Evaluating precipitation products for water resources hydrologic modeling over Germany

Accurate precipitation data are crucial for many sectors and applications, like managing water resources, for agriculture, or assessing the risks of hydrometeorological extreme events like floods and droughts, which are expected to further increase with climate change. This study compares the spatial and temporal characteristics of ten state-of-the-art, commonly used precipitation datasets, with each other and against reference in situ precipitation gauge observations from the European Climate Assessment & Dataset (ECA&D) over Germany. The objectives are to evaluate whether bias adjustment is needed for the European Centre for Medium-Range Weather Forecasts (ECMWF) High Resolution (HRES) meteorological forecasting dataset, which is used in near real-time water resources modeling with the ParFlow integrated hydrologic model, and if so, to assess whether any of the observation-based comparison datasets might be suitable for this bias adjustment. Results show that HRES and Reanalysis v5 (ERA5) capture spatial patterns well, albeit with deficits in reproducing extremes, and over- and underestimation at low and high altitudes, respectively. COSMO-REAnalysis (COSMO-REA6) captures the spatial precipitation patterns less effectively but outperforms HRES and ERA5 in reproducing extreme events. HYRAS-DE-PRE (HYRAS), Radar Online Adjustment (RADOLAN), and Radarklimatologie (RADKLIM) perform very well, showing strong spatial accuracy and potential for bias adjustment, though their limited spatial coverage potentially restricts their use across all river catchments affecting Germany. The Operational Program of the Exchange of Weather Radar Information (OPERA) tends to underestimate mean precipitation quantities and extreme events. Integrated Multi-satellite Retrievals for Global Precipitation Measurement (IMERG) Final shows an improvement over IMERG-Late. EUropean RADar CLIMatology (EURADCLIM) outperforms OPERA due to gauge adjustments. The methodology and findings from this study may also be applicable to similar evaluations in other regions, and may help in the selection of precipitation datasets, e.g., for hydrological model forcing or for bias adjustments.