Scholarlink

<p>This study has evaluated the size-frequency distribution and factor effects on the landslide susceptibility modelling in the Himalayan region of central Nepal. The landslide inventory includes 7096 landslides, covering an area of 49.37&#xa0;km<sup>2</sup>. The frequency-area distribution (FAD) has been computed to analyse the spatial variation of landslides using a historical landslide database. Determination of the FAD scaling factors such as cutoff point and power exponent values of inventory dataset has proven the dominancy of their sizes in the overall process of occurrence and landslide susceptibility modelling. The conditional independence among different contributing factors is examined from the Chi-square test to assess their effects on landslide susceptibility modelling processes. The susceptibility models are prepared by adopting conditionally independent factors and further tested statistically to predict the performance of models and validation. In this study, the calculated power exponent is about − 2.3 in the noncumulative FAD curve which correlates well with a power-law relation within an area (<i>A</i><sub><i>L</i></sub>) range of 0.03&#xa0;km<sup>2</sup> &lt; <i>A</i><sub><i>L</i></sub> &lt; 1.02&#xa0;km<sup>2</sup>. An optimal prediction model for overall landslides derived from the combination of factors such as slope, aspect, elevation, curvature, flow direction, stream density, road proximity, and lithology, provides an area under the receiver-operating curve with a success rate of 0.77, within the valid prediction accuracy range. The computed models of large-scale landslide susceptibility have revealed a distinctive influence of geomorphological (slope, aspect, elevation) and geological (thrust proximity, lithology) factors, contributing to the significant control on the genesis pertinent to large-scale landslides.</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 Size-Frequency Distribution and Factor Effects on Susceptibility Modelling in the Himalayan Region, Central Nepal

  • 期刊论文

  • 发表日期：

    2024年12月31日

查看全文
  • Bikash Phuyal,
  • Madhu S. Acharya,
  • Prem B. Thapa

摘要

This study has evaluated the size-frequency distribution and factor effects on the landslide susceptibility modelling in the Himalayan region of central Nepal. The landslide inventory includes 7096 landslides, covering an area of 49.37 km2. The frequency-area distribution (FAD) has been computed to analyse the spatial variation of landslides using a historical landslide database. Determination of the FAD scaling factors such as cutoff point and power exponent values of inventory dataset has proven the dominancy of their sizes in the overall process of occurrence and landslide susceptibility modelling. The conditional independence among different contributing factors is examined from the Chi-square test to assess their effects on landslide susceptibility modelling processes. The susceptibility models are prepared by adopting conditionally independent factors and further tested statistically to predict the performance of models and validation. In this study, the calculated power exponent is about − 2.3 in the noncumulative FAD curve which correlates well with a power-law relation within an area (AL) range of 0.03 km2 < AL < 1.02 km2. An optimal prediction model for overall landslides derived from the combination of factors such as slope, aspect, elevation, curvature, flow direction, stream density, road proximity, and lithology, provides an area under the receiver-operating curve with a success rate of 0.77, within the valid prediction accuracy range. The computed models of large-scale landslide susceptibility have revealed a distinctive influence of geomorphological (slope, aspect, elevation) and geological (thrust proximity, lithology) factors, contributing to the significant control on the genesis pertinent to large-scale landslides.