Review of seismic-hazard issues associated with the Auburn Dam project, Sierra Nevada foothills, California

U.S. Geological Survey Open File Report 96-0011

RESERVOIR-INDUCED SEISMICITY

Based on analysis of 55 reported cases of reservoir-induced seismicity worldwide and geologic and seismologic data for 16 selected dams and reservoirs in the Sierran foothills, WCC (1977) reached two sets of conclusions about reservoir-induced seismicity associated with Auburn Dam as proposed at that time. First, if the 1975 Oroville earthquake is assumed not to be reservoir induced, then the likelihood of an induced earthquake of the magnitude of the Oroville event (5.7) or larger is 2% to 5 % during the life of the dam. Second, if the Oroville earthquake is assumed to be reservoir induced, then the likelihood of an induced earthquake of the magnitude of the Oroville earthquake is 30 % during the lifetime of the dam. We have not been able to reconstruct the basis for these probabilities and do not endorse them. The specific probabilities of an induced earthquake of M 5.7 or larger associated with an Auburn reservoir are open for revaluation.

Since 1979 the body of data on reservoir induced seismicity has grown. Table 1 1ists four additional case histories of proposed reservoir-induced seismicity as well as those for other types of induced earthquakes. A better understanding of reservoir-induced seismicity has come from theoretical analyses and from studies of earthquakes induced by other means such as fluid injection, gas and oil production, and stress changes associated with other earthquakes. Recent studies have suggested why some reservoirs produce earthquakes immediately upon filling (due to elastic stress changes), some after a delay (a result of pore fluid diffusion), and some after several years but only when the water level is changed (fluid diffusion accompanied by the elastic stress changes) (Roeloffs, 1988; Segall, 1992; Simpson and others, 1988). As currently understood, the essential features of reservoir-induced seismicity, and in some cases other types of induced earthquakes, are:

1. Observable characteristics (stress drop, ground motions, source parameters) of earthquakes induced by reservoir impoundment, fluid injection, or gas and oil production are indistinguishable from those of naturally occurring tectonic events.

2. The magnitude of the maximum induced earthquake is geologically controlled and is independent of reservoir impoundment.

3. Most examples of induced seismicity occur in regions of low tectonic loading rates dominated by normal faulting as is the case in the Sierran foothills, or strike-slip faulting (Scholz, 1990).

4. The probability of the occurrence of earthquakes at and in the region around a reservoir may be increased by impoundment. In a few of the best controlled studies where a reliable record of seismicity before and after reservoir impoundment exists (Segall, Grasso, and Mossop, 1994; Simpson and Negmatullaev, 1981), an increased rate of activity is seen within 10-15 km of impounded reservoirs and gas fields (Figure 2 and Figure 3/).

5. Observations point to a higher rate of earthquakes after reservoir impoundment, especially during the first decade (Anderson and O'Connell, 1993; Scholz, 1990; Simpson, 1986), although the relationship between impoundment time history and reservoir-induced seismicity is not consistent or well understood. In some cases reservoir induced seismicity has been observed shortly after the initiation of filling (Figure 3). For example, a magnitude 6.3 earthquake was induced by the Kremasta reservoir in Greece less than seven months after the initiation of filling in 1966 (Cominakis et al 1968). Alternatively, the Oroville earthquake, if reservoir induced, occurred ten years after reservoir impoundment.

6. Induced earthquakes, reservoir induced or otherwise, appear to be triggered by very small stress changes, sometimes only a small fraction of one bar. For examples of induced earthquakes listed in Table 1, the calculated stress changes in the seismogenic zone are remarkably small, typically a few percent of the stress drop of the induced earthquake. During the past five years, new evidence has emerged that earthquakes can also be induced by the stress transferred from other earthquakes. Earthquake-induced stress changes of 1-3 bars (about the same as the pressure in a car tire) appear to be able to trigger large earthquakes, such as the M 6.5 Big B ear earthquake, which occurred three hours after and 30 km away from the M 7.4 Landers event (Harris and Simpson, 1992; King, Stein, and Lin, 1994), and stress changes of 20.25 bar can trigger abundant smaller earthquakes (Harris and Simpson, 1992; Harris and Simpson, 1995; Harris, Simpson, and Reasenberg, 1995; Stein, King, and Lin, 1994).

We suggest that the proposed Auburn project entails some likelihood of reservoir-induced seismicity. The probability of a resewoir-induced event could vary significantly depending on whether the dam is a flood-control-only detention dam or is used for permanent water storage. The worldwide correlations among reservoir depth, resewoir volume, and seismicity suggest a greater likelihood of reservoir induced-seismicity for the permanent water storage version of the Auburn project because of the longer period of time the region is exposed to the load and pore pressure changes. The flood-control-only version may involve a lower probability of significant reservoir induced-seismicity than the permanent water storage reservoir. However, in view of the observation that earthquakes can be induced during rapid changes in reservoir level and after impoundment of only a few months, the possibility of reservoir induced-seismicity associated with a flood-control-only dam, for which impoundment would be rapid but only during flood conditions, should not be dismissed.

We recognize there are questions and uncertainties in quantifying this hazard. Aside from Oroville, which may have had reservoir-induced seismicity, there are six existing large dams in the Sierran foothills that have the height and reservoir volume necessary to place them in the worldwide classification of dams that have the potential to produce reservoir induced seismicity. These dams and their construction dates are Folsom (1956), Pardee (1929), New Bullards Bar (1970), New Don Pedro (1970), New Exchequer (1966), and New Melones (1978) (Figure 1). Since the late 1970s these dams have experienced several major episodes of drawdown and filling associated with drought and flood cycles. Between 1978 and 1994 the seismic network operating in the Sierran foothills has not recorded the occurrence of any earthquakes of M24 in this region (M. McLaren, PO&E, written communication) and there is no obvious association between recorded seismicity and these large dams.

In summary, we conclude that reservoir-induced seismicity is an issue that will require additional analysis if the Auburn Dam project continues. There is a need to pursue development of approaches to calculate how reservoir impoundment may affect earthquake probabilities at the Auburn site and its environs. If the Auburn Dam Project is authorized, a dense seismic network should be installed around the Auburn Dam site as soon as possible to determine a baseline level of seismicity before impoundment. A downhole, three-component, wide-dynamic-range seismometer array could be very useful for this purpose. Additionally, borehole measurements of permeability, hydraulic head, and stress should be made in the vicinity before, during, and after impoundment.

Auburn Dam Index
USGS Water Resources of California
USGS Headquarters
USGS Water Resources - National