Engineering Geological Factors Affecting Loess Caves Susceptibility Mapping in Gorgan plain, Golestan province, northeastern Iran

Abstract

Loess cave is one of the geological hazards in the Loess Plateau of northeast Iran, in the Gorgan plain. In total, 697 caves were mapped, and an inventory map was prepared. Due to the importance of geo-environmental and soil properties in the development of loess caves, six independent variables, including joint density, land use, soil type, collapse sensitivity, self-weight collapse coefficient, and surface water distance, were designed for modeling the caves susceptibility in the GIS software. Loess cave susceptibility maps were constructed using Density Area (DA), Information Value (IF), and Frequency Ratio (FR) methods in a GIS environment. Loess cave susceptibility map obtained from three models was compared using Relative Operating Characteristic (ROC) and Area Under Curve (AUC). The area under the curve analysis was used to evaluate model compatibility and predictability. The equal interval classification method classified the study area into five ordinal categories, i.e., very low, low, moderate, high, and very high. AUC of the IF model was 0.727. The values of AUC for DR and the FR models were similar and equal to 0.725. Therefore, the loess cave susceptibility maps of Golestan Province is valuable for decision-maker in loess cave-prone area.