Editor’s note: This article originally appeared as the cover story of the May 1994 Discover.
Like most people in Nicaragua, Chris Terry didn’t feel the mild earthquake that shook the country at about 8 p.m. on September 1, 1992. He didn’t notice anything out of the ordinary until some minutes later. Terry and his friend Scott Willson, both expatriate Americans, run a charter fishing business in San Juan del Sur, a sleepy village on Nicaragua's Pacific coast. On the evening of the earthquake they were aboard their boat in San Juan del Sur’s harbor.
”We were down below,” says Terry. “We heard a slam.” The sound came from the keel of their boat, which had just scraped bottom in a harbor normally more than 20 feet deep. Somehow the harbor had drained as abruptly as if someone had pulled a giant plug.
Terry and Willson didn’t have much time to contemplate the novelty of a waterless harbor. Within seconds they were lifted back up by a powerful wave. “Suddenly the boat whipped around very, very fast,” says Terry. “It was dark. We had no idea what had happened.”
The confusion was just beginning. As Willson and Terry struggled to their feet, the boat began dropping once again, this time into the trough of a large wave. Willson was the first to get out to the deck. There he found himself staring into the back side of a hill of water rushing toward the shore. “He was seeing the lights of the city through the water,” says Terry. “And then the swell hit, and the lights went out, and we could hear people screaming.”
One of those on the shore was Inez Ortega, the owner of a small beachfront restaurant. She hadn’t noticed the earthquake either. While preparing dinner she glanced out at the harbor and noticed that the water seemed unusually low. “I didn’t pay much attention at the time,” she says. But when she looked up again a swell of water at least five feet high was racing up the beach toward her restaurant.
“I started running, but I didn’t even get out of the restaurant when the wave hit,” she says. Ortega and several of her customers spent about half an hour swimming in a debris-filled stew before they managed to drag themselves out of the water.
Ortega and everyone else in San Juan del Sur looked about themselves in stunned silence. The waves had swept away restaurants and bars lining the beach, as well as homes and cars—and people—hundreds of yards inland. Terry and Willson managed to ride out the disaster on their boat. Still reeling, they witnessed the receding wake of the last wave.
“When the wave came back out, it was like being in a blender,” says Terry. Collapsed homes bobbed in the water around their boat.
Terry, Willson, and Ortega had survived a tsunami, a devastating wave triggered by an undersea earthquake. Although the waves that hit San Juan del Sur were extremely powerful, they rose only 5 to 6 feet high. Other parts of Nicaragua weren’t so lucky. All told, the offshore earthquake sent tsunamis crashing along a 200-mile stretch of the coast, and newspapers reported 65-foot waves in some places (though seismologists consider that figure unlikely; a more realistic wave height might be about 30 feet). The waves killed about 170 people, mostly children who were sleeping when the waves came. More than 13,000 Nicaraguans were left homeless.
Destructive tsunamis strike somewhere in the world an average of once a year. But the period from September 1992, the time of the Nicaraguan tsunami, through last July was unusually grim, with three major tsunamis. In December 1992 an earthquake off Flores Island in Indonesia hurled deadly waves against the shore, killing more than 1,000 people. Entire villages washed out to sea. And in July 1993 an earthquake in the Sea of Japan generated one of the largest tsunamis ever to hit Japan, with waves washing over areas 97 feet above sea level; 120 people drowned or were crushed to death.
In Japanese tsunami literally means “harbor wave.” In English the phenomenon is often called a tidal wave, but in truth tsunamis have nothing to do with the tame cycle of tides. While volcanic eruptions and undersea landslides can launch tsunamis, earthquakes are responsible for most of them. And most tsunami-spawning earthquakes occur around the Pacific rim in areas geologists call subduction zones, where the dense crust of the ocean floor dives beneath the edge of the lighter continental crust and sinks down into Earth’s mantle. The west coasts of North and South America and the coasts of Japan, East Asia, and many Pacific island chains border subduction zones. There is also a subduction zone in the Caribbean, and tsunamis have occurred there, but the Atlantic is seismically quiet compared with the restless Pacific.
More often than not, the ocean crust does not go gentle into that good mantle. As it descends, typically at a rate of a few inches a year, an oceanic plate can snag like a Velcro strip against the overlying continent. Strain builds, sometimes for centuries, until finally the plates spasmodically jerk free in an earthquake. As the two crustal plates lumber past each other into a new locked embrace, they sometimes permanently raise or lower parts of the seafloor above. A 1960 earthquake off Chile, for example, took only minutes to elevate a California-size chunk of real estate by about 30 feet. In some earthquakes, one stretch of the sea bottom may rise while an adjoining piece drops. Generally, only earthquakes that directly raise or lower the seafloor cause tsunamis. Along other types of faults—for example, the San Andreas, which runs under California and into the ocean—crustal plates don't move up and down but instead scrape horizontally past each other, usually without ruffling the ocean.
Seismologists believe the sudden change in the seafloor terrain is what triggers a tsunami. When the seafloor rapidly sinks—or jumps—during an earthquake, it lowers (or raises) an enormous mountain of water, stretching from the seafloor all the way to the surface. “Whatever happens on the seafloor is reflected on the surface,” says Eddie Bernard, an oceanographer with the National Oceanic and Atmospheric Administration (NOAA). “So if you imagine the kind of deformation where a portion of the ocean floor is uplifted and a portion subsides, then you’d have—on the ocean surface—a hump and a valley of water simultaneously, because the water follows the seafloor changes.”
One major difference between the seafloor and the ocean surface, however, is that when the seafloor shifts, it stays put, at least until the next earthquake. But the mound of water thrust above normal sea level quickly succumbs to the downward pull of gravity. The vast swell, which may cover up to 10,000 square miles depending on the area uplifted on the ocean floor, collapses. Then the water all around the sinking mound gets pushed up, just as a balloon bulges out around a point where it’s pressed. This alternating swell and collapse spreads out in concentric rings, like the ripples in a pond disturbed by a tossed stone.
Although you might think a tsunami spreading across the ocean would be about as inconspicuous as a tarantula walking on your pillow, the wave is, in fact, essentially invisible in deep ocean water. On the open sea, a tsunami might be only ten feet high, while its wavelength—the distance from one tsunami crest to another—can be up to 600 miles. The tsunami slopes very gently, becoming steeper only by an inch or so every mile. The waves so feared on land are at sea much flatter than the most innocuous bunny-run ski slope; they wouldn’t disturb a cruise ship’s shuffleboard game. Normal surface waves hide tsunamis. But that placid surface belies the power surging through the water. Unlike wind-driven waves, which wrinkle only the upper few feet of the ocean, a tsunami extends for thousands of fathoms, all the way to the ocean bottom.
Tsunamis and surface waves differ in another crucial respect: tsunamis can cross oceans, traveling for thousands of miles without dissipating, whereas normal waves run out of steam after a few miles at most. Tsunamis are so persistent that they can reverberate through an ocean for days, bouncing back and forth between continents. The 1960 Chilean earthquake created tsunamis that registered on tide gauges around the Pacific for more than a week.
”You’ve got to remember how much energy is involved here,” says Bernard. “Look at the size of these earthquakes. The generating mechanism is like a huge number of atomic bombs going off simultaneously, and a good portion of that energy is transferred into the water column.”
The reason for tsunamis’ remarkable endurance lies in their unusually long wavelengths—a reflection of the vast quantity of water set in motion. Normal surface waves typically crest every few feet and move up and down every few seconds. Spanning an ocean thus involves millions of wavelengths. In a tsunami, on the other hand, each watery surge and collapse occurs over perhaps 100 miles in a matter of minutes. For a large subduction-zone earthquake—magnitude 8 or more—the earthquake’s impulse can be powerful enough to send tsunamis traveling across the Pacific—from the Chilean coast to Japan, Australia, Alaska, and all the islands en route as well.
For much the same reason, tsunamis can race through the ocean at jetliner speeds—typically 500 miles an hour. To span a sea, they need travel a distance equal to just a few dozen of their own wavelengths, a few swells and collapses. That means the wave only has to rise and fall a handful of times before the surge reaches its destination. The outsize scale of a tsunami makes an ocean seem like a pond.
As a tsunami speeds on its covert way, undersea mountains and valleys may alter its course. During the 1992 Indonesian earthquake, villages on the south side of Babi Island were the hardest hit, even though the source of the tsunami was to the north of the island. Seismologists believe that the underwater terrain sluiced the tsunami around and back toward the island's south coast.
Only when a tsunami nears land does it reveal its true, terrible nature. When the wave reaches the shallow water above a continental shelf, friction with the shelf slows the front of the wave. As the tsunami approaches shore, the trailing waves in the train pile onto the waves in front of them, like a rug crumpled against a wall. The resulting wave may rear up to 30 feet before hitting the shore. Although greatly slowed, a tsunami still bursts onto land at freeway speeds, with enough momentum to flatten buildings and trees and to carry ships miles inland. For every five-foot stretch of coastline, a large tsunami can deliver more than 100,000 tons of water. Chances are if you are close enough to see a tsunami, you won't be able to outrun it.
As Inez Ortega and Chris Terry witnessed in San Juan del Sur, the first sign of a tsunami’s approach is often not an immense wave but the sudden emptying of a harbor. This strange phenomenon results from a tremendous magnification of normal wave motion. In most waves, the water within the crest is actually moving in a circular path; a wave is like a wheel rolling toward the shore, with only the top half of the wheel visible. When that wave is 100 miles long, the water in the crest moves in long, squashed ellipses rather than in circles. Near the front and bottom of the wave, water is actually on the part of the elliptical “wheel” moving backward—toward the wave and out to sea. If you’ve ever floated in front of a wave, you’ve probably felt the pull of the wave as water sloshes back toward the crest. With a tsunami, that seaward pull reaches out over tens of miles, sometimes with tragic results: when an earthquake and tsunami struck Lisbon in 1755, exposing the bottom of the city’s harbor, the bizarre sight drew curious crowds who drowned when the tsunami rushed in a few minutes later. Many people died in the same way when a tsunami hit Hawaii in 1946.
Although seismologists and oceanographers understand in broad terms how tsunamis form and speed across oceans, they are still grappling with some nagging fundamental questions. One of the major mysteries is why sometimes relatively small earthquakes generate outlandishly large waves. Such deceptive earthquakes can be particularly devastating because they may be ignored by civil agencies that are charged with issuing tsunami warnings.
The Nicaraguan earthquake is a case in point. By conventional measures, it shouldn’t have produced a tsunami at all. The earthquake registered magnitude 7.0 on the Richter scale, not puny by any means, but not large enough, seismologists believed, to pose much of a tsunami risk. The quake’s epicenter was 60 miles offshore, distant enough to dampen the tremors on land. Yet people who had not even felt the quake found themselves swept out to sea minutes later.
Hiroo Kanamori, a seismologist at Caltech, has made a point of studying the earthquakes that spring these unexpected tsunamis. Such earthquakes, he says, are responsible for some of the most damaging tsunamis on record. In 1896, for example, an earthquake in Japan was followed by a tsunami that drowned 22,000 people, even though survivors reported only mild shaking before the wave. And a relatively moderate 1946 quake in the Aleutian Islands sent a huge tsunami tearing across the north Pacific and into Hawaii, where it inundated much of the city of Hilo.
Twenty years ago Kanamori proposed an explanation for these surprise tsunamis. But to test his ideas he needed seismometers sensitive to a broad spectrum of ground movement, and the instruments of the 1970s just weren’t up to the job. Only in the past few years, in fact, have seismometers become sophisticated enough for his purposes. And the 1992 Nicaraguan tsunami proved an ideal test case.
Kanamori thinks some earthquakes may release their energy very slowly, over a minute or more, rather than in a brief, spastic lurch. This could happen, he says, if soft ocean sediments were sandwiched between two interlocked crustal plates. The lubricated plates would slide past each other smoothly, without sharp, building-shaking convulsions. “If you have two blocks of hard rock,” says Kanamori, “usually the friction between them is very high, so you can accumulate large amounts of stress. And when it slips, it slips very fast. But if you have lubricating materials in between, it can slip at relatively low stress, and when it slips, it goes slowly.”
The seismic energy from such a quake moves Earth’s surface in long, slow undulations. Humans don’t feel them. We notice only the shorter, sharper shivers from earthquakes, the type that rattle foundations. Nevertheless, the slow “tsunami earthquakes,” as Kanamori calls them, can be as quietly dangerous as a pristine hillside of snow on the verge of an avalanche. The Nicaraguan quake, he estimates, moved a 120-mile-long section of the seafloor more than three feet in about two minutes when part of the Cocos plate, a wedge-shaped slice of the Pacific ocean floor, slid under North America. This slow-motion shift, imperceptible to humans, sent the tsunami on its destructive way.
When Kanamori first suggested this model, seismometers weren’t able to reveal the true nature of these long-period vibrations. Part of the problem was that most seismometers depended ultimately on a pendulum to trace earthquake-generated movements onto a graph. The pendulum typically hangs by a spring from a supporting arm. Nestling right up against it is a recording drum—essentially a revolving scroll of paper—anchored firmly to the ground. During an earthquake, the recording drum bumps up and down. But because the pendulum is suspended in midair by a spring, its inertia makes it lag behind the shaking drum. In a sense, it floats freely, while the earth (and drum) jiggles underneath it. A pen attached to the pendulum scribbles a jagged portrait of the ground motion on the drum’s paper.
In these older seismometers, the long-period waves that Kanamori was looking for tended to get buried under the mountain of squiggles left by the far more numerous short-period waves. Newer instruments also rely on pendulums, but their coils are suspended in magnetic fields and attached to electronic recording systems rather than rolls of paper. When the seismometer shakes during an earthquake, the motion of the coil generates a current that can be analyzed digitally by a computer, revealing even subtle patterns such as the long-wavelength signatures typical of slow tsunami earthquakes.
The newer technology allowed Kanamori to study the deceptive Nicaraguan quake in all its complexity. Any earthquake, large or small, releases a symphony of shock waves. Some of these waves, particularly the jarring, high-frequency tremors that cause most of the damage near an earthquake’s epicenter, die out quickly and don’t travel far through the earth. In the Nicaraguan quake, these tremors never made it to the mainland with enough force to be felt on shore. Lower-frequency waves, however, resonate strongly in the planet’s interior, lingering like the deep boom from a bass drum in a concert hall. These tremors cyclically raise and lower the ground about every 20 seconds or more. Seismometers in Nicaragua and elsewhere in the world were able to pick up some of these signals, but not all of them.
As it turns out, the conventional measure of earthquake magnitude—the Richter scale—isn’t an accurate gauge of all an earthquake’s overtones. “In the 1930s,” explains Northwestern University seismologist Emile Okal, “when Charles Richter designed the scale, he had instruments that recorded 20-second-long seismic waves very well but that didn’t record waves with much longer periods.” As a consequence, standard seismic measuring devices may underestimate the size of an earthquake. The values on the Richter scale are derived primarily from a ratio of two numbers: the amount of ground-surface motion during an earthquake—as recorded on a seismometer, either electronically or on paper—and the number of vibrations per second. Simply put, lots of movement in a short time span means a big earthquake. Some earthquakes, however, like the one in Nicaragua, may release their energy very slowly, generating signals with periods longer than 20 seconds.
Last year, in the British scientific journal Nature, Kanamori argued that the magnitude 7.0 assigned to the Nicaraguan quake was far too low. By including the longer, slower seismic signals in his calculations, Kanamori upped the quake's magnitude to 7.6. Since an increase of one on the scale corresponds to a tenfold increase in the size of an earthquake, the Nicaraguan quake, by Kanamori's calculations, was five times bigger than originally estimated.
Emile Okal often does research at a French seismological station in Tahiti, one of a handful of stations in the world equipped with the technology to detect the longer signals of slow earthquakes. The Tahiti station, says Okal, received signals from the Nicaraguan quake even though it was more than 4,000 miles to the northeast, and immediately recognized the potential for disaster.
“The only problem,” says Okal, “was that by the time we picked up the signal, the Nicaraguan coast had been totally ravaged.” It took about 25 minutes for the signal from the quake to travel through Earth’s crust to Tahiti, Okal explains. And although seismic waves travel 15 to 20 times faster than tsunamis, this tsunami had to cover only 40 miles before hitting the Nicaraguan coast.
Unfortunately, even the network of seismological stations Nicaragua did have had greatly deteriorated during the country’s civil war. Had more modern seismometers been available, and had researchers been aware of the importance of the long-period seismic waves, says Kanamori, there might have been enough warning time to tell people to run for high ground. Kanamori hopes that more countries—especially those on the earthquake-prone Pacific rim—will invest in technology capable of detecting a wider range of earthquake tremors. “No matter what we do, tsunamis are going to happen,” he says. “The question is whether we can have a very effective tsunami warning system.”
The United States has two tsunami warning centers, one near Honolulu and another in Palmer, Alaska, just north of Anchorage. Both were built in the 1960s after tsunamis from two large earthquakes, one in Alaska in 1964 and the other in Chile in 1960, caused millions of dollars’ worth of damage in the two states. The tsunami from the 1964 earthquake also hit Crescent City in northern California, killing 11. The warning centers, staffed around the clock, collect data via satellite from dozens of seismological stations in more than 20 countries bordering the Pacific.
For earthquakes above about magnitude 6.5, says Michael Blackford, a seismologist at the Hawaii warning center, the center alerts the warning systems in other countries around the Pacific. If the earthquake is far away—say in Chile, Japan, or Alaska—the seismic signal, pulsing rapidly through Earth’s crust, will arrive in Hawaii hours ahead of any potential tsunami, giving the center time to alert the state civil defense via a hot line.
In addition to seismographic data, the warning systems also receive readings from tide gauges scattered in harbors in Alaska, Hawaii, and Pacific-rim countries: if, accompanying an earthquake, a harbor’s water level drops a few feet in a few minutes, the quake may have triggered a tsunami. While that information would be too late to help local residents, more distant communities could be forewarned.
Even with seismographic and tide-gauge data, tsunami prediction is a hit-or-miss proposition. False alarms outnumber real tsunamis by more than two to one, and this can cause problems besides simply jading the public. In 1986, after a magnitude 7.7 earthquake in Alaska, the Hawaii center issued a tsunami warning. Television, radio, and even air-raid sirens were used to get people to evacuate coastal areas. The exercise cost some $30 million. Yet the tsunami that arrived was barely three feet high.
“I wasn’t here at the time,” says Blackford, “but I got a lot of feedback from people who were. It’s well remembered. The tsunami arrival time was to be in the late afternoon, so the civil defense just blew the sirens. The word was, everybody should evacuate from the coastal areas. They closed offices and turned everybody loose. This resulted in virtual gridlock on the roads in Oahu. Some of the roads are right next to the sea. People have told me about this chaotic situation: you’re sitting in your car with the ocean lapping alongside you there, and you're wondering, ‘Why am I here? I’m supposed to be away from the ocean and I’m stuck in this traffic jam.’ ”
Part of the reason it’s so hard to issue reliable warnings lies in the inherent difficulty of detecting a tsunami barreling invisibly across the ocean. That’s a problem now being addressed by Eddie Bernard and Frank González, oceanographers with NOAA’s Pacific Marine Environmental Laboratory in Seattle. For the past few years González and Bernard have been working on developing a deep-sea tsunami sensor. Six of their devices now rest on the ocean floor, one 300 miles directly west of the Oregon-Washington border, one 30 miles southwest of Hawaii, and four more spread out a few hundred miles south of the Aleutians. The instruments are remarkably sensitive. “In 12,000 or 15,000 feet of water they are capable of sensing a change in sea level of less than a millimeter,” González says.
These sensors measure the weight of the water column above them. When a wave passes across the surface, it increases the height and therefore the weight of the water column above the sensor. As a trough comes by, the height and weight of the water column decrease. The sensor consists of a small metal tube about four inches long, which floats just above the ocean bottom; it is held in place by an anchor (usually an old iron railroad wheel). Partially enclosed within the metal tube is a small device—called a Bourdon tube—shaped like a comma with a very long tail. The end of the tail sticks out of the bottom of the metal tube, exposed to the ocean, and is open like a straw. The other end of the Bourdon tube is closed. When a wave passes overhead, increasing the weight of the water column, the pressure slightly straightens the Bourdon tube in the same way a paper noisemaker unfurls when you blow into it. When a trough comes by, the tube curls up again. As the Bourdon tube alternately straightens and curls, it pushes and pulls on a sensitive quartz crystal, which in turn produces an electric signal that varies along with the changes in pressure of the water above.
The pressure sensors remain on the bottom for 12 months at a time, storing data electronically every 15 seconds. To recover the sensors—and their data—González and Bernard need to send a ship to the ocean site above the sensor. When the ship reaches the site, it broadcasts a signal telling the sensor to release its anchor. The entire package (which includes an orange marker buoy and a signal transmitter) then floats to the surface.
Tsunamis are easy to spot in the sensors’ records, says González. “Tsunamis typically have periods of anywhere from 3 to 30 minutes or so,” he says. “On the other hand, tides and signals generated by storms have periods on the order of hours and tens of hours.” (Passing waves or ships generate pressure changes too small to reach the ocean floor.)
For now the sensors aren’t tied into any warning system. But with very minor changes, says González, they could be. Instead of being picked up periodically by a ship, they would have to transmit data every few seconds to a receiver on a surface buoy. If a tsunami rolled by, the buoy would relay the steady, maybe half-hour-long increase in pressure to a satellite or directly to a land station, where researchers could quickly work out the wave’s size and heading. Sensors between Hawaii and Alaska could warn Hawaiians several hours in advance of an Alaska-born tsunami heading their way.
“We have experience in all of these components,” says González. “It's just a matter of putting them together to get them to do what we want in this specific application.”
NOAA has proposed setting up seven such stations, four off Alaska and three off the northwest coast. Altogether the stations would cost about $700,000 to install and another $250,000 a year to maintain, according to Bernard.
One of the reasons Bernard is eager to push ahead with a warning system is that he worries a tsunami could catch the West Coast by surprise. Hidden beneath the waves about 50 miles off the northwest coast is the Cascadia subduction zone, where the Pacific floor plunges under North America. The zone stretches from Vancouver Island to northern California, just a few hundred miles northwest of San Francisco. Many seismologists fear that it's just a matter of time before an earthquake convulses the area.
Although no major earthquake or tsunami has battered the Northwest in historical times, some geologists believe they have found evidence of past tsunamis in the region. Sand deposits resembling the debris left by modern tsunamis have been found in a number of places along the northwest coast and appear to have been laid down about 300 years ago, before Europeans settled there. Native Americans in the area, moreover, have legends of great floods from the sea that sound eerily like tsunamis.
At a workshop in Sacramento last spring, Bernard met with a number of seismologists to discuss the risk of a Cascadia zone earthquake and tsunami. “We all talked for an entire day about what is the most probable earthquake, not the worst case, but what is the most probable earthquake you could expect from this area. From those discussions came a scenario earthquake of a magnitude of about 8.4, and its fault dimensions are about 140 miles long by about 50 miles wide. Using this as a basis, we set out to model what tsunami could result from that size earthquake.”
Based on case studies of tsunamis generated by earthquakes of a similar magnitude, Bernard and his colleagues estimate that the earthquake would spawn a tsunami that might be 30 feet high when it hit the coast. A tsunami that size would threaten coastal cities in the Northwest, northern California, and Hawaii. But Bernard cautions that there is no way of predicting whether such a disaster will happen next year or in 300 years.
Does this mean that a tsunami could come rolling through the Golden Gate, drowning Alcatraz, Sausalito, and Fisherman’s Wharf? Probably not. Most seismologists speculate that the headlands outside the Golden Gate would take the brunt of a tsunami’s wrath. Others admit the very remote possibility that a tsunami could funnel right into San Francisco Bay. No one really knows.
“I want you to understand that this is a fairly speculative business. But we do have a public safety issue at hand. We have to play the ‘what if’ game,” says Bernard. “Unlike Alaska and Hawaii, where we have warning systems in place to respond in five minutes, we have no such facility on the West Coast.”
It’s been 30 years since the United States last suffered through a major tsunami, and some seismologists think the lull has fostered a dangerous, false sense of security. Another tsunami is inevitable, seismologists say. And vulnerable coastal areas, like much of Hawaii—where development has skyrocketed since the last tsunami—may be especially hard hit.
Like others, Bernard hopes that NOAA’s proposed warning system will be in place before the next big tsunami breaks on Pacific shores, but he’s not optimistic. “You see, there’s a trend here,” he says. “We always build a warning center after a big tsunami. We built one in Hawaii after the 1964 event. After, I want to underline the word after. The question I have is, Are we going to wait until after the Cascadia subduction zone earthquake to build one in California and Oregon, or are we going to do it in advance?”