Reflections on the NSHA18: rethinking future seismic hazard assessments for Australia

An updated National Seismic Hazard Assessment of Australia was released in 2018 (the NSHA18). This assessment leveraged off advances in earthquake-hazard science in Australia and analogue tectonic regions to offer many improvements over its predecessors. The outcomes of the assessment represent a significant shift in the way national-scale seismic hazard is modelled in Australia, and so challenged long-held notions of seismic hazard amongst the Australian seismological and earthquake engineering community. The NSHA18 is one of the most complex national-scale seismic hazard assessments conducted to date, comprising 19 independent seismic source models (contributed by Geoscience Australia and third-party contributors) with three tectonic region types, each represented by at least six ground motion models each. The NSHA18 applied a classical probabilistic seismic hazard analysis (PSHA) using a weighted logic tree approach, where the model weights were determined through two structured expert elicitation workshops. The response from the participants of these workshops was overwhelmingly positive and the participants appreciated the opportunity to contribute towards the model’s development. Since the model’s publication, Geoscience Australia has been able to reflect on the choices made both through the expert elicitation process and through decisions made by the NSHA18 team. The consequences of those choices on the production of the final seismic hazard model may not have been fully appreciated prior to embarking on the development of the NSHA18, nor during the expert elicitation workshops. The development of the NSHA18 revealed several philosophical challenges in terms of characterising seismic hazard in regions of low seismicity such as Australia. Chief among these are: 1) the inclusion of neotectonic faults, whose rupture characteristics are underexplored and poorly understood; 2) processes for the adjustment and conversion of historical earthquake magnitudes to be consistently expressed in terms of moment magnitude; 3) the relative weighting of different seismic-source classes (i.e., background, regional, smoothed seismicity, etc) for different regions of interest and exceedance probabilities; 4) the assignment of Gutenberg-Richter b-values for most seismic source models based on b-values determined from broad neotectonic domains, and; 5) the characterisation and assignment of ground-motion models used for different tectonic regimes. This paper discusses lessons learned through the development of the NSHA18, identifies successes in the expert elicitation and modelling processes, and explores some of the abovementioned challenges that could be reviewed for future editions of the model. Abstract presented at the 17th World Conference on Earthquake Engineering (17WCEE )