Detection of Abrupt Changes in Precipitation Extremes over South Korea Using a Bayesian Approach

Abstract

Change point (CP) analysis of extreme rainfall plays a key role to consider non-stationarity in predicting flood or drought under climate change. This study provided a Bayesian framework to detect the existence of the CP in extreme rainfalls. Unlike most published works assuming a normal distribution, it allows for the model to use a generalized Pareto distribution (GPD) to fit the extreme rainfall over a high threshold with a CP. The proposed approach was applied to the extreme rainfall data from five selected stations in South Korea. Results indicated that the employed methodology can precisely capture the CP existed in GPD.