THEY are called megathrusts. They shunt entire coastlines centimetres in a matter of seconds, and send devastating tsunamis half way round the world. The largest earthquakes of the past century were megathrusts with magnitudes greater than 9, dwarfing the recent and more deadly 7.4 earthquake in 1999 in Kocaeli, Turkey, which claimed over 17,000 lives and displaced about a quarter of a million people.
Many regions where tectonic plates meet suffer from megathrusts, and tens of millions of people worldwide live in the knowledge that their cities, which include Vancouver and Mexico City, are due, or overdue, for a big one. Yet so far, no one has been able to tell exactly when and where they will strike.
Megathrusts happen when an oceanic plate sliding down underneath a continental plate jams against it. Stress builds, until the jam suddenly and catastrophically gives way, thrusting the oceanic plate under the coast. To predict these events, researchers need to know what’s going on in the heart of the subduction zone where the oceanic plate sinks. They can glean some clues from seismic signals produced as the plates grind against each other, but these haven’t revealed enough to produce predictions.
Advertisement
Recently, however, researchers have spotted an almost imperceptible signal – a kind of “silent quake” – that seems to provide some warning of impending disaster. And with so much at stake, emergency planners in Mexico City, Vancouver and Tonankai, Japan, have already begun to react.
The first clue came on 5 May 2000, when Kazushige Obara of the National Research Institute for Earth Science and Disaster Prevention in Tsukuba, Japan, was looking at one of his seismological charts.
When it comes to earthquake monitoring, Japan is the most wired country in the world. Obara runs a network of more than 500 seismic stations, called Hi-Net. At each Hi-Net station, sensors are buried in 100-metre-deep boreholes to protect them from surface disturbance. Even so, the detectors are so sensitive that Obara prides himself on being able to monitor far more than just tremor activity at each station. “We can easily recognise the weather conditions by the amplitude of background noises,” he says.
The pattern on Obara’s chart that day looked like noise from the Earth’s surface, the sort of thing he would expect if a truck drove near his seismological sensor. Except the pattern lasted for hours. And when Obara checked the charts for other sensors, he found this “truck” appeared simultaneously near five stations over a distance of 100 kilometres.
At the Hi-Net headquarters in Tsukuba, data from the seismic stations are plotted every hour. When there is an earthquake, or a small tremor, it appears on the chart as a distinctive signal, rather like the way an instrumental solo might appear on an orchestral score. But the signal that Obara saw was very different. It looked noisy, as if many instruments were joining in at once, all playing different notes.
Obara’s first thought was that a large landslide might have hit the region. But none was reported. He then thought of some local laboratories that sometimes ran huge “shaking table” machines to test buildings for structural weaknesses that might cause them to give way in an earthquake. But, he remembered, 5 May is Children’s Day, a national holiday during the Golden Week of holidays in Japan. Workers at the shaking-table labs had the day off to spend with their families.
Baffled, Obara resolved to keep an eye out in case anything like it happened again. And sure enough it did, just a few months later, so he decided to systematically monitor the strange quakes for a while. He developed a program that could automatically detect and locate them as they happened.
Meanwhile, Herb Dragert of the Geologic Survey of Canada was following the movement of fixed GPS reference stations in the north-west of the country whose positions are tracked by satellite. Dragert is interested in “silent” movements of the crust, too gradual to generate seismic waves. Off the west coast of Canada and the northern US, the Juan de Fuca ocean plate moves north-east, slipping under the North American plate and pushing the coastal area north-east, so Dragert expected to see GPS stations on the coast moving north-east at about 10 millimetres a year. But over time, he began to notice some of them behaving oddly.
About seven of his inland stations would sometimes take a back step in their motion, moving south-west by two to four millimetres. The ones further towards the sea continued to move north-west. This suggested the inland portion of the continental plate was suddenly slipping past the oceanic plate, instead of being pushed by its motion. The effect was a slow and gradual slip, taking a week or two to complete. Still, a few millimetres is a long way. If a coast moved this far in a matter of seconds, it would constitute an earthquake with a magnitude of about 6.7. But like most researchers, Dragert assumed the slow slip had to be seismologically silent. It was just too gradual to trigger the shockwaves that seismic stations pick up.
That soon changed. In 2002, Obara published his description of the mysterious noisy tremors in Japan (Science, vol 296, p 1679). He argued that the tremors probably came from a region about 30 kilometres down, where the subducting plate reaches the Earth’s mantle. As with the Canadian slips, the epicentres of the tremors were somewhat inland from where the leading edge of the Philippine Sea plate descends north-eastwards under the continental Eurasian plate that carries Japan.
Uncanny coincidence
When Dragert read Obara’s paper, he realised the similarities between the Japanese tremors and the Canadian slips were uncanny. Both were about 30 kilometres deep, and both lasted one or two weeks. Also, both appeared to ripple along the subduction zone at a similar speed, around 13 kilometres per day.
The coincidence suggested an obvious connection: perhaps the seismic signals Obara had picked up were the sound of the “silent” quakes. To check the idea, Dragert and his team needed to know whether and when there had been tremors like Obara’s in western Canada, so they asked Garry Rogers of the Geological Survey of Canada to print out seismological charts for the region from 1996 to 2002, and placed these alongside their record of slow slips.
“There was a one-to-one correlation with elevated seismic noise, which turned out to be identical to the tremors described by Obara,” Dragert says. The higher sensitivity of the Japanese network made them more obvious to Obara, but Dragert could see them too, now he knew roughly what to look for. The “silent” quakes were certainly very, very quiet, but they weren’t silent.
When Obara read Dragert and Rogers’s study (Science, vol 300, p 1942), he knew immediately he had to go in search of the movements that might explain the tremors. He measured the motion of the Earth’s crust at each of his seismic stations. Again, he found it was usually orientated one way, but every few months it reversed for about a week as the continental plate moved relative to the oceanic plate, just at the time when he was detecting the noisy tremors. “At last, we succeeded in detecting the slow slip associated with the tremor activity,” he says.
Matching seismic noise to “silent” quakes raises the hope that they might be useful to emergency planners. GPS monitoring is expensive, but many earthquake zones are already wired up with seismological networks. Although most networks are not nearly as sensitive as the Japanese one, they should be able to detect and follow slow slips.
Dragert has already found several ways that quakes might help to predict when jammed plates will give way in a megathrust. Quakes happen not at the portion of a subduction zone where plates are actually jammed, but many tens of kilometres inland, in a region where it has always been thought plates were slipping freely past each other (see “Softly does it”). The Japanese and Canadian findings suggest that even here, the tectonic plates get stuck, and then move on again.
To find out how regular the slips really are, Dragert went back to older GPS data. Southern Vancouver Island is home to Canada’s second-longest-running continuous GPS station. Since it was established in May 1992, Dragert noticed that this station has moved south-west for a few days roughly every 14 to 15 months. “For a tectonic phenomenon, this is surprisingly regular,” Dragert says. Several times already, Dragert and his colleagues have been able to predict them a few months before they happen. In Japan, Obara found the strange quakes were even more regular, happening around every six months.
The timing of the strange quakes seems to reflect the loading situation closer to the coast, where the two plates are jammed. Imagine a precarious pile of dishes on a table, says Dragert. “If you are pulling slowly on the tablecloth, that pile could tumble at any moment.” Just as in the case of a megathrust earthquake, it is very hard to say when exactly it will happen. “Now,” says Dragert, “start pulling the tablecloth in small sudden repeated jerks. It is more likely that the pile will tumble at the time of a sudden jerk.” Quakes represent sudden pulls on the jammed portion of a subduction zone. So if a subduction zone is due for a big one, chances are it will happen during a period when the plate is about to slip anyway.
So how might this affect people living near subduction zones? Instead of being prepared for a megathrust all the time, they can look forward to the risk being concentrated during the week or two of the quake. “This is good news all around,” Dragert says. “Because we can detect tremor and slip, we can identify the time at which the probability of an earthquake happening goes up.”
“Silent” quakes can also help pinpoint where and how large the coming megathrust will be. In the region of the quakes, the plates are slipping regularly, so any accumulated stress is regularly relieved. The area at risk is the locked portion of the subduction zone “upstream” of the quakes. Dragert thinks this means that the boundary of the quake activity should indicate the boundary of the danger zone (see “Wrath of the thunder gods”). In Canada, this region extends northwards from northern California to central Vancouver Island, and inland almost, but not quite, as far as the city of Vancouver. Using this information, Dragert calculates the next megathrust will be seaward of Vancouver and Seattle, but could be very large, between 8 and 9.5 in magnitude.
No one knows how many subduction zones worldwide experience similar, slow-moving quakes. In Mexico, for example, any kind of short-term advance warning of a megathrust earthquake would be welcome. Seismologist Vladimir Kostoglodov of Mexico’s Independent National University Geophysics Institute says, “we are expecting a large, magnitude-8 thrust earthquake in the so-called Guerrero gap, which extends from Acapulco some 120 kilometres north-west along the Pacific coast.” Now the group is looking for noisy seismic signals like the ones Obara first saw. Kostoglodov says Dragert’s work on using silent quakes in prediction could make a big difference to emergency planning by the authorities. “We are following him step-by-step with our observations in Mexico’s subduction zone,” he says.
One mystery remains, however. Just why do “silent” quakes make any sound at all? Their seismic signal looks like those seen from volcanoes when the magma inside them moves. It is not well understood how the moving magma creates vibrations in the rock nearby, but it could be similar to the way wind instruments resonate as air flows through them. Mounting pressure crushes seawater out of the sinking ocean plate, and Obara calculates that this should happen at about the depth where the quakes seem to originate. So it may be that deep below the ocean surface, gurgling seawater is making the Earth hum, a whispered warning of faraway danger.
Wrath of the thunder gods
There are no written records of large earthquakes in the Pacific north-west of Canada and the US since Europeans reached there. But the myths of native tribes from Vancouver Island in Canada down through Washington and Oregon to northern California do include stories about earthquake or earth-shaking “thunder gods”.
In particular, native people living on Vancouver Island have a story that historians think dates to about 1700, which describes the catastrophic destruction of a village on the west coast of Vancouver Island. Japanese government records mention the unexpected arrival of three-metre-high tsunamis on 27 January 1700, even though no tremors were felt nearby. By comparing the reported heights of tsunamis at different points along the Japanese coast, Kenji Satake of the Active Fault Research Center in Tsukuba, Japan, pointed to the Pacific north-west as the likely source of the waves, providing a good match to the native story.