Direct evidence of site distortion of P- and S-wave microearthquake source spectra at Oroville, California, is presented. The data were gathered by placing vertical and three-component seismometers at 90 m intervals in a 500 m borehole through the Cleveland Hill normal fault, on which the 1975, M_L = 5.7 Oroville earthquake took place. High-pressure, hydraulic locking mechanisms were used to firmly lock the seismometer packages against the borehole wall. Digital, event-triggering GEOS recorders were used to receive the data. Some 30 seismic events, including 12 microearthquakes, were recorded during the 4 months the seismometers were deployed.

By comparing the velocity spectra of microearthquake waves at the different depths, it can be seen that scattering and attenuation in the shallow crust around the borehole dominantly affect high-frequency S waves. Above 15 Hz, the value of the apparent S-wave quality factor, Q_as, for the upper 500 m of crust at this site is 9. One feature of this low Q_as is a gross difference between S-wave “corner” frequencies observed at ground level and 500 m downhole. For example, the uphole and downhole S-wave corner frequencies for the M_coda = 0.4 microearthquake of Julian Day 259 differ by a factor of 2 or more. Low quality factors and depth-dependent corner frequencies were also observed for P waves, but these data are less definitive due to lower signal-to-noise ratios.

Based on their three-component particle motions and polarizations, the direct S waves of the microearthquakes appear to be composed of two phases, which have similar amplitudes but different apparent velocities. It is possible that the velocity differences are a result of anisotropy in the underlying rock.