Researchers Develop Satellite-Based Method to Map Surface Solar Radiation Using Transfer Learning

A research team from the Chinese Academy of Sciences and Sichuan University of Science and Engineering has developed a new transfer learning method to enhance solar radiation mapping. By utilizing data from the Fengyun-4A (FY-4A) geostationary satellite, the model estimates global, direct, and diffuse surface solar radiation with increased accuracy. This technical approach aims to reduce the reliance on auxiliary meteorological datasets and sparse ground-based station observations, providing a more robust foundation for solar energy forecasting, climate research, and land-surface modeling applications globally.

The integration of transfer learning into satellite-based radiation retrieval addresses significant gaps in monitoring surface solar radiation, particularly in regions where ground-based radiometric networks are limited. By applying radiative knowledge from the Himawari-8 satellite's existing data products to the FY-4A platform, the researchers have created a framework that effectively isolates global, direct, and diffuse solar radiation components. This distinction is critical for energy sector applications, as direct radiation is a primary factor for concentrating solar power systems, while diffuse radiation is essential for evaluating photovoltaic performance under varying atmospheric conditions such as cloud cover or aerosol presence. The reduction in dependence on ancillary meteorological datasets further enhances the operational feasibility of this model for near-real-time monitoring.

From a technical standpoint, the ability to derive high-resolution solar data from geostationary satellites supports more precise site assessment and power grid management. As global energy transitions continue to prioritize renewable sources, the availability of granular solar radiation data becomes a necessary component for long-term climate modeling and infrastructure planning. By providing a scalable strategy that can be extended to other satellites, such as the Fengyun-4B, this methodology offers a path toward consistent, wide-area solar energy monitoring. This development provides researchers and energy planners with a standardized tool to improve the accuracy of solar resource assessments without the logistical challenges associated with maintaining extensive ground-based sensor infrastructure in remote or underserved geographical areas.