Spatiotemporal Dynamics and Environmental Driving Mechanisms of Vegetation in Ecologically Fragile Areas Based on kNDVI: A Case Study of the Loess Plateau, China

Against the backdrop of global climate change, understanding vegetation dynamics in ecologically fragile areas (EFAs) has become crucial for assessing climate change impacts. As a representative EFA in China, the Loess Plateau (LP) necessitates continuous vegetation monitoring to inform effective restoration strategies. This study employs the kernel Normalized Difference Vegetation Index (kNDVI) to accurately characterize vegetation changes across the LP from 2001 to 2022. We systematically analyze spatiotemporal trends, vegetation stability (coefficient of variation, CV), and persistence (Hurst exponent) while quantitatively assessing environmental drivers using the Geodetector method. Key findings include: (1) a significant increasing trend in kNDVI, reflecting improved veg-etation coverage; (2) the CV (0.03–4.41) revealed significant spatial heterogeneity and low vegetation stability, while the Hurst exponent (0.21–0.77, mean = 0.67) indicated strong persistence in vegetation changes; and (3) climate factors—particularly mean annual precipitation (MAP) and aridity index—as dominant drivers of kNDVI variation. Notably, here were significant interactions between climatic factors and biophysical and anthropogenic fac-tors—especially between MAP and vegetation type, landform, mean annual potential evapotranspiration (MAPET), soil type, and land use change—all of which explained more than 60% of the variation in kNDVI. These findings deepen our understanding of nonlinear vegetation-environment interactions and provide critical insights for opti-mizing ecological policies in the LP.