CR 82:97-115 (2020)  -  DOI: https://doi.org/10.3354/cr01621

ABSTRACT: To evaluate and clarify the daily precipitation characteristics (i.e. amount, frequency and intensity) of the regional climate models (RCMs) in China, long-term simulations were carried out using RegCM4.5 and Weather Research and Forecasting model (WRF), which were nested within the European Centre for Medium-Range Weather Forecasts (ECMWF)’s 20^th century reanalysis (ERA-20C) between 1901 and 2010. The 2 RCMs were initially run at a resolution of 50 km. Analyses mainly compared the model-simulated climatic means and interannual variations of precipitation characteristics with those of dense and high-quality station observations (STN) from 1961-2010. Both models satisfactorily reproduced the seasonal mean precipitation amount, but they overestimated its frequency and underestimated its intensity. Extreme rainfall frequency was also underestimated by both RCMs. In winter (DJF), the interannual variabilities in dry days, light precipitation and moderate precipitation were well represented by both models. However, they poorly reproduced the counterparts of extreme precipitation in winter. In summer (JJA), the 2 RCMs performed well in simulating the interannual variability of extreme precipitation. Comparably, RegCM outperformed WRF in reproducing the spatial patterns of precipitation amount, interannual variations in extreme precipitation and rain events. By contrast, WRF better represented precipitation frequency in different sub-regions overall. Moreover, when the horizontal resolution of RegCM was increased from 50 to 25 km, there was a slight improvement in the representation of precipitation amount and intensity. Our results show that RCMs perform well in reproducing actual climatic means and interannual variations of daily precipitation characteristics in China, and that high-resolution RCM simulations can produce improved results for precipitation amount and intensity.

KEY WORDS: RegCM · WRF · Dynamical downscaling · Extreme precipitation