Nadeau and Thomas McEvilly report their findings in the July 30, 1999, issue of Science magazine. The two geophysicists, who are both with Berkeley Lab's Earth Sciences Division, have performed new analyses of high-resolution data collected in the Parkfield region of Central California since 1987. The area has been a center of seismic study because magnitude 6 shocks have occurred there at regular intervals in the past.

"Between 1857 and 1966, quakes of magnitude 6 occurred at Parkfield an average of every 22 years. Despite expectations, there hasn't been another one since 1966," Nadeau says. "However, a build-up of activity started in October of 1992 and persisted through 1994, including four events from magnitude 4.2 up to magnitude 5."

Nadeau, McEvilly, and their colleagues had previously noted clusters of repeating small earthquakes occurring at specific locations in the area, with virtually identical waveforms and very regular recurrence times; during the 1992-94 events, the recurrence times of these clusters accelerated noticeably.

"We have since found a highly organized relationship between the intervals of individual microearthquakes in clusters, the occurrence of the larger events, and changes in fault slip on the surface," Nadeau says.

The 1992-94 events and the magnitude 6 earthquakes of the previous century all started within the same region of the San Andreas Fault, a strip eight kilometers long. They resulted from sudden releases of strain built up between the rocks of the Pacific Plate to the west, which is gradually but intermittently sliding northward, and the North American plate to the east.

Slippage starts at a quake's hypocenter, typically 8 to 10 kilometers beneath the surface in the Parkfield region. In the historical magnitude 6 quakes, slippage was widespread; in the smaller 1992-94 events, slippage was localized.

At Parkfield, seismometers are placed at the bottom of boreholes 200 to 300 meters deep. "There's a lot less noise in the data than with surface seismometers, so we can detect many more quakes and smaller quakes," Nadeau says. "We can also measure earthquake vibration across a wide range of frequencies, which allows us to see much more detail in the behavior of the quakes." The timing of seismic waves arriving at the different seismometers, when compared, allows hypocenters to be pinpointed relative to one another, in three dimensions to within a few meters.

SMALL EARTHQUAKES IN EACH CLUSTER AT PARKFIELD TYPICALLY EXHIBIT SIMILAR WAVEFORMS.   THE BOTTOM SIX SEISMOGRAMS HERE ARE FROM SIX DIFFERENT EVENTS IN ONE CLUSTER, RECORDED AT 221 METERS DEPTH.  THE NOISIER WAVEFORM AT TOP WAS RECORDED ON THE SURFACE.

Over time, some two thirds of all the seismic activity in the Parkfield region has been organized into about 300 localized clusters of microearthquakes. "We identified 160 sequences within these clusters, each with three or more repeating quakes. Then we looked at how the recurrence of intervals between quakes in each sequence changed over time."

Nadeau and McEvilly hypothesized that shorter and shorter recurrence intervals indicated accelerating fault slippage; intervals that got longer meant slippage was slowing down.

"In our model, particular clusters of microquakes represent one or more 'asperities' -- small, strongly locked regions where strain repeatedly builds up and is released," Nadeau says. "The rate at which earthquakes recur on any given asperity indicates the average loading from slippage -- earthquakes that recur faster mean slippage is accelerating, and the load is being released more often." Nadeau and McEvilly have found good agreement between the seismic data and direct measurements of slippage on the surface, made by creepmeters looking across the fault.

Where large parts of the fault are locked, as in the region where the magnitude 6 quakes were centered, an increase in the repetition rate of events in specific clusters means that the strain load is building faster. When it reaches some critical level, a swarm of medium-sized quakes may dissipate the load, or a single larger event may do so.

By looking at an eleven-year collection of data, Nadeau and McEvilly were able to track a zone of accelerated slippage as it moved along the fault from northwest to southeast. When, in 1992, this traveling zone of strain reached the hypocenters of the past magnitude 6 quakes, it apparently triggered the subsequent magnitude 4-plus events as it moved through the region.

Nadeau and McEvilly suggest that these events may have occurred in response to the "pulse" of increased slip rate deep in the fault; the beginning of the pulse was detected prior to the events, but once it passed through the region, the larger quakes stopped.

The persistent, distinctive signatures of individual clusters of microearthquakes and the changes in intervals between them provide a new means of correlating measurements taken near the surface of the San Andreas Fault with slip rates from two to 10 kilometers below the surface. Brown University geophysicist Terry Tullis, in a Perspective article in the July 30 issue of Science, compares Nadeau and McEvilly's method to "a creepmeter installed across the fault at a depth of 10 km," showing "that if we look at the fault zone carefully enough we can learn things that we never expected to find."

Nadeau says that "in the Parkfield region we have found a way to use data on the recurrence of microearthquake clusters to determine slip rates at depth. It remains to be seen whether this kind of intriguing correlation exists in other fault zones, or whether it can be used over longer periods of time and space to warn us of damaging earthquakes. But preliminary results using small repeating earthquakes on the Hayward Fault in the San Francisco Bay Area are already showing promise."

Nadeau's and McEvilly's studies in the Parkfield region have been supported by the U.S. Geological Survey.

The Berkeley Lab is a U.S. Department of Energy national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California.