Trends In Frequency And Intensity Of Extreme Precipitation At Four German Stations In Recent 100 Years And Distribution Fit By Extreme Value Theory

Extreme precipitation has been reported more and more in recent years. An attempt is made to study the trends in frequency and intensity of extreme precipitation occurrences at four stations, which are Bremen, Potsdam, Karlsruhe and Zugspitze, in Germany during 1901 to 2007. Data of daily precipitation is used. According to the definitions of heavy rain and light rain, 20 mm day-1 is considered as one threshold. Normally extreme events are regarded as the top 5 percent in probability density distribution. So the 95th percentile is chosen to be the other threshold. An event with precipitation larger than the threshold is defined as an extreme precipitation. The counts of extreme precipitation every year are recorded to show the variation of frequency. As for the intensity of extreme precipitation, the 95th percentile of precipitation each year is considered as its index. Besides, precipitation owns strong seasonality so winter and summer variations in frequency and intensity are also analyzed. Mann-Kendall method is used to test if there is a significantly increasing or decreasing trend for a series of data. Results tell that the four stations have different behaviors no matter in frequency or intensity. As for extreme precipitation over 20 mm day-1, Bremen and Potsdam both have insignificant developing trends in frequency. They keep stable in occurrences. Karlsruhe has a significantly decreasing trend while Zugspitze has a significantly increasing trend in frequency. Furthermore, the shapes of frequency trends of Bremen and Potsdam are similar, and they are almost opposite phase to that of Karlsruhe. Mean decadal frequency of Bremen and Potsdam have strong positive correlation while they have strong negative correlation with Karlsruhe. As for extreme precipitation over the 95th percentile, Bremen and Zugspitze have significantly increasing trends in frequency, while Potsdam and Karlsruhe have significantly decreasing trends. When it comes to intensity of extreme precipitation, only Zugspitze has an increasing trend. Other three stations almost keep still and have no significant trends. In winter, most stations have significantly increasing trends no matter in frequency of in intensity of extreme precipitation, except for Potsdam, who does not show significant variation in counts of extreme occurrences. In summer, Bremen neither has obvious trend in frequency nor in intensity significantly. At Potsdam there are fewer and fewer events of extreme precipitation but their intensities do not vary too much. The amount is becoming smaller at Karlsruhe and so is the intensity of extreme precipitation. Otherwise, Zugspitze has more and more extreme precipitation and it is going to be stronger and stronger.According to study on North Atlantic Oscillation, winter extreme precipitation frequency in Germany has a correlation with NAO signals. The stronger the NAO is, there is more extreme precipitation in winter and heavier; otherwise it is the same. During the time when large precipitation is easy to occur, such as August, the temperature anomaly of North Atlantic Ocean could affect the intensity of extreme precipitation in northern Germany. When there is high temperature anomaly above the upstream of Icelandic Low, the enhanced evaporation and convection make plenty of warm and wet air above the sea surface, which could be transported to the downstream by the trough resulting in a large scale of precipitation. The adjustment of atmospheric circulation plus the topography of middle Germany make differences in extreme precipitation between the north and the south. Were the westerly strong, the south would have a great many extreme precipitation while the north would have little, otherwise the same. Under the background of global warming, the alpine snow smelting is promoted, leading to encouragement of evaporation. Then vertical temperature gradient becomes larger and convection becomes stronger, which provides opportunities to heavy precipitation. Thus alpine extreme precipitation has upward trend in frequency and intensity.According to Extreme Value Theory, General Extreme Value distribution (GEV) and General Pareto Distribution (GPD) are fit to the precipitation data at those four stations. Using the maximum likelihood estimation, all parameters are estimated and all stations are separated into different types according to the values of parameters. Results say Potsdam and Karlsruhe both belong to the Frechet family while Bremen and Zugspitze belong to the Gumbel family, for GEV fit. For GPD fit, Bremen, Potsdam and Karlsruhe all fall into the type of heavy-tail distribution (Pareto distribution). However Zugspitze goes into the light-tail distribution (exponential distribution). Furthermore, for each station, different return levels are also estimated, corresponding to different return periods.