Ecological Risk Dynamics and Associated Factors Revealed by Remotely Sensed Energy Extraction Footprints on China's Loess Plateau

Energy extraction creates substantial land-surface disturbances in ecologically fragile regions, yet its spatial footprints remain insufficiently represented in regional ecological risk assessments. This study mapped energy extraction footprints in Qingyang, a major energy-producing region on China's Loess Plateau, using Sentinel-2 imagery, random forest classification, and high-resolution image interpretation for 2017 and 2025. An integrated ecological risk index (ERI) was developed, and disturbance–risk relationships and associated environmental factors were examined using spatial analysis and XGBoost–SHAP–PDP. Energy extraction land expanded from 34.04 to 63.65 km² (86.98%), while the number of patches increased by 69.78%. Mean ERI increased from 0.1425 to 0.1483, and the proportion of relatively high- and high-risk areas rose from 14.73% to 22.27%. Newly disturbed areas exhibited a clear spatial association with ecological risk increases. Hydroclimatic variables remained important predictors, while the combined relative importance of nighttime lights and PM2.5 increased from 28.7% to 38.3%. Nonlinear responses and changing feature interactions further highlighted the combined relevance of environmental conditions and human activities. These findings demonstrate the value of explicitly mapping energy extraction footprints for identifying localized ecological risks and supporting spatially targeted environmental management.