Transition Zone Azimuthal Anisotropy From Love and Rayleigh Wave Higher Modes

We built a new global 3-D seismic anisotropy model and demonstrated the presence of seismic anisotropy for vertically (SV) and horizontally (SH) polarized shear-waves down to ~800km depth. Seismic anisotropy is caused by the lattice preferred orientation (LPO) of elastically anisotropic minerals and is taken as an indication of mantle deformation. While the presence of seismic anisotropy is well established in the upper 250km of the mantle, its strength and location is more controversial at larger depths due to the lower resolution of commonly used seismic data. Here we used the anisotropic components of higher mode Rayleigh and Love wave phase velocity maps (Visser et al., 2008) to model SV and SH azimuthal anisotropy. The high sensitivity of these data to structure in the deep upper mantle and topmost lower mantle enabled us to resolve anisotropy at greater depths than in previous models. We found that both the topmost mantle and the transition zone (TZ) are seismically anisotropic. The root mean square SV anisotropy amplitude displays several stable peaks: one of about 2% in the uppermost mantle, and most remarkably, about 1% anisotropy above, inside, and below the TZ. The average amplitude for SH anisotropy displays several stable peaks as well, including one in the TZ. In the uppermost mantle, our 3-D model is in general agreement with previous global studies. In the lithosphere our results are compatible with the idea of frozen-in anisotropy due to olivine LPO during past mantle deformation. In the asthenosphere, we can interpret our model in terms of olivine LPO in response to horizontal shear. Remarkably, on average the fast SV wave direction changes around 250km depth, and at the upper and lower transition zone boundaries where phase transitions are known to occur. This previously undetected correlation between changes in seismic anisotropy and the location of phase changes strongly suggests that TZ anisotropy is caused by LPO of wadsleyite and ringwoodite rather than tilted laminated structure. Wadsleyite has been shown to be able to yield 1% seismic anisotropy at TZ conditions (Tommasi et al., 2004), compatible with our findings. Laboratory experiments showed that both anhydrous and hydrous ringwoodite can be anisotropic and that the presence of volatiles increases the ductile strain rates of olivine aggregates, resulting in stronger LPO (Kavner, 2003). Our detection of seismic anisotropy in the TZ might thus indicate the presence of volatiles. In addition, while the change in fast axes across the TZ boundaries could simply be caused by recrystallization during phase changes, a change in the slip system of the olivine structure might also occur due to the presence of water if it has a similar effect on the anisotropy of TZ material as it has on olivine anisotropy at shallower depths. Our results provide thus new constraints on the nature of the TZ and challenges common views on mantle deformation. In order to interpret the origin of the observed signal and determine whether water plays a role in the presence of TZ anisotropy, more mineral physics studies on the anisotropy of TZ minerals are however needed. In particular, the effect of pressure, melting, water content, deformation mechanisms, and slip systems of TZ materials will need to be addressed.