Earthquake Prediction: Research Funding On Shaky Ground?

When the field of earthquake prediction took off in the mid- 1970s, seismologists had high hopes of finding geologic warning signs that would allow them to issue timely evacuation notices. But after years of inconclusive research, only a determined few of these investigators are still engaged in the effort to predict earthquakes, as federal support shifts to general studies designed to minimize quake aftereffects. "It isn't as easy as we thought it was going to be 15 years ago," says Jim Savage, a seismologist with the Menlo Park office of the United States Geological Survey (USGS), which directs earthquake research. "I think it's possible to predict earthquakes, but it will require more basic research before we know for certain." Now, the leading earthquake prediction experiment in the United States, coordinated by USGS, which involves intense scrutiny of a single California fault, is being reviewed by a panel of independent seismologists who will advise the agency on whether the effort should continue. Many seismologists say that before focusing on prediction, more studies are needed of fracture zone dynamics, plate tectonics, and other forces thought to cause or contribute to earthquakes. But prediction advocates say their experiments are the best way to gain those basic insights, and they urge USGS not to abandon the effort simply because prediction is proving more difficult than earlier imagined. Prediction efforts got about 10 percent of USGS's $50 million budget for earthquake research in 1992. Virgil Frizzell, assistant chief of the USGS Office of Earthquakes, Volcanoes and Engineering, says funding for earthquake research is budgeted at a proposed $44 million next year, and prediction experiments will get 10 percent of that declining amount. "The term `prediction' is rather strong," Frizzell says. "We would rather talk about forecasting and probabilities that can influence public policy now." Nowadays, instead of studies aimed at evacuating cities in the nick of time, Frizzell says, USGS is producing hazard maps that show where large earthquakes are likely to occur. Such maps encourage disaster relief planning and changes in building codes to lessen the loss of life and property. Seismologists Leery "Earthquake prediction is a branch of seismology that many seismologists won't have anything to do with," says Arch Johnston, a geophysicist and director of the Center for Earthquake Research at Memphis State University. "A lot of scientists think the goal of prediction may not be approachable." The field certainly looked more promising in 1975, when Chinese scientists predicted an earthquake in Liaoning Province that led to a lifesaving evacuation of the city of Haicheng. Impressed by that success, Congress created the National Earthquake Hazard Reductions Program in 1979. It gave USGS about $30 million in earthquake research funds, with a major emphasis on the search for earthquake precursors. Back then, seismologists thought they could predict earthquakes by detecting precursor events, such as elevated levels of seismic activity, increased strain along fault zones, or changes in local electromagnetic fields. The long list of supposed earthquake precursors included falling levels of ground water and strange animal behavior, as observed by Chinese scientists at Haicheng. Today, seismologists say many of those clues have led to dead ends, and the precursors that may still be valid have proved so subtle that they are nearly impossible to detect. "The good ideas we thought we had about how to predict earthquakes have not panned out," says David Jackson, a seismologist at the University of California, Los Angeles. According to Jackson, most seismologists now believe that earthquakes are so complex and chaotic as to defy prediction, at least for now. "Part of the trouble is, we have never caught an earthquake up close to see what it is we're dealing with," Jackson says. Although seismographs and other instruments record earthquakes all the time, Jackson compares that to seeing someone across the room write in a notebook. "It is quite another thing to read over their shoulder," he says. Earthquake researchers agree they need a close-up of an earthquake before they can say whether precursors are detectable. But the earthquake community is divided over just how to get that close-up view. Tom McEvilly, a seismologist at the University of California, Berkeley, is of the minority who want to concentrate monitoring efforts--and research funds--along faults thought to have a high probability of producing earthquakes. "We don't know how to understand earthquakes without looking at them, for God's sake," says McEvilly. "We can't calculate the [fault zone] failures yet. We don't have the equations." But Seismologist Hiroo Kanamori, director of the Seismological Laboratory of the California Institute of Technology, doubts that earthquakes can be predicted with any precision. To his way of thinking, the more scientists learn about earthquakes, the more theories proliferate on their causes, with no certainty that any one theory will explain all earthquakes. Given that uncertainty, Kanamori says, monitoring equipment should be spread over wide areas, to gather data about many faults, rather than stake out a single fault and wait for it to slip. "If we have a reasonably good instrument network in Southern California, we stand a good chance of catching an event," says Kanamori. Attack On Predicition Efforts Robert Geller, a seismologist at Tokyo University, shook up the earthquake prediction community in a Nature commentary last July (352:275-6, 1991). "Earthquake prediction seems to have come to a dead end," wrote Geller, attacking a 30-year-old Japanese experiment that has concentrated seismographs and other monitoring equipment along a single fault zone, the Tokai Ridge. The experiment has collected data about ground movement and other activity. But since the fault has not yet produced an earthquake, the Japanese cannot pick earthquake precursors out of the data. Geller said too much money has been spent looking for clues that might not exist, while basic research in seismology has been shortchanged. Now similar criticisms are being directed at the American earthquake prediction program, which centers around a Japanese- style concentrated monitoring effort at the Parkfield fault in central California. Since 1985, USGS and university researchers have been monitoring the Parkfield section of the San Andreas fault with an array of costly and sensitive equipment designed to detect signs of an impending quake. USGS seismologist Alann Lindh says the Parkfield fault has caused earthquakes of magnitude 6 on the Richter scale about every 25 years going back to 1881. Based on past behavior, Lindh and other prediction experts estimated a 95 percent chance the fault would rupture again before 1993. Relying on that high probability estimate, USGS and state of California each put up $1 million to begin deploying instruments along the fault zone. But as 1993 approaches, Parkfield has not yet slipped, casting doubt on the wisdom of putting such emphasis on one site. "We put our eggs in that basket, and the earthquake has not cooperated," says Lindh. "Nobody got into this business to do safe experiments." USGS is now assembling a panel of leading seismologists to review the six-year-old Parkfield experiment. "We expect them to give us some guidance by September," says Frizzell. He estimates that $10 million in equipment and staffing has been invested in Parkfield to date. Frizzell says the review panel will help USGS decide whether to fund the effort at approximately $1 million per year, or discontinue some of the monitoring efforts. Duke University geologist Peter Malin, a member of the Parkfield experiment, says, "I fear the government is on the verge of backing off on this very critical and fruitful research." Malin says the Parkfield team has developed instruments sensitive enough to detect earthquake precursors, a breakthrough he likened to the invention of the microscope. "We finally have our microscope built," Malin says. "We're like the first guys looking down the microscope at cells." Selwyn Sacks and Alan Linde of the Carnegie Institution of Washington, D.C., developed a key instrument at Parkfield, a device called a strain meter. A stainless steel cylinder 10 feet long and five inches in diameter, the strain meter has a hydraulic system that records changes in rock pressure the way a water balloon deforms when squeezed. "Is there really a recognizable signal the Earth sends out that it's about to go bang?" Linde asks, posing the hypothesis that guides his work. "Nobody knows if we can find it, if it exists." To detect subtle pressure changes in underground rock, Sacks and Linde bored holes 200 to 400 meters deep. Inside these holes they cemented strain meters that essentially became part of the fault rock. Linde says such bore-hole strain meters can detect changes in rock pressure on the order of 1 part per billion. Similarly, seismologists buried 10 seismographs along the fault to detect micro-earthquakes that might signal a larger rupture. McEvilly says underground seismographs can detect microquakes approaching zero on the Richter scale, a sensitivity far greater than surface instruments, which contend with distance distortion and ground motion caused by wind. "We have never been able to observe rock deformation through the noise," McEvilly says. "You're not going to get to the point of making earthquake predictions without understanding the pathology of the fault zone." Other Parkfield instruments record changes in water levels, a supposed sign of water seeping into fractured rock, and fluctuations in local electromagnetic fields, which are thought to occur when fault slippage sets ionized groundwater in motion. Without disputing the sensitivity of the instruments designed for Parkfield, critics of the experiment question the wisdom of literally burying them all in one place. "Considering the very tight budgets here in seismology, maybe some of the experiments there should be discontinued," Kanamori says. Parkfield proponents say little would be saved by pulling the plug on the experiment, because the expensive equipment is already underground. They say even if the fault is late, it will eventually slip, at which point their instruments will provide the first close-up of an earthquake. Serendipitous Precursor Found But those who advocate spreading instruments around can point to a chance finding that arose during the Loma Prieta earthquake that rocked the San Francisco Bay area in 1989, killing 62 people and causing $6 billion in damage. During that event, Stanford University electronics engineer and geophysicist Antony Fraser- Smith had a magnetometer set up seven miles from the epicenter of the 6.9-magnitude event. He was using the magnetometer to measure ultra-low-frequency (ULF) radio waves, a Navy-funded project aimed at studying the use of ULF waves for submarine communications. Examining their data after Loma Prieta, Fraser-Smith's team discovered that a huge spike in ULF levels occurred three hours before the quake. "This was the biggest signal we've ever seen," he says. He published his findings in Geophysical Research Letters (17:1465-68, 1990), speculating that ULF spikes might be earthquake precursors, caused by the movement of ionized groundwater. He says Russian geophysicists found a similar ULF spike before a 6.9-magnitude earthquake in Spitak, Armenia, in 1988. "I don't want to get overly excited, because I've been burned with these sorts of intuitive leaps before," says Fraser- Smith, who adds that USGS has shown interest in his work. Meanwhile, prediction researchers are planning to spend at least $2 million to line the Hayward fault near Oakland, Calif., with instruments similar to those at Parkfield. They argue that Loma Prieta may be a precursor to a more dangerous slippage on the Hayward fault. They say the tools developed at Parkfield give them at least a chance of detecting it in advance. "People in the area care if there is a big earthquake," says Lindh. "Their lives depend on it." Christopher Scholz, a physicist studying plate tectonics at the Lamont-Doherty Geophysical Observatory in New York, thinks Parkfield-type experiments are the only way to make earthquake prediction possible. "You've got to stake out an earthquake and analyze it in retrospect," Scholz says. "However, I would hate to be one of those guys, because if nothing happens for 20 years, you've blown your career. But I'm glad they're doing what they do." Tom Abate is a freelance writer based in Brooklyn, New York.

Earthquake Prediction: Research Funding On Shaky Ground?

Interested in reading more?

Become a Member of