Optimization of typhoon-wave parameterization schemes using the SWAN model coupled with a genetic algorithm

To address the problems of high parameter sensitivity and large simulation deviation of the SWAN (Simulating Waves Nearshore) model under extreme typhoon conditions, this study adopts the Genetic Algorithm (GA) to optimize its key physical parameters. Taking the significant wave height and mean wave period during two typhoon events in the South China Sea as research objects, the numerical simulation results are compared with buoy observation data. The results show that the GA-optimized parameters can improve the simulation accuracy and alleviate the systematic overestimation seen in the default parameter setup for the two studied typhoon cases. The optimized scheme achieves satisfactory overall simulation performance for significant wave height and can well reproduce the evolution process and extreme characteristics of typhoon waves. Nevertheless, the simulation accuracy of mean wave period is relatively lower than that of significant wave height, which may be partly explained by the relatively high sensitivity of mean wave period to the fine evolution of wave spectral structure, nonlinear wave interactions and non-stationary typhoon wave conditions. The optimal parameter combination achieves the most balanced comprehensive performance under the weighted fitness function defined in this work and outperforms other schemes in terms of significant wave height, the core engineering indicator. The research findings can provide a reference for parameter selection in typhoon wave numerical simulation and numerical forecast for disaster prevention and mitigation in marine engineering.