Object‑oriented landslide identification and susceptibility assessment for the Jiuzhaigou earthquake by integrating NDVI difference

A groundbreaking study identifies and maps earthquake-triggered landslides (EqTLs) with unprecedented accuracy in the Jiuzhaigou Valley. By integrating object-oriented detection with a Random Forest susceptibility model—enhanced by Normalized Difference Vegetation Index (NDVI) differences—the research pinpoints 841 landslides (covering 11.96 km²), primarily concentrated near the Huya Fault and on steep slopes (30–50°). The model achieved high accuracy (AUC = 0.98), highlighting fault distance, peak ground acceleration (PGA), and slope angle as critical controls. Notably, 95% of landslides were found in very high susceptibility zones, with a density of 2.83/km². This approach offers a robust framework for spatial analysis and risk management in alpine valley regions, advancing our understanding of seismic-induced geological hazards.

Tang, Q., Qin, Y., Lai, Y., Lu, Z., Li, Y., Wang, R., & Li, F. (2026). Object‑oriented landslide identification and susceptibility assessment for the Jiuzhaigou earthquake by integrating NDVI difference. Geocarto International, 41(1). https://doi.org/10.1080/10106049.2026.2614145