NON-STATIONARY HYDROLOGIC FREQUENCY ANALYSYS FOR ESTIMATION OF EXTREME PRECIPITATION CHANGE USING GLOBAL WARMING PROJECTION INFORMATION

Abstract

A non-stationary hydrologic frequency analysis method was applied for estimating future extreme precipitation change using JMA global warming projection information vol. 8 (2013). The method uses distribution functions with time dependent model parameters, which were estimated using the method of maximum likelihood. The best fitted hydrologic frequency model was selected from a non-stationary GEV model, a non-stationary Gumbel model, a non-stationary SQRT-ET model and a non-stationary lognormal model according to Takeuchi Information Criterion, TIC. It was found that the 100-years annual maximum 24 hours precipitation shows increasing tendency in most of Japanese islands; and it becomes larger when using bias corrected precipitation data.