Scholarlink

Abstract <p>Landslide caused catastrophic disasters frequently in the Karakoram Mountains wide range. The Hunza Valley, Pakistan in the Northwest of the Karakoram mountains, which is prone to the clustering development of landslide was taken as a case in this study. The updated complete inventory including 53 SBAS-InSAR detected active landslides and optical image interpreted 65 landslides were constructed, based on Sentinel-1A data in 2019–2020 and several field survey until 2023. Twelve factors related to geomorphology, hydrology, vegetation, geology, tectonics, and environment were incorporated into the model training within twelve machine learning models: Generalized Linear models, Navies Bayes, Nearest Neighbors, Support Vector Machines and so on. The Support Vector Classification was selected for landslide susceptibility mapping (LSM) and its characteristics in geomorphologically meaningful landscape partitions called slope units, with the highest accuracy of 0.96, average AUC of 0.99 for tenfold cross-validation, and high computational efficiency of 6.11&#xa0;s. The results revealed that the areas with moderate landslide susceptibility account for 62.14%, followed by the high susceptibility area (24.25%). The slopes of high landslide susceptibility are mainly located on the north side of the Hunza river, because of higher terrain relief, aspect, near distance to epicenters, and to fault. This research reveals that topography and tectonic activities such as relief, aspect, earthquakes and fault movements make a large contribution to landslide formation and development in Hunza, the Karakoram mountains, serving as a crucial step toward understanding and facilitating hazard management and risk reduction in the Hunza Valley and thus the uninterrupted operation of the KKH.</p> Highlights <p>(1) An integrated method for landslide susceptibility assessment in high mountainous areas of the Karakoram in High Mountain Asia.</p> <p>(2) A state-of-the-art comprehensive landslide inventory using SBAS-InSAR monitoring, optical image interpretation and field investigation was completed in Hunza Valley, Pakistan.</p> <p>(3) Comprehensive selection and application of machine learning model and inducing factors in the Hunza Valley, Northern Pakistan.</p> <p>(4) Revealed the factors controlling the landslides susceptibility and its characteristics in Hunza Valley, in the Karakoram Mountains</p> Graphical Abstract <p></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

An integrated landslide susceptibility assessment in the Karakoram Mountains based on SBAS-InSAR and machine learning: a case study of the Hunza Valley

  • 期刊论文

  • 发表日期：

    2025年5月7日

查看全文
  • Xiaojun Su,
  • Yi Zhang,
  • Xingmin Meng,
  • Mohib Ur Rehman,
  • Dongxia Yue,
  • Yan Zhao,
  • Ziqiang Zhou,
  • Fuyun Guo,
  • Qiang Zhou,
  • Baicheng Niu

摘要

Abstract

Landslide caused catastrophic disasters frequently in the Karakoram Mountains wide range. The Hunza Valley, Pakistan in the Northwest of the Karakoram mountains, which is prone to the clustering development of landslide was taken as a case in this study. The updated complete inventory including 53 SBAS-InSAR detected active landslides and optical image interpreted 65 landslides were constructed, based on Sentinel-1A data in 2019–2020 and several field survey until 2023. Twelve factors related to geomorphology, hydrology, vegetation, geology, tectonics, and environment were incorporated into the model training within twelve machine learning models: Generalized Linear models, Navies Bayes, Nearest Neighbors, Support Vector Machines and so on. The Support Vector Classification was selected for landslide susceptibility mapping (LSM) and its characteristics in geomorphologically meaningful landscape partitions called slope units, with the highest accuracy of 0.96, average AUC of 0.99 for tenfold cross-validation, and high computational efficiency of 6.11 s. The results revealed that the areas with moderate landslide susceptibility account for 62.14%, followed by the high susceptibility area (24.25%). The slopes of high landslide susceptibility are mainly located on the north side of the Hunza river, because of higher terrain relief, aspect, near distance to epicenters, and to fault. This research reveals that topography and tectonic activities such as relief, aspect, earthquakes and fault movements make a large contribution to landslide formation and development in Hunza, the Karakoram mountains, serving as a crucial step toward understanding and facilitating hazard management and risk reduction in the Hunza Valley and thus the uninterrupted operation of the KKH.

Highlights

(1) An integrated method for landslide susceptibility assessment in high mountainous areas of the Karakoram in High Mountain Asia.

(2) A state-of-the-art comprehensive landslide inventory using SBAS-InSAR monitoring, optical image interpretation and field investigation was completed in Hunza Valley, Pakistan.

(3) Comprehensive selection and application of machine learning model and inducing factors in the Hunza Valley, Northern Pakistan.

(4) Revealed the factors controlling the landslides susceptibility and its characteristics in Hunza Valley, in the Karakoram Mountains

Graphical Abstract