AIRU-WRF: Spatial-Temporal Wind Datasets and Forecasts for Data-Science-Based Operational Offshore Wind Forecasting in the U.S. East Coast

Accurate short-term forecasts of wind conditions and power generation are critical for the cost-effective and reliable integration of offshore wind farms along the U.S. East Coast. These forecasts, spanning operational horizons from few minutes to several days ahead, are essential inputs to grid- and farm-level operational decision-making. Yet, long-term records of hub-height wind measurements in U.S. offshore regions are scarce, fragmented, or difficult to access. This lack of publicly available, high-quality data presents a significant barrier to the development, benchmarking, and deployment of advanced forecasting models—unlike other regions with significant offshore wind potential (e.g., the North Sea), where well-curated datasets have fueled progress and accelerated the progress in data-driven forecasting. In response, and as part of a NOWRDC-supported project, Rutgers University developed one of the first data-driven offshore wind forecasting models specifically tailored to the U.S. East Coast. This repository contains important datasets related to this project: A year-long record of raw and processed hub-height wind speeds at key offshore wind energy locations in the NY/NJ Bight. Day-ahead forecasts of key meteorological variables from an in-house numerical weather prediction (NWP) model developed by Rutgers University, called RU-WRF. Machine learning–based hub-height wind speed forecasts from AIRU-WRF - Rutgers' newly developed offshore wind forecasting model, trained on both historical measurements and RU-WRF outputs. By making the datasets and forecasts publicly available, we aim to catalyze further research, model development, and benchmarking studies in offshore wind for U.S. coastal regions. Related publications and patents: [1] Ye, Feng, Joseph Brodie, Travis Miles, and Ahmed Aziz Ezzat. “AIRU-WRF: A physics-guided spatio-temporal wind forecasting model and its application to the US Mid Atlantic offshore wind energy areas.” Renewable Energy 223 (2024): 119934 (*An earlier version of AIRU-WRF that primarily focused on six-hour ahead forecast horizons, i.e., <= 6 hours*) [2] Ye, Feng, Travis Miles, and Ahmed Aziz Ezzat. “Improved spatio-temporal offshore wind forecasting with coastal upwelling information.” Applied Energy 380 (2025): 125010 (**An update to AIRU-WRF to model coastal upwelling effects**) [3] Ye, Feng, Travis Miles, and Ahmed Aziz Ezzat. “Offshore Wind Energy Prediction Using Machine Learning with Multi-Resolution Inputs.” In Multimodal and Tensor Data Analytics for Industrial Systems Improvement, pp. 167-183. Cham: Springer International Publishing, 2024 (**A variant of AIRU-WRF that leverages multi-resolution WRF inputs**). [4] Ye, Feng, Joseph Brodie, Travis Miles, and Ahmed Aziz Ezzat. “Ultra-short-term probabilistic wind forecasting: Can numerical weather predictions help?.” In 2023 IEEE Power & Energy Society General Meeting (PESGM), pp. 1-5. IEEE, 2023 (**A method to leverage WRF wind velocity predictions for very-short-term wind forecasting**). [5] Ji, Jiaxiang, Ye, Feng, Travis Miles, and Ahmed Aziz Ezzat. “A Multivariate, Time-Dependent Ensemble Method for Space-Time Wind Energy Forecasting,” Under Review (2025) (**Most recent version of AIRU-WRF -as of June 2025- that primarily focused on day-ahead forecasting, i.e., <= 24 hours**) [6] Walid, K.B., Ye, F., Ji, J., Miles, T., Aziz Ezzat, A., and Jiang, Y., “Economic and Reliability Value of Improved Offshore Wind Forecasting in Bulk Power Grid Operation,” Working paper (2025). (**An economic validation study that quantifies the projected grid benefits of AIRU-WRF**) [7] “Techniques To Provide Improved Wind Input for Operating Offshore Wind Turbines,” U.S. Application 19/111. Inventors: Ahmed Aziz Ezzat, Feng Ye, Travis Miles, Joseph Brodie (**Patent for AIRU-WRF supported by the Rutgers Office of Innovation Ventures**)