The dynamic of precipitation extremes in Berlin and Brandenburg based on a generalized Pareto distribution

In recent years, Berlin and its surrounding area has experienced multiple intensive precipitation events, which caused significant damage and severely impaired the local infrastructure. An extreme value analysis of the available station data could improve the understanding of such events and provide valuable information for risk assessments in the region. Additionally, results gathered from this analysis will serve as a reference for similar evaluations with the Central European Refined analysis (CER), a gridded dataset generated via dynamical downscaling of ERA5 data using the Weather Research and Forecasting model. In this study, we assess the spatiotemporal dynamics of extreme precipitation event days in Berlin and Brandenburg using a generalized Pareto distribution (GPD).  Daily precipitation data of the last 30 years was extracted at 137 stations of the German Meteorological Service. For all stations with a sufficient amount of data a seasonal time-dependent threshold was defined and independent exceedances were extracted using an automatic declustering scheme. The resulting threshold series was used to fit a GPD at each location with a time-dependent scale parameter to investigate temporal changes in extreme event days. After evaluating the goodness of fit the distribution model was used to calculate seasonal 2-, 10-, and 20-year return levels. The results indicate a high regional variability especially for the 20-year precipitation extremes with slightly higher values around Berlin reaching up to 130 mm d^-1 in the summer. Return levels also tended to be higher in the northern parts of Berlin and Brandenburg during winter and in the south during spring.