Ground motion simulations for Dunedin and Mosgiel, Otago, New Zealand
Abstract
We develop large scenario earthquakes on active faults in the vicinity of Dunedin and use them to develop ground motion simulations for a site in Dunedin (St Kilda – St Clair area, referred to as “St Beach”) and Mosgiel (centre of Mosgiel, referred to as “Taieri Basin”). The scenarios are developed to represent large Akatore Fault (within 15 km of Dunedin and Mosgiel) and Hyde Fault (within 40-50 km) earthquakes. The simulations utilise the Southern California Earthquake Centre Broadband Simulation Platform and the Graves–Pitarka simulation method. Site response analysis is conducted with two-dimensional basin models, and the nonlinear finite element software OpenSees. The dynamic response characteristics of the soft sedimentary layers are modelled with a pressure-independent multi-yield plasticity model. Some confidence in the simulation method is gained by undertaking historical validations, using the only instrumentally recorded earthquake of significance in the region (the Mw 4.7 2015 Lees Valley earthquake). The simulations provide close matches to the amplitudes and durations of the recorded time histories. The Akatore and Hyde fault earthquake simulations show peak ground accelerations of up to 0.8 g and 0.3g respectively, with durations of strong shaking of around 10 to 20 seconds. Uncertainty in the simulated ground motions due to source is quantified by comparing the spectra for repeated simulations, in which the range of source parameters are sampled. The resulting range of simulations shows a spread of as much as 0.5g. The Akatore – St Beach spectra are also compared to NZS1170.5 and New Zealand national seismic hazard model 2022 (NZ NSHM 2022) spectra, for site classes relevant to those of the St Beach site. In general, the simulated spectra exceed the NZS1170.5 spectra at the 0.1-0.3 second periods, but are similar to the mean NZ NSHM 2022 spectra at these periods. Future updates to NZS1170.5 based on NZ NSHM 2022 will therefore be expected to produce design spectra that are more consistent with the results of our study. The study represents the first ground motion simulations developed for southern New Zealand, and the simulation methods could be used to further advance understanding of seismic hazard in the region.