Lessons from the Destructive M_w 6.3 Christchurch, New Zealand, Earthquake

doi:10.1785/gssrl.82.3.371

At 12:51 local time on 22 February 2011 New Zealand’s second largest city was struck by a close, shallow M_w 6.3 earthquake. The event caused widespread damage in the city and resulted in about 180 deaths. At the time of this writing parts of the central business district are still cordoned off because of damage to structures (particularly unreinforced masonry). It is estimated that 800 buildings in the central business district will have to be demolished, and that 10,000 of the 140,000 domestic dwellings in the city will suffer the same fate. There has been extensive liquefaction and lateral spreading in the eastern half of the city, and 200,000 tons of ejected silt have already been removed. Costs of the damage are in the NZ$15–20 billion (US$11–15 billion) range. Given that Christchurch contributes 15% of New Zealand’s GDP, this earthquake will have a significant effect on the nation’s economy.

The 22 February event was an aftershock of the M_w 7.1 earthquake of 04 September 2010 at 04:35 local time (see Figure 1). This complex event caused surface rupture of up to 5 m on the previously unrecognized Greendale fault (a preliminary seismological report on this is contained in this issue of SRL (Gledhill et al. 2 011, 3 78−386). The potential for widespread liquefaction of soils in the Christchurch region during strong ground shaking has been well known for decades. Thus, given the high probability of a large earthquake on the Alpine fault (150 km to the west) in the next 40 years, a dense strong-motion network had been installed in Christchurch and the surrounding region. This network has recorded an exceptional set of strong ground motions, with vertical peak ground accelerations reaching 1.26 g close to the September rupture and 2.20 g close to the February rupture. Many of the close stations to this later event show clear evidence of the slapdown phase, which contributed to the extensive liquefaction. Data and information on the earthquake sequence is freely available on the GeoNet website, http://www.geonet.org.nz.

The Canterbury area where this earthquake sequence occurred is a region of continental convergence across the Pacific/Australian plate boundary. GPS measurements indicate that 75% of the 38 mm/yr of relative plate motion is accommodated by the Alpine fault, with a further 20% accommodated in the Southern Alps and their foothills. This leaves 5% (i.e., a few mm/yr) to be accommodated beneath the Canterbury plains (Figure 1). Prior to the September rupture of the Greendale fault, no past surface ruptures had been recognized in the immediate region. But the possibility of large earthquakes had been incorporated in the probabilistic seismic hazard model for New Zealand by assigning a maximum magnitude (Mmax) of 7.2 to distributed seismicity in the Canterbury region. The problem is that the extreme ground motions recorded for both the M_w 7.1 mainshock and M_w 6.3 aftershock were greater than this model predicted (even at 10,000-year return periods for the case of the M_w 6.3 aftershock).

The reason for this is that both these events radiated anomalously high levels of seismic energy relative to their magnitudes. For the M_w 7.1 September event, the energy magnitude (Me) is 7.4, while the M_w 6.3 February event has Me 6.7. This can be understood in terms of the anomalous crustal structure of the region. At approximately 10 km depth, we have a ~100 Myr plate boundary, marking the subduction thrust where the Hikurangi Plateau subducted from the north under the edge of Gondwana. This plateau is extremely strong and is recognized in tomographic images throughout New Zealand by the high seismic velocity (Vp > 8.5 km/s) in the eclogite layer at its base. It was originally part of the Ontong Java Plateau large igneous province. It is capped by schist and greywackes, which contain east-west Cretaceous faults active at the Gondwana margin.

Because of this crustal structure, there is no shallow brittle-ductile transition beneath the Canterbury region— small earthquakes extend down to 35 km depth in the mafic Hikurangi Plateau. Slip on the faults that moved in September and February is restricted to the rocks above the plateau and likely involves reactivation of east-west Cretaceous faults, which are well oriented in the regional stress field. The high levels of radiated seismic energy from the events indicate high stress drop and high fault friction. Both earthquakes show short fault lengths relative to slip amplitude, consistent with high stress drop (and long recurrence interval). GPS measurements more than five to six weeks after the September earthquake also show very little postseismic motion (less than ~2% of coseismic), consistent with the lack of a shallow brittle-ductile transition.

Southwest of Christchurch city, structure is disrupted by Banks Peninsula, an intraplate, basaltic shield volcano that was active over the period 12–6 Myr ago. The lower limit of seismicity in the rocks overlying the plateau decreases as we approach this old volcano, suggesting that the crust may be weaker there. An important question that needs addressing is what part this change in structure played in concentrating changes in Coulomb failure stress resulting from the September earthquake near Christchurch.

We are actively researching the above questions, using both seismological and geodetic data from the permanent GeoNet network and portable deployments after both earthquakes. But a few lessons are already clear:

• Beware the tyranny of the obvious. At an active plate boundary such as New Zealand, it is easy to be seduced by prominent active fault traces (you can see the Alpine fault from space) and belts of high seismicity in the highest strain regions of the plate boundary. But we should not forget that regions further removed from the plate boundary still need to absorb measurable strain, and will eventually produce damaging earthquakes (albeit with long recurrence).
• We need to incorporate crustal structure information into our probabilistic seismic hazard models, to better capture earthquake stress drop variation.   

Martin Reyners GNS Science m [dot] reyners [at] gns [dot] cri [dot] nz

---

To send a letter to the editor regarding this opinion or to write your own opinion, you may contact the SRL editor by sending e-mail to
<srled [at] seismosoc [dot] org>.

---

[Back]

 

Posted: 28 April 2011