Working out where the next big earthquake will come from

Friday, 30 April 2021

You’ve probably logged onto the GeoNet website and seen some startling numbers. If not, go have a look. There’s a 6 per cent chance of a magnitude-7 earthquake in central New Zealand in the next year. There’s an 8 per cent chance of a magnitude-7 earthquake (or larger) in the East Cape in the next 12 months.

You’ve also probably recently heard the next Alpine Fault earthquake will likely happen sooner than we thought.

These are what's called earthquake forecasts. They outline the chance of an earthquake occurring over a certain time period. They’re not predictions. Predictions are different, and we’ll get back to them.

Forecasts have a hidden impact on your life. Some inform the standards behind new buildings. They may even hit your pocket through changes in insurance premiums. So what are they exactly, and how do they work?

**READ MORE:

* The plan to get rid of district health boards and centralise healthcare explained

* Stuff Explained: Demystifying the complicated and making news accessible

* One of the smallest parcels of land in New Zealand explained

**

The basics of earthquake forecasting

Earthquake forecasting is a little like weather forecasting. It’s based on maths, knowledge of what’s happened in the past, and a dollop of human scientific judgment.

GNS Science has produced forecasts since the 1990s, but the 2010 Darfield earthquake, near Christchurch, and the uncertainty that followed, prompted more regular communication with an anxious public.

Think of an earthquake forecast as a completed puzzle or jigsaw. Every puzzle is made up of small pieces. And there are all sorts of puzzle pieces that inform these forecasts.

To give you an example, one piece is a concept called Omori’s Law – which helps define how the rate of aftershocks decays, or drops off, over time. This concept, along with, say, knowledge about a particular fault, could be used to build up a picture of how many aftershocks are expected after an earthquake.

Scientists can also make estimates on how likely a big earthquake is to occur on a particular fault in a certain time period. Soil samples, for instance, can tell the story of historical quakes and give us a glimpse as to what the future holds.

There are people out there who say they can predict the precise date, time, location, and magnitude of a quake. Most of these predictions are fuelled by an event (think animals behaving strangely) that is seen as a forerunner to a big shake.

The USGS website explains this very well. “The so-called precursor is often a swarm of small earthquakes, increasing amounts of radon in local water, unusual behaviour of animals, increasing size of magnitudes in moderate size events, or a moderate-magnitude event rare enough to suggest that it might be a foreshock.”

The problem is these precursor events happen all the time without big earthquakes following. But if individuals throw enough predictions at the wall, usually via social media, some will stick.

Will we ever be able to predict exactly when and where an earthquake will occur?

Don’t hold your breath. There has been some work done in this space in the past, but it hasn’t been successful, GNS Science hazards modeller Dr Matt Gerstenberger​ explains.

Most serious scientific work is now focused on developing better models to forecast the chances of earthquakes.

OK. Is there something that shows the risk for the whole of New Zealand?

There is. It’s called the New Zealand National Seismic Hazard Model (NSHM).

The last version was completed in 2010 and spat out an array of maps and datasets (thousands of them) that show the likelihood of earthquake shaking throughout the country in a 50-year period. The outputs are used by the government, engineers and insurers to plan for the worst, and ensure buildings are up to scratch.

The GNS Science map below is part of that work. It shows expected spectral acceleration (SA) in the next 50 years. This is a measure of how much acceleration, or movement, a building experiences. So, for example, in Wellington there’s a 2 per cent chance of a building experiencing 2.4gs or more over the next 50 years.

What’s a g? A g is a unit of measurement for acceleration. One g is the force gravity exerts on you and me, keeping us on the ground. Seven gs, for instance, is closer to what fighter pilots experience.

The NSHM is a very detailed, complex forecast, but a forecast nonetheless. It is, again, made up of puzzle pieces. But this time there’s a lot more of them. And some are really complicated.

A forecast showing the probability of aftershocks in the week after an earthquake is akin to a children’s puzzle – fairly simple stuff (relatively speaking).

The NSHM is more like that 10,000-piece puzzle of the Paris skyline that takes up your entire coffee table.

It uses all sorts of inputs, including everything we know about 550 faults across the country, along with nearly 200 years of historical earthquake data.

It also uses what’s called geodetic modelling. This is a field of science that measures just how the surface of the Earth is deforming or changing. GPS sensors throughout the country provide a picture of this.

There’ are also faults out there we don’t know about. The assumption in 2010 was that these faults could only cause up to a 7.2-magnitude earthquake. If a fault could cause a larger quake, we’d know about it by now. All of this needs to be factored in.

The fault at the Hikurangi Subduction Zone off the East Coast is an outlier and a big problem. Researchers don’t know much about past ruptures and therefore need to make a best guess at what it could do. It could conceivably cause a 9.0-magnitude earthquake, for example. Another significant puzzle piece.

Right now, scientists are working to create a new NSHM, which will be out in 2022. The model also includes subjective judgment – not as subjective as your opinion on the All Blacks winning the next World Cup, but it still involves extremely smart people making some calls.

For example, the puzzle pieces that make up the NSHM may well throw up slightly different results. One may see there’s a greater chance of a big earthquake in a particular place than another.

So essentially those very bright people need to get together and rank the importance of those different inputs. How do they do that?

“Well, people often talk about consensus but getting consensus in really complex areas is challenging. And sometimes it’s not even reasonable to get consensus. We aim for what’s called rational consensus,” Gerstenberger explains.

That means everyone agrees to the process used to reach consensus. It may well be that not everyone entirely agrees with the final result, but because the process was so good, they’re happy.

Wait, there are faults we don’t know about?

Yes. Earthquakes can occur everywhere in New Zealand. But even if there’s been an earthquake recorded in, say, Northland, you can’t necessarily just map out the responsible fault and what that fault could do in the future.

Some faults are hidden as there’s no evidence on the ground of when they last went off. These faults may have ruptured tens of thousands of years ago. Since then, the land could have “reset itself”, leaving no evidence of the dangers lurking underground. But that doesn’t mean those faults will never rupture again.

For example, scientists knew faults existed below the Canterbury plains. But they didn’t know about the specific fault that caused the 2010 Darfield earthquake. There was no evidence of it on the surface.

The underlying assumptions that these faults could cause a 7.2-magnitude earthquake will likely change in the upcoming model, most likely going up. Why? Well, because in recent years more evidence has emerged of slow-moving faults throughout the country.

Just a reminder, faults are fractures between two blocks of rocks. Faults essentially allow those two blocks to move about next to each other. A rapid slip or movement causes an earthquake. When the blocks are moving more slowly, relative to normal faults, they can still generate big earthquakes but not as often. There isn’t evidence for these faults on the surface. But they are there and they can cause problems.

What about the forecasts on GeoNet?

These forecasts typically outline what is likely to happen in a particular place over a certain time period. They’re called operational earthquake forecasting and are simpler than the NSHM.

They can act as a puzzle piece for the NSHM. Sometimes they are created because they are very much necessary, say for example, after a big and unexpected earthquake.

This is important. Scientists can’t just create a bundle of earthquake forecasts for particular places and then sign off, slapping themselves on the back, saying “job done”. The Earth is constantly moving. We’re constantly learning more about it. There are always new earthquakes and those earthquakes will change up the likelihood of something happening in the future.

Let’s have a look at the forecasts for Christchurch. This model shows the probability of earthquakes in a part of the Canterbury region for the next year.

Within the next year, it is very unlikely (less than 1 per cent) that there will be an earthquake of 7.0-magnitude or greater.

Now let’s look at central New Zealand.

Here, there is a 6 per cent chance of a 7.0-magnitude earthquake in the next year and a 30 per cent chance in the next decade.

You’ll notice some of these forecast what could happen in the next few months, some in the next year, and some reach out to a decade. It’s perfectly normal to have a broad time range. The Christchurch earthquake was so long ago now, there isn't a need for monthly predictions. But there is for the East Cape, after the recent tremors.

What about this Alpine Fault research?

So this research by Te Herenga Waka – Victoria University of Wellington senior lecturer Dr Jamie Howarth​ studied 20 previous Alpine Fault ruptures. It found that there’s a 75 per cent chance of a rupture on the fault in the next 50 years.

Interestingly, they also found a rather unusual “earthquake gate” where the Alpine Fault’s southwestern and central segments meet (there are four segments in total on the fault). This gate appears to define how big an earthquake will be. If an Alpine Fault rupture stops at the gate, the earthquake will typically be about magnitude-7. If the rupture moves through the gate, the quake can be magnitude-8 or more.

Think of these two segments as two distinct roads. The southern segment is a typical city road, with say two lanes, the central a large motorway. Now, think of the earthquake as a car – once that car reaches exits on to the motorway it can go a lot faster.

What this means is the make-up of a fault has a major impact on how earthquakes behave. The thing is, typically earthquake forecasts or models haven’t incorporated this.

Let’s tease this out a touch. We think of faults as lines on a map. But there’s more to them than that. The Alpine Fault, for example, dips and swoops through the earth. It subtly moves east and west. This is called 3D fault geometry.

“One of our observations was that really subtle changes in geometry exerts an impact on earthquakes in terms of magnitude and that isn’t accounted for, even the most advanced source models,” Howarth explains.

“The change in direction of a fault by 10 degrees, for example, or whether it’s dipping, can control whether an earthquake stops or not.” And if an earthquake doesn’t stop, it can cause more dangerous shaking.

The 2022 NSHM will seek to take some of these complexities into account. It will also acknowledge that faults are part of a greater system. You can't just look a single fault and say that could cause a single earthquake. Faults will be seen as cogs in a wheel of an entire system. We saw this manifest in real life during the Kaikōura earthquake in 2016. Then, more than 20 faults ruptured, not just one.

Anything else I should know?

In 2012, six Italian scientists were sentenced to six years in prison for manslaughter. Why? It was alleged they didn't properly communicate the risk of a big shake.

An earthquake subsequently occurred at the 13th-century city of L’Aquila, killing more than 300 people. The convictions were later overturned, but the whole saga was a worthwhile reminder of how fraught earthquake forecasting actually is.

The Earth’s crust is always moving, and our understanding of it continues to evolve. Forecasts may not be perfect. They may sometimes be off. But they are a lot better than nothing.