Spatiotemporal Dynamics and Driving Factors Analysis of Vegetation Net Primary Productivity in China Using Remote Sensing and Machine Learning

Net primary productivity (NPP) is a critical indicator of ecohydrological processes, reflecting vegetation responses—including in croplands—to hydrological and carbon dynamics amid environmental changes. However, existing studies often overlook comprehensive cross-basin analyses of seasonal, interannual, and extreme-event responses in China's major river basins, while insufficiently addressing spatial heterogeneity and autocorrelation, thereby constraining ecohydrological attribution and management. To bridge these gaps, this study integrates remote sensing and climate data (2001–2020) with time-series, statistical, machine learning, and spatial analyses to examine NPP dynamics across nine major basins: SLR, HHR, HER, YWR, YER, PLR, SEB, SWB, and ILB. Spatiotemporal patterns reveal a southeast-to-northwest gradient, with peak NPP in humid PLR and SEB and minima in arid ILB; growing seasons account for 59.5%–98.9% of annual totals, while temperate basins exhibit significant increases (3.60–5.98 g C m⁻² yr⁻¹, p < 0.05) driven by restoration efforts, contrasting ILB's decline (−0.78 g C m⁻² yr⁻¹, p < 0.05) due to drought. The 2011 drought induced 4.6%–12.0% reductions in temperate NPP, with humid basins showing greater resilience. SHAP identifies vegetation metrics (FVC, LAI) and hydrological factors (Precip-Tavg, ET) as dominant drivers, modulated by topography. Spatial heterogeneity is evident from Moran's I (0.32–0.90, p < 0.001), with spatial lag models (SLM) improving predictions (R² = 0.89, RMSE = 70.63 g C m⁻² yr⁻¹) over standalone Light Gradient Boosting Machine (LGM; R² = 0.87, RMSE = 78.76 g C m⁻² yr⁻¹) by incorporating autocorrelation. These findings inform adaptive ecohydrological strategies to enhance ecosystem resilience and support China's ecological civilization goals, with implications for crop monitoring in agriculturally intensive basins such as YWR and YER.