There aren't many observations of an earthquake as large as Kaikoura: GNS

Friday, 21 July 2017

Aftershock forecasting in the constantly-evolving field of seismology is a tricky business.

Interest in GeoNet earthquake aftershock forecasts has grown considerably since the Canterbury earthquakes in 2010 and 2011.

Last year's Kaikoura earthquake was one of the more complex, multi-fault large magnitude shakes on record, extending more than 150 kilometres from the northeast corner of the South Island and beneath Cook Strait, shooting energy northwards to the lower North Island.

An earthquake of such a large magnitude – a magnitude-7.9 destroyed 80 per cent of San Francisco killing at least 3000 people in 1906 – is rare and the average rate of return is, statistically, in decades or hundreds of years.

**READ MORE:

* Complexity of Kaikoura quake may be clue to low number of aftershocks

* How GeoNet models aftershocks

* Kaikoura workers heading to roading jobs, leaving businesses struggling**

Since 1900, there have been almost 11,000 quakes greater than magnitude 6. However, earthquakes don't happen in isolation, and there is always a risk of a large magnitude quake in New Zealand.

Kaikoura was a massive earthquake.

Larger earthquakes have larger and more numerous aftershocks. Overall, for any magnitude, aftershock sequences follow a predictable pattern – the number and size fall away in a rough curve over time.

Aftershocks also decrease with distance from the epicentre.

GNS has been producing forecasts since the 1990s. These aren't predictions – it's not scientifically possible to predict an earthquake – but they do offer sets of likely scenarios after a large-magnitude shake.

A forecast usually has three outlooks, the short-term year ahead, the medium term years ahead, and the long-term decades ahead.

GNS Science seismologist Matt Gerstenberger and his team combine real-world observations with mathematical and statistical models to produce the forecasts.

He said the forecasts were only as good as the timeframe in which they were produced because the aftershock rates decayrf over time – in line with fundamental seismic laws.

Forecasts are an average snapshot in time across a range of probability.

'There's no single expectation of what it will be. We give an average range of what we expect it to be.'

Those averages weren't always in line with reality and the Kaikoura aftershocks were towards the lower range of forecast numbers.

Because the dataset record for big quakes is constantly updating and, therefore, incomplete, there is a larger margin of error and the lower than forecast aftershock numbers are perfectly normal and within the range. In other words, the more incomplete the record and data, the more uncertainty.

'It's generally tracked below the average.

'We have few observations of well recorded large earthquakes this big in New Zealand that we have data for. There are larger uncertainties because of that. The average is also less well constrained.'

Why?

In simple terms, the mathematical models used by seismologists around the world imply the rate of aftershock decay can extend for a century.

Large-magnitude quakes are less frequent and therefore the historical records for big quakes are less complete. There are also regional variations depending on geology, the type of quake, and tectonic plate characteristics.

The rate of aftershock decay isn't entirely smooth – there may be a cluster of aftershocks, or a longer than average quiet period.

Large magnitude earthquakes do not happen as often as the small, everyday events happening beneath New Zealand and, therefore, there aren't as many observations.

GNS handle this knowledge gap by combining models from previous New Zealand quakes and historical earthquake data from other countries. There are worldwide catalogues of seismic data dating back centuries.

But, by its very nature, the earthquake record is incomplete and with averages, probability, and forecasting comes a range of likely scenarios.

'None of them completely capture what we think is the reality.

'We have two main models. One of those models is tracking pretty close to the behaviour of what we're seeing. The other model has forecasts that are a bit higher.

'The forecasts are specific to right now – that's only good for the next two months.

'We've based it on what we know about average earthquakes and aftershocks in New Zealand and around the world.'

That rate of decay, governed by a mathematical principle called Omori's law, is the same for all magnitudes.

But, a larger magnitude quake produces more aftershocks – and larger ones.

Aftershock decay for large aftershocks is the same as small aftershocks but significantly fewer large aftershocks are expected.

Gerstenberger said one way to think about earthquakes and aftershock decay is to realise the enormous forces beneath the surface of the earth are not happening in isolation.

'[Seismology] evolved into thinking about families of earthquakes; every earthquake has some influence on other earthquakes. The international community will be working on this earthquake for a long time.'

Datasets are still being produced, seismologists are still poring over the numbers, and analysts are still crunching those numbers to get a better understanding of what happened and what is likely to happen in the future.

Most of the 17,492 aftershocks since the 7.8 Kaikoura earthquake occurred across north Canterbury and Cook Strait with some in the lower North Island.

As of June 20 – the next forecast is due in August - there were five greater than M6, 61 shakes between M5 and M5.9, and 549 between M4-4.9.

All of the shakes greater than M6 happened within a short time of the 7.9 Kaikoura earthquake.