U.S. researchers said they were able to measure small changes in the density of fractures along the San Andreas fault as much as 10 1/2 hours before an earthquake, a step toward the long-sought goal of predicting potentially devastating quakes.
The team has so far monitored only two events over a two-month period, so a great deal more work will be necessary before their findings can be verified and understood. But the results suggest for the first time that it may be possible to forecast quakes hours before they occur -- giving residents time to prepare or, if necessary, evacuate.
Already teams in Japan and China are gearing up to test the new approach on faults in those countries, and the U.S. team is planning a much longer test to better understand the results, said seismologist Fenglin Niu of Rice University in Texas, who led the study, reported Thursday in the journal Nature.
"It's very encouraging, but we definitely need more experiments," he said.
It has been known for decades that the velocity of seismic waves through rock varies with the stress applied to the rock, presumably caused by the opening and closing of micro-cracks in the rock.
Researchers have tried several times before to exploit these changes for predictive purposes, but the instruments have been inadequate. In particular, the devices used to generate the seismic waves have not displayed sufficient repeatability to give accurate results.
Niu and his colleagues at the Carnegie Institution of Washington and the Lawrence Berkeley National Laboratory in California used a new piezoelectric source that expands when an electric current is applied and can be reproduced, creating a seismic wave, and a highly sensitive accelerometer to detect the resulting wave. The time required for a seismic wave to travel from the source to the accelerometer is a measure of the density of the rock through which it passes.
The two devices were placed in bore holes about 30 feet apart on the San Andreas fault in Parkfield, where quakes occur frequently. The devices were about half a mile underground adjacent to the fault zone.
The team operated the devices for two one-month periods in 2005 and 2006. The devices were so sensitive that they could detect changes in the barometric pressure at the surface: As the air pressure increased, the weight of the air would close micro-fractures, reducing the time for a seismic wave to travel between the source and detector by about three microseconds.
The team observed two larger changes during the monitoring periods. One occurred about 10 1/2 hours before a magnitude 3 quake on the fault, and a smaller change occurred about two hours before a magnitude 1 quake. The changes persisted until after the quakes had occurred, and the team believes that they were an indication of stress building up before the temblors.
"We hope the changes could be used to predict earthquakes, but that's only two data points," Niu said. "We need a much better understanding of the relationship between the timing [of the stress changes] and the size" of a quake.