Global wind farms are associated with land use dependent asymmetric warming and ecological changes

Wind farms are crucial for decarbonization, but their effects on the atmosphere and ecosystems remain unclear. Here, we use explainable machine learning and structural equation modeling applied to decades of observations from tens of thousands of wind farms worldwide. The findings show that global wind farms are associated with land-use-dependent asymmetric warming and ecological changes. Nighttime land surface temperature increases by 0.041–0.125 °C and tends to level off when installed capacity reaches approximately 300 megawatts under observational aggregation. Based on the quantitative comparison, wind farm areas show a mean reduction of 1.75 grams of carbon per square meter in Net Primary Productivity. These relationships vary by land use: wind farms in forests show reduced evapotranspiration and lower vegetation productivity, whereas croplands and barren lands manifest altered moisture fluxes and cooling-related associations. Structural equation modeling supports the hypothesized pathways linking wind farms, vegetation responses, evapotranspiration, and land use. These results highlight the potential value of land-use-specific siting for sustainable wind energy development.