Characteristics of the Indian Summer Monsoon Rainfall Simulated by the Regional Climate Model (RegCM4)

Abstract

The regional climate model (version RegCM4) is used in this study to simulate the Indian summer monsoon rainfall. Model integrations were carried out for 16 monsoon seasons from 1985 to 2000. The model-simulated rainfall at a daily scale as well the seasonal mean were compared with observed rainfall. It was observed that the model underestimates the rainfall amount, especially over the eastern and central parts of India. Frequency distributions of daily rainfall showed that there were more days with light rainfall and fewer days with moderate rainfall in the model as compared to the observation over most parts of India. The ability of the RegCM4 to simulate interannual variability in rainfall over two sub-regions of India (one in western parts and the other in eastern India) was examined for the months of July and August. Composite analyses of rainfall, winds, moisture and evaporation for excess and deficient rainfall years for these two regions were carried out and compared with similar analysis using ERA-Interim data. When the eastern parts of India experienced more rainfall, the cyclonic circulation over Odisha was more intense, with stronger easterlies over the Gangetic plains, enhancing moisture transport to the Indian land mass from the adjoining seas. A cyclonic circulation difference between excess and deficient years over this region and an anticyclonic circulation difference over peninsular India led to the development of a strong convergence zone over parts of Maharashtra and Gujarat in the excess years. Several sensitivity experiments were also carried out using various schemes, domain sizes and diffusion parameters to study the role of these parameters in defining the model bias. The key parameters that emerged from these experiments as affecting the moisture convergence and rainfall bias over the Indian region were the horizontal diffusion and auto-conversion of cloud water into precipitation. However, these parameters were not able to improve the simulation of interannual variability over most parts of India.