Simulation and Prediction of Machine Learning-Based Storm Surge Forecasting Model

Abstract

Forecasts of storm surges are essential to mitigate the damage of extreme events such as land shift, wave height and wave periods. Numerical models that simulate nearshore hydrodynamics and morphodynamics, such as XBeach, have been significant tools for storm impact prediction. However, AI techniques, including ensemble models, deep learning (DL) models, and machine learning (ML) models proved a significant and good accuracy in predicting storm impacts. This study systematically simulate the performance of the numerical model and AI models to forecast storm consequences. The results highlight that each model type numerical or AI has a unique combination of benefits and drawbacks, that gives rise to the urgent need for reliable predictive models as storm intensities and patterns are further altered by climate change. The study also implied that the most effective course of action is to combine both AI driven approach models with enough datasets and algorithms to produce a hybrid approach.