ABSTRACT The genesis of the forearc magmatism in southwest Japan at 14 Ma was studied using geologic and seismic observations. Before the magmatism, the Shimanto accretionary complexes were uplifted by 1000–3000 m between 21 and 17 Ma during the opening of the Japan Sea and the Shikoku Basin. Opening of the Japan Sea and the Shikoku Basin terminated at 15 Ma, when the Kinan Seamount Chain on the Shikoku Basin activated. The magmatic products are distributed at segment boundaries and in aseismic areas of the subducting Philippine Sea plate. The segment boundaries are located on syncline and anticline axes of the subducting slab at Kyushu. The magmatic products at Shikoku are distributed at places where olistostromes uplifted between 21 and 17 Ma. Beneath the Kumano volcanic rocks at Kii, a significant discontinuity in the locations of deep earthquakes is observed. These observations indicate that rupturing of the subducting slab by the load of the overriding plate occurred at around 14 Ma. The slab rupturing would have enabled subslab asthenosphere and/or magma to be injected into the plate interface through the tear and cause the forearc magmatism. Since the oceanic plate has a number of preexisting weaknesses, such as fracture zones, slab rupturing could occur more commonly than previously considered. The forearc magmatism caused by slab rupture is an important process associated with the growth of continental crust in subduction zones.

ABSTRACT To accurately estimate the coseismic rupture area in a Nankai megathrust earthquake and predict seismic and tsunami hazards, various three-dimensional models of the subducting plate geometry and simple seismic velocity models of the subduction zone in SW Japan have been proposed. However, to ensure consistency among studies, more realistic and reliable standard models must be developed. Here, we use wide-angle ocean-bottom seismographic survey results to develop models of the three-dimensional geometry of the subducting plate and of the three-dimensional P-wave velocity structure around the Nankai Trough. We confirmed the reliability of the proposed models by comparing theoretical first arrivals, calculated from two-dimensional structure models sampled from the three-dimensional model along seismic profile lines, with observed traveltime data. The proposed models are the first to be visible and to attempt to represent the actual seismic velocity structures around the entire Nankai Trough in SW Japan. Although the spatial validity of the three-dimensional velocity structure model could not be strictly evaluated, we confirmed that the differences between hypocenter parameters determined from previously published seismic tomography results and those obtained by using our three-dimensional structure model were sufficiently small (latitude and longitude within ±0.1° and depth within approximately ±5 km). Therefore, our three-dimensional structure model is suitable for use as an initial model for hypocenter determination.

ABSTRACT Understanding the factors controlling earthquake rupture areas in subduction zones is a fundamental question in geodynamic research. In the Nankai Trough, Japan, three major controlling factors have been proposed so far: surface roughness, such as seamounts, on the subducting Philippine Sea plate, a locally dense and rigid upper plate, and stable or unstable frictional properties of the plate-boundary megathrust with abnormal pore-fluid pressure. The most prominent rupture segment boundary in the Nankai Trough is located off the Kii Peninsula. When the rupture starts and propagates beyond this boundary, the entire megathrust breaks, as in the 1707 Hoei earthquake. When the rupture does not propagate beyond this portion, the rupture area is segmented, as in the 1854 Ansei, 1944 Tonankai, and 1946 Nankai earthquakes. In this case, the boundary works as a barrier. The asperity or barrier depends on the frictional behavior along the megathrust in this area. Thick and dense plutonic rocks are the main component of the upper plate of the Kii Peninsula. Magmatic activity associated with the proto–Izu-Bonin arc collision is speculated to have occurred in the middle Miocene. Thus, we propose that the long-term tectonic hysteresis of the upper plate is the main factor controlling the rupture area segmentation in the Nankai Trough.