Scholarlink

<p>Mapping and classification of active landslides are essential for landslide risk management. However, current manual visual interpretation methods face the bottleneck of high costs and low efficiency in updating wide-area landslide inventories. This study combined Interferometric Synthetic Aperture Radar (InSAR) with deep learning technology to detect and semi-automatically classify active landslides in the upper reaches of the Yellow River on the eastern Tibetan Plateau between October 2014 and June 2023. Results suggest that the deformation velocity obtained by the Generic Atmospheric Correction Online Service Assisted Stacking-InSAR method exhibits good consistency in the overlapping observation areas of adjacent datasets. A total of 473 active landslides were detected, and the reliability of the detection results was verified through visual interpretation of optical imagery. Spatial distribution analysis of the landslides revealed a distinct belt-like distribution pattern due to long-term river erosion and fault tectonic activities. Furthermore, based on our proposed two-dimensional velocity vector inclination in four quadrants and machine learning technology, we achieved the classification of translational landslides, rotational landslides, and roto-translational landslides. The significant differences in their deformation velocities further illustrate the distinct deformation characteristics. This study achieved wide-area detection and semi-automatic classification of active landslides through technology integration, laying a theoretical foundation for identification of geological hazards.</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

Detection and classification of active landslides along the Yellow River corridor using InSAR

  • 期刊论文

  • 发表日期：

    2026年5月22日

查看全文
  • Yaru Zhu,
  • Haijun Qiu,
  • Zijing Liu,
  • Yijun Li,
  • Wenchao Huangfu,
  • Wanqi Luo,
  • Kailiang Zhao,
  • Yiwen Liang

摘要

Mapping and classification of active landslides are essential for landslide risk management. However, current manual visual interpretation methods face the bottleneck of high costs and low efficiency in updating wide-area landslide inventories. This study combined Interferometric Synthetic Aperture Radar (InSAR) with deep learning technology to detect and semi-automatically classify active landslides in the upper reaches of the Yellow River on the eastern Tibetan Plateau between October 2014 and June 2023. Results suggest that the deformation velocity obtained by the Generic Atmospheric Correction Online Service Assisted Stacking-InSAR method exhibits good consistency in the overlapping observation areas of adjacent datasets. A total of 473 active landslides were detected, and the reliability of the detection results was verified through visual interpretation of optical imagery. Spatial distribution analysis of the landslides revealed a distinct belt-like distribution pattern due to long-term river erosion and fault tectonic activities. Furthermore, based on our proposed two-dimensional velocity vector inclination in four quadrants and machine learning technology, we achieved the classification of translational landslides, rotational landslides, and roto-translational landslides. The significant differences in their deformation velocities further illustrate the distinct deformation characteristics. This study achieved wide-area detection and semi-automatic classification of active landslides through technology integration, laying a theoretical foundation for identification of geological hazards.