Temporal landslide susceptibility modeling using machine learning: validation against multi-year inventories in the NW Himalayas

The region of Muzaffarabad, located in the tectonically active foothills of the Himalayas, is highly vulnerable to landslides due to frequent slope failures exacerbated by both natural and anthropogenic factors. This study assesses landslide susceptibility and hazard in Muzaffarabad, Azad Jammu and Kashmir, Pakistan, using two machine learning (ML) approaches, specifically Logistic Regression (LR) and Random Forest (RF). A total of thirteen landslide conditioning factors, including topographic, geological, environmental and anthropogenic attributes, were analyzed using models. Landslide susceptibility maps were developed and validated with landslide datasets from 2005, 2007, and 2012, showcasing the models’ predictive performance through accuracy metrics such as Area Under Curve (AUC), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), F1-score and Kappa. The study found that RF models consistently outperformed LR models in predictive accuracy, with AUC values of 99% for the 2007 dataset and 96.3% for the 2012 dataset. Temporal validation of these models against subsequent-year landslide inventories demonstrated that RF retained stronger predictive stability over time than LR, although the use of static conditioning layers limits the interpretation of these temporal changes as direct evidence of landscape dynamics. The results underscore the potential of RF-based models for operational landslide hazard assessments, providing critical insights for hazard mitigation and urban planning in tectonically active regions. The study’s findings contribute to the development of more effective risk management strategies, particularly in the face of ongoing seismic activity and climate change, enhancing regional resilience to landslide disasters.

Authors

Publication Details

Journal
Journal of Mountain Science
Published
2026-11-19
DOI
https://doi.org/10.1007/s11629-025-0058-6
Primary Topic
Landslides and related hazards
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Temporal landslide susceptibility modeling using machine learning: validation against multi-year inventories in the NW Himalayas

Łukasz Pawlik, Saad Wani, Muhammad Basharat, Malik Talha Riaz et al.
Journal of Mountain Science
Landslides and related hazards
article

Temporal landslide susceptibility modeling using machine learning: validation against multi-year inventories in the NW Himalayas

Łukasz Pawlik, Saad Wani, Muhammad Basharat, Malik Talha Riaz, Muhammad Tayyib Riaz
article en

Abstract

The region of Muzaffarabad, located in the tectonically active foothills of the Himalayas, is highly vulnerable to landslides due to frequent slope failures exacerbated by both natural and anthropogenic factors. This study assesses landslide susceptibility and hazard in Muzaffarabad, Azad Jammu and Kashmir, Pakistan, using two machine learning (ML) approaches, specifically Logistic Regression (LR) and Random Forest (RF). A total of thirteen landslide conditioning factors, including topographic, geological, environmental and anthropogenic attributes, were analyzed using models. Landslide susceptibility maps were developed and validated with landslide datasets from 2005, 2007, and 2012, showcasing the models’ predictive performance through accuracy metrics such as Area Under Curve (AUC), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), F1-score and Kappa. The study found that RF models consistently outperformed LR models in predictive accuracy, with AUC values of 99% for the 2007 dataset and 96.3% for the 2012 dataset. Temporal validation of these models against subsequent-year landslide inventories demonstrated that RF retained stronger predictive stability over time than LR, although the use of static conditioning layers limits the interpretation of these temporal changes as direct evidence of landscape dynamics. The results underscore the potential of RF-based models for operational landslide hazard assessments, providing critical insights for hazard mitigation and urban planning in tectonically active regions. The study’s findings contribute to the development of more effective risk management strategies, particularly in the face of ongoing seismic activity and climate change, enhancing regional resilience to landslide disasters.

Journal of Mountain Science
Landslides and related hazards
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon