Scientists Debate Feasibility of Earthquake Prediction : Seismology: More than 100 at Irvine conference vent frustrations and hopes about efforts to find accurate precursors. Arguments are made to go on with the search.

At a conference here a few days ago to assess the feasibility of making accurate earthquake predictions, a Japanese scientist said he might have detected a precursor to the Kobe earthquake.

If his suggestion were valid and precursors could be demonstrated as occurring consistently before other temblors, it would be a spectacular discovery, opening the way to quake forecasts.

But in the real world of seismology, quake prediction is still a dream. His theory was quickly challenged--and quashed--by other scientists at the meeting.

Hiroshi Wakita, of Tokyo University’s Laboratory for Earthquake Chemistry, touted his claim of a precursor by showing a graph on chemical changes in ground water recorded by an instrument at a manufacturing plant on Rokko Island, just off Kobe. It showed a fairly steady rise in chlorine readings in the 10 months preceding the Jan. 17 earthquake.

During that time, the chlorine level rose from 13.3 p.p.m. of water to 15.4. Then, when the quake struck, the reading immediately sank to its earlier level.

Unfortunately, though intriguing, this was not the long-hoped-for reliable precursor to a strong quake, said seismologist Lucile M. Jones of the U.S. Geological Survey.

“We would need 20 years of such readings, and then we would have to see how many times such a rise occurs, and how many times it is followed by an earthquake, and how many times it is not,” Jones explained.

In fact, Jones declared in her own presentation, so far no precursors have been discovered that are reliable enough to allow firm short-term earthquake predictions.

But Max Wyss of the Geophysical Institute at the University of Alaska, Fairbanks, responded by listing five occasional precursors--including foreshocks, pre-shocks and ground water changes--that had been recognized at a recent meeting of the International Assn. of Seismology and Physics of the Earth’s Interior.

The only trouble, Wyss acknowledged, is they do not occur all of the time.

Such exchanges marked the conference here. Despite occasional claims, earthquake prediction has not seemed to have gotten very far in recent years. And the more than 100 scientists who had gathered to talk about it both vented their frustrations and expressed their hopes.

The coordinator of the conference, sponsored by the National Academy of Sciences at its Beckman Center here, was UCLA physicist Leon Knopoff, a proponent of pressing ahead with prediction studies. He invited critics of such exercises, as well as boosters, to make presentations.

Hiroo Kanamori, director of Caltech’s Seismological Laboratory, said he is skeptical.

The results of research to date “demonstrate that reliable short-term earthquake prediction . . . is difficult,” he said. “Precursory phenomena such as foreshocks and non-seismological anomalies may occur, but their behavior does not seem to be consistent enough to allow a reliable and accurate short-term prediction.”

*

He noted too that even if quakes could be predicted, significant details about them could not be. “It would have been almost impossible to predict that the Landers earthquake (rupture) would go from south to north. If it had been north to south, it might have triggered a quake on the San Andreas fault.”

Kanamori said, “Everybody in Southern California knows there will be a big earthquake. Can we tell them more?”

But Knopoff had a rebuttal for Kanamori: “Just because you don’t find a cure for cancer immediately doesn’t mean you stop working on it,” he said.

Frank Press, former president of the National Academy of Sciences, was philosophical about the prediction dilemma.

Press said that the French marshal, Louis-Hubert Lyautey, once told an assistant to plant a tree the next morning. “But,” the aide remonstrated, “this tree won’t bear fruit for 100 years.”

“In that case, you’d better plant it this afternoon,” the marshal replied.

Press said an informal survey he took among the scientists attending the conference indicated that most believed it would take from 20 to 100 years before the means of making accurate quake predictions were discovered.

But, even so, he said, too much is at stake not to try, including “thousands of lives and tens of billions of dollars in damages.”

Much of the discussion was devoted to intricate examination of instrumental records indicating what happens before a quake. Various phenomena--from changes in electric conduction near coastal faults to variations in crustal strain--were debated.

Even prediction enthusiasts often acknowledged that the earthquake process too often seems “chaotic” or “incoherent.”

Prediction adherent Lynn R. Sykes of the Lamont-Doherty Geological Observatory at Columbia University argued that even if correct short-term predictions are impossible for now, perhaps intermediate-term or general predictions might have their own value.

“To present prediction as only short-term invites lack of progress in the future and does not take into account . . . (the) chance to make modest progress during the next 20 years,” Sykes contended.

He suggested that if a seven- or eight-year prediction could be made for a segment of the San Andreas fault, certain broad mitigation preparations could be made.

In California, a series of quakes in the magnitude 5 or 6 range occurred in decades leading up to both the 1906 San Francisco quake and the 1989 Loma Prieta quake. It is possible the series of quakes along the San Gabriel and Santa Susana mountains that has been occurring in Southern California since 1971 might point the way to a larger quake too, Sykes said.

But Kanamori and Jones said they feared the outside world will not have much appreciation for anything called prediction that is not short-term.

And even the conference organizer, Knopoff, questioned a recent probability assessment by leading seismologists that there is an 86% chance of a major earthquake in Southern California in the next 30 years.

“They should be questioning their assumptions,” he said, in light of historic records indicating that big quakes are often clustered within short periods, with long periods of nothing happening in between.