Detecting disturbance and recovery in mining landscapes: A novel time-series framework based on improved LandTrendr...

Mining production has driven human development but has also led to sustained ecological damage. Vegetation serves as a critical carrier and indicator of ecosystem conditions in resource-extraction zones. Therefore, monitoring vegetation sustainability in such regions is crucial for both environmental conservation and regional growth. Leveraging the Google Earth Engine (GEE), this study employed trend analysis and an improved LandTrendr algorithm to quantitatively assess the spatiotemporal characteristics of vegetation changes (greening and browning) and abrupt changes (disturbance and recovery) in the Shizhuyuan mining zone and surrounding areas in southern China from 1997 to 2020. The enhanced LandTrendr algorithm resolved the common issue of vegetation recovery being erroneously detected before disturbance, making it particularly suitable for vegetation monitoring in mined landscapes. Additionally, machine learning algorithms were applied to classify vegetation change patterns. The results indicate a general greening trend across the region, with localized browning driven by human activities. Disturbances were primarily concentrated in 1999, 2003, and 2012, with disturbance and recovery processes following a distinct temporal sequence. Over 60% of the disturbed vegetation within the typical mining zone has been restored. Spoil heaps nearest to urban centers showed the most notable recovery, with over 64% of the area achieving high-quality restoration, exhibiting clear spatiotemporal characteristics of damage and recovery, reflecting targeted artificial reclamation efforts. This research provides a practical monitoring approach for tracking vegetation dynamics in mining-affected regions, offering valuable support for ecological studies and the development of environmental management strategies.