Meteorologische Zeitschrift (May 2016)
A Central European precipitation climatology – Part II: Application of the high-resolution HYRAS data for COSMO-CLM evaluation
Abstract
The horizontal resolution of regional climate model (RCM) simulations is increasing constantly in the last years. For the evaluation of these simulations and the further development of the models, adequate observational data sets are required, in particular with respect to the spatial scales. The aim of this paper is to investigate the value of a new high-resolution precipitation climatology, the HYRAS-PRE v.2.0 data set, for the evaluation of RCM output. HYRAS-PRE is available for the time period 1951–2006 at daily resolution and covers ten river catchments in Germany and neighbouring countries at a spatial grid spacing of 5 km. A set of simulations with the regional climate model COSMO-CLM with three different grid spacings (~7$\sim7$, 14 and 28 km) is used for this model evaluation study. In addition, three other data sets with different horizontal resolution are considered in the comparisons: the E‑OBS v.8.0 gridded observations (~25$\sim25$ km grid spacing), the ERA-Interim reanalysis (~79$\sim79$ km) and the analysis of the driving model GME (~40$\sim40$–60 km). For three selected years, different spatial and temporal characteristics of daily precipitation are investigated. In all the analyzed precipitation characteristics, it is found that the variability between the data sets is very large. The benefit of an evaluation with HYRAS-PRE compared to coarser-resolved observations becomes visible especially in the representation of the frequency of occurrence distribution of daily precipitation amounts and in the spatial variability of different precipitation indices. A second goal of this study was to estimate the error when comparing a high resolution simulated precipitation field with coarser resolved observations. Comparing the HYRAS-PRE average over an area of 5×5$5\times5$ grid points with the original HYRAS-PRE data results in a systematic underestimation of high values of all indices considered and an overestimation of small values. The maximum values of the different indices across the whole HYRAS-PRE domain are underestimated by 3–20 %. The regional variability is not very high but with a slight orographic dependence. Problems with areas of low station density can also be detected. In conclusion, this study shows that good-quality observational data sets at high-resolution are essential for a correct evaluation of the performance of RCMs designed for applications at high spatial resolution.
Keywords
