2014 – Comparing Probabilistic and Historical Seismic Hazard Estimates in Southeast Australia

Russell Cuthbertson

Two techniques were used to calculate seismic hazard at a number of locations in southeast Australia. To simplify matters only Peak Ground Accelerations were compared.
The first technique used a seismological model of areal source zones that was based on the recorded seismicity as well as geological and tectonic inputs. Each zone was assigned a rate of earthquake activity that had been calculated from the recorded seismicity and a magnitude completeness function. Known geological faults that are also part of the model had to be excluded to allow a direct comparison with the second technique. A standard probabilistic seismic hazard analysis then gave PGA values versus return periods. This is the approach that has been used for the current Australian earthquake loading code (AS1170.4).
The second technique used a simple historical approach whereby recorded earthquakes were combined with an attenuation function to directly give the estimated return periods. This approach takes no account of tectonics, geological terranes or faulting – it simply uses the known, recorded earthquake catalogue. This is the technique used in the original Australian earthquake loading code (AS 2121).
The same ground motion attenuation function was used in both techniques but for a direct comparison the aleatory variability was set to zero in the probabilistic case because the historical approach did not include this effect.
In the historical approach the variability in completeness of the recorded catalogue was not considered. It was simply assumed that all earthquakes producing accelerations greater than a given value would be recorded over the last 100 years.
The comparisons were made for minimum considered magnitudes of 4 and 5.
There was general agreement between the two approaches especially at shorter return periods (lower PGA amplitudes). At longer return periods (higher PGA amplitudes) where there were higher uncertainties, the results at some sites diverged.
This simple comparison of two approaches to the same problem of estimating earthquake hazard is shown to be of value in ensuring that the AUS5 model used by SRC is producing results that are consistent with the historically recorded data.