Earthquake prediction on shaky ground?

Earthquake predictions on shaky ground?

The U.S. Geological Survey (USGS) last week slightly increased its estimated probability of a major earthquake occurring in southern California within 30 years, while two new studies suggests that forecasting such an event may be even trickier than geologists think.

Representing the work of a dozen scientists over the past year, the USGS document estimates a 60 percent probability of an earthquake of at least 7.5 magnitude along the San Andreas fault near Los Angeles. A 1980 study had put the probability at 50 percent. The report also gives a 50 percent likelihood for a similar earthquake in the San Francisco Bay area during the next 30 years.

The USGS scientists calculated their estimates from the dates and sizes of previous ruptures along the San Andreas fault system. They noted that major quakes occur northeast of Los Angeles on average every 130 years, with the most recent in 1857. But in the July 8 SCIENCE, Gordon Jacoby and Paul Sheppard of Columbia University's Lamont-Doherty Geological Observatory in Palisades, N.Y., and Kerry Sieh of California Institute of Technology in Pasadena say the fault moves with such irregularity that earthquakes there remain too uncertain to forecast from historical records.

After measuring rings from 70 trees, Jacoby and his colleagues believe an earthquake occurred along the Mojave segment of the fault in 1812--the year of a large rupture known as the San Juan Capistrano quake (SN: 4/18/87, p.255). Previously, scientists had blamed this earthquake on a coastal fault, not on the San Andreas, since historical accounts describe coastal damage. However, the researchers found that nine trees growing within 20 meters of the fault suffered a severe shock in 1812 characteristic of a large, growth-stunting earthquake. Assuming trees are accurate historians, they say, only 45 years separate the two most recent major quakes.

In a separate, still-unpublished study, Sieh dated the last 10 earthquakes along the same segment using refined radiocarbon dating methods on sediment layers snapped apart during earthquakes. Although a simple average of the more exact numbers still spaces the temblors 131 years apart, Sieh found they occurred in bursts of two or three, followed by 200 to 330 years of calm.

To increase precision, Sieh's group used larger samples of sediment and left them in a radioactivity counter two weeks, rather than the usual 17 hours or so. Although the new radiocarbon study was mentioned in the USGS report (Sieh was also one of its authors), the data were not used in the calculations.

"We need to study some other localities that can either substantiate or conflict those data," say Randy Updike, deputy director for the USGS Earthquake Hazards Office in Reston, Va. Ironically, this means the segment of the fault traced in detail leaves the most questions.

"We probably would have been more certain [of the USGS probability estimates] not having them," says Sieh of the radiocarbon dates. "But we wouldn't have been better off." He believes that to outsmart the next Los Angeles-area quake, scientists must determine whether the nearby fault is moving in one of these clusters of earthquakes. Although he thinks the Mojave segment is most likely quiet for now, given the length of time since the last rupture, he also cautions that no one can be sure how many earthquakes make up a cluster.

James Dieterich, USGS geophysicist and an author of the agency's report, says that because the USGS "doesn't regard the difference between 50 percent and 60 percent [probability] as significant," the report raises more questions about the measurement of past earthquakes than about the prediction of future ones. Comments Dietrich: "It's more of a scientific issue than a public concern."