Mantle seismic velocity tomography is a powerful tool for exploring the Earth's deep structure. Interpreting the velocity anomalies as thermal anomalies allows us a snapshot glimpse of the dynamics of a convecting mantle.
Isotropic seismic velocities independent of direction are a simplifying assumption made in most tomography models. Velocity anisotropy, however, is often interpreted to exist in many areas. These range from the upper crust due to sedimentary layering, to the upper mantle as a result of viscous interaction between the underlying convecting mantle and the rigid lithospheric slabs, down to the core-mantle boundary, where chemical interactions with the molten outer core or a chemically distinct layer of old slabs may introduce anisotropy.
Anisotropy can have many causes. Often, it is a result of alignment of crystal orientation, or possibly alignment of fractures or pockets of melt within a strain field. In general, while isotropic velocities may only give us snapshots of the current thermal and chemical conditions of the mantle, anisotropy can give us a more dynamic picture by giving us additional information about the stress and strain present in the mantle.
Until recently, much of our work has focused on SH models of the mantle (Li and Romanowicz, 1996; Mègnin and Romanowicz, 2000). Handpicked data from transverse (T) component seismograms were used in these efforts. The transverse component is the easiest to work with because it contains only SH phases without the many more phases present in coupled P-SV system. If we want to look at anisotropy, however, we need all three components of data, so we have developed an automatic picking algorithm to gather wavepackets from the longitudinal (L) and vertical (Z) component seismograms for use in inversions (Figure 28.1).
It is very important to analyze the coverage of the dataset we have gathered. One important area to consider coverage is D", the area immediately above the core-mantle boundary. In the period range in which we gather data (cutoff frequency of 1/32 Hz), we are looking at primarily S energy. This means we can look at the T component as chiefly SH data, and the L and Z components as chiefly SV data. For SH, we have fairly good coverage in D", since ScS and other phases that reflect off the core-mantle boundary multiple times are high energy, since no SH energy can propagate into the fluid outer core. Unfortunately, this is not the case for SV energy. SV phases can convert to P to travel in the outer core, thus reducing the amplitude of the ScS phases. In general, we therefore have poorer coverage for SV in D", than for SH, leading to possible difficulties in our models in the lowermost mantle.
Although a fully anisotropic model requires many more model parameters than an isotropic one, a simplifying assumption of transverse isotropy with a vertical axis of symmetry can be made. In S waves, this shows up as an inconsistency between SH and SV velocities. To get a first order impression of anisotropy present in the mantle, we can separate our dataset into data that is primarily SH sensitive, and that which is primarily SV sensitive. Although this is only an approximation of the actual anisotropy, it does give us some information on some areas of the mantle that are worthy of further investigation. We have done some preliminary inversions of our L and Z component data to obtain an SV model to compare with SAW12D (Li and Romanowicz, 1996), an SH model already in the literature (Figure 28.2).
Although inadequate crustal correction may be causing instability in the uppermost 200 km of the SV model as demonstrated in the RMS profile in Figure 28.2 (C), the correlation is strong in the upper mantle with pronounced dips at the surface (crustal/damping effects?), 300 km, and the transition zone. At 300 km, we can see that the Pacific Ocean is characterized by a fast velocity anomaly in the SH model that is much less prevalent in the SV model. This is consistent with other studies of upper mantle anisotropy (Montagner and Tanimoto, 1990; Ekstrom and Dziewonski, 1998). The decorrelation at 1200 km corresponds with the lowest amplitudes in both models, so while it could be indicative of anisotropy, it might also be due to the relatively small structural signal. The lowermost mantle pattern in SV is probably an artifact of the poor data coverage discussed above.
Dziewonski, A.M., and D.L. Anderson, Preliminary reference Earth model, Phys. Earth Plan. Int., 25, 297-356, 1981.
Ekström, G., and A.M. Dziewonski, The unique anisotropy of the Pacific upper mantle, Nature, 394, 168-172, 1998.
Li, X.D., and B. Romanowicz, Global shear model developed using nonlinear asymptotic coupling theory, Jour. Geophys. Res., 101, 22,245-22,272, 1996.
Megnin, C., and B. Romanowicz, The 3D shear velocity structure of the mantle from the inversion of body, surface, and higher mode waveforms, Geophys. Jour. Inter., 143, 709-728, 2000.
Montagner, J.P., and T. Tanimoto, Global anisotropy in the upper mantle inferred from the regionalization of phase velocities, Jour. Geophys. Res., 95, 4797-4819, 1990.