Recently, the Smart Agriculture Team at the Institute of Agricultural Resources and Regional Planning (IARRP), Chinese Academy of Agricultural Sciences (CAAS), has made significant strides in near-real-time farmland monitoring. The team developed a Near-Real-Time Ground–Satellite Fusion (NRT-GSF) algorithm, which successfully integrates ground-based Internet of Things (IoT) observations with high-spatial-resolution satellite imagery to achieve continuous daily monitoring of crop Green Area Index (GAI) at the field scale. These findings were published in Remote Sensing of Environment under the title "NRT-GSF: A novel near-real-time ground-satellite fusion algorithm to retrieve daily green area index at field scale".

The precision management of crop fields, including irrigation scheduling, fertilization practices, pest management, harvest planning and yield forecasting, relies heavily on high-frequency, fine-grained monitoring of crop growth status. However, while satellite remote sensing offers broad coverage, it often struggles to acquire continuous time-series data in regions with frequent cloud and rain due to cloud obstruction. Conversely, ground-based IoT devices, despite providing daily observations, are limited to single-point monitoring and lack the spatial distribution information required for whole-field assessment. Furthermore, existing spatiotemporal fusion methods typically depend on historical data and lack near-real-time prediction capabilities, making it difficult to meet the immediate monitoring demands of precision agriculture.

To address these challenges, this study proposes a near-real-time daily data fusion framework (NRT-GSF) based on a Bayesian Dynamic Linear Model and Kalman Filtering. By combining Sentinel-2 satellite data with a ground-based agricultural IoT system (IoTA) within a recursive computational framework, NRT-GSF can perform forward prediction using ground IoT data and existing satellite data when satellite images are unavailable, generating daily 10-meter resolution GAI products. Additionally, upon the acquisition of new satellite data, the system supports backward updating to refine historical calculations, further enhancing the accuracy of the time-series data.

Using IoTA data and Sentinel-2 images from the 2019 wheat-growing season in France, the team validated the algorithm's performance. Results indicate that the NRT-GSF algorithm effectively fills gaps in satellite observations, generating daily GAI data that demonstrates high consistency with ground measurements (R = 0.75 - 0.98, RMSE = 0.1 - 0.49), outperforming the existing CACAO (Consistent Adjustment of the Climatology to Actual Observations) algorithm in accuracy. Furthermore, independent ground validation using handheld RGB cameras confirmed the method's reliability (RMSE = 0.5). This research provides a novel approach for resolving scale discrepancies and real-time constraints in multi-source data fusion, offering improved scientific evidence and technical support for precision agriculture management.

Schematic diagram of the NRT-GSF algorithm implementation

Dr. Li Wenjuan, Researcher at IARRP, is the first and corresponding author of the paper. This research was supported by the State Key Laboratory of Efficient Utilization of Arid and Semi-Arid Farmland, the National Key R&D Program of China (2023YFD2300500), and the National Natural Science Foundation of China (42201388). 

Original article link: https://doi.org/10.1016/j.rse.2025.115160