Author(s): Bi Junfang; Sun Weikang; Luo Zheng
Linked Author(s):
Keywords: Changjiang Estuary; Storm surge; Numerical model; Machine learning; SVM
Abstract: This study establishes two storm surge forecasting models in Changjiang Estuary, one is hydrodynamic numerical model based on FVCOM, and the other is SVM regression model based on data-driven machine learning. The two models are used to simulate the storm surge process in Yanglin station which is in the south branch of Changjiang Estuary during the Typhoon Muifa No. 202212, The results show that the two models’ predictions are qualified at 24h storm surge forecasting, both methods can meet the actual operational requirements of disaster mitigation in Changjiang Estuary. The coefficient of determination of the two models are 0.96 and 0.95, and the prediction accuracys of the two models are good, the SVM regression model can also be used as one of the tools for storm surge forecasting in the absence of boundary conditions, terrain data.
Year: 2024