ASSESSMENT OF UNCERTAINTY IN ESTIMATING FUTURE EXTREME STORM SURGE EVENTS IN OSAKA BAY USING LARGE ENSEMBLE TYPHOON DATA

Abstract

Risk assessment of low-frequency catastrophic events requires observation/simulation data of a large number of disaster events. When we consider the risk of storm surge, high waves, and strong wind caused by tropical cyclone, the way of stochastic tropical cyclone model approach has been conducted. The stochastic model can create a lot of synthetic tropical cyclone track data which has same statistic characteristics with historical data got in short period. In addition, climate change impact assessment requires comparison of many ensemble results (models and scenarios) in order to consider the uncertainty of future projections. However, there are few proposals for stochastic models that can predict the future climate change effects. Even if a large amount of tropical cyclone track data is produced, it is practically impossible to simulate disasters for all of them. For this reason, methods for calculating storm surge and storm waves using ANN models and regression models have been proposed, but the validity of data screening using such simplified methods needs to be further investigated. This study examines the valid ity of the newly developed regression model and its applicability to the screening of hazardous storm surge events. Future changes in storm surge risk in Osaka Bay are analyzed by combining the regression model results with detailed numerical methods.