Spatial drivers of atmospheric NO₂ and NH₃ in China: Disentangling source- versus process-driven controls

Understanding the distinct drivers of nitrogen dioxide (NO₂) and ammonia (NH₃) is essential for effective air-quality management. This study characterizes their nationwide dynamics across China from 2015 to 2023 using an interpretable machine-learning framework that integrates satellite observations with Random Forest modeling and SHapley Additive exPlanations (SHAP). Nationally, NO₂ concentrations declined markedly from 6.20 to 4.90 Nμg/m³, whereas NH₃ levels remained broadly stable. Spatial patterns show NO₂ hotspots concentrated in urban-industrial corridors and NH₃ enhancements over agricultural regions. SHAP-based attribution indicates that meteorology is a shared primary driver, yet secondary controls diverge: NO₂ is predominantly source-driven, strongly linked to anthropogenic variables such as population density and urban extent (35.1% of explained importance), while NH₃ is process-driven, governed by ecological and agricultural factors - including vegetation, soil properties, and fertilizer application (30.9%) - with these effects most pronounced in cropland and vegetated areas. The contrasting regimes of the two pollutants underscore the differing efficacy of control approaches. Direct emission policies have successfully reduced NO₂, but the persistence of NH₃ highlights the need for integrated, process-oriented management that couples agricultural best practices with climate and ecosystem adaptation. This differentiated understanding provides a scientific basis for regionally tailored nitrogen-mitigation strategies in China.