To eliminate errors caused by uncertainty of parameters and further improve capability of storm surge forecasting, the variational data assimilation method is applied to the storm surge model based on unstructured grid with high spatial resolution. The method can effectively improve the forecasting accuracy of storm surge induced by typhoon through controlling wind drag force coefficient parameter. The model is first theoretically validated with synthetic data. Then, the real storm surge process induced by the TC 0515 typhoon is forecasted by the variational data assimilation model, and results show the feasibility of practical application.