Regional frequency analysis of streamflow based on partial L-moments approach

Abstract

An attempt has been made to model the annual maximum streamflow in the East Coast of Peninsular Malaysia, utilizing the guidelines in regional flood frequency analysis. The L-moments and partial L-moments (PL-moments) at several censoring levels are employed to estimate the regional parameters of three extreme value distributions, namely, generalized extreme value (GEV), generalized logistic (GLO) and generalized Pareto (GPA) distributions. A total number of 18 streamflow stations located throughout the eastern region of Peninsular Malaysia are used as case study. The performances of the L-moments and PL-moments methods and their corresponding distribution functions are compared using Monte Carlo simulation. The results of relative root mean square error (RRMSE) and relative bias (RBIAS) show that the GLO distribution of PL-moments at censoring level 0.1 is appropriate to model the regional streamflow data in East Coast of Peninsular Malaysia compared to L-moments. The overall simulation results indicated that, in some situation, the PL-moments method improves the streamflow quantile prediction and provide useful tools for application in regional flood frequency analysis.