Statistics of Large Scale Influence on Extreme Precipitation in Germany

Abstract

It is important to investigate the change in probability of extreme precipitation events in order to reduce impact consequences and help to prevent casualties. The main statistical characteristics of extreme precipitation events - return level, return period and time scale - are shown with Intensity-Duration-Frequency (IDF) curves. These curves help to visualize how extreme the event for different durations is by providing information on the probability of exceedance of precipitation intensities. The underlying distribution is modeled with the Generalized Extreme Value (GEV) distribution. The scarce availability of data can be addressed by using the available data more efficiently. Therefore, including maxima from different measurement durations is useful for (1) gathering more information from the data and (2) estimating return periods for different time scales with a consistent modeling approach. Consistency is ensured by modeling all durations in one step which prevents quantile crossing, a common problem of two-step IDF modeling.Large-scale information is included by modeling each of the GEV parameters as a polynomial function of the large-scale variables temperature, blocking situation, humidity, year and North Atlantic oscillation (NAO). All variables are spatially and monthly averaged. We show that the probability of extreme events increases with time, temperature and humidity over all seasons (summer, winter, whole year). The effects of blocking situation and NAO depend on the season with positive NAO leading to stronger events only in winter and blocking leading to stronger events only in summer and vice versa. A cross-validated model verification shows improvement over a reference model without large-scale information. Around 200 measurement stations with temporal resolutions from minutes to days are used to gather precipitation data across Germany.