The purpose of this paper is to document the strong- and weak-motion seismic data from the Loma Prieta earthquake and its aftershocks obtained by Lawrence Livermore National Laboratory (LLNL), and to present some analysis of the spectral seismic response using both weak- and strong-motion recordings. LLNL operates six free-field, digitally recorded, triaxial, strongmotion accelerographs in the vicinity of LLNL; five of these were operating during the Loma Prieta earthquake. Two days after the main event, LLNL initiated a field deployment of 3-component weak-motion instruments to record aftershocks at three LLNL sites and four California Strong Motion Instrumentation Program (CSMIP) sites that recorded strong-motion from the main event.
Spectral ratios of strong- and weak-motion recordings are computed for two pairs of rock and soil sites. One pair of stations is in the vicinity of LLNL, and the other pair is Treasure Island TRI (fill) and Yerba Buena Island YBI (rock) in San Francisco Bay near the Bay Bridge. For the first pair, the weak-motion spectral ratios predict the strong-motion amplification, within 95% confidence limits, for frequencies from 3 to 12 Hz. For TRI and YBI, the strong-motion spectral ratio is much lower than the weak-motion 95% confidence region for frequencies from 1 to 7 Hz. The strong-motion ratio, however, still suggests that the soil underlying TRI resulted in a factor of 3 amplification of energy between 1 and 4 Hz. This is in contrast to the factor of 8 amplification of the weak-motion energy, derived from the spectral ratios of 7 Loma Prieta aftershocks. The large difference between the weak-motion and strong-motion spectral ratios reinforces the limitation that weak-motion cannot be used to directly predict strong-motion amplification at sites underlain by soils that may respond non-linearly at high strain levels.
A further examination of weak-motion recordings indicates that the source effect can be removed and the propagation path effects approximated so that the site response can be isolated. Resulting site specific spectral amplifications reveal that the spectral ratio method can lead to erroneous conclusions if the “rock” site has a complicated geology. At two sites near LLNL the apparent diminishing of spectral amplitudes below 5 Hz observed in the spectral ratios was actually due to amplification of spectral response at the rock site. It appears that the reference site spectral ratios at low frequencies may have been influenced by topography or near-surface geologic features. For the other pair of sites, the spectrum at YBI, the rock site, was flat, so that the features in the spectral ratios are due to the seismic response of the soil at TRI.