Anisotropy at the edges of the mantle: shear wave splitting observations in the upper and lowermost mantle from around the globe

Convection is the mode by which the Earth's mantle transports the majority of its heat outwards towards space. At the edges of the mantle, vertical convection currents give way to horizontal flow, and conduction takes over as the mode for heat to escape. The physical conditions that dominate and the precise nature of the processes that operate at the boundaries are poorly understood, especially in the lowermost mantle, where constraints are lacking. Yet these processes are bound to be of foremost importance in determining the character of mantle convection. One striking feature of the mantle's edges is that they are strongly anisotropic compared to the bulk of the mantle's interior. The anisotropy might be developed by lattice preferred orientation (LPO) under deformation due to convective flow. Or it might be caused by the alignment of sub-wavelength heterogeneities such as fine layers, or elongated melt inclusions. Measurements of anisotropy in the boundary layer regions place strong constraints on the physical conditions and processes that operate in these regions. Measuring the anisotropy in the upper mantle is a crucial first step in making measurements of anisotropy in the lowermost mantle. That is because any signal that samples the lowermost mantle must also sample the upper mantle, and so anisotropy in the upper mantle will contaminate the lowermost mantle signal. We must correct the waveforms for the upper mantle anisotropy in order to measure the signal from the lowermost mantle. A knowledge of upper mantle anisotropy is not only a useful geophysical measurement for studying the upper mantle, it is also the key to unlocking the signal of anisotropy in the lowermost mantle. In this study we take a step towards creating a global map of anisotropy in the upper and lowermost mantle. We use the complete data set of waveforms from the IRIS DMC covering the period from 1998-2010. We present a map of anisotropy in the upper mantle as detected by SKS and S phases beneath stations in our dataset, and some measurements from S phases beneath events. We also present a map of anisotropy in the lowermost mantle, as detected by ScS phases, where the waveforms are corrected for the upper mantle anisotropy. We have used a newly developed technique in making the upper mantle corrections, which incorporates the uncertainty of the correction into the uncertainty of the final measurement. Our measurements do not make any limiting assumptions about the style or orientation of symmetry axis of the anisotropy. This in principle allows us to detect fabrics with lower than hexagonal symmetry, with axes tilted at any arbitrary angle. We compare our measurements with previous studies of anisotropy for the upper and lowermost mantle, and interpret our measurements in light of geophysically plausible mineralogical and geodynamic models, including recent petrophysically generated models of post-perovskite LPO.