Investigating the Impact of Geomorphic Characteristics on Landslide Patterns Using the Random Forest Algorithm (Case Study: Shahid Abbaspour Dam Watershed)

Moavi, Mahshid; Entezari, Mojgan; Elmizadeh, Heeva

doi:10.22111/jneh.2025.51425.2107

Investigating the Impact of Geomorphic Characteristics on Landslide Patterns Using the Random Forest Algorithm (Case Study: Shahid Abbaspour Dam Watershed)

Articles in Press

Document Type : Research Article

Authors

¹ PhD Student, Department of Physical Geography, Faculty of Geographical Sciences and Planning, University of Isfahan, Iran.

² Professor, Department of Physical Geography, Faculty of Geographical Sciences and Planning, University of Isfahan, Iran.

³ Associate Professor, Department of Marine Geology, Faculty of Marine Natural Resources, Khorramshahr University of Marine Sciences and Technology, Iran.

10.22111/jneh.2025.51425.2107

Abstract

Landslides, recognized as complex natural hazards, pose significant threats to infrastructure, human communities, and fragile ecosystems. This study aims to investigate the influence of geomorphic characteristics on landslide occurrence patterns in the Shahid Abbaspour Dam watershed (Deh-Sheikh) using the advanced machine learning Random Forest algorithm. The dataset comprised 15 key layers influencing landslides, sourced from various datasets including 1:50,000 topographic maps, 1:100,000 geological maps, Landsat satellite imagery, and field observations. Initial data processing was conducted in ArcGIS, SAGA-GIS, and ENVI software environments, followed by modeling using machine learning packages in RStudio. The results indicate that geological formations (23.7%), slope (19.5%), and distance from rivers (15.2%) are the primary factors controlling landslide patterns in the region. The model output reveals that approximately 30.4% of the watershed area (75.37 km²) falls within the high-hazard class, predominantly concentrated in the southern and southeastern sectors. Model performance evaluation using statistical metrics demonstrated the algorithm’s robust performance, with an overall accuracy of 0.986, a Kappa coefficient of 0.972, and a root mean square error (RMSE) of 0.0101. This study represents the first comprehensive landslide hazard assessment in this watershed using machine learning techniques, providing a scientifically rigorous framework for risk management, environmental planning, and mitigation of landslide impacts in similar mountainous regions.

Keywords

Main Subjects

Environmental Hazards

References

References [in Persian]

Moharami, M., & Jelokhani Niaraki, M. R. (2023). Mapping landslide-prone areas using statistical and machine learning models: A case study of Austria. Journal of Mapping Sciences and Techniques, 13(2), 79–94. [In Persian]

Mumipour, M., & Moavi, M. (2022). Analysis of tectonic and erosion conditions in the Shahid Abbaspour Dam watershed using geomorphometric techniques. Journal of Geography and Environmental Hazards, 11(1), 1–16. https://doi.org/10.22067/geoeh.2021.70140.1053. [In Persian]

References [in English]

Achour, Y., & Pourghasemi, H. R. (2020). How do machine learning techniques help in increasing the accuracy of landslide susceptibility maps? Geoscience Frontiers, 11(3), 871-883. https://doi.org/10.1016/j.gsf.2019.10.001

Agboola, G., Hashemi Beni, L., Elbayoumi, T., & Thompson, G. (2024). Optimizing landslide susceptibility mapping using machine learning and geospatial techniques. Ecological Informatics, 81, 102583. https://doi.org/10.1016/j.ecoinf.2024.102583

Li, M., Wang, H., Chen, J., & Zheng, K. (2024). Assessing landslide susceptibility based on the random forest model and multi-source heterogeneous data. Ecological Indicators, 158, 111600. https://doi.org/10.1016/j.ecolind.2024.111600

Mondini, A. C., Mondini, F., Guzzetti, F., & Melillo, M. (2023). Deep learning forecast of rainfall-induced shallow landslides. Nature Communications, 14(1), 2446. https://doi.org/10.1038/s41467-023-38135-0

Rodrigues Neto, J. M., & Bhandary, N. P. (2024). Landslide susceptibility assessment by machine learning and frequency ratio methods using XRAIN radar-acquired rainfall data. Geosciences, 14(6), 171. https://doi.org/10.3390/geosciences14060171.