Superweak asthenosphere in light of upper mantle seismic anisotropy

Earth's upper mantle includes a ∼ 200 km thick asthenosphere underneath the plates where viscosity and seismic velocities are reduced compared to the background. This zone of weakness matters for plate dynamics and may be required for the generation of plate tectonics itself. However, recent seismological and electromagnetic studies indicate strong heterogeneity in thinner layers underneath the plates which, if related to more extreme, global viscosity reductions, may require a revision of our understanding of mantle convection. Here, I use dynamically consistent mantle flow modeling and the constraints provided by azimuthal seismic anisotropy as well as plate motions to explore the effect of a range of global and local viscosity reductions. The fit between mantle flow model predictions and observations of seismic anisotropy is highly sensitive to radial and lateral viscosity variations. I show that moderate suboceanic viscosity reductions, to ∼ 0.01 –0.1 times the upper mantle viscosity, are preferred by the fit to anisotropy and global plate motions, depending on layer thickness. Lower viscosities degrade the fit to azimuthal anisotropy. Localized patches of viscosity reduction, or layers of subducted asthenosphere, however, have only limited additional effects on anisotropy or plate velocities. This indicates that it is unlikely that regional observations of subplate anomalies are both continuous and indicative of dramatic viscosity reduction. Locally, such weak patches may exist and would be detectable by regional anisotropy analysis, for example. However, large‐scale plate dynamics are most likely governed by broad continent‐ocean asthenospheric viscosity contrasts rather than a thin, possibly high melt fraction layer.