HR: 1330h
AN: S22A-0416    [PDF]
TI: Earthquake source parameters determined using the SAFOD Pilot Hole vertical seismic array
AU: * Imanishi, K
EM: imani@ni.aist.go.jp
AF: Geological Survey of Japan, Aist, AIST Tsukuba Central 7 1-1, Higashi 1-Chome, Tsukuba, 305-8567 Japan
AU: Ellsworth, W L
EM: ellsworth@usgs.gov
AF: U.S. Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025 United States
AU: Prejean, S G
EM: sprejean@usgs.gov
AF: U.S. Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025 United States
AB: We determined source parameters of microearthquakes occurring at Parkfield, CA, using the SAFOD Pilot Hole vertical seismic array. The array consists of 32 stations with 3-component 15 Hz geophones at 40 meter spacing (856 to 2096 m depth) The site is about 1.8 km southwest of a segment of the San Andreas fault characterized by a combination of aseismic creep and repeating microearthquakes. We analyzed seismograms recorded at sample rates of 1kHz or 2kHz. Spectra have high signal-to-noise ratios at frequencies up to 300-400 Hz, showing these data include information on source processes of microearthquakes. By comparing spectra and waveforms at different levels of the array, we observe how attenuation and scattering in the shallow crust affect high-frequency waves. We estimated spectral level ($\Omega$$_0$), corner frequency ({\it f$_c$}) and path-averaged attenuation ({\it Q}) at each level of the array by fitting an omega squared model to displacement spectra. While the spectral level changes smoothly with depth, there is significant scatter in {\it f$_c$} and {\it Q} due to the strong trade-off between these parameters. Because we expect source parameters to vary systematically with depth, we impose a smoothness constraint on {\it Q}, $\Omega$$_0$ and {\it f$_c$} as a function of depth. For some of the nearby events, take-off angles to the different levels of the array span a significant part of the focal sphere. Therefore corner frequencies should also change with depth. We smooth measurements using a linear first-difference operator that links {\it Q}, $\Omega$$_0$ and {\it f$_c$} at one level to the levels above and below, and use Akaike_fs Bayesian Information Criterion (ABIC) to weight the smoothing operators. We applied this approach to events with high signal-to-noise ratios. For the results with the minimum ABIC, {\it f$_c$} does not scatter and {\it Q} decreases with decreasing depth. Seismic moments were determined by the spectral level and range from 10$^{9}$ and 10$^{12}$ Nm. Source radii were estimated from the corner frequency using the circular crack model of Sato and Hirasawa (1973). Estimated values of static stress drop were roughly 1 MPa and do not vary with seismic moment. {\it Q} values from all earthquakes were averaged at each level of the array. Average {\it Q$_p$} and {\it Q$_s$} range from 250 to 350 and from 300 to 400 between the top and bottom of the array, respectively. Increasing {\it Q} values as a function of depth explain well the observed decrease in high-frequency content as waves propagate toward the surface. Thus, by jointly analyzing the entire vertical array we can both accurately determine source parameters of microearthquakes and make reliable Q estimates while suppressing the trade-off between f$_c$ and {\it Q}.
DE: 7200 SEISMOLOGY
DE: 7209 Earthquake dynamics and mechanics
DE: 7215 Earthquake parameters
DE: 7230 Seismicity and seismotectonics
SC: Seismology [S]
MN: 2003 Fall Meeting