A new model of volcanic surge for New Zealand

Summary

The propagation of particle laden volcanic currents such as pyroclastic surges and base surges can cause extensive damage to exposed populations and structures. Such currents occur frequently in the volcanic record and in New Zealand, but are relatively small and thus generally unpreserved in the geologic record. Being an extreme proximal hazard, the surge extent is crucial to estimate in order to determine exclusion/risk zones. While models exist to estimate the extent of surges, they are generally based on either very simple approaches that neglect important physical processes with limited dimensionality, or are highly complex and require extensive computational resources and estimation of uncertain parameters.

In this project, we propose a new method to modelling surge propagation that can be adapted and extended to more complex physics than simple (e.g. energy line) methods to provide first order approximations to surge extent, but without including computationally expensive physical details. This is achieved through use of a level set approach, where propagation of the surge is calculated from speed functions at the surge front. This method was demonstrated on three case studies: the 2012 Te Maari surges, the 2016 Whakaari/White Island surges and the Maungataketake surges in the Auckland volcanic field. First, an energy line method was reformulated using Bernoulli’s theorem and used to identify areas of critical need and limitations to focus improvements. For the Te Maari case, different coefficients of α for the Energy Grade Line gave different fits for the surge extend perpendicular to and aligned with the estimated directions of the blast. The constraint of large scale topography in directing energy was seen to be important for the Whakaari case, where the runout was approximated better using an ‘effective radius’ approach, while the Maungataketake surge model did not show a good fit to the estimated extents, highlighting potential limitations in the energy line method. To improve upon this model, we took a potential flow approach to account for the effect of terrain relief acting as an energy sink for the surge. In case studies with low relief, the potential flow model had little effect. However, the high relief in the Te Maari case had the effect of limiting the extent of the surge to approximate the extents perpendicular to the blast, but with an underestimation in the direction of the blast. This demonstration shows the applicability of the new modelling approach, while also supporting implementation through model architecture that integrates geospatial libraries.