Objective  In semi-arid regions, intensive coal mining is prone to induce vegetation degradation and reshape the land use pattern. To identify the ecological impacts of coal mining and provide support for differential restoration, this study presents a quantitative comparison of changes in vegetation, soil moisture content, and land use types between mining-affected and non-mining areas.

Methods Yulin City, Shaanxi Province was investigated in this study. Two parallel baseline frameworks were constructed: mining-affected and non-mining areas, which were compared under the sandy-fluvial terrace and loess hilly-gully landforms. The annual maximum normalized difference vegetation index (NDVI) derived from Landsat imagery was used to analyze pixel-level trends and significance (Z and P) using the Theil-Sen estimator and Mann-Kendall significance tests. Soil moisture content data on a spatiotemporally continuous 1-km scale (2000‒2020) were used to calculate average moisture content across various areas. Furthermore, primary land-use transition pathways were quantified by establishing the transfer matrix of land use types using multi-stage land-use data.

Results The NDVI values of Yulin City increased overall, with averages rising from 0.273 5 to 0.497 4 with an average annual increase of 0.011 2. These values revealed that areas with significantly improved vegetation reached 17 224.17 km², those with slightly improved vegetation spanned an area of 23 686.34 km², and degraded areas covered an area of 493.23 km² (accounting for approximately 0.91%). Coal mining exerted significant effects. Specifically, the NDVI values of coal mining-affected areas increased by 92.17%, which was higher than that in non-mining-affected areas (81.31%). The mining-affected areas showed an annual growth rate of NDVI values reaching up to 4.85%, substantially surpassing that of non-mining-affected areas (4.33%). Additionally, the mining-affected areas generally exhibited high Z values and P<0.05, indicating a stable upward trend. In contrast, most non-mining-affected areas in the southern part showed P>0.08, indicating insignificant changes in NDVI values. For the sandy-fluvial terrace landform, the NDVI growth rate of mining-affected areas was 13.64% higher than that of non-mining-affected areas. In contrast, for the loess hilly-gully landform, the NDVI growth rate of mining-affected areas proved 2.11% higher than that of non-mining-affected areas. These findings enhanced the cross-landform comparability. The soil moisture content improved simultaneously in the mining-affected and non-mining-affected areas while remaining low overall. To be specific, the average soil moisture content increased by 25.1% from 0.202 5 m³/m³ to 0.253 5 m³/m³ in the whole study area, increasing by 27.0% from 0.180 0 m³/m³ to 0.228 6 m³/m³ in mining-affected areas and by +24.3% from 0.212 0 m³/m³ to 0.263 5 m³/m³ in the non-mining-affected areas. Furthermore, the soil moisture content in non-mining-affected areas remained higher than that in mining-affected areas, with a stable increased amplitude. Regarding the reshaping of the land use pattern and its correspondence to mining activity, the mining-affected areas showed grassland and forestland retention rates of 88.67% and 88.76%, respectively, lower than those (91.40% and 91.96%) of the non-mining-affected areas. In contrast, the construction land in the mining-affected areas expanded from 42.9 km² to 525.1 km² (increase: 12.2 times), while that in non-mining-affected areas grew from 119.6 km² to 381.0 km²(increase: 3.19 times).

Conclusions From 2000 to 2020, Yulin City showed significantly improved fractional vegetation cover, with an increase of 82%, which, however, was accompanied by an increasingly significant spatial heterogeneity. Compared to the non-mining-affected areas, the mining-affected areas exhibited lower baseline NDVI levels, while their NDVI values fluctuated more significantly under enhanced soil moisture content. Additionally, the mining-affected areas demonstrated lower grassland and forestland retention rates compared to the non-mining-affected areas. Furthermore, these areas showed far higher degrees of construction land expansion, which was highly consistent with localized vegetation degradation spatially. This study quantitatively revealed the superimposed effects of mining disturbance and ecological restoration, highlighted the predominant controlling effect of the soil moisture content on vegetation response mechanisms, and proposed a zonal management and control framework integrating coal mining planning and ecological sensitivity zoning. The results of this study will provide new scientific support for differential restoration and policy formulation for semi-arid mining regions.