In a deep canyon in the foothills of the Himalayas, near the Indian town of Tehri, a thousand-odd workers are building a hydroelectric dam. When it is complete in four years or so, the Tehri High Dam will deliver 2,400 megawatts of electricity to energy-starved northern India. Its location seems perfect: a hydroelectric plant draws its energy from falling water, after all, and the greater the drop, the better. Once the Tehri dam has blocked off the narrow gorge of the Bhagirathi, interrupting that river’s rush toward the Ganges, the water will have a long drop indeed--850 feet from the top of the dam to the canyon floor. Moreover, the 30-mile-long reservoir behind the dam will provide the drinking and irrigation water that Delhi and the rest of the region need to weather the arid months before the summer monsoon. To Indian planners, the Tehri dam makes doubly good sense.
But the very forces that make it such a good idea--the geologic forces that built the deep gorges and high peaks of the Himalayas--might also be the dam’s undoing. The Tehri dam will stand in one of the world’s most active earthquake zones. Less than ten miles beneath the dam site lies a fault, the geologic boundary between India and the Asian continent, that is capable of unleashing a catastrophic quake. Some seismologists think the fault segment around Tehri is overdue for a quake whose magnitude could be as large as 8.9. And they fear the dam will not survive it.
The geologic stage for this human drama was set some 180 million years ago, with the breakup of the supercontinent of Gondwanaland. One of Gondwanaland’s fragments was a crustal plate that carried what is now the Indian subcontinent. Over the next 130 million years, this plate tracked steadily northward, erasing the ocean that then separated India from Eurasia. The oceanic crust was subducted--forced down into the underlying mantle.
But continental crust is too light and buoyant to be subducted. So when India itself finally bulldozed into Eurasia 50 million years ago, it hit a wall. Both continents had to give. On the Eurasian side of the collision zone, the crust was thickened and squeezed upward, creating the high plateau of Tibet. On the Indian side, a great slab of rock was sheared off the leading edge of the subcontinent, along a thrust fault that dipped gently northward toward Eurasia, like a beach toward the sea. As India ground forward the slab was forced back up onto the subcontinent. Eventually that fault became inactive; at that point another detachment fault formed underneath and parallel to the first one, and another slab was shoved underneath the first, lifting it up. Over millions of years, those two slabs became the Himalayas.
Today India still piles headlong into Eurasia, and the Himalayas are still under construction. The active detachment fault is around 1,500 miles long and between 60 and 120 miles wide; from the plains of northern India it dips north underneath the mountains. Most of the time the fault is stuck. But every so often, after enough elastic energy has built up in the fault to force the blocked continent forward, it slips 30 feet or more all at once. That generates a great earthquake: a quake of magnitude 8 or greater.
There have been four such quakes along the detachment fault in just the past 100 years. The quakes--a magnitude 8.7 in 1897, in Assam; a magnitude 8 in 1905, in Kangra; a magnitude 8.4 in 1934, in Bihar; and a magnitude 8.7 in 1950, again in Assam--each ruptured a different segment of the fault. But one segment, located between the Kangra and Bihar rupture zones (roughly between the longitudes of Kathmandu and Delhi), has been strangely reserved. Between 300 and 500 miles long, it is called the Central Himalayan seismic gap.
Most seismologists believe quakes are more likely to occur in seismic gaps, where they haven’t occurred recently. If so, the Central Himalayan gap is more than due. According to the available historical records--mostly Portuguese and English records from India’s colonial days-- it has not had a great quake for at least 300 years. Notes scribbled on some Nepalese religious tracts indicate that a big quake did strike Kathmandu in June 1255. The quake killed one-third of the population of the country, including its king. Assuming it ruptured the detachment fault in the Central gap (which isn’t certain), and assuming it was the last major quake in that segment (ditto), then 740 years’ worth of strain has built up in the gap.
Last year seismologists Roger Bilham of the University of Colorado in Boulder, Roland Bürgmann of Stanford, and their colleagues determined the amount of strain, using the Global Positioning System, a network of satellites stationed around the globe that can pinpoint, within a fraction of an inch, the distances between receivers on the ground. The researchers found that India and Asia are converging at the rate of around eight-tenths of an inch per year. But nearly all that convergence is happening between central Nepal and Tibet, in the Himalayas themselves; Nepal and India are barely getting closer at all--because the detachment fault that separates them is stuck. At the moment, the energy of the continental collision seems to be going into squeezing the rocks in the mountains; that is, it’s being stored as elastic strain. Someday, says Bilham, the rocks will rebound, and the strain will be released as motion along the fault. If the last quake in the Central gap was indeed in 1255, some 50 feet of motion--eight-tenths of an inch per year times 740 years-- has built up and needs to be released.
What would it take to release all that strain? Hundreds of quakes comparable to the magnitude 6.8 temblor that devastated Kobe, Japan, last January, or the magnitude 6.7 that rocked Los Angeles in January 1994--but if the Central Himalayan gap were prone to minor quakes in such number, they would be happening already. It’s far more likely, says Bilham, that the strain will be uncorked by quakes of magnitude 8 or greater, as it has been elsewhere on the detachment fault. Bilham estimates that four magnitude 8.2 quakes would do the job, or three 8.5’s, or a single 8.9. (The energy released by a quake increases by a factor of 30 with each added point in magnitude.)
The extent of the damage would be enormous. A single 8.2 quake in the Central gap, says Bilham, would be felt as far away as Calcutta and Bombay, some 850 miles distant. We’re talking 100 or 200 million people affected. All along the Ganges plain, where there are about a dozen cities with populations of at least a million, including the capital of Delhi, with nearly 9 million inhabitants, the shaking would be intense, because the soft soil would tend to amplify it. And it would be intense too, of course, in the Central gap itself, directly above the ruptured fault segment.
The Tehri High Dam, 200 miles northeast of Delhi, is being built on the edge of the Central gap.
The site was chosen in 1961, before plate tectonics had revolutionized geology--at a time when geologists didn’t understand the fundamental mechanism driving earthquakes in the Himalayan region. When the original design for the dam was approved by the Indian government’s Planning Commission in 1972, engineers still believed the dam would be at risk only from a quake on one of the smaller faults that crisscross the area. Thus they designed the dam to withstand only a magnitude 7.2 quake, which they figured would produce ground accelerations--fast back-and-forth shaking triggered by high-frequency seismic waves--of .25 g, or one-fourth the acceleration due to gravity.
Construction of the Tehri dam began in 1978, but it has proceeded slowly. The tunnels that will house the turbines have been dug, and a small cofferdam has been erected to divert the Bhagirathi while the main dam is being built just downstream. Today the main dam rises just 50 feet off the canyon floor. It is being constructed of rock fill--considered more stable in an earthquake than concrete, which cracks, or soil, which can become saturated with water and liquefy--over a core of impervious claylike material. The dam will have a triangular cross section: it will be more than half a mile wide (down the valley) at its base and just 65 feet wide at the crest. When it reaches its full height of 850 feet, its reservoir will submerge, partially or entirely, the town of Tehri and over a hundred villages, with a combined population of 70,000. Opposition from these people and from environmental groups has slowed the dam project. So has opposition from seismologists.
One of those seismologists is Vinod Gaur, formerly the director of the National Geophysical Research Institute of India and now at the Center for Mathematical Modeling and Computer Simulation in Bangalore. In 1990, Gaur served on an expert committee convened by the government to review the troubled dam project. He succeeded in convincing the committee that a quake much larger than a magnitude 7.2--at least a magnitude 8.5-- was quite likely to occur at Tehri. Such a quake, rupturing the detachment fault under the dam, would be capable of producing a 1-g acceleration, an acceleration that could literally throw a building off its foundation.
But the huge increase in projected quake magnitude--more than tenfold, given the logarithmic nature of the magnitude scale--didn’t faze Tehri’s engineers: they concluded that their dam would still be subject to a ground acceleration of only .25 g. They justified this by arguing that the peak ground acceleration at Tehri would not necessarily be greater in a magnitude 8.5 quake than in a 7.2 quake, and by further assuming that the quake’s peak accelerations would be over too fast to do any damage. With the dam 12 years into construction, and more than $200 million spent, this reasoning allowed the dam’s engineers to keep the effective ground acceleration at the level they had designed for.
The assumption that a quake’s strongest accelerations wouldn’t last long enough to damage a dam, Gaur and other seismologists say, isn’t generally true for great quakes. Moreover, the strong accelerations triggered by high-frequency seismic waves aren’t the only threat such a quake poses to a large structure. A low-frequency ground motion can be even more dangerous if it happens to match the dam’s natural oscillation frequency--which in the case of the Tehri dam is one oscillation every one or two seconds. Then the dam’s motion would be greatly amplified. It’s the same thing as if you are on a swing, says John Hall, a Caltech earthquake engineer. If you want to build up your amplitude, you have to kick your feet with the swing’s natural frequency. When you do that, you get resonance.
Resonance is more likely in large quakes, which not only produce stronger vibrations at all frequencies but also pump them out for a longer time. In a magnitude 8.5, you could get a minute of strong shaking, Hall says. In that minute, you could get a good ten cycles of two-second-period motion. The more cycles, the more chance of resonance, and the more chance of collapse.
Back in 1992 the Tehri dam’s design was tested by a computer simulation in which the dam was exposed to the seismic waves measured from a real magnitude 7.1 quake that had produced ground accelerations of more than 1 g. Although the dam showed deformation, it survived intact. But, says Gaur, the test was run for only 12 seconds. If the experiment had been run longer, this deformation, which rose continuously for the duration of the test, would have continued to rise, Gaur says. At some point it might have shown failure.
But every time such questions were raised, he goes on, there were cavalier statements and arm waving--that this had been taken into account and that taken into account. And in an atmosphere of so much scientific illiteracy, what eventually wins the day is how many people repeat the same statement, rather than how scientifically examined that statement is. And unfortunately, the number of people on the other side is much larger.
The Tehri dam’s designers insist failure is next to impossible even in a magnitude 8.5 quake. The design, declares B. B. Raj, deputy general manager of the Tehri Hydro Development Corporation, has been found to be safe. Last fall the Indian government formally agreed: it granted final approval of the dam’s design.
Should the Tehri dam ever collapse, the 700 billion gallons of water in the reservoir would crash downriver, toward the holy cities of Rishikesh and Hardwar, 50 miles or so away. The two cities have a combined population of 200,000, although on any given day 10,000 or more Hindu pilgrims are there, too. Presumably the flood would kill many people who had not already been killed by the quake itself. And presumably, with the dam gone, much of northern India would be without power and water just as it was trying to recover from a tremendous disaster.
Of course, a great quake on the detachment fault would be a tremendous disaster even if the Tehri dam survived. The population of India has grown explosively since the last great quake in 1950, and most of these people live in poorly constructed homes; building codes are barely enforced. In 1993 a magnitude 6.4 quake centered southeast of Bombay in Maharashtra State, far from the Himalayas, killed more than 7,500 people. Not long ago an Indian seismologist named A. S. Ayra, from the University of Roorkee, tried to estimate the potential damage from a repeat of the magnitude 8 quake that struck Kangra, which is in the western Himalayas, in 1905. That quake killed 20,000 people. If it were to happen today, Ayra calculated, anywhere from 88,000 to 344,000 people might die. Nearly 150,000 buildings would collapse completely, and another 268,000 buildings partially. And if the quake happened in the Central gap instead of in Kangra, the Tehri dam might collapse, too.
No one knows when the next great quake will strike the Central gap; it could be tomorrow, or it could be in a century. There are seismologists, notably David Jackson of UCLA, who don’t accept the seismic gap theory at all--who argue that, if anything, large quakes are more likely to recur in areas that have had quakes recently than in seismic gaps. But even Jackson would warn the Indian government against complacency. The Central gap, he says, is still on the plate boundary, has similar geologic characteristics to places nearby that have had large earthquakes, and has to be viewed as a candidate for a large earthquake.
The Indian government, then, faces a heavy burden. It has a booming population that needs power and water. And it has a poor population, living in one of the world’s worst earthquake zones, that is almost completely unprotected from earthquakes. Roger Bilham thinks the government should scrap the Tehri dam in favor of a series of much smaller hydroelectric projects, any one of which could be knocked out by a quake without doing serious harm to the populace. And he thinks the government, in spite of all the other demands on it, must start preparing its people for the next great quake.
If you are the government and you are concerned about a great earthquake being overdue, you have a terrible responsibility to your people, Bilham says. What are you going to do? Are you going to retrofit the buildings? Have earthquake preparedness exercises like Japan? Prohibit buildings of a certain type? These are very severe restrictions economically. But if you do nothing, and the earthquake happens, I think you are being criminally irresponsible.