Application of regional frequency analysis to non-stationary rainfalls in Korea

Abstract

According to the research using GCM, it is predicted the intensity of storm becomes stronger than ever in most of the area in the Northern hemisphere. Conventional frequency analysis assumes that rainfall series is stationary, however, in practice, all of the rainfall series are not stationary. This assumption could misguide result of frequency analysis. Several studies have tried to consider nonstationarity to regional frequency analysis. Regional frequency analysis which can complement the problem of lack of the number of observations is known as more stable method than at-site frequency analysis. This study selected the rain gauge stations whose rainfall series shows non-stationarity and stations are geometrically close to the stations. In order to consider non-stationarity of rainfalls, this study applied high correlation between parameters and mean of cumulative annual maximum rainfall to process of parameter estimation. Based on this, probability distribution of the specific point in the future can be estimated. © 2010 ASCE.