Geospatial Based Landslide Sensitivity Mapping in Addis Ababa, Ethiopia

Background Numerous triggering variables tend to determine the frequency and intensity of landslides, and human activity frequently makes them worse. Disaster risks stemming from landform features and meteorological factors are frequently reported in the central highlands of Ethiopia. It is where the research region is located, which have a higher possibility of experiencing landslide hazards. Topographic undulation, high rates of urbanization, and the ensuing boom in building activity are characteristics of the current research region. As a result, landslide hazard susceptibility zonation is extremely important for the research area. The objective of the present study is, thus, to delineate the landslide susceptibility map (LSM) using Analytical Hierarchical Process (AHP) model in the study area. Methods The landslide inventory mapping that was conducted through field observations and Google Earth image interpretation identified about 250 landslide locations, and was randomly classified into training datasets (70%) and validation datasets (30%). Major parameters such rainfall intensity, topography, geology, soil type, land use and land cover, distance from the Earthquake Center, stream and road network data, and more were discovered as causative factors for landslide susceptibility in this study. The relative importance of each landslide causative factors was determined using the AHP Model in ArcGIS10x. Results The AHP results indicated that slope steepness, Lineament and/or fault density, rainfall and lithology were found to be the dominant triggering factors of landslides contributing about 83.1% of the area of the study region. Subsequently, a landslide susceptibility index (LSSI) was calculated based on the relative influence of these causative factors in an overall landslide susceptibility analysis. The land slide sensitivity map (LSM) of the study area indicated that around 65% (26885 hectares) of the study area was designated as Very low and low landslide susceptibility category. The remaining 34% (13831 ha), which are often located in regions with high slope gradients, poor geological structure, and high drainage density and rainfall are classed as medium and high landslide vulnerability. The performance of the LSM produced by AHP model, in the present study, was evaluated using Receiver Operating Characteristics (ROC) and area under the curve (AUC) method. The validated AHP model result revealed that the AUC success rate curves was 0.807, which demonstrated a very good model predicting performance. Conclusion Therefore, landslide susceptibility mapping using AHP model and geospatial approaches can deliver very good prediction performance and can be helpful for urban land use planners and other sectorial offices for proper land use planning.