STUDY ON THE EXTREME STORM SURGE INDUCING TYPHOON EXTRACTION METHOD AND FUTURE STORM SURGE RISK CHANGES IN OSAKA BAY

Abstract

 A regression model is proposed to predict the maximum storm surge height at representative locations in Osaka Bay for a large amount of typhoon track data generated by a stochastic typhoon model. First, the accuracy of the regression model was evaluated for individual tropical cyclone scenario. Second, the applicability of the regression model as a data screening method, pre-processing of detailed analysis using a physical model, was examined. Then the applicability of the regression model to a wide range of risk assessment from low frequency catastrophic storm surges to normal one was clarified.

 The data screening of tropical cyclone using a regression model of storm surge was applied to the results of a stochastic typhoon model for the past reproduction and future prediction, and the model ensemble mean future change of 1/50-1/1000 probability values and the range of prediction uncertainty at representative sites in Osaka Bay were shown. The distribution of 1/500 probability values of maximum storm surge height in Osaka Bay shows that the distribution pattern is similar between some models, although the trend of future change differs significantly among models.