Landslides in the Sydney Basin - Geoscience Australia

More documents

Recommendations

Info

Seismic Hazard in SydneyProceedings of the one day workshopTable 6 Results of sensitivity tests on mean recurrence intervalMEAN RECURRENCE INTERVAL (MA)84% MEDIAN MEAN 16%Preferred model 0.53 3.31 13.13 20.91Equal moment rate on strike-slip and dip-slip structures 0.15 0.84 2.68 4.60Moment rate distributed over only reactivated structures 0.19 0.72 1.46 2.41Maximum allowable fault length 10 km 0.48 2.99 10.91 17.78Maximum allowable fault length 18 km 0.53 3.52 14.01 22.42Maximum magnitude 7.3 0.68 4.26 16.50 26.77Maximum magnitude 7.7 0.40 2.57 10.48 16.4310 bar scaling relation 0.17 0.98 3.54 5.77In the section “Derivation of fault parameters and earthquake size from fault offset” we used a limiton maximum allowable rupture length to in turn limit the range of possible results from theprobabilistic model. The preferred model allowed only 50% of rupture lengths in the 10–18 kmrange to be used and none greater than that. We have run two sensitivity tests on the rupture lengthlimit, in the first allowing any rupture lengths less than 10 km and in the second allowing anyrupture lengths less than 18 km. The effect on mean recurrence interval is small compared to theprevious sensitivity tests (Table 6). The effect of allowing longer ruptures is to extend the tail of themean recurrence interval distribution on the long recurrence side.In the section “Derivation of fault parameters and earthquake size from fault offset” we used amaximum magnitude of 7.50.19 for the region to calculate moment release rate using the a- and b-values derived from historical seismicity. We have tested the sensitivity to this parameter byrunning models with the maximum magnitude set at 7.3 and 7.7 (a 1-sigma range). The results(Table 6) show only slight sensitivity to this parameter, with the effect being to move thedistribution of mean recurrence intervals to shorter values if a larger maximum magnitude for theregion is allowed. There is no effect on the shape of the distribution. The distribution of meanmagnitudes is unaffected because that is derived from the SED distribution, not historical seismicity.In the section “New fault data in the Sydney Basin” we chose two appropriate earthquake scalingrelations to derive rupture length and magnitude from displacement. These relations encompassedthe likely range of static stress drops expected for the Sydney Basin region, namely high values (inexcess of 100 bars) expected in continental regions such as continental Australia and Eastern NorthAmerica, and moderate values (~60 bars) derived from crustal New Zealand earthquakes in regionsof low to moderate slip rate. Both of these relationships were derived using modern waveformmodelling techniques. Before the advent of waveform modelling, some researchers attempted toestimate rupture lengths, and hence static stress drops, using aftershock distributions. This approachis now known to overestimate rupture length and underestimate stress drop. For example, stressdrop values of ~10 bars have been determined in this way for continental Australia (Denham et al.,1981; Mumme, 1984). We have used a 10 bar scaling relation in a sensitivity test. The results forthis model (Table 6) show a significant shortening of mean recurrence interval, but there is littlechange in the shape of the distribution. The mean recurrence intervals are still long, even with thisextreme test. The effect on the mean magnitude distribution is more marked with generally highermagnitude values (about half a magnitude unit) and a more sharply peaked distribution.Summary of sensitivity analysisThe sensitivity tests show that the way in which moment rate is partitioned between strike-slip anddip-slip structures and how it is distributed over the dip-slip structures will have a direct effect onthe mean recurrence intervals for surface rupture on any particular fault. However, in our tests of79
Seismic Hazard in SydneyProceedings of the one day workshopthese models mean recurrence intervals remain long. The parameters of allowable fault length andmaximum magnitude have much less effect on the estimation of mean recurrence intervals than thepartitioning and distribution of moment release rate. If extremely low values of stress drop wereused in deriving scaling relations, mean recurrence intervals would be shortened significantly.Given that low published values of stress drop were derived some decades ago using a method thatis now known to have some shortcomings, and more recent seismological data has been used toderive a high stress drop for a continental Australian earthquake, we do not consider a 10 bar stressdrop option as viable.CONCLUSIONSThe main conclusions as a result of this study are:New data on the extent and density of faulting in the 20-30 km wide coastal part of the SydneyBasin has been obtained from coalfield and tunnel log data show that the region is divided bynorthwest-trending strike-slip faults. These are the largest structures and are thus assumed to bedominant. Between these faults there is a pervasive distribution of north- to northeast-trending dipslipfaults, most of which have net normal displacements. Distributed among the normal faults aresome faults with net reverse displacements which we interpret as having sustained reverse faultingsince a mid Tertiary inception of an approximately east-west compressive stress field acrossAustralia.By developing some fault scaling relations and linking these with the seismic moment release ratemanifest as historic seismicity in the Sydney region a relationship between fault geology andseismicity has been achieved. This has been completed in a probabilistic framework such thatuncertainty and sensitivity to model parameters can be assessed.Two self-similar scaling relations have been adopted that embrace the likely range of stress dropsexpected for earthquakes in the Sydney Basin region. These relations are based on seismologicaldata from low slip rate Eastern North America and low-to-moderate slip rate regions in NewZealand.The outputs of the probabilistic modelling are mean and standard deviation recurrence intervals forsurface rupture on the north- to northeast-striking reactivated faults and their associated earthquakemagnitude. The mean recurrence interval is most sensitive to the way in which moment release rateis partitioned between strike-slip and dip-slip structures and the way in which moment rate isdistributed over the dip-slip structures. In constructing the logic tree we have been careful toembrace the full range of likely values for these parameters so that the output distribution accuratelyreflects the uncertainty in knowledge.The preferred result from the logic tree output is a recurrence interval for surface rupture of faults inthe north- to northeast-striking array of 13 +7/-12 million years for the mean and one standarddeviation range. The earthquakes associated with this faulting have magnitudes of M w 5.5±0.5 forthe mean and one standard deviation range.ACKNOWLEDGEMENTSNorm Abrahamson, Paul Somerville, Dougal Townsend, & Russ Van Dissen were contributors tothe project that this paper has been derived from. Keith Simpson and Peter Volk of CoffeyGeosciences Pty Limited provided the Sydney Basin tunnel log data. John Shepherd of ShepherdMining Geotechnics kindly provided information and personal observations on the Southern80
Page 5 and 6:
Seismic Hazard in SydneyProceedings
Page 7 and 8:
Page 9 and 10:
Page 11 and 12:
Page 13 and 14:
Page 15 and 16:
Page 17 and 18:
Page 19 and 20:
Page 21 and 22:
Page 23 and 24:
Page 25 and 26:
Page 27 and 28:
Page 29 and 30:
Page 31 and 32:
Page 33 and 34: Seismic Hazard in SydneyProceedings
Page 65: Seismic Hazard in SydneyProceedings
Page 120: Seismic Hazard in SydneyProceedings
show all

Landslides in the Sydney Basin - Geoscience Australia

Create successful ePaper yourself

Delete template?

Save as template?