Distributed modeling at the catchment scale (Contact: Dan Rosbjerg, Peter Bauer-Gottwein)
There is a growing demand for distributed and semi-distributed hydrological models that can describe the state of a catchment including stream flow, groundwater level, soil moisture etc. Such models depend on the necessary input data and on an accurate catchment conceptualization. The use of GIS tools and remote sensing data are being investigated to improve model performance. Regional-scale hydrological models are key decision support tools to guide the management and allocation of water and land resources. Long-term field studies include the Okavango Delta, Botswana and the Eastern Yucatán karst aquifer, Mexico.
Reservoir design and operation (Contact: Dan Rosbjerg)
In most cases models used for the design of reservoirs are simple continuity models. Synthetic inflow series can be used in conjunction with such models to develop, amongst other things, storage-reliability-yield curves. The simple rule curves used in reservoir operation can be made more effective by introducing optimization techniques and adaptive management that combine real-time data, inflow forecasts and historical data.
Flood forecasting and data assimilation (Contact: Peter Bauer-Gottwein, Dan Rosbjerg)
Accurate flood forecasts are extremely important, but model predictions are uncertain due to imprecise model forcing and imperfect models. Assimilation of measurements, however, can reduce the forecast error. Methods like Kalman filtering are attractive because they can explicitly take model and data uncertainties into account and provide estimates of the resulting predictive uncertainty.
Parameter identification and model uncertainty (Contact: Dan Rosbjerg)
In distributed catchment modeling the various hydrological processes are considered jointly, and this makes parameter identification and model uncertainty assessment a very complex task. Global optimization procedures have proven to be effective and robust calibration tools, but they need to be expanded to include parameter and prediction uncertainty using methods like Bayesian and Markov Chain Monte Carlo analysis.
Flood frequency and reliability analyses (Contact: Dan Rosbjerg, Peter Steen Mikkelsen)
Up-to-date estimation of hydrological design values by means of flood frequency analysis must include an assessment of the predictive uncertainty based on advanced statistical tools. Regionalization methods allow flood frequency analyses to be extended to ungauged and data-poor sites