畜牧密度

The scarcity of long-term, high-resolution datasets limits our understanding of grazing dynamics in heterogeneous grassland landscapes. Accurate, spatially explicit livestock distribution data are essential for quantifying grazing pressure and fostering sustainable pastoral management. This study develops a Downscaling-Optimized Random Forest (DORF) framework, integrating multi-source remote sensing variables and census data to generate a 1-km gridded livestock density dataset for Inner Mongolia from 2000 to 2019. We characterized the spatiotemporal evolution of grazing intensity across different ecological zones and policy implementation periods. To disentangle the complex interactions between management and environment, structural equation modeling (SEM) was employed to quantify the relative contributions of climatic variables and anthropogenic factors to net primary productivity (NPP). Our results reveal significant spatial heterogeneity in grazing pressure, with grassland productivity being jointly regulated by climate and human interventions; notably, the dominant drivers shifted significantly over the two-decade study period. By coupling livestock density with productivity indicators, we identified a widespread alleviation of grazing pressure and a substantial reduction in overgrazed areas across Inner Mongolia, suggesting an improvement in the grass-livestock balance. This research demonstrates that the DORF-based reconstruction provides a robust tool for high-resolution grazing assessment, offering critical data support for adaptive grassland management and the restoration of degraded geo-ecosystems.