Surface Subsidence Monitoring and Interpretable Factor Analysis in Coal Mining Areas of Henan Province Based on SBAS-InSAR

Abstract

Henan Province, a major coal producing region in China, faces severe surface subsidence induced by extensive underground mining, which compromises regional ecological security and infrastructure stability. In this study, small baseline subset interferometric synthetic aperture radar (SBAS-InSAR) was applied to Sentinel-1A imagery acquired from March 2017 to February 2025 to characterize surface deformation in concentrated coal mining areas. A local validation was conducted within a representative mining area in Study Area 3 using measurements from 14 leveling benchmarks acquired between 5 May and 20 July 2023. The comparison yielded an R2 of 0.816 and an RMSE of 9.22 mm, indicating good agreement between the SBAS-InSAR and leveling measurements during the validation interval. The subsidence in the study area exhibits significant spatial heterogeneity and continuous accumulation characteristics. The most negative approximate vertically projected deformation rate reached −371 mm/yr, and the maximum cumulative displacement reached −2101 mm. Scenario-based sensitivity analysis indicated potential projection errors of 6.76–8.34% for a horizontal-to-vertical displacement ratio of 0.10 and 20.28–25.01% for a ratio of 0.30, with larger uncertainty expected near subsidence trough margins. Given the difficulty of quantifying large-scale underground mining parameters, this study employs multisource environmental and topographic variables as auxiliary indicators and develops an XGBoost-SHAP model to evaluate their relative explanatory contributions to the spatial heterogeneity of mining-induced subsidence. Among the selected measurable environmental and topographic variables, groundwater table depth represents the most important measurable explanatory factor for the spatial heterogeneity of subsidence, with distinct response patterns between plain areas with thick unconsolidated layers and piedmont bedrock regions. Furthermore, wavelet coherence analysis identifies scale-dependent spatial associations between topography and subsidence. At the regional scale, elevation exhibits spatial correspondence with the geomorphological framework of contiguous subsidence basins. At the local scale, slope and aspect show localized associations with differential deformation gradients near the margins of subsidence troughs.