Complex, multilayered azimuthal anisotropy beneath Tibet: Evidence for co-existing channel flow and pure-shear crustal thickening

Of the two debated, end-member models for the late-Cenozoic thickening of Tibetan crust, one invokes “channel flow” (rapid viscous flow of the mid-lower crust, driven by topography-induced pressure gradients and transporting crustal rocks eastward) and the other—“pure shear” (faulting and folding in the upper crust, with viscous shortening in the mid-lower crust). Deep-crustal deformation implied by each model is different and would produce different anisotropic rock fabric. Observations of seismic anisotropy can thus offer a discriminant. We use broadband phase-velocity curves—each a robust average of tens to hundreds of measurements—to determine azimuthal anisotropy in the entire lithosphere-asthenosphere depth range and constrain its amplitude. Inversions of the differential dispersion from path pairs, region-average inversions and phase-velocity tomography yield mutually consistent results, defining two highly anisotropic layers with different fast-propagation directions within each: the middle crust and the asthenosphere. In the asthenosphere beneath central and eastern Tibet, anisotropy is 2–4 per cent and has a NNE–SSW fast-propagation azimuth, indicating flow probably driven by the NNE-ward, shallow-angle subduction of India. The distribution and complexity of published shear-wave splitting measurements can be accounted for by the different anisotropy in the mid-lower crust and asthenosphere. The estimated splitting times that would be accumulated in the crust alone are 0.25–0.8 s; in the upper mantle—0.5–1.2 s, depending on location. In the middle crust (20–45 km depth) beneath southern and central Tibet, azimuthal anisotropy is 3–5 and 4–6 per cent, respectively, and its E–W fast-propagation directions are parallel to the current extension at the surface. The rate of the extension is relatively low, however, whereas the large radial anisotropy observed in the middle crust requires strong alignment of mica crystals, implying large finite strain and consistent with high-rate horizontal flow. Together, radial and azimuthal anisotropy suggest eastward mid-crustal channel flow in central Tibet, along the regional topography gradient. In NE high Tibet, mid-crustal azimuthal anisotropy is 4–8 per cent and has WNW–ESE and NW–SE fast-propagation directions, parallel to the net extension at the surface. These fast directions are inconsistent with channel flow following the SW–NE regional topography gradient. Instead, they suggest similar net deformation in the (decoupled) shallow and deep crust. In the brittle upper crust, it is accommodated by strike-slip faulting; in the ductile mid-lower crust—by shear oriented at ∼45° to the faults. Although mid-crustal flow beneath NE Tibet may transport some material towards the plateau periphery at a low region-average rate, the dominant mid-crust deformation pattern is shear parallel to the plateau boundary. This implies that channel flow from central Tibet is not the main cause of the on-going crustal thickening farther northeast