Stanford Scientist Sees Possible Quake Predictor

For the second time, Antony Fraser-Smith has picked up a sharp electromagnetic signal a month before a California earthquake, fueling hope that some quakes may be predictable.

The first time was in 1989, before the magnitude 7.1 Loma Prieta quake struck near San Francisco, killing 63. The signal persisted for two months afterward. The second time was a year ago, before a more benign magnitude 5 quake at Parkfield, about halfway between San Francisco and Los Angeles on the San Andreas Fault. The signal persisted for more than a week.

Explaining his results at the American Geophysical Union fall meeting here, the Stanford University professor of geophysics and electrical engineering said he was noticeably wary of proclaiming that he’s found the elusive quake precursor that scientists have long sought.

But, he said Dec. 11, the tenfold increase in the ultra-low-frequency magnetic field signal that he detected underground makes electromagnetic signals a fertile area for additional research.

“It’s just an extra brick on this pyramid I’m trying to build up,” he said. “Right now, I’m not even saying we’re going to be able to predict [quakes] using these fields.”

Fraser-Smith believes that the signals are generated deep in the earth before the faults actually rupture. He theorizes that, as pressure builds on a fault, cracks open up and water trapped in the earth is pushed through them, creating a flow of electric currents that produces a magnetic field. The water flowing along the crack of the fault acts like an underground antenna.

Fraser-Smith’s latest result came from the magnitude 5 quake at 2:27 a.m. on Dec. 20, 1994. It registered at two monitoring sites just to the north and just to the south of Parkfield.

Parkfield is the site of a nationally sponsored quake-prediction experiment that has yet to yield the magnitude 6 quake that has been expected since 1988. A large quake hits the area about every 22 years.

Fraser-Smith has instruments lying in wait at Parkfield, in Southern California and in Northern California.

“I am fishing for an earthquake,” he joked. “I’m sort of sitting with my bait out. Our biggest problem has been gophers gnawing on our fancy, high-tech cables.”

He was lucky the last time: His instruments were practically sitting atop the earthquake about three miles below. Just two years ago, his instruments picked up a signal change, but no quake followed. Later, he and his colleagues showed that the signal came from outside, not from underground.

The Stanford scientist notes that he’s alone in monitoring magnetic field changes before quakes and that “no one else has been in a position to confirm” his findings.

In the Soviet Union, scientists are “absolutely convinced there are electromagnetic precursors to quakes,” and Greek scientists are showing successes with predictions of magnitude 6 quakes whose results he has examined.

Here in the United States, scientists remain skeptical about electric or magnetic precursor signals.

The U.S. Geological Survey, however, has been interested enough to give him some funding. And the Electric Power Research Institute is funding him to monitor parts of the Hayward Fault near San Francisco for any signals that might precede shakers there.

Alan Lindh, a U.S. Geological Survey scientist at Menlo Park and former chief scientist at the Parkfield prediction project, has been one of Fraser-Smith’s champions.

He called Fraser-Smith’s 1989 observation after the Loma Prieta earthquake “the most important glimmer of hope for short-term earthquake prediction that we have seen, period.”

Andy Michael, a survey scientist who studies Parkfield, was cautious about the few examples that Fraser-Smith has found.

“If you can identify a signal that precedes some quakes and is always followed by a quake, that’s useful, but signals that sometimes are followed by quakes are not useful,” he said. “Tony is collecting good data. Eventually, I hope that will lead to a resolution. But it’s going to take a long time, because you need to build up a large statistical sample.”