Toward a better model of earthquake hazard in Australia
Toward a better model of earthquake hazard in Australia (in Continental intraplate earthquakes; science, hazard, and policy issues, Seth Stein (editor) and Stephane Mazzotti (editor))
Special Paper - Geological Society of America (2007) 425: 263-283
The tectonic setting of Australia has much in common with North America east of the Rocky Mountains because stable continental crust makes up the whole continent. The seismicity is still sufficient to have caused several damaging earthquakes in the past 50 yr. However, uncertainties in the earthquake catalogue limit the reliability of hazard models. To complement traditional hazard estimation methods, alternative methods such as paleoseismic, geodynamic numerical models and high-resolution global positioning system (GPS) are being investigated. Smoothed seismicity analysis shows that seismic recurrence varies widely across Australia. Despite the limitations of the catalogue, comparisons of regional strain rates calculated from the seismicity are consistent with data derived from geodetic techniques. Recent paleoseismic studies, particularly those examining high-resolution digital elevation models, have identified many potential prehistoric fault scarps. Detailed investigation of a few of these scarps suggests that the locus of strain release is migratory on a time scale an order of magnitude greater than the instrumental seismic catalogue, consistent with Australia's lowrelief landscape. Numerical models based on the properties of the Australian plate provide alternative constraints on long-term crustal deformation. Two attenuation models for Australia have recently been developed. Because Australia is an old, deeply weathered continent that has experienced little Holocene glaciation, it has very little material comparable to North American "hard rock" site classification. The combination of relatively low attenuation crust under widespread thick weathered regolith makes the use of ground-motion and site response models derived from Australian data vital for Australian hazard assessment. Risk modeling has been used to assess sensitivities associated with variations in both source and ground-motion models. Systematic analyses allow the uncertainty in these models to be quantified. Uncertainty in most input models contributes a 30%-50% variation in the predicted loss. Where a city lies in a thick sedimentary basin, such as Perth, uncertainties in the behavior of the basin can result in a 500% variation in predicted loss.