COMPUTER STUDIES OF MICROSEISM STATISTICS WITH APPLICATIONS TO PREDICTION AND DETECTION.

摘要

Computational experiments have been performed on seismic data digitized from the records obtained by the Air force during the Logan and Blanca underground nuclear shots,by Dr. Bruce Bogert in New Jersey and by the Wichita Mountain Seismic Observatory. The experiments indicate that microseismic noise of about .3cps frequency is associated with the oceans but the higher frequencies are not. Attempts to identify definite wave types,such as Rayleigh and Love waves,and to follow wave packets from station to station failed,but the failure illustrated the complexity of the microseisms and points out the necessity of a statistical study. For the statistical studies the microseisms were considered to be stochastic time series. It was found that the probability densities of the amplitudes were Gaussian and were not independent. Spectral analysis showed the typical microseism spectrum to have a maximum at about .3cps and often other strong bands at 1.4and 2cps.