| 論文要旨 |
Geometry of the upper boundary of a subducting plate has
been estimated from various kinds of seismic observation,
such as the spatial distribution of seismicity, the
converted or reflected phases from natural or artifical
sources, and receiver function analysis. In this study,
we made use of the converted phase from S wave reflected at
the core-mantle boundary and propagating nearly vertically
called ScS phase, then converted to P wave at the upper
boundary of the subducting plate, which is called ScSp
phase, as a precursor of ScS phase. The traveltime
difference between ScS and ScSp phase enables us to estimate
the depth of the plate boundary where the conversion takes
place. The major difficulty in this approach is very weak
ScSp phase. We estimated the detailed geometry of the upper
boundary of the Pacific plate beneath the Hokkaido region,
by taking the advantage of the dense seismic network named
Hi-net.
Since signals of ScSp phase are generally very weak compared
with those of ScS, we stacked recorded seismograms twice to
detect the traveltime difference between ScS and ScSp phases
clearly. At first, we picked several axises with directions
that the plate in the corresponding areas is subducting
grossly, and divided stations into blocks of the axises. The
seismograms of the stations within a given block were
stacked applying the weight in the shape of a gaussian
function to represent the seismogram in center of each
block. Secondaly, we stacked the seismograms stacked above
arround the apparent slowness of ScSp in each profile, based
on Neighbourhood Algorithm (NA). The final 3-D plate
geometry was estimated by combining these inverted plate
models with a 3-D spline function, and we compared it with
the seismicity in this region.
In order to refine the above result, we performed
frequency-wavenumber analysis. We assumed seismic arrays by
grouping adjacent stations of each staion, which is called
the reference station(the total of 112 stations or seismic
arrays). We conducted the beam-forming of the seismograms
and calculated the f-k power spectrum in each seismic array.
The direction of arrival and slowness were obtained from the
spectrum, estimating the conversion point of ScSp for each
reference station, whose 3-D varieties were visualized more
reliably.
Finally, we compared the plate model obtained by the NA
inversion with the conversion depth refined by f-k analysis.
As a result, the geometry of the upper boundary of the
subducting Pacific plate obtained from f-k analysis seemed
to be consistent more than with the seismicity re-examined
by Katsumata et al(2003), in the slab beneath the Hokkaido
region.
|