AI maps China’s national wind and solar infrastructure

Researchers at Peking University and Alibaba Group’s DAMO Academy applied a deep-learning model to 7.56 terabytes of sub‑meter satellite imagery to identify 319,972 solar photovoltaic facilities and 91,609 wind turbines, creating China’s first national high-resolution inventory of wind and solar infrastructure, the paper published in Nature states.

The team processed imagery covering installations in 1,915 counties and produced a county-level dataset and the code, which are publicly available on Zenodo. The model was trained to detect a wide range of installation types, from rooftop panels in coastal cities to utility-scale wind farms on the Mongolian plateau, and the inventory maps each identified facility.

Using the new dataset, the researchers evaluated solar-wind complementarity-how solar and wind output offset each other’s variability-by testing interactions among actual facilities rather than hypothetical deployments. They report that pairing solar and wind across larger geographic areas reduces aggregate generation variability, with greater gains as pairings expand across provinces. The paper says managing resources at a national scale, rather than at the current provincial level, could make it easier to balance supply and reduce curtailment.

Liu Yu, a professor at Peking University’s School of Earth and Space Sciences, described the inventory as allowing China to view its new-energy landscape from a “God’s-eye view,” and added that grid operators cannot optimize assets they do not know exist.

The study notes rapid growth in electricity demand from data centers and computing. The China Electricity Council reported that power consumption in the data-services and computing sector rose 44% year-on-year in the first quarter of 2026 to 22.9 billion kilowatt-hours. New data centers have tended to locate in northern and western provinces where land costs are lower and wind and solar resources are stronger, the regions the study identifies as having high solar-wind complementarity.

China has long experienced curtailment, where generated renewable energy cannot be absorbed by the grid. The researchers state that coordinating distant solar and wind facilities could smooth aggregate output and allow more efficient use of transmission, and they provide analytical tools to test transmission and dispatch strategies using the mapped infrastructure.

The technical work processed 7.56 terabytes of imagery to produce a nationally consistent inventory spanning nearly 320,000 solar installations and more than 91,000 wind turbines across 1,915 counties. The paper cites an estimate that China’s clean energy industries generated about 15.4 trillion yuan (roughly US$2.26 trillion) in output last year and says the inventory is intended to support planners, grid operators and researchers working on integration, siting and transmission planning.