When the tsunami arrived on Friday, March 11, 2011, Masaru Sakai sat idling in traffic, en route to deliver a load of oysters.
Forty-five minutes earlier, the largest earthquake Sakai had ever felt shook the port city of Kesennuma in northeast Japan like a rag doll. Sakai, now 67, went about his life once the shaking stopped, as did many others. It was another seismic event in a land prone to frequent tectonic convulsion, notwithstanding the length and intensity of it. In his truck bed was a shipment of mollusks from an ocean farming collective he’d helped to establish three years prior. It was the group’s first return on investment and Sakai meant to deliver. Despite having lived nearly his entire life in a maritime community in which it was common knowledge that earthquakes precede tsunamis, even when the warnings began to broadcast over the citywide loudspeakers that afternoon, Sakai stayed the course.
“I chose to interpret the situation as I wanted,” he told me through an interpreter this May at the traditional Japanese inn he and his wife run on the island of Ōshima.
But by the time he saw it in his rearview mirror, it was too late. Sakai recalled the tsunami as kuroi—black—and the approaching wave’s low, tumbling roar sounded like goro-goro-goro. “I knew I was going to die on that street,” Sakai said.
Fill a dinner plate with water. Jam another plate beneath one side, and as the water cascades off the other, you will have crudely modeled the 2011 Japan tsunami. Such a banal simulation, however, would belie the reality that the preceding megathrust earthquake—measuring 9.1M on the Richter scale—released the energy equivalent of roughly 45,000 of the “Little Boy” atomic bomb dropped on Hiroshima in 1945. This power lifted the seabed and ocean above it into a series of waves that measured as high as 128 feet upon making landfall under an hour later, and was dissipated by the lives of roughly 22,000 Japanese, of whom 1,356 died in the Miyagi Prefecture city of Kesennuma.
That last figure includes the singular death of my Japanese grandmother, who died at a rest home for the elderly in the Kesennuma district of Shishiori.
Seven years later, “3/11,” as many Japanese people now refer to it, remains only the latest deadly tsunami of more than 60 in the country’s recorded history, and the fourth in what is roughly a thousand-year cycle associated with the subduction of the Pacific tectonic plate beneath Japan. The Pacific plate is one of many possible tsunamigenic quake sources, and scientists believe the Nankai Trough, another subduction zone much closer to the Japanese coast than the source of 3/11, will produce a similar megaquake and tsunami within the next 30 years. And when the Pacific rears and pitches itself again at Japan, the island nation’s tsunami warning system will sound the alarm having benefitted from vast post-3/11 technological improvements.
But when that day comes, while the difference between life and death will remain at the unpredictable intersection between technologically-aided information and human behavior, officials and experts believe two words could tip the balance toward survival: takai and kyodai.
That is, high and huge.
To understand what happened on 3/11, I met with Kenichi Sato, then-chief of Kesennuma’s municipal Disaster Preparedness office. A soft-spoken man, Sato’s voice gave out when he tried to relate the human cost of 3/11 to me. He removed gold-rimmed glasses to wipe away tears as he recounted the story of a subordinate of his who remained at his post for days after the waves hit, despite having lost both his wife and young daughter to the tsunami.
Sato, who retired from the city the year after, described the Japanese tsunami warning system as essentially a pyramid with three levels. At the top is the Japan Meteorological Agency (JMA), armed with a dense network of sensors that was strictly land-based in 2011, and an information system that enables nationwide tsunami warnings within three minutes. Beneath the JMA, local government Disaster Preparedness offices continuously educate the population through community outreach and response drills, and turn JMA’s warnings into evacuation orders. The at-risk citizen, finally, comprises the foundation of the system. It is this individual who must choose to participate in local tsunami awareness programs, and ultimately decide whether or not to heed an evacuation order.
On 3/11, Sato had staggered through the temblor down two flights of stairs at Kesennuma city hall to his office. By the time he arrived at his desk and the violent shaking ceased, the power had gone out. He knew the clock was ticking. At 2:50 PM local time, roughly four minutes after the quake, Sato issued a general warning over the independently-powered citywide system of loudspeakers: large earthquake, turn off your gas, potential for tsunami, the alert blared. (The warning system can be heard at the 90-second mark in a clip of what appears to be spliced-together cell phone footage taken in Kesennuma's Shishiori neighborhood. The tsunami appears two-and-a-half minutes in, accompanied by warning klaxons, and rises for the next 15 minutes.)
At 2:52 PM, Sato received an email containing JMA’s tsunami alert for Miyagi Prefecture from an online weather service. (At the time, the cell system was still functioning.) Though he was supposed to confer with the mayor first, Sato then broke protocol and immediately issued the first of twelve tsunami evacuation orders broadcasted to the more than 80,000 Kesennuma residents over the next 45 minutes.
At approximately 3:30 PM, the tsunami made landfall.
A lifelong fisherman and resident of Ōshima, Shoji Murakami was a couple kilometers to sea when the earthquake began. At an interview in a small prefabricated home he now rents from the government, he recalled the feeling of the sea shuddering his boat, the sound of his equipment rattling. “I knew a tsunami was coming,” Murakami said.
He made for the island, then ran up a hill overlooking the ocean where he joined four others. What he saw confirmed his belief that a tsunami was on its way: currents reversing away from shore, waves lapping against the wrong side of a breakwater. Murakami remembered hearing a warning about a six-meter tsunami, but thought his home was higher than that and besides, there seemed to be plenty of time, so he descended and drove home. (Everyone else on the hill likewise left, he added, to see if the quake had damaged their homes.) He walked in his door, observed nothing amiss. Then he heard a roar from the side of the home that faced the water, which was normally a football field away.
In what felt like moments, the wave ripped open a wall and the ocean flooded in. The cold water surged around him, swept him across the living room. Murakami’s hand found a pillar and he struggled to hold on, kicking to keep up with the rising water even as it submerged him. His strength gave out as the current slowed. Murakami floated toward the front door, his head now near the ceiling. He recalled hearing his phone off the hook, somehow sounding a dead tone from the depths. I’m next, he thought.
Why, having heeded the initial warning, had he returned home? Like Masaru Sakai, who was about to capitalize on an investment and figured that even if the tsunami materialized, it wouldn’t be a concern, Murakami rationalized not evacuating. He had no objective data to support the hunch that his residence was high enough to be safe, or that time truly was on his side, and yet on both counts Murakami talked himself into believing so. Viewed in this light, the decisions that led him and Sakai to each stay put in their own way, assuming everything would be fine, emerge as a profound lesson on the fallibility of human decision-making in the face of crisis.
An early study of tsunami-affected Japanese behavior on 3/11 surveyed 1,998 survivors who could speak for their own pre-tsunami actions in addition to those who perished. The study, carried out at Waseda University’s Department of Civil and Environmental Engineering in Tokyo, concluded that 48 percent of those who did not, or could not evacuate, died. This might seem to reduce the good fortune of those like Sakai and Murakami to a coin flip, but would elide another, more powerful lesson: Ninety-one percent of tsunami survivors made the conscious decision on 3/11 to either evacuate immediately or at the direction of others, according to the study. The ocean’s surges killed Japanese who were stuck in traffic, trying to help, checking on their homes, or simply doing nothing, but those individuals who evacuated immediately or soon after did increase their odds of survival.
JMA’s standard was, and remains, three minutes from quake to tsunami warning. But on 3/11, the agency’s reliance on seismic data to predict tsunamis met its limit. The method required correlating quake magnitude and location with previous and simulated tsunamis, but the earthquake that day was so large and long that it resulted in “data saturation,” according to a JMA after-action document released in 2013. The initial estimation of 7.9M was too low, which then drove a tsunami height underestimation released at 2:50 PM. Minutes ticked by. With many sensors off-line from the quake and vast amounts of data pouring in from those that remained online, each second eroded attempts to accurately calculate the quake magnitude, and by extension, the resulting tsunami.
Then, at 3:10 PM, a Japanese Coast Guard buoy 10 kilometers off Kamaishi recorded a rapid increase in ocean height, before the floating device ominously went dead. At 3:15, still lacking an accurate magnitude, JMA used the Kamaishi buoy data to update the tsunami height forecast. Miyagi Prefecture went from a six-meter (about 18 feet) tsunami warning to 10 meters (about 30 feet) but few in Kesennuma, if any, were privy to the update for one simple reason: the quake had interrupted the power. Sato received the email update on his phone at 3:21, nine minutes before the tsunami arrived. By that point, he’d already broadcasted as many warnings, which were tweeted out dozens of times by the department account. Whether increasing the forecast height by four meters would have saved more lives in Kesennuma is still anyone’s guess.
“The final decision will belong to those advised to evacuate.”
Two years later, JMA released its findings document and rolled out a $25 billion (USD) investment that increased the number of land-based seismic sensors to 280, all upgraded with off-grid power and satellite communications as a safeguard from future outages like the one on 3/11. The agency installed 80 low-sensitivity Broadband Strong Motion Meters to ensure the sensors are not overwhelmed in the midst of an earthquake of 8.0M or greater, and also now incorporates data produced by two cabled seafloor earthquake and tsunami observation systems owned by Japan’s National Research Institute for Earth Science and Disaster Prevention (NIED).
Additionally, when a tsunami occurs, provided a preceding quake has not shut down cell service, those in Japan will also now receive warning notifications on their phones and through other means of mass communication. On 3/11, Sakai, Murakami, and Sato did not report receiving warning texts. The country’s J-Alert system, a Fire and Disaster Management Agency public alert system rolled out in 2007, is meant to cover a broad range of emergencies to include tsunamis with rapid multi-medium broadcasts (similar to the Wide Area Emergency System in the US) but covered only about 30 percent of the country by 2011. While the system appeared closer to being fully implemented at the time of this writing, the reality is that an at-risk individual in Japan could simultaneously see a tsunami warning on TV, hear it on the radio and over loudspeakers, and see it on their phone. The decision to do something about it will still remain theirs.
Buried in that 2013 JMA report was a subtle change of methodology that attempted to address the human factor described in the Waseda University study and others like it by the inclusion of two attributes: takai (high) and kyodai (huge). “When JMA recognizes the possibility of underestimation in a calculated magnitude and issues an initial tsunami warning,” the report reads, “qualitative terms such as Huge and High are used rather than quantitative expressions.”
Translation: In the event of another tsunami-producing earthquake of 8.0M or greater, JMA will no longer provide initial height estimations with its initial three-minute warning, which, as 3/11 made apparent, might be the only warning at-risk Japanese receive. Instead, a JMA spokesperson told me in an emailed statement that the agency acknowledges the risk of miscalculation for a large quake and that in such a case, the use of subjective words like takai and kyodai in a warning is the best way to motivate evacuation.
In other words, imperfect information delivered immediately is better than perfect information that arrives too late.
Since 2013, Japan’s upgraded post-3/11 tsunami warning system has gone untested in a large earthquake. Only a November 2016 quake that measured 7.4M, well below the threshold of what the new system was designed for, necessitated a tsunami warning. In that incident, the forecasted height of three meters or greater prompted an evacuation of 414 residents from Minamisoma (a city roughly 50 miles south of Sendai, the seat of Miyagi Prefecture), according to a Disaster Preparedness official. The official was unable to identify whether the number of actual evacuees matched what would be expected for a tsunami of that height.
It’s this tricky confluence of human behavior and technology that JMA and Disaster Preparedness personnel will continue to navigate. In the meantime, promising studies of early tsunami warning systems have emerged since 3/11. One offers a low-tech, low-cost solution that reverse-engineers GPS signal shift, while an AI-centric research collaboration promises rapid, high-fidelity tsunami inundation forecasts. Notably, the researchers behind both studies acknowledge their work only addresses the initial challenge of rapid tsunami prediction.
When it comes to how human beings use the information, as Taro Arikawa, a professor at Chuo University’s Tsunami Disaster Prevention Studies Program put it, “The final decision will belong to those advised to evacuate.”
Stuck in traffic as the tsunami barreled down on 3/11, Sakai remembers that he saw a narrow side street and hit the gas as a wall of black closed in. It first lifted the rear of his truck, then the rest of the vehicle—before long he was afloat, watching the timber of destroyed homes slip by amid the overwhelming din of destruction and so much still-rising water. Buoyed maybe 10 meters high, Sakai took his foot off the accelerator. Water had begun to flood into the cab, but somehow the truck’s engine was still running. He lowered his window and climbed out, onto layers of debris that clogged the surface. From this tenuous footing, Sakai managed to scramble to the roof of a nearby building that hadn’t washed away.
He smelled oil. He heard the gurgle and rumble of a world and the lives it contained disappearing by force of nature. It was cold. Sakai knew he would have to save himself. He could see higher ground nearby—he could hear survivors calling out. He’d shiver on that rooftop for what felt like an hour, waiting for a calm-enough window of water to again navigate over floating debris that could support his weight until eventually he reached solid ground once more. He was freezing, soaked, clothing torn, and bleeding. But alive.
Sakai has since published a tsunami awareness book, spoken publicly on the matter, and run for public office, the seeming activities of a survivor with a newfound sense of purpose. But of all these, his lasting legacy might be simply telling the story of why he didn’t evacuate on 3/11. For it is voices like his and Murakami’s, in the absence of those 22,000 who perished, that remain to inform our understanding of the human factors that will influence who will live and die when the next tsunami strikes Japan.
“I don’t believe I survived,” Sakai said. “I was saved.”