Scholarlink

<p>Landslides are among the most dangerous geological disasters, always present along mountainous highways. Damage caused by landslides to road infrastructure and residential areas along these routes will increase the public investment and construction costs. Therefore, proactively assessing the hazard and exposure level of infrastructure and residents to landslides is extremely necessary. This study proposes a deep neural network (DNN) model and geospatial analysis techniques to assess landslide susceptibility and quantify housing and infrastructure exposure along National Highway 279 in Bac Kan Province, Vietnam. The DNN model was developed on Google Colab to build a landslide susceptibility map with a forecast accuracy of 91.1%. Then, this map was integrated with infrastructure and residential data to estimate potential exposure levels. The exposure levels of these objects are only counted in high and very high landslide susceptibility areas. The statistical results in descending order show that the number of culverts potentially exposed to landslides belongs to Ba Be district &gt; Na Ri district &gt; Ngan Son district. The number of traffic signs and convex mirrors potentially exposed to landslides belongs to Na Ri district &gt; Ngan Son district &gt; Ba Be district. The number of houses potentially exposed to landslides located within the 50&#xa0;m and 100&#xa0;m buffer zones belongs to Ba Be district &gt; Ngan Son district &gt; Na Ri district. These results provide important information to support local authorities in proactively implementing solutions to reinforce and enhance infrastructure resilience. This approach can be applied to mountainous areas with similar terrain and geological conditions to support sustainable transport development.</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

Deep neural network-based landslide hazard assessment and infrastructure exposure mapping: a case study of National Highway 279, Bac Kan province, Vietnam

  • 期刊论文

  • 发表日期：

    2025年11月18日

查看全文
  • Quynh Duy Bui,
  • Hang Ha,
  • Tung Hoang,
  • Viet-Phuong Nguyen,
  • Chinh Luu

摘要

Landslides are among the most dangerous geological disasters, always present along mountainous highways. Damage caused by landslides to road infrastructure and residential areas along these routes will increase the public investment and construction costs. Therefore, proactively assessing the hazard and exposure level of infrastructure and residents to landslides is extremely necessary. This study proposes a deep neural network (DNN) model and geospatial analysis techniques to assess landslide susceptibility and quantify housing and infrastructure exposure along National Highway 279 in Bac Kan Province, Vietnam. The DNN model was developed on Google Colab to build a landslide susceptibility map with a forecast accuracy of 91.1%. Then, this map was integrated with infrastructure and residential data to estimate potential exposure levels. The exposure levels of these objects are only counted in high and very high landslide susceptibility areas. The statistical results in descending order show that the number of culverts potentially exposed to landslides belongs to Ba Be district > Na Ri district > Ngan Son district. The number of traffic signs and convex mirrors potentially exposed to landslides belongs to Na Ri district > Ngan Son district > Ba Be district. The number of houses potentially exposed to landslides located within the 50 m and 100 m buffer zones belongs to Ba Be district > Ngan Son district > Na Ri district. These results provide important information to support local authorities in proactively implementing solutions to reinforce and enhance infrastructure resilience. This approach can be applied to mountainous areas with similar terrain and geological conditions to support sustainable transport development.