Data from a network of seismic sensors along the Parkfield stretch of the San Andreas fault is providing important insight into the nature of earthquake cycles.
Earth scientists from LBL and UC Berkeley report that small earthquakes, which have been steadily occurring along the central 25-km stretch of the Parkfield fault zone (northeast of San Luis Obispo), are increasing in frequency and depth and are moving outside of their characteristic clusters.
"For the first time," says Thomas McEvilly of LBL's Earth Sciences Division, "we are seeing something systematic in the process that we presume is the nucleation [initiation] of a magnitude six earthquake."
At the San Andreas fault, two great slabs of the Earth's crust--the Pacific plate and the North American plate--generate seismic activity as they slip past each other. The Parkfield fault zone is the subject of a number of scientific studies because of the magnitude-six earthquakes that have occurred there regularly for more than a hundred years. Based on historical records from 1857 to 1966, the quakes typically happen every 20 to 30 years, but can come as early as 12 years or as late as 32 years after the previous one. Another magnitude-six quake is expected at Parkfield within the next few years.
Since 1987, scientists from LBL and UCB have run a precision high-resolution seismic network at Parkfield. It involves 10 boreholes, each 200 to 300 meters deep, and containing a set of three sensors cemented at the bottom. With these sensors, several thousand seismic events have been recorded over the years. The results have been analyzed at LBL's Center for Computational Seismology.
In the January 27 issue of Science, Robert Nadeau, William Foxall, and McEvilly reported their analysis of phenomena observed between 1987 and 1992. The study is part of Nadeau's dissertation.
They observed that only a small fraction of the fault zone was responding to the motion of the plates; almost two-thirds of the several thousand earthquakes recorded occurred in less than one percent of the active fault zone. The earthquakes were clustered in approximately 300 small cells, each about 20 meters across, unevenly distributed along the fault. When one cell was activated, another would go off as far away as about 200 meters, implying some sort of communication process between the two cells, such as fluid migration in the rocks.
The small earthquakes measure less than magnitude one on the Richter scale and recur with distinct regularity. "These micro-earthquakes are indistinguishable from one to the next--they're identical," McEvilly says. "Most of them occur with a very regular periodicity, of the order of one year. They're like seismic pulsars."
The regularity of the small quakes at Parkfield supports the idea that seismic activity is not chaotic.
"Here's a laboratory experiment that demonstrates that at this magnitude range repeating characteristic earthquakes do occur and at this scale are fairly predictable," McEvilly says.
In the course of the study, conducted during what is thought to be the final portion of the interval between the last big earthquake at Parkfield and the one due to come, the frequency of small earthquakes has increased steadily, from fewer than 200 earthquakes per year to 500. At the same time, they are happening deeper in the earth, and fewer of them reside in the cluster cells. According to the scientists, these changes in seismic activity may be signals of an impending magnitude-six quake.