Regional differences of random inhomogeneities around the volcanic front in the Kanto‐Tokai area, Japan, revealed from the broadening of S wave seismogram envelopes

Abstract

Broadening of seismogram envelopes around direct S waves of deep earthquakes that occurred in a lithospheric slab reveals random inhomogeneities in the upper mantle and crust. Regional differences in the character of envelope broadening were studied in relation to the volcanic front in the central part of Japan. A total of 58 earthquakes that occurred along the subducting Pacific plate ranging from 80 to 500 km in depth were observed at 73 stations of the Kanto‐Tokai seismic observation network and used in this analysis. Time lags of the maximum peak arrival and the half‐maximum arrival were measured from the onset of the direct S wave on root‐mean‐squared traces of horizontal component seismograms in frequency bands of 1, 2, 4, and 8 Hz. The envelope broadening is weak at stations located along the coastline of the Pacific Ocean, however, the dependence of time lags on the hypocentral distance and frequency becomes strong at stations near the volcanic front. Moreover, the time lags become very long with increasing hypocentral distances and frequencies west of the volcanic front. This broadening of the seismogram envelope can be interpreted as the effect of the diffraction and multiple forward scattering on the seismic wave propagation through random inhomogeneities. The parabolic approximation is applied to simulate the envelope broadening when the correlation distance of velocity inhomogeneities is longer than the seismic wavelength. The numerical simulation indicates that the dependence of the envelope broadening on the travel distance and frequency depends greatly on the type of the autocorrelation function for the randomness. By comparing the observed frequency dependence of the envelope broadening with the result of the numerical simulation, the type of the autocorrelation function was estimated to be close to Gaussian at stations east of the volcanic front and close to exponential west of the volcanic front. The regional difference in the randomness on both sides of the volcanic front might reflect the difference in the short‐wavelength component inhomogeneities.