Developing a multi-criteria assessment model for soil primary productivity in double cropping systems: Insights from the North China Plain

Soil, one of the Earth's most critical natural resources, supports global agricultural production and underpins key ecosystem services. Among the multiple functions soil performs, primary productivity stands out as a crucial element, pivotal for ensuring food security as the basis of the agricultural system. This study aimed to develop a multi-criteria assessment model for soil primary productivity at field scale, drawing insights from the winter wheat − summer maize rotation systems in the North China Plain. The development of the model followed the Decision Expert (DEX) methodology, using an integrated approach that combines knowledge graph and data mining techniques. We systematically structured the knowledge underpinning soil primary productivity. Utilising datasets derived from long-term field experiments and smallholder farms, the model was subjected to an iterative process of calibration and validation, enhancing both its predictive accuracy and operational applicability. The developed DEX model consists of 28 input attributes that encompass soil properties, field management practices, and meteorological conditions. The model achieved an accuracy of 71% in assessing soil primary productivity in the experimental field dataset after calibration, and 62% in the smallholder farm dataset as model validation. The developed model can effectively assess soil primary productivity function and facilitate the improvement of soil management. The innovative integration of knowledge-based and data-driven approaches proved to be effective. It is expected that the developed model can be integrated with other soil function models into a soil health decision support system that provides a holistic approach to soil health assessment and optimisation of field practices.