Dynamic analysis of long run-out rock avalanches: A view from the Vaigat Strait, West Greenland

Abstract

Risk assessments of the threat posed by rock avalanches conventionally rely upon numerical modelling of potential run-out. Such models are contingent upon a thorough understanding of the flow dynamics inferred from deposits left by previous events. Few records exist of multiple rock avalanches with boundary conditions sufficiently consistent to develop a set of more generalised rules for behaviour. This thesis uses a numerical modelling approach to investigate the emplacement dynamics of 20 adjacent events in Vaigat, West Greenland, which are sourced from a stretch of coastal mountains of relatively uniform geology and structure. Rheological calibration of the numerical flow code VolcFlow was performed using a well constrained event at Paatuut (AD 2000). The best-fit simulation assumes a constant retarding stress with a collisional stress coefficient and simulates run-out to within ±0.3% of that observed. Despite being widely used to simulate rock avalanche propagation, other models, that assume either a Coulomb frictional or a Voellmy rheology, failed to reproduce the observed event characteristics and deposit distribution at Paatuut. This calibration was then applied to 19 other events, simulating rock avalanche motion across 3D terrain of varying levels of complexity. The findings illustrate the utility and sensitivity of modelling a single rock avalanche satisfactorily as a function of rheology, alongside the validity of applying the same parameters elsewhere, even within similar boundary conditions. VolcFlow can plausibly account for the observed morphology of a series of deposits emplaced by events of different types, although its performance is sensitive to a range of topographic and geometric factors. These exercises show encouraging results in the model’s ability to simulate a series of events using a single set of parameters obtained by back analysis of the Paatuut event alone, suggesting that first-order run-out prediction is possible.