The simulation of the variability and extremes of daily precipitation over Europe by the HIRHAM regional climate model

Abstract

In this study the simulation of various aspects of daily precipitation events over Europe in an ensemble of three simulations with the HIRHAM RCM for the period 1961–1990 is investigated. Daily precipitation events are characterized by their frequency and intensity, and heavy precipitation events are described by 30-year return levels of daily precipitation. Further, extended periods with and without rainfall, the so-called wet and dry spells, are studied, considering frequency, duration and extreme amounts of precipitation associated with wet spells, again described by 30-year return levels. By presenting the results based on the simulations, a comprehensive picture of the different aspects of daily precipitation over Europe is given. By comparing these results with the results based on 142 stations from the ECA data set, the quality of the simulations is evaluated in those regions where observational data are available. Furthermore, the use of an ensemble of simulations allows for estimating the sampling uncertainty due to internal variability. The geographical distribution of daily precipitation simulated by HIRHAM reflects the different nature of precipitation in the different seasons. In winter, precipitation is formed in large-scale frontal systems, while in summer much of the precipitation is formed through convective processes. The distribution of the intensity of extreme daily precipitation events is very similar to the corresponding distribution of the overall precipitation intensity, with the highest values of the 30-year return levels exceeding 100 mm/day in winter and 80 mm/day in summer. The distribution of the 30-year return level based on the precipitation during wet spells reveals the same maxima as the corresponding distribution for daily precipitation with more than 270 mm in winter and more than 210 mm in summer. The comparison with the observations reveals that HIRHAM is better at simulating the characteristics of daily precipitation during the cold than during the warm season. Both the intensity of daily precipitation and the magnitude of extreme daily precipitation events are underestimated in summer but simulated realistically in winter, so that the seasonal variation of these parameters is generally too weak in the simulations. Also, the model does not simulate enough wet days in summer except for central and southern Europe, where wet days occur too often. By using an ensemble of three simulations instead of a single realisation, the effect of the internal variability on the estimates of the 30-year return level has been reduced, in particular for the upper confidence limit by one third.