Under the central part of the San Andreas fault, the deep seismic whisperings known as tremors have increased after two recent large earthquakes, and a surge in tremors appears to have foreshadowed one of them, according to a study to be published today in the journal Science.
"It reaffirms the need to be ready," said Robert Nadeau, a research seismologist at UC Berkeley who led the study. "The San Andreas fault is changing down deep and it's changing down deep in places where large earthquakes have happened in the past."
Among the findings was an unusually strong tremor episode three weeks before the magnitude-6.0 Parkfield earthquake in 2004. If these types of signals are found before other large earthquakes, they could provide a kind of early warning, said Greg Beroza, a seismologist at Stanford University who was not involved in the study.
"There have been plenty of tremor episodes that have not triggered earthquakes in other places," he said. "This one might make the strongest case."
Predicting large quakes with precision is the elusive Holy Grail of seismology. Scientists have only been able to calculate probabilities for quakes in certain areas by analyzing a timeline of ruptures and calculating the amount of stress building on a fault.
The latest study may potentially inch us closer to having an actual predictor, scientists said.
Earthquakes typically generate clear seismic waves with sharp onsets, tailing off after a minute or two, seismologists said. Tremors vibrate quietly and can continue for days. Tremors also tend to happen in a deeper, softer part of the Earth's crust, rather than in the upper part typically thought to generate earthquakes.
Seismologists used to ignore tremors because they looked like noise in the data caused by wind or cars. Until recently, scientists also had difficulty storing the enormous amounts of data required to detect tremors.
About 10 years ago, Japanese seismologists discovered deep tremors when they took a closer look at the background noise. Since then, scientists have detected tremors in the Pacific Northwest and below the San Andreas.
Nadeau's study focused on the San Andreas fault in the Parkfield region, about 170 miles northwest of Los Angeles.
Nadeau and graduate student Aurelie Guilhem combed through seismometer data from July 2001 to February 2009. That period included two strong earthquakes: the magnitude-6.5 San Simeon earthquake of 2003 and the Parkfield temblor the following year.
Even though the San Simeon quake originated on a fault about 40 miles away from the San Andreas fault in Parkfield, the quake appeared to set off stress changes under the San Andreas, Nadeau said.
The tremor activity in the Parkfield area 45 days after the San Simeon quake more than doubled compared with the 45 days before the quake.
About eight months later, there was a five-day spike of tremors about 10 times more active than the average before the San Simeon quake, Nadeau found. The tremors quieted, but the Parkfield earthquake hit about three weeks later on the San Andreas.
This surge can be considered a "foretremor" of the Parkfield quake because the events were close together, Nadeau said. But, he added, "we don't have enough resolution to say there's a mechanical connection between the two."
The Parkfield quake appeared to stimulate the tremors again, raising the activity level to about six times the level before the San Simeon quake. Since then, tremor activity has remained high, varying from 1.5 to two times the level before the San Simeon quake.
Scientists still do not understand the exact relationship between tremors and quakes, Nadeau said. But more tremors probably means more stress on the fault, he said, and that suggests "an increased probability of an earthquake coming sooner rather than later."
Southern California has a huge stake in watching this part of the San Andreas fault, said Susan Hough, a seismologist at the U.S. Geological Survey in Pasadena who was not involved in Nadeau's study.
While the 2004 Parkfield earthquake ruptured northward away from Los Angeles, the massive Fort Tejon earthquake in 1857, estimated to have a magnitude of 7.9, appeared to start just south of Parkfield and roar down to the Cajon Pass, Hough said.
That quake sent long, rolling ground motions into the San Fernando Valley and collapsed a number of houses, she said. It was preceded by two moderate quakes in the Parkfield area.
"We don't know for sure, but there's pretty good evidence that the last Big One started up there," Hough said. "So it's always been in the back of people's minds that maybe a moderate quake at Parkfield could be the foreshock to the next Big One."