Anisotropy in the Shallow Crust Observed around the San Andreas Fault Before and After the 2004 M 6.0 Parkfield Earthquake

Abstract

Local seismic arrays were deployed at two locations along the San An- dreas fault (SAF) near Parkfield, California, before and after the 2004 M 6.0 Parkfield earthquake. Using local earthquakes we determine the anisotropic field within 1- 2 km of the main trace of the SAF at the two array locations separated by 12 km. The initial array, near the SAFOD site, was deployed for six weeks in October and November 2003, and the second array, located near the town of Parkfield, was de- ployed for 3 months following the 28 September 2004 M 6.0 Parkfield earthquake. We find the fast shear-wave polarization direction nearly fault-parallel (N40W) for stations on the main fault trace and within 100 m to the southwest of the SAF at both array locations. These fault-parallel measurements span the 100- to 150-m-wide zone of pervasive cracking and damage interpreted from fault-zone-trapped waves associated with the main fault core (Li et al., 2004, 2006). Outside of this zone, the fast orientations are scattered with some preference for orientations near N10E, roughly parallel to the regional maximum horizontal compressive stress direction (rh). In addition, fast directions are preferentially oriented parallel to a northern branch of the SAF recorded on stations in the 2004 Parkfield deployment. The measured anisotropy is likely due to a combination of stress-aligned micro- cracks away from the fault and shear fabric within the highly evolved fault core. The majority of our measurements are taken outside of the main fault core, and we es- timate the density of microcracks from the measured delay times. Apparent crack densities are approximately 3%, with large scatter. The data suggest weak depth dependence to the measured delay times for source depths between 2 and 7 km. Below 7-km source depth, the delay times do not correlate with depth suggesting higher confining pressure is forcing the microcracks to close. No coseismic variation in the anisotropic parameters is observed, suggesting little to no influence on measured splitting due to the 2004 M 6.0 Parkfield earthquake. However, the premainshock and postmainshock data presented here are from arrays separated by 12 km, limiting our sensitivity to small temporal changes in anisotropy.