Home / News and Events / Media Releases / How GNS Science estimates

How GNS Science estimates the possibility of future earthquakes 02/11/2013

We estimate the possibility of future earthquakes based on models of earth behaviour developed at GNS Science and informed by research done in other seismically-active countries. We typically use three models:

During seismically quiet times, we use the National Seismic Hazard Model to determine how likely earthquakes are in the long-term. This model is helpful in planning and in understanding how much shaking buildings should be able to withstand. The National Seismic Hazard Model is based on knowledge of more than 500 mapped faults in New Zealand, from which we can learn about earthquakes which occurred tens of thousands of years ago, and the more than 200,000 earthquakes we have recorded in New Zealand since 1840. Using these two datasets, we can estimate the average recurrence interval for large earthquakes in New Zealand and how strong the expected shaking might be.

A link to the the National Seismic Hazard Map

There will be aftershocks following a large earthquake. We calculate aftershock probabilities using empirical seismic models1 that yield a future probability for an earthquake of a particular magnitude in a specified time window. These are based on observations of aftershocks all over the world, but are optimised for the geology of New Zealand. These models tell us that after a large earthquake, the probability of an even larger earthquake is increased, but is still small. We used models of this type to provide aftershock probabilities for the Canterbury and Cook Strait earthquake sequences.

Our most recent models suggest that the probability that a magnitude 7 or greater earthquake will occur in central New Zealand is about 2% in the coming year.

To see our most recent probabilities and for more information about the aftershock modelling, please see

The third type of modelling we do is aimed at a better understanding of the probabilities for large earthquakes to occur elsewhere in the region surrounding the main shock. We know from worldwide aftershock histories that aftershocks larger than the main shock can and do occur. We evaluate scenarios for the occurrence of large events on faults we know about in the aftershock region; however because we lack accurate and detailed geological models2, for this type of modelling we do not have well established scientific models to fall back on and are reliant on cutting edge research. This means that we are not yet able to calculate the probability for a particular fault rupture to occur over short-time intervals, such as within the next year. What we can do, however, is look at the faults we know about and study how the state of stress on the a specific fault may be affected by the earthquakes happening in the region around it, which can give us an idea of which faults might rupture, if a larger earthquake were to occur.

How we estimate scenarios for Cook Strait
For the Cook Strait events, the largest stress changes we calculated were on the Awatere Fault in the upper South Island. The last large earthquake occurred on this fault in 1848. This suggests that the fault is currently in a very low stress state and is less likely to produce a damaging earthquake. Stress change on other larger crustal faults, including the Wellington Fault, was relatively minor. Stresses acting on the plate boundary in the Cook Strait region were slightly increased by the Cook Strait earthquake sequence; however, the amount they were increased was similar or less than the amount caused by the periodic (and currently) ongoing slow-slip events along the Kapiti Coast. In the Hikurangi subduction zone offshore the North Island, there is a mostly continual increase in the crustal stress until a big large earthquake occurs. The Cook Strait events were part of this gradual stress changing. At present we cannot calculate a precise probability for earthquakes of this type nor can we determine if the recent Cook Strait earthquakes will be the ‘straw that breaks the camel’s back’ and trigger a much larger earthquake, except to say that the chances are very small.

GNS Science has discussed these in publicly available places. Please see visit the following links for more information:

A quick summary of aftershock behaviour is:
1. After a large earthquake we always experience aftershocks that are usually smaller than the main shock. The aftershocks will decrease in time in a well-understood way.
2. At the same time, there is a small probability of experiencing a later, larger event triggered by the first event. How the probability of a larger event changes changes over tens to hundreds years is much less well understood than the aftershocks described in [1].
3. While we can calculate the likelihood of larger triggered events in the region, and do this routinely, it is difficult to ascribe reliable probabilities to particular scenarios. The best we can say is that in late August, the probability of triggering a M7.7 earthquake in the coming year was probably less than 1%.


1 Examples include the Modified Omori law and the Gutenberg-Richter relationship.
2 Many faults are buried and have little or no surface expression.