Comparative Analysis of Extreme Value Distributions for Flood Risk Assessment in Kelantan Rivers using L-Moments

Abstract

Floods are a recurring natural disaster in Kelantan, Malaysia, posing severe risks to communities and infrastructure. This study presents a comparative flood frequency analysis (FFA) using the L-moment method at the Rantau Panjang station along the Golok River to determine the most suitable extreme value distribution for modeling peak flow data. Three distributions were analyzed: the Three-Parameter Log-Normal (LN3), Generalized Logistic (GLO), and Generalized Pareto (GPA). The research evaluated each model based on statistical performance metrics, L-moment ratio diagrams, and goodness-of-fit tests (Kolmogorov-Smirnov and Anderson-Darling). The GLO distribution emerged as the most accurate, yielding the lowest error values and highest R², and closely aligned with observed data in the L-moment diagram. Its superior performance in estimating return periods and flood magnitudes confirms its reliability for risk assessment in flood-prone regions. This study demonstrates the efficacy of L-moments in enhancing hydrological modeling, offering vital insights for disaster preparedness and water resource management in Kelantan.