Thursday, February 05, 2009

FORECAST: EARTH QUAKE

Fascinated by the implications of what were apparently man-made quakes, USGS scientists in 1969 set up their instruments at the Rangely oilfield in northwestern Colorado. There, Chevron was recovering oil from less productive wells by injecting water into them under great pressure. The recovery technique was setting off small quakes, the strongest near wells subjected to the greatest water pressure. If water was pumped out of the earth, the survey scientists wondered, would the quakes stop? In November 1972, they forced water into four of the Chevron wells. A series of minor quakes soon began, and did not stop until March 1973. Then the scientists pumped water out of the wells, reducing fluid pressure in the rock below. Almost immediately, earthquake activity ended. In a limited way, they had controlled an earthquake.

The results of the Rangely experiments led USGS Geophysicists Raleigh and James Dietrich to propose an ingenious scheme. They suggested drilling a row of three deep holes about 500 yds. apart, along a potentially dangerous fault. By pumping water out of the outer holes, they figured they could effectively strengthen the surrounding rock and lock the fault at each of those places. Then they would inject water into the middle hole, increasing fluid pressure in the nearby rocks and weakening them to the point of failure. A minor quake—contained between the locked areas—should result, relieving the dangerous stresses in the immediate vicinity. By repeating the procedure, the scientists could eventually relieve strains over a wide area. Other scientists feel that such experiments should be undertaken with caution, lest they trigger a large quake. Raleigh is more hopeful. In theory, he says, relatively continuous movement over the entire length of the San Andreas Fault could be maintained—and major earthquakes prevented—with a system of some 500 three-mile-deep holes evenly spaced along the fault. Estimated cost of the gigantic project: $1-$2 billion.

In a time of austerity, the possibility of such lavish financing is remote। As M.I.T.'s Press puts it: "How does one sell preventive medicine for a future affliction to Government agencies beleaguered with current illness?" Ironically, the one event that would release money for the study of earthquake prediction and control is the very disaster that scientists are trying to avert: a major quake striking a highly populated area without any warning. Tens of thousands of people living in the flood plain of the Van Norman Dam had a close call four years ago in the San Fernando Valley quake; had the tremor lasted a few more seconds, the dam might have given way. When the San Andreas Fault convulses again—as it surely must—or when another, less notorious fault elsewhere in the U.S. suddenly gives way, thousands of other Americans may not be so lucky.

The last large-scale killers occurred in Asia. One, last December in northern Pakistan, ravaged nine towns and took nearly 5,000 lives. The other, a February tremor in China, is believed to have killed hundreds. Indeed, not a day passes without earth tremors somewhere on the globe. Some of those quakes are too weak to be felt by humans; they can be detected only by sensitive seismographs. Others are more violent but occur on the ocean floor or in remote areas and do no harm. Some add to the long catalogue of destruction. Last week, for example, a 4.7 earthquake rocked lightly populated Kodiak Island, off the coast of Alaska. In July, a 6.8 quake struck Pagan, Burma, destroying or damaging half of the city's historic temples. Within the past several weeks, strong earthquakes struck Oroville, Calif., Mindanao in the Philippines, the Kamchatka Peninsula in Siberia and the southwest Pacific island of Bougainville.

With good reason, many primitive peoples regarded the terrible quakes they could not understand as the acts of a vengeful deity. As late as 1750, Thomas Sherlock, the Bishop of London, told his flock that two recent earthquakes were warnings that Londoners should atone for their sins. John Wesley agreed. In a 1777 letter to a friend, he wrote:

"There is no divine visitation which is likely to have so general an influence upon sinners as an earthquake." The ancient Japanese believed that the hundreds of quakes that shook (and still shake) their islands every year were caused by the casual movements of a great spider that carried the earth on its back. Natives of Siberia's quake-prone Kamchatka Peninsula blamed the tremors on a giant dog named Kosei tossing snow off his fur. Pythagoras, the Greek philosopher and mathematician, believed that earthquakes were caused by the dead fighting among themselves. Another ancient Greek, Aristotle, had a more scientific explanation. He contended that the earth's rumblings were the result of hot air masses trying to escape from the earth's interior.

In the past decade, the development of a bold new geological theory called plate tectonics—which offers an elegant, comprehensive explanation for continental drift, mountain building and volcanism—seems finally to have clarified the underlying cause of earthquakes। It holds that the surface of the earth consists of about a dozen giant, 70-mile-thick rock plates. Floating on the earth's semimolten mantle and propelled by as yet undetermined forces, the plates are in constant motion. Where they meet, friction sometimes temporarily locks them in place, causing stresses to build up near their edges. Eventually the rock fractures, allowing the plates to resume their motion. It is that sudden release of pent-up energy that causes earthquakes. Off Japan, for instance, the Pacific plate is thrusting under the Eurasian plate, causing the deep-seated quakes characteristic of the Japanese archipelago. In California, along the San Andreas Fault, two great plates are sliding past each other. The sliver west of the fault, which is located on the Pacific plate, is moving toward the northwest. The rest of the state is resting on the North American plate, which is moving westward. The sudden movement of a portion of the fault that had been locked in place for many years is thought to have caused the great San Francisco earthquake of 1906.

When quake centers are marked on a map of the world, it becomes clear that many earthquakes do indeed occur along plate boundaries. The earthquake-marked "ring of fire" around the Pacific Ocean, for example, neatly outlines the Pacific plate. But earthquakes can also occur well within a plate, possibly because the plate structure has been weakened in those places during periods of ancient volcanism. Charleston, S.C., for instance, is more than 1,000 miles away from the edge of the North American plate; yet it lies in a seismically active area (see map page 39) and was hit by a major quake that killed 27 people in 1886. New Madrid, Mo., near the middle of the plate, was the site of three huge quakes in 1811 and 1812. Wrote one resident of the then sparsely populated area: "The whole land was moved and waved like the waves of the sea. With the explosions and bursting of the ground, large fissures were formed, some of which closed immediately, while others were of varying widths, as much as 30 ft."

Long before the plate-tectonics theory was conceived, scientists were aware that rocks fracture only under extreme stress. As early as 1910, Johns Hopkins Geologist Harry Reid suggested that it should be possible to tell when and where quakes were likely to occur by keeping close tab on the buildup of stresses along a fault. But the knowledge, instruments and funds necessary to monitor many miles of fault line and interpret any findings simply did not exist. Earthquake prediction did not draw much attention until 1949, when a devastating quake struck the Garm region of Siberia, causing an avalanche that buried the village of Khait and killed 12,000 people. Stunned by the disaster, the Soviets organized a scientific expedition and sent it into the quake-prone area. Its mission: to discover any geologic changes—in effect, early warning signals—that might occur before future quakes. The expedition remained In Siberia far longer than anyone had expected. But it was time well spent. In 1971, at an international scientific meeting in Moscow, the Soviet scientists announced that they had achieved their goal: learned how to recognize the signs of impending quakes.

The most important signal, they said, was a change in the velocity of vibrations that pass through the earth's crust as a result of such disturbances as quakes, mining blasts or underground nuclear tests। Earth scientists have long known that tremors spread outward in two different types of seismic waves. P waves cause any rock in their path to compress and then expand in the same direction as the waves are traveling. S waves move the rock in a direction that is perpendicular to their path. Because P waves travel faster than S waves, they reach seismographs first. The Russian scientists found that the difference in the arrival times of P and S waves began to decrease markedly for days, weeks and even months before a quake. Then, shortly before the quake struck, the lead time mysteriously returned to normal. The Russians also learned that the longer the period of abnormal wave velocity before a quake, the larger the eventual tremor was likely to be.*

The implication of that information was not lost on visiting Westerners. As soon as he returned home from Moscow, Lynn Sykes, head of the seismology group of Columbia University's Lamont-Doherty Geological Observatory, urged one of his students, a young Indian doctoral candidate named Yash Aggarwal, to look for similar velocity shifts in records from Lamont-Doherty's network of seismographs in the Blue Mountain Lake region of the Adirondacks, in upper New York State, where tiny tremors occur frequently

As it happens, a swarm of small earthquakes had taken place at approximately the time of the Moscow meeting. Aggarwal's subsequent analysis bore out the Russian claims: before each quake, there had been a distinct drop in the lead time of the P waves.

As significant as those changes seemed, U.S. seismologists felt that they could not be really dependable as a quake-prediction signal without a more fundamental understanding of what was causing them. That explanation was already available. In the 1960s, while studying the reaction of materials to great mechanical strains, a team of researchers under M.I.T. Geologist William Brace had discovered that as rock approaches its breaking point, there are unexpected changes in its properties. For one thing, its resistance to electricity increases; for another, the seismic waves passing through it slow down.

Both effects seemed related to a phenomenon called dilatancy—the opening of a myriad of tiny, often microscopic cracks in rock subjected to great pressure. Brace even suggested at the time that the physical changes associated with dilatancy might provide warning of an impending earthquake, but neither he nor anyone else was quite sure how to proceed with his proposal. Dilatancy was, in effect, put on the shelf.

The Russian discoveries reawakened interest in the subject। Geophysicist Christopher Scholz of Lamont-Doherty and Amos Nur at Stanford, both of whom had studied under Brace at M.I.T., independently published papers that used dilatancy to explain the Russian findings. Both reports pointed out an apparent paradox: when the cracks first open in the crustal rock, its strength increases. Temporarily, the rock resists fracturing and the quake is delayed. At the same time, seismic waves slow down because they do not travel as fast through the open spaces as they do through solid rock. Eventually ground water begins to seep into the new openings in the dilated rock. Then the seismic-wave velocity quickly returns to normal. The water also has another effect: it weakens the rock until it suddenly gives way, causing the quake.

Soon California Institute of Technology's James Whitcomb, Jan Garmany and Don Anderson weighed in with more evidence. In a search of past records, they found a distinct drop in the speed of P waves 3½ years before the 1971 San Fernando quake (58 deaths), the largest in California in recent years. The P waves had returned to their normal velocity a few months before the tremor. Besides providing what amounted to a retroactive prediction of that powerful quake, the Caltech researchers demonstrated that it was primarily the velocity of the P waves, not the S waves, that changed. Their figures were significant for another reason: the P-wave velocity change was not caused by a quirk of geology in the Garm region or even in the Adirondacks, but was apparently a common symptom of the buildup of dangerous stresses in the earth.

In fact, dilatancy seems to explain virtually all the strange effects observed prior to earthquakes. As cracks open in rock, the rock's electrical resistance rises because air is not a good conductor of electricity. The cracks also increase the surface area of rock exposed to water; the water thus comes in contact with more radioactive material and absorbs more radon—a radioactive gas that the Soviet scientists had noticed in increased quantities in Garm-area wells. In addition, because the cracking of the rock increases its volume, dilatancy can account for the crustal uplift and tilting that precedes some quakes. The Japanese, for instance, noticed a 2-in. rise in the ground as long as five years before the major quake that rocked Niigata in 1964. Scientists are less certain about how dilatancy accounts for variations in the local magnetic field but think that the effect is related to changes in the rock's electrical resistance.

With their new knowledge, U।S. and Russian scientists cautiously began making private predictions of impending earthquakes. In 1973, after he had studied data from seven portable seismographs at the Blue Mountain Lake encampment, Columbia University's Aggarwal excitedly telephoned Lynn Sykes back at the laboratory. All signs, said Aggarwal, pointed to an imminent earthquake of magnitude 2.5 to 3. As Aggarwal was sitting down to dinner two days later, the earth rumbled under his feet. "I could feel the waves passing by," he recalls, "and I was jubilant." In November 1973, after observing changes in P-wave velocity, Caltech's Whitcomb predicted that there would be a shock near Riverside, Calif., within three months. Sure enough, a tremor did hit before his deadline—on Jan. 30. Whitcomb's successful prediction was particularly important. All previous forecasts had involved quakes along thrust faults, where rock on one side of a fault is pushing against rock on the other. The Riverside quake took place on a strike-slip fault, along which the adjoining sides are sliding past each other. Because most upheavals along the San Andreas Fault involve strike-slip quakes, Whitcomb's forecast raised hopes that seismologists could use their new techniques to predict the major earthquakes that are bound to occur along the San Andreas.

The Chinese, too, were making rapid progress in their earthquake-forecast studies. When a delegation of U.S. scientists headed by M.I.T. Geologist Frank Press toured Chinese earthquake-research centers in October 1974, they were astonished to learn that the country had some 10,000 trained earthquake specialists (more than ten times the American total). They were operating 17 major observation centers, which in turn receive data from 250 seismic stations and 5,000 observation points (some of which are simply wells where the radon content of water is measured). In addition, thousands of dedicated amateurs, mainly high school students, regularly collect earthquake data.

The Chinese have good reason to be vigilant. Many of their people live in vulnerable adobe-type, tile-roofed homes that collapse easily during tremors. And the country shudders through a great number of earthquakes, apparently because of the northward push of the Indian plate against the Eurasian plate. Says Press: "It is probably the one country that could suffer a million dead in a single earthquake."

Chinese scientists read every scientific paper published by foreign earthquake researchers. They also pay close attention to exotic prequake signals—including oddities of animal behavior—so far largely overlooked by other nations. Before a quake in the summer of 1969, the Chinese observed that in the Tientsin zoo, the swans abruptly left the water, a Manchurian tiger stopped pacing in his cage, a Tibetan yak collapsed, and a panda held its head in its paws and moaned. On his return from the China tour, USGS's Barry Raleigh learned that horses had behaved skittishly in the Hollister area before the Thanksgiving Day quake. "We were very skeptical when we arrived in China regarding animal behavior," he says. "But there may be something in it."

Though the U.S. does not have the national commitment of the Chinese, there is no lack of urgency among American scientists. California has not had a great earthquake since the San Francisco disaster in 1906, and seismologists are warily eying at least two stretches of the San Andreas Fault that seem to be "locked." One segment, near Los Angeles, has apparently not budged, while other parts of the Pacific and North American plates have slid some 30 ft. past each other. Near San Francisco, there is another locked section. Sooner or later, such segments will have to catch up with the inexorable-movement of the opposing plates. If they do so in one sudden jolt, the resulting earthquakes, probably in the 7-to 8-pt. Richter range and packing the energy of multimegaton hydrogen bombs, will cause widespread destruction in the surrounding areas.

If one of those quakes occurs in the San Francisco area, the results will be far more calamitous than in 1906 (see box page 40)। A comparable earthquake near Los Angeles could kill as many as 20,000 and injure nearly 600,000.

As a practical start toward earthquake prediction, USGS is constructing a prototype network of automated sensing stations equipped with magnetometers, tiltmeters and seismographs in California's Bear Valley. They are also beginning to make measurements of radon in wells and electrical resistance in rock. Some of the data are already being fed into the USGS's central station at Menlo Park. But analysis is still being delayed by lack of adequate computer facilities.

Other seismic monitoring grids in the U.S. include a 45-station network in the Los Angeles area, operated jointly by the USGS and Caltech; smaller networks in the New York region under the Lamont-Doherty scientists; and those in the Charleston, S.C., area, operated by the University of South Carolina. When completed and computerized, these networks will provide two warnings of impending quakes. If scientists detect changes in P-wave velocities, magnetic field and other dilatancy effects that persist over a wide area, a large quake can be expected—but not for many months. If the dilatancy effects occur in a small area, the quake will be minor but will occur soon. The return to normal of the dilatancy effects provides the second warning. It indicates that the quake will occur in about one-tenth the time during which the changes were measured. If dilatancy changes have been recorded for 70 days and then suddenly return to normal, the quake should occur in about a week.

The networks are far from complete, progress in general has been slow, and seismologists blame inadequate Government funding. The USGS's annual quake budget has remained at about $11 million for the past few years, only about $3 million of it for research in the art of forecasting.

Once in operation, an earthquake warning system will bring with it a new set of problems. If a major quake is forecast for San Francisco, for example, should the Government shut down businesses and evacuate the populace? Where would evacuees be housed? If the quake does not occur, who will be responsible for the financial loss caused by the evacuation? Answers come more easily in totalitarian China. There, says Press, "if an actual quake does not take place, it is felt that the people will understand that the state is acting on their behalf and accept a momentary disruption in their normal lives."

Just such a disruption took place in many Chinese communities on Feb। 4, the day that an earthquake struck an industrialized area in Liaoning province. According to the Chinese publication Earthquake Frontiers, at 6 p.m. that day an announcement was made over the public-address system in the Kuan-t'un commune: "According to a prediction by the superior command, a strong earthquake will probably occur tonight. We insist that all people leave their homes and all animals leave their stables." As an added incentive for people to go outside, the commune leaders also announced that movies would be shown in an outdoor location.

"As soon as the announcement was finished," the article says, "many men and women members with their whole families gathered in the square in front of the detachment gate. The first film was barely finished when a strong earthquake, 7.3 on the magnitude scale, occurred. Lightning flashed and a great noise like thunder came from the earth. Many houses were destroyed at once. Of the 2,000 people in the commune, only the 'stubborn ones,' who ignored the mobilization order, were wounded or killed by the earthquake. All the others were safe and uninjured; not even one head of livestock was lost."

Convinced that "earthquake prediction is a fact at the present time," and worried about the effect of such forecasts, particularly in U.S. cities, the National Academy of Sciences this week released a massive study entitled "Earthquake Prediction and Public Policy." Prepared by a panel of experts headed by U.C.L.A. Sociologist Ralph Turner, the study takes strong issue with the politicians and the few scientists who believe that earthquake predictions and warnings would cause panic and economic paralysis, thus resulting in more harm than the tremors themselves. Forecasting would clearly save lives, the panel states, and that is the "highest priority." Because most casualties during a quake are caused by collapsing buildings, the report recommends stronger building codes' in areas where earthquakes occur frequently, the allocation of funds for strengthening existing structures in areas where earthquakes have been forecast and even requiring some of the population to live in mobile homes and tents when a quake is imminent. Fearful that forecasting could become a political football and that some officials might try to suppress news of an impending quake, the panel recommends that warnings, which would cause disruption of daily routine when an earthquake threatens, should be issued by elected officials—but only after a public prediction has been made by a panel of scientists set up by a federal agency.

Other scientists are already looking ahead toward an even more remarkable goal than forecasting: earthquake control। What may become the basic technique for taming quakes was discovered accidentally in 1966 by earth scientists in the Denver area. They noted that the forced pumping of lethal wastes from the manufacture of nerve gases into deep wells at the Army's Rocky Mountain arsenal coincided with the occurrence of small quakes. After the Army suspended the waste-disposal program, the number of quakes declined sharply.

Fascinated by the implications of what were apparently man-made quakes, USGS scientists in 1969 set up their instruments at the Rangely oilfield in northwestern Colorado. There, Chevron was recovering oil from less productive wells by injecting water into them under great pressure. The recovery technique was setting off small quakes, the strongest near wells subjected to the greatest water pressure. If water was pumped out of the earth, the survey scientists wondered, would the quakes stop? In November 1972, they forced water into four of the Chevron wells. A series of minor quakes soon began, and did not stop until March 1973. Then the scientists pumped water out of the wells, reducing fluid pressure in the rock below. Almost immediately, earthquake activity ended. In a limited way, they had controlled an earthquake.

The results of the Rangely experiments led USGS Geophysicists Raleigh and James Dietrich to propose an ingenious scheme. They suggested drilling a row of three deep holes about 500 yds. apart, along a potentially dangerous fault. By pumping water out of the outer holes, they figured they could effectively strengthen the surrounding rock and lock the fault at each of those places. Then they would inject water into the middle hole, increasing fluid pressure in the nearby rocks and weakening them to the point of failure. A minor quake—contained between the locked areas—should result, relieving the dangerous stresses in the immediate vicinity. By repeating the procedure, the scientists could eventually relieve strains over a wide area. Other scientists feel that such experiments should be undertaken with caution, lest they trigger a large quake. Raleigh is more hopeful. In theory, he says, relatively continuous movement over the entire length of the San Andreas Fault could be maintained—and major earthquakes prevented—with a system of some 500 three-mile-deep holes evenly spaced along the fault. Estimated cost of the gigantic project: $1-$2 billion.

In a time of austerity, the possibility of such lavish financing is remote। As M.I.T.'s Press puts it: "How does one sell preventive medicine for a future affliction to Government agencies beleaguered with current illness?" Ironically, the one event that would release money for the study of earthquake prediction and control is the very disaster that scientists are trying to avert: a major quake striking a highly populated area without any warning. Tens of thousands of people living in the flood plain of the Van Norman Dam had a close call four years ago in the San Fernando Valley quake; had the tremor lasted a few more seconds, the dam might have given way. When the San Andreas Fault convulses again—as it surely must—or when another, less notorious fault elsewhere in the U.S. suddenly gives way, thousands of other Americans may not be so lucky.

* At the current rate of plate movement, Los Angeles will lie directly west of San Francisco in 10 million years.

* Used to measure the strength of earthquakes. Because the scale is logarithmic, each higher number represents a tenfold increase in the magnitude of the tremors, and a 30-fold increase in the energy released. Thus a 2-point quake is barely perceptible, a 5 may cause minor damage, a 7 is severe, and an 8 is a violent quake.

* U.S. scientists now estimate that the change can occur as long as ten years before a magnitude 8 quake, a year before a 7-pointer and a few months before a magnitude 6.