Subduction zones are the source of the world’s most dangerous earthquakes and tsunamis. UTIG’s researchers are on a mission to understand them

By Constantino Panagopulos

On Jan. 26, 1700, a barrage of tsunamis ripped across the Pacific Ocean at the speed of a jet liner. The 100-foot waves slammed into the northwest coast of America and carved a path of destruction 3 miles inland.

Hours later, towering waves destroyed coastal villages in Japan, searing the catastrophic event into the cultural memory of people on both sides of the ocean.

The disaster was the result of a magnitude 9 earthquake at the Cascadia fault in the Pacific Northwest, which struck with such violence that the coastline dropped by at least 3 feet.

Cascadia is one of several subduction zones on the Pacific Ring of Fire. Similar faults encircle all the world’s oceans. By the simplest description, they are the colliding fronts of tectonic plates and the conduits through which most of the world’s tectonic energy is released. It’s no surprise then that they are responsible for the largest earthquakes and tsunamis.

It’s also why they’re the target of researchers at the University of Texas Institute for Geophysics (UTIG), one of the foremost investigators of subduction zone earthquake hazards worldwide.

“There’s a real big push to try and understand the earthquake and tsunami hazard posed by subduction zones,” said UTIG Research Scientist Laura Wallace.

Subduction zone earthquakes are a worldwide problem, but scientists have yet to find a reliable way of forecasting when the next big one will hit. Their unpredictable nature and lack of warning set earthquakes apart among natural disasters.

“They are the only large-scale natural hazard for which we are truly at the mercy of the planet,” said UTIG Director Demian Saffer.

UTIG researchers are investigating subduction zones around the globe in search of insights into how they work, using every tool and method available.

In Japan, they took the deepest- ever measurement of the country’s Nankai fault and installed a network of subseafloor sensors that monitor its every creak.

In New Zealand, where Wallace has a joint position with the country’s national geologic research institute, GNS Science, UTIG researchers are wiring up the Hikurangi fault to assess the risk it poses to the country’s east coast.

And in the Pacific Northwest, UTIG researchers have their sights set on Cascadia, where they recently co-led the first-ever complete subsurface imaging of the fault.

Other ongoing field investigations include subduction zones off the shores of Chile, Costa Rica, Mexico and Alaska.

Back in Austin, UTIG lab researchers have constructed a scaled-down fault zone for studying real earthquakes up close. Others are using supercomputers to re-create major earthquake faults in 3D.

And now a UTIG-led initiative is bringing all the pieces together to create the first physics-based earthquake forecasts, a feat that would put scientists’ understanding of earthquakes on the same track as weather and climate predictions.

“We’re bringing every method at our disposal to the table as part of a concerted effort to better understand subduction zone earthquakes and the hazards they pose,” Saffer said.

Wiring Up Faults in the Deep

Subduction zones are found in the deepest parts of the ocean and stretch for miles under the seafloor. To study them up close, UTIG researchers are using sensor-packed observatories drilled over a thousand feet into the seafloor.

Installing subsurface instruments in the deep ocean is no mean feat, but UTIG researchers have been doing it successfully for over a decade.

In Japan, where the Nankai fault has been eerily quiet for decades, Saffer, together with Japanese colleagues, co-led a series of International Ocean Drilling Program (IODP) expeditions, the last one in 2018, to place a new network of observatories on the seafloor where the fault faces Tokyo.

“The observatories let us hear every tiny creak, even those far out at the trench,” Saffer said. “It puts our finger right on the fault’s pulse.”

Among other details, the observatories revealed previously unknown movement that researchers think lets the fault periodically release some tension and could be important in understanding how it causes tsunamis.

With Japan’s Nankai wired up, Wallace and other UTIG researchers are now leading efforts to do the same at New Zealand’s Hikurangi fault, another Pacific Ring of Fire subduction zone.

Two observatories were installed in 2018 by an IODP expedition led by Wallace, but because of the pandemic, it was several years before the opportunity came up to retrieve the observatories’ recordings.

Finally, in March 2021, Wallace boarded the research vessel Tangaroa to deliver a truck-size remotely operated vehicle called ROPOS that travelled 2 miles down to the seafloor to dock with the observatory.

After 10 years working on the project, Wallace was understandably nervous. But when data finally streamed across her screen, the ship’s dimly lit operations room filled with elation.

There were whoops, high fives, and much relief, recalled Wallace.

“Putting sensors into the deep-sea environment is extremely challenging,” she said. “It’s like trying to put something into space.”

The wait was worth it. The observatory recorded priceless information about the occurrence of slow-motion earthquakes, called slow slip events.

Just like the largest earthquakes, slow slip events release tension between tectonic plates. But because they unfold so slowly — taking days or weeks — slow slip events pose little threat, at least not immediately.

Wallace points to several instances of slow slip events occurring before big earthquakes. In Japan it’s thought a large slow slip event may have triggered the massive 2011 Tohoku earthquake. If such a connection exists, slow slip events could help warn of an imminent dangerous earthquake.

The borehole observatories are a way of investigating whether there’s a connection by looking for patterns between slow slip and other seismic events as the fault moves through its cycle.

The results so far are encouraging and are already factoring into New Zealand’s latest earthquake hazard assessment models.

“We’re seeing where and when strain builds in the Earth’s crust in far greater detail than I ever would have imagined,” Wallace said. “That gets us a step closer to making that link between slow slip events and earthquakes, and potentially using them to forecast larger earthquakes.”

A Geologic Ultrasound

When an undersea earthquake hits in a particular way, it creates a whiplash in the ocean floor that triggers a tsunami. While subduction zone earthquakes are destructive enough on their own, it’s the tsunamis that make them exceptionally dangerous.

To find out which faults have the most potential to cause a tsunami, UTIG scientists are using seismic imaging to peer inside subduction zones and see what makes them tick. They’re finding that what plates are made of plays a big role in the threat they pose.

In 2017, Andrew Gase, a UTIG graduate student researcher, joined an international team of scientists aboard the R/V Marcus Langseth on the longest- ever seismic survey of the Hikurangi fault. For 35 days, the ship sailed along the 300-mile margin, performing an ultrasound of the fault using seismic pulses and a 7-mile-long microphone.

“The seas were calm, the equipment never failed, and we collected over 4,000 kilometers (2,500-plus miles) of excellent 2D seismic data. It was a complete success,” Gase said.

In practical terms, however, it was much like mowing the lawn.

“Well, they say the best cruises are also the most boring,” he said. Uneventful though it was, the data they collected kept Gase busy for the next five years.

What he found was a fault divided. In the north, where the fault is known to cycle harmlessly through regular slow slip events, he found a motley mixture of rocks of all sizes and origins, including ancient volcanoes and sea sediments. In the south, however, where measurements show that the fault is stuck, he found a mile-thick homogenous wedge of sand jammed in the fault right where the largest earthquakes are thought to happen.

Now a postdoctoral researcher at UTIG, Gase thinks that the motley rocks in the northern section mean the fault breaks more easily, releasing the locked plates before tectonic forces can build. Remnant volcanoes that are being sucked into the fault further roughen the fault and halt the occurrence of large earthquakes. In the south, however, the sandy wedge has made the fault rigid and smooth, which might create resistance with the tectonic plates until they slip all at once and smash together in a large earthquake.

For now, the question remains unresolved.

“We don’t yet know exactly why these materials cause different slip behaviors. That’s a challenge for the modelers to figure out,” he said.

Gase’s analysis of the seismic images does not show how close the fault is to failure, but it does confirm for the first time a hypothesis that UTIG scientists have tested for years in lab experiments: Rock type governs earthquake behavior.

Making Earthquakes

Postdoctoral researchers Srisharan Shreedharan and David “Chas” Bolton are tightening bolts on a squat, car-size metal frame. They’re in a warehouse at The University of Texas at Austin’s J.J. Pickle Research Campus, dwarfed by vats, pipes and large equipment, but the instrument they’re working on is by far the most unusual.

It’s called the Earthquake Machine. The device, which Shreedharan and Bolton have designed and constructed themselves, is in fact a replica of a subduction zone.

When fully operational, the machine will push two synthetic tectonic plates together and record what happens. Although its fault is only a little over 3 feet long, that’s still much larger than most lab-scale faults, which are usually only a few inches in length.

“What’s unique with this machine is that because it’s so large, it’s going to allow the actual ruptures to nucleate and propagate in ways that are more analogous to what actually happens along natural fault zones,” Bolton said.

This means that unlike typical lab- scale faults that break all at once, their Earthquake Machine is large enough that earthquakes should break along just parts of the fault.

It also means they can configure it to mimic a divided subduction zone like Hikurangi.

“What we can do is re-create something like a patchy fault, with carbonates here, volcanic clays there, and see how the earthquake evolves,” Shreedharan said.

The Earthquake Machine began simulating life-size earthquakes in late 2022. The researchers said it can be configured to mimic other kinds of faults such as San Andreas or the complex faults found in the Permian Basin.

It’ll soon be joined by the Friction Machine, a similarly large device that’s designed to simulate faults under the kind of pressure found miles underground and is currently being installed at the Jackson School of Geosciences’ GeoMechanics & GeoFluids lab.

Together, the two machines are the latest additions to a suite of instruments UTIG researchers are using to make lab measurements of each stage of an earthquake.

“There are few places with laboratory capabilities that can cover such a wide range of conditions and slip speeds,” Saffer said. “Having these instruments under one roof allows us to explore how earthquakes nucleate and evolve in different types of fault rocks, as they might in the real world, and that’s a pretty unique combination.”

In the Heart of the Subduction Zone

Japan sits at the meeting point of four tectonic plates, so it’s no surprise that each year, the island nation records more earthquakes than anywhere else in the world.

But there’s another reason so much seismicity gets recorded. On land and at sea, Japan’s faults are the most closely monitored and heavily instrumented in the world.

One of those is the Nankai Trough, a 430-mile-long segment of a subduction zone south of Japan that’s been the focus of UTIG scientists’ research for over two decades.

On Dec. 21, 1946, Nankai ruptured in a massive earthquake and tsunami that left thousands dead in southern Japan. History shows that a major earthquake occurs there every 100 to 150 years, so there should be detectable signs that the next earthquake is building.

Beginning in 2007, Saffer and colleagues from Japan, Europe and the U.S. co-led several ocean drilling missions to investigate. So far, the program has taken measurements at 25 locations, including one in 2019 that at the time was the deepest ever drilled into a subduction zone.

“This is the heart of the subduction zone, right above where the fault is locked, where the system should be storing up energy between earthquakes,” Saffer said.

According to Saffer, models predicted that 76 years into its cycle, tectonic stress in this part of the Nankai fault should be noticeably rising.

The models, it seems, were wrong. Saffer and his collaborators found that tectonic stresses were essentially

zero. Eerily, it was just like Tohoku, which researchers investigated soon after it shook Japan with a magnitude 9 earthquake in 2011. The Tohoku measurements made sense — the fault had slipped 100 feet during the earthquake. Nankai’s did not.

“It’s raised more questions than it’s answered,” he said. “It doesn’t violate the physics of the fault as we understand it, but it does revise the way we’re thinking about stress in these systems.”

Saffer said there were many possibilities for why the expected stress was not found. It could be that the locked zone is more slippery than previously thought and slips easily when strain starts to build.

The stresses may also be lurking deeper in the fault than expected.

Or it could be that the tectonic push will come suddenly in the coming years. Either way, Saffer said the drilling showed the need for further investigation and long-term monitoring of the fault.

A Rock in a Hard Place

In 2015, former UTIG research associate Adrien Arnulf made an intuitive leap that seven years later would lead to the most accurate 3D visualization of Japan’s subsurface ever constructed. The premise was simple: What if Japan’s vast seismic network — usually geared to listen for tremors — was detecting seismic pings from nearby ocean surveys?

When a ship performs a seismic survey, it pings sound waves into the Earth, each one of which is meticulously catalogued.

Looking back 20 years through the catalogues, Arnulf and his collaborators found thousands of seismic pings in and around the southern coast of Japan. When they laid the pings over the seismic network’s archives, they found each one had been detected by hundreds of sensors. Individually, each seismic ping was a pinprick in the dark, but when put together, they fully illuminated the subsurface for the first time.

Arnulf’s work lit up a buried mountain-size rock, called a pluton, wedged in the upper plate. It showed how the pluton forces tectonic pressure and buried fluids around it while earthquakes get channeled into its flanks. (Read more about the pluton)

Published in early 2022, the information is already being used for a major new government-funded project in Japan to find out whether another major earthquake is building in the Nankai subduction zone.

Arnulf’s pioneering project was a triumph of big data analysis and supercomputing. He has since left UTIG, but his former collaborators at UTIG have taken up the baton, continuing his work at other subduction zones, beginning with Hikurangi and eventually turning to Cascadia in the Pacific Northwest.

Cascadia

For the people living on Japan’s east coast, the tsunami of Jan. 26, 1700, seemed to come out of nowhere. The merchants who detailed the destruction wrote of an “orphan” tsunami that arrived with no accompanying earthquake.

The wave, of course, was no orphan. It was set off by a massive earthquake on the other side of the Pacific Ocean, at Cascadia.

For the Native American and First Nation communities who lived there, the event was catastrophic. More than 600 miles of the fault slipped, triggering waves that swept away forests and wiped out entire communities.

Since then, Cascadia’s subduction zone has remained mysteriously quiet. No significant earthquakes have been recorded in the hundred years or so since seismometers were introduced.

“The question is when the next big one will happen,” said Shuoshuo Han, a UTIG research associate who studies Cascadia.

According to Han, paleoseismic evidence in the form of lake deposits and petrified tsunami ripples show that Cascadia produces a near-magnitude-9 earthquake every 300 to 500 years. That means the next one could be on the horizon, but until recently not much was known about the fault.

That all began to change in 2021 when Han co-led a 41-day research cruise aboard the R/V Marcus Langseth to image the entire subduction zone — from Vancouver Island to the northern tip of California — for the first time.

Trailing a massive 9-mile-long antenna, the cruise produced spectacular images of the subduction zone, revealing previously unknown details about the geometry and physical properties of the part of the fault responsible for great earthquakes.

It’s taken a year to fully process the data, but Han hopes their findings will soon help bring the fault, and its actions, into focus.

“This data set will provide critical information for future computer modeling work on earthquake rupture along the Cascadia subduction zone,” she said. “I think it will greatly advance our understanding of this margin in particular, but also earthquake and tsunami processes at subduction zones in general.”

Getting Serious About Seismic Hazard Forecasting

Whether it’s observations from state- of-the-art instruments or using big data to pluck clues from archives, UTIG has an impressive record of making break-through discoveries about the world’s major subduction zones. The goal, of course, is to know when the next big earthquake or tsunami is most likely to strike.

That’s where UTIG’s computational modelers come in.

The processes behind earthquakes are, at present, too complex to be useful for predictions. But scientists are hopeful that a new generation of computer models could help them better understand the timing and strength of future earthquakes and even forecast their potential damage.

Among those leading the way is UTIG’s Thorsten Becker, who in 2021 launched a National Science Foundation-funded project to figure out the physics needed to make earthquake forecasting a reality.

“We’ve brought together people with expertise in different parts of the problem to see if we can build a model that captures it all,” said Becker, who is also a professor at the Jackson School’s Department of Geological Sciences.

Known as the Megathrust Modeling Framework, the goal is to develop models that glue everything together: earthquake nucleation, seismic cycles, tectonic strain, subducting materials, anything from the gaps between mud grains to the march of continents.

The models will merge and compare findings from three natural laboratories: Japan, New Zealand and Cascadia.

The five-year project is already developing computational tools to narrow in on uncertainties that are critical to understanding earthquakes. But Becker believes the surest way to achieve a quantum leap in earthquake science is to expand the community of people who are working on the problem.

In 2022, the project held its first summer school, which brought together graduate students and mentors from across the globe to work on a real-world earthquake hazard problem.

Organized and hosted by UTIG, the intense week-long program introduced students to earthquake physics and probabilistic hazard assessment — then asked them to estimate the likelihood of a major earthquake happening at Hikurangi in New Zealand within the coming years. The problem posed was intentional: It was exactly what the New Zealand government had asked Wallace and her GNS Science colleagues after the 2016 Kaikoura earthquake.

The students’ results were impressive, and comparable to the best hazard forecasts for the region. More importantly, said Becker, they identified questions about how to improve models that could be studied for years to come.

“Some really great ideas came out of the students’ projects,” he said. “My hope now is that they take the computational tools we’ve taught them and go and make them better.”

Becker and his collaborators don’t expect to be giving earthquake forecasts any time soon. But with an emerging generation of researchers who are fluent in geophysics and computational geosciences, he’s confident the science and tools to do so are within reach.

Saffer agrees that with UTIG researchers continuing to push the frontiers of subduction zone research in everything from deep ocean measurements and large-scale field experiments to computer modeling, they could soon be on the cusp of a great leap forward.

“That’s a strength of UTIG,” Saffer said. “We have a critical mass of researchers with a diverse set of tools in their toolbelts, approaching the problem from a very interdisciplinary point of view. You would be hard-pressed to find that in many other programs.”

Originally published in the Jackson School of Geosciences 2022 Newsletter.