OPTIMIZATION OF THE RAIN-GAGES FOR A REPRESENTATIVE ALL-INDIA AND SUBDIVISIONAL RAINFALL SERIES

An objective approach similar to the forward selection of independent variables in the multiple linear regression has been attempted to opti mize the network of raingauges for the summer monsoon rainfall (June-S eptember total) series (1871-1984) of India as well as its 29 selected meteorological subdivisions prepared involving the data of 306 rainga uges. For the all-India monsoon rainfall series twenty seven gauges en tered the selection whose mean showed the correlation coefficient (CC) of 0.9869. Keeping in view the difficulties of getting data from all the 306 gauges, 35 India Meteorological Department (IMD) gauges with m ean showing CC of 0.9898 have been identified for updating this series . The constructed all-India monsoon rainfall series for the period 187 1-1992 using 35 selected observations is presented. It was interesting to note that the set of 35 gauges selected for the monsoon total has shown equally promising results for the all-India monsoon monthly (Jun e-September) as well as the annual rainfall series. For the 29 subdivi sional monsoon rainfall series, however, in total 188 IMD-gauges (62% of the total of 306 gauges) entered the selection. For 17 subdivisions the CC exceeded 0.98, for 3 subdivisions it varied between 0.97 and 0 .98, for 5 subdivisions between 0.96 and 0.97 and for the remaining 4 subdivisions between 0.90 and 0.94. They showed equally encouraging re sults for the monsoon monthly and annual rainfall series for the diffe rent subdivisions. Limitations and implications of the optimization te chnique are also briefly discussed.