Multiscale, Finite-Frequency Rayleigh Wave Traveltime Tomography of the Lithospheric Structure beneath South-Central Tibet: A Fully 3-D Approach

Surface wave traveltime tomography is widely used in unraveling shear velocity structure of the Earth's crust and upper mantle. The inversion commonly proceeds in two steps: first inverting measured phase data on a single mode for 2-D phase velocity maps as a function of frequency; secondly using the resulting phase-velocity dispersion curve at each point to constrain shear velocity variations with depth. Previous studies have shown that strong directional- and depth-dependence of scattering are present in surface wave propagation through laterally heterogeneous Earth, and not properly taken into account for 2-D phase-velocity kernels [Zhou et al., 2005]. Here we present a fully 3-D approach using finite-frequency sensitivity kernels for phase delays of fundamental-mode Rayleigh waves to directly invert for spatial variations of lithospheric shear velocities beneath south-central Tibet. Surface wave data used in our study were collected from Project Hi-CLIMB, which deployed a composite linear and regional seismic array of over 200 broadband stations with a large aperture of 800 km and very dense spacing of ~3-8 km across foreland, Himalayan, Lhasa and Qiangtang terranes during 2002-2005. We perform wavelet-based, time-frequency analysis to discern and window fundamental-mode surface waves and use a cross-spectral multitaper method to measure differential phases of Rayleigh waves between two receivers recording common distant earthquakes at periods of 20-120 s. Moreover, we incorporate surface wave empirical Green's functions (EGFs) extracted from ambient noise cross correlation to enhance path coverage and improve resolution of shallow crustal structure. Likewise, phase differences of noise-derived Rayleigh waves between two pairs of stations with one in common are measured at periods of 10-33 s. Taking differential phase residuals with respect to those in a regional 1-D model as input data, we conduct multiscale, finite-frequency tomography in fully 3-D across the active zone of the Indian-Asian continental collision. The 3-D Fréchet kernels of phase residual data with respect to S-velocity perturbations are derived based on surface-wave mode summation and single-scattering approximation for multitaper measurements. A wavelet-based, multi-resolution parameterization is then employed to deal with unevenly-distributed data in the inversion and result in the model with data-adaptive resolution. The high-resolution tomographic images would illuminate the configuration of velocity heterogeneities in the crust and uppermost mantle and help understanding the key geodynamic process in the most prominent Himalayan-Tibetan orogeny.