Links between large-scale heavy precipitation over the Czech Republic and atmospheric circulation in regional climate model outputs

Large-scale heavy precipitation events (LHPEs) are associated with negative impacts on society, mainly as they may trigger floods and landslides. Therefore, it is important to better understand underlying physical mechanisms leading to these extremes and how they are reproduced in climate models. The present study evaluates ability of regional climate models (RCMs) to reproduce relationships between LHPEs over the Czech Republic and atmospheric circulation. We use an ensemble of 32 RCM simulations with the 0.11° resolution, taken from the Euro-CORDEX project. The historical simulations (1951-2005) are compared against observations from the E-OBS gridded data set. LHPEs are defined as days with at least 70% of all grid points over a given area with precipitation amounts exceeding the 90th grid-specific percentile of the seasonal distribution of daily amounts. The association with atmospheric circulation is investigated through circulation types derived from sea level pressure using airflow indices (direction, strength and vorticity). The analysis is carried out separately for summer (JJA) and winter (DJF) season.

In observations, the frequency of all LHPEs is higher in winter than summer, which is associated with larger importance of frontal stratiform precipitation in cold part of the year. The links of LHPEs to circulation differ between the seasons and between the western and eastern part of the examined area with varied Atlantic/Mediterranean influences. In winter, observed LHPEs are connected mainly with cyclonic type and westerly supertype. Connection with non-westerly and cyclonic-nonwesterly supertypes is more frequent only in the eastern region. The majority of RCMs simulate differences in circulation-to-precipitation link between the western and the eastern regions well. However, almost all RCMs overestimate the frequency of cyclonic type during LHPEs. In summer, the importance of cyclonic and non-westerly types in producing LHPEs increases compared to winter, and RCMs are able to capture this pattern. The prevailing type during events simultaneously occurring in both regions is cyclonic type with frequency 56%, which is 6 times higher than its mean summer occurrence. Only few models are able to simulate the strong link between events simultaneously occurring in both regions and cyclonic type. In conclusion, the method proposed for LHPEs is able to capture the precipitation events potentially leading to floods and may be suitable to characterize the flood-risk conditions in future climate scenarios.