A recent study has unveiled an innovative artificial intelligence (AI)-driven remote sensing framework aimed at identifying the potential for forage cultivation in the drylands of northern China, particularly in the middle reaches of the Yellow River. Published in the journal Water Research, this research highlights optimal areas for growing forage at a kilometer scale, providing critical data and decision-making tools to enhance ecological protection, promote sustainable agricultural practices, and support national food and feed security.
Led by Prof. Wang Shudong from the Aerospace Information Research Institute of the Chinese Academy of Sciences, this study represents a collaborative effort with the Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters and the Department of Earth and Environmental Science at the University of Pennsylvania.
The drylands in northern China face significant challenges, including limited water resources and the necessity of ensuring a stable supply of feed and food. To tackle these issues, the researchers developed a comprehensive framework that integrates satellite observations, ecohydrological model outputs, and field measurements, significantly reducing the need for extensive in-situ sampling.
Utilizing a combination of multi-source satellite data and mechanistic models that assess water balance and crop growth, the team generated high-quality training samples. They employed advanced techniques such as ensemble learning and transfer learning to extract essential production factors, including irrigation water usage, vegetation net primary productivity (NPP), and soil organic carbon (SOC), achieving a remarkable retrieval accuracy exceeding 90% for these factors.
Furthermore, the study”s methods for distribution alignment and quantile mapping successfully reduced regional biases by 43%. This enabled the identification of optimal forage-growing belts with a positional accuracy surpassing 85%. Unlike traditional assessments that rely on single metrics, this framework conceptualizes forage planting as a spatial optimization challenge, which concurrently considers water consumption, soil carbon sequestration benefits, and forage production capacity.
By quantifying ecological, economic, and water cost factors on a unified scale, the framework identifies priority areas for planting and optimal input-output ratios. This facilitates the efficient allocation of resources, labor, and funding. The researchers emphasized that the approach is not only replicable and cost-effective but also supports ecosystem restoration and high-quality agricultural development in regions facing stringent water limitations.
For further details, refer to the study by Kai Liu and colleagues titled “Assessing reclamation potential of abandoned drylands using knowledge-guided machine learning (KGML) and remote sensing,” published in Water Research.
