Document Type : Research Paper

Authors

• Shadi Jalilian ¹
• Shaban Shataee Jouibary ¹
• Mohammad Hadi Moayeri ²
• Amir Saddodin ³

¹ Gorgan University of Agricultural Sciences and Natural Resources

² Department of Forestry, Gorgan University of Agricultural Sciences and Natural Resources

³ Department of Watershed Management, Gorgan University of Agricultural Sciences and Natural Resources

Abstract

Abstract
Introduction:
Landslides, as one of the most destructive natural phenomena, annually cause significant human casualties and financial damage. This study aimed to evaluate the performance of two machine learning models, Random Forest (RF) and Support Vector Machine (SVM), as well as two deep learning models, Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN), in modeling landslide risk in a part of Golestan province.
Methodology:
For modeling, layers of 10 influential environmental and human factors were identified and prepared, including elevation, slope, slope direction, Topographic Wetness Index (TWI), Normalized Difference Vegetation Index (NDVI), land use, distance from roads, distance from rivers, distance from faults, and precipitation. From 494 landslide and non-landslide points, 70% were used for the training phase and 30% for the validation phase. To further the research objectives and prepare the information layers, ArcGIS software and the Python programming language were utilized to implement machine learning (ML) and deep learning (DL) algorithms.
Results:
The results demonstrated that the deep learning algorithm CNN, with an AUC score of 0.910, an overall accuracy of 87.72%, and a Kappa coefficient of 0.899 for high-risk classes (high and very high risk), was identified as the most efficient model. Variable importance analysis using the superior model (CNN) revealed that the factors Distance to Fault and Distance to River were respectively the most significant contributors to landslide occurrence in the study area.
Conclusion:
The results of this research can be highly effective in identifying high-risk areas and determining the factors influencing the occurrence of landslides in this area, thereby aiding in reducing potential damages and threats associated with landslides, as well as in implementing effective management strategies.

Keywords

• Landslide susceptibility prediction
• Hyrcanian forests
• Deep learning
• Convolutional Neural Network (CNN)
• Spatial modeling
• Receiver Operating Characteristic (ROC) curve