Spatiotemporal Dynamics and Reach-Dependent Drivers of Vegetation NDVI in the Datong River Basin from 2000 to 2024

We examined spatiotemporal NDVI patterns in the Datong River Basin (DTR) using MODIS normalized difference vegetation index (NDVI) data from 2000 to 2024. We used the Theil–Sen median slope estimator and Mann–Kendall test to characterize NDVI trends, the Hurst exponent as an exploratory indicator of persistence, and the optimal-parameter geographical detector (OPGD) to quantify the spatial explanatory power and interaction patterns of factors associated with NDVI differentiation. We found that: (1) NDVI increased overall, despite pronounced interannual fluctuations, at 0.0012 yr−1. For descriptive comparison, the five equal-length sub-periods differed in both the direction and magnitude of NDVI change. (2) Vegetation coverage showed marked reach-dependent heterogeneity. Medium-high coverage was the dominant class across the basin, and medium-high- and high-coverage areas together accounted for 69.05% of the vegetated area. The middle reach had the highest combined proportion of medium-high and high coverage, whereas the lower reach had the highest proportion of high coverage alone and a more heterogeneous coverage composition. The mean Hurst exponent was 0.71; persistent increase, persistent decrease, anti-persistent decrease, and anti-persistent increase accounted for 76.3%, 14.2%, 1.4%, and 8.1% of the basin, respectively. (3) Mean annual temperature and elevation had the highest q values for basin-wide NDVI differentiation, at 0.636 and 0.580, respectively. All detected two-factor interactions were classified as bi-variable or nonlinear enhancement, and their joint q values exceeded those of the corresponding individual factors. Factor-ranking and interaction patterns differed among reaches: elevation and temperature had relatively high q values in the upper reach; climate- and land-use-related factors had relatively high explanatory power in the middle reach; and soil type, vegetation type, and human footprint had relatively high q values in the lower reach. (4) These results demonstrate reach-dependent spatial association patterns between NDVI and the examined factors and provide a basis for spatially differentiated vegetation monitoring, conservation, and ecological restoration in the DTR.