Performance of general circulation models and their ensembles for the prediction of drought indices over India during summer monsoon

The drought during the months of June to September (JJAS) results in significant deficiency in the annual rainfall and affects the hydrological planning, disaster management, and the agriculture sector of India. Advance information on drought characteristics over the space may help in risk assessment over the country. This issue motivated the present study which deals with the prediction of drought during JJAS through standardized precipitation index (SPI) using nine general circulation models (GCM) product. Among these GCMs, three are the atmospheric and six are atmosphere–ocean coupled models. The performance of these GCM’s predicted SPI is examined against the observed SPI for the time period of 1982–2010. After a rigorous analysis, it can be concluded that the skill of prediction by GCM is not satisfactory, whereas the ability of the coupled models is better than the atmospheric models. An attempt has been made to improve the accuracy of predicted SPI using two different multi-model ensemble (MME) schemes, viz., arithmetic mean and weighted mean using singular value decomposition-based multiple linear regressions (SVD-MLR) of GCMs. It is found that among these MME techniques, SVD-MLR-based MME has more skill as compared to simple MME as well as individual GCMs.