Refining the image of the San Andreas Fault near Parkfield, California using a finite difference travel time computation technique

The Parkfield Area Seismic Observatory (PASO) was a dense, telemetered seismic array that operated for nearly 2 years in a 15 km aperture centered on the San Andreas Fault Observatory at Depth (SAFOD) drill site. The main objective of this deployment was to refine the locations of earthquakes that will serve as potential targets for SAFOD drilling and in the process develop a high (for passive seismological techniques) resolution image of the fault zone structure. A challenging aspect of the analysis of this data set was the known existence of large (20–25%) contrasts in seismic wavespeed across the San Andreas Fault. The resultant distortion of raypaths could challenge the applicability of approximate ray tracing techniques. In order to test the sensitivity of our hypocenter locations and tomographic image to the particular ray tracing and inversion technique employed, we compare an initial determination of locations and structure developed using a coarse grid and an approximate ray tracer Thurber, C., Roecker, S., Roberts, K., Gold, M., Powell, M.L. , and Rittger, K., 2003. Earthquake locations and three-dimensional fault zone structure along the creeping section of the San Andreas fault near Parkfield, CA: Preparing for SAFOD, Geophys. Res. Lett., 30 3, 10.1029/2002GL016004.] with one derived from a relatively fine grid and an application of a finite difference algorithm Hole, J.A., and Zelt, B.C., 1995. 3-D finite-difference reflection traveltimes, Geophys. J. Int., 121, 2, 427–434.]. In both cases, we inverted arrival-time data from about 686 local earthquakes and 23 shots simultaneously for earthquake locations and three-dimensional Vp and Vp/Vs structure. Included are data from an active source seismic experiment around the SAFOD site as well as from a vertical array of geophones installed in the 2-km-deep SAFOD pilot hole, drilled in summer 2002. Our results show that the main features of the original analysis are robust: hypocenters are located beneath the trace of the fault in the vicinity of the drill site and the positions of major contrasts in wavespeed are largely the same. At the same time, we determine that shear wave speeds in the upper 2 km of the fault zone are significantly lower than previously estimated, and our estimate of the depth of the main part of the seismogenic zone decreases in places by  100 m. Tests using “virtual earthquakes” (borehole receiver gathers of picks for surface shots) indicate that our event locations near the borehole currently are accurate to about a few tens of meters horizontally and vertically.