Fractal-Based Evaluation of the Spatial Relationship Between Conditioning Factors and the Distribution of Landslides (a Case Study in Tinh Tuc, Cao Bang Province, Vietnam)

Abstract

This paper presents the assessment results of the spatial relationship between past landslides and four landslide factors in Tinh Tuc town, Cao Bang province, Vietnam, including distance to roads, distance to faults, distance to drainage, and distance to geological boundaries. The fractal dimension values were determined using fractal analyses on three sets of raw data, including the number of landslides, landslide density, and landslide relative density. Using a combination of the frequency ratio (FR) method and the fractal method, landslide susceptibility maps were produced to determine which raw data set more properly analyzes the role of factors in the development of the landslide process. Assessment results using the receiver operating characteristic (ROC) method indicate that the fractal analysis of the raw data sets successfully determines the spatial relationship between conditioning factors and landslides in the study area. In addition, relationship analyses using datasets based on landslide density provide more optimal outcomes. The authors also suggest integrating the fractal method with other quantitative assessment methods to improve the accuracy of landslide prediction in Vietnam and worldwide.