Earth Observation Reveals Spatial Patterns of Farmland Shelterbelts in Three-North Region of China
23 Jan 2026
As a core component of the Three-North Shelter Forest Program in China's Three-North Region (TNR), farmland shelterbelts (FSs), also known as windbreaks, play a crucial role in reducing wind erosion and preventing soil degradation. However, the large-scale spatial patterns of FSs have remain poorly understood, limiting efforts to upgrade and optimize these systems.
A new study led by the CropWatch team from the Aerospace Information Research Institute, Chinese Academy of Sciences (AIRCAS), published in Remote Sensing of Environment, provided the first comprehensive map of farmland shelterbelts across the TNR. The study developed a deep learning-based framework utilizing Sentinel-2 satellite data to extract farmland shelterbelts in the TNR at 10-m resolution for the year of 2022.
A stratified random sampling strategy was adopted to construct training samples. A TransUNet model was trained to extract farmland shelterbelts from Sentinel-2 satellite imagery. The extracted results were then vectorized, and key shelterbelt parameters were calculated, including FS length, width, orientation, area, FS to farmland ratio, FS network connectivity, and FS deviation angle. Finally, the spatial patterns of farmland shelterbelts were analyzed.
The study estimates that the total length of farmland shelterbelts in the region reached 915.09 ± 21.45 thousand kilometers, with a total area of 1575.18 ± 58.75 thousand hectares. Heilongjiang, Xinjiang, and Inner Mongolia had highest FS length and area, while Beijing, Tianjin, and Qinghai showed the lowest values. The average FS to farmland ratio, connectivity, and FS deviation angle were 2.83 % (± 0.11 %), 0.34, and 26°, respectively, indicating potential for structural optimization.
Spatially, shelterbelts were more concentrated, and well-structured in Northeast China and Xinjiang. In contrast, regions such as the Hetao Plain, Guanzhong Plain, and Hexi Corridor showed clear gaps, suggesting the need for further shelterbelt construction and improvement.
This research provides essential baseline data and practical insights for assessing and optimizing farmland shelterbelt infrastructure.
Research News
Earth Observation Reveals Spatial Patterns of Farmland Shelterbelts in Three-North Region of China
As a core component of the Three-North Shelter Forest Program in China's Three-North Region (TNR), farmland shelterbelts (FSs), also known as windbreaks, play a crucial role in reducing wind erosion and preventing soil degradation. However, the large-scale spatial patterns of FSs have remain poorly understood, limiting efforts to upgrade and optimize these systems.
A new study led by the CropWatch team from the Aerospace Information Research Institute, Chinese Academy of Sciences (AIRCAS), published in Remote Sensing of Environment, provided the first comprehensive map of farmland shelterbelts across the TNR. The study developed a deep learning-based framework utilizing Sentinel-2 satellite data to extract farmland shelterbelts in the TNR at 10-m resolution for the year of 2022.
A stratified random sampling strategy was adopted to construct training samples. A TransUNet model was trained to extract farmland shelterbelts from Sentinel-2 satellite imagery. The extracted results were then vectorized, and key shelterbelt parameters were calculated, including FS length, width, orientation, area, FS to farmland ratio, FS network connectivity, and FS deviation angle. Finally, the spatial patterns of farmland shelterbelts were analyzed.
The study estimates that the total length of farmland shelterbelts in the region reached 915.09 ± 21.45 thousand kilometers, with a total area of 1575.18 ± 58.75 thousand hectares. Heilongjiang, Xinjiang, and Inner Mongolia had highest FS length and area, while Beijing, Tianjin, and Qinghai showed the lowest values. The average FS to farmland ratio, connectivity, and FS deviation angle were 2.83 % (± 0.11 %), 0.34, and 26°, respectively, indicating potential for structural optimization.
Spatially, shelterbelts were more concentrated, and well-structured in Northeast China and Xinjiang. In contrast, regions such as the Hetao Plain, Guanzhong Plain, and Hexi Corridor showed clear gaps, suggesting the need for further shelterbelt construction and improvement.
This research provides essential baseline data and practical insights for assessing and optimizing farmland shelterbelt infrastructure.