A detailed receiver function image of the upper mantle discontinuities in the Japan subduction zone

We have imaged the upper mantle discontinuities in a 30×20° large region at the active continental margin of the Japan subduction zone and neighboring areas, using P-to-S converted phases from teleseismic records of permanent broadband stations. The 410 km discontinuity is detected within ±10 km of its global average position. An interesting exception in its observation is a gap near 135°E, very close to the slab penetration of the 410 km discontinuity and in an area where we have rather high data density. The 660 km discontinuity reaches 700 km depth at two places where it is hit directly by the slab. These data generally show good agreement with tomographic results. Both, the 660 km discontinuity depression and the P-velocity anomalies, suggest about 400–500 K temperature deficit within the slab. However, no downward bending of the 660 km discontinuity is observed in eastern China where the flat lying slab is imaged by tomography. This suggests that the slab does not cool the 660 km discontinuity in this region. Therefore the positive buoyancy required to keep the slab lying flat cannot have been provided by the negative Clapeyron slope of the spinel–perovskite phase transition. Another mechanism is needed, which could possibly be metastable olivine in the cold core of the slab. We have also imaged the shallower portion of the slab down to about 150 km underneath some seismic stations, likely because metastable gabbro is still existing to this depth providing a sufficient velocity contrast. A strong negative converted phase is observed at 150–200 km depth underneath a volcanic region in Japan. If real (observed at one station only), the zone which produces this conversion and which may begin exactly above the slab would require fluid-fed melting.