Implications of landslide runout modeling for vulnerability assessment

Abstract

The vulnerability to landslides depends on both the susceptibility of the exposed elements and the intensity of the landslide, which is commonly characterized by its motion mechanism. This study proposes a quantitative evaluation framework to assess the implications of using different models for predicting the landslide runout distance (LRD) on vulnerability, estimated through two distinct vulnerability functions. The analysis focuses on a debris flow that impacted a major highway in the Colombian Andes. The event, with a triggered volume of 340 m³, a runout of 84 m, and a vertical drop of 42 m, serves as a benchmark for evaluating model performance. The findings provide insights into the influence of material type, flow regime, and model uncertainty on LRD and vulnerability estimates. Empirical methods enabled rapid assessments but exhibited high variability (LRD errors up to 120 %). Analytical models, particularly the sliding block model, offered a balance between simplicity and physical realism, overestimating LRD by 14 % without calibration while also providing velocity estimates. Multidimensional (2D/3D) models, though resource-intensive, best reproduced the observed behavior; the 3D model closely matched the measured runout when calibrated with high-friction parameters and GIS-derived inputs. A benchmarking analysis using the Analytic Hierarchy Process (AHP) identified the sliding block model as the most effective overall, combining accuracy, functionality, and usability. These results highlight that model selection should align with the intended application: empirical models for rapid screening, analytical models for design purposes, and multidimensional models for detailed vulnerability assessments.