(April 9, 1994) The earthquake that killed 10,000 people in India last September struck within 15 kilometers of a reservoir filled just 2 years earlier. That proximity in time and space seems more than coincidental to two U.S. seismologists who propose that filling the reservoir may have set off the quake.

Scientists had previously documented cases in which reservoirs induced sizable quakes in India, China, Africa, and the United States. Four of these matched or exceeded the strength of last year’s tremor, which measured magnitude 6.1. But none of the previous shocks linked to reservoirs killed as many people as the September tremor, which leveled more than 20 villages near the city of Latur.

Leonardo Seeber and John G. Armbruster of the Lamont. Doherty Earth Observatory in Palisades, N.Y., presented their controversial theory in Pasadena, Calif. this week at a meeting of the Seismological Society of America. In support of the hypothesis, they note that Earth’s crust in this part of India is extremely old and normally stable. Local records show no sign of earthquake activity until a year after the Killari reservoir was filled, when residents began noticing small tremors. The main quake followed these preshocks a year later. “That doesn’t prove [the connection]. But the timing favors this hypothesis, as does the location,” says Seeber, who visited the region of the epicenter in the weeks following the quake.

Filling a reservoir can trigger a tremor in two ways. The weight of the water can stress the crust underneath, causing it to crack. Or water can seep down into old faults near the reservoir, lubricating rock on either side of the fault and enabling these crustal blocks to slip.

Seeber and Armbruster’s arguments have not convinced other seismologists who have studied the Latur earthquake. Pradeep Talwani, an Indian-born seismologist at the University of South Carolina in Columbia, has investigated numerous reservoir-induced quakes. After visiting the epicentral region last year, he concluded that the Latur shock does not fit the pattern observed in other exampies of reservoir triggering.

Taiwani and others point out that the Killari reservoir was only 10 meters deep at the time of the quake, whereas most reservoirs that spark large quakes measure 100 meters or more in depth. “This was essentially a puddle compared with most cases of reservoir-induced seismicity,” Talwani says.

He also focuses on the unusual pattern of foreshocks and altershocks. When reservoirs set off large quakes, they usually trigger a “swarm” of many small tremors that begin before the main jolt and continue long after. Although the Latur quake did have preshocks, they did not occur as part of a swarm that rattled up until the disaster. Furthermore, the altershocks died off extremely quickly.

“None of the features that we usually associate with reservoir-induced seismicity was present; Talwani says.

Archibald C. Johnston of Memphis State University visited the Latur area in February as part of a United Nations group studying the earthquake. “Our team favored the conclusion that it was not induced by the reservoir, although we acknowledged that it could have been triggered. The difference being that when we say triggered, we mean the earthquake would have happened anyway, but its timing might have been altered by the reservoir. Induced means that the reservoir actually created an earthquake that wouldn’t have happened otherwise:

Regardless of what caused the Latur quake, Seeber notes that many earthquakes are set off by reservoirs, quarries, geothermal power stations, and other artificial structures. Because tremors often precede these quakes, Seeber and Armbruster urge colleagues to consider the seismic potential of such structures when small shocks occur in normally stable continental crust.

Richard Manastersky, April 9, 1994 issue of Science News