Mapping China’s photovoltaic distribution and decarbonization capacity using a customized genetic algorithm-enhanced machine learning approach

Amid increasing extreme weather and China’s carbon neutrality goals, accurate mapping of Photovoltaic (PV) systems remains challenging but essential for assessing their energy benefits and climate resilience. This study develops a customized hybrid genetic algorithm (CHGA) ensemble model for PV extraction, achieving 95.16% accuracy and identifying 7,139.35 km² of PV installations across China in 2024. Spatially, PV deployment follows a distinct “dual-mode, three-tier gradient” pattern: large, centralized PV clusters on bare land in the northwest, and small, high-density rooftop systems in the eastern coastal provinces, with national deployment decreasing stepwise from the northwest and east toward the central regions, and further toward the southwest and northeast. Moreover, our results indicate that previous studies underestimated rooftop PV deployment by approximately 50%. In terms of energy and environmental benefits, the national installed PV capacity reached 624.63 GW, producing roughly 585,977 GWh annually and avoiding about 308.42 MtCO₂e of emissions by 2024, equivalent to about 1.5 times China’s cumulative emission reductions from 2008 to 2018, with inland and coastal provinces contributing nearly equally to these avoided emissions. Typhoons, the most damaging natural disaster for photovoltaic systems, caused direct physical damage to around 3.5 km² of PV installations and latent damage to about 28.3 km², with losses concentrated near landfall tracks. This study provides a comprehensive national-scale quantification to support PV deployment strategies and risk management.