Richard Allen was riding Bay Area Rapid Transit between Berkeley and Oakland when the train suddenly stopped. “We’ve had an warning system of earthquake,” the conductor announced. “We’re going to assess the situation and decide what to do.”

The newest warning system give users ten seconds’ notice. What can be done in that time?

One Saturday afternoon a few years ago, Richard Allen was riding Bay Area Rapid Transit between Berkeley and Oakland when the train suddenly stopped. “We’ve had an alert for an earthquake,” the conductor announced. “We’re going to assess the situation and decide what to do.” Allen, the director of the Seismological Laboratory at the University of California, Berkeley, had spent much of the previous two decades working toward that moment. He studied the faces around him, looking for signs of panic. A couple of locals sat down and inspected their phones; a family of tourists joked about their Bay Area bucket list (Alcatraz, Golden Gate, earthquake—check). After a few minutes’ orderly wait, the train’s journey resumed. Apparently, earthquake early warning, or E.E.W.—the technology that Allen had been championing since 2001—could feel routine, like an everyday inconvenience.

In fact, the warning on the train was quite novel. Allen’s fellow bart riders likely didn’t know that their train could brake in anticipation of shaking, not because of it. Using data derived from every major Californian quake since the late nineteen-eighties, the system had provided the riders with a brief warning of the onrushing quake. It will cost more than sixty million dollars to complete, with an annual maintenance cost of thirty million. Still, Allen and others believe that it can prevent at least half of all injuries in the next major earthquake, while possibly avoiding millions of dollars in damages.

Earthquake early warning is not earthquake prediction. The last time the United States Geological Survey tried to predict a seismic event was in 1984, and the shaking arrived at the prophesied epicenter, in Parkfield, California, eleven years too late. In a 2009 book, “Predicting the Unpredictable: The Tumultuous Science of Earthquake Prediction,” the geophysicist Susan Hough explains that earthquakes seem to “pop off in the crust like popcorn kernels”; seismologists may have “no way to tell which of the many small earthquakes will grow into the occasional big earthquake.” As a result, researchers have essentially given up on the idea of pinpointing when a quake will occur.

As a sort of consolation prize, E.E.W. allows seismologists to estimate, in real time, the potential amount of shaking an earthquake will cause. The principles are relatively simple. Seismometers can scan the ground for the “P waves,” or compression waves, which precede the slower and more powerful “S waves,” or shear waves, that are responsible for severe earthquake damage, like lightning before thunder. Using the P waves, a local data-processing center can calculate the likely reach and magnitude of a quake moments after it begins. The resulting alert, dispatched at the speed of light, usually beats the S waves, which ripple through rock at about two miles per second.

To call the warning “early” is generous. It usually arrives between a few seconds and less than a minute ahead of the quake—advance notice that, in duration, is somewhere between a sneeze and a red light. The Achilles’ heel of the system is what its designers call the “blind zone.” Those who might benefit the most from a warning—the people at the epicenter—won’t get one, because the S waves arrive too soon. Still, if you’re far enough outside that bull’s-eye, a few moments can be meaningful. In the past few decades, more than half of earthquake injuries in the U.S. have been caused by people or things falling—two occurrences easily avoided if you have time to take cover beneath a sturdy piece of furniture. For those who are landing planes, assembling electronics, operating cranes, or drilling into molars, even the smallest warning would be welcome. You might have just enough time to lock the wheels on your wheelchair, or to remove your scalpel from your patient’s chest. The effectiveness of the warning depends on how much can be done in a handful of seconds. The goal of E.E.W., therefore, isn’t just to sound the alarm. It’s to help transform knowledge into action as quickly as possible.

The traditional E.E.W. method was first implemented in Japan, in the nineteen-sixties and seventies, to protect the country’s then new bullet train from derailment. It uses conventional seismometers, which are firmly planted in the ground. By 2007, Japan’s Meteorological Agency had built a nationwide network of more than a thousand seismic sensors; when intensity surpasses a particular threshold, cell phones issue an automatic warning, and radio and television stations broadcast a surprisingly melodious chime based on a snippet of a symphony by the composer who also happens to have written the original score for “Godzilla.” The system isn’t failproof: during the 2011 Tohoku earthquake—a megathrust quake and the biggest test for E.E.W. to date—it managed to halt bullet trains and alert more than fifty million people, but ignored a few affected regions altogether. Still, other countries have followed suit. In the late nineteen-eighties, Mexico used this technology to set up a seismic trip wire between its capital city and the coast, in order to catch offshore tremors. Romania has installed E.E.W. for a nuclear facility, and Turkey monitors Istanbul’s undersea Marmaray Tunnel; South Korea activated nationwide phone alerts in 2015. In 2023, China’s Earthquake Administration is expected to complete a landmark multibillion-dollar detection network, complete with more than five thousand seismic sensors.

Two years ago, the U.S.G.S. brought cell-phone warnings to Californians. Its earthquake tracking network, ShakeAlert, relies on more than seven hundred seismic stations, capable of detecting everything from trees swaying in the wind to distant quakes across the Pacific Ocean. Designed in partnership with Berkeley, the California Institute of Technology, the University of Oregon, the University of Washington, the Swiss Federal Institute of Technology, and emergency-management agencies, ShakeAlert sends out mass public warnings through third-party apps and fema’s Wireless Emergency Alert system; it is set to warn about earthquakes more severe than a relatively moderate magnitude 4.5 (rattling plates, confused car alarms, a high-school math test interrupted but not necessarily cancelled). ShakeAlert expanded to Oregon and Washington—where researchers expect a major earthquake at some point in the future—this spring.

Meanwhile, Google has launched its own system based on crowdsourcing. A few years ago, Marc Stogaitis, an Android software engineer, realized that the company had unintentionally distributed billions of seismometers in the form of the accelerometers inside smartphones, which can differentiate not only between a walk and a jog but—when linked together—between the rumble of a subway car and an earthquake. Stogaitis’s team partnered with a few academic researchers, including Allen and his collaborator Qingkai Kong, who had already collaborated on an E.E.W. smartphone app called MyShake, and the Caltech seismologist Lucy Jones; the result is that Android can now warn its users when the ground beneath their feet is about to start shaking.

It’s nearly impossible to miss the “take action” alert. The phone flashes full-screen instructions to “Drop, Cover and Hold,” overriding any silent settings with an authoritative alarm. In places with established detection systems, such as the West Coast, the Google feature simply delivers—and often speeds up—warnings, using a mixture of Wi-Fi and cellular service. But elsewhere the phones themselves can now serve as monitoring stations. In June, the global rollout of the Android Earthquake Alerts System travelled past New Zealand and Greece to reach Turkey, the Philippines, and most of Central Asia. In those areas, which have a high concentration of seismically vulnerable structures, the free service may be especially helpful—although the warnings will only go out to Android users.

Wherever the warnings come from, the E.E.W. window is short; it’s important to make the most of the alerts. In Santa Monica, California, a company called Early Warning Labs—“Seconds Count,” its Web site warns—has begun helping businesses link the ShakeAlert network to custom P.A. systems. Some of its clients are owners of high-end buildings; the firm has also worked with the Cedars-Sinai Medical Center, NBC Universal film studios, Los Angeles City Hall, Santa Monica College, and a library and public school.

Audible alerts are only part of the promise of E.E.W. Many see it as a path toward automated earthquake resilience—a kind of citywide pause button. In downtown San Francisco, the elevators in the thirty-four-story P.G. & E. headquarters are already programmed to let passengers out on the nearest floor before locking in place; at a high-tech Menlo Park fire station, the electronic bay doors will swing open, averting the risk of a power outage blocking the exits, as happened during the 1994 Northridge quake. Further north, in Vancouver, an alert directs incoming traffic away from the outdated George Massey Tunnel, which is not expected to survive the next serious quake.

Companies licensed by ShakeAlert are piloting programs to shut off gas and water valves, turn on emergency lights, secure sewage tanks, and stop turbines in hydropower plants. Google says that it would like to integrate an application programming interface, or A.P.I., into its system, which would allow developers to build off its earthquake alerts. Construction sites, laboratories, amusement parks, freeways, and factories, among other sites, could theoretically benefit from early-warning automation. As such a system would have more immediate power over a city than any public official, the U.S.G.S. and Google say that they have set up safeguards against hacking—although eventual third-party partners will need equally robust defenses to prevent a malicious or accidental pause.

In advance of such automation, the track record of the system as a whole is promising but mixed. According to unpublished data gathered by MyShake, on September 19, 2020, during a 4.5-magnitude quake east of Los Angeles, ninety per cent of alerted users received their warning before the shaking reached its peak. And yet this summer, when a forceful magnitude-6.0 earthquake hit Antelope Valley, California, causing rock slides and shattering glass in local shops, many people weren’t warned: the epicenter of the quake happened to be in one of the last spots still waiting on a seismic station, and so ShakeAlert misinterpreted the tremors as three separate events. In California and the Pacific Northwest, many faults lie beneath major cities; this makes it especially challenging to deliver alerts quickly. (In other places, a larger geographic buffer between the epicenter and populated areas translates to at least a few extra seconds’ warning.)

Google’s system has its own limitations. It performs best wherever many Android users live and work. The oceans, as well as rural areas, remain a challenge. (Grillo, a Mexico City-based social enterprise, brings cheap seismic sensors to at-risk regions; researchers in Costa Rica have recently tried tacking smartphones to the wall, for a tiny fraction of the cost of a traditional seismic station.) Google has trained its ground-motion algorithms to ignore inputs like thunderstorms, but it’s possible for other phenomena to trick the phones into a false positive. (Even the buzzing earthquake warnings delivered by fema’s Wireless Emergency Alert system could have been identified as a separate quake.) The Android engineers are refining their service, one quake at a time. “You learn at the speed that Mother Nature teaches you,” Stogaitis said.

In 2020, David Wald, a U.S.G.S. seismologist, published a paper in the journal Earthquake Spectra, challenging what he sees as unrealistic expectations for the new technology. “Most EEW projections presume activation and response times to be instantaneous or ignore them altogether,” Wald wrote. The brief window for action created by the systems doesn’t allow for hesitancy or confusion. (In 2019, some residents of Sichuan, China, mistook the region’s first earthquake-alert siren for an air-raid drill, a call for firefighters, an explosion, and the début of a shopping sale, among other non-geological events.) It’s not a given that people will reflexively follow the illustrated instructions displayed on their phones. To some, an advance warning might seem like an invitation to evacuate, or a prompt to check on other family members. Many earthquake myths persist—doorframes are no good, it turns out—and there are still gaps in public awareness: What do you do if you get the alert in bed? (According to fema, you should turn face down and cover your neck and head with a pillow.)

In processing seismic signals, there is a trade-off between timeliness and accuracy. The warnings must arrive quickly, but fast calculations about the reach and intensity of a quake could be wrong. “The public can and will immediately assess if our information was correct or not,” Allen and the seismologist Diego Melgar wrote, in a recent paper. Too many unreliable interruptions might drive users to opt out of the service. Conversely, others may want more alerts: when the system launched, residents of Los Angeles complained that they weren’t being warned before minor shaking, and the intensity threshold had to be adjusted. It will be hard to satisfy everyone.

Still, preliminary studies suggest that people like early-warning systems even when they don’t work perfectly. Evidence is limited to seismically active Japan and Mexico, but in those countries tolerance for false alerts and miscalculations is high. “Japanese people are really patient with the alarms,” Masumi Yamada, a professor of earthquake hazards at Kyoto University, told me. “That is because people really understand the dangers.” Even when warnings arrive too late to make a tangible difference, Yamada said—just two or three seconds ahead of time—they may still matter psychologically. A 2019 study conducted in Japan found that most respondents used those hard-won seconds to prepare for the earthquake mentally, rather than physically.

Collective-behavior researchers have observed that people in crowds often wait for outside confirmation before acting on an emergency warning. In that sense, informative earthquake warnings may help us to react more intelligently to the earthquake itself. And there is something valuable about just knowing what’s coming. Suppose that you are about to get into a car crash, Yamada said. Would you want to be awake or asleep? Personally, she told me, she’d take “a little information” over being “totally unconscious.” Either way, you will get hit by the car. But you might brace yourself and anticipate an impact. The Android alarm has a mute button. You might silence its insistent ringing and take a deep breath.

Source New Yorker

By Arsalan Ahmad

Arsalan Ahmad is a Research Engineer working on 2-D Materials, graduated from the Institute of Advanced Materials, Bahaudin Zakariya University Multan, Pakistan.LinkedIn: