Machine learning improves the modelled wave spectrum in the North Sea
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
To improve the energy density and directional spectra computed with the SWAN model for the North Sea, a data-driven model is trained to correct the SWAN spectra. After training of the data-driven model on a year of observed and modelled data, the energy density and directional spectrum are corrected for three locations in the North Sea. When this correction is applied, the SWAN results are significantly improved. Both the energy density and the directions show a reduction in RMSE of up to 30% for the directions and 26 % for the energy density. Due to the short computational time of the data-driven model, this approach can easily be implemented in an operational forecast system.