Detection of upper mantle flow associated with the African Superplume

A continental-scale, low seismic velocity anomaly in the mid to lower mantle beneath Africa is a robust feature of global tomographic models. Assuming the low velocities are associated with warm, less dense material, the African seismic anomaly has been ascribed to a long-lived thermal upwelling from the lower mantle. Such a large-scale upwelling should also affect the regional horizontal flow field in the upper mantle. To test this model, we compare seismic anisotropy inferred from shear-wave splitting measurements with instantaneous flow calculations that incorporate mantle density structure inferred from seismic tomography. We calculate splitting parameters at 13 ocean island stations surrounding Africa. Splitting measurements from island stations are ideal for interpreting anisotropy induced by asthenospheric flow because they lack a thick overlying lithosphere that may also contribute to the observed anisotropy. We tested for a possible lithospheric contribution by comparing the splitting measurements with the fossil spreading directions. We find that although the fossil lithospheric fabric closely matches the observed fast polarization directions at stations < 500 km from a ridge axis, they are a poor fit to the data at stations located farther off-axis. Thus, we conclude that far from a ridge axis, the observed anisotropy is dominated by asthenospheric flow. To test for an active component of mantle upwelling, we considered several models with varying assumptions about the velocity at the base of the asthenosphere: that it is (1) stationary below plates moving in the no-net-rotation (NNR) or hotspot reference frames, (2) driven by plate motions at the Earth's surface, or (3) driven by a combination of plate-motion and mantle density heterogeneity inferred from either seismic tomography or the history of subduction. We find that the best-fitting flow field is generated by plate motions and density heterogeneity associated with large-scale upwelling originating in the lower mantle beneath southern Africa and is manifest as a radial pattern of flow at the base of the asthenosphere. This model provides a significantly better fit to the observed anisotropy than a model in which mantle flow is driven through a passive response to subduction. The resulting sub-asthenospheric flow field is estimated to have velocities of 0-3 cm/year, and an asthenospheric viscosity of similar to 3.10(19) Pa(.)s is found to be most consistent with the regional anisotropy, geoid height, and dynamic topography. (C) 2004 Elsevier B.V All rights reserved.