Scholarlink

<p>The present study investigates landslide susceptibility in the Bachchangad catchment area, Alaknanda basin of Uttarakhand Himalaya, India. Susceptibility zonation has been done with a conventional Logistic Regression (LR) and smarter Random Forest (RF) algorithm. A landslide inventory of 106 events was compiled using field surveys and satellite imagery. Thirteen causative factors, including distance to streams, lithology, slope, rainfall, and elevation, were assessed for the susceptibility study. The RF model outperformed LR, yielding AUC values of 78% (training) and 76.6% (testing), compared to 71.3% (training) and 72.1% (testing) for LR. Lithology (24.5%), relative relief (20.2%), and elevation (18.4%) were identified as the most influential predictors. Further, kinematic analysis was also carried out for a holistic geotechnical assessment. At four key sites, Fatehpur, Kandai Market, Khankara Bridge, and proximity of Dewansh Hotel, structural instability was observed. This integrated machine learning and kinematic approach was found to enhance the accuracy of susceptibility mapping. The proposed methodology yields a robust framework for landslide risk assessment, disaster management, and sustainable planning in hilly 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

Landslide susceptibility mapping and kinematic analysis in Bachchangad catchment, Uttarakhand, India: a comparative study using machine learning benchmark classifiers

  • 期刊论文

  • 发表日期：

    2026年2月26日

查看全文
  • Khyati Gupta,
  • Aasif Ibni Ahad,
  • Devendra Singh Rawat,
  • Syed Kaiser Bukhari,
  • Bikash Kumar Ram,
  • Govind Singh Rawat,
  • Vikas Yadav,
  • Ashutosh Kainthola,
  • Tariq Siddique,
  • Mohammad Azarafza

摘要

The present study investigates landslide susceptibility in the Bachchangad catchment area, Alaknanda basin of Uttarakhand Himalaya, India. Susceptibility zonation has been done with a conventional Logistic Regression (LR) and smarter Random Forest (RF) algorithm. A landslide inventory of 106 events was compiled using field surveys and satellite imagery. Thirteen causative factors, including distance to streams, lithology, slope, rainfall, and elevation, were assessed for the susceptibility study. The RF model outperformed LR, yielding AUC values of 78% (training) and 76.6% (testing), compared to 71.3% (training) and 72.1% (testing) for LR. Lithology (24.5%), relative relief (20.2%), and elevation (18.4%) were identified as the most influential predictors. Further, kinematic analysis was also carried out for a holistic geotechnical assessment. At four key sites, Fatehpur, Kandai Market, Khankara Bridge, and proximity of Dewansh Hotel, structural instability was observed. This integrated machine learning and kinematic approach was found to enhance the accuracy of susceptibility mapping. The proposed methodology yields a robust framework for landslide risk assessment, disaster management, and sustainable planning in hilly regions.