Application of Neural Networks to the Prediction of Significant Wave Height at Selected Locations on the Baltic Sea

Abstract

This paper describes the application of methodology based on the artificial neural network technique to make short-term wave forecasts. The neural network model is used to predict significant wave height at a selected location on the Baltic Sea based on wave and/or wind data at ten points scattered on the sea. High quality hindcast data were used in the process of developing the forecast methodology. The data originated from the WAM4 wave model. The results show that the neural network technique allowed significant wave height to be predicted accurately. The agreement obtained by a comparison with a testing data set was sufficiently good to confirm the effectiveness of this approach.