PARKFIELD keeps SECRETS
after a LONG-AWAITED QUAKE.
By: SCIENCE Oct. 2004
But they didn’t get it entirely right. When seismologists began the Parkfield Earthquake Prediction Experiment in the 1980s, they expected to capture the next magnitude 6 in unprecedented detail within a few years. Instead, they had to wait 2 decades, a delay that casts additional doubt on models of predictable seismic behavior. And far from providing practical experience in the nascent science of short- term earthquake prediction, Parkfield 2004 seems to have given no warning that would lend hope to the field of short-term quake forecasting. All in all, Parkfield has driven home the point that even one of the world’s best behaved fault segments can be pretty cantankerous.
Twenty years ago, the 25-kilometer section of the San Andreas fault that runs under the town of Parkfield, California seemed like a model seismic citizen. Earthquakes of about magnitude 6, noted two (USGS) seismologists, had ruptured the same Parkfield segment of the San Andreas n 1857, 1881, 1901, 1922, 1934, and 1966. The average of 22 years between recurrences seemed reliable enough (after rationalizing 1934’s “early” arrival), so the next quake in the~series should arrive in 1988, give or take 5 years. The National Earthquake Prediction Evaluation Council, a federal committee advising the (USGS) director, had concurred with that long-term forecast.
But the accuracy of that “give-or-take” forecast had long ago come into question.
Now, 16 years after the forecast’s most probable date, official quake forecasts say the likelihood of the next Parkfield quake occurring in 2004 was just 5% to 10%.. The delay only reinforces the idea that “earthquake reoccurrence is less regular than had been hoped,” says seismologist William Ellsworth of the (USGS) in Menlo Park. “There are real practical limits to the type of forecast we made at Parkfield.”
The limits of quake forecasting became clearer still when seismologists looked at the magnitude-6.0 event on 28 September, which caused little damage to the sparsely populated region 75 kilometers inland from the coast. Seismologist Ross Stein of (USGS) Menlo Park recalls a number of I 980s ideas about quakes that would have favored predictability. They included the idea that quakes could recur with some regularity; that the more time a fault had to build up strain, the larger the eventual quake would be; and that the same fault segment would rupture in the same characteristic quake the same magnitude and same section of fault—each time.
Of these and other optimistic quake ideas, “the only one still alive at Parkfield is the characteristic earthquake,” says Stein. The quakes timing certainly wasn’t regular. And to judge by the amount of fault strain accumulated in the intervening 38 years, Parkfield 2004 should have released 20 times the energy that it did and have been a magnitude 6.7. Even the characteristic aspect does not hold up in detail, Stein notes. The same 25 kilometers of fault broke as in 1966 and 1934, producing a similar-magnitude quake. But in 2004 the rupture started at the south-east end of the segment and ran northwest ward, the opposite direction from those that struck in ‘34 and ‘66. Parkfield earthquakes—once considered among the most regular of quakes—~are certainly not peas in a pod,” observes Menlo Park’s Johnston.
Unfortunately for the prediction experiment at Parkfield, the individuality of quakes there extended to geophysical activity before the main shock, activity that seismologists once hoped couldbe used to predict the main event. The 1966 Parkfield main shock was preceded by a number of possible and even certain precursors. They include a flurry of micro-earthquakes 2 to 3 months before, cracks in the ground along the fault at least 11 days prior, and a magnitude-5.l foreshock 17 minutes ahead of the main shock. A magnitude-5 foreshock preceded the 1934 Parkfield quake by 17 minutes as well.
Nothing obvious heralded the 2004 Parkfield quake. “At the moment, nothing has jumped off the screen,” says Ellsworth. A vastly improved seismometer network at Parkfield detected no fore-shocks down to magnitude 0, says Robert Nadeau of the University of California, Berkeley. (Magnitudes can be even smaller and negative.) Johnston reports nothing obvious from the dense networks of creepmeters, magnetometers, and strain-meters scattered along the fault. The only possible precursor being discussed is a slow, subtle straining around the fault that began on 27 September. Johnston thinks it may be the long-sought signature of a main shock’s very beginnings, so-called nucleation. Colleagues are reserving judgment. Despite all the disappointments, seismologists haven’t lost faith in their quest to understand how earthquakes behave. “The [Geological Survey bet the farm, lost, was humbled, but stuck it out,” says Stein. “In the end, it was the right choice.”
Earthquake prediction aside, the recording of strong ground shaking in unpreced-ented detail creates a great opportunity to learn how to build safer, more quake- resistant buildings, says engineering seismologist Anthony Shakal of the California Geological Survey in Sacramento. “Our science advances on the basis of great data,” adds Stein, and that is what they got.
—RICHARD A. KERR
8 OCTOBER 2004, VOL. 306 (pg. 206)
Church of the Science of God
La Jolla, California 92038-3131
© Church of the Science of GOD, 1993