Underground coal mining-induced subsidence threatens farmland resources and ecological sustainability in coal–grain overlapping regions with high groundwater tables, making concurrent mining and reclamation a critical management need. Previous studies have not systematically compared the integrated effects of mining sequence, extraction method, and panel optimization on subsidence control and reclamation efficiency in such regions. This study designed six mining schemes, integrating these three technical factors to investigate spatiotemporal subsidence evolution and the performance of deep digging–shallow filling reclamation. Findings reveal that mining design synergistically regulates short-to-mid-term subsidence: deep–thin seam-first skip mining eliminates initial severe subsidence damage, while shallow-thick seam-first sequential mining induces the most severe early-stage ecological disturbance. After a full extraction of both coal seams, long-term surface damage converges to 2374 ha (1509 ha severe damage), dictated by total extracted coal volume and inherent geological conditions. Reclamation efficiency depended on earthwork availability and terrain adaptability, with the optimal scheme achieving a reclamation rate of 65.00%. The findings identify mining strategies that balance subsidence mitigation and farmland restoration, providing actionable insights for sustainable mining in high-groundwater coal–grain overlapping regions.

The coal mining subsidence area constitutes a distinct ecotone in the transition from agricultural soil to sediment, yet the microbially mediated nitrogen cycle within it remains inadequately understood. This investigation comprehensively analyzed physicochemical properties, microbial communities, functional genes, and co-occurrence networks along a 0–6500 mm depth gradient. Results indicated that pH transitioned from acidic to alkaline, while TN, TP, OM, and NH₄⁺–N accumulated with depth. NO₃⁻–N decreased rapidly within 1000 mm and then stabilized. Alpha-diversity showed an S-shaped increase in richness, with Shannon index peaking at 1500 mm. Beta-diversity shifted along PC1, and the shallow subsidence area (SS) influenced by NO₃⁻–N; the transition zone (TZ) regulated by OM, TN, and NH₄⁺–N; deep subsidence area (DS) was constrained by TP and pH. Microbial communities transitioned from aerobic/facultative to strictly anaerobic phyla, yet Pseudomonadota remained dominant (24–32%) across depths. With increasing depth, gene abundances for denitrification, assimilatory nitrate reduction to ammonium (ANRA), and nitrate assimilation declined, while those for dissimilatory nitrate reduction to ammonium (DNRA) and nitrification increased; nitrogen fixation remained weak. Co-occurrence networks shifted from highly connected, short-pathlength, and clustered in TZ to highly modular and long-pathlength in DS, with Aminicenantes, Syntrophus, and Methanoregula as key taxa. Overall, the thick and stable reducing zone in the subsidence area restructured the nitrogen cycle, shifting terminal products from N₂ removal to NH₄⁺ retention. These findings advance the understanding of nitrogen transformation in soil-sediment ecotones and provide a mechanistic framework for nitrogen cycling in mining-affected ecosystems.

Annual cropland data are widely used to delimit agricultural land for soil organic carbon (SOC) mapping and spatial aggregation, yet mapped cropland may not coincide with land under agricultural use at soil sampling. Such a mismatch changes the land entering aggregation and may alter both the aggregated mean of predicted SOC and the agricultural land it represents. This difference has rarely been quantified with the SOC prediction surface held fixed. At Xinglongzhuang Coal Mine in eastern China, an agricultural sampling frame established before fieldwork was verified during March 2022 field sampling and finalized as the sampling-period agricultural-use extent (Full) for comparison with Esri 2022 Crops. Within Full, Esri-covered land was Covered and the remainder Uncovered. A random forest fitted to 155 model-development sites generated the SOC prediction surface; it was independently evaluated using 60 probability-sampled validation sites, while area of applicability (AOA) analysis assessed predictor-space support. Uncovered comprised 25.25% of Full; 86.47% occurred >10 m from the Full boundary and 97.84% was within the AOA. Independent validation yielded a root mean square error (RMSE) of 2.272 g kg⁻¹. Mean predicted SOC was 14.338 g kg⁻¹ in Covered and 8.732 g kg⁻¹ in Uncovered; Full averaged 12.922 g kg⁻¹, 1.415 g kg⁻¹ (9.9%) lower than Covered. A design-based residual correction gave a Full–Covered contrast of −1.592 g kg⁻¹ (95% confidence interval: −2.036 to −1.147 g kg⁻¹). The Full–Covered difference equals the Uncovered area proportion multiplied by the mean predicted SOC difference between Uncovered and Covered. Across the three reporting zones, these two components ranked differently, so Uncovered area proportion alone did not track the magnitude of the aggregation difference. With the prediction surface fixed, including Uncovered land with lower predicted SOC reduced the mean for Full relative to Covered. The reported mean is therefore tied to the agricultural extent used for aggregation and, in changing landscapes, to the period represented by that extent.

Mining-induced subsidence monitoring is essential for safe coal production and ecological protection in mining areas. UAV photogrammetry has become a widely adopted technique for constructing Digital Subsidence Models (DSuM); however, multi-scale composite noise significantly limits model accuracy and parameter extraction reliability. Taking the 2S201 working face of Wangjiata Coal Mine in a western arid–semi-arid region as the study area, this study systematically investigates DSuM noise characteristics and proposes a hierarchical multi-scale denoising framework. First, subsidence value interval stratification is employed to analyze the spatial distribution of noise. Based on this analysis, a two-stage strategy is developed. In the first stage, large-scale outliers are identified and removed using an improved DBSCAN algorithm with empirically calibrated and density-adaptive parameter computation. In the second stage, small-scale mixed noise is suppressed through a curvature-adaptive multi-stage denoising method. Validation using 20 ground monitoring points demonstrates that the RMSE decreases from 154 mm to 86 mm after large-scale denoising and further to 59 mm, achieving a 61.5% overall accuracy improvement. The denoised model exhibits enhanced surface continuity, smoother deformation profiles, and clearer subsidence boundaries while preserving overall deformation trends. The proposed framework effectively improves DSuM geometric accuracy and spatial consistency, providing reliable technical support for subsidence monitoring with improved accuracy in complex mining environments.