GMDH-based wave prediction model for one-week nearshore waves using one-week forecasted global wave data

Abstract

Wave forecasting up to 168 h (one-week) ahead is crucial to the operational management of shipping and construction in the marine work site. The present study proposes a method of areal-time One-week Wave Forecast of Nearshore Waves (OWFNW) at 13 stations on the Japanese coast. For each station, two Group Method of Data Handling (GMDH) based wave prediction models for the wave height and period are developed to transform the global wave data into nearshore waves. For each GMDH-based wave prediction model, four models are trained with four different training data sets using the one-week global wave forecast data released in real-time by the Japan Meteorological Agency (JMA), National Oceanic and Atmospheric Administration (NOAA), and European center for Medium-Range Weather Forecasts (ECMWF). The study conducts a series of training and testing GMDH-based wave prediction models at each station, and examines the accuracy of predictions. The best performing GMDH-based wave prediction model selected among the four models is compared to the best global wave forecast data. It was found that the best performance GMDH-based wave prediction model varies for different locations. On average, the best performing GMDH-based wave height model is able to improve the accuracy of nearshore wave predictions up to 38% compared to that of the best global wave height forecast with 168 h lead time. The best GMDH-based wave period model improves the prediction error up to 65% compared to the 168 h ahead error of the best global wave period forecasts. The present methodology to get OWFNW can be applied to other nearshore seas where global wave forecast data and corresponding observed one are available.