Probability Analysis for Prediction of Rainfall for Tamil Nadu Agricultural University, Coimbatore

Abstract

A study was undertaken predict the rainfall by probability analysis for Tamil Nadu Agricultural University (TNAU), Coimbatore. Rainfall data of 30 years (1974 ~ 2003) was collected from local meteorological station of TNAU. Owing to the necessary of the water harvesting, measures, storm water drains and crop planning activities in and around this rain shadow area, the design of such measures envisage the reliable estimates of various rainfall types at different frequencies which are not currently available. The one day maximum, average weekly, average monthly and cumulative seasonal rainfall data were analyzed by fitting different probability distributions like Normal, Log-normal, Gumbel, Log-Pearson Type Il, Gamma and Exponential with the reference to minimum D-Index. The rainfall data had been fitted in the above distributions and their corresponding rainfall events were estimated at 20, 40, 60 and 80 per cent probabilities of exceedence. From the rain fall analysis, it was inferred that Gumbel followed by Normal distribution were identified as the best fit for one day maximum rainfall; Gumbel followed by Exponential, and Normal distribution were identified as the best fit for average weekly rainfall; Gumbel followed by Normal, Exponential and then Log-Pearson type III distribution seemed to be identified the best fit for average monthly rainfall. In case of cumulative seasonal rainfall, it was observed that Gumbel, Log-normal, Log Pearson III and Normal distributions were fitted the best for the chosen area in the respective sequence.