Deep Rumblings May Have Been Tip to ’83 Quake

Scientists have learned that a major earthquake in the Sea of Japan in 1983 was preceded for several months by deep rumblings within the Earth, a finding that offers some hope that at least some earthquakes might be predictable.

Records from the 7.7 quake, which have recently been re-examined, show periodic and measurable variations in strain in rocks near the fault before the quake struck. If that pattern is repeated elsewhere, it might mean that such events could serve as a warning of an impending earthquake.

On the other hand, other fault zones may not behave the same way, according to one of the scientists, Alan T. Linde of the Carnegie Institute in Washington.

“Obviously, we think we are right” in believing that the rumblings from rocks grinding against each other were precursors to the quake, Linde said in a telephone interview. “But we could easily be wrong.”

Some seismologists believe that most earthquakes are preceded by changes in the strain in rocks near the fault, but others believe major quakes result from the sudden release of energy along a locked fault zone and thus would be less likely to foreshadow their future. Linde’s research, reported today in the British journal Nature, could add to that controversy, especially if similar results can be obtained elsewhere.

Such evidence is hard to come by, however, in that the instruments must be placed in wells drilled near the fault before the temblor hits, and no one knows for sure where quakes will hit over the next few years. Linde has placed instruments near Parkfield, a tiny farming community east of San Luis Obispo, where scientists expect a moderate quake to hit at any time, based on what appears to be a periodic pattern of quakes on that segment of the San Andreas Fault.

That expectation has led many scientists to install a wide range of instruments at Parkfield, making it the most intensely instrumented earthquake zone in the world. But even if Linde gets the same results at Parkfield that he found in Japan, the finding may not apply to most other areas. The geophysical mechanics that cause earthquakes vary widely around the world, making it extremely difficult to determine universal causes for quakes.

“If we get an answer in one place, it may not apply someplace else,” Linde said.

Working with scientists in Japan, Linde analyzed records from three bore holes in the region hit by the May 26, 1983, quake. None of the wells were as close to the fault as Linde would have liked, the closest being within about 55 miles.

What the records show, Linde said, is that instruments in the closest well showed about 100 periodic changes in the strain in the rocks during a five-month period preceding the quake.

“These were episodic, a three-hour blip and then nothing, followed by a three-hour blip and then nothing again,” he said.

The scientists believe the episodes show “a relatively rapid stress concentration” building up in the fault zone before the quake hit. That finding is consistent with the fact that the more distant instruments did not detect increases in strain, Linde said, because strain should decrease proportional to the distance from the fault.

Linde, however, is reluctant to suggest that the evidence is of major importance.

“We’re still fairly ignorant about what the Earth does before an earthquake,” he said. “The problem is we are observationists, not experimentalists. We have to sit and wait for the Earth to tell us something.”