Scholarlink

<p>The northwestern of Setif (Algeria) is particularly prone to landslide occurrence due to its steep topography, weak facies, and anthropogenic pressure. This study provides an integrated assessment of landslide susceptibility by combining multi-source geospatial data with statistical modelling approaches. A landslide inventory comprising 61 events was developed through the interpretation of high-resolution satellite imagery, virtual globe platforms, published scientific sources, and field validation supported by GPS measurements. Nine conditioning factors were considered, including slope gradient, slope aspect, lithology, elevation, rainfall, land use/land cover, and distances to rivers, roads, and faults. Landslide susceptibility was modelled using four bivariate statistical techniques: Frequency Ratio (FR), Information Value (IV), Weighted Factor (WF), and Weight of Evidence (WoE). The resulting susceptibility maps were classified into five categories ranging from very low to very high susceptibility. Model performance was evaluated using Receiver Operating Characteristic (ROC) analysis and the Area Under the Curve (AUC). All models exhibited strong predictive capability, with AUC values indicating well to excellent performance. The WoE model achieved the highest accuracy (AUC = 0.925), followed closely by the IV model (AUC = 0.905), confirming the effectiveness of probabilistic approaches in reproducing the spatial distribution of landslide occurrence. Spatial analysis consistently shows that high and very high susceptibility zones are mainly associated with steep slopes, weak marly and clay-rich lithologies, higher rainfall regimes, and proximity to drainage networks and road infrastructures. These convergent patterns across models emphasize the dominant role of both geomorphological controls and human-induced modifications in slope instability. The results highlight the added value of probabilistic modelling frameworks in capturing complex spatial relationships between conditioning factors and landslide occurrence. The proposed approach provides a reliable basis for susceptibility mapping and offers practical support for risk reduction strategies, land-use planning, and monitoring prioritization in Mediterranean environments.</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 prediction in a mediterranean environment using multi-source geospatial data and bivariate probabilistic methods

  • 期刊论文

  • 发表日期：

    2026年7月30日

查看全文
  • Djamel Eddine Abbas,
  • Riadh Boukarm,
  • Riheb Hadji,
  • Younes Hamed,
  • Arif Ismail,
  • Matteo Gentilucci,
  • Maurizio Barbieri,
  • Dehni Abdellatif

摘要

The northwestern of Setif (Algeria) is particularly prone to landslide occurrence due to its steep topography, weak facies, and anthropogenic pressure. This study provides an integrated assessment of landslide susceptibility by combining multi-source geospatial data with statistical modelling approaches. A landslide inventory comprising 61 events was developed through the interpretation of high-resolution satellite imagery, virtual globe platforms, published scientific sources, and field validation supported by GPS measurements. Nine conditioning factors were considered, including slope gradient, slope aspect, lithology, elevation, rainfall, land use/land cover, and distances to rivers, roads, and faults. Landslide susceptibility was modelled using four bivariate statistical techniques: Frequency Ratio (FR), Information Value (IV), Weighted Factor (WF), and Weight of Evidence (WoE). The resulting susceptibility maps were classified into five categories ranging from very low to very high susceptibility. Model performance was evaluated using Receiver Operating Characteristic (ROC) analysis and the Area Under the Curve (AUC). All models exhibited strong predictive capability, with AUC values indicating well to excellent performance. The WoE model achieved the highest accuracy (AUC = 0.925), followed closely by the IV model (AUC = 0.905), confirming the effectiveness of probabilistic approaches in reproducing the spatial distribution of landslide occurrence. Spatial analysis consistently shows that high and very high susceptibility zones are mainly associated with steep slopes, weak marly and clay-rich lithologies, higher rainfall regimes, and proximity to drainage networks and road infrastructures. These convergent patterns across models emphasize the dominant role of both geomorphological controls and human-induced modifications in slope instability. The results highlight the added value of probabilistic modelling frameworks in capturing complex spatial relationships between conditioning factors and landslide occurrence. The proposed approach provides a reliable basis for susceptibility mapping and offers practical support for risk reduction strategies, land-use planning, and monitoring prioritization in Mediterranean environments.