Analog SFF, October 2005

by Dell Magazine Authors

  Nobody had ever before seen a tsunami in the open ocean, and if you'd asked only a few years ago, you'd almost certainly have been told it was impossible. That's because, in deep water, a tsunami is nothing but a low, fast-moving swell. It only becomes dangerous when it hits shallow water and humps up, like breaking surf. But one of the satellites carried a laser altimeter, which easily spotted the deadly waves sloshing around the Indian Ocean.

  Laser altimetry is an extremely precise technology capable of mapping minor variations in sea-surface height with remarkable precision.[4] The satellite saw the tsunami as a fifty-centimeter crest followed by a forty-centimeter trough, with a wavelength of 430 kilometers, explaining why you can't possibly see it from a boat! It also revealed that rather than being a single big wave, the tsunami was an initial big one followed by hours of choppy sloshing.[5]

  [4. See Richard A. Lovett, “The Wired Ocean: Doing Oceanography Without Getting All Wet,” Analog, October 2005.]

  [5. News reports showed the same thing, as the coastlines were hammered by wave]

  Another satellite managed to photograph the waves hitting shore. Again, purely by coincidence, it was passing over the shoreline of India and Sri Lanka about four hours after the earthquake, when the wave action was at its highest. Because the satellite had nine cameras lined up along its flight path—ahead, below, and behind—it could combine shots of each location into sequences that produced crude movies. These showed that the waves struck shore at thirty-eight miles per hour, with a frequency of about 4.8 per hour (about one every 12.5 minutes).

  More importantly, the images showed precisely how the waves interacted with the coastline. Details are still being figured out, but one likely lesson is that waves slam with devastating force into broad, wall-like objects, while diverting around narrow ones with much less impact. At a small scale, this is obvious: a wave that will knock down a house won't affect a telephone pole. (One implication is that in tsunami zones, buildings should be put on stilts, perhaps with breakaway walls on the bottom floor.) As the satellite images are processed, we'll probably also learn useful things about the effects of headlands, harbors, and islands.

  The same effect also works on a larger scale, where it spells the difference between what happened to India and Sri Lanka, compared to the island of Diego Garcia.

  Topographically, the coasts of India and Sri Lanka were the equivalent of a wall that took the full brunt of the wave's force. Diego Garcia by contrast, rises from the ocean bottom like a tall, narrow pole. In Sri Lanka, the tsunami generated waves up to ten meters high; in Diego Garcia, the maximum height was about 2.5 meters. Basically, the tsunami ignored the island and went around it.

  * * * *

  Listening for the Wave

  Diego Garcia is only a tiny speck of land, but its location near the middle of the Indian Ocean, south of India and Sri Lanka, makes it strategically valuable as a listening post for the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO), which is charged with monitoring for illegal bomb tests.[6] It turns out that instruments designed to detect nuclear bombs can also detect lots of other things—including tsunamis.[7]

  [6. It's also a military base.]

  [7. See Richard A. Lovett, “Forensic Seismology, The Big Science of Minor Shakeups,” Analog, April 2004.]

  Two of these instruments are infrasound detectors and hydrophones. Infrasound is extremely low-frequency sound: not the bass thump of a big audio speaker, but lower yet, where frequencies are measured not in cycles per second, but seconds (or longer) per cycle. These super-low sound waves carry extremely well, refracting through the upper atmosphere to tracking stations where they are detected by microphone-like pressure sensors so sensitive that they could sense the air-pressure difference between the top and bottom of a sheet of paper.

  Hydrophones are simply underwater microphones. Like infrasound detectors, they are laid out in arrays that can identify the direction from which the sound is coming; if a signal is received by multiple arrays, it's possible to triangulate on its source.

  The Diego Garcia hydrophones are moored in deep water, at a depth of 1,200 meters. That puts them in what is known as the T-zone, a layer bounded above and below by discontinuities at which the water changes density. These boundary layers reflect low-frequency sound, channeling it for long distances in a waveguide effect similar to that by which light propagates through fiber-optic cables. Whales use the T-zone for long-distance communication.

  The infrasound detectors and hydrophones both detected the earthquake. The infrasound sensors probably heard the sounding-board effect of seawater being bounced up and down, while the hydrophones heard the earthquake more directly, as seismic waves found their way into the T-zone.[8]

  [8. It is possible that the infrasound may also have been created by vibration of nearby landmasses.]

  Infrasound is somewhat blurry, due to the fact that there are multiple paths the waves can take through the upper atmosphere. Hydroacoustic images are sharper. In addition to being heard in Diego Garcia, they were heard by receivers near Crozet Island (in the southern Indian Ocean) and Cape Leeuwin, Australia. This allowed triangulation on the source of the signal and gave enough detail that, if the receivers had all been linked to a central computer, it would have been possible to track the earthquake as it occurred.

  It is widely believed that when faults slip, they do so simultaneously at all points. Actually, earthquakes originate at their epicenters and “unzip” their faults at finite speed.

  In the case of the Boxing Day quake, Catherine de Groot-Hedlin, of Scripps Institution of Oceanography, reported that the rupture progressed northward, moving quickly for 600 kilometers, then slowing for another 550. “Slow,” of course, is a relative term. Initially, the fault unzipped at a speed of 2.4 km/sec (about 5,400 mph); by the end, it was “only” going 1.5 km/sec (about 3,500 mph).[9]

  [9. The March 28 earthquake progressed at a steady speed of about 2.5 km/sec, but only ruptured 340 km of the fault.]

  This change in speed was borne out by seismic observations and suggested that in some important way, the Boxing Day quake went through two phases. But what this means appears to be anybody's guess. Perhaps we'll eventually discover that it was, effectively, two mega-quakes rolled into one. Perhaps multiple phases are a common feature of mega-quakes. With only one well-studied magnitude-9 available, it's difficult to generalize.

  For that matter, the speed change may not even have occurred. A study reported at an October 2005 meeting of the Geological Society of America combined data from 700 seismological stations in Japan in an effort to map the quake's progress with previously impossible levels of precision. It got a very nice map of the manner in which the fault unzipped: and totally failed to find any appreciable slowdown.

  Infrasound and hydroacoustics are interesting, but not worth much as earthquake warning systems because they're limited by the speed of sound in air and water. Because seismic waves travel much more quickly through rock, seismometers will always beat them to the punch. What infrasound and hydroacoustics can do, however, is detect tsunamis in progress.

  In the case of infrasound, the sensors heard the deeply subsonic noise of big waves sweeping around the Bay of Bengal. By then, of course, tide gauges were already registering the signal and newscasters were sending frantic broadcasts, so infrasound still can't provide a useful warning. But hydrophones are different. Positioned between the approaching tsunami and the shore, they can recognize a tsunami passing overhead, even in open water, where you'd expect it to be impossible.

  What the Diego Garcia hydrophones spotted were minute changes in the pressure of the 1,200 meters of overlying water as the tsunami rolled overhead. It was a surprising finding, because hydrophones designed to look for nuclear blasts aren't tuned for the super-low-frequency oscillations of tsunamis (which, as we noted before, are on the order of 4.8 oscillations per hour). But each crest of the tsunami, like a photon, is actually a wave “packet” containing a wide rang
e of frequencies. This allowed the hydrophones to catch what, to them, were extremely low-frequency pressure changes—on the order of half a minute per cycle.

  "In the tsunami community, we call that high-frequency,” says Emile Okal, a geology professor from Northwestern University who was one of many scientists thrilled by the finding. Now that we realize that these arrays can spot tsunamis, he adds, upgrading them to look for even lower tsunami-wave frequencies should be easy.

  Diego Garcia has two sets of hydrophone arrays. One was on the wrong side of the island, but the other was moored about 100 km offshore, more or less in the direction from which the wave arrived. Had it been programmed to sound an alert, it would have been able do so about five minutes before the first wave hit the island. And while five minutes’ notice might not sound like much, it's a lot if the meaning is “run for your life."

  Currently, there are only eleven of these extremely low-frequency sensor arrays in the world, and none are programmed to detect tsunamis. Instead the data go to CTBTO headquarters in Vienna, where computers identify it as not-an-atom-blast and dismiss it as noise. But that too is an easy fix, and there's no reason the agency's computers can't be programmed to sound an alert for tsunamis as well as for nuclear tests. Perhaps someday, hydrophones might lie offshore from all vulnerable shores, waiting to sound automated alarms.

  * * * *

  Saturated Models

  Unfortunately, nobody managed to give a tsunami warning to Thailand, India, Sri Lanka, and other countries in the path of the devastation. In large part, that was because there was no warning system because prior to December 26, nobody thought the Indian Ocean was vulnerable to large tsunamis. But it was also because seismologists, to their chagrin, discovered that it took far too long to realize just how big the earthquake had been.

  The quake began at 12:59 A.M., Greenwich time. Eight minutes later, seismic stations in Australia picked it up and triggered a local alarm. Shortly after that, the Pacific Tsunami Warning Center, in Hawaii, calculated the magnitude at 8.0 and (correctly) concluded that the location precluded a tsunami threat in the Pacific. It took another hour for seismologists to revise their magnitude estimate upward to 8.5, and several more hours before a team at Harvard re-estimated it as 8.9. A month later, the Harvard team upped its estimate to 9.0, and four months after that, geophysicists raised their estimates yet again, into the range of 9.2 to 9.3.

  Thanks to the logarithmic nature of the magnitude scale, the differences between 8.0, 8.5, 8.9, and 9.2 are huge—and by the time the estimate had been increased from 8.5 to 8.9, it was far too late.

  This does not mean seismologists were negligent. Rather, they discovered that much of what they thought they knew wasn't right.

  To understand the difficulty, it's necessary to look more closely at how earthquake magnitudes are calculated. It is frequently said that magnitude measures energy. This is true, but what it really measures is “moment,” which expresses how big a chunk of the Earth's crust shifted, and by how much. (It also takes into account the strength of the rock.) A magnitude 5.0 earthquake, for example, will move a chunk of land the size of New York's Central Park by a fraction of a meter. A magnitude 8.0 would shift a fifty-kilometer by 100-kilometer block by about five meters. The Boxing Day quake did the equivalent of moving the state of California by about ten meters.

  One way to measure that is with surveying teams and a site visit. That, of course, takes a while and isn't feasible in the quake's immediate aftermath.

  Seismology works by correlating seismic waves to moment. Fortunately for humanity, the Earth doesn't see a lot of enormous earthquakes. Unfortunately for seismologists, this means their models for big earthquakes are based on extrapolation, rather than real data. The first test of these extrapolations was the Boxing Day quake, and what it proved was that they didn't work. “Earthquakes,” Okal says wryly, “do not read textbooks in seismology."

  One problem is that nobody realized that big earthquakes could last as long as the Boxing Day quake did or affect such large segments of their faults. The 1994 Northridge quake in Southern California, magnitude 6.9, lasted a mere eight seconds. The 2002 Denali quake, magnitude 7.9, shook the ground for two minutes and covered 200 miles. The Boxing Day quake lasted ten minutes and unzipped 700 miles of its fault.

  The best comparison might be between tapping the surface of a swimming pool with your fingertip and sloshing the water back and forth with a 2 x 4. In one case, you'll create a lot of little, rapidly oscillating waves. In the other, you'll also get big, lower-frequency ones—and the bulk of the energy will be in them.

  That was the problem seismologists faced on Boxing Day. Because it lasted ten minutes, the December 26 temblor put out seismic waves with vibrational frequencies as low as six cycles per second. But the computer models didn't believe such things existed, and didn't look for them—with the result that the models were found to “saturate” for large earthquakes, making it difficult to distinguish big ones from colossal ones. And even if the models had known what to look for, it's still going to take several tens of minutes to wait through enough of these extremely long cycles to get an accurate estimate of how strong they are. By then, the tsunami has covered hundreds of miles.

  So how can you get a decent magnitude estimate without having to wait for the seismometers? Jeff Freymueller of the Geophysical Institute at the University of Alaska thinks the answer lies in GPS.

  Hand-held GPS devices of the type used by hikers and boaters employ a network of satellites to pin down location to within a meter. Surface-mounted units can do orders of magnitude better, quickly enough that if the data is uplinked to the Internet, you can watch an earthquake as it occurs.

  At the Spring 2005 meeting of the American Geophysical Union, Freymueller proved this is possible by using data from GPS units on landmasses within the earthquake zone, even if the GPS stations were many miles from the fault.[10] If enough such units had been mounted throughout Indonesia, he said, it would have been possible to map out land displacements accurately enough to obtain a decent estimate of the earthquake's magnitude within minutes of its onset. Freymueller therefore advocates installing GPS units 100 to 200 kilometers apart near all of the Earth's major faults.[11]

  [10. Earthquakes shift entire blocks of land and transmit that force around the globe. So you don't have to be right next to the fault to get useful GPS data, so long as you know how far from the fault your GPS units actually lie.]

  [11. Three hundred such units already exist in Southern California, installed in the aftermath of the 1994 Northridge Earthquake. Not only can they precisely measure displacements caused by earthquakes, but they reveal the gradual creep of rocks on each side of faults, as they build up the strains that might lead to future tem]

  Big earthquakes, though, don't always generate tsunamis. It appears, for example, that shallow earthquakes are more likely to produce them than are deep ones, because shallow quakes produce greater up-and-down motion: and it is this, rather than side-to side-motion, which appears to be the main culprit. On Boxing Day, for example, the offshore seabed rose, while the near-shore seabed dropped, producing a catastrophic slosh. On March 28, only a relatively small tsunami was produced (small enough that nobody was killed). Partly, that may have been because it was a deeper earthquake, but more importantly, it appears that much of the motion occurred beneath a chain of offshore islands. “Instead of displacing water,” Okal says, “it displaced islands. And every time you have an island, this is that many cubic meters of water that are not displaced."

  The length and orientation of the fault are also important. Big undersea quakes are like guns, aiming energy in specific directions, says De Groot-Hedlin. The Boxing Day quake sent big waves east and west. To the extent that the March 28 one generated a tsunami, it beamed most of its energy to the southwest, where it dissipated in the open ocean.

  * * * *

  The Earthquake that Wasn't ... and One Yet to Come

  Indonesia was unfor
tunate enough to experience the two largest earthquakes in forty-plus years. But there was also a monster that didn't occur.

  To understand, we must begin with a primer on the type of earthquakes that rocked the Indonesian shoreline. Called subduction quakes, they are the result of plate tectonics, which causes crustal plates to bash into each other in long, slow collisions—one of which is happening offshore from Indonesia. The site of the collision is marked by a trench, where the dense seabed rocks are forced downward, preparatory to dipping beneath the lighter rocks of the Indonesian archipelago.

  Between quakes, pressure builds along the boundary, causing the top plate (the land) to bow upward.[12] Then the quake occurs, and the land not only springs forward, but drops. Conversely, the seabed, which has been bowing downward, springs the opposite direction. It's like stepping on one end of a springy ruler to hold it in place, while pushing on the other. The “earthquake” occurs when you lift your foot.

  [12. One of the uses of GPS is to measure these bulges as markers of accumulating strain. But old-fashioned methods also work. In Indonesia, it had long been known that the shallow near-shore waters (which are part of the island's plate, not the seabed's) were rising. This was evident from the fact that coral reefs had been steadily pushed above sea level, killing the coral. By determining the date at which each stratum of coral died, it was possible to chart the rate of uplift. Ironically, in early December 2004, less than three weeks before Boxing Day, a group of geophysicists unveiled a brochure they had been preparing for distribution in Indonesia, explaining what this meant and outlining the earthquake risk.]

  In the big picture of plate tectonics, all sections of a plate are trying to move in the same direction.[13] But slippage isn't always uniform along the rupture zone. In part, that's why big temblors generate aftershocks, as various portions of the rupture zone settle into new equilibrium. On Boxing Day alone, there were 142 aftershocks of magnitude 5.0 or larger.