POWER OF TEST OF THE EVALUATION METHODOLOGY FOR NON-STATIONARITY IN EXTREME HYDROLOGICAL TIME SERIES BASED ON PROBABILITY LIMIT METHOD

Abstract

 It is necessary to develop adaptation measures that take the nonstationarity of extreme rainfall into account. Therefore, we developed a methodology to evaluate nonstationarity in extreme rainfall time series data using the probability limit method test. In this study, we compared the testing power of this methodology with other hypothesis test theories. The results indicated that the probability limit method can estimate the degree of nonstationarity in extreme rainfall time series data more accurately than other methods. Additionally, applying this method to the annual maximum daily rainfall time series observed in Japan revealed a nationwide increasing trend in nonstationarity. Furthermore, it was shown that the rate of transition to nonstationarity increases over time, particularly in the Shikoku, Tohoku, Kinki, and Kyushu regions.