Young Scientists Workshop “Data assimilation, dealing with
Transcription
Young Scientists Workshop “Data assimilation, dealing with
Young Scientists Workshop on Data Assimilation Young Scientists Workshop “Data assimilation, dealing with uncertainties, and the prediction capabilities of models in water research” Seminaris Hotel Lüneburg, Germany October 8th (evening) – 11th (morning) 2007 http://www.dfg-wasserkommission.de Back to contents 1 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 2 of 88 Young Scientists Workshop on Data Assimilation KoWa Newsletter August 2007 Contents Aims of the workshop.........................................................................................................................4 Programme..........................................................................................................................................5 Organizing Committee.......................................................................................................................7 Workshop Venues...............................................................................................................................8 Keynote Speakers Prof. Dr.techn. Günter Blöschl.............................................................................................................9 Dr. Matthias Drusch...........................................................................................................................10 Prof. Dr. Peter Reichert......................................................................................................................11 Prof. dr.ir. Sjoerd E.A.T.M. van der Zee............................................................................................12 Participants Dr. Felix Ament..................................................................................................................................13 Dr. Hans-Stefan Bauer........................................................................................................................15 Dr. rer. nat. Christof Beyer.................................................................................................................17 Dr. Helge Bormann............................................................................................................................19 Dr.-Ing. Jörg Dietrich.........................................................................................................................21 Dipl. Envir. Sciences Martin Frey......................................................................................................23 Dipl.-Geoökol. Bastian Graupner.......................................................................................................25 Dipl.-Hydrol. Jens Grundmann..........................................................................................................27 Dr. Björn Gücker................................................................................................................................29 Dr. Anke Hildebrandt.........................................................................................................................31 Dr. Johan Alexander (Sander) Huisman.............................................................................................32 Dipl.-Geoökol. Sascha Christian Iden................................................................................................34 Dr. Sonja C. Jähnig.............................................................................................................................36 Dr. David Kneis..................................................................................................................................38 Dipl.-Ing. Stefan Krämer....................................................................................................................40 Dipl.-Geoökol. Tobias Krueger..........................................................................................................42 Dipl.-Geol. Christine Kübeck.............................................................................................................44 Dr. Eva Nora Müller...........................................................................................................................46 Dr. Lars Nerger...................................................................................................................................48 http://www.dfg-wasserkommission.de Back to contents 3 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Ing. Benjamin Nitsch................................................................................................................49 Dr. Karsten Rinke...............................................................................................................................51 Dipl.-Biomath. Stefanie Rost.............................................................................................................53 Dr. Nele Schuwirth.............................................................................................................................55 Dipl.-Ing- Leopold Stadler.................................................................................................................56 Dr. Henning Wilker............................................................................................................................59 http://www.dfg-wasserkommission.de Back to contents 4 of 88 Young Scientists Workshop on Data Assimilation Aims of the workshop The workshop is organised by the Senate Commission on Water Research (KoWa) of the German Research Foundation (DFG) for advanced young researchers working on problems of data assimilation, uncertainties, and the prediction capabilities of models in the (water-related) fields of hydrology, hydrogeology, aquatic ecology and meteorology to participate in this interdisciplinary workshop. By data assimilation, we mean the integration of measurement data into models with which system analysis and predictions should be made. If a model can reproduce data, we assume that it can also predict processes. Data are treated with different approaches in various models. In some disciplines, such as meteorology, for example, scientists attempt to objectify the integration of data into models. If measurement data and models are combined, uncertainties in the data as well as in the models need to be considered and quantified. In the workshop, methods from various disciplines for integrating data into models and quantifying the pertaining uncertainties will be discussed. The general intention of the workshop is to develop new ideas for interdisciplinary research projects where the approaches and methods of the disciplines will be exchanged and combined. We assume that it will be easier to initiate interdisciplinary projects focusing on methods rather than having to define the interfaces between the models of the different disciplines when addressing a thematic topic. Aims of the workshop • Bringing together young postdocs and doctoral candidates in the final stage of their thesis work on various water-related areas • Broadening the horizon of participants by learning about methods from neighboring disciplines • Providing a forum for the development of further interdisciplinary research topics and project initiatives The river Elbe near Lüneburg (from www.lueneburg.de) http://www.dfg-wasserkommission.de Back to contents 5 of 88 Young Scientists Workshop on Data Assimilation Programme Monday, 8th October 2007 18:00 Arrival and mounting of posters 20:00 Welcome address by Prof. Dr. Dietrich Borchardt, spokesman of the KoWa workgroup, presentation of KoWa and the aims of this workshop Ice breaker Tuesday, 9th October 2007 09:00 – 09:40 Keynote lecture Meteorology: Dr. Matthias Drusch, Reading, UK Chair: Dr. Felix Ament 09:40 – 10:00 Discussion 10:00 – 10:40 Keynote lecture Hydrogeology: Prof. Dr. Sjoerd van der Zee, Wageningen, The Netherlands Chair: Dr. Anke Hildebrandt 10:40 – 11:00 Discussion 11:00 – 11:30 Coffee break 11:30 – 12:30 Poster session I 12:30 – 13:30 Lunch 13:30 – 14:10 Keynote lecture Aquatic Ecology: Prof. Dr. Peter Reichert, Dübendorf, Switzerland Chair: Dr. Karsten Rinke 14:10 – 14:30 Discussion 14:30 – 15:10 Keynote lecture Hydrology: Prof. Dr. Günter Blöschl, Wien, Austria Chair: Dr. Helge Bormann 15:10 – 15:30 Discussion 15:30 – 16:00 Coffee break 16:00 – 17:00 Poster session II 17:00 – 18:30 Meeting of disciplinary workgroups 19:00 – 20:30 Dinner 20:30 – Short presentation of the disciplinary workgroups' results Arrangement of the interdisciplinary workgroups http://www.dfg-wasserkommission.de Back to contents 6 of 88 Young Scientists Workshop on Data Assimilation Wednesday 10th October 2007 9:00 – 10:15 First meeting of the interdisciplinary workgroups 10:15 – 10:45 Coffee break 10:45 – 12:00 Funding opportunities Dr. Ute Weber, DFG Experience of the Emmy-Noether Programme of the DFG Dr. Insa Neuweiler, Group 'Effective Soil Parameters for Infiltration Processes', Universität Stuttgart 12:00 – 13:00 Lunch 13:00 – 15:00 Continued meeting of the interdisciplinary workgroups 15:00 – 15:30 Coffee break 15:30 – 18:30 Presentations of the results of the interdisciplinary workgroups 19:00 – 20:00 Dinner 20:00 Presentation of the results of the interdisciplinary workgroups (optional) Meeting in a pub in Lüneburg (Mälzer Brau- und Tafelhaus) Thursday 11th October 2007 9:00 – 11:00 Final discussion and feedback Locations: Main room: Plenarsaal 1-2 Additional rooms for workgroups Tuesday, 9th October: Konferenzraum Bad Boll, Gruppenraum 7, Palmengarten Additional rooms for workgroups Wednesday, 10th October: Seminarraum 3, Plenarsaal 3 Market Place in Lüneburg (www.lueneburg.de) http://www.dfg-wasserkommission.de Back to contents 7 of 88 Young Scientists Workshop on Data Assimilation Organizing Committee Prof. Dr. D. Borchardt Prof. Dr. E. Günther Helmholtz-Centre for Environmental Technische Universität Dresden Research – UFZ, Magdeburg Chair for Environmental Management Department Aquatic Ecosystem Analysis and Management Prof. Dr. M. Isenbeck-Schröter Ruprechts-Karls-Universität Heidelberg Institute of Environmental Geochemistry Prof. Dr. K.-O. Rothhaupt Universität Konstanz Limnological Institute Prof. Dr. M. Sauter Universität Göttingen Applied Geology Prof. Dr. C. Simmer Universität Bonn Meteorological Institute Prof. Dr. P. Werner Technische Universität Dresden Institute for Waste Management and Contaminated Site Treatment Prof. Dr. Sabine Attinger Helmholtz-Centre for Environmental Research – UFZ, Leipzig Department Computational Hydrosystems Office of the DFG-Senate Commission on Water Research (KoWa) German Research Foundation Deutsche Forschungsgemeinschaft (DFG) Dr.-Ing S. Manthey Institut für Wasserbau Universität Stuttgart Pfaffenwaldring 61 70569 Stuttgart Tel.: +49 711 685 60076 Fax: +49 711 685 60430 email: [email protected] Dr. U. Weber DFG-Geschäftsstelle Kennedyallee 40 53175 Bonn http://www.dfg-wasserkommission.de Back to contents 8 of 88 Young Scientists Workshop on Data Assimilation Workshop Venues Seminaris Hotel Lüneburg Soltauer Str. 3 21335 Lüneburg Tel.: (04131) 713-0 Fax: (04131) 713-727 [email protected] Mälzer Brau- und Tafelhaus Heiligengeiststraße 43 21335 Lüneburg Back to contents http://www.dfg-wasserkommission.de Back to contents 9 of 88 Young Scientists Workshop on Data Assimilation Prof. Dr.techn. Günter Blöschl keynote speaker Hydrology Institut für Wasserbau und Ingenieurhydrologie Vienna University of Technology Karlsplatz 13/222 A-1040 Wien Austria Tel.: +43-1-58801-22315 Fax: +43-1-58801-22399 Email: [email protected] Research interests • Hydrology and river basin management • Spatially distributed modelling and simulation, the issue of scale and scaling, and estimation in the absence of data • Design of flood estimation for hydraulic structures, hydrological forecasts and flood risk Education 1985 Diploma in Civil Engineering 1990 Doctorate in Hydrology 1997 Habilitation in Hydrology at the Vienna University of Technology 1985 – 1997 Research Assistant 1997 Associate Professor at the Vienna University of Technology http://www.dfg-wasserkommission.de Back to contents 10 of 88 Young Scientists Workshop on Data Assimilation Dr. Matthias Drusch Keynote speaker Meteorology Data Assimilation Section ECMWF Shinfield Park Reading RG2 9AX UK e-mail: [email protected] Research Interests • Remote sensing of the earth's surface and radiative transfer modelling. • Numerical Weather Prediction and hydrological applications. • Data assimilation and surface analyses. Education & Professional Experience 10 / 2002 – now Consultant in the Data Assimilation Section, European Centre for Medium-range Weather Forecasts 06 / 2001 - 12 / 2004* Principal Investigator / Project Manager, Meteorological Institute, University of Bonn, German Climate Research Programm DEKLIM- E: „Hyperspektrale Satellitendaten Analyse über Landoberflächen zur Anwendung in der Klimamodellierung“ (*on secondment from 10 / 2002) 06 / 2000 - 05 / 2001 Research Associate, Meteorological Institute, University of Bonn CLIWANet (EU-FP5) 09 / 1998 - 05 / 2000 Research Associate, Dept. Civil Engineering, Princeton University 10 / 1990 - 04 / 1994 Research Student, Institute for Marine Sciences, Kiel University June 1998 PhD / Promotion Meteorological Institute at the University of Bonn (Prof. Dr. C. Simmer, „Fernerkundung von Landoberflächen mit multispektralen Satellitendaten“) 1996-1998 Research Scientist, Meteorological Institute at the University of Bonn 1994-1996 Research Scientist, Institute for Marine Sciences at Kiel University May 1994 Diploma in Meteoroloy (Prof. Dr. E. Ruprecht,„Vergleich zwischen der optischen Dicke aus ISCCP C1 Daten und dem Flüssigwassergehalt aus SSM/I Messungen“) http://www.dfg-wasserkommission.de Back to contents 11 of 88 Young Scientists Workshop on Data Assimilation Prof. Dr. Peter Reichert keynote speaker Aquatic Ecology Department of Systems Analysis, Integrated Assessment and Modeling (SIAM) Swiss Federal Institute of Aquatic Science and Technology (Eawag) PO Box 611 CH - 8600 Dübendorf Switzerland Tel +41 44 823 5281 Fax +41 44 823 5375 [email protected] Professor at: Department of Environmental Sciences Swiss Federal Institute of Technology (ETH) CH - 8092 Zürich Switzerland Research Interests • • • Identification of models for technical and natural aquatic systems using methods of systems analysis. Assessment of the identifiability of model parameters and the uncertainty of model predictions using frequentist and Bayesian techniques. Application to data from rivers (river water quality modeling files), lakes (lake water quality modeling files), lake sediments, biofilms, sewage treatment plants, soil columns and laboratory reactors. Integration of knowledge from various sources (literature, data, output of more detailed models, expert opinions) in the form of probability network models. Use of such models to support environmental decision making. Design and implementation of software that supports environmental scientists in modelbased evaluation of their data ( AQUASIM, UNCSIM ) Education 1981: M.Sc. in theoretical physics, University of Basel, Basel, Switzerland. 1985: Ph.D. in theoretical solid state physics, University of Basel, Basel, Switzerland. 1995: Habilitation in environmental systems analysis at the Department of Environmental Sciences, Swiss Federal Institute of Technology (ETH), Zürich, Switzerland. http://www.dfg-wasserkommission.de Back to contents 12 of 88 Young Scientists Workshop on Data Assimilation Prof. dr.ir. Sjoerd E.A.T.M. van der Zee keynote speaker Hydrogeology Wageningen University Soil physics, ecohydrology and groundwater management group Droevendaalse steeg 4 Building ATLAS number 104 6700 AA Wageningen The Netherlands Tel.: +31 317 482875 Fax: +31 317-484885 email: [email protected] Research interests • Heterogeneity, spatial variability, and scale problems in soil pollution research. • Multiphase fow and nonlinear biogeochemical interaction affecting transport. • Mathematical and stochastic approaches to deal with flow and transport and their uncertainties, for natural porous media. • Biological availability of contaminants and nutrients for plants and other organisms. • Long term effects of soil management. Scientic education 1988 Doctoral degree at Wageningen Universiteit,Wageningen 1988 - 1992 UD / Assistant professor at Wageningen Universiteit 1992 - 1998 UHD / Associate professor 1998 - present Full professor Professional experience 1981 - 1984 Grondmechanica Delft, Delft 1981 - 1983 Project leader contaminant transport modelling 1983 - 1984 Permanent position, project leader soil and groundwater 1984 - present Wageningen Universiteit,Wageningen 1998 - 2005 Professor (Personal Title) for Soil Pollution and Soil Protection at the chair of Soil Chemistry and Chemical Soil Quality since 2005 Full professor Soil Physics, Ecohydrology, and Groundwater Management http://www.dfg-wasserkommission.de Back to contents 13 of 88 Young Scientists Workshop on Data Assimilation Dr. Felix Ament Dr. Felix Ament *1975 Degree: Dipl. Meteorology (Univ. Bonn, 2001); Ph.D., Meteorology (Univ. Bonn, 2006). Meteorology Curr. Position: Scientist in the group “Numerical Models” at the Federal Office of Meteorology and Climatology MeteoSwiss. Appointed to become Juniorprofessor at the Meteorological Institute, University of Hamburg Research interests: land-atmosphere exchange, representation of the hydrological cycle in atmospheric models, interface between models and observations (assimilation, verification, model devolpment) Motivation huge overlap with my research interests; interested to learn new techniques, viewpoints or questions; gathering scientific and funding ideas to establish a new group devoted to the combination of observations and model data at the University of Hamburg in 2008. Current address MeteoSchweiz, Kraehbuehlstrasse 58, P.O. Box 514, CH-8044, Zurich, Switzerland; [email protected] Selected Publications related to the workshop: Ament, F. and C. Simmer: Improved Representation of Land-Surface Heterogeneity in a NonHydrostatic Numerical Weather Prediction Model. Boundary-Layer Meteorology,121(1), 153ff, 2006 Abstract MAP D-PHASE project: Demonstration of today’s capabilities to predict alpine flood events by atmospheric and hydrological models Keywords: flood forecasting, precipitation forecast, uncertainty The aim of the forecast demonstration project MAP D-PHASE is to demonstrate the performance of today’s models to forecast heavy precipitation events and floods in the Alpine region. Therefore an end-to-end, real-time forecasting system has been installed and is operated during the demonstration phase from June until November 2007 (www.d-phase.info). Part of this system are more than 20 numerical weather prediction models operated by weather services and research institutes, which issue alerts if forecasted precipitation accumulations exceed critical thresholds. Additionally, all relevant model fields of these simulations are stored in a central data archive. This comprehensive data set allows a detailed assessment of today’s quantitative precipitation forecast (QPF) performance. The D-PHASE system is completed by a wealth of local hydrological models using the precipitation forecasts as forcing data. Some hydrological models are operated as ensemble system by performing individual simulations for each member of an atmospheric ensemble system. This technique allows studying the propagation of uncertainties from precipitation forecast to discharge simulations. The first part of the poster will give an overview of the D-PHASE system: its components the interfaces and the new concepts to provide useful information to end-users. The second part is http://www.dfg-wasserkommission.de Back to contents 14 of 88 Young Scientists Workshop on Data Assimilation attributed to first results: The overall QPF performance is assessed by a verification against Swiss Radar data both from a qualitative point of view, in terms of alerts, as well as from a quantitative perspective, in terms of precipitation rate. This analysis is refined and extended considering hydrological models for a severe flood event from 6 until 8 August 2007. Both exercises clearly reveal the great uncertainty in quantitative aspects of QPF and that it might be very dangerous to rely on a single deterministic forecasting system (see Fig. 1). Figure 1: Accumulated precipitation observations by radar (thick black) in comparison to deterministic model predictions for two adjacent D-PHASE warning regions in Switzerland. Note that models being accurate in one region (e.g. COSMOCH7 (thick blue) and ISACMOL (yellow)) do not necessarily perform well in the other one. http://www.dfg-wasserkommission.de Back to contents 15 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 16 of 88 Young Scientists Workshop on Data Assimilation Dr. Hans-Stefan Bauer Meteorology Current affiliation: Institute of Physics and Meteorology University of Hohenheim Garbenstrasse 30, 70599 Stuttgart, Germany phone: +49 711 459 2154 E-mail: [email protected] CV 1996 - 2000 2000 - 2001 PhD student Max Planck Institute for Meteorology / Hamburg Scientist at the Max Planck Institute for Meteorology / Hamburg Abstract Assimilation of high resolution water vapour observations into the MM5 4DVAR system H.-S. Bauer, M. Grzeschik, F. Zus, V. Wulfmeyer and A. Behrendt Institute of Physics and Meteorology, University of Hohenheim, Garbenstrasse 30, 70599 Stuttgart Precipitation strongly affects many aspects of our economy and general livelihood. Therefore, efforts to improve quantitative precipitation forecasting (QPF) have high priority in meteorological research. This is particularly challenging in the warm season where convective precipitation is more important. However, the short-term precipitation forecast of strong local precipitation events is still of very low quality. The reasons for this are manifold and include inaccuracies in parameterization schemes, severe gaps and faults of observations of the pre-convective environment, and an insufficient resolution of the forecast models. The limited availability and accuracy of observational data needed as initial and boundary conditions for running numerical weather prediction (NWP) models is considered as one of the most critical deficiencies in QPF. Recent results in nowcasting and short-range weather forecasting clearly demonstrate that variational data assimilation in combination with high-resolution modelling is essential for optimising the quality of QPF on the mesoscale. Particularly, it is crucial to determine the initial state of the model with respect to dynamics and water in all its phases as accurate as possible. We are intending to make a significant contribution to quantitative precipitation forecasting (QPF) by focusing on the development of new and the extension of existing forward operators for the assimilation of additional types of observations into the MM5 4DVAR system. Here, the existing radiosonde operator was generalized and combined with an observation pre-processor to be able to assimilate data of the airborne water vapour DIAL system LASE (Wulfmeyer et al., 2006), as well as data of ground-based Raman lidar systems (Grzeschik et al., 2007). With recent advances in Global Positioning System (GPS) atmospheric remote sensing, groundbased GPS receivers have become an important instrument that can provide high resolution water vapour measurements operationally at low cost with an accuracy of a few millimetres in all weather conditions. Therefore, a new operator to assimilate GPS slant path delays was developed. At the moment, it is in operational use in a real-time assimilation system during the WWRP Research and Development Project COPS (Convective and Orographically-induced Precipitation Study) and the WWRF Forecast Demonstration Project (FDP) D-PHASE (Demonstration of Probabilistic Hydrological and Atmospheric Simulation of Flood Events in the Alpine Region). http://www.dfg-wasserkommission.de Back to contents 17 of 88 Young Scientists Workshop on Data Assimilation This contribution shall give an overview about the assimilation efforts done so far in Hohenheim as well as an outlook to planned future activities. As an example, Figures 1 shows the impact of the assimilation of GPS slant path data on the specific humidity and temperature fields at 850 hPa for two different time steps from the forecast initialized at 00Z, 27th of June 2007. Figure 1: Difference 4DVAR-CONTROL of the specific humidity [g/kg] (top row) and temperature [K] (bottom row) at 850 hPa for the initial time 20070627 00Z (left) and 12 hours later (right). It is clearly seen, that the assimilation of GPS data has a large-scale impact on both fields, although only water vapour data was assimilated. References: Grzeschik, M. H.-S. Bauer, V. Wulfmeyer, D. Engelbart, U. Wandinger, I. Mathis, D. Althausen, R. Engelmann, M. Tesche, and A. Riede, 2007: Four-dimensional variational analysis of water-vapor Raman lidar data and their impact on mesoscale forecasts. J. Atmos. Oceanic Technol. Accepted. Wulfmeyer, V., H.-S Bauer, M. Grzeschik, A. Behrendt, F. Vandenberghe, E. Browell, S. Ismail, and R. Ferrare, 2006: 4-dimensional variational assimilation of water vapor differential absorpion lidar data. The first case study within IHOP 2002. Mon. Wea. Rev., 134, 1, 209-230. http://www.dfg-wasserkommission.de Back to contents 18 of 88 Young Scientists Workshop on Data Assimilation Dr. rer. nat. Christof Beyer Dr. Christof Beyer *1974 Degree: Geoecologist (Diploma, TU Braunschweig, 2002); Ph.D. (Applied Geology, University of Tübingen, 2007) Hydrogeology Curr. Position: Postdoc, Institute for Applied Geology, Geohydromodeling, Christian Albrechts University of Kiel Research interests: multi-phase flow & reactive transport modeling in heterogeneous media, environmental assessment of recycling materials, stochastic modelling & uncertainty propagation, geostatistics & regionalisation Motivation: to get in contact with colleagues and scientists from neighbouring disciplines, to learn more about how to address issues of integrating data from diverse sources into models and on new strategies or methods to tackle data or model uncertainty and to contribute and participate in an exchange and discussion of these concepts Current address: Angewandte Geologie / Geohydromodellierung, Ludewig-Meyn-Str. 10 D 24118 Kiel, Tel. +49 (0)431 880 2857, email: [email protected] Selected Publications related to the workshop: Beyer, C., Altfelder, S., Duijnisveld, W.H.M., Ingwersen, J., Streck, T. (2004): Räumliche Variabilität und statistische Unsicherheit im Kontext der flächenhaften Verlagerung von Schadstoffen mit dem Sickerwasser. Bodenschutz, 2004, 3, 92-98. Bauer, S., Beyer, C., Kolditz, O. (2006): Assessing measurement uncertainty of first-order degradation rates in heterogeneous aquifers. Water Resour. Res., 42, W01420, Beyer, C., Bauer, S., Kolditz, O. (2006): Uncertainty Assessment of Contaminant Plume Length Estimates in Heterogeneous Aquifers. J. Contam. Hydrol., 87, 73-95, Bauer, S., Beyer, C., Chen, C., Gronewold, J., Kolditz, O. (2006): Virtueller Aquifer (VA) Computergestützte Bewertung von Erkundungs-, Sanierungs- und Monitoringstrategien im Hinblick auf das "Natural Attenuation" (NA) und "Enhanced Natural Attenuation" (ENA) -Potenzial kontaminierter Böden und Grundwässer. In: Mitteilungen des DGFZ e.V., 3/2006, 93-113. Dresden. Beyer, C., Bauer, S., Chen, C., Gronewold, J., Kolditz, O. (2007): Determination of first order degradation rate constants for well investigated contaminant plumes. Ground Water 45(6), 774-785. Abstract Using Virtual Aquifers for uncertainty assessment of site investigation strategies: Application to degradation rate estimation Christof Beyer , Jan Gronewold , Cui Chen , Olaf Kolditz , Sebastian Bauer 1 Center for Applied Geoscience, University of Tübingen, Sigwartstr. 10, 72076 Tübingen; [email protected] 2 Helmholtz Centre for Environmental Research, Permoserstr. 15, 04318 Leipzig 3 Institute for Geosciences, University of Kiel, Ludewig-Meyn-Str. 10, 24118 Kiel 1 1 1 2 3 Keywords: Heterogeneity, site investigation, degradation rate, uncertainty, numerical modelling. Due to the limited accessibility of the subsurface, measurements of pollutant concentrations at contaminated sites usually are sparse and may not be representative of the heterogeneous hydrogeologic or geochemical conditions. As a result, any site investigation is subject to uncertainty, for which three main sources can be identified: incomplete or wrong description of parameter distributions due to site heterogeneity, measurement errors (e.g. in heads or http://www.dfg-wasserkommission.de Back to contents 19 of 88 Young Scientists Workshop on Data Assimilation concentrations), and conceptual model errors from an incorrect identification of the governing processes. Due to this uncertainty, field investigation methods can hardly be verified in the field. The Virtual Aquifer (VA) approach is particularly aimed at investigating this problem. The basic idea is the computer based evaluation of the performance of field investigation methods by application in heterogeneous synthetic (i.e. Virtual) aquifers. In this it resembles the concept of “virtual realities” which is used e.g. in car industry (virtual crash test) or medicine (interactive operation planning). Synthetic site models based on statistical properties of natural aquifers are generated by numerical simulation of contaminant spreading in groundwater. Just like for a “real world” site, the synthetic contamination can be investigated, e.g. by setting up observation well networks. The unique advantage of the VA is that the “true” spatial distribution of the contamination and all physical or geochemical parameters is exactly known, which allows to evaluate the accuracy of the investigation methods used by comparison of their results to the “true” parameter distribution. The VA concept is demonstrated by studying the uncertainty in field scale contaminant degradation rate constants estimated from measured field data: Multiple realizations of contaminant plumes under conditions of biodegradation in heterogeneous aquifers were individually and independently investigated in the computer by individual test persons engaged in hydrogeological research or consulting using an interactive plume investigation and mapping software. Investigation of the plumes was achieved by installation of extensive networks of observation wells and interpolation of measured concentrations to identify the geometry and position of the plumes. From the data measured at the wells, i.e. Contaminant concentrations, hydraulic conductivities and heads, first order degradation rate constants were estimated by (A) three different one-dimensional analytical models, which only use measurements located on the presumed plume center line, and (B) a two-dimensional model, which uses all measurements available downgradient from the contaminant source. Results for (A) show that the “true” rate constant (used for the numerical simulation of the plumes) in general tends to be overestimated by the 1D analytical center line models. Overestimation is strongest for narrow plumes from small source zones and may reach factors as high as 50 in the worst cases. For wider plumes from larger source zones overestimation decreases, but still is significant. The general overestimation is mainly due to three factors: i) deviations of observation well positions from the true plume center line, ii) biased estimates of the mean transport velocity and iii) wrong parameterization of dispersivity. For small source widths overestimation with the 2D model (B) is comparable to or at best slightly less than for (A). For wider plumes, however, incorporation of all observation wells in the estimation procedure yields significantly closer estimates of the degradation rate constant than for (A) on average. Hence, this approach should preferably be used for rate constant estimation. Overestimation of the degradation rate is a nonconservative result and thus a critical point if natural attenuation is considered as an remediation alternative, as the overall attenuation potential is assessed too positive. The VA application thus demonstrates the importance but also the pitfalls of a careful and accurate collection of investigation data, if these are to be used for a model prognosis of contaminant behaviour at a site or for comparing and optimizing alternatives of remediation measures. The main advantage of the VA is, that individual factors, such as site heterogeneity or conceptual model errors can be studied in detail at low costs, either individually or in combination, and under controlled conditions. As the sum of these possibilities can neither be provided by large scale field experiments nor in the laboratory, the VA concept can be considered a valuable contribution complementing state of the art experimental methods. Data scarcity and parameter uncertainty due to the heterogeneity of environmental systems rather than general system comprehension or the availability of advanced process based numerical models is a major problem also in other applied geoscientific disciplines (e.g. surface hydrology, climatology, geothermic reservoir engineering). Here a transfer of the VA concept may be possible and could help in getting a better understanding of the consequences of environmental systems management on basis of imprecise or uncertain data. http://www.dfg-wasserkommission.de Back to contents 20 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 21 of 88 Young Scientists Workshop on Data Assimilation Dr. Helge Bormann Jun.-Prof. Dr. Helge Bormann *1967 Hydrology Degree: Dipl. Geoecology (TU Braunschweig, 1995); Ph.D., Geography (Dr. rer. nat., Univ. Bonn 1999). Curr. Position: Jun.-Prof. for Hydrology, University of Oldenburg, Institute for Biology and Environmental Sciences Research interests: catchment hydrology, hydrological modelling, scaling, uncertainty analysis, model comparison, model ensembles, environmental change (climate, land use, human impacts), coastal hydrology Motivation: disciplinary as well as interdisciplinary research; exchange of techniques and methods between different research fields (learning from each other); future cooperation and project initiatives; landscape management Current address: University Oldenburg, Institute for Biology and Environmental Sciences, Hydrology Working Group, Uhlhornsweg 84, D-26129 Oldenburg Tel. +49 441-798 4459, Email: [email protected] Selected Publications related to the workshop: Bormann, H., Breuer, L., Gräff, T. & Huisman, J.A. (2007): Analysing the effects of soil properties changes associated with land use changes on the simulated water balance: A comparison of three hydrological catchment models for scenario analysis. Ecological Modelling, 209 (1), 29-40. Bormann, H. (2007): Analysis of the suitability of the German soil texture classification for the regional scale application of physical based hydrological model. Advances in Geosciences, 11, 7-13. Bormann, H, Breuer, L., Croke, B., Gräff, T., Hubrechts, L.,Huisman, J.A. Kite, G.W., Lanini, J., Leavesley, G.Lindström, G., Seibert, J., Viney, N.R., & Willems, P. (2007): Reduction of predictive uncertainty by ensemble hydrological modelling of discharge and land use change effects. IHP Technical Documents in Hydrology. In Druck. Bormann, H. (2006): Impact of spatial data resolution on simulated catchment water balances and model performance of the multi-scale TOPLATS model. Hydrology and Earth System Sciences, 10, 165-179. Bormann, H. (2006): Verbesserung der Prognosequalität hydrologischer Modelle durch den Einsatz von Modell-Ensembles für die Simulation von Hochwasserereignissen. In: Disse, M., Guckenberger, K., Pakosch, S., Yörük, A. & Zimmermann, A. (Hrsg.): Risikomanagement extremer hydrologischer Ereignisse. Forum der Hydrologie und Wasserwirtschaft. Heft 15.06, Band 2, Vorträge 2, S. 87-98. Bormann, H. (2005): Evaluation of hydrological models for scenario analyses: Signal-to-noiseratio between scenario effects and model uncertainty. Advances in Geosciences, 5, 43-48. Bormann, H., Giertz, S. & Diekkrüger, B. (2005): Hydrological catchment models between process representation, data availability and applicability for water management - case study for Benin. IAHS Publ. 295, S: 86-93. Stephan, K., Bormann. H. & B. Diekkrüger (Hrsg.)(2002): 5. Workshop zur Hydrologischen Modellierung. Möglichkeiten und Grenzen für den Einsatz hydrologische Modelle in Politik, Wirtschaft und Klimafolgeforschung. Kassel University Press, 158 S. Bormann, H., Diekkrüger, B. & C. Renschler (1999): Regionale Simulationen in der Hydrologie Quantifikation der Fehler und der Unsicherheiten. Zusammenstellung der Beiträge des 2. Workshops "Wasser- und Stofftransportes in großen Einzugsgebieten" in Gießen. Kassel University Press. S. 21-29. http://www.dfg-wasserkommission.de Back to contents 22 of 88 Young Scientists Workshop on Data Assimilation Abstract Reduction of predictive uncertainty by ensemble hydrological modelling of catchment processes and land use change Bormann, H., 2Breuer, L., 3Croke, B.F.W., 4Gräff, T., 5Hubrechts, L., 2Huisman, J.A., 6Kite, G. W., 7 Lanini, J., 8Leavesley, G., 9Lindström, G., 10Seibert, J., 11Viney, N.R., 12Willems, P. 2 University of Gießen, Institute for Landscape Ecology and Resources Management, Germany, 3 ICAM, SRES, The Australian National University, Canberra, ACT, Australia, 4University of Potsdam, Institute of Geoecology, Germany, 5Lisec NV, Genk, Belgium, 6Hydrologic Solutions, Pantymwyn, United Kingdom, 7University of Washington, USA, 8United States Geological Survey, Denver, USA, 9 Swedish Meteorological and Hydrological Institute, Norrköping, Sweden, 10 Stockholm University, Sweden, 12CSIRO Land and Water, Canberra, Australia, 13Katholieke Universiteit Leuven, Belgium. Key words: Multi-model ensemble, catchment hydrology, structural uncertainty Ensemble modelling, whereby predictions from several models are pooled in an attempt to improve prediction accuracy, has often been used in the climate and atmospheric sciences, but until recently, has received little attention in hydrology. Two recent examples of model ensembles are DMIP (Distributed model intercomparison project) and HEPEX (Hydrologic ensemble prediction experiment). One of the key aims of the ensemble approach is to reduce uncertainty in the modelled predictions. This abstract introduces the LUCHEM-initiative (Breuer et al., submitted; Huisman et al., submitted, Viney et al., submitted), initiated by the Working Group on Resources Management of the University of Gießen. Within LUCHEM, predictions of ten different catchment models applied to the mesoscale Dill basin (693km²) in central Germany are compared for a 20 years period (1980-1999) and used to generate ensemble predictions of current catchment characteristics and as well as of the effects of several projected land use changes. The applied models encompass a large range in inherent complexity and input requirements. In approximate order of decreasing complexity, they are DHSVM, MIKE-SHE, TOPLATS, WASIM-ETH, SWAT, PRMS, SLURP, HBV, LASCAM and IHACRES. The models are driven using the same input data sets such as spatial data sets (digital elevation model, land use classification, soil map) and time series (daily values of precipitation and weather data). Furthermore spatial interpolation of rainfall, adjustment of temperature by topography and plant parameters is homogenised to achieve most comparable model results. Streamflow at the catchment outlet is used to compare simulations to observations and to evaluate model ensembles versus individual models using statistical quality measures. Comparing the simulated results of the individual models, the simpler models tend to perform better in both calibration and validation period, but while all models tend to show improved performance during the less extreme validation period. This improvement is greatest for some of the more complex models. Despite the disparity in model performance, ensemble predictions made up of various combinations of the ten model predictions outperform all of the individual models. Different model ensembles of all available models are applied such as daily mean value, daily median value. The two mentioned ensembles are simple ensembles as others represent optimized combinations of all (or only selected) models. Different quality measures are applied to analyse the results of model ensembles and individual models such as Nash-Sutcliffe efficiency, square of correlation coefficient, root mean squared deviation and bias. For the calibration period, as expected, an ensemble based on a multi-variable regression of all models provides the best predictions, but its prediction accuracy declines to a greater extent than all of the models in terms of both its bias and Nash-Sutcliffe efficiency when used for the validation period. For the validation period, the best predictions using an ensemble of all models are provided by an ensemble consisting http://www.dfg-wasserkommission.de Back to contents 23 of 88 Young Scientists Workshop on Data Assimilation of the daily median model predictions. The predictions of this median ensemble improve more between calibration and validation than any of the other models, thus indicating its robustness. Focusing not only on daily water flows and water balances, the study reveals also that the model ensembles show the best performance also for peak flows and for low flow periods. After calibration and validation, the models are then applied to different land use change scenarios. In the scenarios, the projected patterns of land use are based on assumed average field sizes of 0.5 ha, 1.5 ha and 5.0 ha, respectively. There is broad agreement among the models on the expected hydrological change. While only two models predict a significantly different change between current state of land use and the scenarios, all models predict the same trend between the three land use scenarios. The successful validation of the mean and median ensembles suggests that model ensembles can be applied successful not only in atmospheric but also in hydrological sciences, namely with respect to the calculation of catchment water flows, catchment water balances, peak flows and low flow periods. In the case of the Dill catchment, there was no model performing better than the best model ensemble in the Dill catchment. Those model ensembles which were optimized for the calibration period performed worse for the validation period. Based on the agreement of the different models with respect to simulated land use change effects, it seems to be possible to predict both, the direction and the magnitude of streamflow changes associated with land use change scenarios with an increased degree of confidence. Predictive uncertainty of individual models is reduced significantly. Breuer L, Huisman JA, Willems P, Bormann H, Bronstert A, Croke BFW, Frede H-G, Gräff T, Hubrechts L, Jakeman AJ, Kite G, Leavesley G, Lanini J, Lettenmaier DP, Lindström G, Seibert J, Sivapalan M, Viney NR. Assessing the impact of land use change on hydrology by ensemble modeling (LUCHEM) I: model intercomparison of current land use. Adv Water Resour; submitted. Huisman JA, Breuer L, Bormann H, Bronstert A, Croke BFW, Frede H, Gräff T, Hubrechts L, Jakeman AJ, Kite GW, Lanini J, Leavesley G, Lettenmaier DP, Lindstr¨om G, Seibert J, Sivapalan M, Viney NR, Willems P. Assessing the impact of land use change on hydrology by ensemble modelling (LUCHEM) III: Scenario analysis. Adv Water Resour; submitted. Viney NR, Bormann H, Breuer L, Bronstert A, Croke BFW, Frede H-G, Gräff T, Hubrechts L, Huisman JA, Jakeman AJ, Kite G, Leavesley G, Lanini J, Lettenmaier DP, Lindström G, Seibert J, Sivapalan M, Willems P. Assessing the impact of land use change on hydrology by ensemble modeling (LUCHEM) II: Deterministic and probabilistic ensemble combinations and predictions. Adv Water Resour; submitted. http://www.dfg-wasserkommission.de Back to contents 24 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 25 of 88 Young Scientists Workshop on Data Assimilation Dr.-Ing. Jörg Dietrich Dr.-Ing. Jörg Dietrich *1970 Degree: Ph.D., Civil Engineering (Ruhr-Univ. Bochum, 2006); Dipl. Geoecology (Tech. Univ. Braunschweig, 2000). Hydrology Curr. Position: Scientific assistant, project coordinator, BMBF joint research project “Ensemble forecasts in Operational Flood Risk Management – Mulde River Basin”, Ruhr-University Bochum. Research interests: hydrological modelling, decision support models and systems (strategic and operational), integrated river basin management, flood risk management Motivation: My working group has elaborated an ensemble based flood forecast strategy, which combines meteorological and hydrological ensembles. I offer to share the experiences gained in this work with others, and I’m also interested in discussion of the various methods of data assimilation and uncertainty assessment developed or conceptualized by other young researchers. My personal interest is in information fusion methods under uncertainty, including information from different sources. Current address: Lehrstuhl für Hydrologie, Wasserwirtschaft und Umwelttechnik (Institute of Hydrology, Water Resources Management and Environmental Engineering), Ruhr-Universität, 44780 Bochum, Tel. +49-234-32-25875, Email: [email protected], [email protected] Selected Publications related to the workshop: Dietrich, J., Trepte, S., Wang, Y., Schumann, A. H., Voß, F. & Hesser, F. (2007): Probabilistic flood forecast using different types of model ensembles - Mulde case study. Submitted to Nonlinear Processes in Geophysics. Dietrich, J. & Funke, M. (2007): Integrated catchment modelling for strategic planning and decision making: Werra case study. Submitted to Physics and Chemistry of the Earth. Dietrich, J., Voß, F. & Schumann, A. (2007): Operational Flood Risk Management Based on Ensemble Predictions. In: A. Schumann et al. (Ed), Reducing the Vulnerability of Societies to Water Related Risks at the Basin Scale, Proceedings of the 3rd International Conference on Integrated Water Resources Management, Bochum, September 2006. IAHS Publ. 317. Wang, Y., Dietrich, J., Voß, F. & Pahlow, M. (2007): Identifying and reducing model structure uncertainty based on analysis of parameter interaction. Advances in Geos-ciences Vol. 11. Dietrich, J. (2006): Entwicklung einer Methodik zur systemanalytischen Unterstützung adaptierbarer Entscheidungsprozesse bei der integrierten Flussgebietsbewirtschaftung. Schriftenreihe des Lehrstuhls für Hydrologie, Wasserwirtschaft und Umwelttechnik, Band 21, zugleich Dissertation Ruhr-Universität Bochum. Dietrich, J., Schumann, A. H. & Lotov, A. V. (2006): Workflow oriented participatory decision support for integrated river basin planning. In: Castelletti, A. & Soncini Sessa, R. (Eds.): Topics on System Analysis and Integrated Water Resource Management (IWRM). Elsevier. Dietrich, J. & Schumann, A. (Hrsg.) (2006): Werkzeuge für das integrierte Flussgebietsmanagement – Ergebnisse der Fallstudie Werra. Konzepte für die nachhaltige Entwicklung einer Flusslandschaft Bd. 7, Weissensee-Verlag, Berlin, 470 S. Schöniger, M., Dietrich, J., Groth, P. & Hattermann, F. (2002): Geological Reconstruction using Conditional Stochastic Simulation for Uncertainty Analyses of Water Rescources Management (Northwest Germany, Liebenau). In: Steenvoorden, J., Claessen, F. & Willems, J. (Hrsg.): Agricultural Effects on Ground and Surface Waters: Research at the Edge of Science and Society. Proceedings of an International Conference, Wageningen, The Netherlands, 1-4 October 2000, IAHS Publ. No. 273. http://www.dfg-wasserkommission.de Back to contents 26 of 88 Young Scientists Workshop on Data Assimilation Abstract Information Fusion in Probabilistic Flood Forecasting Using Ensembles from Different Sources Dr.-Ing. Jörg Dietrich, Dipl.-Ing. Yan Wang ([email protected]) Keywords: ensemble forecast, data fusion, evidence theory, rainfall-runoff model Flood forecasts are essential to issue reliable flood warnings and to initiate flood control measures on time. The accuracy and the lead time of forecasts for the headwaters primarily depend on the meteorological forecasts. Ensemble flood forecasts aim at framing the possible future development of the hydro-meteorological situation, admitting and communicating the imperfectness of the forecast (Toth et al. 2003). This contribution presents a framework for probabilistic flood forecast simulation based on operational meteorological ensemble prediction systems and rainfall-runoff-models. We propose an adaptive modeling strategy. For medium to longer lead times (2-10 days, COSMO-LEPS and ECMWF ensembles), the face-values of the ensemble members are processed by a rainfall-runoff model or several models to account for structural uncertainty of the hydrological models. For each of the models parameter uncertainty can also be regarded (not subject of this contribution but covered by other researchers). For very short lead times (3 – 21 hours) SRNW-PEPS, COSMOLEPS and LMK forecasts are evaluated and processed by a single conceptual model (fig. 1). Here parameter uncertainty is regarded by the generation of parameter ensembles from efficient solutions obtained by calibration of historic flood events of a similar type. One question to solve in the longer time is: when to release water from a reservoir to increase flood retention capacities, and when to classify an alarm as false alarm and stop measures? In the very short time, a probabilistic forecast of discharge values for vulnerable sites is a main subject of interest. The poster presents the application of evidence theory as a tool for information fusion and decision support. Evidence theory has been developed as a generalization of Bayesian theory, which introduces the concept of belief and plausibility of hypotheses (Shafer 1976, Klir & Smith 2001). The possible contribution of information fusion to support the type of decisions mentioned above is discussed. A challenging and possibly subjective task is the estimation of posterior distributions in extreme situations, where no records from the past exist. Finding a realistic sharpness of a probability distribution or a closed probability density function of the forecasted variable (i.e. discharge) is challenging. On the one hand, the forecasters use uncertainty bands, pdf’s etc. for communication of uncertainty, on the other hand decision makers are often interested in deterministic or even Boolean values (evacuate – yes or no?). As a case study we present flood forecasts for the August 2002 extreme flood event and a false alarm in August 2006, where a flood of similar severity had been forecasted. http://www.dfg-wasserkommission.de Back to contents 27 of 88 Young Scientists Workshop on Data Assimilation 5d 2d COSMO-LEPS ensemble meso-scale, medium range 21 h Probabilistic weather scenario meso-scale, short range Driving Force LMK-LAF ensemble local scale, very short range Driving Force Driving Force Training period: MC historic flood events Multi model ensemble ArcEGMO, NASIM, SWAT, HBV, … Inference Preconditions API, snow, season 12-24 hrly comp. Bayesian Model Average (BMA) Ass. Observations 0 Parameter ensemble ArcEGMO 3 hrly comp. Ass. discharge gauges Sequential ensemble update Probabilistic runoff scenario for the headwaters 2007/2008 flood routing/inundation models Figure 1. Conceptualization of an ensemble-based operational flood risk management – hydrological ensembles. References: Klir, G. J. & Smith, R. M. (2001): On measuring uncertainty and uncertainty-based information: Recent developments. Annals of Mathematics and Artificial Intelligence 32, 5-33. Shafer, G.A. (1976): A Mathematical Theory of Evidence. Princeton, NJ: Princeton University Press. Toth, Z., Talagrand, O., Candille, G. & Zhu, Y. (2003): Probability and Ensemble Forecasts. In: Joliffe, I. T. & Stephenson, D. B. (eds.): Forecast Verification: A Practitioner’s Guide in Atmospheric Science. John Wiley & Sons. http://www.dfg-wasserkommission.de Back to contents 28 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 29 of 88 Young Scientists Workshop on Data Assimilation Dipl. Envir. Sciences Martin Frey Hydrogeology Current affiliation: Environmental Eawag Überlandstrasse 133, P. O. Box 611, 8600 Dübendorf, Switzerland Phone: +41 (0)44 823 50 71 E-mail: [email protected] Chemistry CV 1997 – 2003 since 2004 Study of environmental science at the ETH Zurich with the extension in aquatic physics Diploma thesis: Modeling of the effect of hydropower plants on the temperature regime of the branch river of the Rhône PhD student in the Department of Environmental Chemistry at Eawag in Dübendorf: Hydrological modeling of contributing areas for diffuse herbicide losses to surface waters including an uncertainty analysis. Abstract Spatial prediction of contributing areas for diffuse pollution losses to surface waters Sensitivity and uncertainty analysis Keywords: Hydrological modelling, contributing areas, sensitivity analysis, Monte Carlo simulation Field studies have shown that herbicide losses often originate from a limited part of a catchment only. These areas are characterized by the occurrence of fast flow processes like surface runoff or macropore flow to tile drains. We test the predictability of such critical source areas in a small agricultural catchment in Switzerland, where herbicide losses were monitored in previous field studies. For that purpose, we modified the Soil Moisture Distribution and Routing (SMDR) model. SMDR is a simple distributed water balance model, which predicts saturation-excess overland flow areas in a watershed. The prediction is based on an estimation of the relative saturation of the soil by accounting for infiltration, lateral water flow, percolation and evaporation of water in the soil column. The model predictions are dependent on the spatially variable hydrological parameters of the catchment. The spatial subdivision follows soil map units. Prior estimations of the hydraulic parameters were derived with pedotransfer functions from the soil texture. Other parameters like the efficiency of drainage system were estimated by expert knowledge. Such estimations involve substantial uncertainty, which affects the prediction of the spatially distributed risk areas. http://www.dfg-wasserkommission.de Back to contents 30 of 88 Young Scientists Workshop on Data Assimilation To analyze the sensitivity of the spatial prediction to the individual parameter and to quantify the overall uncertainty of the model predictions the simulation program is linked with the system analysis software tool UNCSIM (Reichert, 2005). First calculations were carried with parameter estimated a priori based on literature values and data from pedotransfer function. The predicted discharge agreed reasonably well with the measured values. The prediction of the spatial distribution of contributing areas was more difficult to evaluate because no explicit validation data exist. The measured herbicide losses at the outlet of three subcatchment agreed partially with the predicted runoff processes on the herbicide application fields. The local sensitivity analysis revealed that the parameter describing the lower boundary of the soil (deep percolation and depth of the soil) are the most sensitive parameters. This is critical because classical soil maps contain only little information about these parameters. To test whether discharge data from the catchment helps to restrict the space of acceptable parameter values Monte Carlo calculations simulation were carried out. Simulations with a NashSutcliffe criterion > 0.6 yield significantly different parameter distributions and spatial distributions of runoff generation. This analysis will be extended in the future by using a Bayesian inference approach. References: Reichert, P, 2005, UNCSIM - A computer programme for statistical inference and sensitivity, identifiability, and uncertainty analysis. In: Teixeira, J.M.F. and Carvalho-Brito, A.E., eds., Proceedings of the 2005 European Simulation and Modelling Conference (ESM 2005), Oct. 24-26, Porto, Portugal, EUROSIS-ETI, pp. 51-55. http://www.dfg-wasserkommission.de Back to contents 31 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 32 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Geoökol. Bastian Graupner Hydrogeology Current affiliation: Dresden Groundwater Research Center (DGFZ e.V.) Meraner Straße 10, 01217 Dresden, Germany Phone: +49 (0)351 – 4050667 Email: [email protected] CV 1997 – 2003 Freiberg University of Technology Diplom-Geoökologe Diploma thesis over reactive transport modelling with MODCHEM and column tests for sulphate reduction Abstract Influence of the Lausitz Post Mining Region on Groundwater Quality – a Large Scale Approach Keywords: sulphate, pyrite oxidation, transport modelling In East Germany large post mining landscapes exist from open pit brown coal mining. They consist of mining dumps, post mining lakes and branched river systems. The research area covers a whole post mining region (about 2000 km²) located in the Lausitz southwest of Cottbus. A groundwater cone of depression has influenced an area of around 1350 km² during the last century whereof about 10 percent in area are mining dumps and about 3 percent are post mining lakes. Rising groundwater table and groundwater recharge releases pyrite oxidation products like iron and sulphur within the mining dumps. Highly concentrated groundwater in these dumps is formed with sulphate concentrations of up to 4000 mg/l. Polluted groundwater endangers surrounding groundwater bodies as well as downstream located surface water bodies. Describing future migration of these solutes with groundwater is necessary basis for required risk assessment and for developing protection strategies according to EU Water Framework Directive. A holistic approach starting with a detailed description of mining dumps as pollutant sources, geochemical alterations over time of deposition to matter transport simulation is used to deal with the problem dimension. Due to site dimension and large time scale questions like preparation of large data sets, assignation of point data to area, uncertainty of data and acceptable degree of simplification rise and have to be answered. And furthermore such a large area can’t be handled without automated procedures. Existing data are the main source of information; therefore effort is necessary for collection, preparation and validation of data. All calculations were carried out on a quasi 4D GIS-grid base with the additional dimensions depth and time. Input data and results are stored in that system providing anytime access to every processing step. Modelling of mining discharge requires a good knowledge of the matter source itself. Mass balance calculations based on a regional geological model were set up using borehole data from mining. Using geochemical properties of each geological layer yield the mean content of selected reactive http://www.dfg-wasserkommission.de Back to contents 33 of 88 Young Scientists Workshop on Data Assimilation minerals in each mining dump. An average primary pyrite oxidation turnover rate of 7 percent for each dump body was assumed representing oxidation during mining activities. Ongoing reactions with oxygen in the mining dumps, referred to as secondary pyrite oxidation, were considered by modelling pyrite oxidation due to oxygen diffusion through the unsaturated dump. Calculation of secondary pyrite oxidation processes were carried out at each grid point over the depth and time. Borehole samples from the dumps were analysed for sulphur species and showed good agreement with calculated results. A detailed knowledge of groundwater flow is a precondition for transport simulation. For the research area five detailed groundwater flow models exist. They were developed and maintained over more than one decade. The groundwater flow scenario was used as external information and gathered in form of data files. Results are put together and used as input for transport modelling using the tool PHT3D (Prommer, 2002) that links MT3D and PHREEQC. The release of matter from mining dump material to the groundwater is described with a double porosity concept. Exchange coefficients are derived from column experiments. All mining dumps within the research area held originally about 66 Million tons of pyrite sulphur. Approximately 15 percent of it was oxidised during and after mining and forms the source of groundwater pollution for the whole region. Transport modelling shows continuing mining impact on groundwater over the next decades with an area of influence of more than 100 km². Post mining lakes are due to their close position immediately affected by polluted groundwater. Rivers are as well influenced during the next decades. My research aims at the quantification of the impact of mining on the aquatic environment. For similar settings (surface mining in humid climate) it may be used in mine planning well ahead of mine closure planning. This contributes to the sustainable use of resources. References: Prommer, Henning (2002): A reactive multicomponent transport model for saturated porous media, Users Manual, www.pht3d.org http://www.dfg-wasserkommission.de Back to contents 34 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Hydrol. Jens Grundmann Jens Grundmann *1973 Hydrology Degree: Dipl. Hydrology (TU Dresden, 2001) Curr. Position: Research associate at the Institute of Hydrology and Meteorology at Dresden University of Technology Research interests: hydrological modelling of rainfall-runoff and watershed processes, optimisation, uncertainty analysis, river basin management and hydrodynamics Motivation The topic of the workshop sounds highly interesting to me since it deals with the topic of my main research field. At the moment I am investigating the uncertainties in rainfall-runoff modeling within the HORIX - project “Development of an Operational Expert System for Flood Risk Management Considering Prediction Uncertainty” founded by the BMBF-RIMAX flood research initiative. In this context I am most of all interested in methodological questions like: How can we minimize the effort of Monte-Carlo Simulations by the superposition of different uncertainties? How can we analyses in a reliable way the extrapolation ability of rainfall runoff models? as well as technical aspects like handling of large amounts of data (storage, analysis) and parallelization of optimization and simulation methods. Thus, from the workshop I expect an intensive exchange of experience and ideas with theother participants. Current address: Institute of Hydrology and Meteorology, Dresden University of Technology, Würzburger Str.46, D- 01187 Dresden Tel. +49 351 46335556 Email: [email protected] Selected Publications related to the workshop: Abstract Uncertainty analysis in rainfall-runoff modelling Jens Grundmann, Gerd H. Schmitz Institute of Hydrology and Meteorology, Dresden University of Technology, Germany Keywords: uncertainty analysis, monte-carlo simulation, optimisation Problem Especially in fast responding catchments the limited reliability and wobbliness of a deterministic flood forecast may easily jeopardise the forecast credibility. This can be overcome by considering in addition to the meteorological forecast uncer-tainties also the uncertainty of the hydrologic modelling. These http://www.dfg-wasserkommission.de Back to contents 35 of 88 Young Scientists Workshop on Data Assimilation latter uncertainty essentially arises from: • insuffisant catchment information (e.g. vague soil data) and • the likelihood to obtain a similar calibration quality (goodness of fit) from different parameter combinations, i.e. different model parameters may arise from a single calibration event (problem of equifinality). These uncertainties are exemplarily quantified for the flash flood-prone catchment Zöblitz in the ore mountain. Aims and methods Within our study we focus on the above mentioned sources of hydrologic model uncertainties. After analysing the uncertainties of individual parameter groups of the rainfall-runoff model (i.e. soil parameters and model parameters), the global uncertainty of rainfall-runoff modelling is evaluated. Finally we analyse the be-haviour pattern of these uncertainties along a growing catchment area. In this con-text we investigate different rainfall events in order to evaluate the impact of vary-ing process dynamics as regards the runoff formation. The uncertainties of individual parameter groups will be quantified according to the properties and the a-priori knowledge as regards these parameter groups. The following methods will be used: • Monte-Carlo simulations of distributed parameters sets to analyse the influ-ence of uncertain soil data with respect to runoff, Based on the range of soil structure information resulting from soil maps we derive soil hydraulic parameters via a pedotransfer function. We use the con-cept of soil similarity to describe the variability of the soil hydraulic character-istics relative to a single reference soil per soil type by only one scaling pa-rameter. This allows reducing the large number of parameters and to include their correlative dependencies in a Monte-Carlo simulation. • Parameter optimizations with evolutionary algorithms and Monte-Carlo simu-lations to analyse the uncertainty of conceptual model parameters, Based on Monte-Carlo simulations the most sensitive parameters will be iden-tified and analysed with respect to their correlative dependencies. Simultane-ously we use these simulations to derive task adequate objective functions. The selected parameters and objective functions serve as an input for different state-of-the-art optimisation algorithms (e.g. MOSCEM (Vrugt 2003), Cross entropy (Kroese 2006)) which provide the uncertainty information. A com-parison with the common GLUE methodology (Beven 1992) serves for verify-ing the obtained model parameter uncertainty. • Rainfall information is only provided at a distinct times and at only a few gauges in the catchment. The necessary interpolation results in a further com-ponent of hydrologic uncertainty. It is modelled by an ensemble of precipita-tion scenarios. These scenarios, describing time and space variability of rain-fall data, are generated by a turning bands method and serve as input for model simulation. Results Results of the above mentioned tasks will be presented for our test and research catchment of gauge Zöblitz,using the distributed hydrological model WaSiMETH (Schulla 1998). For example figure 1 shows the influence of uncertain soil data of a medium range flood event. Further investigations show, that the cor-responding variation coeffi-cient (v) decreases with increasing rainfall intensity. fig 1: flood event 9/1995 gauge Zöblitz http://www.dfg-wasserkommission.de Back to contents 36 of 88 Young Scientists Workshop on Data Assimilation Literature Beven, K.J, and Binley, A.M., 1992: The future of distributed models: model calibration and uncertainty prediction, Hydrological Processes, 6, 279-298. Kroese, D. P., Rubinstein, R. Y and Porotsky, S., 2006: Cross-Entropy Method for Continuous Multiextremal Optimization, Methodology and Computing in Applied Probability, 8, pp.383-407. Schulla, J., & Jasper, K., 1998: Modellbeschreibung WaSiM-ETH. - Geographisches Institut ETH Zürich. Vrugt, J.A., 2003: Effective and efficient algorithm for multiobjective optimization of hydrologic models, Water Resources Research, 39 (8), Artn 1214. http://www.dfg-wasserkommission.de Back to contents 37 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 38 of 88 Young Scientists Workshop on Data Assimilation Dr. Björn Gücker Dr. Björn Gücker *1975 Aquatic Ecology Degree: Dipl. Biol., Zoology and Ecology (Philipps University Marburg, 2001); Dr. rer. nat., Freshwater Ecology (Univ. Potsdam, 2005). Curr. Position: Postdoctoral research fellow of the German Academy of Sciences LEOPOLDINA; researcher at the Federal University of Minas Gerais, Brazil Research interests: ecosystem ecology and limnology, stream ecology (nutrient and organic carbon dynamics, energy flow, primary and secondary production, hydrodynamics) Motivation: In my current research, I am dealing with uncertainties in simple metabolic and biogeochemical models. Further, I am currently planning a project on the mechanistic upscaling of stream reach-scale biogeochemical process rates to river-system scale. Current address: Universidade Federal de Minas Gerais, Instituto de Ciências Biológicas, Depto. Botânica, Av. Pres. Antônio Carlos 6627, Pampulha, 31270-010 Belo Horizonte, Minas Gerais, Brazil, Phone: ++55-31-34944316, Email: [email protected], Web: http://www.bjoernguecker.net Selected Publications related to the workshop: Gücker B, Boëchat IG (2007) Impacts of agricultural land use on ecosystem structure and function of tropical Cerrado streams. Freshwater Biology (in press) Boëchat IG, Weithoff G, Krüger A, Gücker B, Adrian R (2007) A biochemical explanation for the success of mixotrophy in the flagellate Ochromonas sp.. Limnology & Oceanography: 52: 1624-1632 Gücker B, Pusch MT (2006) Regulation of nutrient uptake in eutrophic lowland streams. Limnology & Oceanography 51: 1443-1453 Gücker B, Brauns M, Pusch MT (2006) Effects of wastewater treatment plant discharge on ecosystem structure and function of lowland streams. Journal of the North American Benthological Society: 25: 313-329 Gücker B, Boëchat IG (2004) Stream morphology controls ammonium retention in tropical headwaters. Ecology 85: 2818-2827 Gücker B, Fischer H (2003) Flagellate and ciliate distribution in sediments of a lowland river: relationships with environmental gradients and bacteria. Aquatic Microbial Ecology 31: 67-76 Abstract Uncertainties in estimating and predicting stream ecosystem services Keywords: whole-stream primary production and respiration, nutrient spiraling In river systems, various natural and anthropogenic compounds emitted from the catchments, such as organic carbon (OC) and inorganic nutrients, are not only transported, but also stored, biogeochemically transformed, and thereby partially eliminated (1). Those biogeochemical http://www.dfg-wasserkommission.de Back to contents 39 of 88 Young Scientists Workshop on Data Assimilation processes in river systems play an important role in the global geochemical budget, which is heavily influenced by human impacts, such as eutrophication and climate change (2). Thus, biogeochemical processing of OC and nutrients is regarded as a natural ecosystem service, which significantly contributes to the well-being of human society by increasing the usability of limited water resources, such as stream and river water, floodplain aquifers, riverine lakes, and estuaries. Biogeochemical processing of OC and nutrients in running waters is most commonly estimated as reach-scale rates of whole-stream dissolved oxygen (DO) metabolism, i.e., respiration (R) and gross primary production (GPP), as well as rates of whole-stream gross nutrient uptake, and related subprocesses, such as nitrification, denitrification, etc. (3, 4). However, estimates of R and GPP, as well as those of nutrient uptake can be associated with considerable uncertainties, especially when process rates and stream water nutrient concentrations are low (5, 6). Estimates of R and GPP appear to be most sensitive to on-site estimates of atmospheric pressure and both coefficients of atmospheric streamwater reaeration and streamwater dilution by anaerobic groundwater. Estimates of whole-stream nutrient uptake seem to be most sensitive to the degree of spatial and temporal experimental replication as well as to nutrient concentrations measured at low concentration levels. Thus, efforts to improve the precision in measuring those parameters should significantly reduce uncertainties in estimating whole-stream R, GPP, and gross nutrient uptake. In fact, analyses presented in this contribution will demonstrate that uncertainties in R, GPP, and gross nutrient uptake estimates can be considerably diminished by the combined application of improvements in experimental design, adequate degrees of replication, as well as modern measuring techniques. A mechanistic understanding of ecosystem services, such as OC processing and nutrient uptake, at watershed scale is essential for the integrative and costeffective management of river catchments. This includes knowledge on the factors that control ecosystem services and their spatial and temporal dynamics, which could in turn be used for a mechanistic upscaling of reach-scale rates of ecosystem services to the watershed scale. Models that predict rates of ecosystem services successfully (explaining variations between 73 to 97%) are typically non-linear and involve multiple independent control variables (3, 7, 8). However, models differ strongly between stream systems as well as stream orders and include different hydrogeomorphic (9, 10), chemical (3, 11), and biotic control variables (3, 12). Though there are several approaches to predict stream ecosystem services at small spatial and temporal scales, there is, at present, neither a unifying mechanistic model available to predict long-term patterns of stream ecosystem services, nor to predict patterns of ecosystem services in different watersheds or even along several stream orders. References 1. J. D. Newbold, in The rivers handbook, P. Calow, G. E. Petts, Eds. (Blackwell Scientific, 1992), pp. 379-408. 2. P. M. Vitousek et al., Ecological Applications 7, 737 (1997). 3. B. Gücker, M. T. Pusch, Limnology and Oceanography 51, 1443 (2006). 4. B. Gücker, M. Brauns, M. T. Pusch, Journal of the North American Benthological Society 25, 313 (2006). 5. S. Hanafi, M. Grace, J. A. Webb, B. Hart, Ecosystems, in press. 6. J. R. McCutchan, W. M. Lewis, J. F. Saunders, Journal of the North American Benthological Society 17, 155 (1998). 7. R. O. Hall, J. L. Tank, Limnology and Oceanography 48, 1120 (2003). 8. P. J. Mulholland et al., Freshwater Biology 46, 1503 (2001). 9. B. Gücker, I. G. Boëchat, Ecology 85, 2818 (2004). 10. P. J. Mulholland, E. R. Marzolf, J. R. Webster, D. R. Hart, S. P. Hendricks, Limnology and Oceanography 42, 433 (1997). 11. W. K. Dodds et al., Journal of the North American Benthological Society 21, 206 (2002). 12. J. R. Webster et al., Freshwater Biology 48, 1329 (2003). http://www.dfg-wasserkommission.de Back to contents 40 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 41 of 88 Young Scientists Workshop on Data Assimilation Dr. Anke Hildebrandt Hydrology Current affiliation: Helmholtz Centre for Environmental Research – UFZ Hydrosystemmodellierung / Computational Hydrosystems Permoserstraße 15 / 04318 Leipzig / Germany Tel: +49 341 235 3973 email: [email protected] CV Since June 2005 June 2, 2005 2001-2005 at March 23, 2000 Postdoctoral Researcher at the Helmholtz Centre for Environmental Research UFZ. Graduation with a PhD in the field of Hydrology (MIT, Cambridge, USA) Graduate studies at the Department of Civil and Environmental Engineering the Massachusetts Institute of Technology (MIT, Cambridge, USA) Title of the PhD thesis: „On the ecohydrology of a seasonal cloud forest in Dhofar (Oman).“ Graduation from the Technical University of Dresden (Germany) with the degree of a „Diploma Engineer“ in the field of Water Resources Management. Abstract Modelling water uptake based on a model for a single root According to van Honert’s Model transpiration is a water flux along a potential gradient and through a series of pathways (soil, xylem, stomata) with different resistances. The discussion about which of those resistances, namely plant or soil resistance to water flow, dominates the transpiration process at high atmospheric demand is still ongoing. It is often argued that the root resistance is larger than the corresponding soil resistance to water flow, unless in very dry soils. Therefore the plant resistance is considered the main limitation to the transpirational flux under most soil moisture conditions. Here we argue, based on the investigation of water flow towards a single root, that the soil hydraulic resistance to water flow might be larger than usually assumed, even under conditions where bulk soil water content is elevated. However, plants might adapt to minimizing water shortages stemming from limiting soil hydraulic conductivity, for example by adjusting rooting density. http://www.dfg-wasserkommission.de Back to contents 42 of 88 Young Scientists Workshop on Data Assimilation Dr. Johan Alexander (Sander) Huisman Dr. J.A. (Sander) Huisman *1973 Hydrology Degree: Dipl. Soil physics and Soil chemistry (Univ. Amsterdam, 1997); Ph.D., Soil physics (Univ. Amsterdam, 2002). Curr. Position: Postdoctoral Forschungszentrum Jülich, Germany researcher at the Research interests: Hydrogeophysics, Soil physics, Watershed modelling Motivation. Recently, I have received a grant from the DFG (‘Projektleiterstelle’) to investigate methods to improve the use of geophysical measurements for hydrological model calibration. This project will start in September 2007, and besides me, the research team will consist of a postdoctoral researcher, a PhD student and a scientific programmer. In my poster presentation, I presented a new method to integrate geophysical measurements in hydrological models. I am aware of the fact that other disciplines are dealing with similar problems. For example, it is my (perhaps too simplistic) view that meteorologists use a plethora of data sources to improve their weather forecasts. Amongst these data sources, there might also be tomographic data that are prone to similar problems as the geophysical data of interest to me. During the workshop, I hope to find people dealing with similar problems and perhaps even to come across alternative methods to integrate geophysical data into hydrological models. Current address: Institute of Chemistry and Dynamics of the Geosphere (ICG), Institute 4: Agrosphere, Forschungszentrum Jülich, 52425 Jülich, Germany, Tel:++49-(0)2461-618607. Email: [email protected] Selected Publications related to the workshop: Breuer, L., J.A. Huisman and H-G. Frede (2006). Monte Carlo assessment of uncertainty in the simulated hydrological response to land use change. Environmental Modeling and Assessment, 11 (3), 209-218. Heimovaara, T.J., J.A. Huisman, J.A. Vrugt and W. Bouten (2004). Obtaining the spatial distribution of water content along a TDR probe using the SCEM-UA Bayesian inverse modeling scheme. Vadose Zone Journal, 3(4), 1128-1145. Huisman, J.A., L. Breuer and H.-G. Frede (2004). The sensitivity of simulated hydrological fluxes towards changes in soil properties in response to land use change. Physics and Chemistry of the Earth, 29(11-12), doi: 10.1016/j.pce.2004.05.012, 749-758. Huisman, J.A., S.S. Hubbard, J.D. Redman and P.A. Annan (2003). Measuring soil water content with ground penetrating radar: a review. Vadose Zone Journal, 2(4), 476-491. Huisman, J.A., J.J.J.C. Snepvangers, W. Bouten and G.B.M. Heuvelink (2003). Monitoring temporal development of spatial soil water content variation: comparison of groundpenetrating radar and time domain reflectometry. Vadose Zone Journal, 2(4), 519-529. Huisman, J.A., C. Sperl, W. Bouten and J.M. Verstraten (2001). Soil water content measurements at different scales: accuracy of time domain reflectometry and ground-penetrating radar, Journal of Hydrology, 245(1-4), 48-58. Lambot, S., L. Weihermüller, J. A. Huisman, H. Vereecken, M. Vanclooster, and E. C. Slob (2006). Analysis of air-launched ground-penetrating radar techniques to measure the soil surface water content, Water Resources Research, 42, W11403, doi: 10.1029/2006WR005097. Pohlert, T., L. Breuer, J.A. Huisman and H-G. Frede (2007). Assessing the model performance of an integrated hydrological and biogeochemical model for discharge and nitrate load predictions. Hydrology and Earth System Sciences, 11(2), 997-1011. http://www.dfg-wasserkommission.de Back to contents 43 of 88 Young Scientists Workshop on Data Assimilation Pohlert, T., J.A. Huisman, L. Breuer and H-G. Frede (2007). Integration of a detailed biogeochemical model into SWAT for improved nitrogen predictions – model development, sensitivity and uncertainty analysis. Ecological Modelling, 203(3-4), 215-228. Abstract An integrated data inversion approach to use geophysical measurements in hydrological models J.A. Huisman, S. Lambot, C. Oberdörster, K.Z. Jadoon, L. Weihermüller, J. Vanderborght and H. Vereecken. Keywords: Geophysical measurements, Model inversion, Hydrology. Geophysical measurements are a valuable source of information for the parameterization of hydrological models. Traditionally, relevant information on hydrological properties and/or state variables is obtained in a sequential approach from geophysical measurements: the geophysical survey data are inverted first, and the information thus obtained is used within the hydrological model calibration. When monitoring hydrological state variables, this ill-posed sequential inversion ignores the potential constraints that are provided by available hydrological information (e.g. mass balance, shape of wetting front). Additionally, artifacts of the geophysical data inversion (e.g. due to lack of sensitivity, high measurement and modeling errors, etc.) will directly translate in errors of the hydrological model calibration. The aim of this paper is to explore an alternative method to use geophysical information in hydrological model parameter identification. In the so-called integrated inversion approach, geophysical measurements are directly included in the hydrological inverse problem by coupling a forward model of the geophysical measurements with a hydrological model. The link between hydrological and geophysical state variables is established using petrophysical relationships. Hydrological and petrophysical parameters are obtained by minimizing the difference between modeled and observed hydrological and geophysical data. Figure 1 represents a general example of such integrated inversion scheme. First tests of this integrated hydrogeophysical inversion approach involved numerical test cases with three geophysical methods: Time Domain Reflectometry (TDR), Ground Penetrating Radar (GPR) and Electrical Resistivity Tomography (ERT). For all three methods, the numerical experiment investigated whether soil hydraulic parameters could be determined from time-lapse measurements during an infiltration experiment. In case of TDR and GPR, a one-dimensional http://www.dfg-wasserkommission.de Back to contents 44 of 88 Young Scientists Workshop on Data Assimilation Figure 1. Flow chart of the integrated hydrogeophysical data inversion approach. hydrological model based on the Richards equation with homogeneous soil hydraulic parameters was coupled to an electromagnetic wave propagation model. In the case of the ERT measurements, a homogeneous 2D model based on the Richards equations was coupled to a forward geophysical model describing electrical resistivity measurements. The inversion results demonstrated that the physical constraints provided by the hydrological model resulted in well-posed inverse problems that allowed determination of the soil hydraulic parameters from time-lapse measurements during infiltration. However, there are issues with the initial conditions, boundary conditions and the petrophysical relationships that need further investigation. More information on the GPR study can be found in Lambot et al. (2006). Future applications of the integrated hydrogeophysical data inversion framework will focus on the analysis of existing data sets (Krauthausen tracer experiments monitored with ERT, Selhausen infiltration experiments monitored with ERT and GPR) and the feasibility of joint inversion of a range of geophysical methods. Other frameworks for using geophysical data to calibrate hydrological models, such as ensemble Kalman filtering, might also be considered. References Lambot, S., E.C. Slob, M. Vanclooster, and H. Vereecken. 2006. Closed loop GPR data inversion for soil hydraulic and electric property determination. Geophysical Research Letters 33, L21405, doi:10.1029/2006GL027906. http://www.dfg-wasserkommission.de Back to contents 45 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 46 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Geoökol. Sascha Christian Iden Hydrogeology Current affiliation: Institut für Geoökologie Technische Universität Braunschweig Langer Kamp 19c, 38106 Braunschweig, Germany phone: +49 531 391 5602 E-mail: [email protected] CV 2002 since 10/2002 Graduation in Geoökologie Ph.D. Student, TU Braunschweig Abstract Free-Form estimation of the unsaturated soil hydraulic properties using global optimization Models used to predict the transport and fate of pollutants in soil rely on a proper soil hydraulic characterization because the soil hydraulic properties control the dynamics of water flow in the vadose zone (Simunek, 2005). The term “soil hydraulic properties” (SHP) summarizes a set of two constitutive relationships, namely the soil water retention curve relating water content to water potential, and the unsaturated hydraulic conductivity curve relating unsaturated hydraulic conductivity to water potential. The standard approach to simulate unsaturated water flow in porous media is the Richards equation (van Dam, 2004) which is a nonlinear partial differential equation that needs to be solved numerically under most initial and boundary conditions. Since the solution also requires knowledge of the SHP, their accurate knowledge is crucial for solving many basic and applied problems in environmental research (Vanclooster et al., 2004). Determining the SHP in the laboratory or in the field is preferably be achieved by conducting transient flow experiments in combination with inverse modelling (Hopmans et al., 2002; Finsterle, 2004; Vrugt and Dane, 2005; Durner and Lipsius, 2005). This requires a combination of numerical flow simulation and nonlinear parameter estimation. The advantages over classic steady-state methods are (i) the use of arbitrary boundary conditions, (ii) a less time-consuming experimental analysis, and (iii) the applicability to scales larger than typical for the laboratory. Techniques of nonlinear parameter estimation are, although well established in engineering in science, still a topic of basic and applied research. In particular, considerable effort is spent on improving the computational tools for solving inverse problems, e.g. by the adaptation and application of modern global optimization techniques like genetic algorithms (Holland, 2001), and modern hybrid optimization algorithms (Duan, 1992, Pan and Wu, 1998) to unsaturated flow problems. The inverse modelling approach currently receives even more attention, because the vadose zone scientific community is striving to bridge the gap between small and regional scale simulation studies, and the application to larger scales is an alternative to the bottom-up (upscaling) approach to derive effective properties (Harter and Hopmans, 2004; Vrugt et al., 2004; Durner et al., 2007). Determining SHP by inverse modelling relies on an adequate parameterization of the SHP. This enables the estimation of a set of model parameters that mathematically describe the SHP. In soil physics, a whole universe of functions with a relatively low number of parameters has been developed and tested in practical simulation studies including inverse modelling. However, not all functions are sufficiently flexible to yield adequate descriptions of the observed behaviour of unsaturated flow systems. Moreover, the selection of an appropriate parameterization of the soil water retention and hydraulic http://www.dfg-wasserkommission.de Back to contents 47 of 88 Young Scientists Workshop on Data Assimilation conductivity function remains a challenge (Vrugt, 2003), in particular because strong parameter correlations impede the interpretability of single estimated coefficients. My poster contribution will present an improved algorithm for estimating SHP without assigning an a priori shape to them. It is a further development of an idea presented by Bitterlich et al. (2004) who introduced the term “free-form” functions to the vadose zone community. The new algorithm uses cubic Hermite interpolation between nodal values of water content and unsaturated hydraulic conductivity and a global evolutionary optimization strategy. A multilevel routine identifies the adequate number of degrees of freedom by balancing model performance, the statistical interaction of the estimated model parameters (quantified by a collinearity index inferred from the scaled sensitivity matrix; Brun et al., 2001), and their number. A first-order uncertainty analysis provides a quantitative measure of how well the SHP can be identified in different ranges of pressure head. This offers great potential for designing optimal experimental procedures for identifying the hydraulic properties of porous media. I will demonstrate the effectiveness of the algorithm for the evaluation of classic multi-step outflow experiments by investigating synthetic data sets and real measurements. It will be shown that the freeform approach yields optimal model parameters that show only moderate correlation indicating wellposed inverse problems. Since parameterization errors are almost completely avoided, the algorithm is well suited to identify other error sources in unsaturated flow problems, e.g. limitations in the applicability of Richards equation or problems caused by spatial heterogeneity. The details of the algorithm are described in Iden and Durner (2007). The principal idea to estimate constitutive relationships by free-from functions is applicable to many problems in hydrological and hydrochemical research, like the determination of sorption isotherms (Knabner, 2005), arbitrary dose-response relationships, or rainfall-runoff modelling. The contribution will emphasize the need for (i) efficient global optimization strategies, (ii) uncertainty analysis, and (iii) identifiability analysis to avoid overparameterizations and inadequate model structures. References Bitterlich, S., W. Durner, S. C. Iden, and P. Knabner (2004), Vadose Zone J., 3, 971–981. Brun, R., P. Reichert, and H. R. Künsch (2001), Water Resour. Res., 865 37, 1015– 1030. Duan, Q., S. Sorooshian, and V. Gupta (1992), Water Resour. Res., 28, 1015– 1031. Durner, W., U. Jansen, and S. C. Iden (in press), European Journal of Soil Science Durner, W., and K. Lipsius (2005), in: Encyclopedia of Hydrological Sciences, pp. 1021–1144, John Wiley, Hoboken, N. J. Finsterle, S. (2004), Vadose Zone J., 3, 747– 762. Harter, T., and J. W. Hopmans (2004), in: Unsaturated Zone Modeling: Progress, Applications, and Challenges, Kluwer, Dordrecht, Netherlands. Holland, J. H. (2001), Adaptation in Natural and Artificial Systems, MIT Press, Cambridge, Mass. Hopmans, J. W., J. Simunek, N. Romano, and W. Durner (2002), in: Methods of Soil Analysis Part 4: Physical Methods, Soil Sci. Soc. Am. Book Ser., vol. 5, pp. 963–1004, Soil Sci. Soc. of Am., Madison, Wisconsin. Iden S. C., and W. Durner (2007): Water Resour. Res., 43(7) doi:10.1029/2006WR005845. Knabner, P., B. Igler, K. Totsche, and P. DuChateau (2005), Comput. Geosci., 9(4), 203–217, doi: 10.1007/s10596-005-9008-0. Pan, L., and L. Wu (1998), Water Resour. Res., 34, 2261– 2269. Simunek, J. (2005), in: Encyclopedia of Hydrological Sciences, pp. 1171–1180, John Wiley, Hoboken, N. J. van Dam, J. C., G. H. de Rooij, M. Heinen, and F. Stagnitti (2004), in: Unsaturated Zone Modeling: Progress, Applications, and Challenges, Kluwer, Dordrecht, Netherlands.. Vanclooster, M., J. Boesten, A. Tiktak, N. Jarvis, J. G. Kroes, R. Munoz-Carpena, B. E. Clothier, and S. R. Green (2004), in: Unsaturated Zone Modeling: Progress, Applications, and Challenges, Kluwer, Dordrecht, Netherlands. Vrugt, J. A., W. Bouten, H. V. Gupta, and J. W. Hopmans (2003), Vadose Zone J., 2, 98–113. Vrugt, J. A., G. Schoups, J. W. Hopmans, C. Young, W. W. Wallender, T. Harter, and W. Bouten (2004), Water Resour. Res., 40, W06503, doi:10.1029/2003WR002706. http://www.dfg-wasserkommission.de Back to contents 48 of 88 Young Scientists Workshop on Data Assimilation Dr. Sonja C. Jähnig Dr. Sonja C. Jähnig *1975 Aquatic Ecology Degree: Dipl. Environmental Sciences (Univ. Essen, 2001); Ph.D. Hydrobiology (Univ. Duisburg-Essen, 2007). Curr. Position: Postdoc, Department of Applied Zoology / Hydrobiology at the University of Duisburg-Essen. March 2008 – Feb (Aug) 2009: Postdoc, Chinese Academy of Sciences, Institute of Hydrobiology, Wuhan. Research interests: stream restoration, aquatic ecology, ecological modelling: effects of restoration on biota, stream benthic invertebrates Motivation - Discuss and obtain feedback on the planned research project "modeling stream benthic invertebrates in their field of dreams". - Exchange ideas and requirements, e.g. on data, with scientists from other disciplines. - Get a better idea about uncertainty and variability issues and various approaches to consider them in models. - Contact to other young scientists, start building a network. Current address: Sonja Jähnig, Universität Duisburg-Essen, FB Biologie & Geographie, Abteilung Angewandte Zoologie / Hydrobiologie, Raum: S05 T03 B12, 45117 Essen, Tel: 0201-183-4308, Fax: 0201-183-4442, Email: [email protected] Selected Publications related to the workshop: The question arose during my PhD-thesis: I have obtained a sound background on factors determining invertebrate communities: Jähnig SC, Lorenz A and Hering D. 2007. Hydromorphological parameters indicating differences between single- and multiple-channel mountain rivers in Germany, in relation to their modification and recovery. Aquatic Conservation: Marine and Freshwater Ecosystems: in press. Kail J, Jähnig SC and Hering D. 2007. Relation between floodplain land use and river hydromorphology on different spatial scales - a case study from two lower-mountain catchments in Germany. Fundamental & Applied Limnology: in press. Jähnig SC, Lorenz A and Hering D. 2007. Habitat mosaics and macroinvertebrates – does channel form determine community composition? Aquatic Conservation: Marine and Freshwater Ecosystems (submitted). Jähnig SC and Lorenz A. 2007. Substrate-specific macroinvertebrate diversity patterns following stream restoration. Aquatic Sciences – Research across Boundaries (submitted). Lorenz A, Jähnig SC, Schlachta O and Hering D. 2007. Remeandering German lowland streams – qualitative and quantitative success of restoration measures. Restoration Ecology (submitted). Abstract Modeling stream benthic invertebrates in their field of dreams Keywords: streams, benthic macroinvertebrates, aquatic habitat, scale The 'field of dreams hypothesis' means: 'if you build it, they will come' (Palmer et al. 1997). Consequently, to envisage effects on aquatic habitat conditions from management or restoration http://www.dfg-wasserkommission.de Back to contents 49 of 88 Young Scientists Workshop on Data Assimilation measures and predict the changes in occurrence of certain taxa or community composition of stream benthic macroinvertebrates is an intriguing yet challenging and interdisciplinary task. It is a step forward from recent research, which attempted to model the response of stream ecosystem characteristics to land use changes (Johnson et al. 2007). A model to employ for this purpose is HABITAT (WL|Delft 2003). HABITAT is a GIS-based application, which models the availability and quality of habitats for individual species. Such modeling might be used for prediction of suitability of habitat to flora or fauna in response to changes in environmental factors. Stream habitat parameters influencing benthic macroinvertebrate communities have comprehensively been described. Profound knowledge is available for the various scales, as well as intensity of parameters (Hering et al. 2006). However, the modeling of macroinvertebrate communities (or certain taxa for a start) according to the available habitats within a stream section has not been attempted, due to great complexity and interaction of parameters. The description of aquatic habitat involves the consideration of hydraulic and hydrologic information at site, section and catchment scale (Molnar et al. 2002). Relevant habitat parameters for benthic macroinvertebrates include current velocity, shear stress, water depth, substrate, sediment stability, shading, and physico-chemical parameters (Oxygen, pH, Conductivity, Temperature). On a larger scale the hydromorphological situation of a section, longitudinal, lateral and vertical continuity as well as catchment parameters (sea level, geology, pressures) are relevant (compare Vinson and Hawkins 1998). For organisms relevant information is available from the comprehensive freshwaterecology-database (Euro-Limpacs 2007), which includes autecological characteristics of more than 12.000 European freshwater organisms belonging to macroinvertebrates, fish, diatoms and macrophytes. Particularly for macroinvertebrates information on habitat preference (current preference, substrate), saprobic preferences and life parameters (feeding type, locomotion type, etc.) are available. These preferences need to be transferred into rules processable for HABITAT. Prospects, possibilities, difficulties and interfaces to other aquatic organism groups will be discussed in the workshop contribution. To build such a complex model the expertise of hydraulic and hydrologic scientists is required for implementing the relevant hydraulic models from catchment to habitat level, modeling aquatic habitat and changes of it due to management. Implementing rules for presence / absence of taxa requires hydrobiologists – eventually for all aquatic organism groups such as macroinvertebrates, fish, phytobenthos and macrophytes. Ultimately the effect of management or restoration measures on the habitat configuration of stream sections and finally on biota might be predicted. References. Euro-Limpacs. 2007. Freshwaterecology.info - The Taxa and Autecology Database for Freshwater Organisms. Available from www.freshwaterecology.info (Version 3.1, July 2007). Hering D, Johnson RK, Kramm S, Schmutz S, Szoszkiewicz K and Verdonschot PFM. 2006. Assessment of European streams with diatoms, macrophytes,macroinvertebrates and fish: a comparative metric-based analysis of organism response to stress. Freshwater Biology 51: 1757–1785. Johnson TE, McNair JN, Srivastava P and Hart DD. 2007. Stream ecosystem responses to spatially variable land cover: an empirically based model for developing riparian restoration strategies. Freshwater Biology 52: 680-695. Molnar P, Burlando P and Ruf W. 2002. Integrated catchment assessment of riverine landscape dynamics. Aquatic Sciences 64: 129-140. Palmer MA, Ambrose RF and Poff LN. 1997. Ecological theory and community restoration ecology. Restoration Ecology 5: 291–300. Vinson MR and Hawkins CP. 1998. Biodiversity of stream insects: Variation at local, basin, and regional scales. Annual Review of Entomology 43: 271–293. WL|Delft. 2003. HABITAT – a spatial analysis tool for ecological assessments, Directorate-General of Public Works and Water Management, Copyright © RWS, WL|Delft Hydraulics 2003. http://www.dfg-wasserkommission.de Back to contents 50 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 51 of 88 Young Scientists Workshop on Data Assimilation Dr. David Kneis David Kneis *1977 Degree: Dipl. Geoecology (Univ. Geoecology (Univ. Potsdam 2007). Aquatic Ecology Potsdam, 2002); Ph.D., Curr. Position: Research associate in a project on operational flood forecasting (Saxon flood forecasting center / University of Potsdam) Research interests: Development AND use of models for applications in hydrology and aquatic ecology, GIS, data collection and processing methods, further interest: renewable energies Motivation I attended the YSW for several reasons: - I wanted to learn new methods of data integration/assimilation and uncertainty analysis from other disciplines. - I was interested in the opinions/visions of the invited senior scientists concerning the perspectives of modeling in their fields of specialization. - The workshop was a great opportunity to get in contact and discuss with young scientists from similar and other disciplines. - I was interested in the objectives of the KoWa with respect to future water-related research and funding opportunities offered by the DFG. Current address: Sächsisches Landesamt für Umwelt und Geologie Landeshochwasserzentrum, Postfach 80 01 32, 01101 Dresden Tel. +49 351 89 28 334, Email: [email protected] Selected Publications related to the workshop: Kneis, D. (2007): A water quality model for shallow river-lake systems and its application in river basin management; PhD thesis, University of Potsdam; http://nbnresolving.de/urn:nbn:de:kobv:517-opus-14647 Kneis, D., Knösche, R. & Bronstert, A. (2006): Analysis and simulation of nutrient retention and management for a lowland river-lake system; Hydrology and Earth System Sciences, 10, 575-588 Bronstert, A. & Itzerott, S. (Eds.) (2006): Bewirtschaftungsmöglichkeiten im Einzugsgebiet der Havel, Brandenburgische Umweltberichte 18 Förster, S., Kneis, D., Gocht, M., Bronstert, A. (2005): Flood Risk Reduction by the Use of Retention Areas at the Elbe River, Intl. J. River Basin Management Vol. 3, No. 1, pp. 21-29., http://www.jrbm.net/pages/home/vol3.asp Abstract A flexible water quality model for shallow river-lake systems Keywords: phosphorus, nitrogen, shallow lakes, model development, river basin management In my recently finished PhD thesis (Kneis, 2006 & 2007) I developed a model for simulating mass http://www.dfg-wasserkommission.de Back to contents 52 of 88 Young Scientists Workshop on Data Assimilation transport and turnover in rivers and shallow lakes (TRAM). The simulation tool is intended to complement eco-hydrological catchment models in studies on river basin management. Model development and application were embedded in a research project focusing on options to ease surface water eutrophication in the Havel River Basin (Bronstert & Itzerott, 2006). In this study, TRAM was used for simulating nutrient concentrations in a large-scale river-lake system exhibiting significant internal nitrogen and phosphorus turnover. For conducting scenario analyses with altered external loading, TRAM was coupled with two catchment models specialized on the estimation of nutrient emissions from point and non-point sources. Simulated in-river concentrations of total phosphorus were used for assessing alternative management options in the light of the European Water Framework Directive. The design of the newly developed model was guided by a review of existing water quality models, recommendations published by BfG (2002), personal modeling experience, as well as constraints arising from the targeted spatial/temporal scale of application. A summary of some of the model’s key features is given below. (1) TRAM relies on a strongly generalized transport scheme based on a discretization of the natural system into fully mixed and purely advective elements. This approach proved to be suitable for large-scale applications. Required boundary conditions such as flow rates and storage volumes are supplied by external hydrodynamics or hydrological models. (2) A key feature of TRAM is its flexibility with respect to turnover simulations. The number and kinds of simulated components, reaction rates, as well as the stoichiometry matrix (Reichert et al., 2001) are not predefined but user-supplied. Due to this ‘open-structure approach’ the model’s complexity is adjustable to the present knowledge of processes, the amount and quality of data for parametrization and verification, and the available time budget. This approach is considered essential for efficient systems research since the model structure can be continuously refined by evaluating residuals of simpler approaches and new data can be collected more specifically. (3) Finally, TRAM aims at being efficient with respect to both setup and simulation. For example, the capabilities of geographic information systems are used for model parametrization. The userdefined description of turnover is integrated with the model's generic kernel by automatic code generation. The software was designed for efficient analyses of parameter sensitivities and uncertainties. The main result of the model application to the Havel River is that the targeted "good status" with respect to phosphorus concentrations seems to be achievable in spite of continued P release from sediments if the full potential of emission control was tapped. Although this is a clear statement, the simulation study revealed a number of severe limitations of the model's predictive capabilities. For example, the chosen turnover model neither includes a mechanistic description of the seasonal succession of the plankton nor a process-oriented model of sediment diagenesis. However, in general, these are prerequisites for modeling the impact of altered boundary conditions on aquatic nutrient cycling. Due to the large number of biotic interactions and abiotic control factors, the required fully mechanistic shallow-lake ecosystems model is not (and might never become) available. Consequently, it is believed that research should further focus on the identification of the "appropriate" level of model complexity that balances structural and parameter uncertainties. In this light, the use of open-structure simulation tools is recommended as they facilitate "playing" with alternative process descriptions. Uncertainty analyses might become more convincing by using such alternative model formulations in parallel – instead of "believing" in a single structure and focusing on the investigation of unknown parameter values. Bibliography BfG (2002): Mathematisch-numerische Modelle in der Wasserwirtschaft, Handlungsempfehlung für Forschungs- und Entwicklungsarbeiten, Bundesanstalt für Gewässerkunde, BfG-Mitteilungen 24. Bronstert, A. & Itzerott, S. (Eds.) (2006): Bewirtschaftungsmöglichkeiten im Einzugsgebiet der Havel, Brandenburgische Umweltberichte 18 http://www.dfg-wasserkommission.de Back to contents 53 of 88 Young Scientists Workshop on Data Assimilation Kneis, D. (2007): A water quality model for shallow river-lake systems and its application in river basin management; PhD thesis, University of Potsdam; http://nbn-resolving.de/urn:nbn:de:kobv:517-opus-14647 Kneis, D., Knösche, R. & Bronstert, A. (2006): Analysis and simulation of nutrient retention and management for a lowland river-lake system; Hydrology and Earth System Sciences, 10, 575-588 Reichert, P., Borchardt, D., Henze, M., Rauch, W., Shanahan, P., Somlyody, L. & Vanrolleghem, P. A. (2001): River water quality model No. 1, IWA Publ. http://www.dfg-wasserkommission.de Back to contents 54 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 55 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Ing. Stefan Krämer Dipl.-Ing. Stefan Krämer *1973 Meteorology Degree: Dipl. Ing., Civil Engineering (Leibniz University of Hanover, 2000). Ph.D. Thesis, submitted to the Faculty of Civil Engineering and Geodesy Leibniz University of Hanover (“Quantitative Radardatenaufbereitung für die Niederschlagsvorhersage und die Siedlungsentwässerung”) Curr. Position: Research Scientist, Institute of Resources Management, Leibniz University of Hanover Water Research interests: Radar meteorology, nowcasting and urban drainage Motivation My field of activity has concentrated on the measurement of radar rainfall as input for urban drainage models. This has involved the development of algorithms for quantitative radar data processing in real time (attenuation correction, R-Z relation) as well as the integration of rainfall measurements from different sources (radar, microwave link, rain gauges, distrometer) for hydrological modelling under consideration of their uncertainties. One benefit of radar rainfall data is the possibility for extrapolation of rainfall textures in order to provide runoff and flow forecasts. Future ideas deal with the development of an integrated rainfall model for urban water management. For the combined control of sewer systems, waste water treatment plants and receiving waters different timescales and forecasting horizons have to be considered as well as various sources of rainfall information concerning their spatial resolution and their reliability bounds. Current address: Institut für Wasserwirtschaft, Hydrologie und landwirtschaftlichen Wasserbau, Leibniz Universität Hannover, Appelstr. 9a, D30167 Hannover Tel. +49 511 762 5195, Email: [email protected] Selected Publications related to the workshop: Krämer, S., M. Grum, H.-R. Verworn and A. Redder (2005): Runoff modelling using radar data and flow measurements in a stochastic state space approach, Wat. Sci. Tech., 52 (5), 1 – 8. Krämer, S., H.-R. Verworn and A. Redder (2005): Improvement of X-band radar rainfall estimates using a microwave link. Atmos. Res. 77 (1-4), 278 – 299. Grum; M., S. Krämer, H.-R. Verworn and A. Redder (2005): Combined use of point rain gauges, radar, microwave links and level measurements in urban hydrological modelling. Atmos. Res. 77 (1-4), 313 – 321. Verworn H.-R. and S. Krämer (2005): Aspects and effectiveness of real time control in urban drainage systems combining radar rainfall forecasts, linear optimization and hydrodynamic modelling. Proc. 8th Int. Conf. on Computing and Control for the Water Industry, 5–7 September 2005, University of Exeter, UK, 307 – 312. Rahimi, A.R., A.R. Holt, G.J. Upton, S. Krämer, A. Redder and H.-R. Verworn (2006): Attenuation calibration of an X-band weather radar using a microwave link. J. Atmos. Oceanic Tech., 23 (3) 395 – 405. Krämer, S., H.-R. Verworn, A.Hartung, A.R. Holt, G.J.G Upton and M. Becker (2006): Uncertainty quantification of operational X-band weather rainfall measurements. Proc. 7th Int. Workshop on Precipitation in Urban Areas, 7-10 December, 2006, St. Moritz, Switzerland, 87 - 92. Krämer, S., L. Fuchs and H.-R. Verworn (2007): Aspects of radar rainfall forecasts and their effectiveness for real time control - the example of the sewer system of the city of Vienna. Wat. Pract. Tech. 2 (2), doi10.2166/wpt.2007.042. http://www.dfg-wasserkommission.de Back to contents 56 of 88 Young Scientists Workshop on Data Assimilation Abstract Integration of microwave link attenuation information to correct X-band weather radar reflectivity profiles Keywords: radar, rainfall, microwave link, attenuation correction Rainfall results from complex atmospheric processes and is the driving force for hydrology. Its distribution in space and time is important for meteorological and hydrological applications, (e.g. DWD-LMK, flash flood predictions, real time control of urban drainage systems). Weather radars provide this information highly resolved in space and time. The reliability of quantitative radar rainfall estimation, however, is limited. This is mainly due to the effect of attenuation the radar signal experiences when it passes a cloud of hydrometeors. As a result the radar measured reflectivity profile Za(r) underestimates the true (unknown) reflectivity profile Z(r) with increasing distance (r) from the radar. This error propagates when radar rainfall R is derived using any R-Z relationship. The attenuation phenomenon is severe for X-band, but it is also acute at C-band. Hitschfeld and Bordan (1954) addressed this problem and proposed an analytical solution for the attenuation problem which is often used in a modified version (Delrieu et al. 1997). The specific attenuation k(r), the radar signal experiences, is derived from measured rainfall reflectivity Za(r). Between specific attenuation k and reflectivity Z a power law k = a × Zb is assumed. For correction the choice of the coefficients (a, b) is crucial. They have to be estimated before the correction is made. The common approach is to model attenuation (k) and backscattering (Z) effects on single raindrops for different drop sizes, shapes and temperatures depending on the radar frequency. Then, either modelled drop size distributions (DSD) or historic disdrometer measurements are the basis for integration of the scattering characteristics for different rainfall types. This results in averaged coefficients a,b which are uniformly / constantly used for correction. This methodology, however, does not work for the current rainfall and its characteristics as measured by the radar. As a consequence the Hitschfeld Bordan (HB) algorithm shows the well known instability. Therefore attenuation correction is not applied to operational weather radar systems and the “true” rainfall is continuously underestimated. To investigate this problem a dual frequency microwave link was installed. The receivers were collocated with an X-band weather radar antenna operated by Emschergenossenschaft / Lippeverband near the City of Essen. With one frequency (10.5 GHz) operating close to the radar frequency (9.47 GHz), the link was aimed to provide an attenuation reference (Aref) parallel to one radar beam over a path length of 29.6 km (119 radar gates, length 0.25 km). The resulting attenuation time series is shown in the figure left for the convective rainfall event on June 8, 2003. The parameter space of attenuation coefficients (a,b) has been explored by calculating the radar attenuation (Aradar) with the HB-approach. An optimal combination of a,b coefficients has been accepted for each minute when the difference between radar and reference attenuation dmin = Aref - Aradar(a,b) was found in 0.25 dB bounds (|dmin| ≤ 0.25 dB). The results (figure right) demonstrate a high variability of the a,b coefficients. For each minute many a,b combinations are found which satisfy the objective criterion. They concentrate along “minute-lines”. The characteristics of the minute-lines show a systematic behaviour depending on the rainfall (convective / stratiform) along the radar beam. Attenuation coefficients found in literature and derived from event specific DSD measurements using the Mie theory are plotted for comparison. 15 events have been processed in total which confirm the results. http://www.dfg-wasserkommission.de Back to contents 57 of 88 Young Scientists Workshop on Data Assimilation The high variability of the results demonstrates that the constant and uniform use of attenuation coefficients is non realistic and the main reason for instability of the HB-algorithm. The use of variable a,b coefficients, in contrast, enables attenuation corrections up to 60 dB. In addition, the systematic behaviour of coefficients defines an a priori knowledge to develop an operational attenuation correction methodology. The attenuation correction improves significantly the quantitative radar rainfall estimation (X and C-Band) and has a strong benefit to meteorological and hydrological application of radar data. Delrieu G., Caoudal S. and J.D. Creutin (1997): Feasibility of using mountain returns for the correction of ground based X-band weather radar data. J. Atmos. Oceanic Tech., Vol. 14, 368 – 385. Hitschfeld W. and J. Bordan (1954) Errors inherent in the radar measurement of rainfall at attenuating wavelength. J. Meteorol., Vol. 11, 58 – 67. http://www.dfg-wasserkommission.de Back to contents 58 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 59 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Geoökol. Tobias Krueger Dipl.-Geooekol. Tobias Krueger *1978 Hydrology Degree: Dipl. Geoecology (Tech. Univ. Braunschweig, 2005). Curr. Position: PhD student in Environmental Science, Lancaster University, UK; working on DEFRA funded project PE01020 Phosphorus mobilisation with sediment and colloids through drained and un-drained grasslands in cooperation with IGER, University of Exeter, CEH and University of Plymouth. Research interests: Modelling of water, sediment and nutrient transport at plot to catchment scale; nutrient load estimation techniques; analysis of model parameter and structural uncertainty; quantification of data uncertainty and their incorporation into models. Motivation: In my role as a research student in an interdisciplinary project on phosphorus transfer processes, I am required to model the transfer of phosphorus from agricultural land to receiving waters. Our research has shown that this inevitably needs to include model evaluation within an uncertainty estimation framework. In water quality modelling, the data availability with respect to process complexity will always be more limited than, for example, in hydrological model applications. I am, therefore, actively researching the interplay between data availability and model complexity as well as the quantification of measurement uncertainties .I am keen to exchange ideas with related disciplines which I believe face similar issues. An interdisciplinary discussion of model development, evaluation and uncertainty estimation has great potential for identifying techniques that could be suitable for a given application – especially as I think there is not one technique which is suitable for all applications. As a contribution to the discussion, I can offer the perspective of informal uncertainty estimation techniques along the lines of the Generalised Likelihood Uncertainty Estimation (GLUE). Current address: Lancaster Environment Centre, Department of Environmental Science, Lancaster University, Lancaster, LA1 4YQ, UK Tel. +44 (0)1524 593534, Email: [email protected] Selected Publications related to the workshop: Krueger T., Freer J., Quinton J. N. and C. J. A. Macleod (2007). Processes affecting transfer of sediment and colloids, with associated phosphorus, from intensively farmed grasslands: A critical note on modelling of phosphorus transfers. Hydrological Processes 21(4): 557-562. Abstract http://www.dfg-wasserkommission.de Back to contents 60 of 88 Young Scientists Workshop on Data Assimilation Data availability, uncertainty and process identification in water quality modelling: the case of agricultural phosphorus transfers T. Krueger, J. Freer, J.N. Quinton Keywords: data limitations, scale, process identification, uncertainty analysis In this poster we will give an overview of what we think are the key issues in water quality modelling. The focus will be placed on modelling transfers of phosphorus from agricultural land to receiving waters (see Krueger et al. 2007), but the arguments can easily be transferred to other water quality modelling applications were we face similar data limitations, uncertainties and gaps in process understanding. Our points are: Water quality modelling in practice is restricted by limitations on the amount and scope of data that can be collected in the field to parameterise and drive models. Therefore, the scale of application and the availability of field observations should direct model development for a given study – different scales may require different models. How can robust process descriptions be inferred from incomplete and imperfect field observations? A pragmatic strategy seems to be aiming at the dominant modes of system behaviour, where these can be supported by field observations – some model compartments may be easier to identify than others and for some we might neither have the data nor conceptual knowledge to say anything. Ideally, models should be evaluated by assessing the potential uncertainties in model structure, parameters and data. However, making assumptions about the nature of these uncertainty components may neither be justified nor testable given the available field observations – complexity in uncertainty analysis needs to be balanced against the information content of the data and the level of detail required by the modelling application. To illustrate our arguments, we will present data analyses, uncertainty analyses and modelling work undertaken at the plot, catchment and large catchment scale within the interdisciplinary Grassland sediment and colloid Phosphorus (GrasP) project1 funded by the UK Department for Environment, Food and Rural Affairs as well as in collaboration with Swiss colleagues2. Acknowledgements 1 GrasP are P.M. Haygarth, G.S. Bilotta, R. Bol, P.J. Butler, S.J. Granger, J. Hawkins, C.J.A. Macleod (Institute of Grassland and Environmental Research, North Wyke, UK), R.E. Brazier (University of Exeter, UK), L.J. Gimbert, P. Worsfold (University of Plymouth, UK), P.S. Naden, G. Old (Centre for Ecology and Hydrology, Wallingford, UK), J. Freer, T. Krueger, J.N. Quinton (Lancaster University, UK). 2 M. Schaerer (Federal Office for the Environment, Bern, Switzerland). References Krueger T., Freer J., Quinton J.N., Macleod C.J.A., 2007. Processes affecting transfer of sediment and colloids, with associated phosphorus, from intensively farmed grasslands: A critical note on modelling of phosphorus transfers. Hydrological Processes 21(4), 557-562. http://www.dfg-wasserkommission.de Back to contents 61 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 62 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Geol. Christine Kübeck Dipl.-Geol. Chistine Kübeck Degree: Dipl. Geology (Univ. TU-Clausthal, 2005) Hydrogeology Curr. Position: scientific co-worker, TU-Clausthal Research interests: functional ecology of aquatic ecosystems, ecological modelling, river basin management Motivation: non-uniqueness of model results caused by uncertainties in data assimilation; principle of parsimony; simplification of natural systems as ill-defined process; uncertainties of modelling in general and its influence on model results; differences between application fields Current address: Inst. Geology and Palaeontology, TU-Clausthal, Leibnizstr. 10, 38678, Clausthal-Zellerfeld (Germany) Selected Publications related to the workshop: Kübeck, Ch. & Steding, T. (2006): PhreeqC Modellierung der Laborversuche zur „Bestimmung der Basenkapazität KB 8.2 “ - Ringversuche zur Basisvalidierung und Überarbeitung der Norm DIN 39709 7 (H7) vom Mai 1979. Clausthaler Geowissenschaften Bd. 5 S.: 115-134. Abstract The principle of simplicity in hydrogeochemical modelling Keywords: simplicity, transparency, refutability, groundwater modelling Looking at nature we find very complex systems influenced by and dependent on a huge number of parameters. But we will also find clear structures and processes that can be used for understanding often not the whole system but the main character of nature. Though, finding these structures or processes is troublesome. In this context the development of computer models has been both, boon and bane. In the first place, computer models are a great benefit in solving mathematic mo-dels with high complexity. At the same time they are suggestive of being confident in its results. But the uniqueness of modelling results is often not given due to the lack of system information. Thus increasing the plausibility of modelling results requires the implementation of transparency and refutability into modelling. Hill (2006) discusses the necessity of developing models that are both, refutable by testing its assumptions, as well as transparent by means of understanding its dynamical processes. Carrying on Hill´s fundamental idea of model transparency, should also include technical data of modelling like numerical method, boundary conditions or discretization. Most of the modellers underestimate the influence on model results by using different kinds of mathematic approaches. However, the purpose of modelling is to understand the structures and processes that form the natural system by analysing them, using less complex models. As a consequence, transparency is a major objective of modelling. Described by Oreskes (2000) paradoxically transparency and refutability suffer with increasing complexity of a model. On that condition, modelling is a balancing of keeping the model as simple as possible and integrating all characteristic and http://www.dfg-wasserkommission.de Back to contents 63 of 88 Young Scientists Workshop on Data Assimilation dominating processes. For example, the strategy of modelling hydrogeochemical groundwater systems developed by Hansen and van Berk (2004) implies a high grade of transparency, although the model attains a relatively huge number of hydrogeochemical pro-cesses. By means of starting with a very simple model and stepwise integrating more interacting processes, the model maintains comprehensible. Although the 1D material flux model reproduces the raw water quality with a high conformity, it is not meant to predict future system conditions. This is not generally a matter of non uniqueness, but also a question of oversimplification. Oreskes and Belitz (2001) describe the increase of the likeliness of non-uniqueness with the complexity of a model. But at the same time the validity of the model forecast decreases with increasing level of simplification. Reflecting on that paradox, the determination of the appropriate level of model complexity gains more importance. However, the required determination process is ill-defined, so the level of model complexity depends on the experience of the modeller, the task formulation and the kind of natural system. For instance, the described 1D material flux model implies merely a primitive temporal and spatial discretization. In order to model forecasts, this kind of discretization is non-sufficient, hence a further development has to be implemented. To maintain transparency, the model refinement must comply with the principle of simplification. Transferring the hydrogeochemical model into the three-dimensional space or coupling with the geohydraulical model implies a non-reasonable increase of complexity and modelling uncertainties. Therefore an uncommon basic approach is developed to abstract the three-dimensional space of the groundwater system as quasi-one-dimensional image by using stream tubes. Each stream tube represents a specific area within the catchment area and a defined volume flow. Abstracting the transport of groundwater within the stream tube as a covered path length, the stream tube can be transferred into the 1D space. According to the task formulation further simplification of the temporal and spatial discretization can be implemented by using zones of flow times or generating a representative square section. Keeping in mind, every model is just an abstract image of the complex nature that can be wisely used as an instrument to formulate questions and interpret signals but should not be misused as a verification code of nature. References Hansen C. and van Berk W., Retracing the development of raw water quality in water works applying reactive controlled material flux analyses., Aqu. Sci., Vol. 66, 2004. Hill M. C., The Practical Use of Simplicity in Developing Ground Water Models., Ground Water, Vol. 44, No. 6, pp. 775-781, 2006. Oreskes N., Why believe a computer? Models, measures and meaning in natural world. in: Schneidermann J. S., The Earth Around., pp. 70-82, Freeman & Co., 2000 Oreskes N. and Belitz K., Philosophical Issues in Model Assessment. in: Anderson M. G. and Bates P. D., Model Validation: Perspectives in Hydrological Science., Wiley, 2001. http://www.dfg-wasserkommission.de Back to contents 64 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 65 of 88 Young Scientists Workshop on Data Assimilation Dr. Eva Nora Müller Dr. Eva Nora Müller (1975) Degree: PhD in Hydrology/Geomorphology (2004), MSc in Nonlinear Dynamics (2001), Dipl.-Ing. (2000) Hydrology Curr. Position: Post-doc position within the SESAM-Project at the Institute of Geoecology, University of Potsdam Research interests: Modelling of catchment hydrology, sediment- and nutrient transport processes, dryland environments, overland flow dynamics and soil erosion in arid and semiarid environments, desertification and land degradation, vegetation-change dynamics (ecohydrology) Motivation Current address: Institut für Geoökologie, Universit Potsdam, Karl-Liebknecht-Str. 24-25, 14476 Potsdam, Tel. +49 (0)331-977-2975, Email: [email protected] Selected Publications related to the workshop: E. N. Mueller, J. Wainwright, A. J. Parsons (2007), The impact of connectivity on the modelling of overland flow within semi-arid shrubland environments, Water Resources Research 43, W09412, doi:10.1029/2006WR005006. E. N. Mueller, J. Wainwright, A. J. Parsons (2007) The stability of vegetation boundaries and the propagation of desertification in the American Southwest: A modelling approach, Ecological Modelling 208, 91-101 E. N. Mueller, J. Wainwright, A. J. Parsons (in press), Spatial variability of soil and nutrient characteristics of semi-arid grasslands and shrublands, Jornada Basin, New Mexico, Ecohydrology E. N. Mueller, R. J. Batalla, C. Garcia, A. Bronstert (in revision) Modelling bedload transport during small floods in a gravel-bed river, Journal of Hydraulic Engineering E. N. Müller, R. J. Batalla, A. Bronstert. (2006) Dryland river modelling of water and sediment fluxes using a representative river stretch approach. Book chapter IN: Natural Systems and Global Change, German-Polish Seminar Turew, Poznan T. Francke, A. Güntner, A. Bronstert, G. Mamede, E. N. Müller. (2007, in press) Automated Catena-based Discretisation of Landscapes for Semi-distributed Hydrological Modelling. International Journal of Geographical Information Science A. Bronstert, R. J. Batalla, J. C. de Araújo, T. Francke, A. Güntner, G. Mamede, E. N. Müller (2007) Investigating erosion and sediment transport from head-waters to catchments to reduce reservoir siltation in drylands. In: Reducing the Vulnerability of Societies to Water Related Risks at the Basin Scale (Proceedings of the third International Symposium on Integrated Water Resources Management, Bochum, Germany, September 2006). IAHS Publ. 317, Wallingford (GB) G. L. Mamede, A. Bronstert, T. Francke, E. N. Müller, J. C. de Araújo, R. J. Batalla, A. Güntner. (2006) 1D Process-based modelling of reservoir sedimentation: a case study for the Barasona Reservoir in Spain. Conference Proceeding River Flow 2006. International Conference on Fluvial Hydraulics. Lisbon Sept. 06 Abstract The importance of vegetation dynamics in hydrological and sediment-transport modelling from the macro to the meso-scale: the missing link Keywords: vegetation dynamics; erosion; hydrological modelling Vegetation dynamics plays a vital role for the redistribution processes of water and sediment http://www.dfg-wasserkommission.de Back to contents 66 of 88 Young Scientists Workshop on Data Assimilation resources. However, many current modelling approaches for water, sediment and nutrient transport neglect the reproduction of vegetation dynamics. This is the case, for example, with inter-storm dynamics and the functional relationship between key model parameters such as friction factor, infiltration rate, evaporation and vegetation cover. This poster reviews the importance of vegetation dynamics at different spatial scales, investigates its implementation in current hydrological and erosion models and discusses the uncertainties related to the simplified implementation of ecohydrological processes. Two case studies are presented that signify the importance of the representation of vegetation dynamics in hydrological and erosion models at the micro- and mesoscale. The first case study deals with degradation processes and severe vegetation changes through shrub propagation into grassland in a semi-arid desert ecosystem in the south western part of the US. Spatially distributed, process-based models were used to investigate the redistribution of water and soil resources via overland flow within different grasslands and shrublands. Analysis of a multi-scale field data set showed that shrublands are characterised by lower friction factors, infiltration rates and nutrient content in the interconnected intershrub areas in comparison with higher values within shrub patches. Grasslands exhibit a more homogeneous distribution of flow and soil resource parameters with significantly higher friction factors and nutrient contents than shrubland (Mueller et al. 2007a). Transport models were parameterised for 0.1 km2 catchments with a grid size of 2 x 2 metres by using a connectivity approach that intrinsically represented the pattern formations of model input parameters such as friction factor, infiltration rate and nutrient content related to the patchiness of vegetation (Mueller et al. 2007b). Testing the model results showed that the incorporation of connectivity patterns into model parameterisation is crucial for model performance. The modelling study led to the formulation of a novel hypothesis regarding the ongoing degradation of grasslands (Mueller et al. 2007c): it is hypothesised that a vegetation boundary is stable when the following two conditions are met to balance the lower resistance of grassland towards the existing environmental setting with the higher resistance of shrubland. First, the soil depletion of nutrients by the action of overland flow in the grassland zone close to the boundary is in balance with the replenishment rates of grassland by nutrient cycling. Second, the grassland gains enough water resources from the upslope shrublands. On the contrary, a vegetation boundary has a potential for instability when the grassland acquires a competitive disadvantage towards shrubland regarding water benefit and nutrient depletion due to the combined effects of overland flow dynamics and some external stresses. With reference to the ecosystem stability and resilience theory, the modelling results provided important insights into the potential stability of the grassland-shrubland boundaries as a function of soil-nutrient depletion and water-resource enrichment for the grassland. To test this stability hypothesis, the implementation of vegetation dynamics in the current modelling framework requires substantial development to include interstorm dynamics of nutrient cycling and dynamic adjustment of hydrological and soil parameters as a function of shrub propagation. The second case study demonstrates and discusses the spatial implementation of vegetation distribution for a meso-scale erosion model. This research is part of the SESAM project (Sediment Export from Semi-Arid Regions: Measurement and Modelling), a DFG-funded project of an international research consortium dealing with sediment transport causing severe sedimentation of reservoirs in Spain and Brazil. Modelling of water and sediment transport processes are carried out for domains of 1,200 to 2,000 km2 with a spatial representation of sub-catchments with sizes ranging between 5 and 50 km2 that contain homogeneous landscape units and terrain components representing individual, characteristic hillslopes. Vegetation parameters are described for each terrain component that take into account height, root suction, LAI and stomata resistance and vegetationdependent erosion parameters. The poster presents simulation results for water and sediment transport from a region with different vegetation covers, ranging from semi-arid shrublands such as Caatinga in northeast Brazil to coniferous forest and agricultural fields in the Mediterranean pre-Pyrenees. The deficiencies of the current vegetation implementation include the insufficient temporal variation of vegetation characteristics such as seasonal changes of http://www.dfg-wasserkommission.de Back to contents 67 of 88 Young Scientists Workshop on Data Assimilation the parameters and dynamic linkages between land-use change and flow and erosion parameters. Finally, the poster discusses and develops ideas of how to include ecohydrological processes at the micro- and meso-scale in process based, spatially distributed modelling frameworks to increase the predictive ability of models in water research. References: E. N. Mueller, J. Wainwright, A. J. Parsons (in press 2007a), The impact of connectivity on the modelling of overland flow within semi-arid shrubland environments, Water Resources Research E. N. Mueller, J. Wainwright, A. J. Parsons (in press 2007b) The stability of vegetation boundaries and the propagation of desertification in the American Southwest: A modelling approach, Ecological Modelling E. N. Mueller, J. Wainwright, A. J. Parsons (in press 2007c), Spatial variability of soil and nutrient characteristics of semi-arid grasslands and shrublands, Jornada Basin, New Mexico, Ecohydrology http://www.dfg-wasserkommission.de Back to contents 68 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 69 of 88 Young Scientists Workshop on Data Assimilation Dr. Lars Nerger Aquatic Ecology Current affiliation: Alfred Wegener Institute for Polar and Marine Research Am Handelshafen 12, 27570 Bremerhaven, Germany phone: +49 (471) 4831 1558 E-mail: [email protected] CV 2000 M.S. in Physics, Albert Einstein Institute, Golm and University of Bremen, Germany Thesis title: “Investigations of 3D Binary Black Hole Systems” Advisor: Dr. Bernd Brügmann 2004 Ph.D. in Applied Mathematics, Alfred Wegener Institute, Bremerhaven and University of Bremen, Germany Thesis title: “Parallel Filter Algorithms for Data Assimilation in Oceanography” Advisors: Prof. Dr. Wolfgang Hiller and Dr. Jens Schröter Abstract Assimilation of Satellite Ocean Chlorophyll Data with Bias Correction for Biogeochemical State Estimation Lars Nerger1 and Watson W. Gregg2 1) Alfred Wegener Institute for Polar and Marine Research, Bremerhaven, Germany 2) Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA ([email protected]) Key words: Data assimilation, ecosystem modeling, Kalman filter, SEIK, bias correction, ocean color Chlorophyll data from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) is assimilated into the three-dimensional global NASA Ocean Biogeochemical Model (NOBM) for the period 1998-2004. The ensemble-based SEIK filter is used here with a localized analysis and simplified by the use of a constant covariance matrix. A multivariate configuration is applied, which updates the four phytoplankton groups of the model as well as nitrate, ammonium, and nitrogen detritus. The SEIK filter is combined with an online bias estimation algorithm, which estimates the bias in the model estimate of surface chlorophyll. The results of the daily assimilation are validated by comparison with independent in situ data. The assimilation results in significant improvements of the chlorophyll estimates. These are superior to both the free-run model and SeaWiFS data. The bias correction scheme shows that there are significant model biases in the chlorophyll field. These can be estimate and improve the assimilation compared to a bias-blind scheme. The results are less clear for the nutrients. Without bias correction, the prediction of nitrate is improved in wide regions. However, there are locations in which the assimilation is unstable, resulting in strong overestimates of nitrate. Here, the bias correction stabilizes the assimilation process, but results in smaller improvements of the non-observed fields. http://www.dfg-wasserkommission.de Back to contents 70 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 71 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Ing. Benjamin Nitsch Hydrology Current affiliation: Institute of Waste Management and Contaminated Sites TU Dresden Pratzschwitzer Straße 15, 01796 Pirna, Germany phone: 03501-530039 E-mail: [email protected] CV 2000 – 2006 2004 – 2006 degreed engineer for water management at TU Dresden Dresden Groundwater Research Centre (DGFZ e.V.), second mandatory internship and elaboration of diploma thesis Abstract Using synthetic precipitation time series for transient flow model-ing in the unsaturated zone Benjamin Nitsch & Oliver Kemmesies & Peter-Wolfgang Graeber 1 Institute of Waste Management and Contaminated Sites, Technische Universität Dresden, [email protected] 2 KP Ingenieurgesellschaft für Wasser und Boden mbH, Bahnhofstr. 37, D-91710 Gunzenhausen, [email protected] 1 2 1 Keywords: synthetic precipitation time series, transient flow modeling In the upper unsaturated soil layers, the water balance is mainly affected by pre-cipitation and its intensity. Heavy precipitation produces high infiltration fronts which result in a rapid mass transfer in the soil. This has significant effects on the dimensioning of capping systems, flood protection dams and on the risk assess-ment of the judging of contaminated sites, where so far the approach of average water balance values has generally been applied. For this reason over the last years a weather generator has been developed by the authors, which supplements simulation tools with a corresponding module, for example the program SiWaPro DSS for the simulation of flow – and mass transport processes in the unsaturated zone. On the soil surface, precipitation, root water uptake and potential evapora-tion as atmospheric boundary conditions have to be taken into the calculation. Background for the generation of synthetic time series for those variables in daily values is the analysis of available long term series from climate gauging stations (for example of the DWD). The characteristic of these real time series is described with statistic parameters. These parameters allows firstly a generation of time se-ries for unknown locations via regional interpolation and secondly a generation of time series of any length. After a validation and a trend test, for example the pre-cipitation is described through Markov-chain first order (for the precipitation con-dition of a day) and a Weibull distribution (for the amount of precipitation of a day). These parameters are then determined for a 14-day time period respectively, so that seasonal changes can be displayed. Using the synthetic time series, daily evapotranspiration can be calculated with the method of TURC-WENDLING. Further modules for correction of the evapotranspiration with data for declination, orientation and the value for Albedo value of the surface, for modelling the inter-ception reservoir and for modelling root water uptake reduce the available potential evapotranspiration to the required parameters root water uptake, potential soil evaporation and ground level precipitation. In the practical application, it was possibly to show that for “Süptitzer Berg”, a former military base, only by using transient infiltrations fronts, which are generated by the weather generator, the concentration of contaminants in soil and groundwater could be reproduced with http://www.dfg-wasserkommission.de Back to contents 72 of 88 Young Scientists Workshop on Data Assimilation sufficient accuracy. In the further development, the weather generator should gen-erate time series, which consider the influences of forecast climate change using data from calculated climate scenarios. With their help we will investigate ques-tions for the effect of climate change on the unsaturated soil zone. In the future, the weather generator will be used for questions of the effects of climate change on the water balance. Reference Nitsch, B., Gräber, P.-W.. 2007. Anwendung synthetischer Niederschlagszeitrei-hen bei der Strömungsmodellierung der ungesättigten Bodenzone. Simulation in Umwelt- und Geowissenschaften, Workshop Berlin 2007. Shaker Verlag, Bereich Umweltinformatik Gräber, P.W., Blankenburg, R., Kemmesies, O., Krug, S.. 2006. „SiWaPro DSS – Beratungssystem zur Simulation von Prozessen der unterirdischen Zone“; in: Si-mulation in Umwelt- und Geowissenschaften, Workshop Leipzig 2006; Shaker Verlag, Bereich Umweltinformatik Richter, Gunnar. 2003. Zeitliche Beschreibung der Sickerwassereinträge in den Boden. Diplomarbeit. Technische Universität Dresden. Institut für Abfallwirt-schaft und Altlasten Geng, S., Penning de Vries, F.W.T., Supit, L., 1985. A Simple method for gene-rating rainfall data. Agricultural and Forest Meteorology, 36: 363-376 Richardson, C.W. 1981. Stochastic simulation of daily precipitation, temperature, and solar radiation. Water Resource Research 17, pp. 182-190 Matalas, N.C. 1967. Mathematical assessment of synthetic hydrology. Water Re-source Research 3, pp. 937-945 Richardson, C.W., and D.A. Wright. 1984. WGEN: A model for generating daily weather variables. U.S. Department of Agriculture, Agriculture Research Service, ARS-8. p. 235 Richardson, C.W. 2000. Data Requirement for Estimation of Weather Generation Parameters. Trans. of ASAE 43(4), pp. 877-882 http://www.dfg-wasserkommission.de Back to contents 73 of 88 Young Scientists Workshop on Data Assimilation Dr. Karsten Rinke Dr. Karsten Rinke *1975 Aquatic Ecology Degree: Biologist (Dipl. Biol., TU Dresden, 2001); Ph.D., Faculty of Forest-, Geo-, and Hydrosciences (TU Dresden 2006). Curr. Position: Postdoc at the Limnological Institute (University of Konstanz, Prof. Rothhaupt). Research interests: Evolutionary ecology of zooplankton, functional ecology of phytoplankton, water quality management models, ecological modelling Motivation: Learning techniques about uncertainty analysis and stochastic simulations, meeting scientists from other disciplines for possible interdisciplinary project ideas Current address: University of Konstanz, Limnological Institute, Mainaustr. 252, 78464 Konstanz Tel. +49 7531 882930, Email: [email protected] Selected Publications related to the workshop: Schalau, K., Rinke, K., Straile, D. & Peeters, F. (submitted). Temperature is the key factor explaining interannual variability of Daphnia development in spring a modelling study. Rinke, K., Hülsmann, S. & Mooij, W.M. (accepted). Adaptive value, energetic costs, and underlying resource allocation patterns of predator-induced life-history shifts in Daphnia. Oikos Petzoldt, T. & Rinke, K. (accepted). Simecol: an object-oriented framework for ecological modeling in R. Journal of Statistical Software. Rinke,K., Hübner, I., Petzoldt, T., Rolinski, S., König-Rinke, M., Post, J., Lorke, A. & Benndorf, J. (2007). How internal waves influence the vertical distribution of zooplankton. Freshwater Biology 52: p137-144. Rinke, K & Vijverberg, K. (2005). A model approach to evaluate the effect of temperature and food concentration on individual life-history and population dynamics of Daphnia. Ecological Modelling 186: p326-344. Hülsmann, S., Rinke, K. & Mooij, W.M. (2005). A quantitative test of the size efficiency hypothesis by means of a physiologically structured model. Oikos 110: p43-54. Rinke, K., Vijverberg, J., Petzoldt, T. & Benndorf, J. (2005). Individual and population level dynamics of Daphnia at varying food, temperature, and fish predation: a model approach. Verhandlungen der Internationalen Vereinigung für Limnologie 29: p310-314. Rinke, K. & Petzoldt, T. (2003). Effects of temperature and food on individual growth and reproduction of Daphnia and their consequences on the population level. Limnologica 33: p293-304. Rinke, K., Robinson, C.T. & Uehlinger, U. (2001). A note on abiotic factors that constrain periphyton growth in alpine glacier streams. International Review of Hydrobiology 86: p361-366. Abstract Simulating phytoplankton community dynamics in Lake Constance with a coupled hydrodynamic-ecological model Rinke, Karsten ; Eder, Magdalena , Peeters, Frank , Gal, Gideon and Rothhaupt, Karl-Otto 1 Limnological Institute, University of Konstanz, Germany 1 http://www.dfg-wasserkommission.de 2 1 Back to contents 3 1 74 of 88 Young Scientists Workshop on Data Assimilation 2 Institute of Hydraulic Engeneering, University of Stuttgart, Germany 3 Kinneret Limnological Institute, Migdal, Israel Email addresses: [email protected], [email protected], [email protected], [email protected] keywords: hydrodynamics, limnology, phytoplankton dynamics, water quality management models, functional phytoplankton groups, longterm dynamics Lake Constance is among the largest lakes in central Europe and represents the most important drinking water reservoir in south-western Germany. In total, about 4 million people depend on this lake in terms of their drinking water supply and considerable efforts have been carried out to protect the high water quality of the lake. Besides drinking water supply the lake is also the receiving water body of numerous wastewater plants within its catchment and is moreover used for recreational purposes and fisheries. The pelagic zone of Lake Constance has been studied extensively in the past and detailed data about nutrient and plankton dynamics as well as physical properties are available. In an attempt to provide a model-based decision support system for water managerns of Lake Constance the BodenseeOnline-project was initiated. Within this project a complex model system (Fig.1) is developed enabling to perform online-simulations and scenario analyses with a coupled hydrodynamic-ecological model. The ecological model CAEDYM (Fig. 2) is applied to this purpose, which can either be coupled with a one-dimensional hydrodynamic (DYRESM) or a threedimensional hydrodynamic model (ELCOM). For the application of CAEDYM (Fig. 2) an ecosystem analysis of Lake Constance was carried out and simulations of historical data were conducted in order to find an appropriate model configuration and parameterisation. To account for the functional diversity of phytoplankton four functional phytoplankton groups were established and parameterised. Long-term simulations with the coupled model DYRESM-CAEDYM showed that the model is able to reproduce the typical succession patterns of the phytoplankton and the dynamics of total chlorophyll. However, besides uncertainties in the meteorological and hydrological forcing factors the physiological characterisation of the functional phytoplankton groups remains puzzling. Problems arise due to high variability in measurements of physiological rates and physiological plasticity of the organisms. The current realisation with four groups may represent a compromise between accounting for the prevalent diversity and keeping the model tractable. Fig. 1: Structure of the BodenseeOnline system (for Fig. 2: State variables of and included proceses in detailed information refer to our webpage at the ecological model CAEDYM. www.bodenseeonline.de). http://www.dfg-wasserkommission.de Back to contents 75 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 76 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Biomath. Stefanie Rost Hydrogeology Current affiliation: Potsdam Institute for Climate Impact Research Earth System Analysis Research Domain P.O. Box 60 12 03, 14412 Potsdam, Germany phone: +49-331-288-24 22 E-mail: [email protected] CV since 07.2005 10.2004 – 05.2005 10.2000 – 09.2004 Ph.D. Student, Department of Global Change and Natural Systems, Potsdam Institut of Climate Impact Research Master Thesis (Diplomarbeit) prepared at the TU Dresden Institute for Hydrobiology: „Methodology for the qualitative evaluation of water quality models by means of measured data“ („Methodik zur qualitativen Beurteilung von Gewässergütemodellen anhand von Messdaten“) Diploma Study of Natural Science Biomathematic at the Ernst-Moritz-ArndtUniversität Greifswald major subjects: Stochastic/Statistic, Ecology Degree: Master of Biomathematic (Dipl.-Biomath.Univ.) Abstract Modeling the agricultural green and blue water consumption and determining the anthropogenic influence on the global water system Global food production relies on both "blue" water (available in rivers, lakes and aquifers) and "green" water (soil water evaporating or transpiring through plants). This study quantifies for the first time, spatially explicitly and consistently, how much water currently is consumed by the Earth’s vegetation, differentiated between water consumption by irrigation, rainfed agriculture, and natural ecosystems. We use the dynamic global vegetation and water balance model LPJmL (LundPotsdam-Jena managed land), which simulates the growth and abundance of natural and agricultural vegetation and the related water fluxes in a single framework (Sitch et al, 2003; Gerten et al., 2004; Bondeau et al., 2007). For the present study, the model was enhanced by a river routing model, lakes and reservoirs as additional water storage pools, and modules to consider blue water withdrawals for households and industry (Jachner et al., subm.). Since on irrigated land agricultural plants use both blue water stemming from irrigation and green water as infiltrated from precipitation, a method was developed to quantify their individual fractions. We performed three different simulations: 1) natural and agricultural vegetation with irrigation as dependent solely on computed blue water stored in rivers, lakes and reservoirs; 2) natural and agricultural vegetation with a potential irrigation assuming that enough water is available in every irrigated place; and 3) potential natural vegetation, for which the green water flux compared to 1) and 2) is determined. We show that global agricultural water use is by far dominated by green water (though the blue water component predominates in certain regions). We also find that the conversion of land cover for agricultural use has reduced the global green water flow by ca. 5%. The simulation results highlight the importance of green water in global food production and ecosystem conservation, as well as the need for a focus on better management of green water resources. On this basis the virtual water content of crops (i.e. the amount of water required to produce a unit of crop yield) http://www.dfg-wasserkommission.de Back to contents 77 of 88 Young Scientists Workshop on Data Assimilation differentiated between green and blue will be estimated and the virtual water trade between nations and regions will be assessed in a consistent modeling framework. To evaluate the uncertainties the model results as simulated by two different models (LPJmL and WaterGAP) will be compared. Despite the important role of green water existing water stress indicators considering only blue water resources. For a consistent assessment of human dependencies on the global water system new water stress indicators combining green and blue water consumption as well as virtual water transport will be developed. References Bondeau, A., P. C. Smith, S. Zaehle, S. Schaphoff, W. Lucht, W. Cramer, D. Gerten, H. Lotze-Campen, C. Müller, M. Reichstein, and B. Smith (2007), Modelling the role of agriculture for the 20th century, Glob. Change Biol., 13(3), 679-706. Gerten, D., S. Schaphoff, U. Haberlandt, W. Lucht, and S. Sitch (2004), Terrestrial vegetation and water balance hydrological evaluation of a dynamic global vegetation model, J. Hydrol., 286(1), 249-270. Jachner, S., D. Gerten, A. Bondeau, W. Lucht, J. Rohwer, and S. Schaphoff, Agricultural green and blue water consumption and its influence on the global water system, Submitted to Water Resources Research. In July 2007. Sitch, S., B. Smith, I. C. Prentice, A. Arneth, A. Bondeau, W. Cramer, J. O. Kaplan, S. Levin, W. Lucht, M. T. Sykes, K. Thonicke, and S. Venensky (2003), Evaluationof ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model, Glob. Chang. Biol., 9(2), 161-185. http://www.dfg-wasserkommission.de Back to contents 78 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 79 of 88 Young Scientists Workshop on Data Assimilation Dr. Nele Schuwirth Dr. Nele Schuwirth *1977 Aquatic Ecology Degree: Dipl. Geology (Univ. Mainz, 2002); Ph.D., Geosciences (Univ. Mainz 2006) Curr. Position: Postdoc at the Department of Systems Analysis, Integrated Assessment and Modelling, Eawag: Swiss Federal Institute of Aquatic Science and Technology, Lecturer at ETH Zurich, Institute of Environmental Sciences Research interests: Construction of mechanistic models of biogeochemical and ecological processes in river ecosystems, Application of Bayesian techniques for the assessment of the identifiability of model parameters. Soil and groundwater protection, behaviour of inorganic contaminants in the environment, thermodynamic modelling of hydrochemical processes, application of soil water sampling techniques and leaching tests for the assessment of contaminated sites Motivation: meeting people, learning to know other system analytical strategies, getting new ideas, cooperation perspectives, Current address: Department of Systems Analysis, Integrated Assessment and Modelling, Eawag: Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, P. O. Box 611, CH-8600 Duebendorf/Switzerland, Phone: +41 (0)44 823 55 28, Fax: +41 (0)44 823 53 75, e-mail: [email protected] Selected Publications related to the workshop: Schuwirth, N., Kühni, M., Schweizer, S., Uehlinger, U., and Reichert, P.: A mechanistic model of benthos community dynamics in the River Sihl, Switzerland. Freshwater Biology (in revision). Abstract Combining data analysis and expert knowledge in a mechanistic river model using Bayesian inference Schuwirth, Nele & Reichert, Peter Keywords: Mechanistic models, Bayesian inference, river ecosystems The intention of the development of mechanistic models of aquatic ecosystems is to test hypotheses on ecosystem functionality and to asses the influence of driving forces. With the help of mechanistic models it is possible to estimate transformation rates from substance concentration and organism density data. Mechanistic models can be used to support the communication of quantitative knowledge about processes, to optimize the design of field measurements or laboratory studies, and finally to predict the future development of an ecosystem under changing environmental conditions including an estimation of the prediction uncertainty. A difficulty with the development of mechanistic models for aquatic ecosystems is the requirement of a detailed knowledge on structural relationships and external influence factors. Since these relationships are of complex nature mechanistic models very often tend to be overparameterized so that it is not possible to identify all model parameter from the available measured data. Frequentist parameter estimation procedures require a good identifiability of all calibration parameters. This http://www.dfg-wasserkommission.de Back to contents 80 of 88 Young Scientists Workshop on Data Assimilation prerequisite is often not fulfilled so that significantly different parameter sets lead to a very similar model output. This is known as the equifinality problem (Beven and Binley, 1992) occurring in many fields of environmental modeling. Therefore, it can be advantageous to include prior knowledge on parameter distributions from further experience and literature in the parameter estimation process and to update this prior knowledge with new measured data to calibrate the model. This can be done with the application of Bayesian inference (e.g. Gelman & Carlin, 1995). For this method identifiability is not required and it allows a consideration of uncertainty of model parameters and results. In this study we used Bayesian inference to calibrate a mechanistic river benthos community model, ERIMO (Ecological River Model), developed to simulate the dynamics of the most important functional feeding groups of benthic organisms (periphyton and invertebrates) (Schuwirth et al., submitted). We used process formulations at the lowest level of complexity consistent with a mechanistic, and as far as possible universal, description of the ecosystem. As a first step this model was applied to the River Sihl, Switzerland. The strategy for model development is summarized in Figure 1. Since processes determining the development of the benthic community are highly complex and due to multiple interferences with adjacent ecosystems many influencing factors are difficult to quantify. Figure 1: Concept for model development Despite the availability of a comprehensive data set, the complexity of processes to be considered even in a relatively simple model leads to an overparameterized model that cannot be calibrated uniquely with the available data. We therefore combine expert knowledge, literature data, and the experience from other model studies with the information contained in the data to confine the model parameters as good as possible and to consider uncertainty by applying Bayesian inference. This led to a successful calibration of the model, an analysis of the gain of information about the values of the model parameters, and an analysis of the uncertainty of model predictions. For validation and extension of the range of applicability, data from more investigation sites and more rivers would be required. The method of Bayesian calibration of process-based models turned out to be a promising method for application in aquatic ecology, but it can be easily transferred to a wide variety of disciplines, e.g., hydrology, ecotoxicology, air pollution (Arhonditsis et. al, 2007). We hope that the demonstration of the capabilities of our methodological approach, which leads to a deeper analysis of what we can learn from model-based data analyses, stimulates the use of such techniques in ecological studies. References: Arhonditsis, G.B., Qian, S.S., Stow, C.A., Lamon, C., and Reckhow, K.H. (2007): Eutrophication risk assessment using Bayesian calibration of process-based models: Application to a mesotrophic lake, Ecological Modeling (in press, doi: 10.1016/j.ecolmodel.2007.05.020). Beven, K., Binley, A., 1992. The future of distributed models–model calibration and uncertainty prediction. Hydrological Processes 6, 279–298. Gelman, S., Carlin, J.B., Stren, H.S. & Rubin, D.B. (1995) Bayesian Data Analysis, Chapman and Hall, New York, USA. Schuwirth, N., Kühni, M., Schweizer, S., Uehlinger, U., and Reichert, P. (submitted): A mechanistic model of benthos community dynamics in the River Sihl, Switzerland. Freshwater Biology (submitted). http://www.dfg-wasserkommission.de Back to contents 81 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 82 of 88 Young Scientists Workshop on Data Assimilation Dipl.-Ing- Leopold Stadler Dipl.- Ing. Leopold Stadler *1980 Hydrogeology Degree: Dipl.-Ing. Umweltschutztechnik, Universität Stuttgart Curr. Position: Fachgebiet für Wasserwirtschaft und Hydrosystemmodellierung, Institut für Bauingenieurwesen, TU - Berlin Research interests: - Multiphase flow in porous media - Coupling of software / models - Macropores - Unsaturated Zone + Groundwater + Transport - Model concepts Motivation I am working as a young scientist in the DFG researcher’s group "Natural Slopes, Coupling of Flow and Deformation Processes for Modelling the Movement of Natural Slopes” at the TU-Berlin (Technical University). The interdisciplinary workgroup consists of several sub-projects. In my sub-project “Subsurface Hydraulics” we simulate the infiltration and flow of rainwater in natural slopes. The complex structure of natural slopes, the interaction between macropores and the soil matrix, the geometry of macropores and the soil parameters are mostly unknown. This makes modelling to a challenging work, and the dealing with uncertainties plays an important role. To use models for simulation, we have to know the capabilities and limitations of our models and the uncertainties in the used model parameters, and boundary conditions. Current address: TU-Berlin, Institut für Bauingenieurwesen, Fachgebiet Wasserwirtschaft und Hydrosystemmodellierung, Sekr. TIB1-B14, Gustav-Meyer-Allee 25, 13355 Berlin email: [email protected], phone: +49 30 314 72428 Selected Publications related to the workshop: 2006 PHAM VAN,S., STADLER, L. & HINKELMANN,R.: Comparison of a Micro-scale and a Mesoscale Model Concept for Two-phase Flow in Fractured-porous Media. CMWR XVI Computational Methods in Water Resources, XVI International Conference, Copenhagen, Denmark STADLER,L., HELMIG,R., HINKELMANN,R. & PHAM VAN,S.: A Comparison of Model Concepts for Macropore Infiltration.. 6. Workshop "Porous Media", Blaubeuren Abstract Simulation of the rainwater infiltration via macropores into natural slopes Stadler L., Hinkelmann R., Germer K., Braun J., Helmig R. Keywords: Macropore; Landslides; Double continuum; Two-phase flow, Porous media, http://www.dfg-wasserkommission.de Back to contents 83 of 88 Young Scientists Workshop on Data Assimilation The failure of natural slopes consists of several coupled and complex processes, so that for a prediction hydrological, soil-hydraulic and soil-mechanical processes must be considered together with deterministical and statistical methods in order to take into account the available data on the one hand and the very different scales involved on the other hand. Since there is no overall simulation tool to describe all these processes, the coupling and exchange between different models is necessary. The DFG research’s group 581 Natural Slopes, “Coupling of Flow and Deformation Processes for Modeling the Movement of Natural Slopes” investigates the movement and deformation of natural slopes. Goal of the research is to develop new methods and simulation tools to predict the failure of slopes. The group investigates the movement and deformation of the slope Heumöser Hang in Ebnit (Austria). Therefore, a high number of hydrological and geophysical measurements have been carried out by the group. The field measurement showed that the fast infiltration of rainwater in macropores at the upper parts of the hill and the resulting pressure increase in the lower parts of the slope plays a dominant role for the deformation, which might lead to a failure. Macropores can generate a fast infiltration via the bypassing of rain water through mainly dry parts of a slope. This feature was captured by the group during tracer experiments with Brilliant Blue. The flow regime in macropores is very complex, in literature it is often distinguished between film and pulse flow. To simulate the fast infiltration we investigate different model concepts for two-phase flow (water/air) in porous media, using the numerical simulator MUFTE-UG (Multipahse Flow, Transport and Energy Model Unstructured Grids). The first studies with an easy example of a natural slope have shown that the existing concepts are not suitable to describe the fast infiltration and bypassing via macropores on a small scale. Simulating flow through a single macropore using the two-phase flow equations leads to extremely small time-steps and numerical problems. Also the flow regime is not described correctly by using the extended Darcy’s law. Models that use simple boundary conditions to describe macropores often overestimate the infiltration because the missing limitation of the water inflow into the macropore. The group has carried out a number of experiments to investigate the infiltration over a single macropore. To simulate the experiments with a two-phase flow model, we developed an extended boundary condition model. Therefore we use an external program that simulates the distribution of water in the macropore and the exchange to the soil matrix. The simulation for the macropore is similar to a cascade of storage cells and requires an interface length for the exchange with the soil matrix. For larger scales this concept is not suitable, thus we started to implement a double continuum model. The model consists of two continua, one that describes the soil matrix and a second describing the macropores. For both continua we use the two-phase flow equation with the extended Darcy’s law. The exchange between both continua is formulated with a function similar to the extended Darcy’s law. Therefore, we use a so called “interface length” to be able to compute the pressure gradient between both continua. Primary results of the double continuum model at the http://www.dfg-wasserkommission.de Back to contents 84 of 88 Young Scientists Workshop on Data Assimilation larger scale look very promising; the model can reproduce the fast bypassing of infiltrating water. But for a detailed evaluation benchmarks at different scales are necessary. We plan to transfer the knowledge from the experimental scale continuously to larger scales, till the scale of the natural slope. During this, the number of unknown variables increases rapidly. Not only the structure of the soil and the soil parameters like porosity, permeability and the corresponding soil-function parameters are often unknown, also the boundary conditions are more complex. The infiltration is coupled with the highly variable rainfall and the generated surface runoff. This makes it necessary to couple and exchange the data of different models to get realistic boundary conditions. At the large scale of a natural slope, the assimilation of data will be the main challenge. Since it is not possible to measure the parameters for the whole domain, geostatistical methods will be used. The unknowns of the larger scale create additional uncertainties. To handle the uncertainties arising from the different scales and processes an intelligent use of different methods is important. Without the estimation of the range of the different variables and parameters the use and calibration of the model is not possible. The prediction and the results of the model will be compared with measurements on different scales. Thereby it is important to distinguish between differences of the simulation results and the measurements. The differences can be caused by the limitation of the used model concept, and the uncertainties in the used simulation parameters. http://www.dfg-wasserkommission.de Back to contents 85 of 88 Young Scientists Workshop on Data Assimilation http://www.dfg-wasserkommission.de Back to contents 86 of 88 Young Scientists Workshop on Data Assimilation Dr. Henning Wilker Meteorology Current affiliation: Meteorological Institute University of Bonn Auf dem Hügel 20, 53121 Bonn, Germany phone: +49 228 735101 email: [email protected] CV 1991 – 1998 2001 – 04/2007 Since 04/2007 Higher education in Meteorology, University of Kiel. Diploma thesis: The effects of clouds on the radiation budget over the Baltic Sea. Ph.D. student at the Meteorological Institute of the University of Bonn. Dissertation: Soil moisture analysis based on microwave brightness temperatures: A study on systematic and random errors. Projects: ELDAS, LandSAF, geoland. Research associate at the Meteorological Institute of the University of Bonn. Research focus: Soil moisture analysis in numerical weather prediction models based on passive microwave remote sensing data. Abstract Soil Moisture Analysis Based On Microwave Brightness Temperatures: Single-Column Studies with the ECMWF Weather Prediction Model Henning Wilker1, Matthias Drusch2, Clemens Simmer1 1 Meteorological Institute, University of Bonn Auf dem H¨ugel 20, 53121 Bonn, Germany 2 European Centre for Medium-Range Weather Forecasts Shinfield Park, Reading, RG2 9AX, United Kingdom Keywords: Land Data Assimilation, Numerical Weather Prediction, Root-Zone Soil Moisture, Passive Microwave Remote Sensing Root-zone soil moisture governs the partitioning of available energy at the land surface into latent and sensible heat fluxes and thus plays an important role for the energy and water budget in the atmospheric boundary layer. Consequently, the soil water content can significantly affect the development of the local to regional weather. Numerical weather prediction therefore requires an accurate specification of the horizontal and vertical soil moisture distribution. Due to deficiencies in the soil moisture simulation of current operational weather prediction models (e.g. caused by imperfect parameterizations, uncertainties in radiation and precipitation predictions, inaccurately known soil and vegetation characteristics), modeled soil moisture values can drift away from reality and therefore need to be regularly adjusted by observations. Since a global soil moisture measurement network does not exist at present, proxy observations have to be used. Many current operational weather prediction systems use air temperature and humidity measured at a height of 2 m above ground (screen level) within the framework of the global meteorological observation network. These two variables are coupled to soil moisture through the sensible and latent heat fluxes at the land surface. The proxy observations are assimilated into the forecast models by means http://www.dfg-wasserkommission.de Back to contents 87 of 88 Young Scientists Workshop on Data Assimilation of an objective soil moisture analysis which merges the information of both observations and modeled values by taking into account their respective uncertainties. However, the use of air temperature and humidity requires a sufficient coupling between the land surface and the atmospheric boundary layer (e.g. high insolation and weak winds). Another proxy for soil moisture independent of the weather situation and therefore increasingly under consideration is the land surface brightness temperature at low microwave frequencies. The Soil Moisture and Ocean Salinity (SMOS; Kerr et al., 2001) mission will, for the first time, provide satellite observations at the L-band microwave frequency, which is best suited for monitoring nearsurface soil moisture. In preparation for the launch of the SMOS satellite scheduled for the year 2008, the European Centre for Medium-Range Weather Forecasts developed an soil moisture analysis system (based on an extended Kalman filter) which is able to assimilate both screen-level variables and microwave brightness temperatures (Seuffert et al., 2004). The goal of this new analysis system is to improve weather prediction by more realistic soil moisture values in the surface model. Positive effects are e.g. expected for the precipitation forecasts. Accurate precipitation forecasts are particularly important for hydrologic applications like water discharge prediction in order to assess the possibility and intensity of floodings. The combination of the proxy nearsurface soil moisture information retrieved from SMOS with appropriate coupled soil-atmosphere models can also help in assessing the water absorption capacity of the soil which also highly affects the risk of floodings in case of heavy rain. Single column-studies with the new soil moisture analysis system based on hydrological and meteorological data from field measurements and aircraft-borne brightness temperature observations over parts of the U.S. state of Oklahoma revealed promising results but demonstrate that an accurate specification of the model and observation uncertainties is fundamental (Wilker, 2007). The studies indicate that the new assimilation system is able to improve modeled root-zone soil moisture in comparison to observations. At the same time, however, the modeled surface fluxes partly get worse when compared to the flux observations. Reasons for this seem to be imperfect surface scheme parameterizations, some only roughly known vegetation and soil characteristics and difficulties in deriving exact observations of soil moisture and surface fluxes arising from the measuring techniques applied. References: Kerr, Y. H., P. Waldteufel, J.-P. Wigneron, J.-M. Martinuzzi, J. Font, and M. Berger, 2001: Soil moisture retrieval from space: The Soil Moisture and Ocean Salinity (SMOS) mission. IEEE Transactions on Geoscience and Remote Sensing, 39, 1729–1735. Seuffert G., H. Wilker, P. Viterbo, M. Drusch, and J.-F. Mahfouf, 2004: The usage of screen-level parameters and microwave brightness temperature for soil moisture analysis. Journal of Hydrometeorology, 5, 516–531. Wilker, H., 2007: Soil Moisture Analysis Based On Microwave Brightness Temperatures: A Study on Systematic and Random Errors. Ph.D. thesis, University of Bonn. http://www.dfg-wasserkommission.de Back to contents 88 of 88