A deep learning approach to predict significant wave height using long short-term memory. (February 2023)

• Record Type: Journal Article
• Authors: Minuzzi, Felipe C.; Farina, Leandro
• Is Part Of: Ocean modelling. Volume 181(2023)
• Publication Date: 2023-02
• Subjects: Ocean waves -- Deep learning -- Long–short term memory -- Significant wave height -- Forecast; Oceanography -- Periodicals; Océanographie -- Périodiques; Oceanography; Periodicals; 551.46
• DOI: 10.1016/j.ocemod.2022.102151
• Languages: English
• ISSNs: 1463-5003

Abstract

Abstract: We present a new deep learning training framework for forecasting significant wave height on the Southwestern Atlantic Ocean. We use the long short-term memory algorithm (LSTM), trained with the ERA5 dataset and also with buoy data. The forecasts are made for seven different locations in the Brazilian coast, where buoy data are available. We consider four different lead times, e.g., 6, 12, 18 and 24 h. Experiments are conducted using exclusively historical series at the selected locations. The influence of other variables as inputs for training is investigated. Results of the LSTM forecast show that a data-driven methodology can be used as a surrogate to the computational expensive physical models and also as an alternative to the reanalysis data. Accuracy of the forecasted significant wave height is close to 87% when compared to real buoy data. Highlights: Long short-term memory is used to predict significant wave height. Training is performed using reanalysis data and real observations. Analysis conducted with historical series at seven locations in the Brazilian coast. The influence of other variables as inputs for training is investigated.