From Earthquakes to Inundation: Using Physics-Based Synthetic Earthquake Catalogues to Assess Tsunami Hazard in New Zealand
Tsunamis are an infrequent natural hazard that have the potential to cause extreme damage to infrastructure, as well as significant loss of life. Yet, the rarity of tsunamis in instrumental records makes studying these large scale natural hazards difficult. Therefore, developing new methods to investigate these hazards is important to not only explore what the next event may look like, but also to help mitigate the risks that they pose. This is particularly important for countries like Aotearoa/New Zealand, which have large sections of coastline and are close to subduction zones that have not generated large magnitude earthquake in historical/modern records but are known to have generated them from paleoearthquake evidence.
Physics-based synthetic earthquake catalogues, generated by multi-cycle earthquake simulators, provide a new and alternative pathway for analysing tsunami hazard. The Rate and State Earthquake Simulator (RSQSim) was used to generate a synthetic earthquake catalogue and 2,585 earthquakes, that had a magnitude >Mw 7.0, had a tsunami potential energy greater than 1 × 10^11J and nucleated over a 30,000 year time period were identified. To-the-coast tsunami simulations were run and using these simulations, we constructed a Probabilistic Tsunami Hazard Assessment (PTHA) for all of New Zealand’s coastline. Modelled slip of the Hikurangi and Tonga-Kermadec subduction thrusts generated maximum tsunami heights at-the-coast of up to 28 m and these earthquakes pose the greatest tsunami hazard along Aotearoa/New Zealand’s coastline. The results provided a successful “proof of concept” that physics-based synthetic earthquake catalogues can be used to undertake PTHAs, and the results provide a platform upon which next generation probabilistic tsunami inundation models and multi-hazard risk models can be constructed.
For many subduction zones, uncertainty surrounds the locking distribution, the pattern of high slip-deficit patches, along the subduction interface. Whether the trench is locked or creeping and if these locking distributions are invariant structures through time are additional sources of uncertainty. The extent to which differences in the locking distribution affect the modelled tsunami hazard currently remains unknown. Using three alternative catalogues, generated for the Hikurangi Subduction Margin and the Tonga-Kermadec Subduction Zone, with alternative representations of the locking distribution, to-the-coast tsunami simulations were run and PTHAs were constructed. These assessments showed that spatial variability in the degree of locking, along both the strike and dip of the subduction interface, and locking to the trench, directly impact the tsunami hazard. Our results show that careful consideration of frictional heterogeneities in physical models are necessary before using them for PTHAs. It also shows that by analysing multiple physical models of subduction zones, uncertainty in hazard assessments caused by the unresolved interface properties can also begin to be quantified.
From the physics-based synthetic earthquake catalogues previously analysed, we selected the 100 earthquakes from each catalogue that generated the largest wave height at-the-coast around Te Matau-a-Maui/Hawke’s Bay, and ran tsunami inundation simulations. Using these simulations, five Probabilistic Tsunami Inundation Hazard Assessments (PTIHAs) were successfully constructed to investigate, 1) the inundation around Te Matau-a-Maui/Hawke’s Bay, and 2) how these catalogues could be used to quantify uncertainty surrounding tsunami inundation. The results showed that the PTIHAs are extremely sensitive to changes in the underlying earthquake catalogue. Moreover, the pattern of co-seismic deformation, the amount of offshore uplift and onshore subsidence, primarily controls the tsunami inundation. From this analysis, we determined that inter-catalogue variability in the tsunami inundation is larger than the intra-catalogue variability, highlighting that if physics-based earthquake catalogues were to be used to run PTIHAs, multiple catalogues would be required in order to obtain a comprehensive hazard assessment.
To complement the analysis in Te Matau-a-Maui/Hawke’s Bay, we selected two catalogues and ran tsunami inundation simulations for Te Whanganui-a-Tara/Wellington, both accounting for, and not accounting for, co-seismic deformation in the Digital Elevation Models. These results showed that the location of the pivot line, as well as how far south the earthquake rupture extends along the Hikurangi Subduction Margin, both control the severity of the inundation the occurs across Te Whanganui-aTara/Wellington. These models allow us to investigate the uncertainty surrounding tsunami inundation and what the next large magnitude subduction earthquake and tsunami may look like. This is key as there is limited paleoearthquake/paleotsunami evidence available in Aotearoa/New Zealand. Understanding the pattern of deformation though time, and if this changes between earthquakes, allows for more comprehensive seismic and tsunami hazard assessments, both in New Zealand and globally.