FLOOD FREQUENCY STUDIES USING L-MOMENTS

Abstract

The most important aspect of flood frequency analysis is the estimation of parameters. The method of moments, the oldest and most widely understood parameters estimation technique, exhibits remarkable bias perticularly with skewed and small size samples. Recently new parameter estimation technique called L-moments has emerged which, being linear combination of sample data, is able to yield- relatively unbiased estimates of parameters. The use of L-moments has found significant application in the area of flood frequency analysis. In the present study L-moments technique has been used to estimate parameters of various distributions such as Extreme Value Type I, GEV, - Logistic, Generalised Logistic, Generalised Pareto, Normal and Log Normal for the different sites of subzone 3(c). Goodness of fit of various distribution has -also been evaluated based on various criteria such as efficiency and L-moment ratios etc. Regional flood frequency methods require homogeneity of the area under consideration. Generally more homogeneous a region is greater is the gain in regional flood frequency analysis. Various homogeneity tests such as discordancy measure and X-10 test based on L-moments, Original Dalrymple's test, Revised Dalrymple's, test, Cv of Cv test and MOM-Cv test are studied and these tests are applied to judge the homogeneity of region subzone 3(c). Regional flood frequency analysis has been carried out using modified USGS method, NERC method and method based on standardised L-moments for EV1, GEV, Logistic, Generalised Logistic, Generalised Pareto, Normal and Log Normal distributions for the region subzone 3(c). L-moment ratios have significant application in goodness of fit measure for the regional distributions. For subzone 3(c) L-moment ratios diagram, a very effective and simple tool for the identification of regional distribution, clearly indicate the Generalised Pareto distribution being most appropriate regional distribution. L-moments are very useful in all aspects of RFFA, be it homogeneity, analysis or goodness of fit measure and present study cover all these aspects. (