Abstract Photovoltaic (PV) power plants rapidly expand in drylands because of the high solar potential and efficient land‐use capabilities. However, existing PV data sets are often incomplete and lack installation timestamps, which significantly hinder comprehensive assessments of PV power plants' ecological impacts. By integrating a random forest (RF) algorithm with change detection techniques and using Sentinel‐2 and Landsat imagery, we have created a high‐resolution geospatial data set of PV installations (2010–2023). The data set achieved an overall accuracy of 99.07% and a recall of 84.11% in identifying installation dates. By 2023, PV power plants covered 444.26 km 2 , predominantly on barren land (74.28%) and grasslands (23.18%). Ecological impacts were largely positive, with improvements in vegetation indices, soil moisture, and reductions in diurnal temperature range (DTR) and soil salinization. In hyper‐arid regions, PV power plants reduced evapotranspiration and increased precipitation, while in semi‐arid areas, they lowered albedo and mitigated DTR. In arid regions, vegetation showed notable enhancement. These findings highlight the dual role of PV power plants in renewable energy generation and ecological restoration, offering insights into sustainable energy planning and ecosystem management. The results demonstrate how strategic deployment of PV power plants can align with global climate goals and contribute to achieving the Sustainable Development Goals.
Gao et al. (Sun,) studied this question.