Imaging mid-mantle discontinuities: implications for mantle chemistry, dynamics, rheology, and deep earthquakes

Abstract

Using teleseismic network data and new signal processing techniques, I investigated mantle structure in the vicinity of subduction zones. The 660-km seismic discontinuity in the Izu-Bonin region is depressed by up to 80 km, indicative of material 1000$\sp\circ$K colder than ambient mantle. A broad depression of the discontinuity near 33$\sp\circ$N suggests that the subducting slab extends to the west upon the discontinuity from 139$\sp\circ$E to at least 135$\sp\circ$E. Farther to the south near 26$\sp\circ$N, a narrow and confined discontinuity depression suggests local penetration. Along the entire Mariana-Bonin-Izu-Japan-Kuril Trench, the dip of the slab changes to vertical at the deepest earthquake, regardless of the depth of the earthquakes. This observation implies that the slab loses strength at the seismicity cutoff. A grain size reduction due to the olivine to spinel phase change would create a weak zone within the slab, eliminate the slab membrane strength, and thus provide a mechanism allowing the slab to fall vertically downward. The 660-km discontinuity is less than 3 km thick and the shear wave jump across the discontinuity is.40 km/sec $\pm$.05 km/sec. This jump agrees with the velocity jump predicted of a pyrolite mantle containing 4% cation aluminum at the bottom of the transition zone. I observed the 410-km discontinuity infrequently, suggesting this discontinuity broadens in the subduction zone, possibly due to high water content. There exists no consistent sharp ($<$10 km) discontinuity between 660 km and 410 km depth and between 410 km and the crust. Beneath the 660-km discontinuity in Izu-Bonin, there does not exist another horizontal discontinuity to at least 1300 km. If a chemical change is associated with the 660-km discontinuity, it produces a very small change in shear velocity ($\delta Vs<1\%).$ The absence of a discontinuity in the 660 km-1300 km depth range shows that a hypothesized global discontinuity does not exist at these depths. Applying new migration and stacking techniques to global waveform data, I used S-to-P scattered arrivals to image a deep dipping structure beneath Izu-Bonin, which I interpret as an ancient slab. Reinterpretations of S wave tomography, P wave tomography, and plate reconstructions all confirm this result.