Returning period of nonstationary extreme precipitation under climate change

It is belived that climate change since the industrial period has caused more extreme climate events and natural disasters, especially floods, which have caused a large number of casualties and economic losses. Many studies have already analyzed the changes of extreme precipitation over the past decades and predicted future changes, most of which suggest extreme events may become more sever and frequently. Such nonstaionary changes modifies the returning period of extreme precipitation, while influences possess strong spatial heterogeinity. Therefore, the threshold of the extreme event and returning period should be redefined to reflect nonstaionry changes due to warming. In this study, we use both observed historical data and future precipitation projections from CMIP5 GCMs under four RCP scenarios, to investigate the variations in the changing magnitude and frequencies of the annual maximum 1-day, 3-day, and 5-day precipitation. The spatial GEV distribution and the peak-over-threshold (POT) method to calculate the extreme precipitation frequency by the threshold maps of heavy precipitation (once-in-decade and once-in-century). Our results show that some low-latitude areas tend to suffer both more frequency and heavier precipitation events, while over a global scale, the proportion of small mount of precipitation to total may decrease and heavy precipitation increase in the future. Seen from the distribution of annual maximums, the magnitude of those ranging from 10- to 100- year increase more significantly than those ranging over 100-year events in most humid areas under both RCP 2.6 and RCP 8.5 scenarios.