Sensitivity of CWRF simulations of the China 1998 summer flood to cumulus parameterizations

Better understanding the dynamics of the East Asian monsoon system is essential to address its climate variability and predictability. Regional climate models are useful tools for this endeavor, but require a rigorous evaluation to first establish a suite of physical parameterizations that best simulate observations. To this end, the present study focuses on the CWRF (Climate extension of WRF) simulation of the 1998 summer flood over east China and its sensitivity to cumulus parameterizations on CWRF performance. The CWRF using the Kain-Fritsch and Grell-Devenyi cumulus schemes both capture the observed major characteristics of geographic distributions and daily variations of precipitation, indicating a high credibility in downscaling the monsoon. Important regional differences, however, are simulated by the two schemes. The Kain-Fritsch scheme produces the better precipitation patterns with smaller root-mean-square errors and higher temporal correlation coefficients, while overestimating the magnitude and coverage. In contrast, the Grell-Devenyi ensemble scheme, using equal weights on all closure members, overall underestimates rainfall amount, suggesting for future improvement with varying weights depending on climate regimes.