Global Renewables Watch (github.com)

A new dataset called Global Renewables Watch has been announced, providing a detailed global overview of commercial solar photovoltaic (PV) farms and onshore wind turbines. This dataset, derived from high-resolution satellite imagery and analyzed quarterly from late 2017 through mid-2024, identifies 375,197 wind turbines and 86,410 solar installations. The deep learning segmentation models used for detection achieved impressive correlation values (R2 of 0.96 for solar and 0.93 for wind) compared to the International Renewable Energy Agency's (IRENA) recent capacity estimates. This dataset is set to enhance efforts towards sustainable development by offering critical insights for policymakers and researchers. Significantly, the dataset's accessibility (via GeoPackage files) and accompanying inference scripts enable users to run detections on satellite imagery efficiently. Researchers can extract valuable geospatial data such as construction dates and land-use types, which can be pivotal in assessing renewable energy strategies and progress towards global sustainability goals. This initiative not only bolsters the growing field of AI in environmental monitoring but also encourages collaboration and contributions from the community, making it a vital tool in the push for clean energy solutions.