UNCERTAINTY EVALUATION OF RIVER DISCHARGE USING MULTIVARIATE EXTREME VALUE ANALYSIS

Abstract

 This study presents a method for constructing confidence intervals for multivariate extreme value distributions based on probability limit method test, and multivariate extreme value analysis method that introduces such confidence intervals. The method can estimate the peak flood discharge values and their confidence limits under the rainfall of interest by introducing confidence intervals into the multivariate extreme value analysis, which is useful for the problem of the large uncertainty of various possible spatiotemporal distributions of river discharge in rainfall data that have not been accumulated for several decades. The results are shown below. Furthermore, as an application of this method, it is possible to quantitatively discuss flood risks related to rainfall and peak discharge under a future climate by using a large ensemble of data.