Physical and numerical modeling of rainfall triggered shallow landslides in central highlands, Ethiopia

• Published in: Bulletin of engineering geology and the environment 2023-07, Vol.82 (7), Article 239
• Main Authors: Gidday, Biruk Gissila, Ayothiraman, R., Ramaiah, B. Janaki, Ramana, G. V.
• Format: Article
• Language: English
• Subjects: Earth and Environmental Science; Earth Sciences; Foundations; Geoecology/Natural Processes; Geoengineering; Geotechnical Engineering & Applied Earth Sciences; Hydraulics; Nature Conservation; Original Paper
• ISSN: 1435-9529; 1435-9537
• DOI: 10.1007/s10064-023-03235-y

Rainfall-triggered landslides are the most common natural hazard with serious consequences worldwide, including in Ethiopia. In Ethiopia, GIS-based landslide assessment has been the most common approach, which ignores the underlying effects of controlling factors, such as rainfall characteristics, as well as geotechnical properties, yielding subjective and debatable results. The current study aims to capture the failure mechanism and establish rainfall thresholds through field investigations, laboratory experiments, and physical flume experiments coupled with numerical modeling. Filed hydrogeological parameters (rainfall, groundwater), DEM, and geotechnical parameters were used in numerical models such as HYDRUS and TRIGRS. Saturated and unsaturated hydraulic properties, such as the hydraulic conductivity function and the soil–water characteristic curve (SWCC), as well as hydrogeological factors, were used to establish empirical correlations. Based on multiple rainfall combinations, slopes, and TRIGRS-generated factor of safety data, a rainfall intensity of approximately 93.9 mm/h was identified as the rainfall threshold beyond which landslides are likely to be triggered. TRIGRS outputs were reprocessed using ArcGIS tools to develop a landslide hazard zonation (LHZ) map. The TRIGRS model was validated using receiver operating characteristic (ROC) curve analysis, and the model has a high predictive competence, with an average sensitivity of 83% and specificity of 76%. It can be deduced that the intensity duration (ID) curve, threshold value, and empirical correlations can all be used to develop early warning systems for the study region and neighboring regions with similar catchment characteristics. More integrated research, however, is required to accurately predict the hazards associated with landslides.