Document Type : Research Paper

Authors

• Malihe Mohamadnia ¹
• Abolghasem Amirahmadi ²
• Liela Goli Mokhtari ³

¹ Hakim Sabzevari university

² departement geography and enviroment scinceT university hakim, geomorphology ,sabzevar

³ Hakim Sabzevari University

Abstract

The purpose of this research is to spatially model landslide susceptibility using machine learning methods such as random forest, support vector machine and enhanced regression tree in Razavi Khorasan province. At first, the distribution map of landslides in the region was prepared by field visits and using the country's landslides database. In the next step, 70% of the identified landslides were used for the modeling process and 30% for the evaluation of the said models. The information layers of height, slope, direction of slope, distance from waterway, density of waterway, distance from road, distance from fault, land use, vegetation index, surface curvature, profile curvature, precipitation and selected geology and its maps in ArcGIS environment it was prepared. The results of prioritizing the effective factors using the random forest model showed that the factors of precipitation and altitude had the greatest impact on the occurrence of landslides in the study area. Also, the results of the evaluation of machine learning models using the relative performance detection curve (ROC) showed that the map prepared by the random forest method had the highest accuracy in preparing the landslide potential map in the studied area and based on this, more than 25% The area is in the high and very high risk class.

Keywords

• range of motion
• Risks
• GIS
• Razavi Khorasan province