Landslide Susceptibility Mapping using Statistical Methods in Uatzau Catchment Area, Northwestern Ethiopia

Abstract Landslide susceptibility mapping is important to hazard management and to have planning development activities in the mountainous country like Ethiopia. In the present study, the landslide susceptibility mapping of the Uatzau basin is made using certainty factor, information value and logistic regression methods. Preparation of landslide inventory map from detailed fieldwork and Google Earth image interpretation was the first activity. Thus, 514 landslides were mapped and out of which 490 (70%) of landslides were randomly selected keeping their spatial distribution to build landslide susceptibility models while the remaining 155 (30%) of the landslides were used to models validation. It is clear that the effectiveness of the landslide susceptibility model using GIS and statistical methods is depending on the selection of the causative factors, which have a prevailing effect on landslide occurrence. In this study, six factors including lithology, land use/cover, distance to stream, slope gradient, slope aspect, and slope curvature were the landslide factors that were evaluated. After preparation of these factor maps, the effects of them on slope instability was determined by comparing with landslide inventory raster map using GIS. Finally, the landslide susceptibility model for the Uatzau area was developed and validated using the receiver operating characteristics curve (ROC). The results of ROC showed that for landslide susceptibility map using frequency ratio model (FRM) with an AUC value of 0.8883 has the highest prediction accuracy of 88.83%. The landslide susceptibility map, which is produced using Certainty factor and information value methods also showed that 87.03% and 84.83% of prediction accuracy respectively. Besides the prediction accuracy of the model, the success rate curve for all models was applied and the result showed that more than 80% accuracy (i.e, 80.83% for the information value model, 87.19% for the certainty factor model and 83.27% for frequency ratio model). The present research finds out that all methods/ models, which have employed in this study showed that reasonably very good accuracy in predicting landslide susceptibility of the Uatzau area. Therefore, these landslide susceptibility maps can be used for regional land use planning and landslide hazard mitigation purposes.