Varenna workshop report. Operational earthquake forecasting and decision making ANNALS OF GEOPHYSICS, 58, 4, 2015, RW0434; doi:10.4401/ag-6756 RW0434 Operational earthquake forecasting and decision making Warner Marzocchi1,*, Thomas H. Jordan2, Gordon Woo3, conveners 1 Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy 2 Southern California Earthquake Center, University of Southern California, Los Angeles, USA 3 Risk Management Solutions, London, UK A workshop on Operational earthquake forecasting and decision making was convened in Varenna, Italy, on June 8-11, 2014, under the sponsorship of the EU FP 7 REAKT (Strategies and tools for Real-time EArthquake risK reducTion) project, the Seismic Hazard Center at the Istituto Nazionale di Geofisica e Vulcanologia (INGV), and the Southern California Earthquake Center (SCEC). The main goal was to survey the interdisciplinary issues of operational earthquake forecasting (OEF), including the problems that OEF raises for decision making and risk communication. The workshop was attended by 64 researchers from universities, research centers, and governmental institutions in 11 countries. Participants and the workshop agenda are listed in the appendix. The workshop comprised six topical sessions struc- tured around three main themes: the science of opera- tional earthquake forecasting, decision making in a low-probability environment, and communicating haz- ard and risk. Each topic was introduced by a moderator and surveyed by a few invited speakers, who were then empaneled for an open discussion. The presentations were followed by poster sessions. During a wrap-up ses- sion on the last day, the reporters for each topical session summarized the main points that they had gleaned from the talks and open discussions. This report attempts to distill this workshop record into a brief overview of the workshop themes and to describe the range of opin- ions expressed during the discussions. 1. Background The concept of operational earthquake forecast- ing was developed by the International Commission on Earthquake Forecasting for Civil Protection (ICEF; Jor- dan et al. [2011]) in a report requested by the Italian government after L’Aquila earthquake of April 6, 2009 (available at http://www.annalsofgeophysics.eu/index. php/annals/article/view/5350). According to the ICEF definition, OEF comprises procedures for gathering and disseminating authoritative information about the time dependence of seismic hazards to help communities prepare for potentially destructive earthquakes. The ICEF report has stimulated a broad discussion about the development of OEF and its practical utility in man- aging seismic risk. The search for “diagnostic precursors” has so far been unsuccessful; no signals have yet been discovered that can reliably predict the location, time, and magni- tude of an impending earthquake with high probability. But the ICEF concluded that earthquake clustering models can reliably forecast earthquakes at low proba- bilities over short time intervals. According to these models, the short-term probabilities of large earth- quakes may increase by up to 2-3 orders of magnitude during seismic sequences, though they rarely exceed about few percent per week. Decision making in such a “low-probability environment” is problematic and poorly studied. An even more formidable challenge is how to communicate the hazard information and risk implications to various stakeholders, including the pub- lic at large, in a comprehensible and effective way. 2. Science of operational earthquake forecasting Two sessions on OEF science highlighted the for- mulation and testing of probabilistic forecasting models based on seismic clustering [e.g., Ogata 1988, Reasen- berg and Jones 1989, Michael 2012], and a third sur- veyed how OEF systems are developing in Italy, New Zealand, and the United States [Marzocchi et al. 2014, Gerstenberger et al. 2014, Field et al. 2015]. A hot topic was the seismicity induced by deep fluid injection, which is rapidly changing the seismicity of the U.S. con- tinental interior [e.g. Ellsworth 2013] and has become a Article history Received March 3, 2015; accepted July 22, 2015. Subject classification: Operational earthquake forecasting, Decision-making, Seismic risk, Risk communication. Varenna workshop report political as well as scientific issue in several countries. Speakers pointed out the difficulty of accommodating the rapid rise of induced seismicity into long-term probabilistic seismic hazard assessments (PSHA) that are the basis for most national seismic hazard maps, and this motivated discussions of how OEF systems could be useful for projecting time-dependent earthquake probabilities in regions experiencing anthropogenic seismicity increases. Short-term clustering models, such as the short- term earthquake probability (STEP; Gerstenberger et al. [2005]) and epidemic type aftershock sequence (ETAS; Ogata [1988]) models, currently show the highest in- formation gains relative to long-term forecasts, making them the prime candidates for OEF applications. It was reported that various versions of these models are now being continuously evaluated against seismic data in several testing regions around the world by the Col- laboratory for the Study of Earthquake Predictability (CSEP). CSEP is an international infrastructure for eval- uating earthquake forecasting models through prospec- tive forecasting experiments that are blind, transparent, and reproducible. CSEP testing experiments, which currently involve over 400 variations of statistical fore- casting models, are running in California, New Zealand, Italy, Japan, Western Pacific, and at global scale; other testing regions under consideration include Iceland and part of China. CSEP has been enhancing its procedures to provide more robust statistical evaluation of the models and to assess a wider range of model types [e.g. Jordan 2006, Zechar et al. 2010]. Model developers emphasized the role of CSEP testing in validating models, or model combinations, for the practical purposes of OEF. CSEP experiments are also aimed at addressing important scientific ques- tions, for example testing the hypotheses that clustering is self-similar from small to large earthquakes. The null hypothesis is that earthquakes do not encode any in- formation about the magnitude of the future events; the alternative hypothesis is that the preparatory phase of a large shock is somehow different, implying more predictability. The limitations of CSEP were also dis- cussed, particularly regarding the infrequency of large earthquakes, which limits the power of the prospective tests for that class of events. It was generally agreed that, in order to get better statistics at higher magni- tudes, prospective testing in the existing natural labo- ratories should be augmented with experiments in new natural laboratories and the testing of worldwide fore- casts, as well as through well-structured retrospective testing. It was reported that both activities are under active development at the CSEP testing centers. The clustering models involve generic statistics that do not explain the distinctive features observed in specific seismic sequences. Some participants argued that such deficiencies rendered the models inadequate for operational purposes. Others disagreed, noting that, though the models are uncertain and cannot explain everything, their forecasting performance has been val- idated by prospective testing in the CSEP natural labo- ratories, and they should therefore be useful within their stated limitations and epistemic uncertainties. All participants agreed that the skill of existing short- term forecasting methods is very modest. It is rare for the weekly probability of large shocks to reach 1% (usually after another large shock). Researchers are exploring how to increase forecasting skill by incorporating other types information into the forecasting models, including constraints on stress changes and rupture nucleation processes, as well as non-seismic precursory activity. Potential improvements may also come from the explicit incorporation of fault-based rupture models, such as the short-term extensions of the Third Uniform California Earthquake Rupture Forecast (UCERF3), which is being developed by the Working Group on California Earthquake Probabilities [Field et al. 2015]. Knowledge of the fault geometries and slip rates bet- ter defines the distribution of the large ruptures that could be triggered during seismic sequences, and it al- lows the concept of elastic rebound and stress relax- ation to be incorporated into the clustering models. A major topic in the workshop discussions was the need for consistency among forecasting methods across a full range of forecasting time scales. New Zealand re- searchers reported that, in the Canterbury region, earth- quake forecasting is now being done using hybrid models that incorporate the short-term STEP model, the long- term PSHA model, and a medium-term model called EEPAS (Every Earthquake a Precursor According to Scale), developed in New Zealand [Rhoades and Ger- stenberger 2009]. There was much interest in the lessons learned from recent Canterbury sequence, which began with the 2010 Darfield earthquake and has since provided a data- rich environment for evaluating OEF practice and its impact on society. The New Zealanders provided in- teresting perspectives on the important question of how expert opinion should be used in combination with earth- quake clustering models. In recent practice, experts have defined a set of plausible scenarios that span the pos- sibilities for the evolution of the seismic sequence, at- taching to each scenario a probability that is consistent with the probability given by the earthquake clustering models [Gerstenberger et al. 2014]. Representatives from other advisory groups de- scribed different approaches. In Italy, for example, the MARZOCCHI ET AL. 2 3 Grand Risk Commission (GRC) relies on expert opinion about the peculiarities of each seismic sequence rather than short-term forecasting models, which results in qualitative rather than quantitative assessments of the sequence’s possible evolution. In United States, the Cal- ifornia Earthquake Prediction Evaluation Council and the National Earthquake Prediction Evaluation Council have not yet structured their procedures to release ho- mogeneous information consistently during seismic se- quences, although the U.S. Geological Survey has been routinely, automatically producing generic assessments of aftershock probabilities after every M≥3.5 in Califor- nia for almost 20 years. Presently, USGS is working to expand the approach nationwide and to more rigorously handle situations where a straight Omori-Gutenberg- Richter probability assessment may not be the best rep- resentation of the hazard, and considering approaches for continuous production of probabilities. There was a consensus among the participants that expert opinion should be included in OEF, but some disagreement on how to integrate it with quan- titative models. Most modelers in the audience favored the quantitative formulation of expert opinion; for ex- ample, through well-structured expert elicitations. Some stressed the importance of casting all OEF infor- mation in the form of probabilities, both aleatory and epistemic, to allow an adequate separation between sci- ence of hazard assessment and the more socially com- plex process required to formulate an appropriate spectrum of risk-mitigation options. 3. Decision making in a low-probability environment OEF provides probabilistic forecasting informa- tion, but the formulation of mitigation options requires that the earthquake hazard, as described in terms of faulting, shaking, and secondary effects such as ground failures, be translated into seismic risk, as measured in dollars, casualties, and social functionality. The engi- neering and policy approaches to this risk-analysis prob- lem were the topic of the fourth session, and related decision-making issues were addressed by several speak- ers in the fifth and sixths sessions. It was noted that the term “low-probability” associ- ated with the hazard is an incomplete description of OEF problem, because the consequences for some stake- holders (e.g., critical facility operators) may be much higher than for ordinary citizens. The higher risk levels perceived by such stakeholders will more frequently war- rant mitigation actions. Usually, but not always, a cost- benefit analysis will dictate actions that are relatively low in cost, commensurate with the low probabilities. Earth- quake engineers emphasized the quantification of risk through well-defined loss metrics normalized to quies- cent reference conditions, such as the expected damage rates to structural models or expected fatality rates, and they illustrated plausible metrics with examples from structural engineering [Iervolino et al. 2015]. There was a substantial agreement among the at- tendees to separate the formulation of scientific infor- mation about hazard from the risk assessments that inform decision making. According to the ICEF, the OEF role is to deliver robust, authoritative hazard in- formation, usually cast in terms of probability. This in- formation has to be translated into risk, i.e. into different kinds of expected losses. Risk assessment is the natural framework in which to formulate mitigation options and to choose the proper actions according to a ration- ale cost-benefit analysis. Although widely accepted in principle, the precise nature of this hazard/risk separa- tion, as well as the means to achieve it, was debated by the participants, mostly around the question of whether scientists should act as decision makers in evaluating and prioritizing the mitigation options. One clear example discussed at the meeting is the policy of releasing OEF information only during after- shock sequences [e.g. Wang and Rogers 2014, Jordan et al. 2014]. Most participants agreed that OEF probabilities during vigorous aftershock sequences should be made publically available. After large earthquakes, the proba- bilities are higher and, according to advocates of the “af- tershocks-only” policy, more easily comprehended by the public. Others argued that such a policy violates the haz- ard-risk separation principle, because it puts OEF in an inappropriate decision-making role; namely, of judging at which probability levels the users of OEF should be in- formed. Moreover, it violates the transparency principle, also articulated by the ICEF, which states that authorita- tive scientific information about future earthquake activ- ity should not be withheld from the public. Those issues aside, it was widely agreed that the utility of forecasting is strongly correlated with the am- plitude of the probabilities. Higher probabilities can be more easily used by the decision makers to establish a set of mandatory mitigation actions based on a quanti- tative rational framework based on cost-benefit analysis. The practical use of forecasts with high-hazard proba- bility rates of less than, say, 5% per week to mitigate the risk is much less clear, and the attending issues have not been yet been sufficiently explored. Some attendees argued that current short-term forecasting methods are effectively useless for practical purposes, owing to the high false alarm rates implied by the low-probability rates - the “crying-wolf ” prob- lem [Wang and Rogers 2014]. Others countered that even small probabilities can be helpful in nudging peo- ple towards beneficial self-protection actions that are VARENNA WORKSHOP REPORT consistent with their own aversion to risk [Thaler and Sunstein 2009, Jordan et al. 2014]. Some might choose to minimize their time in seismically unsafe buildings, for example. Once citizens are correctly informed about the earthquake threat and possible actions to mitigate their personal risks, they can act as rational decision makers for their own families and communities. It was noted that nudging citizens towards rational actions through the release of authoritative information is commonly employed to defend against many other low-probabil- ity hazards, such as terrorism threats and the spread of communicable diseases. The recent Canterbury seismic sequence in New Zealand has provided an excellent testbed for the use of OEF in decision making on an urban scale. It was re- ported how quantitative OEF information was deliv- ered, interpreted, and applied to risk mitigation during this sequence, and how the OEF system set up by New Zealand seismologists continues to guide the recon- struction of Christchurch, which was badly damaged by the Darfield aftershocks of 2011 [Wein and Becker 2013, Gerstenberger et al. 2014]. OEF systems will have to service a wide variety of information requirements, most obviously those of the general public and responsible governmental agencies, but also those of other stakeholders, such as those who manage critical infrastructures and insurance/re-insur- ance companies. The various needs and levels of risk aversions among the many stakeholder groups imply that OEF systems will have to provide scientific infor- mation in different formats across a range of forecast- ing intervals. One notable problem is maintaining the consistency of short-term OEF forecasts with the long- term hazards quantified in PSHA. To avoid confusion during seismic crises, it was recommended that the probability thresholds for sig- nificant risk-reduction actions be described in protocols prepared in advance of a crisis through negotiations among scientists, risk analysts, political decision-makers, experts in communication, and interested stakeholders. These protocols should describe the way in which the authoritative scientific information is to be delivered and how high-priority risk mitigation actions are to be implemented. 4. Communicating hazard and risk The discussions of how OEF information should be communicated to decision makers, including the public at large, dominated the fifth and sixth topical sessions. The social scientists attending the workshop outlined some general findings from their research on risk com- munication. Humans have the ability to perceive and bal- ance a large number of risks regularly encountered in the natural environment, including high risks at low levels of probability. However, when new information about low- probability, high-consequence events is received, people tend to dichotomize their response: they are either con- cerned and motivated to act, or they are unconcerned and don’t. Studies of this threshold-type response has led to some general guidelines for risk messaging [e.g., Wood et al. 2012, and references therein]: (i) Communication about specific risks should be layered, with basic information broadcast in high-level messages and more detailed information, including hazard and risk probabilities, made easily available to those who seek it. (ii) High-level messages should contain authorita- tive information about the hazard, identify the sources of the information, and be explicit in recommending specific actions to reduce the risk. (iii) Messaging is most effective when consistent information comes from multiple sources. (iv) The public should be educated about hazards and risks through a continual dialog with scientists and decision makers and within their own communities. Considerable debate was centered on how to com- municate increases in earthquake hazard and risk when the absolute probability of high hazard remains low. Some geoscientists expressed particular concern over the widespread public illiteracy about probability, which may limit the ability of non-scientists, including deci- sion makers, to comprehend the meaning of the aleatory variability and epistemic uncertainty intrinsic to earth- quake forecasting. Techniques for improving public pre- paredness through OEF have not yet received sufficient attention, although relevant research was conducted in California during the Parkfield prediction experiment and following the 1989 Loma Prieta earthquake [Mileti and DeRouen 1995]. Studies were reviewed that show the ability of even primitive societies to comprehend probability-based statements of risk, as long as the statements are properly formulated and clearly communicated [Fontanari et al. 2014]. It was emphasized that communication can build trust in authoritative information over time, and that honesty about uncertainty can enhance credibility. Experts in risk communication argued that, when faced with confusing and possibly hazardous situations, people respond positively to authoritative statements from official sources about the actions they should take, even when the future is highly uncertain; properly deliv- ered, authoritative information about impending threats has rarely, if ever, caused panic among an informed populace [e.g. Clarke 2002]. These considerations underline the need for trans- parency and continuity in broadcasting authoritative sci- MARZOCCHI ET AL. 4 5 entific information about the time-dependence of earth- quake hazards. Several participants noted that the timely communication of authoritative forecasts can benefit the public by filling information vacuums that set the stage for amateur earthquake predictions and misinformation. Seismic crises can also be teachable moments, when peo- ple abandon their usual apathy about earthquake pre- paredness. Timely OEF communication can thus be very effective in teaching people how to reduce seismic risk both in the short and in the long-term. References Clarke, L. (2002). Panic: Myth or reality?, Contexts, 1, 21-26. Ellsworth, W.L. (2013). Injection-Induced Earthquakes, Science, 341, 1225942. Field, E.H., et al. (2015). Long-Term Time-Dependent Probabilities for the Third Uniform California Earth- quake Rupture Forecast (UCERF3), B. Seismol. Soc. Am., 105, 511-543. Fontanari, L., M. Gonzalez, G. Vallortigara and V. Girotto (2014). Probabilistic cognition in two in- digenous Mayan groups, Proc. Natl. Acad. Sci., 111, 17075-17080. Gerstenberger, M., S. Wiemer, L.M. Jones and P.A. Reasenberg (2005). Real-time forecasts of tomorrow’s earthquakes in California, Nature, 435, 328-331. Gerstenberger, M., D. Rhoades, G. McVerry, A. Christo- phersen, S. Bannister, B. Fry and S. Potter (2014). Re- cent Experiences in Aftershock Hazard Modelling in New Zealand, AGU fall meeting, San Francisco. Iervolino, I., E. Chioccarelli, M. Giorgio, W. Marzocchi, G. Zuccaro, M. Dolce, G. Manfredi (2015). Opera- tional (short-term) earthquake loss forecasting in Italy, B. Seismol. Soc. Am., 105 (4), 2286-2298. Jordan. T.H. (2006). Earthquake predictability, brick by brick, Seismol. Res. Lett., 77, 3-6. Jordan, T.H., Y.-T. Chen, P. Gasparini, R. Madariaga, I. Main, W. Marzocchi, G. Papadopoulos, G. Sobolev, K. Yamaoka and J. Zschau (2011). Operational earth- quake forecasting: state of knowledge and guide- lines for utilization, Annals of Geophysics, 54 (4), 315-391. Jordan, T.H., W. Marzocchi, A. Michaela and M. Ger- stenberger (2014). Operational earthquake forecast- ing can enhance earthquake preparedness, Seismol. Res. Lett., 85, 955-959. Marzocchi, W., A.M. Lombardi and E. Casarotti (2014). The establishment of an operational earthquake forecasting system in Italy, Seismol. Res. Lett., 85, 961-969. Michael, A.J. (2012). Fundamental questions of earth- quake statistics, source behavior, and the estimation of earthquake probabilities from possible fore- shocks, B. Seismol. Soc. Am., 102, 2547-2562. Mileti, D.S., and J. DeRouen (1995). Societal response to revised earthquake probabilities in the San Fran- cisco Bay area, Int. J. Mass Emer. Dis., 13, 119-145. Ogata, Y. (1988). Statistical models for earthquake oc- currences and residual analysis for point processes, J. Am. Stat. Assoc., 83, 9-27. Reasenberg, P.A., and L. M. Jones (1989). Earthquake hazard after a mainshock in California, Science, 243, 1173-1176. Rhoades, D.A., and M.C. Gerstenberger (2009). Mix- ture models for improved short-term earthquake forecasting, B. Seismol. Soc. Am., 99, 636-646. Thaler, R.H., and C.R. Sunstein (2009) Nudge: improv- ing decisions about health, wealth and happiness, Penguin Press. Wang, K., and G.C. Rogers (2014). Earthquake pre- paredness should not fluctuate on a daily or weekly basis, Seismol. Res. Lett., 85, 569-571. Wein, A., and J. Becker (2013). Communicating after- shock risk: Roles for reassuring the public, Risk Frontiers, 13 (3), 3-4. Wood, M.W., D.S. Mileti, M. Kano, M.M. Kelley, R. Regan and L.B. Bourque (2012). Communicating Actionable Risk for Terrorism and Other Hazards, Risk Anal., 32, 601-615. Zechar, J.D., D. Schorlemmer, M. Liukis, J. Yu, F. Eu- chner, P. Maechling and T.H. Jordan (2010). The col- laboratory for the study of earthquake predictability perspective on computational earthquake science, Concurrency Comput Pract. Ex., 22, 1836-1847. Corresponding author: Warner Marzocchi, Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy; email: warner.marzocchi@ingv.it. © 2015 by the Istituto Nazionale di Geofisica e Vulcanologia. All rights reserved. VARENNA WORKSHOP REPORT Appendix A: Workshop agenda Session 1: The problem of earthquake forecasting - The problem of earthquake forecasting (T. Jordan) - Earthquake forecasting when the background rate is non-sta- tionary: The case of the U. S. midcontinent (B. Ellsworth) - Forecasting induced seismicity: State of the art and future di- rections (S. Wiemer) - Panel discussion (moderator: I. Main; reporter: M. Page) Session 2: Testing short-term earthquake forecasting models - The collaboratory for the study of earthquake predictability (D. Schorlemmer) - The two-way street between next-day seismicity forecasts and operational earthquake forecasting ( J. Zechar) - CSEP developments in support of operational earthquake forecasting: a future perspective (D. Rhoades) - Panel discussion (moderator: N. Hirata; reporter: M. Werner) Session 3: Operational earthquake forecasting - Operational earthquake forecasting in Italy (W. Marzocchi) - Operational earthquake forecasting in the United States (N. Field) - Recent experiences in OEF in New Zealand: the good and the bad (M. Gerstenberger) - Panel discussion (moderator: R. Stein; reporter: M. Blanpied) Session 4: From short-term hazard to risk - Operational earthquake loss forecasting: framework and proof of concept (I. Iervolino) - Evaluating risk and providing advise for low probability-high impact events in the presence of large uncertainties: the expe- rience of the Italian High Risk Commission, 2011-2013 (D. Giardini) - Current practice in the United States, and plans for improved dynamic earthquake likelihood forecasting (M. Blanpied) - Panel discussion (moderator: M. Dolce; reporter: M. Gerstenberger) Session 5: Decision-making process and risk communica- tion in a low-probability forecasting environment (I) - Participatory decision making under uncertainty (G. Woo) - The structure and content of effective risk messages com- municated to the public (D. Mileti) - Managing seismic sequences in Italy: civil protection current practice and questions for the future (M. Dolce) - Panel discussion (moderator: W. Marzocchi; reporter: T. Sellnow) Session 6: Decision-making process and risk communica- tion in a low-probability forecasting environment (II) - Case histories of communicating uncertain earthquake haz- ard and loss estimates (D. Wald) - Risk communication strategies when you understand that your audience is about as smart as you are (T. Patt) - Strategies for communicating science-based messages in times of uncertainty (T. Sellnow) - Panel discussion (moderator: D. Mileti; Reporter: G. Woo) Session 7: Summary of the meeting - Reports on the topic sessions (M. Page, M. Werner, M. Blan- pied, M. Gerstenberger, T. Sellnow, G. Woo) - Wrap-up discussion Posters sessions: - A short-term earthquake forecasting experiment in Japan (N. Hirata) - Operational short-term earthquake forecasting in Japan (K. Doi) - CORSSA: community online resource for statistical seismic- ity analysis ( J. D. Zechar) - Evaluating earthquake predictions by using the gambling score ( J. Zhuang, J. D. Zechar, C. Jiang) - Effect of data quality on a hybrid Coulomb/STEP model for earthquake forecasting (S. Steacy) - Practice and enlightenment of earthquake prediction in China (Xiadong Zhang) - An operational earthquake forecasting system in metropolitan area around Beijing, China (Yongxian Zhang) - Short-term forecasting and preparation of earthquakes (D. Albarello) - Preparing for an operational earthquake forecast experiment in Iceland (F. Panzera) - Improving earthquake and aftershock risk communication: lessons from the Canterbury earthquakes, New Zealand (S. Potter) - The effect of including aftershocks in probabilistic seismic hazard assessment modelling: A case study for Wellington (A. Christophersen) MARZOCCHI ET AL. 6 7 Appendix B: Workshop participants ALBARELLO, Dario AMATO, Alessandro BLANPIED, Mike CASAROTTI, Emanuele CATTANEO, Marco CHIARABBA, Claudio CHIARALUCE, Lauro CHRISTOPHERSEN, Annemarie COMUNELLO, Francesca CONSOLE, Rodolfo CRISCUOLO, Annamaria D’AMICO, Vera DEICHMANN Nicholas DI BUCCI, Daniela DOI, Keiji DOLCE, Mauro ELLSWORTH, Bill FAENZA, Licia FALCONE, Giuseppe FIELD, Ned GARCIA Alexander GASPARINI, Paolo GERSTENBERGER, Matt GIARDINI, Domenico HERRMANN, Marcus HIRATA, Naoshi IERVOLINO, Iunio JORDAN, Thomas LE GUENAN, Thomas LUSSIGNOLI, Orsola MAIN, Ian MALAFRONTE, Lucia MARZOCCHI, Warner MELETTI, Carlo MILETI, Dennis MONELLI, Damiano MURRU, Maura NAYLOR, Mark PAGE, Morgan PANZERA, Francesco PATT, Anthony POWER, Christopher RHOADES, David ROSELLI, Pamela ROSI, Mauro ROSSI FILANGIERI, Alfonso SABETTA, Fabio SCHORLEMMER, Danijel SEGOU, Margarita SEIF, Stefanie SELLNOW, Deanna SELLNOW, Timothy SHAPIRA, Avi STEACY, Sandy STEIN, Ross TARONI, Matteo VOGFJORD, Kristin WALD, David WALTER, Andre WENZEL, Friedemann WERNER, Maximilian WIEMER, Stefan WOO, Gordon ZECHAR, Jeremy ZHANG, Xiaodong ZHANG, Yongxian ZHUANG, Jiancang ZUCCARO, Giulio VARENNA WORKSHOP REPORT << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles false /AutoRotatePages /None /Binding /Left /CalGrayProfile (Dot Gain 20%) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.3 /CompressObjects /Tags /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.1000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize false /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness true /PreserveHalftoneInfo false /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile (None) /AlwaysEmbed [ true /AndaleMono /Apple-Chancery /Arial-Black /Arial-BoldItalicMT /Arial-BoldMT /Arial-ItalicMT /ArialMT /CapitalsRegular /Charcoal /Chicago /ComicSansMS /ComicSansMS-Bold /Courier /Courier-Bold /CourierNewPS-BoldItalicMT /CourierNewPS-BoldMT /CourierNewPS-ItalicMT /CourierNewPSMT /GadgetRegular /Geneva /Georgia /Georgia-Bold /Georgia-BoldItalic /Georgia-Italic /Helvetica /Helvetica-Bold /HelveticaInserat-Roman /HoeflerText-Black /HoeflerText-BlackItalic /HoeflerText-Italic /HoeflerText-Ornaments /HoeflerText-Regular /Impact /Monaco /NewYork /Palatino-Bold /Palatino-BoldItalic /Palatino-Italic /Palatino-Roman /SandRegular /Skia-Regular /Symbol /TechnoRegular /TextileRegular /Times-Bold /Times-BoldItalic /Times-Italic /Times-Roman /TimesNewRomanPS-BoldItalicMT /TimesNewRomanPS-BoldMT /TimesNewRomanPS-ItalicMT /TimesNewRomanPSMT /Trebuchet-BoldItalic /TrebuchetMS /TrebuchetMS-Bold /TrebuchetMS-Italic /Verdana /Verdana-Bold /Verdana-BoldItalic /Verdana-Italic /Webdings ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 150 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.10000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 150 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.10000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.08250 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown /CreateJDFFile false /SyntheticBoldness 1.000000 /Description << /ENU (Use these settings to create PDF documents with higher image resolution for high quality pre-press printing. The PDF documents can be opened with Acrobat and Reader 5.0 and later. These settings require font embedding.) /JPN /FRA /DEU /PTB /DAN /NLD /ESP /SUO /NOR /SVE /KOR /CHS /CHT /ITA >> >> setdistillerparams << /HWResolution [2400 2400] /PageSize [595.000 842.000] >> setpagedevice