The challenges of probabilistic seismic hazard assessment in stable continental interiors - an Australian example

In stable continental regions (SCR) the undertaking of probabilistic seismic hazard assessment remains a scientific and technical challenge. There is no seismo-tectonic model, akin to the plate tectonic model, to explain seismicity and guide source zonation, ground motion prediction equation (GMPE) selection, recurrence statistics estimation or MMAX selection. A model of seismicity being episodic, in regions of stable continental crust, has emerged in recent years. This model is consistent with Australian seismological, geological and geodetic data and is used to guide the zonation for a new national seismic hazard map of Australia. The selection of MMAX is based on the analysis of fault scarps from palaeo-earthquakes, with MMAX thought to between geological domains The heterogeneous Australian catalogue, arising from: the variability between magnitude scales, (particularly at regional distances >250km), sparse networks and poorly constrained Mc, results in high epistemic uncertainty in the recurrence parameters a & b. To account for changes in magnitude scales around 1990, the magnitudes of pre-1990 earthquakes have been empirically corrected. Due to the heterogeneous catalogue and small numbers of earthquakes, existing methods (e.g. Maximum Likelihood) for estimating frequency-magnitude recurrence parameters (a & b) were found to be unstable. To overcome these problems, a new method, that removes outlier earthquakes before applying a least squares regression , was developed. The sensitivity of PSHA to MMAX, zone boundaries, recurrence parameters and GMPEs was examined. The hazard was insensitive to MMAX in the identified range (e.g. 7.2-7.6 in non-extended SCR and 7.4-7.8 in extended SCR). The uncertainty in recurrence parameters was found to contribute similar variation in hazard as the epistemic uncertainty associated with the different GMPEs used in this study. For sites near zone boundaries, a similar variation in hazard was observed by reasonable changes in the position of the boundary. Aleatory variability and epistemic uncertainty in GMPEs are routinely incorporated in PSHAs, as is variation in MMAX. However, the uncertainty in recurrence parameters and zone boundaries are normally ignored. From MODSIM2013 Conference