The Central Europe Refined analysis version 2 (CER v2): Dynamical downscaling of ERA5 precipitation data for the metropolitan region Berlin-Brandenburg

The Central European Refined analysis (CER) was developed in 2016 as a high-resolution, reanalysis-based, gridded dataset for Central Europe. The second version (CER v2) aims to further improve the performance of the CER with a particular focus on precipitation data for the metropolitan region Berlin-Brandenburg. The simulation setup consists of two-way nested, cascaded domains for Germany (10 km grid spacing) and the region Berlin-Brandenburg (2 km grid spacing) and employs a daily re-initialization approach. Major changes from the precursor version include the use of ECMWF-ERA5 reanalysis forcing data and a newer WRF version, allowing for the production of longer time series. To further improve the precipitation performance for the CER v2 we performed sensitivity experiments with five cumulus and five microphysics schemes. The results of these test simulations were evaluated using one year of daily precipitation data at 244 stations of the German Weather Service (DWD) in the 2 km domain of the model.  The best average performance was achieved for a combination of the conventional Kain-Fritsch cumulus and the Thompson microphysics scheme. Using this setup, we simulated the precipitation conditions for 30 years (1991-2020) and evaluated monthly and annual precipitation averages against station and radar data by the DWD. Here, the CER v2 showed a significant reduction in deviations and mean bias compared to the previous version. Based on the spatial resolution of the ERA5 data, we resampled the CER v2 and observational data to compare the performance of both datasets. We observed a wet bias in the ERA5 precipitation data for this region, which was significantly reduced in the CER v2. Results of monthly averages indicated a comparable performance to ERA5 data throughout most of the year. Deviations from the observational data were typically higher during the summer months. However, due to the significant bias reduction and the high spatial resolution, the CER v2 could provide important insights about the local- to mesoscale precipitation dynamic of this region.