NEW REGRESSION-MODEL FOR INDIAN-SUMMER MONSOON RAINFALL

Based on the study of 45 years (1948-1992) data, the average lowest MS L pressure of heat low over central Pakistan and adjoining northwest I ndia for the month of May is found to have potential as a parameter fo r predicting all India Summer monsoon seasonal rainfall. This new para meter is seen to have stable and significant correlation with monsoon rainfall. Its correlation coefficients for different periods are found significant at 0.1% to 1% level of significance. The stability of the correlation coefficients was tested using 10, 20 and 30 year sliding windows. This test revealed that it is the most dependable parameter i n comparison with 7 of the well known parameters analysed in this stud y. Regression models have been developed considering this new paramete r along with other circulation paramters. The regression models develo ped are seen to perform very well for the independent data. The Root M ean Square Error (RMSE) values of some of these models, for independen t data, are smaller than those of similar regression models reported i n literature.