The general correlation of Salt Lake Valley sites located on soft, saturated unconsolidated silty and clayey deposits (that is, deposits with low S-wave velocity) with high seismic amplification at the ground surface motivated our investigation of the relationship between the P- and S-wave seismic velocity (V_p and V_s) of these units and their corresponding observed site response. We found that low, near surface V_s^−1.5 is proportional to seismic amplification measured on the surface, and suspected that we might be able to predict the site response if the near surface seismic velocity structure were known. With this idea in mind we constructed plane-layered seismic impedance models from borehole data in order to correlate impedance structure with variations in site response. Seismic travel times, measured in a borehole at 2-m intervals, were converted into compressional and shear wave seismic velocity profiles for 22 boreholes (average depth = 59 m) in Salt Lake Valley, Utah. Using this impedance model, we estimated site response on the ground surface of the borehole site to within 12% of the measured value in the 0.7 to 1.0 Hz frequency range for six of the twenty sites, and within 20% for 55% of the sites. All except two of our site response estimates are within a factor of two of the measured value. Thus, high values of seismic amplification appear to be partially explained by a near surface high-impedance contrast produced by the low S-wave velocities.

Comparing the downhole data with published Salt Lake Valley ground motion data derived from Nevada nuclear tests shows that increased site response (sites of relative ground motion amplification) is associated with: (1) a lower value of V_s (110 to 400 m/sec), and (2) high Poisson ratios (0.45 to 0.49) derived from borehole V_p and V_s values. The lowest S-wave velocities found in the Salt Lake Valley are comparable to other regions, such as the muds around San Francisco Bay and the lake sediments of Mexico City, with low S-wave velocities and a record of severe seismic wave amplification in previous earthquakes. The very-low-velocity surface layer in Salt Lake Valley is, however, about half as thick (10 to 14 m) as the low-velocity layers around San Francisco Bay and Mexico City.

We also considered the influence of sedimentary basin fill on site response, because coincidentally, the sites of high seismic wave amplification correspond to the locations of thickest basin fill. The valley fill impedance structure, revealed in seismic reflection profiles we acquired, indicate that a more dynamic impedance structure characterizes sites near the center of the valley where the basin fill is thickest. The reflection data from mid-valley sites typically have more high-amplitude reflectors relative to basin edge sites. The sites with a more dynamic impedance structure suggest that the site response would also be correspondingly different. Thus, the broadband, deep basin effect on site response might account for some of the discrepancy between observed and predicted site response in this study.