Source models

Definitions

The earthquake sources capable of producing significant ground-motion at the site are identified and characterised from available source types. Source characterisation includes definition of source geometry (using a zone or polygon) and the probability distribution of potential rupture locations within the source. The source model selected for the site has been defined from four type of sources:

Finite fault sources represent a single tectonic fault. The complete fault geometry is defined by the upper and lower depth of the fault, dip and rake angles and strike direction. A finite fault source can be defined by its trace projection over the surface as a polyline, or by a 3D grid, representing subduction planes with varying dip and upper and lower depths. The occurrence of earthquakes is assumed to be uniformly distributed within the source. The size of the rupture area for a given earthquake can be estimated from the magnitude,61 or can be specified as covering the entire fault surface. All finite fault sources assume that rupture surfaces have a uniform probability distribution over the entire fault surface. Finally, the entire fault surface is divided in a regular mesh and all possible faults planes of a given size (magnitude) are considered. The productivity associated with that magnitude is assumed to be equally distributed between the corresponding planes.62

Regional source models represent regions of homogenous seismicity and are often used for modelling earthquake patterns with or without tectonic evidence. These models span over a single tectonic region representing a geological unit or tectonic feature: active shallow crust, stable continental crust, subduction in-slab, subduction interface or deep seismicity. The boundaries of the regional sources are adjusted to follow the surface projection of identified faults, avoiding the interruption of a fault system unless major differences are observed. A single productivity is assigned to each regional source under the assumption that there is a uniform likelihood of earthquake occurrence within the source. Therefore, the sources must be consistent with the earthquake catalogue, limiting their extent to regions where the productivity is uniform.

Background seismicity sources are used to incorporate the in-buffer seismicity. The background seismicity incorporates the earthquake pattern observed in regions with no identified faults. These models assume that the seismicity is spatially uniform over wide regions, and hence largely ignores the patterns shown by historical seismicity. The historical seismicity is distributed uniformly over broad regions.

Smoothed seismicity sources also called grid sources, are used to incorporate the in-buffer seismicity. The in-buffer seismicity includes the small magnitude events assumed to occur in the vicinity of the modelled faults. These models also assume that the historical seismicity patterns indicate the likely occurrence of future earthquakes, but rather than distributing the historical uniformly over the source zones, it performs spatial smoothing of the historical seismicity and does not require the definition of source zones.63 Grid sources are a collection of point sources distributed over a region. In principle each point can have a distinctive set of seismic properties (rupture plane geometry, maximum magnitude, ground-motion models, etc.). Grid sources assume that future earthquakes will tend to happen where previous earthquakes have occurred. The area to be modelled is covered in a uniform grid of evenly spaced cells. The productivity of each cell is based on the number of events recorded in its corresponding area. The productivities are then smoothed with a smoothing kernel, such as a Gaussian function with a correlation distance of three cells.64

A uniform probability of events occurrence is assumed inside the boundaries of each source. This assumption generally results in conservative estimates of seismic hazards but reflects the uncertainty that exists in identifying the precise geological features and locations where future earthquakes can occur. For areal sources, a grid of uniformly separated and randomly located epicentres is defined by a meshing algorithm. At each point of the grid the fault rupture geometry is modelled assuming a fault plane,65 its dimensions being derived from a regression analysis.66 Grid sources follow the same procedure, but the epicentre mesh is fixed by the source model.

In the hazard model prepared for the site, the sources are represented by a specific rupture plane associated with a moment magnitude. The minimum possible hypocentral depth is selected, keeping the rupture plane within the upper and lower depth boundaries specified by the source model. If the height of the rupture plane exceeds the maximum possible height for the source, the aspect ratio of the rupture surface is modified to accommodate the required area within the prescribed depth limits. The strike direction of the rupture plane is determined from the fault trace specified in the source model. Once the rupture plane of a scenario has been established, finite fault distances can be computed following the procedure described in Kaklamanos et al..67

Regional sources model

The study considers the seismic model of the Australian continent developed by Geoscience Australia68 for the National Seismic Hazard Assessment 2018 (NSHA 18). It consists of 20 peer-reviewed national coverage source models which were weighted by a panel of experts in order to account for the epistemic uncertainty.69 Figure 14 presents a sample of the different boundaries adopted for the areal models, the national fault-source model distribution (present in seven branches) and two source point models (b12 and b14).

Sample of NSHA 18 seismic source model branches

Figure: 14: Sample of NSHA 18 seismic source model branches

The source models are grouped into the following five seismic source models classes:

Background area-source models that use broad geographic zones within which earthquakes can occur anywhere with equal probability. For the NSHA 18, these are typically source models with 20 or fewer area-source zones on the national scale.

Regional area-source models that assume the spatial distribution of seismicity is not uniform at the scale of background source models and that the distribution of historical seismicity is useful to forecast future earthquake occurrence. These are typically models with 30 or more area-source zones on the national scale.

Smoothed seismicity data-driven models that yield spatially varying earthquake occurrence rates by smoothing the observed rates of earthquake occurrence with a given smoothing kernel.70 These models assume that historical seismicity is a good predictor of future seismic hazard.

Seismotectonic models that combine regional source models with the national fault-source model.71

Smoothed seismicity combined with a national fault-source model. These models combine short-term information from the instrumental catalogue with long-term geological information from the fault-source model.

Additional epistemic uncertainty was accounted for during the model parameterization. The source models are available in the following repository: https://github.com/GeoscienceAustralia/NSHA2018. The PSHA calculation was performed using the OpenQuake engine.72


  1. Donald L Wells and Kevin J Coppersmith, “New Empirical Relationships Among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement,” Bulletin of the Seismological Society of America 84, no. 4 (1994): 974–1002.↩︎

  2. Marco Pagani et al., “OpenQuake Engine: An Open Hazard (and Risk) Software for the Global Earthquake Model,” Seismological Research Letters 85 (2014): 692–702; “The OpenQuake-Engine User Manual.” (Global Earthquake Model (GEM) OpenQuake Manual for Engine version 3.5.0, 2019).↩︎

  3. P. Somerville et al., “Source and Ground Motion Models for Australian Earthquakes” (Proc. 2009 Annual Conference of the Australian Earthquake Engineering Society, 2009), 11–13.↩︎

  4. Arthur Frankel, “Mapping Seismic Hazard in the Central and Eastern United States,” Seismological Research Letters 66 (1995): 8–21.↩︎

  5. Keiiti Aki and Paul G. Richards, Quantitative Seismology, 2002.↩︎

  6. Wells and Coppersmith, “New Empirical Relationships Among Magnitude, Rupture Length, Rupture Width, Rupture Area, and Surface Displacement.”↩︎

  7. “Estimating Unknown Input Parameters When Implementing the NGA Ground-Motion Prediction Equations in Engineering Practice,” Earthquake Spectra 27 (2011): 1219–35.↩︎

  8. T. Allen et al., “The 2018 National Seismic Hazard Assessment for Australia: Model Overview” (Geoscience Australia, 2018).↩︎

  9. Jonathan Griffin et al., “Expert Elicitation of Model Parameters for the 2018 National Seismic Hazard Assessment: Summary of Workshop, Methodology and Outcomes” (Geoscience Australia, 2018).↩︎

  10. Frankel, “Mapping Seismic Hazard in the Central and Eastern United States.”↩︎

  11. Dan Clark et al., “Incorporating Fault Sources into the Australian National Seismic Hazard Assessment (NSHA) 2018,” in Australian Earthquake Engineering Society 2016 Conference, 2016.↩︎

  12. Pagani et al., “OpenQuake Engine.”↩︎