A deep dive into earthquake forecasting

Scientists studying the world’s most dangerous earthquakes are racing against time to get up close and personal to their sources.

The need is urgent: these super-dangerous fault lines are in subduction zones like those that produced devastating tsunamis in Indonesia in 2004 and Japan in 2011. They threaten coastal communities around the globe with earthquakes up to magnitude 9.5.

“It’s really the biggest risk in terms of loss of life,” says Laura Wallace, a research scientist at the University of Texas Institute of Geophysics (UTIG) in the US and GNS Science in New Zealand.

“It’s really the biggest risk in terms of loss of life.”

Laura Wallace, University of Texas

There’s just one problem. These fault lines tend to occur at the bottom of marine trenches, thousands of metres below the ocean surface, where plates collide and the relatively light rocks of the continents slowly slide across denser seabed rocks, forcing them downward into the Earth’s mantle.

For generations, this process and the earthquakes produced have been studied via onshore seismic networks or, more recently, high-precision GPS stations that can monitor crustal movements with millimetre-level precision.

But you can’t learn the nuances of how these processes work from a distance, because the seismic details are muffled by 100km or more of intervening rock.

If the fault unleashes a true earthquake, onshore instruments aren’t going to miss it. But much of the action takes the form of gradual movements, called slow-slip events, in which the fault moves steadily over the course of days, weeks or months, rather than in the type of brief lurches that generate earthquakes.

Slow-slip events don’t pose any immediate threat, and are in fact a good thing because they signal that the fault isn’t locked. Instead, strain is gradually released before it can build up to dangerous levels.

But, Wallace says, the reduction of strain in one region may simply transfer it to a nearby region. “It’s a double-edged sword,” she says. “One region’s reduced risk might lead to someone else’s increased danger.”

You can’t learn the nuances of how these processes work from a distance.

To truly keep tabs on this process, she says, it’s necessary to go to sea. And unfortunately, that isn’t cheap.

The instruments themselves aren’t all that expensive: a few hundred thousand dollars per site, says UTIG’s director, Demian Saffer. But they can’t just be placed on the seabed and left there. They need to be installed in boreholes penetrating hundreds of metres into the rocks below.

Going that deep is important for two reasons, he says: “First, it gets us away from surface noise from ocean currents. And it’s more strongly coupled to the rock itself, so it’s more sensitive.”

Not that the instruments themselves are at the bottom of the borehole. Rather, they are positioned at the top, where they can be maintained, when necessary, by remotely operated submersibles. From there, they measure minute pressure changes conducted up from the base of the borehole, changes that show how fluids in the pore space between the grains of the underlying rock are being affected as the rocks are slowly squeezed, strained, or deformed by tectonic forces.

“It’s extraordinary, the sensitivity that we are able to achieve,” he says. “This allows us to monitor tectonic motions at a place that’s really crucial with a precision that’s really unprecedented.”

“It’s extraordinary, the sensitivity that we are able to achieve.”

Demian Saffer, University of Texas

The real expense is in drilling the boreholes.

That takes about $16 million to $20 million per hole, plus two months at sea. “Eight weeks on a boat with no vegetables,” Saffer quips – though it’s clear that, to an ocean scientist, that’s not really a hardship.

“It’s pretty fun, actually,” Wallace says.

So far, borehole networks have been installed on two subduction zones, one near Japan and the other in New Zealand – two earthquake-aware countries with the financial ability to support such research.

The network in Japan is located south of its main island, where it monitors a subduction zone that directly threatens Tokyo, among other places.

The one in New Zealand is near the Hikurangi Trench along the east coast of North Island. That trench is widely viewed as New Zealand’s most dangerous seismic and tsunami hazard and was the site of a magnitude 7.1 earthquake in 1947 that created an 8- to 10-metre tsunami.

The Japanese borehole instruments are connected to shore via fibre-optic cables that allow real-time monitoring of data. New Zealand hasn’t yet instrumented its seabed to that level, and its data have to be obtained by autonomous submarine vehicles which periodically visit the boreholes, dock with their seabed instruments, and download the data for return to land.

At the moment, the difference in how quickly the data is returned doesn’t really matter, as scientists are still studying it for correlations – something that requires abundant patience.

But already, interesting insights are emerging. In 2015, Saffer says, the Japanese study site was able to monitor a slow-slip event that progressed for three months – and was followed shortly after by a magnitude 7.0 earthquake offshore from the Ryukyu Islands, hundreds of kilometres away.

The Japanese study site was able to monitor a slow-slip event that progressed for three months – and was followed shortly after by a magnitude 7.0 earthquake.

The New Zealand instruments haven’t been operational for as long – only since 2018 – so they’ve not yet had the opportunity to correlate slow-slip events at their borehole sites to larger earthquakes elsewhere. But that doesn’t mean they aren’t returning important data. “There are things we’ve never been able to see before,” Wallace says – noting how it was possible to observe a 2019 slow-slip event occurring in three phases along different parts of the fault, over the course of about two months.

“We see the creaks and groans very close to these events,” she says.

Ultimately, Wallace and Saffer say, the goal is to learn how these slow-slip events correlate to changes in the risk of larger earthquakes. “I don’t think we will ever be able to predict earthquakes,” Wallace says, “but doing a better job of forecasting is certainly, I think, on the horizon.” The distinction, she adds, is the difference between making a definitive prediction (a magnitude “x” earthquake will occur at this point at such-and-such a time, for example) and an assessment of probability – akin to a weather report suggesting a 30% chance of showers.

Such forecasts, however, are only possible for subduction zones where instruments have been installed – and only then if they are connected to fibre-optic cables that can return data in real time.

Proposals are already under consideration for the US Pacific Northwest, Saffer says, and another in Southern Japan. But the goal is clear. “Every subduction zone in the world should have these kinds of instruments,” he says.