Mapping land- and offshore-based wind turbines in China in 2023 with Sentinel-2 satellite data

Abstract

Global wind power generation has grown rapidly in recent years, with China emerging as the world's largest market. Wind turbines, the key devices for this generation, are widely distributed both on land and at sea. Accurate mapping and regular updates of their locations are essential for energy production predictions, efficiency assessment, and environmental impact evaluation. While satellite remote sensing facilitates rapid mapping of offshore wind turbines, methods for detecting land-based wind turbines remain underdeveloped. To address this issue, this study proposes a novel framework for wind turbine detection using Sentinel-2 MSI data and generates the first map of both land- and offshore-based wind turbines in China in 2023. A total of 148,181 land- and 7,541 offshore-based wind turbines are detected with satisfactory accuracy (OA = 0.964, F-score = 0.963). We find that land-based turbines are primarily concentrated in northwest and north China, with the largest numbers found in Inner Mongolia, Xinjiang, Hebei, and Gansu provinces (>10,000 units). Inner Mongolia is the leading contributor, with over 23,000 units. These turbines are mainly located in areas with low altitudes, gentle slopes, strong winds, and surrounding land cover types of grasslands, cropland, and barren land. Offshore turbines are mostly found in nearshore areas with uniform distribution. This wind turbine map provides essential information for predicting wind power production, optimizing wind farm sites, and evaluating environmental impacts. Moreover, the proposed approach relies entirely on Sentinel-2 data, currently the highest-resolution open-access satellite data globally, providing valuable support for wind turbine localization and installation date updates.