Global Anisotropy and the Depth of Continental Roots

The thickness of continental roots has been debated for a long time. Most studies based on heat-flow or the analysis of xenoliths indicate that the coherent, conductive part of continental roots does not exceed 200-250 km depth. Some global seismic tomographic models agree with the estimate, but some indicate much thicker lithospheric roots, reaching at least 400km in depth. We propose that the disagreement between tomographic models is not due to the inability of seismologists to agree, but that it is due to the fact that different models are based on different types of data, which are affected differently by anisotropy in the depth range 250-400 km, under stable continents. Indeed, models based on data with primary sensitivity to SH exhibit deeper roots than those built with mostly SV sensitive data. Because most recent upper mantle tomographic models are based on isotropic inversions, or ignore coupling between SH and SV due to anisotropy, it is important to verify that the differences observed are not an artifact of simplified theoretical assumptions. Indeed, while coupling effects are relatively minor for fundamental mode surface waves, they are non-negligible for higher modes. We have implemented an inversion for transverse isotropy using three component waveform data (~ 85,000 wave-packets of surface waves and ~ 50,000 of body waves) and consider the 6 parameters SH, SV, η (=F/(A-2L)), P(isotropicV_P), φ (=C/A), and ρ, with appropriate kernels for weak transverse anisotropy. We introduce scaling relations for P, ρ, η and φ , as inferred from laboratory experiments (Montagner and Anderson,1989), and invert for SV and SH. We confirm the existence of transverse isotropy with SH>SV under the central Pacific and Indian oceans in the depth range 100-200km (Montagner and Tanimoto, 1991; Ekström and Dziewonski, 1998). At greater depth (200-400km), this signal is replaced by SH>SV located under most continental cratons. Because at these depths frozen lithospheric anisotropy cannot be sustained, we propose that both the oceanic (shallow) and continental (deeper) SH>SV signal indicates horizontal flow in the asthenospheric channel beneath the lithosphere.