Landslide susceptibility mapping by fuzzy gamma operator and GIS, a case study of a section of the national road n°11 linking Mateur to Béja (Nortshern Tunisia)

Abstract

The purpose of this paper is to map the landslide susceptibility using the fuzzy gamma operator and GIS for the section of the national road n°11 linking Mateur to Beja. To minimize the subjectivity of the fuzzy logic approach, the frequency ratio was used to calculate the fuzzy membership. To locate the 147 landslides, Google Earth extracts analysis, and field survey has been done. In this cartography, nine factors controlling landslides were used: slope, aspect, curvature plane, curvature profile, distance to faults, distance to rivers, land use, rainfall, and lithology. Once the fuzziness is calculated in each factor using the frequency ratio (FR), the fuzzy gamma operator can be applied to these nine factors to produce the landslide susceptibility map. Finally, the choice of the nearest susceptibility map of reality is made by applying the ROC curve. By the analysis of the area under curve (AUC), it is seen that the prediction accuracy of model proves that the gamma value of 0.90 yielded the better prediction of landslide susceptibility map (0.89). Finally, our area of interest was classified based on the Jenks natural breaks classification method into five classes, such as non-susceptible zone, low susceptible zone, moderate susceptible zone, high susceptible zone, and very high susceptible zone. Landslide susceptibility mapping based on the fuzzy gamma operator presents a very acceptable result with a reliability of 89%.