Vegetation Recovery, Interannual Variability, and Hydroclimatic Controls in a Hilly Coal-Mining Area of the Middle Yellow River Basin: Implications for Sustainable Land Management

Long-term assessment of vegetation recovery in mining-disturbed landscapes is essential for ecological restoration and sustainable land management. This study assessed fractional vegetation cover (FVC) dynamics from 2000 to 2024 in a hilly coal-mining region of the middle Yellow River Basin using Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data. Annual FVC was retrieved from maximum-value NDVI composites using a pixel dichotomy model. Theil–Sen trend analysis, the Mann–Kendall test, lag-1 autocorrelation assessment, coefficient of variation (CV), partial correlation analysis, and the geographical detector model were combined to quantify vegetation recovery, interannual variability, and hydroclimatic–topographic associations. To avoid ambiguity in spatial interpretation, statistics were calculated for the full study region, coalfield polygons, and the surrounding non-mining area. FVC increased across 67.78% of the full study region, 65.42% of the coalfield polygons, and 68.66% of the surrounding non-mining area. High-FVC zones expanded from 33.50% in 2000–2004 to 71.56% in 2020–2024. Because significant positive lag-1 autocorrelation occurred in 33.36% of valid pixels, nominal Mann–Kendall significance was interpreted cautiously. Low- and very-low-CV classes dominated the coalfields, while high-variability pixels were localized monitoring priorities. Precipitation was broadly positively associated with FVC, whereas the independent temperature effect was weak and mostly non-significant. Actual evapotranspiration (AET) had the highest explanatory power in the full geographical-detector model, but an AET-excluded sensitivity analysis showed that temperature, elevation, and precipitation remained important explanatory variables. Thus, AET should be interpreted as an integrated vegetation–water–energy coupling indicator rather than a fully independent causal driver. These findings support restoration-priority identification and sustainable land management in hilly mining regions.