Geodetic results from very long baseline interferometry (VLBI), satellite laser ranging (SLR), and the Global Positioning System (GPS) are used to estimate angular velocities between the Sierran microplate, Pacific plate, and North American plate. The Sierra-Pacific pole of rotation lies nearer to the San Andreas fault than does the Pacific–North America pole of rotation and leads to different tectonic implications than if the latter is used. The angular velocities show that the San Andreas fault system and central California Coast Ranges accommodate motion of 39 ± 2 mm/yr, mainly by strike-slip faulting. (All confidence limits following ± signs in this paper are 95% confidence limits.) Fault-normal motion is small, is mainly convergent (at rates up to 3.3 ± 1.0 mm/yr), and varies along the coast, but is divergent (at 2.6 ± 1.2 mm/yr) across San Pablo Bay and associated topographic lows across which the Sierran and Central Valley watershed drains to the Pacific Ocean. The mountain ranges tend to be larger where the fault-normal convergence rates are larger. The low convergence rate (0.5 ± 1.8 mm/yr) normal to the San Andreas fault in the Carrizo Plain differs sharply from that previously inferred (8.2 ± 1.2 mm/yr and 4.9 ± 1.6 mm/yr) by Feigl et al. (1993). The difference is due to differences between their and our elastic strain accumulation models and between how their and our Pacific plate reference frames are defined.
The ranges in most places require a minimum of 4 +2/–1 m.y. of fault-normal convergence at the present rate to attain their present cross-sectional area if erosion is neglected, more if it is not. The amount of convergence previously estimated from a balanced cross section across the Diablo Range in central California requires 10 +8/–3 m.y. of convergence at the present rate. The former is consistent with widely held views about the onset of the Coast Range orogeny, but the latter is not. Both are consistent, however, with the recent plate reconstructions by S. Cande, J. Stock, and colleagues, which indicate that Pacific plate motion relative to North America changed to a more convergent direction, 20°–25° clockwise of its prior direction, at ca. 8 to 6 Ma and not at 3.5 Ma, as had been previously inferred. The inferred change in direction of plate motion is large compared with the present angle of convergence across the straight and narrow segment of the San Andreas fault of 0.7°–4.7°, from which we infer that the Sierran microplate changed motion relative to North America at the same time (ca. 8 to 6 Ma) as did the Pacific plate. We further infer that the motion accommodated across the Great Basin must also have changed at the same time.
We also examine the hypothesis that stable sliding occurs along the San Andreas fault and other northwest-striking strike- slip faults in central California where the fault-normal convergence rate is low or negative, and that these faults are unstable where the fault-normal convergence rate is high. Such a relationship appears to hold in general, but fails in detail. In particular, there are substantial sections of fault with small inferred rates of fault-normal convergence across which the San Andreas fault is locked. Moreover, the creeping section of the San Andreas fault (i.e., the section between Parkfield and the Calaveras junction) is the locus of greater fault-normal convergence (3.2 ± 1.4 mm/yr) than is the locked part of the fault (0.5 ± 1.8 mm/yr) south of Parkfield. Thus, this hypothesis is at best a partial explanation for the observed distribution of locked and nonlocked sections of the fault.