Scholarlink

<p>Population growth and expanding infrastructure in landslide-prone areas have intensified human and economic losses worldwide. Landslide susceptibility mapping is widely used to support risk mitigation; however, identifying the optimal combination of conditioning factors remains challenging, especially in large-scale, heterogeneous regions. To this end, this study develops a well-structured workflow to improve factor selection analysis and the accuracy of susceptibility modeling in the Alpine Central Rif Chain of northern Morocco, a geologically diverse and highly landslide-prone environment. We first established an exhaustive landslide inventory map covering the entire study area to enhance the reliability of the results. We then applied multicollinearity diagnostics to 11 initial factors and compared two factor-selection approaches, bivariate Pearson correlation and multivariate multi-criteria analysis, to identify the most influential variables. The latter approach proved more effective in capturing complex interactions between slope dynamics and controlling factors. Using the selected factors, three susceptibility models were computed: GIS Matrix Method (GMM), Logistic Regression (LR), and Artificial Neural Networks (ANN). The results highlight the importance of applying multiple susceptibility approaches, as each method exhibits different sensitivities to variations in input data in such heterogeneous geological contexts.</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 assessment framework in the central Rif chain, Northern Morocco

  • 期刊论文

  • 发表日期：

    2026年1月14日

查看全文
  • Oussama Obda,
  • Reda Sahrane,
  • Ilias Obda,
  • Younes El Kharim

摘要

Population growth and expanding infrastructure in landslide-prone areas have intensified human and economic losses worldwide. Landslide susceptibility mapping is widely used to support risk mitigation; however, identifying the optimal combination of conditioning factors remains challenging, especially in large-scale, heterogeneous regions. To this end, this study develops a well-structured workflow to improve factor selection analysis and the accuracy of susceptibility modeling in the Alpine Central Rif Chain of northern Morocco, a geologically diverse and highly landslide-prone environment. We first established an exhaustive landslide inventory map covering the entire study area to enhance the reliability of the results. We then applied multicollinearity diagnostics to 11 initial factors and compared two factor-selection approaches, bivariate Pearson correlation and multivariate multi-criteria analysis, to identify the most influential variables. The latter approach proved more effective in capturing complex interactions between slope dynamics and controlling factors. Using the selected factors, three susceptibility models were computed: GIS Matrix Method (GMM), Logistic Regression (LR), and Artificial Neural Networks (ANN). The results highlight the importance of applying multiple susceptibility approaches, as each method exhibits different sensitivities to variations in input data in such heterogeneous geological contexts.