Remote sensing diagnosis of vegetation degradation and recovery in mining disturbed domain of China's core ionic rare earth mining area from 1982 to 2025

Spatially detailed knowledge of vegetation disturbance and recovery is important for effective ecological restoration in ionic adsorption rare earth mining areas. By integrating over four decades of Landsat kNDVI time series (1982–2025), we mapped mining domain using temporal detection algorithms (CCDC and LandTrendr) and deep learning (UNet), then diagnosed vegetation recovery and prioritized restoration in China's Lingbei rare earth mining area. UNet delineated domain boundaries with high accuracy (F1-scores of 0.79–0.80). The total disturbed domain was 16.1 km², showing notable spatial clustering, with 50% occurring by 2003 and 90% by 2008, mainly from November to February. The hybrid classification scheme revealed that vegetation was 78% recovering, 9% stable and 13% persistently degrading. Spatial exposure factors of distance from mines and mining density dominated the degradation risk. Restoration priorities were mapped (Tier 1:>0.39, Tier 2: 0.34–0.39, Tier 3:≤0.34), providing actionable targets for interventions of degradation.