'Clockwork' quakes may not keep good time.

'Clockwork' quakes may not keep good time

Parkfield, Calif., has a reputation for moderate earthquakes that strike with clock-like regularity. Expecting the next such quake by 1993, the U.S. Geological Survey (USGS) has installed millions of dollars' worth of quake-monitoring equipment there. This week, however, two groups of geoscientists reported evidence suggesting Parkfield's quakes are less regular than believed, raising doubts about whether the forecast will hold.

"I'm a little less optimistic [than I used to be] for it happening in the next several years," says USGS geophysicist Paul Segall of Menlo Park, Calif., who led one of the new studies.

In 1985, two USGS scientists gave 20-to-1 odds that a magnitude 6 shock would occur before 1993 on the San Andreas fault in Parkfield, a sparsely populated farming town about halfway between Los Angeles and San Francisco. The prediction stemmed from the theory that some portions of the San Andreas rupture repeatedly in "characteristic" earthquakes. In Parkfield, earthquakes of magnitude 5.5 or 6 were known to have struck in 1857, 1881, 1901, 1922, 1934 and 1966. Aside from the 1934 event, this shock sequence suggested an apparently regular period of 21 or 22 years. The 1934 and 1966 quakes had the same epicenter and magnitude, adding weight to the notion of Parkfield's characteristic quakes.

Believing another shock was due soon, the USGS in 1985 set up its Parkfield Prediction Experiment in an attempt to study the earthquake process in unprecedented detail and to issue a prediction days or hours before the quake.

Now, evidence culled from thousands of old newspapers, diaries and letters challenges the assumption that Parkfield quakes occur regularly. Tousson R. Toppozada and his colleagues at the California Division of Mines and Geology in Sacramento say their historical search has uncovered earthquakes in 1877 and 1908 that apparently originated in Parkfield and measured magnitude 5.5. These two quakes clearly do not fit in the presumed 21-year cycle. Toppozada notes. The historical records also suggest that a magnitude 5.5 aftershock followed each of the previously known earthquakes in 1901 and 1922.

Toppozada's group estimates the epicenter and size of quakes described in old documents by comparing the descriptions with well-known modern quakes. They presented their results in Santa Cruz, Calif., at a meeting of the Seismological Society of America.

The new finds suggest that Parkfield's seismic activity has diminished dramatically in recent decades. Toppozada says. If the newly described shocks did in fact originate in Parkfield and were as strong as estimated -- assertions debated by some geoscientists -- then the area experienced seven shocks of magnitude 5.5 or greater between 1870 and 1930, but only two such quakes between 1930 and 1990.

In a separate presentation, Segall and colleagues from the USGS and Stanford University described old geodetic surveys across the fault in Parkfield. The data suggest that the 1934 and 1966 ruptures didn't break identical parts of the fault -- a finding that makes Segall wonder how "characteristic" the Parkfield quakes really are.

Nonetheless, Segall maintains there is a "reasonable chance" that the forecasted earthquake will occur by 1993, and he says the need to concentrater limited equipment makes Parkfield the best place for the experiment.

But as the end of the forecast period draws closer, many observers, including some scientists, have started grumbling that Parkfield has diverted equipment and attention from more dangerous parts of the San Andreas, says Evelyn A. Roeloffs, chief scientist of the Parkfield Prediction Experiment.

In a broader sense, the new evidence "raises questions of whether the concept of characteristics earthquakes is a useful one," Roeloffs says. "If not, then it's going to be very hard to issue earthquake forecasts."