Abstract:
Numerous landslides affect practically a large part of the natural slopes of Souk Ahras region each year, especially during or after heavy rain. The instabilities markers are currently outlined on the slopes in the form of numerous tearing scars, and the bead at the feet of the affected slopes. For the purpose of landslides susceptibility mapping and characterization in Souk Ahras province, the present work focuses on the development of a landslides inventory map for the purpose of compiling a census of landslides. the spatial distribution of these events, making it possible to evaluate the sensitivity of the territory to this phenomenon, and to identify the conditions which favored the appearance of the latter by, i) the bibliographical data exploitation, the systematic interpretation of the satellites images proved by field visits, ii) the geological presentation of slopes affected by landslides and the analysis of the spatial distribution of these events on the one hand, and our knowledge of the geotechnical behavior of geological facies slipping on the other hand ; iii) the application of mineralogical and geotechnical analyzes for some unstable samples, in order to understand there mechanical behavior. The distribution of landslides in the study area is largely governed by a combination of several conditions related to the geological, geomorphological and hydrological as well as human activities. As a result, a spatial database of ten landslides-related factors were identified and used to assess landslides in Souk Ahras region. It aims to provide a scientific assessment of areas susceptible to this phenomenon, basing on the results of the three statistical models : Logistic Regression (RL), Frequency Ratio (Fr) and Weighted of Evidence (WoE). Three landslide susceptibility maps were produced using these statistical models. Each susceptibility map subdivides the study area into five classes of susceptibility to landslides: very low, low, moderate, high and very high. These susceptibility maps were compared and verified with the inventory data. The results of the landslides susceptibility models validation confirm that the statistical methods (LR, Fr and WoE) are all reliable for the mapping of landslide susceptibility, while the first model has a prediction accuracy significantly higher than 90.91 % compared to the other two techniques, as shown in our study. Its performance is the best among the three statistical methods. The results of this research work can serve as a basis of rational decisions for development planning in Souk Ahras region.
