MLP learning-based landslide susceptibility assessment for Kurdistan province, Iran

Abstract

Landslide susceptibility analysis is considered the spatiotemporal pattern of the prone area for landsides occurrences which is mostly used in urban planning, hazard management, and sustainable developments. The presented study tried to analyze the landslide susceptibility condition for Kurdistan province in Iran by using geographic information system (GIS) and multilayer perceptron (MLP). The assessment parameters are categorized in geomorphological, geological, and human-work classes concluded slope aspect, slope angle, elevation, hydraulic condition, distances from faults, weathering, distances from rivers, distances from roads and cities. According to the results of the susceptibility assessment for the studied area, it has appeared the MLP-based model is provided the susceptibility modeling with 85.25% accuracy and 83.79% precision. Also, the middle part and some of the west part of the studied area are located in the high-risk area for landslides failures.