Global Transition Zone Anisotropy and Consequences for Mantle Flow and Earth's Deep Water Cycle

The transition zone has long been at the center of the debate between multi- and single-layered convection models that directly relate to heat transport and chemical mixing throughout the mantle. It has also been suggested that the transition zone is a reservoir that collects water transported by subduction of the lithosphere into the mantle. Since water lowers mantle minerals density and viscosity, thereby modifying their rheology and melting behavior, it likely affects global mantle dynamics and the history of plate tectonics. Constraining mantle flow is therefore important for our understanding of Earth's thermochemical evolution and deep water cycle. Because it can result from deformation by dislocation creep during convection, seismic anisotropy can help us model mantle flow. It is relatively well constrained in the uppermost mantle, but its presence in the transition zone is still debated. Its detection below 250 km depth has been challenging to date because of the poor vertical resolution of commonly used datasets. In this study, we used global Love wave overtone phase velocity maps, which are sensitive to structure down to much larger depths than fundamental modes alone, and have greater depth resolution than shear wave-splitting data. This enabled us to obtain a first 3-D model of azimuthal anisotropy for the upper 800km of the mantle. We inverted the 2Ψ terms of anisotropic phase velocity maps [Visser, et al., 2008] for the first five Love wave overtones between 35s and 174s period. The resulting model shows that the average anisotropy amplitude for vertically polarized shear waves displays two main stable peaks: one in the uppermost mantle and, most remarkably, one in the lower transition zone. F-tests showed that the presence of 2Ψ anisotropy in the transition zone is required to improve the third, fourth, and fifth overtones fit. Because of parameter trade-offs, however, we cannot exclude that the anisotropy is located in the upper transition zone as well. Azimuthal anisotropy in the transition zone could result from tilted laminated structures, or from the LPO of wadsleyite and hydrous ringwoodite. Anhydrous ringwoodite is mostly isotropic, but it becomes more anisotropic in the presence of water [Kavner, 2003]. The presence of significant seismic anisotropy in the lower transition zone may thus indicate the presence of OH--bearing minerals. This would be consistent with the observed high solubility of water in ringwoodite and wadsleyite, and the hypothesis that the transition zone is a water reservoir. In addition, at most locations the fast azimuth of propagation for Vsv forms approximately a 90° angle in the transition zone with the fast direction found at shallower depths. Assuming that LPO causes the anisotropy and that seismic fast directions are a proxy for flow direction in the transition zone, this angle change combined with mineral physics data could help us infer mantle convective pattern. The robustness of this feature is, however, currently difficult to assess as Love wave overtones are unable to reliably constrain 2Ψ anisotropy at shallow depths. The inclusion of Rayleigh wave fundamental mode data in future work will help resolve that issue.