Rainfall frequency analysis using generalized extreme value case study Upper Citarum River Basin

Abstract

Design storm analysis has a significant influence for flood discharge analysis. Design strom derived from historical rainfall data with statistical approach using probability distributions. Generalized Extreme Value (GEV) probability distribution is a distribution which generalizes extreme value theorem type 1, 2, and 3. Generalized Extreme Value Distribution have 3 parameters, shape parameter k, scale parameter a, and location parameter ?, which influence design storm analysis result. To know the influence of those three parameters and the effect of using GEV in Design storm analysis compared to other probability distributions, then Upper Citarum River Basin is used as a case study. Sensitivity analysis for all three GEV parameters show that the shape parameter gives the most significant influence for design storm analysis. The lower shape parameters value is, the greater frequency analysis result will increase exponentially. Rainfall-runoff model result with design storm calculated using GEV is compared with the result using Pearson III and Gumbel for design storm. From these models it was found that the flood discharge generated from design storm calculated using GEV is greater than that of Pearson III and lower than Gumbel. Flood discharge from Upper Citarum River Basin using GEV design storm give exponentially increasing discharge value proportional to its return period. Design storm increase caused by GEV design storm usage for 50 year, 1000 year, 500 year, 1000 year return periods are 2.76%, 3.87%, 6.80%, dan 8.25%, which caused 4.19%, 5.89%, 10.6%, 12.95% increase in flood discharge, respectively