Bias Corrected Near-Surface Wind Data for Solar Field Siting

This dataset catalogs a long-term high-resolution data set characterizing the near-surface hourly wind speed variations at two solar sites in the Southwest U.S. This dataset demonstrates the potential for integrating improved near-surface wind data into CST and PV planning tools for better-informed site selection, improved stow strategies, and structural collector design optimization.

Site-specific wind conditions impact the collector field of concentrating solar power (CSP) plants and tracking photovoltaic (PV) utility plants during the design, installation, operation and maintenance of the field. Despite its importance. Most numerical weather models provide wind data at 10m or higher, and often at coarse spatial and temporal resolution. Wind conditions at solar collector heights of 3m are greatly impacted by surface properties, such as terrain and land use, and historical wind observation data close to the surface is not widely available.

In 2025-2026, NLR developed two Machine Learning (ML) models, Random Forest (RF) and Neural Network (NN), for bias correction of wind speed from the High Resolution Rapid Refresh (HRRR) model input 10m wind speed to more accurate historical near-surface wind data at 3m height relevant to CSP and PV. The corrected output wind speed time series from the ML models were tested and validated against long-term wind observation data records in the Southwest U.S. ranging from 1.5m to 8m heights with an average 20-year operation time and hourly resolution. Finally, the ML model was applied to generate a 10-year historical wind speed time series at two sites in the Southwest U.S. with solar resource potential: (1) Generation 3 Particle Pilot Plant (G3P3) at the National Solar Thermal Test Facility (NSTTF), Sandia National Laboratories, New Mexico, (2) Flatirons Campus (FC) at the National Laboratory of the Rockies, Colorado.