The statistical bias correction of projected precipitation for the extreme events

Abstract:

In this study, we suggest the new method to correct the bias of projected daily precipitation using two probability distributions, which are Gamma and generalized extreme value (GEV) distribution, and spline interpolations. The normal bias correction methods are usually focused on the average annual precipitation. However, the needs of extreme events analysis in climate change are increasingly growing to prevent the future disaster, such as flooding. We conduct the bias corrections of daily precipitation with considering the extreme cases of projected daily precipitation and also maintaining the characteristics of observed annual precipitation. The observed daily precipitation data at four sites are examined to estimate the effects of the new bias correction method for the extreme events. The power of this method is checked with the observed daily precipitation from 1979 to 2005 at Seoul, Daejeon, Daegu and Busan in Korea by the Q-Q plots and statistical diagnostic tools before applying to the projected precipitation. And then this method is applied to estimate the bias of projected precipitation and compared to the results of the previous studies, such as linear scaling and quantile mapping with Gamma distribution. We found that this method can preserve the characteristics of annual precipitation and estimate the appropriate extreme quantiles. These results illustrate the need to consider this method in the bias correction of projected precipitation in extreme analysis, which is very attractive for designing water-related infrastructure in response to future climate change.