A Surrogate Model for Wave Prediction Based on an Artificial Neural Network Using High-fidelity Synthetic Hurricane Modelling

Mun, J.; Kim, S.-W., and Melby, J. A. 2018. A surrogate model for wave prediction based on an artificial neural network using high-fidelity synthetic hurricane modelling. In: Shim, J.-S.; Chun, I., and Lim, H.S. (eds.), Proceedings from the International Coastal Symposium (ICS) 2018 (Busan, Republic of Korea). Journal of Coastal Research, Special Issue No. 85, pp. 876–880. Coconut Creek (Florida), ISSN 0749-0208. A surrogate model for significant wave height and period prediction is developed based on an artificial neural network with synthetic simulations of hurricanes. The neural networks between five input parameters and the wave height or wave period are trained by the Levenberg-Marquardt backpropagation algorithm. A back propagation is a form of supervising training algorithm and the weighting factors and the bias are updated by the back propagation training within a limited epoch. The data consists high intensity 446 storms on west region in Texas. The 14 save points in this area are analysed by the surrogate model. The performance of the static surrogate model for the significant waves is expressed as the correlation coefficient and the mean square error at the 14 save points. The minimum correlation coefficient and the mean square error are respectively 0.981 and 0.011. The predicted wave height and period are good agreement with the target values.