Spatial and temporal characteristics of heat waves over Central Europe in an ensemble of regional climate model simulations

The study examines the capability of regional climate models (RCMs) to reproduce spatial and temporal characteristics of severe Central European heat waves. We analysed an ensemble of seven RCM simulations driven by the ERA-40 reanalysis over the 1961–2000 period, in comparison to observed data from the E-OBS gridded dataset. Heat waves were defined based on regionally significant excesses above the model-specific 95 % quantile of summer daily maximum air temperature distribution and their severity was described using the extremity index. The multi-model mean reflected the observed characteristics of heat waves quite well, but considerable differences were found among the individual RCMs. The RCMs had a tendency to simulate too many heat waves that were shorter but their temperature peak was more pronounced on average compared to E-OBS. Deficiencies were found also in reproducing interannual and interdecadal variability of heat waves. Using as an example the most severe Central European heat wave that occurred in 1994, we demonstrate that its magnitude was underestimated in all RCMs and that this bias was linked to overestimation of precipitation during and before the heat wave. By contrast, a simulated precipitation deficit during summer 1967 in the majority of RCMs contributed to an “erroneous” heat wave. This shows that land–atmosphere coupling is crucial for developing severe heat waves and its proper reproduction in climate models is essential for obtaining credible scenarios of future heat waves.