STUDIES RELATED TO OCCURRENCE OF EARTHQUAKES NEAR NEW RESERVOIRS

Abstract

This is a report on several sets of theoretical and numerical investigations on the occurrence of earthquakes near new reservoirs. A review of the literature of the last two decades on the theme indicated the need and scope for developments in the theory of saturated porous elastic media. Thus the first set of our investigations is theoretical. We simulate reservoir induced stresses and pore pressures in a water saturated porous elastic half space. We consider reservoirs of infinite extent (1D problem), infinite length and finite width (2D problem or strip reservoir problem) and laterally finite extent (3D problem). The 1D and 2D problems were exercises to evolve a tested algorithm for 3D problems. We have attained, in principle, a capability for computer based 3D simulations of stresses and pore pressure in a porous elastic medium under the approximation that elastic stresses can affect pore pressure but not vice-versa. The second investigation is conceptual. Although briefer than the above'work, it is, in our opinion, the most significant. We have explored the notion of fault stability as a measure of the interplay of frictional stresses mobilised and the resolved shear stresses acting on a fault. We suggest that, in the simulations of earthquakes near a reservoir, we should consider not only the fault stability (Sr in Pascals) induced by a reservoir but also, at least notionally, the stability (Sa) induced by the ambient stresses. Although it may be difficult to measure Sa independently, we may assert that the numerical sum of Sr and Sa should be zero on the causative fault at the hypocentre at the time of an earthquake. We note three other facts in this regard. Firstly, a reservoir will have a stabilising or destabilising influence on a nearby fault depending on whether Sr is numerically positive or negative respectively. Secondly, the ambient stresses contributing to Sa would be generally much greater in magnitude than reservoir induced stresses involved in S,. Thirdly, both Sa and Sr may be functions of time. We thus suggest that, in general, a reservoir alone cannot be the cause of an earthquake. Rather, the construction and operation of a reservoir leads to elastic stresses and pore pressure which may merely advance or delay the times of occurrence of natural earthquakes in its vicinity. In other words RIS should stand for reservoir influenced (natural) seismicity. But if the current meaning of RIS as reservoir induced seismicity is to be retained then it would have to be restricted to only those earthquakes that occur due to a destabilising influence of the reservoir at the hypocentres in their causative faults. The above tools and concepts were used to examine the seismicity within a twenty kilometre radius of the Tarbela reservoir in Pakistan Himalaya during the dry seasons of 1977, 1980 and 1981. Contrary to the recent view that they were reservoir induced earthquakes, we lii) suggest that a majority ofthem were marginally delayed natural tectonic earthquakes because they occurred inspite of the stabilising influence exerted on the respective causative faults by the reduced quantity of water remaining in the reservoir in the dry seasons. Asimilar investigation of selected earthquakes from the Koyna reservoir region in peninsular India leads us to conclude that the reservoir may have exerted a destabilising influence on nearby seismogenic faults and caused the earthquakes of the Koyna sequence to occur ahead of their appointed times in the absence of the reservoir. All the foregoing work was, in essence, a preparation for our final investigation. A grossly simplified but suggestive simulation leads us to the view that the impoundment of the reservoir behind the 260 m high Tehri dam in the Garhwal Himalaya may delay by an unspecified time, but not prevent, the occurrence of a great natural earthquake due to build up of stresses of plate tectonic origin in the region.