Development of a typhoon wave overtopping prediction model based on an artificial neural network

Abstract

In this study, an artificial neural network (ANN) model for the typhoon wave overtopping was developed using the database by a numerical wave flume simulation. The developed ANN model is effective for saving calculation time largely. The accuracy of the model is also approached to over 95% of the numerical simulation. This accuracy was evaluated by the correlation coefficient and the root mean square error with the target data of the numerical simulation and output of the ANN model. This model quickly produces the mean wave overtopping rate, maximum wave run-up height, maximum wave overtopping depth and velocity at the middle point in the coastal road without high-fidelity numerical model and high-computing resources. It means that the typhoon warning system including the ANN models is powerful and useful rather than only the monitoring warning system currently in use.