Consideration of Applicability of Stochastic Tropical Cyclone Model for Probability Assessment of Storm Surge

Storm surge risk is basic and important information for design of coastal structure, but its probability assessment is difficult because the frequency of devastating TC is not so high. Furthermore, storm surge damage is very sensitive to not only its intensity but also its track and translation speed. Therefore, the estimation of occurrence probability of devastating disaster from our limited observation data has uncertainty. Recently, we have developed Global Stochastic Tropical cyclone Model (GSTM). The comparison of observation data and simulation results of GSTM showed the reasonable reproducibility of macroscopic statistics of TC parameters such as central pressure, translation direction and speed. However, the verification area of previous study was too large to consider the applicability to risk assessment of local bay scale. Therefore, the microscopic and detailed verification of GSTM is necessary. In this study, we tried to evaluation of reproducibility of TC parameters which was calculated by GSTM at small region as a local bay scale. Furthermore, we improved GSTM by implementation of the cluster analysis of observation data to the process of estimation of the joint Probability Density Function (joint PDF) of temporal correlation of TC parameters. At last of this study, a case study approach by storm surge simulation was performed in order to explain the practical meaning of GSTM. Synthetic TC data based on historical TC track was generated and they were used for input to numerical model for estimating the water height of storm surge at regional scale. Then, the TC track of the worst-case scenario for Yatsushiro bay located in center of west Kyushu Island was decided. Finally, the occurrence probability of the worst-case scenario was calculated from GSTM results.