 3.37 3.43 1 0.04 a11 kaohsiung 6 7 12 5 0.53 0.61 0.56 0.60 0.58 0 0 a12 pingtung 13 13 24 9 1.16 1.14 1.12 1.08 1.12 0 0 a13 taitung 2 2 4 2 0.18 0.18 0.19 0.24 0.20 0 0 a14 hualien 2 2 4 2 0.18 0.18 0.19 0.24 0.20 0 0 a15 yilan 1 1 3 1 0.09 0.09 0.14 0.12 0.11 0 0 total 1125 1140 2150 830 100 100 100 100 100 2482 100 *epicenter. table 8 | proportional division of resources for 15 affected districts districts magnitude resource allocation after earthquake resource allocation rate after earthquake (%) average distribution rate (%) number of deaths death rate (%) rescue workers ambulances fire engines machine and equipments rescue workers ambulances fire engines machine and equipments a1 taipei 4 63 64 121 47 5.63 5.63 5.63 5.63 5.63 134 5.40 a2 taoyuan 4 63 64 121 47 5.63 5.63 5.63 5.63 5.63 1 0.04 a3 hsinchu 4 63 64 121 47 5.63 5.63 5.63 5.63 5.63 0 0.00 a4 miaoli 5 79 80 151 58 7.04 7.04 7.04 7.04 7.04 6 0.24 a5 taichung 6 95 96 182 70 8.45 8.45 8.45 8.45 8.45 1305 52.58 a6 changhua 6 95 96 182 70 8.45 8.45 8.45 8.45 8.45 30 1.21 a7 nantou* 7 111 112 212 82 9.86 9.86 9.86 9.86 9.86 916 36.91 a8 yunlin 6 95 96 182 70 8.45 8.45 8.45 8.45 8.45 82 3.30 a9 chiayi 5 79 80 151 58 7.04 7.04 7.04 7.04 7.04 7 0.28 a10 tainan 5 79 80 151 58 7.04 7.04 7.04 7.04 7.04 1 0.04 a11 kaohsiung 4 63 64 121 47 5.63 5.63 5.63 5.63 5.63 0 0 a12 pingtung 3 48 48 91 35 4.23 4.23 4.23 4.23 4.23 0 0 a13 taitung 3 48 48 91 35 4.23 4.23 4.23 4.23 4.23 0 0 a14 hualien 5 79 80 151 58 7.04 7.04 7.04 7.04 7.04 0 0 a15 yilan 4 63 64 121 47 5.63 5.63 5.63 5.63 5.63 0 0 total 1125 1140 2150 830 100 100 100 100 100 2482 100 *epicenter. 144 c-k. wu and x. hu / journal of risk analysis and crisis response 9(3) 134–144 figure 5 | comparison of our division and proportional division. [2] dixit a, skeath s. games of strategy. new york, ny: w. w. norton & company; 2007. [3] parsons s, wooldridge m. game theory and decision theory in multi-agent systems. autonom agents multiagent syst 2002;5: 243–54. [4] ranganathan n, gupta u, shetty r, murugavel a. an automated decision support system based on game theoretic optimization for emergency management in urban environments. j homeland secur emerg manage 2007;4. [5] cai j, pooch u. allocate fair payoff for cooperation in wireless ad hoc networks using shapley value. in: proceedings of the 18th international parallel and distributed processing symposium. santa fe, nm, usa: ieee; 2004, pp. 219–27. [6] conejo aj, contreras j, lima da, padilha-feltrin a. zbus transmission network cost allocation. ieee trans power syst 2007;22:342–9. [7] weiss g. multiagent systems – a modern approach to distributed artificial intelligence. cambridge: mit press; 1999. [8] burke da. towards a game theory model of information warfare, master’s thesis. air force institute of technology, air university; 1999. [9] paruchuri p, pearce jp, tambe m, ordóñez f, kraus s. an efficient heuristic approach for security against multiple adversaries. in: international conference on autonomous agents and multiagent systems (aamas). 2007, pp. 311–318. [10] alpcan t, basar t. a game theoretic approach to decision and analysis in network intrusion detection. in: 42nd ieee international conference on decision and control (ieee cat. no.03ch37475). maui, hi, usa: ieee; 2003, pp. 2595–600. [11] gupta u, ranganathan n. multievent crisis management using noncooperative multistep games. ieee trans comput 2007;56:577–89. [12] osborne mj, rubinstein a. a course in game theory. cambridge: mit press; 1994. [13] mishra d, rangarajan b. cost sharing in a job scheduling problem using the shapley value. in: proceedings of the 6th acm conference on electronic commerce. vancouver, bc, canada: acm; 2005, pp. 232–9. [14] zolezzi jm, rudnick h. transmission cost allocation by cooperative games and coalition formation. ieee trans power syst 2002;17:1008–15. [15] owen g. game theory, 3rd ed., new york, ny: academic press; 2003. [16] lemke ce, howson jt. equilibrium points of bimatrix games. j soc ind appl math 1964;12:413–23. [17] gibbons r. a primer in game theory. pearson higher education; 1992. [18] hansen t. on the approximation of nash equilibrium points in an n-person non-cooperative game. j soc ind appl math 1974;26:622–37. [19] perry rw. emergency operations centers in an era of terrorism: policy and management functions. j contingencies crisis manage 2003;11:151–9. [20] hsu yw. a study on response of fire department in earthquake focusing on the 921 ‘chi-chi’ earthquake, master thesis, department of fire science, central police university, taiwan; 2000. the proportional division of resources. the results of this comparison support the idea of the proposed model, which creates a minimum set of resource allocation costs when the mas confronts external attacks. the eoc rapidly commands all agents and efficiently mobilizes and redistributes sar resources for seriously affected districts. 6. conclusion ids, hsas, and the emergency response system deploy multiple interacting agents to resist outside attackers. the resources of mas are constrained for defending entire large-scale geographic regions. the emergency manager faces problems in agent resource deployment. the divide-and-conquer idea is utilized to improve the scalability of resource allocation. a two-stage model is proposed in this study to combine a non-cooperative game with a cooperative game to provide acceptable reallocation of resources for the mas. the proposed model is applied to one strong earthquake attack case and is able to allocate resources efficiently in the overall system. future work will design a computer-based decision-support system based on the proposed model. real data will be applied and the feasibility of resource allocation from an organization’s emergency operational center point of view verified. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments the authors appreciate the time and effort of the editors and reviewers in providing constructive comments. references [1] lewis hw. why flip a coin? the art and science of good decisions. new york, ny: john wiley and sons; 1997. https://doi.org/10.2202/1547-7355.1236 https://doi.org/10.2202/1547-7355.1236 https://doi.org/10.2202/1547-7355.1236 https://doi.org/10.2202/1547-7355.1236 https://doi.org/10.1109/ipdps.2004.1303254 https://doi.org/10.1109/ipdps.2004.1303254 https://doi.org/10.1109/ipdps.2004.1303254 https://doi.org/10.1109/ipdps.2004.1303254 https://doi.org/10.1109/tpwrs.2006.889138 https://doi.org/10.1109/tpwrs.2006.889138 https://doi.org/10.1109/tpwrs.2006.889138 https://doi.org/10.1109/cdc.2003.1273013 https://doi.org/10.1109/cdc.2003.1273013 https://doi.org/10.1109/cdc.2003.1273013 https://doi.org/10.1109/cdc.2003.1273013 https://doi.org/10.1109/tc.2007.1023 https://doi.org/10.1109/tc.2007.1023 https://doi.org/10.1109/tc.2007.1023 https://doi.org/10.1145/1064009.1064034 https://doi.org/10.1145/1064009.1064034 https://doi.org/10.1145/1064009.1064034 https://doi.org/10.1145/1064009.1064034 https://doi.org/10.1109/tpwrs.2002.804941 https://doi.org/10.1109/tpwrs.2002.804941 https://doi.org/10.1109/tpwrs.2002.804941 https://doi.org/10.1137/0112033 https://doi.org/10.1137/0112033 https://doi.org/10.1137/0126058 https://doi.org/10.1137/0126058 https://doi.org/10.1137/0126058 microsoft word volume 12, issue 4-2 journal of risk analysis and crisis response, 2022, 12(4), 170-188 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i4.342 170 article research on the spatial effect of government science and technology expenditure on the development of digital economy shu-xian wang 1,* and tao zhang 1 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: wangshuxian0426@163.com; tel.: +86-13598629108 received: september 11, 2022; accepted: october 29, 2022; published: december 31, 2022 abstract: in the context of a unified large market, the development of digital economy has become the focus. based on the panel data of 31 provinces in china from 2011 to 2020, this paper uses the spatial durbin model to examine the impact of government spending on digital economy through different spatial weight matrices. the research results show that: 1) the government science and technology expenditure in all provinces in china has a significant positive spatial correlation with the level of digital economy development; 2) the comparative analysis of different weight matrices show that the impact of local government science and technology expenditure on the level of digital economy development is significant and positive, and there is a certain spatial spillover effect; 3) through the heterogeneity test, the spillover effect of the central and western regions of china is more obvious under the setting of the geographical weight matrix, and the spillover effect of the east is more obvious under the setting of the economic matrix weight. at the same time, this study expands the research on the influencing factors of the development level of the digital economy, and puts forward policy recommendations corresponding to government science and technology expenditure and the development of the digital economy based on the empirical results. keywords: digital economy; government science and technology expenditure; spatial spillover effect 1. introduction the digital economy is a new development trend formed after the agricultural economy and the industrial economy. it promotes greater alignment of fairness and efficiency. the development speed, radiation scope, and degree of influence of the digital economy is all at the forefront, constantly promoting the transformation of production, living and governance methods, and gradually becoming the key to the competition and transformation of world resources, structures and patterns. during the "13th five-year plan" period, china has deepened the digital economy development strategy, strengthens and improves digital infrastructure, and continuously cultivates new business forms and new models, resulting in remarkable achievements in digital industrialization and industrial digitization. during the "14th five-year plan" period, china digital economy has turned to a new stage of development. in 2021, the notice of the state council on printing and distributing the "14th five year" digital economy development plan mentions that the added value of china's digital economy core industry will reach 7.8% of gross domestic product (gdp) in 2020, and the digital shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 171 economy will continue to promote the sustainable and healthy development of the economy and society. the digital economy continues to promote the sustainable and healthy development of the economy and society. according to the data released by the ministry of science and technology recently, in 2021, the whole country's r&d investment will reach 2.79 trillion yuan, a year-on-year increase of 14.2%, and the intensity of r&d investment will reach 2.44%. in 2020, longhua district proposes to accelerate the development of "digital longhua". now, "digital longhua, urban core" has become a business card of longhua district. the development of digital longhua cannot be separated from scientific and technological innovation, which cannot be separated from the support of government scientific and technological expenditure. at the same time, in the context of a unified large market, the government requires to continuously increase digital construction, promote the integration of online and offline, and use the digital economy to break the regional limitations of the region. at present, the development of a new round of technological revolution and industrial transformation has accelerated the digital transformation. through the influence of multiple internal and external factors, the situation facing the development of china digital economy is also constantly changing. so, can increasing local financial support for the development of the digital economy significantly affect the level of development level of the digital economy in the corresponding region? will local financial support have spatial spillover effects? through the exploration of the above problems, this study can provide useful policy inspiration for the development of china digital economy to a certain extent. 2. literature review 2.1. digital economy entering the 21st century, digital technology is advancing by leaps and bounds around the world. since the outbreak of the covid-19 epidemic, the world pattern has undergone tremendous changes. on the one hand, the world economy has been seriously damaged, but on the other hand, the development of the digital economy has also received attention from countries around the world. in view of the existing literature, the relevant literature on the development and impact of the digital economy is excavated, and it is concluded that the research direction of the existing literature is mainly concentrated in the following aspects: the first is the level of macroeconomic development. sun et al (2021) [1] believe that the development level of the regional digital economy shows significant spatial imbalance characteristics. jiao (2021) [2] studies the regional differences and dynamics of the development of the digital economy, and uses the natural fracture method to develop the digital economy. the levels are divided into five classes. the second is industry and governance. the first is industrial digitization. zhang (2018) [3] points out that to promote the development of the digital economy, it is necessary to develop the manufacturing industry firstly, pay attention to the transformation and upgrading of the real economy, and transform the industrial structure to the mid-to-high end. zheng et al (2020) [4] find that the digital development of the industrial chain can be promoted through clustering; followed by the digital industry, liu et al (2022) [5] believe that the focus should be on how to give full play to the penetration of digital technologies such as big data. the role is the main task of the digital industry; shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 172 the last is digital governance. zhao et al (2022) [6] believe that it is necessary for us to systematically analyze the mechanism and dilemma of the digital economy to promote the reform of government governance, seek countermeasures, and promote the digitalization of government governance. the third is to combine the digital economy with urban development, consumption or enterprises. gao et al (2022) [7] link the digital economy with consumption, and believe that the digital economy is an important focus for driving consumption upgrades. liao et al (2021) [8] believe that the digital economy can improve the ability of enterprises to allocate resources and then promote the transformation of the manufacturing industry upgrade. 2.2. financial support at present, there are limited studies on finance and digital economy, and related literatures are mostly studied from both macro and micro aspects. first, at the macro level, fan et al (2020) [9] accelerate the construction of new infrastructure and improve the global competitiveness of china digital economy from the perspective of supply and demand. fan (2021) [10] believes that the reform of the fiscal and taxation system should conform to the trend of economic digitization, make effective efforts in employment and other aspects, carry out macro-coordination in terms of supply and demand, and guide the formation of a unified large market from the perspective of financial system and mechanism. li et al (2021) [11] believes that digital finance can drive the upgrade of public consumption and stimulate the demand for digital consumption in the field of public services. with the help of digital technology, it can accurately grasp the trend of social and economic operation and effectively carry out macro-control. deng et al (2021) [12] obtain through the mediation effect test that the development of digital economy can affect local fiscal sustainability through fiscal revenue. at the micro level, li et al (2021) [13] analyze the financial system from five aspects: the legal system and development environment of the digital economy, the macro layout of infrastructure optimization projects, the integration with traditional industries, the guidance of consumption upgrades, and the synergy of fiscal, taxation and financial policies. combined with the digital economy, we will continue to promote the positive effect of financial support on the development of the digital economy. jiang (2022) [14] combines finance and digital innovation, and believes that government preference can significantly improve the effect of regional digital innovation, but the incentive effect on regional digital innovation efficiency is not significant. by combing the relevant literature on the digital economy through fiscal policy, it is found that the existing literature has the following characteristics: first, regarding the digital economy, most literatures in the past five years have studied the digital economy and the real economy, technological innovation, and industrialization and innovation. the combination of efficiency and the combination of finance and the digital economy is relatively rare; secondly, most of the current research focuses more on the theoretical level, and there is less research on the spatial effect of finance and the digital economy. therefore, after measuring the development level of the digital economy, on the basis of sorting out relevant theories, this paper uses the dynamic panel data of 31 provinces in china from 2011 to 2020 to test the impact of government science and technology expenditure on the development level of the digital economy. from the perspective of spatial spillover effects government science and technology expenditure and the level of development of the digital economy will be studied. the innovation of this paper is to study whether there is a spillover effect of shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 173 government science and technology expenditure on the development level of digital economy from a spatial perspective through a spatial econometric model. 3. theoretical analysis and research assumptions keynesian fiscal expenditure theory believes that due to insufficient effective demand, government intervention is needed, and the main means of government intervention is fiscal expenditure. therefore, keynes put great emphasis on the role of financial expenditure means. according to the classification of government revenue and expenditure by the ministry of finance, as government science and technology expenditure is a part of financial expenditure, this paper studies the impact of government science and technology expenditure on digital economy from the perspective of government science and technology expenditure. lu et al (2020) [15] believe that the development of digital economy can significantly improve the quality of urban economic development in the region, and significantly promote the economic development of neighboring regions. not only that, yang et al (2021) [16] believe that the impact of digital economy development is heterogeneous. the study found that the spatial distribution of my country's digital development level is not completely random, but presents a state of agglomeration, but the problem of regional imbalance in the development of digital economy is prominent. based on the previous research by scholars, it can be concluded that there is a significant spatial effect in the development of the digital economy. fiscal policy, as a basic criterion formulated by the state to know fiscal distribution activities and deal with various fiscal distribution relationships, reflects the state's focus on economic development to a certain extent. therefore, this paper focuses on the study of the impact of government science and technology expenditure on the development level of the digital economy influences. based on the above analysis, this paper proposes the following two hypotheses: hypothesis 1: government science and technology expenditure can promote the level of development of the digital economy, and government science and technology expenditure have a certain spatial effect on the level of development of the digital economy. hypothesis 2: government science and technology expenditure has different effects on the development level of the digital economy in different regions. 4. research design 4.1. variable selection 4.1.1. explained variable digital economy development level (dig). by borrowing the method of zhao et al (2020) [17], the core of the measurement is set to the development of the internet, and the construction idea of the indicator system of digital transactions is added to measure the development level of the digital economy from both the internet and digital financial inclusion. the raw data for the indicators can be obtained from the china urban statistical yearbook. starting from the comprehensive development index of digital economy, this paper selects five secondary indicators. the construction of indicators is shown in table 1. after that, the weight and comprehensive index of the indicators are calculated by the entropy method, and the comprehensive development level of the digital economy in 31 provinces in china from 2011 to 2020 is obtained, which is recorded as dig. shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 174 table 1. construction of regional digital economy comprehensive development indicators. first-level indicator secondary indicators indicator properties index weight digital economy comprehensive development index internet penetration positive indicators 0.079 number of internet-related employees positive indicators 0.315 internet related output positive indicators 0.423 number of mobile internet users positive indicators 0.091 digital financial inclusion development positive indicators 0.094 the main purpose of this paper is to investigate the spatio-temporal evolution pattern of china's digital economy development from 2011 to 2020. in order to ensure the comparability of the time series of relevant indicators, first, the indicators of different properties and units are dimensionless. in order to avoid the uneven distribution caused by the large difference of indicator values, the data shall be standardized as follows. (1) (2) formula (1) is a positive indicator formula, and formula (2) is a negative indicator formula. among them, xij refers to the value of index j of province i before standardization, xij refers to the value of index j of province i after standardization, and n refers to provinces, autonomous regions, and municipalities directly under the central government (due to the availability of data, this paper considers excluding hong kong, macao, and taiwan, so n is 31), and m refers to the number of indicators (building a digital economy development evaluation index system that includes one level one indicator and five level two indicators, so m is 5). the weight is calculated by entropy method. the entropy method reflects the information content of the same index difference, which can effectively avoid the influence of subjective factors in the weight setting process. before calculating the weight of each indicator, first calculate its information entropy. the formula is as follows. (3) secondly, calculate the weight of each evaluation index. the formula is as follows. (4) finally, based on the standardized value xij of each indicator and the weight wj of each indicator, the multi-objective linear weighting function method is used to calculate the digital economy development level at the national and provincial levels from 2011 to 2012. the calculation formula as follows. shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 175 (5) 4.1.2. explanatory variables local government science and technology expenditure (gov): the new infrastructure such as 5g networks, data centers, and industrial internet that china is focusing on promoting the construction of, is essentially the digital economic infrastructure around the science and technology innovation industry. therefore, this paper expresses the proportion of government science and technology expenditure in fiscal spending to measure government science and technology expenditure. 4.1.3. control variables the selection of control variables refers to the practices of sun et al. (2021) and liang et al. (2021), considering the theme of this study. the control variables selected in this paper are as follows: 1) regional r&d investment (r&d): expressed by the ratio of the internal expenditure of scientific research funds in each province to gdp. 2) regional economic development level (eco): by gdp. 3) urbanization level (urb): measured by dividing the urban population of each province by the total population. 4) industrial structure (ind): dividing by the output value of the tertiary industry in each province expressed by gross output value. 5) financial development level (finance): expressed by the ratio of institutional deposit and loan balance to regional gdp. 4.2. econometric model in order to verify the impact of local government science and technology expenditure on the development level of the digital economy, the linear model form of this paper is set as follows. digit=a+ρwdigit+βgovit+β1xcontrol+δwgovit+δ1wxcontrol+μi+θt+εit (6) in formula (6), dig represents the regional digital economy development level of 31 provinces in my country; gov represents the scientific and technological expenditure of local governments, xcontrol is a control variable, a is a constant term, ρ is a spatial autoregressive coefficient, and w is the standardized value. spatial weight matrix, β represents the regression coefficient of the independent variable, δ represents the influence coefficient vector of the spatial lag term of the dependent variable, μi is the time effect, θt is the spatial effect, and εit represents the random error term. government science and technology expenditure contribute to some extent to the level of digital economic development, but endogeneity problems may arise when unobserved variables confound the level of digital economic development and government science and technology expenditure, or when measurement errors in government science and technology expenditure are significant. in order to solve the problem of endogeneity and consider the interdependence of space, this paper constructs a dynamic space panel for analysis, and the model is set as follows. digit=a+τdigit-1+ξwdigit-1+ρwdigit+βgovit+β1xcontrol+δwgovit+δ1wxcontrol+μi+θt+εit (7) shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 176 the digit-1 is the first-order lag term of the explained variable. the implication is the impact of the previous period's local government science and technology expenditure on the level of digital economy development. τ is the regression coefficient of the first order lag term of the development level of the digital economy, ξ is the spatial autoregressive coefficient and other variables are explained in the same way as formula (6). 4.3. data sources and descriptive statistics the data sources and calculation methods of each variable are obtained from the table below. table 2. variable selection and indicator description. variable type variable name variable symbol indicator explanation data sources explained variable digital economy development level dig calculated using the entropy method calculated by the author explanatory variables government science and technology expenditure gov government technology expenditure/fiscal expenditure “china science and technology statistical yearbook” control variable regional r&d investment r&d internal expenditure of scientific research funds in various provinces/gdp regional economic development level eco regional gdp per capita by gdp deflator national bureau of statistics the level of urbanization urb urban population/total population of each city industrial structure ind tertiary industry output value/total output value of each province level of financial development finance institutional deposit and loan balance / gross regional product table 3. descriptive statistics of each variable from 2011 to 2020. variable variable name mean std. dev. min max dig digital economy development level 0.371 0.174 0.0773 0.982 gov government science and technology expenditure 2.039 1.471 0.3 6.76 r&d regional r&d investment 1.628 1.141 0.19 6.44 eco regional economic development level 5.104 2.409 1.602 14.096 urb the level of urbanization 0.577 0.132 0.227 0.896 ind industrial structure 0.494 0.089 0.327 0.837 finance level of financial development 0.031 0.011 0.012 0.073 descriptive statistics for each variable can be obtained from the table below. according to table 3, the average value of the digital economic development level of the explained variable is 0.371, the maximum value is 0.982, and the minimum value is 0.0773, indicating that there are differences in shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 177 the development level of digital economy in different regions. explanatory variables the average value of government science and technology expenditure is 2.039, the maximum value is 6.76, and the minimum value is 0.3, indicating that there are significant differences in the level of government science and technology expenditure in different regions. there are also significant differences in regional r&d investment, regional economic development level, urbanization level, industrial structure and financial development level of control variables. 4.4. spatial econometric methods based on the methods and principles of spatial econometrics, the idea of spatial econometric analysis of factors affecting the development level of digital economy is as follows: first, moran's i index method is used to test whether there is spatial autocorrelation in the level of digital economy development of the dependent variable; if there is spatial autocorrelation, then a spatial econometric model is established to estimate and test the spatial econometric factors influencing the development level of the digital economy. 4.4.1. spatial autocorrelation first, the global moran’s i index is used to test the spatiality of the development level of the digital economy. the formula is as follows. (8) in equation (3), n is the total number of regions, yi is the observation value of i region, wij is the spatial weight and s is the standard deviation, s2 is the variance. the value range of moran's i index is [-1, 1]. if moran's i>0, the variables are considered to have positive spatial correlation; if moran's i<0, it is spatially negative correlation; if moran's i=0, there is no spatial correlation sex. there are three types of weight matrices constructed in this paper: the first one is the geographic distance weight matrix (w1), in which this paper calculates the geographic distance between provincial capitals through latitude and longitude and then takes the reciprocal; the second one is the economic distance weight matrix (w2), which is calculated by the inverse of the absolute value of the difference in per capita gdp between the two regions; the third is the geographic and economic nested weight matrix (w3), the formula is w3=kw1+(1-k)w2, where k is the value the range is between 0 and 1, indicating the proportion of the geographic distance weight matrix. this part refers to the practice of shao shuai et al. (2016), and the value of k is set to 0.5. 4.4.2. spatial measurement model the spatial measurement model mainly includes the spatial durbin model (sdm), the spatial lag model (slm) and the spatial error model (sem). regarding the selection of the spatial econometric model in this paper, it is first assumed that the optimal spatial econometric model in this paper is the spatial doberman model (sdm), which will be selected based on the following tests. 5. analysis of empirical results 5.1. spatial correlation test shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 178 table 4 shows the spatial moran index results of the digital economy development index. it can be seen from the results in the table that under the settings of the three weight matrices, the global moran index of the development level of the digital economy is all greater than 0 and very significant. this results in a significant spatial effect on the distribution of development level of the digital economy in various provinces in my country. in these study samples, however, moran index values show a volatile downward trend over time. this also shows that the spatial agglomeration effect of the development level of the digital economy may gradually weaken over time. at present, the economic level of various regions in china has been significantly improved, and the imbalance between regions is also weakening. in particular, the national policies of "western development" and "the rise of the central plains" have accelerated the economic development of the central and western regions. in addition, the digital economy development pilot zone set by the state covers the east, the central, and the west. chongqing, guizhou and other places in the west have accelerated the development of digital economy, which has weakened the spatial agglomeration effect of the development level of digital economy to a certain extent. therefore, the spatial cluster effect of the development level of digital economy will gradually weaken over time. in addition, the moran value of the digital economy development index is generally larger than the geographic distance weight matrix and the geographic economy nested weight matrix under the economic distance weight matrix, which shows that the economy plays a promoting role in the spatial impact of the development level of the digital economy. the spatial influence of economic development level plays a narrowing role. table 4. spatial moran’s i index of regional digital economy development index. year weight matrix w1 weight matrix w2 weight matrix w3 digital economy digital economy digital economy moran’s i z value moran’s i z value moran’s i z value 2011 0.058*** 3.085 0.405*** 4.945 0.210*** 3.275 2012 0.060*** 3.192 0.403*** 4.969 0.206*** 3.251 2013 0.053*** 2.940 0.425*** 5.202 0.204*** 3.213 2014 0.044*** 2.673 0.421*** 5.271 0.195*** 3.157 2015 0.032** 2.287 0.410*** 5.165 0.184*** 3.014 2016 0.042*** 2.666 0.416*** 5.304 0.192*** 3.165 2017 0.030** 2.219 0.408*** 5.140 0.182*** 2.984 2018 0.023* 1.918 0.380*** 4.720 0.169*** 2.750 2019 0.018* 1.772 0.372*** 4.671 0.162*** 2.677 2020 0.012 0.029 0.364*** 4.620 0.147** 2.504 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. 5.2. analysis of spatial measurement results 5.2.1. model selection and related tests the first step is to test the model. table 5 shows the diagnostic test results of the spatial econometric model. the lm statistics in the table are all significant at the 1% level under the three weight matrices, indicating that the sem model and the sar model can be selected, so we choose the shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 179 sdm model combining the two. secondly, the hausman test is carried out, and it is concluded that the results are significant under the three weight matrices, and the fixed effect model is better than the random effect model, so the fixed effect model is better selected in this paper. after the fixed effect model is determined, it is necessary to test the three fixed effects and choose the most suitable model for this paper. the test results shows that the region fixed effect and the time fixed effect is very significant. therefore, when choosing the sdm model in this paper, it is better to choose the timespace double-fixed effect model. the wald test shows that the null hypothesis that the sdm model can degenerate into sar and sem is rejected, and the sdm model is accepted. from the results of the lr test, it can be concluded that under the three matrix settings, the lr statistic has passed the significance test at the 1% level, which is consistent with the previous wald test results. sdm cannot be degenerated into sar model or sem model. therefore, the model can be extended to a time-space bidirectional fixed effect model. to sum up, this paper chooses the sdm under double fixed effects to test the spatial effect of local government science and technology expenditure on the development level of the digital economy. table 5. diagnostic tests of spatial econometric models. diagnostic tests weight matrix w1 weight matrix w2 weight matrix w3 value p-value value p-value value p-value lm-error 1054.179*** 0.000 37.011*** 0.000 57.565*** 0.000 robust lm-error 767.219*** 0.000 12.446*** 0.000 22.738*** 0.000 lm-lag 353.917*** 0.000 46.446*** 0.000 60.532*** 0.001 robust lm-lag 66.957*** 0.000 21.881*** 0.000 25.705*** 0.000 hausman test 20.33*** 0.0024 14.08** 0.0288 21.57*** 0.0014 wald test-sar 45.29*** 0.0000 105.90*** 0.0000 24.64*** 0.0004 wald test-sem 43.82*** 0.0000 97.97*** 0.0000 24.68*** 0.0004 lr-sdm-sar 42.36*** 0.0000 89.72*** 0.0000 23.72*** 0.0006 lr-sdm-sem 42.12*** 0.0000 89.78*** 0.0000 23.75*** 0.0006 lr-both-ind 46.02*** 0.0000 102.51*** 0.0000 109.64*** 0.0000 lr-both-time 308.85*** 0.0000 805.58*** 0.0000 389.41*** 0.0000 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. 5.2.2. measurement results of the sdm 1) analysis of the impact of government science and technology expenditure on the development level of the digital economy. table 6 shows the estimated results of the sdm. among them, the columns (1), (3) and (5) are the estimation results of the static sdm under the three weight matrix settings, respectively, and the columns (2), (4) and (6) are the three dynamic sdm estimation results under the weight matrix setting. according to table 6, under the economic distance weight matrix and the economic geography nested weight matrix, the coefficient of local government science and technology expenditure is 0.007 and 0.004 are both positive. and both are significant for the weight matrices w2 and w3, indicating that the government's science and technology expenditure. it has a positive role in promoting the development level of the digital economy. under the setting of weight matrix w1 and w2, the shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 180 coefficient of local government science and technology expenditure is significantly -0.043 and -0.012, indicating that local government's financial expenditure on science and technology can restrain the development level of digital economy in regions with similar geographical locations or similar economic conditions to a certain extent. that is to say, the local government's science and technology expenditure have a spatial effect on the development level of the digital economy to a certain extent. so, hypothesis 1 is verified. the continuous increase in the proportion of local governments in science and technology expenditure can promote the transformation and upgrading of enterprises to a certain extent, spawn new digital enterprises, and improve the development level of the local digital economy. due to the scarcity of resources, the increase in the proportion of local government science and technology expenditure will attract many companies to influx. while improving the development level of the digital economy in the region, it will also inhibit the development of the digital economy in areas with similar geographical locations or economic conditions. table 6. estimation results of spatial panel model under three weight matrix settings. variable weight matrix w1 weight matrix w2 weight matrix w3 (1) (2) (3) (4) (5) (6) digt-1 (τ) 1.504*** (32.31) 1.814*** (36.52) 0.928*** (20.12) wdigt-1 (ξ) 4.476*** (11.13) 3.204*** (22.59) 0.638*** (3.52) wgov (ρ) -0.043*** (-2.62) 0.072*** (6.24) -0.012** (-2.25) 0.0006 (0.16) -0.004 (-0.56) 0.008* (1.85) gov -0.0002 (-0.11) 0.006*** (4.06) 0.007*** (3.74) 0.011*** (7.42) 0.004* (1.82) 0.006*** (3.87) r&d 0.001 (0.17) -0.010** (-2.18) -0.0004 (-0.07) -0.016*** (-3.38) 0.002 (0.32) -0.011** (-2.36) eco 0.012*** (3.36) -0.002 (-0.58) -0.018*** (-3.73) -0.015*** (-3.64) 0.012*** (2.88) 0.009*** (3.01) urb -0.289*** (-3.72) 0.218*** (3.66) 0.227*** (2.71) 0.285*** (4.10) -0.100 (-1.28) 0.087 (1.52) ind -0.228*** (-4.77) -0.241*** (-6.70) -0.191*** (-4.66) -0.073** (-2.16) -0.248*** (-5.51) -0.148*** (-4.42) finance -0.519** (-2.24) 1.898*** (11.61) -0.589*** (-2.77) -0.663** (-4.03) -0.533** (-2.23) 0.225 (1.36) n 310 279 310 279 310 279 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. test statistics with z test in ( ). in order to reduce the error of the model and make the dynamic change of the spatial spillover effect of the model more accurate, the following part will focus on analyzing the estimation results of the dynamic spatial model. according to the columns (2), (4) and (6) of table 6, the time lag coefficient under three weight matricesτ1.504, 1.814 and 0.928 respectively, significantly positive at the 1% level, indicating that at the time dimension level, urban there is an obvious path dependence in the development of the digital shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 181 economy, and the development level of the digital economy in the next period can be predicted by the results of the current period, which also means that the improvement of the development level of the digital economy is a long process that requires a lot of accumulation. the spatial lag coefficient ρ is significantly positive when it is 0.07 and 0.008 under the weight matrix w1 and w3, indicating that the level of development level of the digital economy in geographically similar regions has a significant positive impact on the development of digital economy in the region, that is, there is a spatial spillover effect. under the weight matrix of w1, w2 and w3, the space-time lag coefficient ξ 4.476, 3.204 and 0.638 respectively, significantly positive at the 1% level, which indicates that under the influence of the two dimensions of time and space, the number of areas with similar geographical locations or similar economic conditions in the previous period. the level of economic development has a significant positive impact on the current level of digital economy development in the region. this shows that the increase in the level of development level of the digital economy in regions with similar geographical locations or economic conditions in the previous period will affect the current level of development level of the digital economy in this region to a certain extent, causing it to rise to a certain extent. this article attributes the reason of the fact that when a region has a relatively high level of development level of the digital economy, it will produce "spillover effects" and "demonstration effects", which in turn stimulates the level of development level of the digital economy in regions with similar geographical locations or economic conditions. as shown in columns (2), (4) and (6) of the above table, the level of urbanization has a positive and significant relationship with the level of digital economy development. the reason is that the higher the level of urbanization, the higher the level of digital industries in the region the investment and construction of the region will be improved, which will significantly improve the development level of the digital economy in the region. on the contrary, the industrial structure has a significant negative impact on the development level of the digital economy under the three weight matrices. the reason may be that the development of the digital economy in my country at this stage is mainly concentrated in the tertiary industry. the lack of motivation for development can even lead to the preemption of digital economic resources between cities, and then there will be a "crowding-out effect", which will make the industrial structure have an inhibitory effect on the development level of the digital economy. the financial development level and the regional economic development level have different effects under different spatial weight settings, but in general, combined with the geographic and economic weight matrix, the regional economic development level has a significant role in promoting the development of the digital economy, and the level of financial development also has a positive impact on the development of the digital economy. the general view is that regional r&d investment has a promoting effect on the development level of the digital economy, and the results show that regional r&d investment has a certain negative impact on the development level of digital economy in geographically similar regions. the reason may be that the r&d investment in regions with high economic development level is high, while the r&d investment in regions with low economic development level is small. due to the imbalance of economic development, the development level of digital economy in this region has been restrained to a certain extent. 2) the decomposition effect of government science and technology expenditure on the digital economy development. shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 182 under the influence of the spatial spillover effect, changes in one of these factors will not only affect the development level of the local digital economy, but also affect the development level of the digital economy in areas with similar geographical locations or economic conditions. the level of development of the digital economy. because the change of the explanatory variables associated with a spatial individual will affect the spatial individual itself, this impact is the direct effect described in the traditional regression model, and it will also indirectly affect other spatial individuals, producing indirect effects. table 6 is an analysis of the spatial effect of government science and technology expenditure on the development level of the digital economy. however, in order to see the impact of government science and technology expenditure on the development level of the digital economy more intuitively, the results are decomposed into direct effects and indirect effects for analysis. the results are shown in table 7. table 7. decomposition results of government science and technology expenditures on the development level of digital economy under three weight matrices. matrix type effect lngov r&d eco urb ind finance w1 direct effect -0.0002 (-0.10) 0.001 (0.12) 0.013*** (3.58) -0.282*** (-3.38) -0.226*** (-4.67) -0.512** (-2.13) indirect effect -0.045** (-1.97) -0.057 (-1.32) 0.063** (2.07) 2.886*** (3.09) 0.434 (0.94) -8.373*** (-2.70) total effect -0.045* (-1.91) -0.056 (-1.28) 0.076** (2.48) 2.604*** (2.74) 0.208 (0.44) -8.865*** (-2.79) w2 direct effect 0.007*** (3.18) -0.002 (-0.26) -0.015*** (-3.33) 0.173** (2.08) -0.192*** (-4.77) -0.456** (-2.04) indirect effect -0.013 (-1.61) -0.023 (-1.02) 0.071*** (4.34) -1.325*** (-3.71) -0.039 (-0.31) 3.037*** (3.45) total effect -0.006 (-0.63) -0.024 (-0.96) 0.057*** (3.45) -1.152*** (-2.95) -0.231 (-1.64) 2.581*** (2.61) w3 direct effect 0.004* (1.80) 0.002 (0.30) 0.012*** (3.12) -0.100 (-1.25) -0.250*** (-5.68) -0.520** (-2.14) indirect effect -0.004 (-0.51) -0.052** (-2.08) 0.021 (1.42) 0.012 (0.03) -0.098 (-0.78) 2.949*** (3.34) total effect 0.0004 (0.05) -0.050* (-1.88) 0.033** (2.31) -0.086 (-0.23) -0.348*** (-2.65) 2.429** (2.53) note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. test statistics with z test in ( ). from the above results, it can be concluded that under the weight setting of w1, the indirect effect and total effect of local government science and technology expenditure are negative, which indicates that government science and technology expenditure not only has an impact on the development level of the digital economy in the region, but also has an impact on the geographically similar regions. with certain spillover effects, hypothesis 1 is confirmed again. under the weight matrix of w2 and w3, the science and technology expenditure of local governments has only a direct effect, indicating that the increase of science and technology expenditure in a region will improve the development level of the digital economy in the region, but it has no significant effect on the shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 183 development level of the digital economy in regions with similar economic conditions. spillover effect. it can be seen from the above that increasing the development of the digital economy in regions with similar geographical locations or similar levels of economic development will increase government science and technology expenditure in the region in the next period and improve the development level of the region’s digital economy. the increase in the ratio will only have a negative impact on regions with similar geographical locations, and although there is a certain negative impact on regions with similar economic development conditions, the results are not significant. the reason may be that financial agglomeration makes regions with similar geographical locations compete for resources, while regions with similar economic development levels have different strategic orientations, and the level of economic development is similar, so that the competition for resources is not so fierce, and then the level of development level of the digital economy in a region. it does not significantly affect the level of development level of the digital economy in another region. 5.3. heterogeneity analysis due to the agglomeration effect of economic development, this paper divides the eastern, central and western region into three regions and further analyzes the regional heterogeneity of the impact of government science and technology expenditure on the development level of the digital economy. the specific results are shown in table 8. table 8. heterogeneity test in eastern, central and western regions. variable w1 w2 w3 east centra west east centra west east centra west wgov (ρ) 0.033 (1.91) 0.057*** (3.73) 0.144*** (3.54) -0.028** (-3.23) -0.0004 (-0.08) -0.014 (-0.83) 0.002 (0.11) 0.002 (0.39) -0.021 (-1.12) gov 0.009** (2.26) 0.014*** (4.57) 0.026*** (3.05) 0.004 (1.06) 0.005* (1.89) 0.007 (0.87) 0.004 (0.99) 0.003 (1.03) 0.002 (0.23) r&d 0.012 (1.27) -0.11 (-1.16) -0.020 (-1.27) 0.007 (0.79) -0.014 (-1.24) -0.028 (-1.48) 0.013 (1.28) -0.011 (-1.07) -0.026 (-1.57) eco 0.014*** (3.43) -0.039*** (-2.67) -0.015 (-1.57) -0.021** (-2.26) 0.011 (0.86) -0.007 (-0.59) 0.018*** (3.26) 0.022** (1.99) 0.002 (0.25) urb -0.517*** (-3.82) -0.001 (-0.00) -0.558* (-1.83) 0.358* (1.88) 0.559*** (2.89) 0.181 (0.79) -0.417*** (-2.23) 0.419** (2.04) 0.124 (0.53) ind -0.083 (-0.52) 0.151** (2.29) -0.212*** (-3.05) -0.096 (-0.67) -0.079 (-1.44) -0.168** (-2.38) -0.143 (-1.84) -0.045 (-0.85) -0.158** (-2.55) finance 0.346 (0.72) -0.719 (-1.53) -1.341*** (-4.82) -0.166 (-0.34) 0.225 (0.43) -0.790*** (-3.20) 0.384 (0.73) -0.154 (-0.31) -0.736*** (-3.15) n 120 90 100 120 90 100 120 90 100 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. test statistics with z test in ( ). it can be seen from the table that under the setting of the w1 weight matrix, the spatial spillover effect of government science and technology expenditure in the central and western regions is significantly positive, and the government science and technology expenditure in the three regions is very significant. this shows that in the eastern, central and western regions, government science shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 184 and technology expenditure will significantly promote the development level of the digital economy in the region. in the central and western regions, an increase in government science and technology expenditure in one region will lead to an increase in government spending in regions with similar geographical locations, and then drive the level of development of the digital economy to improve. although the government science and technology expenditure in the east, middle and west is very significant, the absolute value of the coefficient is the west > the middle > the east. this shows that the lower the level of economic development in the region, the more the government's technology spending can stimulate the development level of the digital economy in the region. under the setting of the w2 weight matrix, the spatial spillover effect of government science and technology expenditure in the eastern region is significantly negative. the reason may be that due to the better economic development conditions in the eastern region, the increase in government science and technology expenditure in a region will cause more resources to flow into the region, and then it has squeezed the digital economy development market in areas with similar economic conditions in the east, and has a certain negative impact on the development of the digital economy in other areas with similar economic conditions. table 9. replacement core explanatory variables. variable weight matrix w1 weight matrix w2 weight matrix w3 (1) (2) (3) (4) (5) (6) digt-1 (τ) 1.934*** (41.34) 1.583*** (31.37) 5.149*** (110.61) wdigt-1 (ξ) 7.622*** (18.73) 1.755*** (12.15) 29.225*** (158.53) wlngov (ρ) -0.154** (-2.37) 0.238*** (4.98) -0.041** (-2.21) 0.013 (0.88) -0.021 (-0.88) 0.489*** (29.38) lngov -0.004 (-0.51) 0.020*** (3.37) 0.021*** (2.94) 0.026*** (4.36) 0.012 (1.44) 0.172*** (29.94) r&d 0.002 (0.32) -0.003 (-0.60) 0.001 (0.20) -0.015*** (-3.17) 0.003 (0.39) 0.006 (1.17) eco 0.012*** (3.41) 0.007*** (-2.63) -0.018*** (-3.66) -0.007* (-1.75) 0.011*** (2.69) -0.120*** (-38.56) urb -0.269*** (-3.37) 0.236*** (3.86) 0.194** (2.29) 0.164** (2.34) -0.107 (-1.33) -0.327*** (-5.50) ind -0.225*** (-4.65) -0.204*** (-5.54) -0.193*** (-4.71) -0.088** (-2.59) -0.248*** (-5.51) 0.104*** (3.07) finance -0.541** (-2.34) 2.205*** (13.42) -0.604*** (-2.85) -0.343** (-2.07) -0.561** (-2.36) -4.289*** (-25.83) n 310 279 310 279 310 279 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. test statistics with z test in ( ). 5.4. robustness test shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 185 zeng et al. (2022) [18] re measure the development of digital finance by using the logarithm of the digital financial index (indf). this paper uses zeng et al. to re measure the level of government science and technology expenditure by using the logarithm of government science and technology expenditure. in order to avoid differences in conclusions due to the selection of core explanatory variables, this paper refers to the method of zeng et al. to conduct the following robustness tests: 5.4.1. replace the core explanatory variables in order to avoid differences in conclusions due to the selection of core explanatory variables, this paper uses the logarithm of government science and technology expenditure (lngov) to remeasure the level of government science and technology expenditure. the results are shown in table 9. the results are basically consistent with the benchmark regression conclusions, indicating that the results of this paper are robust. 5.4.2. exclude autonomous regions table 10. exclude autonomous regions. variable weight matrix w1 weight matrix w2 weight matrix w3 (1) (2) (3) (4) (5) (6) digt-1 (τ) 2.664*** (56.29) 1.730*** (35.37) 2.664*** (56.29) wdigt-1 (ξ) 3.763*** (9.84) 0.457*** (3.63) 3.763*** (9.84) wgov (ρ) -0.039** (-2.51) 0.165*** (16.39) -0.014*** (-2.21) 0.020*** (6.05) -0.039** (-2.51) 0.165*** (16.39) gov -0.002 (-0.81) 0.012*** (8.66) 0.007*** (3.29) 0.004*** (2.76) -0.002 (-0.81) 0.012*** (8.66) r&d -0.002 (-0.31) 0.016*** (3.55) -0.001 (-0.13) -0.002 (-0.35) -0.002 (-0.31) 0.016*** (3.55) eco 0.010*** (2.72) 0.013*** (5.30) -0.023*** (-4.80) 0.006* (1.77) 0.010*** (2.72) 0.013*** (5.30) urb -0.340*** (-4.08) 1.898*** (31.30) 0.306*** (3.62) -0.074 (-1.17) -0.340*** (-4.08) 1.898*** (31.30) ind -0.238*** (-4.25) -0.777*** (-19.38) -0.160*** (-3.35) -0.115*** (-3.20) -0.238*** (-4.25) -0.777*** (-19.38) finance -0.381 (-1.11) 7.842*** (35.18) -0.652** (-2.19) 0.740*** (3.54) -0.381 (-1.11) 7.842*** (35.18) n 260 234 260 234 260 234 note: ***, **, * indicate statistical significance at the 1%, 5% and 10% levels, respectively. test statistics with z test in ( ). since the government management of autonomous regions is quite different from other regions, this may make the growth of digital economy development different from the influence of government spending. therefore, in order to verify the generality of the conclusions of this study, we chose to exclude the panel data of the five autonomous regions of inner mongolia, guangxi, tibet, shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 186 ningxia and xinjiang. the regression results are shown in table 10. it can be seen that the estimated parameters and significance are basically consistent with the basic regression results, which again shows that the results in this paper are robust. the table 9 and table 10 shows the estimated results of the sdm. among them, columns (1), (3) and (5) are the static sdm estimation results under the three weight matrix settings respectively, and columns (2), (4) and (6) are the dynamic model estimation results under the three weight matrix settings respectively. 6. conclusions and policy recommendations the continuous development of the digital economy not only points out the way for my country's future economic development, but also accelerates my country's socialist modernization to a certain extent, and continuously promotes the steady development of the development path of socialism with chinese characteristics. and how we should promote the development of the digital economy and further promote the economic development of our country is an urgent problem to be solved today. based on china's provincial panel data from 2011 to 2020, this paper uses three weight matrices and a sdm to verify the promoting effect of local government science and technology expenditure on the development of regional digital economy. finally, the following conclusions are drawn: first, based on different spatial weight matrices, the distribution of provincial government science and technology spending and the level of digital economy development shows a positive correlation. the more science and technology spending by local governments, the better the level of local digital economy development; expenditure can significantly improve the development level of the local digital economy. under the three weight matrix settings, the spatial spillover effect of government science and technology expenditure in areas with similar geographical locations or similar economic conditions are significantly positive; finally, from the perspective of location, government spending from the perspective of economic development, the spatial spillover effect of government science and technology expenditure in the eastern region is greater. the research of this paper also has the following shortcomings: because the provincial data of the subdivisional subjects under the government science and technology expenditure project cannot be obtained, this paper lacks mechanism analysis, and does not better present the impact of government science and technology expenditure on the development level of the digital economy. in the future, based on this, we will increase the number of data samples, take a more micro perspective, consider the impact of government science and technology expenditure on the development level of the digital economy from various aspects, and improve its transmission mechanism. based on the above conclusions, this paper puts forward the following policy suggestions: 1) since local government science and technology spending have a significant role in promoting the development level of the digital economy, regional governments should continue to increase financial investment in the development of the digital economy to stimulate the development of digital industries in the region. development, by strengthening policy guidance, constantly regulates the development of the digital economy. in addition, it is necessary to change the talent training plan and cultivate more "digital" talents. 2) it is necessary to increase economic support for the central and western regions, increase research and development expenses, and continuously promote the balanced development of the digital economy. it is necessary to fully consider the laws of spatial heterogeneity, and based on the shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 187 advantages of the region, create a digital economy development circle and increase the demonstration effect of the digital economy development pilot zone, formulate corresponding preferential fiscal policies to enhance the attractiveness of the digital economy. 3) local financial science and technology expenditure will affect the development level of digital economy in regions with similar geographical locations or economic conditions. at the same time, we should consider the time effect of government science and technology expenditure on the development level of digital economy, and consider the effect of regional digital economy development level from the two dimensions of space and time. therefore, when formulating financial budgets, we should fully consider the impact across regions, and build a network platform for digital economy from the overall situation, promote the circulation of digital resources, and accelerate the creation of "digital finance". funding: this research was funded by the guizhou provincial department of education university humanities and social sciences research project, grant number 2020gh023. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] sun l, xu w c. the impact of the digital economy on the regional global value chain embedding: an analysis based on the perspective of spatial spillovers [j]. economic management, 2021,(11):16-34. doi: https://doi.org/10.19616/j.cnki.bmj.2021.11.002. [2] jiao y. china digital regional differences and dynamic evolution of high-quality economic development [j]. economic system reform, 2021,(06):34-40. [3] zhang y z. development ideas and main tasks for digital economy to drive industrial structure to the midto-high end [j]. economics, 2018,(09):85-91. doi: https://doi.org/10.16528/j.cnki.22-1054/f.201809085. [4] zheng j z, li q. the basic connotation, measurement range and development path of the digital economy [j]. journal of zhejiang shu-ren university (humanities and social sciences), 2020,20(06):33-39. [5] liu j q, ru s f. how the digital economy affects high-quality economic development: based on an international comparative perspective [j]. economic system reform, 2022,(01):157-163. [6] zhao j h, du c h. digital economy promotion analysis of the mechanism, dilemma and way out of government governance reform [j]. theoretical discussion, 2022,(02):154-158. doi: https://doi.org/10.16354/j.cnki.23-1013/d.2022.02.017. [7] gao b, yuan h w. the mechanism and path of digital economy driving consumption upgrade under the dual-cycle pattern [j]. jiangsu administration institute journal, 2022,(02):36-44. [8] liao x l, yang z y. effect measurement and realization path of digital economy empowering manufacturing transformation and upgrading in the yangtze river delta region [j]. east china economic management, 2021,(06):22-30. doi: https://doi.org/10.19629/j.cnki.34-1014/f.201124009. [9] fan y x, xu h. finance helps the high-quality development of the digital economy: core mechanism and experience enlightenment [j]. reform, 2020,(08):83-91. [10] fan y x. reflections on promoting the expansion of domestic demand through the reform of the fiscal and taxation system under the digital economy [j]. financial supervision, 2021,(18):5-10. [11] li z, zhang r t. the coordinated development of digital economy and financial governance [j]. local finance research, 2021,(04):8-13. [12] deng d, pan g x, lin x l. the impact of china's digital economy development on the sustainability of local finance [j]. contemporary finance, 2021,(09):38-52. doi: https://doi.org/10.13676/j.cnki.cn361030/f.2021.09.005. [13] li j, han f q. fiscal and tax policy suggestions for china's digital economy development [j]. social scientist, 2021,(12):107-112. shu-xian wang and tao zhang / journal of risk analysis and crisis response, 2022, 12(4), 170-188 doi: https://doi.org/10.54560/jracr.v12i4.342 188 [14] jiang y c. government preference, fiscal decentralization and regional digital innovation [j]. statistics and decision making, 2022,38(04):32-37. doi: https://doi.org/10.13546/j.cnki.tjyjc.2022.04.006. [15] lu y x, fang x m, zhang a q. digital economy, spatial spillover and high-quality development of urban economy [j]. economic j w, 2021,(06):21-31. doi: https://doi.org/10.15931/j.cnki.1006-1096.2021.06.003. [16] yang h m, jiang l. digital economy, spatial effects and total factor productivity [j]. statistical research, 2021,(04):3-15. doi: https://doi.org/10.19343/j.cnki.11-1302/c.2021.04.001. [17] zhao t, zhang z, liang s k. digital economy, entrepreneurial activity and high-quality development: empirical evidence from chinese cities [j]. management world, 2020,36(10):65-76. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2020.0154. [18] zeng y p, jiang c y, cui z b. the impact of digital finance on high-quality economic development—a research based on the spatial dubin model [j]. technological economy, 2022,41(04):94-106. [19] liang q, xiao s p, li m x. development of digital economy, space spillover and improvement of regional innovation quality——also on market threshold effect [j]. shanghai economic research, 2021,(09):44-56. doi: https://doi.org/10.19626/j.cnki.cn31-1163/f.2021.09.004. [20] yao w h, yao z q. the impact of digital economy and r&d investment intensity on industrial structure upgrading [j]. journal of xi'an jiao-tong university (social science edition), 2021,(05):11-21. doi: https://doi.org/10.15896/j.xjtuskxb.202105002. [21] pan w h, he z c, pan h y. china's digital economy the spatiotemporal evolution and distribution dynamics of development [j]. china soft science, 2021,(10):137-147. [22] ren y h, xu ling, you w h. spatial econometric model of influencing factors of financial agglomeration and its application [j]. research on quantitative economy and technology economy, 2010,(05):104-115. doi: https://doi.org/10.13653/j.cnki.jqte.2010.05.011. [23] yu y z, liu d y. the spatial spillover effect and value chain spillover effect of china's regional innovation efficiency: a multi-dimensional spatial panel model study from the perspective of innovation value chain [j]. management world, 2013,(07):6-20+70+187. doi: https://doi.org/10.19744/j.cnki.11-1235/f.2013.07.002. [24] tang y q, shi j. fiscal decentralization, local government catch-up behavior and environmental governance efficiency—analysis of threshold effect and transmission mechanism based on data from 87 cities [j]. journal of guizhou university of finance and economics, 2019,(05):25-34. [25] liu y s. the integration and renewal of urban spatial layout driven by the development of digital economy [j]. regional economic review, 2022,(02):97-103. doi: https://doi.org/10.14017/j.cnki.2095-5766.2022.0024. [26] liu j, yang y j, zhang s f. china digital research on economic measurement and driving factors [j]. shanghai economic research, 2020,(06):81-96. doi: https://doi.org/10.19626/j.cnki.cn31-1163/f.2020.06.008. [27] guo f, wang j y, wang f, kong t, zhang x, cheng z y. measuring china's digital financial inclusion development: index compilation and spatial characteristics [j]. economics (quarterly), 2020,(04):1401-1418. doi: https://doi.org/10.13821/j.cnki.ceq.2020.03.12. [28] xue x y, liu x h. digital globalization, digital risk and global digital governance [j]. northeast asia forum, 2022,(03):3-18+127. doi: https://doi.org/10.13654/j.cnki.naf.2022.03.001. [29] zhang h m, li t, chen y z. dynamic comprehensive evaluation of ecological environment of 12 provinces and cities in western china[j]. journal of risk analysis and crisis response, 2019,9(3):156-161. doi: https://doi.org/10.2991/jracr.k.191024.005. [30] zhang m, luo q. a systematic literature review on the influence mechanism of digital finance on high quality economic development [j]. journal of risk analysis and crisis response, 2022,12(1):45-54. doi: https://doi.org/10.54560/jracr.v12i1.321. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). special issue reconstruction: key to social responsibility risk assessment for tourism investment projects† chen lv1,2, xiaoyan li2,*, 1china tourism academy, room 319, block 2, guanghua chang’an building 7 jianguomennei ave., dongcheng district, beijing 100005, china 2school of tourism, taishan university, 525 dongyue street, tai’an city, shandong province, tai’an 271000, china 1. introduction large-scale tourism investment abroad refers to the economic behavior of undertaking large-scale tourism investment projects abroad under the guidance of the host country or region government, with foreign tourism enterprise group as the main body. characterized by large investment scale, long cycle and complex social influence, it has important use function and great social significance. in recent years, chinese enterprises have invested more than $100 billion in 10 large-scale tourism projects abroad. the strategy of investing large-scale tourism projects abroad has become one of china’s important strategies to help developing countries improve their people’s livelihood, provide assistance and fulfill its international responsibilities [1]. the proposal and implementation of the belt and road (b&r) initiative provides an opportunity for the rapid development of china’s large-scale tourism investment project abroad. in 2017, dalian wanda group invested as much as $10 billion in malaysia’s dama city tourism investment project and india’s haryana industrial city project. in paris, china invested more than 3 billion euros in a large-scale cultural tourism project. however, the loss rate is as high as 90% and the losses run into hundreds of billions of dollars. the $3.6 billion bahamas resort project conducted by china state construction engineering group co. ltd., which would have been completed in 2015, went bankrupt due to labor issues. a 180-million dollar project in cancun, mexico, went bankrupt in 2014 as a result of environmental problems of mangroves and the protection of bird homes. these failures make the complex problem of risks of chinese large-scale investment project abroad once again a hot topic and a global challenge theoretically and practically. current research on investment risks mainly focuses on the economic ones, while social responsibility risk is far less discussed. domestic evaluation of social responsibility risk is mainly based on the perspective of enterprises rather than the b&r largescale tourism investment abroad. there are even fewer systematic and quantitative analysis studies of social responsibility risk of b&r large-scale tourism investment project abroad based on international organization for standardization (iso) 26000. according to the six core subjects of iso 26000, human rights, labor practices, the environment, fair operating practices, consumer issues and community involvement and development, combined with the background of the b&r and the current national conditions of china, this paper a r t i c l e i n f o article history received 16 september 2020 accepted 07 january 2021 keywords tourism investment social responsibility risk reconstruction a b s t r a c t the strategy of investing large-scale tourism projects abroad has become one of the most important strategies for china to help developing countries to improve their people’s livelihood, provide assistance and practice china’s international responsibility. the proposal and implementation of the belt and road initiative (b&r for short) provides an opportunity for the rapid development of china’s large-scale tourism investment project abroad. current research on investment risks mainly focuses on economic ones, while social responsibility risk is far less discussed. there are even fewer risk studies based on the international organization for standardization (iso) 26000 guidance on social responsibility, not to mention a systematic evaluation and quantitative analysis of social responsibility risk of b&r large-scale tourism investment project abroad. based on the six core subjects and 35 issues listed in iso 26000 guidance on social responsibility, with the method of the analytic hierarchy process, this paper fixes and sorts the evaluation indexes of social responsibility risk, so as to reconstruct the assessment for social responsibility risk of large-scale tourism investment project abroad. this assessment consists of three first-level value dimensions, nine second-level value dimensions and 21 third-level value dimensions. the results show: in the social responsibility risk assessment for b&r large-scale tourism projects abroad, stress should be put not only on project implementation and project selection, but also on project follow-up management. among the 21 third-level dimensions, top 10 dominant indexes are listed in turn as follows: dispute treatment, 0.0727, community environment 0.0610, education and awareness 0.0529, community participation 0.0499, skills and technology 0.0426, union organizations 0.0193, vulnerable groups 0.0168, respect for property rights 0.0163, ecological restoration 0.0160, and work environment condition 0.0156. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 348566634@qq.com †supported by: social science planning project of shandong province (digital shandong special project) (20csdj62); china postdoctoral science foundation project (2017m620958). journal of risk analysis and crisis response vol. 11(1); april (2021), pp. 1–9 doi: https://doi.org/10.2991/jracr.k.210111.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-1325-5841 http://creativecommons.org/licenses/by-nc/4.0/ mailto:348566634%40qq.com?subject= https://doi.org/10.2991/jracr.k.210111.002 https://www.atlantis-press.com/journals/jracr 2 c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 develops and innovates the social responsibility risk assessment of b&r large-scale tourism investment project abroad by identifying the essential dimensions to lower investment risks and sorting the key dimensions so as to promote the sustainable development of b&r large-scale tourism investment project abroad. 2. dimensions of social responsibility risk assessment 2.1. the connotation and evaluation of social responsibility risk assessment risk generally refers to the uncertainty that is inconsistent with the expected goal, including the uncertainty of the occurrence of events, the uncertainty of losses, as well as the uncertainty of both occurrence possibility and loss degree. social responsibility risk of investment projects refers to the uncertainty of the future investment income due to the failure to fulfill the social responsibility during the implementation process and the follow-up management of the investment project. investors may take the risk of income loss, or even the loss of principal, which is mainly manifested as natural risk, social risk, economic risk, technical risk, humanistic risk and cultural risk, etc. (table 1). 2.2. dimensions of social responsibility risk assessment ghosh and jintanapakanont [2] applied the method of factor analysis to identify key risk factors in engineering projects. huseby and skogen [3] incorporated decision-making behavior uncertainty into the risk evaluation indexes of investment projects. jiang et al. [4] adopted the method of fuzzy clustering and fuzzy similarity to evaluate the risks of investment projects. irene hall et al. [5] studied the impact of 3215 emergencies on investment projects in the united states from 1990 to 1992. malone et al. [6] considered shared dependency, flow dependency, co-dependency and the interdependence among them as the dominant factors of complex risks of investment project. according to the fuzzy set theory, knight and fayek [7], lee and daniel [8], lin and chen [9], sadiq and husain [10], motawa et al. [11], incorporated cost risk, safety risk, bidding risk, environmental risk and project change risk into investment project evaluation. su and wang [12] pointed out the influence of environment on investment decisions. jia [13] insisted that top-level decision of the government have a great impact on investment projects. han et al. [14], zhou and ren [15], huang and tang [16] believed that social responsibility affected the direction of investment projects. meng et al. [17], zheng [18], lin [19], wang et al. [20], zhou and yu (2005) [21], hong and li [22], hu et al. [23], zhou and wang [24] etc. regarded energy consumption, local residents’ demands, carbon emissions, performance, emergency and income inequality as the key dominant factors of investment projects. feng and zhang [25] designed the risk assessment indexes of agricultural investment projects from the perspectives of external environmental risks, technological risks and internal risks of enterprises. li [26] used the balanced scorecard method to design the index system of complex risk assessment of engineering projects. the above experts and scholars mainly conducted research from the perspective of economic risks. in terms of social responsibility risk, sun et al. [27] conducted the research from the point of institutional quality, green innovation and energy efficiency and energy policy (table 2). sun et al. [28] studied the environmental sustainability performance of south asia. yang and chen (2020) [29], lv and ge (2020) [30], studied social responsibility risks of investment abroad from the perspectives of lacking environmental awareness, improper handling of labor relations, conflicts with local residents, and the environment. zhang et al. (2016) [31], analyzed the risks faced by chinese investment in foreign infrastructure projects based on the environment of the host country, complexity of infrastructure projects and stakeholders. li (2016) [32] analyzed the risk of foreign investment from five dimensions of politics, management, culture, law and nature. pu (2015) [33] analyzed the risks of chinese railway enterprises’ foreign investment projects from the perspectives of politics, economy and society. yang and pi (2019) [34] believed that china’s outward foreign direct investment faced risks of politics, terrorism, nationalism, policies and laws, exchange rate, cultural conflicts and other aspects. other scholars and enterprises have studied china’s foreign large-scale investment projects, including social responsibility risks of some large-scale tourism investment projects from different perspectives, such as, huang (2011) [35] and wang (2013) [36], from the perspective of respects for human rights, feng (2005) [37] and wang (2005) [38] from the point of consumers’ issues, wanhua group, shan dong province, from the angle of fair operating practices, and he (1995) [39], zhao (2005) [40], jiang (2009) [41], and zhang (2013) [42] etc. from the point of environmental protection. however, in terms of social responsibility risks, the above researches are carried out only from a certain aspect or angle without a systematic research based on any international standards of social responsibility, let alone a social responsibility risk assessment for the b&r large-scale tourism investment projects abroad. therefore, based on the standards of iso 26000, the six core subjects, namely human rights, labor practice, environment, fair operation practice, consumer issues, community and development are incorporated as the dimensions of social responsibility risk assessment of the b&r large-scale tourism investment projects abroad. 3. iso 26000 guidance on social responsibility the standard of iso 26000 guidance on social responsibility is the first relatively unified international social responsibility standard in the world. since the 1990s, the concept of social responsibility has table 1 | connotation of investment risks perspectives experts and scholars uncertainty of occurrence of events willett (1901), mowbray (1995), williams (1985), knight (2003) uncertainty of loss rosenb (1972a and b), haynes (1985) possibility of occurrence and loss degree williams (1993), tumer (1992), shuzhen zhu (2002) source: according to the relevant data collation. c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 3 table 2 | typical social responsibility assessment dimensions representative authors/organizations factors for assessment robins philanthropy, pricing issues, employee relations, resource protection, product quality drucker (1992) [43] employees, community, clients the ftse social responsibility index stakeholder, environment, labor in supply chain, human rights, climate change and anti-bribery american economic priorities the status of women, charitable giving, military contracts, the status of people of color, information disclosure, animal testing, environmental records, community service, south africa, nuclear power, and so on organization for economic co-operation and development (oecd) (2001) information dissemination, general policy, environment, labour relations, consumer rights and interests, combating bribery, science and technology, competition and taxation mallin et al. (2014) [44] social responsibility information disclosure index, etc. china textile industry association (2005) labour contracts, management systems, forced labour, child labour, pay and benefits, working hours, discrimination, sexual harassment and abuse, trade unions and collective bargaining, and occupational safety and health li [45] human rights protection, labor rights and interests, social responsibility management, public welfare and business ethics china business council for sustainable development (2007) human rights, resource conservation and ecological protection, competition and cooperation, employees, environmental performance, products and services, cultural diversity, community building and public welfare, and public relations state-owned assets supervision and administration commission of the state council of china (2008) adhering to honest and lawful business practices, continuously improving profitability, effectively improving product quality and service, strengthening resource conservation and environmental protection, advancing independent innovation and technology, ensuring production safety, safeguard the legitimate rights and interests of employees, and participating in public welfare programs source: according to the relevant data collation. set off a great wave and pushed the development of global social responsibility movement. at first, only corporate social responsibility for enterprises was proposed, namely, the production regulations of enterprises. the international labor organization is committed to promoting international labor standard; based on 10 principals of global compact, the united states established the united nations global compact office and the social accountability international; in 1997, the american civil society organization issued the social responsibility standard of thirdparty certification (sa8000), guide to sustainable development reporting (latest version g3) and global reporting initiative, etc. (see table 3). however, all the above standards and guidance target narrowly and have certain limitations. how to uniformly define social responsibility, especially how to practice and measure social responsibility has become a common research subject for the international community. at the request of the committee on consumer policy, iso 26000, an international social responsibility standard, was developed by the technical management bureau and the plan of iso 26000 project was carried out in 2005. after 5 years’ formulation and having reached a consensus between developed and developing countries and among various stakeholders involving more than 90 countries and more than 40 international and regional organizations for social responsibility, iso 26000 guidance on social responsibility was officially promulgated by the iso on november 1, 2010 (see table 4 for details). as a key definition of iso 26000, social responsibility is defined as “responsibility of an organization for the impact of its decisions and activities on society and the environment, through transparent and ethical behaviour that contributes to sustainable development, including health and the welfare of society, takes into account the expectations of stakeholders, is in compliance with applicable law and international norms of behavior, and is integrated throughout the organization and practised in its relationships.” iso 26000 mainly focuses on organizational governance and involves six core subjects and 35 issues (see table 5 for details). 4. social responsibility risk assessment and positioning of b&r large-scale tourism investment project abroad 4.1. the hierarchy model of social responsibility risk assessment theoretically speaking, the construction of an evaluation index system should be carried out from three levels: overall design, application operation and a clear goal. similarly, the principles for constructing a comprehensive evaluation index system for social responsibility risks should also be considered from these three aspects. while the overall design of the assessment is supposed to be scientific, systematic, comprehensive and complete, and reasonably reflect the social responsibilities of the investment project from all aspects, the indicators selected in every level the system need to be measurable, attainable and comparable, so as to facilitate the application and operation of indicators in assessment. in addition, the indicator system should be future-oriented, sustainable and forward-looking, in order to guide the improvement of the sustainable development capacity of investment projects. due to geopolitics, the host country’s political environment and monopoly of chinese large tourism investment project abroad, 4 c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 table 4 | formulating process of iso 26000 guidance on social responsibility stages time contents stage 1 2001 an iso subordinate, consumer policy committee (copolco) submitted an application to the council. the application was approved and a strategic advisory group was established. 2002 copolco presented a feasibility report on the standardization of social responsibility to the international organization for standardization (iso) council. 2003 iso set up a special strategy advisory group to develop social responsibility standards. stage 2 2004 technical management bureau (tmb) of iso decided to launch the project of iso 26000 guidance on social responsibility. 2005 iso 26000 guidance on social responsibility working group was established, and they held the first meeting. the development of iso 26000 guidance on social responsibility was officially initiated. 2008 the expert group submitted the iso 26000 project proposal and draft to the committee of social responsibility, and the committee draft was formed after a second discussion. 2010 the draft was amended by the countries and stakeholders involved in the formulation of iso 26000, and the draft of international standard was confirmed, only to be approved. 2010 iso 26000 guidance on social responsibility standards were officially released. source: according to the relevant data collation. table 3 | representative international norms of social responsibility norms countries using the standard contents so14000 (1996) more than 130 countries and regions use it, thus making it a recognized standard for global quality and environmental management systems 17 elements involving environmental policy, planning, fact and operation, inspection and corrective measures and management review sa8000 (1997) 2919 organizations covering 65 industries of 62 countries; top ten countries and regions of certification: india, china, italy, brazil, pakistan, vietnam, spain, romania, taiwan china, philippines child labour, health and safety, forced labour, discrimination, freedom of association and the right to collective bargaining, remuneration, disciplinary measures, management systems and working hours un global compact (1999) with 8700 businesses in more than 130 countries, it is the world’s largest voluntary social responsibility initiative 10 principles of global consensus in the fields of human rights, labour, environment and anti-corruption oecd guidelines for transnational corporations (2000) signed up by 34 governments, and gradually promoted globally general policy, information disclosure, labour relations, environment, fight against bribery, consumer interests, science and technology, competition and taxation, etc. global reporting initiative (gri) (2000) by november 2014, 20,881 gri reports had been released beyond environment, it expands the reporting framework to society, environment, economy and governance areas source: according to the relevant data collation. there are huge social responsibility risks in investing abroad. china railway construction’s bid for a high-speed rail line in mexico and cosco group’s bid for a 67% stake in greece’s piraeus port failed due to the factional conflicts and regime change in mexico and greek. the reason why japan has relatively fewer social responsibility risks in large investment projects abroad is that japanese enterprises carry out international cooperation with developed countries or host countries of the investment. zhang et al. (2014) [46] studied the influence of host behavior on social responsibility risk of china’s large investment projects abroad, and the results showed that geopolitics and political environment of the host country is one of the main factors for the social responsibility risk of china’s investment. therefore, this paper incorporates international cooperation and joint development, domestic political risks and geopolitical environment of the host country into the indexes of social responsibility risks of the b&r large-scale tourism inv estment projects abroad. what is more, according to the 21 thirdlevel dimension indicators extracted from the six core subjects and 35 issues listed in iso 26000, this paper constructs three first-level indicators including project selection, project implementation and project follow-up management. therein, project selection consists of international collaborative development, domestic political environment and regional political environment of the host country. human rights, labor practice, environment and fair operation practice are included in project implementation, and consumer issues, community involvement and development included in project follow-up management. therefore, the hierarchy model of social responsibility risk assessment for the b&r large-scale tourism investment project abroad is constructed with three firstlevel value dimensions, nine second-level value dimensions and 21 third-level value dimensions. 4.2. weight and consistency test of social responsibility risk indicators 4.2.1. weight measurement and consistency test of stratified target in decision-making target according to the characteristics and evaluation requirements of social responsibility risk assessment of the b&r large-scale tourism investment projects abroad, this paper adopts the c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 5 table 5 | iso 26000 guidance on social responsibility core subjects issues dimensions extracted in this paper human rights a21 due diligence due diligence a211 human rights risk situations avoidance of complicity handling of complaints a212 resolving grievances discrimination and vulnerable groups vulnerable groups a213 civil and political rights economic, social and cultural rights economic, social and cultural rights a214 fundamental principles and rights at work labour practices a22 employment and employment relationships employment and employment relationships a221 social dialogue conditions of work and social protection conditions of work and social protection a222 health and safety at work human development and training in the workplace human development and training in the workplace a223 the environment a23 prevention of pollution prevention of pollution a231 sustainable resource use sustainable resource use a232 climate change mitigation and adaptation global warming a233 protection of the environment, biodiversity and restoration of natural habitats ecological restoration a234 fair operating practices a24 promoting social responsibility in the value chain promoting social responsibility in the value chain a241 anti-corruption political responsibility awareness a242 responsible political involvement fair competition fair competition a243 respect for property rights respect for property rights a244 consumer issues a31 protecting consumers’ health and safety fundamental rights of consumers a311 consumer data protection and privacy access to essential services consumer service, support, and complaint and dispute resolution handling of disputes a312 fair marketing, factual and unbiased information and fair contractual practices education and awareness education and awareness a313 sustainable consumption community involvement and development a32 community involvement community involvement a321 wealth and income creation community environment a322 education and culture health employment creation and skills development technology development a323 technology development and access social investment source: iso 26000 guidance on social responsibility. analytic hierarchy process (ahp) to evaluate the social responsibility risk assessment index. five experts in the field of tourism investment risks were invited to score the importance (1–9) of each value dimension. according to the judgment matrix of the impact degree of projects selection, project implementation and project follow-up management, as proposed in the assessment, the maximum eigenvalue λmax equals 3.0120, and the weight on the reconstructed social responsibility risk assessment is 1.0, the consistency ratio (cr) = 0.0115 < 0.1. the normalized weight vectors corresponding to the maximum eigenvalues are presented in table 6. from the weight vector column, it can be seen that the weight of project selection on reshaping the social responsibility risk assessment is 47.93%, showing that project selection is the dominant factor of the social responsibility risk. project follow-up management accounts for 30.68% and project implementation for 21.39%, both playing considerable roles, between which project follow-up management has a stronger influence on the social responsibility risk of investment project. 4.2.2. weight measurement and consistency test of stratified target indexes based on the assessment constructed above, weight measurement and consistency of the stratified targets in the alternative plan are tested respectively. weight of each index in the stratified target is calculated and the consistency is tested. (1) project selection a1. table 6 | stratified target judgment matrix in decision-making target reshaping risk indexes a1 a2 a3 weight vector w project selection a1 1 5/2 7/5 0.4793 project implementation a2 2/5 1 7/9 0.2139 project follow-up management a3 5/7 9/7 1 0.3068 λmax= 3.0120, cr = 0.0115 < 0.1 weight on the overall target 1.0 cr, consistency ratio. 6 c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 table 8 | judgment matrices of targets in the alternative plan a21 a211 a212 a213 a214 w a22 a221 a222 a223 w a23 a231 a232 a233 a234 w a211 1 5/6 9/8 5/4 0.2554 a221 1 6/5 4/3 0.3874 a231 1 4/5 6/5 3/4 0.2281 a212 6/5 1 1/3 7/5 0.2124 a222 5/6 1 8/9 0.3130 a232 5/4 1 7/5 8/9 0.2765 a213 8/9 3 1 6/5 0.3283 a223 3/4 9/8 1 0.2997 a233 5/6 5/7 1 4/5 0.2057 a214 4/5 5/7 5/6 1 0.2039 – – – – – a234 4/3 9/8 5/4 1 0.2897 λmax= 4.2262, cr = 0.0847 < 0.1 λmax= 3.0055, cr = 0.0053 < 0.1 λmax= 4.0076, cr = 0.0029 < 0.1 weight on the overall target 0.0513 weight on the overall target 0.0498 weight on the overall target 0.0554 a24 a241 a242 a243 a244 w a31 a311 a312 a313 w a32 a321 a322 a323 w a241 1 5/4 1 7/9 0.2470 a311 1 1/2 2/5 0.1809 a321 1 4/5 6/5 0.3253 a242 4/5 1 7/9 7/8 0.2136 a312 2/1 1 9/5 0.4740 a322 5/4 1 7/5 0.3974 a243 1 9/7 1 6/7 0.2548 a313 5/2 5/9 1 0.3451 a323 5/6 5/7 1 0.2774 a244 9/7 8/7 7/6 1 0.2846 – – – – – – – – – – λmax= 4.0125, cr = 0.0047 < 0.1 λmax= 3.0735, cr = 0.0707 < 0.1 λmax= 3.0005, cr = 0.0005 < 0.1 weight on the overall target 0.0574 weight on the overall target 0.1534 weight on the overall target 0.1534 table 7 | stratified target judgment matrices a1 a11 a12 a13 w a2 a21 a22 a23 a24 w a3 a31 a32 w a11 1 1 6/5 0.3539 a21 1 4/5 8/9 6/5 0.2399 a31 1 1 0.5 a12 1 1 8/9 0.3202 a22 5/4 1 7/9 7/9 0.2327 a32 1 1 0.5 a13 5/6 9/8 1 0.3259 a23 9/8 9/7 1 4/5 0.2589 – – – – – – – – – a24 5/6 9/7 5/4 1 0.2685 – – – – λmax= 3.010, cr = 0.0096 < 0.1 λmax= 4.0554, cr = 0.0208 < 0.1 λmax= 2.000, cr = 0.0000 < 0.1 weight on the overall target 0.4793 weight on the overall target 0.2139 weight on the overall target 0.3068 cr, consistency ratio. according to the judgement matrix, the maximum eigenvalue is 3.010, and its weight on the reconstructed social responsibility risk assessment is 0.4793. (2) project implementation a2. according to the judgement matrix, the maximum eigenvalue is 4.0554, and the weight is 0.2139. (3) project follow-up management a3. according to the judgement matrix, the maximum eigenvalue is 2.0000, and the weight is 0.3068. three stratified target judgment matrices and the normalized weight vector corresponding to the maximum eigenvalue are presented in table 7. 4.2.3. weight measurement and consistency test of indexes in the alternative plan the reconstructed assessment of the social responsibility risk of the b&r large-scale tourism investment project abroad includes a total of 21 three-level value dimensions, and the corresponding index judgment matrix is shown in table 8. in this paper, the maximum eigenvalue vector is calculated by integral method. (1) normalization of column elements in the judgment matrix. the general term of elements is � �h h h i j nij ij ij n= = ∑1 1 2( , , , , ). (2) sum of rows in the judgment matrix � � �w h i j ni ij n = =∑1 1 2( , , , , ). (3) normalizing the vector �wi . w w w i j ni i j n= = ∑ � � � 1 1 2( , , , , ). w = (w1, w2, . . . , wn) t, the approximate solution of all eigenvectors. (4) the maximum eigenvalue of the judgment matrix: lmax = ( ) ∑ hw nw i i n 1 . 5. reconstructing and positioning the weight distribution of social responsibility risk assessment for the belt and road large-scale tourism investment projects abroad the final weight distribution of social responsibility risk assessment for the large-scale tourism investment projects abroad is shown in figure 1. the weights of the stratified targets (project selection a1, project implementation a2 and project follow-up management a3) on the decision-making target (social responsibility risk assessment for investment projects) are 0.4793, 0.2139 and 0.3068 respectively. the weights of the sub-layers in the stratified targets are indicated as follows: international collaborative development 0.1696, host domestic political environment 0.1535, host geographic political environment 0.1562, human rights 0.0513, labor practice 0.0498, environment 0.0554, fair operation practice 0.0574, consumer issues 0.1534, community involvement and development 0.1534. in the alternative plan, the dominant indexes are listed in turn as follows: the index of vulnerable group, a sub-layer to human rights, accounts for 0.068, labor organization, sub-layer to labor practice, 0.0193, ecological restoration to the environmental 0.0161, fair operation practice to respect for property rights 0.0163, dispute handling to consumer issues 0.0727, and community environment to community participation and development 0.0610, thus it can be seen that the top 10 dominant indexes in the third level are: dispute handling 0.0727, community environment 0.0610, education and awareness 0.0529, community involvement 0.0499, skills and technology 0.0426, trade union 0.0193, vulnerable groups 0.0168, respect for property rights 0.0163, ecological restoration 0.0160 and working conditions 0.0156. since the second-level indicators c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 7 figure 1 | weight distribution in the reconstructed social responsibility risk assessment for the belt and road large-scale tourism investment project abroad. (a11, a12, and a13) are not sub-divided, project management after the completing the project bears significantly high social responsibility risks, far higher than the risks during the project construction. therefore, under the premise of cautious selection of projects, more focus should be laid on preventing the social responsibility risks in the follow-up operation and management of the b&r large-scale tourism investment projects abroad. 6. conclusion and discussion traditional economic risks are no longer the dominant risks for china’s large-scale tourism investment abroad. instead, social responsibility risks, such as labor conflict, environmental factors, cultural risks, community involvement and development, and educational awareness have taken the place in the b&r largescale tourism investment projects abroad. if corresponding social responsibility risks cannot be effectively handled, safety of china’s b&r large-scale tourism investment projects abroad may be threatened. with the promulgation and implementation of iso 26000, social responsibility standards have become a tool for some countries to carry out green barriers and it has been an irreversible international trend for enterprises to take social responsibility with economic globalization. therefore, it is particularly urgent and significant to study the social responsibility risk of china’s b&r large-scale tourism investment projects abroad based on iso 26000. it is also an urgent need to establish a forward-looking and scientific social responsibility risk assessment mechanism in the current practice of social responsibility risk management. the theoretical contribution of this paper is to propose the assessment mechanism of social responsibility risks in china’s large-scale tourism investment, further stressing the importance of management in not only project implementation, but also in project selection and project follow-up. based on the six core subjects, namely human rights, labor practices, the environment, fair operating practices, the consumers and community involvement and development, and 35 issues listed in iso 26000, this paper reconstructs a social responsibility risk assessment for the b&r large-scale tourism investment projects abroad and positions each index according to their weight on the principle of being scientific and systematic, complete and comprehensive, measurable, comparable, sustainable and forward-looking (xiao, 2016) [47]. the assessment consists of three first-level value dimensions, namely project selection, project implementation and project follow-up management; nine second-level value dimensions including international cooperation and joint development, domestic political environment of host country, regional political environment of host country, etc.; and 21 third-level value dimensions including indicators of due diligence, complaints processing, vulnerable groups, and economic, social and cultural rights. (1) it is an important subject to quantify the social responsibility risks, to reconstruct the comprehensive evaluation system and thus to identify each evaluation index according to iso 26000. social responsibility risk assessment has become an essential part to select, assess and reduce social responsibility risks in the b&r large-scale tourism investment projects abroad. (2) according to the reconstructed social responsibility risk assessment, the management and supervision of social responsibility risks of project investment should be strengthened. in the whole process of large tourism investment project management, the social responsibility risk of project selection is increasingly highlighted. in the first-level index of project selection (a1), the importance of international cooperative development, domestic political environment of the host country, and regional political environment of the host country should be emphasized so as to strengthen the management of social responsibility. in the first-level index of project implementation (a2), social responsibility is to be 8 c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 fulfilled from four aspects: human rights, labor practice, the environment and fair operating practices; while in the third-level index of project follow-up management (a3), social responsibility is to be practiced from the perspectives of consumer issues and community involvement and development. therefore, based on the iso 26000 international standard, conducting whole-process management in b&r large-scale tourism investment projects abroad is of great significance to reduce social responsibility risks and facilitate reinvestment so as to realize the sustainable development of the b&r large-scale tourism investment projects abroad. there are some limitations in this paper. firstly, instead of adopting a comprehensive method to elaborate the research, we use only the method of ahp to construct the assessment mechanism of social responsibilities risks. secondly, it is necessary to further verify the validity of the reconstructed assessment in corresponding cases, which we fail to do due to lack of space. the next step is to further verify the scientificity and rationality of the comprehensive assessment of social responsibility risks constructed in this paper. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution in writing this paper, based on the international organization for standardization (iso) 26000 guidance on social responsibility, cl proposes the key points of social responsibility risk assessment for the belt and road large-scale tourism investment project abroad. with the method of analytic hierarchy process, xl positions and sorts the indicators of social responsibility risk assessment, and reconstructs the assessment for the belt and road large-scale tourism investment project abroad. about the authors chen lv (1981-), male, born in jining, shandong province. doctor of economics, professor of taishan university and post-doctor of china tourism academy. main research interests: industrial upgrading, industrial policy and tourism investment risk, etc. xiaoyan li (1988-), female, born in huize, yunnan province. master of arts, associate professor of taishan university. research interests: tourism safety and tourism culture. acknowledgment as a phased achievement, this paper owes to the support of the talent introduction and education project of youth innovation team in colleges and universities of shandong province (tourism investment risk and management innovation team) (95), social science planning project of shandong province (digital shandong special project) (20csdj62) and china postdoctoral science foundation project (2017m620958). references [1] lv c, lin hc. dynamic monitoring and decision making system for large-scale foreign investment projects. reform 2016;5:89– 100 (in chinese). [2] ghosh s, jintanapakanont j. identifying and assessing the critical risk factors in an underground rail project in thailand: a factor analysis approach. int j project manage 2004;22:633–43. [3] huseby ab, skogen s. dynamic risk analysis: the dynrisk concept. int j project manage 1992;10:160–4. [4] jiang w, deng l, chen l, wu j, li j. risk assessment and validation of flood disaster based on fuzzy mathematics. prog nat sci 2009;19:1419–25. [5] irene hall h, ramana dhara v, price-green pa, kaye we. surveillance for emergency events involving hazardous substances – united states, 1990-1992. surveill summar 1994;43:1–6. [6] malone tw, crowston k, lee j, pentland b, dellarocas c, wyner g, et al. tools for inventing organizations: toward a handbook for organizational processes. manag sci 1999;45:296–454. [7] knight k, fayek ar. use of fuzzy logic for predicting design cost overruns on building project. j constr eng manage 2002;128:503–12. [8] lee s, daniel wh. predictive tool for estimating accident risk. j constr eng manage 2003;129:431–6. [9] lin ct, chen yt. bid/no-bid:decision-making – a fuzzy linguistic approach. int j project manage 2004;22:585–93. [10] sadiq r, husain t. a fuzzy-based methodology for an aggregative environmental risk assessment: a case study of drilling waste. environ model softw 2005;20:33–46. [11] motawa ia, anumba cj, el-hamalawi a. a fuzzy system for evaluating the risk of change in construction projects. adv eng softw 2006;37:583–91. [12] su jj, wang xm. the spatial-temporal evolution characteristics and driving factors of china’s tourism investment structure benefit. geogr geo-inform sci 2020;36:117–24. [13] jia k. government public investment, social investment and economic and social development. j southwest minzu univ (human soc sci) 2010;31:138–41 (in chinese). [14] han pc, xue l, wang wj. corporate innovation, social responsibility and corporate value — a case study of small and medium-sized enterprises. forum sci technol china 2020;93–9 (in chinese). [15] zhou sp, ren jz. corporate social responsibility management theory and practice in china. j chinese acad govern 2010;66: 38–41 (in chinese). [16] huang xj, tang lx. political association, equity financing and the change of the raised funds. soft sci 2012;26:123–6 (in chinese). [17] meng b, li f, yang y. research on the evaluation of corporate social responsibility in transportation industry based on circular correction. sci res manage 2020;41:174–84 (in chinese). [18] zheng yx. don’t be blind in energy conservation and emission reduction. study pract 2011;24–32 (in chinese). [19] lin zm. an imperative trend of the transformation and upgrading of the chinese economy. econ rev j 2014;17–22 (in chinese). [20] wang yn, li zj, zhang br. the influencing factors and mechanism of social responsibility of transnational corporations under “one belt and one road”. econ probl 2020;82–90 (in chinese). [21] cong shh, zhou w, yu n. formulating indicators for the evaluation of public expenditure performance. finance trade econ 2005;37–41+97. https://doi.org/10.1016/j.ijproman.2004.05.004 https://doi.org/10.1016/j.ijproman.2004.05.004 https://doi.org/10.1016/j.ijproman.2004.05.004 https://doi.org/10.1016/0263-7863%2892%2990005-t https://doi.org/10.1016/0263-7863%2892%2990005-t https://doi.org/10.1016/j.pnsc.2008.12.010 https://doi.org/10.1016/j.pnsc.2008.12.010 https://doi.org/10.1016/j.pnsc.2008.12.010 https://www.jstor.org/stable/24675549 https://www.jstor.org/stable/24675549 https://www.jstor.org/stable/24675549 https://doi.org/10.1287/mnsc.45.3.425 https://doi.org/10.1287/mnsc.45.3.425 https://doi.org/10.1287/mnsc.45.3.425 https://doi.org/10.1061/(asce)0733-9364(2002)128:6(503) https://doi.org/10.1061/(asce)0733-9364(2002)128:6(503) https://doi.org/10.1061/(asce)0733-9364(2002)128:6(503) https://doi.org/10.1061/(asce)0733-9364(2003)129:4(431) https://doi.org/10.1061/(asce)0733-9364(2003)129:4(431) https://doi.org/10.1016/j.ijproman.2004.01.005 https://doi.org/10.1016/j.ijproman.2004.01.005 https://doi.org/10.1016/j.envsoft.2003.12.007 https://doi.org/10.1016/j.envsoft.2003.12.007 https://doi.org/10.1016/j.envsoft.2003.12.007 https://doi.org/10.1016/j.advengsoft.2006.01.006 https://doi.org/10.1016/j.advengsoft.2006.01.006 https://doi.org/10.1016/j.advengsoft.2006.01.006 c. lv and x. li / journal of risk analysis and crisis response 11(1) 1–9 9 [22] hong xj, li jc. a review of the polarization measurement method and the polarization of chinese residents’ income. econ res j 2007;139–53 (in chinese). [23] hu b, shen l, gong j. the market value effect of social responsibility of china’s listed tourism enterprises: from the perspective of marketing strategy analysis. tour tribune 2006;35–42 (in chinese). [24] zhou jq, wang ts. foreign direct investment, economic growth and co2 emissions. j beijing inst technol (soc sci ed) 2014;30–8 (in chinese). [25] feng z, zhang sd. research on the dynamic evaluation of risk index and risk coefficient of agricultural high-tech project investment. sci technol manage res 2010;14:88–90 (in chinese). [26] li l. design of performance evaluation index of public works project based on balanced scorecard method. soc sci 2007;168– 70 (in chinese). [27] sun h, edziah bk, sun c, kporsu ak. institutional quality, green innovation and energy efficiency. energy policy 2019;135: 111002. [28] sun h, mohsin m, alharthi m, abbas q. measuring environmental sustainability performance of south asia. j clean prod 2019;251:119519. [29] yang zh, chen j. innovation and governance of corporate social responsibility in the digital and intelligent era. j shanghai univ of finance econ 2020;22:33–51 (in chinese). [30] lv wq, ge xf. social capital theory and community tourism operators’ social responsibility. thinking 2020;46:156–164 (in chinese). [31] zhang shb, yang qb, zhang yw. the risk management of infrastructure investment projects. int project contracting and labour service 2016;74–75 (in chinese). [32] li yw. a rational approach to the influence of tpp on china: opportunities, challenges and countermeasures. contemporary econ 2016;10–13 (in chinese). [33] pu y. analysis of the risk factors of chinese railway enterprises’ foreign investment and countermeasures. business 2015;160 (in chinese). [34] yang hr, pi jc. the analysis of corporate social responsibility in the vertical market structure. mod econ sci 2019;41:97–107 (in chinese). [35] huang zhx. for the hundreds-arces demolition project. people’s political scene 2011;48 (in chinese). [36] wang lf. the human rights policy of china’s overseas investment enterprises. human rights 2013;37–40 (in chinese). [37] feng f. a brief analysis of consumer group litigation——on the protection of consumers in anti unfair competition law. contemporary manager 2005;236–237 (in chinese). [38] wang b, zhou x, yang ch. on consumer problem and its harmfulness. soc sci yunan 2005;62–67 (in chinese). [39] he jw. market demand and product quality strategy. acad res 1995;43–45 (in chinese). [40] zhao xb. deepen the enterprise environment research and improve the enterprise management level. wuhan univ j (philosophy soc sci) 2005;437 (in chinese). [41] jiang fx. academic record of the track of reform and development. jianghai acad j 2009;20–21 (in chinese). [42] zhang xq. interpretation of outdoor heat resource wells for water source heat pump system wells. green building 2013;5:41– 42 (in chinese). [43] drucker pf. management: tasks, responsibilities, practices. beijing: china machine press; 2007. [44] fabrizi m, mallin c, michelon g. the role of ceo’s personal incentives in driving corporate social responsibility. j business ethics 2014;124:311–326. [45] li lq. a theoretical and empirical study on corporate social responsibility evaluation. south china j econ 2006;105–18. [46] zhang xt, zheng yj, yue ys. heterogeneity of the impact of host country’s government behavior on chinese enterprises’ overseas investment decision—based on the perspective of enterprise size. acad forum 2014;37:30–36 (in chinese). [47] xiao hj. a study on the reform of corporate governance of public-policy-related state-owned enterprises. reform of econ sys 2016;5–11 (in chinese). https://doi.org/10.1016/j.enpol.2019.111002 https://doi.org/10.1016/j.enpol.2019.111002 https://doi.org/10.1016/j.enpol.2019.111002 https://doi.org/10.1016/j.jclepro.2019.119519 https://doi.org/10.1016/j.jclepro.2019.119519 https://doi.org/10.1016/j.jclepro.2019.119519 microsoft word decision-making model in the environment of complex structure data(no. 1301005) decision-making model in the environment of complex structure data fusheng yu school of mathematical sciences, beijing normal university beijing, 100875, china e-mail: yufusheng@bnu.edu.cn shihu liu school of mathematical sciences, beijing normal university beijing, 100875, china e-mail: liush02@126.com abstract for decision makers, the data property has a direct influence on the selection of decision making approaches and the reliability of decision results. because of the complexity and diversity of practical decision data, some traditional decision approaches are not very good at reflecting the actual problem. for this, we propose a decision making model in the environment of complex structure data. the aim of this model is to discover the underlying community structure of the data by taking all aspects of original information into account. in this paper, the considered data is diversity, not only in structure but also in representation. what is more, a missing data compensation method is proposed by considering the information losing situation in practical decision making problem. research shows that this model has great maneuverability. especially, the proposed decision model seems more consistent with the actual decision problem, than decision model with single data structure. keywords: graph data, weighting determination, information fusion, completion of incomplete data, decision analysis. 复杂结构数据环境下的决策模型 于福生 刘士虎 北京师范大学/数学科学学院，北京 100875 摘要：对决策者而言，数据的特性直接影响决策工具的取舍和决策质量的可靠性。由于现实决策数据的复 杂多变性，使得传统的一些决策模型不能够很好地实现决策目的。鉴于此，本文给出了一种基于复杂结构 数据的决策模型。该模型旨在充分利用现有数据信息的基础上，实现挖掘数据潜在社团结构的目的。本文 考虑的数据具有两大特性：一是结构的多样性，二是表示的多样性。甚至，考虑到实际问题中信息有损情 况的存在，本文给出了一种有损数据补偿的方法。不难发现，该决策模型可操作性强，相对于单一数据格 式下的决策模型，更贴近实际需要。 关键词：图数据，权重确定，信息融合，不完备数据完备化，决策分析 1. 引言 大到宇宙自然界，小到生活中的点点滴滴，风险无 处不在，无时不有。这其中，有的风险是不以人的 意志为转移的，如地震、台风、海啸等；有的风险 是伴随着人类的活动而出现的，如工程建设风险、 人员伤害风险、社会声誉风险等 1, 2。对于有人类参 与的风险，风险的产生既有技术层面上的主观原 因，又有社会层面上的客观原因。不管风险源自何 处，归于何类，风险的最终承担者都是我们人类。 于是，如何降低风险对人类的危害程度，是每一个 风险管理者需要解决的首要问题。 journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 103-109 published by atlantis press copyright: the authors 103 willieb typewritten text received 14 june 2013 willieb typewritten text accepted 28 june 2013 willieb typewritten text fusheng yu, shihu liu 为了最大程度地降低风险带来的损失，就需要 对风险做出一个科学合理的评估。通常用定性或者 定量的分析方法 3 研究风险发生的可能性及其后果 的严重程度。但是常言道，天灾人祸不单行。对于 实际问题的风险分析，往往存在诸多不确定性因 素。诸如问题的描述、数据的采集、分析方法或工 具的取舍，决策结果可靠性分析等。尤其对于问题 的描述，一贯采用多指标描述的策略。但是，指标 之间的交互影响，为问题的分析带来了困难。 简言之，不管问题有多么复杂，在风险分析过 程中，通常是把抽象的问题模型化。对于不同的实 际问题，可以结合具体的背景知识，建立相应的数 学模型。然后利用该模型，对潜在的风险做一量化 分析，在这个过程中对数据的处理显得至关重要。 因为人们所面对的数据，往往具有复杂多变且不确 定等诸多特点。甚者，对同一个问题，不同的数据 分析员所面对的是不同的数据库，而且这些数据库 之间是互相保密的。同时，不同的数据库，描述数 据的方式也可能不相同。如有的数据库采用多指标 描述法，有的采用关系描述法等。究其一点，数据 类型大体上可以分为两类：向量型数据和关系型数 据。通常，两种类型的数据是混合出现的。在风险 分析的过程中，如何充分利用这两种不同类型的数 据，对分析结果的可靠性有着巨大的影响。 基于此，本文对于具有复杂结构的数据的风险 分析问题，从三个方面展开了相关研究。(1) 对于向 量型数据中指标的权重确定方法的研究。熟知，对 于多指标问题，不同的指标对于知识的认知程度， 贡献通常是不一样的。故在分析过程中对于指标的 区别看待是很有必要的。故在本文中我们给出了基 于粒度的指标权重确定方法。同时，基于不同的模 式对指标权重的贡献也不一样这一假设，我们给出 了基于聚合算子的指标权重确定方法。(2) 对于数据 损失的问题，给出了一种损失数据的补偿方法。在 数据的采集、传输与存储过程中，出现数据的丢失 是不可避免的现象。在此，基于数据的相似性，我 们建立了一种基于“局部-整体”相似的丢失数据补 偿方法。并用虚拟的知识表达系统，分析了该方法 的可行性。(3) 对于复杂结构数据环境下的风险问 题，我们从聚类的角度，做了相关分析。旨在挖掘 出问题中潜在的“社团结构”或者“块结构”，为 决策者的进一步决策提供一个指导作用。 作为本节的结束，接下来我们给出本文的基本 框架。第 2 节简单的回顾一下本文需要的一些基本 概念，如复杂结构数据、信息粒度、信息聚合与聚 类分析等。第 3 节主要是对于向量型数据中指标权 重的确定，从信息粒度和聚合算子两个方面给出确 定方法。第 4 节是介绍向量型数据完备化的方法。 第 5 节是利用聚类的思想，挖掘对于具有复杂结构 数据的风险分析问题。 2. 预备知识 在该部分，我们对本文中用到的一些基本概念及其 相关知识做一简要介绍。如：图数据，知识粒度， 数据信息集成算子和聚类分析。对于其详细的介 绍，可以参阅相关文献 4,5。 定义 1. (复杂结构数据)一个图数据可以表示为 二元组 ( , )g v e  ，其中 v 代表该数据的模式集， e 代表模式之间的关系集。 对于一个只有有限个模式组成的图数据 g 而 言，我们用 1 2( )={ , , , }nv g x x x   表示顶点集，如果 每个模式 ix 是用 m 个指标来刻画的，则 1 1 11 1 1 ( ) . m m n n nm a a x x x v g x x x                   = (1) 进一步，用 ( )e g 表示顶点关系集，即对于任意的两 个对象 , ( )i jx x v g ， ( , ) ( )i jx x e g 当且仅当 ix 和 jx 存在某种关系。 显然，对于一个具有复杂结构的数据而言，对 于不同的指标，其值域表示也不尽相同。同时，描 述顶点之间关系的数据集 ( )e g 也未必只有一个。 定义 2. (知识粒度)给定向量型数据 ( )v g ，及在 任意指标 ia 下的划分 1 2{ , , , }i i i inp p p p  ，则关于 指标 ia 的知识粒度定义为 2 1 1 ( ) | |. n i ij j g a p n    (2) 针对决策数据表示的异同性，许多学者提出了 各种不同的数据集成算子 5，其一般的形式可以定义 如下。 定义 3. (数据集成算子)设 : nf r r  ，则称 1 2( , , ..., )nf a a a   为 1 2( , , ..., ) n na a a r   的加权集成，其 中 为权系数且满足 1 2 n+ + + =1   . 显然，集成算子 f 以及权系数 的取舍，依赖 于决策者的决策需要以及具体的决策问题。 数据的聚类分析 4 ，旨在把给定的数据集 x 分 成所期望的 k 类  1 2, ,..., kc c c c ，使得类间尽可能 远离，类内尽可能抱团。纵观现有的聚类算法，大 致可以划分为硬划分(hard partition）和软划分(soft partition)两种。在硬划分下，每个模式只能属于某 个特定的类。而在软划分下，对象是以一定的隶属 度(介于 0 和 1 之间的某个数)归属于某个类。如对于 向量型数据 x ，划分成 k 类，最优聚类使下述误差 函数 published by atlantis press copyright: the authors 104 decision-making model in the environment of complex structure data | | 2 1 1 ( , ) || || x k i c i c j c m x v     (3) 达到最小值，其中  1 2, ,..., km v v v 为类的中心。相 应的，其软划分，如经典的 fcm 算法，为求下述目 标函数 | | 2 1 1 ( , ) || || x k m ci i c i c j c m u x v     (4) 的最小值，其中 m 为模糊化因子。 在不引起歧义的前提下，下面符号在接下来的 节中是通用的： g 代表图数据， e 代表图数据中顶 点之间的关系， x 代表向量型数据， n 代表 g 或者 x 中顶点的个数， m 代表向量型数据的维数。在叙 述的过程中，我们对向量型数据 x 和向量型数据 ( )v g 在符号表示上不做进一步的区别。 3. 决策模型中指标权重的确定 该部分我们主要讨论决策问题中，衡量决策目标的 指标在决策过程中所扮演的角色。如果所有的指标 所扮演的角色相同，则认为其对决策结果的贡献一 样。这样，从权重的角度而言，所有指标的权重应 该相等。一般而言，不同的指标对于决策的结果， 所起的作用往往是不同的。于是，对不同的指标赋 予不同的权重，对决策结果认知度的提高，就显得 很有必要。决策的过程，实质上就是一个不断挖掘 已有数据集中潜在信息的过程。下面，我们给出两 种不同的权重确定方法，一是基于粒度思想的指标 权重确定方法；二是基于聚合算子的指标权重确定 方法。 3.1 基于粒度的权重确定方法 粒度这一概念，在特定层面上反映的是人类对知识 的认知程度。对于一个向量型的数据，在某一指标 下的粒度，可以刻画关于该指标对知识的认知能 力。而且，如果粒度值越小，则代表该指标对知识 的认知能力越强，反之，则认知能力越弱。 定义 4. 给定向量型数据集 x 以及关于任意指 标 ia 的粒度值 ( )ig a ，则该指标 ia 的权重定义为 1 1 ( ) (1 ( )) i i m j j g a g a       . (5) 显然，上式定义的指标的权重是归一化的。但 是，对于某个特定的决策问题，往往考虑的指标很 多。在此情况下，则有可能会出现很多指标的权重 很小的现象。对此，我们给出一种基于权重的指标 取舍方法。 定义 5. 给定向量型数据集 x 及任意指标 ia 的 权重 i 。对于事先假定的阈值  ，若存在指标 ia 满 足条件 i  ，则称该指标 ia 是 - 可删除的。 记 1 2=( , ,..., )m    是向量型数据集 x 中指标   1 m i i a  的权重，指标 ia 是 - 可删除的，以及 * =  * * * * *1 2 1 1, ,i i m     , . . . , , . . . , ，则根据定义 2 和定 义 4，下面的结论显然成立。 命题 1. 对于指标 ja ( i j )，成立 * j j  。 证明：根据公式 (2) 可知， * j1 ( ) 1 ( ) (1 ( )) (1 ( )) 0 j j j j l ll i l g a g a g a g a                 , (6) 即 *j j  。 □ 命题 2. 对于指标 1j a 及 2j a ( 1 2i j j  )， 若 1 2 ( ) ( )j jg a g a 成立，则 1 1 2 2 * * j j j j      。 证明：由公式 (2) 及命题 1 可直接得证。 □ 由上述命题可知：对于在阈值  下不可删除的 某个指标，删除权重小于阈值  的指标后，它的权 重不小于初始的权重值。还有， - 不可删除指标关 于知识认知能力的单调性是不变的。 例 1. 选取一些病人的看病记录，具体数据见下 表 1. 用定义 4 给出的粒度的方法确定相应指标的权 重。其中指标记为：头疼( 1a )、肌肉痛( 2a )、体温 ( 3a )、咳嗽( 4a )、睡眠质量( 5a )。 表 1. 一个关于某些病人的知识表达系统 病人 头疼 肌肉痛 体温 咳嗽 睡眠质量 1 否 是 正常 否 正常 2 是 是 正常 否 一般 3 是 否 偏低 否 正常 4 否 否 高 是 正常 5 是 是 很高 否 差 6 否 否 正常 是 正常 7 是 是 很高 是 差 8 否 是 较高 是 一般 9 否 是 高 否 一般 10 是 是 较高 否 差 我们约定：对于任意的指标 ia ，若 si tix x ，则 six 和 tix 应该划分为一类。经计算可知关于指标 1a 的 粒度划分为  1 = {1,4,6,8,9} {2,3,5,7,10}p ， 。根据定义 2 可知 1( ) 0.5g a  。类似地，通过计算可知： 2 3 4 5 ( ) 0.22 ( ) 0.22 ( ) 0.52 ( ) 0.34 g a g a g a g a       published by atlantis press copyright: the authors 105 fusheng yu, shihu liu 于是，由公式 (5) 可知，指标集 1 2 5{ , ,..., }a a a 的权重 向量为 1 2 3 4 5= ( , , , , ) = 0.1563, 0.2437, 0.2437, 0.1500, 0.2062)          . 3.2 基于聚合算子的权重确定方法 在该部分，针对决策信息是不确定的情况，给出一 种考虑决策目标偏好关系的指标权重确定方法。从 (1) 可以看出，决策信息 ijx 一般都是事先通过一定的 方法采集到的。实际问题的不确定性往往会导致数 据 ijx 表示的多样性与不确定性。在此，假设数据 ijx 的不确定性由两部分构成：对目标 ix 关于指标 ja 的 认可程度和否定程度。这样 ijx 就可以表述为一个直 觉模糊数 3  ,ij ij ijx    。在不改变向量型数据集 x 符号表述的基础上，下面我们给出确定指标权重的 相关定义与方法。 定义 6. 给定数据集 x ，则指标 ia 基于聚合算 子的权重定义为 1 ( ) ( ) i i m i j agg a agg a     . (7) 其中， ( )iagg a 是数据 1{ } n i ix  关于指标 ia 的聚合值。 根据定义 3 可知，如果决策者看重整体数据的 影响，则采用公式   1 1 1 (1 ) j j n n i ji ji j j agg a           . (8) 计算聚合值。若决策者想突出单个数据的作用，则 采用公式   1 1 1 (1 )j j n n i ji ji j j agg a          . (9) 计算聚合值，其中 1 2=( , ,..., )n    为决策者对 1{ } n i ix  的偏好，且满足归一化条件 1 1 n i i    . 在文章5中，我们就对基于直觉模糊集的聚合算 了做了一个简单的应用。当然，还存在诸多聚合算 子，具体的表述可以参考文献 6和7. 例 2. 表 2 是一个关于 4 个病人的医疗诊断数据 表。 下面我们计算关于 5 个体表特征的权重。在此取 =(0.25 0.25 0.25, 0.25)  ， ， ，即对 4 个病人同等看待。 经计算，基于公式 (8)的特征权重为 1 2 3 4 5 0.2121 0.2136 0.1872 = 0.2046 =0.1825            从上述两种确定权重的方法不难发现，基于粒 度的权重确定方法，侧重于从分类的角度，考察关 于某个指标对知识认知能力的强弱。而基于聚合算 子的权重确定方法，考虑了决策者对决策目标的偏 好程度。总之，两种方法各有优点。但是同时存在 一个缺点，就是如果决策问题的指标个数很多，则 指标之间的差异就有可能变得很小。对于此类问 题，我们在接下来的部分，将从数据变换的角度， 展开详细研究 4. 决策模型中有损数据的完备化 虽然发达的科技能够为数据的采集与存储提供便捷 的方式，但是对于实际问题，在数据采集、传输与 存储的过程中，出现部分信息损坏或丢失的情况是 不可避免的。接下来我们从一个全新的角度，对缺 失数据给出一种补偿的方法。该方法未必是最好 的，但从下面的实例可以看出，它是行之有效的。 问题描述：给定向量型数据 x ，存在部分信息 丢失的情况，假定所有指标的权重是已知的。 有损信息补偿方法： 1. 把存在有损信息的对象归为一类：  , {1, 2, }i ijms x x j m     . (10) 2. 计算对象 ix 和 jx 关于指标 pc 的局部相似性。数 据为实数时：(1) ipx 和 jpx 均存在，若 ip jpx x ， 则 ( , , ) 1ls i j p  ，否则 ( , , ) 0ls i j p  . (2) 对于其余 的任何情况，取 ( , , ) 0.5ls i j p  . 数据为集值时： (1) 若 ipx 和 jpx 均存在，则 | | ( , , ) | | ip jp ip jp x x ls i j p x x    ; (11) 否则 ( , , ) 1ls i j p  . (2)其余取 ( , , ) 0.5ls i j p  . 3. 计算任意两个对象之间的整体相似度 表 2. 医疗诊断问题 病人 temperature headache stomach pain cough chest pain al (0.8,0.1) (0.6, 0.1) (0.2,0.8) (0.6,0.1) (0.1,0.6) bob (0.0,0.8) (0.4, 0.4) (0.6,0.1) (0.1,0.7) (0.1,0.8) joe (0.8,0.1) (0.8, 0.1) (0.0,0.6) (0.2,0.7) (0.0,0.5) ted (0.6,0.1) (0.5, 0.4) (0.3,0.4) (0.7,0.2) (0.3,0.4) published by atlantis press copyright: the authors 106 decision-making model in the environment of complex structure data 1 ( , ) ( , , ) m p p gs i j ls i j p    . (12) 4. 对于存在信息丢失的任意对象 ix ，计算  ( , )i j i jc x x x t  , (13) 其中   , ( , )i jt x x gs i j   。 5. 对于 ix ms ，如果 ix 缺失的信息是关于 pc 的， 先计算  ( ) ,p i tp t i tpc x x x c x    , 然后(1)若 pc 是实数型的，则 ( ) [ ( ), ( )]p i p i p ic x c x c x   ；(2) 若 pc 是集值型的，则 ( )p i tp t c x x  。 例 3. 表 3 是一个包含 5 个对象和 6 个指标的一 个不完备信息系统。 显然，第 3 节介绍的权重的确定办法是无法确 定相应指标的权重的。在此不妨假定所有指标的权 重是相等的，即 1 / 6i  .通过一系列的计算，丢失 数据补偿后的信息系统表示为 4. 从中可以看出，该 方法是可行的。比如对指标 3c 而言，关于 1x 的补偿 完全不等同于 3x 的补偿值。这也说明，对于该方 法，不再是单一的取最大值，最小值或者平均值来 补偿丢失的数据。 5. 基于聚类的复杂结构环境下的决策模型 我们知道，风险分类8-10 是风险管理中的一项基本工 作。它是根据不同标准，对已知风险进行分类，旨 在提高风险管理效率的基础上降低风险管理成本。 聚类分析的基本思想，已经被广泛应用到风险 分析中，诸如震后灾情评估11，财务风险分析12，风 险投资13 等诸多领域。在此我们就实际问题的复杂 性，从聚类的角度作进一步分析研究。 对于一个实际问题，除了可以利用多个指标来 刻画一个对象的特性外，所要研究的对象之间往往 还存在千丝万缕的关系。这种关系，构成了所谓的 关系数据。于是，对于实际问题的建模，就是对于 一个具有复杂结构数据 ( , )g v e  的再分析过程。其 中， v 代表所要研究的对象集，是一向量型数据； e 代表该数据对象之间的关系集，是一关系型数 据。综上所述，对于一个实际问题的分析，就转变 成对多结构数据的分析。 通常情况下，所研究的问题都只是对 ( )v g 或者 ( )e g 的单独分析。分析过程简单明了，但是处理结 果有失偏颇。为了避免这一点，最直接的方法就是 对 ( )v g 和 ( )e g 中的数据实现对位加权处理。权重 往往是通过以往实验或者经验值估计出来的。 对于任意两个对象 , ( )i jx x v g  ，不妨记 ijd 为 关于向量型数据集 ( )v g 的不相似描述， ije 为关于关 系型数据集 ( )e g 的不相似描述，下面我们给出一种 不加权的信息融合方法。 定义 7. 给定向量型数据 ( )v g 和关系型数据 ( )e g ，则称   1 1 1ij ij ijt d e    (14) 为对象 ix 和 jx 无权的信息融合值，记为t . 表 3. 一个不完备的信息系统 编号 1c 2c 3c 4c 5c 6c 1 3 2 * * {1, 3} {3} 2 2 * [0.0, 1.5] [0.5, 1.0] {1, 2} * 3 1 2 * [1.5, 2.0] {1, 2, 3} * 4 * 0 [1.0, 2.0] * {1, 2} {2, 3} 5 2 1 [0.5, 1.5] [0.3, 1.8] {3} {1, 2} 表 4.完备化的不完备信息系统 编号 1c 2c 3c 4c 5c 6c 1 3 2 [0.5, 1.5] [0.3, 2.0] {1, 3} {3} 2 2 [0,2] [0.0, 1.5] [0.5, 1.0] {1, 2} {1, 2, 3} 3 1 2 [0.0, 2.0] [1.5, 2.0] {1, 2, 3} {2, 3} 4 [1,2] 0 [1.0, 2.0] [0.5, 2.0] {1, 2} {2, 3} 5 2 1 [0.5, 1.5] [0.3, 1.8] {3} {1, 2} published by atlantis press copyright: the authors 107 fusheng yu, shihu liu 显然，上述信息融合矩阵t 满足以下性质： 命题 3. 对于任意 , ( )i jx x v g  ，成立 ij jit t 。 命题 4. 对于任意 ( )ix v g ，成立 0iit  。 显然，融合后的数据 t 是一个关系型数据，描 述的是任意两个对象之间的不相似程度。从本质上 而言，和最初的关系型数据 ( )e g 在数据表示上没有 本质的区别。可是在数据信息的蕴含方面， t 却包 含了向量型数据 ( )v g 所描述的信息，要比 ( )e g 蕴 含的信息丰富。 对于一个具体的风险分析问题，一旦 ( )v g 和 ( )e g 给定，就可以由公式 (14)，计算出融合后的关 系型数据 t 。接下来，我们就需要对关系型数据 t ，挖掘其潜在的社团结构，对数据分析人员对问 题的进一步研究提供有用的信息。 问题描述：对于某个风险分析问题，其模型化 的数据表示为 ( , )g v e  。希望把 | ( ) |v g 个对象划 分成 k 个团体，使得团体之间的差异尽可能的大， 但是团体内部差异尽可能的小。 模型建立： 1. 向量型数据集 ( )v g 的处理：如果 ( )v g 存在部分 信息丢失的情况，则利用第 4 节给出的有损信息 补偿方法，补全丢失的信息。如果 ( )v g 不存在 信息丢失的情况，则利用第 3 节介绍的方法，确 定向量型数据 ( )v g 中相应指标的权重。 2. 利用定义 7 介绍的方法，把 ( )v g 和 ( )e g 融合成 新的关系型数据 t 。其中上一步计算出来的指标 的权重，作用于关于向量型数据集 ( )v g 中任意 对象的不相似性度量方面。 3. 根据问题的需要，利用谱聚类的方法，将关系型 数据 t 划分成需要的 k 类。 由上可知，对于具体的风险决策问题，由采集 到的数据，上述模型可以实现进一步挖掘该问题中 潜在“社团”结构的目的。相对于利用单一结构数 据寻找“社团”而言，复杂数据结构环境下的决策 模型，提供的结果更可靠，更具有说服力。这不仅 能为对该数据的进一步分析提供良好的指导，同时 也能降低接下来处理数据时人为带入的不确定性。 例 4. 表 5 是一个复杂结构数据 ( , )g u v 的向 量型数据集 ( )v g 的表示，表 6 是对应的关系型数据 ( )e g 的表示。 表 5：向量型数据 ( )v g 编号 1c 2c 3c 4c 5c 6c 1 14.2 1.7 2.4 11.4 127 2.8 2 13.2 1.7 2.1 14.0 100 2.6 3 13.1 2.3 2.6 18.6 101 2.8 4 14.3 1.9 2.5 16.8 113 3.8 5 13.2 2.5 2.8 21.0 118 2.8 6 14.2 1.7 2.4 15.2 112 3.2 7 14.2 1.8 2.4 14.6 96 2.5 8 14.0 2.1 2.6 17.6 127 2.6 9 14.0 1.6 2.1 14.0 96 2.8 10 13.8 1.3 2.2 16.0 98 2.9 表 6： 关系型数据 ( )e g 编号 1 2 3 4 5 6 7 8 9 10 1 0 1 0 0 0 0 1 1 0 0 2 1 0 1 1 0 1 0 0 0 0 3 0 1 0 0 1 0 1 1 0 0 4 0 1 0 0 0 0 1 1 0 0 5 0 0 1 0 0 0 0 0 0 0 6 0 1 0 0 0 0 1 1 0 0 7 1 0 1 1 0 1 0 0 0 0 8 1 0 1 1 0 1 0 0 0 1 9 0 0 0 0 0 0 0 0 0 0 10 0 0 0 0 0 0 0 1 0 0 在表 6 中， 数值 1 表示对象 ix 和对象 jx 恒不相 似，即 1ije  ,反之，0 表示对象 ix 和 jx 绝对相似， 即 0ije  .接下来我们从三个方面对模型加以说明： （i）只考虑数据 ( )v g 利用经典的 k-means 聚类算法把向量型数据集 ( )v g 划分成 3k  类： 1 2 3 7 9 10 2 4 5 6 3 1 8 { , , , , } { , , } { , } c x x x x x c x x x c x x     . (15) （ii）只考虑数据 ( )e g 利用谱聚类算法(spectral clustering)把关系型数 据集 ( )e g 仍然划分成 3k  类：       1 2 7 2 1 3 4 5 6 9 10 3 8 c = , c = , , , , , , c = x x x x x x x x x x     . (16) （iii）同时考虑数据 ( )v g 和 ( )e g 首先计算各指标的权重，在这里我们采用基于 粒度的权重确定方法。经计算可知  = 0.1619, 0.17, 0.1619, 0.1781, 0.1741, 0.1538   . 然后在计算关于 ( )v g 的加权不相似性度量。在此基 础上，根据定义 7，计算数据集 ( )v g 和 ( )e g 无权 融合后的信息 t 。由于 t 在结构表示上仍然是一个 关 系 型 数 据 ， 故 采 用 谱 聚 类 算 法 (spectral clustering)，可知当 3k  时，得到的划分为 published by atlantis press copyright: the authors 108 decision-making model in the environment of complex structure data 1 2 7 9 10 2 1 3 4 5 6 3 8 { , , , }, { , , , , } { } c x x x x c x x x x x c x       (17) 从式子 (15)-(17) 不难发现，对象 4 5,x x 和 6x 一致划 分在同一类，对象 2x 和 7x 一致划分在同一类。相对 于综合考虑两种结构的数据而言，在只考虑向量型 数据集 ( )v g 时，原本划分在同一类的对象 1x 和 3x ，将被划分到其余不同的类中。而当只考虑关系 型数据 ( )e g 时，对象 9x 和 10x 将被划分到其余的同 一类中。 6. 结论 风险无处不在，无时不有。在面对具有复杂结构数 据的决策问题时，进一步挖掘其潜在的数据结构， 能够为降低风险带来巨大的帮助。本文正是基于这 一点，首先探讨了相应指标的权重确定方法。其 次，对于数据丢失的情况，建立了一种数据补偿的 方法，并用相应的实例验证了该方法的可行性。对 于问题的数学模型所面对的是一个具有复杂结构的 数据时，我们利用聚类的思想挖掘其潜在的社团结 构，期望达到进一步降低问题处理中带来的不确定 性。 参考文献 1. c. f. huang, natural disaster risk analysis and management. (science press, beijing, 2012). 黄崇福.自然灾害风险分析与管理[m].北京:科学出版 社，2012. 2. l. b. tao, y. s. li, z. l. feng, et al., project risk analysis theory and practice.(tongji university press, shanghai, 2006). 陶履彬，李永盛，冯紫良等. 工程风险分析理论与实 践[m]. 上海:同济大学出版社，2006. 3. c. h. le, h. y. ding, g. h. dong, et al., risk analysis of failure damage to marine riser based on fuzzy fault tree, journal of natural disasters,21(2)(2012) 173-179. 乐丛欢，丁红岩，董国海等.基于模糊故障树的海洋 立管破坏失效风险分析[j]. 自然灾害学报，2012,21(2): 173-179. 4. w. pedrycz, knowledge-based clustering: from data to information granules. (john wiley & sons, new jersey, 2005). 5. s. h. liu, f. s. yu, aggregation operators based mcdm with intuitionistic fuzzy information, in proceedings of the fifth annual meeting of risk analysis council of china association for disaster prevention, eds. c. f. huang and g. f. zhai (atlantis press, paris, 2012), pp. 411-416. 6. z. s. xu, intuitionistic fuzzy information aggregation theory and application. (science press, beijing, 2008). 徐泽水.直觉模糊信息集成理论及其应用[m].北京:科 学出版社，2008. 7. z. s. xu, intuitionistic fuzzy aggregation operators, ieee transactions on fuzzy systems, 15(6) (2007) 1179-1187. 8. p. j. shi, c. f. huang, t. ye, et al., constructing china’s comprehensive risk management system, disaster reduction in china, 1(2)(2005), 164-167. 史培军，黄崇福，叶涛等.建立中国综合风险管理体 系[j].中国减灾, 2005,1(2): 35-37. 9. c. f. huang, a trapezoid framework for integrated risk management, journal of natural disasters, 14(6)(2005) 9-14 黄崇福.综合风险管理的梯形架构[j].自然灾害学报， 2005,14(6): 9-14. 10. p. j. shi, t. ye, j. a. wang, et al., integrated governance of natural disaster risk, journal of beijing normal university,5(2006), 130-136. 史培军，叶涛，王静爱等.论自然灾害风险的综合行 政管理[j].北京师范大学学报，2006, 5: 130-136. 11. l. g. tian, y. li, fuzzy cluster analysis in the application of reservoir evaluation after earthquake, yellow river, 32(1)(2010),130-131. 田林刚，李洋.模糊聚类分析在震后水库风险评价中 的应用[j].人民黄河，2010, 32(1): 130-131. 12. x. g. zhou, r. zhu, analysis of corporate financial risk based on fuzzy clustering and pattern recognition, science and technology management research, 8(2012),115-123. 周晓光，朱荣.基于模糊聚类和模式识别的企业财务 风险分析[j].科技管理研究，2012, 8: 115-123. 13. y. p, yang, j. wang, research on industry clustering of venture capital in china, science and technology management research, 12(2012), 164-167. 杨艳萍，王静. 我国风险投资的行业聚类研究[j]，科 技管理研究，2012, 12: 164-167. published by atlantis press copyright: the authors 109 microsoft word the research on applicability of amended kmv model with different indus-tries the research on applicability of amended kmv model with different industries yang yang school of management and economics, university of electronic science and technology of china, chengdu 610054, china li li1, 2, zongfang zhou1, wenying fei3 1school of management and economics, university of electronic science and technology of china, chengdu 610054, china 2school of mathematics, yunnan normal university, kunming 650092, china 3research center of corporate governance, naikai university, tianjin 300071, china abstract the determination of the default point is the key to applying the kmv model. in the traditional research, the default point is equal to the sum of short term debt and half of the long term debt empirically based on the credit data of american companies. but whether it is fit for the chinese companies and whether it is sensitive to industries depend on further research. this paper amends the parameters of default point. further the paper also analyzes the applicability of amended kmv model in different industries based on the data of chinese listed companies. keywords: kmv model; default point; different industries; credit data 1. introduction the kmv model is a credit risk measurement established by kmv company in 1997. there are four main advantages which led this model widely used in the areas of credit risk assessment and the forecasting of financial distress. first, the kmv model is superior to the timeliness of the assessment models. the kmv model uses the real-time data. it can update the probability of default in real time based on the data of securities market; second, the assumption of kmv model is week. the efficient market assumption is not required. this is very applicable for the week effective securities market of china; third, the kmv model is a forward-looking method. the data used in the model reflects the expected value of the company and the judgment of the company’s future development trends of the investors; fourth, the kmv model is a base method which is different from the ordinal method. it can not only reflect the credit risk level of the order but also reflect the credit risk level of the degree of difference. a large number of empirical researches have shown that the kmv model integrates the advantage of both financial reports and market price. it reflects the company’s history situation on one hand, and reflects the trends of the company’s future development on the other hand.1 jeff bohn (2003) had compared the credit risk evaluation results of kmv model and the standard & poor’s. the research concluded that the kmv is suitable for assessing the credit risk with low quality of information.2 kealhofer and kurbat (2001) had also compared the credit risk evaluation results of kmv model and the roa method of moody’s corporation. their results illustrated that the forecast accuracy of kmv model is higher than the forecast accuracy of the roa method.3 besides, matthew kurbat and irina korablev (2002)4 , peter crodbie and jeff bohn (2003)5 , stefan m. denzler (2006)6 all had researched the kmv model from different perspectives. and conclusions of their research consistently indicated that the kmv model is effective and widely applicable. at present, the applicability of the kmv model based on the data of chinese companies remains controversial. many domestic scholars have researched the applicability of the kmv model in chinese context. however their conclusions are not consistent. most chinese scholars believe that the parameters of kmv model needs to be amended in order to suit for assessing credit risk of chinese companies. therefore, many journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 27-33 published by atlantis press copyright: the authors 27 yang yang et al. researches about the amended kmv model emerging. li (2010)7 had introduced the growth rate of total assets into the kmv model in order to describe the characteristics of business group; peng (2011)8 had amended the kmv model under the market segmentation environment based on assumption of assets correlations. more amendments for kmv model focus on the amendments for the default point. that’s because the determination of the default point has a certain degree of arbitrariness and subjectivity. a lot of experience has shown that the default point is equal to the sum of short term debt and half of the long term debt with the credit data of american companies. but many chinese scholars doubt that whether the parameter of default point is suit for chinese companies. for this reason, a number of literatures discussed the parameter of default point. zhang (2010)9 etc. exogenously given parameters based on qualitative analysis. all these method to determine parameter of default point have a certain degree of arbitrariness. only pan (2009)10 gave a quantitative methods to determine the parameter based on multiple regression. however both qualitative and quantitative researches are concluded inconsistent results. the research of ma (2008)11 maybe can give some revelation. different companies have different prediction accuracy with kmv model. the different parameter may adapt to companies in different industries. follow this idea, this paper first amends kmv model through a appropriate clustering method to determine the parameter of default point. then the paper discusses the applicability of amended kmv model with different industries. there are two main innovations in this paper. first, this paper endogenously determines the parameter of default point. it improve the prediction accuracy of kmv model; second, this paper, to some extent, answers the reasons for the inconsistent results of the existing researches with the perspective of different industries. 2. kmv model the essence of kmv model is a kind of option pricing model which is rooted in the option pricing theory by black-scholes12 .the kmv model describes the bankruptcy as an endogenous event for a company. the affair of bankruptcy is determined by both the nominal value of liability of the company and the market value of the assets of the company at maturity date. the kmv assumes that the value of the company follows a geometric brownian motion which is the same as the assumption of black-scholes option pricing formula. suppose the nominal value of liability of the company is f and the market value of the assets of the company at maturity date isv . the company can’t be in full settlement of its liability. the default occurs while the market value of the company’s assets is less than the nominal value of the liability at maturity date, i.e. fv < . if the market value of the company’s assets is higher than the nominal value of the liability at maturity date, i.e. fv > . the default of the company will not occur. the debtor’s debt will be fully repaid. so the commercial banks loans to the company is equivalent to the company opened a put option, i.e. there is a put option that is held by the company. the exercise price of the option is the amount of loan debt. the underlying asset of the option is the market value of the company’s assets. when the market value of the company’s assets dropped to a certain level (called default threshold), the company will choose to implement the option. that is chose to default. similarly, the loans can be also interpreted as a call option. the measure of credit risk with kmv is often divided into three basic steps. the first step is to estimate the market value of the company’s assets v and its fluctuation vσ ; the second step is to calculate the distance of default; and the last step is to convert the distance of default into probability of default. according to the theory of option pricing by black and scholes, the analytical expression for the equity value is )()( 2 )( 1 dnfednve ttr tt −−−= （1） in which tt ttrfv d v v t − −++ = σ σ ))( 2 ()/ln( 2 1 ttdd v −−= σ12 published by atlantis press copyright: the authors 28 the research on applicability of … r is the risk-free rate, te and tv , respectively, the equity value and the value of total assets at the time t . )(⋅n is the distribution function of the standard normal distribution. further, according to the ito lemma, it can get the relationship between the fluctuation of equity value and the total assets, i.e. the relationship between eσ and vσ . ve e dvn σσ )( 1= （2） simultaneous the formula (1) and the formula (2), it can have the key variable of the kmv model, i.e. v and vσ . then the distance of default can be represented as follow. vv dpv dd σ − = （3） if the probability distribution of assets is known, the expected default frequency (edf) can be calculated. the kmv corporation gets the probability distribution of assets based on lot of historical credit data. in scientific research, it often uses standard normal distribution or lognormal distribution to describe the distribution of assets approximately. assume the assets follow the standard normal distribution; the edf can be interpreted as follow. )( ) )( )( ( ))(pr( ddn ve vedp n dpveedf v −= − = <= σ （4） 3. the amendment of kmv model the parameter of the default point in the kmv model is widely controversial because of its empirical and arbitrary. most research defined the default point as the sum of short term debt (std) and half of the long term debt (ltd). however, the parameter 0.5 is based on lots of american credit data. in china, many scholars doubt its adaptability of the parameter for chinese company data of credit. but a great amount of research conclusions are not consistent. there may have two main reasons for this. first reason is that different companies may suit for different parameter of default point; second reason is that many researches amended the parameter of default point exogenously based on qualitative analysis. they also have a certain degree of subjectivity and arbitrariness themselves. because of this, this paper first considered the differentiation of companies, dividing the companies into three basic industries, the traditional manufacturing, the commerce and trade industry, high-tech industry. then this paper divided the samples into two parts. one part is the training samples and the other part is the testing samples. in this way, the parameter of default point is determined endogenously. specifically, the paper applied the thought of clustering; assuming the parameter of default point is a undetermined parameter β . for a given industry and selected training samples, there are n companies including 1n st companies and 2n non-st companies. 21 nnn += （5） denote the default point of the 1n st companies as 1idp , 2idp …… 1indp . then the distance of default and the edf of the 1n st companies are 1idd , 2idd … … 1in dd and 1iedf , 2iedf … … 1inedf . similarly it can denote the distance of default and the edf of 2n non-st companies as 1sdd , 2sdd …… 2sn dd and 1sedf , 2sedf …… 2snedf . for every company, the bivariate vector ),( lplp edfdd , },{ sil ∈ describes its credit situation. as the distance of default and the edf both depend on the parameter of default point β , the distance of default and the edf can be also written as )(βlpdd and )(βlpedf . for 1n st companies, the average distance of default and the average expected default frequency denote as 1 1 1 )( n dd dd n k ik i ∑ ==β （6） and 1 1 1 )( n edf edf n k ik i ∑ ==β （7） similarly it can also define the average distance of default and the average expected default frequency of 2n non-st companies. published by atlantis press copyright: the authors 29 yang yang et al. 2 1 2 )( n dd dd n k sk s ∑ ==β （8） and 2 1 2 )( n edf edf n k sk s ∑ ==β （9） further, denote the ))(),(( ββ iii edfdddf = （10） ))(),(( ββ sss edfdddf = （11） learn from the thought of clustering; define the distance of st companies and non-st companies by euclidean distance 22 edfdd si dfdf θθ += −=π （12） in which ))()(( ββθ sidd dddd −= ))()(( ββθ siedf edfedf −= then determine the parameter of default point through the follow optimal problem ]10,0[.. maxarg ∈ π= β β ts （13） finally, the parameter β is formulated endogenously by the optimal problem (13). this method to determine the parameter has two advantages. first, it increases the degree of freedom of the optimal problem. so that it can improve the accuracy and suitability of the kmv model; second, it correct the deviation of the assumption of standard normal distribution of assets based on sample data. 4. the empirical test as the chinese commercial banks are not open the list of default companies to the public, it’s unable to obtain the actual data of default companies. this paper uses the approach that the most researches used, make the samples of st companies in the shanghai and shenzhen a-share market as the samples of default companies. this paper selects 180 companies as the samples from shanghai and shenzhen a-share market during 2006 to 2010, including 60 st companies and 120 nonst companies. for each industry, it mentioned above, there are 20 st company samples and 40 non-st company samples. in each industry, it selects training samples including 14 st company samples and 30 non-st company samples randomly. then it uses the method which is introduced in the third part of this paper to formulate the parameter of default point. from this it can get parameters of default point in traditional manufacturing, commerce and trade industry and hightech industry respectively. these parameters are noted as 1β , 2β and 3β . from optimal problem (13), it can formulate the value of the parameters of default point which are listed in table 1below. table 1. parameter of default point in different industries traditional manufacturing commerce & trade high-tech industry 89.01 =β 68.02 =β 12.11 =β the parameter concluded in different industries shows that the applicability of kmv model is quite fig.1. roc in traditional manufacturing published by atlantis press copyright: the authors 30 the research on applicability of … different for different industries. it illustrates the discussion of parameter of default point by different industries is effective and may explain the inconsistent of existing researches with different samples from a perspective. figure2 roc in commerce & trade fig.3. roc in high-tech industry it can easily know from table1 that the parameter of default point in every industry of china is higher than the parameter of default point in traditional kmv model (the parameter of default point is equal to 0.5). this illustrates that the long term debt (ltd) is more important to the credit risk assessment for chinese companies than american companies. further, it draws the roc curve of full samples for each industry and the figure1, figure2 and figure3 express the roc curve of full samples for traditional manufacturing, commerce & trade industry and the high-tech industry respectively. from the figure 1 to figure 3, it can easily know that for each industry, the amended kmv models are more effective than the traditional kmv model. these figures also illustrate that the prediction accuracy of amended kmv model is higher than the traditional kmv model under the full samples’ testing. it has already indicated that the prediction accuracy of amended kmv model under the full samples. then this paper will test the prediction of amended kmv model out of the training samples. as it said in above section, for each industry, this paper divided the samples into two parts, one part is the training samples including 14 st company samples and 30 non-st company samples; and the other part is the testing samples including 6 st company samples and 10 nonst company samples. first it uses the training samples to formulate the parameter of default point. then it tests the prediction accuracy of the amended kmv model with the testing samples. in order to be consistent with the previous discussion, it select the same training samples as the paper did above, so the parameter of default point is concluded as table 1shows. compared with the traditional kmv model, the prediction accuracy in different industry is shown as follow, i.e. table 2 to table 4. table 2. prediction accuracy of traditional manufacturing model accuracy traditional kmv model amended kmv model type 1 error 16.7% 0 type 2 error 20% 10% total accuracy 81.25% 93.75% table 2 shows the prediction accuracy of the industry of traditional manufacturing. in this industry, both type 1 error and type 2 error of amended kmv model is less than traditional kmv model. the total accuracy of amended kmv model is higher than traditional kmv model significantly. table 3. prediction accuracy of commerce & trade industry model accuracy traditional kmv model amended kmv model type 1 error 33.3% 16.7% type 2 error 30% 10% total accuracy 69.75% 87.5% table 3 shows the prediction accuracy of the commerce and trade industry. in this industry, both type published by atlantis press copyright: the authors 31 yang yang et al. 1 error and type 2 errors of amended kmv model is less than traditional kmv model. the total accuracy of amended kmv model is higher than traditional kmv model significantly. table 4. prediction accuracy of high-tech industry model accuracy traditional kmv model amended kmv model type 1 error 50% 16.7% type 2 error 30% 0 total accuracy 62.5% 93.75% table 4 shows the prediction accuracy of the hightech industry. in this industry, both type 1 error and type 2 errors of amended kmv model is less than traditional kmv model. the total accuracy of amended kmv model is higher than traditional kmv model significantly. totally, from table 2 to table 4 it can easily conclude that the amended kmv model is better than the traditional kmv model consistently. the table 5 below shows the total accuracy of traditional kmv model and amended kmv model based on the data out of training samples. table 5. prediction accuracy of both kmv models model traditional model amended model accuracy 70.83% 91.67% from the discussion above, it can get two conclusions. first is the parameter of default point has significant difference in different industries. second is prediction accuracy of amended kmv model is higher than traditional kmv model consistently in all industries. 5. conclusions and further discussion the parameter of default point in traditional kmv model is experience concluded by the credit risk data of american companies. because of this, the applicability of the traditional kmv model with chinese background is widely questioned by scholars. there are many researches that had amended the traditional kmv model in chinese background. these researches have two defects. first, most of these researches amended the parameter of default point exogenously which leads to an inevitable problem of subjectivity and arbitrary. second, these researches are concluded inconsistent with different samples. for these reasons, this paper first learns from the thought of clustering to determine the parameter of default point endogenously in different industries respectively. then it compares the amended kmv model with the traditional kmv model with roc curve in different industries based on full samples data. at last, this paper also tests the prediction accuracy out of the training samples. the research concluded two main results. one is the applicability of the kmv model may very difference in different industries; the other is the amended kmv model in this paper is better than the traditional kmv model consistently. these results of this paper may provide a think of application of kmv model in chinese background on one hand and explain the inconsistent of existing researches in a certain extent. limited by the samples, the classification of industries in this paper is very rough. the reasonable of such classification is still need to be considered. further discussion of classification criteria in a broader sense and deeper reasons of applicability of kmv model will help to understand kmv model deeper and applies it in chinese background. references 1. j. a. mcquown, a comment on market vs accountingbased measures of default risk, white paper, moody’s kmv, (1993). 2. j. r. bohn, modeling, methodology, white paper, moody’s kmv, (2003). 3. stephen kealhofer, brian dvorak, modeling the default risk of unlisted firms, research reports of kmv, (2001). 4. matthew kurbat, irina korablev, methodology for testing the level of the edftm credit measure, white paper, moody’s kmv, (2002). 5. peter crodbie, jeff bohn modeling default risk, working paper, kmv corporation, (2003). 6. m. stefan denzler from default probabilities to credit spreads: credit risk models do explain market prices, published by atlantis press copyright: the authors 32 the research on applicability of … finance research, 122(3): 439-462. 7. li li, zongfang zhou, credit risk measure for business group based on modified kmv model, in 4th annual meeting of risk analysis council of china association for disaster prevention(changchun china, 2010): 837-842. 8. daheng peng, congyu zhang, improvement of kmv model under market segmentation and its empirical analysis, statistic & information forum, 26, 5(2011):3338. 9. nengfu zhang, jia zhang, the application of amended kmv model in measuring credit risk of china listed companies, forecasting, 29, 5(2010):48-52. 10. jie pan, zongfang zhou, empirical study on revising default point in kmv model of public companies based on full circulation of stock, in 4th annual meeting of enterprise operation of china (huhehaote china, 2009):197-203. 11. ruowei ma. testing kmv on the financial distress of listed companies in china, application of statics and management, 5(2006):593-601. 12. f. black and m. scholes. the pricing of options and corporate liabilities, journal of political economy, 5(1973):637-659. published by atlantis press copyright: the authors 33 microsoft word volume 12, issue 1-3 journal of risk analysis and crisis response, 2022, 12(1), 36-44 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i1.320 36 article coupling coordinated development of ecological environment and digital inclusive finance in ecologically fragile areas of western china xudong cai 1,2 and hongmei zhang 1,2,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhm1035@qq.com; tel.: +86-0851-88510575 received: march 24, 2022; accepted: april 3, 2022; published: april 15, 2022 abstract: this paper takes 12 ecologically fragile provinces in western china as the research object, calculates the comprehensive evaluation value of ecological environment and the development level of digital inclusive finance of 12 provinces in western china from 2011 to 2020 by constructing the evaluation index system, and analyzes the changes of ecological environment and digital inclusive finance coordination scheduling by using the coupling coordination degree model. the results show that: the overall ecological environment assessment of 12 provinces in the west is on the rise; the level of digital inclusive finance has also increased year by year. the coordination degree is on the verge of maladjustment to primary coordination, and the coordination index is on the rise, which indicates that the relationship between the western ecological environment and digital inclusive finance is gradually easing and tends to coordinated development. keywords: ecological environment; digital inclusive finance; coupling coordination degree model; open grade method 1. introduction ecological environment, the basis of our survival, with the rapid development of the economy, natural resources are subject to excessive exploitation and utilization by human beings, and the problem of ecological deterioration is increasingly prominent, especially in the western ecologically fragile areas ecological environment has become a key issue that restricts regional economic development. with the concept of inclusive finance, some western provinces have gradually encountered financial services to achieve local economic development, so how to coordinate the development of ecological environment and inclusive finance is a problem we urgently need to solve. the inherent mechanism of the coordinated development of ecological environment and digital inclusive finance is to achieve common sustainable development, but coordinated sustainable development does not mean "equal development", but the development of mutual promotion and coupling between the two systems. xiaonan huang and mu zhang (2020) [1] used hesitant fuzzy language, gray correlation projection method measured the level of the two systems of ecological environment and financial agglomeration in chinese provinces, then they used the coupling coordination model to empirically analyzed. the results showed that the coupling coordination xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 37 between ecological environment and financial agglomeration in chinese provinces was not high, but their spatial patterns remained generally stable. shengliang su (2020) [2] made a comprehensive analysis of the development status of ecological environment and socio-economic development, the level of coupling and coordination, and the evolution trend of coupling degree in ningxia hui autonomous region as an example, which better reflects the coupling development process of interactive coercion between ecological environment and socio-economy in ningxia. peiyu jia and jianing chen (2020) [3] used coupled coordination model and coordinated development theory to study the changing pattern of coordinated development level in ecologically fragile areas of shanxi province. the study shows that the overall level of coordinated ecological, resource and economic development in ecologically fragile areas of shanxi province has improved, with basic coordinated areas dominating. peishan tong and shengxu shi (2018) [4] used the psr-gcq model to conduct a comprehensive evaluation of the coupled and coordinated development of ecological environment and economic development in an urban agglomeration, and the study showed that the ecological and economic development of this urban agglomeration is well coordinated in time and space. guofeng gu and xuehui wang (2018) [5] analyzed the coupling relationship between economic development and ecological environment in northeast china and its spatial and temporal evolution characteristics by using the gray correlation model, which showed that the coupling relationship between ecological environment and economic development in northeast china is complex, among which excessive emission of pollutants and lack of water resources become the main factors limiting economic development. to sum up, this paper selects the western ecologically fragile region as the research object, firstly, constructs the relevant index system of ecological environment and digital inclusive finance development level, selects the original data of 12 western provinces, cities and autonomous regions in the last ten years (2011-2020), and uses the pull-out grade method and comprehensive index method to measure the level; secondly, uses the coupled coordination degree model to empirically analyze the coordination of ecological environment and digital inclusive finance development, in order to provide scientific reference for the coordinated development of ecological environment and digital inclusive finance in the western ecologically fragile region. 2. ecologically fragile area of western china in this paper, 12 provinces and autonomous regions (hereinafter referred to as: 12 provinces) in the western ecologically fragile region of china, including guizhou, chongqing and sichuan, are selected and roughly divided into 3 regions according to their different ecological vulnerability characteristics. (1) northwest desert oasis intersection ecologically fragile area, the administrative area involving xinjiang, gansu, qinghai, tibet and inner mongolia 5 provinces and autonomous regions. (2) northern arid and semi-arid grassland region, the administrative area involves shaanxi and ningxia 2 provinces and regions. (3) the southwest karst mountain stone desertification ecological fragile area administrative area involving sichuan, guizhou, yunnan, chongqing, guizhou and other five provinces and cities. the western region is located in an exceptionally fragile ecological environment, and the realization of coupled and coordinated development between ecological environment and socio-economic development in the western region is a key issue that needs to be addressed urgently to achieve regional ecological protection and high-quality development. xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 38 3. measurement and analysis of ecosystem and digital financial inclusion level 3.1. comprehensive evaluation of ecological environment the construction of an index system for western ecologically fragile areas is the basis for a comprehensive evaluation of their ecological environment, jianning yang (2011) [6] established natural environment indicators, economic environment indicators and social environment indicators based on the actual ecological environment in the west. its sub-indicators include forest coverage rate, the total source of investment in forestry system fixed assets, the amount of water resources per capita, and a total of 18 indicators. hongmei zhang et al. (2019) [7] constructed an evaluation system including 15 secondary indicators in four subsystems: resources, environment, economy and society, and comprehensively evaluated the ecological and environmental conditions of 12 provinces, cities and autonomous regions in the western ecologically fragile region. with reference to the above literature, this paper constructs a comprehensive evaluation index system of ecological environment in western ecologically fragile areas from three perspectives: environmental resources, economic development level, and social development level, as shown in table 1. in this paper, 12 provinces from 2011 to 2020 were selected as the research sample, and the data were obtained from the china environmental statistical yearbook, china statistical yearbook, and local statistical yearbooks, water resources bulletins, environmental bulletins, national economic and social development bulletins, resset financial research database, and peking university digital inclusive finance index research reports in previous years. table 1. comprehensive evaluation index system of ecological environment in ecologically fragile areas of western china. indicator name (unit) environment and resources forest coverage (%) green space per capita (m2) comprehensive soil erosion control area (1000 hm2) per capita water holdings (m3) sulfur dioxide emissions (10,000 tons) wastewater discharge (billion tons) total investment in environmental pollution control (billion yuan) area share of nature reserves (%) economic development level per capita disposable income of urban residents (yuan) gross domestic product per capita (yuan) per capita disposable income of rural residents (yuan) total output value of agriculture, forestry, animal husbandry and fishery (billion yuan) level of social development total population (million people) natural population growth rate (%) engel coefficient (%) number of students enrolled in general higher education schools (10,000) xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 39 3.2. ecological environment level measurement and result analysis minghua wei et al. (2010) [8] adopted the pull-out grade method for the comprehensive evaluation of water environment, and the results show that the pull-out grade method can consider the influence of time on the evaluation results and improve the accuracy of the evaluation results, and this type of method is good for the comprehensive evaluation of ecological environment. in this paper, after fully examining the environmental characteristics of the western region, it is proposed to measure and analyze the ecological environment level of 12 western provinces, cities and autonomous regions using the pull-out grade method, and the research process is as follows: suppose there are m evaluation objects 1 2, , , ms s s , n evaluation indicators 1 2, , , nq q q , and arranged 1 2, , , rt t t chronologically from the original data   ij ks t ,   ij ks t composed a timing stereo data table. a comprehensive evaluation problem supported by a time-series stereo data table, called a dynamic comprehensive evaluation problem can be generally expressed as:               1 2 1 2, ,..., ; , ,..., , 1, 2,..., ; 1, 2,...,i k k k m k k k n ky t f w t w t w t q t q t q t i n k r   among them  i ky t indicates the comprehensive evaluation value of the evaluation object is at the moment rt , f denotes the analytic equation of the function,  i kw t indicates weighting factor of the evaluation object iq at the moment rt . the pull-down grade method is to take the composite evaluation function: 1 1 2 2 ... , 1, 2,...,i i i n iny w q w q w q i n     to determine the value of iw is to make the difference between evaluation objects s as large as possible, differences between evaluation subjects can be indicated by 2 2 1 ( ) m i i y y    after standardizing raw data: 22 ( ) ( )t t tiy y y wx wx w hw     we will get, when restrict 1w  , get the eigenvector corresponding to the largest eigenvalue of the symmetric matrix h of w, 2 takes the maximum value the feature vector at this point is the corresponding weight of each index, and then brought into the comprehensive evaluation function to obtain the comprehensive evaluation value of the system. the change trend of the comprehensive evaluation of ecological environment by regions is shown in figure 1. as can be seen from figure 1: (1) the overall eco-environmental assessment values of the 12 provinces increased and decreased during the period 2011-2020 but showed an overall increasing trend with a moderate magnitude, indicating that the ecological environment in the western region tended to improve, the ecological vulnerability decreased, and the structure and function of the ecosystem improved and enhanced during the study period. (2) except for some individual provinces which showed a decline in a certain year, it basically developed in an upward trend. among them, sichuan, qinghai, tibet, inner mongolia's ecological environment is significantly better than the other xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 40 eight provinces, and in recent years, with the country's continued attention to the ecological development of the west, inner mongolia's ecological environment has achieved a major reversal. green lock sand throat, yellow sand into oasis, this is just a microcosm of the ecological realization of a major reversal in inner mongolia. (3) due to the continuous economic development and urbanization construction in recent years, guizhou, ningxia and gansu have paid too much attention to economic development instead of pollution control, resulting in the deterioration of the ecological environment, making their ecological environment comprehensive evaluation value in the 12 provinces fell in the back of the comprehensive ranking, this overall comprehensive ranking is more consistent with the actual ranking and has some reference value. figure 1. comprehensive evaluation value of ecological environment in 12 provinces in western china. 3.3. measurement and analysis of the development level of digital financial inclusion the key to this paper is how to build a digital financial inclusion measurement system that meets the realities of the western region. feng guo et al. (2020) [9] enriched the traditional financial inclusion evaluation system by considering the breadth of digital financial services and the multi-level and diversified nature of digital financial services and compiled the "peking university digital financial inclusion index" covering 31 provinces, 337 cities above prefecture level and about 2,800 counties in mainland china. this paper refers to and cites the peking university digital inclusive finance index research report, and takes logarithms of digital inclusive finance in 12 provinces to obtain the trend of its development level as shown in figure 2. from figure 2, it can be seen that: (1) on the whole, the evaluation value of digital inclusive finance development in western ecologically fragile regions has increased and stabilized year by year, indicating that the development level of digital inclusive finance in western ecologically fragile regions is improving year by year, which is related to the national vigorous development of digital economy in recent years. (2) 2011-2012 to enhance the speed very quickly after 2013 to enhance the rate tends to be stable, the development space is larger, to achieve leapfrog development. by region, the level of digital financial inclusion development in xinjiang, gansu, qinghai, inner mongolia and tibet, which are in the ecologically fragile zone of the northwest desert oasis junction, has been increasing year by year but is lower compared with other ecologically fragile zones, and the relatively poor environment, water shortage and infertile land have limited the development of the local economy. the development of digital inclusive finance in five provinces and autonomous regions, namely sichuan, chongqing, yunnan and guangxi, has performed relatively well, among which 0.000 0.100 0.200 0.300 0.400 0.500 0.600 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 41 sichuan and chongqing have a fragile environment, but chengdu in sichuan, as a typical national strong provincial capital, has stimulated the development of its surrounding economy, and chongqing, as a municipality directly under the central government, has led the western region, while guizhou, yunnan and guangxi, although their initial levels are low, are gradually narrowing the gap through government support and orderly local development. figure 2. the level of development of digital financial inclusion in 12 provinces in western china. 4. coordination degree model construction and result analysis 4.1. coupling coordination evaluation method and result analysis table 2. coupling coordination degree interval and grade. no. coherence value coordination level 1 0.000-0.10 extreme disorder 2 0.1001-0.20 severe disorder 3 0.2001-0.30 moderate disorder 4 0.3001-0.40 mild disorder 5 0.4001-0.50 nearly dysfunctional 6 0.5001-0.60 barely coordinated 7 0.6001-0.70 primary coordination 8 0.7001-0.80 intermediate coordination 9 0.8001-0.90 good coordination 10 0.9001-1.00 quality coordination coupling refers to the phenomenon of interaction and influence of multiple systems. the coupling degree can scientifically measure the strength of this action, and according to the research content of this paper, the coupling degree function can be set as: 1 2 1 2* / ( )c w w w w  where c is the coupling degree. it is used to measure the strength and weakness of the interaction between systems or elements, while the degree of coordination is the relationship between systems 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 42 or elements in a coordinated and virtuous cycle, reflecting the degree of system coupling. the larger the value, the higher the coupling degree. coupling coordination function is as: 1 2t w w   , *e c t where e is the coupling coordination degree and t is the comprehensive evaluation index.  ,  are tbd factors, set them both to 0.5. in this paper, the uniform distribution function method is used to delineate the intervals and levels of coupling coordination as shown in table 2. 4.2. analysis of coupling coordination degree results the coupling coordination model was used to measure the coupling coordination between the ecological environment and the level of digital financial inclusion in 12 provinces from 2011 to 2020, which was analyzed by spss 25.0, and the results are shown in table 3. table 3. coupling and coordination of ecological environment and digital inclusive finance in 12 provinces in western china. coupling coordination province 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 gui zhou 0.3323 0.4821 0.5438 0.5853 0.6219 0.6396 0.6848 0.6977 0.7036 0.7057 si chuan 0.4660 0.5871 0.6518 0.6736 0.7105 0.7232 0.7516 0.7766 0.7863 0.7921 yun nan 0.3811 0.5199 0.5902 0.6130 0.6498 0.6671 0.6936 0.7133 0.7322 0.7413 chong qing 0.4365 0.5337 0.5964 0.6180 0.6453 0.6598 0.6809 0.7081 0.7101 0.7135 guang xi 0.4221 0.5370 0.5982 0.6201 0.6612 0.6651 0.6931 0.7025 0.7053 0.7108 shan xi 0.4404 0.5488 0.6024 0.6325 0.6627 0.6704 0.6959 0.7019 0.7064 0.7134 ning xia 0.3775 0.4860 0.5414 0.5761 0.6173 0.6177 0.6491 0.6605 0.6701 0.6805 xin jiang 0.3417 0.4943 0.5686 0.5927 0.6235 0.6209 0.6551 0.6496 0.6557 0.6001 gan su 0.3347 0.4761 0.5463 0.5730 0.6045 0.6038 0.6420 0.6552 0.6623 0.6712 qing hai 0.3704 0.5017 0.5879 0.6226 0.6685 0.6740 0.7071 0.7272 0.7305 0.7401 inner mongolia 0.4142 0.5512 0.6188 0.6438 0.6792 0.6818 0.6972 0.7241 0.7321 0.7368 xi zang 0.3614 0.5099 0.5907 0.6218 0.6635 0.6812 0.7113 0.7286 0.7345 0.7423 based on the comprehensive evaluation index of ecological environment and socio-economic development, the coupling coordination degree model was used to calculate the coupling coordination degree index of ecological environment and digital inclusive finance in the study area from 2011 to 2020, as shown in table 3 and figure 3. the analysis led to the following conclusions: (1) overall, the coupling coordination index of the ecological environment and the development level of digital inclusive finance in the western region shows an increasing trend year by year, indicating that the relationship between the ecological environment and the development level of digital inclusive finance in the study area is improving and tends to be coordinated, and the coupling coordination degree shows a rapid increase, indicating that the ecological environment system and the digital inclusive finance system gradually tend to develop harmoniously and coordinate under human intervention. the level of coordination has been increasing. the regions develop from near-dissonance and mild dissonance to primary coordination level. (2) by region, the coordination level of the five provinces and autonomous regions of xinjiang, xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 43 gansu, qinghai, inner mongolia, and tibet is relatively low compared with other regions but the gap is gradually narrowing. (3) the coupling and coordination of the five provinces and autonomous regions of sichuan, guizhou, chongqing, yunnan and guangxi is better than that of other ecologically fragile regions, while guizhou, as a national important big data development base, is in an economic upswing and neglects the coordinated development of ecological environment and economy, resulting in its coupling and coordination degree is not high. (4) shaanxi and ningxia provinces face greater environmental pressure, which hinders economic development, but shaanxi has a superior geographical location and a more complete industrial structure, and its coupling coordination degree is also considerable. figure 3. coupling and coordination of ecological environment and digital financial inclusion levels in 12 provinces in western china. 5. conclusions by studying the coupling coordination degree of ecological environment and digital inclusive finance in the western ecologically fragile regions from 2011 to 2020, we find that the level of digital inclusive finance development in the western region has been increasing, the quality of ecological environment has been improving, and the coupling coordination development index has been rising, but in terms of its coupling coordination development type, 12 provinces are still in the primary coupling coordination development stage, compared with coastal areas and other compared with coastal areas and other regions, there is still a large gap. this is not only related to their fragile ecological background and scarce natural resources, but also to their industrial structure and development mode. the coupled and coordinated development of ecological environment and digital inclusive finance is influenced by a variety of factors. this paper mainly analyzes and studies the degree of coupled and coordinated development of digital inclusive finance and ecological environment in western ecologically fragile areas, but does not study its evolutionary characteristics, evolutionary mechanism and driving mechanism, which is also the focus of this paper for further research in the future. funding: this research was funded by the special project of key cultivation disciplines and urgently needed disciplines of guizhou university of finance and economics: "research on credit risk prediction and evaluation of big data enterprises" (2020zjxk20) and the guizhou provincial postgraduate research fund 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 xudong cai and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(1), 36-44 doi: https://doi.org/10.54560/jracr.v12i1.320 44 project: "research on digital inclusive finance and multidimensional poverty alleviation in guizhou province: an empirical analysis based on data from 88 counties in guizhou province" (yjskyjj [2021] 127). conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] huang x n, zhang m. study on the coupling coordinated development between china’s provincial ecological civilization construction and financial agglomeration[j]. forestry economics, 2020,42(4):34-45. doi: https://doi.org/10.13843/j.cnki.lyjj.20200622.002. [2] su s l. study on the coupling and coordination of ecological environment and economic development in ningxia hui autonomous region[j]. soil and water conservation research, 2021,28(2):367-374. doi: https://doi.org/10.13869/j.cnki.rswc.20201023.001. [3] jia p y, cheng j n. spatial and temporal changes in the coordinated development of ecology, resources and economy in ecologically fragile areas of shanxi province[j]. china desert, 2020,40(1):179-186. [4] tong p s, shi s x. coupled and coordinated evaluation of ecological environment and economic development in xiamen-zhangzhou-quanzhou urban agglomeration: based on psr-gcq model[j]. forestry economy, 2018,40(4):90-95+104. doi: https://doi.org/10.13843/j.cnki.lyjj.2018.04.015. [5] gu g f, wang x h. spatial and temporal analysis of the coupling relationship between economic development and ecological environment in northeast china[j]. journal of northeast normal university (philosophy and social science edition), 2018(4):154-160. doi: https://doi.org/10.16164/j.cnki.221062/c.2018.04.025. [6] yang j n. dynamic and comprehensive evaluation study of ecological environment in 12 provinces and cities in western china[d]. chongqing normal university, 2011. [7] zhang h m, li t, chen y z. dynamic comprehensive evaluation of ecological environment of 12 provinces and cities in western china[j]. journal of risk analysis and crisis response, 2019,9(3):156-161. doi: https://doi.org/10.2991/jracr.k.191024.005. [8] wei m h, huang q, qiu l, et al. comprehensive evaluation of water environment based on "longitudinal and transverse" staging method[j]. journal of shenyang agricultural university, 2010,41(1):59-63. [9] guo f, wang j y, wang f, et al. measuring the development of digital inclusive finance in china: indexing and spatial characteristics[j]. economics (quarterly), 2020,19(4):1401-1418. doi: https://doi.org/10.13821/j.cnki.ceq.2020.03.12. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word risk analysis techniques in construction engineering projects risk analysis techniques in construction engineering projects daniel baloi department of civil engineering, eduardo mondlane university, av. de moçambique, km 1,5, maputo, mozambique email: baloi@zebra.uem.mz abstract there is a great deal of risk analysis techniques and tools available for the management of risks. in principle, each risk analysis technique has its strengths and weaknesses. techniques such as probability theory, certainty factors, dempster-shaffer theory of evidence and fuzzy logic are discussed with regard to their application to risk analysis in construction engineering projects. suggestions on the most appropriate tools associated with the techniques are also presented. the strengths and weaknesses of each technique are highlighted and discussed. it has been concluded that the nature of risks under consideration is determinant in the selection of modeling and analysis techniques. keywords: risk analysis, construction, projects, techniques, uncertainty 1. introduction risk management constitutes one of the key elements towards effective project implementation and success. the risk management process can be viewed in different ways. from the project management perspective it is common to consider planning, identification, classification, analysis, response, monitoring and control as the main steps. all these steps are interconnected and are part of a system, which means each should be properly addressed so as to enable an effective operation of the whole. risk analysis aims to estimate or assess the likely outcomes or impacts of risks under consideration, in case they materialize. deciding which courses of action to pursue, risk response strategies, is largely based on the results of risk analysis. there is a great deal of risk analysis techniques and tools available. in principle, each risk analysis technique has its strengths and weaknesses and, as such, the option for one specific technique depends on several factors of which the effectiveness to capture the inherent uncertainty is of paramount importance. very often, though, the selection of a specific modeling and analysis technique is more governed by factors such as familiarity, simplicity and availability, rather than the nature of prevailing risks combined with the power to describe uncertainty. this procedure is likely to hinder the validity of the outcomes seriously and, consequently, the whole risk decision-making process. there are instances where qualitative techniques are more effective than quantitative, although the latter may appear to be the most robust and meaningful for many practitioners. again, the right analysis technique is the one capable of adequately capturing and handling uncertainty. the purpose of this paper is to review and discuss risk analysis techniques that would be most suitable to construction management. some of the risk analysis techniques available include preliminary risk analysis, faulty trees, event trees, sensitivity analysis, probability analysis, certainty factors, dempstershaffer theory of evidence and fuzzy logic. four of these methods, probability analysis, certainty factors, dempstershaffer theory of evidence and fuzzy logic are discussed with regard to their application to risk analysis in engineering projects. the main reason for considering these four techniques is their potential to effectively handle uncertainty inherent in construction projects. the study analyzes how suitable the techniques are for the risk categories under consideration, namely organization-specific, global and acts of god. among these, acts of god are obviously handled in a very different manner as ‘’force majeure’’ under contractual terms. the strengths and weaknesses of the techniques are highlighted and discussed. the remaining techniques are not considered since they rarely apply to project risk management. they are more appropriate for problems commonly found in fields such as safety/hazard, biological, pharmaceutical and health. this work is part of a larger study conducted to develop a framework for managing risk factors affecting construction projects costs. the actual detailed modeling and analysis, including computations, are not part of the present paper due to scope delimitation and space limitations. they will be discussed in the next stage. although the study concentrates on construction engineering project risk, it is believed that the risk analysis techniques under consideration can be employed in various other fields as they are cross-cutting. 2. risks in construction projects construction projects as other types of projects pose serious management challenges. geographical dispersion, significant number of players, technical variability, technical complexity journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 115-123 published by atlantis press copyright: the authors 115 administrateur texte tapé à la machine received 18 april 2011 administrateur texte tapé à la machine accepted 8 june 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine         daniel baloi  and large number of inputs are some of the variables that make construction projects challenging. as a result of these interconnections the number and type of risks tend to be significant. risk classification is not a consensual exercise even among the construction community but the most important issue is the recognition of risks and the need to tackle them so as to increase the chance of project success. many different classifications of risk have been developed. however, most of these have considered the source as the most important criterion. following this criterion, one way of classification for construction is as follows: technical; construction; legal; natural; logistic; social; economic; financial; commercial; and political, thompson and perry1. it is likely that apart from construction, these categories of risks apply to a wide variety of other projects such as mineral, manufacturing and transport. a classification taking into account the location of the impact of risks in the elements of the project has been suggested by flanagan and norman2. it is also usual to categorize risks on a wider perspective as dynamic/static, corporate/individual, internal/external, positive/negative, acceptable/unacceptable, short term/long term and insurable/non-insurable. the adoption of source as a means to risk classification gives rise to a myriad of diverse risks that need to be properly managed to enable effective responses delineation. political risks can hardly be handled, in terms of modeling, measurement of likelihood and impact, in the same way as the technical ones, for example. from this reasoning, it follows that it is fundamental to identify the most effective techniques to perform the operation. 3. nature of construction projects risks as it has been discussed, construction project risks were classified in accordance with source as technical, legal, natural, logistic, social, economic, financial, commercial and political. in order to facilitate the analysis and reduce the number of classes, these risks can be grouped at a higher level in the breakdown structure using the concept of environment layers. before that is done it is important to present some important concepts. all organizations exist within an environment and not in a vacuum. the structure of the environment surrounding a project or organization can be subdivided into three distinct layers: outer layer or general environment; operational environment; and inner layer or internal environment. both the general and operational environments can be considered external environments. the general environment comprises domains that are broad in scope and have little immediate and direct impact on the organization’s activities. the general environment comprises five basic elements or domains, namely economic environment, political environment, social environment, technological environment and physical or natural environment. these domains interact with organizations or projects constantly and determine what they might do. in general, some environment domains are less predictable and difficult to understand than others. the operating environment is the external environment comprising factors that have more specific and immediate impact on the organization or project. these factors include suppliers, clients, sub-contractors, consultants and competitors. finally, the internal environment is the inside environment that has direct, close, and immediate impact on the organization. the internal environment is basically concerned with organization’s resources, which include financial, physical, human, and technological resources, as well as managerial values and ethics. physical resources in construction include labor, equipment and materials. system and technological resources are technical capabilities and models used in the operations of the organization, such as quality control processes, reward policies, patents, brands and technologies. in conclusion, the internal environment of an organization identifies its strengths and weaknesses, i.e., what the organization can do. based on both opportunities/threats/constraints from the external environment and strengths/weaknesses/constraints from the internal environment the organization determines what should be done. opportunities and threats are current or future conditions in the environment, whereas strengths and weaknesses are positive and negative internal conditions of the organization. in this context, opportunities/threats/constraints from the external environment and strengths/weaknesses/constraints from the internal environment are strongly related to risks. following the concept of environmental layers approach, project risks are assumed to be associated with the internal, operational and general environments. projects risks are, therefore, classified according to their primary source or cause and the environmental layers as:  organization specific risk factors;  material related  labor related  equipment related  estimator related  management related  construction related  finance  global risk factors  economic  political  competition  project/design  construction  estimation  acts of god  heavy floods;  massive landslides;  earthquakes,  tsunamis; published by atlantis press copyright: the authors 116                                 risk analysis techniques   hurricanes and  others. this classification is important for risk decision making process, since managers are expected to carefully examine both internal and external environments of their organizations or projects in order to evaluate the factors that pose serious challenges to success. the consideration of the relevant primary sources of risks and the subsequent detailed examination of the risks linked to each primary source is crucial in any project. 4. uncertainty categories understanding uncertainty associated with risk categories is crucial for their effective management. uncertainty in the context of the present work refers to lack of sureness about a phenomenon, process, system or event, and can vary from partial sureness to a complete lack of conviction. there are various types of uncertainty inherent in decision-making problems. more often than not decision-making is a challenging endeavor because data and information are scarce, incomplete, inaccurate or unavailable. research on uncertainty types has been conducted over time, but the advances in computing technologies were a significant boost. in recent years, a great deal of knowledge on uncertainty has been produced. helton3, suggested that there are two main types uncertainty, namely random and epistemic. random uncertainty results from the random behavior of a system. this type of uncertainty is also known as aleatory, stochastic, irreducible, objective or variability. random uncertainty can be described as unknowns that differ each time an experiment is made. this kind of uncertainty cannot be minimized by more precise and accurate measurements. however, the likelihood of specific outcomes occurring can be obtained. according to klir4, randomness refers to a situation where the outcomes of an event are rigorously a matter of chance, that is, it is impossible to predict the outcomes. for example, casting a die has several possible outcomes each with known probability (perfect die) or with unknown probability (defective die). epistemic uncertainty, on the other hand, originates from lack of knowledge about a system, object or process. it is also called reducible, subjective, state of knowledge or ignorance. epistemic or systematic uncertainty derives, to a greater extent, from lack of knowledge about a phenomenon, system or process. very often efforts are directed to turning epistemic into random uncertainty, at least, which has proved a daunting challenge. van gelder5 grouped uncertainty into natural variability and knowledge uncertainty, respectively. this classification is similar to helton’s, where variability and knowledge are equivalent to random and epistemic, respectively. subcategories of variability uncertainty include temporal, spatial and individual heterogeneity, whereas knowledge encompasses model, parameter and decision-related uncertainty. model-related uncertainty arises from oversimplification or from failure to capture important characteristics of the process/system. furthermore, failure to understand uncertainty can lead to significant errors due misrepresentation. surrogate variables, excluded variables, abnormal situations, approximation and incorrect form constitute the main sources of model uncertainty. natural variability uncertainty cannot be reduced as it stems from the variability of natural forces. that is the reason why it is also called irreducible. another perspective on uncertainty defines vagueness and randomness, zadeh6. vagueness arises when the meaning of a statement or word is poorly-defined, that is, it lacks precision or sharpness. for example, risk factor strong competition does not have an exact meaning, because the qualifier ‘’strong’’ may assume several degrees of intensity. strong competition may involve a wide spectrum of human perceptions. therefore, there is no rigorous definition of what strong competition is. a particular type of vagueness is fuzziness. fuzziness is a kind of imprecision where the transition from a membership state to a non-membership of an element to a set is gradual. fuzziness is a general characteristic in many areas such as management, engineering, manufacturing, and medicine. it is, nevertheless, most frequent in situations where human judgment is an essential feature such as reasoning, learning and decision-making process as suggested by zimmermann7. from the review, it can be concluded that there are several types of uncertainty associated with natural, managerial, social, economic, technological and political phenomena. the three groups of construction engineering risk factors differ in nature and the underlying uncertainty. assessing the nature of uncertainty is, therefore, crucial for deciding how modeling should be conducted in order to cater for uncertainty attached to risk factors. 5. appropriateness of techniques for risk modeling and analysis the different groups of risks that permeate construction projects have been presented and briefly described. risk managers have been confronted with the choice of right risk analysis technique. is it effective to employ the same technique to model organization-specific, global and acts of gods risk-related factors? the response is not straightforward, but it appears that due to the difference in these risks nature, the underlying uncertainty is also likely to differ. for example, uncertainty attached to acts of god and organization-specific risks are hardly similar. while phenomena such as earthquakes and 200-year period return river discharge are highly unpredictable (stem from natural variability), the impacts of financial performance or human published by atlantis press copyright: the authors 117         daniel baloi  resources competence on a project being implemented by an organization can be reasonably assessed. the following section discusses the effectiveness of four risk analysis techniques to tackle uncertainty inherent in organization-specific, global and acts of god related risks. furthermore, suggestions on the most appropriate tools that can be used to materialize the above mentioned analysis techniques are briefly advanced. the analysis is expected to shed some light on the diversity of modeling and analysis perspectives as well as their effectiveness. 6. probabilistic analysis technique it is commonplace to model and analyze risks through available specific techniques/tools. most of the tools available in the market are based on probability theory. as such, probabilistic modeling and analysis has been the prominent way to handle risks regardless their nature. those who argue in favor of probability theory as the foremost technique state that random methods are the only effective methods for dealing with uncertainty. on the other hand, uher and toakley8 noted that the uncertainty inherent in real risk situations was epistemic rather than aleatoric (matter of chance). this view corroborates the general perception that risk management in construction engineering is eminently cognitive. indeed, the process is mainly based upon experience, assumptions and human judgment. it appears that reliance on the probability theory as the only effective and reliable methodology to deal with uncertainty has historical roots. indeed, probability theory has wellestablished and sound scientific foundations and has been widely used for centuries in all spheres of science. probability is a branch of mathematics concerned with random phenomena. as such, it deals with stochastic processes and events through frequentist, outcomes of repeated experiments, and subjective views. probability theory has been widely used to model precisely described, repetitive experiments with observable but uncertain outcomes. 6.1 probability basics the basic assumption in the classical theory of probability is that all types of uncertainty are ‘’frequentist’’ measures of randomness or subjective measures of confidence. this reasoning is clear in the definition of probability, as follows: if a random experiment has n possible outcomes which are all equally likely and mutually exclusive and n of these possibilities have outcome a, then the probability of outcome a is n/n. the basics of probability theory is, thus, the elicitation of probabilities, which may either be objective or subjective, in order to predict the likelihood of uncertain events. objective probabilities are those directly derived through experiments or statistical data. objective probabilities apply to repeatable events only. subjective probabilities, on the other hand, represent degrees of belief of the decisionmakers. individuals with diverse backgrounds are entitled to assess the likelihood of the same events differently. 6.2 organization-specific risks organization-specific risks are related to the internal environment of organizations and these are supposed to manage them. organizations’ level of knowledge about the operations is significantly higher than other type of risk. indeed, issues such as planning, finance, human resources, equipment and materials logistics are dealt with on a daily basis. organizations can conduct operations studies in order to build robust databases and learn systematically. very often databases are used as a means for routine programmed decision-making. the large quantity of information obtained in this way can be retrieved in order to make probabilistic estimates. for example, important aspects such as labor productivity and cost can be estimated in probabilistic terms in the planning process through probability distributions. from this point of view, it appears that probabilistic modeling and analysis is suitable for this kind of risks. as a great deal of knowledge is available, predictions about the likelihood and impacts of risks can be made with some degree of certainty. the prominent uncertainty associated with organization-related risks is assumed to be eminently random. in term of tools, simulation and analytical models can be utilized for risk modeling analysis. one of the most effective tools for this purpose would be decision support systems – dss, which are computer information systems that provide information in a given domain of application by means of analytical decision models and databases, in order to support a decision-making in complex and ill-structured problems. 6.3 global risk factors global risk factors are beyond organizations’ control. they relate to a more complex, erratic and dynamic environment. as such, the possibility for collecting data and information in order to draw useful lessons becomes limited. even taking into account the cyclic nature of some factors such as economy ups and downs, it is very difficult to make reasonable estimates about them. although there have some attempts to estimate the degree of risk associated with global risks the amount of available data make the exercise difficult. an example of probabilistic modeling applied to global risk factors is the production of political risk indices covering a large number of countries around the world as explained by bremmer9. the exercise has been conducted and the results marketed systematically over the years. the indices measure the stability perception over specific countries and aim to help investors and other interested parties in their business published by atlantis press copyright: the authors 118                                 risk analysis techniques  decision making. political variables are complex and dynamic and, as consequence, very difficult to grasp. in general, political phenomena are rarely repetitive, which creates serious challenges for effective modeling. the same applies to the economic and social variables. furthermore, political, economic and social events are rarely mutually exclusive, exhaustive and conditionally independent. first, there have been strong interrelationships among these variables. economy, for example, is strongly associated with social and political events. in the light of these characteristics it can be concluded that probabilistic modeling of global risk factors poses serious challenges in terms of robustness and validity. the uncertainty inherent in this group of risk factors is much more epistemic rather than random. 6.4 acts of god acts of god are generally classified as ‘’force majeure’’ under the contractual terms, because no party is supposed to be able to manage them and, consequently, bear the associated costs. they represent extreme events stemming from nature and, as such, tremendously difficult to predict. it would be important for a contractor, for example, to know the likelihood of an earthquake striking and disrupting construction works. unfortunately, such estimates are not available. that is the reason why the best response strategy to this type of risk has been transfer to a third party. attempts have been made to predict the probability of occurrence of these events, but very little progress has been achieved so far. concerns over the problem have led analysts to develop some probabilistic models but, it should recognized, these can hardly be robust due to prevailing nature of uncertainty. therefore, such estimates may be precise but not accurate. in addition, since most nature-related risk factors are poorly defined and fuzzy in nature they cannot be evaluated with such high precision inherent in numerical expressions. these models have used subjective probabilities instead of objective ones, as the latter are difficult to obtain due to lack of data. the fact that subjective probabilities cannot be distinguished from objective ones once in the model, is likely to hinder the quality of the analysis. there is guarantee that subjective probabilities capture uncertainty effectively. as a result, probabilistic modeling and analysis of nature-related risks is debatable. 7. certainty theory analysis technique 7.1 global risk factors and acts of god as previously shown, while organization-related risk factors can be well modeled through probabilistic means, global risks and acts of god are far more difficult to handle in the same way due to the inherent kind of uncertainty and scarcity of data and information. certainty theory is mainly a theory for handling uncertainty in knowledge based systems kbss. it was developed in attempt to overcome some of the weaknesses of the so called idiot bayes approaches for inexact reasoning, according to duda.10 certainty theory relies on defining judgmental measures of belief rather than adhering to strict probability estimates. therefore, certainty factors (cf) are not probabilities but informal measures of confidence for a piece of evidence. they represent the degree to which people believe that the given evidence is true. in other words, they express how accurate, reliable, truthful people judge statements or evidences. certainty theory fundamentals are the concepts of certainty measures which are associated with factual statements. the certainty measures or factors cfs consist of numbers ranging from –1 to +1 and factual statements, (rules). a negative value of the certainty factor indicates that one believes that a fact is not true and a positive value indicates the one believes that a fact is true with complete knowledge. cf = 1, there is complete certainty that a proposition is true cf = -1, there is complete certainty that a proposition is false cf = 0, there is no information at all about or no change in belief -1 < cf < 1, measure of the degree of belief about the proposition with decreasing and increasing beliefs respectively. global risk factors and acts of god in engineering projects can be modeled using cfs within knowledge-based systems, where the following format is common: if a then b with certainty factor cf = cf (rule), where a is the antecedent and b, the consequent. the antecedent comprises facts (evidence) that support the derivation of the consequent (hypothesis). the cf is the net degree of belief in hypothesis, given that the evidence is observed (given). for example: if the rate of inflation increases then the prices will be high cf = 1, where the rate of inflation increases is the evidence and the prices will be high is the consequence. the degree of belief associated with the rule is 1, which means the analyst is 100% sure. in a typical knowledge based system there are numerous rules of this kind that cover most of the project situations. for global risk factors alone, hundreds of rules would be required in the kbss. the formal representation of evidence for, against the hypothesis and the composite certainty factors, that is degree of belief mb, degree of disbelief md and cf, is defined in terms of hypothesized mathematical relationships governed published by atlantis press copyright: the authors 119         daniel baloi  by prior and posterior probabilities of the propositions that they describe, as stated by clark11. in practical applications of certainty factors rules, hypothesis and input data have an associated mb and md. then, the composite certainty factor cf is calculated by aggregating the values of mb and md. if the information contained in the premise is not known with certainty (cf < 1), then the level of belief in the conclusion is reduced. the original function used in the aggregation is as follows: cf[h,e] = mb[h,e] – md[h,e]; this original formula is very often modified for convenience and it becomes cf[h,e]=(mb[h,e]-md[h,e]): {1min(mb[h,e], md[h,e])}. where cf[h,e] is the certainty factor of the hypothesis h given the evidence e, mb the degree of belief, md the degree of disbelief in h given e. both mb and md values vary from 0 to 1. cf is also known as the composite certainty factors and varies between –1 and +1. the value of mb and md are calculated as explained in detail by shortliffe and buchanan: mb[h,e] = 1, if p[h] = 1 mb[h,e] = {p[h | e] – p[h]}: {1 – p[h]}, if p[h]  1 md[h,e] = 1, if p[h] = 0 md[h,e] = { p[h] p[h | e]}: p[h], if p[h]  1 where p[h] is prior probability of an hypothesis h; p[h,e] is the posterior probability of the hypothesis given some evidence e. 7.2 tools knowledge based systems kbss appear to be the more appropriate tools for the application of certainty factors due to the way knowledge (rules) is represented within the system. knowledge based systems are particular applications of the dsss. essentially, a knowledge based system is a computerized system that uses knowledge about some domain to arrive at a solution to a problem from that domain. 7.3 summary there are several weaknesses associated with certainty factor theory. first, the measures of belief seem to be largely subjective. they depend on the individual making the assessment and can vary for same individual based on specific circumstances. therefore, it is not easy to ascertain how accurate the measures are. blockely and baldwin12 argued that certainty factor theory is an ad-hoc approach and therefore, lacking a formal foundation. the authors criticized the use of parallel combination function for assuming conditional independence and the negation of each hypothesis evidence and considered the theory, worse than the idiot bayes model which assumes only one of these propositions. additionally, it appears that the theory does not suit rulebased models because of its non-modular operational requirements. furthermore, it has been suggested that the cf formulae have little validity and the consultation in cf logic mode tends to require more work on the user's part than is required by the normal binary logic mode. 8. dempster-shafer theory of evidence analysis technique dempster-shafer13 theory of evidence is usually called epistemic probability because it provides an alternative model for the assessment of numerical degrees of belief. in principle, it helps to overcome the handicaps of certainty factors and can be applied to model phenomena whose behavior falls under epistemic uncertainty. indeed, dempstershafer attempts to distinguish between uncertainty and ignorance. it uses belief functions instead of probabilities. in this sense it can be considered a generalization of the bayesian theory of subjective probability and that is why it is also called theory of belief functions. these degrees of belief may or may not have the mathematical properties of probabilities. the theory gained momentum in the 80’s when researchers made some attempts to adapt probability in expert systems. the dempster-shafer theory is based on two core ideas, namely obtaining degrees of belief for one question from subjective probabilities for a related question, and dempster's rule for combining such degrees of belief when they are based on independent items of evidence. to illustrate how it works, a company may be interested to assess the fiscal policy risk in a region or a country where it is executing a project. this is usually a crucial factor as project financial/economic viability can be largely influenced. it is obvious that objective probabilities for this kind of risk are hard to estimate. the company can, therefore, resort to subjective probabilities and then develop belief functions. the question being asked would be ‘’ is fiscal stability and conducive to business?’’. if this probability is set at 90%, for example, then, the probability of the contrary (unstable and detrimental to business) would be 10%. in such circumstances, the company can be confident that the system is credible and good. the degree of belief attached to the system’s reliability is then 0.90 and the degree of belief that it is unreliable is consequently 0.10. a second question would be – ‘’was a specific corporate taxation rate fair/balanced?’’. based on the inherent credibility, the fairness of the taxation justifies a degree of belief of 0.90. the degree of belief of it being unfair is 0.0 and not 0.10. the later degree of belief does not necessarily mean sureness about an unlikely unfairness. it simply means that the reliability of the policy gives no reason to believe the contrary. the numbers 0.90 and 0.0 together constitute a belief function. the illustration published by atlantis press copyright: the authors 120                                 risk analysis techniques  presented explains the above-mentioned fundamentals of dempster-shafer theory, namely the degrees of belief attached to two related questions and the rules for combining them. the degrees of belief associated with the specific taxation fairness are determined on the basis of the reliability of the policy/system which has an initial degree of belief. dempster-shafer theory provides a platform (rules) for combining various degrees of belief. for example, different risks with potential to influence project success can be considered for assessment. having attached degrees of belief on these variables would enable development and combination of belief functions following appropriate rules in order to ascertain the likelihood and impact of different factors. the main assumption behind the rule is the a priori independence of the questions for which probabilities have been estimated. 8.1 tools knowledge based systems kbss appear to be the more appropriate tools for the application of certainty factors due to the way knowledge (rules) is represented within the system. 8.2 summary the main distinctions between bayesian models of numerical degrees and dempster-shafer model are the following: belief functions of dempster-shafer are set functions rather than point values; rejection of the law of additivity for belief in disjoint propositions; and dempster-shafer theory has an operation for the pooling of evidence from various sources. dempster-shafer theory is richer in terms of semantics since it allows an expression of partial knowledge. its main shortcoming is, however, the elicitation and interpretation of belief functions. furthermore, the computational methods employed in the theory are very complex and thus, of little practical use. this is the reason why it has had very little application. 9. fuzzy set analysis technique fuzzy set theory is a branch of modern mathematics that was formulated by zadeh14 to model vagueness intrinsic to human cognitive processes humanistic systems. since then, it has been used to tackle poorly-defined and complex problems due to incomplete and imprecise information that characterize the real-world systems. it is therefore suitable for uncertain or approximate reasoning that involves human intuitive thinking. due to the inherent uncertainty, it seems that fuzzy sets are more suitable to handle global risk factors and acts of god than other approaches. in fact, very often, individuals involved in engineering projects deal with these risks in an approximate manner, using natural language, and not probabilities to assess their likelihood and impacts. risk management can, to large extent, be considered cognitive in nature. the process of risk management in construction industry is mainly based upon experience, assumptions and human judgment. zadeh asserted that fuzzy sets handle vagueness type of uncertainty better than any other approach and vagueness is a distinct type of uncertainty to be treated by probabilistic methods. according to the fuzzy set theory a meaning in natural language is a matter of degree. considering one variable of global risk factors a question can be formulated:  is the competition strong?. the answer is not always simply ‘’yes or no’’. it largely depends on the subject who is to respond. the meaning of strong is not precise. it varies from individual to individual. the basic fundamentals of fuzzy set theory are the concepts of linguistic variable and degree of membership. since most construction engineering decision problems are complex and imprecise, they might be better described by linguistic expressions rather than by numbers. numbers are associated with precision, whereas decision problems like managing risks need not or do not have specific outcomes but approximate ones. furthermore, due the imprecision inherent to linguistic expressions the transition from one state to another is smooth. for example, the transition from very heavy floods’’ to heavy floods is not sharp, but gradual. linguistic expressions play an important role in this regard because description is at the very core of risk management in construction. finally, fuzzy sets have the ability to preserve the uncertainty inherent to the problems throughout the analysis instead of making assumptions. to illustrate the use of fuzzy sets in modeling risk factors a variable designated ‘’ low competition’’ is chosen as shown is figure 1. the membership functions were derived from interviews with engineering professionals. figure 1: graphical form of fuzzy set “low competition”, where a is the grade of membership; low competition is fuzzy set a; and1, 2 and 3 are the elements of the universe of discourse. the fuzzy set ‘’low competition’’ can also be represented analytically: alow competition=[1.0/0, 0.6/2, 0.0/3] where 1.0, 0.6 and 0.0 are the degrees of membership. thus, the membership function of low competition is [1.0, 0.6, 0.0]. low compet it ion a 1 5.0 1 2 3 published by atlantis press copyright: the authors 121         daniel baloi  apart from low competition other values would be moderate competition and high competition. the corresponding fuzzy sets for these values may be the following: bmoderate competition=[0.0/2, 0.5/3 1.0/4, 0.5/5, 0.0/6]; and chigh competition = [0.0/5, 0.5/6, 1.0/7, 1.0/8, 1.0/10]. important to notice, though, are the areas of overlap between low and moderate and moderate and high competition membership functions. these overlaps represent the grey zones, that is, the gradual transitions from membership to non-membership properties. in a crisp set an object is either a member of a set or not. the only grades of membership in the crisp set low competition are 1 and 0. thus, competition levels from 0 to 3 are members of the set with a grade 1 and, all other elements of the universe are not, and they have grade membership of 0. as it can be seen, the change in the grades is quite abrupt. using similar reasoning, all global risk factors and acts of god can be modeled and analyzed based on fuzzy sets. the major challenge lies with the development of membership functions for the different variables. there are various ways for that purpose of which interviews and discussions with project team members is the most prominent. existing procedures and rules for combining membership functions allow performance of complex operations. furthermore, there are now many software shells that can be used to develop effective models and run analysis. the inputs and outputs of fuzzy set analysis are natural language expressions which facilitate interpretation. 9.1 tools a combination of knowledge based systems and decision support systems resulting in a kbs-dss can be a useful strategy when employing fuzzy sets approach. the first part houses the vast quantity of knowledge whereas the second one provides support to decision-makers. the need for a kbdss stems from the fact that a system capable of processing data and numeric relationships, on the one hand, and transforming these computations, through reasoning, into judgement, opinions, evaluations and advice, on the other hand, would increase the quality of data processing and thus provide better support to decision-making. 9.2 summary fuzzy set theory is not without problems. there is a strong controversy around the fundamentals of probability theory and fuzzy set theory. some scholars have argued that fuzzy sets theory is no more than a false appearance of probability theory. for most of the critics, fuzzy sets has no well established mathematical or empirical methods to model human judgment as stated by lootsma15 . the main sources of the criticisms are probably the fundamentals of fuzzy sets theory which are clearly in contradiction with the dominant scientific view of the world precision. detractors of fuzzy set theory argue that it is probability theory in disguise, as stated by cheseeman16. 10. summary of the techniques the strengths and weaknesses of the techniques that have been discussed are summarized in table 1. for each of the four risk analysis techniques, the evaluation of type of risk, prominent uncertainty and tools to be employed are shown. the indications provided in the table constitute a guide on what a specific technique can help to manage. for example, probabilistic modeling can be effectively to model organization specific and global risks factors where the underlying uncertainty is mainly random. for that purpose decision support systems dss seem to be the most appropriate tools. on the other hand, probabilistic modeling is very poor in tackling natural events or acts of god as data for these is scarce and frequently unavailable. the same procedure is followed with regard to other risk analysis techniques. 11. conclusions the importance of managing construction engineering risk factors has been discussed. an extensive analysis of table 1: summary of risk analysis techniques and tools  modeling and analysis suitability no technique risk group prominent uncertainty tools (a) organization -specific (b) global (c) acts of god 1 probability very good very poor very poor random (a) dss 2 certainty factor very poor good good epistemic (b and c) kbs 3 dempster_shaffer theory of evidence poor good good epistemic (b and c) kbs 4 fuzzy set theory poor very good very good epistemic (b and c) kb-dss published by atlantis press copyright: the authors 122                                 risk analysis techniques  construction projects implementation with emphasis on cost performance has triggered work on risk modeling and analysis. it has been concluded that the nature of risks under consideration is determinant in the selection of modeling and analysis techniques. as it has been shown, not all uncertainty is random in nature. a great deal of engineering management and technical issues do not comply with randomness properties. they are mainly cognitive and thus do not lend themselves to precise measurement. three groups of risk factors inherent in construction engineering projects have been presented and explained. it is understood that the nature of the risk factors is diverse and thus their handling requires appropriate techniques and tools. indeed, the type uncertainty associated with each of the groups differs. while organization-specific risk factors can be effectively modeled through probabilistic analysis, uncertainty underlying global and acts of god risk factors appears to be much more difficult to capture using the same approach. the uncertainty type associated with the latter risk factors derives mainly from reduced knowledge rather than natural variability. as such, epistemic uncertainty handling techniques are likely to be more effective than probabilistic approaches. several uncertainty modeling techniques namely uncertainty theory, dempster_shaffer theory of evidence and fuzzy logic have been discussed with emphasis on appropriateness and robustness. although there is no consensus on the applicability of these techniques to bridge the gap between the ideal and feasible solutions provided by the probabilistic analysis, it appears that they can be employed for such purpose. in particular, dempster-shafer theory and fuzzy logic are capable of modeling epistemic uncertainty through belief and membership functions. some examples of fuzzy logic knowledge-based-systems using inference have been developed. nevertheless, research is required to ascertain the extent to which these and additional techniques can be applied to improve construction engineering risks modeling and analysis. references 1. thompson, p. and perry, j., engineering construction risks. a guide to project risk analysis and risk management. (london: thomas telford. 1992) 2. flanagan, r and norman, g., risk management and construction. (london: blackwell. 1993) 3. helton,, j uncertainty and sensitivity analysis in the presence of stochastic and subjective uncertainty. “journal of statistical computation and simulation” 57: 3-76 (1997). 4. klir, g. where do we stand on measures of uncertainty, ambiguity, fuzziness and the like?. fuzzy sets and systems, 1987, 24, pp.141-160. 5. van gelder, p.h.a.j.m., "statistical methods for the riskbased design of civil structures", 2000, communications on hydraulic and geotechnical engineering, issn:0169-6548001, 248pages. 6. zadeh, l. a., fuzzy sets. information and control, 1965, 8, pp.338-353 7. zimmermann, h. j. and haley, k.b. theory and applications of fuzzy sets. operational research, 78, (amsterdam: northholland. 1996) 8. uher, t., probabilistic estimating. cost engineering, 1996, 38(4), pp. 33-40. 9. ian bremmer, “how to calculate political risk,” inc. magazine, april 2007, p. 101 10. duda, r. o. et al development of prospector system for mineral exploration. final report, sri projects 5821 and 6415, (menlo park, ca, 1979). 11. clark, d.a., numerical and symbolic approaches to uncertainty management. in ai: artificial intelligence,, 1990 4, pp.59-91. 12. blockley, d. i. and baldwin, j.f., uncertain inference in knowledge-based systems. asce, journal of engineering mechanics, 1987, vii(3),. 13. shafer, g., mathematical theory of evidence. (princeton: princeton university press,,1976). 14. zadeh, l. a. (1978) a meaning representation language for natural languages. international j. of man-machine studies, 10(4), pp. 395-460. 15. lootsma, f.a. (1997) fuzzy logic for planning and decision making. the netherlands: kluwer. 16. cheeseman, p. c. (1965) in defence of probability. proceedings, international joint conference on ai,, pp.10021009. published by atlantis press copyright: the authors 123 microsoft word volume 11, issue 4-5 journal of risk analysis and crisis response, 2021, 11(4), 198-199 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i4.313 198 communication focus on risk crisis management! 2021 symposium on disaster risk analysis and management in chinese littoral regions was held in wenzhou anping pan 1,* and yutong luo 2,3 1 college of civil engineering and architecture, wenzhou university, wenzhou (325035), zhejiang, china 2 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 3 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: pap3888@163.com; tel.: +86-0577-86598000 received: november 28, 2021; accepted: december 28, 2021; published: january 25, 2022 from november 27 to 28, 2021 symposium on disaster risk analysis and management in chinese littoral regions was held in wenzhou. with the theme of "strengthening coastal disaster risk management and serving the 'double carbon' strategic goal", this seminar focused on the latest research results, practical experience, development trends and difficulties in the fields of risk analysis theory and methods, disaster prevention and mitigation, and dual carbon technology. extensive academic exchanges on hot issues are a grand event in the field of risk analysis. affected by the epidemic, this academic conference will adopt a combination of online and offline live broadcasts. in recent years, affected by climate change and sea-level rise, coastal cities have faced severe challenges from a series of disaster risks such as heatwaves, floods, droughts, forest fires, and a sharp decline in biodiversity. social development brings many risks. carbon peaking and carbon neutrality have gradually become hot issues and scientific frontiers that the international community and academia are generally concerned about. this seminar received more than 40 paper submissions from more than a dozen provinces including beijing, tianjin, shanghai, hubei, hunan, guizhou, zhejiang, shaanxi, jiangsu, tibet, guangxi, sichuan, qinghai, and yunnan. this seminar is sponsored by the risk analysis professional committee of china disaster defense association, undertaken by the school of civil engineering and architecture, wenzhou university, and co-organized by the real estate institute of wenzhou university, the institute of disaster prevention and mitigation engineering of wenzhou university, and the journal of risk analysis and crisis response (jracr). the chairman of the risk analysis professional committee of china disaster defense association and professor huang chongfu from beijing normal university served as the chairman of the conference, and professor pan anping from wenzhou university served as the chairman of the conference program committee. the opening ceremony and keynote speeches were held on the first day of the conference, and 16 sub-forums were discussed on the second day. the opening ceremony on the morning of the 27th was presided over by wang jun, dean of the school of civil engineering, wenzhou university, tang bao, secretary-general of china disaster anping pan and yutong luo / journal of risk analysis and crisis response, 2021, 11(4), 198-199 doi: https://doi.org/10.54560/jracr.v11i4.313 199 defense association, professor zhao min, president of wenzhou university, professor shi kan, supervisor of the chinese psychological society, and risk analysis major of china disaster defense association the chairman of the committee, professor huang chongfu, delivered speeches and congratulated on the successful holding of the seminar. in the opening speech, president zhao min introduced the basic situation of wenzhou university, saying: " wenzhou university will take this seminar as an opportunity to further strengthen exchanges and cooperation with leaders, experts and guests, promote the theoretical research and achievement transformation of risk analysis and management in coastal areas, and contribute 'wen da's wisdom' to china's disaster risk research!". this seminar invited 11 big shots who have made outstanding contributions in the field of risk to give presentations. they are chinese academy of sciences, professor kan shi of wenzhou university, professor chongfu huang of beijing normal university, professor xiaojun li of beijing university of technology, northeast normal university professor jiquan zhang, professor guofang zhai of nanjing university, researcher xueliang chen from the institute of geophysics, china earthquake administration, professor ren zhang from national university of defense technology, professor jiahong wen from shanghai normal university, professor hongbin cao from beijing normal university, professor anping pan from wenzhou university, and sen qiao from institute of geophysics, china earthquake administration researcher. the conference report focused on coastal areas risk analysis theories and methods, disaster prevention and mitigation, real estate risk prevention and control and other fields, sharing the latest research results and progress in coastal disaster risk analysis, evaluation, and management, which will greatly promote the development of coastal cities and regions. comprehensive disaster prevention and mitigation and sustainable development. on the 28th, a branch forum was held on the three themes of "natural disaster risk", "risk prevention and control research" and "financial risk". a total of 32 scholars shared their latest research results. the conference report session and each sub-forum attracted the participation of many experts and scholars. the discussion atmosphere in the venue was enthusiastic, which strengthened the exchanges between domestic risk scholars and achieved the expected goals. it is reported that the symposium on disaster risk analysis and management in china's coastal areas has been held every two years since its first conference in shanghai in 2011. this conference is the fifth. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). research article research on information channel of climate change risk perception of shaanxi people si wen xue1,3,*, , qi zhou1,2,3 1college of geography and environment, baoji college of arts and sciences, baoji 721013, china 2shaanxi key laboratory of disaster monitoring and mechanism simulation, baoji college of arts and sciences, baoji 721013, china 3key research center of socialism with chinese characteristics of shaanxi province (baoji base), baoji 721013, china 1. introduction in the context of global climate change, major natural disasters happen frequently, and risks of climate change has further increased. the public are not only the most extensive and direct disaster bearers of climate change risk events, but also the most specific executors of disaster prevention and mitigation policies [1,2]. people’s ability to perceive climate change risks greatly influences their response ability. in other words, perception determines action [3]. an in-depth study of public climate change risk perception is an effective way to improve the public’s ability to cope with climate change risks and reduce their vulnerability [4,5]. it also has certain practical significance for the research on national climate change risk perception and response. as scholars continue to deepen their research on climate change risk perception [6], climate change that attracts worldwide concern has gradually transformed into a scientific topic concerning the public. in this process, due to differences in climate change risk perception and knowledge between scientists and the public, the dissemination of climate change information has become an important platform for communication between scientists and the public, which directly influences whether the public can achieve favorable communication with governments and scientists. the dissemination channels and sources of climate change risk information determine whether people can accurately recognize climate change as a macro-abstract natural phenomenon, thereby influencing their attitudes and behaviors toward climate change risks. smith [7] held that media culture, technology and practice create the opportunity to enhance public’s understanding and identification of climate change risks. studies, such as by maria carmen lemos, indicated that there is a gap between useful information understood by scientists and useful information recognized by users [8]. hmielowski [9] also found through several studies that trust in scientists influences the use of news media, which in turn influences the understanding of global warming. lack of information was repeatedly identified by archie [10], among others, as an obstacle to climate change adaptation planning and implementation. lynch [11], among others, suggested that the multiple utilization of communication tools will facilitate climate change science, as well as mitigation and adaptation policy formulation. carmichael and brulle [12] using structural equation models showed that although media reports play an important role, they are largely the result of elite suggestion and economic factors [12]. julia et al. [13] noted that similar to six inter-american studies of global warming, different attitudes (the five germanys of global warming) result in differences in understanding climate change, media use, and communication behavior. john wiley & sons believed that key aspects of the communication process (including the purpose and scope of communication, the audience, the framework, the message, the messenger, the means and channels of communication, as well as the evaluation of the results and effectiveness of communication) influence climate change risk communication [14]. moser and dilling [15] found in their study that the lack of information and understanding explains the lack of public participation, so more information and explanations are needed to a r t i c l e i n f o article history received 26 december 2020 accepted 29 march 2021 keywords shaanxi province ann neural network model cart decision tree model information channel of climate change risk perception a b s t r a c t we take shaanxi province as the research area, aiming at exploring the information channel or path of climate change risk perception of shaanxi people. it is desirable for us to carry out information channel or path classification of climate change risk perception based on survey data involving 5493 people in shaanxi province. firstly, we use a back propagation (bp) neural network method to fit the information path of climate change risk perception. secondly, a decision tree model is adopted to classify information channels of climate change risk perception. the results show that 300 neurons are needed in the information channel of climate change risk perception of shaanxi people. the first path which influences climate change risk perception of shaanxi people is as follows: indirect perception–direct perception–indirect perception–conductive perception. the second path is indirect perception–conductive perception. the third path is as below: indirect perception–direct perception–conductive perception, which also impacts climate change risk perception. according to varying information channels or paths of climate change risk perception, the public can formulate different risk management strategies to improve the level of climate change risk perception. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 1213268775@qq.com journal of risk analysis and crisis response vol. 11(1); april (2021), pp. 36–44 doi: https://doi.org/10.2991/jracr.k.210331.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-0809-5185 http://creativecommons.org/licenses/by-nc/4.0/ mailto:1213268775%40qq.com?subject= https://doi.org/10.2991/jracr.k.210331.001 https://www.atlantis-press.com/journals/jracr s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 37 motivate people to take action. lazrus and et al. [16] investigated in florida how people express their vulnerability or activity in receiving, interpreting, and responding to hurricane risk information. there are many types of researches on climate change policy influencing factors supported in the united states, but relatively few researches on individual climate change policy support. for illustration, dan m. kahan et al. [17] conducted studies in two countries (usa, n = 1500; united kingdom, n = 1500) to test the potential value of a unique two-channel science communication strategy that combines information content (“channel 1”) with cultural significance (“channel 2”) and is chosen to promote open assessment of information from different communities. contrary to this hypothesis, we believed that subjects exposed to geoengineering information were concerned more about climate change risks than control conditions [17]. xie et al. [18] used recent climate change risk perception models to predict risk perception and willingness of australians to engage in mitigation behavior (n = 921), and highlighted the influence of information emotion, cognition, and sociocultural factors on climate change risk perception. researches based on disaster event network data by lin et al. [19] not only explored the spread of disaster information in time and space, but also investigated in detail the public natural disaster perception. science and technology daily reporter hepeng [20] argued that compared with a large number of international counterparts, chinese scholars paid little attention to the spread of climate change in the theme of communication. therefore, the spread of climate change is worth an urgent and serious study in china. at present, research on climate change risk perception channels at home and abroad has achieved certain results, but in general the most urgent concern to domestic and foreign scholars is a specific report on climate change itself: namely an in-depth analysis of how media under the joint influence of social sciences and local governments rebuild a theoretical framework and a report of china’s climate change. a few scholars have analyzed the dissemination channels of global climate change risk information from the public perspective by investigating climate change perception data, thereby proposing specific information dissemination types and formulating corresponding climate change risk management strategies [21]. according to research methods, structural equation model is the widely used one, and artificial neural network (ann) neural network is frequently adopted to fight climate change risk sensing channel. the ann neural network is a complex network structure formed by a large number of processing units. it is an information processing system established by imitating the function of brain neural network structure, and composed of sensor units with a single-layer structure [22,23]. the unit can not only make most of the calculation and complete linear function, but also simulate the signal process of human brain nerve cells based on nonlinear expression ability according to nonlinear functions. this method is close to the perception process of climate change risk information in this paper, can better simulate the human brain’s perception process of climate change risk information, and is more suitable for the study of the information channel of climate change risk perception. besides, the classification and regression trees (cart) decision tree algorithm is a binomial tree model algorithm, which has the advantages of high precision and fast operation speed. it supports discrete and continuous data and can be applied to classification and regression [24]. at the same time, the cart decision tree model is suitable to classify information channels of climate change risk perception and to locate factors influencing climate change risk perception. on the other hand, shaanxi province suffers repeated meteorological disasters, including drought and frost, and a rainstorm to varying degrees will occur almost every year [25]. as a result, shaanxi people are more sensitive to climate change risks and have conducted certain research [26]. therefore, this paper will focus on the information channel of climate change risk perception, which is innovative in the field of climate communication. this paper will mainly use the ann neural network model to simulate the information channel of climate change risk perception in shaanxi province, and then use the cart decision tree model to study the major information channels that influence climate change risk perception of shaanxi people. hopefully, the information channel classification of climate change risk perception will provide references for climate change risk management. meanwhile, with the information channel of climate change risk perception as a breakthrough point, different policy strategies will be proposed for different information channels. 2. materials and methods 2.1. overview of the research area shaanxi province is located in the inland central part of china, with a total area of 205,600 km2 and a population of about 36 million. there are various landforms, which can be divided into three types: the loess plateau in northern shaanxi, the guanzhong plain and the mountains in southern shaanxi [21]. it belongs to continental monsoon climate in general, and covers three climatic zones from south to north. specifically speaking, northern shaanxi, guanzhong and southern shaanxi belong to the temperate zone, warm temperate zone and north subtropical zone respectively. the annual precipitation is 275–1274 mm, increasing gradually from north to south with obvious precipitation differences. the natural climates of the three landforms vary greatly. with global warming ongoing, hydrothermal conditions are enhanced, and temperature and precipitation in shaanxi are also enhanced. local heavy precipitation is frequent in summer and autumn. due to unique topographic and geomorphic conditions, the frequency and intensity of geological disasters are increased. especially in the loess plateau of northern shaanxi and qinba mountainous area of southern shaanxi, mud-rock flows, wildfires and landslides occur frequently, causing huge loss of personnel and property. worse still, increased rainfall and rising temperature have also led to varying degrees of drought (figure 1). 2.2. data source and the questionnaire content from july 2018 to october 2020, and based on social survey of national natural science foundation of china, the public in 10 prefecture-level cities in shaanxi province were visited and surveyed successively. the field survey was mainly conducted in urban areas with large population increases and in rural households. the interviewees were interviewed face to face by the random survey (figure 1). no guiding answers were given to the interviewees during the interview. in the presence of older or less educated 38 s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 respondents, the interview was conducted with full explanation of the topic. this investigation method also ensured authenticity and reliability of the data and laid a foundation for the follow-up research work. to make the analysis results of the questionnaire truthful and reliable, after manually screening sample data from collected questionnaires, questionable questionnaires were screened out, such as blank questionnaires and questionnaires with inconsistent answers. then, we entered the sample data based on the remaining questionnaires. first, we entered the sample data on the questionnaire into the excel table, and then imported data into spss25.0 for subsequent investigation and analysis. before analyzing data and results of the input questionnaire sample, first of all, options and results of questionnaire sample data were assigned. meanwhile, some missing values and outliers in answers of questionnaire sample were checked and modified or discarded to avoid possible errors in the experimental results. after the initial manual screening, 277 questionnaires were discarded due to too many blank questions or abnormal answers. finally, investigators got 5493 valid questionnaires, and the recovery rate of valid sample questionnaires reached 91.1%. according to previous methods and experience from other experts and scholars, sample size statistics should be about 10 times of the total number of variable samples in the multi-sample size statistics and analysis of the survey. in this paper, 35 samples of final statistical indicators and 5493 valid samples were selected, which fully met the basic requirements for experience accumulation and judgment. the statistical results of the basic characteristics of survey samples were listed in table 1. the main content of the questionnaire was based on the personal and family demographic characteristics of respondents, which could be divided into four parts. the first consisted of basic demographic characteristics of respondents, including sex, age, living environment, educational level and occupation, mainly used to grasp the basic information on respondents, and to facilitate exploration of the relationship between demographic characteristics and various potential variables in the future. the second was to examine people’s capacity to adapt to global climate change risks and economic conditions. the third was the subjective perception of global climate change risks, including climate change disaster perception, exposure degree perception and vulnerability perception. the fourth was composed of the public’s willingness to deal with climate change risks, knowledge of climate change risks, channels to obtain information about climate change risks, and daily behaviors to cope with climate change risks. this paper mainly studied the information channels for people to deal with climate change risks. all questionnaire options followed likert’s five-level quantitative design (with some questions at four or seven levels). in this way, index options could be uniformly assigned, and the simplicity of questionnaire could also be ensured to reduce troubles. figure 1 | topographic map of shaanxi province. table 1 | basic characteristics of the people surveyed survey item category frequency proportion (%) educational level primary school or below 650 11.83 junior high school-bachelor 1174 21.37 high school or technical secondary school 1359 24.74 undergraduate or junior college 2124 38.67 post-graduate 186 3.39 monthly income 500 and below 2997 54.56 500–1000 1107 20.15 1001–2000 946 17.22 2001–3000 662 12.05 3001–5000 400 7.28 more than 5000 67 1.22 sex man 2631 47.90 woman 2862 52.10 occupation animal husbandry and fishery 300 5.46 production transportation work 383 6.46 service industry or business 520 9.47 government institution 580 10.56 professional skill work 473 8.61 medical staff 350 6.37 teacher 557 10.14 soldier 332 6.04 self-employed worker 390 11.47 student 630 6.01 others 978 17.80 age 15–25 1928 35.10 26–36 1384 25.20 37–47 1095 19.93 48–58 679 12.36 59–69 307 5.59 70–80 90 1.64 81–87 10 0.18 s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 39 2.3. research methods 2.3.1. calculation of climate change risk perception index g w fi i i n =å (1) where g was the perception index; wi was the weight coefficient of each factor obtained by entropy method (table 2); i = 1, 2, 3; fi was the corresponding degree of information acquisition; the value range of fn was 1–10, and n = 1, 2, 3, ..., 5493. 2.3.2. ann neural network model and cart decision tree model the ann neural network model consisted of input, hidden and output layers. the number of neurons in the input layer depended on the number of independent variables. it was believed that people’s information acquisition ways of climate change risks mainly were school education, radio and television, information from elders, friends and relatives, information from local governments, the internet and mobile phones, previous personal experience, change of direct natural experience, popular science books and scientists, community emergency drill, and public welfare activities. therefore, the test data on these 10 variables were used as ann input parameters of the constitutive model (table 3). the number of neurons in the output layer depended on the number of dependent variables, so the climate change risk perception index was chosen as an output parameter, and the number of hidden layers and neurons in each layer could be adjusted. the aim was to reduce inefficiency in the learning process and to maintain convergence rate. the cart classification tree model consisted of the root node, intermediate node and end node. the average of classified variables in final node was a predicted value [27,28]. this paper mainly used the cart decision tree model to transform black box model of the ann neural network into a white box model, revealing main information channels that influenced climate change risk perception. 3. result analysis 3.1. fitting results of ann neural network model the ann neural network model was used to fit public’s information channels of climate change risk perception in shaanxi province for many times. when the number of hidden neurons was set as 100 and 200 respectively, the fitting degree r of the model was smaller than 0.5 in both cases. when the number of hidden neurons was set as 300, r was about 0.61, with the highest accuracy of about 53.7% (figures 2 and 3), and the error rate was about 0.42 < 0.5 (figure 4), indicating that this model had a high degree table 2 | summary of weight calculation results by entropy method item information entropy information utility value weight coefficient (%) b1: are you concerned about climate change risks? 0.9900 0.0100 46.53 b2: do you think climate change risks are closely related to you? 0.9940 0.0050 26.83 b3: how do you think climate change risks will influence your region? 0.9940 0.0057 26.64 table 3 | statistical values of influencing factors and target variables in shaanxi item school education (1) radio and television (2) elders, relatives and friends (3) local governments (4) internet and mobile phones (5) previous personal experience (6) experience nature changes (7) popular science books and scientists (8) community emergency drill (9) public environmental welfare activities (10) climate change risk perception index average 4.564 4.914 4.267 4.883 4.701 4.421 4.217 4.333 3.92 3.905 2.101 the standard deviation 1.696 1.538 1.528 1.717 1.581 1.913 1.66 1.732 1.711 1.727 2.383 standard error 0.022 0.02 0.02 0.022 0.02 0.024 0.021 0.022 0.022 0.022 0.03 figure 2 | fitting results of ann neural network. 40 s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 of fitting and could better reflect the public’s climate change risk perception mechanism. this meant that when 10 pieces of climate change risk perception information worked as input, 300 neurons would be called to perceive climate change risks. there were many information channels affecting public perception of climate change risks, which needed to be analyzed in depth. however, the ann neural network could only roughly simulate information paths of climate change risk perception, but failed to reflect detailed information channels of climate change risk perception. therefore, to make this black-box model transparent, the cart decision tree model would be used next to calculate the information channel that influenced public’s climate change risk perception. 3.2. research on information channels of climate change risk perception first, taking the climate change risk perception index in formula 1 as a dependent variable and the climate change risk information acquisition mode in table 3 as an independent variable, 2874 training samples were selected and cart decision tree model was adopted for training. as shown by the clustering results, information channels that affected public’s climate change risk perception could be divided into seven categories (figure 5), and the classified auc value was 0.54 > 0.5 (figure 6), indicating that this model had a small error and could better reflect the information channels that affected public’s climate change risk perception. second, combining the clustering results of climate change risk perception information channels and the content of the questionnaire, the information channels of climate change risk perception could be summarized as follows (table 4): indirect perception–conductive perception–scientific perception; indirect perception–conductive perception–scientific perception–local government information– conductive perception; indirect perception–conductive perception– scientific perception–local government notification; indirect perception–direct perception–third party notification–conductive perception; indirect perception–direct perception; indirect perception–conductive  perception–government notification; and indirect figure 4 | error rate of ann neural network. figure 3 | fitting degree of ann neural network. figure 5 | path classification results of climate change risk perception. s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 41 perception–conductive perception. third, as shown in table 4, indirect perception–conductive perception–scientific perception; indirect perception–conductive perception–scientific perception– local government information–conductive perception; indirect perception–conductive perception–scientific perception–local government notification; indirect perception–direct perception–third party notification–conductive perception; indirect perception– direct perception; indirect perception–conductive perception– government notification; and indirect perception–conductive perception paths accounted for 5.981%, 10.420%, 1.276%, 39.022%, 10.260%, 4.625% and 28.416% respectively. this indicated that the information channel that most affected climate change risk perception was indirect perception–direct perception–third party notification–conductive perception, accounting for 39.022%; the second was indirect perception–conductive perception, taking up 28.416%; and the third was indirect perception–conductive perception– scientific perception–local government information–conductive perception, accounting for 10.420% (figure 7). this suggested that real knowledge came from practice, and that the channel of conductive perception was an important part of the process of people’s climate change risk perception. only by continuously personally experiencing or practicing climate change risk information, such as third-party notification, scientist notification, government notification could people’s climate change risk perception be enhanced. based on the above three typical information channels of climate change risk perception in shaanxi province, the public could make the following adjustments in managing climate change risks. as for the indirect perception–direct perception–third party notification– conductive perception path, we had better strengthen the accuracy of radio and television notification and scientific nature of popular science books. for indirect perception–conductive perception path, we should accumulate more practical experience and increase the number of direct experiences. for indirect perception–conductive perception–scientific perception–local government information– conductive perception path, the public could organize more risk emergency drills to improve people’s conductive perception level, and then improve people’s climate change risk perception. also, we were supposed to ensure the accuracy of people’s second notification or information acquisition, to avoid interruption of communication, thus reducing people’s perception of climate change risks. finally, it was necessary to avoid perception bias and to achieve the accuracy of terminal recovery or the intermediate information conduction by reducing the interference of third-party notification. 4. discussion the ann neural network method could be used to promptly determine the number of information channels of people’s climate figure 6 | roc curve. table 4 | classification of information channels of climate change risk perception path category of information channel 1 and 2 indirect perception–conductive perception–scientific perception (5.981%) 3 indirect perception–conductive perception–scientific perception–local government notification–conductive perception (10.420%) 4 indirect perception–conductive perception–scientific perception–local government notification (1.276%) 5 and 6 indirect perception–direct perception–third party notification–conductive perception (39.022%) 7 indirect perception–direct perception (10.260%) 8 indirect perception–conductive perception–government notification (4.625%) 9 indirect perception–conductive perception (28.416%) 42 s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 change risk perception, because it was more objective than the traditional structural equation model in reflecting brain’s processing of climate change risk information. the finding was consistent with the principle established by guoru et al. [29], who constructed a multi-layer perception model to analyze the factors that affected farmers’ climate change risk perception, when minimizing the prediction error of the target variable. on the basis of the ann neural network method, it was concluded that when the maximum fitting degree of ann neural network model was 0.53, there were 300 hidden neurons in the risk perception process of climate change in shaanxi province, which was similar to the research results obtained by related scholars. however, the ann neural network method was a black-box model, and the intermediate path was not clear. therefore, we used the cart decision tree model to explore the main information channels that influenced public’s climate change risk perception in shaanxi province. there were seven main information channels of climate change risk perception. this may be related to the brain’s self-protection mechanism that processed up to seven pieces of information to protect the body from harm, which agreed with the conclusion by relevant scholars that there were five information channels of climate change risk perception [30]. the training accuracy of the cart decision tree model reached 54%, and the indirect perception–direct perception– third party notification–conductive perception path (39.022%) had the greatest influence on perception of climate change risks. the second followed the indirect perception–conductive perception path (28.416%). finally, the indirect perception–conductive perception–scientific perception-local government notification– conductive perception path also had a major impact on climate change risk perception (10.420%). in short, the conductive perception link was at the end of the climate change risk perception information channel, which was consistent with related studies believing that the conductive perception of community emergency drills had the greatest impact on the climate change risk perception index [31]. however, the first-end of climate change risk perception information channels was the indirect perception method figure 7 | information channels of climate change risk perception. s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 43 in most cases. this showed that mass media and communication exerted a greater impact on climate change risk perception, which was similar to the research results by some scholars [32–36]. these references confirmed the above conclusions in terms of mass media information, daily climate change risk events and agricultural activities, and the public’s trust in information sources. the innovations of this paper were mainly reflected in the following three aspects. first, based on the cart decision tree model, the paper not only located the main information channels influencing climate change risk perception, but also clearly classified information channels of climate change risk perception into seven categories. second, on the one hand, we created more innovative methods than the traditional weighting method or linear regression [37]; on the other hand, we made a breakthrough in content. ofoegbu and new [38] believed that information flow and exchange through organizational collaboration networks had a limited effect on improving farmers’ knowledge about climate risks, impacts and available risk response options. ponce de leon [39] studied the effect of information on the evacuation of typhoon haiyan in the philippines, and found that according to the media and researchers, people did not have enough information about the storm, or did not understand the information given to them, and therefore did not evacuate, and that participants from different locations in the same municipality understood warning information differently. kalafatis et al. [40] provided five sources of climate change information, leading interviewees to reflect on their experiences and to gain new knowledge from them; each interviewee described a reflection system, and increased attention to these tailored reflection systems offered a path to understanding how experiential learning could most effectively enhance climate change decision support. the above-mentioned researches on the information paths of climate change risk perception have focused on the influence of information on climate change risk perception and on the theoretical level of climate change risk perception channel, but there are few researches on the internal mechanism and specific path or information channel of climate change risk perception. therefore, the study of the information channel simulation of climate change risk perception and internal mechanism in this paper was innovative to a certain extent, and could reasonably reveal the principle of climate change risk perception. in other words, most scholars have failed to comprehensively consider factors that affect climate change risk perception from the perspective of the dissemination channels of climate change risk information. and it was indicated that only through repeated practice and exercise could we enhance the climate change risk perception. third, the above conclusions had certain practical significance and application value. people could take different measures to deal with climate change risks, according to different information channels of climate change risk perception in varied regions. 5. conclusion (1) there were 300 hidden neurons in the risk perception process of climate change in shaanxi province, showing that a lot of information channels affected climate change risk perception in shaanxi province. (2) the indirect perception–direct perception–third party notification–conductive perception path (39.022%) had the greatest influence on the perception of climate change risks. the second followed the indirect perception–conductive perception path (28.416%). finally, the indirect perception–conductive perception–scientific perception–local government notification– conductive perception path also had a huge impact on climate change risk perception (10.420%). (3) we could formulate different strategies aimed at different information channels of climate change risk perception to improve people’s climate change risk perception level. as for the indirect perception–direct perception–third party notification– conductive perception path, it was necessary to strengthen the accuracy of radio and television notification and popular science books. for the indirect perception–conductive perception path, it was desirable to accumulate more direct experiences. for the indirect perception–conductive perception– scientific perception–local government notification–conductive perception path, the public should participate in risk emergency drills actively and frequently. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution swx wrote all the contents of the manuscript and paid for the retouches. qz was responsible for only searching some references of the manuscript. acknowledgments this work is supported by the national natural science foundation of china for “regional climate change risk perception and response” (41771215). the authors appreciate the time and effort spent by the editors and reviewers in providing constructive comments which have helped to improve the manuscript. references [1] yifan j. differences in people’s cognition and response to earthquake disasters in tibetan areas. qinghai province, china: qinghai normal university; 2015. [2] guiwu s, zongjin m, ruojia w, yue w, boyang d, shuwei z, et al. the characteristics of people’s cognition and response to earthquake disaster in wenchuan earthquake areas and the significance of publicity and education on disaster reduction — a case study of deyang city, sichuan province. seismogeology 2008;30:877–94. [3] yifan j, benyong w, guiwu s, haifeng z, lei s, yang w. preliminary analysis on the cognitive characteristics of earthquake disaster for government officials in yushu, qinghai. disaster sci 2015;30:229–34. [4] wei d. research on disaster risk perception of residents in guangzhou. lanzhou city, china: lanzhou university; 2014. [5] koresawa a. government’s response to the great east japan earthquake and tsunami. j disaster res 2012;7:517–27. 44 s.w. xui and q. zhou / journal of risk analysis and crisis response 11(1) 36–44 [6] yushuang w. research on the coupling relationship between public perception of climate change and adaptation behavior in northern shaanxi. northeast normal university; 2015. [7] smith j. dangerous news: media decision making about climate change risk. risk anal 2005;25:1471–82. [8] lemos mc, kirchhoff cj, ramprasad v. narrowing the climate information usability gap. nature clim change 2012;2:789–94. [9] hmielowski jd, feldman l, myers ta, leiserowitz a, maibach e. an attack on science? media use, trust in scientists, and perceptions of global warming. public underst sci 2014;23:866–83. [10] archie km, dilling l, milford jb, pampel fc. unpacking the ‘information barrier’: comparing perspectives on information as a barrier to climate change adaptation in the interior mountain west. j environ manage 2014;133:397–410. [11] lynch ah, tryhorn l, abramson r. working at the boundary: facilitating interdisciplinarity in climate change adaptation research. bull am meteorol soc 2008;89:169–79. [12] carmichael jt, brulle rj. elite cues, media coverage, and public concern: an integrated path analysis of public opinion on climate change, 2001–2013. environ polit 2017;26:232–52. [13] metag j, füchslin t, schäfer ms. global warming’s five germanys: a typology of germans’ views on climate change and patterns of media use and information. public underst sci 2017;26:434–51. [14] moser sc. communicating climate change: history, challenges, process and future directions. wires clim change 2010;1:31–53. [15] bolton re. americans and climate change: closing the gap between science and action. j plan edu res 2006;26:123–5. [16] lazrus h, morrow bh, morss re, lazo jk. vulnerability beyond stereotypes: context and agency in hurricane risk communication. weather clim soc 2012;4:103–9. [17] kahan dm, jenkins-smith h, tarantola t, silva cl, braman d. geoengineering and climate change polarization: testing a two-channel model of science communication. ann am acad polit soc sci 2015;658:192–222. [18] xie b, brewer mb, hayes bk, mcdonald ri, newell br. predicting climate change risk perception and willingness to act. j environ psychol 2019;65:101331. [19] lin gs, qi z, mingjie l, yanjun w, sharecropper s, guannan c, et al. spatio-temporal differentiation and influencing factors of new media users’ disaster cognition and response — a case study of shouguang flood in shandong. sci disasters 2020;35:210–16. [20] hepeng j. global warming, science communication and public participation — analysis of the communication of climate change technology in china. popul sci res 2007;39–45. [21] yujie l. information source, channel and content — research on climate communication strategies of chinese people based on investigation. int media 2013;35:67–83. [22] gardner mw, dorling sr. artificial neural networks (the multilayer perceptron)—a review of applications in the atmospheric sciences. atmos environ 1998;32:2627–36. [23] rosenblatt f. the perceptron: a probabilistic model for information storage and organization in the brain. psychol rev 1958;65:386–408. [24] tao l. analysis of enterprise communication activity based on cart decision tree. netw secur technol appl 2020;53–4. [25] kegang w. china meteorological disasters of shaanxi province. china: meteorological press; 2005. [26] yo-chuang y, qi z, chang-yan w. temporal and spatial variation of people’s climate change perception in northern shaanxi. j northwest univ (nat sci ed) 2011;41:134–8. [27] yuan-hui r, wen g, wei-qiang c, yue-hong m. research on economical and intensive benefits of construction land based on analytic hierarchy process — taking shangqiu city as an example. chinese agric sci bull 2016;32:87–91. [28] chunxin l, tianwei l. decision support and analysis technique for cart algorithm based on decision tree. comput sci 2004;31:44–6. [29] guoru w, xingmin s, lifan s, lu w, huanjuan l. the diversity of farmers’ perception and adaptation behaviors of climate change in dry farming areas. j shaanxi norm univ (nat sci ed) 2020;48: 116–24. [30] siwen x, qi z. research on the information channel of climate change risk perception in northern shaanxi based on cart decision tree. henan sci 2020;38:1641–9. [31] bigl b. stop the frack! exploring the media’s portrayal of the social representation of an anti-fracking protest at the baltic sea. environ commun 2020;14:271–86. [32] bo c. research on the impact of climate change media information framework on public risk perception and behavior will. nanjing city, china: nanjing university; 2019. [33] lingyun w. the impact of media sources and information balance on climate change risk perception: research based on experiments. zhejiang university; 2018. [34] fleming k, abad j, booth l, schueller l, baills a, scolobig a, et al. the use of serious games in engaging stakeholders for disaster risk reduction, management and climate change adaption information elicitation. int j disaster risk reduction 2020;49: 101669. [35] hesam m, roshan g, grab sw, shabahrami ar. comparative assessment of farmers’ perceptions on drought impacts: the case of a coastal lowland versus adjoining mountain foreland region of northern iran. theor appl climatol 2021;143:489–503. [36] vainio a, paloniemi r, varho v. weighing the risks of nuclear energy and climate change: trust in different information sources, perceived risks, and willingness to pay for alternatives to nuclear power. risk anal 2017;37:557–69. [37] xingmin s. research progress on public perception and adaptation to climate change. water soil conserv bull 2016;36:25. [38] ofoegbu c, new m. the role of farmers and organizational networks in climate information communication: the case of ghana. int j clim change strat manage 2021;13:19–34. [39] ponce de leon iz. of warnings and waiting: an examination of the path of information for two communities hit by typhoon haiyan. j risk res 2020;23:598–612. [40] kalafatis se, neosh j, libarkin jc, whyte kp, caldwell c. experiential learning processes informing climate change decision support. weather clim soc 2019;11:681–94. https://doi.org/10.1111/j.1539-6924.2005.00693.x https://doi.org/10.1111/j.1539-6924.2005.00693.x https://doi.org/10.1038/nclimate1614 https://doi.org/10.1038/nclimate1614 https://doi.org/10.1177/0963662513480091 https://doi.org/10.1177/0963662513480091 https://doi.org/10.1177/0963662513480091 https://doi.org/10.1016/j.jenvman.2013.12.015 https://doi.org/10.1016/j.jenvman.2013.12.015 https://doi.org/10.1016/j.jenvman.2013.12.015 https://doi.org/10.1016/j.jenvman.2013.12.015 https://doi.org/10.1175/bams-89-2-169 https://doi.org/10.1175/bams-89-2-169 https://doi.org/10.1175/bams-89-2-169 https://doi.org/10.1080/09644016.2016.1263433 https://doi.org/10.1080/09644016.2016.1263433 https://doi.org/10.1080/09644016.2016.1263433 https://doi.org/10.1177/0963662515592558 https://doi.org/10.1177/0963662515592558 https://doi.org/10.1177/0963662515592558 https://doi.org/10.1002/wcc.11 https://doi.org/10.1002/wcc.11 https://doi.org/10.1002/wcc.11 https://doi.org/10.1177/0739456x06292375 https://doi.org/10.1177/0739456x06292375 https://doi.org/10.1175/wcas-d-12-00015.1 https://doi.org/10.1175/wcas-d-12-00015.1 https://doi.org/10.1175/wcas-d-12-00015.1 https://doi.org/10.1177/0002716214559002 https://doi.org/10.1177/0002716214559002 https://doi.org/10.1177/0002716214559002 https://doi.org/10.1177/0002716214559002 https://doi.org/10.1016/j.jenvp.2019.101331 https://doi.org/10.1016/j.jenvp.2019.101331 https://doi.org/10.1016/j.jenvp.2019.101331 https://doi.org/10.1016/s1352-2310(97)00447-0 https://doi.org/10.1016/s1352-2310(97)00447-0 https://doi.org/10.1016/s1352-2310(97)00447-0 https://doi.org/10.1037/h0042519 https://doi.org/10.1037/h0042519 https://doi.org/10.1037/h0042519 https://doi.org/10.1080/17524032.2019.1651367 https://doi.org/10.1080/17524032.2019.1651367 https://doi.org/10.1080/17524032.2019.1651367 https://doi.org/10.1016/j.ijdrr.2020.101669 https://doi.org/10.1016/j.ijdrr.2020.101669 https://doi.org/10.1016/j.ijdrr.2020.101669 https://doi.org/10.1016/j.ijdrr.2020.101669 https://doi.org/10.1016/j.ijdrr.2020.101669 https://doi.org/10.1007/s00704-020-03432-9 https://doi.org/10.1007/s00704-020-03432-9 https://doi.org/10.1007/s00704-020-03432-9 https://doi.org/10.1007/s00704-020-03432-9 https://doi.org/10.1111/risa.12640 https://doi.org/10.1111/risa.12640 https://doi.org/10.1111/risa.12640 https://doi.org/10.1111/risa.12640 https://doi.org/10.1108/ijccsm-04-2020-0030 https://doi.org/10.1108/ijccsm-04-2020-0030 https://doi.org/10.1108/ijccsm-04-2020-0030 https://doi.org/10.1080/13669877.2019.1592212 https://doi.org/10.1080/13669877.2019.1592212 https://doi.org/10.1080/13669877.2019.1592212 https://doi.org/10.1175/wcas-d-19-0002.1 https://doi.org/10.1175/wcas-d-19-0002.1 https://doi.org/10.1175/wcas-d-19-0002.1 journal of risk analysis and crisis response vol. 9(4); january (2020), pp. 163–167 doi: https://doi.org/10.2991/jracr.k.200117.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article the harmonious development of big data industry and financial agglomeration in guizhou junmeng lu1,2,*, mu zhang1 1school of big data application and economics, guizhou university of finance and economics, guiyang, huaxi, china 2guizhou institution for technology innovation and entrepreneurship investment, guizhou university of finance and economics, guiyang, huaxi, china 1. introduction the 18th national congress of the communist party of china clearly pointed out that scientific and technological innovation should be taken as the core of development and innovation-driven development strategy should be adopted to facilitate further economic growth, so as to provide a sustained driving force for the improvement of china’s comprehensive national strength. as a strategic emerging industry, big data industry, with its innovative technology content, can well meet the needs of various enterprises and government agencies for data resources, meet the needs of various industries to follow the trend of internet development, and make positive contributions to economic construction while the industry itself is developing continuously. at present, the big data industry in china presents a completely different development path from the traditional industry. china’s big data industry is mainly concentrated in beijing, shanghai, guangdong and other places with relatively developed economy, but the place with the most rapid development of big data industry is guizhou, where the economic development and industrial foundation are relatively backward. this phenomenon indicates that compared with general industries, big data industry may have unique development influencing factors. domestic and foreign research achievements on the relationship between financial agglomeration and industrial development are relatively abundant. galbis [1] proposed that the inhibited financial system would easily lead to the imbalance of industrial development in the region, and the constantly optimized financial structure would obviously promote the industrial development in the region and improve the efficiency of resource allocation. carlin and mayer [2] studied the impact of various financial sub-industries on various industries from the perspective of financial structure, and found that market-oriented financial structure is conducive to the development of high-tech and high-risk industries. sun and li [3] used moran index and spatial geographic weighted regression model to study and believe that the optimization of domestic industrial structure is conducive to the agglomeration development of the financial industry, and the financial agglomeration, in turn, has some promoting effect on the industrial structure. deng and liu [4] studied the relationship between financial agglomeration and industrial structure upgrading in the eastern, central and western regions of china, and found that regional financial agglomeration had a significant positive impact on industrial structure upgrading. through the analysis of the mechanism of action among financial development, industrial structure and economic development, su and xu [5] believe that regional financial agglomeration and financial development can support and promote the upgrading of regional industrial structure. most scholars at home and abroad believe that financial agglomeration is conducive to optimizing industrial structure and promoting industrial development, so as to promote economic development in the region. however, the rise of big data industry is still a short period, and most of the researches are still at the level of major problems, development trend, competitive advantage and composition and policy system research that big data industry faces [6–10]. there is no systematic and authoritative theory and analysis on the factors influencing the development of big data industry, let alone theoretical and empirical research on the interaction and support between big data industry and financial agglomeration. based on this, this study was to explore big data industry and financial agglomeration a r t i c l e i n f o article history received 31 january 2019 accepted 17 october 2019 keywords big data industry financial agglomeration ifahp coupling model coupled coordination model a b s t r a c t it has important practical and theoretical significance to study the coupling relationship and coordinated development between big data industry and financial agglomeration. this paper used 2015 cross-section data, the intuitionistic fuzzy analytic hierarchy process, the intuitionistic fuzzy number score function, the coupling model and the coupling coordination model to empirically research the coupling and coordination level between guizhou big data industry and financial agglomeration. the empirical research shows that there is an obvious imbalance in the coordinated development and obvious spatial heterogeneity of big data industry and financial agglomeration in guizhou. only guiyang and zunyi can achieve the coupling and coordinated development of big data industry and financial agglomeration. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 463232121@qq.com; rim_007@163.com https://doi.org/10.2991/jracr.k.200117.001 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:463232121%40qq.com?subject= mailto:rim_007%40163.com?subject= 164 j. lu and m. zhang / journal of risk analysis and crisis response 9(4) 163–167 coupling between coordination mechanism, taking guizhou province as a sample, using the intuitionistic fuzzy analytic hierarchy process (ifahp) and the coupling coordination model to empirically analyze the development level and coupling coordination level of big data industry and financial agglomeration. it is expected to put forward corresponding countermeasures and suggestions for the development of regional big data industry, which has certain practical and theoretical significance. 2. analysis of interaction between big data industry and financial agglomeration according to the theory of manufacturers, labour factor, capital factor and technology factor are the main driving forces to promote industrial development, and the development of emerging industries has a huge demand for the three factors, especially the capital factor. referring to china’s big data industrial agglomeration areas, most of which are economically developed areas, and big data industry, as an emerging industry, its capital demand determines the development level of big data industry to a large extent. therefore, it is speculated that the agglomeration of financial industry in a certain region can promote the development of local big data industry to a certain extent. the role of financial agglomeration in promoting the development of big data is mainly reflected in two aspects. on the one hand, as an emerging industry, big data industry is different from the traditional labour-intensive industry, its development requires a lot of capital-intensive research and development activities, which means that sufficient capital investment can effectively drive the structural adjustment and upgrading of big data industry and promote the development of big data industry. on the other hand, financial agglomeration can promote the benign development of big data industry. specifically, the capital allocation function and screening mechanism of the financial market can always make investment funds flow to enterprises with growth space and development potential, so as to improve the use efficiency of investment funds, improve the factor productivity of big data industry and bring economic benefits, and finally promote the stable and healthy development of big data industry. 3. research method 3.1. index system and data source based on the principles of scientificity, representativeness, comparability and accessibility, and references the results of bai et al. [11], this study constructs an evaluation system for the coupling and coordinated development of big data industry–financial agglomeration system according to the mechanism and characteristics of the coupling and coordinated development of big data industry and financial agglomeration. the development level of financial agglomeration is described from the two dimensions of industry agglomeration level and industry development level of financial industry, banking industry, insurance industry and securities industry, including four first-level indicators and 13 second-level indicators. the industrial development level of big data in guizhou is measured from two aspects of industrial scale and infrastructure level, including two first-level indicators and 12 second-level indicators. in this study, cross-sectional data in 2015 were selected from statistical yearbooks of guizhou prefectural cities, statistical yearbooks of guizhou province, china banking and insurance regulatory commission, china securities regulatory commission, statistics bureau of guizhou province, statistics bureau of guizhou cities and official websites of governments. 3.2. comprehensive evaluation steps based on ifahp step 1: data standardization. indicators of the evaluation system for the coupling and coordinated development of the big data industry-financial agglomeration system are all efficiency indicators. the standardized formula is: z x x x x zij ij j j j ij= − − ∈ ( ) ( ) [ ], , , min max min 0 1 xij is the original index value, max xj and max xj is the maximum and minimum value of the original index. step 2: construct the intuitive judgment matrix. by comparing the importance of indexes, the qualitative evaluation words of indicators are converted into intuitionistic fuzzy numbers, and the intuitionistic judgment matrix is constructed r = (rij)n × n. among them, rij = (mij, vij)(i,j = 1, 2, …, n), mij represent the membership degree, vij represent the non-membership degree, mij ∈ [0, 1], vij ∈ [0, 1], mij = vji, vij = mji, mii = vii = 0.5, mij + vij ≤ 1. the pij = 1 − mij − vij is the hesitation degree. step 3: consistency test. the consistency of intuitionistic judgment matrix is tested by constructing consistency intuitionistic judgment matrix. if it passes the consistency test, skip to step 5. if not, proceed to step 4. step 4: revise the intuitive judgment matrix which does not satisfy the consistency test. set the parameter s for iteration, s ∈ [0, 1], until it passes the consistency test. step 5: calculate indicators weight. by using the operator of intuitionistic fuzzy number, the total weight of the combined second-level indicators and first-level indicators is obtained. step 6: the intuitionistic fuzzy number is obtained by combining index data with index weight. step 7: using the score function: u vf f f f= + − −( )m m m1 to convert the intuitionism fuzzy number f into the comparable real number u, and the comprehensive evaluation value of the development level of big data industry and financial agglomeration is obtained. the larger u is, the higher comprehensive evaluation value is. 3.3. coupling degree model and coupled coordination model the coupling coordination model includes coupling degree and coupling coordination degree to reflect the degree of interaction j. lu and m. zhang / journal of risk analysis and crisis response 9(4) 163–167 165 table 1 | the coupling coordination development level of big data industry and financial agglomeration districts u1 u2 c t d coupling coordination level guiyang 0.6274 0.6442 0.9998 0.6413 0.8007 good coordinated development zunyi 0.5487 0.3514 0.9863 0.4709 0.6815 primary coordinated development southern guizhou 0.1268 0.0849 0.9849 0.1081 0.3263 mild disorder recession southeast guizhou 0.1267 0.0446 0.9453 0.0956 0.3006 mild disorder recession tongren 0.1096 0.0589 0.9423 0.0821 0.2781 intermediate disorder recession ansun 0.0436 0.0903 0.9012 0.0769 0.2633 intermediate disorder recession southwest guizhou 0.0953 0.0688 0.9186 0.0683 0.2504 intermediate disorder recession bijie 0.2062 0.0188 0.5435 0.1121 0.2468 intermediate disorder recession liupanshui 0.0660 0.0519 0.9938 0.0594 0.2430 intermediate disorder recession and interaction among multiple systems. the coupling degree model of big data industry-financial agglomeration system is: c u u u u = × +( ) 2 1 2 1 2 1 2( ) . / c is the coupling degree and the value range is [0, 1]. when c = 0, it indicates that the two systems are mutually independent. when 0 < c ≤ 0.3, the system is in the stage of low-level coupling. when 0.3 < c ≤ 0.5, the system is in the antagonism stage. when 0.5 < c ≤ 0.8, the system is in the run-in stage; when 0.8 < c ≤ 1, the system is in the stage of high horizontal coupling. u1 and u2 are the comprehensive evaluation values of big data industry and financial agglomeration. if the two achieve coupling and coordinated development, when u1 − u2 > 0.1, it is the lag type of financial agglomeration lag. when u2 − u1 > 0.1, it is the lag type of big data industry. when 0 ≤ |u1 − u2| ≤ 0.1, it is the synchronous type of financial agglomeration and big data industry. if the coupling between the big data industry and financial agglomeration is in a state of misalignment and recession, when u1 − u2 > 0.1, it is the damage type of financial agglomeration. when u2 − u1 > 0.1, it is the damage type of big data industry. when 0 ≤ |u1 − u2| ≤ 0.1, it is the damage type of both financial agglomeration and big data industry. when the comprehensive evaluation values of the two systems are similar but not high, only calculating the coupling degree will lead to the pseudo-evaluation results with high degree of collaborative development. therefore, in order to accurately reflect the coupling and coordinated development degree of big data industry and financial agglomeration, the coupling and coordination degree model is further constructed: d c t t u u= ´ = +( ) , ./1 2 1 2 a b d is the degree of coupling coordination and the value range is [0, 1], t is the comprehensive evaluation value of big data industry– financial agglomeration system, and a and b are undetermined coefficients (a + b = 1), reflecting the contribution of big data industry and financial agglomeration to the coupling coordination of the overall system. considering that both are equally important to the overall system, a = b = 0.5 is selected. when 0.9 < d ≤ 1, the degree of coupling coordination between big data industry and financial agglomeration is in high quality and coordinated development. when 0.8 < d ≤ 0.9, it is a good coordinated development. when 0.7 < d ≤ 0.8, it belongs to intermediate coordinated development. when 0.6 < d ≤ 0.7, it belongs to primary coordinated development. when 0.5 < d ≤ 0.6, it is barely coordinated development. when 0.4 < d ≤ 0.5, it is an endangered recession. when 0.3 < d ≤ 0.4, it is a mild disorder recession. when 0.2 < d ≤ 0.3, it belongs to intermediate disorder recession. when 0.1 < d ≤ 0.2, it is a severe disorder recession. when 0 < d ≤ 0.1, it belongs to extreme disorder recession. 4. empirical results and analysis 4.1. empirical research based on ifahp the empirical research results are shown in table 1. according to the empirical research results based on ifahp, the comprehensive evaluation value of big data industry in guiyang is 0.6724, with the highest score, and the lowest score is from anshun, which is 0.0436. the value of zunyi, southern guizhou, southeast guizhou, tongren, southwest guizhou, bijie and liupanshui is 0.5487, 0.1268, 0.1267, 0.1096, 0.0953, 0.2062 and 0.0660 respectively. the comprehensive evaluation value of financial agglomeration of guiyang is 0.6442, followed by zunyi, which the value is 0.3514. the value of southern guizhou, southeast guizhou, tongren, ansun, southwest guizhou, bijie and liupanshui is 0.0849, 0.0446, 0.0589, 0.0903, 0.0688, 0.0188 and 0.0519 respectively. the comprehensive valuation of big data industry and financial agglomeration of nine districts in guizhou are large fluctuations. only guiyang and zunyi achieves that the comprehensive evaluation of big data industry development level and financial agglomeration level is more than average, across the rest of the city are below average. it shows that big data industry development and the financial industry agglomeration are very unequal status in guizhou. 4.2. empirical research based on coupling coordination model the coupling degree c of the two systems of the other eight districts is relatively stable, fluctuating within the range of 0.9012–0.9998, except for bijie which is 0.5435. the gap between different districts is not obvious, and they are all in the high level coupling stage, indicating that there is a strong correlation between big data industry development and financial agglomeration in guizhou province. however, in terms of coupling and coordinated development, only guiyang and zunyi have a coupling coordination degree d higher than the average level, which is 0.8007 and 0.6815 and within the acceptable range of the coupling and coordination criteria, while the other seven districts are all in the unacceptable range, indicating that there is a coupling coordination development imbalance 166 j. lu and m. zhang / journal of risk analysis and crisis response 9(4) 163–167 between big data industry development and financial agglomeration in guizhou province. (1) guiyang belongs to the high level stage of coupling and good coordination development type of big data industry and financial agglomeration. the comprehensive evaluation of big data industry, the comprehensive evaluation of financial agglomeration and coupling coordination degree of both are much higher than the rest of districts. moreover, there is little difference between the development level of big data industry and financial agglomeration, so it belongs to the synchronous development type of big data industry and financial agglomeration. this indicates that in the process of big data industrial development, the financial industry agglomeration of guiyang can offer good support for its development, and the big data industry at the same time can promote the development of regional financial industry agglomeration, the coordinated development of both promote each other to form a good situation. (2) zunyi belongs to the high-level coupling stage and primary coordinated development type of big data industry and financial agglomeration. its comprehensive evaluation value of big data industry, comprehensive evaluation value of financial agglomeration and their degree of coupling and coordination are only second to the provincial capital guiyang. moreover, the degree of coupling coordination is close to the intermediate coordinated development, indicating that the big data industry and the financial industry in zunyi can achieve the coordinated development of the two. but the big data industry value is greater than the financial agglomeration, the integrated evaluation and the difference between the two is >0.1, so it belongs to the financial agglomeration lag type. this phenomenon shows that financing needs of some big data enterprises may not being met, which limit the development of big data industry in the area. (3) southern guizhou and southeast guizhou belong to big data industry and the financial agglomeration high coupling phase and mild disorder recession types, which d value is 0.3263 and 0.3006. big data industry and the financial agglomeration development level difference is not big, belong to the big data industry and financial agglomeration joint damage type. although the coupling level is high, there are many big data enterprises in southern guizhou and southeast guizhou without corresponding financial support. at the same time, the limited development of big data industry itself may also hinder the agglomeration of financial industry in the region, making the big data industry and financial agglomeration in southern guizhou and southeast guizhou restrict each other and fail to achieve joint and coordinated development. (4) tongren, anshun, southwest guizhou, bijie and liupanshui belong to the intermediate disorder recession type of big data industry and financial agglomeration, which d value is 0.2781, 0.2633, 0.2504, 0.2468 and 0.243. the coupling of big data industry and financial agglomeration in bijie is in the running-in stage, and there is a big gap in the comprehensive evaluation value between the two, which belongs to the type of financial agglomeration lag. the development level of big data industry and financial agglomeration in the other four districts is approach, they are in the stage of high level coupling and the type of common damage of big data industry and financial agglomeration. showed that it is not only the majority of big data business financial needs are not being met in tongren, anshun, southwest guizhou, bijie and lliupanshui, result to the development of big data industry in the region is greatly restricted, and the low level of development of big data industry may also cause the outward dispersion of regional financial industry. the development level of big data industry and financial agglomeration level in five districts both needs to be greatly increased. 5. conclusion and suggestions 5.1. conclusion based on the coupling coordination mechanism of big data industry and financial agglomeration, this paper established the coupling coordination development evaluation index system of big data industry and financial agglomeration. then using the ifhap to calculate the big data industry and the financial agglomeration evaluation index weight and total weight at all levels. finally, it combined with the coupling coordination model and 2015 cross-section data of each districts of guizhou to measure the coupling coordination degree between big data industry and financial agglomeration. by analyze the coupling coordination development condition, we can conclude that: (1) from the perspective of overall, the coupling coordination development level of big data industry and financial agglomeration in guizhou is low and span to five levels, and the regional difference is more apparent. it has an obvious spatial heterogeneity. (2) from the perspective of space, in terms of coupling stage, except for bijie, the other eight districts in guizhou province all achieve the high-level coupling of big data industry and financial agglomeration. the coupling of big data industry and financial agglomeration in bijie is in the runningin stage. but in terms of the coordinated development stage, only guiyang and zunyi can achieve coupling coordinated development of big data industry and financial agglomeration. the condition of southern guizhou and southeast guizhou are in the mild disorder recession of big data industry and financial agglomeration, while other five districts and regions, tongren, anshun, southwest guizhou, bijie and liupanshui, are all in the intermediate disorder recession. 5.2. suggestions after several years of development, big data industry in guizhou has a certain scale. in the process of construction and development in the future, it still needs further improvement. on the one hand, guizhou should draw lessons from foreign advanced experience of the development of big data industry. at the same time, digestion and absorption of the successful model of guiyang big data exchange. guizhou should vigorously introducing big data industry demonstration enterprise, while should also actively build infrastructure of big data industry. for laying a solid foundation for the development of big data industry, guizhou should try best to build up comprehensive national big data (guizhou) pilot zone, and developing a path for big data industry with guizhou characteristics. j. lu and m. zhang / journal of risk analysis and crisis response 9(4) 163–167 167 on the other hand, with the increasing integration of big data technology and traditional industries, guizhou should take advantage of the development momentum of big data industry and use big data to meet the development needs of government departments, social institutions, enterprises and individuals, which can help to achieve the economic development transformation of various industries in the province and improve the quality and efficiency of economic growth. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments this research was financially supported by the regional project of national natural science foundation of china (71861003) and the second batch projects of basic research program (soft science category) in guizhou province in 2017 (foundation of guizhouscience cooperation [2017] 1516-1). references [1] galbis v. financial intermediation and economic growth in less‐developed countries: a theoretical approach. j develop stud 1977;13:58–72. [2] carlin w, mayer c. finance, investment, and growth. j finan econ 2003;69:191–226. [3] sun j, li h. financial agglomeration and industrial structure upgrading: an empirical analysis of inter-provincial economic data from 2003 to 2007. economist 2012;3:80–6. [4] deng x, liu w. empirical analysis of the role of financial agglomeration in upgrading industrial structure. j nanjing soc sci 2013;10:5–12. [5] sun j, xu z. financial development, industrial structure upgrading and economic growth theoretical and empirical research. j ind technol econ 2014;33:139–49. [6] zhou y, liu y. research on influencing factors of big data industry development. j mod intell 2017;37:129–34. [7] lemen c, guanghui m, haijuan l. analysis of the characteristics of china’s big data industry and policy recommendations. j inform theory pract 2016;39:5–10. [8] li w, liu g, lu z. research on technology competition situation of china’s big data industry based on patent analysis. j inform 2015;34:65–70. [9] hu j, ding z, ren y. research on competitive advantage of china’s big data industry based on the revised diamond model. j ind technol econ 2015;34:17–27. [10] chen b. research on the complementary development of big data industry differences under the beijing-tianjin-hebei collaborative framework. j econ manage 2015;29:27–30. [11] bai c, huang y, song w. study on the coupling and coordinated development of financial agglomeration and ecological efficiency in provincial regions. j resour environ arid areas 2014;28:1–7. microsoft word research on model for evaluating risks of venture capital projects research on model for evaluating risks of venture capital projects yunfei li1,2 zongfang zhou1 1school of management and economics (university of electronic science and technology of china) chengdu 610054, china, liyunfei80@126.com ,zhouzf@uestc.edu.cn 2school of mathematics and information(china west normal university) nanchong 637002, china, abstract an index system for evaluating the risks of venture capital projects is established on the basis of analyzing the factors affecting the risks of venture capital projects. a model for evaluating the risk degree of single venture capital project is presented by applying the uncertain type analysis of hierarchy process and the ewaa operator based on the multiple attribute decision making; another model for evaluating the ranks of the risks of multiple venture capital projects is established based on the interval type ideal point for the uncertain multiple attribute decision making. two examples about practical application are given to show the reasonableness and effectiveness of the models. keywords: venture capital, risk evaluation, interval number, multiple attribute decision making 基于风险投资项目的风险评价模型研究 李云飞 1,2 周宗放 1 1. 电子科技大学经济与管理学院，成都 610054 2. 西华师范大学数学与信息学院，南充 637002 摘要：在对风险投资项目的风险因素进行分析的基础上，首先建立了风险投资项目风险评价指标体系， 然后针对单个风险投资项目，综合运用不确定层次分析法（utahp）和基于 ewaa 算子的多属性决策方法， 构建了一类评价模型用于其风险等级的确定；针对多个风险投资项目，构建了一类基于区间型理想点的不确定 多属性决策模型，并应用于实现多个风险投资项目的风险排序。 后给出算例说明了这两个模型的具体应用。 关键词：风险投资，风险评价，区间数，多属性决策方法 1. 引言 风险投资，又称创业投资，是指向主要属于科 技型的高成长性创业企业（风险项目）提供股权资 本，并以一定的方式参与所投资创业企业或项目的 管理，为其提供经营管理和咨询服务，以期在被投 资企业（风险项目）发展成熟后，通过股权转让获 取中长期资本增值收益的投资行为[1]。风险投资的 性质决定了其主要投资对象是具有高度不确定性的 中小型高新技术企业或项目，具有高收益、高风险 的特点[2]。风险投资运作的核心是风险投资机构， 风险投资机构是从事风险投资具体运作的组织，是 连接风险投资者和风险企业家的金融中介。一般而 言，风险投资者（机构投资者和个人投资者）以集 体投资的方式成立风险投资机构（基金），委托风 险投资机构的风险投资家负责风险投资基金的日常 运作与管理。风险投资家在投入资金之前，必须利 用他们长期积累的经验、知识和信息网络对待选的 风险投资项目的风险构成和大小进行分析评价。风 险项目的风险评价结果不仅是风险投资家选择风险 项目的重要依据，而且直接影响到风险项目运作过 程中的控制权、现金流权的分配，进而影响到风险 投资契约的制定。因此，风险投资项目的风险评价 工作是风险投资家控制与管理风险的前提，直接关 系到风险投资的成败[3]，研究风险投资项目的风险 评价问题具有重要的现实意义。 有关风险投资项目的风险评价问题，许多学者 进行了研究，并取得了一些有价值的研究成果[2-10]。 这些成果的研究思路为：通过对投资项目风险因素 分析的基础上，构建风险评价指标体系，然后建立 评价模型进行风险评价。但是其中有些研究所采用 的方法存在不足之处：（1）在确定评价指标权重的 过程中，有些学者利用传统的层次分析法（ahp） 对评价指标间的相对重要性程度做判断。然而，风 险投资项目大多涉及高新技术，其技术、市场、产 品等都不成熟，具有很大的不确定性，风险投资家 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 142-148 published by atlantis press copyright: the authors 142 由于自身条件的限制或者所掌握的信息不足而对项 目风险因素认识不清，无法用一个确定的数值表示 风险评估指标两两比较的相对重要性程度。因此， 虽然层次分析法作为一种对非定量事件做定量分析 的实用决策方法，能有效解决复杂的决策问题，但 是用一个确定的数表示风险投资项目风险评估指标 间的相对重要性程度，有失妥当。（2）有些学者采 用模糊综合评估法作为风险评估方法，但是作为状 态集函数的模糊隶属度不满足“归一性条件”和 “可加性原则”，导致评估结果的可信度不高[11]。 另外，模糊集的“取大“、“取小“运算也会损失 许多有用的信息，造成评估结果不合理。 针对已有研究的不足和风险投资项目高度不确 定性等特点，结合风险投资运作过程中的实际情 况，本文从风险投资家的角度出发，构建风险评价 指标体系，通过建立两个模型分别解决单个风险投 资项目的风险等级确定问题和多个风险投资项目的 风险排序问题。 2. 评价指标体系的构建 在借鉴国内外风险投资项目风险评价指标的基 础上，根据全面性、系统性、可比性、科学性与可 行性相结合的原则[2,12]，本文考虑将以下几个风险因 素作为风险投资项目风险评价指标： （1）技术风险 1u ：按新技术生命周期理论 [13]， 风险项目由于新思想和新技术本身的先天不足，技 术不成熟、不完善以及可替代的新技术出现等多种 因素面临的风险。 （2）企业家风险 2u ：指由于委托代理关系引发 的风险企业家逆向选择风险、道德风险、风险投资 家与风险企业家目标差异风险等。 （3）管理风险 3u ：指因管理不善而导致项目 失败的风险。主要是指由于组织制度和结构的科学 性、经营管理方式、管理层能力和经验、约束激励 机制以及人才管理水平等引发的风险。 （4）生产风险 4u ：指与知识产权、生产人员结 构、生产设备水平以及能源原材料供应等相关的风 险。 （5）市场风险 5u ：指新产品、新技术市场竞 争力的不确定性引发的风险。比如市场需求程度、 产品可替代性、产品销售网络、产品生命周期等相 关风险。 （6）投资分析风险 6u ：指与投资工具的选 择、投资规模大小、投资决策方法的科学性、企业 资本结构等方面相关的风险。 3. 单个项目的风险等级评价模型 当风险投资机构面临一个风险投资项目时，可 以组织多个风险投资家分析风险企业家提交的商业 计划书，根据该项目的具体情况（如所处的行业） 对其风险进行评价，首先确定该项目的风险评价指 标权重。针对风险投资项目具有较大不确定性的特 点，本文采用区间数对评价指标间的相对重要性程 度进行描述，然后利用互补判断矩阵排序向量确定 指标权重。 3.1 评价指标权重的确定 记 [ , ] { , , }a a a x a x a a a r        ≤ ≤ ，称 a 为一个区间数[14]。 定义 1[14] 设 [ , ] , [ , ],a a a b b b     且记 , , a b l a a l b b       则 min{ , max( , 0)} ( ) a b a b l l a b p a b l l        ≥ (1) 为 a b ≥ 的可能度，且记 a b, 的次序关系为 p a b≥ 。 正如前面所述，由于风险投资家难以把握影响 风险投资风险的各因素的准确状态，从而会产生主 观上的不确定性。因此，用传统的层次分析法构造 判断矩阵对评估指标间的相对重要程度进行分析显 然不合理。而不确定层次分析法（utahp）不需要 风险投资家对各指标的相对重要程度做出精确的判 断，只需要给出某一估计区间即可。由此，本文采 用区间数转换判断矩阵中的元素。 首先，根据各风险评估指标间的相互重要性， 构造评估指标的区间数判断矩阵为 6 6( )ija a  ，其 中 ija  [ , ]ij ija a   { 1 9 ij ijt a t a   ≤ ≤ ≤ ≤ 9} 为区间数， 表示指标 iu 对 ju 的相对重要程度，且满足 1 , ij ji a a  1ij jia a   ( , 1, 2, , 6)i j   。具体方法如 下： 假设有 n 位风险投资家参加评估，任取两个指 标 iu , ju ，让每个风险投资家独立给出 iu 与 ju 相对 重要程度的区间数。设第 h 位风险投资家给出的区 间 数 为 ： ( ) ( ) ( )[ , ]h h hij ij ija a a   ， ( 1, 2, , )h n  。 则 取 ( ) ( ) 1 1 , n n h h ij h ij ij h ij h h a a a a          (1≤ i ≤ )j ≤ 6 published by atlantis press copyright: the authors 143 其中 1 ( 1, n h h h     0 1)h  为第 h 位风险投资家 的 重 要 性 程 度 ， 由 此 得 到 区 间 数 判 断 矩 阵 6 6( )ija a  。 其次，设 6 6 6 6( ) , ( ) ,ij ija a a a        则 [ , ]a a a   利用某一排序方法（如方根法）对矩阵 ,a a  求得 权 重 向 量 分 别 为 ： 1 2 6( , , , )w w w w      ， 1 2 6( , , , )w w w w      ，取 6 6 6 6 1 1 1 1 1 1 , j j ij ij i i k m a a          (2) 得到区间数权重向量： 1 2 6( , , , ) [ , ]w w w w kw mw    (3) 其中 [ , ] ( 1, 2, ,6)i i iw kw mw i     (4) 若 0 1k m ≤ ≤ ，则表明区间数判断矩阵 a 满足一 致性条件，否则需要反馈给风险投资家重新判断。 后，利用定义 1，对所有区间数权重 [ , ] ( 1, 2, ,6)i i iw kw mw i     两两进行比较，建立可 能度矩阵 6 6( )ijp p  ，其中 ( )ij i jp p w w ≥ 。由可能 度的性质可知，可能度矩阵 p 是一个互补判断矩 阵，利用互补判断矩阵排序向量计算公式[14]，可求 出各评估指标的权重为： 1 1 1 ( 1, 2, , ) ( 1) 2 q i ij j q w p i q q q               （在本文中 6q  ） (5) 3.2 基于ewaa算子的风险等级评价 由于风险投资是一种创新投资，项目缺乏历史 数据，因此，专家往往利用语言形式对项目的风险 评估指标进行评估。现将各评估指标按照其风险状 况划分为 9 个等级，相应的语言标度集为 s  { 4s =极低、 3s =很低、 2s =低、 1s =较 低、 0s =一般、 1s =较高、 2s =高、 3s =很高、 4s =极高}。 为了便于计算和避免丢失决策信息, 在原有语言 标度 { 4, 4}s s     的基础上定义一个拓展 语言标度 { [ , ]}s s l l    ，其中 l 是一个充分 大的自然数， 1 2 1 , ( , , ) q i i q i w w w w w      是 语言数据 2, , )( 1, i qs i   的加权向量。 定义 2[14] 假设 ewaa: s s ,令 1 21 2 1 2 ( , , , ) qw ewaa s s s w s w s        qn w s s   (6) 称函数 ewaa 为扩展的加权算术平均算子。 假设第 h 位风险投资家对该项目的第 i 个风险评 价指标的语言评估值为 ( 1, 2, , 6, 1, 2, , ) i hs i h n    ， 利用(6)式对以上指标评估信息进行集结可得第 h 位 风险投资家对该项目的风险语言评估值为： 1 2 61 2 6h h h hs w s w s w s       (7) 其中 6 1 h i i i w     。 进一步利用(6)式对这 n 位风险投资家的项目风 险评估信息进行集结，得到该风险投资项目的风险 语言评估值为 1 21 2 nnbs s s s       (8) 其中 1 ( 4 4, 1, n h h h b u b v v u       ≤ ≤ ≤ ≤ , )u v z 。 借鉴周光明的研究[15]，设 b 为服从区间[ , ]u v 上均匀分布的随机变量的取值，则可得到该项目风 险等级的评估准则： （1）若 0 0.5,b u≤ ≤ 则该项目的风险在置信度 (1 )%u b  下的风险等级为 us ； （2）若 0 0.5,v b≤ ≤ 则该项目的风险在置信度 (1 )%b v  下的风险等级为 vs 。 从以上评估准则可以看出，本文构建的模型与 已有研究的不同之处在于：在一定置信度下确定风 险投资项目的风险等级，避免了评估结果的绝对 化，从而更加符合风险投资项目风险评估的实际。 4. 多个项目的风险排序模型 从第 3 节的风险评价模型可知，针对单个风险 投资项目的风险评价方法 终能够在一定置信度下 published by atlantis press copyright: the authors 144 确定其风险等级。然而风险投资家在风险投资的实 际运作中，经常需要对多个风险投资项目的风险大 小进行比较，从而根据风险程度进行项目选择。一 个很自然的想法是，按照第 3 节的方法，对每个项 目的风险进行评价，然后比较其风险大小。但是， 这样的方法可能面临一个现实的问题：单项目的风 险评价方法没有得到项目风险的具体评分值，只是 确定了其风险等级，因此，如果有若干个项目的风 险等级评价结果相同，那么该如何选择项目呢？为 了解决这个现实问题，有必要对多个风险投资项目 的风险评价方法进行研究。 4.1 不确定多属性决策模型 由于只是解决多个风险投资项目的风险排序问 题和为了计算的方便，现将风险投资项目评价指标 的风险划分为五个等级：低风险 1v ，较低风险 2v ， 一般风险 3v ，较高风险 4v ，高风险 5v 。以上五个 评价等级元素构成等级集合： 1 2 3 4 5{ , , , , }v v v v v v ， 按照 5 分制打分，五个风险等级对应的值分别赋予 为 1、2、3、4、5。若指标等级介于两相邻等级之 间，则相应评分值为 1.5、2.5、3.5、4.5。 下面建立一个基于区间型理想点的不确定多属 性决策模型用于解决多个风险投资项目的风险排序 问题。 定义 3[14] 设区间数 [ , ], [ , ],a a a b b b     令 范数 2 2( ) ( )a b b a b a        ，称 ( , )d a b a b    为区间数 ,a b 的相离度。 显然 ( , )d a b 越大，则区间数 ,a b 相离的程度 就越大。特别地，当 ( , ) 0d a b  时，有 a b  ，即 区间数 ,a b 相等。利用定义 3 可以衡量两个区间数 相似的程度。 现假设有 s 个风险投资项目 1 2, , , sx x x 可供风 险投资家选择，评价指标为 1 2 6, , , ,u u u 风险投资 家对这些项目的风险评价指标按照5分制打分。如前 所述，风险投资家所掌握的信息不足以把握评价指 标的真实状态，以区间数的形式给出指标判断值， 从而得到项目决策矩阵为 6( )ij sa a   ，其中 [ , ] { 1 5}ij ij ij ij ija a a t a t a      ≤ ≤ ≤ ≤ 表示第 i 个项目的第 j 个指标风险等级区间数 ( 1, 2, , , 1, 2, , 6)i s j   。 风险投资项目风险评价指标都为成本型指标， 为了消除不同物理量纲对决策结果的影响，可用以 下公式[14]将决策矩阵 6( )ij sa a   转化为规范化决策矩 阵 6( ) ,ij sr r   其中 [ , ] { 0 1}ij ij ij ij ijr r r t r t r      ≤ ≤ ≤ ≤ ( 1, 2, , , 1, 2, , 6)i s j   。 2 2 1 1 1 1 , (1 ) (1 ) ij ij ij ijn n ij ij i i a a r r a a             ( 1, 2, , , 1, 2, , 6)i s j   (9) 定义 4 称 * * *1 2 6( , , , )r r r r    为区间型理想点，其 中 * * *[ , ] [max( ), max( )]( 1, 2, , 6)j j j ij ij i i r r r r r j         (10) 不确定多属性决策方法的实质是利用已有的决 策信息通过一定的方式对一组（有限个）备选方案 进行排序并择优。在处理风险投资项目风险评价问 题时，用区间数表示指标值较为合理，因此，可以 建立指标权重未知且指标值为区间数的不确定多属 性决策模型对多个风险投资项目的风险进行排序。 由规范化矩阵 6( )ij sr r   、指标权重向量 1 2 6( , , , ) tw w w w  及区间数的运算法则可知，风险 投资项目 ix 的风险综合属性值为 6 6 6 1 1 1 [ , ] ( 1, 2, , )i ij j ij j ij j j j j z r w r w r w i s           (11) 式中 jw 为第 j 个指标 ju 的未知权重，满足单位 化约束条件 6 2 1 1 1, 2, , 6j j j w w j    （ ≥0 ）。 因为风险评价指标权重向量 1 2 6( , , , ) tw w w w  完全未知，所以不能直接由(11)式确定项目风险综合 属性值。 徐泽水[16]针对只有部分权重信息且指标值为实 数的不确定多属性决策问题，提出过一种基于方案 满意度的决策方法。该研究指出对于每个方案 ( 1, 2, , )ix i s  而言，其满意度总是越大越好。利 用该研究思想，本文构建基于区间型理想点的不确 定多属性决策模型解决多个风险投资项目风险评价 问题。 对于风险投资项目 ix 而言，它越接近区间型理 想点就越优。因此，可令项目 ix 与区间型理想点 published by atlantis press copyright: the authors 145 * * * 1 2 6( , , , )r r r r    之间的加权偏差为 6 6 * * 1 1 ( ) ( , )i ij j j ij j j j j d w r r w d r r w          ( 1, 2, , )i s  (12) 对于给定的指标权重向量 1 2 6( , , , ) tw w w w  ， ( )id w 越小则项目 ix 越优。因此可建立如下的多目 标决策模型： 1 2 6 2 1 min ( ) ( ( ), ( ), , ( )) . 1 1, 2, , 6 s j j j d w d w d w d w s t w w j          ≥ 0 (13) 由于每个项目之间是公平竞争，不存在任何偏 好关系，因此，上述多目标决策模型可用等权线性 加权法综合为如下等价的单目标 优化模型： 6 * 1 1 1 6 2 1 min ( ) ( ) ( , ) . 1 1, 2, , 6 s s i ij j j i i j j j j d w d w d r r w s t w w j                   ≥ 0 (14) 求解该模型得到 * 1 6 * 2 1 1 ( , ) ( 1, 2, , 6) ( ( , )) s ij j i j s ij j j i d r r w j d r r              (15) 将上述权重向量 1 2 6( , , , ) tw w w w  作归一化处理， 可得 * * 1 6 * 1 1 ( , ) ( 1, 2, , 6) ( , ) s ij j i j s ij j j i d r r w j d r r             (16) 求出 优权重向量 * * * *1 2 6( , , , ) tw w w w  之后，通 过(11)式计算出各个项目的风险综合属性值 ( 1, 2, , )iz i s  。由于 ( 1, 2, , )iz i s  仍然是区间 数，不便于直接对项目风险进行排序，因此可利用 (1)式计算出 ( 1, 2, , )iz i s  之间的可能度，并建立 可能度矩阵 ( )ik s sp p  ，其中 ( )ik i kp p z z  ≥ 。 ( , 1, 2, , )i k s  矩阵 p 是一个互补判断矩阵，利 用(5)式得到矩阵 p 的排序向量 1 2( , , , ) t s     并按其分量大小对区间数 ( 1, 2, , )iz i s  进行排序， 由此可得到风险投资项目 1 2, , , sx x x 的风险排 序。 5. 算例分析 5.1 单项目风险等级评价算例 假设三个风险投资家对某一风险投资项目的风 险进行评估，评估指标如第 2 节所述。 不妨设三个风险投资家的重要性程度相同，按 照3.1节所述方法，对该三个风险投资家的区间数判 断矩阵进行处理（具体计算过程略），得到区间数 判断矩阵 6 6( )ija a  。 1 [2.33, 3.33] [2.00, 3.00] [0.30, 0.43] 1 [0.33, 0.50] [0.33, 0.50] [2.00, 3.03] 1 [0.27, 0.37] [0.33, 0.50] [0.33, 0.50] [2.00, 3.03] [3.03, 4.00] [2.00,3.03] [0.23, 0.30] [0.30, 0.43] [0.27, 0.37] a         [2.67,3.67] [0.33, 0.50] [3.33, 4.33] [2.00,3.00] [0.25, 0.33] [2.33,3.33] [2.00,3.00] [0.33, 0.50] [2.67,3.67] 1 [0.20, 0.33] [2.00,3.00] [3.03, 5.00] 1 [3.67,5.33] [0.33, 0.50] [0.19, 0.27] 1         则 1 2.33 2.00 2.67 0.33 3.33 0.30 1 0.33 2.00 0.25 2.33 0.33 2.00 1 2.00 0.33 2.67 0.27 0.33 0.33 1 0.20 2.00 2.00 3.03 2.00 3.03 1 3.67 0.23 0.30 0.27 0.33 0.19 1 a                     published by atlantis press copyright: the authors 146 1 3.33 3.00 3.67 0.50 4.33 0.43 1 0.50 3.00 0.33 3.33 0.50 3.03 1 3.00 0.50 3.67 0.37 0.50 0.50 1 0.33 3.00 3.03 4.00 3.03 5.00 1 5.33 0.30 0.43 0.37 0.50 0.27 1 a                     利用方根法求出 ,a a  的归一化权重向量分别 为： (0.231, 0.127, 0.169, 0.095, 0.307, 0.071)w  (0.227, 0.126, 0.171, 0.097, 0.310, 0.069)w  由(2)式得到： 0.913 1, 1.103 1k m    ，因此 区间数判断矩阵 a 满足一致性要求。 由(3)式得到： 3 4 5 6 1 2 [0.116, 0.139], [0.154, 0.189], [0.087, 0.107], [0.280, 0.342], [0.065, 0.076] [0.211, 0.250], w w w w w w       由(1)式计算出各区间数权重之间的可能度，并 建立可能度矩阵为： 0.5 1 1 1 0 1 0 0.5 0 1 0 1 0 1 0.5 1 0 1 0 0 0 0.5 0 1 1 1 1 1 0.5 1 0 0 0 0 0 0.5 p                     由(5)式求出各个评估指标的权重为： (0.217, 0.150, 0.183, 0.117, 0.250, 0.083)w  现假设这三个风险投资家对该项目评估指标的 评估值如表 1 所示： 表 1 三位专家对该风险项目非系统风险的评估值 1u 2u 3u 4u 5u 6u 1d 2d 3d 2s 1s 2s 1s 1s 1s 2s 1s 1s 0s 0s 1s 3s 0s 2s 1s 0s 1s 表1中 ( 1, 2, 3)jd j  表示第 j 位专家。 利用(7)式对以上指标评估信息进行集结，得到 这三个风险投资家对该项目风险的语言评估值分别 为： 1 2 31.317 0.550 1.001, ,s s s s s s       再利用(8)式对这三个风险投资家的评估信息进 行集结，得到该项目非系统风险的语言评估值为 0.8703bs s 。 根据单项目风险等级评估准则(1)可知，该项目 的风险在置信度 87.03% 下的风险等级为 1s ，即该 项目的风险在置信度87.03% 下处于“较低” 等 级。 5.2 多项目风险排序算例 假设现有 5 个备选项目 1 2 5, , ,x x x 可供风险投 资家进行选择，风险投资家从项目风险因素角度对 这些项目进行评价，评价指标如第 2 节所述。评价 对象空间为： 1 2 5{ , , , }x x x x  ， 评价指标空间为： 1 2 6{ , , , }u u u u  。得到项目 决策矩阵 5 6( )ija a   为： [3, 4] [3, 4] [2,3] [2, 3] [2, 4] [4,5] [2, 3] [3, 4] [4, 5] [2, 4] [3, 4] [2, 3] [2, 3] [3, 4] [4, 5] [2, 4] [2, 3] [3,5] [2, 4] [2, 5] [2,3] [3, 4] [3, 5] [2, 3] [3, 4] [2, 5] [2, 4] [3, 5] [4,5] [2, 4] a                  将决策矩阵 a 按(9)式规范化以后得到规范化决 策矩阵 5 6( )ijr r   如下： [0.254, 0.521] [0.274, 0.645] [0.357, 0.828] [0.338, 0.781] [0.274, 0.645] [0.214, 0.414] [0.338, 0.781] [0.274, 0.645] [0.214, 0.414] [0.254, 0.781] [0.219, 0.967] [0.357, 0.828] [0.254, 0.521] [0.219, 0.967] [0.267, r          0.828] [0.338, 0.859] [0.282, 0.889] [0.208, 0.414] [0.254, 0.859] [0.282, 0.593] [0.347, 0.828] [0.254, 0.859] [0.376, 0.889] [0.208, 0.552] [0.254, 0.573] [0.226, 0.593] [0.347, 0.828] [0.203, 0.573] [0.226, 0.445] [0.260, 0.828]         利用(16)式可以求得 优权重向量为 * (0.115, 0.196, 0.176, 0.142, 0.219, 0.152)tw  利用(11)式求得各项目的风险综合属性值： published by atlantis press copyright: the authors 147 1 2( ) [0.287, 0.712], ( ) [0.281, 0.667],z w z w   3 4( ) [0.303, 0.730], ( ) [0.263, 0.737],z w z w   5 ( ) [0.237, 0.635]z w  。 由(1)式计算出各项目综合属性值之间的可能 度，并建立可能度矩阵为： 0.5 0.545 0.490 0.509 0.586 0.455 0.5 0.444 0.466 0.546 0.510 0.556 0.5 0.518 0.598 0.491 0.534 0.482 0.5 0.573 0.414 0.454 0.402 0.427 0.5 p                 由(5)式计算得到可能度矩阵 p 的排序向量为： (0.207, 0.196, 0.209, 0.204, 0.185)t  由排序向量 及矩阵 p 中的可能度，得到区间 数 ( ) ( 1, 2, , 5)iz w i   的排序： 3 1 4 2 5 0.510 0.509 0.534 0.546 ( ) ( ) ( ) ( ) ( )z w z w z w z w z w        因此这5个风险投资项目的排序为（从风险角度 考虑）： 3 1 4 2 5 0.510 0.509 0.534 0.546 x x x x x    . 故 优项目为 3x 。 6. 结束语 在借鉴国内外风险投资项目风险评价指标的基 础上，本文建立了符合我国实际情况的风险投资项 目风险评价指标体系，针对单个风险投资项目的风 险等级确定问题和多个风险投资项目的风险排序问 题，分别构建模型予以解决。在单项目风险评价模 型的构建中，充分考虑了风险投资项目和专家主观 认识上的不确定性，采用区间数对风险投资项目的 风险评估指标相对重要性程度进行描述。本文构建 的模型能够在一定置信度下确定风险投资项目的风 险等级，避免了传统评估方法所得结果的绝对化。 在多项目风险评价模型的构建中，同样用区间数来 描述风险投资项目风险评价指标的风险等级，构建 了基于区间型理想点的不确定多属性决策模型，从 而实现对多个风险投资项目进行风险排序。由于充 分认识到风险投资项目和专家主观认识上的不确定 性，而采用区间数来进行刻画，因此本文构建的风 险评价模型符合风险投资的特点，具有较好的科学 性和合理性。 参考文献 [1] 成思危.成思危论风险投资.北京:中国人民大学出版 社,2008. [2] tyebjee t t, b runo a v. a model of ven ture capitalist investment activity . management science, 1 984,30(9): 1051-1066. [3] wright m. r obbie k. venture capital. bookfiled : dartmouth publishing company ltd. 1997. [4] koski t m .using fuzz y s et theory in assessing the s uccess potential of venture capital investment theory and pilot field study.international journal of entrepreneurship and innovation management,2003,3(5):509-524. [5] cheng c h, y ang k l, hwang c l. evaluating attack helicopters b y ahp based on linguistic vari able weight. european journal of operational research, 199 9, 116(2):423-435. [6] 钱水土,周春喜. 风险投资的风险综合评价研究. 数量经 济技术经济研究,2002,(5):45-48. [7] 徐宪平 . 风险投资的风险评价与控制 . 中国管理科 学,2001,9 (4):7580. [8] 万玉成,盛昭瀚. 基于未确知测度的风险投资非系统风险 的评价与控制研究. 系统工程理论与实践,24(11):22-27. [9] 赵振武,唐万生,宁玉富. 基于模糊模拟的 ahp 在风险投 资项目评价中的应用. 模糊系统与数学,20(4):128-133. [10] 刘德学,樊治平,王欣荣. 风险投资项目非系统风险的分析 与评价. 系统工程理论方法应用,11(3)198-201. [11] 刘开第,庞彦军,吴和琴等. 模糊隶属度定义中隐含的问题. 系统工程理论与实践,2000,20(1):110112. [12] 萧维嘉,仝允桓,王剑雨. 基于阶段门的初创企业评价指标 差异性研究. 中国软科学,2008,(12):121-128 [13] 周宗放等. 新兴技术企业信用风险评价理论与方法.北京: 科学出版社,2010. [14] 徐泽水.不确定性多属性决策方法及应用.北京:清华大学 出版社,2004. [15] 周光明,刘树人.不确定多属性决策中区间数的一种新排 序法.系统工程,2006,24(4):115-117. [16] 徐泽水,孙在东. 一种基于方案满意度的不确定多属性决 策方法. 系统工程,2001,19(3):76-79. published by atlantis press copyright: the authors 148 microsoft word volume 12, issue 3-2 revised journal of risk analysis and crisis response, 2022, 12(3), 124-134 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i3.333 124 article behavioral heterogeneity and excessive volatility of rmb exchange rate zhong-qiang zhou 1,2, jiajia wu 1,2 and sheng yuan 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 key laboratory of green fintech, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: 528072697@qq.com; tel.: +86-0851-88510575 received: september 2, 2022; accepted: september 28, 2022; published: september 30, 2022 abstract: the paper develops a two-type behavioral heterogeneous agent model including fundamentalists and chartists. it examines whether investors’ behavioral heterogeneity is related to the excessive volatility of rmb exchange rate. we use the deviation of the real exchange rate from the fundamental exchange rate as a measure of excessive exchange rate volatility. the fundamental value is calculated by the revised rmb fundamental exchange rate model with cointegration technology. after estimating the behavioral heterogeneous agent model using the monthly rmb exchange rate data from october 2006 to november 2020, we find that the heterogeneity of traders in price and trading strategies can significantly explain excess volatility of the rmb exchange rate. our analysis of two significant fluctuations in 2015-2016 and 2018-2019 further corroborates our key finding that investors’ behavioral heterogeneity plays an important role in explaining excess volatility of rmb exchange rate. keywords: exchange rate volatility; behavioral heterogeneity; agent-based model 1. introduction the exchange rate is considered as a “barometer” of national economic development, reflecting the stability of economic growth and serving as a link between the country’s economy and the world economy. the rmb exchange rate maintained a small fluctuation within roughly 1% per month (except for the 2008 financial crisis) before the reform of china’s foreign exchange system in 2015. after the reform, the rmb exchange rate had been increasing volatile. for example, it hit its maximum monthly volatility of 3.84% in july 2018. a large body of literature considers macroeconomic factors as the main factors which resulted in the rmb exchange rate volatility. however, as one of the typical characteristics in the formation of the exchange rate mechanism, the forex traders’ heterogeneity was underappreciated in existing studies. our study attempts to explain the rmb exchange rate volatility with the behavioral heterogeneity of forex traders. in the empirical study of meese and rogoff (1983), the traditional model of the exchange rate has been difficult to explain various abnormal foreign exchange phenomena, which is no match for the simple random walk model in terms of out-of-sample prediction. de grauwe and grimaldi (2005, 2006) and de grauwe and markiewicz (2013) use behavioral heterogeneity models to simulate the process of excessive exchange rate volatility. these models assume two types (fundamentalists and zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 125 chartists) of traders in the foreign exchange market, which have heterogeneous beliefs about exchange rate expectations and determine the current trading strategy (including strategies of fundamentalists and chartists) based on the profitability of the previous trading strategy. the simulation results show that the investor's behavioral heterogeneity can significantly explain the excess volatility in the foreign exchange market. therefore, based on these studies, we try to provide the empirical evidence from the rmb foreign exchange market. the measurement of rmb fundamental exchange rate is vital in an empirical analysis of exchange rate fluctuations. li and chen (2010) construct a model of the rmb fundamental exchange rate, assuming that the rmb fundamental exchange rate depends on macroeconomic factors that consistent with economic fundamentals. the rmb fundamental exchange rate is empirically measured by cointegration technology. our study calculates the rmb fundamental exchange rate from october 2006 to november 2020 by using cointegration technology based on the revised model according to the work of hu (2014). our paper has two contributions. first, it enriches the empirical literature on the relationship between traders’ behavior heterogeneity and excess volatility in the foreign exchange market. there is a great deal of theoretical research that consider traders’ behavior heterogeneity as an essential factor to explain excessive exchange rate volatility (see manzan and westerhoff (2007); bauer (2009); de jong (2010); buncic (2016); li and wu (2018)), however, few articles focus on empirical research to support the above theoretical studies. furthermore, the previous studies concern the foreign exchange markets of developed economies such as japan, the united states and europe, whereas our study provides empirical evidence from an emerging market by testing the link between traders’ behavior heterogeneity and the rmb exchange rate volatility. second, our study endogenously explains the excessive fluctuation of the rmb exchange rate. there are many articles which study the exogenous factors that affect the volatility of the rmb exchange rate (see chen (2020); zhou (2020); lucey et al (2020); liu (2020); tian and li (2021)), however few articles consider endogenous factors of rmb exchange rate fluctuations (see li and chen (2010); hu (2014)). our study builds a nonlinear heterogeneous agent model to endogenously explain the rmb exchange rate fluctuations. the remaining structure of this paper is arranged as follows. section 2 describes the rmb exchange rate model with two behavioral heterogeneous agents. section 3 calculates the rmb fundamental exchange rate from october 2006 to november 2020. in section 4, we estimate the exchange rate model and analyze the results. section 5 concludes. 2. the model 2.1. model assumptions -there are two types (fundamentalists and chartists) of traders in the exchange market. fundamentalists trade based on mean-reversion, while chartists trade based on trend-following. -the limited rational traders in the exchange market only know the past information about relevant variables in the model. -the expectation rules of traders in the foreign exchange market are dynamically adjusted based on the performance of the previous trading rules. 2.2. fundamental exchange rate zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 126 models to calculate the rmb fundamental exchange rate according to chen and li (2010) are as follows: 𝑙𝑛𝑠∗ = 𝑐 + 𝑐 × (𝑚 − 𝑚∗) + 𝑐 × (𝑦 − 𝑦∗) + 𝑐 × 𝑖 − 𝑖 ∗ + 𝑐 × 𝑖 − 𝑖 ∗ + 𝑐 × (𝑝 − 𝑝∗) + 𝜀 (1) where 𝑙𝑛𝑠 ∗ is the natural logarithm of the fundamental exchange rate, 𝑚 is domestic money supply, and 𝑦 is national output. 𝑖 measures domestic short-term interest rate, while 𝑖 measures domestic long-term interest rate, and 𝑝 measures national inflation rate. the variabls with * on the right of the equation represent variables corresponding to united states. therefore, the rmb fundamental exchange rate is determined by the divergences in the money supply, output, shortterm interest rates, long-term interest rates, and inflation rates between china and the united states. we find multiple collinearities between short-term and long-term interest rates in our analysis, which is consistent with hu (2014). thus, we delete the expected inflation rate difference variable (𝑖 ) and use hu's rmb fundamental exchange rate model as follows: 𝑙𝑛𝑠∗ = 𝑐 + 𝑐 × (𝑚 − 𝑚∗) + 𝑐 × (𝑦 − 𝑦∗) + 𝑐 × 𝑖 − 𝑖 ∗ + 𝑐 × (𝑝 − 𝑝∗) (2) it is difficult to determine whether 𝑐 is positive or negative. according to the elastic price currency analysis method, the relative growth of a country's money supply will lead to an increase in domestic inflation in the same proportion, which ultimately depreciate the local currency. but according to the vicious price currency analysis method, there is an "overshoot" in the exchange rate market that increased domestic money supply will lead to excessive depreciation. we define 𝑐 and 𝑐 negative, while 𝑐 positive. china is an export-oriented economy, which makes exports a significant role in supporting the economy. when the domestic output increases, the national surplus will increase too, which contributes to more foreign currency assets held by our nation. under the foreign exchange settlement and sale system, the foreign currency supply in the foreign exchange market will increase, leading to the depreciation of the foreign currency. similarly, the domestic currency attraction increases as the domestic short-term interest rate rises, leading to a growing domestic currency demand in the foreign exchange market, so the local currency appreciates. for 𝑐 , the rise in domestic inflation will theoretically lead to a decline in the attractiveness of the local currency and a depreciation of the domestic currency. 2.3. heterogeneity expectations based on the previous assumption, two types of traders using different expectation rules are considered in our model. 𝐸 , 𝐸 are heterogeneous beliefs of fundamentalists and chartists. the fraction of fundamentalists is 𝑛 , whereas the fraction of chartists is 𝑛 . the exchange rate expectation 𝐸 (𝑠 ) for the fundamentalists is given by: 𝐸 (𝑠 ) = 𝑠 + 𝛼 × (𝑠 − 𝑠 ∗ ), −1 < 𝛼 ≤ 0 (3) where 𝛼 reflects traders’ sensitivity to the deviation between the exchange rate and its fundamental values. chartists ' expectation of exchange rates is given by: 𝐸 (𝑠 ) = 𝑠 + 𝜑 × (𝑠 − 𝑠 ), 𝜑 > 0 (4) where 𝜑 is the extrapolation coefficient. the dynamic fractions 𝑛 and 𝑛 are updated as shown below: zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 127 𝑛 = 𝑒𝑥𝑝(𝛽 × 𝑈 ) 𝑒𝑥𝑝(𝛽 × 𝑈 ) + 𝑒𝑥𝑝(𝛽 × 𝑈 ) (5) 𝑛 = 1 − 𝑛 (6) where 𝛽 is the intensity of choice, reflecting the degree to which traders switch between different expectation rules. 𝑈 is traders’ risk-adjusted profitability in period 𝑡 − 1. inconsistent rules adopted by traders to gain will create uncertainty, so the traders should adjust the corresponding risk when calculating and selecting returns from the different expected rules. given risk 𝜎 , , and the traders’ risk aversion coefficient 𝜇 = 1, risk-adjusted profit is given by: 𝑈 = 𝜋 − 𝜇 × 𝜎 , (7) where 𝜋 is the realized profit on the investment at 𝑡 − 1 and the maturity of 𝑡 period, as shown below: 𝜋 = (𝑠 − 𝑠 ) × 𝑝𝑙𝑢𝑠[𝐸 (𝑆 ) − 𝑠 ] (8) where 𝑝𝑙𝑢𝑠[𝑥] = 1, 𝑓𝑜𝑟 𝑥 > 0 0, 𝑓𝑜𝑟 𝑥 = 0 −1, 𝑓or 𝑥 < 0 , 𝑖 = 𝑐, 𝑓. according de grauwe (2005), risk 𝜎 , is measured by the square of the prediction error as follows: 𝜎 , = [𝐸 (𝑠 ) − 𝑠 ] 1 + (𝑠 − 𝑠∗ ) (9) 𝜎 , = [𝐸 (𝑠 ) − 𝑠 ] (10) where (𝑠 − 𝑠 ∗ ) is the misalignment. as the misalignment increases, fundamentalists attach less importance to the short-term volatility measured by the one-period forecast error. they become increasingly confident that the exchange rates will revert to their fundamental values. combined with equations (3) to (10), the expectation of the exchange rate is the weighted average of the expectation of the two types of heterogeneous traders: 𝐸 (𝑠 ) = 𝑠 + 𝑛 × 𝛼 × (𝑠 − 𝑠 ∗ ) + 𝑛 × 𝜑 × (𝑠 − 𝑠 ) (11) so, the estimation model in this paper can be expressed as: 𝑠 = 𝑠 + 𝑛 × 𝛼 × (𝑠 − 𝑠 ∗ ) + 𝑛 × 𝜑 × (𝑠 − 𝑠 ) + 𝜀 (12) where 𝜀 is independent and identical distributed shock with a gaussian distribution of 𝑁~(0,1). 3. data this paper uses monthly data of the rmb exchange rate from october 2006 to november 2020, with 170 samples in total, which was obtained from the national bureau of statistics of china. the rmb exchange rate s is the average monthly exchange rate against the dollar under the direct price method. the broad money supply m2 represents the money supply of china and the united states. the unit of china's broad money supply m is rmb 1 billion, and the unit of the u.s. money supply m∗ is $ 1 billion. the outputs of china and the united states are expressed in broad terms of total domestic production, which are converted from quarterly data to monthly data. the unit of china’s gdp y is rmb 1 billion, and the unit of us gdp y∗ is dollars 1 billion. the above-mentioned data was all derived from the wind. china's short-term interest rate i uses the shanghai interbank zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 128 offered rate, and the u.s. short-term interest rate i∗ uses the u.s. three-month treasury bill interest rate, which comes from the ecic. in the cointegration analysis, it is found that the difference between the monthly consumer price index and producer price index between china and the united states is stable from october 2006 to november 2020. therefore, our paper uses industrial producers' monthly ex-factory price index that comes from the ecic to measure the two countries' inflation levels. the index forms a fixed base of 100 in 2005. we use model (2) to calculate the dynamic rmb fundamental exchange rate, which can be divided into three steps. first, the monthly exchange rate of rmb against the u.s. dollar and the macroeconomic variables in the model are used to establish a long-term cointegration equation. second, we perform an h-p filtering analysis on the macroeconomic variables in the cointegration equation to obtain the long-term values of the macroeconomic variable, because the rmb fundamental exchange rate depends on the long-term values of macroeconomics. third, we calculate fundamental exchange rate based on the estimated cointegration equation and the long-term values of the macroeconomic variable. the adf unit root test is performed on the variables in the formula (2) of the model, and the test results are shown in table 1. the variables lns∗ , (m − m∗), (y − y∗), (i − i ∗ ), (p − p∗) are all first-order unit covariates. table 1. the unit root test of the variables. the variables include lns∗, (m − m∗), (y − y∗), (i − i ∗ ) and (p − p∗). the test types (c, t, l) represent constants, trend items, and lag items. *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. variable the type of test pro-check value adf statistics p value 1% 5% 10% 𝐥𝐧𝐬𝐭 ∗ （c, t, 13） -4.0136 -3.4368 -3.1425 -2.4230 0.3664 𝐝(𝐥𝐧𝐬𝐭 ∗) （c, 0, 13） -3.4695 -2.8786 -2.5760 -7.1821 0.0000*** (𝐦𝐭 − 𝐦𝐭 ∗) （c, t, 13） -4.0133 -3.4366 -3.1425 0.8853 0.9998 𝐝(𝐦𝐭 − 𝐦𝐭 ∗) （c, 0, 13） -3.4695 -2.8786 -2.5760 -9.8527 0.0000*** (𝐲𝐭 − 𝐲𝐭 ∗) （c, t, 13） -4.0172 -3.4385 -3.1436 -2.6961 0.2398 𝐝(𝐲𝐭 − 𝐲𝐭 ∗) （c,0, 13） -3.4720 -2.8797 -2.5765 -3.2586 0.0185** (𝐢𝐭 𝐬 − 𝐢𝐭 𝐬∗ ) （c, t, 13） -4.0133 -3.4366 -3.1425 -2.5226 0.3169 𝐝(𝐢𝐭 𝐬 − 𝐢𝐭 𝐬∗ ) （c, 0, 13） -3.4695 -2.8786 -2.5760 -11.2959 0.0000*** (𝐩𝐭 − 𝐩𝐭 ∗) （c, t, 13） -4.0176 -3.4387 -3.1437 -2.024 0.5678 𝐝(𝐩𝐭 − 𝐩𝐭 ∗) （c, 0, 13） -3.4723 -2.8798 -2.5766 -3.5005 0.0092*** therefore, there may be cointegration relationships among variables lns∗, (m − m∗), (y − y∗), (i − i ∗ ) and (p − p∗). the results of the johansen cointegration test are shown in table 2 and table 3. zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 129 table 2. johansen trace test results (trace) of the variables. the variables include lns∗, (m − m∗), (y − y∗), (i − i ∗ ) and (p − p∗). *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. the original assumption eigenvalue trace statistics 5% threshold p-value note 0.322697 141.0014 69.81889 0.0000*** at most 1 0.172072 75.54238 47.85613 0.0000*** at most 2 0.133366 43.81904 29.79707 0.0007*** at most 3 0.076086 19.77175 15.49471 0.0106** at most 4 0.037819 6.476917 3.841466 0.0109** table 3. johansen max feature test results of the variables. the variables include lns∗, (m − m∗), (y − y∗), (i − i ∗ ) and (p − p∗). *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. the original assumption eigenvalue trace statistics 5% threshold p-value note 0.322697 65.45902 33.87687 0.0000*** at most 1 0.172072 31.72334 27.58434 0.0138** at most 2 0.133366 24.04729 21.13162 0.0189** at most 3 0.076086 13.29483 14.26460 0.0707* at most 4 0.037819 6.476917 3.841466 0.0109** a cointegration equation is estimated, and the standardized cointegration coefficients are shown in table 4. table 4. standardized cointegration equation. using cointegration technology, we obtain an equation of cointegration. the variables include lns∗, (m − m∗), (y − y∗), (i − i ∗ ), and (p − p∗). variable 𝐥𝐧𝐬𝐭∗ (𝐦𝐭 − 𝐦𝐭∗) (𝐲𝐭 − 𝐲𝐭∗) (𝐢𝐭𝐬 − 𝐢𝐭𝐬 ∗ ) (𝐩𝐭 − 𝐩𝐭∗) constant items estimate 1.000000 -2.285026 2.389038 0.004184 -1.615043 -1.165414 standard deviation 0.30603 0.31853 0.01423 0.97248 the cointegration equation is as follows: lns∗ = 1.165414 + 2.285026 × (m − m∗) − 2.389038 × (y − y∗) − 0.004184 × (i − i ∗ ) + 1.615043 × (p − p∗) (13) to calculate the rmb fundamental exchange rate, we performed the h-p filtering analysis on the economic variables in the cointegration equation. the long-term values of the rmb fundamental exchange rate are given by substituting the long-term values of these variables into the model (13). according to figure 1, the rmb exchange rate always fluctuates around its fundamental values from october 2006 to july 2018. since august 2018, the deviation has continued to increase. although the rmb appreciated relatively to the us dollar before the reform, the change was relatively small compared to the change in its fundamental value. the pressure of the continuous appreciation of the rmb should not be neglected although it developed slowly, which led to a gradual increase in the deviation of the rmb exchange rate from 2006 to mid-2007. there were two reasons as following. on the one hand, the rmb was under pressure to appreciate as china’s domestic economy was growing rapidly and commodity prices were falling. on the other hand, the united states economy was in the zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 130 doldrums and called for a rapid appreciation of the rmb (hu, 2014). for a longer period after mid2007, the rmb exchange rate was overvalued. more seriously, since the end of 2018, the deviation of the rmb exchange rate has continued to increase. figure 1. panel (a) is the rmb exchange rate, panel (b) is the rmb fundamental exchange rate, and panel (c) is the deviation between the rmb exchange rate and its fundamental value from october 2006 and november 2020. 4. estimated results this section presents results of the estimation of the model (12). we estimate the model by the nonlinear least squares method (nlls), and the parameter estimation results are shown in table 5. table 5. estimation results of the model. this table shows the estimation results of the behavioral heterogeneous agent’s model given in equation (3) – (13) that, which includes the belief coefficient of fundamentalists (α), the belief coefficient of chartists (φ), and the intensity of choice (β). *, ** and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively. parameter numeric value standard deviation t-value p-value α -0.02892 0.01261 2.292 0.0232** φ 1.19516 0.19945 5.992 0.0000*** β 0.00113 0.00095 1.186 0.2375 the results show that the two belief coefficients α and φ are significant and differ at the 5% confidence level, which means there are two distinct types of expectorants of heterogeneity in the rmb foreign exchange market. the belief coefficient of fundamentalists α is -0.02892, while the belief coefficient of chartists (φ) is 1.19516, which complies with the economic theory hypothesis mentioned above. we also found that the estimated result of the intensity of choice β is not significant because our sample size is too small, as explained by hommes and in ’t veld (2017). if it exists a significant heterogeneity in the estimated mechanism, it should not be worrisome if heterogeneity indicators are not significant in the estimation. zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 131 figure 2. panel (a) shows the changes of the rmb exchange rate, panel (b) is the exchange rate expectations of fundamentalists, and panel (c) is the exchange rate expectations of chartists from february 2007 to november 2020. figure2 (b) and (c) show the exchange rate expectations of fundamentalists and chartists. fundamentalists have obvious regression expectations: during the period when the rmb currency was undervalued, that is, before february 2008, and after july 2018, fundamentalists expected the rmb to appreciate. as the rmb exchange rate approached its fundamental values, the exchange rate expectations of fundamentalists decreased gradually. when the rmb exchange rate was overvalued, especially from september 2012 to august 2016, fundamentalists expected the rmb to depreciate. besides, as the exchange rate deviated from its fundamental values, the expected depreciation of fundamentalists was getting bigger and bigger. furthermore, comparing the trend of the exchange rate expected by chartists in figure 2(c) with the trend of the rmb exchange rate in figure 2(a), we discover that the expected trend of the exchange rate by chartists lags the rmb exchange rate. according to historical change trends, the expected method of chartists has inferred expectations. figure 3 shows the ratio changes between fundamentalists and chartists during the sample period. before march 2008, the rmb exchange rate was undervalued, and the proportion of fundamentalists increased gradually. the fundamentalists’ mean-reversion expectations further increased the appreciation of the rmb exchange rate. during this period, both types of traders expected the exchange rate to appreciate in the foreign exchange market. however, the proportion of fundamentalists did not continue to rise. on the one hand, as the rmb exchange rate appreciated significantly, the deviations decreased gradually, so as their profitability. on the other hand, both types of traders believed that the rmb exchange rate would appreciate, the herding effect led to an overvaluation of the rmb exchange rate. as a result, the proportion of fundamentalists declined from march to november 2008, coupled with the impact of the financial crisis. zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 132 figure 3. panel (a) shows the ratio of fundamentalists, and panel (b) shows the ratio of chartists from february 2007 to november 2020 in the foreign exchange market. influenced by the financial crisis, the two types of traders in the foreign exchange market have been in a stable state for a long time. since the belief value of fundamentalists is much lower than the value of chartists, the exchange rate expectations of chartists will significantly influence the changes in the rmb exchange rate. thus, although the rmb exchange rate has been overvalued from november 2008 to august 2015, it was still slowly appreciating. this is mainly due to the rapid development of the chinese economy after the financial crisis. while the rmb exchange rate has appreciated, its fundamental value has also appreciated at a higher rate. however, since late 2013, there has been an economic “bubble” and there was a stock market crash in august 2015. the impact of the stock market crash also spread to the foreign exchange market, so the rmb exchange rate began to fall. as the deviation in the exchange rate decreased, the proportion of fundamentalists decreased accordingly. as a result, fundamentalists played a dominant role in the foreign exchange market from october 2015 to may 2017, but their dominant power diminished as deviations decreased. after the stock market crash, the social economy returned to normality, and the foreign exchange market entered a short-term stable situation. the rmb fundamental exchange rate has entered a state of appreciation, and the proportion of fundamentalists has increased gradually. however, the proportion of fundamentalists will not remain at a high level for a long time. as the deviation decreases, the proportion of fundamentalists will decrease. as we all know, the united states has imposed several rounds of large-scale economic sanctions against china since 2018. figure 1 shows that the deviation continued to increase after june 2018. at the same time, the proportion of fundamentalists has risen rapidly in the foreign exchange market. the fundamentalists' mean-reversion expectations led to an appreciation of the rmb exchange rate. however, the united states continues to increase sanctions against china, which led to an increase in the proportion of chartists. as a result, since then, the fluctuations in the foreign exchange market have appeared in two directions. on the one hand, the increased united states sanctions against china will cause the rmb exchange rate to depreciate. on the other hand, the rmb exchange rate has been stayed under its fundamental value for a long time and there will be expectations of appreciation. the fundamentalists believe that the rmb exchange rate will appreciate, while the chartists believe that the rmb exchange rate will not appreciate due to the strong pressure from the united states. as a result, the ratio of fundamentalists to chartists fluctuates during this period, with the ratio of fundamentalists slightly higher than the ratio of chartists. zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 133 5. conclusions our study examines whether the excessive volatility of rmb exchange rate is associated with investors’ behavioral heterogeneity. consistent with the heterogeneous agent model prediction, results show that investors’ heterogeneity in price trends and trading strategies leads to rmb exchange rate volatility. this finding is further confirmed by our analysis of the two significant fluctuations of rmb exchange rate in 2015–2016 and 2018–2019. the switching of investors with heterogeneous beliefs contributed to the excess volatility of rmb exchange rate. our results offer an important implication. our evidence provides insights into the underlying sources that can explain the excess volatility of rmb exchange rate. we find empirical evidence that there is a strong linkage between investors’ heterogeneity and the fluctuation of the rmb exchange rate. thus, reducing the degree of heterogeneity in investors' behavior is conducive to maintaining the stable development of the rmb foreign exchange market. funding: this research was funded by the youth foundation for humanities and social sciences research of the ministry of education of china, grant number 22yjc790190. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] richard, a.m.; rogoff, k. empirical exchange rate models of the seventies: do they fit out of sample? journal of international economics,1983, 14(1-2), 3-24. doi: https://doi.org/10.1016/0022-1996(83)90017-x. [2] de grauwe, p.; grimaldi m. heterogeneity of agents, transactions costs and the exchange rate. journal of economic dynamics and control, 2005, 29(4), 691-719. doi: https://doi.org/10.1016/j.jedc.2004.01.004. [3] de grauwe, p.; grimaldi m. exchange rate puzzles: a tale of switching attractor. european economic review, 2006, 50, 1-3. doi: https://doi.org/10.1016/j.euroecorev.2004.07.005. [4] li x, chen h. traders' expected heterogeneity, the effectiveness of central bank intervention, and changes in rmb exchange rate. journal of financial research, 2010, 8, 49-67. doi：10.1360/972010-1084 (in chinese) [5] frankel, j.a.; froot, k.a. chartists, fundamentalists and the demand for dollars. greek economic review, 1991, 10(1). isbn: 1010-9994. [6] neuberg, l.; louargant, c.; protin, p. from heterogeneous expectations to exchange rate dynamics. luxembourg, 2004, sce, no. 310. [7] manzan, s.; westerhoff, f.h. heterogeneous expectations, exchange rate dynamics and predictability. journal of economic behavior & organization, 2007, 64, 111-128. doi: https://doi.org/10.1016/j.jebo.2006.08.005. [8] bauer, c.; de grauwe, p.; reitz, s. exchange rate dynamics in a target zone—a heterogeneous expectations approach. journal of economic dynamics and control, 2009, 33(2), 329-344. doi: https://doi.org/10.1016/j.jedc.2008.05.008. [9] de grauwe, p.; markiewicz, a. learning to forecast the exchange rate: two competing approaches. journal of international money and finance, 2013, 32, 42-76. doi: https://doi.org/10.1016/j.jimonfin.2012.03.001. [10] buncic, d.; piras, g.d. heterogeneous agents, the financial crisis and exchange rate predictability. journal of international money and finance, 2016, 60, 313-359. doi: https://doi.org/10.1016/j.jimonfin.2015.09.006. [11] li x, wu c. carry trade, heterogeneous expectations, and micro determination of exchange rate. journal of management sciences in china, 2018, 06, 1-11.doi: cnki:sun:jcyj.0.2018-06-001 (in chinese). [12] hu z y. the theoretical model construction and empirical study of the monthly average exchange rate of rmb to u.s. dollar. inquiry into economic issues, 2014, 08, 68-78. doi: 10.3969/j.issn.1006-2912.2014.08.012 (in chinese). zhong-qiang zhou, jiajia wu and sheng yuan / journal of risk analysis and crisis response, 2022, 12(3), 124-134 doi: https://doi.org/10.54560/jracr.v12i3.333 134 [13] flaschel, p.; hartmann, f.; malikane, c.; et al. a behavioral macroeconomic model of exchange rate fluctuations with complex market expectations formation. computational economics, 2015, 45(4), 669-691. doi: https://doi.org/10.1007/s10614-014-9437-8. [14] stanek, f.; kukacka, j. the impact of the tobin tax in a heterogeneous agent model of the foreign exchange market. computational economics, 2018, 51(4), 865-892. doi: https://doi.org/10.1007/s10614-017-9649-9. [15] şengül, o. does exchange rate volatility affect financial depth? evidence from brics countries. theoretical & applied economics, xxviii (2021), no. 1(626), spring, 247-258. [16] feng g, yang h, gong q. et al. what is the exchange rate volatility response to covid-19 and government interventions? economic analysis and policy, 2021, 69, 705-719. doi: https://doi.org/10.1016/j.eap.2021.01.018. [17] lu b, wang y, tan x. exchange rate volatility, heterogeneous firms and market concentration. china & world economy, 2020, 28(4), 51-75. doi: https://doi.org/10.1111/cwe.12326. [18] kurihara, y. exchange rate determination: which is appropriate, chartist model or fundamental model? journal of applied finance & banking, 2013, 3, 169-178. mag: https://oa.mg/work/2184372596. [19] teräsvirta, t. specification, estimation, and evaluation of smooth transition autoregressive models. journal of the american statistical association,1994, 89(425), 208-218. doi: https://doi.org/10.1080/01621459.1994.10476462. [20] tian, m.; li, w.; wen, f. the dynamic impact of oil price shocks on the stock market and the usd/rmb exchange rate: evidence from implied volatility indices. the north american journal of economics and finance, 2021, 55, 101310.1-101310.21. doi: https://doi.org/10.1016/j.najef.2020.101310. [21] chen l, du z, hu z. impact of economic policy uncertainty on exchange rate volatility of china. finance research letters, 2020, 32, 101266. doi: https://doi.org/10.1016/j.frl.2019.08.014. [22] lucey, b.; wang x, wang y. et al. can financial marketization mitigate the negative effect of exchange rate fluctuations on exports? evidence from chinese regions. finance research letters, 2020, 34, 101250. doi: https://doi.org/10.1016/j.frl.2019.07.023. [23] liu t, lee c. exchange rate fluctuations and interest rate policy. international journal of finance & economics, 27(3), 2020, 3531-3549. doi: https://doi.org/10.1002/ijfe.2336. [24] zhou z, fu z, jiang y. et al. can economic policy uncertainty predict exchange rate volatility? new evidence from the garch-midas model. finance research letters, 2020, 34, 101258. doi: https://doi.org/10.1016/j.frl.2019.08.006. [25] de jong, e.; verschoor, w.f.c.; zwinkels, r.c.j. heterogeneity of agents and exchange rate dynamics: evidence from the ems. journal of international money and finance, 2010, 29(8), 1652-1669. doi: https://doi.org/10.1016/j.jimonfin.2010.05.007. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word the nature event's probabilistic risk assessment of nuclear power plants with the nature event's probabilistic risk assessment of nuclear power plants with improvements chen yan, yu shaoqing, zhang chunming, chai jianshe* nuclear and radiation safety center, ministry of environmental protection beijing, 100082, china *e-mail: chaijshe@163.com abstract probabilistic risk assessment is the systematic and comprehensive methodology to evaluate the risk of nuclear power plants. currently some aspects, which are the hazard analysis and fragility evaluation in the probabilistic risk assessment of the nature event in nuclear power plants need to be improved. this paper introduces the principle of risk assessment of the nature disaster, and studies the relations between the two risk analysis methods, finally discusses the application of the technologies, which are the hazard factor, vulnerability analysis, loss risk and risk zone map in the nature disaster’s risk assessment, in the nature event pra of nuclear power plants. keywords: pra, risk assessment of nature disaster, hazard analysis, fragility evaluation. 一种改进的核电站自然灾害风险评价方法 陈妍, 余少青, 张春明, 柴建设 环境保护部核与辐射安全中心，北京 100082 摘要：概率风险技术是核电站风险定量评价常用的方法，目前核电站自然灾害概率风险评价方法在危害性 分析、脆弱性评估等方面亟需补充完善。本文引入了自然灾害风险分析的基本原理，研究了自然灾害风险 分析的基本原理与核电站自然灾害概率风险评价方法的联系，探讨了其在核电站自然灾害风险中的危害 性、脆弱性、损失分析和风险区划图方面的借鉴。该方法在完善核电站自然灾害概率风险评价方面能发挥 重要作用。 关键词：pra，自然灾害风险分析，危害性，脆弱性 1. 引言 二十世纪自然灾害在全球范围内造成了几百万 人的死亡，在我国仅因地震造成的人员死亡约 56 万 人。二十一世纪以来全球自然灾害频发,印度洋海 啸、美国卡特里那飓风、汶川地震、海地地震、澳 大利亚洪灾、日本大地震海啸等自然灾害给人类造 成了灾难性的损失。联合国国际减灾战略发布的全 球自然灾害统计报告指出，在 2000 年至 2009 年 间，全球共发生自然灾害事件 3852 起，直接导致超 过 78 万人死亡，近 20 亿人受到影响，经济损失高 达 9600 亿美元。 目前全球共有 440 多座在运核电站，各核电站 虽然在设计中充分考虑了防御自然灾害的能力，采 取了预防和缓解超设计基准事故的措施，但是自然 灾害仍然会对核电站造成一定风险。1992 年的安德 鲁飓风使得迈阿密南部的土耳其角核电站断电，电 站的应急柴油发电机确保了设备正常运行；2007 年 日本中越冲地震，引起柏崎刈羽核电厂 3 号机组主 变着火、6 号机乏燃料水池的水溅出、固体废物贮 存罐破损，致使微量放射性物质泄漏；2011 年日本 9.0 级大地震及其引起的海啸造成福岛核电站的严重 损坏，导致大量放射性物质释放。国际原子能机构 （iaea）在对福岛核事故的调查报告中指出对于超 设计基准的自然灾害，自然事件概率风险评价方法 的研究有助于核电站的外部灾害风险评估和事故管 理措施的改善 [1]。 journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 139-145 published by atlantis press copyright: the authors 139 administrateur texte tapé à la machine received 31 december 2011 administrateur texte tapé à la machine accepted 14 march 2012 administrateur texte tapé à la machine 陈妍 et.al 概率风险评价（probabilistic risk assessment, pra）是核电站风险定量评价常用的方法。pra 方 法首次大规模应用于核电站的安全研究是 1975 年的 《反应堆安全研究》（wash-1400）[2]。此后经过 三十多年的发展和完善，pra 已经被公认为核电站 安全分析的有效工具。iaea 指出“pra 技术已经 成为核电站安全评价的一个标准化工具，使人们能 够深入地了解核电站的设计、性能和环境影响，包 括对支配性风险因素的鉴别以及对可降低风险的各 种方案进行比较。pra 为安全有关问题的决策提供 了协调一致、完整的方法”。核电站 pra 按始发事 件范围可以分为内部事件和外部事件的 pra，外部 事件 pra 进一步分为自然事件和人为事件的 pra。 核电站自然事件 pra 是一种对地震、外部火 灾、台风、洪水等自然灾害造成电站风险的定量评 价方法，其在内部事件概率风险评价模型的基础上 加入自然灾害造成的影响。核电站自然事件 pra 主 要包括自然灾害的危害性分析、脆弱性评估和系统 分析及定量化三部分内容。 1988 年美国核管会 （nrc）要求各核电站对严重事故的易损性进行检 查评价，其中外部事件的单个电厂检查（ipeee） 推动了自然事件概率风险评价的发展。 1991 年 nureg-1407 总结出进行 ipeee 的核电站的外部事 件造成的堆芯损伤频率（cdf）在 10-6-10-4/堆•年的 范围，与内部事件造成的 cdf 在同一量级。我国 1984 年开始研究秦山和大亚湾核电站的概率风险评 价方法，目前国内核电站一级内部事件 pra 已经基 本完成。外部事件 pra 的研究自 2008 年初步开 展，在自然灾害的危害性分析和脆弱性评估等方面 亟需与各学科加强合作研究，以补充完善风险评价 方法中的数据和分析方法，推进我国核电站自然灾 害风险评估工作。 核电站自然事件 pra 属于自然灾害风险分析的 范畴。自然灾害风险分析是对风险区遭受不同强度 自然灾害的可能性及其可能造成的后果进行定量分 析和评估，其能够科学认识自然灾害的孕育、发 生、发展和可能造成的影响，避免防灾减灾行动的 盲目性，是防灾减灾的重要内容之一 [3]。自然灾害 风险分析自联合国 1987 年提出“国际减轻自然灾害 十年”以来得到了进一步的研究发展，其重要内容 包括致灾因子分析、承灾体易损性评价、灾情损失 分析和减灾对策等方面。目前国内已有较为系统地 自然灾害风险评价和风险分析的理论和方法体系[3,4], 但自然灾害风险分析理论和技术发展仍较缓慢 [5]。 基于自然灾害的多因性、系统性和不可预期 性，需要综合多个学科的技术来减轻自然灾害造成 的损失[6]。同时，基于核电站放射性危害的特殊 性，需要结合目前的 pra 技术和自然灾害风险分析 原理与实践对自然灾害造成的核电站风险进行深入 研究，以减轻可能的放射性后果。因此，本文将回 顾自然灾害风险分析的基本原理和核电站的自然事 件概率风险评价方法，并以地震灾害为例研究两种 风险分析方法的关键技术要素的联系，探讨自然灾 害风险分析的基本原理在核电站自然灾害风险中的 危害性、脆弱性和损失分析方面的借鉴。 2. 自然灾害风险分析的基本原理 [3,7] 自然灾害风险分析的基本原理是正视自然灾害 系统本身所固有的复杂性和不确定性，从基本的元 素着手分析，对其进行组合，进行不确定性意义下 的量化分析。自然灾害风险分析的基本内容主要包 括致灾因子风险分析、承灾体易损性评价、灾情损 失分析和减灾对策四个方面内容，其关系如图 1。 图 1. 自然灾害风险分析流程图 [3] （1） 致灾因子风险分析 可能造成灾害的因素称为致灾因子。在自然灾 害系统中，致灾因子通常指地震、洪水、干旱等灾 源。致灾因子风险分析主要研究给定区域内各种强 度的自然灾害发生的可能性，即估计时段 t 内，在 给定区域 s 内，致灾因子以 h 强度发生的条件可能 性 p(h|t, s)。通常说的地震危害性分析、洪水风险实 质上都是致灾因子风险分析。 （2） 承灾体易损性评价 承受灾害的对象称为承灾体。承灾体是复杂的 能量转化系统，担当着将自然灾害的破坏性能量转 化为破坏现象的角色。承灾体易损性评价的核心是 根据给定的致灾因子强度 h，计算承灾体破坏程度 d 的破坏模型 d=f(h)，主要包括以下三方面内容： 风险区确定：研究一定强度自然灾害发生时的 受灾范围。 选定灾种 选定承灾体 选定损失测度空间 致灾因子风险分 析估计 p（h|t,s） 承灾体易损性评价 破坏模型 d=f(h) 灾情损失评估 损失模型 l=g(d) 破坏风险 p(d) 损失风险 p(l)孕灾环境 风险等级划分 规划 减灾对策 社会经济特性 结果输出 书面报告 风险区划图 published by atlantis press copyright: the authors 140 一种改进的核电站自然灾害风险评价方法 风险区特性评价：对风险区内主要建筑物、其 它固定设备和建筑内部财产，风险区内的人口数 量、分布和经济发展水平等进行分析和评价。 抗灾性能分析：对风险区内的财产进行抗灾性 能分析（也称脆弱性分析）。许多承灾体的破坏机 理还没有完全掌握，函数关系 f(h)通常使用实验模 拟、历史灾情统计回归、模糊关系矩阵等方法进行 不确定的脆弱性分析。 （3） 灾情损失评估 灾情损失评估指评估风险区内一定时段内可能 发生的一系列不同强度自然灾害给风险区造成的可 能后果。损失包括经济损失和人员伤亡损失。灾情 损失评估的核心是更具破坏程度 d 计算损失程度模 型 l=g(d)。自然灾害系统几乎不可能用物理模拟的 方式寻找关系、更多地是用统计方法并用专家经验 来加以评估。 （4） 减灾对策 减灾对策是为减轻自然灾害的损失或影响程度 而采取的对策。如对建筑物进行抗震设计、制定紧 急救灾预案、建立巨灾保险基金等。减灾风险分析 需要评估，如果采取了某种减灾对策 f，当发生强度 为 h 的自然灾害打击时，造成额外损失 l 的可能性 p(l|f, h)。 综合自然灾害风险分析的主要内容，可以看到 自然灾害风险分析需要由各灾种的专家提供给定区 域内自然致灾因子发生时、空、强的可能性数值， 由防灾减灾工程师依据致灾因子强度，提供人类社 会系统各种破坏的可能性数值，由经济学家和社会 学家们依据破坏程度，推测各种损失的可能性数 值，最后，由自然灾害风险分析人员将三个环节的 可能性数值组合起来，给出损失风险。 3. 核电站自然事件 pra 方法 pra 是以概率论为基础、事件树/故障树为工具 的风险定量评价技术。核电站概率风险评价以真实 而非保守的方式将电站所有有关的信息，包括设 计、建造、运行、维修、设备可靠性、人因可靠 性、事故进程等综合考虑到一个电站 pra 模型中， 评价电站对公众健康与安全的潜在影响。核电站 pra 能够定量评估核电站风险，给出构筑物、系 统、部件（ssc）的相对风险及重要度排序，进而 发现电站的薄弱环节，并有效配置资源进行管理和 控制风险。 pra 一般分为三级 [8]，pra 的研究内容框架如 图 2。其中一级 pra 的目的是计算堆芯损坏频率。 二级 pra 是在一级的分析基础上研究堆芯损坏后的 事故进程及安全壳响应，评价各种放射性核素向环 境的释放量及释放频率，该级可以分析减缓堆芯损 坏后的事故后果途径和提出事故管理的具体意见。 三级 pra 是对二级 pra 结果加上厂外后果进行评 价进一步研究放射性物质在环境中的扩散及其对公 众和环境的影响，三级 pra 能够对后果减缓措施的 相对重要性做出分析，也能对应急响应计划的制定 提供支持。自然事件 pra 是在内部事件 pra 模型 图 2. 核电站 pra 的内容框架 基础上加入由自然灾害造成的影响，如危害性分析 为自然灾害造成的始发事件频率提供数据，脆弱性 分析为始发事件和系统分析中设备失效提供数据 等。自然事件 pra 主要包括三方面内容：自然灾害 的 危 害 性 分 析 (hazard analysis) 、 脆 弱 性 评 估 （fragility evaluation）和系统分析及定量化。 4. 自然灾害风险评价方法的联系 核电站自然事件 pra 属于自然灾害风险分析的 范畴，在研究方法内容上与自然灾害风险分析多有 相似之处。本节以地震 pra 为例 [9,10]，比较这两种 方法的关键技术要素，探讨关键要素之间的联系。 地震 pra 的研究方法流程如图 3，其关键技术要素 是地震危害性分析、地震脆弱性评估、电厂系统和 事件序列分析及堆芯损坏频率定量化评估。 4.1. 地震概率风险分析 （1） 地震危害性分析 核电站地震危害性分析的目的是研究核电站所 在厂址的不同水平地震地面运动（如峰值地面加速 度，pga）的发生频率。核电站地震危害性分析的 步骤是首先确定研究区域的地震源，然后统计该区 域的地震历史数据评估不同震级/烈度的地震发生频 率，研究衰减关系以评估地震导致的厂址的地面运 动的强度，最后综合上述信息评估研究区域的地面 运动参数超出某值的超越频率。地震危害性分析的 结果是描述某厂址不同水平地面运动发生频率的曲 published by atlantis press copyright: the authors 141 陈妍 et.al 线图，通常用 h(g)表示，如图 4，其不确定主要来 自衰减关系、地震源地质构造和可能发震强度的假 设。 图 3. 地震 pra 的研究方法流程 [10] 图 4. 地震危害性曲线 [10] （2） 地震脆弱性评估 地震脆弱性评估目的是评估电厂重要系统、构 筑物和部件在各级别的地面运动影响下（给定地震 动参数，如 pga、强度、加速度谱值）的条件失效 概率，其反映了地震导致的始发事件频率（如应急 交流电源丧失，强迫循环冷却系统丧失）以及不同 缓解系统或设备的条件失效概率（如辅助给水系 统）。地震脆弱性评估需要电厂设计基准、设计阶 段 ssc 的响应分析以及材料特性等，同时电厂巡访 对于地震脆弱性评估也非常重要。评估结果是描述 ssc 在不同 pga 下的条件失效概率的曲线图，如图 5。核电站中部件的脆弱性曲线 [11,12,13]通常用 eq.(1) 描述：                r u m qφ a a φqaf   )(ln ),( 1 . （1） 其中，φ为标准高斯累积分布函数，φ-1 是其逆函 数。am 是地面运动参数中位数，βu 描述模型不确 定性（认知不确定性），βr 描述随机不确定性（地 震及其影响的不确定性），q 为曲线的置信度，取 值（0,1），a 是地面运动强度参数。 图 5. 地震脆弱性评估曲线 [10] （3） 系统/事故序列分析和定量化 系统/事故序列分析将综合电厂逻辑图、部件脆 弱性评估和地震危害性分析，对导致并传递地震堆 芯损坏事故序列的各种构筑物和设备失效的组合进 行建模，然后对事故序列频率进行定量化。即危害 性分析为始发事件频率提供数据，脆弱性分析为始 发事件和系统分析中设备失效提供数据，继而通过 始发事件、事件序列、故障树和相关性等分析，使 用软件建立核电站 pra 模型进而得到堆芯损坏频率 定量化结果。堆芯损坏频率实质上是由灾害性曲线 和脆弱性曲线卷积成的一族数值{<pij,fij>}。其中，fij 是地震造成的电站损坏频率，pij 是这种损坏频率的 可能性： ji pqijp . （2） da da dh aff j iij    0 )( . （3） 其中，hj 代表第 j 个危害性曲线，fi 代表第 i 个电厂 脆弱性曲线；qi 是第 i 个脆弱性曲线的可能性，pj 是 第 j 个危害性曲线的可能性 [10,11,14]。 4.2. 自然灾害原理与地震 pra 方法的联系 基于地震 pra 属于自然灾害分析原理的范畴， 可以把地震 pra 中的技术要素纳入自然灾害风险分 析的基本原理框架，如图 6。可以看到，两种风险 分析方法中的技术要素有许多有相似之处，如致灾 因子、易损性评价、破坏风险方面是等价的；损失 published by atlantis press copyright: the authors 142 一种改进的核电站自然灾害风险评价方法 风险、孕灾环境、减灾对策方面原理上一致，但范 围不同。 图 6 地震 pra 与自然灾害风险分析基本原理的关系图 a) 致灾因子风险分析与危害性分析：致灾因子 风险分析主要研究给定区域内各种强度的自然灾害 发生的可能性, 核电站地震危害性分析主要研究核电 站所在厂址的不同水平地震地面运动的发生频率,因 此从统计角度致灾因子发生的条件可能性 p(h|t, s)与 核电站厂区地震危害性曲线 h(g)相对应，即地震危 害性分析实质上是致灾因子（地震）风险分析。地 震危害性曲线为核电站地震 pra 的始发事件频率提 供输入数据，因此危害性分析环节需要 pra 人员提 出需求，地震地质专家提供数据信息。 b) 承灾体易损性评价与核电站地震脆弱性评 价：承灾体易损性评价是计算承灾体在给定致灾因 子强度下的破坏程度，核电站地震脆弱性评价是评 估 ssc 在各级别的地面运动影响下的条件失效概 率。因此，核电站中的 ssc 是自然灾害原理中的承 灾体，破坏程度 d=f(h)与失效概率 f(a)等价描述， 承灾体（ssc）易损性评价与核电站地震脆弱性评 价一致。基于 ssc 失效概率是地震导致的始发事件 频率以及不同缓解系统或设备的条件失效概率的输 入，因此地震脆弱性评价环节也需要 pra 人员提出 需求，力学结构专家提供数据信息。 c) 破坏风险与系统定量化结果：从自然灾害风 险分析原理中可以看到破坏风险 p(d)是致灾因子、 易损性评价以及孕灾环境的函数；核电站系统定量 化结果（cdf/lerf）是在危害性曲线和脆弱性评价 作为输入的基础上，结合以事件树/故障树为工具的 电站状态分析得到风险定量结果。因此，p(d)与堆 芯 损 坏 频 率 cdf （ 早 期 大 量 放 射 性 释 放 频 率 lerf）基本等价。cdf/lerf 由核电站风险分析 pra 分析人员用软件建模获得。 d) 自然灾害基本原理的损失风险分析与核电站 的风险评估：自然灾害基本原理中损失风险通常包 括人员伤亡损失和经济损失。而在核电领域，目前 国内外核电站绝大多数只可以量化人员死亡损失风 险，即每年死亡人数（集体风险）或年人均死亡率 （个人风险），这一方面因为核电目前关注核安全 目标，另一方面由于技术模型发展的限制。因此与 自然灾害基本原理中的损失风险相比，范围较小。 此外，核电站中的纵深防御、三级应急响应体系的 方法属于自然灾害基本原理中的减灾对策的范畴。 4.3. 自然灾害风险分析原理在核电站自然事件 pra 中的借鉴 自然灾害危害性分析和 ssc 的脆弱性评估是核 电站自然事件 pra 模型不可或缺的输入环节。基于 自然灾害风险分析的基本原理与自然事件 pra 的联 系，以下将探讨其在核电站自然事件 pra 中的借 鉴： （1） 自然灾害危害性分析 自然灾害危害性分析给出的年超越频率等参数 一方面是特定核电站外部灾害筛选的重要依据，即 当自然事件频率低于定量筛选准则，且其后果不会 显著影响核电站安全时，对此自然事件可以不进行 pra 分析。另一方面，纳入核电站外部灾害分析的 事件，其年超越频率及其不确定度是 pra 模型的重 要参数。目前核电站自然事件 pra 需要各种自然灾 害在各种强度下的始发频率，但是自然灾害危害性 分析中存在统计样本较少，统计数据回归外推时不 确定度大等问题。模糊集方法是处理不完备信息条 件下风险分析常用的方法，借鉴自然灾害风险评价 理论中的信息分配方法，计算超越概率及其模糊风 险πm(p) [3]，并理解建立模糊风险和 pra 模型的接 口，可以补充完善核电站部分自然灾害危害性分析 结果。 （2） ssc 自然灾害脆弱性评估 核电站 ssc 的脆弱性评估的结果是 pra 模型中 自然灾害造成的始发事件及基本事件的条件失效概 率的输入。目前国内核电站 ssc 的脆弱性评估处在 pra 分析人员给出 ssc 的脆弱性评估的概率需求， 联合地震工程、结构力学等方向专家建立核电站 ssc 脆弱性模型及联合相关试验机构进行分析验证 阶段。核电站 ssc 的脆弱性评估一方面可以借鉴更 多的参数作为脆弱性评价的输入参数，如从单一的 地震动参数过渡到反应谱参数等 [15]，另一方面可以 借鉴其它领域的脆弱性分析中的经验法、解析/力学 等分析方法及破坏准则等，加快我国核电站 ssc 自 然灾害的脆弱性评估工作。 灾源：地震 核电站 sscs 人员伤亡损失风险 地震危害性分析 h（g） 地震脆弱性分析 f（a） 损失模型 系统定量化 人员死亡风险 气象、人口、健康 风险等级划分 规划缓解措施和应急响应 社会经济特性 结果输出 书面报告风险区划图 published by atlantis press copyright: the authors 143 陈妍 et.al （3） 核电站的风险评估 核电站的风险定义为后果与造成这种后果的事 故发生的频率的乘积，即 r(后果/单位时间)＝p(事 件/单位时间)×c(后果/事件)。目前国内外核电站用 三级 pra 模型可以量化人员死亡损失风险，还没有 考虑经济损失和人员辐照损伤损失，以及放射性物 质泄漏造成的长期的环境影响损失。此方面可以充 分借鉴自然灾害的基本原理及实践，在自然灾害危 害性分析、脆弱性评估和系统定量化基础上，联合 经济、环境、社会学方面的专家对核电站的综合损 失风险进行研究，以全面评价核电站的风险。 （4） 风险区划图 值得注意的是，自然灾害基本原理中的风险区 划图的概念方法在核电站 pra 评价风险还没有，可 以尝试其应用于核电站 pra 的风险评价中。风险区 划图是以地图的形式表述自然灾害风险大小及其区 域分异规律的一种方式，也是自然灾害风险评价的 终极产品 [16]。风险分析结果通过恰当的制图表达了 度量的风险信息，并支持决策的有效性。一方面， 核电各厂址的自然灾害危害性分析可以借鉴风险区 划图的分析结果，另一方面，风险区划图的概念可 以用于特定厂址的风险描述，即在 pra 的定量分析 结果基础上，使用重要度参数，画出特定厂址的风 险区划图，以直观描述在某种自然灾害下，电厂内 各系统设备的风险分布情况。但是在这一可能的应 用过程中，需要注意通常的地震区划图作为国标， 其对象为民用建筑，基本地震动参数的概率水平为 50 年超越概率 10%，与核电站抗震设计基准的概率 水平-不低于年平均超越概率 1e-4 之间存在差异。而 且区划图的比例尺在 1:100 万的数量级上，而核电 站概率地震危害性分析，在厂址周边 5km 范围内， 其地质调查结果的比例尺为 1:5000。因此二者的风 险区划图在工作内容、工作深度上将存在较大差 异。 5. 展望 pra 自上世纪七十年代提出，历经近四十年的 发展，目前已成为与确定论方法互为补充风险评价 方法。我国核安全法规 haf102《核动力厂设计安 全规定》5.9 中规定“必须对核动力厂设计进行安全 分析，在分析中必须采用确定论和概率论分析方 法”。2010 年 2 月,我国国家核安全局正式发布了 《概率安全分析技术在核安全领域中的应用》，提 出了八条技术政策以鼓励 pra 的应用研究工作。 我国核电站 pra 目前已在不同程度上完成了内 部事件的一级概率安全分析，部分核电站初步开展 了二级 pra 和外部事件 pra 的研究，三级 pra 的 研究初步展开。我国是易于受灾的国家，且具有灾 害类型多样、影响大和分布区域广等特征，因此国 内核电站外部事件的 pra 工作需要进一步深入研 究。 基于自然事件 pra 和自然灾害基本原理方法的 关键技术要素的联系，核电站自然事件 pra 可以充 分借鉴自然灾害基本原理中致灾因子风险分析、承 灾体易损性评价、损失风险、风险区划图等方法和 概念，并且与各灾种的专家、防灾减灾工程师、经 济学家和社会学家充分合作交流，合理评价核电风 险，提高核电应对自然灾害的能力。同时核电领域 的纵深防御、三级应急响应体系、pra 故障树等方 法可以在自然灾害风险分析中应用。 综上，核电站自然事件 pra 的危害性、脆弱性 和损失分析是电厂风险评价的重要环节。目前国内 核电站的自然灾害风险分析尚处于起步阶段，亟需 加强与各种自然灾害研究机构的联系，推进自然灾 害风险分析的致灾因子风险，危害性、易损性、损 失分析、风险区划图等技术在核电站自然事件 pra 中的应用，进一步完善核电站自然事件风险评价技 术及全范围 pra 的应用工作。 附加注解 在国内和 iaea，通常称 pra 为概率安全评价 （probabilistic safety assessment, psa）。 致谢 感谢张佳佳和荆旭对本文内容的有益讨论和建议。 参考文献 [1] iaea international fact finding expert mission of the fukushima dai-ichi npp accident following the great east japan earthquake and tsunami. mission report. 2011. [2] usnrc, reactor safety study: an assessment of accident risks in u.s. commercial nuclear power plant, wash-1400 nureg 75/014 .oct. 1975. [3] 黄崇福, 自然灾害风险评价：理论与实践[m], 科学出 版社, 2005. [4] 张继权, 李宁, 主要气象灾害风险评价与管理的数量 方法及应用[m], 北京师范大学出版社, 2007. [5] 庞西磊, 黄崇福, 赵思健, 自然灾害风险分析方法与相 关技术的研究进展跟踪[c],中国视角的风险分析和危 机反应-中国灾害防御协会风险分析专业委员会第四 届年会论文集, 2010. [6] 张继权, 冈田宪夫, 多多纳裕一, 综合自然灾害风险管 理—全面整合的模式与中国的战略选择[j], 自然灾害 学报, 15(1) (2006) 29-37. [7] 黄崇福, 自然灾害风险分析的基本原理[j], 自然灾害 学报, 8(2) (1995) 21-30. published by atlantis press copyright: the authors 144 一种改进的核电站自然灾害风险评价方法 [8] 陈捷飞，郭建兵.概率安全分析方法综述[j], 大亚湾 核电-核电概率安全评价技术专刊, (2006) 18-24. [9] m. k. ravindra, seismic individual plant examination of external events of us nuclear power plants: insights and implications, nuclear engineering and design, 175 (1997) 227-236. [10] standard for level 1/large early release frequency probabilistic risk assessment for nuclear power plant applications, asme/ans ra-sa–2009. [11] r. p. kennedy, m. k. ravindra, seismic for nuclear power plant risk studies, nuclear engineering and design, 79 (1984) 47–68. [12] nrc, handbook of nuclear power plant seismic fragilities, seismic safety margins research program, nureg/cr-3558. [13] j. h. kim, i. k. choi, j. h. park, uncertainty analysis of system fragility for seismic safety evaluation of npp, nuclear engineering and design, 241 (2011) 2570– 2579. [14] american nuclear society and national institute of electrical and electronic engineers, pra procedures guides: a guide to the performance of probabilistic risk assessment for nuclear power plants, nureg/cr-2300. [15] 于晓辉, 吕大刚, 王光远, 土木工程结构地震易损性分 析的研究进展[c], 第二届结构工程新进展国际论坛, 2008. [16] 张俊香, 黄崇福, 自然灾害软风险区划图模式研究[j], 自然灾害学报, 14(6) (2005) 20-25. published by atlantis press copyright: the authors 145 microsoft word a fuzzy systematic approach to construction risk analysis.doc a fuzzy systematic approach to construction risk analysis babak a. samani department of engineering, california state university east bay, 25800 carlos bee boulevard, hayward, ca 94542, usa e-mail: baghaebrahimisamani@horizon.csueastbay.edu farzad shahbodaghlou* department of engineering, california state university east bay, 25800 carlos bee boulevard, hayward, ca 94542, usa *e-mail: farzad.shahbodaghlou@csueastbay.edu abstract this article emphasizes the need for a fuzzy systematic structural approach to the risk assessment of construction projects and introduces the processes required to form a hierarchical systematic structure based on fuzzy logic using the fuzzy decision making trial and evaluation laboratory (fuzzy dematel) method. in this research, for the first time, the fuzzy dematel method has been successfully applied to the risk assessment of the persian gulf bridge megaproject, connecting qeshm island, located in the high risk strait of hormuz, to the mainland of iran, and produced logical results. in this case study, forty-five major risks, classified in ten groups of risks, were found, analyzed, and structured within a systematic hierarchical model. the final causal hierarchical structure of risks has been briefly analyzed and the risks have been prioritized with respect to their values of prominence in the model and their influence on the structure of risk factors. a result of this analysis confirms that country risks, including the state economy and multinational sanctions against the country are the most important risks affecting the project. another result shows that the most destructive risk can arise from the planning and budgeting of the sea bridge project. keywords: construction risk analysis, fuzzy dematel method, persian gulf bridge, construction risks. 1. introduction due to the new integrated approach to the construction project lifecycle and the vast implementation of turnkey contracts, the construction project is newly considered an integrated system rather than one of distinct phases. on the other hand, the number of projects is rapidly increasing every day, especially in developing countries with natural resources. the environment in which the project is executed is becoming ever more complex. in these integrated and complex environment projects, both the number of risks increases, along with the number of interrelationships between them. this increase in the number and interrelationships of risks creates a need for a systematic integrated approach to risk assessment that leads us to the use of systematic group decision making methods, which produce systematic structures of risks and their interrelationships. because of the fuzzy logic of the human way of thinking, which is to tradeoff between significance and precision, values presented in the procedure of risk assessment, by risk management team members, are not absolute. due to the recent discussion, fuzzy logic is accepted as a governing theory over the systematic structure. 2. a need for a fuzzy systematic approach to construction risk assessment in construction projects, since risk factors arise from varying areas, such as political, economic, social, technological and geographical environments, as well as from the project implementation, risk assessment is a group decision making process in which its system journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 275-284 published by atlantis press copyright: the authors 275 administrateur texte tapé à la machine received 4 february 2012 administrateur texte tapé à la machine accepted 20 november 2012 administrateur texte tapé à la machine babak a. samani & farzad shahbodaghlou criteria are risk factors that, even though they come from different areas, intensify each other. in these highrisk projects, in addition to directly impacting the project, individual risks indirectly impact the project objectives through their interrelationships. thus risks can no longer be classified by a product of two quantities known as probability of occurrence and direct impact on project goals. in other words, the priority of a risk is not determined solely by its prominence, a combination of its direct and indirect impacts on the project goals; rather, the priority of a risk is determined by taking into account both its prominence and its influence on other risks. this complicated nature of risk prioritization in the uncertain environment of newer projects creates a need for a systematic approach to risk assessment. this systematic approach encourages the use of group decision making based on a pair-wise comparison method in which experts from various areas determine the existence or lack of interrelationships between any pair of risks and the strength of the influence of that relation. in systematic risk assessment based on pair-wise comparison decision making methods, after risk identification, experts from various areas determine the existence or lack of interrelationships between any pair of risks and the strength of the influence of that relation. that strength of influence is described by a linguistic term such as “none,” “very low,” “low,” “high,” or “very high.” these terms are vague statements that are not absolutely precise, however meaningfully understandable and acceptable. traditionally, systematic group decision making leads to the assignment of a crisp quantity between 0 and 5, rather than the linguistic terms stated. the human way of thinking conflicts with the traditional assignment method because of its core principles. the human way of thinking is a decision making process that is a tradeoff between precision and significance; the results of this process are vague linguistic terms. vagueness is a form of epistemic uncertainty that is brought about by the human way of thinking that results in the imprecise meaning of linguistic terms. the tolerance of uncertainty in the process of thinking in human beings proposes that the logic behind the process of thought is logic based on fuzzy truths, fuzzy connectedness and fuzzy rules of inference. the various factors stated lead to the authorization of fuzzy logic as a governing theory over the systematic structure. fuzzy logic is a consequence of the developments of fuzzy sets by lotfi zadeh, the father of fuzzy logic. a fuzzy set is an extension of a classical (crisp) set. the allowance of membership in crisp sets is either full membership or no membership at all. in other words, in crisp sets an element is pertained to a set or is not pertained to it. fuzzy sets represent commonsense linguistic labels. a further explanation of fuzzy logic is out of the scope of this paper. 3. fuzzy dematel method decision making trial and evaluation laboratory, dematel, is a decision making method based on a pair-wise comparison using experts’ judgment. this method originated from the geneva research center of battelle memorial institute by fontela & gabus 1976. they developed the dematel method to find integrated solutions for the fragmented and antagonistic phenomena of world societies. using the dematel method, and by distinguishing system elements based on the experts’ idea generation, incorporating their assessments and judgments, and using the pair-wise comparison matrix and graph theory, a hierarchical structure of elements in a complex system can be achieved in a way that the cause and effect relationship of system elements and the strength of relationships’ influence is illustrated in the final structure. in addition to the practical feature of the dematel method, which is the structural visualization of complex casual relationships with matrix and diagraphs, another advantage of the dematel method over other pairwise comparison methods is that in the hierarchical structure each element can influence all other elements whether they are at the same level, an upper level or a lower level and are capable of being influenced by them. therefore, the value of each system is assessed by each and every element within that system. applying linguistic variables, a fuzzy extension of the dematel method, enables the crisp method to be more appropriate for solving multicriteria and multi expert problems in complex and fuzzy environments. recently, the fuzzy dematel method has been successfully applied in many fields. for the first time in construction risk assessment, in 2005, babak a. samani, successfully applied crisp dematel as a tool for risk assessment in iranian oil and gas projects and produced logical results. the final hierarchical model was valid and practical for years. recently, the fuzzy dematel method was found to be a better approach. therefore, this method was applied to the persian gulf sea bridge project. this bridge is located in the strait of hormuz and connects qeshm island to the mainland of iran, a high-risk developing country. the dematel method, whether crisp or fuzzy, is of greater service to larger projects than smaller ones due to the fact that risks arisen from various areas and different levels in larger projects have more interrelationships and dematel has the ability to take that into account. published by atlantis press copyright: the authors 276 a fuzzy systematic approach to construction risk analysis chi-jen lin and wei-wen wu (2004) clearly explained all essential fuzzy theorems to insure the convergence of this requirement. they discussed the mathematical procedure of computing the total relation fuzzy matrix t as well. the result of their method will suffice to meet our project goal. based on lin and wu’s paper, the analytical procedure of the fuzzy dematel method, omitting theorems and proofs for clarity in application, is described as follows: step (1) – problem identification, expert group, evaluation criteria the project management team should clearly identify its point of view and the project phase at which the risk assessment is being performed. the authors suggest that this stage occur as part of the feasibility study and economic justification. a group of experts from various fields relating to risks should be formed. applying one of the idea generation methods within the expert group, they must prepare a list of evaluation criteria, risks or uncertainties. step (2) – fuzzy pair-wise comparison matrix a pair-wise comparison matrix (matrix ( )kx ) from the evaluation criteria, which have been identified in step (1), should be generated and presented to the members of the expert group. then, the experts are asked to make sets of pair-wise comparisons to assign a degree of influence to each cell of matrix ( )kx . to deal with imprecision of human assessments of causal relations, which have several complicated aspects by nature, li r.j. (1999) developed a fuzzy scale with five different degrees of influence. in his proposed fuzzy scale, the degree of influence is one of five linguistic terms {no, very low, low, high, very high}. each linguistic term has its own corresponding positive triangular fuzzy number. in other words, instead of the crisp values, which have been used in the crisp dematel method, experts state their assessments with fuzzy values. the applied fuzzy scale including its linguistic terms and their corresponding positive triangular numbers is shown in table 1. if a group of p experts is asked to make sets of pairwise comparison, between criteria { }| ,1, 2, ,ic c i n= = … then the number of p pairwise matrixes as (1) ( 2) ( ), ,..., nx x x , each corresponding to a member of the expert group, will be obtained. the fuzzy scale should be applied to assign corresponding fuzzy numbers to linguistic terms in ( )kx obtained pair-wise comparison matrixes. so pairwise matrix ( )kx is denoted as ( ) 1 ( ) ( ) 1 0 0 k n k k k x x x ⎡ ⎤ ⎢ ⎥ = ⎢ ⎥ ⎢ ⎥ ⎣ ⎦ , k = 1, 2, 3, … , p, where p is the number of experts ( )( ) , , .kij ij ij ijx l m u= table 1. fuzzy scale for pair-wise comparison of evaluation criteria step (3) – fuzzy direct-relation matrix the directrelation matrix x is the average matrix of p pair-wise comparison matrixes, which are corresponded to the number of p experts, and calculated by the following formula: (1) ( 2) (3) ( )px x x x x p + + +…+ = fuzzy direct-relation matrix n nx × of criteria { }| 1, 2, ,ic c i n= = … is denoted as 1 2 1 12 1 2 21 2 1 2 0 0 0 n n n n n n n n c c c c x x x c x x c x x × ⎡ ⎤ ⎢ ⎥= ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎣ ⎦ where ( , , )ij ij ij ijx l m u= are triangular fuzzy numbers. elements , 1, 2,...,ijx i n= will be regarded as a triangular fuzzy number (0,0,0) wherever it is necessary. step (4) – normalization degree of influence linguistic term fuzzy value crisp value very high influence (vh) (0.75,1.0,1.0) 4 high influence (h) (0.5,0.75,1.0) 3 low influence (l) (0.25,0.5,0.75) 2 very low influence (vl) (0,0.25,0.5) 1 no influence (no) (0,0,0.25) 0 published by atlantis press copyright: the authors 277 babak a. samani & farzad shahbodaghlou in order to convert the criteria scales to comparable scales, the liner scale transformation is used to normalize the fuzzy directrelations matrix. if [ ]ij n nx x ×= and 1 1 1 1 ( , , ), n n n n ij ij ij ijj j j j r x l m u = = = = = =∑ ∑ ∑ ∑ then 11 , max n ijji j n r r u =≤ ≤ = = ∑ assuming that at least there is one i such that 1 n ijj u r = 〈∑ , then the normalized direct-relation fuzzy matrix denoted by m, equals 11 12 1 21 22 21 1 2 n n n n nn z z z z z z z r x z z z z − ⎛ ⎞ ⎜ ⎟ ⎜ ⎟= × ⇒ = ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ … where , ( , , )ij ij ij ijij x i m u z r r r r = = according to essential definitions, which are given before, the multiplication of two triangular fuzzy numbers results in another triangular fuzzy number. thus, the elements of matrix m are triangular fuzzy numbers. step (5) –total relation fuzzy matrix to figure the total-relation fuzzy matrix t we need to meet the requirement of lim [0]k n nk z ×→∞ = first. according to the crisp dematel method, the total relation matrix t is defined as 1 2lim( ... )k k t z z z →∞ = + + + when lim [0]k n nk z ×→∞ = . let define three crisp matrixes, whose elements are extracted from z , as 12 1 21 2 1 2 0 0 0 n n l n n l l l l z l l ⎛ ⎞ ⎜ ⎟ ⎜ ⎟= ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ … 12 1 21 2 1 2 0 0 0 n n m n n m m m m z m m ⎛ ⎞ ⎜ ⎟ ⎜ ⎟= ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ … 11 12 1 21 2 1 2 0 0 n n u n n u u u u u z u u ⎛ ⎞ ⎜ ⎟ ⎜ ⎟= ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ … and 11 12 1 21 22 2 1 2 t t t n t t t nt t t tn n nn ⎛ ⎞ ⎜ ⎟ ⎜ ⎟ = ⎜ ⎟ ⎜ ⎟ ⎜ ⎟⎜ ⎟ ⎝ ⎠ … where ( '' , '' , '' )ij ij ij ijt l m u= , and '' 1[ ] ( )ij i il z i z −= × − '' 1[ ] ( )ij m mm z i z −= × − '' 1[ ] ( )ij u uu z i z −= × − step (6) – prominence and influence of each criterion, hierarchical structure assuming id and ir are the sum of columns and the sum of the rows of matrix t , respectively, the prominence of each criterion can be obtained by adding id to ir and the influence of each criterion is calculated by subtracting ir from id . to acquire the causal diagram one of the defuzzification methods should be applied. then, the causal diagram and analysis will be drawn like crisp dematel method just by assigning ( )defi id r+ and ( ) def i ir d− values to the horizontal and vertical axis’s, respectively. 4. persian gulf sea bridge project the persian gulf sea bridge’s project objectives are construction of a bridge facility including marine, civil, structural, piping, electrical and any other necessary amenities and equipment to transport cars, trucks, trains, water, oil, gas and liquid products transporting pipe lines and cable connections (electricity, telecommunications, internet)from qeshm island to iran’s main land and vice versa. the specific location of qeshm island in the strait of hormuz, makes it important for the iranians. the strait of hormuzis a narrow waterway to the persian gulf, published by atlantis press copyright: the authors 278 a fuzzy systematic approach to construction risk analysis which is the most significant region of oil, gas, petrochemicals and minerals in the world. iran is an underdeveloped country in which several huge construction projects, such as oil and gas field developments, urban developments, power plants, dams, infrastructure improvements, and etc. are being planned and implemented. due to the importance of qeshm island for the country, there are many development plans that are being designed and executed on the island and its surrounding. 5. project risks based on field studies and experts opinions, 45 major risks to the general contractor in the persian gulf bridge project affecting the company’s engagement in the project have been identified from the private general contractor’s point of view at the very early stage of the project. then, based on the most important feature of each risk, they have been classified into 10 risk categories as shown in table 2. explanation of all the risk factors is beyond the scope of this paper. risk categories are briefly explained in the following: 5.1. country risks according to advanced engineering associate international (aeai) “country risk is a particularly general term, which refers to risks affecting all companies operating within a particular country. primarily, it refers to the risk of investing in a country.�major country risks of iran that affect the persian gulf bridge project are as follows. • multinational sanctions • possibility of regional war or conflicts • state economy • political instability of the government • high inflation rate • recession • corruption • out of date labor, tax, insurance, trade and environment laws • slow process of governmental permits issuance • slow and costly customs process • incapable it infrastructure 5.2. companies’ capabilities risks the success of a project depends on the capabilities of its participant companies, the project’s executive system, and the existing infrastructure, inclusive of legal, informative, cultural, and educational infrastructures. as a matter of fact, many project risks, which are usually called company’s internal risks, arise from within company parameters that compose the capability of a company. company’s capability is defined by parameters such as: strategy, policy, cash flow, organizational structure, management systems, and information technology infrastructure. it also requires leadership, skills and proficiency, personnel motivation and organizational culture. the more successful management operations are in these areas, the more a company’s capability will increase and the fewer hazards will arise. there are complex direct and indirect causal relationships between component parameters of a company’s capability, elements of its executive system and infrastructural factors of the environment in which the project is implemented. also, unsatisfactory results in time, quality, and cost of project implementation result in a drop in client/ governmental confidence levels causing that company to lose future projects and consequently intensifying the company’s weakness. see figure 1. due to several causal relationships within the system shown in figure 1, an extensive research is required to analyze the risks arisen from internal factors of a company. also, due to the large size of this research, the feedback links from project results are not taken into account. fig. 1.the effective parameters in projects success risks associated with companies' capabilities in persian gulf bridge project are as follows. • low level capability of contractors • lack of powerful foreign companies to participate in domestic projects 5.3. management risks risks associated with ineffective, destructive or underperforming management, which affects project goals are called management risks. management risks refer to the risks of the circumstance in which project parties would have been better off without the choices made by management. basically, management risks refer to the chance that project managers will put their own interests ahead of the interest of the project published by atlantis press copyright: the authors 279 babak a. samani & farzad shahbodaghlou stakeholders and project goals. major management risks in persian gulf bridge project are as follows. • frequent replacement of project managers • poor management • debility in project planning and budgeting • delay in choosing contractors and consultants 5.4. human capital risks construction projects essentially are reliant on people to succeed. employees, the human capital asset, with the right skill profile and capability, are one of the critical success factors of projects. according to seymour adler, (2008); human capital risk is employee-related events or behaviors that can affect the operations and/or value of the company. these risks may be generated by activities either inside or outside the workplace. major human capital risks in iran are as follows: • lack of professional and developed manpower • low level motivation and efficiency of existing manpower • low efficiency of training programs • debility in hiring foreign professionals 5.5. financial risks advanced engineering associate international (aeai, 2010) defines financial risks as follows; “financial risks are associated with any form of financing and reflect the uncertainty of whether the return on an investment will be less than that required for the repayment of lending”. lack of funds to provide the necessary budget to manage costly construction projects along with the emergence of financial problems during project execution are the primary challenge facing iranian industry. risks arisen from project financing area are as follows. • insufficient economic justification • weakness of banking system • absence of foreign banks 5.6. insurance risks insurance risks reflect the uncertainty of insurance and bonding. since the low capable iranian companies fail to meet the project primary goals, many insurance companies refuse to insure their projects. risks associated with project insurance in iranian construction projects are as follows. • weakness of bonding and insurance companies • absence of foreign bonding and insurance companies 5.7. procurement risks risks originating from procurement activities are known as procurement risks. in huge projects 40% to 60% of budget is assigned to procurement activities. thus, risks associated with procurement activities have huge impacts on project primary goals. major procurement risks in persian gulf bridge project are as listed below. • low quality of domestic products • manufacturers’ debility in manufacturing of some project equipment • obstacles of abroad procurement • impediments of software purchasing 5.8. health, safety and environmental risks similar to all construction projects, health, safety and environmental (hse) risks are highlighted in persian gulf bridge project. lack of health and safety culture within iranians makes the situation worse. because of unique environment of site location and pure tourist attractions, environmental risks are highly required to be taken into consideration. two major risks arisen from hse area are as listed below. • insufficiency of health, safety and environmental regulations • high risk project in terms of health, safety and environmental risks 5.9. site condition risks differing site conditions has always been an issue in construction contracts. hence, risks arisen from site condition should be taken into account in terms of risk assessment. major site condition risks are as follows. • geotechnical risks (intertidal zones floored by unconsolidated sandy mud and clay, over laid on marine calcareous terraces) • earthquake-prone area • hydrological risks (tsunami waves, strong two-way water -flow and high water – salinity) • mangrove forests • traffic of large vessels and huge ships • distance from primary sources and materials • distances from major manufacturers which are concentrated in central states 5.10. design, construction, operation and maintenance risks this category of risks is all about risks arisen during the project implementation; those are more affected by the other risk categories and have less impact on them. these types of risks are very typical so that they can be identified in all construction projects. main execution risks in persian gulf bridge are listed below. • lack of experience in this type of project • design risks • change orders • force majeure • operational risks published by atlantis press copyright: the authors 280 a fuzzy systematic approach to construction risk analysis • maintenance risks 6. applying the fuzzy dematel method to the persian gulf sea bridge project in this research, the fuzzy dematel method has been applied twice. one time it was applied for all 45 risks of the project, while another time it was applied for only the 10 major risks. for brevity, the risk categories and application for the risk categories will be discussed. applying step 2 of the fuzzy dematel method, the fuzzy pair-wise comparison matrix has been obtained and shown in figure 2. the comparison matrix reflects the strength of influence of each risk category on the other risk categories by the linguistic terms. the authors believe that while the project risks gradually damage the project primary goals (time, cost, quality and safety), the back influences from the project primary goals on the project risks intensify the project risks subsequently. in other words, there are bilateral relationships between the project risks and the project’s primary goals. since discussion about this novel idea is beyond the scope of this research, the back influences from primary goals on risk categories are not taken into account. nevertheless, the final hierarchical model is still valid, because the back influences have very low strengths plus risk analyzers unconsciously consider the back influences during the pair-wise comparison between risk categories. there are also interrelationships between cost, quality, time, and safety, which are to be taken into account in the risk assessment process. since different risks have different direct and indirect effects on each of the project objectives that in turn has different influences on the rest of project objectives, the project management team should prioritize the project goals and make a trade-off between cost, time, quality and safety throughout the risk assessment process. otherwise the risk prioritization will not be sensible. the trade-off between project goals is done in lower right cells of pair-wise comparison matrix 1. using matlab programming software and assigning the corresponding positive triangular numbers, specified in table 1, to the given linguistic terms of comparison matrix 1 and completing the method’s steps, the fuzzy values of ( )i iw r d= + and ( )i iq r d= − have been calculated for each risk category. using defuzzification operation, the crisp values of w and q have been obtained. tables 3 and 4 show the fuzzy and crisp vales of w and q for each risk category, respectively. as discussed in step 6 of the fuzzy dematel method, the crisp values of w and q should be assigned to the horizontal and vertical axis of coordinating system of final hierarchical model, respectively. the final hierarchical model has been acquired by mapping the dataset of ( , )def defw q proving final systematic structure of the risk categories. figure 3 demonstrates the final hierarchical model. table 2. project risk categories and risks 7. analyzing the final systematic structure of risk categories the crisp value of w indicates weight of each risk category importance and the crisp value of q specifies influence of each risk category and divides the risk categories into a cause or effect group. when q is published by atlantis press copyright: the authors 281 babak a. samani & farzad shahbodaghlou positive, the risk category belongs to the cause group. otherwise, if the q is negative, the risk category belongs to the effect group. in the top right hand part of figure 3 where the highest values of w and q of risks exist (w=1.67, q=1.17), country risks (cn) having the high prominence in model, would be less affected by the other risks within the system, but would influence them strongly. hence, to mitigate country risks, we should directly use external areas from the final model. addressing these big risks, which affect 90 percent of other criteria within the final model directly and intensively, would be a great help in mitigating the most of risks. at the right hand side of the coordinate system, where we see the highest number of w and a relatively large amount of q, management risks (mg) are placed (w=1.85,q=0.60). occupying the area of highest prominence values, management risks are the most important risks of persian gulf bridge project and highly intensify the other project risks. in addition, management risks are also tremendously influenced by the other risks because of their large value of q. many practices show that numbers of management risks are originated from country risks. due to the vertical fig.2. pair-wise comparison matrix of risk categories tables 3. fuzzy and crisp values of ( )i iq r d= − for risk categories tables 4. fuzzy and crisp values of ( )i iw r d= + for risk categories fig.3. the final systematic structure of risk categories location of human capital risks (hc) and financial risks (fn) between country risks (cn) and management risks (mn) in the coordinate system, it is inferable that similar to every other project the management risks (mn) are largely influenced by country risks (cn) before by the other risks. two major factors of any management system are human capital and financial capital. thus any risk associated with these two areas can influence the management system. this concept is illustrated well in the final hierarchical model. locations of these two categories of risks in figure 3 perfectly demonstrate this hypothesis. the capability of a construction company is associated with its management system, financial and human capitals and insurance resources. if we look at the place of contractors’ capabilities risks (cm) in figure 3, we published by atlantis press copyright: the authors 282 a fuzzy systematic approach to construction risk analysis can see that the contractors’ capability risks (cm) are located in the middle of the casual diagraph. this category of risks is influenced by the lack of sufficient human (hc) and financial (fn) capitals as well as management risks. companies’ capabilities risks also influence the risks arisen from project execution process such as procurement, design, construction, health and safety, environmental, operational and maintenance risks. noticing the positions of the design, build and operational (db) risks category and the health, safety and environmental risks category (hs), both risk categories have negative values of influence (q<0) which indicate that these categories of risks belong to the effect group of casual diagraph. although the influence value of procurement risks category (pr) is positive, it is still very close to the other project execution risks (db and hs). thus procurement risks might belong to the effect group of risks if they are evaluated comparison to all other 41 risks. generally project execution risks are highly influenced by upstream risks such as procurement, construction, operational and maintenance risks. risks associated with the project implementation process have a lower value of prominence than the upstream risks such as country, management, finance and human capital and companies’ capabilities risks and have less back influence on them. as it is shown in figure 3, since the risk assessment has been done at the early stages of the persian gulf bridge project, the health, safety and environmental risk category (hs) has less value of prominence comparison with the other risks originated from project execution process (db and pr). at the mid-left side of figure 3, where we see the positive calculated amounts of influence (q>0) and small values of prominence )75.0( ≤w , the category of site condition risks (sc) is located. typically, site conditions risks (sc) are independent risks that do not receive major influence from other risk categories. due to their positive influence strength (q = 0.54), site condition risks, however, have significant influences on the several risks associated with project implementation system in the areas of design, procurement and construction. site condition risks also intensify the planning and budgeting risks and have some effects on risks associated with companies’ capabilities. 7.1. trade-off between project objectives at the lowest level of figure 3, with the most negative values of q, the project objectives are strongly impacted by the project risks. in this area, the health, safety and environment (hse) goal is located in the lower left position of the other objectives. this means that the hse goal is the least important and the most impacted project objective. there are many problems that should be resolved to mitigate the hse risk and there is a danger of abandonment of this goal. cost objective (c) has taken a place slightly below the time objective and on its left side. since the time goal (t) has a higher position than the other project objectives, it can be inferred that governmental stakeholders of project are extremely concern about meeting the project milestones. this could be due to their effort to take a partisan advantage of the sea bridge project in the approaching presidential election in 2012. the project’s quality objective (q) is located on the bottom right side of the casual diagraph. although the quality objective has a relatively large value of prominence, it is in danger to be scarified for the other project’s goals because of its highly negative value of influence. arrows drawn in the coordinating system (figure 3) help with the understanding of influence flow in the casual diagraph. obviously a dematel model is not adjustable from bottom up. 8. conclusion and recommendation 8.1 conclusion after reviewing the concept of uncertainty, risk, and fuzzy logic, the authors concluded that the logic inspiring risk assessment is neither absolute nor probabilistic, rather that it is fuzzy. the authors strongly believe that the traditional decision making methods widely used in the risk assessment process are better when shifted to fuzzy operations. due to the new integrated approach to the construction project lifecycle and the vast implementation of turnkey contracts, construction project is newly considered an integrated system rather than one of distinct phases. on the other hand, the number of megaprojects is increasing every day, especially in developing countries with natural resources. the environment in which the project is executed is becoming ever more complex. in theses integrated and complex environment of megaprojects, the number of risks increases, along with the number of interrelationships between them. this increase in the number and interrelationships of risks creates a need for a systematic integrated approach to risk assessment that leads us to the use of systematic group decision-making methods. this in turn produces systematic structures of risks and their interrelationships. because of the fuzzy logic of the human way of thinking, which is to tradeoff between significance and precision, values presented in the procedure of risk assessment, by risk management team members, are not absolute. due to the recent published by atlantis press copyright: the authors 283 babak a. samani & farzad shahbodaghlou discussions, fuzzy logic is accepted as a governing theory over the systematic structure. in this research, for the first time, the fuzzy dematel method has been successfully applied to the megaproject of the persian gulf bridge project and has produced logical results. in this case study, forty-five major risks, classified in ten groups of risks, were found, analyzed, and structured within a systematic hierarchical model. a result of this analysis confirms that country risks, including political instability of the government, the state economy, and multinational sanctions against the country are the most important risks affecting the project. another result shows that the most serious risk can arise from the planning and budgeting of sea bridge project. trading-off between project goals is an activity that should be done in the process of risk analysis, applying pair-wise comparison methods. the casual relationships between the project objectives should be taken into account because each of the project objectives can be impacted by any direct impact of each risk and by the indirect impact of risks through the other project goals. the use of the fuzzy dematel method in risk assessment is highly recommended. 8.2 recommendations for further research first, evaluating the validation of the final hierarchical model structured in this research can be a new topic. applying neural network models can be an approach to this evaluation. secondly, defining fuzzy functions for the project objectives and their relationships with project risks can be done as further research. thirdly, complementary research can follow this project so that the back-influences from the project goals on the risks are taken into account. 9. references 1. babak a. samani (2012). “applying the fuzzy dematel method to the risk analysis of construction projects”. m.s. thesis, california state university east bay. 2. rongjun li (1999). “fuzzy method in group decision making”. department of decision science, central south university changsha, hunan 410083, p.r. china 3. chi-jen lin, wei-wen wu (2004). “a fuzzy extension of the dematel method for group decision-making”, department of industrial engineering and management, ta hwa institute of technology. 4. advanced engineering associates international, inc. (aeai) (2010) sheberghan gas field development plan risk analysis report – http://www.sheberghangas.com 5. seymour adler, amy mills & matt shadrick (2008) why you should be a human capital risk manager – aon consulting & kellogg school of management published by atlantis press copyright: the authors 284 the errors estimate of optimal combined forecasting* thethethethe errorserrorserrorserrors estimateestimateestimateestimate ofofofof thethethethe multistagemultistagemultistagemultistage combinedcombinedcombinedcombined iiiinvestmentnvestmentnvestmentnvestment riskriskriskrisk assessmentassessmentassessmentassessment∗∗∗∗ yyyyuuuu jikejikejikejike zhouzhouzhouzhou zongfangzongfangzongfangzongfang1 school of management and economics, university of electronic science & technology chengdu sichuan, 610054, p.r.china abstractabstractabstractabstract investment risk is economic development faced serious risk. the multistage combination investment risk assessment (mcira) can reduce the assessment error, but how to survey the error which produces by the mcira models, has the important significance. from theoretical side, the errors upper-bound of the mcira models is determined in this paper. we also give the relationships between the errors of the general mcira models, the simple average models and the errors of each investment risk assessment model in the combination. keywords: investment risk; errors estimate; simple average models; mcira models. ∗this research has been supported by national natural science foundation of china (no. 70971015). 1zhou zongfang, professor, zhouzf@uestc.edu.cn journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 106-109 published by atlantis press copyright: the authors 106 mailto:zhouzf@uestc.edu.cn zhou zongfang and yu jike 1.1.1.1. introductionintroductionintroductionintroduction since the combined investment risk assessment methods were brought forward, a number of scholar have studied the manifold methods of determining the combined weighted coefficients, such as: the equal-weight or the simple average method, the recursive equal-weight method, the superiority-matrix method, novel particle swarm, real options and so on[1]-[8]. to our great regret, there were few research results available on the errors bounds of multistage combined investment risk assessment problems [9]-[10]. in this paper, we first determine the errors maximum of the simple average combined investment risk assessment model, then the errors maximum of the general multistage combined investment risk assessment model will be discussed, at last we will give the relationships between the errors of the general multistage combined investment risk assessment (mcira) model and the simple average models and the errors of each investment risk assessment model in the combined assessment. we utilize n investment risk assessment models for the same investment risk assessment problem within t given stages. to determining the combined risk assessment weights, the multistage combined investment risk assessment (mcira) problem is described as below: on the assumption that yt (t=1,2, …, t) are actual observations of investment risk in the t-th stage; fit (t=1,2, …, t; i=1,2, …, n) are the risk assessment values in advance of the i-th investment risk assessment model; eit = ytttt ---fit are errors between the observations and risk assessment values of the i-th model in the t-th stage. the combined investment risk assessment values are: ∑ = = n i itit ff 1 µ , t=1,2, …, t (1) where µ i (i=1,2, …, n) are called the combined risk assessment weights, which satisfies∑ = n i i 1 µ =1, (i =1,2, …, n); et =∑ = n i ite 1 = ytttt ---ft (t =1,2, …, t) are called errors of the combined risk assessment. let ∑∑∑ === == n i iti t t t t t eej 1 2 11 2 )( µ µµµµ )( 1 1 1 )]([ n n i n j t t jtitji eee ′== ∑∑ ∑ = = = (2) be the errors square sum of the mcira model, where )( 21 ′= n,μ,,μμμ ⋯ is called weight vector; )( 21 ′= itiii ,e,,eee ⋯ (i=1,2,…, n) are error vectors of the i-th model, jijiij eeee ′== , nnijn ee ×= )()( is a n × n symmetric matrix, i t t itii jee == ∑ =1 2 (i=1,2, …, n) is just the investment risk assessment errors square sum of the i-th investment risk assessment model. the matrix )(ne provides the error messages of each investment risk assessment model. we call )(ne the risk assessment errors information matrix of the mcira model and assume the matrix )(ne is invertible (or replace by the errors vectors ei (i=1,2, …, n) are linear independence). definitiondefinitiondefinitiondefinition 1:1:1:1: if *(n) ** μeμj )( ′= , ( *iμ ≥ 0, i=1,2, …, n) is minimal about µ , the µ * is called the optimal combined assessment weight vector, the j* is called the minimal errors square sum of the optimal mcira model. 2.2.2.2. thethethethe errorserrorserrorserrors boundsboundsboundsbounds ofofofof simplesimplesimplesimple averageaverageaverageaverage modelmodelmodelmodel definitiondefinitiondefinitiondefinition 2:2:2:2: in equation (1), if n μi 1 = , i=1,2, …, n , then the corresponding mcira model is called the simple average model. let errors square sum of simple average model is: )0()0( )( μeμj (n)a ′= (3) where ) 1 ,, 1 , 1 ()0( ′= nnn μ ⋯ . if the errors square sum of the i-th investment risk assessment model is ji, let their maximum and minimum published by atlantis press copyright: the authors 107 the errors estimate of the multistage combined investment risk assessment be respectively: { }ini jj ≤≤= 1max max and { }ini jj ≤≤= 1min min . theoremtheoremtheoremtheorem 1:1:1:1: if the mcira model is the simple average model, then maxjja ≤ . (4) corollarycorollarycorollarycorollary 1:1:1:1: if the mcira model is the simple average model, then mina jj < if and only if ∑∑ = = < n i n j ij jne 1 min 2 1 (5) theoremtheoremtheoremtheorem 2:2:2:2: supposing (n)e is an positive definite matrix, if the errors square sum of each investment risk assessment model in the combination is the same constant c ( i.e. ji = c, i=1,2,…, n), then aj < c (6) above theorem indicates that the simple average model can reduce the assessment errors, if ji = c (constant), i=1,2, …, n. for example, under the hypothesis of theorem 2, if ),,,( 21 ′= itiii eeee ⋯ , i=1,2,…,n are n (t ≥ n) orthogonal vectors in t-dimension space (i.e. nnjinnij(n) eeee ×× ′== )()( is a scalar matrix, and its diagonal elements are constant c), then the errors square sum of simple average model is c n 1 . 3.3.3.3. thethethethe errorserrorserrorserrors estimateestimateestimateestimate ofofofof thethethethe generalgeneralgeneralgeneral mciramciramciramciramodelsmodelsmodelsmodels let the errors square sum of the general mcira model (in order to distinguish other peculiar circumstances, said that it is the general mcira model） be: μeμμjj (n)′== )( (7) where µ is a weight vector. lemma:lemma:lemma:lemma:[11] if nξξξ ,,, 21 ⋯ are n different (n-dimension) weight vectors, and constants nααα ,,, 21 ⋯ satisfy 01 1 >=∑ = i n i i ,αα , then ∑∑ == < n i ii n i ii ξjαξαj 11 )()( (8) or ∑∑∑ === <′ n i i(n)ii n i ii(n) n i ii ξeξαξαeξα 111 )()()( theoremtheoremtheoremtheorem 3:3:3:3: if µ = ),,( 21 ′nμ,μμ ⋯ is a weight vector, and 01 1 >=∑ = i n i i ,μμ , then ∑ = <′== n i ii(n) jμμeμμjj 1 ）（ . (9) in other words, the errors square sums of the general mcira model don’t exceed the weight sum of errors square sums of each investment risk assessment models in the combination. corollarycorollarycorollarycorollary 2:2:2:2: if µ is any nonnegative weight vector, then i) max(n) jμeμj <′= (10) ii) max n i ia jjn j ≤< ∑ =1 1 (11) form the above results, we know that the errors square sum of the general mcira model doesn’t exceed the maximum of errors square sum of each model in the combination. we can reduce the investment risk assessment errors by combining assessment model. the general mcira model's errors that showed by errors square sum are bounded above, and the superbound is maxj . 4.4.4.4.anananan exampleexampleexampleexample considering a general mcira model: the reciprocal variance weight multistage combined investment risk assessment model, its weight vector is: published by atlantis press copyright: the authors 108 zhou zongfang and yu jike ) 1 ,, 1 , 1 ( 1 ),,,( 21 1 21 ′=′= ∑ = n n i i n jjjj μμμμ ⋯⋯ , from theorem 3, its errors square sum j* satisfies inequation as below: ) 111 ( 1 1 2 2 1 1 1 1 n n n i i n i ii * j j j j j j j jμj +++=< ∑ ∑ = = ⋯ cn i i m j n == ∑ =1 1 where mc is called the harmonic mean of j1, j2 , …,jn. consequently the errors square sum j* of the reciprocal variance weight model is smaller then the harmonic mean mc of each investment risk assessment model in the combination. in particular, if ji >0, i=1,2,…,n, since the harmonic mean don’t exceed the arithmetic average value, we express: max n i ic * jj n μj ≤≤< ∑ =1 1 (12) the above inequation is just inequation (11). 5.5.5.5. conclusionsconclusionsconclusionsconclusions in this paper, we have estimated the errors boundary of the simple average method and the general mcira models, and indicated that the errors is the bounded to the above theoretically. we have also applied the mathematics analysis techniques to determine the existent maximum of the errors square sum of the general mcira models. this article research to carries on the appraisal accurately to the investment risk, dodges the investment risk effectively, has the very important theory and the practical significance. the research conclusion is also suitable for the general combination risk assessment domain, such as combination credit risk assessment[12], insurance risk assessment, disaster risk assessment and so on. we certainly believe that these studies will be of a great significant theoretical value and potential practical significance in the risk management domain. referencesreferencesreferencesreferences 1. j.m.bates and c.w.j.granger, combination of forecasts, journal of forecasting, 20(4)(1969)451-468. 2. j.i.munoz; j.contreras; j.caamano and p.f. correia, risk assessment of wind power generation project investments based on real options, powertech, 2009 ieee bucharest, june 28-july 2(2009). 3. jun sun,wei fang etc, solving the multi-stage portfolio optimization problem with a novel particle swarm optimization, expert systems with applications: an international journal, 6(23)( 2011) 67266735. 4. g.q.liu, z.f.zhou and y.shi, a multi-dimensional forward selection method for firms' credit sale ， computers & mathematics with applications, 54(2007)1228-1233. 5. z.f.zhou,t.y. mu and y.shi, the mathematical structure on credit evaluation, fur east journal of applied mathematics,1(20)(2005)113-119. 6. x.w. tang, z.f.zhou and y.shi, multi-objective constrained nonlinear optimization: an ode approach, international journal of information, 7(4)(2004)487-497. 7. z.f.zhou and x.w.tang, the research on ordering structure of credit, journal of systems science and information, 3(2)(2004)531-534. 8. x.w. tang, z.f.zhou and y.shi, the research of variable weights in combined forecasting, computers & mathematics with applications, 45(2003)723-730. 9. z.f.zhou and x.w.tang, the error estimates of continuous variable weight combined forecasting, journal of systems science and information, 3(1)(2003)235-242. 10. x.w. tang, z.f.zhou and y.shi, the errors bounds of combined forecasting, journal of mathematical and computer modelling, 36(2002) 997-1005. 11. x.w.tang, forecast theory with its applications, (university of electronic science & technology of china publisher,1992). 12. z.f.zhou,y.zhang and l. chen, emerging technology enterprise credit risk evolution mechanism and evaluation method,(science press,beijing,2010). published by atlantis press copyright: the authors 109 http://ieeexplore.ieee.org/xpl/mostrecentissue.jsp?punumber=5271856 results.cfm?query=name: results.cfm?query=name: 1.introduction 2.theerrorsboundsofsimpleaveragemodel 3.theerrorsestimateofthegeneralmciramodels microsoft word volume 12, issue 2-2 journal of risk analysis and crisis response, 2022, 12(2), 77-87 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i2.326 77 article-infus 2021 special issue machine learning sales forecasting for food supplements in pandemic era funda ahmetoğlu taşdemir 1,* 1 department of industrial engineering, izmir university of economics, izmir (35330), izmir province, turkey * correspondence: funda.tasdemir@izmirekonomi.edu.tr; tel.: +90-0232-488-8261 received: november 13, 2021; accepted: march 9, 2022; published: july 10, 2022 abstract: the covid-19 pandemic has brought a lot of concerns about the operational and financial situation of businesses. forecasting is crucial as it guides businesses through these critical points. forecasting has become even more critical in the pandemic environment and therefore the necessity of using an accurate forecasting method has increased. taking this into consideration, in this study, intelligent machine learning methods, namely; grey model (gm), artificial neural network (ann) and support vector machine (svm) are applied to make a short-term prediction of a food supplement, a product whose demand increased with the pandemic situation. eighty-five percent of the historical data is used for training purposes and fifteen percent of the data is used for measuring accuracy. the accuracy of the models employed is improved with parameter optimization the accuracy performance indicator mean absolute percentage error (mape) showed that all methods give superior results when the historical data has an increasing sales trend. this study presents an important consideration for businesses and has a potential to be generalized for a business whose sales have an increasing trend not only because of the pandemic but also for any reason. keywords: machine learning; grey model; artificial neural network; support vector machine; sales forecasting 1. introduction sales forecasting is essential for a company to make strategic decisions, especially to achieve better production planning, inventory control and financial estimation [1]. sales forecasts especially form the basis for supply policies, therefore poor forecasting causes too much or too little stocks, directly affecting competitiveness and revenue [2]. therefore, improving the accuracy of forecasts will provide better operational efficiency and financial savings. taking this into consideration, many algorithms have been developed to take the advantages of more accurate forecasts [3]. in many decades, statistical models have been widely applied in conducting forecasting. statistical models are effective and the study can be completed in a very short time even if the data set is large, but they may underperform with complex data pattern [4]. machine learning methods, which have been used more recently, are more efficient and flexible than traditional statistical techniques for forecasting since they have a better processing power [5]. tarallo et al. [6] made a comprehensive investigation for the machine learning techniques emphasizing their superiority to funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 78 traditional statistical techniques. many forecasting studies were conducted to evaluate and compare statistical and machine learning methods, both within themselves and with each other. ansuj et al. [7] applied autoregressive integrated moving average (arima) and ann methods to forecast sales in a medium-sized enterprise and found more accurate results with ann according to mean absolute error (mae) measurement. kotsialos et al. [8] compared holt-winter’s method and ann for sales forecasting of two german companies and found slightly better results with ann. tanaka [9] proposed a sales forecasting model with a knowledge-based database observing correlations for new-released and nonlinear sales trend products and results showed that their proposed model is superior to applied statistical methods; moving average (ma) and exponential smoothing (es) according to mape indicator. kandananond [10] applied ann, svm and arima to predict demand for six different consumer products and found out that for each product, svm performs better than ann and arima according to mape measurement. choi et al. [11] examined ann, ann+gm hybrid model and gm to forecast color trend of a fashionable product by employing real sales data and found that the latter gave best mape value. yu et al. [12] used svm for newspaper/magazine sales forecasting and their experiment showed svm is very successful in this kind of task. du et al. [13] used svm and ann for forecasting demand of perishable farm products and results show that svm outperformed ann. kitapçı et al. [14] forecasted automobile sales with linear regression (lr) and ann; the mean square error (mse) and mape measurements showed ann was more successful than lr. xia and wong [15] established a seasonal discrete gm for onestep-ahead forecasting sales of fashion products using real sales data of three different retailers and compared their model with ann, autoregressive (ar), gm (1,1), markov gm (1,1), tseng’s model and fuzzy grey regression model; and they found the lowest mse and mape with their proposed model. pillo et al. [16] applied svm and the statistical methods; es, holt-winter’s es (hwes) and arima to forecast sales of a commodity in retail and they got the lowest mse value from svm, suggesting any sales manager to consider using the svm method rather than conventional methods. vhatkar and dias [17] forecasted the sales of three different types of oral-care goods with ann. they used mean absolute deviation (mad), mse and root mean square error (rmse) performance measurements and concluded that ann gave accurate results. karmy and maldonado [18] applied svm for forecasting sales of duty free and travel retail industry. the mape indicator showed that svm gave higher forecasting accuracy compared to arima and holt-winter methods applied. güven and şimşir [19] used ann and svm for sales forecasting of different products in retail garment sector and according to mse, both methods yielded good results, with ann slightly more successful. zhang et al. [20] forecasted the monthly sales of cigarettes by applying arima, seasonal arima, winter’s multiplicative model and their own model in which they take seasonal and trend factors into consideration. their proposed model gave the lowest error percentage. taşdemir and şeker [21] applied gm (1,1), ann and fuzzy time series to find one-step ahead forecast of a cleaning product and the ape measurement showed that all the models gave superior results. based on the literature review above, machine learning methods have been very successful in sales forecasting. obviously, the covid-19 pandemic has changed the sales tendencies of some products in a positive way and change in trends has disrupted the forecasting process. to minimize the effects of this change, the necessity of the reliable methods increased. therefore, this study has aimed to make a short-term forecasting for the sales of a food supplement with three reliable machine learning methods, namely; gm, ann and svm. although there are important studies in the funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 79 literature on sales forecasting with gm, not much research has been conducted on comparison of gm with ann and svm methods together. this study contributes to the literature with comparative application of these 3 models. in addition, all models have been improved by optimizing their parameters before the final comparison is made. the underlying challenge of the task was utilizing these methods with limited historical data and this is especially may not be a good option for ann, however, this problem can be reduced to some extent by parameter optimization. the remainder of this paper is structured as follows: section 2 describes the methodology of the methods applied to make short-term forecasting and evaluation measurement used for the accuracy of these methods. section 3 describes the application of the methods with optimized parameters in details and summarizes the outputs. section 4 presents the conclusion for the whole study. 2. methodology 2.1. grey theory the grey system model is first proposed by deng [22] and gm (1,1) is the simplest form of this model which is widely used for forecasting. gm (1,1) requires at least four observations and has high forecasting capability with small samples. in gm (1,1), an accumulated generation operation (ago) is applied to construct differential equations to get the forecasted output from the system [23]. the gm (1,1) formulation is given in the following [24]: step 1: consider an original series to be x (0) = x (0)(1), x (0)(2), x(0)(3),…, x(0)(n) (1) step 2: a new sequence of x(1) series is obtained using the ago. x(1) = (x(1)(1), x(1)(2), x(1)(3),…, x(1)(n)) (2) 𝑥 ( )(𝑘) = ∑ 𝑥 ( )(𝑖) (3) step 3: a first-order differential equation is generated as follows: (dx (1)/dt) + αz=v (4) z(1)(k) =αx(1)(k) +(1-α)x(1) (k+1), k=1, 2, n-1 (5) α indicates a horizontal adjustment coefficient where 0 < α < 1. the selection criterion of α value is targeted to obtain the minimum forecasting error. step 4: 𝑥 ( )(𝑘 + 1) = 𝑥 ( )(1) − 𝑒 + (6) 𝜃 = 𝑎 𝑢 = (𝐵 𝐵) 𝐵 𝑌 (7) 𝐵 = ⎣ ⎢ ⎢ ⎡ −𝑧( )(2) −𝑧( )(3) … −𝑧 ( )(n) 1 1 … 1 ⎦ ⎥ ⎥ ⎤ (8) y = (x (0)(2), x (0)(3), …, x(0)(n)) t (9) step 5: inverse ago. since the grey forecasting model is conducted by applying ago data instead of the original data, iago can be employed to reverse the forecasting value as given in the following: 𝑥 ( ) = (𝑥 ( )(𝑘) − 𝑥(𝑘 − 1), 𝑘 = 2,3, … , 𝑛 (10) funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 80 2.2. artificial neural network artificial neural networks are computational models inspired by human brain and are used in forecasting problems [11]. a common ann structure consists of some elements called input, hidden and output layers. information is provided in the input layer, while the output layer produces the forecasting results. the hidden layer between those two layers establishes nonlinear relationships between inputs and outputs by adjusting weights which is called learning. the general structure of an ann is given in figure 1. figure 1. this structure of ann. a special type of feedforward artificial neural network is multilayer perceptron (mlp) which is employed in this study. mlp consists of one input and one output layer and one or more hidden layers [25]. mathematical calculation of mlp is given by following formula: y= 𝑓 { 𝑤 𝑓 ∑ ℎ 𝑋 + 𝑏 + 𝑏 } (11) where y is the output, x is the input vector, hij is the matrix of weight, bj is the vector of bias, and fh is the activation function of hidden layer, wj is weight vector, b0 is the bias scalar and f0 is activation function of output layer [26]. 2.3. support vector machine the support vector machine was introduced by vapnik [27] and it is one of the machine learning techniques used both for classification and regression. although they were underestimated in the past, currently they can yield successful results which are comparable to those of anns and other intelligent methods [28]. svm employs a linear function in a large space to test the forecasting regression. in svm, forecasting is calculated by following formula: f(x)=wtx+b (12) where x is the input vector, b is bias and w is the weight vector [26]. an iterative algorithm for solving regression problems is called sequential minimal optimization (smo) and later improved by shevade et al. [29] to solve regression problems which named smoreg [30]. smoreg is the function which will be the scope of this study. 2.4. evaluation of accuracy funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 81 one of the robust forecasting accuracy measurements is mape, which is proposed by lewis [31]. according to lewis, if the mape value is less than or equal to 10%, forecasting accuracy is evaluated high. the formulation for mape is given as follows: 𝑀𝐴𝑃𝐸 = ( ) ( ) ( ) × 100% (13) where y(i) is the actual value and 𝑦(𝑖) is the predicted value of y for period i [32]. 3. application and discussions this study focuses on a food supplement whose sales have increased due to the pandemic and aims to provide an accurate forecast of the relevant product. the company never experienced any increase like this; especially an unprecedented one like amid pandemic, therefore the necessity of an accurate forecasting system has become much more critical than ever. to employ in the forecasting models, the original data was extracted from the company’s data base and the data starts from march 2019, when the first covid case was seen in turkey and ends in june 2020 which creates a total of 16 observations. based on this data set, it is aimed to make a short-term forecast for the following 3 months. the gm (1,1), ann and svm methods mentioned in section 2 were used to perform the estimation and mape is used for evaluating the performance of these methods. by analyzing the results of these forecasting models, the company will decide whether to replace their legacy forecasting method since it is not satisfactory. 3.1. gm (1,1) model results according to the formulations steps used in gm (1,1) given in section 2.1, the model is established using the original sales data of the food supplement. step 1: the original series belonging to related product are x(0) = {13280, 15045, …, 32870}. step 2: x(1) cumulative series are then calculated as x(1) = {13280, 28865, …, 366806}. step 3: the calculations showed that the best value for α is 0.16. −26458 −43215 . . −361547 1 1 … 1 , 𝑌 = 15045 17083 … 32870 , 𝜃 = (𝐵 𝐵) 𝐵 𝑌 = −0,0525 14353,1 step 4: when we take 𝜃 into consideration: 𝑥 ( ) (𝑘) = 𝑥 ( )(1) − 𝑢 𝑎 (1 − 𝑒 )𝑒 ( ) the equation of gm (1,1) becomes: 𝑥 ( ) = 13280 − 143523,1 −0,0525 (1 − 𝑒 , )𝑒 ( , )( ), 𝑘 = 2,3, … , 𝑛, 𝑛 + 1 … for next 3 months, gm (1,1) gives the following forecast values respectively: 34080, 35290, 37860. 3.2. mlp results the mlp model is established to forecast the 3 months ahead sales of food supplement product in waikato environment for knowledge analysis (weka) 3.8.5 software. the software is developed in java by university of waikato for machine learning studies. when the software is run, a screen with five different interfaces is displayed. the software also has a package manager where it is funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 82 possible to access various tools [28]. forecast plugin is one of these tools which is installed to apply mlp and smoreg models in this study (figure 2). figure 2. weka forecasting plug-in. mlp models have some parameters to be optimized to reach best accuracy. these parameters are number of epochs (n), number of hidden layers, number of neurons, learning rate (l) and momentum (m). to train mlp, 85% of the data is used and several experiments have been applied to minimize mape. table 1 shows the list of experiments and the obtained mape value accordingly. the first experiment shows the default parameters in weka. firstly, the experiment was started with the first three trials of epoch number, and it was observed that the forecasting accuracy decreased due to the overfit of the model when the epoch number increased too much. when the number of epochs is reduced, the forecasting accuracy decreased again due to the inability of the model to learn enough. in the first, fourth and fifth experiments, other parameters were kept constant and the number of neurons was changed, and it was found that these changes decreased the forecasting accuracy again. the first and sixth experiments were examined to see how the number of hidden layers affected the prediction accuracy, and the results show that increasing the number of hidden layers from one to two significantly reduced the accuracy. in the transition from the sixth experiment to the seventh experiment, it is shown that the accuracy decreased when the number of neurons is decreased again while the number of hidden layers is still two. in the first, eighth and ninth experiments, the learning rate was changed and other parameters were kept constant. accuracy increased when the learning rate was 0.2, unlike the default value. finally, in the first, tenth and eleventh experiments, the momentum was changed and the other values were kept constant, and it was observed that the accuracy increased when the momentum value was 0.3. based on the experiments conducted, the best value for mape is obtained with 500 epochs, 1 hidden layer, 10 neurons, 0.3 learning rate and 0.3 momentum as shown in table 1. figure 3 shows the actual sales between period 1 and 19 and the forecasted sales for period 17, 18 and 19 for the best set of parameters in mlp. the forecasted values are 34080, 35751 and 37132 respectively while the actual values are funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 83 34067, 35789 and 37893. it is revealed from figure 3 that forecast values of the mlp are close to actual sales values. table 1. mlp parameters and corresponding mape values. experiment number number of epoch number of hidden layer number of neurons learning rate momentum mape 1 500 1 10 0.3 0.2 0.86% 2 1000 1 10 0.3 0.2 1.43% 3 100 1 10 0.3 0.2 1.56% 4 500 1 5 0.3 0.2 1.41% 5 500 1 15 0.3 0.2 1.34% 6 500 2 10 0.3 0.2 4.01% 7 500 2 5 0.3 0.2 4.10% 8 500 1 10 0.4 0.2 1.47% 9 500 1 10 0.2 0.2 1.03% 10 500 1 10 0.3 0.3 0.72% 11 500 1 10 0.3 0.4 0.81% figure 3. actual sales and forecasted sales of mlp. 3.3. smoreg results the regression algorithm of svm, which is called smoreg is also established in weka environment. as in mlp, the same percentage in smoreg is used for training purposes. in addition to this, smoreg has also key parameters to be optimized, such as kernel and complexity (c) parameters. kernel function determines how to allocate data in the future space and c determines the flexibility of the line to separate different classes [33]. kernel function can be selected in weka interface and the c parameter must be set by the user. the best setting of the parameters can only be found by experimentation [34]. the kernels used in the experiments are normalized polynomial (normalized poly), polynomial (poly), pearson vii (puk) and radial basis function (rbf). the c values applied are, 1 (the default parameter in weka), 2, 0.5 and 3. table 2 shows 16 experiments performed with different kernel and c parameters. the best mape value is obtained with poly 13000 18000 23000 28000 33000 38000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 s al es periods actual sales forecasted sales funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 84 kernel and c = 3. figure 4 shows the actual sales for periods 1-19 and the forecast results for period 17, 18 and 19 for the best experiment in smoreg. smoreg forecasted 34460, 35993 and 37854 respectively against actual values of 34067, 35789 and 37893. figure 4 shows that smoreg yields estimated values that approximate actual sales values. table 2. mlp parameters and corresponding mape values. experiment number kernel function c mape 1 normalized poly 1 26.78% 2 poly 1 2.91% 3 puk 1 28.78% 4 rbf 1 1.30% 5 normalized poly 2 19.12% 6 poly 2 1.46% 7 puk 2 28.78% 8 rbf 2 1.11% 9 normalized poly 0.5 36.14% 10 poly 0.5 3.28% 11 puk 0.5 28.77% 12 rbf 0.5 15.96% 13 normalized poly 3 11.97% 14 poly 3 0.85% 15 puk 3 28.78% 16 rbf 3 1.13% figure 4. actual sales and forecasted sales of smoreg. 3.4. forecasting accuracy measurement for evaluating the forecasting accuracy of the three methods mentioned above, 15% of the data patterns is not used in the models for testing purpose and the output of the models are compared with the actual value to calculate mean absolute percentage error (mape). table 3 shows the shortterm forecasting accuracy of the applied methods based on mape measurement mentioned in section 2.4. the table is a summary of best experiments performed in each method. the mape rates are 0.16%, 13000 18000 23000 28000 33000 38000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 s a le s periods actual sales forecasted sales funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 85 0.72% and 0.85% for gm (1,1), mlp and smoreg respectively. according to results gm (1,1) outperformed mlp and smoreg models by providing lower mape value. the superiority of the gm (1,1) is not an unexpected result as it gives successful results with a small amount of data with increasing trend. however, the high forecasting accuracy of ann vs svm shows that these methods can also be applied in cases where there is little historical data with trend. table 3. mape value of gm (1,1), mlp and smoreg. models mape gm (1,1) 0.16% mlp 0.72% smoreg 0.85% 5. conclusions in recent years sales forecasting has been critical for determining major activities and strategic decisions of companies. the sales volumes of many companies are likely to be positively or negatively affected by extraordinary circumstances. the pandemic has been an unprecedented example of these extraordinary situations and people started to pay much more attention to their health than ever before, and therefore, they began to buy products that will boost their health. the increased purchase of these products has positively impacted some specific food industries. the company worked in this paper operates in the referred food industry and requires a guiding method to deal with their increasing sales. in the light of above, a short-term estimate was made for a food supplement product of the company. gm (1,1), ann and svm are assessed as sales forecasting tools, which are known as intelligent machine learning methods. all the models have been developed with their own optimization parameters. ann and svm experiments were conducted in weka environment. the accuracies of the employed methods were compared using the mape criteria. the mape results show that all the models gave superior results, with the gm (1,1) slightly better than ann and svm. the challenge of this study is that due to the novelty of the pandemic situation, very little historical data is available. therefore, it is not surprising that the gm (1,1), which gives very successful results with limited data is superior one. however, it should not be ignored that ann and svm also gave very successful results. while forecasting is essential for decision makers in normal conditions, more robust methods is required in existence of extraordinary situations like pandemic. this paper examines the usability of three intelligent forecasting models by employing increasing sales data of the relevant company affected by pandemic. the above-mentioned findings show that the company can replace the legacy forecasting method with these intelligent forecasting methods, especially with gm (1,1). for further studies, the proposed models can be applied also for forecasting other products in different fields such as personal care products and organic products, which have an increasing sales trend due to pandemic or any other reason. in this way, companies will benefit from accurate results of models, holding too much or too little inventory can be prevented and, as a result, costs can be reduced. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 86 references [1] wang, c-h., yun, y. demand planning and sales forecasting for motherboard manufacturers considering dynamic interactions of computer products. computers & industrial engineering 2020, 149. doi: https://doi.org/10.1016/j.cie.2020.106788. [2] chu, c-w., zhang, g.p. a comparative study of linear and nonlinear models for aggregate retail sales forecasting. int. j. production economics 2003, 86, 217–231. doi: https://doi.org/10.1016/s0925-5273(03)000689. [3] abolghasemi, m., hurley, j., eshragh, a., fahimnia, b. demand forecasting in the presence of systematic events: cases in capturing sales promotions. int. j. production economics 2020, 230. doi: https://doi.org/10.1016/j.ijpe.2020.107892. [4] yu, y., choi, t-m., hui, c-l. an intelligent fast sales forecasting model for fashion products. expert systems with applications 2011, 38, 7373–7379. doi: https://doi.org/10.1016/j.eswa.2010.12.089. [5] tsoumakas, g. a survey of machine learning techniques for food sales prediction. artif intell rev 2019, 52, 441–447. doi: https://doi.org/10.1007/s10462-018-9637-z. [6] tarallo, e., akabane, g., shimabukuro, c.i., mello, j. amancio, d. machine learning in predicting demand for fast-moving consumer goods: an exploratory research. ifac papersonline 2019, 52, (13), 737–742. [7] ansuj, a.p., camargo, m.e., radharamanan, r., petry, d.g. sales forecasting using time series and neural networks. computers ind. engng 1996, 31, 421-424. [8] kotsialos, a., papageorgiou, m., poulimenos, a. holt-winters and neural-network methods for mediumterm sales forecasting. proceedings of 16th triennial world congress, czech republic, elsevier ifac publications 2005, 133-138. [9] tanaka, k. a sales forecasting model for new-released and nonlinear sales trend products. expert systems with applications 2010, 37, 7387-739. doi: https://doi.org/10.1016/j.eswa.2010.04.032. [10] kandananond, k. a comparison of various forecasting methods for autocorrelated time series. international journal of engineering business management 2012, 4, (4). doi: https://doi.org/10.5772/51088. [11] choi, t-m, hui, c-l, ng, s-f., yu, y. color trend forecasting of fashionable products with very few historical data. ieee transactions on systems, man, and cybernetics, part c (applications and reviews), 2012, 42, (6), 1003-1010. [12] yu, x., zhiquan, q., yuanmeng, z. support vector regression for newspaper/magazine sales forecasting. procedia computer science 2013, 17, 1055-1062. doi: https://doi.org/10.1016/j.procs.2013.05.134. [13] du, x-f., leung, s-c-h., zhang, j-l, lai, k-k. demand forecasting of perishable farm products using support vector machine. international journal of systems science 2013, 44, (3), 556-567. doi: https://doi.org/10.1080/00207721.2011.617888. [14] kitapcı, o., özekicioğlu, h., kaynar, o., taştan, s. the effect of economic policies applied in turkey to the sale of automobiles: multiple regression and neural network analysis. procedia-social and behavioral sciences 2014, 148, 653-661. doi: https://doi.org/10.1016/j.sbspro.2014.07.094. [15] xia, m., wong, w.k. a seasonal discrete grey forecasting model for fashion retailing. knowledge-based systems 2014, 57, 119-126. doi: http://dx.doi.org/10.1016/j.knosys.2013.12.014. [16] di pillo, g., latorre, v., lucidi, s., procacci, e. an application of support vector machines to sales forecasting under promotions. 4or-qj oper res 2016, 14, 309-325. doi: https://doi.org/10.1007/s10288-0160316-0. [17] vhatkar, s., dias, j. oral-care goods sales forecasting using artificial neural network model. procedia computer science 2016, 79, 238-243. doi: https://doi.org/10.1016/j.procs.2016.03.031. [18] karmy, j.p., maldonado, s. hierarchical time series forecasting via support vector regression in the european travel retail industry. expert systems with applications 2019, 137, 59-73. doi: https://doi.org/10.1016/j.eswa.2019.06.060. [19] güven, i̇., şimşir, f. demand forecasting with color parameter in retail apparel industry using artificial neural networks (ann) and support vector machines (svm) methods. computers & industrial engineering 2020, 147. doi: https://doi.org/10.1016/j.cie.2020.106678. [20] zhang, mu., huang, x-r., yang, c-b. a sales forecasting model for the consumer goods with holiday effects. journal of risk analysis and crisis response 2020, 10, (2), 69-76. doi: https://doi.org/10.2991/jracr.k.200709.001. funda ahmetoğlu taşdemir / journal of risk analysis and crisis response, 2022, 12(2), 77-87 doi: https://doi.org/10.54560/jracr.v12i2.326 87 [21] ahmetoğlu taşdemir, f., şeker, s. fuzzy time series and grey theory forecasting for the sales of cleaning products. international conference on intelligent and fuzzy systems, 2021, istanbul, 24-26.08.2021, turkey, springer, 641-648. [22] deng, j.-l. control problem of grey system. systems and control letters 1982, 1, (5), 288-294. doi: https://doi.org/10.1016/s0167-6911(82)80025-x. [23] ene, s., öztürk, n. grey modelling-based forecasting system for return flow of end-of-life vehicles. technological forecasting and social change 2017, 115, 155-166. doi: http://dx.doi.org/10.1016/j.techfore.2016.09.030. [24] wang, c-h. predicting tourism demand using fuzzy time series and hybrid grey theory. tourism management 2004, 25, 367-374. doi: https://doi.org/10.1016/s0261-5177(03)00132-8. [25] toğa, g., atalay, b., toksari, m.d. covid-19 prevalence forecasting using autoregressive integrated moving average (arima) and artificial neural networks (ann): case of turkey. journal of infection and public health 2021, 14, 811-816. doi: https://doi.org/10.1016/j.jiph.2021.04.015. [26] ballı, s. data analysis of covid-19 pandemic and short-term cumulative case forecasting using machine learning time series methods. chaos, solitons & fractals 2021, 142. doi: https://doi.org/10.1016/j.chaos.2020.110512. [27] vapnik, v., chervonenkis, a. a note on one class of perceptrons. automation and remote control 1964, 25. [28] merlini, d., rossini, m. text categorization with weka a survey. machine learning with applications 2021, 4. doi: https://doi.org/10.1016/j.mlwa.2021.100033. [29] ballı, s. data analysis of covid-19 pandemic and short-term cumulative case forecasting using machine learning time series methods. chaos, solitons & fractals 2021, 142. doi: https://doi.org/10.1016/j.chaos.2020.110512. [30] shevade, s., keerthi, s., bhattacharyya, c., murthy, k. improvements to the smo algorithm for svm regression. ieee trans. neural networks 2000, 11, (5), 1188-1193. [31] meenal, r., selvakumar, a. i. assessment of svm, empirical and ann based solar radiation prediction models with most influencing input parameters. renewable energy 2018, 121, 324-343. doi: https://doi.org/10.1016/j.renene.2017.12.005. [32] lewis, c.d. industrial and business forecasting methods. butterworths-heinemann, london,1982. [33] jena, p. r., majhi, r., kalli, r., managi, s., majhi, b. impact of covid-19 on gdp of major economies. application of the artificial neural network forecaster, economic analysis and policy 2021, 69, 324-339. doi: https://doi.org/10.1016/j.eap.2020.12.013. [34] villavicencio, c. n., macrohon, j. j. e., inbaraj, x. a., jeng, jh., hsieh, jg. covid-19 prediction applying supervised machine learning algorithms with comparative analysis using weka. algorithms 2021, 14. doi: https://doi.org/10.3390/a14070201. [35] witten, i. h., frank, e., hall, m. a., pal, c. j. data mining practical machine learning tools and techniques, elsevier, 4th edition, 2017. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 川、陕、甘三省大熊猫分布区犬瘟热风险评估 risk assessment of canine distemper in the distribution area of giant panda in sichuan, shaanxi and gansu provinces, china weigeng shao1, feng jiang2,3, liya huang4, xiaolong wang2,3,* 1heilongjiang academy of forestry and environment, qiqihar 161005, china 2college of wildlife resources, northeast forestry university, harbin 150040, china 3center of conservation medicine & ecological safety, northeast forestry university, harbin 150040, china 4changbai mountain academy of sciences, antu 133613, china abstract giant panda is the world-class precious endangered species, facing the canine distemper and other important infectious diseases on its wild and captive population of a serious threat. in this study, we used maxent model and combined with arcgis analysis to predict the potential risk of canine distemper to giant panda habitat in sichuan, gansu and shaanxi provinces, china. the results showed that 35.05% and 19.47% of the distribution areas of the giant pandas were in the high risk and medium risk of canine distemper, respectively. the canine distemper pose a great risk to the healthy survival of giant pandas in china. in future, epidemic prevention, vaccine development and application of wild animals should be enhanced so as to effectively protect the giant panda. keywords: giant panda, canine distemper, maxent model, risk assessment 川、陕、甘三省大熊猫分布区犬瘟热风险评估 邵伟庚 1 ，江峰 2,3 ，黄利亚 4 ，王晓龙 2,3,* 1.黑龙江省森林与环境科学研究院，齐齐哈尔 161005，中国； 2.东北林业大学野生动物资源学院，哈尔滨 150040，中国； 3.东北林业大学保护医学与生态安全研究中心，哈尔滨 150040，中国； 4.长白山科学研究院，安图 133613，中国 摘要：大熊猫是世界级珍贵濒危物种，正面临犬瘟热等重要传染病对其野生和圈养种群的严重威 胁。本研究采用 maxent 模型并结合 arcgis 分析，预测犬瘟热对我国四川、甘肃和陕西大熊猫 栖息地潜在风险。结果显示，35.05%和 19.47%的大熊猫活动区分别处于犬瘟热高风险和中风险 威胁之下，犬瘟热对我国大熊猫健康生存构成极大的风险。因此应加强对大熊猫活动区域内野生 动物疫情防范，提升野生动物的疫苗研制与应用，以有效保护大熊猫。 关键词：大熊猫，犬瘟热，maxent 模型，风险评估 * corresponding author: e-mail: yttuhh@yeah.net. post address: college of wildlife resources, northeast forestry university, harbin 150040, china journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 225–229 ___________________________________________________________________________________________________________ 225 received 3 september 2017 accepted 10 november 2017 copyright © 2017, the authors. published by atlantis press. this is an open access article under the cc by-nc license (http://creativecommons.org/licenses/by-nc/4.0/). 大熊猫（ailuropoda melanoleuca）是我国特有 的生物多样性保护旗舰物种 [ 1] ，目前野生种群数量 为 1864 只，栖息地面积为 258 万公顷[2]，分布于川、 陕、甘三省的六大山系 [ 3] 。自 1983 年至 2015 年， 我国至少有 12 只大熊猫感染犬瘟热（ canine distemper, cd），其中 10 只死亡[ 4, 5, 6]，从而引发关 注。cd是由犬瘟热病毒（canine distemper virus, cdv）引发的一种急性、烈性的病毒性传染病，动 物感染后致死率极高，可达 100%[7]。cdv全球均有 分布，无特定宿主屏障，可感染多种动物 [ 8, 9, 10] ，也 是引发大熊猫感染性腹泻的主要病原体 [ 11] 。野生大 熊猫异常珍贵、数量稀少，踪迹很难发现，开展大 规模野外研究困难。因此本研究采用目前广泛使用 的maxent (maximum entropy)模型[ 12, 13]，结合地理 信息系统分析大熊猫栖息地内的犬瘟热空间分布及 风险情况，为大熊猫保护对策的制定提供理论依据。 1. 材料与方法 1.1 研究区域 研究区域位于川、陕、甘三省的岷山、邛崃、 大相岭、小相岭、凉山和秦岭这六大山系的大熊猫 的 栖 息 地 ， 地 理 位 置 为 e101.85°~108.799° ， n28.2°~34.01°之间，行政区域包括 45 个县（市， 区） [ 14] 。 1.2 川、陕、甘三省 cd 空间分布分析 （1）cd 样本数据环境数据的收集和处理 在 ncbi 的基因库（genbank）以及世界动物 卫生组织（oie）等网站中收集 cd 地理坐标信息， 以 excel 表格存储为".csv" 格式数据。从世界气象 数据库（global climate date）下载 68 个空间分辨 率为 30arc-seconds 的环境因子数据，并转换成 maxent 软件所需要的".asc"格式。 （2）maxent 模型运算及处理 maxent模型是由steven phillips等人基于最大 熵原理开发的一种生态位模型，它是通过物种已知 实际地理分布点和限制该物种分布的环境变量等约 束条件，并建立二者之间的关系模型，探寻在环境 变量等约束条件下的最大分布，熵达到最大时即为 该物种出现的概率分布作为物种实际分布 [ 15, 16] 。因 此maxent模型可以评价或预测物种给定环境等约 束条件下适宜性潜在分布区域 [ 17] ，并且该模型在濒 危动植物 [ 18, 19] 、病虫害、入侵物种 [ 20] 、经济作物等 方面均有相关研究。而模型给出的auc（areas under receiver operator characteristic curves)值越接近 1，则 表 明 预 测 结 果 越 好 。 使 用 maxent 3.3.3k 版 本 （http://www.cs.princeton.edu/~schapire/maxent/），将 采集的数据按测试集 25%和训练集 75%随机分组， 选择刀切法，重复次数为 10，其他参数为默认值。 运算结果采用arcgis 10.2 显示，叠加川、甘、陕三 省的包括省、市、县（区）行政区划图层，剪裁后 分析。 1.3 大熊猫分布区 cd 空间分布分析 利用 arcgis 10.2 软件将大熊猫主要栖息地区 域叠加到川、甘、陕三省 cd 分布图中，沿区域边 界线剪裁并分析。打开大熊猫栖息地区域属性表， 在“property”中选择添加面积（area）字段，并在 “calculate system”中选择投影坐标系统(pcs)，把 前述栖息地区域属性表复制、导入 excel 中。 通过软件中“数据框属性”经纬度网格将栖息 地均分为最小单元格（0.1°× 0.1°），并提取单元格 中 0（蓝）~1（红）的属性值，该属性值即为 cd 的概率值，值越大（越红）则表明该地区的风险越 高。将该属性值数据导出，复制入 excel 表格中， 然后统计大熊猫栖息地中六大山系最小单元格个数 以及属性值为高、中、低的单元格个数，最后计算 各个山系不同风险的面积以及风险比例。 2 结果分析 2.1 川、陕、甘三省 cd 风险分析 川、陕、甘三省整体 cd 风险水平呈现为从东 南向西北逐级降低的趋势，四川省高风险区域面积 较大（图 1）。四川省中部和东部（e102.31°~106.44°, n28.61°~31.22° ）、 北 部 局 部 （ e103.92°~105.28°, n31.79°~32.66° ）， 以 及 甘 肃 省 东 南 部 （e104.24°~105.71°, n32.62°~33.93°）处于 cd 高风 险区域。其他地区则分别处于 cd 中等风险至极低 风险概率区域。 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 225–229 ___________________________________________________________________________________________________________ 226 auc 检测显示，cd 的 auc 平均值为 0.949（图 2）。 2.2 大熊猫主要栖息地 cd 风险分析 大熊猫主要活动区域可分为 a、b 这 2 大区域 进行分析（图 3）。a 区域包括岷山、邛崃、大相岭、 小相岭、凉山，此区域 cd 风险同样呈现从东南向 西北逐级降低的趋势，其东部、东北部和东南部处 于 cd 高风险等级，高、中风险区域分别占总研究 区域面积的 35.05%和 13.78%（表 1）。b 区域包括 秦岭山系（陕西境内），cd 风险处于中等风险水平， 中风险区域占总研究区域面积的 5.96%。 图 1 川、陕、甘三省犬瘟热风险图 figure 1. the canine distemper risk diagram in sichuan, shanxi and gansu provinces 图 2 犬瘟热风险分布 roc 曲线 figure 2. roc curve of canine distemper risk distribution 注：roc 曲线是“受试者工作特征曲线”（receiver operating characteristic curve）的简称。 journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 225–229 ___________________________________________________________________________________________________________ 227 表 1 大熊猫分布六大山系风险等级面积百分比 table 1 risk rating area percentage of six mountain systems in giant panda habitat 山系 mountain system 高风险 high risk 中风险 medium risk 低至极低风险 low to extremely low risk 面积（km2） area 比例 proportion 面积（km2） area 比例 proportion 面积（km2） area 比例 proportion 岷 山 minshan 9421.32 7.09% 9889.61 7.45% 31180.06 23.48% 邛 崃 qiong lai 12047.08 9.07% 5511.74 4.15% 16106.54 12.13% 大相岭 daxiangling mountains 6117.47 4.61% 0.00 0.00 0.00 0.00 小相岭 xiaoxiangling mountains 6468.20 4.87% 1758.08 1.32% 6770.00 5.10% 凉 山 liangshan 12492.23 9.41% 1138.94 0.86% 0.00 0.00 秦 岭 qin ling 0.00 0.00 7555.20 5.69% 6361.80 4.79% 合 计 total 46546.28 35.05% 25853.58 19.47% 60418.40 45.49% 图 3 大熊猫主要活动区域内犬瘟热风险分析图 figure 3. analysis of canine distemper risk in the main active area of giant panda journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 225–229 ___________________________________________________________________________________________________________ 228 3. 讨论 结果研究显示，大熊猫栖息地中约 54.51%区域 处于 cd 高风险和中风险水平（图 3，表 2），整体 呈现为西北向东南方向 cd 风险强度逐渐增加的趋 势，与川、陕、甘三省整体风险趋势相同。 大熊猫栖息地约 85%的面积在四川境内，且野 生种群数量（1387 只）占全国总数的 74.4%[2]；四 川省 cd 风险等级高于甘肃和陕西省（图 1），且高 风险区域主要在四川境内，甘肃省东南部也处于高 风险区域（图 3，表 1），这表明 cd 对该区域大熊 猫的生境健康存在较高的潜在风险。 我国对大熊猫病毒性疾病的流行病学调查研究 尚显不足，家犬以及其它野生易感动物的cd流行情 况亦多不清楚，家犬免疫率偏低 [ 21] ，都给大熊猫保 护带来了无法克服的困难。 关于野生动物疫苗的研制及应用在我国还未成 为公共卫生的重点，对野生动物免疫研究投入并不 充足 [ 22, 23] ，野生动物的疫苗研制工作还需提升 [ 24] 。 因此在大熊猫活动区域内，cd预防工作应成为保护 大熊猫的重要内容。在大熊猫cd高风险区域，需要 采取有效的措施，例如可将大熊猫活动区域内的家 犬或肉犬进行强制免疫或者统一管理，加强大熊猫 疫苗研制工作，有效保护大熊猫。 致谢： 国家重点研发计划重点专项（2017yfd0501702）资助。 参考文献 [1] wen z, yang s, shan t, et al. virome comparisons in wild-diseased and healthy captive giant pandas. microbiome, 2017, 5(1):90. [2] http://www.forestry.gov.cn/ [3] 李俊清，申国珍，等．大熊猫栖息地研究．北京：高 等教育出版社，2012：17~18. [4] 李金中，夏咸柱，何洪彬，等．基因序列分析确诊大 熊猫的犬瘟热病毒感染．中国兽医学报，1999，19 ( 5 ) ：448~450． [5] 高玉伟，夏咸柱，鞠会艳，等．大熊猫血清犬瘟热病 毒中和抗体调查．第十一次全国养犬学术研讨会论 文集，2005：240~243． [6] 胡彦博，秦晓东，梁晚枫．大熊猫犬瘟热研究进展．吉 林畜牧兽医，2016，37(2)：12~15． [7] wyllie se, kelman m, ward mp. wyllie se, kelman m, ward mp. epidemiology and clinical presentation of canine distemper disease in dogs and ferrets in australia, 2006-2014. australian veterinary journal, 2016, 94(7), pp.215-22. [8] lempp c, spitzbarth i, puff c, et al. new aspects of the pathogenesis of canine distemper leukoencephalitis. viruses, 2014, 6(7):2571-601. [ 9 ] martinezgutierrez m, ruizsaenz j. diversity of susceptible hosts in canine distemper virus infection: a systematic review and data synthesis. bmc veterinary research, 2016, 12(1):1-11. [10] carpenter ma, appel m.j, roelke-parker m.e, et al. genetic characterization of canine distemper virus in serengeti carnivores. vet immunol immunopathol, 1998, 65 (2-4): 259−266. [11] feng n, yu y, wang t, et al. fatal canine distemper virus infection of giant pandas in china . scientific reports, 2016, 6:27518. [12] johnson t l, bjork j k h, neitzel d f, et al. habitat suitability model for the distribution of ixodes scapularis (acari: ixodidae) in minnesota. journal of medical entomology, 2016(3):3~8. [13] 王运生，谢丙炎，万方浩，等．roc 曲线分析在评 价入侵物种分布模型中的应用．生物多样性，2007， 15 (4) ：65 -372． [14] 周洁敏. 大熊猫生境质量评价体系研究[d]. 北京林 业大学, 2008. [15] phillips s j, dudik m. modeling of species distributions with maxent: new extensions and a comprehensive evaluation. ecography, 2008, 31(2): 161~175. [16] 邢丁亮，郝占庆．最大熵原理及其在生态学研究中 的应用．生物多样性，2011，(03)：295~302． [17] 刘振生，高惠，滕丽微，等．基于 maxent 模型的 贺兰山岩羊生境适宜性评价．生态学报，2013，（22）： 7243-7249． [18] 万基中，王春晶，韩士杰，等．应用 gis 和 maxent 预测南方红豆杉潜在地理分布及其保护建议．江苏 农业科学，2014，42 (7) ：349~352． [19] 马松梅，聂迎彬，耿庆龙，等．气候变化对蒙古扁 桃适宜分布范围和空间格局的影响．植物生态学报， 2014，38（7）：262~269． [20] 杨会枫，郑江华，吴秀兰，等．基于 maxent 模型的 西伯利亚蝗虫在新疆潜在分布预测研究．新疆农业 科学，2016，53（1）43~50． [21] 张菲，张守峰,唐青，等．我国狂犬病现状与防控意 见．中国人兽共患病学报，2016，26（4）：381~387． [22] 王磊，李天松，刘玉秀，等．犬瘟热新型疫苗研究 进展．动物医学进展，2011，32（2）99-103． [23] 夏咸柱．野生动物疫病与公共卫生．中国自然医学 杂志，2007，9（1）：72~77． [24] 夏咸柱，俞永新，侯云德，等．加强我国狂犬病防 控．中国病毒病杂志，2011，1（3）：161~162． journal of risk analysis and crisis response, vol. 7, no. 4 (december 2017) 225–229 ___________________________________________________________________________________________________________ 229 1. 材料与方法 2 结果分析 3. 讨论 << /ascii85encodepages false /allowtransparency false /autopositionepsfiles true /autorotatepages /none /binding /left /calgrayprofile (dot gain 20%) /calrgbprofile (srgb iec61966-2.1) /calcmykprofile (u.s. web coated \050swop\051 v2) /srgbprofile (srgb iec61966-2.1) /cannotembedfontpolicy /error /compatibilitylevel 1.4 /compressobjects /tags /compresspages true /convertimagestoindexed true /passthroughjpegimages true /createjobticket false /defaultrenderingintent /default /detectblends true /detectcurves 0.0000 /colorconversionstrategy /leavecolorunchanged /dothumbnails false /embedallfonts true /embedopentype false /parseiccprofilesincomments true /embedjoboptions true /dscreportinglevel 0 /emitdscwarnings false /endpage -1 /imagememory 1048576 /lockdistillerparams false /maxsubsetpct 100 /optimize true /opm 1 /parsedsccomments true /parsedsccommentsfordocinfo true /preservecopypage true /preservedicmykvalues true /preserveepsinfo true /preserveflatness true /preservehalftoneinfo false /preserveopicomments true /preserveoverprintsettings true /startpage 1 /subsetfonts true /transferfunctioninfo /apply /ucrandbginfo /preserve /useprologue false /colorsettingsfile () /alwaysembed [ true ] /neverembed [ true ] /antialiascolorimages false /cropcolorimages true /colorimageminresolution 300 /colorimageminresolutionpolicy /ok /downsamplecolorimages true /colorimagedownsampletype /bicubic /colorimageresolution 300 /colorimagedepth -1 /colorimagemindownsampledepth 1 /colorimagedownsamplethreshold 1.50000 /encodecolorimages true /colorimagefilter /dctencode /autofiltercolorimages true /colorimageautofilterstrategy /jpeg /coloracsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /colorimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000coloracsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000colorimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasgrayimages false /cropgrayimages true /grayimageminresolution 300 /grayimageminresolutionpolicy /ok /downsamplegrayimages true /grayimagedownsampletype /bicubic /grayimageresolution 300 /grayimagedepth -1 /grayimagemindownsampledepth 2 /grayimagedownsamplethreshold 1.50000 /encodegrayimages true /grayimagefilter /dctencode /autofiltergrayimages true /grayimageautofilterstrategy /jpeg /grayacsimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /grayimagedict << /qfactor 0.15 /hsamples [1 1 1 1] /vsamples [1 1 1 1] >> /jpeg2000grayacsimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /jpeg2000grayimagedict << /tilewidth 256 /tileheight 256 /quality 30 >> /antialiasmonoimages false /cropmonoimages true /monoimageminresolution 1200 /monoimageminresolutionpolicy /ok /downsamplemonoimages true /monoimagedownsampletype /bicubic /monoimageresolution 1200 /monoimagedepth -1 /monoimagedownsamplethreshold 1.50000 /encodemonoimages true /monoimagefilter /ccittfaxencode /monoimagedict << /k -1 >> /allowpsxobjects false /checkcompliance [ /none ] /pdfx1acheck false /pdfx3check false /pdfxcompliantpdfonly false /pdfxnotrimboxerror true /pdfxtrimboxtomediaboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxsetbleedboxtomediabox true /pdfxbleedboxtotrimboxoffset [ 0.00000 0.00000 0.00000 0.00000 ] /pdfxoutputintentprofile () /pdfxoutputconditionidentifier () /pdfxoutputcondition () /pdfxregistryname () /pdfxtrapped /false /createjdffile false /description << /ara <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> /bgr <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> /chs <feff4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e9ad88d2891cf76845370524d53705237300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /cht <feff4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc9ad854c18cea76845370524d5370523786557406300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /cze <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> /dan <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> /deu <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> /esp <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> /eti <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> /fra <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> /gre <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> /heb <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> /hrv (za stvaranje adobe pdf dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. stvoreni pdf dokumenti mogu se otvoriti acrobat i adobe reader 5.0 i kasnijim verzijama.) /hun <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> /ita <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> /jpn <feff9ad854c18cea306a30d730ea30d730ec30b951fa529b7528002000410064006f0062006500200050004400460020658766f8306e4f5c6210306b4f7f75283057307e305930023053306e8a2d5b9a30674f5c62103055308c305f0020005000440046002030d530a130a430eb306f3001004100630072006f0062006100740020304a30883073002000410064006f00620065002000520065006100640065007200200035002e003000204ee5964d3067958b304f30533068304c3067304d307e305930023053306e8a2d5b9a306b306f30d530a930f330c8306e57cb30818fbc307f304c5fc59808306730593002> /kor <feffc7740020c124c815c7440020c0acc6a9d558c5ec0020ace0d488c9c80020c2dcd5d80020c778c1c4c5d00020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /lth <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> /lvi <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> /nor <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> /pol <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> /ptb <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> /rum <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> /rus <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> /sky <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> /slv <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> /suo <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> /sve <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> /tur <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> /ukr <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> /enu (use these settings to create adobe pdf documents best suited for high-quality prepress printing. created pdf documents can be opened with acrobat and adobe reader 5.0 and later.) /nld (gebruik deze instellingen om adobe pdf-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. de gemaakte pdf-documenten kunnen worden geopend met acrobat en adobe reader 5.0 en hoger.) >> /namespace [ (adobe) (common) (1.0) ] /othernamespaces [ << /asreaderspreads false /cropimagestoframes true /errorcontrol /warnandcontinue /flattenerignorespreadoverrides false /includeguidesgrids false /includenonprinting false /includeslug false /namespace [ (adobe) (indesign) (4.0) ] /omitplacedbitmaps false /omitplacedeps false /omitplacedpdf false /simulateoverprint /legacy >> << /addbleedmarks false /addcolorbars false /addcropmarks false /addpageinfo false /addregmarks false /convertcolors /converttocmyk /destinationprofilename () /destinationprofileselector /documentcmyk /downsample16bitimages true /flattenerpreset << /presetselector /mediumresolution >> /formelements false /generatestructure false /includebookmarks false /includehyperlinks false /includeinteractive false /includelayers false /includeprofiles false /multimediahandling /useobjectsettings /namespace [ (adobe) (creativesuite) (2.0) ] /pdfxoutputintentprofileselector /documentcmyk /preserveediting true /untaggedcmykhandling /leaveuntagged /untaggedrgbhandling /usedocumentprofile /usedocumentbleed false >> ] >> setdistillerparams << /hwresolution [2400 2400] /pagesize [612.000 792.000] >> setpagedevice microsoft word volume 13, issue 1-1 journal of risk analysis and crisis response, 2023, 13(1), 1-18 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i1.352 1 article cross-cultural research of the perceived risk during the covid-19 pandemic krum krumov 1,*, johann f. schneider 2, jin liu 3, albena k. krumova 1, eko widodo 4, alexander l. gungov 5, marta juhasz 6, magdalena z. garvanova 7, sanjay kumar 8 and rita repaczki 6 1 department social psychology, sofia university “st. kliment ohridski”, sofia (1463), sofia, bulgaria 2 department of psychology, saarland university, saarland (66123), saarbrücken, germany 3 school of humanities and social sciences, beijing institute of technology, beijing (100081), beijing, china 4 faculty of business administration and communication sciences, atma jaya catholic university of indonesia, jakarta (12930), jakarta, indonesia 5 dept. of logic, ethics and aesthetics, sofia university “st. kliment ohridski”, sofia (1463), sofia, bulgaria 6 dept. ergonomics & psychology, budapest uni. of technology & economics, budapest (1111), budapest, hungary 7 department of public communications, ulsit, sofia (1463), sofia, bulgaria 8 western coalfields ltd., nagpur (440001), maharashtra, india * correspondence: prof.krumov@gmail.com received: february 7, 2023; accepted: march 30, 2023; published: march 31, 2023 abstract: one of the essential aspects in analyzing the topic of managing people's risk behavior in extreme situations is related to the issue of risk perception. the subject of this study was to reveal whether certain factors, such as cultural context, gender, age, education, religious beliefs, etc., mediate the perceived risk during the covid-19 pandemic. the study was conducted online from april to june 2020. the two samples included a total of 2617 participants aged between 18 and 70 years, where 1412 of which were drawn from asia and 1205 were representatives of europe. participants were asked to complete an internet-based version of a short questionnaire that described the covid-19 pandemic situation, as one of the questions referred to the hazard perception of the infection. results obtained showed that there was a higher level of perceived risk among members of the asian culture group in comparison with the european study participants. furthermore, it was revealed that gender is not a significant factor in the perception of risk in either the european or the asian cultural group (p> 0.05). the results showed that levels of the perceived risk increased with age and that there were significant differences between young and old in threat perception for the european sample (p<0.001). regarding the asian sample, it turned out that all age groups perceive the risk of infection almost equally. several demographic characteristics also emerged as important mediators of risk perception. for example, people with the lowest education, married, religious, and families with children fear the hazard of covid-19 infection to the greatest extent, compared to people with the highest education, singles, atheists, and families without children, who do not perceive the virus as a high-risk factor. keywords: asia; covid-19; cultural differences; europe; hazard perception; pandemic; perceived risk 1. introduction perceived risk of the virus during the covid-19 pandemic is a central point in the regulation and self-regulation of individuals' health behaviors. whether people will perceive the virus as a real danger or treat the threat lightly and irresponsibly depends on whether they will engage in krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 2 protective or risky behavior. in fact, risk behavior and risk perception are important scientific fields, well-developed in psychology, sociology, economics, management, and anthropology [1, 2, 3, 4, 5, 6]. this issue emerged with great importance and became the focus of research during the covid-19 pandemic when several scientists began to study the perceived risk of infection and the risk behavior of people [7, 8, 9, 10, 11, 12]. this statement sounds logically acceptable, however, from a psychological point of view, it is untenable, since, in practice, the behavior depends not only on the objective characteristics of the danger but above all on the way in which people perceive the threatening agent. the result of the risk perception process, i.e., the perceived risk, is precisely the one on which the risk behavior depends. perceived risk is defined as an individual’s subjective assessment of the level of risk associated with a particular hazard (e.g., health threat) and varies according to factors such as past experiences, age, gender, and culture defined as an individual’s subjective assessment of the level of risk associated with a particular hazard (e.g., health threat), and it varies according to factors such as past experiences, age, gender, and culture [13]. ultimately, perceived risk determines an individual's engagement in risk/protective behavior. several studies reveal that there are many factors on which risk management and risk behavior depend. for example, according to chionis and karanikas [14], the factors influencing risk perception are personality, experience, beliefs, age, gender, level of education, knowledge, culture, and psychological context. individuals form their inner vision of risk in a specific social and cultural context and under the influence of many external factors – friends, family, institutions, history, traditions, symbols, ideology, etc. the formation of their attitudes toward risk also depends on whether they live in a high-risk or low-risk environment and on whether their encounter with risk is of high or low intensity. on the other hand, it should also be borne in mind that the way of perceiving risk is mediated by internal personal factors such as knowledge, emotions, and attitudes, which means that the perceived risk ultimately depends on past experiences [15]. during their experiences, people learn what is harmful or beneficial to their well-being and what is safe or dangerous for their survival. in other words, it is the personal experience that teaches an individual when and what to fear or not fear. if the current situation is related to the health and life of the individual, the resulting risky behavior will be non-standard, unaccustomed, and extreme, as it is based on instinctive fear for one's own survival. fear is defined as a primary emotion that arises from the detection of an imminent threat, involving an immediate alarm reaction that mobilizes the organism by triggering a set of physiological changes. in a word, it is a short-term response to a present, clearly identifiable threat [16]. being a primary emotion and a short-term response to a clearly identifiable threat, fear is an attribute of any critical life-threatening situation. in this sense, it is an indicator and determinant of risk behavior since the greater the fear of external threats, the higher the degree of perceived risk. however, the risk (risky behavior/perceived risk), in general, is personal, and at the same time, it is a social and cultural construct [17]. that is why fear is a universal regulator of human behavior, which means it is a common element, a common attribute of both external and internal control (self-control). internal control is, to some extent, relative because the socio-cultural environment (the context) is a demiurge of personal phenomena, such as attitudes, values, and beliefs, through which krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 3 self-control is exercised. this circumstance presupposes the existence of a strange paradox: fear generally is subject to complete purposeful control by external factors, however, on a personality level this is not the case and the control, itself, is relative, although people imagine that they themselves have entire control of their own lives. when discussing the determinants of perceived risk in a critical situation, risk perception is usually seen as a personal process based on the interaction between available information and the individual’s set of attitudes, competencies, and experiences developed over a lifetime. as a result of this interaction, the individual assesses whether the stimulus that has arisen, or the event that has occurred, poses a danger to his/her well-being, and on the bases of this, makes a final decision about his/her own behavior. it should be borne in mind, however, that the danger itself does not directly impact the individual's behavior. before that, it must be perceived as an arousing fear threat, and in turn, the fear must provoke an awareness of the risky situation, threatening individual well-being. even though a situation is perceived as threatening and there is a perceived risk, there is no guarantee that the individual will act rationally and will not engage in risky behavior. this means that risk-taking behavior also depends on the functioning of several other psychological phenomena that are beyond the intentional conscious control of the individual. for example, individuals may realize that their action would be risky, but under the pressure of a momentary emotion it is possible for them to engage with it, and vice versa, to avoid it. this also means that emotional responses to risk are often independent of cognitive assessments and that emotions are stronger determinants of people's behavior [18]. for example, people tend to react with fear not only to the visible, familiar characteristics of an external object, but they may perceive as harmful such objects that are unobservable, unknown, and new to them [5]. these automatic reactions of the individual, based on unconscious emotional processes, can be caused by a variety of external stimuli – loud sounds, bright light, huge objects, and words whose connotation is associated with danger. they, however, affect the different social groups in different ways, depending on the experience of individuals, their attitudes, beliefs, etc. this ultimately means that the degree of perceived risk will vary between individuals and social groups. for example, slovic and his colleagues confirm that the word "chemicals", which is perceived as a danger and provokes a fearful reaction in members of the general public, is not perceived in the same way by scientists and experts. this study of “intuitive vs. scientific toxicology”, proving the discrepancy between expert and lay views about chemical risks, is an illustration of the existence of internal determinants (experience, knowledge, conceptions, attitudes, values) that make individuals perceive the same danger differently [19]. based on the analysis in the preceding lines, two main conclusions could be drawn: firstly, the magnitude of fear generated in an extreme (pandemic) situation that threatens the lives and health of individuals is an indicator of the degree of perceived risk. secondly, being an indicator of perceived risk, fear as such, its magnitude, characteristics, and control functions depend on internal (innate and acquired) and external factors. however, external factors dominate risk behavior to the greatest extent in extreme situations such as the pandemic. 2. materials and methods 2.1. purpose of the study and hypotheses krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 4 according to the previous analysis, the risk perception of covid-19 infection is mediated and depends on different factors, such as age, gender, social experience, cultural context, etc. this is precisely the purpose of the present study: to reveal whether factors such as cultural context, gender, age, and several demographic characteristics, mediate the hazard perception of covid-19 infection, resulting in varying degrees of individuals' perceived risk. in accordance with the purpose of the study, the following hypotheses were formulated: hypothesis 1. various cultural contexts mediate in different ways the hazard perception of covid-19 infection, as a result of which the degree of perceived risk is higher among representatives of the collectivist culture compared to representatives of the individualistic culture. hypothesis 2. sex differentiation mediates hazard perception of covid-19 infection, and as a result, the degree of perceived risk among women is higher than among men. hypothesis 3. age mediates hazard perception of covid-19 infection, resulting in varying degrees of perceived risk among different age groups. hypothesis 4. some demographic characteristics, such as education, financial security, religious orientation, health status, etc., mediate the hazard perception of covid-19 infection and significantly influence the perceived risk by individuals during the covid-19 pandemic. 2.2. participants the subjects of the study were representatives of two cultural groups drawn from europe and asia. the total number of participants included 2617 subjects, aged between 18 and 70 (m = 37.98, sd = 15.20), 1412 of which were drawn from asia (45.9% identified as women, 49.6% as men, and 4.5% preferred not to answer), and 1205 were representatives of europe (64.6% identified as women, 34.5% as men, and 0.9% preferred not to answer). the number of asian participants is higher due to the large population in the three asian countries. most of the research participants are of active working age. the average age of the participants from asia was 32 years (m = 31.50, sd = 12.77). the average age of the participants from europe was 45 years (m = 44.96, sd = 14.51). the distribution of participants in the study is as follows: asia: the study included representatives of three countries – china (n = 500), india (n = 500), and indonesia (n = 412). these are the three asian countries, influenced by collectivistic values with the largest populations but different religious systems – mainly buddhism, hinduism, islam, taoism, confucianism, and catholicism. europe: the study included participants from bulgaria (n = 405), germany (n = 400), and hungary (n = 400). these three countries, in general, are typical representatives of european christian civilization and culture. a section with various demographic characteristics is included in the questionnaire. all research participants were required to indicate their age, educational background, work experience, number of children in the family, financial status, qualifications, marital status, religious beliefs, various party affiliations, etc. data obtained make it possible to examine the influence of demographic factors on perceived risk. moreover, the data obtained show that the respondents are from diverse social groups, which reflects well enough the different cultural profiles of both samples. 2.3. materials and procedure krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 5 the study was conducted online from april to june 2020. participants were asked to complete an internet-based version of a short questionnaire that described the extreme covid-19 pandemic situation. the original version of the questionnaire is written in english, and for each country, it is translated into bulgarian, chinese, german, hindi, hungarian, and indonesian, respectively. this makes it possible for representatives of different social groups to fill out the questionnaire in their native language. the items were related to the main characteristics of the virus and the pandemic situation, such as the virus origin, functions of the virus, desire for vaccination, precautions against infection, etc. subjects had to present their answers on a six-point likert scale. one of the items was related to the fear of a virus contracting and aimed to measure the extent to which the subjects perceived the running pandemic situation as a risk to their health and well-being. the wording of this item is as follows: i'm not afraid of the hardships of life – i'm only afraid of the disease infection. the responses to this question were expected to show that a significant part of the entire sample perceived the virus as a real threat to human health. in addition, the results had to reveal the extents to which separate groups, differentiated according to sex, age, cultural context, and certain specific demographic characteristics, perceive the pandemic situation as risky to their health and well-being. for this purpose, respondents had to answer questions separated into a section, including demographic characteristics, as follows: sex, age, profession, family status, biological or adopted children, number of children in a family, severe childhood illnesses, current employment, financial status, religion, political orientation, etc. to reveal the influence of age on risk perception, according to the concept of armstrong [20], the subjects were divided into three age groups: early adulthood – up to 35 years; midlife – from 36 to 50; mature adulthood – over 50. the research ethics approval procedure was not applied in this study, as the survey was anonymous, and respondents completed the online questionnaire voluntarily. to examine the hypotheses, the following statistical methods were employed: descriptive statistics and independent samples t-test, paired-samples t-test, one-way anova, and post hoc tests. 3. results the results of the study showed the ways in which people of different genders, ages, experiences, and cultures perceive the hazard of infection during the covid-19 pandemic. it is reasonable to expect in this extreme situation, the external stimulus (covid-19) is perceived as a life-threatening danger, which arouses anxiety and fear in people for their lives. it depends, however, on the degree of the aroused fear, whether an individual will perceive the invisible virus as an extreme life-threatening danger, as a real risk to their life. the presence of a high-level virus fear indicates the presence of a high level of perceived risk, and conversely, the low level or lack of virus fear indicates a low level or lack of perceived risk. covid-19 hazard information, provoked by external agents (media, state institutions, doctors, etc.), does not directly affect individuals and is not a direct source of their fear as it is mediated by various factors such as age, gender, social experience, etc. this suggests that the degree of induced fear, as a determinant of perceived risk, will be mediated by these factors and will be dependent on them. one important factor that mediates fear and on which perceived risk depends is the cultural context. we assume, according to the formulated hypothesis 1, cultural contexts mediate the hazard perception of covid-19 infection, krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 6 and the degree of perceived risk is higher among representatives of the collectivist culture compared to representatives of the individualistic one. the results show (fig. 1) that in the total sample, the perceived high risk prevails – 53% (n = 855) vs. low risk 47% (n = 758). the assumption that collectivist cultures, such as asian, are dominated by a high degree of perceived risk, and individualistic cultures, such as european, are dominated by a low degree of perceived risk has been confirmed. the results show that europe is dominated by a low level of perceived risk – 57.3% (n = 460) vs. a high level of perceived risk – 42.7% (n = 343). the trend in asia is exactly the opposite: the high degree of perceived risk dominates – 63.2% (n = 512) vs. the low level of perceived risk – 36.8% (n = 298). figure 1. levels of perceived risk in different cultures. in order to accurately prove hypothesis 1, another statistical criterion was applied the student’s t-test for independent samples. the results confirm the assumption that the cultural context is a significant determinant (mediator) of the risk perception process during the covid-19 pandemic. in this case, the level of perceived risk for the asian sample is higher – x = 3.82, sd = 1.54 (n = 1400), than the european sample – x = 3.27, sd = 1.72 (n = 1204) at t = -8.52, df = 2440, p < 0.001, d = 0.35. one of the assumptions in the present study concerns sex as a mediator of perceived risk. according to hypothesis 2, sex is a significant determinant (mediator) in the risk perception process during the covid-19 pandemic, where the degree of perceived risk in women is higher than in men. although there are theoretical prerequisites for such an assumption, which will be discussed in the next section, this hypothesis was not confirmed. based on the results of the comparative analysis by sex for the entire sample, it can be concluded that sex is not a significant factor for variability in the level of perceived risk – t(2431) = 1.01, p = 0.312, which is de facto р > 0.05, n.s. furthermore, sex is not a significant factor in either the european cultural community – t(1191) = 1.16, p = 0.246, or the krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 7 asian group – t(1336) = -1.78, p = 0.075. the two-factor anova also confirmed that the cumulative effect between the variables “sex” and “type of culture” was not statistically significant – p > 0.05. according to the third hypothesis, age is a significant determinant (mediator) of risk perception during the covid-19 pandemic, as different age groups affect perceived risk to varying degrees. it is logical to assume that different ages affect risk perception differently when people are in a life-threatening situation, especially as the media pressure on the elderly was greatest. as noted in the previous section, respondents were divided into three age groups, according to the concept of armstrong [20]. he postulates the human life cycle, in general, is divided into twelve stages, and we accepted it as the most appropriate in the case. it is believed, concerning hypothesis 3, that during the covid-19 pandemic, the three stages of the human life cycle – early adulthood ( up to 35 years), midlife ( from 36 to 50), and mature adulthood (over 50), impact in different ways the risk perception process. the results reveal (fig. 2) that age is a significant factor that introduces statistically significant variability in relation to the level of perceived risk: one-way anova shows that f(2,2469) = 13.37, p < 0.001. the application of games-howell as a post-hot test reveals that the emerging group is the third age group – people over 50 years old (n = 594, x = 3.88, sd = 1.61), where the level of perceived risk is highest compared to early adulthood (n = 1205, x = 3.52, sd = 1.56), and midlife (n = 673, n = 3.43, sd = 1.70). two-way anova supports the findings regarding the impact of age on perceived risk levels. at the same time, other dependencies are revealed. the results show that both separate and combined effects of "age" and "type of culture" are statistically significant: age – f(2,2466) = 24.54, p < 0.001, η = 0.14; type of culture – f(1,2466) = 87.11, p < 0.001, η = 0.18; age/culture type – f(2,2466) = 12.30, p < 0.001, η = 0.10. figure 2. influence of life cycles on perceived risk levels. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 8 figure 2 shows that in the european sample, the level of perceived risk increases with age. it is lower among young people, while people over the age of 50 perceive the virus as highly threatening. the opposite trend is visible in the asian sample: all three age groups perceive the risk almost equally. however, the small differences between them show that representatives of the midlife group perceive the virus as more dangerous, while younger people perceive the virus as less threatening. one of the research objectives was to examine whether certain specific demographic characteristics mediate risk perception. hypothesis 4 suggests that several demographic characteristics, like education, financial security, religious orientation, health status, etc., mediate the process of hazard perception of covid-19, and in this way influence the perceived risk levels. we assumed that education was a factor, affecting hazard perception of covid-19 infection, as a result of which expected respondents with different levels of education (primary education, secondary education, bachelor’s degree, master’s degree ph.d., or a higher degree) to indicate different levels of perceived risk. two-way anova shows that the cumulative effects of the “culture type” and “education level” are not statistically significant – f(4,2572) = 1.87, and p = 0.114, n.s., so we used one-way anova, involving the independent variable "education" and the dependent variable "perceived risk level". the one-factor analysis of variance revealed that education introduces statistically significant variations in the assessment of perceived risk – f (4,2577) = 6.60 at p < 0.001, η = 0.10. figure 3. perceived risk and level of educational qualification. the games-howell post hoc test was used for the intergroup comparison of the means, as the levene test showed that there was no equality between the variances. the results, in this case, show that there are statistically significant differences between “primary education”, on the one hand, and “phd or a higher degree” (p < 0.01), on the other hand, as well as between "bachelor's degree", on the krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 9 one hand, and "master's degree" (p < 0.01) and “phd or a “higher degree” (p < 0.001), on the other hand. it can be seen in figure 3 that with increasing stages of educational qualification, the level of the perceived risk of infection with covid-19 generally decreases, and this was observed particularly when comparing respondents with the lowest and highest levels of education. we assumed that family/marital status mediated the perception of risk. when using two-way anova, the cumulative effects of “culture type” and “marital status” were not statistically significant f(3.2566) = 2.24 at p = 0.082, n.s. one-way anova, however, with the independent variable "family status" and dependent variable "level of perceived risk" reveals that family status has a statistically significant effect on the perceived risk level – f(3,2570) = 13.67 for p < 0.001, η = 0.13. figure 4. perceived risk and family status. the levene test shows equality of variances, so the post hoc tests use the scheffe test to estimate the differences in the mean values between the compared groups. the results show (fig.4) that the group of divorcees does not differ from other groups – p > 0.05. the group of married differs from unmarried/single – p < 0.05 and those living in cohabitation – p < 0.001, and vice versa. in addition, the group of unmarried/single differs from those living in cohabitation, and vice versa p < 0.001. an important characteristic that mediated the perception of risk is the existence of children in the family. in this regard, respondents had to answer the question: do you have any biological or adopted children? two-way anova shows that there are statistically significant individual and cumulative effects of the cultural type, and the presence of children in the family concerning perceived risk: type of culture – f(1,2572) = 83.02, p < 0.001, η = 0.18; children in family – f(1,2572) = 25.64, p < 0.001, η = 0.10; cumulative effect – f(1,2572) = 11.78, p < 0.001, η = 0.07. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 10 figure 5 shows that having children in the family increases the perceived risk, especially in european cultures. in the asian sample is observed, a higher level in both groups families with children, and without children, due to the higher perceived risk in general in this cultural group. another demographic characteristic included in the questionnaire refers to respondents' religiosity. respondents had to indicate the degree of their religiosity, whether they believed in something else or were outspoken atheists. figure 5. perceived risk in families with and without children. two-way anova shows that there are statistically significant individual and cumulative factor effects of "culture type" and "degree of religiosity" on "perceived risk". for “culture type” – f(1,2567) = 77.00 at p < 0.001, η = 0.17, for “degree of religiosity” – f(3,2567) = 9.93 at p < 0.001, η = 0.10 and “culture x religiosity” – f(3,2567) = 8.54 at p < 0.001, η = 0.10. the graphical visualization of the results clearly outlines the differences between the studied groups. as figure 6 shows, between representatives of asian culture, there are no statistically significant differences, in relation to the perceived risk, depending on religious beliefs (p > 0.05). for europeans, however, religious beliefs influence perceived risk. there are significant differences (p < 0.001) between europeans with deep and moderate religious feelings, on the one hand, and europeans who are atheists or believe in something indefinite, on the other hand. the former indicated high values of the perceived risk of infection, but the latter stated that they are not afraid of virus hazards. while deeply and moderately religious europeans and asians do not differ in relation to the perceived risk, there are significant differences between those representatives of the two cultures, who are atheists, and who believe in something indefinite (p < 0.001). europeans, who are atheists and believe in something indefinite, have significantly lower levels of perceived risk than asians, who are members of the same religious groups but perceive the situation as highly risky. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 11 figure 6. perceived risk and religiosity of respondents. 4. discussion the results of the study showed (fig.1) that, at the very beginning of the covid-19 pandemic, in representatives of both cultures, asian and european, high levels of anxiety and fear of the virus were induced. it, although invisible, was perceived as a real threat, as a risk to the life and health of individuals. findings show, however, that there are significant differences, between asian and european representatives, concerning levels of the hazard perception of covid-19 infection. since the invisible external threat (covid-19) is a constant quantity for the entire study population, the conclusion is that differences in degrees of perceived risk are due to the fact the impact of induced fear is mediated in different ways from the two cultures – asian and european. the results obtained unequivocally confirm hypothesis 1, according to which the degree of perceived risk is higher among representatives of the collectivistic culture compared to representatives of the individualistic culture. the higher level of perceived risk among representatives of the collectivist culture, compared to representatives of the individualistic one, can be explained by differences in characteristics between the two cultures. as is known, according to the cultural perspective [21, 22, 23, 24, 25, 26, 27, 28], societies can be divided into two main types collectivistic and individualistic. based on this division lies one of the main characteristics of societies, related to the degree to which people perceive themselves as integrated into groups [29, 30]. in collectivist societies, individuals are closely related to the group – they are loyal to the group, accountable for its achievements, and rely on mutual support. in individualistic societies, people are poorly connected to each other, and they are responsible primarily for themselves and their own goals. it is generally assumed that asian culture is collectivistic and european culture is individualistic, which, therefore, means that different levels of the perceived risk of covid-19, revealed in the two samples, can be explained by cultural differences between asian and european participants. representatives of asian culture fear not only krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 12 for their own lives and the lives of those closest to them, but they also fear for the well-being of the group they are a part of and the society to which they belong. in other words, for them, the degree of perceived risk is significantly higher, as the virus poses an existential threat not only to themselves and their families but also to their groups of membership and to the society they are affiliated with. for representatives of individualistic societies, the degree of perceived risk is primarily due to fear for one's own life and the well-being of one's family. for europeans, the existential threat seems to be rather personalized, as the ties to other people, to membership groups, and to society at large, are not strong enough. we hypothesized that along with cultural context, gender differentiation would also be a significant mediator in the risk perception process. according to hypothesis 2, we assumed, in an extreme pandemic situation, sex differentiation mediates hazard perception of infection, as a result of which the degree of perceived risk among women is higher than among men. the reasons for this assumption are related to the popular belief that the ordinary mind perceives images of women and men differently: a woman is usually perceived as weak, defenseless, and timid, while a man is perceived as strong, fearless, and tough. it was expected that the results of the present study would reveal that, in a critical pandemic situation, females would be more fearful of the virus than males, and, respectively, their level of perceived risk would be significantly higher. however, the results of the comparative analysis for the whole sample showed that gender was not a significant factor in the variability in the level of perceived risk between women and men. furthermore, it turned out that when comparing men and women as members of the same cultural group, gender also had no significant influence on the degree of perceived risk. the finding that there were no significant differences between both asian male and female and european male and female participants in the study leads to the conclusion that in asian and european cultural contexts, during the extreme covid-19 pandemic situation, the perceived risk of infection is the same for both women and men. the scientific explanation of this fact has to do with the nature of fear as a primary emotion that arises in the presence of a lethal threat. it is the human brain that automatically reacts to all dangers and threats coming from the environment [31]. in the case of the kovid-19 pandemic, fear of the deadly virus was induced from the outset, activating in women and men alike the survival instinct formed at the physiological level over millennia. although there are generally some differences in the physiological and psychological profiles of men and women, these do not appear to be significant enough to cause representatives of the two sexes to perceive lethal threats differently. according to hypothesis 3, during the covid19 pandemic, age mediates hazard perception infection, resulting in varying degrees of perceived risk among different age groups. in line with this assumption, we expected that in the three distinct age groups – early adulthood (up to 35 years), midlife (from 36 to 50), and mature adulthood (over 50), the levels of induced fear, respectively, the degrees of perceived risk, would be different. the finding for the whole sample is (fig. 2) that age is a significant factor that influences the risk perception process, whereby the level of perceived risk is highest in the group of people over 50 years old, compared to the early age groups. these differences between the groups can be explained in two ways by the information invasion of the media on people, on the one hand, and by the "distance to death" on the other. as is well known, according to medical instructions, the media around the world have described the virus as particularly deadly for people over the age of 60 since the beginning of the pandemic. it is only logical that people in this age group assume that they are the most at risk in the pandemic situation. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 13 “distance to death” is the second reason why perceived risk is higher in older compared to younger people. while young people have a long-life horizon, the elderly have a short life perspective, and the distance to death is shortened for them. the shorter the time to the natural end of human life, the older people think about death and, as a result, perceive even minimal dangers as high-risk to their lives. the findings become very exciting when the analysis focuses on combined effects between age groups and types of cultures. the results show that in europeans, the degree of perceived risk increases dramatically with increasing age, where representatives of the mature adulthood group, compared to the other two groups, demonstrate a much higher fear of the virus infection. although, in general, asians over 50 also perceived the virus as a major life-threatening danger, there were no significant differences between the three age groups of the asian sample. all three groups have a high degree of hazard perception of contracting the virus, and in this, they differ dramatically from europeans. this means that in the two cultures, european and asian, the age of people mediates and affects in different ways the perceived risk (fig. 2). the fact that, unlike the europeans, representatives of the asian culture of all age groups perceive the pandemic situation as high-risky can be explained by the characteristics of the asian cultural model. as representatives of the collectivist culture, asians of the studied age groups feel responsible not only for themselves personally but also for the group of membership and for the society to which they belong. the lack of significant differences in the degree of risk perception between the distinct age groups in the asian sample means that although the collectivist cultural model is most strongly cultivated in the older generation, this model dominates a significant part of the identity of young asians. the results in the previous paragraph suggest that, during the covid-19 pandemic, certain demographic factors mediated hazard perception of the infection, thereby influencing the perceived risk. one of these factors is education. study results show, regarding education, that during the covid-19 pandemic, it emerges as a factor influencing the perception of life-threatening danger. findings show that different educational levels mediate the hazard perception of covid-19 infection in different ways and thus influence the degree of perceived risk (fig.3). it is evident that between individuals with the highest level of education, and those with the lowest level of education, there are statistically significant differences in levels of perceived risk, revealing a trend to increase in induced fear of the virus in the former and a decrease in the latter. individuals with a bachelor's degree or less perceive the threat of contracting the virus as high risk, while the trend is the opposite for those with a master's degree or doctorate. in this case, however, we could argue that this is only a trend and would not be able to draw precise conclusions regarding the different levels of education, as the educational degrees in the three asian and three european countries where the study was conducted are not identical. nevertheless, it could be concluded, in general, that with increasing years of training, levels of the perceived risk of contracting the virus decrease. these findings support the thesis on the relationship between education and risk in an extreme pandemic situation, according to which educational levels are a significant factor in mediating the perceived risk of contagion [32, 33, 34, 15]. likely, the low educational level of individuals makes them rely predominantly on their primal instincts in extreme situations, which increases levels of fear and anxiety, while accumulated pieces of knowledge and experience of individuals with a high level of education mediate the impact of the threat, and reduces levels of the perceived risk. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 14 another factor, which mediates the hazard perception of infection and influences the perceived risk during the covid-19 pandemic, is the family status of individuals. study results show (fig.4) that family status has a statistically significant effect on the individual's level of perceived risk. among individuals in the married group, the degree of perceived risk is the highest, and this is logical since they fear not only for themselves but also for their families (spouses and children). the degree of perceived risk among unmarried individuals is also high. loneliness, in this case, seems to be the main reason for the high fear level of life-threatening danger. the results show that the degree of perceived risk among divorced individuals is also high. this is probably because individuals have separated from their former spouse and are living in loneliness, or perhaps this is coupled with fear for their children, living with or apart from them. findings suggest that loneliness and awareness of existing health hazards to children are very important mediators of perceived risk, during extreme situations, like the covid-19 pandemic. this is categorically proven by the data obtained for the individuals who live with a partner in cohabitation. the results show that during the covid-19 pandemic, these individuals were the least likely to perceive the situation as threatening or risky. likely, the low-level induced fear of the virus among them is because these individuals do not live alone and that they do not have children in common to care for. the finding that individuals who live alone are more afraid of death, in general, is not surprising. however, the important in this case is whether the presence of biological or adopted children is a significant factor mediating perceived risk in an extreme situation. in addition, it is curious whether representatives of both cultures who have children are equally afraid of the virus and perceive to an equal extent the covid-19 pandemic as risky. the results obtained from the study showed (fig. 5) that for the entire sample, the presence of biological or adopted children is a significant factor that mediates the perceived risk. results show that both european and asian participants who have biological or adopted children in their families fear the virus to a great extent, as a result of which the degree of the perceived risk of covid-19 infection is very high. nevertheless, differences between the two cultures emerge. for example, in the asian sample, both individuals having children, and individuals without children, demonstrated high levels of perceived risk. at the same time, only european participants, who have children, perceive the covid-19 pandemic as risky, indicating in their responses that they intensely feared contracting the virus. this difference in the results of the two samples can be explained by the different characteristics of the two cultures. obviously, representatives of the individualistic culture fear primarily for their own children. in contrast, representatives of the collectivist culture feel responsible not only for their own but to a large extent for all children in the society to which they belong. it could be argued, during a pandemic situation, in an individualistic context, there is a personalization of perceived risk, whereas, in a collectivistic context, the life-threatening danger is perceived by individuals not only as a risk to themselves and their loved ones but also as a risk to society. the fear of mortal illness and death in all cultures is associated with belief in a god or other higher powers on which the existence of the human being depends. therefore, it was interesting to check the ways of threat perception of the virus in subjects who are religious and non-religious. incidentally, the relationship between people's religiosity, on the one hand, and the fear and anxiety generated by the covid-19 pandemic, on the other, has been the subject of analysis in a number of studies [35, 36, 37, 38, 39, 40]. the results of the current study show that there are differences in the krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 15 perception of the risk of contracting the virus between the groups identified on the basis of religiosity – deeply religious, moderately religious, believing in something undefined, and atheists (fig. 6). in this respect, the individuals in the european sample stand out most dramatically. there are significant differences between representatives of european culture with deep and moderate religious feelings, on the one hand, and atheist europeans, on the other. the former feel great fear of contracting the virus, while the latter do not perceive the pandemic situation as high-risk at all. the high level of the perceived risk of virus infection, in religious europeans, can be explained by the fact that they trust in god, think about death, feel responsible for others, and, ultimately, the way they will in the future leave this world, is of great importance for them. at the same time, european atheists are sure that there is no god and do not feel responsible either for how they live or for how they will leave this world. it is important to note that individuals in both the european and asian samples, who are profoundly and moderately religious, do not differ from each other in their assessment of the pandemic situation. however, in comparison with other religious groups, they demonstrate higher levels of perceived risk. while religious europeans and asians are similar in that they perceive the pandemic situation as high risk, for atheist europeans and atheist asians, the picture is precisely the opposite. europeans, who are atheists or believe in something undefined, do not fear contracting the virus, as has been pointed out, while asians belonging to the same religious group perceive the pandemic situation as very risky. this difference could be explained by the specificity of asian culture and its influence on this religious group: asian atheists, dominated by a collectivist culture, feel responsible not only for their own lives but also for the lives of other members of the society to which they belong, and this is why they perceive the covid-19 pandemic as high-risky. 5. conclusions the present study, conducted in three european, and three asian countries, revealed that different cultures, as well as certain demographic factors, mediate the hazard perception of the virus infection, resulting in varying levels of perceived risk in individuals during the covid-19 pandemic. the results showed increased levels of perceived risk in the whole study population. when comparing the two samples, however, it appeared that the level of the perceived risk of virus infection was higher in asians compared to europeans, and this supports the thesis for a mediating role of cultural context. in addition, age differences between individuals are also revealed as a significant mediator of risk perception. for europeans, however, fear of contagion was found to increase with increasing age, while asians of almost all age groups perceived the virus equally as a high risk to their well-being. contrary to expectations, it turned out that, unlike age, gender was not a significant mediator in hazard perception of covid-19 infection. results also reveal that the ways people perceive the threat of viruses in an extreme pandemic situation depends on certain demographic factors such as education, marital status, presence of own or adopted children, and religiosity. however, the influence of these factors is different in different cultural contexts. the conclusions drawn from the study will contribute to more effective management of people's risk behavior in extreme situations. this is because the way in which risk is perceived should be seen as key to the successful management of people's health behavior during pandemics. it is the perceived risk that determines whether individuals will engage in risk-taking or in risk-avoidance behavior following the precautions imposed by medical institutions. exploring risk perception in situations of krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 16 health uncertainty is, therefore, a method to successfully manage future pandemics and a way to save human lives. in addition to the current risk perception study, the international research team investigated early in the covid-19 pandemic also the health locus of control, hopelessness, coping strategies, and other phenomena, which will be the subject of future analyses. the authors hope that the study will contribute to more effective management of human health behavior in future pandemics. the strengths, weaknesses and future research goals are as follows: the most important strength of this work is that the research results will help to manage people’s risk behavior more effectively in future pandemic situations. the findings that differences in cultures, genders, age, education, religious beliefs, etc., influence the perception of the risk of infection, show that in the future it is necessary to use differentiated approaches in the management of the health behavior of individuals in extreme situations. another advantage is that the study represents an original snapshot of the beginning of a pandemic in two different cultures that will never start the same way again. a weakness of the study was the lack of time to create a sufficiently efficient research design to ensure the use of representative (stratified) national samples for each country. this prevented the possibility of making a comparative analysis between the countries, which is why it was made only between the two cultural groups the asian and the european. contributions: conceptualization, k.k.; methodology, k.k. and a.k.; software, a.k. and m.g.; formal analysis, k.k., j.s., a.k. a.g. and m.g.; investigation, k.k., j.s., j.l., a.k., e.w., m.j., s. k. and r.r.; resources, k.k., j.s., j.l., a.k., e.w., a.g., m.j., m.g., s.k. and r.r.; data curation, m.g. and a.k.; writing—original draft preparation, k.k.; writing—review and editing, k.k., j.s., j.l., a.k., e.w., a.g., m.j., m.g., s.k. and r.r.; visualization, m.g.; supervision, k.k. acknowledgments: the authors thank all anonymous survey participants who voluntarily completed the online questionnaire. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] boholm, a. anthropology and risk. routledge, london, 2015; 190 pp. doi: https://doi.org/10.4324/9781315797793. [2] gigerenzer, g. simply rational: decision making in the real world. oxford university press, 2015. doi: https://doi.org/10.1093/acprof:oso/9780199390076.001.0001. [3] hopkin, p. fundamentals of risk management: understanding, evaluating and implementing effective risk management. london: kogan pag, 2010; 436 pp. http://dspace.vnbrims.org:13000/xmlui/bitstream/handle/123456789/4826/fundamentals%20of%20risk%2 0management%20understanding%2c%20evaluating%20and%20implementing%20effective%20risk%20 management.pdf?sequence=1&isallowed=y. [4] kahneman, d. thinking, fast and slow. farrar, straus, and giroux. new york, 2011. http://dspace.vnbrims.org:13000/xmlui/bitstream/handle/123456789/2224/daniel-kahneman-thinking-fas t-and-slow-.pdf?sequence=1&isallowed=y. [5] slovic, p. perception of risk. science, 1987, 236, (4799), 280–285. doi: https://doi.org/10.1126/science.3563507. see also: slovic p. the feeling of risk: new perspectives on risk perception. routledge, london, 2010; 456 pp. doi: https://doi.org/10.4324/9781849776677. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 17 [6] walaski, p. risk and crisis communications: methods and messages. new york : john wiley & sons, incorporated, 2011; 224 pp. https://www.wiley.com/en-ie/9780470592731. [7] adachi, m., murakami, m., yoneoka, d., kawashima, t., hashizume, m., sakamoto, h., eguchi, a., ghaznavi, c., gilmour, s., kaneko, s. et al. factors associated with the risk perception of covid-19 infection and severe illness: a cross-sectional study in japan. ssm popul. health, 2022, 18, 101–105. doi: https://doi.org/10.1016/j.ssmph.2022.101105. [8] cipolletta, s., andreghetti, g.r., mioni, g. risk perception towards covid-19: a systematic review and qualitative synthesis. international journal of environmental research and public health, 2022, 19, (8), 4649. doi: https://doi.org/10.3390/ijerph19084649. [9] dryhurst, s., schneider, c.r., kerr, j., freeman, a.l.j., recchia, g., vander bles, a.m., spiegelhalter, d., van der linden, s. risk perceptions of covid-19 around the world. journal of risk research, 2020, 23, (7-8), 994–1006. doi: https://doi.org/10.1080/13669877.2020.1758193. [10] heydari, s.t., zarei, l., sadati, a.k., moradi, n., akbari, m., mehralian, g., lankarani, k.b. the effect of risk communication on preventive and protective behaviours during the covid-19 outbreak: mediating role of risk perception. bmc public health, 2021, 21,(1), 54. pmcid: pmc7787415. doi: https://doi.org/10.1186/s12889-020-10125-5. [11] siegrist, m., luchsinger, l., bearth, a. the impact of trust and risk perception on the acceptance of measures to reduce covid-19 cases. risk analysis, 2021, 41, 787–800. doi: https://doi.org/10.1111/risa.13675. [12] yang, x.y., gong, r. n., sassine, s., morsa, m., tchogna, a. s., drouin, o., chadi, n., jantchou, p. risk perception of covid-19 infection and adherence to preventive measures among adolescents and young adults. children, 2020, 7, (12), 311. doi: https://doi.org/10.3390/children7120311. [13] apa. apa dictionary of psychology. washington, dc: american psychological association. vandenbos, gary, r, apa dictionary of psychology. washington, dc: american psychological association, 2015. available online at: https://dictionary.apa.org/risk-perception (accessed october 10, 2022). [14] chionis, d., karanikas, n. differences in risk perception factors and behaviours amongst and within professionals and trainees in the aviation engineering domain. aerospace, 2018, 5, (62), 1–23. doi: https://doi.org/10.3390/aerospace5020062. [15] traczyk, j., zaleskiewicz, t. implicit attitudes toward risk: the construction and validation of the measurement method. journal of risk research, 2015, 19, (5), 632–644. doi: https://doi.org/10.1080/13669877.2014.1003957. [16] apa. apa dictionary of psychology. washington, dc: american psychological association. vandenbos, gary, r, apa dictionary of psychology. washington, dc: american psychological association, 2015. available online at: https://dictionary.apa.org/fear (accessed october 10, 2022). [17] douglas, m., wildavsky, a. b. risk and culture: an essay on the selection of technical and environmental dangers. berkeley, ca: university of california press, 1982. https://faculty.washington.edu/mccurdy/sciencepolicy/douglas%20and%20wildavsky.pdf. [18] loewenstein, g. f., weber, e. u., hsee, c. k., welch, n. risk as feelings. psychological bulletin, 2001, 127, (2), 267–286. doi: https://doi.org/10.1037/0033-2909.127.2.267. [19] slovic, p., malmfors, t., krewski, d., mertz, c. k., neil, n., bartlett, s. intuitive toxicology. ii. expert and lay judgments of chemical risks in canada. risk analysis, 1995, 15, (6), 661–675. doi: https://doi.org/10.1111/j.1539-6924.1995.tb01338.x. [20] armstrong, t. the human odyssey: navigating the twelve stages of life. ixia press, 2019. https://www.institute4learning.com/resources/articles/the-12-stages-of-life/. [21] gudykunst, w.b., gao, g., schmidt, k.l., nishida, t., bond, m., leung, k., wang, g., barraclough, r. a. the influence of individualism-collectivism, self-monitoring, and predicted-outcome value on communication in ingroup and outgroup relationships. journal of cross-cultural psychology, 1992, 23, 196– 213. doi: https://doi.org/10.1177/0022022192232005. [22] lee, h.o., boster, e.j. collectivism-individualism in perceptions of speech rate: a cross-cultural comparison. journal of cross-cultural psychology, 1992, 23, 377–388. doi: https://doi.org/10.1177/0022022192233008. [23] kemmelmeier, m., burnstein, e., krumov, k., genkova, p., kanagawa, c., hirshberg, m. s., erb, h-p., wieczorkowska, g., noels, k. a. individualism, collectivism and authoritarianism in seven societies. krum krumov, et al. / journal of risk analysis and crisis response, 2023, 13(1), 1-18 doi: https://doi.org/10.54560/jracr.v13i1.352 18 journal of cross-cultural psychology, 2003, 34, (3), 304–322. doi: https://doi.org/10.1177/0022022103034003005. [24] krumov, k., larsen, k. cross-cultural psychology: why culture matters. iap information age publishing, po box 79049, charlotte, nc 28271, usa, 2013, 478 рр. https://www.infoagepub.com/products/cross-cultural-psychology. [25] matsumoto, d., weissman, m., preston, k., brown, b., kupperbusch, c. context-specific measurement of individualism-collectivism on the individual level: the ic interpersonal assessment inventory (iciai). journal of cross-cultural psychology, 1997, 28, 743–767. doi: https://doi.org/10.1177/0022022197286006. [26] singelis, t.m., triandis, h.c., bhawuk, d.p.s., gelfand, m.j. horizontal and vertical dimensions of individualism and collectivism: a theoretical and measurement refinement. cross-cultural research, 1995, 29, 240–275. doi: https://doi.org/10.1177/106939719502900302. [27] triandis, h.c. new directions in social psychology: individualism and collectivism. boulder, co: westview press, 1995. https://psycnet.apa.org/record/1995-97791-000. [28] triandis, h.c., mccusker, c., hui, c.h. multimethod probes of individualism and collectivism. journal of personality and social psychology, 1990, 59, (5), 1006–1020. doi: https://doi.org/10.1037/0022-3514.59.5.1006. [29] hofstede, g. cultures and organizations: software of the mind. london/new york: mcgrawhill, 1991. https://www.worldcat.org/title/cultures-and-organizations-software-of-the-mind/oclc/23015181. [30] hofstede, g. what did globe really measure? researchers’ minds versus respondents’ minds. journal of international business studies, 2006, 37, 882–96. doi: https://doi.org/10.1057/palgrave.jibs.8400233. [31] ledoux, j. rethinking the emotional brain. neuron, 2012, 73, (4), 653–676. doi: https://doi.org/10.1016/j.neuron.2012.02.004. [32] rattay,p., michalski, n., domanska, o., kaltwasser,a., de bock, f., wieler, l.h., jordan, s. differences in risk perception, knowledge and protective behaviour regarding covid-19 by education level among women and men in germany. results from the covid-19 snapshot monitoring (cosmo) study. plos one, 2021, 16, (5): e0251694. pmid: 33979413. doi: https://doi.org/10.1371/journal.pone.0251694. [33] pförtner, t. k., dohle, s., hower, k.i. trends in educational disparities in preventive behaviours, risk perception, perceived effectiveness and trust in the first year of the covid-19 pandemic in germany. bmc public health, 2022, 22, 903. doi: https://doi.org/10.1186/s12889-022-13341-3. [34] soltan, e.m., el-zoghby, s.m., salama, h. m. knowledge, risk perception, and preventive behaviors related to covid-19 pandemic among undergraduate medical students in egypt. sn compr. clin. med., 2020, 2, 2568–2575. doi: https://doi.org/10.1007/s42399-020-00640-2. [35] bentzen, j. s. in crisis, we pray: religiosity and the covid-19 pandemic. journal of economic behavior and organization, 2021, 192, 541–583. doi: https://doi.org/10.1016/j.jebo.2021.10.014. [36] kranz, d., niepel, c., botes, e., greiff, s. religiosity predicts unreasonable coping with covid-19. psychology of religion and spirituality, 2020, advance online publication. doi: https://doi.org/10.1037/rel0000395 // https://psycnet.apa.org/fulltext/2020-71379-001.pdf. [37] lee, s.a., mathis, a.a., jobe, m.c., pappalardo, e.a. clinically significant fear and anxiety of covid-19: a psychometric examination of the coronavirus anxiety scale. psychiatry research, 2020, 290: 113112. pmid: 32460185. doi: https://doi.org/10.1016/j.psychres.2020.113112. [38] pomerleau, j. m., pargament, k. i., krause, n., ironson, g., hill, p. religious and spiritual struggles as a mediator of the link between stressful life events and psychological adjustment in a nationwide sample. psychology of religion and spirituality, 2020, 12, (4), 451–459. doi: https://doi.org/10.1037/rel0000268. [39] rigoli, f. the link between covid-19, anxiety, and religious beliefs in the united states and the united kingdom. j. relig. health, 2021, 60, 2196–2208. doi: https://doi.org/10.1007/s10943-021-01296-5. [40] saraei, m. johnson, k.a. disappointment with and uncertainty about god predict heightened covid-19 anxiety among persian muslims. religions, 2023, 14, (1), 74. doi: https://doi.org/10.3390/rel14010074. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word dynamic vulnerability analysis of population for flood disaster in urban area dynamic vulnerability analysis of population for flood disaster in urban area xilei pang emergency management department, dalian adminstrative college dalian 116013, china, pxl_bang@163.com abstract vulnerability analysis is a key step of risk analysis of natural disaster. based on information diffusion theory and incomplete information, it analyzes the changes of the affected population in urban flood and identifies the function relationship between the hazard (rainstorm) and disaster-bearing body (population), which is called vulnerability curve of population in flood disaster.this curve not only can present some changing extent of urban flood risk, but also could provide an important basis for integrated risk assessment of urban flood. keywords：vulnerability，flood，disaster-bearing body，population，dynamic 城市洪涝灾害的人口脆弱性动态分析 庞西磊 大连行政学院应急管理教研部，大连，116013 摘 要：承灾体脆弱性分析是自然灾害风险分析的关键步骤之一。基于信息扩散技术，本文利用小样本数据， 在框定时间约束的前提下，较客观地分析了城市洪涝灾害系统中单一承灾体（受灾人口）的动态变化情况， 并识别和近似推理出致灾因子（暴雨）与承灾体（人口）之间的函数关系，即人口脆弱性变化曲线。该曲线 在一定程度上可以反映出研究案例区的洪涝灾害风险发展变化情况，并为自然灾害的综合风险评估提供重要 依据。 关键词：脆弱性，洪涝灾害，承灾体，人口，动态 1 研究背景 有史以来，洪涝灾害一直是许多国家和地区面 临的主要自然灾害之一。由于全球变化引起的区域 气候波动，导致极端暴雨天气的频繁出现，洪涝灾 害的发生频次已经超过台风、地震、泥石流、干旱 等自然灾害（参见图1），逐渐成为危害城市安全和 社会、经济发展的头号公敌。有权威统计报告[1]显示， 中国近十年来的洪涝灾害造成的受灾人口和经济损 失规模均列世界首位，而这些洪涝灾害有很大一部 份都发生在人口集中和经济发展迅速的城镇地区。 图1 21世纪初全球巨灾发生频率对比图 长期以来，国内的城市洪涝灾害风险研究主要集中 在暴雨降水预测、洪水动力模型、致灾因子危险性 分析等方面，这些研究重在计算和分析灾害发生的 可能性方面，多是利用暴雨洪涝形成的物理数学模 型和计算机仿真模拟技术，从不同侧面对城市洪涝 灾害风险进行了不同层次的探讨和研究。 随着认知水平的提高，人类对自然灾害的本质 和演变规律的认识也发生着变化。即便是在相当长 一段时间内某种自然灾害的演变规律是不变的，但 时刻都会发展变化的社会经济系统，也会导致区域 自然灾害风险发生明显的变化。自然灾害的风险水 平，主要与致灾因子的强弱和人类社会的脆弱性有 关。脆弱性是系统暴露在因气候变化引起的灾害事 件之后调节风险变化的因素[2]，当致灾因子的破坏性 强且社会系统的脆弱性也高时，自然灾害的风险水 平就高。从时间尺度上看，脆弱性研究不仅要关注 长时间内全球气候变化的缓慢过程，也关注短期内 突发性环境事件的影响，脆弱性会在较短时间内从 极其脆弱突变为极其安全[3]。从系统学角度看，社会 -经济-生态系统的脆弱性、恢复力和适应能力三者之 间是相互影响和制约的[4]。 journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 166-174 published by atlantis press copyright: the authors 166 willieb typewritten text received 1 october 2013 willieb typewritten text accepted 12 december 2013 willieb typewritten text xilei pang 当前，自然灾害领域的脆弱性研究越来越关注 自然系统与社会系统之间的复杂作用关系，其未来 的发展越来越取决于多学科研究方法、研究视角的 交叉以及多源数据多学科知识的融合。 关于脆弱性研究，有人关注于概念和框架的理 解，也有人热衷于运用加权处理、层次分析等方法 进行脆弱性评价。例如，有研究重在用多指标体系 建立脆弱性分析模型[5-6]；也有研究从自然灾害系统 入手进行机理分析的探索，并得出“脆弱性是区域灾 害系统中致灾因子、承灾体和孕灾环境综合作用过 程的状态量，它主要取决于区域的经济发达程度与 社区安全建设水平”的结论[7]。有研究认为，经济的 发展使我们进入“系统的系统”时代，许多关键设施 面临高风险，社会经济脆弱性并没有降低 [8] 。也有研 究人为：随着收入的增加，灾害的死亡人数也在增 加，一旦超过某个拐点，灾害死亡人数就开始下降， 因此，转变经济发展方式，实现“包容性增长”，才 是降低社会经济脆弱性的根本[9]。虽然人们发现灾害 损失和 gdp 之间存在着经验关系[10]，但人口密度等 数据在研究生命易损性时也很重要[11]。理论上讲， 经济社会发展评价指标体系越完善，脆弱性的评价 就越可靠，但从成本角度看，有限的几个主要指标 就可以支撑脆弱性分析。 脆弱性的形成原因及表现特征在空间上具有很 强的区域差异性，在时间上具有动态变化性。因此， 要想跨区域、跨时段、跨学科地建立科学合理的脆 弱性评价指标体系是非常困难的。此外，指标体系 评价法所得出的评价结果的有效性也很少被验证。 因此，仅仅依靠指标值计算和权重赋值而得到的脆 弱性分析结果，会显得有些单薄，而且权重的大小 从某种角度讲过于主观。 自然灾害是自然和人类社会相互作用的复杂 现象。从系统论角度看，灾害风险是由风险源、风 险载体和人类社会的防减灾措施等三方面因素相 互作用而形成的[12]；自然灾害风险研究应该从孕灾 环境、致灾因子、承灾体、承灾能力、灾害损失等 多项构成自然灾害系统的基本要素着手研究[13]。 由上可知，承灾体脆弱性研究是自然灾害风险 分析不可或缺的关键步骤。因此，为了解决人们在 日常研究中经常碰到的数据缺乏、过分依赖主观赋 分评价和难以准确量化多因素关系的现状，本文基 于综合风险评估的基本模式[14]，在尽量避免主观评 价的前提下，利用信息扩散技术，以城市洪涝灾害 为例，开展单一指标的脆弱性分析研究。 2 承灾体脆弱性分析 2.1 承灾体脆弱性的动态分析模式 进行承灾体脆弱性分析时，需要考虑承灾体的 “时”、“空”、“度”等因素。其中，“时”指承灾体遭受 灾害影响的时间，此时间可以是时间点，也可以是 时间段；“空”指承灾体所处的空间地理位置；“度” 指承灾体的受影响程度，比如财产的损失量、建筑 物的倒塌数量和破损程度、人员伤亡的数量等等。 如果用 vi 表示承灾体脆弱性指标，用 h 代表致 灾因子变量，t 为承灾体变化的时间变量。则，承灾 体脆弱性动态评估的基本模式可以用以下示意图表 达（见图 2）。 承灾体一般是指社会经济系统中的经济、人口、 居民地、建筑物、土地、道路、机场、军用设施以 及地下管线等公共基础设施在内的，一切与人类活 动有关的客观存在。人们一般会用城市形态结构、 人口结构、基础设施结构，或是人口密度、人口年 龄结构、经济密度、建筑物密度、生命线工程密度、 道路敏感度等作为承灾体脆弱性的评价指标[15-16]。 图2 承灾体脆弱性动态评估的三维示意图 承灾体的脆弱性直接反映了社会经济系统中不 同层次的承灾体遭遇过或有可能在未来遭遇灾害时 所表现的某种性质，但是由于不同承灾体的度量单 位不一样，往往会造成脆弱性的评估量纲难以统一 标准化。因此，本文建议用经济损失率、受影响人 口比例等表示承灾体的脆弱性指标，即灾情与相应 社会经济发展指标的比值。美国保险业就是用受淹 水深增加时不同类型建筑的损失率来表达建筑物的 洪水脆弱性[17]。如果我们用 d 表示灾情，s 表示社 会经济发展指标，用 v 表示承灾体的脆弱性指标， 则有：v=d/s。 2.2 脆弱性曲线 承灾体脆弱性曲线不是一条主观曲线，而是要 通过大量因果关系研究给出的。评估脆弱性的一种 基本方法是：用同一发展时期的灾害数据统计出一 条脆弱性曲线，再由多个发展时期的多条脆弱性曲 线统计出曲线族的变化规律，然后引入庞加莱-契达 耶夫方程，建立社会发展的广义力学系统运动微分 方程，在相关边界条件下求得理论意义上的脆弱性 曲线，最后再用理论意义上的脆弱性曲线去修订统 计意义上的脆弱性曲线[18-19]。然而，这种方法需要 大量的多领域的历史统计资料，同时建立一系列的 广义力学系统运动微分方程也相当复杂，目前只是 在理论构想阶段，尚没有具体算法和模型出现。 鉴于上述情况，人们利用相对简便易懂的剂量反应函数（dose response functions）来表示致灾因 子强度与受灾程度之间的关系，其几何表达称为剂 量-反应曲线（见图 3）。 published by atlantis press copyright: the authors 167 dynamic vulnerability analysis of population for flood disaster in urban area 图 3 “暴露-死亡率”的剂量-反应曲线示意图 剂量-反应原本是用来表达剂量和不良健康之 间数学关系的一种函数[20-21]。美国科学院国家研究 委员会在 1983 年的“联邦政府的风险评估报告”中， 正式将其用于一般性的风险评估中。 剂量-反应函数的特征表现形式主要有三种，分 别是离散剂量-反应函数、指数分布剂量-反应函数和 概率分布剂量-反应函数。 （1）离散的剂量-反应函数 最简单的剂量-反应函数的特征形式是离散的 剂量-反应函数。关于溃坝洪水灾害分析[22]和地震灾 害中的房屋倒塌破坏性分析 [23] 等相关研究，就是利 用多情景的离散剂量-反应曲线表达不同承灾体的 脆弱性。 （2）指数分布剂量-反应函数 很多情况下，人们是得不到足够的数据完全反 映[0, 1]之间的剂量-反应函数关系的，“反应”也不一 定都是当“剂量”达到某一临界值才会发生变化的， 一些剂量-反应的变化过程是曲线的、循序渐进的。 于是，连续型的剂量-反应函数更多地得以应用。例 如，有研究利用正态分布函数计算人在流水中不稳 定性的不确定性概率问题[24]；也有研究用历史数据 的实证统计分析，指出了洪水水深与死亡率之间的 相关性[25]，可以用上升连续的指数分布函数（曲线） 描述。 （3）概率分布剂量-反应函数 概率分布剂量-反应函数是最常用的剂量-反应 函数，也称之为对数正态分布剂量-反应函数[26]。该 模型认为“剂量”和“反应”之间的关系可以用累计正 态分布表示（见图 4）。这里，概率可以用来描述不 同环境条件下的致命或非致命健康影响。概率分布 剂量-反应函数常用来表示人类在爆炸、毒气扩散、 热辐射等环境下的“反应”情况[27]。 图 4 概率分布/对数正态分布剂量-反应函数 一般的，剂量-反应函数的研究建立在大量的真 实物理实验之上，但是在一些领域（比如自然灾害 领域的人员伤亡和建筑物损毁），不管是从伦理上还 是成本角度上看，进行大量的物理实验显然是不可 能的。因此，我们获得相关数据的途径只能是收集 历史统计数据或开展仿真模拟实验。 2.3 信息扩散理论 从某种意义上讲，人们掌握的用于进行风险分 析的信息往往是不完备的，如果人们能获得足以认 清未来的信息时，风险也就不存在了。针对自然灾 害风险分析领域经常遇到不完备信息的这一情况， 有学者经过长期的试验、验证以及多领域的应用研 究[28-31]，系统地提出了一种处理不完备信息的理论 方法，即信息扩散技术。 信息扩散技术是一种模糊分析技术，它在不假 设样本属于任何分布的前提下，可以尽可能多地挖 掘出数据资料本身所蕴藏的有用信息，提高系统识 别的精度，其解决了从普通样本转变为模糊样本的 问题，从而超越了传统模糊集技术依赖专家选定隶 属函数的随意性，在一定程度上可以保证分析结果 的客观性。该技术可以有效地利用有限信息，较合 理地分析系统中单个因素的变化规律，并可以识别 和近似推理出系统因素间的函数关系（见图 5）。 基于信息矩阵法的函数关系识别和计算过程， 可以归纳为构造信息矩阵、生成模糊关系矩阵和近 似推理三个步骤。 步骤 1：构造信息矩阵 首先，需要准备样本 g，如果 g 中含有 n 个样 本点（n 也称为 g 的样本容量），每个样本点有两个 分量，这两个分量分别是输入 i 和输出 o。则该样本 可记为： 1 1 2 2{( , ) ( , ) ... ( , )}n ng i o i o i o ， ，， 然后，制定样本 i 和 o 对应的监控空间 1 2{ , ,..., }su u u u 和 1 2{ , ,..., }tv v v v 。监控空间 u 和 v 的大小应根据研究需要设定，我们称相邻监控点 之间的距离为步长，用δ表示，δ可以是等长的， 也可以不等长。但原则上监控点的个数须小于样本 容量，以便所有样本点的信息都能全部分配到监控 空间中。 最后，利用分明区间和模糊区间的信息矩阵以 及正态信息扩散扩散函数的计算公式[32]，计算每个 样本点在监控空间上的信息增量 qst，进而组成由 u 和 v“控制”的信息矩阵 im。 1 2 1 11 12 1 2 21 22 2 1 2 t t t s s s st v v v u q q q u q q q im u q q q                      步骤 2：生成模糊关系矩阵 published by atlantis press copyright: the authors 168 xilei pang 自然灾害风险分析领域常常遇到小样本问题， 虽然可使用的样本点不多，但不能因为难以统计出 其规律而放弃研究。样本点少，同样会有一定的规 律，其提供的信息一般可以用来进行模糊关系分析。 假设我们通过小样本计算得到了信息矩阵 im， u 和 v 可以被看作是描述和控制 的两个向量，我们 也可以称以这种形式出现的 u 和 v 为状态因素空 间，因素空间就像是几何中的坐标系，它提供了一 种用来描述事物状态的参考环境。当我们把 u×v 看成是一个因素空间时，v 的元素则可被视为概念， u 可被视为概念的论域，而概念的形式大小则由信 息矩阵 { }jk s tim q  来限定。 令： 1 max{ } 1, 2,...,k jkj s s q k t    ， 且： ( ) 1, 2,...,jkk j k q u j s s   ， 那么，通过信息矩阵 im，v 中的任意一个元素 vk 都可以用由 u 表示的一个模糊集 k 来表示。如果 用 rjk 表示 ( )k ju 、fm 代表由所有 ( )k ju 组成的矩 阵。则我们说通过信息矩阵 im 得到了一个关于 u 和 v 的因果模糊关系矩阵  jk s tfm r  。 步骤 3：近似推理 当我们通过信息矩阵计算得到了一个因果模糊 关系矩阵后，大体上相当于得到了一个函数关系 y=f(x)。然而，模糊关系矩阵只能表达输入-输出的 近似关系，那么就需要用模糊近似推理的办法，将 模糊关系矩阵 fm 用隶属函数的形式表达为 y=f(x) （见图 5） 图 5 模糊近似推理的简单示意图 有关近似推理的原理和具体步骤，请参阅相关 参考文献[32]。 3 城市洪涝灾害的人口脆弱性案例分析 汕头市在赤道低气压带和副热带高气压带之 间，属亚热带海洋性气候，雨量丰沛，但降雨量时 空和地域分布不均匀，呈增多态势。按照国家气象 局的降雨强度（24 小时降雨）等级划分标准，汕头 市历史上（1951 年 1 月 1 日~2010 年 12 月 31 日） 发生过暴雨 348 次、大暴雨 103 次、特大暴雨 4 次， 年均 7.6 次，强降水频率逐年增多。在炎热多雨、地 面排水不畅和防灾减灾措施缺乏等情况下，在经济 发达、人口密集和建筑物林立的区域就容易造成暴 雨洪涝灾害。 基于以上情况，本研究选择暴雨内涝灾害频繁 发生的汕头市为案例区。 3.1 技术路线与数据准备 （1）技术路线 基于收集整理的汕头市历年的人口及洪涝灾情 数据，首先选取受灾人口比例为脆弱性指标；然后 根据政府制定政策的时间规律，以 5 年为单位框定 时间尺度；最后，利用信息扩散技术计算和分析汕 头市洪涝灾害的人口脆弱性动态变化情况。 洪涝灾情数据人口数据 t = t + t ; t <=t 图 6 城市洪涝灾害的人口脆弱性分析流程 （2）时间约束 城市洪涝灾害的承灾体包括人口、建筑物、地 下管线、道路及其他基础设施等。承灾体的空间分 布、数量、大小都会随着时间的变化而发生变化。 大多数情况下，人们会根据研究需要有针对性地选 取相应时间尺度的资料数据进行评估和分析。以人 口为例，有统计意义的人口数据一般是以年为单位 存储的，我国公共管理部门的社会、经济发展规划 又常常会以 5 年为单位进行更新。 结合承灾体可用数据的统计情况，基于以年为 单位的人口、洪涝灾情等数据，并根据研究需要， 从我国经济发展的“五年”计划角度出发，对相关数 据进行分段分析，最终分别以年和“五年计划”为单 位开展人口脆弱性的动态分析。 （3）数据准备 用于承灾体脆弱性分析的数据主要包括研究 区的历年洪涝灾情数据和人口数据。汕头市历年的 洪涝灾害灾情数据（受灾人口）则是通过对汕头市 历年的防灾减灾年鉴、气象志、县志以及汕头市地 方各级政府的三防办、民政局、水利部门和农业局 等部门的统计资料进行收集整理得到。汕头市的人 口数据来自于汕头市历年统计年鉴。 published by atlantis press copyright: the authors 169 dynamic vulnerability analysis of population for flood disaster in urban area 3.2 承灾体脆弱性的动态分析 3.2.1 承灾体脆弱性指标的选取 承灾体的脆弱性直接反映社会经济系统中不同 层次的承灾体在遭遇过或有可能将遭遇灾害是表现 的某种性质，本研究采用受灾人口比例作为城市人 口的脆弱性指标，即受灾人口与区域人口的比值。 表 1 汕头市暴雨洪涝灾害的人口脆弱性指标 时段 年份 降雨量 脆弱性指标 (单位：mm) 七五 1986 146 0.002 1987 153 0.027 1988 185 0.011 1989 130 0.017 1990 198 0.025 八五 1991 119 0.01 1992 105 0.045 1993 141 0.019 1994 182 0.106 1995 152 0.024 九五 1996 117 0.021 1997 150 0.062 1998 103 0.03 1999 139 0.007 2000 80 0.033 十五 2001 113 0.032 2002 118 0.026 2003 131 0.021 2004 114 0.007 2005 128 0.055 十一五 2006 216 0.059 2007 101 0.036 2008 230 0.004 2009 81 0.039 2010 124 0.009 3.2.2 人口脆弱性曲线 在分析人口脆弱性曲线的变化之前，需要先制 定人口脆弱性指标样本和样本监控空间。针对 1986-2010 年的样本，我们用 pv 表示其人口脆弱性 指标样本，用 x 和 y 分别表示样本的两个分量。其 中，x25=[x1,x2,x3,…,x25]表示受灾人口比例，y25=[y1, y 2, y 3,…, y 25]表示降雨量。 用 u24=[u1,u2,u3,…,u25]表示 x25 的监控空间，用 v24=[v1,v2,v3,…,v25]表示 y25 的监控空间，监控空间 u 的步长δu=（maxx-minx）/15，v 的步长为：δv= （maxy-miny）/15。然后，利用正态信息扩散法， 生成由 u24 和 v24 监控的信息矩阵为： 25q  0.002 0.009 0.016 ... 0.093 0.100 0.107 80 0.260 0.530 0.980 ... 0.000 0.000 0.000 90 0.720 1.250 1.890 ... 0.000 0.000 0.000 100 1.600 2.480 3.310 ... 0.000 0.000 0.000 ... ... ... ... ... ... ... ... 210 0.610 0.790 0.860 ... 0.090 0.140 220 230 0.160 0.740 0.790 0.680 ... 0.030 0.040 0.050 0.800 0.790 0.570 ... 0.010 0.010 0.010                       然后，将信息矩阵 q25 转化为模糊关系矩阵 r25， 则“1986-2010”期间的降雨量与受灾人口比例的模糊 关系矩阵为： 25r  0.002 0.009 0.016 ... 0.093 0.100 0.107 80 0.067 0.096 0.156 ... 0.000 0.000 0.000 90 0.185 0.226 0.300 ... 0.000 0.000 0.000 100 0.411 0.449 0.525 ... 0.000 0.000 0.000 ... ... ... ... ... ... ... ... 210 0.157 0.143 0.137 ... 0.220 0.206 220 230 0.203 0.190 0.143 0.108 ... 0.073 0.059 0.063 0.206 0.143 0.090 ... 0.024 0.015 0.013                       事实上，当我们得到了降雨量与受灾人口比例 的模糊关系矩阵时，已经大致可以看出两者之间的 变化关系了（见图 7）。 图 7 中方格的色彩值从蓝色到红色，表示降雨 强度与受灾人口比例的关系由弱到强。由上图可以 看出，受灾人口比例的变化并不是随着降雨量的升 高而增加的，也就是说降雨量和受灾人口比例之间 并不是呈明显的正比或反比线性关系。为了能够通 过有限的数据，较清晰地表达致灾因子与承灾体之 间的关系，也就是脆弱性的变化趋势，我们需要将 相对模糊的关系矩阵转化为一维的向量，并转化为 更直观的曲线形式予以表达。 图 7 降雨强度与受灾人口比例的模糊关系色谱图 根据函数关系的模糊推理原理，对模糊关系矩 阵 r25 进行近似推理，计算出 1986-2010 期间的人口 脆弱性变化向量 l25： published by atlantis press copyright: the authors 170 xilei pang l25=[0.041,0.039,0.038,0.037,0.038,0.041,0.045,0.051, 0.060,0.070,0.075,0.074,0.069,0.064,0.060,0.055] l25 实质上是经过信息扩散和近似推理得到的 “受灾人口比例”在“降雨量”监控下的趋势向量，这个 趋势向量就是人口脆弱性变化的数学表达，其曲线 形式见图 8。 图 8 汕头市洪涝灾害人口脆弱性变化曲线 (1986-2010) 从图 8 中，我们就可以看到受灾人口比例与降 雨强度之间的波动变化关系，但该曲线没有能够反 应出脆弱性曲线随着时间的推移而发生的动态变化 情况。因此，下面我们将 1986-2010 平均划分为 5 个阶段，通过对 5 个阶段各自脆弱性的变化分析， 找出人口脆弱性在时间轴上的动态变化性。 按照上述计算方法，分别计算出 “七五”到“十 一五”各个阶段汕头市的降雨量与受灾人口比例之 间的模糊关系矩阵和人口脆弱性变化向量，并得到 其变化曲线（见图 9-图 13）。 通过对 5 个“五年”计划期间的人口脆弱性分析， 我们分别得到了汕头市的“七五”、“八五”、“九五”、 “十五”和“十一五”期间的洪涝灾害人口脆弱性曲线， 这 5 条形状各异的脆弱性曲线，不仅反映了每个五 年计划期间不同降雨强度条件下的人口脆弱性变化 情况，同时也指出了 5 个五年计划时段之间的脆弱 性变化也存在着很大的差异，而且人口脆弱性的变 化并不是随着降雨强度的增加而线性增大或者线性 减小的。也就是说，降雨和受灾人口之间存在着一 种复杂的模糊关系，通过对已有的不完备信息进行 推理计算，就可以得到一些离散的洪涝灾害人口脆 弱性值（见表 8），将其连接，便得到了近似意义上 的脆弱性曲线。 图 9 “七五”期间汕头市洪涝灾害人口脆弱性曲线 图 10 “八五”期间汕头市洪涝灾害人口脆弱性曲线 图 11 “九五”期间汕头市洪涝灾害人口脆弱性曲线 图 12 “十五”期间汕头市洪涝灾害人口脆弱性曲线 表 8 汕头市洪涝灾害人口脆弱性动态变化值 降雨均值 降雨强度（mm） 时间段 脆弱性值 脆弱性均值 165 mm 130 153 176 199 1986-1990 0.014 0.015 0.015 0.019 0.016 144 mm 105 131 157 183 1991-1995 0.048 0.033 0.046 0.087 0.053 116 mm 80 104 128 152 1996-2000 0.036 0.030 0.026 0.042 0.034 122 mm 113 119 125 131 2001-2005 0.024 0.026 0.042 0.036 0.032 157 mm 81 131 181 231 2006-2010 0.031 0.022 0.037 0.032 0.031 171 dynamic vulnerability analysis of population for flood disaster in urban area 图 13 “十一五”期间汕头市洪涝灾害人口脆弱性曲线 3.2.3 时间轴上的人口脆弱性动态分析 为了更清晰地表达人口脆弱性和降雨强度之 间的关系，以及脆弱性随时间推移的变化规律，将 5 个不同时期的人口脆弱性曲线合制成为三维立体图 （见图 14）。 图 14 汕头市洪涝灾害的人口脆弱性动态变化曲线 从图 14 中，我们可以看出每条脆弱性曲线都是 由 4 个离散点连接而成的，但控制这些曲线变化的 降雨强度区间值是不一样的。 由表 8 可知，“七五”（1986-1990）期间的脆弱 性值变化范围是从 0.014 到 0.019，其对应的降雨强 度变化范围是从 130mm 到 199mm；“八五”期间的脆 弱性值范围是从 0.033 到 0.087，其对应的降雨强度 范围是 105mm 到 183mm；“九五”期间的脆弱性值范 围是从 0.026 到 0.042，其对应的降雨强度范围是 80mm 到 152mm；“十五”期间的脆弱性值范围是从 0.024 到 0.042，其对应的降雨强度范围是 113mm 到 131mm； “十一五”期间的脆弱性值范围则是从 0.022 到 0.037，其对应的降雨强度范围是从 81mm 到 231mm。 如果我们分别求出汕头市“七五”、“八五”、“九 五”、“十五”和“十一五”期间的降雨强度均值和脆弱 性均值，则可以得到均值意义上的随时间变化的人 口脆弱性动态变化曲线。因为 5 个发展时期的降雨 均值均大于 100mm 小于 250mm，属于大暴雨范围， 因此我们称此时的脆弱性曲线是大暴雨情景下的人 口脆弱性变化曲线（见图 15）。 图 15 显示了从 1986-2010 期间汕头市暴雨洪涝 灾害人口脆弱性的变化趋势是先上升后减小，然后 趋于平缓。 图 15 大暴雨（均值）条件下的汕头市洪涝灾害人口 脆弱性动态变化曲线 基于表 8，也可以找出各个时期降雨强度相近 （130±2mm）的脆弱性值，将这些脆弱性值绘成时 间轴上的人口脆弱性动态变化曲线后（见图 16），可 以看到在 130mm 左右降雨强度条件下的人口脆弱性 曲线变化趋势与大暴雨均值意义下的人口脆弱性变 化趋势稍有不同，即从“七五”到“九五”，人口脆弱性 变化的趋势都是先升后降，但到了“十五”和“十一 五”，均值意义上的脆弱性曲线基本趋于平滑，变化 幅度不大，而 130mm 降雨条件下的脆弱性曲线却是 从“九五”到“十五”升高，然后从“十五”到“十一五” 降低。 图 16 定量降雨（130mm）条件下的汕头市洪涝灾害 人口脆弱性动态变化曲线 4 结论与讨论 从1986-2010期间的多条脆弱性曲线的变化上 可以看出：汕头市洪涝灾害的人口脆弱性并不是随 着经济的发展而持续降低或持续升高；也不是人口 多了脆弱性就大，人口少了脆弱性就小；更不是降 雨量越大脆弱性就越大。同时，致灾因子强度均值 意义上的脆弱性动态变化曲线和固定强度值对应的 脆弱性动态变化曲线也存在差异。在科学计算的前 提下，脆弱性研究结果应该多样化，以供决策者在 决策分析时可以根据自身需要多一些选择。 在没有主观赋分的前提下，利用信息扩散技术 和小样本数据分析得出的上述结果，一方面说明人 口脆弱性的发展变化不只是单纯随着时间的变化、 人口的增长和致灾因子强度的大小而有规律的变化 的。另一方面说明承灾体脆弱性的变化受诸多因素 影响（如：居民地的易损性、人口结构、人员流动、 published by atlantis press copyright: the authors 172 xilei pang 人口分布等），致灾因子的变化并不能控制脆弱性的 变化。由于目前尚没有出现验证灾害风险评估和脆 弱性分析结果有效性的科学手段，本文只能从理论 角度展开研究，希望能够为自然灾害的脆弱性评估 研究提供一种新的思路。 注：脆弱性和风险是两个不同的概念，自然灾 害风险分析过程是致灾因子、承灾体和孕灾环境等 多因素综合分析的系统工程，本文仅仅针对城市暴 雨洪涝灾害中的人口这一单一承灾体进行了脆弱性 分析研究，而非系统性的风险分析。 参考文献 [1] allianz. allianz risk pulse-focus: floods[r]. 2011，1-6 [2] wilhelmi o v, wilhite d a. assessing vulnerability to agricultural drought: a nebraska case study[j]. natural hazards，2002，25：37-58 [3] dow k. exploring differences in our common futures: the meaning of vulnerability to global environmental change[j]. geoforum，1992，23： 417-436 [4] gilberto c gallopin. linkages between vulnerability, resilience and adaptive capacity. global environmental change，2006，16：293-303 [5] 石勇. 灾害情景下城市脆弱性评估研究[d]. 博 士学位论文，上海：华东师范大学，2010 yong shi. research on vulnerability assessment of cities on the disaster scenario[d]. doctoral dissertation, shanghai: east china normal university, 2010 [6] 张灵,陈晓宏,王兆礼. 防洪系统脆弱性的模糊 极大熵诊断[j]. 系统工程理论与实践，2011， 31（8）：1600-1606 zhang ling, chen xiaohong, wang zhaoli. fuzzy diagnosis of vulnerability to flood based on maximum entropy[j]. systems engineering theory & practice, 2011, 31(8): 1600-1606 [7] 史培军. 四论灾害系统研究的理论与实践[j]. 自然灾害学报，2005，14(6)：1-7 shi peijun. theory and practice on disaster system research in a fourth time[j]. journal of natural disasters, 2005，14(6)：1-7 [8] kröger w. critical infrastructures at risk: a need for a new conceptual approach and extended analytical tools[j]. reliability engineering & system safety，2008，93(12)：1781-1787 [9] kellenberga d. k. and mobarakb a.m. does rising income increase or decrease damage risk from natural disasters[j]. journal of urban economics，2008，63(3)：788-802 [10] 陈颙,彭文涛,徐文立. 21 世纪地震灾害的一些 新特点[j]. 地球科学进展，2004，19(3)： 359-363 chen yong, peng wentao, xu wenli. new feature of earthquake disaster in 21st century[j]. advance in earth sciences, 2004， 19(3)：359-363 [11] 刘吉夫,陈颙,史培军等. 中国大陆地震风险分 析模型研究(ⅱ)：生命易损性模型[j]. 北京师 范大学学报(自然科学版)，2009，45(4)： 404-407. liu jifu, chen yong, shi peijun et. al. study on seismic risk assessment in mainland china (ii): life vulnerability model, journal of beijing normal university (natural science) , 2009，45(4)：404-407 [12] 苏桂武,高庆华. 自然灾害风险的分析要素[j]. 地学前缘，2003，10 (特刊)：272-279 su guiwu, gao qinghua. the elements for analyzing natural disaster risk[j]. earth science frontiers，2003，10 (suppl)：272-279 [13] 赵思健. 情景驱动的区域自然灾害风险分析 原理与应用[d]. 博士后出站报告，2010，北 京：北京师范大学 zhao sijian. scene-driven risk analysis to regional natural disasters[d]. postdocteral research report, 2010, beijing: beijing normal university [14] 黄崇福. 综合风险评估的一个基本模式[j]. 应用基础与工程科学学报，2008，16(3)： 371-381 huang chongfu. a basic model for assessing integrated risk. journal of basic science and engineering, 2008，16(3)：371-381 [15] 石勇,石纯,孙蕾等. 沿海城市自然灾害脆弱性 评价研究-以上海浦东新区为例[j]. 中国人口. 资源与环境，2008，18(4)：24-27 shi yong, shi chun, sun lei et.al.. vulnerability assessment of natural disasters in pudong district[j]. china population, resources and environment, 2008，18(4)：24-27 [16] 张斌,赵前胜,姜瑜君. 区域承灾体脆弱性指标 体系与精细量化模型研究[j]. 灾害学，2010， 25（2）：36-40 [17] zhang bin, zhao qiansheng and jiang yujun. research on indexes system about regional vulnerability of hazard-affected bodies and fine spatial quantitative model[j]. journal of catastrophology，2010，25(2)：36-40 [18] 石勇,许世远,石纯等. 洪水灾害脆弱性研究进 展[j]. 地理科学进展，2009，28(1)：41-46 shi yong, xu shiyuan, shi chun et.al. a review on development of vulneralbility assessment of floods[j]. progress in geography, 2009，28(1)：41-46 [19] huang c.f. differences between disaster prediction and risk assessment in natural disasters [j]. human and ecological risk assessment，2011，17(3)：542-553 [20] burton i, kates r w, white g f. the environment as hazard[s]. oxford univ. press， oxford，1978 [21] covello v.t., merkhoff er m.w. risk assessment methods, approaches for assessing health and environmental risks[m]. plenum press，new york，1993 [22] graham, w.j. a procedure for estimating loss of life caused by dam failure[r]. dam safety office report，dso-99-6，1999 [23] coburn a, spence r. earthquake protection （2nd edition）[m]. john wiley & sons，2002， chichester [24] lind n., hartford d. probability of human instability in a flooding: a hydrodynamic model[a], in: melchers e and stewart m.g. (ed) published by atlantis press copyright: the authors 173 dynamic vulnerability analysis of population for flood disaster in urban area proc. of icasp 8, applications of statistics and probability[c]，2000，1151-1156 [25] jonkman s.n. methode voor de bepaling van het aantal slachtoff ers ten gevolge van een grootschalige overstroming[r]. dww report， 2004-042，2004 [26] finney d.j. probit analysis（3rd edition）[m]. london cambridge university press，1972 [27] cpr methods for the determination of possible damage: people and goods through the release of hazardous substances[m]. cpr， voorburg，netherlands，1990 [28] chongfu huang, claudio moraga. extracting fuzzy if-then rules by using the information matrix technique[j]. journal of computer and system sciences，2005，70：26-52 [29] chongfu huang. information dissusion techniques and small-sample problem[j]. international journal of information technology & decision making. 2002，1(2)：229-249 [30] huang chongfu. information matrix and application[j]. international journal of general systems，2001，30(6)：603-622 [31] 黄崇福. 非完备样本知识的优化处理[j]. 北 京师范大学学报(自然科学版)，1992，28（2）： 129-135 huang chongfu. optimality processing to the sample knowledge of non-completeness[j]. journal of beijing normal university (natural science), 1992，28(2)：129-135 [32] 王家鼎. 地理学研究中的模糊信息优化处理 方法[j]. 地理学与国土研究，1999，15(1)： 75-80 wang jiading. fuzzy information optimization technology for geography research[j]. geography and territorial research，1999， 15(1)：75-80 [33] 黄崇福. 自然灾害风险评价：理论与实践[m]. 北京：科学出版社，2005 huang chongfu. risk assessment of natural disaster: theory & practice[m]. beijing: science press, 2005 published by atlantis press copyright: the authors 174 dynamics of urban fire correlations with detrended fluctuation analysis jinghong wang 1,2 1 state key laboratory of fire science, university of science and technology of china, hefei 230026, china, arain@mail.ustc.edu.cn 2 ustc-cityu joint advanced research centre (suzhou), suzhou 215123, china abstract to explore and understand the intrinsic dynamic mechanisms of urban fire system is an important prerequisite for the fire risk analysis and management. by analyzing the time series of urban fire system, it is helpful to achieve this goal. in this paper, the detrended fluctuation analysis (dfa) is applied to study the long-range correlations of urban fires in both small and large spatial scales with a case of chinese city. it is found that the time series of direct economic loss caused by urban fires presents stable anti-power law correlations, while the time interval series of urban fire occurrence presents persistent long-range power-law correlations with two scaling exponents. based on the characteristics of these scaling exponents, the intrinsic dynamics of urban fires is discussed. it is found that the effect of human activities on the dynamics of urban fire system may become prevalent at small timescales, while at large timescales the effect of human activities is not obvious in influencing the long-range correlation patterns of urban fire system, which may be controlled mainly by some general characteristics of a city such as the meteo-climatic conditions. keywords: urban fire, dynamics, detrended fluctuation analysis, long-range correlation. 基于去趋势波动分析的城市火灾相关性动态研究 王静虹 1,2 1.中国科学技术大学火灾科学国家重点实验室，合肥 230026 2.中科大-香港城大联合高等研究中心（苏州），苏州 215123 摘要：对城市火灾系统的内在动力学机制进行探讨和研究，是城市火灾风险分析与管理的重要前提，而对 城市火灾系统的时间序列进行分析，则有助于实现这一目标。本文运用去趋势波动方法，在不同的空间尺度上对 城市火灾发生及损失规模的时间序列进行分析，探讨了城市火灾在时间和空间上的长程相关性特征。研究发现， 城市火灾造成的直接经济损失时间序列表现出稳定的反幂律相关性，而城市火灾发生的时间间隔序列表现出稳定 的长程幂律相关性，且这种长程相关性具有两个分段的标度指数。基于这种分段标度指数的特征，本文进一步讨 论了城市火灾系统的内在动力学机制，发现在较短的时间尺度上，人类活动对城市火灾系统动态特征的影响应是 普遍存在的，而在较长的时间尺度上，人类活动对城市火灾系统的这种长程相关性特征并无明显的影响，而是一 些城市的普遍特征如气象气候条件等起着主导作用。 关键词：城市火灾，动态特征，去趋势波动分析，长程相关性 1. 引言 城市火灾系统即城市火灾和它的影响因素所构成 的整体，是一个复杂系统,涉及到人类活动、城市结构、 社会经济、气象环境等多种因素。城市火灾与这些因 素之间存在着复杂的相互作用,城市火灾事件之间也 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 126-132 published by atlantis press copyright: the authors 126 存在着相互关联。对城市火灾系统进行科学合理的风 险评估和预测，是关系到城市火险控制及保障居民生 命财产安全的重要工作。作为一个复杂的非线性动力 学系统，城市火灾系统通常以其原始的输出信号（火 灾发生时间及规模）表现出一种非稳定性的随机特征。 因此，在分析城市火灾的内在动力学机制、预测未来 的发展趋势以及研究影响城市火灾发生的各种社会经 济、人类活动和环境因素之前，必须首先找出城市火 灾在时间和空间尺度上的分布形式。已有研究发现， 城市火灾系统在时空尺度上呈现幂律分布的形式，即 具有一种自组织临界性（soc）特征 [1-3] 。这些研究大 多基于一些大空间范围内的原始火灾数据，例如，在 [4] 的研究中，采用了中国和日本全国范围内的火灾数据 进行了火灾分布的对比研究，其研究结果对于整体上 理解城市火灾系统的分布特征具有积极的意义。然而， 当涉及到某个具体的城市乃至城市中某个具体的区域 时，这些研究所提供的方法并不足以帮助我们了解其 具体的火灾分布形式以及内在的动力学机制。目前， 基于社区的防火概念（community-oriented fire protection）已经越来越受到国内外的重视 [5] ，其中的 一个重要方面就是需要在更小的空间尺度上对城市火 灾系统进行更为精细的分析，进而更有助于掌握某些 具体城市区域的火险特征，以及区域和整体的火灾分 布形式之间的区别和联系。 从分析方法上来说，已有的研究通常采用传统的 功率谱方法和hurst分析方法对城市火灾的静态数据序 列进行分析。这些方法对于序列的变化趋势和分形特 征有着较好的分析效果，且在不考虑一个较长时间尺 度下序列背景发生变化的前提下，其结果具有较高的 可靠性。但是城市火灾系统附属于城市系统之中，其 时间序列本身具有很强的不稳定性，这种不稳定性往 往与城市结构的变化、社会经济的发展等背景因素紧 密耦合，因此采用传统的分析手段，其结果的可靠性 将在很大程度上受到这些不稳定因素的影响，同时也 无法反映出城市火灾系统在某一时间尺度内演化的长 程特征 [6] 。一种优化的方法就是peng [7] 等人提出的去趋 势波动（detrended fluctuation analysis，简称dfa）分 析方法。该方法可以有效挖掘出一个看似不稳定序列 内部蕴含着的自相似性，以及序列是否具有长程相关 性。与传统方法相比，dfa方法最大的优势在于可以 消除序列的局部趋势，从而避免将时间序列的短程相 关、非平稳性虚假地检测为长程相关性。该方法目前 已经被广泛运用于诸多领域，如生物医学领域中的 dna序列研究 [8] 、心率变异性研究 [9,10] 、气候 [11,12] 、经 济领域 [13,14] 、地震 [15] 、森林火灾 [16] 等，但是在城市火 灾领域尚较少见诸报导。本文运用dfa方法，分析了 某些城市区域中火灾时空分布形式的长程趋势，以及 这种长程趋势与城市区域特征之间的定性联系，以期 对城市火灾系统的内在动力学机制进行一个初步的探 讨。 2. 数据和方法 本文分析的数据来源于安徽省消防总队，是合 肥市 2000-2009 年的全部火灾数据，数据包括起火 时间、起火场所、火灾原因及直接财产损失等。运 用 dfa 方法对合肥市全市以及不同区域的火灾时间 序列进行长程相关性分析，同时为解释这种长程相 关性的潜在原因，或者说动力学机制，文中进一步 考虑了不同城市区域的特征，如人口、社会经济活 动等因素。 dfa方法已被证明可以有效揭示时间序列的长 程相关性程度。对于一个非稳定时间序列，应用dfa 方法进行分析的主要步骤为： 首先计算该时间序列的累积离差：    k i avexixky 1 ])([)( (1) 其中 avex 是该序列中所有数据的平均值。接着将 这个新的序列 )(ky 等分成n/n个具有同样长度n的区 间，在每个区间中通过最小二乘法进行拟合，所获 得的拟合直线斜率代表了该区间中数据的趋势。用 )(kyn 表示每个区间中通过拟合获得的趋势函数，则 去趋势后的子区间时间序列为 )()( kyky n 。最后计 算去趋势后时间序列的波动均方根：    n k n kyky n nf 1 2 )]()([ 1 )( (2) 如果每个区间中的拟合多项式 )(kyn 采用了 published by atlantis press copyright: the authors 127 线性、二次、三次乃至更高阶m的多项式，那么就将 算法分别记为dfa1，dfa2，dfa3，…dfam等。显 然m阶的dfa滤去了累积离差中的m阶趋势成分以 及原始序列中的m-1阶趋势成分。本文采用dfa3进 行数据的分析，可以消除序列中的线性、二阶和三 阶趋势。n的取值范围与原始序列的长度n有关，初 始值（最小值）设置均为4，此时的区间数（n/n） 最多；将希望得到的区间数最小值设置为4，此时对 应的n值即为最大值。根据这个算法，就可以获得f(n) 和n之间的关系。 在不同的时间尺度（不同的区间长度n）上重复 这样的算法，就可以获得平均波动程度f(n)与时间尺 度n之间的关系。通常f(n)会随着时间尺度n的增加而 递增。如果在双对数坐标中f(n)与n存在幂律关系（自 相似性），即  nnf )( (3) 此处的拟合直线斜率α，就可以表征该时间序列 的标度指数，表明该时间序列中的信号并不代表纯 粹的随机过程，而是具有一定的长程相关性特点。 不同的α值表示原始时间序列具有不同的分布形式： 当α=0.5时，该序列即为白噪声； 当α=1.0时，该序列即为1/f噪声； 当0.5<α< 1时，说明该序列具有稳定的长程幂律 特征，如果在时刻t以前存在上升(或下降)趋势，那 么在时刻t以后总体上也存在上升(或下降)趋势，即 过去发生的事件对未来事件的发生产生影响； 当0 <α< 0.5时，该序列表现出反相关性，或具 有非持久性； 当α> 1时，时间上的长程相关性存在，但是不 再服从幂律分布。 3. 结果和讨论 3.1 全市火灾数据的dfa分析结果（较大空间尺度） 合肥市位于北纬31.52 °、东经117.17 °，作为安徽省省 会，是全省政治、经济、文化、金融和商业中心，同 时也是长江流域的一个重要城市。合肥全市下辖瑶海 区、庐阳区、蜀山区和包河区四个中心城区，肥东县、 肥西县和长丰县三个郊县（在本文中不作为分析对 象），以及高新技术产业区和经济开发区两个经济开 放区域。合肥市行政区划如图1所示。 图1 合肥市行政区划示意图 图2所示为合肥市2000-2009年间共6328起火灾的 直接经济损失时间序列。为了判断城市火灾的长程相 关性是否受特殊值的影响，比如某一年中损失数值特 别大的重特大火灾，本文除了真实火灾序列之外，根 据真实火灾损失数据构建了一个随机序列，即将真实 序列中非零的数值用(0, 10 6 )这样一个随机分布进行取 代。表1列出了合肥市2000-2009年间每年单起火灾造 成的直接经济损失前三大的数值，可以看到仅有很少 的数据（以下划线标出）超出了百万元量级，换句话 说，百万元这个量级可以反映合肥市2000-2009年火灾 造成的直接经济损失的一个普遍最大数量级水平，因 此将随机分布函数选择为(0, 10 6 )。图3所示为真实序列 和随机序列的dfa分析结果。可以看到，虽然图中两 条曲线整体形状并不完全一致，但是拟合直线的斜率 几乎一致。真实序列和随机序列的拟合直线斜率分别 为0.48和0.49，均小于0.5，这就说明城市火灾的损失 规模具有长程时间尺度上的反相关性，即所分析序列 具有非持久性，过去发生的事件对未来事件具有不确 定的影响。城市火灾损失规模受多种不确定因素的影 响，一些极端阈值的出现往往受到各种不确定因素的 作用，这也正体现了火灾损失规模序列的非持久性特 征。但不管是真实序列还是模拟的随机序列，这种空 published by atlantis press copyright: the authors 128 间尺度上的长程特征均比较稳定存在，而损失规模一 定程度上反映了火灾在空间上的分布特征，由此可以 认为，城市火灾在空间上的长程特征应是受到某种动 力学机制的作用，即与时间上的分布形式有所联系。 因此，本文进一步对城市火灾发生的时间间隔序列进 行了去趋势波动分析，以发掘其在时间尺度上的分布 特征。 表1. 2000-2009年合肥市每年单起火灾造成的前三大 直接经济损失（万元） 年份 最大 第二大 第三大 2000 2178.94 64.35 18.98 2001 62.06 18.68 12.00 2002 2692.09 56.10 24.28 2003 33.36 27.04 16.54 2004 80.42 38.05 28.71 2005 25.08 25.01 17.72 2006 33.60 29.23 23.52 2007 445.22 17.05 15.66 2008 654.87 115.62 87.76 2009 500.00 231.90 50.00 图2 合肥全市2000-2009年火灾时间序列 横轴代表每起火灾的发生时间，单位为分钟；竖轴 代表每起火灾造成的直接经济损失，单位为元人民 币。 图3 合肥全市2000-2009年火灾直接经济损失时间 序列的dfa分析结果 根据合肥市2000-2009年间6328起火灾的发生时 间， 可以形成一个时间间隔序列，其中的每一个数据 均代表两起时间上相邻的火灾事件的等待时间。运用 dfa方法分析这个时间间隔序列，结果如图4所示。可 以看到，全市火灾时间间隔序列表现出分段的标度指 数特征，分段处约在64天左右的位置，短程标度α1为 0.73，小于长程标度α2为0.97。与此类似，对单个年份 的全市火灾时间间隔序列进行dfa分析，选取2007、 2008和2009三年的火灾数据进行分析，表2所示为这三 年每年发生的火灾起数。图5所示为这三年火灾时间间 隔序列的标度指数特征，发现单个年份的火灾时间间 隔序列同样具有分段的标度指数特征，分段处约在15 天左右的位置，如图5所示。2007年的短程标度指数α1 为0.58，长程标度指数α2为0.74；2008年分别为0.54和 0.75；2009年分别为0.56和0.72。短程标度指数和长程 标度指数均具有较好的稳定性，且每一年份的短程标 度指数α1均小于长程标度指数α2。α1 和 α2均在0.5~1.0 之间，表明全市火灾的时间间隔序列具有稳定的长程 幂律相关性。 published by atlantis press copyright: the authors 129 图4 2000-2009年全市火灾发生时间间隔序列的dfa 分析结果 表2. 2007-2009年合肥市每年火灾次数 year no. of fires 2007 1203 2008 1458 2009 1375 图5 单个年份全市火灾发生时间间隔序列的dfa分 析结果 图4和图5的结果显示，对于城市火灾发生时间 间隔，在整个分割区间上存在交叉点，即图中箭头 所指处，这说明城市火灾时间序列存在两个不同数 值的标度指数。某个系统的标度指数α并不总是常 数，经常会随着不同的标度范围而改变，即出现交 叉点。其原因通常认为是由于在不同时间和空间标 度下，信号的相关性质发生改变而引起的，反映不 同的物理含义。但在本文中，在整个分割区间内城 市火灾发生时间间隔序列的两个标度指数的数值都 在0.5~1之间，是同一性质的标度指数，只是强弱程 度的区别。这说明城市火灾发生时间的序列具有长 期记忆特征，当前发生的城市火灾和未来火灾事件 之间存在长期相关性，从而在时间尺度上根据现在 及以前的城市火灾对未来的城市火灾进行预测是科 学可行的，且所分析的时间区间越长，结果会越合 理。 3.2 具有不同社会经济特征的城市区域火灾数据的 dfa分析结果（较小空间尺度） 一个城市往往具有不同的功能区域，对于基于社 区的防火概念来说，需要找出不同城市区域中火灾发 生的特征，而不仅仅是关注于整个城市的火灾数据。 因此，本文将dfa方法进一步应用于分析2000-2009 年合肥市的六个不同辖区内的火灾时间序列，结果 如图6所示。发现这六个辖区的火灾分布与全市的 火灾分布具有一致的形式，即每个辖区内的火灾发 生时间间隔序列均表现出一种稳定的长程相关性， 且具有分段的标度指数特征，分段处的位置分别 为：蜀山区23天，庐阳区32天，包河区38天，瑶海 区25天，高新技术产业区16天，经济开发区13天， 均小于图4中全市火灾时间序列分段点处的分割时 间值，64.8天，这说明城市火灾的空间背景越大， 对其发生时间进行分析和预测时所需要的过去的连 续事件数越多。对于这四个中心城区，短程标度指 数α1数值在0.65至0.72，与全市火灾发生时间间隔 序列的短程标度指数0.73十分接近，而长程标度指 数α2数值在0.81至0.91，比全市火灾序列的长程标 度指数0.97稍小。对于两个新近设立的开发区，短程 标度指数与长程标度指数比较接近，且短程标度指 数均明显高于四个中心城区，而长程标度指数与四 个中心城区及全市的火灾序列比较类似，均接近于 1.0，差距不大。 published by atlantis press copyright: the authors 130 四个中心城区即包河区、蜀山区、庐阳区和瑶 海区几乎包括了合肥市90%的居住人口，以及全市的 大部分商业活动、社会活动和传统工业活动，而高 新技术产业区和经济开发区主要通过一些高科技产 业来促进合肥经济的发展，在这两个区域中无论是 人口还是传统的社会活动，都大大少于四个中心城 区。结合四个中心城区与两个开发区在短程标度指 数上的明显差别，可以认为与人口和社会经济因素 密切相关的人类活动对城市火灾在较短时间尺度上 的动态特征会造成影响。因此，在这两个开发区 中，能在较短时间尺度上体现出来的人口与社会经 济因素对城市火灾分布的影响并不明显，从而短程 标度指数α1与长程标度指数α2在数值上更为接近， 亦即在这两个区域内，无论是从较短还是较长时间 尺度下进行分析，其城市火灾的动态发展是类似 的。然而，对于四个中心城区来说，短程标度指数 α1明显小于长程标度指数α2，这就意味着在这四个 区域中，当从较短时间尺度进行分析时，城市火灾 的分布形式与人口及社会经济方面的因素具有比较 强的内在联系。当从较长时间尺度进行分析时，全 部六个区域均表现出稳定的长程相关性，长程标度 指数接近于1.0，这就意味着在长时间尺度上，城市 火灾系统的动态特征与人类活动关系不大，而是主 要受到一些与城市的普遍特征如城市结构、气候等 因素有关的内在机制的控制。 (a) (b) 图 6 2000-2009 年不同区域城市火灾发生时间间隔 序列的 dfa 分析结果 4 结论 本文运用去趋势波动分析方法，在不同的空间尺 度上对一个典型的中国发展中城市合肥的火灾发生及 损失规模时间序列的长程相关性特征进行了分析，发 现城市火灾造成的直接经济损失时间序列在长程时间 尺度上表现出稳定的反幂律相关性，即所分析序列具 有非持久性，过去发生的事件对未来事件具有不确定 的影响。损失规模一定程度上反映了火灾在空间上的 分布特征，而这种空间上的长程特征应是受到某种动 力学机制的作用，即与时间上的分布形式有所联系。 进一步的研究发现，城市火灾在时间上确实具有长程 相关性的特征。通过对比研究全市及不同城市区域的 火灾序列，发现无论是在全市的大空间尺度范围或是 某个区域的小空间尺度范围内，城市火灾发生的时间 间隔序列都表现出稳定的长程幂律相关性，且具有分 段的标度指数特征，短程标度指数α1与长程标度指数 α2均位于0.5至1.0之间。 结合这种分段标度指数的特征，本文进一步讨论 了城市火灾系统的内在动力学机制。对于两个经济开 发区来说，短程标度指数α1明显高于四个中心城区， 而长程标度指数α2与四个中心城区及全市相差不大， 均接近于1.0。由于四个中心城区与两个经济开发区之 间的主要差异在于人口规模及社会经济活动等方面， published by atlantis press copyright: the authors 131 因此本文认为人类活动对城市火灾系统的动态影响主 要普遍存在于较短的时间尺度上，而在较长的时间尺 度上，人类活动对城市火灾系统的这种长程相关性特 征并无明显的影响。在较长时间尺度上，城市火灾系 统的动态特征主要受到一些与城市的普遍特征如城市 结构、气候等因素有关的内在机制的控制。 致谢 本文受中国香港特别行政区研究委员会项目基金资助 （项目号：cityu118708），在此表示感谢。 另：本文的原始英文版本已发表在第三届风险分析与 危机响应国际会议论文集。 参考文献 [1] bak p, tang c, wiesenfeld k. self-organized criticality: an explanation of 1/ f noise. phys rev lett, 1987, 59: 381-384 [2] hwa t, kardar m. fractals and self-organized criticality in dissipative dynamics. phys d, 1989, 38(1-3): 198-202 [3] song w g, wang q, wang j. characteristics of fire distribution in huludao city and its enlightenment to city fire danger analysis. sci & technol rev, 2005, 23(8): 35-37 [4] wang j, song w g, zheng h y. temporal scaling behavior of human-caused fires and their connection to relative humidity of the atmosphere. ecol model, 2010,221(1):85-89 [5] guo t n, fu zh m. the fire situation and progress in fire safety science and technology in china[j]. fire safety j, 2007,42:171-182. [6] j alvarez-ramirez, r escarela-perez, g espinosa-perez, et al. dynamics of electricity market correlations. phys a, 2009,388:2173-2188. [7] peng c k, buldyrev s v, havlin s, et al. mosaic organization of dna nucleotides. phys rev e, 1994,49:1685-1689. [8] nagarajan r, upreti m, govindan r b. qualitative assessment of cdna microarray gene expression data using detrended fluctuation analysis. phys a, 2007,373:503–510. [9] perakakis p, taylor m, martinez-n e, et al. breathing frequency bias in fractal analysis of heart rate variability. biol psychol, 2009,82(1): 82-88 [10] stubsjøen s m, bohlin j, skjerve e, et al. applying fractal analysis to heart rate time series of sheep experiencing pain. physiology & behavior, 2010,101(1): 74-80 [11] koçak k. examination of persistence properties of wind speed records using detrended fluctuation analysis. energy, 2009,34(11):1980-1985 [12] yuan n m, fu z t, mao j y. different scaling behaviors in daily temperature records over china. phys a, 2010,389(19): 4087-4095 [13] yudong wang, li liu, rongbao gu, et al. analysis of market efficiency for the shanghai stock market over time. phys a, 2010,389(8):1635-1642 [14] rongbao gu, hongtao chen, yudong wang. multifractal analysis on international crude oil markets based on the multifractal detrended fluctuation analysis. phys a, 2010,389(14): 2805-2815 [15] l. telesca, lovallo m, lapenna v, et al., long-range correlations in two-dimensional spatio-temporal seismic fluctuations, phys a, 2007,377:279–284. [16] hongyang zheng, weiguo song, jian wang. detrended fluctuation analysis of forest fires and related weather parameters. phys a, 2008, 387 :2091–2099 [17] yue j h, zhao x j, shang p j. effect of trends on detrended fluctuation analysis of precipitation series. math probl eng,2010:1-15. published by atlantis press copyright: the authors 132 http://apps.isiknowledge.com/daisyoneclicksearch.do?product=wos&search_mode=daisyoneclicksearch&db_id=&sid=x1ahed1kbdm3pl8@jc8&name=wang%20j&ut=000272929100010&pos=1 http://apps.isiknowledge.com/daisyoneclicksearch.do?product=wos&search_mode=daisyoneclicksearch&db_id=&sid=x1ahed1kbdm3pl8@jc8&name=song%20wg&ut=000272929100010&pos=2 http://apps.isiknowledge.com/daisyoneclicksearch.do?product=wos&search_mode=daisyoneclicksearch&db_id=&sid=x1ahed1kbdm3pl8@jc8&name=zheng%20hy&ut=000272929100010&pos=3 http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6t4t-4wm0jf4-1&_user=14084&_coverdate=09%2f30%2f2009&_alid=1464160259&_rdoc=18&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=4983&_st=5&_docanchor=&_ct=58&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=680501441404bc4f22b108d024f8444c&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6t4t-4wm0jf4-1&_user=14084&_coverdate=09%2f30%2f2009&_alid=1464160259&_rdoc=18&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=4983&_st=5&_docanchor=&_ct=58&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=680501441404bc4f22b108d024f8444c&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6t0p-50106xj-1&_user=14084&_coverdate=08%2f04%2f2010&_alid=1464169720&_rdoc=21&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=4868&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=4391cbe6c0667d58cba7efc8efc6dc5c&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6t0p-50106xj-1&_user=14084&_coverdate=08%2f04%2f2010&_alid=1464169720&_rdoc=21&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=4868&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=4391cbe6c0667d58cba7efc8efc6dc5c&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6t0p-50106xj-1&_user=14084&_coverdate=08%2f04%2f2010&_alid=1464169720&_rdoc=21&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=4868&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=4391cbe6c0667d58cba7efc8efc6dc5c&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6v2s-4x49mw7-6&_user=14084&_coverdate=11%2f30%2f2009&_alid=1464160259&_rdoc=2&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5710&_st=5&_docanchor=&_ct=58&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=93549d3fbc89bb796cb76f13695348e5&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6v2s-4x49mw7-6&_user=14084&_coverdate=11%2f30%2f2009&_alid=1464160259&_rdoc=2&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5710&_st=5&_docanchor=&_ct=58&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=93549d3fbc89bb796cb76f13695348e5&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6v2s-4x49mw7-6&_user=14084&_coverdate=11%2f30%2f2009&_alid=1464160259&_rdoc=2&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5710&_st=5&_docanchor=&_ct=58&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=93549d3fbc89bb796cb76f13695348e5&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-507bhny-2&_user=14084&_coverdate=10%2f01%2f2010&_alid=1464161951&_rdoc=23&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=0c435fc1f7797293466f830f2d08b8f6&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-507bhny-2&_user=14084&_coverdate=10%2f01%2f2010&_alid=1464161951&_rdoc=23&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=0c435fc1f7797293466f830f2d08b8f6&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4y34khx-1&_user=14084&_coverdate=04%2f15%2f2010&_alid=1464161951&_rdoc=37&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=c6321cd362853d79910a514f92242e05&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4y34khx-1&_user=14084&_coverdate=04%2f15%2f2010&_alid=1464161951&_rdoc=37&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=c6321cd362853d79910a514f92242e05&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4y34khx-1&_user=14084&_coverdate=04%2f15%2f2010&_alid=1464161951&_rdoc=37&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=c6321cd362853d79910a514f92242e05&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4ykgj4y-2&_user=14084&_coverdate=07%2f15%2f2010&_alid=1464161951&_rdoc=5&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=f50271241d30e19c0a59de79eed2ca3b&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4ykgj4y-2&_user=14084&_coverdate=07%2f15%2f2010&_alid=1464161951&_rdoc=5&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=f50271241d30e19c0a59de79eed2ca3b&searchtype=a http://www.sciencedirect.com/science?_ob=articleurl&_udi=b6tvg-4ykgj4y-2&_user=14084&_coverdate=07%2f15%2f2010&_alid=1464161951&_rdoc=5&_fmt=high&_orig=search&_origin=search&_zone=rslt_list_item&_cdi=5534&_st=5&_docanchor=&_ct=68&_acct=c000001598&_version=1&_urlversion=0&_userid=14084&md5=f50271241d30e19c0a59de79eed2ca3b&searchtype=a http://apps.isiknowledge.com/full_record.do?product=wos&search_mode=generalsearch&qid=3&sid=r1njne7m@ifgdknhm7@&page=1&doc=5 http://apps.isiknowledge.com/full_record.do?product=wos&search_mode=generalsearch&qid=3&sid=r1njne7m@ifgdknhm7@&page=1&doc=5 http://apps.isiknowledge.com/full_record.do?product=wos&search_mode=generalsearch&qid=3&sid=r1njne7m@ifgdknhm7@&page=1&doc=5 atlantis press journal style a new risk management model dean kashiwagi* and jacob kashiwagi del. e webb school of construction, arizona state university po box 870204 tempe, az 85287, usa *e-mail: dean.kashiwagi@asu.edu abstract the identification of the source of project risk has been a troubling issue in construction management. a new risk management model has been developed. analysis of the risk in the it industry, construction industry, and also industry structure and supply chain delivery has resulted in the identification of the major source of risk: the major source of risk is the buyer. the buyer's attempt to transfer this risk to the contractor causes confusion and an increase of risk. the owner's lack of expertise and accountability, their use of management, direction and control to minimize risk, their inability to listen to the expert contractors who have no risk and the owner's decision making and expectations has resulted in increased project risk. this study uses deductive logic to design the new risk model. the new risk model is used by the expert contractor to minimize the risk they do not control. the new model identifies risk, mitigates risk by use of transparency and dominant information. the new model has been tested and the results have been documented. the preliminary test results are encouraging. keywords: risk, deductive logic, risk model, dominant information, construction management, it management 1. introduction the delivery of construction has been studied by construction management researchers for the past twenty years. the performance of the delivery of construction services has been identified as poor (cahill and puybaraud, 1994; lepatner, 2007; cfma, 2006; simonson, 2006; flores and chase, 2005; adrian, 2001; hamel, 2007; post, 2000; egan, 1998; chan, 2004; davis et. al., 2009). many studies have been published which identifies the poor performance and suggest the need for risk management. risk management has been a hot topic, and potential solutions have included preproject planning (hillson, 1997; cii, 1995; gibson et. al., 2006; griffith, 2001; wang et. al., 2010; hamilton, 1996), alignment of experts and expertise (kashiwagi, 2009; sullivan, 2010), more defined contracts (davis, et. al., 2009; sweet, 2011), or the hiring of professional project managers (pmi, 2010; ice & ap, 2005). construction delivery systems have evolved from the design-bid-build model, to design-build, to cm@risk and lastly to integrated project delivery (ipd) to minimize the risk of non-performance (matthews and howell, 2005; konchar and sanvido, 1998; nellore, 2001; hopper and goldman, 2004; yinglai, 2009; chan, 2002; williams et. al., 2003). other systems such as indefinite quantity, indefinite delivery (idiq) which includes job order contracting, and cost plus fixed fee, cost plus variable fee, guaranteed maximum price (gmp) and other variations have been created to overcome procurement transactions and control the project time and price journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 233-251 published by atlantis press copyright: the authors 233 administrateur texte tapé à la machine received 21 april 2012 administrateur texte tapé à la machine accepted 11 august 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine d. kashiwagi, j. kashiwagi deviations. other creative processes such as public private partnerships (ppp), design, build and operate (dbo), and privately financed initiatives (pfi) have been used to generate the funding for public projects (grout, 1997; grimsey, 2002; kumaraswamy and morris, 2002; pietroforte, 2002; wong, 2006; boukendour, 2001; papajohn et. al., 2011). however, there is little documentation to show that the overall cost, time, quality performance and value of services have increased (williams et. al., 2003; konchar and sanvido, 1998; hale et. al., 2009; ling et. al., 2004). various concepts such as business process reengineering, continuous improvement, quality control/quality assurance, and lean thinking have been suggested to improve the quality and minimize the cost of construction projects (green, 2011). however, risk and nonperformance remains a stubborn issue in the delivery of construction. the perception of high risk remains. however, when looking for risk data that confirms that there is high risk in delivering construction, evidence remains anecdotal. evidence includes the identification of failed projects, identification of construction quality issues, identification of the lack of performance, identification of the continuing decrease in craftsperson quality and the decreasing productivity of construction workers. 2. problem risk management and non-performance in construction has been a hot research topic (akintoye and macleod, 1997; baloi and price, 2003; bernstein et al., 2011; cooke and davies, 2001; crawford et al., 2006; del cano and de la cruz, 2002; dikmen et al., 2008; hillson, 1998; muller and turner, 2001; raz and michael, 2001; themistocleous and wearne, 2000; tummala et al., 1997; turner and muller, 2003; williams, 1995; wood and ellis, 2003; zou et al., 2007). this research proposes to answer the following questions:  what is risk?  is there risk on every construction project?  what is the major source of risk?  how is risk minimized? 3. definition of risk risk is defined by the authors as when project cost, time, or owner's quality expectation are not met by the contractor (hillson, 2002; hillson, 2009; williams, 1995; wharton, 1992). in a project environment risk is an uncertain event or condition that if it occurs has a negative or positive effect on the project objectives (pmi, 2010). according to hillson (2009) risk is uncertainty that matters; it can affect one or more objectives. risk is defined by others as (hillson and murray-webster, 2006; project management institute, 2010; institution of civil engineers et. al., 2005; williams, 1995; wharton, 1992):  not achieving the expected goal.  the difference between what is expected and what resulted.  unforeseen events that impacted cost, time, and quality.  negative impacts. risk is relative due to the expectations and understanding of the owner, contractor, professional or inspector. some identify the risk level in construction as high. however, when compared with the delivery of it systems, the relative risk level of construction is much lower than in the it industry (little, 2012; jacob kashiwagi, 2012; al-ahmad, et.al., 2009; brown, 2001; connolly, 2006; masing, 2009; natovich, 2003; peppard, 2006; schneider, et. al., 2009; schwaig, et. al., 2006; skulmoski, 2010; vries, 2006; vital smarts, et. al., 2006). the failure rate of the delivery of services in the it industry has been documented at 70% (kerzner, 2011) based on the following surveys:  the standish group chaos reports (19952010).  robbins-gioia, llc (2001).  the oasig study (1995).  chaos report (1995).  the bull survey (1998).  the kpmg canada survey (1997). the bull survey of 1998 identified the following on it projects:  missed deadlines (75%).  exceeded budget (55%). published by atlantis press copyright: the authors 234 a new risk management model  poor communications (40%).  inability to meet project requirements (37%). the survey revealed that the major causes of project failure during the lifecycle of the project is a breakdown in communication (57%), lack of planning (39%) and poor quality control (35%). the standish group’s chaos report was the result of 365 surveys of it practitioners and they identified cost overruns, time overruns, and quality issues as the main reason for nonperformance. the reasons for nonperformance as: table 1. standish group’s result: performance report (chaos report, 1995) performance metrics % 1 incomplete requirements 13.10% 2 lack of user involvement 12.40% 3 lack of resources 10.60% 4 unrealistic expectations 9.90% 5 lack of executive support 9.30% 6 changing requirements & specifications 8.70% 7 lack of planning 8.10% 8 didn't need it any longer 7.50% 9 lack of it management 6.20% 10 technology illiteracy 4.30% 11 other 9.90% “the standish group research shows 31.1% of the projects will be cancelled before they get completed while 52.7 % of the projects will cost two folds (189%) of the original estimates.” another survey was carried by the kpmg (canada) 1997. the study revealed over 61% of the projects that were analyzed was deemed to have failed. the main causes of project failure being:  poor project planning.  weak business case.  lack of top management involvement and support. the key finding of oasig study (1995) was 7 out of 10 it projects ‘fails’ in some respect. robbinsgioia, llc, a provider of management consulting services located in alexandria virginia, made a study over the perception by enterprises of their implementation of an e.r.p. (enterprise resource planning) package (2001.) this included 232 survey respondents spanning multiple industries including government, information technology, communications, financial, utilities, and healthcare. a total of 36 % of the companies surveyed had, or were in the process of, implementing an erp system. key findings included:  51% viewed their erp implementation as unsuccessful.  46% of the participants noted that while their organization had an erp system in place, or was implementing a system, they did not feel their organization understood how to use the system to improve the way they conduct business.  56% of survey respondents noted their organization has a program management office (pmo) in place, and of these respondents, only 36% felt their erp implementation was unsuccessful. the authors propose to use the clearly identified sources of risk in it projects as a starting point in designing a new risk and project management model. this source of information will then be combined with the deductive logic results of the analysis of the industry structure and supply chain issues. the above results from it performance studies list the following as reasons for project failure:  incomplete requirements.  unrealistic expectations.  lack of planning.  changing requirements.  lack of project management. construction risk has previously been identified in terms of cost deviation, time deviation, and not meeting quality expectations. comparing construction risk and published by atlantis press copyright: the authors 235 d. kashiwagi, j. kashiwagi it industry risk, the capability to accurately capture requirements, identify project scope, preplanning and manage the project from beginning to end are lacking in both industries. due to the lack of success in mitigating the risk issue despite years of research and publication, the authors propose that project risk may not be inherent in the project as many have believed. based on the results of the literature search in both the construction and the it industries, the authors propose that risk may be due to the lack of expertise and decision making of the project managers in identifying the requirement, delivering the project from beginning to end, and managing the project to mitigate risk. the authors also propose that the project managers may not understand a workable and accurate definition of risk. the authors also propose that if a project manager had expertise, and a good understanding of the project, by definition of being an expert, the amount of their risk would be negligible. 4. proposal from the literature search, the authors identify that minimizing risk in the construction and it industries have been stubborn issues. the proposal of this paper is that the source of risk may not be project complexity, but a systems or environmental issue that has not previously been sufficiently explored. by observation of potential sources of risk in the it and construction industries, a lack of expertise may be a major source of risk. the authors, therefore, propose that the current systems and environment may be preventing the identification and efficient use of expertise at the proper time. instead of analyzing the risk in the supply chain silos of design, procurement, construction and inspection, the authors are proposing to use a supply chain approach and allow the movement of expertise where it is needed. this practice has been utilized in the design-build and cm@risk delivery systems. the authors also propose to identify new sources of risk based on deductive logic and thinking. 5. methodology the authors propose to 1) identify the source of risk of the existing delivery system using deductive logic and observation instead of inductive research (identifying the proposed solution by using the perceptions and expertise of industry personnel), 2) design a new risk management model and environment and run case study tests to verify the validity of the deductive logic and potential of the new risk management system. the proposed new system will have a measurement system to show the performance of the contractors and to identify the source of project deviations or risk. the objective of the new risk management model will be to measure and minimize contractor risk of project deviation to increase project performance and to identify the sources of project deviation and risk. 6. deductive logic the authors will use a deductive approach. the deductive approach is the use of already accepted logic constructs, the building of a new risk management model using the constructs, and the testing of the risk management model by case study testing. the proposed new construct would be a starting point of future research. the authors will use a series of deductive logic concepts from the information measurement theory (imt) (kashiwagi, 2012). they include:  all event outcomes are predictable if all information is known about the initial conditions.  natural law regulates the change in one set of conditions to another set of conditions over a period of time.  anything that happens over time is an event. an event has initial conditions and final conditions.  the change in the event initial conditions to the final conditions is regulated by natural law. if someone could accurately observe and identify the initial conditions, they could predict how the initial conditions will change to the final conditions using the natural laws.  once the initial conditions are set, the final conditions are also set due to natural laws. in other words, the natural laws dictate how the initial conditions will change to the final conditions.  there is no possibility to impact, influence, or control the event to result in an outcome that is not related to the initial conditions once the initial conditions are set. all event outcomes are singular, predictable and regulated by natural laws. published by atlantis press copyright: the authors 236 a new risk management model the above progression of deductive logic to identify that no one party has influence or control over another party, has been analyzed by jacob kashiwagi (2007). kashiwagi showed that there was no documentation that showed conclusive evidence or logic that one party can have influence, impact or control over another party (kashiwagi, 2010; kashiwagi, 2008; kashiwagi, 2007). major events in society that support kashiwagi's concept include the failure of prohibition, the high cost and negative impacts of the ongoing drug war, the lack of effectiveness of prisons to rehabilitate criminals, the lack of success of welfare programs to change individuals, the failure of one country's attempt to change another country’s government structure through war (vietnam war and iraq war.) in all cases, the attempt to control an entity or group of entities to a desired outcome proved costly with the expectations being unfulfilled (white, 2010; szalavitz, 2009; langan and levin, 2002; gregory, 2006; greer, 2009; caulkins et. al., 1997). the authors, therefore, propose that a system that uses management, direction and control as a means to mitigate risk may be ineffective. therefore, if owners/buyers of construction are attempting to compensate for the lack of expertise in contractor's project managers by using direction and control, the result may be increased risk. the methodology to show the potential validity of this concept will be deductive in nature, using logic which can be readily observed, and case study testing. when one party attempts to manage, direct, and control another party, the second party usually becomes more reactive and, therefore, has a higher risk. if the testing shows potential merit of the concept, more research may be required. 7. development of the concept of no “control” according to the proposed deductive logic based on observation, the centerpiece of the proposed new system is the concept that the buyer:  has no effective control over a contractor or vendor.  cannot utilize a contract to efficiently and effectively control the contractor to meet their expectations.  if the owner is a decision maker, and attempts to manages, direct, and control the contractor, the expertise to deliver construction is a requirement of the buyer. this assumes that the owner is a better expert than the expert vendor, thus increasing the risk of the project.  if owners are attempting to manage, direct and control, the system becomes riskier due to the lack of precedence that control can be effectively used to mitigate risk. in other words, if an owner attempts to control a contractor the project is perceived as a riskier project. if the project was not initially perceived as risky, there would be no need to control the contractor. an alternative method to control risk will be utilized in a new risk management system. the new system will use the opposite of the "control" concept. the new concept will use concepts that are not found in traditional models. the new concepts include (kashiwagi, 2011; kashiwagi, 2009; child, 2010; meyer, et. al., 2010):  the owner's control will be replaced by the alignment of expertise. the owner will select and employ an expert contractor. an expert contractor should have minimal risk in delivering construction.  the expert contractor will prove that they can do the project based on past experience, performance metrics and the ability to see into the future project before it is done.  the expert contractor will compete with other experts to do the work.  the expert contractor will plan out the project from beginning to end, before the project is started.  the contractor will control the project. they will identify the delivered scope. the contractor will identify what is out of the scope, and that will be identified as risk to the project.  an expert contractor has no risk within their defined scope.  the contractor shall have a risk mitigation plan to mitigate the risk that they do not control. published by atlantis press copyright: the authors 237 d. kashiwagi, j. kashiwagi  the contractor will be the offeror of the contract, and the owner/buyer will be the acceptor of the offer.  if the contractor does not clearly define their scope, schedule, risk and risk mitigation, and track their risk and resulting cost and time deviations, the contractor’s proposal is not acceptable as an expert and their contract is. they are not an expert, and the owner should not accept the proposal.  the contractor will identify all deviations to the project, the source of the deviations, and mitigate the risk at the lowest cost during the project. the use of control by the owner over a contractor through a contract will not be used. the contract will be written by the contractor to identify how the expert will deliver the project. the contractor will be the offeror of the contract, and the owner or buyer will be the acceptor of the contract. the owner will assume that the competing contractors are experts. if they are not an expert, the owner’s delivery system will not allow the hiring of the non-expert contractor. this new approach also identifies risk as "what is outside the scope of the contract." by definition, an expert should have no risk in doing their expertise, within an offered scope. their only risk is outside of their control, scope or where the owner’s intent and project information is not clear. an expert contractor, who writes their scope, should not volunteer to do anything that they cannot do. therefore, by deductive logic, the expert contractor should not have risk in the scope of a project which they determine. this identifies the new structure as having the following characteristics:  the owner uses a selection process which identifies expert contractors who can deliver construction.  the owner identifies the best value contractor who can deliver an acceptable product at the lowest price to the owner.  the owner will identify the best value expert contractor as one who makes them the most comfortable that they are delivering a product which the owner can accept, who can identify the project risk and how they will mitigate the risk of a project which is outside of their control, and can do it for the lowest price within their budget. the new structure will be tested by case study. the success of the tests would be defined by the following:  the new system will measure project deviation.  the clients will be satisfied with the delivered construction.  the risk created by the expert contractor is negligible.  if project deviations occur, the deviations will most likely be caused by the owner (outside the control of the expert contractor.)  if the owner attempts to exert control, the risk of the project will increase.  if the contractor creates project deviations and customer dissatisfaction, the deductive concepts of no control are probably not accurate. the results of the case studies will identify if there is potentially a new risk system. if the proposed system based on the new concepts produces positive results, the system and the concepts can be further analyzed. these concepts force the research into a new area that research has not heavily investigated. the solution is a radical departure from conventional construction management research. concepts that are not well published include:  expert contractors have no risk.  risk does not come from projects, but from owners hiring contractors who are not experts in defining the scope and not mitigating risk that they do not control.  owners using decision making, management, direction, and control to minimize risk will increase the risk of the project. first, if they need to utilize control, they have hired a nonexpert who has high risk. then they are exacerbating the risk by trying to manage, direct and control.  decision making increases risk. when the initial conditions are understood, they predict the final conditions. when decision making is used, the initial conditions are not understood, and the final outcome is in doubt. published by atlantis press copyright: the authors 238 a new risk management model  contractors should be responsible to mitigate risk they do not control, but are not financially responsible for the risk.  risk is caused by the owner's delivery system, unforeseen events due to a lack of information and non-experts and thus can be attributed to the owner.  contracts cannot mitigate risk.  contracts are used to control the contractor. contracts are a source of risk due to the perception that the owner can successfully use it to control and direct the contractor. deductively, if a buyer makes decisions, directs, and controls a contractor, the buyer is increasing the risk of project failure. if the contractor is an expert, and observes and can accurately identify the initial conditions of the event (buyer expectation, environmental conditions, budget, time, quality), and can identify the final conditions and how to get from the initial conditions to the final conditions, they minimize the risk of the buyer/client. if the owner is the expert, and is directing the contractor, the risk of project failure increases because by default, the contractor is a nonexpert and requires direction. if a contractor is an expert, by definition they can identify the following:  initial project conditions (owner expectation, resources, and environmental constraints).  final conditions.  how to get from the initial conditions to the final conditions.  the scope of the project (which brings no risk to the contractor) and what is outside of the scope of the project, which is the financial responsibility of the owner/buyer.  minimize the risk of those who they do not control or risk caused by a lack of information. project risk is being defined as items which are not in the scope of the project. project risk cannot be transferred to the contractor because it is not in the scope of the project. the expert contractor can mitigate risk, but they cannot be financially responsible for risk that they do not control. therefore, project risk cannot be transferred from the owner to the contractor. therefore, contractors only have risk when they cannot see their project from beginning to end cannot clearly define their scope and cannot proactively mitigate risk that they do not control. risk is caused by the inability to see into the future. risk is, therefore, related to the level of expertise and experience of the contractor. if they are not expert, they increase the project risk. risk is, therefore, minimized by employing the most expert contractor, having them identify the scope of the project, and having the owner accept the scope of work as defined by the contractor. the proposal is that risk is not caused by the complexity level of a project, but risk is defined by the lack of experience and expertise of personnel who are participants of a project. personnel with more experience (performed like projects in the past) and expertise (ability to observe and accurately perceive the initial conditions and natural laws) will have less risk. risk, therefore, is more personnel related and not project related. 8. dominance the term dominance is defined as (kashiwagi, 2012):  simple.  obvious.  it is a no brainer.  predicts the future outcome.  results in minimal risk.  brings consensus among people, and minimizes decision making. when dominance is maximized, decision making is minimized. when the need for decision making is maximized, the conditions are complex, and participants must use their own experience and expertise to determine future actions. by using deductive reasoning, risk increases when decision making is maximized. experts who have expertise and experience, minimize decision making, can accurately identify initial conditions and can accurately predict the changing of conditions into the future. the clearer the expert's vision of the future project, the simpler they can communicate and plan for the project. experts plan the project from beginning to end. experts also minimize the risk that they do not control or identify areas where they lack information. in these instances, the contractor published by atlantis press copyright: the authors 239 d. kashiwagi, j. kashiwagi has risk, and must make a decision (using their experience and expertise), to set a tentative plan. therefore, experts explain things in clear, dominant and simple terms that are easily understood by everyone. experts can bring consensus in a group with their dominant perceptions, explanations and proposals. they minimize the decision making of all the other participants by making things simple, by connecting the dots between the present and the future. when they make things dominant, it relieves all other participants of decision making to predict the future outcome. this deductive reasoning results in the following proposals:  experts minimize the need for decisions if they can see into the future.  decision making is done when someone cannot see into the future and they must choose between multiple options.  the degree of decision making is relative to the vision or blindness of the person. the visionary who can see into the future, does not depend as much on decision making. if someone cannot see, they depend heavily on decision making.  decision making increases risk.  risk is when someone cannot see into the future and cannot predict the outcome.  risk and decision making are related. people who are more dependent on decision making, have higher risk.  the new environment will increase dominance, simplicity and transparency, and thereby, minimize decision making and risk. the new environment will minimize decision making by forcing everyone to look into the future and preplan. when people pre-plan they minimize decision making and risk and see the future more clearly. 9. industry structure figure 1 is the industry structure model. it identifies the price based and the best value environments. the major difference between the two quadrants is that in the price based quadrant, the buyer or client determines the requirement and the solution, and minimizes risk by managing, directing, and controlling the contractor. in this environment, the client is responsible to mitigate the risk. fig. 1. industry structure (kashiwagi, 2012) in the best value environment, the client identifies the best value contractor, but the contractor identifies what they will deliver and how they will deliver it. the contractor does the risk management and the quality control, and the buyer does the quality assurance (ensures that the contractor is doing what they proposed they would do including risk management and quality control.) the contractor also defines the scope of the work and "what is in" the scope, and "what is out." the contractor must ensure that what is not in scope is clearly defined, and does not cause risk to the scope of the project. an expert contractor will not have risk in their defined scope. the expert contractor, therefore, accomplishes the following:  must compete to identify their company as the best value.  identifies what is in and what is out of the project.  has a detailed project schedule.  identifies the activities in the schedule that they do not control or which they do not have sufficient information. these are risk activities. these are not in their scope, but they will work to mitigate these risks.  identify how they will mitigate the risk in a risk mitigation plan.  identify how they will track the risk, measure the risk and report the risk to the owner/buyer. these actions reinforce the concepts that: published by atlantis press copyright: the authors 240 a new risk management model  an expert contractor has no financial risk. they are not financially responsible for accomplishing a task that the client has not given sufficient or accurate information.  a best value contractor is the offeror and the buyer is the acceptor of the offer.  the expert contractor is in the best position to identify what will be done.  a contract does not have to be all inclusive, but merely identifies what the contractor will provide in the project.  an expert contractor minimizes the project risk.  the contract is the contractor's offer and how they will mitigate risk. 10. system solution the deductive logic proposes that the price based environment, or the traditional method of project delivery is a major source of project risk. as defined above, the price based environment has the following characteristics:  the wrong party is doing the talking (the party with less expertise.)  buyer/owner controls the contractor and the project.  the owner uses their own expertise, experience and decision making to identify the construction project scope, cost, time duration, and quality.  the owner and their representatives use management, direction and control to mitigate project risk.  the number of participants to deliver the project increases.  the risk increases as communications, meetings and directions increase.  as management, direction and control increases, contractor's personnel become more reactive.  the contract becomes a more vital part of the project, as the client attempts to control the contractor. the authors propose that in the current environments where construction is being delivered, the buyers control the environment, and are managing, directing, and controlling the contractors to minimize risk. the authors propose that if the control of the projects was transferred to the expert contractors, and if the sources of deviation were identified and measured, the following results would occur:  the clients and buyers would be identified as the major source of risk and deviations.  the buyers of construction would be very satisfied with the performance of the contractors.  the contractors would become more proactive and would not be the source of risk. 11. creation of the best value environment and test results the authors have been developing and testing the new risk management model, the best value performance information procurement system (pips) for the past 18 years at arizona state university. the best value pips was designed with the characteristics of the new “no control” alignment model. pips testing has the following results (pbsrg, 2012):  over 1000 tests.  industry funding of $12m to conduct the tests.  delivering of $4.4b of services.  bv pips is copyrighted and licensed to 26 different research clients (most licensed technology developed at arizona state university (asu) by az tech, the innovation group at asu.)  vendors increase their profit margin by 5%. (state of hawaii report, 2002; university of minnesota report, 2011).  minimize management transactions by as much as 90%.  50% of the time the awarded best value is the lowest cost.  customer satisfaction is at 98%. published by atlantis press copyright: the authors 241 d. kashiwagi, j. kashiwagi  testing has been conducted outside of construction, and results show that the same problems in delivering construction exist in other industries. best value pips is characterized by the following unique characteristics:  identifies the contractor as the expert, the offeror of the proposal, and the client as the acceptor of the offer.  transfers the control of the project from the buyer to the contractor.  minimizes the need to manage, direct and control the contractor.  minimizes decision making of all parties.  the contractor identifies the scope of the project and how it will be accomplished.  the contractor tracks all project deviations and the source of the deviations (risk.)  forces the contractor to have a detailed plan, and to improve quality by minimizing project deviations. over the past 18 years, multiple clients (motorola, honeywell, mcdonnell douglas, international rectifier, ibm, neogard, state of hawaii, united airlines, federal aviation administration, u.s. coast guard, u.s. air force, u.s. corps of engineers, u.s. army medical command, university of minnesota, tremco, state of idaho, state of oklahoma, state of alaska, rijkswaterstaat (dutch infrastructure agency), university of alberta and brunsfield) have participated in tests to change the environment from an "owner controlled" price based environment, to a "contractor controlled" best value environment (pbsrg, 2012; kashiwagi, 2012). two efforts, the university of minnesota and related tests in the state of minnesota, and a large government organization (lgo) used the best value pips environment to mitigate risk. both clients ran repeated project tests over a number of years and were able to document the contractor project performance, the project deviations, and the source of the deviations. the results of tests run in the state of minnesota by the university of minnesota and others are shown below in table 2. note the following results (pbsrg, 2012):  the client and their representatives were responsible for 95% of all project cost and time deviations.  client satisfaction of the contractor quality was 100% and the average rating for performance of the contractor was 9.5.  over 50% of the time, the best value was the lowest cost. it is also important to note that the best value pips system was being continually improved during the testing. the project testing was done on design-bidbuild projects, design-build projects, and multiple prime contractor projects. the lgo tests were different from the university of minnesota tests due to the following:  the lgo did not use pips for the selection phase, but only for the risk management phase. the selection was run more in an owner controlled price based manner, even though other factors than price were considered. the buyer was non-transparent, negotiated the contractor's price, and heavily considered price in the awards, which is not the best value pips process.  the lgo used the clarification phase after award, by having the contractor conduct a site survey, and hold a clarification meeting.  the risk documentation was kept on more projects for a longer period of time giving longitudinal information the lgo application was called the performance information risk management system (pirms). even though the selection phase was not used, the results mirror the results of the state of minnesota (table 3):  over time, as the new system was implemented, the project deviation (cost and time) decreased.  the use of the risk management plan (rmp) which identified the scope and the risk that the vendor did not control resulted in lower project deviations.  the major project deviations were caused the buyer/clients.  the contractors caused very little risk. published by atlantis press copyright: the authors 242 a new risk management model the buyer's users were very satisfied with the contractors' performance (table 4.) table 2. university of minnesota results criteria rating number of projects 326 awarded cost $296m percent of projects where bv had lowest cost 53.00% overall change order rate 15.80% due to client 14.30% due to designer 00.60% due to contractor 00.00% due to unforeseen 00.90% overall delay rate 42.90% due to client 31.20% due to designer 04.10% due to contractor 02.20% due to unforeseen 05.40% average contractor performance rating 9.6/10 number of completed project ratings 206 table 3. lgo wrr/rmp performance metrics of completed projects estimated project overview without wrr with wrr with wrr and rmp # of projects 130 265 72 average project awarded cost ($m) $1.92 $1.31 $1.33 % of projects on time (%) 28 32 40 % of projects on budget (%) 40 48 65 % over awarded budget (%) 7.02 6.87 3.73 % due to owner (%) 4.77 4.77 3.48 % due to contractor (%) 0.02 0.06 0.00 % due to unforeseen (%) 2.23 2.03 0.26 % delayed 39.1 37.29 29.11 % due to owner (%) 25.0 26.07 29.14 % due to contractor (%) 3.58 2.21 -2.61 % due to unforeseen (%) 10.52 9.01 2.58 table 4. owner’s satisfaction with contractor contractor a b c d e number of projects 68 121 51 51 24 average project awarded cost ($m) .78 1.08 2.31 1.65 1.29 % over awarded budget (%) 8.25 6.29 5.83 2.46 0.82 % due to owner (%) 7.29 4.92 3.27 1.30 -0.12 % due to contractor (%) 0.32 -0.09 0.00 0.00 0.12 % due to unforeseen (%) 0.63 1.45 2.56 1.16 0.82 % delayed (%) 42.00 45.88 22.34 33.38 21.86 % due to owner (%) 32.29 34.15 19.49 28.05 11.98 % due to contractor (%) 3.85 1.62 -2.08 0.82 0.94 % due to unforeseen (%) 5.86 10.11 4.92 4.50 8.94 owner satisfaction (1-10) 9.6 9.1 9.3 9.5 9.5 the lgo started testing pirms in 2006. the initial objectives were (kashiwagi, 2009):  make the environment transparent.  transfer the control of the projects to the vendors.  educate the lgo personnel on how to use the system to minimize their efforts.  educate the contractors on how to use the system to maximize their profit and minimize project cost and transactions (meetings, communications, and problems). by 2010, four out of the six areas of the lgo, were using the best value approach. half of the contractors were using the best value approach in their own organizations. table 5 and figure 2 show the improvement in the understanding of the change in paradigm. however, in 2010, a change occurred at the lgo. the visionary who understood the concepts of the best value approach retired, and the personnel who replaced him did not have the same understanding or capability of being a visionary. starting in the beginning of 2010 the lgo management decided to do the following:  control the information, no longer making all information available to all contractors. it became a "need to know" approach.  stop education to the lgo personnel on how to change their paradigm and use the best value published by atlantis press copyright: the authors 243 d. kashiwagi, j. kashiwagi approach in all their work functions. they did not understand that unless the lgo policy was consistent with the best value approach of "nocontrol" and the lgo personnel could implement a paradigm across all their job functions, there would not have sufficient motivation to change and thereby be comfortable with the best value risk management approach.  stop educating the contractors in the best value  manage, direct and control the entire system at the headquarter's level and make it a "top down" approach. the lgo attempted to automate the best value risk management system. they attempted to turn the risk management system into an automated performance information system that they could control. they were attempting to keep the performance and project deviation information, but not use the paradigm of "no control" of the best value model. comparing the difference between projects with the best value risk management approach and the traditional process, the best value risk management approach increases in value from 2009-2010, then decreases in 2011 when the organization attempted to control the system and the information. the authors propose that the change is not implementing an information system, but changing their paradigm to move to a "no-control" paradigm. table 5. best value conference attendees best value conference attendees year 06'-07' year 08'9' entity # of representatives idiq contractors 23 79 clients 9 37 certification program year 06'-07' year 08'-09' certified best value contractors 0/7 2/7 awareness & best value education/training year 06'-07' year 08'-09' contractors requesting education/training 2/7 5/7 fig. 2. comparison of survey results from 2006 and 2009 table 6 shows the impact of the decision to change back into the control mode. the overall deviation rate has increased and there is no difference between using a risk management plan and not using it. it shows signs of contractors becoming more reactive. table 6 also reinforces the reactive behavior of the contractors. the contractors can take control of the project by starting a weekly risk report (wrr) as soon as they are awarded a project. the wrr goes to all project participants once a week. the wrr does the following: table 6. contractor performance rmp analysis by year completed project overview without rmp with rmp without rmp with rmp without rmp with rmp year 3 3 4 4 5 5 # of projects 82 36 29 102 30 108 average project actual project cost ($m) 1.5 1.0 2.6 1.5 2.0 2.1 % over budget 6.50 2.48 4.18 4.19 4.76 5.74 % due to owner (%) 5.32 1.79 2.57 3.72 4.71 4.79 % due to contractor (%) 0.07 0.00 0.00 0.18 0.00 -0.03 % due to unforeseen (%) 1.11 0.69 1.61 0.29 0.05 0.97 % delayed 43.88 5.17 44.39 34.84 46.38 53.43 % due to owner (%) 29.53 10.11 27.86 33.59 42.31 38.16 % due to contractor (%) 5.14 -6.90 3.17 -1.01 -0.11 5.35 % due to unforeseen (%) 9.21 1.96 13.36 2.26 4.19 9.93 published by atlantis press copyright: the authors 244 a new risk management model  establishes the contractor's project milestone schedule.  identifies any deviations to the contractor's schedule and the source of the deviations.  it motivates all the rest of the participants in the project to be accountable to do their task at the right time. if they do not, they are identified as the source of risk with an accompanying cost and time to the project.  informs all project participants on the status of their project, and who may be a source of risk by not performing their task. by creating the wrr at the time of award, the contractor can ensure that the notice to proceed (ntp) is accomplished by the contracting office in a timely manner due to the accurate and repeated documentation of the contractor. if the contracting office was not processing them in a timely manner, and the lengthy time could impact the contractor’s schedules and costs. even though the wrr was not required until the ntp, by creating the wrr at the time of award, the contractors kept the users and all interested parties notified if the contracting office is not proceeding with a timely ntp. the wrr goes to all interested parties, and they can easily see why their project is not on time if the ntp is not processed in a timely manner. by tracking the time of creation of the wrr, the attitude of the contractors and users can be identified. table 7 shows that only a small percentage of wrr were done before the ntp in 2009/2010. the majority of the wrr are created 1.5 months after the ntp (when the wrr is legally required to be created by the contractor). this shows that the client and vendors were not diligent in doing their responsibilities. the potential impact of the lgo to not continue the education to change the paradigm and training of the clients and vendors resulted in the contractors not acting in their own best interest. table 7. delivery of the wrr wrr creation data 2009/2010 # of projects 226 # of projects start with weekly 64 % of projects start with weekly 28% average length of time a project was without a weekly (weeks) 6.43 the performance of the contractors in the system by notice to proceed date is shown in table 8. the following conclusions can be drawn from the data:  the majority of the project deviations are caused by the client.  as the client gets accustomed to the new paradigm, the deviation rates continue to fall. 12. other results of best value approaching test the best value approach was tested by the dutch agency rijkswaterstaat who is responsible for water and ground infrastructure. they tested the best value pips environment on a $1b test in 2008-2012. the process minimized the government procurement and transaction costs by 50%. the participating contractors also identified savings in competing and securing tenders (50-75%). the project results have been that deviations can be identified back to the owner, most projects finished a year ahead of schedule, and the system accurately identified that the vendors were not the source of risk in the delivery of construction. the rijkswaterstaat is now using the best value risk management approach to reorganize their internal structure to mitigate risk, transactions, and cost (brandsen, 2011; witteveen, 2011). the rijkswaterstaat also recently won the 2012 dutch sourcing award (dsa) for public procurement and overall for public and private procurement achievement for their success with the best value pips delivery system. the nevi, the dutch professional industry procurement group has been licensed by arizona state university to educate and certify all dutch practitioners. this achievement is unparalleled in university research published by atlantis press copyright: the authors 245 d. kashiwagi, j. kashiwagi developed systems and concepts in the delivery of construction services. table 8. contractor’s performance by year of notice to proceed (ntp) ntp year 1 2 3 4 # of projects 108 127 109 72 average project awarded cost ($m) 1.6 1.4 1.3 1.0 total % over budget (%) 6.30 5.92 3.63 3.25 % due to owner (%) 4.73 5.54 2.90 2.17 % due to contractor (%) 0.00 -0.17 0.00 0.31 % due to unforeseen (%) 1.46 0.41 0.78 0.72 total % delayed (%) 48.44 45.56 28.33 20.82 % due to owner (%) 40.29 39.68 22.12 11.59 % due to contractor (%) 1.74 -0.09 1.11 -0.54 % due to unforeseen (%) 6.40 5.47 5.10 7.93 the brunsfield company from malaysia is one of the largest and most successful development/contracting groups in asia. they are implementing the best value risk mitigating environment in their entire supply chain. what makes the effort significant is that brunsfield is one of the most successful entities in delivering construction performance, often minimizing delivery time and cost by 33%, while also minimizing defects (kashiwagi, 2011). the brunsfield company currently has a research grant to implement the best value approach not only in their own supply chain, but with all contractors, subcontractors, and vendors who do business with them. the state of oklahoma has been testing the best value risk management approach for the last three years on both construction and services projects. their results are shown in table 9. the following are significant characteristics:  did not have internal expertise, but depended on the external vendors' expertise.  minimized protest behavior. encountered three protests and easily won all three.  high customer satisfaction.  lower costs than anticipated.  better quality and performance than anticipated.  issues were caused by the owner's representatives. table 9. state of oklahoma results oklahoma best value project information # of best-value procurements 20 estimated value of best-value procurements $100m protest success rate (# of protest won / # of protests) 3/3 # of different services 13 % where identified best-value was lowest cost 71% project performance # of completed projects 8 average customer satisfaction 9.5 (out of 10) cost savings $29m % on-time 100% % on-budget 100% the state of oklahoma did not confine their testing to construction projects. it included the following projects (pbsrg, 2012; kashiwagi, 2012):  commercial off the shelf (cots) tax software  enhancement of workforce job website  electronic document management for construction documents.  computer to plate printer  state wide light bulb and lighting fixture contract  emergency hazardous waste removal contract  construction commissioning services  state mental health services  performance measurement of federal grants  new construction and renovation projects  combination juvenile center and services (cancelled due to political issues) published by atlantis press copyright: the authors 246 a new risk management model the state of oklahoma used the positive results to change their construction procurement law to allow the new risk management environment of the best value pips structure. the state is also encouraging all procurement agents in the state to get educated and trained in the new process. they are also implementing the certification program that the dutch are using. 13. conclusion and future recommendation the researchers propose that project risk is a function of the following:  an owner is making decisions and creating expectations.  an owner's delivery environment where the wrong party is doing the decision making, management, direction and control.  the owner is attempting to pass the risk of a project to a contractor.  the owner is under the illusion that they can minimize project risk by directing and controlling a contractor.  owners/buyers are attempting to direct and control expert vendors to meet inaccurate expectations.  the contractors in response become reactive and will not be financially responsible for risk. the owner’s delivery system creates risk by attempting to use decision making and control to minimize risk. the owner does not realize that risk is caused by a lack of expertise. by attempting to manage, direct and control, the owner is encouraging the contractors to be more reactive, thus increasing the owner's risk and delivering lower quality. the owners are essentially minimizing the usage of the contractor's expertise. the researchers propose a new delivery model that utilized:  no control by the owner/buyer.  alignment of expertise.  hiring an expert contractor by ensuring the expert can accurately identify the initial conditions, identify how to go from initial conditions to final conditions, and can identify and mitigate risk that they do not control.  transferring project control to the expert.  determination of scope by the contractor.  identification of risk as only factors that are outside of the control of the contractor.  minimization of decision making, deferring to the expert contractor to identify solutions. a model was built using the above characteristics. the model is the best value performance information procurement system (pips) or the performance information risk management system (pirms). pips/pirms was used to minimize risk, minimize project cost, time and quality deviations. the testing of the new risk management and project management model resulted in:  identification that the owner/buyer and their representatives as the biggest source of risk.  minimization of project risk.  a change of paradigm, from a management based approach to a leadership based approach.  both owners and contractors required education. the new risk management model has the following characteristics:  minimized communication and decision making in the selection process and during the project.  the best value vendor identifies the scope and writes the contract.  the vendor is the offeror and the client/buyer is the acceptor of the offer.  the best value vendor is identified through competition of performance and price.  the best value vendor must clearly identify their project capability, their price, their scope, project risk and risk mitigation.  the contract is written by the contractor, and includes all owner legal requirements, the scope of the work, the schedule, quality control and risk mitigation. it also includes the weekly risk report. published by atlantis press copyright: the authors 247 d. kashiwagi, j. kashiwagi the testing of the new risk model by the lgo and the university of minnesota had the following results:  the clients have been satisfied.  the contractors get high customer satisfaction ratings.  the major source of project risk and deviation is the client.  the new risk model is a paradigm shift and requires the client/buyers and the contractors to be educated in the new paradigm. the contractors become better with practice.  the new model is a transfer of control to the contractors. the amount of management, direction and control transactions is reduced to as much as 90%. the testing of the new risk model in the netherlands by the rijkswaterstaat on a $1b infrastructure fast track project had the following results:  projects tendered twice as fast.  costs for both contractors and rijkswaterstaat were reduced by 50%.  the rijkswaterstaat won the prestigious 2012 dpa for both the private sector and the public sector for their testing of the new risk model.  projects finished a year ahead of initial schedule completion dates.  the nevi (ism and nigp counterpart in the netherlands) licensed to educate and certify professionals delivering services in the netherlands in 2012. the state of oklahoma ran 20 projects ($100m) on construction and non-construction projects. the results included 100% customer satisfaction, no vendor caused deviations, cost savings of $29m and the mitigation of all risk caused by protests. three of the projects were protested; however the protests were successfully overcome by the transparent measurements of the selection system. these preliminary results are sufficient to warrant future research and testing of the “no control” risk management model. testing is ongoing in canada, malaysia, netherlands, and u.s. future publications will cover the details of the new risk model and the test results. references 1. adrian, j. j. (2001) improving construction productivity, construction productivity newsletter, 12. 2. akintoye, a. and macleod, m. (1997). risk analysis and management in construction. international journal of project management. 15(1), 31-38. 3. al-ahmad, w., al-fagih, k., khanfar, k., alsamara, k., and abuleil, s. (2009). a taxonomy of an it project failure. root causes, 5(1), 93-106. 4. baloi, d. & price, a. d. f. (2003). modeling global risk factors affecting construction cost performance, international journal of project management. 21, 261269. 5. bernstein, h.m., russo, m. a., and laquidara-carr, d. (2011). mitigation of risk in construction: strategies for reducing risk and maximizing profitability. mcgrawhill construction. 6. boukendour, s. (2001). the guaranteed maximum price contract as call option, construction management and economics, 19 (6), pp 563. 7. brandsen, c. (2011). innovation in public procurement: priority road investment programme. cees brandsen, managing director, rijkswaterstaat, competence centre infrastructure, ministry of infrastructure and environment, the netherlands, london, uk. july, 24, 2011. 8. brown, t. (2001). modernisation or failure?, it development projects in the uk public sector, 17(4), 363-381. 9. cahill, d. and puybaraud, m. (1994). constructing the team: the latham report. construction reports 1944-98. blackwell science ltd, pgs. 145-160. 10. caulkins, j. p., rydell , p. c., schwabe, w. l. and chiesa, j. (1997). mandatory minimum drug sentences: throwing away the key or the taxpayers’money, rand, 1-192. 11. cfma’s. (2006). construction industry annual financial survey, moss-adams, llp, eighteenth edition. 12. chan, a. (2002). framework of success criteria for design/build projects, journal of management in engineering, vol. 18, iss. 3, pp. 120. 13. chan, a.p.c., and chan, a.p.l. (2004). key performance indicators for measuring construction success, benchmarking an international journal, emerald group publishing limited, 11(2), 203-221. 14. child, g., kashiwagi, d., and sullivan, k. (2010). us department of defense procurement systems approaching best value, cib world congress 2010, the lowry, salford quays, united kingdom, 1580, pp. 1-14 (may 10-13, 2010). 15. connolly, m. (2006). strategies to avoid failure of it projects, 11(3), 6-7. (also solutions for everything) 16. construction industry institute (cii). (1994). pre-project planning: beginning a project the right way, construction industry institute, austin, tx, publication 39-1. published by atlantis press copyright: the authors 248 a new risk management model 17. construction industry institute (cii). (1995). pre-project planning handbook, construction industry institute, austin, tx, special publication 39-2. 18. cooke davies, t. (2001). the real success factors in projects. in: a project management odyssey, proceedings of the pmi europe 2001, the 4th pmi european project. 19. crawford, l., pollack, d., and england, d. (2006). uncovering the trends in project management: journal emphases over the last 10 years. international journal of project management. 24(2), 175-184. 20. davis, b., and sebastian, r. (2009a). the relationship between contract administration problems and contract type, journal of public procurement, 9(2), 262-282. 21. davis, b., and sebastian, r. (2009b). an analysis of the consequences of contract administration problems for contract types, journal of management research, 1(2). 22. del cano, a. and de la cruz, m.p. (2002). integrated methodology for project risk management. journal of construction engineering and management. 128(6), 473485. 23. dikmen, i., birgonul, m.t., anac, c., and aouad, g. (2008). learning from risks: a tool for post-project risk assessment. automation in construction, 18(1), 42-50. 24. egan, s.j. (1998). rethinking construction: the report of the construction task force to the deputy prime minister, john prescott, on the scope for improving the quality and efficiency of uk construction, the department of trade and industry,1 victoria street, london, sw1h 0et. 25. chaos report. (1995). standish groups chaos report, it cortex, accessed from http://www.itcortex.com/stat_failure_rate.htm#. 26. flores, v. and chase, g. (2005). project controls from the front end, cost engineering, april 2005, vol. 47, no. 4; pgs 22-24. 27. gibson, g. e., wang, y., cho, c. and pappas, m. (2006). what is preproject planning, anyway?, journal of management in engineering, 22 (1). 28. green, stuart d. (2011). making sense of construction improvement, john wiley & sons ltd, the atrium, southern gate, chichester, west sussex, po19 8sq, uk. 29. greer, m. (2009). money spent on the war on drugs this year, drug war clock, retrieved january 8, 2010 from http://www.drugsense.org/wodclock.htm. 30. gregory, a. (2006). the real reason behind overcrowding in prisons, contra costa times, retrieved january 8, 2010 from http://www.independent.org/newsroom/article.asp?id=17 75. 31. griffith, a. and gibson, g., jr. (2001). alignment during preproject planning, journal of management in engineering, 17(2), 69–76. 32. grout, p. (1997). the economics of the private finance initiative, oxford review of economic policy, vol. 13, iss. 4, pp. 53. 33. grimsey, d. (2002). evaluating the risks of public private partnerships for infrastructure projects, international journal of project management, vol. 20, iss. 2, pp. 107. 34. hale, d., shrestha, p., gibson, g., and migliaccio, g. (2009). empirical comparison of design/build and design/bid/build project delivery methods, journal of construction engineering and management asce, july 2009 pp. 579-587. 35. hamel, g. (2007). the future of management, harvard business school publishing, harvard, massachusetts. 36. hamilton, m.r. & gibson, g.e. (1996). benchmarking preproject planning effort, journal of management in engineering, 12(2), 25-33. 37. hillson, d. a. (1997).towards a maturity model, international journal of project & business risk management, 1(1), 35–46. 38. hillson, d.a. (2002). extending the risk process to manage opportunities, international journal of project management. 20(3), 235-240. 39. hillson, d.a. and murray-webster, r. (2006). managing risk attitude using emotional literacy. proceedings of the pmi congress 2006 emea, presented in madrid, spain, 9 may. 40. institution of civil engineers & actuarial professionals (ice & ap). (2005). risk analysis and management for projects (ramp) 2nded., thomas telford, london. 41. university of minnesota report, email chain october 11, 2011, jake, s., kashiwagi, d., sylvia r. and perkins, m., university of minnesota, summary of pips program implementation and performance 2005-2011,university of minnesota, personal communications (attached ppt final umn summary package). 42. hopper, n. and goldman, c. (2004). the federal market for esco services: how does it measure up?, energy analysis department, ernest orlando lawrence berkeley national laboratory, university of california, berkeley, lbnl-54952. 43. kashiwagi, d. (2011). project management model of the future: aligning expertise instead of managing and controlling, 5th annual ut dallas project management symposium, august 11-12, 2011, the university of texas at dallas richardson, texas, p. 1-7. 44. kashiwagi, d. and kashiwagi, j. (2011). solving the malaysian construction industry research environment, sixth international conference on construction in the 21st century, construction challenges in the new decade, july 5-7, 2011, kuala lumpur, malaysia, cd 35. 45. kashiwagi, d. (2012). the best value standard, performance based studies research group, tempe, az. 46. kashiwagi, d. (2012). information measurement theory, performance based studies research group, tempe, az. 47. kashiwagi, d. and kashiwagi, j. (2011). state of the art research model: brunsfield research center, sixth international conference on construction in the 21st published by atlantis press copyright: the authors 249 d. kashiwagi, j. kashiwagi century, construction challenges in the new decade, july 5-7, 2011, kuala lumpur, malaysia, cd 34. 48. kashiwagi, d., kashiwagi, j., and savicky, j. (2009). industry structure: misunderstood by industry and researchers, journal of research, ned university of engineering and technology, 6 (2) pp.59-76 49. kashiwagi, d., sullivan, k., kashiwagi, j., and davisjohnson, m. (2010). supply chain analysis impacts delivery of it services: potential impact to construction, rics cobra 2010, university of dauphine, paris, france, paper 3 (september 2-3, 2010). 50. kashiwagi, j. (2007). leadership is alignment not influence, arizona state university. tempe, az: united states of america. master thesis. 51. kashiwagi, j., sullivan, k. and kashiwagi, d. (2010). the decision-less environment of the future pm, pm-05 advancing project management for the 21st century "concepts tools & techniques for managing successful projects, heraklion, crete, greece (may 29-31, 2010). 52. kashiwagi, j., sullivan, k., badger, w. and kashiwagi, d. (2007). changing the paradigm of leadership, cobra 2007 construction and building research conference, georgia institute of technology, atlanta, ga, usa, cdt36 (september 6, 2007). 53. kashiwagi, j., sullivan, k., kashiwagi, d.t. and badger, w. (2008). is project management alignment or influence/control, 4th scientific conference on project management (scpm) & 1st international project management association (ipma) / mediterranean network (mednet) conference on pm advances, training & certification in the mediterranean, chios island, greece, pp. 186-191 (may 29, 2008). 54. kashiwagi, j. and kashiwagi, d. (2011). the seamless bureaucracy, sixth international conference on construction in the 21st century, construction challenges in the new decade, july 5-7, 2011, kuala lumpur, malaysia, cd 36. 55. kashiwagi, j., malhotra, n., luna, e., kashiwagi, d., and sullivan, k. (2009). creating organizational change: minimizing client generated construction inefficiencies at the us army medical command, crc conference, seattle, wa, pp. 370-379, (april 5-7, 2009). 56. kashiwagi, j., sullivan, k. and kashiwagi, d. (2009). risk management system implemented at the us army medical command, journal of facilities management, vol. 7 no.3, 2009 pp. 224-245. 57. kerzner, h. (2011). project management metrics, kpis, and dashboards: a guide to measuring and monitoring project performance, john wiley and sons, inc., hoboken, new jersey. 58. konchar, m. and sanvido, v. (1998). comparison of u.s. project delivery systems. journal of construction engineering and management, november/december, 435-444. 59. kumaraswamy, m. and morris, d. (2002). build-operatetransfer-type procurement in asian megaprojects, j. construction engineering manage, 128, 93. 60. langan, p. & levin, d. (2002). two thirds of former state prisoners rearrested for serious new crimes, time magazine, retrieved from http://www.time.com/time/magazine/article/0,9171,9145 66,00.html. 61. lepatner, b.b. (2007). broken buildings, busted budgets, the university of chicago press, united states of america: chicago. 62. ling, f., chan, s., chong, e., and ee, p. (2004). predicting performance of design-build and design-bidbuild projects, journal of construction engineering & management, jan/feb 2004, vol. 130. 63. little, m. and kashiwagi, d. (2012). state of idaho procurement of it services, 5th international public procurement conference (ippc5), proposed paper, seattle, wa, august 18, 2012. 64. masing, e. (2009). technical support: improving performance and reducing costs with it risk, risk management, rims, 56(8), 48-51. 65. matthews, o. and howell, g. (2005). integrated project delivery an example of relational contracting, lean construction journal, vol. 2, #1 april 2005, issn: 15551369. 66. meyer, j., witt, s., kashiwagi, j., and kashiwagi, d. (2010). general services administration streamlines the procurement of construction services, proceedings seventh annual acquisition research symposium volume ii, naval postgraduate school, monterey, california, pp. 609-625, (may 12-13, 2010). 67. muller r. and turner, j.r. (2001). the impact of performance in project management knowledge areas on earned value results in information technology projects. project management: international project management journal, project management association finland, norwegian project management forum, 7(1), 44–51. 68. natovich, j. (2003). vendor related risks in it development. a chronology of an outsourced project failure, 15(4), 409-419. 69. nellore, r. (2001). factors influencing success in integrated product development (ipd) projects, ieee transactions on engineering management, vol. 48, iss. 2, pp. 164. 70. oakes, m. (1986). statistical inference: a commentary for the societal and behavioral sciences. john wiley, uk. 71. papajohn, d., qingbin, c. and mehmet, b. (2011). public private partnerships in the us transportation: research overview and a path forward, asce, journal of managmeent in engineering, 27 (3), 126. 72. pbsrg. (2012). performance based studies research group internal research documentation, arizona state university, unpublished raw data. 73. peppard, j. (2006). the conundrum of it management, acis 2006 keynote address, 16, 336–345. 74. pietroforte, r. (2002). procurement methods for us infrastructure: historical perspectives and recent trends, building research and information: the international published by atlantis press copyright: the authors 250 a new risk management model journal of research development and demonstration, vol. 30, iss. 6, pp. 425. 75. post, n.m. (1998). building teams get high marks, engineering news record, 240[19], 32-39. 76. post, n. (2000). no stamp of approval on building plans: contractors sound off over difficulties with bid documents, engineering news record, 244(17), 34-46. 77. project management institute (pmi). (2010). a guide to the project management body of knowledge (pmbok guide). 4th 609 ed. project management institute, newtown square, pa. 78. raz, t. and michael, e. (2001). use and benefits of tools for project risk management. international journal of project management. 19, 9-17. 79. schneider, g, lane, s, and bruton, c. (2009). monitoring risk in information, 13(1), 63-67. 80. schwaig, k., and gillam, s., and leeds, e, (2006). project management issues in it offshore outsourcing, international journal of e-collaboration, vol. 2, issue 4, pg. 53 (21). 81. simonson, k. (2006). quick facts, association of general contractors, chief economist report. 82. skulmoski, g, and hartman, f. (2010). information systems project manager soft competencies. a projectphase investigation, 41(1), 61-80. 83. state of hawaii pips advisory committee (2002). report for senate concurrent resolution no. 39 requesting a review of the performance information procurement system (pips). honolulu, hi: u.s. government, accessed from http://hawaii.gov/dags/rpts/pips.pdf. 84. sullivan, k., kashiwagi, d., chong, n. (2010). the influence of an information environment on a construction organization’s culture: a case study, advances in civil engineering, 2009, article 387608, (10). 85. sullivan, w., and ngwenyama., o. (2005). how are public sector organizations managing its outsourcing risks? an analysis of outsourcing guidelines from three jurisdictions, the journal of computer information systems, spring 2005, 45(3), 73. 86. sweet, j. (2011). standard construction contracts: academic orphan. the construction lawyer, 31(1), 3842. 87. szalavitz, m. (2009). drugs in portugal: did decriminalization work?, time magazine, retrieved january 8, 2010 from http://www.time.com/time/health/article/0,8599,1893946, 00.html. 88. the bull survey. (1998).state failure cause, it cortex, accessed from http://www.itcortex.com/stat_failure_cause.htm. 89. themistocleous, g. and wearne, s.h. (2000). project management topics coverage in journals, international journal of project management. 18, 7-11. 90. tummala, v.m.r., leung, h.m., mok, c.k., burchett, j.f., and leung, y.h. (1997). practices, barriers and benefits of using risk management approaches in selected hong kong industries. international journal of project management. 15(5), 297-312. 91. vries, h. (2006). best practice in company standardization, international journal of it standards and standardization research, 4(1), 62-85. 92. vital smarts and the concours group. (2006). silence fails: the five crucial conversations for flawless execution, vitalsmarts, accessed from http://www.pmi.org/knowledgecenter/~/media/pdf/surveys/pp_maxfield.ashx. 93. yinglai, y. (2009). a comparative analysis of agentconstruction system and cm@r, construction economy, cnkichina national knowledge infrastructure journal, doi: cnki:sun:jzjj.02009-02-006. 94. wang, y. & gibson, g.e. (2010). a study of pre project planning and project success using anns and regression models, automation in construction, 19, 341-346. 95. wharton, f. (1992). risk management: basic concepts and general principles, in: ansell, j. & wharton, f. (eds.). risk: analysis, assessment and management, wiley, new york, 1-14, 28. 96. white, d. (2010). iraq war facts, results & statistics at january 6, 2010, about.com: us liberal politics, retrieved january 6, 2010 from http://usliberals.about.com/od/homelandsecurit1/a/iraqnu mbers.htm. 97. williams, t. (1995). a classified bibliography of recent research related to project risk management, european journal of operational research. 85, 18-38. 98. williams i., young hoon, k., tzeyu, n. and murat, o. (2003). project delivery systems and project change: quantitative analysis, asce, j. construction engineer manage, 129, 382. 99. witteveen, w. (2011). dutch pips testing results, presentation at the eden postgraduate conference, skema, lille, france, august 21, 2011. 100. wong, m. (2006). current problems with multiple award indefinite delivery/indefinite quantity contracts: a primer, heinonline, ross-blakley law library, army law, 17. 101. wood, g.d. and ellis, r.c.t. (2003). risk management practices of leading uk cost consultants, engineering, construction, and architecture management, 10(4), 254262. 102. zou, p., zhang, g. and wang, j. (2007). understanding the key risks in construction projects in china, international journal of project management, 25, 601–61. published by atlantis press copyright: the authors 251 prespective risk perception biases and the resilience of ethics for complying with covid-19-pandemic-related safety measures bako rajaonah1,*, , enrico zio2,3, 1laboratory of industrial and human automation control, mechanical engineering and computer science (lamih umr cnrs 8201), université polytechnique hauts-de-france, valenciennes f-59313, france 2centre de recherche sur les risques et les crises, mines paristech/psl université paris, sophia antipolis, france 3department of energy, politecnico di milano, milan, italy 1. motivation after 1 year and a half of the covid-19 pandemic, almost four million deaths,1 and considerable socioeconomic damage worldwide, we extend a prior short communication2 to more thoroughly explore the relationships between risk judgment and the decision making that risk judgment informs. our goal is to raise the debate about risk communication, starting from the influences on risk perceptions and risk decisions of cognitive, affective and social factors – with an emphasis on mental models and trust – and ending with the concept of the ethics of resilience. 2. influences of risk judgments and risk decisions we choose the work of breakwell [1] to summarize the determinants of risk judgments and decisions regarding individual protective behavior (i.e., mask wearing, physical distance keeping, contact tracing app usage, vaccination, movement restriction, and lockdown acceptance) (table 1). note that even though these influences are categorized here in terms of cognitive, personality, affective, and social factors for their readability, they interact with each other and provide different levels of explanation from the individual level to the institutional level. in brief, risk judgments, and correlatively, risk decisions, are influenced by (i) heuristics; (ii) affective factors (e.g., feelings of confidence); (iii) mental models of risk including knowledge (e.g., virus transmissibility) and beliefs (e.g., self-efficacy), social and cultural norms (e.g., attitude toward mask wearing) and moral values (e.g., altruism); and (iv) trust as an engagement in action (e.g., intention to get vaccinated). we decided to focus on mental models of risk and trust because the goal of risk communication is to (i) fill the gaps in the mental model with regard to risk understanding, reinforce correct beliefs, and correct misconceptions [2] and (ii) correct noncalibrated trust. however, let us begin with their common thread, the resilience of ethics. 3. resilient ethics and risk communication resilience is defined by the united nations office for disaster risk reduction as the ability of a system, community or society exposed to hazards to resist, absorb, accommodate, adapt to, transform and recover from the effects of a hazard in a timely and efficient manner […]. resilience relies not only on groups and communities but also on their members as individual responsible moral entities who are distinct from each other but influence each other within a given group or community. the challenge of covid-19-related risk communication is not only to change individuals’ attitudes until the pandemic is mastered but also to make people aware that there a r t i c l e i n f o article history received 30 november 2020 accepted 06 july 2021 keywords risk perception resilience ethics mental models of risk trust covid-19 a b s t r a c t this perspective paper presents factors that bias covid-19-related risk judgments and risk decisions, such as cognitive biases, affect heuristics, mental models of risk and trust. the goal is to gain knowledge about the difficulty of risk communication in inducing attitudinal and behavioral changes regarding protective measures. talking about morality and ethics appears to be less obsolete and more necessary than ever; it could even be seen as a ‘spare tire’ after one and a half years of risk communication and almost 4 million deaths. perhaps it is time to think in terms of resilience at all levels, from the citizen of humanity to the highest institutions. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: bako.rajaonah@uphf.fr 1who: https://covid19.who.int/. 2hal archives ouvertes: https://hal-uphf.archives-ouvertes.fr/hal-02533290v2. journal of risk analysis and crisis response vol. 11(2); june (2021), pp. 87–90 doi: https://doi.org/10.2991/jracr.k.210707.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0001-8070-9308 http://orcid.org/0000-0002-7108-637x http://creativecommons.org/licenses/by-nc/4.0/ mailto: bako.rajaonah@uphf.fr https://covid19.who.int/ https://hal-uphf.archives-ouvertes.fr/hal-02533290v2 https://doi.org/10.2991/jracr.k.210707.001 https://www.atlantis-press.com/journals/jracr 88 b. rajaonah and e. zio / journal of risk analysis and crisis response 11(2) 87–90 table 1 | cognitive, personality, affective, and social factors that influence risk judgment and decision-making from breakwell [1]. the factors are in italics cognitive factors availability heuristic: tendency to assume that the frequency or the probability of an event is greater if people can easily remember an instance of the event (p. 87). it can interact with confirmation bias, i.e., to favor paying attention to information that confirms existing hypotheses or beliefs (pp. 108–109). anchoring bias: tendency to rely too heavily on the first piece of information offered (the anchor) about a hazard (p. 90). confirmation bias: tendency to favor paying attention to information that confirms existing hypotheses or beliefs (p. 108). representativeness heuristic: tendency to assume that the probability that an object a belongs to a category b (categorical judgment) is greater if a resembles b in some fashion (p. 86). optimistic bias: tendency to believe to be less likely to experience negative events, and more likely to experience positive events, than other people (p. 91). mental model of risk: a system of beliefs, feelings and attitudes that the individual holds about the risk (p. 16), which can influence the way people use information and bring into play their heuristic biases (p. 103). personality factors perceived self-efficacy and locus of control: representation of how much control individuals believe they have over what they do or what happens to them (p. 63). affective factors affect heuristic: refers to the influence of the affect associated with the object of risk judgment and decision-making (p. 12 and 131). feeling of risk: refers to the primacy of emotion in the response to the hazard (p. 123). social factors group norms: perceived norms of the group to which the individual belongs (p. 113). cultural biases: attitudes and beliefs shared by a group (p. 80). social trust: comprising public trust (i.e., the amalgamated feelings of trust toward all societal institutions and leaders, the diffuse communal acceptance of the values of the system, p. 158), institutional trust (i.e., the general feeling of trust about a particular organization or social category, p. 109), and specific trust (i.e., the acceptance of the morals and feeling of confidence in a particular institution as it addresses a specific issue at a single time, p. 109). are critical contexts in which their perceived right way to behave is not necessarily the best with regard to these contexts. resilient ethics refers here to some flexibility and adaptability3 of the moral values to fit with very critical situations that require attitudinal and behavioral change. 3.1. a bit of morality communication must therefore be educative to make people see the hazard in a different way and persuasive to induce informed behavioral changes. civic education programs could help to increase civic engagement, which would be beneficial in the long term [3]. message design should be bidirectional and collaborative between government officials, experts and the public at large to acknowledge motives and needs and to avoid discordance between messages [4]. messages should contain some types of moral lessons that address both individual and shared social and moral values to truly make messages an opportunity for attitude and behavior change: advocating altruism and prosocial behavior; emphasizing the effectiveness of protective measures at both personal and societal levels; emphasizing the sense of sacrifice for the greater benefit of society; developing a shared sense of identity; and encouraging the sense of cooperation [3,5,6]. repeating messages over and over is a good strategy to avoid familiarization with risk (especially relevant regarding the new variants!), which induces less perceived health risk and thus increases nonprotective behaviors, but it should also serve to avoid discouragement of people who make sacrifices; when they observe that other people “don’t bother,” they could feel unfairness and become less cooperative [7]. communication should make the public aware of the impacts of the behavior of a handful of individuals across the whole society, for example, the extreme exhaustion of healthcare workers, the despair of people who lost their jobs, and increasing poverty: in short, all aspects of the socioeconomic fallout. looking at these societal impacts, defending the value of individual freedom, for example, through organization of or participation in public demonstrations against covid-19 restrictions, could be perceived as insulting given the millions of deaths. however, values such as altruism are not sufficient. on the one hand, covid-19 is considered to be challenging to humanity [8]; on the other hand, the philosopher hans jonas considered that humans were responsible for their actions regarding the future of nature and humanity [9]. his imperative of responsibility is a moral obligation that should be considered in education: we are responsible “for and to a distant future,” and behaviors toward compliance with protective measures are part of respect for humanity. given the threat due to the new variants, the morality issues in covid-19 risk communication are less than ever obsolete and more than ever crucial. 3.2. updating mental models of risk the work on moral values can be reinforced by working on mental models, by filling the gap with regard to risk understanding, reinforcing correct beliefs, and correcting misconceptions [1]: knowledge about covid-19 disease, especially the mechanisms of transmission and infectiousness, helps people to be more 3https://www.collinsdictionary.com/dictionary/english/resilient. https://www.collinsdictionary.com/dictionary/english/resilient b. rajaonah and e. zio / journal of risk analysis and crisis response 11(2) 87–90 89 4bmj opinion: https://blogs.bmj.com/bmj/2020/03/11/slowing-down-the-covid-19outbreak-changing-behaviour-by-understanding-it/. 8who: https://www.who.int/director-general/speeches/detail/who-director-general-sopening-remarks-at-148th-session-of-the-executive-board. 9gavi: https://www.gavi.org/covax-facility. 10reuters: https://graphics.reuters.com/world-coronavirus-tracker-and-maps/vaccinationrollout-and-access/. table 2 | recommendations on communications that address vaccine hesitancy and vaccine refusal regarding the audience of messages [14,15] • localizing educational messages to specific audiences, without ‘scientific slang’. • involving the public in the communication campaign. for example, involving students to elaborate messages and communicate them to their peers whom they trust (much more than they trust official institutions) via their usual social media platform; for an example, see: http://fckitwontcutit.com/ (accessed 4 may 2021). regarding the fears related to vaccination and vaccines [15,16] • comparing vaccine risks to other daily risks and providing nonmathematical information using an emotive, empathic approach. • acknowledging that the transient side effects of vaccination mean that the vaccine is preparing the body to fight the disease, and that the risk of the covid-19 disease is by far more important than the risks related to the covid-19 vaccines. • educating people to weigh risks against benefits for vaccines, for example, by means of fact boxes.5 fact boxes present health information in an understandable and transparent manner (see: https://www.hardingcenter.de/sites/default/files/2020-05/methods_paper_harding-center_en.pdf, accessed 5 may 2021). regarding the awareness of the necessity of vaccination • acknowledging that the new coronavirus is also harmful and deadly in younger age groups, i.e., the harmfulness is not limited to older populations and people with comorbidities. this aspect is particularly true with the new variants. • acknowledging that at this time, there is no scientific evidence that supports “natural” population immunity,6 i.e., being protected through naturally acquired infection. • acknowledging that as a result, the best solution, as well as the most ethical solution, to achieving population immunity is vaccination.7 regarding after vaccine uptake [17] • acknowledging the need for continuing protective behavior such as mask wearing and physical distancing after vaccination because of uncertainties regarding vaccine protection. 5scientific american: https://www.scientificamerican.com/article/few-would-fear-covid-vaccines-if-policy-makers-explained-their-risks-better/. 6bmj opinion: https://blogs.bmj.com/ bmj/2020/12/17/natural-herd-immunity-should-not-be-used-as-a-means-of-pandemic-control/. 7who: https://www.who.int/publications/i/item/10665338400. reflective in their judgments and to use less heuristics [10]. one way to enhance individuals’ mental models of risk is to provide them with tools to better understand risk and its management. in the context of the present situation, donnarumma and pezzulo [11] suggest new educational strategies that make available to the public the relevant knowledge on cognitive and social sciences— e.g., regarding cognitive biases—to help them make informed decisions; providing statistical tools is also relevant, for example, to understand the concept of exponential growth, a mechanism that underlies the pandemic and that the lay public understands with difficulty [12]. the transmission process should be clearly explained to the public, for example, in the form of a mental model, such as the model suggested by michie et al.4 3.3. trust: whom? what? finally, working on confidence and trust is also important because both appear to play an important role in the containment of the pandemic, even though there are no clear causal relationships: is noncompliance with protective measures truly due to a lack of trust? a strong sense of adherence to the values of a group (e.g., a political or religious group) could lead to unconditional affective support for or hostility toward, for example, behavioral recommendations. is vaccine reluctance and hesitancy due to a lack of confidence toward the health system or to a lack of trust in a particular vaccine? confidence is rather a feeling, while trust is a cognitive engagement in action [13]: the difference implies distinct communication guidelines. it is also important for appropriate communication to distinguish the object of confidence or trust, especially when different objects coexist for the same question and, therefore, different expectations, such as with covid-19 vaccination, the literature reports various objects of trust: science, the health system, health care professionals, the localization of vaccine manufacturers, the effectiveness and safety of vaccines. future studies could aim to clarify the content of expectations beyond all of these aspects of trust and confidence to design appropriate messages. 3.4. specific recommendations at the time of the writing of this paper, covid-19 vaccination is the center of attention in countries all over the world, because its success is necessary for the restart and recovery of normal social, economic, and political life. therefore, table 2 provides examples of recommendations to shape or update the beliefs that underlie both trust and mental models of risk regarding covid-19 vaccination. 4. conclusion let us conclude in terms of the ethics of resilience. even the most sophisticated and successfully tailored messages will remain useless if vaccines are not available to everyone in all parts of the world. on 18 january 2021, who’s director-general declared that the world was “on the brink of a catastrophic moral failure,”8 referring to the fact that higher-income countries had gone around covax (i.e., a collaborative project that aims at guaranteeing fair access to covid-19 vaccines for every country in the world)9 by dealing directly with the manufacturers. six months after this declaration, reuters reported that approximately 47% of people who had received at least one dose of a vaccine were from high-income countries (with at least approximately 43% from europe and north america).10 https://blogs.bmj.com/bmj/2020/03/11/slowing-down-the-covid-19-outbreak-changing-behaviour-by-understanding-it/ https://blogs.bmj.com/bmj/2020/03/11/slowing-down-the-covid-19-outbreak-changing-behaviour-by-understanding-it/ http://fckitwontcutit.com/ https://www.hardingcenter.de/sites/default/files/2020-05/methods_paper_harding-center_en.pdf https://www.scientificamerican.com/article/few-would-fear-covid-vaccines-if-policy-makers-explained-their-risks-better/ https://blogs.bmj.com/bmj/2020/12/17/natural-herd-immunity-should-not-be-used-as-a-means-of-pandemic-control/ https://blogs.bmj.com/bmj/2020/12/17/natural-herd-immunity-should-not-be-used-as-a-means-of-pandemic-control/ https://www.who.int/publications/i/item/10665338400 90 b. rajaonah and e. zio / journal of risk analysis and crisis response 11(2) 87–90 therefore, even if citizens are more flexible regarding their moral values and accept changing their habits to comply with protective behaviors, societal resilience cannot be achieved without ethics at higher levels than that of the citizen. we call this concept the ethics of resilience, in support of an integrated approach to risk and resilience management of disasters.11 concretely, the ethics of resilience implies respecting some moral imperatives throughout the four phases of disaster management (i.e., prevention/mitigation; protection/preparation; emergency and crisis response activities; recovery for restoring a normal or improved situation). such an ethics could consider the seven principles of ethics for disasters proposed by the philosopher naomi zack [18]: general moral obligations to plan for and respond to disaster; adequacy (i.e., proportionality to benefit and inverse proportionality to risk) and fairness, with the rule of fairly save all who can be saved with the best preparation; individual responsibility; social contract obligations, i.e., governments are obligated to assist individuals in disaster-preparation and -response planning and implementation; safety and security, with a clear distinction regarding the treatment of public safety and public security; dignity, i.e., the human dignity of all disaster victims should be preserved as a primary oral value; and needs, i.e., the needs of all victims must be addressed. the ethics of resilience could also account for jonas’s collective imperative of responsibility for the distant future [9], as well as the englobing view of the interdependences between the various dynamics of a disaster, the various sources and natures of risk, and the spatial worldwide distribution of causes and effects [19,20]. clearly, an enormous amount of work is waiting for the discipline of disaster risk reduction. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution both br and ez authors conceived the idea and substance of the article and contributed to its structure and development. br edited the manuscript and ez supervised the work and its writing. acknowledgment the authors warmly thank the reviewers for their constructive comments. references [1] breakwell gm. the psychology of risk, 2nd edition. cambridge, united kingdom: cambridge university press; 2014. [2] fischhoff b. risk perception and communication unplugged: twenty years of process. risk anal 1995;15:137–45. [3] alessandri g, filosa l, tisak ms, crocetti e, crea g, avanzi l. moral disengagement and generalized social trust as mediators and moderators of rule-respecting behaviors during the covid-19 outbreak. front psychol 2020;11:2102. [4] zhang l, li h, chen k. effective risk communication for public health emergency: reflection on the covid-19 (2019-ncov) outbreak in wuhan, china. healthcare 2020;8:64. [5] dryhurst s, schneider cr, kerr j, freeman alj, recchia g, van der bles am. risk perceptions of covid-19 around the world. j risk res 2020;23:994–1006. [6] van bavel jj, baicker k, boggio ps, capraro v, cichocka a, cikara m, et al. using social and behavioural science to support covid-19 pandemic response. nat hum behav 2020;4: 460–71. [7] lunn pd, belton ca, lavin c, mcgowan fp, timmons s, robertson da. using behavioral science to help fight the coronavirus: a rapid, narrative review. j behav public adm 2020;3:1–15. [8] jones a, strigul n. is spread of covid-19 a chaotic epidemic?. chaos solitons fractals 2021;142:110376. [9] jonas h. technology and responsibility: reflections on the new tasks of ethics. soc res 1973;40:31–54. [10] thoma v, weiss-cohen l, filkuková p, ayton p. cognitive predictors of precautionary behavior during the covid-19 pandemic. front psychol 2021;12:589800. [11] donnarumma f, pezzulo g. moral decisions in the age of covid-19: your choices really matter. soc sci human open 2021;4:100149. [12] banerjee r, bhattacharya j, majumdar p. exponential-growth prediction bias and compliance with safety measures related to covid-19. soc sci med 2021;268:113473. [13] hunyadi m. au début était la confiance [at the beginning was trust]. lormont, france: éditions le bord de l’eau; 2020. [14] contera s. communication is central to the mission of science. nat rev mater 2021;6:377–8. [15] lewandowsky s, cook j, schmid p, holford dl, finn a, leask j, et al. the covid-19 vaccine communication handbook. a practical guide for improving vaccine communication and fighting misinformation (version 6). available from: https://osf.io/f6a48/ (accessed 30 june 2021). [16] poland cm, matthews aks, poland ga. improving covid-19 vaccine acceptance: including insights from human decision making under conditions of uncertainty and human-centered design. vaccine 2021;39:1547–50. [17] su z, wen j, mcdonnell d, goh e, li x, šegalo s, et al. vaccines are not yet a silver bullet: the imperative of continued communication about the importance of covid-19 safety measures. brain behav immun health 2021;12:100204. [18] zack n. ethics for disasters. lanham, md: rowman & littlefield publishers, inc.; 2010. [19] thoradeniya t, jayasinghe s. covid-19 and future pandemics: a global systems approach and relevance to sdgs. global health 2021;17:59. [20] aven t, zio e. globalization and global risk: how risk analysis needs to be enhanced to be effective in confronting current threats. reliab eng syst saf 2021;205:107270. 11prof. enrico zio. “la resilienza etica, di necessità virtù.” il mondo, la casa comune, solidarietà e sostenibilitaà, le opportinità per il nuevo ruolo, dell’imprenditore, forum pmi cesano maderno, 17 giugno 2021. https://doi.org/10.1111/j.1539-6924.1995.tb00308.x https://doi.org/10.1111/j.1539-6924.1995.tb00308.x https://doi.org/10.3389/fpsyg.2020.02102 https://doi.org/10.3389/fpsyg.2020.02102 https://doi.org/10.3389/fpsyg.2020.02102 https://doi.org/10.3389/fpsyg.2020.02102 https://doi.org/10.3390/healthcare8010064 https://doi.org/10.3390/healthcare8010064 https://doi.org/10.3390/healthcare8010064 https://doi.org/10.1080/13669877.2020.1758193 https://doi.org/10.1080/13669877.2020.1758193 https://doi.org/10.1080/13669877.2020.1758193 https://doi.org/10.1038/s41562-020-0884-z https://doi.org/10.1038/s41562-020-0884-z https://doi.org/10.1038/s41562-020-0884-z https://doi.org/10.1038/s41562-020-0884-z https://doi.org/10.30636/jbpa.31.147 https://doi.org/10.30636/jbpa.31.147 https://doi.org/10.30636/jbpa.31.147 https://doi.org/10.30636/jbpa.31.147 https://doi.org/10.1016/j.chaos.2020.110376 https://doi.org/10.1016/j.chaos.2020.110376 https://doi.org/10.3389/fpsyg.2021.589800 https://doi.org/10.3389/fpsyg.2021.589800 https://doi.org/10.3389/fpsyg.2021.589800 https://doi.org/10.1016/j.ssaho.2021.100149 https://doi.org/10.1016/j.ssaho.2021.100149 https://doi.org/10.1016/j.ssaho.2021.100149 https://doi.org/10.1016/j.socscimed.2020.113473 https://doi.org/10.1016/j.socscimed.2020.113473 https://doi.org/10.1016/j.socscimed.2020.113473 https://doi.org/10.1038/s41578-021-00316-w https://doi.org/10.1038/s41578-021-00316-w https://doi.org/10.1016/j.vaccine.2021.02.008 https://doi.org/10.1016/j.vaccine.2021.02.008 https://doi.org/10.1016/j.vaccine.2021.02.008 https://doi.org/10.1016/j.vaccine.2021.02.008 https://doi.org/10.1016/j.bbih.2021.100204 https://doi.org/10.1016/j.bbih.2021.100204 https://doi.org/10.1016/j.bbih.2021.100204 https://doi.org/10.1016/j.bbih.2021.100204 https://doi.org/10.1186/s12992-021-00711-6 https://doi.org/10.1186/s12992-021-00711-6 https://doi.org/10.1186/s12992-021-00711-6 https://doi.org/10.1016/j.ress.2020.107270 https://doi.org/10.1016/j.ress.2020.107270 https://doi.org/10.1016/j.ress.2020.107270 microsoft word the spatial symmetry axis of earthquake hazard in china the spatial symmetry axis of earthquake hazard in china junping yan college of tourism and environment science, shaanxi normal university, xi’an, shaanxi, 710062, china e-mail: yanjp@snnu.edu.cn shuangshuang li, jing bai, xinyan liu college of tourism and environment science, shaanxi normal university, xi’an, shaanxi, 710062, china abstract natural disaster risk assessment is one of the hot fields of the study of natural disaster. it is of theoretical and practical significance to accurately identify the trend of natural disasters. not all the natural disaster has the quality of spatial symmetry, only some disasters during some certain periods (active period) and some certain regions share this quality, which requires substantial research work to distinguish. this paper takes the spatial symmetry axis of earthquake hazard in china as the key of assessing natural disaster risk to emphatically analyze the rationality of the existence of the spatial symmetry axis of the regions of western china, yunnan province and taiwan and find out the dynamics basis of spatial symmetry axis combing with those presentational features, so as to provide some theoretical thinking and methods for reference to identify the time, spatial trends of the earthquake disaster occurrence. the results show that the earthquakes have the quality of spatial symmetry in certain regions and the symmetry has the basis of dynamics, which can work as the basis of the regional earthquake trend. and further improvements are needed to judge the earthquake trend to combine the spatial imagery, dynamics basis and physical mechanism.. keywords: earthquake; the spatial symmetry; dynamics; physical basis 中国地震空间对称轴探讨 延军平 李双双 白晶 刘新颜 陕西师范大学/旅游与环境学院，陕西西安，710062 摘要：在灾害的综合研究中，自然灾害风险评价是灾害研究的一个热点领域，对自然灾害进行准确的趋势 判断，其理论和现实意义重大。不是所有灾害均具有时空对称性，只有部分灾害在部分时期（活跃期）和 部分空间具有一定的对称性。这正需要我们进行大量的研究甄别。本文以中国地震空间对称轴为切入点， 着重探讨了中国西部地震空间对称轴、云南地震空间对称轴、台湾地震空间对称轴等存在的合理性，结合 表象特征寻找空间对称轴的动力学基础，从而为判断地震灾害未来发生的时间、空间趋势提供一些理论思 考和方法借鉴。结果表明，地震在部分区域具有空间对称性特征，其对称性有一定的动力学基础，并且可 以作为区域地震未来趋势判断的依据。如何通过地震对称性的空间表象、动力学基础以及其物理机理有机 结合来判断地震趋势仍需进一步探索。 关键词：地震; 空间对称; 动力学; 物理基础 1. 问题的提出 在灾害的综合研究中，自然灾害风险评价是灾 害研究的一个热点领域，其研究的关键是准确判断 重大自然灾害的发生趋势，即研究未来 2~20 年内发 生重大自然灾害的可能年份和趋势。针对目前灾害 研究单一手段多，综合方法少；害的风险多，灾的 风险少；历史评价多，趋势判断少；学术研究多， 减灾价值少的现象，如果能够提取出灾害的对称性 结构，自此由表及里地逼近灾害发生的物理机理， 逐步提高重大自然灾害的趋势判断水平，这可能是 地理学对灾害研究的重要贡献。 通常所说的对称性多指几何空间配置上的对称 性，如数学中的点对称、轴对称等等，是被观察对 象相对于某一事物经过空间位置转换后和变化前相 似的性质。时间对称性是指某些地理现象的产生和 变化存在一定的时间周期或多重时间周期叠加组 合，在周期的任意对应时刻都存在相同或相似的性 journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 59-64 published by atlantis press copyright: the authors 59 j.yan et al 状或状态。对称性体现的就是一种万变中的不变。 值得一提的是，笔者认为只有部分灾害在部分时期 （活跃期）和部分空间具有一定的对称性，并不是 所有的灾害都具有时空对称性。这就需要通过深入 研究，对不同案例进行甄别，寻找对称性规律 1。 物理原理、物理基础与物理机理是三个既有联 系又存在不同的概念。物理原理是宏观机理，物理 机理是具体的动力学过程，物理基础则是介于二者 之间的中间状态，物理基础是认识物理机理的基 础。认识地震时空对称结构的物理基础，是探索地 震对称性物理机理的前提，空间等间距现象则是形 成对称轴的基础。物理基础是强震灾害趋势判断的 理论前提，物理机理是强震灾害趋势预测的理论基 础（图 1）。我们的研究在此定位为物理基础。 强震灾害风险分析、风险评价、趋势判断的关 系。强震灾害风险分析是指充分利用人类对各种致 灾因子（强震）、承灾体和社会系统的研究成果， 对一定区域、工程项目等可能遭受强震灾害的程度 进行可能性意义下的量化分析，并对采取减灾措施 后的可能效果也进行分析。强震灾害风险评价 (risk assessment of major earthquake disaster) 则是指通过 风险分析的手段或观察外表法，对尚未发生的强震 灾害之致灾因子强度、受灾程度，进行评定和估 计。评定是根据致灾因子强度和承灾体脆弱性推断 出受灾程度。估计是不确定意义下的估计。强震灾 害风险分析是自然灾害学和风险科学的集成体，重 点在科学理论方面。强震灾害风险评价是风险分析 技术在自然灾害学中的应用，重点在具体的模型方 面，属于技术层面 2。 强震趋势判断与强震风险分析一样，均属于理 论方面研究，但是两者的侧重点有所不同。强震趋 势判断侧重于未来强震什么时间发生、地点在哪里 和强度如何；强震风险分析侧重于怎么办，即未来 强震发生后，承灾体的脆弱性如何，承载力如何， 能否应对；如果不能，如何改善、提高等问题。如 果强震趋势判断能取得关键性的突破，其对强震风 险分析将有非常大促进。 马宗晋认为：“地震分区是探索区域地震成因 和区域地球动力学的一种途径”。地震分区的原则 之一就是对每一地震分区给予适当的动力学解释。 笔者曾通过可公度法计算，于 2008 年在《陕西师范 大学学报》（哲学社会科学版）发表“关于地震预测 体系构建的可能性及验证”论文 3，认为 2010 年中国 西部（即川滇地区）将发生 ms≥7.2 地震，结果是 2010 年 4 月 14 日青海玉树发生 7.1 地震，其震级误 差在 0.1 级内，只是位置略偏西北。本文在已有的 地震分区研究基础上，着重探讨了中国西部地震空 间对称轴、云南地震空间对称轴、台湾地震空间对 称轴等存在的合理性，即结合表象特征寻找空间对 称轴的动力学基础，从而为判断地震灾害未来发生 的时间、空间趋势提供一些理论上的思考和方法上 的借鉴。 2. 中国西部地震空间对称轴 2.1. 中国西部地震空间特征 在大陆地质结构中，中国西部及附近地区是一 个特殊的构造区域，是一个相对完整的较大的构造 系。中国西部是印度板块运动的 大受力区，是全 球板内地震 活跃的地区，也是板内地震灾害的 大受害区。在中国西部地区，在地形结构上存在一 定的对称性分布，大致以新疆拜城—青海德令哈— 四川广元三点连线为界，可把中国西部划分为东 北、西南两个次一级的地震空间活动单元，这条线 就是西北—东南地震空间对称轴，简称为西北对称 轴（图 2）。 该对称轴位于地块与活动带的交界处，包括塔 里木块体与天山山地之间、柴达木块体与祁连山之 间、华南块体与秦岭山地之间的缓冲带。此对称轴 从地形结构上来看，位于天山山地—祁连山山地— 秦岭山地前坡。 图 2 中国西部地质构造 图 1 强震灾害分析阶段与体系 published by atlantis press copyright: the authors 60 spatial symmetry axis of earthquake hazard in china 西北对称轴是中国 大一级的空间对称轴，是 重大地震在两边交替活动的分界线，这正恰似于跷 跷板的支点。西北地震带是地震活跃区，而对称轴 则是一个相对稳定的弱作用区及地震平静带，与郭 增建研究的时间“静中动”不同 4，这是一种空间 “动中静”的结构现象，是一个少震的“空心廊 道”，是地震波浪活动的一种表现。此处，空间对 称轴很有可能成为中国西部窥视地震活动空间规律 的重要窗口，即“非应力窗口”。 2.2. 空间对称轴动力学基础 已有研究发现中国西部地质构造具有对称性特 点，新疆拜城—青海德令哈—四川广元连线对称轴 是西部地区地块之间的过渡带 5（图 3）。准噶尔地 块、吐哈地块、阿拉善地块、鄂尔多斯地块等位于 对称轴的东北方向，而塔里木地块、柴达木地块、 四川地块等位于对称轴的西南方向，这些地块均属 刚性地块。ms≥8 地震基本发生在地块边缘地带， 在中心少有发生。在对称轴上，ms≥8 地震发生的 次数几乎为 0。由此说明，中国西部地震空间对称 轴的动力学基础，主要是由地质构造基础决定的。 新疆拜城—青海德令哈—四川广元连线对称 轴，位于亚洲板内地震的南北应力的碰撞带后部 位，是剪切断裂带的缓冲线，其主要力源来自印度 板块导致的青藏块体向北运动（图 4），这里也是 地质构造及地震活动的相对空档；这条线附近也是 现今地壳运动速度比较缓慢或运动方向的转折区 域。据 gps 站点监测 6,7,8,9,10，水平运动速度场在一 定位置出现了转向，即由西南方向转为东北方向， 而这个位置就是我们描述的西北对称轴。换言之， 该对称轴就是水平运动场的方向转折线，是速度波 浪的波谷区。 3. 云南地震空间对称轴 3.1. 云南地震空间对称轴特征 云南省位于印度板块与欧亚板块中国大陆碰撞 带东缘，地震活动频度高、震级大、分布广，属于 板缘、板内地震混合型地区 11。1500 年以来，云南 共发生 20 次 ms≥7 级地震 12，且基本发生于块体边 缘。以 101.5°e、24.2°n 为对称轴，4 个象限区均已 发生 5 次地震，具有明显的对称性。总体看来，以 三次大震的震中位置为顶点，构成三角形，同一三 角形的三个顶点基本出现在经向对称轴 101.5°e 的 异侧(第三组除外)，而在纬向对轴 24.2°n 两侧对称 性变化规律并不明显 13。由此可以看出，东西震荡 性是云南地区 ms≥7 级地震空间活动的主要特点， 且 101.5°e 对称轴两侧地震活动相对均衡。 3.2. 空间对称轴动力学基础 云南地区地质结构复杂，是不同大地构造交汇 地带，其中可分出东南地洼区的滇桂地洼系、南北 地洼区川滇地洼系、巴颜喀拉地洼区、滇西地洼 区、腾冲地洼区。据研究，云南地区平均热流值 高，说明该地区地壳活动频繁，且强度大于全国平 均值 14。从 500m 深地温图和 1000m 深地温图上来 看，以 101.5°e 为轴，发现其东西两侧等温线分布 均具有东疏西密的特点。笔者推测此处存在地质构 造的分界线。 从 中 国 西 南 地 区 现 代 水 平 活 动 图 来 看 （ 图 5），在 101.5°e 附近存在南北向的动力来源，中部 块体自北向南运动，造成了块体间的错动，从而引 发了大震。为了满足 小作用量原理，云南 ms≥7 地震呈东西震荡性分布，两侧的地震活动发生也较 均匀。 图 4 中国大陆及周边地区现今地壳运动速度场 (据文献 8 改绘) 图 3 1500 年以来中国大陆 16 次 ms≥8 地震空间对称性 published by atlantis press copyright: the authors 61 j.yan et al 4. 台湾地震空间对称轴 4.1. 台湾地震空间对称轴特征 利用 1900 年来，台湾及近海 ms≥7.2 地震数据 （数据来源于中国强震目录），探讨台湾地震 ms≥7.2 地震的空间对称性分布。对 17 次 ms≥7.2 地震进行空间标注，共顺次形成 4 个四边形结构，5 个三边形结构，归纳震源点的迁移规律，发现五个 三边形均是一个顶点在南部，其他两点在北部，四 个四边形的第二点均跨轴发生 16 。 4.2. 空间对称轴动力学基础 台湾地震空间对称轴的动力学基础，笔者认为 亦与台湾地区的构造分布有密切的联系。台湾地区 的构造分为北部的琉球弧沟系统和南部的吕宋弧沟 系统。图 7 为对称分布的北部的 b-d 区和南部的 ac 区，从图中可以看到区域内的地震空间分布有较 好的对称性。特别是深源地震都分布在两个相反方 向的俯冲带前缘的冲绳海槽和吕宋海槽，而 d、a 区域分别为琉球海沟、马尼拉海沟后的弧后盆地， 这两个区域内的地震基本都为浅源地震 17。 5. 本研究关注核心问题 1）强震灾害趋势判断、强震预测、强震预报的 区别是什么？ 强震灾害趋势判断、强震预测、强震预报是三 个不同的概念。 图 6 台湾 ms≥7.2 地震三边形（a）、四边形（b）空间分布图（见参考文献 16） 图 7 台湾地震空间对称轴分布图（见参考文献 17） 图 5 中国西南地区现代水平活动图(据文献 15 改绘) published by atlantis press copyright: the authors 62 spatial symmetry axis of earthquake hazard in china 地震预报按照时间尺度可以分为 5 个阶段：第 一、长期预报，指几年至几十年或更长时间内的地 震危险性及其影响的预报；第二、中期预报，指几 个月至几年内将要发生的破坏性地震的预报；第 三、短期预报，指几个天至几月内将要发生的破坏 性地震的预报；第四、临震预报，指几天内将要发 生的破坏性地震的预报或预警；第五、震后震区余 震及趋势预报。 一般而言，预报时间越长，其预报空间范围越 大。上述预报，特别是短期预报和临震预报，只有 国家指定的专门机构发布，是政府行为。无论是专 业还是业余，个人或是集体，对于上述地震预报内 容所持的分析结果和判定意见，只能向有关上级主 管部门上报，不能自行向公众发布或散布。强震预 测是有关人员为这种地震预报提供的参考资料和意 见，《中华人民共和过防震减灾法》对地震预测和 地震预报两者概念做出了明确界定 18。 强震趋势判断，是学者们研究的内容，一般讨 论中长期地震趋势时，即在 1~2 年或 10 年左右的宏 观地震可能性，其地震三要素(时间、地点、震级)一 般不会十分精确，重在给人们以预警信号。基于上 述考虑，我们的研究定位为强震趋势判断，而非强 震预测、更不是强震预报。 2）对称性与部分强震灾害时空结构及机理分 析，是否可以上升到数学阶段，构建数学模型或概 念模型，用数学语言定量描述其逻辑关系，将研究 由定性向定量进一步深化？ 在科学研究中，数学语言是定量化研究的 好 体现，也是其研究成熟度的 有力说明。数学模型 是对研究内容高度抽象的概括，是每个研究孜孜不 倦的追求，我们研究也不例外。从目前研究现状 看，数学模型的建立还是存在很大的困难的，这个 定位应该是清楚的，但是概念模型突破很有可能， 如构建对称性的概念模型，强震灾害机理的概念模 型等，这些都是下一步研究工作的重点和难点，需 要进一步的努力和探索。 3）文中列举了大量的重大自然灾害趋势判断案 例，如果存在错报、漏报，如何看待？ 我们对部分重大自然灾害未来趋势做出判断， 其存在错报、漏报是可能的，或许说是不可避免 的。我们必须坚持 “实事求是”的态度，是什么就 是什么，不必刻意地追求准确，而有意的将漏报、 错报隐去。应坚信一点，失败不可怕，要勇于借鉴 失败的经验。在正确的方向上，失败是一次又一次 成功的开始。在现阶段，对称性与自然灾害的研 究，尚处于初步阶段，这个定位应该是明确的。任 何理论和研究发展都需要一个过程，这个过程有长 有短，但都需坚持不懈，只有坚持才能有突破。 4）以对称性为切入点，未来强震灾害发生能否 进行预测？ 国际公认，有科学和应用意义的地震预测必须 较准确地事先估计强震震中位置、发震时间和震级 三个参数 19]。曾有科学工作者提出以下判断标准： a）对地震预测依据的可观测量有定量描述； b）对未来地震的时间、地点和震级给出定量描 述，包括误差范围； c）有事先预测的详细文字记录； d）过去曾做过详细预测机理，包括成功的和失 败的（虚报、漏报）。 至今人类还不能准确预测强震发生趋势，地震 预测研究成为全球 复杂和 具有挑战向的重大科 学和社会问题之一。 以对称性为切入点，探讨强震灾害发生时空结 构，借助时间对称性方法确定重大自然灾害发生时 间趋势，通过对空间对称性及时间有序性的探索， 进一步逼近灾害发生的可能区域，是本文研究的核 心思路。我们也需要清晰的认识到，本研究的弊端 和局限性。唯象向唯理发展是一种必然趋势，地震 机理不明是制约地震预测的 大瓶颈。如果没有清 晰地物理机理做基础，准确预测还是存在很大难度 的。在现有技术手段和对地震规律认识下，对强震 未来发生趋势准确预测是难以做到的。但是，不排 除未来关键技术的突破，地震资料积累的丰富，准 确预测强震的可能。我们的观点是：应该有信心， 坚信“强震应可预测”，只是现阶段还需要不断的 探索和积累，在唯象基础上寻求唯理机理是一项有 意义的探索。 6. 结论 本研究以对称性为切入点，通过大量事实积 累，探索地理学与对称性的关系，揭示重大自然灾 害时空对称结构内在规律，形成公理化结论，在唯 象基础上寻求唯理机理，为地球科学认识对称性规 律提供一种新的思路。得到初步结论如下： 1）地震在部分地区有空间对称性现象。中国西 北地区地震存在西北—东南向空间对称轴、云南地 区地震存在南北向空间对称轴、台湾地区地震存在 西北—东南向空间对称轴。 2）对称有一定的动力学基础。中国部分地区地 震具有空间对称性，其存在动力学基础，其内在物 理机理还有待进一步探索，空间对称性应是动力学 基础和内在物理机理的外在表现，三者应该是有机 统一的。 3）强震的对称性可以作为地震趋势判断的依 据，为后续研究提供了启示，但仍有待完善。笔者 认为强震的对称性是时间对称性和空间对称性的有 published by atlantis press copyright: the authors 63 j.yan et al 机统一，在某些情况下，其空间对称性会在时间对 称性中得以体现，如何有效的抓住表象对称，进而 深入研究内部更深层次的对称机理，是一个值得探 索的问题。 致谢 本研究得到了国家自然科学基金项目“部分重大自 然灾害的时空对称性：结构、机理与适应对策” (41171090)的资助。. 参考文献 1. li shuangshuang, yan junping, liu lishan, et al. spatiotemporal symmetry and tendency judgment of the ms ≥ 7.8 strong earthquake in indonesia. journal of natural disaster, (01) (2013) : 190-197. 李双双, 延军平, 刘栎杉,等.印度尼西亚 ms≥7.8 强震 时空对称特征及其趋势判断. 自然灾害学报, (01) (2013): 190-197. 2. huang chongfu. natural disaster risk assessment-theory and practice. (science press, beijing, 2005, 3-4). 黄崇福.自然灾害风险评价——理论与实践, 北京:科学 出版社, 2005, 3-4. 3. yan junping, yan na.probability and verification of constructing an earthquake predicting system. journal of shaanxi normal university (philosophy and social sciences edition), 37(5) 2008: 19-23. 延军平, 闫娜. 关于地震预测体系构建的可能性及验证. 陕西师范大学学报(哲学社会科学版), 37(5) (2008): 1923. 4. guo anning, guo zengjian. forecast review of wenchuan earthquake. (xian map publishing press, xian, 2009, 148-159) 郭安宁, 郭增建. 5.12 汶川地震预报回顾, 西安: 西安地 图出版社, 2009, 148-159. 5. liang haihua. occurrence conditions for great earthquakes. seismology and geology, 32(2) (2010): 295-301. 梁海华. 大地震发生的条件. 地震地质. 32(2) (2010): 295-301. 6. wang qi, zhang peizhen, ma zongjin. gps database and velocity field of contemporary tectonic deformation in continental china. earth science frontiers (china university of geosciences, beijing), 9(2) (2002): 416-428. 王琪, 张培震, 马宗晋. 中国大陆现今构造变形 gps 观测数据与速度场, 地学前缘, 9(2) (2002): 416-428. 7. wang qi, zhang peizhen, niu zhijun, et al. crustal movement and tectonic deformation in recent chinese mainland. science in china (series d), 31(7) (2001): 529-536. 王琪, 张培震等, 中国大陆现今地壳运动和构造变形. 中国科学(d 辑), 31(7) (2001): 529-536. 8. qi wang, peizhen zhang et al. present-day crustal deformation in china constrained by global positioning system (gps) measurements. science, (294) (2001): 574577. 9. jian wang, zheng-ren ye, jian-kun he. threedimensional mechanical modeling of large scale crustal deformation in china constrained by the gps velocity field. tectonophysics, (446) (2008): 51–60. 10. ma zongjin, chen xinlian, ye shuhua, et al. gps research of crustal movement in recent chinese mainland. chinese science bulletin, 46(13) (2001): 11181120. 马宗晋, 陈鑫连, 叶叔华, 等. 中国大陆现今地壳运动 的 gps 研究. 科学通报, 46(13) (2001): 1118-1120. 11. huang fugang. a study on seismic activity in yunnan area, phd thesis, anhui: university of science and technology of china, (2009). 皇甫岗. 云南地震活动性研究. 中国科学技术大学博 士学位论文(2009). 12. qin jiazheng, qian xiaodong, liu lifang. study on interval-time characteristics of strong earthquake activities in yunnan region. journal of seismological research, 32(3) (2009): 221-227. 秦嘉政, 钱晓东, 刘丽芳. 云南强震活动间隔时间特征 研究. 地震研究, 32(3) (2009): 221-227. 13. yan junping, bai jing, su kunhui, et al. research on symmetry and tendency of several major natural disasters. geographical research, 30(7) (2011): 11591168. 延军平, 白晶, 苏坤慧等. 对称性与部分重大自然灾害 趋势研究. 地理研究. 30(7) (2011): 1159-1168. 14. xu qing, wang jian, wang jichang, et al. terrestrial heat flow and its tectonic significance in yunnan, china. geotectonica et metallogenia, 16(3) (1992): 285-299. 徐清, 云南大地热流及其大地构造的意义. 大地构造与 成矿学. 16(3) (1992): 285-299. 15. guo anning, chen jiachao. problem on strong earthquake in china. (earthquake press, beijing, 2002). 郭安宁, 陈家超. 中国大震问题. 北京:地震出版社,2002. 16. yan junping, li shuangshuang, liu xinyan, et al. the trend judgment of earthquake in taiwan based on symmetry theory and its physical basis. journal of catastrophology, 28(1) (2013): 11-14. 延军平, 李双双, 刘新颜, 等. 基于的中国台湾地震趋势 判断及物理基础, 灾害学, 28(1) (2013): 11-14. 17. chen daqing, yang maling. the symmetrical partition of taiwan earthquake and its relation to moderatestrong earthquakes in adjacent area of guangdong and fujian provinces. south china journal of seismology, 30(sup.) (2010): 69-75. 陈大庆, 杨马陵. 台湾地震对称分区及与粤闽交界中强 地震的对应关系. 华南地震, 30(增刊) (2010): 69-75. 18. ma zongjin. question of earthquake knowledge. (science press, beijing, 2008). 马宗晋. 地震知识问答. 北京:科学出版社,2008. published by atlantis press copyright: the authors 64 microsoft word access risk management for arabian it company for investing based on prediction of supervised learning journal of risk analysis and crisis response, 2021, 11(3), 91-103 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i3.300 91 article access risk management for arabian it company for investing based on prediction of supervised learning bhupinder singh 1,* and santosh kumar henge 1 1 department of computer science and engineering, lovely professional university, phagwara (144401), punjab, india * correspondence: bhupinder0741@gmail.com received: september 7, 2021; accepted: october 27, 2021; published: october 30, 2021 abstract: the study focuses on chances of profit from saudi it company to increase with few losing trade and a less margin winning investing decisions. fear and greed are two psychological points that dominates the investing decisions. the main objective of the research to study the risk management related to al moammar information systems that is listing on saudi share market. previous research relied on limited methods for prediction of accurate price for investing in the current bullish markets. the research also emphasizes on predicting the right price for investing on the basis of supervised learning methods involving support vector machine, random forest regression, xgboost, auto arima and quasi poisson regression. research has found that the right price to investing in this company comes out to be 106.945 on the prediction of previous 6 months period data. data is sourced though yahoo finance api in form of date, open, high, low, close, volume, dividends and stock splits. this solution can be fruitful for newly trained investors who are willing to invest for long term basis. keywords: support vector machine; random forest regression; xgboost; auto arima; quasi poisson regression; risk management 1. introduction stock market risk management has played a predominant role in surviving in the volatility scenario. saudi stock markets have revolutionized the flow of trend in the international market. supervised learning is focused on predicting the ideal price for the investing in the bullish market. al moammar information systems has revolutionized the saudi stock market by showing 100percentage growth in consecutive years. the problem arises when new investing is struggling to find the appropriate stage to enter into the market. predicting the stock based on sentiment analysis [5] leads to the complexity of the model and thus makes it difficult to take a decision in financial stock market. risk management based on previous knowledge may increase in uncertainty that leads to financial loss in day trading or long-term investments. previous studies indicates that investors relied on the fundamental aspects of the companies but machine based supervised learning [9] is a booster for effective as per the technical data of the company. data calibrate for one year for risk management and further technical prediction has used for latest six month of data to increase the execution time of the system. previous year’s data has formulated as the training data [6]. the yahoo finance symbol of al moammar information systems is 7200.sr and yfinance model is required to install as external bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 92 library in google collab framework. classification of various supervised learning are represented in form of flowchart. fig. 1 supervised learning methods. a. related work d. wei (2019) accessed the role of time series in historical information. he also stated that lstm model has some limitations such as time lag for prediction [2]. m. vijh, d. chandola, v. a. tikkiwal and a. kumar (2020) considered the artifical neural based network and random forest methods for next day closing price of different sectors of companies. they stated that low values of rmse, mape concluded that models are quite efficient for prediction [5]. h. r. patel, a. m. patel and s. m. parikh (2020) concerned about the role of machine learning to find the particular knowledge from uncertain data by analysis of different methods [6]. seungho baek, sunil k. mohanty and mina glambosky (2020) focused on understanding the change related to volatility using markov switching ar model. they also confirmed that covid-19 news affected the stock prices of company [7]. a. f. wagner (2020) elaborates the after effects of covid-19 on the complexity of the company profit or loss ratios in relation to expected cash flows [9]. d. shah, h. isah and f. zulkernine (2019) provided the review for the different taxonomy methods of stock prices movements and discussed various challenges associated with machine learning methods [10]. fuli feng, huimin chen, xiangnan he, ji ding, maosong sun, tat-seng chua (2019) deployed adversial training to enhance the generalization of the prediction model and proposed to add perturbation to implement price variable to work well [15]. e. letizia and f. lillo (2019) implemented predominant correlation among the local topology attributes of firm and their associated risks [17]. h. shah (2019) concluded that levenberg marquardt methods perform better for prediction than the anfis in terms of memory allocation and accuracy [20]. 2. simulation of risk management the investment in stock market requires assuming a certain amount of risk. intelligent investor [19] utilizes efficient methods to minimize the market risk and the boost the gains associated with the market trend. the biggest challenge is to find the appropriate stage to enter into the market in the bullish markets. there exists an enormous risk for an over scribed levels and possess higher chances bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 93 to come down to some extent. table 1 shows the evaluation of average closing price of the stock for n number of days [7] as calibrated in al moammar information systems stock price. the average price has calculated for 1 week, 2 weeks, 1 month, 3 months, 6 months and 1 year eventually. author has focused on following the trend of the direction of stock market and stop loss leads to financial loss. table 1. evaluation of average closing price and profit or loss for the n number of days. period average price profit or loss 1 week 50.62358038766043 6.903356352695258 2 weeks 50.89297458103725 -2.497292811798067 1 month 54.29145151774089 9.923665858258898 3 months 63.86256421407064 6.534656256543506 6 months 84.87345790863037 4.550047454134997 1 year 102.67734532129197 49.826950441055025 furthermore, the table 1 calculate the profit or loss for the n number of days. the profit/ loss has calculated for 1 week, 2 weeks, 1 month, 3 months, 6 months and 1 year respectively. table 2. evaluation of average closing price and profit or loss for the n number of days. daily returns trend 0.5 && 0.5 very slight change or no change 0.5 && 1.0 slight change on the positive side -3 && -1 slight change on the negative side 1 && 3 change on the positive side -3 && -1 change on the negative side 3 && 7 top gains -3 && -7 top losses > 7 bull run (stock prices are on rise) < -7 bear (stock prices are on decline) next, the trends of the stock are perceiving based on the daily returns i.e., daily percentage change in closing price of particular stock. the trends analyzed based on the following factors such as dependable parameters and non-dependable parameters. a. outline of bear and bull market the words bull and bear market used to explain how financial prices usually do whether or not they increase or depreciate. at the same time, given that the economy is driving by the behavior of consumers, these words often reflect how investors feel about the business and the patterns that follow. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 94 table 3. evaluation of trends of the stock is observed based on the daily returns. trend mean median counts among top gainers 1.039402e+06 835992 27 among top losers 7.932692e+05 587088 11 bear 1.463943e+06 1463943 2 bull run 1.747939e+06 1585293 7 negative 2.770925e+05 226514 42 positive 4.743291e+05 376900 42 slight negative 2.522934e+05 171047 27 slight or no change 2.189553e+05 144980 62 slight positive 2.998117e+05 219735 31 fig. 2 exhibit stem plot in form of discrete series. the above graph is a stem plot, which is a discrete series plot. it helps to gain insights from the dataset and highlighting the outliers in the dataset [22]. the daily returns of the closing price of the stock are plotting over the time. if the 14-day rolling average is closer to upper band, it indicates that the stock overbought whereas if it closer to lower band it indicates that the stock is oversold [12]. the above plots are bar graphs. it plots the mean and median respectively of the total traded quantity of the stock grouped by the trend it follows. fig. 3 shows the mean of the total traded quantity of the stock grouped by the followed trend. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 95 fig. 4 shows the median of the total traded quantity of the stock grouped by the followed trend. the fig.5 shows the percentage of each type of trend in the stocks. the trend constituting the highest percentage is the slight or no change in the daily returns of the stock indicating that the stock is non-volatile [15]. fig. 5 evaluation of average closing price and profit or loss for the n number of days. 𝑈𝑝𝑝𝑒𝑟 𝐵𝑎𝑛𝑑 = 𝑀𝐴 + 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ∑ (𝑦 − 𝑀𝐴) 𝑛 (1) 𝐿𝑜𝑤𝑒𝑟 𝐵𝑎𝑛𝑑 = 𝑀𝐴 − 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ∑ (𝑦 − 𝑀𝐴) 𝑛 (2) where 𝑀𝐴 is moving average. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 96 fig. 6 shows plot of bollinger bands. the above plot is of bollinger bands. bollinger bands are a type of statistical chart characterizing the prices and volatility over time period for given interval. equation 1-represented computation of upper band and equation 2 is labeled as lower band while ma is moving average it consists of 3 bands described: 14 day rolling/moving average as green band; upper band (blue band) calculated using the addition of 2 times the standard deviation of the 14-day rolling standard deviation from the 14-day rolling mean. fig. 7 describes the total traded quantity of the stock for the entire timeframe over the stem plot of daily returns. lower band (red band) calculated using the subtraction of 2 times the standard deviation of the 14-day rolling standard deviation from the 14-day rolling mean. the fig.7 plots the total traded quantity of the stock for the entire timeframe over the stem plot of daily returns. it helps to derive a relation between the daily returns of the stocks and the total traded quantity. as it is visible in the plot, the total traded quantity usually increases when the daily returns are high indicating a positive relationship [13] but it not always trues in some parts. table 4 observed the structure of category of parameters with respect to its role in volatility of share price and dependency of its value alter the situation of investor mindset to enter into the market [8]. apart from the stock value variations, there are certain parameters are associated with stock market environment. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 97 table 4. describes the structuring of dependable parameter and non-dependable parameters. dependable parameters non-dependable parameters invested budget (ib) political crises (pc) number of projects in hand (nph) climate and uncertainty (c-uc) number of projects take over in future (nptf) data leakage (dl) manpower (mp) war inside country (wic) infrastructure (in) brand image worldwide (biw) oil prices (op) company ceo instability (c-ceo-i) electronic prices (ep) corporate governance (cg) 3. evolution of supervised learning fig. 8 shows time series forecasting in form of correlogram. fig. 9 describes the modest variation of mean and standard deviation. the practical approach to machine learning is focusing using supervised learning. commonly, input variables )( x and output variables )( y are implementing into mapping function )(xf to predict the future predictions. learning concludes when appropriate method achieves desired level of performance. initially, time series forecasting is done to study the pattern of data into four bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 98 decompositions such as trend, seasonal variations, cyclic variations and random movements. fig.8 is representation of the time series forecasting of share price.       n kni i nknkn k p kk ppp ma 1 21 1... (3) where 1 2, ,..., np p p are the data points that are considered as closing price. 𝑠 = 1 𝑁 − 1 (𝑥 − �̅�) (4) where n-1 represented degree of freedom, ),...,( 1 xxxx n  are the deviations from the mean and fig.9 shows the modest range of comparison between standard deviation with respect to mean. support vector machine can be considered best-supervised learning method for both classification and regression problems. normally, support vector machine [9] builds hyper plane in a highly space in form of dimensions. new data point has to decide which class has to be included called linear classifier. author has implemented linear svm. let assume a training data set of n points in ),(),...,,( 11 nn yxyx where iy are classified as 0 or -1 which belongs to each point ix . thus, hyperplane can be expressed as for given set of points x . 𝑊 𝑥 − 𝑏 = 0 (5) where w is the normal vector with respect to the hyperplane. further move, compute the support vector machine classifier minimizing a normal form. though, precision in equation 7 anticipates about how much model is accurate in form of predicted positive. precision is an excellent way to signify the cost of high value of false positive. recall in equation 8 evaluates how many of actual positives our method simulating through labeling it as bests case true positive [14]. 𝐶𝑙𝑎𝑠𝑠𝑖𝑓𝑖𝑒𝑟 = [ 1 𝑛 𝑚𝑎𝑥 (0,1 − 𝑦 (𝑊 𝑥 − 𝑏))] + 𝛾||𝑤|| (6) 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = 𝑇𝑃 𝑇𝑃 + 𝐹𝑃 (7) 𝑅𝑒𝑐𝑎𝑙𝑙 = 𝑇𝑃 𝑇𝑃 + 𝐹𝑁 (8) 𝐹1 𝑆𝑐𝑜𝑟𝑒 = 2 × 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 × 𝑅𝑒𝑐𝑎𝑙𝑙 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 + 𝑅𝑒𝑐𝑎𝑙𝑙 (9) where tp is true positive, fp is false positive, fn is false negative. the normal f1-score measure as 0.483415 and jaccard similarity score is 0.48. author has realized average f1-score is 0.3933 and jaccard score is 0.4400. f1-score as label in equation 9 is required to study the exact balance between recall and precision [10]. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 99 table 5. evaluation of accuracy of svm in terms of precision, recall andf1-score. precision recall f1-score support 0 0.57 0.53 0.55 15 1 0.36 0.40 0.38 10 accuracy 0.48 25 macro avg 0.47 0.47 0.47 25 weighted avg 0.49 0.48 0.48 25 fig. 10 shows the confusion matrix for support vector simulation with respect to predicted label. random forest regression is most flexible machine learning algorithm that does not require hyper tuning is referred as random forest regression. one of the tiresome problems is over fitting of classifier data that is overcoming by this type of supervised learning. the out of bag r-2 score estimate comes out to be 0.9948780115436955 and test data r-2 score has calibrated as 0.9969415555766163. 𝜌 = 1 − 6 ∑ 𝑑 𝑛(𝑛 − 1) (10) 𝑟 = ∑(𝑥 − 𝑚 )(𝑦 − 𝑚 ) ∑(𝑥 − 𝑚 ) ∑(𝑦 − 𝑚 ) (11) 𝑓 = 1 𝐵 𝑓 (𝑥) (12) after training, predictions [23] for unseen samples x' can be made by averaging the predictions from all the individual regression trees on x': random forests regression score: 0.9969415555766165. test data spearman correlation is 0.9921138864039075 as computed in equation 10. test data pearson correlation is 0.9989167770310116 as evaluated in equation 12. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 100 xboost for time series forecasting [12] is used as automatic parametric selection of features and there exist a availability of extra parameters related to randomization. the version used is 0.90 in this experiment. initialize model with a constant value as represented in equation 13.    n i iylxf 1 )0( ),(minarg)(ˆ   (13) 𝑂𝑢𝑡𝑝𝑢𝑡 𝑓(𝑥) = 𝑓( )(𝑥) = 𝑓 (𝑥) (14) prediction of price of al moammar information systems has observed as 106.94523 that is optimum for investing in current bullish market. quasi poisson regression requires the count variable as represented as dependent variable [22]. it is also known for the generalization of most popular poisson regression consumed while modeling an over dispersed in form of count variable. specific variables based on the poisson regression process. author has considered number of events per unit of space and eventually the number of events dependents on the size of the specific events. table 6 represented the regression results for it company. furthermore, author has modeled y as the specific number of events as poisson distribution, as specified in equation 15. 𝑃(𝑌 = 𝑦) = 𝑒 ⋋ ⋋ 𝑦! (15) for y= 0, 1, ∞, ⋋ is the particular mean with respect to space related to interest. the prominent step in supervised learning is fetching the accuracy of the model. the mean absolute error [14] computes the average of the absolute differentiation between the actual and predicted values in the series of data as formulated in equation 16. mae evaluates the average of the residuals in series of data calibrated from yahoo finance api. table 6. evaluation of quasi poisson regression. gee regression results coef std error z p >|z| [0.025 0.975] intercept -1.0147 0.147 -6.912 0.000 -1.302 -0.727 returns -8.3425 1.567 -5.325 0.000 -11.41 -5.272 buy, sell, open 0.3501 0.335 1.046 0.295 -0.306 1.006 open 0.0006 0.001 0.639 0.523 -0.001 0.003 skew 0.1346 kurtosis -1.8786 centered skew 0.1587 centered kurtosis -1.5741 𝑀𝐴𝐸 = ∑ |𝑦 −𝑥 | 𝑛 (16) 𝑀𝑆𝐸 = 1 𝑛 (y − y ) (17) bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 101 𝑅𝑀𝑆𝐸 = ( 𝑦 − 𝑦 ) 𝑛 (18) mean squared error as reprised as the average of the squared dissimilarity across the original and predicted values in the sourced data. it quantifies the variance of specific residuals in large quantities of irregular data as shown in equation 17. root mean squared error [17] is contemplating as the square root of altogether mean squared error as formulated in equation 18. it administers the standard deviation of standard residuals. fig.11 is concerned with histogram comparison between three errors [19]. previous studies focused on fundamental aspects of the company thus making it difficult to enter into peaked stage of market with higher chance of downward trend in the future. future work extends by involving more machine learning methods with multiple companies’ implementation with long-term period of investment evolution [17]. table 7. comparison analysis of supervised learning methods. supervised learning mae mse rmse xgboost time series 1.3756160 4.040635 2.0101332 support vector machine 0.56 0.56 0.74833147 quasi-poisson regression 0.474412 0.231796 0.481452327 random forests regression 0.844807 1.50751 1.227807 fig. 11 represented of mse, mae and rmse on basis of supervised learning. 4. conclusions the research also emphasizes on predicting the right price for investing based on supervised learning methods involving support vector machine, random forest regression, xgboost, auto arima and quasi poisson regression. research has found that the right price to investing in this company comes out to be 106.945 on the prediction of previous 6 months period data. data is sourced though yahoo finance api in form of date, open, high, low, close, volume, dividends and stock splits. through this analysis, xboost algorithm has outperformed as compare to other supervised bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 102 learning methods due to its less complexity and sequence procedure. this solution can be fruitful for newly trained investors who are willing to invest for long-term basis. funding: there is no source of funding for this research work. acknowledgments: dr. amar singh supports this study and we are very thankful to his valuable reviews to our research work. he examines our work. we really appreciate his remarks related to our work. conflicts of interest: authors share no conflict of interest in the research analysis and prediction of stock market. we confirm that the manuscript has read and approved by all named authors and that are no other persons who satisfy the criteria for authorship but are not listing. references [1] k. zhang, g. zhong, j. dong, s. wang and y. wang, "stock market prediction based on generative adversarial networks", procedia computer science, vol. 147, pp. 400-406, 2019. [2] d. wei, "prediction of stock price based on lstm neural network", 2019 international conference on artificial intelligence and advanced manufacturing (aiam), pp. 544-547, 2019. [3] l. sayavong, z. wu and s. chalita, "research on stock price prediction method based on convolutional neural network", 2019 international conference on virtual reality and intelligent systems (icvris), pp. 173-176, 2019. [4] h. abbasimehr, m. shabani and m. yousefi, "an optimized model using lstm network for demand forecasting", computers & industrial engineering, pp. 106435, 2020. [5] m. vijh, d. chandola, v. a. tikkiwal and a. kumar, "stock closing price prediction using machine learning techniques", procedia computer science, vol. 167, pp. 599-606, 2020. [6] h. r. patel, a. m. patel and s. m. parikh, "challenges and applications for implementing machine learning in computer vision: machine learning applications and approaches" in challenges and applications for implementing machine learning in computer vision, hershey, pa: igi global, pp. 119-135, 2020. [7] s. baek, s. k. mohanty and m. glambosky, "covid-19 and stock market volatility: an industry level analysis", finance research letters, vol. 37, pp. 101748, 2020. [8] s. mohan, s. mullapudi, s. sammeta, p. vijayvergia and d. c. anastasiu, "stock price prediction using news sentiment analysis", 2019 ieee fifth international conference on big data computing service and applications (big data service), 2019. [9] a. f. wagner, "what the stock market tells us about the post-covid-19 world", nature human behaviour, vol. 4, no. 5, pp. 440-440, 2020. [10] d. shah, h. isah and f. zulkernine, "stock market analysis: a review and taxonomy of preddiction techniques", international journal of financial studies, pp. 2-22, 2019. [11] n. n. reddy, v. k. bp and n. e., "prediction stock price using sentimental analysis through twitter data", utc from ieee xplore, 2020. [12] g. s. amidwar, r. kulkarni and m. muthya, "stock prediction through news sentiment analysis", journal of architecture & technology, vol. 9, pp. 36-40, 2019. [13] g. d. and l. qin, "study on the prediction of stock price based on the associated network model of lstm", international journal of machine learning and cybernetics 11, vol. 6, pp. 1307-1317, 2020. [14] t. s. kumar, "data mining based marketing decision support system using hybrid machine learning algorithm", journal of artificial intelligence 2, vol. 3, pp. 185-193, 2020. [15] f. feng, h. chen, x. he, j. ding, m. sun and t.-s. chua, "enhancing stock movement prediction with adversarial training", proceedings of the 28th international joint conference on artificial intelligence, pp. 5843-5849, 2019. [16] c. chen, l. zhao, j. bian, c. xing and t.-y. liu, "investment behaviors can tell what inside: exploring stock intrinsic properties for stock trend prediction", sigkdd, pp. 2376-2384, 2019. [17] e. letizia and f. lillo, "corporate payments networks and credit risk rating", epj data sci., vol. 8, no. 1, pp. 21:1-21:29, 2019. [18] m. g. sousa, k. m. sakiyama, l. de souza rodrigues, p. h. moraes, e. r. fernandes and e. t. matsubara, "bert for stock market sentiment analysis", ictai, pp. 1597-1601, 2019. bhupinder singh and santosh kumar henge / journal of risk analysis and crisis response, 2021, 11(3), 91-103 doi: https://doi.org/10.54560/jracr.v11i3.300 103 [19] c. li, d. song and d. tao, "multi-task recurrent neural networks and higher-order markov random fields for stock price movement prediction: multi-task rnn and higer-order mrfs for stock price classification", sigkdd, pp. 1141-1151, 2019. [20] h. n. shah, "prediction of stock market using artificial intelligence", 2019 ieee 5th int. conf. converg. technol. i2ct 2019, pp. 1-6, 2019. [21] s. o. ojo, p. a. owolawi, m. mphahlele and j. a. adisa, "stock market behaviour prediction using stacked lstm networks-", proc. 2019 int. multidiscip. inf. technol. eng. conf. imitec 2019, pp. 1-5, 2019. [22] m. vijh, d. chandola, v. a. tikkiwal and a. kumar, "stock closing price prediction using machine learning techniques", procedia comput. sci., vol. 167, no. 2019, pp. 599-606, 2020. [23] w. khan, m. a. ghazanfar, m. a. azam, a. karami, k. h. alyoubi and a. s. alfakeeh, "stock market prediction using machine learning classifiers and social media news", j. ambient intell. humaniz. comput., no. 0123456789, 2020. [24] s. e. gao, b. s. lin and c. m. wang, "share price trend prediction using crnn with lstm structure", proc. 2018 int. symp. comput. consum. control. is3c 2018, pp. 10-13, 2019. [25] a. nakayama, k. izumi, h. sakaji, h. matsushima, t. shimada and k. yamada, "short-term stock price prediction by analysis of order pattern images", cifer 2019 ieee conf. comput. intell. financ. eng. econ., pp. 1-5, 2019. [26] v. sharma, r. khemnar, r. kumari and b. r. mohan, "time series with sentiment analysis for stock price prediction", 2019 2nd int. conf. intell. commun. comput. tech. icct 2019, pp. 178-181, 2019. [27] c. peng, z. yin, x. wei and a. zhu, "stock price prediction based on recurrent neural network with long short-term memory units", 2019 int. conf. eng. sci. ind. appl. icesi 2019, pp. 1-5, 2019. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). a cooperative game theoretical approach to risk analysis, using network structure jun-ichi takeshita 1 ∗, hiroaki mohri 2 1 research institute of science for safety and sustainability national institute of advanced industrial science and technology 16-1, onogawa, tsukuba, ibaraki 305-8569, japan e-mail: jun-takeshita@aist.go.jp 2 faculty of commerce , waseda university 1-6-1, nishi-waseda, shinjuku-ku, tokyo 169-8050, japan e-mail: mohri@waseda.jp abstract the principal aim of this paper is to introduce the framework for a cooperative game theoretical approach to risk analysis using network structure. probabilistic risk analysis (pra) is a common methodology for evaluating risks associated with real-world network structure. although there are numerous studies on pra from a physical engineering perspective, hausken (2002) noted that human behavior is a significant factor in estimating risk using pra, and he integrated pra and game theory. while his and related works focused on non-cooperative game theory, in some situations, such as chemical plants, cooperative structures are the norm. therefore, we here provide a risk analysis method based on cooperative game theory, and especially so-called shapley values. keywords: game theory, cooperative game, shapley value, risk management, network structure 1. introduction probabilistic risk analysis (pra) is a common methodology for evaluating risks associated with network structures (bedford and cooke, 2001). although there are numerous studies on pra from a physical engineering perspective, human behavior has increasingly been noted as a significant factor in estimating risks using pra. hausken (2002) initiated this latter approach, and integrated pra and game theory. roughly speaking, game theory is a mathematical theory of decision-making in situations involving two or more relevant players, where decisions depend on competitors’ behavior. such situations are referred to as game situations. originating in von neumann and morgenstern’s theory of games and economic behavior (1944), game theory analyzes conflicts of interest and cooperation among relevant players, by formulating game situations as mathematical models (e.g., fudenberg and tirole, 1991); and therefore, game theoretical approaches typically involve considerations of human behavior. game theory typically focuses on two types of ∗corresponding author journal of risk analysis and crisis response, vol. 4, no. 1 (march 2014), 43-48 published by atlantis press copyright: the authors 43 willieb typewritten text received 17 december 2013 willieb typewritten text accepted 10 march 2014 willieb typewritten text j. takeshita, h. mohri games: cooperative and non-cooperative. in social sciences such as economics, non-cooperative game theory is often employed in modern microeconomics, for example in industrial organization theory (aumann and hart, 1994c, ch. 49), and has recently gained increasing popularity in faculties of economics and management. cooperative game theory, however, has not yet attained such widespread use, and one reason may be that economists typically think of market companies as competitors in a noncooperative game. in this paper, we wish to draw attention to real-world situations in which relevant players are cooperative in terms of risk consideration. we shall take as an example, a complex chemical plant, whose various segments represent relevant players working for a single company. these segments may be numerous and varied, but all must cooperate toward the common production ends of the single plant and company. some economists would argue that even if relevant players are cooperative, the cooperative game would be converted into a noncooperative game by the nash program (nash, 1951, 1953), and this is true in some areas, but not in all. in the chemical plant described above, it is reasonable and unproblematic to posit a sustained cooperative game situation. in light of the foregoing, though most previous studies adopted a non-cooperative game theoretical approach to risk analysis using network structure, a cooperative game theoretical approach is considered more suitable to some real-world situations. the principal aim of this paper, then, is to introduce the framework for a cooperative game theoretical approach to risk analysis with network structure. more precisely, we provide a method to allocate risk among the various segments of a chemical plant with a network structure, using so-called shapley values, solution concepts widely employed in cooperative game theory. the paper is organized as follows. first, we introduce shapley values through concrete examples involving simple-structure graphs. second, we consider two chemical plant toy models: a petroleum refining plant and a medicinal chemical manufacturing plant. finally, we discuss the proposed framework and offer concluding remarks. 2. shapley values here we introduce shapley values by means of two simple examples. throughout this section, we shall denote segments using the letters a, b, and c, and consider each segment to represent a specific factory operation. example 1 (graph with a simple edge) here we consider a graph with a simple edge (figure 1), whose characteristic functions are as follows: a b fig. 1. graph with a simple edge. in this example, we set so-called characteristic functions as follows: c(a) = 8, c(b) = 8, c(ab) = 15. the values c(a) and c(b) indicate the separate costs involved when each segment is working independently and experiences a loss, while c(ab) shows the total cost when a and b are working in concert and an accident occurs or plant production is impaired. the reason that c(ab) < c(a) + c(b) holds is that segments a and b work cooperatively, so that the total risk decreases. the primary aim of this paper is to provide a framework wherein the respective segments are involved in cost sharing. let us calculate the shapley values for each segment. when b’s work is added to that of a, the total cost is 15. hence, b would claim that he only pays 7 (= 15 − 8) if an accident occurs. if we treat their respective contributions as defrayment costs, (a, b) = (8, 7) may be allocated for risk. however, this procedure does not take into account the coalition order of a and b; if only b is working initially, and then a adds to the line, a would claim that he only pays 7 (= 15 − 8) if an accident occurs; and if we treat their respective contributions in this case published by atlantis press copyright: the authors 44 cooperative game to risk analysis as defrayment costs, (a, b) = (7.5, 7.5) may be allocated to risk. the shapley values represent the average of each order’s contribution, given any segment order, for the total segment set. in this example, the shapley values of a and b are 7.5 = (8 + 7)/2 and 7.5 = (7 + 8)/2, respectively. note that we can illustrate the procedure by using table 1: table 1. table for calculating the shapley values of example 1. coalition order a’s contr b’s contr a → b 8 7 = 15 − 8 b → a 7 = 15 − 8 8 shapley value 7.5 7.5 *contr – contribution example 2 (tandem graph) here again we consider a graph with a simple edge (figure 2), whose characteristic functions are as follows: a b c fig. 2. tandem graph. c(a) = 8, c(b) = 8, c(c) = 8, c(ab) = 15, c(bc) = 15, c(ca) = 16, c(abc) = 20. the values c(ab) and c(bc) are both equal to c(ab) in example 1, since their structure is identical. on the other hand, the relations c(ca) = c(c) + c(a), c(ca) > c(ab), c(ca) > c(bc) are satisfied, because segments a and c do not directly connect. this means that cooperation between a and b, and b and c, will reduce the segment cost of each, but that cooperation between c and a will not reduce the segment cost of either. from table 2, the shapley values are (a, b,c) = ( 41 3 , 38 3 , 41 3 ) . note that as b is directly connected to a and c, the shapley value of b is less than that of a and c. table 2. table for calculating the shapley values of example 2. coalition a’s b’s c’s order contr contr contr a → b → c 8 7 5 a → c → b 8 4 8 b → a → c 7 8 5 b → c → a 5 8 7 c → a → b 8 4 8 c → b → a 5 7 8 shapley value 41/3 38/3 41/3 *contr – contribution to take an example, if the coalition order is a → b → c, we can calculate each segment’s contribution as follows: v(a) = c(a) = 8, v(b) = c(ab) − c(a) = 15 − 8 = 7, v(c) = c(abc) − c(ab) = 20 − 15 = 5. 3. chemical plant toy models here we consider two chemical plant toy models, one a petroleum refining plant, and the other a medicinal chemical manufacturing plant. roughly speaking, in the former the primary process is crude oil separation, and in the latter, the repetition of chemical reactions. they are both characterized by a structure typical of chemical plants, which involves a process of either deriving or adding chemicals. expressed mathematically, this is a tree structure; see e.g., lawler (2011, sec. 5, ch. 2) for mathematical elaboration. model 1 (petroleum refining plant model) the process of petroleum refinement involves the separation of crude oil into heavy oil, gas oil, heating oil, and gas. we can model this process as 3, with reservoirs for crude oil (a), heavy oil (d), and the remaining oil (b). published by atlantis press copyright: the authors 45 j. takeshita, h. mohri a b ced fig. 3. petroleum refining plant model. we assume that a and b are reservoirs for oil mixtures, and the others (d, c, and e ) are purified products. we may define the characteristic functions of this model as follows: c(a) = c(b) = 5, c(c) = c(d) = c(e) = 3, c(ab) = c(ac) = c(ae) = c(bd) = 8, c(bc) = c(cd) = c(ce) = c(ad) = c(be) = c(de) = 6, c(ace) = 11, c(abc) = c(abd) = c(abe) = 10, c(acd) = c(ade) = c(bcd) = c(bde) = c(cde) = 9, c(bce) = 8, c(abcd) = c(abce) = c(acde) = 12, c(abde) = c(bcde) = 11, c(abcde) = 13. a and b are involved in the production of two or more products, and c, d, and e are primarily responsible for only one product. hence, the characteristic functions of a and b are greater those of the others. the characteristic functions of more than two segments are governed by the following rule: let and n be a set of segments and a segment, respectively; then (i) c( n) < c( ) + c(n) if there exists a segment m of the set such that m connects to n directly; (ii) c( n) = c( ) + c(n) is satisfied unless case the (i) holds. for example, suppose that = ad and n = b, then a (a member of ) connects to b directly. thus, this is an example of case (i); and as c(ad) = 6, c(b) = 5, and c(abd) = 10, then we have 10 = c(abd) < c(ad) + c(b) = 10. a further example: suppose that = ade and n = c, then there is no segment in such that the segment connects to c directly. thus, this is an example of the case (ii); and as c(ade) = 9, c(c) = 3, and c(acde) = 12, then we have 12 = c(acde) = c(ade) + c(c). the shapley value of this model is (a, b,c, d, e) = ( 420 5 , 342 5 , 290 5 , 248 5 , 260 5 ) . note that we must calculate each segment’s contribution for each coalition order when we generate the shapley values, and there are 5! = 120 orders. remark 1 the results represent each segment’s marginal contribution to the plant’s production (detailed in the final section). therefore, when applying this method to chemical plants, we allocate responsibility to each segment according to its shapley value. specifically, when an accident occurs in a given plant, each segment covers the damage cost according to this value. also, in a risk management context, the plant manager can allocate costs for safety measures to each segment based on its respective shapley value. model 2 (medicinal chemical manufacturing plant model) the process of manufacturing medicine involves the repeated addition of chemical substances. we can model this process as figure 4. published by atlantis press copyright: the authors 46 cooperative game to risk analysis a c e d b f g fig. 4. medicinal chemical manufacturing plant model. the model may at first seem somewhat complex; however, the process works in tandem, and by labeling the intersection points from left to right as α , β , and γ , we can represent this model as a tandem graph (figure 5). a c e d b f g ˺ ˻ ˼ ˺ ˻ ˼ fig. 5. medicinal chemical manufacturing plant model as a tandem graph. therefore, we may treat the medicinal chemical manufacturing plant model in the same manner as example 2, discussed in the previous section. more precisely, by specifying the value of the characteristic functions, c(α), c(β ), c(γ), c(α β ), c(β γ), c(γ α), and c(α β γ), we can calculate the shapley value for this model. for example, when we specify the value of these functions in line with example 2, the shapley value for the model is (α, β , γ) = ( 41 3 , 38 3 , 41 3 ) , which is, of course, the same result as in example 2. remark 2 in the two examples above, we designed the characteristic functions to satisfy the rule described in model 1. however, their specific values were chosen for convenience; and, though we have focused specifically on the construction of network models in this study, we are also concerned with the manner of specifying the value of such characteristic functions, however this is a consideration for the future. 4. discussion and concluding remarks we provided a cooperative game theoretical approach to risk analysis using network structure, based on shapley values. the approach would appear to have potential for solving risk allocation problems, since, if the total risk is known, the shapley values help us determine each segment’s respective contribution, and thereby allocate segmental risk according to these values. in general, we may calculate shapley values as follows. let n be the set of all segments. first, we define a characteristic function c : 2n → r, where 2n and r denote the sets of all subsets of n and real numbers, respectively; and the function has the following properties: (i) c( /0) = 0, (ii) c(s ∪ t ) 6 c(s) + c(t ), where s and t are subsets of n satisfying s ∩ n = /0. note that s ∩ t = /0 means that there no segment belonging to both sets s and t . then, segment i’s shapley value is defined by φi(c) := ∑ s⊂n\{i} |s|!(n −|s|− 1)! n! (c(s ∪{i}) − c(s)) , published by atlantis press copyright: the authors 47 j. takeshita, h. mohri where n is the total number of segments, and |s| denotes the number of segments belonging to the set s (i.e., |n| = n). in the equation, c(s ∪ {i}) − c(s) means that the marginal contribution of i to s. hence, the shapley value indicates a weighted average of the respective marginal contributions. in addition, a shapley value has the following five properties: (i) for any i ∈ n, φi(c) 6 c(i) is satisfied. (ii) ∑ i∈n φi(c) = c(n). (iii) if c(s ∪ i) = c(s ∪ j) holds for any s ⊂ n satisfying i, j 6= s, then φi(c) = φ j(c). (iv) for any s ⊂ n satisfying i 6= s, c(s ∪ {i}) = c(s) is satisfied. (v) for any characteristic functions c1 and c2, φi(c1 + c2) = φi(c1) + φi(c2) is satisfied. it has been well established that a shapley value is the only solution concept that satisfies the five properties above (aumann and hart, 1994c, ch. 53). we used the shapley value as a solution concept in our cooperative game models, but an alternative solution concept, that of the nucleus, is also available for this purpose (aumann and hart, 1994a, ch. 18). a shapley value is defined based on each segment’s contribution when that segment cooperates with others, while the nucleus concept is based on a given segment’s dissatisfaction with its respective coalitions. since the approaches are so different, it is difficult to generalize about their respective superiority; this will likely be determined by the specific context of the problem faced. although the framework introduced in this paper is the first attempt of its kind for cooperative game theory, it appears to offer a promising approach to risk analysis, especially in the context of risk allocation. acknowledgments the authors wish to thank dr. ryoji makino for his support and fruitful discussion in improving this paper. 1. r. j. aumann and s. hart, handbook of game theory with economics applications, volume 1, north holland, amsterdam, (1994a). 2. r. j. aumann and s. hart, handbook of game theory with economics applications, volume 2, north holland, amsterdam, (1994b). 3. r. j. aumann and s. hart, handbook of game theory with economics applications, volume 3, north holland, amsterdam, (1994c). 4. t. bedford and r. cooke, probabilistic risk analysis: foundations and methods, combridge university press, cambridge, (2001). 5. d. fudenberg and j. tirole, game theory, mit press, ma, (1991). 6. k. hausken, “probabilistic risk analysis and game theory,” risk analysis, 22, 17–27 (2002). 7. e. lawler, combinatorial optimization: networks and matroids, dover publications, new york, usa, (2011). 8. j. f. nash, “non-cooperative games,” the annals of mathmatics, second seties, 54 286–295 (1951). 9. j. f. nash,“two person cooperative games,” econometrica, 21, 128–140 (1953). published by atlantis press copyright: the authors 48 microsoft word risk assessment for a chemical spill into a river risk assessment for a chemical spill into a river david mccready* the dow chemical company 2001 union carbide drive south charleston, wv, 25303 usa dmccready@dow.com james b. williams the dow chemical company south charleston, wv, 25303 usa jim-bob@dow.com abstract a chemical spill into a river presents a significant threat to human health and the aquatic environment. a mathematical model is presented for predicting the chemical concentrations in a river resulting from a chemical spill. a screening risk assessment approach is presented to characterize the risk to humans and aquatic life resulting from a single spill. human exposure is evaluated using two approaches; exposure from drinking contaminated water and aggregate exposure. the potential for adverse effect to aquatic organisms is evaluated by comparing the predicted environmental concentrations to the lethal effect concentration for the most sensitive species as well as to the no observed effect concentration. the model and the approach are used to evaluate a small hypothetical spill of methanol. the example provides guidance for evaluating other chemical spill scenarios. keywords: river dilution model, aggregate exposure, aquatic exposure, methanol. * corresponding author 1. introduction the capability to determine the human and aquatic exposure from a chemical spill into a river is necessary for industry to evaluate the risk from an actual spill. this capability is also useful for emergency response planners to evaluate hypothetical spill scenarios. the first step of a chemical spill analysis may use a mathematical model to predict the chemical concentrations in a river. models are particularly useful when measurements cannot be taken, for example when predicting the effects of a chemical not actually spilled or perhaps not even manufactured yet. a short literature review 1 of models for predicting the chemical concentration in a river after a spill has been previously reported. models can be classified as simple or refined; the model complexity can range from a single equation based on complete mixing to a refined model for evaluating the chemical distribution in an entire river system. a simple model can be applied when chemical concentrations are reduced primarily by dilution. one of the simplest models is the volumetric dilution equation 2 to estimate the chemical concentration after complete mixing in a river. the river dilution model (rdm) 1 provides more information about the spill than the volumetric dilution equation while it requires less input data and training than a refined model. a refined model needs more input data so it is typically applied to journal of risk analysis and crisis response, vol. 3, no. 3 (november 2013), 116-126 published by atlantis press copyright: the authors 116 willieb typewritten text received 26 september 2013 willieb typewritten text accepted 17 october 2013 willieb typewritten text willieb typewritten text david mccready, james b. williams a well defined river. another way to classify the models is by the release type, either a continuous release or an instantaneous release; each release type uses a different model formulation. the second step of a spill analysis is to characterize the risk using the model predicted chemical concentrations. this step provides guidance on whether the potential exposure to humans and aquatic life is acceptable or unacceptable. human exposure is sometimes evaluated using the assumption that the contaminated water is the primary drinking water source. this is a conservative assumption but it may not be realistic since most people do not drink untreated river water. this study presents a second approach for assessing human exposure based on aggregate (oral, dermal, and inhalation) exposure. 2. spill requirements and legislation a chemical spill may result from a tank rupture, equipment failure, overfilling, vandalism, transportation accident, or improper operation. at a stationary facility, a secondary containment structure is required around a storage tank so any spilled liquid can be recovered. however, spilled or leaked liquids without secondary containment may flow into surface water or infiltrate into the ground. a transportation accident on land may result in a spill which flows into surface water if secondary containment is not immediately installed. industry has a financial incentive to prevent spills because they have to pay to remove contaminated soil and also have to replace the spilled material. industry also wants to prevent spills for regulatory reasons since they may have to cease operations during an inspection and they may be fined if the spill was preventable. thus, there is a strong interest to prevent chemical spills and contain a spill if it occurs. there are many federal regulations 3 in the u. s. a. concerning chemical spill reporting. section 311 of the clean water act requires notifying the national response center (nrc) for any release that violates applicable water quality standards, causes a discoloration of the water, or deposits sludge. section 103 of the comprehensive emergency response, compensation, and liability act requires notifying the nrc for release of any listed hazardous substance with a reportable quantity. any person in charge of a facility (or a vessel) must immediately notify the nrc as soon as there is knowledge of a release greater than the reportable quantity. section 304 of the emergency planning and community right-to-know act requires notification to each local emergency planning committee for a spill over the reportable quantity. section 1808 of the hazardous material transportation act of 1974 requires notification to nrc for a release of a dot hazardous material that occurs during transport. the oil pollution prevention regulation applies to a spill containing oil which flows into a surface water; reporting such a spill to the national response center is required. the u.s. coast guard (uscg) has jurisdiction concerning spills on navigable waters. most state regulatory agencies also regulate chemical spills but the specific regulations are too numerous to describe here. this study used the model and the approach for risk characterization to evaluate a hypothetical spill of methanol and its subsequent environmental fate. methanol (methyl alcohol), cas 67-56-1, was selected to evaluate because it is commonly used, its chemical properties are well known 4, and its toxicity is well characterized. methanol is used as feed stock for manufacturing organic chemicals; it is an ingredient in antifreeze products, solvents, gasoline, and hydraulic fracturing fluids. the u.s. production is about 2.25 billion gallons of methanol per year.4 it is transported as 99-100% (w/w) methanol by barge, railcar, tank truck, totes, drums, and smaller containers. for the purposes of demonstrating the utility of the proposed model we have selected the example of a small instantaneous spill with a mass well below the reportable quantity of 2,272 kg; it is not intended to reflect a worst case release. this study is not intended to provide a rigorous evaluation of all possible methanol spill scenarios. the objective of this paper is to present a screening model to predict concentrations from a chemical spill in a non-tidal river and how to use the predicted concentrations to characterize the potential human and aquatic risks using appropriate toxicity benchmarks. it presents an example screening level evaluation which can be applied to other chemical spill scenarios. this study is a significant contribution because little information has been published on how to use model predicted concentrations to evaluate the risk from an instantaneous chemical spill into a river. published by atlantis press copyright: the authors 117 risk assessment for a chemical spill into a river 3. methods the following are discussed below: exposure scenario, model description, model input data, model calibration, and risk characterization. 3.1. exposure scenario this study evaluated a 150 kilogram spill of pure methanol which is equivalent to an entire 190 liter (50 gal) drum. this study assumed that the entire volume flowed instantaneously into the river. while there is low probability that the entire spill mass would flow instantaneously into the water, a screening model has not been identified for a quasi-instantaneous release 1. the instantaneous release model provides a conservative prediction of the peak concentration for a quasi-instantaneous release. although it may be more common for methanol to be spilled onto the ground, such an evaluation is beyond the scope of this study. 3.2. model description the mixing zone is the area where a chemical spill is initially mixed and diluted as it is transported downstream; this study assumes the chemical is not instantaneously mixed across the river width. the potential exposure is high in the mixing zone because the concentrations are high. a descriptive analysis of an instantaneous chemical release into a river can be obtained using a model. rdm2d, an improved version of rdm, is used in this analysis. the model equations have been previously derived and described, see refs. 5-7 for details. the peak concentration, c1, in the mixing zone is predicted as a function of the time after release using eq. (1):          td y td x dddt m c yxyx 44 exp )(4 22 5.01  (1) where c1 is the peak concentration (mg/l), m is the chemical mass (mg) spilled into river, d is river depth (m), dx is the longitudinal mixing coefficient (m 2/s), dy is the lateral mixing coefficient (m2/s), x is the longitudinal distance (m) downstream of the release point, y is the lateral distance (m), and t is the time (s) after the chemical spill enters the river. eq. (1) shows the peak concentration is directly proportional to the mass spilled. the chemical dilution is due to the water turbulence and it is quantified using lateral and longitudinal mixing coefficients whose values depend primarily on the river velocity which is estimated by the model. the vertical mixing is assumed to be fast compared to the longitudinal mixing and the vertical mixing (dz) is ignored in the two-dimensional dilution calculations. a second equation is used to predict the peak concentration, c2, for distances beyond the mixing zone where the chemical is uniformly mixed across the river width, as shown in eq. (2): 5.02 )4( xtda m c   (2) if one wants to predict the chemical concentration, c3, versus time at a fixed distance downstream of the release point after uniform mixing, then eq. (3) is used:           td utx tda m c xx 4 exp )4( 2 5.03  (3) where u is the river velocity (m/s). the three equations listed above are easy to solve in an excel spreadsheet. rdm2d has options for a release from the river bank and a release at the center of river. although the model assumes the river flow, width, and depth are uniform in a reach, it can evaluate sequential reaches with different property values. it can estimate the travel time for a spill to reach a specified distance downstream from the release point, such as a drinking water intake. the dilution model applies only to soluble chemicals; it does not apply to slightly soluble chemicals such as heavy oils which may sink or to light oils which may float on the water surface. rdm2d conservatively assumes there is no chemical loss due to hydrolysis, evaporation, reaction, or sorption. this assumption is acceptable for a screening analysis which predicts concentrations near the release point where concentrations are reduced primarily by dilution. 3.3. model input data one needs values for several parameters to execute the model: the total mass of chemical released, chemical properties, river flow rate, river width, and river depth. the material safety data sheet (msds) is the first place to search for information and it can typically provide the physical and chemical properties. the msds should be reviewed closely to determine if the model is applicable. the msds may also have human published by atlantis press copyright: the authors 118 david mccready, james b. williams and aquatic toxicity information for the chemical of interest. the usepa has excellent websites, such as iris (integrated risk information system) 8 for human toxicity information and ecotox (ecotoxicology) 9 database, for aquatic toxicity information. the mass of chemical released can be estimated from the container volume and the chemical density. the mass of each component of a mixture is modeled separately. the usgs 10 website has river flow rate and other data measured at over 8000 sites. if one is modeling an actual event then the river conditions at the time of the event should be used. if one is modeling a hypothetical event, then a range of river conditions 1, 11 could be evaluated such as small, medium, and large rivers. this study evaluates a hypothetical spill into two surface waters; granny creek which discharges into the elk river, a small river located in central west virginia, usa. this study illustrates a common occurrence where a spill initially flows into a creek with a small flow rate which discharges into a river with a larger flow. this study also illustrates how the river flow conditions can vary significantly. table 1 presents usgs 10 measured values for six flow conditions. the flow, width, and depth vary with the flow conditions. the minimum flow rate represents the worst case since there is the least amount of water available for chemical dilution and it would result in the highest chemical concentration while the maximum flow rate represents the best dilution case. table 1. flow conditions for the elk river and granny creek. river flow cond. flow (m3/s) width (m) depth (m) elk min. 0.01 32.9 0.4 river 10th pct. 12 63.7 1.8 50th pct. 40 70.7 2.3 mean 59 73.2 2.5 90th pct. 142 79.3 3.0 max. 1646 98.8 4.9 granny min. 0.001 1.8 0.001 creek 10th pct. 0.05 3.3 0.02 50th pct. 0.2 4.0 0.06 mean 0.3 4.1 0.09 90th pct. 0.6 4.6 0.15 max. 8.0 .6 1.04 3.4. model calibration the rdm2d performance is evaluated by comparing its predictions to measured concentration values reported in ref. 12 for releases of radioactive tracers into copper creek near gate city, va, usa. the copper creek data set measured concentrations as close as 0.2 km to the release point for flow rates in the same range as those for granny creek. rdm2d was executed to match the reported experimental conditions as closely as possible for each of the four sets of experimental data. the model has been previously calibrated with other chemicals in other rivers. environment canada 13 developed the envirotips manual for use by specialists to assess spill effects on the environment. the rdm2d predictions were compared to the envirotips example for a 20 tonne spill of acetic acid in a river. rdm2d predicted concentrations were compared to the measured and predicted concentrations reported in ref. 6 for an actual 791,000 kg spill of ethylene dichloride into a river after a train derailment. 3.5. risk characterization the usepa2, 14 guidance documents provide detailed discussions of the risk characterization for exposure to contaminated water. although this paper provides a simple discussion of risk characterization and the selection of a toxicity benchmark, this complex topic has been ignored in most papers on modeling exposure to an instantaneous chemical spill. to quantitatively characterize the risk to humans and aquatic life for a chemical spill, the predicted chemical concentration should be compared to an appropriate toxicity benchmark, exposure duration, and health effect. because a spill typically has a short release duration and it occurs infrequently, a spill is difficult to evaluate because the concentration and duration of exposure vary with distance from the release point and the time after the release. this section is intended to provide guidance to emergency response planners on how to better characterize the health risk. although it is common to use a national primary drinking water regulation (e.g., a drinking water standard) or perhaps use a discharge limit from a wastewater treatment plant to characterize the exposure, these apply to a continuous release and they may provide an overly conservative risk characterization for an instantaneous spill. the most commonly evaluated human exposure scenario for a chemical spill into a river is drinking contaminated water, perhaps because a large population could potentially be exposed via a contaminated potable water distribution system. one should compare the oral exposure to the appropriate oral toxicity benchmark to published by atlantis press copyright: the authors 119 risk assessment for a chemical spill into a river evaluate this scenario. dermal exposure may also occur via a contaminated potable water system and the dermal exposure should be compared to an appropriate dermal toxicity benchmark. if the chemical causes skin irritation, then the predicted chemical concentration may be compared directly to a concentration based on a skin irritation benchmark. inhalation exposure may occur via a contaminated drinking water system due to the chemical evaporation in a shower or bathtub, and the inhalation exposure should be compared to an appropriate inhalation toxicity benchmark. the ingestion of contaminated fish is not typically considered a significant source of exposure for a single chemical spill into a flowing river because the fish would likely have negligible uptake within the limited time frame. the human risk can be simply characterized using the hazard quotient (hq) described in eq. (4): tox exp hq  (4) where exp is the calculated exposure (mg/kg body weight/d ) and tox is the toxicity benchmark (mg/kg body weight/d). a hq value less than one is typically acceptable provided the toxicity benchmark includes appropriate uncertainty factors (uf) for the extrapolation of the test animal response to humans. the individual chemicals in a mixture are evaluated separately because the human health impact is based on the effect for the individual chemical. the total exposure may be characterized by summing the hq value for each of the chemicals. there is potential for risk if an individual hq is greater than one or the sum of all hqs (called the hazard index or hi) is greater than one. if the hq or hi is unacceptable then the exposure analysis may be refined. one would prefer to compare the predicted exposure to a toxicity benchmark derived from an acute dose which had no adverse health effect to the tested species or perhaps a temporary health effect, such as irritation. although the lc50 is a commonly available acute toxicity concentration, this is not an appropriate effect for human exposure since it is based on death of 50% of the test animals. if appropriate acute human toxicity data are not available, then one may have to use a toxicity benchmark based on repeated dose (i.e., chronic) animal exposure, have a study performed to determine the toxicity, or use a surrogate with similar chemical properties. the usepa15 has established an enforceable standard, the maximum contaminant level (mcl), in drinking water for about 80 chemicals. the mcl has a low value because it applies to long term, repeated oral exposure. the chemical concentration to which one may be exposed should ideally be lower than the mcl value. however, it is conservative to compare a chemical concentration from a single spill in a river with short exposure duration to the mcl which is based on long term exposure and this point is missed in some risk assessments. as an alternative to the mcl, the usepa 16 established health advisories (ha) as an estimate of the acceptable drinking water level which can be used for evaluating a chemical spill. although the ha provides guidance on acceptable exposure, it is not a legally enforceable federal standard. the one-day, tenday, and lifetime ha values are considered protective of adverse noncancer health effects in a child who may receive a greater dose (e.g., on mg/kg basis) than an adult. a short duration toxicological study may be used to derive a one-day ha although a long duration toxicological study may be substituted. the one-day ha is derived using eq. (5): dwiuf bwnl ha    (5) where the nl is the no observed adverse effect level, noael, (mg/kg/d) from a study of an appropriate duration, bw is the body weight (kg), uf (unitless) is the total uncertainty factor, and dwi is the default daily water intake for a child (1 l/day). the default body weight of 10 kg correlates to a one year old child. the entire chemical dose is assumed to be from drinking water contaminated by the chemical spill. only the lifetime ha has an adjustment for possible carcinogenicity. the usepa 12 provided guidance on selecting the uncertainty factors; their values depend on the quality of the available toxicity studies, the extent of the chemical toxicity database, and scientific judgment. the integrated risk information system (iris) 8 may be helpful for deriving a ha value since it has a compilation of human health effects including the oral noael for approximately 540 chemicals. human exposure to a chemical spill via a contaminated potable water system may have a low probability of occurrence, so this study also evaluated published by atlantis press copyright: the authors 120 david mccready, james b. williams the aggregate exposure (oral, dermal, and inhalation) for a fisherman. it is assumed the fisherman incidentally ingested contaminated water while wading in the creek and the oral dose is estimated using eq. (6): bw ic dose dwworal   (6) where doseoral is the oral dose (mg/kg/d), cw is the water concentration (mg/l), idw is the drinking water intake rate (l/d), and bw is the adult body weight. the dermal dose is estimated using eq. (7): 131000    lcmbw edatsac dose filmskinw dermal (7) where dosedermal is the dermal dose (mg/kg/d), cw is the water concentration (mg/l), saskin is surface area of the skin (cm2), tfilm is the water film thickness on the skin (cm), da is the relative dermal absorption of the chemical (%), and e is the number of events per day. the inhalation dose is estimated using eq. (8): bw dic dose airairinhalation   (8) where doseinhalation is the inhalation dose (mg/kg/d), cair is the air concentration (mg/m3), iair is the default inhalation rate (m3/h), and d is the exposure duration (h/day). the methanol evaporation rate is estimated using an evaporation model 17. the methanol concentration in the air over the spill is estimated using the evaporation rate in a simple air dispersion model 18. the usepa19 characterizes the acute exposure to fish or invertebrate species using the risk quotient (rq) approach described in eq. (9): tox exp rq  (9) the rq is a single point estimate of acute risk based on exp, the peak chemical concentration (mg/l) in the river and tox, the single point estimate of the toxicity (mg/l), either the lc50 or the ec50. the lc50 is a lethal effect and the lowest value (most sensitive) of all tested freshwater species is typically used. the usepa19 interprets a rq less than 0.1 as acceptable in most cases. however, the usepa requires a rq less than 0.1 for an endangered species and it may be appropriate to use a lower rq for commercially or recreationally important species. the risk characterization is more complicated because there are many aquatic species and test data are reported for different test conditions, effects, and units. one should select a species which could be present in the river of interest. 4. results rdm2d is applied in a case study to illustrate the approach to evaluate human and aquatic exposure for a hypothetical spill. the results are discussed below. 4.1. model performance rdm2d conservatively over predicts the measured concentrations reported in ref. 12 by up to a factor of 2.8 at distances greater than 0.7 km downstream of the release point in the model calibration, as shown in fig. 1. rdm2d under predicts the measured concentrations by a factor of 0.4 to 0.7 at 0.2 km from the release point. fig.1. rdm2d normalized (predicted/measured) concentrations versus distance from the release point. another measure of model performance is how well it predicts the time for the arrival of the peak concentrations. rdm2d under predicts the measured time for the arrival of the peak concentrations; it under predicts as much as a factor of two at distances beyond 5 km. 4.2. toxicity benchmarks methanol does not have a mcl or a ha so a one-day ha is developed. the reported rfd 8 is 0.5 mg/kg/d based on the noel of 500 mg/kg/d from a subchronic oral study in rats using a total uf of 1,000. the current analysis used the noel and a uf of 10 to account for published by atlantis press copyright: the authors 121 risk assessment for a chemical spill into a river interspecies extrapolation (from animals to humans) and a uf of 10 to account for intraspecies extrapolation (potentially sensitive individuals). this analysis did not include the usepa 8 proposed uf of 10 to account for extrapolation from subchronic to chronic exposure since this analysis is determining an acceptable exposure for one-day of exposure rather than chronic exposure. the one-day ha is calculated as 50 mg/l using eq. (5). the aggregate exposure calculation requires an inhalation dose and a toxicity benchmark. the usepa has not established an inhalation toxicity benchmark for methanol. a chronic reference exposure level (rel) 20 of 4 mg/m3 was established after observation of developmental malformations in mice. this study converted the rel to an allowable daily dose of 1 mg/kg/d for an adult with a mean inhalation rate 21 of 20 m3/d and a mean body weight 21 of 80 kg. the ecotox 9 database reports over 400 values for methanol. the msds 22 reports methanol toxicity results for several freshwater aquatic species. daphnia pulex, a common species of the water flea, has a lc50 of 19,500 mg/l for the 18 h test duration. daphnia obtuse, another common species of the water flea, has an ec50 of 23,500 mg/l for the 24 h test duration. pimephales promelas (fathead minnow) has a lc50 of 29,700 mg/l for the 24 h test duration. daphnia has a no observed effect level (noel) of 10,000 mg/l for the 48 h test duration. the no-observed-effect concentration (noec) to a freshwater aquatic ecosystem is reported in ref. 23 as 23.75 mg/l based on sublethal effects of methanol in a mesocosm study for a 90 day period. the lc50 of 19,500 mg/l for daphnia pulex was selected to calculate the rq for lethal effects and the noec of 23.75 mg/l was selected as the no effect concentration. 4.3. model predictions rdm2d is used to predict concentrations versus distance downstream from the release point for the hypothetical methanol spill scenario because a specific location in a river could not be identified where a person might be exposed (e.g., a drinking water intake). figure 2 presents the predicted methanol concentrations versus distance for a spill into granny creek. table 2 summarizes the predicted distances to methanol toxicity benchmarks for the different flow conditions. the spill would be diluted below the lc50 concentration of 19,500 mg/l within about 130 m of the release point for the 10th percentile flow, within 28 m for the mean flow rate, and within 17 m for the 90th percentile flow rate in granny creek. 1000 10000 100000 1000000 0 200 400 600 800 1000 c o n ce n tr a ti o n ( m g /l ) distance from release point (m) 10th percentile flow mean flow 90th percentile flow lc50 = 19, 500 mg/l fig.2. rdm2d predicted methanol concentrations versus distance in granny creek. the spill would be diluted below a concentration of 1,950 mg/l (which is equal to an rq of 0.1) within about 13,700 m for the 10th percentile flow, within 1,430 m for the mean flow, and within 560 m for the 90th percentile flow. the predicted distances to concentrations below the noec of 23.75 mg/l far exceed the actual length of the creek. the distance to the ha value of 50 mg/l is not predicted because granny creek is not a drinking water source. table 2. summary of model predicted distances to the toxicity benchmarks for various flow conditions. river flow cond. toxicity benchma rk value (mg/l) distance to (m) granny 10th pct. lc50 19,500 130 creek mean lc50 19,500 28 90th pct. lc50 19,500 17 10th pct. rq=0.1 1,950 13,700 mean rq=0.1 1,950 1,430 90th pct. rq=0.1 1,950 560 10th pct. noec 23.75 >10 6 mean noec 23.75 >10 6 90th pct. noec 23.75 >10 6 elk 10th pct. lc50 19,500 <1 river mean lc50 19,500 <1 90th pct. lc50 19,500 <1 10th pct. rq=0.1 1,950 7 mean rq=0.1 1,950 4 90th pct. rq=0.1 1,950 3 10th pct. 1-d ha 50 70 mean 1-d ha 50 40 90th pct 1-d ha 50 30 10th pct. noec 23.75 120 mean noec 23.75 80 90 th pct. noec 23.75 70 published by atlantis press copyright: the authors 122 david mccready, james b. williams figure 3 shows the predicted concentrations are much lower for the same spill into the elk river. the spill would be diluted below the lc50 of 19,500 mg/l at a distance of about one m from the release point for all flow rates. the spill would be diluted below 1,950 mg/l (equivalent to an rq of 0.1) for the 10th percentile flow rate within about 7 m from the release point, within 4 m for the mean flow rate, and within 3 m for the 90th percentile flow rate. the spill would be diluted below the ha of 50 mg/l for the 10th percentile flow rate within about 70 m from the release point, within 40 m for the mean flow rate, and within 30 m for the 90th percentile flow rate. the spill would be diluted below the noec of 23.75 mg/l for the 10th percentile flow rate within about 120 m from the release point, within 80 m for the mean flow rate, and within 70 m for the 90th percentile flow rate. 1 10 100 1000 10000 100000 0 50 100 150 200 c on ce n tr at io n ( m g/ l ) distance from release point (m) 10th percentile flow mean flow 90th percentile flow lc50 = 19, 500 mg/l ha = 50 mg/l fig.3. rdm2d predicted methanol concentrations versus distance in elk river. figure 4 presents some details on travel time and methanol concentrations versus time for the spill into granny creek during the 10th percentile flow rate. the spill length is stretched and the peak concentration decreases as the spill flows downstream. this type of analysis might be performed at a drinking water intake.  at 300 m downstream, the leading edge of the spill arrives after 60 s, the peak arrives at 220 s, and the trailing edge passes at 760 s, the potential exposure duration is about 700 s, and the time weighted average (twa) concentration is estimated as 4,075 mg/l using eq. (3).  at 600 m downstream, the leading edge of the spill arrives after 200 s, the peak arrives at 470 s, and the trailing edge passes at 1,000 s; the potential exposure duration is about 800 s and the twa methanol concentration is 3,150 mg/l. 0 2000 4000 6000 8000 10000 12000 14000 0 200 400 600 800 1,000 c o n ce n tr a ti o n ( m g /l ) time after release (s) x = 300 m x = 600 m fig.4. rdm2d predicted methanol concentrations versus time after release into granny creek. 4.4. aggregate human exposure the aggregate exposure is evaluated for an adult fisherman wading in granny creek using the worst case scenario with the highest predicted concentrations for the spill during the 10th percentile flow rate. it evaluated the twa concentration at 300 m downstream of the spill. granny creek is only 0.02 m deep and 3.3 m wide during the 10th percentile flow rate. it is assumed that the fisherman immersed his feet and the dermal dose is calculated using the following assumptions: cw is the twa methanol concentration of 4,075 mg/l, saskin is mean surface area of the adult male feet 21 of 1,370 cm2, tfilm is the water film thickness on the skin 21 of 0.005 cm, e is one event per day, da is the default relative dermal absorption of 100 %, and bw is the default mean adult body weight 21 of 80 kg. the predicted dermal dose is calculated as 0.35 mg/kg/d using eq. (7) and the hqdermal is calculated as 0.13 using eq. (4). the inhalation dose calculation assumed the fisherman inhaled methanol evaporating from the spill. the methanol evaporation rate is estimated as 0.025 g/s at 298k for a 1 m/s wind speed using the evaporation model. the methanol concentration in the breathing zone over the spill, cair, is estimated as 2.9 mg/m 3 using the air dispersion model. the default inhalation rate 21, iair, is 0.013 m 3/min and d is the exposure duration of 11.7 min. the predicted inhalation dose is calculated as 0.006 mg/kg/d using eq. (8) and the hqinhalation is calculated as 0.01 using eq. (4). published by atlantis press copyright: the authors 123 risk assessment for a chemical spill into a river the usepa 21 reported an incidental water ingestion of 0.053 l/event (maximum value) for an adult during swimming or diving activities and this value is extrapolated to 0.053 l/d for a fisherman wading in granny creek. the worst case oral dose is calculated for incidental ingestion of methanol contaminated water using the following assumptions: the water concentration, cw, is 4,075 mg/l, and the incidental water ingestion, idrw, is 0.053 l/d. the predicted oral dose is calculated as 2.7 mg/kg/d using eq. (6) and the hqoral is calculated as 0.54 using eq. (4). the aggregate hq (or hi) of 0.62 is the sum of hqdermal of 0.07, the hqinhalation of 0.01, and the hqoral of 0.54; the hi value is acceptable since it is less than one. 5. discussion the rdm2d model allows for a quick screening analysis of a chemical spill by providing details about its dilution and travel time. the model shows the highest concentrations occur within the mixing zone and decrease with distance from the release point. rdm2d calculates the peak concentration on the release centerline at the river surface as a function of downstream distance in the mixing zone. this provides a conservative screening estimate since the maximum predicted concentration is on the release centerline. the release centerline location is influenced by river flow irregularities and one cannot simply predict its exact location. the rdm2d model under predicted measured concentrations at 0.2 km from the release and it over predicted as much as a factor of three at distances greater than 0.7 km from the release. ref 5. states if the model predicted concentration value is within a factor of four of the observed value then this is considered a reasonably good agreement. the rationale is that dilution varies in a river due to irregularities (bends, sandbars, dead zones, riffles, structures, etc.) which alter the dilution but cannot be precisely defined so the mixing coefficients cannot be accurately defined nor can their variability be considered in a simple model. the rdm2d predictions compared well to predictions from the two other screening models6, 13. the model performance may be improved by tuning it to the river of interest using a better representation of the channel shape and cross-sectional area. rdm2d conservatively assumes there is no chemical loss due to hydrolysis, evaporation, reaction, or sorption. although this assumption may cause it to over predict concentrations for long travel times, it is a reasonable assumption because the travel time is shorter (highest travel time is three h in granny creek) than the chemical decay half-life in the river (lowest value is about one d) as discussed below in environmental fate. this study identified several key parameters for dilution of a chemical spill and exposure. the river flow rate is a critical parameter; the highest concentration occurs for the lowest flow condition. table 1 shows a wide range in flow conditions, for example the 90th percentile flow rate is about 12 times greater than the 10th percentile flow rate for both granny creek and the elk river. granny creek provided much less dilution especially at low flow conditions. the spill travel time is important because the slowest travel time and the highest potential exposure duration occurs for the low flow conditions. the predicted spill travel time is also important for emergency response immediately after a spill; the fastest travel time gives less time to respond and it occurs for high flow conditions. the spill mass is a key exposure parameter because the predicted chemical concentration is directly proportional to the mass. for the small methanol spill example, the human impact based on the one-day ha value of 50 mg/would be limited to a distance less than 70 m from the release point in elk river. the elk river has a greater capacity to dilute a spill since its 10th percentile flow rate is 240 times greater than that for granny creek. for a methanol spill into granny creek, the concentrations would be quickly diluted as the stream flows into the elk river; the spill dilution below 50 mg/l is estimated to occur within 70 m from the above results. if the usepa has not developed a ha value for the chemical of interest then the risk characterization for a chemical spill can be complex. one can see why it is common to simply use a drinking water standard, such as a mcl, to characterize the exposure to a spill even though it applies to a continuous exposure. however, the ha provides a higher allowable exposure for a spill than the mcl. for example, the one-day ha for toluene is 20 times higher than its mcl.16 the one-day ha derived in this study provides a conservative toxicity benchmark since it is based on a noael from a repeat dose study rather than a short duration study. the aggregate exposure approach provides another way to characterize the potential human exposure to a spill that is more realistic if exposure via contaminated published by atlantis press copyright: the authors 124 david mccready, james b. williams drinking water has a low probability of occurrence. the aggregate exposure calculation, which evaluated an adult fisherman wading in the creek, is more relevant to granny creek since it is not used as a drinking water source. oral exposure has the highest hazard quotient (hqoral of 0.54). however, the hi of 0.08 for inhalation and dermal exposure would better characterize the potential exposure if there is no incidental ingestion of the contaminated creek water. figure 4 shows the exposure duration is about 0.2 h for the spill into granny creek at the 10th percentile flow rate. because the lc50 value is based on an 18 h test duration, it provides a very conservative risk characterization; the calculated exposure duration of only 0.2 h is 90 times lower than that for the lc50 test. others 24 have reported the inconsistency of using lc50 results based on 48 or 96 h tests to characterize the aquatic risk for a spill with a short exposure duration. however from a practical standpoint, one must either use the available lc50 test results or perform additional toxicity testing. it is assumed that the remaining aquatic population will recover, but the recovery could take a long time during which the rest of the aquatic community and food chain are disturbed. the spill concentration would never be diluted below the no effect concentration, the noec, in granny creek for the 10th percentile flow rate but it would be diluted within about 120 m after it flows into the elk river. ref. 4 reports the fate of methanol spilled into surface water. methanol is completely miscible in water and it will dissolve quickly. methanol will biodegrade after its concentration is diluted below the toxic level. its biodegradation half-life (the time required for 50% mass reduction) is 1 to 7 days in surface water. methanol is not expected to adsorb to suspended solids and sediment. its bioconcentration potential in aquatic organisms is low. 22 methanol evaporates slowly from an aqueous solution because the methanol molecules form a strong bond to the water molecules. the usepa 25 epi suite™ v4.1 program predicted the volatilization half-life as three days using its river model with the default conditions. 6. conclusions the rdm2d model is suitable for a practical, screening level analysis of an instantaneous chemical spill into a non-tidal river. the model predicted concentrations provided reasonably good agreement in the calibration to the measured values reported in the literature. guidance is provided on quantifying river parameters, evaluating chemical properties, and characterizing the human and aquatic risks. if one is modeling an actual spill event then the river conditions at the time of the spill should be used. the flow conditions in a river can vary widely and this needs to be accounted for in modeling a hypothetical spill scenario. the low flow rate represents the worst case since it results in the highest chemical concentrations while the high flow rate predicts the fastest travel time giving less time for emergency response immediately after a spill. the methanol spill example provides guidance which can be used to evaluate other chemical spill scenarios. two approaches are presented to characterize the potential human risk from a hypothetical spill; exposure via drinking water and aggregate exposure. rdm2d predicted the small methanol spill in the elk river would be diluted below the one-day ha value for human exposure within 30 to 70 m of the release point. the aggregate exposure (oral, dermal, and inhalation) for a fisherman wading in granny creek during the worst case flow conditions had a hi of 0.62 which is acceptable. the aquatic risk, based on 0.1 of the lc50 concentration (corresponding to an rq of 0.1), could extend the entire length of granny creek while the spill into the elk river would be diluted below this value within about 7 m. the spill would never be diluted below the no effect concentration in granny creek before it flowed into the elk river, but the same spill into the elk river would be diluted below the no effect concentration within about 120 m. the human toxicity benchmark selection and the risk characterization can be a complex process. characterization of the aquatic risk is complicated by the wide range of species, toxicity values, test durations, and effects. the approach described in this paper provides a better risk characterization for a spill. the human and aquatic toxicity benchmarks suitable for a continuous chemical release may be used when they are readily available but they may provide an overly conservative characterization of a spill and this point is missed in some risk assessments. acknowledgements the dow chemical company supported this paper for good product stewardship. published by atlantis press copyright: the authors 125 risk assessment for a chemical spill into a river references 1. d. mccready and j. williams, a simplified approach to evaluate human and aquatic exposure to a chemical spilled in a river, j. hazard. mater. 193 (2011) 225– 232. 2. united states environmental protection agency (usepa), technical support document for water quality-based toxics control, epa/505/2-90-001, washington, d.c. 1991. 3. gaia corporation, chemical spill plans & accidental release notification requirements (2013) available on line: http://www.ehs-world.net/spill-plan-notifications. 4. malcolm pirnie, inc. evaluation of the fate and transport of methanol in the environment, prepared for american methanol institute, (1999) available on line: http://www.methanol.org/environment/resources/enviro nment/mp-methanol-fate.aspx. 5. h. fischer, e. list, r. koh, j. imberger and n. brooks, mixing in inland and coastal waters (academic press inc. new york, ny, 1979). 6. w. neely and r. lutz, estimating exposure from a chemical spilled into a river, j. hazard. mater. 10 (1985) 33–41. 7. c. van leeuwen, t. vermeire (eds.). risk assessment of chemicals: an introduction (springer, dordrecht, the netherlands, 2007) pp. 171–173. 8. united states environmental protection agency (usepa), integrated risk information system (iris), (2013) available on line: http://www.epa.gov/ncea/iris/. 9. united states environmental protection agency (usepa), ecotoxicology (ecotox) database, (2013) available on line: http://cfpub.epa.gov/ecotox/ 10. united states geological survey (usgs), surface water data for the nation, (2013) available on line: http://waterdata.usgs.gov/nwis/. 11. m. jamali, g. lawrence and k. maloney, fate of methanol spills into rivers of varying geometry. csce/ewri of asce environmental engineering conference, niagara, 2002. 12. h. jobson, prediction of traveltime and longitudinal dispersion in rivers and streams, u.s. geological survey, office of surface water, water-resources investigations report 96-4013, (1996) available on line: http://water.usgs.gov/osw/pubs/disp/dispersion.pdf, 1996. 13. environment canada, introduction manual, environmental and technical information for problem spills (envirotips), technical services branch, environmental protection programs directorate; ottawa, ontario, canada, 1985. 14. united states environmental protection agency (usepa), 2002. a review of the reference dose and reference concentration processes. epa/630/p-02/002f, washington, d.c. 15. united states environmental protection agency (usepa), drinking water contaminants, (2013) available on line: at http://www.epa.gov/safewater/conta minants/index.html#organic/. 16. united states environmental protection agency (usepa), drinking water standards and health advisories, (2013) available on line: http://water.epa.gov /action/advisories/drinking/. 17. d. mccready and j. foster, estimating spill quantities and emissions, 89-55.4, presented at the 82nd annual meeting of the air and waste management association in anaheim, ca, june, 1989. 18. m. scheringer, t. vögl t, j. von grote, b. capaul, r. schubert r, and k. hungerbühler, scenario-based assessment of multi-use chemicals: application to solvents. risk analysis, 21 (2001) 481-497. 19. united states environmental protection agency (usepa), ecological risk assessment website, (2013) available on line: http://www.epa.gov/oppefed1/ecorisk _ders/toera_risk.htm. 20. oehha (california environmental protection agency office of environmental health hazard assessment). 2001. chronic toxicity summary for methanol, cas 6756-1, (2013) available on line: oehha.ca.gov/air/chronic _rels/pdf/67561.pdf. 21. united states environmental protection agency (usepa), exposure factors handbook: 2011 edition, available on line: http://cfpub.epa.gov/ncea/risk/recor display.cfm?deid=236252. 22. methanex corporation. material safety data sheet for methanol, (2008) available on line: http://www.columbus.k12.wi.us/cms_files/resources/mx -methanol-msds-us-english-final-sept08.pdf. 23. a. kaviraj, f. bhunia and n. saha, toxicity of methanol to fish, crustacean, oligochaete worm, and aquatic ecosystem, int j toxicol. 23 (2004) 55-63. 24. b. kontic and m. gerbec, the role of environmental accidental risk assessment in the process of granting development consent, risk anal. 29 (2009) 1601-1614. 25. united states environmental protection agency (usepa), estimation program interface (epi) suite, (2013) available on line: http://www.epa.gov/opptintr/ exposure/pubs/episuite.htm published by atlantis press copyright: the authors 126 microsoft word environmental risk discrimination and assessment for municipal solid waste secure landfill site environmental risk discrimination and assessment for municipal solid waste secure landfill site cao yonghong urban and environmental sciences school, northeast normal university, changchun 130024, china abstract with development of urbanization, the amount of municipal solid waste is increasing. landfill is the main method for disposing the urban garbage at present in china. but the secure landfill site is not safe completely because it has a lot of risks threatening surrounding environment. after types of environmental risk in municipal solid waste secure landfill site were analyzed, i summarized that the risks in waste secure landfill site include water environment risks, air environment risks and waste pile sliding risks which are results caused by anthropic, equipment and natural factors. the methods of calculating risk probability, affecting scope and intensity were presented. the ways and means of risk prevention and management were put forward. keywords: municipal solid waste; secure landfill site; environmental risk. 城市生活垃圾安全填埋场环境风险因素识别与评价 曹勇宏 东北师范大学城市与环境科学学院 , 长春 130024, 中国 摘要：随着城市化进程的加快，各个城市的生活垃圾产生量也日益增多。卫生填埋仍然是我国处理生活 垃圾的主要途径。但安全填埋场并不安全，其存在着很多风险因素威胁着周围区域环境的安全。分析了城市生 活垃圾安全填埋场存在的各种风险类型，认为人为风险、设备风险以及自然风险是诱因，水环境风险、大气环 境风险、垃圾堆体滑动风险是结果，提出了估算垃圾场风险发生概率、风险影响范围及影响强度的方法，最后 阐明了填埋场风险防范措施。 关键词：生活垃圾，安全填埋场， 环境风险 环境风险是指在一定区域或环境单元内，通过 介质传播的，由人为活动或自然等原因引起的“意 外”事故对人类、社会与生态等造成的不良影响。 而环境风险评价是对环境风险事件发生的概率、影 响的范围、后果的严重程度、可能带来的损失等进 行评估并提出相应对策的过程[1,2,7]。 随着我国城市化进程的加快，城市垃圾的产生 量也在迅速增长，而且伴随着居民生活水平的提 高、城市工业结构和能源结构的调整，城市垃圾的 组成成分也越来越复杂，增加了城市垃圾的处理难 度。目前，垃圾的处理方式主要有堆肥、焚烧、回 收利用及填埋。由于填埋法成本低、技术简单、处 理迅速，仍然是我国应用最为广泛的垃圾处置方 式。 虽然垃圾填埋场使生活垃圾得到了有效的处 置，但并没有消除其污染性与有害性，并且这种集 中堆放的方式还会产生一些新的污染风险。比如垃 圾渗滤液的外泄事故[3]、填埋气火灾爆炸事故[4]， 都严重威胁着填埋场以及填埋场周围的环境安全。 这意味着垃圾安全填埋场并不是绝对地安全，其中 存在着许多风险及不安全因素。如能有效地识别风 险、规避风险，并在此基础上进行有效的风险管理 对保障一个区域的环境安全具有十分重要的意义。 1. 城市垃圾安全填埋场正常运转时的污染与风 险防治措施 1.1 填埋场主要污染与环境风险 垃圾填埋场的主要污染因素为垃圾渗滤液、扬 尘、恶臭、随风刮动的塑料、纸张、垃圾堆体滑动 等。主要的风险因素为垃圾渗滤液的外泄、以甲烷 为主要成分的填埋气体的爆炸。 journal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 149-155 published by atlantis press copyright: the authors 149 1.1.1 垃圾渗滤液 垃圾渗滤液中含有大量的高浓度有毒物质，如 果不经处理排入到环境中，则会使生态系统受到严 重污染和损害，并通过食物链对人体健康产生威胁 [5]。垃圾渗滤液来源主要有三种，一是垃圾内的含 水成分在垃圾堆体的强大压力下，水分被挤压出 来；第二部分为垃圾内的有机成分发生生化反应产 生的水分；第三部分为降落到垃圾堆体上的雨水以 渗滤液的形式排出。 1.1.2 填埋气体 填埋气体主要是由于垃圾内的有机成分在微生 物生化降解作用下，产生甲烷、二氧化碳、氨气、 水以及硫化氢等混合性气体(填埋气)。 垃圾填埋场对上述两种主要污染与风险因素通 常采取以下工程与管理措施进行防治。 1.2 工程防治措施 1.2.1 垃圾渗滤液污染防治措施 (1) 基底与边坡防渗措施 垃圾场底部以及边坡的水平防渗层主要是利用 聚乙烯材料、土工布或者是其它防渗效果好的材料 进行防渗，同时辅以粘土压实等方法，可以增加防 渗性并可有效防止防渗材料被石头等硬物划破。垃 圾堆放场地在没有边坡拦挡的开口处要建设具有防 渗性能的垃圾坝，并在坝下采用垂直帷幕灌浆的方 法进行垂直防渗。从而形成垂直与水平相结合的防 渗系统。 （2）导流措施 在垃圾堆埋体内设置竖向集水石笼(兼导气管)， 在底部的防渗层上设置以碎石或卵石为材料的盲 沟，把垃圾渗滤液导入收集池内。 （3）收集池 渗滤液收集池除了要具有良好的防渗性能外，其 最大设计容量要充分考虑到垃圾渗滤液的正常产量 以及在大雨、垃圾液处理设施不能正常运转等风险 因素。 （4）雨水截流措施 为了减少雨水汇入垃圾堆体而增加渗滤液的产生 量，需要在填埋区四周的边坡上挖设雨水截流沟， 把降落在填埋区以外的雨水导出填埋区。 （5）处理设施 经过集中收集的渗滤液处理方法主要有三种： ①将渗滤液输送到城市污水处理厂与城市生活污水 合并处理；②采用渗滤液回灌技术处理，其主要原 理是把渗滤液回灌到垃圾堆体上，增加蒸发量，从 而使渗滤液减量；③由填埋厂自建渗滤液处理站处 理，采用的工艺包括生物法、物化法与土地法等 [5]。 1.2.2 填埋气的疏导措施 主要是在垃圾堆体内合理布设石笼和排气井， 使填埋气体能得到有效排散。 1.3 风险管理措施 填埋区要划分为若干作业区，垃圾被倾倒入某作 业区后，要立即进行平整压实，以降低垃圾空隙 率，减少沉降。每达到隔2－5m厚度时要进行覆土 并压实，可有效减少进入垃圾层的降雨量，避免垃 圾层因雨水过多而失稳，并可减少渗滤液的产生 量。 2. 城市垃圾安全填埋场的风险因素识别 垃圾填埋场的风险类型按照不同的分类标准可 以有多种。按风险定义的内涵可以分为广义风险与 狭义风险。按风险诱发因素可以分为人为风险、设 备风险以及自然风险。按风险类别可以分为水环境 风险、大气环境风险、垃圾堆体滑动风险等。而这 些风险又存在着必然的联系，人为风险、设备风险 以及自然风险是诱因，水环境风险、大气环境风 险、垃圾堆体滑动风险是结果。人为风险与设备风 险属于事故风险范畴，而自然风险属于潜在风险范 畴。各种风险之间的联系见图1。 2.1 广义风险与狭义风险 广义风险是指凡是对生态系统或者是人体健康 造成威胁的所有污染事件，无论大小都统称为风 published by atlantis press copyright: the authors 150 险。按风险发生的机制又可以为分常规风险、事故 风险与潜在风险[6]。狭义风险是指具有一定发生概 率的，一旦发生会造成生态系统、人员伤害的，或 者造成一定经济损失的环境事件，特指广义风险中 的事故风险以及潜在风险。 2.1.1 常规风险 常规风险是指垃圾填埋场各种设施在正常运行 时，向环境排放的有害物质。例如运输与填埋机械 作业时发出的噪音，运输、倾倒、碾压、平整垃圾 过程中随风飘散的纸张、塑料、扬尘，垃圾堆体、 垃圾渗出液产生的恶臭以及甲烷气体，还有渗滤液 收集系统收集的垃圾液等。因为这些因素对环境以 及人群尚不能构成严重威胁，所以人们常把他们称 为污染因素而不是风险因素。 2.1.2 事故风险 由于设备的老化、丢失以及人为操作失误，使 垃圾填埋场设施发生故障，致使经济受到损失、环 境遭到损害、以及人员造成伤亡的事件。例如垃圾 渗滤液防渗系统、收集系统、处理系统发生损坏或 故障，造成垃圾渗滤液的外泄；可燃气体收集与导 气系统发生故障，造成可燃气体郁集以至于爆炸 等；能源供给系统发生故障造成填埋场无法正常作 业而被迫关闭等。 2.1.3 潜在风险 是指垃圾填埋场中那些环境危害而暂时还没有 条件发生的风险。它们发生的机率很小，但是一旦 发生却具有很大的破坏性以及危害性。例如地震造 成填埋区防渗层破裂使渗滤液外泄事故。 2.2 人为风险、设备风险、自然风险 2.2.1 人为风险 是指填埋场的设备在安装、使用、管理过程 中，或者在垃圾场运营过程中，工作人员违规作业 造成填埋场设施损坏、不能正常运转，从而造成垃 垃圾填埋场环境风险 水环境风险 垃圾堆体滑动风险 大气环境风险 防渗层 断裂 收集系 统故障 处理系 统故障 收集系 统堵塞 排气系 统故障 人为风险 设备风险 自然风险 安装 使用 管理 作业 地震 洪水 泥石流 雷电 事故风险 潜在风险 图 1 生活垃圾填埋场风险类型图 published by atlantis press copyright: the authors 151 圾填埋场的污染外泄与爆炸等世故。例如进行垃圾 填埋区底部防渗时，防渗膜的铺设需要有严格的操 作规程，如果不按规程进行操作，极易出现防渗膜 接合不严，或者石头等硬物的垫、硌造成防渗膜破 损等，都会引起垃圾渗滤液外漏，进而发生污染地 下水事故。如果在垃圾堆埋过程中，垃圾堆体不能 及时覆土压实，在突遇大雨的情况下，雨水就会进 入垃圾堆体转变为垃圾渗滤液，使垃圾渗滤液的产 生量迅速增加，一旦超出收集池的收集容量或者超 过污水处理设施的处理能力，就会造成渗滤液外 溢，造成污染地表水的环境风险。 2.2.2 设备风险 是指用于垃圾填埋场基础设施建设的材料、设 备等质量存在问题，或者是材料、设备年久失效， 从而使垃圾场在运营过程中，发生污染或爆炸事 故。 2.2.3 自然风险 是指自然界发生的偶然事件，使填埋场发生了 爆炸与污染事故。例如由洪水、地震、滑坡、泥石 流、自然雷击等自然灾害造成垃圾场设备毁坏或者 是无法正常运转而造成的垃圾渗滤液泄露、可燃气 体爆炸等事故。 2.3 水环境风险、大气环境风险、垃圾堆体滑动风险 2.3.1 水环境风险 是指由于人为或者是自然原因引起的垃圾渗滤 液的外泄事件，从而对周围环境构成了一定的威 胁。水环境风险的可能性包括以下两方面。 （1）防渗层破裂 地震、防渗层敷设时由于施工不规范导致基础 不均匀沉降、聚乙烯膜接合处衔接不严密、机械碾 压时造成聚乙烯膜破裂、防渗材料质量低劣都可以 造成防渗层破裂，其威胁对象主要是地下水。 （2）集水系统及水污染处理系统失效 当集水系统发生损坏，或者污水处理站因故无 法正常运行，以及突遇大雨、洪水，超出集水池的 负荷或者超出污水处理系统的处理能力使污水未经 处理而排放。其威胁的对象主要是地表水环境以及 附近的生态系统。 2.3.2 大气环境风险 当垃圾堆体内收集系统堵塞或者是排气系统设 计有缺陷时，造成以甲烷为主要成分的填埋气在垃 圾场内郁集，达到爆炸极限时，一旦遇上高温、明 火、雷击、电火花等，就会发生火灾或者是爆炸。 2.3.3 垃圾堆体沉降或滑动引起的环境风险 当垃圾堆总体高度过大，垃圾中的有机组分经 过降解，会导致垃圾堆的自压缩与沉降，由此带来 堆场的不稳定风险。一旦遇较大降雨，垃圾场产生 滑坡的几率就会增大，周边的自然以及人工生态系 统都将受到危害[2]。 3. 城市垃圾卫生填埋场风险大小及影响范围评 价方法 环境风险评价应该主要包括对风险大小的预 测、事故后污染影响范围的预测以及污染物质扩散 浓度的预测[8]。生活垃圾填埋场风险大小，以及事 故发生后能否对一定时间内其影响范围大小进行合 理估测，对环境风险应急预案的制定以及风险发生 后处理措施的有效执行可以提供重要的依据。 3.1 垃圾填埋场污染风险的度量 垃圾填埋场的风险是多种因素综合作用的结 果，应该根据其所在区域的自然条件、社会经济条 件及垃圾场的工艺条件进行综合评价。其综合评价 指数计算方法如下： esodrci  其中： ci：垃圾填埋场风险评价综合指数； r：填埋场运营过程中的工艺风险指数； d：区域自然灾害风险指数，是指区域地震、洪水 等自然灾害对填埋场风险发生的影响，根据各 自然灾害发生的概率与可能性确定； so：区域社会指数，指社会消防能力、交通情况、 管理部门的组织应急能力对垃圾填埋场环境风 险的影响。 e：环境指数，表示事故发生后，事故对环境及人群 影响的程度。 published by atlantis press copyright: the authors 152 工艺风险指数 r 是由多种因素综合决定的，其 由下式来进行计算， 5 bmstcr  其中： c：危险指数，表示填埋场固有的、潜在的污染事故 的危险程度。它决定于填埋场拥有的污染物的数 量、毒性、燃烧爆炸性等。     m j n i ijij mac 1 1 （a 为事故隐患指数，m 为权重，i 为评价项目个 数，j 为第 j 种污染因子）； t：工艺指数，是指填埋工艺过程对环境风险的影 响。    n i itt 1 （i 为有环境事故风险的工艺环 节）； s：安全指数，是指垃圾场安全等级的指标，它反映 填埋场生产装置、设备的可靠程度、安全环保管 理系统的水平。    n i iss 1 （i 为影响垃圾场安 全的设施因子）； m：安全管理指数，主要反映管理者和操作者素质 的高低，对职工的安全教育状况和日常安全管 理状况； b：殉爆指数：是指爆炸时引起一定距离外另一爆炸 的发生，在填埋场指一填埋单元发生事故后能 否引起另一单元事故的发生； 区域自然灾害风险指数 d 由各自然灾害发生概 率与可能性共同决定，其大小由下式计算， 4/)( lsleqfd  其中： f：区域洪水风险指数，是指区域洪水对填埋场风险 发生的影响，根据洪水发生的概率与可能性确 定； eq：区域地震风险指数，根据地震裂度级别与地震 发生的可能性确定； l：滑坡风险指数，填埋区周围山体滑坡对填埋场风 险发生的影响，根据滑坡发生的概率确定； ls：雷击风险指数，指雷击事件对填埋场发生爆炸 风险的影响，根据雷击事件发生的概率确定。 so、e 以及 r、d 指数计算过程中的指标、标 准应参考各种相关资料并向专家咨询而得到。 3.2 事故发生后污染物影响范围与浓度预测 垃圾场发生的污染风险中，大气事故风险主要 表现为爆炸，为瞬时性事件；而水环境污染事故风 险影响范围大、影响时间长，所以对风险事故影响 范围与浓度的预测，应主要集中在对水污染事故的 预测与评估上，包括对地下水和地表水的污染预测 与评估。 3.2.1对地下水的影响范围与污染物浓度预测 当垃圾场底部防渗层破损失效时，垃圾渗滤液 就会经过包气带进入含水层。污染物在岩土层中， 会产生对流和弥散、机械过滤、吸附和解析、化学 反应、沉淀和溶解、降解和转化等一系列复杂的物 理、化学、水文地球化学和生物作用。因此，对地 下水影响范围与浓度的预测，除了要考虑渗滤液的 化学成分、浓度、以及入渗条件之外，还要考虑包 气带的地质结构、岩土成分、厚度、饱和和非饱和 渗透性能和对污染物的吸附滞留能力等因素的影 响，因此需要根据渗滤液的特点以及当地的自然条 件，而建立复杂的预测模式。为了反应一般情况下 的普遍规律，我们需要设立一些假设条件，从而建 立起理想状态下的一般预测模型。 （1）下游某点处的污染物浓度预测 假设地下含水层为各向同性的均匀孔隙介质， 污染物在固相上吸附遵从 henry 吸附平衡关系，污 染物的降解或衰变常数为 0，即污染物是守恒的。 污染物在下游某一点处浓度 c 可按下式求得：   t dtcfc 0 1 )()(  式中： f( )： 时刻源的释放率； c1（t- ）：在（t- ）时刻，下游某点处的浓度； (2）污染物的渗透速度 对于非饱和及饱和状态流场，可以采用统一的达 西定律。 i iji x h kq    式中， iq ：渗透速度； kij：渗透系数张量； h：水头。 3.2.2 地表水的影响范围与浓度预测 published by atlantis press copyright: the authors 153 当垃圾渗滤液进入到地表水体以后，污染物的 浓度会随着时间和空间的推移而逐渐降低，即发生 了水体的自净作用，自净作用的机制可以分为物理 自净、化学自净和生物自净三类，它们往往是同时 发生而又相互影响。上述各种过程要受到受纳水体 的大小、水量、水速、水质等一系列水文特征的影 响。因此，进入到河流、湖泊、海洋等不同的水体 中，要采用不同的预测模式，下面以污染物排入均 匀恒定的河流为例，说明对地表水影响范围与浓度 预测的方法。 （1）持久性污染物的浓度预测 当持久性污染物随垃圾渗滤液排入均匀恒定的 河流后，经过混合过程段到达充分混合段时，污染 物的浓度可按河流完全混合模式求得： c＝（cpqp+chqh）/(qp+qh) 其中： ch：河流上游污染物浓度，mg/l； cp：污染物排放浓度，mg/l； qp：废水排放量，m 3/s； qh：河流流量，m 3/s。 （2）河流一维稳态混合衰减模式 当在平直河流稳态排入非持久性污染物时，污 染物的浓度可采用 s-p 模式估算： ) 86400 exp( 10 u x kcc  式中： c0：：计算初始断面污染物浓度，mg/l； u ：河流断面平均流速，m/s； k1：衰减系数； x：衰减距离。 4. 城市生活垃圾安全填埋场环境污染风险管理 措施 对垃圾安全填埋场的风险管理要贯彻全过程管 理的原则，即从垃圾场的选址直到垃圾场的封场， 均要有风险意识。在认真贯彻落实《城市生活垃圾 卫生填埋技术规范》的前提下，通过有效的管理， 提高填埋场对自然风险的抗御与应对能力，减少以 至于消除设备风险及人为风险，从而为填埋场所在 区域提供一个安全的环境。 4.1选址阶段的风险管理 填埋场所选地点首先要保证地质基础稳定，要 避开褶皱、断裂发育地点、洪泛区以及滑坡、泥石 流易发地点，而且附近要无大的地表水体与水源 地。同时要保证与居民点之间有足够的卫生防护距 离。 4.2 设计阶段的风险管理 垃圾场的设计要有备用电源、设备、机械部 件，对垃圾渗滤液收集池容量的设计要充分考虑到 处理设施故障、洪水等风险因素，从而提高垃圾场 的抗风险能力。 垃圾填埋场的设计要具有抗震性，充分考虑到 一旦遭到地震的损坏后的补救措施。 要填埋场内以及周围地区要布设地下水监测 井，定期监测地下水的成分，一旦发生防渗层破 损，做到早发现早处理。 4.3 建设施工阶段的风险管理 垃圾场建设材料的选择上要严把质量关，建设 施工要严格执行工程设计规范，尤其是建设垃圾填 埋场防渗层、导气装置、污水收集与处理设施时， 要加强建设施工的监理，以降低甚至杜绝这些环节 出现故障而出现的风险。 4.4 运营阶段的风险管理 垃圾填埋场在日常的运营过程中，要贯彻风险 意识，要按照操作规程进行日常作业，对入场垃圾 要及时进行平整、压实，当垃圾达到 2m 左右的厚 度时，要及时覆土以减少雨水的进入量，并且要根 据所在区域的自然环境状况控制垃圾堆体的填埋厚 度。在大雨、暴雨等恶劣天气条件下，填埋场要临 时关闭。 对污水处理设施要定期进行检测，对污水浓 度、填埋气体的浓度要定期监测，发现异常要寻找 问题根源及时进行处理。 4.5 封场阶段的风险管理 封场后要及时在垃圾堆体表面覆盖防渗层，并 覆土进行植被恢复。应继续处理填埋场产生的渗滤 液和填埋气，并定期进行监测。 4.6 风险意识的培养与应急能力的训练 针对垃圾场职工和公众广泛开展风险宣传和教 育, 组织安全培训与演习，提高员工和公众的风险意 识和应急能力。 制定包括监测、报警及污水事故排放等情景组 合下的应急预案并定期演习, 减少突发环境事件带来 的灾难性后果。 published by atlantis press copyright: the authors 154 参考文献 [1] 朱东升,黄佺,胡海兰,环境风险评价的进展,职业与健康. 25(14) (2009) 1537-1539. [2] 韩东升,肖国芽,谢雯, 城市垃圾填埋场环境风险评价浅析, 污染防治技术. 36(10) (2007) 20-23. [3] 段小丽,王宗爽,于云江等,垃圾填埋场地下水污染对居民 健康的风险评价,环境监测管理与技术. 20(3) (2008)20-24. [4] 王炜,张小梅,刘茂,垃圾填埋场火灾爆炸风险分析,环境卫 生工程. 13(5) (2005) 41-44,47. [5] 龙腾锐,易洁,林于廉等,垃圾渗滤液处理难点及其对策研 究,土木建筑与环境工程. 31(1) (2009)114-119. [6] 曹希寿,区域环境风险评价与管理初探,中国环境科学. 14(6) (1994)465-470. [7] 宋国君,马中,陈婧等.论环境风险及其管理制度建设，环 境污染与防治. 28(2) （2006）100-103. [8] 徐惠民,区域环境风险综合评价理论与方法探讨,东北师 范大学硕士学位论文.1995. published by atlantis press copyright: the authors 155 microsoft word manuscript_the assessment of risk caused by fire and explosion_ a domino effect-based study the assessment of risk caused by fire and explosion in chemical process industry: a domino effect-based study farid kadri univ. lille nord de france, f-59000 lille, france uvhc, tempo lab., "production, service, information" team, f-59313 valenciennes, france farid.kadri@univ-valenciennes.fr eric chatelet, patrick lallement university of technology of troyes (utt), france laboratory of systems modeling and dependability umr stmr/icd cnrs n°6279 {eric.chatelet, partrick.lallement}@utt.fr abstract in the field of risks analysis, the domino effect has been documented in technical literature since 1947. the accidents caused by the domino effect are the most destructive accidents related to industrial plants. fire and explosion are among the most frequent primary accidents for a domino effect due to the units under pressure and the storage of flammable and dangerous substances. heat radiation and overpressure are one of major factors leading to domino effect on industrial sites and storage areas. in this paper we present a method for risk assessment of domino effects caused by heat radiation and overpressure on industrial sites. this methodology is based on the probabilistic models and the physical equations. it allows quantifying the effect of the escalation vectors (physical effects) in industrial plants, the three areas defined in this study may be useful in the choice of safe distances between industrial equipments. the results have proven the importance of domino effect assessment in the framework of risk analysis. keywords: domino effect; quantitative risk assessment; explosions; fires; storage areas. 1. introduction the accidents caused by the domino effect are those that cause the most catastrophic consequences. the consequences of these latter are at various levels and may affect not only the industrial plants, but also people, environment and economy. the probability of domino effect is increasingly high due to development in industrial plants, the proximity of such establishments and their inventories of dangerous substances. the potential risk of domino effect is widely recognized in the legislation since the first "seveso-i" directive (82/501/eec), which required the assessment of domino effects in the safety analysis of industrial sites whose activities are subject to this directive. furthermore, the "seveso-ii" (96/82/ec) extended these requirements to the assessment of domino effects not only within the site under consideration, but also to nearby plants [1]. an inventory of the past domino accidents [2], reveals that explosion are the most frequent cause of domino effect (57%), followed by fires (43%). a study of 225 accidents involving domino effects [3], shows that storage areas are the most probable starters of a domino effect (35%), followed by process plant (28%). also, the journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 66-76 published by atlantis press copyright: the authors 66 willieb typewritten text willieb typewritten text received 7 november 2012 willieb typewritten text accepted 19 february 2013 willieb typewritten text farid kadri, eric chatelet and patrick lallement most frequent accident sequences are explosion-fire (27.6%), fire-explosion (27.5%) and fire-fire (18%). to address the problem posed by the assessment and/or analysis of domino effects in industrial sites, several methods and software tools have been developed [4, 5, 6, 7]. an analytic methodology for the quantitative assessment of industrial risk due to accidents triggered by seismic events has been developed [8]. this procedure is based on the use of available data (historical data) to assess the expected frequencies and magnitude of seismic events. a method for assessing domino effects based on monte carlo simulation has been developed by [9]., the authors developed an algorithm, which is based on conducting several hypothetical experiments to simulate the actual behavior of a multi-unit system. recently, a review of methodologies and software tools used in the literature to the study of the cascading events [10], shows that, in the last decade, the available methodologies for the assessment of domino effects caused by heat load and overpressure to process equipments are based on the probit models [11, 12, 13]. the objective of this article is to present a methodology for the quantitative assessment of domino effects caused by heat radiation and overpressure to industrial/chemical plants and storage sites. next-subsection is dedicated to a brief definition of the domino effect and its main features, potential sources of domino effects and the propagation process. next, brief analysis of previous works is presented. in the third section we present a methodology for quantitative assessment of domino accidents in industrial sites. the fourth section uses a case study to illustrate the proposed model and to present typical results. the last section concluded this paper. 1.1. domino effect and escalation there is no generally accepted definition of what constitutes domino effects in the context of accidents in the industrial plants, although various authors have provided suggestions [14, 15, 16, 17]. a domino accidental event may be considered as an accident in which a primary event propagates to nearby equipment (units), triggering one or more secondary events resulting in overall consequences more severe than those of the primary event [18]. the concept of escalation is a process that promotes the degradation of property (materials, equipments, systems industrials, ecosystems) and injury to people during development of the domino effect (increase damages). thus, in the industrial field, we consider that any event spreading from equipment and/or industrial unit to another or from one site to another site should be classified as a domino event. according to the case histories concerning past domino accidents, all the accidental sequences where a relevant domino effect has took place have three common features [19]:  a primary accidental scenario, which initiates the domino accidental sequence;  the propagation of the primary event, due to an escalation vectors, generated by the physical effects of the primary scenario, that results in the damage of at least one secondary target;  one or more than one secondary accidental scenarios or events, involving the same or different plant units causing the propagation of the primary event. 1.2. potential sources of domino effects potential sources of domino effects are of different nature and are also linked to various initiating events. in general, they are distinguished by the nature of risks, from natural or anthropogenic. in the latter category, there are technological and organizational risks (unintentional) and the risks of malevolence (intentional), knowing that the purpose of study of domino effects takes into account the combination of these two risks. it is therefore possible to propose the decomposition of the nature of risks as follows: a) natural origins (geological origins and/or atmospheric mainly) [20, 21]:  climate origin: forest fires, runoff and floods, avalanches, hurricanes and tornadoes, storms;  geological origin: landslides and earthquakes, tsunamis, volcanic eruptions and other natural emissions (gas, etc.). published by atlantis press copyright: the authors 67 the assessment of risk fire and explosion in chemical process industry b) human origins (organizational and malevolence) [22, 23, 24]:  organizational origin: humans failures (incorrect human action, lack of human action), defects in design, procedures and/or organizational;  malevolence origin, thefts, sabotage and/or revenge action, damage of any kind attacks. these actions may touch or affect the material, but also the personal or sensitive information. c) technological origin (fire, explosion and toxic releases):  fire: pool fire, flash fire, fireball and jet fire;  explosion: confined vapor cloud explosions (cvce), boiling liquid expanding vapor explosion (bleve), vented explosion, vapor cloud explosion (vce), dust explosion and mechanical explosion;  toxic chemicals release: from process or storage sites and transportation accidents. these risks can be combined which significantly complicates the analysis. sometimes, the very different nature of risks involves varied propagation processes. this also leads to the exploitation of different analysis methods (deterministic, probabilistic and quantitative methods). 1.3. the propagation process the propagation process is directly related to the potential source and the initiating event, but also to its immediate environment (field of danger). it is described by a physical-chemical process, but also informational whose evolution conditions are guided by features such as: physical (atmospheric, geological, hydrological) and material (buildings, sites, facilities, roads,...), ecological (vegetation, animals), informational (detections, observations and information systems) and human (individual behavior, organization and logistics, local demography). for more detailed about the propagation of danger from potential source to a potential target and the concepts of "source" and "target" and systemic approaches, it is advisable to refer to references [25, 26, 27]. 2. domino effect analysis in the framework of domino effect analysis, the risk of explosion and fire, characterized by the possibility of an accident in an industrial site may lead to damage and serious consequences for the surrounding process equipment, people, goods and environment. these latter can generate four main events that may affect and/or cause the failure of the surrounding process equipments/units:  overpressure/blast waves;  heat load;  projection of fragments (missiles);  toxic release. although several studies were dedicated to the assessment of domino effect caused by fires and explosions, only few models based on very simplistic assumptions are available for the assessment of equipment damage caused by heat load and overpressure in the framework of domino effect. the more simple approach proposed for the assessment of damage to equipment caused by fires and explosions. several authors propose to consider zero probability of damage to equipment if the physical effect is lower than a threshold value for damage, and to assume a probability value of one if the physical effect is higher than a threshold value for damage [28,29, 30, 31]. a quantitative study, however, of the domino effect has been made by [32]. they have described possible approaches for quantifying the consequences of domino effects resulting from events giving rise to thermal radiation. a first approach evaluating the frequency accidental explosions was proposed by [33]. they provided a methodology for predicting domino effects from pressure equipment fragmentation. a simplified model proposed by [34].allows to assess the damage probability of process equipment caused by blast wave. the model is based on "experimental" evaluation of equipment displacement with the subsequent deformation and breakage of connections. the author defines the "probit function" (y) relating equipment damage to the peak static overpressure (p^0) as follow: ln (1) where y is probit function for equipment damage, p^0is peak static overpressure (pa), a and b are the probit coefficients. published by atlantis press copyright: the authors 68 farid kadri, eric chatelet and patrick lallement the probit approach has been followed by [18, 35, 36], the authors have been published articles in which they analyzed and reviewed the existing models to develop a probabilistic model for damage evaluation of specific categories of industrial equipments. the damage probability model proposed by the authors takes into account four categories of industrial equipments (atmospheric vessels, pressurized vessels, elongated vessels, and small equipments). the probit coefficients and thresholds for overpressure damage probabilities for four equipment categories are represented in the table 1. table 1. probit coefficients for different equipment categories [36] equipment category a b threshold atmospheric vessels -18.96 +2.44 22 kpa pressurized vessels -42.44 +4.33 16 kpa elongated equipments -28.07 +3.16 31 kpa small equipments -17.79 +2.18 37 kpa to estimate the time to failure ttf of industrial equipments exposed to fire. a well known simplified model proposed by [37] is based on the probit approach. the authors proposed damage probability models that take into account the categories of industrial equipments. table 2 presents the thresholds and probit models for two equipment categories. a methodology for domino effect analysis has been developed by [4] and, some applications in [38, 39]. the authors have cited that the intensity of heat radiation of 37 kw/m2 is sufficient to cause severe damage to process equipment in other installations that operate under atmospheric conditions. also, a peak overpressure of 70 kpa is enough to cause severe damage to process equipment and may generate new accidents, either associated to new explosions or new events involving fires. a systematic procedure for the quantitative assessment of the risk caused by domino effect to industrial plants has been developed by [19]. this methodology aims to calculate the propagation probability of primary scenarios, the expected frequencies of domino events, and allowed to estimate the contribution of domino scenarios to individuals as well as societal risk. on industrial sites/storage areas, the heat load and overpressure generated by bleve explosions of tanks containing gas or highly pressurized liquids are threats to other surrounding equipment and can lead to successive explosions and fires. several studies have been done on modeling the impact of bleve explosions on industrial installations [40, 41, 42, 43, 44]. boiling liquid expanding vapor explosions (bleves) are among the diverse major accidents which can occur in process industries. it is usually associated with the explosion of tanks containing flammable liquids (lpg). therefore, to the effects of the bleve, one must add those corresponding to the fireball often occurring immediately after the explosion. on the whole, then, the physical effects from this type of explosion are usually i) thermal radiation, ii) overpressure (blast) and ii) fragments projection. the bleves mechanism, the causes and consequences are presented by [45, 46]. different formulas are used to quantify the heat radiation generated by fire. the radiation from fireball or pool fire on a receptor body located at a distance r from the center of this latter may be expressed by the following equation [47]: table 2. probability models and threshold values for the heat radiation, y is the probit function, ttf is the time to failure (sec), v is the vessel volume (m3), and i is the amount of heat radiation received by the target vessel ( / ) [36] equipment category threshold correlation atmospheric vessels 15 / 12.54 1.847 10 1.128 2.667 10 9.887 pressurized vessels 50 / 12.54 1.847 10 0.947 8.835 . published by atlantis press copyright: the authors 69 the assessment of risk fire and explosion in chemical process industry i r (2) where i(r) is the heat radiation flow (kw/m ), f is the fraction of the generated heat radiated from the flame surface, m is the combustion velocity per unit surface area of the pool [ kg/ m .s)], τ , is the atmospheric transmissivity coefficient, h is a combustion heat (kj/kg), d is the pool diameter. in experiments with explosives framework, the equivalent mass of tnt (m ) was used to evaluate the effects of potential damage of a quantity of fuel (hydrocarbon) given. the combustion energy available in a cloud of steam was converted into an equivalent mass of tnt (kg). m may be evaluated assuming that an exploding fuel mass behaves like exploding tnt on equivalent energy basis. hence, the equivalent mass of tnt is estimated by using the following equation [48]: m ∆ (3) where μ is the explosion efficiency (0.03 to 0.1), m is the mass of fuel involved in the explosion (kg), ∆h is the energy of explosion of the flammable gas (energy/mass) (mj/kg), e is the energy of explosion of tnt (mj/kg). in an explosion, the peak overpressure may be estimated using the following equation: p r . . . . . . (4) where is the peak of overpressure (kpa), and is atmospheric pressure (101.3 kpa), is a scaled distance ( / ) which may be estimated using an equivalent mass tnt ( ) as follow: (5) where r is distance from the center of the explosion. note that, can be calculated by setting the threshold of peak of overpressure for each equipment categories. 3. methodology an industrial site and storage areas contains many storage equipments/units under pressure that may be subjected to an external and/or internal incident. the escalation vectors (physical affects) generated after a unit rupture (explosion), may affect the surrounding units, building, personnel and environment. if the affected targets are damaged, these latter, may also explode and generate another threats to other surrounding facilities and so on. this accident chain is a domino effect and may lead to catastrophic consequences in an industrial plant. 3.1 domino system we define a domino system as a system which consists at least of two subsystems ( , ), a source subsystem and a target subsystem (see fig. 1):  a source subsystem: its failure may generate a danger (physical effects) that may affect other surrounding subsystems (heat load, overpressure, fragments, toxic releases), and  a target subsystem: it may be affected by the failure of sub-system sources. in addition to these physical effects, we may include the influencing factors that can influence or aggravate the target system damage (malicious acts, human and organizational factors, intervention system and weather conditions). in the case of domino effect analysis, the failure of a subsystem depends on the dynamic characteristics of the escalation vectors (input vector), threshold values and the aforementioned influence factors. then, the domino system can be described by the following vector function: , , (6)  , …, : is a real vector (input vector) with p dimension in a space of physical state at time t. may be divided into two types of parameters, random physical parameters (physical effects) and influence factors (intervention system and human factor);  , …, : is a real vector (input vector) with g dimension, represents the deterministic input parameters of the system (physical characteristics of system like thresholds); published by atlantis press copyright: the authors 70 farid kadri, eri  outp depe 3.2 represen an industria ( , ,…, analysis, each three main sta  stat corre syste resp affec by a  stat esca corre affec  stat vect thres to study the as starting p initiating eve least one faile transitions sta ic chatelet and p , …, put with k dim ending on inpu ntation of syste al system com ). in the f h unit can be c ates: te 1: in norma esponding to em are less ectively. in cted by the e a primary even te 2: while th alation vecto esponding th cted, and te 3: while or(s) is gre shold value. domino accid point, the fai ent. based on ed subsystem. ates in the cas fig. 1. dofig. 1. do patrick lallement : is the v mensions, is ut parameters em states mposed of sev framework of characterized 1, 2, 3, al operation, t the input pa than the t this state, th escalation vec nt, he intensity o or(s) is hreshold value the value o ater than it dental sequen ilure of at le n the assump . figure 2 pres e of two subsy omino system omino system ector of syst random varia . veral subsyste f domino ef by the follow the output val arameters of threshold val he unit may ctor(s) genera or the value(s) equal to e, the unit s of the escalat s correspond nce, one can t east one unit ption, there is sents the possi ystems/units. tem able ems ffect wing lues the lues be ated ) of its says tion ding take t as s at ible 3.3 in to thr out cor fai of wh sys for aft pro cal th vec cal 3.4 wh sub dam wh bet the dam eac 3 failure prob normal opera the input par reshold values tput of syst rresponding t led. then, the the system ma here is the stem) and is r which , fter calculatin obability lculated by the he total failur ctors that aff lculated with t 4 domino effe hile failure bsystem, the mage radii (a hole system. tween a minim e impacted zo mage level is ch impacted z f bability ation, the outp rameters of th s respectiv tem for an en threshold val e failure funct ay define as fo e threshold c s the output of 0 , then ng the failu for each es e following eq re probability fects the targe the following ⋃ ect probability probability probability o ffected zones domino effe mum of two z ones are pres increased in one. fig. 2. transiti put values c he system are vely. while the ntry point is g lue , the tion that desc ollows: criterion (defi f target system the system sa ure function, scalation vec quation: , 0 y for all t et subsystem equation: 0 y/affected zon is know of domino e ) may be eva ect consists ones. the dam ented in the involved area ion between sta corresponding e less than the e value of any greater than its system says cribes the state (7) fined for each m. if it exists ays failed. , the failure ctor may be (8) the escalation ( ) may be (9) nes wn for each effect and the aluated for the in interaction mage radii and figure 3. the as, but also on ates g e y s s e ) h i e e ) n e ) h e e n d e n published by atlantis press copyright: the authors 71 according to affected zone equipment lo probability this zone the and iii) saf process equip the probabil accidental seq where n is involving in probability th 4. the case-s the above d study in orde storage area. this case stud inventory are o the figure 3 es i) zone of c ocated in this 1, ii) zo failure probab fety zone; th pment 0 lity of each do quence) may b the number the domino hat each unit fr study efined metho er to assess figure 4 sho dy. the type e shown in the fig.3. the a 3, we can de certain destruc s area are fai one of possibl bility is betwe he failure pro 0. omino scenario be calculated ∏ r of the fai o sequence, from sequence dology was u domino effec ows the lay-o of equipmen table 3 bellow affected zones efine three m ction; all proc iled with fail le destruction een 0 obability of o (domino as follows: ( led sub-syste is the jo e i fails. used in the ca ct in the case out considered ts/units and th w. the a main cess lure n; in 1, the (10) ems oint asee of d in heir tab tank tk1 tk2-7 4.1 we rup gen ove aff som the bee cas in we pro exp can des saf assessment of ris ble 3. equipme k typ pressurize 7 atmospher 1 effects on su e assume tha pture (catastro nerate three e erpressure wa fect the surrou me simplifica e effects of he en considered se are tabulate the case-stud ere considere obability in fu plosion of the n define thre struction, ii) fety zone. fig sk fire and explo ent considered i e sub ed tank l ric tank eth urrounding e at a primary ophic failure) escalation vec ave and iii) f unding equipm ations are used eat radiation a d. the influen ed in the table dy, only prim ed. the figu unction of the tk1 in case ee types of zone of pos g.4. lay-out use osion in chemical in the case-stud bstance cont lpg 1 hanol 3 quipments y scenario ha of one tank. ctors; i) heat fragments, the ments. d in the prese and overpress nce parameter 4. mary and seco ure 5 shows distance resu of overpressu zone: i) zon ssible destruc ed for the case s l process industry dy tent (t) failu 150 9 315 as caused the the latter can t radiation, ii) ese latter may ent study, only sure wave has rs used in this ondary events s the failure ulting from the ure effects. we ne of certain ction, and iii) study ry ure frequency 9 10 10 e n ) y y s s s e e e n ) published by atlantis press copyright: the authors 72 fa farid kadri, eri we remark th the area limit failure proba overpressure failure proba radiation. the failure p escalation v waves) are r figure 6 pres catastrophic f heat radiation (zone of pos probability table 5. proba ailed tank es tk1 h tk1 h tk1 o tk1 o tk2 h tk2 h tk2 o tk2 o fig. 5. failure taic chatelet and p hat the whole ted by the radi ability waves, and ability probability, ectors (heat represented in sents the affe failure (ruptur n, z1 (zone o sible destruct 0.9 10 ability due t over scalation vector heat radiation heat radiation overpressure overpressure heat radiation heat radiation overpressure overpressure e probability res the case able 4. the infl random : e m: mass : explo : atmo : frac d: pool d patrick lallement process equip ius of 132 m 9.4 10 the radius o 10 in th due to the e radiation an n the followi ected zones g re) of the tan of certain des tion) estimate 0 respective to the effects of rpressure target tank tk2-6 tk4 tk2-6 tk4 tk2-5-6 tk3-7 tk2-5-6 tk3-7 sulting from the of overpressur uence paramete m parameters explosion energ s involved in th osion efficiency ospheric transm ction of the ge diameter pment that are have failed w in the case of 420 m w he case of h effects of the t nd overpress ing table 5. t generated by ks in the case struction) and ed for the fail ly. f heat radiation failure proba 1.21 10 1.32 10 7.51 10 2.2 10 2.84 10 1.75 10 9.46 10 8.91 10 e rupture of the e ers used in the c gy he explosion y missivity enerated heat e in with of with heat two sure the the e of d z2 lure and ability 15 4.2 to tha wa dom tab tab scen 1 2 3 4 fig. 6 tk1 in case of heat rad probabilis ~ 50 2. domino eff o estimate the at the two ev aves) are ind mino sequen bulated in the t ble 6. the prob arios domin 1 t 2 t 3 t 4 t 6. affected zo diation and over stic distributio ~ ~ 0. ~ ~ ~ 2 fect scenarios e domino eff vents (heat r dependents. ce, for table 6. ability for each no effect sequenc tk1-tk4-tk5 tk4-tk6-tk7 tk4-tk6-tk7 tk5-tk4-tk1 ones z1 and z rpressure waves on 4.50,0.15 80,0.04 0.65,0.18 0.20,0.80 0.26,0.08 2 ,0.26 fect sequences radiation and the probabi each domino h considered dom ce failure pr 1.17 1.08 8.75 2.02 z2 in the case o s, r is spherical 2 s, we assume d overpressure ility of each o scenario is mino scenario obability ( ) 7 10 8 10 5 10 2 10 of heat radiati l tank rayon. e e h s ion. published by atlantis press copyright: the authors 73 the assessment of risk fire and explosion in chemical process industry 5. conclusion a quantitative method for the assessment of domino effects in industrial sites has been developed in this paper. it allows quantifying the effect of heat load and over pressuring waves in industrial plants and/or storage areas. based on this method, we can evaluate the failure probability for each subsystem (unit), after the probability of domino scenario (domino sequence) may be evaluated for all the system. the three areas defined in this study (zone of certain destruction, zone of possible destruction, and safety zone) may be useful in the choice of safe distances between industrial equipments. domino effect caused by fragments is not studied in this paper. however, the projectiles generated by an explosion of a tank (unit) containing gas or highly pressurized liquids are threats to other surrounding equipment and can lead to successive explosions and a chain of accidents. hence, domino effect caused by fragments must be considered to evaluate the total failure probability for each equipment resulting from the combination of these events (heat load, overpressure and fragments). also, heat radiation and overpressure effects can affect not only the industrial equipments but also environment and people. so, a human vulnerability models to the heat radiation and overpressure effects should be developed to estimate the individual and societal risk. the analysis above shows the importance of domino effect assessment in the framework of risk analysis. hence, it shows that must much more importance be attached to the study of this phenomenon. finally, domino effects need more scientific investigations, particularly in terms of quantitative assessment of risks and damage with probabilistic and deterministic modeling. acknowledgements this work has been developed in the disc project and supported by the anr (agence nationale de la recherche) from the french research ministry (http://www.agence-nationale-recherche.fr). references 1. council directive 96/82/ec on the control of majoraccident hazards involving dangerous substances. official journal of the european communities, january, 1997. 2. abdolhamidzadeh b., abbasi t., rashtchian d., and abbasi s.a., 2011, domino effect in process-industry an inventory of past events and identification of some patterns. journal of loss prevention in the process industries, 24(5), 575–593. 3. darbra r.m., palacios a., and casal j., 2010, domino effect in chemical accidents: main features and accident sequences. journal of hazardous materials, pages 556573. 4. khan f.i., and abbasi s.a., 1998, models for domino effect analysis in chemical process industries. process safety progress-aiche, 17(2):107-113. 5. tixier j., rault-doumax j., dandrieux a., dimbour j.p., and dusserre g., 2002, geosiris : domino effects software. gosiris: logiciel de gestion des effets domino. in 13-esrel, 1-5. 6. lee, j.y, lee jw., ko, j., and en sup yoon, es., 2005. optimization for allocating the explosive facilities in order to minimize the domino effect using nonlinear programming. korean journal of chemical engineering, 22(5):649-656. 7. reniers g. l.l., and dullaert w., 2007, domprevplanning: user-friendly software for planning domino effects prevention. safety science, 45(10):1060-1081. 8. antonioni, g., spadoni g., and cozzani v., 2007, a methodology for the quantitative risk assessment of major accidents triggered by seismic events. journal of hazardous materials, 147(1-2):48-59. 9. abdolhamidzadeh b., abbasi, t., rashtchian, d., and abbasi, s.a., 2010, a new method for assessing domino effect in chemical process industry. journal of hazardous materials, 182(1-3):416-426. 10. kadri f., châtelet e., and elegbede c., 2013, domino effect analysis and assessment of industrial sites: a review of methodologies and software tools, international journal of computers & technology (accepted, 2013). 11. cozzani, v., and salzano e., 2004, the quantitative assessment of domino effects caused by overpressure part i. probit models. journal of hazardous materials, a107:67-80. 12. mingguang, z., and juncheng j., 2008, an improved probit method for assessment 13. of domino effect to chemical process equipment caused by overpressure. journal of hazardous materials, (158), 208-286. 14. landucci g., gubinelli, g., antonioni, g., and cozzani v., 2009, the assessment of the damage probability of storage tanks in domino events triggered by fire. accident analysis and prevention, 41(6):1206-1215. 15. delvosalle, ch., 1996, domino effects phenomena: definition, overview and classification, european seminar on domino effects. leuven, belgium, federal ministry of employment, safety administration, published by atlantis press copyright: the authors 74 farid kadri, eric chatelet and patrick lallement direction chemical risks, brussels, belgium, pages 515. 16. gledhill j. and lines, i., 1998, development of methods to assess the significance of domino effects from major hazard sites, cr report 183, health and safety executive. 17. khan fii., and abbasi s.a., 1999, major accidents in process industries and an analysis of causes and consequences. journal of loss prevention in the process industries, 12(5):361-378. 18. ccps (centre for chemical process safety), 2000, guidelines for chemical process quantitative analysis, second edition, american institute of chemical engineers, new york. 19. cozzani,v., and salzano e., 2004, threshold values for domino effects caused by blast wave interaction with process equipment. journal of loss prevention in the process industries,(17)437447. 20. cozzani v., et al., 2005, the assessment of risk caused by domino effect in quantitative area risk analysis. journal of hazardous materials, 127:14-30. 21. [20] krausmann, e., and mushtaq f., 2008, a qualitative natech damage scale for the impact of floods on selected industrial facilities. natural hazards, 46(2):179-197. 22. steinberg l.j., sengul h., and cruz, a.m., 2008, natech risk and management: an assessment of the state of the art. natural hazards, 46(2):143-152. 23. dimattia d.g., khan f.l., and amyotte p.r., 2005, determination of human error probabilities for offshore platform musters. journal of loss prevention in the process industries, 18:488-501. 24. mohaghegh, z., and mosleh, a., 2009, incorporating organizational factors into probabilistic risk assessment of complex socio-technical systems: principles and theoretical foundations. safety science, 47:1139-1158. 25. piwowar j., châtelet e., and laclémence p., 2009, an efficient process to reduce infrastructure vulnerabilities facing malevolence. reliability engineering and system safety, 94:1869-1877. 26. kervern g.y., 1995, eléments fondamentaux des cindyniques, editions economica, paris 27. perilhon p., 2000, mosar, techniques de l'ingénieur, traité sécurité et gestion des risques. se 4 060, pages 116. 28. le moigne j. l., 1994, théorie du système général, théorie de la modélisation. ed. puf, paris (4e édition). 29. lees f.p., and ang m. l., 1989, safety cases within the control of industrial major accident hazards (cimah) regulations. butterworths scientific in london. 30. bagster d.f., and pitblado r.m., 1991, the estimation of domino incident frequenciesan approach. trans icheme, 69:195-199. 31. purdy g., pitblado r.m., and bagster d.f., 1992, tank fire escalation modeling and mitigation. in internaional symposium on loss prevention and safety promotion in the process industries. 32. pettitt g. n., 1993, evaluating the probability of major hazardous incidents as a result of escalation events. journal of loss prevention process industries, 6(1):3746. 33. lath p., gautam g., and raghavan k. v., 1992, strategies for quantification of thermally initiated cascade effects. journal of loss prevention process industries, 5(1):15-21. 34. [33] scilly n. f., and crowther j. h., 1992, methodologies for predicting domino effects from pressure vessel fragmentation. in international conference on hazard identification and risk analysis, human factors and human reliability in process safety, florida, ccps, aiche, 15-17. 35. eisenberg n. a., lynch c. j., and breeding r.j., 1975, vulnerability model: a simulation system for assessing damage resulting from marine spills, report cg-d-136-75, enviro control inc., rockville, md. 36. cozzani, v., and salzano, e., 2004, the quantitative assessment of domino effect caused by overpressure part ii, case studies. journal of hazardous materials, a107:81-94. 37. cozzani, v., gubinelli g., and salzano, e., 2006, escalation thresholds in the assessment of domino accidental events. journal of hazardous materials, 129:1-21. 38. cozzani, v., antonioni g., and spadoni g., 2006, quantitative assessment of domino scenarios by a gisbased software tool. journal of loss prevention in the process industries, 19(5):463-477. 39. khan f.i., and abbasi s.a., 2001, an assessment of the likelihood of occurrence, and the damage potential of domino effect (chain of accidents) in a typical cluster of industries. journal of loss prevention in the process industries, 14(4):283-306. 40. khan f.i, and abbasi s.a., estimation of probabilities and likely consequences of a chain of accidents (domino effect) in manali industrial complex. journal of cleaner production, 9(6):493-508. 41. baum m. r., 1988, disruptive failure of pressure vessels: preliminary design guidelines for fragment velocity and the extent of the hazard zone. journal of pressure vessel technology, 110(2):168-176. 42. birk a.m., 1996, hazards from propane bleves: an update and proposal for emergency responders. journal of loss prevention in the process industries, 9:173-181. 43. venart j.e.s., 2000, boiling liquid expanding vapor explosions (bleve); possible failure mechanisms and their consequences. in: proceedings of the icheme symposium series, 147:121-137. 44. planas-cuchi, e., salla j.m., and casal j., 2004, calculating overpressure from bleve explosions. journal of loss prevention in the process industries, 17(6):431-436. 45. ogle r.a., ramirez j.c.,, and smyth, s.a., 2012, calculating the explosion energy of a boiling liquid expanding vapor explosion using exergy analysis. process safety progress, 31(1):51-54. 46. tellez c., and pena j.a., 2002, boiling-liquid expanding vapor explosion (bleve): an introduction to consequence and vulnerability analysis. chem. eng.educ, 36:206-211. 47. abbasi, t., and abbasi s.a., 2007, the boiling liquid expanding vapor explosion (bleve): mechanism, published by atlantis press copyright: the authors 75 the assessment of risk fire and explosion in chemical process industry consequence assessment, management. journal of hazardous materials, 141(3):489-519. 48. van den bosh c. j. h. and weterings r. a. p. m., 1996, methods for the calculation of physical effects due to releases of hazardous materials (liquids and gases) yellow book. the hague. 49. daniel a.c., and louvard j.f., 2002, chemical process safety: fundamentals with applications. prentice hall international series in the physical and chemical engineering sciences, 2002. published by atlantis press copyright: the authors 76 microsoft word a_risk_assessment_model_based_on_attribute_theory a risk assessment model based on attribute theory jiali feng information engineering college, shanghai maritime university,1550,address shanghai, 201306, china tao sheng information engineering college, shanghai maritime university,1550,address shanghai, 201306, china e-mail: michaelsheng88@sina.com www.shmtu.edu.cn abstract since the global financial crisis in 2008, the chinese government has invested 4000 billion to maintain the economic growth. and the investment made the inflation became more seriously. however, the investment has led to high inflation rate. to fight inflation, it is important to understand the effects of changes in interest rates, which is itself a complicate issue. this article investigated the roles and the interrelationship of five main parties that involved in the issue using attribute theory method. these 5 main parts are banks, government, enterprises, domestic capital and international capital. each of the 5 main parties was considered to consist of 5 attributes, including agriculture, construction, service, manufacturing and mining. in principle, these 5 main parties each directly play a role in the changes of interest rate, and, at the same time, affect each other. here we used the attribute theory method to investigate how the five parties affect the interest rate and their impacts on each other. as a result, we built a mathematical model to provide guidance on determining the interest rates to properly deal with various economical issues. keywords: qualitative theory; interest rate; economy 1. current economic situation since the global financial crisis in 2008, the global economy has experienced serious depression. in china, during a considerably long time, the central bank has issued too much currency to the market that becomes one of the main problems in china economy and leads to serious inflation. especially due to the price increase in staple commodities, the input inflation worsens the chinese inflation. in july 2012, the cpi consumer price index rate reached 6.5%, and kept running high until now. the high price of commodities lowers the living standard of chinese people. hence, the government has to adjust the price of commodities to release the pressure of inflation. national central bank, for example, the federal reserve (fed) can influence the inflation through adjusting interest rate as well as other currency policy. raising the interest rate, as a typical method to control the inflation, will also leads to high unemployment rate and lowers production. recently, the china, government tried to handle the inflation issue by increasing the reserve requirement and interest rate to withdraw the extra currency and slow down its negotiability. however, we have to realist inflation is not the only issue in chinese economy. as a consequence of the financial crisis, chinese government has invested 4000 billion in total to protect the local economies and markets. till now, this policy combines with other local government investment has reach 15000 billion yuan. in 2011, there is big demand to repay capital with interest. rising interest rate will make the government facing higher pressure to repay the capital with interest. as a result, the government would encounter the risk of serious debt crisis, which has happened to greece government. on the other hand, continuous rise of the interest rate will stop the enterprise from financing. in the worst case, this will result in bankruptcy of a great number of medium-and small sized enterprises. this in turn will cause a reduction in government tax income. . the data released by the state administration of taxation (sat) in 2010 revealed journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 22-26 published by atlantis press copyright: the authors 22 a risk assessment model based on attribute theory that the national tax revenue reached 7739 billion yuan, constituting 89% of the overall government revenue. therefore, mass bankrupt of enterprises will seriously impact government income. 2. methods and attribute theory 2.1. the partial order category and topes of attribute set based on information received by the human brain, and can only feel the feeling of the property to respond to the fact that we made the human brain is deciphering the properties of things, and encode the information for the brain, so as to get the construction of thinking, the basic ideas and the formation of intelligence. we concluded that decipher the properties of object is the basic way for the brain to construct mechanism. at the same time we study the basic structure of the property set of things. finally, we made a conclusion that: the attribute set of any object with level order between both attributes constructs a partial order category, and it with the conjuction becomes a attribute inference lattice category and a monoid category. they not only constitute the corresponding topos respectively, but also there is representation model of an attribute baracentric coordinate system k(m)(u)for for them. it is not only proved the mathematical description and discussion tool for the reasoning and the generation attribute based on conjuction between both attributes, but also using the functor categories, topos and attribute coordinate system k(m)(u). example: let x(t)∈x be the borrower’s income, y(t)∈y be the loans he need to repaid, f is the function that describe the relationship between x and y, f:x→y. we will get y’∈y and y’(t)=f’(x’(t)) to all x’∈x. let x’(t)∈x’ be the loans that can be repossessed by the bank, y’(t)∈y’ be the bank’s risk, f is the function that describe the relationship between the x’(t) and y’(t), f’:x’→y’, we will get y’∈y’ and y’(t)=f’(x’(t)).1 with the borrower and commercial bank as an example, as shown in fig.1 and the connection of the five deep attribute can be seen in fig.2. )(cmk f )(bmk a4(t-1, c) g a4(t,b) a3(t-1,c) e0(t,b)) a3(t,b) e0(t-1,c) f states(t-1,c) states(t, b) g a1(t-1,c) a2(t-1,c) a1(t,b) a2(t,b) fig.1. the attribute category between the borrower and the bank’s loans ,the model that show the functor f and g (note: is the borrower’s attribute category, f is the functor between them). 2.2. a attribute topos consist of the category of functors func(l(au,≤), s) we can see from the category theory that to any category c, it will be a topos by the dual category cop and the functor category func(cop, s) of the set category s. l(au,≤) is the dual category of the l(au,≥). if func(l(au,≤),s)=hom(l(au,≤), s), the func(l(au,≤), s) will be an attribute topos. in fact, as functor hom(-,b)∈hom(l(au,≤), s),so, not only hom(-, b) can be seen as a functor from functor category func(l(au,≤), s) itself, but also attribute topos func(l(au,≤), s) can be seen as category, its object is the hom-set of the hom(-,b).it will be the state shot with two functor natural transformation τ:hom(-,a)→hom(-,b) between two functors. the attribute topos func(l(au,≤), s) have the key properties: assume that a is the top attribute of l(au,≥), then a will be the bottom attribue of l(au,≤), so it can be the final object 1u of func(l(au,≤),s). if a is the performance properties of b, that is we have a state shot f:a→b in l(au,≥). it induces a synthesis of partial order sa={fn=g⋅f|fn: cn→a} in func(l(au,≤), s), as shown in fig.3, which is called the sieve screen or filter with the target a. to the attribute of the partial order or category l(au,≤), because all the synthesis partial order fn is the arrow fn from different attributes cn to the top attribute or final object a, sa is the cluster of the partial order fn which is the way from ca to a. as shown in fig.4, they are not only constitute a pullback but also exist a limit. fig.3. the sieve sa, sb , s1 consist of { fi } ,{ hj } and { pj }. fig.2. möbius strip published by atlantis press copyright: the authors 23 j. feng, t.sheng fig.4. the limit lim ci = b of sieve sa 3. model for the relationship of bank interest rate and china economy this model was developed to investigate the role and interrelationship of 5 main parties: bank, government, enterprise, domestic and international capital. in addition to the 5 main parts, we introduce another two variable, time dimension and fundamental economy.the ranges of interest rate studied are four sets :(10%~8%) 、 (8%~6%) 、 (6%~4%) 、 (under 4% we provide each of themthe bank interest, a different weight which represent its effects in real economic situation.2 furthermore, we also see that interest rate impact the economy slowly and usually accumulate in a long run. hence we should not only concern the impact of an interest rate to the inflation when it is set, but also consider the effect of accumulation in the future. time dimension has the unit of year.. then using the model to work out how the interest rate in this year can influent the coming years. meanwhile, different fundamental economy behaves differently in different interest rate situation. here we set 3 fundamental economy: inflation (set to 1), normal (set to 2), deflation (set to 3). we are aware that same economic behaviors have very different result in these 3 situations. here we try to give a proper weight to the 3 fundamental economy in order to simulate the real world economic behavior .3 above all, based on our analysis of all the main parts and dimensions, we built the mathematics model from macroscope to microscope, and from entirety to parts. table 1 represent the condition (“represent sth” add object here) when only considering 1 year interest rate impact on various economy situation. in this case, we need to analyze in deep. table 3 describes in normal economy, how overall macroscopic economy behave when interest rate changes. we can find the reasons by analyzing the through the table. table 2 describes in normal economy, how the 5 main parts behave when interest rate change. however, in inflation or deflation, we should not only consider the effect of interest rate, but also add proper weight. for example, interest rate of 4% in normal economic situation should benefit to domestic capital. but in inflation case, 4% interest rate will promote capital circulate and exacerbate the inflation. that will badly impact other main parts.4 through the analysis of table 2, we understand that each main partier’s behavior toward the change of interest rate in deep is judged by their interactions. all of them impact each other. and the accumulation of the impact represents real world's behavior in certain interest rate. the relationship table below represents typical ways that they impact each other. certainly, we have to consider the economic situation, (foreign hot money is sensitive to the interest rate and exchange rate, in short is the high interest rates will infux and low interest rates will flow out. so the article will not calculation specific but it needs to be considered.)3 as shown in table 3. meanwhile, we should realize that economy is not a shorter process. we should not only concentrate on the effects of interest rate changes in short time but also consider its influence to future economy. of course, the longer the accumulation time, the smaller its effects. s represents the impact and t represent the time.5 txts /= （1） and obviously, a certain weight should be adopted in different fundamental economy. table 1. year interest rate impact on various economy situation deflation normal inflation interest rates（ 10%~8%） it would be devastating if we take a high interest rates under deflation. if we take a high interest rates for a long time, may be it would cause deflation. it is benefit to the overall economy and will help to reduce the economic mobility interest rates（ 8%~6%） it would be adverse to the overall economy and will aggravating inflation. it would increase the uncertainty to the economy and likely to cause deflation. it would help to ease high liquidity interest rates（ 6%~4%） a littlt higher interest rates would be a block for the economy turn back to normal. the interest rates has both advantages and disadvantages. it would be adverse to the overall economy and will make the inflation to be more serious interest rates（ under 4%） it would be beneficial to the economy and will increase the domestic liquidity. low interest rates is good for economic development ,we suggest to maintain. it would be devastating if we take a low interest rates under inflation published by atlantis press copyright: the authors 24 a risk assessment model based on attribute theory table 2. the 5 main parts’ reaction in interest rate change table 3. the deep impact between 5 main parts after realize the relationship of all the main parts, the mathematics model can be achieved by java. fig.5 shows the simulation results. fig.5 describes the experiment result in time dimension t=3(year), and deflation environment. interest rate in first year is 2.6%, second year is 6.8%, and the third year is 2.2%. we assumed all the economic parts can process properly in the range of [0,100]. the lower of the score, the less dangerous it is. if exceed 100, it means the main part is in danger. from picture 1, we see that in deflation environment, keep the low interest rate in 3 years, the overall economy can maintain a smooth progress, and in low risk. same precondition, let’s adjust the interest rate as 7.1% in first year and 6.6% in second year, 6.0% in the third year. then we can get the result shown as fig.6: obviously we see that under a high interest rate, the accumulation effect is very clear. all the main parts are facing the risk of break up. overall risk value also exceeds the preset 100. it correctly represents the requirement of interest rate in different economy situation. after the basic model we also want to build is a model that can show the instant contact of the five parts, so , we use the neatbeans ide 6.9.1 to build out a real-time visual program as shown in the fig.7. interest rates（50%~30%） interest rates（30%~20%） interest rates(20%~10%） interest rates（under 10%） government the revenue of the government have an significanty decrease and expenditure increased significanty which will increase the risk. the government’s revenue decreased and expenditure increased the revenue of the government have an slight decrease and expenditure increased the revenue of the government increase and expenditure decreased bank the bank’s revenue decreased significanty and defaults increased(equals to expenditure) which increase the risk the bank’s revenue decreased and defaults increased(equals to expenditure) the bank’s revenue increased slightly and defaults decreased(equals to expenditure) the bank’s revenue increased and defaults decreased(equals to expenditure) enterprise the enterprise faces a serious problem of cash flow and circulation of goods slow down. the enterprise faces a problem of cash flow and circulation of goods slow down. the enterprise faces a little problem of cash flow and circulation of goods improve slightly. the enterprise have a good cash flow and smooth circulation of goods domestic financial the liquidity of the domestic capital declined obviously and overall economy become weak the liquidity of the domestic capital declined and overall economy become weak the liquidity of the domestic capital declined slightly and have a slightly adverse effect on overall economy it is good for overall economy and the circulation of the domestic captial. foreign hot money the foreign hot money pour into and will have an impact to the domestic financial. the foreign hot money pour into the foreign hot money have little desire to get in. the foreign hot money have little desire to get in. fig.5. result in the rate(2.6%,6.8%,2.2%) fig.6. result in the rate(7.1%, 6.6%, 6.0%) published by atlantis press copyright: the authors 25 j. feng, t.sheng fig.7. real-time demonstration figure 4. conclusion to sum up, this article built a mathematics model for the 5 main parts and their relationship based on qualitative theory method, correctly represents how these main parts behave and their impacts on each other in different interest rates, and properly estimates the risk value for these main parts. the relationship functions provided in this article was based on various related papers, which would require further practice. next, i will work on the relationship functions as well as the weights in order to make it represent closely to real world economy, and let the model work properly. acknowledgements we thank for jingjuan feng, ph.d candidate of us berkeley, for her help in english transformation. references 1. jiali feng, qualitative mapping orthogonal system induced by subdivision transformation of qualitative, criterion and biomimetic pattern recognition，chinese journal of electronics, special issue on biomimetic pattern recognition, 85 (6a), (2006). 2. h. b. malmgren, the credit crisis is not over, the international eeonomy, (2007): 58-61. 3. (switzerland) manuel oman, credit risk assessment, yang yuming beijing: tsinghua university press, (2004). 4. vasicek, o: an equilibrium characterization of the term structure of interest rates, journal of financial economics, 01(5) (1977): 177-188. 5. lihu wang, jiali feng. classifying and learning based on qualitative mapping, proceedings international conference on intelligent information technology, (2002): 559563. published by atlantis press copyright: the authors 26 microsoft word urban seismomorphoses seismic vulnerabilities, an embarrassing legacy urban seismomorphoses seismic vulnerabilities, an embarrassing legacy stephane cartier* laboratory for public policies, political actions and territories, national council for scientific research 14 bis rue marie reynouard 38100 grenoble, france cloe vallette laboratory for public policies, political actions and territories 14 bis rue marie reynouard 38100 grenoble, france cloe.vallette@unicaen.fr hafida mediene laboratory for geographic spaces and country planning cité du chercheur, exiap locale du rectorat, oran, algeria abstract seismic vulnerability challenges sustainable urbanism. antioch, manosque and oran, three mediterranean cities destroyed by earthquakes, demonstrate how preservation of urban patrimony protects populations. the methodological pattern “urban seismic patrimonial strategies” cross investment and patrimonial care to explain natural hazards mitigation as a factor of urban policy. urban patrimonial managers are unaware to seismic threat, which obliges to explain liabilities. buildings evolution observation indicates urban phases. according urban policy but without fatality, natural phenomena reshape urban morphology and amplify social vulnerability. keywords: seismic urbanism patrimony antioch manosque oran. 1. introduction seismic collapses reveal anti-seismic gaps, the inequalities of vulnerability, the laxity of urban control, the shortcomings of protection, demographic imbalances, the dilapidation of buildings and hindrances to evacuation. these weaknesses create a variable urban vulnerability in time and space. unequally distributed among districts (vallette1, 2006), this vulnerability is also dynamic, showing how the uses of the city may develop. thus certain urban sectors combine factors of risks linked to an industrial neighbourhood (beck2, 2006) while others hold major economic (lutoff3, 2000 ; cartier4 2004), administrative or institutional issues (d’ercole, et al.5 2005). a disaster adds a crisis into decision-making and an entanglement of materials, *scartier@ujf-grenoble.fr bodies, activities and responsibilities to ordinary urban complexity (hewitt6, 1983 ; wisner, et al.7, 2004). in order to avoid chaos social sciences suggest a simplification of the urban decision-making process to prevent the tetanising anxiety about an unpredictable threat. in this way, the paper focuses on local manager as key-stakeholders. after some synthetic efforts (coburn and spence8, 1992), different approaches and scales exist among the growing experiences of seismic vulnerability assessment according the site, the architecture type, the scale, the seismic scenarios, etc. (benedetti, et al.9, 1988 ; reiter10, 1990 ; augusti, et al.11, 2001 ; sepe, et al.12, 2008 ; lantada13, 2009). to prognostic damage eventuality, four technical degrees are often used: single building examination in an architectural spirit (zaçek14, 1996), calculation of ductility and retrofitting journal of risk analysis and crisis response, vol. 2, no. 2 (august 2012), 96-106 published by atlantis press copyright: the authors 96 administrateur texte tapé à la machine received 16 december 2011 administrateur texte tapé à la machine accepted 8 june 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine cartier et al. propositions by engineers for a type of building (with or without shaking table tests), identification of a clue point (the design, the age, the material, the roof) and statistical extrapolation with or without use of urban images (gueguen, et al.15, 2008), statistical analyse according resistance of types of building (design, age, material) and urban density (rojahn16, 1986 ; wyss17, 2005). between singular building diagnosis to global overview of urban vulnerability and resilience capacity at universal scale, a middle scale approaches, by neighbourhood or by historical type, helps managers of ancient buildings to integrate safety standards according local geography and population interest to legacy. but these methods are confronted with some difficulties: monographic local approaches doesn’t give elements to compare and class priorities; universal approaches are very generous but doesn’t consider practical difficulties to implement (lack of data, level of expertise, time consuming, political disagreement); statistical overview of seismic loss assessment leads to un-efficient discourses; extreme simplification of technical criteria induces great disappointment of managers. considering the rich literacy of methodology to assess building vulnerability, the missing point is how to get attention from local managers to the protection of patrimony and population, despite of financial cost of retrofitting and ... before improving disaster (lavell18, 1994). confronted with the long time return of seismic hazard, the interest of local managers needs to be stimulated with more ordinary preoccupations: people safety, housing, real estate rent, employment, economical growth, patrimonial proud, tourism impact, city image, community peace, personal interest, electoral triumph ... such as transform seismic safety policy into political ambition. considering the risk situation as a political result, we propose to examine the vulnerability according a typology of urban management of patrimonial buildings, which directly gives information of potential danger for users and ruin of legacy. on the whole technical sciences, social sciences and the authorities ask themselves about how to implement antiseismic policies. in this framework as for the restoration of the qsûr (ben hounet and guinand19 2007), the renovation of dilapidated buildings is sensitive, as much from a technical point of view (consolidation, demolition, reconstruction) as from a political point of view (priority buildings, criteria, identity significance). the complexity of seismic vulnerability of old cities (coburn and spence8, 1992 ; lutoff3, 2000 ; mouroux and lebrun20, 2006) leads to the analysis of their history, the diversity of uses and re-uses of buildings and the delicate financial issues (pico21, 2006). understanding the fragility of old buildings, designed without anti-seismic engineering and badly maintained, requires knowledge on heterogeneous architectures (gueguen et al.15, 2008; leroi22, 2005). mediterranean cities are particularly affected by this seismic situation imposed to inhabitants, real estate managers and civil service administrators. in istanbul, the 1999 earthquakes and the threat of tremors provoked variable reactions. some anchor the protection of the heritage and of activities to the local civic life: the characterization of soils, the inventory of buildings, the census of vulnerable populations or even the analysis of rescue capabilities (pérouse23, 2006). some do not face up to the extent of the diagnosis and reinforcement work in apathy (visier and polo24, 2006). in the city of cairo, after 1992, the vulnerability to earthquakes of some dilapidated districts enhances the constitution of peripheral private cities (denis25, 2006). carved in the stone of mediterranean ruins and the memory of writings (poursoulis et al.26, 2006), earthquake history leaves its mark on the urban architectural heritage, as much for monuments as for ordinary heritage. according to a. le blanc27 (2006), the collective architectural heritage inherits the marks of disasters. it represents a memory resource to guide a long-lasting protection. in this way, in assisi, the partial collapse of the basilica in 1997 is a significant issue for the cohesion of the local community (juilliard28, 2006). following the same logic, but at international level, the reconstruction of the stronghold of bam, an iranian heritage classified by unesco, was at the origin of a strong commitment (afps29, 2004). the symbolic media coverage emphasizes the dilemma between a renovation of the heritage and the reconstruction of housing for the victims of the disaster (juilliard28, 2006). two themes emerge from this brief review of the vulnerability to earthquakes of mediterranean cities: the local economic trend in the city (investment, published by atlantis press copyright: the authors 97 urban seismomorphoses seismic vulnerabilities speculation, needs for infrastructure, economic development) and the management of the architectural heritage (issues for identity and tourism). despite local seismic specificities, a common logic influences the way the heritage should be preserved, left as is or deteriorated. anti-seismic safety yet falls within other factors of urban heritage management. social, economic, political factors guide phases of maintenance against dilapidation, of investment or of abandonment at building, block, district, or city level. urban phases combine decision-making cycles and physical (architectural and mechanical), economic and social transformations. besides, the history and heterogeneity of districts shape each city. in addition to this problem, seismic safety requires simple criteria of selection, maintenance, abandonment or wait, to prioritize tasks. to avoid emotional, business or megalomania management of urban heritage financiers of programmes of international cooperation, national ministries, local elected representatives, architects, directors of town planning or heritage services expect a scientific, technical and organizational answer to their fundamental constraints: knowledge gaps, lack of availability, need for action, lack of financing, difficulty to express the arguments of a reasoned choice, responsibility of priorities... to that end a simple analysis method of the issues and limits to the preservation of ancient buildings exposed to a seismic risk is applied for various mediterranean cities that already were the victims of earthquakes: manosque (france), antioch (turkey) and oran (algeria). through the observation (field visit, photographic inventory, architectural typology, safety survey, mapping) of heritage practices (maintaining, transformation or reinforcement) and the analysis of decisions (public documents, codes, rules, professional recommendations and standards, interviews with city managers, real estate managers, landlords, inhabitants, patrimony protectors), the comparison distinguishes the local conciliation abilities between anti-seismic safety and heritage investment to understand the place of the anti-seismic criterion in the methods of heritage conservation. based on field investigation, this empirical approach tries to confront declarations of managers and in situ observations to prognostic the transformation of urban shape exposed to seismic hazard according evident tendencies inscribed in urban patrimony management. contributions are to be in english. authors are encouraged to have their contribution checked for grammar. american spelling should be used. abbreviations are allowed but should be spelt out in full when first used. integers ten and below are to be spelt out. italicize foreign language phrases (e.g., latin, french). 2. method and fieldwork focused on management of safety in old buildings, these analyse investigate three urban situations to indicate tendencies of local policy according simple criteria. this approach helps to characterize other situation, like beirut (lebanon) in order to mobilise local managers before the collapse. it represents only one point of global analyse of the urban system confronted to seismic hazard. in that way, the libris30 (lebanon risk seismic) scientific program represents the next step of integration of geologic, architectural and human data. the « urban seismic heritage strategies » analysis table (table 1) matches the economic, often tourist, profitability of buildings with the maintenance logic of the architectural heritage (investments, uses, maintenance abilities). from a definition of the criteria decided with the heritage manager (contracting authority, public heritage manager, municipality, administrative institutions, international silent partners), the objective is to qualify the urban phases going along with key decision-making moments, which mark the development of buildings' dilapidation (first deterioration marks, mechanical effects, irreversible thresholds, acceleration of the ruin) in order to develop a common clarity on the state of buildings, the vulnerabilities, the opportunities and perspectives. as a source of dialogue between the actors concerned, this approach allows to progressively share the necessity for a protection of the heritage and the activities with the users and inhabitants: works, financing, rehousing and transfer of hazardous activities. peculiar to each situation observed the management criteria however always imply that a diagnosis of physical elements (location, state of buildings, materials), of mechanical published by atlantis press copyright: the authors 98 cartier et al. elements (development of materials under hydric and seismic strains) and of social elements (density of occupancy, types of use, demographic perspectives, economic resources, rehousing alternatives) be established. more than anything, with or without establishing a model, the matter is to report the possible developments with simple criteria in order to help managers choose a protection approach which should preserve the safety of the heritage, of the use and the user. without a seismic hazard, dilapidation weakens buildings and technical maintenance mitigates vulnerability in particular for low magnitude earthquakes. this position leads to question the motivations of owners to invest and to question the decisions to implement public policies (regulations, investments, vocational training and information). in 1708, the medieval centre of manosque, near the durance fault, was partly destroyed (quenet31, 2005). after a quick reconstruction, some suburbs increased the urban pattern in the 19th century (angignard32, 2005). after 1950, the establishment of the cea in cadarache and the settlement of rapatriates from algeria increased the demographic impetus and urban extension (angignard32, 2005). industries (géométhane, géosel, cea) make manosque an attractive but vulnerable city. heterogeneous constructions hinder the bringing up to standards by multiplying the parameters of diagnosis and reinforcement, all the more so as twelve buildings are listed as historical monuments (drac paca33, 2008). and yet the iter nuclear project attracts new populations. the need for housing and equipments conflicts with the limitation of building areas under prefectural surveillance. this restriction of the offer increases the price of land and intensifies urban density. antioch, the administrative centre of the hatay province located in the southeast of the fault system of anatolia is exposed to a high seismic risk. the long accumulation of seismic strains increases tension for this growing urban space (apame34). following the destruction of the ancient city by an earthquake in 526, the medieval architectural heterogeneity diversified seismic vulnerability: densely populated narrow alleyways of the souk, composite constructions (cut stones, bricks, wood, concrete and metal), projecting stories over the streets, damage traces, ruins... under the influence of haussmann during the french mandate in syria the avenues lined with cut stone and brick buildings encircled the medieval heart of the city. on the left bank, at the expense of the protective forest cover, unplanned contemporary urbanization clings to hills that are vulnerable to streaming, landslides and earthquakes. on the right bank the large avenues of the modern city structured by investments reduce the seismic threat. as a prefectural administrative centre, oran is the second economic city of algeria with a demographic density of 10,676 inhabitants per km2 in 2006 (oran wilaya35, 2006). since it was founded in 903 ad, oran has been marked by demolitions and reconstructions. the earthquake of october 1790 destroyed the main part of the spanish heritage. during the reconquest of oran in 1792, bey mohamed el kabîr initiated the reconstruction of the city. then french colonization (1830-1962) led the city to sprawl. after independence, urbanization was an extension of the colonial period. the historical heterogeneity of constructions led to different phases of urban deterioration. the oldest districts host the biggest part of buildings in a bad state, or even ruined. moreover from 1987 to 1995, despite numerous construction programmes, the housing crisis increased. the moderate earthquake of june 2008, the table 1. analysis table: urban seismic heritage strategies (typical factors) dilapidation/ threat main vulnerability building maintenance financial rentability urban qualification urban examples earthquake owner yes / no yes / no earthquake and rain inhabitants yes / no yes / no earthquake and flood urban users yes / no yes earthquake and war institutions yes / no yes / no published by atlantis press copyright: the authors 99 urban seismomorphoses seismic vulnerabilities daily collapses during winter rains and the perspective of a great international congress in 2010 led the authorities to transfer the populations to the suburbs, to diagnose, demolish or restore the buildings. manosque, antioch and oran present different urban and social characteristics but were demolished and reconfigured by earthquakes. failing to be maintained dilapidated buildings become dangerous while their users (inhabitants, tourists, shop employees, public services, architecture experts) expect a protection but fail to have the same adaptation abilities. the number and degree of dilapidation of the buildings require political choices of conservation to be made according to the available means (technical and financial) and to the needs of each population (reinforcing the sense of belonging, getting a source of income with tourism, maintaining urban coherence...). in order to understand how, why and to which extent each of those cities meets the challenge of the anti-seismic safety of old buildings, we propose a synthesis of researches fitting into various programmes (archeo-seismicity of aps in manosque, anti-seismic protection of the monuments in antioch and the prevention of the earthquake risk in oran) by comparing information via the « urban seismic heritage strategies ». 3. the earthquake risk: a major concern? the uncertainties in manosque resulting from the experts' debates on the localization and direction of the durance fault (peulvast et al.36,1999) make the seismic hazard a suspicious subject. in this context, the urban constraints imposed by the 1997 ppr37 (risk prevention plan) relating to earthquakes carry a lot of weight. to support its consents and refusals of planning or development projects, the local council needs proven scientific results, in particular about the limits of unconstructible « strips » along « suspicious active seismic faults ». all the more so as the variety of local threats (soil movements, flooding, drought, forest fires, industrial installations) superimposes numerous standards. this superimposition is a constraint for public authorities and a source of anxiety for private investors. urban planning then depends on a clarification of geological assessment and of administrative and legal interpretations. in antioch, the mobilization at school (rescue equipment on each floor, weekly «civil protection clubs», regular simulation exercises) and the activities of the engineers chamber (promotion of the antiseismic standards established in 2001 and control) show the seismic awareness. benefiting from strong media coverage this mobilization promotes local solutions aided by the turkish state. the new construction procedures include a study of soils, civil engineering calculations and architectural plans checked by a private controller, with a municipal agreement and paid for by the owner. according to engineers 70% of the new constructions comply with the 2001 paraseismic regulations, a success compared to the previous 40%. in oran the dilapidation of the major real estate heritage is a direct danger to the population's daily life: « oran still loses its old buildings one by one at a worrying pace and the spectre of victims haunts ancient buildings in the old districts of the city. as a reminder, during the last three years, oran has recorded over 400 building collapses with some 1,200 families stricken. » (oran newspaper38 from august 7th, 2008). the coastal natural hazards (storms, rains, flooding, landslides, earthquakes and the effects of corrosion) increase the mechanisms that ruin buildings in connection with a lack of maintenance. the quick deterioration of buildings increases urban vulnerability (135 collapses in 2003). threatened by the slightest rain shower the population can not protect themselves against earthquakes. as a genuine warning shot the june 2008 earthquake in oran confirms how much ordinary dilapidation is a constant vulnerability for precarious populations. since 1997, in order to define areas of intervention, the urban development plan (pdau39, 1997) of oran has listed the « deteriorated » constructions of the city: 528 buildings, i.e. 3,300 houses. the old districts (sidi el houari, el derb and saint-antoine) concentrate half of these very populated collective buildings. the aging of ancient districts increases the housing shortage. besides, following the colonial destruction of historical buildings, this ruined heritage implies preservation choices. emergency or wild planning operations, started without any historical, architectural or archaeological awareness destroy entire districts, like the calère district. published by atlantis press copyright: the authors 100 cartier et al. 4. desperate times call for small measures 4.1. architectural heritage: which profitability criteria? the municipality of manosque tries to reconcile the preservation of buildings, the needs for infrastructure40 and the constraining hazards by proposing to restructure the networks, to develop the city centre and to build collective equipment. according to the director of the mayor's cabinet and to the director of the municipal services, the renovation initiated in the 1970's represents a political showcase to reduce the desertion of the city centre (saunier41, 2004). certain salvaged buildings are reused as public equipment, others provide renovated housing. despite the success achieved the results are mixed: dilapidation of buildings, lack of attractiveness and utility (saunier41, 2004). the cost of renovations encourages the municipality to sell the heritage to private or mixed operators. the drac (regional direction of cultural affairs) however freezes the sales of listed historic buildings. moreover few purchasers are able to accept the maintenance of buildings, the antiseismic safety and profitability. finally the double supervision on certain buildings makes it difficult to share responsibilities. all these elements are in line with a difficulty to generate a process of preservation and development of the historical heritage which suggests a difficulty to integrate the anti-seismic safety of ancient, composite and little ductile buildings. as a tourist asset and identity resource, the architectural heritage of antioch also results from distinct interventions according to the listing status. private homeowners favour their needs as users (housing, shops and workshops) over safety. thereby old buildings are preserved to become tourist shops or hotels but others are modified for business needs regardless of the solidity of the structure. the fragmentation of responsibilities spreads investments. in a cave on a cliff the saint-pierre church is under the authority of the catholic bishop but the visit of this national heritage is under the state supervision of the administration of antiquities and tourism. places of worship are under a different supervision: the direction of religious affairs for mosques under the authority of the prime minister, for imams are civil servants; private properties, parishes, dioceses, patriarchates for churches. only in case of a denunciation the heritage commission controls the decisions made by property owner dioceses. whatever their owner the majority of ancient monuments are protected by the administration of antiquities and tourism of the state as a national heritage, as a symbol of national unity. the common administration of antiquities and tourism reflects the will to promote monuments as a cultural pillar for the tourist industry. the inventory of monuments determines the functions of each building, the rules of conservation and the price of the visit. the monuments should generate the means of their preservation but far from istanbul's tourist godsend, the price of the visit does not make the isolated monuments of hatay profitable, for they depend on state subsidies. the development of the heritage also depends on contextual elements. even if national regulations promote local investment in the anti-seismic safety of the heritage, the tourist potential is hindered by the distance to the airport. besides, the safety of users is unequal. if the safety instructions and the vigilance of the guards protect the visitors of the colossal titus tunnel, on the contrary the citadel of the crusaders of antioch, listed in the national heritage without being a tourist site, is an accessible ruin laying itself open to a seismic threat. the transfer requires to clarify the responsibilities in terms of safety and anti-seismic reinforcement:  which criteria for heritage protection?  which experts should diagnose safety?  which seismic diagnoses? each question fuels shared competences between public entities while state intervention is now considered as a normal thing. under the advice of ministries (ankara), the director of the museum of antiquities (antioch) rather has the ordinary competence for the organization of mosaic exhibitions, the direction of the team and the cultural promotion of monuments. in case of an urgent problem the director requests a technical inspection to the state administration for the inspection of buildings (adana) which will assess safety, decide on investments and choose the contracting company. but there, remains a global concern about traditional know-how disappearing for instance for the roof structure of wooden houses. programmes of urban restoration favour a romantic preservation of the heritage. but restoration techniques can involve seismic mitigation as the care provided for the safety of listed monuments in published by atlantis press copyright: the authors 101 urban seismomorphoses seismic vulnerabilities the turkish heritage shows. in oran, two programmes show the issues of the renovation of ancient buildings. from 1997, the pdau assessed that « in 1987, more than 80 % of houses were over 25 years of age and even if the current construction impetus was maintained the number of houses over 50 years of age by 2015 would represent more than 45% of the whole estimated stock. » (pdau38, 1997:193). in march 2008, before the renovation works started, the office for real estate promotion and management (opgi) of oran listed 1,990 buildings requiring « imminent intervention » (opgi42, 2008). the financial limits of opgi require selecting the buildings according to their architectural interest and their visibility, in order to improve the city's image and encourage the neighbouring owners. focused on 200 buildings in the historical districts this subjective strategy favours heritage and tourist promotion. in addition the european cooperation programme archimedes43 applied to four mediterranean cities (beirut, el mina, istanbul, oran) promotes the transfer of competences for urban renovation to develop the heritage and sustainable tourism. in oran this project applies to the sidi el houari district, the old city centre that has become a neglected transit district. relying on the architecture and beaux-arts schools the objective is to reintegrate this district into the city by teaching specialists the techniques of heritage restoration and conservation, of heritage inventory, by developing a work plan, renovating a building, planning other renovations in partnership with public developers. if the primary function of housing is to protect people a state of ruin remains a disaster. urban pauperization goes beyond any aesthetic concern and turns the central districts into very hazardous ruins for the inhabitants and activities (shops, markets) traumatized by the threat of the lightest rain. discontent and disappointment are such that even efficient rehousing operations go through tension and require massive police interventions to prevent riots. if « (…) the heritage policies can be tools for the management of the seismic risk. » (le blanc27, 2006:246), the contribution of oran projects is indirect by the involvement of safety in the political agendas, favoured in emergency by the prospect of an international congress of the gas industry in 2010. this sudden urban awareness goes along with luxurious intentions made of saudi investments with a koreainspired model... between an anarchic concreting and the mirages of the « city of the sea », the suburban growth leaves the popular city centre behind. pedagogically concentrated on an exemplary building the method of the archimedes43 programme suggests protecting the district of sidi el houari with a will to preserve its architectural aspect. vividly the programme plans to eradicate the already ruined buildings or the ones that could possibly collapse. if the ambition is to really transfer to oran a methodology of urban regeneration favourable to the economic development and the tourist promotion of the cultural heritage, nothing on the other hand seems to indicate a specific attention to the local natural threats. 4.2. urban decentring: which efficiency? in manosque, the issue of the city centre and the needs for equipment drive the municipality to build new public buildings (hospital, schools) outside the historical defensive walls. this strategy is common: «the cost of the compliance to standards of heritage buildings and needs for surfaces have led the companies to look for locations outside «historic» spaces of city centres and ancient suburbs. » (garat, gravari-barbas and veschambre44, 2008 : 6). the new buildings then respect all the safety standards, of which anti-seismic standards. however, outside, they separate activities and ignore the revitalization of the historical centre. in antioch, the fragility of the old city clashes with the efforts of seismic safety in other districts. yet, despite real estate conflicts and the abuse of sea sand in the constructions, the demolition of hazardous buildings allows to reconstruct private new housing complying with the 2001 seismic codes, however without any concern for homogeneity in the historical style. the difference of intervention according to districts is then essentially linked to the heritage value of the old city. in oran, the geographical transfer, particularly marked towards the east, generates new tower districts with collective housing succeeding the individual detached houses of the suburbs. urban programmes involve the demolition of ruined housing and the rehousing of inhabitants : « the planning policy is centred upon the production of new housing in the suburbs to the detriment of an urban centre that deteriorates alarmingly. » (opgi42, 2008: 1). favoured by the « 1 published by atlantis press copyright: the authors 102 cartier et al. million houses» presidential plan (2004-2009), these rehousing programmes are slowed down by construction problems, the difficulties to establish fair lists of beneficiaries, slow decision-making and experts' meticulousness. besides, these programmes increase with a list of new disaster victims as new crises, exceptional seismic tremors (june 2008) or ordinary autumn rains dictate. through lack of money inventories are just the initial step of programmes that are announced but rarely achieved. socially the objective is to simultaneously carry out the rehousing of hundreds of families as during summer 2008 and to commit the population into transforming the districts. a rare initiative in algeria, inhabitants' associations invited to the international symposium on urban revitalization try to get informed and to spread the projects (autumn 2008). they gather inhabitants threatened by the collapse of their houses, inhabitants waiting for architectural restoration and owners. the commitment of real estate owners is still hard to achieve due to the unclear real estate status of inhabitants. far from being rural squatters fleeing terrorism the resident families have often settled for a long time in the districts, at least since independence. very often born in the buildings or in the neighbourhood, the inhabitants hold certificates of occupancy and not property certificates. in the absence of clear criteria for public rehousing, the tenants who give up paying the rents for ruins and legatees without property certificates, the rehousing lists exclude a large population coming from ancient districts to the benefit of real estate speculators. 4.3. suspense in manosque, the pro-active renovation of the city centre places the seismic and heritage question on the municipal agenda. nevertheless, administrative difficulties, the deterioration of buildings, heritage listing and the renovation costs involved by restricting standards, both anti-seismic and aesthetic, make the task heavy. the strategies of transfer of responsibilities and spatial transplantations throw a city already lacking expansion off balance. besides the identity issue of the city centre is not reinforced by a tourist, commercial or political economic profitability which increases the municipal debt. as an urban showcase the effort of cultural development of the heritage has no immediate profitability despite the past choices of the municipality to purchase historical buildings. in antioch, profitability encourages heritage maintenance. the anti-seismic effort promotes the tourist potential without the collapse issue. however the slow pace of profitability slows down the anti-seismic reinforcement of urban monuments. under the supervision of the culture and tourism ministry, the direction of heritage and museums comes to a compromise with archaeologists, art historians, town planners, lawyers and municipalities without experts on seismic safety. the direction monitors restoration projects without relying on a surveillance organization. always delicate in case of an anti-seismic reinforcement the restoration can lead to a denunciation and judicial proceedings. facing this situation and the multiplicity of supervision authorities for worship places, the turkish state plans to transfer some monuments to local authorities which raise the question of their ability to finance studies, to set up a conservation policy, to organize visits and safety measures, to produce expertise, works and control. indirectly insurance companies could become the managers of historical buildings protection, of tourist development and of investment safety. oran suffers from a disastrous lack of maintenance and from the chronic weakness of the urban project for the historical heritage. after the current phase of ruins the tragic state of dilapidation naturally leads to a phase of speculative reinvestment through a complete renewal of buildings and populations. however the international involvement in a programme to salvage the heritage mitigates this destiny and provides for an opportunity of tourist promotion while preserving the façade urban identity. breaking the deadlock requires means. as a historical opportunity the public authorities have the petrodollar financial godsend at their disposal to end the deadlock and the auspicious archimedes43 international cooperation. in the framework of the « 1 millions houses» programme (2004-2009), the construction of hundreds of social houses in oran suburbs provides for a solution to the victims of a disaster. but the mobilization of means remains complex. in algeria, heritage protection remains the subject of paradoxical virulent political injunctions and of a tangible administrative laxity. governments succeed one another, the national heritage passes away. published by atlantis press copyright: the authors 103 urban seismomorphoses seismic vulnerabilities 5. conclusion in the context of mediterranean transformations, the maintenance of a world heritage requires to realistically admit contemporary social, economic and political tensions. if the heritage deserves an honorary place, economic transitions tend to ignore it due to its low profitability. little useful to provide accommodations to populations, ancient buildings bother real estate speculators always tempted by the tabula rasa or a « neapolitan-like » façade monopolizing (montesano45, 2003). confronted with an anarchic urbanization the public authorities hardly place the protection of monuments, a fortiori against a ghostly seismic threat, on their agendas and budgets. as a now classical solution for ancient cities tourist promotion nevertheless requires an effort of seismic mitigation. all the more so relevant that it is easily readable for managers the «urban seismic heritage strategies» (table 2) matches investments and the protection of the heritage to select renovation policies in the light of natural risks. the information gathered characterize urban vulnerabilities and renovation efforts: interests of investments (maintenance, profitability, speculation), tools for public policies (regulation, subsidies, training, information), links between real estate interests and protection policies, decision-making process (abandonment/freezing of the heritage, safety promotion, urban renewal). the developments in the degradation of buildings (first deterioration marks, mechanical effects, irreversible thresholds, acceleration of the ruin) show the cycles of urban decisions, arrangements of political initiatives and of private opportunities. by increasing vulnerability natural events mark the development and morphology of a city but are absolutely not inevitable. the opportunities for a symbolic and financial reinvestment of the heritage remain scarce and limited. renovation efforts must however methodically integrate the adaptation to seismic conditions to all the phases of the architectural or urban project (cartier and elassad46, 2010). to reconcile the protection of the heritage and the safety of users it is then essential to identify which item requires the most safety and why. understanding their motivations helps prioritize the means according to safety objectives, financial resources, acceptable responsibilities and local abilities. stabilizing real estate and land statuses is the first step to prevent a ruining of the heritage through the harmful joint ownership between private or public joint owners. even if the necessary cost for an anti-seismic renovation excludes investments from the poorest people, involving inhabitants into the project and sharing the benefits of tourism are a guarantee for a further maintenance of the heritage. hope of community involvement supposes the capacity of leaders to be proud of the heritage and aware of danger. inhabitants confronted with all year long collapses, for example in oran, express directly their fear and needs by reclaiming retrofitting or new homes. the general tendency of decrease of density in historic vulnerable quarters leads to push people and activities outside historical perimeter in suburbs. but central historical quarters represent also some amenities and resources (familial and community ties, employment facilities, trade opportunities, education and health services). settling on dangerous housing is also a mean to keep rights on land ownership, real estate speculation in city, reclaim for new housing. more than techniques, energies must be devoted to a dialogue between actors so as to understand safety objectives, available means, acceptable constraints and table 2. urban seismic heritage strategies dilapidation/ threat main vulnerability building maintenance financial rentability urban qualification urban examples earthquake owner yes no cultural promotion manosque earthquake and rain inhabitants urban users no no ruin oran earthquake and flood inhabitants owners yes yes investment antioch earthquake and war inhabitants institutions no yes land speculation beirut30 published by atlantis press copyright: the authors 104 cartier et al. agendas. but thinking in terms of « negotiated » risk requires thinking on the temporality of the project to reduce the vulnerability of ancient buildings with a view towards sustainable tourism. in this framework seismic safety can be supported by heritage policies. this strategy allows to associate the symbols of collective identity with the protection of the users of monuments. references 1. c. vallette, et chaque année, au retour des pluies…, (master thesis, lyon ii university, lyon, france, 2006). 2. e. beck, approche multi-risques en milieu urbain, application à l’agglomération de de mulhouse (hautrhin) (phd thesis, geography, louis pasteur university, strasbourg, france, 2006). 3. c. lutoff, le système urbain niçois face à un séisme. méthode d’analyse des enjeux et des dysfonctionnements potentiels, (phd thesis, geography, savoie university, chambéry, france, 2000). 4. s. cartier, contraintes sismiques et décisions économiques : de la crise à la réduction de la vulnérabilité, revue du service géologique de l’algérie, 12 (2004) 193–204. 5. r. d’ercole, p. metzger and n. bermudez arboleda, espaces enjeux, espaces vulnérables. le cas de quito (equateur), systèmes d’information géographique et gestion des risques, (2005) 16–19. 6. k. hewitt, the idea of calamity, in the technocratic age in interpretations of calamity: from the viewpoint of human ecology (allen and unwin, london,1983), pp. 332. 7. b. wisner, p. blaikie, t. cannon and i. davis, at risk, natural hazards, people vulnerability and disasters, (routledge, oxon, 2004 [1994]). 8. a. coburn and r. spence, earthquake protection, (wiley, chixhester, 1992). 9. d. benedetti, g. benzoni, and m.a. parisi, seismic vulnerability and risk evaluation for old urban nuclei, earthquake engineering & structural dynamics, vol.16 num.2 (1988) 183–201. 10. l. reiter, earthquake and hazard analysis – issues and insights, (columbia university press, new york. 1990). 11. g. augusti, m. ciampoli and p. giovenale, seismic vulnerability of monumental buildings, structural safety, vol.23 num. (2001) 253-274. 12. v. sepe, e. speranza and a. viskovic, a method for large-scale vulnerability assessment of historic towers, structural control and health monitoring, 15 (2008) 389-415. 13. n. lantada, l.g pujades and a.h barbat, vulnerability index and capacity spectrum based methods for urban seismic risk evaluation. a comparison, natural hazards, 51 (2009) 501 – 524. 14. m. zacek, construction parasismique : risque sismique, conception parasismique des bâtiments, réglementation, (editions parenthèses, arles, france, 1996). 15. p. gueguen, c., lutoff, p.a., davoine, g., talierco, f., cotton and s. cartier, analyse de la vulnérabilité sismique dans un pays à sismicité modérée: le cas de grenoble, actes du colloque « vulnérabilités sociétales, risques et environnement, comprendre évaluer » (toulouse, france, 2008). 16. c. rojahn, r.l sharpe, r.e scholl, a.s kiremidjian, r.v nutt ans r.r wilsan, earthquake damage and loss evaluation for california, earthquake spectra, vol.2 num.4 (1986) 767-782. 17. m. wyss, human losses expected in himalayan earthquakes, natural hazards, 34 (2005) 305 – 314. 18. a. lavell, opening a policy window : the costa rican hospital retrofit and seismic insurance programs 19861992, international journal of mass emergencies and disasters, vol.12 num.1 (1994) 95-115. 19. y. ben hounet and s. guinand, la restauration des qsûr : institution du patrimoine et enjeux de mémoire, espaces et sociétés, 128-129 (2007) 151–170. 20. * p. mouroux et b. le brun, presentation of “risk-ue project”, bulletin of earthquake engineering, 4(4) (2006) 323–339. 21. l. pico, géographie et assurance : le risque sismique dans les espaces urbains mal documentés : le cas de beyrouth ((phd thesis, geography, sorbonne university, paris, france and saint-jospeh university, beyrouth, libano, 2006). 22. e., leroi, réalisation d’une étude sur les critères pour la définition de mesures de réduction du risque sismique du bâti existant en france, (urbater, ministère de l’écologie et du développement durable, france, 2005). 23. j-f, pérouse, catastrophes, risques sismiques et redécouverte de la dimension locale à istanbul, in pouvoirs urbains et sociétés urbaines face aux risques, sur le pourtour de la méditerranée (anthropos– economica, paris, 2006), pp. 56–78. 24. c. visier and j.f. polo, les séismes de 1999 en turquie, un laboratoire de la société civile, in pouvoirs urbains et sociétés urbaines face aux risques, sur le pourtour de la méditerranée (anthropos–economica, paris, 2006), pp. 79–98. 25. e. denis, risque et actualisation de l’écologie métropolitaine : la figure emblématique des citées privées au caire, in pouvoirs urbains et sociétés urbaines face aux risques, sur le pourtour de la méditerranée (anthropos–economica, paris, 2006), pp.168–195. 26. g. poursoulis, a. levret, n. lambert, a. rideau and b. helly, the 1708 manosque earthquake (france): a reading of its archaeological traces as a contribution to estimate the effects on buildings, journal of seismology, 10 (2006) 511–526. 27. a. le blanc, les politiques de conservation du patrimoine urbain comme outils de gestion du risque sismique. trois exemples en italie : noto, assise, gémone published by atlantis press copyright: the authors 105 urban seismomorphoses seismic vulnerabilities (phd thesis, geography, provence university, aixmarseille, france, 2006). 28. c. julliard–macian, le tremblement de terre d’assise du 26 septembre 1997, ou l’écriture d’un drame parfait in in pouvoirs urbains et sociétés urbaines face aux risques, sur le pourtour de la méditerranée (anthropos– economica, paris, 2006), pp. 33–55. 29. afps, le séisme de bam, iran, du 26 décembre 2003 (association française du génie parasismique, paris, 2004). 30. libris, contribution à l'étude du risque sismique au liban (anr, paris, 2009). 31. g. quenet, les tremblements de terre aux dix-septième et dix-huitième siècles. la naissance d’un risque (édition champ vallon, seyssel, france, 2005). 32. m. angignard, la perception du risque sismique en provence, exemple de collégiens à manosque, (master thesis, paul valery university, montpellier, france, 2005). 33. drac paca, immeubles et jardins en provence-alpescôte-d’azur, (ministère de la culture et de la communication, france, 2008). 34. apame programme apame (programme européen) : the impact of large earthquakes on the archeological sites and cultural heritage in west syria and lebanon, m. meghraoui (dir.) 35. oran wilaya, directory of oran wilaya, (oran wilaya, algeria, 2006). 36. j–p. peulvast, e. baroux, o. bellier ans m. sébrierlien, le problème de l'activité des failles de nîmes, de saloncavaillon et de la moyenne durance (se de la france) : apports de la géomorphologie structurale, géomorphologie, 5–4(5) (1999) 327–358. 37. manosque municipality, fiche synthétique sur les risques recensés (risk prevention plan’s extract approved by prefect, april 24th 1997, 2005). 38. j. boukraâ, oran : un enfant de 11 ans tué dans un effondrement, quotidien d’oran (oran newspaper from august 7th 2008). 39. pdau oran, plan directeur d’aménagement et d’urbanisme (pdau) du groupement d’oran, (wilaya oran, oran, algeria, 1997). 40. national institute for statistic and economic studies, schéma de cohérence territoriale (scot) de la région de manosque, région provence alpes côtes d’azur (national institute for statistic and economic studies, france, 2005). 41. p. saunier, manosque, alpes de haute-provence, actualisation du plan de référence du centre historique (ville de manosque / urb’action, marseille, france, 2004). 42. opgi oran, rapport de présentation, opération de réhabilitation de 200 imeubles à oran (ministère de l’habitat et de l’urbanisme, office de promotion et de gestion immobilière d’oran, département développement de la promotion immobilière et foncière et de réhabilitation, algeria, 2008). 43. archimedes (actions to regenerate cities and help innovative mediterranenan economic development enhancing sustainability), newsletters (euromed, eurpoean union), 1/2/3 (2007–2008). 44. i. garat, m. gravari-barbas and v. veschambre, préservation du patrimoine bâti et développement durable : une tautologie ? les cas nantes et angers », développement durable et territoires, 4 (2008) url : http://developpementdurable.revues.org/4913. 45. g. montesano, di questa vita menzognera, (universale economica feltrinelli, milan, 2003). 46. s. cartier and k. elassad, enjeux et procédures de protection sismique du patrimoine historique, la réhabilitation des ex-galeries de france en musée d’art moderne d’alger, annales de la recherche urbaine, 106 (2010) 171–178. published by atlantis press copyright: the authors 106 小麦大田节水灌溉气象等级指标研究* meteorological grading indexes of water–saving irrigation for corn jingjing xiao zhejiang climate center, meteorological bureau of zhejiang province, hangzhou, 310017, china zhiguo huo chinese academy of meteorological sciences, china meteorological administration, beijing, 100081, china dapeng huang national climate centre, china meteorological administration, beijing, 100081, china fenfen guo the second institute of oceanography, soa, hangzhou, 310012, china yan jiang department of emergency response, disaster mitigation and public services, china meteorological administration, beijing, 100081, china gaofeng fan zhejiang climate center, meteorological bureau of zhejiang province, hangzhou, 310017, china youhao e, caixia yu chinese academy of meteorological sciences, china meteorological administration, beijing, 100081, china abstract to optimize irrigation and provide a technical guide on agricultural production, the water-saving irrigation meteorological grading indexes of different development stages of maize were built in this paper, based on the principle of farmland water balance, the theories of insufficient irrigation and regulated deficit irrigation, and by using multidimensional information including day-by-day meteorological data in 1961 － 2008 from 500 meteorological stations, soil moisture data in ten-day periods in 1993－2008 from 239 agro-meteorological stations (of which, the data in 1980－2008 came from 68 agro-meteorological stations)，irrigation amount and yield data were collected to analyze the relationship between the moisture deficiency rates, the quantity of water deficit and yield reduction rates at sowing-jointing stage, jointing-heading stage, heading-mature stage and the whole growth stage of maize referring to the product-water function(fao-1979). results indicated that water-saving irrigation meteorological grading indexes from ⅰ－ ⅲ levels of moisture deficiency rate and the quantity of water deficit (secondary indicators) were [10%, 25%), [25%, 40%) and ≥40%, corresponding to 1 － 2 irrigations (600m3∙hm–2 each time), 2－3 irrigations and >3 irrigations for the whole growth stage of maize, respectively. for sowing-jointing stage, the indicators were [15%, 30%), [30%, 45%) and ≥45%, corresponding to 0.5 irrigation, 1 irrigations and ＞1 irrigation, respectively. for jointing-heading stage, the indicators were [15%,30%), [30%, 45%) and ≥45%, corresponding to 0.5 irrigation, 1 irrigation and ＞1 irrigation, respectively. for heading-mature stage, the indicators were [15%, 30%), [30%, 50%) and ≥50%, corresponding to 1 irrigation, 1 － 2 irrigations and ＞2 irrigations, respectively. these indices could supply scientific support for the optimization of irrigation，and give guidance to agricultural production. key words: maize, water-saving irrigation, meteorological index 玉米节水灌溉气象等级指标研究 肖晶晶 1，霍治国 2 ※ ，黄大鹏 3，郭芬芬 4，姜燕 5，樊高峰 1，俄有浩 2，于彩霞 2 1.浙江省气候中心，杭州 310017 corresponding author: huozhigg@cams.cma.gov.cn 基金项目：“十二五”国家科技支撑计划课题“重大农业气象灾害预测预警关键技术研究（2011bad32b02）”、国 家标准项目（20083116-t-416） journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 95-102 published by atlantis press copyright: the authors 95 willieb typewritten text willieb typewritten text willieb typewritten text willieb typewritten text received 20 may 2013; accepted 27 june 2013 2.中国气象科学研究院，北京 100081 3.国家气候中心，北京 100081 4.国家海洋局杭州海洋工程勘测设计研究中心，杭州 310012 5.中国气象局应急减灾与公共服务司，北京 100081 摘要：基于农田水分平衡原理、非充分灌溉理论和调亏灌溉理论，利用全国 500 个地面气象观测站 1961－ 2008 年逐日气象资料、239 个农业气象观测站 1993－2008 年（其中 68 个站为 1980－2008 年）逐旬土壤湿 度，以及实际灌溉量、逐年作物产量等资料，结合 fao-1979 作物水分–产量关系函数，分析全国玉米种植 区播种–拔节、拔节–抽穗、抽穗–成熟和全生育期的水分亏缺率、水分亏缺量与减产率之间的关系，构建玉 米不同发育期和全生育期节水灌溉气象等级指标。结果表明：玉米全生育期节水灌溉气象ⅰ－ⅲ级水分亏 缺率和水分亏缺量（辅助指标）指标分别为[10%, 25%）、[25%, 40%）、≥40%和 1－2 水、2－3 水、＞3 水；播种–拔节为[15%, 30%）、[30%, 45%）、≥45%和 0.5 水、1 水、＞1 水；拔节–抽穗为[10%, 20%）、[20%, 35%）、≥35%和 1 水、1－2 水、＞2 水；抽穗–成熟为[10%, 25%）、[35%, 40%）、≥40%和 0.5 水、1 水、 ＞1 水。指标体系可以为优化农业灌溉、指导农业生产提供科学支撑。 关键词：玉米；节水灌溉；气象指标 1. 引 言 玉米是中国重要的粮食作物和饲料作物，其生育 期耗水多且产量对水分亏缺的敏感指数较大。中国 95%以上的玉米种植在华北、东北、西北等干旱和半 干旱地区，水分不足一直是玉米高产、稳产的限制因 素 1。干旱是目前影响中国玉米生产最主要的农业气象 灾害，其发生的频率、强度都有增大的趋势，1997、 1999 和 2000 年中国玉米因旱减产分别为 20.6%、11.4% 和 27.5%2。 农业干旱是指由于外界环境造成作物体内水分失 去平衡而发生水分亏缺，影响作物正常生长发育，进 而导致减产甚至绝收的一种农业气象灾害 3。如何准 确、定量评估农业干旱对产量的影响、农业措施的增 产效应及投入产出成本优化配置，提升农田水分利用 效率，对实现农业可持续发展，开展实时、有针对性 的农业节水灌溉对策和措施的制定具有积极意义。中 国气象、农业、水利部门对农业干旱的防灾减灾工作 一直高度重视，但受国情因素限制，农业靠天吃饭的 局面依然没有得到根本改变，干旱依然是制约中国粮 食安全生产的最大自然灾害。目前中国农田灌溉水的 有效利用系数仅为 0.5，远不及发达国家的 0.7－0.84。 如何提高农田水分利用效率，即探寻适宜灌溉时间、 灌溉量的节水灌溉指标成为节水灌溉的关键问题。节 水灌溉以非充分灌溉理论和调亏灌溉理论为依据 5，其 指标构建以土壤-植物-大气连续体（soil-plant atmosphere continuum，简称 spac）中的水分运移机 理为基础，主要包括土壤水分指标、植物生理指标、 综合效应评价指标等 6–8，其中土壤含水率是节水灌溉 最常用的指标之一，即基于土壤供水量来判定作物是 否发生水分亏缺，通过土壤水分上、下限指标值合理 安排灌溉时间和灌溉量来实现节水 9–10。然而，由于资 料获取难易程度、土壤水分参数差异，且现有的节水 灌溉指标一般是通过田间试验得到，局地性较强。该 类指标不仅推广应用难度较大，且难以与实时降水监 测、未来降水预报进行耦合集成应用。目前试验获取 指标主要针对作物全生育期，实际上不同发育期干旱 对作物产量的影响不同，相同的旱情在不同时期进行 灌溉的增产效应也有所差别。 本文在当前大田生产现状下，以农田水分平衡原 理、节水灌溉理论和水分敏感系数为基础，耦合地面 气象观测、农业气象观测、作物产量、统计年鉴等多 元信息资料，分析气象条件对玉米不同发育阶段水分 亏缺率及其对减产率影响的量化关系，筛选导致玉米 不同减产程度的水分亏缺临界指标、变化范围，构建 节水灌溉气象等级指标。为通过对玉米不同发育阶段 的需水规律及其当前时段作物水分亏缺率、亏缺量等 的实时分析，结合未来 3 天的降水定量预报，进行区 域节水灌溉气象等级预报，开展实时、有针对性地玉 米节水灌溉气象服务提供科学依据。 2. 资料与方法 2.1 资料来源 图 1 是论文气象与棉花研究站点分布图。 气象资料、土壤湿度资料和作物发育期资料取自 国家气象信息中心，气象资料包括 500 个站点 1961－ 2008 年逐日的气压（hpa）、降水量（mm）、最高温 图 1 气象与玉米研究站点分布 published by atlantis press copyright: the authors 96 度（℃）、最低温度（℃）、水汽压（hpa）、风速（m·s–1）、 日照时数（h）等，个别资料缺失值采用多年平均值代 替。土壤湿度资料为 239 个农业气象观测站 1993－ 2008 年（其中 68 个站为 1980－2008 年）的逐旬土壤 相对湿度、土壤水分常数等。作物发育期资料取自各 地农业气象观测站，产量资料来自各地的统计年鉴。 干旱资料来自《中国气象灾害大典·北京卷》和《中国 气象灾害大典·天津卷》11,12。 2.2 研究方法 论文从当前大田生产水平和管理水平出发，依据 农田水分平衡和作物水分敏感系数，耦合气象、产量 和灾情历史资料，分析不同年份下作物产量与水分的 关系。计算玉米种植区不同年份下玉米播种–拔节、拔 节–抽雄、抽雄–成熟和全生育期水分亏缺率、水分亏 缺量及对应年份的减产率，按照自然灾害等级划分方 法 13，以减产率 5%、10%和 20%为界限点筛选导致玉 米不同减产程度的水分亏缺临界指标、变化范围，构 建玉米节水灌溉气象等级指标。囿于篇幅，论文以玉 米全生育期的指标构建为例，具体说明玉米节水灌溉 气象等级指标的构建和验证情况。 2.2.1 研究区域、样本的选择 选取站点遵循以下原则：①选取发生干旱、减产 的年份，且减产率不少于 3%；②选取灌溉能力有限的 区域；因为灌溉能力强的区域，作物减产往往是由于 病虫害、田间管理、其他气象灾害等多方面因素造成， 干旱不是主要的减产因素；③选取灌溉能力强、干旱 严重、有灌溉量记载的地区年份进行指标验证。 2.2.2 干旱减产致灾因子的识别 干旱减产是多个因子共同影响的结果，干旱减产 致灾因子的识别是本论文首先要解决的问题。论文涉 及灌溉资料、未灌溉或者少量灌溉资料和噪音资料等 三部分资料。分析过程中须剔除噪音资料，分离灌溉 资料，剔除涝渍害和干热风等非干旱原因为主而减产 的噪音资料。灌溉资料即有记录的干旱年份资料，该 部分资料作为验证资料单独处理。为此论文引入 fao-1979 作物水分–产量关系函数式 14 和先验知识判 断对数据进行初步识别。 2.2.3 发育期水分亏缺量/率的计算 基于实际生产的玉米不同发育期水分亏缺量/率 的计算，采用参考文献 15 给出的综合考虑自然供水 量、灌溉量和作物需水量的算法，计算公式如下：      m j m j jjljj egpwz 1 1 01d （1）              m j j m j m j jljjj eegpwd 11 1 01 %100/ （2） 式中：dz/d 为作物某发育期的农田水分盈亏量/ 率 （mm/%）。当 dz/d＞0 时，农田水分盈余，称为 农田水分盈余量/率；当 dz/d＜0 时，农田水分亏缺， 称为农田水分亏缺量/率。w01 为时段始第一天的土壤 有效底墒量（mm）。p 为时段内的日降水量（mm）。 gl 为当前发育期内的灌溉量。e 为时段内的作物日需 水量（mm），结合 fao–56 推荐系数计算 16，如表 1。 j 为时段日数序号，j=1,2, …,m，m 为时段日数。 表中：季初阶段为从发芽到覆盖 10％的土地；作 物发育阶段为从覆盖土地 10％到覆盖土地 80％；季中 阶段为从覆盖土地 80％到开始成熟；季末阶段为从开 始成熟到收获。 2.2.4 水分亏缺率与减产率理论关系式的引入 fao-1979年提出关系式14：    maxmax /1/y-1 wwky  （3） 式中：y为作物实际产量；ymax为作物最高产量； w为作物实际水分供给量；wmax为作物理论需水量；k 为产量反应系数。由于玉米全生育期的水分–产量响应 系数为“1.25”17，当k=1.25时，其水分–产量关系函数为 一条斜率1.25的直线，则（3）式可变换为：    maxmax /125.1/1 wwyy  （4） 在作物干旱减产年，作物实际产量低于其趋势产 量，作物实际水分供给量低于其理论需水量；因此， 可近似地取作物趋势产量为最高产量，作物需水量为 理论需水量。则（4）式可变换为：     maxmaxmaxmax /25.1/ wwwyyy  （5） （5）式左边为作物减产率，右边则为作物水分 亏缺率。理论上可以认为在一定范围内二者近似存在 上式的关系。通过比较二者的实际值与理论值的相似 关系，可进行作物干旱减产年验证站点的合理选择、 剔除非主要由干旱导致作物减产的异常站点，以及进 行等级指标确定的合理性分析等。 需要说明的是，直线模型理论存在一定的缺陷， 即在水分亏缺率超过作物忍受的范围时，也会获得相 应的减产率。因此，在实际验证时，主要将其作为指 标合理性的参照进行评价。 3. 结果与分析 3.1 玉米水分亏缺率等级指标 基于全国玉米产区逐站、逐年的水分亏缺率、减 产率计算结果，筛选由干旱导致玉米减产且灌溉能力 有限的站点和年份，表2和表3按照不同减产率等级给 出了不同站点、不同年份春玉米和夏玉米全生育期水 分亏缺率与减产率的对应情况。 玉米不同生育期水分亏缺敏感系数变化范围为 [0.07，0.55]18，不同发育期对水分亏缺反应不同，相 同的水分亏缺造成减产率大小顺序为抽雄吐丝期>拔 节期>苗期19，玉米高产节水试验也得到类似的结论6。 表 1 玉米发育期作物系数(fao-56) 季初 发育 季中 季末 收获 生育期 0.3 0.5 0.7 0.85 1.05 1.2 0.8 0.95 0.55 0.6 0.75 0.9 published by atlantis press copyright: the authors 97 4 适宜的水分亏缺可增加玉米抵御干旱的能力。田间试 验表明，苗期中度调亏、关键期轻度调亏下玉米产量 和水分利用效率最高20，而不适当灌溉甚至可能导致减 产21。玉米耗水量与产量关系的研究表明，春玉米和夏 玉米产量与生育期耗水量呈良好的二次抛物线关系 （r=0.9277和r=0.9479）22,23，表2、3数据二次函数拟 合通过显著性检验（r=0.893和 r=0.834），与该结论 相符。由表2、3可以看出，减产率＞20%、10%－20% 和5%－10%对应春玉米和夏玉米样本水分亏缺率（d） 的平均值分别为50.28%、33.85%、16.79%和49.87%、 32.84%、19.06%。落在d≥40%、25%－40%和10%－ 25%区间内的样本分别占三个减产率区间内所有样本 的91.18%、64.29%、100%和77.78%、81.82%、75.00%， 落 在 对 应 d 区 间 ±2.5% 所 占 比 率 分 别 达 87.80% 、 52.63%、100%和63.64%、73.33%、75.00%。按表2、3 给出的玉米水分亏缺率等级区间，对全部站点、区间 站点分别进行直线拟合（r=0.938和r=0.980），比较 拟合直线与1.25理论线的平行关系，区间站点与理论 线几近平行，表明构建的等级指标具有较好的合理性。 因此，将指标阈值分别设定于10%、25%和40%。 表3 夏玉米全生育期水分亏缺率与减产率的对应关系 减产率/% 站名 年份 亏缺率/% 减产率/% 站名 年份 亏缺率/% 减产率/% 站名 年份 亏缺率/% 减产率/% 5 10 唐河 2001 22.44 9.53 房县 1997 13.62 7.99 唐河 1988 12.63 7.23 西平 1992 31.17 9.16 正阳 1988 19.22 7.24 兖州 1997 27.6 8.76 阆中 1997 14.35 6.13 10 20 郑州 1997 49.57 19.85 通县 2003 39.69 13.79 黄骅 1999 31.03 15.00 景县 1997 35.73 19.51 易县 2003 39.66 13.73 通县 2000 23.78 14.48 静海 1999 50.82 19.48 黄骅 1992 29.45 12.24 唐河 1997 28.19 14.10 三台 2006 28.38 18.47 隰县 1994 35.70 12.14 香河 2000 34.66 11.92 正阳 1994 23.75 11.22 >20 唐河 2001 22.44 9.53 正阳 1988 19.22 7.24 房县 1997 13.62 7.99 西平 1992 31.17 9.16 唐河 1988 12.63 7.23 唐河 1992 22.70 4.30 兖州 1997 27.60 8.76 阆中 1997 14.35 6.13 表2 春玉米全生育期水分亏缺率与减产率的对应关系 减产率/% 站名 年份 亏缺率/% 减产率/% 站名 年份 亏缺率/% 减产率/% 站名 年份 亏缺率/ 减产率/% 5 10 白城 1980 17.53 7.63 沈阳 1972 15.99 6.59 营口 1965 10.69 5.83 白城 1977 16.11 9.57 通辽 1995 14.11 7.40 右玉 1975 13.55 6.83 朝阳 1999 29.33 5.69 通辽 1967 23.47 9.09 右玉 1994 14.08 8.30 东胜 1979 14.16 6.41 围场 1980 20.29 5.54 原平 1970 11.15 8.27 东胜 1989 21 9.05 围场 1981 21.9 6.63 彰武 1968 29.94 5.60 河曲 1970 25.36 8.27 围场 1963 19.82 6.72 彰武 1973 10.11 7.79 黄骅 1983 21.74 5.43 翁牛特 1995 15.13 9.85 彰武 1963 22.48 8.86 泸西 1992 11.24 7.24 西峰镇 2005 12.28 5.30 隰县 2001 22.6 5.04 洛川 2007 15.87 5.74 叶柏寿 1999 29.62 5.69 隰县 1971 16.67 7.65 平凉 2006 13.36 5.57 叶柏寿 1988 19.46 8.42 延安 1986 17.47 6.97 前郭 1973 17.51 6.13 扎贲特 2007 19.19 6.63 延安 2001 13.34 7.75 前郭 1980 11.08 8.42 扎贲特 1977 22.72 7.11 延安 1995 28.73 9.78 前郭 1992 21.26 8.84 扎贲特 1975 16.75 9.98 10 20 安达 1982 33.08 18.07 平凉 2000 36.28 12.72 阜新 1968 31.4 16.33 白城 1968 35.51 12.86 泰来 1975 41.33 16.07 阜新 1981 35.63 12.67 朝阳 1980 30.26 13.23 泰来 1977 34.74 16.80 黄骅 1986 29.92 13.61 大同 1962 37.62 15.48 西峰镇 1969 35.62 19.16 介休 2000 35.15 15.05 右玉 1986 38.77 11.88 西峰镇 2000 29.14 12.38 开原 2000 28.93 16.84 彰武 2001 40.76 18.54 隰县 1994 29.4 12.14 临汾 1986 36.99 15.75 中卫 1976 42.63 16.90 叶柏寿 1961 36.72 16.42 平凉 1991 36.29 19.34 >20 安达 2000 38.3 34.99 泰来 1968 54.53 37.94 阜新 2000 50.97 73.06 安达 2001 41.68 23.37 泰来 1972 56.02 39.77 阜新 2006 47.12 28.66 白城 1972 51.74 62.75 泰来 1982 44.27 31.75 河曲 1962 55.59 51.54 白城 1995 44.58 27.22 泰来 1995 44.11 23.53 河曲 1968 54 46.58 白城 2000 42.49 41.66 泰来 2001 65.41 49.50 河曲 1974 50.65 29.67 白城 2004 58.34 66.68 泰来 2004 60.32 23.89 河曲 2000 52.94 54.47 大同 1972 50.24 24.16 焉耆 1967 76.22 86.47 黄骅 1997 46.4 31.74 大同 1984 43.13 43.75 焉耆 1972 77.26 87.30 洛川 1997 58.78 27.52 榆林 2006 46.23 24.38 延安 1997 39.1 21.49 前郭 1982 44.23 31.82 榆社 1986 40.84 33.73 叶柏寿 1972 40.8 49.65 前郭 2004 38.14 24.62 原平 1968 39.1 46.58 叶柏寿 1981 42.38 29.20 彰武 1982 45.73 37.40 原平 1974 38.66 29.67 叶柏寿 2000 49.77 60.14 彰武 2000 47.96 63.32 published by atlantis press copyright: the authors 98 由表4可以看出春玉米和夏玉米全生育期水分亏 缺率指标阈值相同，播种-拔节、拔节-抽雄、抽雄-成 熟等发育阶段的指标构建也得到了同样的结论。因此， 后文的辅助指标构建即论证将不再分开讨论。 3.2 水分亏缺量等级辅助指标 玉米节水田间试验表明，在中等干旱或者水文年 为75%的年型下，华北地区灌溉2水可以实现经济用水 6,24。由于灌溉定额与当地的土壤类型、水文情况等密 切相关，故本文以灌水次数来表征灌溉量。参考华北 地区的灌溉制度，按灌溉1水为600m3·hm–2，即每单位 面积60mm计算18。 选取玉米生育期有旱情记录的京、津干旱较严重 的年份（27个数据）。灌溉次数通过田间土壤水分平 衡计算得到。参考国家标准《气象干旱等级gb/t 20481–2006》25，季尺度降水距平25%－50%为轻旱， 50%－70%为中旱，＞70%为重旱。研究1982－2007 年的干旱、农田灌溉与减产率的对应关系发现重旱年 份没有充分灌溉时减产率一般大于20%，中旱年份灌2 －3水，轻旱年份灌溉1－2水。按照站点选取原则统计 有灌溉记录的玉米站点如表5。由表5可以看出，水分 亏缺1－2水、2－3水和>3水分别占3个减产率等级站点 的80%、23.81%和75%。不同减产率等级对应的水分 亏缺率平均值分别为55.90%、40.22%和19.62%，落在 上文设定的d等级指标阈值范围内。对应等级区间缺水 量平均值分别为93、150和193mm。综上分析结果确定 水分亏缺量等级辅助指标为轻旱1－2水、中旱2－3水、 重旱＞3水。 3.3 基于实际灌溉的间接验证 为了验证上文得到指标的可行性，选取没有参与 指标构建、灌溉条件好、有灌溉记载的地区的独立样 本，通过计算无灌溉时的玉米生育期水分亏缺率和水 分亏缺量、统计对应年份的实际灌溉量与减产率样本 （表6），进行实际生产验证。 表6给出了玉米全生育期水分亏缺率、灌溉量与减 产率的验证样本，按前文给出的玉米全生育期水分亏 缺率等级指标，分等级统计落在不同灌水次数区间中 的样本数、以及灌溉后增减产的样本数。由表7可以看 出：全生育期水分亏缺率等级为ⅰ级时，灌溉样本数2 个；其中落在灌1－2水区间1个（吻合率50%）、灌2 表4 玉米全生育期水分亏缺率等级指标的回代验证情况 水分亏缺率 （d）/% 减产率 /% 验证样本 d 区间 d 区间±2.5 吻合率 /% 亏缺率 平均值/% 减产率 平均值/% 吻合率 /% 亏缺率 平均值/% 减产率 平均值/% 春玉米 10≤d＜25 5 -10 100 16.79 7.35 100 17.04 7.38 25≤d＜40 10 20 64.29 33.85 16.54 52.63 34.39 18.33 d≥40 ＞20 91.18 50.28 41.00 87.80 48.27 38.51 夏玉米 10≤d＜25 5 10 75.00 19.06 8.52 75.00 19.06 8.52 25≤d＜40 10 20 81.82 32.84 13.53 73.33 31.50 13.55 d≥40 ＞20 77.78 49.87 29.81 63.64 48.02 26.89 表6 玉米全生育期水分亏缺率/量与实际灌溉量、减产率的验证样本 地区 年份 亏缺量/mm 亏缺率/% 灌溉量/mm 减产率/% 地区 年份 亏缺量/mm 亏缺率/% 灌溉量/mm 减产率/% 原阳 1997 315.97 72.43 240 –5.89 三台 2002 185.68 48.90 180 –3.96 夏县 2006 206.86 46.00 180※ 6.48 静海 2007 122.34 33.32 120 0.430 夏县 2005 204.45 47.19 180※ 14.54 静海 1999 186.46 50.82 120 19.480 夏县 2001 192.52 46.45 180※ 11.70 黄骅 1993 113.17 30.45 120 5.85 夏县 2000 183.56 43.47 180※ 7.83 岱岳区 2002 221.10 45.98 120 5.63 隰县 1997 386.77 70.94 180※ 14.17 宝坻 1997 95.61 24.73 60 2.100 宝坻 1992 63.15 17.34 180 –0.530 原阳 1986 191.29 47.74 90 12.75 宝坻 1989 115.23 31.74 180 1.720 虞城 1997 203.85 50.24 60※ 14.88 通县 1997 147.11 38.54 180 –2.17 表5 玉米全生育期水分亏缺量等级回代验证情况 缺水量 /mm 缺水次数 /次 减产率 /% 资料样本 所占比率/% 平均值/mm 最小值/mm 最大值/mm 60 120 1 2 5 10 80 93 51 138 120 180 2 3 10 20 23.81 150 66 206 ＞180 ＞3 ＞20 75 193 148 236 published by atlantis press copyright: the authors 99 －3水区间1个；灌溉后增产的1个、减产在5%以下的1 个；灌溉减灾效果较好。全生育期水分亏缺率等级为 ⅱ级时，灌溉样本数4个；全部落在灌2－3水区间（吻 合率100%）；灌溉后增产的1个、减产在5%以下的2 个、减产在5%以上的1个；灌溉减灾效果总体较好。 全生育期水分亏缺率等级为ⅲ级时，灌溉样本数11个； 其中落在灌1－2水区间2个、灌2－3水区间8个、灌＞3 水区间1个（吻合率9.09%）；灌溉后增产的2个、减产 在5%以上的9个；灌溉减灾效果总体较差。其中1个灌 ＞3水的样本，虽然水分亏缺率高达72.43%，灌溉后在 所有样本中增产幅度最大，达5.89%；由于少灌1水而 造成减产率超过5%的为90.01%，其中减产率超过10% 的达到54.55%，从反面说明指标的可行性（表7）。 综上，全生育期水分亏缺率等级为ⅰ级、ⅱ级、 ⅲ级时，灌水次数分别以1－2水、2－3水、＞3水。综 合实际灌溉样本完全落在指标区间的吻合率、完全落 在指标区间样本实际灌水后玉米增产或减产在5%以 下的验证结果，表明论文构建的玉米水分亏缺率等级 指标是合理的、可行的。 3.4 玉米节水灌溉气象等级指标 采用同样的方法构建玉米播种–拔节、拔节–抽雄 和抽雄–成熟等不同发育阶段节水灌溉气象等级指标， 结果见如表8。表8中玉米全生育期与不同发育阶段的 指标上下阈值并不完全一致，这主要是由于不同阶段 的需水量、水分亏缺敏感指数和“补偿和超补偿效应” 不同造成的26。jensen模型研究表明，玉米拔节前发生 水分亏缺并不直接影响作物籽粒的形成，该阶段的产 量对水分亏缺反应的敏感指数较小，苗期旱后拔节期 复水对产量补偿效应显著；拔节期以后，水分亏缺敏 感指数上升较快，抽雄到灌浆阶段，产量水分亏缺敏 感指数达到最大，且该阶段干旱复水补偿效应较小， 因此随着干旱加剧，减产呈高次函数变化18。水分胁迫 并非全是负效应，前期适量干旱有利于玉米根系深扎 健长，提高后期水肥利用率，只有当水分亏缺率超过 一定限度时，尤其是苗期、拔节期连续干旱时，会导 致严重减产26。因此该发育期水分亏缺率指标值较其他 发育期略大。拔节–抽雄为玉米的水分关键期，因此ⅲ 级指标下限要低于其他发育阶段。全生育期综合了各 发育期的水分敏感系数、需水量等因素，ⅰ级下限与 拔节–抽穗、抽穗–成熟下限相同、ⅲ级下限值在播种– 拔节和拔节–抽穗的下限值之间是符合实际的。 4. 结论与讨论 全球变暖已经成为不争的事实。当温度上升 1－ 4℃时，玉米生长期内需水量将增加 1.7%－18.1%27。 节水灌溉是在水分供求矛盾日益加剧的背景下提出的 与传统丰水灌溉相对的一种经济、可持续的生产方式， 节水条件下玉米需水量可减少13%左右28。节水灌溉气 象等级因子是通过气象、农业、生物等综合因素反映 作物生长期的水分亏缺程度，耦合土壤、作物、气象 信息，以水分亏缺–减产率对应关系反映不同水分亏缺 程度对产量的影响，从作物角度揭示农业干旱。研究 表明，玉米不同生育期不同干旱方式和干旱程度对玉 米生长发育和产量影响不同，不同生育时段对水分亏 缺的敏感度差异很大，导致的减产程度也明显不同3； 前期适度的水分亏缺利于玉米根系深扎稳健生长，利 于提高后期水肥利用率；水分关键期的水分亏缺敏感 指数大，较轻程度的干旱也可能造成作物减产，抽雄乳熟期的旱灾致灾风险明显高于其他生育期29，基于水 分盈亏指数的玉米生育期干旱风险分析也得出了类似 的结论30。 本次研究首次构建了适用于大田生产的玉米不同 发育期和全生育期玉米节水灌溉气象ⅰ－ⅲ级等级指 标。研究遵循科学性、针对性、适用性、可行性的原 则，依据农田水分平衡原理，通过对前人相关研究成 果的综合比较分析，在充分考虑计算因子所涉及的数 据资料的易获取性、计算方法的可操作性、实用性及 普适性等的基础上，确定计算方法，并对其涉及的模 型、指标、因子进行筛选、修正和调整，构建基于农 业、气象、水文等多元信息的玉米不同发育阶段的节 水灌溉气象等级指标。研究结论具有一定的普适性、 空间和时间可比较性，为有效缓解农用水资源的短缺、 表8 玉米节水灌溉气象等级指标 水分亏缺率/% 水分亏缺量（辅助指标) ⅰ级 ⅱ级 ⅲ级 ⅰ级 ⅱ级 ⅲ级 播种-拔节 15≤d＜30 30≤d＜45 d≥45 0.5 水 1 水 >1 水 拔节-抽雄 10≤d＜20 20≤d＜35 d≥35 1 水 1 2 水 ≥2 水 抽雄-成熟 10≤d＜25 25≤d＜40 d≥40 0.5 水 1 水 >1 水 全生育期 10≤d＜25 25≤d＜40 d≥40 1 2 水 2 3 水 >3 水 表 7 基于实际灌溉的玉米水分亏缺率等级指标的回代验证情况 水分亏缺率/% 灌溉 样本数 落在不同灌水次数区间中的样本数 灌溉后增减产的样本数 1～2 水 2～3 水 ＞3 水 增产 减产 5%以下 减产 5%以上 样本数 吻合率 样本数 吻合率 样本数 吻合率 10≤d＜25 2 1 50% 1 0 1 1 0 25≤d＜40 4 0 4 100% 0 1 2 1 d≥40 11 2 8 1 9.09% 2 0 9 published by atlantis press copyright: the authors 100 提高农田水分利用效率、实现作物节水增产提供技术 支撑。论文的指标体系采用作物水分亏缺率，基于该 指标体系，通过对玉米不同发育阶段的需水规律及其 当前时段作物水分亏缺率、亏缺量等的实时分析，结 合未来 3 天的降水的定量预报，评估分析区域节水灌 溉气象等级预报，实现分时段和滚动干旱监测、预警 和响应，为开展实时、有针对性的玉米节水灌溉气象 服务提供科学依据。 实际生产中，农田用水并不完全按照可供水量来 计算，必须考虑用水存在风险，过度开采地下水来满 足作物灌溉的生产方式是不可持续的 31。本次研究结 论从农业生产出发，基于作物 fao–1979 水分–产量关 系函数得到，主要考虑自然供水、人工灌溉对作物产 量的影响。但受资料可获取等因素限制，分发育期指 标构建仍不够精细。研究中未考虑不同地区间土壤类 型、根系发育、品种等差异，在计算作物耗水量时统 一采用了 fao–56 推荐作物系数值；在计算土壤有效 底墒时采用平均状态下的根系发育情况；有效降水文 中采用降水量来计算；fao–1979 产量–水分关系函数 本文取干旱年份下的趋势产量来代替最高产量。此外， 有关结合当地实际灌溉能力、用水风险分析和指标细 化等问题有待下一步研究。 参考文献 1. l. y. hu and y. f. ding, crop cultivation, beijing, higher education press, 2008. 胡立勇, 丁艳锋, 作物栽培学, 北京, 高等教育出版社, 2008. 2. c. y. wang, x. r. lou, j. l. wang, influence of agricultural meteorological disasters on output of crop in china, journal of natural disasters, 16(5)(2007)37–43. 王春乙, 娄秀荣, 王建林, 中国农业气象灾害对作物产 量的影响, 自然灾害学报, 16(5)(2007)37–43. 3. z. g. huo and s. l. wang, agricultural and biological meteorological disasters, beijing, china meteorological press, 2009. 霍治国, 王石立, 农业和生物气象灾害, 北京, 气象出 版社, 2009. 4. s. z. peng and x. l. gao, discussion on improvement of irrigated water use coefficient, china water resources, (1) (2012)33–35. 彭世彰, 高晓丽, 提高灌溉水利用系数的探讨, 中国水 利, (1) (2012)33–35. 5. y. d. du, l. l. song, z. x. liu, an overview on theoretic research on high efficient water use in agriculture, chinese journal of applied ecology, 14(5)(2003)808–812. 杜尧东, 宋丽莉, 刘作新, 农业高效用水理论研究综述, 应用生态学报, 14(5)(2003)808–812. 6. b. z. zhang, g. x. yin, x. zhang, determination of several water–saving irrigation indicators for wheat and corn with high yield in guanzhong irrigated region, bulletin of soil and water conservation, 29(5)(2009)142–145. 张柏治, 殷格侠, 张学, 关中灌区小麦、玉米高产节水 灌溉的几个指标确定, 水土保持通报, 29(5)(2009)142–145. 7. d. s. intrigliolo and j. r. castel, performance of various water stress indicators for prediction of fruit size response to deficit irrigation in plum, agricultural water management, 83(1) (2006)173–180. 8. j. s. wu, s. z. kang, j. l. wang, study on selection and gradation of the evaluation index of comprehensive effects in the water–saving irrigation, journal of irrigation and drainage, 23(5)(2004)17–19. 吴景社, 康绍忠, 王景雷, 节水灌溉综合效应评价指标 的选取与分级研究, 灌溉排水学报, 23(5)(2004)17–19. 9. s. z. kang, w. j. shi, x. t. hu, et al, effects of regulated deficit irrigation on physiological indices and water use efficiency of maize, transactions of the chinese society of agricultural engineering, 14(4) (1998)82–87. 康绍忠, 史文娟, 胡笑涛, 等, 调亏灌溉对玉米生理指 标及水分利用效率的影响, 农业工程学报, 14(4) (1998)82–87. 10. h. j. cai, s. z. kang, z. h. zhang, et al, proper growth stages and deficit degree of crop regulated deficit irrigation, transactions of the chinese society of agricultural engineering, 16(3)(2000)24–27. 蔡焕杰, 康绍忠, 张振华, 等, 作物调亏灌溉的适宜时 间与调亏程度的研究, 农业工程学报, 16(3)(2000)24–27. 11. k. g. wen, p. xie, disasters in china∙beijing, beijing, china meteorological press, 2005. 温克刚, 谢璞, 中国气象灾害大典∙北京卷, 北京, 气象 出版社, 2005. 12. k. g. wen and z. x. wang, disasters in china∙tianjing, beijing, china meteorological press, 2005. 温克刚, 王宗信, 中国气象灾害大典∙天津卷, 北京, 气 象出版社, 2005. 13. s. s. yang, study on the model of grade division of natural disaster and comparison of disastrous conditions, journal of natural disasters, 6(1)(1997)8–13. 杨仕升, 自然灾害等级划分及灾情比较模型探讨, 自 然灾害学报, 6(1)(1997)8–13. 14. j. doorenbos, a. h. kassam, yield response to water, fao irrigation and drainage paper 33, princeton, rome, fao, 1979. 15. z. g. huo and y. jiang, risk analysis of water demand and supply of winter wheat in north china on irrigation , transactions of the chinese society of agricultural engineering, 22(11)(2006)79–84. 霍治国, 姜艳, 基于灌溉的北方冬小麦水分供需风险 研究, 农业工程学报, 22(11)(2006)79–84. 16. r g allen, l spereira, d raes, et al, crop evapotranspiration–guidelines for computing crop water requirements fao irrigation and drainage paper 56, princeton, rome, fao, 1998. 17. g. l. hou and s. b. chen, estimating the maximum yield of main crops in wageningen method in the loess plateau, resources science, 12(4)(1990)30–39． 侯光良, 陈沈斌, 根据瓦赫宁根方法估算黄土高原地 区一些主要作物最大产量, 自然资源, 12(4)(1990)30–39． published by atlantis press copyright: the authors 101 http://www.cqvip.com/qk/90712x/199804/3287974.html http://www.cqvip.com/qk/90712x/199804/3287974.html 18. y. m. chen, g. s. guo, g. x. wang, et al, mai crop water requirement and irrigation of china, beijing, china water & power press, 1995. 陈玉民, 郭国双, 王广兴, 等, 中国主要作物需水量与 灌溉, 北京, 水利电力出版社, 1995. 19. x. l. bai, s. x. sun, g. h. yang, et al, effect of water stress on maize yield during different growing stages, journal of maize sciences, 17(2) (2009)60–63. 白向历, 孙世贤, 杨国航, 等, 不同生育时期水分胁迫 对玉米产量及生长发育的影响, 玉米科学, 17(2) (2009)60–63. 20. m. x. wang, s. z. kang, h. j. cai, et al, study on the mechanism of saving water of regulated deficit irrigation in maize, journal of northwest science technical university of agricultural and forum(natural science edition), 32(12)(2004)87–90. 王密侠, 康绍忠, 蔡焕杰, 等, 玉米调亏灌溉节水调控 机理研究, 西北农林科技大学学报(自然科学版), 32(12)(2004)87–90. 21. s. q. xu, j. song, s. y. ning, spring corn water requirement law and non–full irrigation sprinkler system exploration, saving irrigation, (3) (2002)11–12. 徐淑琴, 宋军, 宁少予, 春玉米需水规律及非充分灌溉 条件下喷灌制度探索, 节水灌溉, (3) (2002)11–12. 22. q. hou, j. g. shen, study on root development characteristics of spring maize and its relationships with canopy growth, act ecological sinica, 21(9) (2001)1536–1542. 侯琼, 沈建国, 春玉米根系发育期特征与冠层关系的 研究, 生态学报, 21(9) (2001)1536–1542. 23. j. s. sun, j. f. xiao, j. y. zhang, et al, yield–water relation and irrigation schedule of high efficient water use for summer maize, irrigation and drainage, 17(3) (1998)17–21.孙景生, 肖俊夫, 张寄阳, 等, 夏玉米产量 与水分关系及其高效用水灌溉制度, 灌溉排水, 17(3) (1998)17–21. 24. z. d. liu, h. z. niu, y. m. jia, et al, study on insufficient irrigation schedule of winter wheat and spring maize under different groundwater depth, water–saving irrigation, (6)(2010)36–41. 刘战东, 牛豪震, 贾云茂, 等, 不同地下水埋深下冬小 麦和春玉米非充分灌溉制度研究, 节水灌溉, (6)(2010)36–41. 25. q. zhang, x. k. zou, f. j. xiao, et al. classification of meteorological drought gb/t20481–2006, beijing, standards press of china.2006. 张强,邹旭恺,肖风劲,等, 气象干旱等级 gb/t20481–2006, 北京, 中国标准出版社, 2006. 26. f. y. li and z. b. huang, a study on compensatory effect on corn at different growth stages to water deficit change, chinese journal of eco–agriculture, 9(3)(2001)61–63. 李凤英, 黄占斌, 夏玉米不同生育阶段干湿变化的补 偿效应研究, 中国生态农业学报, 9(3)(2001)61–63. 27. x. y. liu and e. d. lin, impact of climate change on water requirement of main crops in north china, journal of hydraulic engineering, 35(2)(2004)77–82． 刘晓英, 林而达, 气候变化对华北地区主要作物需水 量的影响, 水利学报, 35(2)(2004)77–82. 28. s. z. peng and c. l. zhu, experimental research on crop water requirement under water–saving irrigation, journal of irrigation and drainage, 22(2) (2003)21–25. 彭世彰, 朱成立, 节水灌溉的作物需水量试验研究, 灌 溉排水学报, 22(2) (2003)21–25. 29. h. c. jia, j. a. wang, d. h. pan, et al, maize drought disaster risk assessment based on epic model, a case sdudy of maize region in northern china, acta geographica sinica, 66(5)(2011)643-652. 贾慧聪, 王静爱, 潘东华, 等, 基于 epic 模型的黄淮海 夏玉米旱灾风险评价, 地理学报, 66(5)(2011)643-652. 30. y. f. zhang, m. t. wang, j. liu, et al, spatio-temporal characteristics of drought at different maize growth stages in sichuan province as determined by water budget index, chinese journal of eco-agriculture, 21(2)(2013)236-242 张玉芳, 王明田, 刘娟, 等, 基于水分盈亏指数的四川 省玉米生育期干旱时空变化特征分析,中国生态农业学 报, 21(2)(2013)236-242. 31. m. li, p. guo, y. h. yin, optimization irrigation procedure and risk analysis of the main crops in minqin country, journal of irrigation and drainage, 32(1)(2013)91-955. 李茉, 郭萍, 付银环, 民勤县主要作物优化灌溉制度制 定及风险分析, 灌溉排水学报, 32(1)(2013)91-955. published by atlantis press copyright: the authors 102 journal of risk analysis and crisis response vol. 9(3); october (2019), pp. 145–148 doi: https://doi.org/10.2991/jracr.k.191024.004; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article evaluation on technology innovation efficiency of big data enterprises based on dea xinpu wang1,2, mu zang1,* 1college of big data application and economics, guizhou university of finance and economics, guiyang 550025, china 2guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang 550025, china 1. introduction data become the basic strategic resources of the country. the platform for action to promote big data development and the big data industry development plan (2016–2020) issued by the state council and the ministry of industry and information technology in 2015 have aroused strong repercussion. “outline” points out that encouraging financial institution to strengthen and improve financial services, increase support for big data enterprises, and continue to enhance the prosperity of the big data industry. the development planning of big data industry (2016–2020) [1] proposes that promoting the development of big data industry is of great significance to improving government governance ability, optimizing public services for people’s livelihood, promoting economic transformation and innovation and development; studying and establishing an evaluation system for the development of big data industry is of great significance to the construction of big data resources, the degree of openness and sharing, the ability of industrial development and the level of application in china and other regions. it is of great significance to monitor, analyze and evaluate the development of the big data industry, compile and publish the development index of the big data industry, and guide and evaluate the development of the national big data. as for the definition of big data industry, foreign scholars have not made a clear distinction between big data industry and big data. laney [2] earlier proposed to use volume, variety and velocity to define big data. later scholars improved the definition of big data from different perspectives. in addition, gaff et al. [3] believes that in the development of the big data industry, the protection of big data privacy is very important, and the government should strengthen the protection. johnson (2012) [4] put forward the formula of “big data + big analytics = big opportunities”, pointing out that the two key driving factors of big data industry are big data and big data technology. sun [5], a domestic scholar, first defined the data industry. he believed that the data industry can be divided into broad sense and narrow sense. the broad sense of the data industry includes the information industry, while the narrow sense of the data industry only refers to the information processing industry and the information service industry. zhong and zhang [6] proposed that the big data industry is an information service industry based on the collection of a large amount of information through the internet, the internet of things, cloud computing and other channels. document issued by guizhou province [7] points out that the big data industry refers to the synthesis of all economic activities related to the generation and agglomeration of big data, organization and management, analysis and discovery, application and service. the “big data industry development plan (2016–2020)” [1] issued by the ministry of industry and information technology defines the big data industry as the activities of data resource construction, development, sale and leasing of big data hardware and software products, and related information technology services. the definition of big data enterprises in this paper refers to guizhou province’s “outline of the development and application planning of guizhou’s big data industry” [7], ministry of industry and information technology’s “development planning of big data industry (2016–2020), state council’s” action plan for promoting a r t i c l e i n f o article history received 28 january 2019 accepted 18 march 2019 keywords dea big data enterprise technological innovation efficiency a b s t r a c t discussing the technological innovation efficiency of big data enterprises and carrying out in-depth research on big data enterprises will help to accurately evaluate the development status of big data industry and guide the development of big data industry. this paper uses dea-bcc model to evaluate the technological innovation efficiency of 21 big data enterprises. it is found that the technological innovation efficiency of big data enterprises is better. the average values of technological efficiency, pure technological efficiency and scale efficiency are 0.587, 0.772 and 0.750, respectively. the technological efficiency of big data enterprises is lower, and the individual differences are larger. at the same time, the research shows that there are 15 big data enterprises need to further expand the scale, and then improve production efficiency. © 2019 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: rim_007@163.com https://doi.org/10.2991/jracr.k.191024.004 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ rim_007%40163.com 146 x. wang and m. zang / journal of risk analysis and crisis response 9(3) 145–148 big data development” [8] and hefei’s definition of big data enterprises [9]: those engaged in big data storage, big data. according to data collection and management, big data analysis and mining, big data presentation and application, traditional industry big data fusion, forming the core independent intellectual property rights of enterprises, and on this basis, enterprises carrying out business activities can be identified as big data enterprises. as for the research on the efficiency of technological innovation, the academia generally believes that dea is a mature and universal method to measure the efficiency of technological innovation. cha and cai [10] think the innovation ability and performance of big data enterprises are the important supporting force to cultivate the core competitiveness of big data industry. to carry out the research on innovation performance evaluation of big data enterprises is of great practical significance for promoting the healthy and sustainable development of china’s big data industry. cai [11] uses dea to measure the comprehensive efficiency, pure technical efficiency and scale efficiency of innovation in the whole big data industry, and further analyses the influencing factors of innovation performance in the big data industry, so as to improve the effectiveness of innovation activities and decision-making basis for big data enterprises and promote the development of big data enterprises. li [12] based on dea-bcc, super-efficiency dea and malmquist index, empirically measures the operating performance of 36 big data enterprises in china from 2013 to 2016. the related research on big data enterprises is conducive to realizing the optimal allocation of big data enterprises’ resources, saving expenditure, thus promoting the vigorous development of big data industry and contributing to the economic development of the country. in summary, scholars have made fruitful achievements in the research of big data industry, and have carried out a lot of empirical research on the evaluation of technological innovation efficiency, and discussed in detail the evaluation index system and evaluation methods, which has great guiding significance. however, these studies rarely conduct in-depth research on big data enterprises, nor do they have literature to measure the technological innovation efficiency of big data enterprises. therefore, this paper uses dea to measure the technological innovation efficiency of listed big data enterprises, which is helpful for further in-depth study of big data enterprises. 2. dea-bcc model dea model is used to evaluate the relative efficiency of decision making units with multi-input and multi-output structure. dea-bcc is based on increasing returns to scale or decreasing returns to scale. in 1984, banker, charnes and cooper established four axioms of convexity, inefficiency, irradiation unrestricted and minimum extrapolation for the possible set of production. the concept of shepherd distance function was introduced to decompose technology efficiency (te) into pure te (pte) and scale efficiency (se), namely: te = pte * se. this paper uses bcc model to measure the technological innovation efficiency of big data enterprises. assuming that there are n production decision making units (dmu), j = 1, 2, …, n, each dmu has m input xj = (x1j, x2j, …, xmj) and s output yj = (y1j, y2j, …, ysj), the efficiency evaluation model of the jth dmu is established by adding restrictions on the weight of l, and il = 1, i = (1, 1, …, 1)1 × i: min . . , , , , q l q l l l s t x x x y i j t j j j t j j j t j ≤ ≥ = ≥ =       = = ∑ ∑ 0 1 0 1 1 0 1 2 …     pure technical efficiency can be obtained from the objective function. scale efficiency can be calculated from te/pte. pte measures the distance between dmu and production frontier when scale reward is variable; se measures the distance between production frontier when scale reward is variable and production frontier when scale reward is constant. 3. empirical analysis 3.1. evaluation index system of technological innovation efficiency and sample data the definition of enterprise technological innovation efficiency is defined as: the utilization ratio of innovation resources invested by enterprises relative to the innovation results of output, reflecting the contribution degree of input resources of technological innovation to output, i.e., the allocation efficiency of technology innovation resources [13]. therefore, the selection of technical innovation efficiency indicators for big data enterprises should take into account the input of innovation resources and the results of innovation output. based on the principles of index system establishment, this paper establishes an index system which includes three input indicators and two output indicators when evaluating the technological innovation efficiency of big data enterprises. the evaluation index system of technological innovation efficiency of big data enterprises is shown in table 1. among them, the input indicators include the amount of research and development (r&d) investment, the number of r&d personnel and the capital r&d investment. the amount of r&d investment includes capitalized r&d investment and expenditure r&d investment. the number of r&d personnel reflects the investment of innovators. capitalized r&d investment is counted into cost of products; output indicators include the number of patent applications for invention and the increase rate of intangible assets. the number of patent applications, including invention patents, utility model patents and design patents, refers to the number of patents that have applied for patents but have not yet obtained patent rights; intangible assets, such as table 1 | evaluation index system of technological innovation efficiency of big data enterprises input r&d investment amount (million yuan) number of r&d (person) capitalized r&d investment (million yuan) output number of patent applications (piece) increase rate of intangible assets (%) x. wang and m. zang / journal of risk analysis and crisis response 9(3) 145–148 147 intellectual property, technology, management, brand, system and innovation ability, measure the promotion and protection of intangible assets to technological innovation. the selection of big data enterprise, referring to the definition mentioned above, and combined with the main business scope published in the annual report, this paper selected 21 big data enterprises to evaluate their technological innovation efficiency. the index data come from the annual report of listed big data enterprises in 2017 and the state intellectual property office. 3.2. the evaluation results of technological innovation efficiency of big data enterprises as part of the index data is 0 or negative, while the dea data need to be all positive, so the data need to be standardized. the standardized formulas for data are as follows: y x x x x = + − − 0 1 0 9. . min max min among them, y is the index value after standardization; xmax and xmin are the maximum and minimum of a certain index respectively. x is the value of an index in a dmu, i.e., the original value. standardized values range from 0 to 1. then using deap version 2.1 software; the dea evaluation values of technological innovation efficiency of big data enterprises are calculated. software calculates the input redundancy value and output deficiency value as shown in table 2. 3.3. efficiency analysis and analysis of input redundancy and output insufficiency technical efficiency is a comprehensive measure and evaluation of resource allocation ability and resource utilization efficiency of dmu. te is the product of pte and scale efficiency. dea efficiency, i.e. technical efficiency reaches 1, is an optimal state in which both pte and scale efficiency reach 1. according to the results of the technical efficiency evaluation in table 1, in 2017, the technical efficiency of haikvision, si-tech, dahua technology, vastdata and goodix technology and technology reached 1, which shows that the input and output of these five big data enterprises are comprehensive and effective, i.e., technology and scale are effective at the same time, accounting for 23.81% of the decision-making unit. on average, in 2017, the average technical efficiency was 0.587. the technical efficiency of zte, ultrapower software, venustech, taiji computer, join-cheer software, thunisoft, navinfo, dawning information, baosight software, nsfocus information technology, beyondsoft and tsinghua tongfang did not reach 0.587, indicating that big data enterprises as a whole have not yet fully realized effective. the technical efficiency of the above 13 big data enterprises needs to be improved. pure technical efficiency reflects the production efficiency of input factors of dmu under certain conditions (optimal scale). pte is influenced by management and technology, and reflects the utilization rate of resources. pte is equal to 1, which means that resources have been fully utilized at the current level of technology; or there is a waste of resources. in 2017, the pte of zte, haikvision, si-tech, dahua technology, vastdata and goodix technology reached 1, table 2 | dea value of technological innovation efficiency and input redundancy and output deficiency big data enterprise name technical efficiency pure technical efficiency scale efficiency scale remuneration input redundancy output insufficiency r&d investment amount (million yuan) number of r&d (person) capitalized r&d investment (million yuan) number of patent applications (piece) increase rate of intangible assets (%) zte 0.318 1.000 0.318 drs 0 0 0 0 0 hikvision 1.000 1.000 1.000 – 0 0 0 0 0 inspur software 0.609 0.703 0.865 irs −63.828 −280.548 −0.042 0 0.236 iflytek 0.228 0.246 0.924 irs −863.205 −4325.445 −0.306 0 0.145 digital china 0.654 0.945 0.691 irs −5.401 −11.393 −0.006 0.021 0 ultrapower software 0.510 0.583 0.875 irs −85.274 −607.460 −0.072 0 0.214 venustech 0.351 0.726 0.484 irs −128.905 387.512 −0.038 0.089 0.341 taiji computer 0.421 0.754 0.558 irs −70.835 −398.185 −0.033 0.050 0.425 join-cheer software 0.531 0.932 0.570 irs −16.671 −113.19 −0.007 0.058 0.354 thunisoft 0.327 0.601 0.543 irs −131.86 −864.661 −0.066 0.049 0.451 si-tech 1.000 1.000 1.000 – 0 0 0 0 0 merit data 0.996 1.000 0.996 irs 0 0 0 0 0 dahua technology 1.000 1.000 1.000 – 0 0 0 0 0 navinfo 0.435 0.466 0.934 irs −487.53 1436.29 −0.114 0.002 0 dawning information 0.320 0.476 0.673 irs −121.637 −520.217 −0.110 0.015 0.444 baosight software 0.462 1.000 0.462 irs 0 0 0 0.134 0.087 nsfocus 0.515 0.683 0.753 irs −91.752 −416.218 −0.046 0.018 0.313 beyondsoft 0.552 0.920 0.600 irs −10.828 −148.101 −0.009 0.024 0.508 tsinghua tongfang 0.093 0.182 0.515 irs −1136.684 −1045.217 −0.451 0 0.728 vastdata 1.000 1.000 1.000 – 0 0 0 0 0 goodix technology 1.000 1.000 1.000 – 0 0 0 0 0 average value 0.587 0.772 0.750 – – – – – – drs, diminishing returns to scale; irs, increasing returns to scale. 148 x. wang and m. zang / journal of risk analysis and crisis response 9(3) 145–148 and the enterprise resources were fully utilized, accounting for 33.33% of the dmu. in 2017, 10 big data enterprises, including inspur software, iflytek, ultrapower software, venustech, taiji computer, thunisoft, navinfo, dawning information, nsfocus information technology and tsinghua tongfang, were lower than the average net technical efficiency value of 0.772. these big data enterprises made less use of resources and needed further to improve utilization level of resources. scale efficiency reflects the production efficiency affected by the scale of dmus. the relative effectiveness of scale efficiency means that there is no difference between the te of a dmu under fixed scale reward and that under variable scale reward, namely scale efficiency is 1. in 2017, the scale efficiency of haikvision, si-tech, dahua technology, vastdata and goodix technology, the five big data enterprises, was 1, and the scale efficiency accounted for 23.81%. the average scale efficiency of 21 big data enterprises is 0.750. among them, zte, venustech, taiji computer, join-cheer software, thunisoft, baosight software and their counterparts have low scale efficiency, which needs further improvement. in 2017, zte showed diminishing returns to scale (drs), indicating that the size of enterprises should be moderately reduced to match input and output; haikvision, si-tech, dahua technology, vast data and goodix technology showed that returns to scale remained unchanged (−), indicating that these five big data enterprises should maintain their current size. at present, the scale of enterprises matches input and output; the other 15 big data enterprises all show increasing returns to scale (irs), which indicates that these 15 big data enterprises should expand their scale moderately to match input and output, so as to improve production efficiency. those 13 big data enterprises have the problems of input redundancy and insufficient output. after data standardization, the input redundancy and output insufficiency rate should be used for analysis. references for calculating input redundancy rate and output deficiency rate [14]. taking navinfo as an example, in 2017, if the amount of r&d input is reduced by 487.52 million yuan, r&d personnel are reduced by about 1437 people, capitalized r&d investment is reduced by 54.3%, while the number of patents applied for invention is increased by 1.67%, and the navinfo can achieve dea validity. there are great differences in the number of r&d personnel and r&d input among big data enterprises. some enterprises have big redundancy in r&d input, which indicates that enterprises may have blind investment, redundancy of personnel and insufficient output, which needs to be paid attention to by enterprises. 4. conclusion this paper uses dea-bcc to measure the technological innovation efficiency of big data enterprises. specifically, firstly, it establishes the index system of input and output, standardizes the index data, and uses deap version 2.1 software to measure the technological efficiency, pte, scale efficiency, scale reward, input redundancy and output deficiency of big data enterprises, and makes analysis. empirical research finds that: big data enterprises are not effective as a whole; there are not many big data enterprises with high technical efficiency and scale efficiency, only 6; pte of big data enterprises is slightly higher than technical efficiency and scale efficiency; from the perspective of scale compensation, most big data enterprises should continue to expand their scale to ensure that the size of enterprises matches input and output. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments this research was financially supported by the regional project of national natural science foundation of china (71861003) and the second batch projects of basic research program (soft science category) in guizhou province in 2017 (foundation of guizhouscience cooperation [2017] 1516-1). references [1] ministry of industry and information technology. big data industry development plan (2016–2020). 2016-1-17. [2] laney d. 3d data management: controlling data volume, velocity and variety. meta group research note, 6. [3] gaff bm, sussman he, geetter j. privacy and big data. ieee comput 2014;47:7–9. [4] johnson je. big data + big analytics = big opportunity. finan exec 2012;28:50–3. [5] sun q. china data industry development report. econ inform 1998;44–9. [6] zhong y, zhang h. the origin, impact and application of big data. modern commun (j china media univ) 2013;104–9. [7] guizhou provincial committee on economy and information. outline of the development and application planning of big data industry in guizhou province (2014–2020). available from: http:// www.cbdio.com/bigdata/2014-11/18/content_1913510.htm. [8] state council. platform for action to promote big data development. 2015-8-31. [9] hefei data resources bureau. in 2018, the identification of big data enterprises in hefei started. available from: http://www. sohu.com/a/253103862_181366. [10] cha h, cai g. research on innovation performance evaluation of big data enterprises in china based on dea method. j beijing univ posts telecommun (soc sci ed) 2017;19:71–8. [11] tea h, cai g. research on innovation performance evaluation of chinese big data enterprises based on dea method. j beijing univ posts telecommun (soc sci ed) 2017;19:71–8. [12] li j. research on the performance evaluation of big data enterprises based on dea. harbin engineering university; 2018. [13] li y. research on technological innovation efficiency of regional industrial enterprises based on network dea. stat decision making 2015;23:85–9. [14] sheng z, zhu q, wu g. dea theory, method and application. beijing: science press; 1996, pp. 22–41, 65–72, 155. microsoft word risk assessment of radio frequencies and public information risk assessment of radio frequencies and public information marc poumadère institut symlog risk governance . 262, rue saint-jacques 75005 paris, france tel: 331 40 46 00 29 e-mail: poumadere@wanadoo.fr anne perrin biomedical research institute of the army crssa, department of radiobiology 24, avenue des maquis du grésivaudan 38700 la tronche, france abstract the rapid, worldwide development of mobile phones is accompanied in many countries by public concern over the possible health risks of radio frequencies (rf). national and international health agencies have undertaken risk assessments, and an in-depth study was rec ently conducted with the french agency on environment and health safety (afsset). this paper builds upon this experience, considering not only risk assessment as it bears upon a physical object (rf), but also concerning activities largely characterized along psychosocial dimensions (mobile phone use). it becomes a theoretical and methodological challenge, however, to integrate these characteristics into the assessment process. the research of causal links between rf exposure and subsequent biological or health effects is the main focus of expology. applied in the afsset assessment, the results are in line with previous reviews of research: biological response appears nonexistent or limited to a few cases, and there is no evidence of rf health effects. considering these results, we specifically look at how information may play a role in the construction of the psychological and social relationship to risk, with a particular emphasis on potential health impacts. in that context, we explore the lesser-known notion of socio-cognitive exposure, char acterized by chronic exposure of populations to potentially worrying information when various health consequences are evoked in the literature and by the media. th is raises the possibility of a link between risk information and health. several specific explanations are explored here, notably: nocebo, stress, and the symmetry rule. keywords: radio frequencies; mobile phone; risks; environment; health; psychological – social; chronic exposure; france; electromagnetic hypersensitivity 1. introduction risks can be positioned along a dimension, ranging from confirmed risks t hat can be objectively controlled, to suspected risks that are more difficult to assess and manage. in the case of rf and mobile phones, with health risks remaining outside quantification, the mere suspicion of risk nonetheless impacts the collective and individual imagination of a very large population. risk perception research has shown how information, more so in most cases than direct personal experience, shapes the individual and collective relationship to risk objects or activities.1 research in social amplification of risk has shown that the attention given to a risk ob ject or activity can be either soc ially amplified or attenuated, regardless sometimes of its dangerous nature as m easured by health or environmental impacts.2 practically speaking, this often means that the scale of the activity in question and the degree of controversy over its perceived riskiness are frequently uncorrelated with the actual magnitude of t he hazard. when there is insu fficient evidence to indicate that a hazard exists, additional scientific data is required, more so to re duce uncertainty than to provide a c lear conclusion. uncertainty, especially as it relates to public health risks, can fuel controversy, as politicians await the reports and recommendations from their health authorities. yet, no matter how essential a step scientific risk assessment may be, it does not provide a ready-made decision for risk managers. meanwhile, studies reveal that part of the public perceives radio signals transmitted by mobile phones and base stations as a potential health risk.3 in this situation it is tempting for politicians to invoke the precautionary principle: policy assumptions at that level are that this approach cannot hurt and will reassure populations. however, this policy assumption ought to be verified: as we sha ll discuss in section 5, precautionary advice can impact risk perceptions. to position the risks of rf within a larger perspective, several levels of environmental risk analysis issues over time can be examined. the development of the chemical and nuclear industries during the second half of the 20th century journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 3-12 published by atlantis press copyright: the authors 3 m. poumadère,, a. perrin presented notable benefits to society, along with a new type of risk: while the probability of accident remained low, the catastrophic consequences of a potential accident were quite high. this presented several challenges in terms of risk assessment and decision-making, and the m andatory disclosure of haz ard information became part of communicating risk t o the public4 subsequent studies into the relationship between environment and health were undertaken, particularly in light of the presence of low levels of dangerous substances in the air, soil or water. this type of pollution can stem from various sources, including industries, transportation, or the agricultural use of pesticides, to name a few. for this sort of risk analysis, the difficulty lies in assessing the health effects of c hronic exposure to l ow doses of potentially dangerous substances, as well as the hazards posed by the combination of low-dose exposure to various substances.5 a third type of risk can be seen with electromagnetic fields as a striking example. though the c haracteristics of rf as a physical agent have been known for more than a cent ury, they were not previously identified as h azardous to human health, even despite public concern over the ea rliest applications of wireless technology at the beginning of th e 20th century. that mobile telephone usage has become so global, with an estimated 5 billion or m ore users, is an un precedented phenomenon which leads us to fra me the question of r isk differently. while the majority of risky objects and activities are assessed through the routine application of t he appropriate processes (which involve the separation of assessment and management), some risks appear embedded in a larger social context (e.g. the sudden omnipresence of mobile phones in everyday life). when this is the case, risk assessment faces additional challenges, and new approaches must be found. the shift from risk gov ernance to the governance of public concern6 illustrates this search for new approaches, but it might overlook the role of information as a mediating variable between a risk issue and health. with the objective of dealing with both the health aspects and the social dimension of rf, this paper presents first an assessment of risk based upon human exposure to rf, along with the guidelines for regulating this exposure. in addition, we consider the fact that populations are also exposed to the massive and rapid penetration of mobile phones in t he private, professional and public spheres, in addition to being exposed to rf as a physical threat. the notion of socio-cognitive exposure is then introduced, along with a discussion of the possible effects of large populations being exposed to worrying information when potential health impacts of r f are reported. certain ca usal relationships, which could possibly be related to symptoms reportedly associated with radio frequencies, are also explored: stress, the nocebo effect, and the symmetry rule. 2. radio frequencies and health in this section, we offer a general overview of the interaction of radiofrequency emitting sources with biological systems, and of the existing guidelines for public safety legislation. we discuss more spec ifically the level of exposure of the population through mobile phones, the state of current scientific knowledge allowing risk assessment, and electromagnetic hypersensitivity. 2.1. radio frequency emitting sources and interaction with biological systems electromagnetic fields are increasingly used for methods of wireless communication, such a s telephones, wi-fi, radiofrequency identification (rfid), etc. usually called radio frequencies (rf), they take place in the non ionizing radiation part of the total electromagnetic spectrum due to the fact that they do not carry enough energy to induce damage within molecules through chemical ionizing reactions. nevertheless, there is an interaction of rf with the matter it encounters; for example, when rf interacts with the human body, the radiation energy is partly absorbed by the body’s tissues. since 1975, this absorption has come to be characterized by t he specific absorption rate (sar), expressed in watt per kilogram (w/kg). this value depends on the rf power received by the matter itself (incident power in watt) related to the power and the distance of t he rf emitting source. the sar also de pends on th e physicochemical characteristics of the ex posed object, as well as its shape, volume and environment. the sar value is not easy to assess, especially for living models (animal, human). it must be calculated by numerical simulation and validated by experimental measurements of electric fields or temperatures of the exposed matter. when the power level of the emitting device is sufficient, interaction between matter and rf can lead to a high sar level (thermal level). a subsequent increase in the temperature of the matter and molecular and/or biological effects might then occur (thermal effects). 2.2. safety guidelines and public exposure in the leg islation, safety reference levels regarding the general public’s exposure to rf are the basis for the different legal and regulatory approaches being pursued in many countries. for the most part, the reference levels are based on the international commission on non-ionizing radiation protection (icnirp) guidelines.7 in the us, the guidelines are established by the institute of electrical and electronics engineers (ieee), following a similar procedure which consists of an ongoing, critical and rigorous survey of scientific literature on the s ubject8 the t hermal effects of radio frequencies are described and serve as a reference for the safety guidelines, which include a safety factor of 50 for published by atlantis press copyright: the authors 4 risk assessment of radio … the general public. the aim is to ensure that human exposure to rf emitting devices does not induce a hazardous thermal sar level even in the case of excessive use, with particular attention paid to m ore vulnerable subpopulations (the elderly, children, etc.). 2.3. radio frequency exposure of the population to mobile telephones in mobile phone technology, the cell phone handset, which is both a receptor and emitter, represents the main source of rf exposure for the general public; the resulting sar int o the head is 10 to 100 thousand times higher than that resulting from base station rf emissions. the environmental exposure level due to the latter can be assessed by measuring the electric component of the electromagnetic field with appropriate probes (expressed in volt per meter: v/m). the electric field must not surpass the limits designated by the current rules regarding mobile phone frequencies, ensuring that the icnirp guidelines will not be exceeded in terms of sar: e.g. in france, 41 v/m (gsm 900 mhz), 58 v/m (gsm 1800 mhz) and 61 v/m (umts 2100 mhz and wi-fi 2400 mhz). in the environment, the ambient values are low, generally below 1 or 2 v/m, and of ten smaller than those resulting from radio emitters. in accordance with the laws of physics, the intensity is greater at distances of 50 to 200 meters from the base stations as opposed to directly under the station, in light of the fact that they are designed to be directional .9 the measurement of the electric field is quite easy to conduct with appropriate probes, but i t is o nly appropriate at a distance from the emission source where the electromagnetic wave is well organized (far field). it is not suitable for the cell phone handset which is used in near field close to the head. in this case, the sar value is the only reference unit. 2.4. scientific research, knowledge and risk assessment in terms of risk policy and management, taking into account the very large number of people exposed, it is clear that even a very small health risk cannot be neglected. the objective of current research is therefore to find answers to the following question: do non thermal effects due to rf exposure exist below the current established limits? for telephones, scientific experts do no t consider that mobile phone base station exposure may be haz ardous for human health, considering the ve ry low exposure level. scientific research is prima rily focused on exposure to cell phone handsets, given that the exposure levels generated are higher and that the emitting source is used close to the head. this research activity became intense in the 1990s, following a lawsuit where cell phone use was blamed in the death of a woman diagnosed with brain cancer. experimental studies performed in vivo and in vitro, as well as epidemiological studies on human populations, have be en carried out worldwide to investigate the effects of r f exposure on bi ological reactions, physiological functions, behaviour and diseases, especially cancers. since then, radiofrequency health risk assessment has been regularly carried out all over the world, giving rise to expert’s reports. the conclusions rely on critical reviews of relevant scientific literature by multidisciplinary panels of co nfirmed experts. an observed effect can be considered established, however, only when it has been described in high quality peer-reviewed scientific papers, and could be replicated, and/or when the results of different complementary studies are consistent with each other. for example, this means that if one in vitro study showed a dna damaging effect, it would not necessarily constitute a h azard (especially if the effect is weak) if no effect of exposure in animal or on human health is found. furthermore, it should also be noted that a biological effect is not necessarily related to an adverse (or beneficial) health effect. in france, following reports edited in 2 000 and 20 05, afsset’s report of october 2009 deals with all studies published between january 2005 and april 2009, including 12 reports produced during this period in d ifferent countries10 taking into account the anterior status of the knowledge, the general conclusion was that “the data from experimental research thus far indicates neither short nor long-term health effects. the same goes for epidemiology concerns: no short-term effects have been discovered, though questions do remain over the potential effects of longer-term exposure, even though no biological mechanism suggests such effects”. the r eport also mentioned that decreasing the ambient rf exposure limits to the often cited 0.6 v/m value was not scientifically justified. since 2009, the interphone was published.11 it is the largest international epidemiological study conducted to date, focusing on brain cancers – i.e. glioma, meningioma, accoustic neuroma – and individual mobile phone use. it remains inconclusive without providing proof of an increased risk. several reports have also been published by other groups, such as, for instance, the icnirp12, the executive agency for health and consumers13, and the swedish independent expert group on electromagnetic fields .14 in examining the existing data, no causal relationship between low level rf exposure and adverse health effects can be established. there is a large consensus that the risk, if existant, would be very small. nonetheless, debates over the possibility of long term effects continue. more generally, a recent who summary on the health effects of rf exposure corroborates this current state of the available knowledge .15 2.5. electromagnetic hypersensitivity (ehs) it has been estimated to affect anywhere from 1 to 10 % o f the population, and exhibit quite noticeable geographical published by atlantis press copyright: the authors 5 m. poumadère,, a. perrin variability in terms of pr evalence and the reporting of relevant symptoms. individuals with ehs suffer from a variety of non specific and common health symptoms self-attributed to rf. some of them are so severely affected that they must stop working and may become withdrawn from society. of all groups, women and mi ddle-aged individuals with a higher education are the most affected. a set of convergent associated signs suggests that individual neuro-psychic factors may play a part in this condition.16 a number of stu dies were carried out to inv estigate whether ehs volunteers are able to distinguish rf exposure compared to sim ulated exposure (sham) and if they have a different perception of rf exposure when compared with control subjects. the results showed no evidence for a causal relationship between these symptoms and rf exposure in double blind experiments, and several studies have moreover suggested the existence of a noc ebo effect and somatisation .17,18 similar symptoms were reported for ca ses of m ultiple chemical sensitivity (mcs), sick building syndrome (sbs), chronic fatigue syndrome and gulf war syndrome. for each of these, a more ge neral applicable name, which describes sensitivity to environmental factors, is referred to as idiopathic environmental intolerance (iei) by the who .19 3. social concerns and risk assessment of mobile phones in several countries around the world, the use of mo bile phones has become a controversial issue. some argue that a significant and continuously increasing number of people worldwide are threatened by the biological effects of radio frequencies as they may negatively impact one’s health. in light of this potential threat and the scientific uncertainty associated with attempts to assess the actual risk, the situation warrants action that would lower rf emissions and reduce public exposure to such hazards, in particular for children, in application of the precautionary principle. the precautionary principle, which entered the french constitution in 2005, stipulates: “when the realization of a damage, although uncertain on the basis of available scientific knowledge, could affect the environment severely and irreversibly, public authorities insure that, by application of the precautionary principle and within their attributed domains, proper risk assessment is undertaken and that temporary and proportionate measures are adopted so as to prevent the realization of the damage.” from this excerpt, it is clear that the e valuation of risks represents an important application of the precautionary principle as prescribed in the french constitution. in fact, it could even be argued that risk evaluation is indeed the first priority, even before considering the eventual application of protections or regulatory measures with which the precaution principle is more typically associated. risk assessment entails the identification of dangers to health resulting from exposure to an agent identified as a risk factor, and aims at measuring the magnitude and likelihood of a negative impact on health or the environment. in order to prevent an evaluation from being influenced or pre determined by a political power, economic issues, or the social amplification of risk, an established principle in riskanalysis consists of sepa rating the evaluation of risk from the way in which this information is communicated and t he risk is managed. this separation was not always the norm, but became more standard after developments towards the end of the 1970s. dividing up the different components of ri sk-analysis (i.e. assessment, management, and communication) allows for a m ore thorough, systematic process and thus facilitates more careful analysis of t he various issues at play, most notably between science and po litics. this does no t exclude the possibility of a pluralist approach, based on a n interdisciplinary or tra nsversal methodology with the participation of associated representatives and the actors who are directly involved. members of the local population at the source of the “signal” can be included, taking into account their relationship to the environment. in stu dying these subjects, it is important to take an approach sensitive to the linguistic and interpersonal dimensions of their expression. if the thre e steps of risk analysis appear to be formally defined and separated, the governance of risk grants those directly affected a form of r ecourse in deliberation. in france, afsset is situated at the intersection of science and policy, as on the one ha nd it is in charge of organizing collective and multidisciplinary scientific expertise of ris k and promoting areas of research, and on th e other han d it provides policy advice. this process is specifically appropriate for objects or activities marked by complexity, uncertainty, and ambiguity.20 it can be equally useful when several agents and substances are evaluated simultaneously, taking into account their possible combined effects on the environment, though the formal method of risk analysis was initially developed to treat one risk at a time. the aim here is to attain a higher level of efficacy, whether via an evaluation of risks t hrough pluralist scientific assessments or throu gh management and communication with different forms of collective deliberation, given the reality of complementary and otherwise not easily accessible resources. the efficacy depends on a commitment by those persons potentially affected to cooperate in co nfronting the complexity, the incertitude, and the ambiguity of risk. this type of involvement with stake holders, based on an organized and collective initiative centered on their own concerns, would require surpassing the existing polarization between opponents and investigators, who tend to take a reductionist stance on the issues at hand, and transform the debates into trials and spectacles attracting media attention. within the context of france and other european union countries, there exists a high level of public concern about published by atlantis press copyright: the authors 6 risk assessment of radio … the health risk of rad io frequencies, unless there are other risks which are considered more pressing and which have yet to be dealt with, such as chemicals, foods or drinking water, as measured by risk perception surveys.21 the risk of mobile phones has become a topic of controversy, with each side of the debate featuring strongly opposing points of view. such controversy is neither surprising, nor a r ecent phenomenon: the dangers of w ireless technology were acknowledged almost one century ago, as documented in an article published by the new york times in 1914.22 the ar ticle, “persistent theory of wireless peril”, reports on french newspaper articles suggesting the capacity of hertzian waves to ignite gun powder and cause an explosion at a coal mine, a notion strongly rejected by the scientists of those days. from very early on, fierce debate and controversy plagued the fields of risk assessme nt and risk management. in 1984, no less than 162 controversies appear to be associated with technological risks.23 it should be noted, however, that controversies are not unique to the risk domain and may potentially arise whenever there exist opposing points of view or interests. controversy provides a traditional base for media coverage and, more importantly, attracts the attention of the public, regardless of any ac tual correlation to t he degree of risk of a particular potential hazard. conversely, the absence of controversy does not mean the absence of risk, or the absence of severe impacts in terms of public health. some acute th reats, while being scientifically identified as such, can be socially minimized, as wa s the c ase, for example, with the heat waves of 2003, which caused some 20,000 deaths in france.24 in other w ords, social controversies are not indicative of the nature or severity of a risk. this poses a political dilemma: while attention to the public’s concerns is surely important, public health must also be managed according to objective criteria like mortality and morbidity. this nec essitates a carefully balanced approach that prioritizes which risks a re most pressing, a task best performed by a neutral entity like the government. it is important to di fferentiate assessments dealing with risks associated with future projects from those being applied to existing equipments, such as the c ase of wireles s communication systems. in the case of the former, the analysis can be performed in ta ndem with public hearings and inquiries preceding the realization of technical projects, which allows for the consideration of i ssues such as acceptable risk and choices (both of technology and society) and associated value judgments. in the latter case, populations can feel confronted with an invasive technology and its possible risks without having previously participated in public hearings and debates. mobile phone technology falls into this second category. the implementation of network coverage over an entire country was framed, in most countries, as a public good, with no associated health risk. in france, it was shown that public ire was mainly directed against network antennas and or iginated in local protest movements in the absence of a ny proven adverse health effects.25 when, after a risk c ontroversy, public concern comes to be a major issue, the scientific assessment can play a social role of retroactive adjustment: it can contribute indirectly to the social debate and, in some cases, facilitate social appropriation of the new technology at hand. nevertheless, the sit uation is paradoxical because while mobile telecommunication has sp read rapidly around the world, illustrated by massive increases in public subscriptions to this technology, public concern over possible associated health risks has simultaneously increased in certain countries. it is to be note d that people expressing greater concerns are among those estimating they are sufficiently informed.26 4. characterization of mobile phones as a social object social science research in the risk domain is based on three distinct levels. the first deals with methods and concepts allowing for the definition and im plementation of ris k management procedures and is linked to decision analysis and policy. the sec ond level concerns the orga nization of collective deliberation and the structuring of debate in such a way as to incorporate the relevant stakeholders into the process of risk governance. the third level is the characterization of ob jects and sit uations associated with risk, in particular through studying the social and psychological aspects of risk, thus contributing to risk assessment. within our approach, we consider psychosocial repercussions not only to be a consequence of the uncertain nature of the hazards of rf, but also as just one of the factors which could have an impact on health. this could occur in particular through public exposure to t he large amount of information suggesting health damage associated with wireless technologies, especially antennas and cell phone handsets. the massive and r apid worldwide diffusion of m obile phones appears to be an unprecedented technological event in human history.27 that these mobile networks and the ir physical installations exist quite literally everywhere has a profound impact on society and has changed the lifestyles of the populations affected. this mere fact is rarely considered in the relevant analyses and debates, even though it is probably a key ca use of significant observed cultural changes around the world. studies on mobile phone diffusion often focus on the main advantages of the technology, which explain its widespread success: users are able to communicate at a distance (which was not previously the case for a large part of the global population), they can access information, and they can be reached anywhere and at any moment. among the risks some times acknowledged, cell phone use wh ile driving is bes t known, although researchers often qualify published by atlantis press copyright: the authors 7 m. poumadère,, a. perrin their concerns by emphasizing the simultaneous benefits of having a mobile phone in the car in case of an accident or emergency. regarding the health effects of human exposure to th e electromagnetic fields inherent to m obile phones, this technology can be con sidered a double risk: an i ndividual risk associated with the handset that can be controlled; and a collective, or environmental, risk associated with the base stations that no one individual can control and to which everyone is subje cted. in th is context, cell phone handsets and base stations give rise to a paradoxical management situation with ample opportunity for confusion in the general population. subsequent to their massive diffusion, radio frequencies have become the object of numerous scientific studies. as a conse quence, one can also note the para llel development of worldwide mobile phone networks and the publication of studies related to health risk. what conclusions can we draw from this correlation? scientific attempts to investigate radio frequencies as a physical agent are nothing new. however, it does appear that the growing number of studies on the biological effects of rf are linked to the development of mobile telephones, not only as a technology, but above all as a widely diffused social object. we can therefore see these publications as a form of response to the worries and concerns of affected populations. this highlights the soc ial function of the scientific research and analysis, and draws attention to the fact that it is not limited to scientific issues. one byproduct of this research is for its part the uneven political treatment of risk and of its assessment around the world. for example, in some places, local p opulations recognize the he alth concerns of wind energy installations (noise, stroboscopic effect, infrasound). however, due to the exposed populations not being large enough (in contrast with populations exposed to radio frequencies), and in the absence of a significant social movement, no coordinated action is undertaken. with regards to the content of studies on radio frequencies, table 1 give s an ide a of the objects studied when the biological effects are researched(first column), and when epidemiological studies are conducted (second column). the third column presents the symptoms most commonly attributed to electromagnetic fields by individuals experiencing them; 57 symptoms have been listed to date.28 table 1. biological responses searched, epidemiology and symptoms attributed to radio frequencies biological and clinical effects (radio frequencies) epidemiology (antennas, mobile phone handset) attributed symptoms (exposure to electromagnetic fields) • genetic expression and protein synthesis • oxidative stress and production of free radicals • genetic material (dna) • apoptosis • development of cancer in the animal • immunological system • nervous system • development • reproduction • hearing • ocular system • cardiovascular system • melatonin • cellular proliferation • erythropoiesis in the rat • stem cell precursors in the mouse • ornithine decarboxylase (odc) activity • endocytosis • mitosis • aggregates of cancer cases (antennas) • temporal trends of impact and of mortalities from brain tumors • brain tumors linked to proximity of dect base stations • glial tumors • meningiomas • acoustic neuroma • parotid tumors • pituitary gland tumors • testicular cancer • breast cancer in men • ocular melanoma • cutaneous symptoms (face) • fatigue • palpitation • headaches • difficulties concentrating • sleep disorders • nervousness • osteomuscular pain • respiratory disorders • balance disorders • tinnitus • hearing disorders • vision disorders • cutaneous signs (general) • numbness in head • memory disorders • irritation • agitation • intestinal spasms • restless legs published by atlantis press copyright: the authors 8 risk assessment of radio … while these lists are not exhaustive, they nonetheless show the com plexity of the topic and of the various approaches taken. the question remains how the results of scientific studies can become informative content to be broadcast by the media and brought to the attention of the public and how, in com bination with other social representations, they can b ecome issues of concern. research has shown that both the media and the public devote their attention more frequently to the information related to the absence of established causality and to the unlikelihood of the health risk. this also appears to be the case for radio frequencies. moreover, risk information is not limited to input from research; there is indeed a wide array of anec dotal accounts on m obile phones in the media as w ell as reports from activists opposed to this technology. 5. links between risk information and health in this section, we will confront the question of a possible link between health and ris k information concerning radiofrequency technology, and in particular the mobile phone, by considering the various levels of information available to the public. the first level of information on a risk is provided by the actual presence of physical objects in the hum an environment. a study on this subject deals with radon and high voltage lines, as t wo cases of radiat ion risk which produced various levels of atte ntion among local populations and experts.29 considering their similarities, these cases and that of rf might be lumped together, qualifying this grouping with the fact that the hazardous nature of radon was scientifically confirmed. the stu dy shows that public attitudes towards power lines differ according to the level of estimated “exposure”. this exposure was defined in this context in terms of spat ial proximity to high tension lines more or less visible to a community. the researchers involved in the study concluded that the results were based on the visual information used by the “ exposed” subjects, their awareness of the threat due to elec tromagnetic fields being measured by the following question: “can you see the high tension lines from one or seve ral windows of your house?” this observation regarding the e ffects of visual exposure is int eresting in that it reveals that concerns about and occasional complete rejection of t he radiation emitting source begins their visual identification. the situation might be similar for mobile phone base stations as self declared base station neighbors were found be more strained than others. the notion of exposure in the domain of risk first relates to the approaches used in toxicology or epidemiology to look for dose-effect relation. taken in another sense, the notion of exposure allows the effec ts linked to the proximity of the antennas to be taken into consideration when they are stigmatized as potentially harmful to health. during an audition for the afsset risk assessment10 , an expert from the u.k., l. challis, reported that a) 90% of the concerns in five different cities of united kingdom related to antennas and b) general practitioners mentioned an association between the fear of the base station and the production of symptoms: “we also talked to doctors and they said that fears about base stations could produce real physical symptoms. we cannot ignore these fears and the physical effects they can lead to”. though dependent on t he accounts of do ctors, this evaluation remains imprecise and the idea that fear may possibly impact health must be considered hypothetical. as a “ signal” necessitating further investigation, these reactions raise questions about the mechanisms associated with the production of symptoms. in the present paper, the etiological aspect of the clinical symptoms is disc ussed relative to t heir position at t he intersection between human beings, with cognitive and physiologic components, and their environment, in the broadest sense of the word. the underlying biological mechanism, poorly documented to date, will not be addressed. we propose to explore three distinct categories of concepts: nocebo, stress, a nd the symmetry rule. the nocebo effect describes the negative side of the pla cebo effect, where a m edication or a neutral medical action produces symptoms (adverse to h uman health in this case). as previously mentioned several studies have shown this effect where symptoms appeared when the ehs subjects were submitted to simulated exposure but believed to be exposed to radio frequencies. the effect of stress could also apply if we consider that the network antennas present in the environment, along with the disturbing stigma and controversies which surround them, constitute a chronic stressor. various symptoms of stress could then also be p otentially associated with the prevalence of the se antennas as a cause. in contrast to these two relatively established concepts, attempts to demonstrate the l ink between cognition (in the larger sense of risk inf ormation being received and processed by individuals) and health are more recent. it has been found in studies concerning the rule of s ymmetry that such a link does indeed exist. briefly, this finding can be presented with the help of a case involving transgenic corn in the united states.31 an association of consumers alerted the public that traces of transgenic corn were present in food products like taco shells, while the implicated industry maintained that transgenic corn was only used to feed livestock. the news of this incident attracted the attention of the media, and published by atlantis press copyright: the authors 9 m. poumadère,, a. perrin was quickly diffused. the center for disease control (cdc) received 51 cases of persons displaying symptoms (moderate: weakness, dizziness; or se rious: fainting, symptoms requiring hospitalization), which they attributed to their exposure to transgenic corn. the cdc subsequently conducted a thorough investigation of these cases through rigorous interviews and serological tests. the results 32 revealed that none of these persons had been in fact exposed to the transgenic corn, and that an allergic reaction linked to the exposure could not explain the reported symptoms. two possibilities, not mutually exclusive, were used to ex plain this phenomenon. the first suggested that the persons who experienced an unexplained symptom prior to the alert were able to find a satisfying explanation by assuming that they were victims of transgenic corn. according to the second explanation, the warning initially disseminated by the media was itself able to incite a heightened vigilance in those who believed that they had consumed the transgenic corn and those who were thought to be thus “exposed”. these two theoretical explanations illustrate the symmetry rule, ac cording to which awareness of the disease depends on the duality of representations in memory: one concrete, and the other abstract. the mental construction of a state of hea lth includes an abstract representation (for example, having consumed food products containing transgenic corn), and concrete symptoms (for e xample nausea, weakness), knowing that the presence of one of these representations could lead to the other. thus, “persons searching for and finding representations to explain their symptoms, and searching for and finding symptoms to render their representation of the disease concrete.” 33 extrapolating from the analogy above, the transposition of t he symmetry rule to symptoms associated with radio frequencies would require additional studies. first, it would be necessary to specify which symptoms could be attributed to visual exposure to antennas, and which symptoms could be attributed to socio-cognitive exposure (the latter referring to receiving worrisome information about the risks of radi o frequencies) and also to what extent these two exposure levels are or are not linked. another dimension of the exposure to information would also need to be evaluated: that of presenting general advice on prudence regardless of a pr oven health hazard. in the case of the mobile phones, precautionary measures are often used as a means of responding to the pre occupations of t he public; however, some studies explore how these measures could themselves cause or e xacerbate such worries. an experimental study conducted in g ermany34, 35 allowed the authors to conclude that these precautionary measures could arouse concern and amplify the perception of risk. in order to ascertain the impact of the information on the potential risks of m obile phones that is diffused by t he government of the united kingdom, researchers employ focus groups (these grou ps offer a diverse range of individuals, with different backgrounds, ages, interests, levels of sensitivity, types of preoccupations, propensities to use a m obile phone, etc.). the results show that the comprehension of the m essages does appe ar to be principally rooted in prec aution, and tha t the precautionary advice is generally perceived as worrying rather than reassuring information.36 another study applied as well to mobile phones in denmark (n= 1 687) shows that information can increase concern in a lar ge part of t he population.37 these results, obtained through different methodologies and in several countries, converge and reveal the active role of population in their interpretation of risk in formation. in this context is introduced the new iarc classification38 of rf in the 2b category ("possibly carcinogenic for humans" ), in relation to an increased risk for glioma, a malignant type of brain cancer, associated with wireless phone use. the effect of this information upon risk perception of rf is uncertain and ought to be studied. finally, it must be said that these three concepts (nocebo, stress and the symmetry rule) are not exclusive and can overlap. although they cannot fully explain the complex underlying physiological mechanisms, they allow us to think about a transversal approach. research hypotheses relevant to any discussion of the link between information and health in the case of mobile phones need to be explored even before taking into account the many methodological and theoretical difficulties. the notion of socio-cognitive exposure also dese rves serious consideration, particularly so as to more precisely define the information in its individual and collective dimensions. indeed, in this complex context, the information simultaneously comprehends the var ious content, be it technical or symbolic, as w ell as the processes and interactions within society that contribute to the construction of mobile phones as a social object. properly addressing this issue would require an interdisciplinary approach involving analysis in both the social and biomedical domains. 6. conclusion expanding on the assessment of radiofrequency health risks, several points have been analyzed in consideration of the relationship between risk information and health. in sum, the relationship remains paradoxical: despite many studies that have failed to prove the existence of adverse health consequences, concerns over their potential danger persist even as cell phone handsets continue to se ll well and new wireless communication networks continue to grow. published by atlantis press copyright: the authors 10 risk assessment of radio … the first element of th is somewhat counterintuitive reality is the d ifficulty of communicating the absence of an established health risk w hen the opposite belief prevails in a society. this contrasts with the past roughly forty years, which were characterized by difficulty in managing and communicating to the public the threat of an accidental or chronic risk even when it was c learly established in various contexts. this ne w era, w hich began with the dawn of mobile phones, is defined in many countries by a general awareness and attentiveness to the possibility of health risks. made r elevant by mobile phones, it would need to be verified if this phenomenon occurs with other objects also characterized by sim ilar technical and symbolic dimensions (potential social objects), as would be the case with the largely invisible, yet ever-present nanotechnologies. it was interesting to no te that since the 1 990s the number of s cientific papers on mobile telephones’ biological and health effects closely follows the exponential development of this technology during this period. furthermore, social considerations were responsible for the ini tiation of the first st udies and have also been used, alongside scientific arguments, to justify the need for further studies (and related fund). the social role of sc ientific research is evi dent when risk assessments are repeated, though scientific knowledge is for the most part alre ady established. this social role ought to be better understood. notably, it can contribute indirectly to the appropriation of a t echnology by the population since it takes into consideration concerns and social mobilizations. however, controversies and public mobilization are not indicative of the nature or severity of risk, and public health efficiency relies upon mortality and morbidity indicators. the public is not only exposed to the physical agent constituted by radio frequencies, but also to the mobile phone as a social object whose prevalence throughout society has been rapidly imposed and ha s become normal in most professional and private contexts. in that sense, a special "a wareness" of the presence of mobile phone base stations has set in. aside from the findings of the many research studies and collective reports looking for bio logical and hea lth effects, including studies regarding the symptoms and various phenomena attributed to exposure to radio frequencies, another level of exposure is left to consider. socio-cognitive exposure is manifested by t he quantity and content of risk information provided to large parts of the population. the question of the social importance and health impact of such w orrying information has been investigated and se veral concepts have been discussed: nocebo, stress, symmetry rule. from the perspective of their concrete integration into risk assessment, the notions explored here should be further studied in expology and through interdisciplinary research, in pa rticular to better understand the individual and collective dimensions of the link between risk information and health. acknowledgements: as members of the french agency for health and environmental safety (afsset) working group, the authors are indebted to the work setting provided by this institution and to the exchanges with the other experts. the points of view expressed in this article are those of the authors and involve their sole responsibility. the full expertise report is available on l ine: htpp//www. afsset. fr/upload/bibliotheque/049737858004877833136703438 564/ rapport_rf_final_25_091109_web.pdf. on line references were last consulted in october 2011. references: 1. p. slovic, the perception of risk. (earthscan publication ltd, london, 2000). 2. n. pidgeon, r. kasperson, & p. slovic (eds.), risk communication and social amplification of risk. (cambridge: cambridge university press, 2003) 3. siegrist, m., earle, t. c., gutscher, h . and keller, c., perception of mobile phone and base station risks. risk analysis,25(2005):1253–1264.doi:10.1111/j.1539-6924.2 005.00672.x. 4. m. baram, the right to know and the duty to disclose hazard information, american journal of public health, 74(4) (1984): 385-390. 5. j. l. greeno and j. s. wilson, new frontiers in environmental, health and safety management, in r. kolluru, s. bartell, r. pitblado and s. stricoff (eds.) risk assessment and management handbook for environmental, health, and safety professionals. (mcgraw-hill, new-york, 1995). 6. l. kheifets, j. swanson, s. kandel, and t. f. malloy, risk governance f or mobile phones, power lines, and other emf technologies. risk analysis, 30(2010): 1481–1494. doi: 10.1111/j.1539-6924.201 0.01467.x. 7. icnirp, icnirp statement on the guidelines for limiting exposure to time varying electric, magnetic and electromagnetic fields (up to 300 ghz), icnirp statement(2009)http://www.icnirp.org/documents/state mentemf.pdf. 8. institute of electrical and electronics engineer, ieee standard for safety levels with respect to human exposure to radio frequency electromagnetic field, 3 khz to 30 0 ghz. s.l. : ieee std c95.1(2005). 9. j.f. viel, e. cardis, m. moissonnier, r. de seze and m. hours, radiofrequency exposure in the french general population: band, time, location and activity variabi lity, environ int, 35(8) (2009). 10. affset, mise à jour de l’expertise relative aux radiofréquences(2009)www.afsset.fr/upload/bibliothequ e/049737858004877833136703438564/rapport_rf_fina l_25_091109_web.pdf. 11. the interphone study group, brain tumour risk i n relation to mobile telephone use: results of the published by atlantis press copyright: the authors 11 m. poumadère,, a. perrin interphone international case–control study, int j epid, (2010). 12. icnirp, “blue book” exposure to high frequency electromagnetic fields, biological effects and health consequences (100 khz-300 ghz) review of the scientific evidence and health consequences, (icnirp , munich, 2009). 13. executive agency for health and consumers, promoting healthy environment with a focus on the impact of actions on electromagnetic fields, final report august 2010 (2010). 14. ssms independent expert group on electromagnetic fields for swedish radiation safety authority, seventh annual report, recent research on emf and health risk (2010). 15. who, electromagnetic fields and public health: mobile phones, fact sheet (193) (2011), http://www.who.int/ mediacentre/factsheets/fs193/en/index.html. 16. j.-p marc-vergnes, electromagnetic hypersensitivity: the opinion of an observer neurologist, c.r. physique, 11 (2010). 17. g. j. rubin, r. nieto-hernandez and s. wessely, idiopathic environmental intolerance attributed to electromagnetic fields (formerly 'electromagnetic hypersensitivity'): an updated systematic review of provocation studies, bioelectromagnetics, 31(1) (2010). 18. r. szemerszky, f. köteles, r. lih i and g. bárdosa, polluted places o r polluted minds? an experimental sham-exposure study on background psychological factors of symptom formation in “idiophatic environmental intolerance attributed to electromagnetic fields”, int j hyg envir heal, 213 (2010). 19. who, electromagnetic hypersensitivity, fact sheet n°296.(2005)http://www.who.int/mediacentre/factsheets/ fs296/en/index. 20. o. renn, risk governance: coping with uncertainty in a complex world. (earthscan, london, 2008). 21. european commission. electromagnetic fields, ec, brussels, (2007), available on li ne: available on line: http://ec.europa.eu/public_opinion/.../ebs_272a_en.pdf. 22. new york times, persistent theory of wireless (peril french press attributes many explosions to hertzian waves, may 10, 1914). 23. d. von winterfeldt, w. edwards, patterns of conflict about risky technologies, risk analysis, 4 (1) (1984): 55-68. 24. m. poumadère, c. mays, s. le mer and r. blon g, the 2003 heat wave in france: dangerous climate change here and now, risk analysis, 25(6) (2005): 1483-1494. 25. o. borraz , m. devigne and d. salomon et, controversy and protest around mobile phone antennas in france, in c. del pozo, d. papameletiou, p. wiedemann, and p. ravazzani (dir). risk perception and risk communication: tools, experiences and strategies in electromagnetic fields exposure, (rome, consiglio nationale delle ricerche (cnr), 2006) 94-106. 26. european commission. special eurobarometer: electro magnetic fields, ec,( brussels, 2010), available on line:http://ec.europa.eu/public_opinion/archives/ebs/ebs_ 347 _fr.pdf. 27. oecd (2008) oecd in numbers year 2008. paris : oecd editions. 28. s. eltiti, d. wallace, k. zougkou et al., development and evaluation of the electromagnetic hypersensitivity que stionnaire, bioelectromagnetics, 28(2) (2007): 137-51. 29. w. poortinga, p. cox, n. pidgeon, the perceived health risks of indoor radon gas and overhead powerlines: a comparative multilevel approach, risk analysis, 28(1) (2008): 235-248. 30. c. augner and gw. hacker, are people living next to mobile phone base stations more strained? relationship of health concerns, self-estimated distance to base station , and psychological parameters. indian j occup environ med. 13(3) (2009): 141-145. 31. s. brownlee, h. leventhal, leventhal e.a. regulation, self-regulation, and construction of the s elf in the maintenance of physical health. in:m. boekartz, p.r. pintrick, m. zeidner, handbook of self-regulation. research, theory and applications, (san diego, ca: elsevier academy press, 2000): 369-93. 32. cdc. investigation of human health eff ects associated with potential exposure to genetically modified corn. center for disease control: report to the u.s. food and drug administration, (2001) 24. 33. n. t. brewer, w. k. hallman , h. m. kipen, the symmetry rule: a s even-year study of symptoms and explanatory labels among gulf war veterans, risk analysis, 28(6) (2008): 1737-48. 34. p. m. wiedemann, h. schütz, the precautionary principle and risk perception: experimental studies in the emf area. environ health perspect, 113(4) (2005): 402-5. 35. p. m. wiedemann, h. schütz, m. clauberg, influence of information about specific absorption rate(sar) upon customers' purchase decisions and safety evaluation of mobile phones, bioelectromagnetics, 29(2) (2008): 133-44. 36. j. barnett, l. timotijevic, r. shepherd et al ., public responses to precautionary information from the department of health (uk) about possible heal th risks from mobile phones, health policy, 82(2) (2007): 240-50. 37. j. b. nielsen, a. elstein, d. gyrd-hansen, h.w. kildemoes, i.s. kristiansen, h. stovring, effects of alternative styles of risk information on emf risk perception, bioelectromagnetics, 31(2010) 504-512. 38. iarc, press release n° 208, may 2011 "iarc classifies radiofrequency electromagnetic fields as po ssible carcinogenicto humans". http://www.iarc. fr/en/ media centre/pr /.../pr208_e.pdf. published by atlantis press copyright: the authors 12 microsoft word efficient applications of risk analysis in the chemical industry and emergency response efficient applications of risk analysis in the chemical industry and emergency response roberto bubbico dipartimento di ingegneria chimica – università di roma “la sapienza” via eudossiana 18, 00184, roma (italy) abstract despite it is now used in many technical and industrial areas, risk analysis is sometimes still considered by many plant managers as a methodology too complex and too timeconsuming to be of practical use. conversely, it will be shown that risk analysis can often provide very useful information on the safety of selected industrial activities. also, the introduction of proper simplifying assumptions can even wider the range of application with great benefit of all stakeholders. keywords: risk analysis, safety, chemical industry, hazardous materials, emergency response 1. introduction the introduction of the first techniques for the calculation of the risk posed by a given technical activity, dates back to the end of the 1970s in the area of the nuclear and aeronautics industries. some of them are nowadays quite well known among practitioners, such as the probabilistic risk assessment (pra)1,2, the preliminary hazard analysis (pha), and so on3,4. however, in the recent years, these techniques, or some modifications of their original versions, have been adopted to a large extent, also in many new areas, such as finance, medicine, biology, etc., different from the native ones5-7. this is because of the high flexibility of the methods and to the capability of a very detailed and powerful analysis of the system under study, whether an industrial installation, an informatics network, a financial system and so on. among the many methodologies available, the quantitative approaches are able to provide more detailed and useful injournal of risk analysis and crisis response, vol. 1, no. 2 (november 2011), 92-101 published by atlantis press copyright: the authors 92 formation but, on the other hand, they are usually much more time-consuming and demanding in terms of computational effort. furthermore, due to the uncertainties in the input data or to the lack of adequate information, in some cases the accuracy and reliability of the results are a matter of discussion. despite these difficulties, risk analysis has gained a good reputation among practitioners, especially in its quantitative version (quantitative risk analysis, qra)8,9. with specific reference to the area of the chemical and process industries, the main problems involved in its use will be here critically analysed and a number of useful examples of application presented. 2. risk analysis risk analysis is a powerful and comprehensive methodology that can be adopted to assess of the level of the risk associated with a given activity (industrial, economical, etc.)5,6. it is a stepwise procedure and more than one single component technique are available for performing each of the tasks within the whole procedure. the need for a stepwise procedure is given by the many aspects which must be considered to estimate the value of the risk. these aspects depend on the area of application, the risk possibly being linked to a financial loss/income, to the safety of an industrial activity and so on. in the specific case of the chemical industry, different variations can be found in the definition of risk in the literature (see for example the paper by kaplan and garrick9, for a more thorough analysis), but it is generally agreed that risk is a function of a combination of the impact of a number of selected hazardous events, and of their probability/frequency of occurrence. of course, based on this definition, the value of this function strongly depends on the identified harmful events (accident scenarios). as a consequence, in order to estimate the risk, the following four traditional main steps composing the whole analysis are required:  hazard identification  consequence calculation  frequency estimation  risk evaluation as mentioned above, in order to accomplish each of these steps, different techniques can be adopted, thus providing the methodology with a high level of flexibility. in fact, on one hand, the methodology can be applied to any stage of the lifetime of an industrial installation: from the initial preliminary design stage, to engineering phase, up to the actual operation of the plan, possibly taking into account all structural and procedural changes (management of changes). on the other hand, the methodology can be applied to a variable level of detail for the same stage of the plant: all the specific pieces of equipment can be studied, along with the associated accidents and their frequency of occurrence and magnitude of consequence; or larger subsystems of the plant can be, at least preliminarily, considered as a black box, to allow a more general and quicker analysis to be carried out. of course, once the critical subsystems have been identified, they can be analysed to a higher level of detail. another very important aspect is that the methodology can be either qualitative or published by atlantis press copyright: the authors 93 quantitative. from this point of view, it worth noting that different conclusions can be drawn by these two approaches. a typical example: it is generally assumed that an accident deriving from one single initiating event, is more critical than one occurring from the combination of two or more initiating events. this sounds reasonable. however, when quantitatively analysed, the frequency of occurrence of the first accident can result even much smaller than that of the second one, thus revealing a misleading conclusion of the qualitative approach. therefore, when planning a risk analysis study, it is important to clearly state since the beginning the scope and aim of the analysis. besides this aspect, it must be said that some of the techniques and models used for performing these tasks can be rather complex and often require skilled and experienced personnel. and even in the case of relatively simple models, a minimum level of uncertainty and approximation is implicit. as examples of such complex models, in the area of consequence calculation, the mathematical models adopted for the assessment of the fraction of aerosol generated after a liquid release, or those used for the calculation of the dispersion of a heavy gas, can be mentioned. in the case of the dispersion of a heavy gas in a complex environment, such as a urban area or a congested industrial site, the application of these models becomes even more difficult and usually require a trained analyst, familiar with the fluid-dynamics and thermodynamics. toxicological information are available only for a limited number of substances, and therefore, even when dispersion can be assessed rather accurately, effect models cannot be used without a high level of uncertainty. other difficulties arise when there is a lack of historical data required as input parameters for some models. for example, the values of the frequency of occurrence of some events (release of a material from a containment system, rupture or failure of a piece of equipment, the ignition of a flammable mixture, and so on) are hardly found in the literature, or, when available, they are referred to systems/conditions other than those under investigation. in these cases the needed parameters have to be estimated by means of specific techniques (fault tree analysis, event tree analysis, etc.), which represent another important source of uncertainty. in fact, besides the quality of the data used to quantitatively solve these models, even the definition of the model itself is prone to uncertainties and errors. based on the above considerations, it appears that the inherent structure of risk analysis, on one hand gives rise to a very flexible and powerful methodology for the assessment of the safety of an industrial activity, but, on the other hand, is a source of uncertainty and errors. also, the higher the accuracy of the results required, the longer the time and the calculation burden for performing the analysis. just to represent more clearly the uncertainties we are dealing with in this area, it can be useful to mention a famous study carried out in the past at a european level10,11: a number of risk analysis teams were provided with the same system to analyse, with specific indications on the techniques to use, and so on. yet, the published by atlantis press copyright: the authors 94 possible assumption of hypotheses, the selection of the incidents to study and of the input data (frequencies, and so on) were on their own judgment. at the end of the study, the results obtained by the different teams ranged over several order of magnitudes, and the influence of the many uncertainties was not enough to explain this variability. only when much more restricting instructions were given to the teams, a much more acceptable variability was obtained. based on these considerations, a strong debate sometimes arises among practitioners, especially process engineers and plant personnel, about the actual reliability and effectiveness of the methodology. different arguments are also debated in the literature12,13. despite these difficulties and sources of uncertainties, risk analysis still represents a useful and sometimes unique methodology for the assessment of the safety of a given industrial activity, especially when a comparative use is to be carried out: e.g. when different layouts of the plants or alternative production processes have to be evaluated. this is now generally recognized and in fact the use of this technique is nowadays also recommended by different regulations in force in europe14 as well as in other countries all over the world. 3. examples of application in the following paragraphs, a number of examples of application of the methodology to various industrial activities will be reported. however, besides the “conventional” use of risk analysis, it must be noted that its component techniques are widely used in a number of other applications: consequence evaluation15-17 , decision making18-20, process design21 and others. 3.1. transportation risk analysis any process industry managing hazardous substances as raw materials or final products to be sent to the market, requires the transportation of these materials to/from the production site. this specific task has to be considered as an integral part of the general production activity. as a consequence, in the framework of the assessment of the risk connected with the whole production activity, the analysis of the risk associated with the transportation phase has to be taken into consideration along with that connected with the other production units. this has been recognized recently, and a specific version of the original quantitative risk analysis technique has been set up, usually referred to as transportation risk analysis (tra)22. this is an extension of the traditional quantitative risk analysis technique, with the distinctiveness that the risk source is not in a stationary location, but is moving along a given route in a continuously changing environment. therefore, in order to apply the technique to transportation activities, the knowledge of a large amount of information is required, partly strictly connected with the territory characteristics, and partly relative to other parameters. just to mention a few examples: the local distribution of population along the route, the site-specific accident rates, the local weather conditions, the accidents evolution in relation with the local specific characteristics, and so on. the need of this considerable amount of information often prevent the application published by atlantis press copyright: the authors 95 of this very useful methodology to practical cases, with heavy detriment of the safety for the involved working personnel, for the exposed population or for the environment. in order to overcome these complexities, different approaches have been proposed in the literature, and a number of methods of implementation of the technique have been devised (see [15] for some references). in most of the cases, the proposed variations are simplified versions of the more rigorous approach23,24. for example, a first solution can be to distinguish the parameters which are really routedependent from those which are not23. in the first group can be considered the release scenarios, the probability of a release scenario after a given accident and the probability of an outcome case following a given release scenario. other parameters, such as the accident rate, the population distribution and the weather conditions, have to be considered as route-dependent factors. however, some simplifying assumptions are possible also in this case, at least as a first approximation: the accident rate can be considered as a function of the type of road (highway, state-road, etc.); typical values of population densities can be associated to a reduced number of different types of built-up areas ranging from urban to rural, and so on. all these assumptions allow to reduce the amount of information to be collected. of course, each simplification will involve advantages and disadvantages. in summary, the pros consist in the availability of a less demanding procedure, requiring a reduced amount of input data, and in a more “user-friendly” approach, allowing the technique to be applied even by less experienced practitioners. on the other hand, the use of too many or too “heavy” simplifications can lead to a less than acceptable accuracy of the results; this can finally result in the adoption of erroneously biased decisions during risk management. even if the accuracy of the results cannot be expected to be very high, in some cases, such as a preliminary analysis, it can provide very useful information and allow an efficient use of time. in figure 1 a comparison between a simplified and a rigorous approach is shown. fig. 1. comparison of f-n curves obtained by a simplified and a rigorous approach [10] in fact, a similar simplified approach has been proposed in the literature25 to identify critical areas along a specific route and thus provide effective strategies, such as a different distribution of the dangerous goods traffic, alternative routes and/or limitation to the transportation itself. also, the adoption of techniques based on the graph theory allowed to improve the planning of the emergency phase. a more efficient way to manage the tra calculations involves the use of geographic information systems (gis). by 1e-09 1e-08 1e-07 1e-06 1e-05 1e-04 1e-03 1e-02 1 10 100 1000 10000 n f ( # /y r) rail road published by atlantis press copyright: the authors 96 this methodology, a tool capable of performing in a simple and relatively quick way all the steps of a tra can be devised. the consequences of an accident are influenced by many parameters, and some of them depend on the specific location (e.g. accident rate, on-route and off-route population, weather conditions and so on). gis databases are able to provide some of these data already directly linked to the location along the route network. different commercial gis databases are already available for almost any country in the world, providing at least some basic information (road and rail network, land use, and so on). nonetheless, the values of other important parameters (accident rates, population and meteorology) have to be collected from different sources and then manually introduced in the gis system. depending on the available data and their initial format, these phases (data acquisition and manipulation) may be quite long (e.g. see [26, 27]). besides the gis system, a database containing information on the probabilities of occurrence of all the possible accident outcomes, and on the magnitude of the consequences of such dangerous events under the selected reference weather conditions, has to be prepared. of course, these data are available only after the chemical products of interest and the weather conditions to be associated with the considered accident have been defined. the larger the number of chemicals and/or the number of weather conditions adopted, the “heavier” the database and the more accurate the calculation results. by such a gis system, the calculation procedure is very simple: once the product and the schedule of the transportation (number of trips per season) have been defined, and once the route of interest has been selected, all the related data are automatically uploaded by the software and used for the calculations. in figure 2, the selection of a specific route between given origin and destination is shown. fig. 2: selection of the route of interest on the gis map [26] the local information about many different parameters (population, road characteristics, etc.) can be automatically displayed on the same map by simply clicking on the proper gis tool buttons (figure 3). fig. 3: site-specific information on the gis map [26] road segment id probab. wind direction weather conditions population d published by atlantis press copyright: the authors 97 both the individual risk as a function of the distance from the route, for each route segment, and the societal risk, in terms of the cumulative f-n curve of the route, can be finally displayed (see figures 4 and 5). sometimes, the procedure can be also speeded up by using one of the gis selection tools, such as the fastest or shortest route selection option. fig. 4: individual risk as a function of the distance from the road. fig. 5: societal risk: f-n curve for the whole route selected. if different routes can be chosen for the same transportation activity, the construction of the risk curves for each of them would allow to identify the one which is characterized by the lower level of risk. at the same time, the knowledge of the individual risk profiles along the selected route make it possible to locate the most hazardous spots: this would allow, for example, to slightly modify the route, achieving a lower risk value with minor changes and, correspondingly, with negligible cost increases. it is thus apparent that such a tool is able to permit a fast and relatively accurate investigation of the alternative routes or transportation modalities for a given activity, and to provide a comprehensive risk management tool. different examples of application can be found in the literature. here, reference will be made to a previous study28 where the overall risk associated with the transportation activities of hazardous materials in a whole region of italy has been reduced by simply modifying the distribution of the transported chemicals in terms of transport modalities and schedule. the data concerning the transportation of dangerous goods in sicily (italy) were obtained from different sources (civil defence office of the prefettura of messina, federchimica, italian railways company, harbour offices, etc.). also, the routes travelled by the products were identified with a reasonable degree of accuracy, thus allowing to get the specific accident frequencies and other characteristics of the travelled routes. the analysis of the data allowed to identify 51 road transportation cases and 4 rail ones, and the use of a specifically devised gis tool led to the construction of the following overall f-n curve (figure 6): published by atlantis press copyright: the authors 98 fig. 6: societal risk for the initial transportation activity. by re-arranging and optimizing the distribution of the transport among road, rail and intermodal and, consequently, modifying the amounts of materials transported and the routes travelled, a new risk curve is obtained. the advantages of the new organization of the transport, is demonstrated by the risk reduction shown in figure 7, where a comparison between the two curves (before and after redistribution) is reported. fig. 7: societal risk for the optimized transportation activity. 3.2. emergency support tool a gis application quite similar to that set up for the analysis of the risk associated with the transportation of hazardous materials, can represent a useful support tool in the case of an emergency. in fact, in the case of an accident, if its location is identified on the relevant gis map, the impact areas of all the possible outcome cases associated to the corresponding accidental scenario (dependent on the substance, the type of release, the environmental conditions, etc.) can be easily displayed on the map itself. if specific information, such as the actual wind direction at the time of the accident, are available, a rather accurate prediction of the territory at risk (of fire in the case of a flammable release, of toxic exposure in the case of a toxic cloud, etc.) can be identified and very useful information can correspondingly be obtained: e.g. the urban areas to be evacuated by the population, the roads and/or railways to be prohibited or, conversely, those to be used by the emergency teams for reaching the location of the accident; similarly, if the location of centres of interest (fire brigade and police stations, civil protection offices, hospitals, etc.) are reported on the map, it might be possible to know what centres need to be activated following the emergency. in figure 7 the possible impact areas following the release of ammonia from a road tank are represented. the examples reported here are only a few of the many possibilities offered by such a kind of tool. 1e-06 1e-04 1e-02 1e+00 1 10 100 1000 10000 n f ( # /y r) rail total road 1e-06 1e-04 1e-02 1e+00 1 10 100 1000 10000 n f ( # /y r before mitigationafter mitigation published by atlantis press copyright: the authors 99 fig. 8: impact areas following a release of ammonia from a road tank. 4. conclusions a quick presentation of some of the most common applications of the risk analysis technique in the industrial sector and in the civil protection and emergency planning area has been reported in the present paper. it should be apparent that, despite the complexity of the procedure and the considerable effort in terms of calculation and data collection, risk analysis still represents in many cases the only method able to provide useful information on the safety and quality of a selected industrial activity. it represents a quantitative approach and, despite the absolute values obtained cannot be considered as “the” value of the investigated risk, especially when applied in a comparative analysis, the results provided can be trusted with a sufficient level of confidence. furthermore, when a number of appropriate simplifying assumptions are introduced, the application can be definitely easier and thus the range of applications and of possible users becomes quite larger with great benefit for all stakeholders: the responsible of the activity under investigation and the exposed population. references 1. nureg, u.s. nuclear regulatory commission, “pra procedures guide: a guide to the performance of probabilistic risk assessment for nuclear power plants”, nureg cr-2003, washington d.c., 1984. 2. nureg, u.s. nuclear regulatory commission, “probabilistic safety analysis procedures guide”, nureg cr-2815, washington d.c., 1985. 3. n. hyatt, “guidelines for process hazards analysis, hazards identification & risk analysis”, dyadem press, richmond hill, ontario, canada, 2003. 4. ccps, center for chemical process safety, “guidelines for hazard evaluation procedures”, 2nd edition, aiche, new york, 1992. 5. g. koller, “risk assessment and decision making in business and industrya practical guide”, chapman and hall/ crc, boca raton, fl, 2005. 6. t. aven, o. renn, “the role of quantitative risk assessment for characterizing risk and uncertainty and delineating appropriate risk management options, with special emphasis on terrorism risk”, risk analysis, vol 29/4, pp. 587-600, 2009. 7. h. li, g.e. apostolakis, j. gifun, w. van schalkwyk, s. leite, d. barber, “ranking the risks from multiple hazards in a small community”, risk analysis, vol 29/3, pp. 438-456, 2009. 8. ccps, center for chemical process safety, “guidelines for chemical process quantitative risk analysis”, 2nd edition, aiche, new york, 2000. 9. s. kaplan, b.g. garrick, “on the quantitative definition of risk”, ”, risk analysis, vol 1/1, pp. 11-27, 1981 10. amendola, “uncertainties in systems reliability modeling: insight gained through european benchmark exercise”, nuclear engineering design, 93, 215-225. amsterdam, the netherlands, elsevier science, 1986 imp a ct a re a s tox ic clo ud fla sh fire uvce jet fire po ol fire 2 0 2 4 kilom e ters ew s n published by atlantis press copyright: the authors 100 11. amendola, s. contini, i. ziomas, “uncertainties in chemical risk assessment: results of a european benchmark exercise”, j. of hazardous materials, 29, 347-363, 1992 12. z. ciechanowicz, “risk analysis: requirements, conflicts and problems”, computers and security, 16/3, 223-232, 1997 13. m. elisabeth paté-cornell, “uncertainties in risk analysis: six levels of treatment”, reliability engineering and systems safety, 54, 95-111, 1996 14. european commission council directive, 96/82/ec, on the control of major accident hazards involving dangerous substances, 9 december 1996 15. j. tixier, g. dusserre, s. rault-doumax, j. ollivier, c. bourely, “osiris: software for the consequence evaluation of transportation of dangerous goods accidents”, environmental modeling & software, vol. 17/7, 627-637, 2002 16. r. bubbico, m. marchini, “assessment of an explosive lpg release accident: a case study.” j. of hazardous materials, 155, pp 558-565, 2008 17. p.a. davies, f.p. lees, “the assessment of major hazards: the road transport environment for conveyance of hazardous materials in great britain”, j. of haz. mat., 32, pp. 41-79, 1992 18. i.a. papazoglou, g.s. bonanos, z.s. nivolianitou, n.j. duijm, b. rasmussen, “supporting decision makers in land use planning around chemical sites. case study: expansion of an oil refinery”, j. of hazardous materials, 71, pp 343-373, 2000 19. k.h. harrington, s.e. rose, “wisely make investment decisions via risk mapping”, chemical engineering progress, 95/12, 63-66, 1999 20. fabiano, f. currò, e. palazzi , r. pastorino, “a framework for risk assessment and decision-making strategies in dangerous good transportation”, j. of loss prevention in the proc ind., 93, pp. 1-15, 2002 21. f.i. khan, s.a. abbasi, “inherently safer design based on rapid risk analysis”, j. of loss prevention in the proc ind., 11, pp. 361-372, 1998 22. ccps, center for chemical process safety, “guidelines for chemical transportation risk analysis”, new york: aiche, 1995. 23. r. bubbico, s. di cave, and b. mazzarotta, “risk analysis for road and rail transport of hazardous materials: a simplified approach”, j. of loss prevention in the proc ind., vol 17/6, pp. 477-482, 2004. 24. c.b. goh, c.b. ching, r. tan, “risk analysis for the road transportation of hazardous chemicals in singapore – a methodology”, j. of loss prevention in the proc ind., 8, pp. 35-39, 1995 25. fabiano, f. currò, a.p. reverberi, r. pastorino, “dangerous good transportation by road: from risk analysis to emergency planning”, j. of loss prevention in the proc ind., 18, pp. 403-413, 2005 26. r. bubbico, s. di cave, b. mazzarotta, “risk analysis for road and rail transport of hazardous materials: a gis approach”, j. of loss prevention in the proc ind., vol 17/6, pp. 483-488, 2004. 27. r. bubbico, s. di cave, b. mazzarotta, “la gestione del rischio nel trasporto di merci pericolose via terra”, convegno scientifico nazionale “sicurezza nei sistemi complessi”, iii edizione, bari, 19-21 ottobre 2005, sessione iii/2, (in italian), 2005. 28. r. bubbico, g. maschio, b. mazzarotta, m.f. milazzo, e. parisi, “risk management of road and rail transport of hazardous materials in sicily”, j. of loss prevention in the proc ind., vol 19/1, pp. 32-38, 2006. 29. g. purdy, “risk analysis of the transportation of dangerous goods by road and rail”, j of haz. mat., 33, pp. 229-259, 1993 published by atlantis press copyright: the authors 101 microsoft word volume 12, issue 3-3 journal of risk analysis and crisis response, 2022, 12(3), 135-142 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i3.334 135 article covid-19 mental health risks a critical survey of africa emma anyika 1,* 1 school of computing and mathematics, the co-operative university of kenya, karen (00502), nairobi, kenya * correspondence: eanyika@cuk.ac.ke received: march 14, 2022; accepted: may 20, 2022; published: september 30, 2022 abstract: covid-19 is still being experienced worldwide, with other variants also cropping up, including delta and omicron, etc. many countries are learning to cope with the mental, physical, and economic impacts of this pandemic by applying recommended health or otherwise protocols in their daily undertakings. the research critically surveyed covid-19 mental health risks in africa. its objectives were to determine the extent to which covid-19, directly and indirectly, affected the mental health of citizens and to estimate the mental health risk levels due to the covid-19 pandemic in africa. in doing so, participants in africa were sent a google survey form by whatsapp. and seventy-two responses were received. the depression, anxiety and stress scale 21 (dass-21) was used to measure participants' mental health risk levels of depression, anxiety, and stress during the covid-19 pandemic periods up to november 7, 2021. the study discovered that over 90 % of individuals had one form of mental health disorder during the pandemic. in addition, many participants experienced severe depression and anxiety resulting in mental health issues such as dysphoria, anhedonia, and inertia that assesses autonomic arousal, skeletal muscle effects, situational anxiety, and subjective experience of anxious affect. keywords: covid-19; mental health; mental health risks; survey; africa 1. introduction the world is still grappling with the covid-19 pandemic almost two years since its onset. according to the worldometer (wm) statistics, as of monday, november 8, 2021, 5075739 people have died of covid-19, including 4706 on the same day. in addition, wm has reported a staggering 251333353 cases of covid-19 infections, with 270,493 new cases reported on monday, november 8, 2021. africa had reported 220237 deaths, 24 of them occurring on monday, november 8, 2021, and 8612365 reported cases of infection, 1022 of these appearing on monday november 8, 2021. the daily covid-19 incidences had significantly dropped worldwide, but new cases continue to be reported, and new outbreaks necessitate stringent containment measures. these scenarios continue to drain resources and endurance for the disease, which may lead to stress and anxiety. despite the decline in the general crime rate, homicides and all types of violence against women and girls had increased. this was the shadow pandemic growing amidst the covid-19 crisis, and a global collective effort was needed to stop it (un, 2021). in may 2020, during the covid-19 pandemic, emergency department (ed) visits for suspected suicide attempts began to increase among adolescents aged 12–17 years, especially girls. by february 21–march 20, 2021, suspected suicide emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 136 attempt ed visits were 50.6% higher among girls aged 12–17 years than during the same period in 2019; among boys aged 12–17 years, suspected suicide attempt ed visits increased 3.7% yard & radhakrishnan, (2021). this paper surveys the african case to establish the mental health impact of covid-19 pandemic considering reports of its effects in other jurisdictions. the various and complex factors impacting our mental health and wellbeing are often defined as either risk factors or protective factors (mental health commission, 2022). risk factors adversely impact a person's mental health. covid-19, like other public health emergencies, may affect the health and wellbeing of individuals. these may cause problems like emotional isolation, confusion, and stigma in communities due to economic losses, closure of schools, job losses, and inadequate resources for medical response; hence it is a mental health risk factor. these were associated with significant psychopathological effects, including depression and anxiety disorders (pfefferbaum & north, 2020). certain groups are at an increased risk of adverse psychosocial outcomes due to covid-19. health workers, especially those on the frontline fighting the virus, are vulnerable to emotional stress due to their prolonged exposure to covid-19 patients, particularly those severely ill who do not make it out alive. the health workers also worry about infecting their families and loved ones. sometimes they must make difficult ethical decisions that border between life and death and often work for long hours with insufficient medical resources, including personal protective equipment in some settings. other groups of people predisposed to adverse psychosocial effects include people with underlying psychiatric and medical disorders and individuals whose immune systems are compromised. those with drug and substance abuse challenges, children and adolescents due to stressful situations in their families, including abuse, disrupted learning, and uncertainty about the future, are similarly affected. women bear the most significant burden of a pandemic due to their caregiving role and those in peaceful humanitarian and conflict environments. the latter special groups of people are likely to have their mental health needs ignored (pfefferbaum & north, 2020). rumors, wrong information concerning the virus, and uncertainty about the future also cause distress among people. therefore, there is a need to investigate the effect of these encounters with the covid-19 disease indirectly or directly, particularly its impact on their day-today undertakings. various countries instituted a raft of public health measures. these included lockdowns, social and physical distancing, quarantining, isolating the infected cases, and putting on masks. some measures like quarantine had been used before during the previous sars-cov outbreak in 2002/2003 and were largely successful in containing and terminating the outbreak. however, it is essential to note that many african countries have never experienced such lockdown measures affecting their social and economic wellbeing. thus, there is a lack of data on mental health risks due to similar pandemics and their effects on citizens' wellbeing, a factor this research will pursue. according to the african mental health research foundation (2020) covid-19 response effort, likely mental health patients were expected to make appointments to visit their clinics for assistance or book a teleconference appointment. considering the extent of the covid disease and the stringent lockdown measures, the mitigation measures for mental health covid-19 related issues seem not to be far-reaching in its penetration. given this, the paper will also investigate how citizens have coped with these mental health risks. 1.1. general objective emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 137 to assess the impact of the covid-19 pandemic on the mental health of africans. 1.2. specific objective (1) to determine the extent to which covid-19 has directly affected the mental health of citizens. (2) to examine the extent to which covid-19 has indirectly affected the mental health of citizens. (3) to estimate the mental health risk levels due to the covid-19 pandemic in africa. 1.3. research questions (1) to what extent has covid-19 has directly affected the mental health of citizens. (2) to what extent has covid-19 has indirectly affected the mental health of citizens. (3) what are the mental health risk levels due to the covid-19 pandemic in africa? 1.4. justification of the study africa has never experienced high risks levels of contagious pandemics of the standards of covid-19. and furthermore, there is lack of documentation on the same. this paper thus not only seeks to determine effect of the covid-19 pandemic on the mental health of citizens but will also determine its risk levels and document both findings. the paper also determines the mitigation or coping methods of citizens. this will form a reference point for similar pandemics as well as form a basis for the determination of suitable current and future mental health risk mitigating measures. 1.5. limitations of the study accessing respondents was the main limitation of the study particularly with the lockdown measures in place. this was mitigated using a google form that was circulated on the whatsapp social media. 2. literature review the psychological effects of quarantines were studied among a group of quarantined individuals in canada. the study identified symptoms of post-traumatic stress disorder (ptsd) and depression in 28.9% and 31.2% of the respondents respectively. the study also found that longer durations of quarantine as well as exposure to sars patients were associated with a higher prevalence of ptsd and depression (hawryluck, et al., 2004). a qualitative study conducted in china on the impacts of health, society and economy of the sars and h7n9 outbreaks found that the sars outbreak impacted negatively on the people’s mental health. two interviewed participants who were doctors mentioned that the outbreak had caused a lot of emotional sadness and those being treated in the hospital could not see their families and were afraid of treatment. it was also found that people went on a drug buying spree with a focus on banlangen (radix isatidis) which was thought to cure the sars virus (qiu, chu, mao, & wu, 2018). zürcher, et al. (2020) reviewed many surveys on mental health problems during virus epidemics and realized that many studies were conducted during or shortly after the peak phase of the epidemic, results were regarded as acute stress reactions that do not allow for inference of longer-lasting mental health problems (mhp). this paper used the depression, anxiety and stress scale 21 (dass-21) by lovibond & lovibond (1995) to measure the mental health risk levels of depression, anxiety and stress of participants during the covid-19 pandemic during the covid-19 pandemic periods up to emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 138 7th november 2021 thus captures results of over twelve months of the epidemic hence capable of providing inference of longer – lasting mhp. 3. methodology 3.1. research design the depression, anxiety and stress scale 21 (dass-21) by lovibond & lovibond (1995) was used to measure participants' mental health risk levels of depression, anxiety, and stress during the covid-19 pandemic periods up to 7th november 2021. dass-21 is a set of three self-report scales designed to measure the emotional states of depression, anxiety, and stress. the depression scale assessed dysphoria, hopelessness, devaluation of life, self-deprecation, lack of interest/involvement, anhedonia, and inertia. the anxiety scale assesses autonomic arousal, skeletal muscle effects, situational anxiety, and subjective experience of anxious affect. the stress scale is sensitive to levels of chronic non-specific arousal. it assesses difficulty relaxing, nervous arousal, and being easily upset/agitated, irritable / over-reactive, and impatient. scores for depression, anxiety, and stress are calculated by summing the scores for the relevant items. 3.2. data collection the population of the study was the african continent. the participants were sent a google survey form via whatsapp, and seventytwo forms were received by the researcher. for in-depth qualitative studies, griffin et al. (1991) found that “20-30 in-depth interviews are necessary to uncover 90-95% of all customer needs for the product categories studied. the questions used in this study were twenty-one in-depth questions which, due to the nature of the covid-19 pandemic, could not be administered by interviews and thus surveyed by a questionnaire. this study used seventy-two responses; thus, it is expected that over 95 % of the respondents’ experiences during the pandemic were captured. moreover, according to hogg et al. (2015), the central limit theorem can be used for approximating specific probabilities concerning the mean x or the sum y of a random sample. that is if x is approximately n (μ, σ2/n), and y is approximately n (nμ, nσ2), when n is “sufficiently large,” where μ and σ2 are, respectively, the mean and the variance of the underlying distribution from which the sample arose. generally, these approximations will be good if n is greater than 25 or 30. this study used the sums of similar responses and their percentages as probabilities to make inferences. since the african population is large enough to tend to be normal, the sample of seventytwo which was randomly picked is representative of the population and good enough to draw inferences conclusions. 3.3. ethical issues no participant was coerced into participating in the survey since in responding one was expected to willingly access the response form on line and freely respond to the survey questions. 3.4. data analysis emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 139 dass matrix by lovibond & lovibond, (1995) is modified to reflect a likert scale of five. depression, anxiety and stress scale are summed for each individual participant, doubled and weighed against the scale to determine the different mental health levels of participants. 4. results and discussion the results of mental health risks surveyed in africa were determined and presented as follows: sample questionnaire: table 1. table of the sample questionnaire. s/n & code survey question scores 1(s) i found it hard to wind down 0/1 2 3 4 5 2(a) i was aware of dryness of my mouth 0/1 2 3 4 5 3(d) i couldn’t seem to experience any positive feeling at all 0/1 2 3 4 5 4(a) i experienced breathing difficulty (e.g., excessively rapid breathing, breathlessness in the absence of physical exertion) 0/1 2 3 4 5 5(d) i found it difficult to work up the initiative to do things 0/1 2 3 4 5 6(a) i tended to over-react to situations 0/1 2 3 4 5 7(a) i experienced trembling (e.g., in the hands) 0/1 2 3 4 5 the das score card matrix was adjusted by proportional comparison to reflect the five tier likert scale as follows: table 2. dass five tier score card matrix. depression anxiety stress normal 0-15 0-12 0-23 mild 16-22 13 -15 24-30 moderate 23-33 16-23 31-42 severe 33-45 24-32 43-55 extremely severe 46+ 33+ 56+ table 3. an extract of results of the survey for das mental health risk levels. 5. how frequent do you leave your house 6. to what extend did you find it hard to wind down 7. to what extend were you aware of dryness of the mouth c) less frequent 3 moderate 3 smallest 1 d) least frequently 2 moderate 3 moderate 3 d) least frequently 2 smallest 1 nil 0 d) least frequently 2 small 2 smallest 1 a) very frequent 5 moderate 3 smallest 1 c) less frequent 3 great 4 great 4 4.1. determination of depression anxiety and stress (das) risk levels 4.1.1. determination of risk of depression table 4. a table of depression risk levels of respondents in volume and percentages. risk levels normal mild moderate severe extremely severe volume 14 16 15 22 5 percentage 19.4 22.2 20.8 30.5 6.94 emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 140 table 5. a table of risk of getting depressed due to the covid-19 pandemic. those affected those not affected percentages 80.6 19.4 depression risk levels were represented graphically as follows: figure 1. a graph showing the risk levels of getting depressed due to the covid-19 pandemic. 4.1.2. determination of risk levels of anxiety table 6. a table of anxiety risk levels of respondents in volume and percentages. risk levels normal mild moderate severe extremely severe volume 9 15 8 20 20 percentage 12.5 11.11 20.83 27.7 27.7 table 7. a table of risk of getting anxious due to the covid-19 pandemic. risk of getting anxious those affected those not affected percentages 87.5 12.5 anxiety risk levels were represented graphically as follows: figure 2. a graph showing the risk levels of getting anxious due to the covid-19 pandemic. 4.1.3. determination of risk of stress 0 5 10 15 20 25 normal mild moderate severe extremely severe 0 5 10 15 20 25 normal moderate mild severe extremely severe emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 141 table 8. a table of stress risk levels of respondents in volume and percentages. risk levels normal mild moderate severe extremely severe volume 28 21 14 9 0 percentage 38.88 21.16 19.44 12.5 0 table 9. a table of risk of getting stressed due to the covid-19 pandemic. risk of getting stressed those affected those not affected percentages 61.62 38.38 stress risk levels were represented graphically as follows: figure 3. a graph showing the risk levels of getting stressed due to the covid-19 pandemic. 4.2. other personal surveyed experiences during the covid-19 pandemic fear, pressure, randomly tested for the virus because due to contact with those infected, job loss, relocation due to loss of livelihood, the most difficult experience, never had any difficulty, overwhelming loss of life, mild and severe symptoms of covid-19 infections. 4.3. coping strategies studying online. working online. working from home. follow recommended protocol for prevention from infection and containment. 5. conclusions many times, africa lacks data for reference and relies on foreign or imported data to discern a lot of forecasted events and happenings. the same occurred during the onset of the pandemic since many african countries didn't have a lot of past data or research on airborne diseases, particularly viral ones. these studies' results will create a database for future reference in case of similar happenings. 0 5 10 15 20 25 30 normal moderate mild severe extremely severe emma anyika / journal of risk analysis and crisis response, 2022, 12(3), 135-142 doi: https://doi.org/10.54560/jracr.v12i3.334 142 the study's results indicate that over 80% of persons were depressed during the covid-19 pandemic, over 87% suffered anxiety disorders, and over 61 % were stressed. over 90% of participants had one form of mental health disorder during the covid-19 pandemic period investigated. many experienced severe depression and anxiety resulting in mental health issues such as dysphoria, anhedonia, inertia assesses, autonomic arousal, skeletal muscle effects, situational anxiety, and subjective experience of anxious affect. these are mental health disorders that require attention. with the stigma associated with mental health issues, there is a high chance that many of these mental health disorders have been left unattended. this article recommends further research be undertaken on mitigating measures. furthermore, a lot of effort and resources by entities need to be dedicated to address the mental health issues determined in this study, including but not limited to outreach, establishment, and dissemination of relevant mitigating methods as well as diagnosis and treatment for severe cases. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] commission, mhi. (2022). about mental health issues. western australia: https://www.mhc.wa.gov.au/your-health-and-wellbeing/about-mental-health-issues/. [2] griffin, a., and hauser, j.r., 1993. the voice of the customer. marketing science, 12(1), pp.1-27. doi: https://doi.org/10.1287/mksc.12.1.1. [3] hawryluck, r., gold, l., robinson, w., pogorski, s., galea, s., & styra., s. (2004, july). sars control and psychological effects of quarantine. centers for disease control and prevention, pp. 1206-1212. doi: https://doi.org/10.3201/eid1007.030703. [4] hogg, r. v., tanis, e. a., & zimmerman, d. (2015). probability and statistical inference. new jersey : pearson education. isbn 978-0-321-92327-1. [5] lovibond, s., & lovibond, p. (1995). manual for the depression anxiety & stress scales. sydney: psychology foundation. isbn: 733414230. [6] pfefferbaum, b., & north, c. (2020). mental health and the covid-19 pandemic. the new england journal of medicine, 510-512. doi: https://doi.org/10.1056/nejmp2008017. [7] qiu, w., chu, c., mao, a., & wu, j. (2018). the impacts on health, society, and economy of sars and h7n9 outbreaks in china: a case comparison study. journal of environmental and public health. doi: https://doi.org/10.1155/2018/2710185. [8] un. (2021). un women raises awareness of the shadow pandemic of violence against women during covid-19. new york: un women. [9] yard, e., & radhakrishnan, l. (2021, june 18). morbidity and mortality weekly report. emergency department visits for suspected suicide attempts among persons aged 12–25 years before and during the covid-19 pandemic — united states, january 2019–may 202, p. 1. [10] zürcher, s., kerksieck, p., adamus, c., burr, c., lehmann, a., huber, f., & richter, d. (2020). prevalence of mental health problems during virus epidemics in the general public, health care workers and survivors: a rapid review of the evidence. frontiers in public health. doi: https://doi.org/10.3389/fpubh.2020.560389. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 12, issue 2-3 journal of risk analysis and crisis response, 2022, 12(2), 88-94 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i2.327 88 article research on the influence of fintech development on the operating benefit of banks yong huang 1,2 and hongmei zhang 1,2,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhm1035@qq.com; tel.: +86-0851-88510575 received: may 14, 2022; accepted: june 9, 2022; published: july 10, 2022 abstract: this paper first theoretically analyzes the influence of the development of fintech on the operating benefit of banks from the perspective of competition and diminishing marginal benefit, then the annual data of 38 banks listed in china’s a-share market from 2015 to 2020 were selected for empirical and robustness tests by system gmm. the results show that there is a significant "inverted u-shaped" relationship between the development of fintech and the operating benefit of banks, which is robust and dependable. finally, relevant feasible suggestions for the development of fintech in banks are provided, in order to help banks' digital transformation. keywords: fintech; bank; operating benefit; gmm 1. introduction with the digital economy has become a national strategy in china, the financial industry is accelerating into an important period of the technological revolution, the transformation of economic paradigms and the reset of production factors. in the context of the financial industry entering a new stage of comprehensive digital transformation, fintech is becoming an important engine driving economic growth, and will also become the industry growth point and competition point of the financial industry in the future. as the ballast of the financial industry, banks have begun to try to use fintech as the driving force for innovation-driven development, speed up the strategic deployment and safe application of fintech, actively seek changes, and take multiple measures simultaneously, such as deepening scientific and technological cooperation, promoting reorganization, and other multi-dimensional exploration of technological empowerment, which means that banks is entering a new track of digital transformation and upgrading. therefore, it is of great research significance to explore how banks can implement new development concepts according to their own characteristics and local conditions to realize high-quality development. however, while fintech drives the digital transformation of banks, the development of fintech also faces many severe challenges, and the problem of unbalanced and insufficient development cannot be ignored. the digital gap brought about by the application of fintech under the wave of digitalization has become increasingly prominent. the "matthew effect" of digital development between large and small financial institutions has yet to be eliminated. fintech applications are blooming and the key technologies need to be broken through. thus, it has extremely important yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 89 practical value that studying the problem of the imbalance and inadequacy in the process of deepening the integration of finance and technology for promoting the development of fintech. regarding the connotation of fintech, there is no consensus in the academic community. from the perspective of technology, any technology that can be used to optimize financial services can be understood as fintech [1]; from the perspective of financial attributes, fintech is a financial activity and financial innovation that creates new products, improves the supply of financial services by using a series of emerging technologies such as big data technology, artificial intelligence, and cloud computing [2-3]; from the perspective of the sum of the relationship between finance and technology, the financial stability board defines fintech as a technological means that can promote financial innovation and have a significant impact on business models, technology applications, business processes and innovative products in financial markets, institutions and financial services the interpretation of the connotation of fintech solves the problem of what fintech is, and the internal logic and mechanism of how financial technology drives financial innovation and development has become a new research content for scholars. some scholars put forward from the perspective of creative effect, fintech has created a new financial ecosystem by restructuring organizations [4], reshaping service processes [5], and expanding service channels [6-7]. from the perspective of information effect, some scholars have proposed that fintech reduces information asymmetry between financial institutions and customers [8-9], reduces transaction costs [10-11], and improves the efficiency of financial operations [12] by innovating data and information acquisition and utilization mechanisms. from the perspective of inclusive effect, some scholars have proposed that through technological innovation, fintech improves financial institutions' accurate assessment of customers' credit levels, lowers the threshold for financing, improves the convenience of financial services, and realizes inclusive finance [13-14]. above all, it can be found that most of the academic research on fintech focuses on the connotation and mechanism of fintech, and few literatures empirically explore the impact of financial technology on the operation and management of banks from a micro perspective. so, this paper theoretically analyzes the influence of fintech on the operating efficiency of banks, and verify it by constructing fintech development index and establishing a regression model in order to facilitate the digital transformation of banks. 2. theoretical analysis 2.1. competitive effect with the continuous advancement of interest rate liberalization, the competition between traditional businesses of banks is intensifying. the fierce competition makes it difficult for commercial banks to maintain growth in their traditional operating benefits, which in turn enhances banks’ willingness to innovate and promotes traditional business and fintech better integration [15]. by providing financial services with low prices, low thresholds and wide coverage, fintech enhances banks’ ability to reach long-tail customers such as ordinary personal financial consumers, small and medium-sized enterprises that were previously excluded by banks, increasing banks’ operating benefit [16]. due to the gradual penetration of fintech enterprises in the supply of financial services such as payment and settlement and the demand for financial services such as internet wealth management, and more and more banks entering the racing track of fintech development, banks are yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 90 faced with competition from other financial institutions and emerging fintech companies. there will be more intense competition among banks to maintain the existing market share and compete for the remains. therefore, fintech will reduce the operating benefit of banks through internal and external competition. 2.2. diminishing marginal benefit effect the traditional economic theory believes that there is information asymmetry between banks and customers, and banks cannot accurately grasp all the information of customers, which will lead to moral hazard and adverse selection. by using fintech, banks can reduce information asymmetry and expand potential customer groups, thereby improving the operating benefit of banks. with the passage of time, the value of potential customer groups has been gradually divided up by banks, and the incremental benefits brought by fintech have gradually subsided. while the original fintech has been iteratively updated, making it that its operating and maintenance costs have continued to increase. besides, the technology threshold of new fintech is also increasing, resulting in the increasing development period and cost. so, banks get the diminishing marginal benefit. based on the analysis of the above two effects, this paper proposes the following theorem. theorem 1: there is an "inverted u-shaped" relationship between fintech and the operational benefit of banks. in order to verify the relevant conclusions of proposition one, the system gmm model will be introduced for empirical research in the following. 3. empirical analysis 3.1. sample selection and data sources this paper selects the annual data of 38 banks listed in china’s a-share market from 2015 to 2020 as a sample, and the data sources is from winder database and other public websites. 3.2. definition of variables 3.2.1. explanatory variables the academic community has not yet formed a unified consensus on measuring the development index of fintech. this paper draws on relevant methods [17] and relevant references [18] to construct the development indicators of fintech. the index system is shown in table 1. table 1. the development index system of fintech. primary indicators secondary indicators industry resource input fintech capital investment ability of research and application number of fintech patents degree of social cognition keyword searching 3.2.2. explained variable this paper uses roa to measure the performance of banks. this index focuses more on the profitability of enterprises and can better reflect the operating benefit of banks. yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 91 3.2.3. control variables when exploring the operational benefits of banks, from a macro perspective, the influence of social and economic development on the operational benefits of banks cannot be ignored. therefore, this paper selects gdp to quantify the level of macroeconomic development. from a microscopic perspective, this paper selects the total deposits, non-performing loans, and net interest margins of banks to measure their influence on operating benefit. they are shown in table 2. table 2. definition of control variables. variable name variable notation gross domestic product gdp total deposits td non-performing loans bl net interest margins nim 3.3. model setting this paper uses dynamic panel data and sets the systematic gmm model to empirically analyze the influence of the development of fintech on the operating benefit of banks. the model expression is shown in formula (1). itiitititttttiit nimbltdgdpftiftiroaroa    7654 2 321,10 (1) 4. results 4.1. the results of regression table 3. the results of regression and robustness test. variable notation formula (1) test (1) l.roa 0.8223*** (0.1212) 0.7338*** (0.1240) fti 0.0099*** (0.0022) 0.0105*** (0.0028) fti2 -0.0038*** (0.0008) -0.0040*** (0.0010) gdp 0.0519*** (0.0162) 0.0431*** (0.0189) td 0.0121* (0.0063) 0.0192** (0.0090) bl -0.0115** (0.0054) -0.0172** (0.0076) nim 0.0085 (0.0083) 0.0136 (0.0100) ar(2) 0.389 0.413 sargan 0.430 0.315 1 standard error are in parentheses under the regression coefficients. 2 ***, ** and * indicate significance at the 1%, 5% and 10% confidence levels, respectively. 3 ar(2) and sargan show the p value. yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 92 4.2. discussion 4.2.1. the formula (1) as the formula (1) shows, it can be included that: first, the p value of ar(2) is greater than 0.1, and it has passed the autocorrelation test, indicating that there is no second-order serial correlation in the random disturbance term difference in the model;and in the instrumental variable over-identification test, the sargan test p value is greater than 0.05, it has passed the sargan test, indicating that all the instrumental variables in the model are valid; therefore, the system gmm model constructed in this paper is reasonable, and the estimation results are valid. second, the coefficient of fti is positive but that of fti2 is negative, and they are significant at the 1% level, which proves that there is an obvious "inverted u-shaped" between the development of fintech and the operating benefit of banks,which is consistent with the previous theoretical analysis. third, the operation will have inertia that the operation of the past period will have a cumulative effect, which will have a significant co-directional effect on the current period; the coefficient of l.roa is greater than 0, which is significant at the 1% confidence level, which is consistent with the operating inertia. fourth, from a macro-level analysis, the better the macroeconomic development and the better the market environment, the more individuals and enterprises need to carry out deposit and loan business. banks can more fully identify the credit status of lenders and carry out long-term data tracking by fintech, reducing the possibility of bad loans in banks, so that banks can expand the customer group of loans and improve the operating benefit of banks. the empirical results show that the coefficient of gdp is significant and greater than 0 at the 1% confidence level. fifth, from the micro-level analysis of banks, the more deposits held by banks, the more funds banks use for lending, and the stronger the ability of banks to create deposits. in addition, the higher the net interest income of banks, the less non-performing loans, and the better the operating benefit of banks. in the empirical results, the coefficient of total deposits and non-performing loans are significantly positive and negative at the confidence level of 10% and 5%, respectively, and that of net interest margins is negative, which is consistent with the theoretical analysis. 4.2.2. the robustness test (1) in order to ensure the reliability of the empirical model and the empirical results, this paper adjusts the sample data capacity, excludes the data of large state-owned banks, and conducts the regression again. the results are shown in table 3 test (1). it can be found that there is no significant change in the test (1) compared with the formula (1), and no specific analysis will be made here. the test (1) in table 3 show that the empirical model is reasonable and reliable, and the empirical results are robust and correct. 5. conclusions and suggestions this paper analyzes the influence mechanism of competition effect and diminishing marginal benefit effect on bank's operating benefit, and establishes a system gmm model to regress the relationship between the development of fintech and bank's operating benefit. the results show that yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 93 there is a significant "inverted u-shaped" relationship between the development of fintech and the operating benefit of banks. accordingly, this paper proposes the following suggestions. first, banks should precisely position themselves, take the initiative to incorporate fintech into their business systems based on local economic development and customer needs, and rely on digital transformation to achieve precise empowerment of fintech. second, banks need to focus their investment on weak links, research, and development (r&d) priorities, select technology r&d and application models that meet their own endowments and actual needs, and build a distinctive financial technology innovation system. third, banks should continue to keep fintech-business development-internal management go hand in hand, deepen the communication and linkage between business and technology, and promote the integration of technology and business, development of technology. funding: this research was funded by 2020 the special project of key cultivation discipline and urgently needed discipline direction of guizhou university of finance and economics, "research on credit risk prediction and evaluation of big data enterprises", grant number 2020zjxk20. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] chen, m.a.; wu, q.; yang, b. how valuable is fintech innovation? social science electronic publishing, 2018, 32(5), 2062-2106. doi: https://doi.org/10.1093/rfs/hhy130. [2] milian, e. z.; et al. fintechs: a literature review and research agenda. electronic commerce research and applications, 2019, 34(2), 1-21. doi: https://doi.org/10.1016/j.elerap.2019.100833. [3] tang, s.; lai, x.b.; huang, r. how can fintech innovation affect tfp: facilitating or inhibiting? — theoretical analysis framework and regional practice. china soft science, 2019, (07), 134-144. doi: https://doi.org/10.3969/j.issn.1002-9753.2019.07.013. [4] sklyar, a.; et al. organizing for digital servitization: a service ecosystem perspective. journal of business research, 2019, 104(11), 450-460. doi: https://doi.org/10.1016/j.jbusres.2019.02.012. [5] peter, g.; kauffman, r.j.; arker, c. on the fintech revolution: interpreting the forces of innovation, disruption and transformation in financial services. journal of management information systems, 2018, 5(1), 220-265. doi: https://doi.org/10.1080/07421222.2018.1440766. [6] drasch, b. j.; et al. integrating the ‘trouble makers’: a taxonomy for cooperation between banks and fintechs, journal of economics and business, 2018, 100(4), 26-42. doi: https://doi.org/10.1016/ j.jeconbus.2018.04.002. [7] zhou, g.y.; luo, s.m.; lian, s.t. research on fintech innovation, interest rate decision of network lending and financing of small and micro enterprises—also discusses the management of “macmillan gap”. studies of international finance, 2020, (3), 76-86. doi: https://doi.org/10.16475/j.cnki.10061029.2020.003.008. [8] vives, x. digital disruption in banking. annual review of financial economics, 2019, (11), 243-272. doi: https://doi.org/10.1146/annurev-financial-100719-120854. [9] hou, s.y; song l.r. fintech, techfin and regional r&d innovation. the theory and practice of finance and economics, 2020, 41(5), 11-19. doi: https://doi.org/10.16339/j.cnki.hdxbcjb.2020.05.002. [10] jagtiani, j.c.; lemieux. the roles of alternative data and machine learning in fintech lending: evidence from the lending club consumer platform. financial management, 2019, 48(4), 1009-1029. doi: https://doi.org/10.1111/fima.12295. yong huang and hongmei zhang / journal of risk analysis and crisis response, 2022, 12(2), 88-94 doi: https://doi.org/10.54560/jracr.v12i2.327 94 [11] wang, x.y.; zhang, j.y.; wang, x.l. fintech, corporate lifecycle, and technological innovation: heterogeneous characteristics, mechanism test and governmental regulation performance evaluation．financial economics research, 2019, 34(5), 93-108. [12] fuster, a.; et al. the role of technology in mortgage lending. review of financial studies, 2019, 32(5), 18541899. doi: https://doi.org/10.2139/ssrn.3127175. [13] zhang, x.; wan g.h.; et al. digital economy, financial inclusion, and inclusive growth. economic research journal, 2019, 54(8), 71-86. [14] hasan, m.m.; et al. regional development of chinese inclusive finance through financial technology. sage open, 2020, 10(1), 1-16. doi: https://doi.org/10.1177/2158244019901252. [15] ye, l.; wang, r. fintech, banks competition and enterprise technological innovation. modern economic research, 2021, (6), 49-57. doi: https://doi.org/10.13891/j.cnki.mer.2021.06.007. [16] wan, j.y.; zhou, q.; xiao, y. digital finance, financial constraint and enterprise innovation. economic review, 2020, (1), 71-83. doi: https://doi.org/10.19361/j.er.2020.01.05. [17] shen, y.; guo, p. internet finance, technology spillover and commercial banks tfp. journal of financial research, 2015, (3), 160-175. [18] qiu, h.; huang, y.p.; ji, h. how does fintech development affect traditional banking in china? the perspective of online wealth management products. journal of financial research, 2018, (11), 17-29. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). research article infection waves in pandemics and risk prediction: physical diffusion theory and data comparisons romney b. duffey* idaho falls, id 83404, usa 1. introduction: typical infection patterns and repeating “wave” phases health crisis policy and decisions are based on infection rates so need quantitative predictions of risk trends, while medical system pandemic planning and response often depends on projected emergency bed count capacity which lag infection rates. since viral pandemics exhibit successive waves or peaks of infections, the key question is what physical model can explain and predict the risk of their occurrence, size trends and timing? the present note expands the numerical study idea by aciola [1] by examining if the dynamic community-wide spread can be described and understood using classical diffusion theory. to avoid confusion, the relevant fundamental physics is completely distinct from the use of the term “diffusion” by sociologists and political scientists to qualitatively explain the varying implementation of pandemic countermeasures and policies between nations [2]. the onset and subsequent progress of pandemics such as h1n1 in 1918 and covid-19 (aka sars-cov-2) in 2020, are known to be characterized by multiple increases or “waves” of infections, peaking, declining and returning over many 100’s of days. we do not re-iterate all the well-documented features of pandemics and their behavior [see, e.g. 3–6], or the many and various person-to-person infection pathways and mechanisms [see, e.g. 7,8]. we distinguish between: the dominant mechanisms of initial (direct and local transmission) rapid spreading for the first wave; and subsequent slower community spreading and transmission for producing the second and more waves (within region/country). we postulate the first peak or “wave” is the result of rapid but mainly externally introduced infections as the virus opportunistically attacks the first susceptible and unaware hosts, and increases limited according to initially rapid (few days) transmission and incubation timescales. infections grow exponentially by unconstrained random person-to-person transmission in an initial sample population or local neighborhood without prior resistance or effective countermeasures. the initial peak (which we term here peak 1) is reached in about 30 days until social and non-pharmaceutical intervention (npi) countermeasures (e.g. improved hygiene, social distancing, improvised quarantines, public awareness, etc…) effectively counterbalance the growth in the infection rate. the peak 1 wave subsequently declines universally according to learning theory from adopting such simple countermeasures [9,10]. however, zero risk or rates are not achieved or achievable (as known for perennial influenza) due to hidden residual infections, asymptomatic cases, and/or undetected importations or “clusters”, which ensure the virus continues to be slowly, inexorably and a r t i c l e i n f o article history received 15 november 2020 accepted 08 june 2021 keywords covid-19 infection waves community spreading diffusion theory data predictions policies a b s t r a c t we predict the magnitude and estimate the uncertainties of the spread, growth and maximum expected long-term infection rates that affect emergency policies and plans. for the covid-19 and 1918 viral pandemics, large second or successive peaks, waves or plateaux of increased infections occur long after the initial rapid onset. the key question is what physical model can explain and predict their occurrence trends and timing? we establish the principal that the timing and magnitude of such increases can be based on the well-known physics of classical diffusion theory, so is fundamentally different from the commonly used multiparameter epidemiological methods. this physical model illuminates our understanding of the societal viral progress, providing quantitative predictions, estimates and uncertainties supporting risk decision-making and resilient medical planning. we obtain an approximate relation for predicting the risk of the observed magnitudes, timing and uncertainties of second and more waves, as needed for proactive emergency pandemic planning, bed count and decision-making purposes. the dynamic results and characteristics are compared and fitted to data using just two physical parameters for a number of countries and regions, and the concept shown to apply for both entire national and local regional populations. the present analysis quantitatively shows how much the timing and magnitude are reduced by more learning and effective countermeasures. the medical system and health policy must recognize and pro-actively plan for such inexorable diffusive spread and large residual infection waves. © 2021 the author. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *email: duffeyrb@gmail.com journal of risk analysis and crisis response vol. 11(2); june (2021), pp. 67–74 doi: https://doi.org/10.2991/jracr.k.210609.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:duffeyrb%40gmail.com?subject= https://doi.org/10.2991/jracr.k.210609.001 https://www.atlantis-press.com/journals/jracr 68 r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 randomly distributed. the virus progressively becomes distributed throughout the community, and largely undetected or hidden progressively continues to spread. this initially low level of mainly internally introduced or enabled community-wide spread inside countries is widely observed, even in island states where complete entry and viral import restrictions, mandates or controls are possible (like hawaii, australia and new zealand). since the initial infective decline is competing against diffusive increase, the second wave onset only starts to be discernable or noticeable many weeks after the peak 1. eventually, after about 100 days, a second distinct “wave” or increase toward a plateau (which we term here peak 2) is more clearly observed, which is also reached before some balance is achieved between increased diffusive transmission and effective countermeasures. while not exactly followed by all regions, these overall data patterns, trends and surges are compelling and apparently independent of continent, culture, society or century, as shown in figure 1 (the rates per day are normalized to the first peak and the temporal evolution measured in days exceeding 100 initial infections as a nominal start or pandemic threshold). we hypothesize that the second and subsequent waves have a fundamentally different physical spreading character, having moved from the early rapid infection phase of initial local populations dominated by rapid personal incubation timescales (3–5 days) and contacts to slower and more extensive overall community spreading (50–100 days) dominated by slower societal interactions, more cautious behaviors and residual infections.1 these phases may of course overlap or not be entirely distinct everywhere, so we look to data for guidance, quantification and verification. the usual susceptibility-exposed-infected-recovery (seir) and r0 models for the infection rate [11–13] have been fitted using some seven parameters to first peak rise and decline data [e.g. 14,15] where the infection increase rate is limited by personal incubation timescales. previous work on second waves [16] solved the standard seir model plus a further two equations, giving six first-order differential equations with eight variable parameters, four representing various assumed socio-economic countermeasures (e.g. “lockdown”, school closure, perception and learning). despite the seir approach being known to be inaccurate and not predictive [13], three parameters were first fitted or “calibrated” to data, while varying the five remaining of the eight parameters gave differing peak 2 to peak 1 ratios. the results showed oscillations, with peak 2 at 240 days and then a decline, and that “in some cases, the second peak was higher than the first peak, while for other parameter combinations it was lower.” it was concluded “a second wave of covid-19 on account of the coupled behavior-disease feedbacks… will characterize many populations” [16]. there were no comparisons given to actual data as in our new approach. to model the societal medico-physics, and avoid so many empirical parameters, we postulate that after the initial peak this subsequent community-wide spreading is dominated and governed by random diffusion, with infections seeping steadily, systematically and inevitably throughout the population, and largely independent of countermeasures. as first pointed out by aciola [1] for the first peak using purely numerical solutions: “a simple diffusion model treats each individual in a population as a brownian particle …added to this model is the incubation period of the virus and a probability of transmission of the virus if individuals are closer than a certain distance.” this fundamental idea reflects the reality that individual infection transfers and opportunities are statistically random; and of not being able to identify, track, eliminate or monitor all the encounters and social interactions of every infected person and prevent such transfers. for a whole society, we are not looking at individuals but entire populations and overall societal behaviors and the resulting infection trends. therefore, for second or other successive waves the problem and equation to be solved is essentially fick’s or fourier’s law, a second order differential equation, where the net rate of change of infections at any location is proportional to the incremental infection gradient. for any infection number, n, at any time, t, conventionally, without additional internal or imported infection sources:2 � � � n t d n� 2 (1) the community-wide diffusion coefficient or diffusivity, d, physically represents all random person-to-person and intra societal cross infections and is the key parameter to be deduced figure 1 | general idealized trends of pandemic infection waves over the last century, with daily infection numbers normalized to the first peak. (1) initial random infections, n0, from external introduction with rapid rise (incubation limited timescale, g). (2) exponential rise to first wave peak, nm1, in circa 20–30 days, dm1. (3) decline due to learning (countermeasure like awareness, hygiene, and social distancing…), k. (4) minimum achievable infection rate (or detection threshold), nm. (5) second “wave” apparent onset, n02, at circa 50–100 days (community spread limited timescale). (6) second rise, n(d), from random intercommunity spread (societally embedded). (7) second peak or plateau, nm2, at about, dm2, circa 100+ days often larger than first peak. (8) plateau or decline due to learning, countermeasures plus human host spreading limits. (9) minimum “acceptable” or achievable rate achieved again. (10) additional waves possible (annual, seasonal, social…). 1we utilize infection rates as a leading indicator of spread, while public health officials usually focus on death numbers and rates which lag infections; and which fraction of infections is variable being highly dependent on the propensity, vulnerability, medical treatment and ages of the population so cannot be solely diffusivity dominated. 2for a source add to the rhs a term, say ns′, being the number per unit time. r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 69 from observational data. importantly, equation (1) is a second order differential equation, so: (a) is fundamentally more general than the first order differential equations used in classic r0 and seir epidemiological spread models with multiple adjustable parameters; (b) contains a very limited number of variable but physically-based coefficients that can be directly tested against or fitted to data; and (c) therefore able to draw on the foundations and prior knowledge of classical physics and methods for heat and concentration diffusion. to illuminate the physics, we seek an analytical solution to equation (1) that is testable against publically available data for the second and/or subsequent waves. 2. materials and methods 2.1. basic theory and postulates exact solutions of equation (1) are numerical [1] whereas we only seek a working correlation to aid our understanding and enable rapid estimates for risk decision-making. for the second and successive “waves”, we make simplifying assumptions and reasonable approximations to derive a basic analytical solution form for the dynamic infection trends and the limiting physical societal transmission mechanism, the accuracy and applicability of which can be determined by comparisons to data. hence, even if and as countermeasures are deployed, without complete isolation or elimination the virus spreads: (a) by diffusion in any local region(s) or cities, becoming inevitably embedded in the wider community (regions of higher infection numbers naturally infect regions with lower); (b) throughout any national or regional location, all the population is equally able to be randomly infected so the distribution of the virus is to first order homogenous and any and all infections are equally possible; (c) with fundamental physics limiting the extent the rate of subsequent community spreading is described by the classic fick’s law with diffusivity parameters averaged over the population/ region/society; and, (d) being equally possible so affording a simple homogenous and one-dimensional approximation, with the number of infections, n, some fraction of the total possible, n, where usually n <<< n, depending on the overall community transmission mechanisms, societal behaviors, and countermeasure effectiveness. for some overall societal characteristic effective transmission scale, l, after an elapsed time measured in days, d, the rate of change of infections or number counted on any day, n(d, l), in one dimension equation (1) becomes, � � � � n d d n l � 2 2 (2) using the usual non-dimensional formulation, q � � � n n n n m m02 , with nominal minimum achievable and second wave initial numbers, nm, and n02, respectively we have, � � � � q q d d l � 2 2 (3) prior physics and classic texts provides the analogous fourier flux solution satisfying equation (3) [1,17,18] and is the usual error function, q( ) ( ) / d l dd = −       1 4 1 2 erf (4) where, the term, ld = (4dd) 1/2, is an effective overall community viral diffusive penetration distance or societal transmission “scale”. in contrast to the classic multi-parameter seir models [13,16] there are now only two governing and physically-based parameters to be determined. 2.2. approximate correlation derivation for wave timing and magnitude to guide our thinking for present correlation and risk trending purposes, for the large-scale ratios, l/(4dd)1/2, relevant to whole communities and regions during diffusion, we can retain just the first term in the series expansion of equation (4) [see 18 #586], so: q p ( ) ( ) d n d n n n l dd em m l dd= − − ≈     − −       02 1 1 2 88 2 (5) usually, the minimum rate number, nm << n(d) and n02, so q( ) ( ) d n d n ≈ 02 , and increases initially as (dd)1/2. the validity, accuracy and limitations of this (or any other) approximate analytical solution must be demonstrated by data. for any chosen region this simple solution equation (5) has the sensible limiting conditions, n → nm as l → ∞, n → n02 as d → 0, and n → ∞ as d → ∞ for any given diffusivity, d. after the decrease from peak 1, we can expect to observe an initially slow square root increase in infection numbers with time, followed by an inevitable exponential increase at longer elapsed times which is a prediction consistent with the observations (figure 1). writing equation (5) in a convenient non-dimensional form, with l* = ld/l, q p ( )* * *l l e l� � �� � �� � � �� � ��2 1 2 1 2 2 (6) using equations (5) and (6) for describing the second wave requires determining just two adjustable but physically based parameters: the diffusion coefficient, d (m2/day), and the assumed characteristic or effective length scale, l (m), assuming the peak 1 parameters (g and k) still prevail. both of course depend on some integration/aggregation of all the exact overall societal viral transmission mechanisms (person-to-person, aerosols, crowds, contamination, random exposure…), which we simply assume to exist. we can then also estimate the relative size and timing, dm2, for attaining peak 2 when the diffusive growth is eventually balanced, assuming the same npi learning countermeasures are used (e.g. social distancing, improved hygiene, public awareness etc...). for peak 1, the decrease in the increasing rate decreases when the daily transmission and incubation increase number, nm1, is comparable to or balanced by countermeasures and public awareness. for some 70 r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 learning constant, k, representing the overall countermeasures effectiveness and resilience during and after peak 1 [10], n d n em g k dm 1 01 1( ) ~ ( )− − (7) here, the parameters for incubation growth for any initial viral variant, g, learning recovery, k, and peak day, dm1, are known from the peak 1 prior data trends (figure 1), and allows quantification of countermeasures by evaluating, k. this second wave or “curve flattening” semi-plateau has an asymptote given by dn dd 2 0→ , so from equation (5), peak 2 tends to flatten as, d l kdm 2 2 1 2 8 →     , i.e. governed by the ratio of the learning countermeasure reduction to the diffusion increase e-folding rates. the more effective the countermeasures (k increasing) the earlier the infection curve “flattens”. so, with countermeasures, the ratio, r*, of the peak 1 and peak 2 infection rate numbers is the key measure, and for the nominal initial base or threshold daily rate numbers, n01 and n02, respectively, from equations (5) and (7), r n n n n dd l e m m m l dd g k d m m* ( )= ≈       − − −      2 1 02 01 2 2 1 2 88 2 2 1 p       (8) given this approximate second plateau, from equation (8) the predicted peak 2 to peak 1 rate size ratio is of order, r n n dd l em g k dm* / (( ) )( /≈ { }− −02 01 2 2 1 2 18 1p ) , and is usually >1 depending on effective initial learning and the ratio of the low base infection numbers, n n 02 01 . using the ratio removes any dependency on uncertainties in actual case counts, testing variations and reporting protocols. the magnitude ratio for any additional successive (third plus) waves, e.g. n n m m 3 2 , and timing, dm3, will now depend on the subsequent prevailing virus variant growth rate, g, and effective countermeasure, k, values after peak 2, which latter factor can now include vaccination administration extent and effectiveness. 3. results 3.1. comparisons of theory to pandemic second wave onset data trends excellent data for daily infection numbers are available in downloadable spreadsheet format [19] from january 2020, and from who, johns hopkins university and other reliable sources, including local state data from departments of public health websites. as a reference origin, we adopt the initial detection/reportable threshold of n > 100 total cases as a datum for the definite onset of significant infections, d = 0, and varying this threshold has only a minor impact as peak 1 usually occurs within 20–30 days. the origin of the minimum at the second wave onset, q (d02), is the lowest daily infection count continuously achieved following the exponential decline from peak 1, and used to normalize all subsequently increasing infection rates, r* = q (d)/q (d02). to encompass widely separated and differing societies, we performed test calculations for ongoing and continuing infection data showing distinct second waves as listed in table 1 for uk, italy, france, canada, and australia (as of october 15, 2020). the latter is specifically chosen as being a distinct “island state” with comparatively low (but nonzero) rates but second waves (or “clusters”) despite extensive internal controls and border/entry restrictions. comparisons to the theory are shown in figure 2, where the fits were derived using the range of 0.012 < d < 0.014 m2/s and 7 < l < 12 m-values as shown in table 1 and are largely independent of country or societal culture. these fitted parameter values in table 1 indeed are consistent with a whole society reaching a peak or plateau at d l kdm 2 2 1 2 8= ( ) / ~ 200 days using k ~ 0.02/day as we previously derived for the recovery from peak 1 in italy [9] and also applicable for uk and turkey. the values for the long-term diffusion coefficient in table 1 and figure 2 are all o(10−2) m2/day, and for l ~ 10 m and d ~ 100 days we have, l/(4dd)1/2 ~ 5, so the first term approximation for equation (5) is reasonable. the overall agreements with such disparate data are not perfect (canada and uk exhibit over predictions) but the salient overall slow growing trend and 100–200 days timescales are reasonable (especially for australia, france and italy) given the approximations in the theory. capturing the steep exponential onset is also inexact at present. the australia case with literally different boundary conditions on infection control shows clearly that peak 2 can also be reached, while terminating the second wave growth as also observed in the classic 1918 pandemic [4,20,21]. 3.2. estimating second wave magnitude and uncertainties when recovery from peak 1 is not always complete, n02 > n01, ensuring that peak 2 daily rate is larger and what is actually observed, affecting emergency pandemic planning and decision-making. for countries already known to exhibit second wave peaks with recovery/decline, australia actually had a peak 1 to peak 2 infection rate magnitude ratio, r n n m m * = 2 1 , of 2.2 and japan of 3. the still evolving usa data had a first peak of circa 30,000 at 40 days which, after an initial decline at 150 days merged into a second “wave”, almost trending to a wavy persistent plateau and a still increasing rate of over some 2,000,000 per day, giving a peak ratio of at least n n m m 2 1 ~ 6. within any given society there is wide variation in local infection numbers and rates, so we made a more detailed “within country” analysis for the 10 us states which have exhibited a distinct or discernable early peak 1 plus evidence of already reaching peak 2 or some semi-plateau in daily rates after about 100+ days (and as of table 1 | country data with first peak and distinct second wave country/ region start d = 0, 2020 peak 1 rate p (d0)day d (m 2/day) l (m) australia 18 march 715 81 0.013 10 canada 12 march 2760 53 0.012 7 france 5 march 4610 107 0.014 10 italy 25 february 6560 100 0.014 12 uk 7 march 5290 120 0.014 9 r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 71 june 22, 2021, available state-by-state at https://coronavirus.jhu.edu “by region”), as shown in table 2, there is a wide range of ratios with an average of n n m m 2 1 4 1= . , comparable to the above national average. these observations can be compared to the predicted peak 1 to peak 2 number ratio from the diffusion model [equation (8)]. using the above observed and fitted values of dm1 ~ 30 days, dm2 ~ 100 days, with (g–k) ~ 0.15 per day, gives r n n n n m m * = ≈     2 1 02 01 8 , for d ~ 0.01 m2/day, l ~ 10 m, a second peak 2 or semi-plateau of nearly an order of magnitude larger than the peak 1 long preceding it. the uncertainty in this simple estimate depends on parameters which are location and policy specific, namely the balance, (g–k), between infection growth and societal countermeasure effectiveness. following these predictions and estimates, we examined the apparently unique case of china3 as: (a) being the original source of the pandemic reaching peak 1 in 26 days of 4000 cases per day by february 13, 2020; (b) exhibiting a rapid decline from peak 1 which followed learning theory and presaged global recovery trends [9]; (c) thereafter reporting low daily infection rates (claimed to be mainly importations), with a barely discernable second wave peak 2 reported as 276 cases per day on july 31; and (d) maintaining almost complete control of population movement and mobility, with enforced npi mandates and personal contact tracking. with low and fluctuating numbers, it is difficult to uniquely pinpoint the second wave onset day, dn02, and rate value, n02, taken here as nominally day d = 100 with just two reported cases per day, suggesting n02/n01 ~ 2/100 = 0.02, which would give nm2/nm1 ~ 0.16, compared to that reported or observed of 276/4000 ~ 0.07. the comparison of the data to the fitted theory [equation (5)] is shown in figure 3 using the same d = 0.012 m2/day and l = 10 m values found for the table 1 countries. this similarity suggests the fundamental physical diffusion and societal transmission processes are globally the same; and that china exhibits the same long-term embedding in the community as everywhere else. in addition since the peak 1 initial recovery in china is similar to italy and others [10], the second peak should be r n n * ≈     8 02 01 ~ 8, close to that observed. as further confirmation, currently, a wide range of second defined waves or plateaux after peak 1 are on-going and have not yet 3this case study was suggested by professor francesco d’auria as being of particular importance in terms of the total population at risk. table 2 | us state data with first peak and distinct second wave us state peak 1 rate, nm1 peak 2 rate, nm2 ratio, r * alabama 346 2143 6.2 arkansas 323 992 3.1 colorado 526 1139 2.2 florida 1269 13965 11 georgia 1333 4813 3.6 idaho 131 878 6.7 kansas 345 1040 3 louisiana 1857 3840 2.1 nebraska 645 1286 2 washington 584 959 1.6 figure 2 | second wave trends compared to theory. 72 r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 declined [22]. current ratio examples include: austria (8), belgium (9), canada (2), denmark (4), france (11), germany (4), indonesia (4), iran (4), israel (5), italy (6), nepal (9), spain (3), sweden (3), and uk (6). these overall ratios cover a similar range (2–12) to the us internal/regional state ratios in table 2, and encompass the predicted estimate, r n n n n m m * = ≈     2 1 02 01 8 , which surely is not a coincidence. 3.3. local and regional peak ratio trends and predictions insignificant initial infections may occur locally outside the initial peak regions (e.g. large cities like new york or wuhan) but increase only after about 100–150 days solely due to slow progressive internal diffusive spread throughout the region, notably seen for us states like alabama, alaska, wisconsin and some european countries. being important for local public health risk planning purposes, we examined if an even small regional population internal to the country followed the same trends. data were obtained directly from the state-run east idaho public health district (eiphd) covering bonneville county, idaho (where the author lives), with a local population only circa 120,000, a rural and urban setting without a first local first peak despite idaho overall having a peak 1 from infections cases occurring in other locations (figure 1). the second wave arrived and emerged after about 100 days after the first onset, a similar timeframe to the second wave onset for the whole state data in figure 2. figure 4 compares the data to equation (8), with the starting, d = 100 days after first exceeding 100 initial cases in the whole state (figure 1). the “best” theory fit shown uses the same values for d (0.012 m2/day) and l (10 m) as for china and all the large countries in table 1 and encompasses the data. this is a major and quite unexpected confirmation that the diffusion concept represents infection risk growth throughout whole and totally disparate societies with large differences in population and enforced or “mandated” countermeasures. in passing, note that despite continued controls and countermeasures, new york state is currently exhibiting the symptoms of the onset of a second wave after day 150 (august 7, 2020), with cases slowly and inexorably rising. this is a pure prediction for a possible second wave of similar magnitude to italy.4 for successive (third or more) waves arising in the ever-widening infected community, the assumption is each wave is superimposed on (or builds upon) any base or remnants of predecessor infections from the earlier wave(s). in usa and japan, despite being totally distinct and distanced societies, distinct peak 3 occurred after a further 100 days after peak 2, so an initial scoping analysis was performed using the identical l and d values derived for peak 2 to predict the usa peak 3 growth, as shown in figure 5. although a simplified and pessimistic over-prediction, the overall trends are reclaimed, with the ratio, n n m m 3 2 � �� � �� ~ 8/3 ~ 2.7 in accord with table 2 n n m m 2 1 � �� � �� values, implying again that diffusion represents the limiting or controlling processes for any and all additional waves independent of societal countermeasures.figure 3 | comparison of theory to second wave data in china. figure 4 | internal second wave onset in local region of a usa state. figure 5 | initial scoping third wave predictions for the entire usa (data for march 4 to november 10, 2020). 4note added in march 2020: as predicted in october, such a second wave in fact occurred in italy with peak 2 in mid-november with rate magnitude ratio nm2/nm1 ~ 8. r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 73 4. discussion: the risk and uncertainty of large successive waves and vaccine effectiveness we need to pro-actively plan, but as far we know there have been no other second wave size and timing predictions with diffusion adopted as the dominant limiting physics for community spreading. predicting second wave timing and magnitude is a difficult task with considerable uncertainty so we have tried to find a simple solution to this hard problem while not claiming complete accuracy. despite significant variability and uncertainty, after a few 100 days second waves (up to a factor of 10 larger than peak 1) are to be expected, and this elevated risk should be pro-actively planned and managed. given diffusion represents the limiting societal transmission phenomena, the present simple one-dimensional theory cannot replicate every detail for more complex scenarios but can plausibly explain as follows: (a) no discernable first peak appears because multiple numerous initially rapid infections dominate initial national counts, followed by more widespread cases over the 100’s of days of slow diffusive timescale, notably as observed in argentina and brazil as well as in local regions. (b) only a slow recovery occurred after the initial rapid peak 1, so many thousands of cases were still occurring during the diffusion masking the second wave onset which is then superimposed after about 150 days, as notably demonstrated by russia and the usa. while beyond the scope of the present paper, these detailed trends really require more complex intra-regional analysis. the limiting overall societal viral transmission mechanism being diffusion via widespread societal contamination and viral presence replaces the usual paradigm of traditional seir and c0 epidemiological models which assume the new infection rate is directly proportional to the number originally infected [e.g. 11–13,16]. ascribing physical significance, the observed characteristic diffusive length scale of approximately 10 m is an effective overall societal transmission distance not inconsistent with feasible transmission dimensions [21]; and not coincidentally is a typical room, home or store dimension where people and societies interact. the order of magnitude for the coefficient, d, is a factor of more than 10,000 smaller than the 10−2 to 10−3 m2/s associated or expected for local airborne atmospheric particulate or direct aerosol spreading mechanisms [see, e.g. 23,24] reflecting the slow onset of second and other waves are simply the consequences of normal human and personal interactions and overall societal risk-taking behavior [25], as exemplified in the usa by the contrasting weak and strong policies adopted in florida and california, respectively. as to the important question of quantifying effectiveness of vaccines as a pharmaceutical intervention or countermeasure, in principle our learning model [9,10] allows direct estimation of the revised, k, value in equation (7) describing the exponential decline from peak 3, an exercise that is on-going. the new predictions and comparisons illuminate public health policy and risk planning. the present analysis quantitatively shows timing and learning and adopting effective countermeasures reduce magnitudes but cannot eliminate infections completely. since pandemic second and more waves persist, the only perfect diffusion “barrier” or so-called “circuit breaker” is indefinite isolation for everyone everywhere, which is neither feasible nor desirable. 5. conclusion: being pro-active in risk predictions we have established the principal that the timing and magnitude of long-term infection increases or successive waves can be based on the well-known physics of classical diffusion theory. this simple concept explains and predicts the successive “waves” or recurrences of higher infection numbers 100’s of days after the first peak in case rates, with the subsequent inexorable dynamic community-wide spread and embedding of infections. if the pandemic started in localized centers, this diffusion process represents the inevitable but “imperfect” subsequent transmission of viral infections independent of strain due to prevalent, pervasive and persistent societal behaviors, and is a physical model fundamentally different from the proportional rate assumptions of traditional seir and c0 epidemiological models. to enable original trend, size and timing predictions and obtain physical insight, we have derived explicit approximate equations for dynamic infection numbers using the simplest homogenous one-dimensional diffusion model. two physically-based variables, a characteristic societal length scale and an effective community wide diffusion coefficient, have been fitted to the infection rate data for a wide range of countries and local regions of differing societal behaviors and countermeasures, showing that the effective diffusivity and characteristic scale are universal. the timing and relative size of the second and subsequent larger peaks or semi-plateau is also limited by the relative (in)effectiveness of npi countermeasures and societal learning effects. medical system planning and emergency bed count capacity projections encompass peak wave magnitude ratios, r*, of up to about 10 times the initial peak infection rate after about 100 days. we know of no other estimation method or published number for the risk of second and subsequent peaks and for estimating uncertainties. the policy and risk planning implications of such inexorable diffusive spread must be by recognizing potentially larger infection waves will inevitably occur and pro-actively planning for that eventuality. conflicts of interest the author declares no conflicts of interest. acknowledgments the author is indebted to all the contributors to infection data collection and reporting that have made this work possible, and ms. geri rackow for supplying the eiphd data files. he also thanks dr. e.d. hughes for his important critique and suggestions; and professors enrico zio and francesco d’auria for their comments, encouragement and extended dialogs on this topic. 74 r.b. duffey / journal of risk analysis and crisis response 11(2) 67–74 nomenclature d – effective community diffusion coefficient; d – elapsed time in days; g – initial characteristic infection rate; k – learning countermeasure characteristic rate; l – characteristic transmission societal scale; l*– diffusive scale ratio, (4dd/l2)1/2; npi – non-pharmaceutical intervention; n – number of infections in any day; r* – infection number ratio, r* = q (d)/q (d0); t –time; q – non-dimensional number ratio; [equation (5)]. subscripts m1 – maximum value at peak 1; m2 – maximum or asymptotic value of second wave; m – minimum achieved or attainable value; 0 – initial or beginning value; 01 – initial value for onset of peak 1; 02 –initial value for onset of second wave. references [1] aciola ph. diffusion as a first model of spread of viral infection. 2020. arxiv:2003.11449v2 [physics.ed-ph]. [2] sebhatu a, wennberg k, arora-jonsson s, lindberg si. explaining the homogeneous diffusion of covid-19 nonpharmaceutical interventions across heterogeneous countries. proc natl acad sci u s a 2020;117:21201–8. [3] glezen wp. emerging infections: pandemic influenza. epidemiol rev 1996;18:64–76. [4] barry jm. the great influenza: the story of the deadliest pandemic in history. new york: penguin random house books; 2005. [5] us centers for disease control (cdc). coronavirus disease 2019. us centers for disease control; 2020 (accessed at www. cdc.gov). [6] binti hamzah fa, lau c, nazri h, ligot dv, lee g, tan cl, et al. coronatracker: world-wide covid-19 outbreak data analysis and prediction. bull world health organ 2020 [preprint]. [7] national academies of sciences, engineering, and medicine. airborne transmission of sars-cov-2. in: proceedings of a workshop—in brief. washington, dc: the national academies press; 2020. [8] covello vt, hyer rn. covid-19: simple answers to top questions, risk communication guide. arlington, virginia: association of state and territorial health officials (astho); 2020. [9] duffey rb, zio e. analysing recovery from pandemics by learning theory: the case of covid-19, ieee eng med biol soc sect 2020;8:110789–95. [10] duffey rb, zio e. prediction of covid-19 infection, transmission and recovery rates: a new analysis and global societal comparisons. saf sci 2020;129:104854. [11] heesterbeek jap. a brief history of r0 and a recipe for its calculation. acta biotheor 2002;50:189–204. [12] holme p, masuda n. the basic reproduction number as a predictor for epidemic outbreaks in temporal networks. plos one 2015;10:e0120567. [13] li j, blakeley d, smith rj. the failure of r0. comput math methods med 2011;2011:527610. [14] bertuzzo e, mari l, pasetto d, miccoli s, casagrandi r, gatto m, et al. the geography of covid–19 spread in italy and implications for the relaxation of confinement measures. nat commun 2020;11:4264. [15] krishna mv. mathematical modelling on diffusion and control of covid-19. infect dis model 2020;5:588–97. [16] pedro sa, ndjomatchoua ft, jentsch p, tchuenche jm, anand m, bauch ct. conditions for a second wave of covid-19 due to interactions between disease dynamics and social processes. front phys 2020;8:574514. [17] carslaw hs, jaeger jc. conduction of heat in solids. second edition. oxford, uk: oxford university press; 1959, p. 310. [18] mills af. heat transfer. chapter 3. illinois, usa: irwin; 1992. [19] ritchie h, ortiz-ospina e, beltekian d, mathieu e, hasell j, macdonald b, et al. coronavirus (covid-19) cases. 2020. published online at ourworldindata.org. available from: https:// ourworldindata.org/covid-cases [online resource]. [20] andreasen v, viboud c, simonsen l. epidemiologic characterization of the 1918 influenza pandemic summer wave in copenhagen: implications for pandemic control strategies. j infect dis 2008;197:270–8. [21] duffey rb, zio e. covid-19 pandemic trend modeling and analysis to support resilience decision-making. biology (basel) 2020;9:156. [22] dwight hb. tables of integrals and other mathematical data. new york: macmillan; 1968. [23] gorbunov b. aerosol particles laden with covid-19 travel over 30m distance. preprints 2020;2020040546. [24] korovina nv, zharova ik, kudryashova ob, titov ss, diffusion coefficient when fine aerosol media propagate in a confined volume. epj web conferences 2016;110:01029. [25] duffey rb. learning about risk: life, death...and money. 2020. https://doi.org/10.1073/pnas.2010625117 https://doi.org/10.1073/pnas.2010625117 https://doi.org/10.1073/pnas.2010625117 https://doi.org/10.1073/pnas.2010625117 https://doi.org/10.1093/oxfordjournals.epirev.a017917 https://doi.org/10.1093/oxfordjournals.epirev.a017917 https://doi.org/10.2471/blt.20.255695 https://doi.org/10.2471/blt.20.255695 https://doi.org/10.2471/blt.20.255695 https://doi.org/10.2471/blt.20.255695 https://doi.org/10.1109/access.2020.3001344 https://doi.org/10.1109/access.2020.3001344 https://doi.org/10.1109/access.2020.3001344 https://doi.org/10.1016/j.ssci.2020.104854 https://doi.org/10.1016/j.ssci.2020.104854 https://doi.org/10.1016/j.ssci.2020.104854 https://doi.org/10.1023/a:1016599411804 https://doi.org/10.1023/a:1016599411804 https://doi.org/10.1371/journal.pone.0120567 https://doi.org/10.1371/journal.pone.0120567 https://doi.org/10.1371/journal.pone.0120567 https://doi.org/10.1155/2011/527610 https://doi.org/10.1155/2011/527610 https://doi.org/10.1038/s41467-020-18050-2 https://doi.org/10.1038/s41467-020-18050-2 https://doi.org/10.1038/s41467-020-18050-2 https://doi.org/10.1038/s41467-020-18050-2 https://doi.org/10.1016/j.idm.2020.08.009 https://doi.org/10.1016/j.idm.2020.08.009 https://doi.org/10.3389/fphy.2020.574514 https://doi.org/10.3389/fphy.2020.574514 https://doi.org/10.3389/fphy.2020.574514 https://doi.org/10.3389/fphy.2020.574514 https://doi.org/10.1086/524065 https://doi.org/10.1086/524065 https://doi.org/10.1086/524065 https://doi.org/10.1086/524065 https://doi.org/10.3390/biology9070156 https://doi.org/10.3390/biology9070156 https://doi.org/10.3390/biology9070156 https://doi.org/10.20944/preprints202004.0546.v1 https://doi.org/10.20944/preprints202004.0546.v1 https://doi.org/10.1051/epjconf/201611001029 https://doi.org/10.1051/epjconf/201611001029 https://doi.org/10.1051/epjconf/201611001029 microsoft word volume 13, issue 1-4 journal of risk analysis and crisis response, 2023, 13(1), 49-69 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i1.355 49 review a systematic literature review on financial support for the development of strategic emerging industries su-juan xu 1,2 and mu zhang 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: october 23, 2022; accepted: march 13, 2023; published: march 31, 2023 abstract: in order to further promote the theoretical research on the effective financial support for the high-quality development of strategic emerging industries, this paper reviews the research status at home and abroad on the mechanism and path, influence effect, support efficiency, support system, problems and countermeasures of financial support for the development of strategic emerging industries. at present, the research on financial support for the development of strategic emerging industries at home and abroad is mainly conducted from the perspective of financial support and financing. under the two perspectives, the influence effect can be subdivided into direct financing, indirect financing and comprehensive financing, and the research on support efficiency can be divided into national and regional studies. the support system can be divided into three aspects: horizontal, vertical, and horizontal and vertical synthesis. problems and countermeasures are subdivided into four aspects: countermeasures, policies, strategies and optimization paths. future research directions include: deepening the mechanism of financial support for strategic emerging industries from the perspective of financing; expand the research of vertical support system for the development of strategic emerging industries. keywords: strategic emerging industries; financial support; literature review 1. introduction under the influence of the global financial crisis in 2008, significant changes have taken place in the industrial system all over the world. to get out of the financial crisis as soon as possible, all countries regard the development of emerging industries as the key point to promote economic development. strategic emerging industries are the combination of emerging technologies and emerging industries. they play a leading role in the overall and long-term development of the economy and society. with huge market potential, strong driving capacity and high comprehensive income, they represent the direction of scientific and technological innovation and industrial development. the development of strategic emerging industries is a major strategic choice made by china with an international perspective and strategic thinking. however, the development of strategic emerging industries cannot be achieved without financial support, which requires a large amount of financial support in the process of development. the financial industry can promote the development of strategic emerging industries by giving full play to the role of financing channels. the state also attaches great importance to the development of strategic emerging industries. the su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 50 central and local governments have successively issued policies on the capital, human resources and intellectual property rights to promote the healthy, sustained and rapid development of strategic emerging industries. at present, the domestic and foreign scholars have abundant researches on financial support for the development of strategic emerging industries, which are carried out theoretically and empirically from different perspectives. the concept of strategic emerging industries was first put forward in china in 2010. since then, many scholars have studied the development of financial support for strategic emerging industries. however, there are no direct studies on strategic emerging industries in foreign countries, but more studies on emerging industries and financial support for emerging industries and economic growth. this paper reviews the domestic and foreign literatures on financial support for the development of strategic emerging industries, and the structure of the remaining parts is arranged as follows: the second part introduces the mechanism and path of financial support for the development of strategic emerging industries. the third part introduces the effect of financial support for the development of strategic emerging industries. the fourth part introduces the support efficiency of finance to the development of strategic emerging industries. the fifth part introduces the financial support system for the development of strategic emerging industries. the sixth part introduces the problems and countermeasures of financial support for the development of strategic emerging industries. part 7 is a brief comment. 2. the financial support mechanism and path of strategic emerging industries 2.1. mechanism study 2.1.1. financial support perspective relevant scholars have studied the financial support mechanism for the development of strategic emerging industries, analyzed the organizational structure and form of financial resources to support the development of economy and industry, and the specific aspects to promote the development of strategic emerging industries. gu haifeng (2013) [1] from the perspective of capital market, established the "one body, two systems" policy support system with capital market as the main body and policy-based financial support mechanism and market-based financial support mechanism as the two systems. pan juan (2013) [2] believes that capital flow to strategic emerging industries through financial support mechanism, resource-oriented mechanism and risk sharing mechanism to optimize the allocation of financial resources and realize the adjustment and allocation of financial elements. shi zhangming (2014) [3] believed that the development of strategic emerging industries could be promoted through three mechanisms: promoting technological innovation, easing financing constraints and improving capital allocation efficiency. fang lai (2018) [4] believes that the mechanism of finance on strategic emerging industries is mainly that finance provides strategic emerging industries with different capital allocation methods, multiple financing channels, and improves the efficiency of capital allocation and risk control. jin guanghui and liu zhiyuan et al. (2016) [5] studied from the perspective of investors and found that policy uncertainty increased the risk perception and ambiguity aversion of investors, which had a negative impact on their sentiment, and investor sentiment became an important medium for policy uncertainty to affect corporate investment. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 51 2.1.2. financing perspective from the perspective of financing, hu haifeng and hu jiya (2011) [6] compared and analyzed the financing mechanisms of strategic emerging industries dominated by the us market, japanese banks and german universal banks. gu haifeng (2011) [7], based on the industry life cycle theory, realized the cultivation goal through the initial optimal allocation of credit resources in the cultivation stage of strategic emerging industries, achieved the development goal through the secondary optimal allocation of credit resources in the development stage, and achieved the upgrading goal through the third optimal allocation of credit resources in the upgrading process. xie qinghe (2013) [8] analyzed the internal mechanism of financial support for the cultivation and development stage of strategic emerging industries, to guide the internal dynamic factors and financial system innovation to promote the development of strategic emerging industries. ma junwei (2013) [9] believed that the financial system affects and supports the development of strategic emerging industries mainly through capital formation, technological innovation and the mechanism of resource allocation. liu guowei and shao yunfei (2021) [10] believe that the game between the innovation subject of strategic emerging industries and the financial support subject is a double-helix coupling relationship including power and opportunity behavior. 2.2. path study 2.2.1. financial support perspective zhao tianyi (2013) [11] designed a combination of the capital chain of science and technology finance, the value chain of technology innovation, and the industrial chain of strategic emerging industries. through policy guidance and institutional innovation, various social capital was invested in strategic emerging industries. tang youwei (2012) [12] believes that financial support for the development of strategic emerging industries can be considered from the aspects of policy system, financing service system, external policy environment, and a financing service system suitable for the characteristics of strategic emerging industries can be established by broadening the channels of enterprise equity financing service, establishing the docking platform of finance and technology, integrating financial resources, and strengthening the cooperation between bank and government and insurance. 2.2.2. financing perspective from the perspective of the insurance industry, tian juanjuan and liang feng (2015) [13] believe that the insurance industry should support the development of strategic emerging industries with diversified investment forms, design special insurance products for them, improve the risk supervision mechanism, and build a trinity financing support and risk guarantee system of "government-strategic emerging industries insurance". 3. the influence effect of financial support on strategic emerging industries 3.1. financial support perspective 3.1.1. direct financing su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 52 li fuyou and yin haifeng (2014) [14] concluded that financial support can promote the development of strategic emerging industries. zhang hongyan and wang lei (2015) [15] studied the quarterly data of more than 400 listed companies in strategic emerging industries and reached the same conclusion: in the short term, there is a one-way causal relationship between the two, and financial support is the reason for the development of strategic emerging industries. in the long run, they cause and affect each other. zhang lu (2016) [16] believes that multi-level capital market plays an important supporting role in accelerating the development of strategic emerging industries, but there is a problem that the proportion of direct financing is small. chen jin and liu xinzi (2016) [17] believed that in the process of supporting the development and internationalization of strategic emerging industries, development finance played a positive role in promoting the development and internationalization of strategic emerging industries by taking advantage of the exemplary role of policy-based finance, risk control, sufficient capital and credit enhancement. some studies have found that financial support has little or no effect on the development of strategic emerging industries. gu xin (2012) [18] investigated the industrial performance effect of the capital market of china's high-end equipment manufacturing industry and found that china's capital market had a negative effect on the development of high-end equipment manufacturing industry in strategic emerging industries. luo jin and zeng lu (2015) [19] took 38 listed enterprises of strategic emerging industries in hunan province as research objects. the research showed that direct financial support had little effect on the promotion of strategic emerging industries in the initial and growth stages. yan junzhou et al. (2020) [20] found that financial support is not conducive to improving innovation efficiency, and financial support has a masking effect between government subsidies and innovation efficiency. 3.1.2. indirect financing wang jing and hu lijun (2019) [21] collected the financial data of 30 listed companies in strategic emerging industries in hubei province from 2010 to 2015. using the gmm method, they studied the influence of bank credit financing, equity financing and retained earnings financing on profitability, and found that different financial support means have different effects. 3.1.3. comprehensive financing huang jiankang et al. (2015 [22], 2016 [23]) believed that although the capital market system was gradually improved, the proportion of direct financing was relatively low. they established a dynamic panel model for empirical analysis of the data of listed companies in strategic emerging industries in zhejiang province and jiangsu province. the results of zhejiang province show that equity financing has a negative effect on the development of strategic emerging industries in zhejiang province, retained earnings financing has a positive effect, and bank credit financing has no significant effect. the research results of jiangsu province show that retained earnings financing and commercial credit financing play a positive role in supporting the development of strategic emerging industries in jiangsu province, while bank credit financing and equity financing play a negative role in supporting the development of strategic emerging industries in jiangsu province, and bond financing plays an insignificant role in the development of strategic emerging industries in jiangsu province. gao linrui et al. (2018) [24] found that indirect financial system dominated by bank credit has a negative effect on the development of strategic emerging industries in gansu province, while su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 53 bond financing as a direct financial support plays a positive role in the development of strategic emerging industries in gansu province, but equity financing as a direct financial support mode plays a negative role. the research results of wang jian and zhang zhuo (2015) [25] show that the direct financial system dominated by capital market has a more stable positive supporting effect than the indirect financial system dominated by bank credit. net fixed assets and net intangible assets have different effects on the development of strategic emerging industries. the labor input level has a significant positive effect on the development of strategic emerging industries. from the perspective of corporate governance, yang ronghai and li yabo (2017) [26] thought from a long-term perspective, the impact shows a steady increase. the impact of indirect finance on the growth of enterprises in emerging industries is higher than that of direct finance, while the impact of direct financial support is steadily increasing. 3.2. financing perspective 3.2.1. direct financing dong xuemei and yang qian (2014) [27] compared the impact of private equity investment on the development of strategic emerging industries. the business performance of enterprises with private equity investment was not better than that of enterprises without private equity investment, and the average profitability was even lower than that of enterprises without private equity investment. tian juanjuan and liang feng (2016) [28] studied the policy-based financing effect of strategic emerging industries, indicating that the financing effect of chinese policy-based loans on strategic emerging industries is not obvious, while that of government venture capital guidance fund is obvious. geng chengxuan and zeng gang (2019) [29] argued that most enterprises face relatively serious financing constraints, and financing constraints caused by cash flow and debt financing have a significant impact on financing efficiency, and the number of direct government subsidies and the proportion of state-owned property rights can help alleviate financing constraints. wang jinfu and zhang yaoling (2019) [30] found that internal financing has the strongest promoting effect on r&d and innovation of strategic emerging enterprises, followed by government subsidies, followed by debt financing and equity financing. r&d and innovation can effectively improve the external financing ability of enterprises, but has little promoting effect on internal financing and government subsidies. lin xiaoling and chen ming (2020) [31] believe that the r&d incentive effect of fiscal subsidies on non-free trade zone enterprises is better than that of free trade zone enterprises, the r&d incentive effect of fiscal subsidies on strategic emerging industries is more effective than that of traditional industrial transformation, and the r&d incentive effect of fiscal subsidies on private enterprises is better than that of state-owned enterprises. 3.2.2. indirect financing shi zhangming et al. (2015) [32] used panel quantile regression method to make an empirical analysis of the easing effect of financial development on the financing constraints of strategic emerging industries. the result showed that strategic emerging industries still had the problem of financing constraints. jiang yongling and lei xiaoyu (2016) [33] found that the involvement of venture capital has a significant positive driving effect on the financing of strategic emerging su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 54 industries. strategic emerging enterprises with venture capital background are more likely to obtain external debt financing represented by short-term debt financing and external equity financing. hu jiya (2020) [34] empirically analyzed the role of external financing methods in the development of strategic emerging industries and found that the variable of bank loan financing rate was negatively correlated with the return on equity, the variable of commercial credit financing rate was positively correlated with the return on equity. 3.2.3. comprehensive financing zhang hongyan and wang lei (2016) [35] investigated the influence of the development of china's banking industry and stock market on the development of strategic emerging industries. the development of banks and stock markets can effectively improve the financing constraints of strategic emerging industries, and the development of stock market can ease the financing constraints of strategic emerging industries more than the development of banks. liu yibei and liu hongwei (2020) [36] argued that only the endogenous financing mode has a significant positive impact on the total factor productivity of enterprises. the equity financing mode of enterprises with higher asset specificity is conducive to improving the total factor productivity of enterprises. the debt financing model will reduce the total factor productivity of enterprises, while the internal financing model is less affected by this. xu k. et al. (2019) [37] constructed an indicator system of financing ecology to analyze the impact of external and internal financing ecology on financing efficiency. the results show that the financing ecology of strategic emerging industries, especially the external financing ecology, is constantly improving. xu k. and geng c. (2020) [38] believe that both financial intermediation and the development of the stock market play an important role in alleviating restrictions on r&d financing, and the development of the stock market can better alleviate restrictions on r&d financing. 3.3. the influence of financing structure from the perspective of financing ling jianghuai and hu wenrong (2012) [39] found that equity capital reduces the operating performance of traditional industries but improves the operating performance of strategic emerging industries, and their scale expansion is more likely to be implemented through equity financing. huang qingshan et al. (2013) [40] found that debt financing had a significant positive correlation with corporate performance, while equity financing had a significant negative correlation with corporate performance, and internal financing had no significant impact on business performance. sun zao and xiao liping (2016) [41] found that internal financing and equity financing have a significant positive effect on the independent innovation of listed enterprises in strategic emerging industries. debt financing has an inhibitory effect on the independent innovation of listed enterprises in strategic emerging industries. pan haiying and hu qingfang (2019) [42] believe that there is an interactive relationship between corporate financing structure and innovation level, but this interaction only exists in enterprises in growth and maturity stages, while enterprises in recession stage only have a one-way influence from financing structure to innovation level. zeng fanrong et al. (2020) [43] found that there was an inverted u-shaped relationship between financing structure and corporate performance in strategic emerging industries, and an inverted u-shaped relationship between financing structure and innovation efficiency. innovation efficiency played a significant role in partially mediating the relationship between financing structure and corporate performance of listed su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 55 companies in emerging industries. 4. the financial support efficiency of strategic emerging industries 4.1. financial support perspective 4.1.1. national level the research of xiong zhengde and lin xue (2010) [44] shows that the realization of financial support efficiency is accompanied by the increase of risks. in the case of insufficient indirect financing driving force, direct financing becomes the main support for the development of strategic emerging industries. li donglin (2016) [45] believes that the efficiency of financial support for strategic emerging industries in china has not reached the optimal level, the economies of scale of financial support industry development need to be improved, and the resource allocation function of the financial system still needs to be further optimized and improved. li meng and yang yang (2017) [46] calculated the comprehensive efficiency and dynamic efficiency of financial support for strategic emerging industries respectively, indicating that the financial support efficiency of strategic emerging industries under the new normal of the economy is relatively effective, but the overall downward trend should be vigilant. indirect and direct financing is conducive to the development of strategic emerging industries. liu zhengqiao (2018) [47] concluded that the number of years of establishment and market competition had a significant positive effect on the financial support efficiency. corporate scale and asset-liability ratio have significant negative effects on financial support efficiency, while ownership concentration degree and ownership balance degree have certain positive effects on financial support efficiency. ren zhengyu (2021) [48] takes listed companies in china's strategic emerging industries from 2011 to 2017 as the research object. the results show that the financial support efficiency of listed companies in strategic emerging industries is low, and the comprehensive efficiency shows a u-shaped trend of first decreasing and then rising. scale efficiency is higher than technical efficiency. from the strategic emerging industry of seven sub-sectors, ma junwei (2013 [49], 2014 [50]) believed that from the perspective of sub-industries, there were differences in the financial support efficiency of seven strategic emerging industries. the efficiency of resource allocation in the development process is not optimal. li meng and wang anqi (2016) [51] also believe that the technical efficiency of china's strategic emerging industries is still relatively low. although the efficiency of financial support for strategic emerging industries has been on the rise since the new normal of the economy, there are great differences among the seven major industries. xiong zhengde et al. (2011) [52] believed that it was difficult to achieve financial support efficiency in the three industries of highend equipment manufacturing, new generation information technology and biology, and the influencing factors of financial support efficiency highlighted the characteristics of stages. zhai huayun (2012) [53] showed that the equity financing efficiency of energy conservation and environmental protection, new energy, new materials and information technology industries increased greatly in 2011, while the equity financing efficiency of high-end technology equipment industry remained at a high level although it did not increase. and the efficiency of equity financing of biological and new energy automobile industry decreased slightly. 4.1.2. regional level su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 56 some scholars have studied the financial support efficiency of strategic emerging industries in some regions. tan zhongming et al. (2015) [54] uesd relevant data of 105 listed companies in jiangsu, zhejiang and shanghai in emerging industries, and they found that to improve the development efficiency of financial support industries, technical efficiency and scale efficiency of the industries must be improved at the same time. li donglin (2016) [55] used data envelopment method (dea) combined with malmquist model to measure the financial support efficiency of the capital market for listed companies in the field of strategic emerging industries in shanghai and shenzhen, and found that the application of high and new technologies in the financial field and the reengineering of products and business processes played a good supporting role in the development of strategic emerging industries in china. ma junwei and wang jianhua (2019) [56] took the yangtze river delta region as the research object, the results showed that the financial support efficiency of the development of strategic emerging industries presented a large inter-provincial difference, and technical factors were significantly correlated with the efficiency. chen xiaorong et al. (2020) [57] took 237 listed companies of strategic emerging industries in beijing, tianjin and hebei as research samples, tested the efficiency of financial support and its dynamic changes by using dea model and malmquist productivity index. the research results found that the strategic emerging industries in beijing, tianjin and hebei had not yet realized the optimal allocation of financial resources and the overall efficiency of financial support showed a downward trend. the overall technical efficiency is low because of the low scale efficiency, and the overall efficiency of financial support is reduced because of the delay in updating the technical level. some studies take a province as the research object and find that the financial support efficiency of most enterprises is low or ineffective. cheng li (2014) [58], deng yan and lu pengguang (2016) [59] took the strategic emerging industries in guangdong province as the research object to measure their financial support efficiency. the research showed that the average comprehensive efficiency of financial support for the strategic emerging industries in guangdong province was low. wang shuhua (2016) [60] calculated the financial support efficiency of strategic emerging industries in gansu province based on dea model, and found that the financial support efficiency of biological industry and new energy industry was the highest, while other industries were not ideal, mainly because of low pure technical efficiency. ma junwei and wang jianhua (2016) [61] studied the financial support efficiency of strategic emerging industries in jiangsu province and its influencing factors, and the results showed that the comprehensive efficiency of financial support for strategic emerging industries in jiangsu province was not optimal, and the overall efficiency trend was also declining. fang lai (2018) [4] took 20 listed companies with the characteristics of strategic emerging industries in gansu province as the research object and found that the capital allocation of strategic emerging industries in gansu province was basically in an inefficient state, and the development of strategic emerging industries in gansu province was mainly restricted by the financing dilemma. 4.2. financing perspective 4.2.1. national level lv zhiyuan and zhang mu (2020) [62] believe that in addition to common financial and nonfinancial indicators, big data indicators of baidu index play a prominent role in influencing the equity financing efficiency of listed companies in strategic emerging industries. wang qiong and geng su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 57 chengxuan (2017) [63] proposed a six-stage dea model based on the four-stage dea model, considering the information of input relaxation variables and the influence of external environment and random factors. the empirical results show that the financing efficiency of most enterprises is still at a low level. efficiency differences among chinese provinces also provide another support for the environmental impact of strategic emerging industries. from the perspective of subdivided industries, huang yaxi and zhang mu (2017) [64] show that equity financing efficiency of listed companies in strategic emerging industries is low, different industries have different degrees of input redundancy and output shortage, and the development of equity financing efficiency is unbalanced among different industries. zeng gang and geng chengxuan (2019) [65] studied the internal motivation of insufficient financing efficiency of enterprises. enterprises can improve financing efficiency by improving financing structure and reducing investment redundancy. among the seven sub-industries, the biological industry has the highest efficiency, while the new energy automobile industry has the lowest efficiency. xiong zhengde et al. (2014) [66] showed that the debt financing efficiency of listed companies in the new energy automobile industry was low, and the relatively low debt financing efficiency was the main reason. with the passage of time, the two-stage efficiency of most listed companies in the new energy automobile industry gradually tends to double efficiency. chen xiangyuan and wang ying (2020) [67] showed that biomedicine industry and energy conservation and environmental protection industry had the highest comprehensive level, while high-end equipment manufacturing industry and new energy industry had the lowest financing efficiency. li sisi and zhang mu (2020) [68] measured the equity financing efficiency of listed companies in strategic emerging industries, analyzed the equity financing efficiency in different years and compared it among seven industries, and found that the equity financing efficiency of listed companies in strategic emerging industries was generally low, mainly distributed in a low efficiency range. the pure technical efficiency and scale efficiency are mostly distributed in the higher efficiency range every year. the equity financing efficiency has unbalanced development among industries. 4.2.2. regional level li jingwen et al. (2014) [69] took 51 listed companies of strategic emerging industries in beijing as samples and found that the financing efficiency of strategic emerging industries in beijing was generally low. wang qiong and geng chengxuan (2016) [70] carried out static and dynamic evaluation on the financing efficiency of 29 listed companies in strategic emerging industries in jiangsu province, and the research showed that the improvement of financial ecology and innovation environment could significantly improve the financing efficiency. the overall financing efficiency showed an upward trend, but the average annual growth was slowing down. zeng gang and geng chengxuan (2018) [71] conducted static and dynamic evaluation on the financing efficiency of strategic emerging industries in beijing-tianjin-hebei, and the results showed that the average financing efficiency of the strategic emerging industries in beijing-tianjin-hebei was small, with significant capital input overcapacity. the average financing efficiency of the strategic emerging industries in tianjin was the highest, followed by hebei, and the lowest was beijing. 5. the financial support system of strategic emerging industries 5.1. financial support perspective su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 58 5.1.1. horizontal the construction of financial support system for strategic emerging industries needs the combination and coordination of finance and finance. he jiye (2016) [72] and hu zhaoju (2016) [73] believe that the construction of financial support system for strategic emerging industries needs to strengthen the coordination between finance and finance, strengthen macro-policy regulation and control, build a financial risk rating system, cultivate compound financial talents, and build a risk investment and risk avoidance mechanism. the policy system of strategic emerging industries should be effectively connected with the policy system of financial support. ma junwei and wang jianhua (2018) [74] believe that the "going out" strategy can generate many cross-border and offshore financial service demands, increase the utilization mode and demand of foreign capital, and build a policybased financial support, risk-oriented financial support, and a systematic and multi-level integrated financial support system at home and abroad. 5.1.2. vertical related scholars have studied from the perspective of industry life cycle. gu haifeng (2011) [7578], based on the perspective of market-based finance and policy-based finance, believed that the development of strategic emerging industries supported by market-based finance and policy-based finance includes the cultivation process, development process and upgrading process. the cultivation process and development process pay attention to giving full play to the supporting role of marketbased finance and policy-based finance to guide the promoting effect of internal dynamic factors on the scale of industrial development. the upgrading process should pay attention to the role of market finance and policy finance in supporting technological innovation. zhang zhuo (2012) [79] believes that in the growth stage of enterprises, a financial support system with bank credit, capital market and private financing as the theme should be built, and the role of private financing should be played to provide funds for the growth stage. liu hongchang and yan shuai (2013) [80] built a multi-level and diversified financial support system based on the financing needs of strategic emerging industries in different cycles, including the support of credit market and capital market, as well as the collaborative support of policy finance and commercial finance. yan zeying (2014) [81] integrated various financial funds according to the capital requirements characteristics of strategic emerging enterprises at different stages of independent innovation, combined with the financial and financial support modes of strategic emerging enterprises, and built a financial and financial linkage support mode for strategic emerging enterprises. from the perspective of enterprise life cycle, li yabo (2018) [82] explored the financial support factors most consistent with different stages of the life cycle, and found that enterprises in the growth and maturity stages tend to prefer indirect finance, while enterprises in the later maturity or recession stage are more willing to choose the form of equity to get financial support, and direct financing plays a more significant role in this stage. 5.1.3. horizontal and vertical synthesis hu haifeng and sun fei (2010) [83] proposed that the supporting role of finance in the cultivation of strategic emerging industries could be fully brought into play through policy means such as establishing the coordination mechanism between finance and fiscal and building a multilevel financial support supply system. zhao tianyi (2013) [11] designed a science and technology finance support system for strategic emerging industries from the perspectives of public science and su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 59 technology finance, mixed technology finance and market science and technology finance. hu chi (2014) [84, 85] believes that multi-form and multi-channel investment and financing systems and institutional arrangements should be built to adapt to the different financial needs of different stages of the development of strategic emerging industries and different industrial chain links, and effectively transform them dynamically. 5.2. financing perspective 5.2.1. horizontal hu jiya (2020) [86] divides the seven strategic emerging industries into three tiers. the three industries in the first tier mainly rely on financial support and equity financing, the two industries in the second tier mainly rely on financial support, and the two industries in the third tier tend to rely on debt financing and financial support. 5.2.2. vertical based on the perspective of industrial development, deng yan et al. (2011) [87] studied the financing mode of strategic emerging industries mainly from the development cycle of strategic emerging industries, and believed that different development cycles and different industrial chain links have different financing modes, but policy-based financing runs through the whole development mode. liu cunping et al. (2014) [88] proposed financing policies at different stages based on the life cycle theory of emerging industries. lin songchi (2014) [89] proposed the development path of strategic emerging industries in different periods. in the initial stage, direct financing is the main source of government support; in the growth stage, debt financing is the main source of other financing methods; in the mature stage, various financing methods are combined; and in the recession stage, they are supported by their own funds. 6. the financial support problems and countermeasures of strategic emerging industries 6.1. financial support perspective 6.1.1. countermeasures tan zhongming and li zhanqi (2012) [90] put forward some countermeasures and suggestions on innovating the credit business model of commercial banks, establishing science and technology banks, further improving the capital market, developing the bond market of strategic emerging industries, developing new insurance varieties and establishing the insurance mechanism of strategic emerging industries. ma junwei (2013) [91] proposed to realize the effective combination of finance and strategic emerging industries by improving market mechanism, optimizing financial structure, promoting rational return of equity investment, and innovating financial technology or financial products. wang jian and zhang zhuo (2014) [92] put forward suggestions from the aspects of industrial development, industrial process and industrial objectives, including whole-process financial support, exploring diversified financing channels, giving full play to the financing guarantee function of policy-based guarantee institutions and innovating financial support modes. wang hui (2019) [93] put forward countermeasures and suggestions on financial support for the development of strategic emerging industries from three aspects: increasing the scale of financial su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 60 support, optimizing the structure of financial support and improving the efficiency of financial support. from the perspective of commercial banks, relevant scholars put forward relevant countermeasures for financial support of strategic emerging industries. liu zhibiao (2011) [94] believes that commercial banks should increase credit input, adjust credit structure, accelerate innovation of scientific and technological financial products and services, and realize strategic transformation of commercial banking business while supporting the transformation of economic development mode. lv tie and yu jian (2012) [95] believe that financial innovation is needed to further transform the traditional financing mode and create new financing mode to promote the development of strategic emerging industries more effectively. commercial banks should be guided to increase substantive support for strategic emerging industries, expand financing channels and tools, and increase policy support. chen chen (2013) [96], aiming at the problem of the banking industry supporting the development of strategic emerging industries, proposed countermeasures such as strengthening policy support, improving bank credit process management, perfecting risk sharing mechanism and external incentive mechanism. in view of financial support for strategic emerging industries in some provinces, liu jianmin et al. (2012) [97] put forward countermeasures and suggestions to support and promote the development of strategic emerging industries in hunan, including strengthening government guidance, giving full play to the initiative of financial institutions, improving capital market and financial infrastructure construction. in view of the development of strategic emerging industries in jiangsu province, mao zesheng and wang hongmian (2012) [98] suggested that jiangsu province should continue to increase government investment to actively develop venture capital, innovate the credit mechanism of commercial banks, vigorously support the listing of enterprises, and try to improve various supporting measures to promote the development of strategic emerging industries. she zhihong (2013) [99] proposed the following optional countermeasures to promote the healthy development of strategic emerging industries in wuhan: establishing a green channel to increase the supply of financial support resources, strengthening the supporting and guiding role of policy-based financial support, and optimizing the operation mode of commercial financial institutions. 6.1.2. policy some scholars put forward policy suggestions from various aspects. based on the life cycle theory, gu haifeng (2011) [100, 101] designed policies for different development stages of strategic emerging industries. he designed financial support policies for the evolution of strategic emerging industries from the level of marketable direct finance and market-based indirect finance, and designed ideas for financial support for the development of strategic emerging industries. li ming (2011) [102] put forward policy suggestions on encouraging financial service innovation, expanding financing channels, optimizing financing environment and strengthening coordination among government departments. yuan qingsong (2011) [103] studied the credit policies of commercial banks to support the development of strategic emerging industries. according to the development status and prospect of strategic emerging industries, different measures should be taken for different industries, ideas and products should be innovated, and professional talents and teams should be cultivated. lv tie and yu jian (2012) [104] believed that there was a mismatch between the industrial characteristics and financial demands of strategic emerging industries and the traditional financing su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 61 mode, and financial innovation was needed to further transform the traditional financing mode and create new financing mode, to promote the development of strategic emerging industries more effectively. sun xiaojuan (2013) [105] put forward specific policy suggestions, such as innovating the way of financial investment in science and technology, improving the risk investment mechanism, designing the standards of the open market, improving the bond and bill market, encouraging the investment of listed companies through policies, innovating the financing service system of commercial banks, and promoting the cooperation between finance and science and technology. relevant scholars put forward policies from the perspective of capital market. liu hongchang and yan shuai (2013) [80] proposed that the policy orientation of financial support for the development of strategic emerging industries includes the establishment of multi-level capital market and venture capital market, the establishment of a coordination mechanism between finance and finance, the innovation of financial supply mode, and the strengthening of legislation. xia zhe and wang xiaodong (2014) [106] proposed relevant financial innovation policy suggestions: develop equity investment and strengthen the direct financial system, focus on the government's guiding funds for entrepreneurship, industry and mergers and acquisitions, coordinate policy-based financial instruments and capital markets, build regional multi-level bond markets, improve the intermediary service system for investment and financing, and promote indirect financing of new industries, innovate financial derivatives and consumer credit models. zhang lu and gu xiaogang (2014) [107] explored and designed specific policies and measures for the venture capital market to support the front-end cultivation of strategic emerging industries from the aspects of perfecting the multi-level capital market system, constructing the scientific and technological financial system, optimizing the operating environment of venture capital and improving the supporting policies and incentive mechanism. kan jingyang (2017) [108] put forward some countermeasures and suggestions to develop investment and loan linkage, venture capital leasing and bill pool business, and improve the financing function of incubator. 6.1.3. strategy hu jianbo and gu shikui (2012) [109] proposed strategies to boost china's strategic emerging industries, strengthen institutional innovation, accelerate and improve green credit, broaden financing channels, and optimize human resource allocation. zan donghai (2012) [110] believed that financial support for science and technology should be strengthened in order to promote the process of property rights, marketization and industrialization of proprietary intellectual property achievements in strategic emerging industries. in view of the financial service system in wuhan, yu zhen et al. (2012) [111] proposed that financial agglomeration should be further adopted to strengthen the construction of financial organizations and service platforms, develop multi-level direct financing market, improve the construction of credit system, and further promote the cluster development of strategic emerging industries in the east lake demonstration zone. taking hunan as an example, zeng xiandong (2012) [112] made an empirical analysis of the development status and financial support of strategic emerging industries. the supporting policies and environment of financial services still need to be improved, the corresponding strategies are put forward in the aspects of bank credit management system, investment and financing channels, enterprise management behavior and intermediary service system. chu haiying (2014) [113] put forward strategic suggestions on the support of science and technology finance for the development of su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 62 strategic emerging industries in inner mongolia. it is necessary to give full play to the role of financial finance, promote the development of bank credit technology, make full use of capital market and venture capital, promote science and technology insurance services, and improve the science and technology financial service platform. 6.2. financing perspective 6.2.1. countermeasures hu jiya (2013 [114], 2014 [115]) believed that the imperfect financing service system and the uneven distribution of financing funds are the most prominent problems in the financing process of china's strategic emerging industries. expanding financing methods, strengthening financial innovation and strengthening financial supervision are the key to solve the financing problems of china's strategic emerging industries. it is suggested to release the policy restrictions, actively encourage the participation of various financing subjects, vigorously expand financing channels, improve the policy system and market system, make full use of social capital to realize the optimal allocation of funds, guide the combination of venture capital and strategic emerging industries to improve the financing mechanism of strategic emerging industries. bu xiaoning and huang ruliang (2013) [116] put forward relevant countermeasures for the establishment of credit guarantee system for strategic emerging industries: the guarantee risk compensation fund for strategic emerging industries can be established, the re-guarantee mechanism of guarantee institutions can be improved, and the guarantee cooperation alliance for strategic emerging industries can be established. based on the game analysis of "harsanyi transformation", hu jiya (2021) [117] believes that the establishment of information exchange platform and penalty mechanism for breach of contract can partially solve the problem of information asymmetry, and proposes to reduce government discount, establish information exchange platform and risk mitigation mechanism, carry out multi-party cooperation mode, and build reward and penalty mechanism to reduce moral hazard of credit subjects. 6.2.2. optimization paths gu shengzu et al. (2014) [118] believe that it is very difficult for traditional financing channels to obtain financial support, and equity investment and financing chain dominated by angel investment, venture capital and private equity investment should be constructed to provide financial support. private capital investment channels should be further expanded to promote the healthy development of angel investment, venture capital and private equity investment, and a complete equity investment chain should be constructed. and promote institutional innovation to improve the availability and effectiveness of equity financing in strategic emerging industries. zhang zongcheng and wang yun (2014) [119] analyzed the current situation and existing problems of investment and financing development of strategic emerging industries, and then proposed the ideas of fiscal, tax and financial support policy support and strengthening development financial support, to improve the investment and financing mechanism of strategic emerging industries. hu jiya (2019) [120], aiming at the problems of financing structure imbalance, capital resource mismatch, slow institutional construction and insufficient financial innovation in the financing process of strategic emerging industries, put forward suggestions: financial support should be phased and specific, investment and financing companies should be actively established in strategic emerging industries, capital market system su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 63 should be improved, and innovation should be strengthened to vertical deepening and systematic development. 7. brief comment since the concept of strategic emerging industries was first proposed by the chinese government, there is almost no direct study on financial support for the development of strategic emerging industries in foreign literature. the object of foreign research is emerging industries, mainly starting from a certain financial market body or a certain financing method, to study the impact of financial support on the development and innovation of emerging industries or small and medium-sized enterprises, the relationship between financial support and economic development, the relationship between banking and the development of emerging industries, the relationship between venture capital and the development of emerging industries. therefore, there are many indirect literatures on financial support for the development of strategic emerging industries. domestic scholars mainly from the theoretical and empirical aspects of financial support for the development of strategic emerging industries. in theory, it focuses on the study of the mechanism and path of financial support for the development of strategic emerging industries, discusses the importance of financial support for the development of strategic emerging industries, studies the existing problems and puts forward countermeasures and suggestions accordingly, but rarely studies the financial system from the perspective of financial structure. most of them put forward relevant countermeasures without indepth analysis of the reasons for doing so. in terms of empirical research, efficiency measurement tools such as dea and sfa are mainly used to measure the efficiency of financial support, and relevant data are used to build models to study the impact of financial support on the development of strategic emerging industries. however, most relevant research data in china are based on the influence of listed companies on sub-industries, which leads to insufficient representation of research objects. moreover, research is mostly conducted on single factors, and there are few studies on the supporting effects of different ways. in the theory and practice of financial support for strategic emerging industries, more attention may be paid to the relevance of financial support for the development of strategic emerging industries, the space for innovation and risk control in the practice of financial support for strategic emerging industries in the future. funding: this research was funded by the regional project of national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] gu haifeng. industrial life cycle, evolution of strategic emerging industries and financial support: based on the analysis of capital market [j]. theory journal, 2013 (04): 56-60, doi: https://doi.org/10.14110/j.cnki.cn-37-1059/d.2013.04.011. [2] pan juan. analysis on the mechanism of financial support for the development of strategic emerging industries [j]. journal of commercial economics, 2013 (10): 121-122. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 64 [3] shi zhangming. research on the mechanism of financial supporting to strategic emerging industry based on financial development [d]. wuhan university of technology, 2014. [4] fang lai. financial support mechanism and efficiency evaluation of the development of strategic emerging industry: a case study of gansu province [j]. journal of harbin university of commerce (social science edition), 2018 (02): 28-37. [5] jin guanghui, liu zhiyuan, hua guiru. policy uncertainty, investor sentiment and corporate investment: an empirical study on strategic and emerging industry [j]. journal of central university of finance and economics, 2016 (05): 60-69. [6] hu haifeng, hu jiya. the comparative analysis and revelations on the financing mechanisms of strategic emerging industries in u.s., japan and germany. [j]. economic theory and business management, 2011 (08): 62-74. [7] gu haifeng. research on the mechanism and policy of credit system supporting the evolution of strategic emerging industries [j]. contemporary economic research, 2011 (08): 78-82. [8] xie qinghe. research on chinese financial support for the development of strategic emerging industries [j]. new finance, 2013 (02): 54-57. [9] ma junwei. internal mechanism of financial support for development of strategic emerging industries [j]. science and technology management research, 2013, 33 (17): 16-19. [10] liu guowei, shao yunfei. study on the financial support mechanism and t-s evolutionary game of the strategic emerging industries. [j]. operations research and management science, 2021, 30 (04): 87-95. [11] zhao tianyi. research on the path and system of financial support for science and technology in strategic emerging industries [j]. science and technology progress and policy, 2013, 30 (08): 63-67. [12] tang youwei. financial support strategic emerging industry development path to study [j]. china national conditions and strength, 2012 (08): 9-11. doi: https://doi.org/10.13561/j.cnki.zggqgl.2012.08.016. [13] tian juanjuan, liang feng. financing risks of strategic emerging industries and ways to support insurance industry [j]. sub national fiscal research, 2015 (11): 84-89+96. [14] li fuyou, yin haifeng. an empirical research on financial support for the development of strategic emerging industries based on granger test and co-integration analysis [j]. science and technology progress and policy, 2014, 31 (15): 53-58. [15] zhang hongyan, wang lei. dynamic correlation of chinese strategic emerging industries and financial support -based on the shanghai and shenzhen a-share listed companies' empirical data [j]. the theory and practice of finance and economics, 2015, 4(06): 14-19. doi: https://doi.org/10.16339/j.cnki.hdxbcjb.2015.06.003. [16] zhang lu. research on multi-level capital market supporting the development of strategic emerging industries -based on the shenzhen emerging index sample [j]. communication of finance and accounting, 2016 (02): 14-16. doi: https://doi.org/10.16144/j.cnki.issn1002-8072.2016.02.005. [17] chen jin, liu xizi. research on development finance to support the development and internationalization of strategic emerging industries [j]. modern management science, 2016 (12): 15-17. [18] gu xing. on the promotion effect of multi-level capital market on strategic emerging industries [j]. journal of commercial economics, 2012 (30): 121-122. [19] luo jin, zeng lu. research on interactive dynamic financial support for strategic emerging industries: based on empirical analysis of hunan province [j]. finance and accounting monthly, 2015 (05): 102-105. doi: https://doi.org/10.19641/j.cnki.42-1290/f.2015.05.026. [20] yan junzhou, qi niannian, tong chao. how do government subsidies and financial support influence innovation efficiency? -empirical evidence from listed companies of strategic emerging industries in china [j]. soft science, 2020 (12): 41-46. doi: https://doi.org/10.13956/j.ss.1001-8409.2020.12.07. [21] wang jing, hu lijun. research on the influence of financial support on the development of strategic emerging industries: evidence from listed companies in hubei province [j]. hubei social sciences, 2019 (01): 56-61. doi: https://doi.org/10.13660/j.cnki.42-1112/c.014947. [22] huang jiankang, zhao zongyu, shi jiamin, jiang meihui. research on financial support for the development of strategic emerging industries in jiangsu province -based on the empirical analysis of listed companies [j]. science and technology management research, 2016, 36 (02): 95-100. [23] huang jiankang, shi jiamin, huang binbin. a study on financial support for the development of strategic emerging industry in zhejiang province--an empirical analysis based on the dynamic panel of 30 listed companies [j]. east china economic management, 2015, 29 (12): 26-31. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 65 [24] gao linrui, li shenglan, zhang jianqi. financial support for the influence of the strategic emerging industry development in gansu province [j]. research on development, 2018 (01): 124-131. doi: https://doi.org/10.13483/j.cnki.kfyj.2018.01.020. [25] wang jian, zhang zhuo. the effect of financial support on the development of strategic emerging industries: an empirical analysis based on chinese listed companies [j]. the theory and practice of finance and economics, 2015, 4 (04): 62-67. doi: https://doi.org/10.16339/j.cnki.hdxbcjb.2015.04.011. [26] yang ronghai, li yabo. the source analysis of the financial support to the enterprise management for the strategic emerging industry [j]. soft science, 2017, 31 (01): 47-51. doi: https://doi.org/10.13956/j.ss.10018409.2017.10.11. [27] dong xuemei, yang qian. business performance and private equity financing of strategic emerging industries: based on gem data [j]. finance and economy, 2014 (03): 68-70. doi: https://doi.org/10.19622/j.cnki.cn36-1005/f.2014.03.020. [28] tian juanjuan, liang feng. policy-oriented financing effect analysis of the strategic emerging industry in china [j]. rural financial research, 2016 (04): 31-36. doi: https://doi.org/10.16127/j.cnki.issn10031812.2016.04.006. [29] geng chengxuan, zeng gang. government subsidies, financing constraints and financing efficiency of strategic emerging industries -empirical research based on bilateral random boundary model [j]. modernization of management, 2019, 39 (04): 5-8. doi: https://doi.org/10.19634/j.cnki.11-1403/c.2019.04.002. [30] wang jinfu, zhang yaoting. research on the interaction between financing activities and research and innovation from the perspective of corporate ownership -an empirical study of 543 listed companies based on strategic emerging industries [j]. science and technology progress and policy, 2019, 36 (08): 83-92. [31] lin xiaoling, chen ming. how does external financing effect innovation input in strategic emerging industries? [j]. credit reference, 2017, 38 (12): 76-82. [32] shi zhangming, cao haijuan, zhang zhangcheng. discussion on the relationship between financing constraints, financial development and strategic emerging industries [j]. journal of commercial economics, 2015 (21): 96-98. [33] jiang yongling, lei xiaoyu. spillover effects of venture capital on financing strategic emerging industries [j]. science and technology management research, 2016, 36 (14): 160-164. [34] hu jiya. an empirical study on external financing model and business performance of strategic emerging industries: based on panel data of 120 listed companies in strategic emerging industries [j]. academia bimestris, 2020 (01): 141-149. doi: https://doi.org/10.16091/j.cnki.cn32-1308/c.2020.01.019. [35] zhang hongyan, wang lei. financial development, financing constraints and the development of strategic emerging industries [j]. journal of beijing technology and business university (social sciences), 2016, 31 (01): 94-101. doi: https://doi.org/10.16299/j.1009-6116.2016.01.011. [36] liu yibei, liu hongwei. asset specificity, financing model and enterprise total factor productivity: based on the empirical analysis of listed companies in china' s strategic emerging industries [j]. journal of xiangtan university (philosophy and social sciences), 2020, 44 (01): 98-103. doi: https://doi.org/10.13715/j.cnki.jxupss.2020.01.016. [37] xu k., geng c., wei x. research on financing ecology and financing efficiency of strategic emerging industries in china [j]. journal of business economics and management, 2019, 20 (02), 311-329. https://doi.org/10.3846/jbem.2019.9592. [38] xu k., geng c. financial development, financing constraints and financing efficiency of strategic emerging industry [j]. journal of technical economics and management, 2020 (04): 102-106. [39] ling jianghuai, hu wenrong. firm size, capital structure and operating performance comparative study on traditional industries and strategic emerging industries [j]. finance and trade economics, 2012 (12): 7177. doi: https://doi.org/10.19795/j.cnki.cn11-1166/f.2012.12.012. [40] huang qingshan, deng yan, zhao tianyi. the relationship between financing structure and business performance of strategic emerging industries: a case study of listed companies in the pearl river delta [j]. friends of accounting, 2013 (18): 44-48. [41] sun zao, xiao liping. financing structure and innovation -evidence from a-share of listing corporation of strategic emerging industries in china [j]. economic theory and business management, 2016 (03): 45-58. [42] pan haiying, hu qingfang. a research on the interaction effect between enterprise financing structure and innovation level from the perspective of life cycle: based on the empirical evidence of a-share listed companies in strategic emerging industries [j]. journal of nanjing audit university, 2019, 16 (04): 81-92. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 66 [43] zeng fanrong, zhang xuedi, fang yu. financing structure, innovation efficiency and firm performance: an empirical study of listed companies in strategic emerging industries [j]. communication of finance and accounting, 2020 (01): 79-84. doi: https://doi.org/10.16144/j.cnki.issn1002-8072.2020.01.015. [44] xiong zhengde, lin xue. research on the efficiency and its factors of financial support about listed companies in strategic emerging industry [j]. business and management journal, 2010, 32 (11): 26-33. doi: https://doi.org/10.19616/j.cnki.bmj.2010.11.007. [45] li donglin. research on financial support efficiency of shanghai and shenzhen stock markets for listed companies in strategic emerging industries: based on dea-malmquist index decomposition [j]. new finance, 2016 (09): 27-31. [46] li meng, yang yang. research on financial support efficiency and influence factors of strategic emerging industries under the new normal [j]. reform of economic system, 2017 (01): 129-135. [47] liu zhengqiao. the research of the financial support efficiency measure and its effecting factors of china's strategic emerging industries [d]. hunan university, 2018. [48] ren zhengyu. the research of the financial support efficiency of strategic emerging industries: based on the evidence of listed companies in seven industries [j]. communication of finance and accounting, 2021 (04): 160-163. doi: https://doi.org/10.16144/j.cnki.issn1002-8072.2021.04.031. [49] ma junwei. research on the difference of financial support efficiency of seven strategic emerging industries and its influencing factors: based on empirical evidence of listed companies [j]. reform of economic system, 2013 (03): 133-137. [50] ma junwei. efficiency measure of financial support strategic emerging industry in china [j]. statistics and decision, 2014 (05): 153-155. doi: https://doi.org/10.13546/j.cnki.tjyjc.2014.05.040. [51] li meng, wang anqi. evaluation and analysis of financial support efficiency of strategic emerging industries under the new normal of economy [j]. inquiry into economic issues, 2016 (05): 83-87. [52] xiong zhengde, zhan bin, lin xue. the efficiency of financial support about strategic emerging lndustry based on dea and logit model [j]. systems engineering, 2011, 29 (06): 35-41. [53] zhai huayun. financial support efficiency of listed companies in the strategic emerging industries [j]. securities market herald, 2012 (11): 20-25. [54] tan zhongming, tong jie, sheng zhuyun. empirical analysis on the development efficiency of financial support for strategic emerging industries: based on the data of 105 listed companies in jiangsu, zhejiang and shanghai [j]. wuhan finance, 2015 (12): 38-41. [55] li donglin. path exploration to enhance the efficiency of financial support to strategic emerging industries [j]. hebei academic journal, 2016, 36 (06): 209-213. [56] ma junwei, wang jianhua. on efficiency of financial support for strategic emerging industries based on empirical analysis of the yangtze river [j]. forum on science and technology in china, 2019 (10): 52-58, doi: https://doi.org/10.13580/j.cnki.fstc.2019.10.007. [57] chen xiaorong, han jingwang, ren aihua, sun zhongyan. financial support efficiency of strategic emerging industries: a case study of 237 listed companies in beijing-tianjin-hebei region [j]. finance and economy, 2020 (05): 81-88. doi: https://doi.org/10.19622/j.cnki.cn36-1005/f.2020.05.011. [58] cheng li. analysis on financial support efficiency of strategic emerging industries based on dea tobit model [j]. applied mechanics and materials, 2014, 3277 (608-609): 145-150. [59] deng yan, lu pengguang. financial support efficiency of strategic emerging industries: a case study of guangdong province [j]. friends of accounting, 2016 (24): 71-74. [60] wang shuhua. financial support efficiency analysis of strategic emerging industries: based on the empirical study of listed companies in gansu province [j]. journal of commercial economics, 2016 (24): 170172. [61] ma junwei, wang jianhua. efficiency evaluation and influencing factors in financial support of strategic emerging industries in jiangsu province [j]. forum on science and technology in china, 2016 (08): 3137, doi: https://doi.org/10.13580/j.cnki.fstc.2016.08.025. [62] lv zhiyuan, zhang mu. influencing factors of equity financing efficiency of the listed companies in strategic emerging industry based on baidu index[j]. journal of risk analysis and crisis response, 2020, 10 (04):130-137. [63] wang qiong, geng chengxuan. research on financing efficiencies of strategic emerging listed companies by six-stage dea model [j]. mathematical problems in engineering, vol. 2017, 2017. doi: https://doi.org/10.1155/2017/3284657. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 67 [64] huang yaxi, zhang mu. research on evaluation of equity financing efficiency of listed companies in strategic [j]. journal of risk analysis and crisis response (jracr), 2017, 7(04): 189-213. [65] zeng gang, geng chengxuan. empirical study of measurement of financing efficiency and influencing factors of listed companies in strategic emerging industries: based on super sbm-logit model [j]. science and technology management research, 2019, 39(16): 135-143. [66] xiong zhengde, yang fangjuan, wan jun. a study on the efficiency of debt financing of listed company based on two-stage dea: an application to strategic emerging industry of new energy vehicles [j]. the theory and practice of finance and economics, 2014, 35 (05): 51-56. [67] chen xiangyuan, wang ying. research on financing efficiency of china's strategic emerging industries based on super efficiency dea and tobit model [j]. international journal of emerging markets, 2020, 17(02): 485-504. [68] li sisi, zhang mu. equity financing efficiency measurement of listed companies in strategic emerging industries based on dea [j]. journal of risk analysis and crisis response, 2020, 10 (03): 101-112. [69] li jingwen, wang yuchun, yang zhengdong. strategic emerging industries listed companies financing efficiency research in beijing [j]. research on economics and management, 2014 (06): 74-82. doi: https://doi.org/10.13502/j.cnki.issn1000-7636.2014.06.010. [70] wang qiong, geng chengxuan. a study on financing efficiency of jiangsu's listed companies in strategic emerging industries based on the panel data from 2009 to 2014 [j]. east china economic management, 2016, 30 (07): 14-20. [71] zeng gang, geng chengxuan. financing efficiency measure and strategy of collaborative development of beijing-tianjin-hebei strategic emerging industries [j]. forum on science and technology in china, 2018 (12): 142-149 + 172. doi: https://doi.org/10.13580/j.cnki.fstc.2018.12.016. [72] he jiye. theory of financial support system to build strategic emerging industry in china [j]. shandong social science, 2016 (11): 160-164. doi: https://doi.org/10.14112/j.cnki.37-1053/c.2016.11.023. [73] hu zhaoju. construction of financial support system for strategic emerging industries in china [j]. social scientist, 2016 (09): 56-60. [74] ma junwei, wang jianhua. research on financial support for strategic emerging industries under "going out" strategy [j]. journal of commercial economics, 2018 (09): 167-169. [75] gu haifeng. research on financial support system and policy for the evolution of strategic emerging industries: based on the support perspective of market finance [j]. inquiry into economic issues, 2011 (11): 74-78. [76] gu haifeng. on financial support for incubation and growth of strategic new industries [j]. china opening journal, 2012 (04): 71-74. doi: https://doi.org/10.19625/j.cnki.cn44-1338/f.2012.04.023. [77] gu haifeng. state developing and finance supporting for china's strategic emerging industries [j]. securities market herald, 2011 (04): 57-61+66. [78] gu haifeng. the cultivation, upgrading and financial support of strategic emerging industries [j]. reform, 2011 (02): 29-34. [79] zhang zhuo, zhang yuan, hu ting. financial support and optimization path of strategic emerging industries based on business evolution [j]. wuhan finance, 2012 (06): 18-20+8. [80] liu hongchang, yan shuai. financial support and policy orientation for the development of strategic emerging industries [j]. modern economic research, 2013 (01): 60-63+88. [81] yan zeying. analysis on financial support modes of strategic emerging enterprises at different innovation stages [j]. modernization of management, 2014 (02): 54-56. [82] li yabo. financial support on different life cycle stages for firms in strategic emerging industry [j]. journal of industrial technological economics, 2018, 37 (05): 3-10. [83] hu haifeng, sun fei. research on financial support system in the cultivation of strategic emerging industries [j]. expanding horizons, 2010 (06): 24-26. [84] hu chi. financial support for the development of strategic emerging industries during the 12th five-year plan period [j]. economic review journal, 2014 (08): 17-20. doi: https://doi.org/10.16528/j.cnki.221054/f.2014.08.018. [85] hu chi. financial support for the development of strategic emerging industries since the 12th five-year plan [j]. review of economic research, 2014 (27): 27-33. doi: https://doi.org/10.16110/j.cnki.issn20953151.2014.27.009. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 68 [86] hu jiya. study on heterogeneity and financing pattern matching of strategic emerging-panel data of 120 listed companies in strategic emerging industries [j]. journal of social sciences, 2020 (04): 44-57, doi: https://doi.org/10.13644/j.cnki.cn31-1112.2020.04.005. [87] deng yan, li qingyun, li huajun. research on financing model of strategic emerging industries based on industrial development cycle [j]. friends of accounting, 2011 (30): 13-15. [88] liu cunping, et al. study of financing policy on the development of strategic emerging material industry [j]. applied mechanics and materials, 2014, 3468 (644-650): 4848-4851. [89] lin songchi. enterprise life cycle perspective strategic emerging industries financing mode selection [j]. communication of finance and accounting, 2014 (14): 14-15. doi: https://doi.org/10.16144/j.cnki.issn10028072.2014.14.027. [90] tan zhongming, li zhanqi. theory of the development of strategic emerging industry financial support countermeasures [j]. enterprise economy, 2012, 31 (02): 172-175. doi: https://doi.org/10.13529/j.cnki.enterprise.economy.2012.02.031. [91] ma junwei. efficiency of financial support for strategic emerging industries based on malmquist model [j]. journal of xi'an university of finance and economics, 2013, 26 (02): 11-15. doi: https://doi.org/10.19331/j.cnki.jxufe.2013.02.002. [92] wang jian, zhang zhuo. development efficiency measurement and financial support of strategic emerging industries [j]. journal of zhongnan university of economics and law, 2014 (01): 76-81. [93] wang hui. financial support for strategic emerging industries in the new era and countermeasures: based on the perspective of government promotion [j]. theoretical investigation, 2019 (05): 123-128. doi: https://doi.org/10.16354/j.cnki.23-1013/d.2019.05.018. [94] liu zhibiao. technology banking functions construction: research on some critical problems of commercial banks supporting strategic emerging industries development [j]. nanjing journal of social sciences, 2011 (04): 1-7. doi: https://doi.org/10.15937/j.cnki.issn1001-8263.2011.04.017. [95] lv tie, yu jian. practical innovation, existing problems and policy suggestions of financial support for the development of strategic emerging industries [j]. macroeconomics, 2012 (05): 18-26, doi: https://doi.org/10.16304/j.cnki.11-3952/f.2012.05.013. [96] chen chen. the innovations, problems and countermeasures in the development of strategic emerging industries supported by the banking sector [j]. securities market herald, 2013 (06): 21-26. [97] liu jianmin, wu fei, wu jinguang. the research on financial support strategy about strategic emerging industry of hunan province [j]. journal of hunan university (social sciences), 2012, 26 (06): 67-72. [98] mao zesheng, wang hongjin. analysis on financial support for of strategic emerging industries in jiangsu province [j]. science and technology and economy, 2012, 25 (01): 82-85. [99] she zhihong. financial support to promote the development of strategic emerging industries in wuhan health countermeasures [j]. hubei social sciences, 2013 (02): 58-61. doi: https://doi.org/10.13660/j.cnki.421112/c.012081. [100] gu haifeng. study on finance supporting system and policy to strategic emerging industry developing [j]. modern finance and economics-journal of tianjin university of finance and economics, 2011, 31 (09): 76-83. doi: https://doi.org/10.19559/j.cnki.12-1387.2011.09.011. [101] gu haifeng. financial support system and policies for the cultivation of strategic emerging industries [j]. wuhan finance, 2012 (12): 15-18. [102] li ming. giving full play to the role of financial support to promote the development of emerging industries -a survey of the financial support for the development of strategic emerging industries in xiangtan city [j]. wuhan finance, 2011 (11): 57-59. [103] yuan qingsong. support for the development of strategic emerging industry credit policy research [j]. macroeconomic management, 2011 (05): 51-52. doi: https://doi.org/10.19709/j.cnki.11-3199/f.2011.05.020. [104] lv tie, yu jian. problems and countermeasures of financial support for the development of strategic emerging industries [j]. chinese cadres tribune, 2012 (09): 30-32. [105] sun xiaojuan. research on financial support in strategic emerging industries development for gansu province [j]. lanzhou academic journal, 2013 (11): 163-168. [106] xia zhe, wang xiaodong. the policy research on financial support for guangzhou's strategic emerging industry [j]. science and technology management research, 2014, 34 (24): 37-40. [107] zhang lu, gu xiaogang. front-end cultivation of strategic emerging industries and policy design of venture capital market [j]. south china finance, 2014 (03): 10-14. su-juan xu and mu zhang / journal of risk analysis and crisis response, 2023, 13(1), 49-69 doi: https://doi.org/10.54560/jracr.v13i1.355 69 [108] kan jingyang. for the development of strategic emerging industry financial support policy research [j]. review of economic research, 2017 (34): 83-88. doi: https://doi.org/10.16110/j.cnki.issn2095-3151.2017.34.012. [109] hu jianbo, gu shikui. the development and financial support strategy of china's strategic emerging industries [j]. practice in foreign economic relations and trade, 2012 (11): 27-31. [110] zan donghai. financial support for intellectual property of the strategic emerging industries [j]. chongqing social sciences, 2012 (08): 73-79. doi: https://doi.org/10.19631/j.cnki.css.2012.08.012. [111] yu zhen, li chunzhi, wu ying. financial support for the development of strategic emerging industries in wuhan: based on the perspective of co-agglomeration [j]. wuhan university journal (philosophy & social science), 2012, 65 (06): 107-111. [112] zeng xiandong. financial support strategy for the rapid development of strategic emerging industries based on the investigation of hunan province [j]. wuhan finance, 2012 (08): 29-31. [113] chu haiying. research on sci-tech finance supportive strategies of the development of strategic emerging industries in inner mongolia [j]. scientific management research, 2014, 32 (04): 68-71. doi: https://doi.org/10.19445/j.cnki.15-1103/g3.2014.04.018. [114] hu jiya. strategic emerging industry development in our country's financing problem research [j]. journal of chinese youth social science, 2013, 32 (06): 108-113. doi: https://doi.org/10.16034/j.cnki.101318/c.2013.06.007. [115] hu jiya. strategic emerging industry in our country the financing present situation, problems and countermeasures [j]. journal of chinese youth social science, 2014 (06): 95-100. doi: https://doi.org/10.16034/j.cnki.10-1318/c.2014.06.046. [116] bu xiaoning, huang ruliang. financing predicament and credit guarantee system reconstruction of strategic emerging industries [j]. reform, 2013 (09): 33-39. [117] hu jiya. the current situation, problems and countermeasures of credit financing in strategic emerging industries game analysis based on "harsanyi transformation" [j]. soft science, 2021, 35 (02): 1-6. doi: https://doi.org/10.13956/j.ss.1001-8409.2021.02.01. [118] gu shengzu, ma junwei, gao mei. the development of strategic emerging industry need perfect equity investment and financing chain [j]. forum on science and technology in china, 2014 (10): 5-10. doi: https://doi.org/10.13580/j.cnki.fstc.2014.10.002. [119] zhang zongcheng, wang yun. research on related issues of improving investment and financing mechanism of strategic emerging industries [j]. wuhan finance, 2014 (02): 18-21. [120] hu jiya. analysis of the financing status of strategic emerging industry and choice of innovation path based on 120 sample enterprise data [j]. social sciences in xinjiang, 2019 (05): 57-70+153. copyright © 2023 by the authors. this is an open-access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 11, issue 4-2 journal of risk analysis and crisis response, 2021, 11(4), 163-175 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v11i4.310 163 article data ecology and accurate portrait: optimization of credit risk system for smes in supply chain finance based on big data technology chenyang wu 1,2, jinyue liu 1,2 and hongmei zhang 1,2,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhm1035@qq.com; tel.: +86-0851-88510575 received: may 18, 2021; accepted: november 20, 2021; published: january 25, 2022 abstract: big data technology can collect data, store it, mine it, and create an accurate portrait. it can assist financial institutions in resolving information asymmetry between banks and enterprises, as well as lowering the likelihood of default of small and medium-sized financing enterprises (smes). the credit risk system for smes in supply chain finance can realize “visualization” management of credit risk with the help of open public data in government affairs, collaborative development of various technologies, and the establishment of an ecological platform with transparent and accurate data portraits. the platform with accurate risk warning capability can reduce the risk monitoring cost and improve the risk management efficiency of financial institutions. the core enterprises are more willing to grant credit to smes through the big data technology supervision platform, which significantly improves the financing efficiency of smes. moreover, a better financing credit circumstance also could improve transaction efficiency of enterprises and deeply connect the business relationship between enterprises. the main conclusion of this research: big data technology has a significant impact on supply chain in the digital economy era. firstly, big data technology can identify credit risks accurately, which narrows the "information gap" between financial institutions and supply chain financing enterprises, and lower the likelihood of credit default. secondly, financial institutions can allocate funds accurately based on the “visualization” information provided by the big data platform, and strengthen supervision of the use of funds. lastly, the supply chain finance credit risk supervision system based on big data technology promotes the deep integration of big data and real economy. therefore, in order to ensure the sustainable development of supply chain finance and financing risk management, it is necessary to create a digital ecosystem of supply chain finance with supply chain finance control tower as its core, as well as a supply chain finance credit risk control system based on big data in the context of the continuous development of big data technology. keywords: supply chain finance; big data technology; data ecology; accurate portrait; risk management visualization 1. introduction science and technology promote the economic changes of the times, and every technological innovation would create a new mode of economic operation. in the context of the third wave of information technology marked by big data, cloud computing and the internet of things, the chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 164 economy mode of china is gradually upgrading from resource-led to data-led, and data has become an important factor of production. the state council issued the opinions on building a better institutional mechanism for factor market allocation on march 20, 2020, which explicitly included data as a new factor of production. on october 10, 2021, the "national standardization development outline" clearly put forward the implementation of the new industry standardization pilot project. this policy document also emphasizes that standards should be used to enhance the stability of the industrial chain supply chain, and enhance the comprehensive competitiveness of the industry in the era of digital economy is significant. smes play an important role in economy but they still face many financing problems. the proportion of loans from financial institutions to smes has been increasing year by year, and the total amount of loans has been expanding, while it is still difficult and high-cost for smes to finance. according to the sme loan demand index in the wind database, this index showed an overall upward trend from 2016 to 2020. in recent years, it has become an important topic to improve the regulatory environment of credit risk of supply chain finance through big data, thus improving the dilemma of financing difficulty and high financing cost of smes. as a new financing method based on the upstream and downstream of the industry, supply chain finance can meet the short-term financing needs of enterprises due to real trade, and provide financing services for smes in a flexible manner. in the guiding opinions on actively promoting supply chain innovation and application issued by the general office of the state council in 2017, it is clearly pointed out that supply chain finance should be actively and properly developed, and commercial banks and other financial institutions should be encouraged to assist core enterprises in establishing supply chain finance platforms to provide convenient financing channels for smes. however, it must be acknowledged that all kinds of financial innovation methods, including supply chain finance, cannot avoid the theoretical logic among financial innovation, financial risk, and financial supervision. meanwhile financial supervision usually lags behind, so it is imperative to regulate supply chain finance based on big data technology. while in reality, only a few leading enterprises, such as mybank and webank, have realized supervision of credit risks by utilizing their own mature digital technologies. even if financial institutions have part of data element resources, they cannot give full play to the value of data elements because they have not mastered a series of big data technologies such as collection and acquisition, storage and management, data mining and analysis. according to luo hao [1], big data thinking is gradually becoming an integrated innovation tool applied to the technology supply chain, brand ecological chain, and user experience chain. by building a supply chain finance ecology, big data technology can narrow the "information gap" among the stakeholders in the supply chain, promote credit risk management in supply chain finance, and help smes to manage risk. song hua [2] believed that the relationship between supply chain finance and internet finance, the elements of industrial ecology on which supply chain finance operations rely and the next development trend are important to promote supply chain finance. wang jun [3] implicated that digital technology or digital technology can propel supply chain finance to bridge the data gap, and multiple parties should collaborate to build a supply chain finance ecology that meets the needs of china's industrial development. ding qianlan [4] confirmed that data plays an important role in the development of smart supply chain and proposed the creation of a general framework of digital-driven smart supply chain ecosystem. wen lu [5] pointed out that big data has chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 165 promoted the development process of supply chain integration, and the acquisition, screening, analysis, and integration of business data by enterprises have promoted cross-border integration of member enterprises. gou yanjie [6] confirmed that to innovate agricultural supply chain finance needs to build an agricultural industry ecology of "cross-border, empowerment, integration, symbiosis and synergy", and to aggregate various elements, so that agricultural supply chain finance can complete the evolution from "point" to "ecological chain". further, the supply chain finance could strengthen credit risk management tools. based on the cooperative and symbiotic ecological relationship among supply chain subjects, zhao chengguo [7] establishes a cooperative and symbiotic mechanism of internal and external supply chain finance ecological risk management from the perspective of resource, organization, and knowledge analysis. jing jun [8], based on a case study of the supply chain finance ecosystem mode built by "car payment", showed that the industrial ecology and platform based on financial technology can greatly reduce the transaction costs and risks of supply chain finance. to construct a “control tower” based on big data technology could boost credit risk management of supply chain finance. the supply chain control tower serves as a central hub, serving as a "dashboard" of data flows, key business metrics and events connected across the supply chain. the supply chain control tower allows the supervision platform to gain more comprehensive understanding of the information and solve key supply chain problems in real time. several famous consulting companies, such as gartner, accenture, and capgemini, regard the control tower as seamless, end-to-end, holistic visibility that provides real-time data analysis, prediction, and decision-making to solve problems quickly and build collaborative, consistent, agile, and demanddriven supply chains. wang chuanlei [9] believed that supply chain control tower is an integrated control center of supply chain data, and an effective tool to realize digital transformation. supply chain control tower can realize end-to-end visualization of supply chain through coordinated development of various financial technologies. song hua [10] believed that the intelligent control tower can not only effectively master the operation status of business and assets, but also detect abnormal situations in time, and implement risk early warning, monitor the compliance of supply chain and its financial activities meanwhile to ensure the safety and effectiveness of supply chain finance. tang longji [11] found out that the core framework of applying digital control tower to supply chain was first reported in the white paper industry 4.0 published by pwc in 2016. scholars have much less explored the use of digital technology to embed digital control tower into the core framework of supply chain ecosystem. therefore, the supply chain financial ecosystem coined in this paper is innovative. supply chain finance is still in its early stages in china, and the depth and breadth of credit risk management of supply chain finance need to be improved. the supply chain finance mode has the characteristics of diversified transaction subjects and complicated transaction scenarios. asymmetry information makes it difficult to identify and measure credit risks. this study aims to apply big data technology to the credit risk management of smes in supply chain finance. first to analyze the feasibility of applying big data technology to supply chain finance credit risk management. then take big data technology as the core factor, to design a supply chain financial ecosystem which regards supply chain finance control tower as the core. last to realize intelligent management of credit risks in supply chain finance. 2. analysis of credit risk sources of traditional supply chain finance mode chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 166 2.1. analysis of credit risk sources of accounts receivable financing mode the accounts receivable financing mode is a financing method in which the upstream enterprises pledge to financial institutions the outstanding receivable accounts guaranteed by the core enterprises of the supply chain, in order to obtain liquidity in time. enterprises can adopt this mode to solve capital shortage after selling goods but before receiving payment. credit risk supervision of accounts receivable financing mainly includes the following two aspects: (1) the authenticity of trade is difficult to supervise. the fabrication process of accounts receivable contract is so easy that counterfeit trade contracts directly increase the difficulty of supervision. for example, core enterprises would collude with some of smes to sign false trade contracts to attain credit loan. thus, banks and other financial institutions shall determine the authenticity of transaction contracts firstly when smes submit accounts receivable contracts to prevent such problem. (2) the use of funds is difficult to supervise. the phenomenon of "borrowing to support lending" is serious in smes. in the case of insufficient funds, enterprises often "rob peter to pay paul". if smes use accounts receivable financing to repay other loans, it will be difficult for financial institutions to supervise, which will have a negative impact on the financing availability of supply chain enterprises. therefore, credit risk supervision of accounts receivable financing should attach great importance to trade authenticity and use of financing. 2.2. analysis of credit risk sources of prepayment financing mode the prepayment financing mode emerged to relieve liquidity pressure of downstream smes. enterprises in the downstream of the supply chain often need to pay the core enterprises in advance in order to obtain the raw materials and some other required products. the core enterprises use the right of delivery or materials in transit as the credit basis to grant prepayment financing credit to smes. the credit risk of prepayment financing mainly comes from the sales of goods of smes and the repurchase ability of the core enterprises. under the prepayment financing mode, smes sell goods to obtain payment for the goods, and then pay the bank deposit to obtain the right to take delivery of the goods in the next step, and then obtain more quantities of goods. therefore, in this financing environment, the sales of smes become the key to influence the credit risk of prepayment. for smes, having an efficient sales and marketing mode will not only assist them in increasing their market share, but will also lead to higher financing capacity with high sales volume. if the smes' sales are poor or the goods are not selling due to certain market factors, smes are unable to sell the products and thus cannot obtain stable repayment. at this point, the ability of the core company to execute the repurchase commitment becomes the key to the stable development of the smes under the prepayment financing mode. 2.3. analysis of credit risk sources of inventory pledge financing mode the inventory financing mode can improve the operating conditions of smes. however, banks are generally reluctant to use inventory as pledges because of the characteristics of inventory such as market price variability and difficulty in valuation. through the guarantee of core enterprises and the supervision of storage and logistics, small and medium-sized financing enterprises can realize inventory financing through supply chain finance. in view of the credit problems of traditional inventory pledge financing, it is concluded that the credit risk of inventory pledge financing mode mainly comes from the review of transportation qualification of third-party logistics enterprises and chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 167 the pledge qualification of storage institutions. (1) the qualification of third-party logistics companies in transporting pledged goods. the main problems are whether the logistics company can properly store the pledged goods, the destination of the pledged goods and the ability to assess the value of the pledged goods. after receiving the goods, there are cases of third-party logistics companies running away with the goods. therefore, in order to ensure the effective supervision of credit risk in the inventory pledge financing mode, the transportation qualification of the third-party logistics enterprises should be examined in focus. (2) the pledge qualification of the storage institution. it is mainly manifested in three aspects: the problem of storage institution's ability to keep the pledged goods, the problem of repeated pledges of pledged goods and the ability to assess the value of pledged goods. the loss of value in the process of storage of pledged goods by the storage institution happens from time to time. for the problem of duplicate pledges of pledges, banks and other financial institutions do not have their own warehouses, so banks need to sign supervision agreements with third-party pledge supervisors. since banks and other financial institutions are not able to supervise pledges in real time, they can only rely on the credit of the pledging institution. when the third-party warehousing department colludes with the financing smes, it can cooperate with multiple banks and other financial institutions, and a real pledge can be financed several times, which is a credit default situation of "repeated warehouse receipt financing". 3. feasibility analysis of combining credit risk of supply chain finance with big data technology 3.1. big data technology can solve the difficulties of credit risk management in supply chain finance the difficulty of credit risk management in supply chain finance lies in the limited access to data by risk managers. in recent years, there have been a lot of collapses in supply chain finance, the root cause of which is the serious information asymmetry among the stakeholders in the chain. banks and other financial institutions have limited ability to identify and manage risks, and are unable to understand the actual operating conditions of smes in the supply chain in real time. supply chain finance has gone through three stages. under the supply chain finance 1.0, financial institutions as risk managers have limited access to data and can only assess credit risk through financial data of historical transactions between upstream and downstream enterprises. under the supply chain finance 2.0, the core enterprises can grasp the transaction flow, logistics and information flow, but the core enterprises are unwilling to share these information. and under the supply chain finance 3.0, the third-party professional service platform is trying to build its own data platform. since there is no unified implementation standard, diverse and heterogeneous platforms and data types further exacerbate the barriers to information. 3.2. big data technology drives "visualization" of credit risk management in supply chain finance chinese president xi jinping has proposed, "the in-depth development of digitization, networking and intelligence is enhancing the level of wisdom of social operation and empowering the modernization of national governance capacity." as a productive force, big data technology boosts the evolution of modern economy to intelligence and wisdom. with the continuous improvement and development of information infrastructure, 5g, as a strategic resource in the current global concentrated competition, will widely empower banks and other financial institutions. chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 168 big data technology with the help of new infrastructure such as 5g, block chain, cloud computing and artificial intelligence technology can break through the limitations of time and space and boost the intelligent development of economy. data transaction process visualization has become a basic requirement for supply chain finance credit risk management, and the emergence of big data technology has proposed a new solution for supply chain finance credit risk management. the key technologies of big data include the following aspects: (1) real-time data collection. through the data interface or with the help of web crawler technology, the credit data related to the credit assessment subject can be obtained, ensuring the traceability of the data; (2) transaction process traceability. all transaction processes are stored in the database in real time, and the transaction status of the transaction subject can be traced to ensure the traceability of the data; (3) accurate risk early warning. using data mining technology, the credit evaluation subject's historical data is repeatedly trained, which means that the machine is allowed to simulate various user behaviors. transaction subjects of supply chain finance can be interconnected through data interface interchange or sharing. nowadays, various algorithms are being gradually improved, and big data technology has been able to accurately fit the behavior of users within a certain range, which can make accurate portraits of credit evaluation subjects. combining with historical data of financial operation activities, big data technology can carry out effective risk early warning, meeting the "visualization" demand of supply chain finance risk management. banks and other financial institutions also can obtain valuable credit risk information with the help of big data supervision platform and provide various financial services for supply chain enterprises. 3.3. analysis of the driving effect of big data technology on supply chain finance 3.3.1. strengthen the risk supervisory capacity of financial institutions financial institutions can use big data technology to know about the activity data and comprehensive information data of the financial operation, and effectively reduce the opportunism and moral risk caused by information asymmetry, the difficulty of project evaluation and supervision, and high financing cost in supply chain finance. various financial activities happen in real time, big data technology crawls massive data for data collection, then each stakeholder can obtain information in a zero-delay context. banks and other financial institutions can promote the real-time monitoring and early warning capability to credit subjects in supply chain finance, by retrieving historical credit data, then attaining accurate credit status of customers through machine learning models and set credit risk early warning level respectively. 3.3.2. accelerate the development of cost-effectiveness intensification of supply chain supervision big data technology speeds up breaking down the information barriers of each subject in supply chain finance gradually, and alleviates the asymmetric effect of information among various subjects in supply chain finance gradually. diversified big data technology reduces the acquisition cost and storage cost of each transaction subject. accurate data mining technology strengthens the risk monitoring ability of financial subjects and reduces various credit risk supervision costs, including human resource and audit costs. therefore, in the whole supply chain finance management, using of big data technology has realized the cost reduction and efficiency of risk management for each chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 169 transaction enterprise, and driven the cost-effectiveness intensive development of supply chain financial risk supervision. 3.3.3. promote the in-depth integration of industry and finance supply chain finance is a financing business established between the upstream and downstream of the industrial chain in order to meet the transaction. financial institutions in the traditional mode do not participate in the supply chain transaction business, which leads to the inability to effectively integrate and supervise the transaction flow, logistics and information flow of supply chain finance business. supply chain finance business does not develop completely between the upstream and downstream of the industry. use of big data technology can effectively solve the problem of information asymmetry between financial institutions and smes. its risk warning ability can effectively reduce the cost of risk monitoring of financial institutions, improve the efficiency of risk management of financial institutions, and reduce the probability of default of smes. core enterprises are more willing to grant credit to smes on the big data technology supervision platform, which results in the financing efficiency of smes has been significantly improved. better financing and credit status will reversely increase the transaction efficiency between the upstream and downstream of the industry and deepen the business relationship between enterprises. song hua (2020) confirmed that the real combination of industry and finance is to apply financial thinking to the actual scenes of the industrial chain, so that finance becomes the blood flowing in the industry. with the enhancement of credit risk supervision capability promoted by big data technology, the tight integration of industrial subject and supply chain finance is improved. 4. optimization design of supply chain finance credit risk ecosystem based on big data technology 4.1. overall layout of supply chain finance ecosystem supply chain finance has evolved through three development stages, with the 1.0 mode dominated by financial institutions, the 2.0 mode dominated by core enterprises and the 3.0 mode dominated by third-party professional service institutions. the advancement of big data technology has the potential to accelerate the transition of supply chain finance from financial institutions to an industrial internet technology platform. the credit risk ecosystem of supply chain finance is designed with the concept of "embracing openness and cooperation for mutual benefit" of data elements. this paper expands on the traditional supply chain finance model by introducing the role of a big data platform. the supply chain finance control tower with big data platform can monitor big data of multiple agents and achieve accurate portrait of credit risk of supply chain finance, as shown in figure 1. 4.2. functions of each subject constitution in supply chain finance ecosystem subject constitutions are the fundamental building blocks of supply chain finance credit risk system. the main subject constitutions of the ecosystem designed by this study include core enterprises, upstream small and medium-sized suppliers, downstream small and medium-sized chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 170 distributors, financing institutions, big data platforms, third-party logistics companies and warehousing departments. figure 2 depicts the fundamental logical structure. figure 1. supply chain finance ecosystem including big data platforms. figure 2. supply chain finance credit risk management ecosystem including big data platforms. the following are the primary functions of each subject constitution in the supply chain finance credit risk ecosystem with big data platforms: the big data platforms are the overall risk masters of supply chain finance. each subject constitution's relevant data is aggregated into big data, and valuable credit risk information is mined using the data risk control model. this function is assumed by financial institutions in the supply chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 171 chain 1.0 mode, by core enterprises in the supply chain 2.0 mode, and by third-party service platforms in the supply chain 3.0 mode. the industry chain ecosystem mainly includes credit providers, as well as upstream and downstream capital demanders. in the supply chain finance mode, core enterprises play the role of credit providers while upstream and downstream smes are capital demanders. in the accounts receivable financing mode, smes need to submit cooperative orders to obtain credit financing. in the prepayment financing mode, smes need to submit margin to obtain the right to pick up goods for credit financing. in the inventory pledge financing mode, smes need pledges to obtain credit financing. the financial institutions ecosystem includes commercial banks and internet banks and other financial institutions. in the supply chain finance, the ecosystem plays the role of capital providers, mainly responsible for issuing funds to smes. in the traditional mode, the capital providers are in charge of supervising risk, formulating standards and issuing capitals. the warehouse logistics ecosystem mainly includes in-transit material transport service providers and pledge custodians. in-transit material transport service providers generally refers to the third-party logistics company, and pledge custodians generally refer to the warehouse storage institutions. the logistics service providers or the pledge custodians shall properly place the pledges and are able to evaluate the value of the pledges. 4.3. operation mechanism of big data in supply chain finance ecosystem in traditional supply chain finance, banks and other financial institutions need to assume the responsibility of credit supervision and fund release. the supply chain finance ecosystem including big data platform is still based on traditional supply chain finance business mode. however, in comparison to the traditional business model, the innovation of the system designed in this paper lies in the risk manager's transition from the traditional model of banks and other capital providers to a big data platform. big data has the characteristics of large quantity, high value. relying on data storage technology, big data platforms can record various information of all participants in the supply chain finance ecosystem, which include capital demanders, credit providers and capital providers. all the data between the different participants form big data. through the control tower, appropriate risk control models such as machine learning and deep learning are selected to conduct multi-agent association relationship mining to obtain accurate risk portraits of this node, and ultimately provide intuitive decision-making assistance for decision makers, based on the risk assessment rules set by experts. the big data platform enables financial institutions to effectively control information flow, logistics and capital flow, thus improving the operation efficiency of the entire supply chain finance ecosystem, so as to realize the coordinated development of the industrial chain ecosystem, financial institutions ecosystem and logistics and storage ecosystem. figure 3 shows the risk early warning system integrated that is integrated with big data. risk managers use big data technology to collect relevant data of credit assessment subjects, and use data mining technology to comprehensively depict the credit risk of each participant in the supply chain network, so as to realize risk management visualization. simply relying on big data technology to monitor all the smes’ credit status is insufficient. the actual financial scenario must be digitized, i.e., transaction terminals and other iot sensors obtain data to form a large data stream into the control chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 172 tower, and the data is processed and monitored in real time through blockchain and distributed storage technology, thus realizing the digitization of the transaction process and the physical world. with the help of blockchain technology and internet of things technology, the supply chain finance credit risk monitoring system including big data platforms can effectively avoid the problem of associated risk caused by a financing transaction participant breaking trust and the problem that the logistics and storage service provider breaking trust that cannot be detected. figure 3. risk early warning system of big data integration. 4.4. application of supply chain finance credit risk management ecosystem including big data technology 4.4.1. distributed data storage drives digitization of supply chain finance transaction processes enterprise "triangular debt" can cause a chain reaction due to the default of funds between enterprises. big data technology can digitize transaction processes with the help of distributed storage technologies such as blockchain. this process mainly includes three parts: the common identification of participant, the common identification of asset right and the common identification of transaction. the transactions process in each stage of the system realize the digitization of the whole supply chain transactions with the help of standardized digital vouchers. digital transaction vouchers are similar to newly issued standardized bills, which can play a role as a means of payment. digital vouchers can be split and transferred among various subject constitutions of supply chain finance, thus realizing multi-pole credit penetration among enterprises. in addition, each transfer of digital transaction vouchers will offset a part of accounts payable and resolve a debt, so as to help enterprises in the supply chain effectively clear the triangle debt, optimize the financial situation between enterprises, and reduce the associated risk caused by the trust-breaking problem of a financing transaction participant. 4.4.2. digital twins: leading the physical world by one virtual world of data with the help of the internet of things technology, the supply chain finance credit risk ecosystem including big data platforms can achieve effective supervision of the pledge transport process. sensors and other tools become the link between logistics supervision and digital management. the big data platforms can monitor the stored goods through sensors and other tools, such as temperature sensors, humidity sensors and other gps positioning devices, and digitizes the physical environment of the products. the dynamic matching of transportation logistics, goods storage and financing chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 173 transaction orders with the help of artificial intelligence technology can solve the problem of separation of logistics supervision and financing orders in the traditional state, that is, the dynamic combination of each financing order and each logistics document. when credit default occurs in financing orders, the intelligent system can lock the pledged chattel in real time. 5. conclusions and suggestions the information asymmetry between the participants of supply chain finance restricts the supervision of credit risk. emerging fin-tech provides a new solution to the problem of credit risk management in supply chain finance. big data technology can realize the functions of data collection, data storage, data mining and accurate portrait, which can realize the "visualization" management of credit risks in supply chain finance. the accurate early warning function of big data technology can reduce the cost of supervision and management and increase the efficiency of credit risk management. the conclusions are summarized as follows: 5.1. bridging the "information gap" to reduce credit risk the supply chain finance credit risk management system based on big data technology integrates the business transaction flow, information flow and logistics of traditional supply chain finance to form a big data flow. combined with block chain technology and distributed storage technology, it fully mines the internal information of big data flow and achieves accurate portrait of big data. real-time monitoring and early warning also can be carried out on the operating conditions and credit risks of smes participating in financing. it effectively narrows the "information gap" between the credit subjects upstream and downstream of the supply chain and banks and other financial institutions, ensures the authenticity of the full-cycle trade transaction process, reduces the negative impact of information asymmetry in the financing of supply chain smes. moreover, it solves the credit mismatch problem that cannot be avoided by traditional supply chain finance, and reduces credit risk. further, the system effectively reduces the audit cost and supervision cost of banks and other financial institutions. it has significant practical value and reference significance for maintaining national financial security, as well as preventing and resolving systemic risks. 5.2. realize the precise issuance and supervision of capital flow banks and other financial institutions need to take the responsibility of credit supervision and fund issuance in traditional supply chain finance. the risk supervision system of supply chain finance based on big data technology separates the supervision work from banks and makes banks become professional fund providers. banks can issue funds to upstream and downstream enterprises in the supply chain according to their accurate credit risk portraits, realizing the service goal of universality and precision that traditional supply chain finance cannot achieve. it effectively solves the problem of difficult and expensive financing for smes in the supply chain and lowers the threshold for obtaining funds for long-tail enterprises. in addition, through the accurate portrait of credit subjects, it can fit their liquidity needs, improve the efficiency of capital use, and effectively avoid the liquidity mismatch problem faced by traditional supply chain finance. it as well as highly fits the demand for financial services for high-quality development and is conducive to traditional financial reform. 5.3. promote the further integration of big data technology and the real economy chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 174 supply chain finance is an important field to carrying out supply-side structural reform. the supply chain finance supervision system based on big data technology can promotes the further integration of big data technology and the real economy. the traditional supply chain is empowered by big data technology, and the vendors, third-party logistics and storage institutions in the supply chain are jointly connected to the big data platform, which can enhance the competitiveness of key links in the industrial chain, improve the supply chain system of key industries, and become the fulcrum for transforming and upgrading traditional industries. the credit participants of the supply chain get the funds needed for developing can enhance the competitiveness of enterprises, which is conducive to deepening the structural reform on the supply side. it as well as can stimulate the vitality of the real economy, enlarge and strengthen the domestic real economy. as a result, it is easier to build a new development pattern with the domestic grand cycle as the main body and the domestic and international double cycles promoting each other. based on the above research conclusions, the following suggestions are put forward. first, to accelerate the construction of new infrastructure to build a firm foundation of digital economy. upstream and downstream enterprises in the supply chain should actively promote the digitization of enterprise and access their respective data to the supply chain finance supervision platform and realize the interconnection of data. the supply chain finance supervision system based on big data technology cannot be separated from new infrastructure networks such as 5g, internet of things and block-chain, while the construction of new infrastructure will also drive the rapid growth of local economy in the short term. central government and local governments should pay attention to the huge potential of the digital economy and do their best to support the construction of new infrastructure driven by technological innovation and based on information networks to help upgrade the structure of traditional industries and realize the digitization and digital development of the economy. second, to promote data interconnection and construction of information-based regulatory platform. the government should follow the principle of "open and sharing, multi-technology synergy and diversification of adaptive subjects" for the future development mode of supply chain finance. drawing on the current case of nanjing economic development zone and the bank of jiangsu's "yuan rong project" and the hangzhou municipal government's collaboration with boer data to share the "anti-fraud risk list" to banking institutions for free, government departments should take the initiative to seek changes and increase support for new financial technology. integrate various kinds of data, such as the data of industry and commerce, tax data and other data scattered outside the supply chain finance, into the supervision platform and converge them into big data. through data mining, make full use of the high-value characteristics of big data to realize accurate risk portraits of credit subjects and achieve the purpose of credit risk supervision and early warning. third, to attach importance to data security and formulate industry standards. it is considerable to attach importance to cyber security and strengthen data regulation. the aggregation of various data into big data streams through multiple paths is likely to infringe personal privacy and data security. regulatory platforms should attach importance to the contradiction between information security and data sharing, so as to make data "invisible when available". the government should adhere to the idea of promoting the development of digital economy with equal emphasis on data security and utilization, increase the supervision of big data platforms, strengthen the ability to protect key data resources, and guarantee data security. to build a supply chain finance credit risk chenyang wu, jinyue liu and hongmei zhang / journal of risk analysis and crisis response, 2021, 11(4), 163-175 doi: https://doi.org/10.54560/jracr.v11i4.310 175 management system based on the big data platform and establish a supply chain finance ecology suitable for the development of china's industrial chain, industry standards and codes of conduct should also be formulated as soon as possible to promote the collaborative development of technology. the standardized and orderly development of supply chain finance products will help achieve the goal of high-quality supervision of credit risk for all subjects in supply chain finance. this section may be divided by subheadings. it should provide a concise and precise description of the experimental results, their interpretation as well as the experimental conclusions that can be drawn. funding: this research was funded by 2020 the special project of key cultivation discipline and urgently needed discipline direction of guizhou university of finance and economics, "research on credit risk prediction and evaluation of big data enterprises", grant number 2020zjxk20 and the research fund project of guizhou university of finance and economics, "research on credit risk evaluation index system of prepayment financing in supply chain finance based on big data", grant number 2019zxsy96. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] luo, h.; he, r.k. design innovation driven by big data thinking. packaging engineering 2017, 38(12), 136140. [2] song, h. innovative trends of supply chain finance based on industrial ecology. china circulation economy 2016, 30(12), 85-91. [3] wang, j.; yang, l.l. the ecological construction of supply chain finance in the context of data-driven technology revolution. banker 2019, (06), 68-70. [4] ding, q.l.; zhang, s.w.; mei, y.; bao, q. data-driven intelligent supply chain ecosystem construction. business economics research 2020, (18), 38-41. [5] wen, l. research on the construction of supply chain ecosystem of aviation manufacturing industry in the context of big data. business and management 2020, (08), 95-99. [6] guo, y.j. research on innovation of agricultural supply chain finance model from the perspective of industrial internet. journal of sichuan light chemical university (social science edition) 2020, 35(02), 33-52. [7] zhao, c.g.; jiang, w.x. construction of risk management system of supply chain finance from the perspective of financial ecology. finance and accounting communication 2021, (06), 130-133+171. [8] jing, j.; feng, l.; song, x.l. research on supply chain finance model based on industrial ecological platform: theoretical analysis and case empirical evidence. financial development research 2021, (02), 8087. [9] wang, c.l.; hu, c.h.; zhang, y.; wu, h.h.; chen, x. supply chain control tower enables digital transformation of enterprises. intelligence theory and practice 2019, 42(09), 28-34. [10] song, h. innovation of supply chain finance model empowered by digital platform. china circulation economy 2020, 34(07), 17-24. [11] tang, l.l. theory and practice of digital supply chain control tower. supply chain management 2020, 1(02), 60-72. copyright © 2021 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word analysis of disputes in transportation projects.doc * corresponding author analysis of disputes in transportation projects m. emre bayraktar associate professor, ohl school of construction, florida international university, 10555 west flagler street, ec 2953 miami, fl 33174, usa bayrakm@fiu.edu cagri cinkilic former graduate student, ohl school of construction, florida international university 10555 west flagler street, ec 2900 miami, fl 33174, usa ccink001@fiu.edu farrukh arif* ph.d. candidate, ohl school of construction, florida international university, 10555 west flagler street, ec 2946 miami, fl 33174, usa *farif001@fiu.edu abstract disputes in construction projects are fairly common. they most often lead to increased costs, delays, and damaged business relationships. disputes can also result in loss of reputation, personnel resources, and potential business opportunities. previous studies have shown that the frequency of disputes in construction projects can be reduced through the timely identification of potential dispute sources and the implementation of effective lessons-learned strategies. this paper presents an analysis of the disputes for transportation projects in the state of florida and the related lessons-learned in order to prevent these disputes from recurring in future projects. two hundred and sixtytwo (262) cases of disputes in transportation projects between 1994 and 2008 stored in the dispute resolution board (drb) database of the florida department of transportation (fdot) were analyzed. it was found that majority of disputes were due to unforeseen conditions. also, more than half of the disputes (58%) occurred during the main construction stage of the projects. most of the lessons-learned were derived from the evaluation of the drb reports aimed to address ambiguous specifications, a responsibility-assignment mechanism, and poor conceptual planning. keywords: dispute, lessons learned, transportation projects, dispute resolution board. 1. introduction the construction business appears to be one of the most problematic as opposing ideas may lead to disputes in projects. this is a result of the risks and complexities inherent to the nature of construction projects, as well as the divergent interests of the parties involved; as such, disputes are often an unavoidable consequence of the construction process (levin 1998). disputes during construction projects are counterproductive to the progress of the project. a dispute takes place when there is a disagreement regarding the understanding and application of the contract (hibberd and newman 1999). disputes in construction projects may be financial, legal, or take on other forms. they often lead to loss of finances, time, and market share. not only that, but journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 215-222 published by atlantis press copyright: the authors 215 administrateur texte tapé à la machine received 29 may 2012 administrateur texte tapé à la machine accepted 24 august 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine m.e.bayraktar, c.cinkilic and f.arif disputes can also result in a loss of reputation. however, previous studies have shown that the frequency of disputes on construction projects can be reduced through the proper identification of the dispute and an effective lessons-learned mechanism (groton 1997, mitropoulos and howell 2001). in the case of transportation projects, these lessons-learned can be applied at different stages of the project in a variety of ways; such processes are helpful for dealing with issues at the local, regional, state, and national levels. this paper analyzes and evaluates the disputes in transportation projects in florida and presents a lessonslearned document. first, the sources of disputes in construction and use of a lessons-learned approach to dispute resolution is examined in the literature review. then, a close examination and analysis of disputes reported in the dispute resolution board (drb) database of the florida department of transportation (fdot) is conducted. it is hoped that this paper will provide a document that, if needed, can be used throughout the entire sequence of transportation decision-making, from transportation planning to project development to prevent future disputes. 2. effectiveness of dispute resolution boards a drb is a board of impartial professionals formed at the beginning of a project to follow construction progress, and is available on short notice to resolve disputes for the duration of the project. usually, a drb consists of three members selected jointly by the contractor(s) and owner, to monitor the progress of construction works, and to recommend a resolution for disputes in a timely fashion. drb members are generally familiar with the type of construction involved, are respected in the industry, and approach their responsibilities with neutrality and impartiality (eladaway 2007). it is found that the drb process is more successful than any other alternative technique for dispute resolution in construction disputes. the success rate is very high, 98% until 2007, in resolving disputes without appealing to litigation (drb manual 2007). since the first successful implementation in 1975, drbs have gained popularity as a standing neutral alternative dispute resolution (adr) technique. menassa (2010) reports that drbs have been successfully implemented in all construction sectors in the united states. the effectiveness of drbs as a prevention technique was observed in approximately 50% of the 810 projects in which no disputes were ever heard through a drb formal hearing. for the remaining 50% of the projects, the effectiveness of drb as an adr technique was found to exceed 90% when comparing the number of disputes that were settled due to drb recommendation to those that were actually heard during a drb session. with the efficacy of drbs as discussed above in mind, a lessons-learned approach can be applied to get the highest future benefits from the case histories of previous disputes resolved through the drbs. for this purpose, the drb database of the florida department of transportation has been utilized herein. 3. lessons-learned approach for dispute resolution throughout the construction of any facility, knowledge is obtained and lessons are learned from both positive and negative experiences. harrison (2003) defines lessons learned as “a good work practice or innovative approach that is captured and shared to promote repeat application, or an adverse work practice or experience that is captured and shared to avoid recurrence.” the lessons-learned process includes three key steps: collection, analysis, and implementation (caldas et al. 2009). many organizations in the construction industry have come to recognize the importance of a lessonslearned program (llp) as a vital asset in knowledge management systems. in these organizations, project team members generally acquire new knowledge as their careers progress. this knowledge, which includes both the successes that organizations want to repeat and the problems that they wish to prevent, may not be routinely disseminated. the benefits to be gleaned from, the optimized dissemination of such knowledge highlights the importance of llp (caldas et al. 2009). in the field of transportation, the data that would be required to develop meaningful lessons-learned from claims are typically available in the dispute resolution board (drb) database. there are, however, some barriers that prevent effective implementation of lessons-learned. these barriers are as follows: (i) the lesson is too general to be passed from one case to another, (ii) the lesson is ambiguous, not mutually exclusive, and collectively exhaustive to implement, (iii) the lesson is not typically linked to the project stage, (iv) lacks a meaningful classification system, (v) has published by atlantis press copyright: the authors 216 analysis of disputes in transportation projects difficulty in integrating new systems into existing procedures and operations, (vi) has an unmanageable format that limits the access, retrieval, and updating of a potentially enormous volume of lessons, etc. (marlin 2008). on the other hand, lessons-learned provide many benefits if used effectively. lessons-learned allow other practitioners to learn from previous experiences and avoid “reinventing the wheel”. they help stakeholders at different levels to understand the relevance of other activities and achievements, thus improving collaboration and coordination. moreover, lessons inform decision-makers about common mistakes and assist in the facilitation of a more efficient working environment. 4. research scope and objectives in order for a lessons-learned document to be successful, it must have a clear “application domain”. the purpose of this paper is to identify, analyze, and evaluate the current dispute reports in the fdot drb database for the development of such a document with a focus on transportation projects. to this end, 262 reports in the fdot drb database from between 1994 and 2008 were reviewed against the above criteria 5. research methodology in the first step of this research, a thorough literature review was conducted (fig. 1). its purpose was to determine the construction dispute resolution strategies most effective in the united states for the development of a lessons-learned approach. as demonstrated in the literature, the dbr approach has consistently proven successful. another purpose was to determine the criteria required to produce a quality lessons-learned document; various sources such as academic journals, technical reports, news articles, and online resources were used in this regard. in the second step, a quantitative analysis was performed to determine the basic characteristics of the 262 disputes in the fdot drb database, such as number of disputes, time value of disputes, monetary value of disputes, and dispute results. in the third step, disputes were categorized by contractual aspect and construction stage. finally, a qualitative and critical analysis was carried out to develop a lessons-learned document for the various disputes that may occur in transportation projects. also, conclusions were made and recommendations put forward. fig. 1. research methodology. 6. analysis and discussions two hundred and sixty--two (262) reports stored in the fdot drb database and containing dispute information between 1994 and 2008 were reviewed. all of the disputes submitted by the contractors to the drb were resolved, except one. the reports were quantitatively analyzed to determine the general characteristics of the disputes and to categorize disputes both by contractual aspect and construction stage. the analysis and a discussion thereof are presented in the following sections. 6.1. general characteristics of disputes out of the 262 disputes, the contractors won 119 (45.42%) times, while fdot won 133 (50.76%). in addition to these numbers, 10 of the disputes (3.82%) were concluded in negotiation. fig. 2 shows information about the disputes in terms of monetary and time values. as far as the monetary value of the disputes is concerned, the majority of disputes, i.e. 188 disputes (72%), did not have any monetary value specified. out of the remaining 74 disputes, 39 disputes were in the range of $0 to $49,999, 13 disputes were in the range of $50,000 to $99,999, 7 disputes were in the range of published by atlantis press copyright: the authors 217 m.e.bayraktar, c.cinkilic and f.arif $100,000 to $149,000, and the remaining 15 disputes were $150,000 and above. regarding time value, the records showed that 185 of the disputes (71%) did not have this characteristic specified. out of the remaining 77 disputes, 43 had a time value of 0 to 25 days, 9 disputes had a value of 26to 50 days, 12 disputes had a value of 51 to 75 days, and the remaining 13 disputes had a time value of above 76 days. fig. 2. disputes in terms of monetary and time values 6.2. categorization of disputes by contractual aspect ten categories of disputes, as per contractual aspects were defined through an extensive literature review. these categories included material, quality, safety, plans and specifications, construction methods, equipment, third party hindrances, quantity variation, unforeseen conditions, and permits. the 262 disputes were then cataloged under these respective aspects. table 1 shows the number of disputes generated with respect to the different contractual aspects. table 1. number of disputes by contractual aspect category number of disputes material 32 quality 3 safety 4 plans & specifications 43 construction methods 22 equipment 2 third party hindrance 21 quantity variation 58 unforeseen conditions 67 permit 10 total 262 as shown in table 1, 67 of the disputes (26%) occurred due to unforeseen conditions. quantity variation is the second most encountered dispute characteristic, with 58 disputes (22%), while the third leading characteristic is plans and specifications with 43 disputes (16%). 6.3. categorization of disputes by construction stage through an extensive literature search, six typical project stages in transportation and their subcategories were identified. the six main categories included 1) permit, 2) site work, 3) foundation, 4) main construction, 5) landscaping, and 6) other were. fig. 3 shows the distribution of disputes in fdot projects among these six main stages. it was found that more than half of the disputes occurred during the main construction stage. the foundation stage was found to be the second most critical stage for the occurrence of disputes. fig. 3. disputes in different construction stages the detailed categorization of disputes according to subcategories is shown in table 2. most of the disputes in the permit stage were related to environmental permit issues (50%), while in the foundation stage, sheet pile (23%), bridge joints (15%), and additional unforeseen condition (12%) -related issues occurred more frequently. during the main construction stage, additional unforeseen condition (17%), utility conflicts (8%), and concrete/slab/asphalt (8%) -related issues were found to cause disputes. in the landscaping stage, driveways (30%), sod installation, seeding, fertilizing, mulching, mowing (30%), and additional unforeseen condition (30%) –related issues were the contributing factors. other than this, delay (33%) problems were also a significant cause of disputes. published by atlantis press copyright: the authors 218 analysis of disputes in transportation projects table 2. detailed categorization of disputes category no of disputes permit environmental permit 5 lane closure 3 site access 1 other 1 site work fence 3 foundation defective specifications/plan 3 base material/other 8 base material/shortage 3 earth wall 4 footing (replacement) 1 footing (shaft) 2 sheet pile 12 excavation 4 bridge joints 8 additional unforeseen condition 6 hauling 1 construction concrete work 9 concrete/asphalt/slab 12 truncated domes 2 utility work electrical rough-in 8 water rough-in 7 specialty rough-ins (phone) 1 specialty rough-ins (cable ) 2 gas utility 2 relocation 3 conflict 13 drainage utility/sanitary sewer utility 9 insulation(coating) 5 material unsuitable material 3 extra 3 shortage 3 traffic signals 5 repair/replace 2 restriction 1 bridge deck 1 defective specifications/plan 3 control of work 2 maintenance of traffic 7 additional unforeseen work 26 equipment (idle) 1 equipment (other) 4 tests 2 noise ordinance suspension 1 changed site conditions 8 non-payment 4 workman compensation cost 2 landscaping driveways 6 sod installation, seeding, fertilizing, mulching, mowing 6 pond 1 other 1 additional unforeseen 6 other speciality engineer 1 contractual document 2 change in scope of work 1 delay 8 fire hydrant 1 discharge of superintendant 1 overhead expense 1 liquated damage 1 incentive-disincentive 1 stand-by cost of crew 1 off-duty law enforcement 1 traffic accident 1 bridge clearance 1 schedule interpretation 1 vandalism/stolen 1 changed market price 1 7. lessons-learned and lessons-learned matrix the lessons-learned developed from the fdot drb reports are presented in this section. these lessonslearned are mainly aimed at avoiding future disputes. they were derived through an intense critical analysis of the reported cases, a thorough literature review, expert opinion of the construction industry, monitoring, evaluation, and documentation techniques. it aims at building a strong foundation of knowledge and its objective is to serve as an important input, not only for dispute resolution, but also during the conceptual and detailed planning stages of transportation projects. the data were collected for the maximum benefit of users and coded as l1, l2, …., l24 in the lessons-learned matrix, which is explained later in the text. based on this, a lessons-learned matrix for the fdot projects was developed (table 3). l1. make sure to procure all environmental permits required by federal, state, county, and local regulatory agencies. l2. make sure to procure the proper water permits to cover all necessary project work on site. l3. proper access to the site shall be arranged. l4. fdot shall clearly define any restrictions before the time of bidding and changes shall only be made if fdot is willing to provide compensation. l5. any change in site clearance work or fencing work characteristics shall be clearly delineated. the effect shall be determined both in terms of materials and the nature of work. published by atlantis press copyright: the authors 219 m.e.bayraktar, c.cinkilic and f.arif table 3: lessons-learned matrix for fdot projects l6. to avoid defective specifications, fdot experts should have multiple early site visits before defining the specifications. also, other regulatory agencies shall be taken in confidence regarding the specifications and shall be involved in such visits. this exercise will prove to be extremely valuable from a cost estimation perspective as well. l7.site surveys and tests related to geotechnical conditions of the site shall also be performed. l8. foundation drawings shall provide clear and complete details to avoid any disputes. l9. clear instruction shall be given to contractor(s) to prevent the flow of turbid water into the canals. contractor(s) shall also submit a notice of intent for any extra work he/she has planned in this regard. l10. the responsibility of repair work during the construction shall be properly communicated to the contractor(s) l11. specifications shall clearly mention that materials proposed by the contractor(s) must be submitted in samples and approved by fdot. l12. there shall be no unnecessary delays in payment from fdot once its specifications have been met and the quality of work approved. l13.the contract shall mention any temporary arrangements required, such as a glare screen or temporary barrier wall. the documents drawn shall also address payment issues for these arrangements. l14. the contract shall clearly address the responsibility for any unforeseen conditions. it shall also address the payment issues for such conditions. l15. contractor(s) shall try to anticipate any delay in advance, and shall also apply for any time extension in advance. l16. any problem encountered by the contractor(s) at site shall be communicated to fdot immediately. this will expedite the compensation process if required. responsibility applicability for the avoidance of disputes in different construction stages lesson fdot contractor permit site work foundation main construction landscaping others l1 x x x x x x x - l2 x -x x x x x - l3 x --x ---- l4 x x x x x x -- l5 -x -x ---- l6 x ---x x -- l7 x ---x --- l8 x ---x --- l9 x x --x x x - l10 x ----x -- l11 x x ---x -- l12 x ---x x x x l13 x ---x x -- l14 x ---x x -- l15 -x --x x x - l16 -x -x x x x - l17 x ---x x x - l18 x ----x -- l19 -x ---x -- l20 x ---x x x - l21 x x -x x x -- l22 x -----x - l23 x x x x x x x x l24 x x -x x x x x published by atlantis press copyright: the authors 220 analysis of disputes in transportation projects l17. non-working days shall be clearly communicated in the contract. it shall also mention the rate of pay and other details regarding payment if these days are utilized during the construction stage. l18. any restrictions related to burning operations shall be mentioned in the contract. l19. lump sum items shall be clearly calculated with market trends in mind. l20. any change in scope shall be communicated to all parties in writing. l21. fdot is responsible for providing an alternative traffic control plan. any changes in the plan by the contractor(s) due to prevailing conditions shall be properly communicated to all parties in writing. l22. fdot shall contractually bind the contractor(s) to maintain the sodden areas in a satisfactory condition until final acceptance of the project. l23. communication among the parties shall always be formal, in accord with contract terms and conditions. this will avoid any misinterpretation of the responsibilities, scope, and standards of the project. l24. it shall be clearly outlined in the contract that the primary responsibility for any accident on-site lies with the contractor(s), except in case of special circumstances. this matrix defines two things: first, the actions suggested by the lessons-learned to avoid disputes between fdot and the contractor(s); second, it also defines the applicability of the lessons for avoiding disputes at various construction stages of the project. for instance, for the first lesson-learned (l1), the matrix in table 4 shows that the procurement of all environmental permits required by federal state, county, and local regulatory agencies is the joint responsibility of both the fdot and the contractor(s). this means that these stakeholders must work cooperatively to obtain the permits, in contrast to previous cases in which one blamed the other when problems occurred; hence, the resulting disputes cost the project both in terms of time and money. the applicability of these lessons-learned, then, is to avoid the disputes which are present in almost all construction stages. 8. conclusions and recommendations two hundred and sixty two (262) disputes in fdot between 1994 and 2008 were analyzed. the contractors won 119 (45.42%) times, while fdot won 133 (50.76%). in addition to these numbers, 10 of the disputes (3.82%) were concluded in negotiation. furthermore, the disputes were categorized from a contractual perspective and according to construction stages. it was found that 67 disputes (26%) were due to unforeseen conditions. the reason behind this is that the language of the contract is for the most part unrelated to the actual costs borne by fdot and the contractors. according to the construction stages, it was found that more than half of the disputes in florida’s transportation projects occurred during the main construction stage. this is because during the main construction stage, contractor(s) begin to enact their ideas, which may lead to various unforeseen conditions and utility transferrelated conflicts. in answer to these issues, this study has presented many recommendations regarding lessons-learned. these lessons-learned aim to avoid disputes while keeping in mind both the contractual perspective and different construction stages. most of the lessons-learned indicate that the majority of disputes were a result of ambiguous specifications, weak responsibility-assignment mechanism, and poor conceptual planning. therefore, a lessons-learned matrix has also been developed to assign the responsibility clearly and apply the aforementioned lessons. this matrix is also suggested to maintain a more formalized record of lessons-learned for fdot projects. continued improvement to this lesson-learned matrix will benefit future transportation projects, and inclusion of these lessons-learned in the conceptual and detailed planning stages can save both time and money. furthermore, it is recommended that this study be extended to update and measure the effectiveness of the lessons-learned and presented herein. acknowledgement the authors would like to thank the florida department of transportation for providing data for this research. references 1. levin, p. (1998). “construction contracts, claims, and disputes”, asce, reston, va. mark 2. hibberd, peter r., and paul newman (1999), “adr and adjudication in construction disputes”. blackwell science publisher. 3. groton, j. p. (1997). “alternative dispute resolution in the construction industry.” dispute resolution journal, summer, 49-57. published by atlantis press copyright: the authors 221 m.e.bayraktar, c.cinkilic and f.arif 4. mitropoulos, p., howell, g. (2001), "model for understanding, preventing, and resolving project disputes." journal of construction engineering and management, volume 127 issue 3, pp.223-233. 5. el-adaway, islam h., and a. samer ezeldin (2007), “dispute review boards: expected application on egyptian large-scale construction projects.” journal of professional issues in engineering education and practice, volume 133 issue 4, pp.365-372 6. menassa, carol c., and feniosky peña mora (2010). “analysis of dispute review boards application in u.s. construction projects from 1975 to 2007.” journal of management in engineering, volume 26, issue 2, pp. 6577. 7. harrison, w. (2003). “a software engineering lessons learned repository”. proceedings of 27th annual nasa goddard/ieee software engineering workshop, institute of electrical and electronics engineers, los alamitos, ca. 8. caldas, carlos h., g. edward gibson, jr., runi weerasooriya, and angela m. yohe (2009), “identification of effective management practices and technologies for lessons learned programs in the construction industry.” journal of construction engineering and management, vol. 135, no. 6, pp. 531539. 9. marlin (2008), “implementing effective lessons learned process in a global project environment.” published as a part of utd 2nd annual project management symposium proceedings. dallas, texas. published by atlantis press copyright: the authors 222 microsoft word review of the strengths and weaknesses of risk matrices (1) 1 review of the strengths and weaknesses of risk matrices mustafa elmontsri department of primary care and public health, school of public health, imperial college london st dunstan’s road, london, w6 8rp, united kingdom e-mail: m.elmontsri10@imperial.ac.uk www.imperial.ac.uk abstract risk assessment and risk matrices are powered tools used in risk management and help guide in the process of decision-making in organisations. nevertheless, risk matrices have their own weaknesses and strengths. this paper provides a critical overview of the development and use of risk matrices in different field with an example of the risk matrix used by the national health service (nhs) in england. risk matrices are helpful tools for risk assessment as they use quantitative measures to ensure consistent method of determining risk but organisations should adjust the design and size of risk matrices to suit their needs. keywords: risk assessment matrix, risk matrices, nhs risk matrix, quantitative risk matrix 1. introduction all over the world, nations and organisations are attempting to reduce risks, to improve safety and to extend lives. indeed risk reduction has become a principle goal of modern governments and almost in every organisation. it is obvious that people, including government officials, often lack risk-related information. they often know little about the nature and magnitude of the risks at issue, and they often know little about the various consequences of risk reduction (sunstein, 2002). since risk cannot be eliminated, the main problems people face, individually and collectively, are how much risk they should live with and how they should go about managing the risk. if a set of strategies have been chosen that will allow the abatement of a particular risk, the question of what level of risk should be chosen arises. if abating the risk costs nothing, the obvious answer is zero, get rid of the risk. but risk abatement almost always does cost money and time (glickman and gough, 1990). to answer these questions, analytical tools and risk ranking schemes must be used to distinguish lower risk activities / incidents from higher risk activities / incidents. one of the risk ranking methodologies is known as the risk assessment matrix. 2. risk management risk management is the process of assessing risks and taking steps to either eliminate or to reduce them (as far as is reasonably practicable) by introducing control measures. risk management refers to the process of reducing the risks to a level deemed tolerable by society and to assure control, monitoring, and public communication (morgan, 1990). there are more questions than answers when people talk about risks. the career of the term ‘risk’ is a rather recent phenomenon, however (fischhoff et al., 1984), states that, “risk has always been part of human journal of risk analysis and crisis response, vol. 4, no. 1 (march 2014), 49-57 published by atlantis press copyright: the authors 49 willieb typewritten text received 5 july 2013 willieb typewritten text accepted 29 january 2014 willieb typewritten text mustafa elmonts existence and as human bei own deaths dangerous sit the internatio developed a distinguishes risk – for wh – and decidi management 2005). fig. 1.irgc’s 3. definition there is no c risk – neit understanding common, how and possibilit the definitio (1992) is “the occurs during particular ch statistical risk probabilities” fischhoff et a like that of inherently co affect the ou resources am of political po 4. risk asse all of the pe have to be m hazard. it fo sri d the field of ings started to and conte tuations”. onal risk gov a framework between ana ich risk appra ng what to d is the key risk governanc n of risk commonly acc ther in the g. all risk co wever: there is ty (renn, 1998 on of risk acc e probability t g a stated per hallenge. as k obeys all th ” al (1984), sta any other ke ontroversial. t utcome of pol mongst safety m ower in societ essment eople, things, managed can, ollows therefo risk research o reflect the po emplated act vernance cou (figure 1 alysing and u aisal is the ess do about a ri activity (irg e framework (i cepted definit e sciences oncepts have s a distinction 8). cording to th that a particul riod of time o a probability he formal law ates that the d ey term in p the choice o icy debates, t measures, and ty. , activities an in some circ ore, that unle h started as ea ossibility of th tions to av uncil (irgc) h below), wh understanding sential proced sk – where r gc white pap irgc, 2005) tion for the te nor in pub one element n between real he royal soci lar adverse ev or results from y in a sense ws of combin efinition of ri policy issues, of definition c the allocation d the distribut nd places wh cumstance, be ess it is kno arly heir oid has hich g a ure risk per, erm blic t in lity iety vent m a of ing isk, , is can n of ion hich e a own ex ide ac ide ide oth th de an 19 de pre tec is co mu eff ri ap mo tha ma the in co ap pro ris ran be sce inc su w (c nu ris ris th nu qu ma co in ma ris ma de xactly what is entify all of th ccording to b entification is entified, with hers. he mathemat eveloped mor nalyses were p 998). risk as efining the edominantly chnical risk a at stake, calc onsequences, ultiplying the ffects (kolluru isk assessmen pproaches thro ore traditiona at constraints anagement pe e public, and which risk as ox (2005), sta pplied to risk oduce subject sk only. he nk the risk fro e greater tha enarios. whe cluded in the ubjectively. whereas quant cox, 2005), th umerically, all sk relative to sk measured o hese determin umerical scal uantitative ris ainly in the onsequence lev 2008, ton athematical p sk tool often r atrix or, more evices come being manage he sources of h boyle, one of s that a large h some obvi tical tools f re than a ce performed on ssessment is components quantitative t assessments, t culating the p and aggrega e probabilities u and brooks, nt techniques v ough a regime al quantitative such as time erceptions, ris political pres ssessment are ates that whe k assessments tive and very l argues that q om one scenar an some oth en all the sce ranking, the titative risk he risk from lowing the an all scenarios on whatever sc nations can b les. jeffery k assessment form of boar vels. ny cox wrot problems asso referred to as e simply, a “ri in the form ed it will not hazards (boyl f the problem number of ha iously more for risk asse entury before n technical sy the scientifi s of risk terms. it is a this means sp probabilities fo ting both co s by the mag 1995). vary from pure e of semi-qua e. altenbach ( , money, man sk result com ssures all affe carried out. en quantitativ s, it can be limited relativ qualitative ju rio or group o her scenario enarios from ranking can assessment, each scenario nalyst to determ in the system cale of units ar be made obje (2006) state t may use so rd ranges of te about ser ociated with a s a consequen isk matrix. ge m of qualitat be possible to le, 2002). ms with hazard azards will be serious than essment were e actual risk ystems (renn ic process o in precise argued that in pecifying wha for (un)wanted omponents by gnitude of the ely qualitative alitative to the (1995), argues npower, skills mmunication to ect the manner e approach is considered to ve sense of the udgments may of scenarios to or group o a system are only be done according to o is estimated mine not only m, but absolute re chosen. ectively using es the semi ome numbers frequency or rious techno a widely used nce probability enerally, these tive or semi o d e n e k n, f e, n at d y e e e s s, o r s o e y o f e e o d y e g s, r d y e published by atlantis press copyright: the authors 50 strengths and weaknesses of risk matrices quantitative instruments in which hazards are first identified and then allocated to a box on a twodimensional grid for which one axis measures the likelihood of a specific incident and the other the potential severity of consequences. the issues identified by cox are certainly not confined to the united states, and indeed usage of risk matrices has spread in the united kingdom and europe from industry to all manner of public and private agencies ranging from hospitals to smalland medium-sized enterprises, local and central government bodies, and professional institutions 4.1. risk assessment matrix a common method used for risk ranking utilises risk matrices; these are typically 4x4 or 5x5 matrices, having event consequences along one axis and event frequency along the other. each block on the risk matrix represents some level of risk, and blocks presenting similar risk are often grouped together into one of four or five risk regions (altenbach&brereton, 1998) risk matrix is defined as “a mechanism to characterise and rank process risks that are typically identified through one or more multifunctional reviews (e.g. process hazard analysis, audits, or incident investigation” (markowski and mannan, 2008), and is also defined by cox (2008) as “a table that has several categories of “probability,” “likelihood,” or “frequency” for its rows (or columns) and several categories of “severity,” “impact,” or “consequences” for its columns (or rows, respectively)”. in most cases, the frequency axis of the matrix has numerical values associated with it, typically spanning several orders of magnitude. often, the consequence axis is based on a qualitative scale, where consequences are judgment based. however, the consequence scale generally has implicit quantitative values associated with it, which may or may not be recognised. risk regions are often arbitrarily assigned (or assigned on the basis of symmetry). this presents a problem in that if the blocks of the risk matrix are incorrectly grouped, then incorrect conclusions can be drawn about the relative risk presented by events at a facility (woodruff, 2005).three types of risk matrices are commonly used for risk ranking. a purely qualitative risk matrix will have its blocks defined in descriptive or qualitative terms. a purely quantitative risk matrix has its blocks defined in measurable or quantitative terms. relative or absolute numerical scales are used on quantitative matrices, whereas scales on qualitative matrices are relative but not numerical. the third type of risk matrix is a hybrid: a semi-quantitative matrix with one scale (usually frequency) expressed quantitatively, while the other scale is expressed qualitatively (emblemsvag and kjølstad, 2006).integ-risk (2008) clearly states the importance of using scoring systems in risk assessment and management which generally requires the application of specific scores or scales. they highlight that in practical use, conventions such as using 5x5 risk matrices and/or a colour-code can be beneficial 4.1.1.qualitative risk matrix the qualitative risk matrix is basically task and or hazard analysis with some relative judgments made in order to categorise the hazards. when the 3x3 matrix is used, both the frequency and consequence of each accident scenario are then estimated on simple relative scales, such as low, medium and high. the risk for each scenarios is the product of the frequency rating and consequence rating, this indicates that the qualitative risk in this case falls into nine distinct regions or frequency x consequence pairs: low x low, low x medium, low x high, medium x low, medium x medium, medium x high, high x low, high x medium, high x high. clearly low x low region has the lowest risk, while the high x high region has the highest risk. the intermediate regions are more difficult to interpret because some regions are directly comparable and others are not (altenbach, 1995) in the environmental protection agency in the usa (epa) technical guidance for hazards analysis adapted by doe-std-3009-94, the risk levels from the 3 by 3 matrix are grouped into three categories: high (major concern), medium (concern) and low (no concern), as indicated in the figure.1 below, and also table 1 shows the risk groupings from the epa. published by atlantis press copyright: the authors 51 mustafa elmonts fig. 2.qualitat regions connec it is notable designing dir risk regions. a numerical highest. som are denoted indicate that w with respect t risk of these may in fact b only relative connected by is risk grade 2 is grade 1, an and it is note have risk gra implied equ information c risk than me medium. table 1. risk g 1994) sri tive risk matrix cted by arrows ( in the figure rections from the relative r grade, with 1 me regions wi by prime (’ while they ha to nearby regi regions is no be significantl e when appli y the arrows. f 2 and is highe nd lower risk ed that high x ade 3 and 3’ uivalence b common to b edium x low grouping from x: risk levels a (altenbach, 199 e above, that lower risk re risk of each re 1 being the lo th the same n ’) and doubl ve the same r ions connecte ot necessarily ly different. t ed to those for example, er risk than lo k than high x x low and me respectively, between them both is that th and lower ri epa (us depa are relative to 95) t the arrows egions to high egion is given owest and 5 numerical gra le-prime (’’) elative risk le ed by arrows, y equivalent, a the risk grade regions direc medium x l ow x low wh x low (grade edium x mediu but there is m. the on hey have grea isk than high artment of ener the are her by the ade to evel the and e is ctly low hich 3). um no nly ater h x rgy, th inc reg in eq eq sec wh no reg 19 it de ab ca an to 4.1 fe qu eff res tri ev the (c it co ha ris al 4.1 no qu co dim all req as qu lab ba arg ea as an he epa gro consistencies, gions of diffe the same g quates risk gr quates risk gra cond type o hich are not d ote the conc gions which 994). is also argu ecember 15, bility to ran ategorising the nd severity or inherently am 1.2.semi-qua ew serious ri ualitative appr ffort to enhan sults, many ed. there are ven though the e frequency cox, babayev is argued be omparison, qu ave very limit sk groupings ltenbach, 1998 1.3.quantitati ot all hazardo ualitative eve onsequences sc mensionless u l regions in quirement fo sessments is uantitative risk bels, or at lea abayev and gues that by u ach accident s sociated with nd ranked. co uping presen , the first typ erent and dire group. note t rade 4 and 5 ades 1, 2 and f logical inc directly comp cern group co are not direc ued by cox 2008), that r nk quantitat e two axes of probability an mbiguous risk antitative matr isk assessmen roach, due to i nce the usefu semi-quantita often referred ere is a quantit axis, consequ and huber, 20 cause of the ualitative and ted value, it m of the blocks 8). ive risk matrix ous situations nt tree/fault cale quantitat units, relative n the matri or qualitativ soundness, ks should rec ast, should no huber, 2005 using a quantit scenario will h it, then all s ox (2008), al nts two type pe of inconsi ectly comparab the major c 5, and no c one region of consistency p parable in the ontains two r ctly comparab (personal co risk matrices ive risks c f the matrix (e nd consequen classification rix nts actually its limited use ulness of the ative scheme d to as qualita tative foundat uences axis, 005). limitations in semi-quantita makes no sen s (cox, 2008: x need to be an analysis. by ive, even if o e risk can be ix. a basic ve and quan which states eive higher q ot receive low 5).simmons e tative risk ma have a relati scenarios can so states that es of logica istency places ble risk grade oncern group concern group f grade 3. the places regions e same group risk grades 3 ble (us doe ommunication have limited correctly and e.g. frequency nce) often lead . use a purely efulness. in an e comparative es have been ative methods tion applied to or even both n making risk ative matrices nse to attemp brereton and nalysed with a y making the nly in relative calculated for c consistency ntitative risk s that higher qualitative risk wer ones (cox et al (2005) atrix approach ive risk value be compared t, for the risk al s e p p e s p. 3 e, n, d d y d y n e n s, o h k s t d a e e r y k r k x, ), h, e d k published by atlantis press copyright: the authors 52 matrix to be discriminate r risks, so that to focus risk m 5. nhs ris the nationa kingdom rec and it is a t control measu grade or ra assessment m management of risk assoc incident. fig matrix that is fig. 3.nhs risk according to possible, the a predicted outcome. if t assigned to given time fr patient care e numerical pr descriptions w (www.npsa.n 6. discussio organisations defensible ran based on com risk ranking i frequency ax frequency pro e most usefu reliability bet it can be used management a sk assessmen al health ser cognises the u tool used for ures to be put anking that matrix. an process is to iated with a w gure 3 below widely used b k matrix availa o the nhs r score of the li frequency of this is not po the adverse rame, such as episode. if it i robability then will determin nhs.uk) on s are recomm nking system mpany’s defin is the risk ma xis. the pr ovides a meas ul it should, tween very hig d as an effectiv attention and r nt rvice (nhs) use of risk ass r making dec t in place acco is calculated essential pa be able to m work activity w shows the by the nhs. able at (www.np risk managem ikelihood will f occurrence ossible, a pro outcome occ the lifetime o is not possibl n the use of ne the most a mended to ad to allow for d ed safety goa atrix that has c roduct of co ure of risk. at a minimu gh and very l ve screening t resources. in the uni sessment matr cisions on w ording to the r d by the r art of the r measure the le y or a workpla risk assessm psa.nhs.uk) ment strategy, l be by assign of the adve obability will curring within of a project o e to determin f the probabil appropriate sc dopt a standa decision maki als. the basic consequence a onsequence a um, low ool ited rix, what risk risk risk evel ace ment , if ing erse be n a or a ne a lity ore ard, ng, for and and co dif tha rat cle eq be inc an ap ac de jud reg an lev it gra by mi wa qu qu in in the cla nu hid ha sta are ris loc mi th bo ac al est sce co ad en wi ac co on onsidering tha fferent risk r at groupings c ting are not ear, in gene quivalent risk e found. to o cident scenar nalysed and ppropriate re ceptance leve efine the risk dgments base gion. on th nalyst open to vels to suit som is argued that asp and comm y examples of isunderstandin all which bloc ualitative rea uantitative real my experienc appendix a& e consequen assifications a umerical estim dden agendas ave. the risk andardise the e involved in sk matrix mak cate the risk ight put them he nhs risk m oard and all tion that shou lso, the quant timates of r enario is sa ost/benefit tra ddress the per nough?” the ill easily justif ccording to c onsultants bel nly rough appr strength at groupings atings are no consisting of logically con eral, logical from the qua obtain the bes rio is evalua placed on egion withou els or judgme acceptance le ed on the sc he other han o criticism fo me hidden age t the concept municate, even f probability, ng and misuse cks the jump f alm to the lm. ce within the & b, shows t nces and th assist all nhs mates of the s that the staf k matrix in t process of gr n the assessing kes it easier f grading with off completin matrix also as risk-owners t uld be taken de titative appro isk instead o afe. it can adeoffs of a rplexing ques increased uti fy the extra bo cox (2008), m ieve that, wh roximate tool hs and weaknesse consisting of ot logically co regions from nsistent, one groupings o alitative risk st use of risk ated for the the risk m ut any pre ents. then the evels by maki cenarios that nd this proce or adjusting th enda. of probability n though we a such as lotte e of the princi from the fuzz e precise y nhs sector, the possible d he likelihood s staff to be ab risk, while ff member/dep the risk asse rading the ris g of risks, ha for them to un hout any com ng the risk ass ssists the risk to decide on epending on th ach can addr of some fee be used to risk reducti stion of “how ility of quant other in many many decisio hile risk mat s for risk ana es of risk matrice f regions from onsistent, and the same risk conclusion is of regions o matrix canno k matrix, each system being matrix in the edefined risk e analysts can ing subjective fall in each ss leaves the he acceptance y is difficult to are surrounded ery picks, the iples provide a zy comfortable yet uncertain table 2 and 3 descriptors for d that, these ble to allocate avoiding any partment may ssment forms sk, as all staf aving a simple nderstand and mplexity which essment form managers, the n the level o he score. ress numerica eling like the analyse the ion plan and w safe is safe titative results applications. on-makers and trices may be alysis, they are es m d k s f t h g e k n e h e e o d e a e n 3 r e e y y s ff e d h m. e f al e e d e s d e e published by atlantis press copyright: the authors 53 mustafa elmontsri very useful for distinguishing qualitatively between the most urgent and least urgent risks in many settings and are certainly much better than doing nothing, for example, than purely random decision making. donoghue (2001) also supports the idea that, the risk assessment matrices are effective tools in making decisions in regard to the control of occupational health risks. he states that, the control measures can be applied in an iterative fashion until the risk has been reduced to an acceptable residual. the imagery of risk matrices is powerful, which may, along with their alleged and apparent simplicity, explain their popularity among agencies that are responsible for mainly lesser hazards,1 and therefore are likely less qualified in risk, but who nonetheless feel the need to be seen to be proactive in managing risk. inter alia, and as observed, though not sanctioned, in the new international guidance on risk assessment (iso 31010), it is said that matrices are also widely used to determine if a risk posed by a given hazard is or is not acceptable. ball and watt (2013) also concur with cox (2008) that one of the leading arguments in support of risk matrices, which is that they are simple to use and transparent, is false. as determined here, all positionings of hazards on the matrix are subject to innumerable considerations, some of which even the rater may not be wholly aware. yet, and it is another serious matter, requisite explanations and justifications are seldom, if ever, attempted. it is this latter issue, of the consistency of use of risk matrices as applied to what are normally seen as beyond-the-workplace hazards. a growing number of authors, highly experienced in risk assessment, have questioned or had cause to investigate alleged shortcomings of risk matrices, mainly on technical grounds. in addition, standards-setting institutions have warned of the potential for subjectivity and inconsistencyas have researchers in occupational safety (ball and watt, 2013). 7. conclusion risk assessment and risk management techniques are being developed in many fields as an aid to safety investment decision making. expanding responsibilities and limited resources compel policy makers to make difficult choices about the prioritisation of risk reduction measure and what safety standards to aim for. the need for mechanisms to help policy makers set priorities has been increasingly felt, and during the last few decades techniques of risk assessment and philosophies of optimisation have been developed. risk matrices are very effective and widely used tool in making and improving risk management decisions, however the question of how ideally risk matrices should be constructed to improve risk management decisions is ongoing. it is not easy to answer, because risk matrices are typically used as only one component in informing eventual risk management decisions and also because their performance depends on the joint distribution of the two attributes probability and consequence. a risk matrix can be a useful tool to present the results of simplified risk analysis, helping one to gain insight into the relative risk of various scenarios that might be encountered in a given system. when developed quantitatively with axes constructed to be relevant to the facility and operations being studied, risk evolutions can be defined logically. logic based risk evaluations can facilitate management decisions such as the authorisation of operations. it can also help optimise resources by showing where to concentrate efforts for more detailed analysis or for risk reduction activities. using 3x3, 4x4 or 5x5 matrix, will be useful to some organisations and might not be for others i.e. when 5x5 matrix is used, the matrix will have 25 blocks (risk grades), the more blocks for representation, the more likelihood of the risk matrix producing different levels which would produce more risk ranking grades. therefore, organisations would be able to allocate the low, moderate, high and extreme risk groups to the appropriate levels of responsibilities within the organisations. the wider options for the probability and consequence scores on a risk matrix should give more scope to differentiate within the risk group the probability of a certain risk occurring and the consequence of the risk occurring within the low, moderate, high and extreme groups for the different levels of responsibility. whereas by having 3x3 matrix, there will be only 9 blocks for the risk grades, which in some cases might not be useful when making decisions or allocating resources. however, if the descriptions of the consequence and likelihood scores are difficult to classify then the scores cannot always be well interpreted. for example, table 1 published by atlantis press copyright: the authors 54 strengths and weaknesses of risk matrices (appendix 1 nhs risk matrix), where it shows the consequence scores, by looking at the column where it says; service business interruption; the difference between major and catastrophic scores; catastrophic score leads the business to a permanent loss of the business while major score can only cause the business to be interrupted for one week. in such a case, the extreme description should be more than one week and permanent loss. cox (2009) argues that risk priority scoring systems, although widely used (and even required in many current regulations and standards), ignore essential information about correlations among risks. this information typically consists of noting common elements across multiple targets (e.g., common vulnerabilities).these common features induce common, or strongly positively correlated, uncertainties about the effectiveness of different risk-reducing measures. it is easy to use this information, in conjunction with well-known decision analysis and optimization techniques, to develop more valuable risk reduction strategies, for any given risk management budget, than can be expressed by a priority list. thus, there appears to be abundant opportunity to improve the productivity of current risk-reducing efforts in many important applications using already well-understood optimization methods.to sum up, risk matrices are a useful way of ranking risks, but organisations should adjust the design and size of risk matrices to suit their needs. references 1. altenbach, t (1995). “a comparison of risk assessment techniques from qualitative to quantitative”, proceedings of the joint asmeijsme pressure vessels and piping conference, honolulu hi. 2. altenbach, t & brereton s. (1998). “risk ranking methodology for chemical release events. probabilistic safety assessment and management. 4, international conference on probabilistic safety assessment and management new york city, ny 3. boyle, t. (2002). health and safety: risk management. england: iosh services limited 4. cox, l. a. jr., & huber, w. (2008). optimal design of qualitative risk matrices to classify binary quantitative risks [abstract]. in: proceedings of the annual meeting of the society of risk analysis; 2008 december 7-10, boston 5. cox, l. a. jr., babayev, d., &huber,w. (2005). some limitations of qualitative risk rating systems. risk analysis, 25(3), 651–662. 6. cox, l. a. jr. (2008). what is wrong with risk matrices?, risk analysis, vol. 28, no. 2, 7. donoghue, a, m. (2001). the design of hazard risk assessment matrices for ranking occupational health risk and their application in mining and minerals processing. society of occupational medicine. vol. 51 no. 2, pp 118-123 8. dietz, t., scott frey, r. and rosa, e. (1996) risk, technology, and society, in: r.e. dunlapand w. michelson (eds) handbook of environmental sociology, westport: greenwood press. 9. emblemsvag, j. &kjølstad, l.e. (2006), qualitative risk analysis: some problems and remedies. management decision, vol. 44 no. 3 10. fischhoff, b., watson, s.r. and hope, c. (1984) defining risk, policy sciences 17, 123–29. 11. glickman, t, s & gough, m. (1990). readings in risk. usa: resources for the future 12. kolluru, r.v & brooks, d.g. (1995) integrated risk assessment and strategic management, in: r. kolluru, s. bartell, r. pitblade and s. stricoff (eds) risk assessment and management handbook. for environmental, health, and safety professionals, pp. 2.1–2.23, new york: mcgraw-hill. 13. morgan, m.g. (1990) choosing and managing technology-induced risks, in: t.s. glickman and m. gough (eds) readings in risk, pp. 5–15, washington: resources for the future. 14. markowski, a, s. &mannan, m, s. (2008). fuzzy risk matrix. journal of hazardous materials. pp 152-157 15. mcilwain, j, c. (2006). a review : a decade of clinical risk management and risk tools. journal of clinician in management, volume 14, no 4, pp. 189-199(11) 16. renn, o. (1998) three decades of risk research: accomplishments and new challenges, journal of risk research 1 (1), 49–71 17. simmons, j.dwyer, s &pfitzer, t. (2005). the rac matrix: a universal tool or a toolkit. journal of system safety. vol. 41 no. 2. pp 14-19 18. sunstein, c, r. (2000). risk and reason: safety, law and the environment. england: cambridge university press 19. the royal society. (1992). risk: analysis, perception and management. england: the royal society published by atlantis press copyright: the authors 55 mustafa elmontsri 20. the national patient safety agency (www.npsa.nhs.uk) accessed on 26th august 2013 at 17:40hrs 21. the international risk governance council. (2005). white paper on risk governance: towards an integrative approach, can be downloaded from www.irgc.org 22. us department of energy. (1994). doe standard preparation guide for us department of energy nonreactor nuclear facility safety analysis reports, washington, dc, doe-std-3009-94 23. woodruff, j, m. (2005). consequence and likelihood in risk estimation: a matter of balance in uk health and safety risk assessment practice. safety science. 43, 345–353 24. ball, d. j. and watt, j. (2013), further thoughts on the utility of risk matrices. risk analysis, 33: 2068–2078 25. cox, jr., l. a. (2009), what's wrong with hazardranking systems? an expository note. risk analysis, 29: 940–948 26. iso (international standards organisation). risk management:risk assessment techniques, 2009. iso/iec 31010.geneva: iso. 27. integ-risk (2013). integ-risk-early recognition, mointoring and integrated management of emerging, new technology related risks (2007-2013), www.integrisk.eu-vri.eu published by atlantis press copyright: the authors 56 strengths and weaknesses of risk matrices table 2. the consequence scores used by the (national patient safety agency in england) table 3. the likelihood scores 1 2 3 4 5 descriptor rare unlikely possible likely almost certain frequency not expected to occur for years. expected to occur at least annually. expected to occur at least monthly. expected to occur at least weekly. expected to occur at least daily. <1% 1-5% 6-20% 21-50% >50% probability will occur in exceptional circumstances. unlikely to occur. reasonable chance of occurring. likely to occur. more likely to occur than not. 1 2 3 4 5 descriptor insignificant minor moderate major catastrophic objectives / projects insignificant cost increase/schedule slippage. barely noticeable reduction in scope or quality <5% over budget/schedule slippage. minor reduction in quality/scope 5-10% over budge/schedule slippage. reduction in scope or quality 10-25% over budget/schedule slippage. does not meet secondary objectives >25% over budget/schedule slippage. does not meet primary objectives injury minor injury not requiring first aid minor injury or illness, first aid treatment needed riddor/agency reportable major injuries or long incapacity/disability (loss of limb) death or major permanent incapacity patient experience unsatisfactory patient experience not directly related to patient care unsatisfactory patient experience – readily resolvable mismanagement of patient care. serious mismanagement of patient care totally unsatisfactory patient outcome or experience complaint/cla ims locally resolved justified complaint peripheral to clinical care below excess claim. justified complaint involving lack of appropriate care claim above excess level. multiple justified complaints multiple claims or single major claim service business interruption loss/interruption > 1 hour loss/interruption >8 hours loss/interruption >1 day loss/interruption >1 week permanent loss of service of facility staffing & competence short-term low staffing level temporarily reduces service quality (< 1 day) ongoing low staffing level reduces service quality late delivery of key objective/service due to lack of staff. minor error due to poor training. ongoing unsafe staffing level. uncertain delivery of key objective/service due to lack of staff. serious error due to poor training non-delivery of key objective/service due to lack of staff. loss of key staff. critical error due to insufficient training financial small loss loss >0.1% of budget loss >0.25% of budget loss>0.5 of budget loss >1% of budget inspection/aud it minor recommendations minor non-compliance with standards recommendations given. noncompliance with standards reducing rating. challenging recommendations. non-compliance with core standards enforcement action. low rating. critical report. major noncompliance with core standards prosecution. zero rating. severely critical report adverse publicity/reput ation rumours local media – short term. minor effect on staff morale local media – long term. significant effect on staff morale national media <3 days national media >3 days. mp concern (question in house of parliament) published by atlantis press copyright: the authors 57 research article analysis of long-term care insurance demand of megalopolis in china under the background of aging of population—take shanghai as an example† guo jing1, xiaojun pan2, chengyi pu1,* 1school of insurance, central university of finance and economics, beijing 100081, china 2institute of disaster prevention and mitigation, beijing normal university, beijing 100028, china 1. introduction china’s aging has the characteristics of getting older before getting rich, growing rapidly, and having a large population base. as of the end of 2018, according to the data of the national bureau of statistics, the number of people above 65 in china is 165.65 million, accounting for 11.94% of the total population. among them, the old-age dependency ratio is 16.8%, which means that every 100 working population in china needs to support 16.8 old people. the aging disability phenomenon in china’s mega-cities represented by shanghai is particularly serious. according to the data compiled from the shanghai sixth census bulletin, the disability rates of the 60–80-year-olds in shanghai are 5.0275%, 14.8139%, and 32.7914%, the severe aging trend has promoted the demand for elderly care. judging from the current development of the pension system in china, although the elderly, the disabled, and the need for care for diseases are extremely high, there is no directly compatible policy system and corresponding insurance products, the supporting facilities for elderly care need to be resolved urgently. firstly, china’s social security system is not perfect. although the medical insurance of the five social insurances can solve the reimbursement of patients’ medical expenses, most of the conditions that are not reimbursable are the exclusion of nursing and rehabilitation treatment expenses, that is, the expenses of nursing after the patient is discharged are not involved. secondly, the supply of commercial nursing insurance products in china is in short supply. the major illness insurance in the current commercial insurance, its claims are mainly illness treatment and loss of income, etc., as a supplement to current social insurance, it also has limitations. on the one hand, there is a threshold for the payment of critical illness insurance. if the insured does not meet the payment conditions, commercial insurance companies a r t i c l e i n f o article history received 12 october 2020 accepted 10 march 2021 keywords the demand for long-term care insurance rapid aging of the population megacity megalopolis shanghai a b s t r a c t the development of long-term care insurance (ltci), which matches that people get old before becoming rich lag behind. the mega-urban agglomeration formed in the metropolitan area under the process of urbanization intensifies the demand for insurance driven by long-term care needs. this paper constructs a theoretical model to sort out the micro-action mechanism of ltci products entering the production field through the consumption of corporate security services, and at the same time entering the consumption field in the form of security services demand from families, individuals and government agencies, it reveals that the differences in the micro-paths of ltci products entering the enterprise production field and the household consumption field under different regional economic and social conditions. then, using the gm (1,1) model to analyze the demand for ltci under the aging population trend of shanghai, a super large city in china, we found that the situation and characteristics of shanghai and japan are similar, and the number of people who need care under the distribution of the elderly population in shanghai from 2017 to 2027 is much lower than the number of people who need ltci. meanwhile, the current long-term nursing service has a large labor gap and high professional skills requirements, but the current salary is low. besides, the share of elderly care institutions is small, which is not enough to support the current occupancy needs of severely disabled people. the actual development has a greater potential impact on public health expenditures, especially the increase in public health expenditures after the outbreak of the covid-19 has brought uncertainty. based on the above findings, we propose to introduce the successful mode of american political and commercial cooperation and the japanese social security model to alleviate the rapid growth in the demand for ltci accompanied by the aging trend in the mega-urban clusters formed in the urbanization process. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: pucy@cufe.edu.cn †this article is supported by: 1the “strategic economy and military-civilian integration” interdisciplinary discipline (gjj2019163) of beijing universities’ high-precision disciplines, the “first-class discipline” construction project of the central university of finance and economics, “research on modernization of catastrophe risk management and natural disaster governance”. 2the national ethnic affairs commission in 2020 supported by the project “research on rural revitalization and built-in financial innovation in western tibetan areas in the post-poverty era” (2020-gmb-030). 3national natural science foundation of china, general project, project name: research on the mechanism and early warning mechanism of nursing risk in elderly care institutions; project batch number: 71871147. journal of risk analysis and crisis response vol. 11(1); april (2021), pp. 16–25 doi: https://doi.org/10.2991/jracr.k.210310.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:pucy%40cufe.edu.cn?subject= https://doi.org/10.2991/jracr.k.210310.001 https://www.atlantis-press.com/journals/jracr g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 17 will not pay. on the other hand, even if the payment conditions are met, the insured amount of the major illness insurance is a one-time payment, which does not apply to people who need long-term care. therefore, to respond to the increasing demand for care under the background of china’s aging, and to realize the social vision of “the elderly are dependent”, china is currently conducting a pilot program of long-term care insurance (ltci). shanghai is china’s super first-tier city, and its economic development is at the forefront of my country. the implementation of the system has exemplary significance for the country. therefore, based on the goal of establishing a ltci system for elderly care, it is necessary to start with the pre-judgment of the elderly disabled population to further clarify the care needs, to establish a scientific and reasonable ltci system. this article chooses shanghai as the research object, and mainly discusses and forecasts population as the introduction point. 2. literature review 2.1. summary of the evolution of the long-term care insurance system the original ltci model originated in the united states in the 1970s. with the deepening of domestic aging in the united states, the care needs of the elderly are increasing year-by-year. in 1996, the health insurance portability and accountability act came out and made a far-reaching reference to long-term insurance costs. subsequently, many countries in europe and asia also have different forms of ltci [1–4]. for example, in the dual mechanisms represented by france, although critical illness insurance covers most of the hospital’s health expenditures in its medical system, the direct funding of family assistance only accounts for 9%, requiring the introduction of compulsory care insurance. commercial nursing insurance was introduced as a supplementary mechanism in subsequent years [5]. in the universal social security system represented by japan, the form of payment is to provide benefits regardless of income or family status [6]. generally speaking, the european region is mainly represented by the compulsory insurance of germany and france, while the more developed countries in asia are mainly japan. the overall trend of global ltci development is to change from the early government-sponsored model to a government-led corporate business model and then to a market led corporate business model. there is no mandatory ltci system in the united states. the market-oriented model represented by the united states is essentially a joint guarantee of social insurance and commercial insurance, it is mainly operated by commercial insurance companies in the insurance market, with government intervention and subsidies. there are three main categories in social and commercial insurance [7]. the government mainly provides limited care protection to certain groups of people through medicare and medicaid, while the rest of the population is covered by commercial ltci: (1) medical care (medicare), which mainly covers the elderly above 65 years old, providing nursing care for the elderly for daily medical treatment, medication, and hospitalization within a certain number of days, individuals do not need to bear the cost. (2) medical assistance (medicaid) is mainly operated by institutions, and the implementation subject is still the government. the subsidy fee is mainly for the operator of the institution since its original intention is to medical help, participants do not need to make personal payments. (3) long-term care cooperative plan is a form of cooperation between the government and the enterprise: the government takes care of the bottom line, the company conducts private operations, and the insured voluntarily insure, and the cost of insurance is higher than the previous two forms. the form of commercial insurance is mainly operated by commercial insurance companies, it is aimed at middleand high-income groups and it is also a supplement to social insurance and aims to meet the willingness of higher-demand groups to purchase insurance. next is the universal social insurance model represented by japan. the japanese government introduced the “long-term care preventive measures” in 2006. in the past 10 years of development, several bills have been introduced [8]. among them, the “longterm care insurance law” was reformed in 2015, covering all people above 40 years of age. there are two types of insurance subjects, divided by age: nationals between 40 and 65 years old are called no. 2 insurers, and nationals 65 and above are called no. 1 insurers; among them, according to the family and income, it is divided into five levels, as shown in table 1. while providing long-term care protection, this measure is organizing community salons for the elderly above 65 years old. the cost of participation is relatively low, which can effectively screen them regularly to identify high-risk groups and promote social activities for the elderly. 2.2. research progress of long-term care insurance demand theory in the early discussion of ltci theory by foreign scholars, it was mainly based on the assumption of rational people [9,10]. most of them are thinking about the family’s intergenerational relationship. it is found that the consumption of ltci is affected by the income level of the children. low-income children are more inclined to table 1 | overview of the implementation of long-term care insurance in japan coverage objects grades main component no. 1 insurers (people aged above 65) level 1: recipients of social assistance home services: such as home care, daycare, etc.; community services: daily life care, etc.; institutional services: nursing homes, health, and medical institutions, etc. among them, the no. 2 insurers are paid for specific 16 diseases (such as advanced cancer, early-stage alzheimer’s disease, cerebrovascular disease, etc.). level 2: municipalities tax exempt households, etc. level 3: municipalities tax exempt individuals, etc. level 4: a taxon insured income of fewer than 2.5 million yen level 5: a taxon insured income of more than 2.5 million yen no. 2 insurers (people aged 40–65) other categories 18 g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 take care of their parents in person [11–14] using hrs data analysis found that children’s care for their parents will be replaced with income. some scholars, based on the sociology of population, found that individuals with higher incomes and married people have more urgent needs for ltci [15,16]. although china’s ltci theory research started late, the results of research on aging are relatively rich, and it is generally believed that my country’s aging has three major characteristics [17]. firstly, is get old before getting rich. our country has entered an aging society when the overall national strength is not high. secondly, is unprepared to get old. china’s pension system will face the strong impact of aging when it is not perfect. thirdly is loneliness. due to the widespread phenomenon of “fewer births” today, the elderly are prone to getting old and difficult to support due to various reasons such as disease and loneliness. in the government’s response strategy, in the face of a major change in the demographic structure, the government needs to build strong public service measures for elderly care, centering on the actual needs of the elderly to ensure the quality of elderly care. furthermore, with the acceleration of aging in china, the concept of “long-term care” has begun to be studied in depth. many scholars have compared the experience of foreign ltci implementation and believe that china should establish a ltci mechanism with chinese characteristics [18–21]. in recent years, the content of the current development model across the country is quite different, the development of ltci can be carried out by providing customers with protection and avoiding moral hazard [22]. furthermore, in the research methods of ltci, the existing literature generally focuses on the demand factors of ltci [23–25]. for example, cross-sectional data from the share database (european survey of health, aging and retirement) to estimate the determinants of physical health for purchasing ltci in france, the sample variables include bmi index and disease status, risky behavior and disability experience explain the characteristics of french ltci demand [25], however, the mechanism design of how to design public policies to encourage the growth of long-term insurance is very lacking [26]. domestic research is still in a situation of following foreign countries, for example, panel data analysis of the linear relationship between independent variables such as household savings, oldage dependency ratio, and the dependent variable ltci premium income suggests the trend of the macro demand characteristics of ltci [27]. economic factors and education status substitute microscopic investigation logistics probability model analysis to insure ltci change factors and whether respondents prefer [28,29]. however, due to the short period of domestic pilot projects and the lack of data compared with foreign countries, domestic scholars mainly use questionnaire surveys to obtain data, and most often use the logistics model for analysis. in summary, regardless of the evolution of the ltci system model or the development of ltci theory, china lags behind foreign countries. however, domestic urbanization and aging have completed the rapid transformation of more than 300 years abroad in just above 30 years. in particular, the rate of aging is faster than abroad, and the rate of an ultra-high aging population is higher. the mega-urban agglomeration formed in the metropolitan area under the process of urbanization has intensified the demand for insurance driven by the demand for long-term care. in reality, there is an urgent need for ltci system model reform and ltci theoretical research to keep up with the needs of actual development, and build a true and worry-free life for the people. because the ltci has not been implemented for a long time, the relevant data is still blank, and the relevant empirical analysis is all based on questionnaire surveys, and there are few predictive analyses on the population. this article starts with the aging trend and the nursing needs of the elderly disabled population and uses the gm (1,1) model to predict the population simulation to explore the insurance demand potential driven by the long-term care demand in the process of urbanization. 3. analysis of the potential mechanism of long-term care insurance 3.1. a theoretical model of consumer economic activity in demand for long-term care insurance from the perspective of insurance demand acting on insurance consumption, as the current pilot ltci in china is mainly social security mode, insurance consumption and welfare payment lag behind. therefore, this paper puts insurance as a safety service of production and people’s life into the general consumption category, and the demand for ltci is measured by consumption function. (1) the insured behavior is expressed by the production function as the input in the young age [30]. (2) the payment in the old age is measured by the utility function. (3) assuming that the production factor is a necessary condition of the production function, physical capital can be used as a risk backup when it is not insured, and insurance backup does not require excessive material backup, the insurance compensation (contingent claim) is a kind of market-oriented pre-preparation in advance. therefore, physical capital and insurance consumption are replaceable. based on the academic works of related scholars [31,32] this article improves the cobb–douglas production function: q = alakb. let k(t) be the capital stock, c(t) be the consumption of general commodities, and e(t) be the consumption of special insurance commodities. among them, er(t) is the consumption of insurance products entering life, and ep(t) is the consumption of insurance products entering production. insurance investment cost is measured by consumption. (4) assuming that the production function y(k, l, ep) is strictly concave, first-order homogeneous and second-order continuously differentiable. since the population changes little in the short term, it can be assumed that the labor force l is constant. then, the production function should be a function of the physical capital k and the insurance consumption ep(t) (calculated as a cost) into production, set to y(k, ep). and there are: f e f k f fp ke e kp p( , ) ( , ) ,0 0 0 0= = = > (1) if consumption is divided into ordinary consumption and special consumption, it is assumed that special consumption is only insurance consumption. then the total insurance consumption e composed of insurance entering production and consumption and insurance entering life consumption in a certain period, e remains unchanged. g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 19 generally speaking, insurance for production and consumption in poor areas is more than insurance for living consumption, developed regions have more insurance for production and consumption. which is: e e er p+ = (2) defining the consumption function u(c, er) : r+→ r+ is a non decreasing function, and the consumption function is a second order continuous differentiable function with diminishing marginal utility. let r be the discount rate, then the utility function is u c e e dtr t( , ) .¥ ò r0 (3) then the above problem becomes to solve the following dynamic optimization problem: max ( , ) . : ( , ) _ . u c e e e dt st k f k e c k p t p = ì í ï î ï -¥ ò r d 0 (4) let k(0) = k0 > 0, c > 0, ep ≥ 0, l be the shadow price of capital, d be the depreciation rate of physical assets, and the depreciation rate of monetary assets is 0, then the hamilton equation: h u c e e f k e c kp p= + -éë ùû( , ) ( , )l d (5) the necessary and sufficient conditions for the optimal path are: l l l l r d = = = + ì í ï î ï u c e e u c e e f k e f e e c p e p e p k p r p ( , ) ( , ) ( , ) [ ( , )]� (6) then there is u u fe c er = p, indicating that the marginal utility value of insurance used for personal household consumption must be equal to the marginal utility of ordinary consumption multiplied by the value of the marginal product of insurance for production and consumption of the enterprise. let � �k = =l 0 , * a represents the special solution of the variable in the steady-state, and the results are as follows: k f f e c k f k e p k p = = == = + -éë ùû ì í ï îï 0 0 ( , ) ( , ) * * * * * * * d l l r d� (7) when λ* > 0 then r d+ = f k ek p( , ) * * , c f k e e k ep p p= −[ ( ), ] ( ) * * *d , from (6a) and (6b) we can get: u f k e e k e e e f k e e u f k e c p p p p e p p e p p p [ ( ), ] ( ), [ ( ), ] [ ( ) * * * * * * * -{ } = d ,, ] ( ),* * *e k e e ep p p-{ }d (8) from equation (8), the left is the total marginal productivity of insurance consumption (vmpep ), and the right is the marginal cost of insurance products for production and consumption (mc ep ). then: d de u c f dk de f u d de e p e k p e e e e p p r p r r p mc vmp = −     +         − = ( )d uu f f dk de f u f u f f cc e k p e ce e c e e ke p p r p p p p ( )−     +         − + − d 2 ffkk               <            0 (9) it means that the production function is strictly concave, so in the steady-state, there are (1) fk – d = r is a positive value and ucep > 0. (2) in the interval [ , ]0 e , mc ep is a monotonically increasing function, and vmpep is a monotonically decreasing function. then: ¶ ¶ = -æ èç ö ø÷ < ¶ ¶ = -æ èç ö ø÷ < = c k f u e k f u f e k ce p e k cc e p p p p l l l d d d ( ) ( ) 0 0 ee cc e e cc cep p r r u u u u+ > ì í ï ï î ï ï 2 0 (10) because equation (7b) shows that fkep > 0, the insurance of production and consumption is increasing. therefore, equations (10a) and (10c) show that when the stock of wealth capital increases, the marginal productivity of wealth capital decreases, and the marginal productivity must be less than r + d . moreover, there is an alternative relationship between insurance for production consumption and insurance for life consumption, in other words, there is an alternative relationship between a corporate annuity and personal commercial ltci. also, since u²(·) < 0, then when consumption decreases, uc must increase. the larger the investment in the early stage of the production function, the greater the welfare for the elderly in the later stage, so the introduction of ltci is extremely necessary. as one of the leading cities in the megalopolis, shanghai has strong regional advantages and economic strength. policy subsidies and the participation of individuals can increase the benefits for the elderly. 3.2. analysis and prediction of the shanghai population-based on gm (1,1) model the source of the data in this article is the total population of shanghai registered household registration and the number of elderly people above 60 years of age who are registered in the shanghai statistical yearbook from 2004 to 2017. it can be seen from figure 1 that the degree of aging in shanghai has been increasing year-by-year. especially in 2014, its aging growth rate reached 6.33%. shanghai has reached a state of deep aging in 2015. among them, the proportion of the population above 60 years old is 30.21%. according to the data in figure 1, a corresponding population prediction model is established by the gray prediction analysis theory below. figure 1 | the growth rate of aging from 2004 to 2016. data source: shanghai statistical yearbook 2005–2017. 20 g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 first use the data of the elderly above 60 to form the original sequence: x k x x x( ) ( ) ( ) ( )( ) ( ), ( ), , ( ) . , . , . 0 0 0 01 2 14 260 78 266 37 275 6 = ¼éë ùû = 22 286 83 300 57 315 70 331 02 347 76 367 32 387 62 413 98 , . , . , . , . , . , . , . , . ,, . , . , .435 95 457 80 481 61 é ë ê ù û ú accumulate x(0) once and generate 1-ago sequence: x ( ) ( ), ( ), , ( )( ) ( ) ( ) ( )1 1 1 11 2 14k x x x= …  where x k x i k i k ( ) ( )( ) ( ), , , ,1 1 0 1 2 14= = ¼ = å let z(1) = [z(1)(2), z(1)(3), …, z(1)(14)] be the sequence generated by x(1)(k) where z k x k x k k( ) ( ) ( )( ) ( ) ( ) , , , ,1 1 1 1 2 1 1 2 14= + -éë ùû = ¼ . construct two sets of matrices, respectively: b z z z = ¼ ¼ é ë ê ù û ú ( ) ( ) ( )( ) ( ), ( ), , , , , ; 1 1 12 143 1 1 1 y x x xn = ¼ éë ùû ( ) ( ) ( )( ), , , ( ) .( )0 0 02 143 the vector of coefficients to be estimated as a b b b yt t n �= × ×-( ) 1 establish gm (1,1) gray estimation model, its differential equation is: x k a z k b( ) ( )( ) ( )0 1+ ´ = after generating the corresponding sequence of time, the corresponding variance is: x k x b a e b a kak� ( ) ( )( ) ( ) , , , , 0 01 1 1 2 13+ = é ë ê ù û ú× + = ¼ that is, the value of a b� �= =0 00564432 1354 727722. , . . then do cumulative subtraction reduction x k x k� � ( ) ( ) ( ) ( ) 1 1 1+ to calculate the corresponding value. in the same way, for the estimation of shanghai’s registered population, let another original sequence: p k p p p( ) ( ) ( ) ( )( ) ( ), , , ( ) . , . , ( )0 0 0 01 14 1352 39 1360 26 136 2= ¼éë ùû = 88 08 1378 86 1391 04 1400 70 1412 32 1419 36 1426 93 1432 . , . , . , . , . , . , . , .. , . , . , . , .34 1438 69 1442 97 1450 00 1455 13 é ë ê ù û ú test the two levels: s ( ) ( ) ( ) . , . , . , ( ) ( )k x k x k = = 0 0 1 0 994214341 0 99428396 0 992181947 0.. , . , . , . , . 991243961 0 993103448 0 991772403 0 995040018 0 9946949044 0 996222964 0 995586263 0 997033895 0 995151724 0 99647 , . , . , . , . , . 44542 é ë ê ê ê ê ê ù û ú ú ú ú ú ∈ −    = − − 2 2 0 875173319 1 142630812 1 1e en n , ,( . . ). similarly, g ( ) ( ) ( ) , ( ) ( )k p k p k e en n = î -æ èç ö ø÷0 0 1 1 1 2 2 . then use a� and s (k) to test the level ratio deviation value. the value is to meet higher requirements. therefore, x(0)(k) and p(0)(k) can be used to construct a satisfactory gray prediction model. since only the data before 2018 is obtained, this article aims at the medium and long-term forecast, predicting the population distribution in the next 10 years (2017–2027) (tables a1–a3). it can be seen that because this gray prediction model level ratio deviation value test has reached higher requirements, the population prediction is basically consistent with the original data, and the conclusion that the prediction data is credible can be drawn. the following analysis of the ltci model is based on this data (figure 2). 3.3. calculation of nursing needs 3.3.1. calculation of disability of the elderly population based on the above results of using the gm (1,1) gray prediction model, this paper selects the sixth population census in shanghai, and statistically collates the health and disability of the elderly population. by calculating the corresponding distribution ratio, age is divided into three stages: 60–79 years old and above 80 years old. health status is measured by self-care ability, it is divided into the elderly who are not disabled, partially disabled and completely disabled and the latter two are the composition of the disability rate as shown in table 2. according to calculations, among the distribution of the elderly above 60 years old, 87.2711% are not disabled, 9.0208% are partially disabled; 3.7081% are completely disabled, and the disability rate is 12.289%. the number of elderly disabled people and the number of people who need ltci for the next 10 years from 2016 to 2027 are shown in table 3. figure 2 | simulation of the original population above 60 and the predicted population. data source: shanghai statistical yearbook 2005–2017. table 2 | distribution of disability among the elderly population age distribution 60–69 years old (%) 70–79 years old (%) above 80 years old (%) total 100.0000 100.0000 100.0000 not disabled 94.9725 85.1861 67.2086 partially disabled 4.1806 11.3786 19.6793 completely disabled 0.8469 3.4353 13.1121 disability rate 5.0275 14.8139 32.7914 data source: the sixth population census in shanghai. g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 21 it can be seen from figure 3 that both the completely disabled population and the partially disabled population are increasing. in particular, the partially disabled population has increased significantly, indicating that the number of people in need of moderate care in the future is relatively large. under the rising trend of the elderly population in the future, it can be concluded that in the next 10 years, the number of ltci needs will be about 60, accounting for 1/8 of the elderly population. 3.3.2. elderly disabled population forecast for different care category the ratings of the disabled elderly in various regions of china are mainly assessed concerning the barthel index, which is a commonly used internationally accepted activities of daily living (adls). the treatment standards are mainly divided into three categories according to the degree of completely disability: (1) the first care category: the assessment level is two or three; (2) the second care category: the assessment level is four; (3) the third care category: the assessment level is five or six. with reference to the national research on the status of the disabled elderly in urban and rural areas issued by the research group of the china research center on aging in 2011, it is pointed out that in 2010, the proportions of mild, moderate and severe disability of the completely elderly who in china were 84.3%, 5.1% and 10.6%, respectively. combining the nursing treatment standards in shanghai, and based on the values in table 3, we can get the predicted number of total disabled elderly care grade distribution in shanghai from 2017 to 2027. it should be noted that since it is difficult to divide the standard of the number of partially disabled people, the number of partially disabled people is listed separately here, as shown in table 4. 3.3.3. calculation of the number of home care workers required according to the payment policy of the shanghai ltci fund, the payment levels for community care at home and elderly care institutions are 90% and 85% respectively. inpatient medical care is mainly determined by actual conditions since the latter two contents vary greatly with actual conditions, this article selects the home care to estimate the nursing staff. the first care category, the second care category, and the third care category of nursing standards are mainly performed in the form of door-to-door services. the standards are three times a week, five times a week, and seven times a week and each service last 1 h. if part of the disabled population is considered in the first type of care standard, based on the 40 h/week working hours per person, the number of nursing staff who need home care can be roughly estimated, as shown in table 5. 4. the dilemma of long-term care insurance in shanghai 4.1. the population who has received reimbursement from long-term care insurance is inadequate according to the data published in the “shanghai elderly population and elderly career monitoring statistics information” in 2018–2019, figure 3 | the evolution of the disabled elderly population forecast in shanghai from 2016 to 2027. table 3 | 2016–2027 elderly disabled population and long-term care insurance (ltci) demand forecast year population above 60 (10,000 people) partially disabled (10,000 people) completely disabled (10,000 people) ltci demand forecast (10,000 people) the current number of participants in ltci (10,000 people) 2016 457.8000 41.2970 16.9758 58.2728 * 2017 481.6100 43.4448 17.8587 61.3036 * 2018 505.2430 45.5767 18.7351 64.3118 23.4 2019 531.9568 47.9865 19.7256 67.7122 49.3 2020 560.0831 50.5237 20.7686 71.2923 * 2021 589.6965 53.1951 21.8667 75.0618 * 2022 620.8757 56.0077 23.0229 79.0305 * 2023 653.7034 58.9690 24.2402 83.2091 * 2024 688.2668 62.0868 25.5218 87.6087 * 2025 724.6577 65.3696 26.8712 92.2408 * 2026 762.9727 68.8259 28.2920 97.1179 * 2027 803.3136 72.4649 29.7879 102.2528 * note: forecast data after 2017. *represents data not yet available. table 4 | 2016–2027 elderly disabled population forecast for different care category year partially disabled (10,000 people) completely disability (10,000 people) the first care category the second care category the third care category 2016 41.297 14.3106 0.865766 1.799435 2017 43.4448 15.05488 0.910794 1.893022 2018 45.5767 15.79369 0.95549 1.985921 2019 47.9865 16.62868 1.006006 2.090914 2020 50.5237 17.50793 1.059199 2.201472 2021 53.1951 18.43363 1.115202 2.31787 2022 56.0077 19.4083 1.174168 2.440427 2023 58.969 20.43449 1.23625 2.569461 2024 62.0868 21.51488 1.301612 2.705311 2025 65.3696 22.65242 1.370431 2.848347 2026 68.8259 23.85016 1.442892 2.998952 2027 72.4649 25.1112 1.519183 3.157517 note: forecast data after 2017. 22 g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 the number of people above 60 years old in shanghai receiving nursing services in 2018 totaled 234,000, accounting for 4.6% of the elderly population, the scope of the pilot program has been greatly expanded in 2019, and the number of elderly people above 60 years old in the city receiving nursing services totaled 493,000, accounting for 9.5% of the elderly population. from the results in table 3, it can be concluded that the number of people in need of ltci in 2018–2019 is much greater than the number of people in the pilot, the demand potential of ltci is huge. 4.2. supporting facilities for ltci are relatively absent it can be seen from the above-predicted value that the difficulty of follow-up by nursing staff and nursing institutions is an important reason for the relatively lagging development of ltci in china. firstly, there is a large shortage of nursing staff in ltci. combined with the data from the shanghai statistical yearbook and table 5, the predicted number of nursing home care in 2018 alone will require more than 50,000 people, while the total number of health technicians in 2018 was only 206,500. if we add institutional care and inpatient medical care, it is unrealistic to rely solely on the current 206,500 people. therefore, it can be judged that there is a huge shortage of nursing staff for ltci. in terms of salaries, the 2015– 2017 salaries of shanghai residents’ service, repair and other service industries were 33,720, 37,292, and 40,478 rmb per year respectively, which were lower than other industries such as agriculture, forestry, and animal husbandry. currently, there are relatively few people working in health and social work, and people are more inclined to move to tertiary industries with higher salaries such as wholesale and retail, it can be seen that the promotion of ltci still needs to be achieved by attracting salary. at the same time, if the government intends to train professional nursing talents, it needs to invest a lot of money. secondly, the number of nursing institutions is still in a small share. in 2018, the number of elderly nursing homes and hospitals in shanghai was only 38 and two, and the number of inpatients was 26,001 and 1863. currently, the people who can participate in institutional care are elderly people with the third type of nursing standard. according to table 4, the predicted number of people with the third care category in 2018 is close to 20,000, which is equivalent to the overall value of the current elderly care institution. the number is not enough to support the occupancy needs of the elderly with full weight disability, and the elderly with a severe disability still largely rely on home care. 4.3. the burden structure of health expenses is under certain pressure since the implementation of the ltci pilot program is closely related to medical resources, public health expenditures can be used as a reference standard. as shown in figure 4, shanghai’s public health expenditures have been on the rise, and the trend is that the proportion of personal health expenditures has fallen, staying at about 20%, while the proportions of government health expenditures and social health expenditures have been increasing. the health expenses of shanghai ltci are mainly paid by the special ltci fund, which is a new fiscal expenditure item. the long-term care policy based on aging will have a certain impact on shanghai’s fiscal expenditure. correspondingly, due to the impact of the covid-19 in china in 2020, the national medical resources may further participate in the fight against the epidemic, which will greatly increase the corresponding health expenses. therefore, the expenditure on health expenditure is uncertain. despite this, we can still analyze the severity of the elderly population, and more public expenditure on health can be foreseen in the next 10 years (table a4). in summary, ltci needs corresponding aging care needs to exist objectively, meanwhile, the implementation of the policy is under great pressure: the shortage of nursing staff and the low salary, the impact of the special fund for ltci on fiscal expenditures, and the collateral effect of covid-19 have increased the uncertainty of public health expenditure. 5. conclusion and prospect this paper uses the gm (1,1) model to predict and analyze the total registered population and the number of elderly people aged 60 years and above in shanghai in the next 10 years, combined with the health of the elderly population above 60 years old in the sixth table 5 | 2016–2027 calculation of the number of home care workers required year the first care category (10,000 people) the second care category (10,000 people) the third care category (10,000 people) total (10,000 people) 2016 4.17057 0.108221 0.314901 4.59369188 2017 4.387476 0.113849 0.331279 4.8326041 2018 4.602779 0.119436 0.347536 5.06975168 2019 4.846139 0.125751 0.36591 5.3377992 2020 5.102372 0.1324 0.385258 5.62002973 2021 5.372155 0.1394 0.405627 5.91718225 2022 5.6562 0.146771 0.427075 6.23004573 2023 5.955262 0.154531 0.449656 6.55944868 2024 6.270126 0.162702 0.473429 6.90625693 2025 6.601652 0.171304 0.498461 7.2714161 2026 6.950705 0.180362 0.524817 7.6558826 2027 7.318208 0.189898 0.552565 8.06067085 note: forecast data after 2017. figure 4 | 2004–2015 shanghai health expenditure structure. data source: shanghai statistical yearbook 2005–2017. g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 23 census of shanghai in 2010. based on the census data, this article mainly draws the following conclusions: theoretical derivation shows that ltci is affected by regional economic and socio-economic conditions in megacities. the micropath and difference between ltci products entering the enterprise production field and the household consumption field, among them, production insurance consumption depends on changes in wealth and capital stock. the larger the capital stock, the more production insurance consumption. empirical research shows that, first of all, the predicted population of long-term care needs in shanghai can be obtained by calculation, which is much higher than the actual participating population in 2018 and 2019. the current gap in ltci is large, especially for some disabled population. the largest increase indicates that the number of people in need of moderate care in the future is large, and the supply of mid-term care should be considered. secondly, through the calculation of the expected number of people with disabilities in the nursing level and the corresponding number of people in shanghai, combined with the current service industry salary, it is found that the nursing staff of ltci has a large gap and low salary. moreover, the number of corresponding nursing institutions is extremely small, and the degree of institutional care is low, which is not enough to support the current occupancy needs of disabled people for institutional care. third, there is a certain pressure on the burden structure of shanghai’s health expenses. lastly, there is a certain pressure on the burden structure of shanghai’s health expenses. the special fund expenditure for ltci and the effect of covid-19 have increased the uncertainty of public health expenditures. if the government increases the implementation of ltci, it will inevitably increase the corresponding public resource expenditures. several factors may exert a certain pressure on the development of long-term protection insurance. based on the above conclusions, this article gives the following suggestions: to begin with, based on the results of theoretical derivation, the promotion of long-term protection insurance should be carried out with the idea of “large–medium–small” urban agglomerations. good regional economy and material conditions are conducive to the consumption of ltci. the first promotion of ltci in big cities is to lay the foundation and to explore suitable development experience and consumption stock, which may be more beneficial to small and medium-sized cities. secondly, is to build a medical security system for ltci for the elderly and improve supporting facilities. the increase in the degree of aging will promote health expenditures, especially after the outbreak of the new crown epidemic, the larger increase in health expenditures is foreseeable. in terms of public resources, china’s future trend is to further reduce personal health expenditures, the government should fully consider the proportion of personal contributions and medical pooling funds, and plan the corresponding resource allocation. in terms of the supply of supporting facilities, the salary of nursing staff should be appropriately increased based on the investment in pension welfare institutions and medical institutions. last but not least, it is to encourage corresponding commercial insurance policies to complement. on the one hand, from the above data, it can be seen that the government is facing greater pressure to build a comprehensive ltci social security system, thus, we suggest to cooperate with commercial insurance companies to fully mobilize resources for ltci exploration and play a complementary role in social security. on the other hand, is to actively try to combine with the corresponding pilot system, such as the possibility of combining with the housing pension system: is it possible for the elderly to reverse mortgage the property and move into the designated institution of ltci? in summary, whether it is the form of universal social insurance in japan or the mode of political and commercial cooperation in the united states, its core cannot be separated from the needs of people with ltci. china’s current ltci policy, in essence, is to draw lessons from the foreign pension system, the mode and development process of the implementation of foreign countries, are worth learning experience, especially on the elderly insurance group needs and the potential number of insurance. in the exploration of ltci, for the insured, the degree of policy awareness is one aspect, and the definition of the insured population is another aspect. on the national side, the needs for the construction of pilot nursing service manpower, material resources, and the elderly care environment are worthy of in-depth exploration. starting from the perspective of participants, it is extremely necessary to build a system that covers their health awareness, economic awareness, and policy awareness. conflicts of interest the authors declare they have no conflicts of interest. references [1] tamiya n, noguchi h, nishi a, reich mr, ikegami n, hashimoto h, et al. population aging and wellbeing: lessons from japan’s long-term care insurance policy. lancet 2011;378:1183–92. [2] wasem j. a study on decentralizing from acute care to home care settings in germany. health policy 1997;41:s109–s29. [3] rhee jc, done n, anderson gf. considering long-term care insurance for middle-income countries: comparing south korea with japan and germany. health policy 2015;119:1319–29. [4] launois r. pooling public and private funds in the patient’s interest: the case for long-term care insurance. soc sci med 1996;43:739–44. [5] german insurance association. statistical yearbook of german insurance 2016. 2016. available from: https://www.nextdeal.gr/ pdfs/ereunes/stat_yearbook_2016.pdf. [6] umegaki h, yanagawa m, nonogaki z, nakashima h, kuzuya m, endo h. burden reduction of caregivers for users of care services provided by the public long-term care insurance system in japan. arch gerontol geriatr 2014;58:130–3. [7] limra, u.s. group long-term care insurance executive summary. annual 2005. 2005. available from: https://www. limra.com/en/research/research-abstracts/classics/2006/u.s.group-long-term-care-insurance-2005-annual/. [8] ministry of health, labor and welfare. long-term care insurance system of japan, health and welfare bureau for the elderly ministry of health, labor and welfare. 2016. available from: https://www.mhlw.go.jp/english/policy/care-welfare/care-welfare elderly/index.html. [9] gupta a, li l. a modeling framework for optimal long-term care insurance purchase decisions in retirement planning. health care manag sci 2004;7:105–17. https://doi.org/10.1016/s0140-6736%2811%2961176-8 https://doi.org/10.1016/s0140-6736%2811%2961176-8 https://doi.org/10.1016/s0140-6736%2811%2961176-8 https://doi.org/10.1016/s0168-8510(97)00046-8 https://doi.org/10.1016/s0168-8510(97)00046-8 https://doi.org/10.1016/j.healthpol.2015.06.001 https://doi.org/10.1016/j.healthpol.2015.06.001 https://doi.org/10.1016/j.healthpol.2015.06.001 https://doi.org/10.1016/0277-9536(96)00118-9 https://doi.org/10.1016/0277-9536(96)00118-9 https://www.nextdeal.gr/pdfs/ereunes/stat_yearbook_2016.pdf https://www.nextdeal.gr/pdfs/ereunes/stat_yearbook_2016.pdf https://doi.org/10.1016/j.archger.2013.08.010 https://doi.org/10.1016/j.archger.2013.08.010 https://doi.org/10.1016/j.archger.2013.08.010 https://doi.org/10.1016/j.archger.2013.08.010 https://www.limra.com/en/research/research-abstracts/classics/2006/u.s.-group-long-term-care-insurance-2005-annual/ https://www.limra.com/en/research/research-abstracts/classics/2006/u.s.-group-long-term-care-insurance-2005-annual/ https://www.limra.com/en/research/research-abstracts/classics/2006/u.s.-group-long-term-care-insurance-2005-annual/ https://www.mhlw.go.jp/english/policy/care-welfare/care-welfare-elderly/index.html https://www.mhlw.go.jp/english/policy/care-welfare/care-welfare-elderly/index.html https://www.mhlw.go.jp/english/policy/care-welfare/care-welfare-elderly/dl/ltcisj_e.pdf. https://doi.org/10.1023/b:hcms.0000020650.16096.23 https://doi.org/10.1023/b:hcms.0000020650.16096.23 https://doi.org/10.1023/b:hcms.0000020650.16096.23 24 g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 [10] pauly mv. the rational nonpurchase of long-term-care insurance. j polit econ 1990;98:153–68. available from: https://www. jstor.org/stable/2937646. [11] zweifel p, strüwe w. long-term care insurance and bequests as instruments for shaping intergenerational relationships. j risk uncertain 1996;12:65–76. [12] zissimopoulos j. resource transfers to the elderly: do adult children substitute financial transfers for time transfers?. santa monica, ca: rand corporation; 2001. available from: https:// www.rand.org/pubs/drafts/dru2542.html. [13] sloan fa, zhang hh, wang j. upstream intergenerational transfers. south econ j 2002;69:363–80. [14] pestieau p, sato m, 主光 佐藤. long-term care: the state, the market and the family. ssrn electronic j 2004. [15] barnett ae, stum ms. spousal decision making and long-term care insurance. j financial couns plan 2013;24:5–19. [16] coe nb, skira mm, van houtven ch. long-term care insurance: does experience matter? j health econ 2015;40:122–31. [17] guangzong m, tuan z. the development trend of my country’s population aging and its strategic response. j central china norm univ (human soc sci ed) 2011;50:29–36. [18] jie c. nursing insurance in japan and its enlightenment. market demogr anal 2002;2:69–73. [19] tao j. long-term care insurance research. beijing, china: university of international business and economics; 2005. [20] weidong d. the theory and model of long-term care insurance system. people’s forum 2011;29:31–4. [21] man z, li h. the choice of long-term care insurance system: a research review. china popul sci 2015;1:97–105+128. [22] minglai z, xinping l. the exploration of nursing insurance in china. beijing: china financial and economic publishing house; 2017, pp. 108–22. [23] amilon a, ladenburg j, siren a, østergaard sv. willingness to pay for long-term home care services: evidence from a stated preferences analysis. j econ ageing 2020;17;100238. [24] he aj, chou kl. what affects the demand for long-term care insurance? a study of middle-aged and older adults in hong kong. j appl gerontol 2020;39:413–22. [25] courbage c, roudaut n. empirical evidence on long-term care insurance purchase in france. geneva pap risk insur issues pract 2008;33:645–58. [26] brown jr, finkelstein a. the private market for long-term care insurance in the united states: a review of the evidence. j risk insur 2009;76:5–29. [27] youlu x. analysis of factors affecting the demand of long-term nursing insurance in china. finance econ 2012;11:75–8. [28] xinbang c, qiang c. analysis of the factors affecting the demand for long-term care insurance in china. china popul sci 2014;4: 102–109+128. [29] zhihong d, haiwei w. the willingness of chinese urban elderly to purchase long-term care insurance and its influencing factors. popul res 2016;40:76–86. [30] pu c. china’s insurance industry optimal growth research. beijing: china finance and economics publishing house; 2008, pp. 116–19. [31] hongye g. western economics (science micro part). 5th ed. beijing: renmin university of china press; 2000, pp. 103–4. [32] nicholson w, snyder c. microeconomic theory: basic principles and extensions. beijing: peking university press; 2008, pp. 91–113. appendix table a1 | shanghai’s total registered population and aging growth rate year total population (10,000 people) population above 60 (10,000 people) proportion above 60 (%) aging growth rate (%) 2004 1352.39 260.78 19.28 – 2005 1360.26 266.37 19.58 1.55 2006 1368.08 275.62 20.15 2.88 2007 1378.86 286.83 20.80 3.25 2008 1391.04 300.57 21.61 3.87 2009 1400.70 315.70 22.54 4.3 2010 1412.32 331.02 23.44 3.99 2011 1419.36 347.76 24.50 4.54 2012 1426.93 367.32 25.74 5.06 2013 1432.34 387.62 27.06 5.13 2014 1438.69 413.98 28.77 6.33 2015 1442.97 435.95 30.21 4.99 2016 1450.00 457.80 31.57 4.50 2017 1455.13 481.61 33.10 4.83 data source: “shanghai statistical yearbook 2004–2018” obtained after calculation, corresponding to figure 1 in the article. table a2 | population forecast year original population (10,000 people) predicted population (10,000 people) 2004 1352.39 1352.39 2005 1360.26 1373.946237 2006 1368.08 1381.723156 2007 1378.86 1389.544095 2008 1391.04 1397.409303 2009 1400.70 1405.31903 2010 1412.32 1413.273528 2011 1419.36 1421.273051 2012 1426.93 1429.317853 2013 1432.34 1437.408192 2014 1438.69 1445.544324 2015 1442.97 1453.726508 2016 1450.00 1461.955006 2017 1455.13 1470.23008 2018 * 1478.551993 2019 * 1486.92101 2020 * 1495.337398 2021 * 1503.801426 2022 * 1512.313362 (continued) https://www.jstor.org/stable/2937646 https://www.jstor.org/stable/2937646 https://doi.org/10.1007/bf00353331 https://doi.org/10.1007/bf00353331 https://doi.org/10.1007/bf00353331 https://www.rand.org/pubs/drafts/dru2542.html https://www.rand.org/pubs/drafts/dru2542.html https://doi.org/10.2307/1061677 https://doi.org/10.2307/1061677 https://doi.org/10.1016/j.jhealeco.2015.01.001 https://doi.org/10.1016/j.jhealeco.2015.01.001 https://doi.org/10.1016/j.jeoa.2020.100238 https://doi.org/10.1016/j.jeoa.2020.100238 https://doi.org/10.1016/j.jeoa.2020.100238 https://doi.org/10.1177/0733464818766598 https://doi.org/10.1177/0733464818766598 https://doi.org/10.1177/0733464818766598 https://doi.org/10.1057/gpp.2008.30 https://doi.org/10.1057/gpp.2008.30 https://doi.org/10.1057/gpp.2008.30 https://doi.org/10.1111/j.1539-6975.2009.01286.x https://doi.org/10.1111/j.1539-6975.2009.01286.x https://doi.org/10.1111/j.1539-6975.2009.01286.x g. jing et al. / journal of risk analysis and crisis response 11(1) 16–25 25 table a4 | health expenditure in shanghai year total health expenditure (billion yuan) health expenditure increase in volume (billion yuan) 2004 315.48 – 2005 362.13 46.65 2006 401.46 39.33 2007 485.67 84.21 2008 559.83 74.16 2009 656.66 96.83 2010 751.99 95.33 2011 931 179.01 2012 1092.35 161.35 2013 1248.68 156.33 2014 1347.79 99.11 2015 1536.6 188.81 2016 1838 301.4 data source: shanghai bureau of statistics, corresponding to figure 4 in the article. table a3 | distribution of disability among the elderly population age distribution total (10,000 people) health (10,000 people) not disabled (10,000 people) partially disabled (10,000 people) completely disabled (10,000 people) total 345592 150971 150631 31175 12815 60–64 years old 113829 71877 37371 3842 739 65–69 years old 66121 34007 27648 3681 785 70–74 years old 52401 19720 26785 4679 1217 75–79 years old 55363 15428 29867 7583 2485 80–84 years old 34796 6714 18462 6468 3152 85–89 years old 16979 2606 8119 3607 2647 90–94 years old 5071 538 2050 1105 1378 95–99 years old 958 76 301 201 380 above 100 years old 74 5 28 9 32 data source: shanghai bureau of statistics-“the sixth census bulletin” is compiled and calculated. table a2 | population forecast—continued year original population (10,000 people) predicted population (10,000 people) 2023 * 1520.873478 2024 * 1529.482046 2025 * 1538.139342 2026 * 1546.84564 2027 * 1555.601219 year original population above 60 (10,000 people) predicted population above 60 (10,000 people) 2004 260.78 260.78 2005 266.37 258.5896295 2006 275.62 272.2621027 2007 286.83 286.657484 2008 300.57 301.8139958 2009 315.7 317.7718816 2010 331.02 334.5735126 2011 347.76 352.2635005 2012 367.32 370.8888154 2013 387.62 390.4989111 2014 413.98 411.1458563 2015 435.95 432.8844725 2016 457.8 455.77248 2017 481.61 479.8706506 2018 * 505.2429698 2019 * 531.9568059 2020 * 560.0830893 2021 * 589.6965005 2022 * 620.875669 2023 * 653.7033813 2024 * 688.2668013 2025 * 724.6577016 2026 * 762.9727068 2027 * 803.3135509 year original proportion above 60 (%) predicted proportion above 60 (%) 2004 19.28 19.28 2005 19.58 18.82 2006 20.15 19.70 2007 20.80 20.63 2008 21.61 21.60 2009 22.54 22.61 2010 23.44 23.67 2011 24.50 24.79 2012 25.74 25.95 2013 27.06 27.17 2014 28.77 28.44 2015 30.21 29.78 2016 31.57 31.18 2017 33.10 32.64 2018 * 34.17 2019 * 35.78 2020 * 37.46 2021 * 39.21 2022 * 41.05 2023 * 42.98 2024 * 45.00 2025 * 47.11 2026 * 49.32 2027 * 51.64 note: forecast data after 2017, corresponding to figure 2 in the article. *represents data not yet available. journal of risk analysis and crisis response vol. 10(1); april (2020), pp. 1–5 doi: https://doi.org/10.2991/jracr.k.200421.002; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article review the parameters of historical strong earthquakes in tianshui and its surrounding areas wu qing* institute of geophysics, china earthquake administration, #5 minzu university south road, haidian district, beijing, china a r t i c l e i n f o article history received 3 march 2020 accepted 25 march 2020 keywords historical strong earthquake elliptical intensity distribution model magnitude macro-epicenter 关键词 历史强震 椭圆烈度分布模型 震级 宏观震中 a b s t r a c t there is a high intensity and frequency of earthquake occurrence in tianshui and its surrounding areas and it is very important to study on the historical strong earthquakes occurred in that area for knowing about the seismicity and earthquake risk prevention. this paper proposed an elliptical intensity distribution model based on modern events with both instrument records and macro investigation records that suitable for the western region of china, then estimated the historical strong earthquake parameters in tianshui and its surrounding areas. *email: wuqing908@sina.com 天水及周边地区历史强震参数复核 吴清 中国地震局地球物理研究所北京市海淀区民族大学南路5号北京，100081，中国 摘要 天水盆地及其周边地区是地震发生强度和频度均极高的地区，研究该地区发生过的历史强震对于掌握天水及周边地 区地震活动性规律和地震风险防控非常重要。本文通过既有仪器测定记录又有宏观考察数据的现代大震资料建立适 用于中国西部地区的椭圆烈度分布模型，进而对天水及周边地区历史强震参数进行估算复核。 © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). 1. 引言 天水盆地位于南北活动构造带中段北端，西秦岭断裂带和祁 连山-六盘山断裂带的交汇处附近，也是青藏高原东北缘与黄 土高原接壤地带，是地震发生强度和频度均极高的地区。有 史料记载以来，天水盆地及周边附近地区发生过734年天水西 7级地震、1654年天水南8级地震、1718年通渭7½级地震等7级 以上大震，7级以下中强地震也发生频繁。由于地震仪器记录 的历史较短，不足以在短时间、小尺度范围内弄清楚天水及 周边地区地震活动规律，因而对历史地震的研究非常重要。 尤其是6.5级以上的历史强震，其参数的不确定性，将直接对 地震活动性规律和地震安全性评价等问题产生重大影响，对 地震风险防控非常重要。 吴清等[1]根据6.5级以上强震烈度分布特征，提出了基于烈 度数据点的考虑断层破裂长度的烈度椭圆分布模型，在此基 础上联立椭圆数学方程确定了强震参数估计方法，并采用蒙 特卡洛方法定量分析了所得参数的不确定性。该方法计算得 到的震级精度在1级以内，震中计算精度可达2类。本文采用 此方法，建立了适用于中国西部地区震级大于等于6.5级地震 的烈度椭圆分布模型，并对天水及周边地区的5个历史大震基 本参数进行了估算。本文模型基于既有仪器测定记录又有宏 观考察数据的现代大震建立，对天水及周边地区历史强震的 重建和参数的校核具有重要意义。 2. 中国西部地区强震烈度椭圆分布模型 2.1. 强震烈度估计等值线拟合 我国目前公开发表的烈度调查资料大多采用等震线的形式发 布，原始烈度调查点通常难以获得。本文选取已经出版的《 中国震例（1966-2002）》[2–10]，2008年后的强震烈度资料 mailto:wuqing908%40sina.com?subject= http://creativecommons.org/licenses/by-nc/4.0/ 2 w. qing / journal of risk analysis and crisis response ( ) 从中国地震局官方网站的地震专题上获取(http://www.cea.gov. cn/publish/dizhenj/468/553/index.html),收集了20世纪60年代 以来中国西部地区所发生的震级大于等于6.5级地震的等震线 图，提取了其中既有仪器测量数据又有宏观烈度调查数据的 现代地震资料(表1)，并将烈度分布图数字化。这些地震的等 震线图上部分明确标示了宏观烈度调查点位置；对于未明示 烈度调查点的，本文将能在等震线图上明确获取烈度信息的 城、县、镇、村所在地作为烈度点。经数字化配准后获取各 地震烈度点的空间分布。 考虑中心点和方向性的椭圆参数方程为： x r t r t x y r t r t i a i b i i a i b i = + = + × × × × × × × cos cos sin sin cos sin sin q q q 0 ×× + ì í ï îï cosq y0 � �(1) 式中，（xi, yi）为椭圆上点的坐标，(x0, y0)为椭圆中心坐 标；q(0° £ q £ 180°)是椭圆长轴逆时针与x轴正方向的夹 表1 | 地震数据及各烈度区椭圆烈度估计等值线长、短半轴长度和椭圆中心 序号 发震时间(年-月-日) 宏观震中 烈度区 烈度估计等值线 椭圆中心直角坐标 仪器震级 (ms) 参考地点 log(°e) lat(°n) 长半轴ra (km) 短半轴rb (km) x0 y0 1 1970-01-05 102.683 24.017 ⅹ 24.45 8.25 0 0 7.7 云南通海 ⅸ 28.68 14.75 0 0 ⅷ 36.28 24.55 0 0 ⅶ 55.43 40.99 −3.90 −2.27 2 1973-02-06 100.533 31.483 ⅹ 21.95 4.3 0 0 7.6 四川炉霍 ⅸ 27.5 6.2 0 0 ⅷ 41.62 11.1 0 0 ⅶ 54.82 22.15 −4.97 0.62 ⅵ 84.57 45.75 −2.41 8.0 3 1996-02-03 100.267 27.083 ⅸ 25.6 14.1 0 0 7.0 云南丽江 ⅷ 49.2 18.2 0 0 ⅶ 62.0 35.0 0 0 ⅵ 88.2 54.9 0 0 4 2000-01-15 101.1 25.6 ⅷ 24.52 6.75 0 0 6.5 云南姚安ⅶ 33.42 14.57 0 0 ⅵ 60.46 30.27 0 0 5 2000-09-12 99.53 35.4 ⅷ 7.45 7.0 −1.11 1.28 6.6 青海兴海 ⅶ 16.85 15.49 2.02 7.10 ⅵ 33.92 28.59 0.24 6.15 ⅴ 54.65 51.75 3.17 5.45 6 2008-05-12 103.4 31 ⅺ 33.6 19.87 0 0 8.0 四川汶川 ⅺ 34.5 16.83 0 0 ⅹ 122.6 35.54 78.7 71.5 ⅸ 158.75 38.71 100.8 85.7 ⅷ 200.69 67.07 7.76 −108.20 ⅶ 220.94 77.26 65.57 82.53 ⅵ 392.81 151.40 100.63 84.86 7 2008-10-06 90.3 29.8 ⅷ 23.06 6.74 0 0 6.6 西藏当雄ⅶ 23.26 20.12 0 0 ⅵ 54.61 54.31 0 0 8 2010-04-14 96.6 33.2 ⅸ 18.12 0.83 26.66 −9.00 7.1 青海玉树 ⅷ 40.79 8.53 5.26 0.42 ⅶ 60.68 19.90 8.05 0.33 ⅵ 101.34 52.49 18.20 −16.64 9 2013-04-20 103.0 30.3 ⅸ 11.45 3.57 −0.59 −0.28 7.0 四川芦山 ⅷ 27.30 12.34 −0.75 −11.23 ⅶ 45.53 21.28 −4.22 −22.09 ⅵ 68.67 61.72 −14.36 −36.84 10 2013-07-22 104.2 34.5 ⅷ 18.52 7.45 −6.71 7.23 6.6 甘肃岷县ⅶ 37.83 22.42 −6.60 5.75 ⅵ 72.84 48.47 0.26 1.61 11 2014-08-03 103.3 27.1 ⅸ 6.75 4.68 10.53 −1.94 6.5 云南鲁甸 ⅷ 11.08 10.48 7.52 −2.66 ⅶ 21.15 17.38 5.54 −0.21 ⅵ 54.54 38.23 −4.26 −2.22 12 2014-10-07 100.5 23.4 ⅷ 9.84 8.53 0 0 6.6 云南景谷ⅶ 19.74 18.51 0 0 ⅵ 49.93 44.22 0 0 13 2015-07-03 78.2 37.6 ⅷ 21.65 12.02 −8.39 0.03 6.5 新疆皮山ⅶ 40.70 29.41 0 0 ⅵ 76.24 51.93 0 0 w. qing / journal of risk analysis and crisis response ( ) 3 角；ra、rb分别为椭圆长半轴和短半轴长度，ti(0 £ q £ 2p ) 为椭圆上各点所对应的参数。 吴清等[1]根据我国现代大震资料统计拟合了地震破裂长度震级关系式 lgl m= − + ⋅1 9227 0 4691. . � �(2) s = 0 5099. 式中，m为震级，l为地表破裂长度。据此可以估计相应震级 的地震可能伴随的地表破裂长度l。以式（2）约束椭圆强震 烈度分布模型，要求最内圈烈度估计等值线椭圆长轴要大于 由公式（2）估计的地表破裂长度即 2r la > � (3) 对于既有仪器测量数据又有宏观调查烈度数据的现代地震， 可获得地震的宏观震中位置(x0,y0)、极震区走向q和烈度点坐 标（xi,yi)。根据各烈度区调查点的空间分布(xi, yi)，结合 (1）式和（3）式利用最小二乘原理采用通用全局优化算法拟 合出各烈度区椭圆估计线长半轴和短半轴ra和rb，由此获得 各烈度区的椭圆烈度估计等值线。由于地震的宏观震中并不 一定都是该地震各烈度区的几何中心，因此对某些地震需要 同时拟合出各烈度区的几何中心(x0,y0)。这里烈度估计线长 轴方向均取极震区走向q。 以各自宏观震中坐标为原点(0,0)，将各地震对应烈度点都换 入平面直角坐标系。表1给出了拟合所得的各地震各烈度区椭 圆烈度估计等值线的长半轴和短半轴长度，拟合烈度估计等 值线几何中心不同于宏观震中的也同样列在了表1中。表1中 拟合烈度估计等值线的椭圆几何中心与宏观震中重合的，坐 标即为(0,0)；几何中心与宏观震中不重合的，具体列出。 图1是� 2013年7月22日甘肃岷县6.6级地震烈度估计等值线拟 合结果，以宏观震中为平面坐标原点。图中小三角是烈度数 据点，椭圆长轴方向为极震区走向，烈度估计等值线是对原 始烈度点空间分布进行最小二乘拟合所得，可以看到烈度估 计等值线趋向于原始烈度点空间分布的平均估计。 2.2. 中国西部地区强震烈度分布模型 本文据陈达生等[11]关于地震烈度椭圆衰减关系长短轴统一 回归的思想，采用如下烈度分布模型： i a b m c r r c r ra a b b= + ⋅ + + + + +1 0 2 0log log( ) ( ) e � �(4) 式中，i为地震烈度，m为震级；系数a�、b�、c1、c2均为回归 常数；r0a、r0b分别为椭圆长、短轴两方向烈度近场饱和因 子；e� 为回归分析中表示不确定性的随机变量，通常假定为 对数正态分布，其均值为0，标准差为s。该模型在震中处的 烈度值随震级的变化率为常数b，而中间距离仍保持长短轴烈 度的差别，同时在远场也使烈度分布成圆形。需要强调指出 的是，这里ra、rb分别是烈度i的烈度估计等值线长短半轴长 度，不再是等震线的长、短半轴长度。 将表1中各地震各烈度区的烈度估计等值线，按照公式（4） 进行最小二乘统计回归。由于拟合的烈度估计等值线已经是 对烈度点分布的综合估计，所以没有采取近场补点。但为了 体现远场区发震构造影响消失，烈度分布趋于圆形的特点， 汪素云等[12]提出取有感范围的半径作为远场控制点，有 感烈度值通常为ⅲ-ⅳ度，在计算中取为3.5度，称为远场补 点。这里提出的有感半径有外包线的性质，而本文烈度模型 主要是对烈度分布的综合估计，因此计算时远场控制烈度取 为3度。有感半径与震级的关系见表2。 由原始烈度点拟合得到烈度区的估计等值线，进而统计回归得 到地震烈度分布模型，其中ia为长轴方向，ib为短轴方向： i m ra a= + +4 1428 1 821 5 1339 25. . . ( )lg � �(5) i m rb b= + +0 4550 1 821 3 8636 8. . . ( )lg � �(6) s = 0 692. 3. 强震参数估计方法及验算 3.1. 强震参数估计方法 吴清等[1]给出了强震参数估计方法： ( ) cos ( ) sin( ) ( )x x y y r i i c c m i c a a i a × + ×éë ùû æ è ç + × + 0 0 010 1 2 3 q q e öö ø ÷ + × ×éë ùû + × + 2 2 0 0 10 1 2 3 ( ) sin ( ) cos( ) ( )x x y yi i c c m i c b i b q q e -æ è ç ö ø ÷ = r b0 2 2 1 � �(7) 图1 | 2013年7月22日甘肃岷县6.6级地震烈度估计等值线分布图. 表2 | 有感半径与震级的关系[12] r(km) 340 390 450 520 600 690 800 900 1100 m 6 1 2 6 3 4 7 7 1 4 7 1 2 7 3 4 8 8 1 4 8 1 2 4 w. qing / journal of risk analysis and crisis response ( ) 表3 | 内符验算计算结果 发震时 （年-月-日） 宏观震中 仪器震级 (ms) 计算震级 (ms) 计算震中平面坐标 δm δr (km) 参考地点 log(°e) lat(°n) x0 y0 1996-02-03 100.27 27.08 7.0 7.4 2.45 −3.35 0.4 4.15 云南丽江 2000-01-15 101.1 25.6 6.5 6.6 7.91 3.47 0.1 8.63 云南姚安 2000-09-12 99.53 35.4 6.6 6.2 0.02 7.75 −0.4 7.75 青海兴海 2010-04-14 96.85 33.06 7.1 7.0 1.1 −1.27 −0.1 1.68 青海玉树 2013-04-20 103.0 30.3 7.0 7.0 −2.51 −4.71 0 5.33 四川芦山 2013-07-22 104.2 34.5 6.6 6.7 −5.16 6.32 0.1 8.15 甘肃岷县 2014-08-03 103.3 27.1 6.5 6.8 8.2 −2.9 0.3 8.69 云南鲁甸 2014-10-07 100.5 23.4 6.6 6.6 2.16 −2.92 0 3.63 云南普洱 注：δm: 计算震级-仪器震级；δr: 计算震中到宏观震中平面距离。 式中，c1a、c1b、c2、c3a、c3b、r0a、r0b由烈度分布模型统计 回归可得，见式（5）（6）。(xi,yi,ii)为烈度数据点，(x0, y0) 为震中，m为震级，q为方向。若已知一个地震的多个烈度 信息点(xi,yi,ii)，带入到（7）式，即可联立方程组求得震中 (x0,y0)、震级m和方向q。有4个未知数，则至少需要4个方程 确定一组解，但x0与y0不是相互独立的，因此在已知烈度分布 模型下，3个烈度信息点就能确定一组震中、震级和方向。烈 度信息点增多，变成求解超定非线性方程组，采用通用全局 优化法的数值计算方法寻求最优解，可非常直观的求得地震 宏观震中位置、震级和方向。 参数估计方程是以椭圆数学方程为基础，因此用于计算的烈度 点不能分布于一条或接近于一条直线上，否则在极端情况下将 造成方程无解，或所得数值解不满足地震参数的物理意义。 由于在建模过程中，8.0级以上特大地震的资料太少，因此本 方法只适用于6.5～8.0级的历史强震，对于8.0级以上的特大 地震需要另行讨论。 中国大陆的地震绝大部分发生在地壳以内，其震源深度差别 不大[13]，因此本文暂不考虑震源深度对烈度分布的影响。 3.2. 震例验算 3.2.1. 內符检验 带入由1.2节得到的强震烈度分布模型系数，公式（7）即变 为： ( ) cos( ) ( ) sin( ) . . . x x y yi i m ii × + ×éë ùû + × 0 0 2 4 1428 1 821 5 13310 q q 99 2 0 0 2 0 4550 1 821 25 10 -( ) + × ×éë ùû + × ( ) sin( ) ( ) cos( ) . . x x y yi iq q mm ii-( ) = 3 8636 2 8 1 . � （8） 为了验证此算法的可行性，以参与强震烈度分布模型拟合的 其中8个地震为例，将各自的烈度数据点（xi，yi，ii）直接带 入到方程（8）中进行试算，计算时所有烈度点以各自地震的 宏观震中为原点（0,0）转换到平面坐标下，计算结果见表3。 由表3可以看到，参与验算的8个地震，计算震级与仪器震 级的差值不超过0.5级，计算震中到宏观震中的距离不超过 10km，由此可见此方法对参与模型拟合的地震集是可行的。 3.2.2. 外推检验 为了验证地震参数估计方程的外推有效性，我们提取了3个没 有参与模型拟合的强震烈度数据点，将所有烈度数据点以各自 地震的宏观震中为原点（0,0）转换到平面坐标下（xi, yi, ii)， 然后直接带入到方程（8）中进行验算，计算结果见表4。� 由表4同样可以看到，参与验算的3个地震，计算震级与仪器 震级的差值不超过0.5级，计算震中到宏观震中的距离不超过 15km，由此可见此方法外推也是可行的。 与吴清等[1]得到的适用于全国的烈度分布模型相比，由适用 于西部地区的烈度分布模型估算得来的地震参数较为准确。 4. 天水及周边地区历史大震参数估计 通过前面的分析，本文建立了一套适用于中国西部地区的烈 度分布模型，以此来估算天水及周边地区几个7.0级以上历史 大震，结果见表5。历史大震的烈度数据点均从《中国历史地 震图集》（明、清时期）[14–15]中提取。由表5可以看到， 计算震级与图集震级相差不超过0.5级，计算震中与图集震中 相差不超过30km。由此可见，《中国历史地震图集》里给出 的强震参数还是比较可靠的。 5. 结论与讨论 本文建立了适用于中国西部地区的椭圆烈度分布模型，并对 天水及周边地区5个历史强震进行了地震参数估算，结果表明� 《中国历史地震图集》里给出的强震参数还是比较可靠的。 此方法对历史强震参数的估定颇为有效，直接利用烈度数据 点，跳过对等震线的勾画，处理过程直接明了，一定程度上 减少了主观不确定性，提高了科学性。 本文旨在通过既有仪器测定记录又有宏观考察数据的现代大 震建立参数估计模型，但现代大震资料毕竟有限。模型是原 型的抽象和简化表征，与任何一个真实地震之间都存在偏 差。随着技术的发展、科学的进步以及基础数据的积累，需 要对模型不断的改进以期更接近真实情况。 w. qing / journal of risk analysis and crisis response ( ) 5 表4 | 外推验算计算结果 发震时间 （年-月-日） 宏观震中 仪器震级 (ms) 计算震级 (ms) 计算震中平面坐标 δm δr (km) 参考地点 log(°e) lat(°n) x0 y0 1974-05-11 103.92 28.2 7.1 6.8 −0.44 −0.61 −0.3 0.75 云南大关 1995-10-24 100.25 25.88 6.5 7.0 −3.55 −1.71 0.5 3.94 云南武定 1996-03-19 101.15 25.95 6.9 7.0 −7.99 −8.49 0.1 11.65 新疆阿图什 注：δm: 计算震级-仪器震级；δr: 计算震中到宏观震中平面距离。 表5 | 天水及周边地区历史大震参数估算结果 发震时间 （年-月-日） 图集宏观震中 图集震级 (ms) 计算震级 (ms) 计算震中平面坐标 δm δr (km) 参考地点 log(°e) lat(°n) x0 y0 1352-04-26 105.3 35.6 7.0 7.2 −24.6 −14.2 0.2 28.4 甘肃会宁东南 1654-07-21 105.5 34.3 8.0 8.2 13.5 −2.41 0.2 19.8 甘肃天水南 1718-06-19 105.2 35.0 7.5 8.0 −5.57 −5.56 0.5 7.87 甘肃通渭南 1879-07-01 104.7 33.2 8.0 8.2 8.58 −21.82 0.2 23.4 甘肃武都 1927-05-23 102.7 37.5 8.0 8.2 2.16 −2.69 0.2 3.45 甘肃古浪 注：δm: 计算震级-图集震级；δr: 计算震中到图集宏观震中平面距离。 致谢 本文由科技部基础资源调查专项（2018fy100504）资助。 参考文献 [1]�吴清，高孟潭，基于烈度数据点的历史强震参数估计研究.� 中国地震�2016;32:11–27. [2]�张肇诚，罗兰格，李海华等，中国震例（1966–1975）.� 北 京：地震出版社;�1988. [3]�张肇诚，罗兰格，李海华等，中国震例（1976–1980）.� 北 京：地震出版社;�1990a. [4]�张肇诚，罗兰格，李海华等，中国震例（1981–1985）.� 北 京：地震出版社;�1990b. [5]�张肇诚，郑大林，罗平等，中国震例（1986–1988）.� 北 京：地震出版社;�1999. [6]�张肇诚，郑大林，徐京华等，中国震例（1989–1991）.� 北 京：地震出版社;�2000. [7]�陈棋福，郑大林，车时，中国震例（1992–1994）.� 北 京：地震出版社;�2002a. [8]�陈棋福，郑大林，刘桂萍等，中国震例（1995–1996）.� 北 京：地震出版社;�2002b. [9]�陈棋福，郑大林，高荣胜，中国震例（1997–1999）.� 北 京：地震出版社;�2003. [10]�陈棋福，郑大林，车时等，中国震例（2000–2002）.� 北 京：地震出版社;�2008. [11]�陈达生，刘汉兴，地震烈度椭圆衰减关系.� 华北地震科学,� 1989;7:31–42. [12]�汪素云，武宦英，华北地区地震烈度的衰减关系，中国地 震区划文集.�北京：地震出版社;�1993. p. 185–91. [13]�刘百篪，郑文俊，郭华等，活断层工作方法在中早期历史 地震研究中的作用.�中国地震�2002;18:283–8. [14]�国家地震局地球物理研究所，复旦大学中国历史地理研 究所，� 中国历史地震图集（明时期）.� 北京：地图出版 社，1989. [15]�国家地震局地球物理研究所，复旦大学中国历史地理研究 所，� 中国历史地震图集（清时期）.� 北京：中国地图出 版社，1990. available from: http://www.cea.gov.cn/publish/ dizhenj/468/553/index.html http://www.cea.gov.cn/publish/dizhenj/468/553/index.html http://www.cea.gov.cn/publish/dizhenj/468/553/index.html microsoft word a new approach for contingency determination in a portfolio of construction projects.doc a new approach for contingency determination in a portfolio of construction projects payam bakhshi* assistant professor, department of construction management, wentworth institute of technology 550 huntington ave., anxso 008-f, boston, ma 02115, usa *e-mail: bakhship@wit.edu ali touran professor, department of civil & environmental engineering, northeastern university 360 huntington ave., 400 sn eng., boston, ma 02115, usa atouran@coe.neu.edu abstract in this paper, a new approach for contingency determination in a portfolio of construction projects is proposed. the model proposed helps an agency find the level of confidence needed for individual projects to ensure that the portfolio budget will meet the minimum level of confidence based on available funding and the agency’s policy goals. the promise of this model is to protect a portfolio of projects against cost overrun by adjusting their original budgets. a bayesian approach is employed to update the model on regular intervals. as more information becomes available in the future, the required adjustment in portfolio budget will be reduced, because the accuracy of estimating the contingency is improved. the proposed model is an effective tool for the agencies/owners to develop contingency budgets. keywords: contingency, risk, probabilistic model, portfolio of projects, bayesian. 1. introduction large transportation capital projects all around the globe have been experiencing cost and schedule overruns. nearly 50% of the large active transportation projects in the united states overran their initial budgets (sinnette 2004). to overcome the cost overrun issue, identifying cost escalation factors have been the subject of much research (shane et al 2009; flyvbjerg et al 2003; pickrell 1990). for instance, shane et al (2009) identified 14 risk factors classified in two categories: 1. internal sources such as bias, poor estimating, and contract document conflicts; 2. external sources such as effects of inflation, market conditions, and unforeseen events/ conditions. contingency is a reserve budget for coping with risks and uncertainties and to help keep the projects on budget. an owner agency usually adds contingency to the estimated project cost to account for the uncertainties. risks and uncertainties associated with a project are impediments to reach an accurate cost estimate. contingency is traditionally estimated as a predetermined percentage of project base cost depending on the project phase. in recent years, some agencies have started conducting formal probabilistic risk assessment to estimate contingency budget rather than deterministic approach (touran 2010; molenaar 2005). however, to establish the contingency budget, an agency must make all effort to set aside a budget which is optimized. this becomes more important when an agency is dealing with a portfolio of projects. allocation of an excess budget for a project will use up the money that can be spent on other projects. for instance the current practice in the u.s. to estimate the contingency journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 223-232 published by atlantis press copyright: the authors 223 administrateur texte tapé à la machine received 5 march 2012; accepted 4 august 2012 budget in transit projects called top-down model is based upon a probabilistic method using lognormal distributions for different cost categories in the project. research shows the way that cost categories are ranged is very conservative resulting in a contingency budget far larger than what is indeed needed (bakhshi and touran 2009). nevertheless, despite all claims regarding improved models, budget estimating for transit projects have been inaccurate for several decades (flyvbjerg 2006). also, for large capital programs consisting of several projects, establishing contingency has not been well studied. 2. proposed model for calculating contingency this model is a continuation and major improvement on an earlier model developed by touran (2010) for calculating contingency for a portfolio of projects. the new model uses a bayesian approach for updating the calculations based on new data that becomes available. the application of the model is shown on a group of transit projects. even though the application of this model in this paper is on transit projects, it is a mathematically flexible model that can be applied on any type of construction project. in the model proposed here, the portfolio consists of projects with different owners who have requested funding from the same source. for example, in the case of transit in the united states, these are projects submitted by state agencies for obtaining federal funding. for each of these projects, it is assumed that a formal risk assessment has been conducted based on the specific risks affecting each project as required by the regulations. the objective of the model presented here is then to adjust the overall portfolio budget based on the historical data on budget shortfalls. in other words, it is assumed that a detailed risk assessment has been conducted at the individual project level and that the requested funding reflects that. the model assumes normal distribution for the cost overruns/ underruns and truncated normal distribution for the cost of each project in the portfolio. these assumptions are based on the following factors: first, the cost overrun/underrun distribution will be used as a prior distribution in the bayesian approach. as more information becomes available, the distribution becomes more refined and converges to the true distribution regardless of the initial assumption about prior, and second, the use of normal distribution allows the derivation of closed form solution for the calculation of contingency based on desired confidence levels. furthermore, tests of goodness of fit showed that assumption of normality was adequate for the project cost data that was available. to form truncated normal distribution of cost for each project, it is assumed that the probability of experiencing underrun m is α as the discrete portion of distribution. the parameter m is added because project owners have the tendency to spend most of the budget by enhancing and embellishing of the projects when they realize their projects will be completed under the budget. this parameter equips the model to consider the fact that the project may be completed under the initial budget by a certain percent. m is an arbitrary number based upon agency’s objectives and α can be determined by reviewing the historical cost overruns/underruns. an agency may decide to input m equal to zero and find the α corresponding to that. fig. 1 illustrates the truncated normal x~ ),( σμ ′′n , where the normal component of the distribution is ),( σμn . fig. 1: truncated normal distribution we presume that there is a database of construction projects comprising of ni ...,,1= projects each with the initial budget of ib . it is found through this historical data that there is α % chance to have m percent underrun and get the project done with ii bmc ).1( −= . the model is constructed using the following parameters: =ib initial budget allocated for project i ; =ix actual cost of project i ; =m the maximum expected underrun; =ic minimum expected project i cost which is ibm).1( − ; =α percent of projects in the historical data having underrun more than or equal to m ; published by atlantis press copyright: the authors 224 =δ cost overrun/ underrun; =δ average of cost overruns/underruns in the historical data; =σ standard deviation of cost overruns/ underruns in the historical data; =β average rate of cost overrun/ underrun relative to b which is δ+1 ; =ρ average rate of cost overrun/ underrun relative to c which is )1/( m−β ; =iμ mean of underlying normal distribution in project i ; =iσ standard deviation of underlying normal distribution in project i ; =′iμ mean of hybrid normal distribution in project ;i =′iσ standard deviation of hybrid normal distribution in project i ; =ϕ a constant coefficient which is equal to ii c/σ′ ; =*ib revised budget of project i ; =b sum of all individual initial budgets, ∑ ib ; =η percent of confidence that individual projects’ cost will not be more than *b ; =*b sum of all revised individual budgets based on η , ∑ *ib ; =γ probability that portfolio of projects’ cost will not be more than *b ; project cost ix is defined as follows: ⎪ ⎩ ⎪ ⎨ ⎧ −=> == =< α α 1)( )( 0)( ii ii ii cxp cxp cxp (1) the probability distribution function (pdf) of project cost which is a truncated normal is: ⎪ ⎪ ⎩ ⎪⎪ ⎨ ⎧ >= ≤= −− ii x i ii cxforexf cxforxf i ii 2 2 2 )( . 2 1 )( )( σ μ σπ α (2) the mean μ′ and standard deviation σ ′ of truncated normal can be calculated using the following equations: 2 )]([ 2 )( 21 2 2 . 2 ).1(. . 2 1 .)( α σ μ π σ μαα σπ αμ −φ− ∞ −− +−+ =×+==′ ∫ ec dxexcxe i ii c i x i iiii i i ii (3) )).(1( ].2)(..[ 2 . . 2 1 .)( 22 2 )]([ 122 22 )( 22222 21 2 2 ii i i ii i b i x i iiiii ec dxe xcxe i i i ii μσα μασ π σ μα μ σπ αμσ α σ μ +−+ +φ+′− =′− ×+=′−=′ −φ− − ∞ −− ∫ (4) where: i iic σ μ α − =φ − )(1 (5) )(1 α−φ is the inverse of cumulative function for standard normal distribution. β is the average rate of cost overruns/underruns and can be calculated from the historical data as: ∑ = − +=+= n i i ii b bx n 1 . 1 11 δβ (6) also, we know iμ′ is the expected value of the final cost of project i , so we can model it as a multiplier of its budget: iii cb .. ρβμ ==′ (7) and knowing that ii bmc ).1( −= , we have: mc b cb i i ii − =⇒=→= 1 . .. β ρ β ρρβ (8) by rearranging eq. (5), mean of the underlying normal distribution is calculated. also, by substituting eq. (7) and (9) in eq. (3) and rearranging, the standard deviation of underlying distribution is found: )(. 1 ασμ −φ−= iii c (9) published by atlantis press copyright: the authors 225 π αα ρ σ α 2 )().1( ).1).(1( 2 )]([ 1 21−φ− − −φ− −− = e bm i i (10) reviewing eq. (4) shows that all terms of 2iσ′ are comprised of a constant coefficient times 2ic . therefore, 2iσ ′ can be written in the form of: ii or ii cc .. 222 ϕσϕσ =′⎯→⎯=′ (11) where ϕ is a constant coefficient for all values of iσ ′ and ic that can be computed using eqs. (4) and (11). referring to fig. 1, if a budget ii bb > * for each project is selected, the chance of shortfall of budget would be limited to η . so: η α σ μ α σ μ σ μ α = − − φ+= − φ− − φ+= =≤ ) * ( )() * ( *)(*)( i ii i ii i ii iii b cb bfbxp (12) if we rearrange eq. (12), we obtain: )(.* 1 ησμ −φ+= iiib (13) we know that the original portfolio budget is: bmc c mm c bb i then i i i ).1( )1( 1 )1( −=⎯⎯ →⎯ − = − == ∑ ∑ ∑∑ (14) using eq. (13), (14) and substituting iμ using eq. (9), the new portfolio budget can be computed as follows: ∑ ∑∑ ∑ ∑ −− −− − φ−φ+−= φ+φ−= φ+= = i iii ii i bm c bb σαη σηασ ησμ )].()([).1( ]).([)](.[ )](.[ ** 11 11 1 (15) substituting eq. (10) in eq. (15), the ratio of bb /* is found as follows: ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎣ ⎡ −φ− φ−φ− +−= −φ− − −− π αα αηρ α 2 )().1( )]()().[1( 1).1( * 2 )]([ 1 11 21 e m b b (16) eq. (16) gives the required portfolio budget increase with respect to η . we now assume that project costs are statistically independent and the total actual cost of all projects in the portfolio is t . the assumption of independence is reasonable because these projects are scattered throughout the country and the owners and management structure are different, as these are various state agencies using the federal funds. however, it should be noted that pairwise correlation between costs of any two concurrent projects may exist when there is a belief that they are using common resources, common management, or being affected by other common factors such as statutory/ regulatory constraints, political conditions, or unemployment. based on central limit theorem, t will follow an approximate normal distribution with the mean tμ and the standard deviation tσ . therefore: ),(~ tt clt i ntxt σμ∑ ⎯⎯→⎯= (17) ∑ ∑ −==′= bmciit ).1.().( ρρμμ (18) ∑ ∑ ∑∑ =⎯⎯ →⎯ ==′= 2 22222 . .).( it then iiit c cc ϕσ ϕϕσσ (19) defining γ as the percent of confidence that portfolio of projects cost will not be more than *b , we have: γ σ μ = − φ=< ) * (*)( t tbbtp (20) rearranging eq. (20) and using eq. (18) and (19), we obtain: published by atlantis press copyright: the authors 226 )(..).1.(* * )( 12 1 γϕρ σ μ γ − − φ+−=⎯⎯ →⎯ − =φ ∑ ithen t t cbmb b (21) by equating eq. (15) and (21) and substituting iσ from eq. (10), γ can be found as follows: ]} 2 )().1( )]()([ 1[ . ).1).(1( { 21 )]([ 1 11 2 π αα αη ϕ ρ γ α−φ− − −− −φ− φ−φ + × −− φ= ∑ e c bm i (22) rearranging eq. (22) gives: ]}1 ).1).(1( )(.. [ ] 2 )().1[( )({ 12 )]([ 1 1 21 − −− φ ×−φ− +φφ= − φ− − − ∑ − bm c e i ρ γϕ π αα αη α (23) eqs. (16) and (23) are used to calculate the require percent increase in portfolio budget and probability of overrun for each individual project in the portfolio based on probability of having sufficient budget for the portfolio of projects. in the next section we will see how these values are updated when new completed projects become available. 3. fundamentals of bayesian approach an agency employing the proposed model is expecting to experience less or even no cost overrun in the newly funded projects. since the model has been constructed based on limited observed data, it is conceivable that the overrun will not be eliminated in the first attempt. therefore the model needs to be updated on a yearly or bi-yearly basis, depending on the number of completed projects. to this end, a bayesian approach is utilized to update the model as the information regarding the costs of new projects become available. when the observed data are limited and making decision on the available information is required, the bayesian approach can help update the system as more data is acquired. the fundamentals are based on eq. (24) (ang and tang 2006): )().(.)( δδδ flkf ′=′′ (24) where: 1 1 )().()....,,( −∞ ∞− ⎥⎦ ⎤ ⎢⎣ ⎡ ′== ∫ δδδδδδ dfpk kj is the normalizing constant; =)(δl the likelihood of observing the new cost overruns/ underruns assuming a given δ , mean of the distribution; =′ )(δf prior distribution of δ ; and =′′ )(δf posterior distribution of δ . eq. (24) can be used to update the proposed model in light of new information acquired through newly completed projects during a certain period of time. δ is the average of cost overruns/underruns required to calculate the parameters β and ρ of the model. 4. bayesian approach for k completed projects let’s assume that k new projects are recently completed. further, assume that these k new projects with the cost overrun/underrun of jδ are statistically independent of each other. the probability of observing kδδ ...,,1 coming from a population δ having an underlying normal distribution ),( σδδn is: )(.)()...,,( 1 1 δδδδδδδ dnp j k j kj ∏ = δ== (26) therefore, the likelihood function can be written as: ∏ ∏ = = δ ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − −= = k j j k j jnl 1 2 1 . 2 1 exp .2 1 ),()( σ δδ σπ σδδδ (27) it should be noted that eq. (27), the joint likelihood of the sample, is the product of k individual observed normal likelihoods. it is known that the product of k normal likelihoods has the shape of a normal distribution as follows (bolstad 2007): ( )llk n kk nl σδ σδδ δ ,, ... )( 1 =⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ ++ ∝ (28) the posterior distribution is now the product of likelihood )(δl and prior )(δf ′ . if the prior is a published by atlantis press copyright: the authors 227 normal pdf such as ),(~)( σδδ ′′′ nf , then the posterior has also a normal pdf with the mean and standard deviation as follows (ang and tang 2006): ),(~)().(.)( σδδδδ ′′′′′=′′ nflkf (29) ⎪ ⎪ ⎩ ⎪ ⎪ ⎨ ⎧ +′ ′ =′′ +′ ′+′ =′′ 22 22 22 )()( . )()( ).().( l l l ll σσ σσ σ σσ σδσδ δ (30) by finding the updated distribution of cost overruns/ underruns, the main model is revised to calculate new required percent increase in portfolio budget bb* . this is done by substituting βα , , and ρ parameters of the model with the following values: ⎪ ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ ⎨ ⎧ − =⇒′′+= ′′ ′′−− φ= ′′ ′′−− < =−<= m mm zp mxp new newnew new 1 1 )()( )( β ρδβ σ δ σ δ α (31) 5. numerical example 5.1. main model to show the application of the model, a set of 28 transit projects (booz allen hamilton 2005) is selected (table 1). these projects have been funded by federal transit administration (fta) of the u.s. department of transportation in the past 20 years. cost overrun/underrun of each project is defined to be the percent of difference between actual final cost (as-built cost) and estimated cost at the end of final design (fd), when in transit projects full funding grant agreement (ffga) is established by the fta. reviewing table 1 shows that 22 out of 28 projects have been completed before 2004, five projects in 2004 and one project in 2005. we set aside the 22 projects as historical data to find the primary values for α and β used in the model. then the model is applied to the set of five projects completed in 2004 to see the effect of model on cost overruns/underruns. the project completed in 2005 is not considered in the application process of the model as it would fall in another fiscal year. from this point on, for consistency and ease of referencing, we call the set of 22 projects “historical dataset”, the set of five projects “first dataset”. table 1: cost overrun/underrun of 28 transit projects (booz allen hamilton 2005) proj. id project name year completed cost at the ffga (in m$) actual cost (in m$) cost overrun/ underrun 1 atlanta north line extension 1999 $381.3 $472.7 23.97% 2 boston old colony rehabilitation 1997 $551.9 $565.0 2.37% 3 boston silver line (phase 1) 2004 $413.4 $604.4 46.20% 4 chicago southwest extension 1989 $350.9 $474.6 35.25% 5 dallas south oak cliff extension 2002 $517.2 $437.2 -15.47% 6 denver southwest line 1999 $176.3 $177.1 0.45% 7 los angeles red line mos 1 1991 $960.3 $1,490.1 55.17% 8 los angeles red line mos 2 1994 $1,524.6 $1,921.6 26.04% 9 los angeles red line mos 3 1995 $1,345.6 $1,227.6 -8.77% 10 minneapolis hiawatha line 2004 $675.4 $715.3 5.91% 11 new jersey hudson-bergen mos1 2000-2002 $992.1 $1,113.0 12.19% 12 new york 63rd street connector 2001 $645.0 $632.3 -1.97% 13 pasadena gold line 2003 $693.9 $677.6 -2.35% 14 pittsburgh airport busway (phase 1) 2000-2002 $326.8 $326.8 0.00% 15 portland airport max extension 2001 $125.0 $127.0 1.60% 16 portland banfield corridor 1984 $286.6 $246.8 -13.89% 17 portland interstate max 2004 $314.9 $349.4 10.96% 18 portland westside/hillsboro max 1998 $910.2 $963.5 5.86% 19 salt lake north-south line 1999 $312.0 $311.8 -0.06% 20 san francisco sfo airport ext. 2003 $1,167.0 $1,550.2 32.84% 21 san juan tren urbano 2005 $1,250.0 $2,250.0 80.00% 22 santa clara capitol line 2003 $159.8 $162.5 1.69% 23 santa clara tasman east line 2001 $275.1 $276.2 0.40% 24 santa clara tasman west line 1999 $332.5 $280.6 -15.61% 25 santa clara vasona line 2004 $313.6 $316.8 1.02% 26 seattle busway tunnel 1990 $395.4 $611.1 54.55% 27 st louis saint clair corridor 2000 $339.2 $336.5 -0.80% 28 washington largo extension 2004 $433.9 $456.0 5.09% to verify the assumption of normality, a test of goodness of fit using @risk (palisade corp. 2008) software is conducted on 22 cost overruns/ underruns of historical dataset. the test using the chi-squared statistic passed at 1% level of significance (p-value= 0.0219). fig. 2 depicts the superposition of the normal distribution on the original data histogram. 9.1% 85.9% 12.3% 86.4% -0.150 0.546 -0 .4 -0 .3 -0 .2 -0 .1 0. 0 0. 1 0. 2 0. 3 0. 4 0. 5 0. 6 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 fitted normal dis tribution on cos t overruns /underruns of 22 trans it projects risknormal(0.087940,0.20527) input m inimum -0.1561 m aximum 0.5517 m ean 0.0879 s td d ev 0.2053 v alues 22 n ormal m inimum 8 m aximum +8 m ean 0.0879 s td d ev 0.2053 fig. 2: fitted normal distribution on cost overruns/underruns of 22 transit projects published by atlantis press copyright: the authors 228 fig. 2 demonstrates the limitation in the cost underrun values. it means that in the real world we are dealing with projects that their costs would not be less than a certain value. this certain value can be approximated using historical data. reviewing the historical data, we assume that the fta defines %15=m as the maximum expected underrun. using fig. 2, it is found that the value of α corresponding with %15=m is %1.9=α and the average of cost underruns/overruns is %79.8=δ ; thus 0879.1=β and 2799.1)15.01(0879.1 =−=ρ . after estimating values of α and β from the historical data, the model is ready to be applied on any prospective set of projects which here is the first dataset. from historical dataset (fig. 2), %1.9=α , 0879.1=β , and 2799.1=ρ were estimated. using eqs. (4) and (11), 1875.0=ϕ is calculated. then by the means of eq. (23), the corresponding η s for different γ s are calculated. this is done for γ between 5% and 95% and the result is depicted in fig. 3. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% η γ probability of budget sufficiency in the portfolio vs. in individual projects analytical fig. 3: probability of budget sufficiency in the portfolio of independent projects (γ ) vs. in individual projects (η ) then, eq. (16) is employed to compute the required percent increase in portfolio budget based on the η values (probability that individual projects are sufficiently funded) found from eq. (23). the required percent increase in budget is graphed versus γ and shown in fig. 4. in order to make sure that the results are accurate, we simulated the model to find increasing factor which is superimposed on the analytical curve found using the analytical approach. these two curves are very similar. 0.95 1.00 1.05 1.10 1.15 1.20 1.25 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% b */ b γ required percent increase in the portfolio (first dataset) budget analytical simulation fig. 4: required percent increase in budget )( * bb vs. probability of budget sufficiency in the portfolio of independent projects ( γ ) for example, one can see in fig. 3 that if the fta wants to have 85% confidence that allocated budget for the portfolio of projects will not fall short, it needs to consider a minimum level of confidence of 68.78% ≅ 69% in each individual project risk assessment. also, fig. 4 illustrates that the fta needs to increase the portfolio budget by 16.52% in order to have 85% level of confidence that the budget for the portfolio is sufficient. this finding is significant because the proposed methodology provides a method for calculating the percent increase over existing portfolio budget levels to achieve a certain confidence level in individual projects. in table 2, a comparison is made between the actual cost overrun/underrun of projects in the first dataset and cost overrun/underrun if the budget had been adjusted with the estimated increasing factors. even though the required budget increase in the portfolio can be distributed differently between the projects, we assume all will be increased proportionally by multiplying the required increase factor ( =bb* 1.1652) by the cost at the ffga to reach adjusted cost at the ffga. published by atlantis press copyright: the authors 229 table 2: comparison of cost overrun/underrun of projects in the first dataset using the proposed model cost overrun/ underrun proj. id cost at the ffga (in m$) adj. cost at the ffga (in m$) actual cost (in m$) actual adjusted 3 $413.40 $481.71 $604.40 46.20% 25.47% 10 $675.40 $786.99 $715.30 5.91% -9.11% 17 $314.90 $366.93 $349.40 10.96% -4.78% 25 $313.60 $365.42 $316.80 1.02% -13.30% 28 $433.90 $505.59 $456.00 5.09% -9.81% total $2,151.20 $2,506.64 $2,441.90 13.84% -2.31% in the second column (from the left) of table 2, the original costs at the ffga of all five projects in the portfolio (first dataset) are presented. as it was stated earlier, the proposed model considering the performance of past projects in the historical dataset suggests the fta to increase the total portfolio budget of the first dataset by 16.52% in order to have 85% confidence that the budget for the portfolio (first dataset) is sufficient. then the costs at the ffga of all projects in the portfolio are adjusted by the increase factor of 1.1652 and are shown in the third column of table 2. this column represents the budget of each project if the fta had used the proposed model. the actual costs of projects are given in the fourth column of table 2. the fifth column shows the actual cost overruns/ underruns considering the original budgets at the ffga and the sixth column gives the cost overruns/ underruns if the fta had used the model. table 2 depicts that the model could alleviate cost overrun of some projects in the first dataset. however, the promise of this model is to protect a portfolio of projects against cost overrun. the last row of table 2 shows that if the fta had used the proposed model to allocate budget for five new projects, they could prevent occurring cost overrun of 13.84% with experiencing -2.31% cost underrun. we expect by updating the model and considering the performance of the recently completed projects, we reach more accurate and optimized increasing factor for budgeting of future projects. 5.2. bayesian updating in this step, we use the information collected from completed projects (actual costs) in the first dataset to update the model. the cost overruns/ underruns of five projects are considered new observations and serve to form the underlying distribution. the prior distribution is the normal distribution fitted to the histogram of 22 cost overruns/ underruns in the historical dataset with a mean of 8.79% and standard deviation of 0.2053. considering 85% confidence as a reasonable level, we found that 16.52% increase on the total budget was required. by means of bayesian updating and recent performance of the transit projects sponsored by the fta, the α and β of the model can be updated. the prior distribution comes from the historical dataset as follows: ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − − ==′ 2 2053.0 0879.0 . 2 1 exp )2053.0(2 1 )2053.0,0879.0()( δ π δ nf (32) five new observations are the cost overruns/underruns of projects with adjusted cost at the ffga using 16.52% increasing factor shown in table 2. using eq. (28), the joint likelihood function, the product of five individual normal pdfs, is calculated as: )0708.0,0231.0() 5 1582.0 , 5 0981.01330.00478.00911.02542.0 ( , ... )( 1 −= −−−− =⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ ++ ∝ n n kk nl k σδδ δ (33) published by atlantis press copyright: the authors 230 to find the posterior distribution, eq. (30) is used: ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎧ = + × = +′ ′ =′′ − = + ×+×− = +′ ′+′ =′′ 0669.0 )0708.0()2053.0( )0708.0(2053.0 )()( . %13.1 )0708.0()2053.0( )0708.0(0879.0)2053.0(02313.0 )()( ).().( 22 22 22 22 22 22 l l l ll σσ σσ σ σσ σδσδ δ (34) prior, likelihood, and posterior distributions of cost overrun/ underrun are shown in fig. 5. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 p d f v al ue cost overrun/underrun prior, likelihood, and posterior curves prior likelihood posterior fig. 5: the prior, likelihood, and posterior distributions of cost overrun/ underrun using projects in the first dataset the posterior distribution parameters can now be used to update α , β and ρ parameters considering %15=m : ⎪ ⎪ ⎪ ⎪ ⎩ ⎪ ⎪ ⎪ ⎪ ⎨ ⎧ == − = ⇒=−+=′′+= = +− φ = ′′ ′′−− <=−<= 1632.1 85.0 9887.0 1 9887.0)0113.0(11 %90.1) 0669.0 0113.015.0 ( )()( 1 1 1 1 m m zpmxp new new new new β ρ δβ σ δ α (35) replacing the new values of parameters ( 1newα , 1newβ and 1newρ ) in the model, it is ready and updated to be applied to any prospective dataset. one can see in fig. 5 that the posterior curve has become narrower and moved to the left compared to the prior curve. this means when the performance of the recently completed projects came into consideration, the average cost overruns/ underruns ( δ ) from 8.79% in the historical dataset (prior) decreased to 1.13% underrun in the posterior curve. moreover, as an advantage of bayesian updating, when more data becomes available, the dispersion (standard deviation) of the parameter under consideration diminishes and that the posterior curve becomes narrower. it should be noted that if one integrates the newly completed projects into historical dataset and calculates the parameters of the model, this will give an equal weight to all projects. in other words, this approach will not distinguish between a project completed in 1984 and a project completed in 2004. using bayesian approach, the performance of recently completed projects (first dataset) will have more influence on updating the parameters of the model than projects in the historical dataset. in summary, this procedure enables the model to suggest the required increase in the portfolio budget of the future projects considering mostly the performance of the most recent projects used for the updating as well as the performance of historical projects. 6. conclusion in this paper, a model is proposed which uses a truncated normal distribution and utilizes historical data to assist the agencies to estimate the required confidence level for risk assessment in the projectlevel in order to get a desired confidence for the sufficiency of portfolio budget. it also calculates the required increase in the portfolio budget based on the desired confidence level. for instance, in the given numerical example, 22 transit projects (historical dataset) were used to initialize the model and calculate the primary parameters of the model. then the model was applied to five transit projects (first dataset) to estimate the necessary project-level confidence level for risk assessment and the required increase in the portfolio budget while the fta has 85% confidence that the portfolio budget will not fall short. it was found that using the model, in order to have 85% confidence that the portfolio budget for the first dataset is sufficient, each individual project in the portfolio should have a contingency such that %69=η ; moreover, the fta needs to increase the published by atlantis press copyright: the authors 231 original portfolio budget by 16.52%. this model considers the recent performance of the newly completed projects and is updated as new project data becomes available employing a bayesian approach. the example provided here is updated with the actual costs of five projects in the first dataset. the primary parameters (α , β and ρ ) of the model are updated and it becomes ready to be used for any upcoming portfolio of projects. this process can be repeated on regular intervals (e.g. every two years) so that the future projects can be budgeted considering the most recent performance of projects. the proposed model can be used by agencies such as t he fta which is funding a port folio of transit projects every year as an effective tool to develop contingency budget. 7. references 1. ang, a., and tang, w. (2007). probability concepts in engineering: emphasis on applications in civil and environmental engineering, 2nd edition, john wiley and sons, inc, hoboken, nj. 2. bakhshi, p. and touran, a. (2009). “comparison of current probabilistic approaches for b udget estimating for transportation projects.” 7th international probabilistic workshop, delft, the netherlands, 25-26 november. 3. bolstad, w. (2007). introduction to bayesian statistics, 2nd edition, john wiley and sons, hoboken, nj. 4. booz allen hamilton inc. (2005). “managing capital costs of major federally funded public transportation projects.” web-only document 31, transit cooperative research program (tcrp), washington, d.c. 5. flyvbjerg, b. (200 6). “from nobel prize to project management: getting risk right.” project management journal, 5-15, august. 6. palisade corporation. (2008). “@risk: risk analysis add-in for microsoft excel.” version 5.0.1: professional edition, ithaca, new york. 7. pickrell, d. h. (1990). “urban rail transit projects: forecast versus actual ridership and cost.” dot-t91–04, u.s. dept. of transportation, washington, d.c. 8. molenaar, k. r. (2005). “programmatic cost risk analysis for highway mega-projects.” journal of construction engineering and management, asce, vol. 131, no. 3, 343–353, march. 9. sinnette, j. (2004). “accounting for megaproject dollars.” public roads, federal highway administration, vol. 68, no. 1, july/august, available at: http://www.fhwa.dot.gov/publications/publicroa ds/04jul/07.cfm (accessed on feb 08, 2010). 10. shane, j., molenaar, k., anderson, s. and schexnayder, c. (2009). “construction project cost escalation factors.” journal of management in engineering, asce, vol. 25, no. 4, 221-229, october. 11. springer, m. (1979). the algebra of random variables, 1st edition, john wiley and sons, inc, hoboken, nj. 12. touran, a. (2010). “a probabilistic approach for budgeting in a portfolio of projects.” journal of construction engineering and management, asce, vol. 136, no. 3, 361-366, march. published by atlantis press copyright: the authors 232 microsoft word types of credit risks and strategies to improve risk identification by internet of intelligences2 types of credit risks and strategies to improve risk identification by internet of intelligences chongfu huang1,2 1academy of disaster reduction and emergency management, beijing normal university, beijing 100875, china. 2beijing cazl technology service co., ltd., beijing 100088, china e-mail:hchongfu@gmail.com abstract today, people mainly consider credit risks from view of powerful players. the role of the internet in the credit risk management is not fully. in this paper, we have a comprehensive look at the credit risks, so that, suggest two concepts: binary credit risk and triad credit risk. the internet of intelligences is introduced to improve the recognition of credit risk. this paper shows the intensions of three binary credit risks, respectively, as well as overviews the main methods to analyze them. the study shows that, the bank’s customer credit risk is a simple credit risk. the indirect feeling strategy in internet of intelligences can improve the recognition of credit risk that would be seriously affected by changes in the political, economic and social environment. keywords: credit risk; customer; bank; government; internet of intelligences 信用风险的种类和用智联网改善风险识别的方略 黄崇福 1,2 北京师范大学减灾与应急管理研究院，北京 100875，中国 北京崇安智联科技服务有限责任公司，北京 100088，中国 摘要：今天，人们主要是从强势者的角度看待信用风险，互联网在信用风险管理中的作用也没有充分发 挥。为此，本文全面地审视了信用风险，提出了二元信用风险和三元信用风险的概念，并提出了用智联网 改善信用风险识别的方略。本文分别阐述了三种二元信用风险的内涵，概述了主要分析方法。研究表明， 银行的客户信用风险只是一种简单的信用风险。对与政治、经济和社会环境等变化密切相关的信用风险， 智联网的间接感受方略可改善信用风险识别。 关键词：信用风险；客户；银行；政府；智联网 1. 引言 多数学者谈及信用风险，主要是从商业银行的 角度评判贷款者的信用度，进而论及银行承担的风 险。也有专家从政府的角度评判银行的不良贷款余 额，进而论及金融风险。其实，有一种与金融相关 的信用风险更为重要，即：政府信用风险。如果政 府的金融信用急剧贬值，民众财富必然大量缩水。 政府信用风险，等价于恶性通货膨胀风险。显然， 信用风险不仅仅是银行和政府所面临的主要风险， 也是民众不得不关注的重要风险。银行、政府和民 众，三者之间的信用风险，无时不刻不在发挥作 用。 现有的信用风险模型，主要用于商业银行依据 大量法人客户的信息数据，从统计意义上评估具体 客户违约的可能性。商业银行对非上市公司进行信 用风险度量则更多地是考查其历史违约数据和企业 财务指标。 巴塞尔新资本协议对商业银行承担的信用风 险、市场风险、操作风险等，做了严格的规定并建 议了一些具体的分析方法。本质上，巴塞尔协议制 定了以最低资本要求、监管当局的监督检查和市场 纪律为三大支柱的监管框架，是为了管理银行的信 用风险，以期有效地控制和管理国家、公司、项 目、零售、机构、股权等六类风险敞口。 主权信用评级，是评级机构对国家的中央政府 作为债务人履行偿债责任的信用意愿与信用能力的 一种判断。其国内生产总值增长趋势、对外贸易、 journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 44-51 published by atlantis press copyright: the authors 44 黄崇福 国际收支情况、外汇储备、外债总量及结构、财政 收支、政策实施等，均影响国家的偿还能力。一旦 该国发行的货币量大大超过其偿还能力，就发生严 重的政府信用危机。 本文按下述逻辑结构展开研究：在较全面地审 视了信用风险的本质特征后，在第 2 节中对信用风 险进行了分类。在第 3、4、5 节中分别对 3 种二元 信用的内涵和主要分析方法进行了概述。本文在第 6 节中对三元信用的研究方法进行了探讨。在第 7 节中引入了智联网工具，并提出了用直接识别法和 间接感受法改善信用风险识别的方略。第 8 节总结 了存在 6 类信用风险及智联网能改善信用风险识别 的结论。 2. 信用风险的种类 本质上，信用风险是指契约的一方或多方不能 履约而给其他方造成经济损失的风险。现实中，人 们主要是从强势者的角度看待信用风险。特别地， 借款者违约给银行造成经济损失的风险，成为了信 用风险的主要研究内容。 事实上，在一个复杂的社会系统中，存在着大 大小小的银行，林林总总的政府部门，以个体或组 织出现的民众。令 b，g 和 p 分别为银行集合、政 府部门集合和民众集合，则表征信用风险的基本空 间是 { , , }b g pω = 。显然，ω 是一个离散点空间。 ,x y∀ ∈ω , 将 x 中的个体 x 承担由于 y 中的个 体 y 之信用不良产生的信用风险记为 ( )xy xr yϕ= （1） 不失一般性， x∀ ∈ω ，记其容量（元素个数） 为 n（x），即：|x|= n（x）。易知， ,x y∀ ∈ω ， 其卡氏积 x y× 上存在一个信用风险矩阵： 1 1 1 2 1 ( ) 2 1 2 2 1 ( ) ( ) 1 ( ) 2 ( ) ( ) n y n y n x n x n x n y x y x y x y x y x y x y x y x y x y x y r r r r r r r r r r × ⎛ ⎞ ⎜ ⎟ ⎜ ⎟ = ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ l l l l l l l （2） 通常， x yr × 不是对称矩阵，即：如果 ,i j j ix y y xr r 不同时为 0，则 , i j j ix y y x r r≠ i=1, 2, …, n(x); j=1, 2, …, n(y)。大部分信用风险的研究工作，都在于认识 ( )x yϕ 的内涵，并尽可能建立数学模型，根据 y 的属 性资料，计算出 ( )x yϕ 的数值。 事实上，在一个开放的政治、经济系统中， ( )x yϕ 常常会受来自基本空间 ω 中另一元素中的个 体影响。例如，银行承担房贷者的信用风险，可能 由于政府就业政策的改变而改变。 , ,x y z∀ ∈ω , 将 x 中的个体 x 承担由于 y 和 z 中的个体 y, z 之信用不良产生的信用风险记为 ( , )xyx xr y zϕ= （3） 称 ( ) ( ) ( )( )x y z xyz n x n y n zr r× × × ×= 为一个三维信用风险矩 阵。显然，认识 ( , )x y zϕ 的内涵，并能计算其值，则 更有利于进行信用风险管理。 由基本空间 ω 的构成和式（1）可知，二元的信 用风险可分为 3 类： i．客户信用风险：当 ,x b y p= = 时，我们称 ( )x yϕ 为客户信用风险。 ii．银行信用风险：当 ,x g y b= = 时，我们 称 ( )x yϕ 为银行信用风险。 iii．政府信用风险：当 ,x p y g= = 时，我们 称 ( )x yϕ 为政府信用风险。 由式（3）知，在 ω 上还存在 3 类三元信用风 险，它们分别为： iv．民众承担的信用风险，即：x=p，y=g， z=b 时的 ( , )x y zϕ 。 v．银行承担的信用风险，即：x=b，y=p， z=g 时的 ( , )x y zϕ 。 vi．政府承担的信用风险，即：x=g，y=b， z=p 时的 ( , )x y zϕ 。 在基本空间 { , , }b g pω = 中，民众集合 p 的容 量虽大，但个体的影响较小，由其产生的信用风险 最容易被管理。银行集合 b 的容量不大，但影响远 远超过 p 中的个体，由其产生的信用风险，不可小 觑。政府部门集合 g 的容量很小，但影响巨大，一 旦失控，信用体系很可能崩溃。 三元信用风险常常有传递和反馈作用，评价和 管理并非易事。 3. 客户信用风险的内涵和分析方法 虽然随着信用方式的不断变化、信用工具的不 断创新，商业银行对法人客户信用风险本质的认识 也在不断加深，对客户信用风险的理解也在不断深 化，但内涵依然是稳定的。这里的客户，是指银行 的放贷对象，即：借款人。通过大量文献的总结， 我们给出下述定义： 定义 1：借款人不能按期还本付息而给贷款人 造成损失的风险称为客户信用风险。 具有贷款业务的非银行机构，是一种广义的银 行 。 因 此 ， 一 个 合 法 的 贷 款 人 ， 是 基 本 空 间 { , , }b g pω = 中银行集合 b 中的一元。 由于损失只有当违约实际发生时才会产生，因 此信用风险也称为违约风险。 published by atlantis press copyright: the authors 45 信用风险的种类和用智联网改善风险识别的方略 目前，客户信用风险的研究正处在百家争鸣、 百花齐放的阶段。基于不同数据来源，很多学者采 用不同技术原理开展了大量的、多角度的实证模型 研究。其中，数理统计方法（包括多元判别技术、 logit 模型、主成分分析等）、资产组合原理结构模 型、数据深度拟合技术和层次分析法等较为常见。 统计方法常用于提炼出信用风险的主要标识及 风险要素、风险内核等，从而使人们能对标度风险 的量化要素进行分析研究 1。 客户信用风险分析使用的数据信息主要来自于 所研究商业银行的内部信贷管理系统，即：海量的 法人客户信息数据、内部管理的风险信息数据、辅 助判断的参数指标信息、银行内部风险计量参数等 等。其中，客户的基本信息、财务报表、信用等 级、贷款数量、所属类别（行业、地区等）、银行 管理要件、历史数据等是最重要的信息。 同样的数据，使用不同形式的模型，就会得出 不同的客户信用风险。例如，多元判别、logit 模 型、主成分分析、结构模型、简约模型、混合模型 等，常常结果不同，为此，人们用有效性检验来判 别模型的适用性。“回判准确率”是最常用的检验 指标。 值得注意的是，大多数客户信用风险分析是间 接分析法，并不能计算出客户信用风险值（期望损 失）。 例如，由于陷入财务困境的公司其财务比率和 正常公司的财务比率显著不同，从而人们认为企业 的财务比率能够反应企业信用风险状况，所以用财 务比率作为风险指标。例如，早在 1968 年 altman 就提出了所谓的“z” 财务比率模型 2： z=0 .012x1+0.014x2+0.033x3+0.006x4+0.999x5 (4) 式中的 5 个变量分别是 x1＝营运资金/资产总额，x2 ＝留存收益/资产总额，x3＝息税前利润/资产总额， x4＝股东的收益资产/负债总额，x5＝销售额/资产总 额。通过用破产企业和非破产企业各 33 家共 66 个 样本回归统计出式（4），再进行回判比较，可以判 定，当 z＜1.81 时，企业具有很高的破产几率；当 1.81≤z＜2.675 时, 企业处于财务状况不确定之列， 即处于“灰色地带”；当 z≥2.675 时，企业属于正 常企业之列。 4. 银行信用风险的内涵和分析方法 随着各国金融监管当局对系统性风险的日趋重 视,银行信用风险成为了人们的关注点之一。 对于风险控制管理来说, 除了要考虑正常情况下 金融机构的可能损失, 更重要的是必须确保在极端市 场情况下, 金融机构所持有的金融资产部分不会让该 机构出现破产的风险。1997 年亚洲金融危机爆发 前，风险价值(var)法广泛用于测量银行的风险。之 后，人们认为远远不够，还须考虑市场风险和操作 风险等。巴赛尔委员会提出了压力测试，找出金融 机构在极端市场情况下的承受能力，对风险作出判 断。 显然，相对于客户信用风险而言，人们对银行 信用风险的研究，仍处于起步阶段。由于银行的破 产将导致许多社会问题，并且要由政府来承担，尽 管最后是用纳税人的钱来收拾烂摊子。因此，政府 将承担由于银行之信用不良产生的信用风险。据 此，本文给出下述定义： 定义 2：银行因管理不善而失去信用给政府造 成损失的风险称为银行信用风险。 投资银行为了追求高额回报，致使金融衍生产 品不断泛滥，银行信用风险严重时将导致金融危 机。例如，美国的两大房贷融资机构房利美和房地 美，向还贷能力不高的低收入者发放大量按揭贷 款，再通过资产证券化等手段将这些高风险资产包 装上市，用此金融衍生产品转移风险牟取暴利，最 终于 2008 年陷入危局，并至使雷曼兄弟公司破产， 形成的金融风暴席卷全球。 今天，全球资本主义所向披靡，流动性泛滥， 很难监管的大量影子银行呼风唤雨，其信用风险， 远远超过正规银行。 除了一些正规不正规的金融机构涉足银行业务 成为影子银行外，银行的表外业务风生水起，也是 影子银行。 由中国的地方政府发起设立，通过划拨土地、 股权、规费、国债等资产，迅速包装出一个资产和 现金流均可达融资标准的公司，虽然被称为“地方 融资平台”，其实是一种新型的影子银行。官方的 报导是，截至 2012 年 9 月末，平台贷款的余额为 9.25 万亿元，地方融资平台目前的风险整体可控 3。 民间研究资料则表明 4 ，早在 2010 年，地方政府投 融资平台的贷款余额，相比 2009 年的 6 万亿增长了 42%（达 8.52 万亿）。2010 年末地方政府投融资平 台的银行贷款余额是 gdp 的 20%，相当于全国财政 收入的 102%，地方财政收入的 209%。显然，这种 庞然大物似的影子银行，其信用风险，几乎让人不 敢谈及。 由风险价值、市场风险和操作风险为主产生的 银行信用风险，很难全面系统地进行分析。仅以操 作风险为例，涉及不完善或有问题的内部程序、人 员及系统或外部事件所造成损失的风险5。虽然新巴 塞尔资本协议给出了从简单到复杂的三种方法，即 基本指标法、标准法和高级计量法，但目前仍缺乏 成熟的数量模型来度量操作风险。新巴塞尔协议 6 推 published by atlantis press copyright: the authors 46 黄崇福 荐使用的高级计量法就有多种模型，式(5)是其中的 一个模型。 ∑ ×××= ji jilgejipejieijior , ),(),(),(),(γ (5) 式中，i为第i种操作; j为第j类风险事件; γ 为预期损 失转化成资本配置要求的转换因子； ei 为风险暴露 的规模和金额； pe为损失事件的发生概率; lge为 事件的损失程度。 5. 政府信用风险的内涵和分析方法 2012 年 5 月，高额财政赤字的希腊全面暴发政 府债务危机，人纷纷到银行去提领存款，银行出现 严重的挤兑。随后，一轮又一轮的罢工活动，使经 济发展雪上加霜。值得宽慰的是，作为欧元区国家 的希腊，没有了自主发钞权，政府不得不采取紧缩 措施，eu 和 imf 的救援，基本稳住了势态。 与之形成鲜明对比的是，西方从 2002 年开始对 津巴布韦实施经济制裁，并不断加强。这导致津巴 布韦的经济每况愈下，通货膨胀失控，100 万亿元 面值的大钞于 2008 年问世，通胀率超过百分之两 亿。 建立在经济发展和财政收入基础上的政府信 用，通常都较为坚实。但是，内因和外因都有可能 削弱或破坏这种信用，其严重问题给人民造成的风 险，常常难以承担。据此，本文给出下述定义： 定义 3：政府因重大失误导致经济或财政问题 而失去信用给民众造成损失的风险称为政府信用风 险。 显然，因被动卷入战争或极端自然灾害而使人 民承受苦难，不属于政府信用风险问题。而以全球 经济危机为借口维持重大失误致使人民承受苦难， 则属于政府信用风险问题。 衡量政府信用风险的两大指标分别是主权信用 评级和通货膨胀率。 主权信用评级，是信用评级机构对一国政府作 为债务人履行偿债责任的信用意愿与信用能力进行 的评判。惠誉评级、标准普尔和穆迪是国际三大评 级机构。除了要对一个国家国内生产总值增长趋 势、对外贸易、国际收支情况、外汇储备、外债总 量及结构、财政收支、政策实施等影响国家偿还能 力的因素进行分析外，主权信用评级还要对种种改 革所造成的财政负担进行分析。主权信用评级一般 从高到低，分为 aaa，aa，a，bbb，bb，b， ccc，cc，c。 目前并不存在一个定量模型可以对政府信用风 险进行分析，这是因为根据定量因素所确定的纯模 型化的方法不能够掌握政治、经济、金融和社会因 素之间相互关系的复杂性，这些因素之间的关系是 很难预测。穆迪采取了分步骤评级的方法，共分为 三大步骤、四大因素。 虽然主权信用评级是对国家的信用进行严格评 级，但聚焦于“来自某个特定国家的债务人能否得 到他们需要的外汇以便偿付它们未来的外币债务” 7 。因此，充足的外汇储备能使国家获得很高的主权 信用评级。但是，能对国外履行还款义务并不意味 着本国民众要承担的政府信用风险就小。例如，本 币对外明显升值而对内严重贬值，意味着政府的信 用风险正在增加。通货膨胀率，正好能补充主权信 用评级的不足。 传统的通货膨胀，指在纸币流通条件下，因货 币供给过大，导致货币贬值而引起的一段时间内物 价持续而普遍地上涨现象。今天，由于电子货币的 出现和授信方式的多样化，通货膨胀率常常反映在 消费者物价指数cpi上。cpi由市场上的货物价格增 长百分比来定义。在正常经济体中，人们的平均工 资增长速度很难超越3-4%，因此大多数国家都将cpi 超过3%视为严重的信号。 cpi是一个滞后性的数据，用其分析政府信用风 险，并不理想。加之某些国家为了社会的稳定，刻 意将价格上涨严重的商品排除在cpi成分之外或根据 需要调整各部分比重。用这种人为cpi分析出的结 果，更是没有意义。在美国，构成cpi的主要商品共 分八大类，其中包括：食品、酒和饮品、住宅、衣 着、教育和通讯、交通、医药健康、娱乐、其他商 品及服务，市场敏感度非常高。 一些学者也探讨用各种货币供应量来分析政府 信用风险。我国的货币供应量是以m0、m1、m2为框 架体系。与消费变动密切相关且最活跃的货币供应 量是现金，称为m0。m0与非金融性公司的活期存款 之和为m1，它反映了社会的直接购买能力。m2=m1+ 非金融性公司的定期存款+储蓄存款+其他存款，m2 反映了现实的购买力，也反映了潜在的购买力。 其实，政府并不发放货币，政府是为央行发行 的货币提供信用担保。货币供应量过大，必然引起 通货膨胀率，政府信用降低。为了给过多的货币寻 找出路，虚拟经济常常受到政府的鼓励，认为资本 市场的发展、金融创新的加速，使货币流通速度不 断变化，从而货币供应量与通货膨胀短期和中期的 关系减弱。 用通胀对政府信用风险进行量化分析的主要困 难是通胀指数常常被操纵。cpi可能反映不了真实的 通胀是最好的例证。有研究认为，mo、ml和m2口径 的货币供应量对我国的通货膨胀均没有影响 8 ；但也 有研究认为，m2与通货膨胀之间存在长期均衡关系 9 。从货币交易方程推导出来用于测度通货膨胀压力 published by atlantis press copyright: the authors 47 信用风险的种类和用智联网改善风险识别的方略 的p-star模型10，历经二十多年的发展，目前仍停留 在研究阶段。 6. 三元信用风险的研究方法 本文提出的三元信用风险 ( , )x y zϕ ，远比任何的 二元风险 ( )x yϕ 复杂。原则上来讲，用演绎推理或统 计学习总是可以构建出二元风险的分析模型，但难 以构建三元分析模型。原因是演绎推理中很难处理 两个自变量之间的相互影响，而统计学习的样本基 本无从获取。 从理论上讲，研究三元信用风险，只能从社会 经济动力学入手，由构造微分方程起步，用各种约 束简化问题。从实用的角度讲，将自变量区分为主 自变量和从自变量，则多态灾害链模型 11 可发展为 三元信用风险分析模型。我们以“民众承担的信用 风险”为例来加以说明。 从概率风险的角度看，一个风险现象的形式化 描述就是不利事件发生的概率 p 和后果 v。特别 地，当我们假定它们与某个参数 y 有关时，风险分 析的主要工作就是识别概率分布 p(y)和脆弱曲线 v(y)。当风险的内涵是破坏的期望值时，风险值为 0 ( ) ( )r p y v y dy ∞ =∫ （6） 例如，假定政府的信用完全由通胀率 y 决定， 又假定民众因 y 成的财富蒸发是 v(y)。当 y 在研究时 期内的概率分布是 p(y)时，风险值由式（6）算出。 分析三元信用风险的困难在于，民众财富的蒸 发不仅受政府信用的影响，而且可能受银行破产的 影响。高通胀不仅使财富蒸发，还会使企业破产造 成大量失业。在将自变量区分为主自变量和从自变 量后，三元信用风险分析的问题就被大大简化。 在某些国家，高通胀刺激房货，银行潜在风险 增加，上市银行的股民除承受政府信用风险外又承 受银行股价下跌的风险。外观上看，如果能分别计 算出财富蒸发的期望值和股价下跌造成损失的期望 值，两者相加则是民众承受的风险。问题是，通胀 率和银行信用并不独立，简单的相加可能严重失 真。 将多态灾害链模型发展为多态风险分析模型， 关键是找出一个链结构，通过用各风险环的研究， 分析出民众承担的信用风险。 在多态灾害链中，我们用 m,o,d,l 分别表示致 灾因子、承灾体、破坏、损失参数（或矢量），用 p(l)表示损失 l 的概率值。一个灾害环的形式化描 述是： ⎩ ⎨ ⎧ = = ),( ),()( omfd mdslp （7） 式中，函数 f 的形式和参数由承灾体所处环境和承 灾体类型决定；函数 s 的形式和参数由承灾体社会 属性和致灾因子 m 的不确定性属性决定。 如果有灾害链存在，则破坏 d 将诱导出新的致 灾因子，记为 )(1 dm φ= 。例如，地震引起山崩， 山崩形成拦河大坝，大坝溃决造成洪水灾难。山崩 是破坏 d，洪水是由地震诱导出的致灾因子 1m 。诱 导关系 φ ，可能非常复杂，也可能十分简单。 显然，不失一般性，在三元信用风险 ( , )x y zϕ 中，我们可以假定 y 是主自变量，z 是从自变量，并 假定 z 由 y 诱导，且 y 在发展的过程中有足够的时 间允许人类对其加以影响。与放任不管的形态相 比，可能会有另外的形态出现。这样，系统风险就 会出现多态性，本文称其为多态风险链。 设 x 承担由于主自变量 y 之信用不良产生的信 用风险为 ( )x yϕ ，从自变量 z 由 y 诱导的关系式为 0=φ(y,z) ，则一个风险环的形式化描述是： ⎩ ⎨ ⎧ =φ = 0),( )( zy yr xϕ （8） 由于 x 承担从自变量 z 之信用不良产生的信用 风险可记为 ( )x zϕ ,于是，由 y 诱导的风险环是 ⎩ ⎨ ⎧ =φ = 0),( )(1 zy zr xϕ （9） 当有 n 个手段可对 y 的发展过程加以影响时，y 的诱导形式和结果都会有所不同，这些诱导式可分 别记为 nφφ ,,1 l 。于是，一个多态风险分析模型 可表为： ⎩ ⎨ ⎧ =φ = 0),( )( zy yr xϕ ， ⎪ ⎪ ⎪ ⎩ ⎪⎪ ⎪ ⎨ ⎧ ⎩ ⎨ ⎧ =φ = ⎩ ⎨ ⎧ =φ = 0),( )( 0),( )( 1 1 1 zy zr zy zr n x x ϕ ϕ m （10） 在 y 失信前，人们并不知道什么样的影响手段 将会被采用。于是，分析三元信用风险，必须考虑 概率分布 )(,),(),( 21 nφppφp lφ 。这显然不是一件 容易的工作。 7. 用智联网改善信用风险识别 信用风险的分析工作是一项智能工作。如何发 挥互联网的作用，集众人之所长，生超级之智能， 解风险之万千，是一个值得研究的问题。 面对一个具体的信用风险问题，目前普遍依赖 利益相关者自己的经验和各种途径得来的专家指导 published by atlantis press copyright: the authors 48 黄崇福 去应对。但是，世界越来越复杂，大数据时代已经 到来，各种风险瞬息万变。无论多复杂的信用风险 分析工具，无论多大的数据库，都会越来越难以满 足需要。借助互联网强大的功能，集百家之智，历 万家之险，解动态之迷，才有可能大大提高人们应 对信用风险的能力，使复杂变简单，不变应万变。 基于人类智慧远远高于机器智能而互联网能集 成高级智慧的认识，文 12 首次提出“智联网”的概 念，它由智能体、互联网和一个数学模型构成，以 汇集和处理相关信息，有望提高人们对各种风险的 分析能力。为此，本文提出用智联网改善风险识别 的方略。 7.1. 智联网概念 “智联网”(internet of intelligences)，是由各种智 能体，通过互联网形成的一个巨大网络。其目的是 集小智慧为大智慧，群策群力，帮助人们更好地认 识世界，获得更好的生活质量。 设 a={a1,a2,…,an}是含有 n 个智能体 a1,a2, …, an 的一个集合。设 n={s, c1,c2,…,cn}是含有服务器 s 和 n 台计算机网络终端 c1, c2,…,cn 的，一个可以独立工 作的子系统。设 m 是处理 n 个智能体所提供的信息 的模型。三元体（a, n, m）称为一个智联网。 设（a, n, m）是一个智联网，智力水平为 q。 再设 a 中个体的最大智力水平为 q。当 q 大于 q 时，称（a, n, m）为一个正向智联网。智联网是一 个智力提升系统，只有正向智联网才有意义。简单 的网络调查系统不是智联网。 最早的计算机网络于 1969 年正式启用，由美国 军事部门 arpa 研制，称为 arpanet，仅连接了 4 台计算机。1974 年，人们开发了 tcp/ip 协议，继而 产生了世界上第一个网页浏览器和第一个网页服务 器，人类进入互联网时代。随后搭建的网络高速公 路，实现了组织机构间的联网，电子商务应运而 生。这个时期的网页模式可称为 web 1.0。 在二十世纪 90 年代，人与人之间的联网得以实 现，产生了个体对他人跨界响应的“社会技术”， 名目繁多的网络社区层出不穷。这个时期网络界面 的模式可称为 web 2.0。 网络技术的飞速发展，为人们从网络虚拟空间 走向网络实体空间创造了条件，并刺激人们用算法 从网页中提取信息。于是物联网应运而生，语义网 呼之欲出。帮助人们在任何时间、任何地点、以任 何方式进行信息的获取与处理的普适计算，成为今 天的研究热点，云计算等正在普及。这种建设中的 网络被称为 web 3.0。 然而，无论是千奇百怪的单体计算机还是纵横 全球的计算机网络系统，虽然能较好地处理事务性 和计算性的工作，但不能处理未知情况，难以适应 快速变化的环境，无法进行创造性的工作。所以， 现在能够完成复杂任务的网络，只不过速度更快、 数据更多，花样更新，但其智力还不如 3 岁的小 孩。 未来的网络系统，是智与智之间联网，这就是 智联网。为构建正向智联网而开展的增强思维之理 论和方法的研究，内涵丰富，挑战无限。这种未来 的网络，基本模式可称为 web 4.0。图 1 总结了从 web 1.0 到 web 4.0 的发展历程。 图 1. web 1.0 到 web 4.0 的发展历程 由图 1 可知，互联网发展的必然趋势是智联 网。今天我们处于从 web 2.0 向 web 3.0 发展的过 程中。 7.2. 用智联网改善风险识别的方略 征信系统大量投入商业应用说明，银行对普通 客户信用风险的识别，技术上已无障碍，只剩精度 和成本问题。然而，对涉及新兴产业市场的客户， 征信系统可能失效。更一般地，一旦政治、经济和 社会环境等的变化较大，基于客户历史信用记录的 传统风险识别技术就会失效。商业银行风险和政府 信用风险，相当隐蔽且变化多端，不仅个体民众很 难对其加以识别，评估机构也常常误判。三元信用 风险，更为复杂，目前尚无有效识别技术。 对此，智联网有两个应对方略：直接识别和间 接感受。 i．直接识别法 设 o 的信用风险是待识别对象。传统的识别方 法是将 o 的属性数据放入某种模型，经分析而辩识 出风险的大小。与利益相关者的关系数据、抗压能 力等也可在属性数据之列。用什么样的属性数据， 由所谓的风险评估指标体系确定。 用智联网对 o 的信用风险进行识别，依据的是 网上智能体对问题的认识，而非对属性数据的群体 分析。 人与人之间联网 社会技术 web 2.0 物与物之间联网 普适计算 web 3.0 智与智之间联网 增强思维 web 4.0 组织机构间联网 电子商务 web 1.0 published by atlantis press copyright: the authors 49 信用风险的种类和用智联网改善风险识别的方略 当 o 的信用风险与公众生活密切相关且有大量 公开资料时，可用智联网对其加以直接识别。例 如，某大企业 o 拟生产一种全新的日用品，其巨额 贷款的信用风险，不仅与征信系统中以往 o 的属性 数据有关，还与消费者现今对 o 的认可程度和该新 产品的市场有关。用智联网直接对 o 的认可程度和 新产品进行评价，并辅之以对以往信用的分析，就 可以识别出对此项贷款的信用风险。 此时，商业银行由智联网发布的问题形式是： 对 o 的认可程度如何？消费此产品的愿望如何？并 要求给出理由。智联网（a, n, m）中的模型 m，首 选拼图模式。 任何智联网都是一个交易平台。银行使用智联 网对大客户进行信用风险识别，须有相应的投入， 大部分支付给有效应答的智能体集合 a。由于智联 网具有过滤虚假信息和处理不完备信息的能力，其 投入产出将会远远高于传统的征信系统。 ii．间接感受法 许多信用风险不可计算，只可感受。例如， 2012 年欧洲债务危机，不可避免地会影响到政府信 用和银行信用，但其风险难以计算，只可感受。对 于大型项目的投资者而言，识别出这类信用风险， 有利于科学决策。如果政府信用风险过大，社会可 能不稳，自有资金不宜投入；如果银行信用风险过 大，计划中的贷款会有变故。 显然，任何投资者都无力针对政府或银行建立 征信系统，只有从媒体信息才能感受相关风险。由 于不同环境下的语义差别很大，现有网页提取的信 息不能用于风险识别。 当人们无休止地争论用什么样的标准来建设语 义网，便于机器识别和理解时，智联网跳过了这障 碍，用工具箱模式解决了这问题。当人们大力开发 生物算法以图实现计算智能时，智联网直接使用人 的智能而简化过程。当人们热情不减地进行词语计 算的研究，以图用网页上的词语进行推理时，智能 体天生的词语处理能力正被智联网的各种模式所使 用。 于是，用智联网间接感受，并辅之以必要的统 计分析，就可以识别出难以计算但涉及公共利益的 信用风险。 以项目投资者对政府信用的间接感受而言，其 由智联网发布的问题形式，可类似于：非你个人原 因的生活压力近期是否会有缓解？给出判断的理 由。如果压力进一步增大，三年后的极端情况会是 什么？并要求给出理由。智联网（a, n, m）中的模 型 m，首选路况模式。 8. 结论 金融风险防范的基础工作之一是信用风险识 别。以往人们过多关注微观层次的信用风险，大量 征信系统投入应用。宏观性较强的银行和政府信用 风险，正在引起人们的关注。 由银行、政府和民众组成的系统中，存在 6 类 信用风险：客户信用风险、银行信用风险、政府信 用风险、民众承担的信用风险、银行承担的信用风 险和政府承担的信用风险。前 3 类是二元的信用风 险，后 3 类是三元的信用风险。 传统上，客户信用风险主要是根据属性数据进 行识别。巴塞尔资本协议推荐使用的方法有助于对 银行信用风险进行识别。主权信用评级可作为政府 信用风险的一项重要指标。三元信用风险的识别因 两个自变量的关联而相当复杂。多态风险分析模型 可略为简化问题。 由智能体集合、计算机网络和信息处理模型组 成的智联网，由于具有动态跟踪和提升智能的能 力，十分有助于改善信用风险的识别。直接识别方 略和间接感受方略都可以将传统的静态模式推进到 动态，提供风险识别的准确度。 由于受与“北京崇安智联科技服务有限责任公 司”签订的保密协议约束，智联网中的相关技术， 无法在本文加以介绍。 参考文献 1. gu qianping, research of credit risk model for companies in commercial banks, dissertation for the doctoral degree in tsing-hua university tsing-hua university, (2009). 顾乾屏, 商业银行法人客户信用风险模型研究, 清华大 学经济管理学院博士学位论文, 清华大学, (2009) 2. e. altman, financial ratios: discriminant analysis and the prediction of corporate bankruptcy. the journal of finance. (23) (1968): 589-609. 3. wei yutao, emphasis on legal risks of local financing platform, china economic times, 2012.11.30, third version. 魏域涛, 重视地方融资平台的法律风险, 中国经济时报, 2012 年 11 月 30 日, 第 003 版. 4. wu wei, ai junwei , risk management research of local government investment and financing platform foreign and economic relations & trade, (3) (2012): 57-59. 吴蔚, 艾军卫, 地方政府投融资平台风险管理研究, 对 外经贸, (3) (2012): 57-59. 5. basel committee on banking supervision. sound practices for the management and supervision of operational risk. (basel, switzerland, bank for international settlements, 2003). 6. basel committee on banking supervision. international convergence of capital measure-ment and capital standards: a revised framework. (basel, switzerland, bank for international settle-ments, 2004). published by atlantis press copyright: the authors 50 黄崇福 7. hu weiming, sovereign credit rating model of moody and its enlightenment, china finance, (9) (2012): 78-80. 胡伟民, 穆迪主权信用评级模式及其启示, 中国金融 , (9) (2012): 78-80. 8. chen yanbin, tang shilei, li du. can money supply forecast inflation in china? economic theory and business management, v(2) (2009): 22-28. 陈彦斌, 唐诗磊, 李杜, 货币供应量能预测中国通货膨 胀吗? 经济理论与经济管理, v(2) (2009): 22-28. 9. yang jianming, china's money supply empirical rese arch on output, price forecasting ability, naikai economic studies, (1) (2003). 杨建明, 我国货币供应量对产出、物价预测能力的实 证研究, 南开经济研究, (1) (2003). 10. j. j. hallman, r. d. porter and d. h. small, m2 per unit of potential gnp as an anchor for the price level. staff studies, no.157, board of governors of the federal reserve system, 1989 11. huang chongfu, status of comprehensive risk management framework for design and multi-state,disaster, jo urnal of basic science and engineering, 14(sup) (2006): 29-37. 黄崇福, 综合风险管理的地位、框架设计和多态灾害 链风险分析研究, 应用基础与工程科学学报, 14(增刊) (2006): 29-37. 12. 黄崇福. 风险分析在线服务的智联网. journal of risk analysis and crisis response, 1(2) (2011): 110-117. published by atlantis press copyright: the authors 51 microsoft word volume 12, issue 4-4 journal of risk analysis and crisis response, 2022, 12(4), 204-221 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i4.344 204 review fintech, bigtech credit and economic growth: a bibliometric review and meta analysis ephias munangi 1 and athenia bongani sibindi 1,* 1 department of finance risk management and banking, university of south africa (unisa), pretoria (0003), gauteng, south africa * correspondence: sibinab@unisa.ac.za; tel.: +27-(0)-12-429-3757; fax: +27-(0)-86-569-8848 received: october 16, 2022; accepted: november 26, 2022; published: december 31, 2022 abstract: the financial sector plays a vital role to improve economic performance. development of innovative lending strategies is necessary since debt is the largest form of finance. fintech and bigtech credit are such innovative ways in the new era of extreme technological advancements that help ensure the economy is inclusive and grows in a sustainable manner. this paper applies a bibliometric review and meta-analysis by utilizing the scopus database to explore the academic research that has been conducted on fintech/bigtech credit and economic growth. the vosviewer software was utilized to construct and visualize the bibliometric networks from the extracted data. our findings show a massive increase in documents published on this topic between 2019 and 2022. bank concentration, bank minimum capital requirements, market inefficiencies, financial inclusion and regulation on technological innovation were identified as key factors to influence fintech/bigtech credit adoption. the significance of the study is that we identify the research gaps from previous studies and demonstrate the importance of establishing the nexus between fintech/bigtech credit and economic growth. keywords: fintech; bigtech credit; economic growth; sustainability; financial inclusion; regulation; bibliometric review 1. introduction the three main forces that drive the economy are productivity growth, short-term debt cycle and the long-term debt cycle (berger & udell, 1998). financial services integrating with digital technologies has significantly bridged the gap that existed previously between the banked and unbanked (liu, luan, wu, zhang & hsu, 2021). the credit granting process is now different because technology could analyze a plethora of data and work with individuals considered high risk with a huge probability of default in the past (beck, 2020). therefore, assessment by lenders to safeguard their interests on creditworthiness to determine the default rate is more diverse across various variables (munangi & sibindi, 2020). technology helps to produce more as better improved tools are developed which makes work to be done more efficiently over a shorter period (prisecaru, 2016). economics subscribes to the notion that financial sector development or supply of finance results in economic growth (ahmad & malik, 2009). the financial sector connects savers to borrowers and provides impetus for economic growth (sufian & habibullah, 2012). credit granting is one tool that ensures people acquire capital and can be productive, therefore, necessitates new lending strategies and new approach to conducting customer due diligence and credit score assessment (turlais, 2016). ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 205 advanced methodologies adopted with improved analysis cater for idiosyncratic risks of borrowers. bigtech/fintech credit help to achieve inclusive finance development that stimulates economic growth. policies for monitoring purposes are needed to ensure there is no reckless lending. structured policy reforms that create opportunities for everyone to attain regenerative and distributive economies (rahman & alam, 2021). the balance of supply and demand in the credit market leads to sustainable economic development through prevention of over-indebtedness (nyaruwata, 2009). lending must be approached with caution since inequalities remain in institutional processes on type of credit and terms and conditions for the loans to previously marginalized groups on assessment of ability to repay (dwyer, 2018). the financial sector reduces income inequality where there are poor rules of law (úbeda, forcadell, aracil & mendez, 2022). debt should not rise faster than income, and income not rise faster than productivity (rode, 2020). when total money and credit increases but does not match the goods produced this leads to inflation (girdzijauskas, streimikiene, griesiene, mikalauskiene & kyriakopoulos, 2022). inflation is a macroeconomic factor that influences interest rates and policy adoption that alters economic growth (njoki, 2014). monetary policies set interest rates to control inflation (wambui, 2013). borrowers find it difficult to borrow when interest rates are high, so money supply grows slowly, which then assists to control inflation (duncan, 2011). on the other hand, people find it easier to borrow when interest rates are low, so money supply grows quickly, which ultimately raises inflation. raising productivity leads to a flow of economic growth which is sustainable. sustainable growth is pivoted on a financial system that prioritizes the best investment projects (gros & alcidi, 2013). this involves being agile to adapt to inclusive innovation, such as fintechs that advocate for green financing (vergara & agudo, 2021). bigtech/fintech credit lending platforms have grown across the world (cornelli, frost, gambacorta, rau, wardrop & ziegler, 2020). these platforms have resolved the information asymmetry problem. they are managing to expand credit to households and small businesses reducing transaction costs and improving efficiency (mishkin & strahan, 1999). even though lending decisions and efficiency of processes improves, it is however argued that the results show an increase of information asymmetry problems by other authors (fasano & cappa, 2022). technology has been disruptive; however, competition must not be destructive but instead encourage efficiency and collaboration. fintechs/bigtechs should complement banks so that society benefits through synergy (suprun, petrishina & vasylchuk, 2020). bank strength being resources and security even against cyber risks, and fintechs on the other hand bringing new technologies. fintechs also assist with economic growth and economic development considering welfare and the well-being of a country (musabegovic, ozer, djukovic & jovanovic, 2019). fintech/bigtech credit volumes, their percentage share of the credit markets and potential impact on economic growth must be investigated (cornelli, frost, & gambacorta, 2020). their growth in credit volumes has been observed to be driven by regulation, competition in the banking sector and an economy’s development (frost & turner, 2018). a research gap has been observed to exist on the impact of fintechs on the financial sector as well as on the entire economy (bahrini & qaffas, 2019). against this backdrop the aim of the study is to determine the factors that drive alternative credit and to establish the link between fintech/bigtech credit and economic growth. a bibliometric literature review is employed in this regard to establish the productive trends, identify some top ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 206 influential authors, most influential documents and the most influential sources in this realm. furthermore, it helps to have a network analysis and visualization that includes citation analysis of documents, co-citation analysis of authors and analysis of co-occurrence. meta-analysis is also conducted to map a direction for future studies. the rest of the article is organized as follows: section 2 reviews the related literature. section 3 presents the research methodology employed to execute the study. the research findings are presented and discussed in section 4. lastly, section 5 concludes the article. 2. literature review 2.1. overview of fintech/bigtech credit developments in technology through innovation is rapidly transforming the financial, monetary and credit systems (gomber, kauffman, parker & weber, 2018). online platforms make use of digital technology to interact with customers to offer credit making use of huge amounts of data. fintech credit has predominantly been more successful in countries with higher income levels but low competitive banking system (claessens, frost, turner & zhu, 2018). in addition, less stringent regulation encourages fintech credit advancement. contrarily, the regulations have built consumer confidence adding to new digital tools and automation pushing marginal costs down and reducing transaction costs (arner, barberis & buckley, 2017). other factors besides regulation and technological advancements that affect fintech adoption include competitive environment, unmet demand and changing demographics (frost, 2020). technology companies due to their nature acquire huge amounts of data which they obtain and can process in real time (shamim, zeng, shariq & khan, 2019). information obtained allows the tech firms to effectively measure loan quality and potentially reduce loan defaults. business models that drive alternative finance are in most cases technology enabled and offering funds quickly from a wide range of investors. technology has become a useful tool to enhance existing financial capabilities (dawei, anzi & gen, 2018). it improves results from previous studies that highlighted stability of macroeconomic environment leading to sustainable economic growth, well measured economic policies positively impacting growth, inflation rate managed and budget deficit reduced for the economy to grow (fischer, 1993). 2.2. theoretical literature review 2.2.1. bank market power competition eradicates monopoly previously enjoyed by traditional institutions (gąsiorkiewicz, monkiewiczj & monkiewiczm, 2020). monopoly without competition through barriers to entry led to inflated prices and resulted in financial exclusion. competitive behavior of banks affects individual and firm access to financial services. banks with market power will have resources to invest in technologies to acquire necessary information. however, higher bank market power seems to increase the cost of credit for firms, but the effect is very small. it was only disruptive innovation that could force itself in the financial services market. this led to competition induced innovations and that increases productivity and ultimately grows the economy (hunt, 2012). fintechs through their innovative methods have disrupted the financial sector and come up with methods to reach many ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 207 customers who were not regarded as credit worthy by using the vast amount of data at their disposal which was not regarded useful. in developing countries, bank market power has resulted in a negative relationship with cost efficiency, but positive with profit efficiency and stability (turk ariss, 2010). the results align to the objective of those already in the market blocking new entrants to maintain huge profits. established institutions on the other hand enjoy cost asymmetry due to economies of scale. there is also competition for customers when market concentration is low. this leads to low interest rates being charged by those offering credit which makes many people to be able to afford to take loans. therefore, there is a direct link between bank market power, financial inclusion and economic growth. 2.2.2 . credit creation theory banks create money and increase money supply through fractional reserve banking. money is created collectively through multiple deposits (starkey, 2018). when savers deposit money and it is lent out, it in most cases is deposited into banks again by the borrowers through expenditure (more borrowing leads to more expenditure). banks create credit (freimanis & šenfelde, 2019). reserve requirement is money that should be left in the bank when deposited by a saver that should not be lent. stability is enabled when borrowers pay on time. money multiplier which is equal to 1 divided by the reserve ratio. it is the calculation of the amount of money generated by the banks from their reserve requirement. the concept is because we do not all go to the bank at the same time and demand for our money, that causes a bank run leading to default. 2.2.3. quiet life theory trading off higher profits for less risk to enjoy a quiet life. market power influences risk return preference behavior towards risk avoidance (tabak, gomes & da silva medeiros, 2015). banks would therefore hold less loans since loans are riskier. this is more the case in monopolistic markets rather than competitive markets. however, a monopolistic creditor in a concentrated credit market is more likely to offer credit to upcoming firms since they can leverage on time, future rents (petersen & rajan, 1995). a mixed relationship was regarded the best description between bank competition and risktaking as results from previous research produced conflicting results (boyd & de nicoló, 2005). 2.2.4. intermediation and technological adoption financial intermediation is the process by which financial institutions accept savings and lends them out. the theory was developed from the works of gold smith, gurley and shaw in 1955 who discussed institutionalization of the process of financial intermediation (amaira & amairya, 2014). financial intermediaries are the institutions that transfer the funds between the savers and borrowers. it is the dominant theory of banking with liquidity, risk and information as the three main intermediation functions (freimanis & šenfelde, 2019). fintech credit is credit activity facilitated by electronic (online) platforms. new lending tools and various ways to evaluate credit worthiness have become useful to attain the financial inclusion agenda. third party payment service providers, that is, account information service providers (aisps) and payment initiation service providers (pisps) have been introduced as addition to the traditional method (braun, 2018). the new services offered by new players in the financial markets space has led to an increase in fintech adoption. information and communications technology (ict) has been found to have a ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 208 positive relationship on economic growth in both developing and developed countries (appiah-otoo & song, 2021). results further revealed that the increased influence is dominant in poor countries. in support, ict adoption increases economic growth both in the short and long run (hussain, batool, akbar & nazir, 2021). however, it is one of many drivers of economic growth. other studies revealed the need for an integrated policy since ict infrastructure development, financial inclusion and economic growth are interdependent (pradhan, arvin, nair, hall & bennett, 2021). technology adoption is a complex matter that goes beyond just innovation in technology (sharma & mishra, 2014). it involves understanding users’ attitude, personality, trust in the system and the society in which the people live among other factors. technical adoption model can play a significant role in fintech/bigtech adoption. the determinants of adoption are based on perceived usefulness and perceived ease of use and user acceptance of information technology (davis, 1989). 2.3. determinants of fintech/bigtech credit 2.3.1. bank access financial inclusion is when both individuals and businesses have access to finances and use financial services in a sustainable manner (yengeni, 2020). bank access has improved spending decisions of many individuals through budgeting apps that also help with savings. algorithms and computer programs for delivery of financial services entails digital financial inclusion (ozili, 2018). financial inclusion goes hand in hand with poverty alleviation (kebede, naranpanawa & selvanathan, 2021). statistics show many people without bank accounts in africa (demirgüç-kunt, klapper, singer, ansar & hess, 2017). access to and availability of the facilities due to poor infrastructures continues to be a limiting factor. 2.3.2. government intervention government is a critical part of the fintech ecosystem that handles financial regulators and legislature. the level of governance improves financial inclusion and maximizes economic growth (emara & el, 2021). government policies also play a role to attain the financial inclusion objective (evans & adeoye, 2016). the policies that governments enforce can reduce the tax burden on small businesses and encourage them to operate (khanh & loc, 2018). businesses are generally inclined to invest where there is less red tape and no unnecessary stringent regulations. 2.3.3. business models (ease of doing business) the six (6) fintech business models are insurance services, crowdfunding, payment, lending, wealth management, and capital markets ( lee & shin, 2018). to lead in the current market therefore requires use of technology to gather relevant data, which creates an opportunity to target individuals directly to meet their needs accordingly. there is a significant shift in banking business models from the traditional model. we witness collaboration in the fintech ecosystem with developments such as open banking (sitea, 2020). open banking is a model in which banks share data between two or more affiliated parties to deliver and enhance capabilities to the market place. the business model canvas involves nine (9) elements of customer segments, customer relationships, channels, revenue streams, cost structures, value propositions, key activities, key resources, and key partners (keane et al., 2018). this enables platform-based online services to ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 209 become more personalized, seamless, quick and safe. however, implementation differs because of heterogeneity in business models of the various fintech segments of payments, insurtech, regtech, wealthtech, blockchain/cryptocurrency, cybersecurity (kpmg, 2021). 2.3.4. bank efficiency a bank is efficient when it is offering excellent services and has reduced its costs. the increase in digitalization is useful to improve effectiveness and minimize costs (beck, 2020). systems are refined to lower operational costs and reduce turnaround time as there is real-time managerial information provided. virtual banks as an example reduce significant costs of maintaining a branch network. fintechs have low-fee services because they have no offices and incur no maintenance costs on buildings. this in turn lowers the fees and interest rates they charge customers and has resulted in financial institutions performing well despite the competition (nilsson & fredholm, 2020). furthermore, fintechs have offered fee versus interest driven models in credit. 2.3.5. trust age, educational background, and ethnicity are among the elements that play a role on the numbers for those who choose fintech credit (ghosh, 2021). there is a direct nexus between bank misconduct and the expansion of online lending where there is generalized trust (bertsch et al., 2020). the authors’ further mention that perceived unfair treatment by banks and the great recession is a reason why other borrowers explore online alternatives. security of data, integrity of institutions, privacy, accountability and reliability of service offering are key as this will allow for acceptance and build trust to enable execution (zhang & lee, 2003). 2.3.6. financial regulation the interests of all stakeholders must be safeguarded ensuring consumer protection and encouraging fintech innovation (jagtiani & john, 2018). this happens when governments create a stable financial environment to support the economic development. the balance of digitized regulation around innovation and financial stability is very important (arner, zetzsche, buckley & barberis, 2017). this necessitates standardization and governance of application programmable interfaces (apis) and expedite supervisory oversight. the main challenge is that innovation comes before regulation, hence, plays catch up. the ideal is for regulation to move at the same pace with innovation to support changes. regulation is slow and incompatible with digital transformation, therefore rethinking new approaches to regulation is required since the link between regulation and innovation ensures sustainability in any industry. it even deals with issues of bias where regulatory risk models are involved. stringent legislative requirements can be a barrier to entry in the financial market for fintechs even though it does cement financial stability and improves resilience (claessens, coleman & donnelly, 2018). regulatory sandboxes allow for risk management especially in cases where new business models are implemented (restoy, 2020).therefore, it is important to determine the correlation that exist between fintech credit and regulatory stringency. 2.3.7. bank credit ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 210 bank credit had not been easily accessible in the past until the emergence of bigtech and fintech credit. failure to access bank credit has been negatively affecting countries’ economic growth due to the limited number of people to participate in economic activities. this is because the lending standards had been too stringent (berlin, 2009). however, the alternative credit seems to complement bank credit because it has reached out to the unbanked (cornelli, frost, gambacorta, rau, wardrop & ziegler, 2020). 2.3.8. bank competition when there is no competition among banks, they are in a stronger position to charge high interest rates. there are greater chances for collusive pricing (kaluwa & chirwa, 2017). since the banks will be holding all the cards, they will not be willing to take any risks (mishi, sibanda & tsegaye, 2016). bank competition improves bank’s efficiency and results in higher profitability (amidu & wolfe, 2013). there is greater stability in a competitive environment with higher level of diversification. 2.3.9. stock market development it is vital to establish the role stock market plays in economic growth. the stock market is not the economy, it is simply a forward-looking pricing machine. hence, volatility in the market does not always get attention to affect monetary policy. locals benefit when they buy, sell and issue securities in a well-functioning stock market (demirgüç-kunt & levine, 1996). it is however important to ascertain the relationship between gdp growth and stock returns. it can be argued that economic growth does not benefit stockholders and has a negative correlation with stock market returns (ritter, 2012). on the contrary however, other authors found increase in gdp to have a positive impact on stock market performance (indangasi, 2017). economic growth has a positive relationship with stock market development (vazakidis & adamopoulos, 2009). 3. methodology 3.1. bibliometric review and meta-analysis a bibliometric review and meta-analysis were applied in this study. the bibliometric review relates to the use of quantitative studies on research publications to assess scientific research. the application uses both statistical and mathematical methods to evaluate the quality and quantity of published documents to establish trends or pattern of a specific research area. the bibliometric review method measures impact of published articles and citations after statistical analysis thereof that show linkages between articles (fetscherin & heinrich, 2015). meta-analysis on the other hand combines the results of multiple scientific studies. performance of articles is illustrated through matrices such as number of citations and total h index. vosviewer software has been used to conduct citation analysis of documents, co-citation analysis of authors and analysis of co-occurrence bibliometric techniques. 3.2. data and methods ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 211 for the purpose of this study, we used: (1) scopus database to acquire data for analysis; (2) microsoft excel to calculate the frequencies of the published materials and to design the relevant chart and graph; and (3) vosviewer software to construct and visualize the bibliometric networks; the data was retrieved from the scopus database using an appropriate boolean mix of keywords with query string as “fintech credit” or “bigtech credit” and “economic growth”. these keywords were entered into the scopus advanced search title-abs-key interface as a part of a title or words mentioned in the abstract. the focus on title and abstract narrows our search to obtain relevant topic that is significant to our research area together with assisting to attain objective of the study. this step returned 50 articles, and after screening only for those written in the english language, a total of 49 documents have been obtained to conduct the bibliometric analysis. the bibliometric methodological approach reviews the past and assesses the current situation to map beneficial course of action on fintech/bigtech credit that can improve economic growth if the relevant link is established and various elements of work identified that can add value. 4. research findings 4.1. main information about data the sample data for this study was collected for the period 2019 to 2022 and found 49 documents limited to the english language that consist of 41 articles, 2 review studies, 2 books and 4 book chapters. subject area of economics, econometrics and finance as expected has the most documents with a total of 33,00%, followed by business, management and accounting with 20,90% due to the nature of the study. 4.2. production trends from the sample of the study source: authors’ own compilation (data sourced from scopus). figure 1. number of documents and year of publishing. 5 8 12 24 0 5 10 15 20 25 30 2019 2020 2021 2022 d o c u m e n t s year annual scientific production documents ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 212 figure 1 shows the state of publishing from our sample in the research area on fintech/bigtech credit and economic growth. there has been an exponential increase from 2019 to 2022 the timeframe the data was analyzed. this shows that the nexus between fintech/bigtech credit and its potential impact on economic growth has become more relevant in recent years. this could be as a result of covid-19 that escalated the need to be more reliant on technology, therefore making users comfortable and knowledgeable of its use. in addition, the world economy has been under tremendous strain experiencing stunted growth, therefore it has become crucial to establish and develop creative ways that can provide solutions. 4.3. most influential authors from sample of the study table 1 illustrates the impact of authors based on the total citations. these authors act as respected reputable voices around area of study. a list of 6 most prolific authors based on our sample in the research area is presented with details of their first year of publishing, total published papers, h-index (a metric that evaluates author’s research performance based on quantity and quality of work), total citations, current affiliation and country. table 1. influential authors from sample of study. source: authors’ own compilation (data sourced from scopus). 4.4. most influential documents from sample of the study table 2 is arguably the most important as it helps to identify areas for future research. literature that speaks to the connections relevant to the topic under the research study that paves a way for future research exposing the research gaps. scope for future research can be retrieved from two of the most influential documents from our sample with the highest field-weighted citation impact (fwci) figures. the first document published in 2022 with a fwci score of 13,92 by allen, f., gu, x., and jagtiani, j identified challenges in regulation of technology as a future area of study (allen, gu & jagtiani, 2022). this is in line with emphasis on the importance of regulatory supervision to protect customers and maintain financial stability when enjoying opportunities of financial innovation as stated by frost j. with the third highest fwci score of 7,13 in 2020 (frost, 2020). the other influential document by hodula, m., with an fwci score of 10,37 identified the areas for future research to emanate from market inefficiencies such as higher interest margins, slow loan application procedures that may lead to room for fintech credit (hodula, 2022). author scopus author id year of 1st publication tp h-index tc current affiliation country allen franklin 7102515412 1982 122 47 12458 imperial college london united kingdom huang yiping 55716988300 1995 96 18 1371 peking university china le tu dq 57195804943 2017 32 9 263 viet nam national university viet nam lechman ewa 56499101500 2014 29 8 209 gdansk university of technology poland marszk adam 56499260300 2014 17 7 130 gdansk university of technology poland nguyen dat t. 57221266939 2020 11 3 25 viet nam national university viet nam tp : total publication tc : total citation list of 6 most prolific authors in research area note: h-index is based upon number of documents and number of citations ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 213 table 2. most cited authors and the article title. 4.5. most influential sources from sample of the study table 3. the highly published journals and their most cited article. author title source year citations fwci croutzet, a., dabbous, a. do fintech trigger renewable energy use? evidence from oecd countries renewable energy 179, pp. 1608-1617 2021 15 2.22 hua, x, huang, y understanding china's fintech sector: development, impacts and risks european journal of finance 27(4-5), pp. 321333 2021 13 4.75 kanungo, r.p., gupta, s. financial inclusion through digitilisation of services for wellbeing technological forecasting and social change 2021 12 2.76 allen, f., gu, x., jagtiani, j. fintech, cryptocurrencies, and cbdc: financial structural transformation in china journal of international money and finance 2022 10 13.92 jiao, z., shahid, m.s., mirza, n., tan, z. should the fourth industrial revolution be widespread or confined geographically? a country-level analysis of fintech economies technological forecasting and social change 2021 9 2.21 hodula, m. does fintech credit substitute for traditional credit? evidence from 78 countries finance research letters 2022 8 10.37 yeo, e., jun, j. peer-to-peer lending and bank risks: a closer look sustainability (switzerland) 12(15) 2020 8 0.82 petralia, k. philippon, t., rice, t., veron, n. banking disrupted?: financial intermediation in an era of transformational technology geneva reports on the world economy 2019(22) 2019 7 1.08 lechman, e. marszk, a ict-driven economic and financial development: analyses of european countries ict-driven economic and financial development: analyses of european countries pp. 1-311 2019 7 1.95 tritto, a., he, y., junaedi, v.a. governing the gold rushinto emerging markets: a case study of indonesia's regulatory responses to the expansion of chinese-backed online p2p lending financial innovation 6(1), 51 2020 6 0.97 allen, f., qian, j., qian, m a review of china's institutions annual review of financial economics 11, pp. 39-64 2019 6 0.54 nguyen, l., tran, s., ho, t. fintech credit, bank regulations and bank performance: a cross-country analysis asia-pacific journal of business administration 2021 5 1.69 frost, j the economic forces driving fintech adoption across countries (book chapter) the technological revolution in financial services: how banks, fintechs, and customers win together pp. 70-89 2020 5 7.13 marszk, a., lechman, e., kato, y. the emergence of etfs in asia-pacific (book) the emergence of etfs in asia-pacific pp. 124 2019 5 1.39 fwci : field-weighted citation impact (how well cited document is when compared to similar documents journal tp tc citescore sept 2022 the most cited article times cited publisher sustainability switzerland 11 416 206 310 5.1 the impact of financial development and fdi on renewable energy in the uae: a path towards sustainable development 32 multidisciplinary digital publishing institute technological forecasting and social change 734 29 960 13.6 investigating the spill overs and connectedness between financial globalization, high-tech industries and environmental footprints: fresh eidence in context of china 69 elsevier accounting economics and law a convivium 15 100 1.6 unreliable accounts: how regulators fabricate conceptual narrative to diffuse criticism 2 walter de gruyter annual review of financial economics 19 154 3.4 climate finance 22 annual reviews inc applied economics letters 646 2 536 2.1 is there a pattern in how covid-19 has affected australia's stock returns 36 taylor & francis the 5 most productive journals with their most cited article tp: total publications tc: total citations ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 214 table 3 below shows the 5 most productive journals from our sample with their most cited article and publisher. the cite score as of 5 september 2022 is also provided based on scopus updates. details of total publications and total citations up to the period in september is also presented for each journal. 4.6. network analysis and visualisation the following section is based on network analysis and visualization. citation analysis of documents, co-citation analysis of authors and occurrence analysis of keywords are conducted using vosviewer application to visualize the research trends in the fintech/ bigtech credit and economic growth research. 4.6.1. citation analysis of documents figure 2 presents the impact of research by measuring the total number of citations of documents (ahmi, tapa & hamza, 2020). it has a total of 49 items and 47 clusters. cluster 1 (red): 2 items comprising of chorzempa m. and hua x. cluster 2 (green): 2 items comprising of hodula m. and stankeviciene j. the remainder of the clusters all have 1 item each. the closer the documents are to each other translates to the relatedness of the documents. size of bubbles illustrate the number of citations of documents. figure 2. citation analysis of documents. 4.6.2. co-citation analysis of authors table 4 groups authors into the 4 different clusters based on contents of the documents and citation’s relationship. figure 3 shows measurement of the similarity of the contents of the documents based on the citation’s relationship. the analysis has been conducted using the vosviewer software for network visualization based on bibliometric data. the minimum number of citations selected of an author is 10. out of 5617 authors, 63 met the threshold. ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 215 table 4. co-citation of authors and the clusters. cluster number of items authors 1 (red) 21 allen f, chen i, gambacorta i, hau h, huang y, jagtiana j, li j, lin c, liu j, qian j, qian m, shan h, shen y, sheng z. troise c, wang j, wang x, wang y, xiong w, zhang x, zhou h. 2 (green) 21 arner d. w, asongu s.a, buckley r.p, freeman c, haddad c, helpman e, hornuf i, katz ml, kauffman r, klapper i, lechman e, li x, madhavan a, marszk a, mokyr j, perez c, pradhan r, rosenberg n, shapiro c, trajtenberg m, zhang y. 3 (blue) 16 becker t, berger a, cihak m, claessens s, demirguc-kunt a, frost j, le td, levine r, liu j, ngo t, nguyen dt, pagano m, philippon t, shahbaz m, thakor av, turner g, zhu f. 4 (yellow) 5 agarwal a, agarwal j, agarwal m, agarwal y, sahay r. figure 3. co-citation analysis of co-authors. 4.6.3. analysis of co-occurrence table 5 shows further analysis using vosviewer software to construct and visualize bibliometric networks. figure 4 illustrates the network visualization of the authors’ keywords produced by vosviewer, showing color, circle size, font size, and thickness of connecting lines as the nexus of strength of the relationship among key words. out of the 1279 terms, 79 meet threshold when occurrence is kept at 5 – 60% default choice for most relevant terms, therefore 47 selected and further removed 7 that were less relevant and remained with 40 items in 6 clusters. ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 216 table 5. occurrence of key words and their clusters. cluster number of items occurrence 1 (red) 9 digital financing, disintermediation, financial innovation, financial literacy, financial service, financial stability, policy, structure, systemic risk 2 (green) 8 banking, digital transformation, efficiency, internet finance, investment, overcapacity, relationship, sustainable development goals (sdgs) 3 (blue) 7 digital finance, fintech start-up, happiness, indonesia, islamic bank, performance, practical implication 4 (yellow) 6 bank performance, bank regulation, bank stability, credit information sharing, fintech credit, income inequality 5 (purple) 5 economy, exchange-traded funds (etfs) market, etfs, financial market, market 6 (light blue) 5 bank risk, charter value, government, p2p lending, risk figure 4. occurrence analysis of keywords. 5. discussion data was collected from the scopus database and thereafter we conducted a bibliometric metaanalysis. this method was effective to unpack how fintech and bigtech credit and economic growth research has evolved from 2019 to 2022. furthermore, the analysis shows most cited articles together with the most influential authors giving awareness of the relevant journals. the results of study documented that financial innovation that comes from digital technology has been on the rise in recent years. many people have embraced the changes coming with machine learning, artificial intelligence and robotics in the fourth industrial revolution. initially fintechs/bigtechs had been viewed as disruptors in the financial industry, however, they seem to be moving more towards collaboration with banks establishing exciting synergies for clients. previous studies do point to key issues around monitoring regulation that is requiring more attention ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 217 due to rampant cyber-attacks that may result as a major setback to fintech adoption (solms & niekerk, 2013). big data and ability to decode useful information where credit risk assessment will be concerned plays a huge factor going forward. table 5 on occurrence of keywords groups a cluster with financial service, financial innovation, digital financing and financial stability. furthermore, fintech credit is linked to addressing income inequality issues, maximizing on credit information sharing and impact closely related to bank performance and regulation. citation analysis of documents on figure 2 however, showed results with no linkages which illustrates that area of study is growing and more work must be done. economic growth and sustainability are topical issues considering many countries are not performing well and there is economic pressure. credit being the driver of the financial industry and technology advancements on the rise leaves fintech/bigtech credit a viable option for future sustainable economies. the results of the study are instrumental on the significance of the study and help establish key areas to focus on based on relevance. knowing keywords and most cited articles together with co-citations helps identify not only gaps in knowledge but gives ability to add value to previous research. 6. conclusions the primary goal was to establish trends over the period 2019 to 2022 and identify research gaps in the area of fintech/bigtech credit and economic growth. the study adopted a bibliometric and meta-analysis by utilizing studies that were extracted from the scopus database. the study reports the trend of the previous studies using the boolean mix of ‘fintech credit” or “bigtech credit” and “economic growth’ as bibliometric indicators obtained from the scopus database. in total, the bibliometric details of 49 documents were extracted from the scopus database. the data was then exported onto vosviewer software and used to provide network analysis and visualization. the results of the study showed an increase in the number of documents relevant to the topic from 2019. it is interesting to note that even though the keywords are connected, there are no strong links of citation between the relevant documents. in addition, few documents show close relatedness from the analysis. establishing the linkage between the related topics on area of study does indeed open doors for future research. even though bibliometric and meta-analysis is a good scientific statistical analysis method, it remains subjective and therefore may affect results. limitations are as a result of time period which may differ coupled with the variety of terms to search for data. in addition, the results of our findings are based on what we obtained from the scopus database, and other authors may have their articles not indexed in scopus and hence excluded from analysis. however, despite the stated limitations, it is important to note that this study presents a holistic overview based on our sample of current research on a global scale of fintech/bigtech credit and economic growth. the scope for future research therefore is regulation that extends to monitoring of the technological innovation and enforcement that is financially inclusive. the other areas for future research include exploring bank concentration and market inefficiencies nexus with fintech credit, and considering the relationship between fintech credit and bank capital regulation. the above ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 218 analysis solidifies the relevance to conduct further research on the connectedness of fintech/bigtech credit and economic growth. contributions: “conceptualization, e.m. and a.s.; methodology, e.m and a.s.; software, e.m.; validation, e.m., and a.s..; formal analysis, e.m.; investigation, e.m.; resources, e.m.; data curation, e.m.; writing—original draft preparation, e.m.; writing—review and editing, a.s.; visualization, e.m.; supervision, a.s.; project administration. all authors have read and agreed to the published version of the manuscript.” funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] ahmad, e., & malik, a. financial sector development and economic growth: an empirical analysis of developing countries, 2009. [2] ahmi, a., tapa, a., & hamzah, a. h. mapping of financial technology (fintech) research: a bibliometric analysis. international journal of advanced science and technology, 2020, 29(8): 379–392. [3] allen, f., gu, x., & jagtiani, j. fintech, cryptocurrencies, and cbdc: financial structural transformation in china. journal of international money and finance, 2022, 124. doi: https://doi.org/10.1016/j.jimonfin.2022.102625. [4] amidu, m., & wolfe, s. does bank competition and diversification lead to greater stability? evidence from emerging markets. journal of advanced research, 2013, 3(3): 152–166. doi: https://doi.org/10.1016/j.rdf.2013.08.002. [5] appiah-otoo, i., & song, n. the impact of ict on economic growth-comparing rich and poor countries. telecommunications policy, 2021, 45(2), 102082. doi: https://doi.org/10.1016/j.telpol.2020.102082. [6] arner, d. w., barberis, j., & buckley, r. p. regtech: building a better financial system. in handbook of blockchain, digital finance, and inclusion, volume 1: cryptocurrency, fintech, insurtech, and regulation, 2017, (1st ed., vol. 1). elsevier inc. doi: https://doi.org/10.1016/b978-0-12-810441-5.00016-6. [7] arner, d. w., zetzsche, d. a., buckley, r. p., & barberis, j. n. fintech and regtech: enabling innovation while preserving financial stability. georgetown journal of international affairs, 2017, 18(3): 47–58. doi: https://doi.org/10.1353/gia.2017.0036. [8] bahrini, r., qaffas, a. a. impact of information and communication technology on economic growth: evidence from developing countries [j]. economies, 2019, 7(1): 21. [9] beck, t. fintech and financial inclusion, 2020. 1165. doi: https://doi.org/10.1007/978-1-349-67278-3_116. [10] berger, a. n., & udell, g. f. the economics of small business ® nance: the roles of private equity and debt markets in the ® nancial growth cycle, 1998. 22. [11] berlin, m. bank credit standards. business review, 2009, 1–10. www.philadelphiafed.org. [12] bertsch, c., hull, i., qi, y., & zhang, x. bank misconduct and online lending. journal of banking and finance, 2020, 116, 105822. doi: https://doi.org/10.1016/j.jbankfin.2020.105822. [13] bouzid, a., amaira, b., & amairya, r. financial intermediation and economic growth in tunisia: an econometric investigation econometric model view project financial liberalization index of tunisia: factorial method approach view project financial intermediation and economic growth in tunisia: an econometric investigation. in international journal of business and behavioral sciences, 2014, (vol. 4, issue 3). https://www.researchgate.net/publication/312164371. [14] boyd, j. h., & de nicoló, g. (n.d.). the theory of bank risk-taking and competition revisited. [15] braun, s. third party payment service providers, 2018. [16] claessens, s., coleman, n., & donnelly, m. “low-for-long” interest rates and banks’ interest margins and profitability: cross-country evidence. journal of financial intermediation, 2018, 35: 1–16. doi: https://doi.org/10.1016/j.jfi.2017.05.004. [17] claessens, s., frost, j., turner, g., & zhu, f. (2018). fintech credit markets around the world: size, drivers and policy issues. https://ssrn.com/abstract=3288096. [18] cornelli, g., frost, j., gambacorta, l., rau, r., wardrop, r., & ziegler, t. fintech and big tech credit: a new database, 2020. 887. ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 219 [19] davis, f. d. perceived usefulness, perceived ease of use, and user acceptance of information technology. mis quarterly, 1989, 13(3): 319–340. doi: https://doi.org/10.5962/bhl.title.33621. [20] dawei, l., anzi, h., & gen, l. big data technology: application and cases. in handbook of blockchain, digital finance, and inclusion, 2018, (1st ed., vol. 2). elsevier inc. doi: https://doi.org/10.1016/b978-0-12812282-2.00004-8. [21] demirgüç-kunt, a., klapper, l., singer, d., ansar, s., & hess, j. the global findex database, 2017. [22] demirgüç-kunt, a., & levine, r. stock markets, corporate finance, and economic growth: an overview. in source: the world bank economic review, 1996, (vol. 10, issue 2). https://www.jstor.org/stable/3990061?seq=1&cid=pdf-. [23] duncan, f. interest rates and inflation what gives? economy [j]. personal finance, 2011, (362): 13-14. [24] dwyer, r. e. credit, debt, and inequality [j]. annual review of sociology, 2018, 44: 237-261. doi: https://doi.org/10.1146/annurev-soc-060116. [25] emara, n., & el, a. financial inclusion and economic growth: the role of governance in selected mena countries. international review of economics and finance, 2021, 75(february): 34–54. doi: https://doi.org/10.1016/j.iref.2021.03.014. [26] evans, o., & adeoye, b. munich personal repec archive determinants of financial inclusion in africa: a dynamic panel data approach determinants of financial inclusion in africa: a dynamic panel data approach, 2016. 43 utc university of mauritius research journal, 22(81326). [27] fasano, f., & cappa, f. how do banking fintech services affect sme debt? journal of economics and business, 2022. doi: https://doi.org/10.1016/j.jeconbus.2022.106070. [28] fetscherin, m., & heinrich, d. consumer brand relationships research: a bibliometric citation metaanalysis. journal of business research, 2015, 68(2): 380–390. doi: https://doi.org/10.1016/j.jbusres.2014.06.010. [29] fischer, s. the role of macroeconomic factors in growth [j]. journal of monetary economics, 1993, 32(3): 485-512. [30] freimanis, k., & šenfelde, m. credit creation theory and financial intermediation theory: different insights on banks’ operations, 2019, may 13. doi: https://doi.org/10.3846/cibmee.2019.033. [31] frost, j. the economic forces driving fintech adoption across countries, 2020. [32] frost, j., & turner, g. fintech credit markets around the world: size, drivers. 2018, september, 29–49. [33] gąsiorkiewicz, l., monkiewicz, j., & monkiewicz, m. technology-driven innovations in financial services: the rise of alternative finance. foundations of management, 2020, 12(1): 137–150. doi: https://doi.org/10.2478/fman-2020-0011. [34] ghosh, s. journal of behavioral and experimental finance how important is trust in driving financial inclusion? journal of behavioral and experimental finance, 2021, 30, 100510. doi: https://doi.org/10.1016/j.jbef.2021.100510. [35] girdzijauskas, s., streimikiene, d., griesiene, i., mikalauskiene, a., & kyriakopoulos, g. l. new approach to inflation phenomena to ensure sustainable economic growth. sustainability (switzerland), 2022, 14(1): 1–21. doi: https://doi.org/10.3390/su14010518. [36] gomber, p., kauffman, r. j., parker, c., & weber, b. w. on the fintech revolution: interpreting the forces of innovation, disruption, and transformation in financial services. journal of management information systems, 2018, 35(1): 220–265. doi: https://doi.org/10.1080/07421222.2018.1440766. [37] gros, d., & alcidi, c. the global economy in 2030: trends and strategies for europe. ceps (paperbacks), european union, 2013. http://ssrn.com/abstract=2422275. [38] hodula, m. does fintech credit substitute for traditional credit? evidence from 78 countries. finance research letters, 2022, 46. doi: https://doi.org/10.1016/j.frl.2021.102469. [39] hunt, s. d. understanding the drivers of economic growth: grounding endogenous economic growth models in resource-advantage theory, 2012, 6(4): 8–13. doi: https://doi.org/10.5709/ce.1897-9254.62. [40] hussain, a., batool, i., akbar, m., & nazir, m. (2021). is ict an enduring driver of economic growth? evidence from south asian economies. telecommunications policy, 45(8), 102202. doi: https://doi.org/10.1016/j.telpol.2021.102202. [41] indangasi, j. the relationship between economic growth and stock market performance in kenya. dba africa management review, 2017, 1(1): 27–41. [42] jagtiani, j., & john, k. fintech: the impact on consumers and regulatory responses. journal of economics and business, 2018, 100: 1–6. doi: https://doi.org/10.1016/j.jeconbus.2018.11.002. ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 220 [43] kaluwa, b., & chirwa, g. c. competition, regulation and banking industry pricing conduct in malawi, 2017, 9(2): 2042–1478. [44] keane, s. f., cormican, k. t., & sheahan, j. n. comparing how entrepreneurs and managers represent the elements of the business model canvas. journal of business venturing insights, 9(october 2017): 65–74. doi: https://doi.org/10.1016/j.jbvi.2018.02.004. [45] kebede, j., naranpanawa, a., & selvanathan, s. financial inclusion: measures and applications to africa ✩. economic analysis and policy, 2021, 70: 365–379. doi: https://doi.org/10.1016/j.eap.2021.03.008. [46] khanh, t. van, & loc, t. b. fintech credit — opportunities for smes in vietnam, 2018, 9(6): 537–542. doi: https://doi.org/10.15341/jbe(2155-7950)/06.09.2018/008. [47] kpmg. pulse of fintech, 2021, february. [48] lee, i., & shin, y. j. fintech: ecosystem, business models, investment decisions, and challenges. business horizons, 2018, 61(1): 35–46. doi: https://doi.org/10.1016/j.bushor.2017.09.003. [49] liu, y., luan, l., wu, w., zhang, z., & hsu, y. can digital financial inclusion promote china’s economic growth? international review of financial analysis, 78(july 2020), 101889. doi: https://doi.org/10.1016/j.irfa.2021.101889. [50] mishi, sibanda, & tsegaye. industry concentration and risk taking: evidence from the south african banking sector. african review of economics & finance, 2016, 8(2): 112–135. [51] munangi, e., & sibindi, a. b. an empirical analysis of the impact of credit risk on the financial performance, 2020, 24(3): 1–15. [52] musabegovic, i., özer m, djukovic s, et al. influence of financial technology (fintech) on financial industry[j]. економика пољопривреде, 2019, 66(4): 1003–1021. doi: https://doi.org/10.5937/ekopolj1904003m. [53] nilsson, p., & fredholm, l. competitors or collaborators? a qualitative study of the evolving relationship between banks and fintech companies in the swedish retail banking industry, 2020, june. https://lup.lub.lu.se/student-papers/record/9018931. [54] njoki, k. the relationship between interest rates and mortgage default rate among financial institutions in kenya (doctoral dissertation), 2014. [55] nyaruwata, g. addressing over-indebtedness in south africa: what role should supply-side and demandside interventions play [j]. unpublished paper. university of cape town. cape town, 2009. [56] ozili, p. k. impact of digital finance on financial inclusion and stability. borsa istanbul review, 2018, 18(4): 329–340. doi: https://doi.org/10.1016/j.bir.2017.12.003. [57] petersen, m. a., rajan, r. g., berger, a., chevalier, j., dunham, c., flannery, m., gibson, m., gron, a., hart, o., kaplan, s., mcdonald, r., pennacchi, g., prendergast, c., repullo, r., welch, n., white, l., wolken, j., & sebastian, s. the effect of credit market competition on lending relationships* and the utah winter finance conference for their insights, 1995. https://academic.oup.com/qje/article/110/2/407/1826345. [58] pradhan, r. p., arvin, m. b., nair, m. s., hall, j. h., & bennett, s. e. sustainable economic development in india: the dynamics between financial inclusion, ict development, and economic growth. technological forecasting and social change, 2021, 169(april), 120758. doi: https://doi.org/10.1016/j.techfore.2021.120758. [59] prisecaru, p. challenges of the fourth industrial revolution. terra economicus, 2016, 8(1): 57–62. doi: https://doi.org/10.23683/2073-6606-2019-17-2-6-22. [60] rahman, m. m., & alam, k. exploring the driving factors of economic growth in the world’s largest economies. heliyon, 2021, 7(5): 1–9. doi: https://doi.org/10.1016/j.heliyon.2021.e07109. [61] restoy, b. f. regulating fintech: what is going on, and where are the challenges? 2020, 134: 1–9. [62] ritter, j. in this issue: growth, value, and payout policy. journal of applied corporate finance, 2012, 24(3): 1–4. [63] rode, s. advanced macroeconomics. advanced macroeconomics, 2020. doi: https://doi.org/10.1142/q0267. [64] shamim, s., zeng, j., shariq, s. m., & khan, z. role of big data management in enhancing big data decisionmaking capability and quality among chinese firms: a dynamic capabilities view. information and management, 2019, 56(6), 103135. doi: https://doi.org/10.1016/j.im.2018.12.003. [65] sharma, r., & mishra, r. a review of evolution of theories and models of technology adoption. 2014, 6(2): 1–23. [66] sitea, d. m. the fintech revolution. banking for the new generation. 2020, may. ephias munangi and athenia bongani sibindi / journal of risk analysis and crisis response, 2022, 12(4), 204-221 doi: https://doi.org/10.54560/jracr.v12i4.344 221 [67] solms, r., & niekerk, j. from information security to cyber security. computers and security, 2013, 38: 97102. doi: http://dx.doi.org/10.1016/j.cose.2013.04.004. [68] starkey, m. credit creation theory of banking, 2018. http://shura.shu.ac.uk/23013/. [69] sufian, f., & habibullah, m. globalization and bank efficiency nexus. 2012, 139–155. [70] suprun, a., petrishina, t., & vasylchuk, i. competition and cooperation between fintech companies 2020. 13028. [71] tabak, b. m., gomes, g. m., & da silva medeiros jr, m. the impact of market power at bank level in risktaking: the brazilian case [j]. international review of financial analysis, 2015, 40: 154-165. [72] turk ariss, r. on the implications of market power in banking: evidence from developing countries. journal of banking and finance, 2010, 34(4): 765–775. doi: https://doi.org/10.1016/j.jbankfin.2009.09.004. [73] turlais, v. (2016). business scenario planning for declining industry. journal of business management, 11, 14-. [74] úbeda, f., forcadell, f. j., aracil, e., & mendez, a. how sustainable banking fosters the sdg 10 in weak institutional environments. journal of business research, 2022, 146: 277–287. doi: https://doi.org/10.1016/j.jbusres.2022.03.065. [75] vazakidis, a., & adamopoulos, a. stock market development and economic growth. american journal of applied sciences, 2009, 6(11): 1933–1941. doi: https://doi.org/10.3844/ajassp.2009.1932.1940. [76] vergara, c. c., & agudo, l. f. fintech and sustainability: do they affect each other? in sustainability (switzerland), 2021, (vol. 13, issue 13). mdpi ag. doi: https://doi.org/10.3390/su13137012. [77] wambui, n. s. the effect of interest rate volatility on mortgage default rate in kenya. international journal of business, humanities and technology, 2013, 3(1). [78] yengeni, s. myth or magic: the impact of financial technology on financial inclusion in africa, 2020. [79] zhang, y., & lee, w. intrusion detection techniques for mobile wireless networks, 2003, 1–16. copyright © 2022 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word volume 13, issue 1-2 journal of risk analysis and crisis response, 2023, 13(1), 19-34 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v13i1.353 19 article the sociotechnical construction of risks, and principles of the proactive approach to safety washington barbosa 1,2,*, luiz ricardo moreira 2, gilson brito 3, assed n. haddad 4 and mario cesar vidal 2 1 oswaldo cruz foundation, rio de janeiro (21040360), rio de janeiro, brasil 2 production engineering program, federal university of rio de janeiro, rio de janeiro (21941914), rio de janeiro, brasil 3 production engineering program, federal university fluminense, niterói (24210240), rio de janeiro, brasil 4 environmental engineering program, federal university of rio de janeiro, rio de janeiro (21941909), rio de janeiro, brasil * correspondence: washington.fiocruz@gmail.com received: december 17, 2022; accepted: february 17, 2023; published: march 31, 2023 abstract: this proposal presents the sociotechnical construction of risks, ergonomics, and the two principles of the proactive approach to safety, risks, and emergencies, the structured sociotechnical approach and dynamics of proactive safety intending to complement traditional risk assessments, and prevent and mitigating major and fatal negative events, the in organizations such as cases of the explosion of the space shuttle challenger, the nuclear accident in fukushima, the texas city refinery and the explosion in the port of beirut, among others. to propose these two principles, case studies were developed at fiocruz, and in organizations, sectors, and activities, a bibliographic review on theses, dissertations, reports from regulatory bodies, books, scientific articles, and media articles, on major and fatal negative events, and ergonomics, socio-technical approach, and resilience engineering. a tragedy prevention course was created, with four free online consultation modules, based on cases of major negative events. these principles redirect the focus from human error to focus on the structured sociotechnical system and focus on the dynamics of proactive safety. it is proposed that these two principles can provide us with bases for analysis, to prevent and minimize major and fatal negative events, and are a complement to traditional risk assessments. keywords: safety; risks; ergonomics; sociotechnical construction of risks; proactive approach to safety 1. introduction we live in a dynamic and complex environment, safety management is an important tool to manage this environment. it is recommended that organizations that seek to achieve their goals incorporate security management throughout their life and activities, including strategies, decisions, operations, processes, functions, projects, products, services, and assets (dekker, 2006 [1]; figueiredo, 2018 [2]; filho, 2021 [3]; furuta, 2015 [4]; hollnagel, 2019 [5]; hopkins, 2005 [6], 2008 [7]; levenson, 2020 [8]; llory, 2014 [9]; pidgeon, 2000 [10]; perrow, 1999 [11]; rasmussen, 2000 [12]; reason, 2016 [13]; turner, 1997 [14]; vaughan, 1996 [15]). safety management can be divided into two auxiliary functions: risks and emergencies. the first aims to control latent factors and the second is the manifestations of risks in real facts. therefore, washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 20 there are two complementary forms of action: preventive and corrective, and the proactive security proposal seeks to prevent the organization from acting only in a reactive way. the use of risk management, risk assessment, and risk analysis emerged more or less independently in several areas: nuclear industry, insurance, oil industry, safety at work, corporate security, financial systems, information security, and security of products and processes. the word risk is used in many areas and with different meanings, such as in mathematics, economics, engineering, and the field of public health. safety is a state of low probability of occurrence of events that cause damage or loss. the term safety culture was conceptualized for the first time in the technical report on the accident at the chernobyl nuclear power plant in ukraine, in the 1980s, as being: “set of characteristics and attitudes of organizations and individuals, which guarantee that the safety of a nuclear plant, due to its importance, will have the highest priority”. accident is defined as: “an undesirable event that results in death, health problems, injuries, damages and other losses”. near-miss is defined as: “an unforeseen event that had the potential to cause accidents”. this definition is intended to include all occurrences that do not result in death, ill health, injury, harm and other benefits. the term “incident” cited is defined as: “an unsafe occurrence arising from or in the course of work, in which no personal injury is generated”. this term was added to include all occurrences that generated only material damage and near-accidents in the organizations' focus of action. despite the efforts made by companies, organizations, private sectors, and the government, a series of major and fatal negative events have happened, such as the explosion of the space shuttle challenger, the nuclear accident in fukushima, and the explosion in port of beirut, among others (barbosa, 2022) [16]. turner (1994) [17] analyzed serious technical accidents over a long period and concluded that approximately 20 to 30% of the causes of accidents were technical, with 70 to 80% involving social, administrative, or managerial factors. a series of studies on air and maritime accidents in qureshi (2008) [18] showed human and organizational factors as the main contributors to accidents and incidents. an analysis of major air and maritime accidents in north america during 1996-2006 concluded that the proportion of causal and contributing factors related to organizational issues exceeds those due to human error. for example, the combined causal and contributory factors of aviation accidents in the us showed: 48% related to organizational factors, 37% to human factors, 12% to equipment, and 3% to other causes; and the analysis of maritime accidents classified causal and contributory factors as 53% due to organizational factors, 24-29% as human error, 10-19% for equipment failures and 2-4% as other causes. why do negative events happen? these complex events require both a socio-technical approach and a working conceptualization of these systems. according to llory (2014) [9], however diverse the causes of these accidents are, they all have an organizational dimension, that is, their root causes must be sought to verify what caused the accident. they also confirm that the non-occurrence of serious accidents and good performances in everyday life can hide an important issue, as a catastrophe may be about to happen. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 21 in this way, the objective of the research can be presented as follows: principles can be developed, with analysis of these accidents and case studies, according to barbosa (2022) [16], in search of factors and variables, which present proposals for the prevention and minimization of these accidents, which happen repeatedly, and that can be transmitted to the organizations. initially, a case study was developed, conducted by the author that originated a monograph of the specialization course in ergonomics: "ergonomic analysis of risk management of residues of dangerous products from fiocruz" and an article by the author on the "contribution of ergonomics to the development of proactive safety, risks and emergencies of waste from fiocruz dangerous products”, presented in the panel of articles approved at the abergo 2020 congress (barbosa, 2020) [19]. as a continuation of this research, an in-depth literature review was carried out on theses, dissertations, reports from regulatory bodies, books, scientific articles, and media articles, on major and fatal negative events, ergonomics, socio-technical approaches, and resilience engineering. the initial cases presented in barbosa (2022) [16] were selected, and the research continued in units, sectors, and services at fiocruz and in organizations. regarding fiocruz, as the author is an employee of fiocruz, he can carry out several visits to these places, and talk to the management, department heads, researchers, engineers, architects, and technicians in the areas of research, infrastructure, and management, about the other organizations, confidentiality was requested. this work began in 2016, with the evaluation of the management of fiocruz's hazardous products, and has continued in the research laboratories and fiocruz units, in other teaching and research institutions, and in other organizations, until the date of presentation of this work because one of the author's main activities is the safety assessment of the facilities and services provided by organizations. a tragedy prevention course was also created, with four free online consultation modules, in a blog by the author, with a base of cases of major negative events that are hosted in module three (barbosa, 2022) [16]. 2. literature review traditionally, in the analysis of negative and fatal events, the blame is directed towards workers, who are the most fragile elements of the companies' chains of command, and there is little analysis of the activities performed by workers, and their consequences in procedures and adequate working conditions, supervision and management of activities, investments in the maintenance of facilities, analysis, and adaptation of projects, company policies, remuneration bonuses for directors and managers, social and economic requirements, and analysis of the legislation applied to the activity, among other issues. safety management researchers have focused on this topic in recent decades and have presented their proposals for analyzing the factors that give rise to these negative events. 2.1. evolution of the periods of the analysis of negative and fatal events according to dechy (2011) [20], we can present the evolution of these periods: technical period until the 1970s: the source of problems is seen as technology; security was primarily based on technical reliability. period of “human error” in the 1980s: the source of the problem is seen as the person in particular the operators after the three mile island accident in 1979; allowed improvements in the washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 22 domains of the human-machine interface, design of operational procedures, training, among other activities. socio-technical period in the nineties: after bhopal (1984), challenger and chornobyl (1986) the source of the problem is seen as the interaction between the social and the technical subsystems; furthermore, the concept of “safety culture” emerged after the chornobyl accident. interorganizational relationship period from the 2000s: the source of the problem is dysfunctional relationships between organizations, with the controlling role of authorities, subcontractors, competitors, and other departments within an organization. the results of this evolution are cumulative, not exclusive, and none of these dimensions should be neglected when analyzing an event, as they all provide useful information for the world to understand the dynamics that gave rise to the accident. 2.2. theories and research related to the analysis of accidents and management of organizations we highlight theories and research related to the analysis of accidents and management of organizations, which have worked with the history of safety and also contribute to the work presented in this article. according to hollnagel (2006) [21], we need to have the etiology of accidents, a study of possible causes or origins of accidents, we also need to have a safety etiology – more specifically what safety is and how it can be in danger. this is essential for system safety work in general and resilience engineering in particular. however, for reasons that are not entirely clear, development is lacking. the different perceptions of the accident phenomenon are what in current terminology are called accident models. accident models appear to have started with a relatively single factor from simple models, and developed via simple and complex linear causality models to present-day systemic or functional models. greenwood and woods [22], presented in 1919 the theory that proposed an individual propensity of workers to accidents at work, in 1931, heinrich [23] proposed another theory in which a sequence of factors can cause the accident, in a linear sequence of falling domino pieces, aligned side by side, in which the fall of one piece triggers the fall of the other pieces on the side, in a linear sequence of events, called the domino theory. it is a linear cause-and-effect model. in this theory, it was argued that it would be possible to avoid the accident, even after the first domino piece had fallen if one of the stones in the sequence was removed. heinrich states that about 88% of accidents are due to unsafe acts, 10% to dangerous conditions, and 2% to fortuitous situations, this perspective remains one of the preponderant theories in the area of safety in organizations. turner (1978) [24] analyzed 84 accidents and disasters in all sectors, presenting the idea that social, technical, and administrative interactions systematically produced disasters; and developed the concept of accident incubation, with a six-stage development sequence: 1. normal state, initially accepted beliefs about the world and dangers. precautionary norms in laws, codes of practice, or traditional customs. 2. incubation period, accumulation of a set of unnoticed events at odds with accepted beliefs about hazards and norms for controlling them. 3. precipitating event, disaster begins, general perception changes, surprise, and disturbances occur. 4. events escalate, consequences become apparent, and collapse occurs. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 23 5. rescue and rescue. 6. complete cultural readjustment. investigation. beliefs and norms of precaution are adjusted to suit the newly acquired understanding of the world (“this must never happen again”). perrow (1984) [25] analyzed large-scale accidents, which are a problem for society. according to perrow, high-risk organizations with complex technological systems have structural properties that make these large-scale accidents impossible to predict and avoid. for this reason, in these complex systems, accidents are considered “normal” events, and on this basis, he named the theory of normal accidents, where he concludes that these accidents will repeat themselves, and suggested that some of these systems should be eliminated, due to risks of the occurrence of these accidents, the interaction of multiple failures stands out in these normal accidents, whose operational sequence is not direct. the difficulty in anticipating these situations. it is due to the infinite number of possible interactions between failures in the various components of complex systems. reason’s model (1997) [26], known as “swiss cheese” or the theory of multiple causes, does not defend a single cause as the trigger for a sequence of events that would lead to the accident, but linear combinations of latent conditions and active failures that constitute several chains. and, after overcoming safety barriers by aligning their vulnerabilities, they culminate in an accident. in this theory, the influence of the organization in the occurrence of accidents stands out. thus, investigations must look for latent conditions that may induce situations conducive to active failures. thus, the most effective prevention should identify hazards or threats and manage the risks. rasmussen (1997) [27] developed the accimap, which focuses on failure analysis at the following six organizational levels: government policy and budget; regulatory bodies and associations; company planning and budgeting; technical and operational management; physical processes and activities; and equipment, it is a proposal with a generic approach and does not use failure taxonomies at different levels of analysis. leveson's stamp model (2004) [28] is based on levels of control of the socio-technical system. according to the theory behind stamp, accidents occur due to the violation of the conditions in which the system was designed, to support the identification of violations, a taxonomy of control failures is proposed. fram (hollnagel, 2004 [29], 2012 [30]) is a method that aims to understand how systems work and how variability propagates between their functions, to develop more resilient systems. using this model can identify conditions that can lead to accidents in four steps: • identify and characterize the essential functions of the system, for example, based on the six connectors described; • characterize the variability potential of these connectors; • define functional resonance based on identified dependencies between functions; • identify barriers to variability (reduction factors) and specify required performance monitoring. table 1 is a theoretical framework of contributions. 2.3. social construction of risk it must be accepted that the risk is derived from the organization, through its decision-making processes at the strategic, tactical, and operational levels, that is, the risk is a technically constructed partner, according to dechy (2011) [20], figueiredo (2018) [2], filho (2021) [3], hollnagel (2019) [5], washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 24 hopkins (2008) [7], le coze (2013) [31], levenson (2020) [8], llory (2014) [9], pidgeon (2000) [10], perrow (1999) [11], rasmussen (2000) [12], reason (2016) [13], turner (1997) [14], vaughan (1996) [15]. and to evaluate it, adequate qualitative methods are needed for the socio-technical question. it is necessary to go beyond the analysis of human and technical factors, compliance with legislation, and good practices to improve risk management. these questions are important and basic for understanding risk management and for preventing major and fatal negative events. table 1. accident model theoretical framework (self-elaboration). author year contribution to safety spatial, school, georeferenced major contribution to proactive safety greenwood & woods 1919 theory about the existence of individual workers' propensity, sought to explain the causality of accidents at work. united states historical view. heinrich 1931 theory in which the accident originates in a linear sequence of events, which he called the domino theory. united states beginning of a more technical analysis, and based on negative events. turner 1978 accident incubation concept, and a six-stage development sequence. england dynamic risk management concept, and based on a series of case studies of major negative events. perrow 1984 normal accident theory. united states social construction of risk, and that accidents are inevitable, as alignment of its causes is unique and not repeatable. reason 1997 swiss cheese model” or the theory of multiple causes, does not defend a single cause as triggering a sequence of events that would lead to the accident, but linear combinations of latent conditions and active failures that constitute several chains and, after overcoming safety barriers by the alignment of their vulnerabilities, culminate in the accident. england evolution of domino theory and concepts of safety barriers. rasmussen 1997 accimap model, which focuses on failure analysis at the six organizational levels. denmark the concept of performance levels in risk management evolves to the proposal of exogenous and endogenous variables. leveson 2004 the stamp theory is that accidents occur due to the violation of the conditions in which the system was designed. united states evolution of the accimap model. hollnagel 2004 fram is a method that aims to understand how systems work and how variability propagates between their functions, aiming to develop more resilient systems. denmark the complexity of systems, but it is important to seek the representation of complexity, so i present the proposal of the 2 models and principles of proactive security. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 25 3. methodology the principles of proactive safety, risks, and emergencies are developed through two models. 3.1. analysis of modeling study on the elements of the general organization of work in the organization, which were based on the study of major and fatal negative events presented in barbosa, 2020 [19] and 2022 [16], to know and initiate an analysis, through modeling. 3.1.1. structured sociotechnical approach the socio-technical approach is divided into organizational, human, and technological factors, which i define as endogenous variables. as a contribution to this proposal for a sociotechnical approach, i present, based on the case studies research for this work, major and fatal negative events, at the international, national, and local levels (barbosa, 2022) [16] and the accimap model rasmussen (1997) [27], a proposition of the structured sociotechnical approach, where they are included in this analysis are the contributors: social, economic and other requirements; norms and legislation at the world, country, state, municipality and sector levels, which i define as exogenous variables. as a result of the interaction between exogenous and endogenous variables, positive and negative events will occur, which will be shown in figure 1. figure 1. structural sociotechnical approach (self-elaboration). the exogenous variables are the contributors to the event, external to the organization, a possible classification of level can be at the world, country, state, municipality, sectors, and others, as examples, we can highlight international, national, sectorial, state, municipal standards of security, the economic requirements of recession and economic growth, events of nature, and other variables, which were not verified in the case studies analyzed in barbosa (2022) [16], such as terrorism, sabotage, theft, and vandalism, among others, present in other unanalyzed negative events. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 26 the endogenous variables are the organizational, human and technological factors. the organizational factors are related to the actions of the senior management, administrative council, management, senior management, and advisory/staff, these functions are in the corporate instance, as an example of actions of this factor are: the definition of investments, corporate procedures, and the decisions that affect the area of operations of the organization, pressures for profitability, continuity, and discontinuity of the business. organizational factors are constitutive elements for human and technological factors issues, an adequate analysis of the organization's risks and emergencies is of vital importance for the prevention of major negative events, and for the success and continuity of the organization's operations. the human factors are related to the actions of technicians, supervisors, and middle management who work in the operation of the company's activity; as an example of a hierarchical level we can exemplify the case of an oil rig manager, director of a mining company's site and a supervisor of a manufacturing line; cases related to fatigue, stress, and pressure for results are issued to be analyzed in this factor. the technological factors are related to the entire infrastructure for the company's operation, they are the machines, equipment, software, and production and support facilities; equipment failures are related to this factor. human error is the tip of the iceberg, it is what initially appears in major and fatal negative events, it is important to understand the relevance of exogenous and endogenous variables in the systemically structured socio-technical system. “focus on the structured sociotechnical system and not on human error”. first principle of proactive security. 3.1.2. dynamics of proactive safety to present a dynamic model for safety management, the following model is proposed, shown in figure 2, as an adaptation of the boundaries defined by rasmussen (1997) [27], separating the activity to be analyzed into three areas: area of normality place where the organization must be positioned; occurrence of nonconformities without criticality for a major or fatal negative event; danger area occurrence of non-conformities that are critical for a major or fatal negative event, but which have not yet led to the accident. area of action of the company's management systems, normality must be sought, diagnoses must be developed to seek endogenous and exogenous variables, which may have led to this dangerous area, and through planning, minimize the possibility of recurrence of these issues; accident area apply the emergency and mitigation plans, to seek a return to the area of normality, as in the diagnosis of incidents in accidents, the endogenous and exogenous variables that may have led to the incident must be sought, and through planning to minimize the possibility of accidents reoccurring. the model presents us with an arrow with increased risk, due to social and economic pressures, for profitability, achievement of goals, granting financial bonuses, increased workload, and others, which threaten acceptable limits, for safe and good performance of activities, leading the organization to incidents and accidents. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 27 figure 2. proactive safety dynamics model (self-elaboration). it is important to understand systemically the dynamics of safety management. “focus on the dynamics of proactive safety and not on human error”. second principle of proactive security. 4. results based on the research carried out, the following cases were selected, which represent negative events relevant to the research. next, we will present the accidents of fukushima, challenger, and port of beirut. 4.1. nuclear accident in fukushima the fukushima nuclear accident was a nuclear disaster that occurred at the fukushima nuclear power plant on march 11, 2011, caused by the meltdown of three of the plant's six nuclear reactors (see figure 3). a 9.0 mw earthquake occurred at 2:46 pm on friday, march 11, 2011, with the epicenter near honshu, japan's largest island. according to hollnaghel (2013) [32], immediately after the earthquake, all the nuclear reactors in operation at the fukushima plant, three of the six, were successfully turned off, but soon after that the external power was lost because the electrical line was shorted, the electrical panel and the transformer went out of order, and a power transmission tower was brought down by the earthquake. after the loss of external electricity supply, the emergency standby diesel generators were successfully started, but approximately fifty minutes after the earthquake, the tsunami hit the unit, with the wave reaching fourteen to fifteen meters at the perimeter of the plant, the waves broke the ten meters wall of the plant. as the emergency backup generators were located underground, they were flooded with seawater, and electrical equipment, pumps, and fuel tanks were washed away or damaged, as a result, the plant suffered a total loss of electrical power. the immediate consequence of the loss of electrical energy was the core melting in reactors one, two, and three, which in turn caused the massive release of radioactive materials into the environment, within a few days, of the reactor buildings of reactors 1, 3 and 4 exploded because washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 28 hydrogen that was produced inside the reactor pressure vessels leaked into the buildings and exploded. figure 3. nuclear accident in fukushima (source: https://brasil.elpais.com/internacional/2021-0310/10-anos-de-fukushima-golpe-na-reputacao-de-uma-energia-em-retrocesso.html). the plant began releasing significant amounts of radioactive material on march 12, making it the biggest nuclear disaster since the chornobyl nuclear accident. the area became contaminated by the presence of radioactive material released over it and such exposure caused the site to be continuously irradiated. the fukushima nuclear accident independent investigation commission ruled that the nuclear disaster was "artificial" and that its direct causes were all predictable. the report also found that the plant was unable to withstand the earthquake and tsunami. two employees of tokyo electric power company died from injuries caused by the earthquake and another six received radiation exposure above the acceptable limit for a lifetime. an ongoing intensive cleaning program to decontaminate the affected areas and dismantle the plant will take 30 to 40 years. a barrier in the ground, built in an attempt to prevent further contamination of groundwater, decreased the amount of contaminated water collected. in august 2013, however, a huge amount of radioactive water was detected. there were continuous leaks of contaminated water at the plant and some at sea. factory workers are trying to reduce the leaks through some measures, such as building chemical underground walls, but they still have not significantly improved the situation. 4.2. the challenger case in 1986, 73 seconds after its launch, the space shuttle challenger exploded (see figure 4), it was the first accident of the nasa space shuttle program, and all 7 astronauts died (vaughan, 1996 [15]; reason, 1997 [26], 2016 [13]). washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 29 figure 4. explosion of the challenger (source: https://noticias.r7.com/tecnologia-e-ciencia/acidentecom-onibus-espacial-challenger-ha-30-anos-moldou-nova-geracao-de-espaconaves-29062022). after 6 delays and with the warning that the temperature was below the ideal for launch, made by mission engineers, and that these low temperatures could cause an accident, nasa decided to launch challenger. the o-rings of the space shuttle rockets expand and contract as the temperature varies, and on the day of the accident, the temperature at the nasa space center was below freezing, causing the rings to contract, and with this contraction, there was a leakage of fuel from the rockets, which, upon finding a source of heat, caused the explosion. the issue of o-ring safety dates back to 1977 when engineers at the marshall space flight center repeatedly reported to the solid rocket booster (srb) project manager, george hardy, that the design of the o-rings provided by morton thiokol was unacceptable. hardy never forwarded these suits to thiokol, and the o-rings were accepted in 1980. still, in the space shuttle design phase, mcdonnell douglas reported that a “burn through” near the fuel tank would result in a failure that would make it impossible to abort the mission. the o-rings were then rated criticality 1, meaning their failure would result in the spacecraft being destroyed. evidence of serious erosion of the o-rings was verified as early as the second space shuttle mission, with the spacecraft columbia, by the marshall center. however, contrary to nasa regulations, the marshall center did not report the fact to nasa's senior management, keeping the problem limited to its technical area. in 1985, convinced of the catastrophic potential of the problem, marshall center and thiokol began redesigning the o-rings but did not request a suspension of flights or the use of o-rings. they treated the problem as an acceptable risk. thiokol's management initially supported their engineers' recommendation to postpone the challenger's departure, but in a telephone conversation with a nasa manager, the latter said: "for god's sake thiokol, when do you want challenger to be launched? in april?" (npr, 2016) [33]. nasa's arguments would apparently be that if one o-ring failed, there was a second o-ring. however, nasa's own standards defined that for criticality 1 components, the second element should be redundancy in case of unpredictable failures, and not as a backup of the primary element. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 30 4.3. explosion in the port of beirut on august 4, 2020, around 6:08 pm, an explosion occurred in the port region in beirut, the capital of lebanon, resulting in more than two hundred deaths and more than six thousand injured (see figure 5). hours after the event, the news already reported that the catastrophe had occurred in warehouse 12, where 2,750 tons of pure ammonium nitrate were stored. figure 5. explosion in the port of beirut (self-elaboration). in the explosion at the port of beirut, the lebanese authorities were informed of the risk of storing the 2.7 tons of ammonium nitrate, and the necessary measures were not taken to transfer this material to a suitable storage location that could avoid this tragedy (human rights watch, 2021) [34]. from 2014 to 2020, documents were presented to the authorities of the port of beirut, the prime minister, and the president of lebanon, evidence of the organizational factor as a precursor to this great tragedy in which more than 200 people died and 6 thousand were injured in an explosion in the port of beirut, lebanon, which completed one year on 08/04/2021. storage of ammonium nitrate, without proper port security for years, is what caused the explosion. no member of the government has yet been penalized for the explosion. the ngo human rights watch (2021) [34] accuses the lebanese authorities of criminal negligence. in a 126-page report, the entity documented the numerous violations by politicians and the country's security bodies in the management of this hazardous materials warehouse. 5. discussion parameterization and highlights, based on the proactive safety framework, in the case studies. 5.1. in the case of fukushima washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 31 the authorities responsible for the plant were aware of the possibility of larger waves than those designed to contain flooding of the plant by tsunami waves. a historical study revealed that a large tsunami occurred in the middle of the 9th century, estimated at 869 ad and that a researcher had made a strong recommendation to refurbish the plant in 2006, but the recommendation was reportedly declined on the grounds that the tsunami was hypothetical and because the claimed evidence was not accepted by nuclear industry experts. recommendations from the iaea report (2015) [35] included a few, which specifically address the issue of overconfidence: the assessment of natural hazards needs to be sufficiently conservative. the consideration of primarily historical data in establishing the design basis of nuclear power plants is not sufficient to characterize the risks of extreme natural hazards. even when comprehensive data are available, due to relatively short observation periods, large uncertainties remain in predicting natural disasters. the safety of nuclear plants needs to be reassessed periodically to consider advances in knowledge, and necessary corrective actions or compensatory measures need to be implemented promptly. operations experience programs need to include experience from national and international sources. security improvements identified through operational experience programs need to be implemented promptly. the use of operational experience needs to be evaluated periodically and independently. regarding the structured sociotechnical approach, the following stand out: economic pressures in relation to the need for high investments to adjust the height of the walls may have been a prominent variable for this and other adjustments. in relation to dynamic security management, the following stand out: the recommendation to adjust the height of the wall was made, but there was a lack of planning and execution of actions to address this issue. 5.2. in the case of the challenger the pressure exerted on nasa by society and the government, of 24 launches per year, was not achieved, as they did not even reach 5 per year. in order to ensure that its billionaire budget was maintained, and perhaps increased because despite being reusable, the maintenance of the space shuttle cost millions of dollars with each launch, which were preponderant issues for the erroneous decision to authorize the launch of the space shuttle. space shuttle. after the accident, nasa was prevented from making new missions, while carrying out safety studies and adaptations. it took 3 years for a new launch to be made, and only 22 years later, it sent a civilian into space, not by chance, but another teacher. regarding the structured sociotechnical approach, the following stand out: the social and economic pressures exerted on nasa may have been a prominent variable for the effective decision to launch the rocket. in relation to dynamic security management, the following stand out: the warning was given by the rocket engineers, but it was not accepted in a decision by the nasa directorate and the rocket company. 5.3. in the case of the port of beirut explosion washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 32 in this case, the lebanese authorities were unable to recognize the risk and transfer the ammonium nitrate to a suitable warehouse. around the world, countless numbers including large amounts of the same agricultural fertilizer that detonated in beirut began to appear: in dakar, authorities found 3,000 tons of ammonium nitrate in warehouses, in chennai, port officials admitted they were unsafely storing 800 tons of the chemical, romanian authorities discovered nearly 9,000 tons, including 5,000 tons in a single warehouse. disaster prevention is not just about preventing distributors from improperly storing and transporting large amounts of dangerous goods, it is important to check several issues such as supervision, communication, and preventive maintenance. regarding the structured sociotechnical approach, the following stand out: political and management disorganization may have been a prominent variable, due to the noneffectiveness of adequate storage of ammonium nitrate. in relation to dynamic security management, the following stand out: the alert was made to the authorities, but the necessary adjustments were not made. 6. conclusions from the cases presented of major and fatal negative events, and from the propositions presented in this article, it is suggested that traditional risk assessments need to be reassessed. the assessment of exogenous and endogenous pressures on organizations, the structured socio-technical system, the dynamic management of safety, and the systemic view of safety, provided us with a way to identify contributing factors to these major accidents. in this sense, it is a complement to traditional risk assessments. in risk management, it is important to use the precautionary principle and conservative measures, and when in doubt, re-evaluate and use the opinion of experts, to avoid the major accidents that were described in the cases presented in this article. a decision-making process that prioritizes the production process, achievement of goals, and financial issues, and puts safety in the background, can lead to bigger and more fatal negative events. these two principles of proactive safety, risks, and emergencies are proposed: focus on the structured sociotechnical system and not on human error and focus on proactive security dynamics and not on human error. those principles are a complement to traditional risk assessments and can provide us with bases for analysis, to prevent and minimize these major and fatal negative events. contributions: washington barbosa: conceptualization, methodology, writing—original draft preparation, visualization, investigation; luiz ricardo moreira: visualization, writing—review and editing; gilson brito: conceptualization, writing—review and editing, validation; assed n. haddad: conceptualization, supervision, writing—review and editing, validation; mario cesar vidal: conceptualization, supervision, writing—review and editing, validation. all authors have read and agreed to the published version of the manuscript. funding: this research received no external funding. conflicts of interest: the authors declare no conflict of interest. references [1] dekker, s.w. the field guide to understanding human error. ashgate, 2006. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 33 [2] figueiredo, m.g.; alvarez, d.; adams, r.n. o acidente da plataforma de petróleo p-36 revisitado 15 anos depois: da gestão de situações incidentais e acidentais aos fatores organizacionais. caderno de saúde pública 34 (4), 1–12, 2018. doi: https://doi.org/10.1590/0102-311x00034617. [3] filho, a.p.g.; ferreira, a.m.s.; ramos, m.f.; pinto, a.r.a.p. are we learning from disasters? examining investigation reports from national government bodies. safety science, 2021. doi: https://doi.org/10.1016/j.ssci.2021.105327. [4] furuta, k. "resilience engineering: a new horizon of systems safety". in: ahn, j.; carson, c.; jensen, m.; et al. (eds.), reflections on the fukushima daiichi nuclear accident: toward social-scientific literacy and engineering resilience, part v, chapter 24, new york, usa, springer open, 2015. [5] hollnagel, e. the fram model interpreter fmi: software for fram model analysis. jul. 2019. available at: https://functionalresonance.com/the-fram-model-interpreter.html. accessed: oct. 10, 2021. [6] hopkins, a. safety, culture and risk: the organizational causes of disasters, cch, sydney, australia., 2005. [7] hopkins, a. failure to learn: the bp texas city refinery disaster. cch, sydney, australia, 2008. [8] leveson, n. "safety iii: a systems approach to safety and resilience", mit engineering systems lab, working paper, jul. 2020. available at: http://sunnyday.mit.edu/safety-3.pdf. [9] llory, michel. o acidente e a organização/michel llory e rené montmayeul; tradução de marlene machado zica vianna belo horizonte: fabrefactum, 2014. 192p. [10] pidgeon, n.; o’leary, m. man-made disasters: why technology and organizations (sometimes) fail. saf. sci. 34 (1–3), 15–30, 2000. [11] perrow, c. normal accidents: living with high-risk technologies. princeton university press, 1999. [12] rasmussen, j.; svedung, i. proactive risk management in a dynamic society. 1. ed. karlstad, swedish rescue services agency, 2000. [13] reason, j. organizational accidents revisited. crc press taylor & francis group, 2016. [14] turner, b.a.; pidgeon, n.f. man-made disasters, 2nd edition. butterworthheinemann, london, uk, 1997. [15] vaughan, d. the challenger launch decision: risky technology, culture and deviance at nasa. university of chicago press, chicago, 1996. [16] barbosa, w.r. module 3 case studies of major negative and fatal events internationally and in brazil. proactive management blog, 2022. available at: https://gestaoproativawb.blogspot.com/2022/02/modulo3-estudos-de-casos.html. (in portuguese) [17] turner, b.a. causes of disaster: sloppy management. british journal of management, 5, pp.215-219, 1994. doi: https://doi.org/10.1111/j.1467-8551.1994.tb00172.x. [18] qureshi, zahid h. a review of accident modelling approaches for complex critical sociotechnical systems, 2008. [19] barbosa, w.r. contribuição da ergonomia para o desenvolvimento da segurança proativa, riscos e emergências dos resíduos dos produtos perigosos da fiocruz. congresso da abergo 2020. available at: www.even3.com.br/anais/abergo2020/294483-contribuicao-da-ergonomia-para-odesenvolvimento-da-seguranca-proativa-riscos-e-emergencias-dos-residuosdos-pro. (in portuguese) [20] dechy, n.; et al. learning lessons from accidents with a human and organizational factors perspective: deficiencies and failures of operating experience feedback systems. eurosafe forum 2011. available at: https://www.researchgate.net/publication/233997934_learning_lessons_from_accidents_with_a_human_ and_organisational_factors_perspective_deficiencies_and_failures_of_operating_experience_feedback_sy stems. [21] hollnagel, e.; woods, d.d; leveson, n. resilience engineering: concepts and precepts. 1st. ed. burlington, ashgate, 2006. [22] greenwood, m.; woods, h.m. the incidence of industrial accidents upon individuals with special reference to multiple accidents. industrial fatigue research board, amedical research committee, report no. 4. her britannic majesty's stationary office, london, 1919. [23] heinrich, h. industrial accident prevention. mcgraw-hill, new york, 1931. [24] turner, b.a. man-made disasters, wykeman, london, 1978. [25] perrow, c. normal accidents: living with high-risk technologies. new york: basic books, 1984. [26] reason, j. managing the risk of organisational accidents. ashgate, 1997. [27] rasmussen, j. "risk management in a dynamic society: a modelling problem", safety science, v. 27, n. 2–3, pp. 183–213, 1 nov. 1997. doi: https://doi.org/10.1016/s0925-7535(97)00052-0. washington barbosa, et al. / journal of risk analysis and crisis response, 2023, 13(1), 19-34 doi: https://doi.org/10.54560/jracr.v13i1.353 34 [28] leveson, n. “a new accident model for engineering safer systems,” saf. sci., vol. 42, no. 4, pp. 237-270, 2004. doi: https://doi.org/10.1016/s0925-7535(03)00047-x. [29] hollnagel, e. barriers and accident prevention. 1st. ed. surrey, ashgate, 2004. [30] hollnagel, e. fram the functional resonance analysis method: modeling complex socio-technical systems. farnham, ashgate, 2012. [31] le coze, j.c. what have we learned about learning from accidents? post-disasters reflections. safety science 51 (1), 441–543, 2013. doi: https://doi.org/10.1016/j.ssci.2012.07.007. [32] hollnagel, e.; fujita, y. "the fukushima disaster-systemic failures as the lack of resilience", nuclear engineering and technology, v. 45, n. 1, pp. 13–20, feb. 2013. doi: https://doi.org/10.5516/net.03.2011.078. [33] npr. challenger engineer who warned of shuttle disaster dies. 2016. available at: https://www.npr.org/sections/thetwo-way/2016/03/21/470870426/challenger-engineer-who-warned-ofshuttle-disaster-dies. [34] human rights watch. “they killed us from the inside”. 2021. available at: https://www.hrw.org/report/2021/08/03/they-killed-us-inside/investigation-august-4-beirut-blast. [35] iaea. the fukushima daiichi accident, international atomic energy agency, vienna, austria, 2015. available at: https://www.iaea.org/publications/10962/the-fukushima-daiichi-accident. copyright © 2023 by the authors. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word re jracr-1_1_intro.doc editor's introduction to the first issue of jracr this inaugural issue formally announces the establishment of the international journal of risk analysis and crisis response (jracr) to disseminate research findings and applications on a variety of topics in risk analysis (ra) and improved crisis response (cr). the advisory board and editorial board consist of internationally acclaimed scholars in their respective fields, who bring their combined talents to bear on critical challenges in ra and cr, coming from universities, companies, government agencies and non-governmental organizations in 21 different countries. today, we live in a high-risk society with more insecurity and greater challenges than ever. risk, associated with a future incident accompanied by some adverse effect1, not only arising from natural phenomena but also produced by human activity, has become everywhere linked to potentially more destructive outcomes. our ability to collect information, form knowledge and control systems has been unable to keep up with the speed of change in the world. ra has been currently used to analyze future events characterized by uncertainty that may involve deleterious consequences, whether caused by natural forces or by human short-sightedness and negligence. cr has reached an advanced level in saving lives, reducing losses and preventing the deterioration of perilous situations. so far, no theory can fully explain risks; no technology can cope with every complex crises. the presence of multiple potential threats stimulates the development of new theories and methods for risk analysis. the pursuit of more effective responses to crisis situations requires a continuous improvement of cr technology. the goal of jracr is to enable scientists, engineers and technical officers to share research results and valuable experience, to build a foundation for a better tomorrow. this issue contains eight papers. five contributions are written in english and the remaining are in chinese with english abstracts. the papers can be divided into four categories: emerging risk, crisis planning, environmental risk and disaster risk. in the first category, the paper “biogas a european perspective on safety and regulation” by salvi, delsinne and evanno2, presents a review of the situation on biogas production and uses, and points out that proper safety management and adequate regulation will ease the quick deployment of this technology, but there are still some issues to be solved. the paper “towards the development of regional risk profiles and adaptation measures for sea level rise” by ayyub and kearney3, discusses the sea level threat posed by global warming and suggests a framework for quantifying the risk for addressing many of the aspects of the sea level challenge, especially coastal infrastructure. the second category includes one paper “risk individuality in crisis planning: the case of gender in amercan agriculture” by hoag, keske and goldbach4, shows that, in some cases, it is possible to improve crisis response when we consider differences in risk perception. for example, women were more risk averse and less confident in a case study of agricultural producers in the usa. there are two papers in the environmental risk category. the paper “fuzzy multicriteria risk assessment for hazardous waste management: the case of istanbul” by uğurlu and kahraman5, suggests a multicriteria evaluation method for various hazardous waste treatment alternatives. under the main criteria, economic, risks, and technical, the authors consider 11 sub-criteria. the paper “airborne levels of uranium in the surroundings of various industrial facilities: human health risks” by rovira, linares, bellés, nadal and domingo6, analyzes the airborne concentrations of natural uranium in the vicinity of different industrial plants of catalonia, spain. the health risks associated to the exposure to this element through air inhalation were also assessed for adult people living close to those facilities. there are tree papers in the disaster risk category. the paper “a discrete model of the expected loss for catastrophe insurance in natural disasters” by huang and shi7, suggests a discrete model of catastrophe insurance to calculate the expected losses. it was successfully applied to calculate the premium rate of the typhoon in fujian province, china. the paper “the risk assessment of the fog disaster in beijing” by hu and pan8, assesses the fog disaster risk in beijing by spatial grid with recent 10 years observatory data. the result indicates that the high risk area is consistent to the extending of highway and ring-roads, as parts of urban center area and airport, and so on. the paper “the combustible materials remote sensing ration calculation and fire risk dynamic monitor in the seasons of withered grass in xilingoule grassland” by zhuo, liu and yu9, suggests a fire risk index from the superposition of the meteorological drought index, the combustible stock, vegetation continuous and snow conditions to express the risk level of surface fuel. we sincerely hope our reader will find this issue’s information on ra and cr useful, and we warmly invite readers around the world who are committed to research and practice in ra and cr to submit their quality papers to this journal for dissemination of their vital contribution for constructing a safer world! chongfu huang jracr editor-in-chief references [1] c.f. huang and d. ruan. fuzzy risks and an updating algorithm with new observations. risk analysis, 28(3) (2008) 681-694. [2] o. salvi, s. delsinne and s. evannod. biogas a european perspective on safety and regulation. journal of risk analysis and crisis response, 1(1) (2011) 1-20. [3] b.m. ayyub and m. kearney.towards the development of regional risk profiles and adaptation measures for sea level rise. journal of risk analysis and crisis response, 1(1) (2011) 75-89. [4] d. hoag, c. keske and r. goldbach. risk indviduality in crisis planning: the case of gender in amercan agriculture. journal of risk analysis and crisis response, 1(1) (2011) 21-28. [5] s. uğurlu and c. kahraman. fuzzy multicriteria risk assessment for hazardous waste management: the case of istanbul. journal of risk analysis and crisis response, 1(1) (2011) 29-41. [6] j. rovira,v. linares, m. bellés, m. nadal and j. l. domingo. airborne levels of uranium in the surroundings of various industrial facilities: human health risks. journal of risk analysis and crisis response, 1(1) (2011) 42-47. [7] c.f. huang and s x. shi. a discrete model of the expected loss for catastrophe insurance in natural disasters. journal of risk analysis and crisis response, 1(1) (2011) 48-58. [8] h.b. hu and j.j. pan. the risk assessment of the fog disaster in beijing. journal of risk analysis and crisis response, 1(1) (2011) 59-64. [9] y. zhuo, g.x. liu and f.m. yu. the combustible materials remote sensing ration calculation and fire risk dynamic monitor in the seasons of withered grass in xilingoule grassland. journal of risk analysis and crisis response, 1(1) (2011) 65-74. atlantis press journal style gezi park crisis ali ekşi atatürk medical technology vocational training school, ege university, bornova, izmir, 35040, turkey abstract the objective of this article is to evaluate the gezi park incident which started on 28 may 2013 and continued until the end of june in terms of crisis management. the reasons why a civil opposition act that started against the decision taken by the government regarding the environment transformed into a serious crisis due to an unmanageable process by the government will be questioned. the reasons for the incident and why it could not be prevented will be evaluated from the perspective of crisis management. problem areas focusing especially on crisis communication during the crisis management process, the effect of messages given by the administrators along with the use of disproportionate force in deepening the crisis along with the lessons learned from the crisis will be evaluated. the most important factors that transformed the events starting with representation problem into a crisis have been communication problems and the disproportionate force used on the protesters. turkey should develop a proactive approach in crisis management and should develop permanent solutions to solve the communication problem. the study is important since this was the first time that an environmentalist action transformed into a social movement causing a serious crisis in a country like turkey. keywords: crisis management, crisis communication, disproportionate force, participation, representation problem. 1. introduction turkey has reached a stability regarding growth and inflation during the past decade and the national income has increased from about three thousand to more than ten thousand dollars. along with development targets, assertive projects are also brought to the agenda and serious structural reforms are made. in accordance with the eu inclusion targets, studies regarding public administration reform particularly localization and participation in the decision making processes have become current issues. in this process, it is expected that in parallel to economic development and structural reform, democratic governance will be used more effectively in decision making processes. on the other hand, there are still serious issues regarding inclusive participation which is one of the basic principles of democratic governance. especially, mechanisms required to ensure the participation of the public to decision making processes regarding environment have not yet been fully formed1,2. gezi park is a public park located in the center of istanbul near the taksim square which is one of the important symbols of the republic. actually there were a military barracks in the park area during the ottoman rule. the barracks was demolished in 1940 and the gezi park of today was built3. a decision was taken by the istanbul metropolitan municipality council in 2011 and it was decided to rebuild the military barracks in accordance with its original4. however, as the park which is one of the rare green areas in the center of istanbul was opened to structuring, the intended use of the military barracks became a topic of discussion. the fact that the central government had stated many times that the barracks could be used as a shopping mall or a residence resulted in the increase of the reaction by nongovernmental organizations related with the environment5,6. the perception and management method of the government regarding the gezi park incident which started as an “ecological civil opposition movement” became an important topic of discussion7. especially the disproportionate force used on the activists during the first days of the protests along with the problems experienced by the government in managing the process resulted in the spreading of the protests to all corners of the country. the violence of violence prone groups that blended in with the activists along with the disproportionate force inflicted by the police combined to transform the incident into an unmanageable crisis8. the gezi park events started on 28 may 2013 and continued until the end of june. the after effects of the events continued in july as well. during the events, protest rallies were made in 79 of the total of 81 cities in turkey and about two and a half million people participated in these protests. over five thousand people were taken into custody during the events, five people died one of which was a police officer and about four thousand people got injured. the resulting financial harm on public and privately owned goods was about 100 million $9. the objective of this study is to evaluate the gezi park events in terms of crisis management. answers will be sought regarding how a civil opposition movement that started against a governmental decision transformed into a social movement and became a journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 158-165 published by atlantis press copyright: the authors 158 willieb typewritten text received 27 august 2013 willieb typewritten text accepted 12 december 2013 willieb typewritten text willieb typewritten text ali ekşi serious crisis as a result of a process that was unmanageable by the government. the study also aims to examine the reasons why the crisis could not be managed by the government. the fact that an environmental movement due to a representation problem has for the first time caused a crisis of this magnitude in the region where turkey is located in makes this study important. in the first section, the restructuring process of crisis management in turkey will be evaluated with its problem areas. in the second section, the reasons for the gezi park crisis will be evaluated along with the problem areas regarding public participation in environment related decision making processes. whereas in the third section the management of the gezi park crisis and problems of management will be evaluated in the light of the lessons taken out of the crisis. 2. crisis management in turkey crisis management in turkey was executed by four different institutions in turkey prior to 2009. due to the problems that this scattered structuring in crisis management caused in establishing coordination between these institutions, these four different institutions were closed down with law number 5902 issued on 29 may 2009 and the prime ministry disaster and emergency management presidency was established. among the responsibilities of the prime ministry crisis management center that was closed during this process were tasks related with prevalent violence actions for removing basic rights and freedoms along with the disruption of public order due to violence events. the job definition of the prime ministry disaster and emergency management presidency that was newly established during the restructuring process of public management was limited to natural and technological disasters. the responsibilities of the closed prime ministry crisis management center regarding social events were not defined in this new institution10,11. crisis management in turkey is centralist and hierarchical. this may sometimes decrease the effectiveness of the institutions working in crisis management, causing conflicts of role and responsibilities among the institutions. crisis management is carried out in a more reactive manner. there is especially a widespread notion that the endeavors of the rural organization of crisis management are limited to intervention only12. actually, positive works are carried out during the restructuring of the relevant management regarding these problem areas. however, serious problems still prevail in the strengthening of the rural organizations and the prevention of events with a proactive approach13. it is vital in crisis management that there be a spokesman that can form the link between management and public opinion while ensuring the trust of the public. this spokesman should have been appointed by the management to ensure consistent flow of information to the public and to provide coordination between crisis management and media. it is important that the spokesman can give messages that take into account the concerns while reassuring the public14,15. when the sample cases are evaluated, it is observed that there are no communication plans in the crisis management of turkey and this causes new problems during times of crisis. in addition, the ineffective use of spokesmen during times of crisis and the inability to give the right messages at the right times may sometimes bring with it new crises. there are no governmental units for crisis communication within the corporate crisis management structuring of turkey and there are also no corporate appointments made for professional level spokesmanship13. in ordinary times, the speeches given by the deputy prime minister who is the government spokesperson or the party vice president who is also the justice and development party (akp) spokesman is followed closely when an explanation or opinion is expected from the government. however, the speeches made by these representatives may be quite different than what is stated by the prime minister even in ordinary times. therefore, the president’s speeches are given more importance as the voice of the management during ordinary or extraordinary times and the opinions of other officials are generally not enough to remove concerns or reassure the public16,17. it is vital that the experiences gained during times of crisis are evaluated. the problems that occur during crisis management should be determined, the troubles regarding roles and responsibilities should be defined and the data formed should be shared with all parties related with crisis management18,19. the problem areas of crisis management in turkey and the management performance are not evaluated effectively in turkey and similar errors are repeated in consecutive cases of crisis. for each crisis, new reactive solutions are found to solve the problem under the conditions of that particular day. this result in the continuation of the failure in crisis management and the inability to transfer the experiences gained during crisis management to future crisis situations13,20. 3. the reasons of the gezi park crisis the lack of mechanisms for the declaration of suggestions and concerns for the section of the public who did not give any votes or who did not vote for the ruling parties may sometimes cause new problem areas and a representation problem. in time, this may cause a published by atlantis press copyright: the authors 159 gezi park crisis problem of trust to appear between the government and the citizens resulting in concerns related with whether the government will make decisions for the good of the public thereby effecting respect and trust for the administrators21,22. systems that do not form participatory mechanisms in government and systems that perceive every opposition movement as a threat to themselves will want to prevent the formation of effective non-governmental organizations. conflicts of interest may arise in systems where participatory mechanisms and non-governmental organizations are not strong enough which in turn results in making the oppression and fear among the public to become more pronounced23. in this section of the study, the problem areas in turkey regarding the representation problem which is the most important reason of gezi park events will be handled. turkey is not a party to decision international agreements such as “aarhus”, utrecht protocol” and “espoo” which make public participation in decision making processes more effective. the only democratic legal tool that the public can present its opinions and suggestions in a limited manner regarding environmental decision making processesis the environmental impact assessment (eia) processes. the eia legislation in turkey has been renewed four times since it was first issued on 7 february 1993 and there have been countless changes made on the issued regulations. changes were made in the new eia legislation which was last issued on 17 july 2008 to increase public participation taking into account the eu eia directive. in addition, non-governmental organizations have gone to court many times stating that the issued eia legislations have deviated from their aim of protecting the environment24,25. eia exemption demand by the government still continues in projects that might have serious effects on the environment in turkey. in addition, positive decisions might be taken for the eia reports for some projects despite severe protests from the local public. the final change made in the eia legislation along with the reaction of the administration against eia decisions results in the formation of a perception that eia is used mostly as a tool to convince the public instead of being a participatory tool26,27. it is important that the citizens are informed by the government and that the citizens may demand to be informed by the government in order to ensure effective participation in decision making processes. citizens can participate in decision making processes more actively and support sustainable growth when the society is informed sufficiently regarding the projects and is given the right to access information22,28. it cannot be stated that a systematic briefing is given to the public in turkey especially for the decision making processes of large projects. the projects may sometimes be announced as “mad project” during election campaigns. the legal regulation regarding the information demand by the citizens is the “right to information act” issued in 24 october 2003. even though the law openly states that everyone has the right to obtain information, some difficulties may arise during implementation. governmental organizations may sometimes refuse to give information without any justification even though there is an obligation stated by the law and sometimes a negative response may be given to the information demand by showing the 7th article of the law as justification which states that the required information demand cannot bring any additional work or research load to the administration29,30. it is possible to see some examples of public consultation in turkish public administration which can be evaluated as a stage of participation in public administration. special provincial administrations and municipalities can carry out public opinion polls and opinion researches to determine the opinions and thoughts of the public. this method is not used frequently and does not have a direct impact on decision making processes. another application in which the public can find opportunities for participation at local government level is the city councils. many municipalities in turkey have city councils, however contrary to well-functioning examples all over the world, they are used more for counseling rather than active participation in decision making processes31. the problem areas regarding public participation in decision making processes related with environment in turkey had attracted the attention of international public before. there are infringement decisions given by the european court of human rights against turkey due to the fact that public opinion is not taken in accordance with the standards during decision making processes32,33. the european commission has emphasized both in 2011 and 2012 “turkey progress reports” that the methods for taking public opinion regarding issues related with environment have not been made fully compatible and have not been carried into action34. in may 2013, the european parliament greens group declared its environmental protection concerns regarding the large projects in turkey and have criticized that some of these projects have been exempt from eia processes35. as mentioned abote, the main reason for the start of the gezi park events is representation problem. in addition, there are other factors that caused this representation problem to escalate rapidly and transform into a social event and crisis all over the country. the language used by the government in recent times while carrying out their activities caused serious reaction and concerns in a section of the society. the process that published by atlantis press copyright: the authors 160 ali ekşi started when the president announced that one of their objectives is to raise a faithful generation during the education system change in2012 continued with the declaration by the president that abortion is murder36,37,38. lastly, the name “yavuz sultan selim” selected for the 3rdbosphorusbridge the foundation of which was laid on 29may 2013 which is mentioned in various historical resources as “alewi massacrer” resulted in harsh response especially from the alawite community39,40. many statements given in 2012 and 2013 were perceived by a section of the society as intervention to their way of life and frankly these concerns were not sufficiently removed41. the part of the public that was discomforted by the declarations and messages of the president supported the gezi park events. it was striking that the study carried out by genar research company during the events pointed out that 58 % of the activists declared the president as the cause of the events while 13,7 % declared the government as the cause42. 4. gezi park crisis management process the activists organized via social media and started setting up tents in the camp and staged a sit-in on 28 may 2013 after a wall around the park was started to be demolished and some trees were started to be cut. on the morning of 30may 2013 at about five o’clock in the morning, police intervened to the group that refused to leave the park and resisted the police. the intervention caused more protesters to come to gezi park and its surroundings. on 31may 2013 the protests increased and the disproportionate force used on the protesters carried the events to the top in the agenda of the country and from there the protests expanded to cover the whole country. after this stage, the events were no longer an “ecological civil opposition” to administrative decisions and transformed into a multi-dimensional crisis7,43,44. it is undoubtedly accepted by everyone that the problem of representation during the decision making processes regarding the environment was effective in the starting of these events. the fact that about 35 % of the activists stated that they did not vote before along with the fact that among those who state to have voted before almost all declare to have voted for opposition parties supports the importance of representation problem in the transformation of the events to a crisis42. the problem areas regarding participation in decision making processes in turkey were frequently discussed both nationally and internationally, however no one expected that a participation based representation problem could cause such a crisis45. the events might not have escalated to a crisis of this magnitude if turkey had been able to solve its problem areas related with participation. the lack of a corporate body that will develop proactive management models and strategies during times of crisis caused by social movements was greatly felt during these events46,47. even though the problems that it caused are well known, the inability to prevent the crisis and how the crisis is managed are topics of discussion. the most important problem area in crisis management is the lack of a spokesman appointed to provide efficient communication with the public and the inability to give the right messages at the right time. as the events continued, istanbul administrative court adopted a motion for the stay of execution on 31may 2013 for the lawsuit filed against the barracks project planned for the park. this court order was actually seen by many as a positive development that could decrease the intensity of the events48. despite the positive messages given on the same day by the government spokesperson regarding the court order and statements meaning excuses from the protesters for the disproportionate intervention on the first day of the events, the president gave messages criticizing the decision given by the court. from then on, the public opinion started to take into consideration the declarations of only the president among the statements made by the relevant administrations. the information or messages given by other administrators were not sufficient in reassuring49. the president preferred to make statements during the rallies that his party decided to start as a response to the events. the statements declared during the rallies held in large cities of turkey with the participation of hundreds of thousands of people were more like political messages for party members rather than aiming to solve the problem50,51. throughout the crisis management period, a proper communication with the public could not be established and the right messages were not given at the right time. the mayor of istanbul who was the local administrator for the location that the events took place in did not make any declarations for the first four days45. this caused the intensity of the events to increase. the president got together with non-governmental organizations which could be thought of as one of the parties of the gezi park events on 12 june 2013 that is 15 days after the start of the events. the most important messages that came out of that meeting were the decisions to postpone the barracks project planned for gezi park until the ongoing court gave an order and to consult public opinion even if the court favored the project. these messages that were for the most part accepted among the public were late in timing. identifying the protesters as marginal along with definitions of political identity resulted in making crisis management more difficult instead of solving the problem52. studies carried out by survey companies during the events indicated that the vast majority of the published by atlantis press copyright: the authors 161 gezi park crisis protesters did not have any political identity. the study carried out by konda survey company during the events put forth that about 80 % of the protesters were not members of any association or party and that the gezi park protests were their first protest for 45 % of the protesters53. especially the messages that had to be given by the administration during the first days of the protests should have been aimed to secure environmental protection approaches and to solve representation problem instead of being of political content. in addition to the problem areas in crisis communication, the fact that the vast majority of visual media did not broadcast any of the events during the first days of the events was thought of as autocensorship and thus caused public response. this silence in visual media resulted in an increased use of social media. the study carried out by konda during the events put forth that about 80 % of the participants followed news related with the events in social media and the internet53. from time to time the unreal and exaggerated posts in social media caused various speculations to occur and most of the time made crisis management more difficult54,55. the administration had to be more transparent during crisis management and should have strived to share real information with the public in a systematic manner in order to prevent speculations. various media institutions were also targeted by the protesters during the events. protests were made in front of the buildings of some of the largest media institutions56. in turkey, there is a serious mistrust in the public regarding the media broadcasts made especially for the environment. undoubtedly, the gezi park events should have had negative impact on media trust. when this is evaluated for future crisis events from the perspective of crisis management, it might have a negative impact on the efficient use of media which is an important means of acquiring information during crisis management13. another topic of discussion in crisis management was the occasional use of disproportionate force on the protesters. the use of disproportionate force can along with the problems in crisis communication be shown to be one of the two factors that deepened the crisis and made it difficult to manage57,58,59. even though the protests started as an environment protection protest, the study carried out by konda during the events puts forth that only 15 % of the protesters have explained their participation with reasons related to the environment whereas about % 50 showed the disproportionate use of force by the police as the main reason. the occasional disproportionate force used by the police went out of control with the effect of groups inclined to use violence and quickly transformed into rallies against the government53. whereas the reason for the crisis was accepted by a section of the country as a representation problem, the lessons that the administration learned from this crisis are open to discussion. first, there is a widespread opinion among the administration that the events started with the support of outside forces to pave the way to a coup60,61. even if these opinions are completely true, when events are evaluated from a risk management perspective, the government first had to strengthen their own weak spots. the attitude of the government following the events unfortunately shows that the problem of representation has not been clearly understood. the only non-governmental organization that embraced the civil opposition since the beginning of the events, the “taksim solidarity platform” was evaluated as an illegal organization in the summary of proceedings prepared by the istanbul directorate of security62. the members and administrators of the association were taken into custody63. the authorities of the “union of chambers of turkish architects and engineers” which is one of turkey’s largest trade bodies that supported the events since the beginning that would allow participation in decision making processes were taken away with a proposal in the assembly and its financial power was decreased by taking away its income. this was evaluated by a section of the community as response to the gezi park events64. the most important suggestion regarding the solution of the representation problem following the crisis was made by the istanbul metropolitan municipality mayor. it was stated in the suggestion that public opinion was going to be taken for all projects to be carried out by the municipality. however, according to the current laws this can only cover consulting the public and would definitely not be sufficient to ensure the efficient participation of the public in decision making processes. serious reforms should be made in legal regulations and corporate structuring in order to solve the representation problem. in addition, the acceptance of standards that have been shaped in the international arena which might aid in the solution of the representation problem in environmental problems should be discussed more seriously. it is widely accepted by the vast majority of the public that the inability to establish the right communication and the use of disproportionate force are two of the most important factors that deepened the crisis and made it unmanageable44,65. however, these problem areas in crisis management have not been clearly accepted by the government after the crisis. especially among the government members, there are those who see excessive force as unnecessary and think that it makes a big thing out of such events along with published by atlantis press copyright: the authors 162 ali ekşi those that state just the opposite66,67. suggestions related to the solution of problems in crisis management are developed specific to gezi park events; for instance the suggestions for the solution of the chronic communication problem in crisis management are not even mentioned68. turkey crisis management should discuss the lessons taken from this event more seriously and carry out studies to find permanent solutions to the problem areas in order to put an end to the after effects of the gezi park crisis and to ensure that similar mistakes are not made again during the possible events that might occur in the future. 5. conclusion gezi park events started as an ecologic civil opposition movement against a decision taken by the government but has propagated to the whole country as a rights and freedom movement since it was not perceived and managed properly by the administration. the main reason for the events is the representation problem arising due to lack of participation of the public in decision making processes. whereas the inability to establish proper communication with the protesters and the disproportionate force used on the protesters during the first days of the events have transformed the events into a crisis. the coupling of the violence of violence inducing groups that mixed in with the protesters along with the use of disproportionate force has resulted in transforming the events into an uncontrollable crisis. the gezi park events showed that the social events can turn into major crises if they are not managed properly. turkey should solve this representation problem in order to avoid experiencing the same crises again. the existing problems in crisis management should be evaluated by taking into account the gezi park crisis example. permanent solutions should be developed for problem areas such as prevention of crises with proactive approaches and crisis communication. references 1. m. eğilmez, akp’nin on yılının ekonomik değeri, available at: http://www.mahfiegilmez.com/2013/01/ akpnin-on-ylnn-ekonomik-resmi.html, accessed: 8 july 2013. 2. n. ilıcak, ak parti’nin karnesi, available at: http://www.sabah.com.tr/yazarlar/ilicak/2012/12/12/akpartinin-karnesi accessed: 5 july 2013. 3. k. tayfur, yeryüzü çocukları, atlas aylık coğrafya ve keşif dergisi, 7(2013) 37-56. 4. sabah, belediye meclisi'nden oy birliği ile geçti, available at: http://www.sabah.com.tr/gundem/2013/06/ 01/belediye-meclisinden-oy-birligi-ile-gecti, accessed: 5 july 2013. 5. hürriyet, gezi parkı’ndaki 'kavga'nın sebebi avm planı, available at: http://www.hurriyet.com.tr/ekonomi/23405 125.asp, accessed: 9 july 2013. 6. k. çalışlar, gezi parkı alışveriş merkezi, available at: http://www.radikal.com.tr/yazarlar/koray_caliskan/gezi_ parki_alisveris_merkezi-1131582, accessed: 5 july 2013. 7. z. toprak, yerel yönetimler ve sivil toplum, ortadoğu ve sivil toplum diyalogu çalıştayı, 6-7 april 2013, gaziantep. 8. k. tayfur, gezi parkı kitlelerin evi, atlas aylık coğrafya ve keşif dergisi, 7(2013) 70-76. 9. t. şardan, 2.5 milyon insan 79 ilde sokağa indi, available at: http://gundem.milliyet.com.tr/2-5-milyoninsan-79-ilde-sokaga/gundem/detay/1726600/default. htm?ref=haberici, accessed: 10 july 2013. 10. n. demir, türk kamu yönetiminde kriz yönetimi, i̇çişleri bakanlığı personel genel müdürlüğü kaymakamlık tezi, ankara, 2011. 11. a. ş. özcan, afet ve acil durum yönetimi mahiyeti, i̇lgili kurumları ve kapsamı, i̇darecinin sesi, 144(2011) 11-16. 12. ş. erten, türk kamu yönetiminde kriz yönetimi anlayışı, süleyman demirel üniversitesi sosyal bilimler enstitüsü yüksek lisans tezi, isparta, 2011. 13. a. ekşi, nükleer kaza ve saldırılarda bütünleşik kriz yönetimi, dokuz eylül üniversitesi sosyal bilimler enstitüsü doktora tezi, i̇zmir, 2013 14. m. tyrone, crisis communication: a commanders guide to effective crisis communicatio, air command and staff college air university, alabama, 1998. 15. nhpco, crisis communications planning a guide for providers, available at: http://www.nhpco.org/files/publ ic/communications/crisis_guide.pdf, accessed: 8 july 2013. 16. t. akyol, olaylardan dersler, available at: http://www.hurriyet.com.tr/yazarlar/23421681.asp, accessed: 6 july 2013. 17. milliyet, bülent arınç'tan 'gezi parkı' açıklaması, available at: http://siyaset.milliyet.com.tr/bulent-arinctan-gezi-parki-/siyaset/detay/1717157/default.htm, accessed: 8 july 2013. 18. r. brewster, natural disaster recovery planning, university of technology, kingston, 2005. 19. m. kadıoğlu, afet yönetimi beklenilmeyeni beklemek en kötüsünü yönetmek, t.c. marmara belediyeler birliği yayını, i̇stanbul, 2011, 20. h.h. güler, zarar azaltmanın temel i̇lkeleri, in kadıoğlu, m. and özdamar, e. (ed.), afet zararlarını azaltmanın temel i̇lkeleri, jica türkiye ofisi yayınları no: 2, ankara, 2008, pp. 35-50. 21. s. bilgen, katılımcılık ve temsiliyet, available at: http://www.oed.org.tr/oed/images//katilimciliktemsiliyet.pdf, accessed: 7 july 2013. 22. a. özyol, sürdürülebilir kalkınma ve katılımcı demokrasi, available at: http://www.bpwturkey.org/eklentiler/surdurulebilirkalkinmavekatilimcid emokrasi.doc, accessed: 7 july 2013. 23. z. toprak, algıda seçicilik ve ekoloji, available at: http://www.kazete.com.tr/makale_algida-secicilik-veekoloji_853.aspx, accessed: 6 july 2013. published by atlantis press copyright: the authors 163 gezi park crisis 24. s.s. alıca, (2011), ‘çevresel etki değerlendirmesinin yargı kararları çerçevesinde i̇rdelenmesi’, gazi üniversitesi hukuk fakültesi dergisi, volume 11, number 3, pp. 97-130. 25. chamber of environmental engineers, çed yönetmelik değişikliği üzerine açıklama, available at: http://www.tmmob.org.tr/genel/bizden_detay.php?kod= 8963&tipi=9, accessed: 6 july 2013. 26. artı ivme magazine urban transformation commission, akp’nin çılgın projeleri ve çed engeli, artı i̇vme dergisi, available at: http://www.i̇vmedergisi.com/ akp%e2%80%99nin-%c3%a7ilgin-projeleri-ve%c3%a7ed-engeli.html, accessed: 5 july 2013. 27. enerjiline, artık çed raporu i̇stenmeyecek, available at: http://www.enerjiline.com/haberdetay/artik-cedraporu-istenmeyecek/544, accessed: 6 july 2013. 28. r. demir, yerel yönetimlerde halkın yönetime katılımı, yönetim ve denetim dergisi, 11(1) (2006) 6073. 29. z. aras and e.b. altınok, türkiye’de bilgi edinme hakkı ve uygulama sorunlar, ankara barosu dergisi, 67(1) (2009)107-112. 30. s.y. koçak, kamu yönetiminde açıklık i̇çin bilgi edinme hakkı, selçuk üniversitesi sosyal bilimler enstitüsü dergisi, 23(2010) 115-125. 31. a. t. özdemir, mahalli i̇darelerde halk katılımı bağlamında kent konseyleri, sayıştay dergisi, 83(2011) 31-56. 32. s. cengiz, avrupa i̇nsan hakları mahkemesi kararı işığında bergama davası, http://www.inisiyatif.net/ document/12.asp, accessed: 7 july 2013. 33. r. türmen, çoğulcu ve katılımcı demokrasiye doğru, available at: http://gundem.milliyet.com.tr/cogulcu-vekatilimci-demokrasiye/gundem/ydetay/1722280/default. htm, accessed: 6 july 2013. 34. european commission, türkiye 2012 yılı ilerleme raporu, available at: http://www.abgs.gov.tr/files/ strateji/2012_ilerleme_raporu.pdf, accessed: 6 july 2013. 35. p. cengiz, çılgın proje avrupalı yeşiller’in markajında, available at: http://www.taraf.com.tr/pelin-cengiz/ makale-cilgin-proje-avrupali-yesiller-in-markajinda.htm, accessed: 6 july 2013. 36. c. akgüloğlu, 4+4+4 dindar neslin doğuşu, available at: http://www.radikalgenc.com/politika/444-dindar-neslindogusu, accessed: 5 july 2013. 37. i̇. çoşkun, 4+4+4 dindar nesil yetiştirme projesi mi? available at: http://www.sabah.com.tr/perspektif/ yazarlar/icoskun/2012/09/08/444-dindar-nesilyetistirme-projesi-mi, accessed: 7 july 2013. 38. milliyet, erdoğan: kürtaj bir cinayettir, available at: http://siyaset.milliyet.com.tr/erdogan-kurtaj-bircinayettir/siyaset/siyasetdetay/26.05.2012/1545254/defa ult.htm, accessed: 9 july 2013. 39. n. yılmaz, yavuz sultan selim alevileri katletti mi? available at: http://www.internethaber.com/yavuzsultan-selim-alevileri-katletti-mi-541047h.htm, accessed: 10 july 2013. 40. dogan news agency, alevi dernekleri 3. köprü i̇smine tepki gösterdi, available at: http://www.haberler.com/ alevi-dernekleri-3-kopru-ismine-tepki-gosterdi4697428-haberi/, accessed: 8 july 2013. 41. a. öymen, alkol için iki ayrı söylem, available at: http://www.radikal.com.tr/yazarlar/altan_oymen/alkol_ic in_iki_ayri_soylem-1135390, accessed: 11 july 2013. 42. genar, gezi parkı profili, available at: http://www.genar.com.tr/files/geziparki_profilson.pdf, accessed: 9 july 2013. 43. h. hacaloğlu, taksim gezi parkı’nda polis müdahalesi, available at: http://www.amerikaninsesi.com/content/ taksim-gezi-parkinda-polis-mudahalesi/1672061.html, accessed: 6 july 2013. 44. i̇. berkan, esas orantısız güç kullanımı, available at: http://www.hurriyet.com.tr/yazarlar/23501729.asp, accessed: 8 july 2013. 45. vatan, ben öyle bir kabataş videosu görmedim, available at: http://haber.gazetevatan.com/ben-oyle-birkabatas-videosu-gormedim/553520/1/gundem, accessed: 12 july 2013. 46. t. bengin, hani istihbarat gelmişti, available at: http://gundem.milliyet.com.tr/hani-istihbarati-gelmisti/gundem/ydetay/1730225/default.htm, accessed: 7 july 2013. 47. a. bayramoğlu, bir demokrasi krizi olarak gezi, available at: http://yenisafak.com.tr/yazarlar/alibayra moglu/bir-demokrasi-krizi-olarak-gezi/38243, accessed: 9 july 2013. 48. anadolu agency, gezi parkı ile ilgili mahkemeden karar, available at: http://www.haber7.com/guncel/ haber/1033404-gezi-parki-ile-ilgili-mahkemeden-karar, accessed: 6 july 2013. 49. akşam, başbakanlıkta gezi parkı zirvesi, available at: http://www.aksam.com.tr/siyaset/basbakanlikta-geziparki-zirvesi-iste-aciklamalar/haber-215602, accessed: 7 july 2013. 50. m.a. berber, i̇hanetin şebekesini milletimize açıklayacağız, available at: http://www.sabah.com.tr/ gundem/2013/06/16/ihanet-sebekesini-milletimizeaciklayacagiz, accessed: 6 july 2013. 51. m. alphan, milli irade, http://www.hurriyet.com.tr/ yazarlar/23545140.asp, accessed: 5 july 2013. 52. e. kongar, gezi parkı ve nefret suçları, available at: http://www.cumhuriyet.com.tr/?hn=427004&kn=37&ka =4&kb=5&kc=37, accessed: 11 july 2013. 53. konda, konda anketine göre gezi parkı, available at: http://www.aksam.com.tr/guncel/konda-anketine-go re-gezi-parki/haber-215337, accessed: 10 july 2013. 54. m. yılmaz, gezi olaylarında otosansür habercilik refleksini bastırdı, available at: http://www.medyagun lugu.com/ozelhaber/473-otosansur-habercilik-refleksinibastirdi.html, accessed: 11 july 2013. 55. b. akçura, günümüz medyasında sansür ve otosansür, available at: http://www.milliyet.com.tr/gunumuzmedyasinda-sansur-ve/ombudsman/haberdetay/10.06.20 13/1720764/default.htm, accessed: 6 july 2013. 56. vatan, ntv'de büyük protesto, available at: http://haber.gazetevatan.com/ntvde-buyukprotesto/543260/1/gundem, accessed: 9 july 2013. 57. anadolu agency, mahkemenin kararı isabetli, available at: http://www.ntvmsnbc.com/id/25446574/, accessed: 8 july 2013. published by atlantis press copyright: the authors 164 ali ekşi 58. anadolu agency, bülent arınç’tan gül görüşmesi sonrası gezi parkı özrü, available at: http://gundem.bugun.com.tr/hukumet-ozur-diledihaberi/651420, accessed: 7 july 2013. 59. ü. kılınç, orantısız güç ai̇hm’den döner, available at: http://www.radikal.com.tr/radikal2/orantisiz_guc_aihmd en_doner-1139892, accessed: 8 july 2013. 60. m.a. şahin, gezi parkı eylemlerinin sebebi ağaç değil, darbeye zemin hazırlamak, available at: http://t24.com.tr/haber/mehmet-ali-sahin-gezi-parkieylemlerinin-sebebi-agac-degil-darbeye-zeminhazirlamak/231114, accessed: 8 july 2013. 61. m. çubukçu, gezi’deki dış güçler, available at: http://www.radikal.com.tr/radikal2/gezideki_dis_gucler1137599, accessed: 8 july 2013. 62. cumhuriyet, dayanışma yasa dışı örgüt oldu, available at: http://www.cumhuriyet.com.tr/?hn =428052, accessed: 6 july 2013. 63. radikal, taksim dayanışma serbest, available at:http://www.radikal.com.tr/turkiye/taksim_dayanisma_ serbest-1141451, accessed: 11 july 2013. 64. ihlas news agency, tmmob'dan basın açıklaması, available at: http://www.haberler.com/tmmob-danbasin-aciklamasi-4818748-haberi/, accessed: 5 july 2013. 65. e. kongar, üç erdoğan, available at: http://www.cum huriyet.com.tr/?hn=425192&kn=37&ka=4&kb=5&kc=3 7, accessed: 11 july 2013. 66. dogan news agency, nabi avcı'dan gezi parkı olayları için dikkat çekici yorum, available at: http://www. aksam.com.tr/siyaset/nabi-avcidan-gezi-parki-olaylariicin-dikkat-cekici-yorum/haber-211903, accessed: 6 july 2013. 67. e. kongar, akp’nin büyük başarısı, available at: http://www.cumhuriyet.com.tr/?hn=425192&kn=37&ka =4&kb=5&kc=37, accessed: 11 july 2013. 68. m. sarıkaya, 2014 seçim senaryoları, available at: http://www.haberturk.com/yazarlar/muharremsarikaya/846158-2014-secim-senaryolari, accessed: 11 july 2013. published by atlantis press copyright: the authors 165 1. introduction 2. crisis management in turkey 3. the reasons of the gezi park crisis 4. gezi park crisis management process 5. conclusion references microsoft word risk analysis of water resources crisis-revised.doc risk analysis of water resources crisis in the lancang mekong river drainage basin under the background of climate change ren zhang1,2 1college of meteorology and oceanography, pla univ. of sci. & tech. , nanjing 211101, china 2nanjing information technology university, klme, nanjing 210044, china email: zren63@126.com hengwan zou college of meteorology and oceanography, pla univ. of sci. & tech. , nanjing 211101, china mei hong college of meteorology and oceanography, pla univ. of sci. & tech. , nanjing 211101, china aixia zhou college of meteorology and oceanography, pla univ. of sci. & tech. , nanjing 211101, china gang zeng nanjing information technology university, klme, nanjing 210044, china abstract aiming at the deteriorating variation trends and potential crisis of water resources in the lancang-mekong river basin under climate change, from the view of risk analysis, a conceptual framework of tropical cyclone disaster risk is established, an assessment index system of water resources crisis risk in langcang-mekong river basin are explored based on the thought of ahp, and using the risk index assessment method, an assessment index system is established. then, the risk assessment of water resources crisis in the lancang mekong river basin under climate change is carried out. keywords: lancang mekong river; water resources; risk assessment. 气候变化背景下澜沧江-湄公河流域水资源危机风险分析 张韧 1,2 ，邹恒琬 1 ， 洪梅 1，周爱霞 1，曾刚 2 1 解放军理工大学气象海洋学院，南京 211101，中国 2 南京信息工程大学 klme，南京 210044，中国 摘要：针对气候变化背景下澜沧江-湄公河流域水资源变化趋势和潜在危机，从风险分析的角度建立了水资 源危机的风险概念框架；基于层次分析法思想探索了澜沧江-湄公河流域水资源危机风险的体系结构，采用 风险评价指数方法初步构建了水资源危机风险评估模型，进行了气候变化背景下澜沧江-湄公河流域水资源 危机风险的实验评估。 关键词：澜沧江-湄公河，水资源，风险评估 journal of risk analysis and crisis response, vol. 2, no. 3 (november 2012), 209-213 published by atlantis press copyright: the authors 209 administrateur texte tapé à la machine received 24 june 2012 administrateur texte tapé à la machine accepted 1 october 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine ren zhang 1. 引言 澜沧江-湄公河发源于我国青海省唐古拉山脉， 在我国境内称为澜沧江，自云南南腊出境后称为湄 公河，在越南胡志明市以南流入南海。该河流从北 向南依次流经我国青海、西藏和云南三省以及缅 甸、老挝、泰国、柬埔寨和越南五国。澜沧江-湄公 河是亚洲第六大河流，是亚洲流经国家最多的国际 河流[1]，被称为“东方的多瑙河”。 澜沧江-湄公河流域位于亚洲热带季风区，受季 风影响，5~10 月为降水充沛的雨季，11 月~次年 4 月为干燥少雨的旱季。雨季流域区域暴雨频繁，易 引发洪水泛滥；而在旱季，流域各地则易遭受干旱 [2] 。冰雪融水是澜沧江-湄公河径流主要补给源，近 年来随着全球气候变化，青藏高原的冰雪消融响应 出现了加快趋势，短期内它可能增加流域径流量， 使洪水灾害的风险加大；但从长远来看，这种趋势 可能会加剧高原地区水资源的短缺甚至枯竭。由于 人口的持续增长和社会经济的迅速发展，流域各国 都在加大对河流水资源的开发利用，生产、生活对 该河流依赖性更大，这无疑会对澜沧江-湄公河流域 水环境造成更大负担，既制约河流水质和流域社会 经济发展，同时对流域区水资源的无度使用、恶性 争夺和掠夺性开发等行为也会愈演愈烈，进而引发 流域区的资源争端、地区稳定和环境生态危机。 目前，气候变化背景下水资源危机风险的研究 尚不多，且多以定性描述为主。为此本文基于风险 分析理论，开展澜沧江-湄公河流域水资源危机的风 险评估研究，初步建立了气候变化背景下澜沧江-湄 公河流域水资源危机风险的量化评估模型。 2. 资料与方法 2.1. 研究对象 本文从澜沧江-湄公河流域上游到下游依次选取 七个代表性的城市作为研究对象（图1），分别是： 我国境内的德钦、大理、景洪；泰国的清莱；老挝 的万象和巴色以及柬埔寨的金边。 2.2. 研究资料 （1）气象水文数据：根据中国气象局气候中心和世 界气象组织的资料整理得到澜沧江-湄公河流域的降 水、气温 、暴雨、干旱等气象要素数据；根据云南 省水文信息网和世界江河数据库整理获得澜沧江-湄 公河的径流量、输沙量等水文要素。 （ 2 ） 社 会 人 文 数 据 ： 根 据 国 际 货 币 基 金 组 织 (imf)、斯德哥尔摩国际和平研究所（sipri）和世 界水资源发展报告等整理得到研究对象的水利发展 状况和社会经济等资源。 以上资料统计数据时间范围为19612008年。 图 1 澜沧江-湄公河流域及城市示意图 2.3. 研究方法 根据灾害风险评价指数方法 [3-4] ，水资源危机风 险的形成是危险性（h）和脆弱性（v）共同作用的 结果，数学表达式为： 水资源危机风险指数(r)=危险性(h) × 脆弱性(v) (1) 采用层次分析法（ahp）并结合delphi法[5-6]建 立指标体系和确定指标权重。为消除指标量纲差 异，采用如下标准化公式对各指标标准化处理： max min ' x x x x = − (2) 式中， 'x 为某指标标准化后的值，x为该指标 的原始值，xmin为研究范围内该指标的最小值， xmax为研究范围内该指标的最大值。 对于危险性和脆弱性各指标的合成采用加权综 合评价法(wca)[7]，计算公式为： published by atlantis press copyright: the authors 210 risk analysis of water resources crisis 1 n j i i i c w x = = ⋅∑ (3) 式中，cj表示评估项目j（危险性或脆弱性）的 计算值；xi是对应于评估项目j的指标i的标准化值； wi是指标i权重；n是对应于评估项目j的指标个数。 3. 概念模型与评价指标 3.1. 概念模型 风险是指遭受不利事件影响损失的可能程度， 是孕险环境、致险因子和承险体相互作用的结果。 澜沧江-湄公河流域的水资源区域风险评估包括如下 三方面内容： （1）危险性评估。通过对研究区域降水、温度和洪 涝干旱灾害强度、频率分布情况及变化趋势的统计 分析和预测结果，给出气候致险因子的危险性大小 [8] ；针对地理人文环境进行成险机理分析，从而确 定孕险环境的危险性。在致险因子和孕险环境危险 性分析的基础之上，综合得出研究对象的水资源危 机的危险性评估[9] 。 （2）脆弱性评估。即承险体在出现水资源短缺威胁 时可能的易损程度，主要包括四方面内容：1）承险 体暴露性评估，即对处于争端风险中的人员、财 产、经济活动等承险体的数量（或价值量）及其分 布进行分析[10]；2）承险体敏感性评估，评估承险体 对水资源短缺和洪水干旱灾害响应能力；3）区域承 险能力评估，即研究对象应对水资源短缺能力、防 治生态恶化和抵御洪水干旱灾害能力的评估，它反 映的是人类社会应对风险的主观能动性。 （3）区域风险评估。是在危险性评估和脆弱性评估 的基础上，对区域风险进行综合评估，并以相对等 级的方式加以划分。 本文中气候变化背景下澜沧江-湄公河流域的水 资源危机风险是指由于冰雪融化和暴雨、干旱等气 候变化极端事件可能导致的水资源短缺、流域生态 恶化、水资源污染和旱涝灾害给澜沧江-湄公河流域 带来的损失风险。它是在一定的孕险环境中，由致 险因子危险性和区域承险体脆弱性共同作用形成 的。基于上述分析，建立了气候变化背景下澜沧江湄公河流域水资源危机风险概念模型（图2）。 3.2. 指标体系 根据上述水资源危机风险概念模型以及风险评 价指数法，综合考虑指标选取的科学性、完备性、 相对独立性、系统性和现实可操作性原则，建立了 如下水资源危机风险评价指标体系（图3）。 水资源危机风险为评估的目标层，由危险性指 标体系和承险体脆弱性指标体系两个子目标构成。 危险性指标体系包含致险因子和孕险环境两个评价 模块，分别包含了降水平均值、干旱频率和植被覆 盖率等 8 个指标；承险体脆弱性指标体系由物理暴 露性、敏感性和区域承险能力等三个评价模块构 成，分别包含了人口密度、gdp 增长率、水资源利 用率等 12 个指标。 图 2 澜沧江-湄公河流域水资源危机风险概念模型 各指标的权重采用层次分析法（ahp）确定， 在构造判断矩阵时为了使计算结果更加合理可靠， 还可采用delphi法广泛征求专家意见，具体操作步 骤可参考相关文献[4-5]。经一致性检验的权重计算结 果见图3中标注。 3.3. 评估模型 基于上述指标体系和公式（1）、（3），可建 立如下水资源危机风险评估模型： published by atlantis press copyright: the authors 211 ren zhang vh ww vhr •= （6） 21 21 ww xxh •= （7） ∑ = •= 6 3i ii xwv （8） 其中，x为各指标标准化之后的量化值。 图 3 水资源危机风险指标体系（包括权重） 4. 风险实验评估 基于风险评估模型（6）～（8）和澜沧江-湄公 河流域七个城市的数据资料，分别计算得到各评估 城市的水资源危机风险的危险性、脆弱性和风险度 指数（如表 1）。为了比较不同评估单元的危险性 程度，用自然断裂法[11]，将危险度分为 5 级，各等 级划分基准及各等级划分结果如表 2。同理，可得 到脆弱度和风险度的等级划分结果如表 3 和表 4。 表 1 水资源危机风险指数评估结果 城 市 德钦 大理 景洪 清莱 万象 巴色 金边 危险性指数 0.2285 0.7943 0.4879 0.157 0.6527 0.3819 0.7091 脆弱性指数 0.1877 0.565 0.3479 0.2206 0.6937 0.4586 0.5077 水资源风险度 0.1994 0.6416 0.3946 0.2013 0.68 0.433 0.5748 表 2 水资源危机风险危险度等级划分 等 级 极高 较高 中等 较低 极低 危险性指数 ≥0.7 0.55~0.69 0.35~0.549 0.0.2~0.349 ≤0.19 城 市 大理、金边 万象 景洪、巴色 德钦 清莱 表 3 水资源危机风险脆弱度等级划分 等 级 极高 较高 中等 较低 极低 脆弱性指数 ≥0.6 0.5~0.59 0.4~0.49 0.25~0.39 ≤0.249 城 市 万象 大理、金边 巴色 景洪 清莱、德钦 published by atlantis press copyright: the authors 212 risk analysis of water resources crisis 表 4 水资源危机风险度等级划分 等 级 风险极高 风险较高 风险中等 风险较低 风险极低 风险度 ≥0.6 0.5~0.59 0.4~0.49 0.25~0.399 ≤0.249 城 市 万象、大理 金边 巴色 景洪 清莱、德钦 万象和金边均为世界上重要的粮食生产区，对水 资源的需求较大，万象和金边作为老挝和柬埔寨的首 都，是国际性旅游城市，人口密度大、水资源消耗 多，导致其危险性、脆弱性偏高 [12] 。相比以上两个城 市，虽然大理的灌溉面积要小许多，但其作为著名的 旅游城市，外来人口多、水资源消耗大。近年来，大 理地区频繁遭遇干旱和洪涝等极端天气事件，导致水 资源危机的危险性和脆弱性均较高。综上因素综合评 价：大理、万象为水资源高风险城市，金边为水资源 较高风险城市，巴色为中等风险城市。泰国清莱和我 国景洪虽然较易发生洪灾，但人口相对较少、经济欠 发达、且森林覆盖率较高，脆弱性较小，进而水资源 危机风险偏小。我国德钦为山区，灌溉面积少、植被 覆盖率高，干旱的强度和频率较小，故其水资源危机 风险最小。 气候变化对水资源的影响因子众多、影响机理复 杂；水资源对于社会经济发展的作用过程既包含自然 环境因素、也包含社会与人文因素。气候变化风险首 先是通过自然环境响应的物理机理影响流域降水和径 流；水资源的变异再通过社会机理等过程对流域边境 各国的生产生活、政治经济、军事外交等产生影响， 进而诱发流域地区风险和危机事件。本文由于资料有 限，评价指标比较简单，仅对流域水资源危机风险作 了初步探讨。下一步工作中，我们将进一步考虑自然 和社会因素，充实、完善和扩展指标体系与相应的评 估模型，以使评价体系和评估模型更加科学、合理。 致谢：气象灾害教育部重点实验室开放课题（klme1112） 参考文献 [1] 李丽娟, 李海滨, 王娟. 澜沧江水文与水环境特征及其时 空分异. 地理科学, 2002, 22(1): 49-56. li lijuan, li haibin, wang juan. analysis on hydrological and water quality character and theirspatial and temporal distribution in lancangjiang river. scientia geographica sinica,2002, 22(1): 49-56. [2] he daming,zhao wunjuan,feng yan. research progress of international rivers in china. journal of geographical sciences,2004,21:21-28. [3] davidson r a, lamber k b. comparing the hurricane disaster risk of u．s．coastal counties. natural hazards review，2001, 8:132-142． [4] 杜栋, 庞庆华, 吴炎. 现代综合评价方法与案例精选. 北 京:清华大学出版社, 2008.du dong, pang qinghua, wu yan. modern comprehensive assessment method and case selection. beijing, tsinghua university press, 2008. [5] 何逢标. 综合评价方法 matlab 实现. 北京: 中国社会 科学出版社, 2010. he fengbiao. implementation of comprehensive evaluation method by using matlab language. bejing: chinese social science press, 2010. [6] 文世勇, 赵冬至, 陈艳拢. 基于 ahp 法的赤潮灾害风险 评估指标权重研究. 灾害学, 2007, 22(2): 9-14. wen shiyong, zhao dongzhi, chen yanlong. research on index weighting of risk assessment of harmful algal bloom hazard based on ahp. journal of catastrophology. 2007, 22(2): 9-14. [7] 张继权, 魏民. 加权综合评分法在区域玉米生产水平综 合评价与等级分区中的应用. 经济地理, 1994, 14(5): 1921. zhang jiquan, wei min. the implementation of weighted comprehensive evaluation method in corn production level and grades. economic geography,1 994, 14(5): 19-21. [8] 黄崇福. 综合风险评价的一个基本模式. 应用基础与工 程科学学报, 2008, 16(3): 371-381. huang chongfu. a basic model for assessing integrated risk. journal of basic science and engineering, 2008, 16(3): 371-381. [9] 黎鑫, 张韧. 南海-印度洋海域海洋环境风险评估与区划. 气象水文装备, 2010, 21(6): 14-20. li xin, zhang ren. risk assessment and regionalization of ocean environment in the south china sea and indian ocean. meteorological and hydrological equipments, 2010, 21(6): 14-20. [10] 葛全胜, 邹铭, 郑景云等. 中国自然灾害风险综合评价初 步研究. 北京: 科学出版社, 2008. ge quansheng, zou ming, zheng jingyun, et al. integrated assessment of natural disaster risks in china. beijing: science press, 2008. [11] 黎鑫. 南海-印度洋海域海洋环境风险分析体系与评估技 术研究. 南京: 解放军理工大学, 2010. li xin. the research of risk assessment and regionalization of ocean environment in the south china sea and indian ocean. nanjing: pla university of science and technology, 2010. [12] 何大明, 王玉朝, 何希吾等. 西南国际河流跨境水资源的 合理利用和可再生性维持研究.中国自然资源学会 2004 年学术年会论文集, 2004, 410-417. he daming, wang yuchao, he xiwu, et al. study for the rational use and sustainability of international rivers in southwestern china. paper compilation of china society of natural resources in 2004, 2004, 410-417. published by atlantis press copyright: the authors 213 microsoft word mcsim-based occupational health risk assessment on benzene   mcsim-based occupational health risk assessment on benzene minyan li1 deyin huang1,* mao liu2 1institute of occupational health, tianjin bohai chemical industry group co. ltd, tianjin 300051, china 2centre for urban public safety research (nankai university), tianjin 300071,china abstract nowadays, the acute and chronic adverse health effects caused by benzene exposure have attracted wide attention. in this paper, we evaluate the health risk based on the four-step method recommended by u.s. environmental protection agency (usepa). we present a method using the monte carlo simulating software (mcsim) to build a pbpk model to simulate the concentrations in target tissues of human in distribution forms. the internal dose will be used in the dose-response model to figure out the cancer risk. in the case of workers exposed to benzene in oil refinery workplace, we take the advantage of fluent software to simulate the external dose, treat the concentration data by spss, and compute the internal dose by mcsim software, so that the lifetime cancer risk range of exposed workers can be calculated if the benzene occupational exposure scenarios are complicated. this method will be used to provide the basis for the occupation health management of enterprises. keywords: health risk assessment, benzene occupational exposure, physiologically based pharmacokinetic model, dose-response relationship, mcsim software 基于 mcsim 软件的苯职业暴露健康风险分析研究 李敏嫣 1 黄德寅 1,* 刘茂 2 1.天津渤海化工集团有限责任公司劳动卫生研究所，天津 300051 2. 天津市南开大学城市公共安全研究中心，天津 300071 摘要：现今，苯暴露引发的急性和慢性健康损害越来越引起社会关注。本文从风险定量评价角度，采用美 国国家环境保护局（usepa）推荐的风险评价四步法，对工人在苯职业暴露下的健康风险进行定量评价。利用 蒙特卡洛模拟软件 mcsim 建立苯的生理药代动力学模型，预测苯暴露工人体内各器官中苯及其代谢物质的内剂 量浓度分布，并将得到的内剂量数据代入拟合的剂量-反应函数，求得致癌风险值。在某凝析油炼制作业场所的 案例分析中，利用 spss 软件处理环境监测实验室数据，采用 fluent 计算机模拟方法预测多种混合暴露场景 下工人苯职业暴露的外剂量浓度，并进一步计算工人在此暴露条件下的终生致癌风险，为企业的职业卫生管理 提供依据。 关键词：健康风险，苯职业暴露，生理药代动力学模型，剂量-反应关系，mcsim 软件 1. 引言 随着工业生产的迅速发展，苯作为原料和产品，                                                                 基金项目：天津市科技支撑计划重点项目 （13zczdsy02300） *通讯作者: 黄德寅（1962-），女，主任医师 研究 方向：职业卫生与职业健康风险。 email: huang_deyin@126.com 在工作场所的应用越来越广泛。工人在职业环境中长 期接触苯，可能出现不同程度的中毒症状或癌症，严 重威胁了工人的身体健康。 健康风险评价通过对职业危害因素的潜在作用 进行定量估计，推断在一定暴露条件下，有害物质对 人体健康造成损害的概率和程度，预测职业危害因素 潜在影响，为职业病的预防和防护提供科学依据。 目前国外很多国家在污染物扩散评估技术上多 journal of risk analysis and crisis response, vol. 3, no. 3 (november 2013), 135-145 published by atlantis press copyright: the authors 135 willieb typewritten text received 28 august 2013 willieb typewritten text willieb typewritten text accepted 17 october 2013 willieb typewritten text willieb typewritten text willieb typewritten text   采用综合评估技术，其中数学模型和计算机数值模 拟，尤其是cfd技术的应用发展很快。 我国的健康风险评价工作始于20世纪90年代初， 以介绍和应用国外的研究成果为主。当前，国内健康 风险问题的研究，基本上采用美国国家环保局推荐的 健康风险评价模型。国内职业卫生领域对职业健康风 险方面的研究主要集中在职业暴露监测技术及流行 病学调查研究方面，通常采用环境监测得到的浓度作 为剂量-反应的剂量数据。 随着健康风险评价方法的不断发展，在致癌风险 评价过程中，开始使用内剂量代替外剂量。利用生理 药代动力学（pbpk）模型，模拟有害物质在人体内 的分布、运输和代谢，并用模拟得到的靶器官内剂量 来代替环境中的暴露剂量进行风险分析，将其作为剂 量-反应关系中的剂量更具说服力和准确性。 施婕、王阳、张静、张倩等人利用健康风险评价 四步法，结合生理药代动力学模型与蒙特卡洛模拟软 件，分别评估了三氯乙烯急性暴露[1]、苯慢性暴露 [2-9] 、 氯乙烯[10]和1,3-丁二烯慢性暴露[11,12]的健康风险。 2. 方法 本文依据风险评价的四步法，对苯职业暴露下的 致癌风险进行定量评价。通过对工作场所中的健康危 险源进行辨识，收集相关资料，建立剂量-反应模型， 再根据工作场所中fluent模拟出的暴露浓度，对暴 露人群在此环境下长期工作的潜在致癌风险做出定 量的估计。 2.1风险辨识 风险辨识用于对有害因子引起不良健康反应的 可能性进行定性评价，包括确定暴露于某物质能否引 起不良的健康影响，并定性描述影响的性质和强度。 长期接触较低浓度苯可引起慢性中毒，多数患者 出现神经系统症状。慢性苯中毒主要损害造血系统。 苯已确认为人类致癌物，可引起各种类型的白血病。 我国卫生部颁布的《高毒物品目录》也将苯列入其中。 《职业病危害因素分类目录》中，苯被列为可能导致 苯中毒和白血病的职业病危害因素。 苯被吸入后首先在肝脏内细胞色素氧化酶 p4502e1（cyp2e1）的作用下，氧化成苯环氧化物， 并经非酶性重排自发形成苯酚；也可催化生成开环 产物反式，反式-粘康醛，进而形成反式，反式-粘 康酸（t,t-ma）；或在谷胱甘肽-s 转移酶（gst）的 催化作用下，与谷胱甘肽（gsh）结合形成低毒苯 硫醇尿酸（pha）；或经环氧化物水化酶催化，形成 苯二氢二醇。其中，t,t-ma 和 pha 均可通过尿液排 出体外。苯酚经 cyp2e1 代谢成苯的双羟基或三羟 基化合物，如氢醌（hq）、儿茶酚（cat），苯三醇 （bt）。氢醌、儿茶酚，苯三醇通过血液循环到达 骨髓，在骨髓内经髓过氧化物酶（mpo）催化成苯 醌（hq）。苯醌可以被还原性辅酶ⅱ——醌氧化还 原酶（nqo1）还原为低毒的多羟基苯[13]。 苯代谢产物通过血液循环运送到骨髓或其他器 官，可能表现为骨髓毒性和致白血病作用。迄今为 止，苯的毒作用机制仍未完全阐明。 有关苯的毒性、致癌性及诊断、治疗、预防等 问题受到了全世界的广泛关注，并在土耳其制鞋工 人[14,15]、俄亥俄州盐酸橡胶工厂工人 [16-19] 、dow 化 学公司工人[16,20]、某地区化工厂男性工人（wong et al.）[21,22]和中国工人[23,24]中间开展了流行病学调查。 2.2暴露评价 2.2.1外暴露评价——基于cfd技术的计算机模拟方法 cfd 技术可用于研究复杂条件下的职业暴露 评估方法，通过 fluent 软件实现有毒物质泄漏后 浓度场分布的计算机模拟，实现多种情况下浓度场 分布的计算机模拟，并与 pbpk 技术结合进行致癌 风险分析技术的进一步开发应用。 在利用 fluent 软件进行模拟时，首先利用 gambit 或者其他前处理器完成模拟对象集合结 构的建模以及计算网格的生成与划分，然后将网格 导入 fluent 中进行求解计算， 后对计算结果进 行处理和分析。 2.2.2内暴露评价 2.2.2.1 生理药代动力学（pbpk）模型的选择 本研究中使用的人类pbpk模型基于苯在小鼠体 内代谢的模型[25,26]。该模型追踪了苯、苯环氧化物、 苯酚、氢醌、粘康酸、苯巯基尿酸、苯酚轭合物、氢 醌轭合物、儿茶酚和苯三醇总量在人体各组织中的浓 published by atlantis press copyright: the authors 136   度水平。为研究苯、苯环氧化物、苯酚和氢醌的生理 药代动力学过程，将人体划分为以下房室：血液（苯 生理药代动力学过程不包含血液部分）、肾脏、肝脏、 脂肪、充分灌注组织（包括大脑、心脏、脾脏、肠和 骨髓），以及不充分灌注组织（包括肌肉和皮肤）。 肝脏被划分为三个体积相等的房室，以表示肝脏的三 个区域。此前的研究中，肾脏通常被视作充分灌注组 织的一部分。本文将肾脏作为单独的房室加入模型， 其余部分和单一肝脏房室（scl）模型的基本房室结 构相同。另外，本文假设非酶代谢在所有组织中发生。 scl模型以口服和吸入为苯暴露的途径，该模型 包括了胃房室和肺泡气体交换区域。尿液的代谢物 中，考虑所有单一代房室的代谢物，使用单独的公式 来表示尿液中代谢物排出的量。所有的房室被认为是 充分混合的。 模型使用各组织房室内的静脉平衡模型，即假定 静脉血离开房室时的浓度等于组织内的浓度，静脉血 中化学物质i离开房室j的浓度为：   i ji j i j c cv p  （1） 总的心脏流量等于每个组织房室内的流量之和：   card f s r l kq q q q q q     （2） 在混合的静脉血中苯的浓度用以下公式给出：   ( / bz bz bz bz f f s s r r bz bz l l k k card cv cv q cv q cv q cv q cv q q      ）   （3） 由于苯的血液-空气分配系数相对较低，可用标准 的准稳态空气交换公式表示苯在动脉血中的浓度（当 分配系数大于等于100时，这种关系是不正确的，因 为到达气体交换区域的浓度明显小于吸入的浓度）：   : bz bz bz avv i card avv card bl air q c q cv ca q q p    （4） 口服摄入的苯从胃部到达肝脏，并代谢或分布到 身体其它部分。这是一个单向的交换，即一旦苯到达 肝脏，就不会再回到胃部。摄入量表示为：           8 stomk am （5） 该模型考虑了肝脏和其他组织中的代谢，以及代 谢在肝脏中发生位置的特殊性。多项研究确认了肝脏 的异质性，肝脏能够依据酶的不同组成划分为多个区 域。各区域的氧气和养分物质组成不同。 p4502e1（cyp2e1）主要存在于中心肝细胞，标 记为区域3。因此，cyp2e1介导的代谢仅考虑发生在 肝脏的区域3。cyp2e1在肝脏中的活动： 2 1 3 , 3 1 3 bz bz mpe l l bo l l v c t rm k c d  （6） 2 1 3 , 3 5 3 ph ph mpe l l hq l l v c t rm k c d  （7） 2 1 3 , 3 6 3 ph ph mpe l l cat l l v c t rm k c d  （8） 2 1 3 , 3 7 3 hq hq mpe l l thb l l v c t rm k c d  （9） 2 1ev 表示整个肝脏匀浆中每毫克胞浆蛋白中 cyp2e1的活性，该活性由p硝基代苯酚到p硝基儿茶 酚的氧化确定。 mpc 表示每克肝脏中肝微粒体蛋白质 的量，并且 3 3 31 bz bz ph ph hq hq l l l ld a c a c a c    （10） cyp2e1 在肾脏中的活动表示为： 2 1 , 1 10 bz bz mpe k bo k k k v c rm k c t d  （11） 2 1 , 5 10 ph ph mpe k hq k k k v c rm k c t d  （12） 2 1 , 6 10 ph ph mpe k cat k k k v c rm k c t d  （13） 2 1 , 7 10 hq hq mpe k thb k k k v c rm k c t d  （14） published by atlantis press copyright: the authors 137   其中， 1 bz bz ph ph hq hqk k k kd a c a c a c    （15） 肝脏或肾脏的总质量： 310 /j jt v g l  （16） 其中j可以表示为肝脏（l）或肾脏（k）。 由于苯氧化为苯酚的代谢不涉及酶，可以认为该 代谢发生在所有房室中，并用一阶公式表示。 组织j中bo转化为ph的代谢： , 2 bo bo ph j j jrm k c v （17） 苯环氧化物代谢为苯巯基尿酸时发生了谷胱甘 肽的s-转移。它在肝脏、肾脏、肌肉和心脏中发生。 肝脏中的谷胱甘肽s-转移主要发现在中央静脉的肝 板，因此考虑在充分、不充分灌注组织、脂肪、肾脏、 血液和肝脏的第三个区域中一阶代谢可以用公式 （18）表示。 组织 j 中 bo 转化为 pma 的代谢： , 3 bo bo pma j j jrm k c v （18） 苯环氧化物转化为粘康酸的代谢速率在先前的 模型中认为是一阶的。在这些代谢中介导的环氧化物 酶位于肝脏的小叶中心区（区域3）。 组织j中bo转化为ma的代谢： , 3 4 3 3 bo bo l ma l l v rm k c （19） 与scl模型类似，硫酸盐化作用和糖脂化作用的 假设也应用到模型中。 ph的轭合： 1 1 2 1 , 1 ,1 1 ,2 1 ( ) 3 ph ph ph ph l ph l conj l ph ph ph ph m l m l cp l v c v c rm k c k c t c      （20） 氢醌和它的轭合物的代谢发生在区域 3 中，因 为糖脂化作用在此区域更强。 hq 的轭合： 3 , 3 3 3 hq hq lhq mp l conj l hq hq m l v c t rm c k c   （21） 苯的吸入浓度： (1 ) [ ( ) ] / bz bz e alv i alv card bz bz bz avv i avv c f c f q cv ca q c q         （22） 本文采用的苯及其代谢物在人体内的生理药 代动力学模型框架如图 1。 2.2.2.2 苯生理药代动力学模型参数的选择 （1）分配系数 pbpk 模型需要设定不同组织和血液之间苯环 氧化物、苯酚和氢醌的分配系数。苯的分配系数可 以用于苯环氧化物。此外使用苯酚和氢醌的肝脏血液分配系数作为充分灌注组织-血液的分配系数， 苯、苯环氧化物、苯酚和氢醌的肝脏-血液分配系数 作为肾脏-血液分配系数。苯、苯酚和氢醌的模型中 使用的分配系数来自 karen yokley 等人[27]的研究， 于表 1 中列出。 表 1 苯、苯酚和氢醌的动力学模型中使用的分配系数 （2）其他模型参数 为修正 pbpk 模型使其适合描述苯在人体内代 谢的药代动力学过程，模型中的其他参数需要调整， 见表 2。 人类苯代谢的 pbpk 模型中，部分参数的取值 服从一定的分布，需要根据样本数据估计。需要确 定的模型参数包括受到 p 硝基代苯酚到 p 硝基儿茶 酚的氧化过程确定的 cyp2e1 特定活性 2 1e v ；两种 硫转移的 ph 大代谢速率 1ph v 和 2ph v ；氢醌轭合 参数 数值 参数 数值 : bz bl airp 7.80 ph sp 1.22 ,bz bof fp p 54.50 , ph ph r kp p 2.17 ,bz bol lp p 2.95 hq fp 4.06 ,bz bos sp p 2.05 hq lp 1.04 , , ,bz bo bz bor r k kp p p p 1.92 hq sp 0.94 ph fp 27.63 , hq hq r kp p 1.04 ph lp 2.17 published by atlantis press copyright: the authors 138   （主要是糖脂化作用）的最大速率 hq v 。苯氧化为 苯巯基尿酸（pma）和粘康酸（ma）的一阶 表 2 pbpk 模型中使用的其他参数 代谢速率 k3 和 k4 也因人而异。yokley [27] 利用 mcsim 软件中的蒙特卡洛‐马尔科夫链（mcmc）模块拟合 了这些数据，得到几个模型参数的贝叶斯后验分布。 2 1ev 、 1phv 、 2phv 和 hqv 通过人体肝脏样本的 体外数据分析研究确定，表 3 列出了模拟所需的参 数 分 布 。 正 常 成 年 人 安 静 时 肺 泡 通 气 量 约 为 450.0l/h，心输出量为 336.0l/h。 表 3 pbpk 模型中代谢参数的分布 2.2.2.3 苯的生理药代动力学模型求解 （1）mcsim软件介绍[28] mcsim 软件是代数或微分方程中常用的模拟和统 计工具。该软件是为了实现优化的蒙特卡洛分 参数 值 单位 参数 值 单位 lq 0.02370qcard l/hour lv 0.025bw l fq 0.0425qcard l/hour fv 0.1429bw l kq 0.2027qcard l/hour kv 0.004bw l sq 0.1717qcard l/hour sv 0.4371bw l rq 0.3461qcard l/hour rv 0.040bw l blv 0.07429bw l cpc 14.5 mg/g 1k 4.20×10 -2 l/μmol mpc 58 mg/g 2k 32.16 1/hour ph mk 1, 1.4 μm 5k 4.00×10 -2 l/μmol ph mk 2, 220 μm 6k 2.13×10 -3 l/μmol hq mk 746 μm 7k 2.03×10 -4 l/μmol bza 0.0397 1/μm 8k 374.9598 1/hour pha 1.3×10-2 1/μm 9k 0.1163 1/hour hqa 10-7 1/μm 10k 0.1443 1/hour 代谢参数 参数的分布 3k log-normal,μ=0.60,σ=0.10 4k log-normal,μ=12.1,σ=1.69 2 1ev log-normal,μ=0.012,σ=0.013 1phv log-normal,μ=0.038,σ=0.032 2phv log-normal,μ=0.48,σ=0.38 hqv log-normal,μ=1.05,σ=1.05 吸入  呼出  动      脉            血  脂肪  不充分灌注组织  充分灌注组织  肾脏  胃部  肾脏  cyp2e1 静      脉            血    苯  苯环氧化物 苯酚 对苯二酚    肺泡空间  肺部血液  肝 脏 3  肝 脏 2 肝 脏 1 gst cyp2e1 cyp2e1 cyp2e1 硫化作用 gst cyp2e1 eh gst 脂  肪  不充分 灌注  组织  充分  灌注  组织 血        液 肾脏 pma ma ph 轭合物 血        液  hq 轭合物 儿茶酚 苯三醇 脂 肪 不充分 灌注 组织 充分 灌注 组织 血       液 肾脏 肝 脏 3  肝 脏 2 肝 脏 1 肝 脏 3  肝 脏 2 肝 脏 1 肝 脏 3  肝 脏 2 肝 脏 1 充分  灌注  组织  不充分 灌注  组织  脂 肪 cyp2e1 cyp2e1 糖脂化作用   图 1 苯在人体内代谢的 pbpk 模型 published by atlantis press copyright: the authors 139   析。该软件包括模型生成程序和模拟程序。 （2） 苯的生理药代动力学模型求解 根据上文所述苯的 pbpk 模型及人体模型的相 关参数，利用 mcsim 软件编写代码作为苯暴露的 模型文件。 2.3剂量-反应评价 目前，在剂量-反应评价中，usepa 推荐使用 多阶模型。多阶模型反映了目前普遍认可的致癌作 用过程，即正常细胞在成为恶性细胞前，必须经过 一系列不可逆的转变或阶段。 多阶模型是目前在致癌风险评价当中 广泛运用的 模型。其形式可以表示为：  0 ( ) 1 exp( ) k i i i p d a d     （23） 这里 i=1,2…k 表示 k 阶段过程的不同阶段，d 为暴露剂量。 2.4风险表征 风险表征是定量风险评价四步法的 后步骤， 它通过综合风险辨识、暴露评价和剂量-反应评价的 结果，对有害因子的风险大小做出定量估算与表达， 并对评价结果进行解释和讨论，对评价的可信度或 不确定性加以阐述[29,30]。 根据风险评价四步法的定义，将实际工作场所 中有害物质的暴露水平数据，或通过数学模型、计 算机模拟得到的预测剂量，代入有关剂量-反应关系 模型中，求得暴露人群可能出现不良反应的概率， 即该化学物质对人群造成的健康风险。 0 0pr( ) (1 ) ( )  d p p f d （24） 其中，f（d）为剂量-反应模型。 2.4.1不确定性分析 从风险评价的整个过程不难看出，评价中的不 确定性主要来自剂量-反应评价及暴露评价。许多因 素都会不同程度地影响到评价结果对实际风险的真 实反映，即造成了评价结果的不确定性。 2.4.2蒙特卡洛模拟方法 蒙特卡洛模拟法是目前用于定量风险分析中参 数不确定性研究的 常用方法之一 [31-33] 。利用 mcsim 软件的 montecarlo 部分进行蒙特卡洛模拟， pbpk 模型中的相应代谢参数取表 3 中列出的分布， 输出静脉血和呼出气体中苯的浓度随时间的变化。 输入部分代码如下： montecarlo ("ben.out", 10000, 56761.1164); distrib (k3, lognormal, 1.8221,1.1052); distrib (k4, lognormal, 179871.862, 5.4195); distrib (v_2e1, lognormal, 1.0121,1.0131); distrib (v_ph1, lognormal, 1.0387, 1.0325); distrib (v_ph2, lognormal, 1.6161, 1.4623); distrib (v_hq, lognormal, 2.8577, 2.8577); simulation { # 1: c_bz_i = ndoses（48,dose,0…828,834); printstep (cv_bz, 792, 828, 36); printstep (auc, 792, 828 ,36); } end. 3. 苯职业暴露案例分析 本文选用某石化集团凝析油炼制作业场所作为 苯暴露职业健康风险评价的案例。 3.1项目介绍 该项目为满足增产乙烯裂解原料的需要， 利用常减压蒸馏装置加工南帕斯凝析油。南帕 斯凝析油中硫含量 0.22~0.28w%，为低硫原油。原 油中硫醇的含量不多，一般多存在于轻质馏分中。 生产装置区包括原油预处理及换热系统、 初塔系统、常压系统、吸收塔系统等工艺单元 及储罐区、泵房等公用工程系统等。该工作场 所装置由常减压蒸馏、一脱三注等部分组成。 一脱三注系指原油脱盐、塔顶挥发线注中和剂 （胺）、注缓蚀剂和注水。 工艺原理：石油的常减压蒸馏是利用多元系精 馏原理来实现的。采用常压、减压蒸馏的简单物理 方法将原油切割为沸点范围不同、密度大小不同的 多种石油馏分。各种馏分的分离顺序主要取决于分 子大小和沸点高低。 装置区劳动定员 28 人，分为内操作工及外操作 published by atlantis press copyright: the authors 140   工。内、外操作工均四班二运转，外操作人员承担 巡视、现场仪表控制操作、开关阀门、现场检维修 监管、采样协助与监管、电脱盐系统注有机胺及中 和缓蚀剂、装置区清洁卫生等工作。巡检频率为每 2h 巡检一次，每次 1h 左右，每班在现场累计 4~6h。 采样基本采用密闭采样器，每班采样 2 次。 3.2风险辨识 3.2.1职业危害因素 装置区采用凝析油为炼制原料，凝析油的成分 很复杂，遇热可分解出有毒烟雾，基本成分与其他 原油相似。 原油及其加工过程产生的油蒸气均具有一定的 毒性，因其化学成分、结构、蒸发速度等的不同而 不同，挥发或逸散的油气中主要成分为环烷烃、芳 香烃、硫醇、硫化氢等。在石油炼制过程中产生硫 化氢、硫醇、苯、甲苯、二甲苯、噪声、高温等多 种职业病危害因素。 3.2.2暴露途径 生产过程中挥发或逸散的油气含有芳香烃（以 苯为代表）。苯为明确致癌物质，凝析油加工过程中 石脑油中苯含量高，通过跑冒滴漏等逸散到工作环 境空气中，作业工人长期低浓度职业暴露，不能忽 视苯的致癌风险。 苯在生产环境中主要以蒸气形式经呼吸道进入 人体，皮肤吸收很少，经消化道吸收完全，但实际 意义不大。 3.3暴露评价 3.3.1外暴露评价 基于 fluent 计算机模拟的结果[34]，对于电脱 盐、含硫污水和初顶轻石脑油采样时的苯暴露，均 匀取距离采样点半径 0.5m 圆上的 10 个点；对于跑 冒滴漏状态下的苯暴露，取 8 个跑冒滴漏位置在无 风状态下巡检路线上的 60 个点。8 个点跑冒滴漏情 况下装置区内的苯浓度场分布如图 2 所示。利用 spss 软件分析，上述情况下苯的浓度分布见表 4。 3.3.2 内暴露评价  3.3.2.1生理药代动力学模型 本案例中采用的吸入途径的苯 pbpk 模型包括 6 个房室：脂肪、充分灌注室、不充分灌注室、肝 脏和肾脏（见图 1），不考虑口服和皮肤接触途径。 假设代谢只在肝脏和肾脏中发生。 表 4 正常工况下空气中苯的模拟外暴露浓度和分布 依据流入各房室的血流速率、化学物在房室内 的组织/血分配系数和房室所占的组织体积，可建立 每一房室化学物浓度变化率的微分方程。各房室和 血液中苯浓度的微分方程以及相关模型参数已详细 介绍，此处不再赘述。 3.3.2.2内剂量分布 为了讨论苯职业暴露的终生致癌风险，还要引 入职业暴露的终生平均内剂量浓度的概念，假定工 人寿命为 70 年，在其 30 年的职业中暴露于苯的时 间加权平均内剂量浓度称为终生平均内剂量浓度， 用 cv_bz 终生表示。在 pbpk 模型中，表示为 30 年 模拟场景 分布 参数 苯的浓度 mg/m3 均值 279.25 电脱盐采样 normal 标准差 474.47 均值 285.59 含硫污水采 样 normal 标准差 491.02 均值 745.67 初顶轻石脑 油采样 normal 标准差 285.99 均值 1.92 跑冒滴漏下 的巡检 lognormal 标准差 1.60 图 2 八个点跑冒滴漏情况下装置区内的苯浓度场分布  published by atlantis press copyright: the authors 141   职业暴露中靶器官中内剂量浓度的曲线下面积 （auc）之和除以 70 年寿命的总时长。 30 70 cv _ auc bz  年靶器官内剂量 终生 （25） 假设装置区外操作工实行四班二运转，承担巡 视、现场仪表控制操作、开关阀门、现场检维修监 管、采样协助与监管、电脱盐系统注有机胺及中和 缓蚀剂、装置区清洁卫生等工作。即以 36h 为一个 周期，每周期内各采样点分别采样两次，每次 10min，一班共巡检 5h。 则根据表 4 统计所得，工人每班暴露在苯中的 时间加权平均剂量为： dose=((normalrandom(279.5,474.47)/3+normalran dom(285.59,491.02)/3+normalrandom(745.67,285.9 9)/3+lognormalrandom(1.92,1.64)*5)/6)/78.11; 进行 10000 次蒙特卡洛模拟，输出结果为初次 暴露后第 792h 和 828h 静脉血中的浓度以及这段时 间内的曲线下面积（△auc），见表 5。图 3 所示为 上述暴露条件下，工人体内静脉血中苯浓度随时间 的变化。 职工按照上述条件暴露的静脉血中苯的终生平 均内剂量浓度 cv_bz 终生（mg/m 3 ）可以通过曲线下 面积△auc 计算。 _ = 30 78.11 36 70 auc cv bz    终生 （26） 其中 auc 表示暴露足够长时间后一个周期 内的曲线下面积，μmol/l。 利用 spss 软件对 cv_bz 终生进行统计，经检验 符合均值为 10.1761，标准差为 5.6153 的正态分布， 单位 mg/m3。 3.3.2.3 敏感性分析 为考虑 3k 、 4k 、 2 1ev 、 1phv 、 2phv 、 hqv 对 终生平均内剂量的影响，对这几个参数做敏感性分 析。其方法是，对每个参数取分布的平均值，在保 持其余 5 个参数不变的情况下，观察另一个参数变 化 1%对终生平均内剂量浓度变化造成的影响，选 择 大的敏感性系数作为该变化参数对 auc 的敏 感性系数。敏感性的计算公式如下： δ / = δ / 预 测 值 预 测 值 敏 感 性 参 数 值 参 数 值 （27） 计算可得，这 6 个参数中 2 1ev 的变化对静脉血 中苯浓度及其终生平均内剂量的影响 大。由此可 见， 2 1ev 的取值对苯在人体内的代谢具有十分关键 的影响，在 2 1ev 的取值上应该更加谨慎。在将来利 用苯的生物标志物估计这些代谢参数的过程中， 2 1ev 的估计应该成为研究的重点。 3.4剂量-反应评价 利用 infante et al.和 rinsky et al.对俄亥俄州三 家盐酸橡胶生产工厂工人、ott et al. 和 bond et al. 对 dow 化学工厂工人、中国预防医学会对中国 12 个城市多个行业工厂工人患白血病的流行病学调查 数据进行剂量-反应模型的参数拟合，根据王阳[2]的 研究，确定多阶模型中的参数值 a1、a2。得到内剂 量-反应模型为： 表 5 上述暴露条件下，各参数的 10000 次蒙特卡洛模拟值 iter k3 k4 v_2e1 v_ph1 v_ph2 v_hq cv_bz_1.1 cv_bz_1.2 auc_1.1 auc_1.2 0 1.70912 748329 1.02137 1.07587 2.39201 0.28818 0.00117 0.00117 9.7416 10.1866 1 1.767 167311 1.0246 1.10395 3.65919 4.01923 0.01381 0.01381 115.3 120.567 2 2.04616 1.58624e+06 1.01105 1.02465 1.0261 2.37975 0.01625 0.01625 135.606 141.8 … … … … … … … … … … .. 图 3 外操作工体内静脉血中苯浓度随时间的变化 published by atlantis press copyright: the authors 142   2 2( ) 1 e x p ( )f d a d= − − × （28） 其中， a2= -4.59×10 -5 ，d为内剂量，单位mg/m3。   图 4 多阶模型方法致癌风险分布频率图 根据陈可欣、何敏等人对天津市区 1981 年 1 月~2000 年 12 月白血病流行情况的统计分析，选择 背景反应值为p0=4.71×10 -5 ，拟合后公式可表示为： 5 5 5 2 pr( ) 4.71 10 (1 4.71 10 ) (1 exp(-4.59 10 )) d d − − − = × + − × × − × × （29） 3.5风险表征 利用mcsim软件可以计算得到苯职业暴露的 终生平均内剂量浓度cv_bz 终 生 = normalrandom （10.1761, 5.6153），单位mg/m3。 根据式（29），利用 crystal ball 水晶球软件计 算多阶模型的风险范围如图 4、表 6 和表 7 所示。 表 6 多阶模型致癌风险的统计值 统计量 预测值 均值 0.006164 中值 0.004877 标准差 0.005458 表 7 多阶模型各百分位数对应致癌风险值 3.6 结论 1）通过修正的人类 pbpk 模型，细化了房室的 划分，考虑了酶的影响和代谢产物之间的相互作用。 针对 pbpk 模型参数的不确定性，根据蒙特卡洛模 拟方法，利用 mcsim 软件为从外剂量到内剂量的 计算提供了方法。 2）以某凝析油炼制作业场所为例，利用此研 究方法得出：采样过程中，空气中苯暴露浓度较大， 但接触时间短，跑冒滴漏情况下在装置区内巡检， 暴露的空气中苯浓度较小，但接触时间长；在装置 区 8 个点发生跑冒滴漏以及正常采样的条件下，如 果不采取个体防护，1#常减压生产装置区工人的致 癌风险在 10-5-10-2之间，高于化工行业可接受风险 的水平 10-4的概率在 90%以上。因此，在场区内工 作时要注意以下几点：第一，采样时尽量站在采样 口的上风向，以减少苯的暴露。第二，在浓度较高 的位置采样时，应根据需要佩戴呼吸面罩。第三， 当跑冒滴漏点和跑冒滴漏的量增加，或者工人操作 不当时，各个位置的暴露浓度可能高于假设的模拟 数值，相应的致癌风险也会增加。因此工人要严格 按照规定操作，如果发生泄漏或者严重的跑冒滴漏， 要及时处理和防护。 3）pbpk 模型的人类模型参数仍有部分选用国 外数据，可以通过生物标志物的检测，利用贝叶斯 估计结合 mcmc 模块完善。此外，应用于 pbpk 模型的 mcsim 软件仍有许多功能待开发，将来该 软件可以用于健康风险评价的多个方面。 参考文献 [1] 施婕，突发毒气泄漏事故的急性健康风险分析研究：【硕 士学位论文】，天津：南开大学，2009 j. shi, acute exposure health risk assessment on emergent toxic gas leakage accidents, master thesis, nankai university, 2009 百分位数 预测值 百分位数 预测值 0% 0.000047 10% 0.000468 60% 0.006138 20% 0.001534 70% 0.007906 30% 0.002502 80% 0.009836 40% 0.003606 90% 0.013301 50% 0.004877 100% 0.030686 published by atlantis press copyright: the authors 143   [2] y. wang, m. liu, d.y. huang. health risk assessment for benzene occupational exposure using physiologically based pharmacokinetic model and dose-response model. 3rd international conference on bioinformatics and biomedical engineering, icbbe 2009. [3] 王阳，刘茂，基于生理毒代动力学模型和剂量-反应模 型的苯暴露健康风险评价方法，中国工业医学杂志， 2009，22（1）：34~37 y. wang, m. liu. study on health risk assessment for benzene-exposed workers by physiologically based toxicokinetic model and dose-response model. chinese journal of industrial medicine, 2009, 22(1):34~37 [4] 王阳，刘茂，黄德寅，某涂料生产车间苯暴露的蒙特卡 罗方法评价，中国工业医学杂志，2010，23（2）：141~143 y. wang, m. liu, d.y. huang. assessment on benzene exposure level of certain paint workshop using monte carlo method. chinese journal of industrial medicine. 2010, 23(2):141~143 [5] d.y. huang, j. zhang, m. liu. application of a health risk classification method to assessing occupational hazard in china. 3rd international conference on bioinformatics and biomedical engineering, icbbe 2009. [6] d.y. huang, j zhang, m. liu. research on risk assessment based on monte carlo simulation and dose-response multistage model. 2010 3rd international conference on biomedical engineering and informatics. [7] 黄德寅，张倩，刘茂，苯作业职业暴露评估与致癌风险 模拟，中国工业医学杂志，2011，24（3）：163~168 d.y. huang, q. zhang, m. liu. occupational exposure assessment and carcinogenic risk simulation in benzene workers, chinese journal of industrial medicine. 2011, 24 (3):163~168 [8] 张倩，刘茂，黄德寅，苯暴露的致癌风险评价，中国安 全科学学报，2011，21（5）：143~147 q. zhang, m. liu, d.y. huang. study on carcinogenic risk evaluation of benzene -exposed workers, china safety science journal.2011, 21(5):143~147 [9] d.y. huang, q. zhang, j. zhang.et al. modeling of occupational exposure to benzene and health risk of workers in styrene production. 2011 5th international conference on bioinformatics and biomedical engineering [10] 张倩，黄德寅，刘茂，氯乙烯代谢的生理药代动力学研 究，风险分析和危机反应的创新理论和方法——中国 灾害防御协会风险分析专业委员会第五届年会论文 集，2012-10-27 q. zhang, d.y. huang, m. liu. study on physiologically based pharmacokinetic model of vinyl chloride metabolism. innovative theories and methods for risk analysis and crisis response—proceedings of the 5th annual meeting of risk analysis council of china association for disaster prevention. 2012-10-27 [11] q. zhang, d.y. huang, m. liu. study on carcinogenic risk evaluation for 1,3-butadiene-exposed workers.2012 5th international conference on environmental pollution and public health (epph) [12] 黄德寅，张倩，刘茂.基于内暴露模拟的 1,3-丁二烯致 癌风险评价，风险分析和危机反应的创新理论和方法 ——中国灾害防御协会风险分析专业委员会第五届年 会论文集，2012-10-27 d.y. huang, q. zhang, m. liu. study on carcinogenic risk evaluation based on internal dose simulation. innovative theories and methods for risk analysis and crisis response—proceedings of the 5th annual meeting of risk analysis council of china association for disaster prevention. 2012-10-27 [13] 陈艳，李桂兰，尹松年，苯致血液毒性遗传易感性与外 源化合物代谢酶基因多态性，卫生研究，2002，31（2）： 130~133 y. chen, g.l. li, s.n. yin, individual susceptibility to hematotoxicity from benzene exposure and the genetic polymorphism of metabolic enzymes, journal of hygiene research, 2002,31(2): 130~133 [14] aksoy m. hematotoxicity and carcinogenicity of benzene. environmental health perspectives,1989,82:193~197 [15] aksoy m, erdem s. follow-up study on the mortality and the development of leukemia in 44 pancytopenic patients with chronic exposure to benzene. blood,1978,52(2):285~292 [16] paustenbach d j, bass r d, price p. benzene toxicity and risk assessment,1972-1992: implications for future regulation. environmental health perspectives,1993,101(6):177~200 [17] kaneko t, wang p y, sato a. benzene-associated leukemia and its risk assessment. journal of occupational health,1997,39:159~178 [18] brett s m, rodricks j v, chinchilli v m. review and update of leukemia risk potentially associated with occupational exposure to benzene. environmental health perspectives. 1989,82:267~281 [19] rinsky r a. benzene and leukemia: an epidemiologic risk assessment. environmental health perspectives,1989,82:189~191 [20] bond g g, mclaren e a, baldwin c l, et al. an update of mortality among chemical workers exposed to benzene. british journal of industrial medicine,1986,43:685~691 [21] wong o. an industry wide mortality study of chemical workers occupationally exposed to benzene. i general results. british journal of industrial medicine,1987,44:365~381 [22] wong o. an industry wide mortality study of chemical workers occupationally exposed to benzene. ⅱ dose response analyses. british journal of industrial medicine,1987,44:382~395 [23] yin s n, li g l, tain f d, et al. leukaemia in benzene workers: a retrospective cohort study. british journal of industrial medicine,1987, 44:124~128 [24] yin s n, li g l, tain fd, et al. a retrospective cohort study of leukemia and other cancers in benzene published by atlantis press copyright: the authors 144   workers. environmental health perspectives, 1989, 82: 207~213 [25] cole, c., schlosser, et al. a multi-compartment liver-based pharmacokinetic model for benzene and its metabolites in mice. technical report crsc-tr02-17, center for research in scientific computation, north carolina state university, raleigh, nc. available at www.ncsu.edu/crsc/reports/ftp/pdf/crsc-tr02-17.pdf.2002 [26] cole, c., schlosser, et al. a multi-compartment liver-based pharmacokinetic model for benzene and its metabolites in mice.journal of mathematical biology. [27] karen yokley, hien t. tran, kaija pekari, et al. physiologically-based pharmacokinetic modeling of benzene in humans: a bayesian approach. risk analysis,2006,26:925~943 [28] bois, f. y., maszle, et al. mcsim: a monte carlo simulation program. user’s guide. journal of statistical software,1997,2(9).available at http://www.gnu.org/software/mcsim/mcsim.html. [29] u.s. environmental protection agency. draft final guidelines for carcinogen risk assessment. epa/630/p-03/001a,ncea-f-0644a,2003 [30] u.s．environmental protection agency. guidance for risk characterization, 1995 [31] l.a. cox. reassessing benzene risks using internal doses and monte-carlo uncertainty analysis. environmental health perspectives,1996,104(6):1413~1429 [32] chen s c, liao c m. health risk assessment on human exposed to environmental polycyclic aromatic hydrocarbons pollution sources. science of the total environment,2006,366:112~123 [33] sassi g, magnetti vernai a, ruggeri b. quantitative estimation of uncertainty in human risk analysis. journal of hazardous materials,2007,145:296~304 [34] 黄德寅，刘茂，李敏嫣等，基于 cfd 技术的工作场所 苯职业暴露模拟分析，2012，25（4）：243~246 d.y. huang, m. liu, m.y. li, et al. cfd-based modeling and analysis of occupational benzene exposure at workplace. chinese journal of industrial medicine, 2012, 25 (4): 243~246 published by atlantis press copyright: the authors 145 microsoft word biogas a european perspective on safety and regulation.doc biogas a european perspective on safety and regulation olivier salvi, samuel delsinne european virtual institute for integrated risk management (eu-vri) stuttgart; 70174, germany email: salvi@eu-vri.eu www.eu-vri.eu sébastien evanno accidental risk division, institut national de l’environnement industriel et des risques (ineris) verneuil-en-halatte, france e-mail: sebastien.evanno@ineris.fr www.ineris.fr abstract the production of biogas is positioned as energy which can not only generate a source of energy known as "green" but also which can recycle waste. in the context of sustainable development, the place of biogas is therefore essential. the risks corresponding to the production processes and uses of are still too little known. several questions about the optimization of the production, the safety, the harmonization of the regulations and the need to develop standards are discussed in this paper, based on the results of a workshop on biogas safety and regulation organized in november 2010 by the authors. keywords: biogas, safety, regulation, anaerobic digestion. 1. introduction the development of new energies is experiencing a great development in the world and particularly in europe. as a result, solar panels, wind turbines and other ecological technologies are more and more installed in many european countries and are constantly evolving. the main purpose of this development is to find an alternative energy replacing the fossil energy dependence which is more sustainable and reducing co2 emissions. thus, the impact on the environment is reduced and energy consumption is sustainable. biogas has a special major advantage in addition to other renewable bio-energies. indeed, it reuses the waste as raw materials. the production of biogas is positioned as energy which can not only generate a source of energy known as "green" but also which can recycle waste. in a context of sustainable development, the place of the biogas is therefore essential. however, the processes of anaerobic digestion are continually improving and new ideas for the uses of this gas continue to emerge. however, the risks corresponding to processes of biogas production from biomass or waste are still too little known. several issues have appeared and need to be solved: the optimization of the production, the safety of the processes, the harmonization of the regulations and the need to develop standards (at european or international level). journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 1 salvi, delsinne, evanno to address these issues and find answers to these questions, ineris and eu-vri took the initiative to organize a european working group and a workshop on 24th novembre 2010 in paris. the present document helps to share knowledge, to structure questions, and to propose answers and further actions aiming at improving safety of biogas production and supporting the quick and sustainable deployment of this energy. 2. context of the development of biogas 2.1. background information producing biogas is a recent technology which has two main advantages. indeed, for few decades, governments have tried to develop this new energy because it is a sustainable energy and its raw materials are waste. unlike other renewable energy sectors, biogas production did not emerge from concerns on energy but rather from the environment (elimination of pollution, treatment of waste, control of greenhouse gas emissions). [ref. 1] the development of renewable energy is an important element of the energy policy of the european union. the renewable electricity directive (directive 2001/77/ec[ref. 15]) sets an indicative target of 22% renewable electricity in gross consumption of the eu in 2020 and the biofuel directive (directive 2003/30/ec[ref. 16]) provides indicative targets of 5.75% substitution by biofuels by 2010. the european strategic energy technology plan (setplan) is part of the energy and climate change policy framework. it contributes to the overall policy objectives by proposing, developing and implementing an energy technology policy for europe. it complements eu policy in energy and climate change and can enable cost effective compliance with legally binding targets. these include the realization of the internal market in electricity and gas; compliance with emission reductions and the revised and strengthened emissions trading scheme; the increased contribution of renewable energies to the eu electricity generation; the measures to enhance energy efficiency; the eu car emissions standards; the negotiation of a post-kyoto international agreement; and the development of an external energy policy. [ref. 2] methanization makes it possible to produce biogas from organic elements of vegetal or animal origin. biogas is a gas rich in methane, the same element that constitutes natural gas. biogas can be directly tapped in subsurface containment centers (landfills) or produced using digesters (we can also speak about methanisers in this case). all organic materials except lignin are capable of being transformed into biogas. effluents can be methanized in sewage purification plants. liquid manure, agricultural waste and energy crops can be methanized in small-scale biogas units on farms or in co-digestion units (collective units that treat different type of wastes associated with a significant share of liquid manure). solid municipal waste and green waste can also be transformed into biogas in large-scale solid waste methanization units. european primary energy production from biogas reached 6 million toes (tons of oil equivalents) in 2007, i.e. a 21.2% increase with respect to 2006 (an additional 1 mtoe) [ref. 12]. the increase in biogas production principally benefited electricity produced in chp (combined heat and power) units. with a 24.3% increase in 2007 (+ 2.3 twh produced) chp plant production reached 11.6 twh, and represented a total electricity production which reached 19.9 twh (+ 17.4%). rubbish dump biogas continues to be the principal deposit exploited today: 50.6% of the total. with a 14% share, sewage purification biogas is still behind all of the “other deposits” (principally agricultural biogas units). this type of biogas, the current driving force behind biogas growth in the eu, has the particularity of relying more and more on the development of dedicated energy crops (corn, etc.). these statistics only take into consideration biogas intended to be valorized, and not biogas that is burnedoff in flare stacks. current growth is not strong enough to reach the objective of the european commission white paper (15 mtoe in 2010). at the same time, the trend towards increases in the price of raw agricultural materials should curb growth of agricultural biogas production. taking this situation into consideration, it is estimated a production at 8 mtoe in 2010 (10% annual growth). this production will represent 5.4% of the target of the european published by atlantis press copyright: the authors 2 biogas safety and regulations commission “biomass action plan” (sec(2005) 1573 – [ref. 24]) set at 185 mtoe in 2010. for the future, the target for the share of renewable energies set by the european union is 20% in 2020. in addition, biogas is one of the renewable energies that have most benefited from the strong increase in fossil fuel prices these last two years [ref. 3]. moreover, european union regulations, limiting the discharge of waste in dump/landfills has played a large role in developing this sector, which today has widened its field of action into energy crops. for example, in germany, biogas employs 13 500 persons in direct and indirect full-time jobs and generates a turnover of 650 m€. however, this source of energy is still too poorly studied from a safety point of view. indeed, only few studies have been conducted to define the potential risks involved in producing and using biogas [ref. 3]. 2.2. production 2.2.1. the various raw materials and processes the main idea of the anaerobic digestion is to produce gas from waste. biogas is the name of the gas produced and a biogas plant or anaerobic digester is the place where it is produced. the anaerobic digestion is the process which transforms organic matter into biogases such as methane and carbon dioxide. wastes used in the processes must be organic and biodegradable in order to be digested by microorganisms in the absence of oxygen. various types of organic waste can be found and therefore the following step of the production of biogas will differ because of the various raw materials. so there are several processes for the production of biogas, because it depends on the raw materials, but the main idea is still the same: to produce gas from waste. indeed, biogas could come from several sorts of raw materials [ref. 4]: • sewage sludge • food waste • waste from food industry • manure from cows, pigs etc. • residues from agriculture • ”energy” herbs and plants like maize • distillery by products • organic fraction of municipal solid wastes fig. 1 biogas processes there are different sorts of raw materials and the amount of biogas or the quality of the biogas which will be produced from manure from animals will be different from the one produced from waste in the food industry. thus, the biogas production is very variable from one plant to another. it is also important to highlight the difference between biogas plant and anaerobic digester (biogas from landfills and biogas from digesters). the biogas production is sliced up in four major steps [ref. 1]: • hydrolysis: large polymers are broken down by enzymes • acidogenesis: acidogenetic fermentations are most important, acetate is the main end product. volatile fatty acids are also produced along with carbon dioxide and hydrogen. • acetogenesis: breakdown of volatile acids to acetate and hydrogen. • methanogenesis: acetate, hydrogen are converted to methane and carbon dioxide. fig. 2 process of anaerobic digestion those steps are included in the main processes that are described in a bibliographical study on the anaerobic digestion [ref. 5]. but some projects propose different processes like using aerobic/anaerobic digestion. thus, there is not only one process but several pathways. for anaerobic digestion of farm wastes, usually, there is not “aerobic/anaerobic/aerobic”, but anaerobic and sometimes anaerobic as a post treatment of digestate (composting). in this context, one can wonder if all these processes are equivalent regarding safety and what are the critical phases / steps. 2.2.2. relevant parameters moreover in the anaerobic digestion process we can find several parameters [ref. 1]: • humidity of the substrate: wet or dry • range of temperature: mesophilic or thermophilic • stage of fermentation: single stage or multiple stage • stock flow: continuous or discontinuous • ph control and temperature control published by atlantis press copyright: the authors 3 salvi, delsinne, evanno those examples illustrate the production variability, and it is probable that the yield of the production will depend not only on the raw materials used but also on the process. in addition to the constant improvement of the technical aspects, technology becomes increasingly complex and it is then necessary to control, to measure and to monitor the continuous and instantaneous production with special devices. those devices need to be reliable, accurate and long-lasting. quality or quantity of the gases produced seems to be also different. it depends on raw materials, sort of process, if gases are upgraded and how, or if other gases are removed and how. we previously saw that different raw materials or processes can produce biogas with different yields. but in general, the biogases produced are methane (5075%), carbon dioxide (25-50%), nitrogen (0-10%), hydrogen sulfide (0-4%), hydrogen (0-1%), ammonia (0-500 ppm) and oxygen (0-2%) [ref. 1]. sometimes industries add air in the biogas in order to remove h2s. this is one reason of the residual oxygen in the biogas. see table 1 in source [ref. 31]. biogas contains small amounts of h2s and some other pollutants. h2s is poisonous when inhaled. furthermore, when water is present, h2s forms sulphuric acid (h2so4), which is highly corrosive, rendering the biogas unusable for pipes and reactors. there are different ways to remove the unwanted gases, which mostly target the removal of co2 [ref. 19]: • chemical/ physical absorption • high pressure water scrubbing • pressure swing adsorption • cryogenic separation • membrane separation • addition of air into the biogas • addition of fecl2 in the digester in this part there are a lot of technical aspects which depend on each other: various raw materials mean different processes, different processes mean different amounts of gases, and different amount of gases mean different sorts of up-grading or removing… and finally, it appears impossible to fix a global yield in order to study all uses of biogas or to identify clearly hazards and risks in a biogas plant. then, it is important to understand: what are the safety critical parameters? is it relevant to establish a regulation on the production of biogas or to think about a possible standardization? 2.2.3. profile of biogas plant/ producers throughout the world, a countless number of designs of biogas plants have been developed under specific climatic and socio-economic conditions. choosing a design is essentially part of the planning process. it is, however, important to familiarize with basic design considerations before the real planning process begins. this refers to the planning of a single biogas unit as well as to the planning of biogas-programs with a regional scope. the performance of a biogas plant is dependent on the local conditions in terms of climate, soil conditions, the substrate for digestion and building material availability. the design must respond to these conditions. in areas with generally low temperatures, insulation and heating devices may be important. if bedrock occurs frequently, the design must avoid deep excavation work. the amount and type of substrate to be digested have a bearing on size and design of the digester and the inlet and outlet construction. the choice of design will also be based on the building materials which are available reliably and at reasonable cost. table 1. trace compounds of biogas from different biogas production plants biogas ch4 (%) co2 (%) o2 (%) n2 (%) h2s (ppm) benzene (mg/m^3 toluene (mg/m^3) landfill 47–57 37–41 <1 <1–17 36–115 0,6–2,3 1,7–5,1 sewage digester 61–65 36–38 <1 <2 b.d. 0,1–0,3 2,8–11,8 farm biogas plant 55–58 37–38 <1 <1–2 32–169 0,7–1,3 0,2–0,7 note: b.d.: below detection published by atlantis press copyright: the authors 4 biogas safety and regulations high sophistication levels of biogas technology require high levels of skills, from the planner as well as from the constructor and operators. with a high training input, skill gaps can be bridged, but the number of skilled technicians will get smaller the more intensive the training has to be. in addition, training costs compete with actual construction costs for scarce (project) resources. higher technical sophistication also requires more expensive supervision and, possibly, higher maintenance costs. to which extent prefabricated designs are suitable depends largely on the cost of labor and transport. [ref. 6] there are various types of plants. concerning the feed method, three different forms can be distinguished: • batch plants • continuous plants • semi-batch plants batch plants are filled and then emptied completely after a fixed retention time. each design and each fermentation material is suitable for batch filling, but batch plants require high labor input. as a major disadvantage, their gas-output is not steady. continuous plants are fed and emptied continuously. they empty automatically through the overflow whenever new material is filled in. therefore, the substrate is generally fluid and homogeneous, even though solid feeding is quite common for farm biogas plants. continuous plants are suitable for rural households as the necessary work fits well into the daily routine. gas production is constant, and higher than in batch plants. today, nearly all biogas plants are operating on a continuous mode. up to now, usually farm digesters are fed in semicontinuous mode (several feedings per day). if straw and dung are to be digested together, a biogas plant can be operated on a semi-batch basis. the slowly digested straw-type material is fed in about twice a year as a batch load. the dung is added and removed regularly. [ref. 6] concerning the construction, three main types of simple biogas plants can be distinguished: • fixed-dome plants • floating-drum plants • lagoon-based plants but also other types of plants play a role, especially in past developments. in developing countries, the selection of appropriate design is determined largely by the prevailing design in the region. typical design criteria are space, existing structures, cost minimization and substrate availability. the designs of biogas plants in industrialized countries reflect a different set of conditions. biogas could be produced at different scales: a smallscale by farmers and large-scale by industrials. a lot of parameters depend of the scale such as design, process, and raw materials. moreover, using the technology of biogas production requires knowledge in order to control the process of the production. but in switzerland, courses exist at different levels teaching biogas plant operators and managers of wastewater treatment plants. these have a federal certification. then, in this context, it is important to understand: what is the education in the various countries? how to obtain a permit to operate a plant? is education for operator appropriate? 2.3. uses one of the most interesting advantages of producing biogas is that it has several applications and uses. indeed, because biogas is composed of multiple gases, it is not only used for energy. 2.3.1. combustion the first application of biogas is heat when it burned. the lower heating value (lhv) of the combustion is between 6.0 and 7.0 kwh/m3, whereas the lhv of the natural gas is about 10.3 kwh/m3 [ref. 7]. this heat could be used in the process but biogas could also generate electricity. the electricity yield of cogeneration is around 35%. the yield depends on raw materials and type of biogas plant. and during combustion, some toxic gases are generated. hence, it is necessary to study the composition of the gas and to remove toxic gases or future toxic gases before combustion. 2.3.2. injection in natural gas grid from a technical point of view, those yields are worse than some other forms of energy production and that is published by atlantis press copyright: the authors 5 salvi, delsinne, evanno why the one from biogas still needs to be improved. but for the moment, biogas is a good alternative for replacing fossil energy, because it is a renewable and ecological energy. as yields need to be improved, the safety in biogas applications needs to be further studied. for example, the integration in gas-grid biogas might cause risks as the anses (agence nationale de sécurité sanitaire de l’alimentation, de l’environnement et du travail) explains in its report in 2008, but there is no additional risk when compared to natural gas [ref. 20]. the next paragraphs describe the state of the art in some european countries: in germany, the feeding of biogas into the natural gas grid is an efficient energy solution, even if the sites where gas is used are far away from the sites where it is produced. feeding gas into the grid is facilitated via a compressor, a device raising the pressure level of the biomethane to that of the gas in the closed pressurized lines of the grid. given european regulatory realities, new gas producers have the opportunity to feed gas into the conventional gas grid. for biogas generators, this multiplication of the possible number of consumers is attractive. for purposes of feeding-in, however, the gas must be up to the quality specifications of the relevant legal provisions and may only deviate within the range of these quality standards. such standards are obtained using technologies for reconditioning gas. because a non-negligible quantity of energy is necessary for gas compression, the energy balance and the economic feasibility of the compression and feed-in process must be reviewed on a case-by-case basis. with regard to feeding biomethane into the natural gas grid, it is necessary to distinguish the exchange gas and accessory gas. the difference lies in the quality of the gases. an exchange gas has the same qualitative standards as conventional natural gas and can be exchanged in the grid as such. accessory gas possesses a composition that is not equivalent to that of the natural gas, and can therefore only be mixed into the grid beneath a certain threshold. regulations distinguish between low-quality natural gas (“erdgas l”) and high-quality natural gas (“erdgas h”). erdgas h possesses higher methane content, and is used mainly in the gus federal states and extracted principally in the north sea. erdgas l contains roughly 89% flammable gases (primarily methane, but also small amounts of ethane, propane, butane, and pentane), while erdgas h contains about 97% flammable gases (the same as those listed for erdgas l). the types of natural gas available in germany vary with geography. similarly, the degree to which biomethane is upgraded depends on the region from which it comes [ref. 8]. in sweden, there are 38 sites of biogas injection (laholm, helsingborg, göteborg, stockholm) and several projects of injection because of an economic context very propitious. it is possible to add propane in order to increase the wobbe index [ref. 20]. in switzerland, since 1995, some experiences of injection have been gained. it is not allowed to inject biogas from landfills in gas-grid and to improve the quality of combustion for the other biogases. also there are currently 15 sites of biogas injection. there are two types of injection: • unlimited injection, for a biogas pure at 96% of methane • limited injection, for a biogas composed at least 60% of methane [ref. 20]. in austria, biogas is injected since 2005 in lisbod and there are several projects in progress. however, it is not allowed to inject biogas from landfills [ref. 20]. in france, some experimentation has been performed since 1998 and a project is confirmed in lille sequedin [ref. 20]. in the netherlands, since 1987, landfill biogas is injected in four sites and biogas from sewage sludge since 2006. moreover, experimentation on several upgrading processes has been managed [ref. 20]. in luxembourg 3 large cooperative biogas units are in construction. they will mostly rely on the treatment of organic household/municipal wastes and agricultural residues. the biogas will be upgraded to biomethane using either the water scrubbing or amine absorption techniques. furthermore at least 7 operations in sweden, more than 30 in germany, 5 in austria have been reported but to compare them it is necessary to clarify the definition of an injection site. published by atlantis press copyright: the authors 6 biogas safety and regulations in a nutshell, as reported in ref. 19 several projects have been carried out in order to establish specifications for integrating biogas in gas-grids. they have to be assessed and the results shared to propose a common approach in europe. 2.3.3. biogas for vehicles the biogas could also be used as a fuel in natural gas vehicle. there are currently strong incentives for increased use of renewable fuels in the transport sector worldwide. some bioethanol and biodiesel production routes have limitations with regard to resource efficiency and reduction of greenhouse gases. more efficient biofuel systems are those based on lignocelluloses and novel conversion technologies. a complementary strategy to these is to increase the production of biogas from the digestion of organic residues and energy crops, or from byproducts of ethanol and biodiesel production. compared with other biomass-based vehicle fuels available so far, biogas often has several advantages from an environmental and resource-efficiency perspective. this provides the motivation for further technological development aiming to reduce costs and thereby increase economic competitiveness of biogas as a vehicle fuel [ref. 35]. but the biogas needs to be highly purified. thanks to this technique, replacement of diesel or petrol by biogas reduces the emissions and also the engine noise considerably [ref. 20]. this is an interesting aspect, because the number of vehicle users is growing days after days. but the integration of biogas as a fuel is a recent application and only few projects work on it in europe [ref. 22]. those projects try to study the application with a city scale in 14 cities. but is this possible in a larger scale? thus, biogas vehicle fuel stands out as a promising alternative, together with the ‘second generation’ vehicle fuels that are based on lignocellulose. compared to the majority of the liquid biofuels in use today, biogas often has a far better performance with regard to both area efficiency and life cycle emissions, and it is therefore a strong potential candidate for becoming one of the most sustainable vehicle fuels in the near future [ref. 36] using biogas for vehicle is a promising solution but before using this method for public vehicle, some research on safety is necessary. 2.3.4. some limitations the biogas process is the most complicated of the three biotechnological processes. a consortium of microorganisms catalyses the degradation of complex organic molecules and it results in the production of methane and carbon dioxide as well as some heat. this process remains technically underdeveloped and several crucial aspects need to be addressed. the volumetric productivities need to be raised substantially and the conversion rate of generated digestate improved to reduce the volume of biofertilizer generated. overcoming these limitations will result in the production of increased amounts of gas, while generating less biofertilizer with a higher nutrient content. the volumetric productivities of processing the biomass feedstock to the final biofuels will strongly influence the investments needed to make the technology commercially viable. in the future there will be an increased demand for land to produce biomass for food and animal feed, chemicals, materials and energy. therefore, it is important to prioritize processes, production systems and products that are efficient with regard to the land area used and the use of organic byproducts and wastes, and also according to their environmental impact, particularly in terms of reduction of greenhouse gas emission. 3. safety aspects 3.1. main hazards there are risks in a biogas plant such as explosion, toxicity (leak of h2s) and microbiological risks. the construction of a biogas plant and its maintenance should be well monitored in order to manage risks. prevention of people from being exposed to those risks and checking of all materials (including corrosion) should be realized with the aim of making the production of biogas safer. published by atlantis press copyright: the authors 7 salvi, delsinne, evanno 3.1.1. explosion the risk of explosion is the most studied because it is related to the production and use of an explosive gas which is composed mainly with methane. but some explosions happen regularly. it means that the risk of explosion is always present in a biogas plant because of the mixture air/biogas. a recent incident in uk was caused by the substrate foaming up and ripping the digester hood [ref. 14]. table 3: example of some gases explosion properties gases lel (%v) uel (%v) flame speed (m/s) acetylene 2.5 82 2.66 carbon monoxide 12.5 74 hydrogen 4.1 74.8 2.6 methane 5 15 0.39 source: [ref. 14] 3.1.2. toxicity the second major risk is toxicity related to the leak of h2s, because this gas is very toxic and we know that this gas is one of the most abundantly produced in anaerobic digestion. the most important case of this risk was the accident which happened in zeven in november 2005. four people died because of the exhaust of a cloud of hydrogen sulfide in unusually high concentrations. it has been reported also that biogas plants could cause environmental disasters such as in barßel in september 2002. 20 tons of fish were killed because high quantities of substrate were spread in the environment. in order to have a better use of biogas or to avoid the spread of contaminant gases, the methane has to be removed from other gases such as ammonia, siloxanes… then those gases once separated might be accidentally released into the atmosphere and generate toxic risks. 3.1.3. microbiology the microbial risk is also important and several studies have emphasized this risk. introducing the biogas produced into systems constructed for natural gas is currently causing a debate about the risks of introducing pathogens to the gas systems. finally, risks for disease transmission from processed (upgraded and dried) biogas can be judged as being low. the possible exposure of working personnel is probably the most significant risk. however, this group could be easily identified and informed about potential risks and how they can be managed. the risk of inhaling pathogens when using gas is overshadowed by the risk of gas intoxication and explosions or similar, since these effects would probably occur before a dose of pathogens high enough to cause an infection had been ingested. the biogas produced in the systems analyzed is therefore considered safe to use even in kitchen cookers [ref. 34]. furthermore the microbiological risks also exist at the beginning of the process. i.e. in the handling by the personnel of organic (waste products) therefore it is important to keep in mind the following questions: are the risks of biogas known by the operators? are the safety management systems in place adequate? table 4. example of some gases toxicity thresholds: in european union in united states gases tlv stel tlv stel ld(50) in mg/ m3 ammonia 14 36 17 24 7600 during 2 hours carbon monoxide 55 29 2670 during 4 hours dichloride 1.5 1.5 1 1185 during 1 hour hydrogen sulfide 7 14 14 21 617 during 4 hours tlv: threshold limit value based on normal 8-hour workday – stel: short-term exposure limit. (in mg/ m3) ld(50): lethal dose measured with a group of rats those values might be carefully compare because of the various exposure times. source: [ref. 13] published by atlantis press copyright: the authors 8 biogas safety and regulations 3.2. learning from experience – accident records 3.2.1. accident data bases generally, there is a lack of communication of accidents and their analysis. in fact, when accidents occurred in biogas plants, in general, only the local media describes the circumstances of the accident and little information on the analysis of the causes is disseminated. there is a lack of feedback and learning from experience. 3.2.2. results of research in some data bases several accident databases report about accidents involving biogas productions or uses, such as ari, facts. after a search on aria [ref. 9] and facts [ref. 10] websites, here are feedbacks of some accidents: 08/11/2005 germany – rhadereistedt in a production of biogas from organic waste recovery, an offshoot of hydrogen sulfide (h2s) kills three people and a truck driver who came to unload waste from a slaughterhouse. a severely intoxicated person is hospitalized. the extremely high concentration of h2s in the lobby complicates the response of firefighters and a dozen have suffered of poisoning. an important aeration (more than 24 h) will be required before allowing access to the building. arriving in the evening, the truck from the netherlands was parked outside the hotel until morning. the tragedy occurred when the load of the truck was unloaded within a closed hall to reduce odor nuisance, in a pit of 100 m3 with 2 agitators whose lid can be closed because of the failure electric motor that actuates it. the materials discharged (waste slurries sulfide, ph close to 8.5 and a temperature of 60 °), are guts and viscera of pig. (loaded 24 hours earlier and as usual). the reaction between these substances and materials already in the pit (waste animal or dairy, low ph according to the analysis conducted after the accident) would be the cause of the strong release of h2s. the temperature of the environment and the functioning of the agitation would have favored the dispersal of toxic gas in the lobby of unloading. moreover, the extraction device located at the bottom of the pit which rejects stale air outside through a biofilter would have been insufficient. an investigation has been conducted. 07/01/1999 france la rochette in a recycling unit of biogas from the anaerobic treatment plant of a paper mill, an explosion (5 kg of tnt) buffer destroyed a balloon of flexible material 10 m and their associated piping supplying a boiler or steam flare safety. the balloon exploded, the railings are bent in a radius of 3 m, the tiles are destroyed within a radius of 20 m, cladding and windows on the unit up to 130 m fly into pieces away. there has been no victim. the balloon will be blocked and downhill into depression. air would be entered by teflon joints rubbing on the central axis. the biogas has come back and then has formed the explosive mixture which has been ignited by the pilot flame of the flare. an accidental production of hydrogen in the digester and an act of malice were also discussed. expertise was made. safety devices were then installed (analyzers, valves, etc.). 12/03/1997 italy peschiera del garda in a municipal sewage plant wastewater, an explosion occurred during repair work in a concrete silo of a biogas plant. residue gas and welding operations are the cause of the accident. two workers were thrown out and killed, and a third one falls to the bottom of the building and was seriously injured. the roof of the silo has been blown. 3.2.3. conclusions we can see that safety devices are sometimes not sufficient and some problems appear during maintenance. it is necessary to learn from those experiences in order to never do these mistakes again. that is why it is necessary to communicate accident feedback and risks assessment. the establishment of a permanent network sharing information on causes of accidents involving biogas would be helpful to avoid repeating the same errors. this network should operate at eu or international level. published by atlantis press copyright: the authors 9 salvi, delsinne, evanno 4. regulation 4.1. situation in different european countries the upgrading and feeding of biogas into the natural gas grid is currently not subject of consistent european regulation. the parameters and basic conditions, however, are phrased in directive 2003/55/ec of the european parliament and of the european council of 26 june 2003 concerning common rules for the internal market in natural gas and repealing directive 98/30/ec. in chapter 1 of this directive, biogas and gas from biomass are explicitly included in the scope of the regulation. in paragraph 24 of the preamble, the european parliament and the european council declare the following: “member states should ensure that, taking into account the necessary quality requirements, biogas and gas from biomass or other types of gas are granted nondiscriminatory access to the gas system, provided such access is permanently compatible with the relevant technical rules and safety standards. these rules and standards should ensure that these gases can technically and safely be injected into and transported through the natural gas system and should also address the chemical characteristics of these gases” [ref. 8]. as of 2009, not all european countries have implemented regulations on the upgrade and feed-in of biogas. 4.1.1. germany the political framework for the feed-in of biogas to the natural gas grid in germany is defined by a variety of regulations which reflect the complex value chain stages with different challenges on every stage. an overview can be found here. the most influential parameter is the ongoing implementation of the energy and climate programme of the german federal government. since upgraded and fed-in biogas is currently not yet on a competitive basis with natural gas, the energy and climate programme provides a toolbox to develop the market demand. the legal framework envisages a broadly diversified application field for biomethane, including pure heat applications, combined heat and power generation with state-guaranteed feed-in tariffs according to the renewable energy source act (eeg), as well as applications as biofuel in natural gas-dedicated vehicles. the energy and climate programme consists of 29 courses of action focusing on climate protection. the central measures in context with the injection of biogas are listed in the following: section 2: amendment to the renewable energy source act (eeg) the amendment to the eeg became effective on january 1 2009. section 9: regulations on the injection of biogas into natural gas grids this section was implemented on april 12 2008 by means of the ordinance to promote the use of biogas in the existing gas network. the ordinance includes the following measures: • gas network access ordinance (gasnzv) • gas network tariffs ordinance (gasnev) • incentive regulation ordinance (aregv) and • electricity network tariffs ordinance (stromnev) with the commencement of the adapted gas network access ordinance on april 12 2008, the conditions for the feed in of biogas to the natural gas grid were decidedly improved. moreover, in paragraph 41a, the ordinance defines the goal early phrased in the energy and climate programme on the promotion of the injection of biogas: until the year 2020, up to 60 billion kwh biogas are to be fed every year into the natural gas network, and 100 billion kwh biogas until 2030. this dimension corresponds with the potential for this technology as determined in an influential study by the german technical and scientific association for gas and water (dvgw) of 2006. section 14: renewable energy heat act (eewärmeg) the renewable energy heat act (eewärmeg) also became effective on january 1 2009. section 17: promotion of biofuels • amendment of the biofuel quota act • ordinance on requirements regarding the sustainable generation of biomass for biofuel applications: biomass sustainability ordinance (bionachv) the law on renewable energy was amended on the 6th of june 2008. from 2009 onwards, the digression of the feed in tariff for biomass will be changed from 1.5% to published by atlantis press copyright: the authors 10 biogas safety and regulations 1%. the feed in tariff for one kwh of electricity coming from biomass for plants under 150 kwe will amount to €c 11.67/kwh. the premium for electricity produced from agricultural biogas will be raised from €c 6/kwh to 7 for plants below 500 kwe. to this will be added a premium for production units using more than 30% liquid manure (€c 4/kwh for plants under 150 kwe and €c 1/kwh for plants under 500 kwe) or a premium of €c 2/kwh if the main source is waste coming from the cleaning of natural green spaces for units below 500 kwe. the premium for cogeneration will be raised from €c 2/kwh to 3 for plants below 20 mwe and the technology premium will be maintained. another innovation will be that units producing purified biogas will benefit from a premium which will depend on the plant efficiency: €c 2/kwh for units producing 350 nm3 per hour and €c 1/kwh for units producing up to 700 nm3 per hour. the new law also foresees the increase of the feed in tariff for electricity produced from landfill gas for units up to 500 kwe, thus passing from €c 7.11/kwh to €c 9/kwh and the continuation of the €c 6.16/kwh tariff for plants up to 5mw. the tariff will remain unchanged for sewage sludge biogas: €c 7.11/kwh up to 500 kwe and €c 6.16/kwh up to 5 mw. in parallel, the german government passed a new law on 12 march 2008 concerning injection into the gas network, in order to promote the injection of biomethane (purified biogas). the aim of this new law is to replace 10% of natural gas consumption with biogas as of 2030. as for electricity produced from renewable sources, the law gives biomethane suppliers priority for injecting their output into the natural gas network. it also stipulates that a large share of the costs shall be borne by the network operator and not by the supplier. 4.1.2. switzerland in switzerland, the svgw (“schweizerischer verband des gasund wasserfaches“, swiss association for gas and water) has published the directives g 13, “directives on the injection of biogas into the natural gas network”. these directives define requirements for the quality of biogas from fermentative processes and of similar gases if an injection into the natural gas network is planned. furthermore, the directives define requirements regarding the feed-in technology plants. the overall goal of the directives is to ensure the reliability of the feed-in plants, the natural gas distribution networks as well as installed gas units. the directive suva (safety in biogas plant) is another regulation of switzerland. 4.1.3. france the renewable energy objectives in france are defined by the 13 july 2005 law which plans that 10% of the french energy needs will be insure by renewable sources by 2010. 3 clear targets are defined in order to meet the overall objective. firstly, the share of national renewable electricity production should reach 21% of the electricity consumption by 2010. in 2007 this share is about 12.8% (source: the 10th worldwide electricity production from renewable energy sources). secondly, a 50% increase in the heat production coming from renewable sources. and thirdly, a 7% increase in the share of biofuels used for transport. two important instruments have been put in place to fulfill the objectives. a system of purchase prices for the renewable electricity sector and an income tax credit (reimbursement by the tax departments) for household equipments. a new law on environment perspectives named ‘grenelle 1’ is in way to be voted by the french assembly. this law confirms the willingness of the country to diversify its energy sources and to meet the long-term target of 20% of renewable energy by 2020. the production of primary energy from biogas is equally divided between biogas from landfill sites (161.3 ktoe in 2007) and biogas from waste treatment plant (144.2 ktoe). other biogas sources represent only 1.2% of the total (3.78 ktoe). an order dated 10 july 2006 defines the feed-in-tariff of electricity produced from methanization biogas. it varies from 9 c€/kwh for installations of less than 150 kwe to 7.5 c€/kwh for installations of more than 2 mwe. in addition, there is an energy efficiency incentive of 3 c€/kwh for installations that optimize heat recovery and/or electricity generation. this rate is not applicable to landfill site biogas which (under an order dated 1st october 2001) benefits from a buying price of between 4.5 c€ and 5 c€ per kwh, plus an energy incentive of between 0 c€ and 0.3 c€ per kwh. currently, the feed-in of biogas to the natural gas grid is not yet permitted in france, although (or possibly published by atlantis press copyright: the authors 11 salvi, delsinne, evanno precisely because) the feed-in of upgraded landfill gas was carried out in context with the montech project in 1998. this project, however, was stopped due to discussions on the necessary gas quality and possible gas network contamination. the "agence nationale de sécurité sanitaire de l’alimentation, de l’environnement et du travail" has spoken in favour of the injection of biomethane into the natural gas grid on october 29 2008 with the prospect of applying it as biofuel. a specific regulatory framework coming from the anses conclusions on the admission of biogas to the natural gas network has not yet been published. the technical conditions for the access to the natural gas grids of the gas network operators are, however, already created in such a way as to principally allow the injection of renewably generated gases. 4.1.4. sweden in sweden, the ss 155438 standard “motorbränslen – biogas som bränsle till snabbgående ottomotorer”, which became effective in 1999, regulates the quality criteria for biogas utilised as biofuel. biogas type a refers to vehicles without oxygen sensor; type b applies to vehicles with an oxygen sensor, also called lambda sensor. similarly, this standard also applies to the injection of biogas into the natural gas networks. comparably to the injection of substitute gas in germany, however, an adaptation of the fuel value is required before access to the gas network is permitted. the backdrop of this regulation focusing on usage as vehicle fuel is twofold: the biomethane produced in sweden is almost exclusively used as biofuel, and the swedish gas network infrastructure is less developed than it is, for example, in germany. thus, it has so far been possible to inject biogas into the natural gas network only in some areas on the swedish west coast. only a few injection projects have therefore been implemented so far. no rules are regulating injection of biogas to the gas grid. upgraded biogas is only injected into the distribution grid and not the transmission grid. in general the rules for natural gas are valid for plants connected to the natural gas grid. deals about injection and connection are closed between plant owners and grid owners. deals also cover amounts, price and quality of the gas. work is in progress concerning e.g. quality responsibility. the green gas principle is used and biogas can be sold anywhere along the natural gas grid even though the costumer does not receive biogas, but gets natural gas. there is work in progress on this subject. in sweden the biogas production has been more or less constant the past years, which is also true for the total number of plants. however, there is a change in where the biogas comes from. the production from landfills decreases as expected, while the production from sewage treatment plants and co-digestion plants increases. upgrading of biogas and injection to the gas grid in sweden, there are 8 sites of injection in sweden among the 40 upgrading plants (it seems that the number of injection sites have been mixed up with the number of upgrading plants). there are 115 filling stations. in sweden the biggest environmental benefit is gained, by upgrading biogas and replacing fossil transport fuels. 19% of the produced biogas is utilized as vehicle fuel. during 2007, 28 million of nm3 biogas was sold as vehicle fuel, which is more than the amount of natural gas (about 25 million nm3) sold as vehicle fuel in 2007. 14400 vehicles (buses, heavy cars, light cars) were riding on gas (biogas and natural gas) in 2007. end of december 2007 there were 86 public gas filling stations and 27 gas fill-in stations for buses, now those numbers have reach a total of 115 filling station policy measures the introduction of biogas as vehicle fuel in sweden has been realised by many governmental and local support measures, such as: • a 30% investment support from governmental investment programmes to many biogas plants; • a special investment support of m€ 15 for biogas filling stations during 2006 and 2007; • exemption of biogas use as vehicle fuel from tax; • relative low natural gas tax of 1.4 €ct/kwh (3.9 €/gj) compared to a tax of 7.7 €ct/kwh (21.4 €/gj) for petrol; • free parking for biogas cars in several cities like gothenburg; • readiness of the gas suppliers in trying to keep the biogas price about 20-30% below the equivalent price of petrol, as long as the market for biogas as vehicle fuel is under development. published by atlantis press copyright: the authors 12 biogas safety and regulations • new car tax since october 2006. tax based on fuel and co2 emissions. gas models reduced with about 50% compared to petrol. • obligation to provide biofuels at filling stations. • 1100 € subsidy for new eco-cars. • exception from congestion tax in stockholm from 1 august 2007 for gas vehicles, ethanol and electricity hybrids. regarding the environmental targets, the government’s policy is that 85% of all new governmental cars should be eco friendly, as well as 25% of all new emergency services vehicles. 4.1.5. austria renewable energy policy in austria exists on three levels: the federal level, the regional level of the provinces (bundesländer) and the local level of municipalities. important contributions from renewable energy sources regard large hydropower for electricity, biomass for heat and power and solar thermal installations. on the federal level the programme klima:aktiv (started 2004 by the federal ministry of agriculture, forestry, environment and water management) aims at reducing co2-emission and increasing the penetration of renewable energy sources. regarding renewables, focus is on biogas and biomass, solar thermal and heat pumps. typical for the austrian programmes is that they are easily accessible through the internet and through telephone hotlines. the klima:aktiv programme has biogas as one of the priorities. special focus in this programme is on providing information (conferences, networking events, and training), communication and research. in 2007 62 waste landfill gas recovery plants, 134 sewage sludge digesters, about 350 (agricultural) codigestion plants, 25 industrial anaerobic waste water treatment plants, and about 15 biowaste digestion plants from municipalities were in operation in austria. these plants produced between 265 and 414 million nm3 biogas, from which 45% from (agricultural) codigestion plants, and 47% from landfills and sewage sludge. based on the feed-in tariffs set in 2006, it is possible to get a subsidy of 17 €cents/kwh with an installed capacity of 100 kw or less guaranteed for 10 years. as a result of economic pressure through increased raw material costs, rarely new plants were opened in 2007. in austria, the injection of biogas is possible either via the existing gas network or via a local micro gas network. the injection of landfill gas or sewage gas into the public gas grid is currently generally not permitted. the political framework for the injection into the public gas grid is defined in the following legal regulations: the gas economy act (“gaswirtschaftsgesetz“) obliges gas grid operators to grant grid access to biogenic gases. these gases do need to fulfill the quality criteria defined in the “general distribution network conditions” to ensure safe transport within the austrian gas network. regulations g 31 and g 33 define these criteria. övgw regulation g 31 (övgw = austrian association for gas and water) defines all those parameters and quality criteria, which are necessary to ensure safe transport. moreover, the regulation determines the required gas fuel value data. the quality criteria phrased in the regulation refer to the characteristics of the imported gas. övgw regulation g 33 applies to the injection of biogas based on renewable processes into the natural gas networks. this regulation therefore defines the quality of renewable gases and the quality control which are preconditions for the feed-in. besides the fact that some aspect of the production of biogas in general has a regulation, there is no specific regulation for biogas plants. indeed, there are regulations at the plant construction, at the safety level as atex or at economic level. but as regards the safety to the production and use, the regulation is virtually nonexistent. 4.2. atex directive because there is methane in biogas, biogas production is concerned by the atex directives. there are two atex directives (one for the manufacturer and one for the user of the equipment) [ref. 11]: • the atex 95 equipment directive 94/9/ec, equipment and protective systems intended for use in potentially explosive atmospheres; • the atex 137 workplace directive 99/92/ec, minimum requirements for improving the safety and health protection of workers potentially at risk from explosive atmospheres. published by atlantis press copyright: the authors 13 salvi, delsinne, evanno ineris is a major referent in the implementation of these guidelines at european and international levels. regarding equipments, some companies provide devices, which are used in biogas production, with the atex labels. but at small-scale of biogas production, those devices are hardly used. concerning the workplace, obligations are complex and require special attention. safety measures have to be well managed because of the presence of possible explosive atmosphere and in order to satisfy the minimum requirements. 4.3. bref – “slaughterhouses and animals byproducts industries” [ref. 22] the bref has been published in may 2005, by the european commission. this report is a reference document on best available techniques (co-called bref) in the slaughterhouses and animal by-products industries and particularly on biogas production. in single stage digesters, biogas cannot be produced from pure animal material because the nitrogen content is too high. animal waste must, therefore, be mixed with other organic matter to reduce the nitrogen content. in denmark, approximately 75 % of the biomass resource for anaerobic digestion is animal manure, with the remainder mainly originating from food processing, including slaughterhouses, although some segregated domestic waste is also treated. animal by-products, manure and the sewage sludge from slaughterhouses can all be treated. it is also reported in this report that there is a risk of an accident release of ch4, which is a greenhouse gas and has a global warming potential 30 times greater than that of co2. but the advantages associated with biogas production from slaughterhouse by-products include: a reduction in the concentration of impurities in waste water, low excess sludge production and the production of a biologically stable excess sludge that can be used as a fertilizer. the solid residues from biogas production from animal by-products may be composted. the use of such compost is subject to the restrictions specified in abp regulation 1774/2002/ec. for each unit of electricity generated from biogas, 1.5 units of heat may be produced as hot water at over 80 ºc. the energy-rich gas can be used, e.g. in slaughtering or animal by-products production, as a substitute for conventional primary energy. biogas is reported to make no net contribution to the greenhouse effect. this document reports that a biogas plant could generate noise pollution. indeed, large-scale mechanical equipment such as compressors, used to aerate the process fluid and filtration plant may be potential sources of noise pollution. 4.4. results of commonalities and differences the first commonality which we can see in the different regulations is that governments want to increase the number of biogas plants and the use of biogas for the next years in order to enforce the european directive and energy plans. but differences appear at the next step: the price of energy, such as electricity, is not the same in the various countries of europe. it results that specifications and requirements are different, because legal amount of energy production is linked to its price. concerning the thresholds of dangerous trace compounds in biogas, regulations are broadly similar. finally, the main problem of biogas related to safety is that regulations are enforced for large-scale production but there is a lack of enforcement of those one for small-scale production and, in general, it is because of a lack of skills. in addition, the enforcement practices can vary from one member state to the other, and therefore they have to be harmonized to guarantee the same development of biogas in a single european market. 5. standardization/ guidelines the application of standards makes an industry more efficient and safer and this could be an interesting solution to improve safety in biogas plants. for larger biogas programs, especially when aiming at a self-supporting dissemination process, standards in dimensions, quality and pricing are essential. standard procedures, standard drawings and forms and standardized contracts between the constructor, the planner, the provider of material and the customer avoid mistakes and misunderstandings and save time. there is, published by atlantis press copyright: the authors 14 biogas safety and regulations however a trade-off between the benefits of standardization and the necessity of individual, appropriate solutions. but nowadays there is no biogas standard. the standardization administration of china has made a proposal to the international organization for standardization (iso) for a new field of technical activity on biogas in january [ref. 23]. the table of contents of this standard would be: • biogas glossary • designing, construction, commissioning and check and test of small biogas facilities (household biogas pool) • designing, construction, commissioning and check and test of largeand middle-scale biogas plants • designing, manufacturing, installation and inspection of biogas equipment • designing, manufacturing and inspection of products for biogas application • designing, manufacturing, installation and inspection of equipment and facilities for biogas power generation • comprehensive use of digested solid and liquid • appraisal on technical, economical and environmental benefit of biogas facilities the european committee for standardization has created a working group in order to find a common standard on the injection of non-conventional gases into gas networks. the reference of this group is cen/tc 234/wg 9. the injection of biogas into gas grid is concerned by this research. 6. outcomes of the workshop as part of the work described in this paper a workshop on “biogas – safety and regulation” took place in paris on november 24, 2010. the workshop was organized by ineris and eu-vri and gathered 34 experts from all over europe. the main objectives of the workshop were: • share information, in particular on projects, processes but also on safety and regulation in order to identify the state of the art in these fields across europe. • find and understand the gaps in terms of safety and regulation. • propose solutions to fill those gaps and possibly prepare the base of a permanent network on safety aspects. the main outcomes of the workshop are summarized in the next paragraphs. 6.1. safety issues the first part of the event was dedicated to the safety issues. indeed, some research on safety has been led by institutes like ineris and by industrial companies such as gdf-suez or rhodia in order to analyze the risks and to improve the safety along the production and the uses of biogas. some risks appear important and well studied such as explosion and hydrogen sulfide release but it has been reported that the microbiological risk is also important and not well studied. for example, the safety distances based on the effects corresponding to the hazardous phenomena (heat flux, overpressure and toxicity) have to be considered in order to analyze and quantify the risks. those distances depend on the pressure, the composition of the biogas and the type of biogas (from: agricultural waste, municipal waste, wastewater treatment plant…). the study made by ineris shows that the agricultural process can be in some conditions more dangerous than the other processes [ref. 36] regarding the uses of biogas, gdf-suez has carried out some studies which demonstrated that the cng (compressed natural gas) could be less dangerous than the diesel, looking at the criticity which combines the severity and the probability of the events. the tank of vehicle which uses biogas contains safety devices. moreover, there exists a regulation r110 which describes the requirements of the tank aspects and its tests. the risk analysis is essential to develop safely the uses of biogas. considering the current situation, from the discussions during the workshop, it can be concluded that safety has to be improved based on the existing regulations that have to be better enforced. it also appeared necessary to increase the awareness of the workers and promote prudent behaviors by involving them in the safety management process: participation in risk analysis (hazop, preliminary risk published by atlantis press copyright: the authors 15 salvi, delsinne, evanno analysis…), analyze near-misses and accidents; raise level of knowledge related to safety… (see 6.3). biogas production or use generates several various risks such as explosion, toxic releases but also microbiological risks, often neglected. risk analysis is a key element of the safety management system of biogas facilities and has to be performed involving the workers. 6.2. key production parameters and data the risk analysis enables to identify the key parameters which are relevant to ensure safe production facilities and safe uses. some of these key parameters are subject to regulation. in the framework of several projects, data and key production parameters are produced and compiled in some european institutes such as hsl, ineris or sp. these projects are running in parallel without exchanges at the moment. some parameters are relevant regarding the safety or the production: for example, the pressure and the temperature in the digester. it has also been reported that the presence time in the digester and the load balancing are important for the production efficiency and the safety. the amount of contaminants in the biogas is also relevant but there is no special best upgrade and moreover it depends on the raw materials. thus the composition of the raw materials has to be controlled and certified. the composition of the raw material and of the biogas is very important because they play a major role in the risks analysis. pooling data is important to ensure a quick return on investment and to start up collaborative projects on production, safety and uses connected with biogas. during the workshop, based on the presentations, it was suggested to share some findings and data when possible and to explore the opportunity and benefits to compile those data in a database at the european scale and on all sectors of biogas taking into account that those pieces of information are sensitive. this database could support the regulations and the standardization. from the workshop it appears necessary to create at the european scale a database gathering relevant key production parameters, data and safety methods to ensure the quick and safe development of the biogas sector. 6.3. learning from experience learning from experience based on near-misses and accidents is necessary to avoid repeating mistakes or errors. it is part of the risk analysis process to improve safety of the operation. the feedback of siaap presented during the workshop was a great opportunity to understand the benefits of the learning from experience process. siaap had an accident some years ago and they have worked on a method to learn from this experience and have started the investigation and analysis. then, siaap has updated all the plants: production devices, maintenance, and approach of risks. in addition, during the investigation, they realized that the regulation applying to their installations is very fragmented and the enforcement difficult at the design phase (by engineering companies). thus they decided to create a guideline on regulations and practices in order to make easier the implementation of the safety requirements. the example of the naturalgaskielen in luxembourg shows that agricultural facilities can be safe if they implement standards (german standards). with its strict method of risks analysis derived from the chemical industry, rhodia has observed a reduction of the number of accidents. mostly, accident root causes are linked to human factors, mistakes or errors in the design, when building or during operations of the facilities. thus rhodia has focused on the safety program on its employees. the safety approach used by rhodia consists in using tools in order to analyze the risks, to qualify and quantify objectively the effects. this approach helps to consider all safety aspects. learning from experience and sharing it may force people to consider safety as a major challenge. published by atlantis press copyright: the authors 16 biogas safety and regulations the workshop concluded that safe production or uses improve the image of the biogas sector and will improve its efficiency. creating a network and a platform on biogas safety and sharing best practices will improve safety performance and support the quick development of the biogas industry. 6.4. safety of biogas production processes safety aspects are difficult to assess because they depend on several elements, such as the types of technology, types of facilities, the size of the site, the upgrading technology and the lay-out of the installations, and the staff and organization in place. there is also a combined productivity and safety problem due to the raw materials inconstancy (amount and composition). thus it is important to perform a specific risk analysis which will depend on the composition of the products and on the specific features of the facility. considering accident investigations it appears that often accident sequences are similar in their root causes and consequences. thus there is a clear added value in sharing experience, feedback and results of accident investigations, and also the risk analysis in a global approach for all situations. in a nutshell, in spite of the fact that each plant is specific, it is possible to define a global safety approach and a common risk analysis by sharing information, methods and data. for safety in the regulation, it is relevant to notice the difference between the production and the uses and between the physical process and the biological one regarding safety. the complexity of the safety assessment is due to the fact that each plant is specific because of the technology, the staff and organization in place and the local context. thus it appears necessary to perform risk analysis for each installation and to share results of accident investigations among the personnel, but also in a larger forum. 6.5. implementation of the existing regulation after a short introduction on this issue, it was obvious that the implementation of safety regulations on biogas plants is not harmonized throughout europe. in addition, the main message from the discussion on the implementation of the existing regulation is that the main problem regarding the regulation is not the regulation itself but its implementation. ignorance, fragmentation and in some cases lack of enforcement are observed. most of the participants agreed that the regulation on biogas has to be developed in a way that it could support the safe development of the biogas industry. because of the diversity of the processes and uses, it seems necessary to adopt rather a goal-based approach instead of a descriptive approach. this will give more flexibility and efficiency to achieve a high safety performance. at the end of the workshop, a remark on the quality of waste was pointed out. waste is becoming a “precious raw material”, for example luxembourg imports waste. thus it could be necessary to have a certification of “quality” for the waste. indeed a certification on waste quality will allow a better control of the digestion and improve the biogas quality. it is important to notice that the presence of toxic gases depends on the waste “quality”. the existing regulation seems appropriate but should be better implemented. goal-based regulation appears more effective than descriptive regulation because of the diversity of technologies and configurations, and because the technological improvement is still on-going. 6.6. sharing best practices and education of the operators one of the current problems which have been identified is the lack of knowledge related to risks for the persons involved in the biogas sector, in particular for the agricultural biogas. thus, it is necessary to improve education and dissemination of knowledge on safetyrelated aspects. however, it has to be done in a careful way without demonizing biogas. there is also a technical problem due to detection equipments which operate in a “contaminated environment” and not in a “clean air”. indeed, the measurement of hydrogen sulfide in the biogas is published by atlantis press copyright: the authors 17 salvi, delsinne, evanno different as in the air. sensors for detections and measures have to be carefully selected. it is important to disseminate information related to the risks in biogas to operators, or farmers but also design engineers because everyone has to be aware, and sometimes, even at high-level, risks are not well-known. indeed, in general, simple-looking facilities such as biogas plants might contain similar safety critical situations, as complex facilities. in all cases, risk analysis has to be performed and risk analysis methods and tools have to be made available to engineers and operators working on biogas. there was an agreement that a single accident with severe consequences happening in the biogas industry may damage the whole sector, the whole industry image. thus, the participants agreed to play an active role in sharing information and promoting safety. it is important to define guidelines on best practices at european level to improve convergence if not harmonization in the various european countries. in france, regulations (icpe 2781-1 regulation and atex regulation) adopted recently requires that staff working in the field of biogas has to undergo safety courses. this is an interesting example of measures to address the problem of lack of safety knowledge and awareness. dissemination of information on risks connected with biogas is necessary. every employee should be involved in the safety assessment. best practices have to be better shared to develop rapidly and safely the biogas energy in europe. the participants recommended to create a platform to exchange on safety issues. 7. conclusions and perspectives this paper presents a review of the situation on biogas production and uses, and the perspectives that it offers in the context of the recent european energy policies. it points out that proper safety management and adequate regulation will ease the quick deployment of this technology, but there are still some issues to be solved. regarding the production and uses of biogas, it is important to discuss and share best practices and to reach a harmonization at least at eu level. the main insurance for the safety in the biogas sector is the risk analysis. thus it is important to work on the improvement of the systematic performance of risk analysis. it is important to create a network and a database on biogas safety in order to develop safely the biogas sector. because a major accident could have a negative impact for each actor in the biogas industry, these network and database will give the impulse to share knowledge and best practices to prevent any accident. thus it is necessary to pursue in this way and create a platform to share safety information and best practices. the workshop was fruitful and experts have shared information and ways of thinking about safety and regulation. this action should be repeated in a broader context. acknowledgements this paper was elaborated with the contribution of the members of the programme committee of the workshop presented in the paragraph 6. they are here acknowledged for their contribution: arrhenius karine (sp technical research institut, sweden), coenen joep (rhodia, netherlands), fouquet rémi thiébaut charles gauchet lucille (meddtl, france), greninger aude (gdf-suez, france), hoffmann lucien & delfosse philippe (crp gabriel lippmann, luxembourg), kay jake (hsl, united kingdom), marchais caroline (atee / club biogaz, france), membrez yves (erep sa, switzerland), moletta rené (moletta methanisation sas, france), nedelec ronan & riotte michel (siaap, france). published by atlantis press copyright: the authors 18 biogas safety and regulations references references are to be listed in the order cited in the text. since the paper addresses a topic in quick development a lot of the information was taken from the web. a. websites visited last time in november 2010 1. www.eubia.org/108.0.html 2. http://ec.europa.eu/energy/technology/set_plan/set_plan_ en.html 3. www.energies-renouvelables.org/barometre.asp 4. www.unu.edu/unupress/unupbooks/80434e/80434e0j.ht m 5. www.3a-biogas.com/infocenter/executive_summary.htm 6. www.arizonaenergy.org/altenergyclub/biogas%20%20application%20and%20product%20development.ht m 7. www.biogas-renewableenergy.info/biogas_composition.html 8. www.biogaspartner.de/index.php?id=10093&l=1 9. www.aria.developpement-durable.gouv.fr/ 10. www.factsonline.nl/ 11. www.ineris.fr/fr/ressources/recherche/atex 12. www.eurobserv-er.org 13. www.osha.gov/sltc/healthguidelines 14. www.reptox.csst.qc.ca/rechercheproduits.asp b. document in portable document file (pdf) version 15. http://eurlex.europa.eu/lexuriserv/lexuriserv.do?uri=oj:l:2001 :283:0033:0033:en:pdf 16. http://eurlex.europa.eu/lexuriserv/lexuriserv.do?uri=oj:l:2003 :123:0042:0046:en:pdf 17. www.lth.se/fileadmin/energiportalen/filer/om_energipor talen/invigning/posters/energy_analysis_of_biogas_prod uction_maria_berglund_p_l_b_rjesson.pdf 18. http://students.chem.tue.nl/ifp24/biogasposterweb.pdf 19. www.afsset.fr/upload/bibliotheque/058412092985118807 049572120143/biogaz_oct08.pdf 20. www.ieabiogas.net/dokumente/biogas%20upgrading.pdf 21. http://213.131.156.10/xpo/bilagor/20040115134708.pdf 22. http://eippcb.jrc.ec.europa.eu/reference/brefdownload/do wnload_sa.cfm 23. http://publicaa.ansi.org/sites/apdl/documents/news%20a nd%20publications/links%20within%20stories/iso%2 0tsp%20211%20%28biogas%29.pdf 24. http://eurlex.europa.eu/lexuriserv/lexuriserv.do?uri=com:200 5:0628:fin:en:pdf 25. www.kas-bmu.de/publikationen/kas/kas_12.pdf 26. www.ineris.fr/centredoc/dradrc93-scenarios-accidmethanisation-web.pdf 27. www.ineris.fr/centredoc/rapport_biogaz_web.pdf 28. www.ineris.fr/centredoc/methaniseur_agricoles_medad_ biogaz_web.pdf 29. www.ineris.fr/centredoc/reacteur-pilote-methanisationweb.pdf 30. www.ineris.fr/centredoc/risques-sanitaires-biogazweb.pdf c. books 31. s. rasi, a. veijanen, j. rintala. trace compounds of biogas from different biogas production plants, energy, volume 32, issue 8, august 2007, pages 1375-1380 32. a. jaffrin, n. bentounes, a. m. joan, s. makhlouf. landfill biogas for heating greenhouses and providing carbon dioxide supplement for plant growth, biosystems engineering, volume 86, issue 1, september 2003, pages 113-123. 33. b. vinneras, caroline schonning, annika nordin. identification of the microbiological community in biogas systems and evaluation of microbial risks from gas usage, science of the total environment, volume 367, issues 2-3, 31 august 2006, pages 606-615. 34. m. poeschl, shane ward, philip owende, prospects for expanded utilization of biogas in germany, renewable and sustainable energy reviews, volume 14, issue 7, september 2010, pages 1782-1797 35. p. borjesson, bo mattiasson. biogas as a resourceefficient vehicle fuel, trends in biotechnology, volume 26, issue 1, january 2008, pages 7-13. 36. s. evanno, accidental scenario and modeling of effects distances connected with agricultural-scale and industrial-scale biogas plant. ref. ineris dra-09101660-12814a, available at www.ineris.fr published by atlantis press copyright: the authors 19 salvi, delsinne, evanno fig. 1 biogas processes fig. 2 process of anaerobic digestion biogas process raw materials production final products byproduct waste •sewage sludge •food waste •waste from food industry •manure from cows, pigs… •residues from agriculture fermentation/ anaerobic digestion gases solid / liquid biogas  cleaning biogas  utilization cogeneration • electrical energy • thermal energy biogas upgrading • feeding in gas grid • vehicle fuel direct burning • cooking • lighting • cooling • heating storage separationtransport liquid :   •fermentation • spreading solid : compost  spreading gardeningagriculture anaerobic digestion carbohydrates fats proteins sugars fatty acids amino acids carbonic acids and alcohols hydrogen carbon dioxide ammonia hydrogen carbon dioxide acetic acid carbon dioxide and methane hydrolysis acidogenesis acetogenesis methanogenesis published by atlantis press copyright: the authors 20 microsoft word re-the volatility of viet nam listed banking the volatility of market risk in viet nam listed banking, insurance and financial services company groups after the financial crisis 2009-2011 dinh tran ngoc huy banking university, hcmc, viet nam international university of japan, niigata, japan e-mail: dtnhuy2010@gmail.com abstract the viet nam economy, especially the stock exchange, has been influenced by the global crisis during and after the period 2007-2009. how much risk for a typical industry in an emerging market such as viet nam? for specific industries, such as banking, insurance, investment and security industries, the risk re-analysis and estimation for the listed firms in these industries become necessary, esp. after the crisis period 2009-2011. first of all, by using quantitative and analytical methods to estimate asset and equity beta of four (4) groups of 32 financial service listed companies in viet nam banking, insurance, investment and security industries with a proper traditional model, we found out that the beta values, in general, for most companies are acceptable, excluding a few cases. there are 81% of listed firms with lower risk, among total 32 firms, whose beta values lower than (<) 1. this number is higher than that during the crisis period 2007-2009 (just 69%, see exhibit).secondly, through the comparison of beta values among four (4) above industries, we recognized there are still 19% of total listed firms in the above group companies with beta values higher than (>) 1and have stock returns fluctuating more than the market index. finally, this paper generates some outcomes that could provide both internal and external investors, financial institutions, companies and government more evidence in establishing their policies in investments and in governance. keywords: equity beta, financial structure, financial crisis, risk, asset beta, financial service industry. 1. introduction together with such profitability parameters as roi, roe, firm risk can be used as an investment parameter for investors before they make an investment on either developing or developed financial markets. hence, we perform a market risk analysis based on asset and equity beta of total 32 listed companies in the category of banking, insurance, investment and security firms. this paper emphasizes on analyzing un-diversifiable risk in the above industry in one of emerging markets: vietnam stock market during the financial crisis 2007-2009. after the previous published article on estimated beta for listed construction company groups, we will compare the estimated beta results of listed viet nam banking institutions to those in its comparative activities such as insurance, investment and security companies to make a comparative analysis and risk evaluation after financial crisis impacts. no research, so far, has been done on the same topic. this paper is organized as follow. the research issues and literature review will be covered in next sessions 2 and 3, for a short summary. then, methodology and conceptual theories are introduced in session 4 and 5. session 6 describes the data in empirical analysis. session 7 presents empirical results and findings. then, session 8 gives analysis of risk. lastly, session 9 will conclude with some policy suggestions. this paper also provides readers with references, exhibits and relevant web sources. 2. research issues we mention a couple of issues on the estimating of beta for banking, insurance, investment/financial service and stock investment companies in viet nam stock exchange as following: hypothesis/issue 1: among the four (4) companies groups, under the financial crisis impact and high inflation, the beta or risk level of listed companies in the stock investment group will relatively higher than those in the rest three (3) industries. hypothesis/issue 2: because viet nam is an emerging and immature financial market and the stock market still in the recovering stage, there will be a large disperse distribution in beta values estimated in these four (4) industries. hypothesis/issue 3: with the above reasons, the mean of equity and asset beta values of these 4 listed group companies tend to impose a high risk level, i.e., beta should higher than (>) 1. 3. literature review fama, eugene f., and french, kenneth r., (2004) also indicated in the three factor model that “value” and “size” are significant components which can affect stock returns. as luis e. peirero (2010) pointed, the journal of risk analysis and crisis response, vol. 3, no. 3 (november 2013), 127-134 published by atlantis press copyright: the authors 127 willieb typewritten text willieb typewritten text willieb typewritten text willieb typewritten text received 18 october 2013 willieb typewritten text accepted 21 october 2013 willieb typewritten text willieb typewritten text dinh tran ngoc huy task of estimating cost of equity in emerging markets is more difficult because of problems such as collecting data in short periods. grullon (2012) pointed, there is evidence that the positive relation between firm-level stock returns and firm-level return volatility is due to real options that firms possess. then, becker and schmidt (2012) stated that the influence of past stock price movements on volatilities and correlations is essential for understanding diversification and contagion in financial markets. joshi (2012) found the results which reveal that the asian stock markets exhibit the persistence of volatility, mean reverting behavior and volatility clustering.. next, wang et all (2013) failed to find a significant relationship between volatility of volatility and the variance risk premium. bekaert (2013) found out the variance premium predicts stock returns while the conditional stock market variance predicts economic activity and has a relatively higher predictive power for financial instability than does the variance premium. and fernandez (2013) also stated that industry betas are very unstable. 4. conceptual theories determinants of equity and asset beta in financial markets, systematic risk relates to the overall risk of the whole market, is affected by some factors such as: interest rate fluctuations or economic crisis, can not be avoided by diversification, and is measured by a financial metric, beta which is also called systemic risk. when an investor decides to make an investment in a single company and in a specific stock market, he or she will think of how much risk of the investment. or what is the beta value of the stock or investment? this research will answer that question. of course, risk involves 2 parts: systematic risk (beta) and unsystematic risk, which is business risk or financial risk or diversifiable risk in which investors can reduce it by diversification. another application of beta is that it is used in the capital asset pricing model (capm) to value a firm before making an m&a and for pricing assets. one example to see the meaning of beta is that, if beta of a real estate firm equals to 1,5, the risk of the firm will be 1,5 times higher than the entire market and the return from the investment into this firm might be high equivalently. 5. methodology we select the 2009-2011 period as it is the time highlighting impacts from the financial crisis. therefore, we use the data from the stock exchange market in viet nam (hose and hnx) during these two (2) years to estimate systemic risk results. firstly, we use the market stock price of total 32 listed companies in the banking, insurance, investment and security industries in viet nam stock exchange market to calculate the variability in monthly stock price in the same period; secondly, we estimate the equity beta for these four (4) listed groups of companies and make a comparison. thirdly, from the equity beta values of these listed companies, we perform a comparative analysis between equity and asset beta values of these 4 companies groups in viet nam. finally, we use the results to suggest policy for both these enterprises, financial services institutions and relevant organizations. the below table 1 gives us the number of banking and other financial service (insurance, investment and security) firms used in the research of estimating beta: 6. general data analysis we analyze the data (table 1) of a sample of total 32 firms in 4 categories of industries: banking, insurance, investment and security groups, and the mean of equity beta is valued at 0.648 while that of asset beta is about 0.347. these data are acceptable values during the crisis. furthermore, the sample variance of asset beta is quite low (0.101) which is a good number, while that of equity beta is a little bit higher (0.181). this shows us that the effectiveness of using financial leverage has decreased the systemic risk for the whole industry. however, the max and min values of beta are still somewhat large (table 2). max equity beta value is up to 2.025 that is a little bit high, compared to max asset table 1 – descriptive data of listed firms in the research market listed banking companies (1) listed insurance companies (2) listed investment &financial service companies (3) listed stock investment companies (3) note (4) viet nam 6 4 7 4 estimating by traditional method 3 3 3 2 estimating by comparative method total 9 7 10 6 total firms in groups: 32 (note: the above data is at the december 12th, 2010, from viet nam stock exchange) published by atlantis press copyright: the authors 128 the volatility of viet nam listed bank beta value is just 1.493 that is acceptable (see table 2). looking at table 3, we can see there is 19%, or 6 listed firms still have beta values larger than (>) 1, whereas there is 81% or 26 firms whose beta values lower than (<) 1 and higher than (>) 0. table 2 – estimating beta results for four (4) viet nam listed banking and other financial service companies groups (as of dec 2011) (source: viet nam stock exchange data) statistic results equity beta asset beta (assume debt beta = 0) difference max 2.025 1.493 0.5318 min 0.114 0.037 0.0764 mean 0.648 0.347 0.3008 var 0.1814 0.1011 0.0803 note: sample size : 32 table 3 – the number of companies in research sample with different beta values and financial leverage equity beta no. of firms financial leverage (average) ratio <0 0 0,00% 0% 0<beta<1 26 51.32% 81% beta > 1 6 58.32% 19% total 32 52.6% 100% asset beta no. of firms financial leverage (average) ratio <0 0 0,00% 0% 0<beta<1 31 53.48% 97% beta > 1 1 26.26% 3% total 32 0.7% 100% value of equity beta varies in a range from 2.025 (max) to -1.592 (min) and that of asset beta varies in a range from 1.478 (max) to 0.114 (min). some companies still has larger risk exposure than most of the others. there are no listed companies whose betas are lower than (<) 0, which means no firms has the stock return moving in an opposite direction to the market index. next, asset beta max value is 1.493 and min value is 0.037 which show us that if beta of debt is assumed to be zero (0), the company’s financial leverage contributes to a decrease in the market risk level. lastly, we can see the relatively high difference between max equity and max asset beta values, which is about 0.532, whereas there is a smaller difference between equity and asset beta variance values which is just 0.076; so, there is certain impact on systemic risk of certain firms in term of using leverage while it indicates for most of firms that financial leverage can enable them to reduce market risk. and there is not quite big effect from financial leverage on the gap between company’s beta variance values. 7. empirical research findings and discussion a-banking listed companies group after the crisis 2007-2009, the market for these institutions still exists and grows, but has certain difficulties. the rising inflation and rising lending interest rates and higher opportunity costs made input materials or production costs increasing. so, the market for borrowing firms has been affected because selling prices increase. table 4 below shows us the research of 9 listed firms in this category during the above period. in general, the mean of equity beta and asset beta are 0.492 and 0.130, accordingly. these values are good numbers in term of indicating a low and acceptable un-diversifiable risk. the market demand for financial services is still high. besides, the variance of equity and asset beta of the sample group equals to 0.073 and 0.005 accordingly which are lower than the variance of the entire sample equity and asset beta of 0.181 and 0.101. the effect from financial leverage makes these beta values fluctuate a little bit less than the sample beta mean. we might note that equity beta values of 9 firms in this material category are a little lower than those of firms in the rest two (2) groups: finance and stock investment, but higher than that of insurance firms. this might be considered as one characteristic of these industries. among four (4) industries, the systemic risk of banking group companies is a bit higher than that of insurance group. besides, the estimated equity beta mean is 0.492 and sample variance is 0.130, which is not supporting our 2nd research hypothesis or issue that there would be a large disperse distribution in beta values estimated in this industry as well as our 3rd research hypothesis or issue that the mean of equity and asset beta values of these listed companies tend to impose a high risk level or beta should higher than (>) 1. (see table 5). binsurance listed companies group in an emerging market such as viet nam, the market for insurance firms is definitely potential because of the public need for such necessary vital products and though it may be affected by impacts from the financial crisis. table 6 below shows us the equity and asset beta mean of 7 listed insurance companies, with values of 0.469 and 0.214, accordingly. this result, which means the risk is low and negative because there are only one among 7 firms with beta > 0. we note that equity and asset beta var have acceptable values of 0.162 and published by atlantis press copyright: the authors 129 dinh tran ngoc huy table 6 – estimating beta results for viet nam listed insurance companies (as of dec 2011) (source: viet nam stock exchange data) order no. company stock code equity beta asset beta (assume debt beta = 0) note financial leverage 1 bvh 0.966 0.252 73.9% 2 pvi 0.558 0.345 38.1% 3 abi 0.288 0.104 63.8% 4 bic 0.114 0.037 abi as comparable 67.3% 5 bmi 1.063 0.627 41.0% 6 pgi 0.150 0.067 abi as comparable 55.2% 7 pti 0.145 0.063 abi as comparable 56.7% average 56.6% 0.046. this indicates the good effect from using financial leverage. table 5 – statistical results for vietnam listed banking  companies  statistic results equity beta asset beta (assume debt beta = 0) difference max 1.004 0.267 0.7371 min 0.125 0.037 0.0878 mean 0.492 0.130 0.3623 var 0.0733 0.0050 0.0682 note: sample size : 9 the table 6 below shows us the equity and asset beta mean of 7 listed insurance companies, with values of 0.469 and 0.214, accordingly. this result, which means the risk is low and negative because there are only one among 7 firms with beta > 0. we note that equity and asset beta var have acceptable values of 0.162 and 0.046. this indicates the good effect from using financial leverage. beside, this is the group with the lowest equity beta mean. please refer to table 6 and table 7 for more information. cinvestment & financial service listed companies group among 4 groups, this is the group with the highest number of listed firms (sample size = 10) and with the highest equity beta value of about 0.965. however, the asset beta mean of about 0.524 is a little lower than that of stock investment industry. the using of leverage has influenced these firms’ risk a bit more than the stock investment group. different from firms in the insurance industry, 10 listed firms has the highest equity and asset beta var values, estimated at 0.265 and 0.167, which implies there is a more dispersion in market risks among firms in this industry. the equity and asset beta values are distributed in a proper range, from 0.230 to 2.025, and from 0.116 to 1.493 which are acceptable, esp., asset beta values are low, indicating the effectiveness of using financial leverage. (see table 8 and table 9). table 4 – estimating beta results for viet nam listed banking companies (as of dec 2011) (source: viet nam stock exchange data) order no. company stock code equity beta asset beta (assume debt beta = 0) note financial leverage (f.s reports) 1 acb 0.636 0.152 95.2% 2 ctg 0.554 0.136 94.4% 3 eib 0.385 0.106 90.5% 4 hbb 0.346 0.098 shb as comparable 89.7% 5 mbb 0.234 0.061 stb as comparable 92.5% 6 nvb 0.125 0.037 hbb as comparable 87.7% 7 shb 1.004 0.267 91.8% 8 stb 0.734 0.204 90.3% 9 vcb 0.408 0.106 92.7% average 91.6% note: raw data, not adjusted published by atlantis press copyright: the authors 130 the volatility of viet nam listed bank dstock investment listed companies group many firms in this category have difficulties in their operation during these years. different from firms in the other three (3) industries, 6 listed stock investment firms has the lowest value of equity beta var of 0.0375 and the 2nd lowest value of asset beta var of 0.048, showing market risk with less dispersion. max beta values of 0.76 and 0.756 are the low and acceptable values with the small difference of beta means of 0.028. this indicates a less impact from using leverage in level of market risks among firms in this industry (compared to 0.440 in the investment and finance group). the asset beta mean value is 0.533 (the highest) and equity beta mean value is 0.561 (the 2nd highest in 4 groups) shows the financial leverage has less impacts on market risk exposure in this category during the crisis period, compared to the other industries. please refer to table 10 and table 11 for more information. table 7 – statistical results for vietnam listed insurance companies statistic results equity beta asset beta (assume debt beta = 0) difference max 1.063 0.627 0.4354 min 0.114 0.037 0.0764 mean 0.469 0.214 0.2554 var 0.1622 0.0462 0.1160 note: sample size : 7 table 8 – estimating beta results for viet nam listed investment & financial service companies (as of dec 2011) (source: viet nam stock exchange data) order no. company stock code equity beta asset beta (assume debt beta = 0) note financial leverage 1 agr 1.370 0.313 77.2% 2 apg 0.648 0.630 cls as comparable 2.8% 3 aps 0.895 0.382 57.4% 4 avs 0.546 0.425 cls as comparable 22.1% 5 bsi 1.125 0.873 agr as comparable 22.4% 6 bvs 2.025 1.493 26.3% 7 cls 0.662 0.331 50.0% 8 cts 0.812 0.546 32.8% 9 pvf 1.334 0.116 91.3% 10 vnr 0.230 0.131 43.0% table 9 – statistical results for vietnam listed investment & financial service companies statistic results equity beta asset beta (assume debt beta = 0) difference max 2.025 1.493 0.5318 min 0.230 0.116 0.1135 mean 0.965 0.524 0.4407 var 0.2650 0.1672 0.0978 note: sample size : 10 table 10 – estimating beta results for viet nam listed stock investment companies (as of dec 2011) (source: viet nam stock exchange data) order no. company stock code equity beta asset beta (assume debt beta = 0) note financial leverage 1 asiagf 0.419 0.274 mafpf1 as comparable 34.6% 2 mafpf1 0.586 0.583 0.4% 3 prubf1 0.265 0.264 0.3% 4 vfmvf1 0.760 0.751 1.2% 5 vfmvf4 0.758 0.756 0.4% 6 vfmvfa 0.580 0.572 mafpf1 as comparable 1.4% published by atlantis press copyright: the authors 131 dinh tran ngoc huy table 11 – statistical results for vietnam listed stock investment companies statistic results equity beta asset beta (assume debt beta = 0) difference max 0.760 0.756 0.0050 min 0.265 0.264 0.0009 mean 0.561 0.533 0.0281 var 0.0375 0.0481 -0.0106 note: sample size : 6 e-comparison among 4 groups of banking and other financial service companies in the below charts (figure 1, figure 2 and figure 3), we can see among the 4 groups, equity beta values of the banking and insurance groups are the lowest (0.49 and 0.47) and asset beta values of these groups are also the lowest (0.13 and 0.21), while equity beta mean of investment and finance group is the highest (0.96) and asset beta mean of the stock investment is the highest (0.53). assuming debt beta is 0, financial leverage has helped many listed firms in these industries lower the un-diversifiable risk. 0,96 0,52 0,27 0,17 0,000 0,200 0,400 0,600 0,800 1,000 1,200 equity beta mean asset beta mean equity beta var asset beta var banking insurance investment & finance stock investment figure 1 – statistical results of four (4) groups of 32 listed vn banking and other financial service firms after the crisis period 2009-2011 additionally, we could see the asset beta mean values of all 4 groups are not big and lower than (< 0.6) and acceptable. therefore, it also rejects our 3rd hypothesis that the mean values of equity/asset beta of all 4 groups impose higher risks. next, we can recognize from the chart that, the risk in the stock investment industry higher than those in the other industries. so, it supports our 1st hypothesis. last but not least, from the calculated results, variance of asset beta in the financial service industries are low (vary in range of 0.01 to 0.17). this also rejects our 2nd hypothesis. 0,235 0,000 0,200 0,400 0,600 0,800 1,000 1,200 e q u ity b e ta m e a n a ss e t b e ta m e a n e q u it y b e ta v a r a ss e t b e ta v a r banking insurance investment&fina nce stock investment figure 2 – statistical results of four (4) groups of 32 listed vn banking and other financial service firms during/after the crisis period 2007-2011. 1,15 0,41 0,81 0,41 0,37 -0,40 -0,20 0,00 0,20 0,40 0,60 0,80 1,00 1,20 1,40 equity beta mean asset beta mean equity beta var asset beta var banking insurance investment&finance stock investment figure 3 – statistical results of four (4) groups of 32 listed vn banking and other financial service firms during/after the crisis period 2007-2009 8. risk analysis generally, the financial crisis will affect the whole economy and slowdown the speed of oda and fdi capital invested into financial service industries in viet nam. additionally, it has some negative impacts on viet nam stock market which is in the down turn. and it also might have indirect impacts on banking industry through macro factors such as interest rates, exchange rates. from financial service industry and banking industry, the level of crisis impacts can move to other markets such as real estate market. however, there are positive efforts from market participants. for example, banks and insurance firms have created many more joint products and services. this enables companies to enhance their brand names. published by atlantis press copyright: the authors 132 the volatility of viet nam listed bank 9. conclusion and policy suggestion a-banking industry this is the group with the 2nd highest number of listed companies (9 firms). in is noted that beta mean values are fine (0.49 and 0.13 accordingly), and this is the industry which has both the 2nd lowest asset beta mean value and the lowest asset beta var (0.13 and 0.01) (see figure 1). during the crisis, this industry has lower market risk and beta values of firms in the group are less fluctuated. after increasing rates period, financial services industries, the government and central banks have certain efforts and proper policies to support businesses and internal investors, and stabilize inflation. b-insurance industry generally speaking, this is the industry which has the lowest value of equity bet mean, among 4 groups (0.47) and the 2nd lowest value of asset beta var (0.046). all 7 firms have beta values lower than (<) 1 and there are one firm whose beta value is higher than (>) 0. the market is established and data is acceptable for the period 2009-2011. furthermore, the risk dispersion decreases from 1.15 during the period 2007-2009 (see figure 3) to 0.17 during the period 2009-2011 and 0.24 for the whole period 2007-2011 (see figure 2). c-investment & financial service industry through our comparative analysis on asset beta values, this is the industry which has the highest number of firms (10 firms) and which has the highest equity beta mean and the 2nd highest asset beta mean (0.96 and 0.52 accordingly), as well as the highest values of equity and asset beta var (0.27 and 0.17). but it has lower asset beta mean than that of the stock investment industry. d-stock investment industry in our comparative analysis on asset beta values, this is the industry which has the smallest number of companies (6 firms) and which has the lowest values of equity beta var of 0.037 and the 2nd lowest asset beta var of 0.05. this shows us the lower level of dispersion of market risk compared to other industries although the leverage tends to expand the risk exposure. on the other hand, asset beta mean value of 0.533 ranks the highest and equity beta mean of 0.56 ranks the 2nd highest among 4 groups. when we consider impacts of financial leverage, market risk exposure and dispersion is lower than that of the other three (3) industries. but the leverage has more effects on the investment and financial service industry when we note the equity/asset beta mean values of these 2 groups. in general, our empirical findings state that they are not in favor of our 2nd and 3rd hypotheses or research issues but support our 1st hypothesis. in short, though viet nam is an emerging market with growing financial system, the equity beta values estimated are at acceptable level with 81% firms in the research sample while just a few companies’ beta values are risky (about 19% firms). in term of asset beta values, there are 97% of firms whose beta in a range 0 1, and only 3% or 1 firm whose beta > 1. additionally, it indicates the higher the using of financial leverage, the lower the beta values. in reality, there are 81% of financial service firms (26 among 32 firms) which has 0<beta<1 in this research sample. if used effectively, using leverage can be good for risk management. furthermore, if we compare these data and values to those of construction and real estate firms, and to those of computer and electrical companies in our previous research (see exhibit 4 and 5), for the crisis period 20072009, we might see that in here, the equity beta mean of investment and finance groups can be a little bit higher than those of computer and electrical firms but those of banking and insurance could be lower, whereas asset beta mean of these firms are lower than those of construction and real estate firms and while the impacts from the crisis happens on the overall market. so, the leverage becomes more meaningful and the crisis might have less influence on the firms in the above research. last but not least, looking at figure 1 and 3, it is noted that both equity and asset beta mean values of the insurance firms varied from negative to positive and have the same movement direction with the market. finally, this paper suggests implications for further research and policy suggestion for the viet nam government and relevant organizations, economists and investors from current market conditions. acknowledgements my sincere thanks are for the editorial office and lecturers/doctors at banking university, hcmc, viet nam and at international university of japan, niigata, japan and arhus business school, denmark, as well as amos tuck school of business, usa. through the qualitative analysis, please kindly email me if any error found references 1. becker, c., and schmidt, w.m., (2012), volatility, correlation, and the market trend, ssrn working paper 2. bekaert, g., and hoerova, m., (2013), the vix, the variance premium and stock market published by atlantis press copyright: the authors 133 dinh tran ngoc huy volatility, columbia business school research paper no. 13-32 3. eugene, fama f., and french, kenneth r., (2004), the capital asset pricing model: theory and evidence, journal of economic perspectives 4. eugene, fama f., and french, kenneth r., (2006), the value premium and the capm, journal of finance 5. fernandez, pablo., (2008), levered and unlevered beta, ssrn working paper series 6. grullon, g., lyandres, e., and zhdanov, a., (2012), real options, volatility, and stock returns, journal of finance 7. huy, dinh t.n., (2012), estimating beta of viet nam listed construction companies groups during the crisis, journal of integration and development 8. joshi, p., (2012), financial crisis and volatility behaviour of stock markets of asia, questjournal of management and research 9. pereiro, luis e.,(2010), the beta dilemma in emerging markets, journal of applied corporate finance 10. wang, r., kirby, c., and clark, s.p., (2013), volatility of volatility, expected stock return and variance risk premium, ssrn working paper 11. http://www.mofa.gov.vn/vi/ 12. http://www.hsx.vn/hsx/ published by atlantis press copyright: the authors 134 atlantis press journal style on the renewal risk model with constant interest force k.k. thampi department of statistics, snmc, m.g.university, kerala-683516, india. email : thampisnm@yahoo.co.in m.j. jacob department of mathematics, nitc, calicut-673601, india. email : mjj@nitc.ac.in abstract in this paper, we consider a renewal risk model with constant interest force for an insurance portfolio. we discuss equations for the survival probability and its laplace-stieltjes transforms have been obtained. we provide recursive algorithm for the upper and lower bounds for the ruin/survival probability under interest force. finally, we derive an exponential integral equation for the survival probability. some special cases are also discussed. keywords: constant interest force, generalized exponential distribution, laplace-stieltjes transform, probability of ruin, recursive calculation. 1. introduction the classical compound poisson surplus process, it is often assumed that the surplus receives no interest over time. but the large portion of the surplus of the insurance companies comes from investment income. the impact of investment risk on the ruin probability and other issues of both theoretical interest and practical importance. in risk theory, there is considerable interest in the study of investment income. in this paper we use results given by sundt and tuegels (1995, 1997) and the ideas given in jingmin et.al (2011) and cai and dickson (2002) to study the ruin problems for the generalized exponential distribution. we aim to find the bounds for the ultimate ruin probability by recursive technique. yang (1999) considered a discrete time risk model with a constant interest force and a non-exponential upper bounds for ruin probabilities were obtained. paulsen and gjessing (1997) considered a diffusion perturbed classical model, a lundberg type inequality was obtained by assuming a stochastic investment income. recently there has been considerable interest in extending results from classical risk theory, to more flexible general models. a general model involves the assumption that the inter-claim times are independent and identically distributed, but not necessarily exponential. the resulting surplus process is referred to as a renewal risk process or so called sparre andersen model proposed by sparre andersen (1957). analysis of sparre andersen process is more difficult than the classical model, but remarkable progress has been made . one of such class of distribution for the interclaim times is erlang distribution. various aspects of ruin in erlangian risk models are studied in dickson and hipp (1998, 2001), cheng and tang (2003), sun and yang (2004) and li and garrido (2004a). thampi et al.(2007) have considered another renewal risk process in which claims occur as generalized exponential distribution.the most related reference on the ruin probability for renewal risk model corresponding to our case is dong and wang (2006) which delivered integrodifferential equation for the survival and ruin probabilities with negative risk sums. they obtained journal of risk analysis and crisis response, vol. 3, no. 2 (august 2013), 88-94 published by atlantis press copyright: the authors 88 mailto:thampisnm@yahoo.co.in mailto:mjj@nitc.ac.in willieb typewritten text received 24 december 2012 willieb typewritten text willieb typewritten text willieb typewritten text accepted 11 june 2013 willieb typewritten text exact expression and upper and lower bounds for the ruin probability. there exist a vast literature on the classical risk model with constant interest force. compared to this there are a remarkably few papers on renewal risk model under interest force. one obvious reason is that the theory is much more complicated in the renewal risk model. we mention that there are papers which are devoted to the ruin probability of renewal risk model with interest force. we do not plan to cite here a complete list of references. recently, konstantinides et al. (2010) considered the ruin probability with the constant interest force δ . they established an asymptotic expression for the ultimate ruin probability. here we apply the methodology in sundt and tuegels (1995), cai and dickson (2002) and dong and wang (2006) to derive bounds for ruin/survival probabilities for a particular class of renewal risk model under interest force. the purpose of this paper is to derive some explicit expression for ruin/survival probabilities for a particular class of renewal risk process under interest force. we assume that the claims inter-arrival times have generalized exponential distribution. we show that techniques that can be applied to produce explicit results for ruin probabilities in classical risk model under interest force can also be applied when interclaim times is generalized exponential. we have obtained an integral equation satisfied by the ruin/survival probability under interest force. also we derive a couple alternative expressions, an exponential integral equation for the survival probability and the other a second order differential equation satisfied by the laplace-stieltjes transform of the non-ruin probability. δψ (u) > 0 the outline of the paper is summarized as follows. in section 2, we set out the mathematical preliminaries that help to describe the model in subsequent sections. in section 3, we apply the technique developed in sundt and tuegel (1995) to derive a differential equation for the survival probability in the special case when α . in section 4, we obtain bounds for the ruin probability by recurssive technique and find that the derivation is some what complicated in general, but less so when we assume α . an exponential integral equation for the survival probability is derived in section 5. = 2 = 2 the generalized exponential distribution has been introduced by gupta and kundu (1999). a random variable x has the generalized exponential distribution with parameters α and if it has distribution function λ -λx αf(x; α, λ) = (1e ) , x > 0, α > 0, λ > 0. with corresponding density function -λx α-1 -λxf(x;α, λ) = αλ(1-e ) e , x > 0, α > 0, λ > 0. the generalized exponential distribution is denoted by . the ge distribution has many nice properties and it can be used as an alternative to gamma and weibull distributions in many situations(gupta and kundu(2001)) ge(α, λ) 2. preliminaries we consider a risk process in which claim occur as a renewal process. let be a sequence of independent and identically distributed random variables, where denotes the time until the first claim and, for i > , denotes the the time between the and claim. we assume that has a distribution with density function i i=1{t } ∞ ti 1 th(i -1) thi it ge (n, λ) -λt n -1 -λtg(t) = nλ(1e ) e , t > 0, λ > 0 where is a positive integer, and distribution function n -λt ng(t) = (1e ) , t > 0, λ > 0. for most of this paper, we illustrate ideas by restricting our attention to the case in which n = . of course, when n = , we have the classical risk model. 2 1 let denotes the value of the reserve at time t . is governed by δu (t) δu (t) δ δdu (t) = cdt + u (t)δ dt dx(t), that is t(δ)δt δ(t-v) δ t 0 u (t) = ue + cs e dx(v).∫ (1) where and is a premium that insurance company receives per unit time. in addition to the premium income, the insurance company also receives interest of its reserves with constant force δ and denotes the accumulated amount of claims occurring in the time interval , that is δu = u (0) c x(t) (0, t] n(t) j j=1 x(t) = x∑ we assume that is comprised of renewal claim number process whose inter-claim times have distribution. the individual claim amounts independent of are positive, independent and identically distributed random variables with common cumulative t 0{x(t)} ≥ t 0{n(t)} ≥ 1 2{t , t . . .} ge(2, λ) 1 2x , x . . . t 0{n(t)} ≥ published by atlantis press copyright: the authors 89 distribution function f( and mean . x) = p{x x}≤ m = e(x) finally, t(δ) δv δt t 0 t if δ = 0 s = e dv = e -1 if δ > 0 δ ⎧ ⎪ ⎨ ⎪ ⎩ ∫ for convenience we will drop the index δ , when the force of interest is zero. then (1 is reduced to the usual risk process ) u (t) = u + ct x(t). (2) now we define the time of ruin by δ δτ = inf{t: u (t) < 0}. let denote the probability that the company is ruined at sometime starting with initial reserve . so δψ (u) u δ δ δ δ t 0 ψ (u) = p{ u u (t) < 0} = p{τ < | u (0) = u}. ≥ ∞ we denote , the survival probability, that is, the probability that ruin never occurs. δ δφ (u) = 1-ψ (u) consider the risk reserve process u let and be the moment generating function of . consider the equation (t) = u + ct x(t). i i iy = c t x , i = 1, 2, . . . syi ym (s) = e{e } iy s c t -s xi i ym (s) = e{e }e{e } = 1. (3) clearly . this equation may have a second root. if such a solutions to (3), s = 0 exist, then it is unique and positive. this equation is the defining equation for the adjustment coefficient. ym (0) = 1 3. integral equation for survival probability in this section, we derive an integral equation for . suppose that the first claim occurs at time and the amount of claim is , the surplus just after the payment of first claim is δφ (u) 1t = t 1x = x (δ)δt t ue + cs x . by considering the time and amount of first claim, the conditional probability that the company will survive is (δ)δt δ t φ (ue + cs x) . thus we have (δ)δt1 δ δ 1 δ 1t φ (u) = e{φ (u(t ))} = e{φ (ue +cs -x )}, that is (δ)δte +c s (δ)δtt δ δ t0 0 φ (u) = g (t) φ (ue + cs x) df(x) dt. ∞ ∫ ∫ the change of variable (δ)δt t s = e + cs , we obtain ( ) δ λ/δ -1-λ/δ 2λ/δ -1-2λ/δ u u δ 0 φ (u) = 2λ (c +δu) (c+δs) (c +δu) (c +δs) φ (s x) df(x) ds. ∞ ×∫ ∫ (4) differentiating expression (1) with respect u gives ( ) δ λ/δ -1-λ/δ 2λ/δ -1-2λ/δ u s δ 0 (c+δu)φ (u) = 2λ (c +δu) (c +δs) λ 2λ (c +δu) (c +δs) φ (s x) df(x) ds. ∞ ′ ×∫ ∫ (5) differentiating expression (5) again with respect to u and rearrange the terms δ λ/δ -1-λ/δ 2λ/δ 2 u φ (u) = 2λ λ (c +δu) (c + δs) λ ( -1) 2δ (c+δu) δ(c + δu) ∞ ′′ ⎡ ⎢ ⎣∫ s -1-2λ/δ δ 20 u δ 0 2λ λ (c +δs) ( -1) φ (s-x)df(x)ds δ (c+δu) φ (u-x) df(x). ⎤ ⎥ ⎦ ∫ ∫ (6) inserting δφ (u)′ and into (6) and making a couple of simplification we arrive at the integrodifferential equation δφ (u) 2 δ u 2 2 δ δ δ 0 (c+δu) φ (u) = (c+δu) (3λ-δ)φ (u) 2λ φ (u) + 2λ φ (u-x)df(x). ′′ ′ ∫ (7) integrating both sides of (7) with respect u and after simplifications (8) 2 δ 2 2 δ δ δ u u 2 2 δ δ 0 0 (c+δu) φ (u) = (δc+δ u + 3λδu + 3λc)φ (u) + c φ (0) (3λc+δc)φ (0)(3λδ +δ ) φ (v)dv -2λ φ (u-x)(1-f(x))dx, ′ ′ ∫ ∫ which we rewrite as u 2 2 2 δ δ 0 (c+δu) φ (u) (δ +3λδ)uφ (u) + (3λδ+δ ) φ (v)dv′ ∫ δ published by atlantis press copyright: the authors 90 2 δ δ δ u 2 δ 0 = (δc+3λc) φ (u) + c φ (0) (3λc+δc) φ (0) 2λ φ (u-x)(1-f(x))dx. ′ ∫ (9) as limit u , the left hand side of the expression (9) tend to a constant → ∞ δ 0 b(δ) = δ(δ+3λ) ψ (u)du, ∞ ∫ and the right hand side becomes 2 2 δ δ(δ c+3λc) (1-φ (0)) + c φ (0) 2λ m.′ it is easily seen that , we obtain b(0) = 0 2 δ δ 2 b(δ) (δc+3λc)(1-φ (0)) + 2λ m φ (0) = . c ′ (10) inserting (10) into (8), we get 2 δ 2 2 δ u u 2 2 δ δ 0 0 (c+δu) φ (u) = (δc+δ u+3λδu+3λc)φ (u) + 2λ m + b (δ) (δc+3λc) (3λδ + δ ) φ (v)dv 2λ φ (u-x) (1-f(x))dx. ′ ∫ ∫ (11) integrating this expression again with respect to u gives )( u 2 δ δ2 0 u u 2 2 δ δ δ 0 0 u v 2 2 δ 0 0 1 φ (u) = (3δc+3λc) φ (v) dv + (6λδ+4δ ) (c+δu) vφ (v) dv -(3λδ+δ )u φ (v)dv + c φ (0) + b (δ) (δc+3λc) + 2λ m u 2λ φ (v-x) (1-f(x)) dxdv ⎡ ⎢ ⎣ ⎤ ⎥ ⎦ ∫ ∫ ∫ ∫ ∫ (12) when δ , (12) reduces to = 0 (13) 2 u u 2 2 2 1 0 0 c φ(u) = 3λc φ(v)dv + c φ(0) + (2λ m-3λc)u 2λ m φ(u-x) f (x)dx,∫ ∫ with the integrated tail distribution of is given by f x 1 0 1 f (x) = [1-f(y)]dy. m ∫ and 2 2 3λc 2λ mc φ (0) = r where is the positive solution of the equation (3). r 3.1 laplace transform the appearance of convolution in (13) suggest that it is better to use laplace transform to get an explicit expression for non-ruin probability when the interest force is zero. so we introduce -s v 0 φ̂(s) = e φ(v)dv, ∞ ∫ and -sy 0 f̂ (s) = e df(y). ∞ ∫ taking the laplace transform of the integral equation (13), we obtain ( ) 2 2 2 2 2 c sφ(0) + 2λ m 3λc φ̂(s) = . ˆc s 3λcs + 2λ 1-f (s) (14) such expressions can be easily inverted using mathematical software packages. recursive calculation of ruin probabilities in this section, we derive bounds for the ruin probability in renewal risk model by recursive techniques. exact solutions for the ruin probability are difficult to find. we find the bounds for by discretizing the integral equation (12) . for any h > and we have δψ (u) δφ (u) 0 k = 1, 2, . . . +(h ) (h ) δδ δφ (h k) φ (h k) φ (h k),≤ ≤ with ( ) +(h ) δ 2 2 1 3 k-1 k-1+ -(h ) (h ) 1 2δ δ j=1 j=0 k-1 2 +(h ) 2 3 δ 4δ j=1 k -(h ) 5 jδ j=1 1 φ (h k) = h (c+δk h) p h p (2k-1) 2 p φ (jh)h p kh φ (j h)h + h p (2j-1) φ (jh) + c φ (0) + p kh 2 p φ (k-j)h f (h) ⎧⎪ ⎨ ⎪⎩ ⎫⎪ ⎬ ⎪⎭ ∑ ∑ ∑ ∑ ( ) -(h ) δ 2 2 2 5 1 k-1 k-1+(h ) (h ) 1 2δ δ j=0 j=1 k 2 -(h ) 2 3 δ 4δ j=1 1 φ (h k) = (c+δk h) + p kh + p f (h) p φ (jh)h p kh φ (jh)h + h p (2j-1) φ (j-1)h + c φ (0) + p kh 2 ⎧⎪ ⎨ ⎪⎩ ∑ ∑ ∑ published by atlantis press copyright: the authors 91 ( ) k-1 +(h ) 5 jδ j=1 p φ (k-j)h f (h) , ⎫⎪ ⎬ ⎪⎭ ∑ +1 hk k h(k-1) f (h)= (1-f(y)) dy.∫ where , , , and . 1p = 3δc+3λc 2 2p = 3λδ + δ 2 3p = 6 λδ + 4δ 2 4p = b(δ) (δc+3λc) + 2λ m 2 5p = 2λ m the algorithm implements lower and upper bounds successively. the two sided bounds for the survival probability is of theoretical importance, as it can hardly be used for direct numerical computations. for the interest free model, however, the algorithm give useful two sided bounds for . φ(u) a slightly different method is used to obtain the bounds for the ruin probability in a classical risk model with interest in which -λtg(t) = 1 e , t 0, λ > 0.≥ for compound poisson model with interest δ , there exist a unique positive solution, γ , such that -γx γx λ0 +1 δ 1 λ e = ce{e } δx 1+ c ∞ ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ∫ dx. (15) the solution to equation (15) is called the adjustment coefficient for compound poisson risk model modified by interest. the right hand side of equation (15) is simplified to λcγ λ +1-γ y δδ δ λ0 +1 δ λ e λ δ λ cγ dy = e γ γ , , c c c δ δ δy 1+ c ⎛ ⎞ ⎜ ⎟∞ ⎝ ⎠⎛ ⎞ ⎛ ⎞ ⎜ ⎟ ⎜ ⎟ ⎝ ⎠ ⎝ ⎠ ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ∫ where , , is the incomplete gamma function. n -1 y x γ(n, x) = y e dy ∞ ∫ n > 0 x 0≥ theorem 4.1 assume that the adjustment coefficient , exist, then γ > 0 (16) -γu γuδ +a e ψ (u) a e ,≤ ≤ for , where u 0≥ γx x x [0, x )0 γ y x e f(y)dy a = inf e f(y)dy ∞ ∈ ∞ ∫ ∫ and γ x x + x [0, x )0 γ y x e f(y)dy a = sup e f(y)dy ∞ ∈ ∞ ∫ ∫ . proof: the proof can be given by going along the same lines of the proof of (theorem 5.4.1, rolski et.al) with some obvious modifications. this is known as two sided lundberg bounds for the ruin function . furthermore, for all δψ (u) u 0≥ -γu δ +ψ (u) < a e if and if . + δa > ψ (0) -γu δa e < ψ (u) δa < ψ (0) note that, 0 + δ γ y x 0 f(y) dy γm a = ψ (0), m (γ) -1e f(y) dy ∞ ∞ ≥ ∫ ∫ = (17) and analogously for δa ψ (0)≤ . let have an exponential distribution with . in this case, an explicit formula for is also available, namely x -β xf(x) = 1 e δψ (u) δ λ c λ γ , β(u + ) δ δ ψ (u) = . λ cβ δ γ +1, δ δ ⎛ ⎞ ⎜ ⎟ ⎝ ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ⎠ (18) we use frobenius series method to derive this formula, which is considered to be the simplest method. numerical examples are given to illustrate the application of explicit formula and its upper bounds in compound poisson risk model with interest. for numerical illustration, we set c = , and β . for interest factor, we consider two different values for and . the corresponding values of the adjustment coefficient 1.1 λ = 2 = 3 δ = 0.05 0.10 γ are 1.20879 and 1.23466 respectively. we compare the upper bounds with exact values of ruin probabilities in classical risk model under interest force in table 1 and 2. table: 1 ruin probabilities when δ = 0.05 u exact ruin upper bounds ruin ( ) δ = 0 0 0.586092 0.586092 0.606061 1 0.156639 0.174982 0.185891 2 0.039133 0.052242 0.057017 3 0.009184 0.015597 0.017488 4 0.002034 0.004657 0.005364 5 0.000427 0.001390 0.001645 published by atlantis press copyright: the authors 92 table: 2 ruin probabilities when δ = 0.10 u exact ruin upper bounds ruin ( ) δ = 0 0 0.570308 0.570308 0.606061 1 0.136032 0.165922 0.185891 2 0.028915 0.048272 0.057017 3 0.005560 0.014044 0.017488 4 0.000979 0.004086 0.005364 5 0.000160 0.001189 0.001645 5. alternative expressions for δφ (u) in this section we derive a couple of alternative expressions for the non-ruin probability. after simplifications and rearrangement on (11), we obtain the equation 2 δ δ 2 2 2u u δ δ2 20 0 3λ+δ 2λ m + b(δ) (δc+3λc) φ (u) φ (u) = (c+δu) (c+δu) (3λδ + δ ) 2λ φ (v)dv φ (u-x)(1-f(x))dx. (c+δu) (c+δu) ′ ∫ ∫ (19) let u 3λ+δ d c +δv0c (u) = e ,∫ v which is simplified to 3λ -1δδu c(u) = 1+ . c ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ multiplying by c( on both sides of (19), and integrate from 0 to , we get u) u ( ) 3λ -1δ δ δ 3λ -2δ 2 3λ -1-u vδ2 δ 0 0 3λ -3-2 u vδ δ2 0 0 δu 1+ φ (u) φ (0) = c 1 δu 1 + 1+ 2λ m-3λ c δc + b(δ) 3λ c+2δ c c δv (3λδ + δ ) 1+ φ (y)dy dv c 2 λ δ v 1+ φ (v-y)(1-f(y)) dy dv. cc ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ⎛ ⎞ ⎛ ⎞⎜ ⎟ ⎜ ⎟⎜ ⎟⎝ ⎠⎜ ⎟ ⎝ ⎠ ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ⎛ ⎞ ⎜ ⎟ ⎝ ⎠ ∫ ∫ ∫ ∫ (20) li , (20) becomes mδ 0→ 3λ 3λ u u c c 3λ2 u vv c 12 0 0 2λm e φ(u) = φ(0) + ( -1)(1-e ) 3c 2λ m {e φ(v-x)df (x))}dv. c ∫ ∫ (21) therefore the multiplication of c ( with leads to an exponentially diminishing survival probability function. u) δφ (u) theorem: the laplace-stieltjes transform of equation satisfies the following second order differential equation: δφ (u) 2 2 2 2 2 δˆ ˆδ s φ (s) (2δcs 3δ 3δλ)sφ (s) + [c s -3cs(λ+δ) +δ′′ ′ (22) 2 2 2 δ δ ˆ ˆ(δ+λ)(δ+2λ) 2λ f(s)]φ (s) = (c s-δc-3λc)φ (0) + c φ (0),′δ where , -s uδ δ 0 φ̂ (s) = e φ (u)du ∞ ∫ δ dˆ ˆφ (s) = φ (s) dsδ ′ , 2 δ δ2 dˆ ˆφ (s) = φ (s) ds ′′ and δ δ u=0 d φ (0) = φ (u)| du ′ . proof: consider the second order differential equation obtained in (7). 2 δ u 2 2 δ δ δ 0 (c+δu) φ (u) = (c+δu)(3λ-δ)φ (u) 2λ φ (u) + 2λ φ (u-x)df(x). ′′ ′ ∫ taking laplace transform on both sides of the equation, we get 2 2 2 2 s δ s δ s δ 2 2 δ δ s δ s δ 2 2 s δ s δ s δ 2 s δ c s l {φ (u)}+2cδl {uφ (u)} + δ l {u φ (u)} c sφ (0) c φ (0) =3λcl {φ (u)} +3 λδl {uφ (u)} cδl {φ (u)} δ l {uφ (u)} 2λ l {φ (u)} + 2λ l {φ (u-x)f(x)}, ′′ ′′ ′ ′ ′ ′ ′ (23) where . then we have the following identities s δ δˆl (φ (u)) = φ (s) s δ δ s δl (φ (u)) = -φ (0) + sl (φ (u)),′ s δ s δ δˆl (uφ (s)) = l (φ (u)) sφ (s),′ ′ s δ s δ s δ δˆl (uφ (s)) = l (φ (u)) s(l (φ (u)) + sφ (s)),′′ ′ ′ 2 2 s δ s δ δˆ ˆl (u φ (s)) = 2l (φ (u)) + 4sφ (s) + s φ (s).δ′′ ′ ′′ (22) can be easily proved by inserting these identities in (23). the identities are proved by integration by parts. remarks: when , we get the same expression as in (23). that is δ = 0 2 2 2 2 2 (c s-3λc)φ(0) + c φ (0) φ̂(s) = . ˆc s 3λcs + 2λ (1-f(s)) ′ published by atlantis press copyright: the authors 93 we can eliminate the term φ (0)′ by considering the maximum of the aggregate loss process associated with the surplus process when and it follows that δ = 0 2c φ (0) 3λcφ(0) = 3λc + 2λ m.′ 2 in conclusion, the results in this paper give bounds for the ruin/survival probability in a sparre andersen risk model with constant interest force. the technique that is applied to produce explicit results for the ruin probability in the classical risk model under interest force can also be applied when inter-claim times have generalized exponential. in section 4, a recursive algorithm is derived to obtain the the bounds for . the algorithm yields both lower and upper bound successively. the bounds for the ruin probability in the compound poisson model are given with numerical illustrations. a couple of alternative expressions for are also derived. δφ (u) δφ (u) acknowledgements we would like to thank the anonymous referee for the constructive suggestions and comments on the previous versions of this paper. references 1. andersen, s.e. on the collective risk theory of risk in the case of contagion between the claims. trans. of the xv intnal. congress of acturies, 2, (1957) : 212-219. 2. cai, j., dickson, d.c.m. upper bounds for ultimate ruin probabilities in the sparre andersen model with interest. insurance: mathematics and economics, 32(1), (2003): 61-71. 3. cheng, y., tang, q moments of surplus before ruin and deficit at ruin in the erlang(2) risk process. north american actuarial journal, 7, (2003) 1-12. 4. dickson, d.c.m., hipp, c. ruin probabilities for erlang(2) risk process. insurance: mathematics and economics, 22, (1998): 25-262. 5. dickson, d.c.m., hipp, c. on the time to ruin for erlang(2) risk process. insurance: mathematics and economics, 29, (2001): 333-344. 6. dong, y., wang, g. ruin probability for renewal risk model with negative risk sums. journal of industrial and management optimization, 2(2), (2006): 229-236. 7. gupta, r.d., kundu, d., generalized exponential distribution, astrl & new zealand j. statist., 41(2), (1999): 173 88. 8. gupta, r.d., kundu, d., generalized exponential distribution: different methods of estimation. j. statist. comput. simul., 69, (2001) : 315 337. 9. jingmin, h.e.,rong, w.u., jiafeng, cui. upper bounds for the ruin probability in a risk model with interest force whose premiums depend on the backward recurrence time process. advances in mathematics, 40(4), (2011) : 501 511. 10. konstantinides, d.g., ng, k.w., tang, q. the probabilities of absolute ruin in the renewal risk model with constant force of interest, journal of applied probability, 47(2), (2010): 323-334. 11. li, s. and garrido, j. on ruin for the erlang(n) risk process. insurance: mathematics and economics, 31, (2004a): 391-408. 12. paulson. j and gjessing. h.k. optimal choice of dividend barriers for a risk process with stochastic pattern on investments, insurance: mathematics and economics, 20, (1997): 215-223. 13. rolski, t., schmidli, h., schmidt, v., tuegels, j. stochastic processes for insurance and finance. john wiley & sons (1999) 14. sun, l. and yang, h. on the joint distribution of surplus immediately before ruin and deficit at ruin for erlang(2) risk processes. insurance: mathematics and economics, 34, (2004): 121 125. 15. sundt, s., and tuegels, j. ruin estimates under interest force. insurance: mathematics and economics, 16, (1995): 7 22. 16. sundt, s., and tuegels, j. the adjustment function in ruin estimates under interest force.insurance: mathematics and economics, 19, (1997): 85-94. 17. 17. thampi,k.k., jacob, m.j., raju, n. ruin probabilities under generalized exponential distribution. asia-pacific journal of risk and insurance, 2(1), (2007) : 22-32. 18. yang, h. non-exponential bounds for ruin probabilities under interest force. scandinavian actuarial journal, 1, (1999): 66 79. published by atlantis press copyright: the authors 94 1. introduction 2. preliminaries 3. integral equation for survival probability 3.1 laplace transform recursive calculation of ruin probabilities 5. alternative expressions for microsoft word stochastic look-ahead scheduling method for linear construction projects.doc stochastic look-ahead scheduling method for linear construction projects lingguang song* and sang-hoon lee department of construction management, university of houston, 375 t2 houston, tx 77204, usa *e-mail: lsong5@uh.edu fitria h. rachmat bechtel corporation houston, tx 77056, usa e-mail: fhrachma@bechtel.com abstract many construction projects involve repetitive operations that are exposed to various constraints and uncertainties. during construction field execution, schedulers must maintain and update look-ahead schedules, which coordinate work for the next several weeks based on the most recent productivity achieved. this study enhances the traditional linear scheduling method used for scheduling repetitive construction operations with stochastic simulation to incorporate uncertainty in activity performance during field operations. the proposed stochastic linear scheduling method (slsm) was implemented and demonstrated in a pipeline construction project. keywords: linear construction projects, linear scheduling, simulation, pipeline construction. 1. introduction linear construction projects refer to construction work that involves a number of correlated activities performed in succession and repeated throughout the entire project duration1, such as the construction of high-rise buildings, tunnels, roadways, and pipe/sewer lines. due to technical, resource, and space constrains, these repetitive activities are typically inter-dependent, i.e. delay or slow progress of one activity will affect its successors’ performance, and eventually impact the overall project duration. an effective coordination of various repetitive activities and their interactions through careful planning and scheduling is critical to project success. poor scheduling can result in considerable waste as laborers and equipment become idle due to delayed completion of preceding tasks, limited availability of needed resources, or other space and time constraints. two types of schedules are frequently used in a construction project: master schedules and look-ahead schedules1. a project master schedule provides management with the big picture of project time and procedure. a look-ahead schedule, or short-interval schedule, a schedule maintained during the actual construction operation, is an extension from the master schedule that specifies the day-to-day details of work scopes, resource allocation, and project team coordination for the upcoming several weeks or months. a look-ahead schedule should be developed based on the most recent project productivity data, and it must be updated periodically so that it can accurately reflect the current project status and provide a reliable forecast for future project performance. compared with a master schedule, a look-ahead schedule provides the operationlevel managers a more detailed and realistic view of future performance and potential problems so that waste and delays can be identified and eliminated. therefore, journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 252-260 published by atlantis press copyright: the authors 252 administrateur texte tapé à la machine received 16 march 2012 administrateur texte tapé à la machine accepted 16 september 2012 administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine administrateur texte tapé à la machine song, lee, and rachmat effective look-ahead scheduling based on the most recent performance data can thus help to ensure a smooth and continuous work flow. traditionally, look-ahead scheduling has been presented in a bar chart or critical path method (cpm) formats. however, the linear scheduling method (lsm) provides an alternative approach1. lsm is a graphical tool designed specifically for scheduling repetitive linear construction projects, such as pipeline projects. in this research, lsm is used as the underlying algorithm for look-ahead scheduling. pipeline construction projects and lsm are further described in the next section. regardless of the schedule format, the current industry practice of look-ahead scheduling is based on a deterministic analysis of actual project data—i.e., forecasting future performance using average past performance. this deterministic method is unable to consider the variability of productivity in forecasting future performance, which may result in a considerable amount of work interruption and crew idle time. this study enhances the traditional lsm with stochastic simulation to incorporate uncertainty in look-ahead scheduling. further, this approach is demonstrated in an actual pipeline construction project case study. an introduction to pipeline construction, lsm, and past relevant studies is presented in the next section. this is followed by a description of the proposed stochastic linear scheduling method and the case study. 2. background 2.1. pipeline construction pipeline construction is a highly repetitive and resourceintensive process that is exposed to various constraints and uncertainties in the working environment. a pipeline construction project operates like a moving assembly line, involving a variety of sequenced tasks and specialized crews, as shown in figure 1. the specific location of the pipeline route is first identified and marked with stakes. the pipeline right-of-way is cleared of vegetation and graded to ensure a leveled base. backhoes or trenching machines are then used to excavate a pipeline trench. this is followed by stringing the individual pipe sections along the trench line. individual joints of pipe may also be bent to a desired angle, if required. after stringing and bending are completed, the pipe sections are welded together, coated, inspected, and placed on temporary supports along the edge of the trench. the pipe assembly is then lowered into the trench and tied into the preceding pipe section. the trench is then backfilled. after backfilling and cleaning up, the pipeline is hydrostatically tested for strength and leaks. fig. 1. pipeline construction process. 2.2. linear scheduling method (lsm) lsm is designed for scheduling linear construction projects that contain a family of repetitive and nearly identical tasks1. it represents a repetitive activity as a production line in a two-dimensional time and space graph, as illustrated in figure 2. the horizontal axis represents time, and the vertical axis is the location of an activity or a crew. the slope of a production line represents its productivity rate. a production line may be a straight line if the productivity rate is constant. however, in real-world operations, a production line typically features varying slopes due to the variability in productivity rates, which can be attributed to many factors, such as quantity and complexity of work and weather conditions. the horizontal distance between two lines is a graphic representation of the time float, or time buffer, between the activities. similarly, the vertical distance represents the physical distance, or space buffer, between the activities. much previous research has shown that lsm allows better representation and visualization of scheduling information than the conventional cpm or bar charts in terms of time and space constraints, activity location, and productivity rates2. it is important that work continuity be maintained in order to achieve efficiency in a linear construction project. in this regard, lsm is particularly useful in visualizing work flow and time and space constraints graphically so that a scheduler can easily adjust activity start time or balance productivity rates to achieve work continuity. therefore, lsm is survey clearing grading trenching stringing bending welding lower-in tie-in backfill cleanup hydro-test published by atlantis press copyright: the authors 253 stochastic linear scheduling method used in this research for look-ahead scheduling of linear construction projects, e.g. pipeline construction presented in the case study. fig. 2. a sample lsm chart. 2.3. related research though look-ahead scheduling has been used for many years in construction, few publications provide information on it. in fact, the few sources that addressed the topic did little more than provide definitions1. schedulers typically rely on readily available project progress data and their own subjective judgments in forecasting future performance, and the resulting lookahead schedule is deterministic in nature. past research in lsm has been focused on applying lsm in early project planning, but uncertainty that is frequently encountered during field operations has not been fully addressed. srisuwanrat and ioannou3 studied the optimization of repetitive schedules when activity durations are probabilistic. in this study, only precedence logic relationships between activities were modeled, and productivity rates per day were assumed to follow the normal distribution only, which may not represent real-world cases. similarly, liu et al.4 also studied the linear schedule optimization problem. to calculate the production rate, a chaotic function is used to simulate the uncertainties and the correlations in a time series of points. however, for look-ahead scheduling, the most accurate data can be obtained from real-world operations and they should be used to model and forecast production variations. duffy et al.5 studied lsm with varying production rates in regards to when and where they occur in time and space for a given linear construction project. the changes of productivity rate is either subjectively defined or through regression analysis based on historical data. while complementary to the effort in our study, their work did not explicitly model the uncertainty in productivity changes and, hence, the output from the method is still a point estimate of future performance. el-sayegh6 developed deterministic and probabilistic models for calculating resource-based linear schedules. the deterministic model can be used to produce a linear schedule based solely on user input. the probabilistic model may be used to produce a linear schedule based on monte carlo simulation, which accounts for variability and uncertainty of construction projects. to make lsm suitable for practical look-ahead scheduling of horizontal linear construction projects (e.g., pipeline construction), our research here enhances the previous research in two areas: (1) formalizing the procedure of productivity data collection and analysis for measuring productivity variability in repetitive activities, and (2) modeling both time and space buffer constraints between activities while considering productivity variability. the following two sections describe the proposed stochastic linear scheduling procedure along with a case study applying the method to pipeline construction. 3. stochastic linear scheduling method 3.1. system procedure the proposed stochastic linear scheduling method (slsm) enhances the traditional lsm with the capability of modeling the variability in activity performance based on actual productivity data. simulation input modeling provides a way for a scheduler to measure and evaluate the variability in activity performance and later to incorporate it into look-ahead scheduling. to achieve this, actual project performance data must be properly collected and analyzed in order to measure the variability. once productivity data are collected, they will be analyzed through a standard simulation input modeling procedure, which involves fitting the productivity data to standard statistical distributions and then determining the bestfitted distribution based on goodness-of-fit tests. the information received from input modeling allows schedulers to represent the variability of activity performance in the form of probability distributions. interested readers can refer to law and kelton7 for more technical details about the aforementioned simulation input modeling procedure. the data collection and input published by atlantis press copyright: the authors 254 song, lee, and rachmat modeling procedures are described and demonstrated through the case study later in this paper. for look-ahead scheduling purposes, the impact of the variability of activity productivity rates on future project performance is further evaluated through discrete-event simulation in slsm. discrete-event simulation is a mathematical-logical model representing a real-world system that evolves over time, allowing users to experiment with the model to analyze and predict system performance4. to facilitate the use of slsm, a computer program, stochastic linear scheduling method (slsm), was designed and implemented in this study. the program allows a scheduler to model and analyze repetitive linear construction projects using graphic modeling tools while considering variability in activity performance. this computerized tool and its development and application is described in the following section. once the model is established, the scheduler can conduct “what-if” analysis or risk analysis by experimenting with the computerized simulation and observing its behavior. for example, experiments can be conducted to evaluate baseline schedule, total project duration, individual activity performance, and activity idle time. accurate look-ahead scheduling can help schedulers to better anticipate problem areas and formulate new plans to improve overall project performance. 3.2. simulation program the slsm program was implemented within a simulation tool development environment, simphony, which allows a developer to design highly flexible simulation tools supporting graphical and hierarchical modeling8. the developed slsm program can accurately model repetitive activities, precedence relationships, and time and space buffers, and it allows users to experiment with the model and observe outputs through a variety of output reports. slsm contains two fundamental modeling elements: project and activity, as described in table 1. using these two elements, a scheduler can assemble a model to graphically represent repetitive activities involved in a linear construction project. table 1. slsm modeling elements. element description project the project element represents the linear construction project and is the parent element of any activity element. users can specify the total quantity of work for the project in the project element. the element also handles collecting simulation results and exporting these results to an external database. activity this is a generic element that can represent either a predecessor or successor. each instance of the activity element has an input port and an output port. when the output of an activity element is connected to the input of another activity element, the former activity element becomes the predecessor, while the latter element becomes the successor. if the input of an element does not connect to any activity element, this activity becomes the first activity of the project and its daily start and finish events are controlled only by the total work quantity. the activation of other activity elements with predecessors is determined by the performance of their predecessors and userdefined time and/or space buffers. when two activity elements are connected, the preceding activity becomes predecessor and the other activity is recognized as successor. as discussed in the background section on lsm, the predecessor and the successor usually cannot overlap, and proper space, time buffers, or both must be maintained so that the successor can be performed efficiently and safely. the time and space buffers are user-defined based on past project experience. once determined, schedulers can specify these buffer values and associated them with the successor activity. the simulation procedure of this predecessorsuccessor interaction through time and space buffers is illustrated in figure 3. the time unit used in this simulation is one day, i.e. the simulation resolution is one day. this means the simulation clock advanced one day at a time. the simulation starts on day 1 and the activity without any predecessor is recognized as the first activity in the project. due to time and/or space buffers, we assume no other activity will start on day 1. for the first activity during the first day, its progress during the day (e.g. footage of trenching completed) is determined by a random number (e.g. footage/day) according to the activity’s input model. then the simulation advances to the second day and a check of successor activities is conducted. if the time and/or space buffer requirements are satisfied, the successor published by atlantis press copyright: the authors 255 stochastic linear scheduling method can start and make a progress for one day, which is determined by their input models. this process is repeated until project completion. this completes one simulation experiment, which represents one possible occurrence in the real-world. this experiment should be repeated, e.g. 30 times or more, so that we can determine the “average” system performance. fig. 3. simulation procedure. for the current version of the program, the model output data are exported to a microsoft access database where graphic results such as an lsm chart can be produced. 4. case study of pipeline construction a pipeline construction project was selected to demonstrate and evaluate the applicability of the proposed slsm for look-head scheduling. the sample project involves the construction of approximately 130 miles of 30-inch pipeline. the case study was focused on understanding the current practice of look-ahead scheduling, collecting actual performance data, and demonstrating the usage of slsm. in this project, the contractor was responsible for main pipe installation activities such as trenching, stringing, bending, pipe laying, backfilling, tie-in, cleaning, and hydro-testing. several subcontractors were also involved in this project for activities such as horizontal directional drilling, concrete work, and coating. the original look-ahead schedule was prepared by the contractor in a table format on a weekly basis. the look-ahead schedule contains information on upcoming activities, mile-post targets, and comments. three sources of information are collected in preparing the look-ahead schedule. the first source of information is the most recent activity productivity data, monitored daily by the contractor and expressed in terms of footage per day. the second source of information is a list of potential operation problems identified through discussion among field managers, superintendents, and subcontractors at the weekly meeting. the third source of data is the forecast of upcoming work performance based on a scheduler’s evaluation of each activity’s recent productivity and its predecessor’s progress. the productivity forecast is calculated deterministically as a moving average of the previous 3 to 5 weeks’ productivity rates. in a linear construction project, maintaining a crew’s continuous performance is essential to achieving the activity milestones because successor activities can only be performed after their predecessors are completed. therefore, achieving reliable milestones for each activity is extremely desirable for maintaining a smooth operation and minimizing the overall project duration. as mentioned previously, lsm is an effective scheduling method for linear construction work and, therefore, it was proposed to the contractor as an alternative look-ahead scheduling method. furthermore, the proposed slsm allows schedulers to evaluate the impact of the variability of project performance and take proactive measures to prevent work interruptions and thus reduce crew idle time. the application of slsm in the sample project is described in the following sections. 4.1. data collection in the sample project, the contractor already had a wellstructured progress-monitoring and reporting system that continuously captured performance data that could be used to model the variability of productivity rates. physical construction progress data were collected through the construction daily progress report prepared by superintendents. in addition to progress data, this report also contained information about working conditions, quality, safety, and regulatory compliance. daily construction progress is measured by “from” and “to” stations, which represent the start location and the end location of a crew in a working day, start identify 1st activity determine 1st day progress advance activity progress for the day check successor time & space buffers determine successor progress satisfied? proj. completed? end experiments done? yes no yes no yes no generate result report advance to the next day published by atlantis press copyright: the authors 256 song, lee, and rachmat respectively. table 2 shows a sample of daily progress data, from which the productivity rate measured by footage per day can be easily derived. productivity rates were reported in the same way for other activities, such as clearing, grading, trenching, bending, welding, lower-in, tie-in, backfill, and cleanup. at the time of this study, about 2 months of data were collected for the above-mentioned activities as they progressed to different locations. table 2. daily progress and productivity. station date task from to footage productivity (ft/d) 9/15 stringing 5484+00 5636+00 15,000 15,000 9/16 stringing 5636+00 5705+83 6,983 6,983 9/17 stringing 5705+83 5806+00 10,017 10,017 9/18 stringing 5806+00 5972+00 16,600 16,600 9/19 stringing 5972+00 6140+00 16,800 16,800 traditionally, the contractor analyzes productivities on a weekly basis for look-head scheduling. as mentioned above, the productivity of the future work is estimated as the average of the 3 to 5 proceeding weeks’ productivity rates. in the proposed slsm method, it is suggested that daily productivity rates be used in lieu of weekly rates in order to better reflect the variability of project performance. the following section discusses how to model this variability based on the collected daily productivity data 4.2. input modeling in order to accurately represent the variability of an activity’s performance, the productivity rate of an activity can be modeled as a probabilistic distribution4. the process of determining the underlying statistical distribution of an activity’s productivity rate is an example of simulation input modeling, which provides a way for a scheduler to evaluate the variability in activity performance. to establish a statistical distribution to represent an activity’s productivity, the abovementioned daily productivity data was used. the input modeling procedure involves two steps: distribution fitting and a goodness-of-fit test. the burden of the input modeling procedure can be greatly reduced by using commercially available input-modeling software. for this project, bestfit®9, recently incorporated into a risk-analysis program called @risk, was used. for the purposes of this research, the program will be referred to hereinafter as bestfit. 4.2.1. distribution fitting the first step in modeling an activity’s productivity rate is to generate a histogram to provide understanding of the characteristics related to type and shape of possible underlying distributions. theoretical standard distributions, such as normal, beta, and triangular distributions, can be used to “test-fit” the data set and the histogram. parameters of these distributions can be determined through this fitting process. because of the use of computerized input-modeling software, a large number of standard distributions can be automatically fitted to the data set in a short period of time. for example, bestfit provides 28 different probability distributions for fitting a data set. these candidate distributions are ranked by the quality of fit, as described below. 4.2.2. goodness-of-fit test while any of the standard distributions can be used to represent the raw data, they can be very different in terms of how closely they represent the true underlying distribution of the productivity data. to test the quality of fit, there are heuristic procedures based primarily on visual inspection as well as goodness-of-fit methods based on statistical hypothesis tests7. to make input modeling easier to learn and use by industry practitioners, the goodness-of-fit test approach was adopted, because these tests are typically automated in input-modeling software. a goodness-of-fit test is a statistical hypothesis test that is used to determine if the raw data points are an independent sample from a particular statistical distribution. it describes statistically how well a standard distribution fits a set of raw data. frequently used tests include the chi-square test and the kolmogorov-smirnov test9. while determining distribution parameters and selecting the best-fitted distribution is difficult to perform manually, input-modeling software such as bestfit automates this process and makes input modeling much more accessible to practitioners who lack knowledge in statistics. the input to bestfit is the daily productivity rate of an activity, as discussed previously. the output from bestfit is a list of ranked and parameterized distributions. a user can then pick the best-fitted distribution to represent the activity’s performance. table 3 shows the selected distributions along with their parameters for a group of repetitive published by atlantis press copyright: the authors 257 stochastic linear scheduling method activities. it should be noted that the data were scaled uniformly for confidentiality reasons. table 3. selected input models. task name statistical distributions surveying exponential with mean =16629 clearing exponential with mean = 9527 grading normal with mean = 2874 and standard deviation = 1363 trenching triangular with low limit = 670, most likely = 1809, and high limit = 10720 stringing normal with mean = 4837 and standard deviation = 3011 bending beta with a = 2.3, b = 3.4, low = 670, and high = 13812 welding beta with a = 1.2, b = 1, low = 700, and high = 9800 lower-in normal with mean = 5882 and standard deviation = 3033 tie-in exponential with mean = 2007 backfill beta with a = 1.2, b = 2.9, low = 804, and high = 15758 clean up normal with mean = 3688 and standard deviation = 1221 4.3. simulation modeling the pipeline construction operation can be modeled using the developed slsm program. a project element must first be defined. a user determines the default total work quantity in footage and the location of the external database for exporting simulation output. within this project element, a user can add multiple activity elements to represent individual activities. these activity elements can be linked together to show the precedence relationships between them. for each activity element, four user input parameters must be defined: planned start time, time buffer, space buffer, and productivity rate (e.g. ft/d). an activity’s planned start time is measured as the number of days between the first day of the project and the first day of the planned start of an activity. time buffer and space buffer are measured by days and space (e.g. footage), respectively, and are optional. the productivity rate can be specified as either a constant value or a statistical distribution, as discussed in the previous section. to demonstrate the use of the slsm program, two models were implemented: one model for the deterministic baseline schedule and the other for a simulated look-ahead schedule using slsm. for the baseline model, activities’ planned start times and productivity rates were quoted from the contractor’s baseline schedule. for the look-ahead model, actual data and the distributions defined in table 3 were used to represent the productivity rates. a screenshot of a sample slsm model and a data-entry form for an activity element are shown in figure 3. fig. 3. an slsm model and model inputs. 4.4. output analysis after a model and its parameters are set, users can experiment with the simulation model and collect outputs. a simulation experiment usually contains a large number of runs, often as many as more than 30. each simulation run represents one observation of the possible future project performance and it predicts the project performance based on a set of random productivity rate values generated according to the distributions defined in table 3. the results from these simulation runs can then be gathered and used to forecast project as well as individual activity performance. this section demonstrates some sample outputs from simulation experiments. figure 4 shows a comparison of the deterministic baseline schedule (solid lines) and a simulated lookahead schedule (dotted lines). it can be seen that the actual productivity rates deviate from the deterministic estimates in the baseline schedule. crew waiting or idle time are represented as horizontal line sections, which can be easily observed. the look-ahead schedule can be improved in different ways. for example, trench and tiein activity appears to be slow, and therefore, a more published by atlantis press copyright: the authors 258 song, lee, and rachmat efficient configuration of the crew and equipment should be used. in addition, the planned start of some activities (e.g. hydro test) is too early, which will result in significant waste of crew idling due to conflict with the predecessors. fig. 4. a comparison of baseline schedule and simulated lookahead schedule. outputs from the simulated look-ahead schedule model can also be used to analyze the uncertainty of the total project duration. each of the many simulation runs predicts a possible total project duration scenario. the duration data can be used to determine either the probability of completing a project within a certain number of days or the project duration corresponding to a certain desired risk level. figure 5 shows the probability distribution of the total project duration. for example, according to this distribution, the probability of completing the project within 300 days is only 26%, but the project has a 90% chance to complete within 334 days. fig.5. probability distribution of total project duration. many observations related to an individual activity’s productivity rate and idle time can also be obtained. as an example, the performance range chart can forecast the performance of an activity on a certain confidential interval. figure 6 shows the performance range of the clean-up activity. fig. 6. activity performance range chart. 5. conclusions accurate and timely look-ahead scheduling based on the most up-to-date project data can greatly improve project execution and control. the traditional use of lsm is primarily limited to early project planning or master scheduling. after field operations commence, the most recent project performance data should be collected and analyzed for timely updating of the look-ahead schedule for more accurate forecasting that reflects the dynamic project environment. this research introduces lsm integrated with real-world performance data and computer simulation as an alternative tool for lookahead scheduling in linear construction projects. in the proposed method, actual performance data is first collected from ongoing construction operations and used to model the variability of activity performance. this variability is then incorporated into a process simulation model that can be used to generate lookahead schedules. a computer program, slsm, was designed and implemented. slsm allows schedulers to model and analyze repetitive linear construction projects while considering variability in activity performance. various data analyses can be performed based on the simulation output data—e.g., evaluating baseline schedule, the total project duration, individual activity performance, and idle time. the case study demonstrates that the proposed approach can effectively utilize real-world data to model and forecast the impact of uncertainty on future performance. furthermore, better capability in forecasting future project performance based on actual productivity data can help a scheduler anticipate problem areas and formulate new plans to improve project performance. published by atlantis press copyright: the authors 259 stochastic linear scheduling method references 1. j.w. hinze, construction planning and scheduling, 3rd ed., (pearson prentice hall, upper saddle river, nj, 2008). 2. r.a. yamin and d. j. harmelink, comparison of linear scheduling method (lsm) and critical path method (cpm). journal of construction engineering and management, 127(5) (2001), pp. 374–381. 3. c. srisuwanrat and p. g. ioannou, the investigation of lead-time buffering under uncertainty using simulation and cost optimization”, proc. iglc-15 (ann arbor, mi, 2007). 4. w. liu, i. flood, r. a. issa, simulation and optimization of linear construction projects. proc. of computing in civil engineering, asce (cancun, mexico, 2005). 5. g. a. duffy, g. d. oberlender, d. jeong, linear scheduling model with varying production rates, journal of construction engineering and management, 137(8), (2011), pp. 574–582. 6. s. m. el-sayegh, linear construction planning model (lcpm): a new model for planning and scheduling linear construction projects. ph.d. dissertation, texas a&m univ., (college station, tx, 1998). 7. a.m. law and w. d. kelton, simulation modeling and analysis, 3rd ed., (mcgraw-hill, new york, 2000). 8. d. hajjar and s.m. abourizk, unified modeling methodology for construction simulation, journal of construction engineering and management, 128(2) (2002), pp. 174–185. 9. palisade corporation, bestfit user manual, (palisade corporation, ithaca, ny, 2004). published by atlantis press copyright: the authors 260 microsoft word necc amcm2 fnl3 anon3 edit format new england compounding center meningitis outbreak: a compounding public health crisis. bolanle a. olaniran department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, united states of america juliann c. scholl department of communication studies, texas tech university, p. o. box 43083 lubbock, texas 79409-3083, united states of america e-mail: juliannscholl@yahoo.com  www.ttu.edu abstract the new england compounding center’s (i.e., necc) meningitis outbreak represents a major public health crisis of 2012 that led to over 40 deaths, along with infection of several hundred individuals. the outbreak necessitates the need for how such a crisis could be prevented.. the research focuses on assessment of the necc meningitis outbreak using the anticipatory model of crisis management (amcm). using the amcm principles, it was found that necc did not engage in an adequate vigilant decision making process. keywords: crisis management, crisis preparedness, anticipatory model, meningitis outbreak 1. introduction crises are fairly commonplace in today’s society, either because we are more aware of their occurrence through the media, including social media, or because their impacts are often close to home. although inevitable in organizational life, crises are preventable1. the idea that crises are avertable represents the hallmark of the anticipatory model of crisis management1-2. notwithstanding, organizations rarely do what it takes to prevent crisis. some would rather deal with crises after they occur than commit valuable resources to prevent crises from happening. this leads to an assessment and evaluation of the meningitis epidemic fueled by the new england compounding center in late 2012. in essence, the goal of this study is to evaluate and assess the management of the meningitis crisis while offering important implications for crisis managers and practitioners, as well as public health experts. 2. case overview new england compounding center (necc) is a pharmaceutical company embroiled in a crisis of great proportion. since september 21, 2012, more than 30 americans have died of fungal meningitis and almost 300 people have become infected from tainted steroid injections they received to treat problems with their spines3-5. necc is a pharmaceutical company that does compounding, a practice that involves preparing a medication to suit small but specific needs that are unmet by larger pharmaceutical companies. necc is reported to have shipped close to 18,000 vials of the contaminated steroid to 75 facilities in 23 states, and as a result, many more people stand to be journal of risk analysis and crisis response, vol. 4, no. 1 (march 2014), 34-42 published by atlantis press copyright: the authors 34 willieb typewritten text received 21 july 2013 willieb typewritten text accepted 20 january 2014 willieb typewritten text       olaniran and scholl infected with the fungal contaminated product5. the next section presents a brief overview of the anticipatory model of crisis management which is used to analyze the necc meningitis case. 3. anticipatory model of crisis management the anticipatory model of crisis management (amcm) posits that while one might not be able to prevent all crises from occurring, prevention should be a major priority. the essence of the anticipatory model is that attempts should be made to put in place programs that prevent error, disaster, and crisis while also implementing plans to handle any crises if and when they do occur. an understanding of the amcm requires a clear definition of organizational crisis, which is as an unpredictable or a major threat that could have a negative effect on the credibility of the organization, the industry or its stakeholders 6-8. in keeping with the impact crises can have on public health, crisis is characterized as an event that compromises the health and safety of employees, customers, or and community, that threatens to destroy public trust in the organization, thus damaging the company’s reputation 9. the definition of crisis offered by pearson et al. reflect the sense that the prevention of crisis not only safeguards the public’s health and safety, but also preserves the trust that the public has for the company to prevent crises by ensuring their products are safe and that their business practices and communication with the public are honest. with regard to public health preservation, the anticipatory model implies that best practices are maintained through competent communication within the organization and with the public. the notion of anticipation deserves attention here. specifically, an anticipatory focus not only thwarts disasters, but it reduces their magnitude if and when they do happen10. furthermore, despite the rare occurrence of many organizational and health crises, the brunt is significant when they do occur, which necessitates an anticipatory approach when planning for them and preparing for their potential impacts. finally, decision makers should possess a deep understanding and knowledge of conditions, situations, or events that could signal danger or trigger crisis. this understanding can be achieved through the specific components of amcm, which are outlined subsequently. weick11 warns that the very action that enables people and organizations can also cause destruction. in other words, crisis is not just something that happens to an organization, but rather, organizational members and leaders enact thoughts and engage in actions that can intensify the conditions that make crises more likely to happen. this idea pertains to the principles of enactment and expectation which are germane to the anticipatory model1-2,12-13. enactment represents a process whereby a given action is brought about14. in other words, enactment represents any activity that puts a plan into motion, embodies an organizational policy, or reflects a pre-determined response to an event, such as an anticipated crisis. based on the amcm, enactments are proactive rather than reactive, meaning that they should not just reflect response to disasters that have already occurred, but are actions that are pre-determined. weick11 supports extending the notion of enactment to consequences from such actions. for example, failure to put in place a crisis plan might hinder eventual crisis management. with enactment conceived as a retrospective sense-making process, the model contends that the notion of “anticipation” (of crisis) in and of itself is an action, given that it determines the subsequent choices an organization makes based on available information. justification of this argument lies in the fact that decision-makers often find themselves in situations where they have to anticipate opportunities, threats, and weaknesses in their environment and then take appropriate measures to safeguard their interests. therefore, the model asserts that decision-makers’ actions or inactions with anticipation would result in different outcomes. expectation, a second amcm principle, involves assumptions that people make about certain situations, environments, technologies, or their combination1-2, 10, 12-13. for example, expectations about the potential occurrence of a crisis would determine whether or not a preventive countermeasure is put in place. notwithstanding, assumptions have the potential to bring about a selffulfilling prophecy. for example, when organizational decision-makers assume that a technology is fail safe, they err and might relax other safety measures such that published by atlantis press copyright: the authors 35       necc meningitis outbreak  additional counter measures are never put in place to create necessary redundant procedures1-2, 10, 12. the third element in the model is the idea of control, which is the degree of power an organization has over events or crises. a multifaceted term, control is used to describe how well organizations and decision makers are in command of crisis situations, as well as conditions that might lead to crisis. control, however, is measured in relative terms, which include the perceptions of stakeholders and the public10. for example, if it is discovered that a shipment of medicine leaving a factory is tainted, control might refer to the company’s ability to stop the shipment before it hits pharmacy shelves. the control component intertwines with expectation and enactment to the extent that expectations influence enactments (decisions or actions) and actions exert control over crisis situations. acknowledging the notion of control enhances the ability to prepare and implement effective crisismanagement programs, including positive media relations, information dissemination, and adaptability to make changes when necessary. in sum, crisis prevention requires a thorough understanding of the complexity of relationships and their environmental contexts. nevertheless, enactment, expectation, and control must be acknowledged to understand the process1,11. while enactment consists of specific actions, expectation about an object determines the type of action taken in the enactment process and provides organizations the needed control to handle a crisis. furthermore, control reflects an organization or decision-maker’s ability to respond to an expectation with the appropriate enactment. taken together these factors constitute the crisis anticipation process where managers foresee the occurrence of crisis and work to eliminate or reduce the degree of the catastrophe. the relational perspective as it pertains to public health is useful to facilitate the pre-crisis planning stage of crisis communication. beyond gaining and maintaining the public trust, crisis prevention can be a matter of life and death for community members. 4. methods as is traditionally the norm in crisis management research, the method employed in this study consists of a case study, namely the necc. in an attempt to trace steps, missteps, and processes culminating in the eventual crisis at necc, the researchers utilize and analyze available news materials from media outlets including blogs, news reports, reports of interviews from government agencies (e.g., centers for disease control and prevention, federal bureau of investigations, etc.) and stories. scholars and crisis communication researchers have employed this particular method for decades9,15-19. furthermore, the use of case studies is considered crucial to crisis management and theory development20, and case studies often necessitate the use of any news articles and documents that provide context. therefore, the necc meningitis case materials or information were viewed to create a timeline of events surrounding and leading to the eventual crisis (i.e., deaths and illnesses). analyzing and creating timelines regarding news stories provided the researchers an opportunity to go back in time and assess what went wrong. this analysis also enabled them to identify windows of opportunity where necc could have engaged in crisis planning and/or put in place an adequate crisis management program. moreover, the creation of the timeline provides a systematic approach that uses assumptions and premises of the anticipatory model of crisis management (amcm), which were used to analyze and evaluate necc’s crisis communication and management. thus, the study assesses and evaluates available news materials and interviews from government agencies through the lens of amcm. from the analysis, discussions and recommendations were offered. the next section offers results from the analysis and evaluation of the necc meningitis crisis. 5. analysis and evaluation of necc meningitis case the necc meningitis outbreak offers critical lessons and links between crisis preparation and issues management. while crisis management involves three major stages including pre-crisis, during crisis, and post crisis, issues management focuses on public policy processes to advance organizational interests21-22, as well as protect the health and safety of the public. while crisis management encompasses all crisis stages, issues management is a proactive method of published by atlantis press copyright: the authors 36       olaniran and scholl crisis preparedness and prevention, hence focusing on the pre-crisis stage. in assessing the meningitis outbreak one needs to look at policies or rules in effect governing compounding pharmacies. in so doing, it has been suggested that one of the policy issues that contribute to the crisis is the current administration’s recent tightening of oversight of the manufacturing of generic versions of sterile injectable drugs5. the tightening was called for due to legitimate concerns that the food and drug administration (fda) had about the reliability of the manufacturing facilities. however, it is argued that while the new regulation is aimed at safety issues, it has also created shortages of drugs. consequently, the shortages have forced doctors to seek needed medicines from compounding pharmacies, which are less regulated than the generic drug firms5. notwithstanding this argument, the recent outbreak of meningitis has emboldened government agencies in instituting even tighter laws and regulations. for instance, federal health agencies (e.g., food and drug administration, centers for disease control and prevention) are focused on recalling the drugs, and congress members are called for investigations and the creation of new laws and tightened controls over compounding pharmacies5. an investigation of the role of compounding pharmacies such as necc indicates that compounding medicines never go through formal fda review. compounders frequently distribute these unapproved, older medicines. however, many of the drugs used by compounders have been on the market for decades. these drugs have established proven efficacy with little to no side effects and have been in existence before the establishment of the modern fda review requirements. these drugs include sterile, injected drugs, such as the steroid behind the meningitis outbreak5. the point being made by gotlieb is that tough oversight comes with or results in product shortages. however, we beg to differ because the first responsibility of a government agency is to preserve and protect the safety of the public. we are not arguing that government agencies never go overboard in their regulatory practices. instead, we contend that ensuring manufacturing facilities are in compliance and are operated safely is not too much to ask any pharmaceutical or medical company, especially a compounding pharmaceutical company like necc. after all, the principle of non malfeasance – which focuses on the intent to not engage in harm or bring injury to others is at the core of assessing medical operational guidelines16. this case study analysis reveals that necc may have failed this test of non malfeasance by not adhering to the basic rule governing compounders, which is never to circumvent basic pharmacy compounding to full-scale manufacturing5,23. in other words, a compounding pharmacy must follow specific prescriptions being compounded rather than make a large batch and ship the products without prescription. to this end, hicks23 reports that massachusetts state regulators accused necc of violating state laws by not requiring patient prescriptions before shipping products. essentially, this practice raised questions about the pharmacy’s operation. as a result, necc is said to engage in drug manufacturing (regulated by fda) rather than compounding (not regulated by fda). as a result, necc found itself under investigation and was subject to calls for further investigations by state representative edward makey and governor deval patrick, along with u.s. attorney carmenn ortiz. furthermore, the entire necc entity has been called into question and greater scrutiny due to the meningitis crisis. the fda is now investigating the company’s medication pipelines, including treatments used during heart surgery, for possible contamination attributed to a different fungal meningitis in at least two patients23. 6. necc response to crisis necc’s response to the meningitis crisis has been, for the most part, nonexistent at worst and lackluster at best. as a whole, their response violates the basic tenets of the anticipatory model of crisis management, which is to anticipate and prepare for crisis in an effort to buffer the impact of crisis upon occurrence. in the days following the outbreak of fungal meningitis, necc was nowhere to be found. as a matter of fact, it was reported that after the fda raided its facilities, the company reacted to the raid through its attorney and unplugged its website for several days until october 6, 2012, when it replaced its website with a statement about the recall23-24. as of october 17, the website has never been updated. necc maintains that the recall came "out of an abundance of caution due to the published by atlantis press copyright: the authors 37       necc meningitis outbreak  potential risk of contamination. while there is no indication at this time of any contamination in other necc products, this recall is being taken as a precautionary measure."25. the lack of direct communication from necc to the public and other stakeholders conveys a lack of care. also, not admitting that the fda recall was based on the meningitis outbreak resulting from the contaminated vials is a pattern of behavior that showed necc’s deliberate attempts to be deceptive to the general public in an effort to circumvent potential lawsuits and impending prosecution16. the claim that the steroid recall by the fda is precautionary is questionable, especially given that over 300 people have been infected and more than 30 have died23. more importantly, the resulting behavior by the necc is less than vigilant; it neither conforms to good public relations practices nor conveys good corporate citizenry. with regard to principles of the anticipatory model, necc have failed to analyze potential consequences of a failure to accept responsibility, and they pursued a silence strategy. a vigilant and prior assessment of such a strategy would have suggested that it is only a matter of time before affected members of the public would start filing lawsuits, as is currently the case3,16,23. it would have been more prudent for necc to be out in the public helping affected people and conveying the impression that the company is doing everything in its power to make things right rather than keeping silent. furthermore, silence and denial usually embolden other stakeholders, and especially whistleblowers, to come forward either to get their media time or to vilify the company. at the same time, organizations that remain out of sight in times of crisis are perceived to be guilty and attract more lawsuits and closer media scrutiny as did exxon, shell, and bp16,19,23. it also appears that necc did very little to help itself. a report indicated that when investigators raided its facilities, it was unimaginable what they found in what was supposed to be a sterile facility25. for instance, mats at sterile entrances used to wipe dirt, dust, and other possible contaminants from shoes were dirty and soiled with assorted debris, according to dr. madeline biondolillo of the massachusetts department of public health25. perhaps the most damaging evidence is meningitis-linked fungus (i.e., visible black fungus) found in unopened vials that were shipped to several states25-27, which by any standard is unjustifiable and constitutes irresponsible behavior on the part of necc. clarke and mclaughlin4 report that the problem of unsanitary and questionable practices at necc dates back to 1999 with the most recent reprimand occurring in 2006 when necc was evaluated by an outside firm that found inadequate documentation and process controls involving sterilization at necc. it stands to reason that a company that underwent a severe reprimand in 2006 would be more vigilant and take necessary and periodic measures to make sure that it never repeated the same mistake that could ultimately force it out of business. unfortunately, this was not the path necc chose. sheldon bradshaw (a former fda chief counsel), who signed the enforcement letter in 2006 warning necc about potential microbial contamination, puts it succinctly when interviewed on cbs’s 60 minutes: "well, it's clear the company [necc] made a conscious effort to disregard that warning letter because only a few years later they were engaged in the very behavior the fda warned them about"28. during that same broadcast, a former necc employee indicated that he witnessed technicians handling drug vials without hair nets or gloves. another former employee indicted the owners by indicating that the company's owners discussed as far back as 2009 as to whether necc had crossed the line from compounding, which involves mixing approved drugs for individual prescriptions which is allowed to manufacturing, which would require more oversight. sheldon bradshaw suggested that necc is among other large drug manufacturers who operate under the pretense of traditional pharmacy compounding but are clearly violating the law, and that they do so simply because they think they can get away with it28 (doane, 2012). the above illustrates the amcm principle of expectation and enactment about crisis facilitate or hinder crisis preparedness and response1-2,12. for instance, the fact that necc management considered the possibility of crossing the line between small-scale compounding to largescale manufacturing suggests they were aware of the possibility of breaking the law. notwithstanding, there is something to be said about the fact that it is possible that published by atlantis press copyright: the authors 38       olaniran and scholl some of the employees in the 60 minutes interview may be disgruntled, but the public’s perception in times of crisis is more likely to be anti-establishment or antiorganization than favorable. at the same time, the impression that necc management continues to operate its business as usual and consequently, it has managed to get away with some violations up to that point, might continue to operate with the expectation of going undetected. otherwise, necc would have taken the necessary steps to go through fda requirements for manufacturing or take corrective measures. this would appear as a faulty assumption on the part of necc and its management. as is the case with any faulty assumption, it often results in fallible enactment or actions whose consequences can be dire2,11,13,16. from another standpoint, the amcm contends that assumptions and expectations presented by the model constitute vigilant decision making13,29 only when decision makers like necc are able to efficiently and effectively analyze and evaluate outcomes1. furthermore, some scholars have discussed the fact that most crises are preceded by clear warning signals but often frequently ignored22,30-31. unfortunately, necc fell short as did other organizations, given that it did not appropriately anticipate any negative consequences from blurring the line between pharmaceutical compounding and manufacturing. similarly, failure to engage in appropriate precautionary measures undermines the degree of control that the organization has over the crisis and its management1-2,10,13. in terms of amcm’s principle of control, necc engaged in some missteps that made it lose control in its crisis management effort. for instance, its decision not to take seriously the fda warnings in 2004 and 2006 creates the perception that the company does not care28. the failure to engage or enact a voluntary recall of its steroidal injection as soon as the first victim died, and the decision to wait for directions from the fda, and then suggesting that the fda’s recall is only a precaution does not bode well for the company12,16,23. the failure to engage the public and communicate in a manner that demonstrates care to victims and stakeholders did not represent a competent control of the crisis16,19. also, the fact that one of the co-owners of necc (i.e., conigliaro) operates a company that recycles foam insulation that is run from the same complex as the necc28 violates safety requirements that a sterile pharmaceutical operation demands. 7. implications crisis preparedness is important and essential for any organization, particularly when public health is at stake. an organization need not wait for crisis to occur before developing a crisis management plan. it appears that necc, despite several warning signs, considered itself invulnerable to crisis. vigilant anticipation requires more than just organizing far in advance and preparing speeches to minimize crisis effects on an organization32. successful anticipation and vigilance involves risk assessment and critical evaluation of potential vulnerabilities from different causes of actions and inactions. effective and proactive communication is important to maintain with the affected publics and stakeholders in times of crisis. failure to select appropriate spokespersons other than lawyers can hinder any crisis management effort. lawyers or legal counsel are paid to deal with liability, but adequate crisis management goes beyond liability avoidance; it must convey sincerity, adherence to the principle of non malfeasance, and genuine apology. the moment of crisis is not the time to put financial considerations ahead of public safety16. from a pre-crisis standpoint, it is important that organizations identify potential source(s) of crisis and develop a plan of action to prevent them and consequently enact the plan when a crisis occurs. in the case of necc, contamination prevention is one such area of crisis management, and efforts or programs ought to be in place to put sanitation as a top priority and not wait until regulators identify problems. furthermore, operating in tandem within the boundary of regulations is prudent. if the management can catch itself or speculate that there is the potential of violating the law, chances are it is. therefore, the management must take appropriate measures to ensure that the company is in compliance. doing the right thing might be difficult at first, but in the long run it would save the company from a bigger headache. thus, it is imperative that organizations should make every effort to do things right the first time. published by atlantis press copyright: the authors 39       necc meningitis outbreak  the necc meningitis outbreak crisis underscores the importance of prevention and crisis management to safeguard and protect the community from a public health standpoint. the protection can be done from an issues management perspective, which emphasizes the need for increased regulations and the demand for stricter compliance with current regulations to protect the public and health consumers. the public needs to demand better from pharmaceutical companies and retail pharmacies regarding how their medications, and in particular compounding medicine, might affect them. lack of knowledge and information from this standpoint is likely to result in more episodes like the meningitis outbreak that will further put the public and their health at risk. moreover, the risk might not necessarily be limited to compounded drugs but other manufactured medicines, as illustrated in the johnson and johnson phantom recall case16. in essence, the necc crisis is only symptomatic of what could potentially unfold in the public health arena. the overall goal of public health is to educate the public on matters of health and safety whenever necessary, monitor industrial health standards, and prevent or at least control the outbreak of illness and infection33. necc has a responsibility to the public to process and disseminate products that are safe and effective, and to take the best measures possible not to hamper the health of the population segment that needs their drugs. moreover, necc has the obligation to be transparent about product flaws that may have unfortunately made their way to the market. given these responsibilities and obligations, this study contends that necc failed in its part to promote and safeguard public health. hence individuals, government agencies, and communities at large would be better served in identifying such potential sources of healthcare crises and develop appropriate steps to safeguard against them as specified by the tenets of amcm. finally, this case brings to the forefront that issues management and crisis management often go hand in hand. issues management, in essence, represents a strategic approach to crisis management, and in terms of amcm and crisis prevention, crisis is indicative of a distinct pattern of unfolding occurrences, and the patterns are identifiable even prior to a triggering event2,22. for instance, in the case of necc, there are the warnings based on fda regulations and guidelines for pharmaceutical compounders versus manufacturers, and then there are previous violations and compliance notifications prior to the actual cases of fungal meningitis infection. these circumstances could have prompted necc to make some pivotal decisions to prevent a full-scale meningitis outbreak16. how organizations choose to respond to a crisis could make or break the success of its crisis management plan because publics are more likely to remember not only that the crisis occurred, but more importantly, how an organization responded to the crisis. 8. limitations this present study utilizes a case study methodological approach. as is with any case study approach, generalization is problematic and should be done with caution. more empirical studies should be conducted as to the specific causes of miscommunication with the public and the lacking precautionary measures. nevertheless, the attempt to understand what happened with the necc, regardless of methodology, still offers crisis managers a general glimpse of what to do and what not to in crisis planning and management. as discussed in the paper, the idea of anticipation of crisis potentials and possibilities is something from which any organization, regardless of size, can benefit. furthermore, anticipation by itself is not sufficient until organizational leaders develop a plan of actions to prevent or at least minimize catastrophe should crisis ensue. finally, waiting for crisis to occur before engaging a crisis plan is reactionary rather than proactive. these conclusions, regardless of idiosyncrasies of organizations or environments, are valuable to any crisis prevention and management. references  1. b. olaniran and d. williams, anticipatory model of crisis management: a vigilant response to technological crises. in handbook of public relations, eds. r. l. heath and g. vasquez (thousand oaks, ca: sage, 2001), pp. 487-500. 2. b. olaniran and d. williams, burkian counternature and the vigilant response: an anticipatory model of crisis management and technology. in responding to crisis: a rhetorical approach to crisis communication, eds. d. millar published by atlantis press copyright: the authors 40       olaniran and scholl and r. heath (lawrence erlbaum publishers, 2004), pp. 75-94. 3. m. brown, pharmacy poisons hundreds with deadly fungal meningitis. business insider, (2012, october 17). retrieve online oct. 24 2012 from: http://www.businessinsider.com/pharmacy-poisonshundreds-with-deadly-fungal-meningitis-2012-10 4. t. clark and t. mclaughlin, before meningitis outbreak, firm avoided sanctions. reuters (2012, october 23). retrieve online october 24, 2012 from: http://www.reuters.com/article/2012/10/23/us-usa-healthmeningitis-idusbre89m08v20121023 5. s. gotlieb, compounding a crisis. the american (2012, october 16).retrieve online october 16 2012 from: http://www.american.com/archive/2012/october/compou nding-a-crisis/article_print 6. l. barton, crisis in organizations ii. (south-western college publishing, ohio, 2001). 7. w. coombs, ongoing crisis communication: planning, managing, and responding. (thousand oaks, ca: sage publications, 1999). 8. k. fern-banks, crisis communications: a casebook approach 2nd edn. (mahwah, nj: lawrence erlbaum, 2001). 9. c. pearson s. misra j. clair and i. mitroff, managing the unthinkable. organizational dynamics, 26(2), (1997) 5164. 10. j. scholl, d. williams and b. olaniran, preparing for terrorism: a rationale for the crisis communication center. in community preparedness and response to terrorism [volume iii]: communication systems and the media’s role in community service, eds. d. o’hair, r. heath, and g. ledlow (greenwood publishing group, 2005), pp. 243-268. 11. k. weick, enacted sensemaking in crisis situations. j. management studies, 25, (1988) 305 317. 12. b. olaniran, the role of perception in crisis management: a tale of two hurricanes. multicutural education, 15(2) (2007), 13-16. 13. b. olaniran and d. williams, applying anticipatory and relational perspectives to the nigerian delta region oil crisis. public relations review, 34, (2008) 57-59. 14. l. smircich and c. stubbart, strategic management in an enacted world. acad of manage. review, 10, (1985) 724 736. 15. j. dutton, the processing of crisis and non-crisis strategic issues. journal of management studies, 23(5), (1986) 501-517. 16. b. olaniran, j. scholl, d. williams and l. boyer, johnson and johnson phantom recall: a fall from grace or a re-visit of the ghost of the past. public relations review, 38(1) (2012) 153-155. 17. b. richardson, why we need to teach crisis management and to use case studies to do it. management learning, 24(2) (1993) 138-148. 18. l. wilkins, shared vulnerability: the media and american perceptions of the bhopal disaster. contributions to the study of mass media and communications, 8 (1987). 19. d. williams and b. olaniran, exxon's decision-making flaws: the hypervigilant response to the valdez grounding. public relations review, 20(1) (1994) 5-18. 20. george and a. bennett, case studies and theory development in the social sciences. (boston, ma: mit press, 2005). 21. r. heath and w. coombs, today’s public relations: an introduction. (thousand oaks, ca: sage, 2006). 22. t. jacques, issue management as strategic aspect of crisis prevention. in pre-crisis planning communication and management: preparing for the inevitable, eds. b. olaniran d. williams, and w. coombs, (peter lang, new york, 2012), pp. 17-36. 23. d. hicks, communicating through crisis: meningitis outbreaks grow, necc communication shrinks. blogspot (2012, october, 23). retrieved from: http://crisisexperts.blogspot.com/2012/10/meningitisoutbreak-grows-necc.html 24. t. mclaughlin, pharmacy tied to meningitis deaths says worked with regulators. thomson reuters (2012, october, 23). retrieved online october 24, 2012 from: http://newsandinsight.thomsonreuters.com/legal/news/2012/ 10__october/pharmacy_tied_to_meningitis_deaths_says_worked _with_regulators/ 25. necn.com, meningitis fallout: mass. officials launch surprise investigations (2012, october, 24). retrieved from: http://www.necn.com/10/24/12/meningitis-fallout-massofficialslaunch/landing.html?blockid=792707&feedid=11106 26. reuters, meningitis outbreak: congress widens scrutiny to regulator (2012, october, 24). retrieved from: http://www.huffingtonpost.com/2012/10/12/meningitisoutbreak-2012-congress_n_1962985.html 27. m. stout, cdc: fungus tied to outbreak found in necc’s vials. boston herald (2012). retrieved from: http://www.bostonherald.com/news/regional/view.bg?articleid =1061168778&srvc=rss 28. s. doane, ex-workers: company linked to meningitis outbreak had questionable practices. cbs news (2012, october 19). retrieved from: http://www.cbsnews.com/830118563_162-57536518/ex-workers-company-linked-tomeningitis-outbreak-had-questionable-practices/ 29. r. hirokawa and k. rost, effective group decision making in organizations.management communication quarterly, 5, (1992) 267-288. published by atlantis press copyright: the authors 41       necc meningitis outbreak  30. e. james and l. wooten, leadership as (un)usual: how to display competence in times of crisis. organizational dynamics, 34(2), (2005) 147-157. 31. i mitroff, crisis learning: the lessons of failure. the futurist, 36(5), (2002) 19-21. 32. r. heath, crisis preparation: planning for the inevitable. in responding to crisis: a rhetorical approach to crisis communication, eds. r. l. heath and d. p. millar (hillsdale, nj: lawrence erlbaum associates, 2004) pp. 33-35. 33. m. schneider, introduction to public health, 2nd edn. (boston: jones and barlett, 2006). published by atlantis press copyright: the authors 42 microsoft word jracr_hao_revised_8 attitudes of property owners to climate change considerations attitudes of property owners to climate change considerations and their effects on future property values in coastal communities huili hao* center for sustainable tourism, east carolina university greenville, north carolina, 27858, usa email:haoh@ecu.edu www.ecu.edu patrick long center for sustainable tourism, east carolina university greenville, north carolina, 27858, usa e-mail: longp@ecu.edu www.ecu.edu scott curtis department of geography, east carolina university affiliate faculty, center for sustainable tourism greenville, north carolina, 27858, usa email: curtisw@ecu.edu www.ecu.edu abstract the purpose of this study is to examine property owners’ attitudes regarding the impacts of climate and weather on property ownership and future property values in currituck county, north carolina, and determine whether their attitudes vary according to property owner groupings. the study profiles the segments using a factor-cluster grouping approach that identifies three property owner clusters. these clusters represent various perceptions of property owners toward the effects of climate on property ownership and future property values. a comparative analysis is then conducted among these three property groups, resulting in significant differences being found between them in terms of both attitudinal and demographical variables. group one respondents believe climate and weather affect property ownership and property value, are moderately educated, practice sustainable actions, and there is an equal frequency of second home owners and full time residents in this group. group two respondents believe climate and weather do not affect property ownership but do affect property value, are highly educated, practice sustainable actions, and there is a larger proportion of second home owners. finally, group three respondents believe climate and weather do not affect property ownership or property value, have very little education, practice sustainable actions to a lesser degree than the other groups, and there is a larger proportion of full time residents. this information is useful for currituck county to better communicate with and educate its high-risk and high-end, property owners. keywords: coastal property ownership and value, impacts of climate and weather, property owners’ attitude, cluster analysis. * huili hao, center for sustainable tourism, rw208c rivers building, east carolina university, greenville, north carolina, 27858, united states of america. journal of risk analysis and crisis response, vol. 2, no. 4 (december 2012), 285-291 published by atlantis press copyright: the authors 285 administrateur texte tapé à la machine received 26 september 2012 administrateur texte tapé à la machine accepted 18 october 2012 huili hao, patrick long and scott curtis 1. introduction the social value of a location ranges from the tacit, including sense of place, to the pragmatic, such as property values (anthony et al. 2009). weather and climate play an important role across this spectrum. for example, the psychological concept of a place and one’s attachment to it are intimately linked to its climate (knez 2005). thus, climate change has the potential to both affect property values and the appreciation of place. coastal areas are highly vulnerable to environmental change, in particular through sea level rise, air and ocean temperatures, precipitation, and hurricanes and nor’easters. there also exists an underlying tension between long-term residents with their local customs and the often more wealthy part-time residents with second homes. these environmental and social issues affect sense of place and its management. burley et al. (2007) found that both full-time and temporary coastal residents of louisiana have a constant and heightened sense of place due to the fragility of their environment, and that “attachment to places, perceiving them as under threat, and perceptions that fellow members are willing to engage in environmentally sustainable behaviors, means that residents are more likely to act and demand a greater say in place management”. with regards to climate change, the north carolina lagoon system is vulnerable to barrier island loss and increased salinity from sea level rise and storm surge. links between these environmental impacts and the economy of north carolina (tourism, fisheries, agriculture) has been studied (bin et al. 2007), but little work has been done in north carolina or elsewhere relating sense of place to climate change (adger et al. 2011). this is problematic as adger and co-authors contend that cultural impacts of climate change are equal to economic impacts, have growing importance, and can induce action. because of this lack of attention and the fact that economic values are easier to quantify, coastal management strategies do not normally include tacit social values (anthony et al. 2009). anthony et al. conclude that “tools that articulate and quantify tacit values are needed to provide a more balanced data set to coastal managers, and an appeal to tacit values may better engage society as managers strive to develop and implement mitigation or adaptation strategies”. second home owners are a substantial stakeholder group in coastal counties. their spending is recognized by local officials as important to the economy. thus, land use and economic policies often capitalize on this market group. understanding the thinking of both the resident and non-resident (second home) property owner groups about climate effects can be used to set the stage for communication and education activities with these groups. limited research has been conducted to examine how climate and weather affects property ownership and property values in coastal areas. this study fills the gap by investigating property owners’ attitudes toward climate change and how it may impact future property ownership and values in a high-risk, high end coastal community with a second home vacation economy. specifically, the purpose of this study is: 1) to identify comparatively homogeneous property owner categories using their perceptions of impacts of climate and weather on property ownership and property values; 2) to profile and describe property owner groupings using a factor-cluster approach; and, 3) to examine whether there are any differences between the clusters in terms of their socioeconomic, demographic and other attitudinal characteristics. 2. study area often, the impacts of tourism and second home development, both positive and negative, dominate decisions regarding the economy, environment and community culture of amenity-rich destinations. at the forefront of such a tourism-oriented economic structure can be issues relating to land use, real estate prices, cost of living, transportation, business, workforce, housing, water and the general natural environment, among others. currituck county, a significant part of north carolina’s outer banks tourism destination region, is just such a place. it is located in northeastern north carolina adjacent to the atlantic ocean (figure 1) and is bordered by virginia to the north and dare and camden counties to the south and west respectively. currituck county is well known for its beaches, nature and recreational activities including kayaking, fishing and boating. almost 50 percent of the total land area of currituck county is surface water (currituck county, nc community profile 2012). the county’s population is 23,547 people (census, 2010), but increases three published by atlantis press copyright: the authors 286 climate change and property values fold during the summer due the influx of vacationers. the county has recently experienced substantial growth as indicated by an approximately 30% increase in population between the 2000 and 2010, censuses. the north carolina department of commerce, division of tourism, film and sports development, reported that in 2010 domestic tourism generated $117.12 million of economic impact with 1,380 jobs directly attributable to tourism. additionally, tourism in currituck county resulted in $21.84 million in employee payroll and $11.37 million in state and local tax revenue. forty-three (43%) percent of the single family housing stock is considered second home property. fig. 1. currituck county, north carolina, usa. 3. methodology the geographic information system (gis) tax records of currituck county provided a list of the county’s housing stock from which a random sample was selected from both resident and second home property owners. members of this sample were then sent a cover letter inviting them to visit the study’s website, insert a participant code number and complete a questionnaire. participants were also offered the option of a paper copy or a telephone interview. the questionnaire sought to assess the attitudes and perceptions of these property owners regarding: 1) the importance of sustainable actions on future economic success; 2) the impacts of climate and weather on their property ownership and property value/use and on recreational choice; and 3) degree of community attachment. the sample includes 2,350 second home property owners and 2,408 full time / permanent property owners. four hundred and fifty-nine (459) useable questionnaires were completed and used for this paper (62% were second home owners and 38% full time / permanent property owners). the degree to which the response from full time residents from currituck county is representative of the general resident population was investigated using the census demographic categories of the overall population. the median age for currituck county reported by the u.s. census was 41 years in 2010. among the full-time resident respondents, 13.3% fall in the age range of 35-44 and slightly over fifty percent (55.4%) fall within the age range of 45 to 64 years. approximately thirteen percent (12.9%) of the population in currituck county was 65 years and older according to the 2010 us census while over eighteen percent (18.5%) of the full time respondents for this study in currituck county are 65 years and over. percent male population in currituck county in 2010 was 49.6%; 55.1% of the full time resident response category for this study is male. the median household income for currituck county in 2010 was $55,376 (us census, 2010). twenty-seven (27%) of the full time resident respondents fall within the household income range of $50,000 to $74,999 while 18% fall within the household income range of $75,000 to $99,999. approximately seventeen percent (17.2%) of the population in currituck county has a bachelor’s or higher degree whereas in this study, thirty-nine percent (39.4%) of the full time resident sample has a bachelor’s degree or higher. although the demographic characteristics of the resident sample are similar to those of the full time resident population in currituck county, the sample for full time property owners was older, with a higher level of male representation, as well as a higher education and income level than the currituck county population in general. it is difficult to investigate the representative level of the sample for second home property owners compared to the general second home property owners’ population in currituck county due to the lack of demographic information from the u.s. census or other state and local agencies for this widely geographically distributed group of “residents”. however, according to the national association of realtors’ (nar) 2011 investment and vacation home buyers survey, the typical vacation home buyer in 2010 was 49 years old and had a median household income of $99,500 (national association of realtors, 2011, “vacationand published by atlantis press copyright: the authors 287 huili hao, patrick long and scott curtis investment-home shares hold even in 2010,” para. 7 and 8). among the second home property owner respondents, approximately 68% fall in the age range of 45-64 years and almost seventy percent (69.7%) of them have household income $75,000 and over. the nar’s survey results also showed that nearly half of the vacation-home buyers indicated they were seeking an investment opportunity, while sixty-three percent of the respondents in this study indicated they purchased second home property in currituck county for investment value. study participants were asked to indicate their level of agreement on how climate and weather affects their property ownership in currituck county using a five point likert scale (1 = strongly disagree, 2 = disagree, 3 = neither agree or disagree, 4 = agree, 5 = strongly agree). principal component analysis was performed on the seven items that measure property owners’ perceptions of the impact of climate on their property ownership. four out of seven items loaded highly on one factor (loadings range from .591 to .860) named “climate and weather affect property ownership” (table 1), which explained 38% of the variance. the kaiser-meyer-oklin (kmo) statistic was .757 and the bartlett’s test was significant (p=.000), suggesting that the principal component analysis was necessary and appropriate. a summed scale was then created for this “climate and weather affect property ownership” factor. property owners’ perceptions of the impact of climate change on their future property values were measured by five items using a five point likert scale (1 = not at all, 2 = to s small extent, 3 = somewhat, 4 = to a great extent, 5 = to a very great extent). they were asked to what extent changes in precipitation and temperature, availability of freshwater, number and intensity of coastal storms, as well as sea level rise and coastal flooding affect their future property values. principal component analysis was performed on the five items. all of the five items loaded highly on one factor as shown in table 1 (loadings range from .756 to .872) named “climate and weather affect property values”, which explained 69% of the variance. the kaiser-meyer-oklin (kmo) statistic was .789 and the bartlett’s test was significant (p=.000), suggesting that the principal component analysis was necessary and appropriate. a summed scale was then created for this “climate and weather affect property values” factor. sustainable tourism development places great emphasis on identifying, analyzing and enhancing the characteristics and processes that give destinations a unique character—a sense of place and attachment. in this series of questions, respondents were asked to indicate their level of agreement with the items regarding their attachment to currituck county. principal component analysis was performed on the five items. three out of the five variables had high loading scores (>.5) for one factor named “community sense of place”, which explained 48% of the variance. the kaiser-meyer-oklin (kmo) statistic was .716 and the bartlett’s test was significant (p=.000), suggesting that the principal component analysis was necessary and appropriate. a summed scale was then created for a “community sense of place” factor (see table 1). knowing the rapidly growing importance of integrating sustainability within the tourism industry as well as the propensity of increasing numbers of individuals to do the same within everyday life, survey participants were asked their opinion of the importance of fifteen sustainable actions to the future economic success of the county’s tourism industry. principal component analysis was performed on the 15 sustainable action items. fourteen out of fifteen variables had high loading scores (>.5) for one factor named “sustainable actions”, which explained 50% of the variance. one variable, providing economic benefits from tourism to locals, had a loading score of .43, but this loading score was considered close enough to .5 for this item to be included in the factor. the kaiser-meyer-oklin (kmo) statistic was .918 and the bartlett’s test was significant (p=.000), suggesting that the principal component analysis was necessary and appropriate. 4. results and discussion in order to identify groups of respondents based on similar perceptions of the impacts of climate and weather on property ownership and property values, the summed constructual scores from the factor analyses were used to group the respondents using cluster analysis. three clusters were identified and they each contained an adequate number of cases with the most interpretable outcome. there are 185 respondents in published by atlantis press copyright: the authors 288 climate change and property values table 1. factor loadings of four dimensions of community living. table 2. clusters of respondents based on similar perceptions of the impacts of climate and weather on property ownership and property values. dimension and factored items factor loadings factor: climate and weather affect property ownership weather conditions have changed enough in currituck county that i would not consider buying property here in the future 0.744 climate change will have a noticeably negative impact on my property values in the next 25 years 0.840 changing climate conditions will make currituck county no longer attractive to new residents 0.860 impacts of climate change are evident in currituck county 0.808 factor: climate and weather affect property value changes in precipitation 0.833 changes in temperature and/or humidity 0.838 availability of freshwater 0.756 number and intensity of coastal storms 0.872 sea level rise and coastal flooding 0.849 factor: community sense of place i feel that i can really be myself here 0.777 i really miss it when i am away too long 0.868 this is the best place to do the things i enjoy 0.845 factor: sustainable actions reducing and managing greenhouse gas emissions 0.758 managing, reducing and recycling solid waste 0.753 reducing consumption of freshwater 0.690 managing wastewater 0.650 being energy efficient 0.798 conserving the natural environment 0.738 protecting our community's natural environment for future generations 0.757 protecting air quality 0.786 protecting water quality 0.761 reducing noise 0.607 preserving culture and heritage 0.666 providing economic benefits from tourism to locals 0.430 purchasing from companies with certified green practices 0.772 training and educating employees on sustainability practices 0.748 full access for everyone in the community to participate in tourism development decisions 0.557 clusters climate affect on property ownership climate affect on property value n mean sd mean sd 1 (yespo-yespv) 3.11 0.467 3.79 0.591 185 2 (nopo-yespv 1.93 0.398 3.26 0.553 147 3 (nopo-nopv) 1.95 0.762 1.63 0.493 120 published by atlantis press copyright: the authors 289 cluster 1, 147 respondents in cluster 2, and 120 respondents in cluster 3 as shown in table 2. people in cluster 1 felt climate and weather affected both their property ownership and property values. this cluster will be referred to as yespo-yespv. property owners in cluster 2 responded that climate and weather did not affect their property ownership but still affected their property values. this cluster will be referred to as nopo-yespv. respondents in cluster 3 neither believed climate and weather would affect their property ownership nor their property values. this cluster will be referred to as nopo-nopv. in order to profile the three clusters in terms of their demographic characteristics, cross-tabulation analysis was conducted. the chi-square statistic in cross-tabulation analysis was employed to assess whether there were statistical differences among the clusters for categorical level measurements and dichotomy variables such as gender, residential status, and if employed in tourism-related organizations. one-way anova tests were also carried out to evaluate the differences of the respondents in three clusters on continuous variables such as age, education (less than high school = 1; high school or ged = 2; 2-year college/technical school = 3; some college but no degree = 4; 4-year college = 5; post graduate = 6), sustainable actions (1 = not at all important, 2 = not important, 3 = neither important nor unimportant, 4 = important, 5 = very important), and community sense of place. the chi-square statistics and anova tests revealed that the three clusters were statistically different from each other based on the following variables: residential status, education level, and sustainable action factor as shown in table 3. tables 4 and 5 also illustrate the results of mean difference tests on these three variables. there were statistically significant differences between the three clusters in terms of their residential status, education level and perceptions on the sustainable action factor. table 3. statistically significant differences among three clusters based on demographics and sustainable actions and sense of place. variables f sig. residential status 13.431 0.001* education 5.011 0.007* sustainable actions 40.56 0.000* table 4. numbers and percentages of second home owners and full time residents that fell in 3 clusters. residential status yespo-yes pv nopo-yes pv nopo-no pv second home owners 118 (64%) 104 (71%) 59 (49%) full time residents 67 (36%) 43 (29%) 61 (51%) table 5. mean score for education and sustainable actions for the three clusters. clusters education (mean) sustainable actions (mean) yespo-yespv 4.33 4.04 nopo-yespv 4.99 4.33 nopo-nopv 1.02 3.7 yespo-yespv has the most second home property owners (118) and full time residents (67). however, the ratio of second home owners to full time residents (64% to 36%) is lower than nopo-yespv (71% to 29%), and higher than nopo-nopv (49% to 51%). property owners in yespo-yespv are more educated than those in nopo-nopv but less educated than those in nopo-yespv. they also feel sustainable actions are more important to the success of their county’s tourism economy than people in nopo-nopv do. nopo-yespv is dominated by second home property owners. respondents in this cluster are more educated and perceive sustainable actions are more important to the success of local economy than those in yespo-yespv and nopo-nopv. nopo-nopv is made up of equal numbers of second home owners and full time residents. property owners in nopo-nopv have the lowest level of education among the three clusters. this group of people did not feel sustainable actions are as important as those in the two other clusters perceived. 5. conclusion this study investigated the attitudes of property owners, both full time residents and second home property owners, toward the impact of climate on their property ownership and future property values and determined whether their attitudes vary according to property owner groupings. typologies were determined based upon responses to questions related to property ownership, published by atlantis press copyright: the authors 290 climate change and property values such as whether climate change was evident, whether such change would affect the desirability of the destination, and to what extent it would affect future decisions to retain or purchase property. other climate-related considerations explored perceived impact on property values from changes to precipitation and temperature, availability of freshwater, the number and intensity of storms, and sea level rise and coastal flooding. the study profiled the segments using a factor-cluster grouping approach that identified three property owner clusters. there were statistically significant differences among the three clusters in terms of their residential status, education level and perceptions on the sustainable action factor. first, people who perceived that climate and weather affect both their property ownership and property values have a comparatively high level of education and feel sustainable actions are relatively important to the success of the tourism economy in their community. forty-two percent (42%) of second home owners and 39% of full time residents surveyed fall into this group. second, property owners who perceived climate and weather does not affect their property ownership but still affects their property values are the most educated among three clusters, and also perceive sustainable actions to be very important to the success of the future tourism economy in currituck county. thirty-seven percent (37%) of second home owners and 25% of full time residents surveyed fall into this group. finally, respondents who perceived climate and weather neither affect their property ownership nor their future property values has by far the lowest level of education and places the least amount of importance on sustainable actions. twenty-one percent (21%) of second home owners and 36% of full time residents surveyed fall into this group. this information will allow the decision-making entities in currituck county to adjust their current property ownership practices, recognize their vulnerabilities to the future impacts of climate change, and develop adaptation strategies as necessary, particularly as it relates to investment in home ownership. mitigation of climate change is everyone’s responsibility and policies should be adopted to reduce the carbon footprint of second homes and their respective destinations. there is a pressing need for further discussion among developers, tourism industry leaders, scientists, planners, investors and policy makers on both mitigation and adaptation in second home intensive locations. note this paper is an extension of research data initially presented at the 19th international congress of biometeorology, auckland, new zealand, december 2011 6. references 1. adger, w.n., j. barnett, f.s. chapin iii, and h. ellemor, this must be the place: underrepresentation of identity and meaning in climate change decision-making. global environmental politics, (2011), pp.1-25. 2. anthony, a., j. atwood, p. august, et al., coastal lagoons and climate change: ecological and social ramifications in u.s. atlantic and gulf coast ecosystems. ecology and society (2009), 14, art 8. 3. bin, o., c. dumas, b. poulter, and j. whitehead, measuring the impacts of climate change on north carolina coastal resources. final report for national commission on energy policy (2007) p.101. 4. burley, d., r. jenkins, s. laska, and t. davis, place attachment and environmental change in coastal louisiana. organization and environment 20 (2007), pp. 347-366. 5. currituck county, nc community profile (2012). retrieved on september 8, 2012 from <http://www.thinkcurrituck.com/files/doc/communitypro file2012.pdf>. 6. knez, i., attachment and identity as related to a place and its perceived climate. journal of en vironmental psychology 25 (2005), pp. 207-218 7. u.s. census bureau state and county quickfacts -- currituck county, north carolina. retrieved on april 9, 2012. <http://quickfacts.census.gov/qfd/states/37/37053.html>. 8. vacationand investment-home shares hold even in 2010. retrieved on february 28, 2012. <http://www.realtor.org/press_room/news_releases/2011/ 03/second_homes>. published by atlantis press copyright: the authors 291 microsoft word risk scenes of managerial decision-making with incomplete information an assessment in forecasting models base risk scenes of managerial decision-making with incomplete information: an assessment in forecasting models based on statistical and neural networks approach dusan marcek1,2 1 research institute of the it4i centre of excellence, faculty of philosophy and science, silesian university in opava, bezruc. square 13, 746 01 opava, czech republic e-mail: dusan.marcek@fpf.slu.cz 2 the faculty of management science and informatics, university of zilina, slovakia e-mail: dusan.marcek@fri.uniza.sk abstract the paper is concerned with measuring of risks in managerial decision-making. it builds upon the uncertainty of economic information, which is converted into the concept of risk expressed in terms of probability and using confidence intervals and standard deviations of the predicted quantities. the paper explains the relation of a degree of risk expressed by the classical information measure, bit, by the concept of confidence intervals, or possibly by the standard deviation. forecasting models are applied which are based on a statistical theory and a neural approach. the aim is also to examine whether potentially highly non-linear neural network models outperforms the advanced statistical methods and better reduce risk in managerial decision-making, or they yield competitive results. a method for finding the forecasting horizon within which the risk is minimal is also presented. keywords: confidence interval; entropy; prediction models; neural networks; arima/arch models; managerial decision; risk assessment 1. introduction an important sphere of information necessary for management of economic processes on all managerial levels is the information about the future development of quantities expressed quantitatively, which is used to characterize the state and the development of the object or process. evidence shows that it is possible to reduce uncertainty by a suitable choice and use of forecasting models based on statistical methods, soft computing and artificial intelligence methods. in comparison with the manager's expert estimates, these models based on statistical and soft computing methods or artificial intelligence methods are capable of providing information in the form of forecasts of quantities with an acceptable degree of uncertainty. the manager using these forecasts is able to make better decisions, i.e. such decisions whose risks in achieving targets are minimized. mathematical statistics cox and hinkley and weisberg (see refs. 1, 2) offer the theory of point estimates and confidence intervals. the manager can set and influence the span of these confidence intervals. the confidence interval indicates the span of possible values into which falls the future estimate of the forecasted quantity with the chosen probability defined by the manager. this way the limits of the possible future values are set. point or interval estimates of the future values of various economic indicators are important for the strategic manager's decision-making. when determining information entropy in decision-making, it is useful to focus on how the confidence interval for the forecasted economic quantity can be made more precise, i.e. narrowed by using the forecasting model. a significant prerequisite for the application of such a model in management is that apart from the increased reliability of decision-making, the model output results in uncertainty reduction, which makes decision-making easier and less weighted with risk. the fact or statement that uncertainty reduction facilitates the manager's decision-making is not sufficient. the crucial factor is how specifically the entropy change manifests itself in the consequences of the decision. not only will it be “easier” to make the decision, but more importantly the decision will be more effective in the long run. journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 13-21 published by atlantis press copyright: the authors 13 d.marcek one of the approaches to understanding uncertainty in forecasting models is understanding it as the standard deviation σ of the forecasted quantity used by marcek (see refs. 3, 4). this approach is used with measuring risks of prognoses of economic time series and of many economic and financial models and methods. e. g. applebaum, bertoin (see refs. 5, 6) used it in lévy processes, havlicky; medova and kriacou (see refs. 7, 8) in methods based on the extreme value theory. the standard deviation as a degree of uncertainty or risk of forecasted quantity values estimates is proportional to the statistical degree of accuracy of the forecast defined as root mean square error (rmse). this approach is used with measuring risks of prognoses of many economic and financial forecasting models, and in forecasting models of economic time series, models for managing financial risk. for management, the approach based on the statistical analysis of the dispersion of the quantity values, or on the standard deviation analysis, is the most comprehensible way of representing uncertainty. it is needed to be stated that the standard deviation does not reflect entropy in its true substance as uncertainty which is indicated in bits (binary digits). on the other hand, uncertainty is closely related to how precise are the estimates of the future values of quantities that managers have at their disposal. the less precise the estimate, the larger the standard deviation, and the higher the uncertainty that the information is weighted with. this view of uncertainty does not articulate it in its true sense, however, it expresses very well its inner essence and the mutual relation of entropy and decision-making. the tasks dealing with the influence of information entropy on managerial decision-making have not been described in literature yet. an important prerequisite is whether the calculation of the degree of uncertainty itself has a sufficient informative value for managers in order for them to follow it during the process of decisionmaking. if it were only a very abstract value insufficiently explained and vaguely described in literature, it would hardly be applied in the process of managerial decision-making, where the data serving as the basis for managerial decision-making must be relevant and clearly presented. due to the possibility of achieving effects by applying forecasting models in managerial decision-making, an exceptionally interesting, topical and supporting tool offers itself. our motivation for elaborating it in theory was the fact that to this day there is no theory that would describe the way to reduce or possibly remove uncertainty in management and to use it to calculate effects. at present, entropy as a measure of uncertainty is described in several publications. the entropy, alongside with two of its derivatives, the relative entropy and the mutual information, have been used with great success in the information-theoretic analysis of communication systems. our recent research was concerned with the elaboration of the missing theory of entropy as a category of uncertainty and risk in managerial decision-making. decision-making is one of the basic human activities, whose quality influences the result of the subsequent activities. the objective of the paper is to point out specific outlines of entropy and risk categories, determine their content in structured decision-making as a category which conditions the ways and methods of management not only at the stage of decision-making but also the impacts at the stage of implementation of the decision, where the consequences of these decisions (effects or losses) will manifest themselves. the issue of measuring risk in management and its accompanying phenomena is divided into six chapters in the present paper. chapter two is devoted to characterizing risk and its manifestation in decisionmaking in uncertainty conditions. in the third chapter we present some statistical measures of uncertainty to probe the risk scenes of forecasting models for financial/economic time series. chapter four is devoted to briefly characterizing two latest modelling methodologies used for high financial time series. in the fifth chapter, risk reduction with the use of forecasting models based on the classical (statistical) methods and models based on artificial intelligence is documented and assessed. chapter sixth summarizes the main topics results. 2. risk and its manifestation in decision-making in uncertainty conditions decision-making has a dominant position in management. implementation of planning, organization, coordination and work with people and the results of these activities are based on decision-making. managerial decision-making is understood as the reaction of the manager to the incurred problems, i.e. it is a process of analyzing and thinking, the result of which is a decision. there are many approaches to decision-making. published by atlantis press copyright: the authors 14 risk scenes of managerial… selecting one of them depends on the character of the problem, on the time available, and on the manager's abilities. herbert simon (see refs. 9, 10) distinguishes two ways of decision-making according to the occurrence of the problem at hand: programmed decision-making deals with problems that the manager has already dealt with before. these are routine and recurring problems. the process of solving these problems is well-established, it is usually possible to convert in into an algorithm, it is often programmed, and standard procedures are usually used. non-programmed decision-making deals with problems that the manager has not dealt with before. these are more complex and unique problems. the manager does not know in advance how to proceed, and a creative solution is required of him. decision-making on the level of lower management usually involves theories and tools such as linear and non-linear programming, dynamic programming, game theory, queuing theory, inventory theory, probability theory, renewal theory, graph theory etc. decision making on the level of top management is significantly influenced by time. top management uses tools not only from management but also from other science branches such as mathematical statistics, fuzzy set theory, econometrics, operational research, etc. top managers use these tools to obtain the most precise estimates of the future development of quantities and processes possible. these estimates represent important information on which managers base their decisions. specific choice of tools and models for decisionmaking depends on whether the manager has precise and complete or imprecise and incomplete information at their disposal. the complexity of managerial decisionmaking relates to decision-making with incomplete information. most of the real systems can only be described incompletely, i.e. with information which cannot be formally expressed by unequivocally set parameters. this is uncertain information then. in practice, according to ref. 11, there are mainly two types of such information: according to the first type, uncertain information makes it impossible to exactly determine the future behaviour of the examined system. this type of uncertainty is called stochastic, and it can usually be modelled using the probability theory. stochastic uncertainty is concerned with the category of the probability risk, which is determined as a scene in the future associated with the specific adverse incident that we are able to predict it using probability theory and a lot of data. in this manuscript, we will concern with this type models, which may be described by huang (see ref. 12) as follows. let d be a managerial prediction system including explanatory variables v to explain the behaviour of the variable to be forecast, and faults represented as forecast errors te in time t = 1, 2,…n. a risk function r in term of the conceptual model d for having a risk scene can be represented as ( , ), 1, 2,...tr d v e t n= = to assess the managerial prediction risk r we apply different forecasting models which parameters are estimated by statistical tools. the second type of uncertainty is connected with the description or formulation of the actual meaning of the phenomena or statements about them. this is semantic uncertainty. natural language words semantics with uncertainty, i.e. with meanings of words and individual statements not being exact, is typical of natural language. this uncertainty has the character of possibility rather than probability, and it is most often modelled by fuzzy systems. for more details see refs. 13 – 16. as far as decision-making with risk is concerned, this is the case of decision-making where actual information about real systems is uncertain, and it is not important if the uncertainty is caused by incomplete information about the system's behaviour, or if it is semantic uncertainty. in the further text, in accordance with ref. 12, the risk connected with managerial decision-making will be modelled using probability models and understood as a statistical term of the expected value between two extreme states of decision, i.e. with full uncertainty and decision with certainty. 3. statistical measures of uncertainty we present some statistical measures of uncertainty to probe the risk scenes of forecasting models for financial time series. the fact that the classical forecasting models are based on the probability theory makes it obvious that the models are affected by stochastic uncertainty. it is natural that mangers try to obtain maximum utilizable information, i.e. the most precise values of the future estimate possible. for the assessment of the estimate uncertainty degree, the method of confidence intervals for point forecasts can be used. with determining forecast confidence intervals based on the classical statistical forecasting models such as the models of regression analysis, exponential smoothing, published by atlantis press copyright: the authors 15 d.marcek and winter's seasonal models is concerned e.g. by gaynor and kirkpatrick (see ref. 17). a more complicated situation is in case of models based on artificial intelligence such as gmdh or the classical neural networks with adaptation of parameters by the gradient method using back-propagation algorithm. in this case it is possible to test the h0 hypothesis of the expected type of probability distribution to determine confidence intervals provided that residuals have a normal probability distribution according to ref. 18, and this hypothesis can be verified using χ2 test of good fit on levels of significance set in advance. it is a well-known, widely used and relatively universal method of mathematical statistics. to obtain the correct result, it is necessary for the statistical data file to have at least 50 values. the span of the confidence interval is related to the estimate precision. the more precise the input from the forecasting model, the more precisely it is possible to set the span of the confidence interval, and the larger part of uncertainty is removed. using the χ2 test of good fit, the h0 hypothesis must be verified on the level of significance α = 0.05 and α = 0.01, and this hypothesis claims that the residuals of the forecasted values from the actual values can be considered as a data file with a normal probability distribution. the confidence interval can be then calculated according to the following expression n kx n kxx σσ αα .,. +−∈ （1） where αk is the critical value of the standardized normal probability distribution, α is the level of significance, σ is the standard deviation, n is the number of observations, and x is the expected value. for the chosen probability p = 0.95, the confidence interval determines the limits which the estimate value will not exceed with 95% probability. the value α = 1−p = 0.05 is the so-called level of significance, which means the probability that a random variable of the estimate will acquire a value outside the interval n kx n kx σσ αα +− , interesting about the support of the preference of forecasting models based e.g. on neural networks to managers' expert estimates in managerial decisionmaking is the information about the probability change. the calculation of this probability is possible from expression (1) as the level of significance k 1 e s t n k x a σ − = − （2） where α is the lower limit of the forecast interval of the prognosis calculated by neural network, estσ is the standard deviation and x is the expected value (mean). another measure of uncertainty used in the theory of information is entropy. the entropy, alongside with two of its derivatives, the relative entropy and the mutual information, have been used with great success in the information-theoretic analysis of communication systems (see refs. 19, 20 for details). entropy and also uncertainty is expressed by the amount of information that we get after performing an experiment. for example, if we get a message that an event a has occurred with probability p(a), we also get information i(a) equal 2log ( )p a− bit. in case the event a consists of a finite amount of measured events, i.e. subsets of probabilistic space ω while ia a∈ for 1, 2,...i n= , 1 n ii a = ω =u and 0i ja a =i for i j≠ is valid, then the entropy expressed by sannon´s definition is (see ref. 20) 2 1 1 ( ) ( ). ( ) ( ).log ( ) n n i i i i i i h p i a p a i a p a = = = = −∑ ∑ （3） in this connection, a very important question is, how will the entropy change if the estimate is more precise? the probability used in the relation for the calculation of entropy is the probability that the estimate value will fall into the narrower 95% confidence interval. the most frequently measure of uncertainty and risk used in literature is the standard deviation. as far as relevancy is concerned, it is probably the easiest and, for managerial practice, the most comprehensible way of expressing and quantification of uncertainty. while the entropy indicated in the information unit bit is at present a still relatively abstract and almost non-used measure for expressing risk in the sphere of managerial decisionmaking. uncertainty in the sense of the standard deviation has a higher informative value for managers. uncertainty expressed by the standard deviation has one drawback, which is unit incompatibility. entropy is indicated in bits. despite this fact, as we could see in the given examples, it is easier to work with entropy as the standard deviation. it is possible to state that reduction of entropy of the forecast system was achieved when its standard deviation of forecast errors was reduced. it can be clearly seen in expression (1). in technical systems, rule 3 σ is used which in the figurative meaning provides published by atlantis press copyright: the authors 16 risk scenes of managerial… information about which interval the forecast will almost certainly fall into. therefore, it provides certainty instead of uncertainty. but it is a certainty which will not push the manager forward with his decision-making if there is a big standard deviation. the real solution leading to the support of decision-making is reducing uncertainty of the forecast system by using a better forecasting model which will achieve lesser variability of prognosis errors. described in ref 21, on the basis of prognosis errors analysis, is a method of searching for such a forecast horizon for which entropy and thus also prognosis risk is minimal. a demonstration the procedure of quantifycation of effects arising from the entropy reduction by using different forecasting models can be found in ref. 22. 4. exchange rate forecasting: a statistical and neural approach in the remainder of this paper we investigate the application of forecasting models including the two representations based on the statistical/econometric approach and the neural network one respectively to the high frequency forecasting of the exchange rate, the czech crown (czk) against the slovak one (czk/skk). we analyse, discuss and compare the forecast accuracy from models which are derived from competing statistical and neural network specifications. for time series modelling of high frequency data, the statistical regression models with disturbances following arch (autoregressive conditional heteroscedastic) type processes are used. the method of building a statistical appropriate time series forecasting model is an iterative procedure which consists of the implementation of several steps. the main four steps in are: identification, estimation, diagnostic checking, and forecasting. in the iden-tification step a tentative model is identified by studying the behaviour of the autocorrelation function (acf) and partial autocorrelation (pacf) functions of the time series. in the estimation step, the estimates for the parameters of the tentative model are computed. in the next step, the diagnostic checks are performed. the langrange multiplier test is used for testing for heteroscedasticity. the adequacy of the model to the data is checked by testing the significance and relationship of the parameters. to detect non-liner hidden patterns in stochastic financial time series of generally high frequency the fitted residuals are subjected to the bds test. if any of the tests or residuals is un-acceptable, the model must be re-specified and the previous steps repeated. once the appropriate model has been found, it can be interpreted and future forecast can be found. as an alternative to statistical models we use the radial basic function neural networks (rbf nn) and their fuzzy logic and granular variants. the neural network in this study was used as a non-linear supplement of the statistical linear arch models. the structure of the rbf nn is defined by its architecture (see fig. 1). in fig. 1 each circle or node represents the neuron. this neural network consists an input layer with input vector x and an output layer with the output value tŷ . the output signals of the hidden layer are ( ),2 jjo wx −=ψ （4） where x is a k-dimensional neural input vector, jw represents the hidden layer weights, 2ψ are radial basis (gaussian) activation functions (rbf) or cloud activation functions (caf). note that for an rbf network, the hidden layer weights jw represent the centres jc of activation functions 2ψ . fig. 1. rbf neural network architecture. to find the weights wj or centres of activation functions we used the adaptive (learning) version of kmeans clustering algorithm for s clusters. cluster analysis is regarded as one of the granulation methods, i.e. it say why and how to put objects into same granule. published by atlantis press copyright: the authors 17 d.marcek granules are extracted from data in the form of clusters, i.e. these entities receive collections of numerical data that exhibit some functional or descriptive commonalities. the final number of clusters provides the number of granules representing rbf neurons in the hidden layer. the centers of clusters are regarded as the means of granules. a family of granules containing every value of input data is called as a granulated view. the granulated view of input data consists of a family of overlapping granules. the above mentioned learning algorithm based on the clustering is regarded as one of the granular methods presenting bottom-up granulation, i.e. input data are combined into larger granules.the rbf network computes the output data set as ( ) ( ) ,= , = ,,= ˆ 1 , 1 2, ∑∑ == s j tjj s j jttjtt ovcxvvcxgy ψ n ..., 2, 1, = t （5） where n is the size of data samples, s denotes the number of the hidden layer neurons (rbf neurons) and tŷ corresponds to the estimated variable tr used in models (4) and (5). according to ref. (23) the network with one hidden layer and normalised output values tjo , , where the normalisation means that the sum of the outputs from the hidden layer is equal to 1, is the fuzzy logic model or the soft rbf network. in our case, the subjects of learning are the weights tjv , only. these weights can be adapted by the error back-propagation algorithm. for detailed computational algorithm for the ex post forecast rmse values and the weight update rule in the granular network see ref. 24. 5. empirical studies and forecasting performan-ce evaluation as we mentioned above, we will investigate the application of statistical and rbf nns forecasting models on the high frequency time series data of exchange rate czk/skk (the czech crown against the slovak crown). the data is available at www.oanda.com. the data was collected for the period january 1, 2007 to february 12, 2008 which provided of 702 observations (see fig. 2). to build a forecast model the sample period (training data set denoted α )for analysis r1, ..., r670 was defined, i.e. the period over which the forecasting model was developed and the ex post forecast period (validation data set denoted ε ) r671, ..., r702 as the time period from the first observation after the end of the sample period to the most recent observation. by using only the actual and forecast values within the ex post forecasting period only, the accuracy of the model can be calculated. in our case, the basic unit of time for which the forecasts were made was a day. economists and financial managers are primarily concerned with short-term forecasting, mostly one month (that is 31 days) in the future. hence, we chosen the length of validation data set, as the last 31 periods. all 670 previous observations were included into the training data set. input selection is crucially important for the successful development of an arch-garch model. potential inputs were chosen based on traditional statistical analysis: these included the actual exchange rates czk/skk and lags thereof. fig. 2. actual and fitted values of the czk/skk exchange rates-model (6). residuals are at the bottom. actual time series represents the solid line, the fitted values represents the dotted line. the relevant lag structure of potential inputs was analysed using traditional statistical tools, i.e. using the autocorrelation function (acf), partial autocorrelation function (pacf) and the akaike/bayesian information published by atlantis press copyright: the authors 18 d.marcek criterion (aic/bic): we looked to determine the maximum lag for which the pacf coefficient was statistically significant and the lag given the minimum aic. according to these criteria the ar(7) model was specified. for estimation the parameters of an ar+garch model the maximum likelihood procedure was used. the estimation of the model parameters was performed by means of the r-system software and resulted into ar(7) + garch(1,1) process with mean equation 21 0929.07269.0 0.0012ˆ −− ++= ttt rrr 543 0651.00821.00843.0 −−− −++ ttt rrr （6） 76 0218.005591.0 −− ++ tt rr and the variance function 1 2 1 3616.03049.00000169.0 −− ++= ttt heh （7） where te are estimated residuals of tε from eq. (6). after these findings, we calculated predictions for next 32 trading days these predictions are displayed in fig. 3. the same data used for estimating the statistical ar(7) + garch(1,1) model was used to train the neural network above. the variables forming the right hand site of the eq. 6 were used as the input units. the results of the ex post forecasts evaluation are shown in table 1. according to the rmse measure neural networks show slightly worse forecasting results than the ar(7) + garch(1,1) model. 1.18 1.20 1.22 1.24 1.26 1.28 1.30 11 /0 1 11 /0 3 11 /0 5 11 /0 7 11 /0 9 11 /1 1 11 /1 3 11 /1 5 11 /1 7 11 /1 9 11 /2 1 11 /2 3 11 /2 5 11 /2 7 11 /2 9 12 /0 1 czkskk predicted_arma+garch_student prediction arma(7,0)+garch(1,1) w ith t-distribution fig. 3. actual (solid) and forecast (dotted) values of the czk/skk exchange rates time series. table 1. ex post forecast rmses for arch-garch model and rbf nns (see text for details). model numb. of rbf neurons (s) rmse (risk) ar(7)+ garch(1,1) 0.019 5 0.367 rbf nn(classic) 10 0.367 15 0.368 5 0.029 rbf nn(soft) 10 0.030 15 0.033 5 0.029 rbf nn(granular) 10 0.029 15 0.032 now, the question arises whether there is a method which can help to find an optimal forecasting horizon producing minimal forecast errors. in ref 25 the exponential smoothing method is described for continually revising of the forecast error by accounting for more recent changes in the data. this method was also applied in ref. 21 to find the optimal forecasting horizon. suppose that we have a history a sequence of forecast errors 1e , 2e , …, ne . we could average all past errors to obtain ∑ == nt tt nee 1 / （8） and to estimate the standard deviation te σ̂ of the last n forecast errors as 1 )( ˆ 1 2 − − = ∑ = n ee n t tt et σ （9） we logically might wish to give more weight to recent forecast errors than to older data. hence, we can obtain the smoothed error ts as follows (see refs. 25, 26) 1)1( −−+= ttt ses αα （10） where alpha is the smoothing constant, which falls between zero and one. the smoothing procedure (10) allows us to filter local fluctuations out of our forecast error sequence. this procedure is based on averaging past forecast error values in a decreasing (exponential) manner. the term ts is interpreted as a weighted average. the forecast errors are weighted, with more published by atlantis press copyright: the authors 19 risk scenes of managerial… weight given to the more recent forecast errors. the operation defined by eq. (10) is also called simple exponential smoothing. given that we have an estimate of 2ˆ te σ , we can estimate the variance of the smoothed forecast errors, 2ˆ ts σ , as follows 1 )( ˆ 1 2 2 − − = ∑ = n ss n t tt ts σ （11） where ∑ == n t tt nss 1 / , and the standard deviation tsσ̂ of the last n forecast errors as ts σ̂ = 1 )( 1 2 − −∑ = n ss n t tt （12） as an illustration of this concept, consider the analysis of simple exponential smoothing of the fo-recast errors from ar(7) + garch(1,1) process expressed by equations (5) and (6). using equations (8) and (9), we can estimate the standard deviations te σ̂ for the last forecast errors of validation data set (t = 1, 2,…, 31) and these one, when simple exponential smoothing for α = 0.05, 0.010,…, 0.3 is used respectively. these values are graphically depicted in figure 4. in figure 4 we can see that from standard deviations of last forecast errors point of view the minimal risk value of our forecasting model can be reached when the forecasting horizon t is lesser than six. comparing the risk values of our forecasting model based on mse´s criterion in which each forecast error has the same weights and the method based on exponential smoothing concept, we see that the exponential smoothing concept is better. both methods can be used for forecast control. when the risk value exceeds the control limit for two or more successive forecasting periods, this may be a strong indication that something is wrong with the forecasting model. e.g. the parameter estimates used in the forecasting model are not accurate and must be improved. fig. 4. estimated values of standard deviations of forecast errors for validation data set and standard deviations, when simple exponential smoothing used. 6. conclusion in the present paper we showed the procedure of quantitative assessment of risk scene based on probability terms using confidence intervals for point estimates of economic quantities. we build upon measuring uncertainty based on information entropy indicated in bits and on measuring based on prognosis confidence interval, where uncertainty is expressed in terms of the span of the confidence interval and the probability that by using forecasting model the set prognosis limits around the expected value will not be exceeded. we have proposed two approaches for determining the forecast accuracy of the forecasting system applied to exchange rates time series for czech crown (czk) against slovak crown (skk). the first one was based on the latest statistical arch-garch methodology, the second one on the rbf nn. the results of the study showed that there are more ways of approaching the issue of measuring risk in managerial decision-making in companies. it was also proved that it is possible to achieve significant risk reduction in managerial decision-making by applying modern forecasting models based on latest statistical methods and information technologies such as neural networks developed within artificial intelligence. published by atlantis press copyright: the authors 20 d.marcek in direct comparison between statistical archgarch models and fuzzy logic and granular rbf nn, the experiment with the daily data indicates that both methodologies yield very little and similar rmse´s values. but our experiment shows that rbf nn models are economical and computational very efficient, well suited for high frequency data forecasting. we have also shown that applying the exponential smoothing methodology for analysis of past forecast errors in forecasting systems enables as to set an optimal forecasting horizon in which the risk in decision-making process is a minimum. acknowledgements this paper has been elaborated in the framework of the it4innovations centre of excellence project, reg. no. cz.1.05/1.1.00/02.0070 supported by operational programme 'research and development for innovations' funded by structural funds of the european union and state budget of the czech republic. references 1. d. cox d and d. hinkley, d. teoretical statistics, chapman and hall, (london, uk, 1974). 2. s. weisberg, applied linear regression, wiley, (new york, usa, 1980). 3. d. marcek, m. marcek, j. babel, granular rbf nn approach and statistical methods applied to modelling and forecasting high frequency data, international journal of computational intelligence systems, 2-4, (2009):353-64. 4. m. marcek, statistical and rbf nn models, providing forecast and risk assessment, central european review of economic issues, 12, (2009):175-82. 5. d. applebaum d. lévy processes and stochastic calculus, cambridge, (cambridge university press, 2004). 6. j. bertoin, lévy processes of normal inverse gaussian type, finance and statistics, 2, (1998): 41-68. 7. j. havlicky, loss distribution approach by applying theory of the extreme values, in proc. managing and modelling of financial risk, eds. d. dluhosova, 4, czech republic, ostrava, ( 2008): 36-41. 8. e. a. medova, m. n. kriacou, extremes in operational risk management, working paper, university of cambridge, (2001). 9. h. a. simon, rational decision making in business organizations, aer, (1979). 10. h. a. simon, decision making and problem solving, (1986). 11. v. olej, modelling of economic processes based on computational intelligence, pardubice, (the university press of pardubice, czech republic, 2003). 12. c. f. huang, a note on the difference between disaster prediction and risk assessment in natural disasters, new perspectives on risk analysis and cisis response, 9, (atlantis press, paris, france, 2009): 1-7. 13. c. f. huang, d. ruan, fuzzy risks and an updating algorithm with new observation， risk analysis, 28 (3): 681-94. 14. s. ugurlu, c. kahraman, fuzzy multicriteria risk assessment for hazardous waste management: the case for istanbul, journal of risk analysis and crisis response, 1(2001): 29-41. 15. n. mahan, risk assessment is fuzzy business-fuzzy logic provides the way to assess off-site risk from industrial installations, risk, 206, (2004). 16. j. h. m. tah and a. v. carr, proposal for construction project risk assessment using fuzzy logic, construction management and economics , 18 (2000) :491-500. 17. p. e. gaynor, r. c. kirkpatrick, introduction to timeseries and forecasting in business and economics, (mcgraw-hill, inc., new york, usa, 1994). 18. a. p. da silva ap, l. s. moulin, confidence intervals for neural network based short-term load forecasting, power systems, ieee transactions, 15(4) (2000) :1191-1196. 19. r. m. gray, entropy and information theory, springer, new york, (1990). 20. c. e. shannon. a mathematical theory of communication, bell system tech. j., 27 (1948): 379423, 623-656, reprinted in book form with introduction by w. weaver, (univ. of illinois pres, urbana, 1949). 21. d. marcek, m. marcek, p. matusik, high frequency data: making forecasts and looking for an optimal forecasting horizon, in proc. of the ieee 2010 sixth international conference on natural computation-icnc 2010, eds. shigang yue et al, 4 china, yantai, chandong, (2010) : 1686-1691. 22. d. marcek, m. frano, and m. marcek, managerial decision-making: measuring and manifestations of risks and the possibilities of their reducing, journal of economics, 59(4) (2011): 392-411. 23. v. kecman, learning and soft computing: support vector machines, neural networks, and fuzzy logic, (massachusetts, the mit press, 2001). 24. m. marcek, d. marcek, granular rbf neural network implementation of fuzzy systems: application to time series modelling, journal of mult.-valued logic & soft computing, 14 (2008): 101-14. 25. d. c. montgomery, l. a. johnson, j. s. gardiner, forecarting and time series analysis, (new york: mcgraw-hill, inc., 1990). d. marcek czk/skk 26. r. g. brown, smoothing, forecasting, and prediction of discrete time series, (englewood cliffs, prenticehall, 1963). published by atlantis press copyright: the authors 21 microsoft word jracr-20130525_ submit_revise_for_publish_5_an operational drought risk management framework based on stream-f an operational drought risk management framework based on stream-flow intelligent internet control* rongfang li college of information sciences and technology, donghua university, shanghai 201620, p. r. china jiangxi provincial institute of water sciences, nanchang, jiangxi 330000, p. r. china lijun cheng1,2,4, yongsheng ding1,2*, k. khorasani4, yunxiang chen3, and wei wang1,2 1. college of information sciences and technology, donghua university, shanghai 201620, p. r. china 2. engineering research center of digitized textile & fashion technology, ministry of education, donghua university, shanghai 201620, p. r. china 3. jiangxi provincial institute of water sciences, nanchang, jiangxi 330000, p. r. china 4 department of electrical & computer engineering, concordia university, montreal, quebec, h3g1m8 canada abstract in this paper, an operational drought risk management framework based on the stream-flow intelligent internet control is proposed. in the proposed framework drought can be predicted, evaluated and mitigated by using a dynamic stream-flow control under the sensors detection. the framework mainly includes four sequential steps: (i) the stream-flow prediction, (ii) the stream-flow deficit index (sdi) analysis, (iii) the drought multiple regions response, and (iv) the stream-flow balance control. in order to instantiate a specific framework management, intelligence methods are utilized in these processes, namely the generalized regression neural network (grnn) algorithm for the stream-flow prediction and the collaborative particle swarm optimization (cpso) for the reservoirs water collaborative operation. finally, a specific case study corresponding to the fu basin in china is investigated to test the operability and reliability of the proposed drought risk management. keywords: drought risk assessment; generalized regression neural network; dynamic stream-flow prediction; data-driven methods; collaborative particle swarm optimization *corresponding author :yongsheng ding, phd & professor, email: ysding@dhu.edu.cn journal of risk analysis and crisis response, vol. 3, no. 1 (may 2013), 34-43 published by atlantis press copyright: the authors 34 rongfang li et al. 1. introduction water scarcity has frequently occurred in recent years in the world. numerous studies show that the condition is being further aggravated by the rising water demands of municipalities and industries, the growing population, and the gradual climate changes 1, 2. as a result, drought hydrology has been receiving much attention not only in research, but also in public lives 3. drought is a long period of meteorological anomaly characterized by extreme lack of precipitation, which causes significant damages in both natural environment and human life 4, 5. consequently, it is imperative to construct an entity drought management system to explore the inherent laws and manage the risk holistically like a business in order to mitigate the impacts of drought on water resource systems. stream-flow observations collected from near-natural catchments are of paramount importance for water detection and water attribution studies. any organizational regulation actions for the stream-flow has a direct impact on the lives of the residents, the industrial production, the agricultural irrigation as well as the adaptation and policy options 6. the drought management is fully integrated into the stream-flow development efforts, and the manner to control the stream-flow at different lead times often is a first step 7. such a processing with a stream-flow drought management is often referred to as a drought risk management, which can ensure the development process for risks of short-term climate variability and long-term water flow changes 6, 8. in recent years, much effort has been made in the drought prediction, preparedness, adaptation and mitigation, such as the stream-flow drought time series forecasting 5, 9, 10, the real-time drought forecasting in multi-reservoir operations 11, simulating time series of climate variables for drought management 12, etc. all of the above works have detected an aspect of the drought risk by certain spatial variability. the drought risk management has been dispersed at several separable portions sets. however, risk management, as is known as a consecutive procedure which includes risk monitoring, risk identification, risk evaluation and sequential decision-making. each of the sessions depends on the previous stage and has a mutual interaction with the other parts. if they are separated into pieces for investigation, it becomes difficult to understand, apply, evaluate and trust risk management capabilities 12, 13. therefore, there is a need for a comprehensive drought risk management system. in the integrated internal control-risk management (iicrm) framework that is borrowed from business risk management issued by coso in 2004 14, 15, “internal control” is involved in and is an inseparable part of the risk management system. the risk management is no longer only concerned with the prediction of the risk occurrence in each segment, but is also concerned with the analysis of the risk with certain related portions that are all properly integrated. with the advent of internet, more objects are becoming embedded within wireless sensors and are gaining the ability to communicate in the internet 16. to implement a quantitative drought risk management system, one should consider how to use either discrete or continuous measurement systems to understand the operational risk across each of the drought risk prediction, evaluation, decision and regulatory processes and steps. in this paper, an automated and fully data driven drought risk and integrity management system embedded in the internet is proposed which is derived from iicrm. in the proposed framework, the drought variable parameters including the rainfall, the streamflow, and the reservoir can be automatically detected and recorded by a variety of sensors, and then a datadriven empirical modeling technique 17, 18 is used to evaluate the stream-flow as a drought risk degree. finally, a dynamic collaborative model attempts to control all the reservoirs to discharge or bump and achieve a dynamic stream-flow balance in different regions. through a specific case application in the fuhe basin in china, it is confirmed that the risk management mode of the drought can improve water control processes and reduce the drought prediction costs effectively. the main contributions of this paper are as follows: (i) an operational framework for the drought risk management is proposed which is based on the datadriven intelligent internet control; (ii) in the proposed framework, the drought can be measured, evaluated and mitigated by using a dynamic stream-flow adjusting in a life cycle of risk management with intelligent system models, which includes the generalized regression neural network (grnn), drought stream-flow deficit index (sdi) risk identification published by atlantis press copyright: the authors 35 an operational drought risk.. algorithm and the collaborative particle swarm optimization (cpso); (iii) a case scenario is used to clarify and verify the framework applicability and operability in the streamflow drought risk management. the paper is organized as follows. the operational drought risk management framework which is based on data-driven intelligent internet control is presented in section 2. the basic intelligent research methodology is reviewed in section 3. the details of the drought risk management system based on the dynamic stream-flow management and analysis are illustrated in section 4. a case study scenario in the fuhe basin in china is examined through the proposed framework in section 5. finally, concluding remarks and further works are provided in section 6. 2. a data-driven operational drought risk management framework a dynamic drought risk management system is presented according to a simple chronological order in fig. 1 part 1, which is referred to as an operational drought risk management of the internet control (odrmic). the operational drought risk management framework based on the data-driven intelligent internet control consists of seven interrelated steps that are shown in fig. 1 part 1. fig. 1. the operational drought risk management framework based on the data-driven intelligent internet control it should be emphasized that the framework is based on the assumed conditions in the study that the regional stakeholders have established the corresponding proportion of the watershed for coordinating water under the water shortage scenarios. they will provide the support to the coordination group member organizations in assuming their agreed upon roles and responsibilities under the drought situation. the drought risk management system is derived from the manner the management runs an enterprise and is integrated with the management process 19. these components are as follows: (i) setting observable objectives – before drought risk management systems, the studied region would require a careful analysis, and then the observable objectives (known also as control sections) are set, that include the steam-flow monitor points, the reservoirs dam monitor points, and the rainfall hydrometric station monitor points etc. these monitoring points will be used to collect the data to identify the potential drought events. (ii) sensors information and communication – the sensors information are identified, captured, and communicated in a time slice form under different sensors from the monitoring points. effective communication also occurs in a broader sense, flowing down, across, and up to the backend information platform through the network communications. (iii) drought event identification – based on the sensors information, certain methods are used to forecast the future stream-flow, and the predicting results will be channelled back to a drought management’s strategy or the drought plan-setting processes. (iv) drought risk assessment – risks are analyzed through a drought risk index calculation. the index values will impact and then determine the drought warning range of the regions. drought events may be classified according to the index deficit severity. (v) drought risk response – management system can select responses – avoiding, accepting, mitigating, or sharing risk according to the value of the drought severity assessment as performed in step (iv). for a short-term drought, if the drought can be mitigated through controlling the gate opening of the reservoir dam – then activate the remote reservoirs collaborative control process, or else develop a set of actions to align risks with the region government’s risk tolerances as well as the risk response measures. (vi) drought control activities – to modify the drought risk, start remote collaborative control of the reservoirs and then drain the water in the control. however, the (2) remote real-time data collection (1) analyze study basin, set mainstreaming monitor points (control sections) (4) backend information platform (5)the development of scheduling schemes (3) network communications (iv) the remote collaborative control of reservoirs (i) stream-flow forecasting (ii) drought risk index (iii) drought warning level sensors information and communication drought control activities observation objectives setting drought event identification drought risk assessment drought risk response drought management monitoring part 3. prediction and control part 1. data-driven operational drought risk management framework part 2. drought risk management based on dynamic stream-flow (2) remote real-time data collection (1) analyze study basin, set mainstreaming monitor points (control sections) (4) backend information platform (5)the development of scheduling schemes (3) network communications (iv) the remote collaborative control of reservoirs (i) stream-flow forecasting (ii) drought risk index (iii) drought warning level sensors information and communication drought control activities observation objectives setting drought event identification drought risk assessment drought risk response drought management monitoring part 3. prediction and control part 1. data-driven operational drought risk management framework part 2. drought risk management based on dynamic stream-flow published by atlantis press copyright: the authors 36 rongfang li et al. uncontrolled long-term drought, policies and procedures need be established and implemented to help ensure the risk responses for an effective execution. (vii) drought management monitoring – when the system performs the reservoirs dispatching water scheduling schemes, the sensors can monitor in realtime all the monitoring points of the water variance until each monitoring point satisfies the minimum required water, so that the reservoirs drainage action will then stop. drought risk management is an iterative and sequential process in which almost any component can and does influence another one, but it is not strictly performing each process. in the following section, we will instantiate a specific case to explain the drought event identification, the drought risk assessment, the drought risk response and the drought control activities using an intelligent operational method, which corresponds to the fig. 1 part 2. 3. drought risk management based on dynamic stream-flow management and analysis the stream-flow is the volume of water that moves on a designated point over a fixed period of time 20. the flow of a stream is directly related to the amount of water moving off the watershed into the stream channel. it is affected by the weather conditions, whereas it is increasing during the rainstorms and decreasing during the dry periods. on the other hand, it changes with the seasons of the year; decreases during the summer months and reaches the lowest flow for most streams and rivers in most of the country in the months of august and september 10 . in the following sections, the time series stream-flow analysis is represented as the drought behavior and is researched under different periods. there are a number of other factors for drought detection, such as the nature of the soil moisture deficit and the groundwater deficit or the precipitation deficit, etc. the deficits indices are only a part of the droughts measure in any specific application. they are included in this framework as a specific illustrative case study. due to space limitations, the readers can refer to the multi-index drought prediction for more details. fig. 1 part 2 shows a specific stream-flow case to explain the drought risk management process. the blue box sections in fig. 1 part 3 are the refinements of the entire process, and they will be elaborated in detail in the following sections. fig. 2 shows the refinement details of the steam-flow implementation analysis for the drought risk management by using data-driven intelligence methods. in fig. 2, each counterpart of fig.1 part 2 has been marked. all of these processes in fig. 2 are performed in sequence as follows. (i) monitor the real-time stream-flow, the rainfall and the reservoir water level in each control section. (ii) the grnn model is established to forecast the short-term and the long-term stream-flow based on the historical and real-time sensor data, respectively. the detail model construction is illustrated in section 3.1. (iii) a stream-flow deficit index (sdi) is calculated as a drought evaluation standard for measuring the degree and level of the water resource shortage. through the sdi, drought-warning level in different regions is obtained and it will yield a drought warning level, corresponding to a drought response degree. the details on the calculations are illustrated in section 3.2. (iv) the reservoirs control system will decide whether to perform the water dispatching or not as well as which reservoirs need to be discharged with the drought degree levels established in step (iii). (v) when the drought degree reaches or exceeds the early drought warning grade and level, a real-time water regulation is initiated to perform by the cpso in parallel. cpso will deploy the water in each reservoir and decide which one need to drain and how long the reservoirs drainage last, the details on the process of the overall water allocation is illustrated in section 3.3. (vi) the stream-flow remote sensing device can detect water variation in every control section and renews these real-time data into the backend information platform. (vii) when the real-time stream-flow of the control sections in the information platform satisfies the minimum water usage, then the reservoirs discharging stops. (viii) output the final water dispatch scheduling schemes to the water managers as well as the impacted water departments. published by atlantis press copyright: the authors 37 an operational drought risk.. fig. 2. a detailed process of the data-driven steam-flow control for drought control. in the following, we are going to describe in detail the key steps that are described under items (ii), (iii), (v). 3.1. data-driven stream-flow prediction model the stream-flow is relevant to hydrological and meteorological problems and tasks. by integrating the long-term historical data with the real-time collected data in every hydrological station a time series analysis can be performed in order to detect the drought risk in future several days or a month. many data-driven models 21, including linear autoregressive (ar) 22, nonparametric or nonlinear chaos-based neural networks (nn) 23, have made great progress in hydrologic time series prediction in the past decades. grnn algorithm is a typical chaos-based forecast method for dynamic time series data 24. it uses a non-parametric probability density estimation to approximate the underlying regression surface of the training data available. for this reason it is also called the probabilistic neural networks. given a dataset t having n examples, t= 1 1 2 2{( , ), ( , ),..., ( , )n nx y x y x y , ,i n ix r y r∈ ∈ , i=1,2,…,n}, where ix is an input variable and iy is the desired target. let us define the euclidean distance di, which is used as a measure of how well the training sample ix can represent the point position of prediction input x . the estimate ˆ ( )y x can be visualized as a weighted average of all of the observed value iy , where each observed value is weighted exponentially according to its euclidean distance from x 24, that is: 2 2 1 2 2 1 exp( || || /2 ) ˆ ( ) exp( || || /2 ) n i i i n i i y d y x d δ δ = = − = − ∑ ∑ (1) where ||di|| 2=(x-xi) t(x-xi), andδ is the square deviation, also known as the smoothing parameter. grnn functions through the four-layers, namely: input layer, hidden layer, summation layer and output layer. the input layer only passes the input vector to the rbf hidden layer; in the hidden layer, the calculation performed in each pattern neuron is 2 2exp( || || /2 )id δ− , with a normal distribution centers at each training sample. in the summation layer, the signals of the i-th pattern neuron, applied to the denominator neuron are weighted with the corresponding values of the training samples yi. the weights on the signals applied to the numerator neuron are one. in the output layer, the output is simply a weighted average of the target values of the training cases close to the given input case as given by the expression in eq.(1). 3.2. drought risk identification and evaluation drought is generally considered as periods with insignificant water resources for sustaining activities of a region. there are a variety of indices to definite the drought degrees, generally the most adopted degree of drought is what the american meteorological society has proposed in 2004 25, namely the degree of the water deficit. in the present study, the stream-flow deficit index (sdi) is used as an index to evaluate the drought degree. sdi is statistically similar to most other commonly used standard precipitation index (spi) 26 , which is based on the basic method used to derive drought events from continuous or discrete records of precipitation. the sdi is calculated by scaling the cumulative stream-flow values by using the median, the maximum, and the below minimum cumulative values as expressed in eqn. (2)-(3). it overcomes the problem of fitting a statistical distribution to the cumulative stream-flow data 30, specifically, if ,i j jsw msw≤ , , , 100min . i j j i j j j sw msw sdi msw sw − = − × − (2) multi-reservoirs drainage with collaborative pso algorithm, the overall water allocation (fig.1 part 3 (iv)) renew stream-flow for each control sections n output the finial water dispatch scheduling schemes (fig.1 part 2 (5)) y determine the time slices, the grnn model for a stream flow prediction based on data-driven (fig.1 part 3 (i)) monitor stream-flow, rainfall in each control section(fig.1 part 2 (1)(4)) start n y satisfy the minimum water usage calculate fitness function determine pbest, gbest satisfy ending condition set reservoir discharged volume as particles, initialize particles velocity, position and parameters in threshold velocity satisfy constrains n y n particles velocity mutation renew particles y gbest constrains stream-flow deficit index (sdi) and drought risk response (fig.1 part 3 (ii)) activate water dispatching (fig.1 part 3 (iii)) end published by atlantis press copyright: the authors 38 rongfang li et al. if ,i j jsw msw> , , , 100max . i j j i j j j sw msw sdi sw msw − = × − (3) where i denotes the years, j denotes the day of the year (1 to 365), ,i jsdi denotes the stream-flow deficit index (%), ,i jsw denotes the cumulative stream-flow for the period of interest (mm), jmsw denotes the long-term median cumulative stream-flow for the period of interest (mm), max . jsw denotes the long-term maximum cumulative stream-flow for period of interest (mm), and min . jsw is the long-term minimum cumulative stream-flow for period of interest (mm). these equations are used for scaling in order to create an index which has a lower bound of -100 and an upper bound of +100. 3.3. optimal allocation of water by reserves dispatching the stream-flow can be divided into the two parts, the main-streaming and the tributary. the water shortage in the tributary part can pass its information to the main-streaming level by level so that ultimately the information will be fed back to its control reservoir. the reservoirs in different study areas cooperate and coordinate their activities to maintain a water balance and the stream-flow obtains the overall water allocation. the behavior of the reservoir is known as a collaborative optimal scheduling of the multi-level water. the particle swarm optimization (pso) algorithm, which is derived from the cooperative behavior of decentralized, self-organized systems 27, is an evolutionary computational technique 31. it has been widely applied to multivariate function optimization problems. here, the collaborative pso (cpso) algorithm is used to determine in parallel the collaborative optimal water discharging in different reservoirs in multi-regions. due to space limitations the reader can refer to the literature 28 and 29 for detailed processes and usage of this algorithm. we assume that the reservoir outflows are represented as particles and the solution hyperspace dimensionality d of each reservoir is associated with the dispatch period t. given that for scheduling the time is 12 months a year, with month as a dispatching unit, then d is set to 12 and for day as a dispatching unit, then d is set to 365. the particle iq at the i-th iteration is defined according to 1 2 1 2 1 2 1 1 1 2 2 2( , , , , , , , , , , , , ) d d d i i i i i i i ik ik ikq q q q q q q q q q= l l l l (4) where tijq represents the j-th reservoir discharge at the tth time corresponding to the i-th iteration, and k denotes the number of reservoirs. the particle velocity vector iv defines the velocity change rate of the reservoir discharge corresponding to the i-th iteration, and the velocity of each particle represents the transition rate of every reservoir outflow according to 1 2 1 2 1 2 1 1 1 2 2 2( , , , , , , , , , , , , ) d d d i i i i i i i ik ik ikv v v v v v v v v v= l l l l (5) in practice, the velocity rate of the reservoir discharge is directly proportional to the reservoir storage water capacity. hence, in the reservoir discharged process, the reservoir storage water always becomes smaller and smaller than the previous time. this will cause the reservoir flow rate to have different variation ranges in different times. in this case, the reservoir outflow capability has a variable scope. the minimum and maximum outflow velocities can be calculated according to the following equations, respectively, corresponding to the t-th period, t t min min max(1 ) t i i iv q qα α= + − (6) t t max min max(1 ) t i i iv q qβ β= + − (7) where tmaxiq denotes the maximum reservoir outflow capability of the i-th reservoir, tminiq denotes the minimal reservoir outflow in t-th period, (1 ()) / 2randα = + , (1 ()) / 2randβ = − , 1 j n≤ ≤ , 1 t d≤ ≤ , and ()rand is a random variable. the collaborative optimal water discharge in different reservoirs in multi regions is given as follows, which is shown in the right part of fig.2.: step 1: initialize n particles in the population { }1 2, ,..., ,...,i nu p p p p= , set the parameters in the pso, namely the particle population size, the solution hyperspace dimensionality, and the maximum number of iterations; step 2: calculate the values of the fitness function which can be defined as the minimum shortage of the water storage in each control section: min [ ( ) ( )]s s s fitness d t r t= −∑ (8) where ( )sr t , ( )sd t denote the s-th control section’s water supply capacity (the recommended reservoir outflow) and water demand capacity, respectively. the goal is to have the particles move towards a small fitness direction. step 3: select the particle position with the lowest fitness value that is observed in all particles in a certain generation as the best overall position ( gbest ) gdp and the smallest fitness value in each generation as the best current generation position ( pbest ) bdp ; published by atlantis press copyright: the authors 39 an operational drought risk.. step 4: check the termination condition. if the maximum number of iterations has been reached or the required minimal fitness error accuracy has been obtained,then the best individual is obtained in the current population, output the ( gbest ) gdp and terminate the process; otherwise, go to the next step; step 5: perform the particle velocity mutation using eq. (9) : 1 1 1 2 2( ) ( ) iter iter iter iter iter iter iter iter i i ibest i best iv wv c r p q c r g q + = + − + − (9) where iter denotes the iteration time; 1 i n≤ ≤ ; 1c and 2c are the acceleration coefficients, that are commonly set to 2; 1 iterr and 2 iterr denote two random variables in the range [0, 1]; and 1 w iter = denotes an inertia weight whose values decline linearly over iter . step 6: set the velocity in its maximum or minimum value range if it exceeds the velocity limitation given eqn. (6) and (7). step 7: update particle’s position vectors by using eq. (10), then return to step 2. 1 1iter iter iter i i iq q v + += + (10) 4. a case study 4.1. study area the fuhe basin that is located in the jiangxi province of china has an area of 15,811km2 and is covered with significant agricultural fields. the fu river is its main water supply source which originates from the south guangchang city and ends in the north poyang lake, covering a total length of 312 km, with a yearly average flow of 483 m3/s. the basin has much rainfall every year. however, it suffers from the subtropical high pressure every year, and the spatial and the temporal distribution of water resources are extremely uneven. the yearly mean of precipitation is 838mm from april to june, accounting for 48% of the annual precipitation. however, during the agricultural peak months of july to september, it only has 329mm of precipitation, which accounts for only 17.5% of the annual precipitation. during the period 2003-2009, thousands of people had to face with the drinking water shortages in that period. in order to have an appropriate arrangement usage of the water, a drought risk forecast and water resource management are drawn up in advance so that governments have a good measurement plan to confront with the upcoming water usage contradictions. data of the rainfall and the stream-flow from 1970 to 2008 are collected from 9 hydrometric stations (these are also caller control sections) and the data of the water level and tumble flow are collected in two big reservoirs. fig. 3 shows the network schematic diagram of the fuhe basin. the control sections, the large reservoirs and the main water usage regions are marked in the diagram. at the same time, the control sections as detection points will collect real-time runoff (steam flow) and precipitation data from the sensors, which lie in the key points of the river. the yearly statistical properties of the steam-flow of each control sections are shown in table 1. it can be seen that there exist uneven water distributions in different regions in different seasons. 4.2. results and discussion when the data of the stream-flow sensors is entered into the information database, the grnn will predict the stream-flow variation and determines whether the control section has a lack of water or not according to its sdi degree. if one of the control sections detects a water shortage, it will send the information to the inflow direction river converge reservoir control section 4 5 8 9 1 2 6 r2 r1 main inflow direction river source 3 7 fig. 3. the schematic distribution of the control sections, the large reservoirs and the main water usage regions in the fuhe basin. published by atlantis press copyright: the authors 40 rongfang li et al. table 1. statistical properties of control sections for the annual average runoff and rainfall series in the fu basin (1970-2008) control sections annual average stream-flow series annual average rainfall series no min (m3/s) max (m3/s) mean (m3/s) standard deviation min (mm) max (mm) mean (mm) standard deviation 1 shaziling 5.3748 496.1733 103.1576 78.1151 3.4822 4.3126 6.20 2.0645 2 nancheng 2.0258 423.66 33.8196 22.9884 3.6561 4.5747 7.1258 2.0709 3 miaofang 2.3676 260.9882 39.8958 28.0788 3.1025 4.0379 4.2709 1.9535 4 miaojiawan 1.4154 127.8235 23.5434 14.6113 3.6024 4.6659 7.720 2.0581 5 lijiadu 0 44.2049 2.9765 1.5482 3.4132 4.3996 4.7741 1.8387 6 lichuan 0 325.17 64.0218 35.6321 4.0577 5.0091 7.27 2.4096 7 hongmen 0.3451 100.3733 7.3417 4.6551 3.7376 4.6337 6.8133 1.5903 8 loujiacun 0.5251 207.1633 16.9233 9.3239 3.4835 4.7675 7.7036 2.06 9 mawei 0.1004 560 31.9848 19.1619 3.8131 4.9411 7.6 2.4484 average 0 560 28.9251 49.5302 3.6437 3.7852 9.36 1.9865 table 2. the water dispatch proportion for each control section station no.1 no. 2 no. 3 no. 4 no. 5 no. 6 no. 7 no. 8 no. 9 proportion 4.78% 0.60% 10.44% 16.64% 5.33% 14.79% 16.00% 12.11% 8.66% table 3. the reservoir storage outflow for the water dispatching in different months in 2008 (unit: m3/s) reservoir jan. feb. mar. apr. may june july aug. sept. oct. nov. dec. r1 12.57 18.69 21.63 35.02 97.42 355.80 479.07 189.78 210.74 92.56 126.15 14.028 r2 51.00 68.24 87.93 91.21 141.46 459.10 437.65 284.85 136.53 77.72 68.73 42.40 table 4. the cooperative results of the water dispatch proportion for the entire control sections control sections jan. feb. mar. apr. may june july aug. sept. oct. nov. dec. no.1 4.4% 4.3% 4.3% 4.8% 4.6% 4.5% 4.6% 4.6% 4.7% 4.4% 4.5% 4.2% no.2 0.5% 0.5% 0.5% 0.6% 0.6% 0.6% 0.5% 0.5% 0.6% 0.5% 0.5% 0.6% no.3 10.2% 9.9% 10.1% 10.4% 10.4% 10.5% 9.4% 10.1% 10.3% 10.1% 10.2% 9.6% no.4 14.6% 15.1% 16.3% 15.4% 16.6% 16.7% 16.4% 16.2% 15.9% 16.3% 16.4% 15.5% no.5 4.8% 4.8% 5.0% 5.0% 5.3% 5.1% 5.3% 4.9% 4.8% 5.0% 4.7% 4.2% no.6 13.7% 14.3% 14.5% 12.5% 14.8% 14.5% 14.5% 14.6% 14.5% 14.3% 14.6% 14.0% no.7 14.1% 15.4% 15.4% 13.5% 16.0% 16.1% 15.7% 15.3% 15.9% 15.9% 15.8% 15.8% no.8 10.1% 11.4% 9.7% 9.5% 12.1% 12.1% 11.7% 12.1% 11.7% 11.9% 12.1% 10.5% no.9 8.2% 8.4% 8.5% 7.6% 8.7% 8.5% 8.5% 7.8% 8.0% 8.3% 8.3% 7.4% published by atlantis press copyright: the authors 41 an operational drought risk.. reservoir controlling system. after the discharging instruction is accepted and agreed upon, a dispatch water plan is initiated using the cpso algorithm by adjusting the outflow of each reservoir. it will optimize the water volume for every stream segment to avoid a water shortage. in the following, we simulate a cooperative result of the water dispatch proportion in a month. the primary parameters of the cpso algorithm are set as follows: 12d = (month), 50m = , max 1000iter = , max 0.9w = , min 0.4w = , 1 2 2c c= = , where d represents the solution hyperspace’s dimensions; m denotes the particle size of population; maxiter denotes the maximum iterations; maxw and minw denote respectively the maximum and the minimum inertia weight; and c1 and c2 denote learning factors. the maximum and the minimum particle velocities can be calculated according to eq. (6) and eq. (7), respectively. table 2 shows the corresponding proportion of the watershed for coordinating the water during its shortage which is established by the study regional stakeholders. table 3 shows the computational results of the reservoir storage outflow operation for the drought mitigation by using the cpso algorithm for the next anticipated month. it can be seen that the reservoir storage outflow from the months of june to september are larger than the other months. table 4 shows the simulations for the water dispatch proportion of each control section after dispatching water for the next anticipated month. comparing table 2 with table 4, the overall results in table 4 are almost close to those in table 2. furthermore, the dispatching results have reached the water balance in different regions and have met the different region water demand, especially in the agricultural water requirement peaks from the months of june to september. 5. conclusions and further work with the advent of telecommunication revolution, it is a huge challenge for decision-makers to reconsider the drought risk management in the internet environment. in this paper, we have proposed an operational drought risk management framework based on data-driven and intelligent internet-based control solutions. the framework that is outlined addresses the drought identifying risk, the drought early warning system, the drought dynamic response and the drought mitigation solutions by means of management concepts. it transforms the new idea of applying an automated, consistent, data-driven methodology to ensure that the drought risk management is effective and autonomous. by utilizing the proposed management system, the drought mitigation becomes simple and operational by handling the captured dynamic data. in addition, water administrators will have better information on not just deciding on the current dry state, but also easily forecasting the anticipated drought in future times with confidence. for the drought risk management framework, a specific stream-flow management and analysis is developed and presented in detail for each part of the drought risk management process. through the use of the generalized regression neural network (grnn) the stream-flow forecast for the drought event identification is achieved; by using the stream-flow deficit index (sdi) analysis in different monitoring points, the drought risk assessment is obtained, and finally, the remote collaborative control of reservoirs using the collaborative particle swarm optimization (cpso) algorithm is implemented. the case scenarios corresponding to the fuhe basin in china illustrates that the drought risk management is feasible and operational, and has a potential benefit for many thousands of households and livestock in the fuhe basin. the framework for the drought risk management is easy to operate, and the alternative implemented models are feasible in managing both short-term as well as long-term drought processes in another territory even when the available input datasets is different. acknowledgements this work was supported in part by the key project of the national nature science foundation of china (no. 61134009), the national nature science foundation of china (no. 60975059), specialized research fund for the doctoral program of higher education from ministry of education of china (no. 20090075110002), specialized research fund for shanghai leading talents, project of the shanghai committee of science and technology (nos. 11xd1400100, 11jc1400200, 10jc1400200), and the fundamental research funds for the central universities. published by atlantis press copyright: the authors 42 rongfang li et al. references 1. a. k. mishra, v. p. singh, v. r. desai, drought characterization: a probabilistic approach, stoch environ res risk assess, 23 (2009): 41-55. 2. b. c. bates, z. w. kundzewicz, s. wu and s. palutik eds., climate change and water, intergovernmental panel on climate change secretariat, (ipcc, geneva, 2008): 210. 3. k. m. ashok and p. s. vijay, drought modeling a review, journal of hydrology, (403) (2011): 157-175. 4. jr. r. r. heim, drought indices: a review, chapter 11. in drought: a global assessment (donald a. wilhite, ed. volume 1), (routledge: london, 2000) ,159-167. 5. a. k. mishra and v. p. singh, a review of drought concepts. j. hydrol. 391(1-2) (2010): 202-216. 6. united nations development programme drylands development centre united nations , mainstreaming drought risk managementa primer, undp 2011: 7-11. 7. c. fernández, j. a. vega, t. fonturbel and e. jiménez, streamflow drought time series forecasting: a case study in a small watershed in north west spain. stoch. environ. res. risk assess, 23(2009): 1063-1070. 8. k. stahl, h. hisdal, j. hannaford, l. m. tallaksen, h. a. j. van lanen, e. sauquet, s. demuth, m. fendekova, and j. jódar, streamflow trends in europe: evidence from a dataset of near-natural catchments, hydrol. earth syst. sci. discuss., 7 (2010): 5769-5804. 9. c. l. wu, k. w. chau, data-driven models for monthly streamflow time series prediction, engineering applications of artificial intelligence 23 (2010): 13501367. 10. r. modarres, stream-flow drought time series forecasting, stoch environ res risk assess 21 (2007): 223-233. 11. w. c. huang and c. c. chou, drought early warning system in reservoir operation: theory and practice, water resources research, 41 (2005): 12. 12. a. steinemann, using climate forecasts for drought management. j. appl. meteorol. climatol, 75 (2006): 1353-1361. 13. j. g. carbone and k. dow, water resource management and drought forecasts in south carolina. j. am. water resour. assoc. 41 (1) (2005): 145-155. 14. enterprise risk management-integrated framework, committee of sponsoring organizations of treadway commission report, aicpa , 2004. 15. y. h. pang and d. j. shi, integration of internal control and risk management, in proc. international conference on business intelligence and financial engineering, (2009): 369-372. 16. l. coetzee and j. eksteen, internet of things – promise for the future? an introduction, ist africa 2011, gaborone, botswana, 11-13(2011): 9. 17. j. i. ansell and t. w. archibald, data driven risk based management of assets in the water industry, international journal of perform ability engineering, 4(3) (2008): 215224. 18. r. s. levine, data-driven operational risk managementrisk, (amazon publish, 2008). 19. blackwood and b. kristen, a data-driven approach to managing operational risk: by identifying, analyzing, and managing operational risks, organizations can undertake new initiatives with greater confidence despite a turbulent economy, the rma journal, 91(7) (2009): 68-72. 20. t. murdoch and m. cheo, adopt-a-stream foundation. field guide: watershed inventory and stream monitoring methods, (everett, wa, 1996). 21. j. h. david and s. m. barbara, anomaly detection in streaming environmental sensor data: a data-driven modeling approach, environmental modeling & software, 25(9) (2010): 1014-1022. 22. m. morawietz, c. y. xu, l. gottschalk and l. tallaksen, systematic evaluation of autoregressive error models as post-processors for a probabilistic streamflow forecast system , journal of hydrology , 407(1-4)(2011): 58-72. 23. r. modarres, multi-criteria validation of artificial neural network rainfall-runoff modeling, hydrol. earth syst. sci., 13(2009): 411-421. 24. f. s. donald, a general regression neural network, ieee trans. neural networks, 2(6)(1991): 568-576. 25. e. koleva and v. alexandrov, drought in the bulgarian low regions during the 20th century, theoretical and applied climatology, 92(1-2) (2008): 113-120. 26. f. fiorillo and f. m. guadagno, karst spring discharges analysis in relation to drought periods, using the spi, water resources management, 24(9)(2010): 1867-1884. 27. r. c. eberhart, y. h. shi and j. kennedy , swarm intelligence (the morgan kaufmann series in evolutionary computation), usa: academic press, usa: (morgan kaufmann publishers, 2001), 369-392. 28. w. wang, y. s. ding, r. f. li and l. j. cheng, a modified particle swarm optimization algorithm for optimal scheduling of multi-reservoir system, in proc. 3rd international conference on risk analysis and crisis response, ( laredo, texas, usa, 2011) 29. r. f. li, l. j. cheng, w. wang and y. s. ding, a particle swarm optimization algorithm based collaborative optimal scheduling for multi-level water basin in non-flood season, in proc. 2011 international conference on modeling, identification and control, (shanghai, china, 2011),376381. 30. drought monitoring index for texas, technical report by the texas a&m research foundation (rf-468511) to the texas water development board on behalf of the project investigators, 2007. 31. j. kennedy and r. c. eberhart, particle swarm optimization, proc. ieee int'l conf. on neural networks, piscataway, nj: ieee press, iv, (1995): 1942-1948. published by atlantis press copyright: the authors 43 microsoft word risk indviduality in crisis planning.doc 
 risk indviduality in crisis planning: the case of gender in amercan agriculture dana l.k. hoag colorado state university professor, agricultural and resource economics fort collins, co 80523-1172 dhoag@colostate.edu catherine m h keske colorado state university assistant professor, soil and crop sciences associate director, institute for livestock and environment fort collins, co 80523-1170 usa catherine.keske@colostate.edu rebecca goldbach research assistant, agricultural and resource economics, fort collins, co 80523-1172 crisis response can be improved by accounting for differences in risk perception. we confirm that gender individuality can be an important divider by using a case study of agricultural producers. women were more risk averse and less confident. however, we found less difference than studies have found in other sectors. our research shows that context matters. results will vary by geographical location. gender is one division that should be considered in crisis planning; however, it will not always be important. keywords: agriculture, gender, risk management, crisis response. 1. introduction as related to crisis response, "risk management … encompasses a broad range of activities to increase capacities and reduce vulnerabilities to prepare for, cope with, respond to and recover from disaster events.1” but exactly whose risk is being addressed? no single response plan can be universally effective since everyone perceives and reacts differently to risk. the effectiveness of a plan, however, can be improved by targeting groups when there are clear and systematic divisions or subcultures. for example, a plan for responding to extreme weather might group livestock pastoralists in kenya differently than crop producers. it is difficult and sometimes impossible, however, to accurately aggregate, homogenize or stereotype social systems. for example, pastoralists from differing tribes might respond to risk as differently from each other as they do from crop producers. nevertheless, it is important to develop crises response plans that properly account for the social contexts in which risks occur, especially when the differences are large and/or important1. journal of risk analysis and crisis response, vol.1, no. 1 (july, 2011). published by atlantis press copyright: the authors 21 hoag,
keske
and
goldbach
 
 
 a growing body of research shows that there might be a systematic division between men and women when it comes to risk. according to the hyogo framework for action 2000-2015: on building resilience of nations and communities to disaster, “a gender perspective should be integrated into all disaster risk management policies, plans and decision making processes, including those related to risk assessment, early warning, information management, and education and training.2" however, an unsettled question in the literature is whether men and women perceive and react to risk in the same way. for example, some studies show that women are more risk averse than men3 or that women are less confident.4 others have found no differences.5 whether differences are found likely depends on the context. if, when and where differences are found, they should be integrated into risk analysis and crises response. that is, women and men are a good example of two sub-cultures that may need very different plans as directed by the hydro framework for action. however, as we will show, every case could be unique. therefore, no category, even gender, is automatically different enough to justify unique risk planning. risk analysis in economics and other financial fields can be a useful guide where there can be wide differences in risk attitudes.6 the goal of this study is to apply that knowledge to examine the differences in risk perception and management between men and women in one setting that can confirm whether risk gender individuality is important to account for in crisis planning. specifically, we examine farm or ranch opperators/producers in the united states. there is no information about risk management differences across gender in agriculture, despite the fact that u.s. census data shows that female-operated farms are a rapidly growing component of the sector.7 we base our analysis on two surveys conducted at risk management programs in 2008.8 data were collected in a self assessment survey and a revealed preference survey connected to a computerized risk management simulation program called ag survivor.9 we examine differences between men and women in the areas of risk preference, confidence, motivation, and influence . the purpose of this particular examination is to demonstrate that differences do occur, but that gender differences in agriculture are different than gender differences in other business sectors. due to these differences, results might not necessarily transfer to females in kenya. however we can learn some lessons about why women and other sub-cultures might need individual attention. 2. literature review female business ownership has increased steadily in the united states. according to a 2006 study10, women now own 41% of all privately held firms in the united states. census data found that the number of women who owned businesses grew 43% from 19901994.11 according to a 2001 article, “the relative surge of women entrepreneurs in the last decade has been nothing short of revolutionary.11” census data did not start differentiating male and female farm operators until 1978, but it is assumed that females have been involved in agriculture for much longer. commercial agriculture in the united states is widely dominated by males; as of 2002, only 5% of commercial farms in the us were operated by women. the numbers for commercial farms show greater differences in farm operation across gender than do private farms, as this is where females have been employed for decades, often as co-operators with their husbands. the number of women operated private farms is increasing. women were principle operators of 209,784 farms.12 this number increased to 237,819 in 2002, an increase of 13.4%. 13 the national agricultural statistics service created a publication entitled women in agriculture: quick facts from the 2002 census of agriculture which highlight women’s impact on agriculture.14 the data show that female-operated farms are smaller than maleoperated farms both in acres and sales; women’s farms also have a smaller average income than male farms, indicating a less favorable financial position. finance theory suggests that the higher the risk, the higher the expected rate of return on investment. as will be shown later, females have been found to be more risk averse than males. this would suggest that males, who have a higher risk preference, will have higher returns on their investments; thus, differences in risk taking published by atlantis press copyright: the authors 22 
 risk
individuality
 
 propensity may help explain differences in the financial performance of maleand female-owned businesses. a 2003 study5 found that although profits are significantly higher for males, so are the variations in profits. after adjusting for the variability associated with risk, this study found no significant differences between the performances of maleand female-owned businesses. the lower risk-taking propensity in females can also help explain the differing level of growth of maleand female-owned firms, where male-owned firms show a faster rate of growth. based on their greater risk aversion, female entrepreneurs seem to be more concerned than male entrepreneurs with the risks associated with fast-paced growth and tend to deliberately adopt a slow and steady rate of expansion.15 it has also been found that women accumulate less savings for retirement and that this may, at least in part, be connected to their risk preferences.16 a 1998 study16 displays how risk aversion can affect wealth accumulation later in life, and that women’s lower risk preference may result in their having less wealth in retirement then men. their review of the literature finds that when wealth equals $20,000, single women are predicted to hold 43% of that wealth in risky assets and single men are predicted to hold 51% of their wealth in risky assets. when wealth increases to $100,000, both sexes will allocate a greater percentage into risky assets but for males this increase is greater than for women (an increase for women of 19% and an increase for men of 28%). one study looked at differences across gender related to the optimal storage rule.17 the results from this analysis were as follows: women sell less often than men, men sell earlier than women, and the longer a producer stores the lower the price the producer is expected to receive. the authors state that the most likely explanation for women having fewer sales than men is that men gain more utility from trading than women. in other words, the authors attribute this difference to men’s higher propensity to enjoy competition. overall, the authors found that because women store longer, they tend to make 1.4 cents less per bushel than men. 2.1. risk preference past research, based on finance theory, found that greater risk-taking preferences are positively correlated to performance in small firms; however, this past research has defined performance in terms of sales and profit. when results are controlled for risk and the variations in financial profit caused by risk, differences in performance between genders disappear. women appear to be more risk averse than men.3,18 studies found that females stay in the market less across all levels of cost than males, indicating that females have a lower risk preference.18 2.2. confidence studies have also found that women are less confident and estimate a higher likelihood of the negative outcome, making them less likely to take the risk.4 psychological studies show that most people are overconfident about their own relative abilities, and this overconfidence affects business performance. research on confidence and business performance has found that overconfidence is not necessarily a good thing,19 finding that overconfidence can lead individuals to undertake activities that are not in their best interest. research has also shown that men are more overconfident than women.20,21 one study that finds such results is grain price expectations of illinois farmers and grain merchandisers.22 these researchers found that the phenomenon of overconfidence holds in their sample 2.3. motivation individual motivations can highly influence business success. studies found that women indicate their most important entrepreneurial motivations are: desire for challenge, self-determination and desire to balance family and work responsibilities. for women, profits and growth were important, but not as important as the above-stated motivators. in contrast, male entrepreneurs were most motivated by the need to improve their positions in society for themselves and their families.23 ” if females’ main motivations are not sales and growth, it could explain some of the performance gap between the genders. however, published by atlantis press copyright: the authors 23 hoag,
keske
and
goldbach
 
 
 research on women entrepreneurs has found differences in motivations across women.24-27 2.4. influence according to one study, “for decades, researchers in various domains of the bio-behavioral sciences have had an enduring affinity for the idea of the dominant male….females, by contrast, are often portrayed as being far less aggressive, more communal, and less interested in social politics.28" if this statement is true, it might affect how individuals influence the decisionmaking process in their business. an individual’s influence over business decisions will affect their ability to impact the performance of their operation. research indicates that males exert more influence over the decision making process.29 the literature confirmed many common gender findings, such as men describing themselves as having a higher ability to influence others, and peers viewing males as having higher influencing ability.28 3. research methods the data in this study was collected from agricultural producers that live and work in colorado. two survey instruments were developed and administered to measure gender differences in agricultural producers. the first instrument, a selfassessment survey, asked questions about how individuals perceive themselves with regard to risk preferences and decision-making variables. the second survey was a revealed behavior instrument developed to accompany a computerized simulation game called ag survivor,9 which simulates realistic farm management decisions. using experimental risk preferences based on a computer simulation to compare risk seeking by females and males has been done before.18 we agreed with these authors that a computerized approach was the best option because it “reduces the likelihood of gender effects from peer group pressure, public performance and perception of others relative to self.18" four hypotheses were developed that correspond to the four areas where we found differences between men and women in the literature: 1. h1: risk aversion: women show more risk aversion than men. 2. h2: confidence: women will display less confidence about their decisions than men 3. h3: motivation: women are less motivated by profit than men 4. h4: influence: women will have less influence than men. each survey addressed these four hypotheses. the self-assessment survey assigned a numerical scoring system using a scale of one to five for 10 questions. the ag survivor survey ranked individual risk preference on a scale of zero to three, and confidence on a likert scale of one to five based on the way each survey respondent played the ag survivor scenario. the averages of scores for both surveys were compared across gender and analyzed using a oneand/or twosample t-test assuming unequal variances in microsoft excel. there were 82 usable self-assessment surveys and 81 useable ag survivor surveys. the self-assessment survey included 61 females (74%) and 21 males (26%). the ag survivor survey had a more equal distribution with 44 females (54%) and 37 males (46%). a summary of the participant demographics is presented in table 1. three demographic variables were collected: age, number of children, and marital status. marital status was the only statistically significant variable that was different between the genders. sixtyfive men and 45 women reported their age on the survey, with the male average age being 45 years and the female average 48 years. the male age range was 23-77 years, while the female age range was 21-84 years. sixty-one men and 45 women reported their number of offspring. the average number of children for males was 1.8 while the average number for females was 1.4. the range of children was also similar across gender, with males ranging from 0-6 children and females ranging from 0-5 children. published by atlantis press copyright: the authors 24 
 risk
individuality
 
 male female average age 45 48 percent married 81 57 average number of children 1.8 1.4 table 1: demographic features of survey respondents 3.1. self-assessment survey the self-assessment survey was administered to individuals at risk conferences around northern colorado. to conceal their identity, but link the two surveys, individuals were assigned a common number for the self-assessment surveys and the ag survivor survey. a linked number also saved subjects the trouble of filling out demographic information twice. the self-assessment survey consisted of ten questions, some of which had multiple parts, leading to 16 individual parts. two of the parts aimed to assess whether the individual had a business partner and, if so, what relationship the partner had with them (spouse, friend, etc.). five of the parts asked individuals questions about their investment portfolio or about hypothetical investments they could make. there was also an investment question asking about the business partner’s investment portfolio. two questions asked about the respondent’s perceived capability of running the business alone, one about the individuals perceived capability and another about their business partner’s capability. there was also a part addressing with overall business satisfaction, influence, correctness of decisions made, positive motivations, and negative motivations for the respondent’s actual business. the survey ended with a question asking individuals who they seek input from when making a business decision. 3.2. ag survivor survey the revealed behavior, ag survivor survey was administered at the same conferences as the selfassessment survey. at each of these conferences, individuals participated in a farm computer-simulation program called ag survivor.9 participants were asked to track both their group decisions and the decisions they would have made had they been acting on their own. the survey also asked participants to document how confident they were in their decisions. surveys were administered at conferences that used different versions of the ag survivor game, so the decisions had to be normalized across versions to allow comparison across conferences. the conferences had different versions since the ag survivor program is targeted to provide realistic risk management decisions to different groups. participants played the lazy u ranch scenario (about sheep) at four conferences, the king family ranch (about cattle ranching) at one conference, and the wheatfields (wheat farming) at two conferences. the ag survivor program9 is an innovative risk research and education effort to help farm managers understand and explore risk management decisions and evaluate the financial consequences of those decisions with the presence of risk. ag survivor uses real world farm/ranch settings, probabilities, and impacts to depict the financial effects of risk. participants must make management decisions for the operation through several decision making periods. after each period, random outcomes are generated based on probability and the management teams receive updated prices, yield information, inventories, etc. risk management decisions included buying and storing hay for the winter, buying crop insurance, forward pricing, and hiring a shepherd to guard sheep from predators. a brief description of each program is provided below. all three scenarios are set up over two years of decisions, with each year having four periods and each period involving one to four decisions. the lazy u ranch is a hypothetical sheep operation in the mountains of colorado, which runs 1500 head of ewes in a range flock operation utilizing a mixture of public and private land. each year, participants may choose one of three options for marketing their lamb crop: forward contract weaned lambs to a feedlot for october delivery, sell weaned lambs to a feedlot on the october cash market, or retain ownership of the weaned lambs through the feedlot and sell them on the february cash market for slaughter. the king family ranch is a hypothetical cow/calf/hay operation in the mountains of published by atlantis press copyright: the authors 25 hoag,
keske
and
goldbach
 
 
 colorado. they raise 350 acres of hay each year and use much of their production as winter feed for their 500 cows. they typically calve in march-april and sell weaned calves in october. the wheatfields is a hypothetical wheat/fallow farming operation in eastern colorado. they farm 6,000 acres of farmland, typically harvesting 3,000 acres of winter wheat annually. they also have 100 head of cows that typically calve in march-april with weaned calves sold in october. 4. results 4.1. risk preferences three questions in the self-assessment survey addressed risk preferences. two of three questions, 1 and 5, showed no statistical difference between men and women. question 1 asked about the riskiness of the respondent's investment portfolio. both male and female agricultural producers report their investment portfolio to contain very little or little risk. self assessment question 5 asked individuals to select between four investment choices that presented different risk levels. on the other hand, self-assessment question 4 asked individuals to list their financial investment portfolio by levels of risky assets. when the averages of males and females were compared, the p-value was significant using a one-tailed test but not significant using a twotailed test. the two-tailed test would be significant at 0.10. when the male and female averages were compared in the ag survivor survey, using a two sample t-test assuming unequal variances, the p-value was significant and showed women are slightly more risk averse than men. considering all results from both surveys, women show a slightly higher aversion to risk than men. 4.2. confidence there was no statistically significant difference in the questions directly aimed at differences in male and female confidence. however, when asked about their capability of taking over the business if a partner became unable to, men scored significantly higher. therefore, males stated that they feel more capable about taking over the business on their own. it is believed that an individual’s confidence will affect their perceived personal capabilities; specifically, the higher an individuals’ confidence the more capable their perception of oneself. the literature indicates that males are more confident than females, which would lead to males perceiving themselves as more capable than females. our combined results indicate that men show some signs of more confidence, but less so than implied in studies of other types of businesses. 4.3 influence we also analyzed whether there was a gender difference in perceived influence over the business decision making process. the influence score for males was 2.76, while the influence score for females was 2.72, which presented no significant difference. this contradicts previous literature;28,29 their findings indicate that males will exert more influence than females. 4.4. motivation we examined a total of 12 potential places where we might find different motivations between men and women: negative motivations: receiving a poor price, producing a low yield, not having enough money for retirement, having to add an additional employee to the operation, facing a legal action against you, you or your business partner getting sick/injured and not able to work. positive motivations: growth of business, savings for children’s future, sense of accomplishment or recognition, cushion to deal with illness or injury, to make ends meet, independence only three variables showed significant differences. males were more motivated to avoid the risk of a poor price and poor yield than females. therefore, men place more value on avoiding low profits than women. female operated farms have been found to have a lower net income than male farms. if females are less motivated to reduce the risk of a poor price or yield, instead valuing other motivators as greater influences, it published by atlantis press copyright: the authors 26 
 risk
individuality
 
 could help explain this difference in net income. the other significant variable was that men are more concerned saving for their children's future. 5. discussion there appears to be some differences in risk preferences between men and women, although they do not appear very strong. some questions did not reveal any statistical difference between risk preferences, but differences were measurable in one question on the self assessment survey and in behavior as men and women played the ag survivor scenarios. men were more confident than women in only one way, their perceived capability of taking over the business on one’s own. in terms of motivation, men were found to be statistically different from women in three cases. men were much more concerned about poor prices or yields, which are highly related to profits, than women. finally, there appears to be no difference in influence when it comes to agricultural producers. our purpose was to take a close look at whether female and male agricultural producers in the united states had significant differences in the way they perceive and react to risk. based on the body of literature from other studies, we should have found bigger differences. we can only speculate, but perhaps agricultural women have fewer differences because they, like most men, grew up on a farm or ranch and are therefore relatively comfortable with how the property is managed. this shows, circumstantially, that context matters and that results will therefore be very different from one location to the next. returning to our example of kenyan pastoralists, women have an extremely less say in management issues than american women,30 and face different risks if catastrophe strikes. about 40 percent of smallholder households are managed by women and women are the primary managers of small livestock such as poultry and goats. despite their role in managing livestock, they hold title to less than 1 percent of the land and are often purposefully excluded from making management decisions.30 female-headed households have lower income and are more vulnerable than other households. therefore, it is imperative that the unique risk needs of these women be individually investigated and addressed to assure that their welfare is equally addressed in any crisis response plan. 6. references 1. center for disaster preparedness. integrating gender into community based disaster risk management. (community based disaster risk management: training and learning circlephilippines, 2009). 2. hyogo framework for action 2005-2015: building the resilience of nations and communities to disasters. (world conference on disaster reduction, kobe, japan, 18-22 january 2005) 3. l. karakowsky and a. elangovan. risky decision making in mixed-gender teams: whose risk tolerance matters? small group research 32(1) (2001) 94-111 4. gysler, m., j. kruse, and r. schubert. ambiguity and gender differences in financial decision making: an experimental examination of competence and confidence effects. center for economic research, swiss federal institute of technology, working paper 02-23. (http://ideas.repec.org/p/eth/wpswif/0223.html#provider,2002). 5. j. watson and s. robinson. adjusting for risk in comparing the performances of maleand female-controlled smes. journal of business venturing 18(6) (2003) 773-788 6. j. pennings, b. wansink and m. meulenberg. a note on modeling consumer reactions to a crisis: the case of the mad cow disease. intern. j. of research in marketing 19(1) (2002) 91–100. 7. u.s. census. national agricultural statistics service, u.s. department of agriculture. 2007 8. r. goldbach. gender influence of agricultural performance. masters thesis, department of agricultural and resource economics, colorado state university, fort collins, (2008). 9. d. hoag. applied risk management in agriculture. crc press, taylor & francis group, boca raton, florida, usa (2010) 403 p. published by atlantis press copyright: the authors 27 hoag,
keske
and
goldbach
 
 
 10. j. cohoon and w. aspray. the state of social science research on gender and it entrepreneurship. national center for women and information technology entrepreneurial report series 5, (2006) 11. s. weiler and a. bernasek. dodging the glass ceiling? networks and the new wave of women entrepreneurs. the social science journal 38(1) (2001) 85-103 12. us census. national agricultural statistics service, u.s. department of agriculture. 1997 13. us census. national agricultural statistics service, u.s. department of agriculture. 2002 14. national agricultural statistics service. women in ag quick factos (2002 census of agriculture (ref. 13)) 15. g. johnson, r. mcmahon. owner-manager gender, financial performance and business growth amongst smes from australia’s business longitudinal survey. international small business journal 23(2) (2005) 115-142 16. n. jianakoplos, and a. bernasek, are women more risk averse? economic inquiry 36(4) (1998) 620-630. 17. l. cunningham, brorsen, b., anderson, k., and tostao, e. (2007). gender differences in marketing styles, agricultural economics 38(1) 1-7. 18. m. powell and d. ansic. gender differences in risk behavior in financial decision making: an experimental analysis. journal of economic psychology 18(6) (1997) 605-628 19. c. camerer and lovallo, d. overconfidence and excess entry: an experimental approach. the american economic review 89(1) (1997) 306-318 20. b. barber and odean, t. 2001. boys will be boys: gender, overconfidence, and common stock investment. the quarterly journal of economics 116 (1) (2001): 261-292 21. c. bengtsson, persson, m., and willenhag p. 2004. gender and overconfidence. journal of economic letters. 86(2) 199-203. 22. j. eales, engel, b., hauser, r., thompson, s. grain price expectations of illinois farmers and grain merchandisers. american j. of agricultural economics 72(3) (1990) 701-708 23. h. buttner and d. moore. women’s organizational exodus to entrepreneurship: self-reported motivations and correlates with success. journal of small business and management 35(1) (1997) 34-46. 24. j. harwood, r. heifner, k. coble, j. perry, and a. somwaru. managing risk in farming: concepts, research, and analysis. u.s. department of agriculture, agricultural economics report # 774 (washington dc. 1999) 25. j.e. cliff. does one size fit all? exploring the relationship between attitudes towards growth, gender, and business size. journal of business venturing. 13(6)(1998) 523-542. 26. g. powell and k. eddleston, the paradox of the contented female business owner. journal of vocational behavior 73(1) (2008) 24-36 27. e. doty, tomkiewicz, j. bass, k.. sex differences in motivational traits and ethical decision making among graduating accounting majors. college student journal 39(4) (2005) 817-826 28. p. hawley, t. little, and n. card. the myth of the alpha male: a new look at dominancerelated beliefs and behaviors among adolescent males and females. international journal of behavioral development 32(1) (2008) 76-88. 29. s. colarelli, j. spranger, and r. hechanova. women, power, sex composition in small groups: an evolutionary perspective. journal of organizational behavior 27(2) (2006) 163184 30. united states government: feed the future initiative. kenya fy 2010 implementation plan. www.feedthefutue.gov (accessed january 22, 2011). published by atlantis press copyright: the authors 28 international journal of computational intelligence systems vol. xx(z); month (year), pp. xx–yy doi: 10.1080/xxxxxxxxxxxxxx; issn xxxx–xxxx online https://www.atlantis-press.com/journals/ijndc journal of risk analysis and crisis response vol. 10(1); april (2020), pp. 6–11 doi: https://doi.org/10.2991/jracr.k.200421.003; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr research article kernel density estimation of white noise for non-diversifiable risk in decision making emma anyika shileche1,*, patrick weke2, thomas achia3 1school of computing and mathematics, the co-operative university of kenya, nairobi, kenya 2school of mathematics, university of nairobi, nairobi, kenya 3school of public health, university of witwatersrand, johannesburg, south africa a r t i c l e i n f o article history received 20 october 2019 accepted 14 december 2019 keywords curve estimation non-diversifiable risk random variable probability estimates a b s t r a c t many businesses make profit yearly and tend to invest some of the profit so that they can cushion their organizations against any future unknown events that can affect their current profit making. since future happenings in businesses cannot be predicted accurately, estimates are made using experience or past data which are not exact. the probability element (which is normally determined by experience or past data) is important in investment decision making process since it helps address the problem of uncertainty. many of the investment decision making methods have incorporated the expectation and risk of an event in making investment decisions. most of those that use risk account for diversifiable risk (non-systematic risk) only thus limiting the predictability element of these investment methods since total risk are not properly accounted for. a few of these methods include the certainty (probability) element. these include value at risk method which uses covariance matrices as total risk and the binning system which always assumes normal distribution and thus does not take care of discrete cases. moreover comparison among various entities lacks since the probabilities derived are for individual entities and are just quantile values. finite investment decision making using real market risk (non-diversifiable risk) was undertaken in this study. non-diversifiable risk (systematic risk) estimates of a portfolio of stocks determined by a real risk weighted pricing model are used as initial data. the variance of non-diversifiable risk is estimated as a random variable referred to as random error (white noise). the estimator is used to calculate estimates of white noise (wn). a curve estimation of the wn is made using kernel density estimation (kde). kde is a non-parametric way to estimate the probability density function of a random variable. kde is a fundamental data smoothing problem where inferences about the population are made, based on a finite data sample. this is used to derive probability estimates of the non-diversifiable risks of the various stocks. this enables determination of total risk with given probabilities of its occurrence thus facilitating decision making under risky and uncertain situations as well as accentuating comparison among the portfolio of stocks. © 2020 the authors. published by atlantis press sarl. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: mmnk55378@gmail.com 1. introduction in the past few years there has been evidence of collapse of well established business entities. this has been attributed to lack of accurate methods of preventing or measuring risk and uncertainty as opposed to lack of the same methods. many companies on wall street in 2008 went under despite having extensive measures of mitigating risk such as futures and forward. an investigation into some of these methods reveals the lack of a well-estimated market risk measure in the models. it is an obvious fact now that it was the external reactions that brought down the companies on wall street. once the markets got a hint of the internal financial and investment affairs of the companies this spread so rapidly and in a matter of hours these companies had collapsed. a good example is the lehman brothers holdings limited, merrill lynch and companies, and american investment group as explained by lucchetti et al. [1]. these indicate that market environments are so critical in the existence of business entities such that variables affecting the business entities from the market environments should be estimated with a lot of precision. this paper determines total risk which has both the systematic and non-systematic components. it should be noted that the non-systematic risk is internal in nature, and in most cases well known and relatively less difficult to estimate while systematic is external and in most cases is embodied in market risk. this paper will determine the risk factor of systematic risk a phenomenon lacking in many risk models. since we have seen from most examples that market risk is the precursor of most companies down falls, it is hoped that investors and companies will be able to easily estimate riskiness of the risk measures thus enabling them make informed decisions. jorion [2] determines the value at risk (var) measure as the forecasted volatility, st multiplied by standard normal deviate, a for the selected confidence level (e.g. a = 2.33 for a one-tailed confidence level of 99%). the portfolio variance then becomes s w wt t t t 2 = ¢s where σt is the forecasted covariance matrix for the market risk factors as of the close day t. hence we have, vart = a st. although this research takes care of all the other shortcomings of previous researches, the portfolio variance is determined as a covariance which goes against the definition of market risk as https://doi.org/10.2991/jracr.k.200421.003 https://www.atlantis-press.com/journals/jracr http://creativecommons.org/licenses/by-nc/4.0/ mailto:mmnk55378%40gmail.com?subject= e.a. shileche et al. / journal of risk analysis and crisis response 10(1) 6–11 7 that which cannot be diversified. this has been clearly addressed in this paper by using non-diversifiable risk which is determined without covariance. dennis et al. [3] in estimating density dependence process noise and observation error offers a statistical approach for jointly estimating density dependence, process error and observation error. although this model is relatively easy for ecologists to use and is applicable in many population systems, this process noise has a normal distribution with mean m and variance s 2 (et ~ n(0, s 2)). this paper looks at a case of no assumption of normality for the noise process. white noise is determined as a random variable on the precincts of sklar [4] where he says no common probability space can be found for a given set of random variables, but such common probability spaces exists for arbitrary proper subsets of the given set. in this study the subsets were the portfolios of different companies used giving a common probability space that is estimated. the results of wu [5] show that for finite parameters the consistency of the least squares estimator is equivalent to the existence of a consistent estimator thus the estimator of white noise derived in this paper is an unbiased estimator. 2. risk of non-diversifiable risk 2.1. determination of white noise of non-diversifiable risk white noise refers to a purely random process whose random variables are a sequence of mutually independent, identically distributed random variables. thus it describes an event and is a function with a domain that makes some real number correspond to each outcome of the experiment. in this paper white noise is taken as the random error of non-diversifiable risk ngwi of an investment i. to add credence to this study it is imperative to show that white noise is a random variable. proposition 2.1 below seeks to do so. proposition 2.1: let vi be the white noise of the non-diversifiable risk ngwi, then vi(.) is a random variable. proof: given ngwi and i. the domain ω v vi n j j i 1£ £ -¥£ £¥ ¹ { } ì í î ü ý þ , ii. the counter domain r is such that 0 ≤ r ≤ 1. iii. the range of returns i is −∞ ≤ i ≤ ∞. then vi(.) is an event. that is vi(.) is such that the subset wr = {s: vi(s) ≤ r}, where s is a subset of the domain. this is true since 0 ≤ vi(s) ≤ 1. if wr belongs to w for every real number r, where w is the set of all outcomes of event vi(.). then the probability of vi(.), p[vi(.)] is a set function having domain vi(.) and counter domain the interval [0,1]. therefore vi(.) has a probability space (ω, w, p[vi(.)]). also w consists of four subsets; f, ,v vi n j j i 1£ £ -¥£ £¥ ¹ { } ì í ï îï ü ý ï þï and ω. such that if: i. r < 0, then s:vi(s) ≤ r = ϕ. ii. 0 ≤ r < 1, then s:vi(s) ≤ r, where v j i j -¥£ £¥ ¹ . iii. r ≥ 1, then {s:vi(s) ≤ r} = ω = v vi n j i j 1£ £ -¥£ £¥ ¹ , . since vi(.) has a probability space, and w consists of the four subsets above. then for each r the set {s:vi(s) ≤ r} belongs to w, thus vi(.) is a random variable. since vi(.) is a random variable and it is independent with unique parameters. therefore the parameters of white noise for example its mean and variance as well as its unique probability distribution can be determined. the probability estimates of non-diversifiable risk for investment decisions are then estimated as shown in the following subsections. 2.1.1. determination of random error the non-diversifiable variance estimator n w s s gwi i i i n e i n i 2 2 2 1 2 1 = + = = å å (1) derived from the non-diversifiable risk estimated in anyika et al. [6] indicates the presence of random error in the risk estimator. this error is taken to be white noise (wn) thus it can be said to be a random variable v1,v2,v3, ...,v∞ which is mutually independent and identically distributed. this is estimated from a sample of data by first varying the variance of individual return values of ri resulting in wn t s si r gw i n i  = = å 2 2 1 (2) where t z z = -( ) 2 1 2 , z being the total number of returns and (2) is the predicted random error. from (2) the actual value of wni is given by wn w s c li i ri i n = + = å 2 2 1 (3) where c and l are values representing the location (mean) and scale (variance) parameters. these parameters are determined such that the bias and variance of the actual and predicted values of wn are minimized as follows; let the variance between actual and sample white noise be var ( , )wn wn z wn z wn wn z wni i i i i i  = æ è ç ö ø ÷ + 2 1 2 1 2 1 2 2 2 (4) the values of c and l which will minimize variance are given by the partial derivatives of c and l, fc and fl respectively. after several iterations; f wn z wn lc i i= =2 2 1  f z wn wn ll i i= = 2 1  8 e.a. shileche et al. / journal of risk analysis and crisis response 10(1) 6–11 thus the value of wn z wni i= 4 3 1( )  (5) proposition 2.2: wni is an unbiased estimator of wni. proof: from equation (2) wn t s s e wn z z e v v z i r gw i z i i r i z i � � 2 2 1 2 2 2 1 2 1 2 ( ) ( ) (zz z e v e v z z wn z v i z i r i i 1 1 1 2 1 1 2 1 2 2 2 2 ) ( ) ( )� 22 11 2 22 1 vv z z zu zwn i j z i z i z i ( ) � 11 1 2 1 2 2 2 2 2 z zu zwn z z z z zu zwn u wn z i i i ( ) ( ) ( ) � � � (( ) ( ) ( ) ( ) z z z zu zwn u wn z z z z i i 1 2 1 1 2 1 2 2 2 2 2 2 � � zzu zwn u wn z z z z wn z z wn i i i 2 2 2 2 2 2 1 1 2 1 � � � � ( ) ( ) ( ) ii (6) where wni and m are the actual variance and mean of non-diversifiable risk respectively. dividing equation (6) by z results in = æ è ç ö ø ÷ = æ è ç ö ø ÷ z z z wn z z z z wn z i i 2 1 1 2 1 1   lim as z → ∞ = ¥ æ è ç ö ø ÷ ¥ æ è ç ö ø ÷ = = 1 2 1 1 1 0 1 0 2 wn wn wn i i i    ( ) ( ) thus wni is an unbiased estimator of wni. from the results of wu [7], equation (6) and proposition 2.2, wni is a consistent estimator of wni. 3. derivation of a probability density function for random error of non-diversifiable risk 3.1. kernel density estimation of white noise of non-diversifiable risk let v1,v2, ...,vn denote a sample of size n from the random variable vi(.) with density f. the kernel density estimates of f at the point v is given by f v nh k v v hn i i n  ( ) ( ) = -æ è ç ö ø ÷ = å 1 1 (7) where the kernel k satisfies k v dv( ) = -¥ ¥ ò 1 and the smoothing parameter h is known as the bandwidth. vi is the mean of v. proposition 3.1: vi i n = å 1 is a minimum sufficient statistics of vi _ . proof: given the function f hn v v hcn i i nl = é ë ê ê ù û ú ú = å 1 1 1 ( ) , the likelihood (l) of v is l v v nh v v hc l v v v in nh v i i i n i i i i( : ) ( : ) = -( )é ë ê ê ù û ú ú = ¶ ¶ å å = 1 1 1 1 1 vv hc i i n i n = = å å é ë ê ê ê ê ù û ú ú ú ú 1 1 (8) since, v vi iå = , then v vi i= 0 let v v v vn. ( , ,..., ) . . .= 1 2 be a point in v v v v vn: ( , ,..., )= 1 2 then l v v l v v vi in hn v i in hn i i i n i n ( : ) ( . : . ) ( ) . ( ) = ¶ ¶ ¶ ¶ = = å å 1 1 1 1 1 1 and therefore, l v v l v v i i ( : ) ( . : . ) = 1 meaning that it is independent of vi and thus, vi i n = å 1 is a minimum sufficient statistics of vi _ . lehmann and scheffé [8] remarks that if the sample space is discrete or a finite dimensional euclidean space then a minimal sufficient statistic will always exists. since a minimum sufficient statistic exists then the sample space is discrete and the probability density function exists. e.a. shileche et al. / journal of risk analysis and crisis response 10(1) 6–11 9 it is generally known that the value of the bandwidth is of critical importance while the shape of the kernel function has little practical impact. thus we estimate bandwidth and use a given kernel function to get the density estimation of the white noise of non diversifiable risk. 3.2. bandwidth selection for kernel density estimation of the wn of non-diversifiable risk assuming that the underlying density is sufficiently smooth and that the kernel has fourth moment using the taylor series bias f v h k f v hh � ( ) ( ) ( ) ( ){ } = + 2 2 2 2 m ² (9) var f v nh r k f v nh ( ) ( ) ( ){ } = +1 0 1 (10) where r k k v dv( ) ( )= ò 2 [9]. adding the leading variance and squared bias terms produces the asymptotic mean squared bias squared error (amse) amse ={ } = + éë ùûf v nh r k f v h k f vn � ( ) ( ) ( ) ( ) ( ) 1 4 4 2 2 2m ² (11) the overall measure of the discrepancy between f l and f is the mean integrated squared error (mise) which is given by mise bias var f e f y f y dy f v dv f v n h h h  ( ) = -( ){ } = + ò ò ò ( ) ( ) ( ( )) ( ( ) l l 2 ))dv (12) under an integrability assumption on f, integrating the expression for amse gives the expression for the asymptotic mise (amise), i.e., amise f nh r k h k r fh læ è ç ç ö ø ÷ ÷ = + 1 4 4 2 2( ) ( ) ( )m ² (13) where r f f v dv( ) [ ( ) ] .² ²= ò 2 the value of the bandwidth that minimizes the amise is given by h r k k r f namise = é ë ê ê ù û ú ú -( ) ( ) ( )m2 2 1 5 1 5 ² (14) using the rule of thumb method a global bandwidth h is based on replacing r(f ″) the unknown part of hamise, by its value for a parametric family expressed as a multiple of a scale parameter, which is then estimated from the data. the method dates back to deheuvels [10] and scott [11]. it has been popularized for kernel estimates by silverman [12]. the plug-in method is used to estimate hamise in this study. here the unknown quantity r(f ″) in the expression for hamise is replaced by an estimate. the “solve the equation” plug-in approach developed by sheather and jones [13] is based on deriving, the pilot bandwidth for the estimate r(f ″), as a function of h, namely g h c k r f r f h( ) ( ) ( ) ( ) = é ë ê ù û ú ² ¢² 1 7 5 7 (15) the unknown functions of f are estimated using kernel density estimates with bandwidth based on normal rules of thumb resulting in h r k k r f g ns j h = ( ) é ë ê ê ê ù û ú ú ú -( ) ( ) ( )( )m2 2 1 5 1 5 �² (16) where hsj is known as the sheather–jones plug-in bandwidth. under smoothness assumption on the underlying density, n5/14 (hsj/hamise−1) has an asymptotic n(0, ssj 2) distribution. thus, the sheather–jones plug-in bandwidth has a relative convergence rate of order n−5/14, which is much higher than that of biased cross validation. the triangle kernel is used for smoothing this is given by k t t c t c t ctri ( ) / , / , / = £ > ì í î 1 1 0 1 (17) where c is the constant used to scale the resulting kernel so that the upper and lower quartiles occur at ±0.25. substituting the kernel in equation (16) and the unknowns, h and n into the density function (7) gives the function f hn v v hcn i i nl = é ë ê ê ù û ú ú = å 1 1 1 ( ) (18) this function is used to generate probabilities of non-diversifiable risks of given portfolio thus ensuring that actual systematic risk is determined. 4. results 4.1. calculating actual non-diversifiable risk the sample white noise is estimated by varying the variance of individual return values ri as given by equation (2). the non diversifiable risk estimates in table 1 are substituted into equation (3) to give equation (5). sample wn estimates are then substituted into equation (5) to determine the actual white noise. 4.2. density estimates of actual white noise r statistical software is used to calculate sheather–jones (sj) bandwidth and hence the density estimates of actual wn as plotted in figure 1. 10 e.a. shileche et al. / journal of risk analysis and crisis response 10(1) 6–11 a summary of statistics resulting from sj density estimation in table 2 enables us apportion densities of the different quartile ranges. f-values are calculated as follows: f v vi v = s where vi = white noise of portfolio i, v = mean of white noise of all the portfolios, and sv = variance of all the portfolios. the probability density estimate of a portfolio i is determined by comparing the f-values with the apportioned densities of the different quartile figure 1 | a plot of the density estimates of actual white noise. call: density (x = x, bw = 0.7559, x lim = c(−2, 2)). data: x (20 obs.); bandwidth “bw” = 0.7559. table 1 | the non-diversifiable risks of 20 stocks from nyse company rn yh 29.33 tif 63.5 tm 538.1 hm 28.23 ponard 27.97 vic 0.561 dawt 25.58 bp 105.2 suntb 108.3 pnc 7.876 aig 7164 ford 1898 amr 25.17 bph 1.752 ctl 5.547 pfe 46.58 rti 9.054 gsk 35.81 table 2 | final results of white noise and kernel density estimation of portfolios of stocks company wn f probabilities actual sn yh 0.000729 0.827722 0.63566 18.64391 tif 0.00027 −0.27869 0.491 31.1785 tm 0.00011 −0.66196 0.4414 237.5173 hm 0.000128 −0.62098 0.4467 12.61034 ponard 0.001551 2.809139 0.979 27.38263 vic 0.00046 0.179302 0.5511 0.309222 dawt 0.001456 2.580143 0.9388 24.0145 bp 0.000113 −0.65714 0.442 46.4984 suntb 0.00011 −0.66437 0.441 47.7603 pnc 0.000142 −0.58723 0.4511 3.552864 aig 0.000657 0.654167 0.613 4390.919 ford 0.000308 −0.18709 0.5033 954.7601 amr 0.000491 0.254027 0.521 13.11357 bph 0.000227 −0.38234 0.4778 0.837106 ctl 0.0000884 −0.71655 0.4342 2.408507 pfe 0.0000988 −0.69146 0.4375 20.37875 rti 0.000238 −0.35582 0.4813 4.35769 gsk 0.0000872 −0.71933 0.4339 15.53796 bce 0.00022 −0.39921 0.4756 99.44796 stgi 0.000227 −0.38234 0.4778 7.601798 ranges and the maximum value. an f-score of positive 0.827722 has its density calculated as follows: 0.527674 + (1 – 0.827722)0.130402 = 0.63556 where: 0.527674 is the maximum value, 0.130402 is the value apportioned to the first quartile and 1 − 0.827722 represents the fraction occupied in the first quartile. final results of the survey are tabulated in table 3. 4.3. wilcoxon signed rank test wilcoxon signed rank test of hypothesis is used to compare the var method of determining risk and kernel white noise method. here we test the hypothesis that risks obtained by kernel white noise are a reflection of actual risks than those obtained by var. ho: the population difference are centered at 0. ha: the population differences are centered at a value <0. based on a significance level of a = 0.01, the proper test is to reject ho if z < −za. determining z and za 1 0 0 1 æ è çç ö ø ÷÷ table 3 | a summary of the results of a kernel density estimation of a portfolio white noise x y min.:−2.2676128 min.: 0.005892 1st qu.: 1.133396 1st qu.: 0.041992 median: 0.00081 median: 0.170879 mean: 0.0008191 mean: 0.219614 3rd qu.: 1.135035 3rd qu.: 0.397272 max.: 2.269251 max.: 0.527674 e.a. shileche et al. / journal of risk analysis and crisis response 10(1) 6–11 11 z t t t = = -åtest statistic m s where mt n n = +( )1 4 , st n n n = + +( )( )1 2 1 24 , and t n a n d= + -( )å 2 2 1( ) . using normal tables −za = −1.645, using the difference in risks and their ranks in table 4 (z = −3.88). since, z < za we reject the null hypothesis, so there is sufficient evidence to conclude that the kernel white noise risks are a reflection of actual risks as compared with those obtained by var. 5. conclusion an estimate of random error is made with the least bias and variance. probability estimates of the asset parameters are made thus boosting the level of surety. these are made in comparison i.e. looking at given portfolios one is able to make a decision among a variety of them. methods like var use generated variances to give probability estimates using extreme values. this lacks the comparability factor and assumes the central limit theory leading to application of normality conditions. they also use covariance parameters as market risks thus going against its definition. a case study of new york stock exchange (nyse) dow index in 2008 indicates that the portfolios with the highest actual non diversifiable risks were aig with 4390.919%, ford; 954.7661%, and tm; 237.5173%. these are corporates which experienced financial difficulties during the credit crunch in the usa in 2008. aig and tm had to be given some financial rescue packages to stay afloat until the financial crump was reversed. from the analysis of the results from the nyse case study of the dow index in 2008 it is clear that there is a relationship between the determined actual non-diversifiable and the actual market risk on the ground over the past 2 years. these research findings can aid investors make solid investment decisions as well as the different corporate cut ion themselves against any financial stress currently and in future. conflicts of interest the authors declare they have no conflicts of interest. authors’ contribution the original idea and subsequent analysis of the research was undertaken by eas with the guidance of pw and ta. acknowledgments i wish to acknowledge the university of nairobi staff in the school of mathematics as well as my family members for their encouragement and support during the period of undertaking this research. references [1] lucchetti a, ng s, craig s, mollenkamp c. lehman files for bankruptcy, merrill sold, aig seeks cash. wall str j 2008. [2] jorion p. value at risk: the new benchmark for managing financial risk. new york: mcgraw-hill; 2000. [3] dennis b, ponciano jm, lele sr, taper ml, staples df. estimating density dependence, process noise, and observation error. ecol monogr 2006;76:323–41. [4] sklar a. random variables, distribution functions, and copulas. a personal look backward and forward. j math stat 1996;28:1–14. [5] wu cf. characterizing the consistent directions of least squares estimates. ann stat 1980;8:789–801. [6] anyika e, waweru r, odhiambo r. non-diversifiable risk in investment portfolio as an aid to investment decision making. j sci technol 2005;59:13–27. [7] wu cf. asymptotic theory of nonlinear least squares estimation. ann stat 1981;9:501–13. [8] lehmann el, scheffé h. completeness, similar regions, and unbiased estimation. sankhyā indian j stat 1950;10:305–40. [9] wand mp, jones mc. kernel smoothing. london: chapman and hall; 1994. [10] deheuvels p. estimation non paramétrique de la densité par histogrammes généralisés. rev stat appl 1977;25:5–42. [11] scott dw. on optimal and data-based histograms. biometrika 1979;66:605–10. [12] silverman bw. density estimation for statistics and data analysis. london: chapman and hall; 1986. [13] sheather sj, jones mc. a reliable data-based bandwith selection method for kernel density estimation. j roy statist soc ser b 1991;53:683–90. table 4 | a table of wilcoxon signed rank test for large samples paired company actual rn(i) var at ` = 0.01 (ii) (i) – (ii) rank yahoo 18.64 0.4699 18.17 10 tiffany 31.18 0.2693 30.91 14 toyota 237.52 0.1790 237.34 18 hm 12.61 0.1688 12.44 7 ponardph 27.38 0.8349 26.55 13 vical inc 0.31 0.4629 −0.15 1 data watch 24.01 0.8398 23.17 12 bp 46.50 0.1391 46.36 15 suntrust 47.76 0.2145 47.55 16 pnc 3.55 0.1914 3.36 4 aig 4390.92 0.5951 4390.32 20 ford 954.76 0.3662 954.39 19 amr 13.11 0.4806 12.63 8 bph 0.84 0.2036 0.63 2 ctl 2.41 0.1978 2.211 3 pfe 20.38 0.1607 20.22 11 rti 4.358 0.3184 4.039 5 gsk 15.54 0.1297 15.41 9 bce 99.45 0.1775 99.27 17 sbg 7.602 0.3714 7.230 6 https://doi.org/10.1890/0012-9615(2006)76[323:eddpna]2.0.co;2 https://doi.org/10.1890/0012-9615(2006)76[323:eddpna]2.0.co;2 https://doi.org/10.1890/0012-9615(2006)76[323:eddpna]2.0.co;2 https://doi.org/10.1890/0012-9615%282006%2976%5b323:eddpna%5d2.0.co%3b2 https://doi.org/10.1890/0012-9615%282006%2976%5b323:eddpna%5d2.0.co%3b2 https://www.jstor.org/stable/25048038 https://www.jstor.org/stable/25048038 microsoft word internal exposure simulation based on exposure related dose estimating model internal exposure simulation based on exposure related dose estimating model* qian zhang1, deyin huang1, mao liu2, minyan li1 1tianjin bohai chemical industry group limited liability company institute of occupational health, tianjin 300051, china, huang_deyin@126.com 2 center for urban public safety research, nankai university, tianjin 300071, china abstract in this paper, occupational chemical poisons’ physiologically based pharmacokinetic (pbpk) models are established to simulate metabolic processes in vivo based on exposure simulation softwareexposure related dose estimating model (erdem). pbpk model establishing progress is described based on the development and utilization of erdem. as well known, benzene, 1, 3-butadiene and vinyl chloride are definitely carcinogenic to humans. the study on these carcinogens’ internal exposure has important significance to carcinogenic risk. internal exposure of these three kinds of chemicals are simulated with parameters of pbpk model, organizations and organism from literature data, which proves erdem’s broad applicability. simulation results are represented by different kinds of results forms such as toxicant concentration in blood changes over time, toxicant average concentration in blood, metabolite amount in target organ changes over time, toxicant concentration in exhaled air changes over time and so on. and the simulation results are verified by comparing with the experimental results in previous literature. the good consistency between simulation results and experimental date reflects that output results of erdem are reliable and can be characterized in various forms. keywords: physiologically based pharmacokinetic model, exposure related dose estimating model, internal exposure, occupational cancer risk. 基于暴露相关剂量估算模型的内暴露模拟研究 张倩 1 黄德寅 1 刘茂 2 李敏嫣 1 1.天津渤海化工集团有限责任公司劳动卫生研究所，天津 300051 2. 南开大学城市公共安全研究中心，天津 300071 摘要：在职业健康风险评价中，越来越多研究人员将内剂量代替外剂量进行风险计算。针对职业危害化学 毒物，本文以美国环保署推荐的内暴露模拟软件——暴露相关剂量估算模型(erdem)构建化学毒物的生理 药代动力学（pbpk）模型，模拟计算毒物经人体代谢后的内暴露数据。基于我们前期研究，本文阐述如 何利用 erdem 软件构建 pbpk 模型，并对苯、1,3-丁二烯、氯乙烯等多个人类明确致癌物进行内暴露研 究。内暴露结果以代谢后静脉血毒物浓度、呼出气毒物浓度、代谢脏器内毒物的质量等多种形式输出，并 与前人实验数据进行比较验证。研究表明，erdem 模型在进行内暴露模拟应用时，操作简便、适用广 泛、结果可靠。 关键词：生理药代动力学，暴露相关剂量估算模型 erdem，内暴露，职业致癌风险 1. 引言 近年来，随着我国工业的快速发展，工业生产 中有害因素对人体健康的影响正日益引起社会各界 的普遍关注。随着健康风险评价方法的不断发展， *基金项目：“十二五”国家科技支撑计划（2012baj24b04）；天 津市科技支撑计划重点项目（13zczdsy02300） 内剂量的概念被引入风险评价中。越来越多的研究 人员将生理药代动力学(pbpk)模型应用于化学物的 剂量—反应评价研究中，用内剂量来代替环境中的 暴露剂量进行风险分析，认为这种做法更具说服力 和准确性。 生理药代动力学( pbpk) 模型是建立在机体生 理、生化、解剖、毒物热力学性质基础之上的一种 整体模型，pbpk 模型将每个组织器官单独视为一 journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 175-184 published by atlantis press copyright: the authors 175 willieb typewritten text received 28 august 2013 willieb typewritten text accepted 12 december 2013 willieb typewritten text willieb typewritten text willieb typewritten text qian zhang et al. 个房室，房室间借助血液循环相互连接，且化学物 质在各个房室间的转化与转运遵循质量守恒的原 理。 pbpk 模型可以描述各组织或器官中化学物及 代谢物浓度随时间的变化，可以模拟肝脏等组织或 器官的代谢转化过程，可以提供化学物在体内的分 布和转化数据。此外，国外相关研究表明[1]pbpk 模 型能够很好的适用于不同物种间健康风险的外推计 算。目前国内针对 pbpk 模型的研究应用主要集中 在药物治疗、毒理学等领域，而在职业健康研究方 面涉及较少。 pbpk 模型计算化学物质内剂量时涉及到人体 内多个房室内物质代谢的微分方程，计算过程复 杂，因此研究人员通常利用编程的方法来实现 pbpk 模型的计算，计算程序一般采用 acsl 语言 编写[4]。 基于我们前期研究[3-13]，本文将对美国环保署 usepa 推荐的 pbpk 模型模拟软件——暴露相关剂 量估算模型(erdem)进行开发利用，选取苯、丁二 烯、氯乙烯等三种较常见的职业病危害因素，建立 其 pbpk 模型，进行内暴露模拟分析，并利用现有 动物、人体实验数据对模拟结果进行验证。 2. 利用 erdem 软件模拟内剂量 2.1 erdem软件简介 暴露相关剂量估算模型（exposure related dose estimating model, erdem）是由美国国家暴露研究 实验室（nerl）开发、以acsl语言为基础的pbpk 模拟系统。该软件可以通过选择与构建需要模拟的 房室，输入各种生理及代谢参数，来实现有毒物质 在体内的代谢模拟[7]。erdem中自带的数据库包含 了研究人员根据流行病学研究案例和毒理学实验数 据而构建、并已得到广泛认可的多种毒物及其代谢 产物的pbpk模型，因此使用者可以根据现有的模型 来计算内剂量。同时该软件还支持新的pbpk模型建 立，以此来解决特定问题。 如图1给出了erdem软件建立pbpk模型、模拟 预测内剂量的运行流程。主要分为以下几个步骤： ①新建模型—在相应的界面中设置待模拟化学 物质的名称及其摩尔质量，填写待建模型的名称、 待研究的暴露对象、使用单位等； ②pbpk模型组织及参数设置—选择pbpk模型 所需的房室、组织等并输入各组织参数，选择代谢 过程及其代谢速率、代谢常数等，若有酶反应过程 还可进行酶反应设置； ③暴露参数设置—输入暴露途径、时间、浓度 等； ④输出选项设置—erdem软件可以分析不同房 室中原化学物质及其代谢物情况，并且可以给出不 同的输出形式，如浓度、auc、总量等，因此在此 步骤中可根据所期望的结果形式设置输出选项； ⑤模拟运行、输出结果—erdem根据设置的输 出选项，可以给出随时间变化的内剂量值，还可以 给出内剂量随时间变化的曲线图。 2.2 pbpk模型建立 本文分别对苯、丁二烯、氯乙烯的内暴露进行 模拟，在工业生产中，工人接触这三种职业危害物 质的方式主要以蒸气的形态经呼吸道吸入，口服的 概率很小，而经皮肤吸收的量也很小，对致癌作用 并不突出，因此本文主要考虑这三种物质的暴露途 径是经呼吸摄入，研究此后毒物在体内的分布、吸 收、代谢过程，不考虑口服和皮肤接触等其他暴露 途径。 根据相关文献给出的苯、丁二烯、氯乙烯的 pbpk模型[16-19]，五房室模型是这三种物质运用较普 遍的pbpk模型，即气体交换室(肺) 、脂肪、充分灌 注室、不充分灌注室、肝脏，假设只在肝脏中发生 代谢过程，代谢速率用michaelis-menten方程表示。 如图2所示。 图1 erdem软件运行流程 新建模型 pbpk 模型 组织及参数设置 暴露参数 设置 输出选项 设置 模拟运行 输出结果 published by atlantis press copyright: the authors 176 internal exposure simulation 图2 苯、丁二烯、氯乙烯pbpk模型图 pbpk模型所需的生理参数主要为人体和大鼠的 体重、心输出量、肺通气量、各组织容积、血流量 及分配系数等，这些参数多是通过实验得出的[1619]。目前国外研究者已经发布了许多由实验获得的 较完善的参数，在实际工作中往往直接引用这些数 据。由于国内目前还没有针对中国人个体特征的相 关数据，因此本文参考了国外的成人数据。 3. 基于 erdem 模拟的苯内暴露研究 3.1 苯的职业健康影响 苯是工业生产中广泛使用的有机溶剂，在生产 环境中以蒸汽状态存在，主要通过呼吸系统进入机 体。长期接触苯或含苯溶剂，可引起血液系统以白 细胞减少为主的慢性毒作用，而且可致染色体损 伤、白血病、再生障碍性贫血等。苯被国际癌症研 究机构( iarc) 列为明确人类致癌物。苯在我国《职 业病危害因素分类目录》中被列为可能导致苯中 毒、苯所致白血病的职业病危害因素。 迄今，苯的毒作用机制仍未完全阐明，目前认 为主要涉及[20]：①干扰细胞因子对骨髓造血干细胞 的生长和分化的调节作用。骨髓基质是造血的微环 境，在调节正常造血功能上起关键作用。苯代谢物 以骨髓为靶部位，降低造血正调控因子白介素il-1 和il-2的水平；活化骨髓成熟白细胞，产生高水平 的造血负调控因子肿瘤坏死因子(tnf-α)。②氢醌 与纺锤体纤维蛋白共价结合，抑制细胞增殖。③ dna损伤，其机制有二，一是苯的活性代谢物与 dna共价结合；二是代谢产物氧化产生活性氧有 关，对dna造成氧化性损伤。通过上述两种机制引 发突变或染色体的损伤，引起再生障碍性贫血或因 骨髓增生不良， 终导致急性髓性白血病。④癌基 因的激活。肿瘤的发生往往并非单一癌基因的激 活，通常是两种或两种以上癌基因突变的协同作 用。苯致急性髓性白血病可能与ras、c-fos、c-myc等 癌基因的激活有关。 现阶段苯生物接触标志物主要有血苯、尿苯、 苯巯基尿酸(s-pma)、全反式黏糠酸(t，t-ma)等[2123]。目前，我国还未设置苯的生物接触限值。相关 研究表明[21-23]，血苯可直接发映体内的苯吸收情 况，血苯浓度与环境接触苯浓度具有良好的相关 性，可以更好地反应苯作业人员的实际接触水平， 能更好地评价剂量——反应关系。本文利用erdem 建立苯pbpk模型进行内暴露模拟，实现苯吸入后人 体血苯浓度的预测，为今后生物接触限值和风险评 价的进一步研究作铺垫。 3.2 基于erdem建立苯pbpk模型 根据上文所述，本文考虑苯以蒸汽形态经呼吸 道吸入的暴露方式，利用建立如图2结构的pbpk模 型，研究人体吸入后氯乙烯在体内的分布、吸收、 肺泡空间 肺血 脂肪室 充分灌注室 不充分灌注室 肝脏 动 脉 血 静 脉 血 活性代谢产物 吸入 呼出 代谢 published by atlantis press copyright: the authors 177 qian zhang et al. 代谢过程。根据相关研究者发布的经实验或种间推 导获得的参数[16-17]，表1给出了本文中苯pbpk模型 的所需参数，主要为人体体重、心输出量、肺通气 量、各组织容积、血流量及分配系数等。 3.3 苯内暴露模拟结果验证 可用来进行模型验证的人体实验数据非常少， sato等人[24]曾对暴露于浓度为25ppm苯环境后的成人 血苯浓度进行测定，本文以此内暴露数据进行pbpk 模型验证，验证结果如图3所示。 表1 苯的人体pbpk模型参数 参数 参数 人体参数 组织容积百分比/% 体重/kg 70 动脉血 6 心输出量/ l/h 336 脂肪 20 肺通气量/ l/h 450 肝脏 2.6 分配系数 充分灌注室 5 血/气 7.8 不充分灌注室 62 脂肪/血 19 静脉血 3 充分灌注室/血 1.92 静态肺 1.4 不充分灌注室/血 2.05 血流百分比/ % 肝脏/血 2.95 脂肪 8 代谢系数 肝脏 25 肝脏 大代谢速率/ mg/h 13.89 充分灌注室 38.5 肝脏 michaelis-menten 常数 /mg/l 0.35 不充分灌注室 28.5 资料来源： bois f y, jackson e t, pekari k, et al. population toxicokinetics of benzene. environmental health perspectives, 1996, 104 （s6）: 1405-1411. elizabeth a brown, michael l shelley, jeffrey w fisher. a pharmacokinetic study of occupational and environmental benzene exposure with regard to gender. risk analysis, 1998, 18( 2): 205-213. 图3 成人2h连续暴露于25ppm苯的静脉血中苯浓度的模拟结果与实验测量值 实验数据资料来源： sato a, nakajuma t, fugiwara y, et al. kinetic studies on sex difference in susceptibility to chronic benzene intoxication with special regard to body fat content. british journal of industrial medicine, 1975, 32（4）: 321-328. 为了验证利用erdem软件得到的pbpk模型模 拟结果的准确性，按照sato等人的暴露条件设置模 拟过程的暴露时间和浓度，对成人连续2h暴露于环 境浓度为25ppm苯的人体苯代谢pbpk模型进行模 拟，预测吸入苯后的人体血苯浓度变化，并将模拟 结果与sato等人的实验数据[24]进行对比。 由图3可看出，由此方法得到的血苯预测结果与 实验测量值基本一致，说明利用erdem软件可以较 0 0.05 0.1 0.15 0.2 0.25 0.3 0 1 2 3 4 5 6 7 8 时间/h 静 脉 血 中 苯 浓 度 / m g / l 模拟值 实验值 published by atlantis press copyright: the authors 178 为准确的预测人体接触苯后的内剂量数据。如图3， 在持续吸入苯的两小时过程中，人体内血苯浓度随 时间的增加而递增，且在1h时，递增速率有所降 低，直到2h时达到 大值。2h后，随着苯环境的脱 离，体内代谢的血苯浓度逐渐下降。 4. 基于 erdem 模拟的丁二烯内暴露研究 4.1 1,3-丁二烯的代谢机理及致癌机制 丁二烯( 1,3-butadiene,bd )作为一种重要的有机 化工原料，广泛应用于合成橡胶、合成树脂、合成 纤维、增塑剂和乳胶漆等等石油化工与制造行业， 主要以气体的形式被人体吸入。1,3-丁二烯现未列入 《职业病危害因素分类目录》及卫生部颁布的《高 毒物品目录》，但丁二烯已被国际癌症研究机构 ( iarc) 列为人类明确致癌物。值得关注的是现代流 行病学研究数据表明，职业接触丁二烯而患淋巴癌 和白血病的死亡率很高[25]。小鼠接触丁二烯引发淋 巴癌和因职业接触丁二烯而引发淋巴癌和造血系统 癌，二者部位的一致性特别值得注意。 虽然1,3-丁二烯已被国际公认为致癌物质，但至 今对其致癌机制尙不十分清楚。相关文献表明，bd 的致癌机制与其在体内代谢产生的代谢物有关。本 文主要考虑1,3-丁二烯的暴露途径是经呼吸摄入， 1,3-丁二烯进入生物体内，首先将代谢成活性环氧化 物——1,2-环氧丁烯，1,2-环氧丁烯将继续代谢生产 多种活性代谢物，这些代谢物可以和生物体内的分 子加和生产共价化合物， 终导致遗传物质改变， 引发肿瘤产生[26-28]，因此，1,3-丁二烯到1,2-环氧丁 烯这一代谢过程是后续代谢过程的首要步骤，本文 建立的丁二烯pbpk模型即考虑这一代谢步骤。 4.2 基于erdem建立1,3-丁二烯pbpk模型 本研究根据相关文献参数数据[18]，建立了小 鼠、人的1,3-丁二烯的pbpk模型，如上文图2所示， 模型包含了：肺、脂肪、肝脏、充分灌注室和非充 分灌注室。代谢发生在肝脏中，代谢反应考虑1,3-丁 二烯转变成1,2-环氧丁烯。s. fustinoni等人[29]指出当 bd暴露浓度较低时，可以采用bd在人体内代谢后 血液中的残留浓度作为生物标志物，本文即利用 pbpk模型计算分析1,3-丁二烯经代谢后在静脉血中 的浓度。具体参数如表2，pbpk模型如图2。 表.2 1,3-丁二烯的pbpk模型参数 参数 小鼠 人 生理参数 体重(kg) 0.028 70 心输出量(l/h) 1.044 660b 换气速率(l/h) 2.64 1,200b 血容积比例 0.05 0.077 脂肪容积比例 0.04 0.144 肝脏容积比例 0.062 0.025 内脏容积比例 0.05 0.037 肌肉容积比例 0.78 0.547 脂肪血流比例 0.05 0.036 肝脏血流比例 0.16 0.16 内脏血流比例 0.52 0.446 肌肉血流比例 0.19 0.361 分配系数 空气分配系数 1.5 脂肪分配系数 118.2 肝脏分配系数 5.49 内脏分配系数 5.34 肌肉分配系数 5.26 生化参数 肝脏 v (nmol/h/mg) 155.4 70.8 肝脏 km (mm) 0.002 0.00514 资料来源： u.s. environmental protection agency. health assessment of 1,3-butadiene (epa/600/p-98/001f). washington, dc, 2002 published by atlantis press copyright: the authors 179 qian zhang et al. 0 1 2 3 4 5 6 0 500 1000 1500 b d 在 血 液 中 的 内 剂 量 浓 度 / m g / l 外暴露浓度/ppm 实验值与erdem软件模拟值对比 实验值 模拟值 图4 1,3-丁二烯实验值与模拟值对比 实验数据资料来源： matthew w. himmelstein, max j. turner, bahman asgharian, et al.comparison of blood concentrations of 1,3-butadiene and butadiene epoxides in mice and rats exposed to 1,3-butadiene by inhalation. carcinogenesis, 1994, vol. 15(8): 1479~1486 4.3 1,3-丁二烯内暴露模拟结果验证 为了验证 pbpk 模型模拟结果，本文将通过 erdem软件模拟得到bd在体内代谢内剂量值与生 物实验数据进行对比。但未找到丁二烯人体内剂量 实验数据，本文仅以小鼠内剂量实验数据进行模拟 结果的验证。 实验验证数据采用himmelstein等人[30]对暴露于 1,3-丁二烯不同浓度环境下6h，小鼠6小时之间采集 到的血液样本浓度平均值；当采集血液样本时动物 继续吸入1,3-丁二烯。 本文利用erdem建立丁二烯的pbpk模型后， 按照实验条件进行内暴露模拟条件的设定，在丁二 烯外暴露浓度为71、603、1282ppm的条件下，分别 对小鼠吸入丁二烯后静脉血中丁二烯平均浓度进行 模拟，结果如表3示。 表.3 1,3-丁二烯的内暴露模拟和实验值对比 外暴露 ppm 6 小时内剂量实 验值 mg/l 6 小时内剂量模 拟值 mg/l 71 0.13 0.25 603 2.58 2.11 1282 4.97 4.89 实验数据资料来源： matthew w. himmelstein, max j. turner, bahman asgharian, et al.comparison of blood concentrations of 1,3butadiene and butadiene epoxides in mice and rats exposed to 1,3-butadiene by inhalation. carcinogenesis, 1994, vol. 15(8): 1479~1486 由结果对比图4可以看出，利用erdem软件模 拟的bd内剂量与实验值相吻合，具有一致性，因此 利用pbpk模型能够较准确地预测分析1,3-丁二烯在 体内代谢情况。 5. 基于 erdem 模拟的氯乙烯内暴露研究 5.1 氯乙烯的职业健康影响 氯乙烯，在常温、常压下为无色气体，略有芳 香气味。微溶于水，可溶于乙醇、丙酮、乙酸、二 氯乙烷等。是一种常用的工业生产原料，在生产和 使用过程中, 工作人员长期接触氯乙烯后可引起多种 职业伤害。研究表明氯乙烯在体内的毒性与其代谢 密切相关，其主要靶器官是肝脏，可导致长期接触 人员患肝血管肉瘤。氯乙烯在我国《职业病危害因 素分类目录》中被列为可能导致氯乙烯中毒、肝血 管肉瘤的职业病危害因素。 早期研究显示氯乙烯可被体内肝微粒体细胞色 素p450酶(cyp2e1)所代谢，产生亲核性活性代谢产 物。氯乙烯在啮齿动物的代谢主要通过饱和代谢过 程，产生氧化氯乙烯（ceo）和其重排产物氯乙醛 （caa）[31]。ceo和caa均能作用于dna、rna和 蛋白质，导致体内产生遗传毒性，致使人体肿瘤发 生。相关研究[32]表明由于氯乙烯具有多器官损害效 应，因此其效应标志物较多。但是，在目前研究水 平上，多数生物标志物缺乏特异性，而且在各国存 在较大的分歧。 5.2 基于erdem建立氯乙烯pbpk模型 根据上文所述，本文考虑氯乙烯以蒸汽形态经 呼吸道吸入的暴露方式，利用erdem软件建立如图 2结构的pbpk模型，研究大鼠和人体吸入后氯乙烯 published by atlantis press copyright: the authors 180 internal exposure simulation 在体内的分布、吸收、代谢过程。根据相关研究者 发布的经实验获得的参数，表4出了本文中氯乙烯 pbpk模型的所需参数。 表.4大鼠、人体的氯乙烯pbpk模型参数 参数 人 大鼠 参数 人 大鼠 一般参数 组织容积百分比/% 体重/kg 70.0 0.225 动脉血 6.0 6.0 心输出量/ l/h 363.0 5.97 脂肪 20.0 7.0 肺通气量/ l/h 363.0 5.97 肝脏 2.6 2.53 分配系数 充分灌注室 5.0 4.0 血/气 1.16 1.68 不充分灌注室 62.0 76.07 脂肪/血 20.0 20.0 静脉血 3.0 3.0 充分灌注室/血 1.6 1.6 静态肺 1.4 1.4 不充分灌注室/血 2.1 2.1 组织血流速率/(心输出量百分比) 肝脏/血 1.6 1.6 脂肪 5.0% 5.0% 代谢系数 肝脏 24.0% 24.0% 肝脏 大代谢速率/ mg/h 138.95 0.968 充分灌注室 52.0% 52.0% 肝脏 michaelis-menten 常 数/mg/l 0.04 0.04 非充分灌注室 19.0% 19.0% 资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. e. d., stewart, r. d., and mutchler, j. e. (1969). monitoring exposures to vinyl chloride vapor: breath analysis and continuous air sampling. am. ind. hyg. assoc. j. 30, 537–544. 5.3 大鼠经呼吸摄入氯乙烯的模拟结果及验证 本文利用watanabe等人[33]对大鼠代谢进行的实 验 研 究 ， 验 证 大 鼠 氯 乙 烯 吸 入 的 pbpk 模 型 。 watanabe等人将大鼠暴露于经放射性标记的一系列 浓度（1.4~4600ppm）氯乙烯环境中，持续暴露c14 氯乙烯6小时，并收集这些动物72小时内的放射性排 泄物。本文利用erdem软件对大鼠pbpk模型进行 模拟，分析氯乙烯在肝脏代谢后72小时内产生的代 谢物——氯乙烯氧化物总量（氧化氯乙烯和氯乙 醛），并进行不同浓度下代谢物总量的多次重复模 拟，拟合得到代谢物总量与外暴露浓度变化的曲 线。结果如表5、图5所示。 从图5可以看出，pbpk模型的模拟结果很好地 验证了watanabe等人[33]的实验数据。在1.4~5000ppm 的浓度范围内，本文的大鼠吸入氯乙烯的pbpk模型 能精确地预测氯乙烯在大鼠体内的代谢情况，并成 功地证实了氯乙烯代谢发生饱和的暴露浓度范围 （200-500ppm）。 表.5大鼠吸入氯乙烯后肝脏内代谢物浓度模拟值 外暴露浓度 /ppm 肝脏内代谢物 浓度/mg 外暴露浓度 /ppm 肝脏内代谢物 浓度/mg 1.4 0.04 400 2.86 5 0.15 500 2.93 10 0.30 600 3.00 20 0.60 700 3.05 30 0.89 800 3.10 40 1.17 900 3.14 50 1.42 1000 3.18 60 1.65 2000 3.46 70 1.85 3000 3.63 80 2.02 4000 3.76 90 2.15 5000 3.86 100 2.25 6000 3.95 200 2.63 7000 4.02 300 2.76 published by atlantis press copyright: the authors 181 qian zhang et al. 图5不同浓度下氯乙烯代谢物产生量的实验值与模拟值 图6 成人暴露于59、261 ppm 氯乙烯7.5 h后呼出气中氯乙烯浓度的模拟结果与实验测量值 实验数据资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 5.4 人体吸入氯乙烯的pbpk模型验证 baretta等人[34]对成人暴露于氯乙烯后的生理药 代动力学反应进行了实验研究，并实际测量了暴露 结束后成人呼出气中氯乙烯的浓度值。baretta等人 令实验对象暴露于实验控制下的氯乙烯暴露条件 下，分别暴露于59、261 ppm 氯乙烯7.5 h。工人首 先暴露3.5小时；然后他们离开到没有氯乙烯暴露的 区域吃午饭，大约为0.5hr； 后他们再回到氯乙烯 暴 露环 境 持续 4 小 时 （总 共暴 露 于氯 乙 烯 7.5 小 时）。 为了验证利用erdem软件得到的氯乙烯pbpk 模型模拟结果的准确性，本文对成人暴露于59、 261ppm氯乙烯7.5小时后呼出气氯乙烯浓度进行模 拟，在8.5小时到28小时之间重复多次模拟过程，形 成模拟值拟合曲线，结果如表6、图6所示。 表.6氯乙烯呼出浓度模拟值与实验值对比 时间 59ppm 261ppm 模拟值 /ppm 实验值 /ppm 模拟值 /ppm 实验值 /ppm 8.5 3.33 1.01 4.58 8.807 9.5 1.97 0.69 2.93 5.41 11.5 0.84 0.41 1.52 3.00 13.5 0.49 0.29 1.03 1.95 15.5 0.36 0.20 0.82 1.35 17.5 0.30 0.17 0.69 0.99 19.5 0.26 0.13 0.59 0.83 21.5 0.22 0.10 0.52 0.67 23.5 0.19 0.09 0.45 0.54 25.5 0.16 0.08 0.40 0.46 27.5 0.15 0.06 0.35 0.39 实验数据资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 如图6所示，当工人脱离氯乙烯暴露环境后，呼 出气中氯乙烯浓度随时间递增而降低，但达到28h 时，呼出气中仍含有少量氯乙烯。 将erdem软件模拟结果与baretta等人得到的实 验数据[34]进行对比，结果由表6、图6可看出，由本 published by atlantis press copyright: the authors 182 图6 成人暴露于59、261 ppm 氯乙烯7.5 h后呼出气中氯乙烯浓度的模拟结果与实验测量值 实验数据资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 5.4 人体吸入氯乙烯的pbpk模型验证 baretta等人[34]对成人暴露于氯乙烯后的生理药 代动力学反应进行了实验研究，并实际测量了暴露 结束后成人呼出气中氯乙烯的浓度值。baretta等人 令实验对象暴露于实验控制下的氯乙烯暴露条件 下，分别暴露于59、261 ppm 氯乙烯7.5 h。工人首 先暴露3.5小时；然后他们离开到没有氯乙烯暴露的 区域吃午饭，大约为0.5hr； 后他们再回到氯乙烯 暴 露环 境 持续 4 小 时 （总 共暴 露 于氯 乙 烯 7.5 小 时）。 为了验证利用erdem软件得到的氯乙烯pbpk 模型模拟结果的准确性，本文对成人暴露于59、 261ppm氯乙烯7.5小时后呼出气氯乙烯浓度进行模 拟，在8.5小时到28小时之间重复多次模拟过程，形 成模拟值拟合曲线，结果如表6、图6所示。 表.6氯乙烯呼出浓度模拟值与实验值对比 时间 59ppm 261ppm 实验数据资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 如图6所示，当工人脱离氯乙烯暴露环境后，呼 出气中氯乙烯浓度随时间递增而降低，但达到28h 时，呼出气中仍含有少量氯乙烯。 将erdem软件模拟结果与baretta等人得到的实 验数据[34]进行对比，结果由表6、图6可看出，由本 实验数据资料来源： watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 如图6所示，当工人脱离氯乙烯暴露环境后，呼 出气中氯乙烯浓度随时间递增而降低，但达到28h 时，呼出气中仍含有少量氯乙烯。 将erdem软件模拟结果与baretta等人得到的实 验数据[34]进行对比，结果由表6、图6可看出，由本 internal exposure simulation 文方法得到的结果与实验测量值基本一致，说明利 用erdem软件可以较为准确的预测工人接触氯乙烯 后的代谢数据。 6. 结论与讨论 本文利用 usepa 推荐的 pbpk 模型模拟软件 erdem计算人体内剂量值，用该软件模拟分析一定 暴露浓度下生物体内剂量。 具体阐述了对erdem的开发利用，建立了苯、 丁二烯、氯乙烯等多种不同人类明确致癌物的人 体、动物的pbpk模型，说明erdem在毒物种类、 不同生物体具有广泛的适用性、操作相对简便。 本文在进行化学毒物的内暴露模拟时，以血液 中毒物浓度的经时变化、血液中毒物一定时间内的 平均浓度、靶器官代谢物含量的经时变化、呼出气 毒物浓度的经时变化等多种输出结果形式，进行内 暴露表征。且模拟结果与前人文献中的实验结果进 行比较验证，均有较好的一致性，体现了erdem的 输出功能丰富、输出结果可靠。 本文采用的pbpk模型、各组织参数、人体参数 等均来自于文献数据，且由于国内相关数据较少， 数据大多来自于国外文献，存在种种不确定性，在 今后的研究中，还需进一步讨论利用本土化数据进 行研究，以期与实际情况更相符。 对致癌物进行内暴露研究对其健康风险评价有 着重要意义。在今后的研究中，还需进一步寻找化 学毒物的生物标志物，对生物标志物代谢进行模 拟，研究生物标志物与致癌效应之间的剂量反应关 系，从而进行职业健康风险评价。 参考文献 1. andersen, m. e., clewell, h. j., gargas, m. l., smith, f. a., and reitz, r. h.. physiologically-based pharmacokinetics and the risk assessment process for methylene chloride. toxicol. appl. pharmacol. 1987, 87: 185-205. 2. francis cp law, 何绍雄. 生理药代动力学模型的进展与 应用. 药学学报, 1997, 32(2): 151-160. francis cp law, s. x. he. research progress and application of physiologically based pharmacokinetic model. acta pharmaceutica sinica, 1997, 32(2): 151-160. 3. 姚美村, 姜晓飞, 陆亚松等. 生理药动学模型及其在中药 研究中的应用. 世界科学技术—中医药现代化, 2007, 9(3): 55-59. yao meicun, jiang xiaofei, lu yasong, et al. pbpk model and its application in tcm studies. world science and technology-modernization of traditional chinese medicine and materia medica, 2007, 9(3): 55-59. 4. 陆亚松, raymond s h yang. 以 1,1,1-三氯乙烷为例介绍 生理药代动力学模型的建立. 环境与职业医学, 2006, 23(4): 330-338. lu ya-song, raymond s. h. yang. introduction to physiologically based pharmacokinetic model construction: an example of 1,1,1-trichloroethane. journal of environmental & occupational medicine, 2006, 23(4): 330-338. 5. 施婕. 突发毒气泄漏事故的急性健康风险分析研究. 天 津: 南开大学, 2009. shi jie. analysis of acute health risk brought by accidental toxic gas leakage, master thesis, tianjin: nankai university , 2009. 6. 王阳, 刘茂. 基于生理毒代动力学模型和剂量—反应模 型的苯暴露健康风险评价方法. 中国工业医学杂志, 2009, 22(1): 34-37. wang yang, liu mao. study on health risk assessment for benzene-exposed workers by physiologically based toxicokinetic model and dose-response model. chinese journal of industrial medicine, 2009, 22(1): 34-37. 7. y. wang, m. liu, d.y. huang. health risk assessment for benzene occupational exposure using physiologically ba sed pharmacokinetic model and doseresponse model. 3rd international conference on bioinfor matics and biomedical engineering, icbbe 2009. 8. d.y. huang, j. zhang, m. liu. application of a health ris k classification method to assessing occupational hazard in china. 3rd international conference on bioinformatics a nd biomedical engineering, icbbe 2009. 9. d.y. huang, j zhang, m. liu. research on risk assessme nt based on monte carlo simulation and doseresponse multistage model. 2010 3rd international confer ence on biomedical engineering and informatics. 10. 黄德寅，张倩，刘茂，苯作业职业暴露评估与致癌风 险 模 拟 ， 中 国 工 业 医 学 杂 志 ， 2011 ， 24 （ 3 ） ： 163~168. 11. d.y. huang, q. zhang, m. liu. occupational exposure as sessment and carcinogenic risk simulation in benzene w orkers, chinese journal of industrial medicine. 2011, 24 (3 ):163~168. 12. d.y. huang, q. zhang, j. zhang.et al. modeling of occupa tional exposure to benzene and health risk of workers in styrene production. 2011 5th international conference on bioinformatics and biomedical engineering. 13. q. zhang, d.y. huang, m. liu. study on carcinogenic ris k evaluation for 1,3-butadieneexposed workers.2012 5th international conference on en vironmental pollution and public health (epph) 14. 张倩, 刘茂, 黄德寅. 苯暴露的致癌风险评价. 中国安全 科学学报, 2011, 21(5): 143-147. zhang qian, liu mao huang de-yin. study on carcinogenic risk evaluation of benzene-exposed workers. china safety science journal, 2011, 21(5): 143147. 15. 黄德寅, 张倩, 刘茂;.基于内暴露模拟的 1,3-丁二烯致癌 风险评价.风险分析和危机反应的创新理论和方法—— published by atlantis press copyright: the authors 183 qian zhang et al. 中国灾害防御协会风险分析专业委员会第五届年会论 文集.2012-10-27 d.y. huang, q. zhang, m. liu. study on carcinogenic risk evaluation based on internal dose simulation. innovative theories and methods for risk analysis and crisis response — — proceedings of the 5th annual meeting of risk analysis council of china association for disaster prevention. 2012-10-27 16. 张倩, 黄德寅, 刘茂;.氯乙烯代谢的生理药代动力学研究. 风险分析和危机反应的创新理论和方法——中国灾害 防御协会风险分析专业委员会第五届年会论文集.201210-27 q. zhang, d.y. huang, m. liu. study on physiologically based pharmacokinetic model of vinyl chloride metabolism. innovative theories and methods for risk analysis and crisis response—proceedings of the 5th annual meeting of risk analysis council of china association for disaster prevention. 2012-10-27 17. bois f y, jackson e t, pekari k, et al. population toxicokinetics of benzene. environmental health perspectives, 1996, 104（s6）: 1405-1411. 18. elizabeth a brown, michael l shelley, jeffrey w fisher. a pharmacokinetic study of occupational and environmental benzene exposure with regard to gender. risk analysis, 1998, 18( 2): 205-213. 19. u.s. environmental protection agency. health assessment of 1,3-butadiene (epa/600/p-98/001f). washington, dc, 2002 20. reitz rh, gargas ml , andersen me, et al. predicting cancer risk from vinyl chloride exposure with a physiologically based pharmacokinetic model. toxicology and applied pharmacology , 1996 , 137(2): 253-267. 21. agency for toxic substances and disease registry. toxicological profile for benzene . u.s. department of health and human services, atlanta, ga, 2007. 22. 蒙进怀, 李少旦. 苯接触人群的生物标志物研究进展. 中 国职业医学, 2008, 35(4):327-328. meng jin-huai, li shao-dan. study progress on biomarker of people exposed to benzene. china occupational medicine, 2008,35(4):327-328. 23. 胡宣扬,倪波. 呼出苯和血苯作为生物接触指标的研究. 中华劳动卫生职业病杂志,1994,12(3):18-20+65. hu xuan-yang,ni bo. study of benzene concentration in blood and expired air as biomonitoring index. chinese journal of industrial hygiene and occupational diseases,1994,12(3):18-20+65. 24. 朱钧,张东普. 时间加权平均浓度与血苯相关性的研究. 中国工业医学杂志,2002,15(3):169-170. zhu jun, zhang dong-pu. the relationship between timeweighted average concentration of benzene and benzene concentration in blood. chinese journal of industrial medicine, 2002,15(3):169-170. 25. sato a, nakajuma t, fugiwara y, et al. kinetic studies on sex difference in susceptibility to chronic benzene intoxication with special regard to body fat content. british journal of industrial medicine, 1975, 32（4）: 321-328. 26. elizabeth delzell, nalini sathiakumar, mary hovinga, et al. a follow-up study of synthetic rubber workers. toxicology, 1996, 113: 182~189 27. 谭红汕, 缪文彬, 孙品等人. 1,3-丁二烯加合物 新研究 进展. 职业卫生与应急救援, 2009, vol. 27(6): 303~308. tan hong-shan, miu wen-bin, sun pin, et al. updated research on 1,3-butadiene adducts. occupational health and emergency rescue, 2009, vol. 27(6): 303~308. 28. 刘楠, 程娟, 李斌等人.1,3-丁二烯生物标志物的研究进展. 国外医学(卫生学分册), 2007, vol.34(6): 352~357. liu nan, cheng juan, li bin, et al. updated research on 1,3-butadiene biomarker. foreign medical sciences(section hygiene) , 2007, vol.34(6): 352~357. 29. 刘胜学. 1,3-丁二烯职业暴露人群 hprt 基因分子突变 机理的初步研究. 重庆: 第三军医大学, 2004 liu sheng-xue. preliminary studies on the hprt gene mutation mechanism in lymphocytes from 1, 3-butadieneexposed workers, phd thesis, third military medical university, 2004 30. s. fustinoni, l. perbellini, l. soleo, et al. biological monitoring in occupational exposure to low levels of 1,3butadiene. toxicology letters, vol. 149(1-3):353-360. 31. matthew w. himmelstein, max j. turner, bahman asgharian, et al.comparison of blood concentrations of 1,3-butadiene and butadiene epoxides in mice and rats exposed to 1,3-butadiene by inhalation. carcinogenesis, 1994, vol. 15(8): 1479~1486 32. 缪文彬, 王威, 夏昭林. 氯乙烯致癌机制研究进展. 中国 职业医学, 2007, 34(5): 420-422. miao wen-bin, wang wei, xia zhao-lin. research progress on mechanisms of carcinogenesis induced by vinyl chloride. china occupational medicine, 2007, 34(5): 420-422. 33. 王爱红. 接触氯乙烯的生物标志物. 国外医学卫生学分 册, 2003, 30(2): 87-90. wang ai-hong. a review of vinyl chloride’s biomarker. foreign medical sciences (section hygiene), 2003, 30(2): 87-90. 34. watanabe, p. g., mcgowan, g. r., madrid, e. o., and gehring, p. j.. fate of 14c-vinyl chloride following inhalation exposure in rats. toxicol. appl. pharmacol, 1976(37): 49-59. 35. e. d., stewart, r. d., and mutchler, j. e. (1969). monitoring exposures to vinyl chloride vapor: breath analysis and continuous air sampling. am. ind. hyg. assoc. j. 30, 537–544. published by atlantis press copyright: the authors 184 microsoft word water security risk assessment of the yellow river basin based on constrained-random weight and cloud model water security risk assessment of the yellow river basin based on constrained-random weight and cloud model song chenye, zhang ren*, zhou aixia, hong mei college of meteorology and oceanography, pla university of science and technology, nanjing 211101, china *e-mail: songchenye@126.com；zren63@126.com abstract present risk assessment methods so far were majorly based on the primary knowledge and were incapable of combining uncertainty into risk assessment outputs. in order to solve this problem, a new methodology for risk assessment using constrained-random weight method and cloud model is introduced here to perform the risk assessment of water security in the yellow river basin. the effect of knowledge deficiency on risk assessment can be illustrated on the instability of risk series by the constrained-random weight method; the result of assessment can incorporate the randomness and fuzziness of risks and corresponding assessments, and transition of quantified risk series into qualitative risk level can be possible as the presence of the cloud model. experiment about the water security risk assessment of the yellow river baisin proved the validity and applicability of this method. keywords: water security;cloud model;constrained-random weight;risk assessment 基于随机赋权法和云模型的黄河流域水安全风险评估 宋晨烨 张韧 周爱霞 洪梅 解放军理工大学气象海洋学院，江苏 南京 211101 摘要：针对现有风险评估方法大多着眼于已有知识背景下对风险大小的评估，难以将不确定性融入评估结 果的问题，引入约束随机赋权和云模型相结合的方法途径，对黄河流域的水安全风险进行了评估。约束随 机赋权法的引入可将知识不完备性对风险评估的影响体现于风险序列的不稳定性上，云模型方法的引入使 评估结果可包含风险及风险评估的随机性和模糊性信息，并将定量的风险序列与定性的风险等级区划合理 转换。以黄河流域为例，建立了水安全风险评估的指标体系和风险评估模型，并进行了仿真试验，验证了 方法的有效性和适用性。 关键词：水安全，云模型，随机赋权，风险评估 1. 引言 风险评估工作的目的是为决策者提供决策建议。 目前，“风险”尚未有明确统一的定义，纵观现有观点， 离不开风险是危险（不安全）程度“大小”及其发生 可能性的“不确定性”结合的分析途径[1]。 目前的风险评估研究主要是通过建立指标体系， 对指标赋权再通过相应评估方法对风险大小进行评 估。目前指标赋权主要有主观赋权法和客观赋权法， 以及主客观综合的赋权方法。主观赋权法由于主观随 意性过强，其结果受打分专家的主观性影响较大使其 风险评估的科学性不稳定；而客观赋权法则存在着评 估结果解释性和适用性差的缺点，所得的结果难以给 出明确的解释[2-3]。无论何种赋权方法，都旨在寻求一 种给予指标体系一个合理的权重分配。但由于在风险 识别阶段，风险因子的作用机制常不明确，只能得到 其中的部分致险因子相对重要程度。鉴于客观现实条 件下的知识不完备性，以及风险本身模糊性和随机性， 用精确的权重来反映各指标的重要性，进而对风险进 行分析和评估，将会忽略这种不确定性，反而不能合 理反映风险因子对风险的作用[4]。随机赋权法通过重 复试验产生权重序列，将现有知识信息充分利用，同 时将知识的不完备性和风险的随机性体现于风险序列 中。目前随机赋权法已成为一种常用的综合评价方法。 另一方面，将风险量化成一个确定的数值，一定程度 上有利于人们对风险大小进行理解，但单一的数值并 不能代表风险的全部特征，在实际应用中，需要寻找 一个更能全面体现风险特征的评估方法，可以在给出 定量结果的同时也给出易于直观把握的定性特征。常 用的定性定量转换方法中，如层次分析法、专家群体 打分法、控制论和定性分析等，不能同时兼顾随机性 和模糊性，李德毅院士提出的云理论可将语言值表示 journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 201-209 published by atlantis press copyright: the authors 201 willieb typewritten text willieb typewritten text received 29 october 2013 willieb typewritten text accepted 12 december 2013 willieb typewritten text 基于随机赋权法和云模型的黄河流域水安全风险评估  的定性概念和其定量表示之间的不确定性联系起来， 把风险在不确定性问题上的模糊性和随机性集成到一 起，构成定性概念和定量数值之间映射，目前，云理 论已被广泛应用于复杂系统的综合评估之中。近十年 来，“水安全”概念在政治领域和学术领域都得到广泛 的重视[5]，由于水安全是一个涉及水文学、环境科学、 经济学、哲学等多学科领域的复杂概念，同时涉及政 治和政策决策，不同专业背景的研究者对其研究角度 不同，目前还没有统一的定义，其内涵主要包括人类 生存与发展不存在水资源问题的危险和威胁的状态， 外延则与国家安全相关，包括满足人类需求、粮食供 应、生态系统安全、水资源供需、水灾害风险管理、 社会安全与稳定等[6]，涵盖了资源、环境、生态、社 会、政治、经济等多方面。本文所指的水安全以人类 社会的稳定和发展对水的需求为水安全问题的主体， 水资源为客体，以黄河流域的水安全为例，引入随机 赋权法，将现有知识信息充分利用，同时也将知识的 不完备性和风险的随机性融入评估结果之中；云模型 方法主要用于定性概念和定量数值的转化，在风险序 列分析及获取评估对象风险不同特征等方面，表现出 较好的客观性和适用性。 2. 水安全风险评估指标体系 黄河是我国第二大河，整体位于我国中北部，自 西向东分为三个地形阶梯，其中最高级海拔较高，属 于高寒地区，土地覆盖以牧草地为主，水质受水污染 影响较小；第二阶梯地势落差超过 1000km，水文气象 现象复杂，包括几个较大的地貌单元，其中水土流失 较为严重的黄土高原是黄河中游洪水和泥沙主要来源 区；第三阶梯为黄河冲击大平原，地势平坦，大量泥 沙在此淤积，排水条件差，旱涝灾情严重。黄河流域 大多数地区属于干旱半干旱气候区，蒸发强度高，流 域内水资源紧缺，不同区间径流量悬殊，兰州站以上 和三门峡至花园口区间水资源量丰富，兰州站至花园 口之间地区水资源贫乏，是干旱事件的多发区。整体 而言，黄河流域的水资源量、气候特点、地理环境以 及社会结构、经济发展、人口分布都存在着严重的分 布不均现象，例如，龙门以下的地区占有黄河流域人 口数量的 70%，然而该区域面积只占全流域 32%。水 资源短缺、生态环境脆弱、水土流失严重以及社会发 展结构有待调整是黄河流域水资源管理工作的重点。 对风险评估工作而言，无论采用何种评估方法， 首先要建立一个合理的风险评估指标体系[7]。水安全 评估系统是一个由众多子系统构成的复杂系统，具有 层次性，指标体系设置须考虑区域自身特点、水资源 系统特性、可操作性以及实际可获得的数据[8]。按照 构建指标的科学性、完备性、动态性与静态性结合、 定性与定量结合及可比性、可操作性等原则[9]，以水 安全风险作为指标体系的目标层，将风险的致灾因子 危险性，承灾体脆弱性，风险防范能力三个方面作为 指标体系的准则层[10]。致灾因子的危险性表示导致风 险的直接因素；承灾体脆弱性表示在致灾因子作用下， 承灾体受危害影响的程度；风险防范能力是指承灾体 对致险因子的抵御和防范，以减轻灾害后果的能力。 参考前人的工作 [11-14] ，从水资源利用过程中“自然” 和“社会”两个角度出发，针对黄河流域水资源短缺、 中下游水土流失严重的特点，对危险性构建包括水资 源本身的丰沛程度、水资源可利用程度、水资源受污 染程度三个一级指标的指标体系；对承灾体脆弱性从 暴露性和敏感性两个一级指标，从经济，人口，农业 进行分析，构建相应的二级指标；风险防范能力，将 水利工程调节能力、节水能力、废水处理能力作为一 级指标，最终得到的指标体系见表 1。 表1 黄河流域水安全风险评估指标体系 table 1 risk assessment index system of the huang river basin water security 目标层 准则层 一级指标 二级指标 序号 水资源 安全 致灾因 子 危险性 水资源丰 沛度 地下水资源量 1 地表水资源量 2 水资源可 利用程度 城市面积比 3 水污染程 度 生活污水 排放量 4 工业污水 排放量 5 农药施用量 6 承灾体 脆弱性 暴露性 人口数量 7 人口增长率 8 评估单元 gdp 9 耕地面积 10 敏感性 农业用水量 11 人均用水量 12 风险防 范 能力 水利工程 调节能力 大中型水库 个数 13 大中型水库 储水量 14 节水能力 水重复利用率 15 节水灌溉 面积比 16 废水处理 污水处理厂污 水处理比 17 3. 约束随机赋权法 权重是指标重要性的量化表示，可根据目标的重 要性赋予不同的值，且所有指标权重之和应等于 1[15]。 但由于知识不完备性和风险本身的随机性及其他各 15 不确定性，各个指标之间具体的相互关系无法得到明 确的结论，此时若对指标赋予精确的权值不仅会使赋 权结果失真，还会导致评估结果可参考性降低[16]。 约束随机赋权将风险的知识不完备性和指标之间 关系的模糊性通过随机对指标赋权的方式体现；对指 标随机赋权重复多次可以体现风险因子作用的随机 性；根据风险已知的并且确定的知识，通过对指标体 系分析并以约束条件的形式对随机赋权进行约束。以 在约束条件下的随机多次赋权，模拟本身具有模糊性 published by atlantis press copyright: the authors 202 宋晨烨，张韧，周爱霞，等.   各个风险因子的随机作用产生风险的过程，得到的风 险序列则可包含评估对象风险的总体特征信息。 3.1 指标分析和约束条件确定 考虑指标间的部分可知性，在专家建议的基础上， 对已知部分作有弹性约束，对相互作用以及重要性程 度无法明确对比的部分指标随机生成。具体方法如下： 若  1 2, , , p   是指标体系的 p 个评估指标的 权重值，根据指标权重的定义，所有指标权值在 0 和 1 之间，且所有指标权重之和为 1，由此产生约束条件： 约束 1： 1 1 p j j    ，且  0,1j  。 其他的约束条件，可以结合对风险因子的分析和 专家的建议，根据对指标体系最大程度的掌握得到。 在确定约束条件的过程中，约束条件个数没有硬性范 围，但给出约束条件的前提是可以较为明确地判断指 标之间重要性程度的关系，对于无法明确判断的，不 必给出约束。 本研究以  1 2 17, , ,   代表水安全指标体系中 17 个指标的权重。除基本约束条件约束 1 外，根据对 水安全风险因子分析，结合专家建议，可以得到针对 黄河流域水安全风险指标体系的指标权重约束条件： 约束 2：  6 4 5,   ；水污染的主要来源是人类 工业的发展和人口的增多导致的生活污水排放增加， 而农业生产对水资源和水循环的影响主要是由于对水 资源的消耗以及下垫面植被覆盖的改变导致的水循环 改变，因此在“水污染程度”的三个指标中，“农药施 用量”的权重比“生活污水排放量”和“工业污水排 放量”小 约束 3：  10 7 8 9 11 12, , , ,      ；虽然农业用水 是水资源开发利用中耗水最大的方向，但考虑农业生 产的经济产值较工业生产较低，其对水安全风险的暴 露性应较小，因此将耕地面积在所有承灾体脆弱性的 指标中的权重最小。 约束 4：    13 17 1 12max ~ min ~    ；黄河流 域对水安全问题的防范能力仍然较弱，风险的大小更 多是受自然资源和自然环境本身的危险性以及人类社 会系统脆弱性的影响，因此风险防范能力的指标权重 应小于致险因子危险性和承灾体脆弱性的指标权重。 3.2 权重矩阵的生成 约束随机权重法不明确给出所有指标的具体权 重，对一个有 p 个指标组成的指标体系，设生成的指 标序列组数为 m（本研究中取为 100）,则在约束条件 下随机生成相当规模的矩阵 ij m p  ， ij 表示第 j 个 指标的第 i 次随机生成权重值。以表 1 中所列的 17 个 指标组成的指标体系为例，具体做法如下： （1）由 matlab 中 random 函数生成 m=100 个容 量为 p=17 的一维随机数组，得到规模为100 17 的随 机矩阵； （2）依上一步得到的约束条件调整每一行数组元 素的排序，使矩阵每一行数组都满足 2.1 所列的约束 条件； （3）将所有行的行数组归一化，得到满足约束条 件的随机矩阵  100 17ij   ，即指标权重矩阵（表 2）。 表 2 中的每一组权值序列仅代表一种可能的权值 分配。由于权值分配的不确定性是知识不完备和致险 因子本身的不确定性造成的，而知识不完备性同样也 是风险的一部分，因此在约束条件下产生的各种权值 序列在风险的产生机制上是等可能的。 须说明的是，上述的操作方法得到的权重序列并 不是按先确定约束条件，再随机生成指标值的顺序得 到，而是用约束条件对已生成的随机数组调整，进而 给指标权值赋值。同时，由于在计算机上产生的随机 数都是按照一定的计算方法产生的，不可能是真正的 随机数[1]，事实上得到的只是“伪随机数”。然而，对 指标值随机产生大量权重序列的意义在于在已知条件 下最大限度模拟风险产生的不同可能，这样的得到的 伪随机数列可以很大程度上实现对风险的模拟，因此 上述的操作方法仍是可用的。 表 2随机赋权法生成的部分权重序列 table 2 risk series generated by constrained-random weight 第一组权值序列 1 j 0.035454 0.089632 0.110402 0.065976 0.042139 0.034933 0.148333 0.063914 0.026279 0.021192 0.174896 0.135357 0.01558 0.003082 0.016746 0.010018 0.006068 第二组权值序列 2 j 0.092578 0.087565 0.088072 0.071285 0.07504 0.066425 0.05521 0.101312 0.050307 0.048926 0.064301 0.045173 0.018263 0.018857 0.030383 0.04173 0.044573 …… 3.3 风险序列的生成 ( 1 ) j a j p  表示指标体系的 p 个指标，共有 n 个 评 估 对 象 。 指 标 的 量 化 值 结 果 为  jk p n  ( 1 ; 1j p k n   ， j k 指第 k 个评估对象的 第 j 个指标量化结果。根据随机生成的指标权重  ij m p  ，可得到风险矩阵:      ik ij jkm n m p p nr      ， published by atlantis press copyright: the authors 203 基于随机赋权法和云模型的黄河流域水安全风险评估  i kr 表示根据第 i 组指标权重值对第 k 个评估对象的评 估结果,即  ik m nr  的每一列代表对应的评估对象的风 险序列,每一行则对应一次指标赋权情况下所有评估 对象的风险。 本 研 究 中 ， 根 据 指 标 量 化 结 果 存 放 于 矩 阵   , 17, 8jk p n p n    （8 个评价对象/省）。其中人口 数据来自各省第六次全国人口普查数据公报，水资源 量，水污染水利工程相关数据来自《中国环境统计年 鉴 2011》。由公式:       , 100, 8, 17ik ij jkm n m p p nr m n p        得到。表 3 列出了在 2.2 中得到的 100 组权值序列对 应的 100 次指标融合得到的部分风险值，每一个评价 对象的 100 个风险值组成该评价对象的风险序列。 每一次赋权得到的风险值代表一种可能的致险机 制产生的风险值；不同的赋权导致的风险序列的波动 既包括了自然条件致险因子本身的不确定性，也包含 了知识不完备产生的风险。也可以认为，风险序列本 身就是现有知识水平下的评价对象量化以后的风险。 表 3 黄河流域水安全风险序列（部分） table 3 water security risk series of the yellow river basin 赋权次数 山西 内蒙古 山东 河南 陕西 甘肃 青海 宁夏 1 0.591319 0.60923 0.650574 0.597163 0.497486 0.584994 0.118856 0.445937 2 0.604665 0.530121 0.605908 0.542874 0.479991 0.510023 0.13183 0.486615 3 0.648038 0.491033 0.633445 0.551778 0.488425 0.498057 0.064303 0.506169 4 0.636975 0.56264 0.6 70558 0.594341 0.49139 0.530755 0.068905 0.481557 5 0.598728 0.53621 0.603228 0.508707 0.429022 0.540665 0.101174 0.563585 6 0.552852 0.522702 0.55233 0.477412 0.40931 0.565649 0.124374 0.538443 …… …… …… …… …… …… …… …… …… 100 0.616333 0.544584 0.646446 0.576116 0.490108 0.519891 0.086582 0.476727 4. 云模型风险区划建模 风险序列可看作是量化的风险（包括危险程度的 大小和不确定性），但单纯的数字序列通常无法体现风 险特征，也无法对辅助决策起到有效的帮助。因此将 风险序列与人类在衡量事物好坏时所习惯使用的语言 值(如“好”，“坏”)对应起来，是风险区划必须的工作。 云“远看有形，近看无边”的特征同人类对事物 的理解和评价思维方式有相似处。对于某一个语言值， 人们通常只能大致判断哪些属于“好”，哪些属于“坏”， 但不会给出精确的边界。因此，云也可以看作是人类 的语言值和自然的数字值之间互相对应的桥梁。 云模型由李德毅院士提出，是用语言值表示的定 性概念与其定量表示之间的不确定性转换模型[18]。设 u 是用精确数值表示的定量论域，c 是 u 上的定性概 念 ， 对 x u  存 在 一 个 有 稳 定 倾 向 的 随 机 数 ( ) [0,1]y x  ，表示 x 对 c 的确定度，则每一个 x 称为 一个云滴，x 在论域 u 上的分布称为云，用期望 ex, 熵 en 和超熵 he 三个数字特征来整体表征一个概念 [19-23]。期望值代表了云的整体属性，熵和超熵则是对 云的形状的描述：熵是对云代表的定性概念确定程度 的度量，描述了云的离散程度，在最后得到的风险云 图中显示为风险云的覆盖范围；超熵则是熵的不确定 程度，也就是熵的熵，可认为是对云的厚度的描述。 云模型方法适用于对某一语言概念在定性和定量上的 转化，可用于对已生成的风险序列做进一步分析，将 风险序列的量化数组对应成一朵语言上的评估云，最 终将评估对象的风险序列和评估结果（评语）对应起 来，并可进一步做风险的区划。 利用云模型的风险评价方法，首先将水安全风险 分为由好至差的奇数个语言等级，称为评语集。每个 风险等级对应一个评估云。所有的评估云置于同一个 坐标构成评估的云标尺。将某一评估对象的风险云与 云标尺对比即可得到的评估结果。最终所获得的风险 云，其期望值 ex 代表了评估对象所具有的风险的整体 水平，熵 en 代表了风险的离散程度，可以认为是评估 对象具有的风险对于某一评语等级的确定程度，风险 云离散程度越高，对应于某一评语等级的确定程度越 低。超熵 he 是风险云的厚度，代表评估对象所属的评 语等级在语言上的模糊性。 4.1 评语集的生成 云标尺的获得首先要得到各评估云的三个特征 值，再将特征值用正向云发生器云化而得。本研究中， 将水安全风险分为九个语言等级，对应的语言值为[极 好，很好，好，较好，中等，较差，差，很差，极差]。 其中[极好，好，中等，差，极差]五个等级的评语集 由变量云化过程和云发生器得到对应的评估；对[很 好，较好，较差，很差]四个评语集，则利用浮动云得 到。具体做法如下： 4.1.1 定量变量云化得到评语集 根据云的定义，对某一过程表示定性概念的定量 变量进行云描述，包括获取云的数字特征和形状，这 一过程被称为变量云化过程[24]。设 ( 1 )ja j p  表示 published by atlantis press copyright: the authors 204 宋晨烨，张韧，周爱霞，等.   指标体系的 p 个指标，共有 n 个评估对象。指标的量 化值结果为 jk p n  ( 1 ; 1 )j p k n   ， j k 指第 k 个评估对象的第 j 个指标量化结果。在此次研究中， 分析指标量化结果，取所有评估对象的各指标值的 （0,20,40,60,80,100）百分位数，组成各风险评估云 的上界和下界的指标值集合，用上一节的约束随机赋 权法得到的权重矩阵，生成评语集风险序列，得到评 语集定量变量的云化初始值，取其期望值作为对应风 险等级指标集的上界和下界，并通过正向云发生器[25] 得到五个评语集的云。云参数计算通过下式得到： ( ) / 2 ( ) / 6 ex a b en b a he k        (1) 其中，a 为约束的下边界，b 为约束的上边界，k 是常 数，通过试验确定取值;对于只有单边约束条件的定量 变量，可根据测试数据的最大上限和下限确定缺省边 界参数或期望值，然后参照上式计算云参数，用半升 半降云来描述[21]。 在此次研究中，经过定量指标云化产生的评语集 [极好，好，中等，差，极差]对应的云模型分别是： (0.0200 0.0288 0.001 )cloud x ex 极好 ， ， ， (0.3204 0.0492 0.0016)cloud 好 ， ， ， (0.5432 0.0250 0.0026)cloud 中等 ， ， ， (0.6967 0.0262 0.0016)cloud 差 ， ， ， 0.9200, 0.0167, 0.001 )cloud x ex 极差 （ 。 上述云模型中的三个数字为其三个特征值，例如 (0.5432 0.0250 0.0026)cloud 中等 ， ， 是对应评语为“中等” 的云模型，其期望为 0.5432，熵为 0.0250，超熵为 0.0026。 4.1.2 浮动云得到评语集 浮动云用于生成论域给定的语言值中没有覆盖的 空白区域的虚拟语言值[26]。设论域 u 中的两朵相邻基 云 1 1 1 1( , , )a ex en he 和 2 2 2 2( , , )a ex en he ，则位于两朵 基云之间任意位置 u 存在一浮动云的数字特征由下式 计算： 1 2 2 1 2 1 1 2 2 1 2 1 ( ) ( ) ( ) ( ) ex u en ex ex en ex ex en ex ex he ex ex he ex ex he ex ex                (2) 将定量变量的云化的结果和浮动云得到的结 果还原的云模型置于同一坐标系中，得到指标值取 在不同区间的云标尺。在本研究中，以相邻两朵评 语集的云模型生成浮动云对应的评语集云模型分 别是： (0.1702, 0.0390, 0.0013)cloud 很好 ， (0.4318, 0.0371, 0.0021)cloud 较好 ， (0.6199, 0.0256, 0.0021)cloud 较差 ， (0.8084, 0.0214, 0.0013)cloud 很差 。 其含义与定量变量云化得到的评语集云模型一致，例 如， (0.8084, 0.0214, 0.0013)cloud 很差 指评语为“很 差”的云，其对应的云模型的期望值为 0.8084，熵为 0.0214，超熵为 0.0013. 4.1.3 正向云发生器和云标尺生成 正向云发生器是从定性到定量的映射（图 1），根 据云的数字特征产生云滴。变量云化结果得到的云参 数通过正向云发生器，便得到对应评语云。 ( , )i idrop x  图 1 正向云发生器 fig.1 cloud generator 正向云发生器的具体步骤： 输入：由 3.1.1 和 3.1.2 产生的风险序列对应的特 征值 ( , , )ex en he ；输出：n 个云滴 x 及其确定度 y。 step1:生成以 en 为期望值，he 为方差的正态随机 数 enn；step2:生成以 ex 为期望值，enn 为方差的正 态随机数 ix ；step3：计算     2 2 2 ix ex enn iy e    ；step4：( , )i ix y 为一个云滴；step5：重复步骤（1）~（4），直到产生 n 个要求的云滴。 本研究中用指标值云化以及浮动云通过正向云发 生器得到的云标尺(图 2)，图中的底部横坐标为风险大 小，纵坐标为确定度。 published by atlantis press copyright: the authors 205 基于随机赋权法和云模型的黄河流域水安全风险评估  图 2 黄河流域水安全风险评语集云模型（云标尺） fig.2 cloud model of the huang river basin’s water security risk level 4.2 风险序列生成风险云 4.2.1 逆向云发生器 逆向云发生器用于从已知的云滴中求出正向发生 器的三个特征值 ( , , )ex en he [18]，实现定量值到定性概 念的转换（图 3）。将风险序列作为无确定度信息的逆 向云发生器的输入，得到该风险序列对应的风险云特 征值，再用正向云发生器，得到特征值相同的云作为 该风险序列的风险云。 ( , )i idrop x  图 3 逆向云发生器 fig.3 reversed cloud model 设  ik m nr  是所有评估对象的水安全风险矩阵， i kr 表示根据第 i 组指标权重值对第 k 个对象评估结果, 即 ik m nr  的每一列代表对应的评估对象的风险序列。 以 kr 表示第 k 个评估对象的风险序列，作为第 k 个评 估对象对逆向云发生器的输入项。具体做法如下[27]： 输入：样本点 , 1, 2, ,i ikx r i n  其中 。输出： 反应定性概念的数字特征 ( , , )ex en he 。 step1:计算样本平均值 1 1 n i i x x n    ，一阶样本绝对中 心距 1 1 n i i x x n   ，样本方差   2 2 1 1 1 n i i s x x n      ； step2: ex x ； setp3: 1 1 2 n i i en x ex n      ； step4: 2 2he s en  。 根据表 3 中各评估对象风险序列，通过上述步骤 得到相应云模型数字特征值： 表 4 黄河流域八个省份风险云数字特征值 table 4 eigenvalues of the assessed provinces’risk clouds 评估对象 山西 内蒙古 山东 河南 陕西 甘肃 青海 宁夏 ex 0.616 0.545 0.646 0.576 0.490 0.520 0.087 0.477 en 0.029 0.020 0.038 0.043 0.027 0.015 0.019 0.047 he 0.025 0.018 0.020 0.008 0.020 0.020 0.015 0.019 4.2.2 风险云再生成 再次通过 3.1.3 中的正向云发生器，得到特征值 与评估对象风险序列相同的风险云。将各评价对象的 风险云置于图 2 所示的云标尺中，可得到特征值与风 险序列所含信息对应的风险云分布图（图 4）。由此 产生特征值与风险序列所含信息对应风险云分布图。 由表 4 中的云模型数字特征，通过正向云发生器 还原出个评估对象的云模型图像，云的期望用云的整 体位置表示，熵是图中云的离散程度，超熵是云的厚 度，其中，横坐标为水安全风险的大小，纵坐标为确 定程度： published by atlantis press copyright: the authors 206 宋晨烨，张韧，周爱霞，等.   图 4 黄河流域八省水安全风险云模型 fig.4 cloud models of the eight provinces of the huang river basin’s water security risk 4.3 结论与分析 图 4 给出八个省份的风险云形状及风险云在评估 云标尺中所在位置。风险云的整体位置表示了风险的 大小，云滴离散程度和厚度表示云滴对评语的隶属程 度和风险云所对应的语言概念的清晰程度。从风险云 和云标尺的对应关系来看，山东水安全为八个省份中 最差，其风险云分布于“较差”和“差”之间；山西 的水安全分布介于“中等”到“差”之间，其期望值 与“较差”等级十分贴近，可将山西水安全风险等级 确定为“较差”；河南省位于“较差”和“中等”之间， 接近中等；内蒙古风险云几乎覆盖了整个“中等”评 语集，因此内蒙古的风险等级为“中等”；甘肃风险云 形状与内蒙接近，中心位置较内蒙古稍偏左，因此仍 将其风险等级定为“中等”，事实上其风险小于内蒙古； 陕西和宁夏的风险云都位于“中等”和“较好”之间， 偏向于“较好”，将此二者风险等级定位“较好”，但 需注意的是宁夏风险云覆盖范围比山西更广；青海的 风险云位于“极好”和“很好”之间，稍偏向于“很 好”，可将其风险等级确定为“很好”。 图 5 为 arcgis 软件下绘制出的风险区划图。黄河 发源于青藏高原，青海省位于黄河源区，水资源较为 丰沛，同时人口稀少，经济发展较下游地区相对缓慢， 社会需水量及水资源污染情况都相对较好，水安全风 险的暴露性和脆弱性都很小，因此水安全风险等级很 低；山东省位于黄河流域的下游，人口密集，水资源 消耗大，工业发展水平较高，水资源的危险性和脆弱 性都很大；山西省的水资源量本身较小，而人口基数 大，同时节水措施不够完善，危险性和脆弱性都很大， 而防范能力较低，导致山西省的水安全风险等级较高。 图5 黄河流域水安全风险区划 fig 5 water security risk zoning of the huang river basin 5 小结 通过水安全的风险因子分析，建立了风险评估指 标体系，在此基础之上对指标值特性进行分析并结合 经验知识和专家建议，确定了四个指标赋权的约束条 件，在满足约束条件的基础上随机大量产生权重序列， 得到各评估对象的风险序列。 对指标权重进行分析，通过定量指标云化和浮动 云生成算法，产生了风险评语集合云模型。对风险序 published by atlantis press copyright: the authors 207 基于随机赋权法和云模型的黄河流域水安全风险评估  列通过逆向和正向云发生器得到与各指标风险序列特 征值相对应的八朵风险云。最终得到所有评估对象的 风险等级。 常规方法对水安全的评估一般从水安全的定义和 内涵出发，构建指标体系，利用层次分析法、模糊综 合评价等方法确定评价对象的水安全风险大小，最终 得到的结果一般是将某区域的水安全定为“非常安 全”、“基本安全”、“不安全”等明确的评语等级。本 文尝试使用随机赋权和云模型结合的方法，最终得到 的结果也包含了水安全风险评语等级的划分，但在生 成风险云得到评估对象的风险等级的同时，风险云模 型图中同时可得到风险等级以外的信息，如风险评估 结果的离散性、评语的确定程度。约束随机赋权充分 利用现有知识体系的已知信息，将风险量化成风险序 列并通过云模型将其云化成易于直观感受的风险云， 将风险的大小和不确定性融合并在定量-定性之间合 理转化，在水安全风险评估表现出良好的适用性。 参考文献 1. christina cook,karen bakker.water security:debating an emerging paradigm.global environmental change.2012(22):94-102. 2. pang yan-jun,liu kai-di,zhang bo-wen,the metho of determing the objective index weight in the synthetic evaluation sysstem[j]. systems engineering-theory& practice.2001,8(8):37-42. 庞彦军，刘开第，张博文.综合评估系统客观性指标权重 的确定方法[j].系统工程理论与实践.2001,8(8):37-42. 3. wang ze-yan,gu hong-fang,yi xiao-xin,et al.a method of determining the linear combination weights based on entropy[j].systems engineering-theory&practice.2003,3:112-116. 汪泽焱，顾红芳，益晓新，等.一种基于熵的线性组合赋 权法[j].系统工程理论与实践,2003,3:112-116. 4. ma zhipeng,chen shoulun, li xiaoying. grey decision model of flood control program based on random weight.water resources and power, 4(2007):41-44. 马志鹏，陈守伦，李晓英.基于随机赋权法的洪水调度方 案灰色决策模型[j]. 水电能源科学,4(2007):41-44. 5. han yuping,ruan benqing.a preiminary study on index system of regional water security assessment. acta scientiae circumstantiae,2(2003):267-272. 韩宇平，阮本清.区域水安全评估指标体系初步研究[j]. 环境科学学报,2(2003):267-272. 6. li lei.intelligent multi-attribute evaluation of water security[d].hefei university of technology.2007. 李磊.水安全智能多属性评价[d].合肥工业大学.2007. 7. tang jun, zhao wenjie,li dejun.comprehensive effectiveness evaluation of warfare intelligence based on cloud model[j].ship electronic engineering.6(2012):37-39,70. 汤君，赵文杰，李德军.基于云模型的战场情报综合效能 评估[j].舰船电子工程. 6(2012):37-39,70. 8. wang shuyun. study on water security comprehensive assessment[d].nanjing:nanjing hydraulic research institute.2010. 王淑云.水安全综合平价研究[d].南京：南京水利科学研 究院.2010. 9. 王远坤.水安全评价体系及其应用研究[d].青岛：山东大 学.2006. 10. zou hengwan.risk assessment of climate change impacts on water resources and water disputes in china[d].nanjing:pla university of science and technology.2011. 邹恒婉.气候变化对中国水资源的影响与水资源争端风 险评估[d].南京：解放军理工大学.2011. 11. jia shaofeng,zhang junyan,zhang shifeng.regional water resources stress and water resources security appraisement indicators[j].progress in geography, 2002,21（6）：538-545. 贾绍风，张军岩，张士峰，等.区域水资源压力指数与水 资源安全评价指标体系[j].地理科学进展，2002,21（6）： 538-545. 12. xia jun,zhu yizhong.the measurement of water resources security:a study and chanllenge on water resources carrying capacity[j].journal of natural resources.2002,17(3),262-269. 夏军，朱一中.水资源安全的度量：水资源承载力的研究 与挑战[j].自然资源学报，2002,17（3），262-269. 13. zeng changyun,lee guibao.study on water environment security and evaluation index system_a case study of beijing[d].capital normal university,2004. 曾畅云，李贵宝.水环境安全及其指标体系研究——以北 京市为例[d].北京：首都师范大学,2004. 14. zhang qiaoxian,ouyang zhiyun,wang rusong,et,al.modeling water security in china and comparision of the strategies[j].advances in water science,2002,12(5):452-457. 张巧显，欧阳志云，王如松，等.中国水安全系统模拟及 对策比较研究[j].水科学进展，2002,13（5）：569-577. 15. liu yidan,dong bidan,et,al. 刘一丹，董碧丹，等.基于模糊评估的等级保护风险评估 模型[j].计算机工程与设计.2012,2（34）:452-457. 16. yacov y. haimes.on the comples definition of risk: a systems-based approach.risk analysis.[j]2009,29（12）. 17. hu haipeng.a new method in producing pseudo random number and their statistical performance[d]. changsha:national university of defense technology,2007. 胡海朋.一种新的伪随机数产生方法及其统计性能分析 [d].长沙：国防科学技术大学.2007. 18. liu changyu,li deyi,pan lily.uncertain knowledge representation based on cloud model[j].computer engineering and applications.2004,2. 刘常昱，李德毅，潘莉莉.基于云模型的不确定性知识表 示[j].计算机工程与应用.2004，2. 19. li deyi,duyi.artificial intelligence with uncertainty[m].national defense industry press.2005,7. 李德毅，杜鹢.不确定性人工智能[m].国防工业出版 社.2005，7. published by atlantis press copyright: the authors 208 宋晨烨，张韧，周爱霞，等.   20. li deyi,shi xuemei,meng haijun.membership cloud and membership cloud producer[j].computer research and advance,1995,32(6):15-20. 李德毅，史雪梅，孟海军.隶属云和隶属云发生器[j].计 算机研究和发展,1995,32(6):15-20. 21. li deyi, shi x m, gupta m m. soft inference mechanism based on cloud models. in: proceedings of the 1st international workshop on logic programming and soft computing: theory and applications, bonn, germany, 1996. 22. li deyi, han j, shi x m. knowledge representation and discovery based on linguistic models. in:lu h j, motoda h eds.kdd:techniques and applications. singapore: world scientific press, 1997, 3-20. 23. li deyi, han jiawei, shi xuemei, et al. knowledge representation and discovery based on linguistic atoms. knowledge-based systeme,1998(10): 431-440. 24. li deyi. knowledge representation in kdd based on linguistic atoms. journal of computer science and technology,1997,12(6):481-496. 25. du xiangyu,yin quanjun,huang kedi,liang diannong.transformation between qualitative variables and quantity based on cloud models and its application[j].systes engineering and electronics,2008,30(4):772-776. 杜湘瑜，尹全军，黄柯棣，梁甸农.基于云模型的定性定 量 转 换 方 法 及 其 应 用 [j]. 系 统 工 程 与 电 子 技 术 [j].2008,30(4):772-776. 26. li guofeng.spatial data mining[d].xi’an university of electrical science,2005. 李国锋.空间数据挖掘技术研究[d].西安电子科技大 学,2005. 27. lu huijun,wang ye,li deyi,liu changyu.the application of backward cloud in qualitative evaluation[j].chinese journal of computers.2003,26(88):1009-1014. 吕辉军，王晔，李德毅，刘常昱.逆向云在定性评价中的 应用[j].计算机学报. 2003,26(88):1009-1014. published by atlantis press copyright: the authors 209 on the coastal erosion risk assessment indexes lifen xu college of urban and environmental sciences & laboratory for earth surface process of ministry of education, peking university, beijing 100871; national marine environment monitoring center, dalian 116023 shiyong wen, dongzhi zhao national marine environment monitoring center, dalian 116023 xuegong xu college of urban and environmental sciences & laboratory for earth surface process of ministry of education, peking university, beijing 100871 received 6 september 2013 accepted 18 october 2013 abstract china is one of the most severely affected countries by coastal erosion. about 70% of the sandy beaches and most of muddy tidal flat in open water regions are erode. coastal erosion results in the width of beach being narrowed, bathing beach destruction, coast-protection facilities and roads collapse, and wetland deterioration. these destructions bring many losses to the local people. however, previous coastal erosion studies focused on the hazard of erosion and paid less attention to the risk of erosion. therefore, it’s necessary to study the mechanism of coastal erosion and build the coastal erosion risk assessment system for the effective coastal management. on the basis of the existed researches, this paper builds the coastal erosion risk assessment index system which contains hazard index system and vulnerability index system. the index system is expected to provide a practical theoretical basis for coastal erosion risk assessment. keywords: coastal erosion, risk assessment, index 海岸侵蚀灾害风险评估指标体系研究 徐丽芬 1,2 , 文世勇 2 ，赵冬至 2,* ，许学工 1 1. 北京大学城市与环境学院 地表过程与模拟教育部重点实验室，北京 100871 2. 国家海洋环境监测中心，辽宁 大连 116023 摘要 我国是海岸侵蚀灾害最严重的国家之一，约有 70% 的砂质海滩和大部分处于开阔水域的泥质 潮滩受到侵蚀。海岸侵蚀使得海滩劣化、浴场破坏、湿地损失、堤坝倒塌等，给沿岸地区的生 产、人们生活造成损失。以往的研究主要关注海岸侵蚀灾害的损失；从风险评估角度开展海岸 侵蚀研究还少见报道，国内外在此方面的研究也相对薄弱。系统全面地认识海岸侵蚀灾害的影 响因素、作用机理、发生过程及其危害，建立海岸侵蚀灾害风险评估指标体系，是海岸侵蚀灾 害风险评估的理论基础。本文主要基于大量的文献，提取相关研究方案中的指标参数，构建海 岸侵蚀灾害风险评估指标体系。指标体系中包含致灾因子的危险性指标以及承灾体易损性指标。 关键词：海岸侵蚀，风险评估，指标体系 * 通讯作者，e-mail: dzzhao@nmemc.gov.cn 基金项目：国家海洋局业务工作海洋灾害监测、预警与预报；国家自然科学基金项目（no.41271102, 40830746） journal of risk analysis and crisis response, vol. 3, no. 3 (november 2013), 146-155 published by atlantis press copyright: the authors 146 mailto:dzzhao@nmemc.gov.cn 1 引言 海岸侵蚀（coastal erosion）指在海洋动 力作用下，导致海岸线向陆迁移或潮间带滩 涂和潮下带底床下蚀的海岸变化过程 [1] 。狭 义的海岸侵蚀仅指自然海岸的侵蚀后退过 程；广义的海岸侵蚀不仅包括自然海岸的蚀 退，而且包括人工海岸的破坏过程 [2] 。我国 海岸变化经历了整体淤进或稳定（1950s 之 前）——局部侵蚀（1950s-1970s）——普遍 侵蚀（1970s 之后）的发展过程 [1] 。据统计， 约有 70%的砂质海滩和大部分处于开阔水 域的泥质潮滩受到侵蚀 [3] 。各海域沿岸侵蚀 岸线占总岸线的比例：在渤海沿岸为 46%， 黄海沿岸为 49%，东海沿岸（包括台湾岛） 为 44%，南海沿岸（包括海南岛）为 21% [4] 。 海岸侵蚀使得海滩劣化、浴场破坏、湿地损 失、堤坝倒塌等，给沿岸地区的生产和人民 财产带来损失。以往的海岸侵蚀防治主要采 用硬工程措施以防堵为主，虽有立竿见影之 效，但也为将来埋下隐患。系统全面地认识 海岸侵蚀灾害的影响因素、作用机理、发生 过程及其危害，建立海岸侵蚀灾害风险评估 体系，对海岸综合管理具有重要意义 [5] 。 目前，绝大多数研究成果集中于海岸侵 蚀现状 [3, 6-18] 、原因 [2, 4, 19-29] 以及未来预测分 析 [30-42] 。少数研究从灾害风险评估视角对海 岸侵蚀开展研究，如张晓龙等定性分析了黄 河三角洲在海面上升、风暴潮、黄河断流等 灾害风险因素作用下，黄河三角洲海岸侵蚀 将加剧，滨海湿地将损失和退化，而没有进 行定量的风险评估 [5] ；王文海等就海岸侵蚀 灾害本身进行评估，提出采用经济总损失 量、单位岸线损失量、人均损失量、国民生 产总值损失率、国民收入损失率、预算内收 人损失率、受灾率和土地损失率等指标作为 评估参数，将海岸侵蚀灾害分为特大灾、大 灾、中灾、小灾和轻灾五种 [43] 。del río 以 及 kane 等对海岸侵蚀灾害进行风险评估， 建立了包含 11 个自然因素（危险性指数） 和 6 个社会经济要素（承灾体的影响指数） 的指标体系，并对各指标进行分级，最终采 用加权平均的方法进行综合评价，给出风险 值 [42, 44] 。domínguez 通过收集 1956，1977， 1984，1989，1992，1994，2001 等 7 期航片， 分析西班牙西南岸的易损性。首先通过校正 影像，用断面法分析岸线后退速率；其次， 将海岸按使用类型分为四类——高密度建 筑区，零星建筑区，农用区，自然区；最后， 综合分析岸线后退速率与海岸使用类型确 定海岸的易损性，将易损性分为五级 [45] 。 然而，这些风险评估多侧重于某一方 面，或侧重致灾因子的危险性评价 [5] ，或侧 重承灾体的易损性评价 [43] 。domínguez 的研 究虽然综合考虑了危险性和易损性，但其危 险性仅仅根据历史岸线后退速率而定，未考 虑其他相关因素。本文拟综合考虑致灾因子 的危险性以及承灾体的易损性，构建海岸侵 蚀灾害风险评估指标体系。研究方法主要基 于大量的文献，提取相关研究方案中的指标 参数予以综合。 2 海岸侵蚀灾害风险评估内涵 不同的学者对灾害风险的认识和表达 有一定的差异。ipcc 的报告指出风险是灾 害发生概率与其对承灾体影响程度的乘积 [46] ；crichton 提出风险取决于 3 个因素：致 灾因子、脆弱性和暴露性 [47] 。联合国指出风 险是致灾因子、脆弱性和恢复力的函数 [48] 。 但无论以何种方式表达，灾害风险的最核心 的含义即为： ( , )r f h v （1） 式中， r 为海岸侵蚀风险； h 为海岸 侵蚀的致灾因子发生的概率，即危险度；v 为承灾体易损度，即暴露在海岸侵蚀风险中 的各种要素自身遭受侵蚀的可能损失。 海岸侵蚀灾害风险指海岸侵蚀灾害发 生的可能性及其可能造成的损失 [49] 。依据海 岸侵蚀灾害分析评估的内涵，可知海岸侵蚀 灾害风险评估指标体系应包括海岸侵蚀灾 害危险度评估指标与承灾体易损度评估指 标。此外，海岸侵蚀灾害相较于其他灾害， 有一定的特殊性。海岸作为一个侵蚀对象的 同时，它还是承载其他承灾体的载体。因此， 在评估其风险时，应首先考虑海岸本身的性 质。海岸性质的不同决定着海岸侵蚀分布格 published by atlantis press copyright: the authors 147 局的不同。在此主要体现在海岸类型上，不 同的海岸类型，遭受侵蚀的可能性不一样， 海岸类型控制着海岸侵蚀空间分布规律。海 岸类型大体上可分为：砂（砾）质海岸、基 岩海岸、淤泥质海岸、生物海岸、人工海岸。 在我国，总体来说，以砂质海岸侵蚀最为明 显，且疏松的砂质海岸较砾质海岸抗侵蚀能 力差。基岩海岸的蚀退速率很小。淤泥质岸 段侵蚀不明显。人工岸段基本保持稳定。在 风险评估实践中，评估指标值会因海岸类型 的不同而有所差异。因此，在具体实施海岸 侵蚀风险评估时，需在评估指标的基础上， 对每一类型的海岸进行分别评估。 3 海岸侵蚀灾害危险度评估指标 3.1 影响海岸侵蚀灾害危险度的主要 因子 造成我国海岸侵蚀灾害的原因多种多 样，但基本可归纳为自然因素和人为因素两 大类。其中，自然因素包括：风暴潮、海浪 侵袭，海平面上升，河流输沙减少；人为因 素包括：海岸采砂，河流水利工程拦截泥沙， 海岸工程加强了水动力，海滩植被的破坏 等。而在这些影响因素中，风暴潮、海浪， 海平面上升，河流入海泥沙量减少，人类活 动四个方面是造成海岸侵蚀最主要的因素。 3.1.1 风暴潮、海浪 在近岸，风暴潮和海浪这两个过程并存 并且相互影响。在此，将两个过程统一成一 个因素来看待。实际上，风暴潮、海浪等对 海岸侵蚀的作用可理解为离岸输沙作用，即 通过海岸增水形成强大的垂直环流或侵蚀 性裂流，把波浪掀起的泥沙和本身侵蚀的底 沙带到离岸很远的深处堆积下来或被潮流 带走 [30] 。有时候一次强风暴造成的侵蚀量甚 至超过正常天气下全年的侵蚀量，如 1992 年的 9216 号强热带风暴致使山东境内的砂 质海岸后退 3-12m，最大达 30m [50] 。 风暴潮引起的海岸侵蚀量，通常用侵蚀 的沉积物体积来表达或者用岸线后退的距 离来表达。对于侵蚀的沉积物体积，往往采 用波浪、海流等的输沙理论予以计算。由于 侵蚀体积的大小与波浪的冲刷力呈正相关 [51] ，与泥沙粒径以及泥沙颗粒容重也非常相 关 [58] 。从构建的定量计算公式看，很多研究 表明风暴潮影响下海滩离岸输沙量 v 与 dean 参数相关，其中 dean 参数是浪高、泥 沙沉积速率以及风暴持续时间的函数 [52-56] 。 而 larson 等人在借鉴已有研究结果的基础 上，推出海岸侵蚀体积的定量计算公式，认 为侵蚀体积是海滩坡度、沙丘高度、波浪爬 高、持续时间等的函数 [57] 。mendoza 等人使 用波高、波周期、风暴持续时间、泥沙粒径 以及岸滩坡度等因素来预测风暴潮对海滩 的侵蚀 [54] 。 相对于侵蚀体积的计算，用岸线后退距 离表达的海岸侵蚀量虽然不乏公式的定量 计算，但更多的是基于统计建立的经验关 系。kriebel 等通过长期对 delaware 附近海 岸剖面观察和测量，提出了风暴潮造成海岸 线后退的经验公式，认为风暴潮引起的岸线 后退是近岸波高、风暴增水以及风暴持续时 间的函数 [59] 。jiménez 等提出用有效波高和 最大波高对应的周期来表达风暴潮触发的 海岸侵蚀潜在指数 [55] 。cooper 等提出风暴 潮对海滩的影响是风暴潮持续时间、风速、 风向综合作用的结果 [60] 。 综合上述分析，风暴潮、海浪等对海岸 侵蚀的指标至少应包含：波高、最高波浪周 期、平均波向、风浪持续时间、波浪方向频 率等。 3.1.2 海平面上升 海平面上升与长期的海岸侵蚀速率有 关，而且其影响是整体性和缓变性的。海平 面上升对海岸的侵蚀作用，主要是由于加大 的水深使波浪对古海岸的扰动作用减小致 使海底的横向供沙减少，却加强了激浪对上 部海滩的冲刷而造成的 [20] 。 在定量计算海平面上升引起的海岸线 后退量的研究中，bruun 法则 [61] 应用最多。 bruun 法则是基于以下假设得出的：海平面 上升过程中，海岸均衡剖面向岸和向上移 动，从而造成海滩侵蚀和滨外浅水区堆积； published by atlantis press copyright: the authors 148 海滩的侵蚀物沉积在滨外浅水区，侵蚀和堆 积的沉积物数量相近；滨外浅水区的沉积厚 度与海平面上升值相等，从而维持了海岸均 衡剖面不变 [62] 。所以 bruun 法则的适用对象 为有明显的相对海平面上升，均匀的砂质沉 积物，沉积物的数量变化较小的砂质海岸。 在这种条件下，岸线后退速率与相对海平面 相对速率呈正相关，与岸滩坡度呈负相关。 王颖等对 bruun 图进行了修正（图 1），认为 相对海平面上升导致的海岸线后退量由两 部分组成：因海平面上升使部分海滩受淹没 而产生的后退量（图 1 中的 1 x ）以及相对 海平面上升而造成的海滩侵蚀后退量（图 1 中的 2 x ） [20] 。其研究结果表明：相对海平 面上升 0.5m，大连、秦皇岛、青岛、北海、 三亚海滩滨线将分别平均后退 39.6m、 36.7m、28.0m、22.5m、13.1m。不仅对于砂 质海岸，相对海平面上升与岸线后退存在这 种关系，基岩海岸与相对海平面上升之间也 存在一个类似的较共通的关系，即相对海平 面上升速率与岸线后退速率呈正相关 [63] 。 通过上述分析可以看出，在衡量海平面 与海岸侵蚀的关系中，相对海平面上升高度 （速率）、海滩近滨坡度是关键指标。 （图中假定海面上升前后的海滩剖面均已达到平衡， y 为海面上升幅度） 图1 修正后的bruun图[20] fig.1 the revised diagrammatic sketch of bruun rule [20] 3.1.3 河流入海泥沙量 河流入海泥沙量是影响海岸侵蚀的最 重要原因。近三四十年来，中国沿岸河流入 海泥沙量减少了一半以上，由入海泥沙 20×10 8 t/a 减为不足 10×10 8 t/a，造成我国海 岸泥沙收支严重失衡，使得部分海岸由淤积 为主，转为以侵蚀为主 [1] 。 虽然很多研究定性分析了河流入海泥 沙量的减少是造成海岸侵蚀的重要原因，但 是定量刻画河流入海泥沙量与岸线后退之 间的关系却少见报道。河流入海泥沙量只是 一个河口断面的泥沙含量值，而泥沙入海后 一部分沉积在河口及沿岸，一部分则被潮 汐、海流扩散输送到近海沉积 [64] 。真正对岸 线塑造起作用的是在河口及沿岸沉积的部 分。泥沙入海之后，到底多大的比例是沉积 在河口及沿岸，却无法单纯从河流入海泥沙 量这个指标得出。所以急需找到一个监测入 海后泥沙含量的指标。悬浮泥沙含量是一个 综合因素，它不仅受到河流入海泥沙量、海 岸工程的影响，而且也会受到风暴潮、海浪、 海平面上升等因素的影响，海流是输沙的主 要动力，风浪（包括风暴潮）是掀沙的主要 动力。但悬浮泥沙的主要来源是径流 [65] 。 秦蕴珊等将悬浮泥沙含量近似地看作平均 输沙率，根据河流入海泥沙量、近河口浅海 区某一断面上的输沙量，推测两断面之间的 冲淤率，即入海泥沙对海岸塑造的影响 [66] 。 由此可以认为对比分析河流入海泥沙 量、悬浮泥沙含量以及沿岸流的输沙率，可 published by atlantis press copyright: the authors 149 得到河口及沿岸的泥沙平衡，进而建立泥沙 平衡与岸线后退速率的关系。 3.1.4 人类活动 在影响海岸侵蚀的人类活动中，人工采 砂对海岸侵蚀造成的影响最为直接和显著。 因建筑上的要求，人们在海岸地区的海 滩、河口和水下采砂，造成海岸泥沙亏损， 打破岸滩剖面平衡，造成侵蚀。采砂活动对 岸线的影响是迅速而且非常剧烈的。据初步 统计，仅 1992 年 4~7 月辽东湾滕房身至田 家崴子一带海滩采砂就达 40 万 t，使这一带 近百米的沙滩消失 [21] 。 因此，人工采砂量是需要考虑的指标。 3.2 海岸侵蚀灾害危险度评估指标体 系的建立 根据上述分析，将风暴潮、海浪，海平 面上升，河流入海泥沙量以及人类活动这四 大因素根据要素属性解析为如下指标，如表 1 所示。 表 1 中的指标与海岸侵蚀的关系，可以 定性描述如下：流速、水深、泥沙颗粒容重、 泥沙粒径是近滨的水文泥沙条件，决定着水 流挟沙能力。若将风暴潮、波浪等对海岸侵 蚀的作用理解为离岸输沙，则水流挟沙能力 越强，海岸侵蚀的可能性则越大。波高、波 向以及波周期是描述海浪特征的重要指标。 其中，波高与海浪的能量有关，浪越高，表 明海浪的能量越大，对海岸侵蚀的潜在破坏 力也越大；波向决定着海岸侵蚀的展布方 向；波周期与海浪的频率有关，频率越高， 侵蚀的可能性也越大。风暴增水值是衡量风 暴能量的重要因素，风暴潮增水值越大，表 明风暴潮所蕴含的能量越高，风暴潮对海岸 侵蚀的可能性越大。风暴潮持续的时间也是 影响海岸侵蚀程度的重要因素，持续时间越 长，侵蚀的程度将会越大。海平面上升高度 越大，岸线后退量也越大；相反，近滨坡度 越大，岸线后退量越小。人工采砂量越大， 对海岸侵蚀的影响也越大。 表 1 海岸侵蚀危险度评估指标与编码 table 1 coastal erosion hazard indexes and codes 影响因素层（i 级） 要素指标层（ii 级） 名称 编码 名称 编码 风暴潮、海浪 h1 风暴增水 h11 风暴潮持续时间 h12 浪高 h13 波浪方向频率（或平均波向） h14 最大波浪周期 h15 海平面上升 h2 相对海平面上升高度 h21 海滩近滨坡度 h22 河流入海泥沙 h3 流速 h31 流向 h32 水深 h33 泥沙粒径 h34 河流入海泥沙量 h35 悬浮泥沙含量 h36 人类活动 h4 人工采砂量 h41 published by atlantis press copyright: the authors 150 4 承灾体易损度评估指标 4.1 承灾体范围的确定 承灾体是海岸侵蚀影响区内的土地以 及附着在其上的生态系统、建设在该区域内 的一切设施如港口码头、房屋、娱乐设施等。 虽然海岸侵蚀的影响范围会因不同类型的 岸段而有所差异，侵蚀后退量多则上百米， 少则几厘米。 根据陈吉余等 [1] 制定的海岸侵蚀强度 等级标准，年均侵蚀速率在 1~2m/a 的砂质 海岸即为强侵蚀。从海岸工程的设计标准为 50 年一遇至 100 年一遇的情景看，调查范围 距海岸线 200m 范围即可。考虑到滨海公路 作为承灾体的同时也是海岸侵蚀的人工阻 隔物（尤其对滨岸的公路），因此在有距岸 线较近的区域内，以滨海公路向海一侧为承 灾体的调查范围。而对于基岩海岸，由于它 的侵蚀速率非常低，统一规定以 200m 为界， 有失偏颇，故在此以高程低于 10m 来给出 调查范围。 综合这些分析，提出以下两条承灾体调 查范围划定方案： （1）距海岸线 200m 之内（辽东湾以滨 海路向海一侧为界） （2）以高程低于 10m 为界（在基岩海 岸区，海岸侵蚀基本影响不到距水边线 200m 左右的区域，此情景下可以采用高程 条件来进行调查范围的划定）。 4.2 承灾体因子分类体系的确定 制定合理的海岸侵蚀承灾体分类体系 是承灾体易损度的基础。海岸侵蚀承灾体的 分类主要遵循以下原则： （1）完整性。海岸侵蚀所影响到且会 受到损失或破坏的地物类型均需包含在内。 （2）可量化。承灾体分类是为易损性 评价做准备，因此，承灾体需易于量化，以 便于下一步的承灾体易损性评价。 依照上述原则，在参考大量相关文献材 料的基础上，结合海岸侵蚀灾害的危害特 点，确定了海岸侵蚀灾害承灾体因子及分类 体系编码等，为开展海岸侵蚀灾害承灾体易 损度评估奠定基础。 表 2 海岸侵蚀承灾体分类与编码 table 2 classifications and codes of coastal erosion risk receptors 一级 二级 三级 名称 编码 名称 编码 名称 编码 建筑物 a1 简易房 a11 平房 a12 多层楼房 a13 高层楼房 a14 港口设施 a2 防波堤 a21 土堤 a211 混凝土堤 a212 砌体堤 a213 码头泊位 a22 仓库 a23 交通设施 a3 高速公路 a31 一般公路 a32 国道 a321 省道 a322 乡村道路 a323 铁路 a33 工矿类 a4 盐田 a41 海岸防护工程及水工设施 a5 海岸防护工程 a51 海堤 a511 护岸 a512 保滩设施 a513 水工设施 a52 闸坝 a521 排水口 a522 published by atlantis press copyright: the authors 151 取水口 a523 养殖类 a6 鱼塘虾池 a61 海养育苗室 a62 旅游类 a7 海滩浴场 a71 旅游设施 a72 滨海公园 a73 农林生态类 a8 农地 a81 水田 a811 旱地 a812 林地 a82 乔木 a821 灌木 a822 园地 a823 草地 a83 苇地 a831 荒草地 a832 人工草地 a833 滩涂湿地 a84 由于海岸侵蚀是缓发灾害。它对承灾体 的损坏，不同于其他一些突发性的灾害如地 震、洪水那般瞬间摧毁性质。海岸侵蚀不至 于造成人员伤亡，因此承灾体的分类表中没 有考虑人口分布指标。 4.3 承灾体易损度评估指标体系 海岸侵蚀造成的承灾体损失主要包括 两部分：土地损失和经济损失。土地损失体 现在由于岸线后退造成如耕地、草地、湿地 等的减少。经济损失主要体现在海岸侵蚀造 成房屋倒塌、报废；公路、海岸防护工程、 码头、港池等的损坏、废弃；粮食的减产； 海滩旅游承载力降低导致的收入减少；以及 由于防止岸线后退投入的防护、加固费用， 城镇、村庄的搬迁费用等 [46] 。 衡量这些承灾体造成的可能损失即为 易损度。易损度的评估包含两个方面：一为 承灾体遭受侵蚀损坏的程度有多大，不同承 灾体，其承灾能力、抗灾能力和救灾能力不 同，可用损失率来表示；二为各承灾体的价 值量，在此以单位面积的货币量来表示。但 并非研究范围内所有的承灾体都会遭受损 失，只有暴露在岸线后退影响范围内的面积 才是成灾面积，因此还需加入暴露面积这一 指标。 ( , , ) p v f t e  (2) 整理成易损度评估指标如表 3 所示。 表 3 海岸侵蚀承灾体易损度评估指标与编码 table 3 vulnerability indexes and codes of coastal erosion risk receptors 5 结束语 本文在参考现有研究成果的基础上，根 据海岸侵蚀灾害风险评估的内涵，建立了包 括海岸侵蚀灾害危险度和承灾体易损度的 海岸侵蚀灾害风险评估指标体系。通过分析 海浪、风暴潮、海平面上升、河流入海泥沙 量减少、人类活动等主要致灾因子与海岸侵 蚀的关系，建立了相应的海岸侵蚀灾害危险 因素层（i 级） 要素层（ii 级） 名称 编码 名称 编码 承灾体类型 v1 价值 v2 单位面积价值 v21 暴露面积 v22 损失率 v3 published by atlantis press copyright: the authors 152 度评估指标；在承灾体的范围、分类的基础 上，提出承灾体易损度评估指标。本文所构 建的海岸侵蚀灾害风险评估指标体系，能为 海岸侵蚀风险评估提供借鉴和指导意义，进 而促进海岸带的综合管理。 6 参考文献 [1] 陈吉余, 夏东兴, 虞志英等, 中国海岸侵 蚀概要, (海洋出版社, 北京, 2010）. [2] 王文海, 吴桑云, 山东省海岸侵蚀灾害研 究, 自然灾害学报 (04) (1993) 60-66. [3] 夏东兴, 王文海, 武桂秋等, 中国海岸侵 蚀述要, 地理学报 (05) (1993) 468-476. [4] 季子修, 中国海岸侵蚀特点及侵蚀加剧 原 因 分 析 . 自 然 灾 害 学 报 (02)(1996) 69-79. [5] 张晓龙与李培英, 黄河三角洲滨海湿地 海岸侵蚀灾害及风险, 应用基础与工程 科学学报 14(s1) (2006) 247-252. [6] davidson-arnott and r.g.d.,j. ollerhead, nearshore erosion on a cohesive shoreline, marine geology122(4) (1995) 349-365. [7] 苗丰民, 李淑媛, 符文侠等, 辽东湾东部 砂岸的近期变化及演变趋势, 海洋学报 (中文版) (02) (1996)74-84. [8] list, j.h., a.h. sallenger jr.,m.e. hansen, et al., accelerated relative sea-level rise and rapid coastal erosion: testing a causal relationship for the louisiana barrier islands, marine geology 140(3–4) (1997) 347-365. [9] 王玉广, 李淑媛, 苗丽娟, 辽东湾两侧砂 质海岸侵蚀灾害与防治, 海岸工程 (01) (2005) 9-18. [10] andriani, g.f. and n. walsh, rocky coast geomorphology and erosional processes: a case study along the murgia coastline south of bari, apulia — se italy, geomorphology 87(3) (2007) 224-238. [11] 王玉广, 张宪文, 贾凯等, 辽东湾绥中海 岸侵蚀研究, 海岸工程 (01) (2007) 1-5. [12] 孔繁德, 秦皇岛市海岸侵蚀研究及防治 对策, 城市发展研究 (01)( 2008)73-76. [13] jabaloy-sánchez, a., f.j. lobo, a. azor, et al., human-driven coastline changes in the adra river deltaic system, southeast spain, geomorphology 119(1–2) (2010)9-22. [14] 徐宗军, 张绪良,张朝晖, 山东半岛和黄 河三角洲的海岸侵蚀与防治对策, 科技 导报 (10) (2010)90-95. [15] cui, b., x. li, coastline change of the yellow river estuary and its response to the sediment and runoff (1976–2005), geomorphology 127(1–2) (2011)32-40. [16] liu, y., h. huang,z. qiu, et al., detecting coastline change from satellite images based on beach slope estimation in a tidal flat, international journal of applied earth observation and geoinformation 23(2013)165-176. [17] 庄振业, 杨燕雄, 刘会欣, 环渤海砂质岸 侵 蚀 和 海 滩 养 护 , 海 洋 地 质 前 沿 (02)( 2013)1-9. [18] rahman, a.f., d. dragoni, b. el-masri, response of the sundarbans coastline to sea level rise and decreased sediment flow: a remote sensing assessment, remote sensing of environment 115(12) (2011)3121-3128. [19] 王文海, 我国海岸侵蚀原因及其对策, 海 洋开发 (01) (1987)8-12. [20] 王颖, 吴小根, 海平面上升与海滩侵蚀, 地理学报 (02) (1995)118-127. [21] 苗丰民, 李光天,符文侠等, 辽东湾东部 砂岸严重蚀退及其原因分析. 海洋环境 科学 (01) (1996) 66-72. [22] budetta, p., g. galietta, a. santo, a methodology for the study of the relation between coastal cliff erosion and the mechanical strength of soils and rock masses, engineering geology 56(3–4) (2000)243-256. [23] 丰爱平, 夏东兴, 谷东起等, 莱州湾南岸 海岸侵蚀过程与原因研究, 海洋科学进 展 (01) (2006) 83-90. [24] 温倩, 方凤满, 高健, 山东半岛海岸侵蚀 成因及防治对策研究, 国土与自然资源 研究 (01) (2006)52-53. published by atlantis press copyright: the authors 153 [25] 张晓龙,李培英, 现代黄河三角洲的海岸 侵蚀及其环境影响, 海洋环境科学 (05) (2008)475-479. [26] brunel, c., f. sabatier, potential influence of sea-level rise in controlling shoreline position on the french mediterranean coast, geomorphology 107(1–2) (2009) 47-57. [27] huang, g., time lag between reduction of sediment supply and coastal erosion, international journal of sediment research 26(1) (2011) 27-35. [28] 郭婷婷, 邵晓昕, 浅析海岸带侵蚀, 环境 保护与循环经济 (05) (2011) 37-39+59. [29] zhang, k.q., b. douglas, s. leatherman, do storms cause long-term beach erosion along the u.s. east barrier coast?, journal of geology 110(4) (2002) 493-502. [30] 杨燕雄, 贺鹏起,谢亚琮等, 秦皇岛海岸 侵蚀研究, 中国地质灾害与防治学报 (s1) (1994)166-170. [31] 陈坚, 谢在团, 未来海平面上升对厦门海 滩 侵 蚀 的 测 算 , 台 湾 海 峡 17(4) (1998)391-395. [32] leont'yev, i.o., modeling erosion of sedimentary coasts in the western russian arctic, coastal engineering 47(4) (2003) 413-429. [33] brown, i., s. jude,s. koukoulas, et al., dynamic simulation and visualisation of coastal erosion. computers, environment and urban systems 30(6) (2006) 840-860. [34] callaghan, d.p., p. nielsen, a. short, et al., statistical simulation of wave climate and extreme beach erosion, coastal engineering 55(5) (2008) 375-390. [35] 韩晓庆, 高伟明, 褚玉娟, 河北省自然状 态沙质海岸的侵蚀及预测, 海洋地质与 第四纪地质 (03)( 2008) 23-29. [36] callaghan, d.p., r. roshanka, s. andrew, quantifying the storm erosion hazard for coastal planning, coastal engineering 56(1)(2009)90-93. [37] hapke, c. and n. plant, predicting coastal cliff erosion using a bayesian probabilistic model, marine geology 278(1–4) (2010) 140-149. [38] trenhaile, a.s., modeling cohesive clay coast evolution and response to climate change, marine geology 277(1–4) (2010) 11-20. [39] loureiro, c., ó. ferreira, j.a.g. cooper, extreme erosion on high-energy embayed beaches: influence of megarips and storm grouping, geomorphology 139–140(2012) 155-171. [40] ranasinghe, r., d. callaghan, m.j.f. stive, estimating coastal recession due to sea level rise: beyond the bruun rule, climatic change 110(3-4)( 2012) 561-574. [41] walkden, m.j.a. and j.w. hall, a predictive mesoscale model of the erosion and profile development of soft rock shores, coastal engineering 52(6)( 2005) 535-563. [42] kane, h.h., c.h. fletcher, b.m. romine, et al., vulnerability assessment of hawai'i's cultural assets attributable to erosion using shoreline trend analysis techniques, journal of coastal research 28(3) (2012) 533-539. [43] 王文海, 吴桑云, 陈雪英, 海岸侵蚀灾害 评估方法探讨, 自然灾害学报 (01)( 1999) 71-77. [44] del río, l. and f.j. gracia, erosion risk assessment of active coastal cliffs in temperate environments, geomorphology 112(1–2) (2009) 82-95. [45] domínguez, l., g. anfuso, f.j. gracia, vulnerability assessment of a retreating coast in sw spain, environmental geology 47(8)(2005) 1037-1044. [46] ipcc, climate change 2001: impacts, adaptation and vulnerability, summary for policymakers, (2001)wmo. [47] crichton, d., the risk triangle. natural disaster management, tudor rose, london, (1999)102-103. [48] un, risk awareness and assessment in living with risk. isdr,un,wmo,asian published by atlantis press copyright: the authors 154 and disaster reduction center,geneva(2002) 39-78. [49]丰爱平, 夏东兴, 海岸侵蚀灾情分级, 海 岸工程(02)(2003) 60-66. [50] 王文海, 吴桑云, 陈雪英, 山东省 9216 号 强热带气旋风暴期间的海岸侵蚀灾害, 海 洋 地 质 与 第 四 纪 地 质 (04) (1994) 71-78. [51] overton, m.f., j.s. fisher, k. hwang, development of a dune erosion model using supertank data, in proceedings of 24 th international conference on coastal engineering, kobe, japan, 1(24)(1994). [52] larson, m. and n.s. kraus, sbeach: empirical numerial model for simulating storm-induced beach change, report 1: empirical foundation and model development, in technical report cerc-89-9 (us army corps of engineers,vicksburg, 1989) [53] wise, r.s., s.j. smith, m. larson, sbeach: numerical model for simulating storm-induced beach change, report 4: cross shore transport under random waves and model validation with supertank and field data, in technical report cerc-89-9 (us army corps of engineers, vicksburg, 1996) [54] mendoza, e.t. and j.a. jiménez, storm-induced beach erosion potential on the catalonian coast, journal of coastal research 48( sl) (2006)81-88. [55] jiménez, j.a., a. sancho-garcía, e. bosom, et al., storm-induced damages along the catalan coast (nw mediterranean) during the period 1958–2008, geomorphology 143–144(2012) 24-33. [56] jiménez, j.a. and a. sánchez-arcilla, medium-term coastal response at the ebro delta, spain, mar. geol. 114 (1993) 105-118. [57] larson, m., l. erikson, h. hanson, an analytical model to predict dune erosion due to wave impact, coastal engineering 51(8–9) (2004) 675-696. [58] overton, m.f., w.a. pratikto, j.c. lu, et al., laboratory investigation of dune erosion as a function of sand grain size and dune density, coastal engineering 23(1) (1994)151--165. [59] kriebel, d., r. dalrymple, a. pratt, et al., a shoreline risk index for northeasters, in asce (1997), pp.251-252. [60] cooper, j.a.g., d.w.t. jackson, f. navas, et al., identifying storm impacts on an embayed, high-energy coastline: examples from western ireland, marine geology 210(1–4)( 2004) 261-280. [61] bruun, p., sea-level rise as a cause of shore erosion, journal of the waterways and harbors division 88(1962) 117-130. [62] 李从先, 王平, 范代读等, 布容法则及其 在中国海岸上的应用, 海洋地质与第四 纪地质 (01) (2000) 87-91. [63] ashton, a.d., m.j.a. walkden, m.e. dickson, equilibrium responses of cliffed coasts to changes in the rate of sea level rise, marine geology 284(1–4) (2011)217-229. [64] 程天文,赵楚年, 我国主要河流入海径流 量、输沙量及对沿岸的影响, 海洋学报(中 文版) (04) (1985)460-471. [65] 姜太良, 房宪英,徐洪达等, 河北省海岸 带浅滩海域悬浮泥沙输送的估算, 黄渤 海海洋 (03) (1986)94-103. [66] 秦蕴珊, 李凡, 渤海海水中悬浮体的研究, 海洋学报(中文版) (02) (1982) 191-200. [67] 冯金良, 人类工程活动对秦皇岛海滩侵 蚀及淤积的影响, 海岸工程 (03) (1997) 41-46. published by atlantis press copyright: the authors 155 research article a comparative analysis of climate change risk response perception paths between northern and southern shaanxi siwen xue1,3,*, , zhou qi1,2,3 1school of geography and environment, baoji university of arts and sciences, baoji 721013, china 2shaanxi key laboratory of disasters monitoring and mechanism simulation, baoji university of arts and sciences, baoji 721013, china 3shaan’xi provincial key research center for socialism with chinese characteristics (baoji base), baoji 721013, china 1. introduction the fifth report of the ipcc pointed out that extreme weather and climate events have changed since 1950, and extreme climates have also occurred frequently [1]. in the context of global climate change [2], meteorological disasters occur frequently and the risks of climate change are increasing. it is thus necessary to step up efforts to address the risks of climate change. currently, there are three main ways to deal with climate change risks: mitigation, adaptation and avoidance [3]. the adaptation challenge grows with the magnitude and the rate of climate change. even the most effective climate change mitigation through reduction of greenhouse gas emissions or enhanced removal of these gases from the atmosphere (through carbon sinks) would not prevent further climate change impacts [4], making the need for adaptation unavoidable [5]. climate change mitigation consists of actions to limit the magnitude or rate of global warming and its related effects [6]. the main challenge is move away from coal, oil and gas and replace these fossil fuels with clean energy sources [7]. as for avoiding dangerous climate change, a study published in 2018 points at a threshold at which temperatures could rise to 4° or 5° through self-reinforcing feedbacks in the climate system, suggesting it is below the 2° temperature target [8]. therefore, different climate change risk response methods have certain challenges and shortcomings. and in the interaction between people and the environment, the perception of environment is the main basis for human decision-making behavior [9]. therefore, it is necessary to explore the formation mechanism of people’s climate change risk response perception, so as to overcome the difficulties and shortcomings in climate change risk response. most scholars believed that behaviors influencing people’s response to climate change risks are diverse. in many instances, there are many factors that cam enhance people’s ability to cope with climate change. these factors can include resources, education and information, gender, poverty, wealth, infrastructure, institutional efficiency as well as local indigenous practices, knowledge, and experiences [10,11]. therefore, factors that influence climate change risk response are diverse. owing to the interaction between behavior and perception. it is believed that the factors that impact climate change risk response perception are also diverse. this shows that structural equation model is suitable to deal with multiple factors affecting climate change risk response perception simultaneously. in this regard, some scholars have done extensive research. momtaz et al. investigated the factors affecting perception and adaptation behavior of farmers in response to climatic changes in hamedan. the findings indicated that knowledge, perception, and a r t i c l e i n f o article history received 23 december 2020 accepted 11 march 2021 keywords northern shaanxi southern shaanxi climate change risk perception structural equation modeling a b s t r a c t the public’s awareness of climate change risks is the basis for their choice of adaptation action. a good understanding of the key factors that affect the public’s perception of climate change risk is critical to climate change risk management. in this paper, a path model was constructed to analyze the path of climate change risk response perception in northern shaanxi based on 1660 public survey data in northern shaanxi, which was compared with that of southern shaanxi. the results showed that (1) there are three causal paths in northern shaanxi, that is, the public’s awareness of climate change issues, awareness of ecological stability, and awareness of climate change causes, to affect response status; there are nine causal paths in southern shaanxi. (2) there are four related routes in northern shaanxi and 19 in southern shaanxi. in short, compared with southern shaanxi, there are fewer perception paths and simpler models for climate change risk response in northern shaanxi. (3) the degree of concern for climate change issues and the perception of the causes of climate change influence the establishment of the causal path of climate change risk perception in northern shaanxi. the major factors that influence climate change risk response perception in southern shaanxi are climate change risk reason perception, industrial structure adjustment perception, and energy conservation, and emission reduction perception. (4) the response perception path in northern shaanxi is simpler than that in southern shaanxi, and there are fewer causal and related paths that impact climate change risk response perception. (5) finally, through the comparative analysis of the path of climate change risk response perception in northern shaanxi and southern shaanxi, this paper provides a reference for coping with climate change risks in northern and southern shaanxi. © 2021 the authors. published by atlantis press b.v. this is an open access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). *corresponding author. email: 1213268775@qq.com journal of risk analysis and crisis response vol. 11(1); april (2021), pp. 26–35 doi: https://doi.org/10.2991/jracr.k.210312.001; issn 2210-8491; eissn 2210-8505 https://www.atlantis-press.com/journals/jracr http://orcid.org/0000-0002-0809-5185 http://creativecommons.org/licenses/by-nc/4.0/ mailto:1213268775%40qq.com?subject= https://doi.org/10.2991/jracr.k.210312.001 https://www.atlantis-press.com/journals/jracr s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 27 belief had the maximum impact on the adaptation behavior, with path coefficients of, respectively, 0.53, 0.32, and 0.18, whereas belief and knowledge had the maximum impact on perception, with path coefficients of 0.56 and 0.35 respectively [12]. xue et al. [13] pointed out in exploring new ecological paradigms and coping with climate change in china, highly educated respondents showed a significantly stronger path between risk perception and behavior than less educated respondents. eriksson examined appraisals of threat (cognitive and emotional), personal resources (cost and self-efficacy), and strategies (response-efficacy) as predictors of proactive management responses (past behavior and future intention) among forest owners in sweden by means of a questionnaire (n = 1482), and found that threat appraisals and response-efficacy are direct predictors of past risk management behavior and the intention to respond in the future [14]. brown et al. studied the impact of cyclone evan in december 2012 on fijian households’ risk attitudes and subjective expectations about the likelihood and severity of natural disasters over the next 20 years, and pointed out the main factors that influence the perception of climate change risk response. their results showed that extreme event substantially changes individuals’ risk perceptions as well as their beliefs about the frequency and magnitude of future shocks [15]. in summary, most scholars believed that education, knowledge, experience and concepts are important in the perception path of climate change risk response. however, few have incorporated environmental and experiences factors into the climate change risk response perception path model at the same time. in fact, some have conducted research on the factors that influence climate change risk response perception from the perspectives of environment or experience. marlon et al. analyzed a representative statewide survey of floridians and compared their risk perceptions of 5-year trends in climate change with local weather station data from the 5 years preceding the survey. their research compared to local experience, risk perceptions of climate change were more strongly predicted by subjective experiences of environmental change, personal beliefs about climate change, and political ideology [16]. retchless used an interactive map of sea level rise in sarasota, florida and an accompanying online survey, it considers how college students from nearby and far away from sarasota, and with different views about climate change, vary in their risk perceptions. the results showed that, consistent with spatial optimism bias, risk perceptions increased more from preto post-map for respondents far away from sarasota than for those nearby [17]. nowadays, although domestic and foreign studies have achieved certain progress in the public’s climate change risk perception and its influencing factors, there are still the following shortcomings. first, most of the research subjects focused on investigating the single relationship between environment or experience and response perception, but failed to combine the two to systematically reflect the interaction between various factors and the impact mechanism of climate change risk response perception. moreover, the research was mostly conducted based on the opinions of peasants. northern shaanxi and southern shaanxi are important geographic regions in china, with relatively frequent meteorological disasters. comparing the research results of northern shaanxi with southern shaanxi can further highlight the perception path of climate change risk response in northern shaanxi, and provide a typical reference case for risk management and response. therefore, based on the structural equation model, this paper explored the path of climate change risk response perception in northern shaanxi and conducted a comparative analysis with southern shaanxi. this paper attempted to find the answers to the following questions: (1) how many paths are there to respond to climate change risk perception in northern shaanxi and southern shaanxi? how does it impact on people’s climate change risk response perceptions? (2) what reference can this regular pattern provide for people in northern and southern shaanxi to deal with the risk of climate change? 2. materials and methods 2.1. study area northern shaanxi is located in the northern part of shaanxi province, between 107°28¢ and 111°15¢ east longitude, and between 35°21¢ and 39°34¢ north latitude (figure 1). the loess hilly and gully area of northern shaanxi is in the middle reaches of the yellow river and the northern part of the loess plateau [18]. it borders gansu province and ningxia hui autonomous region in the west. it is adjacent to the inner mongolia autonomous region in the north and fu county, luochuan and yichuan counties in yan’an city in the south, covering 12 counties (districts) including yuyang district and dingbian county in yulin city, and pagoda district, ansai county, zichang county, yanchuan county, yanchang county, ganquan county, zhidan county, and wuqi county in yan’an [19]. northern shaanxi consists of two regions, yan’an and yulin. the former is a typical dry farming area, and the latter belongs to the agro-pastoral zone in the northern area of china. there are many meteorological disasters in the whole northern shaanxi region. drought, frost, rainstorm, gale, hail of varying degrees occur almost every year, among which drought, hail and frost are particularly serious [20]. the south of shaanxi is close to the qinling mountains in the north, and the bashan mountain in the south, with han river flowing from its west to east. the natural conditions of hanzhong and ankang in southern shaanxi have typical characteristics of the southern region. they are located at 105°30¢–110°01¢e and between 31°42¢ and 34°24¢n, as shown in figure 1. they have a humid climate in the northern subtropical zone, and most of mountains have a warm temperate humid climate. the shallow valleys in southern shaanxi are the warmest areas in the province, with temperatures mostly figure 1 | topography and geomorphology of the study area. 28 s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 ranging from 14 to 15°c. the average temperature in january, the coldest month, is 0–3°c, and the average temperature of july, the hottest month, is 24–27.5°c. the annual precipitation is 700–900 mm. there are many flood disasters in southern shaanxi, and the rainy season is in the autumn, which generally lasts from early and late to mid-early september. the main meteorological disasters there are summer drought, heavy rain, continuous rain, hail, frost, strong wind, cold wave etc. [20]. 2.2. data sources the questionnaire data came from a random sampling of public in northern shaanxi. a total of 1660 valid questionnaires were received, and the response rate was 80%. among the respondents, 829 were male, accounting for 49% of the total, and 831 were female, accounting for 51% 412 were at the age of 20 or below, accounting for 24.8% of the total; 629 aged 21–30, accounting for 37.9%, and 248 aged 31–40, accounting for 14.9% [21]. there were 186 respondents aged 41–50, accounting for 11.2% [21], 168 aged 51–70, accounting for 10.1%, and 24 aged over 70, accounting for 1.4%. in this survey, the data of the shaanxi provincial. statistical yearbook (2018) were used in the design of the population structure of the respondents, and appropriate adjustments were made based on the status of yan’an and yulin and large sample requirements. it is for us to consider the representativeness and validity of the sample as much as possible. table 1 shows other basic characteristics of the surveyed public [22]. the correlation coefficients between the perception of environment beauty and the living environment and risk concepts in northern shaanxi are 0.435 and 0.238 respectively, which are both significant at the level of 0.01. the correlation coefficient between risk perception and perception of environmental stability is 0.174. the correlation coefficients of the degree of concern for climate change issues with the perception of response situations and the perception of climate change causes are 0.245 and 0.149 respectively (table 2), both significant at the level of 0.01. therefore, the questionnaire indicators selected in northern shaanxi have a relatively significant correlation, which indicates that the public’s perception of climate change risks and response paths, and the content validity is high [24]. 2.3. research methods 2.3.1. construction of structural equation model based on the field survey in northern shaanxi and the analysis of the validity of the questionnaire [25], this paper proposes the following hypotheses, and constructs a path model of the role of risk concepts, living environment, and climate change information mastery on public climate change risk perception (figure 2). hypothesis h1: the public’s perception of climate change issues, perception of environmental stability, and perception of the causes of climate change affect the response status [26]. hypothesis h2: risk perceptions are positively correlated with the perception of living environment and environment beauty perception. hypothesis h3: the living environment and the perception of environment beauty perception are positively correlated. hypothesis h4: the degree of concern for climate change issues is positively correlated with the perception of the causes of climate change [27] (table 3). table 1 | basic characteristics of the surveyed public [22] survey item category frequency ratio (%) education elementary school or below [23] 408 29.30 junior high school [23] 284 17.10 high school [23]/technical secondary school 60 3.60 undergraduate/junior college [23] 278 16.70 postgraduate and above 553 33.30 monthly income 500 and below 870 52.40 500–1000 292 17.50 1001–2000 205 12.30 2001–3000 179 10.70 3001–5000 114 6.90 profession agriculture, forestry, animal husbandry and fishery 229 13.70 production and transportation 53 3.1 business services 177 10.1 government institutions 28 1.6 expertise 173 10.4 doctors 135 8.5 teachers 557 33.5 soldiers 132 7.9 self-employed people 99 5.9 students 33 1.9 table 2 | kmo value and bartlett test in northern shaanxi kaiser– meyer–olkin measures sampling suitability kmo value and bartlett test in northern shaanxi kaiser–meyer–olkin measures sampling suitability bartlett’s sphere test 0.678 approximately chi-square 1272.646 df 406 significance 0.000 figure 2 | the impact mechanism model of public climate change risk perception in northern shaanxi (hypothetical model). s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 29 2.3.2. variable selection and descriptive statistics table 4 is the descriptive statistics of independent variables and dependent variables of structural equation model. in addition, it also includes specific questionnaire items corresponding to different indicators. 2.3.3. path analysis in order to identify the key factors that affect the perception of climate change risk in southern shaanxi and the path of these factors, this paper uses path analysis to construct a path map and calculates the effect value (including overall effect, direct effect and indirect effect) in the amos26.0 environment [28]. in the structural equation model, the structural model between latent variables with only one observation variable is called path analysis. it is used to test the accuracy and reliability of the hypothetical causal model, the strength of the causal relationship between the measured variables, and it can accommodate the multi-link causal structure and use a path diagram to express it [29]. the basic expression is: h h x z= + +b γ where x is the exogenous variable matrix [30], h is the endogenous variable matrix [30], b is the structural coefficient matrix that represents the influence between the constituent factors of the endogenous variable matrix h, γ is the structural coefficient matrix [31], which represents the influence of the exogenous variable matrix x on the endogenous variable matrix h [31], and z is the residual matrix which represents the unexplained part [31]. 3. result analysis 3.1. model fit test in amos 26.0 environment, path model framework is established and calculated, original path is debugged according to model correction prompts, and the final model of northern shaanxi is determined (figure 3). when response path model freedom degree in northern shaanxi is 9, its chi-square value is about 9.312. the corresponding significance figure 3 | the impact mechanism model of public climate change risk perception in northern shaanxi (standard model). table 3 | correlation coefficient matrix of climate change risk perception in northern shaanxi index understanding the reasons of climate change coping situation scenic beauty perception environmental stability awareness living environment −0.035 0.07 0.435** 0.062 risk concept 0.01 0.075 0.238** 0.174** concern about climate change 0.149** 0.245** 0.069 0.076 **represents significant at the 0.01 level. table 4 | description of explanatory variables in northern shaanxi variables measurement standard assignment mean standard deviation living environment regional climate comfort c72 strongly agree = 1; agree = 2; uncertain = 3; disagree = 4 2.98 1.078 severe surrounding pollution c73 strongly disagree = 5; strongly agree = 1; agree = 2; uncertain = 3; disagree = 4; strongly disagree = 5 2.694 1.16 regional environmental livability c74 strongly agree = 1; agree = 2; uncertain = 3; disagree = 4; strongly disagree = 5 2.665 0.983 risk concept risk perception b1 strongly agree = 1; agree = 2; uncertain = 3; disagree = 4; strongly disagree = 5 3.2 1.86 risk option b3 there is 80% chance of getting 4000 yuan, 20% chance of getting nothing = 1, 100% chance of getting 3000 yuan = 2 1.611 0.569 understanding the causes of climate change evaluation of causes of climate change c91 natural reasons humanistic reasons = 1–7 4.925 2.073 understanding the causes of climate change c81, c82, c83 very well understanding = 1; relatively understanding = 2; general = 3; not very understanding = 4; not at all = 6 2.291 0.861 concern about climate change issues degree of concern for climate change issues d1 very concerned = 1; more concerned = 2; general = 3; not very concerned = 4; very unconcerned = 5 2.134 0.876 coping situation awareness climate change event participation status d6 very willing = 1; more willing = 2; unclear = 3; reluctant = 4; very unwilling = 5 1.958 0.967 daily coping behavior d7 always = 1; sometimes = 2; not sure = 3; rarely = 4; never = 5 1.99 1.042 scenic beauty perception scenic beauty recognition c71 strongly agree = 1; agree = 2; unsure = 3; disagree = 4; strongly disagree = 6 3.112 1.142 environmental stability awareness environmental stability awareness b2 the natural world is fragile, even a small change can cause catastrophic consequences = 2 2.982 0.919 the natural world is very stable, even if it is greatly disturbed, it can be restored to its original state = 4 30 s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 probability p = 0.811 > 0.05, which does not reach the significance level of 0.05. in addition, the ratio of chi-square freedom degree (cmin/df) is 0.665 < 2; rmsea value is 0.000 < 0.050; the gfi, agfi, ifi, tli, cfi values are 0.993, 0.985, 1.050, 1.081, and 1.000 respectively, all of which are over 0.900, complying with the standard. the preset model’s aic, bcc, bic, caic, ecvi values are all smaller than those of independent model and saturation model, indicating that the hypothetical model fits well with actual data (table 5). 3.2. analysis of results in northern shaanxi test results show that the overall effect of climate change reason perception, climate change problems concern degree, public’s environmental stability perception and public response status perception is 0.217, 0.200, and −0.18 respectively. furthermore, the direct effects are 0.217, 0.200, and −0.18, respectively. the direct effects of the degree of concern for climate issues and the perception of the causes of climate change on the situation are significant at the 0.01 level (figure 3). this shows that the environmental stability perception, climate change reasons perception, and concern degree for climate change issues have a significant positive impact on climate change response perception [32]. it is believed that hypothesis of h1 is valid. in contrast, climate change issues concern degree has a greater impact than the above two (table 6) [33]. as for correlation path in northern shaanxi, risk concern is positively correlated with living environment and environment beauty perception, with covariances of 0.137 and 0.203, respectively, assuming h2 holds. among them, the covariance of risk concepts and living environment, beautiful scenery perception is significant at the level of 0.01, which is inferred to be related to the fragile geographical environment in northern shaanxi. further covariance analysis of living environment and scenic beauty perception is 0.362, among which relationship with scenic beauty perception is significant at the level of 0.01, assuming h3 holds. moreover, the covariance between concern degree of climate change issues and perception of climate change reasons is 0.121, significant at the level of 0.05. therefore, h4 is confirmed. this shows that the better the living environment in northern shaanxi, the stronger risk concept and environmental beauty perception. the higher the concern degree of climate change issues, the better the perception of climate change reasons [34] (table 7). 3.3. comparative analysis the path model of public climate change risk response perception in northern shaanxi was constructed based on risk concepts, living environment, and concern for climate change issues. the climate change risk response path model in southern shaanxi was constructed based on risk concepts, human and land concepts, cultural level, living environment, and concern degree for climate change issues, and they have all passed test. it is inferred that in northern shaanxi region, due to the relatively harsh environment, conservative ideological concepts, serious soil erosion, and frequent disasters, education degree has a smaller impact on climate change risk response perception [35]. instead, concern degree for climate change issues and climate change reason perception influence the causal path of climate change risk perception [36]. in southern shaanxi, the mountains and rivers are beautiful, so it is less hit by natural disasters. therefore, climate change result perception, human and land concepts, risk concepts, educational level, and concern degree for climate change issues impact the establishment of climate change risk perception’s causal path in southern shaanxi. in addition, northern shaanxi is dominated by the secondary industry, whereas southern shaanxi is dominated by the primary and tertiary industries (figure 4). according to research by relevant scholars, the tertiary industry can break through hu huanyong line [37], so industrial structure adjustment perception in southern shaanxi has a significant impact on climate change response perception [26]. from figures 3 and 4, it can table 5 | index parameters of model adaptation in northern shaanxi evaluation index preset model saturation model independent model cmin/df (relative chi-square) 0.665 5.12 rmsea 0 0.107 gfi 0.993 1 0.915 agfi 0.985 0.887 ifi 1.05 1 0 cfi 1 1 0 tli 1.081 0 aic 37.312 56 121.511 bcc 37.947 57.269 121.829 bic 91.795 164.966 148.753 caic 105.795 192.966 155.753 ecvi 0.103 0.155 0.337 table 6 | overall effect, direct effect, and indirect effect among variables reason variable result variable overall effect direct effect indirect effect climate change reason perception coping situation perception 0.217 0.217 0 concern degree for climate change problems 0.200 0.200 0 environmental stability perception −0.108 −0.108 0 table 7 | climate change risk perception covariance matrix variables index estimate se cr p living environment← → risk concept 0.137 0.032 4.319 *** living environment← → scenic beauty perception 0.363 0.049 7.424 *** scenic beauty perception← → cimate change reason perception 0.121 0.043 2.787 0.005 risk concept← → scenic beauty perception 0.203 0.050 4.096 * *, ***represents significant at the 0.05 and 0.001 level. s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 31 figure 4 | the impact mechanism model of public climate change risk perception in southern shaanxi (standard model). table 8 | linear regression analysis results (n = 24) constant nonstandardized coefficient standard error normalized coefficient t p vif r2 adjusted r2 f –0.243 – 0.104 – –2.334 0.030* – 0.864 0.843 (3,20) = 42.217, p = 0.000 longitude 0.28 0.173 0.171 1.615 0.122 1.643 latitude 0.813 0.203 0.613 4.002 0.001** 3.444 altitude 0.42 0.188 0.314 2.24 0.037* 2.887 dependent variable: mms_ganzhi. d-w (durbin-watsonstatistic) value: 1.248. *p < 0.05, **p < 0.01. be seen that there are three causal paths in northern shaanxi: public’s of climate change issues concern degree, environmental stability perception, and climate change reason perception influence climate change response perception. there are nine causal paths in southern shaanxi, namely, climate change consequences perception, human and land concept [38], cultural level for climate change issues concern degree and industrial structure adjustment perception impact on climate change response status perception; public human-land and risk concept influence climate change response perception via impact on of climate change reasons perception; human-land and risk concept influence climate change reason perception. as for related routes, there are four in northern shaanxi and 19 in southern [39]. in short, compared with southern shaanxi, there are fewer perception paths and simpler models of climate change risk response perception in northern shaanxi. 3.4. analysis of influencing factors in order to explore the factors influencing climate change risk response perception in different counties and regions, and to reveal the mechanism of differences in climate change risk response perception path in northern and southern shaanxi, latitude, longitude and average altitude of each county were taken as independent variables, and climate change risk response perception intensity of each county as dependent variable for linear regression analysis. as shown in table 8, linear regression model r squared is 0.864, indicating a high fitting degree of model. further analysis of data in table 8 shows that latitude and altitude are the most influential factors on climate change risk response perception, with regression coefficients of 0.203 and 0.188 respectively, significant at the levels of 0.01 and 0.05 respectively. this result demonstrates that the greater the differences in terrain and latitude, the greater the difference of climate change risk response perception intensity, which probably leads to difference in paths (table 8). 4. discussion domestic and foreign studies have also confirmed that the environment and people’s experience will influence perception of climate change risk response [40]. for example, bradley et al. believed that antecedent psychological and socio-demographic variables predict climate change risk perceptions, which lead to enhancing levels of response efficacy and psychological adaptation in relation to climate change, and ultimately to environmentally-relevant behaviors [41]. the study found that: risk perception (hot), response (both hot and direct) and psychological adaptation (directly) predicted behavior [41]. smith provided some ground-breaking work on human behavior as it relates to perception and response to risks associated with climate change and climatic variability in the rural communities of sandy bay and fancy. the study examined households’ knowledge and perception of the climate change phenomenon and their responses to climate-related events. the results showed that an investigation of responses or the decision to respond to some of the impacts that they have experienced as a result of climate change and climatic variability leads to the development of different types of perceptions, including religious, ill informed, experienced-based, and knowledge-based perceptions. it is argued here that these forms of perception may result in non-adaptive, proactive or reactive adaptive behavior [42]. after studying farmers’ response to and perception of climate change risks, wang et al. [43] believed that extreme climate changes such as rising temperature, decreased precipitation and increased frequency of drought would affect farmers’ perception and response to climate change. in the hutt valley, new zealand et al., through a family survey, as well as seminar and interviews with local government officials, found that flood experience can influence flood risk perceptions, and that flood experience can stimulate increased risk reduction and adaptation actions where climate change risks are likely to occur. it is argued here that these forms of perception may result in non adaptive or reactive adaptive behavior. these studies have confirmed the rationality of using the two major variables of environment and concept to design the climate change risk response pathway model in northern shaanxi and southern shaanxi [44]. to verify the reliability of results of this paper, the climate change risk response perception path model of various cities in northern shaanxi (figure 5) and southern shaanxi (ensure rmse = 0) is calculated. it is found that climate change risk response perception path of yulin and yan’an in northern shaanxi is much simpler than that in southern shaanxi (figure 6). the climate change risk 32 s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 response path in northern shaanxi has four factors included in the model, while southern shaanxi has at least five. finally, geographic detectors are utilized to investigate the factors affecting the perception of climate change risk response in northern shaanxi and southern shaanxi respectively. it is found that in northern shaanxi, the explanatory power of each factor on the perception of climate change risk response is: education level > risk concepts > climate change reason perception > living environment > environmental stability perception > industrial structure adjustment perception > scenic beauty perception = climate change problems perception = energy saving and emission reduction perception, as shown in figure 7). this shows that impact of industrial structure adjustment, energy conservation and emission reduction perception on climate change risk response perception is not much different from that of environmental stability and grace perception. thus, the above factors can be substituted for each other, but they cannot be incorporated into the model of climate change risk response perception in northern shaanxi. this shows that climate change risk response perception path in northern shaanxi is not a complete mediation model, which is more consistent with the conclusions drawn by song and shi [45]. as for the climate change risk response perception paths in southern shaanxi, most of them are fully intermediary or partial intermediary models, and there is no non-intermediary model (figure 6). figure 7 shows the explanatory power of climate change risk response perception factors from small to large. it can be found that energy conservation, emission reduction perception, and industrial structure adjustment in southern shaanxi have greater explanatory power to climate change risk response perception than environmental stability or beautiful scenery perception. therefore, it is believed that energy conservation and emission reduction in southern shaanxi, climate change reasons, and industrial structure adjustment perception are three important intermediary variables that influence their perception of climate change risk response. the view that climate change risk perception path model in southern shaanxi is more complicated can be empirically proved by zhou, who demonstrated that public in hanzhong area influences their perception and response to climate change risks through their perceptions of reasons, knowledge, facts and consequences, which in turn influence their behavior and willingness to climate change risks response [46]. the above discussions indicate that the path of climate change risk response perception in northern shaanxi is simpler than that in southern shaanxi, and corresponding influencing factors are also less. the following conclusions can be drawn from the above discussions. first, the main influencing factors of climate change risk response perception in northern shaanxi [47] are climate change reason perception and climate change issues concern degree. second, figure 5 | a perceived path model for climate change risk response of all cities in northern shaanxi. figure 6 | a perceived path model for climate change risk response of all cities in southern shaanxi. s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 33 northern shaanxi region should increase basic network platforms construction to strengthen publicity of climate change risk information. third, regarding complex perception path in southern shaanxi to deal with climate change risks [48], people’s path of climate change risks response perception is diverse. for this reason, a well understanding of the intermediary variables in climate change risk response perception path model is necessary. fourth, because industrial structure adjustment, climate change reason perception and climate change problems concern degree are important variables in the climate change risk response perception path model, it is necessary to vigorously promote development of tertiary industry in southern shaanxi and understanding of climate change risk information and reasons. in terms of demographic factors, education level, monthly income and age in northern shaanxi have greater explanatory power for climate change risk response perception, and can be considered for inclusion in model in the future. in addition to education level in southern shaanxi, age also has a greater influence on climate change risk response perception. therefore, it is necessary to strengthen exploration of age on climate change risk response perception to reduce systematic errors induced by the model. the reasons for the difference in climate change risk response perception path in northern and southern shaanxi have been well explained in subsection 3.4. the facts that vertical difference in topography in southern shaanxi is more significant than in northern shaanxi, and that they are located in the southern and northern parts of the qinling mountains respectively, further confirm that climate change risk response perception in northern shaanxi is simpler than southern. the contribution of this paper is mainly reflected in the flowing aspects. first, this paper combines the environment and public experience to explore factors influencing the risk perception of climate change. in addition, the public’s perception and experience of risk is divided into two measuring dimensions, which is more innovative than the previous psychology measurement paradigm. second, this paper, by comparing the two regions of northern shaanxi and southern shaanxi, provides a more typical case for public climate change risk management. finally, most scholars tend to study on people’s climate change risk response behavior, whereas this paper directly investigates the path and factors of climate change risk response perception [49], with a better design of the research plan. nevertheless, it should be pointed out that this research has a small problem in the selection of indicators for the perception of climate change risk response. that is, the indicator of the living environment needs further improvement although it can replace the objective environment where people live. for example, temperature and precipitation can be used to replace the indicator of the living environment. there are less paths in northern shaanxi than in southern shaanxi. previous studies have shown that in the hanzhong city in southern shaanxi, age, occupation, education level, income level and public perceptions of climate change knowledge, facts, and reasons perception, perception of consequences, willingness to respond, and response behavior have varying degrees of influence [50]. therefore, the paths that affect the perception of climate change risk in southern shaanxi are diverse. some scholars analyzed the adaptation behaviors and influencing factors of peasants in the hilly loess regions of northern shaanxi and concluded that peasants’ adaptation behaviors are affected by the perception of climate change (figure 7). in addition [51], family socioeconomic attributes have a significant impact on the probability of peasants’ adaptation behaviors, while other attributes such as age and education level are independent of the probability of farmers adopting adaptive behaviors [52] (figure 8). figure 8 | the explanatory power of demographic factors in shaanxi. figure 7 | detection of impact factors in shaanxi. 34 s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 5. conclusion based on the research purpose proposed in the introduction part and the results of the discussion part, the following conclusions can be drawn. firstly, there are three causal paths in northern shaanxi, that is, the public’s awareness of climate change issues, awareness of ecological stability, and awareness of climate change causes, to affect response status. there are nine causal paths in southern shaanxi. secondly, there are four related routes in northern shaanxi and 19 in southern shaanxi. in short, compared with southern shaanxi, there are fewer perception paths and simpler models for climate change risk response in northern shaanxi. thirdly, the degree of concern to climate change issues and the perception of climate change causes affect the establishment of the causal path of climate change risk perception in northern shaanxi. fourthly, the related paths of climate change risk perception in northern shaanxi can be summarized into the following two: the better the living environment, the stronger the risk perception of places; the higher the degree of concern for climate change issues, the better the perception of the causes of climate change. finally, according to the above conclusions, we put forward the following suggestions for northern and southern shaanxi to deal with the risks of climate change. northern shaanxi and southern shaanxi should be different in managing climate change risk. northern shaanxi should strengthen advocacy on the causes of climate change, and southern shaanxi should strengthen publicity on the effects of climate change risk response and increase adjustment of industrial structure, and at the same time, actively carry out energy conservation and emission reduction activities to promote climate change risk response. conflicts of interest the authors declare they have no conflicts of interest. acknowledgments this work is supported by the national natural science foundation of china “regional climate change risk perception and response” (41771215). the authors appreciate the time and effort of the editors and reviewers in providing constructive comments which have helped to improve the manuscript. references [1] shen yp, wang gy. key findings and assessment results of ipcc wgi fifth assessment report. j glaciol geocryol 2013;35:1068–76. [2] solomon s, qin d, manning m, marquis m, averyt kb, tignor m, et al. climate change 2007: the physical science basis. working group i contribution to the fourth assessment report of the ipcc. cambridge, uk: cambridge university press; 2007, pp. 3–53. [3] grüneis h, penker m, höferl km. the full spectrum of climate change adaptation: testing an analytical framework in tyrolean mountain agriculture (austria). springerplus 2016;5:1848. [4] fu s, zou j, zhang x, qi y. review on the latest conclusions of working group iii contribution to the fifth assessment report of the intergovernmental panel on climate change. chin j urban environ stud 2015;3:1550005. [5] mackay a. climate change 2007: impacts, adaptation and vulnerability. contribution of working group ii to the fourth assessment report of the intergovernmental panel on climate change. j environ qual 2008;37:2407. [6] somorin oa, visseren-hamakers ij, arts b, tiani am, sonwa dj. integration through interaction? synergy between adaptation and mitigation (redd+) in cameroon. environ plan c: politics space 2016;34:415–32. [7] salvia m, reckien d, pietrapertosa f, eckersley p, spyridaki na, krook-riekkola a, et al. will climate mitigation ambitions lead to carbon neutrality? an analysis of the local-level plans of 327 cities in the eu. renew sustain energy rev 2021; 135:110253. [8] feng q, yang l, deo rc, aghakouchak a, adamowski jf, stone r, et al. domino effect of climate change over two millennia in ancient china’s hexi corridor. nat sustain 2019;2:957–61. [9] walmsley dj, lewis gj, et al. introduction to behavioral geography [xingzhong w, guoqiang z, guicai l, trans]. xi’an: shaanxi people’s publishing house; 1988. [10] nkya k, mbowe a, makoi jhjr. low cost irrigation technology, in the context of sustainable land management and adaptation to climate change in the kilimanjaro region. j environ earth sci 2015;5:45–56. [11] onofri l, nunes pald. economic valuation for policy support in the context of ecosystem-based adaptation to climate change: an indicator, integrated based approach. heliyon 2020;6:e04650. [12] mahmoodi momtaz a, choobchian s, farhadian h. factors affecting farmers’ perception and adaptation behavior in response to climate change in hamedan province, iran. j agric sci technol 2020;22:905–17. [13] xue w, marks adg, hine dw, phillips wj, zhao s. the new ecological paradigm and responses to climate change in china. j risk res 2018;21:323–39. [14] eriksson l. the importance of threat, strategy, and resource appraisals for long-term proactive risk management among forest owners in sweden. j risk res 2016;20:868–86. [15] brown p, daigneault aj, tjernström e, zou w. natural disasters, social protection, and risk perceptions. world dev 2018;104:310–25. [16] marlon jr, van der linden s, howe pd, leiserowitz a, lucia woo sh, broad k. detecting local environmental change: the role of experience in shaping risk judgments about global warming. j risk res 2019;22:936–50. [17] retchless dp. understanding local sea level rise risk perceptions and the power of maps to change them: the effects of distance and doubt. environ behav 2018;50:483–511. [18] yang l, xie gd, zhen l, leng yf, guo gm. spatio-temporal changes of land use in jinghe watershed. resour sci 2005;27: 26–32 (in chinese). [19] zhao j, li c, yang t, tang y, yin y, luan x, et al. estimation of high spatiotemporal resolution actual evapotranspiration by combining the swh model with the metric model. j hydrol 2020;586:124883. [20] changyan w, yaochuang y. research on the temporal and spatial changes of climate change perception by people in southern shaanxi. chin agric sci bull 2015;31:244–51. [21] orczyk m, firlik b. the assessment of vibroacoustic comfort in trams on the basis of experimental studies and surveys. in: suchanek m, editor. new research trends in transport sustainability and innovation. springer: springer proceedings in business and economics; 2018, pp. 46–55. https://doi.org/10.1186/s40064-016-3542-1 https://doi.org/10.1186/s40064-016-3542-1 https://doi.org/10.1186/s40064-016-3542-1 https://doi.org/10.1142/s2345748115500050 https://doi.org/10.1142/s2345748115500050 https://doi.org/10.1142/s2345748115500050 https://doi.org/10.1142/s2345748115500050 https://doi.org/10.2134/jeq2008.0015br https://doi.org/10.2134/jeq2008.0015br https://doi.org/10.2134/jeq2008.0015br https://doi.org/10.2134/jeq2008.0015br https://doi.org/10.1177/0263774x16645341 https://doi.org/10.1177/0263774x16645341 https://doi.org/10.1177/0263774x16645341 https://doi.org/10.1177/0263774x16645341 https://doi.org/10.1016/j.rser.2020.110253 https://doi.org/10.1016/j.rser.2020.110253 https://doi.org/10.1016/j.rser.2020.110253 https://doi.org/10.1016/j.rser.2020.110253 https://doi.org/10.1016/j.rser.2020.110253 https://doi.org/10.1038/s41893-019-0397-9 https://doi.org/10.1038/s41893-019-0397-9 https://doi.org/10.1038/s41893-019-0397-9 https://doi.org/10.1016/j.heliyon.2020.e04650 https://doi.org/10.1016/j.heliyon.2020.e04650 https://doi.org/10.1016/j.heliyon.2020.e04650 http://jast.modares.ac.ir/article-23-30786-en.html http://jast.modares.ac.ir/article-23-30786-en.html http://jast.modares.ac.ir/article-23-30786-en.html http://jast.modares.ac.ir/article-23-30786-en.html https://doi.org/10.1080/13669877.2016.1200655 https://doi.org/10.1080/13669877.2016.1200655 https://doi.org/10.1080/13669877.2016.1200655 https://doi.org/10.1080/13669877.2015.1121905 https://doi.org/10.1080/13669877.2015.1121905 https://doi.org/10.1080/13669877.2015.1121905 https://doi.org/10.1016/j.worlddev.2017.12.002 https://doi.org/10.1016/j.worlddev.2017.12.002 https://doi.org/10.1080/13669877.2018.1430051 https://doi.org/10.1080/13669877.2018.1430051 https://doi.org/10.1080/13669877.2018.1430051 https://doi.org/10.1080/13669877.2018.1430051 https://doi.org/10.1177/0013916517709043 https://doi.org/10.1177/0013916517709043 https://doi.org/10.1177/0013916517709043 https://doi.org/10.1016/j.jhydrol.2020.124883 https://doi.org/10.1016/j.jhydrol.2020.124883 https://doi.org/10.1016/j.jhydrol.2020.124883 https://doi.org/10.1016/j.jhydrol.2020.124883 s. xue and z. qi / journal of risk analysis and crisis response 11(1) 26–35 35 [22] temelová j, novák j, jíchová j. safe life in the suburbs? crime and perceptions of safety in new residential developments in prague’s hinterland, czech republic. eur urban region stud 2016;23:677–96. [23] ke x, zhang l, li z, tang w. inequality in health service utilization among migrant and local children: a cross-sectional survey of children aged 0–14 years in shenzhen, china. bmc pub health 2020;20:1668. [24] damert m, feng y, zhu q, baumgartner rj. motivating low carbon initiatives among suppliers: the role of risk and opportunity perception. resour conserv recycl 2018;136:276–86. [25] feng s, yin j. the study on villa price volatility of shanghai with structural equation. 2012 international conference on management science & engineering 19th annual conference proceedings. dallas, tx, usa: ieee; 2012, pp. 1819–25. [26] azócar g, billi m, calvo r, huneeus n, lagos m, sapiains r, et al. climate change perception, vulnerability, and readiness: inter-country variability and emerging patterns in latin america. j environ stud sci 2021;11:23–36. [27] joyce re, hunt cl. philippine nurses and the brain drain. soc sci med 1982;16:1223–33. [28] kim s, lee je, kim d. searching for the next new energy in energy transition: comparing the impacts of economic incentives on local acceptance of fossil fuels, renewable, and nuclear energies. sustainability 2019;11:2037. [29] minglong wu. structural equation model-amos operation and application. chongqing: chongqing university press; 2009. [30] li w, wei xp, zhu rx, guo kq. study on factors affecting the agricultural mechanization level in china based on structural equation modeling. sustainability 2019;11:1–16. [31] sun h, fan z, zhou y, shi y. empirical research on competitiveness factors: analysis of real estate industry of beijing and tianjin. eng constr architect manage 2010;17:240–51. [32] kwon sa, kim s, lee je. analyzing the determinants of individual action on climate change by specifying the roles of six values in south korea. sustainability 2019;11:1834. [33] zahran s, brody sd, grover h, vedlitz a. climate change vulnerability and policy support. soc nat resour 2006;19:771–89. [34] lata s, nunn p. misperceptions of climate-change risk as barriers to climate-change adaptation: a case study from the rewa delta, fiji. climatic change 2012;110:169–86. [35] you m, ju y. interaction of individual framing and political orientation in guiding climate change risk perception. j risk res 2019;22:865–77. [36] cutler mj, marlon jr, howe pd, leiserowitz a. the influence of political ideology and socioeconomic vulnerability on perceived health risks of heat waves in the context of climate change. weather climate soc 2018;10:731–46. [37] li l, xueyan z, yaru w, qin z, bing x. research on farmers’ perception of climate change in ecologically fragile alpine region based on structural equation model-a case study of gannan plateau. acta ecol sin 2017;37:3274–85. [38] chengchao w, yusheng y, wen z, jing h, jianbin x. research on the perception and adaptation of domestic and foreign farmers to climate change/variation. geogr sci 2017;37:938–43. [39] yudong z, qi z, rong h, kepeng g. research on the relationship between public perception and response to climate change based on structural equation—a case study of hanzhong city, shaanxi province. henan sci 2019;37:429–38. [40] liu y. chapter 6 logistics development in shaanxi province. springer science and business media llc; 2014. [41] bradley gl, babutsidze z, chai a, reser jp. the role of climate change risk perception, response efficacy, and psychological adaptation in pro-environmental behavior: a two nation study. j environ psychol 2020;68:101410. [42] smith ra. risk perception and adaptive responses to climate change and climatic variability in northeastern st. vincent. j environ stud sci 2018;8:73–85. [43] wang j, kim s. analysis of the impact of values and perception on climate change skepticism and its implication for public policy. climate 2018;6:99. [44] lawrence j, quade d, becker j. integrating the effects of flood experience on risk perception with responses to changing climate risk. nat hazards 2014;74:1773–94. [45] zhen s, xingmin s. farmers’ adaptation behaviors to climate change and path analysis of influencing factors in rain-fed agricultural areas. adv geogr sci 2020;39:461–73. [46] yudong z, qi z, rong h, kepeng g. research on the relationship between public perception and response to climate change based on structural equations: taking hanzhong city, shaanxi province as an example. henan sci 2019;37:429–38. [47] becerra mj, pimentel ma, de souza eb, tovar gi. geospatiality of climate change perceptions on coastal regions: a systematic bibliometric analysis. geogr sustain 2020;1:209–19. [48] theory and practice of climate adaptation. springer science and business media llc; 2018. [49] elshirbiny h, abrahamse w. public risk perception of climate change in egypt: a mixed methods study of predictors and implications. j environ stud sci 2020;10:242–54. [50] echavarren jm, balžekienė a, telešienė a. multilevel analysis of climate change risk perception in europe: natural hazards, political contexts and mediating individual effects. saf sci 2019; 120:813–23. [51] terorotua h, duvat vke, maspataud a, ouriqua j. assessing perception of climate change by representatives of public authorities and designing coastal climate services: lessons learnt from french polynesia. front mar sci 2020;7:160. [52] lu w, xingmin s, lifan s. the adaptation behavior and influencing factors of farm households in the loess hilly and gully area of northern shaanxi. progr climate change res 2017;13:61–8. https://doi.org/10.1177/0969776414534254 https://doi.org/10.1177/0969776414534254 https://doi.org/10.1177/0969776414534254 https://doi.org/10.1177/0969776414534254 https://doi.org/10.1186/s12889-020-09781-4 https://doi.org/10.1186/s12889-020-09781-4 https://doi.org/10.1186/s12889-020-09781-4 https://doi.org/10.1186/s12889-020-09781-4 https://doi.org/10.1016/j.resconrec.2018.05.002 https://doi.org/10.1016/j.resconrec.2018.05.002 https://doi.org/10.1016/j.resconrec.2018.05.002 https://doi.org/10.1109/icmse.2012.6414419 https://doi.org/10.1109/icmse.2012.6414419 https://doi.org/10.1109/icmse.2012.6414419 https://doi.org/10.1109/icmse.2012.6414419 https://doi.org/10.1007/s13412-020-00639-0 https://doi.org/10.1007/s13412-020-00639-0 https://doi.org/10.1007/s13412-020-00639-0 https://doi.org/10.1007/s13412-020-00639-0 https://doi.org/10.1016/0277-9536(82)90147-2 https://doi.org/10.1016/0277-9536(82)90147-2 https://doi.org/10.3390/su11072037 https://doi.org/10.3390/su11072037 https://doi.org/10.3390/su11072037 https://doi.org/10.3390/su11072037 https://doi.org/10.3390/su11010051 https://doi.org/10.3390/su11010051 https://doi.org/10.3390/su11010051 https://doi.org/10.1108/09699981011038042 https://doi.org/10.1108/09699981011038042 https://doi.org/10.1108/09699981011038042 https://doi.org/10.3390/su11071834 https://doi.org/10.3390/su11071834 https://doi.org/10.3390/su11071834 https://doi.org/10.1080/08941920600835528 https://doi.org/10.1080/08941920600835528 https://doi.org/10.1007/s10584-011-0062-4 https://doi.org/10.1007/s10584-011-0062-4 https://doi.org/10.1007/s10584-011-0062-4 https://doi.org/10.1080/13669877.2017.1422785 https://doi.org/10.1080/13669877.2017.1422785 https://doi.org/10.1080/13669877.2017.1422785 https://doi.org/10.1175/wcas-d-17-0105.1 https://doi.org/10.1175/wcas-d-17-0105.1 https://doi.org/10.1175/wcas-d-17-0105.1 https://doi.org/10.1175/wcas-d-17-0105.1 https://doi.org/10.5846/stxb201603090415 https://doi.org/10.5846/stxb201603090415 https://doi.org/10.5846/stxb201603090415 https://doi.org/10.5846/stxb201603090415 https://doi.org/10.1016/j.jenvp.2020.101410 https://doi.org/10.1016/j.jenvp.2020.101410 https://doi.org/10.1016/j.jenvp.2020.101410 https://doi.org/10.1016/j.jenvp.2020.101410 https://doi.org/10.1007/s13412-017-0456-3 https://doi.org/10.1007/s13412-017-0456-3 https://doi.org/10.1007/s13412-017-0456-3 https://doi.org/10.3390/cli6040099 https://doi.org/10.3390/cli6040099 https://doi.org/10.3390/cli6040099 https://doi.org/10.1007/s11069-014-1288-z https://doi.org/10.1007/s11069-014-1288-z https://doi.org/10.1007/s11069-014-1288-z https://doi.org/10.1016/j.geosus.2020.09.002 https://doi.org/10.1016/j.geosus.2020.09.002 https://doi.org/10.1016/j.geosus.2020.09.002 https://doi.org/10.1007/s13412-020-00617-6 https://doi.org/10.1007/s13412-020-00617-6 https://doi.org/10.1007/s13412-020-00617-6 https://doi.org/10.1016/j.ssci.2019.08.024 https://doi.org/10.1016/j.ssci.2019.08.024 https://doi.org/10.1016/j.ssci.2019.08.024 https://doi.org/10.1016/j.ssci.2019.08.024 https://doi.org/10.3389/fmars.2020.00160 https://doi.org/10.3389/fmars.2020.00160 https://doi.org/10.3389/fmars.2020.00160 https://doi.org/10.3389/fmars.2020.00160 https://doi.org/10.12006/j.issn.1673-1719.2016.075 https://doi.org/10.12006/j.issn.1673-1719.2016.075 https://doi.org/10.12006/j.issn.1673-1719.2016.075 microsoft word volume 12, issue 4-5 journal of risk analysis and crisis response, 2022, 12(4), 222-234 https://jracr.com/ issn print: 2210-8491 issn online: 2210-8505 doi: https://doi.org/10.54560/jracr.v12i4.345 222 review summary of research on order financing decision optimization and risk management yi-fan fu 1,2 and mu zhang 1,* 1 school of big data application and economics, guizhou university of finance and economics, guiyang (550025), guizhou, china 2 guizhou institution for technology innovation & entrepreneurship investment, guizhou university of finance and economics, guiyang (550025), guizhou, china * correspondence: zhangmu01@163.com; tel.: +86-0851-88510575 received: october 25, 2022; accepted: december 25, 2022; published: december 31, 2022 abstract: to alleviate the pressure of financing for small and medium-sized enterprises in my country, order financing has become one of the effective ways to solve the financing difficulties of china's small and medium-sized enterprises. order financing is a kind of supply chain financing mode. compared with traditional accounts payable financing, order financing advances the financing link, accelerates the capital turnover efficiency, and better solves the urgent demand for funds of small and medium-sized enterprises. this paper reviews the research status of order financing from three aspects: order financing model evolution, decision optimization, and risk management. through combing the existing literature, it puts forward appropriate policy recommendations for the order financing model under the current development status, which will provide direction for the development of order financing in the future. keywords: order financing; model evolution; decision optimization; risk management; literature review 1. introduction order financing is a kind of supply chain financing mode. compared with traditional accounts payable financing, order financing will advance the financing link and speed up the capital turnover efficiency, to better solve the urgent demand for funds for small and micro enterprises. order financing is a business form of supply chain finance. order financing was initially applied in international trade, also known as package loan, which is a model in which financial institutions such as banks provide loan services to exporters with capital problems during the period from receipt of orders by exporters to shipment of goods ordered, based on orders received by foreign companies by domestic exporters. at present, our country is in a crucial stage of economic transformation and development. to achieve the goal of high-quality development, we must rely on the healthy development of small and medium-sized enterprises. in recent years, order financing, a financial innovation model, not only provides a new solution to the difficult problem of financing and expensive financing for small and medium-sized enterprises, but also expands the scope of banking financial services to a certain extent. as an innovative business in the supply chain financial model developed by banks and other financial institutions, order financing has brought new profit points for the financial industry. the risk brought yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 223 by the order financing model to related enterprises and banks also needs to be studied urgently. this paper will sort out the relevant literature on order financing from three aspects: model evolution, decision optimization, and risk management, and clarify the development context and existing problems of the order financing business. 2. research on the evolution of the order financing model from the perspective of financing methods, order financing is considered to be an effective way to solve the emergency loan financing of small and medium-sized enterprises in china. foreign countries have a relatively early understanding and research on order financing. for example, novembrino et al. (1996) [1] discussed order financing in international transactions through a case and analyzed the call signals sent by the lending institutions to the order financing institutions; marks et al. (2009) [2] believe that the traditional pof refers to a supplier's loan to a financial institution conditional on a purchase contract signed by the buyer. since 2008, some domestic scholars have started in logistics finance and opened a new era of order financing research in china. they believe that order financing, inventory pledge financing, and accounts receivable financing are the three major forms of supply chain finance. based on the risk analysis model of order financing business in the value chain process, this paper analyzes the basic structure of logistics finance in depth and considers that order financing business is an effective financing arrangement to solve the financing difficulties of small and medium-sized enterprises in the supply chain (li yixue and wu lihua, 2008) [3]. liu yaya and zeng youxin (2009) [4] found that suppliers can apply for loans from financial institutions with the orders of core manufacturers in the supply chain as a guarantee. after the financial institutions review the authenticity of the orders and evaluate the value of the orders and the corresponding risks, they can grant certain credit lines to help enterprises purchase the necessary raw materials for production. this model not only has a flexible and fast market commodity information collection and feedback system, but also has a perfect logistics management information system. in the research on the credit guarantee model of small and medium-sized enterprises, the "1+n" supply chain financing model is adopted for enterprises that do not meet the bank pledge requirements and have insufficient individual credit, and the seller's credit is enhanced through the third-party guarantee model (zhao jian and huo jiazhen, 2009) [5]. li juan (2010) [6] research found that the adoption of stage loans can make the relationship among banks, logistics enterprises, and small and medium-sized enterprises reach a balanced state under the condition of incomplete information, produce appropriate and reasonable quantities of products, and complete the normal production and sales of products. through the impact on profit, profit margin, and decision-making caused by the numerical changes of order financing cost, order financing interest, and order financing value ratio, logistics enterprises are promoted to make better choices in the process of order financing business, control their costs and obtain more market profits (hong yitian, 2020) [7]. the "three rural issues" are the basic issues in the overall development of china's reform and opening-up. li han (2014) [8] said that by establishing a loan guarantee fund pool to further expand and improve the loan guarantee insurance business, the ultimate goal is to establish an "order + futures + shortage" order finance model. lan qinggao (2014) [9] through analysis shows that it is concluded that order financing solves the problem of shortage of funds for farmers upstream of the dairy supply chain. during the implementation of breeding, the existing breeding capacity and selfyi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 224 owned fund status must be inspected to reasonably determine the financing proportion and financing period. wu haixia (2015) [10] proposed to apply for loans to three types of subjects, namely, farmers, leading agricultural enterprises, and farmers' professional cooperatives. the financing mode is generally "order + credit evaluation + guarantee" (joint guarantee, mortgage, or guarantee insurance). to address the concerns of the acquisition enterprises about the price risk of order financing, shouguang city has introduced an insurance mechanism to obtain loans in the form of guarantee insurance by insurance companies, which greatly broadens the applicability of "order financing". wang tingrui and chu yongze (2015) [11] proposed a centralized credit model for agricultural order financing, forming a bundled credit model. the "bank + company + agricultural cooperative organization + farmers + insurance + trust" and other models formed by agricultural order financing transform decentralized loans into batch loans through upstream and downstream connections, which is beneficial to the intensive and large-scale operation of agricultural credit, thus realizing mutual benefit and win-win for all parties. taking the order-based agriculture in longhai city as an example, a multi-link and multi-organization order-based financing model of fresh agricultural products supply chain, namely "bank + core processing enterprises + cooperative organizations + farmers + futures market + agricultural insurance + government", has been formed. it introduces the agricultural insurance and futures markets, ensures the effective operation of order-based financing with the help of its financial guarantee function, solves the financing problem of farmers, stabilizes the supply of high-quality goods for the core enterprises, and expands the credit scope for the agricultural bank (song equality, 2016) [12]. taking shouguang city in shandong province as an example, under the "company + farmers + order financing" model, farmers' production behavior meets the requirements of the consumption environment on the quality of agricultural products and plays an important demonstration role in guiding farmers' production from output to quality (hu huiyong, 2020) [13]. based on defining the concepts and relevant theories of the pig supply chain and order financing model, according to the basic logic of the "demand, supply, and demand model", this paper sorts out the financing status of pig breeding subject from the aspects of fund demand, financial supply, and traditional financing model, and analyzes the financing difficulties of pig breeding subject from three aspects of industrial scale process, fund supply channel and credit evaluation display degree. finally, it concludes that the order financing model of pig breeding can alleviate the difficulty of pig breeding subject to a certain extent (wang xi, 2021) [14]. based on the agricultural supply chain and the order financing model of "farmers + cooperatives + banks + core enterprises" of agricultural cooperatives, combined with the financing difficulties of chinese herbal medicine planting in yunnan province, this paper analyzes the financing status of chinese herbal medicine growers in yunnan province, and combining with the application situation in yunnan province, puts forward specific suggestions for promoting their development (wang caifen and li fuchang, 2021) [15]. by constructing the theoretical framework of order financing to increase the income of largescale farmers, this paper uses the generalized structural equation model method based on the zeroexpansion model to conduct the empirical test, which provides the theoretical basis and empirical experience for strengthening the agricultural order financing model and risk prevention (jiang boheng and wen tao, 2022) [16]. with the development of e-commerce, network distributors, which are the intermediate distributors with the network as the main distribution channel, have begun to appear, and the network distributors are mainly small and micro enterprises. liang fei (2015) [17] proposed that the yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 225 difference between the online order financing model and the traditional order financing model is that the traditional order financing model is to finance small and medium-sized enterprises according to the orders of the core enterprises in the supply chain, while the online order financing model does not depend on the core enterprises, and enterprises directly finance from banks according to the orders of customers. xu liqi (2016) [18] proposed electronic order financing for internet supply chain finance, which refers to a business model in which upstream financing enterprises (suppliers) and downstream core manufacturers sign real and effective electronic orders in a third-party b2b ecommerce platform, with which suppliers apply to commercial banks for financing, which can be seamlessly connected with factoring financing or bill discounting. rebecca (2017) [19] actively explored the incentive mechanism, game mechanism, and financing strategy of the electronic order financing model, and put forward effective measures for financing strategies of e-commerce platforms and risk control for banks to carry out electronic order supply chain financing, which provided theoretical guidance for optimizing the financing process of small and medium-sized enterprises and promoting banks to efficiently carry out electronic order financing services. huang yingcan (2018) [20] proposed that the advantages of the supply chain model based on b2b platform over the traditional supply chain finance model are not only in online credit, but also in the ability to obtain a real large amount of basic transaction information; providing various business flow, logistics, and information flow data; diversifying sources of pooled financing; cost savings. based on the three-dimensional trust, the seller's electronic order financing can enhance the bank's trust perception, and a green supply chain seller's electronic order financing transaction model based on the threedimensional trust is proposed to alleviate the manufacturer's financing demand in the green supply chain (yin qin, 2019) [21]. the research shows that the supply chain seller's electronic order financing model under the consideration of institutional trust improves the compliance revenue of the participants, effectively promotes mutual trust among the financial participants in the supply chain, and promotes each participant to establish a long-term and stable strategic cooperation relationship (yang qifeng et al., 2019) [22]. through the internet platform set up by the government, the government has made innovative research on the new mode of internet order financing, made full use of the big data and credit reporting system, introduced partners such as banks and commercial insurance, and made the cooperation orders between enterprises circulate and trade online. compared with the traditional financing mode, it has the advantages of strengthening the credit reporting system, enabling information technology, innovating the evaluation system, dynamic risk control, risk sharing among three parties, and expanding the scale of financing (li haocheng et al., 2021) [23]. 3. research on order financing decision optimization to solve the financial difficulties in the early production process, the enterprise generally adopts internal financing or external financing. internal financing refers to applying for deferred payments from upstream enterprises or accepting advance payments from downstream enterprises, while external financing refers to applying for loans from financial institutions. however, financing in the procurement period takes a long time to repay, and financing enterprises need to pay higher loan interest. the study found that financing companies applying for order financing and suspending payment to upstream suppliers at the same time, and reasonably setting the order quantity and wholesale price of the core companies can alleviate the shortage of funds for suppliers and retailers yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 226 to a large extent, and improve the supply chain performance (zhang xin et al., 2015) [24]. zhao ruixue (2018) [25] studied order financing to reverse the factoring financing model in supply chain finance. combining internal and external financing, considering whether a supplier accepts a retailer's order under different reverse factoring rates, different optimal strategies resulting from the game between the retailer and the supplier were obtained. the profit of financial order financing participants in the supply chain is affected by various factors. from the perspective of financing enterprises, it is found that if the current price of the retailer's capital is higher than the risk-free interest rate, then the bank interest rate will be somewhere between the two, suppliers choose order financing, and the enterprise's profit is larger (wang zongrun and shi jiaxing, 2018) [26]. combining the supply chain finance model of order financing with the supply chain coordination model of the call option contract, and introducing the "call option contract model under no financing model" to compare, this paper discusses the impact of the option strike price and the supplier's initial capital under the three models, to make the profit distribution between the supplier and the retailer change (xu qing, 2016) [27]. combining with the buyer's electronic order financing, this paper constructs a three-dimensional trust mechanism from three aspects of the system, personality, and reputation, and uses the method of game theory to establish a mathematical model of buyer's electronic order financing based on three-dimensional trust perception. the research results show that the three-dimensional trust mechanism enhances the banks' trust perception in the unfamiliar network environment and improves their willingness to participate in the financing. the interest rate decision-making formula of a bank is its trust perception function. when the bank senses a higher degree of trust through rational calculation of threedimensional trust variables, the interest rate it decides will be lower. accordingly, the higher the optimal product quality level decided by the supplier, the larger the optimal order quantity decided by the retailer (wang shuoxin, 2019) [28]. research by yin qin (2019) [21] shows that the threedimensional trust mechanism of the green supply chain can promote banks' perception of trust in the green supply chain and reduce their loan interest rates. institutional trust and personality trust as exogenous trust variables in three-dimensional trust mainly affect banks' perception of trust in the green supply chain by positively affecting reputation trust as decision variables, thus affecting manufacturers' and retailers' transaction decisions, and further affecting the overall profit of green supply chain. based on the analysis of the advantages of the order financing model of pig breeding from three aspects of fund demand, financial supply, and traditional financing model, the study found that when the initial capital level of the pig breeding subject is low, the profit situation of financial institutions under the traditional commercial financing business and order financing business is quite different, and the profit of financial institutions increases with the increase of the initial capital level of pig breeding subject (wang xi, 2021) [29]. the decision of each participant in order financing directly determines the effectiveness of supply chain finance. through numerical analysis, the optimal operation strategy of the supply chain order financing model under the control of a bank risk cap is studied. the impact of bank risk caps on the optimal decision of suppliers and retailers is analyzed (wang wenli et al., 2013) [30]. under the assumption of bank risk neutrality and random demand, the overall decision-making and bank decision-making strategies of the supply chain finance order financing model are studied. it is found that the overall maximum revenue of supply chain finance is only related to the order quantity, and the decision of bank loan interest rate is related to risk-free revenue, supplier's capital demand, and yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 227 default rate of core enterprises, which decrease with the increase of supplier's capital demand and increase with the increase of risk-free interest rate (xu liqi, 2016) [18]. song ping et al. (2018) [31] apply the newsboy model and the stackelberg game to construct the expected profit model for financing participants. the higher the reputation level determined by suppliers, the higher the retailer's order quantity, the higher the bank's trust level, and the lower the interest rate. decentralized decision-making cannot achieve the coordinated results of financial supply chains. taking electronic order financing as the starting point, this paper discusses the impact on the revenue and decision-making behavior of both parties from the fixed fee, royalty fee, and mixed fee respectively, and discusses the optimal financing decision-making problem when the b2b platform, enterprise, and bank obtain the maximum revenue under the entrusted credit model, joint credit model and separate credit model (huang yingcan, 2018) [20]. wang ying and chen zhi (2021) [32] used the numerical analysis method to analyze the factors such as the sensitivity of production decision-makers to the penalties for under-production, the residual value of inventory, and the distribution of defective rate and established a model for banks to make lending decisions through the contents of orders under different distributions, which provided decision-making basis for production and financial decision-makers. the optimal decision is taken by the supplier when the information is incomplete. by constructing the optimal decision model of inventory pledge financing stage loan (piwrsf) with complete information and renegotiation, it is concluded that a stage loan is better than a direct loan. when the moral hazard of logistics enterprises increases, the combination of inventory pledge, contract, and stage financing can make the suboptimal solution under incomplete information tend to the optimal solution under complete information (li juan et al., 2010) [6]. the simulation results of ouyang ni (2016) [33] show that due to the uncertainty of market prices and the absence of a 100% continuous price increase, farmers' continuous performance decision in their orders has a positive and powerful impact on optimizing their scale and revenue. under the conditions of asymmetric information and bounded rationality assumptions, an evolutionary game model of financing behavior between banks and nvocc is constructed, and the stability analysis of different behavior strategies is carried out. the results show that the bank's pledge rate, the amount of penalty for default, and the strength of nvocc's willingness to repay are the important influencing factors of whether nvocc will default. the bank's appropriate pledge rate, the amount of penalty for default is large enough, and a strong willingness to repay will effectively reduce the probability of default of nvocc, thus avoiding financing credit risk (huang weixing and gao gengjun, 2020) [34]. to sum up, the initial capital of the supplier and the trust degree of corporate reputation have an important impact on the final revenue of all parties. taking the form of stage loans and continuous performance have a positive impact on their scale and revenue; the retailer's profit is affected by the current price of capital, market demand, and the predetermined quantity of products; banks should pay close attention to whether suppliers' capital requirements are reasonable, default rates of core enterprises, risk-free interest rates and reasonable adjustment of risk caps. all the above decisions will affect the overall effectiveness of supply chain finance order financing. 4. research on risk management of order financing order financing business refers to a business in which a financial institution, based on a valid sales order signed by an enterprise, applies the risk control method of "one single loan, one collection yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 228 settlement" to issue a fully-enclosed loan for that order. at present, domestic research on order financing risk mainly focuses on risk analysis and prevention, risk identification and evaluation, etc. in terms of risk analysis and prevention and control, based on the understanding of the connotation of order financing business, li yixue (2008) [35] believes that the pre-loan assessment of order financing business should focus on the assessment of the risks in the order fulfillment process, which can be specifically analyzed from three parts: the assessment of the borrower's entity, the assessment of the order and the assessment of the supervision level of the financier. based on the risk analysis model of order financing business in the value chain process, li yixue and wu lihua (2008) [3] describe the risks encountered by borrowing enterprises in the process of realizing order financing more comprehensively. the research shows that the risk of the order fulfillment process is the most important factor affecting the loan repayment of order financing business; environmental risk, information system risk, the default risk of downstream manufacturers, and supervision risk of logistics enterprises may affect the financing business by affecting the order fulfillment process; to ensure the safety of order financing, the financier should carefully analyze and manage the five types of risks. based on the risk analysis of each link in the order financing business process, for order financing of products meeting the bank pledge requirements, the financing risk can be effectively controlled through the combination of several supply chain financing methods without limiting the core enterprise status of its downstream procurement enterprises (zhao jian and huo jiazhen, 2009) [5]. wang qian (2014) [36] proposed that commercial order financing products have the following risks: the access standard of the order financing core business enterprise is relatively low, the borrower is a trade-oriented enterprise, the investigation requirements on the relevance between the core customer and the borrower are lacking, the order type applicable to the products is not clear, the regulations on financing operation links are not detailed, and the capital flow and trade logistics do not match. zhu liefu and ding nanxi (2015) [37] explained the connotation and significance of the order financing business of rural financial institutions for agricultural enterprises, deeply analyzed the systematic risks of order financing of agricultural enterprises, and revealed the non-systematic risk characteristics such as credit risk, order risk and operation risk of order financing of agricultural enterprises, and gave specific countermeasures for rural financial institutions to control the systematic risk and non-systematic risk business of order financing business of agricultural enterprises. based on the perspective of banks, dong zhenning et al. (2015) [38] elaborated on the concept and business process of order financing and pointed out that its business risks are divided into four categories and 11 subcategories from four different perspectives: raw material suppliers, production enterprises, downstream core enterprises, and 3pl, and elaborated the control methods of various risks respectively. comparing order financing with traditional financing, from the perspective of business risk control methods, banks in order financing actively control risks, control capital flow by opening special closed accounts for participants, and control logistics by relying on real-time feedback information from cooperating 3pl enterprises, thus achieving maximum risk control in the process (liu wenjuan et al., 2015) [39]. xu liqi (2016) [18] constructed a risk control strategy with typical internet characteristics and combined with the specificity of the internet supply chain financial order financing model, specifically proposed a whole-process risk control mechanism including outsourcing of operation links, deconstruction of key variables and introduction of corresponding mechanisms, as well as collaboration and cooperation among platforms through information integration, matching, and control. taking the electronic order financing model under yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 229 b2b e-commerce supply chain finance as the research object, this paper identifies and analyzes the operational risks of electronic order financing from the four links of b2b e-commerce transaction link, electronic order financing link, online payment link, and logistics warehousing supervision link, and starting from the human factors and systematic factors in reducing the operational risks, puts forward the financial operational risk control strategy of b2b e-commerce supply chain (wang jin, 2017) [40]. taking the order financing guarantee business of gs nongdan company as the research object, this paper points out the problems existing in the risk control of order financing guarantee business, such as the unsound risk control system and inadequate implementation of rules and regulations, through the on-site investigation of the applicant farmers and all aspects of gs nongdan company's business, and puts forward suggestions for improving the business risk control, including constructing the risk control system of order financing guarantee business, standardizing the implementation of the risk management system, strengthening the performance ability construction of the risk management department, improving the talent team and performance evaluation mechanism, and accelerating the construction of the risk management information system (guo yundong, 2020) [41]. in terms of risk identification and evaluation, deng aimin, xiong jian, and zhangfan (2010) [42] based on the research results of domestic and foreign experts on order financing and the definition of its connotation, identified the risk types of order financing business using questionnaires and indepth interviews. on this basis, established an order financing risk early warning index system, and constructed an order financing risk early warning model based on bp neural network. zhao qi (2013) [43] established an order financing risk assessment index system, and established an order financing risk assessment model combining rough set theory and credibility theory, established a risk assessment information system through expert scoring, reduced the assessment index system, calculated the weight of each secondary index after reduction, and finally obtained a comprehensive evaluation value combining credibility theory. zhao qi (2013) [44] has established an index system for evaluating the order financing risk, and uses the method of combining the extension level analysis with the fuzzy comprehensive evaluation to evaluate the order financing risk. it is more reasonable and convenient than the traditional analytic hierarchy process to use the extension interval number to express the relative importance degree among the indexes quantitatively. through expert scoring, an extension interval judgment matrix is established, the weights of evaluation indexes at each level are calculated, and a comprehensive evaluation value is finally obtained by combining fuzzy comprehensive evaluation. under different circumstances, chang qing (2014) [45] uses the elastic coefficient method to conduct quantitative research on the financial risk transmission effect of order financing. for each different situation, it uses the elastic coefficient method to conduct quantitative calculations on each risk transmission effect, and studies the impact on the revenue of financing enterprises, core enterprises, banks, and logistics enterprises when the raw material price of financing enterprises changes and the resulting supply chain financial risk is transmitted along the supply chain, and studies the risk transmission effect to which they are subjected. combined with other participants in the online order financing model, the credit risk evaluation index system is established, the weight of the evaluation index does not depend on the expert evaluation method, and the credit risk evaluation system of online distributors is constructed by the logistic regression method combined with factor analysis. the credit risk evaluation results are more objective and scientific (liang fei, 2015) [46]. taking "orders" as the lead, the risk of different stages is analyzed and a twostage risk level evaluation model is constructed. the first stage judges the authenticity of the business yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 230 order through two first-level indicators; in the second stage, the index system including subject credit, pledge rate, order fulfillment ability, inventory, etc. is constructed, and the measurement methods of various indexes are described in detail. the various quantitative indexes are uniformly projected to 0-10 points by using the processing methods of the exponential function, cosine function, power function, etc. then the anp method and ahp method are used to determine the weights of various indicators, and a risk level evaluation model is constructed to classify the risks into five levels (dong zhenning et al., 2017) [47]. the research results of han yahui (2019) [48] show that the research on credit risk evaluation of order financing needs dynamic analysis from multiple perspectives. relevant credit risk managers should not only pay attention to the qualification of financing enterprises, but also pay attention to the core enterprise qualification in the supply chain. in addition, as the cooperative supervisor and guarantor of banking and other finance, the third-party logistics enterprise is also the focus of the credit risk monitoring department. based on combing the evolution process of the electronic order model and its risks, tian ran and xu kun (2020) [49] describe the risk relationship among various entities from a systematic perspective, determine the key risk factors, and perform simulation analysis by using the principle of system dynamics. it is pointed out that the electronic order financing system consists of multiple subsystems, and a feedback relationship is formed among each subsystem. controlling the risks of each subsystem is helpful to reduce the overall risk of electronic order financing. huang weixing and gao gengjun (2020) [50] applied the order financing model under supply chain finance to the field of non-vehicle carriers. using the evolutionary game method, under the assumptions of information asymmetry and bounded rationality, they constructed an evolutionary game model of financing behavior between banks and non-vehicle carriers, analyzed the stability of different behavior strategies, and carried out numerical simulations. through the discussion on the model of supply chain finance of marine water products, he jing and gu limin (2021) [51] used the fuzzy analytic hierarchy process (fahp) to evaluate the financial risks of the supply chain of marine water products, providing the basis for the next step of risk prevention and control, and striving to promote the application of supply chain finance in marine water products industry, with more scientific and objective evaluation results. according to the order financing business process of the case company, the influencing factors of the order financing risk of the e-commerce platform are identified. based on the risk identification results of the case, an evaluation index system of the order financing risk of the e-commerce platform is constructed, and then the order financing risk of the case company is evaluated by using the comprehensive analytic hierarchy process and expert scoring method. finally, the proposed measures are obtained, such as setting up a hierarchical access standard for order financing vehicles, strengthening the performance inspection of pledged vehicles, establishing a platform dealer transaction database, designing a dealer reward and punishment mechanism, building an automobile industry database and an emergency early warning mechanism (mou baishen, 2021) [52]. 5. brief review order financing is currently an important way for small and medium-sized enterprises to obtain loans. retailers submit orders to suppliers through their good reputation, thus applying to banks for order financing, thus avoiding the situation of capital fragmentation. the order financing model has become the mainstream supply chain financing model at the current stage. through the research results of the existing literature, the following policy recommendations are put forward. yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 231 first, improve laws and regulations and strengthen government supervision. the order financing model is different from the traditional financing model, which requires the government and relevant departments to establish and perfect the laws and regulations of supply chain financing. the government and relevant departments should clarify the business scope of order financing, standardize the corresponding process and the scope of responsibilities of each subject, and formulate sound risk prevention and control strategies to provide legal basis and institutional guarantee for financing risks. in addition, it is also necessary to establish a perfect supervision mechanism to supervise each entity and process in the order financing business, so as to avoid regulatory loopholes that may affect the healthy development of order financing. second, establish and improve the credit system. the credit of small and medium-sized enterprises is directly related to whether they can successfully apply for loans. therefore, it is necessary to establish and perfect the credit database of enterprises and realize information sharing with the government, core enterprises, banks and other financial institutions. the government and relevant departments should strictly review their transaction records with the core enterprises and the development status of the supply chain where they are located, to improve the efficiency of order financing and reduce risks, to enable more small and medium-sized enterprises to obtain financing and enable enterprises to develop healthily. third, build an e-commerce financing service platform. the government and relevant departments should build an order financing service platform according to the actual situation, grasp the production and sales data of enterprises in real time, and discover abnormal data in time. the service platform can effectively solve the problems of small and medium-sized enterprises such as accounts, inventories, difficult and expensive financing and lack of collateral. the government needs to formulate relevant laws and regulations, clarify the legal effect of electronic certificate, safeguard the interests of various subjects in the order financing process from the legal source, strengthen supervision and punish the dishonesty, to promote the order financing mode dominated by ecommerce platforms to become an important financing channel for small and medium-sized enterprises. funding: this research was funded by the regional project of the national natural science foundation of china, grant number 71861003. conflicts of interest: the authors declare no conflict of interest. the funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. references [1] novembrino t a. financing international transactions[j]. secured lender, 1996,52(6):38. [2] marks k h, l e robbins, g fernandez, j p funkerhouser, d l williams. the handbook of financing growth: strategies, capital structure, and m&a transactions[m]. new jersey nj: john wiley&sons, 2009:356-358. doi: https://doi.org/10.1002/9781118267691. [3] li yi xue, wu lihua. risk analysis and management of order financing business under logistics finance innovation [j]. contemporary finance and economics, 2008(12):66-70. doi: cnki: sun: ddcj.0.2008-12013. [4] liu yaya, zeng youxin. research on implementation of order financing business in logistics enterprises [j]. modern business, 2009(03):180. doi: https://doi.org/10.3969/j.issn.1673-5889.2009.03.114. yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 232 [5] zhao jian, huo jiazhen. research on order financing model independent of core enterprises [j]. journal of kunming university of technology (social science edition), 2009,9(12):62-65. [6] li juan, xu yu, jia tao. risk management of order financing business under logistics finance innovation [j]. statistics and decision, 2010,(19):171-173. doi: cnki: sun: ddcj.0.2008-12-013. [7] hong yitian. research on order financing strategy of the supply chain under logistics finance [j]. statistics and decision, 2020,36(01):176-180. doi: https://doi.org/10.13546/j.cnki.tjyjc.2020.01.039. [8] li han, wang zhaoxu, zou lei. a study on feasible approaches to order financing of agricultural products: theoretical analysis and concrete practice [j]. credit information, 2014,32(04):67-71. [9] lan qinggao, wang tingrui. research on order financing of agriculture based on system dynamics: a case study of order financing of dairy farming [j]. journal of the capital university of economics and business, 2014,16(04):32-40. doi: https://doi.org/10.13504/j.cnki.issn1008-2700.2014.04.004. [10] wu haixia. weifang city, the development of order financing and other new forms of financing, benefiting rural small and medium-sized enterprises [j]. shang, 2015(12):228. doi: cnki: sun: shng.0.2015-12-190. [11] wang tingrui, chu yongze. analysis of agricultural order financing [j]. journal of dalian nationalities university, 2015,17(06):574-577. doi: https://doi.org/10.3969/j.issn.1009-315x.2015.06.010. [12] song ping, jiang yixin, yang qifeng. an example of the design of order financing model for fresh agricultural products supply chain [j]. journal of finance and accounting, 2016,(26):86-89. doi: cnki: sun: ckyk.0.2016-26-018. [13] hu huiyong. "research on farmers' production behavior under the" company + farmers + order financing "management model [d]. nanchang university, 2019. [14] wang xi. research on order financing model of pig breeding from the perspective of supply chain [d]. northeast agricultural university, 2021.doi: https://doi.org/10.27010/d.cnki.gdbnu.2021.000626. [15] wang caifen, li fuchang. research on order financing in supply chain of chinese herbal medicine in yunnan province [j]. business exhibition economics, 2021(19):65-67. [16] jiang boheng, wen tao. how order financing promotes income increase of large-scale farmers: theory and evidence [j]. research on financial issues, 2022,(06):81-90. [17] liang fei. credit risk assessment of online distributors under online order financing model [d]. university of electronic science and technology, 2015. [18] xu liqi. internet supply chain finance, electronic order financing risk and banking decision research [d]. wuhan university of technology, 2016. [19] rebecca. research on supply chain financial risks and models based on electronic orders [d]. university of electronic technology, 2017. [20] huang yingcan. decision analysis of electronic order financing based on b2b platform [d]. xiamen university, 2018. [21] yin qin. e-order financing transaction decision of sellers in the green supply chain based on threedimensional trust [d]. wuhan university of technology, 2019. [22] yang qifeng, yang xiaoci, song ping. game analysis of electronic order financing for supply chain financial sellers considering institutional trust [j]. journal of finance and accounting, 2019,(10):132-138. doi: https://doi.org/10.19641/j.cnki.42-1290/f.2019.10.019. [23] li haocheng, wen renling, li leping, xu chang. innovative research on the financing model of small and micro agricultural enterprises takes the construction of an internet order financing platform as an example [j]. china business theory, 2021,(01):105-108. doi: https://doi.org/10.19699/j.cnki.issn20960298.2021.01.105. [24] zhangxin, lin qiang. research on the optimal decision of a three-echelon supply chain under delayed payment and order financing [j]. journal of wuhan university of technology (transportation science and engineering edition), 2015,39(02):405-409. [25] zhao ruixue. optimal operation strategy of the supply chain under the financing strategy from order financing to reverse factoring [d]. beijing foreign studies university, 2018. [26] wang zongrun, shi jiaxing. research on supply chain order financing interest rate under asymmetric information [j]. financial theory and practice, 2018(01):81-85. doi: https://doi.org/10.3969/j.issn.10034625.2018.01.017. [27] xu qing. research on supply chain coordination of call option contracts under order financing [d]. tianjin university, 2016. yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 233 [28] wang shuoxin. trust perception and transaction decision-making of buyer's electronic order financing participants [d]. wuhan university of technology, 2019. doi: https://doi.org/10.27381/d.cnki.gwlgu.2019.000540. [29] wang xi. research on order financing model of pig breeding from the perspective of supply chain [d]. northeast agricultural university, 2021.doi: https://doi.org/10.27010/d.cnki.gdbnu.2021.000626. [30] wang wenli, luo jianwen, zhang qinhong. research on supply chain order financing strategy under bank risk control [j]. china management science, 2013,21(03):71-78. doi: https://doi.org/10.16381/j.cnki.issn1003-207x.2013.03.021. [31] song ping, wang shuoxin, yang qifeng. an electronic order financing contract coordination model based on reputation trust [j]. journal of beijing university of posts and telecommunications (social science edition), 2018,20(06):9-17+51. [32] wang ying, chen zhi. research on order financing decision under uncertainty of output and payment return [j]. journal of guangdong open university, 2021,30(06):105-111. doi: 10.19722/j.cnki.10087729.2018.0211 [33] ouyang ni. research on the decision of farmers' order fulfillment in order financing of agricultural products supply chain [d]. hunan university, 2016. [34] huang weixing, gao gengjun. order financing risk decision of nvocc based on evolutionary game [j]. manufacturing automation, 2020,42(05):10-15. [35] li yixue. logistics finance innovation-pre-loan evaluation of order financing business [j]. statistics and decision, 2008(24):168-170. doi: cnki: sun: jxgy.0.2020-05-003. [36] wang qian. risk analysis and management suggestions on order financing products of commercial banks [j]. modern economic information, 2014(19):355+362. [37] zhu liefu, ding nanxi. risk control strategies for the order financing model of agricultural enterprises [j]. hubei agricultural sciences, 2015,54(04):1007-1009. doi: https://doi.org/10.14088/j.cnki.issn04398114.2015.04.061. [38] dong zhenning, liu wenjuan, he bin, zhuo wang. research on the risk of order financing business [j]. business economics research, 2015(11):99-100. doi: https://doi.org/10.3969/j.issn.1002-5863.2015.11.037. [39] liu wenjuan, dong zhenning, he bin, zhuo wang. comparative analysis of order financing and traditional financing [j]. logistics technology, 2015,34(17):33-35. [40] wang jin. research on financial risk control in b2b e-commerce supply chain [d]. southeast university, 2017. [41] guo yundong. research on risk control of order financing guarantee of gs nongdan company [d]. xidian university, 2020. doi: https://doi.org/10.27389/d.cnki.gxadu.2020.003320. [42] deng aimin, xiong jian, zhangfan. an early warning model of order financing risk based on bp neural network [j]. journal of information, 2010,29(11):23-28. [43] zhao qi. risk assessment of order financing based on rough set and credibility theory [j]. logistics technology, 2013,32(13):278-280. [44] zhao qi. risk assessment of order financing based on fuzzy extension ahp [j]. journal of chongqing university of technology (social sciences), 2013,27(02):35-39. [45] chang qing. analysis of risk transmission effect of order financing based on the elastic coefficient method [d]. southwest jiaotong university, 2014. [46] liang fei. credit risk assessment of online distributors under online order financing model [d]. university of electronic science and technology, 2015. [47] dong zhenning, liu wenjuan, zhuo wang, he bin. research on the risk level evaluation of order financing business [j]. logistics and management, 2017,26(02):140-145. doi: https://doi.org/10.12005/orms.2017.0043. [48] han yahui. research on credit risk assessment of order financing of small and medium-sized enterprises under supply chain [d]. hunan university, 2018. doi: cnki: cdmd: 2.1018.103005. [49] tian ran, xu kun. risk formation mechanism and simulation analysis of electronic order financing under e-commerce background [j]. journal of finance and accounting, 2020,(19):139-145. doi: https://doi.org/10.19641/j.cnki.42-1290/f.2020.19.020. [50] huang weixing, gao gengjun. order financing risk decision of nvocc based on evolutionary game [j]. manufacturing automation, 2020,42(05):10-15. doi: cnki: sun: jxgy.0.2020-05-003. yi-fan fu and mu zhang / journal of risk analysis and crisis response, 2022, 12(4), 222-234 doi: https://doi.org/10.54560/jracr.v12i4.345 234 [51] he jing, gu limin. research on financial risk assessment of marine water supply chain based on order financing model [j]. ocean economics, 2021,11(04):54-62. doi: https://doi.org/10.19426/j.cnki.cn121424/p.2021.04.005. [52] mou bochen. risk identification and assessment of order financing for e-commerce platform [d]. beijing jiaotong university, 2021. doi: https://doi.org/10.26944/d.cnki.gbfju.2021.003303. copyright © 2022 by the authors. this is an open-access article distributed under the cc by-nc 4.0 license (http://creativecommons.org/licenses/by-nc/4.0/). microsoft word after modificatiion ??ok?? assessment research of bijie drought risk based on cloud model ying he school of finance, guizhou university of finance & ecnomics, guiyang 400020, china *e-mail: obying@163.com abstract this paper chooses 8 cities and counties of bijie area as the research target. with the research foundation of natural disaster risk theory and drought risk formation principle, we start from dangerousness, exposure , vulnerability and the ability to prevent disaster to filter out 23 indexes range from social status, economy, weather and geology to construct the assessment system of bijie drought risk. to solve the uncertainty of quantitative description for drought risk assessment index and the judge of assessment result, cloud model and entropy weight method are implemented to decide how much percentage the city or county belong on each index. in the end, we use the weight of 23 indexes to get the drought risk level each city or county belongs to. the result reveal that weining has the highest risk while bijie, qianxi, zhijin, hezhang have higher risk of drought occurence and dafang and jinsha have normal risk, nayong has lower risk. this research aims to help governments at all levels to prevent drought and provide decision basis to reduce drought loss. keywords: bijie; drought risk; normal cloud model; cloud generator; entropy weight method; 基于云模型的毕节地区干旱灾害风险评判研究 ** 贺颖 贵州财经大学/金融学院，贵阳 400020 摘要：本文以贵州省毕节地区的 8 个市、县作为研究对象，从形成干旱灾害风险的危险性、暴露性、脆弱 性和防灾减灾能力等 4 个因素考虑，选取气温、降水、日照百分率等方面的 23 个指标，构建了毕节地区干 旱灾害风险评价指标体系。针对区域干旱灾害风险评价指标定量描述的不确定性和评价结果等级判定的不 确定性，建立了基于正态云模型和熵权法的区域干旱灾害风险评价模型，并利用历史数据对毕节地区 8 个 市、县的干旱灾害风险进行了实证测评。结果显示，威宁县的干旱灾害风险高，毕节市、黔西县、织金 县、赫章县的干旱灾害风险较高，大方县、金沙县的干旱灾害风险一般，纳雍县的干旱灾害风险较低。从 而为各级政府更为有效的指导防灾和备灾，减少旱灾损失提供了决策依据；有助于更为有效的对干旱灾害 进行早期预警；有助于抗旱应急预案的编制。 关键词：毕节地区；干旱灾害风险；正态云模型；云发生器；熵权法 1. 引言 由于特殊的地理位置和人均水资源匾乏，旱灾 成为我国影响范围最广、发生最频繁的气象灾害， 制约了国民经济的可持续发展[1]。干旱灾害也是贵 州省毕节地区最主要的自然灾害之一。2009 年入秋 以来，毕节地区发生的四季连旱，持续时间长、危 害程度深、受灾面积大、经济损失重，是有气象记 **基金项目：国家自然科学基金项目（71263011）；教育部人文 社会科学研究规划基金项目（11yja630196）；毕节地区 2011 年 科技发展计划项目（[2011]01） 录以来最严重的一次干旱[2]。旱灾造成全地区 619.7 万人受灾，425.24 万人、139.43 万头牲畜饮水困 难；585.44 万亩农作物受灾，经济损失达 32.32 亿 元。因此，通过对毕节地区干旱灾害风险进行综合 评价，识别干旱高敏感区和高风险区，可以为各级 政府更为有效的指导防灾和备灾，减少旱灾损失提 供决策依据；可以更为有效的对干旱灾害进行早期 预警，指导各级政府抗旱救灾；同时，对当地政府 编制抗旱应急预案，增强抗旱应急管理能力具有重 要的参考价值[3]。 美国国家干旱减灾中心（ndmc）主任 wilhite 博士将干旱灾害风险定义为：干旱灾害的出现（即 journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 192-200 published by atlantis press copyright: the authors 192 willieb typewritten text received 20 november 2013 willieb typewritten text accepted 13 december 2013 willieb typewritten text ying he 发生的可能性）和社会脆弱性的结果 [4] 。综合来 看，干旱灾害风险可以定义为：干旱影响范围内暴 露于干旱威胁下的承灾体，在干旱致灾因子和孕灾 环境的作用下，由于承灾体自身的脆弱性，所遭受 损失的可能性及大小。干旱灾害风险是由内部和外 部因素共同导致的，内部因素是承灾体自身和人类 活动所导致的脆弱性，外部因素是干旱致灾因子和 孕灾环境的影响。目前，对于干旱灾害风险评价主 要是运用数学方法，对干旱灾害的孕灾环境、危险 性和承灾体脆弱性等因素进行综合分析，最终用干 旱灾害风险度来刻画干旱灾害风险的大小。目前， 评价自然灾害风险的方法可以归纳为如下几种：概 率统计[5-6]、模糊数学[7-9]、信息扩散理论[10-11]、层次 分析法[12-13]、灰色系统理论 [14-17] 、人工神经网络 [1819] 、加权综合评价 [20-21] 等。上述方法各有其优缺 点，但均未考虑在综合评价中针对干旱灾害风险评 价指标定量描述的不确定性和评价结果等级判定的 不确定性。 针对干旱灾害风险评价指标定量描述的不确定 性和评价结果等级判定的不确定性，模糊性和随机 性并存，单纯只考虑模糊性容易造成评价结果的失 真，有鉴于此，引入云模型这样一个以自然语言值 为切入点，实现定性概念与定量数值之间的不确定 性转换的模型；它把模糊性与随机性这二者完全集 成在一起构成定性和定量相互间的映射它同时反映 了客观世界中概念的两种不确定性，即随机性(发生 的概率)和模糊性(亦此亦彼性)[22-24]。之后，拟从危 险性、暴露性、脆弱性和防灾减灾能力等四个方面 构建毕节地区干旱灾害风险评价指标体系，建立了 基于正态云模型和熵权法的区域干旱灾害风险评价 模型，并利用历史数据对毕节地区 8 个市、县的干 旱灾害风险进行了实证测评。 2. 云模型概述 云模型是李德毅院士提出的作为不确定性的定 性定量转换的模型，目的是用于处理定性概念中广 泛存在的随机性和模糊性，已成功应用于智能控 制、决策分析、图像处理、数据挖掘等领域。云模 型是具有普遍适用性的正态云，其数字特征用期望 ex (expected value)，熵 en (entropy)，超熵 he (hyper entropy)三个数值来表征[22]，其中： 期望 ex 表示云滴在论域空间分布的期望，是 最能够代表定性概念的点，或者说是这个概念量化 的最典型样本；熵 en 代表定性概念的可度量度， 定性概念的不确定性随着熵越大而变大，由概念的 随机性和模糊性共同决定。用同一个数字特征来反 映随机性和模糊性，也必然反映他们之间的关联 性；超熵 he 是熵的不确定性度量，即熵的熵，由 熵的随机性和模糊性共同决定，反映了云滴的离散 程度。云模型如图 1 所示: 图1 正态云及数字特征 (ex=5,en=1.5,he=0.07) fig.1 normal cloud and digital characteristics (ex=5,en=1.5,he=0.07) 如果在论域u 中确定点 x ，通过云发生器可以 生成这 x 个特定点属于概念 c 的确定度分布，这时 的云发生器称为正向云发生器。正向正态云发生器 构建了定性到定量的映射关系，根据正态云的数字 特征 ( , , )ex en he 产生云滴,并确定一个特定点处于 某种概念的确定度分布[24]。其具体算法为： （1）给定熵 en 和超熵 he ,生成正态分布的随 机数 ' 2~ ( , )en n en he （2）利用输入值 x 计算隶属度 2 2 ( ) ( ) exp[ ] 2 ' x ex x en     (1) 3. 构建基于正态云模型和熵权的综合评估模型 根据能客观反映指标权重大小的熵权，结合定 量到定性的正态云模型，建立综合评估模型如下 [27] ： 步骤 1，对于待评估对象建立因素论域 1 2{ , ,..., }nu u u u ,以及定性的评价论域 1 2{ , ,..., }nv v v v ； 步骤 2，用熵权法对指标权重进行计算，得到 权向量 1 2{ , ,..., }nw w w w 。具体计算步骤如下所 示。首先，将第 i 个区县的第 j 个指标的数据 ijx 转 化为该值占第 j 个指标的百分比 ijp 1 ij ij m ij i x p x    ，其中 1, 2,..., ; 1, 2,...,i m j n  (2) published by atlantis press copyright: the authors 193 assessment research of bijie drought risk based on cloud model 然后，定义第 j 个指标的熵为 1 m j ij ij i h k p lnp     ， 11, 2,..., ;j n k lnm  (3) 引入常数 k 保证第 j 个指标的 ijp 都相等，满足 1jh  ，此时该指标不能提供任何信息。当 0ijp  时，令 0ij ijp lnp  ，从而保证 [0,1]jh  。最后，定义第 j 个指标的熵权重为 1 1 1 1 1 (1 ) n j j j n n j j j j j h h w w h n h             ， (4) 步骤 3，进行单因素评估，建立模糊关系矩阵 r 。单因素评估即产生从评估对象的因素论域到评 语论域的一对一映射， r 中元素 ijr 表示论域u 中 第 i 个因素 iu 对应于评语论域v 中第 j 个等级 jv 的 隶属度，即一个区县的某一指标（因素论域）对于 若干评价等级（评语论域）进行映射。设定 1 2,ij ijx x 为因素 ( 1, 2, 3, , )i i n l 对应评级等级 ( 1, 2, 3, , )j j n l 的上、下边界值，则因素 i 对应 等级 j 这一定性概念可以用正态云模型表示如下 1 2( ) / 2ij ij ijex x x  (5) 边界值是两个等级的模糊边界，则此过渡值隶 存在相同概率隶属于两种等级： 21 2 2 ( ) exp[ ] 0.5 8( ) ij ij ij x x en    (6) 即 1 2 2.355 ij ij ij x x en   (7) 超熵 ijhe 度量了熵的不确定性，即云滴的凝聚 程度，超熵值越小代表云的厚度越小，反之亦然， 本文根据经验对超熵进行确定[28]。 步骤 4，利用正向云发生器生成待评价项目的 各个指标对应每个等级的云模型隶属度矩阵 ( )ij n mz z  。另外，为提高评估的可信度，重复运 行 n 次正向云发生器以计算某指标对于某等级的隶 属度的平均值： 1 / n k ij ijk z z n    (8) 步骤 5，利用隶属度矩阵 z 与权重集w 进行模 糊转换得到评价集上的模糊子集 b ， 1 2( , ,..., )mb z w b b b   1 , 1, 2,..., n j i ij i b w z j m    (9) 其中 jb 表示待评价对象（某区县）对第 j 条评语 （风险等级）的综合隶属度。根据最大隶属度原 则，最大隶属度所对应的评价等级 i 即为综合评价 结果。 4. 构建毕节干旱灾害风险综合评价 4.1 指标体系和样本数据 本文以自然灾害风险的形成理论和干旱灾害风 险的形成原理为理论基础，借鉴自然灾害风险指数 来构建干旱灾害风险评价指标体系。自然灾害风险 指数指未来若干年内可能达到灾害程度及其发生的 可能性，是危险性、暴露性和脆弱性相互作用的结 果，而防灾减灾能力由于代表社会对于灾害的认知 和对于自然灾害风险大小也有一定的影响，即某地 区内的自然灾害风险形成过程是这 4 个因素综合影 响的结果[29]。（表 1）。整个指标体系分为准则 层、指标层和子指标指标层，各指标的权重为熵权 法计算结果[30]。 本文所使用的气象指标数据来自毕节市气象局 的统计报告，水资源指标数据来源于《毕节地区水 资源公报》（2010），地形指标数据来源于《毕节 地区志--土地志》，在校学生比例数据来源于《贵州 教育》，社会经济指标数据来源《毕节地区统计年 鉴 1988-2008》以及《毕节试验区—辉煌十一五》。 指标原始值如表 2 所示。 published by atlantis press copyright: the authors 194 ying he 表1 干旱灾害风险评价指标体系 tab.1 the assessment index system of drought risk 准则层 指标层 子指标层 权重 准则层 指标层 子指标层 权重 危险性 气象 气温 1u 0.0187 脆弱性 人口 人均生活用水量 14u 0.0319 降水 2u 0.0457 人均农业用水量 15u 0.0533 日照百分率 3u 0.0304 脆弱性人口比例 16u 0.0214 水资源 土壤湿度 4u 0.0539 农村贫困发生率 17u 0.0184 地下水资源量 5u 0.0425 经济 易旱耕地比率 18u 0.0204 地表水资源量 6u 0.0441 单位 gdp 工业用水量 19u 0.0521 年末蓄水量 7u 0.0573 防灾能力 灌溉能力 水田面积 20u 0.0595 地形 耕地坡度 8u 0.0663 社会经济实力 城镇人均可支配收入 21u 0.0716 水田坡度 9u 0.0650 农民人均纯收入 22u 0.0595 旱地坡度 10u 0.0448 教育水平 在校学生比例 23u 0.0471 暴露性 人口 人口密度 11u 0.0235 经济 农林牧渔业产值 12u 0.0254 工业总产值 13u 0.0471 表2 毕节八个地区指标原始值 tab.2 the data of 8 counties/cities in bijie district 指标 毕节市 大方县 黔西县 金沙县 织金县 纳雍县 威宁县 赫章县 1u 13.50 12.70 14.70 15.40 14.90 14.40 14.30 11.90 2u 749.70 818.80 729.40 1020.0 1195.8 995.40 796.7 815.6 3u 39.2 29.1 28.3 24.5 26.3 31.7 40.3 31.6 4u 44.40 40.00 97.20 48.00 90.40 45.00 71.20 72.20 5u 4.35 4.97 4.18 3.93 6.31 4.01 8.23 5.23 6u 14.50 17.50 14.00 12.00 21.50 17.00 27.50 16.00 7u 1969.00 835.00 1268.00 1118.00 643.00 565.00 2009.00 1200.00 8u 16.95 28.58 14.20 15.150 15.52 16.69 14.46 16.62 9u 7.54 9.86 7.06 8.12 7.86 7.93 18.80 9.93 10u 17.92 18.96 15.19 16.99 16.50 17.44 14.37 16.81 11u 356.00 267.00 312.00 230.00 274.00 311.00 164.00 187.00 12u 133800.43 83710.62 100116.67 91814.33 77922.48 64804.10 101831.48 55552.57 13u 195450.85 50360.57 62994.19 140309.19 53677.48 100596.71 84175.05 77912.24 14u 29.8200 25.1600 27.3700 28.3300 26.1000 25.6700 31.5900 29.9600 15u 38.8600 115.66 40.9000 99.1300 38.46 23.01 24.8800 36.4200 16u 10.40 15.10 10.50 8.60 15.80 14.80 14.90 14.80 17u 6.90 9.9750 7.3000 4.4750 11.00 12.5250 9.4250 10.4750 18u 0.8923 0.8857 0.7918 0.6909 0.7596 0.8696 0.9867 0.9639 19u 76.8100 105.24 102.38 139.61 69.69 224.67 57.29 59.81 20u 6.8433 6.0314 9.5495 10.9786 10.7695 4.4024 0.9748 1.3895 21u 4064.17 3453.50 3831.37 3455.68 3536.21 3398.21 3628.21 3825.16 22u 1203.13 1067.88 1164.88 1443.5 1087.88 1101.38 961.88 1083.75 23u 0.1963 0.2064 0.2136 0.2022 0.1995 0.1856 0.1422 0.1580 published by atlantis press copyright: the authors 195 assessment research of bijie drought risk based on cloud model 4.2 干旱灾害风险评价等级划分 对于 23 个指标，为使得在云模型中隶属度的计 算更为准确，需要对每个指标的数值进行区划，构 建针对风险等级的低、较低、一般、较高、高五个 数值范围。由于影响干旱的因素很多，造成干旱的 原因不同，各地气候、地理条件差异很大，目前难 以采用全国统一的干旱评判标准。因此在参考若干 文献基础上，结合气象学、地质学、社会学等学科 知识分析各个指标与干旱之间的联系，进行干旱灾 害风险评价等级的划分[31-33]，如表 3 所示。 表3 干旱灾害风险评价等级划分 tab.3 level grading of drought risk assessment 指标 低风险 较低风险 一般风险 较高风险 高风险 指标 低风险 较低风险 一般风险 较高风险 高风险 1u 0-10 10-12 12-14 14-16 16-20 13u 0-50000 5000090000 90000 130000 130000 170000 170000 220000 2u 1100-1400 900-1100 650-900 500-650 0-500 14u 0-20 20-24 24-28 28-31 31-38 3u 0-14 14-22 22-30 30-38 38-50 15u 0-30 30-55 55-80 80-100 100-130 4u 91-105 77-91 63-77 44-63 0-44 16u 0-5 5-8 8-11 11-15 15-20 5u 7.5-9 6.5-7.5 5.5-6.5 3.5-5.5 0-3.5 17u 0-4 4-6 6-9 9-12 12-16 6u 26-34 22-26 17-22 12-17 0-12 18u 0-0.4 0.4-0.6 0.6-0.8 0.8-1 1-1.3 7u 1900-2300 1400-1900 900-1400 500-900 0-500 19u 0-60 60-105 105-155 155-205 205-250 8u 0-15 15-19 19-22 22-26 26-32 20u 10.5-14 8-10.5 5.5-8 2.5-5.5 0-2.5 9u 0-8 8-10 10-14 14-18 18-24 21u 4000-4500 3650-4000 3300-3650 3000-3300 0-3000 10u 0-7 7-11 11-15 15-19 19-23 22u 1500-2000 1250-1500 1000-1250 800-1000 0-800 11u 0-130 130-200 200-270 270-330 330-400 23u 0.21-0.3 0.19-0.21 0.17-0.19 0.15-0.17 0-0.15 12u 0-50000 5000080000 80000 100000 100000 120000 120000 160000 4.3 毕节地区各区县干旱灾害风险计算 本文通过对搜集的数据进行整理，并根据公式 (5)、公式(7)计算出期望、熵，结合干旱灾害风险评 价指标体系和评价等级划分，将五个等级的正态云 模型（包含期望、熵、超熵）表示如表 4，其中超 熵根据经验选定。 例如，将日照百分率这一指标利用公式(1)和云 矩阵 r 中 3u (如表 4)的数值,建立五个风险等级的正 态云隶属度函数，遵循清晰化原则来确定日照百分 率的超熵，如图 2 所示。 假定步骤 4 反复计算次数 200n  ，例如，将 毕节市干旱灾害风险量化数据代入上述正向正态云 发生器，重复计算 200 次，计算在不同隶属度情况 下的平均综合评估值，即得出毕节市各个指标对于 不同风险等级的隶属情况如表 5 所示。 最终，根据步骤 5 利用各指标权重与隶属度矩 阵 z 进行模糊转换得出评价集v 上的模糊子集 b 。 依据最大隶属度原则，选择最大的隶属度所对应的 第 i 个评价等级作为综合评价的结果。根据收集到 的所有县市指标数据，同理可得计算出毕节地区 8 个县、市的干旱灾害风险等级隶属度，如表 6 所 示。 根据计算结果可知，由于毕节地区乃至贵州省 都是经济水平、教育水平以及地理环境较差的地 域，所以没有任何市县干旱风险低，威宁县的干旱 灾害风险高，毕节市、黔西县、织金县、赫章县的 干旱灾害风险较高，大方县、金沙县的干旱灾害风 险一般，纳雍县的干旱灾害风险较低。 4.4 云模型与模糊综合评价法差异比较 为验证云模型评价的有效性，引入模糊综合评 价法进行评价结果对比。模糊综合评价一般采取专 家评审打分的方法建立模糊关系矩阵，本文试图根 据指标的等级划分数据，为指标建立单因素与隶属 度的映射关系：设某一个指标的某一等级的上下限 为 1 2,ij ijx x ，设定如下函数式(10)，并将原始值带入各 个隶属度函数进行运算得出表 7 结果。 published by atlantis press copyright: the authors 196 ying he 表4 干旱灾害风险正态云标准值 tab.4 the normal cloud standard value on drought risk 指标 低风险 较低风险 一般风险 较高风险 高风险 1u (18,1.7,0.1) (15,0.85,0.1) (13,0.85,0.1) (11,0.85,0.1) (5,4.25,0.1) 2u (250,212.31,7) (575,63.69,7) (775,106.16,7) (1000,84.93,7) (1250,127.39,7) 3u (44,5.1,0.1) (34,3.4,0.1) (26,3.4,0.1) (18,3.4,0.1) (7,5.94,0.1) 4u (22,18.68,0.3) (53.5,8.07,0.3) (70,5.94,0.3) (84,5.94,0.3) (98,5.94,0.3) 5u (1.75,1.49,0.05) (4.5,0.85,0.05) (6,0.42,0.05) (7,0.42,0.05) (8.25,0.64,0.05) 6u (6,5.1,0.1) (14.5,2.12,0.1) (19.5,2.12,0.1) (24,1.7,0.1) (30,3.4,0.1) 7u (250,212.31,8) (700,169.85,8) (1150,212.31,8) (1650,212.31,8) (2100,169.85,8) 8u (29,2.55,0.1) (24,1.7,0.1) (20.5,1.27,0.1) (17,1.7,0.1) (7.5,6.37,0.1) 9u (21,2.55,0.1) (16,1.7,0.1) (12,1.7,0.1) (9,0.85,0.1) (4,3.4,0.1) 10u (21,1.7,0.04) (17,1.7,0.04) (13,1.7,0.04) (9,1.7,0.04) (3.5,2.97,0.04) 11u (365,29.72,2) (300,25.48,2) (235,29.72,2) (165,29.72,2) (65,55.2,2) 12u (140000,16985.14,1500) (110000,8492.57,1500) (90000,8492.57,1500) (65000,12738.85,1500) (25000,21231.42,1500) 13u (195000,21231.42,1500) (150000,16985.14,1500) (110000,16985.14,1500) (70000,16985.14,1500) (25000,21231.42,1500) 14u (34.5,2.97,0.05) (29.5,1.27,0.05) (26,1.7,0.05) (22,1.7,0.05) (10,8.49,0.05) 15u (115,12.74,0.8) (90,8.49,0.8) (67.5,10.62,0.8) (42.5,10.62,0.8) (15,12.74,0.8) 16u (17.5,2.12,0.1) (13,1.7,0.1) (9.5,1.27,0.1) (6.5,1.27,0.1) (2.5,2.12,0.1) 17u (14,1.7,0.08) (10.5,1.27,0.08) (7.5,1.27,0.08) (5,0.85,0.08) (2,1.7,0.08) 18u (1.15,0.13,0.005) (0.9,0.08,0.005) (0.7,0.08,0.005) (0.5,0.08,0.005) (0.2,0.17,0.005) 19u (227.5,19.11,1) (180,21.23,1) (130,21.23,1) (82.5,19.11,1) (30,25.48,1) 20u (1.25,1.06,0.1) (4,1.27,0.1) (6.75,1.06,0.1) (9.25,1.06,0.1) (12.25,1.49,0.1) 21u (1500,1273.89,8) (3150,127.39,8) (3475,148.62,8) (3825,148.62,8) (4250,212.31,8) 22u (400,339.7,5) (900,84.93,5) (1125,106.16,5) (1375,106.16,5) (1750,212.31,5) 23u (0.075,0.06,0.007) (0.16,0.01,0.007) (0.18,0.01,0.007) (0.2,0.01,0.007) (0.255,0.04,0.007) 图2 正态云隶属度 fig.2 the normal cloud membership published by atlantis press copyright: the authors 197 assessment research of bijie drought risk based on cloud model 表6 毕节地区8个县、市干旱灾害风险隶属度结果 tab.6 the evaluation result of drought risk level for 8 counties/cites 县、市 低 较低 一般 较高 高 最终等级 毕节市 0.2 0.24 0.22 0.28 0.19 较高 大方县 0.24 0.23 0.42 0.14 0.07 一般 黔西县 0.06 0.19 0.31 0.32 0.21 较高 金沙县 0.13 0.27 0.33 0.26 0.16 一般 织金县 0.07 0.17 0.31 0.35 0.24 较高 纳雍县 0.19 0.35 0.28 0.21 0.13 较低 威宁县 0.23 0.17 0.23 0.17 0.24 高 赫章县 0.14 0.27 0.3 0.35 0.11 较高 表5 毕节市云模型平均综合评估值 tab.5 the average comprehensive evaluation value of cloud model for bijie city 指标 低 较低 一般 较高 高 指标 低 较低 一般 较高 高 1u 0.03 0.21 0.84 0.02 0.14 13u 1 0.03 0 0 0 2u 0.06 0.03 0.97 0.02 0 14u 0.29 0.97 0.08 0 0.07 3u 0.64 0.31 0 0 0 15u 0 0 0.03 0.94 0.17 4u 0.49 0.53 0 0 0 16u 0 0.31 0.78 0.01 0 5u 0.22 0.98 0 0 0 17u 0 0.02 0.89 0.09 0.02 6u 0.25 1 0.06 0 0 18u 0.13 1 0.08 0 0 7u 0 0 0 0.32 0.74 19u 0 0 0.04 0.96 0.18 8u 0 0 0.02 1 0.33 20u 0 0.08 1 0.08 0 9u 0 0 0.03 0.22 0.58 21u 0.13 0 0 0.27 0.68 10u 0.19 0.86 0.02 0 0 22u 0.06 0 0.76 0.27 0.04 11u 0.95 0.09 0 0 0 23u 0.17 0.03 0.25 0.77 0.3 12u 0.93 0.03 0 0 0 1 1 1 2 2 1 2 0, ( ) , 1, ij ij ij ij ij ij ij x x x x x x x x x x x x            (10) 对比表 7 中云模型综合评价和模糊综合评价的 结果，可以发现一般模糊综合评价中，风险度普遍 都较低且不同风险等级的隶属度之间区别不大，无 法科学地解释最终隶属度如何选择，因此与现实生 活中各个县市的风险情况存在差异，缺乏一定精确 度。 5. 结论 本文借鉴国内外关于自然灾害风险评价的相关 研究成果，从危险性、暴露性、脆弱性和防灾减灾 能力等四个方面构建毕节地区干旱灾害风险评价指 标体系。考虑干旱灾害风险评价指标定量描述的不 确定性和评价结果等级判定的不确定性的特点，构 建了基于熵权和正态云模型的综合评价模型，以毕 节地区的 8 个市、县为研究对象，利用历史数据进 行实证分析，结果表明威宁县的干旱灾害风险高， 毕节市、黔西县、织金县、赫章县的干旱灾害风险 较高，大方县、金沙县的干旱灾害风险一般，纳雍 县的干旱灾害风险较低。 表7 模糊综合评价结果对比 tab.7 the evaluation result of fuzzy evaluation 县、市 低 较低 一般 较高 高 模糊综合 毕节市 0.53 0.40 0.57 0.47 0.53 一般 大方县 0.44 0.27 0.61 0.47 0.53 一般 黔西县 0.54 0.34 0.45 0.39 0.48 低 金沙县 0.55 0.41 0.46 0.34 0.48 低 织金县 0.53 0.33 0.50 0.45 0.48 低 纳雍县 0.38 0.28 0.48 0.29 0.51 高 威宁县 0.54 0.38 0.56 0.44 0.46 一般 赫章县 0.36 0.24 0.47 0.29 0.45 一般 采用云模型评价区域干旱灾害风险，相对于一 般模糊综合评价方法，较好地解决了干旱灾害风险 评价指标定量描述的不确定性和评价结果等级判定 的不确定性问题。通过对毕节地区干旱灾害风险进 行综合评价，识别干旱高敏感区和高风险区，为各 级政府更为有效的指导防灾和备灾，减少旱灾损失 published by atlantis press copyright: the authors 198 ying he 提供了决策依据；有助于更为有效的对干旱灾害进 行早期预警，指导各级政府抗旱救灾；同时，对当 地政府编制抗旱应急预案，增强抗旱应急管理能力 具有重要的参考价值。 参考文献 1. c.jiawi, w.hao. survey of water resource[m].beijing, china water conservancy and hydropower press, 1996:14-15 陈家琦,王浩. 水资源学概论[m]. 北京:中国水利水电出 版社, 1996:14-15. 2. shang chongju, wang qun, hao zhibin. from 2009 to 2010 in guizhou province drought disaster cause, characteristics and influence of [j]. chinese water resources, 2010 (17): 11-13 商崇菊,王群,郝志斌. 贵州省 2009—2010 年特大干旱 灾害成因、特点及影响浅析[j]. 中国水利, 2010(17): 11-13. 3. chen jizu 3. henan province regional drought risk assessment [d]. zhengzhou: zhengzhou university, 2010 陈继祖. 河南省区域干旱灾害风险评估[d]. 郑州: 郑州 大学, 2010. 4. wilhite d a. drought as a natural hazard: concepts and definitions[c]. wilhite d a. drought: a global assessment[m]. london & new york: routledge, 2000, 3-18. 5. ma yinsheng, zhang yecheng, zhang chunshan, wang jinshan. theory and method of risk assessment of geological hazards [j]. journal of geological mechanics, 2004, 10 (1): 7-18 马寅生,张业成,张春山,王金山.地质灾害风险评价的理 论与方法[j].地质力学学报, 2004, 10(1): 7-18. 6. korkmaz k a. earthquake disaster risk assessment and evaluation for turkey[j]. environmental geology, 2009, 57(2): 307-320. 7. chen k, blong r, jacobson c. mce-risk: integrating multicriteria evaluation and gis for risk decision-making in natural hazards[j]. environmental modelling & software, 2001, 16(4) : 387-397. 8. hsu c, hong z. an intelligent typhoon damage prediction system from aerial photographs[j]. springerverlag berlin heidelberg, 2007, 747-756. 9. cao yongqiang, li xiangyun, ma jing, iggy beauty. assessment of agricultural drought risk dalian variable fuzzy algorithm based on [j]. resources science, 2011,33 (5).983-988 曹永强,李香云,马静,伊吉美.基于可变模糊算法的大连 市农业干旱风险评价[j].资源科学, 2011,33(5): 983-988. 10. feng l, luo g. flood risk analysis based on information diffusion theory[j]. human and ecological risk assessment,2008,14(6):1330-1337. 11. huang c. integration degree of risk in terms of scene and application[j]. stochastic environmental research and risk assessment,2009, 23(4): 473-484. 12. zhang jiquan, liu xingpeng, tong zhijun. the grassland fire disaster risk assessment and regionalization of [j]. geographical research, 2007, 26 (4): 755-762 张继权,刘兴朋,佟志军.草原火灾风险评价与分区[j].地 理研究, 2007, 26(4): 755-762. 13. castellanos abella e a, van westen c j. generation of a landslide risk index map for cuba using spatial multicriteria evaluation[j]. landslide, 2007, 4(4):311-325. 14. liu hexiang, xu qingjuan. regional flood disaster risk fuzzy comprehensive evaluation and prediction of [j]. disaster science, 2007,22 (4): 38-42 刘合香,徐庆娟.区域洪涝灾害风险的模糊综合评价与 预测[j].灾害学, 2007, 22(4): 38-42. 15. xinhu mao, liu zhankui, li xiuqing. application of [j]. journal of southwest university for nationalities in the mine geology disaster risk zoning evaluation of grey clustering method: natural science edition, 2008,34 (3): 545-551 毛新虎,刘占魁,李秀青.灰色聚类法在矿山地质灾害危 险性分区评价中的应用[j].西南民族大学学报:自然科 学版, 2008, 34(3): 545-551. 16. chen nanxiang, li shihui, yang li. grey fuzzy clustering theory application in the arid evaluation [j]. north china institute of water conservancy and hydroelectric power, 2007, 28 (2): 4-7 陈南祥,李士辉,杨莉.灰色模糊聚类理论在干旱评价中 的应用[j].华北水利水电学院学报,2007, 28(2):4-7. 17. ma jianqin, wei rui. fuzzy comprehensive evaluation of the agricultural drought evaluation index system and its application [j]. anhui agricultural science, based on the 2010, 38 (19): 10138-10140 马建琴,魏蕊.基于模糊综合评判的华北地区农业干旱 评价指标体系及其应用[j].安徽农业科学, 2010, 38, (19): 10138-10140. 18. aleotti p, chowdhury r. landslide hazard assessment: summary review and new perspectives[j]. bulletin of engineering geology and the environment, 1999, 58(1): 21-44. 19. chen k, blong r, jacobson c. towards an integrated approach to natural hazards risk assessment using gis :with reference to bushfires[j].environmental management,2003,31(4):546-560. 20. mansor s, shariah m a, billa l, et al. spatial technology for natural risk management[j]. disaster prevention and management, 2004, 13 (5):364-373. 21. raaijmakers r, krywkow j, veen a. flood risk perceptions and spatial multi-criteria analysis: an exploratory research for hazard mitigation[j].natural hazards,2008,46(3):307-322. 22. li deyi, liu changyu. on the normal cloud model of the universal [j]. china engineering science,.2004,6 (8): 2834 published by atlantis press copyright: the authors 199 assessment research of bijie drought risk based on cloud model 李德毅,刘常昱.论正态云模型的普适性[j].中国工程科 学.2004,6(8):28-34. 23. li deyi. knowledge representation in kdd based on linguistic atoms[j]. journal of computer science and technology, 1997,12(6):481—496. 24. li deyi, du yi. uncertainty in artificial intelligence [m]. beijing: national defence industry press, 2005.34-56 李德毅,杜鹢.不确定性人工智能[m].北京:国防工业出 版社,2005.34-56. 25. xie chi, zhong zan. entropy method in comprehensive evaluation of bank performance in the [j]. chinese soft science,.2002 (9): 108-110 谢赤,钟赞.熵权法在银行经营绩效综合评价中的应用 [j].中国软科学,.2002 (9):108-110. 26. tang yungang. entropy method of real estate listing corporation financial performance evaluation research of real estate on [j]..2012,10:53-58 汤云刚.熵权法下房地产上市公司财务绩效评价研究 [j].房地产开发.2012,10:53-58. 27. gong yanbing, zhang jiguo. comprehensive evaluation of population modernization development of the normal cloud model and entropy [j]. china population resources and environment based on.2012,1:138-143 龚艳冰,张继国.基于正态云模型和熵权的人口发展现 代 化 程 度 综 合 评 价 [j]. 中 国 人 口 资 源 与 环 境.2012,1:138-143. 28. gong yanbing 28. normal cloud model and entropy weight comprehensive evaluation of ecological risk city in hexi corridor [j]. journal of arid land resources and environment based on, 2012, 5: 169-174 龚艳冰.基于正态云模型和熵权的河西走廊城市化生 态风险综合评价[j].干旱区资源与环境, 2012, 5: 169174. 29. zhao jing, zhang jiquan, yan denghua et al..2012,1:5558. gis in the north area of henan province drought disaster risk zoning of disaster based on [j] 赵静,张继权,严登华等.基于 gis 的豫北地区干旱灾害 风险区划[j].灾害学, 2012,1:55-58. 30. yang chunyan, wang jingai, su jun, wang zhiqiang. assessment of agricultural drought vulnerability in the northern ecotone xinghe county as an example [j]. journal of natural disasters, 2005,14 (6): 89-93 杨春燕,王静爱,苏筠,王志强.农业旱灾脆弱性评价-以 北方农牧交错带兴和县为例［j］.自然灾害学报, 2005, 14(6): 89-93． 31. huang chongfu, liu xinli, zhou guoxian et al. to the historic disaster data of agricultural natural disaster risk assessment method according [j]. journal of natural disasters, 1998,7 (2): 1-9 黄崇福,刘新立,周国贤等.以历史灾情资料为依据的农 业 自 然 灾 害 风 险 评 估 方 法 [j]. 自 然 灾 害 学 报,1998,7(2):1-9. 32. ye jin, lin guangfa, zhang mingfeng. research progress of [j]. 2010,9:20-25. journal of institute of disaster prevention natural disaster risk assessment 叶金玉,林广发,张明锋.自然灾害风险评估研究进展[j]. 防灾科技学院学报, 2010, 9: 20-25. 33. zhang jiquan, liang jingdan, zhou daowei. gis based risk assessment of ecological disasters in jilin province [j] chinese journal of applied ecology 2007,8:1765-1770 张继权,梁警丹,周道玮. 基于 gis 技术的吉林省生态灾 害风险评价[j]. 应用生态学报,2007,8:1765-1770. published by atlantis press copyright: the authors 200 microsoft word wd-rbf model and its application of hydrologic time series prediction wd-rbf model and its application of hydrologic time series prediction dengfeng liu, dong wang*, yuankun wang key laboratory of surficial geochemistry, ministry of education; department of hydro-sciences, school of earth sciences and engineering, nanjing university nanjing, 210046, china lachun wang, xinqing zou school of geographic and oceanographic sciences, nanjing university nanjing, 210046, china * corresponding author e-mail: wangdong@nju.edu.cn abstract accurate prediction for hydrological time series is the precondition of water hazards prevention. a method of radial basis function network based on wavelet de-nosing (wd-rbf) was proposed according to the nonlinear problem and noise in hydrologic time series. wavelet coefficients of each scale were calculated through wavelet transform; soft-threshold was used to eliminate error in series. reconstructed series were predicted by rbf network. the simulation and prediction of wd-rbf model were compared with arima and rbf network to show that wavelet de-nosing can identify and eliminate random errors in series effectively; rbf network can mine the nonlinear relationship in hydrologic time series. examples show that wd-rbf model has superiority in accuracy compared with arima and rbf network. keywords: hydrologic time series, rbf network, wavelet de-noising, water hazards wd-rbf 模型在水文时间序列模拟预测上的应用 刘登峰 1，王栋*1，王远坤 1，王腊春 2，邹欣庆 2 1. 表生地球化学教育部重点实验室，南京大学地球科学与工程学院水科学系，南京，210046 2. 南京大学地理与海洋科学学院，南京 210046 *通讯作者 email: wangdong@nju.edu.cn 摘要：对水文时间序列科学的模拟预测是有效防范水灾害的前提。针对水文时间序列的非线性和噪声污染 问题，提出了一种基于小波消噪理论的径向基函数网络(wd-rbf)方法。通过小波变换得到水文时间序列 各尺度下的小波系数，利用软阈值消噪技术消除序列随机误差。重构小波还原水文时间序列，将消噪后的 序列进行的 rbf 网络模拟预测。将 wd-rbf 模型与 arima 模型和 rbf 网络进行对比分析，小波消噪能 够对序列中的随机误差进行有效识别并消除，rbf 网络能够挖掘水文时间序列中的非线性映射关系。实例 表明 wd-rbf 方法在序列模拟预测上表现出优于 arima 模型和 rbf 的性能，具有更高的准确性。 关键词：水文时间序列，rbf 神经网络，小波消噪，水灾害 1. 引言 在全球气候极端变化的背景下，以旱涝为代表 的水灾害的频繁发生严重威胁着人类的生命财产安 全。水文时间序列作为水文系统的数据化体现，对 其加以准确的模拟预测是有效防范水灾害，保证流 域和地区可持续发展的重要前提。由于影响水文系 统的因素错综复杂，难以用精确的数学语言来描 述，导致水文时间序列预测建模异常困难。 针对水文时间序列的非线性特征，人工智能领 域内的神经网络方法能够较好地予以解决。径向基 函数神经网络(rbf-nn)因具有较强的非线性拟合能 journal of risk analysis and crisis response, vol. 3, no. 4 (december 2013), 185-191 published by atlantis press copyright: the authors 185 willieb typewritten text received 15 september 2013 willieb typewritten text accepted 12 december 2013 willieb typewritten text dengfeng liu, dong wang, yuankun wang et al. 力和局部逼近的特性，近年来被应用于时间序列的 模拟预测中。李军和刘君华(2005)采用基于卡尔曼滤 波算法的新型广义 rbf 网络预测模型；张军峰和胡 寿松(2007)提出一种基于高斯基的距离度量，并联合 了输入输出聚类的策略；柳海涛、孙双科、刘之平 等(2005)在激发函数中引入 sign-d 函数，以改善模 型的稳定性和泛化能力；杨晓华、杨志峰、沈珍瑶 等(2005)用自相关技术分析时间序列的延迟特性，用 以确定 rbf 网络的输入、输出向量；刘俊萍和畅明 琦(2007)采用最近邻聚类学习算法确定径向基函数的 宽度、选取聚类中心和权值；路剑飞和陈子燊(2010) 利用引入滞后因子的 de-rbf 方法，探讨建立多步 预测模型的可行性，等等。 除了非线性特征以外，水文时间序列另一个无 法回避的问题为潜在的噪声污染。水文系统受自然 及人为因素干扰，实测序列往往受到不同程度的噪 声污染，对模拟预测工作产生了不利影响。对于水 文时间序列存在的噪声污染问题，传统的维纳滤 波、卡尔曼滤波等只适用于线型系统，且严格依赖 于状态空间函数的建立；而 fourier 变换仅适用于平 稳时间序列（王文圣，丁晶，李跃清，2005）。小 波消噪 (wa)是消除非平稳时间序列噪音的有效工 具。近年来许多学者对小波消噪方法进行了研究和 改 进 ： coifman 和 wickerhauster (1992) 基 于 shannon 熵提出了小波函数选择的依据和方法； lou 和 hu (2003) 基于二进制离散小波变换和 karhunenloeve 变换提出了用于分析非平稳宽带信号的消噪 方法；桑燕芳和王栋（2008）将小波变换中常用的 7 个小波系共 54 个小波函数作为选择范围，探讨建 立合理选择小波函数的依据和方法，等等。 针对水文时间序列的非线性和噪声污染问题， 提出了 wd-rbf 模拟预测方法。引入小波函数对水 文序列进行小波变换，并对各层小波系数进行阈值 消噪。消噪处理后的序列代入 rbf 网络进行建模预 测。以黄河花园口站和利津站的年径流序列，北京 市和南京市的年降水序列为例，对 wd-rbf 模型与 arima 模型、rbf 网络进行对比分析，验证方法的 准确性。 2. 基本理论 2.1 rbf 网络 人工神经网络包括单层前馈网络，多层前馈网 络和递归网络(s.haykin，2004)。径向基函数 (rbf) 网络是一种前馈式网络，具有唯一最佳逼近点，保 证全局收敛，因而具有较高的训练速度和仿真效 率。 rbf 神经网络基于径向对称的基函数，如式(1) 所示高斯核函数，实现输入空间到隐含层的非线性 变换。通过线性变换将隐含层数据映射至输出空 间，如式(2)所示。 ) 2 |||| exp(),( 2 2 i i i   cx cx   (1) 式(1)中， ic 为神经元 的核函数中心向量； i 为核函数的宽度参数。 wφy  (2) 式.(2)中，y 为 1r  输出向量，w 为 r n 阶权 重矩阵， 为 1n 向量。 2.2 小波阈值消噪 小波消噪技术始于小波阈值消噪方法的提出 (donoho，1995)。此后阈值消噪的方法广泛应用于 光谱分析（高国荣，刘艳萍，潘琼，2012），流量 预测（窦慧丽，刘好德，吴志周等，2009），信号 处理（赵莉，冯稷，翟光杰等，2005）等领域。 阈值消噪的思想即对小波分解后的各层系数中 模大于和小于某阈值的系数分别处理，然后对系数 进行反变换，重构出消噪后的信号。阈值确定准则 有： (1) 固定阈值 选择式为： nt n ln2 (3) 其中 n 为噪声强度， n 为小波系列的长度或 尺寸。 (2) stein 的无偏风险阈值 将某一层小波系数的平方由小到大排列，得到 一 个 向 量  nwwww ,...,, 21 ，  nwww  ...21 。 由 此 计 算 风 险 向 量  nrrrr ,...,, 21 。其中， n wwinin r i k ki i     1 )(2 (4) 以 r 中最小元素 br 作为风险值，由 br 的下标找 到对应的 bw ，则阈值： bwt  (5) 式中 为噪声强度。 (3) heuristic sure 阈值 比较两个变量： published by atlantis press copyright: the authors 186 wd-rbf model and its application of hydrologic time series prediction nnxeta n j j         1 2  (6) 3 2ln ln1       n n crit  (7) 如果 eta<crit，则选用固定阈值；否则取固定阈 值和无偏风险阈值中的较小者作为本准则的阈值。 固定阈值处理小波系数的缺点是会歪曲原始信 号包含的弱特征成分，从而引起重构信号的失真； 而当序列信噪比较小时，sure 估计会有较大误 差。heuristic sure 阈值是固定阈值和 sure 阈值 的综合，所选择的是最优预测变量阈值。本文选择 heuristic sure 作为阈值确定准则。 3. wd-rbf 水文时间序列模拟预测方法 3.1 基本思路 (1) 小波消噪阶段： ①小波分解。选择合适的小波函数，确定小波 分解的层次 n，对水文时间序列进行 n 层小波分 解。序列自身的变化特性是小波函数选择的重要影 响因素。而对于小波层数的确定，若层数偏小，低 频子序列仍是多个时间尺度上信号的混叠，且由于 序列能量压缩不明显，无法有效去除噪声；若层数 过大，较大分解层上会产生一些缺乏物理意义的虚 假信号，且会造成误差累积等问题（chou c m， 2011）。基于序列变化特点选择截频性能出色小波 函数 dmey(桑燕芳，王栋，2008)，并依据自相关白 化检验确定分解层数（李炜 , 陈晓辉 , 毛海杰， 2009）。②阈值消噪。对 1～n 层的每一层高频系数 选择合适的阈值 t 进行量化处理，得到去噪后的序 列成分。因为在硬阈值方法中，信号在 t 处是不连 续的，会给重构序列带来一定振荡。采取 heuristic sure 的软阈值方法对实例序列进行阈值处理。③ 小波重构。根据小波分解的第 n 层的低频系数和经 过处理后的第 l 层到第 n 层的高频系数进行重构， 得到消噪后的水文时间序列。 (2) rbf 网络模拟预测阶段： ①确定输入层神经元数。对原始数据进行归一 化处理。分别计算实例序列的自相关系数，判断序 列自身的延迟特性。取自相关系数显著趋于 0 所对 应的的延迟阶数 k 作为 rbf 网络输入层神经元数。 ②训练网络。在 rbf 网络训练中，隐层神经元数的 确定是一个关键问题，过去的做法是使其与输入矢 量数相等，显然过多的隐层单元数增加了运算的负 担。提出改进方法：从 1 个神经元开始训练，通过 检查输出误差使网络自动增加神经元；每循环计算 一次后，用使网络产生最大误差所对应的训练样本 作为权值向量产生一个新的隐层神经元，然后重新 计算，并检查新网络的误差，重复此过程直到达到 误差要求或最大隐层神经元数为止，网络训练完 成。③模拟预测。将消噪后的实例序列代入 rbf 网 络，以预测步长 k 进行预测模拟，以检验预测方法 的有效性和准确性。 wd-rbf 水文时间序列模拟预测方法流程如图 1 所示。 图1 wd-rbf模拟预测方法流程图 3.2 方法实现 选取黄河花园口站 1950-2003 年径流量，利津 站 1950-2003 年径流量，北京市 1951-2011 年降水 量，南京市 1951-2011 年降水量为实例。序列最后 5 年值作为验证数据不带入建模。 对实测水文时间序列进行 dmey 离散小波分解， 得到各个尺度下的小波系数。对各个尺度下的小波 系数进行 heuristic sure 软阈值消消噪处理，消噪 残差序列如图 2 所示。 published by atlantis press copyright: the authors 187 dengfeng liu, dong wang, yuankun wang et al. 图 2 显示各消噪残差呈现出一个矩形散点图且 没有明显的模式，表明没有出现明显偏离噪声随机 性 外 观 的 情 况 （ jonathan d.cryer, kung-sik chan,2008）。对各消噪之后的序列计算不同延迟阶 数 k 下的自相关系数(acf)。北京市 k-acf 计算结 果如表 1 所示。 表1 北京市年降水量k-acf表 k 1 2 3 4 acf 0.1670 0.0895 0.1811 0.1624 k 5 6 7 8 acf 0.1821 -0.0506 -0.0476 0.0393 表 1 所示， 延迟阶数 k 增大至 6 时，序列的自 相关系数显著降低。因此选取北京年降水量序列的 rbf 网络预测步长（即输入层神经元个数）为 5 （路剑飞，陈子燊，2010）。同理依次确定各时间 序列的网络输入层神经元数，结果如表 2 所示。 表2 rbf网络输入层神经元数对应表 序列 花园口 年径流 利津 年径流 北京 年降水 南京 年降水 神经元数 3 5 5 5 图2 各序列消噪残差图 (a) 花园口站年径流 (b) 利津站年径流 (c) 北京年降水 (d) 南京年降水 published by atlantis press copyright: the authors 188 wd-rbf model and its application of hydrologic time series prediction 确定输入层神经元个数，训练并建立相应 rbf 网络，运用建立好的 rbf 网络对序列进行模拟预 测。各水文时间序列 wd-rbf 模拟预测结果如图 3 所示。 4. 对比分析 选取的对比方法为基于序列线性非平稳假设的 arima 模型及非线性假设的 rbf 网络。对原实测 序列直接进行 rbf 网络模拟预测，并建立 arima 模型模拟预测(box g e p, jenkins g m, reinsel g c. 2011)。将 wd-rbf 模型与 arima 模型、rbf 网络 的模拟预测结果进行对比。 选取预测的绝对误差 ae(表 3)，预测的相对误 差 re(表 4)和模拟预测的均方误差 mse(表 5)三组指 标对各方法的准确性进行比较分析。 图3 各序列wd-rbf模拟预测图 (a) 花园口站年径流 (b) 利津站年径流 (c) 北京年降水 (d) 南京年降水 表3 arima/rbf/wd-rbf时间序列预测ae表 预测 年份 花园口站年径流序列/108m3 利津站年径流序列 /108m3 arima 模型 rbf 网络 wd-rbf arima 模型 rbf 网络 wd-rbf 1 62.27 47.66 -43.09 100.25 8.61 -63.42 2 53.66 280.63 260.7 36.26 30.99 -4.79 3 79.87 60.17 -12.14 203.54 105.18 34.05 4 107.65 89.58 2.02 69.75 92.63 6.15 5 -67.17 117.02 45.74 36.79 -63.15 -139.13 published by atlantis press copyright: the authors 189 dengfeng liu, dong wang, yuankun wang et al. 由表 5，wd-rbf 模型和 rbf 网络对于水文时 间序列的模拟精度较 arima 模型有显著提高。以 花园口年径流量为例，arima 模型模拟的均方误差 为 41.21，rbf 网络模拟的均方误差为 1.77，而 wd-rbf 模拟之后的均方误差最小，为 1.36。从模 拟的对比分析可以得出，wd-rbf 模型和 rbf 网络 的模拟精度高于 arima 模型， wd-rbf 模型的结 果较 rbf 网络有一定程度提高。 对于水文时间序列的预测，以花园口年径流量 为例，由表 3，arima 模型 5 年预测最小绝对误差 为 53.66，rbf 网络的最小绝对误差为 47.66， wdrbf 模型的最小误差为 2.02。又以北京年降水为 例，由表 4， arima 模型预测相对误差范围在 15%-20%之间，最小相对误差绝对值 6.53%，最大 为 20.51%；rbf 网络预测相对误差范围在 10%15% ， 最 小 相 对 误 差 绝 对 值 1.07% ， 最 大 为 16.82% ；而 wd-rbf 模型的相对误差绝对值为 0.78%，最大为 17.04%。在序列的预测上，wdrbf 模型和 rbf 网络的预测精度高于 arima 模 型，wd-rbf 模型在 rbf 网络基础上有所提高，但 提高幅度不大。 5. 结论 对水文时间序列加以科学的模拟预测是水灾害 防范的前提和依据。针对水文时间序列模拟预测中 的非线性特征和噪声污染问题，提出了 wd-rbf 模 拟预测模型。rbf 网络能够模拟出水文时间序列中 内在的非线性机制，小波消噪能够一定程度上识别 消除序列中的随机噪声，达到还原真实序列的目 的。实例表明，wd-rbf 模型在模拟预测中的精度 显著高于 arima 模型，并在 rbf 网络基础上有所 提高，因而表现出优于传统 arima 模型和 rbf 网 络的性能，具有更高的准确性。 然而 wd-rbf 网络较 rbf 网络并没有表现出 明显的优越性，其原因是否与消噪过程有关需要进 一步分析论证。针对不同信噪比的水文时间序列， 如何选择合适的小波函数和消噪阈值，仍值得进一 步研究。 致谢 资 助 项 目 ： 国 家 自 然 科 学 基 金 (no. 41071018) 、 国 家 重 点 基 础 研 究 发 展 计 划 ( 2013cb956503)、教育部新世纪优秀人才支持计划 (ncet-12-0262) 、 教 育 部 博 士 点 基 金 (20120091110026, 20100091120059)、江苏省教育厅 青蓝工程。 参考文献 1. powell, m. j. d. radial basis functions for multi-variable interpolation: a review, ima conference on algorithms for the approximation of functions and data, rmcs shrivenham, uk, 1985. 表4 arima/rbf/wd-rbf时间序列预测re表 预测 年份 北京市年降水序列 南京市年降水序列 arima 模型 rbf 网络 wd-rbf arima 模型 rbf 网络 wd-rbf 1 -19.26% -8.00% -7.78% -25.08% 0.03% -2.86% 2 -15.25% -11.35% -11.13% -8.85% 9.88% 6.77% 3 -6.53% 16.82% 17.04% -5.19% -27.80% -27.40% 4 20.51% 9.66% 9.42% -51.58% -17.63% -19.93% 5 -20.44% -1.07% -0.78% -29.23% -1.35% -4.98% 表5 arima/rbf/wd-rbf时间序列模拟预测mse表 方法 花园口站年径流序列 /108m3 利津站年径流序列 /108m3 北京市年降水序列 /mm 南京市年降水序列 /mm mse ① mse ② mse ① mse ② mse ① mse ② mse ① mse ② arima 41.21 76.47 43.31 108.64 57.57 101.96 343.59 355.58 rbf 1.77 65.23 2.38 31.41 2.93 25.06 0.56 90.70 wdrbf 1.36 53.69 2.11 31.37 2.94 24.85 0.33 92.73 mse ① ：模拟值均方误差 mse ② ：预测值均方误差 published by atlantis press copyright: the authors 190 wd-rbf model and its application of hydrologic time series prediction 2. d. s. broomhead and d. lowe. multivariate functional interpolation and adaptive networks [j], complex systems, vol. 2, 1988, pp. 321-355. 3. s. haykin. neural networks: a comprehensive foundation[m]. macmillan college publishing company: englewood cliffs, 1994. 4. donoho, d.l. de-noising by soft-thresholding [j]. ieee transactions on information theory, 1995. 41(3): p. 613627. 5. d. l. donoho and i. m. johnstone. adapting to unknown smoothness via wavelet thrinkage [j]. j. amer. statist. assoc., vol. 90, pp. 1200–1224,1995. 6. box g e p, jenkins g m, reinsel g c. time series analysis: forecasting and control [m]. wiley, 2011. 7. coifman r r, wickerhauser m v. entropy-based algorithms for best basis selection [j]. information theory, ieee transactions on, 1992, 38(2): 713-718. 8. lou h w, hu g r. an approach based on simplified klt and wavelet transform for enhancing speech degraded by non-stationary wideband noise [j]. journal of sound and vibration, 2003, 268(4): 717-729. 9. chou c m. wavelet-based multi-scale entropy analysis of complex rainfall time series[j]. entropy, 2011, 13(1): 241-253. 10. cryer j d, chan k s. time series analysis with applications in r[m]. springer, 2008. 11. zhang junfeng, hu shousong. chaotic time series prediction based on rbf neural networks with a new clustering algorithm [j]. acta phys. sin., 2007(2):713719. 张军峰，胡寿松. 基于一种新型聚类算法的 rbf 神经 网络混沌时间序列预测[j]. 物理学报, 2007(2): 713-719. 12. li jun, liu junhua. on the prediction of chaotic time series using a new generalized radial basis function neural networks [j]. acta phys. sin., 2005, 54(10): 45694577. 李军，刘君华. 一种新型广义 rbf 神经网络在混沌时 间序列预测中的研究[j]. 物理学报, 2005, 54(10): 45694577. 13. sang yanfang, wang dong. wavelets selection method in hydrologic series wavelet analysis [j]. journal of hydraulic engineering, 2008, 39(3): 295-300. 桑燕芳，王栋. 水文序列小波分析中小波函数选择方 法[j]. 水利学报, 2008,39(3): 295-300,306. 14. lu janfei, chen zichen. salinity prediction at modaomen waterway in estuary of pearl river. journal of china hydrology, 2010(5): 69-74. 路剑飞，陈子燊. 珠江口磨刀门水道盐度多步预测研 究. 水文, 2010(5): 69-74. 15. liu junping, chang mingqi. water demand prediction based on radial basis function neural network [j]. journal of china hydrology, 2007(5): 12-15，11. 刘俊萍，畅明琦. 径向基函数神经网络需水预测研究 [j]. 水文, 2007(5): 12-15，11. 16. yang xiaohua, yang zhifeng, shen zhenyao, et al. high precision radial basis function network model for predicting water environment nonlinear time series[j]. advances in water science, 2005(6): 788-791. 杨晓华，杨志峰，沈珍瑶等. 水环境非线性时序预测 的高精度 rbf 网络模型[j]. 水科学进展, 2005(6): 788791. 17. liu haitao, sun shuangke, liu zhiping, et al. atomization prediction based on artificial neural networks for flood releasing of high dams [j]. journal of hydraulic engineering, 2005(10): 1241-1245. 017. 柳海涛，孙双科，刘之平等. 泄洪雾化预测的人工神 经网络方法探讨[j]. 水利学报, 2005(10): 1241-1245. 18. gao guorong, liu yanping, pan qiong. a differentiable thresholding function and an adaptive threshold selection technique for pulsar signal denoising [j]. acta phys. sin., 2012(13): 549-553. 高国荣，刘艳萍，潘琼. 基于小波域可导阈值函数与 自适应阈值的脉冲星信号消噪[j]. 物理学报, 2012(13): 549-553. 19. zhao li, feng ji, zhai guangjie, et al. wavelet transformation for magnetocardiography signal [j]. acta phys. sin., 2005(4): 1943-1949. 赵莉，冯稷，翟光杰等. 小波变换在心磁信号处理中 的应用[j]. 物理学报, 2005(4): 1943-1949. 20. dou huili, liu haode, wu zhizhou, et al. study of traffic flow prediction based on wavelet analysis and autoregressive integrated moving average model [j]. journal of tongji university(natural science), 2009(4): 486-489，494. 窦慧丽，刘好德，吴志周等. 基于小波分析和 arima 模型的交通流预测方法[j]. 同济大学学报(自然科学 版), 2009(4): 486-489，494. 21. li wei, chen xiaohui, mao haijie. a study of adaptive optimal decomposition level in threshold de-noising algorithm based on wavelet transform [j]. computer simulation, 2009, 26(3): 311-314. 李炜, 陈晓辉, 毛海杰. 小波阈值消噪算法中自适应确 定分解层数研究[j]. 计算机仿真, 2009, 26(3): 311-314. published by atlantis press copyright: the authors 191 atlantis press journal style on risk and disability – investigating the influence of disability and social capital on the perception and digital communication of risk jörgen sparf department of social sciences, mid sweden university, shv, östersund, 831 52, sweden e-mail: jorgen.sparf@miun.se susanna öhman department of social sciences, mid sweden university, shv, östersund, 831 52, sweden abstract does living as a disabled person affect how risks are perceived? studies of various population groups show that not only macro level structures but also micro social contexts and individual conditions influence risk perception. this knowledge is important for formulating, designing, and communicating risk information. the aim of this study is to examine how d isabled people’s risk perception and pref erence for digit al risk communication channels are influenced by disability in itself and by social capital. a quantitative survey (n=6 500) was carried out in sweden showing that while disability is not influential, social capital is. this differs from the results of previous studies on other population groups, which show that group specific factors do influence risk perception. this study suggests that due to the importance of social capital, institutions communicating risk information should build strong relationships with (local) disability associations and networks in order to communicate more effectively. keywords: disability; risk perception; risk communication; social capital. 1. introduction risk is a cent ral aspect of modern life. much of today’s political life and news media revolve around risk-issues and individuals are c onstantly informed about concerns – from local to global. not only are timeless risks such as natural hazards, shortages of food and water supplies, and emergen-cies of immediat e interest but also riskissues that are a result of modern lifestyle characterized by affluence, industrial food production, sedentary work and leisure habits, and increased longevity; along with surveillance and cyber risks, macro economic shocks, climate change, urbanization, and the downsides of new technologies. the pervasiveness of m odern risk issues has led to an increased interest in how risks are c ommunicated with the population. to facilitate communicators’ and policy makers’ decision-making on communication strategies and risk priorities,for example in social work, and for analyzing social vulnerability, knowledge about how people perceive risk is important. in order to prioritize and decide on a cceptable levels of ris k, statistical risk assessments need to be balanced towards perceptions of risk as a reflection of values, for example in order to establish reliability and confidence. thus, the multidisciplinary research area of ris k perception has contributed greatly to the understanding and developing of communication about risk and hazard. the research project ‘risk perceptions and sensemaking of risk in european societies’, which this study is part of, focuses on the risk perception of various population groups. studies show that different ethnic groups, groups of different sexual orientations, and young people perceive risks in different ways compared to a majority of the population (e.g. olofsson & rashid 2011; wall & ol ofsson 2008). to a larg e extent these journal of risk analysis and crisis response, vol. 4, no. 1 (march 2014), 20-33 published by atlantis press copyright: the authors 20 willieb typewritten text received 5 january 2014 willieb typewritten text willieb typewritten text accepted 10 february 2014 willieb typewritten text j. sparf & s. öhman differences are explaine d by a com bination of sociostructural differences (income, education, demography, religion etc.) and ind ividual conditions (cognition, emotion, traits etc.). however, some group-specific differences remain after adjusting for these factors. the differences have been found to emerge from variations in fundamental values, the character of risk-related experiences, the effects of structural discrimination, and changing cultural contexts. as part of the project this study takes a closer look at disabled people. there are two particular reasons for studying disabled people in th is context. firstly, th e chronological evolvement of sociological risk theory has moved from a soci etal perspective to an organizational governance-perspective (from macro to meso). at the same time, studies of risk perception have evolved the other way, from a co gnitive heuristics perspective to studies of social psychological explanations (from micro to meso). at the point of intersection, where these developments meet, cutting edge research on risk perception is to b e found. examining group-specific characteristics have proved to contribute to a more nuanced understanding of risk perception, which may serve positive purposes for th e group in question. secondly, in emergency management guidelines various population groups are commonly identified as vulnerable – disabled being one of the most pervasive ones. the research project therefore seeks more knowledge on the risk perceptions of these groups in order to provide knowledge to develop adjusted guidelines. in everyday life, ris k is a ub iquitous phenomenon for disabled people. a qualitative study of immobilized people showed that not only where they subject to riskrelated decisions like anyone else, they also had to make constant risk-related decisions directly connected t o their physical state (spa rf, 2013). for ins tance, on a daily basis the interviewees choose between avoiding or not avoiding activities and situations referring to their physical abilities, and whether or no t to display their disability-related vulnerability. this recurrence of risk in everyday life makes the necessity and utility of social capital more concrete and direct than for many other people. however, disabled people generally have l ess social capital than non-disabled people (statistics sweden 2006; 2009). the combination of a high demand for and low levels of social capital presumably influences how risks are perceived. the research question is therefore whether living as a disabled person really does influence how risks are perceived– if there is a group specific character of risk perception – and whether the level of social c apital affects how disabled people perceive risks. this knowledge can then be applied by policy makers and practicians in working with social inclusion and safety. if low so cial capital results in h igh risk p erception, fighting social isolation by enhancing the social ca pital can serve as a ris k remedial strategy to reach both actual and perceived personal safety. the survey in this project focused on self-defined disability. therefore, a subjective definition of disability is applied (grönvik 2007; szebehely, fritzell & lundberg 2001). since ‘definitions of d isability’ is no t the topic of study, the term is used here in a ge neral sense. the effects on risk perception from various specific medical conditions are not examined. the main focus is on general patterns for disabled people as a whole. knowledge from risk perception studies are applied in risk information, such as health, environment, and personal security. authorities, for example, often want to target speci fic groups. so me typical exa mples are: teenagers regarding alcohol, drugs and sexually transmittable diseases; i mmigrants regarding police work and fire safety; middle aged men about prostate cancer; and disabled people about instrumental aid, social services and accessibility. in order to produce risk information that can b e appropriately assimilated by individuals in different targeted groups, knowledge is needed of how perceptions vary and what is influencing the perception. by the same token, misunderstandings and disinformation can possibly be identified and corrected. knowledge regarding group-specific effects on risk perception is valuable when formulating, designing and communicating risk information, for e xample, what to stress in texts, what photos and illustrations to choose, and what channels of communication should be used. from a civil rights perspective, all people have the right to get appropriate risk information communicated in an accessible way (olofsson 2007). however, swedish authorities have seldom taken measures to adapt risk communication to match the nee ds of people with different types of i mpairments (sparf, forthcoming[b]) even though digital communication offers a hug e potential for a djustments. to create a ba sic knowledge published by atlantis press copyright: the authors 21 on risk and disability of the demand for digital communication, the preference of disabled people to use various digital channels for risk information is also studied. in short, the aim of the study is to examine how disabled people’s risk perception and preference for digital risk communication channels is influenced by disability and by social capital 2. risk perception the concept of risk perception refers to the su bjective mindset and attitudes regarding risks (slovic 2000). these attitudes, which are m ore or less exp licitly articulated in judgements regarding the characteristics and severity of risks, are derived from and expressed in individual concerns and needs in various situations and fields of interests. it is well established that, in a given population, risk perception differs between various groups (olofsson and rashid 2011). studies of gender, age, various ethnicities, and urban-rural residency show that groupspecific factors such as cultural values, experiences, and physical contextual features all infl uence risk perceptions. in disability research, st udies of risk perception are lacking. risk-related studie s primarily concern speci fic conditions such as learn ing disabilities, and mental and cognitive conditions (e.g. alaszewski&alaszewski 2002; dowse 2009; heyman & huckle 1993; seale & ni nd 2010), or more sociological issues, such as marginalization, normalization, independence, equality, and gender (french gilson 2004; barron 2001; kvalsund and velsvik bele 2010; gustavsson 2005), economy and em ployment (mouridsen and hauschild 2009; zissi et al. 2007), violence and crime (cederborg and gumpert 2010; johnston 2002; olsvik 2006), depression and su icide (meltzer et al. 2012), and living conditions (tøssebro and kittelsaa 2004). one long-standing debate concerns how subjective risk perceptions come about – what factors influence how various risks are perceived? possible explanations are often divided into three major schools: cognitive/psychometric, social/cultural, and interdisciplinary. the cognitive school argues th at risk perception is to be understood in terms of information processes (tversky and kahneman 1974). people use cognitive heuristics in sorting and sim plifying information, which leads to biases in comprehension (e.g. finucane et al. 2000). a more developed version, the psychometric paradigm, adds a num ber of explanatory factors, such as dread, newness, and stigma to understand the complexity of risk perception (slovic, fischhoff, and lichtenstein 2000; slovic 2000). the cognitive/psychometric school focuses on mental aspects and moves away from co gnition only as far as understanding the relative influences of di fferent kinds of knowledge e.g. intuitive knowledge, formal knowledge, and social learning. recently though, the paradigm has been expanded with research concerning the emotional perspectives on risk perception (slovic 2010). in 1982 mary douglas and aaron wildavsky added social construction arguments to the risk p erception debate, primarily drawing on ‘cultural theory’ (adams 1995; douglas 1985; 1992; douglas and wildavsky 1982; lupton 1999) and the grid-group framework (douglas 2003). this approach argues that risk perception is highly affected by constraints in the individual’s social roles and by feelings of solidarity towards groups that he / she belongs to (wildavsky and dake 1990). in these constraining structures of grids and groups, risk perceptions are socially constructed by social institutions and cultural values. the third school is an attemp t to in tegrate research from psychology, sociology, anthropology, and communications theory in order to present a ge neral model for risk perception. the social amplification of risk framework (sarf) shows how risk information is communicated through a number of individual and social ‘stations’ such as experiences, social groups, government agencies, and news media (renn et al. 1992; pidgeon, kasperson, and slovic 2003). all of these stations act as filters b y amplifying or attenuating the information with reference, for example, to heuristics, values, politics, and social and societal considerations. both the amplification/attenuation process itself and the actual ‘outcomes’ – financial losses, regulatory actions, organizational changes, changes in confidence etc. – are believed to influence the individual perception of risks (pidgeon, kasperson, and slovic 2003). although all three schools significantly contribute to the explanation of risk perception, one level of analysis is m issing. while the individual level is covered by cognitive explanations and the macro level by the interdisciplinary research, po ssible explanations between these two levels has not been covered. t his could be ac hieved by examining the indivi dual’s social published by atlantis press copyright: the authors 22 j. sparf & s. öhman capital and its in fluence on risk perception. this factor is to be found on a m eso level and concerns the individual resources and context in terms of the physical, emotional and social life– in short, the intimate (social) life of every day. social capital see ms to fall b etween the cognitive/psychometric paradigm and th e social/cultural paradigm. sarf does include ‘social context’ as a c omponent in the individual ‘station’ but the content of th e component is neither thoroughly explained nor defined with reference to any specific analytic level 3. social capita theories of social capital have been formulated and adapted in se veral academic disciplines, ranging from coleman’s individual approach (coleman 1988), via bourdieu’s classand culture-perspective (bourdieu 1984; 1986), to putnam’s focus on issues of social cohesion and democracy (adkins 2008; putnam 1995; 2002). one common objective, regardless of discipline and perspective, is to understand and describe the interaction between social structures and social behaviour (cook, burt, and lin 2001; lin 1999, 2002). most scholars argue that social capital boils down to networks,trust, and norms of reciprocity (lin 1999; isham, kelley, and ramaswamy 2002; skrabski, kopp, and kawachi 2003; staveren 2003). to gain access t o the capital embedded in s ocial networks, time and commitment are needed. since th is entails some sort of risk-taking, trust is a fundamental prerequisite for every investment initiative. trust primarily refers to in tegrity and confidence, but also c oncerns norms of reciprocity – all members of the network expect more or less explicitly to get something in return. when these norms work smoothly, the strength of the network is enhanced. networks are basically related to how many people an individual knows and the kind of relationships. several categorizations of s ocial relations a re prevalent in empirical research: family, friends, ac quaintances, co-workers, neighbours etc. granovetter (1973) distinguishes between strong ties and weak ties i n network relations. while strong ties refer to tight relations in well-defined groups, for example a family, a criminal gang, some close friends or co-workers, weak ties refer to relations between groups as well as between ‘second order’ relations, acquaintances, a friend of a friend etc. a similar but more functionalistic approach is taken by putnam (1995) who distinguished between bridging social capital and bonding social capital. bridging is the connectedness that is formed across diverse social groups, while bonding refers to the cementation of homogenous groups. trust is usually divided into private trust and public trust. private trust concerns the individual’s trust in other individuals, for instance regarding people's intentions and whether they actually act ou t these intentions. public trust refers to the expectations of public organizations, for instance the efficient delivery of appropriate services, and the fair and legal h andling of cases (e.g. within s ocial security or the judicial system). in sociological theory this division of trust is commonly called individual trust and system trust (giddens 1991; 1990). in so cial capital research, macgillivray and walker (2000) call them informal and formal trust, uphoff and wijayaratna (2000), cognitive and structural trust, and stone (2002), social and civil/institutional trust, respectively. this study adheres to these divisions by examining trust in people and trust in institutions. finally, norms of reciprocity, refer to t he expectations and practices among network members regarding the flow of knowledge and information (dekker and uslaner 2001; woolcock 1998). the types of exchanges in this flow spans from hands-on practical favours and help, to in tangible confidence concerning thoughts, ideas, feelings, experiences etc. exchanges are not to be understood as one-to-one or type-to-type, but rather they are c ontinuously and dynam ically intertwined in a co mplex weave of bo th explicit and tacit reciprocal exchanges. in some texts it is difficult to see the distinction between trust and norms of reciprocity. the d istinction can be illustrated by a b ank metaphor (cf. portes 1998). trust means that the ba nk stores your money safely, while reciprocity refers to the mutual interest for you (e.g. return on investments) and for the bank (e.g. shares of th e return). in this study reciprocity is operationalized by measuring regular contacts with fam ily, friends and neighbours, and whether or not there is someone to turn to for practical help or emotional support. the interrelatedness between social capital and disability is w ell established. this study is b ased in sweden, where statistics sh ow that both t he structural and individual resources of disabled people are published by atlantis press copyright: the authors 23 on risk and disability generally lower than for non-disabled people (statistics sweden 2006; 2009). this could be explained by the fact that the majority of disabled people are elderly (national board of health and welfare 2011; szebehely, fritzell and lundberg 2001) and that the bulk of social capital becomes less with o lder age (statistics sweden 2006; 2009). international studies of social capital show that the further north and west in europe, the higher the social capital (carlsson 2004; kääriäinen & lehtonen 2006; pichler & wallace 2007(oorschot, arts, and gelissen 2006)). considering the similarity in cu lture and level of welfare in th e nordic countries, it is plausible to assume that the national distribution of social capital betwee n various groups is comparable, which would therefore also apply to the results of th is study. 4. risk communication as an accessibility issue one practical matter connected to risk perception is how information about risks is communicated in an effectiv e way. matters of risk communication have to a large extent been studied from a sender-perspective, e.g. public bodies and private companies, and a wide range of aspects related to communication has been adapted to a risk and crisis context (c.f. heath and o'hair 2009). in spite of the vast research-based body of knowledge concerning variations in risk perception, organizations have treated the communication of risk and crisi s information rather uniformly. a co mmon statement is ‘equal information for everyone’ (olofsson 2007). in this article a slightly different approach is taken by addressing the question of the preferred digital communication channel from a receiver-perspective. physical and mental conditions could affect a person’s ability to assimilate information. although a wide selection of ad justable digital hardware and software that can facilitate th e transfer of vital information is av ailable, it is sel dom used fo r risk information (sparf, forthcoming[b]). the reasons include a lack o f knowledge about the needs from the receiver-side, and t hat, public bodies and other organizations on the sender-side are reluctant to invest in technology if it d oes not realistically meet the needs from the receiver-side (ibid.). by studying the disable d as a g eneral target group, knowledge is g ained that could highlight the possible changes needed for the communication strategies of organizations. 5. method as part of the research project ‘risk perceptions and sense-making of risk in european societies’ the authors and colleagues at the risk a nd crisis research centre, mid sweden university, developed and coordinated the survey ‘society and values’. the main idea for the survey was to explore connections between risk perception, risk information, risk behaviour, values, and resources, specifically looking at various groups of people. this study is part of the project. the survey is based on a number of earlier surveys (primarily enander and joh ansson 2002; warg and wester-herber 2001; sjöberg 2000) and a pilot study of five focus group interviews (olofsson, öhman, and rashid 2005). as a consequence the survey consisted of well-established items and qu estions regarding risk issues, as well as completely new ones. risk perception was measured through 16 claims about risk-posing threats to th e individual that respondents were to take a stand on (sjöberg 2000). no specific type of risk was fo cused on in t he survey – rather the set of questions comprised a vast range from accidents during leisure tim e activities and smoking, to climate change and terrorism. the majority of common risk typologies were c overed in the survey: individualcollective, local-global, short term -long term, rapid onset-slow onset, and influenceable-not influenceable. the answers indicated the respondent’s estimates of risk of harm, which by using factor analysis formed three factors: known risks, controllable risks and dr ead risks. examples of ‘known risks’ are the individual’s risk of having a traffic acci dent or devel oping cancer, examples of ‘controllable risks’ are drinking alcohol and smoking. dread risks are risks which are not easily controllable and which pose a wider threat to humans, such as climate change and terrorism. these factors are taken from a l ong-standing research study on risk perception (e.g. slovic, fischhoff, and lichtenstein 2000). the questions on social capital were derived from surveys and reviews with empirical evidence of stabile measuring (harper 2001; spellerberger 2001; eriksson 2003; ruston and akinrodoye 2002). following this level of analysis and the three social capital dimensions, a range of questions re garding everyday life were chosen. table 1 s hows the indicators and t he operationalization. published by atlantis press copyright: the authors 24 j. sparf & s. öhman personal characteristics known t o influence risk perception were included as control variables: sex (male/female), age (six equally sized categories from 16 to 75 years ) and experiences of ris ks-/crisis situations. the last variable was measured by asking if the respondent had personal experiences of fire, natural disaster, violence, accidents during leisure time activities, serious diseases or traffic accid ents. the answers were indexed ranging from 0 (none) to 6 (all). disability was id entified by asking ‘are you disabled?’.more detailed questions about the respondents’ bodily state o r quality of disability were omitted from this study for two reasons. firstly, the aim of the study was to descriptively explore the relation between self-perceived disability and the risk issues in general, not to investigate details on how different kinds of medical conditions influence risk p erception. secondly, the space allocated in the s urvey for thi s study was unfortunately extremely limited. the detailed items would simply have required too much space. this is a weakness in t he study. there are a number of factors possibly affecting risk perception in various ways, e.g. variations of mental and ph ysical conditions, general degree of disability, root causes for disability etc. to gain deepe r knowledge in this area, future studies would definitely need to include more detailed questions. table 1 operationalization of variables. indicators operationalization independent variables* activities regular attendance at organized events and visits to public events/places. social network regular contacts amount of contact with family, friends and neighbours. trust in people extent of trust in other people trust trust in institutions extent of trust in a number of organizations and institutions. norms of reciprocity social inclusion feelings of being part of, or belonging to, the society/community. dependent variables known risks e.g. the risk of having a traffic accident or developing cancer. controlled risks e.g. drinking alcohol and smoking. individual (54.4 % explained variance) dread risks e.g. climate change and terrorism. national welfare risks school quality, financial crisis, ageing population, living conditions, corporate migration, children without moral standards, public health. international risks environmental hazards, nuclear power, unknown illnesses, chemical accidents. risk perception societal (55.6 % explained variance) global challenges migration, population growth, religious fundamentalism. risk communication digital information channel assumed use of: webpages, sms, email, and expert via email/webpage for getting risk information. * income was also included in the analysis of capital. the questions on ri sk communication concerned to what extent a number of digital tools for communicating risk information were likely to be used by the respondent. the answers were given on a five-graded scale ranging from “not likely at all” to “very likely”. a first round of the survey was distributed from november 2005 to january 2006, and a second round in september 2008. for each r ound of the su rvey, a national random sample of postal addresses was selected from statens pe rsonadressregister (spar) * , which is an ad ministrative section of th e swedish tax agency. the official regional ethical review b oard† investigated the survey and sample without comment. after reminders by mail and telephone call, th e total response rate for each round was 47%, and 39% respectively. to reach a l arger absolute number of disabled respondents, the total of both rounds of the survey was used for analysis. all analyses were ca rried out with dummy variables for the two rounds to ensure *http://www.statenspersonadressregister.se/om-spar/in-english.html †http://www.epn.se/en/start/startpage published by atlantis press copyright: the authors 25 on risk and disability that no si gnificant differences between the rounds existed. the respondents were ev enly distributed between men and women and t he average age was somewhat older for disabled people (51) than for nondisabled (46). eight per cent (n=223) reported themselves as suffering from disability in some way. even though different operationalizations of disability in survey research often yields widely varied results (hugaas molden & tøssebro 2010), for an app lication of self defined disability the figure is st rikingly low.two possible explanations were identified. firstly, the survey did not explicitly examine disability and included the disability question in the background items. surveys explicitly dedicated to disability issues might attend more interest from people perceiving themselves disabled – especially if there is reason to believe that the study could enhance the general situation for disabled. secondly, especially elderly people, among which a vast majority of disabled are found, deterred from taking the survey due to the extensiveness with over 800 items. the lower average age of disabled in this study points at that. the total response rate of 43% and 223 disabled respondents is sufficient to analyze whether there were any significant correlations for this group and in what directions they run. however, the low response rate raises a warning that the results and conclusions should be considered more as i ndications for future research rather than generalizable to a wider population. 6. results in this section, after a sho rt description of the small differences in social capita l between the groups, a thorough description is given of the connections between social capital vis-à-vis risk perception and risk communication. concerning social capital, a gene ral variation between disabled and non-disabled was apparent which confirms the results from other studies. although the ingroup-variation was high regarding network, in our survey disabled people participated (passively or actively) to a lowe r degree than non-disabled people in activities such as culture, entertainment, sport, religious work, and ev ening courses (1.74 and 1.96 respectively, index 0-8). disabled people had a spouse or partner to a lower extent than non-disabled, they lived in smaller households, and had less regular contact with family and friends than the non-disabled (3.98 and 4.05 respectively, index 1-5). also, disabled people were more alone during leisure time than the non-disabled. for trust/security the pattern showed th at disabled people had significantly lower general expectations of other people, other peoples’ willingness to help, as well as trust in other people (1.79 and 2.15 respectively, index 0-3). the same pattern appeared in relation to trust in institutions, regardless of the type of institution, such as au thorities, political bodies, trade unions, private enterprises, mass media or political parties (3.16 and 3.34 respectively, index 1-5). social inclusion also varied between disabled and non-disabled. disabled respondents regarded the swedish society to a higher extent as not being open for them. in addition, although they stated that they were living like the average (“normal”) swede, they did not feel as th at they fitted into the community (3.35 and 3.37 respectively, index 1-5). all of t hese differences are rather sm all, and a s stated earlier is base d on a s mall sample. however, as all the regressions have the same directions we s hould pay attention to the differences. regarding incomes, only 21.6% of the disabled had an income of more than sek 20,000 (€ 2,250) per month, compared to 40.9% for the non-disabled. the a verage income in swe den year 2008 was sek 29,275 (€ 3,293) per month (statistics sweden 2009). 6.1. perception of risks for the individual for each of the three risk factors (controllable risks, known risks, and drea d risks), only minor differences between the examined groups were found. one interesting finding is th at the in-group variations differed markedly, showing that disabled people were much more scattered in their perceptions than t he nondisabled. the three risk factors were analyzed in relation to social capital (plus income: control group=low income) in a multiple regression (table 2). the results clearly show that social ca pital is important for understanding the perception of risks for the individual, especially reg arding controllable risks and dread risks. the only resource with a significantly negative influence on perception in all three risk factors was trust in other people, i.e. the higher the trust in other published by atlantis press copyright: the authors 26 j. sparf & s. öhman people, the lower the risk perception. social inclusion showed the same pattern while for trust in organizations, the influence was reverse – th e higher the trust in organizations, the higher the risk perception. notably, experience of risk also had a strong positive one-way influence. 6.2. perception of societal risks concerning societal risks, disabled people had a slightly higher risk perception of all three factors, with the largest being for global challenges; however these results were not significant. the three risk factors were analyzed in t erms of social capital (plus income: control group=low income) in a multiple regression (table 3). most notable from this table is that, once again, trust is the most dominant factor influencing risk perception. in addition, activities seem to have some influence – the more socially active a pe rson, the lower the perce ption of international risks and global challenges. just as in the case of risks for the individual, previous risk experiences had a on e-way positive correlation with societal risk perception. table 2 multiple regression: individual risk perception and capital. (p=/<0.05, non significant beta within brackets). n= 2805. controllable risks n= 2813 known risks n= 2812 dread risks n= 2805 group beta beta beta disabled people (0.01) (0.02) (0.19) social capital activities -0.088 (0.01) (-0.03) family -0.032 (0.02) (-0.02) friends 0.069 (0.02) (0.00) trust. people -0.126 -0.087 -0.163 trust. organizations 0.061 (0.02) 0.053 income -0.061 0.061 -0.089 social inclusion -0.041 (-0.04) -0.046 control variables sex 0.093 -0.042 -0.123 age 0.038 -0.049 0.117 experiences of risks 0.045 0.158 0.083 adjusted r2 0.048 0.040 0.070 6.3. risk communication in line with the other findings, disabled people reported a lower probability of using various information techniques (table 4). once again resources seem to have a high explanation rate. on ly this time income and social inclusion were the resources with th e strongest influence, showing that the higher the income and the higher the feeling of being socially included, the higher the likelihood of using digital risk information channels. table 3 multiple regression: societal risk perception and capital. (p=/<0.05 non significant beta within brackets). national welfare risks n= 2811 international risks n= 2812 global challenges n= 2812 group beta beta beta disabled people (-0.01) (0.01) (0.03) social capital activities (0.01) -0.049 -0.113 family (0.03) (0.00) (0.00) friends (0.01) (0.02) 0.046 trust: people -0.087 -0.076 -0.146 trust: organizations -0.086 0.056 -0.095 income (0.02) -0.150 (0.00) social inclusion 0.100 -0.075 (0.02) control variables sex -0.141 -0.215 0.093 age -0.044 0.034 0.132 experiences of risks 0.085 0.018 0.039 adjusted r2 0.049 0.102 0.087 published by atlantis press copyright: the authors 27 on risk and disability for some risks, the significance of the family seems to matter. the results showed that an intimate relationship with close family results in a lower probability of using the web channels. the regression were definite both for social capital and for the c ontrol variables, where the positive influences were in come and social inclusion and the negative being family, age a nd experiences of risk. to sum up, out of 27 capital indicators in the study, 13 had a si gnificant influence on individual risk perception, 12 on societal risk perception, and 13 on risk communication channels. the m ost influential indicators in the study were social inclusion and income (9 each), then in a falling scale: trust in people (6), trust in organizations (5), family (4), activ ities (3), and friends (2). for risk p erception, of the 25 significant correlations, 18 were negative and only 7 positive. this tendency is also the same for trust, which is the social capital component with t he biggest influence on risk perception. out of 11 significant correlations, 8 were negative and only 3 positive. this means that the higher the trust in people and organizations, the lower the risk perception both for the individual and society. although the analysis is carried out on a small sample and the pvalue of each analysis was rather low, the pattern is clear: social capital does matter and is i mportant in understanding risk perception. table 4 multiple regression: assumed use of digital information channel for various risks and capital. (p=/<0.05 non significant beta within brackets) n= 2801. domestic fire n= 2807 leisure time accident n= 2807 chemical accident n= 2807 epidemics n= 2806 natural disaster n= 2806 group beta beta beta beta disabled people (0.01) (-0.01) -0.051 (0.00) (-0.02) social capital activities (0.02) (0.01) (0.01) (0.02) (0.01) family (-0.02) -0.032 (-0.02) -0.032 -0.057 friends (0.00) (0.02) (0.00) (0.01) (0.00) trust: people (0.00) (-0.01) (-0.01) (-0.02) (0.00) trust: organiza tions (0.01) (0.01) (0.00) (0.01) (0.01) income 0.089 0.103 0.042 0.066 0.060 social inclusion 0.299 0.268 0.215 0.314 0.286 control variables sex 0.041 0.034 (0.02) 0.031 0.046 age -0.325 -0.326 -0.209 -0.302 -0.281 experiences of risks -0.037 -0.036 (0.00) -0.035 -0.050 adjusted r2 0.208 0.192 0.092 0.202 0.178 7. discussion the main finding of the study is that risk perception and the use of digital risk communication channels are affected by social cap ital but not by disability. this is thus different from previous studies on other population groups that showed that group specific factors do influence risk perception. the assumption that being disabled affects how risks are perceived has not been empirically confirmed. regardless of whether a person is disabled from birth or from later in life, the bodily state becomes an integrated part of bo th the p erson’s identity and everyday life. on the other hand sparf (2012) shows that stressful, disability-specific situations, revealing vulnerability, do occur in e veryday life which possibly affect ris k perceptions. in practical ri sk communication, the disabled are usually treated as a homogeneous target group. however, due to varieties in medical conditions, the ingroup variation is large regarding needs and preferences regarding how to receive risk information and how t o facilitate two-way c ommunication. because of t he non precise nature of th e disability data in this study, published by atlantis press copyright: the authors 28 j. sparf & s. öhman analyzing in-group differences is not feasible. therefore the results do not lend themselves to inform adjustments of emergency guidelines, as this would probably require more specific data. what is possible, however, is a discussion of the implications for communicating with the disabled with regard to the influence of social capital. the pattern from the multiple regressions is quite clear: the m ore social capital an individual can access, the lower the perception of risks and the higher the likelihood of using digital risk information channels. this pattern is discussed below for each of the social capital factors. trust is the stronge st social capital factor influencing risk perception. disabled people generally seem to have a l ower degree of t rust both in other people and in in stitutions. since low trust correlates with high risk p erception, a reason able implication is that risk communicators must work with trust issues in parallel with risk communication, in order to enhance the effectiveness of the communication. the function of trust operates on two levels: societal / systemic and interpersonal. the societal / syste mic level concerns (1) trust in modern societal systems of experts and technology (giddens 1992) and (2) trust as a socio-cognitive mechanism to reduce social complexity in the modern society (luhmann 1979:150). social complexity demands individuals to accepta certain amount of uncertainty. in a con stant flow of information trust helps us to rely on t he world to functioning without us having a to tal knowledge about everything or to have control over the course of events. trust on an interpersonal level, concerns integrative aspects, for instance being part of a community, a social group or a society in general. parsons (1978) asserts that trust resides in the individual’s belief that others will put their self-interest aside in favour of a collective orientation. the in tegration thus refers to attitudes and experienced feelings rather than a factual belonging, e.g. citizenship. trust can also be a lubricant for cooperation. this is a return to co leman and his rational choiceinterpretation of social networks. coleman (1990) assumes that actors are not only rational but a re also unconstrained by norms and are purely self-interested. trust within a social network serves as a replacement for normative monitoring and sanctioning. this means that the smaller the network, the better the trust serv es as a lubricant. for institutions communicating risk information, both these levels can be approached by establishing reliability and confidence. by b uilding strong relationships with (local) disability associations and networks, they show trustworthiness and consistency. a confident source of in formation is quite simply more likely to be listened to. the fact that disabled people take part in social activities to a lesser ex tent and have a smaller so cial network raises another challenge for risk communicators. the c hallenge is not that of reaching disabled individuals – t his can be acc omplished by personally ad-dressed mail and via radio, television, the internet, email etc. – th e challenge is rather to ensure that the information is fu lly assimilated. the more controversial or important the issue, the more we tend to talk about it with friends and co lleagues. by discussing issues, we internalize information and socially construct opinions and knowledge. this process is possibly harder to fulfil for socially disconnected or isolated individuals. connecting this reasoning with granovetter’s (1973) assertion on strong and weak ties, risk communicators could make active use of the weak ties in the disability community. for instance, people engaging in disability associations, usually do no t only have strong ties to other highly engaged people but they also know about other disabled people that are less, or not at all, engaged. these weak ties could be utilized as a catalyst for risk communication and ensure that risk information is assimilated within the disability community. the fact that disabled people are formally organized is a strength compared to other groups w ith low social capital. by liaising with local disability associations, subgroups are bridged together (c.f. putnam 1995), thus enhancing the effectiveness of risk communication. the third social capital factor in this study, norms of reciprocity, has the function of fostering solidarity and cohesion, and th ereby creating the stability and permanence of the net -work. the c ollective norms are reinforced by social control and will only function for initiated and active people. this study shows that disabled people are s ocially included to a lesser extent than the non-disabled. for a socially disconnected person, any reciprocal aspe ct with beari ng on risk information is d ifficult to collectively reflect u pon. just as for networks, the process of internalizing information and socially constructing opinions and knowledge can published by atlantis press copyright: the authors 29 on risk and disability be hard to achieve. here too, the aspect of weak ties and relationship-building is relevant for risk communication. final remarks: although defining the disabled as a group within a larger population, with regard to socio-economic status, ‘disability-culture’, or shared experiences can be of theoretical significance, this study does not validate such a d efinition empirically. yet th is is co ntinuously done in practical work on risk communication. as the results from the study indicate,identifying group-specific variables and analyzing their effects can be important in developing effective risk communication. c oncerning disabled various medical conditions and i mpairments might affect th e ability to assimilate risk information, hence there is a larg e potential to use adapted digital communication channels. however, in order to improve risk communication, further research regarding the connections between social capital, risk p erception and v arious impairments are needed. one limitation of th is study is the small number of relevant respondents (n=223). however, the group is big enough to show a clear pattern of significant correlations in the analysis. in order to validate the results from this study more research would be necessary on a l arger sample of disabled people. another limitation is th at the group is treated as homogeneous. although the respondents happen to bear a common feature, the in-group variation is probably very wide, both in term s of individuality and types of disability. therefore, since risk information is often communicated to s pecific groups, and si nce the res ults from the analysis show a m arkedly big in-group variation, a reasonable next step would be to exam ine any in-group patterns related to risk perception. the more detailed knowledge available, the better adapted the risk information and communication channels will be. references 1. adams, j. 1995. risk, london: university college london. 2. adkins, l. 20 08. social capital put to the test. sociology compass, 2(4):1209-1227. 3. alaszewski, a. andh. alaszewski, 2002. towards the creative management of ris k: perceptions, practices and policies,british journal of learning disabilities, 30(2): 56-62. 4. barron, k. 2001. autonomy in everyday life, for whom? disability & society, 16(3): 431 447. 5. bourdieu, p. 1984. distinction : a social crit ique of the judgement of taste, london: routledge. 6. bourdieu, p. 1986.the forms of capital.in handbook of theory and research for the sociology of education, ed. j. richardson, new york: greenwood. 7. carlsson, p. 200 4.the european health divide: a matter of financial or social capital? social science & medicine, 59(9): 1985–1992. 8. cederborg, a-c. and c.h. gumpert, 2010. the challenge of assessing credibility when children with intellectual disabilities are alleged victims of abuse. scandinavian journal of disability research, 12(2): 125 140. 9. coleman, j. s. 1988. social capital in the creation of human capital. the american journal of so ciology (supplement: organizations and institutions: sociological and economic appr oaches to the analysis of social structure), 94: 95-120. 10. coleman, j. s. 1990. foundations of so cial theory. cambridge, mass: harvard university press. 11. cook, k., burt, r. s. and n. lin, 2001 .social capital : theory and re search, sociology and economics. new york: aldine de gruyter. 12. dekker, p. ande.m. uslaner, 2001. introduction. in social capital and participation in everyday life. ed. p. dekker and e. m. uslaner, london: routledge. 13. douglas, m. 19 85.risk acceptability according to th e social sciences. new york: russell sage foundation. 14. douglas, m. 20 03. natural symbols : explora tions in cosmology. london: routledge. 15. douglas, m. an d a. wild avsky, 1982. how c an we know the r isks we face? why risk selection is a social process. risk analysis, 2(2): 49-51. 16. dowse, l. 2009. some people are never going to be able to do that: challenges for people with intellectual disability in the 21st century. disability& society, 24(5): 571-584. 17. enander, a, and a. johansson,2 002.säkerhet och risker i vardagen: en studie av uppfattningar, värderingar och beteenden hos allmänheten i sverige [safety and risks in everyday: a study of perceptions, values, and behaviours among the public in sweden]. karlstad: räddningsverket. 18. eriksson, m. (2003) socialt k apital. teori, b egrepp och mätning en ku nskapsöversikt [social capital. theory, published by atlantis press copyright: the authors 30 j. sparf & s. öhman terms, and measurement – a knowledge r eview]. cerum, working paper, 60. 19. finucane, m. l. , alhakami, a., slovic, p, ands.m. johnson, 2000. the affect heuristic in judgments of risks and benefits. journal of behavioral decision making, 13(1): 1-17. 20. flap, h. andb. völker, 2004 .creation and r eturns of social capital: a n ew research program. london: routledge. 21. french gilson, s. 2004. disability, identity, and cultural diversity. review of disability studies, 1(1): 16-23. 22. giddens, a. 19 90.. the consequences of modernity . cambridge: polity in association with blackwell. 23. giddens, a. 1991. modernity and self-identity. cambridge: polity press. 24. giddens, a. 1 992.the transformation of i ntimacy: sexuality, love and eroticism in modern societies. cambridge: polity press. 25. granovetter,m. 1973. the strength of weak ties. the american journal of sociology78(6): 1360–1380. 26. grönvik, l. (20 07) definitions of disability in social sciences : methodological perspectives. uppsala: actauniversitatisupsaliensis. phd diss. 27. gustavsson, a. 2005.resistance, reflection and change : nordic disability research, social research on disability. lund: studentlitteratur. 28. harper, r. 2001.social capital. a review of the literature. newport: office for national statistics. 29. heath, r. l.and d. o 'hair, 2009.handbook of risk and crisis communication. new york: routledge. 30. heyman, b. and s. huckle, 1993. not worth the risk? attitudes of adults with learning difficulties, and their informal and formal carers to the h azards of everyday life.social science & medicine, 37(12): 1557-1564. 31. hugaasmolden, t. and tøssebr ojan, 2010. measuring disability in survey research: comparing current measurements within one data set. alter european journal of disability research, 4 (2010), pp. 174-189. 32. isham, j., kelley , t. ands. ramaswamy, 2002. social capital and well-being in developing countries: an introduction. in social capital and economic development: well-being in developing countries. ed s. ramaswamy,cheltenham: edward elgar. 33. johnston, s.j. 2 002. risk a ssessment in offenders with intellectual disability: the evidence base. journal of intellectual disability research, 46 (supplement s1): 47-56. 34. kvalsund, r. andi. velsvik bele, 2010 .adaptive situations and social marginalization in early adult life: students with special educational needs. scandinavian journal of disability research, 12(1): 59 76. 35. kääriäinen, j. andh. lehtonen, 2006.the variety of social capital in welfare state regimes – a comparative study of 21 countries. european societies 8(1): 27-57. 36. lin, n.1999. building a netw ork theory of social capital. connections, 22(1): 28-51. 37. lin, n. 2002.social capital : a theory of social structure and action.new york: cambridge university press. 38. luhmann, n. 1979.trust and power: two works. chichester: wiley. 39. lupton, d. (ed .) 1999.risk and soc iocultural theory. new directions and perspectives . cambridge: cambridge university press. 40. macgillivray, a. andp. walk er, 2000. local social capital: making it work on the ground. insocial capital: critical perspectives, ed t. schulleroxford: oxford university press. 41. meltzer, h., brugha, t., dennis, m. s., hassiotis, a., jenkins, r., mcmanus, s., r ai, d., and bebbington, p. 2012. the influence of d isability on suicidal behaviour.alter european journal of d isability research, 6 (2012) pp. 1-12. 42. mouridsen, s. e. andk-m. hauschild, 2009 . disability pensions in individuals d iagnosed with a dev elopmental language disorder as children . scandinavian journal of disability research, 11(4): 275 285. 43. national board of health and welfare. 2011.funktionshindradepersoner insatserenligt lss år 2010 [persons with certain functional impairments measures specified by lss 2010]. stockholm: nationalboard of health and welfare. 44. olofsson, a.,öhman, s. ands. rashid, 2005. views of risk in late modernity : do peop le's view of risk define our time? conference paper, international sociological association: environment, knowledge and democracy , july 6-7, 2005, at facu lty of sciences of lumin y, marseille. 45. olofsson, a. 2007.the p reparedness of local authorities for crisis communication with people who have foreign backgrounds: the case of sweden. international journal of mass emergencies and disasters, 25(2): 145-160. 46. olofsson, a. and s. rash id,2011. the white (m ale) effect and risk perception: can equality make a difference?risk analysis, 31(6): 1016-1032. published by atlantis press copyright: the authors 31 on risk and disability 47. olsvik, v. m. 2006. vulnerable, exposed and invisible: a study of violence and abuse against women with physical disabilities. scandinavian journal of disability research, 8(2): 85 98. 48. oorschot, wim van, wil ar ts, and john gelissen. 2006. "social capital in europe: measurement and social and regional distribution of a multifaceted phenomenon." acta sociologica 49:149-167. 49. parsons, t. 1978.research with human subjects and the professional complex. new york: free press. 50. peters, s. 200 0.is there a disability culture? a syncretisation of three possible world views.disability & society,15(4): 583-601. 51. pichler, f. andc . wallace,2007. patterns of formal and informal social c apital in europe. european sociological review, 23(4): 426-435. 52. pidgeon, n. f., kasperson, r. e. andp. slovic, 2003.the social amplification o f risk. cambridge: cambridge university press. 53. portes, a. 19 98. social capital: its origins and applications in modern socio logy. annual review o f sociology, 24: 1-24. 54. putnam, r. d . 1995. bowling alone: america' s declining social capital. journal of democracy , 6(1): 65 78. 55. putnam, r. d. 2 002.democracies in flux : the evolution of social capital in contemporary society. oxford: oxford university press. 56. renn, o, burns , w. j., kasp erson, r. e. andp. slovic, 1992. the social amplif ication of risk: theoretical foundations and empirical allocations. journal of social issues, 48(4): 137-160. 57. ruston, d. andl. akinrodoye, 2002.social capital question bank. questions from social capi tal surveys included in the socia l capital survey matrix 2002 . newport: office for national statistics. 58. seale, j. and m. nind , 2010.understanding and promoting access for peopl e with learning difficulties: seeing the opportunities and challenges of risk.london: routledge. 59. sjöberg, l. 20 00. factors in risk percep tion. risk analysis, 20(1): 1-12. 60. skrabski, a, kopp, m. and i . kawachi,2003. social capital in a changing society: cross sectional associations with middle aged fem ale and male mortality rates. journal of ep idemiology and community health, 57(2): 114-119. 61. slovic, p. 20 00.the perception of risk. london: earthscan. 62. slovic, p. 2010.the feeling of risk : new perspectives on risk perception.london:earthscan. 63. slovic, p. fischhoff, b. and s. lichtenstein, 2000. cognitive processes and soc ial risk taking. in the perception of risk, ed. p. slovic, london: earthscan. 64. spellerberger, a. 2001.framework for the measurement of social capital in new zealand. wellington: statistics new zealand. 65. statistics sweden 2006 and 2009. undersökningarna av levnadsförhållanden (ulf) [surveys about living conditions]. http://www.scb.se/pages/subjectarea____12184.aspx. 66. staveren, i. van 2003. beyond social capital in poverty research. journal of economic issues, 37(2): 415-424. 67. stone, w. and j. hughes, 2002.social capital: empirical meaning and measurement validity. research paper no. 27. melbourne: australian institute of family studies. 68. szebehely, m., fritzell, j . ando. lundberg, 2001.funktionshinderochvälfärd.sou 2001:56 [disability and welfare]. stockholm: fritzes. 69. tversky, a. andd. kahneman,1974. judgment under uncertainty: heuristics and biases. science,185(4157): 1124-1131. 70. tøssebro, j. anda. kittelsaa, 20 04.exploring the living conditions of disabled people, social research on disability. lund: studentlitteratur. 71. uphoff, n. andc.m. wijayaratna, 2000. demonstrated benefits from social c apital: the productivity of farmer organizations in gal oya, sri lanka. world development, 28: 1875-1890. 72. wall, e.and a. olofsson, 2008. young p eople making sense of r isk: how meanings of risk are materialised within the social context of every-day life.young nordic journal of youth research.16(4): 431-448. 73. warg, l-e. andm. wester-herber, 2001.riskkommunikation. att implementera seveso iidirektivet i sverige [risk communication. implementing the seveso ii directive in sweden], karlstad: räddningsverket. 74. wildavsky, a. andk. dake, 1990. theories of r isk perception: who fears what and why. dædalus, 119(4): 41-60. 75. woolcock, m. 1998. social capital and economic development: toward a th eoretical synthesis and policy framework. theory and society, 27(2): 151-208. published by atlantis press copyright: the authors 32 j. sparf& s. öhman 76. zissi, a., ronto s, c., pap ageorgiou, d., pierr akou, c. and s. chtouris , 2007. greek employers’ attitudes to employing people with disabilities: effects of the type of disability. scandinavian journal of disability research, 9(1): 14 25. 77. sparf, j. (2013a ccepted) disability and vulnerability : interpretations of risk in everyday life.journal of contingencies and crisis management. 78. sparf, j. (forthcoming[b] , submitted) a parasite to care about : integrating a freshwa ter contamination with the everyday work of a caring organization published by atlantis press copyright: the authors 33 1. introduction 2. risk perception 3. social capita 4. risk communication as an accessibility issue 5. method 6. results 6.1. perception of risks for the individual perception of societal risks 6.3. risk communication 7. discussion references