Home / Our Science / Natural Hazards and Risks / Earthquakes / Ground motion simulations / Alpine Fault and Hope Fault modelling

Alpine Fault and Hope Fault modelling

AF8 Alpine Fault modelling

This proposed model already published in Holden and Kaiser (2016) satisfies the conditions required by the source panel. The model is 500 by 12 km long, striking 52 degrees, dipping 45 degrees and with a rake angle of 173 degrees. As suggested by the source panel, it ruptures unilaterally from south to north with a rupture velocity of 2km/s. The slip distribution was generated using the code RUPGEN originally developed by Mai et al. (2000) and Mai et al. (2002). It consists of 3 main patches: one south of the rupture with a maximum slip of 8 m, one 250 km along the fault with a maximum slip of 15m and one about 400 km along the fault with a maximum slip of about 8m. The source is also characterized by tapered slip down to zero meter up to the surface and down to the depth of 12km.

Figure 16: Alpine fault heterogeneous slip model scenario employing RUPGEN toolkit developed by Mai et al. (2000) and Mai et al. (2002). The source model is 500 km long by 12 km wide with a maximum slip of 15 m half-way along the fault. The star represents the earthquake hypocentre

Alpine Fault heterogeneous slip model scenario employing RUPGEN toolkit developed by Mai et al. (2000) and Mai et al. (2002). The source model is 500 km long by 12 km wide with a maximum slip of 15 m half-way along the fault. The star represents the earthquake hypocentre

Figure 17: peak horizontal ground velocities (m/s) over the South Island for an alpine fault rupture scenario as illustrated in Fig. 1. Extreme peak velocities of 3.8m/s are modelled in the Haast region. Note: as described in paragraph 2.a, far-field results are to be taken with caution due to the regional model and approach employed in this particular calculation of ground motion

Peak horizontal ground velocities (m/s) over the South Island for an Alpine Fault rupture scenario. Extreme peak velocities of 3.8m/s are modelled in the Haast region. Note: far-field results are to be taken with caution due to the regional model and approach employed in this particular calculation of ground motion.

Preliminary broadband modelling of an Alpine Fault earthquake in Christchurch. (Holden and Zhao, 2011):

We have computed broadband synthetic seismograms in Christchurch for a large possible Alpine fault earthquake. In this preliminary study, we chose conservative values for all source parameters based on our current understanding of large crustal earthquake source mechanics. By using conservative parameters, we are attempting to model maximum possible ground shaking intensity. We computed the ground motion for generic rock sites in Christchurch. We subsequently superimposed the effects of soft soil condition on the modelled ground motions. Calculations for synthetic seismograms are based on a validated algorithm for large crustal earthquakes. Large accelerations are generated from localized asperities while the ground motions resulting from the rest of the fault rupture area are negligible. We distributed asperities where large surface fault displacements have been inferred from paleoseismic studies. For each asperity, records from a small or moderate earthquake were used as proxies for the Green’s functions. As such, they account for path effects incurred during propagation of the waves from the earthquake to the receiver site. Site effects for soft ground conditions were also added to account for possible amplification of ground shaking by soil layers in Christchurch.

The preliminary estimates for peak horizontal acceleration are less than 4% g. These results are reasonably consistent with recorded values from recent large earthquakes (Mw > 7) and distances of 150 km+.

Figure 14 The NS component of the synthetic Alpine fault  rock site acceleration time history in (a), and soil surface acceleration time history  in (b)

Figure 14 The NS component of the synthetic Alpine fault rock site acceleration time history in (a), and soil surface acceleration time history in (b)

Ground motion modelling of an Alpine fault earthquake and a Hope fault earthquake for main South Island cities (NZ) (Holden, 2011)

The large September 2010 and the tragic February 2011 Canterbury earthquakes caused widespread damage by ground shaking and sand liquefaction in the Canterbury region. Both earthquakes were less than 50 km from the Christchurch central business area and had a magnitude that is much smaller than that expected from the Alpine Fault (Mw=8.2) and that is similar to a potential Hope Fault event (Mw7+). At present, the response spectra from a great scenario earthquake from the Alpine fault can only be estimated from ground-motion prediction equations (GMPEs) based on local and overseas earthquake records. A major problem of all GMPEs, including even the latest GMPEs from the next generation attenuation (NGA), for example, Abrahamson and Silva (2008), is their very large variability in ground motion predictions for large and great earthquakes.

Recent advances in earthquake mechanics allow us to compute seismograms for realistic earthquake scenarios, at specific locations, and with specific site conditions. Such simulations can provide very useful alternative estimates of possible ground motions from large faults for major population centres. In this study, instead of using GMPE, we carry out synthetic broadband simulations to derive synthetic strong-motion records.

The synthetic broadband strong-motion records are produced for both a possible large Alpine Fault earthquake (Mw8.2) and a large Hope Fault earthquake (Mw7.1) at sites in a number of selected population centres that may be strongly affected. The synthetic records show that ground motion accelerations in Greymouth and Hokitika are expected to exceed 20%g and 50%g respectively during an Alpine fault earthquake, while ground motions in Christchurch are expected to be moderate, with peak ground accelerations (PGAs) of 8%g expected from an Alpine event and 6%g from a Hope fault event.

Synthetic ground motions from the broadband simulations are generally consistent with PGAs estimated from GMPEs. However the modelled PGA from an Alpine Fault event in Hokitika is 59%g, twice the level expected from a GMPE for a PGA of 27%g. This high PGA is likely due not only to non-linear soil response not accounted for in this study but also to the presence of a modelled asperity nearby and to strong directivity effects, neither of which are accounted for in current GMPE modelling for New Zealand.

Response spectra are computed from the synthetic ground motions, and are compared to those estimated from the New Zealand GMPE. In general the response spectra from the simulations exceed the spectra derived from GMPE except for spectral periods between 0.5 and 1.2 seconds. The duration of shaking is expected to last over 3 minutes for an Alpine Fault earthquake and at least 20 seconds for a Hope fault earthquake. Such a duration is comparable to the duration observed in Christchurch from the 2010 Mw 7.1 Darfield earthquake.

Figure 15: Horizontal and vertical synthetic acceleration histories for a Site Class D site as modelled from a Mw 8.2 Alpine fault earthquake in Greymouth, Hokitika and Christchurch (top to bottom). The number above each trace represents PGA in m/s/s.