A Puzzle Deep Down


Pity the poor scientist who takes on earthquakes as a calling. Any other natural disaster is a piece of cake by comparison. Hurricanes are tracked from birth. Floods can be anticipated. Progress has been made even in predicting tornadoes. Not earthquakes.

Take a drive out to the San Andreas fault and the most you will feel is a sense of awe at the fact that you are standing on one of the most dangerous borders on Earth, with more destructive power locked into the rock than any nation has ever massed against a foe. The twisted fault line, between the North American and Pacific tectonic plates, reveals nothing else.

Russians and the Chinese claim an ability to predict earthquakes. Most scientists don’t believe it. A team of experts, including a Caltech official and Robert Geller, the chief seismologist at Tokyo University, were even more blunt in a 1997 article in the magazine Science entitled “Earthquakes Cannot Be Predicted.”


Why? The playing fields where most big quakes originate are inaccessible, miles below humankind’s deepest underground exploration. But the pressure for forecasts and for some kind of early warning system continues. The scientists strive, perhaps too hard.

Just a year after the magnitude 6.7 Northridge earthquake in 1994, for example, a team of scientists from Caltech and other renowned research centers released their “earthquake deficit” forecast. Too much seismic strain had built up in the region and the result would be a series of Northridge-size quakes or a smaller number in the range of 7.2 to 7.5.

By March 1998, however, that theory had vanished beneath a spate of technical revisions and acknowledged miscalculations. In fact, some of the “deficit” theorists admitted succumbing to pressure to appear more definitive in public than the evidence suggested. “We have no clue what the next 30 years will bring,” said one geophysicist.

The problem, which ought to be acknowledged far more often, is that earthquake study is still in the discovery phase. Scientists only recently uncovered the phenomenon of nearly imperceptible and slow earthquakes that take place over hours or even days. And it was just last summer that they discovered the first direct evidence of magnitude 7.6 quakes near what is now Los Angeles. Then last week scientists using data from decades-old oil company drilling maps revealed a 25-mile-long fault system capable of magnitude 7.0 quakes running from downtown Los Angeles to northern Orange County.

The best that scientists can do for the foreseeable future is to flood the region with sensors to find as many hidden faults as possible, then use the discoveries to complete a better map of seismic hazards. Meanwhile, one of the most important and expensive earthquake prediction experiments continues in Parkfield, in Monterey County. Scientists said that there was a 95% probability that a magnitude 6.0 quake would hit there by 1993. They are still waiting for it.