Jin Sato is the mayor of a town that no longer exists.
Minamisanriku, a quiet fishing port north of Sendai in northeastern Japan, disappeared last March 11. Sato nearly did too. The disaster started at 2:46 p.m., about 80 miles east in the Pacific, along a fault buried deep under the seafloor. A 280-mile-long block of Earth's crust suddenly lurched to the east, parts of it by nearly 80 feet. Sato had just wrapped up a meeting at the town hall. "We were talking about the town's tsunami defenses," he says. Another earthquake had jolted the region two days earlier—a precursor, scientists now realize, to the March 11 temblor, which has turned out to be the largest in Japan's history.
When the ground finally stopped heaving, after five excruciating minutes, Minamisanriku was still mostly intact. But the sea had just begun to heave. Sato and a few dozen others ran next door to the town's three-story disaster-readiness center. Miki Endo, a 24-year-old woman working on the second floor, started broadcasting a warning over the town's loudspeakers: "Please head to higher ground!" Sato and most of his group headed up to the roof. From there they watched the tsunami pour over the town's 18-foot-high seawall. They listened to it crush or sweep away everything in its path. Wood-frame houses snapped; steel girders groaned. Then dark gray water surged over the top of their building. Endo's broadcasts abruptly stopped.
Some 16,000 people died that day, most of them along hundreds of miles of coast in the Tohoku region, and nearly 4,000 are still missing. The tsunami eradicated several towns and villages in Tohoku and left hundreds of thousands homeless. In Minamisanriku the killed or missing number about 900 of 17,700 inhabitants, including Miki Endo, whose body was not found until April 23. Sato survived by climbing a radio antenna on the roof and clinging to it. "I think I was underwater for three or four minutes," he says. "It's hard to say." Many of the 30 or so other people on the roof tried to hang on to the iron railings at its edge. The waves kept coming all night long, and for the first few hours they repeatedly inundated the three-story building. In the morning only ten people remained on the roof.
Japan leads the world in preparing for earthquakes and tsunamis. It has spent billions retrofitting old buildings and equipping new ones with shock absorbers. High seawalls shield many coastal towns, and well-marked tsunami evacuation routes lead to high ground or to tall, strong buildings. On March 11 government seismologists had barely stopped hugging their computer monitors to keep them from crashing to the floor when their first tsunami warning went out.
Together these measures saved many thousands of lives; Miki Endo alone may have saved thousands. The Tohoku earthquake itself—a magnitude 9—did much less damage than it would have in other countries. But between 16,000 and 20,000 died because of the tsunami—a death toll comparable to that caused by an earthquake and tsunami in the same region in 1896.
Japan's defenses have improved tremendously since then, but its population has tripled. Its coasts are far more crowded. The same is true all over the world, in countries that are much less prepared. In the Indian Ocean, where the deadliest tsunami in history killed nearly 230,000 people in 2004, most of them in Indonesia, a similar disaster has been forecast for sometime within the next 30 years. In the United States, where a tsunami devastated the Pacific Northwest 300 years ago, when it was sparsely inhabited, geologists say another is inevitable. It's likely there will be many Minamisanrikus in the decades ahead.
Sato had survived a big tsunami before. In 1960, when he was eight, a 14-foot wave killed 41 people in Minamisanriku. The seawall was built after that, to a height of 5.5 meters, a little over 18 feet. "We thought we would be safe," Sato says. "Seismologists had told us to prepare for a tsunami that might be five and a half to six meters high. But this one was three times that height." Afterward, in the landscape of debris that had been his town, almost the only thing that remained intact was the seawall.
Tsunamis strike somewhere in the world almost every year, and giant ones have arguably changed history. Some archaeologists have argued, for instance, that a Mediterranean tsunami struck the north shore of Crete a bit over 3,500 years ago; the disaster, they say, sent Minoan civilization, one of the most sophisticated of the age, into a tailspin, leading it to succumb to Mycenaean Greeks. In 1755, when an earthquake and tsunami killed tens of thousands in Lisbon, the tragedy had a lasting impact on Western thought: It helped demolish the complacent optimism of the day. In Voltaire's novel Candide the blinkered philosopher Pangloss arrives in Lisbon during the catastrophe, persists in arguing that "all is for the best in the best of all possible worlds," and gets hanged for his trouble. Voltaire's withering satire made it a little harder to be Panglossian—to believe that a benevolent God designed an optimal Earth.
In the fifth century B.C. the Greek historian Thucydides was the first person to document the connection between earthquakes and tsunamis. He noticed that the first sign of a tsunami is often the abrupt draining of a harbor, as the sea pulls away from the coast. "Without an earthquake I do not see how such things could happen," he wrote. Actually they can. The Minoan tsunami was triggered by the cataclysmic eruption of Thira, a volcanic island 70 miles north of Crete in the Aegean. And landslides can cause local tsunamis, such as the one that surged 1,700 feet up a hillside in Lituya Bay, Alaska, in 1958 (see photo). All it takes is a large mass of rock moving abruptly in a large mass of water—not necessarily the ocean.
The vast majority of tsunamis, however, including the Tohoku one, are caused by seafloor earthquakes along faults called subduction zones. Most are in the Pacific and Indian Oceans. Along those boundaries two of Earth's tectonic plates collide, and the one carrying dense oceanic crust dives under the more buoyant continental one, forming a deep-ocean trench. Normally this happens smoothly, at a rate of a few inches a year. But at some times and places the plates become stuck—the peak of a subducting seamount might snag on the bottom of a continent, for example. After centuries the accumulated strain overwhelms the friction, and the plates shudder past each other. Off Japan last March the quake began miles below the seafloor and then spread up the sloping contact between the plates to the Japan Trench at the seafloor. It released the energy equivalent of 8,000 Hiroshima bombs. A sizable fraction of that went into motion of the seafloor, which raised and lowered the water above it—thus creating a tsunami.
Ordinary ocean waves are mere wind-driven wrinkles in the sea surface, but a tsunami moves the entire water column, from the seafloor up. The initial disturbance spreads out in opposite directions from the fault, in long wave fronts that may be a few hundred miles apart. In deep water offshore they're barely noticeable. They grow to dangerous heights only in shallow water, as they pile up against a coast—and they can remain dangerous even after they've crossed a whole ocean, barreling at the speed of a jetliner. The tsunami that savaged Japan last March swept a man in California out to sea; it broke Manhattan-size blocks of ice off the frozen margins of Antarctica. The tsunami that took 41 lives in Minamisanriku in 1960 was triggered by a magnitude 9.5 earthquake off Chile, the largest quake on record.
The Indonesian tsunami of December 26, 2004, killed people all around the Indian Ocean. It began off the northwest coast of Sumatra with a sudden, thousand-mile-long rupture—and magnitude 9.1 quake—on the Sunda megathrust, a fault along which part of the Indian Ocean floor subducts under Indonesia. Indonesia suffered more than any other country, with nearly 170,000 dead—more than half of them in Banda Aceh, the capital of the north Sumatran province of Aceh. But some 60,000 more died in Sri Lanka, in India, and in other countries around the basin, as far away as Africa.
In the wake of that unprecedented disaster several countries worked together to expand the use of a tsunami-detecting system that had been developed in the United States by the National Oceanic and Atmospheric Administration (NOAA). The system consists of an instrument anchored to the seafloor—called a tsunameter—that measures pressure changes caused by a passing tsunami. The tsunameter sends a signal to a surface buoy, which relays the data to a satellite, which broadcasts the information to warning centers around the world.
By 2004 only six such detectors had been deployed, all in the Pacific. There were none in the Indian Ocean, and in any event many countries in the region had no national warning centers that could have alerted local communities. That policy blunder had tragic consequences. In Sumatra people had only a few minutes to run, but the tsunami took two hours to reach India, and some 16,000 people died there. "It was totally unnecessary," says Paramesh Banerjee, a geophysicist at Nanyang Technological University in Singapore. "Technically it would have been relatively easy to install a tsunami warning system for the Indian Ocean."
There are now 53 detector buoys operating in the world's oceans, including 6 of a planned 27 in the Indian Ocean. So a repetition of the 2004 horror, in which the tsunami traveled for hours and still caught people by surprise, is less likely. But buoys would not have helped in Sumatra. People living on coasts near a rupturing fault can't wait for confirmation that a tsunami is on its way, which it often isn't; they must flee as soon as the quake hits. The Japanese warning system relies not only on tsunameters but also on seismometers—a thousand of them blanket the country, the densest network anywhere—combined with a computer model that forecasts the scale of a tsunami from the magnitude and location of the quake.
In March the system, which is run by the Japan Meteorological Agency (JMA), did not work perfectly. JMA's crucial first estimate, while the ground was still shaking, put the quake magnitude at 7.9—whereas later analysis revealed a quake that, at magnitude 9, was 12 times larger. The tsunami forecast warned of waves of ten feet or more—whereas they reached 50 feet in Minamisanriku and in some places perhaps even higher. But the human response to the warning was imperfect as well. "I think this time many people who lived above the high-water mark of the 1960 tsunami didn't bother to run," Jin Sato says. "Many of them died." The town's seawall, he thinks, also gave people a false sense of security.
The size of the earthquake and tsunami shocked seismologists. The Indonesian quake had ruptured a thousand miles of fault, the Tohoku quake only 280 miles—and yet the latter produced a magnitude 9 quake. Most geologists didn't think the Japan Trench could do that, even with a longer rupture. The oceanic crust there is old, cold, and dense, and scientists reasoned it would sink beneath Japan too readily and with too little friction to generate such a big quake.
Yet there was evidence that such a quake was possible. More than a decade ago scientists from Tohoku University, in Sendai, dug into the black mud around their coastal city and discovered three separate layers of sand that extended almost three miles inland. Abundant marine plankton in the sand layers showed they had been deposited by giant tsunamis at intervals of 800 to 1,100 years over the previous 3,000 years. The researchers' paper was published in 2001 in the Japanese Journal of Natural Disaster Science. It concluded with a warning: Because the last tsunami had struck Sendai more than 1,100 years earlier, the risk of another soon was very high. But to Japanese policymakers the uncertainty in that forecast seemed high too. When the tsunami came last March, it deposited another layer of sand at least two and a half miles inland.
"I think all subduction zones are guilty until proven otherwise," says Kerry Sieh. Sieh, director of the Earth Observatory at Singapore's Nanyang Technological University, is one of the world's leading paleoseismologists—he plumbs the geologic record for evidence of ancient earthquakes and tsunamis. He's a delicately built man of 61, with graying, neatly trimmed hair. The historical record—and especially the modern instrumental record—is too short, he says. It absolves long-dormant faults around the world that very likely could generate killer tsunamis. "We must assume every long subduction zone is capable of producing great earthquakes and tsunamis," Sieh says. "We can't assume that any megathrust is gradually and harmlessly releasing strain."
Sieh pulls up a map on his computer. "This is the Manila Trench," he says, pointing to a line that begins off the west coast of the Philippines and continues north to Taiwan. "It's 800 miles long and hasn't done anything big in 500 years. If it broke in a magnitude 9, it would have very serious consequences along the Chinese coast—the tsunami would focus right on Hong Kong and Macau. We don't know if it will break, but I think we have to assume that it can. And there are many others."
Among them is the Cascadia subduction zone, a 600-mile-long offshore fault that runs from northern California to southern British Columbia. Geologists have found sand deposits up and down the coast that were laid down by a tsunami 312 years ago, in 1700. Recent evidence from seafloor sediment cores suggests to some that about 40 earthquakes have occurred along the Cascadia fault zone over the past 10,000 years, an average of one every 250 years; other researchers estimate the recurrence interval at 500 years. When the fault does rupture, most agree, the earthquake could be as large as the one that hit Japan last March, and the tsunami could reach the coast in 20 minutes.
A lot will depend on the season, says Nathan Wood, a geographer with the U.S. Geological Survey in Vancouver, Washington. "The Pacific Northwest coast is sparsely populated for the most part, and many people are less than a mile from high ground," Wood says. "But in the summer there can be 100,000 people on the coast. We could have tens of thousands of deaths."
In Washington there are tsunami evacuation signs, tall towers on the beaches to broadcast warnings, and tsunami information booklets in hotel rooms, next to the Gideon Bibles. But evacuation centers are sparse, and not everyone has access to high ground. Ocean Shores, a resort town that NOAA lists as "tsunami-ready," lies on a narrow peninsula with no high ground and just one two-lane road to safety; 5,500 people live there year-round, many more in the summer. One evening last summer I drove around the town with Jody Bourgeois, a geologist at the University of Washington. "These people are toast, soggy toast," she said glumly.
Seattle, tucked away in Puget Sound behind the Olympic Peninsula, would probably not be hard hit by the tsunami, though it would certainly feel the shaking from a Cascadia quake. But geologists have discovered smaller, shallower cracks in the crust that extend under Puget Sound. "The picture has just started to come together in the last two decades," says Bourgeois. "It's a major, major hazard." The earthquake from a shallow fault could be extremely destructive, and a moderate tsunami launched right off Seattle might be even more damaging than a giant one off the coast. It's not clear how often such events happen in Puget Sound. The last one was about a thousand years ago.