See our project:


Near-Realtime Quantitative Precipitation Estimation and Prediction

©Mohamed Saadi

The Research Unit (RU) RealPEP develops a flash-flood prediction system for small to medium-size river catchments for Germany including significant improvements at all stages along the process chain (precipitation estimation, nowcasting, numerical prediction, hydrological modelling) exploiting the full observational state evolution information of the soil and the precipitation-generating atmosphere.

RealPEP makes use of radar polarimetric and Commercial Microwave Link information for quantitative precipitation estimates, radar and satellite data for quantitative precipitation nowcasting of the future one to three hours, the assimilation of all this information in numerical weather prediction for quantitative precipitation forecasting, and feeds the results in hydrological models which provides forecasts for water levels and runoff.

While the first funding phase did develop the individual components to a first application-ready state, phase 2 will further advance in these efforts and combines all parts into one ensemble-based flash-flood prediction system and provides for it in-depth evaluation.

Animation (by courtesy of Mohamed Saadi): Evolution of the nowcasted hydrograph for the flooding event on 14 July 2021 in the Ahr river at Altenahr. Simulations were obtained using the hydrological model ParFlow-CLM (3D, physically-based, 0.6 km resolution). ParFlow-CLM was fed by polarimetric rainfall retrievals and 20 ensembles of 3-h nowcasts, spawned each hour between 1:00 UTC and 18:00 UTC. Q100 indicates the 100-year flood, QHistPeak indicates the historically highest measured peakflow, and QEventPeak the last measured flow of the event, occurred at the time indicated by the vertical red line.

RealPEP is a joint effort by scientists of the University of Bonn, Deutscher Wetterdienst, Free University of Berlin, Forschungszentrum Jülich, and Karlsruhe Institute of Technology, Campus Alpin.

The research unit (RU-2589) is funded by the German Research Foundation (DFG).

For more information see