Small seismic events occur continuously at random time intervals within the earth. These events have been referred to as ambient noise, microtremors, or microseisms. Randomly occurring microseisms can be compressed into a single source-pulse and the earth's reflectivity function by taking the autocorrelation of a long sequence of data (Sosie method). In effect, numerous microseisms are compressed into a single earthquake record which can then be used to analyze the response of a site to ground motions. In the frequency domain, earthquake seismologists refer to the reflectivity function as the site amplification response. Near- surface soil layers are known to cause frequency-dependent amplification of ground motions during an earthquake. Ground amplification as high as 25 times bedrock motions were reported at Treasure Island from the Loma Prieta aftershocks. The objective of this proposed study is to show that microseism PSD estimates can be used to predict ground-amplification without the need for ex- pensive soil testing. These direct measurements would provide a more accurate and cost effective means for predicting ground motion.
