Key points are not available for this paper at this time.
Rainfall-induced shallow landslides strongly affect slope stability and hazard potential in mountainous areas. However, the spatiotemporal evolution of landslide scars under repeated rainfall events remains poorly understood. Using Beijing mountainous areas as a case study, we combined remote sensing, time-series NDVI analysis, and a visual foundation model to quantify landslide scar evolution and identify its controlling mechanisms. Two typical patterns have been found. Pattern I follows a “decline–outburst–overcompensation–scarring” sequence: pre-event NDVI declines by 25.1%; during the event, NDVI drops to extreme lows and 21.7% of pixels are masked; after the event, surviving vegetation shows 8.5% overcompensatory growth, but permanent scars form. Pattern II follows a “growth–acceleration–stabilization–masking” sequence: pre-event NDVI increases by 13.6%, reducing landslide risk; rainfall drives NDVI to a peak (+23.4%); post-event NDVI remains high, and landslide areas account for only 0.53%, with damage masked within a new, higher steady state. These findings demonstrate that topographic conditions, vegetation type, and phenological stage jointly control landslide scar characteristics. Steep slopes with shallow-rooted vegetation tend toward Pattern I (explicit damage, persistent scars), while gentle slopes with vegetation in active growing seasons tend toward Pattern II (masked damage, rapid recovery). Pre-event NDVI anomalies provide identifiable precursory information and should be incorporated into early warning and risk assessment systems.
Mu et al. (Fri,) studied this question.