Is San Francisco Next?

Tokyo is more likely, says a scientist whose work on aftershocks may revolutionize quake forecasting.

Jian Lin was 14 years old in 1973, when the Chinese government under Mao Zedong recruited him for a student science team called “the earthquake watchers.” After a series of earthquakes that had killed thousands in northern China, the country’s seismologists thought that if they augmented their own research by having observers keep an eye out for anomalies like snakes bolting early from their winter dens and erratic well-water levels, they might be able to do what no scientific body had managed before: issue an earthquake warning that would save thousands of lives.

In the winter of 1974, the earthquake watchers were picking up some suspicious signals near the city of Haicheng. Panicked chickens were squalling and trying to escape their pens; water levels were falling in wells. Seismologists had also begun noticing a telltale pattern of small quakes. “They were like popcorn kernels,” Lin tells me, “popping up all over the general area.” Then, suddenly, the popping stopped, just as it had before a catastrophic earthquake in 1966 that killed more than 8,000. “Like ‘the calm before the storm,’” Lin says. “We have that exact same phrase in Chinese.” On the morning of February 4, 1975, the seismology bureau issued a warning: Haicheng should expect a big earthquake, and people should move outdoors.

At 7:36 p.m., a magnitude 7.0 quake struck. The city was nearly leveled, but only about 2,000 people were killed. Without the warning, easily 150,000 would have died. “And so you finally had an earthquake forecast that did indeed save lives,” Lin recalls. “People were excited. Or, you could say, uplifted. Uplifted is a great word for it.” But uplift turned to heartbreak the very next year, when a 7.5 quake shattered the city of Tangshan without so much as a magnitude 4 to introduce it. When the quake hit the city of 1.6 million at 3:42 a.m., it killed nearly 250,000 people, most of whom were asleep. “If there was any moment in my life when I was scared of earthquakes, that was it,” Lin says. “You think, what if it happened to you? And it could. I decided that if I could do anything—anything to save lives lost to earthquakes, it would be worth the effort.”

Lin is now a senior scientist of geophysics at Woods Hole Oceanographic Institution, in Massachusetts, where he spends his time studying not the scurrying of small animals and fluctuating electrical current between trees (another fabled warning sign), but seismometer readings, GPS coordinates, and global earthquake-notification reports. He and his longtime collaborator, Ross Stein of the U.S. Geological Survey, are champions of a theory that could enable scientists to forecast earthquakes with more precision and speed.

Some established geophysicists insist that all earthquakes are random, yet everyone agrees that aftershocks are not. Instead, they follow certain empirical laws. Stein, Lin, and their collaborators hypothesized that many earthquakes classified as main shocks are actually aftershocks, and they went looking for the forces that cause faults to fail.

Their work was in some ways heretical: For a long time, earthquakes were thought to release only the stress immediately around them; an earthquake that happened in one place would decrease the possibility of another happening nearby. But that didn’t explain earthquake sequences like the one that rumbled through the desert and mountains east of Los Angeles in 1992. The series began on April 23 with a 6.2 near the town of Joshua Tree; two months later, on June 28, a 7.3 struck less than 15 miles away in the desert town of Landers. Three and a half hours after that, a 6.5 hit the town of Big Bear, in the mountains overlooking the Mojave. The Big Bear quake was timed like an aftershock, except it was too far off the Landers earthquake’s fault rupture. When Lin, Stein, and Geoffrey King of the Paris Geophysical Institute got together to analyze it, they decided to ignore the distance rule and treat it just as a different kind of aftershock. Their ensuing report, “Static Stress Changes and the Triggering of Earthquakes,” became one of the decade’s most-cited earthquake research papers.

Rocks can be subject to two kinds of stresses: the “clamping” stress that pushes them together, and the “shear” stress they undergo as they slide past each other. Together, these stresses are known as Coulomb stress, named for Charles-Augustin de Coulomb, an 18th-century French physicist. Coulomb calculations had been used for years in engineering, to find the failure points of various building materials, but they’d never been applied properly to faults. It turned out, though, that faults in the ground behave much like rocks in the laboratory: they come unglued when shear stress exceeds the friction and pressure (the clamping stress) holding them together. When Stein, Lin, and King applied the Coulomb model to the California sequence, they found that most of the earthquakes had occurred in areas where the shifting of the ground had caused increased stress.

In 1997, Stein and two other geologists using the model found that there was a 12 percent chance that a magnitude 7 or greater would hit near Izmit, Turkey, within 30 years; two years later, on August 17, 1999, a magnitude 7.4 destroyed the city, which wasn’t designed to withstand such a tremor. A Turkish geologist named Aykut Barka quickly wrote up a paper warning that Coulomb stress from the Izmit quake could trigger a similar rupture near Düzce, a town roughly 60 miles east. His work persuaded authorities there to close school buildings damaged during the Izmit shaking. On November 12, a segment of the North Anatolian Fault gave way, in a magnitude 7.2. The empty school buildings collapsed.

Lin and Stein both admit that Coulomb stress doesn’t explain all earthquakes. Indeed, some geophysicists, like Karen Felzer, of the U.S. Geological Survey, think their hypothesis gives short shrift to the impact that dynamic stress—the actual rattling of a quake in motion—has on neighboring faults.

In the aftermath of the disastrous March 11 Tōhoku quake, both camps are looking at its well-monitored aftershocks (including several within 100 miles of Tokyo) for answers. Intriguingly, it was preceded by a flurry of earthquakes, one as large as magnitude 7.2, that may have been foreshocks, although no one thought so at the time; the researchers are trying to determine what those early quakes meant.

When I ask Lin whether California, where I live, is next, he laughs. “I understand that the public now thinks that we’ve entered a global earthquake cluster. Even my own mother in China thinks that. But there’s no scientific evidence whatsoever to suggest that the earthquake in New Zealand triggered the earthquake in Japan, or Japan will trigger one in California.” Still, Lin and his colleagues do wonder whether Tōhoku has pushed neighboring faults closer to rupture. “I am particularly interested in how this earthquake might have changed the potential of future earthquakes to the south, even closer to Tokyo,” Lin tells me. “There, even a much smaller earthquake could be devastating.”