Heavy and extreme precipitation events in the Sichuan Basin during the 2020 summer season in a set of kilometre-scale simulations

Abstract

The Sichuan Basin (SB), a lowland region in southwest China located at the eastern slope of the Tibetan Plateau (TP), regularly experiences heavy and extreme precipitation events. These extreme events often lead to flooding that can pose a threat to people’s lives and livelihoods in this densely populated area. A notable example is the summer of 2020, during which large parts of East Asia experienced anomalously high precipitation. In the SB, these events broke the previous record of daily accumulated rainfall at multiple stations. Since such events are expected to increase in both frequency and intensity in a warmer climate, understanding their causes and the physical processes involved is of high relevance in the SB region. Modelling the climate in mountainous regions with complex topography is challenging but recent developments in convection-permitting modelling make it possible to perform process-based studies. The CORDEX Flagship Pilot Study Convection-Permitting Third Pole (CPTP) aims to improve our understanding of the water cycle over the TP and its surrounding regions using a multi-model ensemble of kilometre-scale simulations. Recent results using a set of CPTP simulations for one extreme precipitation event suggest that an accurate representation of the large-scale forcing is crucial to correctly simulate the event. In this study, we assess how well the kilometre-scale CPTP models WRF, ICON-CLM and COSMO-CLM capture multiple observed heavy and extreme precipitation events that occurred in the SB during the summer of 2020 by validating them against observations and reanalysis data. In addition, we investigate how the simulations differ among each other in representing the observed events and related important physical factors, e.g. large- and mesoscale circulation and moisture transport. We find that all model systems fail to reproduce the observed intensity and location of the precipitation events as they either underestimate the intensity or simulate the event in a different location. Our analysis shows that model performance in simulating precipitation depends on their ability to reproduce the jet stream and the transport of water vapour. We conclude that ICON-CLM is most skillful for all analysed events, but there is need for further improvement in convection-permitting modelling in the TP region.