Shear wave polarization anisotropy in the upper mantle beneath Honshu, Japan

Abstract

Shear wave polarization anisotropy in the wedge portion of the upper mantle between a subducting plate and the earth's surface is investigated using three‐component seismograms of intermediate depth and deep earthquakes recorded at 14 local stations in Honshu, Japan. Eighty nine high‐quality seismograms were selected from a period of 3 years. The data used in this study are restricted such that incidence angles are smaller than the critical angle of 30° to the earth's surface in order to avoid phase shifts in the shear wave train. To find directions of the maximum and minimum velocities in split shear waves, where shear waves are resolved into two phases with the maximum time separation, each set of the two horizontal component seismograms is rotated in the horizontal plane. The split shear waves thus obtained are again recombined after the correction of anisotropy, and the anisotropy‐corrected particle motion is compared with the focal mechanism for a cross‐check of the observed anisotropy. Directions of the maximum axes are plotted on azimuthincidence angle stereograms at each station. The stereograms and the cross sections of seismic ray paths show that (1) the anisotropic material is distributed at intermediate location between earthquake sources and receiving stations, and (2) the anisotropic region is separated into two parts: one in the north of the present study area with the polarization of the maximum velocity shear wave trending 0° to 30° from the north (north anisotropy) and the other in the south with it trending 90° to 120° (south anisotropy). The maximum time delays between the two shear waves along a vertical seismic ray is about l s for both the anisotropic regions. The horizontal extent of the anisotropic area in the north is 50 km at depths of 50 to 150 km. perhaps prevalent in west Honshu. However, the depth of anisotropic material is not well constrained because of insufficient coverage of seismic ray paths and angles. If we assume the vertical extent of the anisotropic material to be 100 km, the maximum velocity contrast would be 4%. If we adopt a crack alignment model, for the observed anisotropy, cracks are inferred to be distributed vertically trending 0° to 30° in the north and 90° to 120° in the south. If we assume an olivine alignment model, olivine crystals are inferred to be aligned in the north with the a axis of the slip direction trending 0° to 30° and the slip plane normal to the b axis being vertical and in the south anisotropy with the axis trending 90° to 120° and the slip plane being horizontal.