Attenuation of seismic waves in the aftershock region - A case of the Off-Izu Peninsula earthquake of 1974.

Abstract

This paper reports a possible effect of attenuative properties of an aftershock region on the nature of seismic body waves which travel through it, by taking as an example of the aftershock events of the Off-Izu Peninsula earthquake, which occurred on May 9, 1974 with magnitude 6.9. For the events in the northwestern end of the aftershock region, the two stations IRT and NHT are chosen for spectral analyses as both stations are located at nearly equal distances (14km) from these events, but the wave path to IRT passes through the aftershock region whereas that to NHT does not. The differences of overall frequency response of the observation systems at NHT and IRT, including station effects, recording, playback, compilation and digitization of the seismic traces on magnetic tape, have been found to be insignificant for the present study. The ratio of the Fourier transform of direct waves observed at IRT to that at NHT is expected to give us information on the attenuative properties in the aftershock region in terms of QI/QN, where QI implies the dissipation factor in the aftershock region and QN that in the undisturbed region of the crust. The spectral ratio was estimated by applying a Wiener filtering. When sampled time-series are short-truncated and random errors are involved in the time-series, the spectral ratio in a least squares sense can be obtained by regarding the observed seismic trace at NHT as an input and that at IRT as an output through the Wiener filter. The Fourier transform of the obtained impulse responses can be considered as the optimum spectral ratio of the transfer characteristics. If we extract the slope κ from the optimum spectral ratio in the graph of amplitude in decibels plotted against frequency, then QI/QN can be obtained as QI/QN≅(1-κln10QN/20πTN)-1 by taking QN as a parameter, where TN denotes the travel time of P-waves from the source to NHT. The frequency decay in aftershock activities obeys the modified Omori's formula with p=1.3, and the spectral analyses are made for four different stages which can be specified by the decay curve. The most probable value of κ is determined from the superposed spectral ratios of appropriate events in each stage. If QN is assumed to be 500-2, 000, then QI is shown to be 100-300 in the first stage, one week after the main shock. This low value of Q in the aftershock region still holds in the second stage, three weeks after the main shock. However, QI tends to recover and will probably take the value from 300 to 600 in the third stage, five weeks after the main shock. The difference between QI and QN becomes rather small in the fourth stage, six months after the main shock. The temporal change of the dissipation factor in the aftershock region described above is closely related to the decay curve in aftershock activities.