Hazard management of rock slopes is utilized to ensure safety of people and infrastructure that could be affected by slope deformation events, such as rockfall. 3D modelling and point clouds are effective tools for analyzing rockfall hazards. Pre-failure rockfall events represent a challenge to fully 3D model given the incomplete geometric information caused by lack of block exposure. This incomplete information can lead to inaccurate predictions of block volumes. The motivation of this work is two-fold: to develop better methods of evaluating in-situ rock block volumes, and to inform mitigation designs regarding likely block orientation for rockfall events. In this paper, we develop a statistical shape modelling process capable of estimating the volume and orientation of in-situ blocks using Guassian Process Morphable Modelling.
Characteristic block shapes for a slope have been identified from samples collected in a catchment ditch and modelled to provide input data into Statismo (Lüthi el al, 2012), a framework for PCA-based statistical models. After enforced extrapolation along the known characteristic dimension of collected samples, the generated statistical models produced volume predictions ranging from 89-99% of a partitioned rock block.
Slope design and hazard management are important for infrastructure design and operation. Increasing demand for transportation surmounting natural obstacles has led to an increasing trend in the number and variety of slopes that may require hazard management and mitigation in the future. Improving current technologies regarding cost, efficiency, accuracy, and utility is an important objective for cost effective and safe infrastructure development and maintenance.
This paper proposes a method to expand the utility of 3D modelling as a volume and block orientation estimation tool for rockfall hazard characterization and monitoring. Previous research has developed the capability to model rockfall events considering the shape characteristics of rocks, and the impact on the coefficients of restitution (Glover 2015). Developing methods of predicting the shape and orientation of rock blocks prior to failure allows for the potential prediction of event paths, which in turn can be used to inform mitigation designs.