A research team led by Prof. ZENG Jiangyuan from the Aerospace Information Research Institute of the Chinese Academy of Sciences (AIRCAS) has developed a new algorithm, the Heterogeneity-Aware Downscaling Algorithm (HADA), which significantly improves global soil moisture observation by enhancing satellite-derived data resolution from approximately 25 km to 5 km.

The study was recently published in IEEE Transactions on Geoscience and Remote Sensing.

Soil moisture is critical for understanding crop growth conditions, water resource availability, and the occurrence of natural hazards such as droughts, floods, and landslides. However, existing global soil moisture datasets—primarily derived from passive microwave satellites—often lack sufficient spatial detail, limiting their applicability in regional-scale research and practical applications.

To address this limitation, the HADA algorithm incorporates surface heterogeneity, including variations in land cover, soil properties, topography, and vegetation. By leveraging machine learning techniques, the algorithm captures complex soil moisture dynamics and applies a heterogeneity-weighted correction, thereby improving both the accuracy and physical consistency of the data.

"The key challenge in soil moisture downscaling lies in effectively capturing surface heterogeneity," Prof. ZENG noted. "Our approach integrates this heterogeneity directly into the model, enabling us to generate higher-resolution data without sacrificing physical consistency."

The team used brightness temperature data from NASA's Soil Moisture Active Passive (SMAP) satellite and a newly developed soil moisture index to generate a baseline dataset. They then refined this dataset using the HADA algorithm to produce a global soil moisture product with a resolution of 0.05° (approximately 5 km).

Validation against 1,260 ground stations worldwide demonstrated that the new dataset outperforms existing products in both accuracy and the ability to capture temporal dynamics of soil moisture. It also offers finer spatial detail, fewer data gaps, and strong consistency with global aridity patterns.

Beyond soil moisture observation, the framework of the HADA algorithm could be applied to enhance the spatial resolution of other satellite-derived environmental datasets affected by surface heterogeneity. The new 5 km resolution soil moisture product is expected to provide support for precision irrigation, drought and landslide monitoring, and climate modeling.

This research was supported by the National Natural Science Fund for Excellent Young Scholars and other related programs.