Developing a Regression Tree for Mapping Landslide Susceptibility in a GIS Environment

Landslides are natural disasters that can cause extensive damage to both property and life every year, and therefore, generating landslide susceptibility map is essential for planning future developmental activities. The objective of this paper is to analyze the relationship between landslide locations and landslide related factors by applying regression tree algorithm in a GIS environment. In this research Classification and Regression Trees (CART) is used for generating landslide susceptibility map. Eight causal factors in landslide occurrence are chosen, namely; slope, aspect, curvature, land cover, lithology, distance to river, distance to faults and distance to roads. Data of 104 sample points of observed landslides in Mazandaran Province, Iran, are applied as training data and the rest are used as test data for pruning regression tree. The results indicated that the landslide susceptibility map has the most intensity with two factor maps; lithology and slope.