Tensor Electromagnetic Profiling of the San Andreas fault 1434-94-6-2423 - Martyn Unsworth and John Booker, P.I.'s Dept. of Earth and Space Sciences, University of Washington Seattle, Washington 98195 ph: (206) 543-4980; e-mail: unsworth@ess.washington.edu Jan. 1, 1993 - Jun. 14, 1995

Component II.1

Investigations

In November 1994 a party from University of Washington and Oregon State University collected continuous magnetotelluric (MT) data across the San Andreas fault (SAF) in central California. The goal of the project was to obtain in situ measurements of electrical resistivity in the fault zone, to help constrain structural models of the fault and to characterize potential locations for deep scientific drilling. At Carrizo Plains an 8 km profile was acquired that crossed the SAF on the Dragonsback, a pressure ridge some 20 miles south of Wallace Creek. At Parkfield a 5 km profile was recorded across Middle Mountain close to the epicenter of the 1966 characteristic earthquake. On both profiles measurements were made every 100 m, in contrast to conventional MT, which has observations every few kilometers. The dense spatial sampling yields a high resolution image of the subsurface electrical resistivity which can be used to map subsurface structures and fluids.

Magnetotelluric data consist of a time recording of the natural variations of the earth's electric and magnetic fields. To convert these observations into information about the subsurface structure requires significant data processing. The first step is a spectral analysis of the time-series. Since high frequency signals contain information about shallow structure and low frequency signals are sensitive to deep structure, this process gives information about the depth variation of electrical resistivity. This analysis yields estimates of the frequency dependent apparent resistivity and phase. A fuller description of the magnetotelluric method is given by Vozoff (1991). The second processing step is to determine the electrical structure that gives rise to the observed apparent resistivity and phase. This process is referred to as inversion and throughout this study we have used the Rapid Relaxation technique of Smith and Booker (1991).

Figure 1 shows the resistivity model that best fits the data collected at Carrizo Plain. The model is only shown to 4 km depth, since a survey line can only reliably be interpreted to a depth approximately equal to half its length. The main features of the model are indicated in the diagram beneath the model. Note that there is a deep change in resistivity across the fault. In the upper kilometer a zone of low resistivity exists, bounded on the west by the fault. This may be due to ground water trapped uphill by the fault gouge. East of the SAF a thrust fault intersects the vertical San Andreas at a depth of 4 km. A second fault dips towards the SAF. These faults, and others that have been mapped west of our survey, may form a flower structure. Alternatively, the upper fault may continue downward, resulting in a beheaded San Andreas with the fault at depth offset slightly to the east. Beneath a depth of 2 km the SAF shows no evidence of having a low resistivity. This indicates that the fault zone contains limited fluids. If present, it would be expected that they would have a significant resistivity signature as described by Eberhart-Philips et al (1995).

Figure 1: Electrical resistivity model of the San Andreas Fault at Carrizo Plains. Ticks indicate measurement locations. The model was derived by inversion of tensor magnetotelluric and is the smoothest model that fits the data. Lower panel shows a structural interpretation.

An identical analysis was applied to the data collected at Parkfield, and the best fitting resistivity model is shown in Figure 2. Note that the shorter profile has limited our depth of investigation to approximately 2 km. The structure at this location is very different to that at Carrizo Plain. West of the fault the resistive block is known from geological mapping by Dibblee (1980) and others to be Salinian granite. This is overlain by approximately 700 metres of sediments from the Paso-Robles formation. East of the fault are less resistive Tertiary and Franciscan rocks. Directly beneath the surface trace of the San Andreas fault a vertical zone of low resistivity is imaged. This feature is required by the magnetotelluric data, and is interpreted as the damaged zone surrounding the fault gouge. It appears that this low resistivity fault zone extends to at least 4 km. The very low resistivities are due to high porosity due to extensive fracturing.

Well logs to the east of the fault indicate that brines may be present that would also contribute to the very low resistivity. The resistivity is lowest vertically beneath the surface trace, indicating the most highly fractured part of the fault zone. This implies a vertical fault to at least 4 km, in agreement with micro-earthquake locations. The width of this low resistivity correlates very well with the seismic low velocity zone observed by Leary and Ben-Zion (1992). Compression normal to the fault may well be responsible for the generation of the topography and land slips that are observed on Middle Mountain.

Figure 2 : Electrical resistivity model of the San Andreas Fault at Parkfield. The resistivity model was derived by identical processing to that shown in Figure 1. Note that the fault is characterized by very low resistivities in a 500 m wide zone directly beneath the surface trace. This zone is interpreted as being the damaged zone adjacent to the fault gouge.

These two images of the San Andreas fault are strikingly different. The Carrizo Plain profile is located on a fault segment that has been locked since the 1857 Fort Tejon Earthquake. In contrast Parkfield lies on a segment that has produced a series of characteristic earthquakes. The presence of significant amounts of fluids in the fault zone at Parkfield may well be linked to the generation of these characteristic earthquakes. It is more difficult to relate the fault zone resistivites to fault mechanics at Carrizo Plain since the seismogenic zone is beyond the depth of investigation of this survey. This is further complicated by the fact that we only have electrical resistivity information at one point on the Carrizo segment. The properties of the fault zone may well show significant along strike variation. However the lack of fluids in the near-surface on this survey line may reflect the fault zone at depth and explain why this segment has been able to accumulate significant stress since 1857. Joint analysis of the conventional MT data described by Mackie (1994) may be able to address this issue on the Carrizo segment.

References

Dibblee, T.R., Geology along the San Andreas fault from Gilroy to Parkfield, in Special Report 140, California Division of Mines and Geology, Sacramento, 1980

Eberhart-Phillips, D., Stanley, W.D., Rodriguez, B. and Lutter, W.J., Surface seismic and electrical methods to detect fluids related to faulting, J. Geophys. Res., 100, 12,919-12,936, 1995.

Leary, P. and Ben-Zion, Y., A 200m wide fault zone low velocity layer on the San Andreas Fault at Parkfield: results from analytic waveform fits to trapped wave groups, Seismo. Res. Lett., 63, p62, 1992.

Mackie, R.L. and Unsworth, M.J., Booker, J.R, and Madden, T.R., 1994, Images of the San Andreas Fault from recent Magnetotelluric measurements EOS Transactions , 76, p. F168

Smith, J.T. and Booker, J.R., Rapid Inversion of Two and Three Dimensional Magnetotelluric data, J. Geophys. Res., 96,3905-3922,1991.

Vozoff, K., The Magnetotelluric Method, in Electromagnetic Methods in Applied Geophysics, edited by M.N. Nabighian, Society of Exploration Geophysicists, 1991.

Publications

Mackie, R.L. and Unsworth, M.J., Booker, J.R, and Madden, T.R., 1994, A preliminary analysis of a LIMS/EMAP survey across the San Andreas Fault in the Carrizo Plains natural Area, abstr., EOS Transactions , 75, p. 200.

Unsworth, M.J., Wu, N., Booker, J.R, Madden, T.R., and Egbert, G., 1995, High Resolution Electromagnetic Imaging of the San Andreas Fault, abstr., IUGG General Assembly, Boulder, Colorado.

Unsworth, M.J., Booker, J.R, and Egbert, G., 1995, The structure of the San Andreas fault at Middle Mountain, Parkfield from a magnetotelluric survey, abstr., EOS Transactions, 76, p. F168

Mackie, R.L. and Unsworth, M.J., Booker, J.R, and Madden, T.R., 1994, Images of the San Andreas fault from recent Magnetotelluric measurements EOS Transactions , 76, p. F168