We estimate the velocity field in central and southern California using Global Positioning System (GPS) observations from 1986 to 1992 and very long baseline interferometry (VLBI) observations from 1984 to 1991. Our core network includes 12 GPS sites spaced approximately 50 km apart, mostly in the western Transverse Ranges and the coastal Borderlands. The precision and accuracy of the relative horizontal velocities estimated for these core stations are adequately described by a 95% confidence ellipse with a semiminor axis of approximately 2 mm/yr oriented roughly north-south, and a semimajor axis of approximately 3 mm/yr oriented east-west. For other stations, occupied fewer than 5 times, or occupied during experiments with poor tracking geometries, the uncertainty is larger. These uncertainties are calibrated by analyzing the scatter in three types of comparisons: (1) multiple measurements of relative position (''repeatability''), (2) independent velocity estimates from separate analyses of the GPS and VLBI data, and (3) rates of change in baseline length estimated from the joint GPS+VLBI solution and from a comparison of GPS with trilateration. The dominant tectonic signature in the velocity field is shear deformation associated with the San Andreas and Garlock faults, which we model as resulting from slip below a given locking depth. Removing the effects of this simple model from the observed velocity field reveals residual deformation that is not attributable to the San Andreas fault.

Baselines spanning the eastern Santa Barbara Channel, the Ventura basin, the Los Angeles basin, and the Santa Maria Fold and Thrust Belt are shortening at rates of up to 5¿1, 5¿1, 5¿1, and 2¿1 mm/yr, respectively. North of the Big Bend, some compression normal to the trace of the San Andreas fault can be resolved on both sides of the fault. The rates of rotation about vertical axes in the residual geodetic velocity field differ by up to a factor of 2 from those inferred from paleomagnetic declinations. Our estimates indicate that the ''San Andreas discrepancy'' can be resolved to within the 3 mm/yr uncertainties by accounting for deformation in California between Vandenberg (near Point Conception) and the westernmost Basin and Range. Strain accumulation of 1--2 mm/yr on structures offshore of Vandenberg is also allowed by the uncertainties. South of the Transverse Ranges, the deformation budget must include 5 mm/yr between the offshore islands and the mainland. ¿ American Geophysical Union 1993