Scholarlink

<p>In the context of landslide susceptibility mapping and geological disaster risk mitigation, this study develops an efficient computational framework for regional landslide susceptibility assessment (LSA) based on the probabilistic physical modelling and geotechnical reliability updating approach (RUA). As a core part of LSA, the fast computation of spatially distributed probability of landslide at a regional scale is a necessity. Specifically, the RUA technique is firstly explored, of which the core innovation lies in its use of low-discrepancy sequences to construct a physical modelling-based sample pool that can effectively cover the entire uncertainty space. The weight index is introduced to quantify each sample’s contribution to the&#xa0;failure probability. Considering landslide probability changes over the regional space&#xa0;due to varying uncertainties, the proposed RUA framework facilitates rapid LSA (probability) updating for all landslide mapping units without much extra computational cost in terms of physical modelling. In comparison with the first-order reliability method (FORM) and Monte Carlo simulation (MCS), the proposed RUA achieves comparable LSA accuracy with significantly higher computational efficiency. For example, RUA can achieve an area under the curve (AUC) of 0.783 for LSA in only 3&#xa0;min, whereas MCS requires 55&#xa0;h to achieve an AUC of 0.781. In conclusion, this study refines the theoretical framework for fast regional LSA while maintaining accuracy, eliminating redundant computations associated with traditional random sampling methods, and providing a pathway for real-time regional landslide risk assessment.</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

A novel perspective on fast landslide susceptibility mapping at regional scale: Probabilistic modelling and reliability updating

  • 期刊论文

  • 发表日期：

    2026年2月21日

查看全文
  • Bin Tong,
  • Jian Ji,
  • Tao Wang,
  • Hongzhi Cui

摘要

In the context of landslide susceptibility mapping and geological disaster risk mitigation, this study develops an efficient computational framework for regional landslide susceptibility assessment (LSA) based on the probabilistic physical modelling and geotechnical reliability updating approach (RUA). As a core part of LSA, the fast computation of spatially distributed probability of landslide at a regional scale is a necessity. Specifically, the RUA technique is firstly explored, of which the core innovation lies in its use of low-discrepancy sequences to construct a physical modelling-based sample pool that can effectively cover the entire uncertainty space. The weight index is introduced to quantify each sample’s contribution to the failure probability. Considering landslide probability changes over the regional space due to varying uncertainties, the proposed RUA framework facilitates rapid LSA (probability) updating for all landslide mapping units without much extra computational cost in terms of physical modelling. In comparison with the first-order reliability method (FORM) and Monte Carlo simulation (MCS), the proposed RUA achieves comparable LSA accuracy with significantly higher computational efficiency. For example, RUA can achieve an area under the curve (AUC) of 0.783 for LSA in only 3 min, whereas MCS requires 55 h to achieve an AUC of 0.781. In conclusion, this study refines the theoretical framework for fast regional LSA while maintaining accuracy, eliminating redundant computations associated with traditional random sampling methods, and providing a pathway for real-time regional landslide risk assessment.