Under the dual challenges of global climate change and agricultural sustainability, spatially precise characterization of agricultural nitrogen surplus has become a key technical bottleneck in balancing food security and water environmental protection. Existing global and national nitrogen datasets remain restricted to individual nitrogen species or loss pathways, with no unified, long-term, spatially continuous gridded product for total agricultural nitrogen surplus—particularly at the national scale for China. Compounding these limitations, current modeling approaches are plagued by coarse statistical resolution, short temporal coverage, and overly simplified process representations that distort realistic nitrogen migration and cycling pathways. To address these critical gaps, this study constructs a 0.5° spatial resolution dataset of agricultural nitrogen surplus in China for the period 2000–2022 (CANSD v1.0). The grid-scale nitrogen surplus (NS) is operationally defined as a spatially continuous variable that quantifies agricultural nitrogen cycling heterogeneity at 0.5° resolution. Gridded values are post-processed by rescaling to align with prefecture-level total nitrogen surplus accounting, while realizing spatially continuous characterization based on the inherent spatial heterogeneity of agricultural nitrogen cycling. This approach overcomes the scale mismatch between administrative statistics and field-scale processes by integrating machine learning with spatial downscaling, thereby mitigating potential circular causality and enhancing the characterization of cross-scale transfer patterns. Our framework provides a practical basis foundation for evidence-based agricultural nitrogen management and environmental risk mitigation.
Li et al. (Mon,) studied this question.