'Preshock' pattern may foretell quakes.

'Preshock' pattern may foretell quakes

Investigations of the Oct. 17 Loma Prieta earthquake in northern California hint at a pattern that may help scientists predict some major quakes one to several years before they strike, seismologist Karen C. McNally reported this week at a meeting of the American Geophysical Union in San Francisco. At the same session, other researchers discussed why most structures fared so well in the quake while others collapsed.

McNally, from the University of California, Santa Cruz, observed a trend in the moderate-sized shocks occurring in the year and a half before the Bay area's magnitude 7.1 quake, and noted that the same pattern preceded a magnitude 5.8 temblor in 1986 near Livermore, Calif. In both instances, after a long period of quiet, a series of progressively deeper and larger "preshocks" led up to the main shock.

"I find this an encouraging lead in our effort to track down earthquakes," McNally says. She adds, however, that much more work is needed to determine whether this progression represents a chance occurrence or a reliable sign of an impending quake.

Loma Prieta, the strongest earthquake in the Bay area since 1906, was centered beneath the Santa Cruz mountains, striking the San Andreas fault at the southernmost section of the 1906 rupture (SN: 10/28/89, p. 277). For decades the section had remained inactive. But in June 1988, a magnitude 5.1 shock broke it at a depth just shy of 14 kilometers. The next moderate temblor hit the area in August 1989, at a depth of almost 17 km. The October quake ruptured the fault at 18 km.

McNally discerned a similar pattern in three smaller shocks occurring in the months before the Livermore-area quake along the Calaveras fault. She calls the earlier quakes "preshocks" to distinguish them from the foreshocks that can appear hours or weeks before a main shock. She also notes that some quakes in Mexico and Costa Rica apparently have followed the deepening preshock pattern.

Geoscientists say the Loma Prieta quake verified the reliability of their techniques for making rough forecasts several decades before a main shock. McNally's observations now suggest a method for intermediate-term predictions: watching for quiescence followed by progressively deeper and larger preshocks. Data from Loma Prieta, however, have not offered hope for making predictions just weeks before a large quake.

William H. Prescott of the U.S. Geological Survey in Menlo Park, Calif., reports that several instruments near the fault detected no precursory signs of the main shock. One such instrument was a dilatometer, which measures strain in the ground and can detect changes one-thousandth the strength of those occurring during the main shock. Located 35 km from Loma Prieta's center, the device may have been too distant to pick up early signals, he suggests. Prescott notes that researchers are now conducting experiments in Parkfield, Calif., to learn what kinds of short-term signals might precede major quakes. Parkfield sits on a San Andreas segment that ruptures at regular intervals, leading scientists to predict a magnitude 6 temblor there within the next three years.

In terms of structural damage, engineers and geologists maintain Loma Prieta held few surprises. As in the 1906 quake, areas on landfills and soft soil shook the hardest. Overall, most buildings fared well, in part because of improved building codes and the relatively isolated location of the quake's epicenter. But peculiar aspects of the shaking also help explain the limited damage, researchers say. The fault broke quickly, producing a shaking that lasted only 6 seconds near the epicenter, says Hiroo Kanamori of the California Institute of Technology in Pasadena. In contrast, last year's magnitude 6.9 earthquake in Armenia shook for 30 to 40 seconds, he says. Usually, the longer the shaking, the more damage wrought. So, although many buildings withstood Loma Prieta, engineers caution that this is no proof they could survive another 7.1 temblor centered just as far away.