The Birth, Life, and Death of the Oceanic Asthensphere from Seismic Anisotropy

Oceanic asthenosphere represents one of the basic components of plate tectonics. It permits the surface plates to move rapidly by partially decoupling them from the mantle below. The existence of such a mechanical asthenosphere is best demonstrated by seismic anisotropy. Indeed, models for mantle flow including such an asthenosphere, driven by known mantle density heterogeneity and the motions of the surface plates, predict surprisingly well the orientation of anisotropy from both shear-wave splitting fast polarization directions and azimuthal anisotropy in surface waves. What is the fate of asthenosphere upon subduction of the oceanic plate? A global survey of below-slab anisotropy from splitting (Long and Silver, 2008) suggests that all but two subduction zones exhibit trench-parallel fast polarization directions that argue for the dominance of trench-parallel flow, rather than the asthenospheric shear flow found beneath oceanic plates, or the broader slab-entrained flow that would be expected in the absence of any asthenosphere. Such a result is most consistent with the existence of a thin (of order 10 km) decoupling zone beneath the slab throughout the upper mantle, which we take to be an attenuated asthenosphere. The two subduction zones that exhibit below-slab trench-normal fast polarization directions are Cascadia and Mexico. It is significant that these are subduction zones where ridge-trench separation is the smallest. These two cases can be accounted for if asthenosphere has a formation stage that requires a minimum amount of finite strain. If this minimum strain is not reached prior to subduction, then slab-entrained flow, rather than trench-parallel flow, would dominate the below-slab flow field and lead to trench-normal fast polarization directions. We thus seek a mechanism for the asthenosphere that explains its formation stage, its maintenance beneath an oceanic plate, and upon subduction, its persistence throughout the upper mantle but in an attenuated form. One popular explanation for a weak asthenosphere is that it marks a maximum in homologous temperature and a corresponding minimum in viscosity. Yet, such a mechanism, being solely dependent on pressure (depth), does not explain the observed anisotropic characteristics, because it predicts that the asthenosphere would be fully developed at the ridge, and would completely disappear upon subduction when the bottom of the slab reaches the base of the asthenosphere. A more promising mechanism is strain- weakening by shear heating. While shear-heating is known to be a negligible source of heat for typical asthenospheric viscosities (e.g. 1019 Pa s, Turcotte and Schubert, 1982), if asthenosphere formed from shear heating of ambient mantle possessing an order of magnitude higher viscosity, then this could produce, by itself, an increase in temperature of order 100°C and a reduction in viscosity of about an order of magnitude. This mechanism would explain the persistence of asthenosphere at greater depth in subduction zones, due to the advection of this heat, and the attenuation of the subducted asthenosphere, since the positive buoyancy of the warm asthenosphere would resist subduction. This hypothesis has several testable predictions. For example, it predicts that midplate asthenosphere is warmer than near-ridge asthenosphere by about 100°C. Second, it predicts that other subduction zones with very small ridge- trench separation would have below-slab trench-normal splitting fast polarization directions.