Tomographic images of P wave velocity variation at Parkfield, California

Tomographie inversion is applied Io delay times from local earthquakes to image three dimensional velocity variations near Parkfield, California. The 25×20 square km region is represented by nearly cubic blocks of 0.5 km per side. Arrival times of P waves from 551 local earthquakes, with depths of 0 to 15 km, were used as sources producing 3135 rays covering the target region. The data were recorded on low‐noise downhole seismographs. A conjugate gradient method is used to invert the resulting sparse system of simultaneous equations. To diminish the effects of noisy data, the Laplacian of the model parameters is constrained to be small within horizontal layers, providing smoothing of the model. The resolution of the model is estimated by calculating point spread functions at blocks of interest. Estimates of standard errors of the model parameters are calculated by the jackknife statistical procedure. The results of the inversion show correlation with some of the local geological and geophysical features. Station corrections removed the long‐wavelength anomaly associated with the contrast of the Salinian block southwest of the San Andreas fault versus the Franciscan to the northeast. A velocity low located a few kilometers northwest of Parkfield (depth 2.5–3.5 km), appears to lie along the gradient of the large Bouguer gravity anomaly associated with the Parkfield syncline. The south‐southeastward extension of the low velocities may relate to reflections observed on the Parkfield, Consortium for Continental Reflection Profiling (COCORP) lines. We speculate on the geological meaning of these features and interpret them either as part of the local strike slip tectonics or a shallow crustal detachment. The correlation of higher‐velocity features and seismic activity may indicate that earthquakes are occurring in more competent zones while aseismic slip takes place in zones of lower‐velocity, less competent rocks.