Accelerograms from the 1952 and 1954 Kern County and the October 1933 Signal Hill earthquakes were used with detailed site models in an attempt to isolate the principal site, path, and source contributions in the recorded motions. Site models 20,000 ft deep were developed from soil, water and oil borings, geological reports, and near-surface refraction surveys at four accelerograph sites in Los Angeles and one at Taft, California. Analysis was based on linear system theory, using Fourier spectra and hypothesizing vertically incident shear waves.
The four subsurface site models at Los Angeles were adjusted within the uncertainties of the model properties to minimize the standard deviation of computed 1952 bedrock motions. These stations were close together relative to the epicentral distance. Resulting subsurface transfer functions showed, generally, highest amplifications at the Hollywood station and lowest at UCLA, with a 100 per cent range based on the latter.
Attenuation in the crustal path was evaluated using computed bedrock motions and incorporating the data at Taft. The exponent of distance for geometrical spreading was close to minus one. The damping along the path, expressed as Q, was 550 for the 1952 case and 62 for the 1933 case. A spectrum was obtained for the ratio of the source functions for the 1952 and 1954 earthquakes.