Layered structure persists at the tops of subducting plates to depths in excess of 150 km. A low-velocity channel several km thick produces guided seismic guided waves, which indicate that the top of the downgoing plate remains coherent and probably has not eclogitized to > 150 km depth. The surrounding medium has seismic velocities 3-6% faster than standard earth models, as inferred from direct travel times, presumably because the waveguide lies atop cold descending lithosphere. However, guided waves travel more slowly. Several tests with full waveform synthetics show that the guided waves can be observed only if excited by sources that lie near or within the low-velocity channel. Thus, the frequent observation of dispersed body waves indicates that intermediate-depth earthquakes tend to lie in waveguides. Also, receivers must be favorably placed. Observations for seven circum-Pacific subduction zones show such signals, which can be explained as low-velocity channels 2-8 km thick. Velocity anomalies within low-velocity channels are largest at shallowest depths, up to 14% slower than surroundings at depths less than 150 km, diminishing to < 3% at greater depths. This variation in depth indicates that low-velocity channels metamorphose at depths that can exceed 150 km to a rock closely resembling eclogite or peridotite. Metamorphism probably involves dehydration of a layer initially abundant in hydrous minerals, although equilibration of variable fractions of metastable gabbro may explain some observations. The size of the low-velocity anomaly correlates with slab dip, with steeply dipping slabs exhibiting the greatest anomalies, but does not correlate with thermal parameter or convergence velocity. This correlation may reflect a greater ability of steep slabs to funnel fluid up the slab rather than into the overlying wedge, so that slow seismic velocities integrate effects of subduction of hydrous materials and their subsequent up-dip fluid transport. At depths exceeding 150 km, the waveguide velocities place an upper bound on the amount of water that could be delivered to deeper mantle, to 2wt.% or less within this 4 km thick layer. (c) 2004 Elsevier B.V. All rights reserved.
909JKTimes Cited:13Cited References Count:58