How does the Australian National Seismic Hazard Map affect your PSHA study?

Why is the NSHA18 Project important?

The National Seismic Hazard Assessment 2018 (NSHA18) was a collaborative effort of up to twenty specialist Australian Seismologists and Earthquake Engineers from private, public, and academic sectors in combining knowledge to produce an enhanced National Seismic Hazard Map of Australia.

Over several years, the group workshopped and created new model development, revised earthquake catalogues, updated magnitude conversions from original magnitude measures, and defined an ambitious logic-tree to incorporate more epistemic uncertainties than done previously. The final product is a National Seismic Hazard Map produced by Geoscience Australia (GA), as shown below.

Atc News Hazardmap Map 1

The issues with using the NSHA18 datasets in their final form

  • Misleading activity rates and b-values

Original model developers only provided the zone boundaries to their contiguous Seismic Source Models (or other types of models such as Gridded and Layered Seismic Sources) and did not have sufficient time to provide derived parameters such as “activity rates” and “b-values”. GA generally produced these in an automated statistical manner without an in-depth review of whether these parameters had merit among international scholars and upper and lower bounds of what is scientifically acceptable given the tectonic environment.

One problem that contributed to this issue is the lowest magnitude earthquake publicly useable for such a large continental scale. Together with lack of in-depth understanding of Stable Continental Regions (SCR) expected and acceptable lower and upper bounds of the “b-value” has led to over-estimated “b-values” in most of the more active earthquake regions within Australia, namely SE Australia and South Australia.

Since Australia does not experience significant or as many earthquakes as other active regions around the globe, we have a much smaller pool of available recorded earthquake data. A magnitude cut-off at ML 2.5 meant that some of the smaller seismic networks and their earthquake density were ignored and impacted the ability to accurately determine suitable “b-values” due to limited available data.

Higher than expected “b-values” especially in SE Australia (as high as 1.22) are documented for many zones within most of the models in this region which should be comparable to a subduction zone “b-value” rather than for an SCR. This erroneous use of “b-values” en masse effectively reduces the longer-term larger magnitude earthquakes that cannot be ruled out and increase the short-term smaller magnitude earthquakes. This creates lower seismic hazard overall and gives the impression that since AS1170.4 (2007), seismic hazard has significantly reduced, especially in active pockets of Australia, where in fact, this may not be the case. In the future, there could be a significant impact on assets evaluation in Risk Assessments, causing a decline in maintenance or importance and exposing these assets to higher risks.

  • Redefining an events magnitude

A significant revision to the national earthquake catalogue undertaken by GA involved a complex reclassification and understanding of original definitions of Seismic Moment (Mo) and Moment Magnitude (Mw) when applied to locally derived magnitude (ML) earthquakes, with varying definitions used over the last fifty years in Australia in quantifying an events “magnitude”. For example, ML has not been consistently measured by public and private sectors within Australia. This essentially means that not only definitions are blurred but also historical memory of how these were measured at that reference time have changed and never formally documented or consistently agreed upon by all those who record local earthquakes within Australia.

The definition of Mw (Mw = 2/3 log Mo – 6) is based on the observation that the stress drop for large earthquakes on active plate boundaries is “remarkably constant” (i.e. actually between 2 and 6 MPa). The stress drops in SCR such as Australia, where faults are stronger and dominantly revers, are much higher than this, up to tens of MPa, with significant effects on ground motion complicating the conversion from Seismic Moment to Ground Motion. The use of Mo and Mw are related to the earthquake source and permanent deformation, rather than the ground motion generated by the earthquake that we are dealing with in a ground motion hazard study. If all appropriate conversions are made, an equivalent ground motion hazard should be possible, but it would require more data, along with research and analysis, all with high uncertainties.

  • Accounting for epistemic uncertainty

Another fundamental issue is in accounting for the epistemic uncertainty. The NSHA18 has allowed for any fault mechanism to be considered appropriate even though in geological terms this is quite impractical and against the basic laws of physics. Considering each seismic source zone as one of three fault mechanisms: Reverse, Normal or Strike-Slip, and assigning an equal weighting to all is stating that all three very different tectonic processes can be at play at any one time. Earthquakes in Australia are nearly all on dominant reverse faults, which give higher stress drops and stronger ground motion. Therefore, looking at irrelevant faults to not an appropriate manner will apply “epistemic uncertainty”.

Given the Australian continent is under high internal stress with compression coming from all surrounding plates, the most probable fault mechanism (i.e. Reverse) is the by-product of such high stress and strain. So, by attributing two-thirds of the epistemic uncertainty to alternative fault mechanisms (Normal = pulling apart; Strike-Slip = moving in opposite directions next to each other), no wonder the overall seismic hazard has decreased for the Nation as a whole.

Atc News Hazardmap Deaggregation 1

How to correctly undertake a routine PSHA Study for your site

Probabilistic Seismic Hazard Assessment (PSHA) is conducted by specially trained Seismologists. In most cases, Seismologists start with declustering a complete earthquake catalogue to convert magnitudes to a single magnitude type for uniformity in their application and usage. Further, zones are defined, faults are identified, and earthquake magnitude recurrence rates are determined statistically before performing seismic hazard computations and provided results in a written report.

Extracting a full earthquake catalogue sufficient to cover all extents of the zone boundaries within a 400 km radius of the site means the total region we need to decluster from the earthquake catalogue could hold up to 10,000 individual earthquakes that need checking. Understandably, this is the most time-consuming component of a seismic hazard study, followed by the development of a seismic source model.

Once the catalogue is declustered and updated in the database, the earthquake magnitude recurrence statistics may begin and could take up to a week in a high volume of dense zones per model. Once finalised, these values are then populated in the software and inform our seismic hazard computations.

Talk to us about your PSHA requirements.

By Vicki-Ann Dimas BA/BSc(Hons) (Arch/Geol), PhD (in progress), ANCOLD Member

Vicki-Ann is ATCW’s Senior Associate Seismologist working in seismic hazard evaluation field throughout Australia and globally (primarily in Southeast Asia) with an academic background in geology. She has previously worked as a sub-contractor within the GIS industry and gained experience in a broad range of GIS applications, while also completing academic studies in GIS. Vicki-Ann’s professional skill-set within seismic hazard involves research and presentation at recent conferences including ANCOLD, ICOLD & SLNCOLD.

Expert Opinion