Wave Condition Measured at an Offshore Tower and Wave Prediction by Using XGBoost

The observation site is an offshore tower for the oceanographic and meteorological observation belonging to Shirahama Oceanographic Observatory (SOO), Disaster Prevention Research Institute (DPRI), Kyoto University. High wave conditions in summer and winter seasons are caused by typhoons, strong low pressures and strong monsoon wind. High wave conditions described above are clearly obstacles to maritime traffic and offshore operation as well as on-site work at the observation site. Therefore, wave prediction information is essential for security and reliability for maritime traffic, offshore operation and the planning of on-site work. To obtain wave prediction information, regression trees of XGBoost are used to predict the significant wave height at the observation site by using the forecast data of the Grid Point Value (GPV) data of coastal wave numerical forecast model (CWM). The simulated results by trained model have a good agreement with the observed data (R squared is around 0.8. MAE is around 0.13 m). The forecasted data of GPV-CWM are used to predict wave height, and the predicted results are compared with the observed data. Mean and median values of wave height difference are small enough, and 75 percentile of wave height difference is around 0.20 m in the first day of GPV-CWM data.