Application of statistical methods for the prediction of extreme wave events

A statistically-based model is applied to forecast sea states for severe storms. The model is based on the application of a neural network and predicts extreme sea state parameters at specified locations. The results show that the neural network can be applied to forecast extreme sea state parameters. This requires a special treatment of the input data. The analysis shows that different types of input data and training data sets should be considered and the representativity of the training data set must be improved. Moreover, a sensitivity analysis should be conducted to remove excess information from the input data. The processing of data sets significantly reduces the number of parameters applied in the model and improves the prediction for most severe storms. The analysis indicates that this neural network model may be helpful in the selection of a measurement system for the forecasting of extreme sea state parameters. This is important because typical installations of wave buoys along the coast have a limited forecasting applicability range.