Tsunami caused by submarine mass failure - case study of Heard Island and McDonald Island, Australia

Worldwide, volcano sourced tsunami account for around 6% of tsunami events. Tsunami generated by volcanoes can be a far-field source which is often overlooked in hazard assessments. There are several mechanisms for volcanic-tsunami events; underwater explosion, blasts, pyroclastic flows, caldera collapse, subaerial failure and submarine failure. A key component in including these events in hazard assessments is being able to identify and characterize the potential source volcanoes. Analysis of remote sensing data can be used to determine topography, collapse scars, thermal anomalies, past lava flows and slope profiles. We have developed a framework using the remotely sensed data combined with volcano type and water depth to determine the possible tsunami generating mechanism. Australia is often seen as a country which does not have high exposure to tsunami or volcanic hazards. The framework described above was applied to Heard Island and McDonald Island (HIMI), an Australia external territory approximately 4000 km south west of Perth, Australia, in the Kerguelen Plateau. There are multiple volcanic edifices in this area, including those with evidence of previous collapse events suggesting that the occurrence of a submarine mass failure is possible and could potentially generate a tsunami. Remote sensing analysis and data from past events in other locations was used to determine the volume that could potentially fail. Two scenarios were then determined for modelling; the first being a submarine mass failure on McDonald Island with a volume of 40km3 (analogous to a past event near Reunion Island), with the second being a submarine mass failure on the Northern Kerguelen seamount with a volume of 230 km3 (analogous to a past event on the Hawaiian Ridge). The initial mass movement that generates the tsunami was modelled as a slide using the predictive submarine slide equations of Grilli and Watts (2005) and Watts et al. (2005). The deep water propagation of the tsunami has been modelled using URSGA which solves the linear shallow water wave equation; with inundation modelled using ANUGA which solves the non-linear shallow water wave equations. Preliminary results show that it is possible for both of these scenarios to trigger a tsunami that can be detected along the south west Australian coast and in Antarctica. The modelling predicts that the first wave of the tsunami wave train takes approximately six hours to reach near shore Perth. This poses a potential tsunami hazard to Australia, but more work needs to be undertaken to understand the mechanism and likelihood of such events. Presented at the 2017 International Tsunami Symposium