A research team has developed a transfer learning framework that allows China's Fengyun-4A (FY-4A) geostationary satellite to estimate surface solar radiation and its global, direct, and diffuse components with high accuracy, according to a study published in the Journal of Remote Sensing on April 29, 2026. The method adapts knowledge from the Himawari-8-based Cloud, Atmospheric Radiation and Renewal Energy Application (CARE) product, reducing dependence on auxiliary meteorological datasets and providing stronger data for solar power forecasting, climate research, and sustainable energy planning.
Surface solar radiation is critical for Earth's energy balance, hydrological cycles, ecosystem processes, and the performance of solar photovoltaic and concentrating solar power systems. Ground-based radiometric networks offer reliable observations but are sparse, especially across oceans and developing regions. Reanalysis products provide broad coverage but may lose accuracy due to coarse resolution and simplified cloud–aerosol–radiation interactions. Satellite observations can fill this gap, but many existing algorithms are sensor-specific and focus mainly on global radiation rather than separately estimating direct and diffuse components.
Researchers from the Aerospace Information Research Institute, Chinese Academy of Sciences; Sichuan University of Science and Engineering; and the Institute of Atmospheric Physics, Chinese Academy of Sciences reported the study (DOI: 10.34133/remotesensing.1044). The team first developed a deep neural network model using Himawari-8 Level 1 observations and the CARE radiation product, then fine-tuned it with FY-4A data. The model uses top-of-atmosphere reflectance and solar–satellite geometry as dynamic inputs, with Bayesian optimization selecting key hyperparameters.
Validation using 33 ground stations from the Baseline Surface Radiation Network, Bureau of Meteorology, and Global Tropical Moored Buoy Array during 2018–2020 showed strong performance. At representative BSRN sites, FY-4A achieved instantaneous root mean square errors of 102.2, 117.5, and 83.1 W m⁻² for global, direct, and diffuse radiation, respectively. At the daily mean scale, RMSEs dropped to 28.5, 30.1, and 22.6 W m⁻².
The authors emphasized that the framework allows FY-4A to estimate direct and diffuse components separately, which determine how solar energy systems perform under clear, cloudy, and hazy conditions. Reducing reliance on auxiliary meteorological data makes the method more practical for near-real-time monitoring. The new FY-4A radiation product could help improve PV site assessment, power forecasting, grid management, climate modeling, and land-surface simulations. The same transfer learning strategy could be extended to other Chinese geostationary satellites, including Fengyun-4B, supporting more reliable solar-energy monitoring across East Asia and beyond.
