Crustal attenuation and site effects at Parkfield, California

At Parkfield, California, the San Andreas fault separates strongly contrast ing rock types: high velocity Salinian granite to the SW and lower velocity Franciscan basement to the NE. I use the recordings of nine earthquakes in 10 boreholes (less than or equal to 1000 m deep) and calculate P and S wav e spectral ratios between borehole recordings at different depths to elimin ate the source. I then invert for Q on each side of the San Andreas fault. I find that the attenuation on the NE side df the fault (Q(NE)similar to 10 0) is approximately twice that on the SW side (Q(SW)similar to 200) in the depth range similar to 200 m to 5 km. Q(P) and Q(S) are similar, but their ratio is not well resolved. I estimate Q(P)similar to 50 and Q(S)similar to 80 in the fault zone itself by combining recordings at a site within the f ault zone with the inversion results and confirm that it is a low -Q zone. Q is also observed to increase with depth, and the rate of this increase de creases as the depth increases. In the upper 1 km at the Varian well (on th e NE side of the San Andreas fault), Q(P)similar to 30 and Q(S)similar to 2 0, with Q(P) increasing from similar to 20 between 0 and 300 m to similar t o 55 between 600 and 900 m. This study demonstrates that the borehole seism ometers at Parkfield are not deep enough for the effects of attenuation to be ignored in analysis of small earthquakes. Also, the attenuation below 1 km on the NE side of the fault is actually larger than that below similar t o 200 m on the SW side. The results of this study suggest that the attenuat ion at depths of a few hundred to a few thousand meters should be considere d when estimating site effects in seismic hazard studies.