Sensitivity of simulated extreme precipitation and temperature to convective parameterization using RegCM3 in China

Abstract

In this study, the regional climate model of RegCM3 is applied to investigate the sensitivity of regional climate over China using four cumulus parameterizations, the modified Anthes-Kuo (AK), the Grell with Arakawa-Schubert closure, the Grell with Fritsch–Chappell closure, and the MIT-Emanuel (EM). The model was integrated over the period of 1982 to 2001 using the NCEP Reanalysis data NNRP2 as boundary conditions. RegCM3 coupled with various cumulus parameterizations is evaluated firstly as for its ability to represent regional climatology and climate extreme indices, and the results show that simulated regional climate in China is sensitive to the option of cumulus parameterizations. All the cumulus schemes produce a northward expansion of heavy rain area and an underestimation of surface air temperature. For precipitation, the AK scheme simulates relatively better magnitude, while the EM scheme has more reliable performance on the spatial distribution. RegCM3 can represent the spatial distributions of extreme indices for both precipitation and temperature, as well as their decadal trends irrelevant to the cumulus parameterizations. However, the model underestimates the consecutive dry days and overestimates the three extreme wet indices, with the EM scheme giving the worst result. Slight underestimations of extreme temperature indices are detected in all cumulus parameterization scheme runs. The shapes of probability distribution functions for extreme indices are correctly produced, though the probabilities of extreme dry and warm events are underestimated.