Analysis and prediction of short period droughts during summer monsoon over India

Abstract

Climatic characteristics of occurrence of dry spells over 7 meteorological subdivisions lying in semi-arid parts of India are studied by using 23 years of spatially averaged sub-divisional rainfall data. Five different thresholds have been used to decide the occurrence of dry or wet day over the sub-divisions. Similar characteristics in occurrence of dry spells have been studied for 6 stations lying in the same area. The Markov chains of order 1 and 2 have been fitted to the frequency of occurrence of dry spells over the stations and on testing the chain of order 2 are found to fit better.

Composite charts of probability of occurrence of rain around depressions located in various geographical areas have been prepared by using daily rainfall data of 220 stations for 19 years. With the help of these charts the role of depressions in alleviating drought situation is examined.

Finally multiple linear regression equations are developed by stepwise screening method to predict the daily rainfall amounts over these subdivisions by using large scale synoptic parameters. When  tested on an independent sample of equal size these equations showed skill scores varying from .22 to .54 in forecasting the terciles of rainfall.