Sensitivity of Southeast Asia rainfall simulations to cumulus and air-sea flux parameterizations in RegCM4

Liew Juneng
Fredolin Tangang
Jing Xiang Chung
Sheau Tieh Ngai
Tze Wei Tay
Gemma T. Narisma, Ateneo de Manila University
Faye T. Cruz, Manila Observatory
Tan Phan-Van
Thanh Ngo-Duc
Jerasorn Santisirisomboon
Patama Singhruck
Dodo Gunawan
Edvin Aldrian

Abstract

We investigated the performance of RegCM4 in simulating rainfall over Southeast Asia with different combinations of deep-convection and air-sea flux parameterization schemes. Four different gridded rainfall datasets were used for the model assessment. In general, the simulations produced dry biases over the equatorial region and slightly wet biases over mainland Indo-China, except those experiments with the MIT Emanuel cumulus schemes, in which large positive rainfall biases were simulated. However, simulations with the MIT schemes were generally better at reproducing annual rainfall variations. The simulations were not sensitive to the treatment of air-sea fluxes. While the simulations generally produced the rainfall climatology well, all simulations showed stronger inter-annual variability compared to observations. Nevertheless, the time evolution of the inter-annual variations was well reproduced, particularly over the eastern Maritime Continent. Over mainland Southeast Asia, all simulations produced unrealistic rainfall anomaly responses to surface temperature. The lack of summer air-sea interactions in the model resulted in enhanced oceanic forcing over the regions, leading to positive rainfall anomalies during years with warm ocean temperature anomalies. This shortcoming in turn caused much stronger atmospheric forcing on the land surface processes compared to that of the observation. A robust score-ranking system was designed to rank the simulations according to their performance in reproducing different aspects of rainfall characteristics. The results suggest that the simulation with the MIT Emanuel convective scheme and the BATS1e air-sea flux scheme performs better overall compared to the rest of the simulations.