Scholarlink

<p>This paper takes the extreme rainfall event in eastern Guangdong in August 2018 as a case study. We systematically compiled an inventory of 6194 rainfall-triggered landslides and combined the AutoGluon framework with SHapley Additive exPlanations (SHAP) analysis to reveal the multi-factor coupling mechanisms governing landslide occurrence. The research identifies low-to-medium elevation (74–374 m), gentle slopes (10–30°), high vegetation coverage (NDVI 210–255), and cumulative rainfall exceeding 380&#xa0;mm as key characteristics of high-incidence landslide zones. Landslides were particularly pronounced in weathered Jurassic granite formations and within 2000 m of rivers. Landslide distribution is significantly controlled by the synergistic interaction of four dominant factors: elevation, cumulative rainfall, slope angle, and normalized difference vegetation index (NDVI). Binary and ternary factor coupling analyses further indicate that low-to-medium elevation topography enhances local rainfall intensity through the ‘windward slope rainfall amplification effect,’ while the combination of colluvial deposits on gentle slopes and root channels in vegetated areas accelerates rainwater infiltration. This process disrupts the critical equilibrium state of slopes, triggering landslides. The research proposes two ternary threshold frameworks: ‘low-to-medium elevation–gentle slope–intense rainfall’ and ‘low-to-medium elevation–gentle slope–high vegetation coverage’. These frameworks provide a scientific basis for rainfall-induced cluster landslide risk assessment and early warning systems in coastal South China.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Analysis of a comprehensive inventory of rainfall-induced landslides and multi-factor coupling mechanisms in Eastern Guangdong, China, in August 2018

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Chenchen Xie,
  • Chong Xu,
  • Xiwei Xu,
  • Yuandong Huang,
  • Huiran Gao

摘要

This paper takes the extreme rainfall event in eastern Guangdong in August 2018 as a case study. We systematically compiled an inventory of 6194 rainfall-triggered landslides and combined the AutoGluon framework with SHapley Additive exPlanations (SHAP) analysis to reveal the multi-factor coupling mechanisms governing landslide occurrence. The research identifies low-to-medium elevation (74–374 m), gentle slopes (10–30°), high vegetation coverage (NDVI 210–255), and cumulative rainfall exceeding 380 mm as key characteristics of high-incidence landslide zones. Landslides were particularly pronounced in weathered Jurassic granite formations and within 2000 m of rivers. Landslide distribution is significantly controlled by the synergistic interaction of four dominant factors: elevation, cumulative rainfall, slope angle, and normalized difference vegetation index (NDVI). Binary and ternary factor coupling analyses further indicate that low-to-medium elevation topography enhances local rainfall intensity through the ‘windward slope rainfall amplification effect,’ while the combination of colluvial deposits on gentle slopes and root channels in vegetated areas accelerates rainwater infiltration. This process disrupts the critical equilibrium state of slopes, triggering landslides. The research proposes two ternary threshold frameworks: ‘low-to-medium elevation–gentle slope–intense rainfall’ and ‘low-to-medium elevation–gentle slope–high vegetation coverage’. These frameworks provide a scientific basis for rainfall-induced cluster landslide risk assessment and early warning systems in coastal South China.