Tuning in to Earthquake Warnings

Coincidence is a bane for scientists. When two apparently unrelated events occur in such a way that they imply connection, it's awfully easy to jump to the wrong conclusion. Yet fear of making the wrong jump may lead researchers to reject a valuable new idea.

Consider an odd detail reported after last year's Loma Prieta earthquake. That earthquake's source, the San Andreas fault, is one of the most studied geological features in the world. Laser beams shoot across it, measuring tiny increments in southern California's slow voyage toward Alaska Tiltmeters dot its flanks, registering minuscule bulges and subsidences in the stressed rock sensitive seismometers surround it, noting its every twitch.

Still, on October 17, 1989, geophysicists were as surprised as everyone else when a 32 kilometer-long section of the San Andreas suddenly broke free. Their studies had given them fair confidence that the section would break sometime within the next 30 years. That particular afternoon, none of their high-tech widgetry gave any hint that something dramatic was afoot. The most bitter disappointment for the experts was their instruments' inability to detect any deformation whatsoever of the rock surrounding the fault just before the quake hit. Well-developed theory, supported by convincing laboratory experiments, had led them to believe that a highly likely precursor to a major earthquake would be a perceptible increase in crustal distortion. The unceasing northwestward creep of the crust along the Pacific side of the fault stresses the jammed rock, which distorts in response. The expectation was that the distortion would accelerate in the days or weeks before the fault failed, and the instruments would surely detect that change.

The instruments were ready for the October 17 event. Geophysicists had placed an array of them in boreholes along the fault; the devices are so sensitive that they can measure deformation to an accuracy of one part in 100 billion. One instrumented borehole lay only 35 kilometers from the epicenter of the Loma Prieta quake, just 10 kilometers beyond the southern tip of the eventual rupture on the fault. It was closer than any such instrument had ever been to a major earthquake. It gave no warning, nor did any of its fellows.

While researchers with the U.S. Geological Survey and other organizations sifted their data, trying to find a useful pattern (or at least evidence that the theory could be saved), some electrical engineers from Stanford University were trying to explain an odd glitch that appeared in their records.

Anthony Fraser-Smith and his colleagues were going through data collected by their ultralow-frequency radio receiver. It was part of a study on sun-related radio noise that could interfere with submarine communications. They had found a site for their receiver well protected from San Francisco area radio frequency pollution. The site turned out to be just seven kilometers from the earthquake epicenter.

The receiver survived the earthquake undamaged, and its data were recovered in fine shape. When Fraser-Smith's team reviewed the information they had gathered, they found a spurt of unusual radio background noise. Starting on October 5, the noise at all frequencies in the range from 0.01 to 10 hertz (cycles per second) went up. The noise declined about a day before the quake, then shot up to 30 times its previous level during the three hours just before the event.

The Stanford engineers had never seen anything like that anomalous record during their two years of monitoring. They passed on the information to researchers in earthquake prediction--who hardly know what to do with it. (It's hard not to imagine the conversation: "Well, er, we've got this anomaly, and it's probably just coincidence, but it sure is strange, and--here. It's all yours. Have fun.")

Other sorts of peculiar events, from howling dogs to surging water in wells, have preceded earthquakes in the past; all have proved unreliable predictors. No one knows yet whether the radio outburst was the first record of a genuine and reliable earthquake precursor, was unique to this one earthquake, or was somehow a mere coincidence. For now, the problem is back in the laps of theoreticians looking for reasons to stay tuned to Earth's unexpected broadcasts.