Physically-based catchment-scale prediction of slope failure volume and geometry

The application of physically-based approaches for slope failure analysis at a catchment scale remains a difficult challenge, and several new models have been proposed in recent years. The assumptions of these models vary significantly. Tools such as random ellipsoid sampling provide detailed assessment of failure probability but due to numerical cost can not be applied beyond site-investigations. In this research, an iterative method for finding progressive slope failure surfaces is developed. Additionally, this method includes a description of lateral forces that occur due to weight of the fractured failure material. This development extends a similar approach that was developed as part of the OpenLISEM multi-hazard modelling tool. Study cases are presented for a set of hypothetical slopes and the storm-induced disaster that occurred in Messina (Italy), on the first of October 2009. Here, a large number of shallow slope failures transformed into debris flows. The model outcomes are compared with the outcomes of other free and open-source methods that are currently available within the scientific community (Infinite Slope, random ellipsoid sampling (r.slope.stability) and random spheroid sampling (Scoops3D)). Finally, finite element modelling is performed for the hypothetical slopes using ADONIS). Analysis of predicted failures show that the presented method is able to better predict the occurrence of smaller failures but ellipsoid sampling methods provide a more robust option for larger slope failures. Predicted failure surface patterns for the Messina 2009 event show complex variation between rotational and translational geometry depending on the topography. The ability to provide catchment-scale deterministic failure volume and geometry can potentially be used to quantify predicted source volumes during future disasters.

Automated summary

The research introduces an iterative method for finding progressive slope failure surfaces while considering lateral forces due to weight of the fractured failure material. The method performed well in predicting smaller failures, but ellipsoid sampling methods proved to be more robust for larger slope failures. Comparison with other free and open-source methods showed complex variation in failure surface patterns for the 2009 Messina event, highlighting the potential of the method to provide catchment-scale deterministic failure volume and geometry for future disasters.