Vertically deflected mantle flow at the eastern edge of the African Large Low Shear Velocity Province

Thursday, 18 December 2014
Heather A Ford1, Maureen D Long1, Xiaobo He2 and Colton Lynner1, (1)Yale University, New Haven, CT, United States, (2)Yonsei Univ., Seoul, South Korea
Despite the abundantly clear evidence for the existence of two Large Low Shear Velocity Provinces (LLSVPs) in the lower mantle, the origin, composition, and dynamics are still not well understood, nor is the relationship of the LLSVPs to the greater mantle. Observations of seismic anisotropy are often used to better understand flow in the mantle, but their utility is limited in the mostly isotropic lower mantle. An exception to this is in the lowermost mantle (D”) where observations of seismic anisotropy are numerous; however, the interpretation of such measurements in terms of dynamic processes remains challenging. Here we use observations of seismic anisotropy at the eastern edge of the African LLSVP beneath the Afar hotspot to 1) test the consistency of different mechanisms for D” anisotropy with observations, and 2) constrain the geometry of mantle flow at an LLSVP edge. Our observational data set consists of measurements of differential S-ScS and discrepant SKS-SKKS splitting, totaling 22 SKS, SKKS and ScS phases, corrected for upper mantle anisotropy. By utilizing different raypath combinations, we are able to sample the edge structure of the African LLSVP from five distinct raypath orientations, an improvement over previous work. Through forward modeling using mineral physics constraints, we find that our splitting observations are best fit by a model of crystallographically aligned post-perovskite, with the [100] axis oriented either vertically, or highly oblique to the horizontal plane. Such an alignment suggests that mantle flow at the eastern edge of the African LLSVP has an upward component, consistent with either the vertical deflection of material at the LLSVP or with sheet-like upwellings. Placing our preferred flow scenario(s) in context with other recent work, we propose that the edge of the African LLSVP acts as a barrier to mantle flow, ultimately deflecting flow upwards along the margins.