Scholarlink

<p>Landslides in mountainous regions, such as Son La province in northern Vietnam, pose significant risks to human life, property, and infrastructure. Rapid urbanization and deforestation in Vietnam exacerbate landslide risks, making effective landslide risk assessment crucial for disaster mitigation and management. This study introduces a new integrated framework combining machine learning models and multi-criteria decision analysis to estimate landslide risk. We utilized 1,771 landslide locations from various sources and fifteen landslide influencing factors as model input to create landslide susceptibility maps using advanced hybrid machine learning models. The Analytic Hierarchy Process is used to weight indicators of social-economic and infrastructure impacts, resulting in a comprehensive landslide risk assessment map. The final risk map showed the landslide susceptibility and consequences in a matrix, highlighting that 3.25% of the area is at very high risk, 14.25% at high risk, 23.46% at medium risk, and 59.05% at low and very low risk. This study emphasizes the consideration of landslide consequence indicators in landslide risk modelling to more accurately reflect potential loss degrees. This comprehensive approach enhances our understanding of the physical and socio-economic impacts, thereby significantly contributing to landslide mitigation and adaptation strategies.</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

Landslide risk assessment using an integrated framework of machine learning algorithms and multi-criteria decision analysis

  • 期刊论文

  • 发表日期：

    2025年8月18日

查看全文
  • Hang Ha,
  • Chinh Luu,
  • Quynh Duy Bui,
  • Viet-Phuong Nguyen,
  • Ngoc-Tru Vu,
  • Matthieu Kervyn

摘要

Landslides in mountainous regions, such as Son La province in northern Vietnam, pose significant risks to human life, property, and infrastructure. Rapid urbanization and deforestation in Vietnam exacerbate landslide risks, making effective landslide risk assessment crucial for disaster mitigation and management. This study introduces a new integrated framework combining machine learning models and multi-criteria decision analysis to estimate landslide risk. We utilized 1,771 landslide locations from various sources and fifteen landslide influencing factors as model input to create landslide susceptibility maps using advanced hybrid machine learning models. The Analytic Hierarchy Process is used to weight indicators of social-economic and infrastructure impacts, resulting in a comprehensive landslide risk assessment map. The final risk map showed the landslide susceptibility and consequences in a matrix, highlighting that 3.25% of the area is at very high risk, 14.25% at high risk, 23.46% at medium risk, and 59.05% at low and very low risk. This study emphasizes the consideration of landslide consequence indicators in landslide risk modelling to more accurately reflect potential loss degrees. This comprehensive approach enhances our understanding of the physical and socio-economic impacts, thereby significantly contributing to landslide mitigation and adaptation strategies.