Data Analysis and Results

Figure 7.2: Velocities relative to stable North America for the BARD stations derived from 1993-2005 data (dark arrows) and for the PBO and Nucleus stations from 2004-2005.5 (white arrows). Ellipses show 95% confidence regions, assuming white noise only. The 35 mm/yr motion between Columbia and the Farallones is primarily due to shear across the San Andreas fault system.

Figure 7.3: Daily position timeseries for POMM, a continuous GPS station located close to the epicenter of the 28 September 2004 Parkfield earthquake. The coseismic displacement, based on one-second solutions, is about 24 mm to the primarily to the west. The postseismic deformation after one-month is about 61 mm to the northwest. Most sites in the continuous network show significant levels of postseismic deformation following an exponential or logarithmic variation in displacement.

Figure 7.4: Coseismic and postseismic deformation of the 28 September 2004 Parkfield earthquake, which ruptured a 25 km section of the fault within the network (star, epicenter; gray circles, aftershocks ). Coseismic are from one-second solutions, and postseismic are from daily solutions one-month following the earthquake.

We use the GAMIT/GLOBK software developed at MIT and SIO to process data from the BARD and other nearby continuous GPS networks. Recent improvements to GAMIT/GLOBK include better accounting of ocean-tide effects, estimating gradients in atmospheric variations, and applying elevation-dependent weighting to the data observables. We process data from more than 70 stations within hours of the completion of the day using rapid or predicted orbits. We have also reprocessed older data from the present to 1991 using improved orbits, and combined these solutions with global IGS solutions provided by SOPAC. Data from over 20 IGS fiducial sites located in North America and Hawaii are included in the solutions to help define a global reference frame.

The estimated relative baseline determinations typically have 1-3 mm long-term scatter in the horizontal components and the 5-10 mm scatter in the vertical. Average velocities for the longest running BARD stations relative to stable North America during 1993-2004 are shown in Figure 7.2, with 95% confidence regions assuming white noise. Also shown are preliminary velocities determined for the PBO and PBO Nucleus networks during 2004-2005.5. In a study using only the continuous GPS stations in northern California and Nevada (Murray and Segall, 2001), we found that the northern Sierra Nevada-Central Valley is tectonically stable with 3 mm/yr right-lateral shear across the Walker Lane-Mt. Shasta seismicity trend. Deformation along the coast in central California is dominated by the active SAF system, which accommodates about 35 mm/yr of right-lateral shear. Our recent Bay Area Velocity Unification (BAVU) study combines survey-mode and continuous GPS solutions into a self-consistent velocity field in the San Francisco Bay area and develops a three-dimensional block model appropriate for the complex fault geometries (d'Alessio et al., 2005). This study finds interseismic rates on faults that are consistent with geologic estimates, undetectable San Andreas Fault-normal compression in the Bay Area, and evidence for up to 5 mm/yr right-lateral shear deformation just east of the central San Andreas Fault.

Transient signals have been observed on BARD stations from at least 3 earthquakes in northern California, including the August 1998 $M_{w}$ 5.1 San Juan Bautista earthquake (Uhrhammer et al., 1999), and the December 2003 $M_{w}$ 6.5 San Simeon earthquake (Rolandone et al., 2004; Murray et al., 2004). The long-anticipated 28 September 2005 $M_{w}$ 6.0 Parkfield significantly displaced all the continuous BARD stations in the Parkfield region, many of which collect at 1-second sampling intervals, allowing very accurate discrimination between coseismic and postseismic displacements. Significant postseismic displacements have been observed at all the stations within 30 km of the fault, most exceeding the coseismic displacements (Figures 7.3 and 7.4). At CARH, which is located between two strands of the San Andreas fault, the postseismic displacements towards the southeast are opposite in direction to the northwest coseismic displacement. Together with creepmeters and borehole strainmeters, the 2004 Parkfield earthquake was one of the most well recorded earthquakes using geodetic data (Langbein et al., 2005). Preliminary results from our investigations of finite-fault rupture models for the Parkfield earthquake from inversions of seismic and geodetic data (Dreger et al., 2005) are presented in the research section (13.2.).