Scholarlink

<p>Gaochang District of Turpan is a representative extremely arid region at the southern foot of the Tianshan Mountains. The “three mountains and two basins” geomorphology and fragile environment cause frequent geohazards, making fine-scale susceptibility mapping essential for disaster mitigation and spatial planning. This study established an evaluation index system to develop a hybrid AHP-INF model, which was compared with six machine learning models (e.g., RF, SVM, and CatBoost). ROC curves and Frequency Ratio (FR) were utilized to evaluate model performance and analyze geological control mechanisms. The results show that the AHP-INF model performs best, with an AUC of 0.930 and an FR of 22.717 in high-susceptibility zones, demonstrating a significant “small area, high coverage” advantage. Regional geological control effects are prominent; erosional tectonic middle-high mountains (FR = 2.079) and stratified clastic rocks (FR = 4.813) are the core physical foundations of hazards. The model’s zonation aligns highly with geological logic. The assessment accurately identifies a core high-susceptibility zone covering 0.31% of the study area, enabling precise spatial positioning of potential hazards. This study provides a scientific framework for refined geohazard prevention in Gaochang and similar arid 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

Geohazard susceptibility assessment in gaochang district: a comparative study based on AHP-INF and multiple models

  • 期刊论文

  • 发表日期：

    2026年7月28日

查看全文
  • Lifeng Chen,
  • Zizhao Zhang,
  • Kai Chen,
  • Jinyu Chang,
  • Liu Zhiqi,
  • Mengmeng Zhao,
  • Yuhang Huang,
  • Xiaowei Lu,
  • Peng Huang

摘要

Gaochang District of Turpan is a representative extremely arid region at the southern foot of the Tianshan Mountains. The “three mountains and two basins” geomorphology and fragile environment cause frequent geohazards, making fine-scale susceptibility mapping essential for disaster mitigation and spatial planning. This study established an evaluation index system to develop a hybrid AHP-INF model, which was compared with six machine learning models (e.g., RF, SVM, and CatBoost). ROC curves and Frequency Ratio (FR) were utilized to evaluate model performance and analyze geological control mechanisms. The results show that the AHP-INF model performs best, with an AUC of 0.930 and an FR of 22.717 in high-susceptibility zones, demonstrating a significant “small area, high coverage” advantage. Regional geological control effects are prominent; erosional tectonic middle-high mountains (FR = 2.079) and stratified clastic rocks (FR = 4.813) are the core physical foundations of hazards. The model’s zonation aligns highly with geological logic. The assessment accurately identifies a core high-susceptibility zone covering 0.31% of the study area, enabling precise spatial positioning of potential hazards. This study provides a scientific framework for refined geohazard prevention in Gaochang and similar arid regions.