The San Andreas fault system in northern California includes three sub-parallel right-lateral faults: the San Andreas, Ma'acama, and Bartlett Springs. This northernmost segment is the youngest portion of the fault system, forming in the wake of the northwestwardly propagating Mendocino triple junction where the Pacific, North America, and Gorda (southern Juan de Fuca) plates meet. The Pacific plate moves about 35-40 mm/yr relative to central California across a broad 100-km zone in northern California. Additional deformation in eastern California and the Basin and Range province contribute to the total relative Pacific-North America motion of 50 mm/yr. The San Andreas fault itself has been essentially aseismic and accumulating strain since it last ruptured in the great 1906 San Francisco earthquake, and no major earthquakes have occurred during the historical record on the more seismically active Ma'acama, and Bartlett Springs faults, which are northern extensions of the Hayward-Rodgers Creek and Calaveras-Concord-Green Valley faults in the San Francisco Bay area.
In Freymueller et al. (1999) we used GPS data collected in 1991-1995 along two profiles crossing the faults near Ukiah and Willits (Figure 27.1). GPS velocities from these profiles constrain the total deep slip rate on the San Andreas fault system to be mm/yr (68.6% confidence interval). Although deep slip rates on the individual faults are less well determined due to high correlations between estimated slip rates and locking depths, and between slip rates on adjacent faults, the slip rate on the Ma'acama fault ( mm/yr) implies that it has now accumulated a slip deficit large enough to generate a magnitude 7 earthquake and therefore poses a significant seismic hazard.
In this renewed and ongoing study, we are resurveying the original profiles and adding two new profiles to the north and south (Covelo and Healdsburg, respectively, in Figure 27.1). Most of the monuments were last observed in 1993 or 1995, so the new observations significantly improve the velocity estimates, and we expect they will improve models of average interseismic strain accumulation, including possible spatial variations along the fault system. These 10-station profiles every 50 km from Pt. Reyes to Cape Mendocino form a primary monitoring network for future observations to detect temporal variations in deformation. We plan to survey 40 additional stations in the southern portion of the network in Fall 2003 to provide better monitoring along the Rodgers Creek and Ma'acama faults.
The survey of the 4 primary profiles was conducted during January-March 2003 after verifying the benchmarks were still suitable for GPS observations, and picking or installing substitutes at the few that were not. Most of the stations were occupied for 6.5-8 hours on two different days. Some sites in the Central Valley or in the higher portions of the Coast Ranges that were not occupied due to weather or logistical considerations will be included in the Fall 2003 survey. Altogether, 43 site positions were measured during 94 session occupations, with the assistance of students and staff of the BSL.
We processed the data using GAMIT/GLOBK software using many of the same techniques used to process the BARD observations (Murray and Segall, 2001). These distributed processing methods allow the solutions to be combined in a self-consistent fashion with other solutions, such as for the BARD network, and for more global networks provided by the SOPAC analysis center, using Kalman filtering techniques, providing a well-defined velocity reference frame with respect to the stable North America. We are now reprocessing the older observations in GAMIT/GLOBK to tie all the northern California observations together in a self-consistent manner. These data sets include Stanford surveys of the profiles, NGS surveys of the HPGN network, USGS surveys of the Covelo profile, and Caltrans surveys of the HPGN-Densification sites.
Figure 27.2 shows site velocities for the 1994-2003 period relative to stable North America, as defined by a set of 20 fiducial stations. Most of the velocities were derived from data spanning 8-10 years, whereas those with the largest error ellipses include data from only a 4 year span (most of these stations will be reoccupied in Fall 2003). The easternmost stations exhibit motions typically associated with Sierran-Great Valley block (ORLA: 12.5 mm/yr NW). The westernmost sites are moving close to the Pacific plate rate (PTAR: 45.9 mm/yr NW). Fault-normal contraction is observed east of the Ma'acama fault, in the region of the Coast Ranges near the Central Valley where similar contraction has been observed elsewhere (e.g., Murray and Segall, 2001).
The North America reference frame used in this analysis is an improvement over the single-station approach used in the Freymueller et al. (1999) study, and allows us to apply angular velocity-fault backslip modeling techniques (e.g., Murray and Segall, 2001) to account for both far-field plate motions and interseismic strain accumulation. We are modifying a set of algorithms provided by Brendan Meade of MIT that sums backslip on rectangular dislocations to extend our simple 2D method to more complex, 3D fault systems (including subduction zones and extensional provinces). We are currently testing it on a variety of problems, such as the Adriatic region with M. Battaglia and R. Bürgmann, and it appears to be well suited to study the northern San Andreas fault system and its transition to the Cascadia subduction zone.
The velocity orientations do not closely follow the mapped traces of the faults in the northernmost section, as one might expect from pure elastic strain accumulation on the faults. We will assess whether this is a result of the Sierran-Great Valley block impinging on the San Andreas fault system, or a strain effect caused by the Mendocino fracture zone. We will determine realistic uncertainties of our strain accumulation models using the bootstrap techniques, and test methods for adding geologic and other information using Bayesian techniques to test whether the additional information can reduce the correlations and provide better resolution on other parameters. For example, the slip-rate and locking depth (10.5-22.6 mm/yr and 4.7-44.6 km, 95% confidence) on the San Andreas fault should be much better resolved by applying constraints derived from other seismic, geodetic, and paleoseismic observations.
We appreciate support for this project by the USGS NEHRP through grant numbers 02HQGR0064 and 03HQGR0074. We thank André Basset, Maurizio Battaglia, Dennise Templeton, and especially Todd Williams for assistance conducting the survey.
Freymueller, J. T., M. H. Murray, P. Segall, and D. Castillo, Kinematics of the Pacific-North America plate boundary zone, northern California, J. Geophys. Res., 104, 7419-7442, 1999.
Murray, M. H., and P. Segall, Modeling broadscale deformation in northern California and Nevada from plate motions and elastic strain accumulation, Geophys. Res. Lett., 28, 4315-4318, 2001.
Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: firstname.lastname@example.org
© 2004, The Regents of the University of California