Advancing Landslide Susceptibility Modeling with Explainable AI

Researchers at the University of Technology Sydney have made new discoveries in the field of information science, shedding light on the effectiveness of the SHapley Additive exPlanations (SHAP) approach in enhancing the interpretability of landslide susceptibility models. The study focuses on the landslide-prone region of Bhutan and compares the performance of two approaches, one incorporating geoenvironmental factors and the other integrating both geoenvironmental factors and a physically based model. The findings provide valuable insights for stakeholders and decision-makers involved in land use planning and disaster preparedness.

Key Takeaways:

• The SHAP approach was found to be effective in enhancing the interpretability of landslide susceptibility models, particularly in the landslide-prone region of Bhutan.
• The study compared the performance of two approaches: one incorporating geoenvironmental factors and the other integrating both geoenvironmental factors and a physically based model.
• The Random Forest (RF) algorithm was used to develop and compare these landslide susceptibility models.
• Various evaluation metrics, including overall accuracy and precision-recall, were employed to assess the predictive capabilities of each model.
• The findings revealed the strengths and limitations of both models, providing valuable insights for stakeholders and decision-makers.
• The study highlights the role of SHAP and its interaction with geoenvironmental and physically based factors in landslide susceptibility modeling.
• The research aims to advance landslide susceptibility modeling by contributing to more effective risk mitigation strategies.

Statistics:

• 81-85: The page numbers of the journal article "SHapley Additive exPlanations (SHAP) for Landslide Susceptibility Models: Shedding Light on Explainable AI" in the ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences.
• 2025: The year in which the research was conducted.
• X-G-2025: A reference code used in the journal article.
• BHUTAN: A country in the region where the landslide-prone area studied is located.
• 2007: The zip code of the University of Technology Sydney, Faculty of Engineering and IT.

Sources:

• VerticalNews
• Ecology, Environment & Conservation - August 1, 2025
• ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
• SHapley Additive exPlanations (SHAP) for Landslide Susceptibility Models: Shedding Light on Explainable AI