Scholarlink

<p>This study presents a novel workflow for developing spatiotemporal Landslide Hazard Maps (LHMs) using publicly available datasets and physics-based modeling, supplemented with machine learning–based Landslide Susceptibility Maps (LSMs), for shallow colluvial soils in eastern Kentucky. LHMs were constructed using a limit equilibrium infinite slope factor of safety (FS) equation for unsaturated conditions, integrating soil properties from the NRCS Web Soil Survey, high-resolution geomorphic variables from 1.5&#xa0;m LiDAR DEMs, and soil moisture dynamics simulated with HYDRUS-1D, driven by precipitation and evapotranspiration data. Validation against three documented landslide events demonstrated that LHMs consistently identified low FS values on failure dates, capturing both temporal and spatial patterns of instability, with false positives primarily in steep, rocky slopes. Parallel LSMs were developed using a bagged trees algorithm trained on soil, hydrologic, and geomorphic variables, achieving an AUC of 0.88. Comparative analysis revealed that LHMs provided superior event-specific performance with fewer false positives, while LSMs offered broader spatial susceptibility assessment. The findings highlight the complementary roles of LHMs for near real-time hazard monitoring and LSMs for long-term susceptibility evaluation. The proposed workflow supports operational landslide risk assessment and offers a framework for integrating dynamic hydrologic modeling with data-driven classification to enhance predictive capability in landslide-prone regions.</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

Workflow for Developing land Slide Hazard Maps for Shallow Colluvial Soils

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Nathaniel O’Leary,
  • L. Sebastian Bryson,
  • Matthew M. Crawford,
  • Jason M. Dortch

摘要

This study presents a novel workflow for developing spatiotemporal Landslide Hazard Maps (LHMs) using publicly available datasets and physics-based modeling, supplemented with machine learning–based Landslide Susceptibility Maps (LSMs), for shallow colluvial soils in eastern Kentucky. LHMs were constructed using a limit equilibrium infinite slope factor of safety (FS) equation for unsaturated conditions, integrating soil properties from the NRCS Web Soil Survey, high-resolution geomorphic variables from 1.5 m LiDAR DEMs, and soil moisture dynamics simulated with HYDRUS-1D, driven by precipitation and evapotranspiration data. Validation against three documented landslide events demonstrated that LHMs consistently identified low FS values on failure dates, capturing both temporal and spatial patterns of instability, with false positives primarily in steep, rocky slopes. Parallel LSMs were developed using a bagged trees algorithm trained on soil, hydrologic, and geomorphic variables, achieving an AUC of 0.88. Comparative analysis revealed that LHMs provided superior event-specific performance with fewer false positives, while LSMs offered broader spatial susceptibility assessment. The findings highlight the complementary roles of LHMs for near real-time hazard monitoring and LSMs for long-term susceptibility evaluation. The proposed workflow supports operational landslide risk assessment and offers a framework for integrating dynamic hydrologic modeling with data-driven classification to enhance predictive capability in landslide-prone regions.