Cascadia's Fault (47 page)

Now in my mind-movie, I can see several small boats scurrying to get inside the harbor before the giant tsunami gates grind shut against the outer bay. A team of firefighters stands by as the steel doors clang together to form a solid barrier against the coming waves. I've seen footage of this scenario on television several times. The people were proud of how quickly they could seal themselves off from the ocean's fury: fifteen minutes max. And on March 11, 2011, that ponderous machinery worked exactly as planned.

Unfortunately, the plan was the problem. It was no match for what nature dished out. The quake was much, much stronger than expected, and the waves much higher than any tsunami in living memory. The subduction zone did something scientists had seen in computer simulations but never before in real life.

For the first sixty seconds, the fault ripped apart pretty much as expected. The overlying tectonic plate—the Okhotsk plate, upon which Japan's main island of Honshu rests—came unstuck from the Pacific plate, the slab of rock that was pushing its way underneath. From the epicentral area, some nineteen miles (30 km) below the sea floor, the lower plate at first lurched downward and westward—toward Japan. And then it kept on ripping. Segment after adjacent segment of the Japan Trench broke apart, spreading damage farther and farther along the coast. The earthquake kept getting bigger. What started at magnitude 7.9 quickly jumped to 8.8.

After seventy-five seconds, the fault did a stunning turnabout: it started rupturing in the opposite direction. In two directions at once, actually. Now it tore itself apart from the epicenter
uphill
along the fault plane and eastward, farther offshore—out toward the Japan Trench where the two plates first meet.

The world's most sophisticated and densely packed grid of seismometers, pressure gauges, and GPS tracking devices measured the steadily increasing magnitude of the quake and the height of the coming waves and beamed the data to satellites. The data automatically triggered alarm systems in Japan and around the Pacific Rim. Bullet trains screeched to a halt, and nuclear power plants initiated emergency-shutdown procedures.

It took more than 180 seconds for the rocks to release several centuries' worth of stress. An area of sea floor roughly 250 miles (400 km) long and 125 miles (200 km) wide lurched sideways—at least 79 feet (24 m) in most areas and as much as 164 feet (50 m) in others. The Tohoku-Oki earthquake will probably turn out to be the largest fault slip ever observed. It was more than twice the peak slip recorded during the Sumatra quake of 2004 (magnitude 9.4). It had twice as much horizontal movement as seen in the Chile quake of 2010 (magnitude 9.0). It caused an even bigger lateral lurch than the one that occurred during the Chile disaster of 1960, the largest earthquake ever recorded (magnitude 9.5). How could this be?

Out along the Japan Trench, the thin edge of the overlying continental plate popped loose from the plate below and heaved upward ten feet (3 m), hoisting the ocean with it. Think of a volume of seawater 250 miles long, 125 miles across (400 by 200 km), and several miles deep being thrust upward—all that water—roughly ten feet straight up. A seething liquid hump of kinetic energy.

Moments later, gravity forced this saltwater mountain to collapse. A series of concentric waves rippled outward, as if an enormous boulder had been dropped into a shallow pond. The birth of a killer tsunami.

As stress was released from the rocks, the upper plate relaxed. The
land stretched like a piece of taffy, then sank lower into the sea than it had been before, dragging the coastline down with it by at least three feet (1 m). From this moment on, many of Japan's outer beaches, bays, towns, and villages would be three feet below sea level.

Thirty minutes after the main shock, thousands of people in Minamisanriku had scrambled into buildings they thought were safe. But because both the quake and wave were much bigger than expected—and because the entire coastline had just sunk by three feet—those “invincible” seawalls simply vanished beneath the surge. The breakwaters and tsunami gates were overtopped and undermined. It was as if they had never been built. All that money and years of effort had created nothing more than a house of cards, a fatally false sense of security.

An estimated 95 percent of Minamisanriku was destroyed. Of the eighty buildings officially designated as tsunami-evacuation shelters, thirty-one were completely wrecked. Two evacuation sites, one on a headland overlooking town from the south and another farther inland, were also inundated; both were at least sixty-six feet (20 m) above sea level and presumed high enough to be considered safe ground. But the people who gathered there were washed away.

The town's mayor and more than 130 public officials escaped from city hall to the Disaster Management Center, a solidly built steel-frame structure three stories tall with a vertical evacuation shelter on the roof. Only thirty made it to the top. Even there they were not safe.

From her desk on the second floor, twenty-five-year-old Miki Ando, an employee of the Crisis Management Department, continued broadcasting emergency information via the city's extensive network of loudspeakers; she never left her post even as the horrifying waves tore through the streets below her and climbed ever higher against the building's groaning walls. Ando repeatedly urged her fellow citizens to head for higher ground, and she was credited with saving many lives by doing so.

A time-stamped photo taken from the top of the building and discovered days later showed that the Disaster Management Center was
fully engulfed by 3:35 p.m., forty-eight minutes after the earthquake had started. Of the thirty people who made it to the roof, only eleven survived, several of them clinging for life to a radio antenna as the ocean torrent—a wall of debris-filled seawater fifty-two feet (16 m) high—thundered past.

An aftermath photo, taken weeks later by geologist Lori Dengler of Humboldt State University, showed nothing left of the building but a gutted steel skeleton. Clearly visible in the background behind this mangled red-metal hulk are the green, tree-clad hills that could have saved more lives if only people had run a few minutes farther and higher.

Miki Ando was among the 9,500 people of Minamisanriku either killed outright or swept out to sea and presumed drowned. More than half the town's population was gone in less than an hour. The same story was repeated in town after town along Japan's battered northeast coast.

When the main shock ended, the final magnitude measured 9.0, the largest earthquake ever to hit Japan. “They were expecting an 8.2 to an 8.4, which is twenty to twenty-five times smaller than a magnitude 9, in terms of energy,” explained Chris Goldfinger. Japan had prepared itself for the wrong earthquake.

The waves broke records as well, with a high-water mark of 127.6 feet (38.9 m) at Aneyoshi Bay south of Miyako City. That's a bulldozer blade of water more than twelve stories high. It was the largest tsunami ever measured in Japan's long experience with killer waves. On the broad, flat Sendai plain, the surge was about thirty feet (9 m) high and it penetrated roughly two miles (3 km) inland, destroying farms and homes as far as the eye could see. Japan's northern network of seawalls and breakwaters was designed for a worst-case wave of about thirty-three feet (10 m) above mean sea level. But the ground had sunk three feet, so most of the barriers were “severely overtopped or destroyed,” according to a report from the American Society of Civil Engineers (ASCE). They were torn apart by waves that, in many cases, were double the expected height.

Japan's real-time warning system—triggered by a dense array of instruments connected to a supercomputer that can process thousands of potential disaster scenarios—worked, but not as well as expected. Eight seconds after the primary wave from the earthquake reached the first seismic station, the Japan Meteorological Agency (JMA) issued an alarm to people living closest to the epicenter. Almost instantly, twenty-seven bullet trains in the region were stopped without a single derailment.

The computer initially reported a quake of magnitude 7.9 and forecast tsunami surges of about ten feet (3 m) along the Iwate and Fukushima coasts, with waves as high as twenty feet (6 m) in the Miyagi district. This news flashed across the country via television, radio, cell phones, and community loudspeaker systems. But the quake kept getting bigger, and the news kept changing. JMA had to update the bulletin continuously over a period of nearly four hours. In the Greater Tokyo Area far to the south of the quake—where high-rise towers were quite obviously swaying as damaging shockwaves rolled through the ground—there was no official warning. Fortunately, the bullet trains and nuclear power plants in the Tokyo region had their own warning systems to shut the operations down.

Clearly, the supercomputer underestimated the size of the catastrophe. Critics later called this a failure of the system. But the initial notification was based on just the first twenty seconds of data; only part of the fault had ripped within that time, and the computer did not know how much bigger the quake would eventually become. It took nearly three and a half minutes for the whole thing to unzip.

“If you analyze it using twenty seconds of data, you're only going to see about a hundred kilometres [62 miles] of rupture,” explained geologist Lori Dengler. “Because that's how long it takes to rupture that amount. By definition, if you're trying to do an early-warning system based on the initial part of the signal, you are always—100 percent of the time—going to grossly underestimate the magnitude of the event.”

In a post-mortem published in
Nature
, Masumi Yamada of Kyoto University wrote that “the unexpected character of the seismic data at the start of the earthquake fooled the early warning system's algorithms.” But he added that “the system has the potential to work well for the next great earthquake . . . if technical improvements are made to recognize great earthquakes quickly.”

JMA has already revised its warning protocol. From now on, if a quake hits magnitude 8 or higher, the system will not even try to forecast tsunami-wave heights. Instead, bulletins will simply announce the possibility of “a huge tsunami.” The question remains whether people will know what that means and what they ought to do. Will they perceive themselves to be at risk and take appropriate action?

Dengler, one of the leaders of an international team of scientists who rushed to the scene in search of clues about what happened, summed it up this way: “Certainly the sirens were going off, and they were announcing it as a worst-case event. But a worst-case event was not what they got. They got a
worse
-than-a-worst-case event.”

Perhaps more worrisome to Dengler were reports that nearly 40 percent of people intentionally delayed their own escape to safety. “The earthquake was a trigger [for people] to go from a safe area back into a hazard zone,” she told me. “And they did that to either rescue a relative who was at home, or to rescue belongings. And this was common. It was not rare. It was really common.

“And that's the thing that I'm really concerned about. Because I expect we're going to have people doing the same thing
here
[in North America]. The main motivations are to rescue
somebody
or to rescue
stuff
. And that's pretty instinctive human behavior.”

Not all the news was bad, however. In some places, people did know what to do and lived to tell the story. More than five hundred students from the junior high and elementary schools in the town of Unosumai managed to escape despite conflicting information and a rapidly changing disaster scene. Both schools lie outside the mapped tsunami
hazard zone of their community, so the students and teachers there might be forgiven for thinking they were relatively safe to begin with.

The two groups of students had been told completely different things about what to do in an earthquake. The elementary-grade children were trained to go to the roof of their three-story school building. The neighboring junior-high students had been taught to head for higher ground. But when the earth ripped apart on March 11, the younger students saw the older ones leaving the school grounds and decided to follow.

The older kids noticed, took charge of the younger ones, and led the way, always keeping an eye on the churning pulse of dirty-brown seawater rising behind them at a terrifying speed. They had to change their planned destination twice before finally reaching a safe spot in the nearby hills. Coaching the students to head for high ground “to evaluate the situation with their own eyes” and to assist others clearly paid off. All five hundred survived.

Most people who did make it to official evacuation sites were stranded for days. Few of the shelters had stored supplies of food, water, blankets, or bedding. There were not enough toilets and, in many cases, no first-aid stations or emergency medical care. Mid-March in northern Japan means winter weather with temperatures hovering around the freezing mark. In some cases, elderly and injured survivors later died because of the difficult conditions at the evacuation sites.

“There was no catastrophic-response plan,” explained Lori Dengler. “There was no idea that you'd really need to coordinate response on a nationwide level. There was no plan because everything was built on the faith that the engineering works were going to work.”

An estimated 92 percent of the nearly twenty thousand people who did not survive died as a result of drowning. If you count those who were washed out to sea and never found, then 96 percent of the deaths were caused by the tsunami. A significant number of these unfortunate people died because they had put their faith in seawalls and official
disaster plans. The best of intentions had undermined decades of diligent training, preparation, and public awareness.

In November 2011 the Earthquake Engineering Research Institute (EERI), based in Oakland, California, published a summary of the findings from several field reconnaissance teams sent along with two International Tsunami Survey Teams (ITST) to the hardest-hit areas of Japan. In essence the EERI teams concluded that “failure to evacuate” was the primary cause of the high casualty rate. They also noted that the official forecasts had underestimated the strength of earthquakes along the Japan Trench because the projections were based on what had happened during “relatively recent” historic events.