Adjoint tomography of teleseismic direct P and coda waves for high-resolution lithosphere structure: methodology and application

Abstract:

Traditional teleseismic body wave travel tomography calculates sensitivity kernels based on 1D Earth's models, which cannot account for effects from the local heterogeneities of a study region. In this study, we adopt a hybrid method by integrating the frequency-wavenumber technique with a spectral-element numerical solver to obtain 3D structural sensitivity kernels in 3D heterogeneous media. We apply this technique to adjoint tomography of teleseismic direct P and coda waves recorded by a dense linear array along with other regional stations in Central California. We collect about 2850 P arrivals and waveforms from vertical and radial components of 43 teleseismic events. Then, we conduct traveltime and waveform adjoint tomography of the direct P waves and coda waves respectively. As a comparison, we also perform the finite-frequency tomography of the traveltime residuals using the same data set based on the ray theory and paraxial approximation.

Our tomographic results and synthetic tests suggest: (1) Iterative traveltime adjoint tomography reveals more pronounced Vp structure than traditional ray theory based tomography. (2) Waveform inversion of P and coda waves can resolve sub-wavelength density and velocity structures. Vp image from traveltime inversion shows a dominating high velocity body in the upper mantle beneath western Great Valley, renowned as the Isabella anomaly. Benefiting from high frequency scattered waves, our waveform inversion reveals more local heterogeneities and help better delineate the Moho discontinuities and geometry of the Isabella anomaly. Adjoint tomography of teleseismic P waves based on the hybrid method can be widely applied to regional seismic arrays to obtain fine-scale lithosphere structures.