Peninsular Indian rainfall and its association with meteorological and oceanic parameters over the Indian Ocean

The present study attempts to formulate a regression model to predict summer rainfall over Peninsular India (PIR) using some regional predictors. Parameters having significant correlation (99%) with PIR were identified for the period 1975–1997 (training), and a 15-year sliding correlation (90%) was found to check the consistency of the relationship between PIR and predictors. From a set of 14 candidate predictors, 4 were selected using a stepwise regression method and tested over a period from 1998 to 2006. The predictors selected are sea surface temperature during March over Indian Ocean, air temperature at 850?hPa during May over Peninsular India, zonal, and meridional wind at 700?hPa during February and January, respectively, over the Arabian Sea. The model captures a variance of 77.7% and has a multiple correlation of 0.88. The root mean square error, absolute mean error, and bias for the training (test) period were 7.6% (21.5%), 6.6% (17.9%), and 0% (11.4%) of mean rainfall, respectively. Results of the climatological predictions show that the model developed is useful.