Crustal features along the southern Kuril Trench, Japan, obtained by a refraction/reflection seismic survey

The seismogenic zone in the southern Kuril Trench can be divided into two segments by the Kushiro Canyon, the Nemuro segment to the east and the Tokachi segment to the west. Except for the giant compound earthquake in 17th century, [e.g. Sawai et al., 2002], M8 class earthquakes have occurred repeatedly within each of these segments. The 1952 and 2003 Tokachi earthquakes are considered to be repeated rupture of the asperity of the Tokachi-oki segment. In order to reveal the seismic velocity structure related to the rupture propagation or suspension along the plate boundary, we made a seismic survey across the segment boundary between the Nemuro and Tokachi segments. In the experiment, we deployed 16 OBSs along a seismic line with about 180 km length and shot 75 liter airgun to correct wide-angle seismic data, and MCS survey was also made simultaneously. The profile ran through the focal areas of the 2003 Tokachi and the 1973 Nemuro earthquakes along the strike of the Kuril Trench. The first arrival times observed by the OBSs are inverted for 2-D P-wave velocity distribution and locations of major reflectors are imaged by using traveltime mapping method (TMM) [Fujie et al., 2005]. In the obtained crustal velocity model, sedimentary layers with Vp < 4.8 km/s shows significant variation along the profile. In the rupture area of the 2003 Tokachi earthquake, their total thickness is about 8 km, it decrease to about 4 km in the segment boundary zone around the Kushiro Canyon. In the Vp model obtained by Nakanishi et al [2004], the layer with Vp of about 5~6 km/s was interpreted as the upper crustal layer of the Kuril arc. But the present result of the TMM shows that there is a distinct reflective boundary within the layer, which separating the layer into upper and lower units. Judging from its large vertical velocity gradient, the upper unit may be old sedimentary unit. Wells et al [2003] pointed out the correlation between the low gravity anomaly (LGA) zones and areas of large coseismic slip. Based on this relation, they discussed that sedimentary basins are developed above locked portions of the plate boundaries due to basal erosion, including the Tokachi segment. Our structure model demonstrates that a thick sedimentary pond is actually developed in the LGA corresponding to the asperity of the Tokachi segment.