2021 Volume 77 Issue 2 Pages I_139-I_144
In order to improve the accuracy of storm surge prediction in real time, it is necessary to improve the accuracy of typhoon prediction. For this purpose, it is important to properly set the calculation conditions of meteorological models, which have often been determined empirically. In this study, we developed a system using machine learning to automatically suggest combinations of physics options of the WRF to achieve the highest prediction accuracy for newly generated typhoons. As a result, we confirmed the improvement of typhoon forecasting accuracy by using this method, which selects physics options of WRF based on typhoon characteristics such as the position of typhoon occurrence, the maximum diameter of the wind field and so on. In addition, by using the typhoon prediction results as the external force for storm surge estimation, the prediction accuracy of maximum storm surge was improved on average in many locations, and in some locations the accuracy was significantly improved, indicating the effectiveness of our method, although further verification is required.
PROCEEDINGS OF COASTAL ENGINEERING, JSCE
PROCEEDINGS OF THE JAPANESE CONFERENCE ON COASTAL ENGINEERING
