The Potential of Machine Learning for Wind Speed and Direction Short-Term Forecasting - A Systematic Review

Aug 3, 2026

Upload by Maeve

Machine learning has significantly improved the accuracy of wind forecasting, which is vital for operational services and safety. This systematic review evaluates 23 studies published between 1983 and 2023 on nowcasting wind speed and direction. Prediction horizons span from 1 minute to 1 week, with more research at lower temporal resolutions. Neural networks dominate, and deep learning has gained prominence. Reported accuracy metrics commonly include MAE, MSE, and MAPE, with mean values of 0.56 m/s, 1.10 m/s, and 6.72%. Deep learning generally outperforms traditional methods, while adding non-wind weather variables does not improve overall performance. The review recommends future work using diverse spatial data points and high-resolution time data with deep learning models.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Introduction

Wind forecasting context and nowcasting definition

Limitations of numerical weather prediction

Data-driven machine learning approaches

Related work and literature review gap

Abstract and study scope

computers Article The Potential of Machine Learning for Wind Speed and Direction Short-Term Forecasting: A Systematic Review D²cio Alves 1,2, *, F¡bio Mendonça 1,2, *, Sheikh Shanawaz Mostafa 2 and Fernando Morgado-Dias 1,2 Citation: Alves, D.; Mendonça, F.; Mostafa, S.S.; Morgado-Dias, F. The Potential of Machine Learning for Wind Speed and Direction Short-Term Forecasting: A Systematic Review. Computers 2023, 12, 206 . [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)computers12100206 Academic Editors: Phivos Mylonas, Katia Lida Kermanidis and Manolis Maragoudakis Received: 14 September 2023 Revised: 5 October 2023 Accepted: 11 October 2023 Published: 13 October 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://](https://)[ ](https://)[creativecommons.org/licenses/by/](creativecommons.org/licenses/by/)[ ](creativecommons.org/licenses/by/)[4.0/](4.0/)) . 1 Faculty of Exact Sciences and Engineering, University of Madeira, 9020-105 Funchal, Portugal; [morgado@staff.uma.pt](morgado@staff.uma.pt) 2 Interactive Technologies Institute (ITI/LARSyS and ARDITI), 9020-105 Funchal, Portugal; [sheikh.mostafa@tecnico.ulisboa.pt](sheikh.mostafa@tecnico.ulisboa.pt) * Correspondence: [decio.alves@iti.tecnico.ulisboa.pt](decio.alves@iti.tecnico.ulisboa.pt) (D.A.); [fabioruben@staff.uma.pt](fabioruben@staff.uma.pt) (F.M.) Abstract: Wind forecasting, which is essential for numerous services and safety, has signiﬁcantly improved in accuracy due to machine learning advancements. This study reviews 23 articles from 1983 to 2023 on machine learning for wind speed and direction nowcasting. The wind prediction ranged from 1 min to 1 week, with more articles at lower temporal resolutions. Most works employed neural networks, focusing recently on deep learning models. Among the reported performance metrics, the most prevalent were mean absolute error, mean squared error, and mean absolute percentage error. Considering these metrics, the mean performance of the examined works was 0.56 m/s, 1.10 m/s, and 6.72%, respectively. The results underscore the novel effectiveness of machine learning in predicting wind conditions using high-resolution time data and demonstrated that deep learning models surpassed traditional methods, improving the accuracy of wind speed and direction forecasts. Moreover, it was found that the inclusion of non-wind weather variables does not beneﬁt the model's overall performance. Further studies are recommended to predict both wind speed and direction using diverse spatial data points, and high-resolution data are recommended along with the usage of deep learning models. Keywords: deep learning; machine learning; nowcast; wind speed; wind direction; wind 1. Introduction According to the World Meteorological Organization (WMO), nowcasting is the process of providing short-term high-resolution forecasts with a detailed description of current weather conditions over a time horizon of up to six hours. Although nowcasting has implications in diverse application ﬁelds such as hydrology, aviation, road safety, civil protection, and industry or energy, it is notably relevant for examining severe weather phenomena such as high wind conditions [1–4] . The conventional forecast employs numerical weather prediction models that have limited capacity for predicting the timing and location of rapidly evolving weather patterns due to the complexity of the mathematical and physical equations that they must process and solve. Additionally, due to their high computational requirements, these systems are limited in producing rapid results and normally can only perform one or two computations per day [5] . The need to explore alternative solutions has arisen due to these limitations. One promising option is using dat

FAQs

What time horizon and resolution range do the reviewed studies cover for wind nowcasting?

The reviewed work predicts wind speed and direction from 1 minute up to 1 week, with more studies focused on lower temporal resolutions.

Which modeling approaches are most commonly used in the literature?

Most studies rely on neural networks, with deep learning models becoming increasingly prominent in more recent research.

How do key performance metrics compare across the reviewed methods?

The most common metrics are MAE, MSE, and MAPE, and the review reports mean performance values of 0.56 m/s, 1.10 m/s, and 6.72% respectively.

Aug 3, 2026

Upload by Maeve