MACHINE LEARNING AND DEEP LEARNING FOR ENHANCED SPATIO-TEMPORAL WAVE PARAMETERS PREDICTION

Traditional methods of wave prediction, which are mainly reliant on extensive numerical simulations, such as the utilization of spectral wave models SWAN, WaveWatch III, or TOMAWAC, have prompted the question: Can faster wave prediction be achieved? The answer, as demonstrated by this study, lies in the advancements of machine learning and deep neural networks. In this research, the spatio-temporal relationship between wind and wave conditions is established using the XGBoost machine learning method and Informer deep neural networks. This approach enables effective predictions of wave height and wave period within the waters of the North Atlantic and northern Scotland. Ten years of hourly wind data from ECMWF ERA5 (2012-2021) is used as training data, while field measured wave parameters from CEFAS WaveNet buoys are employed for model training and verification. The final output enable a comparison that ultimately leads to wave predictions for the year 2022. Building upon this foundation, a versatile model for typical weather conditions and a specialized model for extreme weather scenarios are devised, facilitating more precise predictions. The data-driven model, rooted in wind data, proves adept at predicting wave characteristics across different times and locations. Notably, the trained machine learning and deep learning model delivers significant efficiency gains compared to traditional numerical models. One year's worth of data can be predicted within a few seconds by machine learning, whereas over 24 hours (on 16 logical CPUs) are required for the same prediction by TOMAWAC spectra wave model. This leap in training efficiency is a crucial development in the realm of wave prediction.