Seismological structure of the upper mantle: A regional comparison of seismic layering

We investigate seismic layering (i.e., discontinuities, regions of anomalous velocity gradients, and anisotropy) and its lateral variability in the upper mantle by comparing seismic models from three tectonic regions: old ( ~ 100 Ma) Pacific plate, younger ( ~ 40 Ma) Phillippine Sea plate, and Precambrian western Australia. These models were constructed by combining two data sets: ScS-reflectivity profiles, which provide travel times and impedance contrasts across mantle discontinuities, and observations of frequency-dependent travel times of three-component turning (S, sS, SS, sSS, SSS, Sa) and surface (R₁, G₁) waves, which constrain the anisotropic velocity structure between discontinuities. The models provide a better fit to observed seismograms from these regions than the current generation of global tomographic models. The Australian model is characterized by high shear velocities throughout the upper 350 km of the mantle, with no low-velocity zone (LVZ) in the isotropically averaged shear velocities. In contrast, the oceanic models are characterized by a thin, high-velocity seismic lid underlain by a distinct LVZ, with a sharp boundary (the G discontinuity) separating them. The G is significantly deeper beneath the western Philippine Sea plate than beneath the (older) Pacific (89 and 68 km, respectively), implying that thermal cooling alone does not control the thickness of the lid. We interpret this discontinuity as a compositional boundary marking the fossilized base of the melt separation zone (MSZ) active during sea-floor spreading. No discontinuity is detected at the base of the LVZ in the oceanic models. The S velocity gradient between 200 and 410 km depth is much steeper in the oceans than beneath Australia. This high oceanic gradient is probably controlled by a decrease in the homologous temperature over this depth interval. The relative depths of the transition zone (TZ) discontinuities are consistent with Clapeyron slopes expected for an olivine-dominated mineralogy. The 660-km discontinuity displays variability in its amplitude that appears to correlate with its depth; shallow and bright beneath the Pacific, deep and dim beneath Australia and the Philippine Sea. Such behavior is possibly caused by the juxtaposition of the olivine and garnet components of the phase transition. Radial anisotropy extends through the upper 250 km of the mantle in the Australia model and through the upper 160 km of the two oceanic models. The magnitude of anisotropy is consistent with that expected for models of horizontally oriented olivine, and the localization of anisotropy in the shallowest upper mantle implies that it reflects strain associated with past or present tectonic events.