In spite of early doubts, it is now well established that large reservoirs can generate earthquakes. When the possibility of reservoir-induced seismicity was first suggested during the mid-1960s, it was greeted with skepticism in some quarters, largely because detailed records of the prereservoir seismic background were incomplete at best. A scientifically rigorous examination of the evidence was therefore impossible.
But now there are no doubts, for well-documented examples of reservoir-induced seismicity are known throughout the inhabited continents.
In general terms, what happens is that the huge mass of water in a reservoir changes the stress patterns in the underlying rocks. The water load exerts a simple vertical pressure on the rocks immediately, while the pore pressure in the rocks increases gradually as the water infiltrates at a rate that depends on their permeability and prior water content.
One such example is the Nurek reservoir in the Soviet Union, which has been generating seismic activity since filling began in the early 1970's. In this case, the timing of the induced seismicity is directly related not to the absolute amount of water present but to variations in that amount. Bursts of seismic activity are triggered when the water level changes, with the frequency of the activity increasing as the rate of level change increases.
The spatial distribution of the seismic activity at Nurek is less straightforward. Most of the induced earthquakes occur beneath the central part of the reservoir and upstream from it. The region containing that half of the reservoir immediately behind the dam, on the other hand, is almost completely aseismic. Yet both the seismic and aseismic zones arc lithologically similar, consisting largely of shallow-water limestones, which are highly fractured and thus relatively permeable, interbedded with mixed components of low-permeability shale and gypsum.
What, then, governs the location of the induced seismicity?
The answer lies in the varying structures of the rock strata and the different permcabilities such structures engender. Immediately upstream from the dam the alternating layers of limestone- and gypsum-shale arc folded into a V-shape (syncline). Any water entering one of the permeable layers could migrate to the bottom of the basin formed by that layer, but would be prevented from moving any further downward by the impermeable rock layer beneath. In this region, therefore, water cannot enter the vast mass of rock below and cannot thereby change the port pressure. The water in the reservoir here acts only as a load and thus is insufficient to induce seismicity.
Further upstream from the darn, on the other-hand, the strata arc folded into an inverted V-shape (anticline). In this region water entering a permeable layer can diffuse along it and thus be carried to various depths without restriction. Here the pore pressure at given depths does change when the water level fluctuates, and this is sufficient to generate seismicity.
