Abstract

This study deals with two statistical downscaling (SD) techniques for their potential to improve the skill of forecasts seasonal rainfall over core monsoonal belt regions over South Asia. It constitute most comprehensive to date intercomparison of SD methods. This work uses Indian Summer Monsoon (ISMR) forecasts from global primary models during 1996 to 2010. Two popular downscaling methods were tested by using unequal weight to all studied models which is termed as Best Linear Unbiased Estimator (BLUE) or by equal weight known as Ensemble Mean (EM). The statistical downscaling methods Bias Correction and Spatial Downscaling (BCSD) and Bias Correction and Constructed Analogue (BCCA) from Model Output statistics (MOS) approach were used. The analysis of direct models forecasts reveals that the bias is high with respect to season across South Asia, hence the necessity of SD. Though the skill of SD methods varies strongly depending on regions and seasons considered, both BCCA and BCSD also reveal higher performance than any primary models. The BCCA SD models are most capable to forecasts ISMR even at local scale than any others.

Author: Aminuddin Ali

Received on: August, 2021

Accepted on: March, 2022