Flood forecasting for fast responding catchments including uncertainty

Flood forecasting for fast responding catchments encounters problems especially in terms of short warning periods and a very limited reliability. We envisage tackling these shortcomings by using a symbiosis between physically based stochastic hydrological modelling and computationally highly efficient artificial intelligence techniques which surpasses current deterministic forecasting practice and/or the high computational burden of hydrologic/meteorological ensemble forecasting. Within a new stochastic decomposition framework based on rigorous rainfall-runoff modelling along with new perturbation and stochastic inference techniques we consider uncertainties of three sources: (i) hydrologic calibration uncertainty, (ii) hydrologic soil data uncertainty, and (iii) the uncertainty of the meteorological rainfall forecast. Mirroring the results of hydrologic stochastic decomposition by a problem specific stochastic Artificial Neural Networks (ANN-S) finally allows the instantaneous computation of the runoff under consideration of hydrological uncertainties. Combining the hydrologic uncertainty with the meteorological uncertainty gained from a large number of ANN-S applications to rainfall scenarios generated by radar based ensemble forecasts then allows for a real-time operation for flood forecasting including a realistic assessment of the uncertainties involved.

Duration:      2007-2010
Financed by: DFG

Cooperation

• Meteorologisches Institut (Uni Bonn)
• Institut für Grundwasserwirtschaft (TU Dresden)
• Inst. Polytecnique de Grenoble (Frankreich)
• DWD
• IIT Kharagpur (Indien)

Project team

Prof. Gerd H. Schmitz, Dr. Niels Schütze, Thomas Krauße