Landslides in the Sydney Basin - Geoscience Australia

Seismic Hazard in SydneyProceedings of the one day workshopThere is no reliable estimate of stress drop available for the Cadoux earthquake from eitherteleseismic body wave modelling or elastic dislocation modelling, it being too small to producegood quality data. The stress drop derived from the surface rupture length is ~28 bars (Denham etal., 1987), although that study also cites overcoring and hydrofracturing measurements that showshallow stresses are high — in the 200–300 bar range. The study concludes that this area hascomparatively high levels of crustal stress. The apparent low earthquake stress drops are thus mostlikely to be due to post-seismic fault expansion. There is even less seismological data for theMarryat Creek earthquake, again because of its small size. The most likely explanation for the longrupture length is, as for the other continental events, expansion of the surface fault extent due tostress triggering effects.Observations of surface fault displacements vary significantly in size along strike. For the historicalAustralian surface ruptures we have plotted the range of observed surface displacements as afunction of magnitude in Figure 6b. We have excluded very small values near to the ends of faults.Our weighted ENA/NZ scaling relation is completely consistent with these observations. Note thatthe uncertainties are far too great to be able to define a relationship in its own right.DERIVATION OF FAULT PARAMETERS AND EARTHQUAKE SIZE FROM FAULT OFFSETWell-determined reverse displacement of 35-50 cm has been observed in two exposures of a fault inthe region and we use this range to explore the range of fault and earthquake parameters that areconsistent with this. Scaling relations can also place some constraint on likely SED’s. To get asurface rupture, we expect rupture widths greater than ~2 km, given that the top layer of crust is lessseismogenic (based on local seismicity being distributed over the 2–18 km range and decreasingfault offsets towards the surface observed in coalfield data). A 2 km rupture width equates to aM w 4.9 earthquake with ~21 cm average subsurface slip. This equates to ~14 cm of surface slip,after a surface/subsurface slip correction of 0.67 has been applied (Stirling et al., 2002). On thisbasis, the 50 cm offset could comprise three SED’s of ~17 cm, two of 25 cm, or one SED.Work by Hecker and Abrahamson (2002) has shown that the distribution of slip at a point on a faulthas a coefficient of variation (the standard deviation divided by the mean), C.V., of ~0.45 (for dipslipfaults). This is a statistical variability that may reflect true variability in the physical processand has been built into the probabilistic modelling.Our aim is to determine mean recurrence interval using the rate of historical seismicity andobservations of surface fault offsets. As part of this process we need to determine the geodeticmoment, for which we need to know the average subsurface SED, not the surface SED, so that it canbe compared with the moment release calculated from the seismicity, which is distributed in the 2–18 km depth range. The subsurface rupture lengths used above to calculate the subsurface SEDwere largely derived from aftershock distributions, which are known to overestimate rupture length.This effect is shown in Figure 7, where magnitude is plotted against subsurface rupture length forthe Stirling et al. (2002) data set and for the seismological data set used in the derivation of the lowto-moderateslip rate New Zealand scaling relation. The New Zealand data have consistently shorterrupture lengths for a given magnitude than the Stirling et al. (2002) data. We have used regressionson the two moment-length data sets shown in Figure 7 to correct for the Stirling et al. (2002) rupturelengths being overestimated. The regression on the New Zealand data included a factor assuming20% of the ruptures were bilateral.