Flood frequency analysis using PL-moments approach

Abstract

Estimation of flood magnitude is a crucial component in planning, designing, and managing of water resources projects. Flood frequency analysis (FFA) provides a practical means of determining a robust probability distribution that fits streamflow data at a location of interest. The main focus in hydrology design is the estimation of high flow quantile. L-moments, popular among hydrologist in FFA is known to be oversensitive towards the lower part of the distribution and give insufficient weight to large sample values. As an alternative, the method of partial L-moments (PLmoments) is proposed to give weightage to the upper part of distribution and large values in censored sample. The aim of this study is to compare the performance of PL-moments and L-moments in FFA. The method of PL-moments was developed for generalized extreme value (GEV), generalized logistic (GLO), generalized pareto (GPA), extreme value type 1 (EV1) and logistic (LOG) distributions. Monte Carlo simulations from population distributions of known and unknown samples were conducted to assess the performance of PL-moments compared to L-moments. Simulation results showed that PL-moments give comparable and slightly better parameter estimates than those by L-moments particularly when estimating the high flow quantiles. In regional flood frequency analysis, new statistical tests based on PL-moments were developed to measure discordancy, regional homogeneity and identify a best regional distribution. The quantile estimates based on the regional distribution using PL-moments are more efficient than L-moments in estimating flood quantiles at higher return periods. The overall results strongly support that PLmoments method would improve the flood quantiles estimation particularly for higher quantiles and thus serves as a useful tool for application in flood frequency analysis.