Expert Says He Predicted 4 Quakes on San Andreas

A respected earthquake authority reported Wednesday that he successfully predicted four moderate earthquakes on the San Andreas fault in Central California between October and January, raising the possibility that researchers may have discovered a way to give advance notice of the devastating shudders in the Earth.

In research published in the journal Nature today, University of Wisconsin geologist Clifford Thurber said small, distinctive movements in the Earth’s crust known as “creep events” could be used to give up to five days notice of moderately large earthquakes along some stretches of the San Andreas.

In addition to the four quakes he predicted, he said, he also detected the telltale seismic signature in an analysis of records of local earthquakes during the past 25 years.

If his work is confirmed on other faults, it might bring scientists closer to the long-sought Holy Grail of seismology--a reliable earthquake forecast.

Several experts said they are intrigued but skeptical about the newest effort.

Historically, there has been no shortage of theories or inspired guesswork attempting to predict the time and location of earthquakes. Some researchers have been tantalized by electrical signals, fluctuations in the Earth’s natural magnetic field or chemical changes in ground water that appear to precede a quake. Others have investigated folklore about dropping levels of well water or an increase in the number of lost-pet reports as precursors to major quakes.

But scientists so far have been unable to find any reliable formula on which to base a quake forecast. The forces involved are too complex, too poorly understood and buried too deeply within the Earth to analyze completely.

Thurber developed his theory based on a study of small earthquakes along a 50-kilometer stretch of the San Andreas between the towns of San Juan Bautista and San Benito. There, the fault winds through the rugged terrain of the Diablo Range and the Gabilan Mountains.

In the days before moderate quakes, he noticed a sudden jump in the slippage along the fault near the surface, with the rate of “creep” briefly increasing by a factor of 10 as recorded by sensitive seismographs in the area. The infinitesimal fault movement characteristically speeded up from a few thousandths of an inch per day to almost one-hundredth of an inch per day for several hours and then just as quickly subsided.

While such movements are too small to be noticed by a casual observer, they leave a recognizable spike on recordings by “creep meters,” as geophysicists sometimes call the seismographs.

“This was astonishing to me,” Thurber said. “I saw the creep event and a couple of days later there was an earthquake. I became addicted to watching the creep meters.”

When he analyzed the records of all the local seismic events between 1970 and 1994, he quickly confirmed the pattern. During years of especially high seismic activity, there was a distinctive creep event in the five days before more than half the earthquakes greater than magnitude 3, he reported.

“Creep events might be precursory at times when a significant portion of the fault segment is near failure,” he said.

He used that pattern to successfully predict a series of quakes between October 1995 and January 1996, he said. He made five predictions, notifying a range of state and federal earthquake researchers of the time, location and approximate strength of the expected temblor.

Four turned out to be accurate, he reported in Nature. The fifth was a “near miss,” he said, because the earthquake occurred 16 hours after his five-day deadline expired.

There were also two moderate earthquakes during the same period that were not preceded by any distinctive seismic signature.

“There are caveats,” Thurber cautioned. “In terms of predicting earthquakes, it is less than ideal, but it is a start.”

Not all faults display such slippage near the surface where it can be easily monitored, Thurber acknowledged. The technique, even if it proves reliable, would not have been able to predict the occurrence of an earthquake on a deeply buried fault, like the 1994 Northridge earthquake. But it eventually could be useful in monitoring such major seismic hazards as the Hayward fault, which underlies the heavily populated Bay Area.

Earthquake experts uniformly shy away from any promise of reliable quake predictions. Nonetheless, several experts said Wednesday that Thurber’s work was “reasonable” and agreed that the theory should be taken to its next logical step--a study leading to an official prediction of the date of a magnitude 6.0 earthquake that researchers have long expected to strike the town of Parkfield, Calif.

The Parkfield segment of the San Andreas fault may have more seismic instruments in place than any other such site in the world. In 1985, it was the scene of the first official earthquake prediction by the U.S. Geological Survey: that a temblor between magnitude 5.5 and 6.3 would occur there before 1993. The experts are still waiting.

Because geophysicists are so confident that a major earthquake will occur there soon, the area has become a laboratory for testing all earthquake prediction techniques.

Hiroo Kanamori, director of the Caltech Seismological Laboratory, said Thurber’s work “is very useful for understanding the physics of earthquakes, but you can’t really say you can predict earthquakes by just monitoring creep events. It would be hard to use this for definitive short-term prediction. There would be a fair number of false alarms.”

James C. Savage, a geophysicist at the U.S. Geological Survey in Menlo Park, Calif., was equally cautious. “If you have a lot of earthquakes and a lot of creep events, you are going to have coincidences. I don’t think it is conclusive, but it is interesting.

“It is going to take more investigation before one can place confidence in it,” he said.