北海道大学地球物理学研究報告 = Geophysical bulletin of Hokkaido University;第69号

FONT SIZE:  S M L

ScSp波による北海道周辺での太平洋プレートの形状の推定

長田, 絹絵;吉澤, 和範;蓬田, 清

Permalink : http://hdl.handle.net/2115/21506
JaLCDOI : 10.14943/gbhu.69.135

Abstract

Geometry of the upper boundary of the subducting Pacific plate or slab was estimated in the Hokkaido region, Japan, using the ScSp phase: the converted phase to P wave at the boundary from the S wave reflected at the core-mantle boundary and propagating nearly vertically (i.e., ScS phase). Taking the advantage of a dense seismic network named “Hi-net” recently deployed across the Japanese islands, we applied several seismic array analyses to the recorded waveform data for a large nearby deep earthquake, in order to enhance weak ScSp signals in record. At first, we set up five blocks for the region in the subducting direction of the plate. After aligning the travel time of the ScS phase and stacking seismograms among stations in a same sub-block perpendicular to the plate subduction, we searched for the optimal plate model (i.e., two-dimensional geometry of the upper boundary of the plate) for each block. The model was parameterized with six depths, and the seismograms were stacked based on the travel time of ScSp as time lag at each sab-block, so that the optimal model would yield the maximum amplitude of ScSp after stacking. The searches were done, using ray tracings of the ScSp phase with a reference velocity model and the non-linear inversion scheme called Neighbourhood Algorithm. THe optimal model of each block was combined each other by cubic spline interpolation, in order to construct three-dimensional geometry of the upper boundary of the plate. We then performed the frequency-wavenumber (f-k)spectral analysis to refine the above result. Assuming each station as a reference point, we made each beam output with adjacent 7 stations as a function of wavenumber(kx,ky) and frequency. The peak of its power spectrum was considered as the ScSp signal, estimating the wavenumber vector, that is, the azimuth of arrival and slowness,so that we can estimate the position and depth of the corresponding S-to-P conversion. In the frequency range of 0.5 to 1.5 Hz, we could estimate the conversion points for 21 stations, and refined the geometry of the upper boundary already obtained by the above non-linear stacking approach. The final plate model was compared with the distribution of intraplate earthquakes in the Pacific plate. This comparison clearly reveals that the upper seismic zone merges with the lower from 150 to 200 Km in depth, by systematically deviating away from the upper boundary of the plate.

FULL TEXT:PDF