CHARACTERISTIC OF DESIGN FLOODS USING L‐MOMENT BASED ON GRADUAL COMPOSITION OF ANNUAL MAXIMUM FLOOD FLOW

Abstract

ABSTRACTIn this study, the characteristics of design floods in Korea were analyzed by selecting 19 watershed regions out of all the water level gauging stations in Korea. In the selected watershed regions, the observed floods were verified and used to derive appropriate design floods. To perform a frequency analysis of the annual maximum flood based on the gradual composition method, in which the maximum flood of each target watershed is increased by one year from the year when the observation began, with 1990 as the reference year, the variation characteristics of the annual flood were analyzed using the moving average method. The basic statistics of the 19 target watersheds were also evaluated based on the composition period of the annual maximum flood series, and the independent, homogeneity and outlier tests were performed. Furthermore, the goodness of fit of the Gumbel (GUM), Pearson type 3 (PT3), generalized extreme value (GEV), generalized logistic (GLO), generalized Pareto (GPA), and lognormal (LN3) distributions were verified based on the L‐moment ratio diagram and the Kolmogorov–Smirnov test. Finally, the parameter of the GEV distribution, which was selected as the appropriate probability distribution, was estimated based on the L‐moment, and the design flood of each watershed was derived from the composition period of the annual maximum flood series. According to the variation rate analysis used in this study, the period when the design conditions for an appropriate agricultural hydraulic structure should be changed must be around the year 2002, considering the recent climate changes in Korea due to global warming. Copyright © 2012 John Wiley & Sons, Ltd.