Scholarlink

<p>China is one of the countries most severely affected by landslides, posing a significant threat to densely populated urban areas along its southeastern coast. In response to the call for hazard-resilient infrastructure under Goal 9 of the SDGs, this study examined the Pearl River Delta using six machine learning models, including CART, RF, Adaboost, GBM, XGBoost, and LightGBM, combined with interpretability techniques such as Shapley additive explanation (SHAP) to produce high-accuracy landslide susceptibility mapping. Moreover, this study presents the first computational application of the Critical Infrastructure Spatial Index (CISI) to the Pearl River Delta, based on the spatial distribution of CI, and provides a systematic assessment of landslide threats to 39 types of CI. Results indicate that the LightGBM model outperformed others, with 42.50% of the study area exhibits relatively high susceptibility to landslides. Elevation and drainage density were identified as the dominant contributing factors, while precipitation and road construction were confirmed as primary triggers. More than half of the CI, particularly those related to energy, health, and waste management, faced threats from landslides, prompting the study to propose landslide prevention strategies. This study provides concrete strategies for landslide prevention and serves as a valuable reference for hazard assessment in large urban agglomerations worldwide.</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 the impact of landslide susceptibility on critical infrastructure in large urban agglomerations: using interpretable boosting models and critical infrastructure spatial index

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Mengyao Hou,
  • Aiguo Zhou,
  • Peng Huang

摘要

China is one of the countries most severely affected by landslides, posing a significant threat to densely populated urban areas along its southeastern coast. In response to the call for hazard-resilient infrastructure under Goal 9 of the SDGs, this study examined the Pearl River Delta using six machine learning models, including CART, RF, Adaboost, GBM, XGBoost, and LightGBM, combined with interpretability techniques such as Shapley additive explanation (SHAP) to produce high-accuracy landslide susceptibility mapping. Moreover, this study presents the first computational application of the Critical Infrastructure Spatial Index (CISI) to the Pearl River Delta, based on the spatial distribution of CI, and provides a systematic assessment of landslide threats to 39 types of CI. Results indicate that the LightGBM model outperformed others, with 42.50% of the study area exhibits relatively high susceptibility to landslides. Elevation and drainage density were identified as the dominant contributing factors, while precipitation and road construction were confirmed as primary triggers. More than half of the CI, particularly those related to energy, health, and waste management, faced threats from landslides, prompting the study to propose landslide prevention strategies. This study provides concrete strategies for landslide prevention and serves as a valuable reference for hazard assessment in large urban agglomerations worldwide.