[1] As a possible explanation of "hot fingers" in the mantle wedge below the Tohoku region, Japan, the existence of small-scale convection has been proposed. In this study, we performed numerical calculations around the subduction zone in 3D with composite rheology (i.e., a combination of linear and non-linear rheology) and found that small-scale convection could arise when the rheology determined from laboratory experiments is considered. We also calculated 3D structure of expected P-wave anisotropy for the case with and without small-scale convection based on a theory of LPO development and an assumption that anisotropy is approximately represented as hexagonal symmetry, and found that the fast axis of the P-wave propagation projected on the horizontal cross-sections is nearly in the same direction as that of plate motion in many places. It implies that effects of large-scale mantle flow associated with subducting slab is still dominant even in the presence of small-scale convection in mantle wedge. However, in vertical crosssections, the projected fast axis of the P-wave propagation could tilt vertically while that without small-scale convection is almost horizontal. Therefore, future seismological studies that determine the fast direction of P-wave propagation in 3D would give us critical information on the possible existence of small-scale convection in the mantle wedge.

No abstract

No abstract

No abstract

The thermal structure in subduction zones has a strong influence on seismogenesis and arc volcanism. Traditional 2-D models have been used to provide reasonable agreement between models and observations, but in a number of cases clear 3-D effects are present. One such case is in the Northern Japan subduction system. At the junction between Japan and Kurile arcs, surface heat flow and the occurrence of intermediate-depth seismicity are different than in the Tohoku and Hokkaido regions. We investigate the effects of 3-D slab geometry and a local deepening of slab-mantle decoupling depth on the thermal structure in this region based on 3-D finite element approach. We find that both effects produce the along-arc variation of slab surface temperature, which could reach $100 C. The warmer region arises through 3-D effects of thermal conduction and the colder region arises through localized slow incoming flow in the case where 3-D slab geometry is taken into account. 3-D flow arises where a local deepening of slab-mantle decoupling depth is assumed, which leads to both warmer and colder regions. The effects on surface heat flow are small. While intermediate-depth seismicity in the subducted crust is suggested to be controlled by temperature-dependent phase transitions, the predicted changes in thermal structure are not sufficient to cause the observed deepening of seismicity. This suggests that the thermal structure of this subduction zone may be more strongly influenced by time-dependent deformation of the overriding crust and slab.