Precursors to ScS Phases and dipping interface in the upper mantle beneath southwestern Japan

Abstract

Longitudinally polarized precursors to ScS phases observed in the Shikoku and Chugoku districts, southwestern Japan, are interpreted as ScSp arrivals, resulting from ScSto-P conversions at a dipping interface in the upper mantle. An ScSp phase recorded in the Tohoku district, northeastern Japan, also is examined. The location of the conversion interface, beneath the Shikoku district, determined from the ScSp observations agrees with the upper boundary of the descending Philippine Sea plate inferred from the seismicity pattern of subcrustal earthquakes. It has been proposed on the basis of no seismic activity in the upper mantle that the leading edge of the downgoing Philippine Sea plate has not reached the upper mantle beneath the Chugoku district. The ScSp observations, however, present a possibility of the existence of an aseismic continuation of the Philippine Sea plate in the upper mantle beneath the Chugoku district. An alternative interpretation of the conversion interface may be possible. The interface inferred in the present study may correspond to a boundary between the asthenosphere and an aseismic dead slab which had descended from the Nankai trough at the previous cycle of plate convergence. This ScS-to-P conversion interface may be closely related to the late Quaternary volcanism in the Chugoku district. Low-velocity zones are required in the vicinity of the inclined ScS-to-P conversion interfaces in the upper mantle beneath southwestern and northeastern Japan in order to explain the observed relative polarity between the ScSp and ScS phases. The first-order discontinuity of the velocity contrast of about 6%, which has been suggested between the lithosphere and overlying asthenosphere, cannot simultaneously explain all of the observed amplitudes, periods, and polarity of the ScSp phases. The simplest model which explains these observations is the low-velocity zone with the sharp upper and transitional lower boundaries. This low-velocity zone may be related to the melting or dehydration of the oceanic crust which has been subducted with the underlying lithospheric plate. The ScS-to-P conversion interface inferred beneath southwestern Japan is more transitional than that beneath northeastern Japan. This difference in nature of the conversion interfaces must correspond to the difference in geological histories of both the regions. A high dip angle of the Philippine Sea plate in the Kyushu district and a low dip angle for the Shikoku district suggest that the normal-faulting subcrustal earthquakes beneath the Bungo channel and Iyonada, the western part of the Sevo Inland Sea, can be explained in terms of tearing and hinge faulting at the junction of the two slabs with different dip angles.