Geological Survey of Denmark and Greenland Bulletin 33, 2015, 29-32 440 40 25 20 200 Precipitation In fil tra tio n 740 S ur fac e f l o w Sea Stream flow Atmosphere Evapotranspiration over sea Unsaturated zone 15 Abstraction 15 Evapotranspiration Groundwater Wind Groundwater flow 29 Th e rapidly increasing impacts of climate change are likely to require changes in relevant institutions (IPCC 2012). An ex- ample is the growing need for immediate information on the entire water cycle (Fig. 1), with quantitative assessments of critical hydrological variables and fl ow interactions between diff erent domains, e.g. atmosphere, plant-soil, surface water, groundwater and the sea, as they take place. Potential measures include early warning systems, risk communication between decision makers and local citizens, sustainable land management, including land use planning as well as ecosystem management and restoration (IPCC 2012). Early warning systems (Kundzewicz 2013) that pro- vide information and monitoring of past and present hy- drological conditions as well as forecasting of hydrological conditions (e.g. water content, fl ow and water levels) are fi rst steps in developing fl ooding indicators for operational use. Fig. 1. Water balance in mm/year for a typical Danish area. With climate change the freshwater cycle is no longer in a steady state. Early warning and monitoring keeps us continuously updated and gives us an overview. This is important for our ability to combat the impact of climate change and manage the water resources proac- tively. It is possible that a nationwide system can be linked to local early warning systems and can make use of national overviews based on national hydro- logical models. Flooding and drought events are complex phenomena with several key mechanisms including intense and long lasting precipitation or lack of precipitation in the case of drought and interaction with water uses. Water consumers use ground- water for drinking water, food pro- duction, households and livestock, energy production and recreational activities. At the same time, water authorities have to manage surface and groundwater to sustain ecological systems, and support vital ecological conditions for plants and animals in rivers and wetlands. It is obvious that it is vital for society to have water in the right amount in the right place at the right time. Th erefore, we must understand the water cycle, i.e. how overland drainage and surface runoff are generated, how water fl ows in the upper soil layers and in the deeper subsurface, how it is discharged in water courses and lakes and how freshwater interacts with the sea. One or two decades ago, water management was developed and operated under the assumption of static conditions, a par- adigm which now is dead according to Milly et al. (2008). It is stated by Milly et al. (2008, p. 573) that “Stationarity – the idea that natural systems fl uctuate within an unchanging en- velope of variability – is a foundational concept that permeates A hydrological early warning system for Denmark based on the national model Hans Jørgen Henriksen, Simon Stisen, Xin He and Marianne B. Wiese © 2015 GEUS. Geological Survey of Denmark and Greenland Bulletin 33, 29–32 . Open access: www.geus.dk/publications/bull 3030 training and practice in water-resource engineering . . . . Sta- tionarity is dead because substantial anthropogenic change of the Earth’s climate is altering the means and extremes of pre- cipitation, evapotranspiration, and rates of discharge of rivers”. In the new paradigm, in the non-stationary world, real-time modelling and a continuity of monitoring systems are critical for dealing with the increasing impact of extremes. Globally natural disaster costs have more than quadrupled since 1985 (Georgieva 2014). Early warning and monitoring systems that can transfer operational hydrological knowledge to communi- ty-based climate change adaptation planning and emergency management, are assumed to be fundamental for building societal resilience, both in the phases of pre-disaster, disaster response and post-disaster, and in general for extending moni- toring techniques with network-based public participation. Th e Geological Survey of Denmark and Greenland (GEUS) fi nanced a two year project (2012–2014), in order to determine the requirements for a hydrological system ca- pable of providing real-time and forecast simulations based on a national hydrological model (the DK-model; www. vandmodel.dk; Henriksen et al. 2003; Stisen et al. 2012; Højberg et al. 2013), and how the system can be linked with local early warning systems. Th is paper describes four scenar- ios discussed at a participatory workshop in October 2014 and the workshop’s outcome. Th e workshop provided GEUS with valuable insight and feedback relevant for the future de- velopment of a nation-wide, real-time modelling and water cycle monitoring system for Denmark, including the possible input to an early warning system and real-time forecasting to operate at local level. Structuring the needs – four scenarios for an early warning system design Th e four scenarios, used in the workshop, for the design of a hydrological, real-time forecasting system linked to local level, community-based, emergency management are shown in Fig. 2. Scenario 1: An updated national model can provide an estimate of the actual status of water resources in Denmark based on the calibrated national model (the DK-model) with updated climatic data, and available on-line with absolute val- ues, indices or anomalies (below, same as, or above normal). Scenario 2: A national model that can forecast the state of the water resources on a short timescale or as a seasonal prognosis, based on forward modelling simulations using weather forecast data. Th is could be used as an early warn- ing system and as a starting point for local, community-based emergency management. Scenario 3: Decision Support is a scenario where local communities use a decision support tool for integrated as- sessment and management (Kelly et al. 2013), for incorporat- ing local knowledge and experience, and where simulation results from the national model in forecasting mode are in- cluded. Th is scenario builds on a combination of monitored data and model output. Th e output from this model might be with physical variables, thematic maps, indices or anomalies. Scenario 4: A complex local model where a local detailed model (such as MIKE 11 or MIKE Urban) uses simulated results from the national model in forecasting mode as input or boundary conditions. Th is can be combined with moni- toring data in a data assimilation framework. Th e national model data in this scenario are physical variables delivered as point- or gridded data. A representative catchment area (Skjern Å; Fig. 3) was se- lected for the workshop to test and demonstrate a prototype of a web-based, hydrological warning system for professional users. Th e system provides hydrological simulations from groundwater levels, stream fl ow and water content in the root zone. Webpages can be tailor-made to meet the require- ments of end-users and continuously adapt to changing user demands. Numerical results from simulations on a national Fig. 2. The four real-time forecasting scenarios discussed in the text. Scenario 1: Updated DK- model, Scenario 2: DK-model forecast, Scenario 3: Local decision support and Scenario 4: Com- plex local model. 31 scale could also be provided. Hourly values of climatic and discharge data at Ahlergaarde Station 250082 were used for the prototype demonstration. Setting up the workshop A participatory workshop (Hare & Krykow 2005) was held in Copenhagen for the purpose of eliciting stakeholder ideas and opinions and get feedback from prospective users. Par- ticipants were planners and emergency managers from local authorities, water supply companies, ministries and consult- ing agencies. Th e workshop had invited speakers from the Holstebro, Aarhus and Fredensborg municipalities and lo- cal authorities that had already implemented local, real-time forecasting and early warning systems. Th e Danish Hydrau- lic Institute (DHI) and the Danish Meteorological Institute (DMI), GEUS’ two partners in the project, also presented their experiences. Both of these institutions have more than 20 years of experience developing and implementing model- based early warning systems in Denmark and abroad. Prior to the workshop, an invitation was sent to those em- ployed with climate change adaptation and emergency man- agement at local and national levels. A questionnaire was attached to the invitation which explained why GEUS had initiated the project, introduced the four scenarios and brief- ly described the hydrological components that the national model can simulate. Th e prototype of the web interface was described. Th e goal of the workshop was an in-depth discus- sion of user requirements as a supplement to the web ques- tionnaire which had 27 respondents. A total of 34 partici- pants signed up for the workshop; eight came from GEUS, one from DMI, one from DHI, nine from local authorities, four from national ministries, ten from consulting compa- nies and one from a large water-supply company. Th e workshop was a one day event. Th e fi rst block con- tained three presentations by GEUS participants: introduc- tion to the real-time project, technical challenges in real-time modelling and web presentation of real-time data for the river Skjern Å catchment area. Th is was followed by two in- vited presentations by DMI (better prediction of heavy rain) and the DHI (early warning systems in relation to hydrol- ogy and the freshwater cycle). Th ree invited speakers from local authorities talked about fl ooding from the river Storå in Holstebro, the Usserød Å project in Fredensborg and an early warning model with emphasis on fl ooding and ground- water monitoring in the urban area of Aarhus. Finally, there was a group session with three groups each addressing the same four questions. (1) What are the requirements for real- time forecasting? Do you miss an overview of the hydrologi- cal state of an area in your daily work or in situations with fl ooding or drought? (2) What information would be ben- efi cial during such a situation, absolute or relative values; and which components of the hydrological cycle are the most important? (3) What is the time perspective in such an ex- treme event: days, hours or other? What would be the most appropriate frequency for updating the forecasts? (4) Would a national overview make any diff erence in forecasting hy- drological events, and how can or should the prototype be developed further? How can the present DK-model be part of such a system? Outcome of the workshop Th e results of the discussions were presented by groups. Group 1 recommended linking national and local systems and presented a variety of requirements from diff erent local authorities posing potential challenges to the entire concept. Th ey noted that problems with fl ooding, drought and emer- gency management are highly site specifi c. Data assimilation is an important component in the early warning systems and there are many diff erent sources of observational data, e.g. from local authorities and the Danish Road Directorate. A merged dataset is preferred. Group 2 noted that real-time data for shallow groundwa- ter levels are lacking but it may be possible to use observations from geotechnical boreholes. An investigation of geotechni- cal boreholes is necessary in order to establish a new opera- tional groundwater level monitoring network. More empha- sis should be given to a real-time early warning system instead of the present forecasting system with a short time horizon of only days. Th e quality of early warnings should be quantifi ed in order to demonstrate how reliable the model can describe 5 km9°E 56°10´N Discharge station Hydraulic head observation Stream Ahlergaarde catchment Station 250082 Fig. 3. Pilot case study for river Skjern Å, Ahlergaarde catchment area. Green: low elevation areas. Dark yellow: high elevation areas. 3232 extremes. It is important to simulate water levels including storm surges. Data and forecasts should be available online. Group 3 noted that in many cases local authorities have their own local data which could provide input to the na- tional model. Local authorities may not have a groundwater model so co-operation between GEUS and local authorities are encouraged. Local authorities experience an increased de- mand for warning and action capabilities. At the same time, local authorities are reluctant to issue warnings because they do not wish to be overcautious or risk subsequent claims. Ab- solute values of physical variables are requested and data of levels of surface water and shallow groundwater are the most urgent. It was also suggested that continuous monitoring of water levels and early warning of changes in these levels are the most interesting for short-term forecasts, especially when local authorities do not have early warning systems, which is consistent with what was proposed by Group 2. From a temporal perspective, precipitation events are highly diverse and it may not always be the short-term cloudbursts that are the most infl uential. Long periods of rainfall (e.g. from a few days in succession to prolonged periods of rain) can signifi cantly increase groundwater levels in western and cen- tral Jylland and snowmelting events can also cause extreme fl ooding. For the moment the national model has a limited applicability with its focus on water fl ow simulations. It is necessary to simulate water levels with local models. Discussion and conclusion A participatory workshop discussion real-time forecasting was held to get feedback and get into dialogue with water planners and emergency managers from local authorities, water companies and national authorities. At the workshop, a prototype website illustrating four scenarios of national and local early warning systems was presented for the Skjern Å, Ahlergaarde catchment area with selected events. Th e workshop recommended that GEUS should focus on real- time modelling with the DK-model (Scenario 1). Th e fi rst step is to update the national coverage of climatic data in- put from DMI; real-time discharge fl ow and groundwater level monitoring are required; and the necessity of data as- similation and other types of uncertainty analyses have to be further evaluated. If a forecast model is included (Scenario 2), complex data assimilation is required, however, this can compromise the water and mass balance of the model. Th e water balance and simulation of the whole water cycle should be addressed by the early warning systems (most participants found that soil moisture, discharge in rivers and groundwa- ter levels should be represented in such a system). Eventually, an improved system for collecting observations of precipita- tion (or use of high-resolution radar measurements adjusted with rain-gauge data from a coarse network) is needed be- cause the current network of rain-gauge stations has a poor national coverage. Th e early warning system should be able to deliver results for discharge stations and boundary condi- tions for subsequent use in local models (Scenarios 3 and 4). Acknowledgements Th e study was conducted as part of the GEUS-funded project: ‘Realtids- varsling’. Th e paper is a NOR DR ESS (nordress.hi.is) contribution. We are grateful for the input and feedback of the participants at the workshop. References Georgieva, K. 2014: Post-Haiyan – a way forward. Speech/14/441 by EU Commissioner for international cooperation, humanitarian aid and cri- sis response. ASEM Conference on Disaster Risk Reduction and Man- agement. Manila, 5 June 2014. Hare, M. & Krykow, J. 2005: Participatory processes for the design of wa- ter storage areas – theme group III inception report of the TRUST pro- ject. Seecon report 1/2005, 68 pp. Osnabrück: Hoogheemraadschap van Schielanden en der Krimpenerwaard. Henriksen, H.J., Troldborg, L., Nyegaard, P., Sonnenborg, T.O., Refs- gaard, J.C. & Madsen, B. 2003: Methodolog y for construction, cali- bration and validation of a national hydrological model for Denmark. Journal of Hydrolog y 280, 52–71. Højberg, A.L., Troldborg, L., Stisen, S., Christensen, B.S.B. & Henriksen, H.J. 2012: Stakeholder driven update and improvement of a national water resources model. Environmental Modelling and Soft ware 40, 202–213. IPCC 2012: Managing the risks of extreme events and disasters to advance climate change adaptation, a special Report of working groups I and II of the Intergovernmental Panel on Climate Change, 582 pp. Cam- bridge: Cambridge University Press. Kelly, R.A. et al. 2013: Selecting among fi ve common modelling approach- es for integrated environmental assessment and management. Environ- mental Modelling & Soft ware 47, 159–181. Kundzewicz, Z.W. 2013: Floods: lessons about early warning systems. In: Gee, D. et al. (eds): Late lessons from early warnings: science, precau- tion, innovation, 347–368. EEA Report no. 1/2013. Copenhagen: Eu- ropean Environment Agency. Milly, P.C.D., Betancourt, J., Falkenmark, M., Hirsch, R.M., Kundze- wicz, Z.W., Lettenmaier, D.P. & Stouff er, R.J. 2008: Stationarity is dead: Whither water management? Science 319, 573–574. Stisen, S., Højbjerg, A.L., Troldborg, L., Refsgaard, J.C., Christensen, B.S.B., Olsen, M. & Henriksen, H.J. 2012: On the importance of ap- propriate precipitation gauge catch correction for hydrological model- ling at mid to high latitudes. Hydrolog y and Earth System Sciences 16, 4157–4176. Authors’ address Geological Survey of Denmark and Greenland, Øster Voldgade 10, DK-1350 Copenhagen K, Denmark. E-mail: hjh@geus.dk