ORCID
- Jiaxin Chen: 0000-0002-0761-4756
- Martyn Hann: 0000-0003-3965-9331
- Robert Rawlinson-Smith: 0000-0001-5830-7054
- Deborah Greaves: 0000-0003-3906-9630
Abstract
Reliable real-time wind estimation in nearshore zones is critical for offshore wind energy development, especially in regions lacking direct meteorological observations. This study presents a physics-guided machine learning framework for estimating 10-m surface wind speed and direction from wave buoy spectra. The model incorporates frequency- and wavenumber-based energy parameters, directional Fourier coefficients, directional fetch, and normalised wave number–depth indicators to account for local wind–wave dynamics.Trained on six long-term buoy–mast station pairs along the UK southwest coast, the model outperforms two established empirical methods, reducing wind speed and direction RMSE by over 50%. Sensitivity analyses confirm the value of directional and bathymetric features. The model also demonstrates strong temporal robustness on unseen years and spatial generalisability when applied to an independent buoy site with complex geographic and directional conditions.This approach offers a low-cost, scalable alternative to traditional wind measurements, particularly suited to data-sparse nearshore areas. It supports key offshore wind applications, including floating platform control, wave–wind misalignment diagnostics, and digital twin integration. The Celtic Sea—an emerging floating wind zone with sparse in situ wind data—illustrates the practical value of this framework for both planning and operational decision-making in nearshore renewable energy systems.
DOI Link
Publication Date
2025-12-08
Publication Title
Renewable Energy
ISSN
0960-1481
Acceptance Date
2025-12-06
Deposit Date
2025-12-12
Funding
The authors gratefully acknowledge that this work was supported by the Engineering and Physical Sciences Research Council (EPSRC) as part of the Supergen Offshore Renewable Energy (ORE) Hub project [EP/S000747/1]. We would also like to express our sincere thanks to Professor Paul Taylor (University of Western Australia) for his insightful discussions on depth-related parameterisation, which greatly contributed to the development of this work.
Recommended Citation
Chen, J., Hann, M., Rawlinson-Smith, R., Raji, M., & Greaves, D. (2025) 'Nearshore wind estimation from buoy wave spectra via physics-guided machine learning: A framework for offshore wind applications', Renewable Energy, . Available at: 10.1016/j.renene.2025.125000
