Sub-hourly precipitation and rainstorm event profiles in a convection-permitting multi-GCM ensemble

Abstract

Extreme precipitation on short, sub-hourly time scales has the potential to trigger flash floods, is a particular threat to urban areas, and is expected to increase with climate change. However, little is known about sub-hourly precipitation extremes in convection-permitting climate models (CPMs). We investigate sub-hourly precipitation in the KIT-KLIWA ensemble — a CPM climate ensemble driven by 3 CMIP5 GCMs coupled to the regional climate model COSMO-CLM. The domain is centred over Germany with a grid resolution of 0.025 °(2.8 km). In an event-based analysis, we compare extreme precipitation down to 5-min resolution for a historical simulation (1971–2000) with the dense radar and raingauge observation network in Germany. We find that 5-min CPM precipitation data adequately reproduces the frequency distribution from radar measurements. To improve the understanding of the precipitation bias in CPM simulations we propose an event-based analysis, that reveals a tendency for the CPM to overestimate the occurrence of longer events, and for a simulation bias in the representation of heavy and short, likely convective, precipitation events. The CPM historical simulations mostly reproduce the event precipitation sum for events leading to 1 h and 6 h annual maxima. Maximum 5-min peak intensities of these extreme precipitation events agree with spatially-aggregated radar data but are well below intensity maxima observed in station data. A dominant (very) front-loaded shape for precipitation events leading to 1 h annual maxima is reproduced by the CPM ensemble. The demonstration that CPMs effectively capture the key features of rainstorm profiles, opens up opportunities for climate change studies and their application in hydrological modelling.