Adjoint tomography of surface wave observables from ambient seismic noise

Adjoint tomography has recently been introduced to ambient seismic noise data as a new and promising tomographic method which highlights simulation-based 3-D sensitivity kernels against the approximated analytical theory used in traditional ambient noise tomography. The basic idea of ambient noise adjoint tomography (ANAT) is to iteratively minimize the traveltime misfits between empirical Green's functions (EGFs) from ambient noise cross correlations and synthetic Green's functions from spectral-element simulations based on misfit gradient. To date, most studies use vertical point-force sources to simulate the synthetics for Rayleigh wave EGFs from vertical-vertical component ambient noise cross correlation functions. In this thesis, I develop a semi-automatically iterative inversion package for ANAT based on the software SPECFEM3D Cartesian. I apply the inversion package to ANATof 5 - 50 s Rayleigh wave EGFs in southern California and construct an improved Vsv model based on a community velocity model constructed by 2 - 30 s earthquake data. Furthermore, a general theoretical framework is developed to calculate sensitivity kernels for multicomponent ambient noise cross correlation functions. It is demonstrated that both eastern and northern point-force sources are required to generate horizontal-component synthetics and associated sensitivity kernels based on rotation relations. Similarly, I apply this methodand the inversion package to ANAT of 5 - 50 s Love wave EGFs in southern California and construct an improved Vsh model. The two obtained Vs models (Vsv and Vsh) also give potential insights into the complicated structures of crustal radial anisotropy in southern California. Finally, I apply the method of calculating multi-component sensitivity kernels to investigate the Rayleigh wave ellipticity kernel at regional scale. I show that the lateral variations of the 3-D ellipticity kernel in horizontal direction are non-negligible and should be considered in future tomographic studies using ellipticity observations to image shallow structures.