The Earthquake Phenomena Observation System, located inside the Japan Meteorological Agency in Tokyo’s central Otemachi district, is usually operated by five teams of seven who work in rotating shifts that span every minute of the year. But at 2:46 p.m. on March 11 this year, all that changed. In an atmosphere that even one of Japan’s famously reserved bureaucrats — an agency staffer — admitted was “extremely intense,” everyone who physically could report for duty did, and some didn’t go home for the next 72 hours.
During the first three days after the Great Eastern Japan Earthquake struck, the EPOS operators in Tokyo detected and monitored what was literally a nonstop barrage of aftershocks. In that 72-hour period, they recorded more than 250 quakes of magnitude 5 or higher — that’s an average of one every 17 minutes. At least 45 of those exceeded magnitude 6, and three topped magnitude 7.
The shocks were concentrated across a massive, 300×500-km area off the northeastern coast of Honshu — extending from waters off Iwate Prefecture in the north, to those off Ibaraki Prefecture in the south. The EPOS operators would have been painfully aware that any one of the shocks they were monitoring had the potential to send tsunami smashing into Japan’s coast — just like the ones that had probably already taken some 20,000 lives on March 11.
Recalling the circumstances of those fearsome 72 hours now, it is easy to slip into the language of combat — with the EPOS operators assuming the role of frontline scouts predicting and warning where the next quake would “strike,” and then scrambling to erect “shields” in the form of earthquake early warnings and tsunami alerts. On March 25, Mitsuhiko Hatori, the agency’s director general, himself said that his staff would continue to do their best to “protect the populace.”
But, if the Meteorological Agency was (and is) engaged in a battle with a restless planet, then a more one-sided contest would be hard to imagine. Really, all they can do is try to keep tabs on an adversary that is almighty, and then try to forward on the information they’ve gleaned as quickly as possible.
To track seismic activity in and around Japan, the Meteorological Agency has established multiple networks of monitoring stations that are the most comprehensive of their kind in the world. At 2:46 p.m. on March 11, they swung into action.
The quake was caused when the edge of the Eurasian Plate sprang back upward after centuries of being dragged down by the west-moving Pacific Plate sliding beneath it. With its epicenter some 130 km east-southeast of the Oshika Peninsula in central Miyagi Prefecture in Japan’s north, as the Eurasian Plate skidded upward it sent seismic waves of energy, known as P-waves, racing through the earth’s crust toward the Miyagi coast at between 6 and 14 km per second.
What might be called the Meteorological Agency’s first line of defense is a network of some 200 P-wave-detecting seismometers placed at strategic locations throughout the country. (The network is bolstered by links to several hundred other seismometers operated by other government agencies and universities.) On the afternoon of March 11 — at precisely 2:46 p.m. and 40 seconds — the EPOS operators in Tokyo started receiving P-wave readings from their seismometer at Ouri in the city of Ishinomaki at the head of the Oshika Peninsula.
In such a circumstance, the primary function of the agency’s EPOS system is to ascertain as quickly as possible whether a major quake is imminent and, if it is, to issue an “earthquake early warning.”
P-waves, which are largely imperceptible to humans, move faster than a second type of wave generated by quakes, known as “S-waves,” which travel at between 1 and 8 km per hour and can be felt as tremors. For the agency’s “early warning” to be effective, it must be issued before the S-waves arrive at a location about to be affected.
As Koji Nakamura, assistant director of the agency’s Earthquake and Tsunami Observations Division, explained to The Japan Times this week, “Every second counts, so at this stage, the entire process is automated.”
On March 11, the EPOS system took 8.6 seconds to ascertain from the amplitude of the P-waves monitored at Ouri — and very soon corroborated by similar readings from other seismometers nearby — that a large quake was occurring. They then issued an early warning to various organizations, including the National Police Agency, nuclear power plants, public transport operators and media outlets. Viewers of public broadcaster NHK would have noticed a large warning flash on their screens at around 2:46 p.m. and 50 seconds.
For many people closest to the epicenter, especially those in and around the city of Sendai in Miyagi Prefecture, the agency’s earthquake early warning likely came too late to be effective, as it was issued almost simultaneously with the next stream of data picked up by the EPOS systems: readings of perceptible tremors.
What might be called the agency’s second line of defense consists of a national network of 600 seismic-intensity meters. (These, too, are bolstered by links to many hundreds of other such meters operated by other government agencies and local governments.) Those meters measure what in Japanese is called shindo — the actual degree to which the ground at that location is shaking.
Nakamura said that the easiest way to imagine the difference between a seismometer and a seismic-intensity meter is to remember that the latter was once the job of a human being. In the past, agency staff posted around the country would ascribe a numeric figure (on a scale of 1 to 7) to the intensity with which they and their surroundings were shaking.
Like seismometer readings, seismic-intensity readings are relayed automatically to the EPOS in Tokyo, and within less than a minute they, too, are forwarded to government agencies, the media and all the rest.
On March 11, viewers of NHK would have seen shindo readings of 6 and 7 logged at Sendai and elsewhere in Miyagi flash up on their screens. By then, however, if those viewers were in Tokyo, they would have started feeling the tremors themselves.
At the same time as their computers were sending out early warnings and then seismic-intensity readings, the EPOS operators at the agency would have been working on one of their most important tasks: checking computer-generated estimations of the quake’s magnitude, epicenter and depth, and then using that information to ascertain the likelihood of tsunami.
This is where the agency’s twin goals of speed and accuracy come into conflict. Agency policy specifies that the particulars of a quake — including any tsunami warnings — should be announced within three minutes of a quake’s commencement. The reasoning is that, depending on a temblor’s location, tsunami could conceivably arrive at Japan’s coasts that soon.
“This means you need to make those assessments on the basis of just the first two minutes of data,” Nakamura said. “The problem was that with this quake, which was particularly large, the Eurasian Plate was still sliding.”
On the basis of the preliminary data available, the EPOS computers determined that the quake was a magnitude 7.9 that had occurred at a depth of 10 km. They also pinpointed its epicenter. An EPOS operator checked this off and then that data was fed directly into the agency’s tsunami-simulation computers.
Those computers identified the preprogrammed quake scenario closest to the events then being detected, and generated tsunami warnings and lesser advisories for Japan’s entire eastern coast — from Hokkaido in the north to Okinawa in the south. Those were then announced along with the magnitude, epicenter and depth. Less than three minutes had passed since the quake began and the agency’s first important goal had been met.
Then, the EPOS operators really started getting busy.
In addition to its network of seismometers and seismic-intensity meters, which were by then sending back literally thousands of readings as the shock waves from the quake spread across the nation, the agency also monitors a series of 170 sea-level gauges positioned at key points around the coast. At 2:54 p.m., readings from the gauge at Ofunato in southern Iwate Prefecture indicated that the tsunami had arrived. Within two minutes of the first reading, which was then of just 20 cm, similar readings started pouring in from other gauges nearby — in Hachinohe, Miyako and Kamaishi in Iwate Prefecture and Ishinomaki Ayukawa in Miyagi Prefecture.
With each sea-level reading, agency staff were able to recalibrate their initial tsunami model and adjust their warnings accordingly.
As all that was going on, the Eurasian Plate that had set it all in motion finally stopped sliding and seismic-intensity meters located around Sendai at last began to return normal readings.
With the quake itself thus over, it became possible to start calculating just how big it had been. At 4 p.m., EPOS operators upgraded it to a magnitude 8.4. (By midday on March 13, they had bumped it up to 9.)
Almost as soon as the initial quake had subsided, however, aftershocks began. Between 3 p.m. to 3:30 p.m. on March 11, the EPOS computers detected three aftershocks exceeding magnitude 7. Each of those necessitated the same early warnings, the same calculations of scale and location — and then the same extrapolations for new tsunami warnings.
However, as they were dealing with those aftershocks, the EPOS operators started receiving the kind of sea-level gauge readings that none of them had ever seen before: over 7.3 meters at Soma, Iwate Prefecture; over 4 meters at Miyako; over 4.1 meters at Kamaishi; over 3.3 meters at Ishinomaki Ayukawa. Then, the unthinkable happened. At around 3:22 p.m., the sea-level gauges at Miyako, Ishinomaki Ayukawa and elsewhere went dead.
“The meters that were supposed to measure tsunami fell victim to them,” Nakamura explained somberly. “They had been destroyed.”
And yet tsunami damage was only one of the agency’s problems. At around the same time, vast areas of northern Japan lost power and, almost immediately, many of their vital seismometers and seismic-intensity meters started going offline.
“They have back-up batteries, but they soon started running out,” explained Nakamura.
By 6 p.m. that night, more than 30 of the agency’s 170 sea-level gauges and 55 of its seismic-intensity meters had either lost power, been damaged or been destroyed.
The “enemy” was stripping the agency of the only weapons it had to monitor its progress.
Of course, with many hundreds of seismometers and seismic-intensity meters at other locations still at their disposal, agency experts were able to continue issuing warnings and shindo readings as aftershocks continued throughout the night. But the fact that they couldn’t rely on the meters closest to the epicenter meant their response times slowed.
It was perhaps out of a desire to be safe rather than sorry that, later that night, at 3:20 a.m. on March 12, the agency revised its tsunami warnings so they covered the entirety of Japan’s coastlines — all 29,751 km of them, even including islands in the Sea of Japan, which are on the opposite side of Japan’s main island of Honshu from where the quakes were originating.
But even that array of preventive measures caused a serious dilemma.
“The agency’s tsunami warnings were relayed to all the emergency services — so as long as tsunami warnings were in place, then rescue teams couldn’t operate freely in the coastal areas where they were needed most,” explained Nakamura. “You have to weigh that consideration with the need to protect people.”
It wasn’t until 1:15 p.m. on March 12 that the agency started rolling back tsunami warnings, as the harsh drumbeat of aftershocks finally started to subside. That process coincided with agency staff stationed in northern Japan being sent to repair and restore power to monitoring equipment.
But still the agency’s headaches weren’t over. On March 12, a separate, large earthquake occurred in Nagano Prefecture — a quake that happened to coincide with continued aftershocks off the northeastern coast.
“Unfortunately, at its current level of sophistication, the early warning system cannot distinguish between two separate quakes,” Nakamura said, explaining that the coincidence of multiple tremors resulted in the issuance of erroneous earthquake early warnings.
Readers who were in Japan on March 12 and 13 will likely recall the trauma of hearing and seeing early warnings for quakes, only for them never to occur.
“In total, only about a third of the early warnings released in that period were accurate,” Nakamura said. “This is something we need to try to improve in the future.”
The EPOS operators who clocked in for duty throughout those fraught 72 hours thus had to throw themselves into a variety of tasks — from tracking aftershocks and analyzing and revising tsunami warnings, to coordinating attempts to get monitoring stations back online.
Still more agency staff were involved in preparing press releases and other documents. The announcement made at 4 p.m. on March 11 — that the quake’s magnitude had been revised up to 8.4 — was combined with no fewer than 17 pages of diagrams and data. All of that had been collected, collated and analyzed in the 74 minutes since the largest quake in Japan’s recorded history had begun.
By the afternoon of Monday, March 14, when the 72-hour period that Nakamura described as “extremely tense” came to an end, the agency had issued 17 such documents — an average of one every two hours. It was only then that things started to calm down, though they still haven’t returned to normal.
Last Monday (April 4), when asked how the mood among the EPOS operators was then, Nakamura looked sternly ahead. “Things are tense,” he said.