This article addresses the variability of site response in the Los Angeles area and possible structural causes for the observations. Aftershock records from 231 sites in the San Fernando and Los Angeles basins and the surrounding mountains are used in this study. Spectral ratios, taken with respect to a low-amplitude reference site, are used to document the variation in site amplification in the frequency range 2 to 6 Hz, both spatially and with backazimuth to the source. At higher frequencies (6 to 10 Hz), spectral ratios are shown to have greater spatial variability. Interstation spectral ratios are used to measure the standard deviation among source as a function of station separation. An increase in the variation in ground motion is shown to take place at a station separation of 1 km. Relative site-response estimates between nearby stations are used to demonstrate that preferred directions of motion can exist even in areas with no surface topographic effects.
Significant variations in site response exist over short baselines (up to a factor of 2 over 200 m) that are not explained by differences in surficial geology or shallow shear-wave velocity. A variety of investigative approaches is used, including spectral ratios, arrival-time variations, 1D and 2D waveform modeling, and comparison with seismic reflection lines, to determine the most likely causes of these observations. A correlation is demonstrated between late arrival times of P and S waves and larger site amplification in Sherman Oaks and Northridge. This observation, in conjunction with waveform modeling and seismic reflection profiles, is used to infer that sedimentary structures in the upper 1 to 2 km and topography on the sediment-basement interface play an important role in determining site amplification. These structures, in the form of folds and buried basins, focus energy in spatially restricted areas at the surface. Comparison of displacement waveforms at nearby stations having disparate site amplifications, complemented by known shallow shear-wave velocities at selected sites, is used to support the argument that these structures, in some cases, can be the dominant factor in the modification of local ground motions.