We present a framework for forecasting significant wave height on the Southwestern Atlantic Ocean using the long short-term memory algorithm (LSTM), trained with the ERA5 database available through Copernicus Climate Data Store (CDS) implemented by ECMWF (European Center for Medium Range Forecast) and also with buoy data. The predictions are made for seven different locations in the Brazilian coast, where buoy data are available, ranging from shallow to deep water. Experiments are conducted using exclusively historical series at the selected locations and the influence of other variables as inputs for training is investigated. The results shows that a data-driven methodology can be used as a surrogate to the computational expensive physical models, with the best accuracy near 95%, compared to reanalysis
@article{arxiv.2201.00356,
title = {A deep learning approach to predict significant wave height using long short-term memory},
author = {Felipe C. Minuzzi and Leandro Farina},
journal= {arXiv preprint arXiv:2201.00356},
year = {2022}
}