Frequency Analysis of Extreme Precipitation in Indian River Basins

Abstract

Precipitation is one of the most important variables in the hydrological cycle which affects floods, droughts, ecosystem services etc. Information regarding precipitation extremes in the water sector to identify climate change consequences and understanding the hydrology of an area both need knowledge of water availability over time and space. Because of increasing/decreasing tendencies in hydro meteorological time series, such as droughts, floods, and so on, climate change is projected to have a significant effect on rainfall and runoff. Because of the high unpredictability of rainfall over time, irrigation water supply and flood control systems are designed and managed based on precise rainfall depths that may be predicted for a given probability or return period rather than a long-term mean of rainfall data. Many researchers have recently employed several parametric and statistical methods to analyse extreme precipitations in hydro meteorological time series in India and around the globe. For analysis of extreme rainfalls, parametric distributions that are generally used include GEV-1, Gumbel's, Log Pearson - III and estimation of their parameters with the most precision is found to play an important role. Method of moments, Maximum Likelihood Estimation, L-moments are one of the few methods which are found to be working suitably for parametric estimation of distribution functions. In this study, Ganga River basin and Narmada River basin have been chosen as study area for the extreme precipitation analysis. The parametric estimation methods that are compared for the accuracy and reliability include, Method of Moments (MOM) and Maximum Likelihood Estimation (MLE). The distribution functions that have been used to fit the basin data and predict the precipitations for fix return periods are Gumbel's Distribution and Log-Pearson - III. The time series data set that is used to perform the methods are received from IMD and CHIRPS to make a final comparison between different distribution functions, methods of parameter estimations and data set precision as well.