Flooding events are one of the most dangerous hazards for society. Central Europe was regularly affected by high-impact river floods within the last few decades, causing high socioeconomic losses. In order to improve flood protection, a detailed understanding of the underlying processes and precise return value estimates of the associated extreme precipitation is from importance, especially when these characteristics change in a warmer climate. For process analysis, past studies have either analysed individual events or systematically studied events with smaller return periods on the order of one year, due to the limited length of observational time series. The main objective of this thesis is, hence, to perform such a systematic analysis of 100-year precipitation events.
In order to improve the understanding of atmospheric processes during extreme precipitation, robust process analyses are performed in this thesis. For this, ensemble simulations are used to generate a large set of daily precipitation events. With this approach, time series can be obtained that are considerably longer than observational time series, resulting in a more systematic assessment of processes (beyond individual events) and in reduced uncertainties of return value estimates. Firstly, ensemble weather prediction (EPS) data from the ECMWF are used to investigate atmospheric processes of realistic 100-year events over central European river catchments and their differences to less extreme events. 100-year events are generally associated with an upper-level cut-off low over central Europe in combination with a surface cyclone southeast of the specific catchment. Differences to less extreme events vary between the catchments, indicating that either dynamic processes (intensified upper-level cut-off low and surface cyclone) or thermodynamic mechanisms (higher lower-tropospheric moisture content) are more relevant. Secondly, 100-year return values of daily precipitation are estimated on a global scale from the EPS data, substantially reducing the uncertainty compared to observational estimates. Despite a general agreement of spatial patterns, the model-generated data set leads to systematically and significantly higher return values in several regions (e.g. the Amazon, western Africa, the Arabian Peninsula). This may point to an underestimation of very extreme precipitation events in observations. Finally, changes of daily 10-year precipitation events (due to the too small sample size for a robust analysis of 100-year events) over central European river catchments are analysed from ensemble climate simulations between the historical and a projected warmer climate. Precipitation extremes are projected to intensify and occur more frequently in spring and autumn. Dynamical changes of these events are often associated with a slight westward shift of the upper-level cut-off low, an intensified ridge over eastern Europe and less pronounced changes near the surface. Thermodynamic changes such as an increased horizontal temperature gradient and higher lower-tropospheric moisture content play an important role as well.
