The response to earthquake motion of a site in one sediment-filled valley was observed to depend strongly on frequency and the position of the site within the valley and weakly on the input signal's azimuth and incidence angles. The observed input motion was weak, varying between 10−5 and 10−3g. Measurements of ground velocity were made on profiles across and along the valley and simultaneously on two nearby rock sites. The valley was 400 m wide and 700 m long, with a maximum sediment thickness of 60 m. The average seismic impedance contrast between the sediments and underlying basement rock was about 6:1.
Ratios of Fourier spectra from soil sites to spectra from nearby rock sites showed apparent site amplifications of as much as a factor of ten, depending strongly on frequency and the distance of the site from the valley edge. The scatter in spectral ratios for earthquakes with different azimuths and angles of incidence was about a factor of two; thus local site effects that changed the amplitude of incident motion by a factor of two or more could be identified in the records of most earthquakes. Ground motion at valley-edge and mid-valley sites, separated by less than 100 m, differed by as much as a factor of five. This difference could cause large differential motion in structures spanning these sites.
A theoretical, flat layer model of sediment response predicts the average behavior of the middle of the valley but not that of the valley edge. The lowest frequency responance apparently involves the sediments across the width of the valley, not just the vertical column of soil beneath each site.