Abstract
At the McGee Creek, California, site, three-component strong-motion accelerometers are located at depths of 166 m, 35 m, and 0 m. The surface material is glacial moraine, to a depth of 30.5 m, overlying hornfels. Accelerations were recorded from two California earthquakes: Round Valley, ML 5.8, 23 November 1984, 18:08 UTC and Chalfant Valley, ML 6.4, 21 July 1986, 14:42 UTC. Anti-plane shear strains for the Round Valley and Chalfant Valley events are less than 1.8 * 10−5 and 1.2 * 10−5, respectively. By separating out the SH components of acceleration, we were able to determine the orientations of the downhole instruments. Peak SH accelerations at the surface were 5.7 times greater than those at 166 m in both earthquakes. A constant phase velocity Haskell-Thomson model was applied to generate synthetic SH seismograms at the surface using the accelerations recorded at 166 m. In the frequency band 0.10 to 10.0 Hz, we compared the filtered synthetic records to the filtered surface data. The onset of the SH pulse and the reflections from the interface at 30.5 m are clearly seen in both the synthetics and the recorded data. The synthetic records closely match the data in both amplitude and phase. The fit between the synthetic accelerograms and the data indicates that the seismic amplification at the surface is a result of the resonance of the surface layers and the contrast of the impedances (shear stiffness) of the near surface materials. The impedance contrast of the materials does not account for all of the amplification of the peak accelerations at the surface. The resonances of the layer system below 10 Hz contribute the most to this amplification. We find that a linear model predicts the soil response for frequencies up to 10 Hz. Material damping is added to the soil layers of the model in the form of a quality factor Q. The damping moderates the strong resonance effects by allowing energy to radiate into the half-space. The quality factor Q = 10 is constant in this frequency range but does not accurately model the attenuation at frequencies above 10 Hz.
